computational studies of structure, stability and
Post on 08-Apr-2022
5 Views
Preview:
TRANSCRIPT
Computational Studies of Structure Stability and Properties
of Nanoporous Framework Materials
by
Binit Lukose
A thesis submitted in partial fulfillment
of the requirement for the degree of
Doctor of Philosophy in Physics
Approved by Thesis Committee
_____________________________________ (Chair Prof Dr Thomas Heine JUB)
_____________________________________ (Prof Dr Ulrich Kleinekathoumlfer JUB)
_____________________________________ (Prof Dr Christof Woumlll KIT)
_____________________________________ (Prof Dr Petko Petkov Univ Sofia)
Date of Defense July 19 2012
School of Engineering and Science
Statutory Declaration
I Binit Lukose hereby declare that I have written this PhD thesis independently
unless where clearly stated otherwise I have used only the sources the data
and the support that I have clearly mentioned This PhD thesis has not been
submitted for conferral of degree elsewhere
Bremen 2012
Signature _________________________
i
List of Articles
1 Binit Lukose Agnieszka Kuc Johannes Frenzel Thomas Heine On the reticular construction
concept of covalent organic frameworks Beilstein J Nanotechnol 2010 1 60ndash70
DOI103762bjnano18
2 Binit Lukose Agnieszka Kuc Thomas Heine The Structure of Layered Covalent-Organic
Frameworks Chem Eur J 2011 17 2388 ndash 2392 DOI 101002chem201001290
3 Binit Lukose Barbara Supronowicz Petko St Petkov Johannes Frenzel Agnieszka B Kuc
Gotthard Seifert Georgi N Vayssilov and Thomas Heine Structural properties of metal-
organic frameworks within the density-functional based tight-binding method Phys Status
Solidi B 2012 249 335ndash342 DOI 101002pssb201100634
4 Binit Lukose Agnieszka Kuc Thomas Heine Stability and electronic properties of 3D covalent
organic frameworks Prepared for publication
5 Binit Lukose Mohammad Wahiduzzaman Agnieszka Kuc Thomas Heine Structure
electronic structure and hydrogen adsorption capacity of porous aromatic frameworks
Prepared for publication
6 Jinxuan Liu Binit Lukose Osama Shekhah Hasan Kemal Arslan Peter Weidler Hartmut
Gliemann Stefan Braumlse Sylvain Grosjean Adelheid Godt Xinliang Feng Klaus Muumlllen Ioan-
Bogdan Magdau Thomas Heine Christof Woumlll A novel series of isoreticular metal organic
frameworks realizing metastable structures by liquid phase epitaxy Prepared for publication
7 Binit Lukose Gabriel Merino Christof Woumlll Thomas Heine Linker guided metastability in
templated Metal-Organic Framework-2 derivatives (SURMOFs-2) Prepared for publication
8 Binit Lukose Thomas Heine Review Covalently-bound organic frameworks Prepared for
publication
ii
Acknowledgment
Foremost I would like to thank my supervisor Prof Dr Thomas Heine for the wonderful opportunity to join his group as his PhD student I am greatly thankful to him for giving me the topic and sharing with me his expertise and research insight His thoughtful advices have served to give me senses of motion and direction His ambitious approach to science has given me motivation as well as chances and exposures to develop in science His constant attention and guidance have led my scientific outputs to the best levels possible I am also thankful for the financial support and the comfortable stay in his group during my PhD time Additionally he is acknowledged for correcting and reviewing my thesis
Prof Dr Ulrich Kleinekathoumlfer deserves special thanks as my Thesis Committee member I am very glad to have him in the Committee and greatly thankful for reviewing and evaluating the thesis I also thank him and Prof Ulrich Kortz for the evaluation of my PhD proposal I am thankful also for their friendly manners and considerations throughout my PhD time
Prof Dr Christof Woumlll Director of Functional Interfaces Karlsruhe Institute of Technology is greatly acknowledged for being the external Thesis Committee member I am greatly thankful for the evaluation and reviewing of the thesis I am very much moved by his research outcomes and thankful for sharing them with us Our collaborations with his group have particularly enriched my thesis
Prof Dr Petko Petkov is also acknowledged for reviewing my thesis I particlulary thank him for the friendship and discussions thoughout my PhD time
I am indebted to Dr Agnieszka Kuc for introducing me to the topic of nanoporous materials Her experience and expertise have helped me to begin a career in this field I extend my gratitude for sharing with me her scientific skills and correcting our joint-articles
Dr Lyuben Zhechkov and Dr Achim Gelessus have been great in providing computational assistance I have benefitted from their knowledge and sincerity through fast and timely helps
I owe my heartfelt thanks to Dr Lyuben Zhechkov Dr Nina Vankova Dr Augusto Oliveira Dr Andreas Mavrantonakis Dr Stefano Borini and Dr Christian Walther for all discussions suggestions support help and particularly their lectures Dr Lyuben Zhechkov and Dr Nina Vankova are specially mentioned for their long-term attentions and helps Dr Akhilesh Tanwar is acknowledged for his helps in the beginning of my PhD
In my daily work I have been blessed with a friendly and cheerful group of fellow students Barbara Jianping Wahid Nourdine Mahdi Lei Rosalba Ievgenia Wenqing Guilherme Farjana Maicon Aleksandar Ionut Yulia and Gabriel Discussions aside I had great fun times with them Our interactions have also helped me to develop in a personal level I thank them from my full heart although just a few words are not enough I specially thank Barbara Wahid and Ionut for the joint works and publications
Mrs Britta Berninghausen our project assistant deserves special thanks for the friendly assistance on all matters with the university administration
I thank all the members of the group for a lot of good things From the supervisor to the newly joined member everyone has contributed for the general good fun and easiness All those ldquobio-fuelrdquo workshops barbecues parties retreats and gatherings are unforgettable The group also kept good phase with other groups and visitors I thank all the members once again for the good times I would not have been happier anywhere else
iii
I extend my thanks to the research groups that I visited during the PhD time Dr Sourav Pal Director of National Chemical Laboratory Pune and Dr V Subrahmanian Central Leather Research Institute Chennai deserve my gratitude for giving me the opportunity to visit and work with their group members Also I am very thankful to Prof D Sc Georgi N Vayssilov Faculty of Chemistry University of Sofia for the interesting collaboration and visit to his group The financial assistance during each stay is greatly acknowledged I also thank the members of the respective groups namely Dr Petko Petkov and his family who made the visit to Bulgaria very much entertaining
Prof Dr Lars Pettersson of University of Stockholm Dr Tzonka Mineva of CNRS Montpellier and all other members of the HYPOMAP research project are acknowledged for the scientific discussions exposures and promotions
I acknowledge several projects of Prof Dr Thomas Heine for the financial support of my work and travel the funding sources include the European Commission Deutsche Forschungsgemeinschaft (DFG) and the joint Bulgarian-German exchange program (DAAD)
I thank all the co-authors of my publications who have contributed their knowledge ideas and work to accomplish our scientific goals Without their efforts all those works would not have been complete
Members of Research III of SES at Jacobs University namely Robert Carsten Joumlrg Bogdan Meisam Niraj Mahesh Vinu Pinky Patrice Mehdi Sidhant and all professors postdocs and students in Nanofun center are thankfully mentioned here
A lot of my friends in the campus deserve my thanks Mahesh Mahendran Vinu Deepa Srikanth Rajesh Arumugam Prasad Dhananjay Sunil Tripti Raghu Suneetha Rami Susruta Niraj Abhishek Ashok Rakesh Sagar Rohan Naveen Yauhen Yannic Mila and Samira are thanked for the gatherings travels making funs and those cricket and volleyball evenings Some of them are specially thanked for the occasional ldquogahn bayrdquo parties I owe many thanks to Yauhen Srikanth and Prasad for being good flat-mates and having talks on any matters Srikanth and Prasad are thanked again for generously extending their cooking skills to me
I wish to thank everybody with whom I have shared experiences in life I am obliged to my MSc lecturer Dr Rajan K John whose dreams have inspired and driven me to research In particular his accomplishments in the George Sudarshan Center CMS College Kottayam have molded me to take up this career My previous research supervisors Prof S Lakshmibala and Prof V Balakrishnan of IIT Madras and Dr Anita Mehta of SNBNCBS Kolkata are also acknowledged for their important influences in my academic life Additionally all my teachers friends and well-wishers from neighborhood school college GS Center IIT-M and SN Bose center are thanked and acknowledged Members of St Antonyrsquos Parish Olassa are also thanked for the regards and encouragement
Jacobs University Bremen and its people have been amazing in all sorts of things I am glad that I have been a member of the University With my full heart I thank the university authority for all its facilities that were open for me I also thank Dr Svenja Frischholz Mr Peter Tsvetkov and Ms Kaija Gruumlenefeld in the administration departments for the timely helps
Lastly and most importantly I wish to thank my dearest ones for all the sacrifices and love My parents K P Lukose and Molly and my brother Anit deserve to be thanked They have always supported and encouraged me to do my best in all matters of life I also wish to thank my entire extended family for providing me a loving environment
iv
Abstract
Framework materials are extended structures that are built into destined nanoscale architectures
using molecular building units Reticular synthesis methods allow stitching of a large variety of
molecules into predicted networks Porosity is an obvious outcome of the stitching process These
materials are classified and named according to the chemical composition of the building blocks For
instance Metal-Organic Frameworks (MOFs) consists of metal-oxide centers that are linked together
by organic entities The stitching process is straight-forward so that there are already thousands of
them synthesized Controlled growth of MOFs on substrates leads to what is known as surface-MOFs
(SURMOFs) A low-weight metal-free version of MOFs is known as Covalent Organic Frameworks
(COFs) They consist of light elements such as boron oxygen silicon nitrogen carbon and hydrogen
atoms Diamond-like structures with a variety of linear organic linkers between tetrahedral nodes is
called Porous Aromatic Frameworks (PAFs)
The thesis is composed of computational studies of the above mentioned classes of materials The
significance of such studies lies in the insights that it gives about the structure-property relationships
Density Functional Theory (DFT) and its Tight-Binding approximate method (DFTB) were used in
order to perform extensive calculations on finite and periodic structures of several frameworks DFTB
provides an ab-initio base on periodic structure calculations of very large crystals which are typically
studied only using force-field methods The accuracy of this approximate method is validated prior to
reasoning
As the materials are energized from building units and coordination (or binding) stability vs
structure is discussed Energy of formation and mechanical strength are particularly calculated Using
dispersion corrected (DC)-Self Consistent Charge (SCC) DFTB we asserted that 2D COFs should have a
layer arrangement different from experimental suggestions Our arguments supported by simulated
PXRDs were later verified using higher level theories in the literature Another benchmark is giving an
insightful view on the recently reported difference in symmetries of two-dimensional MOFs and
SURMOFs The result of it shows an extra-ordinary role of linkers in SURMOFs providing
metastability
Electronic properties of all the materials and hydrogen adsorption capacities of PAFs are discussed
COFs PAFs and many of the MOFs are semiconductors HOMO-LUMO gaps of molecular units have
crucial influence in the band gaps of the solids Density of states (DOS) of solids corresponds to that
of cluster models Designed PAFs give a large choice of adsorption capacities one of them exceeds
the DOE (US) 2005 target Generally the studies covered under the scheme of the thesis illustrate
the structure stability and properties of framework materials
- Dedicated to my Family and Rajan sir
Table of Contents 1 Outline 1
2 Introduction 2
21 Nanoporous Materials 2
22 Reticular Chemistry 3
23 Metal-Organic Frameworks 5
24 Covalently-bound Organic Frameworks 8
3 Methodology and Validation 10
31 Methods and Codes 10
32 DFTB Validation 11
4 2D Covalent Organic Frameworks 13
41 Stacking 13
42 Concept of Reticular Chemistry 15
5 3D Frameworks 17
51 3D Covalent Organic Frameworks 17
52 Porous Aromatic Frameworks 18
6 New Building Concepts 20
61 Isoreticular Series of SURMOFs 20
62 Metastability of SURMOFs 21
7 Summary 23
71 Validation of Methods 23
72 Weak Interactions in 2D Materials 25
73 Structure-Property Relationships 27
List of Abbreviations 31
List of Figures 32
References 33
Appendix A Review of covalently-bound organic frameworks 37
Appendix B Properties of MOFs within DFTB 81
Appendix C Stacking of 2D COFs 96
Appendix D Reticular concepts applied to 2D COFs 105
Appendix E Properties of 3D COFs 122
Appendix F Properties of PAFs 131
Appendix G Isoreticular SURMOFs of varying pore sizes 145
Appendix H Metastability in 2D SURMOFs 160
1
1 Outline
I prepared this cumulative thesis as it is permitted by Jacobs University Bremen as all work has been
published in international peer-reviewed journals is submitted for publication or in a late
manuscript state in order to be submitted soon The list of articles contains three published papers
three submitted manuscripts and two manuscripts that are to be submitted The articles are given in
Appendices A-H in the order of their discussions Each appendix has one paper and its supporting
information
The chapters from ldquoIntroductionrdquo to ldquoNew Building Conceptsrdquo contain general discussions about the
articles and provide a red thread leading through the articles The discussions mainly circle around
the context and the content of the articles
The chapter ldquoIntroductionrdquo provides a general introduction to the topic and a review of the materials
discussed in this thesis As no comprehensive review on ldquoCovalently-bound Organic Frameworksrdquo is
available in the literature we prepared a manuscript (Appendix A) covering that subject The chapter
ldquoMethodology and Validationrdquo discusses one article on validation of DFTB method in Metal-Organic
Frameworks (Appendix B) Each consecutive chapter discusses two articles The chapter ldquo2D
Covalent Organic Frameworksrdquo covers the studies on layer arrangements (Appendix C) followed by
analysis on how reticular chemistry concepts can be used to design new materials (Appendix D) The
chapter ldquo3D Frameworksrdquo discusses the stability and properties of 3D COFs (Appendix E) and PAFs
(Appendix F) It also discusses hydrogen storage capacities of PAFs The chapter ldquoNew Building
Conceptsrdquo has coverage on 2D MOFs that are built on organic surfaces The first article in this chapter
describes the achievement of our collaborators in synthesizing a series of SURMOFs of varying pore
sizes supported by our calculations indicating their matastability Extensive calculations revealing the
role of linkers in creating the unprecedented difference of SURMOFs in comparison with the bulk
MOFs is described in another article
Details of the articles and references to the appendices are given in the respective places in each
chapter The chapter ldquoSummaryrdquo gives an overview of all the results and discussions It also discusses
some impacts of the publications and concludes the thesis Overall the studies bring into picture
different classes of materials and analyze their structural stabilities and properties
2
2 Introduction
21 Nanoporous Materials
The field of nanomaterials covers materials that have properties stemming from their nanoscale
dimensions (1-100 nm) In contrast to nanomaterials which define size scaling of bulk materials the
major determinant of nanoporous materials is their pores Nanoporous materials are defined as
porous materials with pore diameters less than 100 nm and are classified as micropores of less than
2 nm in diameter mesopores between 2 and 50 nm and macropores of greater than 50 nm They
have perfectly ordered voids to accommodate interact with and discriminate molecules leading to
prominent applications such as gas storage separation and sieving catalysis filtration and
sensoring1-4 They differ from the generally defined nanomaterials in the sense that their properties
are mostly determined by pore specifications rather than by bulk and surface scales Hence the
focus is onto the porous properties of the materials
Utilization of the pores for certain applications relies on certain parameters such as pore size pore
volume internal surface area and wall composition For example physical adsorption of gases is high
in a material with large surface area which implies significantly high storage in comparison to a tank
Porosity can be measured using some inert or simple gas adsorption measurements Distribution of
pore size can be sketched from the adsorptiondesorption isotherm
Nanoporous materials abound in nature Zeolites for instance many of them are available as mineals
have been used in petroleum industry as catalysts for decades The walls of human cells are
nanoporous membranes Other examples are clays aluminosilicate minerals and microporous
charcoals Zeolites are microporous materials from the aluminosilicate family and are often known as
molecular sieves due to their ability to selectively sort molecules primarily based on a size exclusion
principle A material with high carbon content (coal wood coconut shells etc) can be converted to
activated carbon (AC) by controlled thermal decomposition in a furnace The resultant product has
large surface area (~ 500 m2 g-1) Porous glass is a material produced from borosilicate glasses having
pore sizes usually in nm to μm range With increasing environmental concerns worldwide porous
materials have become a suitable choice for separation of polluting gases storage and transport of
energy carriers etc Synthesis of open structures has advanced this area5-7 especially since the
invention of MCM-41 (Mobil Composition of Matter No 41) by Mobil scientists89 Furthermore
there are many templating pathways in making nanoporous materials10-13 Currently it is possible to
engineer the internal geometry at molecular scales
3
For more than a decade chemists are able to synthesize extended structures from well-defined and
rigid molecular building units Such designed and controlled extensions provide porosity which can
be scaled and modified by selecting appropriate building blocks The first realization of this kind was
a class of materials known as Metal-Organic Frameworks (MOFs) in which metal-oxides are stitched
together by organic molecules Synthesis of molecules into predicted frameworks have led to the
emergence of a discipline called reticular chemistry14 The new design-based synthetic approaches
have produced large number of nanoporous materials in comparison to the discovery-based
synthetic chemistry
22 Reticular Chemistry
The discipline of designed (rational) synthesis of materials is known as reticular chemistry Desired
materials can be realized by starting with well-defined and rigid molecular building blocks that will
maintain their structural integrity throughout the construction process The extended structures
adopt high symmetry topologies The synthetic approach follows well-defined conditions which
provide general control over the character of solids In short it is the chemistry of linking molecular
building blocks by strong bonds into predetermined structures
The knowledge about how atoms organize themselves during synthesis is essential for the design
The principal possibilities rely on the starting compounds and the reaction mechanisms Porosity is
almost an inevitable outcome of reticular synthesis Solvents may be used in most cases as space-
filling agents and in cases of more than one possibility as structure-directing agents
Thousands of materials in large varieties have been synthesized using the reticular chemistry
principles Reticular Chemistry Structure Resource (RCSR) is a searchable database for nets a project
initiated by of Omar M Yaghi and Michael OrsquoKeeffe15 They define nets as the collection of vertices
and edges that form an irreducible network in which any two vertices are connected through at least
one continuous path of edges Building units are finite nets as in the case of a polyhedron Periodic
structures have one two or three-periodic nets An example case of 3D diamond net (dia) is shown in
Figure 1 Graph-theoretical aspects together with explicative terminology and definitions can be
found in the literature16-18
Figure 1 a) ldquoAdamantinerdquo unit of diamond structure and b) diamond (dia) net
4
In other words a framework can be deconstructed into one or more fundamental building blocks
each of them assigned by a vertex in the net The vertices are the branching points and edges are
joining them The realization of the net in space by representing the vertices and lattice parameters
by co-ordinates and a matrix respectively is called embedding of the net Underlying topology of an
extended structure is the structure of the net inherited from the crystal structure that is invariant
under different embeddings For example Cu-BTC MOF has paddlewheels and benzene rings as
fundamental blocks The MOF structure can be simplified into its underlying topology as shown in
Figure 2
Figure 2 CU-BTC MOF and the corresponding tbo net
Alternatively the topology of a framework can be defined using the convention of so-called
secondary building units (SBUs) The shapes of SBUs are defined by points of extension of the
fundamental building blocks SBUs are invariant for building units of identical connectivity Based on
the number of connections with the adjacent blocks (4 for paddlewheel and 3 for the benzene) SBUs
of Cu-BTC can be represented as shown in Figure 3a Assembly of SBUs following the network
topology and insertion of the fundamental building blocks give the crystal structure (see Figure 3b for
the extension of SBUs to the topology of Cu-BTC)
In addition to MOFs Metal-Organic Polyhedra (MOP)19 Zeolite Imidazolate Frameworks (ZIFs)20 and
Covalent-Organic Frameworks (COFs)2122 are developing classes of materials within reticular
chemistry the first members of all of them were synthesized by the group of Yaghi MOPs are nano-
sized polyhedra achieved by linking transition metal ions and either nitrogen or carboxylate donor
organic units ZIFs are aluminosilicate zeolites with replacements of tetrahedral Si (Al) and bridging
oxygen by transition metal ion and imidazolate link respectively COFs are extended organic
5
structures constructed solely from light elements (H B C and O) The materials synthesized under
the reticular scheme are largely crystalline
Figure 3 a) Fundamental building blocks and SBUs of Cu-BTC and b) extension of SBUs following
crystal structure
23 Metal-Organic Frameworks
MOFs are porous crystalline materials consisting of metal complexes (connectors) linked together by
rigid organic molecules (linkers) MOFs are analogues to a class of materials called coordination
polymers (CPs) However there are primary differences between them CPs are inorganic or
organometallic polymer structures containing metal ions linked by organic ligands A ligand is an
atom or a group of atoms that donate one or more lone pairs of electrons to the metal cations and
thereby participate in the formation of a coordination complex In MOFs typically metal-oxide
centers are used instead of single metal ions as they provide strong bonds with organic linkers This
provides not only high stability but also high directionality because multiple bonds are involved
6
between metal-centers and organic linkers Predictability lies in the pre-knowledge about the
connector-linker interactions Thus the reticular design of MOFs derives from the precise
coordination geometry between metal centers and the linkers Figure 4a shows a schematic diagram
of MOFs By varying the chemical composition of nodes and linkers thousands of different MOF
structures with a large variety in pore size and structure have been synthesized Figure 4b shows
MOF-5 that consists of zinc-oxide tetrahedrons and benzene linkers
Figure 4 a) Schematic diagram of the single pore of MOF b) An example of a simple cubic MOF in
which zinc-oxide tetrahedron are linked together by benzene linkers Atom colors blue ndash Zn red ndash
O grey ndash C white ndash H
The synthesis of MOFs differs from organic polymerizations in the degree of reversible bond
formation Reversibility allows detachment of incoherently matched monomers followed by their
attachment to form defect-free crystals Assembly of monomers occurs as single step hence
synthetic conditions are of high importance The strength of the metal-organic bond is an obstacle
for reversible bond formation however solvothermal techniques are found out to be a convenient
solution23 Solvothermal synthesis generally allows control over size and shape distribution Using
post-synthetic methods further changes on cavity sizes and chemical affinities can be made
Materials that are stable with open pores after removal of guest molecules are termed as open-
frameworks The stability of guest-free materials is usually examined by Powder X-ray diffraction
(PXRD) analysis Thermogravimetric analysis may be performed to inspect the thermal stability of the
material Elemental analysis can detail the elemental composition of the material Physical
techniques such as Fourier transform infrared (FTIR) spectra and nuclear magnetic resonance (NMR)
may be used to verify the condensation of monomers to the desired topology Porosity can be
evidenced from adsorption isotherms of gases or mercury porosimetry
7
The number of linkers that are allowed to bind to the metal-center and the orientations of the linkers
depend exclusively on the coordination preferences of the metal The organic linkers are typically
ditopic or polytopic They are essential in determining the topology and providing porosity Longer
linkers provide larger pore size A series of compounds with the same underlying topology and
different linker sizes are called iso-reticular series The structure of MOFs in principle can be formed
into the requirement of prominent applications such as gas storage gas separation sensing and
catalysis The structural divergence and performance can be further increased by functionalizing the
organic linkers Hence several attempts are on-going in purpose to come up with the best material
possible in each application
Interpenetration of frameworks is an inevitable outcome of large pore volume Interpenetrated nets
are periodic equivalents of mechanically interlocked molecules rotaxanes and catenanes Depending
on topology they are either maximally separated termed as interpenetration or minimally separated
termed as interweaving (see Figure 5) Interpenetration can be beneficial to some porous structures
protecting from collapse upon removal of solvents
Figure 5 Schematic diagram of interpenetrated and interwoven frameworks
Controlled growth of MOFs on organic surfaces is achieved by R A Fischer and C Woumlll24 The then
named SURMOFs allow to fabricate structures possibly not achievable by bulk synthesis The growth
is achieved through liquid phase epitaxy (LPE) of MOF reactants on the surface (nucleation site) A
step-by-step approach which is alternate immersion of the substrate in metal and ligand precursors
supplies control of the growth mechanism
8
Figure 6 Schematic diagram of SURMOF
24 Covalently-bound Organic Frameworks
As a result of the long-term struggle for complete covalent crystallization in the year 2005 Cocircte et
al21 synthesized porous crystalline extended 2D Covalent-Organic Frameworks (COFs) using
reticular concepts The success was followed by the design and synthesis of 3D COFs in the year
200722 By now there are about 50 COFs reported in the literature COFs are made entirely from
light elements and the building blocks are held together by strong covalent bonds Most of them
were formed by solvothermal condensation of boronic acids with hydroxyphenyl compounds
Tetragonal building blocks were used for the formation of 3D COFs Alternate synthetic methods
were also used for producing COFs COFs are generally studied for gas storage applications However
they have also shown potentialities in photonic and catalytic applications
Alternative synthesis methods paved the way to new covalently bound organic frameworks
Trimerization of polytriazine monomers in ionothermal conditions gives rise to Covalent Triazine
Frameworks (CTFs)25 Similarly coupling reactions of halogenated monomers produced Porous
Aromatic Frameworks (PAFs) Albeit the short-range order one of the PAFs showed very high surface
area (5600 m2 g-1) and gas uptake capacity26
Due to low weight the covalently-bound materials show very high gravimetric capacities
Suggestions such as metal-doping functionalization and geometry modifications can be found in the
literature for the general improvement of the functionalities There are also various studies of their
structure and properties
A review on the synthesis structure and applications of covalently bound organic frameworks has
been prepared for publication
Article 1 Covalently-bound organic frameworks
Binit Lukose Thomas Heine
9
See Appendix A for the article
My contributions include collecting data and preparing a preliminary manuscript
Figure 7 SBUs and topologies of 2D COFs
10
3 Methodology and Validation
31 Methods and Codes
The Density-Functional based Tight Binding (DFTB) method2728 is an approximate Kohn-Sham DFT29-31
scheme It avoids any empirical parameterization by calculating the Hamiltonian and overlap matrix
elements from DFT-derived local orbitals (atomic orbitals) and atomic potentials The Kohn-Sham
orbitals are represented with the linear combination of atomic orbitals (LCAO) scheme The matrix
elements of Hamiltonian are restricted to include only one- and two-center contributions Therefore
they can be calculated and tabulated in advance as functions of the distance between atomic pairs
The remaining contributions to the total energy that is the nucleus-nucleus repulsion and the
electronic double counting terms are grouped in the so-called repulsive potential This two-center
potential is fitted to results of DFT calculations of properly chosen reference systems The accuracy
and transferability can be improved with the self-consistent charge (SCC) extension to DFTB32 This
method is based on the second-order expansion of the Kohn-Sham total energy with respect to
charge density fluctuations which are estimated by Mulliken charge analysis In order to account for
London dispersion empirical dispersion energy can be added to the total energy3334 Various reviews
are available on DFTB notably the recent ones by Oliveira et al35 and by Seifert et al36
DFTB is implemented in a large number of computer codes For this work we employed the codes
deMonNano37 and DFTB+38 as they allow DFTB calculations on periodic and finite structures
Typically dispersion corrected (DC) SCC-DFTB calculations were carried out Periodic boundary
conditions were used to represent the crystalline frameworks and as the unit cells are large the
standard approach used the point approximation Electronic density of states (DOS) have been
calculated using the DFTB+ code using k-point sampling where the k space was determined by
reaching convergence for the total energy according to the scheme proposed by Monkhorst and
Pack39
For DFT calculations of finite and periodic structures ADF40 Turbomole41 and Crystal0942 were used
For studies of finite models of COFs the calculations were performed at PBEDZP level However for
extensive calculations on SURMOF-2 models we used B3LYP-D The copper atoms were described
using the basis set associated with the Hay-Wadt43 relativistic effective core potentials44 which
include polarization f functions45 Carbon oxygen and hydrogen atoms were described using the
Pople basis set 6-311G
Details of the individual calculations are given in the individual articles in the appendix of this thesis
11
32 DFTB Validation
Figure 8 Deconstructed building units their schematic diagrams and final geometries of HKUST-1
(Cu-BTC) MOF-5 MOF-177 MOF-205 and MIL-53
12
In the literature MOFs and COFs are largely studied for applications such as gas storage using
classical force field methods46-48 First principles based studies of several hundreds of atoms are
computationally expensive Hence they are generally limited to cluster models of the periodic
structures Contrarily DFTB paves the way to model periodic structures involving large numbers of
atoms Being an approximate method DFTB holds less accuracy hence comparisons to experimental
data or higher level methods should be performed for validation
As MOFs contain metal centers andor metal oxides as well as organic elements the DFTB
parameters for both heavy and light elements as well as their mixtures are required Thus we have
chosen MOFs such as Cu-BTC (HKUST-1) MOF-5 MOF-177 MOF-205 and MIL-53 as our model
structures which contain Cu Zn and Al atoms apart from C O and H The MOFs comprising three
common connector units copper paddlewheels (HKUST-1) zinc oxide Zn4O tetrahedron (MOF-5
MOF-177 DUT-6 (MOF-205)) and aluminum oxide AlO4(OH)2 octahedron (MIL-53) Figure 8 shows
the schematic diagram of the MOFs
The validation calculations have been published
Article 2 Structural properties of metal-organic frameworks within the density-functional based
tight-binding method
Binit Lukose Barbara Supronowicz Petko St Petkov Johannes Frenzel Agnieszka B Kuc Gotthard
Seifert Georgi N Vayssilov and Thomas Heine Phys Status Solidi B 2012 249 335ndash342 DOI
101002pssb201100634
See Appendix B for the article
In this article DFTB has been validated against full hybrid density-functional calculations for model
clusters against gradient corrected density-functional calculations for supercells and against
experiment Moreover the modular concept of MOF chemistry has been discussed on the basis of
their electronic properties Our results show that DC-SCC-DFTB predicts structural parameters with a
good accuracy (with less than 5 deviation even for adsorbed CO and H2O on HKUST-1) while
adsorption energies differ by 12 kJ mol-1 or less for CO and water compared to DFT benchmark
calculations
My contributions to this article include performing DFTB based calculations on the MOFs (HKUST-1
MOF-5 MOF-177 and MOF-205) that is optimizing the geometries simulating powder X-ray
diffraction patterns and calculating density of states and bulk modulus Additional involvement is in
comparing structural parameters such as bond lengths bond angles dihedral angles and bulk
modulus with experimental data or data derived from DFT calculations and preparing the manuscript
13
4 2D Covalent Organic Frameworks
41 Stacking
Condensation of planar boronic acids and hydroxyphenyl compounds leads to the formation of two-
dimensional covalent organic frameworks (2D COFs) The layers are held together by London
dispersion interactions The first synthesized COFs COF-1 and COF-5 were reported to have AB
(P63mmc staggered as in graphite) and AA (P6mmm eclipsed as in boron nitride) stackings
respectively (see Figure 9)21 The synthesis of several further 2D COFs then followed49-51 all of them
were inspected for only two stacking possibilities ndash P6mmm or P63mmc and it was reported that
they aggregate in P6mmm symmetry As framework materials possess framework charges the
interlayer interactions significantly include Coulomb interactions P6mmm symmetry is the face-to-
face arrangement where the overlap of the stacked structures is maximized (maximization of the
London dispersion energy) however atom types of alike charges are facing each other in the closest
possible way With the interlayer separation similar to that in graphite or boron nitride the Coulomb
repulsion should be high in such arrangements One should notice that in the example case of boron
nitride the facing atom types are different We therefore assumed that a stable stacking should
possess layer-offset
Figure 9 AA and AB layer stacks of hexagonal layers
We considered two symmetric directions for layer shift and studied their total energies (see Figure
10) The stable geometries are found to have a layer-shift of ~14 Aring a value corresponding to the
shift of AA to AB stacking in graphite and compatible with the bond length between two 2nd row
atoms This stability-supported stacking arrangement as revealed from our calculations was
14
supported by good agreement between simulated and experimental PXRD patterns Hence
independent of the elementary building blocks any 2D COF should expose a layer-offset Based on
the direction of the shift we proposed two stacking kinds serrated and inclined (Figure 10) In the
former the layer-offset is back and forth while in the latter the layer-offset followed single direction
As serrated and inclined stackings have no significant change in stacking energy our calculations
cannot predict the long-range stacking in the crystal However this problem is known from other
layered compounds and the ultimate answer is yet to be revealed by analyzing high-quality
crystalline phases at low temperature
Figure 10 Stacking energy Es vs shift of adjacent layers for COF-5 Different stacking possibilities
and their energies are also shown
We published our analysis of the stacking in 2D COFs
Article 3 The Structure of Layered Covalent-Organic Frameworks
Binit Lukose Agnieszka Kuc Thomas Heine Chem Eur J 2011 17 2388 ndash 2392 DOI
101002chem201001290
See Appendix C for the article
15
My contributions to this article include performing the shift calculations simulating XRDs and partly
preparing the manuscript The article has made certain impact in the literature Most of the 2D COFs
synthesized afterwards were inspected for their stacking stability The instability of AA stacking was
also suggested by the studies of elastic properties of COF-1 and COF-5 by Zhou et al52 The shear
modulus shows negative signs for the vertical alignment of COF layers while they are small but
positive for the offset of layers Detailed analysis performed by Dichtel53 et al using DFT was
confirming the instability of AA stacking and suggesting shifts of about 13 to 25 Aring
42 Concept of Reticular Chemistry
Reticular chemistry means that (functional) molecules can be stitched together to form regular
networks The structural integrity of these molecules we also speak of building blocks remains in the
crystal lattices Consequently also the electronic structure and hence the functionality of these
molecules should remain similar
2D COFs are formed by condensation of well-defined planar building blocks With the usage of linear
and triangular building blocks hexagonal networks are expected The properties of each COF may
differ due to its unique constituents However the extent of the relationship of the properties of
building blocks in and outside the framework has not been studied in the literature
Reticular chemistry allows the design of framework materials with pre-knowledge of starting
compounds and reaction mechanisms Reverse of this is finding reactants for a certain topology We
intended to propose some building units suitable to form layered structures (see Figure 11) The
building units obey the regulations of reticular chemistry and offer a variety of structural and
electronic parameters
Our strategic studies on a set of designed COFs have been published
Article 4 On the reticular construction concept of covalent organic frameworks
Binit Lukose Agnieszka Kuc Johannes Frenzel Thomas Heine Beilstein J Nanotechnol 2010 1
60ndash70 DOI103762bjnano18
See Appendix D for the article
16
Figure 11 Schematic diagram of different building units forming 2D COFs
Various hexagonal 2D COFs with different building blocks have been designed and investigated
Stability calculations indicated that all materials have the layer offset as reported in our earlier
work54 (see Appendix C) We show that the concept of reticular chemistry works as the Density-Of-
States (DOS) of the framework materials vary with the the DOS of the molecules involved in the
frameworks However the stacking does have some influence on the band gap
My contributions to this article include performing all the calculations and preparing the manuscript
17
5 3D Frameworks
51 3D Covalent Organic Frameworks
First 3D COFs were reported in 2007 by the group of Yaghi22 Although the number of 3D COFs
synthesized so far has not been crossed half a dozen they are of particular interest for their very low
mass density Similar to 2D COFs condensation of boronic acids and hydroxyphenyl compounds led
to the formation of COF-102 103 105 108 and 202 Usage of tetragonal boronic acids leads to the
formation of 3D networks (Figure 12) All of them possessed ctn topology while COF-108 which has
the same material composition as COF-105 crystallized in bor topology COF-300 which was formed
from tetragonal and linear building units possessed diamond topology and was five-fold
interpenetrated55 COF-108 has the lowest mass density of all materials known today Unlike most of
the MOFs the connectors in COFs are not metal oxide clusters but covalently bound molecular
moieties In the synthesized COFs the centers of the tetragonal blocks were either sp3 carbon or
silicon atoms
Schmid et al56 have analyzed the two different topologies and developed force field parameters for
COFs The mechanical stability of COFs was also reported However no further study that details the
inherent energetic stability and properties of COFs was found in the literature Using DFTB we
performed a collective study of all 3D COFs in their known topologies with C and Si centers
Figure 12 Schematic diagram of tetragonal and triangular building blocks forming lsquoctnrsquo or lsquoborrsquo
topologies
Our studies of3D COFs have been prepared for publication
Article 5 Stability and electronic properties of 3D covalent organic frameworks
Binit Lukose Agnieszka Kuc Thomas Heine
18
See Appendix E for the article
My contributions to this article include performing all the calculations and preparing the manuscript
We discussed the energetic and mechanical stability as well as the electronic properties of COFs in
the article All studied 3D COFs are energetically stable materials with formation energies of 76 ndash
403 kJ mol-1 The bulk modulus ranges from 29 to 206 GPa Electronically all COFs are
semiconductors with band gaps vary with the number of sp2 carbon atoms present in the linkers
similar to 3D MOFs
52 Porous Aromatic Frameworks
Porous Aromatic Frameworks (PAFs) are purely organic structures where linkers are connected by sp3
carbon atoms (see Appendix A) Tetragonal units were coupled together to form diamond-like
networks The synthesis follows typically a Yamamoto-type Ullmann cross-coupling reaction those
reactions are known to be much simpler to be carried out than the condensation reactions necessary
to form COFs However as the coupling reactions are irreversible a lower degree of crystallinity is
achieved and the materials formed were amorphous The first PAF was reported in 2009 and
showed a very high BET surface area of 5600 m2g-1 The geometry of PAF-1 was equal to diamond
with the C-C bonds replaced by biphenyl Subsequent PAFs may be synthesized with longer linkers
between carbon tetragonal nodes Nevertheless synthesis of a PAF with quaterphenyl linker
provided an amorphous material of very low surface area due to the short range order
Synthetic complexities aside the diamond-like PAFs are interesting for their special geometries from
the viewpoint of the theorist It is interesting to see to what extent they follow the properties of
diamond Additionally the large surface area and high polarizability of aromatic rings are auxiliary for
enhanced hydrogen storage Thus we have explored the possibilities of diamond topology by
inserting various organic linkers in place of C-C bonds (Figure 13)
Figure 13 Diamond structure and various organic linkers to build up PAFs
Our studies of PAFs have been prepared for publication
19
Article 6 Structure electronic structure and hydrogen adsorption capacity of porous aromatic
frameworks
Binit Lukose Mohammad Wahiduzzaman Agnieszka Kuc Thomas Heine
See Appendix F for the article
In this article we have discussed the correlations of properties with the structures Exothermic
formation energies of PAFs calculated from the dehydrogenation of the linkers with CH4 indicate the
strong coordination of linkers in the diamond topology The studied PAFs exhibited bulk moduli much
smaller than diamond however in line with related MOFs and COFs The PAFs are semi-conductors
with their band gaps decrease with the increasing number of benzene rings in the linkers
Using Quantized Liquid Density Functional Theory (QLDFT) for hydrogen57 a method to compute
hydrogen adsorption that takes into account inter-particle interactions and quantum effects we
predicted a large choice of hydrogen uptake capacities in PAFs At room temperature and 100 bar
the predicted uptake in PAF-qtph was 732 wt which exceeds the 2015 DOE (US) target We
further discussed the structural impacts on the adsorption capacities
My contributions to this article include designing the materials performing calculations of stability
and electronic properties describing the adsorption capacities and preparing the manuscript
20
6 New Building Concepts
61 Isoreticular Series of SURMOFs
The growth of MOFs on substrates using liquid phase epitaxy (LPE) opened up a controlled way to
construct frameworks This is achieved by alternate immersion of the substrates in metal and ligand
precursors for several cycles Such MOFs named as SURMOFs can avoid interpenetration because
the degeneracy is lifted58 and are suited for conventional applications This is an advantage as
solvothermally synthesized MOFs with larger pore size suffer from interpenetration and the large
pores are hence not accessible for guest species
MOF-259 is a simple 2D architecture based on paddlewheel units formed by attaching four
dicarboxylic groups to Cu2+ or Zn2+ dimers yielding planar sheets with 4-fold symmetry The
arrangement of MOF layers possessed P2 or C2 symmetry Recently the group of C Woumlll at KIT has
synthesized an isoreticular series of SURMOFs derived from MOF-2 which has a homogeneous series
of linkers with different lengths (Figure 14) XRD comparisons suggest that these MOFs possess P4
symmetry The cell dimensions that correspond to the length of the pore walls range from 1 to 28
nm The diagonal pore-width of one of the SURMOFs (PPDC) accounts to 4 nm The symmetry of
SURMOFs as determined to be different from bulk MOF-2 should be understood by means of theory
As collaborators we simulated the structures and inspected each stacking corresponding to the
symmetries in order to understand the difference
Figure 14 The building units and schematic diagram of the formation of iso-reticular SURMOF
series
21
This collaborated work has been submitted for publication
Article 7 A novel series of isoreticular metal organic frameworks realizing metastable structures
by liquid phase epitaxy
Jinxuan Liu Binit Lukose Osama Shekhah Hasan Kemal Arslan Peter Weidler Hartmut Gliemann
Stefan Braumlse Sylvain Grosjean Adelheid Godt Xinliang Feng Klaus Muumlllen Ioan-Bogdan Magdau
Thomas Heine Christof Woumlll
See Appendix G for the article
The main contribution of this article was the experimental proof backed up by our computer
simulations that SURMOFs with exceptionally large pore sizes of more than 4 nm can be prepared in
the laboratory and they offer the possibility to host large materials such as clusters nanoparticles or
small proteins The most important contribution of theory was to show that while MOF-2 in P2
symmetry shows the highest stability the P4 symmetry as obtained using LPE for SURMOF-2
corresponds to a local minimum
My contribution to this article includes performing and analyzing the calculations Our theoretical
study went significantly beyond and will be published as separate article (Appendix H)
62 Metastability of SURMOFs
Following the difference in stability of SURMOFs-2 in comparison with MOF-2 we identified the role
of linkers as a constituent of the framework that provides the metastability to SURMOFs-2 (Figure
15) In MOFs linkers are essential in determining the topology as well as providing porosity Linkers
typically contain single or multiple aromatic rings and are ditopic and polytopic The orientation of
them normally undergoes lowest repulsion from the coordinated metal-oxide clusters and provides
high stability In the case of 2D MOFs non-covalent interlayer interactions lead to the most stable
arrangements Paddlewheels are identified as the reasons for the stability of bulk MOF-2 as they
form metal-oxide bridges across the MOF layers In the case of SURMOFs-2 the linkers are rotated in
a tightly-packed environment of layers and each linker in bulk MOF-2 is rotated in such a way that
any attempt of shifting is obstructed by repulsion between the neighboring linkers The energy
barrier for leaving P4 increases with the length of organic linkers Moreover SURMOF-2 derivatives
with extremely large linkers are energetically stable due to the increased London dispersion
interaction between the layers in formula units Thus we encountered a rare situation in which the
linkers guarantee the persistence of a series of materials in an otherwise unachievable state
22
Figure 15 Energy diagram of the metastable P4 and stable P2 structures
Our results on the linker guided stability of SUMORs-2 have been prepared for publication
Article 8 Linker guided metastability in templated Metal-Organic Framework-2 derivatives
(SURMOFs-2)
Binit Lukose Gabriel Merino Christof Woumlll Thomas Heine
See Appendix H for the article
This article is based solely on my scientific contributions
23
7 Summary
Nanotechnology is the way of ingeniously controlling the building of small and large structures with
intricate properties it is the way of the future a way of precise controlled building with incidentally
environmental benignness built in by design - Roald Hoffmann Chemistry Nobel Laureate 1981
Currently it is possible to design new materials rather than discovering them by serendipity The
design and control of materials at the nanoscale requires precise understanding of the molecular
interactions processes and phenomena In the next level the characteristics and functionalities of
the materials which are inherent to the material composition and structure need to be studied The
understanding of the materials properties may be put into the design of new materials
Computational tools to a large extend provide insights into the structures and properties of the
materials They also help to convert primary insights into new designs and carry out stability analysis
Our theoretical studies of a variety of materials have provided some insights on their underlying
structures and properties The primary differences in the material compositions and skeletons
attributed a certain choice in properties The contents of the articles discussed in the thesis may be
summarized into the following three parts
71 Validation of Methods
Simulations of nanoporous materials typically include electronic structure calculations that describe
and predict properties of the materials Density functional theory (DFT)30 is the standard and widely-
used tool for the investigation of the electronic structure of solids and molecules Even the optical
properties can be studied through the time-dependent generalization of DFT MOFs and COFs have
several hundreds of atoms in their unit cells DFT becomes inappropriate for such large periodic
systems because of its necessity of immense computational time and power Molecular force field
calculations60 on the other hand lack transferable parameterization especially for transition metal
sites and are hence of limited use to cover the large number of materials to be studied Apparently
a non-orthogonal tight-binding approximation to DFT called density functional tight-binding
(DFTB)28323561 has proven to be computationally feasible and sufficiently accurate This method
computes parameters from DFT calculations of a few molecules per pair of atom types The
parameters are generally transferable to other molecules or solids With self-consistent charge (SCC)
extension DFTB has improved accuracy In order to account weak forces the London dispersion
energy can be calculated separately using empirical potentials and added to total energy Successful
realizations of DFTB include the studies of large-scale systems such as biomolecules62
24
supramolecular compounds63 surfaces64 solids65 liquids and alloys66 Being an approximate method
DFTB needs validation Often one compares DFTB results of selected reference systems with those
obtained with DFT
Before electronic structure calculations of framework materials can be carried out it is necessary to
compute the equilibrium configurations of the atoms Geometry optimization (or energy
minimization) employs a mathematical procedure of optimization to move atoms so as to reduce the
net forces on them to negligible values We adopted the conjugate gradient scheme for the
optimizations using DFTB A primary test for the validation of these optimizations is the comparison
of cell parameters bond lengths bond angles and dihedral angles with the corresponding known
numbers We compared the DFTB optimized MOF COF and PAF geometries with the experimentally
determined or DFT optimized geometries and found that the values agree within 6 error
The angles and intensities of X-ray diffraction patterns can produce three-dimensional information of
the density of electrons within a crystal This can provide a complete picture of atomic positions
chemical bonds and disorders if any Hence simulating powder X-ray diffraction (PXRD) patterns of
optimized geometries and comparing them with experimental patterns minimize errors in the crystal
model Our focus on this matter directed us to identify the layer-offsets of 2D COFs for the first time
In the case of 3D COFs excellent correlations were generally observed between experimental and
simulated patterns Slight differences in the intensities of some of them were due to the presence of
solvents in the crystals as they were reported in the experimental articles PAFs as experimentally
being amorphous do not possess XRD comparisons MOFs within DFTB optimization have
undergone some changes especially in the dihedral angles in comparison with experimental
suggestion or DFT optimization This was verified from the differences in the simulated PXRD
patterns Another interesting outcome of XRD comparison was the discovery of P4 symmetry of
templated MOF-2 derivatives (SURMOFs-2) made by Woumlll et al
Bulk moduli which may be expressed as the second derivative of energy with respect to the unit cell
volume can give a sense of mechanical stability Our calculations provide the following bulk moduli
for some materials MOF-5 1534 (1537)67 Cu-BTC 3466 (3517)68 COF-102 206 (170)69 COF-
103 139 (118)69 COF-105 79 (57)69 COF-108 37 (29)69 COF-202 153 (110)69 The data in the
parenthesis give corresponding values found in the literature calculated using force-field methods
The bulk moduli of MOFs are comparable with the results in the literature however COFs show
significant differences Albeit the differences in values each type of calculation shows the trend that
bulk modulus decreases with decreasing mas density increasing pore volume decreasing thickness
of pore walls and increasing distance between connection nodes
25
Formation of framework materials from condensation of reactants may be energetically modeled
COFs are formed from dehydration reactions of boronic acids and hydroxyphenyl compounds The
formation energy calculated from the energies of the products and reactants can indicate energetic
stability Both DFTB for periodic structures and DFT for finite structures predict exothermic formation
of 3D COFs Likewise for 2D COFs such as COF-1 and 5 the formation of monolayers is found to be
endothermic within both the periodic model calculation using DFTB and finite model calculation
using DFT The stacking of layers provides them stability
72 Weak Interactions in 2D Materials
AB stacked natural graphite and ABC stacked rhombohedral graphite are found to be in proportions
of 85 and 15 in nature respectively whereas AA stacking has been observed only in graphite
intercalation compounds On the other hand boron nitride (h-BN) the synthetic product of boric
acid and boron trioxide exists with AA stacking 2D COFs were believed to have high symmetric AA
(eclipsed ndash P6mmm symmetry) stacking as boron nitride (h-BN) An exception was COF-1 which was
considered to have AB (staggered ndash P63mmc) stacking as graphite The AA stacking maximizes the
attractive London dispersion interaction between the layers a dominating term of the stacking
energy At the same time AA stacking always suffers repulsive Coulomb force between the layers
due to the polarized connectors It should be noted that in boron nitride oppositely charged boron
atoms and nitrogen atoms are facing each other The Coulomb interaction has ruled out possible
interlayer eclipse between atoms of alike charges This gave us the notion that 2D COFs cannot
possess layer-eclipsing Based on these facts we have suggested a shift of ~14 Aring between adjacent
layers at which one or two of the edge atoms of the hexagonal rings are situated perpendicular to
the center of adjacent rings In other words some or all of the hexagonal rings in the pore walls
undergo staggering with that of adjacent layers These lattice types were found to be very stable
compared to AA or AB analogues AA stacking was found to have the highest Coulombic repulsion in
each COF Within an error limit the bulk COFs may be either serrated or inclined The interlayer
separations of all the COF stacks are in the range of 30 to 36 Aring and increase in the order AB
serratedinclined AA70 The motivation for proposing inclined stacking for COFs is that the
rhombohedral graphite (ABC stacking) is the inclined version of the layer-offset in natural graphite
(AB stacking) The instability of AA stacking was also suggested by the studies of elastic properties of
COF-1 and COF-5 by Zhou et al52 The shear modulus show negative signs for the vertical alignment of
COF layers while they are small but positive for the offset of layers
The change of stacking should be visible in their PXRD patterns because each space group has a
distinct set of symmetry imposed reflection conditions We simulated the XRD patterns of COFs in
their known and new configurations and on comparison with the experimental spectrum the new as
26
well as AA stacking showed good agreement5470 The slight layer-offsets are only able to make a few
additional peaks in the vicinity of existing peaks and some changes in relative intensities The
relatively broad experimental data covers nearly all the peaks of the simulated spectrum In other
words the broad experimental peaks are explainable with layer-offset
A detailed analysis on the interlayer stacking of HHTP-DPB COF made by Dichtel et al53 is very
complementary53 This work uses molecular mechanics extensively to define potential energy
surfaces of stacking of two layers from eclipsed to staggered covering all possible layer offsets Low
energy region was near to the eclipsed stacking which was then zoomed in to examine using DFT for
higher accuracy The far-eclipsed regions encounter no energy barriers with the near-eclipsed regions
which suggest the unlikeness of forming near-staggered layering Within DFT the eclipsed
geometries are at high energy compared to the slight offsets around AA (~ 13 ndash 25 Aring away) For a
hexagonal COF the energetically preferred offset could be in one of the six hexagonal sectors (or
circle sectors) of central angle 60 The preference of falling into one sector over the others is
arbitrary and may be determined by symmetry set by nature or external parameters Hence the
layering order in bulk could be serrated inclined spiral or purely by random The latter case better
known as turbostratic disorder is then more like a rule for 2D COFs rather than an exception caused
by external forces This also explains why the experimental PXRD patters have relatively broad peaks
The layer-offset not only change the internal pore structure but also affect interlayer exciton and
vertical charge transport in opto-electronic applications
About stacking stability the square COFs are expected not to be different from hexagonal COFs
because the local environment causing the shifts is nearly the same The DFTB based calculations
reported along with the synthesis of electron-transporting 2D-NiPc-BTDA COF supports this notion71
Two shifted configurations ndash at 07 and 14 Aring away from AA ndash appeared to be energetically preferred
over the AA stacking It appears that AA stacking is only possible for boron nitride-like structures
were adjacent layers have atoms with opposite charges in vertical direction
SURMOFs-2 a series of paddlewheel-based 2D MOFs possess a different symmetry than
solvothermally synthesized MOF-2 due to the differences in interlayer interactions Typically the
interaction between the paddlewheels favors P2 or C2 symmetry of bulk MOF-2 rather than the P4
symmetry of SURMOFs-2 Bader charge analysis reveals that electron distribution between the Cu-
paddlewheels is the lowest when they follow P4 symmetry This indicates the smallest possibility of
having a direct Cu-Cu bond between the paddlewheels In monolayer organic linkers undergo no
rotation with respect to metal dimers
27
X-ray diffraction of SURMOFs-2 made by Woumlll et al indicated P4 symmetry and a relatively small
interlayer separation This increases the repulsion between the linkers and enforces them to rotate
The rotations of each linker in a layer are controlled globally The rotated linkers in adjacent layers
increase London dispersion however a paddlewheel-led shift towards any side increases repulsion
thereby locking SURMOF-2 in P4 packing Hence with the special arrangement of rotated linkers the
linker-linker interaction overcomes the paddlewheel-paddlewheel interaction
P4 symmetry opens up the pores more clearly than P2 or C2 Additionally the metal sites that
typically host solvents are inaccessible Furthermore improvements such as functionalizing the linker
may be easily carried out
Based on our calculations on 2D materials we emphasize the role of van der Waals interactions in
determining the layer-to-layer arrangements The promise of reticular chemistry which is the
maintainability of structural integrity of the building blocks in the construction process is partly
broken in the case of SURMOFs-2 This is because due to repulsion the linkers undergo rotation with
respect to the carboxylic parts albeit keeping the topology
73 Structure-Property Relationships
We have studied a variety of materials from the classes of MOFs COFs and PAFs The structural
differences arise from the differences in the constituents andor their arrangements Properties in
general are interlinked with structural specifications Therefore it is beneficial to know the
relationship between the structural parameters and properties
The mass density is an intensive property of a material In the area of nanoporous materials a crystal
with low mass density has advantages in applications involving transport Definitely the mass density
decreases with increasing pore volume Still the number of atoms in the wall and their weights are
important factors The pore size does not relate directly to the atom counts The volume per atom
(inverse of atom density) another intensive property of a material obliquely gives porosity Figure
16 shows the variation of mass density with volume per atom (including the volume of the atom) for
MOFs COFs and PAFs MOFs show higher mass density compared to other materials of identical
atom density implying the presence of heavy elements It is to be noted that Cu-BTC has mass
density 184 times higher than COF-102 The presence of metals in the form of oxide clusters in MOFs
increases the mass density and decreases the volume per atom Note that the low-weighted MOF in
the discussion (MOF-205) is heavier than many of the PAFs and COFs The relatively high mass
density and small volume per atom of COF-300 is due to its 5-fold interpenetration whereas COF-202
has additional tert-butyl groups which do not contribute to the system shape but affect the mass
density and the volume per atom COF-102 and 103 have same topology but different centers (C and
28
Si respectively) The Si centers increases the cell volume and decreases mass density COF-105 (Si
centers) and COF-108 (C centers) additionally differ in their topologies (ctn and bor respectively) It
appears that bor topology expands the material compared to ctn PAFs hold the extreme cases PAF-
phnl has the highest atom and mass densities whereas PAF-qtph has the smallest atom and mass
densities
Figure 16 Mass density as a function of volume per atom for MOFs COFs and PAFs
The bulk modulus indicates the materialsrsquo resistance towards compression which should in principle
decrease with increasing porosity At the same time larger number of atoms making covalent
networks in unit volume should supply larger bulk moduli Thus differences in molecular contents
and architectures give rise to different bulk moduli It is interesting to see how the mechanical
stability of nanoporous materials is related with the atom density We have obtained a correlation
between bulk modulus (B) and volume per atom (VA) in the case of the studied MOFs COFs and PAFs
as follows
29
where k is a constant (see Figure 17) The inverse-volume correlation indicates that the materials
close to the fitting curve have average bond strengths (interaction energy between close atoms)
identical to each other independent of number of bonds bond order and branching Only Cu-BTC
COF-202 and COF-300 show significant differences from the fitted curve Cu-BTC has 17 times larger
bulk modulus compared to COF-102 of similar volume per atom which implies the substantially
higher strength of the bond network resulting from paddlewheel units and tbo topology
Interpenetration decreased the volume per atom however increased bulk modulus through
interlayer interactions in the case of COF-300 The tert-butyl groups in COF-202 do not transmit its
inherent stability to the COF significantly however decreases the volume per atom Comparison
between COF-102 (C centers) and COF-103 (Si centers) discloses that Si centers decrease the
mechanical stability Comparison between COF-105 (ctn) and COF-108 (bor) suggests that ctn
topology possess higher stability This indicates that local angular preferences can amend the
strength of the bulk material
Figure 17 Bulk modulus as a function of volume per atom for MOFs COFs and PAFs
Band gaps of all the studied materials show that they are semiconductors except of Cu-BTC which
has unpaired electrons at the Cu centers Our studies of the density of states (DOS) of bulk MOFs and
the cluster models that have finite numbers of connectors and linkers show that electronic structure
30
stays unchanged when going from cluster to periodic crystal In the case of 2D COFs the DOS of
monolayers and bulk materials were identical Interpenetrated COF-300 showed no difference in the
electronic structure in comparison with the non-interpenetrated structure Based on these results
we may reach into a premature conclusion that electronic structure of a solid is determined by its
constituent bonded network sufficiently large to include all its building units
HOMO-LUMO gap of the building units determine the band gap of a framework material We have
observed that PAFs nearly reproduced the HOMO-LUMO gaps of the linkers 3D COFs that are made
of more than one building unit show that the band gap is slightly smaller than the smallest of the
HOMO-LUMO gaps of the building units For example
TBPM (43 eV) + HHTP (34 eV) COF-105COF-108 (32 eV)
TBPM (43 eV) + TBST (139 eV) COF-202 (42 eV)
TAM (41 eV) + TA (26 eV) COF-300 (23 eV)
The compound names are taken from appendix E Additionally the band gaps decrease with
increasing number of conjugated rings similar to HOMO-LUMO gaps of the linkers
I believe that the studies in the thesis have helped to an extent to understand the structure
stability and properties of different classes of framework materials The benchmark structures we
studied have the essential features of the classes they represent Ab-initio based computational
studies of several periodic structures are exceptional and thus have its place in the literature
31
List of Abbreviations
ADF Amsterdam Density Functional code
BLYP Becke-Lee-Yang-Parr functional
B3LYP Becke 3-parameter Lee Yang and Parr functional
COF Covalent-Organic Framework
CP Coordination Polymer
CTF Covalent-Triazine Framework
DC Dispersion correction
DFT Density Functional Theory
DFTB Density Functional Tight-Binding
DOS Density of States
DOE (US) Department of Energy (United States)
DZP Double-Zeta Polarized basis set
GGA Generalized Gradient Approximation
LCAO Linear Combination of Atomic Orbitals
LPE Liquid Phase Epitaxy
MOF Metal-Organic Framework
PAF Porous Aromatic Framework
PBE Perdew-Burke-Ernzerhof functional
PXRD Powder X-ray Diffraction Pattern
QLDFT Quantized Liquid Density Functional Theory
RCSR Reticular Chemistry Structure Resource
SBU Secondary Building Unit
SCC Self-Consistent Charge
TZP Triple-Zeta Polarized basis set
SURMOF Surface-Metal-Organic Framework
32
List of Figures
Figure 1 ldquoAdamantinerdquo unit of diamond structure and diamond (dia) net 3
Figure 2 CU-BTC MOF and the corresponding tbo net 4
Figure 3 Fundamental building blocks and SBUs of Cu-BTC and extension of SBUs following crystal
structure 5
Figure 4 Schematic diagram of the single pore of MOF and An example of a simple cubic MOF in
which zinc-oxide tetrahedron are linked together by benzene linkers Atom colors blue ndash Zn red ndash O
grey ndash C white ndash H 6
Figure 5 Schematic diagram of interpenetrated and interwoven frameworks 7
Figure 6 Schematic diagram of SURMOF 8
Figure 7 SBUs and topologies of 2D COFs 9
Figure 8 Deconstructed building units their schematic representations and final geometries of
HKUST-1 (Cu-BTC) MOF-5 MOF-177 MOF-205 and MIL-53 11
Figure 9 AA and AB layer stacks of hexagonal layers 13
Figure 10 Stacking energy Es vs shift of adjacent layers for COF-5 Different stacking possibilities and
their energies are also shown 14
Figure 11 Schematic diagram of different building units forming 2D COFs 16
Figure 12 Schematic diagram of tetragonal and triangular building blocks forming lsquoctnrsquo or lsquoborrsquo
topologies 17
Figure 13 Diamond structure and various organic linkers to build up PAFs 18
Figure 14 The building units and schematic diagram of the formation of iso-reticular SURMOF series
20
Figure 15 Energy diagram of the metastable P4 and stable P2 structures 22
Figure 16 Mass density as a function of volume per atom for MOFs COFs and PAFs 28
Figure 17 Bulk modulus as a function of volume per atom for MOFs COFs and PAFs 29
33
References
(1) Morris R E Wheatley P S Angewandte Chemie-International Edition 2008 47 4966
(2) Li J-R Kuppler R J Zhou H-C Chemical Society Reviews 2009 38 1477
(3) Corma A Chemical Reviews 1997 97 2373
(4) Seo J S Whang D Lee H Jun S I Oh J Jeon Y J Kim K Nature 2000 404 982
(5) Lee J Kim J Hyeon T Advanced Materials 2006 18 2073
(6) Stein A Wang Z Fierke M A Advanced Materials 2009 21 265
(7) Velev O D Jede T A Lobo R F Lenhoff A M Nature 1997 389 447
(8) Beck J S Vartuli J C Roth W J Leonowicz M E Kresge C T Schmitt K D Chu C T
W Olson D H Sheppard E W McCullen S B Higgins J B Schlenker J L Journal of the
American Chemical Society 1992 114 10834
(9) Kresge C T Leonowicz M E Roth W J Vartuli J C Beck J S Nature 1992 359 710
(10) Ying J Y Mehnert C P Wong M S Angewandte Chemie-International Edition 1999 38
56
(11) Velev O D Kaler E W Advanced Materials 2000 12 531
(12) Stein A Microporous and Mesoporous Materials 2001 44 227
(13) Tanev P T Pinnavaia T J Science 1996 271 1267
(14) Yaghi O M OKeeffe M Ockwig N W Chae H K Eddaoudi M Kim J Nature 2003
423 705
(15) OKeeffe M Peskov M A Ramsden S J Yaghi O M Accounts of Chemical Research
2008 41 1782
(16) Delgado-Friedrichs O OKeeffe M Journal of Solid State Chemistry 2005 178 2480
(17) Delgado-Friedrichs O Foster M D OKeeffe M Proserpio D M Treacy M M J Yaghi
O M Journal of Solid State Chemistry 2005 178 2533
(18) OKeeffe M Yaghi O M Chemical Reviews 2012 112 675
(19) Tranchemontagne D J L Ni Z OKeeffe M Yaghi O M Angewandte Chemie-
International Edition 2008 47 5136
(20) Hayashi H Cote A P Furukawa H OKeeffe M Yaghi O M Nature Materials 2007 6
501
(21) Cote A P Benin A I Ockwig N W OKeeffe M Matzger A J Yaghi O M Science
2005 310 1166
(22) El-Kaderi H M Hunt J R Mendoza-Cortes J L Cote A P Taylor R E OKeeffe M
Yaghi O M Science 2007 316 268
(23) Rowsell J L C Yaghi O M Microporous and Mesoporous Materials 2004 73 3
(24) Hermes S Zacher D Baunemann A Woell C Fischer R A Chemistry of Materials
2007 19 2168
34
(25) Kuhn P Antonietti M Thomas A Angewandte Chemie-International Edition 2008 47
3450
(26) Ben T Ren H Ma S Cao D Lan J Jing X Wang W Xu J Deng F Simmons J M
Qiu S Zhu G Angewandte Chemie-International Edition 2009 48 9457
(27) Porezag D Frauenheim T Kohler T Seifert G Kaschner R Physical Review B 1995
51 12947
(28) Seifert G Porezag D Frauenheim T International Journal of Quantum Chemistry 1996
58 185
(29) Kohn W Sham L J Physical Review 1965 140 1133
(30) Parr R G Yang W Density-Functional Theory of Atoms and Molecules New York Oxford
University Press 1989
(31) Hohenberg P Kohn W Physical Review B 1964 136 B864
(32) Elstner M Porezag D Jungnickel G Elsner J Haugk M Frauenheim T Suhai S
Seifert G Physical Review B 1998 58 7260
(33) Zhechkov L Heine T Patchkovskii S Seifert G Duarte H A Journal of Chemical
Theory and Computation 2005 1 841
(34) Elstner M Hobza P Frauenheim T Suhai S Kaxiras E Journal of Chemical Physics
2001 114 5149
(35) Oliveira A F Seifert G Heine T Duarte H A Journal of the Brazilian Chemical Society
2009 20 1193
(36) Seifert G Joswig J-O Wiley Interdisciplinary Reviews-Computational Molecular Science
2012 2 456
(37) Heine T Rapacioli M Patchkovskii S Frenzel J Koester A M Calaminici P
Escalante S Duarte H A Flores R Geudtner G Goursot A Reveles J U Vela A Salahub D
R deMon deMon-nano edn deMon-nano 2009
(38) BCCMS B DFTB+ - Density Functional based Tight binding (and more)
(39) Monkhorst H J Pack J D Physical Review B 1976 13 5188
(40) SCM Amsterdam Density Functional 2012
(41) University of Karlsruhe and Forschungszentrum Karlsruhe gGmbH - TURBOMOLE V63
2011 2007
(42) Dovesi R Saunders V R Roetti C Orlando R Zicovich-Wilson C M Pascale F
Civalleri B Doll K Harrison N M Bush I J DrsquoArco P Llunell M CRYSTAL09 Users Manual
University of Torino Torino 2009 2009
(43) Wadt W R Hay P J Journal of Chemical Physics 1985 82 284
(44) Roy L E Hay P J Martin R L Journal of Chemical Theory and Computation 2008 4
1029
35
(45) Ehlers A W Bohme M Dapprich S Gobbi A Hollwarth A Jonas V Kohler K F
Stegmann R Veldkamp A Frenking G Chemical Physics Letters 1993 208 111
(46) Garberoglio G Skoulidas A I Johnson J K Journal of Physical Chemistry B 2005 109
13094
(47) Han S S Mendoza-Cortes J L Goddard W A III Chemical Society Reviews 2009 38
1460
(48) Getman R B Bae Y-S Wilmer C E Snurr R Q Chemical Reviews 2012 112 703
(49) Cote A P El-Kaderi H M Furukawa H Hunt J R Yaghi O M Journal of the American
Chemical Society 2007 129 12914
(50) Wan S Guo J Kim J Ihee H Jiang D Angewandte Chemie-International Edition 2008
47 8826
(51) Wan S Guo J Kim J Ihee H Jiang D Angewandte Chemie-International Edition 2009
48 5439
(52) Zhou W Wu H Yildirim T Chemical Physics Letters 2010 499 103
(53) Spitler E L Koo B T Novotney J L Colson J W Uribe-Romo F J Gutierrez G D
Clancy P Dichtel W R Journal of the American Chemical Society 2011 133 19416
(54) Lukose B Kuc A Heine T Chemistry-a European Journal 2011 17 2388
(55) Uribe-Romo F J Hunt J R Furukawa H Klock C OKeeffe M Yaghi O M Journal of
the American Chemical Society 2009 131 4570
(56) Schmid R Tafipolsky M Journal of the American Chemical Society 2008 130 12600
(57) Patchkovskii S Heine T Physical Review E 2009 80
(58) Shekhah O Wang H Paradinas M Ocal C Schuepbach B Terfort A Zacher D
Fischer R A Woell C Nature Materials 2009 8 481
(59) Li H Eddaoudi M Groy T L Yaghi O M Journal of the American Chemical Society
1998 120 8571
(60) Rappe A K Casewit C J Colwell K S Goddard W A Skiff W M Journal of the
American Chemical Society 1992 114 10024
(61) Frauenheim T Seifert G Elstner M Hajnal Z Jungnickel G Porezag D Suhai S
Scholz R Physica Status Solidi B-Basic Research 2000 217 41
(62) Elstner M Cui Q Munih P Kaxiras E Frauenheim T Karplus M Journal of
Computational Chemistry 2003 24 565
(63) Heine T Dos Santos H F Patchkovskii S Duarte H A Journal of Physical Chemistry A
2007 111 5648
(64) Sternberg M Zapol P Curtiss L A Molecular Physics 2005 103 1017
(65) Zhang C Zhang Z Wang S Li H Dong J Xing N Guo Y Li W Solid State
Communications 2007 142 477
36
(66) Munch W Kreuer K D Silvestri W Maier J Seifert G Solid State Ionics 2001 145
437
(67) Bahr D F Reid J A Mook W M Bauer C A Stumpf R Skulan A J Moody N R
Simmons B A Shindel M M Allendorf M D Physical Review B 2007 76
(68) Amirjalayer S Tafipolsky M Schmid R Journal of Physical Chemistry C 2011 115
15133
(69) Amirjalayer S Snurr R Q Schmid R Journal of Physical Chemistry C 2012 116 4921
(70) Lukose B Kuc A Frenzel J Heine T Beilstein Journal of Nanotechnology 2010 1 60
(71) Ding X Chen L Honsho Y Feng X Saenpawang O Guo J Saeki A Seki S Irle S
Nagase S Parasuk V Jiang D Journal of the American Chemical Society 2011 133 14510
37
Appendix A
Review Covalently-bound organic frameworks
Binit Lukose and Thomas Heine
To be submitted for publication after revision
Contents
1 Introduction
2 Synthetic achievements
21 Covalent Organic Frameoworks (COFs)
22 Covalent-Triazine Frameworks (CTFs)
23 Porous Aromatic Frameworks (PAFs)
24 Schemes for synthesis
25 List of materials
3 Studies of the underlying structure and properties of COFs
4 Applications
41 Gas storage
411 Porosity of COFs
412 Experimental measurements
413 Theoretical preidctions
414 Adsorption sites
415 Hydrogen storage by spillover
42 Diffusion and selectivity
43 Suggestions for improvement
431 Geometry modifications
432 Metal doping
433 Functionalization
5 Conclusions
6 List and pictures of chemical compounds
38
1 Introduction
Nanoporous materials have perfectly ordered voids to accommodate to interact with and to
discriminate molecules leading to prominent applications such as gas storage separation and sieving
catalysis filtration and sensoring1-4 They are defined as porous materials with pore diameters less
than 100 nm and are classified as micropores of less than 2 nm in diameter mesopores between 2
and 50 nm and macropores of greater than 50 nm The internal geometry that determines pore size
and surface area can be precisely engineered at molecular scales Reticular synthetic methods
suggested by Yaghi and co-workers5-7 can accomplish pre-designed frameworks The procedure is to
select rigid molecular building blocks prudently and assemble them into destined networks using
strong bonds
Several types of materials have been synthesized using reticular chemistry concepts One prominent
group which is much appreciated for its credentials is the Metal-Organic Frameworks (MOFs)8-13 in
which metal-oxide connectors and organic linkers are judiciously selected to form a large variety of
frameworks The immense potential of MOFs is facilitated by the fact that all building blocks are
inexpensive chemicals and that the synthesis can be carried out solvothermally The progress in MOF
synthesis has reached the point that some of the MOFs are commercially available Several MOFs of
ultrahigh porosity have also been successfully synthesized notably the non-interpenetrated IRMOF-
74-XI which has a pore size of 98 Aring Scientists are also able to synthesize MOFs from cheap edible
natural products14 Since the discovery of MOFs many other crystalline frameworks have been
synthesized using reticular chemistry such as Metal-Organic Polyhedra (MOP)15-19 Zeolite
Imidazolate Frameworks (ZIFs)20-22 and Covalent Organic Frameworks (COFs)23-29
COFs are composed of covalent building blocks made of boron carbon oxygen and hydrogen in
many cases also including nitrogen or silicon stitched together by organic subunits The atoms are
held together by strong covalent bonds Depending on the selection of building blocks the COFs may
form in two (2D) or three (3D) dimensions Planar building blocks are the constituents of 2D COFs
whereas for the formation of 3D COFs typically tetragonal building blocks are involved High
symmetric covalent linking as it is perceived in reticular chemistry was confirmed for the end
products5
Unlike the case of supramolecular assemblies the absence of noncovalent forces between the
molecular building units endorses exceptional rigidity and stability for COFs They are in general
thermally stable above 4000 C Also in COFs the stitching of molecular building units guarantees an
39
increased order and allows control over porosity and composition Without any metals or other
heavy atoms present they exhibit low mass densities This gives them advantages over MOFs in
various applications for example higher gravimetric capacities for gas storage3031 The lowest
density crystal known until today is COF-10824 In addition COFs exhibit permanent porosity with
specific surface areas that are comparable to MOFs and hence larger than in zeolites and porous
silicates
MOF and COF crystals possess long range order although COFs have been achieved so far only at the
μm scale Reversible solvothermal condensation reactions are credited for the high order of
crystallinity Other new framework materials such as Covalent Triazine Frameworks (CTFs) and
Porous Aromatic Frameworks (PAFs) differed in ways of synthesis The former was prepared by
ionothermal polymerization while the latter was produced by coupling reactions PAFs lack long
range order in the crystals as a result of the irreversible synthesis Nevertheless many of the
materials are promisingly good for applications In this review we intend to discuss the synthetic
achievements of COF CTFs and PAFs and studies on their structure properties and prominent
applications
For easy understanding of the atomic geometry of a material the reader is referred to Table 1 which
gives the names of the framework materials the reactants and the synthesis scheme Figure 5 shows
the reactants and Figure 4 shows the reaction schemes Example reactions are given in Figure 1-3
Abbreviations of each chemical compound are given in Section 6
2 Synthetic achievements
21 Covalent Organic Frameworks (COFs)
In the year 2005 Yaghi et al23 achieved the crystallization of covalent organic molecules in the form
of periodic extended layered frameworks The condensation of discrete molecules of different sizes
enabled the synthesis of micro- and mesoporous crystalline COFs27 The selection of molecules with 2
and 3 connection sites for synthesis led to the formation of hexagonal layered geometries32 Yaghi et
al have synthesized half a dozen three-dimensional crystalline COFs solvothermally using tetragonal
building blocks containing 4 connection sites since 2007242829 Figure 1 shows a few examples of 2D
and 3D COFs formed by the self-condensation of boronic acids such as BDBA TBPM and TBPS and co-
condensation of the same boronic acids with HHTP
40
Figure 1 Solvothermal condensation (dehydration) of monomers to form 2D and 3D COFs Atom colors are boron yellow oxygen red yellow (big) silicon grey carbon
Alternate synthetic procedures were also exploited for production and functionalization of COFs
Lavigne et al33 showed that the synthesis of COFs using polyboronate esters is facile and high-yielded
41
Boronate esters often contain multiple catechol moieties which are prone to oxidation and are
insoluble in organic solvents Dichtel et al3435 presented Lewis acid-catalysis procedures to form
boronate esters of catechols and arylboronic acids without subjecting to oxidation Cooper et al36
successfully utilized microwave heating techniques for rapid production (~200 times faster than
solvothermal methods) of both 2D and 3D COFs The syntheses of phthalocyanine3437 or porphyrin38
based square COFs have been reported in literature The latter was noticed for its time-dependent
crystal growth which also affects its pore parameters
Abel et al39 reported the ability to form COFs on metallic surfaces COF surfaces (SCOFs) have been
formed by molecular dehydration of boroxine and triphenylene on a silver surface Albeit with some
defects the materials showed high temperature stability allowing to proceed with functionalization
Lackinger et al40 realized the role of substrates on the formation of surface COFs from the surface-
generated radicals Halogen substituted polyaromatic monomers can be sublimed onto metal
substrates and ultimately turned into a COF after homolysis and intermolecular colligation
Chemically inert graphite surfaces cannot support the homolysis of covalent carbon-halogen bonds
and thus cannot initiate the subsequent association of radicals COF layers can be formed onto
Cu(111)40 and Au(111)41 surfaces but with lower crystal ordering Beton et al41 deposited the
monomers on annealed Au(111) surfaces which supplied surface mobility for the monomers and
subsequently increased porosity and surface coverage of the covalent networks Unlike the bulk form
at the surface the monomers were self-organized onto polygons with 4 to 8 edges Such a template
was able to organize a sublimed layer of C60 fullerenes the number of C60 molecules adsorbed in a
cavity was correlated to the size of the polygon
In 2009 Sassi et al42 studied the problem of crystal growth of COFs on substrates They calculated
the reaction mechanism of 14-benzenediboronic acid (BDBA) on Ag(111) surface self-condensation
of BDBA molecules in solvents leads to the formation of two-dimensional (the planar) stacked COF-1
For the surface COFs the study using Density Functional Theory reveals that there are neither
preferred adsorption sites for the molecules nor a charge transfer between the molecules and the
surface Hence the electronic structure of the molecules remains unchanged and the role of the
metal surface is limited in providing the planar nature for the polymer The free reaction enthalpy
(Gibbs free energy) difference of the reaction ∆G = ∆H ndash T∆S is approximated using the harmonic
approximation taking into account the geometrical restrictions of the metal surface and the entropic
contributions of the released water molecules As result the formation of SCOF-1 has been found to
be exothermic if more than six monomers are involved Ourdjini et al43 reported the polymerization
of 14-benzenediboronic acid (BDBA) on different substrates (Ag(111) Ag(100) Au(111) and Cu(111))
and at different source and substrate temperatures to follow how molecular flux and adsorption-
42
diffusion environments should be controlled for the formation of polymers with the smallest number
of structural defects High flux is essential to avoid H-bonded supramolecular assemblies of
molecules and the substrate temperature needs to be optimized to allow the best surface diffusion
and longest residential time of the reactants Achieving these two contradictory conditions together
is a limitation for some substrates eg for copper Silver was found to be the best substrate for
producing optimum quality polymers Controlling the growth parameters is important since
annealing cannot cure defects such as formation of pentagons heptagons and other non-hexagonal
shapes which involved strong covalent bonds
Ditchel et al44 successfully grew COF films on monolayer graphene supported by a substrate under
operationally simple solvothermal conditions The films have better crystallinity compared to COF
powders and can facilitate photoelectronic applications straightaway Zamora et al45 have achieved
exfoliation of COF layers by sonication The thin layered nanostructures have been isolated under
ambient conditions on surfaces and free-standing on carbon grids
A noted characteristic of COFs is its organic functionality Judiciously selected organic units ndash pyrene
and triphenylene (TP) ndash for the production of TP COF can not only harvest photons over a wide range
but also facilitate energy migration over the crystalline solid25 Polypyrene (PPy)-COF that consists of
a single aromatic entity ndash pyrene- is fluorescent upon the formation of excimers and hence undergo
exciton migration across the stacked layers26 A nickel(II)-phthalocyanine based layered square COF
that incorporates phenyl linkers and forms cofacially stacked H-aggregates was reported to absorb
photons over the solar spectrum and transport charges across the stacked layers37 COF-366 and
COF-66 showed excellent hole mobility of 81 and 30 cm2 V-1s-1 surpassing that of all polycrystalline
polymers known until today46 A first example of an electron-transporting 2D COF was reported
recently47 This square COF was built from the co-condensation of nickel(II) phthalocyanine and
electron-deficient benzothiadiazole With the nearly vertical arrangement of its layers it exhibits an
electron mobility of 06 cm2 V-1s-1 The enhanced absorbance is over a wide range of wavelengths up
to 1000 nm In addition it exhibits panchromatic photoconductivity and high IR sensitivity
Lavigne et al48 achieved the condensation of alkyl (methyl ethyl and propyl) substituted organic
building units with boronic acids forming functionalized COFs The alkyl groups contribute for higher
molar adsorption of H2 however the increased mass density of the functionalized COFs cause for
decreased gravimetric storage capacities Boronate ester-linked COFs are stable in organic solvents
however undergo rapid hydrolysis when exposed in water49 In particular the instability of COF-1
upon H2O adsorption shall be mentioned here50 Lanni et al claimed that alkylation could bring
hydrolytic stability into COFs49
43
Functionalization and pore surface engineering in 2D COFs can be improved if azide appended
building blocks are stitched together in click reactions with alkynes51 Control over the pore surface
is made possible by the use of both azide appended and bare organic building units the ratios of
which is matching with the final amount of functionalization in the pore walls The click reactions of
azide groups with alkynes within the pores lead to the formation of triazole-linked groups on the
pore surfaces This strategy also gives the relief of not condensing the already functionalized building
units Jiang et al have asserted eclipsed layering for most of the final COFs using powder X-ray
diffraction pattern Exceptions are the hexagonal functionalized COFs with relatively high azide-
content (ge 75) Their weak XRD signals suggest possible layer-offset or de-lamellation Although
functionalization on pore surfaces exhibits the nature of lowering the surface area the possibility to
add reactive groups within the pores can be utilized for gas-sorption selectivity The authors have
claimed a 16 fold selectivity of CO2 over N2 using 100 acetyl-rich COF-5 compared to the pure COF-5
The range of the click reaction approach is so wide that relatively large chromophores can be
accommodated in the pores thereby making COF-5 fluorescent
Functionalization of 3D COFs has just reported in 201252 Dichtel et al used a monomer-truncation
strategy where one of the four dihydroxyborylphenyl moieties of a tetrahedral building unit was
replaced by a dodecyl or allyl functional group Condensation of the truncated and the pure
tetrahedral blocks in a certain ratio created a functionalized 3D COF The degree of functionalization
was in proportion with the ratio The crystallinity was maintained except for high loading (gt 37) of
truncated monomers The pore volume and the surface area were decreased as a function of loading
level
A heterogeneous catalytic activity of COFs is achieved with an imine-linked palladium doped COF by
enabling Suzuki-Miyaura coupling reaction within the pores53 The COF-LZU1 has a layered geometry
that has the distance between nitrogen atoms in the neighboring linkers in a level (~37 Aring) sufficient
to accommodate metal ions Palladium was incorporated by a post-treatment of the COF PdCOF-
LZU1 was applied to catalyze Suzuki-Miyaura coupling reaction This coupling reaction is generally
used for the facile formation of C-C bonds54 The increased structural regularity and high insolubility
in water and organic solvents are adequate qualities of imine-linked COFs to perform as catalysts
Experiments with the above COF show a broad scope of the reactants excellent yields of the
products and easy recyclability of the catalyst
The comparatively higher thermal stability of COFs is often noted and is explainable with their strong
covalent bonds The reversible dehydrations for the formation of most of the COFs point to their
instability in the presence of water molecules This has been tested and verified for some layered
COFs with boronate esters49 Such COFs are stable as hosts for organometallic molecules55 COF-102
44
framework was found to be stable and robust even in the presence of highly reactive cobaltocenes
The highly stable ferrocenes appear to have an arrangement within the framework led by π-π
interactions
22 Covalent Triazine Frameworks (CTFs)
In the year 2008 Thomas et al56 synthesized a crystalline extended porous nanostructure by
trimerization of polytriazine monomers in ionothermal conditions With the catalytic support of ZnCl2
three aromatic nitrile groups can be trimerized to form a hexagonal C3N3 ring The resulting structure
known as Covalent Triazine-based Frameworks (CTFs) is similar to 2D COFs in geometry and atomic
composition (see Figure 2) Usage of larger non-planar monomers at higher amounts of ionic salts
however led to the formation of contorted structures Interestingly those structures showed
enhanced surface area and pore volume The trimerization of monomers that lack a linear
arrangement of nitrile groups ended up as organic polymer networks Later the same group
introduced mesoporous channels onto a CTF (CTF-2)57 by increasing temperature and ion content
The resulting structure however was amorphous It is concluded that the reaction parameters and
the amount of salt play a crucial role for tuning the porosity and controlling the order of the material
respectively58
Figure 2 Trimerization of nitrile monomers to form CTF-1 Atom colors blue N grey C white H
Ben et al59 followed ionothermal conditions for the targeted synthesis of a 3D framework using
tetraphenylmethane block and a triangular triazine ring The end product was amorphous thermally
stable up to 400 0C and could be applied for the selective sorption of benzene over cyclohexane On a
later synthetic study Zhang et al60 reported that this polymerization could be accomplished in short
45
reaction times under microwave enhanced conditions The template-free high temperature dynamic
polymerization reactions constitute irreversible carbonization reactions coupled with reversible
trimerization of nitriles The irreversibility encounter in these condensation reactions is responsible
for the production of frameworks as amorphous solids6162
An amorphous yet microporous CTF was found to be able to catalyze the oxidation of glycerol with
Pd nanoparticles that are dispersed in the framework63 This solid catalyst appears to be strong
against deactivation and selective toward glycerate compared to Pd supported activated carbon This
is attributed to the relatively high stability of Pd nanoparticles that benefit from the presence of
nitrogen functionalities in CTF An advancement on selective oxidation of methane to methanol at
low temperature is accomplished by using an amorphous bipyridyl-rich platinum-coordinated CTF as
catalyst64
23 Porous Aromatic Frameworks (PAFs)
a notably high surface area (BET surface area of 5600 m2g-1 Langmuir surface area of 7100 m2g-1)65
PAF-1 has been synthesized in a rather simple way using a nickel(0)-catalyzed Yamamoto-type66
Ullmann cross-coupling reaction67(see Figure 3) The material is thermally (up to 520 0C in air) and
hydrothermally (in boiled water for a minimum of seven days) stable The framework exposes all
faces and edges of the phenyl rings to the pores and hence the high surface area can be achieved
while its high stability is inherited from the parent diamond structure The synthesized material was
verified as disordered and amorphous yet with uniform pore diameters The excess H2 uptake
capacity of PAF-1 was measured as 70 wt at 48 bar and 77 K It can adsorb 1300 mg g-1 CO2 at 40
bar and room temperature PAF-1 was also tested for benzene and toluene adsorption
Figure 3 Coupling reaction using halogenated tetragonal blocks for the formation of diamond-like PAF-1 Atom colors red Br grey C white H
46
An attempt to synthesize a diamondoid with quaterphenyl as linker is described recently68 A
tetrahedral unit and a linear linker each of them has two phenyl rings were polymerized using the
Suzuki-Miyaura coupling reaction54 The product (PAF-11) however stayed behind theoretical
predictions and performed poorly pointing to its shortcomings such as short-range order distortion
and interpenetration of phenyl rings Nevertheless the amorphous solid displayed high thermal and
chemical stabilities proneness for adsorbing methanol over water and usability for eliminating
harmful aromatic molecules
PAF-569 is a two-dimensional hexagonal organic framework synthesized using the Yamamoto-type
Ullmann reaction This material is composed only of phenyl rings however lack long range order
because of the distortion of the phenyl rings and kinetics controlled irreversible coupling reaction It
retains a uniform pore diameter and provides high thermal and chemical stability despite its
amorphous nature PAF-5 was suggested for adsorbing organic pollutants at saturated vapour
pressure and room temperature
Co-condensation reactions with piperazine enabled nucleophilic substitution of cyanuric chloride to
form a covalently linked porous organic framework PAF-670 The synthesis in mild conditions yields a
product with uniform morphology and a certain degree of structural regularity Being nontoxic this
material was tested for drug delivery thereby stepping into biomedical applications of covalently
linked organic frameworks Primary tests suggest that PAF-6 could be promoted for drug release for
its permanent porosity and facile functionalization In comparison with MCM-4171 a widely tested
inorganic framework PAF-6 performed equally or even superiorly
24 Schemes for synthesis
The majority of the COFs were synthesized using solvothermal step-by-step condensation
(dehydration) reactions The method incorporates reversibility and is applicable for supplying long
range order in COF materials The reactants generally consist of boronic acids and hydroxy-
polyphenyl (catechol) compounds Extended porous networks are obtained when O-B or O-C bonds
are formed into 5 or 6-membered rings after dehydration (scheme 1) Dehydration may also be
carried out using microwave synthesis Alternately boronic acid can react with catechol acetonide in
presence of a Lewis acid catalyst The molecules link to each other after the liberation of acetone and
water (scheme 2) The utilization of phenyl-attached formyl and amino groups as building units
results in the formation of imines after dehydration (scheme 3) A desired arrangement of molecular
arrays on surfaces can be fulfilled by depositing planar molecules on metallic surfaces followed by
covalent linking using annealing
47
Isocyanate groups follow trimerization to form poly-isocyanurate rings (scheme 4) Cyclotrimerization
of monomers containing nitrile groups or cyanate groups is prone to generate C3N3 rings (scheme 5)
However the ionothermal synthesis of them resulted with amorphous materials Unique bond
formation is often not achieved throughout the material and thus the crystal lacks long-range order
Analogously coupling reactions such as Ullmann and Suzuki-Miyaura give rise to amorphous
products However they are adequate in producing C-C bonds when halogen-substituted compounds
are reacted (scheme 6) Nucleophilic substitution may also be used for framing networks in cases
like reactants are formed into nucleophiles and ions after release of their end-groups (scheme 7)
48
Figure 4 Different schemes reported for the synthesis of covalently-bound organic frameworks
49
25 List of synthesized materials
Table 1 List of synthesized materials in the literature their building blocks synthesis methods measured surface area [m
2 g
-1] pore volume [cm
3 g
-1] and pore size [Aring]
COF Names Reactants Synthesis Surface
Area
Pore
volume
Pore
size
COF-1 BDBA Solvothermal condensation235072
scheme 1
711 62850 032
03650
9
COF-5 BDBA HHTP Solvothermal condensation23
scheme 1
1590 0998 27
Microwave3673 scheme 1 2019
BDBA TPTA Lewis acid catalysis35 TPTA
COF-6 BTBA HHTP Solvothermal condensation27
scheme 1
980 (L) 032 64
COF-8 BTPA HHTP Solvothermal condensation27
scheme 1
1400 (L) 069 187
COF-10 BPDA HHTP Solvothermal condensation27
scheme 1
2080 (L) 144 341
BPDA TPTA Lewis acid catalysis35 scheme 2
COF-18Aring BTBA THB Facile dehydration3348 scheme 1 1263 069 18
COF-16Aring BTBA alkyl-THB
(alkyl = CH3)
Facile dehydration48 scheme 1 753 039 16
COF-14Aring BTBA alkyl-THB
(alkyl = C2H5)
Facile dehydration48 scheme 1 805 041 14
COF-11Aring BTBA alkyl-THB
(alkyl = C3H7)
Facile dehydration48 scheme 1 105 0052 11
50
SCOF-1 BDBA Substrate-assisted synthesis39
scheme 1
SCOF-2 BDBA HHTP Substrate-assisted synthesis39
scheme 1
TP COF PDBA HHTP Solvothermal condensation25
scheme 1
868 079 314
PPy-COF PDBA Solvothermal condensation26
scheme 1
923 188
TBB COF TBB (on Cu(111) and
Ag(110) surfaces)
Surface polymerisation40 scheme
6
TBPB COF TBB (on Au(111)
surface)
Surface polymerisation41 scheme
6
BTP COF BTPA THDMA Solvothermal condensation72
scheme 1
2000 163 40
HHTP-DPB COF DPB HHTP Solvothermal condensation73
scheme 1
930 47
PICU-A DMBPDC Cyclotrimerization74 scheme 4
PICU-B DCF Cyclotrimerization74 scheme 4
COF-LZU1 DAB TFB Solvothermal condensation53
scheme 3
410 054 12
PdCOF-LZU1 COF-LZU1 PdAc Facile post-treatment53 146 019 12
XN3-COF-5 X N3-BDBA (100-X)
BDBA HHTP
Solvothermal condensation51
scheme 1
2160
(X=5)
1865 (25)
1722 (50)
1641 (75)
1421
(100)
1184
1071
1016
0946
0835
295
276
259
258
227
51
XAcTrz-COF-5 XN3-COF-5 PAc Click reaction51 2000
(X=5)
1561 (25)
914 (50)
142 (75)
36 (100)
1481
0946
0638
0152
003
298
243
156
153
125
XBuTrz-COF-5 XN3-COF-5 HP Click reaction51
XPhTrz-COF-5 XN3-COF-5 PPP Click reaction51
XEsTrz-COF-5 XN3-COF-5 MP Click reaction51
XPyTrz-COF-5 XN3-COF-5 PyMP Click reaction51
COF-42 DETH TFB Solvothermal condensation75
scheme 3
710 031 23
COF-43 DETH TFPB Solvothermal condensation75
scheme 3
620 036 38
CTF-1 DCB Ionothermal trimerization56
scheme 5
791 040 12
CTF-2 DCN Ionothermal trimerization57
scheme 5
90 8
mp-CTF-2 2255 151 8
CTF (DCP) DCP Ionothermal trimerization64
scheme 5
1061 0934 14
K2[PtCl4]-CTF DCP K2[PtCl4] Trimerization (scheme 5) +
coordination64
Pt-CTF DCP Pt Trimerization (scheme 5) +
coordination64
PAF-5 TBB Yamamoto-type Ullmann cross-
coupling reaction69 scheme 6
1503 157 166
52
PAF-6 PA CA Nucleophilic substitution70
scheme 7
1827 118
Pc-PBBA COF PcTA BDBA Lewis acid catalysis34 scheme 2 506 (L) 0258 217
NiPc-COF OH-Pc-Ni BDBA Solvothermal condensation37
scheme 1
624 0485 190
XN3-NiPc-COF OH-Pc-Ni X N3-BDBA
(100-X) BDBA
Solvothermal condensation51
scheme 1
XEsTrz-NiPc-
COF
XN3-NiPc-COF MP Click reaction51
ZnP COF TDHB-ZnP THB Solvothermal condensation38
scheme 1
1742 1115 25
NiPc-PBBA COF NiPcTA BDBA Lewis acid catalysis35 scheme 2 776
2D-NiPc-BTDA
COF
OHPcNi BTDADA Solvothermal condensation47
scheme 1
877 22
ZnPc-Py COF OH-Pc-Zn PDBA Solvothermal condensation
scheme 1
420 31
ZnPc-DPB COF OH-Pc-Zn DPB Solvothermal condensation
scheme 1
485 31
ZnPc-NDI COF OH-Pc-Zn NDI Solvothermal condensation
scheme 1
490 31
ZnPc-PPE COF OH-Pc-Zn PPE Solvothermal condensation
scheme 1
440 34
COF-366 TAPP TA Solvothermal condensation46
scheme 3
735 032 12
COF-66 TBPP THAn Solvothermal condensation46
scheme 1
360 020 249
53
COF-102 TBPM Solvothermal condensation24
scheme 1
3472 135 115
Microwave36
scheme 1
2926
COF-102-C12 TBPM trunk-TBPM-R
(R=dodecyl)
Solvothermal condensation52
scheme 1
12
COF-102-allyl TBPM trunk-TBPM-R
(R=allyl)
Solvothermal condensation52
scheme 1
COF-103 TBPS Solvothermal condensation24
scheme 1
4210 166 125
COF-105 TBPM HHTP Solvothermal condensation24
scheme 1
COF-108 TBPM HHTP Solvothermal condensation24
scheme 1
COF-202 TBPM TBST Solvothermal condensation28
scheme 1
2690 109 11
COF-300 TAM TA Solvothermal condensaion29
scheme 3
1360 072 72
PAF-1 TkBPM-X (X=C) Yamamoto-type Ullmann cross-
coupling reaction65 scheme 6
5600
PAF-2 TCM cyclotrimerization59 scheme 5 891 054 106
PAF-3 TkBPM-X (X=Si) Yamamoto-type Ullmann cross-
coupling reaction76 scheme 6
2932 154 127
PAF-4 TkBPM-X (X=Ge) Yamamoto-type Ullmann cross-
coupling reaction76 scheme 6
2246 145 118
PAF-11 TkBPM-X (X=C) BPDA Suzuki coupling reaction68 704 0203 166
54
scheme 6
3 Studies of structure and properties of COFs
31 2D COFs
Following the literature of the years 2005-2010 2D COFs were believed to have high symmetric AA
(eclipsed ndash P6mmm symmetry) stacking as boron nitride (h-BN) [REF] An exception was COF-1
which was considered to have AB (staggered ndash P63mmc) stacking as graphite[REF] The AA stacking
maximizes the attractive London dispersion interaction between the layers an important
contribution to the stacking energy At the same time AA stacking always suffers repulsive Coulomb
force between the layers due to the polarized connectors as the distance between atoms exposing
the same charge is closest It should be noted that in boron nitride boron atoms serve as nearest
neighbors to nitrogen atoms in adjacent layers The Coulomb interaction has ruled out possible
interlayer eclipse between atoms of alike charges Based on these facts we modeled77 a set of 2D
COFs with different possible stacks Each layer was drifted with respect to the neighboring layers in
directions symmetric to the hexagonal pore and we calculated the total energies and their Coulombic
contributions The AA stacking version was found to have the highest Coulombic repulsion in each
COF The lattice types with the layers shifted to each other by ~14 Aring approximately the bond length
between two atoms in a 2D COF were found to be more stable compared to AA or AB In this low-
symmetry configuration the hexagonal rings in the pore walls undergo staggering with that of
adjacent layers Within an error limit the bulk COFs may be either serrated or inclined as shown in
Figure 5 The interlayer separations of all the COF stacks are in the range of 30 to 36 Aring and increase
in the order AB serratedinclined AA32 The motivation for proposing inclined stacking for COFs is
that the rhombohedral graphite (ABC stacking) is the inclined version of the layer-offset in natural
graphite (AB stacking) The instability of AA stacking was also suggested by the studies of elastic
properties of COF-1 and COF-5 by Zhou et al78 The shear modulus show negative signs for the
vertical alignment of COF layers while they are small but positive for the offset of layers
55
Figure 5 Different stacks of a 2D covalent organic framework COF-1 The atoms are shown as Van der Waals spheres
The different stacking modes should in principle be visible in their PXRD patterns as each space
group has a distinct set of symmetry imposed reflection conditions We simulated the XRD patterns
of COFs in their known and new configurations and on comparison with the experimental spectrum
the new as well as AA stacking showed good agreement3279 However the slight layer-offsets in
conjunction with the large number of atoms in the unit cells change the PXRD spectrum only by the
appearance of a few additional peaks all of them in the vicinity of existing signals and by changes in
relative intensities Unfortunately the low resolution of the experimental data does now allow
distinguishing between the different stackings as the broad signals cover all the peaks of the
simulated spectrum
A detailed analysis on the interlayer stacking of HHTP-DPB COF has been made by Dichtel et al73 is
very complementary73 This work uses molecular mechanics extensively to define potential energy
surfaces of stacking of two layers from eclipsed to staggered covering all possible layer offsets The
low energy region was near to the eclipsed stacking which was then zoomed in to examine using DFT
for higher accuracy The far-eclipsed regions encounter no energy barriers with the near-eclipsed
regions which suggest the unlikeness of forming near-staggered layering Within DFT the eclipsed
geometries are at high energy compared to the slight offsets around AA (~ 13 ndash 25 Aring away) For a
hexagonal COF the energetically preferred offset could be in one of the six hexagonal sectors (or
circle sectors) of central angle 60 The preference of falling into one sector over the others is
arbitrary and may be determined by symmetry set by nature or external parameters Hence the
layering order in bulk could be serrated inclined spiral or purely by random The latter case better
known as turbostratic disorder is then more like a rule for 2D COFs rather than an exception caused
by external forces This also explains why the experimental PXRD patters have relatively broad peaks
The layer-offset may not only change the internal pore structure but also affect interlayer exciton
and vertical charge transport in opto-electronic applications
56
Concerning the stacking stability the square 2D COFs are expected not to be different from
hexagonal COFs because the local environment causing the shifts is nearly the same DFTB based
calculations reported along with the synthesis of electron-transporting 2D-NiPc-BTDA COF supports
this notion47 Two shifted configurations ndash at 07 and 14 Aring away from AA ndash appeared to be
energetically preferred over the AA stacking It appears that AA stacking is only possible for boron
nitride-like structures were adjacent layers have atoms with opposite charges in vertical direction In
analogy to BN layers it might be possible to force AA stacking by the introduction of alkalis in
between the layers
32 3D COFs
3D COFs in general are composed of tetragonal and triangular building blocks So far that their
synthesis leads to two high symmetry topologies ndash ctn and bor - in high yields24 The two topologies
differ primarily in the twisting and bulging of their components at the molecular level The
thermodynamic preference of one topology over the other may result from the kinetic entropic and
solvent effects and the relative strain energies of the molecular components It is straight-forward to
state that the effects at the molecular level crucial crucial in the bulk state since transformation from
one net to the other is impossible without bond-breaking There has not been any detailed study on
this matter experimentally or theoretically
Schmid et al8182 have developed force-field parameters from first principles calculations using
Genetic Algorithm approach The parameters developed for cluster models of COF-102 can
reproduce the relative strain energies in sufficient accuracies and be extended to calculations on
periodic crystals The internal twists of aromatic rings within the tetragonal units are different for ctn
and bor nets which lead to stable S4 and unstable D2d symmetries respectively This together with
the bulging of triangular units in the bor net reasoned for energetic preference of ctn over bor for all
boron-based 3D COFs79
The connectivity-based atom contribution method (CBAC) developed by Zhong et al80 can
significantly reduce computational time needed for quantum chemical calculation for framework
charges when screening a large number of MOFs or COFs in terms of their adsorption properties The
basic assumption is that atom types with same bonding connectivity in different MOFsCOFs have
identical charges a statement that follows from the concept of reticular chemistry where the
properties of the molecular building blocks keep their properties in the bulk After assigning the
CBAC charges to the atom types slight adjustment (usually less than 3 ) is needed to make the
frameworks neutral Simulated adsorption isotherms of CO2 CO and N2 in MOFs and COFs show that
CBAC charges reproduce well the results of the QM charges8384 The CBAC method still requires a
57
well-parameterized force field in order to account correctly for adsorption isotherms as the second
important contribution to the host-guest interaction is the London dispersion energy between
framework and adsorbed moleculesG
The elastic constants calculated for the 3D COFs COF-102 and COF-108 show that COF-102 is nearly
five times stronger than COF-10878 While the bulk modulus (B) of COF-102 (B = 216 GPa) exceeds
that of known MOFs such as MOF-5 MOF-177 and MOF-205 (B = 153 101 107 GPa respectively81)
the least dense COF-108 is at the edge of structural collapse (B = 49 GPa) The calculations were
made using Density Functional Tight-Binding (DFTB) method Another report82 based on the same
level of theory possibly with a different parameter set however reveals lower bulk moduli for both
COFs 177 GPa for COF-102 and 005 GPa for COF-108 The bulk moduli for COF-103 and COF-105 are
110 and 33 GPa respectively Recently we have reported the structural properties of 3D COFs The
calculations were made using self-consistent charge (SCC) DFTB method The bulk modulus of each
COF is as follows COF-102 ndash 206 COF-103 ndash 139 COF-105 ndash 79 COF-108 ndash 37 COF-202 ndash 153 and
COF-300 ndash 144 GPa Force field calculations79 provide slightly different values COF-102 ndash 170 COF-
103 ndash 118 COF-105 ndash 57 COF-108 ndash 29 COF-202 ndash 110 GPa Albeit the differences in values each
type of calculation shows the trend that bulk modulus decreases with decreasing mas density and
increasing pore volume and distance between connection nodes One has to note that the high
mechanical stabilities of 3D COFs as suggested by the calculations correspond always to defect-free
crystals Theory is expected therefore to overestimate experimental mechanical stability for real
materials
COFs in general have semiconductor-like band gaps (~ 17 to 40 eV)32 Layer-offset from eclipsed
layering slightly increases the band gap However the Density-Of-States (DOS) of a monolayer is
similar to that of bulk COF The band gap is generally smaller for a COF with larger number conjugate
rings
The thermal influence on the crystal structure of 3D COFs is also intriguing as negative thermal
expansion (NTE) the contraction of crystal volume on heating has been reported for COF-10283 The
studies were performed using molecular dynamics with the force field parameters by Schmid et al84
However the thermal expansion coefficient α = -151 times 10-6 K-1 is much smaller compared to that of
some of the reported MOFs that also show NTE behavior85 The volume-contraction upon the
increase of temperature arises from the tilt of phenyl linkers between B3O3 rings and sp3 carbon
atom in TBPM (figure ) Later Schmid et alreported that COF-102 103 105 108 exhibit NTE
behavior both in ctn and bor topologies whereas COF-202 only in ctn topology79 A typical
application is the realization of controllable thermal expansion composites made of both negative
and positive thermal expansion materials
58
4 Applications
41 Gas storage
The success in the synthesis of COFs was certainly the result of a long-term struggle for complete
covalent crystallization The discovery of COFs coincided with the time when world-wide effort was
paid to develop new materials for gas storage in particular for the development hydrogen tanks for
mobile applications Materials made exclusively from light-weight atoms and with large surface
areas were obviously excellent candidates for these applications The gas storage capacity of porous
materials relies on the success of assembling gas molecules in minimum space This is achieved by
the interaction energy exerted by storage materials on the gas molecules Because the interactions
are noncovalent no significant activation is required for gas uptake and release and hence the so-
called physical adsorption or physisorption is reversible The other type of adsorption ndash chemical
adsorption or chemisorption ndash binds dissociated hydrogen covalently or interstitially at the risk of
losing reversibility As it requires the chemical modification of the host material chemisorption is not
a viable route for COFs and might become possible only through the introduction of reactive
components into the lattice The total amount of gas adsorbed in the pores gives rise to what is
referred to as lsquototal adsorption capacityrsquo This quantity is hard to be assessed experimentally as a
measurement is always subjected to influence of the materials surface and gas present in all parts of
the experimental setup Therefore the lsquoexcess adsorption capacityrsquo is reported in experiment Here
the gas stored in the free accessible volume is subtracted from the total adsorption In experiment
this volume includes the voids in the material as well as any empty space between the sample
crystals This volume is typically determined by measuring the uptake of inert gas molecules (He for
H2 adsorption N2 or Ar for adsorption studies of larger molecules) at room temperature with the
assumption that the host-guest interaction between the material and He can be neglected The
different definitions of adsorption is given in Figure 6
Typically experiments measure excess values and simulations provide total values Quantities of
adsorbed molecules are usually expressed as gravimetric and volumetric units The former gives the
amount per unit mass of the adsorbent while the latter gives the amount per unit volume of the
adsorbent Explicative definitions and terminologies related to gas adsorption can be found
elsewhere86
59
Figure 6 Hydrogen density profile in a pore A denotes the excess A+B the absolute and A+B+C the total adsorption The skeletal volume corresponds to the volume occupied by the framework atomsCourtesy of Dr Barbara Streppel and Dr Michael Hirscher MPI for intelligent systems Stuttgart Germany
411 Porosity of COFs
It is common to inspect the porosity of synthesized nanoporous materials using some inert or simple
gas adsorption measurements Distribution of pore size can be sketched from the
adsorptiondesorption isotherm N2 and Ar isotherms provide an approximation of the pore surface
area pore volume and pore size over the complete micro and mesopore size range Usually the
surface area is calculated using the Brunauer-Emmet-Teller (BET) equation or Langmuir equation
Total pore volume (volume of the pores in a pre-determined range of pore size) can be determined
from either the adsorption or desorption phase The Dubinin-Radushkevic method the T-plot
method or the Barrett-Joyner-Halenda (BJH) method may also be employed for calculating the pore
volume The pore size is often corroborated by fitting non-local density functional theory (NLDFT)
based cylindrical model to the isotherm90-93 In some cases the desorption isotherm is affected by
the pore network smaller pores with narrower channels remain filled when the pressure is lowered
This leads to hysteresis on the adsorption-desorption isotherms The different methods employed for
pore structure analysis are characteristic for micropore filling monolayer and multilayer formations
capillary condensation and capillary filling
For any adsorbate in order to form a layer on pore surface the temperature of the surface must
yield an energy (kBT) much less than adsorbate-surface interaction energy Additionally the absolute
value of the adsorbate-surface binding energy must be greater than the absolute value of the
adsorbate-adsorbate binding energy This avoids clustering of the adsorbate into its three-
dimensional phase
60
At high pressure the adsorption isotherm shows saturation which means that no more voids are left
for further occupation The isotherms show different behaviors characteristic of the pore structure of
the materials There are known classifications based on these differences type I II III IV and V For
COFs and the related materials discussed in this review type I II and IV have been observed (see
Figure 7) As more adsorbate is attracted to the surface after the first surface layer completion one
can expect a bend in the isotherm Type I implies monolayer formation which is typical of
microporosity If the surface sites have significantly different binding energies with the adsorbate
type II behavior is resulted In type IV each ldquokneerdquo where the isotherm bends over and the vapor
pressure corresponds to a different adsorption mechanisms (multiple layers different sites or pores)
and represents the formation of a new layer
Figure 7 Different types of adsorptions in Covalently-bound Organic Frameworks
Argon and nitrogen adsorption measurements at respectively 87 and 77 K exhibit type I isotherms
for COF-1 6 102 and 103 and type IV isotherms for COF-5 8 10 102 and 10387 Smaller pore
diameters of COF-1 and 6 (~ 9 Aring each) are characteristic of monolayer gas formation on the internal
pore surface The same reasons are responsible for the type I character of COF-102 and COF-103
(pore size = 12 Aring each) isotherms albeit with some unusual steps at low pressure The type IV
isotherms of COF-5 8 10 102 and 103 show formation of monolayers followed by their
multiplication within their relatively larger pores (sizes are approximately 27 16 32 12 and 12 Aring
respectively) Other COFs that exhibit type I isotherms for N2 adsorption are the 3D COFs COF-202 (11
Aring) COF-300 (72 Aring) and COF-102-C12 (12 Aring) the alkylated 2D COF series COF-18Aring COF-16Aring COF-14Aring
COF-11Aring (the nomenclature includes their pore sizes) and a 2D square COF NiPc COF (19 Aring)
Isotherms like Type-II were observed for PPy-COF (188 Aring) COF-366 (176 Aring) COF-66 (249 Aring) Pc-
PBBA COF (217 Aring) ZnP-COF (25 Aring) COF-LZU1 (12 Aring) PdCOF-LZU1 (12 Aring) and some of the azide-
appended COFs 50AcTrz-COF-5 (156 Aring) 75AcTrz-COF-5 (153 Aring) 100AcTrz-COF-5 (125 Aring)
50BuTrz-COF-5 (232 Aring) 50PhTrz-COF-5 (212 Aring) 50EsTrz-COF-5 (212 Aring) and 25PyTrz-COF-5
(221 Aring) The majority of the COFs with mesopores exhibit type-IV isotherms They are TP-COF (314
Aring) BTP-COF (40 Aring) COF-42 (23 Aring) COF-43 (38 Aring) HHTP-DPB COF (47 Aring) CTC-COF (226 Aring) NiPc-PBBA
COF ZnPc-Py-COF (31 Aring) ZnPc-DPB-COF (31 Aring) ZnPc-NDI-COF (31 Aring) ZnPc-PPE-COF (34 Aring) 5N3-
61
COF (295 Aring) 25N3-COF (276 Aring) 50N3-COF (259 Aring) 75N3-COF (258 Aring) 100N3-COF (227 Aring)
5AcTrz-COF-5 (298 Aring) 25AcTrz-COF-5 (243 Aring) 5PyTrz-COF-5 (289 Aring) and 10PyTrz-COF-5
(242 Aring)
The synthesized microporous amorphous materials CTF-1 (12 Aring) CTF-DCP (14 Aring) PAF-1 and PAF-2
(106 Aring) exhibit type-I isotherms with N2 adsorption PAF-3 (127 Aring) PAF-4(118 Aring) PAF-5 (157 Aring)
PAF-6 (118 Aring) and PAF-11 showed surprisingly not the typical type I behavior in spite of their
microporosity but type-II isotherms
Many of the mesoporous COFs showed small hysteresis in the adsorption-desorption isotherm
pointing the possibility of capillary condensation Hysteresis was observed for the amorphous
materials in both mirco and meso-pore range Various reasons have been attributed for the observed
hysteresis including the softness of the material and guest-host interactions
412 Gas adsorption experiments
Hydrogen uptake capacities of some boron-based COFs were measured by Yaghi et al87 The excess
gravimetric saturation capacities of COF-1 -5 -6 -8 -10 -102 and -103 at 77 K were found to be 148
358 226 350 392 724 and 705 mg g-1 respectively The saturation pressure increases with an
increase in pore size similar to MOFs88 Pore walls of 2D COFs consist only of edges of connectors
and linkers The fact that faces and edges are largely available for interactions with H2 in 3D
geometries is a reason for their enhanced capacity Total adsorption generally increases without
saturation upon pressure because the difference between the total and the excess capacities
corresponds to the situation in bulk (or in a tank) COFs show in particular good gravimetric
capacities because of their low mass density while volumetric capacities typically do not exceed
those of MOFs The gravimetric hydrogen storage capacities of COF-102 and -103 exceeds 10 wt at
a pressure of 100 bar
COF-18 and its alkylated versions COF-16 -14 and -1 have H2 excess gravimetric capacities 155 144
123 and 122 wt respectively at hellipK and hellipbar
Excess uptake of methane in COFs at room temperature and at 70 bar pressure is as follows COF-1
and -6 up to10 wt COF-5 -6 and -8 between 10 and 15 wt and COF-102 and -103 above 20
wt 87 Isosteric heat of adsorption Qst calculated by fitting the isotherms to virial expansion with
the same temperature are relatively weaker (lt 10 kJmol) for COF-102 and COF-103 at low
adsorption which implies weaker adsorbate-adsorbent interactions In comparison COF-1 and 6
exhibit twice the enthalpy however the overall hydrogen storage capacity was very limited due to
62
the smaller pore volume High Qst at zero-loading is not desirable because the primary excess amount
adsorbed at very low pressures cannot be desorbed practically89
COF-102 and 103 outperform 2D COFs for CO2 storage87 The saturation uptakes at room
temperature were 1200 and 1190 mg g-1 for COF-102 and 103 respectively
A type IV isotherm has been observed for ammonia adsorption in COF-10 inherent to its mesoporous
nature90 The material showed an uptake of 15 mol kg-1 at room temperature and 1 bar the highest
of any nanoporous materials The repeated adsorption-desorption cycles were reported to disrupt
the layer arrangement of this 2D COF Yaghi et al90 were also able to make a tablet of COF-10 crystal
which performed nearly up to the crystalline powder
Not many COFs have been experimentally studied for gas storage applications in spite of high
expectations This has to be understood together as a result of the powder-like polycrystallization of
COFs The enthalpy Qst at low-loading accounted to only 46 kJmol
The excess and total hydrogen uptake capacity of PAF-1 at 48 bar and 77 K reached 7 wt and 10
wt respectively65 PAF-3 and PAF-478_ENREF_73 follow the same diamond topology as PAF-1 the
difference accounts in the central atoms of the tetragonal blocks PAF-1 3 and 4 have C Si and Ge
atoms at the centers respectively At 1 bar the respective adsorption capacities were 166 207 and
150 wt The highest value being found for PAF-3 is attributed to its relatively high enthalpy (66 kJ
mol-1) compared to PAF-1 (54 kJ mol-1) and PAF- 4 (66 kJ mol-1) At high pressure the surface area is
a key factor and because of the lower BET surface area PAF-3 and PAF-4 performs poor At 60 bar
their respective adsorption capacities were 55 and 42 wt For PAF-11 hydrogen uptake was 103
wt at 1 bar68
Methane uptakes of PAF-1 3 and 4 at 1 bar are 13 19 and 13 wt respectively76 Their enthalpies
are respectively 14 15 and 232 kJ mol-1 This suggests that Ge moiety have strong interactions with
methane
CO2 adsorption in PAF-1 at 40 bar and room temperature was 1300 mg g-1 which was slightly more
than the capacity of COF-102 and 10365 PAF-1 3 and 4 at 1 atm pressure could store 48 80 and 51
wt respectively At 273 K and 1 atm they stored 91 153 and 107 wt respectively The storage
capacities at 273 K and 298 K lead to the following zero-loading isosteric enthalpies 156 192 162
kJ mol-1 for PAF-1 3 and 4 respectively The higher uptake of PAF-3 may also be explained by its
relatively higher surface area with CO2 molecules
The selective adsorption of greenhouse gases in PAFs was discussed by Ben et al76 At 273 K and 1
atm pressure PAF-1 selectively adsorbes CO2 and CH4 respectively 38 and 15 times more in
63
amounts than N2 PAF-3 showed extraordinary selectivity of 871 for CO2 over N2 and 301 for CH4
over N2 For PAF-4 the selectivity was 441 for CO2 over N2 and 151 for CH4 over N2 Generally the
uptake of N2 Ar H2 O2 are significantly lower compared to CO2 and NH4 which makes these PAFs
suitable for separating them
Harmful gases like benzene and toluene can be stored in PAF-1 in amounts 1306 mg g-1 (1674 mmol
g-1) and 1357 mg g-1 (1473 mmol g-1) respectively at saturated pressures and room temperature65
In PAF-11 their adsorptions were in amounts of 874 mg g-1 and 780 mg g-1 respectively68 PAF-2 was
accounted to a much lower benzene adsorption (138 mg g-1) at the same conditions59 Adsorption of
cyclohexane vapor in PAF-2 was only 7 mg g-1 While PAF-11 exhibited hydrophobicity it could
accommodate significant amount of methanol (654 mg g-1 or 204 mmol g-1) at room temperature
and saturated pressure68 PAF-5 has been tested for storing organic pollutants69 At room
temperature and saturated pressure PAF-5 adsorbed methanol benzene and toluene in amounts
6531 cm3 g-1 (292 mmol g-1) 26296 cm3 g-1 (117 mmol g-1) and 2582 cm3 g-1 (115 mmol g-1)
respectively Their adsorptions in PAF-5 were deviated from the typical type-I behavior Methanol
exhibited a large slope at the high pressure region corresponding to the relative size of it Zhu G et
al dedicated PAF-6 for biomedical applications With no toxicity observed for PAF-6 over a range of
concentrations adsorption of ibuprofen was measured in it PAF-6 showed an uptake of 350 mg g-1
The release of ibuprofen was also tested in simulated body fluid which showed a release rate of 50
in 5 hours 75 in 10 hours and 100 in almost 46 hours
413 Theoretical predictions
Grand canonical Monte Carlo (GCMC) is a widely used method for the simulation of gas uptakes in
nanoporous materials In GCMC the number of adsorbates and their motions are allowed to change
at constant volume temperature and chemical potential to equilibrate the adsorbate phase The
motions are random guided by Monte Carlo methods and the energies are calculated by force field
methods The details of it may be found in the literature91
Goddart III WA et al92 simulated the uptake capacities of boron-based COFs using force fields derived
from the interactions of H2 and COFs calculated at MP2 level of theory The saturated excess uptakes
of both COF-105 (at 80 bar) and 108 (at 100 bar) were accounted to 10 wt at 77 K which are more
than the uptakes measured in ultrahigh porous MOF-200 205 and 21093 The predictions for other
COFs include 38 wt for COF-1 (AB stacking) 34 wt for COF-5 (AA stacking) 88 wt for COF-102
and 91 wt for COF-103 At 01 bar COF-1 in AB stacking showed highest excess uptake (17 wt )
compared to other COFs in the present discussion Total uptake capacities of the COFs are in the
following order COF-108 (189 wt ) COF-105 (183 wt ) COF-103 (113 wt ) COF-102 (106
64
wt ) COF-5 (55 wt ) and COF-1 (38 wt ) It is worth noticing the higher gravimetric capacities of
COF-108 and 105 which were not measured experimentally They benefit from their lower mass and
higher pore volume In case of volumetric capacity COF-102 (404 gL excess) surpasses the others at
high pressures The total volumetric capacities of the COFs are 499 498 399 395 361 and 338
gL for COF-102 103 108 105 1 and 5 respectively Their additional calculations predicted benzene
rings as favorite locations for H2 molecules
Similar values and trends have been reported by Garberoglio G94 who also reasoned poor solid-fluid
interactions in a large part of free volume for the low volumetric capacity of COF-108 and 105 A
room temperature excess gravimetric uptake of approximately 08 wt was obtained for COF-108
and 105 Quantum diffraction effects were added to the interaction energies of H2 with itself and the
material which were calculated using universal force-field (UFF) With possible overestimation
Klontzas et al30 and Bassem et al95 have reported total gravimetric uptake of 45 and 417 wt
respectively at 300 K for the same COFs Among the boron-based 3D COFs COF-202 exhibited much
smaller uptake capacity at high pressures due to its smaller pore volume95 Lan et al96 predicted a
very small H2 uptake of 152 wt COF-202 at 300K and 100 bar Studies of Yang et al97 clarifies that
pore filling with H2 in 2D COFs is a gradual process rather than being in steps of layer formation
Garberoglio and Vallauri98 pointed out that the relatively higher mass density and lower surface area
per unit volume of 2D COFs are the reasons for their lower uptakes in comparison with 3D COFs The
surface area of hexagonal 2D COFs (AA stacking) averaged around 1000 m2 cm-3 while that of 3D
COFs were about 1500 m2 cm-3
Simulations of H2 adsorption in the perfect crystalline form of PAF-1 and PAF-11 also known as PAF-
302 and PAF-304 respectively are carried out by Zhu et al99 At 77 K H2 excess uptake of PAF-302 (7
wt ) is slightly higher than COF-102 (684 wt )87 and slightly lower than MOF-177 (740 wt )100 At
room temperature PAF-302 exhibited a poor uptake (221 wt at 100 bar) PAF-304 showed
excellent total uptakes 2238 wt at 77 K and 100 bar 653 wt at 298 K and 100 bar These are
highest among all nanoporous materials
Babarao and Jiang studied room-temperature CO2 adsorption in COFs101 At low pressures COFs with
smaller pore volume exhibit a steep rise in their isotherm while at high pressures COF-105 and 108
(82 and 96 mmol g-1 respectively at 30 bar) dominate the others COF-6 has the highest isosteric heat
of adsoption (328 kJmol) hence it made the first steep rise in the volumetric isotherm followed by
COF-8 102 103 10 105 and 108 Correlations of gravimetric and volumetric capacities with mass
density pore volume porosity and surface area have been excellently manifested in this article101
With increasing framework-density gravimetric uptake falls inversely while volumetric capacity
decreases linearly The former rises with free volume while the latter rises and then drops slightly
65
Both of them rise with porosity and surface area Choi et al102 predicted 45 times more CO2 uptake in
COF-108 (3D) compared to COF-5 (2D) Recently Schmid et al79 also have predicted CO2 adsorption
especially for 3D COFs The uptake of COF-202 was almost half of COF-102 and 103 at room
temperature Type IV isotherm has been found for CO2 adsorption in COF-8 and 10 at low
temperatures (lt 200 K)97 Yang and Zhong97 excluded capillary condensation and dissimilar
adsorption sites but multilayer formation as reasons of the stepped behavior (Yang and Zhong
explained this as a consequence of multilayer formation rather than a result of capillary
condensation or dissimilar adsorption sites)
Methane adsorption in 2D and 3D COFs was first calculated by Garberoglio et al Among COF-6 8 and
10 the former which has smaller pore size and higher binding energy with CH4 shows better
volumetric adsorption (~ 120 cm3cm3 at 20 bar) than the others at room temperature at low
pressure region (lt 25 bar) Larger pore size favors COF-10 for higher volumetric adsorption (~ 160
cm3cm3 at 70 bar) at high pressures COF-8 with intermediate pore size shows largest excess amount
of methane adsorbed upon saturation however still lower than two 3D COFs COF-102 and 103
show excess volumetric adsorption of about 180 cm3cm3 at 35 bar This is significantly higher than
the uptake of two other 3D COFs COF-105 and 108 Entropic effects which primarily arise from the
change of dimensionality when the bulk phase of adsorbate transforms to adsorbed phase are
crucial in adsorption Garberoglio94 shows that solid-fluid interaction potential in large part of volume
of COF-105 and 108 is not strong enough to overcome the entropic loss On the other hand these
two COFs exhibit relatively higher gravimetric uptake (720 and 670 cm3g respectively) Goddard III et
al89 predicted total methane uptakes in the following amounts 415 wt in COF108 405 wt in
COF-105 31 wt in COF-103 284 wt in COF-102 196 wt in COF-10(AA) 169 wt in COF-
5(AA) 159 wt in COF-8(AA) 123 wt in COF-6(AA) and 109 wt in COF-1(AB) Yang and Zhong97
have shown that even at low temperatures (100 K) the pore filling of COF-8 with CH4 is rather
gradual than by step-by-step whereas for COF-10 multilayer formation provides a stepped behavior
in the isotherm Alternately Goddard et al pointed out that layer formation coexists with pore-filling
at room temperature89
414 Adsorption sites
First principle calculations on cluster models are typically employed to find favorite adsorption sites
and binding energies of adsorbates within frameworks Benzene rings are found to be a usual
location for H2 where it prefers to orient in perpendicular fashion3086110 Other favorite locations
include B3O3 and C2O2B rings110111 where H2 orients horizontally on the face as well as on the
edge3086 The central ring of HHTP does not provide enough binding to H2 atoms because of small
amount of charges There are some differences in the adsorption energies and favorite sites
66
calculated at different levels of theory Overall the reported binding energies between H2 and any
COF fragment are less than 6 kJ mol-1 In comparison to MOFs COFs do not offer any stronger binding
energy with H2 The advantage of COFs is their low mass density Lan et al103 reported that CO2 is
more preferred to be adsorbed on B3O3 or C2O2B rings than benzene rings Choi et al102 reported that
the oxygen sites are the primary locations for CO2 and CO2 bound COFs are the second strong binding
sites
415 Hydrogen storage by spillover
Hydrogen spillover is an alternate way to store hydrogen113114 It involves dissociation of hydrogen
gas by supported metal catalysts subsequent migration of atomic hydrogen through the support
material and final adsorption in a sorbent Here surface-diffusion of H atoms over the support is
known as primary spillover Secondary spillover is the further diffusion to the sorbent Bridging the
metal part with the sorbent is a practice to enhance the uptake It increases the contact between the
source and receptor and reduces the energy barriers especially in the secondary spillover As the
final sorption is chemisorption surface area of the sorbent is more important than pore volume
Yang and Li measured H2 storage in COF-1 by spillover They used Pt supported on active carbon
(PtAC) as source for hydrogen dissociation Active carbon was the primary receptor and COF-1 the
secondary receptor COF-1 showed much low uptake 128 wt at 77 K and 1 atm 026 wt at 298
K and 100 bar In comparison to MOFs these are very low104 However they have found that the
uptake could be scaled up by a factor of 26 by bridging COF-1 with PtAC by applying carbonization
They also report that heat of adsorption between H and surface sites is more important than surface
area and pore volume in enhancing the net adsorption by spillover
Ganz et al theoretically predicted that the uptake capacities in COFs by hydrogen spillover can be
higher than the measured value116117 Based on ab initio quantum chemistry calculations and
counting the active storage sites they predicted 45 wt for COF-1 in AB stacking and 55 wt for
COF-5 in AA stacking at room temperature and 100 bar
42 Diffusion and Selectivity
Computed self-diffusion coefficients for H2 and CH4 adsorbates in 2D COFs (in AA stacking) are up to
one order of magnitude higher than that in 3D MOFs such as MOF-598 In comparison to nanotubes
the diffusion of H2 molecules in 2D COFs is one order of magnitude slower The differences in
diffusion coefficients are attributed to different pore structures Interactions of the corners of the
hexagonal pore with the fluid present potential barriers along the cylindrical pore axis Diffusion
occurs with jumps after long waiting The height of the barrier is still smaller than in MOFs
67
Homogeneous pore walls and absence of pore corners in nanotubes contribute much less
corrugation to the solid-fluid potential energy surface Increase of self-diffusivities of H2 and CH4 with
increase in pore size of 2D COFs is also shown by Keskin[Ref] Due to the smaller size of H2 its
diffusivity is higher than that of CH4 in any material For reasons of their relative size In a mixture of
the two the self-diffusivity of CH4 increases while that of H2 decreases
Molecular dynamics simulation studies of the diffusion of gas molecules within a 2D COF performed
by Krishna et al105 revealed its relatively lower binding energy of hydrogen molecules within the pore
walls compared to argon CO2 benzene ethane propane n-butane n-pentane and n-hexane
Binding energy prevents the molecules from diffusing through the pore channels They tested if
Knudsen diffusion is applicable to COFs Knudsen diffusion occurs when the molecules frequently
collide with the pore wall This generally happens when the mean free path is larger than the pore
diameter The simulations had been carried out in BTP-COF which accounts a pore diameter of 4 nm
It is found out that the ratio of zero-loading diffusivity with the Knudsen diffusivity has a significant
correlation with isosteric heat of adsorption the stronger the binding energy of the molecules with
the walls the lower the ratio Hydrogen being an exception among the investigated molecules
exhibited the ratio close to unity while others with their isosteric heat of adsorption higher than 10
kJmol-1 have a value close to 01 For a mixture of two gases with significant differences in binding
energies the ratio of self-diffusivities affirms high diffusion selectivity
Selectivity of CH4 over H2 in COF-5 6 and 10 is studied by Keskin106 Narrow pores support the
selective adsorption of CH4 over H2 however they suffer from entropic effects at high pressures
which favor H2 adsorption Liu et al107 have reported that the adsorption selectivity of COFs and
MOFs are in general similar up to 20 bar Delta loading or working capacity (usually expressed in
molkg) is an important term often used to show the economics of the selective adsorption This is
defined as the difference in loadings of the preferred gas at adsorption and desorption pressures
Adsorption is usually in the range 1 to 100 bar while desorption in 01 to 1 bar High selectivity and
high working capacity are preferential for practical use COF-6 has higher selectivity among the three
studied COFs but lower working capacity106 Best selectivity-working capacity combination is shown
by COF-5106 Nonetheless it is not better than CuBTC or CPO-27-X (X = Zn Mg)121122 Liu et al107
attributed the high isosteric heat of adsorption of COF-6 and Cu-BTC for their high adsorption
selectivity They also pointed out that the electrostatic contribution of framework charges in COFs
are smaller than in MOFs however needs to be taken into account
While CH4 is strongly adsorbed H2 undergoes faster diffusion Hence the product of adsorption
selectivity (gt1) and diffusion selectivity (gt0 lt1) which is membrane selectivity is smaller than
adsorption selectivity Keskin106 has compared the selectivity of COF-membranes with known
68
membranes The selectivity of COF-10 is similar to MOF-177 and IRMOFs COF-5 and 6 outperform
them It is to be noted that COF-5 is CH4 selective while COF-10 is H2 selective although their
topologies are similar CH4 permeability of COF membranes is much higher than several MOFs and
ZIFs COF-5 and 6 exhibited high membrane selectivity together with high membrane permeability
Some discussions on the adsorption and membrane based selectivity of COF-102 in comparison with
IRMOFs and Cu-BTC for CH4CO2H2 mixtures can be found in ref108 Adsorption selectivity of COF-6
and Cu-BTC in CH4H2 and CH4CO2 mixtures surpass other COFs and MOFs107 sdfalf
43 Suggestions for improvement
The level of achievement made by COFs and related materials yet do not practically meet the
practical requirements of several applications Hence thoughts for improvement primarily focused
on overcoming their limitations and enhancing characteristic parameters Some theoretical
suggestions for improved performances are mainly discussed here
431 Geometric modifications
Functionalities may be improved by utilizing the structural divergence of framework materials
Several examples may be found in the literature for MOFs etc Klontzas et al suggested replacement
of phenyl linkers in COF-102 by multiple aromatic moieties while keeping the topology unchanged to
increase surface area and pore volume109 They used biphenyl triphenyl naphthalene and pyrene
linkers in the new COFs All of them showed enhanced gravimetric capacity compared to the parent
COF One of the modified COF-102 with triphenyl linker showed 267 and 65 wt at 77 K and 300 K
respectively at 100 bar This kind of replacement may affect the adsorption of each adsorbate
differently leading to the alteration of the selective adsorption of one component over the other110
Following the reticular construction scheme we modeled some new 2D COFs by cross-linking some
of the familiar and unfamiliar building blocks in the COF literature32 This showed the structural
divergence of COFs however they exhibited structural and electronic properties analogues to other
2D COFs
Similar modifications in 2D and 3D COFs for high CO2 uptake were made by Choi et al102 DPABA
(diphenylacetylene-44ʹ-diboronic acid) and its tetrahedral version TBPEPM (tetra[4-(4-
dihydroxyborylphenyl)ethynyl]phenyl) in place of BDBA in COF-5 and TBPMTBPS in COF-108COF-
105 respectively were used for new COF designs The new 2D COF promised a saturated CO2 uptake
higher than in COF-102 because of its large pore volume The new 3D COFs were worth for an uptake
twice more than in COF-105 and 108
69
Goddard et al also designed about 14 new COFs by substituting the aromatic rings in the tetragonal
part (TBPM or TBPS) of COF-102 COF-103 COF-105 COF-108 and COF-202 with vinyl groups or alkyl-
functionalized extended or fused aromatic rings111 The new designs adopted their parent topology
and their structural stability was confirmed by analyzing molecular dynamics (MD) simulations at
room temperature Fused aromatic rings and vinyl groups provided relatively lower and the highest
zero-loading isosteric heat of adsorption (Qst) respectively however the room-temperature delivery
amounts of methane in the respective COFs crossed the DOE target at 35 bar111 In the latter
methane-methane interaction compensated Qst on high-loading
Lino M A et al112 theoretically inspected the ability of layered COFs to exhibit structural variants of
layered materials The hexagonal 2D COF layers may be rolled into the form of nanotubes (NT) or
may exist in the form of fullerenes with the inclusion of pentagons One could think of a formula unit
which resembles the carbon atoms in carbon nanotubes (CNT) or fullerenes The NTs in general can
have any chirality although the study included only armchair and zigzag NTs Density Functional
Theory based calculations reveal that COF-1 NTs of diameter greater than 15 Aring is energetically
favored This was concluded from the comparison of strain-energy per formula unit of the COF-1 NTs
with that of small diameter CNTs The strain energies of zigzag and armchair nanotubes show similar
quadratic functions of tube-diameter The Youngrsquos modulus of COF-1 (22) NT is found out to be 120
GPa This is much smaller compared to typical CNTs which have Youngrsquos modulus average around
1100 GPa113 Band gaps of COF layers remain unchanged when they roll up into nanotubes A COF-1-
fullerene built by scaling C60 molecule has large diameter and hence much less strain energy
compared to C60 Babarao and Jiang reported that the predicted CO2 adsorption capacity of COF-1 NT
is similar to that of CNTs101
Balance between mass density and surface area and hence high gravimetric and volumetric
capacities can be achieved by proper utilization of pore volume or proper distribution of pores Choi
et al theoretically predicted high gravimetric and volumetric capacities for H2 in a pillared COF114 A
pyridine molecule vertically placed above a boron atom in the boroxine ring of COF-1 layer is found
energetically favorable however with wrinkling of the layer115 Here nitrogen atom in pyridine forms
a covalent bond with the boron atom This pillaring increases the separation between the layers
exposes the buried faces of boroxine and aromatic rings to pores and hence increases surface area
and free volume Accessible surface area and free volume have been tripled and gravimetric and
volumetric capacities were calculated to be 10 wt and 60 g L-1 respectively at 77 K and 100 bar114
This volumetric capacity is higher than in NU-100116 and MOF-21093 which show ultrahigh surface
area
70
432 Metal doping
Miao Wu et al studied doping of COF-10 with Li and Ca atoms for hydrogen storage117 The metal
dopants transferred charges to substrate which in turn provided large polarization to hydrogen
molecules Similar observation has been made by Lan et al96 for Li atoms in COF-202 However they
showed the tendency to aggregate at high concentration Lan et al extensively studied doping of
positively charged alkali (Li Na and K) alkaline-earth (Be Mg Ca) and transition (Sc and Ti) metals in
COF-102 and 105 for enhanced CO2 capture103 They found that benzene rings and the three outer
rings of HHTP are the favorite sites for all the metal dopants Other interested locations are edges of
benzene rings and oxygen atoms B3O3 C2O2B rings and the central ring of HHTP were the unwanted
areas Lithium showed stability on the favorite locations while sodium and potassium tended to
cluster even at low concentrations Alkaline-earth metals only weakly interacted with the COFs
whereas transition metals chemisorbed in them which would possibly damage the substrate Lithium
is found out to be a good dopant for enhanced gas storage
Doping electropositive metals would be of advantage because they provide stronger binding with H2
(~ 4 kcalmol)103125133134 The effect of counterions in COFs is studied by Choi et al118 They found out
that although the binding energy of LiF is smaller than Li ion F counterion can hold one hydrogen
atom Additionally Li+ and Mg2+ ions preferred to be isolated than aggregated A step further
Goddard et al119 have recently shown that in presence of tetrahydrofuran (THF a solvent) an
electron is transferred from Li to the benzene ring whereas in the case of gas phase the electron
remained in the atom Additionally they suggested ways to remove solvents which are weakly
coordinated to the material Zhu et al110 suggested doping Li+ after replacement of hydrogen atom by
oxygen ion in COF-102 Another suggestion was the replacement of hydrogen atom by O--Li+ group
Both the cases exhibited higher CO2 adsorption than in the case of Li doping below 10 bar At 10 bar
the three cases exhibited almost the same uptake capacity Below 1 bar the doped Li+ provided
stronger interaction with quadrupolar CO2 in comparison with the substituted O--Li+ group The
differences at low pressures are attributed to the differences in the magnitude of the charge of Li
The doped Li+ remained to hold nearly +1 charge whereas the inclusion of O--Li+ to the framework
diminished the charge of Li+ to a moderate amount Doping Li atom added only relatively small
amount of charge to Li
Doping with Li+ could increase the H2 binding energies up to 28 kJmol118 With properly distributed
metal ion dopants the H2 uptake capacity in COF-108 is predicted to be 65 wt Cao et al also
predicted 684 and 673 wt of H2 uptake in Li-doped COF-105 and 108 respectively at room
temperature and 100 bar120 In Li-doped COF-202 the predicted uptake was 438 wt for the same
conditions96 Goddard et al119 observed that high performance of Li doped COFsMOFs at low
71
pressures (1 ndash 10 bar) is a disadvantage when calculating delivery uptake capacity Na doping could
overcome this difficulty as its heat of adsorption is lower than in Li doped materials Their predicted
delivery gravimetric capacities from 1 to 100 bar at room temperature for Li and Na doped COF-102
and 103 were higher than the 2010 DOE target of 45 wt at room temperature
Lan et al described that CO2 is polarized in presence of Li cation and the polarization increases when
Li cation is physisorbed in COF103 Their calculated uptake capacities of lithium-doped COF-102 and
COF-105 at low loading is about eight and four times higher than that of nondoped ones respectively
Also a very high uptake of 2266 mgg was predicted in the doped COF-105 at 40 bar Li-doped COF-
102 and COF-103 also showed higher uptakes of methane ndash almost double the amount ndash compared
to nondoped COFs at room temperature and low pressures (lt 50 bar)121 Binding energy between
doped Li cation and CH4 was calculated to be 571 kcalmol
Li-functionalized COF-105 realized by inserting alkoxide groups provided very high volumetric uptake
of H2 Klontzas et al122 replaced three hydrogen atoms in HHTP by oxygen atoms in order to achieve
the functionalization In spite of the increased weight due to the additional oxygen atoms the COF
exhibited gravimetric capacity of 6 wt at 300 K
Incorporation of lithium tetrazolide group into a new PAF having diamond topology and triphenyl
linkers for enhanced hydrogen adsorption is suggested by Sun et al123 The tetrazolide anion (CHN4-)
interacts with the lithium cation via ionic bonds The anion and the cation together can bind upto 14
hydrogen molecules Two lithium tetrazolide groups were introduced in the middle phenyl ring of
each triphenyl linker and GCMC calculations predicted 45 wt for the network at 233 K and 100 bar
With the intention of increasing the H2 binding energy Li et al chemically implanted metals in the
place of light atoms in COF-108124 They replaced the C2O2B rings in COF-108 by C2O2Al C2N2Al and
C2Mg2N and each new COF showed stability within DFT This kind of doping does not allow
aggregation of metal clusters Their GCMC simulations showed that the replacement strategy could
improve the hydrogen storage capacity at room temperature by a factor of up to three124 Zou et al
suggested that para-substitution of two carbon atoms in benzene rings by two boron atoms can
facilitate charge transfer between the rings and metal dopants125 Their work revealed that
dispersion of metals forbid any clustering and the doping could enhance the H2 uptake capacity
significantly
433 Functionalization
Functionalization of porous frameworks with ether groups for selective CO2 capture is suggested by
Babarao et al126 The electropositive C atom of CO2 interacts mainly with the electronegative O or N
72
atom of the functional groups They studied PAF-1 functionalized with ndashOCH3 ndashNH2 and ndashCH2OCH2ndash
groups in its biphenyl linkers The latter one which is the tetrahydrofuran-like ether-functionalized
PAF-1 showed high adsorption capacity for CO2 and high selectivity for CO2CH4 CO2N2 and CO2H2
mixtures at ambient conditions
5 Conclusions
Successful covalent crystallization resulted as a class of materials Covalent Organic Frameworks This
review portrays different synthetic schemes successful realizations and potential applications of
COFs and related materials The growth in this area is relatively slow and thus promotions are
needed in order to achieve its potential
6 List and pictures of chemical compounds
alkyl-THB Alkyl-1245-tetrahydroxybenzene
BDBA 14-benzenediboronic acid
BPDA 44ʹ-biphenyldiboronic acid
BTBA 135-benzene triboronic acid
BTDADA 14-benzothiadiazole diboronic acid
BTPA 135-benzenetris(4-phenylboronic acid)
CA Cyanuric acid
DAB 14-diaminobenzene
DCB 14-dicyanobenzene
DCF 27-diisocyanate fluorine
DCN 26-dicyanonaphthalene
DCP 26-dicyanopyridine
DETH 25-diethoxyterephthalohydrazole
DMBPDC 33ʹ-dimethoxy-44ʹ-biphenylene diisocyanate
DPB Diphenyl butadyenediboronic acid
73
HP 1-hexyne propiolate
HHTP 23671011-hexahydroxytriphenylene
MP Methyl propiolate
N3-BDBA Azide-appended benzenediboronic acid
NDI Naphthalenediimide diboronic acid
NiPcTA Nickel-phthalocyanice tetrakis(acetonide)
OH-Pc-Ni 2391016172324-octahydroxyphthalocyaninato)nickel(II)
OH-Pc-Zn 2391016172324-octahydroxyphthalocyaninato)zinc
PA Piperazine
Pac 2-propenyl acetate
PcTA Phthalocyanine tetra(acetonide)
PdAc Palladium acetate
PDBA Pyrenediboronic acid
PPE Phenylbis(phenylethynyl) diboronic acid
PPP 3-phenyl-1-propyne propiolate
PyMP (3α13α2-dihydropyren-1-yl)methyl propionate
TA Terephthaldehyde
TAM tetra-(4-anilyl)methane
TAPP Tetra(p-amino-phneyl)porphyrin
TBB 135-tris(4-bromophenyl)benzene
TBPM tetra(4-dihydroxyboryl-phenyl)methane
TBPP Tetra(p-boronic acid-phenyl)porphyrin
TBPS tetra(4-dihydroxyboryl-phenyl)silane
TBST tert-butylsilane triol
74
TCM Tetrakis(4-cyanophenyl)methane
TDHB-ZnP Zinc(II) 5101520-tetrakis(4-(dihydroxyboryl)phenyl)porphyrin
TFB 135-triformylbenzene
TFPB 135-tris-(4-formyl-phenyl)-benzene
THAn 2345-Tetrahydroxy anthracene
THB 1245-tetrahydroxybenzene
THDMA Polyol 2367-tetrahydroxy-910-dimethyl-anthracene
TkBPM Tetrakis(4-bromophenyl)methane
TPTA Triphenylene tris(acetonide)
trunc-TBPM-R R-functionalized truncated TBPM
75
Figure 8 Reactants of Covalently-bound Organic Frameworks
76
Figure 9 (Figure 8 continued)
(1) Morris R E Wheatley P S Angewandte Chemie-International Edition 2008 47 4966 (2) Li J-R Kuppler R J Zhou H-C Chemical Society Reviews 2009 38 1477 (3) Corma A Chemical Reviews 1997 97 2373 (4) Seo J S Whang D Lee H Jun S I Oh J Jeon Y J Kim K Nature 2000 404 982 (5) Yaghi O M OKeeffe M Ockwig N W Chae H K Eddaoudi M Kim J Nature 2003 423 705
77
(6) OKeeffe M Peskov M A Ramsden S J Yaghi O M Accounts of Chemical Research 2008 41 1782 (7) Ockwig N W Delgado-Friedrichs O OKeeffe M Yaghi O M Accounts of Chemical Research 2005 38 176 (8) Li H Eddaoudi M OKeeffe M Yaghi O M Nature 1999 402 276 (9) Chen B L Eddaoudi M Hyde S T OKeeffe M Yaghi O M Science 2001 291 1021 (10) Eddaoudi M Moler D B Li H L Chen B L Reineke T M OKeeffe M Yaghi O M Accounts of Chemical Research 2001 34 319 (11) Eddaoudi M Kim J Rosi N Vodak D Wachter J OKeeffe M Yaghi O M Science 2002 295 469 (12) Chae H K Siberio-Perez D Y Kim J Go Y Eddaoudi M Matzger A J OKeeffe M Yaghi O M Nature 2004 427 523 (13) Furukawa H Kim J Ockwig N W OKeeffe M Yaghi O M Journal of the American Chemical Society 2008 130 11650 (14) Smaldone R A Forgan R S Furukawa H Gassensmith J J Slawin A M Z Yaghi O M Stoddart J F Angewandte Chemie-International Edition 2010 49 8630 (15) Eddaoudi M Kim J Wachter J B Chae H K OKeeffe M Yaghi O M Journal of the American Chemical Society 2001 123 4368 (16) Sudik A C Millward A R Ockwig N W Cote A P Kim J Yaghi O M Journal of the American Chemical Society 2005 127 7110 (17) Sudik A C Cote A P Wong-Foy A G OKeeffe M Yaghi O M Angewandte Chemie-International Edition 2006 45 2528 (18) Tranchemontagne D J L Ni Z OKeeffe M Yaghi O M Angewandte Chemie-International Edition 2008 47 5136 (19) Lu Z Knobler C B Furukawa H Wang B Liu G Yaghi O M Journal of the American Chemical Society 2009 131 12532 (20) Park K S Ni Z Cote A P Choi J Y Huang R Uribe-Romo F J Chae H K OKeeffe M Yaghi O M Proceedings of the National Academy of Sciences of the United States of America 2006 103 10186 (21) Hayashi H Cote A P Furukawa H OKeeffe M Yaghi O M Nature Materials 2007 6 501 (22) Banerjee R Furukawa H Britt D Knobler C OKeeffe M Yaghi O M Journal of the American Chemical Society 2009 131 3875 (23) Cote A P Benin A I Ockwig N W OKeeffe M Matzger A J Yaghi O M Science 2005 310 1166 (24) El-Kaderi H M Hunt J R Mendoza-Cortes J L Cote A P Taylor R E OKeeffe M Yaghi O M Science 2007 316 268 (25) Wan S Guo J Kim J Ihee H Jiang D Angewandte Chemie-International Edition 2008 47 8826 (26) Wan S Guo J Kim J Ihee H Jiang D L Angewandte Chemie-International Edition 2009 48 5439 (27) Cote A P El-Kaderi H M Furukawa H Hunt J R Yaghi O M Journal of the American Chemical Society 2007 129 12914 (28) Hunt J R Doonan C J LeVangie J D Cote A P Yaghi O M Journal of the American Chemical Society 2008 130 11872 (29) Uribe-Romo F J Hunt J R Furukawa H Klock C OKeeffe M Yaghi O M Journal of the American Chemical Society 2009 131 4570 (30) Klontzas E Tylianakis E Froudakis G E Journal of Physical Chemistry C 2008 112 9095 (31) Tylianakis E Klontzas E Froudakis G E Nanotechnology 2009 20 (32) Lukose B Kuc A Frenzel J Heine T Beilstein Journal of Nanotechnology 2010 1 60
78
(33) Tilford R W Gemmill W R zur Loye H C Lavigne J J Chemistry of Materials 2006 18 5296 (34) Spitler E L Dichtel W R Nature Chemistry 2010 2 672 (35) Spitler E L Giovino M R White S L Dichtel W R Chemical Science 2011 2 1588 (36) Campbell N L Clowes R Ritchie L K Cooper A I Chemistry of Materials 2009 21 204 (37) Ding X Guo J Feng X Honsho Y Guo J Seki S Maitarad P Saeki A Nagase S Jiang D Angewandte Chemie-International Edition 2011 50 1289 (38) Feng X A Chen L Dong Y P Jiang D L Chemical Communications 2011 47 1979 (39) Zwaneveld N A A Pawlak R Abel M Catalin D Gigmes D Bertin D Porte L Journal of the American Chemical Society 2008 130 6678 (40) Gutzler R Walch H Eder G Kloft S Heckl W M Lackinger M Chemical Communications 2009 4456 (41) Blunt M O Russell J C Champness N R Beton P H Chemical Communications 2010 46 7157 (42) Sassi M Oison V Debierre J-M Humbel S Chemphyschem 2009 10 2480 (43) Ourdjini O Pawlak R Abel M Clair S Chen L Bergeon N Sassi M Oison V Debierre J-M Coratger R Porte L Physical Review B 2011 84 (44) Colson J W Woll A R Mukherjee A Levendorf M P Spitler E L Shields V B Spencer M G Park J Dichtel W R Science 2011 332 228 (45) Berlanga I Ruiz-Gonzalez M L Gonzalez-Calbet J M Fierro J L G Mas-Balleste R Zamora F Small 2011 7 1207 (46) Wan S Gandara F Asano A Furukawa H Saeki A Dey S K Liao L Ambrogio M W Botros Y Y Duan X Seki S Stoddart J F Yaghi O M Chemistry of Materials 2011 23 4094 (47) Ding X Chen L Honsho Y Feng X Saenpawang O Guo J Saeki A Seki S Irle S Nagase S Parasuk V Jiang D Journal of the American Chemical Society 2011 133 14510 (48) Tilford R W Mugavero S J Pellechia P J Lavigne J J Advanced Materials 2008 20 2741 (49) Lanni L M Tilford R W Bharathy M Lavigne J J Journal of the American Chemical Society 2011 133 13975 (50) Li Y Yang R T Aiche Journal 2008 54 269 (51) Nagai A Guo Z Feng X Jin S Chen X Ding X Jiang D Nature Communications 2011 2 (52) Bunck D N Dichtel W R Angewandte Chemie-International Edition 2012 51 1885 (53) Ding S-Y Gao J Wang Q Zhang Y Song W-G Su C-Y Wang W Journal of the American Chemical Society 2011 133 19816 (54) Miyaura N Suzuki A Chemical Reviews 1995 95 2457 (55) Kalidindi S B Yusenko K Fischer R A Chemical Communications 2011 47 8506 (56) Kuhn P Antonietti M Thomas A Angewandte Chemie-International Edition 2008 47 3450 (57) Bojdys M J Jeromenok J Thomas A Antonietti M Advanced Materials 2010 22 2202 (58) Kuhn P Forget A Su D Thomas A Antonietti M Journal of the American Chemical Society 2008 130 13333 (59) Ren H Ben T Wang E Jing X Xue M Liu B Cui Y Qiu S Zhu G Chemical Communications 2010 46 291 (60) Zhang W Li C Yuan Y-P Qiu L-G Xie A-J Shen Y-H Zhu J-F Journal of Materials Chemistry 2010 20 6413 (61) Trewin A Cooper A I Angewandte Chemie-International Edition 2010 49 1533 (62) Mastalerz M Angewandte Chemie-International Edition 2008 47 445
79
(63) Chan-Thaw C E Villa A Katekomol P Su D Thomas A Prati L Nano Letters 2010 10 537 (64) Palkovits R Antonietti M Kuhn P Thomas A Schueth F Angewandte Chemie-International Edition 2009 48 6909 (65) Ben T Ren H Ma S Q Cao D P Lan J H Jing X F Wang W C Xu J Deng F Simmons J M Qiu S L Zhu G S Angewandte Chemie-International Edition 2009 48 9457 (66) Yamamoto T Bulletin of the Chemical Society of Japan 1999 72 621 (67) Zhou G Baumgarten M Muellen K Journal of the American Chemical Society 2007 129 12211 (68) Yuan Y Sun F Ren H Jing X Wang W Ma H Zhao H Zhu G Journal of Materials Chemistry 2011 21 13498 (69) Ren H Ben T Sun F Guo M Jing X Ma H Cai K Qiu S Zhu G Journal of Materials Chemistry 2011 21 10348 (70) Zhao H Jin Z Su H Jing X Sun F Zhu G Chemical Communications 2011 47 6389 (71) Mortera R Fiorilli S Garrone E Verne E Onida B Chemical Engineering Journal 2010 156 184 (72) Dogru M Sonnauer A Gavryushin A Knochel P Bein T Chemical Communications 2011 47 1707 (73) Spitler E L Koo B T Novotney J L Colson J W Uribe-Romo F J Gutierrez G D Clancy P Dichtel W R Journal of the American Chemical Society 2011 133 19416 (74) Zhang Y Tan M Li H Zheng Y Gao S Zhang H Ying J Y Chemical Communications 2011 47 7365 (75) Uribe-Romo F J Doonan C J Furukawa H Oisaki K Yaghi O M Journal of the American Chemical Society 2011 133 11478 (76) Ben T Pei C Zhang D Xu J Deng F Jing X Qiu S Energy amp Environmental Science 2011 4 3991 (77) Lukose B Kuc A Heine T Chemistry-a European Journal 2011 17 2388 (78) Zhou W Wu H Yildirim T Chemical Physics Letters 2010 499 103 (79) Amirjalayer S Snurr R Q Schmid R Journal of Physical Chemistry C 2012 116 4921 (80) Xu Q Zhong C Journal of Physical Chemistry C 2010 114 5035 (81) Lukose B Supronowicz B St Petkov P Frenzel J Kuc A B Seifert G Vayssilov G N Heine T Physica Status Solidi B-Basic Solid State Physics 2012 249 335 (82) Assfour B Seifert G Chemical Physics Letters 2010 489 86 (83) Zhao L Zhong C L Journal of Physical Chemistry C 2009 113 16860 (84) Schmid R Tafipolsky M Journal of the American Chemical Society 2008 130 12600 (85) Han S S Goddard W A III Journal of Physical Chemistry C 2007 111 15185 (86) Suh M P Park H J Prasad T K Lim D-W Chemical Reviews 2012 112 782 (87) Furukawa H Yaghi O M Journal of the American Chemical Society 2009 131 8875 (88) Wong-Foy A G Matzger A J Yaghi O M Journal of the American Chemical Society 2006 128 3494 (89) Mendoza-Cortes J L Han S S Furukawa H Yaghi O M Goddard III W A Journal of Physical Chemistry A 2010 114 10824 (90) Doonan C J Tranchemontagne D J Glover T G Hunt J R Yaghi O M Nature Chemistry 2010 2 235 (91) Getman R B Bae Y-S Wilmer C E Snurr R Q Chemical Reviews 2012 112 703 (92) Han S S Furukawa H Yaghi O M Goddard III W A Journal of the American Chemical Society 2008 130 11580 (93) Furukawa H Ko N Go Y B Aratani N Choi S B Choi E Yazaydin A O Snurr R Q OKeeffe M Kim J Yaghi O M Science 2010 329 424 (94) Garberoglio G Langmuir 2007 23 12154 (95) Assfour B Seifert G Microporous and Mesoporous Materials 2010 133 59
80
(96) Lan J Cao D Wang W Journal of Physical Chemistry C 2010 114 3108 (97) Yang Q Zhong C Langmuir 2009 25 2302 (98) Garberoglio G Vallauri R Microporous and Mesoporous Materials 2008 116 540 (99) Lan J H Cao D P Wang W C Ben T Zhu G S Journal of Physical Chemistry Letters 2010 1 978 (100) Furukawa H Miller M A Yaghi O M Journal of Materials Chemistry 2007 17 3197 (101) Babarao R Jiang J Energy amp Environmental Science 2008 1 139 (102) Choi Y J Choi J H Choi K M Kang J K Journal of Materials Chemistry 2011 21 1073 (103) Lan J Cao D Wang W Smit B Acs Nano 2010 4 4225 (104) Wang L Yang R T Energy amp Environmental Science 2008 1 268 (105) Krishna R van Baten J M Industrial amp Engineering Chemistry Research 2011 50 7083 (106) Keskin S Journal of Physical Chemistry C 2012 116 1772 (107) Liu Y Liu D Yang Q Zhong C Mi J Industrial amp Engineering Chemistry Research 2010 49 2902 (108) Keskin S Sholl D S Langmuir 2009 25 11786 (109) Klontzas E Tylianakis E Froudakis G E Nano Letters 2010 10 452 (110) Zhu Y Zhou J Hu J Liu H Hu Y Chinese Journal of Chemical Engineering 2011 19 709 (111) Mendoza-Cortes J L Pascal T A Goddard W A III Journal of Physical Chemistry A 2011 115 13852 (112) Lino M A Lino A A Mazzoni M S C Chemical Physics Letters 2007 449 171 (113) Krishnan A Dujardin E Ebbesen T W Yianilos P N Treacy M M J Physical Review B 1998 58 14013 (114) Kim D Jung D H Kim K-H Guk H Han S S Choi K Choi S-H Journal of Physical Chemistry C 2012 116 1479 (115) Kim D Jung D H Choi S-H Kim J Choi K Journal of the Korean Physical Society 2008 52 1255 (116) Farha O K Yazaydin A O Eryazici I Malliakas C D Hauser B G Kanatzidis M G Nguyen S T Snurr R Q Hupp J T Nature Chemistry 2010 2 944 (117) Wu M M Wang Q Sun Q Jena P Kawazoe Y Journal of Chemical Physics 2010 133 (118) Choi Y J Lee J W Choi J H Kang J K Applied Physics Letters 2008 92 (119) Mendoza-Cortes J L Han S S Goddard W A III Journal of Physical Chemistry A 2012 116 1621 (120) Cao D Lan J Wang W Smit B Angewandte Chemie-International Edition 2009 48 4730 (121) Lan J H Cao D P Wang W C Langmuir 2010 26 220 (122) Klontzas E Tylianakis E Froudakis G E Journal of Physical Chemistry C 2009 113 21253 (123) Sun Y Ben T Wang L Qiu S Sun H Journal of Physical Chemistry Letters 2010 1 2753 (124) Li F Zhao J Johansson B Sun L International Journal of Hydrogen Energy 2010 35 266 (125) Zou X Zhou G Duan W Choi K Ihm J Journal of Physical Chemistry C 2010 114 13402 (126) Babarao R Dai S Jiang D-e Langmuir 2011 27 3451
81
Appendix B
Structural properties of metal-organic frameworks within the density-functional based tight-binding method
Binit Lukose Barbara Supronowicz Petko St Petkov Johannes Frenzel Agnieszka Kuc
Gotthard Seifert Georgi N Vayssilov and Thomas Heine
Phys Status Solidi B 2012 249 335ndash342
DOI 101002pssb201100634
Abstract Density-functional based tight-binding (DFTB) is a powerful method to describe large
molecules and materials Metal-organic frameworks (MOFs) materials with interesting catalytic
properties and with very large surface areas have been developed and have become commercially
available Unit cells of MOFs typically include hundreds of atoms which make the application of
standard density-functional methods computationally very expensive sometimes even unfeasible
The aim of this paper is to prepare and to validate the self-consistent charge-DFTB (SCC-DFTB)
method for MOFs containing Cu Zn and Al metal centers The method has been validated against
full hybrid density-functional calculations for model clusters against gradient corrected density-
functional calculations for supercells and against experiment Moreover the modular concept of
MOF chemistry has been discussed on the basis of their electronic properties We concentrate on
MOFs comprising three common connector units copper paddlewheels (HKUST-1) zinc oxide Zn4O
tetrahedron (MOF-5 MOF-177 DUT-6 (MOF-205)) and aluminum oxide AlO4(OH)2 octahedron (MIL-
53) We show that SCC-DFTB predicts structural parameters with a very good accuracy (with less than
82
5 deviation even for adsorbed CO and H2O on HKUST-1) while adsorption energies differ by 12 kJ
mol1 or less for CO and water compared to DFT benchmark calculations
1 Introduction In reticular or modular chemistry molecular building blocks are stitched together to
form regular frameworks [1] With this concept it became possible to construct framework
compounds with interesting structural and chemical composition most notably metal-organic
frameworks (MOFs) [1ndash10] and covalent organic frameworks (COFs) [11ndash17] The interest in MOFs
and COFs is not limited to chemistry these crystalline materials are also interesting for applications
in information storage [18] sieving of quantum liquids [19] hydrogen storage [20] and for fuel cell
membranes [21ndash23]
Covalent organic framework and MOF frameworks are composed by combining two types of building
blocks the so-called connectors typically coordinating in four to eight sites and linkers which have
typically 2 sometimes also three or four connecting sites Fig 1 illustrates a simplified representation
of the topology of connectors and linkers by using secondary building units (SBUs) (see Fig 1)
Figure 1 The connector and the linker of the frameworks CuBTC (HKUST-1) MOF-177 MOF-5 DUT-6 (MOF-205) and MIL-53 and their respective secondary building units (SBUs) The arrows indicate the connection points of the fragments Red oxygen green carbon white hydrogen yellow copperzinc and blue aluminum
Linkers are organic molecules with carboxylic acid groups at their connection sites which form
bonds to the connectors (typically in a solvothermal condensation reaction) They can carry
functional groups which can make them interesting for applications in catalysis [24] Connectors are
83
either metal organic units as for example the well-known Zn4O(CO2)6 unit of MOF-5 [8] the
Cu2(HCOO)4 paddle wheel unit of HKUST-1 [10] or covalent units incorporating boron oxide linking
units for COFs [11] As the building blocks of MOFs and COFs comprise typically some ten atoms unit
cells quickly get very large In particular if adsorption or dynamic processes in MOFs and COFs are of
interest (super)cells of some 1000 atoms need to be processed While standard organic force fields
show a reasonable performance for COFs [25] the creation of reliable force fields is not
straightforward for MOFs as transferable parameterization of the transition metal sites is an issue
even though progress has been achieved for selected materials [26 27] The difficulty to describe
transition metals in particular if they are catalytically active as in HKUST-1 limits the applicability of
molecular mechanics (MM) even for QMMM hybrid methods [28]
On the other hand the density-functional based tight-binding (DFTB) method with its self-consistent
charge (SCC) extension to improve performance for polar systems is a computationally feasible
alternative This non-orthogonal tight-binding approximation to density-functional theory (DFT)
which has been developed by Seifert Frauenheim and Elstner and their groups [29ndash32] (for a recent
review see Ref [30]) has been successfully applied to a large-scale systems such as biological
molecules [33ndash36] supramolecular systems [37 38] surfaces [39 40] liquids and alloys [41 42] and
solids [43] Being a quantum-mechanical method and hence allowing the description of breaking and
formation of chemical bonds the method showed outstanding performance in the description of
processes such as the mechanical manipulation of nanomaterials [44ndash46] It is remarkable that the
method performs well for systems containing heavier elements such as transition metals as this
domain cannot be covered so far with acceptable accuracy by traditional semi-empirical methods [47
48] DFTB covers today a large part of the elements of the periodic table and parameters and a
computer code are available from the DFTBorg website
Recently we have validated DFTB for its application for COFs [16 17] and have shown that structural
properties and formation energies of COFs are well described within DFTB Kuc et al [49] have
validated DFTB for substituted MOF-5 frameworks where connectors are always the Zn4O(CO2)6 unit
which has been combined with a large variety of organic linkers In this work we have revised the
DFTB parameters developed for materials science applications and validated them for HKUST-1 and
being far more challenging for the interaction of its catalytically active Cu sites with carbon
monoxide (CO) and water The Cu2(HCOO)4 paddle wheel units are electronically very intriguing on a
first note the electronic ground state of the isolated paddle wheel is an antiferromagnetic singlet
state which cannot be described by one Slater determinant and which is consequently not accessible
for KohnndashSham DFT However the energetically very close triplet state correctly describes structure
and electronic density of the system and also adsorption properties agree well with experiment [32
84
50 51] We therefore use DFT in the B3LYP hybrid functional representation as benchmark for DFTB
validations for HKUST-1 added by DFT-GGA calculations for periodic unit cellsWewill show that the
general transferability of the DFTB method will allow investigating structural electronic and in
particular dynamic properties
2 Computational details All calculations of the finite model and periodic crystal structures of MOFs
were carried out using the dispersion-corrected self-consistent density functional based tight-binding
(DC-SCC-DFTB in short DFTB) method [30 52ndash54] as implemented in the deMonNano code [55] Two
sets of parameters have been used the standard SCC-DFTB parameter set developed by Elstner et al
[53] for Zn-containing MOFs for Cu- and Al-containing materials we have extended our materials
science parameter set which has been developed originally for zeolite materials to include Cu For
this element we have used the standard procedure of parameter generation we have used the
minimal atomic valence basis for all atoms including polarization functions when needed Electrons
below the valence states were treated within the frozen-core approximation The matrix elements
were calculated using the local density approximation (LDA) while the short-range repulsive pair-
potential was fitted to the results from DFT generalized gradient approximation (GGA) calculations
For more details on DFTB parameter generation see Ref [30] Periodic boundary conditions were
used to represent frameworks of the crystalline solid state The conjugate-gradient scheme was
chosen for the geometry optimization An atomic force tolerance of 3x10-4 eV Aring-1 was applied
The reference DFT calculations of the equilibrium geometry of the investigated isolated MOF models
were performed employing the Becke three-parameter hybrid method combined with a LYP
correlation functional (B3LYP) [56 57] The basis set associated with the HayndashWadt [58] relativistic
effective core potentials proposed by Roy et al [59] was supplemented with polarization ffunctions
[60] and was employed for description of the electronic structure of Cu atoms The Pople 6-311G(d)
basis sets were applied for the H C and O atoms The calculations were performed with the
Gaussian09 program suite [61] For optimization of the periodic structure of CuBTC at DFT level the
electronic structure code Quickstep [62] which is a part of the CP2K package [63] was used
Quickstep is an implementation of the Gaussian plane wave method [64] which is based on the
KohnndashSham formulation of DFT
We have used the generalized gradient functional PBE [65] and the GoedeckerndashTeterndashHutter
pseudopotentials [64 66] in conjunction with double-z basis sets with polarization functions DZVP-
MOLOPT-SR-GTH basis sets obtained as described in Ref [67] but using two less diffuse primitives
Due to the large unit cell sampling of Brillouin zone was limited only to the G-point Convergence
85
criterion for the maximum force component was set to 45 x 10-3 Hartree Bohr1 The plane wave
basis with cutoff energy of 400 Ry was used throughout the simulations
The initial crystal structures were taken from experiment (powder X-ray diffraction (PXRD) data) The
cell parameters and the atomic positions were fully optimized using conjugate-gradient method at
the DFTB level For the reference DFT calculations only the atomic positions of experimental crystal
structures were minimized The cluster models were cut from the optimized structures and saturated
with hydrogen atoms (see Fig 2 for exemplary systems of Cu-BTC)
3 Results and discussion
31 Cu-BTC We have optimized the crystal structure of Cu-BTC (HKUST-1) [10] using cluster and the
periodic models The structural properties were compared to DFT results (see Table 1) The
geometries were obtained for the activated form (open metal sites) and in the presence of axial
water ligands (see Table 2) as the synthesized crystals often have H2O molecules attached to the
open metal sites of Cu (Fig 2b) We have also studied changes of the cluster model geometry in the
presence of CO (Fig 2c) and the results are also shown in Table 2 We have obtained good agreement
with experimental data as well as with DFT results
Figure 2 (a) Periodic and cluster models of Cu-BTC MOF as used in the computer simulations (b) cluster model with adsorbed H2O and (c) CO molecules
We have also simulated the XRD patterns of Cu-BTC which show very good structural agreement for
peak positions between the experimental and calculated structures The XRD pattern of DFT
optimized structure is nearly a copy of that of the experimental geometry However for DFTB
optimized patterns the relative intensities of some of the peaks notably that of the peaks at 2θ =
138 and 2θ = 158 are different from the experimental XRDs (see Fig 3) The differences in bond
angles between simulation and experiment may affect the sharpness of the signals and hence the
86
intensity To support this argument we have calculated the radial pair distribution function (g(r))
which details the probabilities of all atomndashatom distances in a given cell volume Figure S3 in the
Supporting Information shows g(r) of CundashCu CundashC CundashO CndashC and CndashO atom pairs of DFTB
optimized DFT optimized and experimentally modeled Cu-BTC unit cells The peaks of DFT as well as
DFTB optimized geometries are much broadened whereas the experimentally modeled geometry
has single peaks at the most predicted atomndashatom distances However it is to be noted that DFTB
optimized geometries give slightly shifted peaks compared to those of the DFT optimized geometry
They are broadened around the experimental values The distances between Cu and C atoms which
are not direct neighbors have the largest deviations from the experiment what indicates
shortcomings of the bond angles
Table 1 Selected bond lengths (Aring ) bond angles (⁰) and dihedral angles (⁰) of Cu-BTC optimized at DFTB and DFT level
Bond Type Cluster Model Periodic Model Exp
Cu-Cu 250 (257) 250 (250) 263 Cu-O 205 (198) 202 (198) 195 O-C 134 (133) 133-138 (128) 125
OCuO 836-971 (898) 892-907 (873-937)
891 896
Cu-O-O-Cu plusmn56 ˗ plusmn62 (plusmn02) plusmn04 (plusmn47 ˗ plusmn131) 0
O-C-C-C plusmn38 - plusmn50 (0) plusmn03 - plusmn05 (plusmn06 - plusmn105) 063
Cell paramet a=b=c=27283 (26343)
α=β=γ=90 (90) a=b=c=26343
α=β=γ=90
In detail the bond lengths and bond angles do not change significantly when going from the cluster
to the periodic model confirming the reticular chemistry approach The only exception is the OndashCundash
O bond angle that differs by 4ndash78 between the two systems at both levels of theory
87
Figure 3 Simulated XRD patterns of Cu-BTC MOF with the experimental unit cell parameters and optimized at the DFTB and DFT level of theory
The bond length between Cu atoms is slightly underestimated comparing with experimental (by
maximum 5) and DFT (maximum by 3) results while all other bond lengths are somewhat larger
at DFTB
All bond lengths stay unchanged or become longer in the presence of water molecules The most
striking example is the CundashCu bond where the change reaches 012ndash017 Aring compared with the
structure with activated Cu sites Also bond angles increase by up to 28 when the H2O is present
The CundashO bond length with the oxygen atom from the carboxylate groups is shorter than that with
the oxygen atoms from the water molecules indicating that the latter is only weakly bound to the
copper ions The OndashC distances (~133 Aring) correspond to values between that for the typical single
(142 Aring ) and double (122 Aring) oxygenndashcarbon bond The calculated CringndashCcarboxyl bond lengths of
146A deg indicate that these are typical single sp2ndashsp2 CndashC bonds while experimental findings give a
slightly longer value (15 Aring) for this MOF When comparing dihedral angles CundashOndashOndashCu and OndashCndashCndashC
of the clusters fairly large deviations between DFTB and DFT calculations and experiment are visible
due to the softer potential energy surface associated with these geometrical parameters In periodic
models however the agreement of DFT and DFTB with experiment and with each other is
significantly improved
The unit cell parameters with and without water molecules obtained at the DFTB level overestimate
the experimental data by less than 4 which gives a fairly good agreement if we take into account
high porosity of the material These lattice parameters increase only slightly from 27283 to 27323 Aring
in the presence of water
We have calculated the binding energies of H2O molecules attached to the Cu-metal sites within the
cluster model At the DFTB level Ebind of 40 kJ mol-1 was obtained in a good agreement with the DFT
results (52 kJ mol-1) as well as results from the literature (47 kJ mol-1 [50]) We have also calculated
88
the binding energy of CO which was twice smaller than that of H2O (165 and 28 kJ mol-1 for DFTB
and DFT respectively) suggesting much stronger binding of O atom (from H2O) than C atom (from
CO) While the bond lengths in the MOF structure do not change in the presence of either H2O or CO
the differences in the binding energy come from much longer bond distances (by around 07 Aring) for
CundashC than for CundashO in the presence of CO and water molecules respectively
Furthermore we have studied the electronic properties of periodic and cluster model Cu-BTC by
means of partial density-of-states (PDOS)We have compared especially the Cu-PDOS for s- p- and d-
orbitals (see Fig 4) The results show that the electronic structure stays unchanged when going from
the cluster models to the fully periodic crystal of Cu-BTC Since the copper atoms are of Cu(II) the d-
orbitals are just partially occupied what results in the metallic states at the Fermi level This is a very
interesting result as other ZnndashO-based MOFs are either semiconductors or insulators [49]
Table 2 Selected bond lengths (Aring) bond angles (˚) and dihedral angles (˚) of Cu-BTC optimized at DFTB and DFT levels with water molecules in the axial positions Cluster models are calculated using water and carbon monoxide molecules The adsorption energies Eads (kJ mol-1) of the guest molecules are given as well The DFT data are given in parenthesis
Bond Type Cluster Model +
H2O Periodic
Model+ H2O Cluster Model +
CO
Cu-Cu 267 (266) 262 (260) 250 (260)
Cu-O 205 (197-206) 210 (196-200) 206 (199)
O-C 134 (127) 133 (128) 134 (127)
OCuO 843-955 (889-905)
871-921 (842-930) 842-967 (896)
Cu-O-O-Cu plusmn59 - plusmn76 (plusmn06 ˗ plusmn10)
plusmn02 ˗ plusmn18 (plusmn40 ˗ plusmn136)
plusmn51 - plusmn58 (plusmn70)
O-C-C-C plusmn39 - plusmn53 (plusmn05 - plusmn11)
plusmn03 - plusmn05 (plusmn06 - plusmn105)
plusmn35 - plusmn43 (plusmn12)
Cu-O(H2O) Cu-C(CO) 237 (219) 244 (233-
255) 307 (239)
Eads -4045 (-5200) -1648
(-2800)
32 MOF-5 -177 DUT-6 (MOF-205) and MIL- 53 We have also studied the structural properties
of MOF structures with large surface areas using the SCC-DFTB method Very good agreement with
the experimental data shows that this method is applicable for MOFs of large structural diversity as
well as for coordination polymers based on the MOF-5 framework which has been reported earlier
[49] Tables 3 and 4 show selected bond lengths and bond angles of MOF-5 [68] MOF-177 [69] DUT-
6 (MOF-205) [70 71] and MIL-53 [72] respectively
89
MOFs-5 -177 and DUT-6 (MOF-205) are built of the same connector which is Zn4O(CO2)6
octahedron The difference is in the organic linker 14-benzenedicarboxylate (BDC) 135-
benzenetribenzoates (BTB) and BTB together with 26-naphtalenedicarboxylate (NDC) for MOF-5 -
177 andDUT-6 (MOF-205) respectively (see Fig 5)
Figure 4 DFTB calculations of PDOS of (top left) Cu in Cu-BTC and Zn in MOF-5 (top right) MOF-177 (bottom left) and DUT-6 (MOF-205) (bottom right) In each plot s-orbitals (top) p-orbitals (middle) and d-orbitals (bottom) are shown as computed from periodic (black) and cluster (red) models Cluster models are given in Fig S4
All three MOFs have different topologies due to the organic linkers where the number of
connections is varied or where two different linker types are present MOF-5 is the most simple and
it is the prototype in the isoreticular series (IRMOF-1) [68] it is a simple cubic topology with
threedimensional pores of the same size and the linkers have only two connection points In the
case of MOF-177 the linker is represented by a triangularSBU that means three connection points
are present This results in the topology with mixed (36)-connectivity [69] DUT-6 (MOF-205) has a
much more complicated topology due to two types of linkers The first one (NDC) has just two
90
connection points while the second is the same as in MOF-177 with three connection points One
ZnndashO-based connector is connected to two NDC and four BTB linkers The topology is very interesting
all rings of the underlying nets are fivefold and it forms a face-transitive tiling of dodectahedra and
tetrahedra with a ratio of 13 [73]
Figure 5 The crystal structures in the unit cell representations of (a) MOF-5 (b) MOF-177 (c) DUT-6 (MOF-205) and (d) MIL-53 red oxygen gray carbon white hydrogen and blue metal atom (Zn Al)
MIL-53 is a MOF structure with one-dimensional pores The framework is built up of corner-sharing
AlndashO-based octahedral clusters interconnected with BDC linkers The linkers have again two
connection points MIL-53 shows reversible structural changes dependent on the guest molecules
[74] It undergoes the so-called breathing mode depending on the temperature and the amount of
adsorbed molecules
In this case also the bond lengths and bond angles are slightly overestimated comparing with the
experimental structures but the error does not exceed 3
91
Table 3 Selected bond lengths (Aring) and bond angles (˚) of MOF-5 (424 atoms in unit cell) MOF-177 (808 atoms in unit cell) and DUT-6 (MOF-205) (546 atoms in unit cell) optimized at DFTB level The available experimental data is given in parenthesis [70-73+ Orsquo denotes the central O atom in the Zn-O-octahedron
Bond Type MOF-5 MOF-177 DUT-6
(MOF-205)
Zn-Zn 330 (317) 322-336 (306-330)
325-331 (318)
Zn-Orsquo 202 (194) 202 (193) 202 (194) Zn-O 204 (192) 202-206
(190-199) 202 205 (193)
O-C 128 (130) 128 (131) 128 (125) ZnOrsquoZn 1095 (1095) 1056-1124
(1055 1092) 107-1118 (1084 1100)
OZnO 1083 1108 (1061)
1048 1145 (981-1281)
1046-1112 (1062 1085)
Zn-O-O-Zn 0 (0 plusmn17 plusmn20) plusmn122 - plusmn347 (plusmn176 - plusmn374)
05 - plusmn62 (0 plusmn29)
O-C-C-C 00 (0 - plusmn24) (plusmn 35 - plusmn 336) (plusmn65 - plusmn374)
plusmn04 plusmn22 (0 plusmn174)
Cell paramet a=b=c=26472 (25832) α=β=γ=90 (90)
a=b=37872 (37072) c=3068 (30033) α=β=90 (90) γ=120 (120)
a=b=c=31013 (30353) α=β=γ=90 (90)
We have calculated PDOS for MOF-5 MOF-177 and DUT-6 (MOF-205) (see Fig 4) Their band gaps
calculated to be 40 35 and 31 eV respectively indicate that they are either semiconductors or
insulators We have calculated it for both periodic crystal structure and a cluster containing a metal-
oxide connector and all its carboxylate linkers
Table 4 Selected bond lengths (Aring) and bond angles (˚) of MIL-53 (152 atoms in unit cell) optimized at DFTB level
Bond Type DFTB Exp
Al-Al 341 331 Al-O 189 183-191 O-C 133 129 130 O-Al-O 884 912 886 898 Al-O-Al 1289 1249 Al-O-O-Al 0 0 O-C-C-O 06 03 Cell paramet a=1246
b=1732 c=1365 α=β=γ=90
a=1218 b=1713 c=1326 α=β=γ=90
4 Mechanical properties Due to the low-mass density the elastic constants of porous materials
are a very sensitive indicator of their mechanical stability For crystal structures like MOFs we have
92
studied these by means of bulk modulus (B) B can be calculated as a second derivative of energy
with respect to the volume of the crystal (here unit cell)
The result shows that CuBTC has bulk modulus of 3466 GPa what is in close agreement with
B=3517 GPa obtained using force-field calculations [73 74] and experiment [75] Bulk moduli for the
series of MOFs resulted in 1534 1010 and 1073 GPa for MOF-5 -177 and DUT-6 (MOF-205)
respectively For MOF-5 B=1537 GPa was reported from DFTGGA calculations using plane-waves
[76 77] The results show that larger linkers give mechanically less stable structures what might be
an issue for porous structures with larger voids For MIL-53 we have obtained the largest bulk
modulus of 5369 GPa keeping the angles of the pore fixed
5 Conclusions We have validated the DFTB method with SCC and London dispersion corrections for
various types of MOFs The method gives excellent geometrical parameters compared to experiment
and for small model systems also in comparison with DFT calculations Importantly this statement
holds not only for catalytically inactive MOFs based on the Zn4O(CO2)6 octahedron and organic linkers
which are important for gas adsorption and separation applications but also for catalytically active
HKUST-1 (CuBTC) This is remarkable as the Cu atoms are in the Cu2+ state in this framework DFTB
parameters have been generated and validated for Cu and the electronic structure contains one
unpaired electron per Cu atom in the unit cell which makes the electronic description technically
difficult but manageable within the SCC-DFTB method SCC-DFTB performs well for the frameworks
themselves as well as for adsorbed CO and water molecules
Partial density-of-states calculations for the transition metal centers reveal that the concept of
reticular chemistry ndash individual building units keep their electronic properties when being integrated
to the framework ndash is also valid for the MOFs studied in this work in agreement with a previous
study of COFs [16] The electronic properties computed using the cluster models and the periodic
structure contains the same features and hence cluster models are good models to study the
catalytic and adsorption properties of these materials This is in particular useful if local quantum
chemical high-level corrections shall be applied that require to use finite structures
We finally conclude that we have now a high-performing quantum method available to study various
classes of MOFs of unit cells up to 10000 atoms including structural characteristics the formation
and breaking of chemical and coordination bonds for the simulation of diffusion of adsorbate
molecules or lattice defects as well as electronic properties The parameters can be downloaded
from the DFTBorg website
93
References
[1] E A Tomic J Appl Polym Sci 9 3745 (1965)
2+ M Eddaoudi D B Moler H L Li B L Chen T M Reineke M OrsquoKeeffe and O M Yaghi Acc Chem Res
34 319 (2001)
[3] M Eddaoudi H L Li T Reineke M Fehr D Kelley T L Groy and O M Yaghi Top Catal 9 105 (1999)
[4] B F Hoskins and R Robson J Am Chem Soc 111 5962 (1989)
[5] K Schlichte T Kratzke and S Kaskel Microporous Mesoporous Mater 73 81 (2004) [6] J L C Rowsell A
R Millward K S Park and O M Yaghi J Am Chem Soc 126 5666 (2004)
7+ O M Yaghi M OrsquoKeeffe N W Ockwig H K Chae M Eddaoudi and J Kim Nature 423 705 (2003)
[8] M Eddaoudi J Kim N Rosi D Vodak J Wachter M OrsquoKeeffe and O M Yaghi Science 295 469 (2002)
9+ M OrsquoKeeffe M Eddaoudi H L Li T Reineke and O M Yaghi J Solid State Chem 152 3 (2000)
[10] S S Y Chui SM F Lo J P H Charmant A G Orpen and I D Williams Science 283 1148 (1999)
11+ A P Cote A I Benin N W Ockwig M OrsquoKeeffe A J Matzger and O M Yaghi Science 310 1166 (2005)
[12] A P Cote H M El-Kaderi H Furukawa J R Hunt and O M Yaghi J Am Chem Soc 129 12914 (2007)
[13] H M El-Kaderi J R Hunt J L Mendoza-Cortes A P Cote R E Taylor M OrsquoKeeffe and O M Yaghi
Science 316 268 (2007)
[14] S Wan J Guo J Kim H Ihlee and D L Jiang Angew Chem 120 8958 (2008)
[15] S Wan J Guo J Kim H Ihlee and D L Jiang Angew Chem 121 5547 (2009)
[16] B Lukose A Kuc J Frenzel and T Heine Beilstein J Nanotechnol 1 60 (2010)
[17] B Lukose A Kuc and T Heine Chem Eur J 17 2388 (2011)
[18] D S Marlin D G Cabrera D A Leigh and A M Z Slawin Angew Chem Int Ed 45 77 (2006)
[19] I Krkljus and M Hirscher Microporous Mesoporous Mater 142 725 (2011)
[20] M Hirscher Handbook of Hydrogen Storage (Wiley-VCH Weinheim 2010)
[21] H Kitagawa Nature Chem 1 689 (2009)
[22] M Sadakiyo T Yamada and H Kitagawa J Am Chem Soc 131 9906 (2009)
[23] T Yamada M Sadakiyo and H Kitagawa J Am Chem Soc 131 3144 (2009)
94
[24] K S Jeong Y B Go S M Shin S J Lee J Kim O M Yaghi and N Jeong Chem Sci 2 877 (2011)
[25] R Schmid and M Tafipolsky J Am Chem Soc 130 12600 (2008)
[26] M Tafipolsky S Amirjalayer and R Schmid J Comput Chem 28 1169 (2007)
[27] M Tafipolsky S Amirjalayer and R Schmid J Phys Chem C 114 14402 (2010)
[28] A Warshel and M Levitt J Mol Biol 103 227 (1976)
[29] G Seifert D Porezag and T Frauenheim Int J Quantum Chem 58 185 (1996)
[30] A F Oliveira G Seifert T Heine and H A Duarte J Braz Chem Soc 20 1193 (2009)
[31] D Porezag T Frauenheim T Koumlhler G Seifert and R Kaschner Phys Rev B 51 12947 (1995)
[32] T Frauenheim G Seifert M Elstner Z Hajnal G Jungnickel D Porezag S Suhai and R Scholz Phys
Status Solidi B 217 41 (2000)
[33] M Elstner Theor Chem Acc 116 316 (2006)
Supporting Information
Figure S4 Cluster models as used for the P-DOS calculations in Fig 4 MOF-5 MOF-177 and DUT-6 (MOF-205) (from left to right)
95
Figure S3 Radial pair distribution function (g(r)) of Cu-Cu Cu-C Cu-O C-C and C-O atom-pairs in a Cu-BTC MOF unit cell
96
Appendix C
The Structure of Layered Covalent-Organic Frameworks
Binit Lukose Agnieszka Kuc and Thomas Heine
Chem Eur J 2011 17 2388 ndash 2392
DOI 101002chem201001290
Abstract Covalent-Organic Frameworks (COFs) are a new family of 2D and 3D highly porous and
crystalline materials built of light elements such as boron oxygen and carbon For all 2D COFs an AA
stacking arrangement has been reported on the basis of experimental powder XRD patterns with the
exception of COF-1 (AB stacking) In this work we show that the stacking of 2D COFs is different as
originally suggested COF-1 COF-5 COF-6 and COF-8 are considerably more stable if their stacking
arrangement is either serrated or inclined and layers are shifted with respect to each other by ~14 Aring
compared with perfect AA stacking These structures are in agreement with to date experimental
data including the XRD patterns and lead to a larger surface area and stronger polarisation of the
pore surface
Introduction In 2005 Cocircteacute and co-workers introduced a new class of porous crystalline materials
Covalent Organic Frameworks (COFs)[1] where organic linker molecules are stitched together by
connectors covalent entities including boron and oxygen atoms to a regular framework These
materials have the genuine advantage that all framework bonds represent strong covalent
interactions and that they are composed of light-weight elements only which lead to a very low
mass density[2] These materials can be synthesized solvothermally in a condensation reaction and
97
are composed of inexpensive and non-toxic building blocks so their large-scale industrial production
appears to be possible
Figure 1 Calculated and experimental XRD patterns of all studied COFs (COF-1 top left COF-5 top right COF-6 bottom left COF-8 bottom right)
To date a number of different COF structures have been reported[1ndash3] From a topological
viewpoint we distinguish two- and three-dimensional COFs In two-dimensional (2D) COFs the
covalently bound framework is restricted to 2D layers The crystal is then similar as in graphite or
hexagonal boron nitride composed by a stack of layers which are not connected by covalent bonds
but held together primarily by London dispersion interactions
98
The COF structures have been analyzed by powder X-ray diffraction (PXRD) experiments The
topology of the layers is determined by the structure of the connector and linker molecules and
typically a hexagonal pattern is formed due to the 2m (D3h) symmetry of the connector moieties
The individual layers are then stacked and form a regular crystal lattice With one exception (COF-
1)[1] for all 2D-COFs AA (eclipsed P6mmm) interlayer stacking has been reported[3andashd f g] This
geometrical arrangement maximizes the proximity of the molecular entities and results in straight
channels orthogonal to the COF layers which are known from the literature[1 3a]
The connectors of 2D-COFs include oxygen and boron atoms which cause an appreciable polarization
The AA stacking arrangement maximizes the attractive London dispersion interaction between the
layers which is the dominating term of the stacking energy At the same time AA stacking always
results in a repulsive Coulomb force between the layers due to the polarized connectors It should be
noted that the eclipsed hexagonal boron nitride (h-BN)[4] has boron atoms in one layer serving as
nearest neighbours to nitrogen atoms in adjacent layers (AB stacking) The Coulomb interaction has
ruled out possible interlayer eclipse between atoms of alike charges In this article we aim at
studying the layer arrangements of solvent-free COFs based on the interlayer interactions and the
minimum variance Various lattice types have been considered all significantly more stable than the
reported AA stacked geometry In all 2D COFs we have studied the Coulomb interaction between the
layers leads to a modification of the stacking and shifts the layers by about one interatomic distance
(~14 Aring) with respect to each other (see Figure 1)
Breaking of the highly symmetric layer arrangement has been observed for COF-1 and COF-10 after
removal of guest molecules from pores leading to a turbostratic repacking of pore channels[1 5]
The authors have confirmed the disorder either by the changes in the PXRD spectrum taken before
and after adsorptionndashdesorption cycle[1] or by the changes in the adsorption behavior[5] The
disruption of layer arrangement is attributed to the force exerted by the guest molecules Formation
of poly disperse pores has also been verified[5] The lattice types that we discuss in this article on
the other hand are neither the result of the pressure from any external molecule in the pore nor
having more than one type of pores They are the resultant of minimum variance guided by Coulomb
and London dispersion interactions For the COF models under investigation perfect crystallinity has
been considered
Methods We have studied the experimentally known structures of COF-1 COF-5 COF-6 and COF-8
Structure calculations were carried out using the Dispersion-Corrected Self-Conistent-Charge
Density-Functional-Based Tight-Binding method (SCC-DFTB)[6] DFTB is based on a second-order
expansion of the KohnndashSham total energy in DFT with respect to charge density fluctuations This
does not require large amounts of empirical parameters however maintains all qualities of DFT The
99
accuracy is very much improved by the SCC extension The DFTB code (deMonNano[6a]) has
dispersion correction[6d] implemented to account for weak interactions and was used to obtain the
layered bulk structure of COFs and their formation energies The performance for interlayer
interactions has been tested previously for graphite[6d] All structures correspond to full geometry
optimizations (structure and unit cell) XRD patterns have been simulated using the Mercury
software[7] To allow best comparison with experiment for PXRD simulations we used the calculated
geometry of the layer and of the relative shifts between the layers but experimental interlayer
distances The electrostatic potential has been calculated using Crystal 09[8] at the DFTVWN level
with 6-31G basis set
Results and Discussion
In order to see the favorite stacking arrangement of the layers we have shifted every second layer in
two directions along zigzag and armchair vertices of the hexagonal pores (see Figure 2) The initial
stacking is AA (P6mmm) and the final zigzag shift gives AB stacking (P63mmc) Considering that the
attractive interlayer dispersion forces and repulsive interlayer orbital interactions are different we
have optimized the interlayer separation for each stacking Figure 2 show their total energies
calculated per formula unit that is per established bond between linkers and connectors with
reference to AA stacking for COF-5 and COF-1 In each direction the energy minimum is found close
to the AA stacking position shifted only by about one aromatic C-C bond length (~14 Aring) so that
either connector or linker parts become staggered with those in the adjacent layers leading to a
stacking similar to that of graphite for either connector (armchair shift) or linker (zigzag shift) For
COF-5 the staggering at the connector or linker depends whether the shift is armchair or zigzag
respectively while for COF-1 the zigzag shift alone can give staggering in both the aromatic and
boron-oxygen rings
The low-energy minima in the two directions are labeled following the common nomenclature as
zigzag and armchair respectively Both shifts result in a serrated stacking order (orthorhombic
Cmcm) which shows a significantly more attractive interlayer Coulomb interaction than AA stacking
(see Table 1) while most of the London dispersion attraction is maintained and consequently
stabilizes the material Still this configuration can be improved if we consider inclined stacking
(Figure 1) that is the lattice vector pointing out of the 2D plane is not any longer rectangular
reducing the lattice symmetry from Cmcm (orthorhombic) to C2m (monoclinic)
Besides we have calculated the monolayer formation energy (condensation energy Ecb) from the
total energies of the monolayer and of the individual building blocks and the stacking formation
energy from the total energies of the bulk structure and of the monolayer for four selected COFs The
100
COF building blocks are the same as those used for their synthesis[1 3a] (BDBA for COF-1 BDBA and
HHTP for COF-5 BTBA and HHTP for COF-6 BTPA and HHTP for COF-8) The final energies given per
formula unit that is per condensed bond are given in Table 1 The serrated and inclined stacking
structures are energetically more stable than AA and AB Interestingly within our computational
model zigzag and armchair shifting is energetically equivalent
Figure 2 Change of the total energy (per formula unit) when shifting COF-5 even layers in zigzag (top) and armchair (middle) directions and COF-1 in zigzag (bottom) direction The vector d shows the shift direction The low-energy minima in the two directions are labelled as zigzag and armchair respectively The equilibrium structures are shown as well
The formation energies of the individual COF structures are in agreement with full DFT calculations
We have calculated using ADF[9] the condensation energy of COF-1 and COF-5 using first-principles
DFT at the PBE[10]DZP[11] level to qualitatively support our results For simplicity we have used a
finite structure instead of the bulk crystal Their calculated Ecb energies are 77 and 15 kJmol-1
respectively for COF-1 and COF-5 These values support the endothermic nature of the condensation
101
reaction and are in excellent agreement with our DFTB results (91 and 21 kJmol-1 respectively see
Table 1)
The change of stacking should be visible in X-ray diffraction patterns because each space group has a
distinct set of symmetry imposed reflection conditions Powder XRD patterns from experiments are
available for all 2D COF structures[1 3andashd f g] We have simulated XRD patterns for AA AB serrated
Table 1 Calculated formation energies for the studied COF structures Ecb = condensation Esb = stacking energy EL = London dispersion energy Ee = electronic energy Energies are per condensed bond and are given in [kJ mol
-1+ lsquoarsquo and lsquozrsquo stand for armchair and zigzag respectively Note that the electronic energy Ee=Esb-EL
includes the electronic interlayer interaction and the formation energy of the monolayer and that the formation of the monolayers is endothermic
Structure Stacking Esb EL Ee
COF-5 AA -2968 -3060 092
AB -2548 -2618 070
serrated z -3051 -3073 022
serrated a -3052 -3073 021
inclined z -3297 -3045 -252
inclined a -3275 -3044 -231
Monolayer Ecb= 211
COF-1 AA -2683 -2739 056
AB -2186 -2131 -055
serrated z -2810 -2806 -004
inclined z -2784 -2788 004
Monolayer Ecb= 906
COF-6 AA -2881 -2963 082
AB -2127 -2146 019
serrated z -2978 -2996 018
serrated a -2978 -2995 017
inclined z -2946 -2975 029
inclined a -2954 -2974 021
Monolayer Ecb= 185
COF-8 AA -4488 -4617 129
102
AB -2477 -2506 029
serrated z -4614 -4646 032
serrated a -4615 -4647 032
inclined z -4578 -4612 035
inclined a -4561 -4591 030
Monolayer Ecb= 263
and inclined stacking forms for COF-5 COF-1 COF-6 and COF-8 (see Figure 1for their comparison
with experimental spectrum) For completeness we have studied XRD patterns of optimized COFs
using the experimentally determined[1 3a] interlayer separations this means we have kept the
layer geometry the same as the optimized structures and different stackings were obtained by
shifting adjacent layers accordingly
COF-1 was reported as having different powder spectra for samples before (as-synthesized) and after
removal of guest molecules with a particular mentioning about its layer shifting after removal We
have compared the two spectra with our simulated XRDs in order to see the stacking in the pure
form and how the stacking is changed at the presence of mesitylene guests Except that we have only
a vague comparison at angles (2θ) 11ndash15 the powder spectrum after guest removal is more similar
to the XRDs of serrated and inclined stacking structures A notable difference from AB is the absence
of high peaks at angles ~12 ~15 and ~19 This gives rise to the suggestion that COF-1 is not a
notable exception among the 2D COFs it follows the same topological trend as all other frameworks
of this class having one-dimensional (1D) pores as soon as the solvent is removed from the pores
This suggestion is supported by the experimental fact that the gas adsorption capacity of COF-1 is
only slightly lower than that of COF-6 for even as large gas molecules as methane and CO2[12] It is
not obvious how these molecules would enter an AB stacked COF-1 framework where the pores are
not connected by 1D channels (see Figure 1) Moreover our calculations suggest that the serrated
and inclined stackings are energetically favorable (see Table 1)
Indeed all the new stackings discussed in this article yield XRD patterns which are in agreement with
the currently available experimental data (see Figure 1) The inclined stackings have more peaks but
those are covered by the broad peaks in the experimental pattern The same is observed for the (002)
peaks of serrated stackings At present 2D COF synthesis methods are not yet capable to produce
crystals of sufficient quality to allow more detailed PXRD analysis in particular for the solvent-free
materials Microwave synthesis of COF-5 by rapid heating is reported[13] as much faster (~200 times)
compared with solvothermal methods however the structural details (XRD etc) remained
103
ambiguous We are confident that better crystals will be produced in future which will allow the
unambiguous determination of COF structures and can be compared to our simulations
Finally we want to emphasize that the suggested change of stacking is not only resulting in a
moderate change of geometry and XRD pattern The functional regions of COFs are their pores and
the pore geometry is significantly modified in our suggested structures compared to AA and AB
stackings First the inclined and serrated structures account for an increase of the surface area by 6
estimated for the interaction of the COFs with helium (3 for N2 5 for Ar 56 for Ne) Moreover
the shift between the layers exposes both boron and oxygen atoms to the pore surface leading to a
much stronger polarity than it can be expected for AA stacked COFs This difference is shown in
Figure 3 which plots the electrostatic potential of COF-5 in AA serrated and inclined stacking
geometries While the pore surface of AA stacked COF-5 shows a positively and negatively charged
stripes the other stacking arrangements show a much stronger alternation of charges indicating the
higher polarity of the surface The surface polarity can also be expressed in terms of Mulliken charges
of oxygen and boron atoms calculated at the DFT level which are COF-5 AA qB = 048 and qO = -048
COF-5 serrated qB = 047 and qO = -049 COF-5 inclined qB = 045 and qO = -048
Figure 3 Calculated electrostatic potential of COF-5 AA (left) serrated (middle) and inclined (right) stacking forms Blue - negative and red-positive values of the 005 isosurface
Conclusion We have studied 2D COF layer stackings energetically and found that energy minimum
structures have staggering of hexagonal rings at the connector or linker parts This can be achieved if
the bulk structure has either serrated or inclined order These newly proposed orders have their
simulated XRDs matching well with the available experimental powder spectrum Hence we claim
that all reported 2D COF geometries shall be re-examined carefully in experiment Due to the change
of the pore geometry and surface polarity the envisaged range of applications for 2D COFs might
change significantly We believe that these results are of utmost importance for the design of
functionalized COFs where functional groups are added to the pore surfaces
104
References
[1] A P Cocircteacute A I Benin N W Ockwig M OKeeffe A J Matzger O M Yaghi Science 2005 310 1166
[2] H M El-Kaderi J R Hunt J L Mendoza-Cortes A P Cote R E Taylor M OKeeffe O M Yaghi Science
2007 316 268
[3] a) A P Cocircteacute H M El-Kaderi H Furukawa J R Hunt O M Yaghi J Am Chem Soc 2007 129 12914 b) J
R Hunt C J Doonan J D LeVangie A P Cocircte O M Yaghi J Am Chem Soc 2008 130 11872 c) R W
Tilford W R Gemmill H C zur Loye J J Lavigne Chem Mater 2006 18 5296 d) R W Tilford S J Mugavero
P J Pellechia J J Lavigne Adv Mater 2008 20 2741 e) F J Uribe-Romo J R Hunt H Furukawa C Klock M
OKeeffe O M Yaghi J Am Chem Soc 2009 131 4570 f) S Wan J Guo J Kim H Ihee D L Jiang Angew
Chem 2008 120 8958 Angew Chem Int Ed 2008 47 8826 g) S Wan J Guo J Kim H Ihee D L Jiang
Angew Chem 2009 121 5547 Angew Chem Int Ed 2009 48 5439
[4] R T Paine C K Narula Chem Rev 1990 90 73
[5] C Doonan D Tranchemontagne T Glover J Hunt O Yaghi Nat Chem 2010 2 235
[6] a) T Heine M Rapacioli S Patchkovskii J Frenzel A M Koester P Calaminici S Escalante H A Duarte R
Flores G Geudtner A Goursot J U Reveles A Vela D R Salahub deMon deMonnano edn Mexico DF
Mexico and Bremen (Germany) 2009 b) A F Oliveira G Seifert T Heine H A Duarte J Braz Chem Soc
2009 20 1193 c) G Seifert D Porezag T Frauenheim Int J Quantum Chem 1996 58 185 d) L Zhechkov T
Heine S Patchkovskii G Seifert H A Duarte J Chem Theory Comput 2005 1 841
[7] a) httpwwwccdccamacukproductsmercury b) C F Macrae P R Edgington P McCabe E Pidcock
G P Shields R Taylor M Towler J Van de Streek J Appl Crystallogr 2006 39 453
[8] R Dovesi V R Saunders C Roetti R Orlando C M Zicovich- Wilson F Pascale B Civalleri K Doll N M
Harrison I J Bush P DrsquoArco M Llunell Crystal 102 ed
[9] a) httpwwwscmcom ADF200901 Amsterdam The Netherlands b) G te Velde F M Bickelhaupt S J
A van Gisbergen C Fonseca Guerra E J Baerends J G Snijders T Ziegler J Comput Chem 2001 22 931
[10] J P Perdew K Burke M Ernzerhof Phys Rev Lett 1996 77 3865
[11] E Van Lenthe E J Baerends J Comput Chem 2003 24 1142
[12] H Furukawa O M Yaghi J Am Chem Soc 2009 131 8875
[13] N L Campbell R Clowes L K Ritchie A I Cooper Chem Chem Mater 2009 21 204
105
Appendix D
On the reticular construction concept of covalent organic frameworks
Binit Lukose Agnieszka Kuc Johannes Frenzel and Thomas Heine
Beilstein J Nanotechnol 2010 1 60ndash70
DOI103762bjnano18
Abstract
The concept of reticular chemistry is investigated to explore the applicability of the formation of
Covalent Organic Frameworks (COFs) from their defined individual building blocks Thus we have
designed optimized and investigated a set of reported and hypothetical 2D COFs using Density
Functional Theory (DFT) and the related Density Functional based tight-binding (DFTB) method
Linear trigonal and hexagonal building blocks have been selected for designing hexagonal COF layers
High-symmetry AA and AB stackings are considered as well as low-symmetry serrated and inclined
stackings of the layers The latter ones are only slightly modified compared to the high-symmetry
forms but show higher energetic stability Experimental XRD patterns found in literature also
support stackings with highest formation energies All stacking forms vary in their interlayer
separations and band gaps however their electronic densities of states (DOS) are similar and not
significantly different from that of a monolayer The band gaps are found to be in the range of 17ndash
40 eV COFs built of building blocks with a greater number of aromatic rings have smaller band gaps
Introduction
In the past decade considerable research efforts have been expended on nanoporous materials due
to their excellent properties for many applications such as gas storage and sieving catalysis
106
selectivity sensoring and filtration [1] In 1994 Yaghi and co-workers introduced ways to synthesize
extended structures by design This new discipline is known as reticular chemistry [23] which uses
well-defined building blocks to create extended crystalline structures The feasibility of the building
block approach and the geometry preservation throughout the assembly process are the key factors
that lead to the design and synthesis of reticular structures
One of the first families of materials synthesized using reticular chemistry were the so-called Metal-
Organic Frameworks (MOFs) [4] They are composed of metal-oxide connectors which are covalently
bound to organic linkers The coordination versatility of constituent metal ions along with the
functional diversity of organic linker molecules has created immense possibilities The immense
potential of MOFs is facilitated by the fact that all building blocks are inexpensive chemicals and that
the synthesis can be carried out solvothermally MOFs are commercially available and the scale up of
production is continuing Since the discovery of MOFs many other crystalline frameworks have been
synthesized using reticular chemistry such as Metal-Organic Polyhedra (MOP) [5] Zeolite
Imidazolate Frameworks (ZIFs) [6] and Covalent Organic Frameworks (COFs) [7]
In 2005 Cocircteacute and co-workers introduced COF materials [7-14] where organic linker molecules are
stitched together by covalent entities including boron and oxygen atoms to form a regular
framework These materials have the distinct advantage that all framework bonds represent strong
covalent interactions and that they are composed of light-weight elements only which lead to a very
low mass density [7-9] These materials can be synthesized by wet-chemical methods by
condensation reactions and are composed of inexpensive and non-toxic building blocks so their
large-scale industrial application appears to be possible From a topological viewpoint we distinguish
two- and three-dimensional COFs In two-dimensional (2D) COFs the covalently bound framework is
restricted to layers The crystal is then similar as in graphite composed of a stack of layers which
are not connected by covalent bonds
COFs compared with MOFs have lower mass densities due to the absence of heavy atoms and
therefore might be better for many applications For example the gravimetric uptake of gases is
almost twice as large as that of MOFs with comparable surface areas [1516] Because COFs are fairly
new materials many of their properties and applications are still to be explored
Recently we have studied the structures of experimentally well-known 2D COFs [17] We have found
that commonly accepted 2D structures with AA and AB kinds of layering are energetically less stable
than inclined and serrated forms This is because AA stacking maximises the Coulomb repulsion due
to the close vicinity of charge carrying atoms alike (O B atoms) in neighbouring layers The serrated
and inclined forms are only slightly modified (layers are shifted with respect to each other by asymp14 Aring)
107
and experience less Coulomb forces between the layers compared to AA stacking This is equivalent
to the energetic preference of graphite for an AB (Bernal) over an AA form (simple hexagonal) if we
ignore the fact that interlayer ordering in serrated and inclined forms are not uniform everywhere A
known example of this is that in eclipsed hexagonal boron nitride (h-BN) boron atoms in one layer
serve as nearest neighbours to nitrogen atoms in adjacent layers (AB stacking) The Coulomb
interaction rules out possible interlayer eclipse between atoms with similar charges in this case
In the present work we aim to explore how far the concept of reticular chemistry is applicable to the
individual building units which define COFs For this purpose we have investigated a set of reported
and hypothetical 2D COFs theoretically by exploring their structural energetic and electronic
properties We have compared the properties of the isolated building blocks with those of the
extended crystal structures and have found that the properties of the building units are indeed
maintained after their assembly to a network
Results and Discussion
Structures and nomenclature
We have considered four connectors (IndashIV) and five linkers (andashe) for the systematic design of a
number of 2D COFs (Figure 1) Each COF was built from one type of connector and one type of linker
thus resulting in the design of 20 different structures Moreover we have considered two
hypothetical reference structures which are only built from connectors I and III (no linker is present)
REF-I and REF-III Properties of other COFs were compared with the properties of these two
structures Some of the studied COFs are already well known in the literature [781314] and we
continue to use their traditional nomenclature while hypothetical ones are labelled in the
chronological order with an M at the end which stands for modified
Figurel 1 The connector (IndashIV) and linker (andashe) units considered in this work The same nomenclature is used in the text Carbon ndash green oxygen ndashred boron ndash magenta hydrogen ndash white
108
Using the secondary building unit (SBU) approach we can distinguish the connectors between
trigonal [T] (connectors I II III) and hexagonal [H] (connector IV) and the linkers between linear [l]
(linkers a b e) and trigonal [t] (linkers c d) Topology of the layer is determined by the geometries
of the connector and linker molecules and typically a hexagonal pattern is formed due to the D3h
symmetry of the connector moieties Based on these topologies of the constituent building blocks
we have classified the studied COFs into four groups Tl Tt Hl and Ht (Figure 2) Hereafter we will
use this nomenclature to describe the COF topologies
Figurel 2 Topologies of 2D COFs considered in this work (from the left) Tl Tt Hl and Ht Red and blue blocks are secondary building units corresponding to connectors and linkers respectively
We have considered high-symmetry AA and AB kinds of stacking (hexagonal) and low-symmetry
serrated (orthorhombic) and inclined (monoclinic) kinds of stacking of the layers The latter two were
discussed in a previous work on 2D COFs [17] As an example the structure of COF-5 [7] in different
kinds of stacking of layers is shown in Figure 3 In eclipsed AA stacking atoms of adjacent layers lie
directly on top of each other whilst in staggered AB stacking three-connected vertices lie directly on
top of the geometric center of six-membered rings of neighbouring layers In both serrated and
inclined kinds of stacking the layers are shifted with respect to each other by approximately 14 Aring
resulting in hexagonal rings in the connector or linker being staggered with those in the adjacent
layers In serrated stacking alternate layers are eclipsed In inclined stacking layers lie shifted along
one direction and the lattice vector pointing out of the 2D plane is not rectangular For COFs made of
connector I due to the absence of five-membered C2O2B rings a zigzag shift leads to staggering in
both connector and linker parts For those made of other connectors staggering at the connector or
linker depends on whether the shift is armchair or zigzag respectively [17]
Structural properties
We have compared structural properties of isolated building blocks with those of the extended COF
structures Negligible differences have been found in the bond lengths and bond angles of the
building blocks and the corresponding crystal structures This indicates that the structural integrity of
the building blocks remains unchanged while forming covalent organic frameworks confirming the
109
principle of reticular chemistry In addition the CndashB BndashO and OndashC bond lengths are almost the same
when different COF structures are compared (see Table S1 in Supporting Information File 1) The
experimental bond lengths are asymp154 Aring for CndashB asymp138 Aring for OndashC and 137ndash148 Aring for BndashO However
in the case of COF-1 the experimental values are slightly larger (160 Aring for CndashB and 151 Aring for BndashO)
This could be the reason why our calculated bond lengths for COF-1 are much shorter than the
experimental values while all the other structures agree within 5 error The calculated CndashC bond
lengths vary in the range from 136ndash147 Aring (Figure S2 in Supporting Information File 1) and are the
same for the COFs and their constituent building blocks at the respective positions of the carbon
atoms In addition the reference structures REF-I and REF-III have direct BndashB bond lengths of 167 Aring
and 166 Aring respectively which is shorter by 014 Aring than a typical BndashB bond length The calculated
bond angles OBO in B3O3 and C2O2B rings are 120deg and 113deg respectively
Figure 3 Layer stackings considered AA AB serrated and inclined
Interlayer distances (d) which is the shortest distance between two layers (equivalent to c for AA
c2 for AB and serrated) are different in all kinds of stacking AB stacked 2D COFs have shorter
interlayer distances than the corresponding AA serrated and inclined stacked structures Among the
latter three AA stacked COFs have higher values for d because of the higher repulsive interlayer
orbital interactions resulting from the direct overlap of polarized alike atoms between the adjacent
layers This results in higher mass densities for AB stacked COF analogues Serrated and inclined
stacks have only slightly higher mass densities compared to AA The differences in mass densities for
all kinds of stacking are attributed to the differences in their interlayer separations The d values of
most of the COFs are larger than that of graphite in AA stacking but smaller in AB stacking
Cell parameters (a) and mass densities (ρ) of all the COFs constructed from the considered
connectors and linkers are shown in Table 1 (and in Table S3 in Supporting Information File 1) Mass
densities of all the COFs are much lower than that of graphite (227 gmiddotcmminus3) and diamond (350
gmiddotcmminus3) AAserratedinclined stacked COF-10s have the lowest mass densities (045046046
gmiddotcmminus3) which is lower than that of MOF-5 (059 gmiddotcmminus3) [4] and comparable to that of highly porous
MOF-177 (042 gmiddotcmminus3) [18]
110
In order to identify the stacking orders we have analyzed X-ray diffraction (XRD) patterns of the well-
known COFs (COF-10 TP COF PPy-COF see Figure 4) in all the above discussed stacking kinds The
change of stacking should be visible in XRDs because each space group has a distinct set of symmetry
imposed reflection conditions The XRD patterns of AA serrated and inclined stacking kinds which
differ within a slight shift of adjacent layers to specific directions are in agreement with the presently
available experimental data [81314] Their peaks are at the same angles as in the experimental
spectrum whereas AB stacking clearly shows differences The slight differences in the (001) angle
between each stacking resemble the differences in their interlayer separations The inclined
stackings have more peaks however they are covered by the broad peaks in the experimental
patterns Similar results for COF-1 COF-5 COF-6 and COF-8 have been discussed in our previous
work [17]
Figure 4 The calculated and experimental [81314] XRD patterns of PPy-COF (top) COF-10 (middle) and TP COF (bottom)
111
Table 1 The calculated unit cell parameter a [Aring] interlayer distance d [Aring] and mass density ρ [gmiddotcmminus3
] for AA and AB stacked COFs Note that the cell parameter a is the same for all stacking types Experimental data [719] is given in parentheses
COF Building
Blocks
a d ρ
AA AB AA AB
COF-1 I-a 1502 (15620) 351 313 (332) 094 106
COF-1M I-b 2241 349 304 068 078
COF-2M I-c 1492 347 312 095 106
COF-3M I-d 0747 349 327 153 164
PPy-COF I-e 2232 (22163) 349 (3421) 297 084 099
COF-5 II-a 3014 (30020) 347 (3460) 326 056 060
COF-10 II-b 3758 (37810) 347 (3476) 318 045 (045) 050
COF-8 II-c 2251 (22733) 346 (3476) 320 071 (070) 077
COF-6 II-d 1505 (15091) 346 (3599) 327 104 110
TP COF II-e 3750 (37541) 348 (3378) 320 051 056
COF-4M III-a 2171 350 318 073 080
COF-5M III-b 2915 350 318 055 061
COF-6M III-c 1833 345 318 083 090
COF-7M III-d 1083 350 330 129 136
TP COF-1M III-e 2905 349 310 065 074
COF-8M IV-a 1748 359 329 140 148
COF-9M IV-b 2176 349 330 117 174
COF-10M IV-c 2254 342 336 127 128
COF-11M IV-d 1512 346 338 168 172
TP COF-2M IV-e 2173 347 332 134 140
REF-I I 0773 359 336 144 148
REF-III III 1445 353 336 104 121
Graphite 247 343 335 220 227
112
Energetic stability
We have considered dehydration reactions the basis of experimental COF synthesis to calculate
formation energies of COFs in order to predict their energetic stability Molecular units 14-
phenylenediboronic acid (BDBA) 11rsquo-biphenyl]-44-diylboronic acid (BPDA) 5rsquo-(4-boronophenyl)-
11rsquo3rsquo1rdquo-terphenyl]-44rdquo-diboronic acid (BTPA) benzene-135-triyltriboronic acid (BTBA) and
pyrene-27-diylboronic acid (PDBA) were considered as linkers andashe respectively with -B(OH)2 groups
attached to each point of extension (Figure 5) Self-condensation of these building blocks result in
the formation of B3O3 rings and the resultant COFs are those made of connector I and the
corresponding linkers This process requires a release of three or six water molecules in case of t or l
topology respectively
Figure 5 The reactants participating in the formation of 2D COFs
Co-condensation of the above molecular units with compounds such as 23671011-
hexahydroxytriphenylene (HHTP) hexahydroxybenzene (HHB) and dodecahydroxycoronene (DHC)
(Figure 5) gives rise to COFs made of connectors II III and IV respectively and the corresponding
linkers Self-condensation of tetrahydroxydiborane (THDB) and co-condensation of HHB with THDB
result in the formation of the reference structures REF-I and REF-III respectively In relation to the
corresponding connectorlinker topologies these building blocks satisfy the following equations of
the co-condensation reaction for COF formation
2 2 3 COF 12 H O Tl T l (1)
113
2 1 1 COF 6 H O Tt T t (2)
2 1 3 COF 12 H O Hl H l (3)
2 1 2 COF 12 H O Ht H t (4)
with a stochiometry appropriate for one unit cell The number of covalent bonds formed between
the building blocks is equivalent to the number of released water molecules we refer to it as
ldquoformula unitrdquo and will give all energies in the following in kJmiddotmolminus1 per formula unit
Table 2 The calculated energies [kJ molminus1
] per bond formed between building blocks for AA and AB stacked COFs Ecb is the condensation energy Esb is the stacking energy and Efb is the COF formation energy (Efb = Ecb
+ Esb) The calculated band gaps Δ eV+ are given as well
COF Building
Blocks
Mono-
layer
AA AB
Ecb Esb Efb ∆ Esb Efb ∆
COF-1 I-a 906 -2683 -1777 33 -2187 -1280 36
COF-1M I-b 949 -4266 -3366 27 -2418 -1469 31
COF-2M I-c 956 -5727 -4771 28 -4734 -3778 30
COF-3M I-d 763 -2506 -1742 38 -2801 -2037 40
PPy-COF I-e 858 -5723 -4866 24 -3855 -2998 26
COF-5 II-a 211 -2968 -2756 24 -2548 -2337 28
COF-10 II-b 317 -3766 -3448 23 -1344 -1026 26
COF-8 II-c 263 -4488 -4224 25 -2477 -2213 28
COF-6 II-d 185 -2881 -2695 28 -2127 -1942 31
TP COF II-e 231 -4453 -4222 24 -1480 -1250 27
COF-4M III-a -033 -1730 -1763 26 -880 -913 26
COF-5M III-b 007 -2533 -2526 25 -972 -965 25
COF-6M III-c 014 -3231 -3217 26 -2134 -2120 28
114
COF-7M III-d -170 -1635 -1805 30 -1607 -1777 32
TP COF-1M III-e -014 -3226 -3240 24 -1277 -1291 24
COF-8M IV-a -787 -2756 -3543 18 -2680 -3467 21
COF-9M IV-b -836 -3577 -4414 17 -3003 -3839 21
COF-10M IV-c -947 -4297 -5244 18 -4192 -5140 22
COF-11M IV-d -403 -2684 -3087 21 -2833 -3236 24
TP COF-2M IV-e 030 -4345 -4315 18 -4117 -4087 21
We have calculated the condensation energy of a single COF layer formed from monomers (building
blocks hereafter called bb) according to the above reactions as follows
tot tot tot totcb m H2O 1 bb1 2 bb2E E n E ndash m E m E (5)
where Emtot ndash total energy of the monolayer EH2O
tot is the total energy of the water molecule Ebb1tot
and Ebb2tot are the total energies of interacting building blocks and n m1 m2 are the corresponding
stoichiometry numbers
We have also calculated the stacking energy Esb of layers
tot totsb nl s mE E n E (6)
where Enltot is the total energy of ns number of layers stacked in a COF Finally the COF formation
energy can be given as a sum of Ecb and Esb (see Table 1 and Table S1)
Electronic properties
All COFs including the reference structures are semiconductors with their band gaps lying between
17 eV and 40 eV (Table 2 and Table S4 in Supporting Information File 1) The largest band gaps are
of the reference structures while the lowest values are of COFs built from connector IV The band
gaps are different for different stacking kinds This difference can be attributed to the different
optimized interlayer distances Generally AB serrated and inclined stacked COFs have band gaps
comparable to or larger than that of their AA stacked analogues
115
We have calculated the electronic total density of states (TDOS) and the individual atomic
contributions (partial density of states PDOS) The energy state distributions of COFs and their
monolayers are studied and a comparison for COF-5 is shown in Figure 6 In all stacking kinds
negligible differences are found for the densities at the top of valence band and the bottom of
conduction band These slight differences suggest that the weak interaction between the layers or
the overlap of π-orbitals does not affect the electronic structure of COFs significantly Hence there is
almost no difference between the TDOS of AA AB serrated and inclined stacking kinds Therefore in
the following we discuss only the AA stacked structures
Figure 6 Total densities of states (DOS) (black) of AA (top left) AB (top right) serrated (bottom left) and inclined (bottom right) comparing stacked COF-5 with a monolayer (red) of COF-5 The differences between the TDOS of bulk and monolayer structures are indicated in green The Fermi level EF is shifted to zero
Figure 7 Partial density of states of carbon (left) boron (center) and oxygen (right) atoms of COFs built from type-I connectors and different linkers The vertical dashed line in each figure indicates the Fermi level EF
116
It is of interest to see how the properties of COFs change depending on their composition of different
secondary building units that is for different connectors and linkers PDOS of COFs built from type-I
connectors and different linkers are plotted in Figure 7 The PDOS of carbon atoms is compared with
that of graphite (AA-stacking) while PDOS of boron and oxygen atoms are compared with that of
REF-I a structure which is composed solely of connector building blocks Going from top to bottom
of the plots the number of carbon atoms per unit cell increases It can be seen that this causes a
decrease of the band gap Figure 8 shows the PDOS of COFs built from type-a linkers and different
connectors where the COFs are arranged in the increasing number of carbon atoms in their unit cells
from top to the bottom Again the C PDOS is compared with that of graphite while both REF-I and
REF- III are taken in comparison to O and B PDOS The observed relation between number of carbon
atoms and band gap is verified
Figure 8 Partial density of states of carbon (left) boron (center) and oxygen (right) atoms of COFs built from type-a linkers The vertical dashed line in each figure indicates the Fermi level EF
Conclusion
In summary we have designed 2D COFs of various topologies by connecting building blocks of
different connectivity and performed DFTB calculations to understand their structural energetic and
electronic properties We have studied each COF in high-symmetry AA and AB as well as low-
symmetry inclined and serrated stacking forms The optimized COF structures exhibit different
interlayer separations and band gaps in different kinds of layer stackings however the density of
states of a single layer is not significantly different from that of a multilayer The alternate shifted
layers in AB serrated and inclined stackings cause less repulsive orbital interactions within the layers
which result in shorter interlayer separation compared to AA stacking All the studied COFs show
117
semiconductor-like band gaps We also have observed that larger number of carbon atoms in the
unit cells in COFs causes smaller band gaps and vice versa Energetic studies reveal that the studied
structures are stable however notable difference in the layer stacking is observed from
experimental observations This result is also supported by simulated XRDs
Methods
We have optimized the atomic positions and the lattice parameters for all studied COFs All
calculations were carried out at the Density Functional Tight-Binding (DFTB) [2021] level of theory
DFTB is based on a second-order expansion of the KohnndashSham total energy in the Density-Functional
Theory (DFT) with respect to charge density fluctuations This can be considered as a non-orthogonal
tight-binding method parameterized from DFT which does not require large amounts of empirical
parameters however maintains all the qualities of DFT The main idea behind this method is to
describe the Hamiltonian eigenstates with an atomic-like basis set and replace the Hamiltonian with
a parameterized Hamiltonian matrix whose elements depend only on the inter-nuclear distances and
orbital symmetries [21] While the Hamiltonian matrix elements are calculated using atomic
reference densities the remaining terms to the KohnndashSham energy are parameterized from DFT
reference calculations of a few reference molecules per atom pair The accuracy is very much
improved by the self-consistent charge (SCC) extension Two computational codes were used
deMonNano code [22] and DFTB+ code [23] The first code has dispersion correction [24]
implemented to account for weak interactions and was used to obtain the layered bulk structure of
COFs and their formation energies The performance for interlayer interactions has been tested
previously for graphite [24] The second code which can perform calculations using k-points was
used to calculate the electronic properties (band structure and density of states) Band gaps have
been calculated as an additional stability indicator While these quantities are typically strongly
underestimated in standard LDA- and GGA-DFT calculations they are typically in the correct range
within the DFTB method For validation of our method we have calculated some of the structures
using Density Functional Theory (DFT) as implemented in ADF code [2526]
Periodic boundary conditions were used to represent frameworks of the crystalline solid state The
conjugatendashgradient scheme was chosen for the geometry optimization The atomic force tolerance of
3 times 10minus4 eVAring was applied The optimization using Γ-point approximation was performed with the
deMon-Nano code on 2times2times4 supercells Some of the monolayers were also optimized using the
DFTB+ code on elementary unit cells in order to validate the calculations within the Γ-point
approximation The number of k-points has been determined by reaching convergence for the total
energy as a function of k-points according to the scheme proposed by Monkhorst and Pack [27]
118
Band structures were computed along lines between high symmetry points of the Brillouin zone with
50 k-points each along each line XRD patterns have been simulated using Mercury software [2829]
We have also performed first-principles DFT calculations at the PBE [30] DZP [31] level to support
our results quantitatively For simplicity we have used finite structures instead of bulk crystals
Supporting Information
Table S1 The calculated bond lengths and angles of all studied COFs Corresponding experimental values found from the literature are shown in brackets
COF Building
Blocks
C-B B-O O-C OBO
COF-1 I-a 1498 (1600) 1393 (1509) 1203 (1200)
COF-1M I-b 1497 1393 1203
COF-2M I-c 1497 1392 1203
COF-3M I-d 1496 1392 1201
PPy-COF I-e 1498 1393 1202 (1190)
COF-5 II-a 1496 (1530) 1399 (1367) 1443 (1384) 1135dagger (1139)
COF-10 II-b 1495 (1553) 1399 (1465) 1443 (1385) 1135dagger (1120)
COF-8 II-c 1495 (1548) 1399 (1479) 1443 1135dagger
COF-6 II-d 1496 1399 1443 1135dagger
TP COF II-e 1496 (1542) 1399 (1396) 1444 (1377) 1135dagger (1182)
COF-4M III-a 1496 1398 1449 1135dagger
COF-5M III-b 1496 1398 1449 1136dagger
COF-6M III-c 1496 1399 1451 1134dagger
COF-7M III-d 1496 1398 1449 1136dagger
TP COF-1M III-e 1496 1398 1450 1136dagger
COF-8M IV-a 1496 1398 1445 1131dagger
COF-9M IV-b 1495 1398 1444 1131dagger
119
COF-10M IV-c 1495 1391 1418 1126dagger
COF-11M IV-d 1498 1399 1450 1134dagger
TP COF-2M IV-e 1499 1399 1447 1134dagger
B3O3 connectivity dagger C2B2O connectivity
It can be noticed that the bond lengths of experimentally known COF-1 is much larger compared to
our optimized bond lengths as well as that of other synthesized COFs
Figure S2 Calculated C-C bond lengths in connectors and linkers Hydrogen atoms are omitted for clarity
Table S3 The calculated unit cell parameters a interlayer distance d Aring+ and mass density ρ gcm-3
] for serrated (Sa serrated armchair Sz serrated zigzag) and inclined (Ia inclined armchair Iz inclined zigzag) stacked COFs
COF Building
Blocks
a d ρ
Sa Sz Ia Iz Sa Sz Ia Iz
COF-1 I-a 1502 343 343 097 097
COF-1M I-b 2241 341 342 069 069
COF-2M I-c 1492 340 339 097 097
COF-3M I-d 0747 341 342 157 156
PPy-COF I-e 2232 341 341 086 086
120
COF-5 II-a 3014 342 342 341 340 057 057 058 058
COF-10 II-b 3758 341 341 342 340 046 046 046 046
COF-8 II-c 2251 341 341 342 342 073 073 072 072
COF-6 II-d 1505 342 341 340 340 105 106 106 106
TP COF II-e 3750 342 341 342 342 052 052 052 052
COF-4M III-a 2171 344 344 345 344 074 074 074 074
COF-5M III-b 2915 343 342 343 343 056 056 056 056
COF-6M III-c 1833 341 341 342 341 084 084 084 084
COF-7M III-d 1083 344 343 340 344 131 131 132 131
TP COF-1M III-e 2905 343 342 343 342 066 067 066 066
COF-8M IV-a 1748 341 341 342 342 142 142 142 142
COF-9M IV-b 2176 341 341 341 342 119 119 119 119
COF-10M IV-c 2254 340 340 340 340 128 128 128 128
COF-11M IV-d 1512 341 341 340 340 171 171 171 171
TP COF-2M IV-e 2173 340 340 340 340 137 137 137 137
REF-I I 0773 349 345 148 15
REF-III III 1445 348 349 106 106
Table S4 The calculated energies [kJ mol-1
] per bond formed between building blocks for serrated (Sa serrated armchair Sz serrated zigzag) and inclined (Ia inclined armchair Iz inclined zigzag) stacked COFs Ecb ndash condensation energy Esb ndash stacking energy and Efb ndash COF formation energy (Efb = Ecb + Esb) The calculated band gaps ∆ eV+ are given as well
COF Sa Sz Ia Iz
Esb Efb Δ Esb Efb Δ Esb Efb Δ Esb Efb Δ
-1 -2810 -1904 36 -2786 -1880 36
-1M -4426 -3477 30 -4389 -3440 30
-2M -5967 -5011 30 -5833 -4877 30
121
-3M -2667 -1904 40 -2591 -1828 40
PPy- -5916 -5058 26 -5865 -5007 26
-5 -3052 -2841 27 -3051 -2840 26 -3275 -3064 26 -3297 -3086 26
-10 -3868 -3551 26 -3869 -3552 25 -3830 -3513 26 -4293 -3976 25
-8 -4615 -4352 27 -4614 -4351 27 -4571 -4308 27 -4573 -4310 27
-6 -2978 -2793 30 -2978 -2793 30 -3117 -2932 30 -3090 -2905 30
TP -4557 -4326 26 -4562 -4331 26 -4511 -4280 26 -4511 -4280 26
-4M -1811 -1844 28 -1813 -1846 28 -1769 -1802 28 -1880 -1913 28
-5M -2626 -2619 27 -2630 -2623 26 -2597 -2590 26 -2600 -2593 26
-6M -3375 -3361 28 -3381 -3367 28 -3487 -3473 28 -3349 -3335 28
-7M -1724 -1894 32 -1726 -1896 31 -2333 -2503 32 -1866 -2036 31
TP -1M -3327 -3341 26 -3334 -3348 26 -3340 -3354 26 -3353 -3367 26
-8M -2897 -3684 21 -2895 -3682 21 -2971 -3758 21 -2983 -3770 21
-9M -3732 -4568 20 -3733 -4569 20 -3684 -4520 20 -3706 -4542 20
-10M -4512 -5459 21 -4513 -5460 21 -4491 -5438 21 -4495 -5442 21
-11M -2831 -3234 24 -2834 -3237 24 -2816 -3219 24 -2830 -3233 24
TP -2M -4500 -4470 20 -4505 -4475 20 -4495 -4465 20 -4496 -4466 20
122
Appendix E
Stability and electronic properties of 3D covalent organic frameworks
Binit Lukose Agnieszka Kuc and Thomas Heine
Prepared for publication
Abstract Covalent Organic Frameworks (COFs) are a class of covalently linked crystalline nanoporous
materials versatile for nanoelectronic and storage applications 3D COFs in particular have very
large pores and low-mass densities Extensive theoretical studies of their energetic and mechanical
stability as well as their electronic properties are discussed in this paper All studied 3D COFs are
energetically stable materials though mechanical stability does not exceed 20 GPa Electronically all
COFs are semiconductors with band gaps depending on the number of sp2 carbon atoms present in
the linkers similar to 3D MOF family
Introduction
Covalent Organic Frameworks (COFs)1-3 comprise an emerging class of crystalline materials that
combines organic functionality with nanoporosity COFs have organic subunits stitched together by
covalent entities including boron carbon nitrogen oxygen or silicon atoms to form periodic
frameworks with the faces and edges of molecular subunits exposed to pores Hence their
applications can range from organic electronics to catalysis to gas storage and sieving4-7 The
properties of COFs extensively depend on their molecular constituents and thus can be tuned by
rational chemical design and synthesis289 Step by step reversible condensation reactions pave the
123
way to accomplish this target Such a reticular approach allows predicting the resulting materials and
leads to long-range ordered crystal structures
Since their first mentioning in the literature13 COFs are under theoretical investigation10-17 mainly for
gas storage applications Methods such as doping18-24 and organic linker functionalization2526 have
been suggested to improve their storage capacities In addition to the moderate pore size and
internal surface area COFs have the privileges of a low-weight material as they are made of light
elements Hence their gravimetric adsorption capacity is remarkably high10 The lowest mass density
ever reported for any crystalline material is that for COF-108 (018 gcm-3) Also their stronger
covalent bonds compared to related Metal-Organic Frameworks (MOFs)27 endorse thermodynamic
strength These genuine qualities of COFs make them attractive for hydrogen storage investigations
Crystallization of linked organic molecules into 2D and 3D forms was achieved in the years 20051 and
20073 respectively by the research group of O M Yaghi Several COFs have been synthesized since
then and some of the 2D COFs has been proven good for electronic or photovoltaic applications428-33
However the growth in this area appears to be slow compared to rapidly developing MOFs albeit
the promising H2 adsorption measurements53435 and a few synthetic improvements736-42
COF-102 and COF-103 were synthesized3 by the self-condensation of the non-planar tetra(4-
dihydroxyborylphenyl)methane (TBPM) and tetra(4-dihydroxyborylphenyl)silane (TBPS) respectively
(see Fig 1 for the building blocks and COF geometries) The co-condensation of these compounds
with triangular hexahydroxytriphenylene (HHTP) results in COF-108 and COF-105 respectively with
different topologies Yaghi et al3 have reported the formation of highly symmetric topologies ctn
(carbon nitride dI 34 ) and bor (boracite mP 34 ) when tetrahedral building blocks are linked
together with triangular ones The topology names were adopted from reticular chemistry structure
resource (RCSR)43 Simulated Powder X-ray diffraction in comparison with experimental powder
spectrum suggested ctn topology for COF-102 103 and 105 and bor topology for COF-108 The
condensation of tert-butylsilanetriol (TBST) and TBPM leads to the formation of COF-20244 This was
reported as ctn topology The COF reactants and schematic diagrams of ctn and bor topologies are
given in Figure1 The assembly of tetrahedral and linear units is very likely to result in a diamond-like
form COF-300 was formed according to the condensation of tetrahedral tetra-(4-anilyl)methane
(TAM) and linear terephthaldehyde (TA)45 However the synthesized structure was 5-fold
interpenetrated dia-c5 topology43
In this work we present theoretical studies of 3D COFs using density functional based methods to
explore their structural electronic energetic and mechanical properties Our previous studies on 2D
COFs4647 had resulted in questioning the stability of eclipsed arrangement of layers (AA stacking) and
124
suggesting energetically more stable serrated and inclined packing In this paper we attempt to
explore the stability and electronic properties of the experimentally known 3D COFs namely COF-
102 103 105 108 202 and 300 In the present studies we follow the reticular assembly of the
molecular units that form low-weight 3D COFs Considering the structural distinction from MOFs
COFs lack the versatility of metal ingredients what results in diverse properties48 A collective study is
then carried out to understand the characteristics and drawbacks of COFs
Figure 1 (Upper panel) Building blocks for the construction of 3D COFs (Lower panel) lsquoctnrsquo and lsquoborrsquo
networks formed by linking tetrahedral and triangular building units
Methods
COF structures were fully optimized using Self-consistent Charge -Density Functional based Tight-
Binding (SCC-DFTB) level of theory4950 Two computational codes were used deMonNano51 and
125
DFTB+52 The first code which has dispersion correction53 implemented to account for weak
interactions was used for the geometry optimization and stability calculations The second code
which can perform calculations using k-point sampling was used to calculate the electronic
properties (band structure and density of states) The number of k-points has been determined by
reaching convergence for the total energy as a function of k-points according to the scheme
proposed by Monkhorst and Pack54 Periodic boundary conditions were used to represent
frameworks of the crystalline solid state Conjugatendashgradient scheme was chosen for the geometry
optimization Atomic force tolerance of 3 times 10minus4 eVAring was applied The optimization using Γ-point
approximation was performed on rectangular supercells containing more than 1000 atoms For
validation of our method we have calculated energetic stability using Density Functional Theory (DFT)
at the PBE55DZP56 level as implemented in ADF code5758 using cluster models The cluster models
contain finite number of building units and correspond to the bulk topology of the COFs XRD
patterns have been simulated using Mercury software5960
In this work we continued to use the traditional nomenclature of the experimentally known COFs All
of the structures have the same tetrahedral blocks and differ only in the central sp3 atom (carbon or
silicon) that is included in our nomenclature
Bulk modulus (B) of a solid at absolute zero can be calculated as
(1) B = 2
2
dV
EdV
where V and E are the volume and energy respectively
Owing to the dehydration reactions we have calculated the formation (condensation) energy of each
COF formed from monomers (building blocks) as follows
(2) EF = Etot + n EH2Otot ndash (m1 Ebb1
tot + m2 Ebb2tot)
where Etot -- total energy of the COF EH2Otot -- total energy of water molecule Ebb1
tot and Ebb2tot -- total
energies of interacting building blocks n m1 m2 -- stoichiometry numbers
Results and Discussions
Structure and Stability
We have optimized the atomic positions and cell dimensions of the COFs in the experimentally
determined topologies Cell parameters in comparison with experimental values are given in Table 1
The calculated bond lengths are C-B = 150 C-C = 139-144 (COF-300) C(sp3)-C = 156 C-O = 142 B-
126
O = 140 Si-C = 188 Si-O = 186 C-N = 131-136 Aring These values agree within 6 error with the
experimental values34445
Mass densities (see Table 1) of COFs range between 02 to 05 g cm-3 and are smaller for the silicon at
the tetrahedral centers This implies that the presence of silicon atoms impacts more on the cell
volume than on the total mass That means the replacement of sp3 C with Si in COF-108 can change
its mass density to a slightly lower value To our best knowledge among all the natural or
synthesized crystals COF-108 has the lowest mass-weight
In order to validate our optimized structures we have simulated X-ray Diffraction of each COF and
compared them with the available experimental spectra (see Figure2) Almost all of the simulated
XRDs have excellent correlation in the peak positions with the experiment Only COF-300 shows
somehow significant differences in the intensities These differences may be attributed to the
presence of guest molecules in the synthesized COF-30045
Table 1 The calculated cell parameters [Aring] mass density ρ gcm-3
+ band gap Δ eV+ bulk modulus B GPa+
and formation energy EF [kJ mol-1
] for all the studied 3D COFs Experimental values are given in brackets
along with the cell parameters HOMO-LUMO gaps [eV] of the building blocks are also given in brackets
along with the band gaps
Structure Building
Blocks
Cell
parameters
ρ Δ B EF
COF-102 (C) TBPM 2707 (2717) 043 39 (43) 206 -14995
COF-103 (Si) TBPS 2817 (2825) 039 38 (41) 139 -14547
COF-105-C TBPM HHTP 4337 019 33 (43 34) 80 -18080
COF-105 (Si) TBPS HHTP 4444 (4489) 018 32 (41 34) 79 -17055
COF-108 (C) TBPM HHTP 2838 (2840) 017 32 (43 34) 37 -17983
COF-108-Si TBPS HHTP 2920 016 30 (41 34) 29 -18038
COF-202 (C) TBPM TBST 3047 (3010) 050 42 (43 139) 143 -7954
COF-202-Si TBPS TBST 3157 046 39 (41 139) 153 -7632
COF-300 (C) TAM TA 2932 925 049 23 (41 26) 144 -40286
127
(2828 1008)
COF-300-Si TAS TA 3039 959 044 23 (40 26) 133 -39930
tetra-(4-anilyl)silane
Figure 2 Simulated XRDs are compared with the experimental patterns from the literature COF-300
exhibits some differences between the simulated and experimental XRDs while others show reasonably
good match
The bulk modulus (B) is a measure of mechanical stability of the materials The calculated values of B
are given in Table 1 Compared to the force-field based calculation of bulk modulus by Schmid et
al61 these values are larger The bulk modulus of COF-105 and COF-108 are relatively small
compared with other COFs Considering that the two COFs differ only in the topology it may be
concluded that ctn nets are mechanically more stable compared to bor COFs with carbon atoms in
the center are mechanically more stable than those with silicon atoms Bulk modulus of COF-102
103 COF-202 and interpenetrated COF-300 are higher than many of the isoreticular MOFs62 and
comparable to IRMOF-6 (1241 GPa) MOF-5 (1537 GPa)63 bNon-interpenetrated COF-300 (single
framework dia-a topology43) has much lower bulk modulus of only 317 GPa
Calculated formation energies (Table 1) are given in terms of reaction energies according to Eq 2
Condensation of monomers to form bulk 3D structures is exothermic in all the cases supporting
reticular approach The presence of C or Si at the vertex center does not show any particular trend in
the formation energies We have calculated the formation energy of non-interpenetrated COF-300
(dia-a) to be -39346 kJ mol-1 which is comparable to the interpenetrated cases For a quantitative
comparison we have performed DFT calculations at the PBEDZP level as implemented in ADF code
on finite structures The obtained formation energies for COF-105-C COF-105 (Si) COF-108 (C) COF-
108-Si are -9944 -9968 -1245 -12436 respectively They are in a reasonable agreement with the
128
DFTB results A comparison between COF-105 and COF-108 suggests that bor nets are energetically
more favored than ctn nets
Electronic Properties
Band gaps (Δ) calculated for the 3D COFs are in the range of 23-42 eV (see Table 1) which show
their semiconducting nature similar to the hexagonal 2D COFs47 and MOFs62 The band gap
decreases with the increase of conjugated rings in the unit cell relative to the number of other atoms
Si atoms in comparison with carbon atoms at the tetrahedral centers give a relatively smaller Δ This
is evident from the partial density of states of C (sp3) in COF-102 and Si in COF-103 plotted in Figure3
Carbon atoms define the band edges of the PDOS (see Figure4) The band gaps of COF-105 and COF-
108 differ only slightly (see Figure5) which means that the band gap is nearly independent of the
topology This is because for each atom the coordination number and the neighboring atoms remain
the same in both ctn and bor networks albeit the topology difference DOS of non-interpenetrated
(dia-a) and 5-fold interpenetrated (dia-c5) COF-300 are plotted in Figure6 It may be concluded from
their negligible differences that interpenetration does not alter the DOS of a framework We have
shown similar results for 2D COFs47
Figure 3 Partial density of states (DOS) of sp3 C in COF-102 (top) and Si in COF-103 (bottom) The latter is
inverted for comparison The Fermi level EF is shifted to zero
129
Figure 4 Partial and total density of states of COF-103 The Fermi level EF is shifted to zero
Figure 5 Density of states of COF-108 (bor) and COF-105 (ctn) The two COFs differ only in the topology
130
Figure 6 Density of states of non-interpenetrated (dia-c1) and 5-interpenetrated (dia-c5) COF-300
We also have calculated HOMO-LUMO gaps of the molecular building blocks (see Table 1) In
comparison band gaps of COFs are slightly smaller than the smallest of the HOMO-LUMO gaps of the
building units
Conclusion
In summary we have calculated energetic mechanical and electronic properties of all the known 3D
COF using Density Functional Tight-Binding method Energetically formation of 3D COFs is favorable
supporting the reticular chemistry approach Mechanical stability is in line with other frameworks
materials eg MOFs and bulk modulus does not exceed 20 GPa Also all COFs are semiconducting
with band gaps ranging from 2 to 4 eV Band gaps are analogues to the HOMO-LUMO gaps of the
molecular building units We believe that this extensive study will define the place of COFs in the
broad area of nanoporous materials and the information obtained from the work will help to
strategically develop or modify porous materials for the targeted applications
131
Appendix F
Structure electronic structure and hydrogen adsorption capacity of porous aromatic frameworks
Binit Lukose Mohammad Wahiduzzaman Agnieszka Kuc and Thomas Heine
Prepared for publication
Abstract
Diamond-like porous aromatic frameworks (PAFs) form a new family of materials that contain only
carbon and hydrogen atoms within their frameworks These structures have very low mass densities
large surface area and high porosity Density-functional based calculations indicate that crystalline
PAFs have mechanical stability and properties similar to those of Covalent Organic Frameworks Their
exceptional structural properties and stability make PAFs interesting materials for hydrogen storage
Our theoretical investigations show exceptionally high hydrogen uptake at room temperature that
can reach 7 wt a value exceeding those of well-known metal- and covalent-organic frameworks
(MOFs and COFs)
Introduction
Porous materials have been widely investigated in the fields of materials science and technology due
to their applications in many important fields such as catalysis gas storage and separation template
materials molecular sensors etc1-3 Designing porous materials is nowadays a simple and effective
strategy following the approach of reticular chemistry4 where predefined building blocks are used to
132
predict and synthesize a topological organization in an extended crystal structure The most famous
and striking examples of such materials are Metal- and Covalent-Organic Frameworks (MOFs and
COFs)56 These new nanoporous materials have many advantages high porosity and large surface
areas lowest mass densities known for crystalline materials easy functionalization of building blocks
and good adsorption properties
Gas storage and separation by physical adsorption are very important applications of such
nanoporous materials and have been major subjects of science in the last two decades These
applications are based on certain physical properties namely presence of permanent large surface
area and suitable enthalpy of adsorption between the host framework and guest molecules
Attempts to produce materials with large internal surface area have been successful and some of the
notable accomplishments include the synthesis of MOF-2107 (BET surface area of 6240 m2 g-1 and
Langmuir surface area of 10400 m2 g-1) NU-1008 (BET surface area 6143 m2 g-1) or 3D COFs9 (BET
surface area 4210 m2 g-1 for COF-103)
More recently a new family of porous materials emerged So-called porous-aromatic frameworks
(PAFs) have surface areas comparable to MOFs eg PAF-110 has a BET surface area of 5640 m2 g-1 and
Langmuir surface area of 7100 m2 g-1 Since they are composed of carbon and hydrogen only they
have several advantages over frameworks containing heavy elements MOFs with coordination bonds
often suffer from low thermal and hydrothermal stability what might limit their applications on the
industrial scale The coordination bonds can be substituted with stronger covalent bonds as it was
realized in the case of COFs6 however this lowers significantly their surface areas comparing with
MOFs
Besides its exceptional surface area PAF-110 has high thermal and hydrothermal stability and
appears to be a good candidate for hydrogen and carbon dioxide storage applications PAFs have
topology of diamond with C-C bonds replaced by a number of phenyl rings (see Figure 1)
Experimentally obtained PAF-110 and PAF-1111 have eg one and four phenyl rings respectively
connected by tetragonal centers (carbon tetrahedral nodes) The simulated and experimental
hydrogen uptake capacities of such PAFs exceed the DOE target12
In this paper we have studied structural electronic and adsorption properties of PAFs using Density
Functional based Tight-Binding (DFTB) method1314 and Quantized Liquid Density Functional Theory
(QLDFT)15 We have found exceptionally high storage capacities at room temperature what makes
PAFs very good materials for hydrogen storage devices beyond well-known MOFs and COFs We have
compared our results to widely-used classical Grand Canonical Monte-Carlo (GCMC) calculations
reported in the literature We have also studied other properties of these materials such as
133
structural energetic electronic and mechanical We explored the structural variance of diamond
topology by individually placing a selection of organic linkers between carbon nodes This generally
changes surface area mass density and isosteric heat of adsorption what is reflected in the
adsorption isotherms
Methods
Using a conjugate-gradient method we have fully optimized the crystal structures (atomic positions
and unit cell parameters) of PAFs The calculations were performed using Dispersion-corrected Self-
consistent Charge density-functional based tight-binding (DFTB) method as implemented in the
deMonNano code1617 Periodic boundary conditions were applied to a (3x3x3) supercell thus
representing frameworks of the crystalline solid state Electronic density of states (DOS) have been
calculated using the DFTB+ code18 with k-point sampling where the k space was determined by
reaching convergence for the total energy according to the scheme proposed by Monkhorst and
Pack19
Hydrogen adsorption calculations have been carried out using QLDFT within the LIE-0 approximation
thus including many-body interparticle interactions and quantum effects implicitly through the
excess functional The PAF-H2 non-bonded interactions were represented by the classical Morse
atomic-pair potential Force field parameters were taken from Han et al20 who originally developed
them for studying hydrogen adsorption in COFs that incorporate similar building blocks as PAFs The
authors adopted the approach to the shapeless particle approximation of QLDFT These PAF-H2
parameters have been determined by fitting first-principles calculations at the second-order Moslashllerndash
Plesset (MP2) level of theory using the quadruple-ζ QZVPP basis set with correction for the basis set
superposition error (BSSE) QLDFT calculations use a grid spacing of 05 Bohr radii and a potential
cutoff of 5000 K
Results and Discussion
Design and Structure of PAFs
We have designed a set of PAFs by replacing C-C bonds in the diamond lattice by a set of organic
linkers phenyl biphenyl pyrene para-terphenyl perylenetetracarboxylic acid (PTCDA)
diphenylacetylene (DPA) and quaterphenyl (see Figure 1) We label the respective crystal structures
as PAF-phnl (PAF-301 in Ref 12) PAF-biph (PAF-302 in Ref 12) PAF-pyrn PAF-ptph(PAF-303 in Ref
12) PAF-PTCDA PAF-DPA and PAF-qtph (PAF-304 in Ref 12) respectively Such a design of
frameworks should result in materials with high stability due to the parent diamond-topology and
pure covalent bonding of the network The selected linkers differ in their length width and the
134
number of aromatic rings These should play an important role for hydrogen adsorption properties
aromatic systems exhibit large polarizabilities and interact with H2 molecules via London dispersion
forces Long linkers introduce high pore volume and low mas-weight to the network while wide
linkers offer large internal surface area and high heat of adsorption Hence long linkers are of
advantage for high gravimetric capacity while wide linkers enhance volumetric capacity Proper
optimization of the linker size should result in a perfect candidate for hydrogen storage applications
Figure 1 a) Schematic diagram of the topology of PAFs blue spheres and grey pillars represent carbon
tetragonal centers and organic linkers respectively b) Linkers used for the design of PAFs i) phenyl ii)
biphenyl iii) pyrene iv) DPA v) para-Terphenyl vi) PTCDA and vii) quaterphenyl
Selected structural and mechanical properties of the investigated PAF structures are given in Table 1
Frameworks created with the above mentioned linkers have mass densities that range from 085 g
cm-3 (for phenyl) to 01 g cm-3 (for quaterphenyl) The lowest mass-density obtained for a crystal
structure was for COF-108 with = 0171 g cm-39 Our results show that PAF-DPA and PAF-ptph have
mass densities close to the one of COF-108 while PAF-qtph with = 01 g cm-3 exhibits the lowest
for all the PAFs investigated in this study
While the large cell size and the small mass density of PAF-qtph are an advantage for high
gravimetric hydrogen storage capacity other PAFs (eg PAF-pyrn with wide linker) would
compromise gravimetric for high volumetric capacity As both of them are important for practical
applications a balance between them is crucial
Table 1 Calculated cell parameters a (a=b=c α=β=γ=60⁰) mass densities () formation energies (EForm) band
gaps () bulk moduli (B) and H2 accessible free volume and surface area of PAFs studied in the present work
In parenthesis are given HOMU-LUMO gaps of the corresponding saturated linkers
PAFs
a
(Aring)
ρ
(g cm-3)
EForm
(kJ mol-1)
Δ
(eV)
B
(GPa)
H2 accessible
free volume
H2 accessible
surface area
135
() (m2 g-1)
PAF-phnl 97 085 -121 47 (55) 360 35 2398
PAF-biphl 167 032 -122 36 (40) 132 73 5697
PAF-pyrn 166 042 -124 26 (28) 192 66 5090
PAF-DPA 210 019 -122 35 (37) 87 84 7240
PAF-ptph 237 016 -119 32 (33) 56 86 6735
PAF-PTCDA 236 024 -122 18 (19) 95 81 5576
PAF-qtphl 308 010 -119 29 (30) 35 91 7275
Energetic and Mechanical Properties
We have investigated energetic stability of PAFs by calculating their formation energies We regarded
the formation of PAFs as dehydrogenation reaction between saturated linkers and CH4 molecules
For a unit cell containing n carbon nodes and m organic linkers the formation energy (EForm) is given
by
( )
where Ecell EL and
are the total energies of the unit cell saturated linkers CH4 and H2
molecules respectively This excludes the inherent stability of linkers and represents the energy for
coordinating linkers with sp3 carbon atoms during diamond-topology formation All the formation
energies calculated in the present work are given in Table 1 Negative values indicate that the
formation of PAFs is exothermic The values per formula unit do not deviate significantly for different
PAF sizes and shapes
Although diamond is the hardest known material insertion of longer linkers diminishes its
mechanical strength to some extent In order to study the mechanical stability of PAFs we have
calculated their bulk moduli which is the second derivative of the cell energy with respect to the cell
volume (see Table 1) The highest value that we have obtained is for PAF-phnl with B = 36 GPa This is
over ten times smaller than the bulk modulus of pure diamond 514 GPa (calculated at the DFTB
level)21 and 442 GPa (experimental value)22 Bulk modulus should scale as 1V (V - volume) if all
bonds have the same strength We have plotted such a function for PAFs and other framework
136
materials23 in Figure 2 As PAFs have various types of CmdashC bonds there are minor deviations from
the perfect trend
Figure 2 Calculated bulk modulus B as function of the volume per atom PAFs are highlighted in blue and
compared with other framework materials Red curve denotes the fitting to 1V function (V - volume)
The stability of PAF-phnl is similar to that of Cu-BTC MOF and much higher than for other MOFs such
as MOF-5 -177 and -20524 and 3D COFs23 Subsequent addition of phenyl rings decreases B the
lowest value of 35 GPa has been found for PAF-qtph which is in the same order as for COF-108 In
general all the studied PAFs except for PAF-phnl have bulk moduli in line with 3D COFs25 When the
organic linker is extended in the width (cf PAF-biph and PAF-pyrn) the mechanical stability increases
Electronic Properties
All PAF structures are semiconductors similar to COFs The band gaps of PAFs range from 18 to 47
eV (see Table 1) Interesting trends may be observed from the band gaps of this iso-reticular series
In comparison with diamond which has a band gap of 55 eV it may be understood that subsequent
insertion of benzene groups between carbon atoms in diamond decreases the band gap This is easily
understood as the sp3 responsible for the semiconducting character become far apart with large
number of π-electrons in between which tend to close the gap More importantly the values of
band gaps are directly related to the HOMO-LUMO gaps of the corresponding saturated linkers
which are just slightly larger Also more extended linkers (cf PAF-biph and PAF-pyrn or PAF-ptph and
PAF-PTCDA) reduce the band gap
In general conjugated rings reduce the band gap more effectively than CmdashC bond networks (cf PAF-
DPA PAF-ptph or PAF-qtphl) The extreme cases for this behaviour are graphene which is metallic
137
and diamond which is an insulator Overall the band gap may be tuned by placing suitable linkers in
the diamond network Similar results have been reported for MOFs2627
We have calculated the electronic density of states (DOS) for all the PAF structures Figure 3 a shows
carbon DOS of PAFs arranged in such a way that the band gaps decrease going from the top to the
bottom of the figure PAFs with larger band gaps have larger population of energy states at the top of
valence band and bottom of conduction band whereas for linkers with smaller band gaps the
distribution of energy states is rather spread In order to study the impact of sp3 carbon atoms in the
DOS we plotted their contributions against the total carbon DOS in the cases of PAF-phnl and PAF-
pyrn as examples (see Figure 3 b and c) In both cases the sp3 carbons exhibit no energy states in the
band gap and in the close vicinity of band edges
Figure 3 (a) Carbon density of states of all calculated PAFs From top to bottom the value of band gap
decreases (b) and (c) projected DOS on sp3 C atoms of PAF-phnl and PAF-pyrn respectively The vertical
dashed line indicates Fermi level EF
Hydrogen Adsorption Properties
One of the potential applications of PAFs is hydrogen adsorption We have calculated the gravimetric
and volumetric capacities and analyzed them to understand the contributions of the linkers on the
138
hydrogen uptake in different range of temperatures and pressures H2 accessible free volume and
surface area of the PAFs are given in Table 1 In order to assess the excess adsorption capacity the
free pore volume is necessary In our simulation the free pore volume is defined to be that where
the H2-host interaction energy is less than 5000K This covers all sites where H2 is attracted by the
host structure and excludes the repulsion area close to the framework The solvent accessible surface
areas of the PAFs were determined by rolling a probe molecule along the van der Waals spheres of
the atoms A probe molecule of radius 148 Aring was used which is consistent with the Lennard-Jones
sphere of hydrogen and commonly used in various H2 molecular simulations28
Very large surface areas per mass unit have been observed for all PAFs except for PAF-phnl PAF-DPA
and PAF-qtph for example have 7240 and 7274 m2 g-1 H2 accessible surface area respectively For
comparison highly porous MOF-210 and NU-100 give values of 6240 and 6143 m2 g-1 BET surface
areas respectively determined from the experimental adsorption isotherms78 It is worth
mentioning that longer linkers expand the pore and increase the surface area per unit volume and
unit mass Wider linkers provide a higher surface area per unit volume however they possess larger
mass density and hence the surface area per unit mass gets lower
Figure 4 shows the gravimetric and volumetric total and excess adsorption isotherms of PAFs at 77K
The results show that the gravimetric (total and excess) H2 uptake is dependent on the linker length
The longest linker in the PAF-qtph structure gives the total and excess gravimetric capacity of 32 and
128 wt respectively which is larger than for highly porous crystalline 3D COFs29 These numbers
are calculated for the limit pressure of 100 bar For the shortest linker in PAF-phnl we have obtained
only around 25 wt for the same pressure Using grand canonical Monte-Carlo methods (GCMC)
Lan et al12 have predicted total and excess adsorption capacities of PAF-qtph of 224 and 84 wt
respectively The deviations in results are attributed to the differences in both methods where
different force fields are used to describe atom-atom interactions
The volumetric adsorption capacities lower with the size of the linker and the largest uptake we have
found for the PAF-pyrn (46 and 35 kg m-3 total and excess respectively) while very low values were
found for PAF-qtph (30 and 145 kg m-3 total and excess respectively) at 35-bar pressure It could be
predicted that for PAF-phnl the volumetric adsorption reaches the larges values however due to its
very compact crystal structure it reaches saturation at the low-pressure region and does not exceed
30 kg m-3 for the total uptake From the excess adsorption isotherms however PAF-phnl is the best
adsorbing system for low-pressure application as it gets saturated at 11 bar by adsorbing 266 kg m-3
of H2 Generally we have observed that PAFs with lower pore sizes exhibited saturation of volumetric
capacities at lower pressures
139
Figure 4 Calculated H2 uptake of PAFs at 77 K (a)-(b) gravimetric and (c)-(d) volumetric total (upper panel)
and excess (lower panel) respectively
We have also calculated the adsorption performance of PAFs at room temperature The gravimetric
total adsorption isotherms are shown in Figure 5 Similar to the results obtained for T=77 K PAF-
qtph exhibits the highest total gravimetric capacity at room temperature which is as high as 732 wt
at 100 bar Lan et al12 have predicted the uptake of 653 wt at 100 bar using GCMC simulations
These results exceed the 2015 DOE target of 55 wt 3031 On the other hand at more applicable
pressure of 5 bar the uptake reaches only 05 wt This suggests however that the delivery amount
(approximately the difference between the uptakes at 100 and 5 bar) is still more than the DOE
target Phenyl linker at room temperature does not exceed the gravimetric uptake of 1 wt at 100
bar
Figure 5 Calculated gravimetric total (left) and excess (right) H2 uptake of PAFs at 300 K
140
At 300K excess gravimetric capacity is rather low and does not exceed 075 wt even for large
pressure (see Figure 5)
Effects of interpenetration
Very often frameworks with large pores crystallize as interpenetrated lattices In most cases this is
an undesired fact due to reduction of the pore size and free volume For instance COF-300 which
has diamond topology was found to have 5-interpenetrated frameworks32
We have studied the effect of interpenetration in the case of PAF-qtph which has the largest pore
volume among the materials in this study Without any steric hindrance PAF-qtph may be
interpenetrated up to the order of four The two three and four interpenetrated networks are
named as PAF-qtph-2 PAF-qtph-3 and PAF-qtph-4 respectively and in each case the interpenetrated
structures possess translational symmetry Table 2 shows calculated mass densities and H2 accessible
free volume and surface area of non-interpenetrated and n-fold penetrated PAF-qtph Obviously the
mass density increases by one unit for each degree of interpenetration PAF-qtph has 91 of its
volume accessible for H2 which means that 9 of volume is occupied by the skeleton of the PAF
Hence each degree of interpenetration decreases the free volume by 9 H2 accessible surface area
per unit mass remains the same up to 3-fold interpenetration However PAF-qtph-4 results in much
less accessibility for H2
Table 2 Calculated mass densities () and H2 accessible free volume and surface area of non-interpenetrated
and n-fold interpenetrated PAF-qtph where n = 2 3 4
PAF
(g cm-3)
H2 accessible
free volume ()
H2 accessible
surface area
(m2 g-1)
PAF-qtph 010 91 7275
PAF-qtph-2 020 82 7275
PAF-qtph-3 030 73 7275
PAF-qtph-4 040 64 5998
Figure 6 shows the excess and total adsorption isotherms (gravimetric and volumetric) of non-
interpenetrated and interpenetrated PAF-qtph at 77 K With the increasing degree of
141
interpenetration saturation occurs at lower values of pressure This is due to the smaller pore size
resulted from the high-fold interpenetration The excess gravimetric capacity decreases by 18 - 2 wt
per each n-fold interpenetration a result of the decreasing free space around the skeleton It is to be
noted that the difference between the total and the excess adsorption capacities of PAF-qtph is quite
large however it decreases less for interpenetrated structures This is because the interpenetrated
frameworks occupy the free space that otherwise would be taken by hydrogen to increase the total
capacity but not the excess
Figure 6 Calculated H2 uptake at 77 K of non-interpenetrated and n-fold interpenetrated PAF-qtph with n = 2
3 4 (a)-(b) gravimetric and (c)-(d) volumetric total (lower panel) and excess (upper panel) respectively
On the other hand the interpenetrated frameworks have larger volumetric capacity what is easily
understandable due to the volume reduction Significant increase in excess volumetric capacity has
been obtained for PAF-qtph-2 in comparison with PAF-qtph whereas only slightly lower increase was
obtained for PAF-qtph-3 in comparison with PAF-qtph-2 (see Figure 6) PAF-qtph-4 exhibited even
lower increase compared to PAF-qtph-3 suggesting that only a certain degree of interpenetration is
appreciable Although non-interpenetrated PAF has a large amount of H2 gas in the bulk phase due
to the high pore volume its total volumetric uptake does not exceed that of the interpenetrated
PAFs
Almost linear gravimetric isotherms were obtained at room temperature for all studied PAFs
including the interpenetrated cases (see Figure 7) Large decrease of gravimetric capacity was noted
142
when the PAF was interpenetrated from around 7 wt down to 2 wt for 4-fold interpenetrated
PAF-qtph The excess gravimetric capacity however is the same independent of the n-fold
interpenetration and maximum of 06 wt was found for 100-bar pressure (see Figure 7)
Figure 7 Calculated gravimetric total (left) and excess (right) H2 uptake of non-interpenetrated and n-fold
interpenetrated PAF-qtph (n = 2 3 4) at 300 K
Conclusions
Following diamond-topology we have designed a set of porous aromatic frameworks PAFs by
replacing CmdashC bonds by various organic linkers what resulted in remarkably high surface area and
pore volume
Exothermic formation energies of PAFs calculated from the dehydrogenation of the linkers with CH4
indicate strong coordination of linkers in the diamond topology The studied PAFs exhibit bulk moduli
that are much smaller than diamond however in the same order as other porous frameworks such
as MOFs and COFs PAFs are semi-conductors with their band gaps dependent on the HOMO-LUMO
gaps of the linking molecules
Using quantized liquid density functional theory which takes into account inter-particle interactions
and quantum effects we have predicted hydrogen uptake capacities in PAFs At room temperature
and 100 bar the predicted uptake in PAF-qtph was 732 wt which exceeded the 2015 DOE target
At the same conditions other PAFs showed significantly lower uptakes At 77 K the PAFs with similar
pore sizes exhibited similar gravimetric uptake capacities whereas smaller pore size and larger
number of aromatic rings result in higher volumetric capacities In smaller pores saturation of excess
capacities occurred at lower pressures Interpenetrated frameworks largely decrease the amount of
hydrogen gas in the pores and increase the weight of the material however they are predicted to
have high volumetric capacities
143
References
(1) Eddaoudi M Moler D B Li H L Chen B L Reineke T M OKeeffe M Yaghi O M
Accounts of Chemical Research 2001 34 319
(2) Seo J S Whang D Lee H Jun S I Oh J Jeon Y J Kim K Nature 2000 404 982
(3) Ferey G Mellot-Draznieks C Serre C Millange F Accounts of Chemical Research 2005 38
217
(4) Yaghi O M OKeeffe M Ockwig N W Chae H K Eddaoudi M Kim J Nature 2003 423
705
(5) Eddaoudi M Kim J Rosi N Vodak D Wachter J OKeeffe M Yaghi O M Science 2002
295 469
(6) Cote A P Benin A I Ockwig N W OKeeffe M Matzger A J Yaghi O M Science 2005
310 1166
(7) Furukawa H Ko N Go Y B Aratani N Choi S B Choi E Yazaydin A O Snurr R Q
OKeeffe M Kim J Yaghi O M Science 2010 329 424
(8) Farha O K Yazaydin A O Eryazici I Malliakas C D Hauser B G Kanatzidis M G
Nguyen S T Snurr R Q Hupp J T Nature Chemistry 2010 2 944
(9) El-Kaderi H M Hunt J R Mendoza-Cortes J L Cote A P Taylor R E OKeeffe M Yaghi
O M Science 2007 316 268
(10) Ben T Ren H Ma S Cao D Lan J Jing X Wang W Xu J Deng F Simmons J M Qiu
S Zhu G Angewandte Chemie-International Edition 2009 48 9457
(11) Yuan Y Sun F Ren H Jing X Wang W Ma H Zhao H Zhu G Journal of Materials
Chemistry 2011 21 13498
(12) Lan J Cao D Wang W Ben T Zhu G Journal of Physical Chemistry Letters 2010 1 978
(13) Oliveira A F Seifert G Heine T Duarte H A Journal of the Brazilian Chemical Society
2009 20 1193
(14) Seifert G Porezag D Frauenheim T International Journal of Quantum Chemistry 1996 58
185
(15) Patchkovskii S Heine T Physical Review E 2009 80
(16) Heine T Rapacioli M Patchkovskii S Frenzel J Koester A M Calaminici P Escalante S
Duarte H A Flores R Geudtner G Goursot A Reveles J U Vela A Salahub D R deMon-nano edn ed
deMon 2009
(17) Zhechkov L Heine T Patchkovskii S Seifert G Duarte H A Journal of Chemical Theory
and Computation 2005 1 841
(18) BCCMS Bremen DFTB+ - Density Functional based Tight binding (and more)
(19) Monkhorst H J Pack J D Physical Review B 1976 13 5188
(20) Han S S Furukawa H Yaghi O M Goddard III W A Journal of the American Chemical
Society 2008 130 11580
(21) Kuc A Seifert G Physical Review B 2006 74
(22) Cohen M L Physical Review B 1985 32 7988
(23) Lukose B Kuc A Heine T manuscript in preparation 2012
(24) Lukose B Supronowicz B Petkov P S Frenzel J Kuc A Seifert G Vayssilov G N
Heine T physica status solidi (b) 2011
(25) Amirjalayer S Snurr R Q Schmid R Journal of Physical Chemistry C 2012 116 4921
(26) Gascon J Hernandez-Alonso M D Almeida A R van Klink G P M Kapteijn F Mul G
Chemsuschem 2008 1 981
(27) Kuc A Enyashin A Seifert G Journal of Physical Chemistry B 2007 111 8179
(28) Dueren T Millange F Ferey G Walton K S Snurr R Q Journal of Physical Chemistry C
2007 111 15350
(29) Furukawa H Yaghi O M Journal of the American Chemical Society 2009 131 8875
144
(30) US DOE Office of Energy Efficiency and Renewable Energy and The FreedomCAR and
Fuel Partnership 2009
httpwww1eereenergygovhydrogenandfuelcellsstoragepdfstargets_onboard_hydro_storage_explanatio
npdf
(31) US DOE USCAR Shell BP ConocoPhillips Chevron Exxon-Mobil T F a F P Multi-Year
Research Development and Demonstration Plan 2009
httpwww1eereenergygovhydrogenandfuelcellsmypppdfsstoragepdf
(32) Uribe-Romo F J Hunt J R Furukawa H Klock C OKeeffe M Yaghi O M Journal of the
American Chemical Society 2009 131 4570
145
Appendix G
A novel series of isoreticular metal organic frameworks realizing metastable structures by liquid phase epitaxy
Jinxuan Liu Binit Lukose Osama Shekhah Hasan Kemal Arslan Peter Weidler Hartmut
Gliemann Stefan Braumlse Sylvain Grosjean Adelheid Godt Xinliang Feng Klaus Muumlllen Ioan-
Bogdan Magdau Thomas Heine and Christof Woumlll
Prepared for publication
Highly crystalline molecular scaffolds with nm-sized pores offer an attractive basis for the fabrication
of novel materials The scaffolds alone already exhibit attractive properties eg the safe storage of
small molecules like H2 CO2 and CH41-4 In addition to the simple release of stored particles changes
in the optical and electronic properties of these nanomaterials upon loading their porous systems
with guest molecules offer interesting applications eg for sensors5 or electrochemistry6 For the
construction of new nanomaterials the voids within the framework of nanostructures may be loaded
with nm-sized objects such as inorganic clusters larger molecules and even small proteins a
process that holds great potential as for example in drug release7-8 or the design of novel battery
materials9 Furthermore meta-crystals may be prepared by loading metal nanoparticles into the
pores of a three-dimensional scaffold to provide materials with a number of attractive applications
ranging from plasmonics10-12 to fundamental investigations of the electronic structure and transport
properties of the meta-crystals13
146
In the last two decades numerous studies have shown that MOFs also termed porous coordination
polymers (PCPs) fulfill many of the aforementioned criteria Although originally developed for the
storage of hydrogen eg in tanks of automobiles MOFs have recently found more technologically
advanced applications as sensors5 matrices for drug release7-8 and as membranes for enantiomer
separation14 and for proton conductance in fuel cells15 Although pore-sizes available within MOFs1
are already sufficiently large to host metal-nanoclusters such as Au5516-17 the actual realization of
meta-crystals requires in addition to structural requirements a strategy for the controlled loading
of the nanoparticles (NPs) into the molecular framework Simply adding NPs to the solutions before
starting the solvothermal synthesis of MOFs has been reported18 but this procedure does not allow
for controlled loading The layer-by-layer growing process of SUFMOFs allows the uptake of
nanosized objects during synthesis including the fabrication of compositional gradients of different
NPs Those guest NPs can range from pure metal metal oxide or covalent clusters over one-
dimensional objects such as nanowires carbon or inorganic nanotubes to biological molecules such
as drugs or even small proteins If the loading happens during synthesis alternating layers of
different NPs can be realized
The introduction of procedures for epitaxial growth of MOFs on functionalized substrates19 was a
major step towards controlled functionalization of frameworks Epitaxial synthesis allowed the
preparation of compositional gradients20-21 and provided a new strategy to load nano-objects into
predefined pores
Unfortunately the LPE process has so far been only demonstrated for a fairly small number of
MOFs141922-23 none of which exhibits the required pore sizes In addition for the fabrication of meta-
crystals the architecture of the network should be sufficiently adjustable to realize pores of different
sizes There should also be a straightforward way to functionalize the framework itself in order to
tailor the interaction of the walls with the targeted guest objects Ideally such a strategy would be
based on an isoreticular series of MOFs in which the pore size can be determined by a choice from a
homologous series of ligands with different lengths1
Here we present a novel isoreticular MOF series with the required flexibility as regards pore-sizes
and functionalization which is well suited for the LPE process This class denoted as SURMOF-2 is
derived from MOF-2 one of the simplest framework architectures24-25 MOF-2 is based on paddle-
wheel (pw) units formed by attaching 4 dicarboxylate groups to Cu++ or Zn++-dimers yielding planar
sheets with 4-fold symmetry (see Fig 1) These planes are held together by strong
carboxylatemetal- bonds26 Conventional solvothermal synthesis yields stacks of pw-planes shifted
relative to each other with a corresponding reduction in symmetry to yield a P2 or C2 symmetry2427-
28
147
The relative shifts between the pw-planes can be avoided when using the recently developed liquid
phase epitaxy (LPE) method22 As demonstrated by the XRD-data shown in Fig 2 for a number of
different dicarboxylic acid ligands the LPE-process yields surface attached metal organic frameworks
(SURMOFs) with P4 symmetry29 except for the 26-NDC ligand which exhibits a P2 symmetry as a
result of the non-linearity of the carboxyl functional groups on the ligand The ligand PPDC
pentaphenyldicarboxylic acid with a length of 25 nm is to our knowledge the longest ligand which
has so far been used successfully for a MOF synthesis303030 A detailed analysis of the diffraction data
allowed us to propose the structures shown in Fig 1 This novel isoreticular series of MOFs hereafter
termed SURMOF-2 also consists of pw-planes which ndash in contrast to MOF-2 ndash are stacked directly
on-top of each other This absence of a lateral shift when stacking the pw-planes yields 1d-pores of
quadratic cross section with a diagonal up to 4 nm (see Fig 2) Importantly for the SURMOF-2 series
interpenetration is absent For many known isoreticular MOF series the formation of larger and
larger pores is limited by this phenomenon if the pores become too large a second or even a third
3d-lattice will be formed inside the first lattice yielding interpenetrated networks which exhibit the
expected larger lattice constant but with rather small pore sizes31-33 the hosting of NPs thus becomes
impossible For SURMOF-2 the particular topology with 1d-pores of quadratic cross section is not
compatible with the presence of a second interwoven network and as a result interpenetration is
suppressed
Although the solubility of some the long dicarboxylic acids ligands used for the SURMOF-2 fabrication
(TPDC QPDC and PPDC see Fig 1) is fairly small we encountered no problems in the LPE process
since already small concentrations of dicarboxylic acids are sufficient for the formation of a single
monolayer on the substrate the elementary step of the LPE synthesis process34-35 Even with the
longest dicarboxylic acid available to us PPDC growth occurred in a straightforward fashion and
optimization of the growth process was not necessary
The P4 structures proposed for the SURMOF-2 series except for the 26-NDC ligand differ markedly
from the bulk MOF-2 structures obtained from conventional solvothermal synthesis2427-28 To
understand this unexpected difference and in particular the absence of relative shifts between the
pw-planes in the proposed SURMOF-2 structure we performed extensive quantum chemical
calculations employing approximate density-functional theory (DFT) in this case London dispersion-
corrected self-consistent charge density-functional based tight-binding (DFTB)36-38 to a periodic
model of MOF-2 and its SURMOF derivatives
Periodic structure DFTB calculations (Fig S1) confirmed that the P2m structure proposed by Yaghi
et al for the bulk MOF-2 material24 is indeed the energetically favorable configuration for MOF-2
while the C2m structure that has also been reported for bulk MOF-227-28 is slightly higher in energy
148
(41 meV per formula unit see Table 1) The optimum interlayer distance between the pw-planes in
the P4 structure is 58 Aring in very good agreement with the experimental value of 56 Aring (as obtained
from the in-plane XRD-data see Fig S2) It is important to note that at that distance the linkers
cannot remain in rectangular position as in MOF-2 but are twisted in order to avoid steric hindrance
and to optimize linker-linker interactions
The P4mmm structure as proposed for SURMOF-2 is less stable by 200 meV per formula unit as
compared to the P2m configuration Although the crystal energy for the P4 stacking is substantially
smaller than for P2 packing simulated annealing (DFTB model using Born-Oppenheimer Molecular
Dynamics) carried out at 300K for 10 ps demonstrated that the P4mmm structure corresponds to a
local minimum This observation is confirmed by the calculation of the stacking energy of MOF-2
where a shifting of the layers is simulated along the P2-P4-C2 path (Figure 3) On the basis of these
calculations we thus propose that SURMOF-2 adopts this metastable P4 structure
In Table 1 the interlayer interaction energies are summarized They range from 590 meV per formula
unit with respect to fully dissociated MOF-2 layers for Cu2(bdc)2 and successively increase for longer
linkers of the same type reaching 145 eV per formula unit for Cu2(ppdc)2 indicating that the linkers
play an important role in the overall stabilization of SURMOF-2 with large pore sizes Even stronger
interlayer interactions are found for different linker topologies (PPDC) A detailed computational
analysis of the role of the individual building blocks of SURMOF-2 for the crystal growth and
stabilization will be published elsewhere
The flat well-defined organic surfaces used as the templating (or nucleating) substrate in the LPE
growth process provide a satisfying explanation for why SURMOF-2 grows with the highly
symmetrical P4 structure instead of adopting the energetically more favorable bulk P2 symmetry3439
The carboxylic acid groups exposed on the surface force the first layer of deposited Cu-dimers into a
coplanar arrangement which is not compatible with the bulk P2 (or C2) structure but rather
nucleates the P4 packing In turn the carboxylate-groups in the first layer of deposited dicarboxylic
acids provide the same constraints for the next layer (see Fig 1) As a result of the layer-by-layer
method employed for further SURMOF-2 growth the same boundary conditions apply for all
subsequent layers The organic surface thus acts as a seed structure to yield the metastable P4
packing not an unusual motif in epitaxial growth40
The calculations allow us to predict that it will be possible to grow SURMOF structures with even
larger linker molecules as used here The stacking will further stabilize the P4 symmetry as the
interaction energy per formula unit is more and more dominated by the linkers (Table 1) At present
149
we cannot predict the maximum possible size of the pores but linker PPDC (see Fig 2) is thus far
unmatched as a component in non-interpenetrated framework structures
Acknowledgement
We thank Ms Huumllsmann Bielefeld University for the preparation of P(EP)2DC Financial support by
Deutsche Forschungsgemeinschaft (DFG) within the Priority Program Metal-Organic Frameworks
(SPP 1362) is gratefully acknowledged
Methods
Computational Details
All structures were created using a preliminary version of our topological framework creator
software which allows the creation of topological network models in terms of secondary building
units and their replacement by individual molecules to create the coordinates of virtually any
framework material The generated starting coordinates including their corresponding lattice
parameters have been hierarchically fully optimized using the Universal Force Field (UFF)41 followed
by the dispersion-corrected self-consistent-charge density-functional based tight-binding (DFTB)
method36-3842-43 that we have recently validated against experimental structures of HKUST-1 MOF-5
MOF-177 DUT-6 and MIL-53(Al) as well as for their performance to compute the adsorption of
water and carbon monoxide37 For all calculations we employed the deMonNano software44444444
We have chosen periodic boundary conditions for all calculations and the repeated slab method has
been employed to compute the properties of the single layers in order to evaluate the stacking
energy A conjugate-gradient scheme was employed for geometry optimization of atomic
coordinates and cell dimensions The atomic force tolerance was set to 3 x 10-4 eVAring
The confined conditions at the SURMOF-substrate interface was modeled by fixing the corresponding
coordinate in the computer simulations All calculated structures have been substantiated by
simulated annealing calculations where a Born-Oppenheimer Molecular Dynamics trajectory at 300K
has been computed for 10 ps without geometry constrains All structures remained in P4 topology
Experimental methods
The SURMOF-2 samples were grown on Au substrates (100-nm Au5-nm Ti deposited on Si wafers)
using a high-throughput approach spray method45 The gold substrates were functionalized by self-
assembled monolayers SAMs of 16-mercaptohexadecanoic acid (MHDA) These substrates were
mounted on a sample holder and subsequently sprayed with 1 mM of Cu2 (CH3COO)4middotH2O ethanol
solution and ethanolic solution of SURMOF-2 organic linkers (01 mM of BDC 26-NDC BPDC and
150
saturated concentration of TPDC QPDC PPDC P(EP)2DC) at room temperature After a given
number of cycles the samples were characterized with X-ray diffraction (XRD)
Figure 1 Schematic representation of the synthesis and formation of the SURMOF-2 analogues
151
Figure 2 Out of plane XRD data of Cu-BDC Cu-26-NDC Cu-BPDC Cu-TPDC Cu-QPDC Cu-P(EP)2DC and Cu-PPDC (upper left) schematic representation (upper right) and proposed structures of SURMOF-2 analogues (lower panel) All the SURMOF-2 are grown on ndashCOOH terminated SAM surface using the LPE method
152
Figure 3 Energy for the relative shift of each layer in two different directions horizontal (P4 to P2) and diagonal (P4 to C2) at a fixed interlayer distance of 56 Aring Structures have been partially optimized and energies are given per formula unit with respect to fully dissociated planes
Supporting information
Synthesis of organic linkers
(1) para-terphenyldicarboxylic acid (TPDC)
To a solution of para-terphenyl (1500 g 651 mmol 1 eq) and oxalyl chloride (3350 mL 3900 mmol
6 eq) in carbon disulfide (30 mL) at 0degC (ice bath) was added aluminium chloride (1475 g 1106
mmol 17 eq) After 1h stirring additional aluminium chloride (0870 g 651 mmol 1 eq) was added
the ice bath was removed and the reaction mixture was stirred at room temperature for 24h The
mixture was poured into crushed ice and stirred until the dark-brown mixture turned to yellow
Carbon disulfide was removed under reduced pressure and the resulting aqueous suspension was
filtered The solid was washed with diluted hydrochloric acid then with diethyl ether and dried
under vacuum overnight with an oil bath at 50degC Pale yellow solid yield 2028 g (98)
(2) para-quaterphenyldicarboxylic acid (QPDC)
153
To a solution of para-quaterphenyl (1000 g 326 mmol 1 eq) and oxalyl chloride (1680 mL 1956
mmol 6 eq) in carbon disulfide (20 mL) at 0degC (ice bath) was added aluminium chloride (0740 g 555
mmol 17 eq) After 1h stirring additional aluminium chloride (0435 g 326 mmol 1 eq) was added
the ice bath was removed and the reaction mixture was stirred at room temperature for 24h The
mixture was poured into crushed ice and stirred until the dark-brown mixture turned to yellow
Carbon disulfide was removed under reduced pressure and the resulting aqueous suspension was
filtered The solid was washed with diluted hydrochloric acid then with diethyl ether and dried
under vacuum overnight with an oil bath at 50degC Yellow solid yield 1186 g (92)
(3) P(EP)2DC
The synthesis of the P(EP)2DC-linker has been described in Ref 46
(4) para-pentaphenly dicarboxylic acid (PPDC)
Scheme 1 Synthesis of para-pentaphenly dicarboxylic acid i) Pd(PPh3)4 Na2CO3 toluenedioxane H2O 85oC ii) MeOH 6M NaOH HCl
para-pentaphenly dicarboxylic acid (H2L) was synthesized via Suzuki coupling of 44rdquo-dibromo-p-
terphenyl and 4-methoxycarbonylphenylboronic acid (Scheme 1) 44rdquo-dibromo-p-terphenyl (776 mg
200 mmol) 4-methoxycarbonylphenylboronic acid (108 g 60 mmol) and Na3CO3 (170 g 160 mmol)
were added to a degassed mixture of toluene14-dioxane (50 mL 221) under Ar [Pd(PPh3)4] (116
mg 010 mmol) was added to the mixture and heated to 85 degC for 24 hours under Ar The reaction
mixture was cooled to room temperature The precipitate was collected by filtration washed with
water methanol and used for next reaction without further purification The final product H4L was
obtained by hydrolysis of the crude product under reflux in methanol (100 mL) overnight with 6M
aqueous NaOH (20 mL) followed by acidification with HCl (conc) After removal of methanol the
final product was collected by filtration as a grey solid (078 g 83 yield) 1H NMR (d6-DMSO
250MHz 298K) δ (ppm) 805 (d J = 75Hz 4H) 789-783 (m 12H) 750 (d J = 75Hz 4H) 13C NMR
cannot be clearly resolved due to poor solubility MALDI-TOF-MS (TCNQ as matrix) mz 47002
cacld 47051 Elemental analysis Calculated C 8169 H 471 Found C 8163 H 479
Br Br MeOOC B
OH
OH
+
COOMe
COOMe
COOH
COOH
i ii
154
Figure S1 MOF-2 in P4 (as reported) P2 and C2 symmetry
155
Figure S2 In-plane XRD data of Cu-BDC Cu-26-NDC Cu-BPDC Cu-TPDC Cu-QPDC and Cu-P(EP)2DC All the
SURMOF-2 are grown on ndashCOOH terminated SAM surface using the LPE method The position of (010) plane
represents the layer distance
Table 1 Stacking energy and geometries of P4 SURMOF-2 derivatives
Symmetry a= c b Stacking Energy
Cu2(bdc)2 C2 1119 50 -076
Cu2(bdc)2 P2 1119 54 -08
Cu2(bdc)2 P4 1119 58 -059
156
Cu2(ndc)2 P2 1335 56 -04
Cu2(bpdc)2 P4 1549 59 -068
Cu2(tpdc)2 P4 1984 59 -091
Cu2(qpdc)2 P4 2424 59 -121
Cu2(P(EP)2DC)2 P4 2512 52 -173
Cu2(ppdc)2 P4 2859 59 -145
Stacking energies (DFTB level in eV) of SURMOFs The energies were calculated within periodic
boundary conditions and are given per formula unit
References
1 Eddaoudi M et al Systematic design of pore size and functionality in isoreticular MOFs and
their application in methane storage Science 295 469-472 (2002)
2 Rosi N L et al Hydrogen storage in microporous metal-organic frameworks Science 300
1127-1129 (2003)
3 Rowsell J L C amp Yaghi O M Metal-organic frameworks a new class of porous materials
Microporous and Mesoporous Materials 73 3-14 (2004)
4 Wang B Cote A P Furukawa H OKeeffe M amp Yaghi O M Colossal cages in zeolitic
imidazolate frameworks as selective carbon dioxide reservoirs Nature 453 207-U206 (2008)
5 Lauren E Kreno et al Metal-Organic Framework Materials as Chemical Sensors Chemical
Reviews 112 1105-1124 (2012)
6 Dragasser A et al Redox mediation enabled by immobilised centres in the pores of a metal-
organic framework grown by liquid phase epitaxy Chemical Communications 48 663-665
(2012)
7 Horcajada P et al Metal-organic frameworks as efficient materials for drug delivery
Angewandte Chemie-International Edition 45 5974-5978 (2006)
8 Horcajada P et al Flexible porous metal-organic frameworks for a controlled drug delivery
Journal of the American Chemical Society 130 6774-6780 (2008)
9 Hurd J A et al Anhydrous proton conduction at 150 degrees C in a crystalline metal-organic
framework Nature Chemistry 1 705-710 (2009)
10 Kreno L E Hupp J T amp Van Duyne R P Metal-Organic Framework Thin Film for Enhanced
Localized Surface Plasmon Resonance Gas Sensing Analytical Chemistry 82 8042-8046
(2010)
11 Lu G et al Fabrication of Metal-Organic Framework-Containing Silica-Colloidal Crystals for
Vapor Sensing Advanced Materials 23 4449-4452 (2011)
157
12 Lu G amp Hupp J T Metal-Organic Frameworks as Sensors A ZIF-8 Based Fabry-Perot Device
as a Selective Sensor for Chemical Vapors and Gases Journal of the American Chemical
Society 132 7832-7833 (2010)
13 Mark D Allendorf Adam Schwartzberg Vitalie Stavila amp Talin A A Roadmap to
Implementing MetalndashOrganic Frameworks in Electronic Devices Challenges and Critical
Directions European Journal of Chemistry (2011)
14 Bo Liu et al Enantiopure Metal-Organic Framework Thin Films Oriented SURMOF Growth
and Enantioselective Adsorption Angewandte Chemie-International Edition 51 817-810
(2012)
15 Sadakiyo M Yamada T amp Kitagawa H Rational Designs for Highly Proton-Conductive
Metal-Organic Frameworks Journal of the American Chemical Society 131 9906-9907 (2009)
16 Ishida T Nagaoka M Akita T amp Haruta M Deposition of Gold Clusters on Porous
Coordination Polymers by Solid Grinding and Their Catalytic Activity in Aerobic Oxidation of
Alcohols Chemistry-a European Journal 14 8456-8460 (2008)
17 Esken D Turner S Lebedev O I Van Tendeloo G amp Fischer R A AuZIFs Stabilization
and Encapsulation of Cavity-Size Matching Gold Clusters inside Functionalized Zeolite
Imidazolate Frameworks ZIFs Chemistry of Materials 22 6393-6401 (2010)
18 Lohe M R et al Heating and separation using nanomagnet-functionalized metal-organic
frameworks Chemical Communications 47 3075-3077 (2011)
19 Shekhah O et al Step-by-step route for the synthesis of metal-organic frameworks Journal
of the American Chemical Society 129 15118-15119 (2007)
20 Furukawa S et al A block PCP crystal anisotropic hybridization of porous coordination
polymers by face-selective epitaxial growth Chemical Communications 5097-5099 (2009)
21 Shekhah O et al MOF-on-MOF heteroepitaxy perfectly oriented [Zn(2)(ndc)(2)(dabco)](n)
grown on [Cu(2)(ndc)(2)(dabco)](n) thin films Dalton Transactions 40 4954-4958 (2011)
22 Shekhah O et al Controlling interpenetration in metal-organic frameworks by liquid-phase
epitaxy Nature Materials 8 481-484 (2009)
23 Zacher D et al Liquid-Phase Epitaxy of Multicomponent Layer-Based Porous Coordination
Polymer Thin Films of [M(L)(P)05] Type Importance of Deposition Sequence on the Oriented
Growth Chemistry-a European Journal 17 1448-1455 (2011)
24 Li H Eddaoudi M Groy T L amp Yaghi O M Establishing microporosity in open metal-
organic frameworks Gas sorption isotherms for Zn(BDC) (BDC = 14-benzenedicarboxylate)
Journal of the American Chemical Society 120 8571-8572 (1998)
25 Mueller U et al Metal-organic frameworks - prospective industrial applications Journal of
Materials Chemistry 16 626-636 (2006)
158
26 Shekhah O Wang H Zacher D Fischer R A amp Woumlll C Growth Mechanism of Metal-
Organic Frameworks Insights into the Nucleation by Employing a Step-by-Step Route
Angewandte Chemie-International Edition 48 5038-5041 (2009)
27 Carson C G et al Synthesis and Structure Characterization of Copper Terephthalate Metal-
Organic Frameworks European Journal of Inorganic Chemistry 2338-2343 (2009)
28 Clausen H F Poulsen R D Bond A D Chevallier M A S amp Iversen B B Solvothermal
synthesis of new metal organic framework structures in the zinc-terephthalic acid-dimethyl
formamide system Journal of Solid State Chemistry 178 3342-3351 (2005)
29 Arslan H K et al Intercalation in Layered Metal-Organic Frameworks Reversible Inclusion of
an Extended pi-System Journal of the American Chemical Society 133 8158-8161 (2011)
30 The MOF with the largest pore size recorded so far MOF-200 (Furukawa H et al Ultrahigh
Porosity in Metal-Organic Frameworks Science 329 424-428 (2010)) used a (trivalent)
444-(benzene-135-triyl-tris(benzene-41-diyl))tribenzoate (BBC) ligand The carboxylic
acid-to carboxylic acid distance is 20 nm compared to 25 nm in case of PPDC The cage size
in MOF-200 amounts to 18 nm by 28 nm clearly smaller than the 1d-channels in the PPDC
SURMOF-2 that are 28 nm by 28 nm
31 Batten S R amp Robson R Interpenetrating nets Ordered periodic entanglement
Angewandte Chemie-International Edition 37 1460-1494 (1998)
32 Snurr R Q Hupp J T amp Nguyen S T Prospects for nanoporous metal-organic materials in
advanced separations processes Aiche Journal 50 1090-1095 (2004)
33 Yaghi O M A tale of two entanglements Nature Materials 6 92-93 (2007)
34 Shekhah O Liu J Fischer R A amp Woumlll C MOF thin films existing and future applications
Chemical Society Reviews 40 1081-1106 (2011)
35 Zacher D Shekhah O Woumlll C amp Fischer R A Thin films of metal-organic frameworks
Chemical Society Reviews 38 1418-1429 (2009)
36 Elstner M et al Self-consistent-charge density-functional tight-binding method for
simulations of complex materials properties Physical Review B 58 7260-7268 (1998)
37 Lukose B et al Structural properties of metal-organic frameworks within the density-
functional based tight-binding method Physica Status Solidi B-Basic Solid State Physics 249
335-342 (2012)
38 Zhechkov L Heine T Patchkovskii S Seifert G amp Duarte H A An efficient a Posteriori
treatment for dispersion interaction in density-functional-based tight binding Journal of
Chemical Theory and Computation 1 841-847 (2005)
159
39 Zacher D Schmid R Woumlll C amp Fischer R A Surface Chemistry of Metal-Organic
Frameworks at the Liquid-Solid Interface Angewandte Chemie-International Edition 50 176-
199 (2011)
40 Prinz G A Stabilization of bcc Co via Epitaxial Growth on GaAs Physical Review Letters 54
1051-1054 (1985)
41 Rappe A K Casewit C J Colwell K S Goddard W A amp Skiff W M UFF a full periodic
table force field for molecular mechanics and molecular dynamics simulations Journal of the
American Chemical Society 114 10024-10035 (1992)
42 Seifert G Porezag D amp Frauenheim T Calculations of molecules clusters and solids with a
simplified LCAO-DFT-LDA scheme International Journal of Quantum Chemistry 58 185-192
(1996)
43 Oliveira A F Seifert G Heine T amp Duarte H A Density-Functional Based Tight-Binding an
Approximate DFT Method Journal of the Brazilian Chemical Society 20 1193-1205 (2009)
44 deMonNano v 2009 (Bremen 2009)
45 Arslan H K et al High-Throughput Fabrication of Uniform and Homogenous MOF Coatings
Adv Funct Mater 21 4228-4231 (2011)
46 Schaate A et al Porous Interpenetrated Zirconium-Organic Frameworks (PIZOFs) A
Chemically Versatile Family of Metal-Organic Frameworks Chemistry-a European Journal 17
9320-9325 (2011)
160
Appendix H
Linker guided metastability in templated Metal-Organic Framework-2 derivatives (SURMOFs-2)
Binit Lukose Gabriel Merino Christof Woumlll and Thomas Heine
Prepared for publication
INTRODUCTION
The molecular assembly of metal-oxides and organic struts can provide a large number of network
topologies for Metal-Organic Frameworks (MOFs)1-3 This is attained by the differences in
connectivity and relative orientation of the assembling units Within each topology replacement of a
building unit by another of same connectivity but different size leads to what is known as isoreticular
alteration of pore size The structure of MOFs in principle can be formed into the requirement of
prominent applications such as gas storage4-8 sensing910 and catalysis11-13 The structural
divergence and the performance can be further increased by functionalizing the organic linkers1415
In MOFs linkers are essential in determining the topology as well as providing porosity A linker
typically contains single or multiple aromatic rings the orientation of which normally undergoes
lowest repulsion from the coordinated metal-oxide clusters providing the most stable symmetry for
the bulk material We encounter for the first time a situation that the orientation of the linker
provides metastability in an isoreticular series of MOFs Templated MOF-2 derivatives (surface-MOF-
2 or simply SURMOFs-2)16 show this extraordinary phenomenon While bulk MOF-2 is determined to
be stable in P217 or C218-20 symmetry we found the existence of SURMOFs-2 in P4 symmetry
161
(Figure 1)16 Our theoretical approach to this problem identifies the role of the linkers in providing
P4 geometry the status of a local energy-minimum
MOF-2 derivatives are two-dimensional lattice structures composed of dimetal-centered four-fold
coordinated paddlewheels and linear organic linkers Solvothermally synthesized archetypal MOF-2
had transition metal (Zn or Cu) centers in the connectors and benzenedicarboxylic acid linkers17 The
derivatives under our investigation contain paddlewheels with Cu dimers and benzenedicarboxylic
acid (BDC) naphthalenedicarboxylic acid (NDC) biphenyldicarboxylic acid (BPDC)
triphenyldicarboxylic acid (TPDC) quaterphenyldicarboxylic acid (QPDC) P(EP)2DC21 and
pentaphenyldicarboxylic acid (PPDC) linkers pertaining to isoreticular expansion of pore size16 The
four-fold connectivity of the Cu dimers together with linearity of the linkers gives rise to layers with
quadratic (square) topology The interlayer separation d is typically much more than that of
graphite or 2D COFS22-24 because all the atoms in one layer do not lie in one plane
In bulk form the nearest layers are shifted to each other either towards one of the four linkers
(P2)171920 or diagonally towards one of the four surrounding pores (C2)18-20 in effort to reduce
the repulsion of alike charges in the adjacent paddlewheels when they are in eclipsed form (P4)
(Figure 1) The metal-dimers often show high reactivity which results in attracting water or
appropriate solvents in their axial positions The stacking along the third axis is typically through
interlayer interactions and through hydrogen bonds established between the solvents or between
the solvents and oxygen atoms in the paddlewheel The lateral shift of layers is inevitable without
additional supports Coordination of pillar molecules such as 14-diazabicyclo[222]octane (dabco) or
bipyridine in the axis of metal dimers between the adjacent layers25 is a practical mean to avoid
layer-offset however with the change of MOF dimensionality
Figure 1 Monolayer of MOF-2 and three kinds of layer packing -P4 P2 and C2- of periodic MOF-2
162
Alternate to solvothermal synthesis a step-by-step route may be enabled for the synthesis of
MOFs26-28 It adopts liquid phase epitaxy (LPE) of MOF reactants on substrate-assisted self-assembled
monolayers This is achieved by alternate immersion of the template in metal and ligand precursors
for several cycles This allows controlled growth of MOFs on surfaces The latest outcomes of this
method are the surface-standing MOF-2 derivatives (SURMOF-2s) This isoreticular SURMOF series
has linkers of different lengths (as given above) The cell dimensions that correspond to the length of
the pore walls range from 1 to 28 nm The diagonal pore-width of one of the MOFs (PPDC) accounts
to 4 nm
After evacuation of solvents (water in the present case) X-ray diffraction patterns were taken in
directions both perpendicular (scans out of plane) and parallel (scans in-plane) to the substrate
surface The presence of only one peak -(010)- in the perpendicular scan implies that the layers
orient perpendicular to the surface and the corresponding angle gives the cell parameter a (or c) In
the latter case the peaks - (100) and (001) ndash are identical and this rules out the possibility of layer-
offset Hence the symmetry of the MOFs was determined to be P4 These peaks give rise to the cell
parameter c which is nothing but the interlayer distance d This value (56 Aring) is relatively small for
P4 geometry to have water molecules in the axial position of paddlewheels Presence of some water
molecules observed using Infrared Reflection Absorption Spectroscopy (IRRAS) might be located near
paddlewheels but exposed to the pore The new observations with the SURMOFs are very intriguing
in the context that P4 symmetry appears to be impossible for solvothermally produced MOF-2
We have primarily reported that the P4 geometry of SURMOF-2 exists as a local energy-minimum16
The verification was made using an approximate method of density functional theory (DFT) which is
London dispersion-corrected self-consistent charge density-functional based tight-binding (DFTB) In
the bulk form absolute energies of P2 and C2 are 210 and 170 meV respectively higher than P4 per
a formula unit which is equivalent to the unit cell For SURMOF the barrier for leaving P4 was nearly
50 meV per formula unit It requires further analysis to unravel the reasons for this unusual
metastability We therefore performed an extensive set of quantum chemical calculations on the
composition of the constituent building units The procedure involves defining SURMOF geometry
and analyzing the translations of individual layers
The major individual contributions to the total energy are the interaction between the paddlewheel
units (favouring P2) London dispersion interaction between tilted linkers (favouring P4) and energy
to tilt the linkers In the iso-reticular series of SURMOF-2 the differences arise only in the
163
contributions from the linkers Hence we performed an extensive study only on the smallest of all
linkers- BDC A scaling might be appropriate for other linkers
RESULTS AND DISCUSSION
In solvothermal synthesis of layered MOFs it is assumed that the nucleation process is associated
with the interaction between two connectors This is rationalized by the fact that two paddlewheels
show the strongest possible noncovalent interaction between the individual MOF building blocks
present in MOF-2 As the orientation of the paddlewheels is restricted by the linkers we studied the
stacking of adjacent layers by determining the attraction of two adjacent interacting paddlewheels
upon their respective offsets Thus we investigated the geometries corresponding to lateral
displacements between the known stable arrangements of the bulk structure ie P4-to-P2 and P4-
to-C2 (Figure 2) In the model calculation the two paddlewheels were shifted from D4h symmetry to
two directions that respectively correspond to P2 and C2 symmetries in the bulk The minima along
the two shifts are referred as min-I and min-II and are having C2h symmetry It is interesting to note
that the interaction is in all cases attractive If only the paddlewheels are studied the D4h
configuration where both axes are concentric can be interpreted as transition state between the
two minima configurations P2 or C2 Comparing the two minima min-I corresponding to MOF-2 in
P2 symmetry indicates the formation of two CuO bonds while for min-II the reactive Cu centers do
not participate in the interlayer bonding
Formation of the diagonally shifted C2 bulk geometry for Zn and Cu based MOF-2 as reported in the
literature18-20 possibly is due to the presence of large solvent molecules such as DMF that
coordinate to the free Cu centers the paddlewheels
Figure 2 Displacements of two paddlewheels from D4h symmetry (corresponding to P4) towards minima that correspond to P2 and C2 bulk symmetries
164
To gain further insight on type of interactions for the three paddlewheel arrangements as found in
the bulk (Figure 3) we performed the topological analysis of the electron density for each
structure2930 An inspection of the paddlewheel shows that the electron density of the Cu-O bond has
a (relatively small) value of 0042 au thus showing only weak character In the forms related to P4
and C2 bulk symmetries all (3-1) critical points between the two paddlewheels show very small
density values (0004 au and less) In the P2 structure it is apparent the formation of a four-
membered Cu-O-Cu-O ring across the paddlewheel Note that the Cu-O interactions inside the
paddlewheel weaken (from 0042 to 0031 au) and two intermolecular Cu-O interactions with a
density value of 0024 au stabilize the P2 orientation These links if formed during nucleation will
be regularly repeated during the MOF-2 formation in solvothermal synthesis and due to its strong
binding of -08 eV they are the main reason for the stabilization of the P2 phase If the paddlewheels
are packed in P4 symmetry there must be additional means of stabilization present and that may
only arise from the linkers Moreover one should note that in order to reach the P4 symmetry a
layer-by-layer growth process is required as otherwise the linkers will readily stack as in the P2 bulk
form
165
Figure 3 Topological analysis of the electron density of fully optimized P4 (top left) C2 (top right) and P4 (bottom) structures Atom (grey) (3-1) (red) and (3+1) (green) critical points along with their charge densities are shown
The optimized geometry of a single-layer MOF-2 is shown in Figure 1 Note that the phenyl rings of
the linkers are oriented perpendicular to the MOF plane Contrary to graphene or 2D COFs the more
complex structure of MOF-2 layers may become subject to change upon the interlayer interactions
This is in particular important for the P4 stacking as reported for SURMOF-2 The interaction energy
of two linkers and two benzene rings as oriented in the monolayer has been computed as function
of the interlayer distance (Figure 4) At the experimental interlayer distance of 56 Aring the linkers are
so close that they repel each other strongly and stacking the monolayer structure at the
experimental interlayer distance would introduce an energy penalty of 08 eV per linker
It would not be exotic if we assume that the anchoring of layers on the substrate plays an important
role in setting d A supporting argument is the fact that all the SURMOFs in the iso-reticular series
have the same d An additional point is that the comparatively wider linkers NDC and LM do not
create any difference in the interlayer distance
166
Figure 4 Energy as a function of interlayer distance [d] in case of two paddlewheels and two benzene molecules The energies are given with respect to the complete dissociation of the building blocks
The best adaptation for the linkers in a closer alignment to avoid the repulsion is to partially rotate
the phenyl rings This reduces the Pauli repulsion and at the same time increases the attractive
London dispersion between the linkers However the rotation is energetically penalized by 06 eV as
accordance with similar calculations found in the literature31 and is with the same order of Zn4O-
tetrahedron clusters of the IRMOFs3233
Figure 5 Energy as a function of rotation of linker between two saturated paddlewheels The dihedral angle O-C1-C2-C3 θ between the linker and the carboxylic part in the paddlewheel Values are with respect to the ground state θ = 0⁰
To model the conditions in SURMOF-2 we need to combine the intermolecular interaction of the
linkers with the barrier associated to the rotation of the linker between two paddlewheel units as
given in Figure 5 We may consider a system of three linkers placed as if they are in three adjacent
layers of bulk P4 SURMOF-2 The linkers are undergoing a rotation along their C2 axis that would be
aligned to the Cu-Cu axes in the paddlewheels (see system-I in Figure 6) The dissociation energy of
167
the system includes four times the repulsion from one adjacent linker If we neglect the interaction
between the end-linkers which is only -35 meV the system represents the situation in bulk P4 MOF-
2 The gain of intermolecular energy due to twisting the stacked linkers is partially compensated by
the energy penalty arising from rotation of the linker between the paddlewheels and the resulting
energy shows a minimum at 22deg (Figure 6)
Figure 6 Model of the linker orientation in SURMOF-2 The stabilization of a linker inbetween two layers is approximated in System-I the arrows indicating the rotation while the energy penalty due to breaking the p conjugation of Figure 5 is given in System-II The resulting energy is the black line showing a minimum at 22deg The energy in System-I is with respect to its dissociation limit
Each linker may undergo rotation either clockwise or anti-clockwise with equal preferences in the
local environment However there may be a global control over the preference of each linker The
most stable structure can be figured out from the total energies of each possible arrangement Since
there are only two choices for each linker it may orient either in same fashion or alternate fashion
along X and Y directions If we expect a regular pattern the total number of possibilities are only
three as shown in Figure 8 where rotation of linkers (violet lines) are marked with lsquo+rsquo signs in one of
its two sides which means that facing (outer)-edges of linkers are tilted towards these sides The
three orderings may be verbalized as follows
(i) projection of the facing edges of oppositely placed linkers are either within the square or outside
(P4nbm) (ii) projection of the facing edge of only one of the oppositely placed linkers is within the
square (P4mmm) (iii) projection of the facing edges of all four linkers are either within the square
or outside (P4nmm)
The adjacent layers follow the same linker-orientations The unit cells of (i) and (iii) are four times
bigger than (ii) The stacking energies calculated at the DFTB level suggest P4nbm as the most stable
168
geometry The energies are respectively -048 -032 and -029 eV per formula unit for P4nbm
P4mmm and P4nmm respectively Within P4nbm symmetry the linker edges undergo lowest
repulsion compared to others It is to be noted that only P4nbm has four-fold rotational symmetry
along Z-axis about the Cu-dimer in any paddlewheel
Figure 7 Three possible arrangements of tilted linkers in the periodic SURMOF-2 Paddlewheels are shown as red diamonds and the linkers as violet lines Facing (outer) edges of the linkers are rotated towards the side with + signs The symmetries and calculated stacking energies (per formula unit) of the arrangements are also given
To quantify the different stacking energies we performed periodic DFT calculations on the structure
of Cu2(bdc)2 MOF-2 As the most stable P4nbm geometry demands high computational resources in
each calculation we used P4mmm geometry which has four times less atoms in unit cell We
explored tight-packing with rotated linkers and loose-packing with un-rotated linkers Two energy-
minima have been found as shown in Figure 8a at 58 and 65 Aring corresponding to rotated and un-
rotated states of linkers respectively The latter is 40 meV more stable than the former which
means that SURMOF-2 exists in a metastable state along the perpendicular axis The relative shift of
adjacent layers in the direction of the lattice vector (or ) represents the transition from P2 to P4
and back to P2 symmetry We shifted the adjacent layers parallel to and calculated the relative
energies Since the shift of adjacent layers in P4mmm geometry is not symmetric for positive and
negative directions of averages of the energies of the shift in both directions are plotted (see
Figure 9b) P4 is a local minimum while P2 is the global minimum and the energy barrier separating
the two minima accounts to 23 meV per formula unit Hence the P4 geometry of SURMOF-2 can be
taken as metastable state of MOF-2
169
Figure 8 a) Energy as function of interlayer separation d and b) Energy as a function of relative shift of adjacent layers for periodic MOF-2 Energies are given in eV per formula unit
The role of linkers in creating a local minimum at P4 is now apparent However when undergoing the
transition from P4 to P2 the relative motions of the horizontal and vertical linkers are different from
each other Hence a qualitative study is essential to accurately determine the role of each building
block for holding SURMOFs-2 in a metastable state We thus resolved the shift of entire adjacent
layers with respect to each other into relative motions of individual building blocks The experimental
interlayer distance (56 Aring) has been fixed and the energy-profiles have been calculated using DFT
The scans include the shift of
i) a paddlewheel over other
ii) a horizontal linker over other
iii) a vertical linker over other
iv) a paddlewheel over a horizontal linker
v) a paddlewheel over a vertical linker
Here vertical and horizontal linkers are the linkers vertical and horizontal to the shift directions
respectively A complete picture of individual shifts of the MOF-2 SBU including their energy-profiles
is given in Figure 9 The shift of the vertical over the horizontal linker was found insignificant and was
omitted A note of warning is that the tilted vertical linker meets different neighborhoods when
shifted to the left and right However an average energy of these two shifts seems sensible because
the nearest vertical linker in the same layer of P4nbm geometry has opposite orientation This
averaging also makes sense in a case that alternate layers undergo shifting to the same direction
leading to a zigzag (or serrated) packing of layers in the bulk As is readily seen in Figure 9 the
formation of the Cu-O-Cu-O rings between the paddlewheels (see Figure 3) is the key reasons for the
layer shift in P2 MOF-2 A support is obtained from the shifting of the paddlewheel over the
170
horizontal linker (Shift IV) A major objection is made by the vertical linkers (Shift III) The total
interaction becomes repulsive when a vertical linker shifts with respect to the other for about 05 Aring
This may alter the tilt of the linker however a minimum is already established The vertical linkers of
a layer in a way are trapped between those of adjacent layers The shift to either side from P4 most
probably decreases the interlayer separation However this demands further rotation of the vertical
linkers in order to reduce their mutual Pauli repulsion Overall the sharp minimum at P4 may be
taken as an inheritance from the lower interlayer distance of P4 geometry fixed by the anchoring on
the substrate
Figure 9 Shift of individual building blocks in the adjacent layers correspond to the situation in bulk The energies of each shifted arrangement and their total energy are given in the graph
The total energy involved in the shifting of two building blocks (one building block over the other) is
equivalent to the energy per one building block when it feels shift from two neighbors Only the
vertical linker is sensitive to the shift-direction of the two neighbors However since averages were
taken as discussed earlier the total energy becomes independent of the direction Besides the
relative motion of the SBU facing each other in P4 symmetry (Shifts I II and III) for strong distortions
we also have to consider the interaction of adjacent linker-connector interactions as represented in
Shifts IV and V The former stabilizes P2 while the latter support P4 Overall the total energy for all
the shifts shows a local minimum at P4 (see figure 9) in agreement with the periodic calculation
shown in Figure 8 Indeed the relative energy between P2 and P4 stackings estimated by the
171
superposition of individual contributions is 43 meV in close agreement to the 40 meV obtained by
the periodic calculations
Our finite-component model successfully provides adequate information on the individual
contributions of building blocks Vertical linkers play crucial role in holding the SURMOF with P4
symmetry which was never achieved with solvothermally synthesized MOF-2 Paddlewheels are
held most responsible for lateral shift The vertical linkers winningly establish a local minimum at P4
for the SURMOF
Recently we have reported SURMOF-2 derivatives with very large pore sizes(Ref) This has been
achieved by increasing the length of the linker units In view of our analysis of the stacking and
stability of MOF-2 we can now safely state that it will be possible to generate SURMOF-2 derivatives
with even larger pores with pore sizes essentially limited by the availability of stiff long organic
linker molecules This is exemplified by a calculation of the stacking energy of a series of phenyl
oligomers that is SURMOF-2 derivatives incorporating chains with 1 to 10 benzene rings in the
linkers The stacking energies per formula unit are -041 -067 -091 -121 -145 -168 -192 -215
-237 and -26 eV respectively Each benzene ring adds more than 02 eV into the stacking energy per
formula unit This energy is due to the London dispersion interaction between the linkers in the
neighboring layers
The drop of SURMOFs into the local minimum perhaps is blended with some parameters related to
synthetic environments This was beyond the scope of this work however we suggest that studies of
the mechanism of substrate-assisted layer-by-layer deposition of metal and ligand precursors may
give some primary insights into it
CONCLUSION
We have analyzed the reason for the different stackings observed for MOF-2 In the traditional
solvothermal synthesis MOF-2 is likely to crystallize in P2 symmetry determined by the strong
interaction between the paddlewheel units The coordination of large solvent molecules to the free
metal centers may lead to MOF-2 in C2 symmetry In contrast the formation of SURMOFs using
Liquid Phase Epitaxy (LPE) allows the formation of high-symmetry P4 MOF-2 This structure requires
a structural modification in terms of the orientation of the linkers with respect to the free monolayer
and the stacking is stabilized by London dispersion interactions between the linkers Increasing the
linker length is a straightforward way for the linear expansion of pore size and according to our
computations the pore size is only limited by the availability of linker molecules showing the desired
length Thus we presented a rare situation in which the linkers guarantee the persistence of a series
of materials in an otherwise unachievable state
172
COMPUTATIONAL DETAILS
The Becke three-parameter hybrid method combined with Lee-Yang-Par correlational functional
(B3LYP)34-37 and augmented with a dispersion term38 as implemented in Turbomole39 has been used
for DFT calculations The copper atoms were described using the basis set associated with the Hay-
Wadt40 relativistic effective core potentials41 which include polarization f functions42 This basis set
was obtained from the EMSL Basis Set Library4344 Carbon oxygen and hydrogen atoms were
described using the Pople basis set 6-311G Geometry optimizations of the MOF fragments were
performed using internal redundant co-ordinates45 A triplet spin state was assigned for each Cu-
paddlewheel46
Periodic structure calculations at the BLYP47 level were performed using the ADF BAND 2012
code4849 Dispersion correction was implemented based on the work by Grimme50 Triple-zeta basis
set was used The crystalline state of MOFs was computationally described using periodic boundary
conditions Bader charge analysis of paddlewheel-pairs was also performed using ADF 2012 code
The analysis was made at the level of B3LYP-D using an all-electron triple-zeta basis set
The non-orthogonal tight-binding approximation to DFT called Density Functional Tight-Binding
(DFTB)51 is computationally feasible for unit cells of several hundreds of atoms Hence this method
was used for extensive calculations on periodic structures This method computes a transferable set
of parameters from DFT calculations of a few molecules per pair of atom types The more accurate
self-consistent charge (SCC) extension of DFTB52-54 has been used for the study of bulk MOFs Validity
of the Slater-Koster parameters of Cu-X (X = Cu C O H) pairs were reported before55 The
computational code deMonNano56 which has dispersion correction implemented57 was used
If not stated otherwise the energies of the periodic structure are given per a formula unit (FU) of the
MOF which includes one paddlewheel and two linkers (one vertical linker and one horizontal linker)
REFERENCES
(1) Eddaoudi M Moler D B Li H L Chen B L Reineke T M OKeeffe M Yaghi O M Accounts of
Chemical Research 2001 34 319
(2) Li H Eddaoudi M OKeeffe M Yaghi O M Nature 1999 402 276
(3) Rowsell J L C Yaghi O M Microporous and Mesoporous Materials 2004 73 3
(4) Eddaoudi M Li H L Yaghi O M Journal of the American Chemical Society 2000 122 1391
(5) Rowsell J L C Yaghi O M Angewandte Chemie-International Edition 2005 44 4670
173
(6) Suh M P Park H J Prasad T K Lim D-W Chemical Reviews 2012 112 782
(7) Murray L J Dinca M Long J R Chemical Society Reviews 2009 38 1294
(8) Rosi N L Eckert J Eddaoudi M Vodak D T Kim J OKeeffe M Yaghi O M Science 2003 300
1127
(9) Kreno L E Leong K Farha O K Allendorf M Van Duyne R P Hupp J T Chemical Reviews 2012
112 1105
(10) Achmann S Hagen G Kita J Malkowsky I M Kiener C Moos R Sensors 2009 9 1574
(11) Lee J Farha O K Roberts J Scheidt K A Nguyen S T Hupp J T Chemical Society Reviews 2009
38 1450
(12) Farrusseng D Aguado S Pinel C Angewandte Chemie-International Edition 2009 48 7502
(13) Corma A Garcia H Llabres i Xamena F X Chemical Reviews 2010 110 4606
(14) Rowsell J L C Millward A R Park K S Yaghi O M Journal of the American Chemical Society 2004
126 5666
(15) Deng H Doonan C J Furukawa H Ferreira R B Towne J Knobler C B Wang B Yaghi O M
Science 2010 327 846
(16) Liu J Lukose B Shekhah O Arslan H K Weidler P Gliemann H Braumlse S Grosjean S Godt A
Feng X Muumlllen K Magdau I-B Heine T Woumlll C submitted to Nature Chemistry 2012
(17) Li H Eddaoudi M Groy T L Yaghi O M Journal of the American Chemical Society 1998 120 8571
(18) Carson C G Hardcastle K Schwartz J Liu X Hoffmann C Gerhardt R A Tannenbaum R
European Journal of Inorganic Chemistry 2009 2338
(19) Clausen H F Poulsen R D Bond A D Chevallier M A S Iversen B B Journal of Solid State
Chemistry 2005 178 3342
(20) Edgar M Mitchell R Slawin A M Z Lightfoot P Wright P A Chemistry-a European Journal 2001
7 5168
(21) Schaate A Roy P Preusse T Lohmeier S J Godt A Behrens P Chemistry-a European Journal
2011 17 9320
(22) Cote A P Benin A I Ockwig N W OKeeffe M Matzger A J Yaghi O M Science 2005 310
1166
(23) Wan S Guo J Kim J Ihee H Jiang D Angewandte Chemie-International Edition 2008 47 8826
174
(24) Lukose B Kuc A Frenzel J Heine T Beilstein Journal of Nanotechnology 2010 1 60
(25) Kitagawa S Kitaura R Noro S Angewandte Chemie-International Edition 2004 43 2334
(26) Shekhah O Wang H Zacher D Fischer R A Woell C Angewandte Chemie-International Edition
2009 48 5038
(27) Shekhah O Wang H Kowarik S Schreiber F Paulus M Tolan M Sternemann C Evers F
Zacher D Fischer R A Woll C Journal of the American Chemical Society 2007 129 15118
(28) Zacher D Schmid R Woell C Fischer R A Angewandte Chemie-International Edition 2011 50 176
(29) Bader R F W Accounts of Chemical Research 1985 18 9
(30) Merino G Vela A Heine T Chemical Reviews 2005 105 3812
(31) Tafipolsky M Schmid R Journal of Chemical Theory and Computation 2009 5 2822
(32) Kuc A Enyashin A Seifert G Journal of Physical Chemistry B 2007 111 8179
(33) Winston E B Lowell P J Vacek J Chocholousova J Michl J Price J C Physical Chemistry
Chemical Physics 2008 10 5188
(34) Becke A D Journal of Chemical Physics 1993 98 5648
(35) Lee C T Yang W T Parr R G Physical Review B 1988 37 785
(36) Vosko S H Wilk L Nusair M Canadian Journal of Physics 1980 58 1200
(37) Stephens P J Devlin F J Chabalowski C F Frisch M J Journal of Physical Chemistry 1994 98
11623
(38) Civalleri B Zicovich-Wilson C M Valenzano L Ugliengo P Crystengcomm 2008 10 405
(39) University of Karlsruhe and Forschungszentrum Karlsruhe gGmbH - TURBOMOLE V63 2011 2007
(40) Wadt W R Hay P J Journal of Chemical Physics 1985 82 284
(41) Roy L E Hay P J Martin R L Journal of Chemical Theory and Computation 2008 4 1029
(42) Ehlers A W Bohme M Dapprich S Gobbi A Hollwarth A Jonas V Kohler K F Stegmann R
Veldkamp A Frenking G Chemical Physics Letters 1993 208 111
(43) Feller D Journal of Computational Chemistry 1996 17 1571
(44) Schuchardt K L Didier B T Elsethagen T Sun L Gurumoorthi V Chase J Li J Windus T L
Journal of Chemical Information and Modeling 2007 47 1045
175
(45) von Arnim M Ahlrichs R Journal of Chemical Physics 1999 111 9183
(46) St Petkov P Vayssilov G N Liu J Shekhah O Wang Y Woell C Heine T Chemphyschem 2012
13 2025
(47) Gill P M W Johnson B G Pople J A Frisch M J Chemical Physics Letters 1992 197 499
(48) SCM Amsterdam Density Functional 2012
(49) Velde G T Bickelhaupt F M Baerends E J Guerra C F Van Gisbergen S J A Snijders J G
Ziegler T Journal of Computational Chemistry 2001 22 931
(50) Grimme S Journal of Computational Chemistry 2006 27 1787
(51) Seifert G Porezag D Frauenheim T International Journal of Quantum Chemistry 1996 58 185
(52) Elstner M Porezag D Jungnickel G Elsner J Haugk M Frauenheim T Suhai S Seifert G
Physical Review B 1998 58 7260
(53) Frauenheim T Seifert G Elstner M Hajnal Z Jungnickel G Porezag D Suhai S Scholz R
Physica Status Solidi B-Basic Research 2000 217 41
(54) Oliveira A F Seifert G Heine T Duarte H A Journal of the Brazilian Chemical Society 2009 20
1193
(55) Lukose B Supronowicz B Petkov P S Frenzel J Kuc A Seifert G Vayssilov G N Heine T
physica status solidi (b) 2011
(56) Heine T Rapacioli M Patchkovskii S Frenzel J Koester A M Calaminici P Escalante S Duarte
H A Flores R Geudtner G Goursot A Reveles J U Vela A Salahub D R deMon-nano edn ed deMon
2009
(57) Zhechkov L Heine T Patchkovskii S Seifert G Duarte H A Journal of Chemical Theory and
Computation 2005 1 841
Statutory Declaration
I Binit Lukose hereby declare that I have written this PhD thesis independently
unless where clearly stated otherwise I have used only the sources the data
and the support that I have clearly mentioned This PhD thesis has not been
submitted for conferral of degree elsewhere
Bremen 2012
Signature _________________________
i
List of Articles
1 Binit Lukose Agnieszka Kuc Johannes Frenzel Thomas Heine On the reticular construction
concept of covalent organic frameworks Beilstein J Nanotechnol 2010 1 60ndash70
DOI103762bjnano18
2 Binit Lukose Agnieszka Kuc Thomas Heine The Structure of Layered Covalent-Organic
Frameworks Chem Eur J 2011 17 2388 ndash 2392 DOI 101002chem201001290
3 Binit Lukose Barbara Supronowicz Petko St Petkov Johannes Frenzel Agnieszka B Kuc
Gotthard Seifert Georgi N Vayssilov and Thomas Heine Structural properties of metal-
organic frameworks within the density-functional based tight-binding method Phys Status
Solidi B 2012 249 335ndash342 DOI 101002pssb201100634
4 Binit Lukose Agnieszka Kuc Thomas Heine Stability and electronic properties of 3D covalent
organic frameworks Prepared for publication
5 Binit Lukose Mohammad Wahiduzzaman Agnieszka Kuc Thomas Heine Structure
electronic structure and hydrogen adsorption capacity of porous aromatic frameworks
Prepared for publication
6 Jinxuan Liu Binit Lukose Osama Shekhah Hasan Kemal Arslan Peter Weidler Hartmut
Gliemann Stefan Braumlse Sylvain Grosjean Adelheid Godt Xinliang Feng Klaus Muumlllen Ioan-
Bogdan Magdau Thomas Heine Christof Woumlll A novel series of isoreticular metal organic
frameworks realizing metastable structures by liquid phase epitaxy Prepared for publication
7 Binit Lukose Gabriel Merino Christof Woumlll Thomas Heine Linker guided metastability in
templated Metal-Organic Framework-2 derivatives (SURMOFs-2) Prepared for publication
8 Binit Lukose Thomas Heine Review Covalently-bound organic frameworks Prepared for
publication
ii
Acknowledgment
Foremost I would like to thank my supervisor Prof Dr Thomas Heine for the wonderful opportunity to join his group as his PhD student I am greatly thankful to him for giving me the topic and sharing with me his expertise and research insight His thoughtful advices have served to give me senses of motion and direction His ambitious approach to science has given me motivation as well as chances and exposures to develop in science His constant attention and guidance have led my scientific outputs to the best levels possible I am also thankful for the financial support and the comfortable stay in his group during my PhD time Additionally he is acknowledged for correcting and reviewing my thesis
Prof Dr Ulrich Kleinekathoumlfer deserves special thanks as my Thesis Committee member I am very glad to have him in the Committee and greatly thankful for reviewing and evaluating the thesis I also thank him and Prof Ulrich Kortz for the evaluation of my PhD proposal I am thankful also for their friendly manners and considerations throughout my PhD time
Prof Dr Christof Woumlll Director of Functional Interfaces Karlsruhe Institute of Technology is greatly acknowledged for being the external Thesis Committee member I am greatly thankful for the evaluation and reviewing of the thesis I am very much moved by his research outcomes and thankful for sharing them with us Our collaborations with his group have particularly enriched my thesis
Prof Dr Petko Petkov is also acknowledged for reviewing my thesis I particlulary thank him for the friendship and discussions thoughout my PhD time
I am indebted to Dr Agnieszka Kuc for introducing me to the topic of nanoporous materials Her experience and expertise have helped me to begin a career in this field I extend my gratitude for sharing with me her scientific skills and correcting our joint-articles
Dr Lyuben Zhechkov and Dr Achim Gelessus have been great in providing computational assistance I have benefitted from their knowledge and sincerity through fast and timely helps
I owe my heartfelt thanks to Dr Lyuben Zhechkov Dr Nina Vankova Dr Augusto Oliveira Dr Andreas Mavrantonakis Dr Stefano Borini and Dr Christian Walther for all discussions suggestions support help and particularly their lectures Dr Lyuben Zhechkov and Dr Nina Vankova are specially mentioned for their long-term attentions and helps Dr Akhilesh Tanwar is acknowledged for his helps in the beginning of my PhD
In my daily work I have been blessed with a friendly and cheerful group of fellow students Barbara Jianping Wahid Nourdine Mahdi Lei Rosalba Ievgenia Wenqing Guilherme Farjana Maicon Aleksandar Ionut Yulia and Gabriel Discussions aside I had great fun times with them Our interactions have also helped me to develop in a personal level I thank them from my full heart although just a few words are not enough I specially thank Barbara Wahid and Ionut for the joint works and publications
Mrs Britta Berninghausen our project assistant deserves special thanks for the friendly assistance on all matters with the university administration
I thank all the members of the group for a lot of good things From the supervisor to the newly joined member everyone has contributed for the general good fun and easiness All those ldquobio-fuelrdquo workshops barbecues parties retreats and gatherings are unforgettable The group also kept good phase with other groups and visitors I thank all the members once again for the good times I would not have been happier anywhere else
iii
I extend my thanks to the research groups that I visited during the PhD time Dr Sourav Pal Director of National Chemical Laboratory Pune and Dr V Subrahmanian Central Leather Research Institute Chennai deserve my gratitude for giving me the opportunity to visit and work with their group members Also I am very thankful to Prof D Sc Georgi N Vayssilov Faculty of Chemistry University of Sofia for the interesting collaboration and visit to his group The financial assistance during each stay is greatly acknowledged I also thank the members of the respective groups namely Dr Petko Petkov and his family who made the visit to Bulgaria very much entertaining
Prof Dr Lars Pettersson of University of Stockholm Dr Tzonka Mineva of CNRS Montpellier and all other members of the HYPOMAP research project are acknowledged for the scientific discussions exposures and promotions
I acknowledge several projects of Prof Dr Thomas Heine for the financial support of my work and travel the funding sources include the European Commission Deutsche Forschungsgemeinschaft (DFG) and the joint Bulgarian-German exchange program (DAAD)
I thank all the co-authors of my publications who have contributed their knowledge ideas and work to accomplish our scientific goals Without their efforts all those works would not have been complete
Members of Research III of SES at Jacobs University namely Robert Carsten Joumlrg Bogdan Meisam Niraj Mahesh Vinu Pinky Patrice Mehdi Sidhant and all professors postdocs and students in Nanofun center are thankfully mentioned here
A lot of my friends in the campus deserve my thanks Mahesh Mahendran Vinu Deepa Srikanth Rajesh Arumugam Prasad Dhananjay Sunil Tripti Raghu Suneetha Rami Susruta Niraj Abhishek Ashok Rakesh Sagar Rohan Naveen Yauhen Yannic Mila and Samira are thanked for the gatherings travels making funs and those cricket and volleyball evenings Some of them are specially thanked for the occasional ldquogahn bayrdquo parties I owe many thanks to Yauhen Srikanth and Prasad for being good flat-mates and having talks on any matters Srikanth and Prasad are thanked again for generously extending their cooking skills to me
I wish to thank everybody with whom I have shared experiences in life I am obliged to my MSc lecturer Dr Rajan K John whose dreams have inspired and driven me to research In particular his accomplishments in the George Sudarshan Center CMS College Kottayam have molded me to take up this career My previous research supervisors Prof S Lakshmibala and Prof V Balakrishnan of IIT Madras and Dr Anita Mehta of SNBNCBS Kolkata are also acknowledged for their important influences in my academic life Additionally all my teachers friends and well-wishers from neighborhood school college GS Center IIT-M and SN Bose center are thanked and acknowledged Members of St Antonyrsquos Parish Olassa are also thanked for the regards and encouragement
Jacobs University Bremen and its people have been amazing in all sorts of things I am glad that I have been a member of the University With my full heart I thank the university authority for all its facilities that were open for me I also thank Dr Svenja Frischholz Mr Peter Tsvetkov and Ms Kaija Gruumlenefeld in the administration departments for the timely helps
Lastly and most importantly I wish to thank my dearest ones for all the sacrifices and love My parents K P Lukose and Molly and my brother Anit deserve to be thanked They have always supported and encouraged me to do my best in all matters of life I also wish to thank my entire extended family for providing me a loving environment
iv
Abstract
Framework materials are extended structures that are built into destined nanoscale architectures
using molecular building units Reticular synthesis methods allow stitching of a large variety of
molecules into predicted networks Porosity is an obvious outcome of the stitching process These
materials are classified and named according to the chemical composition of the building blocks For
instance Metal-Organic Frameworks (MOFs) consists of metal-oxide centers that are linked together
by organic entities The stitching process is straight-forward so that there are already thousands of
them synthesized Controlled growth of MOFs on substrates leads to what is known as surface-MOFs
(SURMOFs) A low-weight metal-free version of MOFs is known as Covalent Organic Frameworks
(COFs) They consist of light elements such as boron oxygen silicon nitrogen carbon and hydrogen
atoms Diamond-like structures with a variety of linear organic linkers between tetrahedral nodes is
called Porous Aromatic Frameworks (PAFs)
The thesis is composed of computational studies of the above mentioned classes of materials The
significance of such studies lies in the insights that it gives about the structure-property relationships
Density Functional Theory (DFT) and its Tight-Binding approximate method (DFTB) were used in
order to perform extensive calculations on finite and periodic structures of several frameworks DFTB
provides an ab-initio base on periodic structure calculations of very large crystals which are typically
studied only using force-field methods The accuracy of this approximate method is validated prior to
reasoning
As the materials are energized from building units and coordination (or binding) stability vs
structure is discussed Energy of formation and mechanical strength are particularly calculated Using
dispersion corrected (DC)-Self Consistent Charge (SCC) DFTB we asserted that 2D COFs should have a
layer arrangement different from experimental suggestions Our arguments supported by simulated
PXRDs were later verified using higher level theories in the literature Another benchmark is giving an
insightful view on the recently reported difference in symmetries of two-dimensional MOFs and
SURMOFs The result of it shows an extra-ordinary role of linkers in SURMOFs providing
metastability
Electronic properties of all the materials and hydrogen adsorption capacities of PAFs are discussed
COFs PAFs and many of the MOFs are semiconductors HOMO-LUMO gaps of molecular units have
crucial influence in the band gaps of the solids Density of states (DOS) of solids corresponds to that
of cluster models Designed PAFs give a large choice of adsorption capacities one of them exceeds
the DOE (US) 2005 target Generally the studies covered under the scheme of the thesis illustrate
the structure stability and properties of framework materials
- Dedicated to my Family and Rajan sir
Table of Contents 1 Outline 1
2 Introduction 2
21 Nanoporous Materials 2
22 Reticular Chemistry 3
23 Metal-Organic Frameworks 5
24 Covalently-bound Organic Frameworks 8
3 Methodology and Validation 10
31 Methods and Codes 10
32 DFTB Validation 11
4 2D Covalent Organic Frameworks 13
41 Stacking 13
42 Concept of Reticular Chemistry 15
5 3D Frameworks 17
51 3D Covalent Organic Frameworks 17
52 Porous Aromatic Frameworks 18
6 New Building Concepts 20
61 Isoreticular Series of SURMOFs 20
62 Metastability of SURMOFs 21
7 Summary 23
71 Validation of Methods 23
72 Weak Interactions in 2D Materials 25
73 Structure-Property Relationships 27
List of Abbreviations 31
List of Figures 32
References 33
Appendix A Review of covalently-bound organic frameworks 37
Appendix B Properties of MOFs within DFTB 81
Appendix C Stacking of 2D COFs 96
Appendix D Reticular concepts applied to 2D COFs 105
Appendix E Properties of 3D COFs 122
Appendix F Properties of PAFs 131
Appendix G Isoreticular SURMOFs of varying pore sizes 145
Appendix H Metastability in 2D SURMOFs 160
1
1 Outline
I prepared this cumulative thesis as it is permitted by Jacobs University Bremen as all work has been
published in international peer-reviewed journals is submitted for publication or in a late
manuscript state in order to be submitted soon The list of articles contains three published papers
three submitted manuscripts and two manuscripts that are to be submitted The articles are given in
Appendices A-H in the order of their discussions Each appendix has one paper and its supporting
information
The chapters from ldquoIntroductionrdquo to ldquoNew Building Conceptsrdquo contain general discussions about the
articles and provide a red thread leading through the articles The discussions mainly circle around
the context and the content of the articles
The chapter ldquoIntroductionrdquo provides a general introduction to the topic and a review of the materials
discussed in this thesis As no comprehensive review on ldquoCovalently-bound Organic Frameworksrdquo is
available in the literature we prepared a manuscript (Appendix A) covering that subject The chapter
ldquoMethodology and Validationrdquo discusses one article on validation of DFTB method in Metal-Organic
Frameworks (Appendix B) Each consecutive chapter discusses two articles The chapter ldquo2D
Covalent Organic Frameworksrdquo covers the studies on layer arrangements (Appendix C) followed by
analysis on how reticular chemistry concepts can be used to design new materials (Appendix D) The
chapter ldquo3D Frameworksrdquo discusses the stability and properties of 3D COFs (Appendix E) and PAFs
(Appendix F) It also discusses hydrogen storage capacities of PAFs The chapter ldquoNew Building
Conceptsrdquo has coverage on 2D MOFs that are built on organic surfaces The first article in this chapter
describes the achievement of our collaborators in synthesizing a series of SURMOFs of varying pore
sizes supported by our calculations indicating their matastability Extensive calculations revealing the
role of linkers in creating the unprecedented difference of SURMOFs in comparison with the bulk
MOFs is described in another article
Details of the articles and references to the appendices are given in the respective places in each
chapter The chapter ldquoSummaryrdquo gives an overview of all the results and discussions It also discusses
some impacts of the publications and concludes the thesis Overall the studies bring into picture
different classes of materials and analyze their structural stabilities and properties
2
2 Introduction
21 Nanoporous Materials
The field of nanomaterials covers materials that have properties stemming from their nanoscale
dimensions (1-100 nm) In contrast to nanomaterials which define size scaling of bulk materials the
major determinant of nanoporous materials is their pores Nanoporous materials are defined as
porous materials with pore diameters less than 100 nm and are classified as micropores of less than
2 nm in diameter mesopores between 2 and 50 nm and macropores of greater than 50 nm They
have perfectly ordered voids to accommodate interact with and discriminate molecules leading to
prominent applications such as gas storage separation and sieving catalysis filtration and
sensoring1-4 They differ from the generally defined nanomaterials in the sense that their properties
are mostly determined by pore specifications rather than by bulk and surface scales Hence the
focus is onto the porous properties of the materials
Utilization of the pores for certain applications relies on certain parameters such as pore size pore
volume internal surface area and wall composition For example physical adsorption of gases is high
in a material with large surface area which implies significantly high storage in comparison to a tank
Porosity can be measured using some inert or simple gas adsorption measurements Distribution of
pore size can be sketched from the adsorptiondesorption isotherm
Nanoporous materials abound in nature Zeolites for instance many of them are available as mineals
have been used in petroleum industry as catalysts for decades The walls of human cells are
nanoporous membranes Other examples are clays aluminosilicate minerals and microporous
charcoals Zeolites are microporous materials from the aluminosilicate family and are often known as
molecular sieves due to their ability to selectively sort molecules primarily based on a size exclusion
principle A material with high carbon content (coal wood coconut shells etc) can be converted to
activated carbon (AC) by controlled thermal decomposition in a furnace The resultant product has
large surface area (~ 500 m2 g-1) Porous glass is a material produced from borosilicate glasses having
pore sizes usually in nm to μm range With increasing environmental concerns worldwide porous
materials have become a suitable choice for separation of polluting gases storage and transport of
energy carriers etc Synthesis of open structures has advanced this area5-7 especially since the
invention of MCM-41 (Mobil Composition of Matter No 41) by Mobil scientists89 Furthermore
there are many templating pathways in making nanoporous materials10-13 Currently it is possible to
engineer the internal geometry at molecular scales
3
For more than a decade chemists are able to synthesize extended structures from well-defined and
rigid molecular building units Such designed and controlled extensions provide porosity which can
be scaled and modified by selecting appropriate building blocks The first realization of this kind was
a class of materials known as Metal-Organic Frameworks (MOFs) in which metal-oxides are stitched
together by organic molecules Synthesis of molecules into predicted frameworks have led to the
emergence of a discipline called reticular chemistry14 The new design-based synthetic approaches
have produced large number of nanoporous materials in comparison to the discovery-based
synthetic chemistry
22 Reticular Chemistry
The discipline of designed (rational) synthesis of materials is known as reticular chemistry Desired
materials can be realized by starting with well-defined and rigid molecular building blocks that will
maintain their structural integrity throughout the construction process The extended structures
adopt high symmetry topologies The synthetic approach follows well-defined conditions which
provide general control over the character of solids In short it is the chemistry of linking molecular
building blocks by strong bonds into predetermined structures
The knowledge about how atoms organize themselves during synthesis is essential for the design
The principal possibilities rely on the starting compounds and the reaction mechanisms Porosity is
almost an inevitable outcome of reticular synthesis Solvents may be used in most cases as space-
filling agents and in cases of more than one possibility as structure-directing agents
Thousands of materials in large varieties have been synthesized using the reticular chemistry
principles Reticular Chemistry Structure Resource (RCSR) is a searchable database for nets a project
initiated by of Omar M Yaghi and Michael OrsquoKeeffe15 They define nets as the collection of vertices
and edges that form an irreducible network in which any two vertices are connected through at least
one continuous path of edges Building units are finite nets as in the case of a polyhedron Periodic
structures have one two or three-periodic nets An example case of 3D diamond net (dia) is shown in
Figure 1 Graph-theoretical aspects together with explicative terminology and definitions can be
found in the literature16-18
Figure 1 a) ldquoAdamantinerdquo unit of diamond structure and b) diamond (dia) net
4
In other words a framework can be deconstructed into one or more fundamental building blocks
each of them assigned by a vertex in the net The vertices are the branching points and edges are
joining them The realization of the net in space by representing the vertices and lattice parameters
by co-ordinates and a matrix respectively is called embedding of the net Underlying topology of an
extended structure is the structure of the net inherited from the crystal structure that is invariant
under different embeddings For example Cu-BTC MOF has paddlewheels and benzene rings as
fundamental blocks The MOF structure can be simplified into its underlying topology as shown in
Figure 2
Figure 2 CU-BTC MOF and the corresponding tbo net
Alternatively the topology of a framework can be defined using the convention of so-called
secondary building units (SBUs) The shapes of SBUs are defined by points of extension of the
fundamental building blocks SBUs are invariant for building units of identical connectivity Based on
the number of connections with the adjacent blocks (4 for paddlewheel and 3 for the benzene) SBUs
of Cu-BTC can be represented as shown in Figure 3a Assembly of SBUs following the network
topology and insertion of the fundamental building blocks give the crystal structure (see Figure 3b for
the extension of SBUs to the topology of Cu-BTC)
In addition to MOFs Metal-Organic Polyhedra (MOP)19 Zeolite Imidazolate Frameworks (ZIFs)20 and
Covalent-Organic Frameworks (COFs)2122 are developing classes of materials within reticular
chemistry the first members of all of them were synthesized by the group of Yaghi MOPs are nano-
sized polyhedra achieved by linking transition metal ions and either nitrogen or carboxylate donor
organic units ZIFs are aluminosilicate zeolites with replacements of tetrahedral Si (Al) and bridging
oxygen by transition metal ion and imidazolate link respectively COFs are extended organic
5
structures constructed solely from light elements (H B C and O) The materials synthesized under
the reticular scheme are largely crystalline
Figure 3 a) Fundamental building blocks and SBUs of Cu-BTC and b) extension of SBUs following
crystal structure
23 Metal-Organic Frameworks
MOFs are porous crystalline materials consisting of metal complexes (connectors) linked together by
rigid organic molecules (linkers) MOFs are analogues to a class of materials called coordination
polymers (CPs) However there are primary differences between them CPs are inorganic or
organometallic polymer structures containing metal ions linked by organic ligands A ligand is an
atom or a group of atoms that donate one or more lone pairs of electrons to the metal cations and
thereby participate in the formation of a coordination complex In MOFs typically metal-oxide
centers are used instead of single metal ions as they provide strong bonds with organic linkers This
provides not only high stability but also high directionality because multiple bonds are involved
6
between metal-centers and organic linkers Predictability lies in the pre-knowledge about the
connector-linker interactions Thus the reticular design of MOFs derives from the precise
coordination geometry between metal centers and the linkers Figure 4a shows a schematic diagram
of MOFs By varying the chemical composition of nodes and linkers thousands of different MOF
structures with a large variety in pore size and structure have been synthesized Figure 4b shows
MOF-5 that consists of zinc-oxide tetrahedrons and benzene linkers
Figure 4 a) Schematic diagram of the single pore of MOF b) An example of a simple cubic MOF in
which zinc-oxide tetrahedron are linked together by benzene linkers Atom colors blue ndash Zn red ndash
O grey ndash C white ndash H
The synthesis of MOFs differs from organic polymerizations in the degree of reversible bond
formation Reversibility allows detachment of incoherently matched monomers followed by their
attachment to form defect-free crystals Assembly of monomers occurs as single step hence
synthetic conditions are of high importance The strength of the metal-organic bond is an obstacle
for reversible bond formation however solvothermal techniques are found out to be a convenient
solution23 Solvothermal synthesis generally allows control over size and shape distribution Using
post-synthetic methods further changes on cavity sizes and chemical affinities can be made
Materials that are stable with open pores after removal of guest molecules are termed as open-
frameworks The stability of guest-free materials is usually examined by Powder X-ray diffraction
(PXRD) analysis Thermogravimetric analysis may be performed to inspect the thermal stability of the
material Elemental analysis can detail the elemental composition of the material Physical
techniques such as Fourier transform infrared (FTIR) spectra and nuclear magnetic resonance (NMR)
may be used to verify the condensation of monomers to the desired topology Porosity can be
evidenced from adsorption isotherms of gases or mercury porosimetry
7
The number of linkers that are allowed to bind to the metal-center and the orientations of the linkers
depend exclusively on the coordination preferences of the metal The organic linkers are typically
ditopic or polytopic They are essential in determining the topology and providing porosity Longer
linkers provide larger pore size A series of compounds with the same underlying topology and
different linker sizes are called iso-reticular series The structure of MOFs in principle can be formed
into the requirement of prominent applications such as gas storage gas separation sensing and
catalysis The structural divergence and performance can be further increased by functionalizing the
organic linkers Hence several attempts are on-going in purpose to come up with the best material
possible in each application
Interpenetration of frameworks is an inevitable outcome of large pore volume Interpenetrated nets
are periodic equivalents of mechanically interlocked molecules rotaxanes and catenanes Depending
on topology they are either maximally separated termed as interpenetration or minimally separated
termed as interweaving (see Figure 5) Interpenetration can be beneficial to some porous structures
protecting from collapse upon removal of solvents
Figure 5 Schematic diagram of interpenetrated and interwoven frameworks
Controlled growth of MOFs on organic surfaces is achieved by R A Fischer and C Woumlll24 The then
named SURMOFs allow to fabricate structures possibly not achievable by bulk synthesis The growth
is achieved through liquid phase epitaxy (LPE) of MOF reactants on the surface (nucleation site) A
step-by-step approach which is alternate immersion of the substrate in metal and ligand precursors
supplies control of the growth mechanism
8
Figure 6 Schematic diagram of SURMOF
24 Covalently-bound Organic Frameworks
As a result of the long-term struggle for complete covalent crystallization in the year 2005 Cocircte et
al21 synthesized porous crystalline extended 2D Covalent-Organic Frameworks (COFs) using
reticular concepts The success was followed by the design and synthesis of 3D COFs in the year
200722 By now there are about 50 COFs reported in the literature COFs are made entirely from
light elements and the building blocks are held together by strong covalent bonds Most of them
were formed by solvothermal condensation of boronic acids with hydroxyphenyl compounds
Tetragonal building blocks were used for the formation of 3D COFs Alternate synthetic methods
were also used for producing COFs COFs are generally studied for gas storage applications However
they have also shown potentialities in photonic and catalytic applications
Alternative synthesis methods paved the way to new covalently bound organic frameworks
Trimerization of polytriazine monomers in ionothermal conditions gives rise to Covalent Triazine
Frameworks (CTFs)25 Similarly coupling reactions of halogenated monomers produced Porous
Aromatic Frameworks (PAFs) Albeit the short-range order one of the PAFs showed very high surface
area (5600 m2 g-1) and gas uptake capacity26
Due to low weight the covalently-bound materials show very high gravimetric capacities
Suggestions such as metal-doping functionalization and geometry modifications can be found in the
literature for the general improvement of the functionalities There are also various studies of their
structure and properties
A review on the synthesis structure and applications of covalently bound organic frameworks has
been prepared for publication
Article 1 Covalently-bound organic frameworks
Binit Lukose Thomas Heine
9
See Appendix A for the article
My contributions include collecting data and preparing a preliminary manuscript
Figure 7 SBUs and topologies of 2D COFs
10
3 Methodology and Validation
31 Methods and Codes
The Density-Functional based Tight Binding (DFTB) method2728 is an approximate Kohn-Sham DFT29-31
scheme It avoids any empirical parameterization by calculating the Hamiltonian and overlap matrix
elements from DFT-derived local orbitals (atomic orbitals) and atomic potentials The Kohn-Sham
orbitals are represented with the linear combination of atomic orbitals (LCAO) scheme The matrix
elements of Hamiltonian are restricted to include only one- and two-center contributions Therefore
they can be calculated and tabulated in advance as functions of the distance between atomic pairs
The remaining contributions to the total energy that is the nucleus-nucleus repulsion and the
electronic double counting terms are grouped in the so-called repulsive potential This two-center
potential is fitted to results of DFT calculations of properly chosen reference systems The accuracy
and transferability can be improved with the self-consistent charge (SCC) extension to DFTB32 This
method is based on the second-order expansion of the Kohn-Sham total energy with respect to
charge density fluctuations which are estimated by Mulliken charge analysis In order to account for
London dispersion empirical dispersion energy can be added to the total energy3334 Various reviews
are available on DFTB notably the recent ones by Oliveira et al35 and by Seifert et al36
DFTB is implemented in a large number of computer codes For this work we employed the codes
deMonNano37 and DFTB+38 as they allow DFTB calculations on periodic and finite structures
Typically dispersion corrected (DC) SCC-DFTB calculations were carried out Periodic boundary
conditions were used to represent the crystalline frameworks and as the unit cells are large the
standard approach used the point approximation Electronic density of states (DOS) have been
calculated using the DFTB+ code using k-point sampling where the k space was determined by
reaching convergence for the total energy according to the scheme proposed by Monkhorst and
Pack39
For DFT calculations of finite and periodic structures ADF40 Turbomole41 and Crystal0942 were used
For studies of finite models of COFs the calculations were performed at PBEDZP level However for
extensive calculations on SURMOF-2 models we used B3LYP-D The copper atoms were described
using the basis set associated with the Hay-Wadt43 relativistic effective core potentials44 which
include polarization f functions45 Carbon oxygen and hydrogen atoms were described using the
Pople basis set 6-311G
Details of the individual calculations are given in the individual articles in the appendix of this thesis
11
32 DFTB Validation
Figure 8 Deconstructed building units their schematic diagrams and final geometries of HKUST-1
(Cu-BTC) MOF-5 MOF-177 MOF-205 and MIL-53
12
In the literature MOFs and COFs are largely studied for applications such as gas storage using
classical force field methods46-48 First principles based studies of several hundreds of atoms are
computationally expensive Hence they are generally limited to cluster models of the periodic
structures Contrarily DFTB paves the way to model periodic structures involving large numbers of
atoms Being an approximate method DFTB holds less accuracy hence comparisons to experimental
data or higher level methods should be performed for validation
As MOFs contain metal centers andor metal oxides as well as organic elements the DFTB
parameters for both heavy and light elements as well as their mixtures are required Thus we have
chosen MOFs such as Cu-BTC (HKUST-1) MOF-5 MOF-177 MOF-205 and MIL-53 as our model
structures which contain Cu Zn and Al atoms apart from C O and H The MOFs comprising three
common connector units copper paddlewheels (HKUST-1) zinc oxide Zn4O tetrahedron (MOF-5
MOF-177 DUT-6 (MOF-205)) and aluminum oxide AlO4(OH)2 octahedron (MIL-53) Figure 8 shows
the schematic diagram of the MOFs
The validation calculations have been published
Article 2 Structural properties of metal-organic frameworks within the density-functional based
tight-binding method
Binit Lukose Barbara Supronowicz Petko St Petkov Johannes Frenzel Agnieszka B Kuc Gotthard
Seifert Georgi N Vayssilov and Thomas Heine Phys Status Solidi B 2012 249 335ndash342 DOI
101002pssb201100634
See Appendix B for the article
In this article DFTB has been validated against full hybrid density-functional calculations for model
clusters against gradient corrected density-functional calculations for supercells and against
experiment Moreover the modular concept of MOF chemistry has been discussed on the basis of
their electronic properties Our results show that DC-SCC-DFTB predicts structural parameters with a
good accuracy (with less than 5 deviation even for adsorbed CO and H2O on HKUST-1) while
adsorption energies differ by 12 kJ mol-1 or less for CO and water compared to DFT benchmark
calculations
My contributions to this article include performing DFTB based calculations on the MOFs (HKUST-1
MOF-5 MOF-177 and MOF-205) that is optimizing the geometries simulating powder X-ray
diffraction patterns and calculating density of states and bulk modulus Additional involvement is in
comparing structural parameters such as bond lengths bond angles dihedral angles and bulk
modulus with experimental data or data derived from DFT calculations and preparing the manuscript
13
4 2D Covalent Organic Frameworks
41 Stacking
Condensation of planar boronic acids and hydroxyphenyl compounds leads to the formation of two-
dimensional covalent organic frameworks (2D COFs) The layers are held together by London
dispersion interactions The first synthesized COFs COF-1 and COF-5 were reported to have AB
(P63mmc staggered as in graphite) and AA (P6mmm eclipsed as in boron nitride) stackings
respectively (see Figure 9)21 The synthesis of several further 2D COFs then followed49-51 all of them
were inspected for only two stacking possibilities ndash P6mmm or P63mmc and it was reported that
they aggregate in P6mmm symmetry As framework materials possess framework charges the
interlayer interactions significantly include Coulomb interactions P6mmm symmetry is the face-to-
face arrangement where the overlap of the stacked structures is maximized (maximization of the
London dispersion energy) however atom types of alike charges are facing each other in the closest
possible way With the interlayer separation similar to that in graphite or boron nitride the Coulomb
repulsion should be high in such arrangements One should notice that in the example case of boron
nitride the facing atom types are different We therefore assumed that a stable stacking should
possess layer-offset
Figure 9 AA and AB layer stacks of hexagonal layers
We considered two symmetric directions for layer shift and studied their total energies (see Figure
10) The stable geometries are found to have a layer-shift of ~14 Aring a value corresponding to the
shift of AA to AB stacking in graphite and compatible with the bond length between two 2nd row
atoms This stability-supported stacking arrangement as revealed from our calculations was
14
supported by good agreement between simulated and experimental PXRD patterns Hence
independent of the elementary building blocks any 2D COF should expose a layer-offset Based on
the direction of the shift we proposed two stacking kinds serrated and inclined (Figure 10) In the
former the layer-offset is back and forth while in the latter the layer-offset followed single direction
As serrated and inclined stackings have no significant change in stacking energy our calculations
cannot predict the long-range stacking in the crystal However this problem is known from other
layered compounds and the ultimate answer is yet to be revealed by analyzing high-quality
crystalline phases at low temperature
Figure 10 Stacking energy Es vs shift of adjacent layers for COF-5 Different stacking possibilities
and their energies are also shown
We published our analysis of the stacking in 2D COFs
Article 3 The Structure of Layered Covalent-Organic Frameworks
Binit Lukose Agnieszka Kuc Thomas Heine Chem Eur J 2011 17 2388 ndash 2392 DOI
101002chem201001290
See Appendix C for the article
15
My contributions to this article include performing the shift calculations simulating XRDs and partly
preparing the manuscript The article has made certain impact in the literature Most of the 2D COFs
synthesized afterwards were inspected for their stacking stability The instability of AA stacking was
also suggested by the studies of elastic properties of COF-1 and COF-5 by Zhou et al52 The shear
modulus shows negative signs for the vertical alignment of COF layers while they are small but
positive for the offset of layers Detailed analysis performed by Dichtel53 et al using DFT was
confirming the instability of AA stacking and suggesting shifts of about 13 to 25 Aring
42 Concept of Reticular Chemistry
Reticular chemistry means that (functional) molecules can be stitched together to form regular
networks The structural integrity of these molecules we also speak of building blocks remains in the
crystal lattices Consequently also the electronic structure and hence the functionality of these
molecules should remain similar
2D COFs are formed by condensation of well-defined planar building blocks With the usage of linear
and triangular building blocks hexagonal networks are expected The properties of each COF may
differ due to its unique constituents However the extent of the relationship of the properties of
building blocks in and outside the framework has not been studied in the literature
Reticular chemistry allows the design of framework materials with pre-knowledge of starting
compounds and reaction mechanisms Reverse of this is finding reactants for a certain topology We
intended to propose some building units suitable to form layered structures (see Figure 11) The
building units obey the regulations of reticular chemistry and offer a variety of structural and
electronic parameters
Our strategic studies on a set of designed COFs have been published
Article 4 On the reticular construction concept of covalent organic frameworks
Binit Lukose Agnieszka Kuc Johannes Frenzel Thomas Heine Beilstein J Nanotechnol 2010 1
60ndash70 DOI103762bjnano18
See Appendix D for the article
16
Figure 11 Schematic diagram of different building units forming 2D COFs
Various hexagonal 2D COFs with different building blocks have been designed and investigated
Stability calculations indicated that all materials have the layer offset as reported in our earlier
work54 (see Appendix C) We show that the concept of reticular chemistry works as the Density-Of-
States (DOS) of the framework materials vary with the the DOS of the molecules involved in the
frameworks However the stacking does have some influence on the band gap
My contributions to this article include performing all the calculations and preparing the manuscript
17
5 3D Frameworks
51 3D Covalent Organic Frameworks
First 3D COFs were reported in 2007 by the group of Yaghi22 Although the number of 3D COFs
synthesized so far has not been crossed half a dozen they are of particular interest for their very low
mass density Similar to 2D COFs condensation of boronic acids and hydroxyphenyl compounds led
to the formation of COF-102 103 105 108 and 202 Usage of tetragonal boronic acids leads to the
formation of 3D networks (Figure 12) All of them possessed ctn topology while COF-108 which has
the same material composition as COF-105 crystallized in bor topology COF-300 which was formed
from tetragonal and linear building units possessed diamond topology and was five-fold
interpenetrated55 COF-108 has the lowest mass density of all materials known today Unlike most of
the MOFs the connectors in COFs are not metal oxide clusters but covalently bound molecular
moieties In the synthesized COFs the centers of the tetragonal blocks were either sp3 carbon or
silicon atoms
Schmid et al56 have analyzed the two different topologies and developed force field parameters for
COFs The mechanical stability of COFs was also reported However no further study that details the
inherent energetic stability and properties of COFs was found in the literature Using DFTB we
performed a collective study of all 3D COFs in their known topologies with C and Si centers
Figure 12 Schematic diagram of tetragonal and triangular building blocks forming lsquoctnrsquo or lsquoborrsquo
topologies
Our studies of3D COFs have been prepared for publication
Article 5 Stability and electronic properties of 3D covalent organic frameworks
Binit Lukose Agnieszka Kuc Thomas Heine
18
See Appendix E for the article
My contributions to this article include performing all the calculations and preparing the manuscript
We discussed the energetic and mechanical stability as well as the electronic properties of COFs in
the article All studied 3D COFs are energetically stable materials with formation energies of 76 ndash
403 kJ mol-1 The bulk modulus ranges from 29 to 206 GPa Electronically all COFs are
semiconductors with band gaps vary with the number of sp2 carbon atoms present in the linkers
similar to 3D MOFs
52 Porous Aromatic Frameworks
Porous Aromatic Frameworks (PAFs) are purely organic structures where linkers are connected by sp3
carbon atoms (see Appendix A) Tetragonal units were coupled together to form diamond-like
networks The synthesis follows typically a Yamamoto-type Ullmann cross-coupling reaction those
reactions are known to be much simpler to be carried out than the condensation reactions necessary
to form COFs However as the coupling reactions are irreversible a lower degree of crystallinity is
achieved and the materials formed were amorphous The first PAF was reported in 2009 and
showed a very high BET surface area of 5600 m2g-1 The geometry of PAF-1 was equal to diamond
with the C-C bonds replaced by biphenyl Subsequent PAFs may be synthesized with longer linkers
between carbon tetragonal nodes Nevertheless synthesis of a PAF with quaterphenyl linker
provided an amorphous material of very low surface area due to the short range order
Synthetic complexities aside the diamond-like PAFs are interesting for their special geometries from
the viewpoint of the theorist It is interesting to see to what extent they follow the properties of
diamond Additionally the large surface area and high polarizability of aromatic rings are auxiliary for
enhanced hydrogen storage Thus we have explored the possibilities of diamond topology by
inserting various organic linkers in place of C-C bonds (Figure 13)
Figure 13 Diamond structure and various organic linkers to build up PAFs
Our studies of PAFs have been prepared for publication
19
Article 6 Structure electronic structure and hydrogen adsorption capacity of porous aromatic
frameworks
Binit Lukose Mohammad Wahiduzzaman Agnieszka Kuc Thomas Heine
See Appendix F for the article
In this article we have discussed the correlations of properties with the structures Exothermic
formation energies of PAFs calculated from the dehydrogenation of the linkers with CH4 indicate the
strong coordination of linkers in the diamond topology The studied PAFs exhibited bulk moduli much
smaller than diamond however in line with related MOFs and COFs The PAFs are semi-conductors
with their band gaps decrease with the increasing number of benzene rings in the linkers
Using Quantized Liquid Density Functional Theory (QLDFT) for hydrogen57 a method to compute
hydrogen adsorption that takes into account inter-particle interactions and quantum effects we
predicted a large choice of hydrogen uptake capacities in PAFs At room temperature and 100 bar
the predicted uptake in PAF-qtph was 732 wt which exceeds the 2015 DOE (US) target We
further discussed the structural impacts on the adsorption capacities
My contributions to this article include designing the materials performing calculations of stability
and electronic properties describing the adsorption capacities and preparing the manuscript
20
6 New Building Concepts
61 Isoreticular Series of SURMOFs
The growth of MOFs on substrates using liquid phase epitaxy (LPE) opened up a controlled way to
construct frameworks This is achieved by alternate immersion of the substrates in metal and ligand
precursors for several cycles Such MOFs named as SURMOFs can avoid interpenetration because
the degeneracy is lifted58 and are suited for conventional applications This is an advantage as
solvothermally synthesized MOFs with larger pore size suffer from interpenetration and the large
pores are hence not accessible for guest species
MOF-259 is a simple 2D architecture based on paddlewheel units formed by attaching four
dicarboxylic groups to Cu2+ or Zn2+ dimers yielding planar sheets with 4-fold symmetry The
arrangement of MOF layers possessed P2 or C2 symmetry Recently the group of C Woumlll at KIT has
synthesized an isoreticular series of SURMOFs derived from MOF-2 which has a homogeneous series
of linkers with different lengths (Figure 14) XRD comparisons suggest that these MOFs possess P4
symmetry The cell dimensions that correspond to the length of the pore walls range from 1 to 28
nm The diagonal pore-width of one of the SURMOFs (PPDC) accounts to 4 nm The symmetry of
SURMOFs as determined to be different from bulk MOF-2 should be understood by means of theory
As collaborators we simulated the structures and inspected each stacking corresponding to the
symmetries in order to understand the difference
Figure 14 The building units and schematic diagram of the formation of iso-reticular SURMOF
series
21
This collaborated work has been submitted for publication
Article 7 A novel series of isoreticular metal organic frameworks realizing metastable structures
by liquid phase epitaxy
Jinxuan Liu Binit Lukose Osama Shekhah Hasan Kemal Arslan Peter Weidler Hartmut Gliemann
Stefan Braumlse Sylvain Grosjean Adelheid Godt Xinliang Feng Klaus Muumlllen Ioan-Bogdan Magdau
Thomas Heine Christof Woumlll
See Appendix G for the article
The main contribution of this article was the experimental proof backed up by our computer
simulations that SURMOFs with exceptionally large pore sizes of more than 4 nm can be prepared in
the laboratory and they offer the possibility to host large materials such as clusters nanoparticles or
small proteins The most important contribution of theory was to show that while MOF-2 in P2
symmetry shows the highest stability the P4 symmetry as obtained using LPE for SURMOF-2
corresponds to a local minimum
My contribution to this article includes performing and analyzing the calculations Our theoretical
study went significantly beyond and will be published as separate article (Appendix H)
62 Metastability of SURMOFs
Following the difference in stability of SURMOFs-2 in comparison with MOF-2 we identified the role
of linkers as a constituent of the framework that provides the metastability to SURMOFs-2 (Figure
15) In MOFs linkers are essential in determining the topology as well as providing porosity Linkers
typically contain single or multiple aromatic rings and are ditopic and polytopic The orientation of
them normally undergoes lowest repulsion from the coordinated metal-oxide clusters and provides
high stability In the case of 2D MOFs non-covalent interlayer interactions lead to the most stable
arrangements Paddlewheels are identified as the reasons for the stability of bulk MOF-2 as they
form metal-oxide bridges across the MOF layers In the case of SURMOFs-2 the linkers are rotated in
a tightly-packed environment of layers and each linker in bulk MOF-2 is rotated in such a way that
any attempt of shifting is obstructed by repulsion between the neighboring linkers The energy
barrier for leaving P4 increases with the length of organic linkers Moreover SURMOF-2 derivatives
with extremely large linkers are energetically stable due to the increased London dispersion
interaction between the layers in formula units Thus we encountered a rare situation in which the
linkers guarantee the persistence of a series of materials in an otherwise unachievable state
22
Figure 15 Energy diagram of the metastable P4 and stable P2 structures
Our results on the linker guided stability of SUMORs-2 have been prepared for publication
Article 8 Linker guided metastability in templated Metal-Organic Framework-2 derivatives
(SURMOFs-2)
Binit Lukose Gabriel Merino Christof Woumlll Thomas Heine
See Appendix H for the article
This article is based solely on my scientific contributions
23
7 Summary
Nanotechnology is the way of ingeniously controlling the building of small and large structures with
intricate properties it is the way of the future a way of precise controlled building with incidentally
environmental benignness built in by design - Roald Hoffmann Chemistry Nobel Laureate 1981
Currently it is possible to design new materials rather than discovering them by serendipity The
design and control of materials at the nanoscale requires precise understanding of the molecular
interactions processes and phenomena In the next level the characteristics and functionalities of
the materials which are inherent to the material composition and structure need to be studied The
understanding of the materials properties may be put into the design of new materials
Computational tools to a large extend provide insights into the structures and properties of the
materials They also help to convert primary insights into new designs and carry out stability analysis
Our theoretical studies of a variety of materials have provided some insights on their underlying
structures and properties The primary differences in the material compositions and skeletons
attributed a certain choice in properties The contents of the articles discussed in the thesis may be
summarized into the following three parts
71 Validation of Methods
Simulations of nanoporous materials typically include electronic structure calculations that describe
and predict properties of the materials Density functional theory (DFT)30 is the standard and widely-
used tool for the investigation of the electronic structure of solids and molecules Even the optical
properties can be studied through the time-dependent generalization of DFT MOFs and COFs have
several hundreds of atoms in their unit cells DFT becomes inappropriate for such large periodic
systems because of its necessity of immense computational time and power Molecular force field
calculations60 on the other hand lack transferable parameterization especially for transition metal
sites and are hence of limited use to cover the large number of materials to be studied Apparently
a non-orthogonal tight-binding approximation to DFT called density functional tight-binding
(DFTB)28323561 has proven to be computationally feasible and sufficiently accurate This method
computes parameters from DFT calculations of a few molecules per pair of atom types The
parameters are generally transferable to other molecules or solids With self-consistent charge (SCC)
extension DFTB has improved accuracy In order to account weak forces the London dispersion
energy can be calculated separately using empirical potentials and added to total energy Successful
realizations of DFTB include the studies of large-scale systems such as biomolecules62
24
supramolecular compounds63 surfaces64 solids65 liquids and alloys66 Being an approximate method
DFTB needs validation Often one compares DFTB results of selected reference systems with those
obtained with DFT
Before electronic structure calculations of framework materials can be carried out it is necessary to
compute the equilibrium configurations of the atoms Geometry optimization (or energy
minimization) employs a mathematical procedure of optimization to move atoms so as to reduce the
net forces on them to negligible values We adopted the conjugate gradient scheme for the
optimizations using DFTB A primary test for the validation of these optimizations is the comparison
of cell parameters bond lengths bond angles and dihedral angles with the corresponding known
numbers We compared the DFTB optimized MOF COF and PAF geometries with the experimentally
determined or DFT optimized geometries and found that the values agree within 6 error
The angles and intensities of X-ray diffraction patterns can produce three-dimensional information of
the density of electrons within a crystal This can provide a complete picture of atomic positions
chemical bonds and disorders if any Hence simulating powder X-ray diffraction (PXRD) patterns of
optimized geometries and comparing them with experimental patterns minimize errors in the crystal
model Our focus on this matter directed us to identify the layer-offsets of 2D COFs for the first time
In the case of 3D COFs excellent correlations were generally observed between experimental and
simulated patterns Slight differences in the intensities of some of them were due to the presence of
solvents in the crystals as they were reported in the experimental articles PAFs as experimentally
being amorphous do not possess XRD comparisons MOFs within DFTB optimization have
undergone some changes especially in the dihedral angles in comparison with experimental
suggestion or DFT optimization This was verified from the differences in the simulated PXRD
patterns Another interesting outcome of XRD comparison was the discovery of P4 symmetry of
templated MOF-2 derivatives (SURMOFs-2) made by Woumlll et al
Bulk moduli which may be expressed as the second derivative of energy with respect to the unit cell
volume can give a sense of mechanical stability Our calculations provide the following bulk moduli
for some materials MOF-5 1534 (1537)67 Cu-BTC 3466 (3517)68 COF-102 206 (170)69 COF-
103 139 (118)69 COF-105 79 (57)69 COF-108 37 (29)69 COF-202 153 (110)69 The data in the
parenthesis give corresponding values found in the literature calculated using force-field methods
The bulk moduli of MOFs are comparable with the results in the literature however COFs show
significant differences Albeit the differences in values each type of calculation shows the trend that
bulk modulus decreases with decreasing mas density increasing pore volume decreasing thickness
of pore walls and increasing distance between connection nodes
25
Formation of framework materials from condensation of reactants may be energetically modeled
COFs are formed from dehydration reactions of boronic acids and hydroxyphenyl compounds The
formation energy calculated from the energies of the products and reactants can indicate energetic
stability Both DFTB for periodic structures and DFT for finite structures predict exothermic formation
of 3D COFs Likewise for 2D COFs such as COF-1 and 5 the formation of monolayers is found to be
endothermic within both the periodic model calculation using DFTB and finite model calculation
using DFT The stacking of layers provides them stability
72 Weak Interactions in 2D Materials
AB stacked natural graphite and ABC stacked rhombohedral graphite are found to be in proportions
of 85 and 15 in nature respectively whereas AA stacking has been observed only in graphite
intercalation compounds On the other hand boron nitride (h-BN) the synthetic product of boric
acid and boron trioxide exists with AA stacking 2D COFs were believed to have high symmetric AA
(eclipsed ndash P6mmm symmetry) stacking as boron nitride (h-BN) An exception was COF-1 which was
considered to have AB (staggered ndash P63mmc) stacking as graphite The AA stacking maximizes the
attractive London dispersion interaction between the layers a dominating term of the stacking
energy At the same time AA stacking always suffers repulsive Coulomb force between the layers
due to the polarized connectors It should be noted that in boron nitride oppositely charged boron
atoms and nitrogen atoms are facing each other The Coulomb interaction has ruled out possible
interlayer eclipse between atoms of alike charges This gave us the notion that 2D COFs cannot
possess layer-eclipsing Based on these facts we have suggested a shift of ~14 Aring between adjacent
layers at which one or two of the edge atoms of the hexagonal rings are situated perpendicular to
the center of adjacent rings In other words some or all of the hexagonal rings in the pore walls
undergo staggering with that of adjacent layers These lattice types were found to be very stable
compared to AA or AB analogues AA stacking was found to have the highest Coulombic repulsion in
each COF Within an error limit the bulk COFs may be either serrated or inclined The interlayer
separations of all the COF stacks are in the range of 30 to 36 Aring and increase in the order AB
serratedinclined AA70 The motivation for proposing inclined stacking for COFs is that the
rhombohedral graphite (ABC stacking) is the inclined version of the layer-offset in natural graphite
(AB stacking) The instability of AA stacking was also suggested by the studies of elastic properties of
COF-1 and COF-5 by Zhou et al52 The shear modulus show negative signs for the vertical alignment of
COF layers while they are small but positive for the offset of layers
The change of stacking should be visible in their PXRD patterns because each space group has a
distinct set of symmetry imposed reflection conditions We simulated the XRD patterns of COFs in
their known and new configurations and on comparison with the experimental spectrum the new as
26
well as AA stacking showed good agreement5470 The slight layer-offsets are only able to make a few
additional peaks in the vicinity of existing peaks and some changes in relative intensities The
relatively broad experimental data covers nearly all the peaks of the simulated spectrum In other
words the broad experimental peaks are explainable with layer-offset
A detailed analysis on the interlayer stacking of HHTP-DPB COF made by Dichtel et al53 is very
complementary53 This work uses molecular mechanics extensively to define potential energy
surfaces of stacking of two layers from eclipsed to staggered covering all possible layer offsets Low
energy region was near to the eclipsed stacking which was then zoomed in to examine using DFT for
higher accuracy The far-eclipsed regions encounter no energy barriers with the near-eclipsed regions
which suggest the unlikeness of forming near-staggered layering Within DFT the eclipsed
geometries are at high energy compared to the slight offsets around AA (~ 13 ndash 25 Aring away) For a
hexagonal COF the energetically preferred offset could be in one of the six hexagonal sectors (or
circle sectors) of central angle 60 The preference of falling into one sector over the others is
arbitrary and may be determined by symmetry set by nature or external parameters Hence the
layering order in bulk could be serrated inclined spiral or purely by random The latter case better
known as turbostratic disorder is then more like a rule for 2D COFs rather than an exception caused
by external forces This also explains why the experimental PXRD patters have relatively broad peaks
The layer-offset not only change the internal pore structure but also affect interlayer exciton and
vertical charge transport in opto-electronic applications
About stacking stability the square COFs are expected not to be different from hexagonal COFs
because the local environment causing the shifts is nearly the same The DFTB based calculations
reported along with the synthesis of electron-transporting 2D-NiPc-BTDA COF supports this notion71
Two shifted configurations ndash at 07 and 14 Aring away from AA ndash appeared to be energetically preferred
over the AA stacking It appears that AA stacking is only possible for boron nitride-like structures
were adjacent layers have atoms with opposite charges in vertical direction
SURMOFs-2 a series of paddlewheel-based 2D MOFs possess a different symmetry than
solvothermally synthesized MOF-2 due to the differences in interlayer interactions Typically the
interaction between the paddlewheels favors P2 or C2 symmetry of bulk MOF-2 rather than the P4
symmetry of SURMOFs-2 Bader charge analysis reveals that electron distribution between the Cu-
paddlewheels is the lowest when they follow P4 symmetry This indicates the smallest possibility of
having a direct Cu-Cu bond between the paddlewheels In monolayer organic linkers undergo no
rotation with respect to metal dimers
27
X-ray diffraction of SURMOFs-2 made by Woumlll et al indicated P4 symmetry and a relatively small
interlayer separation This increases the repulsion between the linkers and enforces them to rotate
The rotations of each linker in a layer are controlled globally The rotated linkers in adjacent layers
increase London dispersion however a paddlewheel-led shift towards any side increases repulsion
thereby locking SURMOF-2 in P4 packing Hence with the special arrangement of rotated linkers the
linker-linker interaction overcomes the paddlewheel-paddlewheel interaction
P4 symmetry opens up the pores more clearly than P2 or C2 Additionally the metal sites that
typically host solvents are inaccessible Furthermore improvements such as functionalizing the linker
may be easily carried out
Based on our calculations on 2D materials we emphasize the role of van der Waals interactions in
determining the layer-to-layer arrangements The promise of reticular chemistry which is the
maintainability of structural integrity of the building blocks in the construction process is partly
broken in the case of SURMOFs-2 This is because due to repulsion the linkers undergo rotation with
respect to the carboxylic parts albeit keeping the topology
73 Structure-Property Relationships
We have studied a variety of materials from the classes of MOFs COFs and PAFs The structural
differences arise from the differences in the constituents andor their arrangements Properties in
general are interlinked with structural specifications Therefore it is beneficial to know the
relationship between the structural parameters and properties
The mass density is an intensive property of a material In the area of nanoporous materials a crystal
with low mass density has advantages in applications involving transport Definitely the mass density
decreases with increasing pore volume Still the number of atoms in the wall and their weights are
important factors The pore size does not relate directly to the atom counts The volume per atom
(inverse of atom density) another intensive property of a material obliquely gives porosity Figure
16 shows the variation of mass density with volume per atom (including the volume of the atom) for
MOFs COFs and PAFs MOFs show higher mass density compared to other materials of identical
atom density implying the presence of heavy elements It is to be noted that Cu-BTC has mass
density 184 times higher than COF-102 The presence of metals in the form of oxide clusters in MOFs
increases the mass density and decreases the volume per atom Note that the low-weighted MOF in
the discussion (MOF-205) is heavier than many of the PAFs and COFs The relatively high mass
density and small volume per atom of COF-300 is due to its 5-fold interpenetration whereas COF-202
has additional tert-butyl groups which do not contribute to the system shape but affect the mass
density and the volume per atom COF-102 and 103 have same topology but different centers (C and
28
Si respectively) The Si centers increases the cell volume and decreases mass density COF-105 (Si
centers) and COF-108 (C centers) additionally differ in their topologies (ctn and bor respectively) It
appears that bor topology expands the material compared to ctn PAFs hold the extreme cases PAF-
phnl has the highest atom and mass densities whereas PAF-qtph has the smallest atom and mass
densities
Figure 16 Mass density as a function of volume per atom for MOFs COFs and PAFs
The bulk modulus indicates the materialsrsquo resistance towards compression which should in principle
decrease with increasing porosity At the same time larger number of atoms making covalent
networks in unit volume should supply larger bulk moduli Thus differences in molecular contents
and architectures give rise to different bulk moduli It is interesting to see how the mechanical
stability of nanoporous materials is related with the atom density We have obtained a correlation
between bulk modulus (B) and volume per atom (VA) in the case of the studied MOFs COFs and PAFs
as follows
29
where k is a constant (see Figure 17) The inverse-volume correlation indicates that the materials
close to the fitting curve have average bond strengths (interaction energy between close atoms)
identical to each other independent of number of bonds bond order and branching Only Cu-BTC
COF-202 and COF-300 show significant differences from the fitted curve Cu-BTC has 17 times larger
bulk modulus compared to COF-102 of similar volume per atom which implies the substantially
higher strength of the bond network resulting from paddlewheel units and tbo topology
Interpenetration decreased the volume per atom however increased bulk modulus through
interlayer interactions in the case of COF-300 The tert-butyl groups in COF-202 do not transmit its
inherent stability to the COF significantly however decreases the volume per atom Comparison
between COF-102 (C centers) and COF-103 (Si centers) discloses that Si centers decrease the
mechanical stability Comparison between COF-105 (ctn) and COF-108 (bor) suggests that ctn
topology possess higher stability This indicates that local angular preferences can amend the
strength of the bulk material
Figure 17 Bulk modulus as a function of volume per atom for MOFs COFs and PAFs
Band gaps of all the studied materials show that they are semiconductors except of Cu-BTC which
has unpaired electrons at the Cu centers Our studies of the density of states (DOS) of bulk MOFs and
the cluster models that have finite numbers of connectors and linkers show that electronic structure
30
stays unchanged when going from cluster to periodic crystal In the case of 2D COFs the DOS of
monolayers and bulk materials were identical Interpenetrated COF-300 showed no difference in the
electronic structure in comparison with the non-interpenetrated structure Based on these results
we may reach into a premature conclusion that electronic structure of a solid is determined by its
constituent bonded network sufficiently large to include all its building units
HOMO-LUMO gap of the building units determine the band gap of a framework material We have
observed that PAFs nearly reproduced the HOMO-LUMO gaps of the linkers 3D COFs that are made
of more than one building unit show that the band gap is slightly smaller than the smallest of the
HOMO-LUMO gaps of the building units For example
TBPM (43 eV) + HHTP (34 eV) COF-105COF-108 (32 eV)
TBPM (43 eV) + TBST (139 eV) COF-202 (42 eV)
TAM (41 eV) + TA (26 eV) COF-300 (23 eV)
The compound names are taken from appendix E Additionally the band gaps decrease with
increasing number of conjugated rings similar to HOMO-LUMO gaps of the linkers
I believe that the studies in the thesis have helped to an extent to understand the structure
stability and properties of different classes of framework materials The benchmark structures we
studied have the essential features of the classes they represent Ab-initio based computational
studies of several periodic structures are exceptional and thus have its place in the literature
31
List of Abbreviations
ADF Amsterdam Density Functional code
BLYP Becke-Lee-Yang-Parr functional
B3LYP Becke 3-parameter Lee Yang and Parr functional
COF Covalent-Organic Framework
CP Coordination Polymer
CTF Covalent-Triazine Framework
DC Dispersion correction
DFT Density Functional Theory
DFTB Density Functional Tight-Binding
DOS Density of States
DOE (US) Department of Energy (United States)
DZP Double-Zeta Polarized basis set
GGA Generalized Gradient Approximation
LCAO Linear Combination of Atomic Orbitals
LPE Liquid Phase Epitaxy
MOF Metal-Organic Framework
PAF Porous Aromatic Framework
PBE Perdew-Burke-Ernzerhof functional
PXRD Powder X-ray Diffraction Pattern
QLDFT Quantized Liquid Density Functional Theory
RCSR Reticular Chemistry Structure Resource
SBU Secondary Building Unit
SCC Self-Consistent Charge
TZP Triple-Zeta Polarized basis set
SURMOF Surface-Metal-Organic Framework
32
List of Figures
Figure 1 ldquoAdamantinerdquo unit of diamond structure and diamond (dia) net 3
Figure 2 CU-BTC MOF and the corresponding tbo net 4
Figure 3 Fundamental building blocks and SBUs of Cu-BTC and extension of SBUs following crystal
structure 5
Figure 4 Schematic diagram of the single pore of MOF and An example of a simple cubic MOF in
which zinc-oxide tetrahedron are linked together by benzene linkers Atom colors blue ndash Zn red ndash O
grey ndash C white ndash H 6
Figure 5 Schematic diagram of interpenetrated and interwoven frameworks 7
Figure 6 Schematic diagram of SURMOF 8
Figure 7 SBUs and topologies of 2D COFs 9
Figure 8 Deconstructed building units their schematic representations and final geometries of
HKUST-1 (Cu-BTC) MOF-5 MOF-177 MOF-205 and MIL-53 11
Figure 9 AA and AB layer stacks of hexagonal layers 13
Figure 10 Stacking energy Es vs shift of adjacent layers for COF-5 Different stacking possibilities and
their energies are also shown 14
Figure 11 Schematic diagram of different building units forming 2D COFs 16
Figure 12 Schematic diagram of tetragonal and triangular building blocks forming lsquoctnrsquo or lsquoborrsquo
topologies 17
Figure 13 Diamond structure and various organic linkers to build up PAFs 18
Figure 14 The building units and schematic diagram of the formation of iso-reticular SURMOF series
20
Figure 15 Energy diagram of the metastable P4 and stable P2 structures 22
Figure 16 Mass density as a function of volume per atom for MOFs COFs and PAFs 28
Figure 17 Bulk modulus as a function of volume per atom for MOFs COFs and PAFs 29
33
References
(1) Morris R E Wheatley P S Angewandte Chemie-International Edition 2008 47 4966
(2) Li J-R Kuppler R J Zhou H-C Chemical Society Reviews 2009 38 1477
(3) Corma A Chemical Reviews 1997 97 2373
(4) Seo J S Whang D Lee H Jun S I Oh J Jeon Y J Kim K Nature 2000 404 982
(5) Lee J Kim J Hyeon T Advanced Materials 2006 18 2073
(6) Stein A Wang Z Fierke M A Advanced Materials 2009 21 265
(7) Velev O D Jede T A Lobo R F Lenhoff A M Nature 1997 389 447
(8) Beck J S Vartuli J C Roth W J Leonowicz M E Kresge C T Schmitt K D Chu C T
W Olson D H Sheppard E W McCullen S B Higgins J B Schlenker J L Journal of the
American Chemical Society 1992 114 10834
(9) Kresge C T Leonowicz M E Roth W J Vartuli J C Beck J S Nature 1992 359 710
(10) Ying J Y Mehnert C P Wong M S Angewandte Chemie-International Edition 1999 38
56
(11) Velev O D Kaler E W Advanced Materials 2000 12 531
(12) Stein A Microporous and Mesoporous Materials 2001 44 227
(13) Tanev P T Pinnavaia T J Science 1996 271 1267
(14) Yaghi O M OKeeffe M Ockwig N W Chae H K Eddaoudi M Kim J Nature 2003
423 705
(15) OKeeffe M Peskov M A Ramsden S J Yaghi O M Accounts of Chemical Research
2008 41 1782
(16) Delgado-Friedrichs O OKeeffe M Journal of Solid State Chemistry 2005 178 2480
(17) Delgado-Friedrichs O Foster M D OKeeffe M Proserpio D M Treacy M M J Yaghi
O M Journal of Solid State Chemistry 2005 178 2533
(18) OKeeffe M Yaghi O M Chemical Reviews 2012 112 675
(19) Tranchemontagne D J L Ni Z OKeeffe M Yaghi O M Angewandte Chemie-
International Edition 2008 47 5136
(20) Hayashi H Cote A P Furukawa H OKeeffe M Yaghi O M Nature Materials 2007 6
501
(21) Cote A P Benin A I Ockwig N W OKeeffe M Matzger A J Yaghi O M Science
2005 310 1166
(22) El-Kaderi H M Hunt J R Mendoza-Cortes J L Cote A P Taylor R E OKeeffe M
Yaghi O M Science 2007 316 268
(23) Rowsell J L C Yaghi O M Microporous and Mesoporous Materials 2004 73 3
(24) Hermes S Zacher D Baunemann A Woell C Fischer R A Chemistry of Materials
2007 19 2168
34
(25) Kuhn P Antonietti M Thomas A Angewandte Chemie-International Edition 2008 47
3450
(26) Ben T Ren H Ma S Cao D Lan J Jing X Wang W Xu J Deng F Simmons J M
Qiu S Zhu G Angewandte Chemie-International Edition 2009 48 9457
(27) Porezag D Frauenheim T Kohler T Seifert G Kaschner R Physical Review B 1995
51 12947
(28) Seifert G Porezag D Frauenheim T International Journal of Quantum Chemistry 1996
58 185
(29) Kohn W Sham L J Physical Review 1965 140 1133
(30) Parr R G Yang W Density-Functional Theory of Atoms and Molecules New York Oxford
University Press 1989
(31) Hohenberg P Kohn W Physical Review B 1964 136 B864
(32) Elstner M Porezag D Jungnickel G Elsner J Haugk M Frauenheim T Suhai S
Seifert G Physical Review B 1998 58 7260
(33) Zhechkov L Heine T Patchkovskii S Seifert G Duarte H A Journal of Chemical
Theory and Computation 2005 1 841
(34) Elstner M Hobza P Frauenheim T Suhai S Kaxiras E Journal of Chemical Physics
2001 114 5149
(35) Oliveira A F Seifert G Heine T Duarte H A Journal of the Brazilian Chemical Society
2009 20 1193
(36) Seifert G Joswig J-O Wiley Interdisciplinary Reviews-Computational Molecular Science
2012 2 456
(37) Heine T Rapacioli M Patchkovskii S Frenzel J Koester A M Calaminici P
Escalante S Duarte H A Flores R Geudtner G Goursot A Reveles J U Vela A Salahub D
R deMon deMon-nano edn deMon-nano 2009
(38) BCCMS B DFTB+ - Density Functional based Tight binding (and more)
(39) Monkhorst H J Pack J D Physical Review B 1976 13 5188
(40) SCM Amsterdam Density Functional 2012
(41) University of Karlsruhe and Forschungszentrum Karlsruhe gGmbH - TURBOMOLE V63
2011 2007
(42) Dovesi R Saunders V R Roetti C Orlando R Zicovich-Wilson C M Pascale F
Civalleri B Doll K Harrison N M Bush I J DrsquoArco P Llunell M CRYSTAL09 Users Manual
University of Torino Torino 2009 2009
(43) Wadt W R Hay P J Journal of Chemical Physics 1985 82 284
(44) Roy L E Hay P J Martin R L Journal of Chemical Theory and Computation 2008 4
1029
35
(45) Ehlers A W Bohme M Dapprich S Gobbi A Hollwarth A Jonas V Kohler K F
Stegmann R Veldkamp A Frenking G Chemical Physics Letters 1993 208 111
(46) Garberoglio G Skoulidas A I Johnson J K Journal of Physical Chemistry B 2005 109
13094
(47) Han S S Mendoza-Cortes J L Goddard W A III Chemical Society Reviews 2009 38
1460
(48) Getman R B Bae Y-S Wilmer C E Snurr R Q Chemical Reviews 2012 112 703
(49) Cote A P El-Kaderi H M Furukawa H Hunt J R Yaghi O M Journal of the American
Chemical Society 2007 129 12914
(50) Wan S Guo J Kim J Ihee H Jiang D Angewandte Chemie-International Edition 2008
47 8826
(51) Wan S Guo J Kim J Ihee H Jiang D Angewandte Chemie-International Edition 2009
48 5439
(52) Zhou W Wu H Yildirim T Chemical Physics Letters 2010 499 103
(53) Spitler E L Koo B T Novotney J L Colson J W Uribe-Romo F J Gutierrez G D
Clancy P Dichtel W R Journal of the American Chemical Society 2011 133 19416
(54) Lukose B Kuc A Heine T Chemistry-a European Journal 2011 17 2388
(55) Uribe-Romo F J Hunt J R Furukawa H Klock C OKeeffe M Yaghi O M Journal of
the American Chemical Society 2009 131 4570
(56) Schmid R Tafipolsky M Journal of the American Chemical Society 2008 130 12600
(57) Patchkovskii S Heine T Physical Review E 2009 80
(58) Shekhah O Wang H Paradinas M Ocal C Schuepbach B Terfort A Zacher D
Fischer R A Woell C Nature Materials 2009 8 481
(59) Li H Eddaoudi M Groy T L Yaghi O M Journal of the American Chemical Society
1998 120 8571
(60) Rappe A K Casewit C J Colwell K S Goddard W A Skiff W M Journal of the
American Chemical Society 1992 114 10024
(61) Frauenheim T Seifert G Elstner M Hajnal Z Jungnickel G Porezag D Suhai S
Scholz R Physica Status Solidi B-Basic Research 2000 217 41
(62) Elstner M Cui Q Munih P Kaxiras E Frauenheim T Karplus M Journal of
Computational Chemistry 2003 24 565
(63) Heine T Dos Santos H F Patchkovskii S Duarte H A Journal of Physical Chemistry A
2007 111 5648
(64) Sternberg M Zapol P Curtiss L A Molecular Physics 2005 103 1017
(65) Zhang C Zhang Z Wang S Li H Dong J Xing N Guo Y Li W Solid State
Communications 2007 142 477
36
(66) Munch W Kreuer K D Silvestri W Maier J Seifert G Solid State Ionics 2001 145
437
(67) Bahr D F Reid J A Mook W M Bauer C A Stumpf R Skulan A J Moody N R
Simmons B A Shindel M M Allendorf M D Physical Review B 2007 76
(68) Amirjalayer S Tafipolsky M Schmid R Journal of Physical Chemistry C 2011 115
15133
(69) Amirjalayer S Snurr R Q Schmid R Journal of Physical Chemistry C 2012 116 4921
(70) Lukose B Kuc A Frenzel J Heine T Beilstein Journal of Nanotechnology 2010 1 60
(71) Ding X Chen L Honsho Y Feng X Saenpawang O Guo J Saeki A Seki S Irle S
Nagase S Parasuk V Jiang D Journal of the American Chemical Society 2011 133 14510
37
Appendix A
Review Covalently-bound organic frameworks
Binit Lukose and Thomas Heine
To be submitted for publication after revision
Contents
1 Introduction
2 Synthetic achievements
21 Covalent Organic Frameoworks (COFs)
22 Covalent-Triazine Frameworks (CTFs)
23 Porous Aromatic Frameworks (PAFs)
24 Schemes for synthesis
25 List of materials
3 Studies of the underlying structure and properties of COFs
4 Applications
41 Gas storage
411 Porosity of COFs
412 Experimental measurements
413 Theoretical preidctions
414 Adsorption sites
415 Hydrogen storage by spillover
42 Diffusion and selectivity
43 Suggestions for improvement
431 Geometry modifications
432 Metal doping
433 Functionalization
5 Conclusions
6 List and pictures of chemical compounds
38
1 Introduction
Nanoporous materials have perfectly ordered voids to accommodate to interact with and to
discriminate molecules leading to prominent applications such as gas storage separation and sieving
catalysis filtration and sensoring1-4 They are defined as porous materials with pore diameters less
than 100 nm and are classified as micropores of less than 2 nm in diameter mesopores between 2
and 50 nm and macropores of greater than 50 nm The internal geometry that determines pore size
and surface area can be precisely engineered at molecular scales Reticular synthetic methods
suggested by Yaghi and co-workers5-7 can accomplish pre-designed frameworks The procedure is to
select rigid molecular building blocks prudently and assemble them into destined networks using
strong bonds
Several types of materials have been synthesized using reticular chemistry concepts One prominent
group which is much appreciated for its credentials is the Metal-Organic Frameworks (MOFs)8-13 in
which metal-oxide connectors and organic linkers are judiciously selected to form a large variety of
frameworks The immense potential of MOFs is facilitated by the fact that all building blocks are
inexpensive chemicals and that the synthesis can be carried out solvothermally The progress in MOF
synthesis has reached the point that some of the MOFs are commercially available Several MOFs of
ultrahigh porosity have also been successfully synthesized notably the non-interpenetrated IRMOF-
74-XI which has a pore size of 98 Aring Scientists are also able to synthesize MOFs from cheap edible
natural products14 Since the discovery of MOFs many other crystalline frameworks have been
synthesized using reticular chemistry such as Metal-Organic Polyhedra (MOP)15-19 Zeolite
Imidazolate Frameworks (ZIFs)20-22 and Covalent Organic Frameworks (COFs)23-29
COFs are composed of covalent building blocks made of boron carbon oxygen and hydrogen in
many cases also including nitrogen or silicon stitched together by organic subunits The atoms are
held together by strong covalent bonds Depending on the selection of building blocks the COFs may
form in two (2D) or three (3D) dimensions Planar building blocks are the constituents of 2D COFs
whereas for the formation of 3D COFs typically tetragonal building blocks are involved High
symmetric covalent linking as it is perceived in reticular chemistry was confirmed for the end
products5
Unlike the case of supramolecular assemblies the absence of noncovalent forces between the
molecular building units endorses exceptional rigidity and stability for COFs They are in general
thermally stable above 4000 C Also in COFs the stitching of molecular building units guarantees an
39
increased order and allows control over porosity and composition Without any metals or other
heavy atoms present they exhibit low mass densities This gives them advantages over MOFs in
various applications for example higher gravimetric capacities for gas storage3031 The lowest
density crystal known until today is COF-10824 In addition COFs exhibit permanent porosity with
specific surface areas that are comparable to MOFs and hence larger than in zeolites and porous
silicates
MOF and COF crystals possess long range order although COFs have been achieved so far only at the
μm scale Reversible solvothermal condensation reactions are credited for the high order of
crystallinity Other new framework materials such as Covalent Triazine Frameworks (CTFs) and
Porous Aromatic Frameworks (PAFs) differed in ways of synthesis The former was prepared by
ionothermal polymerization while the latter was produced by coupling reactions PAFs lack long
range order in the crystals as a result of the irreversible synthesis Nevertheless many of the
materials are promisingly good for applications In this review we intend to discuss the synthetic
achievements of COF CTFs and PAFs and studies on their structure properties and prominent
applications
For easy understanding of the atomic geometry of a material the reader is referred to Table 1 which
gives the names of the framework materials the reactants and the synthesis scheme Figure 5 shows
the reactants and Figure 4 shows the reaction schemes Example reactions are given in Figure 1-3
Abbreviations of each chemical compound are given in Section 6
2 Synthetic achievements
21 Covalent Organic Frameworks (COFs)
In the year 2005 Yaghi et al23 achieved the crystallization of covalent organic molecules in the form
of periodic extended layered frameworks The condensation of discrete molecules of different sizes
enabled the synthesis of micro- and mesoporous crystalline COFs27 The selection of molecules with 2
and 3 connection sites for synthesis led to the formation of hexagonal layered geometries32 Yaghi et
al have synthesized half a dozen three-dimensional crystalline COFs solvothermally using tetragonal
building blocks containing 4 connection sites since 2007242829 Figure 1 shows a few examples of 2D
and 3D COFs formed by the self-condensation of boronic acids such as BDBA TBPM and TBPS and co-
condensation of the same boronic acids with HHTP
40
Figure 1 Solvothermal condensation (dehydration) of monomers to form 2D and 3D COFs Atom colors are boron yellow oxygen red yellow (big) silicon grey carbon
Alternate synthetic procedures were also exploited for production and functionalization of COFs
Lavigne et al33 showed that the synthesis of COFs using polyboronate esters is facile and high-yielded
41
Boronate esters often contain multiple catechol moieties which are prone to oxidation and are
insoluble in organic solvents Dichtel et al3435 presented Lewis acid-catalysis procedures to form
boronate esters of catechols and arylboronic acids without subjecting to oxidation Cooper et al36
successfully utilized microwave heating techniques for rapid production (~200 times faster than
solvothermal methods) of both 2D and 3D COFs The syntheses of phthalocyanine3437 or porphyrin38
based square COFs have been reported in literature The latter was noticed for its time-dependent
crystal growth which also affects its pore parameters
Abel et al39 reported the ability to form COFs on metallic surfaces COF surfaces (SCOFs) have been
formed by molecular dehydration of boroxine and triphenylene on a silver surface Albeit with some
defects the materials showed high temperature stability allowing to proceed with functionalization
Lackinger et al40 realized the role of substrates on the formation of surface COFs from the surface-
generated radicals Halogen substituted polyaromatic monomers can be sublimed onto metal
substrates and ultimately turned into a COF after homolysis and intermolecular colligation
Chemically inert graphite surfaces cannot support the homolysis of covalent carbon-halogen bonds
and thus cannot initiate the subsequent association of radicals COF layers can be formed onto
Cu(111)40 and Au(111)41 surfaces but with lower crystal ordering Beton et al41 deposited the
monomers on annealed Au(111) surfaces which supplied surface mobility for the monomers and
subsequently increased porosity and surface coverage of the covalent networks Unlike the bulk form
at the surface the monomers were self-organized onto polygons with 4 to 8 edges Such a template
was able to organize a sublimed layer of C60 fullerenes the number of C60 molecules adsorbed in a
cavity was correlated to the size of the polygon
In 2009 Sassi et al42 studied the problem of crystal growth of COFs on substrates They calculated
the reaction mechanism of 14-benzenediboronic acid (BDBA) on Ag(111) surface self-condensation
of BDBA molecules in solvents leads to the formation of two-dimensional (the planar) stacked COF-1
For the surface COFs the study using Density Functional Theory reveals that there are neither
preferred adsorption sites for the molecules nor a charge transfer between the molecules and the
surface Hence the electronic structure of the molecules remains unchanged and the role of the
metal surface is limited in providing the planar nature for the polymer The free reaction enthalpy
(Gibbs free energy) difference of the reaction ∆G = ∆H ndash T∆S is approximated using the harmonic
approximation taking into account the geometrical restrictions of the metal surface and the entropic
contributions of the released water molecules As result the formation of SCOF-1 has been found to
be exothermic if more than six monomers are involved Ourdjini et al43 reported the polymerization
of 14-benzenediboronic acid (BDBA) on different substrates (Ag(111) Ag(100) Au(111) and Cu(111))
and at different source and substrate temperatures to follow how molecular flux and adsorption-
42
diffusion environments should be controlled for the formation of polymers with the smallest number
of structural defects High flux is essential to avoid H-bonded supramolecular assemblies of
molecules and the substrate temperature needs to be optimized to allow the best surface diffusion
and longest residential time of the reactants Achieving these two contradictory conditions together
is a limitation for some substrates eg for copper Silver was found to be the best substrate for
producing optimum quality polymers Controlling the growth parameters is important since
annealing cannot cure defects such as formation of pentagons heptagons and other non-hexagonal
shapes which involved strong covalent bonds
Ditchel et al44 successfully grew COF films on monolayer graphene supported by a substrate under
operationally simple solvothermal conditions The films have better crystallinity compared to COF
powders and can facilitate photoelectronic applications straightaway Zamora et al45 have achieved
exfoliation of COF layers by sonication The thin layered nanostructures have been isolated under
ambient conditions on surfaces and free-standing on carbon grids
A noted characteristic of COFs is its organic functionality Judiciously selected organic units ndash pyrene
and triphenylene (TP) ndash for the production of TP COF can not only harvest photons over a wide range
but also facilitate energy migration over the crystalline solid25 Polypyrene (PPy)-COF that consists of
a single aromatic entity ndash pyrene- is fluorescent upon the formation of excimers and hence undergo
exciton migration across the stacked layers26 A nickel(II)-phthalocyanine based layered square COF
that incorporates phenyl linkers and forms cofacially stacked H-aggregates was reported to absorb
photons over the solar spectrum and transport charges across the stacked layers37 COF-366 and
COF-66 showed excellent hole mobility of 81 and 30 cm2 V-1s-1 surpassing that of all polycrystalline
polymers known until today46 A first example of an electron-transporting 2D COF was reported
recently47 This square COF was built from the co-condensation of nickel(II) phthalocyanine and
electron-deficient benzothiadiazole With the nearly vertical arrangement of its layers it exhibits an
electron mobility of 06 cm2 V-1s-1 The enhanced absorbance is over a wide range of wavelengths up
to 1000 nm In addition it exhibits panchromatic photoconductivity and high IR sensitivity
Lavigne et al48 achieved the condensation of alkyl (methyl ethyl and propyl) substituted organic
building units with boronic acids forming functionalized COFs The alkyl groups contribute for higher
molar adsorption of H2 however the increased mass density of the functionalized COFs cause for
decreased gravimetric storage capacities Boronate ester-linked COFs are stable in organic solvents
however undergo rapid hydrolysis when exposed in water49 In particular the instability of COF-1
upon H2O adsorption shall be mentioned here50 Lanni et al claimed that alkylation could bring
hydrolytic stability into COFs49
43
Functionalization and pore surface engineering in 2D COFs can be improved if azide appended
building blocks are stitched together in click reactions with alkynes51 Control over the pore surface
is made possible by the use of both azide appended and bare organic building units the ratios of
which is matching with the final amount of functionalization in the pore walls The click reactions of
azide groups with alkynes within the pores lead to the formation of triazole-linked groups on the
pore surfaces This strategy also gives the relief of not condensing the already functionalized building
units Jiang et al have asserted eclipsed layering for most of the final COFs using powder X-ray
diffraction pattern Exceptions are the hexagonal functionalized COFs with relatively high azide-
content (ge 75) Their weak XRD signals suggest possible layer-offset or de-lamellation Although
functionalization on pore surfaces exhibits the nature of lowering the surface area the possibility to
add reactive groups within the pores can be utilized for gas-sorption selectivity The authors have
claimed a 16 fold selectivity of CO2 over N2 using 100 acetyl-rich COF-5 compared to the pure COF-5
The range of the click reaction approach is so wide that relatively large chromophores can be
accommodated in the pores thereby making COF-5 fluorescent
Functionalization of 3D COFs has just reported in 201252 Dichtel et al used a monomer-truncation
strategy where one of the four dihydroxyborylphenyl moieties of a tetrahedral building unit was
replaced by a dodecyl or allyl functional group Condensation of the truncated and the pure
tetrahedral blocks in a certain ratio created a functionalized 3D COF The degree of functionalization
was in proportion with the ratio The crystallinity was maintained except for high loading (gt 37) of
truncated monomers The pore volume and the surface area were decreased as a function of loading
level
A heterogeneous catalytic activity of COFs is achieved with an imine-linked palladium doped COF by
enabling Suzuki-Miyaura coupling reaction within the pores53 The COF-LZU1 has a layered geometry
that has the distance between nitrogen atoms in the neighboring linkers in a level (~37 Aring) sufficient
to accommodate metal ions Palladium was incorporated by a post-treatment of the COF PdCOF-
LZU1 was applied to catalyze Suzuki-Miyaura coupling reaction This coupling reaction is generally
used for the facile formation of C-C bonds54 The increased structural regularity and high insolubility
in water and organic solvents are adequate qualities of imine-linked COFs to perform as catalysts
Experiments with the above COF show a broad scope of the reactants excellent yields of the
products and easy recyclability of the catalyst
The comparatively higher thermal stability of COFs is often noted and is explainable with their strong
covalent bonds The reversible dehydrations for the formation of most of the COFs point to their
instability in the presence of water molecules This has been tested and verified for some layered
COFs with boronate esters49 Such COFs are stable as hosts for organometallic molecules55 COF-102
44
framework was found to be stable and robust even in the presence of highly reactive cobaltocenes
The highly stable ferrocenes appear to have an arrangement within the framework led by π-π
interactions
22 Covalent Triazine Frameworks (CTFs)
In the year 2008 Thomas et al56 synthesized a crystalline extended porous nanostructure by
trimerization of polytriazine monomers in ionothermal conditions With the catalytic support of ZnCl2
three aromatic nitrile groups can be trimerized to form a hexagonal C3N3 ring The resulting structure
known as Covalent Triazine-based Frameworks (CTFs) is similar to 2D COFs in geometry and atomic
composition (see Figure 2) Usage of larger non-planar monomers at higher amounts of ionic salts
however led to the formation of contorted structures Interestingly those structures showed
enhanced surface area and pore volume The trimerization of monomers that lack a linear
arrangement of nitrile groups ended up as organic polymer networks Later the same group
introduced mesoporous channels onto a CTF (CTF-2)57 by increasing temperature and ion content
The resulting structure however was amorphous It is concluded that the reaction parameters and
the amount of salt play a crucial role for tuning the porosity and controlling the order of the material
respectively58
Figure 2 Trimerization of nitrile monomers to form CTF-1 Atom colors blue N grey C white H
Ben et al59 followed ionothermal conditions for the targeted synthesis of a 3D framework using
tetraphenylmethane block and a triangular triazine ring The end product was amorphous thermally
stable up to 400 0C and could be applied for the selective sorption of benzene over cyclohexane On a
later synthetic study Zhang et al60 reported that this polymerization could be accomplished in short
45
reaction times under microwave enhanced conditions The template-free high temperature dynamic
polymerization reactions constitute irreversible carbonization reactions coupled with reversible
trimerization of nitriles The irreversibility encounter in these condensation reactions is responsible
for the production of frameworks as amorphous solids6162
An amorphous yet microporous CTF was found to be able to catalyze the oxidation of glycerol with
Pd nanoparticles that are dispersed in the framework63 This solid catalyst appears to be strong
against deactivation and selective toward glycerate compared to Pd supported activated carbon This
is attributed to the relatively high stability of Pd nanoparticles that benefit from the presence of
nitrogen functionalities in CTF An advancement on selective oxidation of methane to methanol at
low temperature is accomplished by using an amorphous bipyridyl-rich platinum-coordinated CTF as
catalyst64
23 Porous Aromatic Frameworks (PAFs)
a notably high surface area (BET surface area of 5600 m2g-1 Langmuir surface area of 7100 m2g-1)65
PAF-1 has been synthesized in a rather simple way using a nickel(0)-catalyzed Yamamoto-type66
Ullmann cross-coupling reaction67(see Figure 3) The material is thermally (up to 520 0C in air) and
hydrothermally (in boiled water for a minimum of seven days) stable The framework exposes all
faces and edges of the phenyl rings to the pores and hence the high surface area can be achieved
while its high stability is inherited from the parent diamond structure The synthesized material was
verified as disordered and amorphous yet with uniform pore diameters The excess H2 uptake
capacity of PAF-1 was measured as 70 wt at 48 bar and 77 K It can adsorb 1300 mg g-1 CO2 at 40
bar and room temperature PAF-1 was also tested for benzene and toluene adsorption
Figure 3 Coupling reaction using halogenated tetragonal blocks for the formation of diamond-like PAF-1 Atom colors red Br grey C white H
46
An attempt to synthesize a diamondoid with quaterphenyl as linker is described recently68 A
tetrahedral unit and a linear linker each of them has two phenyl rings were polymerized using the
Suzuki-Miyaura coupling reaction54 The product (PAF-11) however stayed behind theoretical
predictions and performed poorly pointing to its shortcomings such as short-range order distortion
and interpenetration of phenyl rings Nevertheless the amorphous solid displayed high thermal and
chemical stabilities proneness for adsorbing methanol over water and usability for eliminating
harmful aromatic molecules
PAF-569 is a two-dimensional hexagonal organic framework synthesized using the Yamamoto-type
Ullmann reaction This material is composed only of phenyl rings however lack long range order
because of the distortion of the phenyl rings and kinetics controlled irreversible coupling reaction It
retains a uniform pore diameter and provides high thermal and chemical stability despite its
amorphous nature PAF-5 was suggested for adsorbing organic pollutants at saturated vapour
pressure and room temperature
Co-condensation reactions with piperazine enabled nucleophilic substitution of cyanuric chloride to
form a covalently linked porous organic framework PAF-670 The synthesis in mild conditions yields a
product with uniform morphology and a certain degree of structural regularity Being nontoxic this
material was tested for drug delivery thereby stepping into biomedical applications of covalently
linked organic frameworks Primary tests suggest that PAF-6 could be promoted for drug release for
its permanent porosity and facile functionalization In comparison with MCM-4171 a widely tested
inorganic framework PAF-6 performed equally or even superiorly
24 Schemes for synthesis
The majority of the COFs were synthesized using solvothermal step-by-step condensation
(dehydration) reactions The method incorporates reversibility and is applicable for supplying long
range order in COF materials The reactants generally consist of boronic acids and hydroxy-
polyphenyl (catechol) compounds Extended porous networks are obtained when O-B or O-C bonds
are formed into 5 or 6-membered rings after dehydration (scheme 1) Dehydration may also be
carried out using microwave synthesis Alternately boronic acid can react with catechol acetonide in
presence of a Lewis acid catalyst The molecules link to each other after the liberation of acetone and
water (scheme 2) The utilization of phenyl-attached formyl and amino groups as building units
results in the formation of imines after dehydration (scheme 3) A desired arrangement of molecular
arrays on surfaces can be fulfilled by depositing planar molecules on metallic surfaces followed by
covalent linking using annealing
47
Isocyanate groups follow trimerization to form poly-isocyanurate rings (scheme 4) Cyclotrimerization
of monomers containing nitrile groups or cyanate groups is prone to generate C3N3 rings (scheme 5)
However the ionothermal synthesis of them resulted with amorphous materials Unique bond
formation is often not achieved throughout the material and thus the crystal lacks long-range order
Analogously coupling reactions such as Ullmann and Suzuki-Miyaura give rise to amorphous
products However they are adequate in producing C-C bonds when halogen-substituted compounds
are reacted (scheme 6) Nucleophilic substitution may also be used for framing networks in cases
like reactants are formed into nucleophiles and ions after release of their end-groups (scheme 7)
48
Figure 4 Different schemes reported for the synthesis of covalently-bound organic frameworks
49
25 List of synthesized materials
Table 1 List of synthesized materials in the literature their building blocks synthesis methods measured surface area [m
2 g
-1] pore volume [cm
3 g
-1] and pore size [Aring]
COF Names Reactants Synthesis Surface
Area
Pore
volume
Pore
size
COF-1 BDBA Solvothermal condensation235072
scheme 1
711 62850 032
03650
9
COF-5 BDBA HHTP Solvothermal condensation23
scheme 1
1590 0998 27
Microwave3673 scheme 1 2019
BDBA TPTA Lewis acid catalysis35 TPTA
COF-6 BTBA HHTP Solvothermal condensation27
scheme 1
980 (L) 032 64
COF-8 BTPA HHTP Solvothermal condensation27
scheme 1
1400 (L) 069 187
COF-10 BPDA HHTP Solvothermal condensation27
scheme 1
2080 (L) 144 341
BPDA TPTA Lewis acid catalysis35 scheme 2
COF-18Aring BTBA THB Facile dehydration3348 scheme 1 1263 069 18
COF-16Aring BTBA alkyl-THB
(alkyl = CH3)
Facile dehydration48 scheme 1 753 039 16
COF-14Aring BTBA alkyl-THB
(alkyl = C2H5)
Facile dehydration48 scheme 1 805 041 14
COF-11Aring BTBA alkyl-THB
(alkyl = C3H7)
Facile dehydration48 scheme 1 105 0052 11
50
SCOF-1 BDBA Substrate-assisted synthesis39
scheme 1
SCOF-2 BDBA HHTP Substrate-assisted synthesis39
scheme 1
TP COF PDBA HHTP Solvothermal condensation25
scheme 1
868 079 314
PPy-COF PDBA Solvothermal condensation26
scheme 1
923 188
TBB COF TBB (on Cu(111) and
Ag(110) surfaces)
Surface polymerisation40 scheme
6
TBPB COF TBB (on Au(111)
surface)
Surface polymerisation41 scheme
6
BTP COF BTPA THDMA Solvothermal condensation72
scheme 1
2000 163 40
HHTP-DPB COF DPB HHTP Solvothermal condensation73
scheme 1
930 47
PICU-A DMBPDC Cyclotrimerization74 scheme 4
PICU-B DCF Cyclotrimerization74 scheme 4
COF-LZU1 DAB TFB Solvothermal condensation53
scheme 3
410 054 12
PdCOF-LZU1 COF-LZU1 PdAc Facile post-treatment53 146 019 12
XN3-COF-5 X N3-BDBA (100-X)
BDBA HHTP
Solvothermal condensation51
scheme 1
2160
(X=5)
1865 (25)
1722 (50)
1641 (75)
1421
(100)
1184
1071
1016
0946
0835
295
276
259
258
227
51
XAcTrz-COF-5 XN3-COF-5 PAc Click reaction51 2000
(X=5)
1561 (25)
914 (50)
142 (75)
36 (100)
1481
0946
0638
0152
003
298
243
156
153
125
XBuTrz-COF-5 XN3-COF-5 HP Click reaction51
XPhTrz-COF-5 XN3-COF-5 PPP Click reaction51
XEsTrz-COF-5 XN3-COF-5 MP Click reaction51
XPyTrz-COF-5 XN3-COF-5 PyMP Click reaction51
COF-42 DETH TFB Solvothermal condensation75
scheme 3
710 031 23
COF-43 DETH TFPB Solvothermal condensation75
scheme 3
620 036 38
CTF-1 DCB Ionothermal trimerization56
scheme 5
791 040 12
CTF-2 DCN Ionothermal trimerization57
scheme 5
90 8
mp-CTF-2 2255 151 8
CTF (DCP) DCP Ionothermal trimerization64
scheme 5
1061 0934 14
K2[PtCl4]-CTF DCP K2[PtCl4] Trimerization (scheme 5) +
coordination64
Pt-CTF DCP Pt Trimerization (scheme 5) +
coordination64
PAF-5 TBB Yamamoto-type Ullmann cross-
coupling reaction69 scheme 6
1503 157 166
52
PAF-6 PA CA Nucleophilic substitution70
scheme 7
1827 118
Pc-PBBA COF PcTA BDBA Lewis acid catalysis34 scheme 2 506 (L) 0258 217
NiPc-COF OH-Pc-Ni BDBA Solvothermal condensation37
scheme 1
624 0485 190
XN3-NiPc-COF OH-Pc-Ni X N3-BDBA
(100-X) BDBA
Solvothermal condensation51
scheme 1
XEsTrz-NiPc-
COF
XN3-NiPc-COF MP Click reaction51
ZnP COF TDHB-ZnP THB Solvothermal condensation38
scheme 1
1742 1115 25
NiPc-PBBA COF NiPcTA BDBA Lewis acid catalysis35 scheme 2 776
2D-NiPc-BTDA
COF
OHPcNi BTDADA Solvothermal condensation47
scheme 1
877 22
ZnPc-Py COF OH-Pc-Zn PDBA Solvothermal condensation
scheme 1
420 31
ZnPc-DPB COF OH-Pc-Zn DPB Solvothermal condensation
scheme 1
485 31
ZnPc-NDI COF OH-Pc-Zn NDI Solvothermal condensation
scheme 1
490 31
ZnPc-PPE COF OH-Pc-Zn PPE Solvothermal condensation
scheme 1
440 34
COF-366 TAPP TA Solvothermal condensation46
scheme 3
735 032 12
COF-66 TBPP THAn Solvothermal condensation46
scheme 1
360 020 249
53
COF-102 TBPM Solvothermal condensation24
scheme 1
3472 135 115
Microwave36
scheme 1
2926
COF-102-C12 TBPM trunk-TBPM-R
(R=dodecyl)
Solvothermal condensation52
scheme 1
12
COF-102-allyl TBPM trunk-TBPM-R
(R=allyl)
Solvothermal condensation52
scheme 1
COF-103 TBPS Solvothermal condensation24
scheme 1
4210 166 125
COF-105 TBPM HHTP Solvothermal condensation24
scheme 1
COF-108 TBPM HHTP Solvothermal condensation24
scheme 1
COF-202 TBPM TBST Solvothermal condensation28
scheme 1
2690 109 11
COF-300 TAM TA Solvothermal condensaion29
scheme 3
1360 072 72
PAF-1 TkBPM-X (X=C) Yamamoto-type Ullmann cross-
coupling reaction65 scheme 6
5600
PAF-2 TCM cyclotrimerization59 scheme 5 891 054 106
PAF-3 TkBPM-X (X=Si) Yamamoto-type Ullmann cross-
coupling reaction76 scheme 6
2932 154 127
PAF-4 TkBPM-X (X=Ge) Yamamoto-type Ullmann cross-
coupling reaction76 scheme 6
2246 145 118
PAF-11 TkBPM-X (X=C) BPDA Suzuki coupling reaction68 704 0203 166
54
scheme 6
3 Studies of structure and properties of COFs
31 2D COFs
Following the literature of the years 2005-2010 2D COFs were believed to have high symmetric AA
(eclipsed ndash P6mmm symmetry) stacking as boron nitride (h-BN) [REF] An exception was COF-1
which was considered to have AB (staggered ndash P63mmc) stacking as graphite[REF] The AA stacking
maximizes the attractive London dispersion interaction between the layers an important
contribution to the stacking energy At the same time AA stacking always suffers repulsive Coulomb
force between the layers due to the polarized connectors as the distance between atoms exposing
the same charge is closest It should be noted that in boron nitride boron atoms serve as nearest
neighbors to nitrogen atoms in adjacent layers The Coulomb interaction has ruled out possible
interlayer eclipse between atoms of alike charges Based on these facts we modeled77 a set of 2D
COFs with different possible stacks Each layer was drifted with respect to the neighboring layers in
directions symmetric to the hexagonal pore and we calculated the total energies and their Coulombic
contributions The AA stacking version was found to have the highest Coulombic repulsion in each
COF The lattice types with the layers shifted to each other by ~14 Aring approximately the bond length
between two atoms in a 2D COF were found to be more stable compared to AA or AB In this low-
symmetry configuration the hexagonal rings in the pore walls undergo staggering with that of
adjacent layers Within an error limit the bulk COFs may be either serrated or inclined as shown in
Figure 5 The interlayer separations of all the COF stacks are in the range of 30 to 36 Aring and increase
in the order AB serratedinclined AA32 The motivation for proposing inclined stacking for COFs is
that the rhombohedral graphite (ABC stacking) is the inclined version of the layer-offset in natural
graphite (AB stacking) The instability of AA stacking was also suggested by the studies of elastic
properties of COF-1 and COF-5 by Zhou et al78 The shear modulus show negative signs for the
vertical alignment of COF layers while they are small but positive for the offset of layers
55
Figure 5 Different stacks of a 2D covalent organic framework COF-1 The atoms are shown as Van der Waals spheres
The different stacking modes should in principle be visible in their PXRD patterns as each space
group has a distinct set of symmetry imposed reflection conditions We simulated the XRD patterns
of COFs in their known and new configurations and on comparison with the experimental spectrum
the new as well as AA stacking showed good agreement3279 However the slight layer-offsets in
conjunction with the large number of atoms in the unit cells change the PXRD spectrum only by the
appearance of a few additional peaks all of them in the vicinity of existing signals and by changes in
relative intensities Unfortunately the low resolution of the experimental data does now allow
distinguishing between the different stackings as the broad signals cover all the peaks of the
simulated spectrum
A detailed analysis on the interlayer stacking of HHTP-DPB COF has been made by Dichtel et al73 is
very complementary73 This work uses molecular mechanics extensively to define potential energy
surfaces of stacking of two layers from eclipsed to staggered covering all possible layer offsets The
low energy region was near to the eclipsed stacking which was then zoomed in to examine using DFT
for higher accuracy The far-eclipsed regions encounter no energy barriers with the near-eclipsed
regions which suggest the unlikeness of forming near-staggered layering Within DFT the eclipsed
geometries are at high energy compared to the slight offsets around AA (~ 13 ndash 25 Aring away) For a
hexagonal COF the energetically preferred offset could be in one of the six hexagonal sectors (or
circle sectors) of central angle 60 The preference of falling into one sector over the others is
arbitrary and may be determined by symmetry set by nature or external parameters Hence the
layering order in bulk could be serrated inclined spiral or purely by random The latter case better
known as turbostratic disorder is then more like a rule for 2D COFs rather than an exception caused
by external forces This also explains why the experimental PXRD patters have relatively broad peaks
The layer-offset may not only change the internal pore structure but also affect interlayer exciton
and vertical charge transport in opto-electronic applications
56
Concerning the stacking stability the square 2D COFs are expected not to be different from
hexagonal COFs because the local environment causing the shifts is nearly the same DFTB based
calculations reported along with the synthesis of electron-transporting 2D-NiPc-BTDA COF supports
this notion47 Two shifted configurations ndash at 07 and 14 Aring away from AA ndash appeared to be
energetically preferred over the AA stacking It appears that AA stacking is only possible for boron
nitride-like structures were adjacent layers have atoms with opposite charges in vertical direction In
analogy to BN layers it might be possible to force AA stacking by the introduction of alkalis in
between the layers
32 3D COFs
3D COFs in general are composed of tetragonal and triangular building blocks So far that their
synthesis leads to two high symmetry topologies ndash ctn and bor - in high yields24 The two topologies
differ primarily in the twisting and bulging of their components at the molecular level The
thermodynamic preference of one topology over the other may result from the kinetic entropic and
solvent effects and the relative strain energies of the molecular components It is straight-forward to
state that the effects at the molecular level crucial crucial in the bulk state since transformation from
one net to the other is impossible without bond-breaking There has not been any detailed study on
this matter experimentally or theoretically
Schmid et al8182 have developed force-field parameters from first principles calculations using
Genetic Algorithm approach The parameters developed for cluster models of COF-102 can
reproduce the relative strain energies in sufficient accuracies and be extended to calculations on
periodic crystals The internal twists of aromatic rings within the tetragonal units are different for ctn
and bor nets which lead to stable S4 and unstable D2d symmetries respectively This together with
the bulging of triangular units in the bor net reasoned for energetic preference of ctn over bor for all
boron-based 3D COFs79
The connectivity-based atom contribution method (CBAC) developed by Zhong et al80 can
significantly reduce computational time needed for quantum chemical calculation for framework
charges when screening a large number of MOFs or COFs in terms of their adsorption properties The
basic assumption is that atom types with same bonding connectivity in different MOFsCOFs have
identical charges a statement that follows from the concept of reticular chemistry where the
properties of the molecular building blocks keep their properties in the bulk After assigning the
CBAC charges to the atom types slight adjustment (usually less than 3 ) is needed to make the
frameworks neutral Simulated adsorption isotherms of CO2 CO and N2 in MOFs and COFs show that
CBAC charges reproduce well the results of the QM charges8384 The CBAC method still requires a
57
well-parameterized force field in order to account correctly for adsorption isotherms as the second
important contribution to the host-guest interaction is the London dispersion energy between
framework and adsorbed moleculesG
The elastic constants calculated for the 3D COFs COF-102 and COF-108 show that COF-102 is nearly
five times stronger than COF-10878 While the bulk modulus (B) of COF-102 (B = 216 GPa) exceeds
that of known MOFs such as MOF-5 MOF-177 and MOF-205 (B = 153 101 107 GPa respectively81)
the least dense COF-108 is at the edge of structural collapse (B = 49 GPa) The calculations were
made using Density Functional Tight-Binding (DFTB) method Another report82 based on the same
level of theory possibly with a different parameter set however reveals lower bulk moduli for both
COFs 177 GPa for COF-102 and 005 GPa for COF-108 The bulk moduli for COF-103 and COF-105 are
110 and 33 GPa respectively Recently we have reported the structural properties of 3D COFs The
calculations were made using self-consistent charge (SCC) DFTB method The bulk modulus of each
COF is as follows COF-102 ndash 206 COF-103 ndash 139 COF-105 ndash 79 COF-108 ndash 37 COF-202 ndash 153 and
COF-300 ndash 144 GPa Force field calculations79 provide slightly different values COF-102 ndash 170 COF-
103 ndash 118 COF-105 ndash 57 COF-108 ndash 29 COF-202 ndash 110 GPa Albeit the differences in values each
type of calculation shows the trend that bulk modulus decreases with decreasing mas density and
increasing pore volume and distance between connection nodes One has to note that the high
mechanical stabilities of 3D COFs as suggested by the calculations correspond always to defect-free
crystals Theory is expected therefore to overestimate experimental mechanical stability for real
materials
COFs in general have semiconductor-like band gaps (~ 17 to 40 eV)32 Layer-offset from eclipsed
layering slightly increases the band gap However the Density-Of-States (DOS) of a monolayer is
similar to that of bulk COF The band gap is generally smaller for a COF with larger number conjugate
rings
The thermal influence on the crystal structure of 3D COFs is also intriguing as negative thermal
expansion (NTE) the contraction of crystal volume on heating has been reported for COF-10283 The
studies were performed using molecular dynamics with the force field parameters by Schmid et al84
However the thermal expansion coefficient α = -151 times 10-6 K-1 is much smaller compared to that of
some of the reported MOFs that also show NTE behavior85 The volume-contraction upon the
increase of temperature arises from the tilt of phenyl linkers between B3O3 rings and sp3 carbon
atom in TBPM (figure ) Later Schmid et alreported that COF-102 103 105 108 exhibit NTE
behavior both in ctn and bor topologies whereas COF-202 only in ctn topology79 A typical
application is the realization of controllable thermal expansion composites made of both negative
and positive thermal expansion materials
58
4 Applications
41 Gas storage
The success in the synthesis of COFs was certainly the result of a long-term struggle for complete
covalent crystallization The discovery of COFs coincided with the time when world-wide effort was
paid to develop new materials for gas storage in particular for the development hydrogen tanks for
mobile applications Materials made exclusively from light-weight atoms and with large surface
areas were obviously excellent candidates for these applications The gas storage capacity of porous
materials relies on the success of assembling gas molecules in minimum space This is achieved by
the interaction energy exerted by storage materials on the gas molecules Because the interactions
are noncovalent no significant activation is required for gas uptake and release and hence the so-
called physical adsorption or physisorption is reversible The other type of adsorption ndash chemical
adsorption or chemisorption ndash binds dissociated hydrogen covalently or interstitially at the risk of
losing reversibility As it requires the chemical modification of the host material chemisorption is not
a viable route for COFs and might become possible only through the introduction of reactive
components into the lattice The total amount of gas adsorbed in the pores gives rise to what is
referred to as lsquototal adsorption capacityrsquo This quantity is hard to be assessed experimentally as a
measurement is always subjected to influence of the materials surface and gas present in all parts of
the experimental setup Therefore the lsquoexcess adsorption capacityrsquo is reported in experiment Here
the gas stored in the free accessible volume is subtracted from the total adsorption In experiment
this volume includes the voids in the material as well as any empty space between the sample
crystals This volume is typically determined by measuring the uptake of inert gas molecules (He for
H2 adsorption N2 or Ar for adsorption studies of larger molecules) at room temperature with the
assumption that the host-guest interaction between the material and He can be neglected The
different definitions of adsorption is given in Figure 6
Typically experiments measure excess values and simulations provide total values Quantities of
adsorbed molecules are usually expressed as gravimetric and volumetric units The former gives the
amount per unit mass of the adsorbent while the latter gives the amount per unit volume of the
adsorbent Explicative definitions and terminologies related to gas adsorption can be found
elsewhere86
59
Figure 6 Hydrogen density profile in a pore A denotes the excess A+B the absolute and A+B+C the total adsorption The skeletal volume corresponds to the volume occupied by the framework atomsCourtesy of Dr Barbara Streppel and Dr Michael Hirscher MPI for intelligent systems Stuttgart Germany
411 Porosity of COFs
It is common to inspect the porosity of synthesized nanoporous materials using some inert or simple
gas adsorption measurements Distribution of pore size can be sketched from the
adsorptiondesorption isotherm N2 and Ar isotherms provide an approximation of the pore surface
area pore volume and pore size over the complete micro and mesopore size range Usually the
surface area is calculated using the Brunauer-Emmet-Teller (BET) equation or Langmuir equation
Total pore volume (volume of the pores in a pre-determined range of pore size) can be determined
from either the adsorption or desorption phase The Dubinin-Radushkevic method the T-plot
method or the Barrett-Joyner-Halenda (BJH) method may also be employed for calculating the pore
volume The pore size is often corroborated by fitting non-local density functional theory (NLDFT)
based cylindrical model to the isotherm90-93 In some cases the desorption isotherm is affected by
the pore network smaller pores with narrower channels remain filled when the pressure is lowered
This leads to hysteresis on the adsorption-desorption isotherms The different methods employed for
pore structure analysis are characteristic for micropore filling monolayer and multilayer formations
capillary condensation and capillary filling
For any adsorbate in order to form a layer on pore surface the temperature of the surface must
yield an energy (kBT) much less than adsorbate-surface interaction energy Additionally the absolute
value of the adsorbate-surface binding energy must be greater than the absolute value of the
adsorbate-adsorbate binding energy This avoids clustering of the adsorbate into its three-
dimensional phase
60
At high pressure the adsorption isotherm shows saturation which means that no more voids are left
for further occupation The isotherms show different behaviors characteristic of the pore structure of
the materials There are known classifications based on these differences type I II III IV and V For
COFs and the related materials discussed in this review type I II and IV have been observed (see
Figure 7) As more adsorbate is attracted to the surface after the first surface layer completion one
can expect a bend in the isotherm Type I implies monolayer formation which is typical of
microporosity If the surface sites have significantly different binding energies with the adsorbate
type II behavior is resulted In type IV each ldquokneerdquo where the isotherm bends over and the vapor
pressure corresponds to a different adsorption mechanisms (multiple layers different sites or pores)
and represents the formation of a new layer
Figure 7 Different types of adsorptions in Covalently-bound Organic Frameworks
Argon and nitrogen adsorption measurements at respectively 87 and 77 K exhibit type I isotherms
for COF-1 6 102 and 103 and type IV isotherms for COF-5 8 10 102 and 10387 Smaller pore
diameters of COF-1 and 6 (~ 9 Aring each) are characteristic of monolayer gas formation on the internal
pore surface The same reasons are responsible for the type I character of COF-102 and COF-103
(pore size = 12 Aring each) isotherms albeit with some unusual steps at low pressure The type IV
isotherms of COF-5 8 10 102 and 103 show formation of monolayers followed by their
multiplication within their relatively larger pores (sizes are approximately 27 16 32 12 and 12 Aring
respectively) Other COFs that exhibit type I isotherms for N2 adsorption are the 3D COFs COF-202 (11
Aring) COF-300 (72 Aring) and COF-102-C12 (12 Aring) the alkylated 2D COF series COF-18Aring COF-16Aring COF-14Aring
COF-11Aring (the nomenclature includes their pore sizes) and a 2D square COF NiPc COF (19 Aring)
Isotherms like Type-II were observed for PPy-COF (188 Aring) COF-366 (176 Aring) COF-66 (249 Aring) Pc-
PBBA COF (217 Aring) ZnP-COF (25 Aring) COF-LZU1 (12 Aring) PdCOF-LZU1 (12 Aring) and some of the azide-
appended COFs 50AcTrz-COF-5 (156 Aring) 75AcTrz-COF-5 (153 Aring) 100AcTrz-COF-5 (125 Aring)
50BuTrz-COF-5 (232 Aring) 50PhTrz-COF-5 (212 Aring) 50EsTrz-COF-5 (212 Aring) and 25PyTrz-COF-5
(221 Aring) The majority of the COFs with mesopores exhibit type-IV isotherms They are TP-COF (314
Aring) BTP-COF (40 Aring) COF-42 (23 Aring) COF-43 (38 Aring) HHTP-DPB COF (47 Aring) CTC-COF (226 Aring) NiPc-PBBA
COF ZnPc-Py-COF (31 Aring) ZnPc-DPB-COF (31 Aring) ZnPc-NDI-COF (31 Aring) ZnPc-PPE-COF (34 Aring) 5N3-
61
COF (295 Aring) 25N3-COF (276 Aring) 50N3-COF (259 Aring) 75N3-COF (258 Aring) 100N3-COF (227 Aring)
5AcTrz-COF-5 (298 Aring) 25AcTrz-COF-5 (243 Aring) 5PyTrz-COF-5 (289 Aring) and 10PyTrz-COF-5
(242 Aring)
The synthesized microporous amorphous materials CTF-1 (12 Aring) CTF-DCP (14 Aring) PAF-1 and PAF-2
(106 Aring) exhibit type-I isotherms with N2 adsorption PAF-3 (127 Aring) PAF-4(118 Aring) PAF-5 (157 Aring)
PAF-6 (118 Aring) and PAF-11 showed surprisingly not the typical type I behavior in spite of their
microporosity but type-II isotherms
Many of the mesoporous COFs showed small hysteresis in the adsorption-desorption isotherm
pointing the possibility of capillary condensation Hysteresis was observed for the amorphous
materials in both mirco and meso-pore range Various reasons have been attributed for the observed
hysteresis including the softness of the material and guest-host interactions
412 Gas adsorption experiments
Hydrogen uptake capacities of some boron-based COFs were measured by Yaghi et al87 The excess
gravimetric saturation capacities of COF-1 -5 -6 -8 -10 -102 and -103 at 77 K were found to be 148
358 226 350 392 724 and 705 mg g-1 respectively The saturation pressure increases with an
increase in pore size similar to MOFs88 Pore walls of 2D COFs consist only of edges of connectors
and linkers The fact that faces and edges are largely available for interactions with H2 in 3D
geometries is a reason for their enhanced capacity Total adsorption generally increases without
saturation upon pressure because the difference between the total and the excess capacities
corresponds to the situation in bulk (or in a tank) COFs show in particular good gravimetric
capacities because of their low mass density while volumetric capacities typically do not exceed
those of MOFs The gravimetric hydrogen storage capacities of COF-102 and -103 exceeds 10 wt at
a pressure of 100 bar
COF-18 and its alkylated versions COF-16 -14 and -1 have H2 excess gravimetric capacities 155 144
123 and 122 wt respectively at hellipK and hellipbar
Excess uptake of methane in COFs at room temperature and at 70 bar pressure is as follows COF-1
and -6 up to10 wt COF-5 -6 and -8 between 10 and 15 wt and COF-102 and -103 above 20
wt 87 Isosteric heat of adsorption Qst calculated by fitting the isotherms to virial expansion with
the same temperature are relatively weaker (lt 10 kJmol) for COF-102 and COF-103 at low
adsorption which implies weaker adsorbate-adsorbent interactions In comparison COF-1 and 6
exhibit twice the enthalpy however the overall hydrogen storage capacity was very limited due to
62
the smaller pore volume High Qst at zero-loading is not desirable because the primary excess amount
adsorbed at very low pressures cannot be desorbed practically89
COF-102 and 103 outperform 2D COFs for CO2 storage87 The saturation uptakes at room
temperature were 1200 and 1190 mg g-1 for COF-102 and 103 respectively
A type IV isotherm has been observed for ammonia adsorption in COF-10 inherent to its mesoporous
nature90 The material showed an uptake of 15 mol kg-1 at room temperature and 1 bar the highest
of any nanoporous materials The repeated adsorption-desorption cycles were reported to disrupt
the layer arrangement of this 2D COF Yaghi et al90 were also able to make a tablet of COF-10 crystal
which performed nearly up to the crystalline powder
Not many COFs have been experimentally studied for gas storage applications in spite of high
expectations This has to be understood together as a result of the powder-like polycrystallization of
COFs The enthalpy Qst at low-loading accounted to only 46 kJmol
The excess and total hydrogen uptake capacity of PAF-1 at 48 bar and 77 K reached 7 wt and 10
wt respectively65 PAF-3 and PAF-478_ENREF_73 follow the same diamond topology as PAF-1 the
difference accounts in the central atoms of the tetragonal blocks PAF-1 3 and 4 have C Si and Ge
atoms at the centers respectively At 1 bar the respective adsorption capacities were 166 207 and
150 wt The highest value being found for PAF-3 is attributed to its relatively high enthalpy (66 kJ
mol-1) compared to PAF-1 (54 kJ mol-1) and PAF- 4 (66 kJ mol-1) At high pressure the surface area is
a key factor and because of the lower BET surface area PAF-3 and PAF-4 performs poor At 60 bar
their respective adsorption capacities were 55 and 42 wt For PAF-11 hydrogen uptake was 103
wt at 1 bar68
Methane uptakes of PAF-1 3 and 4 at 1 bar are 13 19 and 13 wt respectively76 Their enthalpies
are respectively 14 15 and 232 kJ mol-1 This suggests that Ge moiety have strong interactions with
methane
CO2 adsorption in PAF-1 at 40 bar and room temperature was 1300 mg g-1 which was slightly more
than the capacity of COF-102 and 10365 PAF-1 3 and 4 at 1 atm pressure could store 48 80 and 51
wt respectively At 273 K and 1 atm they stored 91 153 and 107 wt respectively The storage
capacities at 273 K and 298 K lead to the following zero-loading isosteric enthalpies 156 192 162
kJ mol-1 for PAF-1 3 and 4 respectively The higher uptake of PAF-3 may also be explained by its
relatively higher surface area with CO2 molecules
The selective adsorption of greenhouse gases in PAFs was discussed by Ben et al76 At 273 K and 1
atm pressure PAF-1 selectively adsorbes CO2 and CH4 respectively 38 and 15 times more in
63
amounts than N2 PAF-3 showed extraordinary selectivity of 871 for CO2 over N2 and 301 for CH4
over N2 For PAF-4 the selectivity was 441 for CO2 over N2 and 151 for CH4 over N2 Generally the
uptake of N2 Ar H2 O2 are significantly lower compared to CO2 and NH4 which makes these PAFs
suitable for separating them
Harmful gases like benzene and toluene can be stored in PAF-1 in amounts 1306 mg g-1 (1674 mmol
g-1) and 1357 mg g-1 (1473 mmol g-1) respectively at saturated pressures and room temperature65
In PAF-11 their adsorptions were in amounts of 874 mg g-1 and 780 mg g-1 respectively68 PAF-2 was
accounted to a much lower benzene adsorption (138 mg g-1) at the same conditions59 Adsorption of
cyclohexane vapor in PAF-2 was only 7 mg g-1 While PAF-11 exhibited hydrophobicity it could
accommodate significant amount of methanol (654 mg g-1 or 204 mmol g-1) at room temperature
and saturated pressure68 PAF-5 has been tested for storing organic pollutants69 At room
temperature and saturated pressure PAF-5 adsorbed methanol benzene and toluene in amounts
6531 cm3 g-1 (292 mmol g-1) 26296 cm3 g-1 (117 mmol g-1) and 2582 cm3 g-1 (115 mmol g-1)
respectively Their adsorptions in PAF-5 were deviated from the typical type-I behavior Methanol
exhibited a large slope at the high pressure region corresponding to the relative size of it Zhu G et
al dedicated PAF-6 for biomedical applications With no toxicity observed for PAF-6 over a range of
concentrations adsorption of ibuprofen was measured in it PAF-6 showed an uptake of 350 mg g-1
The release of ibuprofen was also tested in simulated body fluid which showed a release rate of 50
in 5 hours 75 in 10 hours and 100 in almost 46 hours
413 Theoretical predictions
Grand canonical Monte Carlo (GCMC) is a widely used method for the simulation of gas uptakes in
nanoporous materials In GCMC the number of adsorbates and their motions are allowed to change
at constant volume temperature and chemical potential to equilibrate the adsorbate phase The
motions are random guided by Monte Carlo methods and the energies are calculated by force field
methods The details of it may be found in the literature91
Goddart III WA et al92 simulated the uptake capacities of boron-based COFs using force fields derived
from the interactions of H2 and COFs calculated at MP2 level of theory The saturated excess uptakes
of both COF-105 (at 80 bar) and 108 (at 100 bar) were accounted to 10 wt at 77 K which are more
than the uptakes measured in ultrahigh porous MOF-200 205 and 21093 The predictions for other
COFs include 38 wt for COF-1 (AB stacking) 34 wt for COF-5 (AA stacking) 88 wt for COF-102
and 91 wt for COF-103 At 01 bar COF-1 in AB stacking showed highest excess uptake (17 wt )
compared to other COFs in the present discussion Total uptake capacities of the COFs are in the
following order COF-108 (189 wt ) COF-105 (183 wt ) COF-103 (113 wt ) COF-102 (106
64
wt ) COF-5 (55 wt ) and COF-1 (38 wt ) It is worth noticing the higher gravimetric capacities of
COF-108 and 105 which were not measured experimentally They benefit from their lower mass and
higher pore volume In case of volumetric capacity COF-102 (404 gL excess) surpasses the others at
high pressures The total volumetric capacities of the COFs are 499 498 399 395 361 and 338
gL for COF-102 103 108 105 1 and 5 respectively Their additional calculations predicted benzene
rings as favorite locations for H2 molecules
Similar values and trends have been reported by Garberoglio G94 who also reasoned poor solid-fluid
interactions in a large part of free volume for the low volumetric capacity of COF-108 and 105 A
room temperature excess gravimetric uptake of approximately 08 wt was obtained for COF-108
and 105 Quantum diffraction effects were added to the interaction energies of H2 with itself and the
material which were calculated using universal force-field (UFF) With possible overestimation
Klontzas et al30 and Bassem et al95 have reported total gravimetric uptake of 45 and 417 wt
respectively at 300 K for the same COFs Among the boron-based 3D COFs COF-202 exhibited much
smaller uptake capacity at high pressures due to its smaller pore volume95 Lan et al96 predicted a
very small H2 uptake of 152 wt COF-202 at 300K and 100 bar Studies of Yang et al97 clarifies that
pore filling with H2 in 2D COFs is a gradual process rather than being in steps of layer formation
Garberoglio and Vallauri98 pointed out that the relatively higher mass density and lower surface area
per unit volume of 2D COFs are the reasons for their lower uptakes in comparison with 3D COFs The
surface area of hexagonal 2D COFs (AA stacking) averaged around 1000 m2 cm-3 while that of 3D
COFs were about 1500 m2 cm-3
Simulations of H2 adsorption in the perfect crystalline form of PAF-1 and PAF-11 also known as PAF-
302 and PAF-304 respectively are carried out by Zhu et al99 At 77 K H2 excess uptake of PAF-302 (7
wt ) is slightly higher than COF-102 (684 wt )87 and slightly lower than MOF-177 (740 wt )100 At
room temperature PAF-302 exhibited a poor uptake (221 wt at 100 bar) PAF-304 showed
excellent total uptakes 2238 wt at 77 K and 100 bar 653 wt at 298 K and 100 bar These are
highest among all nanoporous materials
Babarao and Jiang studied room-temperature CO2 adsorption in COFs101 At low pressures COFs with
smaller pore volume exhibit a steep rise in their isotherm while at high pressures COF-105 and 108
(82 and 96 mmol g-1 respectively at 30 bar) dominate the others COF-6 has the highest isosteric heat
of adsoption (328 kJmol) hence it made the first steep rise in the volumetric isotherm followed by
COF-8 102 103 10 105 and 108 Correlations of gravimetric and volumetric capacities with mass
density pore volume porosity and surface area have been excellently manifested in this article101
With increasing framework-density gravimetric uptake falls inversely while volumetric capacity
decreases linearly The former rises with free volume while the latter rises and then drops slightly
65
Both of them rise with porosity and surface area Choi et al102 predicted 45 times more CO2 uptake in
COF-108 (3D) compared to COF-5 (2D) Recently Schmid et al79 also have predicted CO2 adsorption
especially for 3D COFs The uptake of COF-202 was almost half of COF-102 and 103 at room
temperature Type IV isotherm has been found for CO2 adsorption in COF-8 and 10 at low
temperatures (lt 200 K)97 Yang and Zhong97 excluded capillary condensation and dissimilar
adsorption sites but multilayer formation as reasons of the stepped behavior (Yang and Zhong
explained this as a consequence of multilayer formation rather than a result of capillary
condensation or dissimilar adsorption sites)
Methane adsorption in 2D and 3D COFs was first calculated by Garberoglio et al Among COF-6 8 and
10 the former which has smaller pore size and higher binding energy with CH4 shows better
volumetric adsorption (~ 120 cm3cm3 at 20 bar) than the others at room temperature at low
pressure region (lt 25 bar) Larger pore size favors COF-10 for higher volumetric adsorption (~ 160
cm3cm3 at 70 bar) at high pressures COF-8 with intermediate pore size shows largest excess amount
of methane adsorbed upon saturation however still lower than two 3D COFs COF-102 and 103
show excess volumetric adsorption of about 180 cm3cm3 at 35 bar This is significantly higher than
the uptake of two other 3D COFs COF-105 and 108 Entropic effects which primarily arise from the
change of dimensionality when the bulk phase of adsorbate transforms to adsorbed phase are
crucial in adsorption Garberoglio94 shows that solid-fluid interaction potential in large part of volume
of COF-105 and 108 is not strong enough to overcome the entropic loss On the other hand these
two COFs exhibit relatively higher gravimetric uptake (720 and 670 cm3g respectively) Goddard III et
al89 predicted total methane uptakes in the following amounts 415 wt in COF108 405 wt in
COF-105 31 wt in COF-103 284 wt in COF-102 196 wt in COF-10(AA) 169 wt in COF-
5(AA) 159 wt in COF-8(AA) 123 wt in COF-6(AA) and 109 wt in COF-1(AB) Yang and Zhong97
have shown that even at low temperatures (100 K) the pore filling of COF-8 with CH4 is rather
gradual than by step-by-step whereas for COF-10 multilayer formation provides a stepped behavior
in the isotherm Alternately Goddard et al pointed out that layer formation coexists with pore-filling
at room temperature89
414 Adsorption sites
First principle calculations on cluster models are typically employed to find favorite adsorption sites
and binding energies of adsorbates within frameworks Benzene rings are found to be a usual
location for H2 where it prefers to orient in perpendicular fashion3086110 Other favorite locations
include B3O3 and C2O2B rings110111 where H2 orients horizontally on the face as well as on the
edge3086 The central ring of HHTP does not provide enough binding to H2 atoms because of small
amount of charges There are some differences in the adsorption energies and favorite sites
66
calculated at different levels of theory Overall the reported binding energies between H2 and any
COF fragment are less than 6 kJ mol-1 In comparison to MOFs COFs do not offer any stronger binding
energy with H2 The advantage of COFs is their low mass density Lan et al103 reported that CO2 is
more preferred to be adsorbed on B3O3 or C2O2B rings than benzene rings Choi et al102 reported that
the oxygen sites are the primary locations for CO2 and CO2 bound COFs are the second strong binding
sites
415 Hydrogen storage by spillover
Hydrogen spillover is an alternate way to store hydrogen113114 It involves dissociation of hydrogen
gas by supported metal catalysts subsequent migration of atomic hydrogen through the support
material and final adsorption in a sorbent Here surface-diffusion of H atoms over the support is
known as primary spillover Secondary spillover is the further diffusion to the sorbent Bridging the
metal part with the sorbent is a practice to enhance the uptake It increases the contact between the
source and receptor and reduces the energy barriers especially in the secondary spillover As the
final sorption is chemisorption surface area of the sorbent is more important than pore volume
Yang and Li measured H2 storage in COF-1 by spillover They used Pt supported on active carbon
(PtAC) as source for hydrogen dissociation Active carbon was the primary receptor and COF-1 the
secondary receptor COF-1 showed much low uptake 128 wt at 77 K and 1 atm 026 wt at 298
K and 100 bar In comparison to MOFs these are very low104 However they have found that the
uptake could be scaled up by a factor of 26 by bridging COF-1 with PtAC by applying carbonization
They also report that heat of adsorption between H and surface sites is more important than surface
area and pore volume in enhancing the net adsorption by spillover
Ganz et al theoretically predicted that the uptake capacities in COFs by hydrogen spillover can be
higher than the measured value116117 Based on ab initio quantum chemistry calculations and
counting the active storage sites they predicted 45 wt for COF-1 in AB stacking and 55 wt for
COF-5 in AA stacking at room temperature and 100 bar
42 Diffusion and Selectivity
Computed self-diffusion coefficients for H2 and CH4 adsorbates in 2D COFs (in AA stacking) are up to
one order of magnitude higher than that in 3D MOFs such as MOF-598 In comparison to nanotubes
the diffusion of H2 molecules in 2D COFs is one order of magnitude slower The differences in
diffusion coefficients are attributed to different pore structures Interactions of the corners of the
hexagonal pore with the fluid present potential barriers along the cylindrical pore axis Diffusion
occurs with jumps after long waiting The height of the barrier is still smaller than in MOFs
67
Homogeneous pore walls and absence of pore corners in nanotubes contribute much less
corrugation to the solid-fluid potential energy surface Increase of self-diffusivities of H2 and CH4 with
increase in pore size of 2D COFs is also shown by Keskin[Ref] Due to the smaller size of H2 its
diffusivity is higher than that of CH4 in any material For reasons of their relative size In a mixture of
the two the self-diffusivity of CH4 increases while that of H2 decreases
Molecular dynamics simulation studies of the diffusion of gas molecules within a 2D COF performed
by Krishna et al105 revealed its relatively lower binding energy of hydrogen molecules within the pore
walls compared to argon CO2 benzene ethane propane n-butane n-pentane and n-hexane
Binding energy prevents the molecules from diffusing through the pore channels They tested if
Knudsen diffusion is applicable to COFs Knudsen diffusion occurs when the molecules frequently
collide with the pore wall This generally happens when the mean free path is larger than the pore
diameter The simulations had been carried out in BTP-COF which accounts a pore diameter of 4 nm
It is found out that the ratio of zero-loading diffusivity with the Knudsen diffusivity has a significant
correlation with isosteric heat of adsorption the stronger the binding energy of the molecules with
the walls the lower the ratio Hydrogen being an exception among the investigated molecules
exhibited the ratio close to unity while others with their isosteric heat of adsorption higher than 10
kJmol-1 have a value close to 01 For a mixture of two gases with significant differences in binding
energies the ratio of self-diffusivities affirms high diffusion selectivity
Selectivity of CH4 over H2 in COF-5 6 and 10 is studied by Keskin106 Narrow pores support the
selective adsorption of CH4 over H2 however they suffer from entropic effects at high pressures
which favor H2 adsorption Liu et al107 have reported that the adsorption selectivity of COFs and
MOFs are in general similar up to 20 bar Delta loading or working capacity (usually expressed in
molkg) is an important term often used to show the economics of the selective adsorption This is
defined as the difference in loadings of the preferred gas at adsorption and desorption pressures
Adsorption is usually in the range 1 to 100 bar while desorption in 01 to 1 bar High selectivity and
high working capacity are preferential for practical use COF-6 has higher selectivity among the three
studied COFs but lower working capacity106 Best selectivity-working capacity combination is shown
by COF-5106 Nonetheless it is not better than CuBTC or CPO-27-X (X = Zn Mg)121122 Liu et al107
attributed the high isosteric heat of adsorption of COF-6 and Cu-BTC for their high adsorption
selectivity They also pointed out that the electrostatic contribution of framework charges in COFs
are smaller than in MOFs however needs to be taken into account
While CH4 is strongly adsorbed H2 undergoes faster diffusion Hence the product of adsorption
selectivity (gt1) and diffusion selectivity (gt0 lt1) which is membrane selectivity is smaller than
adsorption selectivity Keskin106 has compared the selectivity of COF-membranes with known
68
membranes The selectivity of COF-10 is similar to MOF-177 and IRMOFs COF-5 and 6 outperform
them It is to be noted that COF-5 is CH4 selective while COF-10 is H2 selective although their
topologies are similar CH4 permeability of COF membranes is much higher than several MOFs and
ZIFs COF-5 and 6 exhibited high membrane selectivity together with high membrane permeability
Some discussions on the adsorption and membrane based selectivity of COF-102 in comparison with
IRMOFs and Cu-BTC for CH4CO2H2 mixtures can be found in ref108 Adsorption selectivity of COF-6
and Cu-BTC in CH4H2 and CH4CO2 mixtures surpass other COFs and MOFs107 sdfalf
43 Suggestions for improvement
The level of achievement made by COFs and related materials yet do not practically meet the
practical requirements of several applications Hence thoughts for improvement primarily focused
on overcoming their limitations and enhancing characteristic parameters Some theoretical
suggestions for improved performances are mainly discussed here
431 Geometric modifications
Functionalities may be improved by utilizing the structural divergence of framework materials
Several examples may be found in the literature for MOFs etc Klontzas et al suggested replacement
of phenyl linkers in COF-102 by multiple aromatic moieties while keeping the topology unchanged to
increase surface area and pore volume109 They used biphenyl triphenyl naphthalene and pyrene
linkers in the new COFs All of them showed enhanced gravimetric capacity compared to the parent
COF One of the modified COF-102 with triphenyl linker showed 267 and 65 wt at 77 K and 300 K
respectively at 100 bar This kind of replacement may affect the adsorption of each adsorbate
differently leading to the alteration of the selective adsorption of one component over the other110
Following the reticular construction scheme we modeled some new 2D COFs by cross-linking some
of the familiar and unfamiliar building blocks in the COF literature32 This showed the structural
divergence of COFs however they exhibited structural and electronic properties analogues to other
2D COFs
Similar modifications in 2D and 3D COFs for high CO2 uptake were made by Choi et al102 DPABA
(diphenylacetylene-44ʹ-diboronic acid) and its tetrahedral version TBPEPM (tetra[4-(4-
dihydroxyborylphenyl)ethynyl]phenyl) in place of BDBA in COF-5 and TBPMTBPS in COF-108COF-
105 respectively were used for new COF designs The new 2D COF promised a saturated CO2 uptake
higher than in COF-102 because of its large pore volume The new 3D COFs were worth for an uptake
twice more than in COF-105 and 108
69
Goddard et al also designed about 14 new COFs by substituting the aromatic rings in the tetragonal
part (TBPM or TBPS) of COF-102 COF-103 COF-105 COF-108 and COF-202 with vinyl groups or alkyl-
functionalized extended or fused aromatic rings111 The new designs adopted their parent topology
and their structural stability was confirmed by analyzing molecular dynamics (MD) simulations at
room temperature Fused aromatic rings and vinyl groups provided relatively lower and the highest
zero-loading isosteric heat of adsorption (Qst) respectively however the room-temperature delivery
amounts of methane in the respective COFs crossed the DOE target at 35 bar111 In the latter
methane-methane interaction compensated Qst on high-loading
Lino M A et al112 theoretically inspected the ability of layered COFs to exhibit structural variants of
layered materials The hexagonal 2D COF layers may be rolled into the form of nanotubes (NT) or
may exist in the form of fullerenes with the inclusion of pentagons One could think of a formula unit
which resembles the carbon atoms in carbon nanotubes (CNT) or fullerenes The NTs in general can
have any chirality although the study included only armchair and zigzag NTs Density Functional
Theory based calculations reveal that COF-1 NTs of diameter greater than 15 Aring is energetically
favored This was concluded from the comparison of strain-energy per formula unit of the COF-1 NTs
with that of small diameter CNTs The strain energies of zigzag and armchair nanotubes show similar
quadratic functions of tube-diameter The Youngrsquos modulus of COF-1 (22) NT is found out to be 120
GPa This is much smaller compared to typical CNTs which have Youngrsquos modulus average around
1100 GPa113 Band gaps of COF layers remain unchanged when they roll up into nanotubes A COF-1-
fullerene built by scaling C60 molecule has large diameter and hence much less strain energy
compared to C60 Babarao and Jiang reported that the predicted CO2 adsorption capacity of COF-1 NT
is similar to that of CNTs101
Balance between mass density and surface area and hence high gravimetric and volumetric
capacities can be achieved by proper utilization of pore volume or proper distribution of pores Choi
et al theoretically predicted high gravimetric and volumetric capacities for H2 in a pillared COF114 A
pyridine molecule vertically placed above a boron atom in the boroxine ring of COF-1 layer is found
energetically favorable however with wrinkling of the layer115 Here nitrogen atom in pyridine forms
a covalent bond with the boron atom This pillaring increases the separation between the layers
exposes the buried faces of boroxine and aromatic rings to pores and hence increases surface area
and free volume Accessible surface area and free volume have been tripled and gravimetric and
volumetric capacities were calculated to be 10 wt and 60 g L-1 respectively at 77 K and 100 bar114
This volumetric capacity is higher than in NU-100116 and MOF-21093 which show ultrahigh surface
area
70
432 Metal doping
Miao Wu et al studied doping of COF-10 with Li and Ca atoms for hydrogen storage117 The metal
dopants transferred charges to substrate which in turn provided large polarization to hydrogen
molecules Similar observation has been made by Lan et al96 for Li atoms in COF-202 However they
showed the tendency to aggregate at high concentration Lan et al extensively studied doping of
positively charged alkali (Li Na and K) alkaline-earth (Be Mg Ca) and transition (Sc and Ti) metals in
COF-102 and 105 for enhanced CO2 capture103 They found that benzene rings and the three outer
rings of HHTP are the favorite sites for all the metal dopants Other interested locations are edges of
benzene rings and oxygen atoms B3O3 C2O2B rings and the central ring of HHTP were the unwanted
areas Lithium showed stability on the favorite locations while sodium and potassium tended to
cluster even at low concentrations Alkaline-earth metals only weakly interacted with the COFs
whereas transition metals chemisorbed in them which would possibly damage the substrate Lithium
is found out to be a good dopant for enhanced gas storage
Doping electropositive metals would be of advantage because they provide stronger binding with H2
(~ 4 kcalmol)103125133134 The effect of counterions in COFs is studied by Choi et al118 They found out
that although the binding energy of LiF is smaller than Li ion F counterion can hold one hydrogen
atom Additionally Li+ and Mg2+ ions preferred to be isolated than aggregated A step further
Goddard et al119 have recently shown that in presence of tetrahydrofuran (THF a solvent) an
electron is transferred from Li to the benzene ring whereas in the case of gas phase the electron
remained in the atom Additionally they suggested ways to remove solvents which are weakly
coordinated to the material Zhu et al110 suggested doping Li+ after replacement of hydrogen atom by
oxygen ion in COF-102 Another suggestion was the replacement of hydrogen atom by O--Li+ group
Both the cases exhibited higher CO2 adsorption than in the case of Li doping below 10 bar At 10 bar
the three cases exhibited almost the same uptake capacity Below 1 bar the doped Li+ provided
stronger interaction with quadrupolar CO2 in comparison with the substituted O--Li+ group The
differences at low pressures are attributed to the differences in the magnitude of the charge of Li
The doped Li+ remained to hold nearly +1 charge whereas the inclusion of O--Li+ to the framework
diminished the charge of Li+ to a moderate amount Doping Li atom added only relatively small
amount of charge to Li
Doping with Li+ could increase the H2 binding energies up to 28 kJmol118 With properly distributed
metal ion dopants the H2 uptake capacity in COF-108 is predicted to be 65 wt Cao et al also
predicted 684 and 673 wt of H2 uptake in Li-doped COF-105 and 108 respectively at room
temperature and 100 bar120 In Li-doped COF-202 the predicted uptake was 438 wt for the same
conditions96 Goddard et al119 observed that high performance of Li doped COFsMOFs at low
71
pressures (1 ndash 10 bar) is a disadvantage when calculating delivery uptake capacity Na doping could
overcome this difficulty as its heat of adsorption is lower than in Li doped materials Their predicted
delivery gravimetric capacities from 1 to 100 bar at room temperature for Li and Na doped COF-102
and 103 were higher than the 2010 DOE target of 45 wt at room temperature
Lan et al described that CO2 is polarized in presence of Li cation and the polarization increases when
Li cation is physisorbed in COF103 Their calculated uptake capacities of lithium-doped COF-102 and
COF-105 at low loading is about eight and four times higher than that of nondoped ones respectively
Also a very high uptake of 2266 mgg was predicted in the doped COF-105 at 40 bar Li-doped COF-
102 and COF-103 also showed higher uptakes of methane ndash almost double the amount ndash compared
to nondoped COFs at room temperature and low pressures (lt 50 bar)121 Binding energy between
doped Li cation and CH4 was calculated to be 571 kcalmol
Li-functionalized COF-105 realized by inserting alkoxide groups provided very high volumetric uptake
of H2 Klontzas et al122 replaced three hydrogen atoms in HHTP by oxygen atoms in order to achieve
the functionalization In spite of the increased weight due to the additional oxygen atoms the COF
exhibited gravimetric capacity of 6 wt at 300 K
Incorporation of lithium tetrazolide group into a new PAF having diamond topology and triphenyl
linkers for enhanced hydrogen adsorption is suggested by Sun et al123 The tetrazolide anion (CHN4-)
interacts with the lithium cation via ionic bonds The anion and the cation together can bind upto 14
hydrogen molecules Two lithium tetrazolide groups were introduced in the middle phenyl ring of
each triphenyl linker and GCMC calculations predicted 45 wt for the network at 233 K and 100 bar
With the intention of increasing the H2 binding energy Li et al chemically implanted metals in the
place of light atoms in COF-108124 They replaced the C2O2B rings in COF-108 by C2O2Al C2N2Al and
C2Mg2N and each new COF showed stability within DFT This kind of doping does not allow
aggregation of metal clusters Their GCMC simulations showed that the replacement strategy could
improve the hydrogen storage capacity at room temperature by a factor of up to three124 Zou et al
suggested that para-substitution of two carbon atoms in benzene rings by two boron atoms can
facilitate charge transfer between the rings and metal dopants125 Their work revealed that
dispersion of metals forbid any clustering and the doping could enhance the H2 uptake capacity
significantly
433 Functionalization
Functionalization of porous frameworks with ether groups for selective CO2 capture is suggested by
Babarao et al126 The electropositive C atom of CO2 interacts mainly with the electronegative O or N
72
atom of the functional groups They studied PAF-1 functionalized with ndashOCH3 ndashNH2 and ndashCH2OCH2ndash
groups in its biphenyl linkers The latter one which is the tetrahydrofuran-like ether-functionalized
PAF-1 showed high adsorption capacity for CO2 and high selectivity for CO2CH4 CO2N2 and CO2H2
mixtures at ambient conditions
5 Conclusions
Successful covalent crystallization resulted as a class of materials Covalent Organic Frameworks This
review portrays different synthetic schemes successful realizations and potential applications of
COFs and related materials The growth in this area is relatively slow and thus promotions are
needed in order to achieve its potential
6 List and pictures of chemical compounds
alkyl-THB Alkyl-1245-tetrahydroxybenzene
BDBA 14-benzenediboronic acid
BPDA 44ʹ-biphenyldiboronic acid
BTBA 135-benzene triboronic acid
BTDADA 14-benzothiadiazole diboronic acid
BTPA 135-benzenetris(4-phenylboronic acid)
CA Cyanuric acid
DAB 14-diaminobenzene
DCB 14-dicyanobenzene
DCF 27-diisocyanate fluorine
DCN 26-dicyanonaphthalene
DCP 26-dicyanopyridine
DETH 25-diethoxyterephthalohydrazole
DMBPDC 33ʹ-dimethoxy-44ʹ-biphenylene diisocyanate
DPB Diphenyl butadyenediboronic acid
73
HP 1-hexyne propiolate
HHTP 23671011-hexahydroxytriphenylene
MP Methyl propiolate
N3-BDBA Azide-appended benzenediboronic acid
NDI Naphthalenediimide diboronic acid
NiPcTA Nickel-phthalocyanice tetrakis(acetonide)
OH-Pc-Ni 2391016172324-octahydroxyphthalocyaninato)nickel(II)
OH-Pc-Zn 2391016172324-octahydroxyphthalocyaninato)zinc
PA Piperazine
Pac 2-propenyl acetate
PcTA Phthalocyanine tetra(acetonide)
PdAc Palladium acetate
PDBA Pyrenediboronic acid
PPE Phenylbis(phenylethynyl) diboronic acid
PPP 3-phenyl-1-propyne propiolate
PyMP (3α13α2-dihydropyren-1-yl)methyl propionate
TA Terephthaldehyde
TAM tetra-(4-anilyl)methane
TAPP Tetra(p-amino-phneyl)porphyrin
TBB 135-tris(4-bromophenyl)benzene
TBPM tetra(4-dihydroxyboryl-phenyl)methane
TBPP Tetra(p-boronic acid-phenyl)porphyrin
TBPS tetra(4-dihydroxyboryl-phenyl)silane
TBST tert-butylsilane triol
74
TCM Tetrakis(4-cyanophenyl)methane
TDHB-ZnP Zinc(II) 5101520-tetrakis(4-(dihydroxyboryl)phenyl)porphyrin
TFB 135-triformylbenzene
TFPB 135-tris-(4-formyl-phenyl)-benzene
THAn 2345-Tetrahydroxy anthracene
THB 1245-tetrahydroxybenzene
THDMA Polyol 2367-tetrahydroxy-910-dimethyl-anthracene
TkBPM Tetrakis(4-bromophenyl)methane
TPTA Triphenylene tris(acetonide)
trunc-TBPM-R R-functionalized truncated TBPM
75
Figure 8 Reactants of Covalently-bound Organic Frameworks
76
Figure 9 (Figure 8 continued)
(1) Morris R E Wheatley P S Angewandte Chemie-International Edition 2008 47 4966 (2) Li J-R Kuppler R J Zhou H-C Chemical Society Reviews 2009 38 1477 (3) Corma A Chemical Reviews 1997 97 2373 (4) Seo J S Whang D Lee H Jun S I Oh J Jeon Y J Kim K Nature 2000 404 982 (5) Yaghi O M OKeeffe M Ockwig N W Chae H K Eddaoudi M Kim J Nature 2003 423 705
77
(6) OKeeffe M Peskov M A Ramsden S J Yaghi O M Accounts of Chemical Research 2008 41 1782 (7) Ockwig N W Delgado-Friedrichs O OKeeffe M Yaghi O M Accounts of Chemical Research 2005 38 176 (8) Li H Eddaoudi M OKeeffe M Yaghi O M Nature 1999 402 276 (9) Chen B L Eddaoudi M Hyde S T OKeeffe M Yaghi O M Science 2001 291 1021 (10) Eddaoudi M Moler D B Li H L Chen B L Reineke T M OKeeffe M Yaghi O M Accounts of Chemical Research 2001 34 319 (11) Eddaoudi M Kim J Rosi N Vodak D Wachter J OKeeffe M Yaghi O M Science 2002 295 469 (12) Chae H K Siberio-Perez D Y Kim J Go Y Eddaoudi M Matzger A J OKeeffe M Yaghi O M Nature 2004 427 523 (13) Furukawa H Kim J Ockwig N W OKeeffe M Yaghi O M Journal of the American Chemical Society 2008 130 11650 (14) Smaldone R A Forgan R S Furukawa H Gassensmith J J Slawin A M Z Yaghi O M Stoddart J F Angewandte Chemie-International Edition 2010 49 8630 (15) Eddaoudi M Kim J Wachter J B Chae H K OKeeffe M Yaghi O M Journal of the American Chemical Society 2001 123 4368 (16) Sudik A C Millward A R Ockwig N W Cote A P Kim J Yaghi O M Journal of the American Chemical Society 2005 127 7110 (17) Sudik A C Cote A P Wong-Foy A G OKeeffe M Yaghi O M Angewandte Chemie-International Edition 2006 45 2528 (18) Tranchemontagne D J L Ni Z OKeeffe M Yaghi O M Angewandte Chemie-International Edition 2008 47 5136 (19) Lu Z Knobler C B Furukawa H Wang B Liu G Yaghi O M Journal of the American Chemical Society 2009 131 12532 (20) Park K S Ni Z Cote A P Choi J Y Huang R Uribe-Romo F J Chae H K OKeeffe M Yaghi O M Proceedings of the National Academy of Sciences of the United States of America 2006 103 10186 (21) Hayashi H Cote A P Furukawa H OKeeffe M Yaghi O M Nature Materials 2007 6 501 (22) Banerjee R Furukawa H Britt D Knobler C OKeeffe M Yaghi O M Journal of the American Chemical Society 2009 131 3875 (23) Cote A P Benin A I Ockwig N W OKeeffe M Matzger A J Yaghi O M Science 2005 310 1166 (24) El-Kaderi H M Hunt J R Mendoza-Cortes J L Cote A P Taylor R E OKeeffe M Yaghi O M Science 2007 316 268 (25) Wan S Guo J Kim J Ihee H Jiang D Angewandte Chemie-International Edition 2008 47 8826 (26) Wan S Guo J Kim J Ihee H Jiang D L Angewandte Chemie-International Edition 2009 48 5439 (27) Cote A P El-Kaderi H M Furukawa H Hunt J R Yaghi O M Journal of the American Chemical Society 2007 129 12914 (28) Hunt J R Doonan C J LeVangie J D Cote A P Yaghi O M Journal of the American Chemical Society 2008 130 11872 (29) Uribe-Romo F J Hunt J R Furukawa H Klock C OKeeffe M Yaghi O M Journal of the American Chemical Society 2009 131 4570 (30) Klontzas E Tylianakis E Froudakis G E Journal of Physical Chemistry C 2008 112 9095 (31) Tylianakis E Klontzas E Froudakis G E Nanotechnology 2009 20 (32) Lukose B Kuc A Frenzel J Heine T Beilstein Journal of Nanotechnology 2010 1 60
78
(33) Tilford R W Gemmill W R zur Loye H C Lavigne J J Chemistry of Materials 2006 18 5296 (34) Spitler E L Dichtel W R Nature Chemistry 2010 2 672 (35) Spitler E L Giovino M R White S L Dichtel W R Chemical Science 2011 2 1588 (36) Campbell N L Clowes R Ritchie L K Cooper A I Chemistry of Materials 2009 21 204 (37) Ding X Guo J Feng X Honsho Y Guo J Seki S Maitarad P Saeki A Nagase S Jiang D Angewandte Chemie-International Edition 2011 50 1289 (38) Feng X A Chen L Dong Y P Jiang D L Chemical Communications 2011 47 1979 (39) Zwaneveld N A A Pawlak R Abel M Catalin D Gigmes D Bertin D Porte L Journal of the American Chemical Society 2008 130 6678 (40) Gutzler R Walch H Eder G Kloft S Heckl W M Lackinger M Chemical Communications 2009 4456 (41) Blunt M O Russell J C Champness N R Beton P H Chemical Communications 2010 46 7157 (42) Sassi M Oison V Debierre J-M Humbel S Chemphyschem 2009 10 2480 (43) Ourdjini O Pawlak R Abel M Clair S Chen L Bergeon N Sassi M Oison V Debierre J-M Coratger R Porte L Physical Review B 2011 84 (44) Colson J W Woll A R Mukherjee A Levendorf M P Spitler E L Shields V B Spencer M G Park J Dichtel W R Science 2011 332 228 (45) Berlanga I Ruiz-Gonzalez M L Gonzalez-Calbet J M Fierro J L G Mas-Balleste R Zamora F Small 2011 7 1207 (46) Wan S Gandara F Asano A Furukawa H Saeki A Dey S K Liao L Ambrogio M W Botros Y Y Duan X Seki S Stoddart J F Yaghi O M Chemistry of Materials 2011 23 4094 (47) Ding X Chen L Honsho Y Feng X Saenpawang O Guo J Saeki A Seki S Irle S Nagase S Parasuk V Jiang D Journal of the American Chemical Society 2011 133 14510 (48) Tilford R W Mugavero S J Pellechia P J Lavigne J J Advanced Materials 2008 20 2741 (49) Lanni L M Tilford R W Bharathy M Lavigne J J Journal of the American Chemical Society 2011 133 13975 (50) Li Y Yang R T Aiche Journal 2008 54 269 (51) Nagai A Guo Z Feng X Jin S Chen X Ding X Jiang D Nature Communications 2011 2 (52) Bunck D N Dichtel W R Angewandte Chemie-International Edition 2012 51 1885 (53) Ding S-Y Gao J Wang Q Zhang Y Song W-G Su C-Y Wang W Journal of the American Chemical Society 2011 133 19816 (54) Miyaura N Suzuki A Chemical Reviews 1995 95 2457 (55) Kalidindi S B Yusenko K Fischer R A Chemical Communications 2011 47 8506 (56) Kuhn P Antonietti M Thomas A Angewandte Chemie-International Edition 2008 47 3450 (57) Bojdys M J Jeromenok J Thomas A Antonietti M Advanced Materials 2010 22 2202 (58) Kuhn P Forget A Su D Thomas A Antonietti M Journal of the American Chemical Society 2008 130 13333 (59) Ren H Ben T Wang E Jing X Xue M Liu B Cui Y Qiu S Zhu G Chemical Communications 2010 46 291 (60) Zhang W Li C Yuan Y-P Qiu L-G Xie A-J Shen Y-H Zhu J-F Journal of Materials Chemistry 2010 20 6413 (61) Trewin A Cooper A I Angewandte Chemie-International Edition 2010 49 1533 (62) Mastalerz M Angewandte Chemie-International Edition 2008 47 445
79
(63) Chan-Thaw C E Villa A Katekomol P Su D Thomas A Prati L Nano Letters 2010 10 537 (64) Palkovits R Antonietti M Kuhn P Thomas A Schueth F Angewandte Chemie-International Edition 2009 48 6909 (65) Ben T Ren H Ma S Q Cao D P Lan J H Jing X F Wang W C Xu J Deng F Simmons J M Qiu S L Zhu G S Angewandte Chemie-International Edition 2009 48 9457 (66) Yamamoto T Bulletin of the Chemical Society of Japan 1999 72 621 (67) Zhou G Baumgarten M Muellen K Journal of the American Chemical Society 2007 129 12211 (68) Yuan Y Sun F Ren H Jing X Wang W Ma H Zhao H Zhu G Journal of Materials Chemistry 2011 21 13498 (69) Ren H Ben T Sun F Guo M Jing X Ma H Cai K Qiu S Zhu G Journal of Materials Chemistry 2011 21 10348 (70) Zhao H Jin Z Su H Jing X Sun F Zhu G Chemical Communications 2011 47 6389 (71) Mortera R Fiorilli S Garrone E Verne E Onida B Chemical Engineering Journal 2010 156 184 (72) Dogru M Sonnauer A Gavryushin A Knochel P Bein T Chemical Communications 2011 47 1707 (73) Spitler E L Koo B T Novotney J L Colson J W Uribe-Romo F J Gutierrez G D Clancy P Dichtel W R Journal of the American Chemical Society 2011 133 19416 (74) Zhang Y Tan M Li H Zheng Y Gao S Zhang H Ying J Y Chemical Communications 2011 47 7365 (75) Uribe-Romo F J Doonan C J Furukawa H Oisaki K Yaghi O M Journal of the American Chemical Society 2011 133 11478 (76) Ben T Pei C Zhang D Xu J Deng F Jing X Qiu S Energy amp Environmental Science 2011 4 3991 (77) Lukose B Kuc A Heine T Chemistry-a European Journal 2011 17 2388 (78) Zhou W Wu H Yildirim T Chemical Physics Letters 2010 499 103 (79) Amirjalayer S Snurr R Q Schmid R Journal of Physical Chemistry C 2012 116 4921 (80) Xu Q Zhong C Journal of Physical Chemistry C 2010 114 5035 (81) Lukose B Supronowicz B St Petkov P Frenzel J Kuc A B Seifert G Vayssilov G N Heine T Physica Status Solidi B-Basic Solid State Physics 2012 249 335 (82) Assfour B Seifert G Chemical Physics Letters 2010 489 86 (83) Zhao L Zhong C L Journal of Physical Chemistry C 2009 113 16860 (84) Schmid R Tafipolsky M Journal of the American Chemical Society 2008 130 12600 (85) Han S S Goddard W A III Journal of Physical Chemistry C 2007 111 15185 (86) Suh M P Park H J Prasad T K Lim D-W Chemical Reviews 2012 112 782 (87) Furukawa H Yaghi O M Journal of the American Chemical Society 2009 131 8875 (88) Wong-Foy A G Matzger A J Yaghi O M Journal of the American Chemical Society 2006 128 3494 (89) Mendoza-Cortes J L Han S S Furukawa H Yaghi O M Goddard III W A Journal of Physical Chemistry A 2010 114 10824 (90) Doonan C J Tranchemontagne D J Glover T G Hunt J R Yaghi O M Nature Chemistry 2010 2 235 (91) Getman R B Bae Y-S Wilmer C E Snurr R Q Chemical Reviews 2012 112 703 (92) Han S S Furukawa H Yaghi O M Goddard III W A Journal of the American Chemical Society 2008 130 11580 (93) Furukawa H Ko N Go Y B Aratani N Choi S B Choi E Yazaydin A O Snurr R Q OKeeffe M Kim J Yaghi O M Science 2010 329 424 (94) Garberoglio G Langmuir 2007 23 12154 (95) Assfour B Seifert G Microporous and Mesoporous Materials 2010 133 59
80
(96) Lan J Cao D Wang W Journal of Physical Chemistry C 2010 114 3108 (97) Yang Q Zhong C Langmuir 2009 25 2302 (98) Garberoglio G Vallauri R Microporous and Mesoporous Materials 2008 116 540 (99) Lan J H Cao D P Wang W C Ben T Zhu G S Journal of Physical Chemistry Letters 2010 1 978 (100) Furukawa H Miller M A Yaghi O M Journal of Materials Chemistry 2007 17 3197 (101) Babarao R Jiang J Energy amp Environmental Science 2008 1 139 (102) Choi Y J Choi J H Choi K M Kang J K Journal of Materials Chemistry 2011 21 1073 (103) Lan J Cao D Wang W Smit B Acs Nano 2010 4 4225 (104) Wang L Yang R T Energy amp Environmental Science 2008 1 268 (105) Krishna R van Baten J M Industrial amp Engineering Chemistry Research 2011 50 7083 (106) Keskin S Journal of Physical Chemistry C 2012 116 1772 (107) Liu Y Liu D Yang Q Zhong C Mi J Industrial amp Engineering Chemistry Research 2010 49 2902 (108) Keskin S Sholl D S Langmuir 2009 25 11786 (109) Klontzas E Tylianakis E Froudakis G E Nano Letters 2010 10 452 (110) Zhu Y Zhou J Hu J Liu H Hu Y Chinese Journal of Chemical Engineering 2011 19 709 (111) Mendoza-Cortes J L Pascal T A Goddard W A III Journal of Physical Chemistry A 2011 115 13852 (112) Lino M A Lino A A Mazzoni M S C Chemical Physics Letters 2007 449 171 (113) Krishnan A Dujardin E Ebbesen T W Yianilos P N Treacy M M J Physical Review B 1998 58 14013 (114) Kim D Jung D H Kim K-H Guk H Han S S Choi K Choi S-H Journal of Physical Chemistry C 2012 116 1479 (115) Kim D Jung D H Choi S-H Kim J Choi K Journal of the Korean Physical Society 2008 52 1255 (116) Farha O K Yazaydin A O Eryazici I Malliakas C D Hauser B G Kanatzidis M G Nguyen S T Snurr R Q Hupp J T Nature Chemistry 2010 2 944 (117) Wu M M Wang Q Sun Q Jena P Kawazoe Y Journal of Chemical Physics 2010 133 (118) Choi Y J Lee J W Choi J H Kang J K Applied Physics Letters 2008 92 (119) Mendoza-Cortes J L Han S S Goddard W A III Journal of Physical Chemistry A 2012 116 1621 (120) Cao D Lan J Wang W Smit B Angewandte Chemie-International Edition 2009 48 4730 (121) Lan J H Cao D P Wang W C Langmuir 2010 26 220 (122) Klontzas E Tylianakis E Froudakis G E Journal of Physical Chemistry C 2009 113 21253 (123) Sun Y Ben T Wang L Qiu S Sun H Journal of Physical Chemistry Letters 2010 1 2753 (124) Li F Zhao J Johansson B Sun L International Journal of Hydrogen Energy 2010 35 266 (125) Zou X Zhou G Duan W Choi K Ihm J Journal of Physical Chemistry C 2010 114 13402 (126) Babarao R Dai S Jiang D-e Langmuir 2011 27 3451
81
Appendix B
Structural properties of metal-organic frameworks within the density-functional based tight-binding method
Binit Lukose Barbara Supronowicz Petko St Petkov Johannes Frenzel Agnieszka Kuc
Gotthard Seifert Georgi N Vayssilov and Thomas Heine
Phys Status Solidi B 2012 249 335ndash342
DOI 101002pssb201100634
Abstract Density-functional based tight-binding (DFTB) is a powerful method to describe large
molecules and materials Metal-organic frameworks (MOFs) materials with interesting catalytic
properties and with very large surface areas have been developed and have become commercially
available Unit cells of MOFs typically include hundreds of atoms which make the application of
standard density-functional methods computationally very expensive sometimes even unfeasible
The aim of this paper is to prepare and to validate the self-consistent charge-DFTB (SCC-DFTB)
method for MOFs containing Cu Zn and Al metal centers The method has been validated against
full hybrid density-functional calculations for model clusters against gradient corrected density-
functional calculations for supercells and against experiment Moreover the modular concept of
MOF chemistry has been discussed on the basis of their electronic properties We concentrate on
MOFs comprising three common connector units copper paddlewheels (HKUST-1) zinc oxide Zn4O
tetrahedron (MOF-5 MOF-177 DUT-6 (MOF-205)) and aluminum oxide AlO4(OH)2 octahedron (MIL-
53) We show that SCC-DFTB predicts structural parameters with a very good accuracy (with less than
82
5 deviation even for adsorbed CO and H2O on HKUST-1) while adsorption energies differ by 12 kJ
mol1 or less for CO and water compared to DFT benchmark calculations
1 Introduction In reticular or modular chemistry molecular building blocks are stitched together to
form regular frameworks [1] With this concept it became possible to construct framework
compounds with interesting structural and chemical composition most notably metal-organic
frameworks (MOFs) [1ndash10] and covalent organic frameworks (COFs) [11ndash17] The interest in MOFs
and COFs is not limited to chemistry these crystalline materials are also interesting for applications
in information storage [18] sieving of quantum liquids [19] hydrogen storage [20] and for fuel cell
membranes [21ndash23]
Covalent organic framework and MOF frameworks are composed by combining two types of building
blocks the so-called connectors typically coordinating in four to eight sites and linkers which have
typically 2 sometimes also three or four connecting sites Fig 1 illustrates a simplified representation
of the topology of connectors and linkers by using secondary building units (SBUs) (see Fig 1)
Figure 1 The connector and the linker of the frameworks CuBTC (HKUST-1) MOF-177 MOF-5 DUT-6 (MOF-205) and MIL-53 and their respective secondary building units (SBUs) The arrows indicate the connection points of the fragments Red oxygen green carbon white hydrogen yellow copperzinc and blue aluminum
Linkers are organic molecules with carboxylic acid groups at their connection sites which form
bonds to the connectors (typically in a solvothermal condensation reaction) They can carry
functional groups which can make them interesting for applications in catalysis [24] Connectors are
83
either metal organic units as for example the well-known Zn4O(CO2)6 unit of MOF-5 [8] the
Cu2(HCOO)4 paddle wheel unit of HKUST-1 [10] or covalent units incorporating boron oxide linking
units for COFs [11] As the building blocks of MOFs and COFs comprise typically some ten atoms unit
cells quickly get very large In particular if adsorption or dynamic processes in MOFs and COFs are of
interest (super)cells of some 1000 atoms need to be processed While standard organic force fields
show a reasonable performance for COFs [25] the creation of reliable force fields is not
straightforward for MOFs as transferable parameterization of the transition metal sites is an issue
even though progress has been achieved for selected materials [26 27] The difficulty to describe
transition metals in particular if they are catalytically active as in HKUST-1 limits the applicability of
molecular mechanics (MM) even for QMMM hybrid methods [28]
On the other hand the density-functional based tight-binding (DFTB) method with its self-consistent
charge (SCC) extension to improve performance for polar systems is a computationally feasible
alternative This non-orthogonal tight-binding approximation to density-functional theory (DFT)
which has been developed by Seifert Frauenheim and Elstner and their groups [29ndash32] (for a recent
review see Ref [30]) has been successfully applied to a large-scale systems such as biological
molecules [33ndash36] supramolecular systems [37 38] surfaces [39 40] liquids and alloys [41 42] and
solids [43] Being a quantum-mechanical method and hence allowing the description of breaking and
formation of chemical bonds the method showed outstanding performance in the description of
processes such as the mechanical manipulation of nanomaterials [44ndash46] It is remarkable that the
method performs well for systems containing heavier elements such as transition metals as this
domain cannot be covered so far with acceptable accuracy by traditional semi-empirical methods [47
48] DFTB covers today a large part of the elements of the periodic table and parameters and a
computer code are available from the DFTBorg website
Recently we have validated DFTB for its application for COFs [16 17] and have shown that structural
properties and formation energies of COFs are well described within DFTB Kuc et al [49] have
validated DFTB for substituted MOF-5 frameworks where connectors are always the Zn4O(CO2)6 unit
which has been combined with a large variety of organic linkers In this work we have revised the
DFTB parameters developed for materials science applications and validated them for HKUST-1 and
being far more challenging for the interaction of its catalytically active Cu sites with carbon
monoxide (CO) and water The Cu2(HCOO)4 paddle wheel units are electronically very intriguing on a
first note the electronic ground state of the isolated paddle wheel is an antiferromagnetic singlet
state which cannot be described by one Slater determinant and which is consequently not accessible
for KohnndashSham DFT However the energetically very close triplet state correctly describes structure
and electronic density of the system and also adsorption properties agree well with experiment [32
84
50 51] We therefore use DFT in the B3LYP hybrid functional representation as benchmark for DFTB
validations for HKUST-1 added by DFT-GGA calculations for periodic unit cellsWewill show that the
general transferability of the DFTB method will allow investigating structural electronic and in
particular dynamic properties
2 Computational details All calculations of the finite model and periodic crystal structures of MOFs
were carried out using the dispersion-corrected self-consistent density functional based tight-binding
(DC-SCC-DFTB in short DFTB) method [30 52ndash54] as implemented in the deMonNano code [55] Two
sets of parameters have been used the standard SCC-DFTB parameter set developed by Elstner et al
[53] for Zn-containing MOFs for Cu- and Al-containing materials we have extended our materials
science parameter set which has been developed originally for zeolite materials to include Cu For
this element we have used the standard procedure of parameter generation we have used the
minimal atomic valence basis for all atoms including polarization functions when needed Electrons
below the valence states were treated within the frozen-core approximation The matrix elements
were calculated using the local density approximation (LDA) while the short-range repulsive pair-
potential was fitted to the results from DFT generalized gradient approximation (GGA) calculations
For more details on DFTB parameter generation see Ref [30] Periodic boundary conditions were
used to represent frameworks of the crystalline solid state The conjugate-gradient scheme was
chosen for the geometry optimization An atomic force tolerance of 3x10-4 eV Aring-1 was applied
The reference DFT calculations of the equilibrium geometry of the investigated isolated MOF models
were performed employing the Becke three-parameter hybrid method combined with a LYP
correlation functional (B3LYP) [56 57] The basis set associated with the HayndashWadt [58] relativistic
effective core potentials proposed by Roy et al [59] was supplemented with polarization ffunctions
[60] and was employed for description of the electronic structure of Cu atoms The Pople 6-311G(d)
basis sets were applied for the H C and O atoms The calculations were performed with the
Gaussian09 program suite [61] For optimization of the periodic structure of CuBTC at DFT level the
electronic structure code Quickstep [62] which is a part of the CP2K package [63] was used
Quickstep is an implementation of the Gaussian plane wave method [64] which is based on the
KohnndashSham formulation of DFT
We have used the generalized gradient functional PBE [65] and the GoedeckerndashTeterndashHutter
pseudopotentials [64 66] in conjunction with double-z basis sets with polarization functions DZVP-
MOLOPT-SR-GTH basis sets obtained as described in Ref [67] but using two less diffuse primitives
Due to the large unit cell sampling of Brillouin zone was limited only to the G-point Convergence
85
criterion for the maximum force component was set to 45 x 10-3 Hartree Bohr1 The plane wave
basis with cutoff energy of 400 Ry was used throughout the simulations
The initial crystal structures were taken from experiment (powder X-ray diffraction (PXRD) data) The
cell parameters and the atomic positions were fully optimized using conjugate-gradient method at
the DFTB level For the reference DFT calculations only the atomic positions of experimental crystal
structures were minimized The cluster models were cut from the optimized structures and saturated
with hydrogen atoms (see Fig 2 for exemplary systems of Cu-BTC)
3 Results and discussion
31 Cu-BTC We have optimized the crystal structure of Cu-BTC (HKUST-1) [10] using cluster and the
periodic models The structural properties were compared to DFT results (see Table 1) The
geometries were obtained for the activated form (open metal sites) and in the presence of axial
water ligands (see Table 2) as the synthesized crystals often have H2O molecules attached to the
open metal sites of Cu (Fig 2b) We have also studied changes of the cluster model geometry in the
presence of CO (Fig 2c) and the results are also shown in Table 2 We have obtained good agreement
with experimental data as well as with DFT results
Figure 2 (a) Periodic and cluster models of Cu-BTC MOF as used in the computer simulations (b) cluster model with adsorbed H2O and (c) CO molecules
We have also simulated the XRD patterns of Cu-BTC which show very good structural agreement for
peak positions between the experimental and calculated structures The XRD pattern of DFT
optimized structure is nearly a copy of that of the experimental geometry However for DFTB
optimized patterns the relative intensities of some of the peaks notably that of the peaks at 2θ =
138 and 2θ = 158 are different from the experimental XRDs (see Fig 3) The differences in bond
angles between simulation and experiment may affect the sharpness of the signals and hence the
86
intensity To support this argument we have calculated the radial pair distribution function (g(r))
which details the probabilities of all atomndashatom distances in a given cell volume Figure S3 in the
Supporting Information shows g(r) of CundashCu CundashC CundashO CndashC and CndashO atom pairs of DFTB
optimized DFT optimized and experimentally modeled Cu-BTC unit cells The peaks of DFT as well as
DFTB optimized geometries are much broadened whereas the experimentally modeled geometry
has single peaks at the most predicted atomndashatom distances However it is to be noted that DFTB
optimized geometries give slightly shifted peaks compared to those of the DFT optimized geometry
They are broadened around the experimental values The distances between Cu and C atoms which
are not direct neighbors have the largest deviations from the experiment what indicates
shortcomings of the bond angles
Table 1 Selected bond lengths (Aring ) bond angles (⁰) and dihedral angles (⁰) of Cu-BTC optimized at DFTB and DFT level
Bond Type Cluster Model Periodic Model Exp
Cu-Cu 250 (257) 250 (250) 263 Cu-O 205 (198) 202 (198) 195 O-C 134 (133) 133-138 (128) 125
OCuO 836-971 (898) 892-907 (873-937)
891 896
Cu-O-O-Cu plusmn56 ˗ plusmn62 (plusmn02) plusmn04 (plusmn47 ˗ plusmn131) 0
O-C-C-C plusmn38 - plusmn50 (0) plusmn03 - plusmn05 (plusmn06 - plusmn105) 063
Cell paramet a=b=c=27283 (26343)
α=β=γ=90 (90) a=b=c=26343
α=β=γ=90
In detail the bond lengths and bond angles do not change significantly when going from the cluster
to the periodic model confirming the reticular chemistry approach The only exception is the OndashCundash
O bond angle that differs by 4ndash78 between the two systems at both levels of theory
87
Figure 3 Simulated XRD patterns of Cu-BTC MOF with the experimental unit cell parameters and optimized at the DFTB and DFT level of theory
The bond length between Cu atoms is slightly underestimated comparing with experimental (by
maximum 5) and DFT (maximum by 3) results while all other bond lengths are somewhat larger
at DFTB
All bond lengths stay unchanged or become longer in the presence of water molecules The most
striking example is the CundashCu bond where the change reaches 012ndash017 Aring compared with the
structure with activated Cu sites Also bond angles increase by up to 28 when the H2O is present
The CundashO bond length with the oxygen atom from the carboxylate groups is shorter than that with
the oxygen atoms from the water molecules indicating that the latter is only weakly bound to the
copper ions The OndashC distances (~133 Aring) correspond to values between that for the typical single
(142 Aring ) and double (122 Aring) oxygenndashcarbon bond The calculated CringndashCcarboxyl bond lengths of
146A deg indicate that these are typical single sp2ndashsp2 CndashC bonds while experimental findings give a
slightly longer value (15 Aring) for this MOF When comparing dihedral angles CundashOndashOndashCu and OndashCndashCndashC
of the clusters fairly large deviations between DFTB and DFT calculations and experiment are visible
due to the softer potential energy surface associated with these geometrical parameters In periodic
models however the agreement of DFT and DFTB with experiment and with each other is
significantly improved
The unit cell parameters with and without water molecules obtained at the DFTB level overestimate
the experimental data by less than 4 which gives a fairly good agreement if we take into account
high porosity of the material These lattice parameters increase only slightly from 27283 to 27323 Aring
in the presence of water
We have calculated the binding energies of H2O molecules attached to the Cu-metal sites within the
cluster model At the DFTB level Ebind of 40 kJ mol-1 was obtained in a good agreement with the DFT
results (52 kJ mol-1) as well as results from the literature (47 kJ mol-1 [50]) We have also calculated
88
the binding energy of CO which was twice smaller than that of H2O (165 and 28 kJ mol-1 for DFTB
and DFT respectively) suggesting much stronger binding of O atom (from H2O) than C atom (from
CO) While the bond lengths in the MOF structure do not change in the presence of either H2O or CO
the differences in the binding energy come from much longer bond distances (by around 07 Aring) for
CundashC than for CundashO in the presence of CO and water molecules respectively
Furthermore we have studied the electronic properties of periodic and cluster model Cu-BTC by
means of partial density-of-states (PDOS)We have compared especially the Cu-PDOS for s- p- and d-
orbitals (see Fig 4) The results show that the electronic structure stays unchanged when going from
the cluster models to the fully periodic crystal of Cu-BTC Since the copper atoms are of Cu(II) the d-
orbitals are just partially occupied what results in the metallic states at the Fermi level This is a very
interesting result as other ZnndashO-based MOFs are either semiconductors or insulators [49]
Table 2 Selected bond lengths (Aring) bond angles (˚) and dihedral angles (˚) of Cu-BTC optimized at DFTB and DFT levels with water molecules in the axial positions Cluster models are calculated using water and carbon monoxide molecules The adsorption energies Eads (kJ mol-1) of the guest molecules are given as well The DFT data are given in parenthesis
Bond Type Cluster Model +
H2O Periodic
Model+ H2O Cluster Model +
CO
Cu-Cu 267 (266) 262 (260) 250 (260)
Cu-O 205 (197-206) 210 (196-200) 206 (199)
O-C 134 (127) 133 (128) 134 (127)
OCuO 843-955 (889-905)
871-921 (842-930) 842-967 (896)
Cu-O-O-Cu plusmn59 - plusmn76 (plusmn06 ˗ plusmn10)
plusmn02 ˗ plusmn18 (plusmn40 ˗ plusmn136)
plusmn51 - plusmn58 (plusmn70)
O-C-C-C plusmn39 - plusmn53 (plusmn05 - plusmn11)
plusmn03 - plusmn05 (plusmn06 - plusmn105)
plusmn35 - plusmn43 (plusmn12)
Cu-O(H2O) Cu-C(CO) 237 (219) 244 (233-
255) 307 (239)
Eads -4045 (-5200) -1648
(-2800)
32 MOF-5 -177 DUT-6 (MOF-205) and MIL- 53 We have also studied the structural properties
of MOF structures with large surface areas using the SCC-DFTB method Very good agreement with
the experimental data shows that this method is applicable for MOFs of large structural diversity as
well as for coordination polymers based on the MOF-5 framework which has been reported earlier
[49] Tables 3 and 4 show selected bond lengths and bond angles of MOF-5 [68] MOF-177 [69] DUT-
6 (MOF-205) [70 71] and MIL-53 [72] respectively
89
MOFs-5 -177 and DUT-6 (MOF-205) are built of the same connector which is Zn4O(CO2)6
octahedron The difference is in the organic linker 14-benzenedicarboxylate (BDC) 135-
benzenetribenzoates (BTB) and BTB together with 26-naphtalenedicarboxylate (NDC) for MOF-5 -
177 andDUT-6 (MOF-205) respectively (see Fig 5)
Figure 4 DFTB calculations of PDOS of (top left) Cu in Cu-BTC and Zn in MOF-5 (top right) MOF-177 (bottom left) and DUT-6 (MOF-205) (bottom right) In each plot s-orbitals (top) p-orbitals (middle) and d-orbitals (bottom) are shown as computed from periodic (black) and cluster (red) models Cluster models are given in Fig S4
All three MOFs have different topologies due to the organic linkers where the number of
connections is varied or where two different linker types are present MOF-5 is the most simple and
it is the prototype in the isoreticular series (IRMOF-1) [68] it is a simple cubic topology with
threedimensional pores of the same size and the linkers have only two connection points In the
case of MOF-177 the linker is represented by a triangularSBU that means three connection points
are present This results in the topology with mixed (36)-connectivity [69] DUT-6 (MOF-205) has a
much more complicated topology due to two types of linkers The first one (NDC) has just two
90
connection points while the second is the same as in MOF-177 with three connection points One
ZnndashO-based connector is connected to two NDC and four BTB linkers The topology is very interesting
all rings of the underlying nets are fivefold and it forms a face-transitive tiling of dodectahedra and
tetrahedra with a ratio of 13 [73]
Figure 5 The crystal structures in the unit cell representations of (a) MOF-5 (b) MOF-177 (c) DUT-6 (MOF-205) and (d) MIL-53 red oxygen gray carbon white hydrogen and blue metal atom (Zn Al)
MIL-53 is a MOF structure with one-dimensional pores The framework is built up of corner-sharing
AlndashO-based octahedral clusters interconnected with BDC linkers The linkers have again two
connection points MIL-53 shows reversible structural changes dependent on the guest molecules
[74] It undergoes the so-called breathing mode depending on the temperature and the amount of
adsorbed molecules
In this case also the bond lengths and bond angles are slightly overestimated comparing with the
experimental structures but the error does not exceed 3
91
Table 3 Selected bond lengths (Aring) and bond angles (˚) of MOF-5 (424 atoms in unit cell) MOF-177 (808 atoms in unit cell) and DUT-6 (MOF-205) (546 atoms in unit cell) optimized at DFTB level The available experimental data is given in parenthesis [70-73+ Orsquo denotes the central O atom in the Zn-O-octahedron
Bond Type MOF-5 MOF-177 DUT-6
(MOF-205)
Zn-Zn 330 (317) 322-336 (306-330)
325-331 (318)
Zn-Orsquo 202 (194) 202 (193) 202 (194) Zn-O 204 (192) 202-206
(190-199) 202 205 (193)
O-C 128 (130) 128 (131) 128 (125) ZnOrsquoZn 1095 (1095) 1056-1124
(1055 1092) 107-1118 (1084 1100)
OZnO 1083 1108 (1061)
1048 1145 (981-1281)
1046-1112 (1062 1085)
Zn-O-O-Zn 0 (0 plusmn17 plusmn20) plusmn122 - plusmn347 (plusmn176 - plusmn374)
05 - plusmn62 (0 plusmn29)
O-C-C-C 00 (0 - plusmn24) (plusmn 35 - plusmn 336) (plusmn65 - plusmn374)
plusmn04 plusmn22 (0 plusmn174)
Cell paramet a=b=c=26472 (25832) α=β=γ=90 (90)
a=b=37872 (37072) c=3068 (30033) α=β=90 (90) γ=120 (120)
a=b=c=31013 (30353) α=β=γ=90 (90)
We have calculated PDOS for MOF-5 MOF-177 and DUT-6 (MOF-205) (see Fig 4) Their band gaps
calculated to be 40 35 and 31 eV respectively indicate that they are either semiconductors or
insulators We have calculated it for both periodic crystal structure and a cluster containing a metal-
oxide connector and all its carboxylate linkers
Table 4 Selected bond lengths (Aring) and bond angles (˚) of MIL-53 (152 atoms in unit cell) optimized at DFTB level
Bond Type DFTB Exp
Al-Al 341 331 Al-O 189 183-191 O-C 133 129 130 O-Al-O 884 912 886 898 Al-O-Al 1289 1249 Al-O-O-Al 0 0 O-C-C-O 06 03 Cell paramet a=1246
b=1732 c=1365 α=β=γ=90
a=1218 b=1713 c=1326 α=β=γ=90
4 Mechanical properties Due to the low-mass density the elastic constants of porous materials
are a very sensitive indicator of their mechanical stability For crystal structures like MOFs we have
92
studied these by means of bulk modulus (B) B can be calculated as a second derivative of energy
with respect to the volume of the crystal (here unit cell)
The result shows that CuBTC has bulk modulus of 3466 GPa what is in close agreement with
B=3517 GPa obtained using force-field calculations [73 74] and experiment [75] Bulk moduli for the
series of MOFs resulted in 1534 1010 and 1073 GPa for MOF-5 -177 and DUT-6 (MOF-205)
respectively For MOF-5 B=1537 GPa was reported from DFTGGA calculations using plane-waves
[76 77] The results show that larger linkers give mechanically less stable structures what might be
an issue for porous structures with larger voids For MIL-53 we have obtained the largest bulk
modulus of 5369 GPa keeping the angles of the pore fixed
5 Conclusions We have validated the DFTB method with SCC and London dispersion corrections for
various types of MOFs The method gives excellent geometrical parameters compared to experiment
and for small model systems also in comparison with DFT calculations Importantly this statement
holds not only for catalytically inactive MOFs based on the Zn4O(CO2)6 octahedron and organic linkers
which are important for gas adsorption and separation applications but also for catalytically active
HKUST-1 (CuBTC) This is remarkable as the Cu atoms are in the Cu2+ state in this framework DFTB
parameters have been generated and validated for Cu and the electronic structure contains one
unpaired electron per Cu atom in the unit cell which makes the electronic description technically
difficult but manageable within the SCC-DFTB method SCC-DFTB performs well for the frameworks
themselves as well as for adsorbed CO and water molecules
Partial density-of-states calculations for the transition metal centers reveal that the concept of
reticular chemistry ndash individual building units keep their electronic properties when being integrated
to the framework ndash is also valid for the MOFs studied in this work in agreement with a previous
study of COFs [16] The electronic properties computed using the cluster models and the periodic
structure contains the same features and hence cluster models are good models to study the
catalytic and adsorption properties of these materials This is in particular useful if local quantum
chemical high-level corrections shall be applied that require to use finite structures
We finally conclude that we have now a high-performing quantum method available to study various
classes of MOFs of unit cells up to 10000 atoms including structural characteristics the formation
and breaking of chemical and coordination bonds for the simulation of diffusion of adsorbate
molecules or lattice defects as well as electronic properties The parameters can be downloaded
from the DFTBorg website
93
References
[1] E A Tomic J Appl Polym Sci 9 3745 (1965)
2+ M Eddaoudi D B Moler H L Li B L Chen T M Reineke M OrsquoKeeffe and O M Yaghi Acc Chem Res
34 319 (2001)
[3] M Eddaoudi H L Li T Reineke M Fehr D Kelley T L Groy and O M Yaghi Top Catal 9 105 (1999)
[4] B F Hoskins and R Robson J Am Chem Soc 111 5962 (1989)
[5] K Schlichte T Kratzke and S Kaskel Microporous Mesoporous Mater 73 81 (2004) [6] J L C Rowsell A
R Millward K S Park and O M Yaghi J Am Chem Soc 126 5666 (2004)
7+ O M Yaghi M OrsquoKeeffe N W Ockwig H K Chae M Eddaoudi and J Kim Nature 423 705 (2003)
[8] M Eddaoudi J Kim N Rosi D Vodak J Wachter M OrsquoKeeffe and O M Yaghi Science 295 469 (2002)
9+ M OrsquoKeeffe M Eddaoudi H L Li T Reineke and O M Yaghi J Solid State Chem 152 3 (2000)
[10] S S Y Chui SM F Lo J P H Charmant A G Orpen and I D Williams Science 283 1148 (1999)
11+ A P Cote A I Benin N W Ockwig M OrsquoKeeffe A J Matzger and O M Yaghi Science 310 1166 (2005)
[12] A P Cote H M El-Kaderi H Furukawa J R Hunt and O M Yaghi J Am Chem Soc 129 12914 (2007)
[13] H M El-Kaderi J R Hunt J L Mendoza-Cortes A P Cote R E Taylor M OrsquoKeeffe and O M Yaghi
Science 316 268 (2007)
[14] S Wan J Guo J Kim H Ihlee and D L Jiang Angew Chem 120 8958 (2008)
[15] S Wan J Guo J Kim H Ihlee and D L Jiang Angew Chem 121 5547 (2009)
[16] B Lukose A Kuc J Frenzel and T Heine Beilstein J Nanotechnol 1 60 (2010)
[17] B Lukose A Kuc and T Heine Chem Eur J 17 2388 (2011)
[18] D S Marlin D G Cabrera D A Leigh and A M Z Slawin Angew Chem Int Ed 45 77 (2006)
[19] I Krkljus and M Hirscher Microporous Mesoporous Mater 142 725 (2011)
[20] M Hirscher Handbook of Hydrogen Storage (Wiley-VCH Weinheim 2010)
[21] H Kitagawa Nature Chem 1 689 (2009)
[22] M Sadakiyo T Yamada and H Kitagawa J Am Chem Soc 131 9906 (2009)
[23] T Yamada M Sadakiyo and H Kitagawa J Am Chem Soc 131 3144 (2009)
94
[24] K S Jeong Y B Go S M Shin S J Lee J Kim O M Yaghi and N Jeong Chem Sci 2 877 (2011)
[25] R Schmid and M Tafipolsky J Am Chem Soc 130 12600 (2008)
[26] M Tafipolsky S Amirjalayer and R Schmid J Comput Chem 28 1169 (2007)
[27] M Tafipolsky S Amirjalayer and R Schmid J Phys Chem C 114 14402 (2010)
[28] A Warshel and M Levitt J Mol Biol 103 227 (1976)
[29] G Seifert D Porezag and T Frauenheim Int J Quantum Chem 58 185 (1996)
[30] A F Oliveira G Seifert T Heine and H A Duarte J Braz Chem Soc 20 1193 (2009)
[31] D Porezag T Frauenheim T Koumlhler G Seifert and R Kaschner Phys Rev B 51 12947 (1995)
[32] T Frauenheim G Seifert M Elstner Z Hajnal G Jungnickel D Porezag S Suhai and R Scholz Phys
Status Solidi B 217 41 (2000)
[33] M Elstner Theor Chem Acc 116 316 (2006)
Supporting Information
Figure S4 Cluster models as used for the P-DOS calculations in Fig 4 MOF-5 MOF-177 and DUT-6 (MOF-205) (from left to right)
95
Figure S3 Radial pair distribution function (g(r)) of Cu-Cu Cu-C Cu-O C-C and C-O atom-pairs in a Cu-BTC MOF unit cell
96
Appendix C
The Structure of Layered Covalent-Organic Frameworks
Binit Lukose Agnieszka Kuc and Thomas Heine
Chem Eur J 2011 17 2388 ndash 2392
DOI 101002chem201001290
Abstract Covalent-Organic Frameworks (COFs) are a new family of 2D and 3D highly porous and
crystalline materials built of light elements such as boron oxygen and carbon For all 2D COFs an AA
stacking arrangement has been reported on the basis of experimental powder XRD patterns with the
exception of COF-1 (AB stacking) In this work we show that the stacking of 2D COFs is different as
originally suggested COF-1 COF-5 COF-6 and COF-8 are considerably more stable if their stacking
arrangement is either serrated or inclined and layers are shifted with respect to each other by ~14 Aring
compared with perfect AA stacking These structures are in agreement with to date experimental
data including the XRD patterns and lead to a larger surface area and stronger polarisation of the
pore surface
Introduction In 2005 Cocircteacute and co-workers introduced a new class of porous crystalline materials
Covalent Organic Frameworks (COFs)[1] where organic linker molecules are stitched together by
connectors covalent entities including boron and oxygen atoms to a regular framework These
materials have the genuine advantage that all framework bonds represent strong covalent
interactions and that they are composed of light-weight elements only which lead to a very low
mass density[2] These materials can be synthesized solvothermally in a condensation reaction and
97
are composed of inexpensive and non-toxic building blocks so their large-scale industrial production
appears to be possible
Figure 1 Calculated and experimental XRD patterns of all studied COFs (COF-1 top left COF-5 top right COF-6 bottom left COF-8 bottom right)
To date a number of different COF structures have been reported[1ndash3] From a topological
viewpoint we distinguish two- and three-dimensional COFs In two-dimensional (2D) COFs the
covalently bound framework is restricted to 2D layers The crystal is then similar as in graphite or
hexagonal boron nitride composed by a stack of layers which are not connected by covalent bonds
but held together primarily by London dispersion interactions
98
The COF structures have been analyzed by powder X-ray diffraction (PXRD) experiments The
topology of the layers is determined by the structure of the connector and linker molecules and
typically a hexagonal pattern is formed due to the 2m (D3h) symmetry of the connector moieties
The individual layers are then stacked and form a regular crystal lattice With one exception (COF-
1)[1] for all 2D-COFs AA (eclipsed P6mmm) interlayer stacking has been reported[3andashd f g] This
geometrical arrangement maximizes the proximity of the molecular entities and results in straight
channels orthogonal to the COF layers which are known from the literature[1 3a]
The connectors of 2D-COFs include oxygen and boron atoms which cause an appreciable polarization
The AA stacking arrangement maximizes the attractive London dispersion interaction between the
layers which is the dominating term of the stacking energy At the same time AA stacking always
results in a repulsive Coulomb force between the layers due to the polarized connectors It should be
noted that the eclipsed hexagonal boron nitride (h-BN)[4] has boron atoms in one layer serving as
nearest neighbours to nitrogen atoms in adjacent layers (AB stacking) The Coulomb interaction has
ruled out possible interlayer eclipse between atoms of alike charges In this article we aim at
studying the layer arrangements of solvent-free COFs based on the interlayer interactions and the
minimum variance Various lattice types have been considered all significantly more stable than the
reported AA stacked geometry In all 2D COFs we have studied the Coulomb interaction between the
layers leads to a modification of the stacking and shifts the layers by about one interatomic distance
(~14 Aring) with respect to each other (see Figure 1)
Breaking of the highly symmetric layer arrangement has been observed for COF-1 and COF-10 after
removal of guest molecules from pores leading to a turbostratic repacking of pore channels[1 5]
The authors have confirmed the disorder either by the changes in the PXRD spectrum taken before
and after adsorptionndashdesorption cycle[1] or by the changes in the adsorption behavior[5] The
disruption of layer arrangement is attributed to the force exerted by the guest molecules Formation
of poly disperse pores has also been verified[5] The lattice types that we discuss in this article on
the other hand are neither the result of the pressure from any external molecule in the pore nor
having more than one type of pores They are the resultant of minimum variance guided by Coulomb
and London dispersion interactions For the COF models under investigation perfect crystallinity has
been considered
Methods We have studied the experimentally known structures of COF-1 COF-5 COF-6 and COF-8
Structure calculations were carried out using the Dispersion-Corrected Self-Conistent-Charge
Density-Functional-Based Tight-Binding method (SCC-DFTB)[6] DFTB is based on a second-order
expansion of the KohnndashSham total energy in DFT with respect to charge density fluctuations This
does not require large amounts of empirical parameters however maintains all qualities of DFT The
99
accuracy is very much improved by the SCC extension The DFTB code (deMonNano[6a]) has
dispersion correction[6d] implemented to account for weak interactions and was used to obtain the
layered bulk structure of COFs and their formation energies The performance for interlayer
interactions has been tested previously for graphite[6d] All structures correspond to full geometry
optimizations (structure and unit cell) XRD patterns have been simulated using the Mercury
software[7] To allow best comparison with experiment for PXRD simulations we used the calculated
geometry of the layer and of the relative shifts between the layers but experimental interlayer
distances The electrostatic potential has been calculated using Crystal 09[8] at the DFTVWN level
with 6-31G basis set
Results and Discussion
In order to see the favorite stacking arrangement of the layers we have shifted every second layer in
two directions along zigzag and armchair vertices of the hexagonal pores (see Figure 2) The initial
stacking is AA (P6mmm) and the final zigzag shift gives AB stacking (P63mmc) Considering that the
attractive interlayer dispersion forces and repulsive interlayer orbital interactions are different we
have optimized the interlayer separation for each stacking Figure 2 show their total energies
calculated per formula unit that is per established bond between linkers and connectors with
reference to AA stacking for COF-5 and COF-1 In each direction the energy minimum is found close
to the AA stacking position shifted only by about one aromatic C-C bond length (~14 Aring) so that
either connector or linker parts become staggered with those in the adjacent layers leading to a
stacking similar to that of graphite for either connector (armchair shift) or linker (zigzag shift) For
COF-5 the staggering at the connector or linker depends whether the shift is armchair or zigzag
respectively while for COF-1 the zigzag shift alone can give staggering in both the aromatic and
boron-oxygen rings
The low-energy minima in the two directions are labeled following the common nomenclature as
zigzag and armchair respectively Both shifts result in a serrated stacking order (orthorhombic
Cmcm) which shows a significantly more attractive interlayer Coulomb interaction than AA stacking
(see Table 1) while most of the London dispersion attraction is maintained and consequently
stabilizes the material Still this configuration can be improved if we consider inclined stacking
(Figure 1) that is the lattice vector pointing out of the 2D plane is not any longer rectangular
reducing the lattice symmetry from Cmcm (orthorhombic) to C2m (monoclinic)
Besides we have calculated the monolayer formation energy (condensation energy Ecb) from the
total energies of the monolayer and of the individual building blocks and the stacking formation
energy from the total energies of the bulk structure and of the monolayer for four selected COFs The
100
COF building blocks are the same as those used for their synthesis[1 3a] (BDBA for COF-1 BDBA and
HHTP for COF-5 BTBA and HHTP for COF-6 BTPA and HHTP for COF-8) The final energies given per
formula unit that is per condensed bond are given in Table 1 The serrated and inclined stacking
structures are energetically more stable than AA and AB Interestingly within our computational
model zigzag and armchair shifting is energetically equivalent
Figure 2 Change of the total energy (per formula unit) when shifting COF-5 even layers in zigzag (top) and armchair (middle) directions and COF-1 in zigzag (bottom) direction The vector d shows the shift direction The low-energy minima in the two directions are labelled as zigzag and armchair respectively The equilibrium structures are shown as well
The formation energies of the individual COF structures are in agreement with full DFT calculations
We have calculated using ADF[9] the condensation energy of COF-1 and COF-5 using first-principles
DFT at the PBE[10]DZP[11] level to qualitatively support our results For simplicity we have used a
finite structure instead of the bulk crystal Their calculated Ecb energies are 77 and 15 kJmol-1
respectively for COF-1 and COF-5 These values support the endothermic nature of the condensation
101
reaction and are in excellent agreement with our DFTB results (91 and 21 kJmol-1 respectively see
Table 1)
The change of stacking should be visible in X-ray diffraction patterns because each space group has a
distinct set of symmetry imposed reflection conditions Powder XRD patterns from experiments are
available for all 2D COF structures[1 3andashd f g] We have simulated XRD patterns for AA AB serrated
Table 1 Calculated formation energies for the studied COF structures Ecb = condensation Esb = stacking energy EL = London dispersion energy Ee = electronic energy Energies are per condensed bond and are given in [kJ mol
-1+ lsquoarsquo and lsquozrsquo stand for armchair and zigzag respectively Note that the electronic energy Ee=Esb-EL
includes the electronic interlayer interaction and the formation energy of the monolayer and that the formation of the monolayers is endothermic
Structure Stacking Esb EL Ee
COF-5 AA -2968 -3060 092
AB -2548 -2618 070
serrated z -3051 -3073 022
serrated a -3052 -3073 021
inclined z -3297 -3045 -252
inclined a -3275 -3044 -231
Monolayer Ecb= 211
COF-1 AA -2683 -2739 056
AB -2186 -2131 -055
serrated z -2810 -2806 -004
inclined z -2784 -2788 004
Monolayer Ecb= 906
COF-6 AA -2881 -2963 082
AB -2127 -2146 019
serrated z -2978 -2996 018
serrated a -2978 -2995 017
inclined z -2946 -2975 029
inclined a -2954 -2974 021
Monolayer Ecb= 185
COF-8 AA -4488 -4617 129
102
AB -2477 -2506 029
serrated z -4614 -4646 032
serrated a -4615 -4647 032
inclined z -4578 -4612 035
inclined a -4561 -4591 030
Monolayer Ecb= 263
and inclined stacking forms for COF-5 COF-1 COF-6 and COF-8 (see Figure 1for their comparison
with experimental spectrum) For completeness we have studied XRD patterns of optimized COFs
using the experimentally determined[1 3a] interlayer separations this means we have kept the
layer geometry the same as the optimized structures and different stackings were obtained by
shifting adjacent layers accordingly
COF-1 was reported as having different powder spectra for samples before (as-synthesized) and after
removal of guest molecules with a particular mentioning about its layer shifting after removal We
have compared the two spectra with our simulated XRDs in order to see the stacking in the pure
form and how the stacking is changed at the presence of mesitylene guests Except that we have only
a vague comparison at angles (2θ) 11ndash15 the powder spectrum after guest removal is more similar
to the XRDs of serrated and inclined stacking structures A notable difference from AB is the absence
of high peaks at angles ~12 ~15 and ~19 This gives rise to the suggestion that COF-1 is not a
notable exception among the 2D COFs it follows the same topological trend as all other frameworks
of this class having one-dimensional (1D) pores as soon as the solvent is removed from the pores
This suggestion is supported by the experimental fact that the gas adsorption capacity of COF-1 is
only slightly lower than that of COF-6 for even as large gas molecules as methane and CO2[12] It is
not obvious how these molecules would enter an AB stacked COF-1 framework where the pores are
not connected by 1D channels (see Figure 1) Moreover our calculations suggest that the serrated
and inclined stackings are energetically favorable (see Table 1)
Indeed all the new stackings discussed in this article yield XRD patterns which are in agreement with
the currently available experimental data (see Figure 1) The inclined stackings have more peaks but
those are covered by the broad peaks in the experimental pattern The same is observed for the (002)
peaks of serrated stackings At present 2D COF synthesis methods are not yet capable to produce
crystals of sufficient quality to allow more detailed PXRD analysis in particular for the solvent-free
materials Microwave synthesis of COF-5 by rapid heating is reported[13] as much faster (~200 times)
compared with solvothermal methods however the structural details (XRD etc) remained
103
ambiguous We are confident that better crystals will be produced in future which will allow the
unambiguous determination of COF structures and can be compared to our simulations
Finally we want to emphasize that the suggested change of stacking is not only resulting in a
moderate change of geometry and XRD pattern The functional regions of COFs are their pores and
the pore geometry is significantly modified in our suggested structures compared to AA and AB
stackings First the inclined and serrated structures account for an increase of the surface area by 6
estimated for the interaction of the COFs with helium (3 for N2 5 for Ar 56 for Ne) Moreover
the shift between the layers exposes both boron and oxygen atoms to the pore surface leading to a
much stronger polarity than it can be expected for AA stacked COFs This difference is shown in
Figure 3 which plots the electrostatic potential of COF-5 in AA serrated and inclined stacking
geometries While the pore surface of AA stacked COF-5 shows a positively and negatively charged
stripes the other stacking arrangements show a much stronger alternation of charges indicating the
higher polarity of the surface The surface polarity can also be expressed in terms of Mulliken charges
of oxygen and boron atoms calculated at the DFT level which are COF-5 AA qB = 048 and qO = -048
COF-5 serrated qB = 047 and qO = -049 COF-5 inclined qB = 045 and qO = -048
Figure 3 Calculated electrostatic potential of COF-5 AA (left) serrated (middle) and inclined (right) stacking forms Blue - negative and red-positive values of the 005 isosurface
Conclusion We have studied 2D COF layer stackings energetically and found that energy minimum
structures have staggering of hexagonal rings at the connector or linker parts This can be achieved if
the bulk structure has either serrated or inclined order These newly proposed orders have their
simulated XRDs matching well with the available experimental powder spectrum Hence we claim
that all reported 2D COF geometries shall be re-examined carefully in experiment Due to the change
of the pore geometry and surface polarity the envisaged range of applications for 2D COFs might
change significantly We believe that these results are of utmost importance for the design of
functionalized COFs where functional groups are added to the pore surfaces
104
References
[1] A P Cocircteacute A I Benin N W Ockwig M OKeeffe A J Matzger O M Yaghi Science 2005 310 1166
[2] H M El-Kaderi J R Hunt J L Mendoza-Cortes A P Cote R E Taylor M OKeeffe O M Yaghi Science
2007 316 268
[3] a) A P Cocircteacute H M El-Kaderi H Furukawa J R Hunt O M Yaghi J Am Chem Soc 2007 129 12914 b) J
R Hunt C J Doonan J D LeVangie A P Cocircte O M Yaghi J Am Chem Soc 2008 130 11872 c) R W
Tilford W R Gemmill H C zur Loye J J Lavigne Chem Mater 2006 18 5296 d) R W Tilford S J Mugavero
P J Pellechia J J Lavigne Adv Mater 2008 20 2741 e) F J Uribe-Romo J R Hunt H Furukawa C Klock M
OKeeffe O M Yaghi J Am Chem Soc 2009 131 4570 f) S Wan J Guo J Kim H Ihee D L Jiang Angew
Chem 2008 120 8958 Angew Chem Int Ed 2008 47 8826 g) S Wan J Guo J Kim H Ihee D L Jiang
Angew Chem 2009 121 5547 Angew Chem Int Ed 2009 48 5439
[4] R T Paine C K Narula Chem Rev 1990 90 73
[5] C Doonan D Tranchemontagne T Glover J Hunt O Yaghi Nat Chem 2010 2 235
[6] a) T Heine M Rapacioli S Patchkovskii J Frenzel A M Koester P Calaminici S Escalante H A Duarte R
Flores G Geudtner A Goursot J U Reveles A Vela D R Salahub deMon deMonnano edn Mexico DF
Mexico and Bremen (Germany) 2009 b) A F Oliveira G Seifert T Heine H A Duarte J Braz Chem Soc
2009 20 1193 c) G Seifert D Porezag T Frauenheim Int J Quantum Chem 1996 58 185 d) L Zhechkov T
Heine S Patchkovskii G Seifert H A Duarte J Chem Theory Comput 2005 1 841
[7] a) httpwwwccdccamacukproductsmercury b) C F Macrae P R Edgington P McCabe E Pidcock
G P Shields R Taylor M Towler J Van de Streek J Appl Crystallogr 2006 39 453
[8] R Dovesi V R Saunders C Roetti R Orlando C M Zicovich- Wilson F Pascale B Civalleri K Doll N M
Harrison I J Bush P DrsquoArco M Llunell Crystal 102 ed
[9] a) httpwwwscmcom ADF200901 Amsterdam The Netherlands b) G te Velde F M Bickelhaupt S J
A van Gisbergen C Fonseca Guerra E J Baerends J G Snijders T Ziegler J Comput Chem 2001 22 931
[10] J P Perdew K Burke M Ernzerhof Phys Rev Lett 1996 77 3865
[11] E Van Lenthe E J Baerends J Comput Chem 2003 24 1142
[12] H Furukawa O M Yaghi J Am Chem Soc 2009 131 8875
[13] N L Campbell R Clowes L K Ritchie A I Cooper Chem Chem Mater 2009 21 204
105
Appendix D
On the reticular construction concept of covalent organic frameworks
Binit Lukose Agnieszka Kuc Johannes Frenzel and Thomas Heine
Beilstein J Nanotechnol 2010 1 60ndash70
DOI103762bjnano18
Abstract
The concept of reticular chemistry is investigated to explore the applicability of the formation of
Covalent Organic Frameworks (COFs) from their defined individual building blocks Thus we have
designed optimized and investigated a set of reported and hypothetical 2D COFs using Density
Functional Theory (DFT) and the related Density Functional based tight-binding (DFTB) method
Linear trigonal and hexagonal building blocks have been selected for designing hexagonal COF layers
High-symmetry AA and AB stackings are considered as well as low-symmetry serrated and inclined
stackings of the layers The latter ones are only slightly modified compared to the high-symmetry
forms but show higher energetic stability Experimental XRD patterns found in literature also
support stackings with highest formation energies All stacking forms vary in their interlayer
separations and band gaps however their electronic densities of states (DOS) are similar and not
significantly different from that of a monolayer The band gaps are found to be in the range of 17ndash
40 eV COFs built of building blocks with a greater number of aromatic rings have smaller band gaps
Introduction
In the past decade considerable research efforts have been expended on nanoporous materials due
to their excellent properties for many applications such as gas storage and sieving catalysis
106
selectivity sensoring and filtration [1] In 1994 Yaghi and co-workers introduced ways to synthesize
extended structures by design This new discipline is known as reticular chemistry [23] which uses
well-defined building blocks to create extended crystalline structures The feasibility of the building
block approach and the geometry preservation throughout the assembly process are the key factors
that lead to the design and synthesis of reticular structures
One of the first families of materials synthesized using reticular chemistry were the so-called Metal-
Organic Frameworks (MOFs) [4] They are composed of metal-oxide connectors which are covalently
bound to organic linkers The coordination versatility of constituent metal ions along with the
functional diversity of organic linker molecules has created immense possibilities The immense
potential of MOFs is facilitated by the fact that all building blocks are inexpensive chemicals and that
the synthesis can be carried out solvothermally MOFs are commercially available and the scale up of
production is continuing Since the discovery of MOFs many other crystalline frameworks have been
synthesized using reticular chemistry such as Metal-Organic Polyhedra (MOP) [5] Zeolite
Imidazolate Frameworks (ZIFs) [6] and Covalent Organic Frameworks (COFs) [7]
In 2005 Cocircteacute and co-workers introduced COF materials [7-14] where organic linker molecules are
stitched together by covalent entities including boron and oxygen atoms to form a regular
framework These materials have the distinct advantage that all framework bonds represent strong
covalent interactions and that they are composed of light-weight elements only which lead to a very
low mass density [7-9] These materials can be synthesized by wet-chemical methods by
condensation reactions and are composed of inexpensive and non-toxic building blocks so their
large-scale industrial application appears to be possible From a topological viewpoint we distinguish
two- and three-dimensional COFs In two-dimensional (2D) COFs the covalently bound framework is
restricted to layers The crystal is then similar as in graphite composed of a stack of layers which
are not connected by covalent bonds
COFs compared with MOFs have lower mass densities due to the absence of heavy atoms and
therefore might be better for many applications For example the gravimetric uptake of gases is
almost twice as large as that of MOFs with comparable surface areas [1516] Because COFs are fairly
new materials many of their properties and applications are still to be explored
Recently we have studied the structures of experimentally well-known 2D COFs [17] We have found
that commonly accepted 2D structures with AA and AB kinds of layering are energetically less stable
than inclined and serrated forms This is because AA stacking maximises the Coulomb repulsion due
to the close vicinity of charge carrying atoms alike (O B atoms) in neighbouring layers The serrated
and inclined forms are only slightly modified (layers are shifted with respect to each other by asymp14 Aring)
107
and experience less Coulomb forces between the layers compared to AA stacking This is equivalent
to the energetic preference of graphite for an AB (Bernal) over an AA form (simple hexagonal) if we
ignore the fact that interlayer ordering in serrated and inclined forms are not uniform everywhere A
known example of this is that in eclipsed hexagonal boron nitride (h-BN) boron atoms in one layer
serve as nearest neighbours to nitrogen atoms in adjacent layers (AB stacking) The Coulomb
interaction rules out possible interlayer eclipse between atoms with similar charges in this case
In the present work we aim to explore how far the concept of reticular chemistry is applicable to the
individual building units which define COFs For this purpose we have investigated a set of reported
and hypothetical 2D COFs theoretically by exploring their structural energetic and electronic
properties We have compared the properties of the isolated building blocks with those of the
extended crystal structures and have found that the properties of the building units are indeed
maintained after their assembly to a network
Results and Discussion
Structures and nomenclature
We have considered four connectors (IndashIV) and five linkers (andashe) for the systematic design of a
number of 2D COFs (Figure 1) Each COF was built from one type of connector and one type of linker
thus resulting in the design of 20 different structures Moreover we have considered two
hypothetical reference structures which are only built from connectors I and III (no linker is present)
REF-I and REF-III Properties of other COFs were compared with the properties of these two
structures Some of the studied COFs are already well known in the literature [781314] and we
continue to use their traditional nomenclature while hypothetical ones are labelled in the
chronological order with an M at the end which stands for modified
Figurel 1 The connector (IndashIV) and linker (andashe) units considered in this work The same nomenclature is used in the text Carbon ndash green oxygen ndashred boron ndash magenta hydrogen ndash white
108
Using the secondary building unit (SBU) approach we can distinguish the connectors between
trigonal [T] (connectors I II III) and hexagonal [H] (connector IV) and the linkers between linear [l]
(linkers a b e) and trigonal [t] (linkers c d) Topology of the layer is determined by the geometries
of the connector and linker molecules and typically a hexagonal pattern is formed due to the D3h
symmetry of the connector moieties Based on these topologies of the constituent building blocks
we have classified the studied COFs into four groups Tl Tt Hl and Ht (Figure 2) Hereafter we will
use this nomenclature to describe the COF topologies
Figurel 2 Topologies of 2D COFs considered in this work (from the left) Tl Tt Hl and Ht Red and blue blocks are secondary building units corresponding to connectors and linkers respectively
We have considered high-symmetry AA and AB kinds of stacking (hexagonal) and low-symmetry
serrated (orthorhombic) and inclined (monoclinic) kinds of stacking of the layers The latter two were
discussed in a previous work on 2D COFs [17] As an example the structure of COF-5 [7] in different
kinds of stacking of layers is shown in Figure 3 In eclipsed AA stacking atoms of adjacent layers lie
directly on top of each other whilst in staggered AB stacking three-connected vertices lie directly on
top of the geometric center of six-membered rings of neighbouring layers In both serrated and
inclined kinds of stacking the layers are shifted with respect to each other by approximately 14 Aring
resulting in hexagonal rings in the connector or linker being staggered with those in the adjacent
layers In serrated stacking alternate layers are eclipsed In inclined stacking layers lie shifted along
one direction and the lattice vector pointing out of the 2D plane is not rectangular For COFs made of
connector I due to the absence of five-membered C2O2B rings a zigzag shift leads to staggering in
both connector and linker parts For those made of other connectors staggering at the connector or
linker depends on whether the shift is armchair or zigzag respectively [17]
Structural properties
We have compared structural properties of isolated building blocks with those of the extended COF
structures Negligible differences have been found in the bond lengths and bond angles of the
building blocks and the corresponding crystal structures This indicates that the structural integrity of
the building blocks remains unchanged while forming covalent organic frameworks confirming the
109
principle of reticular chemistry In addition the CndashB BndashO and OndashC bond lengths are almost the same
when different COF structures are compared (see Table S1 in Supporting Information File 1) The
experimental bond lengths are asymp154 Aring for CndashB asymp138 Aring for OndashC and 137ndash148 Aring for BndashO However
in the case of COF-1 the experimental values are slightly larger (160 Aring for CndashB and 151 Aring for BndashO)
This could be the reason why our calculated bond lengths for COF-1 are much shorter than the
experimental values while all the other structures agree within 5 error The calculated CndashC bond
lengths vary in the range from 136ndash147 Aring (Figure S2 in Supporting Information File 1) and are the
same for the COFs and their constituent building blocks at the respective positions of the carbon
atoms In addition the reference structures REF-I and REF-III have direct BndashB bond lengths of 167 Aring
and 166 Aring respectively which is shorter by 014 Aring than a typical BndashB bond length The calculated
bond angles OBO in B3O3 and C2O2B rings are 120deg and 113deg respectively
Figure 3 Layer stackings considered AA AB serrated and inclined
Interlayer distances (d) which is the shortest distance between two layers (equivalent to c for AA
c2 for AB and serrated) are different in all kinds of stacking AB stacked 2D COFs have shorter
interlayer distances than the corresponding AA serrated and inclined stacked structures Among the
latter three AA stacked COFs have higher values for d because of the higher repulsive interlayer
orbital interactions resulting from the direct overlap of polarized alike atoms between the adjacent
layers This results in higher mass densities for AB stacked COF analogues Serrated and inclined
stacks have only slightly higher mass densities compared to AA The differences in mass densities for
all kinds of stacking are attributed to the differences in their interlayer separations The d values of
most of the COFs are larger than that of graphite in AA stacking but smaller in AB stacking
Cell parameters (a) and mass densities (ρ) of all the COFs constructed from the considered
connectors and linkers are shown in Table 1 (and in Table S3 in Supporting Information File 1) Mass
densities of all the COFs are much lower than that of graphite (227 gmiddotcmminus3) and diamond (350
gmiddotcmminus3) AAserratedinclined stacked COF-10s have the lowest mass densities (045046046
gmiddotcmminus3) which is lower than that of MOF-5 (059 gmiddotcmminus3) [4] and comparable to that of highly porous
MOF-177 (042 gmiddotcmminus3) [18]
110
In order to identify the stacking orders we have analyzed X-ray diffraction (XRD) patterns of the well-
known COFs (COF-10 TP COF PPy-COF see Figure 4) in all the above discussed stacking kinds The
change of stacking should be visible in XRDs because each space group has a distinct set of symmetry
imposed reflection conditions The XRD patterns of AA serrated and inclined stacking kinds which
differ within a slight shift of adjacent layers to specific directions are in agreement with the presently
available experimental data [81314] Their peaks are at the same angles as in the experimental
spectrum whereas AB stacking clearly shows differences The slight differences in the (001) angle
between each stacking resemble the differences in their interlayer separations The inclined
stackings have more peaks however they are covered by the broad peaks in the experimental
patterns Similar results for COF-1 COF-5 COF-6 and COF-8 have been discussed in our previous
work [17]
Figure 4 The calculated and experimental [81314] XRD patterns of PPy-COF (top) COF-10 (middle) and TP COF (bottom)
111
Table 1 The calculated unit cell parameter a [Aring] interlayer distance d [Aring] and mass density ρ [gmiddotcmminus3
] for AA and AB stacked COFs Note that the cell parameter a is the same for all stacking types Experimental data [719] is given in parentheses
COF Building
Blocks
a d ρ
AA AB AA AB
COF-1 I-a 1502 (15620) 351 313 (332) 094 106
COF-1M I-b 2241 349 304 068 078
COF-2M I-c 1492 347 312 095 106
COF-3M I-d 0747 349 327 153 164
PPy-COF I-e 2232 (22163) 349 (3421) 297 084 099
COF-5 II-a 3014 (30020) 347 (3460) 326 056 060
COF-10 II-b 3758 (37810) 347 (3476) 318 045 (045) 050
COF-8 II-c 2251 (22733) 346 (3476) 320 071 (070) 077
COF-6 II-d 1505 (15091) 346 (3599) 327 104 110
TP COF II-e 3750 (37541) 348 (3378) 320 051 056
COF-4M III-a 2171 350 318 073 080
COF-5M III-b 2915 350 318 055 061
COF-6M III-c 1833 345 318 083 090
COF-7M III-d 1083 350 330 129 136
TP COF-1M III-e 2905 349 310 065 074
COF-8M IV-a 1748 359 329 140 148
COF-9M IV-b 2176 349 330 117 174
COF-10M IV-c 2254 342 336 127 128
COF-11M IV-d 1512 346 338 168 172
TP COF-2M IV-e 2173 347 332 134 140
REF-I I 0773 359 336 144 148
REF-III III 1445 353 336 104 121
Graphite 247 343 335 220 227
112
Energetic stability
We have considered dehydration reactions the basis of experimental COF synthesis to calculate
formation energies of COFs in order to predict their energetic stability Molecular units 14-
phenylenediboronic acid (BDBA) 11rsquo-biphenyl]-44-diylboronic acid (BPDA) 5rsquo-(4-boronophenyl)-
11rsquo3rsquo1rdquo-terphenyl]-44rdquo-diboronic acid (BTPA) benzene-135-triyltriboronic acid (BTBA) and
pyrene-27-diylboronic acid (PDBA) were considered as linkers andashe respectively with -B(OH)2 groups
attached to each point of extension (Figure 5) Self-condensation of these building blocks result in
the formation of B3O3 rings and the resultant COFs are those made of connector I and the
corresponding linkers This process requires a release of three or six water molecules in case of t or l
topology respectively
Figure 5 The reactants participating in the formation of 2D COFs
Co-condensation of the above molecular units with compounds such as 23671011-
hexahydroxytriphenylene (HHTP) hexahydroxybenzene (HHB) and dodecahydroxycoronene (DHC)
(Figure 5) gives rise to COFs made of connectors II III and IV respectively and the corresponding
linkers Self-condensation of tetrahydroxydiborane (THDB) and co-condensation of HHB with THDB
result in the formation of the reference structures REF-I and REF-III respectively In relation to the
corresponding connectorlinker topologies these building blocks satisfy the following equations of
the co-condensation reaction for COF formation
2 2 3 COF 12 H O Tl T l (1)
113
2 1 1 COF 6 H O Tt T t (2)
2 1 3 COF 12 H O Hl H l (3)
2 1 2 COF 12 H O Ht H t (4)
with a stochiometry appropriate for one unit cell The number of covalent bonds formed between
the building blocks is equivalent to the number of released water molecules we refer to it as
ldquoformula unitrdquo and will give all energies in the following in kJmiddotmolminus1 per formula unit
Table 2 The calculated energies [kJ molminus1
] per bond formed between building blocks for AA and AB stacked COFs Ecb is the condensation energy Esb is the stacking energy and Efb is the COF formation energy (Efb = Ecb
+ Esb) The calculated band gaps Δ eV+ are given as well
COF Building
Blocks
Mono-
layer
AA AB
Ecb Esb Efb ∆ Esb Efb ∆
COF-1 I-a 906 -2683 -1777 33 -2187 -1280 36
COF-1M I-b 949 -4266 -3366 27 -2418 -1469 31
COF-2M I-c 956 -5727 -4771 28 -4734 -3778 30
COF-3M I-d 763 -2506 -1742 38 -2801 -2037 40
PPy-COF I-e 858 -5723 -4866 24 -3855 -2998 26
COF-5 II-a 211 -2968 -2756 24 -2548 -2337 28
COF-10 II-b 317 -3766 -3448 23 -1344 -1026 26
COF-8 II-c 263 -4488 -4224 25 -2477 -2213 28
COF-6 II-d 185 -2881 -2695 28 -2127 -1942 31
TP COF II-e 231 -4453 -4222 24 -1480 -1250 27
COF-4M III-a -033 -1730 -1763 26 -880 -913 26
COF-5M III-b 007 -2533 -2526 25 -972 -965 25
COF-6M III-c 014 -3231 -3217 26 -2134 -2120 28
114
COF-7M III-d -170 -1635 -1805 30 -1607 -1777 32
TP COF-1M III-e -014 -3226 -3240 24 -1277 -1291 24
COF-8M IV-a -787 -2756 -3543 18 -2680 -3467 21
COF-9M IV-b -836 -3577 -4414 17 -3003 -3839 21
COF-10M IV-c -947 -4297 -5244 18 -4192 -5140 22
COF-11M IV-d -403 -2684 -3087 21 -2833 -3236 24
TP COF-2M IV-e 030 -4345 -4315 18 -4117 -4087 21
We have calculated the condensation energy of a single COF layer formed from monomers (building
blocks hereafter called bb) according to the above reactions as follows
tot tot tot totcb m H2O 1 bb1 2 bb2E E n E ndash m E m E (5)
where Emtot ndash total energy of the monolayer EH2O
tot is the total energy of the water molecule Ebb1tot
and Ebb2tot are the total energies of interacting building blocks and n m1 m2 are the corresponding
stoichiometry numbers
We have also calculated the stacking energy Esb of layers
tot totsb nl s mE E n E (6)
where Enltot is the total energy of ns number of layers stacked in a COF Finally the COF formation
energy can be given as a sum of Ecb and Esb (see Table 1 and Table S1)
Electronic properties
All COFs including the reference structures are semiconductors with their band gaps lying between
17 eV and 40 eV (Table 2 and Table S4 in Supporting Information File 1) The largest band gaps are
of the reference structures while the lowest values are of COFs built from connector IV The band
gaps are different for different stacking kinds This difference can be attributed to the different
optimized interlayer distances Generally AB serrated and inclined stacked COFs have band gaps
comparable to or larger than that of their AA stacked analogues
115
We have calculated the electronic total density of states (TDOS) and the individual atomic
contributions (partial density of states PDOS) The energy state distributions of COFs and their
monolayers are studied and a comparison for COF-5 is shown in Figure 6 In all stacking kinds
negligible differences are found for the densities at the top of valence band and the bottom of
conduction band These slight differences suggest that the weak interaction between the layers or
the overlap of π-orbitals does not affect the electronic structure of COFs significantly Hence there is
almost no difference between the TDOS of AA AB serrated and inclined stacking kinds Therefore in
the following we discuss only the AA stacked structures
Figure 6 Total densities of states (DOS) (black) of AA (top left) AB (top right) serrated (bottom left) and inclined (bottom right) comparing stacked COF-5 with a monolayer (red) of COF-5 The differences between the TDOS of bulk and monolayer structures are indicated in green The Fermi level EF is shifted to zero
Figure 7 Partial density of states of carbon (left) boron (center) and oxygen (right) atoms of COFs built from type-I connectors and different linkers The vertical dashed line in each figure indicates the Fermi level EF
116
It is of interest to see how the properties of COFs change depending on their composition of different
secondary building units that is for different connectors and linkers PDOS of COFs built from type-I
connectors and different linkers are plotted in Figure 7 The PDOS of carbon atoms is compared with
that of graphite (AA-stacking) while PDOS of boron and oxygen atoms are compared with that of
REF-I a structure which is composed solely of connector building blocks Going from top to bottom
of the plots the number of carbon atoms per unit cell increases It can be seen that this causes a
decrease of the band gap Figure 8 shows the PDOS of COFs built from type-a linkers and different
connectors where the COFs are arranged in the increasing number of carbon atoms in their unit cells
from top to the bottom Again the C PDOS is compared with that of graphite while both REF-I and
REF- III are taken in comparison to O and B PDOS The observed relation between number of carbon
atoms and band gap is verified
Figure 8 Partial density of states of carbon (left) boron (center) and oxygen (right) atoms of COFs built from type-a linkers The vertical dashed line in each figure indicates the Fermi level EF
Conclusion
In summary we have designed 2D COFs of various topologies by connecting building blocks of
different connectivity and performed DFTB calculations to understand their structural energetic and
electronic properties We have studied each COF in high-symmetry AA and AB as well as low-
symmetry inclined and serrated stacking forms The optimized COF structures exhibit different
interlayer separations and band gaps in different kinds of layer stackings however the density of
states of a single layer is not significantly different from that of a multilayer The alternate shifted
layers in AB serrated and inclined stackings cause less repulsive orbital interactions within the layers
which result in shorter interlayer separation compared to AA stacking All the studied COFs show
117
semiconductor-like band gaps We also have observed that larger number of carbon atoms in the
unit cells in COFs causes smaller band gaps and vice versa Energetic studies reveal that the studied
structures are stable however notable difference in the layer stacking is observed from
experimental observations This result is also supported by simulated XRDs
Methods
We have optimized the atomic positions and the lattice parameters for all studied COFs All
calculations were carried out at the Density Functional Tight-Binding (DFTB) [2021] level of theory
DFTB is based on a second-order expansion of the KohnndashSham total energy in the Density-Functional
Theory (DFT) with respect to charge density fluctuations This can be considered as a non-orthogonal
tight-binding method parameterized from DFT which does not require large amounts of empirical
parameters however maintains all the qualities of DFT The main idea behind this method is to
describe the Hamiltonian eigenstates with an atomic-like basis set and replace the Hamiltonian with
a parameterized Hamiltonian matrix whose elements depend only on the inter-nuclear distances and
orbital symmetries [21] While the Hamiltonian matrix elements are calculated using atomic
reference densities the remaining terms to the KohnndashSham energy are parameterized from DFT
reference calculations of a few reference molecules per atom pair The accuracy is very much
improved by the self-consistent charge (SCC) extension Two computational codes were used
deMonNano code [22] and DFTB+ code [23] The first code has dispersion correction [24]
implemented to account for weak interactions and was used to obtain the layered bulk structure of
COFs and their formation energies The performance for interlayer interactions has been tested
previously for graphite [24] The second code which can perform calculations using k-points was
used to calculate the electronic properties (band structure and density of states) Band gaps have
been calculated as an additional stability indicator While these quantities are typically strongly
underestimated in standard LDA- and GGA-DFT calculations they are typically in the correct range
within the DFTB method For validation of our method we have calculated some of the structures
using Density Functional Theory (DFT) as implemented in ADF code [2526]
Periodic boundary conditions were used to represent frameworks of the crystalline solid state The
conjugatendashgradient scheme was chosen for the geometry optimization The atomic force tolerance of
3 times 10minus4 eVAring was applied The optimization using Γ-point approximation was performed with the
deMon-Nano code on 2times2times4 supercells Some of the monolayers were also optimized using the
DFTB+ code on elementary unit cells in order to validate the calculations within the Γ-point
approximation The number of k-points has been determined by reaching convergence for the total
energy as a function of k-points according to the scheme proposed by Monkhorst and Pack [27]
118
Band structures were computed along lines between high symmetry points of the Brillouin zone with
50 k-points each along each line XRD patterns have been simulated using Mercury software [2829]
We have also performed first-principles DFT calculations at the PBE [30] DZP [31] level to support
our results quantitatively For simplicity we have used finite structures instead of bulk crystals
Supporting Information
Table S1 The calculated bond lengths and angles of all studied COFs Corresponding experimental values found from the literature are shown in brackets
COF Building
Blocks
C-B B-O O-C OBO
COF-1 I-a 1498 (1600) 1393 (1509) 1203 (1200)
COF-1M I-b 1497 1393 1203
COF-2M I-c 1497 1392 1203
COF-3M I-d 1496 1392 1201
PPy-COF I-e 1498 1393 1202 (1190)
COF-5 II-a 1496 (1530) 1399 (1367) 1443 (1384) 1135dagger (1139)
COF-10 II-b 1495 (1553) 1399 (1465) 1443 (1385) 1135dagger (1120)
COF-8 II-c 1495 (1548) 1399 (1479) 1443 1135dagger
COF-6 II-d 1496 1399 1443 1135dagger
TP COF II-e 1496 (1542) 1399 (1396) 1444 (1377) 1135dagger (1182)
COF-4M III-a 1496 1398 1449 1135dagger
COF-5M III-b 1496 1398 1449 1136dagger
COF-6M III-c 1496 1399 1451 1134dagger
COF-7M III-d 1496 1398 1449 1136dagger
TP COF-1M III-e 1496 1398 1450 1136dagger
COF-8M IV-a 1496 1398 1445 1131dagger
COF-9M IV-b 1495 1398 1444 1131dagger
119
COF-10M IV-c 1495 1391 1418 1126dagger
COF-11M IV-d 1498 1399 1450 1134dagger
TP COF-2M IV-e 1499 1399 1447 1134dagger
B3O3 connectivity dagger C2B2O connectivity
It can be noticed that the bond lengths of experimentally known COF-1 is much larger compared to
our optimized bond lengths as well as that of other synthesized COFs
Figure S2 Calculated C-C bond lengths in connectors and linkers Hydrogen atoms are omitted for clarity
Table S3 The calculated unit cell parameters a interlayer distance d Aring+ and mass density ρ gcm-3
] for serrated (Sa serrated armchair Sz serrated zigzag) and inclined (Ia inclined armchair Iz inclined zigzag) stacked COFs
COF Building
Blocks
a d ρ
Sa Sz Ia Iz Sa Sz Ia Iz
COF-1 I-a 1502 343 343 097 097
COF-1M I-b 2241 341 342 069 069
COF-2M I-c 1492 340 339 097 097
COF-3M I-d 0747 341 342 157 156
PPy-COF I-e 2232 341 341 086 086
120
COF-5 II-a 3014 342 342 341 340 057 057 058 058
COF-10 II-b 3758 341 341 342 340 046 046 046 046
COF-8 II-c 2251 341 341 342 342 073 073 072 072
COF-6 II-d 1505 342 341 340 340 105 106 106 106
TP COF II-e 3750 342 341 342 342 052 052 052 052
COF-4M III-a 2171 344 344 345 344 074 074 074 074
COF-5M III-b 2915 343 342 343 343 056 056 056 056
COF-6M III-c 1833 341 341 342 341 084 084 084 084
COF-7M III-d 1083 344 343 340 344 131 131 132 131
TP COF-1M III-e 2905 343 342 343 342 066 067 066 066
COF-8M IV-a 1748 341 341 342 342 142 142 142 142
COF-9M IV-b 2176 341 341 341 342 119 119 119 119
COF-10M IV-c 2254 340 340 340 340 128 128 128 128
COF-11M IV-d 1512 341 341 340 340 171 171 171 171
TP COF-2M IV-e 2173 340 340 340 340 137 137 137 137
REF-I I 0773 349 345 148 15
REF-III III 1445 348 349 106 106
Table S4 The calculated energies [kJ mol-1
] per bond formed between building blocks for serrated (Sa serrated armchair Sz serrated zigzag) and inclined (Ia inclined armchair Iz inclined zigzag) stacked COFs Ecb ndash condensation energy Esb ndash stacking energy and Efb ndash COF formation energy (Efb = Ecb + Esb) The calculated band gaps ∆ eV+ are given as well
COF Sa Sz Ia Iz
Esb Efb Δ Esb Efb Δ Esb Efb Δ Esb Efb Δ
-1 -2810 -1904 36 -2786 -1880 36
-1M -4426 -3477 30 -4389 -3440 30
-2M -5967 -5011 30 -5833 -4877 30
121
-3M -2667 -1904 40 -2591 -1828 40
PPy- -5916 -5058 26 -5865 -5007 26
-5 -3052 -2841 27 -3051 -2840 26 -3275 -3064 26 -3297 -3086 26
-10 -3868 -3551 26 -3869 -3552 25 -3830 -3513 26 -4293 -3976 25
-8 -4615 -4352 27 -4614 -4351 27 -4571 -4308 27 -4573 -4310 27
-6 -2978 -2793 30 -2978 -2793 30 -3117 -2932 30 -3090 -2905 30
TP -4557 -4326 26 -4562 -4331 26 -4511 -4280 26 -4511 -4280 26
-4M -1811 -1844 28 -1813 -1846 28 -1769 -1802 28 -1880 -1913 28
-5M -2626 -2619 27 -2630 -2623 26 -2597 -2590 26 -2600 -2593 26
-6M -3375 -3361 28 -3381 -3367 28 -3487 -3473 28 -3349 -3335 28
-7M -1724 -1894 32 -1726 -1896 31 -2333 -2503 32 -1866 -2036 31
TP -1M -3327 -3341 26 -3334 -3348 26 -3340 -3354 26 -3353 -3367 26
-8M -2897 -3684 21 -2895 -3682 21 -2971 -3758 21 -2983 -3770 21
-9M -3732 -4568 20 -3733 -4569 20 -3684 -4520 20 -3706 -4542 20
-10M -4512 -5459 21 -4513 -5460 21 -4491 -5438 21 -4495 -5442 21
-11M -2831 -3234 24 -2834 -3237 24 -2816 -3219 24 -2830 -3233 24
TP -2M -4500 -4470 20 -4505 -4475 20 -4495 -4465 20 -4496 -4466 20
122
Appendix E
Stability and electronic properties of 3D covalent organic frameworks
Binit Lukose Agnieszka Kuc and Thomas Heine
Prepared for publication
Abstract Covalent Organic Frameworks (COFs) are a class of covalently linked crystalline nanoporous
materials versatile for nanoelectronic and storage applications 3D COFs in particular have very
large pores and low-mass densities Extensive theoretical studies of their energetic and mechanical
stability as well as their electronic properties are discussed in this paper All studied 3D COFs are
energetically stable materials though mechanical stability does not exceed 20 GPa Electronically all
COFs are semiconductors with band gaps depending on the number of sp2 carbon atoms present in
the linkers similar to 3D MOF family
Introduction
Covalent Organic Frameworks (COFs)1-3 comprise an emerging class of crystalline materials that
combines organic functionality with nanoporosity COFs have organic subunits stitched together by
covalent entities including boron carbon nitrogen oxygen or silicon atoms to form periodic
frameworks with the faces and edges of molecular subunits exposed to pores Hence their
applications can range from organic electronics to catalysis to gas storage and sieving4-7 The
properties of COFs extensively depend on their molecular constituents and thus can be tuned by
rational chemical design and synthesis289 Step by step reversible condensation reactions pave the
123
way to accomplish this target Such a reticular approach allows predicting the resulting materials and
leads to long-range ordered crystal structures
Since their first mentioning in the literature13 COFs are under theoretical investigation10-17 mainly for
gas storage applications Methods such as doping18-24 and organic linker functionalization2526 have
been suggested to improve their storage capacities In addition to the moderate pore size and
internal surface area COFs have the privileges of a low-weight material as they are made of light
elements Hence their gravimetric adsorption capacity is remarkably high10 The lowest mass density
ever reported for any crystalline material is that for COF-108 (018 gcm-3) Also their stronger
covalent bonds compared to related Metal-Organic Frameworks (MOFs)27 endorse thermodynamic
strength These genuine qualities of COFs make them attractive for hydrogen storage investigations
Crystallization of linked organic molecules into 2D and 3D forms was achieved in the years 20051 and
20073 respectively by the research group of O M Yaghi Several COFs have been synthesized since
then and some of the 2D COFs has been proven good for electronic or photovoltaic applications428-33
However the growth in this area appears to be slow compared to rapidly developing MOFs albeit
the promising H2 adsorption measurements53435 and a few synthetic improvements736-42
COF-102 and COF-103 were synthesized3 by the self-condensation of the non-planar tetra(4-
dihydroxyborylphenyl)methane (TBPM) and tetra(4-dihydroxyborylphenyl)silane (TBPS) respectively
(see Fig 1 for the building blocks and COF geometries) The co-condensation of these compounds
with triangular hexahydroxytriphenylene (HHTP) results in COF-108 and COF-105 respectively with
different topologies Yaghi et al3 have reported the formation of highly symmetric topologies ctn
(carbon nitride dI 34 ) and bor (boracite mP 34 ) when tetrahedral building blocks are linked
together with triangular ones The topology names were adopted from reticular chemistry structure
resource (RCSR)43 Simulated Powder X-ray diffraction in comparison with experimental powder
spectrum suggested ctn topology for COF-102 103 and 105 and bor topology for COF-108 The
condensation of tert-butylsilanetriol (TBST) and TBPM leads to the formation of COF-20244 This was
reported as ctn topology The COF reactants and schematic diagrams of ctn and bor topologies are
given in Figure1 The assembly of tetrahedral and linear units is very likely to result in a diamond-like
form COF-300 was formed according to the condensation of tetrahedral tetra-(4-anilyl)methane
(TAM) and linear terephthaldehyde (TA)45 However the synthesized structure was 5-fold
interpenetrated dia-c5 topology43
In this work we present theoretical studies of 3D COFs using density functional based methods to
explore their structural electronic energetic and mechanical properties Our previous studies on 2D
COFs4647 had resulted in questioning the stability of eclipsed arrangement of layers (AA stacking) and
124
suggesting energetically more stable serrated and inclined packing In this paper we attempt to
explore the stability and electronic properties of the experimentally known 3D COFs namely COF-
102 103 105 108 202 and 300 In the present studies we follow the reticular assembly of the
molecular units that form low-weight 3D COFs Considering the structural distinction from MOFs
COFs lack the versatility of metal ingredients what results in diverse properties48 A collective study is
then carried out to understand the characteristics and drawbacks of COFs
Figure 1 (Upper panel) Building blocks for the construction of 3D COFs (Lower panel) lsquoctnrsquo and lsquoborrsquo
networks formed by linking tetrahedral and triangular building units
Methods
COF structures were fully optimized using Self-consistent Charge -Density Functional based Tight-
Binding (SCC-DFTB) level of theory4950 Two computational codes were used deMonNano51 and
125
DFTB+52 The first code which has dispersion correction53 implemented to account for weak
interactions was used for the geometry optimization and stability calculations The second code
which can perform calculations using k-point sampling was used to calculate the electronic
properties (band structure and density of states) The number of k-points has been determined by
reaching convergence for the total energy as a function of k-points according to the scheme
proposed by Monkhorst and Pack54 Periodic boundary conditions were used to represent
frameworks of the crystalline solid state Conjugatendashgradient scheme was chosen for the geometry
optimization Atomic force tolerance of 3 times 10minus4 eVAring was applied The optimization using Γ-point
approximation was performed on rectangular supercells containing more than 1000 atoms For
validation of our method we have calculated energetic stability using Density Functional Theory (DFT)
at the PBE55DZP56 level as implemented in ADF code5758 using cluster models The cluster models
contain finite number of building units and correspond to the bulk topology of the COFs XRD
patterns have been simulated using Mercury software5960
In this work we continued to use the traditional nomenclature of the experimentally known COFs All
of the structures have the same tetrahedral blocks and differ only in the central sp3 atom (carbon or
silicon) that is included in our nomenclature
Bulk modulus (B) of a solid at absolute zero can be calculated as
(1) B = 2
2
dV
EdV
where V and E are the volume and energy respectively
Owing to the dehydration reactions we have calculated the formation (condensation) energy of each
COF formed from monomers (building blocks) as follows
(2) EF = Etot + n EH2Otot ndash (m1 Ebb1
tot + m2 Ebb2tot)
where Etot -- total energy of the COF EH2Otot -- total energy of water molecule Ebb1
tot and Ebb2tot -- total
energies of interacting building blocks n m1 m2 -- stoichiometry numbers
Results and Discussions
Structure and Stability
We have optimized the atomic positions and cell dimensions of the COFs in the experimentally
determined topologies Cell parameters in comparison with experimental values are given in Table 1
The calculated bond lengths are C-B = 150 C-C = 139-144 (COF-300) C(sp3)-C = 156 C-O = 142 B-
126
O = 140 Si-C = 188 Si-O = 186 C-N = 131-136 Aring These values agree within 6 error with the
experimental values34445
Mass densities (see Table 1) of COFs range between 02 to 05 g cm-3 and are smaller for the silicon at
the tetrahedral centers This implies that the presence of silicon atoms impacts more on the cell
volume than on the total mass That means the replacement of sp3 C with Si in COF-108 can change
its mass density to a slightly lower value To our best knowledge among all the natural or
synthesized crystals COF-108 has the lowest mass-weight
In order to validate our optimized structures we have simulated X-ray Diffraction of each COF and
compared them with the available experimental spectra (see Figure2) Almost all of the simulated
XRDs have excellent correlation in the peak positions with the experiment Only COF-300 shows
somehow significant differences in the intensities These differences may be attributed to the
presence of guest molecules in the synthesized COF-30045
Table 1 The calculated cell parameters [Aring] mass density ρ gcm-3
+ band gap Δ eV+ bulk modulus B GPa+
and formation energy EF [kJ mol-1
] for all the studied 3D COFs Experimental values are given in brackets
along with the cell parameters HOMO-LUMO gaps [eV] of the building blocks are also given in brackets
along with the band gaps
Structure Building
Blocks
Cell
parameters
ρ Δ B EF
COF-102 (C) TBPM 2707 (2717) 043 39 (43) 206 -14995
COF-103 (Si) TBPS 2817 (2825) 039 38 (41) 139 -14547
COF-105-C TBPM HHTP 4337 019 33 (43 34) 80 -18080
COF-105 (Si) TBPS HHTP 4444 (4489) 018 32 (41 34) 79 -17055
COF-108 (C) TBPM HHTP 2838 (2840) 017 32 (43 34) 37 -17983
COF-108-Si TBPS HHTP 2920 016 30 (41 34) 29 -18038
COF-202 (C) TBPM TBST 3047 (3010) 050 42 (43 139) 143 -7954
COF-202-Si TBPS TBST 3157 046 39 (41 139) 153 -7632
COF-300 (C) TAM TA 2932 925 049 23 (41 26) 144 -40286
127
(2828 1008)
COF-300-Si TAS TA 3039 959 044 23 (40 26) 133 -39930
tetra-(4-anilyl)silane
Figure 2 Simulated XRDs are compared with the experimental patterns from the literature COF-300
exhibits some differences between the simulated and experimental XRDs while others show reasonably
good match
The bulk modulus (B) is a measure of mechanical stability of the materials The calculated values of B
are given in Table 1 Compared to the force-field based calculation of bulk modulus by Schmid et
al61 these values are larger The bulk modulus of COF-105 and COF-108 are relatively small
compared with other COFs Considering that the two COFs differ only in the topology it may be
concluded that ctn nets are mechanically more stable compared to bor COFs with carbon atoms in
the center are mechanically more stable than those with silicon atoms Bulk modulus of COF-102
103 COF-202 and interpenetrated COF-300 are higher than many of the isoreticular MOFs62 and
comparable to IRMOF-6 (1241 GPa) MOF-5 (1537 GPa)63 bNon-interpenetrated COF-300 (single
framework dia-a topology43) has much lower bulk modulus of only 317 GPa
Calculated formation energies (Table 1) are given in terms of reaction energies according to Eq 2
Condensation of monomers to form bulk 3D structures is exothermic in all the cases supporting
reticular approach The presence of C or Si at the vertex center does not show any particular trend in
the formation energies We have calculated the formation energy of non-interpenetrated COF-300
(dia-a) to be -39346 kJ mol-1 which is comparable to the interpenetrated cases For a quantitative
comparison we have performed DFT calculations at the PBEDZP level as implemented in ADF code
on finite structures The obtained formation energies for COF-105-C COF-105 (Si) COF-108 (C) COF-
108-Si are -9944 -9968 -1245 -12436 respectively They are in a reasonable agreement with the
128
DFTB results A comparison between COF-105 and COF-108 suggests that bor nets are energetically
more favored than ctn nets
Electronic Properties
Band gaps (Δ) calculated for the 3D COFs are in the range of 23-42 eV (see Table 1) which show
their semiconducting nature similar to the hexagonal 2D COFs47 and MOFs62 The band gap
decreases with the increase of conjugated rings in the unit cell relative to the number of other atoms
Si atoms in comparison with carbon atoms at the tetrahedral centers give a relatively smaller Δ This
is evident from the partial density of states of C (sp3) in COF-102 and Si in COF-103 plotted in Figure3
Carbon atoms define the band edges of the PDOS (see Figure4) The band gaps of COF-105 and COF-
108 differ only slightly (see Figure5) which means that the band gap is nearly independent of the
topology This is because for each atom the coordination number and the neighboring atoms remain
the same in both ctn and bor networks albeit the topology difference DOS of non-interpenetrated
(dia-a) and 5-fold interpenetrated (dia-c5) COF-300 are plotted in Figure6 It may be concluded from
their negligible differences that interpenetration does not alter the DOS of a framework We have
shown similar results for 2D COFs47
Figure 3 Partial density of states (DOS) of sp3 C in COF-102 (top) and Si in COF-103 (bottom) The latter is
inverted for comparison The Fermi level EF is shifted to zero
129
Figure 4 Partial and total density of states of COF-103 The Fermi level EF is shifted to zero
Figure 5 Density of states of COF-108 (bor) and COF-105 (ctn) The two COFs differ only in the topology
130
Figure 6 Density of states of non-interpenetrated (dia-c1) and 5-interpenetrated (dia-c5) COF-300
We also have calculated HOMO-LUMO gaps of the molecular building blocks (see Table 1) In
comparison band gaps of COFs are slightly smaller than the smallest of the HOMO-LUMO gaps of the
building units
Conclusion
In summary we have calculated energetic mechanical and electronic properties of all the known 3D
COF using Density Functional Tight-Binding method Energetically formation of 3D COFs is favorable
supporting the reticular chemistry approach Mechanical stability is in line with other frameworks
materials eg MOFs and bulk modulus does not exceed 20 GPa Also all COFs are semiconducting
with band gaps ranging from 2 to 4 eV Band gaps are analogues to the HOMO-LUMO gaps of the
molecular building units We believe that this extensive study will define the place of COFs in the
broad area of nanoporous materials and the information obtained from the work will help to
strategically develop or modify porous materials for the targeted applications
131
Appendix F
Structure electronic structure and hydrogen adsorption capacity of porous aromatic frameworks
Binit Lukose Mohammad Wahiduzzaman Agnieszka Kuc and Thomas Heine
Prepared for publication
Abstract
Diamond-like porous aromatic frameworks (PAFs) form a new family of materials that contain only
carbon and hydrogen atoms within their frameworks These structures have very low mass densities
large surface area and high porosity Density-functional based calculations indicate that crystalline
PAFs have mechanical stability and properties similar to those of Covalent Organic Frameworks Their
exceptional structural properties and stability make PAFs interesting materials for hydrogen storage
Our theoretical investigations show exceptionally high hydrogen uptake at room temperature that
can reach 7 wt a value exceeding those of well-known metal- and covalent-organic frameworks
(MOFs and COFs)
Introduction
Porous materials have been widely investigated in the fields of materials science and technology due
to their applications in many important fields such as catalysis gas storage and separation template
materials molecular sensors etc1-3 Designing porous materials is nowadays a simple and effective
strategy following the approach of reticular chemistry4 where predefined building blocks are used to
132
predict and synthesize a topological organization in an extended crystal structure The most famous
and striking examples of such materials are Metal- and Covalent-Organic Frameworks (MOFs and
COFs)56 These new nanoporous materials have many advantages high porosity and large surface
areas lowest mass densities known for crystalline materials easy functionalization of building blocks
and good adsorption properties
Gas storage and separation by physical adsorption are very important applications of such
nanoporous materials and have been major subjects of science in the last two decades These
applications are based on certain physical properties namely presence of permanent large surface
area and suitable enthalpy of adsorption between the host framework and guest molecules
Attempts to produce materials with large internal surface area have been successful and some of the
notable accomplishments include the synthesis of MOF-2107 (BET surface area of 6240 m2 g-1 and
Langmuir surface area of 10400 m2 g-1) NU-1008 (BET surface area 6143 m2 g-1) or 3D COFs9 (BET
surface area 4210 m2 g-1 for COF-103)
More recently a new family of porous materials emerged So-called porous-aromatic frameworks
(PAFs) have surface areas comparable to MOFs eg PAF-110 has a BET surface area of 5640 m2 g-1 and
Langmuir surface area of 7100 m2 g-1 Since they are composed of carbon and hydrogen only they
have several advantages over frameworks containing heavy elements MOFs with coordination bonds
often suffer from low thermal and hydrothermal stability what might limit their applications on the
industrial scale The coordination bonds can be substituted with stronger covalent bonds as it was
realized in the case of COFs6 however this lowers significantly their surface areas comparing with
MOFs
Besides its exceptional surface area PAF-110 has high thermal and hydrothermal stability and
appears to be a good candidate for hydrogen and carbon dioxide storage applications PAFs have
topology of diamond with C-C bonds replaced by a number of phenyl rings (see Figure 1)
Experimentally obtained PAF-110 and PAF-1111 have eg one and four phenyl rings respectively
connected by tetragonal centers (carbon tetrahedral nodes) The simulated and experimental
hydrogen uptake capacities of such PAFs exceed the DOE target12
In this paper we have studied structural electronic and adsorption properties of PAFs using Density
Functional based Tight-Binding (DFTB) method1314 and Quantized Liquid Density Functional Theory
(QLDFT)15 We have found exceptionally high storage capacities at room temperature what makes
PAFs very good materials for hydrogen storage devices beyond well-known MOFs and COFs We have
compared our results to widely-used classical Grand Canonical Monte-Carlo (GCMC) calculations
reported in the literature We have also studied other properties of these materials such as
133
structural energetic electronic and mechanical We explored the structural variance of diamond
topology by individually placing a selection of organic linkers between carbon nodes This generally
changes surface area mass density and isosteric heat of adsorption what is reflected in the
adsorption isotherms
Methods
Using a conjugate-gradient method we have fully optimized the crystal structures (atomic positions
and unit cell parameters) of PAFs The calculations were performed using Dispersion-corrected Self-
consistent Charge density-functional based tight-binding (DFTB) method as implemented in the
deMonNano code1617 Periodic boundary conditions were applied to a (3x3x3) supercell thus
representing frameworks of the crystalline solid state Electronic density of states (DOS) have been
calculated using the DFTB+ code18 with k-point sampling where the k space was determined by
reaching convergence for the total energy according to the scheme proposed by Monkhorst and
Pack19
Hydrogen adsorption calculations have been carried out using QLDFT within the LIE-0 approximation
thus including many-body interparticle interactions and quantum effects implicitly through the
excess functional The PAF-H2 non-bonded interactions were represented by the classical Morse
atomic-pair potential Force field parameters were taken from Han et al20 who originally developed
them for studying hydrogen adsorption in COFs that incorporate similar building blocks as PAFs The
authors adopted the approach to the shapeless particle approximation of QLDFT These PAF-H2
parameters have been determined by fitting first-principles calculations at the second-order Moslashllerndash
Plesset (MP2) level of theory using the quadruple-ζ QZVPP basis set with correction for the basis set
superposition error (BSSE) QLDFT calculations use a grid spacing of 05 Bohr radii and a potential
cutoff of 5000 K
Results and Discussion
Design and Structure of PAFs
We have designed a set of PAFs by replacing C-C bonds in the diamond lattice by a set of organic
linkers phenyl biphenyl pyrene para-terphenyl perylenetetracarboxylic acid (PTCDA)
diphenylacetylene (DPA) and quaterphenyl (see Figure 1) We label the respective crystal structures
as PAF-phnl (PAF-301 in Ref 12) PAF-biph (PAF-302 in Ref 12) PAF-pyrn PAF-ptph(PAF-303 in Ref
12) PAF-PTCDA PAF-DPA and PAF-qtph (PAF-304 in Ref 12) respectively Such a design of
frameworks should result in materials with high stability due to the parent diamond-topology and
pure covalent bonding of the network The selected linkers differ in their length width and the
134
number of aromatic rings These should play an important role for hydrogen adsorption properties
aromatic systems exhibit large polarizabilities and interact with H2 molecules via London dispersion
forces Long linkers introduce high pore volume and low mas-weight to the network while wide
linkers offer large internal surface area and high heat of adsorption Hence long linkers are of
advantage for high gravimetric capacity while wide linkers enhance volumetric capacity Proper
optimization of the linker size should result in a perfect candidate for hydrogen storage applications
Figure 1 a) Schematic diagram of the topology of PAFs blue spheres and grey pillars represent carbon
tetragonal centers and organic linkers respectively b) Linkers used for the design of PAFs i) phenyl ii)
biphenyl iii) pyrene iv) DPA v) para-Terphenyl vi) PTCDA and vii) quaterphenyl
Selected structural and mechanical properties of the investigated PAF structures are given in Table 1
Frameworks created with the above mentioned linkers have mass densities that range from 085 g
cm-3 (for phenyl) to 01 g cm-3 (for quaterphenyl) The lowest mass-density obtained for a crystal
structure was for COF-108 with = 0171 g cm-39 Our results show that PAF-DPA and PAF-ptph have
mass densities close to the one of COF-108 while PAF-qtph with = 01 g cm-3 exhibits the lowest
for all the PAFs investigated in this study
While the large cell size and the small mass density of PAF-qtph are an advantage for high
gravimetric hydrogen storage capacity other PAFs (eg PAF-pyrn with wide linker) would
compromise gravimetric for high volumetric capacity As both of them are important for practical
applications a balance between them is crucial
Table 1 Calculated cell parameters a (a=b=c α=β=γ=60⁰) mass densities () formation energies (EForm) band
gaps () bulk moduli (B) and H2 accessible free volume and surface area of PAFs studied in the present work
In parenthesis are given HOMU-LUMO gaps of the corresponding saturated linkers
PAFs
a
(Aring)
ρ
(g cm-3)
EForm
(kJ mol-1)
Δ
(eV)
B
(GPa)
H2 accessible
free volume
H2 accessible
surface area
135
() (m2 g-1)
PAF-phnl 97 085 -121 47 (55) 360 35 2398
PAF-biphl 167 032 -122 36 (40) 132 73 5697
PAF-pyrn 166 042 -124 26 (28) 192 66 5090
PAF-DPA 210 019 -122 35 (37) 87 84 7240
PAF-ptph 237 016 -119 32 (33) 56 86 6735
PAF-PTCDA 236 024 -122 18 (19) 95 81 5576
PAF-qtphl 308 010 -119 29 (30) 35 91 7275
Energetic and Mechanical Properties
We have investigated energetic stability of PAFs by calculating their formation energies We regarded
the formation of PAFs as dehydrogenation reaction between saturated linkers and CH4 molecules
For a unit cell containing n carbon nodes and m organic linkers the formation energy (EForm) is given
by
( )
where Ecell EL and
are the total energies of the unit cell saturated linkers CH4 and H2
molecules respectively This excludes the inherent stability of linkers and represents the energy for
coordinating linkers with sp3 carbon atoms during diamond-topology formation All the formation
energies calculated in the present work are given in Table 1 Negative values indicate that the
formation of PAFs is exothermic The values per formula unit do not deviate significantly for different
PAF sizes and shapes
Although diamond is the hardest known material insertion of longer linkers diminishes its
mechanical strength to some extent In order to study the mechanical stability of PAFs we have
calculated their bulk moduli which is the second derivative of the cell energy with respect to the cell
volume (see Table 1) The highest value that we have obtained is for PAF-phnl with B = 36 GPa This is
over ten times smaller than the bulk modulus of pure diamond 514 GPa (calculated at the DFTB
level)21 and 442 GPa (experimental value)22 Bulk modulus should scale as 1V (V - volume) if all
bonds have the same strength We have plotted such a function for PAFs and other framework
136
materials23 in Figure 2 As PAFs have various types of CmdashC bonds there are minor deviations from
the perfect trend
Figure 2 Calculated bulk modulus B as function of the volume per atom PAFs are highlighted in blue and
compared with other framework materials Red curve denotes the fitting to 1V function (V - volume)
The stability of PAF-phnl is similar to that of Cu-BTC MOF and much higher than for other MOFs such
as MOF-5 -177 and -20524 and 3D COFs23 Subsequent addition of phenyl rings decreases B the
lowest value of 35 GPa has been found for PAF-qtph which is in the same order as for COF-108 In
general all the studied PAFs except for PAF-phnl have bulk moduli in line with 3D COFs25 When the
organic linker is extended in the width (cf PAF-biph and PAF-pyrn) the mechanical stability increases
Electronic Properties
All PAF structures are semiconductors similar to COFs The band gaps of PAFs range from 18 to 47
eV (see Table 1) Interesting trends may be observed from the band gaps of this iso-reticular series
In comparison with diamond which has a band gap of 55 eV it may be understood that subsequent
insertion of benzene groups between carbon atoms in diamond decreases the band gap This is easily
understood as the sp3 responsible for the semiconducting character become far apart with large
number of π-electrons in between which tend to close the gap More importantly the values of
band gaps are directly related to the HOMO-LUMO gaps of the corresponding saturated linkers
which are just slightly larger Also more extended linkers (cf PAF-biph and PAF-pyrn or PAF-ptph and
PAF-PTCDA) reduce the band gap
In general conjugated rings reduce the band gap more effectively than CmdashC bond networks (cf PAF-
DPA PAF-ptph or PAF-qtphl) The extreme cases for this behaviour are graphene which is metallic
137
and diamond which is an insulator Overall the band gap may be tuned by placing suitable linkers in
the diamond network Similar results have been reported for MOFs2627
We have calculated the electronic density of states (DOS) for all the PAF structures Figure 3 a shows
carbon DOS of PAFs arranged in such a way that the band gaps decrease going from the top to the
bottom of the figure PAFs with larger band gaps have larger population of energy states at the top of
valence band and bottom of conduction band whereas for linkers with smaller band gaps the
distribution of energy states is rather spread In order to study the impact of sp3 carbon atoms in the
DOS we plotted their contributions against the total carbon DOS in the cases of PAF-phnl and PAF-
pyrn as examples (see Figure 3 b and c) In both cases the sp3 carbons exhibit no energy states in the
band gap and in the close vicinity of band edges
Figure 3 (a) Carbon density of states of all calculated PAFs From top to bottom the value of band gap
decreases (b) and (c) projected DOS on sp3 C atoms of PAF-phnl and PAF-pyrn respectively The vertical
dashed line indicates Fermi level EF
Hydrogen Adsorption Properties
One of the potential applications of PAFs is hydrogen adsorption We have calculated the gravimetric
and volumetric capacities and analyzed them to understand the contributions of the linkers on the
138
hydrogen uptake in different range of temperatures and pressures H2 accessible free volume and
surface area of the PAFs are given in Table 1 In order to assess the excess adsorption capacity the
free pore volume is necessary In our simulation the free pore volume is defined to be that where
the H2-host interaction energy is less than 5000K This covers all sites where H2 is attracted by the
host structure and excludes the repulsion area close to the framework The solvent accessible surface
areas of the PAFs were determined by rolling a probe molecule along the van der Waals spheres of
the atoms A probe molecule of radius 148 Aring was used which is consistent with the Lennard-Jones
sphere of hydrogen and commonly used in various H2 molecular simulations28
Very large surface areas per mass unit have been observed for all PAFs except for PAF-phnl PAF-DPA
and PAF-qtph for example have 7240 and 7274 m2 g-1 H2 accessible surface area respectively For
comparison highly porous MOF-210 and NU-100 give values of 6240 and 6143 m2 g-1 BET surface
areas respectively determined from the experimental adsorption isotherms78 It is worth
mentioning that longer linkers expand the pore and increase the surface area per unit volume and
unit mass Wider linkers provide a higher surface area per unit volume however they possess larger
mass density and hence the surface area per unit mass gets lower
Figure 4 shows the gravimetric and volumetric total and excess adsorption isotherms of PAFs at 77K
The results show that the gravimetric (total and excess) H2 uptake is dependent on the linker length
The longest linker in the PAF-qtph structure gives the total and excess gravimetric capacity of 32 and
128 wt respectively which is larger than for highly porous crystalline 3D COFs29 These numbers
are calculated for the limit pressure of 100 bar For the shortest linker in PAF-phnl we have obtained
only around 25 wt for the same pressure Using grand canonical Monte-Carlo methods (GCMC)
Lan et al12 have predicted total and excess adsorption capacities of PAF-qtph of 224 and 84 wt
respectively The deviations in results are attributed to the differences in both methods where
different force fields are used to describe atom-atom interactions
The volumetric adsorption capacities lower with the size of the linker and the largest uptake we have
found for the PAF-pyrn (46 and 35 kg m-3 total and excess respectively) while very low values were
found for PAF-qtph (30 and 145 kg m-3 total and excess respectively) at 35-bar pressure It could be
predicted that for PAF-phnl the volumetric adsorption reaches the larges values however due to its
very compact crystal structure it reaches saturation at the low-pressure region and does not exceed
30 kg m-3 for the total uptake From the excess adsorption isotherms however PAF-phnl is the best
adsorbing system for low-pressure application as it gets saturated at 11 bar by adsorbing 266 kg m-3
of H2 Generally we have observed that PAFs with lower pore sizes exhibited saturation of volumetric
capacities at lower pressures
139
Figure 4 Calculated H2 uptake of PAFs at 77 K (a)-(b) gravimetric and (c)-(d) volumetric total (upper panel)
and excess (lower panel) respectively
We have also calculated the adsorption performance of PAFs at room temperature The gravimetric
total adsorption isotherms are shown in Figure 5 Similar to the results obtained for T=77 K PAF-
qtph exhibits the highest total gravimetric capacity at room temperature which is as high as 732 wt
at 100 bar Lan et al12 have predicted the uptake of 653 wt at 100 bar using GCMC simulations
These results exceed the 2015 DOE target of 55 wt 3031 On the other hand at more applicable
pressure of 5 bar the uptake reaches only 05 wt This suggests however that the delivery amount
(approximately the difference between the uptakes at 100 and 5 bar) is still more than the DOE
target Phenyl linker at room temperature does not exceed the gravimetric uptake of 1 wt at 100
bar
Figure 5 Calculated gravimetric total (left) and excess (right) H2 uptake of PAFs at 300 K
140
At 300K excess gravimetric capacity is rather low and does not exceed 075 wt even for large
pressure (see Figure 5)
Effects of interpenetration
Very often frameworks with large pores crystallize as interpenetrated lattices In most cases this is
an undesired fact due to reduction of the pore size and free volume For instance COF-300 which
has diamond topology was found to have 5-interpenetrated frameworks32
We have studied the effect of interpenetration in the case of PAF-qtph which has the largest pore
volume among the materials in this study Without any steric hindrance PAF-qtph may be
interpenetrated up to the order of four The two three and four interpenetrated networks are
named as PAF-qtph-2 PAF-qtph-3 and PAF-qtph-4 respectively and in each case the interpenetrated
structures possess translational symmetry Table 2 shows calculated mass densities and H2 accessible
free volume and surface area of non-interpenetrated and n-fold penetrated PAF-qtph Obviously the
mass density increases by one unit for each degree of interpenetration PAF-qtph has 91 of its
volume accessible for H2 which means that 9 of volume is occupied by the skeleton of the PAF
Hence each degree of interpenetration decreases the free volume by 9 H2 accessible surface area
per unit mass remains the same up to 3-fold interpenetration However PAF-qtph-4 results in much
less accessibility for H2
Table 2 Calculated mass densities () and H2 accessible free volume and surface area of non-interpenetrated
and n-fold interpenetrated PAF-qtph where n = 2 3 4
PAF
(g cm-3)
H2 accessible
free volume ()
H2 accessible
surface area
(m2 g-1)
PAF-qtph 010 91 7275
PAF-qtph-2 020 82 7275
PAF-qtph-3 030 73 7275
PAF-qtph-4 040 64 5998
Figure 6 shows the excess and total adsorption isotherms (gravimetric and volumetric) of non-
interpenetrated and interpenetrated PAF-qtph at 77 K With the increasing degree of
141
interpenetration saturation occurs at lower values of pressure This is due to the smaller pore size
resulted from the high-fold interpenetration The excess gravimetric capacity decreases by 18 - 2 wt
per each n-fold interpenetration a result of the decreasing free space around the skeleton It is to be
noted that the difference between the total and the excess adsorption capacities of PAF-qtph is quite
large however it decreases less for interpenetrated structures This is because the interpenetrated
frameworks occupy the free space that otherwise would be taken by hydrogen to increase the total
capacity but not the excess
Figure 6 Calculated H2 uptake at 77 K of non-interpenetrated and n-fold interpenetrated PAF-qtph with n = 2
3 4 (a)-(b) gravimetric and (c)-(d) volumetric total (lower panel) and excess (upper panel) respectively
On the other hand the interpenetrated frameworks have larger volumetric capacity what is easily
understandable due to the volume reduction Significant increase in excess volumetric capacity has
been obtained for PAF-qtph-2 in comparison with PAF-qtph whereas only slightly lower increase was
obtained for PAF-qtph-3 in comparison with PAF-qtph-2 (see Figure 6) PAF-qtph-4 exhibited even
lower increase compared to PAF-qtph-3 suggesting that only a certain degree of interpenetration is
appreciable Although non-interpenetrated PAF has a large amount of H2 gas in the bulk phase due
to the high pore volume its total volumetric uptake does not exceed that of the interpenetrated
PAFs
Almost linear gravimetric isotherms were obtained at room temperature for all studied PAFs
including the interpenetrated cases (see Figure 7) Large decrease of gravimetric capacity was noted
142
when the PAF was interpenetrated from around 7 wt down to 2 wt for 4-fold interpenetrated
PAF-qtph The excess gravimetric capacity however is the same independent of the n-fold
interpenetration and maximum of 06 wt was found for 100-bar pressure (see Figure 7)
Figure 7 Calculated gravimetric total (left) and excess (right) H2 uptake of non-interpenetrated and n-fold
interpenetrated PAF-qtph (n = 2 3 4) at 300 K
Conclusions
Following diamond-topology we have designed a set of porous aromatic frameworks PAFs by
replacing CmdashC bonds by various organic linkers what resulted in remarkably high surface area and
pore volume
Exothermic formation energies of PAFs calculated from the dehydrogenation of the linkers with CH4
indicate strong coordination of linkers in the diamond topology The studied PAFs exhibit bulk moduli
that are much smaller than diamond however in the same order as other porous frameworks such
as MOFs and COFs PAFs are semi-conductors with their band gaps dependent on the HOMO-LUMO
gaps of the linking molecules
Using quantized liquid density functional theory which takes into account inter-particle interactions
and quantum effects we have predicted hydrogen uptake capacities in PAFs At room temperature
and 100 bar the predicted uptake in PAF-qtph was 732 wt which exceeded the 2015 DOE target
At the same conditions other PAFs showed significantly lower uptakes At 77 K the PAFs with similar
pore sizes exhibited similar gravimetric uptake capacities whereas smaller pore size and larger
number of aromatic rings result in higher volumetric capacities In smaller pores saturation of excess
capacities occurred at lower pressures Interpenetrated frameworks largely decrease the amount of
hydrogen gas in the pores and increase the weight of the material however they are predicted to
have high volumetric capacities
143
References
(1) Eddaoudi M Moler D B Li H L Chen B L Reineke T M OKeeffe M Yaghi O M
Accounts of Chemical Research 2001 34 319
(2) Seo J S Whang D Lee H Jun S I Oh J Jeon Y J Kim K Nature 2000 404 982
(3) Ferey G Mellot-Draznieks C Serre C Millange F Accounts of Chemical Research 2005 38
217
(4) Yaghi O M OKeeffe M Ockwig N W Chae H K Eddaoudi M Kim J Nature 2003 423
705
(5) Eddaoudi M Kim J Rosi N Vodak D Wachter J OKeeffe M Yaghi O M Science 2002
295 469
(6) Cote A P Benin A I Ockwig N W OKeeffe M Matzger A J Yaghi O M Science 2005
310 1166
(7) Furukawa H Ko N Go Y B Aratani N Choi S B Choi E Yazaydin A O Snurr R Q
OKeeffe M Kim J Yaghi O M Science 2010 329 424
(8) Farha O K Yazaydin A O Eryazici I Malliakas C D Hauser B G Kanatzidis M G
Nguyen S T Snurr R Q Hupp J T Nature Chemistry 2010 2 944
(9) El-Kaderi H M Hunt J R Mendoza-Cortes J L Cote A P Taylor R E OKeeffe M Yaghi
O M Science 2007 316 268
(10) Ben T Ren H Ma S Cao D Lan J Jing X Wang W Xu J Deng F Simmons J M Qiu
S Zhu G Angewandte Chemie-International Edition 2009 48 9457
(11) Yuan Y Sun F Ren H Jing X Wang W Ma H Zhao H Zhu G Journal of Materials
Chemistry 2011 21 13498
(12) Lan J Cao D Wang W Ben T Zhu G Journal of Physical Chemistry Letters 2010 1 978
(13) Oliveira A F Seifert G Heine T Duarte H A Journal of the Brazilian Chemical Society
2009 20 1193
(14) Seifert G Porezag D Frauenheim T International Journal of Quantum Chemistry 1996 58
185
(15) Patchkovskii S Heine T Physical Review E 2009 80
(16) Heine T Rapacioli M Patchkovskii S Frenzel J Koester A M Calaminici P Escalante S
Duarte H A Flores R Geudtner G Goursot A Reveles J U Vela A Salahub D R deMon-nano edn ed
deMon 2009
(17) Zhechkov L Heine T Patchkovskii S Seifert G Duarte H A Journal of Chemical Theory
and Computation 2005 1 841
(18) BCCMS Bremen DFTB+ - Density Functional based Tight binding (and more)
(19) Monkhorst H J Pack J D Physical Review B 1976 13 5188
(20) Han S S Furukawa H Yaghi O M Goddard III W A Journal of the American Chemical
Society 2008 130 11580
(21) Kuc A Seifert G Physical Review B 2006 74
(22) Cohen M L Physical Review B 1985 32 7988
(23) Lukose B Kuc A Heine T manuscript in preparation 2012
(24) Lukose B Supronowicz B Petkov P S Frenzel J Kuc A Seifert G Vayssilov G N
Heine T physica status solidi (b) 2011
(25) Amirjalayer S Snurr R Q Schmid R Journal of Physical Chemistry C 2012 116 4921
(26) Gascon J Hernandez-Alonso M D Almeida A R van Klink G P M Kapteijn F Mul G
Chemsuschem 2008 1 981
(27) Kuc A Enyashin A Seifert G Journal of Physical Chemistry B 2007 111 8179
(28) Dueren T Millange F Ferey G Walton K S Snurr R Q Journal of Physical Chemistry C
2007 111 15350
(29) Furukawa H Yaghi O M Journal of the American Chemical Society 2009 131 8875
144
(30) US DOE Office of Energy Efficiency and Renewable Energy and The FreedomCAR and
Fuel Partnership 2009
httpwww1eereenergygovhydrogenandfuelcellsstoragepdfstargets_onboard_hydro_storage_explanatio
npdf
(31) US DOE USCAR Shell BP ConocoPhillips Chevron Exxon-Mobil T F a F P Multi-Year
Research Development and Demonstration Plan 2009
httpwww1eereenergygovhydrogenandfuelcellsmypppdfsstoragepdf
(32) Uribe-Romo F J Hunt J R Furukawa H Klock C OKeeffe M Yaghi O M Journal of the
American Chemical Society 2009 131 4570
145
Appendix G
A novel series of isoreticular metal organic frameworks realizing metastable structures by liquid phase epitaxy
Jinxuan Liu Binit Lukose Osama Shekhah Hasan Kemal Arslan Peter Weidler Hartmut
Gliemann Stefan Braumlse Sylvain Grosjean Adelheid Godt Xinliang Feng Klaus Muumlllen Ioan-
Bogdan Magdau Thomas Heine and Christof Woumlll
Prepared for publication
Highly crystalline molecular scaffolds with nm-sized pores offer an attractive basis for the fabrication
of novel materials The scaffolds alone already exhibit attractive properties eg the safe storage of
small molecules like H2 CO2 and CH41-4 In addition to the simple release of stored particles changes
in the optical and electronic properties of these nanomaterials upon loading their porous systems
with guest molecules offer interesting applications eg for sensors5 or electrochemistry6 For the
construction of new nanomaterials the voids within the framework of nanostructures may be loaded
with nm-sized objects such as inorganic clusters larger molecules and even small proteins a
process that holds great potential as for example in drug release7-8 or the design of novel battery
materials9 Furthermore meta-crystals may be prepared by loading metal nanoparticles into the
pores of a three-dimensional scaffold to provide materials with a number of attractive applications
ranging from plasmonics10-12 to fundamental investigations of the electronic structure and transport
properties of the meta-crystals13
146
In the last two decades numerous studies have shown that MOFs also termed porous coordination
polymers (PCPs) fulfill many of the aforementioned criteria Although originally developed for the
storage of hydrogen eg in tanks of automobiles MOFs have recently found more technologically
advanced applications as sensors5 matrices for drug release7-8 and as membranes for enantiomer
separation14 and for proton conductance in fuel cells15 Although pore-sizes available within MOFs1
are already sufficiently large to host metal-nanoclusters such as Au5516-17 the actual realization of
meta-crystals requires in addition to structural requirements a strategy for the controlled loading
of the nanoparticles (NPs) into the molecular framework Simply adding NPs to the solutions before
starting the solvothermal synthesis of MOFs has been reported18 but this procedure does not allow
for controlled loading The layer-by-layer growing process of SUFMOFs allows the uptake of
nanosized objects during synthesis including the fabrication of compositional gradients of different
NPs Those guest NPs can range from pure metal metal oxide or covalent clusters over one-
dimensional objects such as nanowires carbon or inorganic nanotubes to biological molecules such
as drugs or even small proteins If the loading happens during synthesis alternating layers of
different NPs can be realized
The introduction of procedures for epitaxial growth of MOFs on functionalized substrates19 was a
major step towards controlled functionalization of frameworks Epitaxial synthesis allowed the
preparation of compositional gradients20-21 and provided a new strategy to load nano-objects into
predefined pores
Unfortunately the LPE process has so far been only demonstrated for a fairly small number of
MOFs141922-23 none of which exhibits the required pore sizes In addition for the fabrication of meta-
crystals the architecture of the network should be sufficiently adjustable to realize pores of different
sizes There should also be a straightforward way to functionalize the framework itself in order to
tailor the interaction of the walls with the targeted guest objects Ideally such a strategy would be
based on an isoreticular series of MOFs in which the pore size can be determined by a choice from a
homologous series of ligands with different lengths1
Here we present a novel isoreticular MOF series with the required flexibility as regards pore-sizes
and functionalization which is well suited for the LPE process This class denoted as SURMOF-2 is
derived from MOF-2 one of the simplest framework architectures24-25 MOF-2 is based on paddle-
wheel (pw) units formed by attaching 4 dicarboxylate groups to Cu++ or Zn++-dimers yielding planar
sheets with 4-fold symmetry (see Fig 1) These planes are held together by strong
carboxylatemetal- bonds26 Conventional solvothermal synthesis yields stacks of pw-planes shifted
relative to each other with a corresponding reduction in symmetry to yield a P2 or C2 symmetry2427-
28
147
The relative shifts between the pw-planes can be avoided when using the recently developed liquid
phase epitaxy (LPE) method22 As demonstrated by the XRD-data shown in Fig 2 for a number of
different dicarboxylic acid ligands the LPE-process yields surface attached metal organic frameworks
(SURMOFs) with P4 symmetry29 except for the 26-NDC ligand which exhibits a P2 symmetry as a
result of the non-linearity of the carboxyl functional groups on the ligand The ligand PPDC
pentaphenyldicarboxylic acid with a length of 25 nm is to our knowledge the longest ligand which
has so far been used successfully for a MOF synthesis303030 A detailed analysis of the diffraction data
allowed us to propose the structures shown in Fig 1 This novel isoreticular series of MOFs hereafter
termed SURMOF-2 also consists of pw-planes which ndash in contrast to MOF-2 ndash are stacked directly
on-top of each other This absence of a lateral shift when stacking the pw-planes yields 1d-pores of
quadratic cross section with a diagonal up to 4 nm (see Fig 2) Importantly for the SURMOF-2 series
interpenetration is absent For many known isoreticular MOF series the formation of larger and
larger pores is limited by this phenomenon if the pores become too large a second or even a third
3d-lattice will be formed inside the first lattice yielding interpenetrated networks which exhibit the
expected larger lattice constant but with rather small pore sizes31-33 the hosting of NPs thus becomes
impossible For SURMOF-2 the particular topology with 1d-pores of quadratic cross section is not
compatible with the presence of a second interwoven network and as a result interpenetration is
suppressed
Although the solubility of some the long dicarboxylic acids ligands used for the SURMOF-2 fabrication
(TPDC QPDC and PPDC see Fig 1) is fairly small we encountered no problems in the LPE process
since already small concentrations of dicarboxylic acids are sufficient for the formation of a single
monolayer on the substrate the elementary step of the LPE synthesis process34-35 Even with the
longest dicarboxylic acid available to us PPDC growth occurred in a straightforward fashion and
optimization of the growth process was not necessary
The P4 structures proposed for the SURMOF-2 series except for the 26-NDC ligand differ markedly
from the bulk MOF-2 structures obtained from conventional solvothermal synthesis2427-28 To
understand this unexpected difference and in particular the absence of relative shifts between the
pw-planes in the proposed SURMOF-2 structure we performed extensive quantum chemical
calculations employing approximate density-functional theory (DFT) in this case London dispersion-
corrected self-consistent charge density-functional based tight-binding (DFTB)36-38 to a periodic
model of MOF-2 and its SURMOF derivatives
Periodic structure DFTB calculations (Fig S1) confirmed that the P2m structure proposed by Yaghi
et al for the bulk MOF-2 material24 is indeed the energetically favorable configuration for MOF-2
while the C2m structure that has also been reported for bulk MOF-227-28 is slightly higher in energy
148
(41 meV per formula unit see Table 1) The optimum interlayer distance between the pw-planes in
the P4 structure is 58 Aring in very good agreement with the experimental value of 56 Aring (as obtained
from the in-plane XRD-data see Fig S2) It is important to note that at that distance the linkers
cannot remain in rectangular position as in MOF-2 but are twisted in order to avoid steric hindrance
and to optimize linker-linker interactions
The P4mmm structure as proposed for SURMOF-2 is less stable by 200 meV per formula unit as
compared to the P2m configuration Although the crystal energy for the P4 stacking is substantially
smaller than for P2 packing simulated annealing (DFTB model using Born-Oppenheimer Molecular
Dynamics) carried out at 300K for 10 ps demonstrated that the P4mmm structure corresponds to a
local minimum This observation is confirmed by the calculation of the stacking energy of MOF-2
where a shifting of the layers is simulated along the P2-P4-C2 path (Figure 3) On the basis of these
calculations we thus propose that SURMOF-2 adopts this metastable P4 structure
In Table 1 the interlayer interaction energies are summarized They range from 590 meV per formula
unit with respect to fully dissociated MOF-2 layers for Cu2(bdc)2 and successively increase for longer
linkers of the same type reaching 145 eV per formula unit for Cu2(ppdc)2 indicating that the linkers
play an important role in the overall stabilization of SURMOF-2 with large pore sizes Even stronger
interlayer interactions are found for different linker topologies (PPDC) A detailed computational
analysis of the role of the individual building blocks of SURMOF-2 for the crystal growth and
stabilization will be published elsewhere
The flat well-defined organic surfaces used as the templating (or nucleating) substrate in the LPE
growth process provide a satisfying explanation for why SURMOF-2 grows with the highly
symmetrical P4 structure instead of adopting the energetically more favorable bulk P2 symmetry3439
The carboxylic acid groups exposed on the surface force the first layer of deposited Cu-dimers into a
coplanar arrangement which is not compatible with the bulk P2 (or C2) structure but rather
nucleates the P4 packing In turn the carboxylate-groups in the first layer of deposited dicarboxylic
acids provide the same constraints for the next layer (see Fig 1) As a result of the layer-by-layer
method employed for further SURMOF-2 growth the same boundary conditions apply for all
subsequent layers The organic surface thus acts as a seed structure to yield the metastable P4
packing not an unusual motif in epitaxial growth40
The calculations allow us to predict that it will be possible to grow SURMOF structures with even
larger linker molecules as used here The stacking will further stabilize the P4 symmetry as the
interaction energy per formula unit is more and more dominated by the linkers (Table 1) At present
149
we cannot predict the maximum possible size of the pores but linker PPDC (see Fig 2) is thus far
unmatched as a component in non-interpenetrated framework structures
Acknowledgement
We thank Ms Huumllsmann Bielefeld University for the preparation of P(EP)2DC Financial support by
Deutsche Forschungsgemeinschaft (DFG) within the Priority Program Metal-Organic Frameworks
(SPP 1362) is gratefully acknowledged
Methods
Computational Details
All structures were created using a preliminary version of our topological framework creator
software which allows the creation of topological network models in terms of secondary building
units and their replacement by individual molecules to create the coordinates of virtually any
framework material The generated starting coordinates including their corresponding lattice
parameters have been hierarchically fully optimized using the Universal Force Field (UFF)41 followed
by the dispersion-corrected self-consistent-charge density-functional based tight-binding (DFTB)
method36-3842-43 that we have recently validated against experimental structures of HKUST-1 MOF-5
MOF-177 DUT-6 and MIL-53(Al) as well as for their performance to compute the adsorption of
water and carbon monoxide37 For all calculations we employed the deMonNano software44444444
We have chosen periodic boundary conditions for all calculations and the repeated slab method has
been employed to compute the properties of the single layers in order to evaluate the stacking
energy A conjugate-gradient scheme was employed for geometry optimization of atomic
coordinates and cell dimensions The atomic force tolerance was set to 3 x 10-4 eVAring
The confined conditions at the SURMOF-substrate interface was modeled by fixing the corresponding
coordinate in the computer simulations All calculated structures have been substantiated by
simulated annealing calculations where a Born-Oppenheimer Molecular Dynamics trajectory at 300K
has been computed for 10 ps without geometry constrains All structures remained in P4 topology
Experimental methods
The SURMOF-2 samples were grown on Au substrates (100-nm Au5-nm Ti deposited on Si wafers)
using a high-throughput approach spray method45 The gold substrates were functionalized by self-
assembled monolayers SAMs of 16-mercaptohexadecanoic acid (MHDA) These substrates were
mounted on a sample holder and subsequently sprayed with 1 mM of Cu2 (CH3COO)4middotH2O ethanol
solution and ethanolic solution of SURMOF-2 organic linkers (01 mM of BDC 26-NDC BPDC and
150
saturated concentration of TPDC QPDC PPDC P(EP)2DC) at room temperature After a given
number of cycles the samples were characterized with X-ray diffraction (XRD)
Figure 1 Schematic representation of the synthesis and formation of the SURMOF-2 analogues
151
Figure 2 Out of plane XRD data of Cu-BDC Cu-26-NDC Cu-BPDC Cu-TPDC Cu-QPDC Cu-P(EP)2DC and Cu-PPDC (upper left) schematic representation (upper right) and proposed structures of SURMOF-2 analogues (lower panel) All the SURMOF-2 are grown on ndashCOOH terminated SAM surface using the LPE method
152
Figure 3 Energy for the relative shift of each layer in two different directions horizontal (P4 to P2) and diagonal (P4 to C2) at a fixed interlayer distance of 56 Aring Structures have been partially optimized and energies are given per formula unit with respect to fully dissociated planes
Supporting information
Synthesis of organic linkers
(1) para-terphenyldicarboxylic acid (TPDC)
To a solution of para-terphenyl (1500 g 651 mmol 1 eq) and oxalyl chloride (3350 mL 3900 mmol
6 eq) in carbon disulfide (30 mL) at 0degC (ice bath) was added aluminium chloride (1475 g 1106
mmol 17 eq) After 1h stirring additional aluminium chloride (0870 g 651 mmol 1 eq) was added
the ice bath was removed and the reaction mixture was stirred at room temperature for 24h The
mixture was poured into crushed ice and stirred until the dark-brown mixture turned to yellow
Carbon disulfide was removed under reduced pressure and the resulting aqueous suspension was
filtered The solid was washed with diluted hydrochloric acid then with diethyl ether and dried
under vacuum overnight with an oil bath at 50degC Pale yellow solid yield 2028 g (98)
(2) para-quaterphenyldicarboxylic acid (QPDC)
153
To a solution of para-quaterphenyl (1000 g 326 mmol 1 eq) and oxalyl chloride (1680 mL 1956
mmol 6 eq) in carbon disulfide (20 mL) at 0degC (ice bath) was added aluminium chloride (0740 g 555
mmol 17 eq) After 1h stirring additional aluminium chloride (0435 g 326 mmol 1 eq) was added
the ice bath was removed and the reaction mixture was stirred at room temperature for 24h The
mixture was poured into crushed ice and stirred until the dark-brown mixture turned to yellow
Carbon disulfide was removed under reduced pressure and the resulting aqueous suspension was
filtered The solid was washed with diluted hydrochloric acid then with diethyl ether and dried
under vacuum overnight with an oil bath at 50degC Yellow solid yield 1186 g (92)
(3) P(EP)2DC
The synthesis of the P(EP)2DC-linker has been described in Ref 46
(4) para-pentaphenly dicarboxylic acid (PPDC)
Scheme 1 Synthesis of para-pentaphenly dicarboxylic acid i) Pd(PPh3)4 Na2CO3 toluenedioxane H2O 85oC ii) MeOH 6M NaOH HCl
para-pentaphenly dicarboxylic acid (H2L) was synthesized via Suzuki coupling of 44rdquo-dibromo-p-
terphenyl and 4-methoxycarbonylphenylboronic acid (Scheme 1) 44rdquo-dibromo-p-terphenyl (776 mg
200 mmol) 4-methoxycarbonylphenylboronic acid (108 g 60 mmol) and Na3CO3 (170 g 160 mmol)
were added to a degassed mixture of toluene14-dioxane (50 mL 221) under Ar [Pd(PPh3)4] (116
mg 010 mmol) was added to the mixture and heated to 85 degC for 24 hours under Ar The reaction
mixture was cooled to room temperature The precipitate was collected by filtration washed with
water methanol and used for next reaction without further purification The final product H4L was
obtained by hydrolysis of the crude product under reflux in methanol (100 mL) overnight with 6M
aqueous NaOH (20 mL) followed by acidification with HCl (conc) After removal of methanol the
final product was collected by filtration as a grey solid (078 g 83 yield) 1H NMR (d6-DMSO
250MHz 298K) δ (ppm) 805 (d J = 75Hz 4H) 789-783 (m 12H) 750 (d J = 75Hz 4H) 13C NMR
cannot be clearly resolved due to poor solubility MALDI-TOF-MS (TCNQ as matrix) mz 47002
cacld 47051 Elemental analysis Calculated C 8169 H 471 Found C 8163 H 479
Br Br MeOOC B
OH
OH
+
COOMe
COOMe
COOH
COOH
i ii
154
Figure S1 MOF-2 in P4 (as reported) P2 and C2 symmetry
155
Figure S2 In-plane XRD data of Cu-BDC Cu-26-NDC Cu-BPDC Cu-TPDC Cu-QPDC and Cu-P(EP)2DC All the
SURMOF-2 are grown on ndashCOOH terminated SAM surface using the LPE method The position of (010) plane
represents the layer distance
Table 1 Stacking energy and geometries of P4 SURMOF-2 derivatives
Symmetry a= c b Stacking Energy
Cu2(bdc)2 C2 1119 50 -076
Cu2(bdc)2 P2 1119 54 -08
Cu2(bdc)2 P4 1119 58 -059
156
Cu2(ndc)2 P2 1335 56 -04
Cu2(bpdc)2 P4 1549 59 -068
Cu2(tpdc)2 P4 1984 59 -091
Cu2(qpdc)2 P4 2424 59 -121
Cu2(P(EP)2DC)2 P4 2512 52 -173
Cu2(ppdc)2 P4 2859 59 -145
Stacking energies (DFTB level in eV) of SURMOFs The energies were calculated within periodic
boundary conditions and are given per formula unit
References
1 Eddaoudi M et al Systematic design of pore size and functionality in isoreticular MOFs and
their application in methane storage Science 295 469-472 (2002)
2 Rosi N L et al Hydrogen storage in microporous metal-organic frameworks Science 300
1127-1129 (2003)
3 Rowsell J L C amp Yaghi O M Metal-organic frameworks a new class of porous materials
Microporous and Mesoporous Materials 73 3-14 (2004)
4 Wang B Cote A P Furukawa H OKeeffe M amp Yaghi O M Colossal cages in zeolitic
imidazolate frameworks as selective carbon dioxide reservoirs Nature 453 207-U206 (2008)
5 Lauren E Kreno et al Metal-Organic Framework Materials as Chemical Sensors Chemical
Reviews 112 1105-1124 (2012)
6 Dragasser A et al Redox mediation enabled by immobilised centres in the pores of a metal-
organic framework grown by liquid phase epitaxy Chemical Communications 48 663-665
(2012)
7 Horcajada P et al Metal-organic frameworks as efficient materials for drug delivery
Angewandte Chemie-International Edition 45 5974-5978 (2006)
8 Horcajada P et al Flexible porous metal-organic frameworks for a controlled drug delivery
Journal of the American Chemical Society 130 6774-6780 (2008)
9 Hurd J A et al Anhydrous proton conduction at 150 degrees C in a crystalline metal-organic
framework Nature Chemistry 1 705-710 (2009)
10 Kreno L E Hupp J T amp Van Duyne R P Metal-Organic Framework Thin Film for Enhanced
Localized Surface Plasmon Resonance Gas Sensing Analytical Chemistry 82 8042-8046
(2010)
11 Lu G et al Fabrication of Metal-Organic Framework-Containing Silica-Colloidal Crystals for
Vapor Sensing Advanced Materials 23 4449-4452 (2011)
157
12 Lu G amp Hupp J T Metal-Organic Frameworks as Sensors A ZIF-8 Based Fabry-Perot Device
as a Selective Sensor for Chemical Vapors and Gases Journal of the American Chemical
Society 132 7832-7833 (2010)
13 Mark D Allendorf Adam Schwartzberg Vitalie Stavila amp Talin A A Roadmap to
Implementing MetalndashOrganic Frameworks in Electronic Devices Challenges and Critical
Directions European Journal of Chemistry (2011)
14 Bo Liu et al Enantiopure Metal-Organic Framework Thin Films Oriented SURMOF Growth
and Enantioselective Adsorption Angewandte Chemie-International Edition 51 817-810
(2012)
15 Sadakiyo M Yamada T amp Kitagawa H Rational Designs for Highly Proton-Conductive
Metal-Organic Frameworks Journal of the American Chemical Society 131 9906-9907 (2009)
16 Ishida T Nagaoka M Akita T amp Haruta M Deposition of Gold Clusters on Porous
Coordination Polymers by Solid Grinding and Their Catalytic Activity in Aerobic Oxidation of
Alcohols Chemistry-a European Journal 14 8456-8460 (2008)
17 Esken D Turner S Lebedev O I Van Tendeloo G amp Fischer R A AuZIFs Stabilization
and Encapsulation of Cavity-Size Matching Gold Clusters inside Functionalized Zeolite
Imidazolate Frameworks ZIFs Chemistry of Materials 22 6393-6401 (2010)
18 Lohe M R et al Heating and separation using nanomagnet-functionalized metal-organic
frameworks Chemical Communications 47 3075-3077 (2011)
19 Shekhah O et al Step-by-step route for the synthesis of metal-organic frameworks Journal
of the American Chemical Society 129 15118-15119 (2007)
20 Furukawa S et al A block PCP crystal anisotropic hybridization of porous coordination
polymers by face-selective epitaxial growth Chemical Communications 5097-5099 (2009)
21 Shekhah O et al MOF-on-MOF heteroepitaxy perfectly oriented [Zn(2)(ndc)(2)(dabco)](n)
grown on [Cu(2)(ndc)(2)(dabco)](n) thin films Dalton Transactions 40 4954-4958 (2011)
22 Shekhah O et al Controlling interpenetration in metal-organic frameworks by liquid-phase
epitaxy Nature Materials 8 481-484 (2009)
23 Zacher D et al Liquid-Phase Epitaxy of Multicomponent Layer-Based Porous Coordination
Polymer Thin Films of [M(L)(P)05] Type Importance of Deposition Sequence on the Oriented
Growth Chemistry-a European Journal 17 1448-1455 (2011)
24 Li H Eddaoudi M Groy T L amp Yaghi O M Establishing microporosity in open metal-
organic frameworks Gas sorption isotherms for Zn(BDC) (BDC = 14-benzenedicarboxylate)
Journal of the American Chemical Society 120 8571-8572 (1998)
25 Mueller U et al Metal-organic frameworks - prospective industrial applications Journal of
Materials Chemistry 16 626-636 (2006)
158
26 Shekhah O Wang H Zacher D Fischer R A amp Woumlll C Growth Mechanism of Metal-
Organic Frameworks Insights into the Nucleation by Employing a Step-by-Step Route
Angewandte Chemie-International Edition 48 5038-5041 (2009)
27 Carson C G et al Synthesis and Structure Characterization of Copper Terephthalate Metal-
Organic Frameworks European Journal of Inorganic Chemistry 2338-2343 (2009)
28 Clausen H F Poulsen R D Bond A D Chevallier M A S amp Iversen B B Solvothermal
synthesis of new metal organic framework structures in the zinc-terephthalic acid-dimethyl
formamide system Journal of Solid State Chemistry 178 3342-3351 (2005)
29 Arslan H K et al Intercalation in Layered Metal-Organic Frameworks Reversible Inclusion of
an Extended pi-System Journal of the American Chemical Society 133 8158-8161 (2011)
30 The MOF with the largest pore size recorded so far MOF-200 (Furukawa H et al Ultrahigh
Porosity in Metal-Organic Frameworks Science 329 424-428 (2010)) used a (trivalent)
444-(benzene-135-triyl-tris(benzene-41-diyl))tribenzoate (BBC) ligand The carboxylic
acid-to carboxylic acid distance is 20 nm compared to 25 nm in case of PPDC The cage size
in MOF-200 amounts to 18 nm by 28 nm clearly smaller than the 1d-channels in the PPDC
SURMOF-2 that are 28 nm by 28 nm
31 Batten S R amp Robson R Interpenetrating nets Ordered periodic entanglement
Angewandte Chemie-International Edition 37 1460-1494 (1998)
32 Snurr R Q Hupp J T amp Nguyen S T Prospects for nanoporous metal-organic materials in
advanced separations processes Aiche Journal 50 1090-1095 (2004)
33 Yaghi O M A tale of two entanglements Nature Materials 6 92-93 (2007)
34 Shekhah O Liu J Fischer R A amp Woumlll C MOF thin films existing and future applications
Chemical Society Reviews 40 1081-1106 (2011)
35 Zacher D Shekhah O Woumlll C amp Fischer R A Thin films of metal-organic frameworks
Chemical Society Reviews 38 1418-1429 (2009)
36 Elstner M et al Self-consistent-charge density-functional tight-binding method for
simulations of complex materials properties Physical Review B 58 7260-7268 (1998)
37 Lukose B et al Structural properties of metal-organic frameworks within the density-
functional based tight-binding method Physica Status Solidi B-Basic Solid State Physics 249
335-342 (2012)
38 Zhechkov L Heine T Patchkovskii S Seifert G amp Duarte H A An efficient a Posteriori
treatment for dispersion interaction in density-functional-based tight binding Journal of
Chemical Theory and Computation 1 841-847 (2005)
159
39 Zacher D Schmid R Woumlll C amp Fischer R A Surface Chemistry of Metal-Organic
Frameworks at the Liquid-Solid Interface Angewandte Chemie-International Edition 50 176-
199 (2011)
40 Prinz G A Stabilization of bcc Co via Epitaxial Growth on GaAs Physical Review Letters 54
1051-1054 (1985)
41 Rappe A K Casewit C J Colwell K S Goddard W A amp Skiff W M UFF a full periodic
table force field for molecular mechanics and molecular dynamics simulations Journal of the
American Chemical Society 114 10024-10035 (1992)
42 Seifert G Porezag D amp Frauenheim T Calculations of molecules clusters and solids with a
simplified LCAO-DFT-LDA scheme International Journal of Quantum Chemistry 58 185-192
(1996)
43 Oliveira A F Seifert G Heine T amp Duarte H A Density-Functional Based Tight-Binding an
Approximate DFT Method Journal of the Brazilian Chemical Society 20 1193-1205 (2009)
44 deMonNano v 2009 (Bremen 2009)
45 Arslan H K et al High-Throughput Fabrication of Uniform and Homogenous MOF Coatings
Adv Funct Mater 21 4228-4231 (2011)
46 Schaate A et al Porous Interpenetrated Zirconium-Organic Frameworks (PIZOFs) A
Chemically Versatile Family of Metal-Organic Frameworks Chemistry-a European Journal 17
9320-9325 (2011)
160
Appendix H
Linker guided metastability in templated Metal-Organic Framework-2 derivatives (SURMOFs-2)
Binit Lukose Gabriel Merino Christof Woumlll and Thomas Heine
Prepared for publication
INTRODUCTION
The molecular assembly of metal-oxides and organic struts can provide a large number of network
topologies for Metal-Organic Frameworks (MOFs)1-3 This is attained by the differences in
connectivity and relative orientation of the assembling units Within each topology replacement of a
building unit by another of same connectivity but different size leads to what is known as isoreticular
alteration of pore size The structure of MOFs in principle can be formed into the requirement of
prominent applications such as gas storage4-8 sensing910 and catalysis11-13 The structural
divergence and the performance can be further increased by functionalizing the organic linkers1415
In MOFs linkers are essential in determining the topology as well as providing porosity A linker
typically contains single or multiple aromatic rings the orientation of which normally undergoes
lowest repulsion from the coordinated metal-oxide clusters providing the most stable symmetry for
the bulk material We encounter for the first time a situation that the orientation of the linker
provides metastability in an isoreticular series of MOFs Templated MOF-2 derivatives (surface-MOF-
2 or simply SURMOFs-2)16 show this extraordinary phenomenon While bulk MOF-2 is determined to
be stable in P217 or C218-20 symmetry we found the existence of SURMOFs-2 in P4 symmetry
161
(Figure 1)16 Our theoretical approach to this problem identifies the role of the linkers in providing
P4 geometry the status of a local energy-minimum
MOF-2 derivatives are two-dimensional lattice structures composed of dimetal-centered four-fold
coordinated paddlewheels and linear organic linkers Solvothermally synthesized archetypal MOF-2
had transition metal (Zn or Cu) centers in the connectors and benzenedicarboxylic acid linkers17 The
derivatives under our investigation contain paddlewheels with Cu dimers and benzenedicarboxylic
acid (BDC) naphthalenedicarboxylic acid (NDC) biphenyldicarboxylic acid (BPDC)
triphenyldicarboxylic acid (TPDC) quaterphenyldicarboxylic acid (QPDC) P(EP)2DC21 and
pentaphenyldicarboxylic acid (PPDC) linkers pertaining to isoreticular expansion of pore size16 The
four-fold connectivity of the Cu dimers together with linearity of the linkers gives rise to layers with
quadratic (square) topology The interlayer separation d is typically much more than that of
graphite or 2D COFS22-24 because all the atoms in one layer do not lie in one plane
In bulk form the nearest layers are shifted to each other either towards one of the four linkers
(P2)171920 or diagonally towards one of the four surrounding pores (C2)18-20 in effort to reduce
the repulsion of alike charges in the adjacent paddlewheels when they are in eclipsed form (P4)
(Figure 1) The metal-dimers often show high reactivity which results in attracting water or
appropriate solvents in their axial positions The stacking along the third axis is typically through
interlayer interactions and through hydrogen bonds established between the solvents or between
the solvents and oxygen atoms in the paddlewheel The lateral shift of layers is inevitable without
additional supports Coordination of pillar molecules such as 14-diazabicyclo[222]octane (dabco) or
bipyridine in the axis of metal dimers between the adjacent layers25 is a practical mean to avoid
layer-offset however with the change of MOF dimensionality
Figure 1 Monolayer of MOF-2 and three kinds of layer packing -P4 P2 and C2- of periodic MOF-2
162
Alternate to solvothermal synthesis a step-by-step route may be enabled for the synthesis of
MOFs26-28 It adopts liquid phase epitaxy (LPE) of MOF reactants on substrate-assisted self-assembled
monolayers This is achieved by alternate immersion of the template in metal and ligand precursors
for several cycles This allows controlled growth of MOFs on surfaces The latest outcomes of this
method are the surface-standing MOF-2 derivatives (SURMOF-2s) This isoreticular SURMOF series
has linkers of different lengths (as given above) The cell dimensions that correspond to the length of
the pore walls range from 1 to 28 nm The diagonal pore-width of one of the MOFs (PPDC) accounts
to 4 nm
After evacuation of solvents (water in the present case) X-ray diffraction patterns were taken in
directions both perpendicular (scans out of plane) and parallel (scans in-plane) to the substrate
surface The presence of only one peak -(010)- in the perpendicular scan implies that the layers
orient perpendicular to the surface and the corresponding angle gives the cell parameter a (or c) In
the latter case the peaks - (100) and (001) ndash are identical and this rules out the possibility of layer-
offset Hence the symmetry of the MOFs was determined to be P4 These peaks give rise to the cell
parameter c which is nothing but the interlayer distance d This value (56 Aring) is relatively small for
P4 geometry to have water molecules in the axial position of paddlewheels Presence of some water
molecules observed using Infrared Reflection Absorption Spectroscopy (IRRAS) might be located near
paddlewheels but exposed to the pore The new observations with the SURMOFs are very intriguing
in the context that P4 symmetry appears to be impossible for solvothermally produced MOF-2
We have primarily reported that the P4 geometry of SURMOF-2 exists as a local energy-minimum16
The verification was made using an approximate method of density functional theory (DFT) which is
London dispersion-corrected self-consistent charge density-functional based tight-binding (DFTB) In
the bulk form absolute energies of P2 and C2 are 210 and 170 meV respectively higher than P4 per
a formula unit which is equivalent to the unit cell For SURMOF the barrier for leaving P4 was nearly
50 meV per formula unit It requires further analysis to unravel the reasons for this unusual
metastability We therefore performed an extensive set of quantum chemical calculations on the
composition of the constituent building units The procedure involves defining SURMOF geometry
and analyzing the translations of individual layers
The major individual contributions to the total energy are the interaction between the paddlewheel
units (favouring P2) London dispersion interaction between tilted linkers (favouring P4) and energy
to tilt the linkers In the iso-reticular series of SURMOF-2 the differences arise only in the
163
contributions from the linkers Hence we performed an extensive study only on the smallest of all
linkers- BDC A scaling might be appropriate for other linkers
RESULTS AND DISCUSSION
In solvothermal synthesis of layered MOFs it is assumed that the nucleation process is associated
with the interaction between two connectors This is rationalized by the fact that two paddlewheels
show the strongest possible noncovalent interaction between the individual MOF building blocks
present in MOF-2 As the orientation of the paddlewheels is restricted by the linkers we studied the
stacking of adjacent layers by determining the attraction of two adjacent interacting paddlewheels
upon their respective offsets Thus we investigated the geometries corresponding to lateral
displacements between the known stable arrangements of the bulk structure ie P4-to-P2 and P4-
to-C2 (Figure 2) In the model calculation the two paddlewheels were shifted from D4h symmetry to
two directions that respectively correspond to P2 and C2 symmetries in the bulk The minima along
the two shifts are referred as min-I and min-II and are having C2h symmetry It is interesting to note
that the interaction is in all cases attractive If only the paddlewheels are studied the D4h
configuration where both axes are concentric can be interpreted as transition state between the
two minima configurations P2 or C2 Comparing the two minima min-I corresponding to MOF-2 in
P2 symmetry indicates the formation of two CuO bonds while for min-II the reactive Cu centers do
not participate in the interlayer bonding
Formation of the diagonally shifted C2 bulk geometry for Zn and Cu based MOF-2 as reported in the
literature18-20 possibly is due to the presence of large solvent molecules such as DMF that
coordinate to the free Cu centers the paddlewheels
Figure 2 Displacements of two paddlewheels from D4h symmetry (corresponding to P4) towards minima that correspond to P2 and C2 bulk symmetries
164
To gain further insight on type of interactions for the three paddlewheel arrangements as found in
the bulk (Figure 3) we performed the topological analysis of the electron density for each
structure2930 An inspection of the paddlewheel shows that the electron density of the Cu-O bond has
a (relatively small) value of 0042 au thus showing only weak character In the forms related to P4
and C2 bulk symmetries all (3-1) critical points between the two paddlewheels show very small
density values (0004 au and less) In the P2 structure it is apparent the formation of a four-
membered Cu-O-Cu-O ring across the paddlewheel Note that the Cu-O interactions inside the
paddlewheel weaken (from 0042 to 0031 au) and two intermolecular Cu-O interactions with a
density value of 0024 au stabilize the P2 orientation These links if formed during nucleation will
be regularly repeated during the MOF-2 formation in solvothermal synthesis and due to its strong
binding of -08 eV they are the main reason for the stabilization of the P2 phase If the paddlewheels
are packed in P4 symmetry there must be additional means of stabilization present and that may
only arise from the linkers Moreover one should note that in order to reach the P4 symmetry a
layer-by-layer growth process is required as otherwise the linkers will readily stack as in the P2 bulk
form
165
Figure 3 Topological analysis of the electron density of fully optimized P4 (top left) C2 (top right) and P4 (bottom) structures Atom (grey) (3-1) (red) and (3+1) (green) critical points along with their charge densities are shown
The optimized geometry of a single-layer MOF-2 is shown in Figure 1 Note that the phenyl rings of
the linkers are oriented perpendicular to the MOF plane Contrary to graphene or 2D COFs the more
complex structure of MOF-2 layers may become subject to change upon the interlayer interactions
This is in particular important for the P4 stacking as reported for SURMOF-2 The interaction energy
of two linkers and two benzene rings as oriented in the monolayer has been computed as function
of the interlayer distance (Figure 4) At the experimental interlayer distance of 56 Aring the linkers are
so close that they repel each other strongly and stacking the monolayer structure at the
experimental interlayer distance would introduce an energy penalty of 08 eV per linker
It would not be exotic if we assume that the anchoring of layers on the substrate plays an important
role in setting d A supporting argument is the fact that all the SURMOFs in the iso-reticular series
have the same d An additional point is that the comparatively wider linkers NDC and LM do not
create any difference in the interlayer distance
166
Figure 4 Energy as a function of interlayer distance [d] in case of two paddlewheels and two benzene molecules The energies are given with respect to the complete dissociation of the building blocks
The best adaptation for the linkers in a closer alignment to avoid the repulsion is to partially rotate
the phenyl rings This reduces the Pauli repulsion and at the same time increases the attractive
London dispersion between the linkers However the rotation is energetically penalized by 06 eV as
accordance with similar calculations found in the literature31 and is with the same order of Zn4O-
tetrahedron clusters of the IRMOFs3233
Figure 5 Energy as a function of rotation of linker between two saturated paddlewheels The dihedral angle O-C1-C2-C3 θ between the linker and the carboxylic part in the paddlewheel Values are with respect to the ground state θ = 0⁰
To model the conditions in SURMOF-2 we need to combine the intermolecular interaction of the
linkers with the barrier associated to the rotation of the linker between two paddlewheel units as
given in Figure 5 We may consider a system of three linkers placed as if they are in three adjacent
layers of bulk P4 SURMOF-2 The linkers are undergoing a rotation along their C2 axis that would be
aligned to the Cu-Cu axes in the paddlewheels (see system-I in Figure 6) The dissociation energy of
167
the system includes four times the repulsion from one adjacent linker If we neglect the interaction
between the end-linkers which is only -35 meV the system represents the situation in bulk P4 MOF-
2 The gain of intermolecular energy due to twisting the stacked linkers is partially compensated by
the energy penalty arising from rotation of the linker between the paddlewheels and the resulting
energy shows a minimum at 22deg (Figure 6)
Figure 6 Model of the linker orientation in SURMOF-2 The stabilization of a linker inbetween two layers is approximated in System-I the arrows indicating the rotation while the energy penalty due to breaking the p conjugation of Figure 5 is given in System-II The resulting energy is the black line showing a minimum at 22deg The energy in System-I is with respect to its dissociation limit
Each linker may undergo rotation either clockwise or anti-clockwise with equal preferences in the
local environment However there may be a global control over the preference of each linker The
most stable structure can be figured out from the total energies of each possible arrangement Since
there are only two choices for each linker it may orient either in same fashion or alternate fashion
along X and Y directions If we expect a regular pattern the total number of possibilities are only
three as shown in Figure 8 where rotation of linkers (violet lines) are marked with lsquo+rsquo signs in one of
its two sides which means that facing (outer)-edges of linkers are tilted towards these sides The
three orderings may be verbalized as follows
(i) projection of the facing edges of oppositely placed linkers are either within the square or outside
(P4nbm) (ii) projection of the facing edge of only one of the oppositely placed linkers is within the
square (P4mmm) (iii) projection of the facing edges of all four linkers are either within the square
or outside (P4nmm)
The adjacent layers follow the same linker-orientations The unit cells of (i) and (iii) are four times
bigger than (ii) The stacking energies calculated at the DFTB level suggest P4nbm as the most stable
168
geometry The energies are respectively -048 -032 and -029 eV per formula unit for P4nbm
P4mmm and P4nmm respectively Within P4nbm symmetry the linker edges undergo lowest
repulsion compared to others It is to be noted that only P4nbm has four-fold rotational symmetry
along Z-axis about the Cu-dimer in any paddlewheel
Figure 7 Three possible arrangements of tilted linkers in the periodic SURMOF-2 Paddlewheels are shown as red diamonds and the linkers as violet lines Facing (outer) edges of the linkers are rotated towards the side with + signs The symmetries and calculated stacking energies (per formula unit) of the arrangements are also given
To quantify the different stacking energies we performed periodic DFT calculations on the structure
of Cu2(bdc)2 MOF-2 As the most stable P4nbm geometry demands high computational resources in
each calculation we used P4mmm geometry which has four times less atoms in unit cell We
explored tight-packing with rotated linkers and loose-packing with un-rotated linkers Two energy-
minima have been found as shown in Figure 8a at 58 and 65 Aring corresponding to rotated and un-
rotated states of linkers respectively The latter is 40 meV more stable than the former which
means that SURMOF-2 exists in a metastable state along the perpendicular axis The relative shift of
adjacent layers in the direction of the lattice vector (or ) represents the transition from P2 to P4
and back to P2 symmetry We shifted the adjacent layers parallel to and calculated the relative
energies Since the shift of adjacent layers in P4mmm geometry is not symmetric for positive and
negative directions of averages of the energies of the shift in both directions are plotted (see
Figure 9b) P4 is a local minimum while P2 is the global minimum and the energy barrier separating
the two minima accounts to 23 meV per formula unit Hence the P4 geometry of SURMOF-2 can be
taken as metastable state of MOF-2
169
Figure 8 a) Energy as function of interlayer separation d and b) Energy as a function of relative shift of adjacent layers for periodic MOF-2 Energies are given in eV per formula unit
The role of linkers in creating a local minimum at P4 is now apparent However when undergoing the
transition from P4 to P2 the relative motions of the horizontal and vertical linkers are different from
each other Hence a qualitative study is essential to accurately determine the role of each building
block for holding SURMOFs-2 in a metastable state We thus resolved the shift of entire adjacent
layers with respect to each other into relative motions of individual building blocks The experimental
interlayer distance (56 Aring) has been fixed and the energy-profiles have been calculated using DFT
The scans include the shift of
i) a paddlewheel over other
ii) a horizontal linker over other
iii) a vertical linker over other
iv) a paddlewheel over a horizontal linker
v) a paddlewheel over a vertical linker
Here vertical and horizontal linkers are the linkers vertical and horizontal to the shift directions
respectively A complete picture of individual shifts of the MOF-2 SBU including their energy-profiles
is given in Figure 9 The shift of the vertical over the horizontal linker was found insignificant and was
omitted A note of warning is that the tilted vertical linker meets different neighborhoods when
shifted to the left and right However an average energy of these two shifts seems sensible because
the nearest vertical linker in the same layer of P4nbm geometry has opposite orientation This
averaging also makes sense in a case that alternate layers undergo shifting to the same direction
leading to a zigzag (or serrated) packing of layers in the bulk As is readily seen in Figure 9 the
formation of the Cu-O-Cu-O rings between the paddlewheels (see Figure 3) is the key reasons for the
layer shift in P2 MOF-2 A support is obtained from the shifting of the paddlewheel over the
170
horizontal linker (Shift IV) A major objection is made by the vertical linkers (Shift III) The total
interaction becomes repulsive when a vertical linker shifts with respect to the other for about 05 Aring
This may alter the tilt of the linker however a minimum is already established The vertical linkers of
a layer in a way are trapped between those of adjacent layers The shift to either side from P4 most
probably decreases the interlayer separation However this demands further rotation of the vertical
linkers in order to reduce their mutual Pauli repulsion Overall the sharp minimum at P4 may be
taken as an inheritance from the lower interlayer distance of P4 geometry fixed by the anchoring on
the substrate
Figure 9 Shift of individual building blocks in the adjacent layers correspond to the situation in bulk The energies of each shifted arrangement and their total energy are given in the graph
The total energy involved in the shifting of two building blocks (one building block over the other) is
equivalent to the energy per one building block when it feels shift from two neighbors Only the
vertical linker is sensitive to the shift-direction of the two neighbors However since averages were
taken as discussed earlier the total energy becomes independent of the direction Besides the
relative motion of the SBU facing each other in P4 symmetry (Shifts I II and III) for strong distortions
we also have to consider the interaction of adjacent linker-connector interactions as represented in
Shifts IV and V The former stabilizes P2 while the latter support P4 Overall the total energy for all
the shifts shows a local minimum at P4 (see figure 9) in agreement with the periodic calculation
shown in Figure 8 Indeed the relative energy between P2 and P4 stackings estimated by the
171
superposition of individual contributions is 43 meV in close agreement to the 40 meV obtained by
the periodic calculations
Our finite-component model successfully provides adequate information on the individual
contributions of building blocks Vertical linkers play crucial role in holding the SURMOF with P4
symmetry which was never achieved with solvothermally synthesized MOF-2 Paddlewheels are
held most responsible for lateral shift The vertical linkers winningly establish a local minimum at P4
for the SURMOF
Recently we have reported SURMOF-2 derivatives with very large pore sizes(Ref) This has been
achieved by increasing the length of the linker units In view of our analysis of the stacking and
stability of MOF-2 we can now safely state that it will be possible to generate SURMOF-2 derivatives
with even larger pores with pore sizes essentially limited by the availability of stiff long organic
linker molecules This is exemplified by a calculation of the stacking energy of a series of phenyl
oligomers that is SURMOF-2 derivatives incorporating chains with 1 to 10 benzene rings in the
linkers The stacking energies per formula unit are -041 -067 -091 -121 -145 -168 -192 -215
-237 and -26 eV respectively Each benzene ring adds more than 02 eV into the stacking energy per
formula unit This energy is due to the London dispersion interaction between the linkers in the
neighboring layers
The drop of SURMOFs into the local minimum perhaps is blended with some parameters related to
synthetic environments This was beyond the scope of this work however we suggest that studies of
the mechanism of substrate-assisted layer-by-layer deposition of metal and ligand precursors may
give some primary insights into it
CONCLUSION
We have analyzed the reason for the different stackings observed for MOF-2 In the traditional
solvothermal synthesis MOF-2 is likely to crystallize in P2 symmetry determined by the strong
interaction between the paddlewheel units The coordination of large solvent molecules to the free
metal centers may lead to MOF-2 in C2 symmetry In contrast the formation of SURMOFs using
Liquid Phase Epitaxy (LPE) allows the formation of high-symmetry P4 MOF-2 This structure requires
a structural modification in terms of the orientation of the linkers with respect to the free monolayer
and the stacking is stabilized by London dispersion interactions between the linkers Increasing the
linker length is a straightforward way for the linear expansion of pore size and according to our
computations the pore size is only limited by the availability of linker molecules showing the desired
length Thus we presented a rare situation in which the linkers guarantee the persistence of a series
of materials in an otherwise unachievable state
172
COMPUTATIONAL DETAILS
The Becke three-parameter hybrid method combined with Lee-Yang-Par correlational functional
(B3LYP)34-37 and augmented with a dispersion term38 as implemented in Turbomole39 has been used
for DFT calculations The copper atoms were described using the basis set associated with the Hay-
Wadt40 relativistic effective core potentials41 which include polarization f functions42 This basis set
was obtained from the EMSL Basis Set Library4344 Carbon oxygen and hydrogen atoms were
described using the Pople basis set 6-311G Geometry optimizations of the MOF fragments were
performed using internal redundant co-ordinates45 A triplet spin state was assigned for each Cu-
paddlewheel46
Periodic structure calculations at the BLYP47 level were performed using the ADF BAND 2012
code4849 Dispersion correction was implemented based on the work by Grimme50 Triple-zeta basis
set was used The crystalline state of MOFs was computationally described using periodic boundary
conditions Bader charge analysis of paddlewheel-pairs was also performed using ADF 2012 code
The analysis was made at the level of B3LYP-D using an all-electron triple-zeta basis set
The non-orthogonal tight-binding approximation to DFT called Density Functional Tight-Binding
(DFTB)51 is computationally feasible for unit cells of several hundreds of atoms Hence this method
was used for extensive calculations on periodic structures This method computes a transferable set
of parameters from DFT calculations of a few molecules per pair of atom types The more accurate
self-consistent charge (SCC) extension of DFTB52-54 has been used for the study of bulk MOFs Validity
of the Slater-Koster parameters of Cu-X (X = Cu C O H) pairs were reported before55 The
computational code deMonNano56 which has dispersion correction implemented57 was used
If not stated otherwise the energies of the periodic structure are given per a formula unit (FU) of the
MOF which includes one paddlewheel and two linkers (one vertical linker and one horizontal linker)
REFERENCES
(1) Eddaoudi M Moler D B Li H L Chen B L Reineke T M OKeeffe M Yaghi O M Accounts of
Chemical Research 2001 34 319
(2) Li H Eddaoudi M OKeeffe M Yaghi O M Nature 1999 402 276
(3) Rowsell J L C Yaghi O M Microporous and Mesoporous Materials 2004 73 3
(4) Eddaoudi M Li H L Yaghi O M Journal of the American Chemical Society 2000 122 1391
(5) Rowsell J L C Yaghi O M Angewandte Chemie-International Edition 2005 44 4670
173
(6) Suh M P Park H J Prasad T K Lim D-W Chemical Reviews 2012 112 782
(7) Murray L J Dinca M Long J R Chemical Society Reviews 2009 38 1294
(8) Rosi N L Eckert J Eddaoudi M Vodak D T Kim J OKeeffe M Yaghi O M Science 2003 300
1127
(9) Kreno L E Leong K Farha O K Allendorf M Van Duyne R P Hupp J T Chemical Reviews 2012
112 1105
(10) Achmann S Hagen G Kita J Malkowsky I M Kiener C Moos R Sensors 2009 9 1574
(11) Lee J Farha O K Roberts J Scheidt K A Nguyen S T Hupp J T Chemical Society Reviews 2009
38 1450
(12) Farrusseng D Aguado S Pinel C Angewandte Chemie-International Edition 2009 48 7502
(13) Corma A Garcia H Llabres i Xamena F X Chemical Reviews 2010 110 4606
(14) Rowsell J L C Millward A R Park K S Yaghi O M Journal of the American Chemical Society 2004
126 5666
(15) Deng H Doonan C J Furukawa H Ferreira R B Towne J Knobler C B Wang B Yaghi O M
Science 2010 327 846
(16) Liu J Lukose B Shekhah O Arslan H K Weidler P Gliemann H Braumlse S Grosjean S Godt A
Feng X Muumlllen K Magdau I-B Heine T Woumlll C submitted to Nature Chemistry 2012
(17) Li H Eddaoudi M Groy T L Yaghi O M Journal of the American Chemical Society 1998 120 8571
(18) Carson C G Hardcastle K Schwartz J Liu X Hoffmann C Gerhardt R A Tannenbaum R
European Journal of Inorganic Chemistry 2009 2338
(19) Clausen H F Poulsen R D Bond A D Chevallier M A S Iversen B B Journal of Solid State
Chemistry 2005 178 3342
(20) Edgar M Mitchell R Slawin A M Z Lightfoot P Wright P A Chemistry-a European Journal 2001
7 5168
(21) Schaate A Roy P Preusse T Lohmeier S J Godt A Behrens P Chemistry-a European Journal
2011 17 9320
(22) Cote A P Benin A I Ockwig N W OKeeffe M Matzger A J Yaghi O M Science 2005 310
1166
(23) Wan S Guo J Kim J Ihee H Jiang D Angewandte Chemie-International Edition 2008 47 8826
174
(24) Lukose B Kuc A Frenzel J Heine T Beilstein Journal of Nanotechnology 2010 1 60
(25) Kitagawa S Kitaura R Noro S Angewandte Chemie-International Edition 2004 43 2334
(26) Shekhah O Wang H Zacher D Fischer R A Woell C Angewandte Chemie-International Edition
2009 48 5038
(27) Shekhah O Wang H Kowarik S Schreiber F Paulus M Tolan M Sternemann C Evers F
Zacher D Fischer R A Woll C Journal of the American Chemical Society 2007 129 15118
(28) Zacher D Schmid R Woell C Fischer R A Angewandte Chemie-International Edition 2011 50 176
(29) Bader R F W Accounts of Chemical Research 1985 18 9
(30) Merino G Vela A Heine T Chemical Reviews 2005 105 3812
(31) Tafipolsky M Schmid R Journal of Chemical Theory and Computation 2009 5 2822
(32) Kuc A Enyashin A Seifert G Journal of Physical Chemistry B 2007 111 8179
(33) Winston E B Lowell P J Vacek J Chocholousova J Michl J Price J C Physical Chemistry
Chemical Physics 2008 10 5188
(34) Becke A D Journal of Chemical Physics 1993 98 5648
(35) Lee C T Yang W T Parr R G Physical Review B 1988 37 785
(36) Vosko S H Wilk L Nusair M Canadian Journal of Physics 1980 58 1200
(37) Stephens P J Devlin F J Chabalowski C F Frisch M J Journal of Physical Chemistry 1994 98
11623
(38) Civalleri B Zicovich-Wilson C M Valenzano L Ugliengo P Crystengcomm 2008 10 405
(39) University of Karlsruhe and Forschungszentrum Karlsruhe gGmbH - TURBOMOLE V63 2011 2007
(40) Wadt W R Hay P J Journal of Chemical Physics 1985 82 284
(41) Roy L E Hay P J Martin R L Journal of Chemical Theory and Computation 2008 4 1029
(42) Ehlers A W Bohme M Dapprich S Gobbi A Hollwarth A Jonas V Kohler K F Stegmann R
Veldkamp A Frenking G Chemical Physics Letters 1993 208 111
(43) Feller D Journal of Computational Chemistry 1996 17 1571
(44) Schuchardt K L Didier B T Elsethagen T Sun L Gurumoorthi V Chase J Li J Windus T L
Journal of Chemical Information and Modeling 2007 47 1045
175
(45) von Arnim M Ahlrichs R Journal of Chemical Physics 1999 111 9183
(46) St Petkov P Vayssilov G N Liu J Shekhah O Wang Y Woell C Heine T Chemphyschem 2012
13 2025
(47) Gill P M W Johnson B G Pople J A Frisch M J Chemical Physics Letters 1992 197 499
(48) SCM Amsterdam Density Functional 2012
(49) Velde G T Bickelhaupt F M Baerends E J Guerra C F Van Gisbergen S J A Snijders J G
Ziegler T Journal of Computational Chemistry 2001 22 931
(50) Grimme S Journal of Computational Chemistry 2006 27 1787
(51) Seifert G Porezag D Frauenheim T International Journal of Quantum Chemistry 1996 58 185
(52) Elstner M Porezag D Jungnickel G Elsner J Haugk M Frauenheim T Suhai S Seifert G
Physical Review B 1998 58 7260
(53) Frauenheim T Seifert G Elstner M Hajnal Z Jungnickel G Porezag D Suhai S Scholz R
Physica Status Solidi B-Basic Research 2000 217 41
(54) Oliveira A F Seifert G Heine T Duarte H A Journal of the Brazilian Chemical Society 2009 20
1193
(55) Lukose B Supronowicz B Petkov P S Frenzel J Kuc A Seifert G Vayssilov G N Heine T
physica status solidi (b) 2011
(56) Heine T Rapacioli M Patchkovskii S Frenzel J Koester A M Calaminici P Escalante S Duarte
H A Flores R Geudtner G Goursot A Reveles J U Vela A Salahub D R deMon-nano edn ed deMon
2009
(57) Zhechkov L Heine T Patchkovskii S Seifert G Duarte H A Journal of Chemical Theory and
Computation 2005 1 841
i
List of Articles
1 Binit Lukose Agnieszka Kuc Johannes Frenzel Thomas Heine On the reticular construction
concept of covalent organic frameworks Beilstein J Nanotechnol 2010 1 60ndash70
DOI103762bjnano18
2 Binit Lukose Agnieszka Kuc Thomas Heine The Structure of Layered Covalent-Organic
Frameworks Chem Eur J 2011 17 2388 ndash 2392 DOI 101002chem201001290
3 Binit Lukose Barbara Supronowicz Petko St Petkov Johannes Frenzel Agnieszka B Kuc
Gotthard Seifert Georgi N Vayssilov and Thomas Heine Structural properties of metal-
organic frameworks within the density-functional based tight-binding method Phys Status
Solidi B 2012 249 335ndash342 DOI 101002pssb201100634
4 Binit Lukose Agnieszka Kuc Thomas Heine Stability and electronic properties of 3D covalent
organic frameworks Prepared for publication
5 Binit Lukose Mohammad Wahiduzzaman Agnieszka Kuc Thomas Heine Structure
electronic structure and hydrogen adsorption capacity of porous aromatic frameworks
Prepared for publication
6 Jinxuan Liu Binit Lukose Osama Shekhah Hasan Kemal Arslan Peter Weidler Hartmut
Gliemann Stefan Braumlse Sylvain Grosjean Adelheid Godt Xinliang Feng Klaus Muumlllen Ioan-
Bogdan Magdau Thomas Heine Christof Woumlll A novel series of isoreticular metal organic
frameworks realizing metastable structures by liquid phase epitaxy Prepared for publication
7 Binit Lukose Gabriel Merino Christof Woumlll Thomas Heine Linker guided metastability in
templated Metal-Organic Framework-2 derivatives (SURMOFs-2) Prepared for publication
8 Binit Lukose Thomas Heine Review Covalently-bound organic frameworks Prepared for
publication
ii
Acknowledgment
Foremost I would like to thank my supervisor Prof Dr Thomas Heine for the wonderful opportunity to join his group as his PhD student I am greatly thankful to him for giving me the topic and sharing with me his expertise and research insight His thoughtful advices have served to give me senses of motion and direction His ambitious approach to science has given me motivation as well as chances and exposures to develop in science His constant attention and guidance have led my scientific outputs to the best levels possible I am also thankful for the financial support and the comfortable stay in his group during my PhD time Additionally he is acknowledged for correcting and reviewing my thesis
Prof Dr Ulrich Kleinekathoumlfer deserves special thanks as my Thesis Committee member I am very glad to have him in the Committee and greatly thankful for reviewing and evaluating the thesis I also thank him and Prof Ulrich Kortz for the evaluation of my PhD proposal I am thankful also for their friendly manners and considerations throughout my PhD time
Prof Dr Christof Woumlll Director of Functional Interfaces Karlsruhe Institute of Technology is greatly acknowledged for being the external Thesis Committee member I am greatly thankful for the evaluation and reviewing of the thesis I am very much moved by his research outcomes and thankful for sharing them with us Our collaborations with his group have particularly enriched my thesis
Prof Dr Petko Petkov is also acknowledged for reviewing my thesis I particlulary thank him for the friendship and discussions thoughout my PhD time
I am indebted to Dr Agnieszka Kuc for introducing me to the topic of nanoporous materials Her experience and expertise have helped me to begin a career in this field I extend my gratitude for sharing with me her scientific skills and correcting our joint-articles
Dr Lyuben Zhechkov and Dr Achim Gelessus have been great in providing computational assistance I have benefitted from their knowledge and sincerity through fast and timely helps
I owe my heartfelt thanks to Dr Lyuben Zhechkov Dr Nina Vankova Dr Augusto Oliveira Dr Andreas Mavrantonakis Dr Stefano Borini and Dr Christian Walther for all discussions suggestions support help and particularly their lectures Dr Lyuben Zhechkov and Dr Nina Vankova are specially mentioned for their long-term attentions and helps Dr Akhilesh Tanwar is acknowledged for his helps in the beginning of my PhD
In my daily work I have been blessed with a friendly and cheerful group of fellow students Barbara Jianping Wahid Nourdine Mahdi Lei Rosalba Ievgenia Wenqing Guilherme Farjana Maicon Aleksandar Ionut Yulia and Gabriel Discussions aside I had great fun times with them Our interactions have also helped me to develop in a personal level I thank them from my full heart although just a few words are not enough I specially thank Barbara Wahid and Ionut for the joint works and publications
Mrs Britta Berninghausen our project assistant deserves special thanks for the friendly assistance on all matters with the university administration
I thank all the members of the group for a lot of good things From the supervisor to the newly joined member everyone has contributed for the general good fun and easiness All those ldquobio-fuelrdquo workshops barbecues parties retreats and gatherings are unforgettable The group also kept good phase with other groups and visitors I thank all the members once again for the good times I would not have been happier anywhere else
iii
I extend my thanks to the research groups that I visited during the PhD time Dr Sourav Pal Director of National Chemical Laboratory Pune and Dr V Subrahmanian Central Leather Research Institute Chennai deserve my gratitude for giving me the opportunity to visit and work with their group members Also I am very thankful to Prof D Sc Georgi N Vayssilov Faculty of Chemistry University of Sofia for the interesting collaboration and visit to his group The financial assistance during each stay is greatly acknowledged I also thank the members of the respective groups namely Dr Petko Petkov and his family who made the visit to Bulgaria very much entertaining
Prof Dr Lars Pettersson of University of Stockholm Dr Tzonka Mineva of CNRS Montpellier and all other members of the HYPOMAP research project are acknowledged for the scientific discussions exposures and promotions
I acknowledge several projects of Prof Dr Thomas Heine for the financial support of my work and travel the funding sources include the European Commission Deutsche Forschungsgemeinschaft (DFG) and the joint Bulgarian-German exchange program (DAAD)
I thank all the co-authors of my publications who have contributed their knowledge ideas and work to accomplish our scientific goals Without their efforts all those works would not have been complete
Members of Research III of SES at Jacobs University namely Robert Carsten Joumlrg Bogdan Meisam Niraj Mahesh Vinu Pinky Patrice Mehdi Sidhant and all professors postdocs and students in Nanofun center are thankfully mentioned here
A lot of my friends in the campus deserve my thanks Mahesh Mahendran Vinu Deepa Srikanth Rajesh Arumugam Prasad Dhananjay Sunil Tripti Raghu Suneetha Rami Susruta Niraj Abhishek Ashok Rakesh Sagar Rohan Naveen Yauhen Yannic Mila and Samira are thanked for the gatherings travels making funs and those cricket and volleyball evenings Some of them are specially thanked for the occasional ldquogahn bayrdquo parties I owe many thanks to Yauhen Srikanth and Prasad for being good flat-mates and having talks on any matters Srikanth and Prasad are thanked again for generously extending their cooking skills to me
I wish to thank everybody with whom I have shared experiences in life I am obliged to my MSc lecturer Dr Rajan K John whose dreams have inspired and driven me to research In particular his accomplishments in the George Sudarshan Center CMS College Kottayam have molded me to take up this career My previous research supervisors Prof S Lakshmibala and Prof V Balakrishnan of IIT Madras and Dr Anita Mehta of SNBNCBS Kolkata are also acknowledged for their important influences in my academic life Additionally all my teachers friends and well-wishers from neighborhood school college GS Center IIT-M and SN Bose center are thanked and acknowledged Members of St Antonyrsquos Parish Olassa are also thanked for the regards and encouragement
Jacobs University Bremen and its people have been amazing in all sorts of things I am glad that I have been a member of the University With my full heart I thank the university authority for all its facilities that were open for me I also thank Dr Svenja Frischholz Mr Peter Tsvetkov and Ms Kaija Gruumlenefeld in the administration departments for the timely helps
Lastly and most importantly I wish to thank my dearest ones for all the sacrifices and love My parents K P Lukose and Molly and my brother Anit deserve to be thanked They have always supported and encouraged me to do my best in all matters of life I also wish to thank my entire extended family for providing me a loving environment
iv
Abstract
Framework materials are extended structures that are built into destined nanoscale architectures
using molecular building units Reticular synthesis methods allow stitching of a large variety of
molecules into predicted networks Porosity is an obvious outcome of the stitching process These
materials are classified and named according to the chemical composition of the building blocks For
instance Metal-Organic Frameworks (MOFs) consists of metal-oxide centers that are linked together
by organic entities The stitching process is straight-forward so that there are already thousands of
them synthesized Controlled growth of MOFs on substrates leads to what is known as surface-MOFs
(SURMOFs) A low-weight metal-free version of MOFs is known as Covalent Organic Frameworks
(COFs) They consist of light elements such as boron oxygen silicon nitrogen carbon and hydrogen
atoms Diamond-like structures with a variety of linear organic linkers between tetrahedral nodes is
called Porous Aromatic Frameworks (PAFs)
The thesis is composed of computational studies of the above mentioned classes of materials The
significance of such studies lies in the insights that it gives about the structure-property relationships
Density Functional Theory (DFT) and its Tight-Binding approximate method (DFTB) were used in
order to perform extensive calculations on finite and periodic structures of several frameworks DFTB
provides an ab-initio base on periodic structure calculations of very large crystals which are typically
studied only using force-field methods The accuracy of this approximate method is validated prior to
reasoning
As the materials are energized from building units and coordination (or binding) stability vs
structure is discussed Energy of formation and mechanical strength are particularly calculated Using
dispersion corrected (DC)-Self Consistent Charge (SCC) DFTB we asserted that 2D COFs should have a
layer arrangement different from experimental suggestions Our arguments supported by simulated
PXRDs were later verified using higher level theories in the literature Another benchmark is giving an
insightful view on the recently reported difference in symmetries of two-dimensional MOFs and
SURMOFs The result of it shows an extra-ordinary role of linkers in SURMOFs providing
metastability
Electronic properties of all the materials and hydrogen adsorption capacities of PAFs are discussed
COFs PAFs and many of the MOFs are semiconductors HOMO-LUMO gaps of molecular units have
crucial influence in the band gaps of the solids Density of states (DOS) of solids corresponds to that
of cluster models Designed PAFs give a large choice of adsorption capacities one of them exceeds
the DOE (US) 2005 target Generally the studies covered under the scheme of the thesis illustrate
the structure stability and properties of framework materials
- Dedicated to my Family and Rajan sir
Table of Contents 1 Outline 1
2 Introduction 2
21 Nanoporous Materials 2
22 Reticular Chemistry 3
23 Metal-Organic Frameworks 5
24 Covalently-bound Organic Frameworks 8
3 Methodology and Validation 10
31 Methods and Codes 10
32 DFTB Validation 11
4 2D Covalent Organic Frameworks 13
41 Stacking 13
42 Concept of Reticular Chemistry 15
5 3D Frameworks 17
51 3D Covalent Organic Frameworks 17
52 Porous Aromatic Frameworks 18
6 New Building Concepts 20
61 Isoreticular Series of SURMOFs 20
62 Metastability of SURMOFs 21
7 Summary 23
71 Validation of Methods 23
72 Weak Interactions in 2D Materials 25
73 Structure-Property Relationships 27
List of Abbreviations 31
List of Figures 32
References 33
Appendix A Review of covalently-bound organic frameworks 37
Appendix B Properties of MOFs within DFTB 81
Appendix C Stacking of 2D COFs 96
Appendix D Reticular concepts applied to 2D COFs 105
Appendix E Properties of 3D COFs 122
Appendix F Properties of PAFs 131
Appendix G Isoreticular SURMOFs of varying pore sizes 145
Appendix H Metastability in 2D SURMOFs 160
1
1 Outline
I prepared this cumulative thesis as it is permitted by Jacobs University Bremen as all work has been
published in international peer-reviewed journals is submitted for publication or in a late
manuscript state in order to be submitted soon The list of articles contains three published papers
three submitted manuscripts and two manuscripts that are to be submitted The articles are given in
Appendices A-H in the order of their discussions Each appendix has one paper and its supporting
information
The chapters from ldquoIntroductionrdquo to ldquoNew Building Conceptsrdquo contain general discussions about the
articles and provide a red thread leading through the articles The discussions mainly circle around
the context and the content of the articles
The chapter ldquoIntroductionrdquo provides a general introduction to the topic and a review of the materials
discussed in this thesis As no comprehensive review on ldquoCovalently-bound Organic Frameworksrdquo is
available in the literature we prepared a manuscript (Appendix A) covering that subject The chapter
ldquoMethodology and Validationrdquo discusses one article on validation of DFTB method in Metal-Organic
Frameworks (Appendix B) Each consecutive chapter discusses two articles The chapter ldquo2D
Covalent Organic Frameworksrdquo covers the studies on layer arrangements (Appendix C) followed by
analysis on how reticular chemistry concepts can be used to design new materials (Appendix D) The
chapter ldquo3D Frameworksrdquo discusses the stability and properties of 3D COFs (Appendix E) and PAFs
(Appendix F) It also discusses hydrogen storage capacities of PAFs The chapter ldquoNew Building
Conceptsrdquo has coverage on 2D MOFs that are built on organic surfaces The first article in this chapter
describes the achievement of our collaborators in synthesizing a series of SURMOFs of varying pore
sizes supported by our calculations indicating their matastability Extensive calculations revealing the
role of linkers in creating the unprecedented difference of SURMOFs in comparison with the bulk
MOFs is described in another article
Details of the articles and references to the appendices are given in the respective places in each
chapter The chapter ldquoSummaryrdquo gives an overview of all the results and discussions It also discusses
some impacts of the publications and concludes the thesis Overall the studies bring into picture
different classes of materials and analyze their structural stabilities and properties
2
2 Introduction
21 Nanoporous Materials
The field of nanomaterials covers materials that have properties stemming from their nanoscale
dimensions (1-100 nm) In contrast to nanomaterials which define size scaling of bulk materials the
major determinant of nanoporous materials is their pores Nanoporous materials are defined as
porous materials with pore diameters less than 100 nm and are classified as micropores of less than
2 nm in diameter mesopores between 2 and 50 nm and macropores of greater than 50 nm They
have perfectly ordered voids to accommodate interact with and discriminate molecules leading to
prominent applications such as gas storage separation and sieving catalysis filtration and
sensoring1-4 They differ from the generally defined nanomaterials in the sense that their properties
are mostly determined by pore specifications rather than by bulk and surface scales Hence the
focus is onto the porous properties of the materials
Utilization of the pores for certain applications relies on certain parameters such as pore size pore
volume internal surface area and wall composition For example physical adsorption of gases is high
in a material with large surface area which implies significantly high storage in comparison to a tank
Porosity can be measured using some inert or simple gas adsorption measurements Distribution of
pore size can be sketched from the adsorptiondesorption isotherm
Nanoporous materials abound in nature Zeolites for instance many of them are available as mineals
have been used in petroleum industry as catalysts for decades The walls of human cells are
nanoporous membranes Other examples are clays aluminosilicate minerals and microporous
charcoals Zeolites are microporous materials from the aluminosilicate family and are often known as
molecular sieves due to their ability to selectively sort molecules primarily based on a size exclusion
principle A material with high carbon content (coal wood coconut shells etc) can be converted to
activated carbon (AC) by controlled thermal decomposition in a furnace The resultant product has
large surface area (~ 500 m2 g-1) Porous glass is a material produced from borosilicate glasses having
pore sizes usually in nm to μm range With increasing environmental concerns worldwide porous
materials have become a suitable choice for separation of polluting gases storage and transport of
energy carriers etc Synthesis of open structures has advanced this area5-7 especially since the
invention of MCM-41 (Mobil Composition of Matter No 41) by Mobil scientists89 Furthermore
there are many templating pathways in making nanoporous materials10-13 Currently it is possible to
engineer the internal geometry at molecular scales
3
For more than a decade chemists are able to synthesize extended structures from well-defined and
rigid molecular building units Such designed and controlled extensions provide porosity which can
be scaled and modified by selecting appropriate building blocks The first realization of this kind was
a class of materials known as Metal-Organic Frameworks (MOFs) in which metal-oxides are stitched
together by organic molecules Synthesis of molecules into predicted frameworks have led to the
emergence of a discipline called reticular chemistry14 The new design-based synthetic approaches
have produced large number of nanoporous materials in comparison to the discovery-based
synthetic chemistry
22 Reticular Chemistry
The discipline of designed (rational) synthesis of materials is known as reticular chemistry Desired
materials can be realized by starting with well-defined and rigid molecular building blocks that will
maintain their structural integrity throughout the construction process The extended structures
adopt high symmetry topologies The synthetic approach follows well-defined conditions which
provide general control over the character of solids In short it is the chemistry of linking molecular
building blocks by strong bonds into predetermined structures
The knowledge about how atoms organize themselves during synthesis is essential for the design
The principal possibilities rely on the starting compounds and the reaction mechanisms Porosity is
almost an inevitable outcome of reticular synthesis Solvents may be used in most cases as space-
filling agents and in cases of more than one possibility as structure-directing agents
Thousands of materials in large varieties have been synthesized using the reticular chemistry
principles Reticular Chemistry Structure Resource (RCSR) is a searchable database for nets a project
initiated by of Omar M Yaghi and Michael OrsquoKeeffe15 They define nets as the collection of vertices
and edges that form an irreducible network in which any two vertices are connected through at least
one continuous path of edges Building units are finite nets as in the case of a polyhedron Periodic
structures have one two or three-periodic nets An example case of 3D diamond net (dia) is shown in
Figure 1 Graph-theoretical aspects together with explicative terminology and definitions can be
found in the literature16-18
Figure 1 a) ldquoAdamantinerdquo unit of diamond structure and b) diamond (dia) net
4
In other words a framework can be deconstructed into one or more fundamental building blocks
each of them assigned by a vertex in the net The vertices are the branching points and edges are
joining them The realization of the net in space by representing the vertices and lattice parameters
by co-ordinates and a matrix respectively is called embedding of the net Underlying topology of an
extended structure is the structure of the net inherited from the crystal structure that is invariant
under different embeddings For example Cu-BTC MOF has paddlewheels and benzene rings as
fundamental blocks The MOF structure can be simplified into its underlying topology as shown in
Figure 2
Figure 2 CU-BTC MOF and the corresponding tbo net
Alternatively the topology of a framework can be defined using the convention of so-called
secondary building units (SBUs) The shapes of SBUs are defined by points of extension of the
fundamental building blocks SBUs are invariant for building units of identical connectivity Based on
the number of connections with the adjacent blocks (4 for paddlewheel and 3 for the benzene) SBUs
of Cu-BTC can be represented as shown in Figure 3a Assembly of SBUs following the network
topology and insertion of the fundamental building blocks give the crystal structure (see Figure 3b for
the extension of SBUs to the topology of Cu-BTC)
In addition to MOFs Metal-Organic Polyhedra (MOP)19 Zeolite Imidazolate Frameworks (ZIFs)20 and
Covalent-Organic Frameworks (COFs)2122 are developing classes of materials within reticular
chemistry the first members of all of them were synthesized by the group of Yaghi MOPs are nano-
sized polyhedra achieved by linking transition metal ions and either nitrogen or carboxylate donor
organic units ZIFs are aluminosilicate zeolites with replacements of tetrahedral Si (Al) and bridging
oxygen by transition metal ion and imidazolate link respectively COFs are extended organic
5
structures constructed solely from light elements (H B C and O) The materials synthesized under
the reticular scheme are largely crystalline
Figure 3 a) Fundamental building blocks and SBUs of Cu-BTC and b) extension of SBUs following
crystal structure
23 Metal-Organic Frameworks
MOFs are porous crystalline materials consisting of metal complexes (connectors) linked together by
rigid organic molecules (linkers) MOFs are analogues to a class of materials called coordination
polymers (CPs) However there are primary differences between them CPs are inorganic or
organometallic polymer structures containing metal ions linked by organic ligands A ligand is an
atom or a group of atoms that donate one or more lone pairs of electrons to the metal cations and
thereby participate in the formation of a coordination complex In MOFs typically metal-oxide
centers are used instead of single metal ions as they provide strong bonds with organic linkers This
provides not only high stability but also high directionality because multiple bonds are involved
6
between metal-centers and organic linkers Predictability lies in the pre-knowledge about the
connector-linker interactions Thus the reticular design of MOFs derives from the precise
coordination geometry between metal centers and the linkers Figure 4a shows a schematic diagram
of MOFs By varying the chemical composition of nodes and linkers thousands of different MOF
structures with a large variety in pore size and structure have been synthesized Figure 4b shows
MOF-5 that consists of zinc-oxide tetrahedrons and benzene linkers
Figure 4 a) Schematic diagram of the single pore of MOF b) An example of a simple cubic MOF in
which zinc-oxide tetrahedron are linked together by benzene linkers Atom colors blue ndash Zn red ndash
O grey ndash C white ndash H
The synthesis of MOFs differs from organic polymerizations in the degree of reversible bond
formation Reversibility allows detachment of incoherently matched monomers followed by their
attachment to form defect-free crystals Assembly of monomers occurs as single step hence
synthetic conditions are of high importance The strength of the metal-organic bond is an obstacle
for reversible bond formation however solvothermal techniques are found out to be a convenient
solution23 Solvothermal synthesis generally allows control over size and shape distribution Using
post-synthetic methods further changes on cavity sizes and chemical affinities can be made
Materials that are stable with open pores after removal of guest molecules are termed as open-
frameworks The stability of guest-free materials is usually examined by Powder X-ray diffraction
(PXRD) analysis Thermogravimetric analysis may be performed to inspect the thermal stability of the
material Elemental analysis can detail the elemental composition of the material Physical
techniques such as Fourier transform infrared (FTIR) spectra and nuclear magnetic resonance (NMR)
may be used to verify the condensation of monomers to the desired topology Porosity can be
evidenced from adsorption isotherms of gases or mercury porosimetry
7
The number of linkers that are allowed to bind to the metal-center and the orientations of the linkers
depend exclusively on the coordination preferences of the metal The organic linkers are typically
ditopic or polytopic They are essential in determining the topology and providing porosity Longer
linkers provide larger pore size A series of compounds with the same underlying topology and
different linker sizes are called iso-reticular series The structure of MOFs in principle can be formed
into the requirement of prominent applications such as gas storage gas separation sensing and
catalysis The structural divergence and performance can be further increased by functionalizing the
organic linkers Hence several attempts are on-going in purpose to come up with the best material
possible in each application
Interpenetration of frameworks is an inevitable outcome of large pore volume Interpenetrated nets
are periodic equivalents of mechanically interlocked molecules rotaxanes and catenanes Depending
on topology they are either maximally separated termed as interpenetration or minimally separated
termed as interweaving (see Figure 5) Interpenetration can be beneficial to some porous structures
protecting from collapse upon removal of solvents
Figure 5 Schematic diagram of interpenetrated and interwoven frameworks
Controlled growth of MOFs on organic surfaces is achieved by R A Fischer and C Woumlll24 The then
named SURMOFs allow to fabricate structures possibly not achievable by bulk synthesis The growth
is achieved through liquid phase epitaxy (LPE) of MOF reactants on the surface (nucleation site) A
step-by-step approach which is alternate immersion of the substrate in metal and ligand precursors
supplies control of the growth mechanism
8
Figure 6 Schematic diagram of SURMOF
24 Covalently-bound Organic Frameworks
As a result of the long-term struggle for complete covalent crystallization in the year 2005 Cocircte et
al21 synthesized porous crystalline extended 2D Covalent-Organic Frameworks (COFs) using
reticular concepts The success was followed by the design and synthesis of 3D COFs in the year
200722 By now there are about 50 COFs reported in the literature COFs are made entirely from
light elements and the building blocks are held together by strong covalent bonds Most of them
were formed by solvothermal condensation of boronic acids with hydroxyphenyl compounds
Tetragonal building blocks were used for the formation of 3D COFs Alternate synthetic methods
were also used for producing COFs COFs are generally studied for gas storage applications However
they have also shown potentialities in photonic and catalytic applications
Alternative synthesis methods paved the way to new covalently bound organic frameworks
Trimerization of polytriazine monomers in ionothermal conditions gives rise to Covalent Triazine
Frameworks (CTFs)25 Similarly coupling reactions of halogenated monomers produced Porous
Aromatic Frameworks (PAFs) Albeit the short-range order one of the PAFs showed very high surface
area (5600 m2 g-1) and gas uptake capacity26
Due to low weight the covalently-bound materials show very high gravimetric capacities
Suggestions such as metal-doping functionalization and geometry modifications can be found in the
literature for the general improvement of the functionalities There are also various studies of their
structure and properties
A review on the synthesis structure and applications of covalently bound organic frameworks has
been prepared for publication
Article 1 Covalently-bound organic frameworks
Binit Lukose Thomas Heine
9
See Appendix A for the article
My contributions include collecting data and preparing a preliminary manuscript
Figure 7 SBUs and topologies of 2D COFs
10
3 Methodology and Validation
31 Methods and Codes
The Density-Functional based Tight Binding (DFTB) method2728 is an approximate Kohn-Sham DFT29-31
scheme It avoids any empirical parameterization by calculating the Hamiltonian and overlap matrix
elements from DFT-derived local orbitals (atomic orbitals) and atomic potentials The Kohn-Sham
orbitals are represented with the linear combination of atomic orbitals (LCAO) scheme The matrix
elements of Hamiltonian are restricted to include only one- and two-center contributions Therefore
they can be calculated and tabulated in advance as functions of the distance between atomic pairs
The remaining contributions to the total energy that is the nucleus-nucleus repulsion and the
electronic double counting terms are grouped in the so-called repulsive potential This two-center
potential is fitted to results of DFT calculations of properly chosen reference systems The accuracy
and transferability can be improved with the self-consistent charge (SCC) extension to DFTB32 This
method is based on the second-order expansion of the Kohn-Sham total energy with respect to
charge density fluctuations which are estimated by Mulliken charge analysis In order to account for
London dispersion empirical dispersion energy can be added to the total energy3334 Various reviews
are available on DFTB notably the recent ones by Oliveira et al35 and by Seifert et al36
DFTB is implemented in a large number of computer codes For this work we employed the codes
deMonNano37 and DFTB+38 as they allow DFTB calculations on periodic and finite structures
Typically dispersion corrected (DC) SCC-DFTB calculations were carried out Periodic boundary
conditions were used to represent the crystalline frameworks and as the unit cells are large the
standard approach used the point approximation Electronic density of states (DOS) have been
calculated using the DFTB+ code using k-point sampling where the k space was determined by
reaching convergence for the total energy according to the scheme proposed by Monkhorst and
Pack39
For DFT calculations of finite and periodic structures ADF40 Turbomole41 and Crystal0942 were used
For studies of finite models of COFs the calculations were performed at PBEDZP level However for
extensive calculations on SURMOF-2 models we used B3LYP-D The copper atoms were described
using the basis set associated with the Hay-Wadt43 relativistic effective core potentials44 which
include polarization f functions45 Carbon oxygen and hydrogen atoms were described using the
Pople basis set 6-311G
Details of the individual calculations are given in the individual articles in the appendix of this thesis
11
32 DFTB Validation
Figure 8 Deconstructed building units their schematic diagrams and final geometries of HKUST-1
(Cu-BTC) MOF-5 MOF-177 MOF-205 and MIL-53
12
In the literature MOFs and COFs are largely studied for applications such as gas storage using
classical force field methods46-48 First principles based studies of several hundreds of atoms are
computationally expensive Hence they are generally limited to cluster models of the periodic
structures Contrarily DFTB paves the way to model periodic structures involving large numbers of
atoms Being an approximate method DFTB holds less accuracy hence comparisons to experimental
data or higher level methods should be performed for validation
As MOFs contain metal centers andor metal oxides as well as organic elements the DFTB
parameters for both heavy and light elements as well as their mixtures are required Thus we have
chosen MOFs such as Cu-BTC (HKUST-1) MOF-5 MOF-177 MOF-205 and MIL-53 as our model
structures which contain Cu Zn and Al atoms apart from C O and H The MOFs comprising three
common connector units copper paddlewheels (HKUST-1) zinc oxide Zn4O tetrahedron (MOF-5
MOF-177 DUT-6 (MOF-205)) and aluminum oxide AlO4(OH)2 octahedron (MIL-53) Figure 8 shows
the schematic diagram of the MOFs
The validation calculations have been published
Article 2 Structural properties of metal-organic frameworks within the density-functional based
tight-binding method
Binit Lukose Barbara Supronowicz Petko St Petkov Johannes Frenzel Agnieszka B Kuc Gotthard
Seifert Georgi N Vayssilov and Thomas Heine Phys Status Solidi B 2012 249 335ndash342 DOI
101002pssb201100634
See Appendix B for the article
In this article DFTB has been validated against full hybrid density-functional calculations for model
clusters against gradient corrected density-functional calculations for supercells and against
experiment Moreover the modular concept of MOF chemistry has been discussed on the basis of
their electronic properties Our results show that DC-SCC-DFTB predicts structural parameters with a
good accuracy (with less than 5 deviation even for adsorbed CO and H2O on HKUST-1) while
adsorption energies differ by 12 kJ mol-1 or less for CO and water compared to DFT benchmark
calculations
My contributions to this article include performing DFTB based calculations on the MOFs (HKUST-1
MOF-5 MOF-177 and MOF-205) that is optimizing the geometries simulating powder X-ray
diffraction patterns and calculating density of states and bulk modulus Additional involvement is in
comparing structural parameters such as bond lengths bond angles dihedral angles and bulk
modulus with experimental data or data derived from DFT calculations and preparing the manuscript
13
4 2D Covalent Organic Frameworks
41 Stacking
Condensation of planar boronic acids and hydroxyphenyl compounds leads to the formation of two-
dimensional covalent organic frameworks (2D COFs) The layers are held together by London
dispersion interactions The first synthesized COFs COF-1 and COF-5 were reported to have AB
(P63mmc staggered as in graphite) and AA (P6mmm eclipsed as in boron nitride) stackings
respectively (see Figure 9)21 The synthesis of several further 2D COFs then followed49-51 all of them
were inspected for only two stacking possibilities ndash P6mmm or P63mmc and it was reported that
they aggregate in P6mmm symmetry As framework materials possess framework charges the
interlayer interactions significantly include Coulomb interactions P6mmm symmetry is the face-to-
face arrangement where the overlap of the stacked structures is maximized (maximization of the
London dispersion energy) however atom types of alike charges are facing each other in the closest
possible way With the interlayer separation similar to that in graphite or boron nitride the Coulomb
repulsion should be high in such arrangements One should notice that in the example case of boron
nitride the facing atom types are different We therefore assumed that a stable stacking should
possess layer-offset
Figure 9 AA and AB layer stacks of hexagonal layers
We considered two symmetric directions for layer shift and studied their total energies (see Figure
10) The stable geometries are found to have a layer-shift of ~14 Aring a value corresponding to the
shift of AA to AB stacking in graphite and compatible with the bond length between two 2nd row
atoms This stability-supported stacking arrangement as revealed from our calculations was
14
supported by good agreement between simulated and experimental PXRD patterns Hence
independent of the elementary building blocks any 2D COF should expose a layer-offset Based on
the direction of the shift we proposed two stacking kinds serrated and inclined (Figure 10) In the
former the layer-offset is back and forth while in the latter the layer-offset followed single direction
As serrated and inclined stackings have no significant change in stacking energy our calculations
cannot predict the long-range stacking in the crystal However this problem is known from other
layered compounds and the ultimate answer is yet to be revealed by analyzing high-quality
crystalline phases at low temperature
Figure 10 Stacking energy Es vs shift of adjacent layers for COF-5 Different stacking possibilities
and their energies are also shown
We published our analysis of the stacking in 2D COFs
Article 3 The Structure of Layered Covalent-Organic Frameworks
Binit Lukose Agnieszka Kuc Thomas Heine Chem Eur J 2011 17 2388 ndash 2392 DOI
101002chem201001290
See Appendix C for the article
15
My contributions to this article include performing the shift calculations simulating XRDs and partly
preparing the manuscript The article has made certain impact in the literature Most of the 2D COFs
synthesized afterwards were inspected for their stacking stability The instability of AA stacking was
also suggested by the studies of elastic properties of COF-1 and COF-5 by Zhou et al52 The shear
modulus shows negative signs for the vertical alignment of COF layers while they are small but
positive for the offset of layers Detailed analysis performed by Dichtel53 et al using DFT was
confirming the instability of AA stacking and suggesting shifts of about 13 to 25 Aring
42 Concept of Reticular Chemistry
Reticular chemistry means that (functional) molecules can be stitched together to form regular
networks The structural integrity of these molecules we also speak of building blocks remains in the
crystal lattices Consequently also the electronic structure and hence the functionality of these
molecules should remain similar
2D COFs are formed by condensation of well-defined planar building blocks With the usage of linear
and triangular building blocks hexagonal networks are expected The properties of each COF may
differ due to its unique constituents However the extent of the relationship of the properties of
building blocks in and outside the framework has not been studied in the literature
Reticular chemistry allows the design of framework materials with pre-knowledge of starting
compounds and reaction mechanisms Reverse of this is finding reactants for a certain topology We
intended to propose some building units suitable to form layered structures (see Figure 11) The
building units obey the regulations of reticular chemistry and offer a variety of structural and
electronic parameters
Our strategic studies on a set of designed COFs have been published
Article 4 On the reticular construction concept of covalent organic frameworks
Binit Lukose Agnieszka Kuc Johannes Frenzel Thomas Heine Beilstein J Nanotechnol 2010 1
60ndash70 DOI103762bjnano18
See Appendix D for the article
16
Figure 11 Schematic diagram of different building units forming 2D COFs
Various hexagonal 2D COFs with different building blocks have been designed and investigated
Stability calculations indicated that all materials have the layer offset as reported in our earlier
work54 (see Appendix C) We show that the concept of reticular chemistry works as the Density-Of-
States (DOS) of the framework materials vary with the the DOS of the molecules involved in the
frameworks However the stacking does have some influence on the band gap
My contributions to this article include performing all the calculations and preparing the manuscript
17
5 3D Frameworks
51 3D Covalent Organic Frameworks
First 3D COFs were reported in 2007 by the group of Yaghi22 Although the number of 3D COFs
synthesized so far has not been crossed half a dozen they are of particular interest for their very low
mass density Similar to 2D COFs condensation of boronic acids and hydroxyphenyl compounds led
to the formation of COF-102 103 105 108 and 202 Usage of tetragonal boronic acids leads to the
formation of 3D networks (Figure 12) All of them possessed ctn topology while COF-108 which has
the same material composition as COF-105 crystallized in bor topology COF-300 which was formed
from tetragonal and linear building units possessed diamond topology and was five-fold
interpenetrated55 COF-108 has the lowest mass density of all materials known today Unlike most of
the MOFs the connectors in COFs are not metal oxide clusters but covalently bound molecular
moieties In the synthesized COFs the centers of the tetragonal blocks were either sp3 carbon or
silicon atoms
Schmid et al56 have analyzed the two different topologies and developed force field parameters for
COFs The mechanical stability of COFs was also reported However no further study that details the
inherent energetic stability and properties of COFs was found in the literature Using DFTB we
performed a collective study of all 3D COFs in their known topologies with C and Si centers
Figure 12 Schematic diagram of tetragonal and triangular building blocks forming lsquoctnrsquo or lsquoborrsquo
topologies
Our studies of3D COFs have been prepared for publication
Article 5 Stability and electronic properties of 3D covalent organic frameworks
Binit Lukose Agnieszka Kuc Thomas Heine
18
See Appendix E for the article
My contributions to this article include performing all the calculations and preparing the manuscript
We discussed the energetic and mechanical stability as well as the electronic properties of COFs in
the article All studied 3D COFs are energetically stable materials with formation energies of 76 ndash
403 kJ mol-1 The bulk modulus ranges from 29 to 206 GPa Electronically all COFs are
semiconductors with band gaps vary with the number of sp2 carbon atoms present in the linkers
similar to 3D MOFs
52 Porous Aromatic Frameworks
Porous Aromatic Frameworks (PAFs) are purely organic structures where linkers are connected by sp3
carbon atoms (see Appendix A) Tetragonal units were coupled together to form diamond-like
networks The synthesis follows typically a Yamamoto-type Ullmann cross-coupling reaction those
reactions are known to be much simpler to be carried out than the condensation reactions necessary
to form COFs However as the coupling reactions are irreversible a lower degree of crystallinity is
achieved and the materials formed were amorphous The first PAF was reported in 2009 and
showed a very high BET surface area of 5600 m2g-1 The geometry of PAF-1 was equal to diamond
with the C-C bonds replaced by biphenyl Subsequent PAFs may be synthesized with longer linkers
between carbon tetragonal nodes Nevertheless synthesis of a PAF with quaterphenyl linker
provided an amorphous material of very low surface area due to the short range order
Synthetic complexities aside the diamond-like PAFs are interesting for their special geometries from
the viewpoint of the theorist It is interesting to see to what extent they follow the properties of
diamond Additionally the large surface area and high polarizability of aromatic rings are auxiliary for
enhanced hydrogen storage Thus we have explored the possibilities of diamond topology by
inserting various organic linkers in place of C-C bonds (Figure 13)
Figure 13 Diamond structure and various organic linkers to build up PAFs
Our studies of PAFs have been prepared for publication
19
Article 6 Structure electronic structure and hydrogen adsorption capacity of porous aromatic
frameworks
Binit Lukose Mohammad Wahiduzzaman Agnieszka Kuc Thomas Heine
See Appendix F for the article
In this article we have discussed the correlations of properties with the structures Exothermic
formation energies of PAFs calculated from the dehydrogenation of the linkers with CH4 indicate the
strong coordination of linkers in the diamond topology The studied PAFs exhibited bulk moduli much
smaller than diamond however in line with related MOFs and COFs The PAFs are semi-conductors
with their band gaps decrease with the increasing number of benzene rings in the linkers
Using Quantized Liquid Density Functional Theory (QLDFT) for hydrogen57 a method to compute
hydrogen adsorption that takes into account inter-particle interactions and quantum effects we
predicted a large choice of hydrogen uptake capacities in PAFs At room temperature and 100 bar
the predicted uptake in PAF-qtph was 732 wt which exceeds the 2015 DOE (US) target We
further discussed the structural impacts on the adsorption capacities
My contributions to this article include designing the materials performing calculations of stability
and electronic properties describing the adsorption capacities and preparing the manuscript
20
6 New Building Concepts
61 Isoreticular Series of SURMOFs
The growth of MOFs on substrates using liquid phase epitaxy (LPE) opened up a controlled way to
construct frameworks This is achieved by alternate immersion of the substrates in metal and ligand
precursors for several cycles Such MOFs named as SURMOFs can avoid interpenetration because
the degeneracy is lifted58 and are suited for conventional applications This is an advantage as
solvothermally synthesized MOFs with larger pore size suffer from interpenetration and the large
pores are hence not accessible for guest species
MOF-259 is a simple 2D architecture based on paddlewheel units formed by attaching four
dicarboxylic groups to Cu2+ or Zn2+ dimers yielding planar sheets with 4-fold symmetry The
arrangement of MOF layers possessed P2 or C2 symmetry Recently the group of C Woumlll at KIT has
synthesized an isoreticular series of SURMOFs derived from MOF-2 which has a homogeneous series
of linkers with different lengths (Figure 14) XRD comparisons suggest that these MOFs possess P4
symmetry The cell dimensions that correspond to the length of the pore walls range from 1 to 28
nm The diagonal pore-width of one of the SURMOFs (PPDC) accounts to 4 nm The symmetry of
SURMOFs as determined to be different from bulk MOF-2 should be understood by means of theory
As collaborators we simulated the structures and inspected each stacking corresponding to the
symmetries in order to understand the difference
Figure 14 The building units and schematic diagram of the formation of iso-reticular SURMOF
series
21
This collaborated work has been submitted for publication
Article 7 A novel series of isoreticular metal organic frameworks realizing metastable structures
by liquid phase epitaxy
Jinxuan Liu Binit Lukose Osama Shekhah Hasan Kemal Arslan Peter Weidler Hartmut Gliemann
Stefan Braumlse Sylvain Grosjean Adelheid Godt Xinliang Feng Klaus Muumlllen Ioan-Bogdan Magdau
Thomas Heine Christof Woumlll
See Appendix G for the article
The main contribution of this article was the experimental proof backed up by our computer
simulations that SURMOFs with exceptionally large pore sizes of more than 4 nm can be prepared in
the laboratory and they offer the possibility to host large materials such as clusters nanoparticles or
small proteins The most important contribution of theory was to show that while MOF-2 in P2
symmetry shows the highest stability the P4 symmetry as obtained using LPE for SURMOF-2
corresponds to a local minimum
My contribution to this article includes performing and analyzing the calculations Our theoretical
study went significantly beyond and will be published as separate article (Appendix H)
62 Metastability of SURMOFs
Following the difference in stability of SURMOFs-2 in comparison with MOF-2 we identified the role
of linkers as a constituent of the framework that provides the metastability to SURMOFs-2 (Figure
15) In MOFs linkers are essential in determining the topology as well as providing porosity Linkers
typically contain single or multiple aromatic rings and are ditopic and polytopic The orientation of
them normally undergoes lowest repulsion from the coordinated metal-oxide clusters and provides
high stability In the case of 2D MOFs non-covalent interlayer interactions lead to the most stable
arrangements Paddlewheels are identified as the reasons for the stability of bulk MOF-2 as they
form metal-oxide bridges across the MOF layers In the case of SURMOFs-2 the linkers are rotated in
a tightly-packed environment of layers and each linker in bulk MOF-2 is rotated in such a way that
any attempt of shifting is obstructed by repulsion between the neighboring linkers The energy
barrier for leaving P4 increases with the length of organic linkers Moreover SURMOF-2 derivatives
with extremely large linkers are energetically stable due to the increased London dispersion
interaction between the layers in formula units Thus we encountered a rare situation in which the
linkers guarantee the persistence of a series of materials in an otherwise unachievable state
22
Figure 15 Energy diagram of the metastable P4 and stable P2 structures
Our results on the linker guided stability of SUMORs-2 have been prepared for publication
Article 8 Linker guided metastability in templated Metal-Organic Framework-2 derivatives
(SURMOFs-2)
Binit Lukose Gabriel Merino Christof Woumlll Thomas Heine
See Appendix H for the article
This article is based solely on my scientific contributions
23
7 Summary
Nanotechnology is the way of ingeniously controlling the building of small and large structures with
intricate properties it is the way of the future a way of precise controlled building with incidentally
environmental benignness built in by design - Roald Hoffmann Chemistry Nobel Laureate 1981
Currently it is possible to design new materials rather than discovering them by serendipity The
design and control of materials at the nanoscale requires precise understanding of the molecular
interactions processes and phenomena In the next level the characteristics and functionalities of
the materials which are inherent to the material composition and structure need to be studied The
understanding of the materials properties may be put into the design of new materials
Computational tools to a large extend provide insights into the structures and properties of the
materials They also help to convert primary insights into new designs and carry out stability analysis
Our theoretical studies of a variety of materials have provided some insights on their underlying
structures and properties The primary differences in the material compositions and skeletons
attributed a certain choice in properties The contents of the articles discussed in the thesis may be
summarized into the following three parts
71 Validation of Methods
Simulations of nanoporous materials typically include electronic structure calculations that describe
and predict properties of the materials Density functional theory (DFT)30 is the standard and widely-
used tool for the investigation of the electronic structure of solids and molecules Even the optical
properties can be studied through the time-dependent generalization of DFT MOFs and COFs have
several hundreds of atoms in their unit cells DFT becomes inappropriate for such large periodic
systems because of its necessity of immense computational time and power Molecular force field
calculations60 on the other hand lack transferable parameterization especially for transition metal
sites and are hence of limited use to cover the large number of materials to be studied Apparently
a non-orthogonal tight-binding approximation to DFT called density functional tight-binding
(DFTB)28323561 has proven to be computationally feasible and sufficiently accurate This method
computes parameters from DFT calculations of a few molecules per pair of atom types The
parameters are generally transferable to other molecules or solids With self-consistent charge (SCC)
extension DFTB has improved accuracy In order to account weak forces the London dispersion
energy can be calculated separately using empirical potentials and added to total energy Successful
realizations of DFTB include the studies of large-scale systems such as biomolecules62
24
supramolecular compounds63 surfaces64 solids65 liquids and alloys66 Being an approximate method
DFTB needs validation Often one compares DFTB results of selected reference systems with those
obtained with DFT
Before electronic structure calculations of framework materials can be carried out it is necessary to
compute the equilibrium configurations of the atoms Geometry optimization (or energy
minimization) employs a mathematical procedure of optimization to move atoms so as to reduce the
net forces on them to negligible values We adopted the conjugate gradient scheme for the
optimizations using DFTB A primary test for the validation of these optimizations is the comparison
of cell parameters bond lengths bond angles and dihedral angles with the corresponding known
numbers We compared the DFTB optimized MOF COF and PAF geometries with the experimentally
determined or DFT optimized geometries and found that the values agree within 6 error
The angles and intensities of X-ray diffraction patterns can produce three-dimensional information of
the density of electrons within a crystal This can provide a complete picture of atomic positions
chemical bonds and disorders if any Hence simulating powder X-ray diffraction (PXRD) patterns of
optimized geometries and comparing them with experimental patterns minimize errors in the crystal
model Our focus on this matter directed us to identify the layer-offsets of 2D COFs for the first time
In the case of 3D COFs excellent correlations were generally observed between experimental and
simulated patterns Slight differences in the intensities of some of them were due to the presence of
solvents in the crystals as they were reported in the experimental articles PAFs as experimentally
being amorphous do not possess XRD comparisons MOFs within DFTB optimization have
undergone some changes especially in the dihedral angles in comparison with experimental
suggestion or DFT optimization This was verified from the differences in the simulated PXRD
patterns Another interesting outcome of XRD comparison was the discovery of P4 symmetry of
templated MOF-2 derivatives (SURMOFs-2) made by Woumlll et al
Bulk moduli which may be expressed as the second derivative of energy with respect to the unit cell
volume can give a sense of mechanical stability Our calculations provide the following bulk moduli
for some materials MOF-5 1534 (1537)67 Cu-BTC 3466 (3517)68 COF-102 206 (170)69 COF-
103 139 (118)69 COF-105 79 (57)69 COF-108 37 (29)69 COF-202 153 (110)69 The data in the
parenthesis give corresponding values found in the literature calculated using force-field methods
The bulk moduli of MOFs are comparable with the results in the literature however COFs show
significant differences Albeit the differences in values each type of calculation shows the trend that
bulk modulus decreases with decreasing mas density increasing pore volume decreasing thickness
of pore walls and increasing distance between connection nodes
25
Formation of framework materials from condensation of reactants may be energetically modeled
COFs are formed from dehydration reactions of boronic acids and hydroxyphenyl compounds The
formation energy calculated from the energies of the products and reactants can indicate energetic
stability Both DFTB for periodic structures and DFT for finite structures predict exothermic formation
of 3D COFs Likewise for 2D COFs such as COF-1 and 5 the formation of monolayers is found to be
endothermic within both the periodic model calculation using DFTB and finite model calculation
using DFT The stacking of layers provides them stability
72 Weak Interactions in 2D Materials
AB stacked natural graphite and ABC stacked rhombohedral graphite are found to be in proportions
of 85 and 15 in nature respectively whereas AA stacking has been observed only in graphite
intercalation compounds On the other hand boron nitride (h-BN) the synthetic product of boric
acid and boron trioxide exists with AA stacking 2D COFs were believed to have high symmetric AA
(eclipsed ndash P6mmm symmetry) stacking as boron nitride (h-BN) An exception was COF-1 which was
considered to have AB (staggered ndash P63mmc) stacking as graphite The AA stacking maximizes the
attractive London dispersion interaction between the layers a dominating term of the stacking
energy At the same time AA stacking always suffers repulsive Coulomb force between the layers
due to the polarized connectors It should be noted that in boron nitride oppositely charged boron
atoms and nitrogen atoms are facing each other The Coulomb interaction has ruled out possible
interlayer eclipse between atoms of alike charges This gave us the notion that 2D COFs cannot
possess layer-eclipsing Based on these facts we have suggested a shift of ~14 Aring between adjacent
layers at which one or two of the edge atoms of the hexagonal rings are situated perpendicular to
the center of adjacent rings In other words some or all of the hexagonal rings in the pore walls
undergo staggering with that of adjacent layers These lattice types were found to be very stable
compared to AA or AB analogues AA stacking was found to have the highest Coulombic repulsion in
each COF Within an error limit the bulk COFs may be either serrated or inclined The interlayer
separations of all the COF stacks are in the range of 30 to 36 Aring and increase in the order AB
serratedinclined AA70 The motivation for proposing inclined stacking for COFs is that the
rhombohedral graphite (ABC stacking) is the inclined version of the layer-offset in natural graphite
(AB stacking) The instability of AA stacking was also suggested by the studies of elastic properties of
COF-1 and COF-5 by Zhou et al52 The shear modulus show negative signs for the vertical alignment of
COF layers while they are small but positive for the offset of layers
The change of stacking should be visible in their PXRD patterns because each space group has a
distinct set of symmetry imposed reflection conditions We simulated the XRD patterns of COFs in
their known and new configurations and on comparison with the experimental spectrum the new as
26
well as AA stacking showed good agreement5470 The slight layer-offsets are only able to make a few
additional peaks in the vicinity of existing peaks and some changes in relative intensities The
relatively broad experimental data covers nearly all the peaks of the simulated spectrum In other
words the broad experimental peaks are explainable with layer-offset
A detailed analysis on the interlayer stacking of HHTP-DPB COF made by Dichtel et al53 is very
complementary53 This work uses molecular mechanics extensively to define potential energy
surfaces of stacking of two layers from eclipsed to staggered covering all possible layer offsets Low
energy region was near to the eclipsed stacking which was then zoomed in to examine using DFT for
higher accuracy The far-eclipsed regions encounter no energy barriers with the near-eclipsed regions
which suggest the unlikeness of forming near-staggered layering Within DFT the eclipsed
geometries are at high energy compared to the slight offsets around AA (~ 13 ndash 25 Aring away) For a
hexagonal COF the energetically preferred offset could be in one of the six hexagonal sectors (or
circle sectors) of central angle 60 The preference of falling into one sector over the others is
arbitrary and may be determined by symmetry set by nature or external parameters Hence the
layering order in bulk could be serrated inclined spiral or purely by random The latter case better
known as turbostratic disorder is then more like a rule for 2D COFs rather than an exception caused
by external forces This also explains why the experimental PXRD patters have relatively broad peaks
The layer-offset not only change the internal pore structure but also affect interlayer exciton and
vertical charge transport in opto-electronic applications
About stacking stability the square COFs are expected not to be different from hexagonal COFs
because the local environment causing the shifts is nearly the same The DFTB based calculations
reported along with the synthesis of electron-transporting 2D-NiPc-BTDA COF supports this notion71
Two shifted configurations ndash at 07 and 14 Aring away from AA ndash appeared to be energetically preferred
over the AA stacking It appears that AA stacking is only possible for boron nitride-like structures
were adjacent layers have atoms with opposite charges in vertical direction
SURMOFs-2 a series of paddlewheel-based 2D MOFs possess a different symmetry than
solvothermally synthesized MOF-2 due to the differences in interlayer interactions Typically the
interaction between the paddlewheels favors P2 or C2 symmetry of bulk MOF-2 rather than the P4
symmetry of SURMOFs-2 Bader charge analysis reveals that electron distribution between the Cu-
paddlewheels is the lowest when they follow P4 symmetry This indicates the smallest possibility of
having a direct Cu-Cu bond between the paddlewheels In monolayer organic linkers undergo no
rotation with respect to metal dimers
27
X-ray diffraction of SURMOFs-2 made by Woumlll et al indicated P4 symmetry and a relatively small
interlayer separation This increases the repulsion between the linkers and enforces them to rotate
The rotations of each linker in a layer are controlled globally The rotated linkers in adjacent layers
increase London dispersion however a paddlewheel-led shift towards any side increases repulsion
thereby locking SURMOF-2 in P4 packing Hence with the special arrangement of rotated linkers the
linker-linker interaction overcomes the paddlewheel-paddlewheel interaction
P4 symmetry opens up the pores more clearly than P2 or C2 Additionally the metal sites that
typically host solvents are inaccessible Furthermore improvements such as functionalizing the linker
may be easily carried out
Based on our calculations on 2D materials we emphasize the role of van der Waals interactions in
determining the layer-to-layer arrangements The promise of reticular chemistry which is the
maintainability of structural integrity of the building blocks in the construction process is partly
broken in the case of SURMOFs-2 This is because due to repulsion the linkers undergo rotation with
respect to the carboxylic parts albeit keeping the topology
73 Structure-Property Relationships
We have studied a variety of materials from the classes of MOFs COFs and PAFs The structural
differences arise from the differences in the constituents andor their arrangements Properties in
general are interlinked with structural specifications Therefore it is beneficial to know the
relationship between the structural parameters and properties
The mass density is an intensive property of a material In the area of nanoporous materials a crystal
with low mass density has advantages in applications involving transport Definitely the mass density
decreases with increasing pore volume Still the number of atoms in the wall and their weights are
important factors The pore size does not relate directly to the atom counts The volume per atom
(inverse of atom density) another intensive property of a material obliquely gives porosity Figure
16 shows the variation of mass density with volume per atom (including the volume of the atom) for
MOFs COFs and PAFs MOFs show higher mass density compared to other materials of identical
atom density implying the presence of heavy elements It is to be noted that Cu-BTC has mass
density 184 times higher than COF-102 The presence of metals in the form of oxide clusters in MOFs
increases the mass density and decreases the volume per atom Note that the low-weighted MOF in
the discussion (MOF-205) is heavier than many of the PAFs and COFs The relatively high mass
density and small volume per atom of COF-300 is due to its 5-fold interpenetration whereas COF-202
has additional tert-butyl groups which do not contribute to the system shape but affect the mass
density and the volume per atom COF-102 and 103 have same topology but different centers (C and
28
Si respectively) The Si centers increases the cell volume and decreases mass density COF-105 (Si
centers) and COF-108 (C centers) additionally differ in their topologies (ctn and bor respectively) It
appears that bor topology expands the material compared to ctn PAFs hold the extreme cases PAF-
phnl has the highest atom and mass densities whereas PAF-qtph has the smallest atom and mass
densities
Figure 16 Mass density as a function of volume per atom for MOFs COFs and PAFs
The bulk modulus indicates the materialsrsquo resistance towards compression which should in principle
decrease with increasing porosity At the same time larger number of atoms making covalent
networks in unit volume should supply larger bulk moduli Thus differences in molecular contents
and architectures give rise to different bulk moduli It is interesting to see how the mechanical
stability of nanoporous materials is related with the atom density We have obtained a correlation
between bulk modulus (B) and volume per atom (VA) in the case of the studied MOFs COFs and PAFs
as follows
29
where k is a constant (see Figure 17) The inverse-volume correlation indicates that the materials
close to the fitting curve have average bond strengths (interaction energy between close atoms)
identical to each other independent of number of bonds bond order and branching Only Cu-BTC
COF-202 and COF-300 show significant differences from the fitted curve Cu-BTC has 17 times larger
bulk modulus compared to COF-102 of similar volume per atom which implies the substantially
higher strength of the bond network resulting from paddlewheel units and tbo topology
Interpenetration decreased the volume per atom however increased bulk modulus through
interlayer interactions in the case of COF-300 The tert-butyl groups in COF-202 do not transmit its
inherent stability to the COF significantly however decreases the volume per atom Comparison
between COF-102 (C centers) and COF-103 (Si centers) discloses that Si centers decrease the
mechanical stability Comparison between COF-105 (ctn) and COF-108 (bor) suggests that ctn
topology possess higher stability This indicates that local angular preferences can amend the
strength of the bulk material
Figure 17 Bulk modulus as a function of volume per atom for MOFs COFs and PAFs
Band gaps of all the studied materials show that they are semiconductors except of Cu-BTC which
has unpaired electrons at the Cu centers Our studies of the density of states (DOS) of bulk MOFs and
the cluster models that have finite numbers of connectors and linkers show that electronic structure
30
stays unchanged when going from cluster to periodic crystal In the case of 2D COFs the DOS of
monolayers and bulk materials were identical Interpenetrated COF-300 showed no difference in the
electronic structure in comparison with the non-interpenetrated structure Based on these results
we may reach into a premature conclusion that electronic structure of a solid is determined by its
constituent bonded network sufficiently large to include all its building units
HOMO-LUMO gap of the building units determine the band gap of a framework material We have
observed that PAFs nearly reproduced the HOMO-LUMO gaps of the linkers 3D COFs that are made
of more than one building unit show that the band gap is slightly smaller than the smallest of the
HOMO-LUMO gaps of the building units For example
TBPM (43 eV) + HHTP (34 eV) COF-105COF-108 (32 eV)
TBPM (43 eV) + TBST (139 eV) COF-202 (42 eV)
TAM (41 eV) + TA (26 eV) COF-300 (23 eV)
The compound names are taken from appendix E Additionally the band gaps decrease with
increasing number of conjugated rings similar to HOMO-LUMO gaps of the linkers
I believe that the studies in the thesis have helped to an extent to understand the structure
stability and properties of different classes of framework materials The benchmark structures we
studied have the essential features of the classes they represent Ab-initio based computational
studies of several periodic structures are exceptional and thus have its place in the literature
31
List of Abbreviations
ADF Amsterdam Density Functional code
BLYP Becke-Lee-Yang-Parr functional
B3LYP Becke 3-parameter Lee Yang and Parr functional
COF Covalent-Organic Framework
CP Coordination Polymer
CTF Covalent-Triazine Framework
DC Dispersion correction
DFT Density Functional Theory
DFTB Density Functional Tight-Binding
DOS Density of States
DOE (US) Department of Energy (United States)
DZP Double-Zeta Polarized basis set
GGA Generalized Gradient Approximation
LCAO Linear Combination of Atomic Orbitals
LPE Liquid Phase Epitaxy
MOF Metal-Organic Framework
PAF Porous Aromatic Framework
PBE Perdew-Burke-Ernzerhof functional
PXRD Powder X-ray Diffraction Pattern
QLDFT Quantized Liquid Density Functional Theory
RCSR Reticular Chemistry Structure Resource
SBU Secondary Building Unit
SCC Self-Consistent Charge
TZP Triple-Zeta Polarized basis set
SURMOF Surface-Metal-Organic Framework
32
List of Figures
Figure 1 ldquoAdamantinerdquo unit of diamond structure and diamond (dia) net 3
Figure 2 CU-BTC MOF and the corresponding tbo net 4
Figure 3 Fundamental building blocks and SBUs of Cu-BTC and extension of SBUs following crystal
structure 5
Figure 4 Schematic diagram of the single pore of MOF and An example of a simple cubic MOF in
which zinc-oxide tetrahedron are linked together by benzene linkers Atom colors blue ndash Zn red ndash O
grey ndash C white ndash H 6
Figure 5 Schematic diagram of interpenetrated and interwoven frameworks 7
Figure 6 Schematic diagram of SURMOF 8
Figure 7 SBUs and topologies of 2D COFs 9
Figure 8 Deconstructed building units their schematic representations and final geometries of
HKUST-1 (Cu-BTC) MOF-5 MOF-177 MOF-205 and MIL-53 11
Figure 9 AA and AB layer stacks of hexagonal layers 13
Figure 10 Stacking energy Es vs shift of adjacent layers for COF-5 Different stacking possibilities and
their energies are also shown 14
Figure 11 Schematic diagram of different building units forming 2D COFs 16
Figure 12 Schematic diagram of tetragonal and triangular building blocks forming lsquoctnrsquo or lsquoborrsquo
topologies 17
Figure 13 Diamond structure and various organic linkers to build up PAFs 18
Figure 14 The building units and schematic diagram of the formation of iso-reticular SURMOF series
20
Figure 15 Energy diagram of the metastable P4 and stable P2 structures 22
Figure 16 Mass density as a function of volume per atom for MOFs COFs and PAFs 28
Figure 17 Bulk modulus as a function of volume per atom for MOFs COFs and PAFs 29
33
References
(1) Morris R E Wheatley P S Angewandte Chemie-International Edition 2008 47 4966
(2) Li J-R Kuppler R J Zhou H-C Chemical Society Reviews 2009 38 1477
(3) Corma A Chemical Reviews 1997 97 2373
(4) Seo J S Whang D Lee H Jun S I Oh J Jeon Y J Kim K Nature 2000 404 982
(5) Lee J Kim J Hyeon T Advanced Materials 2006 18 2073
(6) Stein A Wang Z Fierke M A Advanced Materials 2009 21 265
(7) Velev O D Jede T A Lobo R F Lenhoff A M Nature 1997 389 447
(8) Beck J S Vartuli J C Roth W J Leonowicz M E Kresge C T Schmitt K D Chu C T
W Olson D H Sheppard E W McCullen S B Higgins J B Schlenker J L Journal of the
American Chemical Society 1992 114 10834
(9) Kresge C T Leonowicz M E Roth W J Vartuli J C Beck J S Nature 1992 359 710
(10) Ying J Y Mehnert C P Wong M S Angewandte Chemie-International Edition 1999 38
56
(11) Velev O D Kaler E W Advanced Materials 2000 12 531
(12) Stein A Microporous and Mesoporous Materials 2001 44 227
(13) Tanev P T Pinnavaia T J Science 1996 271 1267
(14) Yaghi O M OKeeffe M Ockwig N W Chae H K Eddaoudi M Kim J Nature 2003
423 705
(15) OKeeffe M Peskov M A Ramsden S J Yaghi O M Accounts of Chemical Research
2008 41 1782
(16) Delgado-Friedrichs O OKeeffe M Journal of Solid State Chemistry 2005 178 2480
(17) Delgado-Friedrichs O Foster M D OKeeffe M Proserpio D M Treacy M M J Yaghi
O M Journal of Solid State Chemistry 2005 178 2533
(18) OKeeffe M Yaghi O M Chemical Reviews 2012 112 675
(19) Tranchemontagne D J L Ni Z OKeeffe M Yaghi O M Angewandte Chemie-
International Edition 2008 47 5136
(20) Hayashi H Cote A P Furukawa H OKeeffe M Yaghi O M Nature Materials 2007 6
501
(21) Cote A P Benin A I Ockwig N W OKeeffe M Matzger A J Yaghi O M Science
2005 310 1166
(22) El-Kaderi H M Hunt J R Mendoza-Cortes J L Cote A P Taylor R E OKeeffe M
Yaghi O M Science 2007 316 268
(23) Rowsell J L C Yaghi O M Microporous and Mesoporous Materials 2004 73 3
(24) Hermes S Zacher D Baunemann A Woell C Fischer R A Chemistry of Materials
2007 19 2168
34
(25) Kuhn P Antonietti M Thomas A Angewandte Chemie-International Edition 2008 47
3450
(26) Ben T Ren H Ma S Cao D Lan J Jing X Wang W Xu J Deng F Simmons J M
Qiu S Zhu G Angewandte Chemie-International Edition 2009 48 9457
(27) Porezag D Frauenheim T Kohler T Seifert G Kaschner R Physical Review B 1995
51 12947
(28) Seifert G Porezag D Frauenheim T International Journal of Quantum Chemistry 1996
58 185
(29) Kohn W Sham L J Physical Review 1965 140 1133
(30) Parr R G Yang W Density-Functional Theory of Atoms and Molecules New York Oxford
University Press 1989
(31) Hohenberg P Kohn W Physical Review B 1964 136 B864
(32) Elstner M Porezag D Jungnickel G Elsner J Haugk M Frauenheim T Suhai S
Seifert G Physical Review B 1998 58 7260
(33) Zhechkov L Heine T Patchkovskii S Seifert G Duarte H A Journal of Chemical
Theory and Computation 2005 1 841
(34) Elstner M Hobza P Frauenheim T Suhai S Kaxiras E Journal of Chemical Physics
2001 114 5149
(35) Oliveira A F Seifert G Heine T Duarte H A Journal of the Brazilian Chemical Society
2009 20 1193
(36) Seifert G Joswig J-O Wiley Interdisciplinary Reviews-Computational Molecular Science
2012 2 456
(37) Heine T Rapacioli M Patchkovskii S Frenzel J Koester A M Calaminici P
Escalante S Duarte H A Flores R Geudtner G Goursot A Reveles J U Vela A Salahub D
R deMon deMon-nano edn deMon-nano 2009
(38) BCCMS B DFTB+ - Density Functional based Tight binding (and more)
(39) Monkhorst H J Pack J D Physical Review B 1976 13 5188
(40) SCM Amsterdam Density Functional 2012
(41) University of Karlsruhe and Forschungszentrum Karlsruhe gGmbH - TURBOMOLE V63
2011 2007
(42) Dovesi R Saunders V R Roetti C Orlando R Zicovich-Wilson C M Pascale F
Civalleri B Doll K Harrison N M Bush I J DrsquoArco P Llunell M CRYSTAL09 Users Manual
University of Torino Torino 2009 2009
(43) Wadt W R Hay P J Journal of Chemical Physics 1985 82 284
(44) Roy L E Hay P J Martin R L Journal of Chemical Theory and Computation 2008 4
1029
35
(45) Ehlers A W Bohme M Dapprich S Gobbi A Hollwarth A Jonas V Kohler K F
Stegmann R Veldkamp A Frenking G Chemical Physics Letters 1993 208 111
(46) Garberoglio G Skoulidas A I Johnson J K Journal of Physical Chemistry B 2005 109
13094
(47) Han S S Mendoza-Cortes J L Goddard W A III Chemical Society Reviews 2009 38
1460
(48) Getman R B Bae Y-S Wilmer C E Snurr R Q Chemical Reviews 2012 112 703
(49) Cote A P El-Kaderi H M Furukawa H Hunt J R Yaghi O M Journal of the American
Chemical Society 2007 129 12914
(50) Wan S Guo J Kim J Ihee H Jiang D Angewandte Chemie-International Edition 2008
47 8826
(51) Wan S Guo J Kim J Ihee H Jiang D Angewandte Chemie-International Edition 2009
48 5439
(52) Zhou W Wu H Yildirim T Chemical Physics Letters 2010 499 103
(53) Spitler E L Koo B T Novotney J L Colson J W Uribe-Romo F J Gutierrez G D
Clancy P Dichtel W R Journal of the American Chemical Society 2011 133 19416
(54) Lukose B Kuc A Heine T Chemistry-a European Journal 2011 17 2388
(55) Uribe-Romo F J Hunt J R Furukawa H Klock C OKeeffe M Yaghi O M Journal of
the American Chemical Society 2009 131 4570
(56) Schmid R Tafipolsky M Journal of the American Chemical Society 2008 130 12600
(57) Patchkovskii S Heine T Physical Review E 2009 80
(58) Shekhah O Wang H Paradinas M Ocal C Schuepbach B Terfort A Zacher D
Fischer R A Woell C Nature Materials 2009 8 481
(59) Li H Eddaoudi M Groy T L Yaghi O M Journal of the American Chemical Society
1998 120 8571
(60) Rappe A K Casewit C J Colwell K S Goddard W A Skiff W M Journal of the
American Chemical Society 1992 114 10024
(61) Frauenheim T Seifert G Elstner M Hajnal Z Jungnickel G Porezag D Suhai S
Scholz R Physica Status Solidi B-Basic Research 2000 217 41
(62) Elstner M Cui Q Munih P Kaxiras E Frauenheim T Karplus M Journal of
Computational Chemistry 2003 24 565
(63) Heine T Dos Santos H F Patchkovskii S Duarte H A Journal of Physical Chemistry A
2007 111 5648
(64) Sternberg M Zapol P Curtiss L A Molecular Physics 2005 103 1017
(65) Zhang C Zhang Z Wang S Li H Dong J Xing N Guo Y Li W Solid State
Communications 2007 142 477
36
(66) Munch W Kreuer K D Silvestri W Maier J Seifert G Solid State Ionics 2001 145
437
(67) Bahr D F Reid J A Mook W M Bauer C A Stumpf R Skulan A J Moody N R
Simmons B A Shindel M M Allendorf M D Physical Review B 2007 76
(68) Amirjalayer S Tafipolsky M Schmid R Journal of Physical Chemistry C 2011 115
15133
(69) Amirjalayer S Snurr R Q Schmid R Journal of Physical Chemistry C 2012 116 4921
(70) Lukose B Kuc A Frenzel J Heine T Beilstein Journal of Nanotechnology 2010 1 60
(71) Ding X Chen L Honsho Y Feng X Saenpawang O Guo J Saeki A Seki S Irle S
Nagase S Parasuk V Jiang D Journal of the American Chemical Society 2011 133 14510
37
Appendix A
Review Covalently-bound organic frameworks
Binit Lukose and Thomas Heine
To be submitted for publication after revision
Contents
1 Introduction
2 Synthetic achievements
21 Covalent Organic Frameoworks (COFs)
22 Covalent-Triazine Frameworks (CTFs)
23 Porous Aromatic Frameworks (PAFs)
24 Schemes for synthesis
25 List of materials
3 Studies of the underlying structure and properties of COFs
4 Applications
41 Gas storage
411 Porosity of COFs
412 Experimental measurements
413 Theoretical preidctions
414 Adsorption sites
415 Hydrogen storage by spillover
42 Diffusion and selectivity
43 Suggestions for improvement
431 Geometry modifications
432 Metal doping
433 Functionalization
5 Conclusions
6 List and pictures of chemical compounds
38
1 Introduction
Nanoporous materials have perfectly ordered voids to accommodate to interact with and to
discriminate molecules leading to prominent applications such as gas storage separation and sieving
catalysis filtration and sensoring1-4 They are defined as porous materials with pore diameters less
than 100 nm and are classified as micropores of less than 2 nm in diameter mesopores between 2
and 50 nm and macropores of greater than 50 nm The internal geometry that determines pore size
and surface area can be precisely engineered at molecular scales Reticular synthetic methods
suggested by Yaghi and co-workers5-7 can accomplish pre-designed frameworks The procedure is to
select rigid molecular building blocks prudently and assemble them into destined networks using
strong bonds
Several types of materials have been synthesized using reticular chemistry concepts One prominent
group which is much appreciated for its credentials is the Metal-Organic Frameworks (MOFs)8-13 in
which metal-oxide connectors and organic linkers are judiciously selected to form a large variety of
frameworks The immense potential of MOFs is facilitated by the fact that all building blocks are
inexpensive chemicals and that the synthesis can be carried out solvothermally The progress in MOF
synthesis has reached the point that some of the MOFs are commercially available Several MOFs of
ultrahigh porosity have also been successfully synthesized notably the non-interpenetrated IRMOF-
74-XI which has a pore size of 98 Aring Scientists are also able to synthesize MOFs from cheap edible
natural products14 Since the discovery of MOFs many other crystalline frameworks have been
synthesized using reticular chemistry such as Metal-Organic Polyhedra (MOP)15-19 Zeolite
Imidazolate Frameworks (ZIFs)20-22 and Covalent Organic Frameworks (COFs)23-29
COFs are composed of covalent building blocks made of boron carbon oxygen and hydrogen in
many cases also including nitrogen or silicon stitched together by organic subunits The atoms are
held together by strong covalent bonds Depending on the selection of building blocks the COFs may
form in two (2D) or three (3D) dimensions Planar building blocks are the constituents of 2D COFs
whereas for the formation of 3D COFs typically tetragonal building blocks are involved High
symmetric covalent linking as it is perceived in reticular chemistry was confirmed for the end
products5
Unlike the case of supramolecular assemblies the absence of noncovalent forces between the
molecular building units endorses exceptional rigidity and stability for COFs They are in general
thermally stable above 4000 C Also in COFs the stitching of molecular building units guarantees an
39
increased order and allows control over porosity and composition Without any metals or other
heavy atoms present they exhibit low mass densities This gives them advantages over MOFs in
various applications for example higher gravimetric capacities for gas storage3031 The lowest
density crystal known until today is COF-10824 In addition COFs exhibit permanent porosity with
specific surface areas that are comparable to MOFs and hence larger than in zeolites and porous
silicates
MOF and COF crystals possess long range order although COFs have been achieved so far only at the
μm scale Reversible solvothermal condensation reactions are credited for the high order of
crystallinity Other new framework materials such as Covalent Triazine Frameworks (CTFs) and
Porous Aromatic Frameworks (PAFs) differed in ways of synthesis The former was prepared by
ionothermal polymerization while the latter was produced by coupling reactions PAFs lack long
range order in the crystals as a result of the irreversible synthesis Nevertheless many of the
materials are promisingly good for applications In this review we intend to discuss the synthetic
achievements of COF CTFs and PAFs and studies on their structure properties and prominent
applications
For easy understanding of the atomic geometry of a material the reader is referred to Table 1 which
gives the names of the framework materials the reactants and the synthesis scheme Figure 5 shows
the reactants and Figure 4 shows the reaction schemes Example reactions are given in Figure 1-3
Abbreviations of each chemical compound are given in Section 6
2 Synthetic achievements
21 Covalent Organic Frameworks (COFs)
In the year 2005 Yaghi et al23 achieved the crystallization of covalent organic molecules in the form
of periodic extended layered frameworks The condensation of discrete molecules of different sizes
enabled the synthesis of micro- and mesoporous crystalline COFs27 The selection of molecules with 2
and 3 connection sites for synthesis led to the formation of hexagonal layered geometries32 Yaghi et
al have synthesized half a dozen three-dimensional crystalline COFs solvothermally using tetragonal
building blocks containing 4 connection sites since 2007242829 Figure 1 shows a few examples of 2D
and 3D COFs formed by the self-condensation of boronic acids such as BDBA TBPM and TBPS and co-
condensation of the same boronic acids with HHTP
40
Figure 1 Solvothermal condensation (dehydration) of monomers to form 2D and 3D COFs Atom colors are boron yellow oxygen red yellow (big) silicon grey carbon
Alternate synthetic procedures were also exploited for production and functionalization of COFs
Lavigne et al33 showed that the synthesis of COFs using polyboronate esters is facile and high-yielded
41
Boronate esters often contain multiple catechol moieties which are prone to oxidation and are
insoluble in organic solvents Dichtel et al3435 presented Lewis acid-catalysis procedures to form
boronate esters of catechols and arylboronic acids without subjecting to oxidation Cooper et al36
successfully utilized microwave heating techniques for rapid production (~200 times faster than
solvothermal methods) of both 2D and 3D COFs The syntheses of phthalocyanine3437 or porphyrin38
based square COFs have been reported in literature The latter was noticed for its time-dependent
crystal growth which also affects its pore parameters
Abel et al39 reported the ability to form COFs on metallic surfaces COF surfaces (SCOFs) have been
formed by molecular dehydration of boroxine and triphenylene on a silver surface Albeit with some
defects the materials showed high temperature stability allowing to proceed with functionalization
Lackinger et al40 realized the role of substrates on the formation of surface COFs from the surface-
generated radicals Halogen substituted polyaromatic monomers can be sublimed onto metal
substrates and ultimately turned into a COF after homolysis and intermolecular colligation
Chemically inert graphite surfaces cannot support the homolysis of covalent carbon-halogen bonds
and thus cannot initiate the subsequent association of radicals COF layers can be formed onto
Cu(111)40 and Au(111)41 surfaces but with lower crystal ordering Beton et al41 deposited the
monomers on annealed Au(111) surfaces which supplied surface mobility for the monomers and
subsequently increased porosity and surface coverage of the covalent networks Unlike the bulk form
at the surface the monomers were self-organized onto polygons with 4 to 8 edges Such a template
was able to organize a sublimed layer of C60 fullerenes the number of C60 molecules adsorbed in a
cavity was correlated to the size of the polygon
In 2009 Sassi et al42 studied the problem of crystal growth of COFs on substrates They calculated
the reaction mechanism of 14-benzenediboronic acid (BDBA) on Ag(111) surface self-condensation
of BDBA molecules in solvents leads to the formation of two-dimensional (the planar) stacked COF-1
For the surface COFs the study using Density Functional Theory reveals that there are neither
preferred adsorption sites for the molecules nor a charge transfer between the molecules and the
surface Hence the electronic structure of the molecules remains unchanged and the role of the
metal surface is limited in providing the planar nature for the polymer The free reaction enthalpy
(Gibbs free energy) difference of the reaction ∆G = ∆H ndash T∆S is approximated using the harmonic
approximation taking into account the geometrical restrictions of the metal surface and the entropic
contributions of the released water molecules As result the formation of SCOF-1 has been found to
be exothermic if more than six monomers are involved Ourdjini et al43 reported the polymerization
of 14-benzenediboronic acid (BDBA) on different substrates (Ag(111) Ag(100) Au(111) and Cu(111))
and at different source and substrate temperatures to follow how molecular flux and adsorption-
42
diffusion environments should be controlled for the formation of polymers with the smallest number
of structural defects High flux is essential to avoid H-bonded supramolecular assemblies of
molecules and the substrate temperature needs to be optimized to allow the best surface diffusion
and longest residential time of the reactants Achieving these two contradictory conditions together
is a limitation for some substrates eg for copper Silver was found to be the best substrate for
producing optimum quality polymers Controlling the growth parameters is important since
annealing cannot cure defects such as formation of pentagons heptagons and other non-hexagonal
shapes which involved strong covalent bonds
Ditchel et al44 successfully grew COF films on monolayer graphene supported by a substrate under
operationally simple solvothermal conditions The films have better crystallinity compared to COF
powders and can facilitate photoelectronic applications straightaway Zamora et al45 have achieved
exfoliation of COF layers by sonication The thin layered nanostructures have been isolated under
ambient conditions on surfaces and free-standing on carbon grids
A noted characteristic of COFs is its organic functionality Judiciously selected organic units ndash pyrene
and triphenylene (TP) ndash for the production of TP COF can not only harvest photons over a wide range
but also facilitate energy migration over the crystalline solid25 Polypyrene (PPy)-COF that consists of
a single aromatic entity ndash pyrene- is fluorescent upon the formation of excimers and hence undergo
exciton migration across the stacked layers26 A nickel(II)-phthalocyanine based layered square COF
that incorporates phenyl linkers and forms cofacially stacked H-aggregates was reported to absorb
photons over the solar spectrum and transport charges across the stacked layers37 COF-366 and
COF-66 showed excellent hole mobility of 81 and 30 cm2 V-1s-1 surpassing that of all polycrystalline
polymers known until today46 A first example of an electron-transporting 2D COF was reported
recently47 This square COF was built from the co-condensation of nickel(II) phthalocyanine and
electron-deficient benzothiadiazole With the nearly vertical arrangement of its layers it exhibits an
electron mobility of 06 cm2 V-1s-1 The enhanced absorbance is over a wide range of wavelengths up
to 1000 nm In addition it exhibits panchromatic photoconductivity and high IR sensitivity
Lavigne et al48 achieved the condensation of alkyl (methyl ethyl and propyl) substituted organic
building units with boronic acids forming functionalized COFs The alkyl groups contribute for higher
molar adsorption of H2 however the increased mass density of the functionalized COFs cause for
decreased gravimetric storage capacities Boronate ester-linked COFs are stable in organic solvents
however undergo rapid hydrolysis when exposed in water49 In particular the instability of COF-1
upon H2O adsorption shall be mentioned here50 Lanni et al claimed that alkylation could bring
hydrolytic stability into COFs49
43
Functionalization and pore surface engineering in 2D COFs can be improved if azide appended
building blocks are stitched together in click reactions with alkynes51 Control over the pore surface
is made possible by the use of both azide appended and bare organic building units the ratios of
which is matching with the final amount of functionalization in the pore walls The click reactions of
azide groups with alkynes within the pores lead to the formation of triazole-linked groups on the
pore surfaces This strategy also gives the relief of not condensing the already functionalized building
units Jiang et al have asserted eclipsed layering for most of the final COFs using powder X-ray
diffraction pattern Exceptions are the hexagonal functionalized COFs with relatively high azide-
content (ge 75) Their weak XRD signals suggest possible layer-offset or de-lamellation Although
functionalization on pore surfaces exhibits the nature of lowering the surface area the possibility to
add reactive groups within the pores can be utilized for gas-sorption selectivity The authors have
claimed a 16 fold selectivity of CO2 over N2 using 100 acetyl-rich COF-5 compared to the pure COF-5
The range of the click reaction approach is so wide that relatively large chromophores can be
accommodated in the pores thereby making COF-5 fluorescent
Functionalization of 3D COFs has just reported in 201252 Dichtel et al used a monomer-truncation
strategy where one of the four dihydroxyborylphenyl moieties of a tetrahedral building unit was
replaced by a dodecyl or allyl functional group Condensation of the truncated and the pure
tetrahedral blocks in a certain ratio created a functionalized 3D COF The degree of functionalization
was in proportion with the ratio The crystallinity was maintained except for high loading (gt 37) of
truncated monomers The pore volume and the surface area were decreased as a function of loading
level
A heterogeneous catalytic activity of COFs is achieved with an imine-linked palladium doped COF by
enabling Suzuki-Miyaura coupling reaction within the pores53 The COF-LZU1 has a layered geometry
that has the distance between nitrogen atoms in the neighboring linkers in a level (~37 Aring) sufficient
to accommodate metal ions Palladium was incorporated by a post-treatment of the COF PdCOF-
LZU1 was applied to catalyze Suzuki-Miyaura coupling reaction This coupling reaction is generally
used for the facile formation of C-C bonds54 The increased structural regularity and high insolubility
in water and organic solvents are adequate qualities of imine-linked COFs to perform as catalysts
Experiments with the above COF show a broad scope of the reactants excellent yields of the
products and easy recyclability of the catalyst
The comparatively higher thermal stability of COFs is often noted and is explainable with their strong
covalent bonds The reversible dehydrations for the formation of most of the COFs point to their
instability in the presence of water molecules This has been tested and verified for some layered
COFs with boronate esters49 Such COFs are stable as hosts for organometallic molecules55 COF-102
44
framework was found to be stable and robust even in the presence of highly reactive cobaltocenes
The highly stable ferrocenes appear to have an arrangement within the framework led by π-π
interactions
22 Covalent Triazine Frameworks (CTFs)
In the year 2008 Thomas et al56 synthesized a crystalline extended porous nanostructure by
trimerization of polytriazine monomers in ionothermal conditions With the catalytic support of ZnCl2
three aromatic nitrile groups can be trimerized to form a hexagonal C3N3 ring The resulting structure
known as Covalent Triazine-based Frameworks (CTFs) is similar to 2D COFs in geometry and atomic
composition (see Figure 2) Usage of larger non-planar monomers at higher amounts of ionic salts
however led to the formation of contorted structures Interestingly those structures showed
enhanced surface area and pore volume The trimerization of monomers that lack a linear
arrangement of nitrile groups ended up as organic polymer networks Later the same group
introduced mesoporous channels onto a CTF (CTF-2)57 by increasing temperature and ion content
The resulting structure however was amorphous It is concluded that the reaction parameters and
the amount of salt play a crucial role for tuning the porosity and controlling the order of the material
respectively58
Figure 2 Trimerization of nitrile monomers to form CTF-1 Atom colors blue N grey C white H
Ben et al59 followed ionothermal conditions for the targeted synthesis of a 3D framework using
tetraphenylmethane block and a triangular triazine ring The end product was amorphous thermally
stable up to 400 0C and could be applied for the selective sorption of benzene over cyclohexane On a
later synthetic study Zhang et al60 reported that this polymerization could be accomplished in short
45
reaction times under microwave enhanced conditions The template-free high temperature dynamic
polymerization reactions constitute irreversible carbonization reactions coupled with reversible
trimerization of nitriles The irreversibility encounter in these condensation reactions is responsible
for the production of frameworks as amorphous solids6162
An amorphous yet microporous CTF was found to be able to catalyze the oxidation of glycerol with
Pd nanoparticles that are dispersed in the framework63 This solid catalyst appears to be strong
against deactivation and selective toward glycerate compared to Pd supported activated carbon This
is attributed to the relatively high stability of Pd nanoparticles that benefit from the presence of
nitrogen functionalities in CTF An advancement on selective oxidation of methane to methanol at
low temperature is accomplished by using an amorphous bipyridyl-rich platinum-coordinated CTF as
catalyst64
23 Porous Aromatic Frameworks (PAFs)
a notably high surface area (BET surface area of 5600 m2g-1 Langmuir surface area of 7100 m2g-1)65
PAF-1 has been synthesized in a rather simple way using a nickel(0)-catalyzed Yamamoto-type66
Ullmann cross-coupling reaction67(see Figure 3) The material is thermally (up to 520 0C in air) and
hydrothermally (in boiled water for a minimum of seven days) stable The framework exposes all
faces and edges of the phenyl rings to the pores and hence the high surface area can be achieved
while its high stability is inherited from the parent diamond structure The synthesized material was
verified as disordered and amorphous yet with uniform pore diameters The excess H2 uptake
capacity of PAF-1 was measured as 70 wt at 48 bar and 77 K It can adsorb 1300 mg g-1 CO2 at 40
bar and room temperature PAF-1 was also tested for benzene and toluene adsorption
Figure 3 Coupling reaction using halogenated tetragonal blocks for the formation of diamond-like PAF-1 Atom colors red Br grey C white H
46
An attempt to synthesize a diamondoid with quaterphenyl as linker is described recently68 A
tetrahedral unit and a linear linker each of them has two phenyl rings were polymerized using the
Suzuki-Miyaura coupling reaction54 The product (PAF-11) however stayed behind theoretical
predictions and performed poorly pointing to its shortcomings such as short-range order distortion
and interpenetration of phenyl rings Nevertheless the amorphous solid displayed high thermal and
chemical stabilities proneness for adsorbing methanol over water and usability for eliminating
harmful aromatic molecules
PAF-569 is a two-dimensional hexagonal organic framework synthesized using the Yamamoto-type
Ullmann reaction This material is composed only of phenyl rings however lack long range order
because of the distortion of the phenyl rings and kinetics controlled irreversible coupling reaction It
retains a uniform pore diameter and provides high thermal and chemical stability despite its
amorphous nature PAF-5 was suggested for adsorbing organic pollutants at saturated vapour
pressure and room temperature
Co-condensation reactions with piperazine enabled nucleophilic substitution of cyanuric chloride to
form a covalently linked porous organic framework PAF-670 The synthesis in mild conditions yields a
product with uniform morphology and a certain degree of structural regularity Being nontoxic this
material was tested for drug delivery thereby stepping into biomedical applications of covalently
linked organic frameworks Primary tests suggest that PAF-6 could be promoted for drug release for
its permanent porosity and facile functionalization In comparison with MCM-4171 a widely tested
inorganic framework PAF-6 performed equally or even superiorly
24 Schemes for synthesis
The majority of the COFs were synthesized using solvothermal step-by-step condensation
(dehydration) reactions The method incorporates reversibility and is applicable for supplying long
range order in COF materials The reactants generally consist of boronic acids and hydroxy-
polyphenyl (catechol) compounds Extended porous networks are obtained when O-B or O-C bonds
are formed into 5 or 6-membered rings after dehydration (scheme 1) Dehydration may also be
carried out using microwave synthesis Alternately boronic acid can react with catechol acetonide in
presence of a Lewis acid catalyst The molecules link to each other after the liberation of acetone and
water (scheme 2) The utilization of phenyl-attached formyl and amino groups as building units
results in the formation of imines after dehydration (scheme 3) A desired arrangement of molecular
arrays on surfaces can be fulfilled by depositing planar molecules on metallic surfaces followed by
covalent linking using annealing
47
Isocyanate groups follow trimerization to form poly-isocyanurate rings (scheme 4) Cyclotrimerization
of monomers containing nitrile groups or cyanate groups is prone to generate C3N3 rings (scheme 5)
However the ionothermal synthesis of them resulted with amorphous materials Unique bond
formation is often not achieved throughout the material and thus the crystal lacks long-range order
Analogously coupling reactions such as Ullmann and Suzuki-Miyaura give rise to amorphous
products However they are adequate in producing C-C bonds when halogen-substituted compounds
are reacted (scheme 6) Nucleophilic substitution may also be used for framing networks in cases
like reactants are formed into nucleophiles and ions after release of their end-groups (scheme 7)
48
Figure 4 Different schemes reported for the synthesis of covalently-bound organic frameworks
49
25 List of synthesized materials
Table 1 List of synthesized materials in the literature their building blocks synthesis methods measured surface area [m
2 g
-1] pore volume [cm
3 g
-1] and pore size [Aring]
COF Names Reactants Synthesis Surface
Area
Pore
volume
Pore
size
COF-1 BDBA Solvothermal condensation235072
scheme 1
711 62850 032
03650
9
COF-5 BDBA HHTP Solvothermal condensation23
scheme 1
1590 0998 27
Microwave3673 scheme 1 2019
BDBA TPTA Lewis acid catalysis35 TPTA
COF-6 BTBA HHTP Solvothermal condensation27
scheme 1
980 (L) 032 64
COF-8 BTPA HHTP Solvothermal condensation27
scheme 1
1400 (L) 069 187
COF-10 BPDA HHTP Solvothermal condensation27
scheme 1
2080 (L) 144 341
BPDA TPTA Lewis acid catalysis35 scheme 2
COF-18Aring BTBA THB Facile dehydration3348 scheme 1 1263 069 18
COF-16Aring BTBA alkyl-THB
(alkyl = CH3)
Facile dehydration48 scheme 1 753 039 16
COF-14Aring BTBA alkyl-THB
(alkyl = C2H5)
Facile dehydration48 scheme 1 805 041 14
COF-11Aring BTBA alkyl-THB
(alkyl = C3H7)
Facile dehydration48 scheme 1 105 0052 11
50
SCOF-1 BDBA Substrate-assisted synthesis39
scheme 1
SCOF-2 BDBA HHTP Substrate-assisted synthesis39
scheme 1
TP COF PDBA HHTP Solvothermal condensation25
scheme 1
868 079 314
PPy-COF PDBA Solvothermal condensation26
scheme 1
923 188
TBB COF TBB (on Cu(111) and
Ag(110) surfaces)
Surface polymerisation40 scheme
6
TBPB COF TBB (on Au(111)
surface)
Surface polymerisation41 scheme
6
BTP COF BTPA THDMA Solvothermal condensation72
scheme 1
2000 163 40
HHTP-DPB COF DPB HHTP Solvothermal condensation73
scheme 1
930 47
PICU-A DMBPDC Cyclotrimerization74 scheme 4
PICU-B DCF Cyclotrimerization74 scheme 4
COF-LZU1 DAB TFB Solvothermal condensation53
scheme 3
410 054 12
PdCOF-LZU1 COF-LZU1 PdAc Facile post-treatment53 146 019 12
XN3-COF-5 X N3-BDBA (100-X)
BDBA HHTP
Solvothermal condensation51
scheme 1
2160
(X=5)
1865 (25)
1722 (50)
1641 (75)
1421
(100)
1184
1071
1016
0946
0835
295
276
259
258
227
51
XAcTrz-COF-5 XN3-COF-5 PAc Click reaction51 2000
(X=5)
1561 (25)
914 (50)
142 (75)
36 (100)
1481
0946
0638
0152
003
298
243
156
153
125
XBuTrz-COF-5 XN3-COF-5 HP Click reaction51
XPhTrz-COF-5 XN3-COF-5 PPP Click reaction51
XEsTrz-COF-5 XN3-COF-5 MP Click reaction51
XPyTrz-COF-5 XN3-COF-5 PyMP Click reaction51
COF-42 DETH TFB Solvothermal condensation75
scheme 3
710 031 23
COF-43 DETH TFPB Solvothermal condensation75
scheme 3
620 036 38
CTF-1 DCB Ionothermal trimerization56
scheme 5
791 040 12
CTF-2 DCN Ionothermal trimerization57
scheme 5
90 8
mp-CTF-2 2255 151 8
CTF (DCP) DCP Ionothermal trimerization64
scheme 5
1061 0934 14
K2[PtCl4]-CTF DCP K2[PtCl4] Trimerization (scheme 5) +
coordination64
Pt-CTF DCP Pt Trimerization (scheme 5) +
coordination64
PAF-5 TBB Yamamoto-type Ullmann cross-
coupling reaction69 scheme 6
1503 157 166
52
PAF-6 PA CA Nucleophilic substitution70
scheme 7
1827 118
Pc-PBBA COF PcTA BDBA Lewis acid catalysis34 scheme 2 506 (L) 0258 217
NiPc-COF OH-Pc-Ni BDBA Solvothermal condensation37
scheme 1
624 0485 190
XN3-NiPc-COF OH-Pc-Ni X N3-BDBA
(100-X) BDBA
Solvothermal condensation51
scheme 1
XEsTrz-NiPc-
COF
XN3-NiPc-COF MP Click reaction51
ZnP COF TDHB-ZnP THB Solvothermal condensation38
scheme 1
1742 1115 25
NiPc-PBBA COF NiPcTA BDBA Lewis acid catalysis35 scheme 2 776
2D-NiPc-BTDA
COF
OHPcNi BTDADA Solvothermal condensation47
scheme 1
877 22
ZnPc-Py COF OH-Pc-Zn PDBA Solvothermal condensation
scheme 1
420 31
ZnPc-DPB COF OH-Pc-Zn DPB Solvothermal condensation
scheme 1
485 31
ZnPc-NDI COF OH-Pc-Zn NDI Solvothermal condensation
scheme 1
490 31
ZnPc-PPE COF OH-Pc-Zn PPE Solvothermal condensation
scheme 1
440 34
COF-366 TAPP TA Solvothermal condensation46
scheme 3
735 032 12
COF-66 TBPP THAn Solvothermal condensation46
scheme 1
360 020 249
53
COF-102 TBPM Solvothermal condensation24
scheme 1
3472 135 115
Microwave36
scheme 1
2926
COF-102-C12 TBPM trunk-TBPM-R
(R=dodecyl)
Solvothermal condensation52
scheme 1
12
COF-102-allyl TBPM trunk-TBPM-R
(R=allyl)
Solvothermal condensation52
scheme 1
COF-103 TBPS Solvothermal condensation24
scheme 1
4210 166 125
COF-105 TBPM HHTP Solvothermal condensation24
scheme 1
COF-108 TBPM HHTP Solvothermal condensation24
scheme 1
COF-202 TBPM TBST Solvothermal condensation28
scheme 1
2690 109 11
COF-300 TAM TA Solvothermal condensaion29
scheme 3
1360 072 72
PAF-1 TkBPM-X (X=C) Yamamoto-type Ullmann cross-
coupling reaction65 scheme 6
5600
PAF-2 TCM cyclotrimerization59 scheme 5 891 054 106
PAF-3 TkBPM-X (X=Si) Yamamoto-type Ullmann cross-
coupling reaction76 scheme 6
2932 154 127
PAF-4 TkBPM-X (X=Ge) Yamamoto-type Ullmann cross-
coupling reaction76 scheme 6
2246 145 118
PAF-11 TkBPM-X (X=C) BPDA Suzuki coupling reaction68 704 0203 166
54
scheme 6
3 Studies of structure and properties of COFs
31 2D COFs
Following the literature of the years 2005-2010 2D COFs were believed to have high symmetric AA
(eclipsed ndash P6mmm symmetry) stacking as boron nitride (h-BN) [REF] An exception was COF-1
which was considered to have AB (staggered ndash P63mmc) stacking as graphite[REF] The AA stacking
maximizes the attractive London dispersion interaction between the layers an important
contribution to the stacking energy At the same time AA stacking always suffers repulsive Coulomb
force between the layers due to the polarized connectors as the distance between atoms exposing
the same charge is closest It should be noted that in boron nitride boron atoms serve as nearest
neighbors to nitrogen atoms in adjacent layers The Coulomb interaction has ruled out possible
interlayer eclipse between atoms of alike charges Based on these facts we modeled77 a set of 2D
COFs with different possible stacks Each layer was drifted with respect to the neighboring layers in
directions symmetric to the hexagonal pore and we calculated the total energies and their Coulombic
contributions The AA stacking version was found to have the highest Coulombic repulsion in each
COF The lattice types with the layers shifted to each other by ~14 Aring approximately the bond length
between two atoms in a 2D COF were found to be more stable compared to AA or AB In this low-
symmetry configuration the hexagonal rings in the pore walls undergo staggering with that of
adjacent layers Within an error limit the bulk COFs may be either serrated or inclined as shown in
Figure 5 The interlayer separations of all the COF stacks are in the range of 30 to 36 Aring and increase
in the order AB serratedinclined AA32 The motivation for proposing inclined stacking for COFs is
that the rhombohedral graphite (ABC stacking) is the inclined version of the layer-offset in natural
graphite (AB stacking) The instability of AA stacking was also suggested by the studies of elastic
properties of COF-1 and COF-5 by Zhou et al78 The shear modulus show negative signs for the
vertical alignment of COF layers while they are small but positive for the offset of layers
55
Figure 5 Different stacks of a 2D covalent organic framework COF-1 The atoms are shown as Van der Waals spheres
The different stacking modes should in principle be visible in their PXRD patterns as each space
group has a distinct set of symmetry imposed reflection conditions We simulated the XRD patterns
of COFs in their known and new configurations and on comparison with the experimental spectrum
the new as well as AA stacking showed good agreement3279 However the slight layer-offsets in
conjunction with the large number of atoms in the unit cells change the PXRD spectrum only by the
appearance of a few additional peaks all of them in the vicinity of existing signals and by changes in
relative intensities Unfortunately the low resolution of the experimental data does now allow
distinguishing between the different stackings as the broad signals cover all the peaks of the
simulated spectrum
A detailed analysis on the interlayer stacking of HHTP-DPB COF has been made by Dichtel et al73 is
very complementary73 This work uses molecular mechanics extensively to define potential energy
surfaces of stacking of two layers from eclipsed to staggered covering all possible layer offsets The
low energy region was near to the eclipsed stacking which was then zoomed in to examine using DFT
for higher accuracy The far-eclipsed regions encounter no energy barriers with the near-eclipsed
regions which suggest the unlikeness of forming near-staggered layering Within DFT the eclipsed
geometries are at high energy compared to the slight offsets around AA (~ 13 ndash 25 Aring away) For a
hexagonal COF the energetically preferred offset could be in one of the six hexagonal sectors (or
circle sectors) of central angle 60 The preference of falling into one sector over the others is
arbitrary and may be determined by symmetry set by nature or external parameters Hence the
layering order in bulk could be serrated inclined spiral or purely by random The latter case better
known as turbostratic disorder is then more like a rule for 2D COFs rather than an exception caused
by external forces This also explains why the experimental PXRD patters have relatively broad peaks
The layer-offset may not only change the internal pore structure but also affect interlayer exciton
and vertical charge transport in opto-electronic applications
56
Concerning the stacking stability the square 2D COFs are expected not to be different from
hexagonal COFs because the local environment causing the shifts is nearly the same DFTB based
calculations reported along with the synthesis of electron-transporting 2D-NiPc-BTDA COF supports
this notion47 Two shifted configurations ndash at 07 and 14 Aring away from AA ndash appeared to be
energetically preferred over the AA stacking It appears that AA stacking is only possible for boron
nitride-like structures were adjacent layers have atoms with opposite charges in vertical direction In
analogy to BN layers it might be possible to force AA stacking by the introduction of alkalis in
between the layers
32 3D COFs
3D COFs in general are composed of tetragonal and triangular building blocks So far that their
synthesis leads to two high symmetry topologies ndash ctn and bor - in high yields24 The two topologies
differ primarily in the twisting and bulging of their components at the molecular level The
thermodynamic preference of one topology over the other may result from the kinetic entropic and
solvent effects and the relative strain energies of the molecular components It is straight-forward to
state that the effects at the molecular level crucial crucial in the bulk state since transformation from
one net to the other is impossible without bond-breaking There has not been any detailed study on
this matter experimentally or theoretically
Schmid et al8182 have developed force-field parameters from first principles calculations using
Genetic Algorithm approach The parameters developed for cluster models of COF-102 can
reproduce the relative strain energies in sufficient accuracies and be extended to calculations on
periodic crystals The internal twists of aromatic rings within the tetragonal units are different for ctn
and bor nets which lead to stable S4 and unstable D2d symmetries respectively This together with
the bulging of triangular units in the bor net reasoned for energetic preference of ctn over bor for all
boron-based 3D COFs79
The connectivity-based atom contribution method (CBAC) developed by Zhong et al80 can
significantly reduce computational time needed for quantum chemical calculation for framework
charges when screening a large number of MOFs or COFs in terms of their adsorption properties The
basic assumption is that atom types with same bonding connectivity in different MOFsCOFs have
identical charges a statement that follows from the concept of reticular chemistry where the
properties of the molecular building blocks keep their properties in the bulk After assigning the
CBAC charges to the atom types slight adjustment (usually less than 3 ) is needed to make the
frameworks neutral Simulated adsorption isotherms of CO2 CO and N2 in MOFs and COFs show that
CBAC charges reproduce well the results of the QM charges8384 The CBAC method still requires a
57
well-parameterized force field in order to account correctly for adsorption isotherms as the second
important contribution to the host-guest interaction is the London dispersion energy between
framework and adsorbed moleculesG
The elastic constants calculated for the 3D COFs COF-102 and COF-108 show that COF-102 is nearly
five times stronger than COF-10878 While the bulk modulus (B) of COF-102 (B = 216 GPa) exceeds
that of known MOFs such as MOF-5 MOF-177 and MOF-205 (B = 153 101 107 GPa respectively81)
the least dense COF-108 is at the edge of structural collapse (B = 49 GPa) The calculations were
made using Density Functional Tight-Binding (DFTB) method Another report82 based on the same
level of theory possibly with a different parameter set however reveals lower bulk moduli for both
COFs 177 GPa for COF-102 and 005 GPa for COF-108 The bulk moduli for COF-103 and COF-105 are
110 and 33 GPa respectively Recently we have reported the structural properties of 3D COFs The
calculations were made using self-consistent charge (SCC) DFTB method The bulk modulus of each
COF is as follows COF-102 ndash 206 COF-103 ndash 139 COF-105 ndash 79 COF-108 ndash 37 COF-202 ndash 153 and
COF-300 ndash 144 GPa Force field calculations79 provide slightly different values COF-102 ndash 170 COF-
103 ndash 118 COF-105 ndash 57 COF-108 ndash 29 COF-202 ndash 110 GPa Albeit the differences in values each
type of calculation shows the trend that bulk modulus decreases with decreasing mas density and
increasing pore volume and distance between connection nodes One has to note that the high
mechanical stabilities of 3D COFs as suggested by the calculations correspond always to defect-free
crystals Theory is expected therefore to overestimate experimental mechanical stability for real
materials
COFs in general have semiconductor-like band gaps (~ 17 to 40 eV)32 Layer-offset from eclipsed
layering slightly increases the band gap However the Density-Of-States (DOS) of a monolayer is
similar to that of bulk COF The band gap is generally smaller for a COF with larger number conjugate
rings
The thermal influence on the crystal structure of 3D COFs is also intriguing as negative thermal
expansion (NTE) the contraction of crystal volume on heating has been reported for COF-10283 The
studies were performed using molecular dynamics with the force field parameters by Schmid et al84
However the thermal expansion coefficient α = -151 times 10-6 K-1 is much smaller compared to that of
some of the reported MOFs that also show NTE behavior85 The volume-contraction upon the
increase of temperature arises from the tilt of phenyl linkers between B3O3 rings and sp3 carbon
atom in TBPM (figure ) Later Schmid et alreported that COF-102 103 105 108 exhibit NTE
behavior both in ctn and bor topologies whereas COF-202 only in ctn topology79 A typical
application is the realization of controllable thermal expansion composites made of both negative
and positive thermal expansion materials
58
4 Applications
41 Gas storage
The success in the synthesis of COFs was certainly the result of a long-term struggle for complete
covalent crystallization The discovery of COFs coincided with the time when world-wide effort was
paid to develop new materials for gas storage in particular for the development hydrogen tanks for
mobile applications Materials made exclusively from light-weight atoms and with large surface
areas were obviously excellent candidates for these applications The gas storage capacity of porous
materials relies on the success of assembling gas molecules in minimum space This is achieved by
the interaction energy exerted by storage materials on the gas molecules Because the interactions
are noncovalent no significant activation is required for gas uptake and release and hence the so-
called physical adsorption or physisorption is reversible The other type of adsorption ndash chemical
adsorption or chemisorption ndash binds dissociated hydrogen covalently or interstitially at the risk of
losing reversibility As it requires the chemical modification of the host material chemisorption is not
a viable route for COFs and might become possible only through the introduction of reactive
components into the lattice The total amount of gas adsorbed in the pores gives rise to what is
referred to as lsquototal adsorption capacityrsquo This quantity is hard to be assessed experimentally as a
measurement is always subjected to influence of the materials surface and gas present in all parts of
the experimental setup Therefore the lsquoexcess adsorption capacityrsquo is reported in experiment Here
the gas stored in the free accessible volume is subtracted from the total adsorption In experiment
this volume includes the voids in the material as well as any empty space between the sample
crystals This volume is typically determined by measuring the uptake of inert gas molecules (He for
H2 adsorption N2 or Ar for adsorption studies of larger molecules) at room temperature with the
assumption that the host-guest interaction between the material and He can be neglected The
different definitions of adsorption is given in Figure 6
Typically experiments measure excess values and simulations provide total values Quantities of
adsorbed molecules are usually expressed as gravimetric and volumetric units The former gives the
amount per unit mass of the adsorbent while the latter gives the amount per unit volume of the
adsorbent Explicative definitions and terminologies related to gas adsorption can be found
elsewhere86
59
Figure 6 Hydrogen density profile in a pore A denotes the excess A+B the absolute and A+B+C the total adsorption The skeletal volume corresponds to the volume occupied by the framework atomsCourtesy of Dr Barbara Streppel and Dr Michael Hirscher MPI for intelligent systems Stuttgart Germany
411 Porosity of COFs
It is common to inspect the porosity of synthesized nanoporous materials using some inert or simple
gas adsorption measurements Distribution of pore size can be sketched from the
adsorptiondesorption isotherm N2 and Ar isotherms provide an approximation of the pore surface
area pore volume and pore size over the complete micro and mesopore size range Usually the
surface area is calculated using the Brunauer-Emmet-Teller (BET) equation or Langmuir equation
Total pore volume (volume of the pores in a pre-determined range of pore size) can be determined
from either the adsorption or desorption phase The Dubinin-Radushkevic method the T-plot
method or the Barrett-Joyner-Halenda (BJH) method may also be employed for calculating the pore
volume The pore size is often corroborated by fitting non-local density functional theory (NLDFT)
based cylindrical model to the isotherm90-93 In some cases the desorption isotherm is affected by
the pore network smaller pores with narrower channels remain filled when the pressure is lowered
This leads to hysteresis on the adsorption-desorption isotherms The different methods employed for
pore structure analysis are characteristic for micropore filling monolayer and multilayer formations
capillary condensation and capillary filling
For any adsorbate in order to form a layer on pore surface the temperature of the surface must
yield an energy (kBT) much less than adsorbate-surface interaction energy Additionally the absolute
value of the adsorbate-surface binding energy must be greater than the absolute value of the
adsorbate-adsorbate binding energy This avoids clustering of the adsorbate into its three-
dimensional phase
60
At high pressure the adsorption isotherm shows saturation which means that no more voids are left
for further occupation The isotherms show different behaviors characteristic of the pore structure of
the materials There are known classifications based on these differences type I II III IV and V For
COFs and the related materials discussed in this review type I II and IV have been observed (see
Figure 7) As more adsorbate is attracted to the surface after the first surface layer completion one
can expect a bend in the isotherm Type I implies monolayer formation which is typical of
microporosity If the surface sites have significantly different binding energies with the adsorbate
type II behavior is resulted In type IV each ldquokneerdquo where the isotherm bends over and the vapor
pressure corresponds to a different adsorption mechanisms (multiple layers different sites or pores)
and represents the formation of a new layer
Figure 7 Different types of adsorptions in Covalently-bound Organic Frameworks
Argon and nitrogen adsorption measurements at respectively 87 and 77 K exhibit type I isotherms
for COF-1 6 102 and 103 and type IV isotherms for COF-5 8 10 102 and 10387 Smaller pore
diameters of COF-1 and 6 (~ 9 Aring each) are characteristic of monolayer gas formation on the internal
pore surface The same reasons are responsible for the type I character of COF-102 and COF-103
(pore size = 12 Aring each) isotherms albeit with some unusual steps at low pressure The type IV
isotherms of COF-5 8 10 102 and 103 show formation of monolayers followed by their
multiplication within their relatively larger pores (sizes are approximately 27 16 32 12 and 12 Aring
respectively) Other COFs that exhibit type I isotherms for N2 adsorption are the 3D COFs COF-202 (11
Aring) COF-300 (72 Aring) and COF-102-C12 (12 Aring) the alkylated 2D COF series COF-18Aring COF-16Aring COF-14Aring
COF-11Aring (the nomenclature includes their pore sizes) and a 2D square COF NiPc COF (19 Aring)
Isotherms like Type-II were observed for PPy-COF (188 Aring) COF-366 (176 Aring) COF-66 (249 Aring) Pc-
PBBA COF (217 Aring) ZnP-COF (25 Aring) COF-LZU1 (12 Aring) PdCOF-LZU1 (12 Aring) and some of the azide-
appended COFs 50AcTrz-COF-5 (156 Aring) 75AcTrz-COF-5 (153 Aring) 100AcTrz-COF-5 (125 Aring)
50BuTrz-COF-5 (232 Aring) 50PhTrz-COF-5 (212 Aring) 50EsTrz-COF-5 (212 Aring) and 25PyTrz-COF-5
(221 Aring) The majority of the COFs with mesopores exhibit type-IV isotherms They are TP-COF (314
Aring) BTP-COF (40 Aring) COF-42 (23 Aring) COF-43 (38 Aring) HHTP-DPB COF (47 Aring) CTC-COF (226 Aring) NiPc-PBBA
COF ZnPc-Py-COF (31 Aring) ZnPc-DPB-COF (31 Aring) ZnPc-NDI-COF (31 Aring) ZnPc-PPE-COF (34 Aring) 5N3-
61
COF (295 Aring) 25N3-COF (276 Aring) 50N3-COF (259 Aring) 75N3-COF (258 Aring) 100N3-COF (227 Aring)
5AcTrz-COF-5 (298 Aring) 25AcTrz-COF-5 (243 Aring) 5PyTrz-COF-5 (289 Aring) and 10PyTrz-COF-5
(242 Aring)
The synthesized microporous amorphous materials CTF-1 (12 Aring) CTF-DCP (14 Aring) PAF-1 and PAF-2
(106 Aring) exhibit type-I isotherms with N2 adsorption PAF-3 (127 Aring) PAF-4(118 Aring) PAF-5 (157 Aring)
PAF-6 (118 Aring) and PAF-11 showed surprisingly not the typical type I behavior in spite of their
microporosity but type-II isotherms
Many of the mesoporous COFs showed small hysteresis in the adsorption-desorption isotherm
pointing the possibility of capillary condensation Hysteresis was observed for the amorphous
materials in both mirco and meso-pore range Various reasons have been attributed for the observed
hysteresis including the softness of the material and guest-host interactions
412 Gas adsorption experiments
Hydrogen uptake capacities of some boron-based COFs were measured by Yaghi et al87 The excess
gravimetric saturation capacities of COF-1 -5 -6 -8 -10 -102 and -103 at 77 K were found to be 148
358 226 350 392 724 and 705 mg g-1 respectively The saturation pressure increases with an
increase in pore size similar to MOFs88 Pore walls of 2D COFs consist only of edges of connectors
and linkers The fact that faces and edges are largely available for interactions with H2 in 3D
geometries is a reason for their enhanced capacity Total adsorption generally increases without
saturation upon pressure because the difference between the total and the excess capacities
corresponds to the situation in bulk (or in a tank) COFs show in particular good gravimetric
capacities because of their low mass density while volumetric capacities typically do not exceed
those of MOFs The gravimetric hydrogen storage capacities of COF-102 and -103 exceeds 10 wt at
a pressure of 100 bar
COF-18 and its alkylated versions COF-16 -14 and -1 have H2 excess gravimetric capacities 155 144
123 and 122 wt respectively at hellipK and hellipbar
Excess uptake of methane in COFs at room temperature and at 70 bar pressure is as follows COF-1
and -6 up to10 wt COF-5 -6 and -8 between 10 and 15 wt and COF-102 and -103 above 20
wt 87 Isosteric heat of adsorption Qst calculated by fitting the isotherms to virial expansion with
the same temperature are relatively weaker (lt 10 kJmol) for COF-102 and COF-103 at low
adsorption which implies weaker adsorbate-adsorbent interactions In comparison COF-1 and 6
exhibit twice the enthalpy however the overall hydrogen storage capacity was very limited due to
62
the smaller pore volume High Qst at zero-loading is not desirable because the primary excess amount
adsorbed at very low pressures cannot be desorbed practically89
COF-102 and 103 outperform 2D COFs for CO2 storage87 The saturation uptakes at room
temperature were 1200 and 1190 mg g-1 for COF-102 and 103 respectively
A type IV isotherm has been observed for ammonia adsorption in COF-10 inherent to its mesoporous
nature90 The material showed an uptake of 15 mol kg-1 at room temperature and 1 bar the highest
of any nanoporous materials The repeated adsorption-desorption cycles were reported to disrupt
the layer arrangement of this 2D COF Yaghi et al90 were also able to make a tablet of COF-10 crystal
which performed nearly up to the crystalline powder
Not many COFs have been experimentally studied for gas storage applications in spite of high
expectations This has to be understood together as a result of the powder-like polycrystallization of
COFs The enthalpy Qst at low-loading accounted to only 46 kJmol
The excess and total hydrogen uptake capacity of PAF-1 at 48 bar and 77 K reached 7 wt and 10
wt respectively65 PAF-3 and PAF-478_ENREF_73 follow the same diamond topology as PAF-1 the
difference accounts in the central atoms of the tetragonal blocks PAF-1 3 and 4 have C Si and Ge
atoms at the centers respectively At 1 bar the respective adsorption capacities were 166 207 and
150 wt The highest value being found for PAF-3 is attributed to its relatively high enthalpy (66 kJ
mol-1) compared to PAF-1 (54 kJ mol-1) and PAF- 4 (66 kJ mol-1) At high pressure the surface area is
a key factor and because of the lower BET surface area PAF-3 and PAF-4 performs poor At 60 bar
their respective adsorption capacities were 55 and 42 wt For PAF-11 hydrogen uptake was 103
wt at 1 bar68
Methane uptakes of PAF-1 3 and 4 at 1 bar are 13 19 and 13 wt respectively76 Their enthalpies
are respectively 14 15 and 232 kJ mol-1 This suggests that Ge moiety have strong interactions with
methane
CO2 adsorption in PAF-1 at 40 bar and room temperature was 1300 mg g-1 which was slightly more
than the capacity of COF-102 and 10365 PAF-1 3 and 4 at 1 atm pressure could store 48 80 and 51
wt respectively At 273 K and 1 atm they stored 91 153 and 107 wt respectively The storage
capacities at 273 K and 298 K lead to the following zero-loading isosteric enthalpies 156 192 162
kJ mol-1 for PAF-1 3 and 4 respectively The higher uptake of PAF-3 may also be explained by its
relatively higher surface area with CO2 molecules
The selective adsorption of greenhouse gases in PAFs was discussed by Ben et al76 At 273 K and 1
atm pressure PAF-1 selectively adsorbes CO2 and CH4 respectively 38 and 15 times more in
63
amounts than N2 PAF-3 showed extraordinary selectivity of 871 for CO2 over N2 and 301 for CH4
over N2 For PAF-4 the selectivity was 441 for CO2 over N2 and 151 for CH4 over N2 Generally the
uptake of N2 Ar H2 O2 are significantly lower compared to CO2 and NH4 which makes these PAFs
suitable for separating them
Harmful gases like benzene and toluene can be stored in PAF-1 in amounts 1306 mg g-1 (1674 mmol
g-1) and 1357 mg g-1 (1473 mmol g-1) respectively at saturated pressures and room temperature65
In PAF-11 their adsorptions were in amounts of 874 mg g-1 and 780 mg g-1 respectively68 PAF-2 was
accounted to a much lower benzene adsorption (138 mg g-1) at the same conditions59 Adsorption of
cyclohexane vapor in PAF-2 was only 7 mg g-1 While PAF-11 exhibited hydrophobicity it could
accommodate significant amount of methanol (654 mg g-1 or 204 mmol g-1) at room temperature
and saturated pressure68 PAF-5 has been tested for storing organic pollutants69 At room
temperature and saturated pressure PAF-5 adsorbed methanol benzene and toluene in amounts
6531 cm3 g-1 (292 mmol g-1) 26296 cm3 g-1 (117 mmol g-1) and 2582 cm3 g-1 (115 mmol g-1)
respectively Their adsorptions in PAF-5 were deviated from the typical type-I behavior Methanol
exhibited a large slope at the high pressure region corresponding to the relative size of it Zhu G et
al dedicated PAF-6 for biomedical applications With no toxicity observed for PAF-6 over a range of
concentrations adsorption of ibuprofen was measured in it PAF-6 showed an uptake of 350 mg g-1
The release of ibuprofen was also tested in simulated body fluid which showed a release rate of 50
in 5 hours 75 in 10 hours and 100 in almost 46 hours
413 Theoretical predictions
Grand canonical Monte Carlo (GCMC) is a widely used method for the simulation of gas uptakes in
nanoporous materials In GCMC the number of adsorbates and their motions are allowed to change
at constant volume temperature and chemical potential to equilibrate the adsorbate phase The
motions are random guided by Monte Carlo methods and the energies are calculated by force field
methods The details of it may be found in the literature91
Goddart III WA et al92 simulated the uptake capacities of boron-based COFs using force fields derived
from the interactions of H2 and COFs calculated at MP2 level of theory The saturated excess uptakes
of both COF-105 (at 80 bar) and 108 (at 100 bar) were accounted to 10 wt at 77 K which are more
than the uptakes measured in ultrahigh porous MOF-200 205 and 21093 The predictions for other
COFs include 38 wt for COF-1 (AB stacking) 34 wt for COF-5 (AA stacking) 88 wt for COF-102
and 91 wt for COF-103 At 01 bar COF-1 in AB stacking showed highest excess uptake (17 wt )
compared to other COFs in the present discussion Total uptake capacities of the COFs are in the
following order COF-108 (189 wt ) COF-105 (183 wt ) COF-103 (113 wt ) COF-102 (106
64
wt ) COF-5 (55 wt ) and COF-1 (38 wt ) It is worth noticing the higher gravimetric capacities of
COF-108 and 105 which were not measured experimentally They benefit from their lower mass and
higher pore volume In case of volumetric capacity COF-102 (404 gL excess) surpasses the others at
high pressures The total volumetric capacities of the COFs are 499 498 399 395 361 and 338
gL for COF-102 103 108 105 1 and 5 respectively Their additional calculations predicted benzene
rings as favorite locations for H2 molecules
Similar values and trends have been reported by Garberoglio G94 who also reasoned poor solid-fluid
interactions in a large part of free volume for the low volumetric capacity of COF-108 and 105 A
room temperature excess gravimetric uptake of approximately 08 wt was obtained for COF-108
and 105 Quantum diffraction effects were added to the interaction energies of H2 with itself and the
material which were calculated using universal force-field (UFF) With possible overestimation
Klontzas et al30 and Bassem et al95 have reported total gravimetric uptake of 45 and 417 wt
respectively at 300 K for the same COFs Among the boron-based 3D COFs COF-202 exhibited much
smaller uptake capacity at high pressures due to its smaller pore volume95 Lan et al96 predicted a
very small H2 uptake of 152 wt COF-202 at 300K and 100 bar Studies of Yang et al97 clarifies that
pore filling with H2 in 2D COFs is a gradual process rather than being in steps of layer formation
Garberoglio and Vallauri98 pointed out that the relatively higher mass density and lower surface area
per unit volume of 2D COFs are the reasons for their lower uptakes in comparison with 3D COFs The
surface area of hexagonal 2D COFs (AA stacking) averaged around 1000 m2 cm-3 while that of 3D
COFs were about 1500 m2 cm-3
Simulations of H2 adsorption in the perfect crystalline form of PAF-1 and PAF-11 also known as PAF-
302 and PAF-304 respectively are carried out by Zhu et al99 At 77 K H2 excess uptake of PAF-302 (7
wt ) is slightly higher than COF-102 (684 wt )87 and slightly lower than MOF-177 (740 wt )100 At
room temperature PAF-302 exhibited a poor uptake (221 wt at 100 bar) PAF-304 showed
excellent total uptakes 2238 wt at 77 K and 100 bar 653 wt at 298 K and 100 bar These are
highest among all nanoporous materials
Babarao and Jiang studied room-temperature CO2 adsorption in COFs101 At low pressures COFs with
smaller pore volume exhibit a steep rise in their isotherm while at high pressures COF-105 and 108
(82 and 96 mmol g-1 respectively at 30 bar) dominate the others COF-6 has the highest isosteric heat
of adsoption (328 kJmol) hence it made the first steep rise in the volumetric isotherm followed by
COF-8 102 103 10 105 and 108 Correlations of gravimetric and volumetric capacities with mass
density pore volume porosity and surface area have been excellently manifested in this article101
With increasing framework-density gravimetric uptake falls inversely while volumetric capacity
decreases linearly The former rises with free volume while the latter rises and then drops slightly
65
Both of them rise with porosity and surface area Choi et al102 predicted 45 times more CO2 uptake in
COF-108 (3D) compared to COF-5 (2D) Recently Schmid et al79 also have predicted CO2 adsorption
especially for 3D COFs The uptake of COF-202 was almost half of COF-102 and 103 at room
temperature Type IV isotherm has been found for CO2 adsorption in COF-8 and 10 at low
temperatures (lt 200 K)97 Yang and Zhong97 excluded capillary condensation and dissimilar
adsorption sites but multilayer formation as reasons of the stepped behavior (Yang and Zhong
explained this as a consequence of multilayer formation rather than a result of capillary
condensation or dissimilar adsorption sites)
Methane adsorption in 2D and 3D COFs was first calculated by Garberoglio et al Among COF-6 8 and
10 the former which has smaller pore size and higher binding energy with CH4 shows better
volumetric adsorption (~ 120 cm3cm3 at 20 bar) than the others at room temperature at low
pressure region (lt 25 bar) Larger pore size favors COF-10 for higher volumetric adsorption (~ 160
cm3cm3 at 70 bar) at high pressures COF-8 with intermediate pore size shows largest excess amount
of methane adsorbed upon saturation however still lower than two 3D COFs COF-102 and 103
show excess volumetric adsorption of about 180 cm3cm3 at 35 bar This is significantly higher than
the uptake of two other 3D COFs COF-105 and 108 Entropic effects which primarily arise from the
change of dimensionality when the bulk phase of adsorbate transforms to adsorbed phase are
crucial in adsorption Garberoglio94 shows that solid-fluid interaction potential in large part of volume
of COF-105 and 108 is not strong enough to overcome the entropic loss On the other hand these
two COFs exhibit relatively higher gravimetric uptake (720 and 670 cm3g respectively) Goddard III et
al89 predicted total methane uptakes in the following amounts 415 wt in COF108 405 wt in
COF-105 31 wt in COF-103 284 wt in COF-102 196 wt in COF-10(AA) 169 wt in COF-
5(AA) 159 wt in COF-8(AA) 123 wt in COF-6(AA) and 109 wt in COF-1(AB) Yang and Zhong97
have shown that even at low temperatures (100 K) the pore filling of COF-8 with CH4 is rather
gradual than by step-by-step whereas for COF-10 multilayer formation provides a stepped behavior
in the isotherm Alternately Goddard et al pointed out that layer formation coexists with pore-filling
at room temperature89
414 Adsorption sites
First principle calculations on cluster models are typically employed to find favorite adsorption sites
and binding energies of adsorbates within frameworks Benzene rings are found to be a usual
location for H2 where it prefers to orient in perpendicular fashion3086110 Other favorite locations
include B3O3 and C2O2B rings110111 where H2 orients horizontally on the face as well as on the
edge3086 The central ring of HHTP does not provide enough binding to H2 atoms because of small
amount of charges There are some differences in the adsorption energies and favorite sites
66
calculated at different levels of theory Overall the reported binding energies between H2 and any
COF fragment are less than 6 kJ mol-1 In comparison to MOFs COFs do not offer any stronger binding
energy with H2 The advantage of COFs is their low mass density Lan et al103 reported that CO2 is
more preferred to be adsorbed on B3O3 or C2O2B rings than benzene rings Choi et al102 reported that
the oxygen sites are the primary locations for CO2 and CO2 bound COFs are the second strong binding
sites
415 Hydrogen storage by spillover
Hydrogen spillover is an alternate way to store hydrogen113114 It involves dissociation of hydrogen
gas by supported metal catalysts subsequent migration of atomic hydrogen through the support
material and final adsorption in a sorbent Here surface-diffusion of H atoms over the support is
known as primary spillover Secondary spillover is the further diffusion to the sorbent Bridging the
metal part with the sorbent is a practice to enhance the uptake It increases the contact between the
source and receptor and reduces the energy barriers especially in the secondary spillover As the
final sorption is chemisorption surface area of the sorbent is more important than pore volume
Yang and Li measured H2 storage in COF-1 by spillover They used Pt supported on active carbon
(PtAC) as source for hydrogen dissociation Active carbon was the primary receptor and COF-1 the
secondary receptor COF-1 showed much low uptake 128 wt at 77 K and 1 atm 026 wt at 298
K and 100 bar In comparison to MOFs these are very low104 However they have found that the
uptake could be scaled up by a factor of 26 by bridging COF-1 with PtAC by applying carbonization
They also report that heat of adsorption between H and surface sites is more important than surface
area and pore volume in enhancing the net adsorption by spillover
Ganz et al theoretically predicted that the uptake capacities in COFs by hydrogen spillover can be
higher than the measured value116117 Based on ab initio quantum chemistry calculations and
counting the active storage sites they predicted 45 wt for COF-1 in AB stacking and 55 wt for
COF-5 in AA stacking at room temperature and 100 bar
42 Diffusion and Selectivity
Computed self-diffusion coefficients for H2 and CH4 adsorbates in 2D COFs (in AA stacking) are up to
one order of magnitude higher than that in 3D MOFs such as MOF-598 In comparison to nanotubes
the diffusion of H2 molecules in 2D COFs is one order of magnitude slower The differences in
diffusion coefficients are attributed to different pore structures Interactions of the corners of the
hexagonal pore with the fluid present potential barriers along the cylindrical pore axis Diffusion
occurs with jumps after long waiting The height of the barrier is still smaller than in MOFs
67
Homogeneous pore walls and absence of pore corners in nanotubes contribute much less
corrugation to the solid-fluid potential energy surface Increase of self-diffusivities of H2 and CH4 with
increase in pore size of 2D COFs is also shown by Keskin[Ref] Due to the smaller size of H2 its
diffusivity is higher than that of CH4 in any material For reasons of their relative size In a mixture of
the two the self-diffusivity of CH4 increases while that of H2 decreases
Molecular dynamics simulation studies of the diffusion of gas molecules within a 2D COF performed
by Krishna et al105 revealed its relatively lower binding energy of hydrogen molecules within the pore
walls compared to argon CO2 benzene ethane propane n-butane n-pentane and n-hexane
Binding energy prevents the molecules from diffusing through the pore channels They tested if
Knudsen diffusion is applicable to COFs Knudsen diffusion occurs when the molecules frequently
collide with the pore wall This generally happens when the mean free path is larger than the pore
diameter The simulations had been carried out in BTP-COF which accounts a pore diameter of 4 nm
It is found out that the ratio of zero-loading diffusivity with the Knudsen diffusivity has a significant
correlation with isosteric heat of adsorption the stronger the binding energy of the molecules with
the walls the lower the ratio Hydrogen being an exception among the investigated molecules
exhibited the ratio close to unity while others with their isosteric heat of adsorption higher than 10
kJmol-1 have a value close to 01 For a mixture of two gases with significant differences in binding
energies the ratio of self-diffusivities affirms high diffusion selectivity
Selectivity of CH4 over H2 in COF-5 6 and 10 is studied by Keskin106 Narrow pores support the
selective adsorption of CH4 over H2 however they suffer from entropic effects at high pressures
which favor H2 adsorption Liu et al107 have reported that the adsorption selectivity of COFs and
MOFs are in general similar up to 20 bar Delta loading or working capacity (usually expressed in
molkg) is an important term often used to show the economics of the selective adsorption This is
defined as the difference in loadings of the preferred gas at adsorption and desorption pressures
Adsorption is usually in the range 1 to 100 bar while desorption in 01 to 1 bar High selectivity and
high working capacity are preferential for practical use COF-6 has higher selectivity among the three
studied COFs but lower working capacity106 Best selectivity-working capacity combination is shown
by COF-5106 Nonetheless it is not better than CuBTC or CPO-27-X (X = Zn Mg)121122 Liu et al107
attributed the high isosteric heat of adsorption of COF-6 and Cu-BTC for their high adsorption
selectivity They also pointed out that the electrostatic contribution of framework charges in COFs
are smaller than in MOFs however needs to be taken into account
While CH4 is strongly adsorbed H2 undergoes faster diffusion Hence the product of adsorption
selectivity (gt1) and diffusion selectivity (gt0 lt1) which is membrane selectivity is smaller than
adsorption selectivity Keskin106 has compared the selectivity of COF-membranes with known
68
membranes The selectivity of COF-10 is similar to MOF-177 and IRMOFs COF-5 and 6 outperform
them It is to be noted that COF-5 is CH4 selective while COF-10 is H2 selective although their
topologies are similar CH4 permeability of COF membranes is much higher than several MOFs and
ZIFs COF-5 and 6 exhibited high membrane selectivity together with high membrane permeability
Some discussions on the adsorption and membrane based selectivity of COF-102 in comparison with
IRMOFs and Cu-BTC for CH4CO2H2 mixtures can be found in ref108 Adsorption selectivity of COF-6
and Cu-BTC in CH4H2 and CH4CO2 mixtures surpass other COFs and MOFs107 sdfalf
43 Suggestions for improvement
The level of achievement made by COFs and related materials yet do not practically meet the
practical requirements of several applications Hence thoughts for improvement primarily focused
on overcoming their limitations and enhancing characteristic parameters Some theoretical
suggestions for improved performances are mainly discussed here
431 Geometric modifications
Functionalities may be improved by utilizing the structural divergence of framework materials
Several examples may be found in the literature for MOFs etc Klontzas et al suggested replacement
of phenyl linkers in COF-102 by multiple aromatic moieties while keeping the topology unchanged to
increase surface area and pore volume109 They used biphenyl triphenyl naphthalene and pyrene
linkers in the new COFs All of them showed enhanced gravimetric capacity compared to the parent
COF One of the modified COF-102 with triphenyl linker showed 267 and 65 wt at 77 K and 300 K
respectively at 100 bar This kind of replacement may affect the adsorption of each adsorbate
differently leading to the alteration of the selective adsorption of one component over the other110
Following the reticular construction scheme we modeled some new 2D COFs by cross-linking some
of the familiar and unfamiliar building blocks in the COF literature32 This showed the structural
divergence of COFs however they exhibited structural and electronic properties analogues to other
2D COFs
Similar modifications in 2D and 3D COFs for high CO2 uptake were made by Choi et al102 DPABA
(diphenylacetylene-44ʹ-diboronic acid) and its tetrahedral version TBPEPM (tetra[4-(4-
dihydroxyborylphenyl)ethynyl]phenyl) in place of BDBA in COF-5 and TBPMTBPS in COF-108COF-
105 respectively were used for new COF designs The new 2D COF promised a saturated CO2 uptake
higher than in COF-102 because of its large pore volume The new 3D COFs were worth for an uptake
twice more than in COF-105 and 108
69
Goddard et al also designed about 14 new COFs by substituting the aromatic rings in the tetragonal
part (TBPM or TBPS) of COF-102 COF-103 COF-105 COF-108 and COF-202 with vinyl groups or alkyl-
functionalized extended or fused aromatic rings111 The new designs adopted their parent topology
and their structural stability was confirmed by analyzing molecular dynamics (MD) simulations at
room temperature Fused aromatic rings and vinyl groups provided relatively lower and the highest
zero-loading isosteric heat of adsorption (Qst) respectively however the room-temperature delivery
amounts of methane in the respective COFs crossed the DOE target at 35 bar111 In the latter
methane-methane interaction compensated Qst on high-loading
Lino M A et al112 theoretically inspected the ability of layered COFs to exhibit structural variants of
layered materials The hexagonal 2D COF layers may be rolled into the form of nanotubes (NT) or
may exist in the form of fullerenes with the inclusion of pentagons One could think of a formula unit
which resembles the carbon atoms in carbon nanotubes (CNT) or fullerenes The NTs in general can
have any chirality although the study included only armchair and zigzag NTs Density Functional
Theory based calculations reveal that COF-1 NTs of diameter greater than 15 Aring is energetically
favored This was concluded from the comparison of strain-energy per formula unit of the COF-1 NTs
with that of small diameter CNTs The strain energies of zigzag and armchair nanotubes show similar
quadratic functions of tube-diameter The Youngrsquos modulus of COF-1 (22) NT is found out to be 120
GPa This is much smaller compared to typical CNTs which have Youngrsquos modulus average around
1100 GPa113 Band gaps of COF layers remain unchanged when they roll up into nanotubes A COF-1-
fullerene built by scaling C60 molecule has large diameter and hence much less strain energy
compared to C60 Babarao and Jiang reported that the predicted CO2 adsorption capacity of COF-1 NT
is similar to that of CNTs101
Balance between mass density and surface area and hence high gravimetric and volumetric
capacities can be achieved by proper utilization of pore volume or proper distribution of pores Choi
et al theoretically predicted high gravimetric and volumetric capacities for H2 in a pillared COF114 A
pyridine molecule vertically placed above a boron atom in the boroxine ring of COF-1 layer is found
energetically favorable however with wrinkling of the layer115 Here nitrogen atom in pyridine forms
a covalent bond with the boron atom This pillaring increases the separation between the layers
exposes the buried faces of boroxine and aromatic rings to pores and hence increases surface area
and free volume Accessible surface area and free volume have been tripled and gravimetric and
volumetric capacities were calculated to be 10 wt and 60 g L-1 respectively at 77 K and 100 bar114
This volumetric capacity is higher than in NU-100116 and MOF-21093 which show ultrahigh surface
area
70
432 Metal doping
Miao Wu et al studied doping of COF-10 with Li and Ca atoms for hydrogen storage117 The metal
dopants transferred charges to substrate which in turn provided large polarization to hydrogen
molecules Similar observation has been made by Lan et al96 for Li atoms in COF-202 However they
showed the tendency to aggregate at high concentration Lan et al extensively studied doping of
positively charged alkali (Li Na and K) alkaline-earth (Be Mg Ca) and transition (Sc and Ti) metals in
COF-102 and 105 for enhanced CO2 capture103 They found that benzene rings and the three outer
rings of HHTP are the favorite sites for all the metal dopants Other interested locations are edges of
benzene rings and oxygen atoms B3O3 C2O2B rings and the central ring of HHTP were the unwanted
areas Lithium showed stability on the favorite locations while sodium and potassium tended to
cluster even at low concentrations Alkaline-earth metals only weakly interacted with the COFs
whereas transition metals chemisorbed in them which would possibly damage the substrate Lithium
is found out to be a good dopant for enhanced gas storage
Doping electropositive metals would be of advantage because they provide stronger binding with H2
(~ 4 kcalmol)103125133134 The effect of counterions in COFs is studied by Choi et al118 They found out
that although the binding energy of LiF is smaller than Li ion F counterion can hold one hydrogen
atom Additionally Li+ and Mg2+ ions preferred to be isolated than aggregated A step further
Goddard et al119 have recently shown that in presence of tetrahydrofuran (THF a solvent) an
electron is transferred from Li to the benzene ring whereas in the case of gas phase the electron
remained in the atom Additionally they suggested ways to remove solvents which are weakly
coordinated to the material Zhu et al110 suggested doping Li+ after replacement of hydrogen atom by
oxygen ion in COF-102 Another suggestion was the replacement of hydrogen atom by O--Li+ group
Both the cases exhibited higher CO2 adsorption than in the case of Li doping below 10 bar At 10 bar
the three cases exhibited almost the same uptake capacity Below 1 bar the doped Li+ provided
stronger interaction with quadrupolar CO2 in comparison with the substituted O--Li+ group The
differences at low pressures are attributed to the differences in the magnitude of the charge of Li
The doped Li+ remained to hold nearly +1 charge whereas the inclusion of O--Li+ to the framework
diminished the charge of Li+ to a moderate amount Doping Li atom added only relatively small
amount of charge to Li
Doping with Li+ could increase the H2 binding energies up to 28 kJmol118 With properly distributed
metal ion dopants the H2 uptake capacity in COF-108 is predicted to be 65 wt Cao et al also
predicted 684 and 673 wt of H2 uptake in Li-doped COF-105 and 108 respectively at room
temperature and 100 bar120 In Li-doped COF-202 the predicted uptake was 438 wt for the same
conditions96 Goddard et al119 observed that high performance of Li doped COFsMOFs at low
71
pressures (1 ndash 10 bar) is a disadvantage when calculating delivery uptake capacity Na doping could
overcome this difficulty as its heat of adsorption is lower than in Li doped materials Their predicted
delivery gravimetric capacities from 1 to 100 bar at room temperature for Li and Na doped COF-102
and 103 were higher than the 2010 DOE target of 45 wt at room temperature
Lan et al described that CO2 is polarized in presence of Li cation and the polarization increases when
Li cation is physisorbed in COF103 Their calculated uptake capacities of lithium-doped COF-102 and
COF-105 at low loading is about eight and four times higher than that of nondoped ones respectively
Also a very high uptake of 2266 mgg was predicted in the doped COF-105 at 40 bar Li-doped COF-
102 and COF-103 also showed higher uptakes of methane ndash almost double the amount ndash compared
to nondoped COFs at room temperature and low pressures (lt 50 bar)121 Binding energy between
doped Li cation and CH4 was calculated to be 571 kcalmol
Li-functionalized COF-105 realized by inserting alkoxide groups provided very high volumetric uptake
of H2 Klontzas et al122 replaced three hydrogen atoms in HHTP by oxygen atoms in order to achieve
the functionalization In spite of the increased weight due to the additional oxygen atoms the COF
exhibited gravimetric capacity of 6 wt at 300 K
Incorporation of lithium tetrazolide group into a new PAF having diamond topology and triphenyl
linkers for enhanced hydrogen adsorption is suggested by Sun et al123 The tetrazolide anion (CHN4-)
interacts with the lithium cation via ionic bonds The anion and the cation together can bind upto 14
hydrogen molecules Two lithium tetrazolide groups were introduced in the middle phenyl ring of
each triphenyl linker and GCMC calculations predicted 45 wt for the network at 233 K and 100 bar
With the intention of increasing the H2 binding energy Li et al chemically implanted metals in the
place of light atoms in COF-108124 They replaced the C2O2B rings in COF-108 by C2O2Al C2N2Al and
C2Mg2N and each new COF showed stability within DFT This kind of doping does not allow
aggregation of metal clusters Their GCMC simulations showed that the replacement strategy could
improve the hydrogen storage capacity at room temperature by a factor of up to three124 Zou et al
suggested that para-substitution of two carbon atoms in benzene rings by two boron atoms can
facilitate charge transfer between the rings and metal dopants125 Their work revealed that
dispersion of metals forbid any clustering and the doping could enhance the H2 uptake capacity
significantly
433 Functionalization
Functionalization of porous frameworks with ether groups for selective CO2 capture is suggested by
Babarao et al126 The electropositive C atom of CO2 interacts mainly with the electronegative O or N
72
atom of the functional groups They studied PAF-1 functionalized with ndashOCH3 ndashNH2 and ndashCH2OCH2ndash
groups in its biphenyl linkers The latter one which is the tetrahydrofuran-like ether-functionalized
PAF-1 showed high adsorption capacity for CO2 and high selectivity for CO2CH4 CO2N2 and CO2H2
mixtures at ambient conditions
5 Conclusions
Successful covalent crystallization resulted as a class of materials Covalent Organic Frameworks This
review portrays different synthetic schemes successful realizations and potential applications of
COFs and related materials The growth in this area is relatively slow and thus promotions are
needed in order to achieve its potential
6 List and pictures of chemical compounds
alkyl-THB Alkyl-1245-tetrahydroxybenzene
BDBA 14-benzenediboronic acid
BPDA 44ʹ-biphenyldiboronic acid
BTBA 135-benzene triboronic acid
BTDADA 14-benzothiadiazole diboronic acid
BTPA 135-benzenetris(4-phenylboronic acid)
CA Cyanuric acid
DAB 14-diaminobenzene
DCB 14-dicyanobenzene
DCF 27-diisocyanate fluorine
DCN 26-dicyanonaphthalene
DCP 26-dicyanopyridine
DETH 25-diethoxyterephthalohydrazole
DMBPDC 33ʹ-dimethoxy-44ʹ-biphenylene diisocyanate
DPB Diphenyl butadyenediboronic acid
73
HP 1-hexyne propiolate
HHTP 23671011-hexahydroxytriphenylene
MP Methyl propiolate
N3-BDBA Azide-appended benzenediboronic acid
NDI Naphthalenediimide diboronic acid
NiPcTA Nickel-phthalocyanice tetrakis(acetonide)
OH-Pc-Ni 2391016172324-octahydroxyphthalocyaninato)nickel(II)
OH-Pc-Zn 2391016172324-octahydroxyphthalocyaninato)zinc
PA Piperazine
Pac 2-propenyl acetate
PcTA Phthalocyanine tetra(acetonide)
PdAc Palladium acetate
PDBA Pyrenediboronic acid
PPE Phenylbis(phenylethynyl) diboronic acid
PPP 3-phenyl-1-propyne propiolate
PyMP (3α13α2-dihydropyren-1-yl)methyl propionate
TA Terephthaldehyde
TAM tetra-(4-anilyl)methane
TAPP Tetra(p-amino-phneyl)porphyrin
TBB 135-tris(4-bromophenyl)benzene
TBPM tetra(4-dihydroxyboryl-phenyl)methane
TBPP Tetra(p-boronic acid-phenyl)porphyrin
TBPS tetra(4-dihydroxyboryl-phenyl)silane
TBST tert-butylsilane triol
74
TCM Tetrakis(4-cyanophenyl)methane
TDHB-ZnP Zinc(II) 5101520-tetrakis(4-(dihydroxyboryl)phenyl)porphyrin
TFB 135-triformylbenzene
TFPB 135-tris-(4-formyl-phenyl)-benzene
THAn 2345-Tetrahydroxy anthracene
THB 1245-tetrahydroxybenzene
THDMA Polyol 2367-tetrahydroxy-910-dimethyl-anthracene
TkBPM Tetrakis(4-bromophenyl)methane
TPTA Triphenylene tris(acetonide)
trunc-TBPM-R R-functionalized truncated TBPM
75
Figure 8 Reactants of Covalently-bound Organic Frameworks
76
Figure 9 (Figure 8 continued)
(1) Morris R E Wheatley P S Angewandte Chemie-International Edition 2008 47 4966 (2) Li J-R Kuppler R J Zhou H-C Chemical Society Reviews 2009 38 1477 (3) Corma A Chemical Reviews 1997 97 2373 (4) Seo J S Whang D Lee H Jun S I Oh J Jeon Y J Kim K Nature 2000 404 982 (5) Yaghi O M OKeeffe M Ockwig N W Chae H K Eddaoudi M Kim J Nature 2003 423 705
77
(6) OKeeffe M Peskov M A Ramsden S J Yaghi O M Accounts of Chemical Research 2008 41 1782 (7) Ockwig N W Delgado-Friedrichs O OKeeffe M Yaghi O M Accounts of Chemical Research 2005 38 176 (8) Li H Eddaoudi M OKeeffe M Yaghi O M Nature 1999 402 276 (9) Chen B L Eddaoudi M Hyde S T OKeeffe M Yaghi O M Science 2001 291 1021 (10) Eddaoudi M Moler D B Li H L Chen B L Reineke T M OKeeffe M Yaghi O M Accounts of Chemical Research 2001 34 319 (11) Eddaoudi M Kim J Rosi N Vodak D Wachter J OKeeffe M Yaghi O M Science 2002 295 469 (12) Chae H K Siberio-Perez D Y Kim J Go Y Eddaoudi M Matzger A J OKeeffe M Yaghi O M Nature 2004 427 523 (13) Furukawa H Kim J Ockwig N W OKeeffe M Yaghi O M Journal of the American Chemical Society 2008 130 11650 (14) Smaldone R A Forgan R S Furukawa H Gassensmith J J Slawin A M Z Yaghi O M Stoddart J F Angewandte Chemie-International Edition 2010 49 8630 (15) Eddaoudi M Kim J Wachter J B Chae H K OKeeffe M Yaghi O M Journal of the American Chemical Society 2001 123 4368 (16) Sudik A C Millward A R Ockwig N W Cote A P Kim J Yaghi O M Journal of the American Chemical Society 2005 127 7110 (17) Sudik A C Cote A P Wong-Foy A G OKeeffe M Yaghi O M Angewandte Chemie-International Edition 2006 45 2528 (18) Tranchemontagne D J L Ni Z OKeeffe M Yaghi O M Angewandte Chemie-International Edition 2008 47 5136 (19) Lu Z Knobler C B Furukawa H Wang B Liu G Yaghi O M Journal of the American Chemical Society 2009 131 12532 (20) Park K S Ni Z Cote A P Choi J Y Huang R Uribe-Romo F J Chae H K OKeeffe M Yaghi O M Proceedings of the National Academy of Sciences of the United States of America 2006 103 10186 (21) Hayashi H Cote A P Furukawa H OKeeffe M Yaghi O M Nature Materials 2007 6 501 (22) Banerjee R Furukawa H Britt D Knobler C OKeeffe M Yaghi O M Journal of the American Chemical Society 2009 131 3875 (23) Cote A P Benin A I Ockwig N W OKeeffe M Matzger A J Yaghi O M Science 2005 310 1166 (24) El-Kaderi H M Hunt J R Mendoza-Cortes J L Cote A P Taylor R E OKeeffe M Yaghi O M Science 2007 316 268 (25) Wan S Guo J Kim J Ihee H Jiang D Angewandte Chemie-International Edition 2008 47 8826 (26) Wan S Guo J Kim J Ihee H Jiang D L Angewandte Chemie-International Edition 2009 48 5439 (27) Cote A P El-Kaderi H M Furukawa H Hunt J R Yaghi O M Journal of the American Chemical Society 2007 129 12914 (28) Hunt J R Doonan C J LeVangie J D Cote A P Yaghi O M Journal of the American Chemical Society 2008 130 11872 (29) Uribe-Romo F J Hunt J R Furukawa H Klock C OKeeffe M Yaghi O M Journal of the American Chemical Society 2009 131 4570 (30) Klontzas E Tylianakis E Froudakis G E Journal of Physical Chemistry C 2008 112 9095 (31) Tylianakis E Klontzas E Froudakis G E Nanotechnology 2009 20 (32) Lukose B Kuc A Frenzel J Heine T Beilstein Journal of Nanotechnology 2010 1 60
78
(33) Tilford R W Gemmill W R zur Loye H C Lavigne J J Chemistry of Materials 2006 18 5296 (34) Spitler E L Dichtel W R Nature Chemistry 2010 2 672 (35) Spitler E L Giovino M R White S L Dichtel W R Chemical Science 2011 2 1588 (36) Campbell N L Clowes R Ritchie L K Cooper A I Chemistry of Materials 2009 21 204 (37) Ding X Guo J Feng X Honsho Y Guo J Seki S Maitarad P Saeki A Nagase S Jiang D Angewandte Chemie-International Edition 2011 50 1289 (38) Feng X A Chen L Dong Y P Jiang D L Chemical Communications 2011 47 1979 (39) Zwaneveld N A A Pawlak R Abel M Catalin D Gigmes D Bertin D Porte L Journal of the American Chemical Society 2008 130 6678 (40) Gutzler R Walch H Eder G Kloft S Heckl W M Lackinger M Chemical Communications 2009 4456 (41) Blunt M O Russell J C Champness N R Beton P H Chemical Communications 2010 46 7157 (42) Sassi M Oison V Debierre J-M Humbel S Chemphyschem 2009 10 2480 (43) Ourdjini O Pawlak R Abel M Clair S Chen L Bergeon N Sassi M Oison V Debierre J-M Coratger R Porte L Physical Review B 2011 84 (44) Colson J W Woll A R Mukherjee A Levendorf M P Spitler E L Shields V B Spencer M G Park J Dichtel W R Science 2011 332 228 (45) Berlanga I Ruiz-Gonzalez M L Gonzalez-Calbet J M Fierro J L G Mas-Balleste R Zamora F Small 2011 7 1207 (46) Wan S Gandara F Asano A Furukawa H Saeki A Dey S K Liao L Ambrogio M W Botros Y Y Duan X Seki S Stoddart J F Yaghi O M Chemistry of Materials 2011 23 4094 (47) Ding X Chen L Honsho Y Feng X Saenpawang O Guo J Saeki A Seki S Irle S Nagase S Parasuk V Jiang D Journal of the American Chemical Society 2011 133 14510 (48) Tilford R W Mugavero S J Pellechia P J Lavigne J J Advanced Materials 2008 20 2741 (49) Lanni L M Tilford R W Bharathy M Lavigne J J Journal of the American Chemical Society 2011 133 13975 (50) Li Y Yang R T Aiche Journal 2008 54 269 (51) Nagai A Guo Z Feng X Jin S Chen X Ding X Jiang D Nature Communications 2011 2 (52) Bunck D N Dichtel W R Angewandte Chemie-International Edition 2012 51 1885 (53) Ding S-Y Gao J Wang Q Zhang Y Song W-G Su C-Y Wang W Journal of the American Chemical Society 2011 133 19816 (54) Miyaura N Suzuki A Chemical Reviews 1995 95 2457 (55) Kalidindi S B Yusenko K Fischer R A Chemical Communications 2011 47 8506 (56) Kuhn P Antonietti M Thomas A Angewandte Chemie-International Edition 2008 47 3450 (57) Bojdys M J Jeromenok J Thomas A Antonietti M Advanced Materials 2010 22 2202 (58) Kuhn P Forget A Su D Thomas A Antonietti M Journal of the American Chemical Society 2008 130 13333 (59) Ren H Ben T Wang E Jing X Xue M Liu B Cui Y Qiu S Zhu G Chemical Communications 2010 46 291 (60) Zhang W Li C Yuan Y-P Qiu L-G Xie A-J Shen Y-H Zhu J-F Journal of Materials Chemistry 2010 20 6413 (61) Trewin A Cooper A I Angewandte Chemie-International Edition 2010 49 1533 (62) Mastalerz M Angewandte Chemie-International Edition 2008 47 445
79
(63) Chan-Thaw C E Villa A Katekomol P Su D Thomas A Prati L Nano Letters 2010 10 537 (64) Palkovits R Antonietti M Kuhn P Thomas A Schueth F Angewandte Chemie-International Edition 2009 48 6909 (65) Ben T Ren H Ma S Q Cao D P Lan J H Jing X F Wang W C Xu J Deng F Simmons J M Qiu S L Zhu G S Angewandte Chemie-International Edition 2009 48 9457 (66) Yamamoto T Bulletin of the Chemical Society of Japan 1999 72 621 (67) Zhou G Baumgarten M Muellen K Journal of the American Chemical Society 2007 129 12211 (68) Yuan Y Sun F Ren H Jing X Wang W Ma H Zhao H Zhu G Journal of Materials Chemistry 2011 21 13498 (69) Ren H Ben T Sun F Guo M Jing X Ma H Cai K Qiu S Zhu G Journal of Materials Chemistry 2011 21 10348 (70) Zhao H Jin Z Su H Jing X Sun F Zhu G Chemical Communications 2011 47 6389 (71) Mortera R Fiorilli S Garrone E Verne E Onida B Chemical Engineering Journal 2010 156 184 (72) Dogru M Sonnauer A Gavryushin A Knochel P Bein T Chemical Communications 2011 47 1707 (73) Spitler E L Koo B T Novotney J L Colson J W Uribe-Romo F J Gutierrez G D Clancy P Dichtel W R Journal of the American Chemical Society 2011 133 19416 (74) Zhang Y Tan M Li H Zheng Y Gao S Zhang H Ying J Y Chemical Communications 2011 47 7365 (75) Uribe-Romo F J Doonan C J Furukawa H Oisaki K Yaghi O M Journal of the American Chemical Society 2011 133 11478 (76) Ben T Pei C Zhang D Xu J Deng F Jing X Qiu S Energy amp Environmental Science 2011 4 3991 (77) Lukose B Kuc A Heine T Chemistry-a European Journal 2011 17 2388 (78) Zhou W Wu H Yildirim T Chemical Physics Letters 2010 499 103 (79) Amirjalayer S Snurr R Q Schmid R Journal of Physical Chemistry C 2012 116 4921 (80) Xu Q Zhong C Journal of Physical Chemistry C 2010 114 5035 (81) Lukose B Supronowicz B St Petkov P Frenzel J Kuc A B Seifert G Vayssilov G N Heine T Physica Status Solidi B-Basic Solid State Physics 2012 249 335 (82) Assfour B Seifert G Chemical Physics Letters 2010 489 86 (83) Zhao L Zhong C L Journal of Physical Chemistry C 2009 113 16860 (84) Schmid R Tafipolsky M Journal of the American Chemical Society 2008 130 12600 (85) Han S S Goddard W A III Journal of Physical Chemistry C 2007 111 15185 (86) Suh M P Park H J Prasad T K Lim D-W Chemical Reviews 2012 112 782 (87) Furukawa H Yaghi O M Journal of the American Chemical Society 2009 131 8875 (88) Wong-Foy A G Matzger A J Yaghi O M Journal of the American Chemical Society 2006 128 3494 (89) Mendoza-Cortes J L Han S S Furukawa H Yaghi O M Goddard III W A Journal of Physical Chemistry A 2010 114 10824 (90) Doonan C J Tranchemontagne D J Glover T G Hunt J R Yaghi O M Nature Chemistry 2010 2 235 (91) Getman R B Bae Y-S Wilmer C E Snurr R Q Chemical Reviews 2012 112 703 (92) Han S S Furukawa H Yaghi O M Goddard III W A Journal of the American Chemical Society 2008 130 11580 (93) Furukawa H Ko N Go Y B Aratani N Choi S B Choi E Yazaydin A O Snurr R Q OKeeffe M Kim J Yaghi O M Science 2010 329 424 (94) Garberoglio G Langmuir 2007 23 12154 (95) Assfour B Seifert G Microporous and Mesoporous Materials 2010 133 59
80
(96) Lan J Cao D Wang W Journal of Physical Chemistry C 2010 114 3108 (97) Yang Q Zhong C Langmuir 2009 25 2302 (98) Garberoglio G Vallauri R Microporous and Mesoporous Materials 2008 116 540 (99) Lan J H Cao D P Wang W C Ben T Zhu G S Journal of Physical Chemistry Letters 2010 1 978 (100) Furukawa H Miller M A Yaghi O M Journal of Materials Chemistry 2007 17 3197 (101) Babarao R Jiang J Energy amp Environmental Science 2008 1 139 (102) Choi Y J Choi J H Choi K M Kang J K Journal of Materials Chemistry 2011 21 1073 (103) Lan J Cao D Wang W Smit B Acs Nano 2010 4 4225 (104) Wang L Yang R T Energy amp Environmental Science 2008 1 268 (105) Krishna R van Baten J M Industrial amp Engineering Chemistry Research 2011 50 7083 (106) Keskin S Journal of Physical Chemistry C 2012 116 1772 (107) Liu Y Liu D Yang Q Zhong C Mi J Industrial amp Engineering Chemistry Research 2010 49 2902 (108) Keskin S Sholl D S Langmuir 2009 25 11786 (109) Klontzas E Tylianakis E Froudakis G E Nano Letters 2010 10 452 (110) Zhu Y Zhou J Hu J Liu H Hu Y Chinese Journal of Chemical Engineering 2011 19 709 (111) Mendoza-Cortes J L Pascal T A Goddard W A III Journal of Physical Chemistry A 2011 115 13852 (112) Lino M A Lino A A Mazzoni M S C Chemical Physics Letters 2007 449 171 (113) Krishnan A Dujardin E Ebbesen T W Yianilos P N Treacy M M J Physical Review B 1998 58 14013 (114) Kim D Jung D H Kim K-H Guk H Han S S Choi K Choi S-H Journal of Physical Chemistry C 2012 116 1479 (115) Kim D Jung D H Choi S-H Kim J Choi K Journal of the Korean Physical Society 2008 52 1255 (116) Farha O K Yazaydin A O Eryazici I Malliakas C D Hauser B G Kanatzidis M G Nguyen S T Snurr R Q Hupp J T Nature Chemistry 2010 2 944 (117) Wu M M Wang Q Sun Q Jena P Kawazoe Y Journal of Chemical Physics 2010 133 (118) Choi Y J Lee J W Choi J H Kang J K Applied Physics Letters 2008 92 (119) Mendoza-Cortes J L Han S S Goddard W A III Journal of Physical Chemistry A 2012 116 1621 (120) Cao D Lan J Wang W Smit B Angewandte Chemie-International Edition 2009 48 4730 (121) Lan J H Cao D P Wang W C Langmuir 2010 26 220 (122) Klontzas E Tylianakis E Froudakis G E Journal of Physical Chemistry C 2009 113 21253 (123) Sun Y Ben T Wang L Qiu S Sun H Journal of Physical Chemistry Letters 2010 1 2753 (124) Li F Zhao J Johansson B Sun L International Journal of Hydrogen Energy 2010 35 266 (125) Zou X Zhou G Duan W Choi K Ihm J Journal of Physical Chemistry C 2010 114 13402 (126) Babarao R Dai S Jiang D-e Langmuir 2011 27 3451
81
Appendix B
Structural properties of metal-organic frameworks within the density-functional based tight-binding method
Binit Lukose Barbara Supronowicz Petko St Petkov Johannes Frenzel Agnieszka Kuc
Gotthard Seifert Georgi N Vayssilov and Thomas Heine
Phys Status Solidi B 2012 249 335ndash342
DOI 101002pssb201100634
Abstract Density-functional based tight-binding (DFTB) is a powerful method to describe large
molecules and materials Metal-organic frameworks (MOFs) materials with interesting catalytic
properties and with very large surface areas have been developed and have become commercially
available Unit cells of MOFs typically include hundreds of atoms which make the application of
standard density-functional methods computationally very expensive sometimes even unfeasible
The aim of this paper is to prepare and to validate the self-consistent charge-DFTB (SCC-DFTB)
method for MOFs containing Cu Zn and Al metal centers The method has been validated against
full hybrid density-functional calculations for model clusters against gradient corrected density-
functional calculations for supercells and against experiment Moreover the modular concept of
MOF chemistry has been discussed on the basis of their electronic properties We concentrate on
MOFs comprising three common connector units copper paddlewheels (HKUST-1) zinc oxide Zn4O
tetrahedron (MOF-5 MOF-177 DUT-6 (MOF-205)) and aluminum oxide AlO4(OH)2 octahedron (MIL-
53) We show that SCC-DFTB predicts structural parameters with a very good accuracy (with less than
82
5 deviation even for adsorbed CO and H2O on HKUST-1) while adsorption energies differ by 12 kJ
mol1 or less for CO and water compared to DFT benchmark calculations
1 Introduction In reticular or modular chemistry molecular building blocks are stitched together to
form regular frameworks [1] With this concept it became possible to construct framework
compounds with interesting structural and chemical composition most notably metal-organic
frameworks (MOFs) [1ndash10] and covalent organic frameworks (COFs) [11ndash17] The interest in MOFs
and COFs is not limited to chemistry these crystalline materials are also interesting for applications
in information storage [18] sieving of quantum liquids [19] hydrogen storage [20] and for fuel cell
membranes [21ndash23]
Covalent organic framework and MOF frameworks are composed by combining two types of building
blocks the so-called connectors typically coordinating in four to eight sites and linkers which have
typically 2 sometimes also three or four connecting sites Fig 1 illustrates a simplified representation
of the topology of connectors and linkers by using secondary building units (SBUs) (see Fig 1)
Figure 1 The connector and the linker of the frameworks CuBTC (HKUST-1) MOF-177 MOF-5 DUT-6 (MOF-205) and MIL-53 and their respective secondary building units (SBUs) The arrows indicate the connection points of the fragments Red oxygen green carbon white hydrogen yellow copperzinc and blue aluminum
Linkers are organic molecules with carboxylic acid groups at their connection sites which form
bonds to the connectors (typically in a solvothermal condensation reaction) They can carry
functional groups which can make them interesting for applications in catalysis [24] Connectors are
83
either metal organic units as for example the well-known Zn4O(CO2)6 unit of MOF-5 [8] the
Cu2(HCOO)4 paddle wheel unit of HKUST-1 [10] or covalent units incorporating boron oxide linking
units for COFs [11] As the building blocks of MOFs and COFs comprise typically some ten atoms unit
cells quickly get very large In particular if adsorption or dynamic processes in MOFs and COFs are of
interest (super)cells of some 1000 atoms need to be processed While standard organic force fields
show a reasonable performance for COFs [25] the creation of reliable force fields is not
straightforward for MOFs as transferable parameterization of the transition metal sites is an issue
even though progress has been achieved for selected materials [26 27] The difficulty to describe
transition metals in particular if they are catalytically active as in HKUST-1 limits the applicability of
molecular mechanics (MM) even for QMMM hybrid methods [28]
On the other hand the density-functional based tight-binding (DFTB) method with its self-consistent
charge (SCC) extension to improve performance for polar systems is a computationally feasible
alternative This non-orthogonal tight-binding approximation to density-functional theory (DFT)
which has been developed by Seifert Frauenheim and Elstner and their groups [29ndash32] (for a recent
review see Ref [30]) has been successfully applied to a large-scale systems such as biological
molecules [33ndash36] supramolecular systems [37 38] surfaces [39 40] liquids and alloys [41 42] and
solids [43] Being a quantum-mechanical method and hence allowing the description of breaking and
formation of chemical bonds the method showed outstanding performance in the description of
processes such as the mechanical manipulation of nanomaterials [44ndash46] It is remarkable that the
method performs well for systems containing heavier elements such as transition metals as this
domain cannot be covered so far with acceptable accuracy by traditional semi-empirical methods [47
48] DFTB covers today a large part of the elements of the periodic table and parameters and a
computer code are available from the DFTBorg website
Recently we have validated DFTB for its application for COFs [16 17] and have shown that structural
properties and formation energies of COFs are well described within DFTB Kuc et al [49] have
validated DFTB for substituted MOF-5 frameworks where connectors are always the Zn4O(CO2)6 unit
which has been combined with a large variety of organic linkers In this work we have revised the
DFTB parameters developed for materials science applications and validated them for HKUST-1 and
being far more challenging for the interaction of its catalytically active Cu sites with carbon
monoxide (CO) and water The Cu2(HCOO)4 paddle wheel units are electronically very intriguing on a
first note the electronic ground state of the isolated paddle wheel is an antiferromagnetic singlet
state which cannot be described by one Slater determinant and which is consequently not accessible
for KohnndashSham DFT However the energetically very close triplet state correctly describes structure
and electronic density of the system and also adsorption properties agree well with experiment [32
84
50 51] We therefore use DFT in the B3LYP hybrid functional representation as benchmark for DFTB
validations for HKUST-1 added by DFT-GGA calculations for periodic unit cellsWewill show that the
general transferability of the DFTB method will allow investigating structural electronic and in
particular dynamic properties
2 Computational details All calculations of the finite model and periodic crystal structures of MOFs
were carried out using the dispersion-corrected self-consistent density functional based tight-binding
(DC-SCC-DFTB in short DFTB) method [30 52ndash54] as implemented in the deMonNano code [55] Two
sets of parameters have been used the standard SCC-DFTB parameter set developed by Elstner et al
[53] for Zn-containing MOFs for Cu- and Al-containing materials we have extended our materials
science parameter set which has been developed originally for zeolite materials to include Cu For
this element we have used the standard procedure of parameter generation we have used the
minimal atomic valence basis for all atoms including polarization functions when needed Electrons
below the valence states were treated within the frozen-core approximation The matrix elements
were calculated using the local density approximation (LDA) while the short-range repulsive pair-
potential was fitted to the results from DFT generalized gradient approximation (GGA) calculations
For more details on DFTB parameter generation see Ref [30] Periodic boundary conditions were
used to represent frameworks of the crystalline solid state The conjugate-gradient scheme was
chosen for the geometry optimization An atomic force tolerance of 3x10-4 eV Aring-1 was applied
The reference DFT calculations of the equilibrium geometry of the investigated isolated MOF models
were performed employing the Becke three-parameter hybrid method combined with a LYP
correlation functional (B3LYP) [56 57] The basis set associated with the HayndashWadt [58] relativistic
effective core potentials proposed by Roy et al [59] was supplemented with polarization ffunctions
[60] and was employed for description of the electronic structure of Cu atoms The Pople 6-311G(d)
basis sets were applied for the H C and O atoms The calculations were performed with the
Gaussian09 program suite [61] For optimization of the periodic structure of CuBTC at DFT level the
electronic structure code Quickstep [62] which is a part of the CP2K package [63] was used
Quickstep is an implementation of the Gaussian plane wave method [64] which is based on the
KohnndashSham formulation of DFT
We have used the generalized gradient functional PBE [65] and the GoedeckerndashTeterndashHutter
pseudopotentials [64 66] in conjunction with double-z basis sets with polarization functions DZVP-
MOLOPT-SR-GTH basis sets obtained as described in Ref [67] but using two less diffuse primitives
Due to the large unit cell sampling of Brillouin zone was limited only to the G-point Convergence
85
criterion for the maximum force component was set to 45 x 10-3 Hartree Bohr1 The plane wave
basis with cutoff energy of 400 Ry was used throughout the simulations
The initial crystal structures were taken from experiment (powder X-ray diffraction (PXRD) data) The
cell parameters and the atomic positions were fully optimized using conjugate-gradient method at
the DFTB level For the reference DFT calculations only the atomic positions of experimental crystal
structures were minimized The cluster models were cut from the optimized structures and saturated
with hydrogen atoms (see Fig 2 for exemplary systems of Cu-BTC)
3 Results and discussion
31 Cu-BTC We have optimized the crystal structure of Cu-BTC (HKUST-1) [10] using cluster and the
periodic models The structural properties were compared to DFT results (see Table 1) The
geometries were obtained for the activated form (open metal sites) and in the presence of axial
water ligands (see Table 2) as the synthesized crystals often have H2O molecules attached to the
open metal sites of Cu (Fig 2b) We have also studied changes of the cluster model geometry in the
presence of CO (Fig 2c) and the results are also shown in Table 2 We have obtained good agreement
with experimental data as well as with DFT results
Figure 2 (a) Periodic and cluster models of Cu-BTC MOF as used in the computer simulations (b) cluster model with adsorbed H2O and (c) CO molecules
We have also simulated the XRD patterns of Cu-BTC which show very good structural agreement for
peak positions between the experimental and calculated structures The XRD pattern of DFT
optimized structure is nearly a copy of that of the experimental geometry However for DFTB
optimized patterns the relative intensities of some of the peaks notably that of the peaks at 2θ =
138 and 2θ = 158 are different from the experimental XRDs (see Fig 3) The differences in bond
angles between simulation and experiment may affect the sharpness of the signals and hence the
86
intensity To support this argument we have calculated the radial pair distribution function (g(r))
which details the probabilities of all atomndashatom distances in a given cell volume Figure S3 in the
Supporting Information shows g(r) of CundashCu CundashC CundashO CndashC and CndashO atom pairs of DFTB
optimized DFT optimized and experimentally modeled Cu-BTC unit cells The peaks of DFT as well as
DFTB optimized geometries are much broadened whereas the experimentally modeled geometry
has single peaks at the most predicted atomndashatom distances However it is to be noted that DFTB
optimized geometries give slightly shifted peaks compared to those of the DFT optimized geometry
They are broadened around the experimental values The distances between Cu and C atoms which
are not direct neighbors have the largest deviations from the experiment what indicates
shortcomings of the bond angles
Table 1 Selected bond lengths (Aring ) bond angles (⁰) and dihedral angles (⁰) of Cu-BTC optimized at DFTB and DFT level
Bond Type Cluster Model Periodic Model Exp
Cu-Cu 250 (257) 250 (250) 263 Cu-O 205 (198) 202 (198) 195 O-C 134 (133) 133-138 (128) 125
OCuO 836-971 (898) 892-907 (873-937)
891 896
Cu-O-O-Cu plusmn56 ˗ plusmn62 (plusmn02) plusmn04 (plusmn47 ˗ plusmn131) 0
O-C-C-C plusmn38 - plusmn50 (0) plusmn03 - plusmn05 (plusmn06 - plusmn105) 063
Cell paramet a=b=c=27283 (26343)
α=β=γ=90 (90) a=b=c=26343
α=β=γ=90
In detail the bond lengths and bond angles do not change significantly when going from the cluster
to the periodic model confirming the reticular chemistry approach The only exception is the OndashCundash
O bond angle that differs by 4ndash78 between the two systems at both levels of theory
87
Figure 3 Simulated XRD patterns of Cu-BTC MOF with the experimental unit cell parameters and optimized at the DFTB and DFT level of theory
The bond length between Cu atoms is slightly underestimated comparing with experimental (by
maximum 5) and DFT (maximum by 3) results while all other bond lengths are somewhat larger
at DFTB
All bond lengths stay unchanged or become longer in the presence of water molecules The most
striking example is the CundashCu bond where the change reaches 012ndash017 Aring compared with the
structure with activated Cu sites Also bond angles increase by up to 28 when the H2O is present
The CundashO bond length with the oxygen atom from the carboxylate groups is shorter than that with
the oxygen atoms from the water molecules indicating that the latter is only weakly bound to the
copper ions The OndashC distances (~133 Aring) correspond to values between that for the typical single
(142 Aring ) and double (122 Aring) oxygenndashcarbon bond The calculated CringndashCcarboxyl bond lengths of
146A deg indicate that these are typical single sp2ndashsp2 CndashC bonds while experimental findings give a
slightly longer value (15 Aring) for this MOF When comparing dihedral angles CundashOndashOndashCu and OndashCndashCndashC
of the clusters fairly large deviations between DFTB and DFT calculations and experiment are visible
due to the softer potential energy surface associated with these geometrical parameters In periodic
models however the agreement of DFT and DFTB with experiment and with each other is
significantly improved
The unit cell parameters with and without water molecules obtained at the DFTB level overestimate
the experimental data by less than 4 which gives a fairly good agreement if we take into account
high porosity of the material These lattice parameters increase only slightly from 27283 to 27323 Aring
in the presence of water
We have calculated the binding energies of H2O molecules attached to the Cu-metal sites within the
cluster model At the DFTB level Ebind of 40 kJ mol-1 was obtained in a good agreement with the DFT
results (52 kJ mol-1) as well as results from the literature (47 kJ mol-1 [50]) We have also calculated
88
the binding energy of CO which was twice smaller than that of H2O (165 and 28 kJ mol-1 for DFTB
and DFT respectively) suggesting much stronger binding of O atom (from H2O) than C atom (from
CO) While the bond lengths in the MOF structure do not change in the presence of either H2O or CO
the differences in the binding energy come from much longer bond distances (by around 07 Aring) for
CundashC than for CundashO in the presence of CO and water molecules respectively
Furthermore we have studied the electronic properties of periodic and cluster model Cu-BTC by
means of partial density-of-states (PDOS)We have compared especially the Cu-PDOS for s- p- and d-
orbitals (see Fig 4) The results show that the electronic structure stays unchanged when going from
the cluster models to the fully periodic crystal of Cu-BTC Since the copper atoms are of Cu(II) the d-
orbitals are just partially occupied what results in the metallic states at the Fermi level This is a very
interesting result as other ZnndashO-based MOFs are either semiconductors or insulators [49]
Table 2 Selected bond lengths (Aring) bond angles (˚) and dihedral angles (˚) of Cu-BTC optimized at DFTB and DFT levels with water molecules in the axial positions Cluster models are calculated using water and carbon monoxide molecules The adsorption energies Eads (kJ mol-1) of the guest molecules are given as well The DFT data are given in parenthesis
Bond Type Cluster Model +
H2O Periodic
Model+ H2O Cluster Model +
CO
Cu-Cu 267 (266) 262 (260) 250 (260)
Cu-O 205 (197-206) 210 (196-200) 206 (199)
O-C 134 (127) 133 (128) 134 (127)
OCuO 843-955 (889-905)
871-921 (842-930) 842-967 (896)
Cu-O-O-Cu plusmn59 - plusmn76 (plusmn06 ˗ plusmn10)
plusmn02 ˗ plusmn18 (plusmn40 ˗ plusmn136)
plusmn51 - plusmn58 (plusmn70)
O-C-C-C plusmn39 - plusmn53 (plusmn05 - plusmn11)
plusmn03 - plusmn05 (plusmn06 - plusmn105)
plusmn35 - plusmn43 (plusmn12)
Cu-O(H2O) Cu-C(CO) 237 (219) 244 (233-
255) 307 (239)
Eads -4045 (-5200) -1648
(-2800)
32 MOF-5 -177 DUT-6 (MOF-205) and MIL- 53 We have also studied the structural properties
of MOF structures with large surface areas using the SCC-DFTB method Very good agreement with
the experimental data shows that this method is applicable for MOFs of large structural diversity as
well as for coordination polymers based on the MOF-5 framework which has been reported earlier
[49] Tables 3 and 4 show selected bond lengths and bond angles of MOF-5 [68] MOF-177 [69] DUT-
6 (MOF-205) [70 71] and MIL-53 [72] respectively
89
MOFs-5 -177 and DUT-6 (MOF-205) are built of the same connector which is Zn4O(CO2)6
octahedron The difference is in the organic linker 14-benzenedicarboxylate (BDC) 135-
benzenetribenzoates (BTB) and BTB together with 26-naphtalenedicarboxylate (NDC) for MOF-5 -
177 andDUT-6 (MOF-205) respectively (see Fig 5)
Figure 4 DFTB calculations of PDOS of (top left) Cu in Cu-BTC and Zn in MOF-5 (top right) MOF-177 (bottom left) and DUT-6 (MOF-205) (bottom right) In each plot s-orbitals (top) p-orbitals (middle) and d-orbitals (bottom) are shown as computed from periodic (black) and cluster (red) models Cluster models are given in Fig S4
All three MOFs have different topologies due to the organic linkers where the number of
connections is varied or where two different linker types are present MOF-5 is the most simple and
it is the prototype in the isoreticular series (IRMOF-1) [68] it is a simple cubic topology with
threedimensional pores of the same size and the linkers have only two connection points In the
case of MOF-177 the linker is represented by a triangularSBU that means three connection points
are present This results in the topology with mixed (36)-connectivity [69] DUT-6 (MOF-205) has a
much more complicated topology due to two types of linkers The first one (NDC) has just two
90
connection points while the second is the same as in MOF-177 with three connection points One
ZnndashO-based connector is connected to two NDC and four BTB linkers The topology is very interesting
all rings of the underlying nets are fivefold and it forms a face-transitive tiling of dodectahedra and
tetrahedra with a ratio of 13 [73]
Figure 5 The crystal structures in the unit cell representations of (a) MOF-5 (b) MOF-177 (c) DUT-6 (MOF-205) and (d) MIL-53 red oxygen gray carbon white hydrogen and blue metal atom (Zn Al)
MIL-53 is a MOF structure with one-dimensional pores The framework is built up of corner-sharing
AlndashO-based octahedral clusters interconnected with BDC linkers The linkers have again two
connection points MIL-53 shows reversible structural changes dependent on the guest molecules
[74] It undergoes the so-called breathing mode depending on the temperature and the amount of
adsorbed molecules
In this case also the bond lengths and bond angles are slightly overestimated comparing with the
experimental structures but the error does not exceed 3
91
Table 3 Selected bond lengths (Aring) and bond angles (˚) of MOF-5 (424 atoms in unit cell) MOF-177 (808 atoms in unit cell) and DUT-6 (MOF-205) (546 atoms in unit cell) optimized at DFTB level The available experimental data is given in parenthesis [70-73+ Orsquo denotes the central O atom in the Zn-O-octahedron
Bond Type MOF-5 MOF-177 DUT-6
(MOF-205)
Zn-Zn 330 (317) 322-336 (306-330)
325-331 (318)
Zn-Orsquo 202 (194) 202 (193) 202 (194) Zn-O 204 (192) 202-206
(190-199) 202 205 (193)
O-C 128 (130) 128 (131) 128 (125) ZnOrsquoZn 1095 (1095) 1056-1124
(1055 1092) 107-1118 (1084 1100)
OZnO 1083 1108 (1061)
1048 1145 (981-1281)
1046-1112 (1062 1085)
Zn-O-O-Zn 0 (0 plusmn17 plusmn20) plusmn122 - plusmn347 (plusmn176 - plusmn374)
05 - plusmn62 (0 plusmn29)
O-C-C-C 00 (0 - plusmn24) (plusmn 35 - plusmn 336) (plusmn65 - plusmn374)
plusmn04 plusmn22 (0 plusmn174)
Cell paramet a=b=c=26472 (25832) α=β=γ=90 (90)
a=b=37872 (37072) c=3068 (30033) α=β=90 (90) γ=120 (120)
a=b=c=31013 (30353) α=β=γ=90 (90)
We have calculated PDOS for MOF-5 MOF-177 and DUT-6 (MOF-205) (see Fig 4) Their band gaps
calculated to be 40 35 and 31 eV respectively indicate that they are either semiconductors or
insulators We have calculated it for both periodic crystal structure and a cluster containing a metal-
oxide connector and all its carboxylate linkers
Table 4 Selected bond lengths (Aring) and bond angles (˚) of MIL-53 (152 atoms in unit cell) optimized at DFTB level
Bond Type DFTB Exp
Al-Al 341 331 Al-O 189 183-191 O-C 133 129 130 O-Al-O 884 912 886 898 Al-O-Al 1289 1249 Al-O-O-Al 0 0 O-C-C-O 06 03 Cell paramet a=1246
b=1732 c=1365 α=β=γ=90
a=1218 b=1713 c=1326 α=β=γ=90
4 Mechanical properties Due to the low-mass density the elastic constants of porous materials
are a very sensitive indicator of their mechanical stability For crystal structures like MOFs we have
92
studied these by means of bulk modulus (B) B can be calculated as a second derivative of energy
with respect to the volume of the crystal (here unit cell)
The result shows that CuBTC has bulk modulus of 3466 GPa what is in close agreement with
B=3517 GPa obtained using force-field calculations [73 74] and experiment [75] Bulk moduli for the
series of MOFs resulted in 1534 1010 and 1073 GPa for MOF-5 -177 and DUT-6 (MOF-205)
respectively For MOF-5 B=1537 GPa was reported from DFTGGA calculations using plane-waves
[76 77] The results show that larger linkers give mechanically less stable structures what might be
an issue for porous structures with larger voids For MIL-53 we have obtained the largest bulk
modulus of 5369 GPa keeping the angles of the pore fixed
5 Conclusions We have validated the DFTB method with SCC and London dispersion corrections for
various types of MOFs The method gives excellent geometrical parameters compared to experiment
and for small model systems also in comparison with DFT calculations Importantly this statement
holds not only for catalytically inactive MOFs based on the Zn4O(CO2)6 octahedron and organic linkers
which are important for gas adsorption and separation applications but also for catalytically active
HKUST-1 (CuBTC) This is remarkable as the Cu atoms are in the Cu2+ state in this framework DFTB
parameters have been generated and validated for Cu and the electronic structure contains one
unpaired electron per Cu atom in the unit cell which makes the electronic description technically
difficult but manageable within the SCC-DFTB method SCC-DFTB performs well for the frameworks
themselves as well as for adsorbed CO and water molecules
Partial density-of-states calculations for the transition metal centers reveal that the concept of
reticular chemistry ndash individual building units keep their electronic properties when being integrated
to the framework ndash is also valid for the MOFs studied in this work in agreement with a previous
study of COFs [16] The electronic properties computed using the cluster models and the periodic
structure contains the same features and hence cluster models are good models to study the
catalytic and adsorption properties of these materials This is in particular useful if local quantum
chemical high-level corrections shall be applied that require to use finite structures
We finally conclude that we have now a high-performing quantum method available to study various
classes of MOFs of unit cells up to 10000 atoms including structural characteristics the formation
and breaking of chemical and coordination bonds for the simulation of diffusion of adsorbate
molecules or lattice defects as well as electronic properties The parameters can be downloaded
from the DFTBorg website
93
References
[1] E A Tomic J Appl Polym Sci 9 3745 (1965)
2+ M Eddaoudi D B Moler H L Li B L Chen T M Reineke M OrsquoKeeffe and O M Yaghi Acc Chem Res
34 319 (2001)
[3] M Eddaoudi H L Li T Reineke M Fehr D Kelley T L Groy and O M Yaghi Top Catal 9 105 (1999)
[4] B F Hoskins and R Robson J Am Chem Soc 111 5962 (1989)
[5] K Schlichte T Kratzke and S Kaskel Microporous Mesoporous Mater 73 81 (2004) [6] J L C Rowsell A
R Millward K S Park and O M Yaghi J Am Chem Soc 126 5666 (2004)
7+ O M Yaghi M OrsquoKeeffe N W Ockwig H K Chae M Eddaoudi and J Kim Nature 423 705 (2003)
[8] M Eddaoudi J Kim N Rosi D Vodak J Wachter M OrsquoKeeffe and O M Yaghi Science 295 469 (2002)
9+ M OrsquoKeeffe M Eddaoudi H L Li T Reineke and O M Yaghi J Solid State Chem 152 3 (2000)
[10] S S Y Chui SM F Lo J P H Charmant A G Orpen and I D Williams Science 283 1148 (1999)
11+ A P Cote A I Benin N W Ockwig M OrsquoKeeffe A J Matzger and O M Yaghi Science 310 1166 (2005)
[12] A P Cote H M El-Kaderi H Furukawa J R Hunt and O M Yaghi J Am Chem Soc 129 12914 (2007)
[13] H M El-Kaderi J R Hunt J L Mendoza-Cortes A P Cote R E Taylor M OrsquoKeeffe and O M Yaghi
Science 316 268 (2007)
[14] S Wan J Guo J Kim H Ihlee and D L Jiang Angew Chem 120 8958 (2008)
[15] S Wan J Guo J Kim H Ihlee and D L Jiang Angew Chem 121 5547 (2009)
[16] B Lukose A Kuc J Frenzel and T Heine Beilstein J Nanotechnol 1 60 (2010)
[17] B Lukose A Kuc and T Heine Chem Eur J 17 2388 (2011)
[18] D S Marlin D G Cabrera D A Leigh and A M Z Slawin Angew Chem Int Ed 45 77 (2006)
[19] I Krkljus and M Hirscher Microporous Mesoporous Mater 142 725 (2011)
[20] M Hirscher Handbook of Hydrogen Storage (Wiley-VCH Weinheim 2010)
[21] H Kitagawa Nature Chem 1 689 (2009)
[22] M Sadakiyo T Yamada and H Kitagawa J Am Chem Soc 131 9906 (2009)
[23] T Yamada M Sadakiyo and H Kitagawa J Am Chem Soc 131 3144 (2009)
94
[24] K S Jeong Y B Go S M Shin S J Lee J Kim O M Yaghi and N Jeong Chem Sci 2 877 (2011)
[25] R Schmid and M Tafipolsky J Am Chem Soc 130 12600 (2008)
[26] M Tafipolsky S Amirjalayer and R Schmid J Comput Chem 28 1169 (2007)
[27] M Tafipolsky S Amirjalayer and R Schmid J Phys Chem C 114 14402 (2010)
[28] A Warshel and M Levitt J Mol Biol 103 227 (1976)
[29] G Seifert D Porezag and T Frauenheim Int J Quantum Chem 58 185 (1996)
[30] A F Oliveira G Seifert T Heine and H A Duarte J Braz Chem Soc 20 1193 (2009)
[31] D Porezag T Frauenheim T Koumlhler G Seifert and R Kaschner Phys Rev B 51 12947 (1995)
[32] T Frauenheim G Seifert M Elstner Z Hajnal G Jungnickel D Porezag S Suhai and R Scholz Phys
Status Solidi B 217 41 (2000)
[33] M Elstner Theor Chem Acc 116 316 (2006)
Supporting Information
Figure S4 Cluster models as used for the P-DOS calculations in Fig 4 MOF-5 MOF-177 and DUT-6 (MOF-205) (from left to right)
95
Figure S3 Radial pair distribution function (g(r)) of Cu-Cu Cu-C Cu-O C-C and C-O atom-pairs in a Cu-BTC MOF unit cell
96
Appendix C
The Structure of Layered Covalent-Organic Frameworks
Binit Lukose Agnieszka Kuc and Thomas Heine
Chem Eur J 2011 17 2388 ndash 2392
DOI 101002chem201001290
Abstract Covalent-Organic Frameworks (COFs) are a new family of 2D and 3D highly porous and
crystalline materials built of light elements such as boron oxygen and carbon For all 2D COFs an AA
stacking arrangement has been reported on the basis of experimental powder XRD patterns with the
exception of COF-1 (AB stacking) In this work we show that the stacking of 2D COFs is different as
originally suggested COF-1 COF-5 COF-6 and COF-8 are considerably more stable if their stacking
arrangement is either serrated or inclined and layers are shifted with respect to each other by ~14 Aring
compared with perfect AA stacking These structures are in agreement with to date experimental
data including the XRD patterns and lead to a larger surface area and stronger polarisation of the
pore surface
Introduction In 2005 Cocircteacute and co-workers introduced a new class of porous crystalline materials
Covalent Organic Frameworks (COFs)[1] where organic linker molecules are stitched together by
connectors covalent entities including boron and oxygen atoms to a regular framework These
materials have the genuine advantage that all framework bonds represent strong covalent
interactions and that they are composed of light-weight elements only which lead to a very low
mass density[2] These materials can be synthesized solvothermally in a condensation reaction and
97
are composed of inexpensive and non-toxic building blocks so their large-scale industrial production
appears to be possible
Figure 1 Calculated and experimental XRD patterns of all studied COFs (COF-1 top left COF-5 top right COF-6 bottom left COF-8 bottom right)
To date a number of different COF structures have been reported[1ndash3] From a topological
viewpoint we distinguish two- and three-dimensional COFs In two-dimensional (2D) COFs the
covalently bound framework is restricted to 2D layers The crystal is then similar as in graphite or
hexagonal boron nitride composed by a stack of layers which are not connected by covalent bonds
but held together primarily by London dispersion interactions
98
The COF structures have been analyzed by powder X-ray diffraction (PXRD) experiments The
topology of the layers is determined by the structure of the connector and linker molecules and
typically a hexagonal pattern is formed due to the 2m (D3h) symmetry of the connector moieties
The individual layers are then stacked and form a regular crystal lattice With one exception (COF-
1)[1] for all 2D-COFs AA (eclipsed P6mmm) interlayer stacking has been reported[3andashd f g] This
geometrical arrangement maximizes the proximity of the molecular entities and results in straight
channels orthogonal to the COF layers which are known from the literature[1 3a]
The connectors of 2D-COFs include oxygen and boron atoms which cause an appreciable polarization
The AA stacking arrangement maximizes the attractive London dispersion interaction between the
layers which is the dominating term of the stacking energy At the same time AA stacking always
results in a repulsive Coulomb force between the layers due to the polarized connectors It should be
noted that the eclipsed hexagonal boron nitride (h-BN)[4] has boron atoms in one layer serving as
nearest neighbours to nitrogen atoms in adjacent layers (AB stacking) The Coulomb interaction has
ruled out possible interlayer eclipse between atoms of alike charges In this article we aim at
studying the layer arrangements of solvent-free COFs based on the interlayer interactions and the
minimum variance Various lattice types have been considered all significantly more stable than the
reported AA stacked geometry In all 2D COFs we have studied the Coulomb interaction between the
layers leads to a modification of the stacking and shifts the layers by about one interatomic distance
(~14 Aring) with respect to each other (see Figure 1)
Breaking of the highly symmetric layer arrangement has been observed for COF-1 and COF-10 after
removal of guest molecules from pores leading to a turbostratic repacking of pore channels[1 5]
The authors have confirmed the disorder either by the changes in the PXRD spectrum taken before
and after adsorptionndashdesorption cycle[1] or by the changes in the adsorption behavior[5] The
disruption of layer arrangement is attributed to the force exerted by the guest molecules Formation
of poly disperse pores has also been verified[5] The lattice types that we discuss in this article on
the other hand are neither the result of the pressure from any external molecule in the pore nor
having more than one type of pores They are the resultant of minimum variance guided by Coulomb
and London dispersion interactions For the COF models under investigation perfect crystallinity has
been considered
Methods We have studied the experimentally known structures of COF-1 COF-5 COF-6 and COF-8
Structure calculations were carried out using the Dispersion-Corrected Self-Conistent-Charge
Density-Functional-Based Tight-Binding method (SCC-DFTB)[6] DFTB is based on a second-order
expansion of the KohnndashSham total energy in DFT with respect to charge density fluctuations This
does not require large amounts of empirical parameters however maintains all qualities of DFT The
99
accuracy is very much improved by the SCC extension The DFTB code (deMonNano[6a]) has
dispersion correction[6d] implemented to account for weak interactions and was used to obtain the
layered bulk structure of COFs and their formation energies The performance for interlayer
interactions has been tested previously for graphite[6d] All structures correspond to full geometry
optimizations (structure and unit cell) XRD patterns have been simulated using the Mercury
software[7] To allow best comparison with experiment for PXRD simulations we used the calculated
geometry of the layer and of the relative shifts between the layers but experimental interlayer
distances The electrostatic potential has been calculated using Crystal 09[8] at the DFTVWN level
with 6-31G basis set
Results and Discussion
In order to see the favorite stacking arrangement of the layers we have shifted every second layer in
two directions along zigzag and armchair vertices of the hexagonal pores (see Figure 2) The initial
stacking is AA (P6mmm) and the final zigzag shift gives AB stacking (P63mmc) Considering that the
attractive interlayer dispersion forces and repulsive interlayer orbital interactions are different we
have optimized the interlayer separation for each stacking Figure 2 show their total energies
calculated per formula unit that is per established bond between linkers and connectors with
reference to AA stacking for COF-5 and COF-1 In each direction the energy minimum is found close
to the AA stacking position shifted only by about one aromatic C-C bond length (~14 Aring) so that
either connector or linker parts become staggered with those in the adjacent layers leading to a
stacking similar to that of graphite for either connector (armchair shift) or linker (zigzag shift) For
COF-5 the staggering at the connector or linker depends whether the shift is armchair or zigzag
respectively while for COF-1 the zigzag shift alone can give staggering in both the aromatic and
boron-oxygen rings
The low-energy minima in the two directions are labeled following the common nomenclature as
zigzag and armchair respectively Both shifts result in a serrated stacking order (orthorhombic
Cmcm) which shows a significantly more attractive interlayer Coulomb interaction than AA stacking
(see Table 1) while most of the London dispersion attraction is maintained and consequently
stabilizes the material Still this configuration can be improved if we consider inclined stacking
(Figure 1) that is the lattice vector pointing out of the 2D plane is not any longer rectangular
reducing the lattice symmetry from Cmcm (orthorhombic) to C2m (monoclinic)
Besides we have calculated the monolayer formation energy (condensation energy Ecb) from the
total energies of the monolayer and of the individual building blocks and the stacking formation
energy from the total energies of the bulk structure and of the monolayer for four selected COFs The
100
COF building blocks are the same as those used for their synthesis[1 3a] (BDBA for COF-1 BDBA and
HHTP for COF-5 BTBA and HHTP for COF-6 BTPA and HHTP for COF-8) The final energies given per
formula unit that is per condensed bond are given in Table 1 The serrated and inclined stacking
structures are energetically more stable than AA and AB Interestingly within our computational
model zigzag and armchair shifting is energetically equivalent
Figure 2 Change of the total energy (per formula unit) when shifting COF-5 even layers in zigzag (top) and armchair (middle) directions and COF-1 in zigzag (bottom) direction The vector d shows the shift direction The low-energy minima in the two directions are labelled as zigzag and armchair respectively The equilibrium structures are shown as well
The formation energies of the individual COF structures are in agreement with full DFT calculations
We have calculated using ADF[9] the condensation energy of COF-1 and COF-5 using first-principles
DFT at the PBE[10]DZP[11] level to qualitatively support our results For simplicity we have used a
finite structure instead of the bulk crystal Their calculated Ecb energies are 77 and 15 kJmol-1
respectively for COF-1 and COF-5 These values support the endothermic nature of the condensation
101
reaction and are in excellent agreement with our DFTB results (91 and 21 kJmol-1 respectively see
Table 1)
The change of stacking should be visible in X-ray diffraction patterns because each space group has a
distinct set of symmetry imposed reflection conditions Powder XRD patterns from experiments are
available for all 2D COF structures[1 3andashd f g] We have simulated XRD patterns for AA AB serrated
Table 1 Calculated formation energies for the studied COF structures Ecb = condensation Esb = stacking energy EL = London dispersion energy Ee = electronic energy Energies are per condensed bond and are given in [kJ mol
-1+ lsquoarsquo and lsquozrsquo stand for armchair and zigzag respectively Note that the electronic energy Ee=Esb-EL
includes the electronic interlayer interaction and the formation energy of the monolayer and that the formation of the monolayers is endothermic
Structure Stacking Esb EL Ee
COF-5 AA -2968 -3060 092
AB -2548 -2618 070
serrated z -3051 -3073 022
serrated a -3052 -3073 021
inclined z -3297 -3045 -252
inclined a -3275 -3044 -231
Monolayer Ecb= 211
COF-1 AA -2683 -2739 056
AB -2186 -2131 -055
serrated z -2810 -2806 -004
inclined z -2784 -2788 004
Monolayer Ecb= 906
COF-6 AA -2881 -2963 082
AB -2127 -2146 019
serrated z -2978 -2996 018
serrated a -2978 -2995 017
inclined z -2946 -2975 029
inclined a -2954 -2974 021
Monolayer Ecb= 185
COF-8 AA -4488 -4617 129
102
AB -2477 -2506 029
serrated z -4614 -4646 032
serrated a -4615 -4647 032
inclined z -4578 -4612 035
inclined a -4561 -4591 030
Monolayer Ecb= 263
and inclined stacking forms for COF-5 COF-1 COF-6 and COF-8 (see Figure 1for their comparison
with experimental spectrum) For completeness we have studied XRD patterns of optimized COFs
using the experimentally determined[1 3a] interlayer separations this means we have kept the
layer geometry the same as the optimized structures and different stackings were obtained by
shifting adjacent layers accordingly
COF-1 was reported as having different powder spectra for samples before (as-synthesized) and after
removal of guest molecules with a particular mentioning about its layer shifting after removal We
have compared the two spectra with our simulated XRDs in order to see the stacking in the pure
form and how the stacking is changed at the presence of mesitylene guests Except that we have only
a vague comparison at angles (2θ) 11ndash15 the powder spectrum after guest removal is more similar
to the XRDs of serrated and inclined stacking structures A notable difference from AB is the absence
of high peaks at angles ~12 ~15 and ~19 This gives rise to the suggestion that COF-1 is not a
notable exception among the 2D COFs it follows the same topological trend as all other frameworks
of this class having one-dimensional (1D) pores as soon as the solvent is removed from the pores
This suggestion is supported by the experimental fact that the gas adsorption capacity of COF-1 is
only slightly lower than that of COF-6 for even as large gas molecules as methane and CO2[12] It is
not obvious how these molecules would enter an AB stacked COF-1 framework where the pores are
not connected by 1D channels (see Figure 1) Moreover our calculations suggest that the serrated
and inclined stackings are energetically favorable (see Table 1)
Indeed all the new stackings discussed in this article yield XRD patterns which are in agreement with
the currently available experimental data (see Figure 1) The inclined stackings have more peaks but
those are covered by the broad peaks in the experimental pattern The same is observed for the (002)
peaks of serrated stackings At present 2D COF synthesis methods are not yet capable to produce
crystals of sufficient quality to allow more detailed PXRD analysis in particular for the solvent-free
materials Microwave synthesis of COF-5 by rapid heating is reported[13] as much faster (~200 times)
compared with solvothermal methods however the structural details (XRD etc) remained
103
ambiguous We are confident that better crystals will be produced in future which will allow the
unambiguous determination of COF structures and can be compared to our simulations
Finally we want to emphasize that the suggested change of stacking is not only resulting in a
moderate change of geometry and XRD pattern The functional regions of COFs are their pores and
the pore geometry is significantly modified in our suggested structures compared to AA and AB
stackings First the inclined and serrated structures account for an increase of the surface area by 6
estimated for the interaction of the COFs with helium (3 for N2 5 for Ar 56 for Ne) Moreover
the shift between the layers exposes both boron and oxygen atoms to the pore surface leading to a
much stronger polarity than it can be expected for AA stacked COFs This difference is shown in
Figure 3 which plots the electrostatic potential of COF-5 in AA serrated and inclined stacking
geometries While the pore surface of AA stacked COF-5 shows a positively and negatively charged
stripes the other stacking arrangements show a much stronger alternation of charges indicating the
higher polarity of the surface The surface polarity can also be expressed in terms of Mulliken charges
of oxygen and boron atoms calculated at the DFT level which are COF-5 AA qB = 048 and qO = -048
COF-5 serrated qB = 047 and qO = -049 COF-5 inclined qB = 045 and qO = -048
Figure 3 Calculated electrostatic potential of COF-5 AA (left) serrated (middle) and inclined (right) stacking forms Blue - negative and red-positive values of the 005 isosurface
Conclusion We have studied 2D COF layer stackings energetically and found that energy minimum
structures have staggering of hexagonal rings at the connector or linker parts This can be achieved if
the bulk structure has either serrated or inclined order These newly proposed orders have their
simulated XRDs matching well with the available experimental powder spectrum Hence we claim
that all reported 2D COF geometries shall be re-examined carefully in experiment Due to the change
of the pore geometry and surface polarity the envisaged range of applications for 2D COFs might
change significantly We believe that these results are of utmost importance for the design of
functionalized COFs where functional groups are added to the pore surfaces
104
References
[1] A P Cocircteacute A I Benin N W Ockwig M OKeeffe A J Matzger O M Yaghi Science 2005 310 1166
[2] H M El-Kaderi J R Hunt J L Mendoza-Cortes A P Cote R E Taylor M OKeeffe O M Yaghi Science
2007 316 268
[3] a) A P Cocircteacute H M El-Kaderi H Furukawa J R Hunt O M Yaghi J Am Chem Soc 2007 129 12914 b) J
R Hunt C J Doonan J D LeVangie A P Cocircte O M Yaghi J Am Chem Soc 2008 130 11872 c) R W
Tilford W R Gemmill H C zur Loye J J Lavigne Chem Mater 2006 18 5296 d) R W Tilford S J Mugavero
P J Pellechia J J Lavigne Adv Mater 2008 20 2741 e) F J Uribe-Romo J R Hunt H Furukawa C Klock M
OKeeffe O M Yaghi J Am Chem Soc 2009 131 4570 f) S Wan J Guo J Kim H Ihee D L Jiang Angew
Chem 2008 120 8958 Angew Chem Int Ed 2008 47 8826 g) S Wan J Guo J Kim H Ihee D L Jiang
Angew Chem 2009 121 5547 Angew Chem Int Ed 2009 48 5439
[4] R T Paine C K Narula Chem Rev 1990 90 73
[5] C Doonan D Tranchemontagne T Glover J Hunt O Yaghi Nat Chem 2010 2 235
[6] a) T Heine M Rapacioli S Patchkovskii J Frenzel A M Koester P Calaminici S Escalante H A Duarte R
Flores G Geudtner A Goursot J U Reveles A Vela D R Salahub deMon deMonnano edn Mexico DF
Mexico and Bremen (Germany) 2009 b) A F Oliveira G Seifert T Heine H A Duarte J Braz Chem Soc
2009 20 1193 c) G Seifert D Porezag T Frauenheim Int J Quantum Chem 1996 58 185 d) L Zhechkov T
Heine S Patchkovskii G Seifert H A Duarte J Chem Theory Comput 2005 1 841
[7] a) httpwwwccdccamacukproductsmercury b) C F Macrae P R Edgington P McCabe E Pidcock
G P Shields R Taylor M Towler J Van de Streek J Appl Crystallogr 2006 39 453
[8] R Dovesi V R Saunders C Roetti R Orlando C M Zicovich- Wilson F Pascale B Civalleri K Doll N M
Harrison I J Bush P DrsquoArco M Llunell Crystal 102 ed
[9] a) httpwwwscmcom ADF200901 Amsterdam The Netherlands b) G te Velde F M Bickelhaupt S J
A van Gisbergen C Fonseca Guerra E J Baerends J G Snijders T Ziegler J Comput Chem 2001 22 931
[10] J P Perdew K Burke M Ernzerhof Phys Rev Lett 1996 77 3865
[11] E Van Lenthe E J Baerends J Comput Chem 2003 24 1142
[12] H Furukawa O M Yaghi J Am Chem Soc 2009 131 8875
[13] N L Campbell R Clowes L K Ritchie A I Cooper Chem Chem Mater 2009 21 204
105
Appendix D
On the reticular construction concept of covalent organic frameworks
Binit Lukose Agnieszka Kuc Johannes Frenzel and Thomas Heine
Beilstein J Nanotechnol 2010 1 60ndash70
DOI103762bjnano18
Abstract
The concept of reticular chemistry is investigated to explore the applicability of the formation of
Covalent Organic Frameworks (COFs) from their defined individual building blocks Thus we have
designed optimized and investigated a set of reported and hypothetical 2D COFs using Density
Functional Theory (DFT) and the related Density Functional based tight-binding (DFTB) method
Linear trigonal and hexagonal building blocks have been selected for designing hexagonal COF layers
High-symmetry AA and AB stackings are considered as well as low-symmetry serrated and inclined
stackings of the layers The latter ones are only slightly modified compared to the high-symmetry
forms but show higher energetic stability Experimental XRD patterns found in literature also
support stackings with highest formation energies All stacking forms vary in their interlayer
separations and band gaps however their electronic densities of states (DOS) are similar and not
significantly different from that of a monolayer The band gaps are found to be in the range of 17ndash
40 eV COFs built of building blocks with a greater number of aromatic rings have smaller band gaps
Introduction
In the past decade considerable research efforts have been expended on nanoporous materials due
to their excellent properties for many applications such as gas storage and sieving catalysis
106
selectivity sensoring and filtration [1] In 1994 Yaghi and co-workers introduced ways to synthesize
extended structures by design This new discipline is known as reticular chemistry [23] which uses
well-defined building blocks to create extended crystalline structures The feasibility of the building
block approach and the geometry preservation throughout the assembly process are the key factors
that lead to the design and synthesis of reticular structures
One of the first families of materials synthesized using reticular chemistry were the so-called Metal-
Organic Frameworks (MOFs) [4] They are composed of metal-oxide connectors which are covalently
bound to organic linkers The coordination versatility of constituent metal ions along with the
functional diversity of organic linker molecules has created immense possibilities The immense
potential of MOFs is facilitated by the fact that all building blocks are inexpensive chemicals and that
the synthesis can be carried out solvothermally MOFs are commercially available and the scale up of
production is continuing Since the discovery of MOFs many other crystalline frameworks have been
synthesized using reticular chemistry such as Metal-Organic Polyhedra (MOP) [5] Zeolite
Imidazolate Frameworks (ZIFs) [6] and Covalent Organic Frameworks (COFs) [7]
In 2005 Cocircteacute and co-workers introduced COF materials [7-14] where organic linker molecules are
stitched together by covalent entities including boron and oxygen atoms to form a regular
framework These materials have the distinct advantage that all framework bonds represent strong
covalent interactions and that they are composed of light-weight elements only which lead to a very
low mass density [7-9] These materials can be synthesized by wet-chemical methods by
condensation reactions and are composed of inexpensive and non-toxic building blocks so their
large-scale industrial application appears to be possible From a topological viewpoint we distinguish
two- and three-dimensional COFs In two-dimensional (2D) COFs the covalently bound framework is
restricted to layers The crystal is then similar as in graphite composed of a stack of layers which
are not connected by covalent bonds
COFs compared with MOFs have lower mass densities due to the absence of heavy atoms and
therefore might be better for many applications For example the gravimetric uptake of gases is
almost twice as large as that of MOFs with comparable surface areas [1516] Because COFs are fairly
new materials many of their properties and applications are still to be explored
Recently we have studied the structures of experimentally well-known 2D COFs [17] We have found
that commonly accepted 2D structures with AA and AB kinds of layering are energetically less stable
than inclined and serrated forms This is because AA stacking maximises the Coulomb repulsion due
to the close vicinity of charge carrying atoms alike (O B atoms) in neighbouring layers The serrated
and inclined forms are only slightly modified (layers are shifted with respect to each other by asymp14 Aring)
107
and experience less Coulomb forces between the layers compared to AA stacking This is equivalent
to the energetic preference of graphite for an AB (Bernal) over an AA form (simple hexagonal) if we
ignore the fact that interlayer ordering in serrated and inclined forms are not uniform everywhere A
known example of this is that in eclipsed hexagonal boron nitride (h-BN) boron atoms in one layer
serve as nearest neighbours to nitrogen atoms in adjacent layers (AB stacking) The Coulomb
interaction rules out possible interlayer eclipse between atoms with similar charges in this case
In the present work we aim to explore how far the concept of reticular chemistry is applicable to the
individual building units which define COFs For this purpose we have investigated a set of reported
and hypothetical 2D COFs theoretically by exploring their structural energetic and electronic
properties We have compared the properties of the isolated building blocks with those of the
extended crystal structures and have found that the properties of the building units are indeed
maintained after their assembly to a network
Results and Discussion
Structures and nomenclature
We have considered four connectors (IndashIV) and five linkers (andashe) for the systematic design of a
number of 2D COFs (Figure 1) Each COF was built from one type of connector and one type of linker
thus resulting in the design of 20 different structures Moreover we have considered two
hypothetical reference structures which are only built from connectors I and III (no linker is present)
REF-I and REF-III Properties of other COFs were compared with the properties of these two
structures Some of the studied COFs are already well known in the literature [781314] and we
continue to use their traditional nomenclature while hypothetical ones are labelled in the
chronological order with an M at the end which stands for modified
Figurel 1 The connector (IndashIV) and linker (andashe) units considered in this work The same nomenclature is used in the text Carbon ndash green oxygen ndashred boron ndash magenta hydrogen ndash white
108
Using the secondary building unit (SBU) approach we can distinguish the connectors between
trigonal [T] (connectors I II III) and hexagonal [H] (connector IV) and the linkers between linear [l]
(linkers a b e) and trigonal [t] (linkers c d) Topology of the layer is determined by the geometries
of the connector and linker molecules and typically a hexagonal pattern is formed due to the D3h
symmetry of the connector moieties Based on these topologies of the constituent building blocks
we have classified the studied COFs into four groups Tl Tt Hl and Ht (Figure 2) Hereafter we will
use this nomenclature to describe the COF topologies
Figurel 2 Topologies of 2D COFs considered in this work (from the left) Tl Tt Hl and Ht Red and blue blocks are secondary building units corresponding to connectors and linkers respectively
We have considered high-symmetry AA and AB kinds of stacking (hexagonal) and low-symmetry
serrated (orthorhombic) and inclined (monoclinic) kinds of stacking of the layers The latter two were
discussed in a previous work on 2D COFs [17] As an example the structure of COF-5 [7] in different
kinds of stacking of layers is shown in Figure 3 In eclipsed AA stacking atoms of adjacent layers lie
directly on top of each other whilst in staggered AB stacking three-connected vertices lie directly on
top of the geometric center of six-membered rings of neighbouring layers In both serrated and
inclined kinds of stacking the layers are shifted with respect to each other by approximately 14 Aring
resulting in hexagonal rings in the connector or linker being staggered with those in the adjacent
layers In serrated stacking alternate layers are eclipsed In inclined stacking layers lie shifted along
one direction and the lattice vector pointing out of the 2D plane is not rectangular For COFs made of
connector I due to the absence of five-membered C2O2B rings a zigzag shift leads to staggering in
both connector and linker parts For those made of other connectors staggering at the connector or
linker depends on whether the shift is armchair or zigzag respectively [17]
Structural properties
We have compared structural properties of isolated building blocks with those of the extended COF
structures Negligible differences have been found in the bond lengths and bond angles of the
building blocks and the corresponding crystal structures This indicates that the structural integrity of
the building blocks remains unchanged while forming covalent organic frameworks confirming the
109
principle of reticular chemistry In addition the CndashB BndashO and OndashC bond lengths are almost the same
when different COF structures are compared (see Table S1 in Supporting Information File 1) The
experimental bond lengths are asymp154 Aring for CndashB asymp138 Aring for OndashC and 137ndash148 Aring for BndashO However
in the case of COF-1 the experimental values are slightly larger (160 Aring for CndashB and 151 Aring for BndashO)
This could be the reason why our calculated bond lengths for COF-1 are much shorter than the
experimental values while all the other structures agree within 5 error The calculated CndashC bond
lengths vary in the range from 136ndash147 Aring (Figure S2 in Supporting Information File 1) and are the
same for the COFs and their constituent building blocks at the respective positions of the carbon
atoms In addition the reference structures REF-I and REF-III have direct BndashB bond lengths of 167 Aring
and 166 Aring respectively which is shorter by 014 Aring than a typical BndashB bond length The calculated
bond angles OBO in B3O3 and C2O2B rings are 120deg and 113deg respectively
Figure 3 Layer stackings considered AA AB serrated and inclined
Interlayer distances (d) which is the shortest distance between two layers (equivalent to c for AA
c2 for AB and serrated) are different in all kinds of stacking AB stacked 2D COFs have shorter
interlayer distances than the corresponding AA serrated and inclined stacked structures Among the
latter three AA stacked COFs have higher values for d because of the higher repulsive interlayer
orbital interactions resulting from the direct overlap of polarized alike atoms between the adjacent
layers This results in higher mass densities for AB stacked COF analogues Serrated and inclined
stacks have only slightly higher mass densities compared to AA The differences in mass densities for
all kinds of stacking are attributed to the differences in their interlayer separations The d values of
most of the COFs are larger than that of graphite in AA stacking but smaller in AB stacking
Cell parameters (a) and mass densities (ρ) of all the COFs constructed from the considered
connectors and linkers are shown in Table 1 (and in Table S3 in Supporting Information File 1) Mass
densities of all the COFs are much lower than that of graphite (227 gmiddotcmminus3) and diamond (350
gmiddotcmminus3) AAserratedinclined stacked COF-10s have the lowest mass densities (045046046
gmiddotcmminus3) which is lower than that of MOF-5 (059 gmiddotcmminus3) [4] and comparable to that of highly porous
MOF-177 (042 gmiddotcmminus3) [18]
110
In order to identify the stacking orders we have analyzed X-ray diffraction (XRD) patterns of the well-
known COFs (COF-10 TP COF PPy-COF see Figure 4) in all the above discussed stacking kinds The
change of stacking should be visible in XRDs because each space group has a distinct set of symmetry
imposed reflection conditions The XRD patterns of AA serrated and inclined stacking kinds which
differ within a slight shift of adjacent layers to specific directions are in agreement with the presently
available experimental data [81314] Their peaks are at the same angles as in the experimental
spectrum whereas AB stacking clearly shows differences The slight differences in the (001) angle
between each stacking resemble the differences in their interlayer separations The inclined
stackings have more peaks however they are covered by the broad peaks in the experimental
patterns Similar results for COF-1 COF-5 COF-6 and COF-8 have been discussed in our previous
work [17]
Figure 4 The calculated and experimental [81314] XRD patterns of PPy-COF (top) COF-10 (middle) and TP COF (bottom)
111
Table 1 The calculated unit cell parameter a [Aring] interlayer distance d [Aring] and mass density ρ [gmiddotcmminus3
] for AA and AB stacked COFs Note that the cell parameter a is the same for all stacking types Experimental data [719] is given in parentheses
COF Building
Blocks
a d ρ
AA AB AA AB
COF-1 I-a 1502 (15620) 351 313 (332) 094 106
COF-1M I-b 2241 349 304 068 078
COF-2M I-c 1492 347 312 095 106
COF-3M I-d 0747 349 327 153 164
PPy-COF I-e 2232 (22163) 349 (3421) 297 084 099
COF-5 II-a 3014 (30020) 347 (3460) 326 056 060
COF-10 II-b 3758 (37810) 347 (3476) 318 045 (045) 050
COF-8 II-c 2251 (22733) 346 (3476) 320 071 (070) 077
COF-6 II-d 1505 (15091) 346 (3599) 327 104 110
TP COF II-e 3750 (37541) 348 (3378) 320 051 056
COF-4M III-a 2171 350 318 073 080
COF-5M III-b 2915 350 318 055 061
COF-6M III-c 1833 345 318 083 090
COF-7M III-d 1083 350 330 129 136
TP COF-1M III-e 2905 349 310 065 074
COF-8M IV-a 1748 359 329 140 148
COF-9M IV-b 2176 349 330 117 174
COF-10M IV-c 2254 342 336 127 128
COF-11M IV-d 1512 346 338 168 172
TP COF-2M IV-e 2173 347 332 134 140
REF-I I 0773 359 336 144 148
REF-III III 1445 353 336 104 121
Graphite 247 343 335 220 227
112
Energetic stability
We have considered dehydration reactions the basis of experimental COF synthesis to calculate
formation energies of COFs in order to predict their energetic stability Molecular units 14-
phenylenediboronic acid (BDBA) 11rsquo-biphenyl]-44-diylboronic acid (BPDA) 5rsquo-(4-boronophenyl)-
11rsquo3rsquo1rdquo-terphenyl]-44rdquo-diboronic acid (BTPA) benzene-135-triyltriboronic acid (BTBA) and
pyrene-27-diylboronic acid (PDBA) were considered as linkers andashe respectively with -B(OH)2 groups
attached to each point of extension (Figure 5) Self-condensation of these building blocks result in
the formation of B3O3 rings and the resultant COFs are those made of connector I and the
corresponding linkers This process requires a release of three or six water molecules in case of t or l
topology respectively
Figure 5 The reactants participating in the formation of 2D COFs
Co-condensation of the above molecular units with compounds such as 23671011-
hexahydroxytriphenylene (HHTP) hexahydroxybenzene (HHB) and dodecahydroxycoronene (DHC)
(Figure 5) gives rise to COFs made of connectors II III and IV respectively and the corresponding
linkers Self-condensation of tetrahydroxydiborane (THDB) and co-condensation of HHB with THDB
result in the formation of the reference structures REF-I and REF-III respectively In relation to the
corresponding connectorlinker topologies these building blocks satisfy the following equations of
the co-condensation reaction for COF formation
2 2 3 COF 12 H O Tl T l (1)
113
2 1 1 COF 6 H O Tt T t (2)
2 1 3 COF 12 H O Hl H l (3)
2 1 2 COF 12 H O Ht H t (4)
with a stochiometry appropriate for one unit cell The number of covalent bonds formed between
the building blocks is equivalent to the number of released water molecules we refer to it as
ldquoformula unitrdquo and will give all energies in the following in kJmiddotmolminus1 per formula unit
Table 2 The calculated energies [kJ molminus1
] per bond formed between building blocks for AA and AB stacked COFs Ecb is the condensation energy Esb is the stacking energy and Efb is the COF formation energy (Efb = Ecb
+ Esb) The calculated band gaps Δ eV+ are given as well
COF Building
Blocks
Mono-
layer
AA AB
Ecb Esb Efb ∆ Esb Efb ∆
COF-1 I-a 906 -2683 -1777 33 -2187 -1280 36
COF-1M I-b 949 -4266 -3366 27 -2418 -1469 31
COF-2M I-c 956 -5727 -4771 28 -4734 -3778 30
COF-3M I-d 763 -2506 -1742 38 -2801 -2037 40
PPy-COF I-e 858 -5723 -4866 24 -3855 -2998 26
COF-5 II-a 211 -2968 -2756 24 -2548 -2337 28
COF-10 II-b 317 -3766 -3448 23 -1344 -1026 26
COF-8 II-c 263 -4488 -4224 25 -2477 -2213 28
COF-6 II-d 185 -2881 -2695 28 -2127 -1942 31
TP COF II-e 231 -4453 -4222 24 -1480 -1250 27
COF-4M III-a -033 -1730 -1763 26 -880 -913 26
COF-5M III-b 007 -2533 -2526 25 -972 -965 25
COF-6M III-c 014 -3231 -3217 26 -2134 -2120 28
114
COF-7M III-d -170 -1635 -1805 30 -1607 -1777 32
TP COF-1M III-e -014 -3226 -3240 24 -1277 -1291 24
COF-8M IV-a -787 -2756 -3543 18 -2680 -3467 21
COF-9M IV-b -836 -3577 -4414 17 -3003 -3839 21
COF-10M IV-c -947 -4297 -5244 18 -4192 -5140 22
COF-11M IV-d -403 -2684 -3087 21 -2833 -3236 24
TP COF-2M IV-e 030 -4345 -4315 18 -4117 -4087 21
We have calculated the condensation energy of a single COF layer formed from monomers (building
blocks hereafter called bb) according to the above reactions as follows
tot tot tot totcb m H2O 1 bb1 2 bb2E E n E ndash m E m E (5)
where Emtot ndash total energy of the monolayer EH2O
tot is the total energy of the water molecule Ebb1tot
and Ebb2tot are the total energies of interacting building blocks and n m1 m2 are the corresponding
stoichiometry numbers
We have also calculated the stacking energy Esb of layers
tot totsb nl s mE E n E (6)
where Enltot is the total energy of ns number of layers stacked in a COF Finally the COF formation
energy can be given as a sum of Ecb and Esb (see Table 1 and Table S1)
Electronic properties
All COFs including the reference structures are semiconductors with their band gaps lying between
17 eV and 40 eV (Table 2 and Table S4 in Supporting Information File 1) The largest band gaps are
of the reference structures while the lowest values are of COFs built from connector IV The band
gaps are different for different stacking kinds This difference can be attributed to the different
optimized interlayer distances Generally AB serrated and inclined stacked COFs have band gaps
comparable to or larger than that of their AA stacked analogues
115
We have calculated the electronic total density of states (TDOS) and the individual atomic
contributions (partial density of states PDOS) The energy state distributions of COFs and their
monolayers are studied and a comparison for COF-5 is shown in Figure 6 In all stacking kinds
negligible differences are found for the densities at the top of valence band and the bottom of
conduction band These slight differences suggest that the weak interaction between the layers or
the overlap of π-orbitals does not affect the electronic structure of COFs significantly Hence there is
almost no difference between the TDOS of AA AB serrated and inclined stacking kinds Therefore in
the following we discuss only the AA stacked structures
Figure 6 Total densities of states (DOS) (black) of AA (top left) AB (top right) serrated (bottom left) and inclined (bottom right) comparing stacked COF-5 with a monolayer (red) of COF-5 The differences between the TDOS of bulk and monolayer structures are indicated in green The Fermi level EF is shifted to zero
Figure 7 Partial density of states of carbon (left) boron (center) and oxygen (right) atoms of COFs built from type-I connectors and different linkers The vertical dashed line in each figure indicates the Fermi level EF
116
It is of interest to see how the properties of COFs change depending on their composition of different
secondary building units that is for different connectors and linkers PDOS of COFs built from type-I
connectors and different linkers are plotted in Figure 7 The PDOS of carbon atoms is compared with
that of graphite (AA-stacking) while PDOS of boron and oxygen atoms are compared with that of
REF-I a structure which is composed solely of connector building blocks Going from top to bottom
of the plots the number of carbon atoms per unit cell increases It can be seen that this causes a
decrease of the band gap Figure 8 shows the PDOS of COFs built from type-a linkers and different
connectors where the COFs are arranged in the increasing number of carbon atoms in their unit cells
from top to the bottom Again the C PDOS is compared with that of graphite while both REF-I and
REF- III are taken in comparison to O and B PDOS The observed relation between number of carbon
atoms and band gap is verified
Figure 8 Partial density of states of carbon (left) boron (center) and oxygen (right) atoms of COFs built from type-a linkers The vertical dashed line in each figure indicates the Fermi level EF
Conclusion
In summary we have designed 2D COFs of various topologies by connecting building blocks of
different connectivity and performed DFTB calculations to understand their structural energetic and
electronic properties We have studied each COF in high-symmetry AA and AB as well as low-
symmetry inclined and serrated stacking forms The optimized COF structures exhibit different
interlayer separations and band gaps in different kinds of layer stackings however the density of
states of a single layer is not significantly different from that of a multilayer The alternate shifted
layers in AB serrated and inclined stackings cause less repulsive orbital interactions within the layers
which result in shorter interlayer separation compared to AA stacking All the studied COFs show
117
semiconductor-like band gaps We also have observed that larger number of carbon atoms in the
unit cells in COFs causes smaller band gaps and vice versa Energetic studies reveal that the studied
structures are stable however notable difference in the layer stacking is observed from
experimental observations This result is also supported by simulated XRDs
Methods
We have optimized the atomic positions and the lattice parameters for all studied COFs All
calculations were carried out at the Density Functional Tight-Binding (DFTB) [2021] level of theory
DFTB is based on a second-order expansion of the KohnndashSham total energy in the Density-Functional
Theory (DFT) with respect to charge density fluctuations This can be considered as a non-orthogonal
tight-binding method parameterized from DFT which does not require large amounts of empirical
parameters however maintains all the qualities of DFT The main idea behind this method is to
describe the Hamiltonian eigenstates with an atomic-like basis set and replace the Hamiltonian with
a parameterized Hamiltonian matrix whose elements depend only on the inter-nuclear distances and
orbital symmetries [21] While the Hamiltonian matrix elements are calculated using atomic
reference densities the remaining terms to the KohnndashSham energy are parameterized from DFT
reference calculations of a few reference molecules per atom pair The accuracy is very much
improved by the self-consistent charge (SCC) extension Two computational codes were used
deMonNano code [22] and DFTB+ code [23] The first code has dispersion correction [24]
implemented to account for weak interactions and was used to obtain the layered bulk structure of
COFs and their formation energies The performance for interlayer interactions has been tested
previously for graphite [24] The second code which can perform calculations using k-points was
used to calculate the electronic properties (band structure and density of states) Band gaps have
been calculated as an additional stability indicator While these quantities are typically strongly
underestimated in standard LDA- and GGA-DFT calculations they are typically in the correct range
within the DFTB method For validation of our method we have calculated some of the structures
using Density Functional Theory (DFT) as implemented in ADF code [2526]
Periodic boundary conditions were used to represent frameworks of the crystalline solid state The
conjugatendashgradient scheme was chosen for the geometry optimization The atomic force tolerance of
3 times 10minus4 eVAring was applied The optimization using Γ-point approximation was performed with the
deMon-Nano code on 2times2times4 supercells Some of the monolayers were also optimized using the
DFTB+ code on elementary unit cells in order to validate the calculations within the Γ-point
approximation The number of k-points has been determined by reaching convergence for the total
energy as a function of k-points according to the scheme proposed by Monkhorst and Pack [27]
118
Band structures were computed along lines between high symmetry points of the Brillouin zone with
50 k-points each along each line XRD patterns have been simulated using Mercury software [2829]
We have also performed first-principles DFT calculations at the PBE [30] DZP [31] level to support
our results quantitatively For simplicity we have used finite structures instead of bulk crystals
Supporting Information
Table S1 The calculated bond lengths and angles of all studied COFs Corresponding experimental values found from the literature are shown in brackets
COF Building
Blocks
C-B B-O O-C OBO
COF-1 I-a 1498 (1600) 1393 (1509) 1203 (1200)
COF-1M I-b 1497 1393 1203
COF-2M I-c 1497 1392 1203
COF-3M I-d 1496 1392 1201
PPy-COF I-e 1498 1393 1202 (1190)
COF-5 II-a 1496 (1530) 1399 (1367) 1443 (1384) 1135dagger (1139)
COF-10 II-b 1495 (1553) 1399 (1465) 1443 (1385) 1135dagger (1120)
COF-8 II-c 1495 (1548) 1399 (1479) 1443 1135dagger
COF-6 II-d 1496 1399 1443 1135dagger
TP COF II-e 1496 (1542) 1399 (1396) 1444 (1377) 1135dagger (1182)
COF-4M III-a 1496 1398 1449 1135dagger
COF-5M III-b 1496 1398 1449 1136dagger
COF-6M III-c 1496 1399 1451 1134dagger
COF-7M III-d 1496 1398 1449 1136dagger
TP COF-1M III-e 1496 1398 1450 1136dagger
COF-8M IV-a 1496 1398 1445 1131dagger
COF-9M IV-b 1495 1398 1444 1131dagger
119
COF-10M IV-c 1495 1391 1418 1126dagger
COF-11M IV-d 1498 1399 1450 1134dagger
TP COF-2M IV-e 1499 1399 1447 1134dagger
B3O3 connectivity dagger C2B2O connectivity
It can be noticed that the bond lengths of experimentally known COF-1 is much larger compared to
our optimized bond lengths as well as that of other synthesized COFs
Figure S2 Calculated C-C bond lengths in connectors and linkers Hydrogen atoms are omitted for clarity
Table S3 The calculated unit cell parameters a interlayer distance d Aring+ and mass density ρ gcm-3
] for serrated (Sa serrated armchair Sz serrated zigzag) and inclined (Ia inclined armchair Iz inclined zigzag) stacked COFs
COF Building
Blocks
a d ρ
Sa Sz Ia Iz Sa Sz Ia Iz
COF-1 I-a 1502 343 343 097 097
COF-1M I-b 2241 341 342 069 069
COF-2M I-c 1492 340 339 097 097
COF-3M I-d 0747 341 342 157 156
PPy-COF I-e 2232 341 341 086 086
120
COF-5 II-a 3014 342 342 341 340 057 057 058 058
COF-10 II-b 3758 341 341 342 340 046 046 046 046
COF-8 II-c 2251 341 341 342 342 073 073 072 072
COF-6 II-d 1505 342 341 340 340 105 106 106 106
TP COF II-e 3750 342 341 342 342 052 052 052 052
COF-4M III-a 2171 344 344 345 344 074 074 074 074
COF-5M III-b 2915 343 342 343 343 056 056 056 056
COF-6M III-c 1833 341 341 342 341 084 084 084 084
COF-7M III-d 1083 344 343 340 344 131 131 132 131
TP COF-1M III-e 2905 343 342 343 342 066 067 066 066
COF-8M IV-a 1748 341 341 342 342 142 142 142 142
COF-9M IV-b 2176 341 341 341 342 119 119 119 119
COF-10M IV-c 2254 340 340 340 340 128 128 128 128
COF-11M IV-d 1512 341 341 340 340 171 171 171 171
TP COF-2M IV-e 2173 340 340 340 340 137 137 137 137
REF-I I 0773 349 345 148 15
REF-III III 1445 348 349 106 106
Table S4 The calculated energies [kJ mol-1
] per bond formed between building blocks for serrated (Sa serrated armchair Sz serrated zigzag) and inclined (Ia inclined armchair Iz inclined zigzag) stacked COFs Ecb ndash condensation energy Esb ndash stacking energy and Efb ndash COF formation energy (Efb = Ecb + Esb) The calculated band gaps ∆ eV+ are given as well
COF Sa Sz Ia Iz
Esb Efb Δ Esb Efb Δ Esb Efb Δ Esb Efb Δ
-1 -2810 -1904 36 -2786 -1880 36
-1M -4426 -3477 30 -4389 -3440 30
-2M -5967 -5011 30 -5833 -4877 30
121
-3M -2667 -1904 40 -2591 -1828 40
PPy- -5916 -5058 26 -5865 -5007 26
-5 -3052 -2841 27 -3051 -2840 26 -3275 -3064 26 -3297 -3086 26
-10 -3868 -3551 26 -3869 -3552 25 -3830 -3513 26 -4293 -3976 25
-8 -4615 -4352 27 -4614 -4351 27 -4571 -4308 27 -4573 -4310 27
-6 -2978 -2793 30 -2978 -2793 30 -3117 -2932 30 -3090 -2905 30
TP -4557 -4326 26 -4562 -4331 26 -4511 -4280 26 -4511 -4280 26
-4M -1811 -1844 28 -1813 -1846 28 -1769 -1802 28 -1880 -1913 28
-5M -2626 -2619 27 -2630 -2623 26 -2597 -2590 26 -2600 -2593 26
-6M -3375 -3361 28 -3381 -3367 28 -3487 -3473 28 -3349 -3335 28
-7M -1724 -1894 32 -1726 -1896 31 -2333 -2503 32 -1866 -2036 31
TP -1M -3327 -3341 26 -3334 -3348 26 -3340 -3354 26 -3353 -3367 26
-8M -2897 -3684 21 -2895 -3682 21 -2971 -3758 21 -2983 -3770 21
-9M -3732 -4568 20 -3733 -4569 20 -3684 -4520 20 -3706 -4542 20
-10M -4512 -5459 21 -4513 -5460 21 -4491 -5438 21 -4495 -5442 21
-11M -2831 -3234 24 -2834 -3237 24 -2816 -3219 24 -2830 -3233 24
TP -2M -4500 -4470 20 -4505 -4475 20 -4495 -4465 20 -4496 -4466 20
122
Appendix E
Stability and electronic properties of 3D covalent organic frameworks
Binit Lukose Agnieszka Kuc and Thomas Heine
Prepared for publication
Abstract Covalent Organic Frameworks (COFs) are a class of covalently linked crystalline nanoporous
materials versatile for nanoelectronic and storage applications 3D COFs in particular have very
large pores and low-mass densities Extensive theoretical studies of their energetic and mechanical
stability as well as their electronic properties are discussed in this paper All studied 3D COFs are
energetically stable materials though mechanical stability does not exceed 20 GPa Electronically all
COFs are semiconductors with band gaps depending on the number of sp2 carbon atoms present in
the linkers similar to 3D MOF family
Introduction
Covalent Organic Frameworks (COFs)1-3 comprise an emerging class of crystalline materials that
combines organic functionality with nanoporosity COFs have organic subunits stitched together by
covalent entities including boron carbon nitrogen oxygen or silicon atoms to form periodic
frameworks with the faces and edges of molecular subunits exposed to pores Hence their
applications can range from organic electronics to catalysis to gas storage and sieving4-7 The
properties of COFs extensively depend on their molecular constituents and thus can be tuned by
rational chemical design and synthesis289 Step by step reversible condensation reactions pave the
123
way to accomplish this target Such a reticular approach allows predicting the resulting materials and
leads to long-range ordered crystal structures
Since their first mentioning in the literature13 COFs are under theoretical investigation10-17 mainly for
gas storage applications Methods such as doping18-24 and organic linker functionalization2526 have
been suggested to improve their storage capacities In addition to the moderate pore size and
internal surface area COFs have the privileges of a low-weight material as they are made of light
elements Hence their gravimetric adsorption capacity is remarkably high10 The lowest mass density
ever reported for any crystalline material is that for COF-108 (018 gcm-3) Also their stronger
covalent bonds compared to related Metal-Organic Frameworks (MOFs)27 endorse thermodynamic
strength These genuine qualities of COFs make them attractive for hydrogen storage investigations
Crystallization of linked organic molecules into 2D and 3D forms was achieved in the years 20051 and
20073 respectively by the research group of O M Yaghi Several COFs have been synthesized since
then and some of the 2D COFs has been proven good for electronic or photovoltaic applications428-33
However the growth in this area appears to be slow compared to rapidly developing MOFs albeit
the promising H2 adsorption measurements53435 and a few synthetic improvements736-42
COF-102 and COF-103 were synthesized3 by the self-condensation of the non-planar tetra(4-
dihydroxyborylphenyl)methane (TBPM) and tetra(4-dihydroxyborylphenyl)silane (TBPS) respectively
(see Fig 1 for the building blocks and COF geometries) The co-condensation of these compounds
with triangular hexahydroxytriphenylene (HHTP) results in COF-108 and COF-105 respectively with
different topologies Yaghi et al3 have reported the formation of highly symmetric topologies ctn
(carbon nitride dI 34 ) and bor (boracite mP 34 ) when tetrahedral building blocks are linked
together with triangular ones The topology names were adopted from reticular chemistry structure
resource (RCSR)43 Simulated Powder X-ray diffraction in comparison with experimental powder
spectrum suggested ctn topology for COF-102 103 and 105 and bor topology for COF-108 The
condensation of tert-butylsilanetriol (TBST) and TBPM leads to the formation of COF-20244 This was
reported as ctn topology The COF reactants and schematic diagrams of ctn and bor topologies are
given in Figure1 The assembly of tetrahedral and linear units is very likely to result in a diamond-like
form COF-300 was formed according to the condensation of tetrahedral tetra-(4-anilyl)methane
(TAM) and linear terephthaldehyde (TA)45 However the synthesized structure was 5-fold
interpenetrated dia-c5 topology43
In this work we present theoretical studies of 3D COFs using density functional based methods to
explore their structural electronic energetic and mechanical properties Our previous studies on 2D
COFs4647 had resulted in questioning the stability of eclipsed arrangement of layers (AA stacking) and
124
suggesting energetically more stable serrated and inclined packing In this paper we attempt to
explore the stability and electronic properties of the experimentally known 3D COFs namely COF-
102 103 105 108 202 and 300 In the present studies we follow the reticular assembly of the
molecular units that form low-weight 3D COFs Considering the structural distinction from MOFs
COFs lack the versatility of metal ingredients what results in diverse properties48 A collective study is
then carried out to understand the characteristics and drawbacks of COFs
Figure 1 (Upper panel) Building blocks for the construction of 3D COFs (Lower panel) lsquoctnrsquo and lsquoborrsquo
networks formed by linking tetrahedral and triangular building units
Methods
COF structures were fully optimized using Self-consistent Charge -Density Functional based Tight-
Binding (SCC-DFTB) level of theory4950 Two computational codes were used deMonNano51 and
125
DFTB+52 The first code which has dispersion correction53 implemented to account for weak
interactions was used for the geometry optimization and stability calculations The second code
which can perform calculations using k-point sampling was used to calculate the electronic
properties (band structure and density of states) The number of k-points has been determined by
reaching convergence for the total energy as a function of k-points according to the scheme
proposed by Monkhorst and Pack54 Periodic boundary conditions were used to represent
frameworks of the crystalline solid state Conjugatendashgradient scheme was chosen for the geometry
optimization Atomic force tolerance of 3 times 10minus4 eVAring was applied The optimization using Γ-point
approximation was performed on rectangular supercells containing more than 1000 atoms For
validation of our method we have calculated energetic stability using Density Functional Theory (DFT)
at the PBE55DZP56 level as implemented in ADF code5758 using cluster models The cluster models
contain finite number of building units and correspond to the bulk topology of the COFs XRD
patterns have been simulated using Mercury software5960
In this work we continued to use the traditional nomenclature of the experimentally known COFs All
of the structures have the same tetrahedral blocks and differ only in the central sp3 atom (carbon or
silicon) that is included in our nomenclature
Bulk modulus (B) of a solid at absolute zero can be calculated as
(1) B = 2
2
dV
EdV
where V and E are the volume and energy respectively
Owing to the dehydration reactions we have calculated the formation (condensation) energy of each
COF formed from monomers (building blocks) as follows
(2) EF = Etot + n EH2Otot ndash (m1 Ebb1
tot + m2 Ebb2tot)
where Etot -- total energy of the COF EH2Otot -- total energy of water molecule Ebb1
tot and Ebb2tot -- total
energies of interacting building blocks n m1 m2 -- stoichiometry numbers
Results and Discussions
Structure and Stability
We have optimized the atomic positions and cell dimensions of the COFs in the experimentally
determined topologies Cell parameters in comparison with experimental values are given in Table 1
The calculated bond lengths are C-B = 150 C-C = 139-144 (COF-300) C(sp3)-C = 156 C-O = 142 B-
126
O = 140 Si-C = 188 Si-O = 186 C-N = 131-136 Aring These values agree within 6 error with the
experimental values34445
Mass densities (see Table 1) of COFs range between 02 to 05 g cm-3 and are smaller for the silicon at
the tetrahedral centers This implies that the presence of silicon atoms impacts more on the cell
volume than on the total mass That means the replacement of sp3 C with Si in COF-108 can change
its mass density to a slightly lower value To our best knowledge among all the natural or
synthesized crystals COF-108 has the lowest mass-weight
In order to validate our optimized structures we have simulated X-ray Diffraction of each COF and
compared them with the available experimental spectra (see Figure2) Almost all of the simulated
XRDs have excellent correlation in the peak positions with the experiment Only COF-300 shows
somehow significant differences in the intensities These differences may be attributed to the
presence of guest molecules in the synthesized COF-30045
Table 1 The calculated cell parameters [Aring] mass density ρ gcm-3
+ band gap Δ eV+ bulk modulus B GPa+
and formation energy EF [kJ mol-1
] for all the studied 3D COFs Experimental values are given in brackets
along with the cell parameters HOMO-LUMO gaps [eV] of the building blocks are also given in brackets
along with the band gaps
Structure Building
Blocks
Cell
parameters
ρ Δ B EF
COF-102 (C) TBPM 2707 (2717) 043 39 (43) 206 -14995
COF-103 (Si) TBPS 2817 (2825) 039 38 (41) 139 -14547
COF-105-C TBPM HHTP 4337 019 33 (43 34) 80 -18080
COF-105 (Si) TBPS HHTP 4444 (4489) 018 32 (41 34) 79 -17055
COF-108 (C) TBPM HHTP 2838 (2840) 017 32 (43 34) 37 -17983
COF-108-Si TBPS HHTP 2920 016 30 (41 34) 29 -18038
COF-202 (C) TBPM TBST 3047 (3010) 050 42 (43 139) 143 -7954
COF-202-Si TBPS TBST 3157 046 39 (41 139) 153 -7632
COF-300 (C) TAM TA 2932 925 049 23 (41 26) 144 -40286
127
(2828 1008)
COF-300-Si TAS TA 3039 959 044 23 (40 26) 133 -39930
tetra-(4-anilyl)silane
Figure 2 Simulated XRDs are compared with the experimental patterns from the literature COF-300
exhibits some differences between the simulated and experimental XRDs while others show reasonably
good match
The bulk modulus (B) is a measure of mechanical stability of the materials The calculated values of B
are given in Table 1 Compared to the force-field based calculation of bulk modulus by Schmid et
al61 these values are larger The bulk modulus of COF-105 and COF-108 are relatively small
compared with other COFs Considering that the two COFs differ only in the topology it may be
concluded that ctn nets are mechanically more stable compared to bor COFs with carbon atoms in
the center are mechanically more stable than those with silicon atoms Bulk modulus of COF-102
103 COF-202 and interpenetrated COF-300 are higher than many of the isoreticular MOFs62 and
comparable to IRMOF-6 (1241 GPa) MOF-5 (1537 GPa)63 bNon-interpenetrated COF-300 (single
framework dia-a topology43) has much lower bulk modulus of only 317 GPa
Calculated formation energies (Table 1) are given in terms of reaction energies according to Eq 2
Condensation of monomers to form bulk 3D structures is exothermic in all the cases supporting
reticular approach The presence of C or Si at the vertex center does not show any particular trend in
the formation energies We have calculated the formation energy of non-interpenetrated COF-300
(dia-a) to be -39346 kJ mol-1 which is comparable to the interpenetrated cases For a quantitative
comparison we have performed DFT calculations at the PBEDZP level as implemented in ADF code
on finite structures The obtained formation energies for COF-105-C COF-105 (Si) COF-108 (C) COF-
108-Si are -9944 -9968 -1245 -12436 respectively They are in a reasonable agreement with the
128
DFTB results A comparison between COF-105 and COF-108 suggests that bor nets are energetically
more favored than ctn nets
Electronic Properties
Band gaps (Δ) calculated for the 3D COFs are in the range of 23-42 eV (see Table 1) which show
their semiconducting nature similar to the hexagonal 2D COFs47 and MOFs62 The band gap
decreases with the increase of conjugated rings in the unit cell relative to the number of other atoms
Si atoms in comparison with carbon atoms at the tetrahedral centers give a relatively smaller Δ This
is evident from the partial density of states of C (sp3) in COF-102 and Si in COF-103 plotted in Figure3
Carbon atoms define the band edges of the PDOS (see Figure4) The band gaps of COF-105 and COF-
108 differ only slightly (see Figure5) which means that the band gap is nearly independent of the
topology This is because for each atom the coordination number and the neighboring atoms remain
the same in both ctn and bor networks albeit the topology difference DOS of non-interpenetrated
(dia-a) and 5-fold interpenetrated (dia-c5) COF-300 are plotted in Figure6 It may be concluded from
their negligible differences that interpenetration does not alter the DOS of a framework We have
shown similar results for 2D COFs47
Figure 3 Partial density of states (DOS) of sp3 C in COF-102 (top) and Si in COF-103 (bottom) The latter is
inverted for comparison The Fermi level EF is shifted to zero
129
Figure 4 Partial and total density of states of COF-103 The Fermi level EF is shifted to zero
Figure 5 Density of states of COF-108 (bor) and COF-105 (ctn) The two COFs differ only in the topology
130
Figure 6 Density of states of non-interpenetrated (dia-c1) and 5-interpenetrated (dia-c5) COF-300
We also have calculated HOMO-LUMO gaps of the molecular building blocks (see Table 1) In
comparison band gaps of COFs are slightly smaller than the smallest of the HOMO-LUMO gaps of the
building units
Conclusion
In summary we have calculated energetic mechanical and electronic properties of all the known 3D
COF using Density Functional Tight-Binding method Energetically formation of 3D COFs is favorable
supporting the reticular chemistry approach Mechanical stability is in line with other frameworks
materials eg MOFs and bulk modulus does not exceed 20 GPa Also all COFs are semiconducting
with band gaps ranging from 2 to 4 eV Band gaps are analogues to the HOMO-LUMO gaps of the
molecular building units We believe that this extensive study will define the place of COFs in the
broad area of nanoporous materials and the information obtained from the work will help to
strategically develop or modify porous materials for the targeted applications
131
Appendix F
Structure electronic structure and hydrogen adsorption capacity of porous aromatic frameworks
Binit Lukose Mohammad Wahiduzzaman Agnieszka Kuc and Thomas Heine
Prepared for publication
Abstract
Diamond-like porous aromatic frameworks (PAFs) form a new family of materials that contain only
carbon and hydrogen atoms within their frameworks These structures have very low mass densities
large surface area and high porosity Density-functional based calculations indicate that crystalline
PAFs have mechanical stability and properties similar to those of Covalent Organic Frameworks Their
exceptional structural properties and stability make PAFs interesting materials for hydrogen storage
Our theoretical investigations show exceptionally high hydrogen uptake at room temperature that
can reach 7 wt a value exceeding those of well-known metal- and covalent-organic frameworks
(MOFs and COFs)
Introduction
Porous materials have been widely investigated in the fields of materials science and technology due
to their applications in many important fields such as catalysis gas storage and separation template
materials molecular sensors etc1-3 Designing porous materials is nowadays a simple and effective
strategy following the approach of reticular chemistry4 where predefined building blocks are used to
132
predict and synthesize a topological organization in an extended crystal structure The most famous
and striking examples of such materials are Metal- and Covalent-Organic Frameworks (MOFs and
COFs)56 These new nanoporous materials have many advantages high porosity and large surface
areas lowest mass densities known for crystalline materials easy functionalization of building blocks
and good adsorption properties
Gas storage and separation by physical adsorption are very important applications of such
nanoporous materials and have been major subjects of science in the last two decades These
applications are based on certain physical properties namely presence of permanent large surface
area and suitable enthalpy of adsorption between the host framework and guest molecules
Attempts to produce materials with large internal surface area have been successful and some of the
notable accomplishments include the synthesis of MOF-2107 (BET surface area of 6240 m2 g-1 and
Langmuir surface area of 10400 m2 g-1) NU-1008 (BET surface area 6143 m2 g-1) or 3D COFs9 (BET
surface area 4210 m2 g-1 for COF-103)
More recently a new family of porous materials emerged So-called porous-aromatic frameworks
(PAFs) have surface areas comparable to MOFs eg PAF-110 has a BET surface area of 5640 m2 g-1 and
Langmuir surface area of 7100 m2 g-1 Since they are composed of carbon and hydrogen only they
have several advantages over frameworks containing heavy elements MOFs with coordination bonds
often suffer from low thermal and hydrothermal stability what might limit their applications on the
industrial scale The coordination bonds can be substituted with stronger covalent bonds as it was
realized in the case of COFs6 however this lowers significantly their surface areas comparing with
MOFs
Besides its exceptional surface area PAF-110 has high thermal and hydrothermal stability and
appears to be a good candidate for hydrogen and carbon dioxide storage applications PAFs have
topology of diamond with C-C bonds replaced by a number of phenyl rings (see Figure 1)
Experimentally obtained PAF-110 and PAF-1111 have eg one and four phenyl rings respectively
connected by tetragonal centers (carbon tetrahedral nodes) The simulated and experimental
hydrogen uptake capacities of such PAFs exceed the DOE target12
In this paper we have studied structural electronic and adsorption properties of PAFs using Density
Functional based Tight-Binding (DFTB) method1314 and Quantized Liquid Density Functional Theory
(QLDFT)15 We have found exceptionally high storage capacities at room temperature what makes
PAFs very good materials for hydrogen storage devices beyond well-known MOFs and COFs We have
compared our results to widely-used classical Grand Canonical Monte-Carlo (GCMC) calculations
reported in the literature We have also studied other properties of these materials such as
133
structural energetic electronic and mechanical We explored the structural variance of diamond
topology by individually placing a selection of organic linkers between carbon nodes This generally
changes surface area mass density and isosteric heat of adsorption what is reflected in the
adsorption isotherms
Methods
Using a conjugate-gradient method we have fully optimized the crystal structures (atomic positions
and unit cell parameters) of PAFs The calculations were performed using Dispersion-corrected Self-
consistent Charge density-functional based tight-binding (DFTB) method as implemented in the
deMonNano code1617 Periodic boundary conditions were applied to a (3x3x3) supercell thus
representing frameworks of the crystalline solid state Electronic density of states (DOS) have been
calculated using the DFTB+ code18 with k-point sampling where the k space was determined by
reaching convergence for the total energy according to the scheme proposed by Monkhorst and
Pack19
Hydrogen adsorption calculations have been carried out using QLDFT within the LIE-0 approximation
thus including many-body interparticle interactions and quantum effects implicitly through the
excess functional The PAF-H2 non-bonded interactions were represented by the classical Morse
atomic-pair potential Force field parameters were taken from Han et al20 who originally developed
them for studying hydrogen adsorption in COFs that incorporate similar building blocks as PAFs The
authors adopted the approach to the shapeless particle approximation of QLDFT These PAF-H2
parameters have been determined by fitting first-principles calculations at the second-order Moslashllerndash
Plesset (MP2) level of theory using the quadruple-ζ QZVPP basis set with correction for the basis set
superposition error (BSSE) QLDFT calculations use a grid spacing of 05 Bohr radii and a potential
cutoff of 5000 K
Results and Discussion
Design and Structure of PAFs
We have designed a set of PAFs by replacing C-C bonds in the diamond lattice by a set of organic
linkers phenyl biphenyl pyrene para-terphenyl perylenetetracarboxylic acid (PTCDA)
diphenylacetylene (DPA) and quaterphenyl (see Figure 1) We label the respective crystal structures
as PAF-phnl (PAF-301 in Ref 12) PAF-biph (PAF-302 in Ref 12) PAF-pyrn PAF-ptph(PAF-303 in Ref
12) PAF-PTCDA PAF-DPA and PAF-qtph (PAF-304 in Ref 12) respectively Such a design of
frameworks should result in materials with high stability due to the parent diamond-topology and
pure covalent bonding of the network The selected linkers differ in their length width and the
134
number of aromatic rings These should play an important role for hydrogen adsorption properties
aromatic systems exhibit large polarizabilities and interact with H2 molecules via London dispersion
forces Long linkers introduce high pore volume and low mas-weight to the network while wide
linkers offer large internal surface area and high heat of adsorption Hence long linkers are of
advantage for high gravimetric capacity while wide linkers enhance volumetric capacity Proper
optimization of the linker size should result in a perfect candidate for hydrogen storage applications
Figure 1 a) Schematic diagram of the topology of PAFs blue spheres and grey pillars represent carbon
tetragonal centers and organic linkers respectively b) Linkers used for the design of PAFs i) phenyl ii)
biphenyl iii) pyrene iv) DPA v) para-Terphenyl vi) PTCDA and vii) quaterphenyl
Selected structural and mechanical properties of the investigated PAF structures are given in Table 1
Frameworks created with the above mentioned linkers have mass densities that range from 085 g
cm-3 (for phenyl) to 01 g cm-3 (for quaterphenyl) The lowest mass-density obtained for a crystal
structure was for COF-108 with = 0171 g cm-39 Our results show that PAF-DPA and PAF-ptph have
mass densities close to the one of COF-108 while PAF-qtph with = 01 g cm-3 exhibits the lowest
for all the PAFs investigated in this study
While the large cell size and the small mass density of PAF-qtph are an advantage for high
gravimetric hydrogen storage capacity other PAFs (eg PAF-pyrn with wide linker) would
compromise gravimetric for high volumetric capacity As both of them are important for practical
applications a balance between them is crucial
Table 1 Calculated cell parameters a (a=b=c α=β=γ=60⁰) mass densities () formation energies (EForm) band
gaps () bulk moduli (B) and H2 accessible free volume and surface area of PAFs studied in the present work
In parenthesis are given HOMU-LUMO gaps of the corresponding saturated linkers
PAFs
a
(Aring)
ρ
(g cm-3)
EForm
(kJ mol-1)
Δ
(eV)
B
(GPa)
H2 accessible
free volume
H2 accessible
surface area
135
() (m2 g-1)
PAF-phnl 97 085 -121 47 (55) 360 35 2398
PAF-biphl 167 032 -122 36 (40) 132 73 5697
PAF-pyrn 166 042 -124 26 (28) 192 66 5090
PAF-DPA 210 019 -122 35 (37) 87 84 7240
PAF-ptph 237 016 -119 32 (33) 56 86 6735
PAF-PTCDA 236 024 -122 18 (19) 95 81 5576
PAF-qtphl 308 010 -119 29 (30) 35 91 7275
Energetic and Mechanical Properties
We have investigated energetic stability of PAFs by calculating their formation energies We regarded
the formation of PAFs as dehydrogenation reaction between saturated linkers and CH4 molecules
For a unit cell containing n carbon nodes and m organic linkers the formation energy (EForm) is given
by
( )
where Ecell EL and
are the total energies of the unit cell saturated linkers CH4 and H2
molecules respectively This excludes the inherent stability of linkers and represents the energy for
coordinating linkers with sp3 carbon atoms during diamond-topology formation All the formation
energies calculated in the present work are given in Table 1 Negative values indicate that the
formation of PAFs is exothermic The values per formula unit do not deviate significantly for different
PAF sizes and shapes
Although diamond is the hardest known material insertion of longer linkers diminishes its
mechanical strength to some extent In order to study the mechanical stability of PAFs we have
calculated their bulk moduli which is the second derivative of the cell energy with respect to the cell
volume (see Table 1) The highest value that we have obtained is for PAF-phnl with B = 36 GPa This is
over ten times smaller than the bulk modulus of pure diamond 514 GPa (calculated at the DFTB
level)21 and 442 GPa (experimental value)22 Bulk modulus should scale as 1V (V - volume) if all
bonds have the same strength We have plotted such a function for PAFs and other framework
136
materials23 in Figure 2 As PAFs have various types of CmdashC bonds there are minor deviations from
the perfect trend
Figure 2 Calculated bulk modulus B as function of the volume per atom PAFs are highlighted in blue and
compared with other framework materials Red curve denotes the fitting to 1V function (V - volume)
The stability of PAF-phnl is similar to that of Cu-BTC MOF and much higher than for other MOFs such
as MOF-5 -177 and -20524 and 3D COFs23 Subsequent addition of phenyl rings decreases B the
lowest value of 35 GPa has been found for PAF-qtph which is in the same order as for COF-108 In
general all the studied PAFs except for PAF-phnl have bulk moduli in line with 3D COFs25 When the
organic linker is extended in the width (cf PAF-biph and PAF-pyrn) the mechanical stability increases
Electronic Properties
All PAF structures are semiconductors similar to COFs The band gaps of PAFs range from 18 to 47
eV (see Table 1) Interesting trends may be observed from the band gaps of this iso-reticular series
In comparison with diamond which has a band gap of 55 eV it may be understood that subsequent
insertion of benzene groups between carbon atoms in diamond decreases the band gap This is easily
understood as the sp3 responsible for the semiconducting character become far apart with large
number of π-electrons in between which tend to close the gap More importantly the values of
band gaps are directly related to the HOMO-LUMO gaps of the corresponding saturated linkers
which are just slightly larger Also more extended linkers (cf PAF-biph and PAF-pyrn or PAF-ptph and
PAF-PTCDA) reduce the band gap
In general conjugated rings reduce the band gap more effectively than CmdashC bond networks (cf PAF-
DPA PAF-ptph or PAF-qtphl) The extreme cases for this behaviour are graphene which is metallic
137
and diamond which is an insulator Overall the band gap may be tuned by placing suitable linkers in
the diamond network Similar results have been reported for MOFs2627
We have calculated the electronic density of states (DOS) for all the PAF structures Figure 3 a shows
carbon DOS of PAFs arranged in such a way that the band gaps decrease going from the top to the
bottom of the figure PAFs with larger band gaps have larger population of energy states at the top of
valence band and bottom of conduction band whereas for linkers with smaller band gaps the
distribution of energy states is rather spread In order to study the impact of sp3 carbon atoms in the
DOS we plotted their contributions against the total carbon DOS in the cases of PAF-phnl and PAF-
pyrn as examples (see Figure 3 b and c) In both cases the sp3 carbons exhibit no energy states in the
band gap and in the close vicinity of band edges
Figure 3 (a) Carbon density of states of all calculated PAFs From top to bottom the value of band gap
decreases (b) and (c) projected DOS on sp3 C atoms of PAF-phnl and PAF-pyrn respectively The vertical
dashed line indicates Fermi level EF
Hydrogen Adsorption Properties
One of the potential applications of PAFs is hydrogen adsorption We have calculated the gravimetric
and volumetric capacities and analyzed them to understand the contributions of the linkers on the
138
hydrogen uptake in different range of temperatures and pressures H2 accessible free volume and
surface area of the PAFs are given in Table 1 In order to assess the excess adsorption capacity the
free pore volume is necessary In our simulation the free pore volume is defined to be that where
the H2-host interaction energy is less than 5000K This covers all sites where H2 is attracted by the
host structure and excludes the repulsion area close to the framework The solvent accessible surface
areas of the PAFs were determined by rolling a probe molecule along the van der Waals spheres of
the atoms A probe molecule of radius 148 Aring was used which is consistent with the Lennard-Jones
sphere of hydrogen and commonly used in various H2 molecular simulations28
Very large surface areas per mass unit have been observed for all PAFs except for PAF-phnl PAF-DPA
and PAF-qtph for example have 7240 and 7274 m2 g-1 H2 accessible surface area respectively For
comparison highly porous MOF-210 and NU-100 give values of 6240 and 6143 m2 g-1 BET surface
areas respectively determined from the experimental adsorption isotherms78 It is worth
mentioning that longer linkers expand the pore and increase the surface area per unit volume and
unit mass Wider linkers provide a higher surface area per unit volume however they possess larger
mass density and hence the surface area per unit mass gets lower
Figure 4 shows the gravimetric and volumetric total and excess adsorption isotherms of PAFs at 77K
The results show that the gravimetric (total and excess) H2 uptake is dependent on the linker length
The longest linker in the PAF-qtph structure gives the total and excess gravimetric capacity of 32 and
128 wt respectively which is larger than for highly porous crystalline 3D COFs29 These numbers
are calculated for the limit pressure of 100 bar For the shortest linker in PAF-phnl we have obtained
only around 25 wt for the same pressure Using grand canonical Monte-Carlo methods (GCMC)
Lan et al12 have predicted total and excess adsorption capacities of PAF-qtph of 224 and 84 wt
respectively The deviations in results are attributed to the differences in both methods where
different force fields are used to describe atom-atom interactions
The volumetric adsorption capacities lower with the size of the linker and the largest uptake we have
found for the PAF-pyrn (46 and 35 kg m-3 total and excess respectively) while very low values were
found for PAF-qtph (30 and 145 kg m-3 total and excess respectively) at 35-bar pressure It could be
predicted that for PAF-phnl the volumetric adsorption reaches the larges values however due to its
very compact crystal structure it reaches saturation at the low-pressure region and does not exceed
30 kg m-3 for the total uptake From the excess adsorption isotherms however PAF-phnl is the best
adsorbing system for low-pressure application as it gets saturated at 11 bar by adsorbing 266 kg m-3
of H2 Generally we have observed that PAFs with lower pore sizes exhibited saturation of volumetric
capacities at lower pressures
139
Figure 4 Calculated H2 uptake of PAFs at 77 K (a)-(b) gravimetric and (c)-(d) volumetric total (upper panel)
and excess (lower panel) respectively
We have also calculated the adsorption performance of PAFs at room temperature The gravimetric
total adsorption isotherms are shown in Figure 5 Similar to the results obtained for T=77 K PAF-
qtph exhibits the highest total gravimetric capacity at room temperature which is as high as 732 wt
at 100 bar Lan et al12 have predicted the uptake of 653 wt at 100 bar using GCMC simulations
These results exceed the 2015 DOE target of 55 wt 3031 On the other hand at more applicable
pressure of 5 bar the uptake reaches only 05 wt This suggests however that the delivery amount
(approximately the difference between the uptakes at 100 and 5 bar) is still more than the DOE
target Phenyl linker at room temperature does not exceed the gravimetric uptake of 1 wt at 100
bar
Figure 5 Calculated gravimetric total (left) and excess (right) H2 uptake of PAFs at 300 K
140
At 300K excess gravimetric capacity is rather low and does not exceed 075 wt even for large
pressure (see Figure 5)
Effects of interpenetration
Very often frameworks with large pores crystallize as interpenetrated lattices In most cases this is
an undesired fact due to reduction of the pore size and free volume For instance COF-300 which
has diamond topology was found to have 5-interpenetrated frameworks32
We have studied the effect of interpenetration in the case of PAF-qtph which has the largest pore
volume among the materials in this study Without any steric hindrance PAF-qtph may be
interpenetrated up to the order of four The two three and four interpenetrated networks are
named as PAF-qtph-2 PAF-qtph-3 and PAF-qtph-4 respectively and in each case the interpenetrated
structures possess translational symmetry Table 2 shows calculated mass densities and H2 accessible
free volume and surface area of non-interpenetrated and n-fold penetrated PAF-qtph Obviously the
mass density increases by one unit for each degree of interpenetration PAF-qtph has 91 of its
volume accessible for H2 which means that 9 of volume is occupied by the skeleton of the PAF
Hence each degree of interpenetration decreases the free volume by 9 H2 accessible surface area
per unit mass remains the same up to 3-fold interpenetration However PAF-qtph-4 results in much
less accessibility for H2
Table 2 Calculated mass densities () and H2 accessible free volume and surface area of non-interpenetrated
and n-fold interpenetrated PAF-qtph where n = 2 3 4
PAF
(g cm-3)
H2 accessible
free volume ()
H2 accessible
surface area
(m2 g-1)
PAF-qtph 010 91 7275
PAF-qtph-2 020 82 7275
PAF-qtph-3 030 73 7275
PAF-qtph-4 040 64 5998
Figure 6 shows the excess and total adsorption isotherms (gravimetric and volumetric) of non-
interpenetrated and interpenetrated PAF-qtph at 77 K With the increasing degree of
141
interpenetration saturation occurs at lower values of pressure This is due to the smaller pore size
resulted from the high-fold interpenetration The excess gravimetric capacity decreases by 18 - 2 wt
per each n-fold interpenetration a result of the decreasing free space around the skeleton It is to be
noted that the difference between the total and the excess adsorption capacities of PAF-qtph is quite
large however it decreases less for interpenetrated structures This is because the interpenetrated
frameworks occupy the free space that otherwise would be taken by hydrogen to increase the total
capacity but not the excess
Figure 6 Calculated H2 uptake at 77 K of non-interpenetrated and n-fold interpenetrated PAF-qtph with n = 2
3 4 (a)-(b) gravimetric and (c)-(d) volumetric total (lower panel) and excess (upper panel) respectively
On the other hand the interpenetrated frameworks have larger volumetric capacity what is easily
understandable due to the volume reduction Significant increase in excess volumetric capacity has
been obtained for PAF-qtph-2 in comparison with PAF-qtph whereas only slightly lower increase was
obtained for PAF-qtph-3 in comparison with PAF-qtph-2 (see Figure 6) PAF-qtph-4 exhibited even
lower increase compared to PAF-qtph-3 suggesting that only a certain degree of interpenetration is
appreciable Although non-interpenetrated PAF has a large amount of H2 gas in the bulk phase due
to the high pore volume its total volumetric uptake does not exceed that of the interpenetrated
PAFs
Almost linear gravimetric isotherms were obtained at room temperature for all studied PAFs
including the interpenetrated cases (see Figure 7) Large decrease of gravimetric capacity was noted
142
when the PAF was interpenetrated from around 7 wt down to 2 wt for 4-fold interpenetrated
PAF-qtph The excess gravimetric capacity however is the same independent of the n-fold
interpenetration and maximum of 06 wt was found for 100-bar pressure (see Figure 7)
Figure 7 Calculated gravimetric total (left) and excess (right) H2 uptake of non-interpenetrated and n-fold
interpenetrated PAF-qtph (n = 2 3 4) at 300 K
Conclusions
Following diamond-topology we have designed a set of porous aromatic frameworks PAFs by
replacing CmdashC bonds by various organic linkers what resulted in remarkably high surface area and
pore volume
Exothermic formation energies of PAFs calculated from the dehydrogenation of the linkers with CH4
indicate strong coordination of linkers in the diamond topology The studied PAFs exhibit bulk moduli
that are much smaller than diamond however in the same order as other porous frameworks such
as MOFs and COFs PAFs are semi-conductors with their band gaps dependent on the HOMO-LUMO
gaps of the linking molecules
Using quantized liquid density functional theory which takes into account inter-particle interactions
and quantum effects we have predicted hydrogen uptake capacities in PAFs At room temperature
and 100 bar the predicted uptake in PAF-qtph was 732 wt which exceeded the 2015 DOE target
At the same conditions other PAFs showed significantly lower uptakes At 77 K the PAFs with similar
pore sizes exhibited similar gravimetric uptake capacities whereas smaller pore size and larger
number of aromatic rings result in higher volumetric capacities In smaller pores saturation of excess
capacities occurred at lower pressures Interpenetrated frameworks largely decrease the amount of
hydrogen gas in the pores and increase the weight of the material however they are predicted to
have high volumetric capacities
143
References
(1) Eddaoudi M Moler D B Li H L Chen B L Reineke T M OKeeffe M Yaghi O M
Accounts of Chemical Research 2001 34 319
(2) Seo J S Whang D Lee H Jun S I Oh J Jeon Y J Kim K Nature 2000 404 982
(3) Ferey G Mellot-Draznieks C Serre C Millange F Accounts of Chemical Research 2005 38
217
(4) Yaghi O M OKeeffe M Ockwig N W Chae H K Eddaoudi M Kim J Nature 2003 423
705
(5) Eddaoudi M Kim J Rosi N Vodak D Wachter J OKeeffe M Yaghi O M Science 2002
295 469
(6) Cote A P Benin A I Ockwig N W OKeeffe M Matzger A J Yaghi O M Science 2005
310 1166
(7) Furukawa H Ko N Go Y B Aratani N Choi S B Choi E Yazaydin A O Snurr R Q
OKeeffe M Kim J Yaghi O M Science 2010 329 424
(8) Farha O K Yazaydin A O Eryazici I Malliakas C D Hauser B G Kanatzidis M G
Nguyen S T Snurr R Q Hupp J T Nature Chemistry 2010 2 944
(9) El-Kaderi H M Hunt J R Mendoza-Cortes J L Cote A P Taylor R E OKeeffe M Yaghi
O M Science 2007 316 268
(10) Ben T Ren H Ma S Cao D Lan J Jing X Wang W Xu J Deng F Simmons J M Qiu
S Zhu G Angewandte Chemie-International Edition 2009 48 9457
(11) Yuan Y Sun F Ren H Jing X Wang W Ma H Zhao H Zhu G Journal of Materials
Chemistry 2011 21 13498
(12) Lan J Cao D Wang W Ben T Zhu G Journal of Physical Chemistry Letters 2010 1 978
(13) Oliveira A F Seifert G Heine T Duarte H A Journal of the Brazilian Chemical Society
2009 20 1193
(14) Seifert G Porezag D Frauenheim T International Journal of Quantum Chemistry 1996 58
185
(15) Patchkovskii S Heine T Physical Review E 2009 80
(16) Heine T Rapacioli M Patchkovskii S Frenzel J Koester A M Calaminici P Escalante S
Duarte H A Flores R Geudtner G Goursot A Reveles J U Vela A Salahub D R deMon-nano edn ed
deMon 2009
(17) Zhechkov L Heine T Patchkovskii S Seifert G Duarte H A Journal of Chemical Theory
and Computation 2005 1 841
(18) BCCMS Bremen DFTB+ - Density Functional based Tight binding (and more)
(19) Monkhorst H J Pack J D Physical Review B 1976 13 5188
(20) Han S S Furukawa H Yaghi O M Goddard III W A Journal of the American Chemical
Society 2008 130 11580
(21) Kuc A Seifert G Physical Review B 2006 74
(22) Cohen M L Physical Review B 1985 32 7988
(23) Lukose B Kuc A Heine T manuscript in preparation 2012
(24) Lukose B Supronowicz B Petkov P S Frenzel J Kuc A Seifert G Vayssilov G N
Heine T physica status solidi (b) 2011
(25) Amirjalayer S Snurr R Q Schmid R Journal of Physical Chemistry C 2012 116 4921
(26) Gascon J Hernandez-Alonso M D Almeida A R van Klink G P M Kapteijn F Mul G
Chemsuschem 2008 1 981
(27) Kuc A Enyashin A Seifert G Journal of Physical Chemistry B 2007 111 8179
(28) Dueren T Millange F Ferey G Walton K S Snurr R Q Journal of Physical Chemistry C
2007 111 15350
(29) Furukawa H Yaghi O M Journal of the American Chemical Society 2009 131 8875
144
(30) US DOE Office of Energy Efficiency and Renewable Energy and The FreedomCAR and
Fuel Partnership 2009
httpwww1eereenergygovhydrogenandfuelcellsstoragepdfstargets_onboard_hydro_storage_explanatio
npdf
(31) US DOE USCAR Shell BP ConocoPhillips Chevron Exxon-Mobil T F a F P Multi-Year
Research Development and Demonstration Plan 2009
httpwww1eereenergygovhydrogenandfuelcellsmypppdfsstoragepdf
(32) Uribe-Romo F J Hunt J R Furukawa H Klock C OKeeffe M Yaghi O M Journal of the
American Chemical Society 2009 131 4570
145
Appendix G
A novel series of isoreticular metal organic frameworks realizing metastable structures by liquid phase epitaxy
Jinxuan Liu Binit Lukose Osama Shekhah Hasan Kemal Arslan Peter Weidler Hartmut
Gliemann Stefan Braumlse Sylvain Grosjean Adelheid Godt Xinliang Feng Klaus Muumlllen Ioan-
Bogdan Magdau Thomas Heine and Christof Woumlll
Prepared for publication
Highly crystalline molecular scaffolds with nm-sized pores offer an attractive basis for the fabrication
of novel materials The scaffolds alone already exhibit attractive properties eg the safe storage of
small molecules like H2 CO2 and CH41-4 In addition to the simple release of stored particles changes
in the optical and electronic properties of these nanomaterials upon loading their porous systems
with guest molecules offer interesting applications eg for sensors5 or electrochemistry6 For the
construction of new nanomaterials the voids within the framework of nanostructures may be loaded
with nm-sized objects such as inorganic clusters larger molecules and even small proteins a
process that holds great potential as for example in drug release7-8 or the design of novel battery
materials9 Furthermore meta-crystals may be prepared by loading metal nanoparticles into the
pores of a three-dimensional scaffold to provide materials with a number of attractive applications
ranging from plasmonics10-12 to fundamental investigations of the electronic structure and transport
properties of the meta-crystals13
146
In the last two decades numerous studies have shown that MOFs also termed porous coordination
polymers (PCPs) fulfill many of the aforementioned criteria Although originally developed for the
storage of hydrogen eg in tanks of automobiles MOFs have recently found more technologically
advanced applications as sensors5 matrices for drug release7-8 and as membranes for enantiomer
separation14 and for proton conductance in fuel cells15 Although pore-sizes available within MOFs1
are already sufficiently large to host metal-nanoclusters such as Au5516-17 the actual realization of
meta-crystals requires in addition to structural requirements a strategy for the controlled loading
of the nanoparticles (NPs) into the molecular framework Simply adding NPs to the solutions before
starting the solvothermal synthesis of MOFs has been reported18 but this procedure does not allow
for controlled loading The layer-by-layer growing process of SUFMOFs allows the uptake of
nanosized objects during synthesis including the fabrication of compositional gradients of different
NPs Those guest NPs can range from pure metal metal oxide or covalent clusters over one-
dimensional objects such as nanowires carbon or inorganic nanotubes to biological molecules such
as drugs or even small proteins If the loading happens during synthesis alternating layers of
different NPs can be realized
The introduction of procedures for epitaxial growth of MOFs on functionalized substrates19 was a
major step towards controlled functionalization of frameworks Epitaxial synthesis allowed the
preparation of compositional gradients20-21 and provided a new strategy to load nano-objects into
predefined pores
Unfortunately the LPE process has so far been only demonstrated for a fairly small number of
MOFs141922-23 none of which exhibits the required pore sizes In addition for the fabrication of meta-
crystals the architecture of the network should be sufficiently adjustable to realize pores of different
sizes There should also be a straightforward way to functionalize the framework itself in order to
tailor the interaction of the walls with the targeted guest objects Ideally such a strategy would be
based on an isoreticular series of MOFs in which the pore size can be determined by a choice from a
homologous series of ligands with different lengths1
Here we present a novel isoreticular MOF series with the required flexibility as regards pore-sizes
and functionalization which is well suited for the LPE process This class denoted as SURMOF-2 is
derived from MOF-2 one of the simplest framework architectures24-25 MOF-2 is based on paddle-
wheel (pw) units formed by attaching 4 dicarboxylate groups to Cu++ or Zn++-dimers yielding planar
sheets with 4-fold symmetry (see Fig 1) These planes are held together by strong
carboxylatemetal- bonds26 Conventional solvothermal synthesis yields stacks of pw-planes shifted
relative to each other with a corresponding reduction in symmetry to yield a P2 or C2 symmetry2427-
28
147
The relative shifts between the pw-planes can be avoided when using the recently developed liquid
phase epitaxy (LPE) method22 As demonstrated by the XRD-data shown in Fig 2 for a number of
different dicarboxylic acid ligands the LPE-process yields surface attached metal organic frameworks
(SURMOFs) with P4 symmetry29 except for the 26-NDC ligand which exhibits a P2 symmetry as a
result of the non-linearity of the carboxyl functional groups on the ligand The ligand PPDC
pentaphenyldicarboxylic acid with a length of 25 nm is to our knowledge the longest ligand which
has so far been used successfully for a MOF synthesis303030 A detailed analysis of the diffraction data
allowed us to propose the structures shown in Fig 1 This novel isoreticular series of MOFs hereafter
termed SURMOF-2 also consists of pw-planes which ndash in contrast to MOF-2 ndash are stacked directly
on-top of each other This absence of a lateral shift when stacking the pw-planes yields 1d-pores of
quadratic cross section with a diagonal up to 4 nm (see Fig 2) Importantly for the SURMOF-2 series
interpenetration is absent For many known isoreticular MOF series the formation of larger and
larger pores is limited by this phenomenon if the pores become too large a second or even a third
3d-lattice will be formed inside the first lattice yielding interpenetrated networks which exhibit the
expected larger lattice constant but with rather small pore sizes31-33 the hosting of NPs thus becomes
impossible For SURMOF-2 the particular topology with 1d-pores of quadratic cross section is not
compatible with the presence of a second interwoven network and as a result interpenetration is
suppressed
Although the solubility of some the long dicarboxylic acids ligands used for the SURMOF-2 fabrication
(TPDC QPDC and PPDC see Fig 1) is fairly small we encountered no problems in the LPE process
since already small concentrations of dicarboxylic acids are sufficient for the formation of a single
monolayer on the substrate the elementary step of the LPE synthesis process34-35 Even with the
longest dicarboxylic acid available to us PPDC growth occurred in a straightforward fashion and
optimization of the growth process was not necessary
The P4 structures proposed for the SURMOF-2 series except for the 26-NDC ligand differ markedly
from the bulk MOF-2 structures obtained from conventional solvothermal synthesis2427-28 To
understand this unexpected difference and in particular the absence of relative shifts between the
pw-planes in the proposed SURMOF-2 structure we performed extensive quantum chemical
calculations employing approximate density-functional theory (DFT) in this case London dispersion-
corrected self-consistent charge density-functional based tight-binding (DFTB)36-38 to a periodic
model of MOF-2 and its SURMOF derivatives
Periodic structure DFTB calculations (Fig S1) confirmed that the P2m structure proposed by Yaghi
et al for the bulk MOF-2 material24 is indeed the energetically favorable configuration for MOF-2
while the C2m structure that has also been reported for bulk MOF-227-28 is slightly higher in energy
148
(41 meV per formula unit see Table 1) The optimum interlayer distance between the pw-planes in
the P4 structure is 58 Aring in very good agreement with the experimental value of 56 Aring (as obtained
from the in-plane XRD-data see Fig S2) It is important to note that at that distance the linkers
cannot remain in rectangular position as in MOF-2 but are twisted in order to avoid steric hindrance
and to optimize linker-linker interactions
The P4mmm structure as proposed for SURMOF-2 is less stable by 200 meV per formula unit as
compared to the P2m configuration Although the crystal energy for the P4 stacking is substantially
smaller than for P2 packing simulated annealing (DFTB model using Born-Oppenheimer Molecular
Dynamics) carried out at 300K for 10 ps demonstrated that the P4mmm structure corresponds to a
local minimum This observation is confirmed by the calculation of the stacking energy of MOF-2
where a shifting of the layers is simulated along the P2-P4-C2 path (Figure 3) On the basis of these
calculations we thus propose that SURMOF-2 adopts this metastable P4 structure
In Table 1 the interlayer interaction energies are summarized They range from 590 meV per formula
unit with respect to fully dissociated MOF-2 layers for Cu2(bdc)2 and successively increase for longer
linkers of the same type reaching 145 eV per formula unit for Cu2(ppdc)2 indicating that the linkers
play an important role in the overall stabilization of SURMOF-2 with large pore sizes Even stronger
interlayer interactions are found for different linker topologies (PPDC) A detailed computational
analysis of the role of the individual building blocks of SURMOF-2 for the crystal growth and
stabilization will be published elsewhere
The flat well-defined organic surfaces used as the templating (or nucleating) substrate in the LPE
growth process provide a satisfying explanation for why SURMOF-2 grows with the highly
symmetrical P4 structure instead of adopting the energetically more favorable bulk P2 symmetry3439
The carboxylic acid groups exposed on the surface force the first layer of deposited Cu-dimers into a
coplanar arrangement which is not compatible with the bulk P2 (or C2) structure but rather
nucleates the P4 packing In turn the carboxylate-groups in the first layer of deposited dicarboxylic
acids provide the same constraints for the next layer (see Fig 1) As a result of the layer-by-layer
method employed for further SURMOF-2 growth the same boundary conditions apply for all
subsequent layers The organic surface thus acts as a seed structure to yield the metastable P4
packing not an unusual motif in epitaxial growth40
The calculations allow us to predict that it will be possible to grow SURMOF structures with even
larger linker molecules as used here The stacking will further stabilize the P4 symmetry as the
interaction energy per formula unit is more and more dominated by the linkers (Table 1) At present
149
we cannot predict the maximum possible size of the pores but linker PPDC (see Fig 2) is thus far
unmatched as a component in non-interpenetrated framework structures
Acknowledgement
We thank Ms Huumllsmann Bielefeld University for the preparation of P(EP)2DC Financial support by
Deutsche Forschungsgemeinschaft (DFG) within the Priority Program Metal-Organic Frameworks
(SPP 1362) is gratefully acknowledged
Methods
Computational Details
All structures were created using a preliminary version of our topological framework creator
software which allows the creation of topological network models in terms of secondary building
units and their replacement by individual molecules to create the coordinates of virtually any
framework material The generated starting coordinates including their corresponding lattice
parameters have been hierarchically fully optimized using the Universal Force Field (UFF)41 followed
by the dispersion-corrected self-consistent-charge density-functional based tight-binding (DFTB)
method36-3842-43 that we have recently validated against experimental structures of HKUST-1 MOF-5
MOF-177 DUT-6 and MIL-53(Al) as well as for their performance to compute the adsorption of
water and carbon monoxide37 For all calculations we employed the deMonNano software44444444
We have chosen periodic boundary conditions for all calculations and the repeated slab method has
been employed to compute the properties of the single layers in order to evaluate the stacking
energy A conjugate-gradient scheme was employed for geometry optimization of atomic
coordinates and cell dimensions The atomic force tolerance was set to 3 x 10-4 eVAring
The confined conditions at the SURMOF-substrate interface was modeled by fixing the corresponding
coordinate in the computer simulations All calculated structures have been substantiated by
simulated annealing calculations where a Born-Oppenheimer Molecular Dynamics trajectory at 300K
has been computed for 10 ps without geometry constrains All structures remained in P4 topology
Experimental methods
The SURMOF-2 samples were grown on Au substrates (100-nm Au5-nm Ti deposited on Si wafers)
using a high-throughput approach spray method45 The gold substrates were functionalized by self-
assembled monolayers SAMs of 16-mercaptohexadecanoic acid (MHDA) These substrates were
mounted on a sample holder and subsequently sprayed with 1 mM of Cu2 (CH3COO)4middotH2O ethanol
solution and ethanolic solution of SURMOF-2 organic linkers (01 mM of BDC 26-NDC BPDC and
150
saturated concentration of TPDC QPDC PPDC P(EP)2DC) at room temperature After a given
number of cycles the samples were characterized with X-ray diffraction (XRD)
Figure 1 Schematic representation of the synthesis and formation of the SURMOF-2 analogues
151
Figure 2 Out of plane XRD data of Cu-BDC Cu-26-NDC Cu-BPDC Cu-TPDC Cu-QPDC Cu-P(EP)2DC and Cu-PPDC (upper left) schematic representation (upper right) and proposed structures of SURMOF-2 analogues (lower panel) All the SURMOF-2 are grown on ndashCOOH terminated SAM surface using the LPE method
152
Figure 3 Energy for the relative shift of each layer in two different directions horizontal (P4 to P2) and diagonal (P4 to C2) at a fixed interlayer distance of 56 Aring Structures have been partially optimized and energies are given per formula unit with respect to fully dissociated planes
Supporting information
Synthesis of organic linkers
(1) para-terphenyldicarboxylic acid (TPDC)
To a solution of para-terphenyl (1500 g 651 mmol 1 eq) and oxalyl chloride (3350 mL 3900 mmol
6 eq) in carbon disulfide (30 mL) at 0degC (ice bath) was added aluminium chloride (1475 g 1106
mmol 17 eq) After 1h stirring additional aluminium chloride (0870 g 651 mmol 1 eq) was added
the ice bath was removed and the reaction mixture was stirred at room temperature for 24h The
mixture was poured into crushed ice and stirred until the dark-brown mixture turned to yellow
Carbon disulfide was removed under reduced pressure and the resulting aqueous suspension was
filtered The solid was washed with diluted hydrochloric acid then with diethyl ether and dried
under vacuum overnight with an oil bath at 50degC Pale yellow solid yield 2028 g (98)
(2) para-quaterphenyldicarboxylic acid (QPDC)
153
To a solution of para-quaterphenyl (1000 g 326 mmol 1 eq) and oxalyl chloride (1680 mL 1956
mmol 6 eq) in carbon disulfide (20 mL) at 0degC (ice bath) was added aluminium chloride (0740 g 555
mmol 17 eq) After 1h stirring additional aluminium chloride (0435 g 326 mmol 1 eq) was added
the ice bath was removed and the reaction mixture was stirred at room temperature for 24h The
mixture was poured into crushed ice and stirred until the dark-brown mixture turned to yellow
Carbon disulfide was removed under reduced pressure and the resulting aqueous suspension was
filtered The solid was washed with diluted hydrochloric acid then with diethyl ether and dried
under vacuum overnight with an oil bath at 50degC Yellow solid yield 1186 g (92)
(3) P(EP)2DC
The synthesis of the P(EP)2DC-linker has been described in Ref 46
(4) para-pentaphenly dicarboxylic acid (PPDC)
Scheme 1 Synthesis of para-pentaphenly dicarboxylic acid i) Pd(PPh3)4 Na2CO3 toluenedioxane H2O 85oC ii) MeOH 6M NaOH HCl
para-pentaphenly dicarboxylic acid (H2L) was synthesized via Suzuki coupling of 44rdquo-dibromo-p-
terphenyl and 4-methoxycarbonylphenylboronic acid (Scheme 1) 44rdquo-dibromo-p-terphenyl (776 mg
200 mmol) 4-methoxycarbonylphenylboronic acid (108 g 60 mmol) and Na3CO3 (170 g 160 mmol)
were added to a degassed mixture of toluene14-dioxane (50 mL 221) under Ar [Pd(PPh3)4] (116
mg 010 mmol) was added to the mixture and heated to 85 degC for 24 hours under Ar The reaction
mixture was cooled to room temperature The precipitate was collected by filtration washed with
water methanol and used for next reaction without further purification The final product H4L was
obtained by hydrolysis of the crude product under reflux in methanol (100 mL) overnight with 6M
aqueous NaOH (20 mL) followed by acidification with HCl (conc) After removal of methanol the
final product was collected by filtration as a grey solid (078 g 83 yield) 1H NMR (d6-DMSO
250MHz 298K) δ (ppm) 805 (d J = 75Hz 4H) 789-783 (m 12H) 750 (d J = 75Hz 4H) 13C NMR
cannot be clearly resolved due to poor solubility MALDI-TOF-MS (TCNQ as matrix) mz 47002
cacld 47051 Elemental analysis Calculated C 8169 H 471 Found C 8163 H 479
Br Br MeOOC B
OH
OH
+
COOMe
COOMe
COOH
COOH
i ii
154
Figure S1 MOF-2 in P4 (as reported) P2 and C2 symmetry
155
Figure S2 In-plane XRD data of Cu-BDC Cu-26-NDC Cu-BPDC Cu-TPDC Cu-QPDC and Cu-P(EP)2DC All the
SURMOF-2 are grown on ndashCOOH terminated SAM surface using the LPE method The position of (010) plane
represents the layer distance
Table 1 Stacking energy and geometries of P4 SURMOF-2 derivatives
Symmetry a= c b Stacking Energy
Cu2(bdc)2 C2 1119 50 -076
Cu2(bdc)2 P2 1119 54 -08
Cu2(bdc)2 P4 1119 58 -059
156
Cu2(ndc)2 P2 1335 56 -04
Cu2(bpdc)2 P4 1549 59 -068
Cu2(tpdc)2 P4 1984 59 -091
Cu2(qpdc)2 P4 2424 59 -121
Cu2(P(EP)2DC)2 P4 2512 52 -173
Cu2(ppdc)2 P4 2859 59 -145
Stacking energies (DFTB level in eV) of SURMOFs The energies were calculated within periodic
boundary conditions and are given per formula unit
References
1 Eddaoudi M et al Systematic design of pore size and functionality in isoreticular MOFs and
their application in methane storage Science 295 469-472 (2002)
2 Rosi N L et al Hydrogen storage in microporous metal-organic frameworks Science 300
1127-1129 (2003)
3 Rowsell J L C amp Yaghi O M Metal-organic frameworks a new class of porous materials
Microporous and Mesoporous Materials 73 3-14 (2004)
4 Wang B Cote A P Furukawa H OKeeffe M amp Yaghi O M Colossal cages in zeolitic
imidazolate frameworks as selective carbon dioxide reservoirs Nature 453 207-U206 (2008)
5 Lauren E Kreno et al Metal-Organic Framework Materials as Chemical Sensors Chemical
Reviews 112 1105-1124 (2012)
6 Dragasser A et al Redox mediation enabled by immobilised centres in the pores of a metal-
organic framework grown by liquid phase epitaxy Chemical Communications 48 663-665
(2012)
7 Horcajada P et al Metal-organic frameworks as efficient materials for drug delivery
Angewandte Chemie-International Edition 45 5974-5978 (2006)
8 Horcajada P et al Flexible porous metal-organic frameworks for a controlled drug delivery
Journal of the American Chemical Society 130 6774-6780 (2008)
9 Hurd J A et al Anhydrous proton conduction at 150 degrees C in a crystalline metal-organic
framework Nature Chemistry 1 705-710 (2009)
10 Kreno L E Hupp J T amp Van Duyne R P Metal-Organic Framework Thin Film for Enhanced
Localized Surface Plasmon Resonance Gas Sensing Analytical Chemistry 82 8042-8046
(2010)
11 Lu G et al Fabrication of Metal-Organic Framework-Containing Silica-Colloidal Crystals for
Vapor Sensing Advanced Materials 23 4449-4452 (2011)
157
12 Lu G amp Hupp J T Metal-Organic Frameworks as Sensors A ZIF-8 Based Fabry-Perot Device
as a Selective Sensor for Chemical Vapors and Gases Journal of the American Chemical
Society 132 7832-7833 (2010)
13 Mark D Allendorf Adam Schwartzberg Vitalie Stavila amp Talin A A Roadmap to
Implementing MetalndashOrganic Frameworks in Electronic Devices Challenges and Critical
Directions European Journal of Chemistry (2011)
14 Bo Liu et al Enantiopure Metal-Organic Framework Thin Films Oriented SURMOF Growth
and Enantioselective Adsorption Angewandte Chemie-International Edition 51 817-810
(2012)
15 Sadakiyo M Yamada T amp Kitagawa H Rational Designs for Highly Proton-Conductive
Metal-Organic Frameworks Journal of the American Chemical Society 131 9906-9907 (2009)
16 Ishida T Nagaoka M Akita T amp Haruta M Deposition of Gold Clusters on Porous
Coordination Polymers by Solid Grinding and Their Catalytic Activity in Aerobic Oxidation of
Alcohols Chemistry-a European Journal 14 8456-8460 (2008)
17 Esken D Turner S Lebedev O I Van Tendeloo G amp Fischer R A AuZIFs Stabilization
and Encapsulation of Cavity-Size Matching Gold Clusters inside Functionalized Zeolite
Imidazolate Frameworks ZIFs Chemistry of Materials 22 6393-6401 (2010)
18 Lohe M R et al Heating and separation using nanomagnet-functionalized metal-organic
frameworks Chemical Communications 47 3075-3077 (2011)
19 Shekhah O et al Step-by-step route for the synthesis of metal-organic frameworks Journal
of the American Chemical Society 129 15118-15119 (2007)
20 Furukawa S et al A block PCP crystal anisotropic hybridization of porous coordination
polymers by face-selective epitaxial growth Chemical Communications 5097-5099 (2009)
21 Shekhah O et al MOF-on-MOF heteroepitaxy perfectly oriented [Zn(2)(ndc)(2)(dabco)](n)
grown on [Cu(2)(ndc)(2)(dabco)](n) thin films Dalton Transactions 40 4954-4958 (2011)
22 Shekhah O et al Controlling interpenetration in metal-organic frameworks by liquid-phase
epitaxy Nature Materials 8 481-484 (2009)
23 Zacher D et al Liquid-Phase Epitaxy of Multicomponent Layer-Based Porous Coordination
Polymer Thin Films of [M(L)(P)05] Type Importance of Deposition Sequence on the Oriented
Growth Chemistry-a European Journal 17 1448-1455 (2011)
24 Li H Eddaoudi M Groy T L amp Yaghi O M Establishing microporosity in open metal-
organic frameworks Gas sorption isotherms for Zn(BDC) (BDC = 14-benzenedicarboxylate)
Journal of the American Chemical Society 120 8571-8572 (1998)
25 Mueller U et al Metal-organic frameworks - prospective industrial applications Journal of
Materials Chemistry 16 626-636 (2006)
158
26 Shekhah O Wang H Zacher D Fischer R A amp Woumlll C Growth Mechanism of Metal-
Organic Frameworks Insights into the Nucleation by Employing a Step-by-Step Route
Angewandte Chemie-International Edition 48 5038-5041 (2009)
27 Carson C G et al Synthesis and Structure Characterization of Copper Terephthalate Metal-
Organic Frameworks European Journal of Inorganic Chemistry 2338-2343 (2009)
28 Clausen H F Poulsen R D Bond A D Chevallier M A S amp Iversen B B Solvothermal
synthesis of new metal organic framework structures in the zinc-terephthalic acid-dimethyl
formamide system Journal of Solid State Chemistry 178 3342-3351 (2005)
29 Arslan H K et al Intercalation in Layered Metal-Organic Frameworks Reversible Inclusion of
an Extended pi-System Journal of the American Chemical Society 133 8158-8161 (2011)
30 The MOF with the largest pore size recorded so far MOF-200 (Furukawa H et al Ultrahigh
Porosity in Metal-Organic Frameworks Science 329 424-428 (2010)) used a (trivalent)
444-(benzene-135-triyl-tris(benzene-41-diyl))tribenzoate (BBC) ligand The carboxylic
acid-to carboxylic acid distance is 20 nm compared to 25 nm in case of PPDC The cage size
in MOF-200 amounts to 18 nm by 28 nm clearly smaller than the 1d-channels in the PPDC
SURMOF-2 that are 28 nm by 28 nm
31 Batten S R amp Robson R Interpenetrating nets Ordered periodic entanglement
Angewandte Chemie-International Edition 37 1460-1494 (1998)
32 Snurr R Q Hupp J T amp Nguyen S T Prospects for nanoporous metal-organic materials in
advanced separations processes Aiche Journal 50 1090-1095 (2004)
33 Yaghi O M A tale of two entanglements Nature Materials 6 92-93 (2007)
34 Shekhah O Liu J Fischer R A amp Woumlll C MOF thin films existing and future applications
Chemical Society Reviews 40 1081-1106 (2011)
35 Zacher D Shekhah O Woumlll C amp Fischer R A Thin films of metal-organic frameworks
Chemical Society Reviews 38 1418-1429 (2009)
36 Elstner M et al Self-consistent-charge density-functional tight-binding method for
simulations of complex materials properties Physical Review B 58 7260-7268 (1998)
37 Lukose B et al Structural properties of metal-organic frameworks within the density-
functional based tight-binding method Physica Status Solidi B-Basic Solid State Physics 249
335-342 (2012)
38 Zhechkov L Heine T Patchkovskii S Seifert G amp Duarte H A An efficient a Posteriori
treatment for dispersion interaction in density-functional-based tight binding Journal of
Chemical Theory and Computation 1 841-847 (2005)
159
39 Zacher D Schmid R Woumlll C amp Fischer R A Surface Chemistry of Metal-Organic
Frameworks at the Liquid-Solid Interface Angewandte Chemie-International Edition 50 176-
199 (2011)
40 Prinz G A Stabilization of bcc Co via Epitaxial Growth on GaAs Physical Review Letters 54
1051-1054 (1985)
41 Rappe A K Casewit C J Colwell K S Goddard W A amp Skiff W M UFF a full periodic
table force field for molecular mechanics and molecular dynamics simulations Journal of the
American Chemical Society 114 10024-10035 (1992)
42 Seifert G Porezag D amp Frauenheim T Calculations of molecules clusters and solids with a
simplified LCAO-DFT-LDA scheme International Journal of Quantum Chemistry 58 185-192
(1996)
43 Oliveira A F Seifert G Heine T amp Duarte H A Density-Functional Based Tight-Binding an
Approximate DFT Method Journal of the Brazilian Chemical Society 20 1193-1205 (2009)
44 deMonNano v 2009 (Bremen 2009)
45 Arslan H K et al High-Throughput Fabrication of Uniform and Homogenous MOF Coatings
Adv Funct Mater 21 4228-4231 (2011)
46 Schaate A et al Porous Interpenetrated Zirconium-Organic Frameworks (PIZOFs) A
Chemically Versatile Family of Metal-Organic Frameworks Chemistry-a European Journal 17
9320-9325 (2011)
160
Appendix H
Linker guided metastability in templated Metal-Organic Framework-2 derivatives (SURMOFs-2)
Binit Lukose Gabriel Merino Christof Woumlll and Thomas Heine
Prepared for publication
INTRODUCTION
The molecular assembly of metal-oxides and organic struts can provide a large number of network
topologies for Metal-Organic Frameworks (MOFs)1-3 This is attained by the differences in
connectivity and relative orientation of the assembling units Within each topology replacement of a
building unit by another of same connectivity but different size leads to what is known as isoreticular
alteration of pore size The structure of MOFs in principle can be formed into the requirement of
prominent applications such as gas storage4-8 sensing910 and catalysis11-13 The structural
divergence and the performance can be further increased by functionalizing the organic linkers1415
In MOFs linkers are essential in determining the topology as well as providing porosity A linker
typically contains single or multiple aromatic rings the orientation of which normally undergoes
lowest repulsion from the coordinated metal-oxide clusters providing the most stable symmetry for
the bulk material We encounter for the first time a situation that the orientation of the linker
provides metastability in an isoreticular series of MOFs Templated MOF-2 derivatives (surface-MOF-
2 or simply SURMOFs-2)16 show this extraordinary phenomenon While bulk MOF-2 is determined to
be stable in P217 or C218-20 symmetry we found the existence of SURMOFs-2 in P4 symmetry
161
(Figure 1)16 Our theoretical approach to this problem identifies the role of the linkers in providing
P4 geometry the status of a local energy-minimum
MOF-2 derivatives are two-dimensional lattice structures composed of dimetal-centered four-fold
coordinated paddlewheels and linear organic linkers Solvothermally synthesized archetypal MOF-2
had transition metal (Zn or Cu) centers in the connectors and benzenedicarboxylic acid linkers17 The
derivatives under our investigation contain paddlewheels with Cu dimers and benzenedicarboxylic
acid (BDC) naphthalenedicarboxylic acid (NDC) biphenyldicarboxylic acid (BPDC)
triphenyldicarboxylic acid (TPDC) quaterphenyldicarboxylic acid (QPDC) P(EP)2DC21 and
pentaphenyldicarboxylic acid (PPDC) linkers pertaining to isoreticular expansion of pore size16 The
four-fold connectivity of the Cu dimers together with linearity of the linkers gives rise to layers with
quadratic (square) topology The interlayer separation d is typically much more than that of
graphite or 2D COFS22-24 because all the atoms in one layer do not lie in one plane
In bulk form the nearest layers are shifted to each other either towards one of the four linkers
(P2)171920 or diagonally towards one of the four surrounding pores (C2)18-20 in effort to reduce
the repulsion of alike charges in the adjacent paddlewheels when they are in eclipsed form (P4)
(Figure 1) The metal-dimers often show high reactivity which results in attracting water or
appropriate solvents in their axial positions The stacking along the third axis is typically through
interlayer interactions and through hydrogen bonds established between the solvents or between
the solvents and oxygen atoms in the paddlewheel The lateral shift of layers is inevitable without
additional supports Coordination of pillar molecules such as 14-diazabicyclo[222]octane (dabco) or
bipyridine in the axis of metal dimers between the adjacent layers25 is a practical mean to avoid
layer-offset however with the change of MOF dimensionality
Figure 1 Monolayer of MOF-2 and three kinds of layer packing -P4 P2 and C2- of periodic MOF-2
162
Alternate to solvothermal synthesis a step-by-step route may be enabled for the synthesis of
MOFs26-28 It adopts liquid phase epitaxy (LPE) of MOF reactants on substrate-assisted self-assembled
monolayers This is achieved by alternate immersion of the template in metal and ligand precursors
for several cycles This allows controlled growth of MOFs on surfaces The latest outcomes of this
method are the surface-standing MOF-2 derivatives (SURMOF-2s) This isoreticular SURMOF series
has linkers of different lengths (as given above) The cell dimensions that correspond to the length of
the pore walls range from 1 to 28 nm The diagonal pore-width of one of the MOFs (PPDC) accounts
to 4 nm
After evacuation of solvents (water in the present case) X-ray diffraction patterns were taken in
directions both perpendicular (scans out of plane) and parallel (scans in-plane) to the substrate
surface The presence of only one peak -(010)- in the perpendicular scan implies that the layers
orient perpendicular to the surface and the corresponding angle gives the cell parameter a (or c) In
the latter case the peaks - (100) and (001) ndash are identical and this rules out the possibility of layer-
offset Hence the symmetry of the MOFs was determined to be P4 These peaks give rise to the cell
parameter c which is nothing but the interlayer distance d This value (56 Aring) is relatively small for
P4 geometry to have water molecules in the axial position of paddlewheels Presence of some water
molecules observed using Infrared Reflection Absorption Spectroscopy (IRRAS) might be located near
paddlewheels but exposed to the pore The new observations with the SURMOFs are very intriguing
in the context that P4 symmetry appears to be impossible for solvothermally produced MOF-2
We have primarily reported that the P4 geometry of SURMOF-2 exists as a local energy-minimum16
The verification was made using an approximate method of density functional theory (DFT) which is
London dispersion-corrected self-consistent charge density-functional based tight-binding (DFTB) In
the bulk form absolute energies of P2 and C2 are 210 and 170 meV respectively higher than P4 per
a formula unit which is equivalent to the unit cell For SURMOF the barrier for leaving P4 was nearly
50 meV per formula unit It requires further analysis to unravel the reasons for this unusual
metastability We therefore performed an extensive set of quantum chemical calculations on the
composition of the constituent building units The procedure involves defining SURMOF geometry
and analyzing the translations of individual layers
The major individual contributions to the total energy are the interaction between the paddlewheel
units (favouring P2) London dispersion interaction between tilted linkers (favouring P4) and energy
to tilt the linkers In the iso-reticular series of SURMOF-2 the differences arise only in the
163
contributions from the linkers Hence we performed an extensive study only on the smallest of all
linkers- BDC A scaling might be appropriate for other linkers
RESULTS AND DISCUSSION
In solvothermal synthesis of layered MOFs it is assumed that the nucleation process is associated
with the interaction between two connectors This is rationalized by the fact that two paddlewheels
show the strongest possible noncovalent interaction between the individual MOF building blocks
present in MOF-2 As the orientation of the paddlewheels is restricted by the linkers we studied the
stacking of adjacent layers by determining the attraction of two adjacent interacting paddlewheels
upon their respective offsets Thus we investigated the geometries corresponding to lateral
displacements between the known stable arrangements of the bulk structure ie P4-to-P2 and P4-
to-C2 (Figure 2) In the model calculation the two paddlewheels were shifted from D4h symmetry to
two directions that respectively correspond to P2 and C2 symmetries in the bulk The minima along
the two shifts are referred as min-I and min-II and are having C2h symmetry It is interesting to note
that the interaction is in all cases attractive If only the paddlewheels are studied the D4h
configuration where both axes are concentric can be interpreted as transition state between the
two minima configurations P2 or C2 Comparing the two minima min-I corresponding to MOF-2 in
P2 symmetry indicates the formation of two CuO bonds while for min-II the reactive Cu centers do
not participate in the interlayer bonding
Formation of the diagonally shifted C2 bulk geometry for Zn and Cu based MOF-2 as reported in the
literature18-20 possibly is due to the presence of large solvent molecules such as DMF that
coordinate to the free Cu centers the paddlewheels
Figure 2 Displacements of two paddlewheels from D4h symmetry (corresponding to P4) towards minima that correspond to P2 and C2 bulk symmetries
164
To gain further insight on type of interactions for the three paddlewheel arrangements as found in
the bulk (Figure 3) we performed the topological analysis of the electron density for each
structure2930 An inspection of the paddlewheel shows that the electron density of the Cu-O bond has
a (relatively small) value of 0042 au thus showing only weak character In the forms related to P4
and C2 bulk symmetries all (3-1) critical points between the two paddlewheels show very small
density values (0004 au and less) In the P2 structure it is apparent the formation of a four-
membered Cu-O-Cu-O ring across the paddlewheel Note that the Cu-O interactions inside the
paddlewheel weaken (from 0042 to 0031 au) and two intermolecular Cu-O interactions with a
density value of 0024 au stabilize the P2 orientation These links if formed during nucleation will
be regularly repeated during the MOF-2 formation in solvothermal synthesis and due to its strong
binding of -08 eV they are the main reason for the stabilization of the P2 phase If the paddlewheels
are packed in P4 symmetry there must be additional means of stabilization present and that may
only arise from the linkers Moreover one should note that in order to reach the P4 symmetry a
layer-by-layer growth process is required as otherwise the linkers will readily stack as in the P2 bulk
form
165
Figure 3 Topological analysis of the electron density of fully optimized P4 (top left) C2 (top right) and P4 (bottom) structures Atom (grey) (3-1) (red) and (3+1) (green) critical points along with their charge densities are shown
The optimized geometry of a single-layer MOF-2 is shown in Figure 1 Note that the phenyl rings of
the linkers are oriented perpendicular to the MOF plane Contrary to graphene or 2D COFs the more
complex structure of MOF-2 layers may become subject to change upon the interlayer interactions
This is in particular important for the P4 stacking as reported for SURMOF-2 The interaction energy
of two linkers and two benzene rings as oriented in the monolayer has been computed as function
of the interlayer distance (Figure 4) At the experimental interlayer distance of 56 Aring the linkers are
so close that they repel each other strongly and stacking the monolayer structure at the
experimental interlayer distance would introduce an energy penalty of 08 eV per linker
It would not be exotic if we assume that the anchoring of layers on the substrate plays an important
role in setting d A supporting argument is the fact that all the SURMOFs in the iso-reticular series
have the same d An additional point is that the comparatively wider linkers NDC and LM do not
create any difference in the interlayer distance
166
Figure 4 Energy as a function of interlayer distance [d] in case of two paddlewheels and two benzene molecules The energies are given with respect to the complete dissociation of the building blocks
The best adaptation for the linkers in a closer alignment to avoid the repulsion is to partially rotate
the phenyl rings This reduces the Pauli repulsion and at the same time increases the attractive
London dispersion between the linkers However the rotation is energetically penalized by 06 eV as
accordance with similar calculations found in the literature31 and is with the same order of Zn4O-
tetrahedron clusters of the IRMOFs3233
Figure 5 Energy as a function of rotation of linker between two saturated paddlewheels The dihedral angle O-C1-C2-C3 θ between the linker and the carboxylic part in the paddlewheel Values are with respect to the ground state θ = 0⁰
To model the conditions in SURMOF-2 we need to combine the intermolecular interaction of the
linkers with the barrier associated to the rotation of the linker between two paddlewheel units as
given in Figure 5 We may consider a system of three linkers placed as if they are in three adjacent
layers of bulk P4 SURMOF-2 The linkers are undergoing a rotation along their C2 axis that would be
aligned to the Cu-Cu axes in the paddlewheels (see system-I in Figure 6) The dissociation energy of
167
the system includes four times the repulsion from one adjacent linker If we neglect the interaction
between the end-linkers which is only -35 meV the system represents the situation in bulk P4 MOF-
2 The gain of intermolecular energy due to twisting the stacked linkers is partially compensated by
the energy penalty arising from rotation of the linker between the paddlewheels and the resulting
energy shows a minimum at 22deg (Figure 6)
Figure 6 Model of the linker orientation in SURMOF-2 The stabilization of a linker inbetween two layers is approximated in System-I the arrows indicating the rotation while the energy penalty due to breaking the p conjugation of Figure 5 is given in System-II The resulting energy is the black line showing a minimum at 22deg The energy in System-I is with respect to its dissociation limit
Each linker may undergo rotation either clockwise or anti-clockwise with equal preferences in the
local environment However there may be a global control over the preference of each linker The
most stable structure can be figured out from the total energies of each possible arrangement Since
there are only two choices for each linker it may orient either in same fashion or alternate fashion
along X and Y directions If we expect a regular pattern the total number of possibilities are only
three as shown in Figure 8 where rotation of linkers (violet lines) are marked with lsquo+rsquo signs in one of
its two sides which means that facing (outer)-edges of linkers are tilted towards these sides The
three orderings may be verbalized as follows
(i) projection of the facing edges of oppositely placed linkers are either within the square or outside
(P4nbm) (ii) projection of the facing edge of only one of the oppositely placed linkers is within the
square (P4mmm) (iii) projection of the facing edges of all four linkers are either within the square
or outside (P4nmm)
The adjacent layers follow the same linker-orientations The unit cells of (i) and (iii) are four times
bigger than (ii) The stacking energies calculated at the DFTB level suggest P4nbm as the most stable
168
geometry The energies are respectively -048 -032 and -029 eV per formula unit for P4nbm
P4mmm and P4nmm respectively Within P4nbm symmetry the linker edges undergo lowest
repulsion compared to others It is to be noted that only P4nbm has four-fold rotational symmetry
along Z-axis about the Cu-dimer in any paddlewheel
Figure 7 Three possible arrangements of tilted linkers in the periodic SURMOF-2 Paddlewheels are shown as red diamonds and the linkers as violet lines Facing (outer) edges of the linkers are rotated towards the side with + signs The symmetries and calculated stacking energies (per formula unit) of the arrangements are also given
To quantify the different stacking energies we performed periodic DFT calculations on the structure
of Cu2(bdc)2 MOF-2 As the most stable P4nbm geometry demands high computational resources in
each calculation we used P4mmm geometry which has four times less atoms in unit cell We
explored tight-packing with rotated linkers and loose-packing with un-rotated linkers Two energy-
minima have been found as shown in Figure 8a at 58 and 65 Aring corresponding to rotated and un-
rotated states of linkers respectively The latter is 40 meV more stable than the former which
means that SURMOF-2 exists in a metastable state along the perpendicular axis The relative shift of
adjacent layers in the direction of the lattice vector (or ) represents the transition from P2 to P4
and back to P2 symmetry We shifted the adjacent layers parallel to and calculated the relative
energies Since the shift of adjacent layers in P4mmm geometry is not symmetric for positive and
negative directions of averages of the energies of the shift in both directions are plotted (see
Figure 9b) P4 is a local minimum while P2 is the global minimum and the energy barrier separating
the two minima accounts to 23 meV per formula unit Hence the P4 geometry of SURMOF-2 can be
taken as metastable state of MOF-2
169
Figure 8 a) Energy as function of interlayer separation d and b) Energy as a function of relative shift of adjacent layers for periodic MOF-2 Energies are given in eV per formula unit
The role of linkers in creating a local minimum at P4 is now apparent However when undergoing the
transition from P4 to P2 the relative motions of the horizontal and vertical linkers are different from
each other Hence a qualitative study is essential to accurately determine the role of each building
block for holding SURMOFs-2 in a metastable state We thus resolved the shift of entire adjacent
layers with respect to each other into relative motions of individual building blocks The experimental
interlayer distance (56 Aring) has been fixed and the energy-profiles have been calculated using DFT
The scans include the shift of
i) a paddlewheel over other
ii) a horizontal linker over other
iii) a vertical linker over other
iv) a paddlewheel over a horizontal linker
v) a paddlewheel over a vertical linker
Here vertical and horizontal linkers are the linkers vertical and horizontal to the shift directions
respectively A complete picture of individual shifts of the MOF-2 SBU including their energy-profiles
is given in Figure 9 The shift of the vertical over the horizontal linker was found insignificant and was
omitted A note of warning is that the tilted vertical linker meets different neighborhoods when
shifted to the left and right However an average energy of these two shifts seems sensible because
the nearest vertical linker in the same layer of P4nbm geometry has opposite orientation This
averaging also makes sense in a case that alternate layers undergo shifting to the same direction
leading to a zigzag (or serrated) packing of layers in the bulk As is readily seen in Figure 9 the
formation of the Cu-O-Cu-O rings between the paddlewheels (see Figure 3) is the key reasons for the
layer shift in P2 MOF-2 A support is obtained from the shifting of the paddlewheel over the
170
horizontal linker (Shift IV) A major objection is made by the vertical linkers (Shift III) The total
interaction becomes repulsive when a vertical linker shifts with respect to the other for about 05 Aring
This may alter the tilt of the linker however a minimum is already established The vertical linkers of
a layer in a way are trapped between those of adjacent layers The shift to either side from P4 most
probably decreases the interlayer separation However this demands further rotation of the vertical
linkers in order to reduce their mutual Pauli repulsion Overall the sharp minimum at P4 may be
taken as an inheritance from the lower interlayer distance of P4 geometry fixed by the anchoring on
the substrate
Figure 9 Shift of individual building blocks in the adjacent layers correspond to the situation in bulk The energies of each shifted arrangement and their total energy are given in the graph
The total energy involved in the shifting of two building blocks (one building block over the other) is
equivalent to the energy per one building block when it feels shift from two neighbors Only the
vertical linker is sensitive to the shift-direction of the two neighbors However since averages were
taken as discussed earlier the total energy becomes independent of the direction Besides the
relative motion of the SBU facing each other in P4 symmetry (Shifts I II and III) for strong distortions
we also have to consider the interaction of adjacent linker-connector interactions as represented in
Shifts IV and V The former stabilizes P2 while the latter support P4 Overall the total energy for all
the shifts shows a local minimum at P4 (see figure 9) in agreement with the periodic calculation
shown in Figure 8 Indeed the relative energy between P2 and P4 stackings estimated by the
171
superposition of individual contributions is 43 meV in close agreement to the 40 meV obtained by
the periodic calculations
Our finite-component model successfully provides adequate information on the individual
contributions of building blocks Vertical linkers play crucial role in holding the SURMOF with P4
symmetry which was never achieved with solvothermally synthesized MOF-2 Paddlewheels are
held most responsible for lateral shift The vertical linkers winningly establish a local minimum at P4
for the SURMOF
Recently we have reported SURMOF-2 derivatives with very large pore sizes(Ref) This has been
achieved by increasing the length of the linker units In view of our analysis of the stacking and
stability of MOF-2 we can now safely state that it will be possible to generate SURMOF-2 derivatives
with even larger pores with pore sizes essentially limited by the availability of stiff long organic
linker molecules This is exemplified by a calculation of the stacking energy of a series of phenyl
oligomers that is SURMOF-2 derivatives incorporating chains with 1 to 10 benzene rings in the
linkers The stacking energies per formula unit are -041 -067 -091 -121 -145 -168 -192 -215
-237 and -26 eV respectively Each benzene ring adds more than 02 eV into the stacking energy per
formula unit This energy is due to the London dispersion interaction between the linkers in the
neighboring layers
The drop of SURMOFs into the local minimum perhaps is blended with some parameters related to
synthetic environments This was beyond the scope of this work however we suggest that studies of
the mechanism of substrate-assisted layer-by-layer deposition of metal and ligand precursors may
give some primary insights into it
CONCLUSION
We have analyzed the reason for the different stackings observed for MOF-2 In the traditional
solvothermal synthesis MOF-2 is likely to crystallize in P2 symmetry determined by the strong
interaction between the paddlewheel units The coordination of large solvent molecules to the free
metal centers may lead to MOF-2 in C2 symmetry In contrast the formation of SURMOFs using
Liquid Phase Epitaxy (LPE) allows the formation of high-symmetry P4 MOF-2 This structure requires
a structural modification in terms of the orientation of the linkers with respect to the free monolayer
and the stacking is stabilized by London dispersion interactions between the linkers Increasing the
linker length is a straightforward way for the linear expansion of pore size and according to our
computations the pore size is only limited by the availability of linker molecules showing the desired
length Thus we presented a rare situation in which the linkers guarantee the persistence of a series
of materials in an otherwise unachievable state
172
COMPUTATIONAL DETAILS
The Becke three-parameter hybrid method combined with Lee-Yang-Par correlational functional
(B3LYP)34-37 and augmented with a dispersion term38 as implemented in Turbomole39 has been used
for DFT calculations The copper atoms were described using the basis set associated with the Hay-
Wadt40 relativistic effective core potentials41 which include polarization f functions42 This basis set
was obtained from the EMSL Basis Set Library4344 Carbon oxygen and hydrogen atoms were
described using the Pople basis set 6-311G Geometry optimizations of the MOF fragments were
performed using internal redundant co-ordinates45 A triplet spin state was assigned for each Cu-
paddlewheel46
Periodic structure calculations at the BLYP47 level were performed using the ADF BAND 2012
code4849 Dispersion correction was implemented based on the work by Grimme50 Triple-zeta basis
set was used The crystalline state of MOFs was computationally described using periodic boundary
conditions Bader charge analysis of paddlewheel-pairs was also performed using ADF 2012 code
The analysis was made at the level of B3LYP-D using an all-electron triple-zeta basis set
The non-orthogonal tight-binding approximation to DFT called Density Functional Tight-Binding
(DFTB)51 is computationally feasible for unit cells of several hundreds of atoms Hence this method
was used for extensive calculations on periodic structures This method computes a transferable set
of parameters from DFT calculations of a few molecules per pair of atom types The more accurate
self-consistent charge (SCC) extension of DFTB52-54 has been used for the study of bulk MOFs Validity
of the Slater-Koster parameters of Cu-X (X = Cu C O H) pairs were reported before55 The
computational code deMonNano56 which has dispersion correction implemented57 was used
If not stated otherwise the energies of the periodic structure are given per a formula unit (FU) of the
MOF which includes one paddlewheel and two linkers (one vertical linker and one horizontal linker)
REFERENCES
(1) Eddaoudi M Moler D B Li H L Chen B L Reineke T M OKeeffe M Yaghi O M Accounts of
Chemical Research 2001 34 319
(2) Li H Eddaoudi M OKeeffe M Yaghi O M Nature 1999 402 276
(3) Rowsell J L C Yaghi O M Microporous and Mesoporous Materials 2004 73 3
(4) Eddaoudi M Li H L Yaghi O M Journal of the American Chemical Society 2000 122 1391
(5) Rowsell J L C Yaghi O M Angewandte Chemie-International Edition 2005 44 4670
173
(6) Suh M P Park H J Prasad T K Lim D-W Chemical Reviews 2012 112 782
(7) Murray L J Dinca M Long J R Chemical Society Reviews 2009 38 1294
(8) Rosi N L Eckert J Eddaoudi M Vodak D T Kim J OKeeffe M Yaghi O M Science 2003 300
1127
(9) Kreno L E Leong K Farha O K Allendorf M Van Duyne R P Hupp J T Chemical Reviews 2012
112 1105
(10) Achmann S Hagen G Kita J Malkowsky I M Kiener C Moos R Sensors 2009 9 1574
(11) Lee J Farha O K Roberts J Scheidt K A Nguyen S T Hupp J T Chemical Society Reviews 2009
38 1450
(12) Farrusseng D Aguado S Pinel C Angewandte Chemie-International Edition 2009 48 7502
(13) Corma A Garcia H Llabres i Xamena F X Chemical Reviews 2010 110 4606
(14) Rowsell J L C Millward A R Park K S Yaghi O M Journal of the American Chemical Society 2004
126 5666
(15) Deng H Doonan C J Furukawa H Ferreira R B Towne J Knobler C B Wang B Yaghi O M
Science 2010 327 846
(16) Liu J Lukose B Shekhah O Arslan H K Weidler P Gliemann H Braumlse S Grosjean S Godt A
Feng X Muumlllen K Magdau I-B Heine T Woumlll C submitted to Nature Chemistry 2012
(17) Li H Eddaoudi M Groy T L Yaghi O M Journal of the American Chemical Society 1998 120 8571
(18) Carson C G Hardcastle K Schwartz J Liu X Hoffmann C Gerhardt R A Tannenbaum R
European Journal of Inorganic Chemistry 2009 2338
(19) Clausen H F Poulsen R D Bond A D Chevallier M A S Iversen B B Journal of Solid State
Chemistry 2005 178 3342
(20) Edgar M Mitchell R Slawin A M Z Lightfoot P Wright P A Chemistry-a European Journal 2001
7 5168
(21) Schaate A Roy P Preusse T Lohmeier S J Godt A Behrens P Chemistry-a European Journal
2011 17 9320
(22) Cote A P Benin A I Ockwig N W OKeeffe M Matzger A J Yaghi O M Science 2005 310
1166
(23) Wan S Guo J Kim J Ihee H Jiang D Angewandte Chemie-International Edition 2008 47 8826
174
(24) Lukose B Kuc A Frenzel J Heine T Beilstein Journal of Nanotechnology 2010 1 60
(25) Kitagawa S Kitaura R Noro S Angewandte Chemie-International Edition 2004 43 2334
(26) Shekhah O Wang H Zacher D Fischer R A Woell C Angewandte Chemie-International Edition
2009 48 5038
(27) Shekhah O Wang H Kowarik S Schreiber F Paulus M Tolan M Sternemann C Evers F
Zacher D Fischer R A Woll C Journal of the American Chemical Society 2007 129 15118
(28) Zacher D Schmid R Woell C Fischer R A Angewandte Chemie-International Edition 2011 50 176
(29) Bader R F W Accounts of Chemical Research 1985 18 9
(30) Merino G Vela A Heine T Chemical Reviews 2005 105 3812
(31) Tafipolsky M Schmid R Journal of Chemical Theory and Computation 2009 5 2822
(32) Kuc A Enyashin A Seifert G Journal of Physical Chemistry B 2007 111 8179
(33) Winston E B Lowell P J Vacek J Chocholousova J Michl J Price J C Physical Chemistry
Chemical Physics 2008 10 5188
(34) Becke A D Journal of Chemical Physics 1993 98 5648
(35) Lee C T Yang W T Parr R G Physical Review B 1988 37 785
(36) Vosko S H Wilk L Nusair M Canadian Journal of Physics 1980 58 1200
(37) Stephens P J Devlin F J Chabalowski C F Frisch M J Journal of Physical Chemistry 1994 98
11623
(38) Civalleri B Zicovich-Wilson C M Valenzano L Ugliengo P Crystengcomm 2008 10 405
(39) University of Karlsruhe and Forschungszentrum Karlsruhe gGmbH - TURBOMOLE V63 2011 2007
(40) Wadt W R Hay P J Journal of Chemical Physics 1985 82 284
(41) Roy L E Hay P J Martin R L Journal of Chemical Theory and Computation 2008 4 1029
(42) Ehlers A W Bohme M Dapprich S Gobbi A Hollwarth A Jonas V Kohler K F Stegmann R
Veldkamp A Frenking G Chemical Physics Letters 1993 208 111
(43) Feller D Journal of Computational Chemistry 1996 17 1571
(44) Schuchardt K L Didier B T Elsethagen T Sun L Gurumoorthi V Chase J Li J Windus T L
Journal of Chemical Information and Modeling 2007 47 1045
175
(45) von Arnim M Ahlrichs R Journal of Chemical Physics 1999 111 9183
(46) St Petkov P Vayssilov G N Liu J Shekhah O Wang Y Woell C Heine T Chemphyschem 2012
13 2025
(47) Gill P M W Johnson B G Pople J A Frisch M J Chemical Physics Letters 1992 197 499
(48) SCM Amsterdam Density Functional 2012
(49) Velde G T Bickelhaupt F M Baerends E J Guerra C F Van Gisbergen S J A Snijders J G
Ziegler T Journal of Computational Chemistry 2001 22 931
(50) Grimme S Journal of Computational Chemistry 2006 27 1787
(51) Seifert G Porezag D Frauenheim T International Journal of Quantum Chemistry 1996 58 185
(52) Elstner M Porezag D Jungnickel G Elsner J Haugk M Frauenheim T Suhai S Seifert G
Physical Review B 1998 58 7260
(53) Frauenheim T Seifert G Elstner M Hajnal Z Jungnickel G Porezag D Suhai S Scholz R
Physica Status Solidi B-Basic Research 2000 217 41
(54) Oliveira A F Seifert G Heine T Duarte H A Journal of the Brazilian Chemical Society 2009 20
1193
(55) Lukose B Supronowicz B Petkov P S Frenzel J Kuc A Seifert G Vayssilov G N Heine T
physica status solidi (b) 2011
(56) Heine T Rapacioli M Patchkovskii S Frenzel J Koester A M Calaminici P Escalante S Duarte
H A Flores R Geudtner G Goursot A Reveles J U Vela A Salahub D R deMon-nano edn ed deMon
2009
(57) Zhechkov L Heine T Patchkovskii S Seifert G Duarte H A Journal of Chemical Theory and
Computation 2005 1 841
ii
Acknowledgment
Foremost I would like to thank my supervisor Prof Dr Thomas Heine for the wonderful opportunity to join his group as his PhD student I am greatly thankful to him for giving me the topic and sharing with me his expertise and research insight His thoughtful advices have served to give me senses of motion and direction His ambitious approach to science has given me motivation as well as chances and exposures to develop in science His constant attention and guidance have led my scientific outputs to the best levels possible I am also thankful for the financial support and the comfortable stay in his group during my PhD time Additionally he is acknowledged for correcting and reviewing my thesis
Prof Dr Ulrich Kleinekathoumlfer deserves special thanks as my Thesis Committee member I am very glad to have him in the Committee and greatly thankful for reviewing and evaluating the thesis I also thank him and Prof Ulrich Kortz for the evaluation of my PhD proposal I am thankful also for their friendly manners and considerations throughout my PhD time
Prof Dr Christof Woumlll Director of Functional Interfaces Karlsruhe Institute of Technology is greatly acknowledged for being the external Thesis Committee member I am greatly thankful for the evaluation and reviewing of the thesis I am very much moved by his research outcomes and thankful for sharing them with us Our collaborations with his group have particularly enriched my thesis
Prof Dr Petko Petkov is also acknowledged for reviewing my thesis I particlulary thank him for the friendship and discussions thoughout my PhD time
I am indebted to Dr Agnieszka Kuc for introducing me to the topic of nanoporous materials Her experience and expertise have helped me to begin a career in this field I extend my gratitude for sharing with me her scientific skills and correcting our joint-articles
Dr Lyuben Zhechkov and Dr Achim Gelessus have been great in providing computational assistance I have benefitted from their knowledge and sincerity through fast and timely helps
I owe my heartfelt thanks to Dr Lyuben Zhechkov Dr Nina Vankova Dr Augusto Oliveira Dr Andreas Mavrantonakis Dr Stefano Borini and Dr Christian Walther for all discussions suggestions support help and particularly their lectures Dr Lyuben Zhechkov and Dr Nina Vankova are specially mentioned for their long-term attentions and helps Dr Akhilesh Tanwar is acknowledged for his helps in the beginning of my PhD
In my daily work I have been blessed with a friendly and cheerful group of fellow students Barbara Jianping Wahid Nourdine Mahdi Lei Rosalba Ievgenia Wenqing Guilherme Farjana Maicon Aleksandar Ionut Yulia and Gabriel Discussions aside I had great fun times with them Our interactions have also helped me to develop in a personal level I thank them from my full heart although just a few words are not enough I specially thank Barbara Wahid and Ionut for the joint works and publications
Mrs Britta Berninghausen our project assistant deserves special thanks for the friendly assistance on all matters with the university administration
I thank all the members of the group for a lot of good things From the supervisor to the newly joined member everyone has contributed for the general good fun and easiness All those ldquobio-fuelrdquo workshops barbecues parties retreats and gatherings are unforgettable The group also kept good phase with other groups and visitors I thank all the members once again for the good times I would not have been happier anywhere else
iii
I extend my thanks to the research groups that I visited during the PhD time Dr Sourav Pal Director of National Chemical Laboratory Pune and Dr V Subrahmanian Central Leather Research Institute Chennai deserve my gratitude for giving me the opportunity to visit and work with their group members Also I am very thankful to Prof D Sc Georgi N Vayssilov Faculty of Chemistry University of Sofia for the interesting collaboration and visit to his group The financial assistance during each stay is greatly acknowledged I also thank the members of the respective groups namely Dr Petko Petkov and his family who made the visit to Bulgaria very much entertaining
Prof Dr Lars Pettersson of University of Stockholm Dr Tzonka Mineva of CNRS Montpellier and all other members of the HYPOMAP research project are acknowledged for the scientific discussions exposures and promotions
I acknowledge several projects of Prof Dr Thomas Heine for the financial support of my work and travel the funding sources include the European Commission Deutsche Forschungsgemeinschaft (DFG) and the joint Bulgarian-German exchange program (DAAD)
I thank all the co-authors of my publications who have contributed their knowledge ideas and work to accomplish our scientific goals Without their efforts all those works would not have been complete
Members of Research III of SES at Jacobs University namely Robert Carsten Joumlrg Bogdan Meisam Niraj Mahesh Vinu Pinky Patrice Mehdi Sidhant and all professors postdocs and students in Nanofun center are thankfully mentioned here
A lot of my friends in the campus deserve my thanks Mahesh Mahendran Vinu Deepa Srikanth Rajesh Arumugam Prasad Dhananjay Sunil Tripti Raghu Suneetha Rami Susruta Niraj Abhishek Ashok Rakesh Sagar Rohan Naveen Yauhen Yannic Mila and Samira are thanked for the gatherings travels making funs and those cricket and volleyball evenings Some of them are specially thanked for the occasional ldquogahn bayrdquo parties I owe many thanks to Yauhen Srikanth and Prasad for being good flat-mates and having talks on any matters Srikanth and Prasad are thanked again for generously extending their cooking skills to me
I wish to thank everybody with whom I have shared experiences in life I am obliged to my MSc lecturer Dr Rajan K John whose dreams have inspired and driven me to research In particular his accomplishments in the George Sudarshan Center CMS College Kottayam have molded me to take up this career My previous research supervisors Prof S Lakshmibala and Prof V Balakrishnan of IIT Madras and Dr Anita Mehta of SNBNCBS Kolkata are also acknowledged for their important influences in my academic life Additionally all my teachers friends and well-wishers from neighborhood school college GS Center IIT-M and SN Bose center are thanked and acknowledged Members of St Antonyrsquos Parish Olassa are also thanked for the regards and encouragement
Jacobs University Bremen and its people have been amazing in all sorts of things I am glad that I have been a member of the University With my full heart I thank the university authority for all its facilities that were open for me I also thank Dr Svenja Frischholz Mr Peter Tsvetkov and Ms Kaija Gruumlenefeld in the administration departments for the timely helps
Lastly and most importantly I wish to thank my dearest ones for all the sacrifices and love My parents K P Lukose and Molly and my brother Anit deserve to be thanked They have always supported and encouraged me to do my best in all matters of life I also wish to thank my entire extended family for providing me a loving environment
iv
Abstract
Framework materials are extended structures that are built into destined nanoscale architectures
using molecular building units Reticular synthesis methods allow stitching of a large variety of
molecules into predicted networks Porosity is an obvious outcome of the stitching process These
materials are classified and named according to the chemical composition of the building blocks For
instance Metal-Organic Frameworks (MOFs) consists of metal-oxide centers that are linked together
by organic entities The stitching process is straight-forward so that there are already thousands of
them synthesized Controlled growth of MOFs on substrates leads to what is known as surface-MOFs
(SURMOFs) A low-weight metal-free version of MOFs is known as Covalent Organic Frameworks
(COFs) They consist of light elements such as boron oxygen silicon nitrogen carbon and hydrogen
atoms Diamond-like structures with a variety of linear organic linkers between tetrahedral nodes is
called Porous Aromatic Frameworks (PAFs)
The thesis is composed of computational studies of the above mentioned classes of materials The
significance of such studies lies in the insights that it gives about the structure-property relationships
Density Functional Theory (DFT) and its Tight-Binding approximate method (DFTB) were used in
order to perform extensive calculations on finite and periodic structures of several frameworks DFTB
provides an ab-initio base on periodic structure calculations of very large crystals which are typically
studied only using force-field methods The accuracy of this approximate method is validated prior to
reasoning
As the materials are energized from building units and coordination (or binding) stability vs
structure is discussed Energy of formation and mechanical strength are particularly calculated Using
dispersion corrected (DC)-Self Consistent Charge (SCC) DFTB we asserted that 2D COFs should have a
layer arrangement different from experimental suggestions Our arguments supported by simulated
PXRDs were later verified using higher level theories in the literature Another benchmark is giving an
insightful view on the recently reported difference in symmetries of two-dimensional MOFs and
SURMOFs The result of it shows an extra-ordinary role of linkers in SURMOFs providing
metastability
Electronic properties of all the materials and hydrogen adsorption capacities of PAFs are discussed
COFs PAFs and many of the MOFs are semiconductors HOMO-LUMO gaps of molecular units have
crucial influence in the band gaps of the solids Density of states (DOS) of solids corresponds to that
of cluster models Designed PAFs give a large choice of adsorption capacities one of them exceeds
the DOE (US) 2005 target Generally the studies covered under the scheme of the thesis illustrate
the structure stability and properties of framework materials
- Dedicated to my Family and Rajan sir
Table of Contents 1 Outline 1
2 Introduction 2
21 Nanoporous Materials 2
22 Reticular Chemistry 3
23 Metal-Organic Frameworks 5
24 Covalently-bound Organic Frameworks 8
3 Methodology and Validation 10
31 Methods and Codes 10
32 DFTB Validation 11
4 2D Covalent Organic Frameworks 13
41 Stacking 13
42 Concept of Reticular Chemistry 15
5 3D Frameworks 17
51 3D Covalent Organic Frameworks 17
52 Porous Aromatic Frameworks 18
6 New Building Concepts 20
61 Isoreticular Series of SURMOFs 20
62 Metastability of SURMOFs 21
7 Summary 23
71 Validation of Methods 23
72 Weak Interactions in 2D Materials 25
73 Structure-Property Relationships 27
List of Abbreviations 31
List of Figures 32
References 33
Appendix A Review of covalently-bound organic frameworks 37
Appendix B Properties of MOFs within DFTB 81
Appendix C Stacking of 2D COFs 96
Appendix D Reticular concepts applied to 2D COFs 105
Appendix E Properties of 3D COFs 122
Appendix F Properties of PAFs 131
Appendix G Isoreticular SURMOFs of varying pore sizes 145
Appendix H Metastability in 2D SURMOFs 160
1
1 Outline
I prepared this cumulative thesis as it is permitted by Jacobs University Bremen as all work has been
published in international peer-reviewed journals is submitted for publication or in a late
manuscript state in order to be submitted soon The list of articles contains three published papers
three submitted manuscripts and two manuscripts that are to be submitted The articles are given in
Appendices A-H in the order of their discussions Each appendix has one paper and its supporting
information
The chapters from ldquoIntroductionrdquo to ldquoNew Building Conceptsrdquo contain general discussions about the
articles and provide a red thread leading through the articles The discussions mainly circle around
the context and the content of the articles
The chapter ldquoIntroductionrdquo provides a general introduction to the topic and a review of the materials
discussed in this thesis As no comprehensive review on ldquoCovalently-bound Organic Frameworksrdquo is
available in the literature we prepared a manuscript (Appendix A) covering that subject The chapter
ldquoMethodology and Validationrdquo discusses one article on validation of DFTB method in Metal-Organic
Frameworks (Appendix B) Each consecutive chapter discusses two articles The chapter ldquo2D
Covalent Organic Frameworksrdquo covers the studies on layer arrangements (Appendix C) followed by
analysis on how reticular chemistry concepts can be used to design new materials (Appendix D) The
chapter ldquo3D Frameworksrdquo discusses the stability and properties of 3D COFs (Appendix E) and PAFs
(Appendix F) It also discusses hydrogen storage capacities of PAFs The chapter ldquoNew Building
Conceptsrdquo has coverage on 2D MOFs that are built on organic surfaces The first article in this chapter
describes the achievement of our collaborators in synthesizing a series of SURMOFs of varying pore
sizes supported by our calculations indicating their matastability Extensive calculations revealing the
role of linkers in creating the unprecedented difference of SURMOFs in comparison with the bulk
MOFs is described in another article
Details of the articles and references to the appendices are given in the respective places in each
chapter The chapter ldquoSummaryrdquo gives an overview of all the results and discussions It also discusses
some impacts of the publications and concludes the thesis Overall the studies bring into picture
different classes of materials and analyze their structural stabilities and properties
2
2 Introduction
21 Nanoporous Materials
The field of nanomaterials covers materials that have properties stemming from their nanoscale
dimensions (1-100 nm) In contrast to nanomaterials which define size scaling of bulk materials the
major determinant of nanoporous materials is their pores Nanoporous materials are defined as
porous materials with pore diameters less than 100 nm and are classified as micropores of less than
2 nm in diameter mesopores between 2 and 50 nm and macropores of greater than 50 nm They
have perfectly ordered voids to accommodate interact with and discriminate molecules leading to
prominent applications such as gas storage separation and sieving catalysis filtration and
sensoring1-4 They differ from the generally defined nanomaterials in the sense that their properties
are mostly determined by pore specifications rather than by bulk and surface scales Hence the
focus is onto the porous properties of the materials
Utilization of the pores for certain applications relies on certain parameters such as pore size pore
volume internal surface area and wall composition For example physical adsorption of gases is high
in a material with large surface area which implies significantly high storage in comparison to a tank
Porosity can be measured using some inert or simple gas adsorption measurements Distribution of
pore size can be sketched from the adsorptiondesorption isotherm
Nanoporous materials abound in nature Zeolites for instance many of them are available as mineals
have been used in petroleum industry as catalysts for decades The walls of human cells are
nanoporous membranes Other examples are clays aluminosilicate minerals and microporous
charcoals Zeolites are microporous materials from the aluminosilicate family and are often known as
molecular sieves due to their ability to selectively sort molecules primarily based on a size exclusion
principle A material with high carbon content (coal wood coconut shells etc) can be converted to
activated carbon (AC) by controlled thermal decomposition in a furnace The resultant product has
large surface area (~ 500 m2 g-1) Porous glass is a material produced from borosilicate glasses having
pore sizes usually in nm to μm range With increasing environmental concerns worldwide porous
materials have become a suitable choice for separation of polluting gases storage and transport of
energy carriers etc Synthesis of open structures has advanced this area5-7 especially since the
invention of MCM-41 (Mobil Composition of Matter No 41) by Mobil scientists89 Furthermore
there are many templating pathways in making nanoporous materials10-13 Currently it is possible to
engineer the internal geometry at molecular scales
3
For more than a decade chemists are able to synthesize extended structures from well-defined and
rigid molecular building units Such designed and controlled extensions provide porosity which can
be scaled and modified by selecting appropriate building blocks The first realization of this kind was
a class of materials known as Metal-Organic Frameworks (MOFs) in which metal-oxides are stitched
together by organic molecules Synthesis of molecules into predicted frameworks have led to the
emergence of a discipline called reticular chemistry14 The new design-based synthetic approaches
have produced large number of nanoporous materials in comparison to the discovery-based
synthetic chemistry
22 Reticular Chemistry
The discipline of designed (rational) synthesis of materials is known as reticular chemistry Desired
materials can be realized by starting with well-defined and rigid molecular building blocks that will
maintain their structural integrity throughout the construction process The extended structures
adopt high symmetry topologies The synthetic approach follows well-defined conditions which
provide general control over the character of solids In short it is the chemistry of linking molecular
building blocks by strong bonds into predetermined structures
The knowledge about how atoms organize themselves during synthesis is essential for the design
The principal possibilities rely on the starting compounds and the reaction mechanisms Porosity is
almost an inevitable outcome of reticular synthesis Solvents may be used in most cases as space-
filling agents and in cases of more than one possibility as structure-directing agents
Thousands of materials in large varieties have been synthesized using the reticular chemistry
principles Reticular Chemistry Structure Resource (RCSR) is a searchable database for nets a project
initiated by of Omar M Yaghi and Michael OrsquoKeeffe15 They define nets as the collection of vertices
and edges that form an irreducible network in which any two vertices are connected through at least
one continuous path of edges Building units are finite nets as in the case of a polyhedron Periodic
structures have one two or three-periodic nets An example case of 3D diamond net (dia) is shown in
Figure 1 Graph-theoretical aspects together with explicative terminology and definitions can be
found in the literature16-18
Figure 1 a) ldquoAdamantinerdquo unit of diamond structure and b) diamond (dia) net
4
In other words a framework can be deconstructed into one or more fundamental building blocks
each of them assigned by a vertex in the net The vertices are the branching points and edges are
joining them The realization of the net in space by representing the vertices and lattice parameters
by co-ordinates and a matrix respectively is called embedding of the net Underlying topology of an
extended structure is the structure of the net inherited from the crystal structure that is invariant
under different embeddings For example Cu-BTC MOF has paddlewheels and benzene rings as
fundamental blocks The MOF structure can be simplified into its underlying topology as shown in
Figure 2
Figure 2 CU-BTC MOF and the corresponding tbo net
Alternatively the topology of a framework can be defined using the convention of so-called
secondary building units (SBUs) The shapes of SBUs are defined by points of extension of the
fundamental building blocks SBUs are invariant for building units of identical connectivity Based on
the number of connections with the adjacent blocks (4 for paddlewheel and 3 for the benzene) SBUs
of Cu-BTC can be represented as shown in Figure 3a Assembly of SBUs following the network
topology and insertion of the fundamental building blocks give the crystal structure (see Figure 3b for
the extension of SBUs to the topology of Cu-BTC)
In addition to MOFs Metal-Organic Polyhedra (MOP)19 Zeolite Imidazolate Frameworks (ZIFs)20 and
Covalent-Organic Frameworks (COFs)2122 are developing classes of materials within reticular
chemistry the first members of all of them were synthesized by the group of Yaghi MOPs are nano-
sized polyhedra achieved by linking transition metal ions and either nitrogen or carboxylate donor
organic units ZIFs are aluminosilicate zeolites with replacements of tetrahedral Si (Al) and bridging
oxygen by transition metal ion and imidazolate link respectively COFs are extended organic
5
structures constructed solely from light elements (H B C and O) The materials synthesized under
the reticular scheme are largely crystalline
Figure 3 a) Fundamental building blocks and SBUs of Cu-BTC and b) extension of SBUs following
crystal structure
23 Metal-Organic Frameworks
MOFs are porous crystalline materials consisting of metal complexes (connectors) linked together by
rigid organic molecules (linkers) MOFs are analogues to a class of materials called coordination
polymers (CPs) However there are primary differences between them CPs are inorganic or
organometallic polymer structures containing metal ions linked by organic ligands A ligand is an
atom or a group of atoms that donate one or more lone pairs of electrons to the metal cations and
thereby participate in the formation of a coordination complex In MOFs typically metal-oxide
centers are used instead of single metal ions as they provide strong bonds with organic linkers This
provides not only high stability but also high directionality because multiple bonds are involved
6
between metal-centers and organic linkers Predictability lies in the pre-knowledge about the
connector-linker interactions Thus the reticular design of MOFs derives from the precise
coordination geometry between metal centers and the linkers Figure 4a shows a schematic diagram
of MOFs By varying the chemical composition of nodes and linkers thousands of different MOF
structures with a large variety in pore size and structure have been synthesized Figure 4b shows
MOF-5 that consists of zinc-oxide tetrahedrons and benzene linkers
Figure 4 a) Schematic diagram of the single pore of MOF b) An example of a simple cubic MOF in
which zinc-oxide tetrahedron are linked together by benzene linkers Atom colors blue ndash Zn red ndash
O grey ndash C white ndash H
The synthesis of MOFs differs from organic polymerizations in the degree of reversible bond
formation Reversibility allows detachment of incoherently matched monomers followed by their
attachment to form defect-free crystals Assembly of monomers occurs as single step hence
synthetic conditions are of high importance The strength of the metal-organic bond is an obstacle
for reversible bond formation however solvothermal techniques are found out to be a convenient
solution23 Solvothermal synthesis generally allows control over size and shape distribution Using
post-synthetic methods further changes on cavity sizes and chemical affinities can be made
Materials that are stable with open pores after removal of guest molecules are termed as open-
frameworks The stability of guest-free materials is usually examined by Powder X-ray diffraction
(PXRD) analysis Thermogravimetric analysis may be performed to inspect the thermal stability of the
material Elemental analysis can detail the elemental composition of the material Physical
techniques such as Fourier transform infrared (FTIR) spectra and nuclear magnetic resonance (NMR)
may be used to verify the condensation of monomers to the desired topology Porosity can be
evidenced from adsorption isotherms of gases or mercury porosimetry
7
The number of linkers that are allowed to bind to the metal-center and the orientations of the linkers
depend exclusively on the coordination preferences of the metal The organic linkers are typically
ditopic or polytopic They are essential in determining the topology and providing porosity Longer
linkers provide larger pore size A series of compounds with the same underlying topology and
different linker sizes are called iso-reticular series The structure of MOFs in principle can be formed
into the requirement of prominent applications such as gas storage gas separation sensing and
catalysis The structural divergence and performance can be further increased by functionalizing the
organic linkers Hence several attempts are on-going in purpose to come up with the best material
possible in each application
Interpenetration of frameworks is an inevitable outcome of large pore volume Interpenetrated nets
are periodic equivalents of mechanically interlocked molecules rotaxanes and catenanes Depending
on topology they are either maximally separated termed as interpenetration or minimally separated
termed as interweaving (see Figure 5) Interpenetration can be beneficial to some porous structures
protecting from collapse upon removal of solvents
Figure 5 Schematic diagram of interpenetrated and interwoven frameworks
Controlled growth of MOFs on organic surfaces is achieved by R A Fischer and C Woumlll24 The then
named SURMOFs allow to fabricate structures possibly not achievable by bulk synthesis The growth
is achieved through liquid phase epitaxy (LPE) of MOF reactants on the surface (nucleation site) A
step-by-step approach which is alternate immersion of the substrate in metal and ligand precursors
supplies control of the growth mechanism
8
Figure 6 Schematic diagram of SURMOF
24 Covalently-bound Organic Frameworks
As a result of the long-term struggle for complete covalent crystallization in the year 2005 Cocircte et
al21 synthesized porous crystalline extended 2D Covalent-Organic Frameworks (COFs) using
reticular concepts The success was followed by the design and synthesis of 3D COFs in the year
200722 By now there are about 50 COFs reported in the literature COFs are made entirely from
light elements and the building blocks are held together by strong covalent bonds Most of them
were formed by solvothermal condensation of boronic acids with hydroxyphenyl compounds
Tetragonal building blocks were used for the formation of 3D COFs Alternate synthetic methods
were also used for producing COFs COFs are generally studied for gas storage applications However
they have also shown potentialities in photonic and catalytic applications
Alternative synthesis methods paved the way to new covalently bound organic frameworks
Trimerization of polytriazine monomers in ionothermal conditions gives rise to Covalent Triazine
Frameworks (CTFs)25 Similarly coupling reactions of halogenated monomers produced Porous
Aromatic Frameworks (PAFs) Albeit the short-range order one of the PAFs showed very high surface
area (5600 m2 g-1) and gas uptake capacity26
Due to low weight the covalently-bound materials show very high gravimetric capacities
Suggestions such as metal-doping functionalization and geometry modifications can be found in the
literature for the general improvement of the functionalities There are also various studies of their
structure and properties
A review on the synthesis structure and applications of covalently bound organic frameworks has
been prepared for publication
Article 1 Covalently-bound organic frameworks
Binit Lukose Thomas Heine
9
See Appendix A for the article
My contributions include collecting data and preparing a preliminary manuscript
Figure 7 SBUs and topologies of 2D COFs
10
3 Methodology and Validation
31 Methods and Codes
The Density-Functional based Tight Binding (DFTB) method2728 is an approximate Kohn-Sham DFT29-31
scheme It avoids any empirical parameterization by calculating the Hamiltonian and overlap matrix
elements from DFT-derived local orbitals (atomic orbitals) and atomic potentials The Kohn-Sham
orbitals are represented with the linear combination of atomic orbitals (LCAO) scheme The matrix
elements of Hamiltonian are restricted to include only one- and two-center contributions Therefore
they can be calculated and tabulated in advance as functions of the distance between atomic pairs
The remaining contributions to the total energy that is the nucleus-nucleus repulsion and the
electronic double counting terms are grouped in the so-called repulsive potential This two-center
potential is fitted to results of DFT calculations of properly chosen reference systems The accuracy
and transferability can be improved with the self-consistent charge (SCC) extension to DFTB32 This
method is based on the second-order expansion of the Kohn-Sham total energy with respect to
charge density fluctuations which are estimated by Mulliken charge analysis In order to account for
London dispersion empirical dispersion energy can be added to the total energy3334 Various reviews
are available on DFTB notably the recent ones by Oliveira et al35 and by Seifert et al36
DFTB is implemented in a large number of computer codes For this work we employed the codes
deMonNano37 and DFTB+38 as they allow DFTB calculations on periodic and finite structures
Typically dispersion corrected (DC) SCC-DFTB calculations were carried out Periodic boundary
conditions were used to represent the crystalline frameworks and as the unit cells are large the
standard approach used the point approximation Electronic density of states (DOS) have been
calculated using the DFTB+ code using k-point sampling where the k space was determined by
reaching convergence for the total energy according to the scheme proposed by Monkhorst and
Pack39
For DFT calculations of finite and periodic structures ADF40 Turbomole41 and Crystal0942 were used
For studies of finite models of COFs the calculations were performed at PBEDZP level However for
extensive calculations on SURMOF-2 models we used B3LYP-D The copper atoms were described
using the basis set associated with the Hay-Wadt43 relativistic effective core potentials44 which
include polarization f functions45 Carbon oxygen and hydrogen atoms were described using the
Pople basis set 6-311G
Details of the individual calculations are given in the individual articles in the appendix of this thesis
11
32 DFTB Validation
Figure 8 Deconstructed building units their schematic diagrams and final geometries of HKUST-1
(Cu-BTC) MOF-5 MOF-177 MOF-205 and MIL-53
12
In the literature MOFs and COFs are largely studied for applications such as gas storage using
classical force field methods46-48 First principles based studies of several hundreds of atoms are
computationally expensive Hence they are generally limited to cluster models of the periodic
structures Contrarily DFTB paves the way to model periodic structures involving large numbers of
atoms Being an approximate method DFTB holds less accuracy hence comparisons to experimental
data or higher level methods should be performed for validation
As MOFs contain metal centers andor metal oxides as well as organic elements the DFTB
parameters for both heavy and light elements as well as their mixtures are required Thus we have
chosen MOFs such as Cu-BTC (HKUST-1) MOF-5 MOF-177 MOF-205 and MIL-53 as our model
structures which contain Cu Zn and Al atoms apart from C O and H The MOFs comprising three
common connector units copper paddlewheels (HKUST-1) zinc oxide Zn4O tetrahedron (MOF-5
MOF-177 DUT-6 (MOF-205)) and aluminum oxide AlO4(OH)2 octahedron (MIL-53) Figure 8 shows
the schematic diagram of the MOFs
The validation calculations have been published
Article 2 Structural properties of metal-organic frameworks within the density-functional based
tight-binding method
Binit Lukose Barbara Supronowicz Petko St Petkov Johannes Frenzel Agnieszka B Kuc Gotthard
Seifert Georgi N Vayssilov and Thomas Heine Phys Status Solidi B 2012 249 335ndash342 DOI
101002pssb201100634
See Appendix B for the article
In this article DFTB has been validated against full hybrid density-functional calculations for model
clusters against gradient corrected density-functional calculations for supercells and against
experiment Moreover the modular concept of MOF chemistry has been discussed on the basis of
their electronic properties Our results show that DC-SCC-DFTB predicts structural parameters with a
good accuracy (with less than 5 deviation even for adsorbed CO and H2O on HKUST-1) while
adsorption energies differ by 12 kJ mol-1 or less for CO and water compared to DFT benchmark
calculations
My contributions to this article include performing DFTB based calculations on the MOFs (HKUST-1
MOF-5 MOF-177 and MOF-205) that is optimizing the geometries simulating powder X-ray
diffraction patterns and calculating density of states and bulk modulus Additional involvement is in
comparing structural parameters such as bond lengths bond angles dihedral angles and bulk
modulus with experimental data or data derived from DFT calculations and preparing the manuscript
13
4 2D Covalent Organic Frameworks
41 Stacking
Condensation of planar boronic acids and hydroxyphenyl compounds leads to the formation of two-
dimensional covalent organic frameworks (2D COFs) The layers are held together by London
dispersion interactions The first synthesized COFs COF-1 and COF-5 were reported to have AB
(P63mmc staggered as in graphite) and AA (P6mmm eclipsed as in boron nitride) stackings
respectively (see Figure 9)21 The synthesis of several further 2D COFs then followed49-51 all of them
were inspected for only two stacking possibilities ndash P6mmm or P63mmc and it was reported that
they aggregate in P6mmm symmetry As framework materials possess framework charges the
interlayer interactions significantly include Coulomb interactions P6mmm symmetry is the face-to-
face arrangement where the overlap of the stacked structures is maximized (maximization of the
London dispersion energy) however atom types of alike charges are facing each other in the closest
possible way With the interlayer separation similar to that in graphite or boron nitride the Coulomb
repulsion should be high in such arrangements One should notice that in the example case of boron
nitride the facing atom types are different We therefore assumed that a stable stacking should
possess layer-offset
Figure 9 AA and AB layer stacks of hexagonal layers
We considered two symmetric directions for layer shift and studied their total energies (see Figure
10) The stable geometries are found to have a layer-shift of ~14 Aring a value corresponding to the
shift of AA to AB stacking in graphite and compatible with the bond length between two 2nd row
atoms This stability-supported stacking arrangement as revealed from our calculations was
14
supported by good agreement between simulated and experimental PXRD patterns Hence
independent of the elementary building blocks any 2D COF should expose a layer-offset Based on
the direction of the shift we proposed two stacking kinds serrated and inclined (Figure 10) In the
former the layer-offset is back and forth while in the latter the layer-offset followed single direction
As serrated and inclined stackings have no significant change in stacking energy our calculations
cannot predict the long-range stacking in the crystal However this problem is known from other
layered compounds and the ultimate answer is yet to be revealed by analyzing high-quality
crystalline phases at low temperature
Figure 10 Stacking energy Es vs shift of adjacent layers for COF-5 Different stacking possibilities
and their energies are also shown
We published our analysis of the stacking in 2D COFs
Article 3 The Structure of Layered Covalent-Organic Frameworks
Binit Lukose Agnieszka Kuc Thomas Heine Chem Eur J 2011 17 2388 ndash 2392 DOI
101002chem201001290
See Appendix C for the article
15
My contributions to this article include performing the shift calculations simulating XRDs and partly
preparing the manuscript The article has made certain impact in the literature Most of the 2D COFs
synthesized afterwards were inspected for their stacking stability The instability of AA stacking was
also suggested by the studies of elastic properties of COF-1 and COF-5 by Zhou et al52 The shear
modulus shows negative signs for the vertical alignment of COF layers while they are small but
positive for the offset of layers Detailed analysis performed by Dichtel53 et al using DFT was
confirming the instability of AA stacking and suggesting shifts of about 13 to 25 Aring
42 Concept of Reticular Chemistry
Reticular chemistry means that (functional) molecules can be stitched together to form regular
networks The structural integrity of these molecules we also speak of building blocks remains in the
crystal lattices Consequently also the electronic structure and hence the functionality of these
molecules should remain similar
2D COFs are formed by condensation of well-defined planar building blocks With the usage of linear
and triangular building blocks hexagonal networks are expected The properties of each COF may
differ due to its unique constituents However the extent of the relationship of the properties of
building blocks in and outside the framework has not been studied in the literature
Reticular chemistry allows the design of framework materials with pre-knowledge of starting
compounds and reaction mechanisms Reverse of this is finding reactants for a certain topology We
intended to propose some building units suitable to form layered structures (see Figure 11) The
building units obey the regulations of reticular chemistry and offer a variety of structural and
electronic parameters
Our strategic studies on a set of designed COFs have been published
Article 4 On the reticular construction concept of covalent organic frameworks
Binit Lukose Agnieszka Kuc Johannes Frenzel Thomas Heine Beilstein J Nanotechnol 2010 1
60ndash70 DOI103762bjnano18
See Appendix D for the article
16
Figure 11 Schematic diagram of different building units forming 2D COFs
Various hexagonal 2D COFs with different building blocks have been designed and investigated
Stability calculations indicated that all materials have the layer offset as reported in our earlier
work54 (see Appendix C) We show that the concept of reticular chemistry works as the Density-Of-
States (DOS) of the framework materials vary with the the DOS of the molecules involved in the
frameworks However the stacking does have some influence on the band gap
My contributions to this article include performing all the calculations and preparing the manuscript
17
5 3D Frameworks
51 3D Covalent Organic Frameworks
First 3D COFs were reported in 2007 by the group of Yaghi22 Although the number of 3D COFs
synthesized so far has not been crossed half a dozen they are of particular interest for their very low
mass density Similar to 2D COFs condensation of boronic acids and hydroxyphenyl compounds led
to the formation of COF-102 103 105 108 and 202 Usage of tetragonal boronic acids leads to the
formation of 3D networks (Figure 12) All of them possessed ctn topology while COF-108 which has
the same material composition as COF-105 crystallized in bor topology COF-300 which was formed
from tetragonal and linear building units possessed diamond topology and was five-fold
interpenetrated55 COF-108 has the lowest mass density of all materials known today Unlike most of
the MOFs the connectors in COFs are not metal oxide clusters but covalently bound molecular
moieties In the synthesized COFs the centers of the tetragonal blocks were either sp3 carbon or
silicon atoms
Schmid et al56 have analyzed the two different topologies and developed force field parameters for
COFs The mechanical stability of COFs was also reported However no further study that details the
inherent energetic stability and properties of COFs was found in the literature Using DFTB we
performed a collective study of all 3D COFs in their known topologies with C and Si centers
Figure 12 Schematic diagram of tetragonal and triangular building blocks forming lsquoctnrsquo or lsquoborrsquo
topologies
Our studies of3D COFs have been prepared for publication
Article 5 Stability and electronic properties of 3D covalent organic frameworks
Binit Lukose Agnieszka Kuc Thomas Heine
18
See Appendix E for the article
My contributions to this article include performing all the calculations and preparing the manuscript
We discussed the energetic and mechanical stability as well as the electronic properties of COFs in
the article All studied 3D COFs are energetically stable materials with formation energies of 76 ndash
403 kJ mol-1 The bulk modulus ranges from 29 to 206 GPa Electronically all COFs are
semiconductors with band gaps vary with the number of sp2 carbon atoms present in the linkers
similar to 3D MOFs
52 Porous Aromatic Frameworks
Porous Aromatic Frameworks (PAFs) are purely organic structures where linkers are connected by sp3
carbon atoms (see Appendix A) Tetragonal units were coupled together to form diamond-like
networks The synthesis follows typically a Yamamoto-type Ullmann cross-coupling reaction those
reactions are known to be much simpler to be carried out than the condensation reactions necessary
to form COFs However as the coupling reactions are irreversible a lower degree of crystallinity is
achieved and the materials formed were amorphous The first PAF was reported in 2009 and
showed a very high BET surface area of 5600 m2g-1 The geometry of PAF-1 was equal to diamond
with the C-C bonds replaced by biphenyl Subsequent PAFs may be synthesized with longer linkers
between carbon tetragonal nodes Nevertheless synthesis of a PAF with quaterphenyl linker
provided an amorphous material of very low surface area due to the short range order
Synthetic complexities aside the diamond-like PAFs are interesting for their special geometries from
the viewpoint of the theorist It is interesting to see to what extent they follow the properties of
diamond Additionally the large surface area and high polarizability of aromatic rings are auxiliary for
enhanced hydrogen storage Thus we have explored the possibilities of diamond topology by
inserting various organic linkers in place of C-C bonds (Figure 13)
Figure 13 Diamond structure and various organic linkers to build up PAFs
Our studies of PAFs have been prepared for publication
19
Article 6 Structure electronic structure and hydrogen adsorption capacity of porous aromatic
frameworks
Binit Lukose Mohammad Wahiduzzaman Agnieszka Kuc Thomas Heine
See Appendix F for the article
In this article we have discussed the correlations of properties with the structures Exothermic
formation energies of PAFs calculated from the dehydrogenation of the linkers with CH4 indicate the
strong coordination of linkers in the diamond topology The studied PAFs exhibited bulk moduli much
smaller than diamond however in line with related MOFs and COFs The PAFs are semi-conductors
with their band gaps decrease with the increasing number of benzene rings in the linkers
Using Quantized Liquid Density Functional Theory (QLDFT) for hydrogen57 a method to compute
hydrogen adsorption that takes into account inter-particle interactions and quantum effects we
predicted a large choice of hydrogen uptake capacities in PAFs At room temperature and 100 bar
the predicted uptake in PAF-qtph was 732 wt which exceeds the 2015 DOE (US) target We
further discussed the structural impacts on the adsorption capacities
My contributions to this article include designing the materials performing calculations of stability
and electronic properties describing the adsorption capacities and preparing the manuscript
20
6 New Building Concepts
61 Isoreticular Series of SURMOFs
The growth of MOFs on substrates using liquid phase epitaxy (LPE) opened up a controlled way to
construct frameworks This is achieved by alternate immersion of the substrates in metal and ligand
precursors for several cycles Such MOFs named as SURMOFs can avoid interpenetration because
the degeneracy is lifted58 and are suited for conventional applications This is an advantage as
solvothermally synthesized MOFs with larger pore size suffer from interpenetration and the large
pores are hence not accessible for guest species
MOF-259 is a simple 2D architecture based on paddlewheel units formed by attaching four
dicarboxylic groups to Cu2+ or Zn2+ dimers yielding planar sheets with 4-fold symmetry The
arrangement of MOF layers possessed P2 or C2 symmetry Recently the group of C Woumlll at KIT has
synthesized an isoreticular series of SURMOFs derived from MOF-2 which has a homogeneous series
of linkers with different lengths (Figure 14) XRD comparisons suggest that these MOFs possess P4
symmetry The cell dimensions that correspond to the length of the pore walls range from 1 to 28
nm The diagonal pore-width of one of the SURMOFs (PPDC) accounts to 4 nm The symmetry of
SURMOFs as determined to be different from bulk MOF-2 should be understood by means of theory
As collaborators we simulated the structures and inspected each stacking corresponding to the
symmetries in order to understand the difference
Figure 14 The building units and schematic diagram of the formation of iso-reticular SURMOF
series
21
This collaborated work has been submitted for publication
Article 7 A novel series of isoreticular metal organic frameworks realizing metastable structures
by liquid phase epitaxy
Jinxuan Liu Binit Lukose Osama Shekhah Hasan Kemal Arslan Peter Weidler Hartmut Gliemann
Stefan Braumlse Sylvain Grosjean Adelheid Godt Xinliang Feng Klaus Muumlllen Ioan-Bogdan Magdau
Thomas Heine Christof Woumlll
See Appendix G for the article
The main contribution of this article was the experimental proof backed up by our computer
simulations that SURMOFs with exceptionally large pore sizes of more than 4 nm can be prepared in
the laboratory and they offer the possibility to host large materials such as clusters nanoparticles or
small proteins The most important contribution of theory was to show that while MOF-2 in P2
symmetry shows the highest stability the P4 symmetry as obtained using LPE for SURMOF-2
corresponds to a local minimum
My contribution to this article includes performing and analyzing the calculations Our theoretical
study went significantly beyond and will be published as separate article (Appendix H)
62 Metastability of SURMOFs
Following the difference in stability of SURMOFs-2 in comparison with MOF-2 we identified the role
of linkers as a constituent of the framework that provides the metastability to SURMOFs-2 (Figure
15) In MOFs linkers are essential in determining the topology as well as providing porosity Linkers
typically contain single or multiple aromatic rings and are ditopic and polytopic The orientation of
them normally undergoes lowest repulsion from the coordinated metal-oxide clusters and provides
high stability In the case of 2D MOFs non-covalent interlayer interactions lead to the most stable
arrangements Paddlewheels are identified as the reasons for the stability of bulk MOF-2 as they
form metal-oxide bridges across the MOF layers In the case of SURMOFs-2 the linkers are rotated in
a tightly-packed environment of layers and each linker in bulk MOF-2 is rotated in such a way that
any attempt of shifting is obstructed by repulsion between the neighboring linkers The energy
barrier for leaving P4 increases with the length of organic linkers Moreover SURMOF-2 derivatives
with extremely large linkers are energetically stable due to the increased London dispersion
interaction between the layers in formula units Thus we encountered a rare situation in which the
linkers guarantee the persistence of a series of materials in an otherwise unachievable state
22
Figure 15 Energy diagram of the metastable P4 and stable P2 structures
Our results on the linker guided stability of SUMORs-2 have been prepared for publication
Article 8 Linker guided metastability in templated Metal-Organic Framework-2 derivatives
(SURMOFs-2)
Binit Lukose Gabriel Merino Christof Woumlll Thomas Heine
See Appendix H for the article
This article is based solely on my scientific contributions
23
7 Summary
Nanotechnology is the way of ingeniously controlling the building of small and large structures with
intricate properties it is the way of the future a way of precise controlled building with incidentally
environmental benignness built in by design - Roald Hoffmann Chemistry Nobel Laureate 1981
Currently it is possible to design new materials rather than discovering them by serendipity The
design and control of materials at the nanoscale requires precise understanding of the molecular
interactions processes and phenomena In the next level the characteristics and functionalities of
the materials which are inherent to the material composition and structure need to be studied The
understanding of the materials properties may be put into the design of new materials
Computational tools to a large extend provide insights into the structures and properties of the
materials They also help to convert primary insights into new designs and carry out stability analysis
Our theoretical studies of a variety of materials have provided some insights on their underlying
structures and properties The primary differences in the material compositions and skeletons
attributed a certain choice in properties The contents of the articles discussed in the thesis may be
summarized into the following three parts
71 Validation of Methods
Simulations of nanoporous materials typically include electronic structure calculations that describe
and predict properties of the materials Density functional theory (DFT)30 is the standard and widely-
used tool for the investigation of the electronic structure of solids and molecules Even the optical
properties can be studied through the time-dependent generalization of DFT MOFs and COFs have
several hundreds of atoms in their unit cells DFT becomes inappropriate for such large periodic
systems because of its necessity of immense computational time and power Molecular force field
calculations60 on the other hand lack transferable parameterization especially for transition metal
sites and are hence of limited use to cover the large number of materials to be studied Apparently
a non-orthogonal tight-binding approximation to DFT called density functional tight-binding
(DFTB)28323561 has proven to be computationally feasible and sufficiently accurate This method
computes parameters from DFT calculations of a few molecules per pair of atom types The
parameters are generally transferable to other molecules or solids With self-consistent charge (SCC)
extension DFTB has improved accuracy In order to account weak forces the London dispersion
energy can be calculated separately using empirical potentials and added to total energy Successful
realizations of DFTB include the studies of large-scale systems such as biomolecules62
24
supramolecular compounds63 surfaces64 solids65 liquids and alloys66 Being an approximate method
DFTB needs validation Often one compares DFTB results of selected reference systems with those
obtained with DFT
Before electronic structure calculations of framework materials can be carried out it is necessary to
compute the equilibrium configurations of the atoms Geometry optimization (or energy
minimization) employs a mathematical procedure of optimization to move atoms so as to reduce the
net forces on them to negligible values We adopted the conjugate gradient scheme for the
optimizations using DFTB A primary test for the validation of these optimizations is the comparison
of cell parameters bond lengths bond angles and dihedral angles with the corresponding known
numbers We compared the DFTB optimized MOF COF and PAF geometries with the experimentally
determined or DFT optimized geometries and found that the values agree within 6 error
The angles and intensities of X-ray diffraction patterns can produce three-dimensional information of
the density of electrons within a crystal This can provide a complete picture of atomic positions
chemical bonds and disorders if any Hence simulating powder X-ray diffraction (PXRD) patterns of
optimized geometries and comparing them with experimental patterns minimize errors in the crystal
model Our focus on this matter directed us to identify the layer-offsets of 2D COFs for the first time
In the case of 3D COFs excellent correlations were generally observed between experimental and
simulated patterns Slight differences in the intensities of some of them were due to the presence of
solvents in the crystals as they were reported in the experimental articles PAFs as experimentally
being amorphous do not possess XRD comparisons MOFs within DFTB optimization have
undergone some changes especially in the dihedral angles in comparison with experimental
suggestion or DFT optimization This was verified from the differences in the simulated PXRD
patterns Another interesting outcome of XRD comparison was the discovery of P4 symmetry of
templated MOF-2 derivatives (SURMOFs-2) made by Woumlll et al
Bulk moduli which may be expressed as the second derivative of energy with respect to the unit cell
volume can give a sense of mechanical stability Our calculations provide the following bulk moduli
for some materials MOF-5 1534 (1537)67 Cu-BTC 3466 (3517)68 COF-102 206 (170)69 COF-
103 139 (118)69 COF-105 79 (57)69 COF-108 37 (29)69 COF-202 153 (110)69 The data in the
parenthesis give corresponding values found in the literature calculated using force-field methods
The bulk moduli of MOFs are comparable with the results in the literature however COFs show
significant differences Albeit the differences in values each type of calculation shows the trend that
bulk modulus decreases with decreasing mas density increasing pore volume decreasing thickness
of pore walls and increasing distance between connection nodes
25
Formation of framework materials from condensation of reactants may be energetically modeled
COFs are formed from dehydration reactions of boronic acids and hydroxyphenyl compounds The
formation energy calculated from the energies of the products and reactants can indicate energetic
stability Both DFTB for periodic structures and DFT for finite structures predict exothermic formation
of 3D COFs Likewise for 2D COFs such as COF-1 and 5 the formation of monolayers is found to be
endothermic within both the periodic model calculation using DFTB and finite model calculation
using DFT The stacking of layers provides them stability
72 Weak Interactions in 2D Materials
AB stacked natural graphite and ABC stacked rhombohedral graphite are found to be in proportions
of 85 and 15 in nature respectively whereas AA stacking has been observed only in graphite
intercalation compounds On the other hand boron nitride (h-BN) the synthetic product of boric
acid and boron trioxide exists with AA stacking 2D COFs were believed to have high symmetric AA
(eclipsed ndash P6mmm symmetry) stacking as boron nitride (h-BN) An exception was COF-1 which was
considered to have AB (staggered ndash P63mmc) stacking as graphite The AA stacking maximizes the
attractive London dispersion interaction between the layers a dominating term of the stacking
energy At the same time AA stacking always suffers repulsive Coulomb force between the layers
due to the polarized connectors It should be noted that in boron nitride oppositely charged boron
atoms and nitrogen atoms are facing each other The Coulomb interaction has ruled out possible
interlayer eclipse between atoms of alike charges This gave us the notion that 2D COFs cannot
possess layer-eclipsing Based on these facts we have suggested a shift of ~14 Aring between adjacent
layers at which one or two of the edge atoms of the hexagonal rings are situated perpendicular to
the center of adjacent rings In other words some or all of the hexagonal rings in the pore walls
undergo staggering with that of adjacent layers These lattice types were found to be very stable
compared to AA or AB analogues AA stacking was found to have the highest Coulombic repulsion in
each COF Within an error limit the bulk COFs may be either serrated or inclined The interlayer
separations of all the COF stacks are in the range of 30 to 36 Aring and increase in the order AB
serratedinclined AA70 The motivation for proposing inclined stacking for COFs is that the
rhombohedral graphite (ABC stacking) is the inclined version of the layer-offset in natural graphite
(AB stacking) The instability of AA stacking was also suggested by the studies of elastic properties of
COF-1 and COF-5 by Zhou et al52 The shear modulus show negative signs for the vertical alignment of
COF layers while they are small but positive for the offset of layers
The change of stacking should be visible in their PXRD patterns because each space group has a
distinct set of symmetry imposed reflection conditions We simulated the XRD patterns of COFs in
their known and new configurations and on comparison with the experimental spectrum the new as
26
well as AA stacking showed good agreement5470 The slight layer-offsets are only able to make a few
additional peaks in the vicinity of existing peaks and some changes in relative intensities The
relatively broad experimental data covers nearly all the peaks of the simulated spectrum In other
words the broad experimental peaks are explainable with layer-offset
A detailed analysis on the interlayer stacking of HHTP-DPB COF made by Dichtel et al53 is very
complementary53 This work uses molecular mechanics extensively to define potential energy
surfaces of stacking of two layers from eclipsed to staggered covering all possible layer offsets Low
energy region was near to the eclipsed stacking which was then zoomed in to examine using DFT for
higher accuracy The far-eclipsed regions encounter no energy barriers with the near-eclipsed regions
which suggest the unlikeness of forming near-staggered layering Within DFT the eclipsed
geometries are at high energy compared to the slight offsets around AA (~ 13 ndash 25 Aring away) For a
hexagonal COF the energetically preferred offset could be in one of the six hexagonal sectors (or
circle sectors) of central angle 60 The preference of falling into one sector over the others is
arbitrary and may be determined by symmetry set by nature or external parameters Hence the
layering order in bulk could be serrated inclined spiral or purely by random The latter case better
known as turbostratic disorder is then more like a rule for 2D COFs rather than an exception caused
by external forces This also explains why the experimental PXRD patters have relatively broad peaks
The layer-offset not only change the internal pore structure but also affect interlayer exciton and
vertical charge transport in opto-electronic applications
About stacking stability the square COFs are expected not to be different from hexagonal COFs
because the local environment causing the shifts is nearly the same The DFTB based calculations
reported along with the synthesis of electron-transporting 2D-NiPc-BTDA COF supports this notion71
Two shifted configurations ndash at 07 and 14 Aring away from AA ndash appeared to be energetically preferred
over the AA stacking It appears that AA stacking is only possible for boron nitride-like structures
were adjacent layers have atoms with opposite charges in vertical direction
SURMOFs-2 a series of paddlewheel-based 2D MOFs possess a different symmetry than
solvothermally synthesized MOF-2 due to the differences in interlayer interactions Typically the
interaction between the paddlewheels favors P2 or C2 symmetry of bulk MOF-2 rather than the P4
symmetry of SURMOFs-2 Bader charge analysis reveals that electron distribution between the Cu-
paddlewheels is the lowest when they follow P4 symmetry This indicates the smallest possibility of
having a direct Cu-Cu bond between the paddlewheels In monolayer organic linkers undergo no
rotation with respect to metal dimers
27
X-ray diffraction of SURMOFs-2 made by Woumlll et al indicated P4 symmetry and a relatively small
interlayer separation This increases the repulsion between the linkers and enforces them to rotate
The rotations of each linker in a layer are controlled globally The rotated linkers in adjacent layers
increase London dispersion however a paddlewheel-led shift towards any side increases repulsion
thereby locking SURMOF-2 in P4 packing Hence with the special arrangement of rotated linkers the
linker-linker interaction overcomes the paddlewheel-paddlewheel interaction
P4 symmetry opens up the pores more clearly than P2 or C2 Additionally the metal sites that
typically host solvents are inaccessible Furthermore improvements such as functionalizing the linker
may be easily carried out
Based on our calculations on 2D materials we emphasize the role of van der Waals interactions in
determining the layer-to-layer arrangements The promise of reticular chemistry which is the
maintainability of structural integrity of the building blocks in the construction process is partly
broken in the case of SURMOFs-2 This is because due to repulsion the linkers undergo rotation with
respect to the carboxylic parts albeit keeping the topology
73 Structure-Property Relationships
We have studied a variety of materials from the classes of MOFs COFs and PAFs The structural
differences arise from the differences in the constituents andor their arrangements Properties in
general are interlinked with structural specifications Therefore it is beneficial to know the
relationship between the structural parameters and properties
The mass density is an intensive property of a material In the area of nanoporous materials a crystal
with low mass density has advantages in applications involving transport Definitely the mass density
decreases with increasing pore volume Still the number of atoms in the wall and their weights are
important factors The pore size does not relate directly to the atom counts The volume per atom
(inverse of atom density) another intensive property of a material obliquely gives porosity Figure
16 shows the variation of mass density with volume per atom (including the volume of the atom) for
MOFs COFs and PAFs MOFs show higher mass density compared to other materials of identical
atom density implying the presence of heavy elements It is to be noted that Cu-BTC has mass
density 184 times higher than COF-102 The presence of metals in the form of oxide clusters in MOFs
increases the mass density and decreases the volume per atom Note that the low-weighted MOF in
the discussion (MOF-205) is heavier than many of the PAFs and COFs The relatively high mass
density and small volume per atom of COF-300 is due to its 5-fold interpenetration whereas COF-202
has additional tert-butyl groups which do not contribute to the system shape but affect the mass
density and the volume per atom COF-102 and 103 have same topology but different centers (C and
28
Si respectively) The Si centers increases the cell volume and decreases mass density COF-105 (Si
centers) and COF-108 (C centers) additionally differ in their topologies (ctn and bor respectively) It
appears that bor topology expands the material compared to ctn PAFs hold the extreme cases PAF-
phnl has the highest atom and mass densities whereas PAF-qtph has the smallest atom and mass
densities
Figure 16 Mass density as a function of volume per atom for MOFs COFs and PAFs
The bulk modulus indicates the materialsrsquo resistance towards compression which should in principle
decrease with increasing porosity At the same time larger number of atoms making covalent
networks in unit volume should supply larger bulk moduli Thus differences in molecular contents
and architectures give rise to different bulk moduli It is interesting to see how the mechanical
stability of nanoporous materials is related with the atom density We have obtained a correlation
between bulk modulus (B) and volume per atom (VA) in the case of the studied MOFs COFs and PAFs
as follows
29
where k is a constant (see Figure 17) The inverse-volume correlation indicates that the materials
close to the fitting curve have average bond strengths (interaction energy between close atoms)
identical to each other independent of number of bonds bond order and branching Only Cu-BTC
COF-202 and COF-300 show significant differences from the fitted curve Cu-BTC has 17 times larger
bulk modulus compared to COF-102 of similar volume per atom which implies the substantially
higher strength of the bond network resulting from paddlewheel units and tbo topology
Interpenetration decreased the volume per atom however increased bulk modulus through
interlayer interactions in the case of COF-300 The tert-butyl groups in COF-202 do not transmit its
inherent stability to the COF significantly however decreases the volume per atom Comparison
between COF-102 (C centers) and COF-103 (Si centers) discloses that Si centers decrease the
mechanical stability Comparison between COF-105 (ctn) and COF-108 (bor) suggests that ctn
topology possess higher stability This indicates that local angular preferences can amend the
strength of the bulk material
Figure 17 Bulk modulus as a function of volume per atom for MOFs COFs and PAFs
Band gaps of all the studied materials show that they are semiconductors except of Cu-BTC which
has unpaired electrons at the Cu centers Our studies of the density of states (DOS) of bulk MOFs and
the cluster models that have finite numbers of connectors and linkers show that electronic structure
30
stays unchanged when going from cluster to periodic crystal In the case of 2D COFs the DOS of
monolayers and bulk materials were identical Interpenetrated COF-300 showed no difference in the
electronic structure in comparison with the non-interpenetrated structure Based on these results
we may reach into a premature conclusion that electronic structure of a solid is determined by its
constituent bonded network sufficiently large to include all its building units
HOMO-LUMO gap of the building units determine the band gap of a framework material We have
observed that PAFs nearly reproduced the HOMO-LUMO gaps of the linkers 3D COFs that are made
of more than one building unit show that the band gap is slightly smaller than the smallest of the
HOMO-LUMO gaps of the building units For example
TBPM (43 eV) + HHTP (34 eV) COF-105COF-108 (32 eV)
TBPM (43 eV) + TBST (139 eV) COF-202 (42 eV)
TAM (41 eV) + TA (26 eV) COF-300 (23 eV)
The compound names are taken from appendix E Additionally the band gaps decrease with
increasing number of conjugated rings similar to HOMO-LUMO gaps of the linkers
I believe that the studies in the thesis have helped to an extent to understand the structure
stability and properties of different classes of framework materials The benchmark structures we
studied have the essential features of the classes they represent Ab-initio based computational
studies of several periodic structures are exceptional and thus have its place in the literature
31
List of Abbreviations
ADF Amsterdam Density Functional code
BLYP Becke-Lee-Yang-Parr functional
B3LYP Becke 3-parameter Lee Yang and Parr functional
COF Covalent-Organic Framework
CP Coordination Polymer
CTF Covalent-Triazine Framework
DC Dispersion correction
DFT Density Functional Theory
DFTB Density Functional Tight-Binding
DOS Density of States
DOE (US) Department of Energy (United States)
DZP Double-Zeta Polarized basis set
GGA Generalized Gradient Approximation
LCAO Linear Combination of Atomic Orbitals
LPE Liquid Phase Epitaxy
MOF Metal-Organic Framework
PAF Porous Aromatic Framework
PBE Perdew-Burke-Ernzerhof functional
PXRD Powder X-ray Diffraction Pattern
QLDFT Quantized Liquid Density Functional Theory
RCSR Reticular Chemistry Structure Resource
SBU Secondary Building Unit
SCC Self-Consistent Charge
TZP Triple-Zeta Polarized basis set
SURMOF Surface-Metal-Organic Framework
32
List of Figures
Figure 1 ldquoAdamantinerdquo unit of diamond structure and diamond (dia) net 3
Figure 2 CU-BTC MOF and the corresponding tbo net 4
Figure 3 Fundamental building blocks and SBUs of Cu-BTC and extension of SBUs following crystal
structure 5
Figure 4 Schematic diagram of the single pore of MOF and An example of a simple cubic MOF in
which zinc-oxide tetrahedron are linked together by benzene linkers Atom colors blue ndash Zn red ndash O
grey ndash C white ndash H 6
Figure 5 Schematic diagram of interpenetrated and interwoven frameworks 7
Figure 6 Schematic diagram of SURMOF 8
Figure 7 SBUs and topologies of 2D COFs 9
Figure 8 Deconstructed building units their schematic representations and final geometries of
HKUST-1 (Cu-BTC) MOF-5 MOF-177 MOF-205 and MIL-53 11
Figure 9 AA and AB layer stacks of hexagonal layers 13
Figure 10 Stacking energy Es vs shift of adjacent layers for COF-5 Different stacking possibilities and
their energies are also shown 14
Figure 11 Schematic diagram of different building units forming 2D COFs 16
Figure 12 Schematic diagram of tetragonal and triangular building blocks forming lsquoctnrsquo or lsquoborrsquo
topologies 17
Figure 13 Diamond structure and various organic linkers to build up PAFs 18
Figure 14 The building units and schematic diagram of the formation of iso-reticular SURMOF series
20
Figure 15 Energy diagram of the metastable P4 and stable P2 structures 22
Figure 16 Mass density as a function of volume per atom for MOFs COFs and PAFs 28
Figure 17 Bulk modulus as a function of volume per atom for MOFs COFs and PAFs 29
33
References
(1) Morris R E Wheatley P S Angewandte Chemie-International Edition 2008 47 4966
(2) Li J-R Kuppler R J Zhou H-C Chemical Society Reviews 2009 38 1477
(3) Corma A Chemical Reviews 1997 97 2373
(4) Seo J S Whang D Lee H Jun S I Oh J Jeon Y J Kim K Nature 2000 404 982
(5) Lee J Kim J Hyeon T Advanced Materials 2006 18 2073
(6) Stein A Wang Z Fierke M A Advanced Materials 2009 21 265
(7) Velev O D Jede T A Lobo R F Lenhoff A M Nature 1997 389 447
(8) Beck J S Vartuli J C Roth W J Leonowicz M E Kresge C T Schmitt K D Chu C T
W Olson D H Sheppard E W McCullen S B Higgins J B Schlenker J L Journal of the
American Chemical Society 1992 114 10834
(9) Kresge C T Leonowicz M E Roth W J Vartuli J C Beck J S Nature 1992 359 710
(10) Ying J Y Mehnert C P Wong M S Angewandte Chemie-International Edition 1999 38
56
(11) Velev O D Kaler E W Advanced Materials 2000 12 531
(12) Stein A Microporous and Mesoporous Materials 2001 44 227
(13) Tanev P T Pinnavaia T J Science 1996 271 1267
(14) Yaghi O M OKeeffe M Ockwig N W Chae H K Eddaoudi M Kim J Nature 2003
423 705
(15) OKeeffe M Peskov M A Ramsden S J Yaghi O M Accounts of Chemical Research
2008 41 1782
(16) Delgado-Friedrichs O OKeeffe M Journal of Solid State Chemistry 2005 178 2480
(17) Delgado-Friedrichs O Foster M D OKeeffe M Proserpio D M Treacy M M J Yaghi
O M Journal of Solid State Chemistry 2005 178 2533
(18) OKeeffe M Yaghi O M Chemical Reviews 2012 112 675
(19) Tranchemontagne D J L Ni Z OKeeffe M Yaghi O M Angewandte Chemie-
International Edition 2008 47 5136
(20) Hayashi H Cote A P Furukawa H OKeeffe M Yaghi O M Nature Materials 2007 6
501
(21) Cote A P Benin A I Ockwig N W OKeeffe M Matzger A J Yaghi O M Science
2005 310 1166
(22) El-Kaderi H M Hunt J R Mendoza-Cortes J L Cote A P Taylor R E OKeeffe M
Yaghi O M Science 2007 316 268
(23) Rowsell J L C Yaghi O M Microporous and Mesoporous Materials 2004 73 3
(24) Hermes S Zacher D Baunemann A Woell C Fischer R A Chemistry of Materials
2007 19 2168
34
(25) Kuhn P Antonietti M Thomas A Angewandte Chemie-International Edition 2008 47
3450
(26) Ben T Ren H Ma S Cao D Lan J Jing X Wang W Xu J Deng F Simmons J M
Qiu S Zhu G Angewandte Chemie-International Edition 2009 48 9457
(27) Porezag D Frauenheim T Kohler T Seifert G Kaschner R Physical Review B 1995
51 12947
(28) Seifert G Porezag D Frauenheim T International Journal of Quantum Chemistry 1996
58 185
(29) Kohn W Sham L J Physical Review 1965 140 1133
(30) Parr R G Yang W Density-Functional Theory of Atoms and Molecules New York Oxford
University Press 1989
(31) Hohenberg P Kohn W Physical Review B 1964 136 B864
(32) Elstner M Porezag D Jungnickel G Elsner J Haugk M Frauenheim T Suhai S
Seifert G Physical Review B 1998 58 7260
(33) Zhechkov L Heine T Patchkovskii S Seifert G Duarte H A Journal of Chemical
Theory and Computation 2005 1 841
(34) Elstner M Hobza P Frauenheim T Suhai S Kaxiras E Journal of Chemical Physics
2001 114 5149
(35) Oliveira A F Seifert G Heine T Duarte H A Journal of the Brazilian Chemical Society
2009 20 1193
(36) Seifert G Joswig J-O Wiley Interdisciplinary Reviews-Computational Molecular Science
2012 2 456
(37) Heine T Rapacioli M Patchkovskii S Frenzel J Koester A M Calaminici P
Escalante S Duarte H A Flores R Geudtner G Goursot A Reveles J U Vela A Salahub D
R deMon deMon-nano edn deMon-nano 2009
(38) BCCMS B DFTB+ - Density Functional based Tight binding (and more)
(39) Monkhorst H J Pack J D Physical Review B 1976 13 5188
(40) SCM Amsterdam Density Functional 2012
(41) University of Karlsruhe and Forschungszentrum Karlsruhe gGmbH - TURBOMOLE V63
2011 2007
(42) Dovesi R Saunders V R Roetti C Orlando R Zicovich-Wilson C M Pascale F
Civalleri B Doll K Harrison N M Bush I J DrsquoArco P Llunell M CRYSTAL09 Users Manual
University of Torino Torino 2009 2009
(43) Wadt W R Hay P J Journal of Chemical Physics 1985 82 284
(44) Roy L E Hay P J Martin R L Journal of Chemical Theory and Computation 2008 4
1029
35
(45) Ehlers A W Bohme M Dapprich S Gobbi A Hollwarth A Jonas V Kohler K F
Stegmann R Veldkamp A Frenking G Chemical Physics Letters 1993 208 111
(46) Garberoglio G Skoulidas A I Johnson J K Journal of Physical Chemistry B 2005 109
13094
(47) Han S S Mendoza-Cortes J L Goddard W A III Chemical Society Reviews 2009 38
1460
(48) Getman R B Bae Y-S Wilmer C E Snurr R Q Chemical Reviews 2012 112 703
(49) Cote A P El-Kaderi H M Furukawa H Hunt J R Yaghi O M Journal of the American
Chemical Society 2007 129 12914
(50) Wan S Guo J Kim J Ihee H Jiang D Angewandte Chemie-International Edition 2008
47 8826
(51) Wan S Guo J Kim J Ihee H Jiang D Angewandte Chemie-International Edition 2009
48 5439
(52) Zhou W Wu H Yildirim T Chemical Physics Letters 2010 499 103
(53) Spitler E L Koo B T Novotney J L Colson J W Uribe-Romo F J Gutierrez G D
Clancy P Dichtel W R Journal of the American Chemical Society 2011 133 19416
(54) Lukose B Kuc A Heine T Chemistry-a European Journal 2011 17 2388
(55) Uribe-Romo F J Hunt J R Furukawa H Klock C OKeeffe M Yaghi O M Journal of
the American Chemical Society 2009 131 4570
(56) Schmid R Tafipolsky M Journal of the American Chemical Society 2008 130 12600
(57) Patchkovskii S Heine T Physical Review E 2009 80
(58) Shekhah O Wang H Paradinas M Ocal C Schuepbach B Terfort A Zacher D
Fischer R A Woell C Nature Materials 2009 8 481
(59) Li H Eddaoudi M Groy T L Yaghi O M Journal of the American Chemical Society
1998 120 8571
(60) Rappe A K Casewit C J Colwell K S Goddard W A Skiff W M Journal of the
American Chemical Society 1992 114 10024
(61) Frauenheim T Seifert G Elstner M Hajnal Z Jungnickel G Porezag D Suhai S
Scholz R Physica Status Solidi B-Basic Research 2000 217 41
(62) Elstner M Cui Q Munih P Kaxiras E Frauenheim T Karplus M Journal of
Computational Chemistry 2003 24 565
(63) Heine T Dos Santos H F Patchkovskii S Duarte H A Journal of Physical Chemistry A
2007 111 5648
(64) Sternberg M Zapol P Curtiss L A Molecular Physics 2005 103 1017
(65) Zhang C Zhang Z Wang S Li H Dong J Xing N Guo Y Li W Solid State
Communications 2007 142 477
36
(66) Munch W Kreuer K D Silvestri W Maier J Seifert G Solid State Ionics 2001 145
437
(67) Bahr D F Reid J A Mook W M Bauer C A Stumpf R Skulan A J Moody N R
Simmons B A Shindel M M Allendorf M D Physical Review B 2007 76
(68) Amirjalayer S Tafipolsky M Schmid R Journal of Physical Chemistry C 2011 115
15133
(69) Amirjalayer S Snurr R Q Schmid R Journal of Physical Chemistry C 2012 116 4921
(70) Lukose B Kuc A Frenzel J Heine T Beilstein Journal of Nanotechnology 2010 1 60
(71) Ding X Chen L Honsho Y Feng X Saenpawang O Guo J Saeki A Seki S Irle S
Nagase S Parasuk V Jiang D Journal of the American Chemical Society 2011 133 14510
37
Appendix A
Review Covalently-bound organic frameworks
Binit Lukose and Thomas Heine
To be submitted for publication after revision
Contents
1 Introduction
2 Synthetic achievements
21 Covalent Organic Frameoworks (COFs)
22 Covalent-Triazine Frameworks (CTFs)
23 Porous Aromatic Frameworks (PAFs)
24 Schemes for synthesis
25 List of materials
3 Studies of the underlying structure and properties of COFs
4 Applications
41 Gas storage
411 Porosity of COFs
412 Experimental measurements
413 Theoretical preidctions
414 Adsorption sites
415 Hydrogen storage by spillover
42 Diffusion and selectivity
43 Suggestions for improvement
431 Geometry modifications
432 Metal doping
433 Functionalization
5 Conclusions
6 List and pictures of chemical compounds
38
1 Introduction
Nanoporous materials have perfectly ordered voids to accommodate to interact with and to
discriminate molecules leading to prominent applications such as gas storage separation and sieving
catalysis filtration and sensoring1-4 They are defined as porous materials with pore diameters less
than 100 nm and are classified as micropores of less than 2 nm in diameter mesopores between 2
and 50 nm and macropores of greater than 50 nm The internal geometry that determines pore size
and surface area can be precisely engineered at molecular scales Reticular synthetic methods
suggested by Yaghi and co-workers5-7 can accomplish pre-designed frameworks The procedure is to
select rigid molecular building blocks prudently and assemble them into destined networks using
strong bonds
Several types of materials have been synthesized using reticular chemistry concepts One prominent
group which is much appreciated for its credentials is the Metal-Organic Frameworks (MOFs)8-13 in
which metal-oxide connectors and organic linkers are judiciously selected to form a large variety of
frameworks The immense potential of MOFs is facilitated by the fact that all building blocks are
inexpensive chemicals and that the synthesis can be carried out solvothermally The progress in MOF
synthesis has reached the point that some of the MOFs are commercially available Several MOFs of
ultrahigh porosity have also been successfully synthesized notably the non-interpenetrated IRMOF-
74-XI which has a pore size of 98 Aring Scientists are also able to synthesize MOFs from cheap edible
natural products14 Since the discovery of MOFs many other crystalline frameworks have been
synthesized using reticular chemistry such as Metal-Organic Polyhedra (MOP)15-19 Zeolite
Imidazolate Frameworks (ZIFs)20-22 and Covalent Organic Frameworks (COFs)23-29
COFs are composed of covalent building blocks made of boron carbon oxygen and hydrogen in
many cases also including nitrogen or silicon stitched together by organic subunits The atoms are
held together by strong covalent bonds Depending on the selection of building blocks the COFs may
form in two (2D) or three (3D) dimensions Planar building blocks are the constituents of 2D COFs
whereas for the formation of 3D COFs typically tetragonal building blocks are involved High
symmetric covalent linking as it is perceived in reticular chemistry was confirmed for the end
products5
Unlike the case of supramolecular assemblies the absence of noncovalent forces between the
molecular building units endorses exceptional rigidity and stability for COFs They are in general
thermally stable above 4000 C Also in COFs the stitching of molecular building units guarantees an
39
increased order and allows control over porosity and composition Without any metals or other
heavy atoms present they exhibit low mass densities This gives them advantages over MOFs in
various applications for example higher gravimetric capacities for gas storage3031 The lowest
density crystal known until today is COF-10824 In addition COFs exhibit permanent porosity with
specific surface areas that are comparable to MOFs and hence larger than in zeolites and porous
silicates
MOF and COF crystals possess long range order although COFs have been achieved so far only at the
μm scale Reversible solvothermal condensation reactions are credited for the high order of
crystallinity Other new framework materials such as Covalent Triazine Frameworks (CTFs) and
Porous Aromatic Frameworks (PAFs) differed in ways of synthesis The former was prepared by
ionothermal polymerization while the latter was produced by coupling reactions PAFs lack long
range order in the crystals as a result of the irreversible synthesis Nevertheless many of the
materials are promisingly good for applications In this review we intend to discuss the synthetic
achievements of COF CTFs and PAFs and studies on their structure properties and prominent
applications
For easy understanding of the atomic geometry of a material the reader is referred to Table 1 which
gives the names of the framework materials the reactants and the synthesis scheme Figure 5 shows
the reactants and Figure 4 shows the reaction schemes Example reactions are given in Figure 1-3
Abbreviations of each chemical compound are given in Section 6
2 Synthetic achievements
21 Covalent Organic Frameworks (COFs)
In the year 2005 Yaghi et al23 achieved the crystallization of covalent organic molecules in the form
of periodic extended layered frameworks The condensation of discrete molecules of different sizes
enabled the synthesis of micro- and mesoporous crystalline COFs27 The selection of molecules with 2
and 3 connection sites for synthesis led to the formation of hexagonal layered geometries32 Yaghi et
al have synthesized half a dozen three-dimensional crystalline COFs solvothermally using tetragonal
building blocks containing 4 connection sites since 2007242829 Figure 1 shows a few examples of 2D
and 3D COFs formed by the self-condensation of boronic acids such as BDBA TBPM and TBPS and co-
condensation of the same boronic acids with HHTP
40
Figure 1 Solvothermal condensation (dehydration) of monomers to form 2D and 3D COFs Atom colors are boron yellow oxygen red yellow (big) silicon grey carbon
Alternate synthetic procedures were also exploited for production and functionalization of COFs
Lavigne et al33 showed that the synthesis of COFs using polyboronate esters is facile and high-yielded
41
Boronate esters often contain multiple catechol moieties which are prone to oxidation and are
insoluble in organic solvents Dichtel et al3435 presented Lewis acid-catalysis procedures to form
boronate esters of catechols and arylboronic acids without subjecting to oxidation Cooper et al36
successfully utilized microwave heating techniques for rapid production (~200 times faster than
solvothermal methods) of both 2D and 3D COFs The syntheses of phthalocyanine3437 or porphyrin38
based square COFs have been reported in literature The latter was noticed for its time-dependent
crystal growth which also affects its pore parameters
Abel et al39 reported the ability to form COFs on metallic surfaces COF surfaces (SCOFs) have been
formed by molecular dehydration of boroxine and triphenylene on a silver surface Albeit with some
defects the materials showed high temperature stability allowing to proceed with functionalization
Lackinger et al40 realized the role of substrates on the formation of surface COFs from the surface-
generated radicals Halogen substituted polyaromatic monomers can be sublimed onto metal
substrates and ultimately turned into a COF after homolysis and intermolecular colligation
Chemically inert graphite surfaces cannot support the homolysis of covalent carbon-halogen bonds
and thus cannot initiate the subsequent association of radicals COF layers can be formed onto
Cu(111)40 and Au(111)41 surfaces but with lower crystal ordering Beton et al41 deposited the
monomers on annealed Au(111) surfaces which supplied surface mobility for the monomers and
subsequently increased porosity and surface coverage of the covalent networks Unlike the bulk form
at the surface the monomers were self-organized onto polygons with 4 to 8 edges Such a template
was able to organize a sublimed layer of C60 fullerenes the number of C60 molecules adsorbed in a
cavity was correlated to the size of the polygon
In 2009 Sassi et al42 studied the problem of crystal growth of COFs on substrates They calculated
the reaction mechanism of 14-benzenediboronic acid (BDBA) on Ag(111) surface self-condensation
of BDBA molecules in solvents leads to the formation of two-dimensional (the planar) stacked COF-1
For the surface COFs the study using Density Functional Theory reveals that there are neither
preferred adsorption sites for the molecules nor a charge transfer between the molecules and the
surface Hence the electronic structure of the molecules remains unchanged and the role of the
metal surface is limited in providing the planar nature for the polymer The free reaction enthalpy
(Gibbs free energy) difference of the reaction ∆G = ∆H ndash T∆S is approximated using the harmonic
approximation taking into account the geometrical restrictions of the metal surface and the entropic
contributions of the released water molecules As result the formation of SCOF-1 has been found to
be exothermic if more than six monomers are involved Ourdjini et al43 reported the polymerization
of 14-benzenediboronic acid (BDBA) on different substrates (Ag(111) Ag(100) Au(111) and Cu(111))
and at different source and substrate temperatures to follow how molecular flux and adsorption-
42
diffusion environments should be controlled for the formation of polymers with the smallest number
of structural defects High flux is essential to avoid H-bonded supramolecular assemblies of
molecules and the substrate temperature needs to be optimized to allow the best surface diffusion
and longest residential time of the reactants Achieving these two contradictory conditions together
is a limitation for some substrates eg for copper Silver was found to be the best substrate for
producing optimum quality polymers Controlling the growth parameters is important since
annealing cannot cure defects such as formation of pentagons heptagons and other non-hexagonal
shapes which involved strong covalent bonds
Ditchel et al44 successfully grew COF films on monolayer graphene supported by a substrate under
operationally simple solvothermal conditions The films have better crystallinity compared to COF
powders and can facilitate photoelectronic applications straightaway Zamora et al45 have achieved
exfoliation of COF layers by sonication The thin layered nanostructures have been isolated under
ambient conditions on surfaces and free-standing on carbon grids
A noted characteristic of COFs is its organic functionality Judiciously selected organic units ndash pyrene
and triphenylene (TP) ndash for the production of TP COF can not only harvest photons over a wide range
but also facilitate energy migration over the crystalline solid25 Polypyrene (PPy)-COF that consists of
a single aromatic entity ndash pyrene- is fluorescent upon the formation of excimers and hence undergo
exciton migration across the stacked layers26 A nickel(II)-phthalocyanine based layered square COF
that incorporates phenyl linkers and forms cofacially stacked H-aggregates was reported to absorb
photons over the solar spectrum and transport charges across the stacked layers37 COF-366 and
COF-66 showed excellent hole mobility of 81 and 30 cm2 V-1s-1 surpassing that of all polycrystalline
polymers known until today46 A first example of an electron-transporting 2D COF was reported
recently47 This square COF was built from the co-condensation of nickel(II) phthalocyanine and
electron-deficient benzothiadiazole With the nearly vertical arrangement of its layers it exhibits an
electron mobility of 06 cm2 V-1s-1 The enhanced absorbance is over a wide range of wavelengths up
to 1000 nm In addition it exhibits panchromatic photoconductivity and high IR sensitivity
Lavigne et al48 achieved the condensation of alkyl (methyl ethyl and propyl) substituted organic
building units with boronic acids forming functionalized COFs The alkyl groups contribute for higher
molar adsorption of H2 however the increased mass density of the functionalized COFs cause for
decreased gravimetric storage capacities Boronate ester-linked COFs are stable in organic solvents
however undergo rapid hydrolysis when exposed in water49 In particular the instability of COF-1
upon H2O adsorption shall be mentioned here50 Lanni et al claimed that alkylation could bring
hydrolytic stability into COFs49
43
Functionalization and pore surface engineering in 2D COFs can be improved if azide appended
building blocks are stitched together in click reactions with alkynes51 Control over the pore surface
is made possible by the use of both azide appended and bare organic building units the ratios of
which is matching with the final amount of functionalization in the pore walls The click reactions of
azide groups with alkynes within the pores lead to the formation of triazole-linked groups on the
pore surfaces This strategy also gives the relief of not condensing the already functionalized building
units Jiang et al have asserted eclipsed layering for most of the final COFs using powder X-ray
diffraction pattern Exceptions are the hexagonal functionalized COFs with relatively high azide-
content (ge 75) Their weak XRD signals suggest possible layer-offset or de-lamellation Although
functionalization on pore surfaces exhibits the nature of lowering the surface area the possibility to
add reactive groups within the pores can be utilized for gas-sorption selectivity The authors have
claimed a 16 fold selectivity of CO2 over N2 using 100 acetyl-rich COF-5 compared to the pure COF-5
The range of the click reaction approach is so wide that relatively large chromophores can be
accommodated in the pores thereby making COF-5 fluorescent
Functionalization of 3D COFs has just reported in 201252 Dichtel et al used a monomer-truncation
strategy where one of the four dihydroxyborylphenyl moieties of a tetrahedral building unit was
replaced by a dodecyl or allyl functional group Condensation of the truncated and the pure
tetrahedral blocks in a certain ratio created a functionalized 3D COF The degree of functionalization
was in proportion with the ratio The crystallinity was maintained except for high loading (gt 37) of
truncated monomers The pore volume and the surface area were decreased as a function of loading
level
A heterogeneous catalytic activity of COFs is achieved with an imine-linked palladium doped COF by
enabling Suzuki-Miyaura coupling reaction within the pores53 The COF-LZU1 has a layered geometry
that has the distance between nitrogen atoms in the neighboring linkers in a level (~37 Aring) sufficient
to accommodate metal ions Palladium was incorporated by a post-treatment of the COF PdCOF-
LZU1 was applied to catalyze Suzuki-Miyaura coupling reaction This coupling reaction is generally
used for the facile formation of C-C bonds54 The increased structural regularity and high insolubility
in water and organic solvents are adequate qualities of imine-linked COFs to perform as catalysts
Experiments with the above COF show a broad scope of the reactants excellent yields of the
products and easy recyclability of the catalyst
The comparatively higher thermal stability of COFs is often noted and is explainable with their strong
covalent bonds The reversible dehydrations for the formation of most of the COFs point to their
instability in the presence of water molecules This has been tested and verified for some layered
COFs with boronate esters49 Such COFs are stable as hosts for organometallic molecules55 COF-102
44
framework was found to be stable and robust even in the presence of highly reactive cobaltocenes
The highly stable ferrocenes appear to have an arrangement within the framework led by π-π
interactions
22 Covalent Triazine Frameworks (CTFs)
In the year 2008 Thomas et al56 synthesized a crystalline extended porous nanostructure by
trimerization of polytriazine monomers in ionothermal conditions With the catalytic support of ZnCl2
three aromatic nitrile groups can be trimerized to form a hexagonal C3N3 ring The resulting structure
known as Covalent Triazine-based Frameworks (CTFs) is similar to 2D COFs in geometry and atomic
composition (see Figure 2) Usage of larger non-planar monomers at higher amounts of ionic salts
however led to the formation of contorted structures Interestingly those structures showed
enhanced surface area and pore volume The trimerization of monomers that lack a linear
arrangement of nitrile groups ended up as organic polymer networks Later the same group
introduced mesoporous channels onto a CTF (CTF-2)57 by increasing temperature and ion content
The resulting structure however was amorphous It is concluded that the reaction parameters and
the amount of salt play a crucial role for tuning the porosity and controlling the order of the material
respectively58
Figure 2 Trimerization of nitrile monomers to form CTF-1 Atom colors blue N grey C white H
Ben et al59 followed ionothermal conditions for the targeted synthesis of a 3D framework using
tetraphenylmethane block and a triangular triazine ring The end product was amorphous thermally
stable up to 400 0C and could be applied for the selective sorption of benzene over cyclohexane On a
later synthetic study Zhang et al60 reported that this polymerization could be accomplished in short
45
reaction times under microwave enhanced conditions The template-free high temperature dynamic
polymerization reactions constitute irreversible carbonization reactions coupled with reversible
trimerization of nitriles The irreversibility encounter in these condensation reactions is responsible
for the production of frameworks as amorphous solids6162
An amorphous yet microporous CTF was found to be able to catalyze the oxidation of glycerol with
Pd nanoparticles that are dispersed in the framework63 This solid catalyst appears to be strong
against deactivation and selective toward glycerate compared to Pd supported activated carbon This
is attributed to the relatively high stability of Pd nanoparticles that benefit from the presence of
nitrogen functionalities in CTF An advancement on selective oxidation of methane to methanol at
low temperature is accomplished by using an amorphous bipyridyl-rich platinum-coordinated CTF as
catalyst64
23 Porous Aromatic Frameworks (PAFs)
a notably high surface area (BET surface area of 5600 m2g-1 Langmuir surface area of 7100 m2g-1)65
PAF-1 has been synthesized in a rather simple way using a nickel(0)-catalyzed Yamamoto-type66
Ullmann cross-coupling reaction67(see Figure 3) The material is thermally (up to 520 0C in air) and
hydrothermally (in boiled water for a minimum of seven days) stable The framework exposes all
faces and edges of the phenyl rings to the pores and hence the high surface area can be achieved
while its high stability is inherited from the parent diamond structure The synthesized material was
verified as disordered and amorphous yet with uniform pore diameters The excess H2 uptake
capacity of PAF-1 was measured as 70 wt at 48 bar and 77 K It can adsorb 1300 mg g-1 CO2 at 40
bar and room temperature PAF-1 was also tested for benzene and toluene adsorption
Figure 3 Coupling reaction using halogenated tetragonal blocks for the formation of diamond-like PAF-1 Atom colors red Br grey C white H
46
An attempt to synthesize a diamondoid with quaterphenyl as linker is described recently68 A
tetrahedral unit and a linear linker each of them has two phenyl rings were polymerized using the
Suzuki-Miyaura coupling reaction54 The product (PAF-11) however stayed behind theoretical
predictions and performed poorly pointing to its shortcomings such as short-range order distortion
and interpenetration of phenyl rings Nevertheless the amorphous solid displayed high thermal and
chemical stabilities proneness for adsorbing methanol over water and usability for eliminating
harmful aromatic molecules
PAF-569 is a two-dimensional hexagonal organic framework synthesized using the Yamamoto-type
Ullmann reaction This material is composed only of phenyl rings however lack long range order
because of the distortion of the phenyl rings and kinetics controlled irreversible coupling reaction It
retains a uniform pore diameter and provides high thermal and chemical stability despite its
amorphous nature PAF-5 was suggested for adsorbing organic pollutants at saturated vapour
pressure and room temperature
Co-condensation reactions with piperazine enabled nucleophilic substitution of cyanuric chloride to
form a covalently linked porous organic framework PAF-670 The synthesis in mild conditions yields a
product with uniform morphology and a certain degree of structural regularity Being nontoxic this
material was tested for drug delivery thereby stepping into biomedical applications of covalently
linked organic frameworks Primary tests suggest that PAF-6 could be promoted for drug release for
its permanent porosity and facile functionalization In comparison with MCM-4171 a widely tested
inorganic framework PAF-6 performed equally or even superiorly
24 Schemes for synthesis
The majority of the COFs were synthesized using solvothermal step-by-step condensation
(dehydration) reactions The method incorporates reversibility and is applicable for supplying long
range order in COF materials The reactants generally consist of boronic acids and hydroxy-
polyphenyl (catechol) compounds Extended porous networks are obtained when O-B or O-C bonds
are formed into 5 or 6-membered rings after dehydration (scheme 1) Dehydration may also be
carried out using microwave synthesis Alternately boronic acid can react with catechol acetonide in
presence of a Lewis acid catalyst The molecules link to each other after the liberation of acetone and
water (scheme 2) The utilization of phenyl-attached formyl and amino groups as building units
results in the formation of imines after dehydration (scheme 3) A desired arrangement of molecular
arrays on surfaces can be fulfilled by depositing planar molecules on metallic surfaces followed by
covalent linking using annealing
47
Isocyanate groups follow trimerization to form poly-isocyanurate rings (scheme 4) Cyclotrimerization
of monomers containing nitrile groups or cyanate groups is prone to generate C3N3 rings (scheme 5)
However the ionothermal synthesis of them resulted with amorphous materials Unique bond
formation is often not achieved throughout the material and thus the crystal lacks long-range order
Analogously coupling reactions such as Ullmann and Suzuki-Miyaura give rise to amorphous
products However they are adequate in producing C-C bonds when halogen-substituted compounds
are reacted (scheme 6) Nucleophilic substitution may also be used for framing networks in cases
like reactants are formed into nucleophiles and ions after release of their end-groups (scheme 7)
48
Figure 4 Different schemes reported for the synthesis of covalently-bound organic frameworks
49
25 List of synthesized materials
Table 1 List of synthesized materials in the literature their building blocks synthesis methods measured surface area [m
2 g
-1] pore volume [cm
3 g
-1] and pore size [Aring]
COF Names Reactants Synthesis Surface
Area
Pore
volume
Pore
size
COF-1 BDBA Solvothermal condensation235072
scheme 1
711 62850 032
03650
9
COF-5 BDBA HHTP Solvothermal condensation23
scheme 1
1590 0998 27
Microwave3673 scheme 1 2019
BDBA TPTA Lewis acid catalysis35 TPTA
COF-6 BTBA HHTP Solvothermal condensation27
scheme 1
980 (L) 032 64
COF-8 BTPA HHTP Solvothermal condensation27
scheme 1
1400 (L) 069 187
COF-10 BPDA HHTP Solvothermal condensation27
scheme 1
2080 (L) 144 341
BPDA TPTA Lewis acid catalysis35 scheme 2
COF-18Aring BTBA THB Facile dehydration3348 scheme 1 1263 069 18
COF-16Aring BTBA alkyl-THB
(alkyl = CH3)
Facile dehydration48 scheme 1 753 039 16
COF-14Aring BTBA alkyl-THB
(alkyl = C2H5)
Facile dehydration48 scheme 1 805 041 14
COF-11Aring BTBA alkyl-THB
(alkyl = C3H7)
Facile dehydration48 scheme 1 105 0052 11
50
SCOF-1 BDBA Substrate-assisted synthesis39
scheme 1
SCOF-2 BDBA HHTP Substrate-assisted synthesis39
scheme 1
TP COF PDBA HHTP Solvothermal condensation25
scheme 1
868 079 314
PPy-COF PDBA Solvothermal condensation26
scheme 1
923 188
TBB COF TBB (on Cu(111) and
Ag(110) surfaces)
Surface polymerisation40 scheme
6
TBPB COF TBB (on Au(111)
surface)
Surface polymerisation41 scheme
6
BTP COF BTPA THDMA Solvothermal condensation72
scheme 1
2000 163 40
HHTP-DPB COF DPB HHTP Solvothermal condensation73
scheme 1
930 47
PICU-A DMBPDC Cyclotrimerization74 scheme 4
PICU-B DCF Cyclotrimerization74 scheme 4
COF-LZU1 DAB TFB Solvothermal condensation53
scheme 3
410 054 12
PdCOF-LZU1 COF-LZU1 PdAc Facile post-treatment53 146 019 12
XN3-COF-5 X N3-BDBA (100-X)
BDBA HHTP
Solvothermal condensation51
scheme 1
2160
(X=5)
1865 (25)
1722 (50)
1641 (75)
1421
(100)
1184
1071
1016
0946
0835
295
276
259
258
227
51
XAcTrz-COF-5 XN3-COF-5 PAc Click reaction51 2000
(X=5)
1561 (25)
914 (50)
142 (75)
36 (100)
1481
0946
0638
0152
003
298
243
156
153
125
XBuTrz-COF-5 XN3-COF-5 HP Click reaction51
XPhTrz-COF-5 XN3-COF-5 PPP Click reaction51
XEsTrz-COF-5 XN3-COF-5 MP Click reaction51
XPyTrz-COF-5 XN3-COF-5 PyMP Click reaction51
COF-42 DETH TFB Solvothermal condensation75
scheme 3
710 031 23
COF-43 DETH TFPB Solvothermal condensation75
scheme 3
620 036 38
CTF-1 DCB Ionothermal trimerization56
scheme 5
791 040 12
CTF-2 DCN Ionothermal trimerization57
scheme 5
90 8
mp-CTF-2 2255 151 8
CTF (DCP) DCP Ionothermal trimerization64
scheme 5
1061 0934 14
K2[PtCl4]-CTF DCP K2[PtCl4] Trimerization (scheme 5) +
coordination64
Pt-CTF DCP Pt Trimerization (scheme 5) +
coordination64
PAF-5 TBB Yamamoto-type Ullmann cross-
coupling reaction69 scheme 6
1503 157 166
52
PAF-6 PA CA Nucleophilic substitution70
scheme 7
1827 118
Pc-PBBA COF PcTA BDBA Lewis acid catalysis34 scheme 2 506 (L) 0258 217
NiPc-COF OH-Pc-Ni BDBA Solvothermal condensation37
scheme 1
624 0485 190
XN3-NiPc-COF OH-Pc-Ni X N3-BDBA
(100-X) BDBA
Solvothermal condensation51
scheme 1
XEsTrz-NiPc-
COF
XN3-NiPc-COF MP Click reaction51
ZnP COF TDHB-ZnP THB Solvothermal condensation38
scheme 1
1742 1115 25
NiPc-PBBA COF NiPcTA BDBA Lewis acid catalysis35 scheme 2 776
2D-NiPc-BTDA
COF
OHPcNi BTDADA Solvothermal condensation47
scheme 1
877 22
ZnPc-Py COF OH-Pc-Zn PDBA Solvothermal condensation
scheme 1
420 31
ZnPc-DPB COF OH-Pc-Zn DPB Solvothermal condensation
scheme 1
485 31
ZnPc-NDI COF OH-Pc-Zn NDI Solvothermal condensation
scheme 1
490 31
ZnPc-PPE COF OH-Pc-Zn PPE Solvothermal condensation
scheme 1
440 34
COF-366 TAPP TA Solvothermal condensation46
scheme 3
735 032 12
COF-66 TBPP THAn Solvothermal condensation46
scheme 1
360 020 249
53
COF-102 TBPM Solvothermal condensation24
scheme 1
3472 135 115
Microwave36
scheme 1
2926
COF-102-C12 TBPM trunk-TBPM-R
(R=dodecyl)
Solvothermal condensation52
scheme 1
12
COF-102-allyl TBPM trunk-TBPM-R
(R=allyl)
Solvothermal condensation52
scheme 1
COF-103 TBPS Solvothermal condensation24
scheme 1
4210 166 125
COF-105 TBPM HHTP Solvothermal condensation24
scheme 1
COF-108 TBPM HHTP Solvothermal condensation24
scheme 1
COF-202 TBPM TBST Solvothermal condensation28
scheme 1
2690 109 11
COF-300 TAM TA Solvothermal condensaion29
scheme 3
1360 072 72
PAF-1 TkBPM-X (X=C) Yamamoto-type Ullmann cross-
coupling reaction65 scheme 6
5600
PAF-2 TCM cyclotrimerization59 scheme 5 891 054 106
PAF-3 TkBPM-X (X=Si) Yamamoto-type Ullmann cross-
coupling reaction76 scheme 6
2932 154 127
PAF-4 TkBPM-X (X=Ge) Yamamoto-type Ullmann cross-
coupling reaction76 scheme 6
2246 145 118
PAF-11 TkBPM-X (X=C) BPDA Suzuki coupling reaction68 704 0203 166
54
scheme 6
3 Studies of structure and properties of COFs
31 2D COFs
Following the literature of the years 2005-2010 2D COFs were believed to have high symmetric AA
(eclipsed ndash P6mmm symmetry) stacking as boron nitride (h-BN) [REF] An exception was COF-1
which was considered to have AB (staggered ndash P63mmc) stacking as graphite[REF] The AA stacking
maximizes the attractive London dispersion interaction between the layers an important
contribution to the stacking energy At the same time AA stacking always suffers repulsive Coulomb
force between the layers due to the polarized connectors as the distance between atoms exposing
the same charge is closest It should be noted that in boron nitride boron atoms serve as nearest
neighbors to nitrogen atoms in adjacent layers The Coulomb interaction has ruled out possible
interlayer eclipse between atoms of alike charges Based on these facts we modeled77 a set of 2D
COFs with different possible stacks Each layer was drifted with respect to the neighboring layers in
directions symmetric to the hexagonal pore and we calculated the total energies and their Coulombic
contributions The AA stacking version was found to have the highest Coulombic repulsion in each
COF The lattice types with the layers shifted to each other by ~14 Aring approximately the bond length
between two atoms in a 2D COF were found to be more stable compared to AA or AB In this low-
symmetry configuration the hexagonal rings in the pore walls undergo staggering with that of
adjacent layers Within an error limit the bulk COFs may be either serrated or inclined as shown in
Figure 5 The interlayer separations of all the COF stacks are in the range of 30 to 36 Aring and increase
in the order AB serratedinclined AA32 The motivation for proposing inclined stacking for COFs is
that the rhombohedral graphite (ABC stacking) is the inclined version of the layer-offset in natural
graphite (AB stacking) The instability of AA stacking was also suggested by the studies of elastic
properties of COF-1 and COF-5 by Zhou et al78 The shear modulus show negative signs for the
vertical alignment of COF layers while they are small but positive for the offset of layers
55
Figure 5 Different stacks of a 2D covalent organic framework COF-1 The atoms are shown as Van der Waals spheres
The different stacking modes should in principle be visible in their PXRD patterns as each space
group has a distinct set of symmetry imposed reflection conditions We simulated the XRD patterns
of COFs in their known and new configurations and on comparison with the experimental spectrum
the new as well as AA stacking showed good agreement3279 However the slight layer-offsets in
conjunction with the large number of atoms in the unit cells change the PXRD spectrum only by the
appearance of a few additional peaks all of them in the vicinity of existing signals and by changes in
relative intensities Unfortunately the low resolution of the experimental data does now allow
distinguishing between the different stackings as the broad signals cover all the peaks of the
simulated spectrum
A detailed analysis on the interlayer stacking of HHTP-DPB COF has been made by Dichtel et al73 is
very complementary73 This work uses molecular mechanics extensively to define potential energy
surfaces of stacking of two layers from eclipsed to staggered covering all possible layer offsets The
low energy region was near to the eclipsed stacking which was then zoomed in to examine using DFT
for higher accuracy The far-eclipsed regions encounter no energy barriers with the near-eclipsed
regions which suggest the unlikeness of forming near-staggered layering Within DFT the eclipsed
geometries are at high energy compared to the slight offsets around AA (~ 13 ndash 25 Aring away) For a
hexagonal COF the energetically preferred offset could be in one of the six hexagonal sectors (or
circle sectors) of central angle 60 The preference of falling into one sector over the others is
arbitrary and may be determined by symmetry set by nature or external parameters Hence the
layering order in bulk could be serrated inclined spiral or purely by random The latter case better
known as turbostratic disorder is then more like a rule for 2D COFs rather than an exception caused
by external forces This also explains why the experimental PXRD patters have relatively broad peaks
The layer-offset may not only change the internal pore structure but also affect interlayer exciton
and vertical charge transport in opto-electronic applications
56
Concerning the stacking stability the square 2D COFs are expected not to be different from
hexagonal COFs because the local environment causing the shifts is nearly the same DFTB based
calculations reported along with the synthesis of electron-transporting 2D-NiPc-BTDA COF supports
this notion47 Two shifted configurations ndash at 07 and 14 Aring away from AA ndash appeared to be
energetically preferred over the AA stacking It appears that AA stacking is only possible for boron
nitride-like structures were adjacent layers have atoms with opposite charges in vertical direction In
analogy to BN layers it might be possible to force AA stacking by the introduction of alkalis in
between the layers
32 3D COFs
3D COFs in general are composed of tetragonal and triangular building blocks So far that their
synthesis leads to two high symmetry topologies ndash ctn and bor - in high yields24 The two topologies
differ primarily in the twisting and bulging of their components at the molecular level The
thermodynamic preference of one topology over the other may result from the kinetic entropic and
solvent effects and the relative strain energies of the molecular components It is straight-forward to
state that the effects at the molecular level crucial crucial in the bulk state since transformation from
one net to the other is impossible without bond-breaking There has not been any detailed study on
this matter experimentally or theoretically
Schmid et al8182 have developed force-field parameters from first principles calculations using
Genetic Algorithm approach The parameters developed for cluster models of COF-102 can
reproduce the relative strain energies in sufficient accuracies and be extended to calculations on
periodic crystals The internal twists of aromatic rings within the tetragonal units are different for ctn
and bor nets which lead to stable S4 and unstable D2d symmetries respectively This together with
the bulging of triangular units in the bor net reasoned for energetic preference of ctn over bor for all
boron-based 3D COFs79
The connectivity-based atom contribution method (CBAC) developed by Zhong et al80 can
significantly reduce computational time needed for quantum chemical calculation for framework
charges when screening a large number of MOFs or COFs in terms of their adsorption properties The
basic assumption is that atom types with same bonding connectivity in different MOFsCOFs have
identical charges a statement that follows from the concept of reticular chemistry where the
properties of the molecular building blocks keep their properties in the bulk After assigning the
CBAC charges to the atom types slight adjustment (usually less than 3 ) is needed to make the
frameworks neutral Simulated adsorption isotherms of CO2 CO and N2 in MOFs and COFs show that
CBAC charges reproduce well the results of the QM charges8384 The CBAC method still requires a
57
well-parameterized force field in order to account correctly for adsorption isotherms as the second
important contribution to the host-guest interaction is the London dispersion energy between
framework and adsorbed moleculesG
The elastic constants calculated for the 3D COFs COF-102 and COF-108 show that COF-102 is nearly
five times stronger than COF-10878 While the bulk modulus (B) of COF-102 (B = 216 GPa) exceeds
that of known MOFs such as MOF-5 MOF-177 and MOF-205 (B = 153 101 107 GPa respectively81)
the least dense COF-108 is at the edge of structural collapse (B = 49 GPa) The calculations were
made using Density Functional Tight-Binding (DFTB) method Another report82 based on the same
level of theory possibly with a different parameter set however reveals lower bulk moduli for both
COFs 177 GPa for COF-102 and 005 GPa for COF-108 The bulk moduli for COF-103 and COF-105 are
110 and 33 GPa respectively Recently we have reported the structural properties of 3D COFs The
calculations were made using self-consistent charge (SCC) DFTB method The bulk modulus of each
COF is as follows COF-102 ndash 206 COF-103 ndash 139 COF-105 ndash 79 COF-108 ndash 37 COF-202 ndash 153 and
COF-300 ndash 144 GPa Force field calculations79 provide slightly different values COF-102 ndash 170 COF-
103 ndash 118 COF-105 ndash 57 COF-108 ndash 29 COF-202 ndash 110 GPa Albeit the differences in values each
type of calculation shows the trend that bulk modulus decreases with decreasing mas density and
increasing pore volume and distance between connection nodes One has to note that the high
mechanical stabilities of 3D COFs as suggested by the calculations correspond always to defect-free
crystals Theory is expected therefore to overestimate experimental mechanical stability for real
materials
COFs in general have semiconductor-like band gaps (~ 17 to 40 eV)32 Layer-offset from eclipsed
layering slightly increases the band gap However the Density-Of-States (DOS) of a monolayer is
similar to that of bulk COF The band gap is generally smaller for a COF with larger number conjugate
rings
The thermal influence on the crystal structure of 3D COFs is also intriguing as negative thermal
expansion (NTE) the contraction of crystal volume on heating has been reported for COF-10283 The
studies were performed using molecular dynamics with the force field parameters by Schmid et al84
However the thermal expansion coefficient α = -151 times 10-6 K-1 is much smaller compared to that of
some of the reported MOFs that also show NTE behavior85 The volume-contraction upon the
increase of temperature arises from the tilt of phenyl linkers between B3O3 rings and sp3 carbon
atom in TBPM (figure ) Later Schmid et alreported that COF-102 103 105 108 exhibit NTE
behavior both in ctn and bor topologies whereas COF-202 only in ctn topology79 A typical
application is the realization of controllable thermal expansion composites made of both negative
and positive thermal expansion materials
58
4 Applications
41 Gas storage
The success in the synthesis of COFs was certainly the result of a long-term struggle for complete
covalent crystallization The discovery of COFs coincided with the time when world-wide effort was
paid to develop new materials for gas storage in particular for the development hydrogen tanks for
mobile applications Materials made exclusively from light-weight atoms and with large surface
areas were obviously excellent candidates for these applications The gas storage capacity of porous
materials relies on the success of assembling gas molecules in minimum space This is achieved by
the interaction energy exerted by storage materials on the gas molecules Because the interactions
are noncovalent no significant activation is required for gas uptake and release and hence the so-
called physical adsorption or physisorption is reversible The other type of adsorption ndash chemical
adsorption or chemisorption ndash binds dissociated hydrogen covalently or interstitially at the risk of
losing reversibility As it requires the chemical modification of the host material chemisorption is not
a viable route for COFs and might become possible only through the introduction of reactive
components into the lattice The total amount of gas adsorbed in the pores gives rise to what is
referred to as lsquototal adsorption capacityrsquo This quantity is hard to be assessed experimentally as a
measurement is always subjected to influence of the materials surface and gas present in all parts of
the experimental setup Therefore the lsquoexcess adsorption capacityrsquo is reported in experiment Here
the gas stored in the free accessible volume is subtracted from the total adsorption In experiment
this volume includes the voids in the material as well as any empty space between the sample
crystals This volume is typically determined by measuring the uptake of inert gas molecules (He for
H2 adsorption N2 or Ar for adsorption studies of larger molecules) at room temperature with the
assumption that the host-guest interaction between the material and He can be neglected The
different definitions of adsorption is given in Figure 6
Typically experiments measure excess values and simulations provide total values Quantities of
adsorbed molecules are usually expressed as gravimetric and volumetric units The former gives the
amount per unit mass of the adsorbent while the latter gives the amount per unit volume of the
adsorbent Explicative definitions and terminologies related to gas adsorption can be found
elsewhere86
59
Figure 6 Hydrogen density profile in a pore A denotes the excess A+B the absolute and A+B+C the total adsorption The skeletal volume corresponds to the volume occupied by the framework atomsCourtesy of Dr Barbara Streppel and Dr Michael Hirscher MPI for intelligent systems Stuttgart Germany
411 Porosity of COFs
It is common to inspect the porosity of synthesized nanoporous materials using some inert or simple
gas adsorption measurements Distribution of pore size can be sketched from the
adsorptiondesorption isotherm N2 and Ar isotherms provide an approximation of the pore surface
area pore volume and pore size over the complete micro and mesopore size range Usually the
surface area is calculated using the Brunauer-Emmet-Teller (BET) equation or Langmuir equation
Total pore volume (volume of the pores in a pre-determined range of pore size) can be determined
from either the adsorption or desorption phase The Dubinin-Radushkevic method the T-plot
method or the Barrett-Joyner-Halenda (BJH) method may also be employed for calculating the pore
volume The pore size is often corroborated by fitting non-local density functional theory (NLDFT)
based cylindrical model to the isotherm90-93 In some cases the desorption isotherm is affected by
the pore network smaller pores with narrower channels remain filled when the pressure is lowered
This leads to hysteresis on the adsorption-desorption isotherms The different methods employed for
pore structure analysis are characteristic for micropore filling monolayer and multilayer formations
capillary condensation and capillary filling
For any adsorbate in order to form a layer on pore surface the temperature of the surface must
yield an energy (kBT) much less than adsorbate-surface interaction energy Additionally the absolute
value of the adsorbate-surface binding energy must be greater than the absolute value of the
adsorbate-adsorbate binding energy This avoids clustering of the adsorbate into its three-
dimensional phase
60
At high pressure the adsorption isotherm shows saturation which means that no more voids are left
for further occupation The isotherms show different behaviors characteristic of the pore structure of
the materials There are known classifications based on these differences type I II III IV and V For
COFs and the related materials discussed in this review type I II and IV have been observed (see
Figure 7) As more adsorbate is attracted to the surface after the first surface layer completion one
can expect a bend in the isotherm Type I implies monolayer formation which is typical of
microporosity If the surface sites have significantly different binding energies with the adsorbate
type II behavior is resulted In type IV each ldquokneerdquo where the isotherm bends over and the vapor
pressure corresponds to a different adsorption mechanisms (multiple layers different sites or pores)
and represents the formation of a new layer
Figure 7 Different types of adsorptions in Covalently-bound Organic Frameworks
Argon and nitrogen adsorption measurements at respectively 87 and 77 K exhibit type I isotherms
for COF-1 6 102 and 103 and type IV isotherms for COF-5 8 10 102 and 10387 Smaller pore
diameters of COF-1 and 6 (~ 9 Aring each) are characteristic of monolayer gas formation on the internal
pore surface The same reasons are responsible for the type I character of COF-102 and COF-103
(pore size = 12 Aring each) isotherms albeit with some unusual steps at low pressure The type IV
isotherms of COF-5 8 10 102 and 103 show formation of monolayers followed by their
multiplication within their relatively larger pores (sizes are approximately 27 16 32 12 and 12 Aring
respectively) Other COFs that exhibit type I isotherms for N2 adsorption are the 3D COFs COF-202 (11
Aring) COF-300 (72 Aring) and COF-102-C12 (12 Aring) the alkylated 2D COF series COF-18Aring COF-16Aring COF-14Aring
COF-11Aring (the nomenclature includes their pore sizes) and a 2D square COF NiPc COF (19 Aring)
Isotherms like Type-II were observed for PPy-COF (188 Aring) COF-366 (176 Aring) COF-66 (249 Aring) Pc-
PBBA COF (217 Aring) ZnP-COF (25 Aring) COF-LZU1 (12 Aring) PdCOF-LZU1 (12 Aring) and some of the azide-
appended COFs 50AcTrz-COF-5 (156 Aring) 75AcTrz-COF-5 (153 Aring) 100AcTrz-COF-5 (125 Aring)
50BuTrz-COF-5 (232 Aring) 50PhTrz-COF-5 (212 Aring) 50EsTrz-COF-5 (212 Aring) and 25PyTrz-COF-5
(221 Aring) The majority of the COFs with mesopores exhibit type-IV isotherms They are TP-COF (314
Aring) BTP-COF (40 Aring) COF-42 (23 Aring) COF-43 (38 Aring) HHTP-DPB COF (47 Aring) CTC-COF (226 Aring) NiPc-PBBA
COF ZnPc-Py-COF (31 Aring) ZnPc-DPB-COF (31 Aring) ZnPc-NDI-COF (31 Aring) ZnPc-PPE-COF (34 Aring) 5N3-
61
COF (295 Aring) 25N3-COF (276 Aring) 50N3-COF (259 Aring) 75N3-COF (258 Aring) 100N3-COF (227 Aring)
5AcTrz-COF-5 (298 Aring) 25AcTrz-COF-5 (243 Aring) 5PyTrz-COF-5 (289 Aring) and 10PyTrz-COF-5
(242 Aring)
The synthesized microporous amorphous materials CTF-1 (12 Aring) CTF-DCP (14 Aring) PAF-1 and PAF-2
(106 Aring) exhibit type-I isotherms with N2 adsorption PAF-3 (127 Aring) PAF-4(118 Aring) PAF-5 (157 Aring)
PAF-6 (118 Aring) and PAF-11 showed surprisingly not the typical type I behavior in spite of their
microporosity but type-II isotherms
Many of the mesoporous COFs showed small hysteresis in the adsorption-desorption isotherm
pointing the possibility of capillary condensation Hysteresis was observed for the amorphous
materials in both mirco and meso-pore range Various reasons have been attributed for the observed
hysteresis including the softness of the material and guest-host interactions
412 Gas adsorption experiments
Hydrogen uptake capacities of some boron-based COFs were measured by Yaghi et al87 The excess
gravimetric saturation capacities of COF-1 -5 -6 -8 -10 -102 and -103 at 77 K were found to be 148
358 226 350 392 724 and 705 mg g-1 respectively The saturation pressure increases with an
increase in pore size similar to MOFs88 Pore walls of 2D COFs consist only of edges of connectors
and linkers The fact that faces and edges are largely available for interactions with H2 in 3D
geometries is a reason for their enhanced capacity Total adsorption generally increases without
saturation upon pressure because the difference between the total and the excess capacities
corresponds to the situation in bulk (or in a tank) COFs show in particular good gravimetric
capacities because of their low mass density while volumetric capacities typically do not exceed
those of MOFs The gravimetric hydrogen storage capacities of COF-102 and -103 exceeds 10 wt at
a pressure of 100 bar
COF-18 and its alkylated versions COF-16 -14 and -1 have H2 excess gravimetric capacities 155 144
123 and 122 wt respectively at hellipK and hellipbar
Excess uptake of methane in COFs at room temperature and at 70 bar pressure is as follows COF-1
and -6 up to10 wt COF-5 -6 and -8 between 10 and 15 wt and COF-102 and -103 above 20
wt 87 Isosteric heat of adsorption Qst calculated by fitting the isotherms to virial expansion with
the same temperature are relatively weaker (lt 10 kJmol) for COF-102 and COF-103 at low
adsorption which implies weaker adsorbate-adsorbent interactions In comparison COF-1 and 6
exhibit twice the enthalpy however the overall hydrogen storage capacity was very limited due to
62
the smaller pore volume High Qst at zero-loading is not desirable because the primary excess amount
adsorbed at very low pressures cannot be desorbed practically89
COF-102 and 103 outperform 2D COFs for CO2 storage87 The saturation uptakes at room
temperature were 1200 and 1190 mg g-1 for COF-102 and 103 respectively
A type IV isotherm has been observed for ammonia adsorption in COF-10 inherent to its mesoporous
nature90 The material showed an uptake of 15 mol kg-1 at room temperature and 1 bar the highest
of any nanoporous materials The repeated adsorption-desorption cycles were reported to disrupt
the layer arrangement of this 2D COF Yaghi et al90 were also able to make a tablet of COF-10 crystal
which performed nearly up to the crystalline powder
Not many COFs have been experimentally studied for gas storage applications in spite of high
expectations This has to be understood together as a result of the powder-like polycrystallization of
COFs The enthalpy Qst at low-loading accounted to only 46 kJmol
The excess and total hydrogen uptake capacity of PAF-1 at 48 bar and 77 K reached 7 wt and 10
wt respectively65 PAF-3 and PAF-478_ENREF_73 follow the same diamond topology as PAF-1 the
difference accounts in the central atoms of the tetragonal blocks PAF-1 3 and 4 have C Si and Ge
atoms at the centers respectively At 1 bar the respective adsorption capacities were 166 207 and
150 wt The highest value being found for PAF-3 is attributed to its relatively high enthalpy (66 kJ
mol-1) compared to PAF-1 (54 kJ mol-1) and PAF- 4 (66 kJ mol-1) At high pressure the surface area is
a key factor and because of the lower BET surface area PAF-3 and PAF-4 performs poor At 60 bar
their respective adsorption capacities were 55 and 42 wt For PAF-11 hydrogen uptake was 103
wt at 1 bar68
Methane uptakes of PAF-1 3 and 4 at 1 bar are 13 19 and 13 wt respectively76 Their enthalpies
are respectively 14 15 and 232 kJ mol-1 This suggests that Ge moiety have strong interactions with
methane
CO2 adsorption in PAF-1 at 40 bar and room temperature was 1300 mg g-1 which was slightly more
than the capacity of COF-102 and 10365 PAF-1 3 and 4 at 1 atm pressure could store 48 80 and 51
wt respectively At 273 K and 1 atm they stored 91 153 and 107 wt respectively The storage
capacities at 273 K and 298 K lead to the following zero-loading isosteric enthalpies 156 192 162
kJ mol-1 for PAF-1 3 and 4 respectively The higher uptake of PAF-3 may also be explained by its
relatively higher surface area with CO2 molecules
The selective adsorption of greenhouse gases in PAFs was discussed by Ben et al76 At 273 K and 1
atm pressure PAF-1 selectively adsorbes CO2 and CH4 respectively 38 and 15 times more in
63
amounts than N2 PAF-3 showed extraordinary selectivity of 871 for CO2 over N2 and 301 for CH4
over N2 For PAF-4 the selectivity was 441 for CO2 over N2 and 151 for CH4 over N2 Generally the
uptake of N2 Ar H2 O2 are significantly lower compared to CO2 and NH4 which makes these PAFs
suitable for separating them
Harmful gases like benzene and toluene can be stored in PAF-1 in amounts 1306 mg g-1 (1674 mmol
g-1) and 1357 mg g-1 (1473 mmol g-1) respectively at saturated pressures and room temperature65
In PAF-11 their adsorptions were in amounts of 874 mg g-1 and 780 mg g-1 respectively68 PAF-2 was
accounted to a much lower benzene adsorption (138 mg g-1) at the same conditions59 Adsorption of
cyclohexane vapor in PAF-2 was only 7 mg g-1 While PAF-11 exhibited hydrophobicity it could
accommodate significant amount of methanol (654 mg g-1 or 204 mmol g-1) at room temperature
and saturated pressure68 PAF-5 has been tested for storing organic pollutants69 At room
temperature and saturated pressure PAF-5 adsorbed methanol benzene and toluene in amounts
6531 cm3 g-1 (292 mmol g-1) 26296 cm3 g-1 (117 mmol g-1) and 2582 cm3 g-1 (115 mmol g-1)
respectively Their adsorptions in PAF-5 were deviated from the typical type-I behavior Methanol
exhibited a large slope at the high pressure region corresponding to the relative size of it Zhu G et
al dedicated PAF-6 for biomedical applications With no toxicity observed for PAF-6 over a range of
concentrations adsorption of ibuprofen was measured in it PAF-6 showed an uptake of 350 mg g-1
The release of ibuprofen was also tested in simulated body fluid which showed a release rate of 50
in 5 hours 75 in 10 hours and 100 in almost 46 hours
413 Theoretical predictions
Grand canonical Monte Carlo (GCMC) is a widely used method for the simulation of gas uptakes in
nanoporous materials In GCMC the number of adsorbates and their motions are allowed to change
at constant volume temperature and chemical potential to equilibrate the adsorbate phase The
motions are random guided by Monte Carlo methods and the energies are calculated by force field
methods The details of it may be found in the literature91
Goddart III WA et al92 simulated the uptake capacities of boron-based COFs using force fields derived
from the interactions of H2 and COFs calculated at MP2 level of theory The saturated excess uptakes
of both COF-105 (at 80 bar) and 108 (at 100 bar) were accounted to 10 wt at 77 K which are more
than the uptakes measured in ultrahigh porous MOF-200 205 and 21093 The predictions for other
COFs include 38 wt for COF-1 (AB stacking) 34 wt for COF-5 (AA stacking) 88 wt for COF-102
and 91 wt for COF-103 At 01 bar COF-1 in AB stacking showed highest excess uptake (17 wt )
compared to other COFs in the present discussion Total uptake capacities of the COFs are in the
following order COF-108 (189 wt ) COF-105 (183 wt ) COF-103 (113 wt ) COF-102 (106
64
wt ) COF-5 (55 wt ) and COF-1 (38 wt ) It is worth noticing the higher gravimetric capacities of
COF-108 and 105 which were not measured experimentally They benefit from their lower mass and
higher pore volume In case of volumetric capacity COF-102 (404 gL excess) surpasses the others at
high pressures The total volumetric capacities of the COFs are 499 498 399 395 361 and 338
gL for COF-102 103 108 105 1 and 5 respectively Their additional calculations predicted benzene
rings as favorite locations for H2 molecules
Similar values and trends have been reported by Garberoglio G94 who also reasoned poor solid-fluid
interactions in a large part of free volume for the low volumetric capacity of COF-108 and 105 A
room temperature excess gravimetric uptake of approximately 08 wt was obtained for COF-108
and 105 Quantum diffraction effects were added to the interaction energies of H2 with itself and the
material which were calculated using universal force-field (UFF) With possible overestimation
Klontzas et al30 and Bassem et al95 have reported total gravimetric uptake of 45 and 417 wt
respectively at 300 K for the same COFs Among the boron-based 3D COFs COF-202 exhibited much
smaller uptake capacity at high pressures due to its smaller pore volume95 Lan et al96 predicted a
very small H2 uptake of 152 wt COF-202 at 300K and 100 bar Studies of Yang et al97 clarifies that
pore filling with H2 in 2D COFs is a gradual process rather than being in steps of layer formation
Garberoglio and Vallauri98 pointed out that the relatively higher mass density and lower surface area
per unit volume of 2D COFs are the reasons for their lower uptakes in comparison with 3D COFs The
surface area of hexagonal 2D COFs (AA stacking) averaged around 1000 m2 cm-3 while that of 3D
COFs were about 1500 m2 cm-3
Simulations of H2 adsorption in the perfect crystalline form of PAF-1 and PAF-11 also known as PAF-
302 and PAF-304 respectively are carried out by Zhu et al99 At 77 K H2 excess uptake of PAF-302 (7
wt ) is slightly higher than COF-102 (684 wt )87 and slightly lower than MOF-177 (740 wt )100 At
room temperature PAF-302 exhibited a poor uptake (221 wt at 100 bar) PAF-304 showed
excellent total uptakes 2238 wt at 77 K and 100 bar 653 wt at 298 K and 100 bar These are
highest among all nanoporous materials
Babarao and Jiang studied room-temperature CO2 adsorption in COFs101 At low pressures COFs with
smaller pore volume exhibit a steep rise in their isotherm while at high pressures COF-105 and 108
(82 and 96 mmol g-1 respectively at 30 bar) dominate the others COF-6 has the highest isosteric heat
of adsoption (328 kJmol) hence it made the first steep rise in the volumetric isotherm followed by
COF-8 102 103 10 105 and 108 Correlations of gravimetric and volumetric capacities with mass
density pore volume porosity and surface area have been excellently manifested in this article101
With increasing framework-density gravimetric uptake falls inversely while volumetric capacity
decreases linearly The former rises with free volume while the latter rises and then drops slightly
65
Both of them rise with porosity and surface area Choi et al102 predicted 45 times more CO2 uptake in
COF-108 (3D) compared to COF-5 (2D) Recently Schmid et al79 also have predicted CO2 adsorption
especially for 3D COFs The uptake of COF-202 was almost half of COF-102 and 103 at room
temperature Type IV isotherm has been found for CO2 adsorption in COF-8 and 10 at low
temperatures (lt 200 K)97 Yang and Zhong97 excluded capillary condensation and dissimilar
adsorption sites but multilayer formation as reasons of the stepped behavior (Yang and Zhong
explained this as a consequence of multilayer formation rather than a result of capillary
condensation or dissimilar adsorption sites)
Methane adsorption in 2D and 3D COFs was first calculated by Garberoglio et al Among COF-6 8 and
10 the former which has smaller pore size and higher binding energy with CH4 shows better
volumetric adsorption (~ 120 cm3cm3 at 20 bar) than the others at room temperature at low
pressure region (lt 25 bar) Larger pore size favors COF-10 for higher volumetric adsorption (~ 160
cm3cm3 at 70 bar) at high pressures COF-8 with intermediate pore size shows largest excess amount
of methane adsorbed upon saturation however still lower than two 3D COFs COF-102 and 103
show excess volumetric adsorption of about 180 cm3cm3 at 35 bar This is significantly higher than
the uptake of two other 3D COFs COF-105 and 108 Entropic effects which primarily arise from the
change of dimensionality when the bulk phase of adsorbate transforms to adsorbed phase are
crucial in adsorption Garberoglio94 shows that solid-fluid interaction potential in large part of volume
of COF-105 and 108 is not strong enough to overcome the entropic loss On the other hand these
two COFs exhibit relatively higher gravimetric uptake (720 and 670 cm3g respectively) Goddard III et
al89 predicted total methane uptakes in the following amounts 415 wt in COF108 405 wt in
COF-105 31 wt in COF-103 284 wt in COF-102 196 wt in COF-10(AA) 169 wt in COF-
5(AA) 159 wt in COF-8(AA) 123 wt in COF-6(AA) and 109 wt in COF-1(AB) Yang and Zhong97
have shown that even at low temperatures (100 K) the pore filling of COF-8 with CH4 is rather
gradual than by step-by-step whereas for COF-10 multilayer formation provides a stepped behavior
in the isotherm Alternately Goddard et al pointed out that layer formation coexists with pore-filling
at room temperature89
414 Adsorption sites
First principle calculations on cluster models are typically employed to find favorite adsorption sites
and binding energies of adsorbates within frameworks Benzene rings are found to be a usual
location for H2 where it prefers to orient in perpendicular fashion3086110 Other favorite locations
include B3O3 and C2O2B rings110111 where H2 orients horizontally on the face as well as on the
edge3086 The central ring of HHTP does not provide enough binding to H2 atoms because of small
amount of charges There are some differences in the adsorption energies and favorite sites
66
calculated at different levels of theory Overall the reported binding energies between H2 and any
COF fragment are less than 6 kJ mol-1 In comparison to MOFs COFs do not offer any stronger binding
energy with H2 The advantage of COFs is their low mass density Lan et al103 reported that CO2 is
more preferred to be adsorbed on B3O3 or C2O2B rings than benzene rings Choi et al102 reported that
the oxygen sites are the primary locations for CO2 and CO2 bound COFs are the second strong binding
sites
415 Hydrogen storage by spillover
Hydrogen spillover is an alternate way to store hydrogen113114 It involves dissociation of hydrogen
gas by supported metal catalysts subsequent migration of atomic hydrogen through the support
material and final adsorption in a sorbent Here surface-diffusion of H atoms over the support is
known as primary spillover Secondary spillover is the further diffusion to the sorbent Bridging the
metal part with the sorbent is a practice to enhance the uptake It increases the contact between the
source and receptor and reduces the energy barriers especially in the secondary spillover As the
final sorption is chemisorption surface area of the sorbent is more important than pore volume
Yang and Li measured H2 storage in COF-1 by spillover They used Pt supported on active carbon
(PtAC) as source for hydrogen dissociation Active carbon was the primary receptor and COF-1 the
secondary receptor COF-1 showed much low uptake 128 wt at 77 K and 1 atm 026 wt at 298
K and 100 bar In comparison to MOFs these are very low104 However they have found that the
uptake could be scaled up by a factor of 26 by bridging COF-1 with PtAC by applying carbonization
They also report that heat of adsorption between H and surface sites is more important than surface
area and pore volume in enhancing the net adsorption by spillover
Ganz et al theoretically predicted that the uptake capacities in COFs by hydrogen spillover can be
higher than the measured value116117 Based on ab initio quantum chemistry calculations and
counting the active storage sites they predicted 45 wt for COF-1 in AB stacking and 55 wt for
COF-5 in AA stacking at room temperature and 100 bar
42 Diffusion and Selectivity
Computed self-diffusion coefficients for H2 and CH4 adsorbates in 2D COFs (in AA stacking) are up to
one order of magnitude higher than that in 3D MOFs such as MOF-598 In comparison to nanotubes
the diffusion of H2 molecules in 2D COFs is one order of magnitude slower The differences in
diffusion coefficients are attributed to different pore structures Interactions of the corners of the
hexagonal pore with the fluid present potential barriers along the cylindrical pore axis Diffusion
occurs with jumps after long waiting The height of the barrier is still smaller than in MOFs
67
Homogeneous pore walls and absence of pore corners in nanotubes contribute much less
corrugation to the solid-fluid potential energy surface Increase of self-diffusivities of H2 and CH4 with
increase in pore size of 2D COFs is also shown by Keskin[Ref] Due to the smaller size of H2 its
diffusivity is higher than that of CH4 in any material For reasons of their relative size In a mixture of
the two the self-diffusivity of CH4 increases while that of H2 decreases
Molecular dynamics simulation studies of the diffusion of gas molecules within a 2D COF performed
by Krishna et al105 revealed its relatively lower binding energy of hydrogen molecules within the pore
walls compared to argon CO2 benzene ethane propane n-butane n-pentane and n-hexane
Binding energy prevents the molecules from diffusing through the pore channels They tested if
Knudsen diffusion is applicable to COFs Knudsen diffusion occurs when the molecules frequently
collide with the pore wall This generally happens when the mean free path is larger than the pore
diameter The simulations had been carried out in BTP-COF which accounts a pore diameter of 4 nm
It is found out that the ratio of zero-loading diffusivity with the Knudsen diffusivity has a significant
correlation with isosteric heat of adsorption the stronger the binding energy of the molecules with
the walls the lower the ratio Hydrogen being an exception among the investigated molecules
exhibited the ratio close to unity while others with their isosteric heat of adsorption higher than 10
kJmol-1 have a value close to 01 For a mixture of two gases with significant differences in binding
energies the ratio of self-diffusivities affirms high diffusion selectivity
Selectivity of CH4 over H2 in COF-5 6 and 10 is studied by Keskin106 Narrow pores support the
selective adsorption of CH4 over H2 however they suffer from entropic effects at high pressures
which favor H2 adsorption Liu et al107 have reported that the adsorption selectivity of COFs and
MOFs are in general similar up to 20 bar Delta loading or working capacity (usually expressed in
molkg) is an important term often used to show the economics of the selective adsorption This is
defined as the difference in loadings of the preferred gas at adsorption and desorption pressures
Adsorption is usually in the range 1 to 100 bar while desorption in 01 to 1 bar High selectivity and
high working capacity are preferential for practical use COF-6 has higher selectivity among the three
studied COFs but lower working capacity106 Best selectivity-working capacity combination is shown
by COF-5106 Nonetheless it is not better than CuBTC or CPO-27-X (X = Zn Mg)121122 Liu et al107
attributed the high isosteric heat of adsorption of COF-6 and Cu-BTC for their high adsorption
selectivity They also pointed out that the electrostatic contribution of framework charges in COFs
are smaller than in MOFs however needs to be taken into account
While CH4 is strongly adsorbed H2 undergoes faster diffusion Hence the product of adsorption
selectivity (gt1) and diffusion selectivity (gt0 lt1) which is membrane selectivity is smaller than
adsorption selectivity Keskin106 has compared the selectivity of COF-membranes with known
68
membranes The selectivity of COF-10 is similar to MOF-177 and IRMOFs COF-5 and 6 outperform
them It is to be noted that COF-5 is CH4 selective while COF-10 is H2 selective although their
topologies are similar CH4 permeability of COF membranes is much higher than several MOFs and
ZIFs COF-5 and 6 exhibited high membrane selectivity together with high membrane permeability
Some discussions on the adsorption and membrane based selectivity of COF-102 in comparison with
IRMOFs and Cu-BTC for CH4CO2H2 mixtures can be found in ref108 Adsorption selectivity of COF-6
and Cu-BTC in CH4H2 and CH4CO2 mixtures surpass other COFs and MOFs107 sdfalf
43 Suggestions for improvement
The level of achievement made by COFs and related materials yet do not practically meet the
practical requirements of several applications Hence thoughts for improvement primarily focused
on overcoming their limitations and enhancing characteristic parameters Some theoretical
suggestions for improved performances are mainly discussed here
431 Geometric modifications
Functionalities may be improved by utilizing the structural divergence of framework materials
Several examples may be found in the literature for MOFs etc Klontzas et al suggested replacement
of phenyl linkers in COF-102 by multiple aromatic moieties while keeping the topology unchanged to
increase surface area and pore volume109 They used biphenyl triphenyl naphthalene and pyrene
linkers in the new COFs All of them showed enhanced gravimetric capacity compared to the parent
COF One of the modified COF-102 with triphenyl linker showed 267 and 65 wt at 77 K and 300 K
respectively at 100 bar This kind of replacement may affect the adsorption of each adsorbate
differently leading to the alteration of the selective adsorption of one component over the other110
Following the reticular construction scheme we modeled some new 2D COFs by cross-linking some
of the familiar and unfamiliar building blocks in the COF literature32 This showed the structural
divergence of COFs however they exhibited structural and electronic properties analogues to other
2D COFs
Similar modifications in 2D and 3D COFs for high CO2 uptake were made by Choi et al102 DPABA
(diphenylacetylene-44ʹ-diboronic acid) and its tetrahedral version TBPEPM (tetra[4-(4-
dihydroxyborylphenyl)ethynyl]phenyl) in place of BDBA in COF-5 and TBPMTBPS in COF-108COF-
105 respectively were used for new COF designs The new 2D COF promised a saturated CO2 uptake
higher than in COF-102 because of its large pore volume The new 3D COFs were worth for an uptake
twice more than in COF-105 and 108
69
Goddard et al also designed about 14 new COFs by substituting the aromatic rings in the tetragonal
part (TBPM or TBPS) of COF-102 COF-103 COF-105 COF-108 and COF-202 with vinyl groups or alkyl-
functionalized extended or fused aromatic rings111 The new designs adopted their parent topology
and their structural stability was confirmed by analyzing molecular dynamics (MD) simulations at
room temperature Fused aromatic rings and vinyl groups provided relatively lower and the highest
zero-loading isosteric heat of adsorption (Qst) respectively however the room-temperature delivery
amounts of methane in the respective COFs crossed the DOE target at 35 bar111 In the latter
methane-methane interaction compensated Qst on high-loading
Lino M A et al112 theoretically inspected the ability of layered COFs to exhibit structural variants of
layered materials The hexagonal 2D COF layers may be rolled into the form of nanotubes (NT) or
may exist in the form of fullerenes with the inclusion of pentagons One could think of a formula unit
which resembles the carbon atoms in carbon nanotubes (CNT) or fullerenes The NTs in general can
have any chirality although the study included only armchair and zigzag NTs Density Functional
Theory based calculations reveal that COF-1 NTs of diameter greater than 15 Aring is energetically
favored This was concluded from the comparison of strain-energy per formula unit of the COF-1 NTs
with that of small diameter CNTs The strain energies of zigzag and armchair nanotubes show similar
quadratic functions of tube-diameter The Youngrsquos modulus of COF-1 (22) NT is found out to be 120
GPa This is much smaller compared to typical CNTs which have Youngrsquos modulus average around
1100 GPa113 Band gaps of COF layers remain unchanged when they roll up into nanotubes A COF-1-
fullerene built by scaling C60 molecule has large diameter and hence much less strain energy
compared to C60 Babarao and Jiang reported that the predicted CO2 adsorption capacity of COF-1 NT
is similar to that of CNTs101
Balance between mass density and surface area and hence high gravimetric and volumetric
capacities can be achieved by proper utilization of pore volume or proper distribution of pores Choi
et al theoretically predicted high gravimetric and volumetric capacities for H2 in a pillared COF114 A
pyridine molecule vertically placed above a boron atom in the boroxine ring of COF-1 layer is found
energetically favorable however with wrinkling of the layer115 Here nitrogen atom in pyridine forms
a covalent bond with the boron atom This pillaring increases the separation between the layers
exposes the buried faces of boroxine and aromatic rings to pores and hence increases surface area
and free volume Accessible surface area and free volume have been tripled and gravimetric and
volumetric capacities were calculated to be 10 wt and 60 g L-1 respectively at 77 K and 100 bar114
This volumetric capacity is higher than in NU-100116 and MOF-21093 which show ultrahigh surface
area
70
432 Metal doping
Miao Wu et al studied doping of COF-10 with Li and Ca atoms for hydrogen storage117 The metal
dopants transferred charges to substrate which in turn provided large polarization to hydrogen
molecules Similar observation has been made by Lan et al96 for Li atoms in COF-202 However they
showed the tendency to aggregate at high concentration Lan et al extensively studied doping of
positively charged alkali (Li Na and K) alkaline-earth (Be Mg Ca) and transition (Sc and Ti) metals in
COF-102 and 105 for enhanced CO2 capture103 They found that benzene rings and the three outer
rings of HHTP are the favorite sites for all the metal dopants Other interested locations are edges of
benzene rings and oxygen atoms B3O3 C2O2B rings and the central ring of HHTP were the unwanted
areas Lithium showed stability on the favorite locations while sodium and potassium tended to
cluster even at low concentrations Alkaline-earth metals only weakly interacted with the COFs
whereas transition metals chemisorbed in them which would possibly damage the substrate Lithium
is found out to be a good dopant for enhanced gas storage
Doping electropositive metals would be of advantage because they provide stronger binding with H2
(~ 4 kcalmol)103125133134 The effect of counterions in COFs is studied by Choi et al118 They found out
that although the binding energy of LiF is smaller than Li ion F counterion can hold one hydrogen
atom Additionally Li+ and Mg2+ ions preferred to be isolated than aggregated A step further
Goddard et al119 have recently shown that in presence of tetrahydrofuran (THF a solvent) an
electron is transferred from Li to the benzene ring whereas in the case of gas phase the electron
remained in the atom Additionally they suggested ways to remove solvents which are weakly
coordinated to the material Zhu et al110 suggested doping Li+ after replacement of hydrogen atom by
oxygen ion in COF-102 Another suggestion was the replacement of hydrogen atom by O--Li+ group
Both the cases exhibited higher CO2 adsorption than in the case of Li doping below 10 bar At 10 bar
the three cases exhibited almost the same uptake capacity Below 1 bar the doped Li+ provided
stronger interaction with quadrupolar CO2 in comparison with the substituted O--Li+ group The
differences at low pressures are attributed to the differences in the magnitude of the charge of Li
The doped Li+ remained to hold nearly +1 charge whereas the inclusion of O--Li+ to the framework
diminished the charge of Li+ to a moderate amount Doping Li atom added only relatively small
amount of charge to Li
Doping with Li+ could increase the H2 binding energies up to 28 kJmol118 With properly distributed
metal ion dopants the H2 uptake capacity in COF-108 is predicted to be 65 wt Cao et al also
predicted 684 and 673 wt of H2 uptake in Li-doped COF-105 and 108 respectively at room
temperature and 100 bar120 In Li-doped COF-202 the predicted uptake was 438 wt for the same
conditions96 Goddard et al119 observed that high performance of Li doped COFsMOFs at low
71
pressures (1 ndash 10 bar) is a disadvantage when calculating delivery uptake capacity Na doping could
overcome this difficulty as its heat of adsorption is lower than in Li doped materials Their predicted
delivery gravimetric capacities from 1 to 100 bar at room temperature for Li and Na doped COF-102
and 103 were higher than the 2010 DOE target of 45 wt at room temperature
Lan et al described that CO2 is polarized in presence of Li cation and the polarization increases when
Li cation is physisorbed in COF103 Their calculated uptake capacities of lithium-doped COF-102 and
COF-105 at low loading is about eight and four times higher than that of nondoped ones respectively
Also a very high uptake of 2266 mgg was predicted in the doped COF-105 at 40 bar Li-doped COF-
102 and COF-103 also showed higher uptakes of methane ndash almost double the amount ndash compared
to nondoped COFs at room temperature and low pressures (lt 50 bar)121 Binding energy between
doped Li cation and CH4 was calculated to be 571 kcalmol
Li-functionalized COF-105 realized by inserting alkoxide groups provided very high volumetric uptake
of H2 Klontzas et al122 replaced three hydrogen atoms in HHTP by oxygen atoms in order to achieve
the functionalization In spite of the increased weight due to the additional oxygen atoms the COF
exhibited gravimetric capacity of 6 wt at 300 K
Incorporation of lithium tetrazolide group into a new PAF having diamond topology and triphenyl
linkers for enhanced hydrogen adsorption is suggested by Sun et al123 The tetrazolide anion (CHN4-)
interacts with the lithium cation via ionic bonds The anion and the cation together can bind upto 14
hydrogen molecules Two lithium tetrazolide groups were introduced in the middle phenyl ring of
each triphenyl linker and GCMC calculations predicted 45 wt for the network at 233 K and 100 bar
With the intention of increasing the H2 binding energy Li et al chemically implanted metals in the
place of light atoms in COF-108124 They replaced the C2O2B rings in COF-108 by C2O2Al C2N2Al and
C2Mg2N and each new COF showed stability within DFT This kind of doping does not allow
aggregation of metal clusters Their GCMC simulations showed that the replacement strategy could
improve the hydrogen storage capacity at room temperature by a factor of up to three124 Zou et al
suggested that para-substitution of two carbon atoms in benzene rings by two boron atoms can
facilitate charge transfer between the rings and metal dopants125 Their work revealed that
dispersion of metals forbid any clustering and the doping could enhance the H2 uptake capacity
significantly
433 Functionalization
Functionalization of porous frameworks with ether groups for selective CO2 capture is suggested by
Babarao et al126 The electropositive C atom of CO2 interacts mainly with the electronegative O or N
72
atom of the functional groups They studied PAF-1 functionalized with ndashOCH3 ndashNH2 and ndashCH2OCH2ndash
groups in its biphenyl linkers The latter one which is the tetrahydrofuran-like ether-functionalized
PAF-1 showed high adsorption capacity for CO2 and high selectivity for CO2CH4 CO2N2 and CO2H2
mixtures at ambient conditions
5 Conclusions
Successful covalent crystallization resulted as a class of materials Covalent Organic Frameworks This
review portrays different synthetic schemes successful realizations and potential applications of
COFs and related materials The growth in this area is relatively slow and thus promotions are
needed in order to achieve its potential
6 List and pictures of chemical compounds
alkyl-THB Alkyl-1245-tetrahydroxybenzene
BDBA 14-benzenediboronic acid
BPDA 44ʹ-biphenyldiboronic acid
BTBA 135-benzene triboronic acid
BTDADA 14-benzothiadiazole diboronic acid
BTPA 135-benzenetris(4-phenylboronic acid)
CA Cyanuric acid
DAB 14-diaminobenzene
DCB 14-dicyanobenzene
DCF 27-diisocyanate fluorine
DCN 26-dicyanonaphthalene
DCP 26-dicyanopyridine
DETH 25-diethoxyterephthalohydrazole
DMBPDC 33ʹ-dimethoxy-44ʹ-biphenylene diisocyanate
DPB Diphenyl butadyenediboronic acid
73
HP 1-hexyne propiolate
HHTP 23671011-hexahydroxytriphenylene
MP Methyl propiolate
N3-BDBA Azide-appended benzenediboronic acid
NDI Naphthalenediimide diboronic acid
NiPcTA Nickel-phthalocyanice tetrakis(acetonide)
OH-Pc-Ni 2391016172324-octahydroxyphthalocyaninato)nickel(II)
OH-Pc-Zn 2391016172324-octahydroxyphthalocyaninato)zinc
PA Piperazine
Pac 2-propenyl acetate
PcTA Phthalocyanine tetra(acetonide)
PdAc Palladium acetate
PDBA Pyrenediboronic acid
PPE Phenylbis(phenylethynyl) diboronic acid
PPP 3-phenyl-1-propyne propiolate
PyMP (3α13α2-dihydropyren-1-yl)methyl propionate
TA Terephthaldehyde
TAM tetra-(4-anilyl)methane
TAPP Tetra(p-amino-phneyl)porphyrin
TBB 135-tris(4-bromophenyl)benzene
TBPM tetra(4-dihydroxyboryl-phenyl)methane
TBPP Tetra(p-boronic acid-phenyl)porphyrin
TBPS tetra(4-dihydroxyboryl-phenyl)silane
TBST tert-butylsilane triol
74
TCM Tetrakis(4-cyanophenyl)methane
TDHB-ZnP Zinc(II) 5101520-tetrakis(4-(dihydroxyboryl)phenyl)porphyrin
TFB 135-triformylbenzene
TFPB 135-tris-(4-formyl-phenyl)-benzene
THAn 2345-Tetrahydroxy anthracene
THB 1245-tetrahydroxybenzene
THDMA Polyol 2367-tetrahydroxy-910-dimethyl-anthracene
TkBPM Tetrakis(4-bromophenyl)methane
TPTA Triphenylene tris(acetonide)
trunc-TBPM-R R-functionalized truncated TBPM
75
Figure 8 Reactants of Covalently-bound Organic Frameworks
76
Figure 9 (Figure 8 continued)
(1) Morris R E Wheatley P S Angewandte Chemie-International Edition 2008 47 4966 (2) Li J-R Kuppler R J Zhou H-C Chemical Society Reviews 2009 38 1477 (3) Corma A Chemical Reviews 1997 97 2373 (4) Seo J S Whang D Lee H Jun S I Oh J Jeon Y J Kim K Nature 2000 404 982 (5) Yaghi O M OKeeffe M Ockwig N W Chae H K Eddaoudi M Kim J Nature 2003 423 705
77
(6) OKeeffe M Peskov M A Ramsden S J Yaghi O M Accounts of Chemical Research 2008 41 1782 (7) Ockwig N W Delgado-Friedrichs O OKeeffe M Yaghi O M Accounts of Chemical Research 2005 38 176 (8) Li H Eddaoudi M OKeeffe M Yaghi O M Nature 1999 402 276 (9) Chen B L Eddaoudi M Hyde S T OKeeffe M Yaghi O M Science 2001 291 1021 (10) Eddaoudi M Moler D B Li H L Chen B L Reineke T M OKeeffe M Yaghi O M Accounts of Chemical Research 2001 34 319 (11) Eddaoudi M Kim J Rosi N Vodak D Wachter J OKeeffe M Yaghi O M Science 2002 295 469 (12) Chae H K Siberio-Perez D Y Kim J Go Y Eddaoudi M Matzger A J OKeeffe M Yaghi O M Nature 2004 427 523 (13) Furukawa H Kim J Ockwig N W OKeeffe M Yaghi O M Journal of the American Chemical Society 2008 130 11650 (14) Smaldone R A Forgan R S Furukawa H Gassensmith J J Slawin A M Z Yaghi O M Stoddart J F Angewandte Chemie-International Edition 2010 49 8630 (15) Eddaoudi M Kim J Wachter J B Chae H K OKeeffe M Yaghi O M Journal of the American Chemical Society 2001 123 4368 (16) Sudik A C Millward A R Ockwig N W Cote A P Kim J Yaghi O M Journal of the American Chemical Society 2005 127 7110 (17) Sudik A C Cote A P Wong-Foy A G OKeeffe M Yaghi O M Angewandte Chemie-International Edition 2006 45 2528 (18) Tranchemontagne D J L Ni Z OKeeffe M Yaghi O M Angewandte Chemie-International Edition 2008 47 5136 (19) Lu Z Knobler C B Furukawa H Wang B Liu G Yaghi O M Journal of the American Chemical Society 2009 131 12532 (20) Park K S Ni Z Cote A P Choi J Y Huang R Uribe-Romo F J Chae H K OKeeffe M Yaghi O M Proceedings of the National Academy of Sciences of the United States of America 2006 103 10186 (21) Hayashi H Cote A P Furukawa H OKeeffe M Yaghi O M Nature Materials 2007 6 501 (22) Banerjee R Furukawa H Britt D Knobler C OKeeffe M Yaghi O M Journal of the American Chemical Society 2009 131 3875 (23) Cote A P Benin A I Ockwig N W OKeeffe M Matzger A J Yaghi O M Science 2005 310 1166 (24) El-Kaderi H M Hunt J R Mendoza-Cortes J L Cote A P Taylor R E OKeeffe M Yaghi O M Science 2007 316 268 (25) Wan S Guo J Kim J Ihee H Jiang D Angewandte Chemie-International Edition 2008 47 8826 (26) Wan S Guo J Kim J Ihee H Jiang D L Angewandte Chemie-International Edition 2009 48 5439 (27) Cote A P El-Kaderi H M Furukawa H Hunt J R Yaghi O M Journal of the American Chemical Society 2007 129 12914 (28) Hunt J R Doonan C J LeVangie J D Cote A P Yaghi O M Journal of the American Chemical Society 2008 130 11872 (29) Uribe-Romo F J Hunt J R Furukawa H Klock C OKeeffe M Yaghi O M Journal of the American Chemical Society 2009 131 4570 (30) Klontzas E Tylianakis E Froudakis G E Journal of Physical Chemistry C 2008 112 9095 (31) Tylianakis E Klontzas E Froudakis G E Nanotechnology 2009 20 (32) Lukose B Kuc A Frenzel J Heine T Beilstein Journal of Nanotechnology 2010 1 60
78
(33) Tilford R W Gemmill W R zur Loye H C Lavigne J J Chemistry of Materials 2006 18 5296 (34) Spitler E L Dichtel W R Nature Chemistry 2010 2 672 (35) Spitler E L Giovino M R White S L Dichtel W R Chemical Science 2011 2 1588 (36) Campbell N L Clowes R Ritchie L K Cooper A I Chemistry of Materials 2009 21 204 (37) Ding X Guo J Feng X Honsho Y Guo J Seki S Maitarad P Saeki A Nagase S Jiang D Angewandte Chemie-International Edition 2011 50 1289 (38) Feng X A Chen L Dong Y P Jiang D L Chemical Communications 2011 47 1979 (39) Zwaneveld N A A Pawlak R Abel M Catalin D Gigmes D Bertin D Porte L Journal of the American Chemical Society 2008 130 6678 (40) Gutzler R Walch H Eder G Kloft S Heckl W M Lackinger M Chemical Communications 2009 4456 (41) Blunt M O Russell J C Champness N R Beton P H Chemical Communications 2010 46 7157 (42) Sassi M Oison V Debierre J-M Humbel S Chemphyschem 2009 10 2480 (43) Ourdjini O Pawlak R Abel M Clair S Chen L Bergeon N Sassi M Oison V Debierre J-M Coratger R Porte L Physical Review B 2011 84 (44) Colson J W Woll A R Mukherjee A Levendorf M P Spitler E L Shields V B Spencer M G Park J Dichtel W R Science 2011 332 228 (45) Berlanga I Ruiz-Gonzalez M L Gonzalez-Calbet J M Fierro J L G Mas-Balleste R Zamora F Small 2011 7 1207 (46) Wan S Gandara F Asano A Furukawa H Saeki A Dey S K Liao L Ambrogio M W Botros Y Y Duan X Seki S Stoddart J F Yaghi O M Chemistry of Materials 2011 23 4094 (47) Ding X Chen L Honsho Y Feng X Saenpawang O Guo J Saeki A Seki S Irle S Nagase S Parasuk V Jiang D Journal of the American Chemical Society 2011 133 14510 (48) Tilford R W Mugavero S J Pellechia P J Lavigne J J Advanced Materials 2008 20 2741 (49) Lanni L M Tilford R W Bharathy M Lavigne J J Journal of the American Chemical Society 2011 133 13975 (50) Li Y Yang R T Aiche Journal 2008 54 269 (51) Nagai A Guo Z Feng X Jin S Chen X Ding X Jiang D Nature Communications 2011 2 (52) Bunck D N Dichtel W R Angewandte Chemie-International Edition 2012 51 1885 (53) Ding S-Y Gao J Wang Q Zhang Y Song W-G Su C-Y Wang W Journal of the American Chemical Society 2011 133 19816 (54) Miyaura N Suzuki A Chemical Reviews 1995 95 2457 (55) Kalidindi S B Yusenko K Fischer R A Chemical Communications 2011 47 8506 (56) Kuhn P Antonietti M Thomas A Angewandte Chemie-International Edition 2008 47 3450 (57) Bojdys M J Jeromenok J Thomas A Antonietti M Advanced Materials 2010 22 2202 (58) Kuhn P Forget A Su D Thomas A Antonietti M Journal of the American Chemical Society 2008 130 13333 (59) Ren H Ben T Wang E Jing X Xue M Liu B Cui Y Qiu S Zhu G Chemical Communications 2010 46 291 (60) Zhang W Li C Yuan Y-P Qiu L-G Xie A-J Shen Y-H Zhu J-F Journal of Materials Chemistry 2010 20 6413 (61) Trewin A Cooper A I Angewandte Chemie-International Edition 2010 49 1533 (62) Mastalerz M Angewandte Chemie-International Edition 2008 47 445
79
(63) Chan-Thaw C E Villa A Katekomol P Su D Thomas A Prati L Nano Letters 2010 10 537 (64) Palkovits R Antonietti M Kuhn P Thomas A Schueth F Angewandte Chemie-International Edition 2009 48 6909 (65) Ben T Ren H Ma S Q Cao D P Lan J H Jing X F Wang W C Xu J Deng F Simmons J M Qiu S L Zhu G S Angewandte Chemie-International Edition 2009 48 9457 (66) Yamamoto T Bulletin of the Chemical Society of Japan 1999 72 621 (67) Zhou G Baumgarten M Muellen K Journal of the American Chemical Society 2007 129 12211 (68) Yuan Y Sun F Ren H Jing X Wang W Ma H Zhao H Zhu G Journal of Materials Chemistry 2011 21 13498 (69) Ren H Ben T Sun F Guo M Jing X Ma H Cai K Qiu S Zhu G Journal of Materials Chemistry 2011 21 10348 (70) Zhao H Jin Z Su H Jing X Sun F Zhu G Chemical Communications 2011 47 6389 (71) Mortera R Fiorilli S Garrone E Verne E Onida B Chemical Engineering Journal 2010 156 184 (72) Dogru M Sonnauer A Gavryushin A Knochel P Bein T Chemical Communications 2011 47 1707 (73) Spitler E L Koo B T Novotney J L Colson J W Uribe-Romo F J Gutierrez G D Clancy P Dichtel W R Journal of the American Chemical Society 2011 133 19416 (74) Zhang Y Tan M Li H Zheng Y Gao S Zhang H Ying J Y Chemical Communications 2011 47 7365 (75) Uribe-Romo F J Doonan C J Furukawa H Oisaki K Yaghi O M Journal of the American Chemical Society 2011 133 11478 (76) Ben T Pei C Zhang D Xu J Deng F Jing X Qiu S Energy amp Environmental Science 2011 4 3991 (77) Lukose B Kuc A Heine T Chemistry-a European Journal 2011 17 2388 (78) Zhou W Wu H Yildirim T Chemical Physics Letters 2010 499 103 (79) Amirjalayer S Snurr R Q Schmid R Journal of Physical Chemistry C 2012 116 4921 (80) Xu Q Zhong C Journal of Physical Chemistry C 2010 114 5035 (81) Lukose B Supronowicz B St Petkov P Frenzel J Kuc A B Seifert G Vayssilov G N Heine T Physica Status Solidi B-Basic Solid State Physics 2012 249 335 (82) Assfour B Seifert G Chemical Physics Letters 2010 489 86 (83) Zhao L Zhong C L Journal of Physical Chemistry C 2009 113 16860 (84) Schmid R Tafipolsky M Journal of the American Chemical Society 2008 130 12600 (85) Han S S Goddard W A III Journal of Physical Chemistry C 2007 111 15185 (86) Suh M P Park H J Prasad T K Lim D-W Chemical Reviews 2012 112 782 (87) Furukawa H Yaghi O M Journal of the American Chemical Society 2009 131 8875 (88) Wong-Foy A G Matzger A J Yaghi O M Journal of the American Chemical Society 2006 128 3494 (89) Mendoza-Cortes J L Han S S Furukawa H Yaghi O M Goddard III W A Journal of Physical Chemistry A 2010 114 10824 (90) Doonan C J Tranchemontagne D J Glover T G Hunt J R Yaghi O M Nature Chemistry 2010 2 235 (91) Getman R B Bae Y-S Wilmer C E Snurr R Q Chemical Reviews 2012 112 703 (92) Han S S Furukawa H Yaghi O M Goddard III W A Journal of the American Chemical Society 2008 130 11580 (93) Furukawa H Ko N Go Y B Aratani N Choi S B Choi E Yazaydin A O Snurr R Q OKeeffe M Kim J Yaghi O M Science 2010 329 424 (94) Garberoglio G Langmuir 2007 23 12154 (95) Assfour B Seifert G Microporous and Mesoporous Materials 2010 133 59
80
(96) Lan J Cao D Wang W Journal of Physical Chemistry C 2010 114 3108 (97) Yang Q Zhong C Langmuir 2009 25 2302 (98) Garberoglio G Vallauri R Microporous and Mesoporous Materials 2008 116 540 (99) Lan J H Cao D P Wang W C Ben T Zhu G S Journal of Physical Chemistry Letters 2010 1 978 (100) Furukawa H Miller M A Yaghi O M Journal of Materials Chemistry 2007 17 3197 (101) Babarao R Jiang J Energy amp Environmental Science 2008 1 139 (102) Choi Y J Choi J H Choi K M Kang J K Journal of Materials Chemistry 2011 21 1073 (103) Lan J Cao D Wang W Smit B Acs Nano 2010 4 4225 (104) Wang L Yang R T Energy amp Environmental Science 2008 1 268 (105) Krishna R van Baten J M Industrial amp Engineering Chemistry Research 2011 50 7083 (106) Keskin S Journal of Physical Chemistry C 2012 116 1772 (107) Liu Y Liu D Yang Q Zhong C Mi J Industrial amp Engineering Chemistry Research 2010 49 2902 (108) Keskin S Sholl D S Langmuir 2009 25 11786 (109) Klontzas E Tylianakis E Froudakis G E Nano Letters 2010 10 452 (110) Zhu Y Zhou J Hu J Liu H Hu Y Chinese Journal of Chemical Engineering 2011 19 709 (111) Mendoza-Cortes J L Pascal T A Goddard W A III Journal of Physical Chemistry A 2011 115 13852 (112) Lino M A Lino A A Mazzoni M S C Chemical Physics Letters 2007 449 171 (113) Krishnan A Dujardin E Ebbesen T W Yianilos P N Treacy M M J Physical Review B 1998 58 14013 (114) Kim D Jung D H Kim K-H Guk H Han S S Choi K Choi S-H Journal of Physical Chemistry C 2012 116 1479 (115) Kim D Jung D H Choi S-H Kim J Choi K Journal of the Korean Physical Society 2008 52 1255 (116) Farha O K Yazaydin A O Eryazici I Malliakas C D Hauser B G Kanatzidis M G Nguyen S T Snurr R Q Hupp J T Nature Chemistry 2010 2 944 (117) Wu M M Wang Q Sun Q Jena P Kawazoe Y Journal of Chemical Physics 2010 133 (118) Choi Y J Lee J W Choi J H Kang J K Applied Physics Letters 2008 92 (119) Mendoza-Cortes J L Han S S Goddard W A III Journal of Physical Chemistry A 2012 116 1621 (120) Cao D Lan J Wang W Smit B Angewandte Chemie-International Edition 2009 48 4730 (121) Lan J H Cao D P Wang W C Langmuir 2010 26 220 (122) Klontzas E Tylianakis E Froudakis G E Journal of Physical Chemistry C 2009 113 21253 (123) Sun Y Ben T Wang L Qiu S Sun H Journal of Physical Chemistry Letters 2010 1 2753 (124) Li F Zhao J Johansson B Sun L International Journal of Hydrogen Energy 2010 35 266 (125) Zou X Zhou G Duan W Choi K Ihm J Journal of Physical Chemistry C 2010 114 13402 (126) Babarao R Dai S Jiang D-e Langmuir 2011 27 3451
81
Appendix B
Structural properties of metal-organic frameworks within the density-functional based tight-binding method
Binit Lukose Barbara Supronowicz Petko St Petkov Johannes Frenzel Agnieszka Kuc
Gotthard Seifert Georgi N Vayssilov and Thomas Heine
Phys Status Solidi B 2012 249 335ndash342
DOI 101002pssb201100634
Abstract Density-functional based tight-binding (DFTB) is a powerful method to describe large
molecules and materials Metal-organic frameworks (MOFs) materials with interesting catalytic
properties and with very large surface areas have been developed and have become commercially
available Unit cells of MOFs typically include hundreds of atoms which make the application of
standard density-functional methods computationally very expensive sometimes even unfeasible
The aim of this paper is to prepare and to validate the self-consistent charge-DFTB (SCC-DFTB)
method for MOFs containing Cu Zn and Al metal centers The method has been validated against
full hybrid density-functional calculations for model clusters against gradient corrected density-
functional calculations for supercells and against experiment Moreover the modular concept of
MOF chemistry has been discussed on the basis of their electronic properties We concentrate on
MOFs comprising three common connector units copper paddlewheels (HKUST-1) zinc oxide Zn4O
tetrahedron (MOF-5 MOF-177 DUT-6 (MOF-205)) and aluminum oxide AlO4(OH)2 octahedron (MIL-
53) We show that SCC-DFTB predicts structural parameters with a very good accuracy (with less than
82
5 deviation even for adsorbed CO and H2O on HKUST-1) while adsorption energies differ by 12 kJ
mol1 or less for CO and water compared to DFT benchmark calculations
1 Introduction In reticular or modular chemistry molecular building blocks are stitched together to
form regular frameworks [1] With this concept it became possible to construct framework
compounds with interesting structural and chemical composition most notably metal-organic
frameworks (MOFs) [1ndash10] and covalent organic frameworks (COFs) [11ndash17] The interest in MOFs
and COFs is not limited to chemistry these crystalline materials are also interesting for applications
in information storage [18] sieving of quantum liquids [19] hydrogen storage [20] and for fuel cell
membranes [21ndash23]
Covalent organic framework and MOF frameworks are composed by combining two types of building
blocks the so-called connectors typically coordinating in four to eight sites and linkers which have
typically 2 sometimes also three or four connecting sites Fig 1 illustrates a simplified representation
of the topology of connectors and linkers by using secondary building units (SBUs) (see Fig 1)
Figure 1 The connector and the linker of the frameworks CuBTC (HKUST-1) MOF-177 MOF-5 DUT-6 (MOF-205) and MIL-53 and their respective secondary building units (SBUs) The arrows indicate the connection points of the fragments Red oxygen green carbon white hydrogen yellow copperzinc and blue aluminum
Linkers are organic molecules with carboxylic acid groups at their connection sites which form
bonds to the connectors (typically in a solvothermal condensation reaction) They can carry
functional groups which can make them interesting for applications in catalysis [24] Connectors are
83
either metal organic units as for example the well-known Zn4O(CO2)6 unit of MOF-5 [8] the
Cu2(HCOO)4 paddle wheel unit of HKUST-1 [10] or covalent units incorporating boron oxide linking
units for COFs [11] As the building blocks of MOFs and COFs comprise typically some ten atoms unit
cells quickly get very large In particular if adsorption or dynamic processes in MOFs and COFs are of
interest (super)cells of some 1000 atoms need to be processed While standard organic force fields
show a reasonable performance for COFs [25] the creation of reliable force fields is not
straightforward for MOFs as transferable parameterization of the transition metal sites is an issue
even though progress has been achieved for selected materials [26 27] The difficulty to describe
transition metals in particular if they are catalytically active as in HKUST-1 limits the applicability of
molecular mechanics (MM) even for QMMM hybrid methods [28]
On the other hand the density-functional based tight-binding (DFTB) method with its self-consistent
charge (SCC) extension to improve performance for polar systems is a computationally feasible
alternative This non-orthogonal tight-binding approximation to density-functional theory (DFT)
which has been developed by Seifert Frauenheim and Elstner and their groups [29ndash32] (for a recent
review see Ref [30]) has been successfully applied to a large-scale systems such as biological
molecules [33ndash36] supramolecular systems [37 38] surfaces [39 40] liquids and alloys [41 42] and
solids [43] Being a quantum-mechanical method and hence allowing the description of breaking and
formation of chemical bonds the method showed outstanding performance in the description of
processes such as the mechanical manipulation of nanomaterials [44ndash46] It is remarkable that the
method performs well for systems containing heavier elements such as transition metals as this
domain cannot be covered so far with acceptable accuracy by traditional semi-empirical methods [47
48] DFTB covers today a large part of the elements of the periodic table and parameters and a
computer code are available from the DFTBorg website
Recently we have validated DFTB for its application for COFs [16 17] and have shown that structural
properties and formation energies of COFs are well described within DFTB Kuc et al [49] have
validated DFTB for substituted MOF-5 frameworks where connectors are always the Zn4O(CO2)6 unit
which has been combined with a large variety of organic linkers In this work we have revised the
DFTB parameters developed for materials science applications and validated them for HKUST-1 and
being far more challenging for the interaction of its catalytically active Cu sites with carbon
monoxide (CO) and water The Cu2(HCOO)4 paddle wheel units are electronically very intriguing on a
first note the electronic ground state of the isolated paddle wheel is an antiferromagnetic singlet
state which cannot be described by one Slater determinant and which is consequently not accessible
for KohnndashSham DFT However the energetically very close triplet state correctly describes structure
and electronic density of the system and also adsorption properties agree well with experiment [32
84
50 51] We therefore use DFT in the B3LYP hybrid functional representation as benchmark for DFTB
validations for HKUST-1 added by DFT-GGA calculations for periodic unit cellsWewill show that the
general transferability of the DFTB method will allow investigating structural electronic and in
particular dynamic properties
2 Computational details All calculations of the finite model and periodic crystal structures of MOFs
were carried out using the dispersion-corrected self-consistent density functional based tight-binding
(DC-SCC-DFTB in short DFTB) method [30 52ndash54] as implemented in the deMonNano code [55] Two
sets of parameters have been used the standard SCC-DFTB parameter set developed by Elstner et al
[53] for Zn-containing MOFs for Cu- and Al-containing materials we have extended our materials
science parameter set which has been developed originally for zeolite materials to include Cu For
this element we have used the standard procedure of parameter generation we have used the
minimal atomic valence basis for all atoms including polarization functions when needed Electrons
below the valence states were treated within the frozen-core approximation The matrix elements
were calculated using the local density approximation (LDA) while the short-range repulsive pair-
potential was fitted to the results from DFT generalized gradient approximation (GGA) calculations
For more details on DFTB parameter generation see Ref [30] Periodic boundary conditions were
used to represent frameworks of the crystalline solid state The conjugate-gradient scheme was
chosen for the geometry optimization An atomic force tolerance of 3x10-4 eV Aring-1 was applied
The reference DFT calculations of the equilibrium geometry of the investigated isolated MOF models
were performed employing the Becke three-parameter hybrid method combined with a LYP
correlation functional (B3LYP) [56 57] The basis set associated with the HayndashWadt [58] relativistic
effective core potentials proposed by Roy et al [59] was supplemented with polarization ffunctions
[60] and was employed for description of the electronic structure of Cu atoms The Pople 6-311G(d)
basis sets were applied for the H C and O atoms The calculations were performed with the
Gaussian09 program suite [61] For optimization of the periodic structure of CuBTC at DFT level the
electronic structure code Quickstep [62] which is a part of the CP2K package [63] was used
Quickstep is an implementation of the Gaussian plane wave method [64] which is based on the
KohnndashSham formulation of DFT
We have used the generalized gradient functional PBE [65] and the GoedeckerndashTeterndashHutter
pseudopotentials [64 66] in conjunction with double-z basis sets with polarization functions DZVP-
MOLOPT-SR-GTH basis sets obtained as described in Ref [67] but using two less diffuse primitives
Due to the large unit cell sampling of Brillouin zone was limited only to the G-point Convergence
85
criterion for the maximum force component was set to 45 x 10-3 Hartree Bohr1 The plane wave
basis with cutoff energy of 400 Ry was used throughout the simulations
The initial crystal structures were taken from experiment (powder X-ray diffraction (PXRD) data) The
cell parameters and the atomic positions were fully optimized using conjugate-gradient method at
the DFTB level For the reference DFT calculations only the atomic positions of experimental crystal
structures were minimized The cluster models were cut from the optimized structures and saturated
with hydrogen atoms (see Fig 2 for exemplary systems of Cu-BTC)
3 Results and discussion
31 Cu-BTC We have optimized the crystal structure of Cu-BTC (HKUST-1) [10] using cluster and the
periodic models The structural properties were compared to DFT results (see Table 1) The
geometries were obtained for the activated form (open metal sites) and in the presence of axial
water ligands (see Table 2) as the synthesized crystals often have H2O molecules attached to the
open metal sites of Cu (Fig 2b) We have also studied changes of the cluster model geometry in the
presence of CO (Fig 2c) and the results are also shown in Table 2 We have obtained good agreement
with experimental data as well as with DFT results
Figure 2 (a) Periodic and cluster models of Cu-BTC MOF as used in the computer simulations (b) cluster model with adsorbed H2O and (c) CO molecules
We have also simulated the XRD patterns of Cu-BTC which show very good structural agreement for
peak positions between the experimental and calculated structures The XRD pattern of DFT
optimized structure is nearly a copy of that of the experimental geometry However for DFTB
optimized patterns the relative intensities of some of the peaks notably that of the peaks at 2θ =
138 and 2θ = 158 are different from the experimental XRDs (see Fig 3) The differences in bond
angles between simulation and experiment may affect the sharpness of the signals and hence the
86
intensity To support this argument we have calculated the radial pair distribution function (g(r))
which details the probabilities of all atomndashatom distances in a given cell volume Figure S3 in the
Supporting Information shows g(r) of CundashCu CundashC CundashO CndashC and CndashO atom pairs of DFTB
optimized DFT optimized and experimentally modeled Cu-BTC unit cells The peaks of DFT as well as
DFTB optimized geometries are much broadened whereas the experimentally modeled geometry
has single peaks at the most predicted atomndashatom distances However it is to be noted that DFTB
optimized geometries give slightly shifted peaks compared to those of the DFT optimized geometry
They are broadened around the experimental values The distances between Cu and C atoms which
are not direct neighbors have the largest deviations from the experiment what indicates
shortcomings of the bond angles
Table 1 Selected bond lengths (Aring ) bond angles (⁰) and dihedral angles (⁰) of Cu-BTC optimized at DFTB and DFT level
Bond Type Cluster Model Periodic Model Exp
Cu-Cu 250 (257) 250 (250) 263 Cu-O 205 (198) 202 (198) 195 O-C 134 (133) 133-138 (128) 125
OCuO 836-971 (898) 892-907 (873-937)
891 896
Cu-O-O-Cu plusmn56 ˗ plusmn62 (plusmn02) plusmn04 (plusmn47 ˗ plusmn131) 0
O-C-C-C plusmn38 - plusmn50 (0) plusmn03 - plusmn05 (plusmn06 - plusmn105) 063
Cell paramet a=b=c=27283 (26343)
α=β=γ=90 (90) a=b=c=26343
α=β=γ=90
In detail the bond lengths and bond angles do not change significantly when going from the cluster
to the periodic model confirming the reticular chemistry approach The only exception is the OndashCundash
O bond angle that differs by 4ndash78 between the two systems at both levels of theory
87
Figure 3 Simulated XRD patterns of Cu-BTC MOF with the experimental unit cell parameters and optimized at the DFTB and DFT level of theory
The bond length between Cu atoms is slightly underestimated comparing with experimental (by
maximum 5) and DFT (maximum by 3) results while all other bond lengths are somewhat larger
at DFTB
All bond lengths stay unchanged or become longer in the presence of water molecules The most
striking example is the CundashCu bond where the change reaches 012ndash017 Aring compared with the
structure with activated Cu sites Also bond angles increase by up to 28 when the H2O is present
The CundashO bond length with the oxygen atom from the carboxylate groups is shorter than that with
the oxygen atoms from the water molecules indicating that the latter is only weakly bound to the
copper ions The OndashC distances (~133 Aring) correspond to values between that for the typical single
(142 Aring ) and double (122 Aring) oxygenndashcarbon bond The calculated CringndashCcarboxyl bond lengths of
146A deg indicate that these are typical single sp2ndashsp2 CndashC bonds while experimental findings give a
slightly longer value (15 Aring) for this MOF When comparing dihedral angles CundashOndashOndashCu and OndashCndashCndashC
of the clusters fairly large deviations between DFTB and DFT calculations and experiment are visible
due to the softer potential energy surface associated with these geometrical parameters In periodic
models however the agreement of DFT and DFTB with experiment and with each other is
significantly improved
The unit cell parameters with and without water molecules obtained at the DFTB level overestimate
the experimental data by less than 4 which gives a fairly good agreement if we take into account
high porosity of the material These lattice parameters increase only slightly from 27283 to 27323 Aring
in the presence of water
We have calculated the binding energies of H2O molecules attached to the Cu-metal sites within the
cluster model At the DFTB level Ebind of 40 kJ mol-1 was obtained in a good agreement with the DFT
results (52 kJ mol-1) as well as results from the literature (47 kJ mol-1 [50]) We have also calculated
88
the binding energy of CO which was twice smaller than that of H2O (165 and 28 kJ mol-1 for DFTB
and DFT respectively) suggesting much stronger binding of O atom (from H2O) than C atom (from
CO) While the bond lengths in the MOF structure do not change in the presence of either H2O or CO
the differences in the binding energy come from much longer bond distances (by around 07 Aring) for
CundashC than for CundashO in the presence of CO and water molecules respectively
Furthermore we have studied the electronic properties of periodic and cluster model Cu-BTC by
means of partial density-of-states (PDOS)We have compared especially the Cu-PDOS for s- p- and d-
orbitals (see Fig 4) The results show that the electronic structure stays unchanged when going from
the cluster models to the fully periodic crystal of Cu-BTC Since the copper atoms are of Cu(II) the d-
orbitals are just partially occupied what results in the metallic states at the Fermi level This is a very
interesting result as other ZnndashO-based MOFs are either semiconductors or insulators [49]
Table 2 Selected bond lengths (Aring) bond angles (˚) and dihedral angles (˚) of Cu-BTC optimized at DFTB and DFT levels with water molecules in the axial positions Cluster models are calculated using water and carbon monoxide molecules The adsorption energies Eads (kJ mol-1) of the guest molecules are given as well The DFT data are given in parenthesis
Bond Type Cluster Model +
H2O Periodic
Model+ H2O Cluster Model +
CO
Cu-Cu 267 (266) 262 (260) 250 (260)
Cu-O 205 (197-206) 210 (196-200) 206 (199)
O-C 134 (127) 133 (128) 134 (127)
OCuO 843-955 (889-905)
871-921 (842-930) 842-967 (896)
Cu-O-O-Cu plusmn59 - plusmn76 (plusmn06 ˗ plusmn10)
plusmn02 ˗ plusmn18 (plusmn40 ˗ plusmn136)
plusmn51 - plusmn58 (plusmn70)
O-C-C-C plusmn39 - plusmn53 (plusmn05 - plusmn11)
plusmn03 - plusmn05 (plusmn06 - plusmn105)
plusmn35 - plusmn43 (plusmn12)
Cu-O(H2O) Cu-C(CO) 237 (219) 244 (233-
255) 307 (239)
Eads -4045 (-5200) -1648
(-2800)
32 MOF-5 -177 DUT-6 (MOF-205) and MIL- 53 We have also studied the structural properties
of MOF structures with large surface areas using the SCC-DFTB method Very good agreement with
the experimental data shows that this method is applicable for MOFs of large structural diversity as
well as for coordination polymers based on the MOF-5 framework which has been reported earlier
[49] Tables 3 and 4 show selected bond lengths and bond angles of MOF-5 [68] MOF-177 [69] DUT-
6 (MOF-205) [70 71] and MIL-53 [72] respectively
89
MOFs-5 -177 and DUT-6 (MOF-205) are built of the same connector which is Zn4O(CO2)6
octahedron The difference is in the organic linker 14-benzenedicarboxylate (BDC) 135-
benzenetribenzoates (BTB) and BTB together with 26-naphtalenedicarboxylate (NDC) for MOF-5 -
177 andDUT-6 (MOF-205) respectively (see Fig 5)
Figure 4 DFTB calculations of PDOS of (top left) Cu in Cu-BTC and Zn in MOF-5 (top right) MOF-177 (bottom left) and DUT-6 (MOF-205) (bottom right) In each plot s-orbitals (top) p-orbitals (middle) and d-orbitals (bottom) are shown as computed from periodic (black) and cluster (red) models Cluster models are given in Fig S4
All three MOFs have different topologies due to the organic linkers where the number of
connections is varied or where two different linker types are present MOF-5 is the most simple and
it is the prototype in the isoreticular series (IRMOF-1) [68] it is a simple cubic topology with
threedimensional pores of the same size and the linkers have only two connection points In the
case of MOF-177 the linker is represented by a triangularSBU that means three connection points
are present This results in the topology with mixed (36)-connectivity [69] DUT-6 (MOF-205) has a
much more complicated topology due to two types of linkers The first one (NDC) has just two
90
connection points while the second is the same as in MOF-177 with three connection points One
ZnndashO-based connector is connected to two NDC and four BTB linkers The topology is very interesting
all rings of the underlying nets are fivefold and it forms a face-transitive tiling of dodectahedra and
tetrahedra with a ratio of 13 [73]
Figure 5 The crystal structures in the unit cell representations of (a) MOF-5 (b) MOF-177 (c) DUT-6 (MOF-205) and (d) MIL-53 red oxygen gray carbon white hydrogen and blue metal atom (Zn Al)
MIL-53 is a MOF structure with one-dimensional pores The framework is built up of corner-sharing
AlndashO-based octahedral clusters interconnected with BDC linkers The linkers have again two
connection points MIL-53 shows reversible structural changes dependent on the guest molecules
[74] It undergoes the so-called breathing mode depending on the temperature and the amount of
adsorbed molecules
In this case also the bond lengths and bond angles are slightly overestimated comparing with the
experimental structures but the error does not exceed 3
91
Table 3 Selected bond lengths (Aring) and bond angles (˚) of MOF-5 (424 atoms in unit cell) MOF-177 (808 atoms in unit cell) and DUT-6 (MOF-205) (546 atoms in unit cell) optimized at DFTB level The available experimental data is given in parenthesis [70-73+ Orsquo denotes the central O atom in the Zn-O-octahedron
Bond Type MOF-5 MOF-177 DUT-6
(MOF-205)
Zn-Zn 330 (317) 322-336 (306-330)
325-331 (318)
Zn-Orsquo 202 (194) 202 (193) 202 (194) Zn-O 204 (192) 202-206
(190-199) 202 205 (193)
O-C 128 (130) 128 (131) 128 (125) ZnOrsquoZn 1095 (1095) 1056-1124
(1055 1092) 107-1118 (1084 1100)
OZnO 1083 1108 (1061)
1048 1145 (981-1281)
1046-1112 (1062 1085)
Zn-O-O-Zn 0 (0 plusmn17 plusmn20) plusmn122 - plusmn347 (plusmn176 - plusmn374)
05 - plusmn62 (0 plusmn29)
O-C-C-C 00 (0 - plusmn24) (plusmn 35 - plusmn 336) (plusmn65 - plusmn374)
plusmn04 plusmn22 (0 plusmn174)
Cell paramet a=b=c=26472 (25832) α=β=γ=90 (90)
a=b=37872 (37072) c=3068 (30033) α=β=90 (90) γ=120 (120)
a=b=c=31013 (30353) α=β=γ=90 (90)
We have calculated PDOS for MOF-5 MOF-177 and DUT-6 (MOF-205) (see Fig 4) Their band gaps
calculated to be 40 35 and 31 eV respectively indicate that they are either semiconductors or
insulators We have calculated it for both periodic crystal structure and a cluster containing a metal-
oxide connector and all its carboxylate linkers
Table 4 Selected bond lengths (Aring) and bond angles (˚) of MIL-53 (152 atoms in unit cell) optimized at DFTB level
Bond Type DFTB Exp
Al-Al 341 331 Al-O 189 183-191 O-C 133 129 130 O-Al-O 884 912 886 898 Al-O-Al 1289 1249 Al-O-O-Al 0 0 O-C-C-O 06 03 Cell paramet a=1246
b=1732 c=1365 α=β=γ=90
a=1218 b=1713 c=1326 α=β=γ=90
4 Mechanical properties Due to the low-mass density the elastic constants of porous materials
are a very sensitive indicator of their mechanical stability For crystal structures like MOFs we have
92
studied these by means of bulk modulus (B) B can be calculated as a second derivative of energy
with respect to the volume of the crystal (here unit cell)
The result shows that CuBTC has bulk modulus of 3466 GPa what is in close agreement with
B=3517 GPa obtained using force-field calculations [73 74] and experiment [75] Bulk moduli for the
series of MOFs resulted in 1534 1010 and 1073 GPa for MOF-5 -177 and DUT-6 (MOF-205)
respectively For MOF-5 B=1537 GPa was reported from DFTGGA calculations using plane-waves
[76 77] The results show that larger linkers give mechanically less stable structures what might be
an issue for porous structures with larger voids For MIL-53 we have obtained the largest bulk
modulus of 5369 GPa keeping the angles of the pore fixed
5 Conclusions We have validated the DFTB method with SCC and London dispersion corrections for
various types of MOFs The method gives excellent geometrical parameters compared to experiment
and for small model systems also in comparison with DFT calculations Importantly this statement
holds not only for catalytically inactive MOFs based on the Zn4O(CO2)6 octahedron and organic linkers
which are important for gas adsorption and separation applications but also for catalytically active
HKUST-1 (CuBTC) This is remarkable as the Cu atoms are in the Cu2+ state in this framework DFTB
parameters have been generated and validated for Cu and the electronic structure contains one
unpaired electron per Cu atom in the unit cell which makes the electronic description technically
difficult but manageable within the SCC-DFTB method SCC-DFTB performs well for the frameworks
themselves as well as for adsorbed CO and water molecules
Partial density-of-states calculations for the transition metal centers reveal that the concept of
reticular chemistry ndash individual building units keep their electronic properties when being integrated
to the framework ndash is also valid for the MOFs studied in this work in agreement with a previous
study of COFs [16] The electronic properties computed using the cluster models and the periodic
structure contains the same features and hence cluster models are good models to study the
catalytic and adsorption properties of these materials This is in particular useful if local quantum
chemical high-level corrections shall be applied that require to use finite structures
We finally conclude that we have now a high-performing quantum method available to study various
classes of MOFs of unit cells up to 10000 atoms including structural characteristics the formation
and breaking of chemical and coordination bonds for the simulation of diffusion of adsorbate
molecules or lattice defects as well as electronic properties The parameters can be downloaded
from the DFTBorg website
93
References
[1] E A Tomic J Appl Polym Sci 9 3745 (1965)
2+ M Eddaoudi D B Moler H L Li B L Chen T M Reineke M OrsquoKeeffe and O M Yaghi Acc Chem Res
34 319 (2001)
[3] M Eddaoudi H L Li T Reineke M Fehr D Kelley T L Groy and O M Yaghi Top Catal 9 105 (1999)
[4] B F Hoskins and R Robson J Am Chem Soc 111 5962 (1989)
[5] K Schlichte T Kratzke and S Kaskel Microporous Mesoporous Mater 73 81 (2004) [6] J L C Rowsell A
R Millward K S Park and O M Yaghi J Am Chem Soc 126 5666 (2004)
7+ O M Yaghi M OrsquoKeeffe N W Ockwig H K Chae M Eddaoudi and J Kim Nature 423 705 (2003)
[8] M Eddaoudi J Kim N Rosi D Vodak J Wachter M OrsquoKeeffe and O M Yaghi Science 295 469 (2002)
9+ M OrsquoKeeffe M Eddaoudi H L Li T Reineke and O M Yaghi J Solid State Chem 152 3 (2000)
[10] S S Y Chui SM F Lo J P H Charmant A G Orpen and I D Williams Science 283 1148 (1999)
11+ A P Cote A I Benin N W Ockwig M OrsquoKeeffe A J Matzger and O M Yaghi Science 310 1166 (2005)
[12] A P Cote H M El-Kaderi H Furukawa J R Hunt and O M Yaghi J Am Chem Soc 129 12914 (2007)
[13] H M El-Kaderi J R Hunt J L Mendoza-Cortes A P Cote R E Taylor M OrsquoKeeffe and O M Yaghi
Science 316 268 (2007)
[14] S Wan J Guo J Kim H Ihlee and D L Jiang Angew Chem 120 8958 (2008)
[15] S Wan J Guo J Kim H Ihlee and D L Jiang Angew Chem 121 5547 (2009)
[16] B Lukose A Kuc J Frenzel and T Heine Beilstein J Nanotechnol 1 60 (2010)
[17] B Lukose A Kuc and T Heine Chem Eur J 17 2388 (2011)
[18] D S Marlin D G Cabrera D A Leigh and A M Z Slawin Angew Chem Int Ed 45 77 (2006)
[19] I Krkljus and M Hirscher Microporous Mesoporous Mater 142 725 (2011)
[20] M Hirscher Handbook of Hydrogen Storage (Wiley-VCH Weinheim 2010)
[21] H Kitagawa Nature Chem 1 689 (2009)
[22] M Sadakiyo T Yamada and H Kitagawa J Am Chem Soc 131 9906 (2009)
[23] T Yamada M Sadakiyo and H Kitagawa J Am Chem Soc 131 3144 (2009)
94
[24] K S Jeong Y B Go S M Shin S J Lee J Kim O M Yaghi and N Jeong Chem Sci 2 877 (2011)
[25] R Schmid and M Tafipolsky J Am Chem Soc 130 12600 (2008)
[26] M Tafipolsky S Amirjalayer and R Schmid J Comput Chem 28 1169 (2007)
[27] M Tafipolsky S Amirjalayer and R Schmid J Phys Chem C 114 14402 (2010)
[28] A Warshel and M Levitt J Mol Biol 103 227 (1976)
[29] G Seifert D Porezag and T Frauenheim Int J Quantum Chem 58 185 (1996)
[30] A F Oliveira G Seifert T Heine and H A Duarte J Braz Chem Soc 20 1193 (2009)
[31] D Porezag T Frauenheim T Koumlhler G Seifert and R Kaschner Phys Rev B 51 12947 (1995)
[32] T Frauenheim G Seifert M Elstner Z Hajnal G Jungnickel D Porezag S Suhai and R Scholz Phys
Status Solidi B 217 41 (2000)
[33] M Elstner Theor Chem Acc 116 316 (2006)
Supporting Information
Figure S4 Cluster models as used for the P-DOS calculations in Fig 4 MOF-5 MOF-177 and DUT-6 (MOF-205) (from left to right)
95
Figure S3 Radial pair distribution function (g(r)) of Cu-Cu Cu-C Cu-O C-C and C-O atom-pairs in a Cu-BTC MOF unit cell
96
Appendix C
The Structure of Layered Covalent-Organic Frameworks
Binit Lukose Agnieszka Kuc and Thomas Heine
Chem Eur J 2011 17 2388 ndash 2392
DOI 101002chem201001290
Abstract Covalent-Organic Frameworks (COFs) are a new family of 2D and 3D highly porous and
crystalline materials built of light elements such as boron oxygen and carbon For all 2D COFs an AA
stacking arrangement has been reported on the basis of experimental powder XRD patterns with the
exception of COF-1 (AB stacking) In this work we show that the stacking of 2D COFs is different as
originally suggested COF-1 COF-5 COF-6 and COF-8 are considerably more stable if their stacking
arrangement is either serrated or inclined and layers are shifted with respect to each other by ~14 Aring
compared with perfect AA stacking These structures are in agreement with to date experimental
data including the XRD patterns and lead to a larger surface area and stronger polarisation of the
pore surface
Introduction In 2005 Cocircteacute and co-workers introduced a new class of porous crystalline materials
Covalent Organic Frameworks (COFs)[1] where organic linker molecules are stitched together by
connectors covalent entities including boron and oxygen atoms to a regular framework These
materials have the genuine advantage that all framework bonds represent strong covalent
interactions and that they are composed of light-weight elements only which lead to a very low
mass density[2] These materials can be synthesized solvothermally in a condensation reaction and
97
are composed of inexpensive and non-toxic building blocks so their large-scale industrial production
appears to be possible
Figure 1 Calculated and experimental XRD patterns of all studied COFs (COF-1 top left COF-5 top right COF-6 bottom left COF-8 bottom right)
To date a number of different COF structures have been reported[1ndash3] From a topological
viewpoint we distinguish two- and three-dimensional COFs In two-dimensional (2D) COFs the
covalently bound framework is restricted to 2D layers The crystal is then similar as in graphite or
hexagonal boron nitride composed by a stack of layers which are not connected by covalent bonds
but held together primarily by London dispersion interactions
98
The COF structures have been analyzed by powder X-ray diffraction (PXRD) experiments The
topology of the layers is determined by the structure of the connector and linker molecules and
typically a hexagonal pattern is formed due to the 2m (D3h) symmetry of the connector moieties
The individual layers are then stacked and form a regular crystal lattice With one exception (COF-
1)[1] for all 2D-COFs AA (eclipsed P6mmm) interlayer stacking has been reported[3andashd f g] This
geometrical arrangement maximizes the proximity of the molecular entities and results in straight
channels orthogonal to the COF layers which are known from the literature[1 3a]
The connectors of 2D-COFs include oxygen and boron atoms which cause an appreciable polarization
The AA stacking arrangement maximizes the attractive London dispersion interaction between the
layers which is the dominating term of the stacking energy At the same time AA stacking always
results in a repulsive Coulomb force between the layers due to the polarized connectors It should be
noted that the eclipsed hexagonal boron nitride (h-BN)[4] has boron atoms in one layer serving as
nearest neighbours to nitrogen atoms in adjacent layers (AB stacking) The Coulomb interaction has
ruled out possible interlayer eclipse between atoms of alike charges In this article we aim at
studying the layer arrangements of solvent-free COFs based on the interlayer interactions and the
minimum variance Various lattice types have been considered all significantly more stable than the
reported AA stacked geometry In all 2D COFs we have studied the Coulomb interaction between the
layers leads to a modification of the stacking and shifts the layers by about one interatomic distance
(~14 Aring) with respect to each other (see Figure 1)
Breaking of the highly symmetric layer arrangement has been observed for COF-1 and COF-10 after
removal of guest molecules from pores leading to a turbostratic repacking of pore channels[1 5]
The authors have confirmed the disorder either by the changes in the PXRD spectrum taken before
and after adsorptionndashdesorption cycle[1] or by the changes in the adsorption behavior[5] The
disruption of layer arrangement is attributed to the force exerted by the guest molecules Formation
of poly disperse pores has also been verified[5] The lattice types that we discuss in this article on
the other hand are neither the result of the pressure from any external molecule in the pore nor
having more than one type of pores They are the resultant of minimum variance guided by Coulomb
and London dispersion interactions For the COF models under investigation perfect crystallinity has
been considered
Methods We have studied the experimentally known structures of COF-1 COF-5 COF-6 and COF-8
Structure calculations were carried out using the Dispersion-Corrected Self-Conistent-Charge
Density-Functional-Based Tight-Binding method (SCC-DFTB)[6] DFTB is based on a second-order
expansion of the KohnndashSham total energy in DFT with respect to charge density fluctuations This
does not require large amounts of empirical parameters however maintains all qualities of DFT The
99
accuracy is very much improved by the SCC extension The DFTB code (deMonNano[6a]) has
dispersion correction[6d] implemented to account for weak interactions and was used to obtain the
layered bulk structure of COFs and their formation energies The performance for interlayer
interactions has been tested previously for graphite[6d] All structures correspond to full geometry
optimizations (structure and unit cell) XRD patterns have been simulated using the Mercury
software[7] To allow best comparison with experiment for PXRD simulations we used the calculated
geometry of the layer and of the relative shifts between the layers but experimental interlayer
distances The electrostatic potential has been calculated using Crystal 09[8] at the DFTVWN level
with 6-31G basis set
Results and Discussion
In order to see the favorite stacking arrangement of the layers we have shifted every second layer in
two directions along zigzag and armchair vertices of the hexagonal pores (see Figure 2) The initial
stacking is AA (P6mmm) and the final zigzag shift gives AB stacking (P63mmc) Considering that the
attractive interlayer dispersion forces and repulsive interlayer orbital interactions are different we
have optimized the interlayer separation for each stacking Figure 2 show their total energies
calculated per formula unit that is per established bond between linkers and connectors with
reference to AA stacking for COF-5 and COF-1 In each direction the energy minimum is found close
to the AA stacking position shifted only by about one aromatic C-C bond length (~14 Aring) so that
either connector or linker parts become staggered with those in the adjacent layers leading to a
stacking similar to that of graphite for either connector (armchair shift) or linker (zigzag shift) For
COF-5 the staggering at the connector or linker depends whether the shift is armchair or zigzag
respectively while for COF-1 the zigzag shift alone can give staggering in both the aromatic and
boron-oxygen rings
The low-energy minima in the two directions are labeled following the common nomenclature as
zigzag and armchair respectively Both shifts result in a serrated stacking order (orthorhombic
Cmcm) which shows a significantly more attractive interlayer Coulomb interaction than AA stacking
(see Table 1) while most of the London dispersion attraction is maintained and consequently
stabilizes the material Still this configuration can be improved if we consider inclined stacking
(Figure 1) that is the lattice vector pointing out of the 2D plane is not any longer rectangular
reducing the lattice symmetry from Cmcm (orthorhombic) to C2m (monoclinic)
Besides we have calculated the monolayer formation energy (condensation energy Ecb) from the
total energies of the monolayer and of the individual building blocks and the stacking formation
energy from the total energies of the bulk structure and of the monolayer for four selected COFs The
100
COF building blocks are the same as those used for their synthesis[1 3a] (BDBA for COF-1 BDBA and
HHTP for COF-5 BTBA and HHTP for COF-6 BTPA and HHTP for COF-8) The final energies given per
formula unit that is per condensed bond are given in Table 1 The serrated and inclined stacking
structures are energetically more stable than AA and AB Interestingly within our computational
model zigzag and armchair shifting is energetically equivalent
Figure 2 Change of the total energy (per formula unit) when shifting COF-5 even layers in zigzag (top) and armchair (middle) directions and COF-1 in zigzag (bottom) direction The vector d shows the shift direction The low-energy minima in the two directions are labelled as zigzag and armchair respectively The equilibrium structures are shown as well
The formation energies of the individual COF structures are in agreement with full DFT calculations
We have calculated using ADF[9] the condensation energy of COF-1 and COF-5 using first-principles
DFT at the PBE[10]DZP[11] level to qualitatively support our results For simplicity we have used a
finite structure instead of the bulk crystal Their calculated Ecb energies are 77 and 15 kJmol-1
respectively for COF-1 and COF-5 These values support the endothermic nature of the condensation
101
reaction and are in excellent agreement with our DFTB results (91 and 21 kJmol-1 respectively see
Table 1)
The change of stacking should be visible in X-ray diffraction patterns because each space group has a
distinct set of symmetry imposed reflection conditions Powder XRD patterns from experiments are
available for all 2D COF structures[1 3andashd f g] We have simulated XRD patterns for AA AB serrated
Table 1 Calculated formation energies for the studied COF structures Ecb = condensation Esb = stacking energy EL = London dispersion energy Ee = electronic energy Energies are per condensed bond and are given in [kJ mol
-1+ lsquoarsquo and lsquozrsquo stand for armchair and zigzag respectively Note that the electronic energy Ee=Esb-EL
includes the electronic interlayer interaction and the formation energy of the monolayer and that the formation of the monolayers is endothermic
Structure Stacking Esb EL Ee
COF-5 AA -2968 -3060 092
AB -2548 -2618 070
serrated z -3051 -3073 022
serrated a -3052 -3073 021
inclined z -3297 -3045 -252
inclined a -3275 -3044 -231
Monolayer Ecb= 211
COF-1 AA -2683 -2739 056
AB -2186 -2131 -055
serrated z -2810 -2806 -004
inclined z -2784 -2788 004
Monolayer Ecb= 906
COF-6 AA -2881 -2963 082
AB -2127 -2146 019
serrated z -2978 -2996 018
serrated a -2978 -2995 017
inclined z -2946 -2975 029
inclined a -2954 -2974 021
Monolayer Ecb= 185
COF-8 AA -4488 -4617 129
102
AB -2477 -2506 029
serrated z -4614 -4646 032
serrated a -4615 -4647 032
inclined z -4578 -4612 035
inclined a -4561 -4591 030
Monolayer Ecb= 263
and inclined stacking forms for COF-5 COF-1 COF-6 and COF-8 (see Figure 1for their comparison
with experimental spectrum) For completeness we have studied XRD patterns of optimized COFs
using the experimentally determined[1 3a] interlayer separations this means we have kept the
layer geometry the same as the optimized structures and different stackings were obtained by
shifting adjacent layers accordingly
COF-1 was reported as having different powder spectra for samples before (as-synthesized) and after
removal of guest molecules with a particular mentioning about its layer shifting after removal We
have compared the two spectra with our simulated XRDs in order to see the stacking in the pure
form and how the stacking is changed at the presence of mesitylene guests Except that we have only
a vague comparison at angles (2θ) 11ndash15 the powder spectrum after guest removal is more similar
to the XRDs of serrated and inclined stacking structures A notable difference from AB is the absence
of high peaks at angles ~12 ~15 and ~19 This gives rise to the suggestion that COF-1 is not a
notable exception among the 2D COFs it follows the same topological trend as all other frameworks
of this class having one-dimensional (1D) pores as soon as the solvent is removed from the pores
This suggestion is supported by the experimental fact that the gas adsorption capacity of COF-1 is
only slightly lower than that of COF-6 for even as large gas molecules as methane and CO2[12] It is
not obvious how these molecules would enter an AB stacked COF-1 framework where the pores are
not connected by 1D channels (see Figure 1) Moreover our calculations suggest that the serrated
and inclined stackings are energetically favorable (see Table 1)
Indeed all the new stackings discussed in this article yield XRD patterns which are in agreement with
the currently available experimental data (see Figure 1) The inclined stackings have more peaks but
those are covered by the broad peaks in the experimental pattern The same is observed for the (002)
peaks of serrated stackings At present 2D COF synthesis methods are not yet capable to produce
crystals of sufficient quality to allow more detailed PXRD analysis in particular for the solvent-free
materials Microwave synthesis of COF-5 by rapid heating is reported[13] as much faster (~200 times)
compared with solvothermal methods however the structural details (XRD etc) remained
103
ambiguous We are confident that better crystals will be produced in future which will allow the
unambiguous determination of COF structures and can be compared to our simulations
Finally we want to emphasize that the suggested change of stacking is not only resulting in a
moderate change of geometry and XRD pattern The functional regions of COFs are their pores and
the pore geometry is significantly modified in our suggested structures compared to AA and AB
stackings First the inclined and serrated structures account for an increase of the surface area by 6
estimated for the interaction of the COFs with helium (3 for N2 5 for Ar 56 for Ne) Moreover
the shift between the layers exposes both boron and oxygen atoms to the pore surface leading to a
much stronger polarity than it can be expected for AA stacked COFs This difference is shown in
Figure 3 which plots the electrostatic potential of COF-5 in AA serrated and inclined stacking
geometries While the pore surface of AA stacked COF-5 shows a positively and negatively charged
stripes the other stacking arrangements show a much stronger alternation of charges indicating the
higher polarity of the surface The surface polarity can also be expressed in terms of Mulliken charges
of oxygen and boron atoms calculated at the DFT level which are COF-5 AA qB = 048 and qO = -048
COF-5 serrated qB = 047 and qO = -049 COF-5 inclined qB = 045 and qO = -048
Figure 3 Calculated electrostatic potential of COF-5 AA (left) serrated (middle) and inclined (right) stacking forms Blue - negative and red-positive values of the 005 isosurface
Conclusion We have studied 2D COF layer stackings energetically and found that energy minimum
structures have staggering of hexagonal rings at the connector or linker parts This can be achieved if
the bulk structure has either serrated or inclined order These newly proposed orders have their
simulated XRDs matching well with the available experimental powder spectrum Hence we claim
that all reported 2D COF geometries shall be re-examined carefully in experiment Due to the change
of the pore geometry and surface polarity the envisaged range of applications for 2D COFs might
change significantly We believe that these results are of utmost importance for the design of
functionalized COFs where functional groups are added to the pore surfaces
104
References
[1] A P Cocircteacute A I Benin N W Ockwig M OKeeffe A J Matzger O M Yaghi Science 2005 310 1166
[2] H M El-Kaderi J R Hunt J L Mendoza-Cortes A P Cote R E Taylor M OKeeffe O M Yaghi Science
2007 316 268
[3] a) A P Cocircteacute H M El-Kaderi H Furukawa J R Hunt O M Yaghi J Am Chem Soc 2007 129 12914 b) J
R Hunt C J Doonan J D LeVangie A P Cocircte O M Yaghi J Am Chem Soc 2008 130 11872 c) R W
Tilford W R Gemmill H C zur Loye J J Lavigne Chem Mater 2006 18 5296 d) R W Tilford S J Mugavero
P J Pellechia J J Lavigne Adv Mater 2008 20 2741 e) F J Uribe-Romo J R Hunt H Furukawa C Klock M
OKeeffe O M Yaghi J Am Chem Soc 2009 131 4570 f) S Wan J Guo J Kim H Ihee D L Jiang Angew
Chem 2008 120 8958 Angew Chem Int Ed 2008 47 8826 g) S Wan J Guo J Kim H Ihee D L Jiang
Angew Chem 2009 121 5547 Angew Chem Int Ed 2009 48 5439
[4] R T Paine C K Narula Chem Rev 1990 90 73
[5] C Doonan D Tranchemontagne T Glover J Hunt O Yaghi Nat Chem 2010 2 235
[6] a) T Heine M Rapacioli S Patchkovskii J Frenzel A M Koester P Calaminici S Escalante H A Duarte R
Flores G Geudtner A Goursot J U Reveles A Vela D R Salahub deMon deMonnano edn Mexico DF
Mexico and Bremen (Germany) 2009 b) A F Oliveira G Seifert T Heine H A Duarte J Braz Chem Soc
2009 20 1193 c) G Seifert D Porezag T Frauenheim Int J Quantum Chem 1996 58 185 d) L Zhechkov T
Heine S Patchkovskii G Seifert H A Duarte J Chem Theory Comput 2005 1 841
[7] a) httpwwwccdccamacukproductsmercury b) C F Macrae P R Edgington P McCabe E Pidcock
G P Shields R Taylor M Towler J Van de Streek J Appl Crystallogr 2006 39 453
[8] R Dovesi V R Saunders C Roetti R Orlando C M Zicovich- Wilson F Pascale B Civalleri K Doll N M
Harrison I J Bush P DrsquoArco M Llunell Crystal 102 ed
[9] a) httpwwwscmcom ADF200901 Amsterdam The Netherlands b) G te Velde F M Bickelhaupt S J
A van Gisbergen C Fonseca Guerra E J Baerends J G Snijders T Ziegler J Comput Chem 2001 22 931
[10] J P Perdew K Burke M Ernzerhof Phys Rev Lett 1996 77 3865
[11] E Van Lenthe E J Baerends J Comput Chem 2003 24 1142
[12] H Furukawa O M Yaghi J Am Chem Soc 2009 131 8875
[13] N L Campbell R Clowes L K Ritchie A I Cooper Chem Chem Mater 2009 21 204
105
Appendix D
On the reticular construction concept of covalent organic frameworks
Binit Lukose Agnieszka Kuc Johannes Frenzel and Thomas Heine
Beilstein J Nanotechnol 2010 1 60ndash70
DOI103762bjnano18
Abstract
The concept of reticular chemistry is investigated to explore the applicability of the formation of
Covalent Organic Frameworks (COFs) from their defined individual building blocks Thus we have
designed optimized and investigated a set of reported and hypothetical 2D COFs using Density
Functional Theory (DFT) and the related Density Functional based tight-binding (DFTB) method
Linear trigonal and hexagonal building blocks have been selected for designing hexagonal COF layers
High-symmetry AA and AB stackings are considered as well as low-symmetry serrated and inclined
stackings of the layers The latter ones are only slightly modified compared to the high-symmetry
forms but show higher energetic stability Experimental XRD patterns found in literature also
support stackings with highest formation energies All stacking forms vary in their interlayer
separations and band gaps however their electronic densities of states (DOS) are similar and not
significantly different from that of a monolayer The band gaps are found to be in the range of 17ndash
40 eV COFs built of building blocks with a greater number of aromatic rings have smaller band gaps
Introduction
In the past decade considerable research efforts have been expended on nanoporous materials due
to their excellent properties for many applications such as gas storage and sieving catalysis
106
selectivity sensoring and filtration [1] In 1994 Yaghi and co-workers introduced ways to synthesize
extended structures by design This new discipline is known as reticular chemistry [23] which uses
well-defined building blocks to create extended crystalline structures The feasibility of the building
block approach and the geometry preservation throughout the assembly process are the key factors
that lead to the design and synthesis of reticular structures
One of the first families of materials synthesized using reticular chemistry were the so-called Metal-
Organic Frameworks (MOFs) [4] They are composed of metal-oxide connectors which are covalently
bound to organic linkers The coordination versatility of constituent metal ions along with the
functional diversity of organic linker molecules has created immense possibilities The immense
potential of MOFs is facilitated by the fact that all building blocks are inexpensive chemicals and that
the synthesis can be carried out solvothermally MOFs are commercially available and the scale up of
production is continuing Since the discovery of MOFs many other crystalline frameworks have been
synthesized using reticular chemistry such as Metal-Organic Polyhedra (MOP) [5] Zeolite
Imidazolate Frameworks (ZIFs) [6] and Covalent Organic Frameworks (COFs) [7]
In 2005 Cocircteacute and co-workers introduced COF materials [7-14] where organic linker molecules are
stitched together by covalent entities including boron and oxygen atoms to form a regular
framework These materials have the distinct advantage that all framework bonds represent strong
covalent interactions and that they are composed of light-weight elements only which lead to a very
low mass density [7-9] These materials can be synthesized by wet-chemical methods by
condensation reactions and are composed of inexpensive and non-toxic building blocks so their
large-scale industrial application appears to be possible From a topological viewpoint we distinguish
two- and three-dimensional COFs In two-dimensional (2D) COFs the covalently bound framework is
restricted to layers The crystal is then similar as in graphite composed of a stack of layers which
are not connected by covalent bonds
COFs compared with MOFs have lower mass densities due to the absence of heavy atoms and
therefore might be better for many applications For example the gravimetric uptake of gases is
almost twice as large as that of MOFs with comparable surface areas [1516] Because COFs are fairly
new materials many of their properties and applications are still to be explored
Recently we have studied the structures of experimentally well-known 2D COFs [17] We have found
that commonly accepted 2D structures with AA and AB kinds of layering are energetically less stable
than inclined and serrated forms This is because AA stacking maximises the Coulomb repulsion due
to the close vicinity of charge carrying atoms alike (O B atoms) in neighbouring layers The serrated
and inclined forms are only slightly modified (layers are shifted with respect to each other by asymp14 Aring)
107
and experience less Coulomb forces between the layers compared to AA stacking This is equivalent
to the energetic preference of graphite for an AB (Bernal) over an AA form (simple hexagonal) if we
ignore the fact that interlayer ordering in serrated and inclined forms are not uniform everywhere A
known example of this is that in eclipsed hexagonal boron nitride (h-BN) boron atoms in one layer
serve as nearest neighbours to nitrogen atoms in adjacent layers (AB stacking) The Coulomb
interaction rules out possible interlayer eclipse between atoms with similar charges in this case
In the present work we aim to explore how far the concept of reticular chemistry is applicable to the
individual building units which define COFs For this purpose we have investigated a set of reported
and hypothetical 2D COFs theoretically by exploring their structural energetic and electronic
properties We have compared the properties of the isolated building blocks with those of the
extended crystal structures and have found that the properties of the building units are indeed
maintained after their assembly to a network
Results and Discussion
Structures and nomenclature
We have considered four connectors (IndashIV) and five linkers (andashe) for the systematic design of a
number of 2D COFs (Figure 1) Each COF was built from one type of connector and one type of linker
thus resulting in the design of 20 different structures Moreover we have considered two
hypothetical reference structures which are only built from connectors I and III (no linker is present)
REF-I and REF-III Properties of other COFs were compared with the properties of these two
structures Some of the studied COFs are already well known in the literature [781314] and we
continue to use their traditional nomenclature while hypothetical ones are labelled in the
chronological order with an M at the end which stands for modified
Figurel 1 The connector (IndashIV) and linker (andashe) units considered in this work The same nomenclature is used in the text Carbon ndash green oxygen ndashred boron ndash magenta hydrogen ndash white
108
Using the secondary building unit (SBU) approach we can distinguish the connectors between
trigonal [T] (connectors I II III) and hexagonal [H] (connector IV) and the linkers between linear [l]
(linkers a b e) and trigonal [t] (linkers c d) Topology of the layer is determined by the geometries
of the connector and linker molecules and typically a hexagonal pattern is formed due to the D3h
symmetry of the connector moieties Based on these topologies of the constituent building blocks
we have classified the studied COFs into four groups Tl Tt Hl and Ht (Figure 2) Hereafter we will
use this nomenclature to describe the COF topologies
Figurel 2 Topologies of 2D COFs considered in this work (from the left) Tl Tt Hl and Ht Red and blue blocks are secondary building units corresponding to connectors and linkers respectively
We have considered high-symmetry AA and AB kinds of stacking (hexagonal) and low-symmetry
serrated (orthorhombic) and inclined (monoclinic) kinds of stacking of the layers The latter two were
discussed in a previous work on 2D COFs [17] As an example the structure of COF-5 [7] in different
kinds of stacking of layers is shown in Figure 3 In eclipsed AA stacking atoms of adjacent layers lie
directly on top of each other whilst in staggered AB stacking three-connected vertices lie directly on
top of the geometric center of six-membered rings of neighbouring layers In both serrated and
inclined kinds of stacking the layers are shifted with respect to each other by approximately 14 Aring
resulting in hexagonal rings in the connector or linker being staggered with those in the adjacent
layers In serrated stacking alternate layers are eclipsed In inclined stacking layers lie shifted along
one direction and the lattice vector pointing out of the 2D plane is not rectangular For COFs made of
connector I due to the absence of five-membered C2O2B rings a zigzag shift leads to staggering in
both connector and linker parts For those made of other connectors staggering at the connector or
linker depends on whether the shift is armchair or zigzag respectively [17]
Structural properties
We have compared structural properties of isolated building blocks with those of the extended COF
structures Negligible differences have been found in the bond lengths and bond angles of the
building blocks and the corresponding crystal structures This indicates that the structural integrity of
the building blocks remains unchanged while forming covalent organic frameworks confirming the
109
principle of reticular chemistry In addition the CndashB BndashO and OndashC bond lengths are almost the same
when different COF structures are compared (see Table S1 in Supporting Information File 1) The
experimental bond lengths are asymp154 Aring for CndashB asymp138 Aring for OndashC and 137ndash148 Aring for BndashO However
in the case of COF-1 the experimental values are slightly larger (160 Aring for CndashB and 151 Aring for BndashO)
This could be the reason why our calculated bond lengths for COF-1 are much shorter than the
experimental values while all the other structures agree within 5 error The calculated CndashC bond
lengths vary in the range from 136ndash147 Aring (Figure S2 in Supporting Information File 1) and are the
same for the COFs and their constituent building blocks at the respective positions of the carbon
atoms In addition the reference structures REF-I and REF-III have direct BndashB bond lengths of 167 Aring
and 166 Aring respectively which is shorter by 014 Aring than a typical BndashB bond length The calculated
bond angles OBO in B3O3 and C2O2B rings are 120deg and 113deg respectively
Figure 3 Layer stackings considered AA AB serrated and inclined
Interlayer distances (d) which is the shortest distance between two layers (equivalent to c for AA
c2 for AB and serrated) are different in all kinds of stacking AB stacked 2D COFs have shorter
interlayer distances than the corresponding AA serrated and inclined stacked structures Among the
latter three AA stacked COFs have higher values for d because of the higher repulsive interlayer
orbital interactions resulting from the direct overlap of polarized alike atoms between the adjacent
layers This results in higher mass densities for AB stacked COF analogues Serrated and inclined
stacks have only slightly higher mass densities compared to AA The differences in mass densities for
all kinds of stacking are attributed to the differences in their interlayer separations The d values of
most of the COFs are larger than that of graphite in AA stacking but smaller in AB stacking
Cell parameters (a) and mass densities (ρ) of all the COFs constructed from the considered
connectors and linkers are shown in Table 1 (and in Table S3 in Supporting Information File 1) Mass
densities of all the COFs are much lower than that of graphite (227 gmiddotcmminus3) and diamond (350
gmiddotcmminus3) AAserratedinclined stacked COF-10s have the lowest mass densities (045046046
gmiddotcmminus3) which is lower than that of MOF-5 (059 gmiddotcmminus3) [4] and comparable to that of highly porous
MOF-177 (042 gmiddotcmminus3) [18]
110
In order to identify the stacking orders we have analyzed X-ray diffraction (XRD) patterns of the well-
known COFs (COF-10 TP COF PPy-COF see Figure 4) in all the above discussed stacking kinds The
change of stacking should be visible in XRDs because each space group has a distinct set of symmetry
imposed reflection conditions The XRD patterns of AA serrated and inclined stacking kinds which
differ within a slight shift of adjacent layers to specific directions are in agreement with the presently
available experimental data [81314] Their peaks are at the same angles as in the experimental
spectrum whereas AB stacking clearly shows differences The slight differences in the (001) angle
between each stacking resemble the differences in their interlayer separations The inclined
stackings have more peaks however they are covered by the broad peaks in the experimental
patterns Similar results for COF-1 COF-5 COF-6 and COF-8 have been discussed in our previous
work [17]
Figure 4 The calculated and experimental [81314] XRD patterns of PPy-COF (top) COF-10 (middle) and TP COF (bottom)
111
Table 1 The calculated unit cell parameter a [Aring] interlayer distance d [Aring] and mass density ρ [gmiddotcmminus3
] for AA and AB stacked COFs Note that the cell parameter a is the same for all stacking types Experimental data [719] is given in parentheses
COF Building
Blocks
a d ρ
AA AB AA AB
COF-1 I-a 1502 (15620) 351 313 (332) 094 106
COF-1M I-b 2241 349 304 068 078
COF-2M I-c 1492 347 312 095 106
COF-3M I-d 0747 349 327 153 164
PPy-COF I-e 2232 (22163) 349 (3421) 297 084 099
COF-5 II-a 3014 (30020) 347 (3460) 326 056 060
COF-10 II-b 3758 (37810) 347 (3476) 318 045 (045) 050
COF-8 II-c 2251 (22733) 346 (3476) 320 071 (070) 077
COF-6 II-d 1505 (15091) 346 (3599) 327 104 110
TP COF II-e 3750 (37541) 348 (3378) 320 051 056
COF-4M III-a 2171 350 318 073 080
COF-5M III-b 2915 350 318 055 061
COF-6M III-c 1833 345 318 083 090
COF-7M III-d 1083 350 330 129 136
TP COF-1M III-e 2905 349 310 065 074
COF-8M IV-a 1748 359 329 140 148
COF-9M IV-b 2176 349 330 117 174
COF-10M IV-c 2254 342 336 127 128
COF-11M IV-d 1512 346 338 168 172
TP COF-2M IV-e 2173 347 332 134 140
REF-I I 0773 359 336 144 148
REF-III III 1445 353 336 104 121
Graphite 247 343 335 220 227
112
Energetic stability
We have considered dehydration reactions the basis of experimental COF synthesis to calculate
formation energies of COFs in order to predict their energetic stability Molecular units 14-
phenylenediboronic acid (BDBA) 11rsquo-biphenyl]-44-diylboronic acid (BPDA) 5rsquo-(4-boronophenyl)-
11rsquo3rsquo1rdquo-terphenyl]-44rdquo-diboronic acid (BTPA) benzene-135-triyltriboronic acid (BTBA) and
pyrene-27-diylboronic acid (PDBA) were considered as linkers andashe respectively with -B(OH)2 groups
attached to each point of extension (Figure 5) Self-condensation of these building blocks result in
the formation of B3O3 rings and the resultant COFs are those made of connector I and the
corresponding linkers This process requires a release of three or six water molecules in case of t or l
topology respectively
Figure 5 The reactants participating in the formation of 2D COFs
Co-condensation of the above molecular units with compounds such as 23671011-
hexahydroxytriphenylene (HHTP) hexahydroxybenzene (HHB) and dodecahydroxycoronene (DHC)
(Figure 5) gives rise to COFs made of connectors II III and IV respectively and the corresponding
linkers Self-condensation of tetrahydroxydiborane (THDB) and co-condensation of HHB with THDB
result in the formation of the reference structures REF-I and REF-III respectively In relation to the
corresponding connectorlinker topologies these building blocks satisfy the following equations of
the co-condensation reaction for COF formation
2 2 3 COF 12 H O Tl T l (1)
113
2 1 1 COF 6 H O Tt T t (2)
2 1 3 COF 12 H O Hl H l (3)
2 1 2 COF 12 H O Ht H t (4)
with a stochiometry appropriate for one unit cell The number of covalent bonds formed between
the building blocks is equivalent to the number of released water molecules we refer to it as
ldquoformula unitrdquo and will give all energies in the following in kJmiddotmolminus1 per formula unit
Table 2 The calculated energies [kJ molminus1
] per bond formed between building blocks for AA and AB stacked COFs Ecb is the condensation energy Esb is the stacking energy and Efb is the COF formation energy (Efb = Ecb
+ Esb) The calculated band gaps Δ eV+ are given as well
COF Building
Blocks
Mono-
layer
AA AB
Ecb Esb Efb ∆ Esb Efb ∆
COF-1 I-a 906 -2683 -1777 33 -2187 -1280 36
COF-1M I-b 949 -4266 -3366 27 -2418 -1469 31
COF-2M I-c 956 -5727 -4771 28 -4734 -3778 30
COF-3M I-d 763 -2506 -1742 38 -2801 -2037 40
PPy-COF I-e 858 -5723 -4866 24 -3855 -2998 26
COF-5 II-a 211 -2968 -2756 24 -2548 -2337 28
COF-10 II-b 317 -3766 -3448 23 -1344 -1026 26
COF-8 II-c 263 -4488 -4224 25 -2477 -2213 28
COF-6 II-d 185 -2881 -2695 28 -2127 -1942 31
TP COF II-e 231 -4453 -4222 24 -1480 -1250 27
COF-4M III-a -033 -1730 -1763 26 -880 -913 26
COF-5M III-b 007 -2533 -2526 25 -972 -965 25
COF-6M III-c 014 -3231 -3217 26 -2134 -2120 28
114
COF-7M III-d -170 -1635 -1805 30 -1607 -1777 32
TP COF-1M III-e -014 -3226 -3240 24 -1277 -1291 24
COF-8M IV-a -787 -2756 -3543 18 -2680 -3467 21
COF-9M IV-b -836 -3577 -4414 17 -3003 -3839 21
COF-10M IV-c -947 -4297 -5244 18 -4192 -5140 22
COF-11M IV-d -403 -2684 -3087 21 -2833 -3236 24
TP COF-2M IV-e 030 -4345 -4315 18 -4117 -4087 21
We have calculated the condensation energy of a single COF layer formed from monomers (building
blocks hereafter called bb) according to the above reactions as follows
tot tot tot totcb m H2O 1 bb1 2 bb2E E n E ndash m E m E (5)
where Emtot ndash total energy of the monolayer EH2O
tot is the total energy of the water molecule Ebb1tot
and Ebb2tot are the total energies of interacting building blocks and n m1 m2 are the corresponding
stoichiometry numbers
We have also calculated the stacking energy Esb of layers
tot totsb nl s mE E n E (6)
where Enltot is the total energy of ns number of layers stacked in a COF Finally the COF formation
energy can be given as a sum of Ecb and Esb (see Table 1 and Table S1)
Electronic properties
All COFs including the reference structures are semiconductors with their band gaps lying between
17 eV and 40 eV (Table 2 and Table S4 in Supporting Information File 1) The largest band gaps are
of the reference structures while the lowest values are of COFs built from connector IV The band
gaps are different for different stacking kinds This difference can be attributed to the different
optimized interlayer distances Generally AB serrated and inclined stacked COFs have band gaps
comparable to or larger than that of their AA stacked analogues
115
We have calculated the electronic total density of states (TDOS) and the individual atomic
contributions (partial density of states PDOS) The energy state distributions of COFs and their
monolayers are studied and a comparison for COF-5 is shown in Figure 6 In all stacking kinds
negligible differences are found for the densities at the top of valence band and the bottom of
conduction band These slight differences suggest that the weak interaction between the layers or
the overlap of π-orbitals does not affect the electronic structure of COFs significantly Hence there is
almost no difference between the TDOS of AA AB serrated and inclined stacking kinds Therefore in
the following we discuss only the AA stacked structures
Figure 6 Total densities of states (DOS) (black) of AA (top left) AB (top right) serrated (bottom left) and inclined (bottom right) comparing stacked COF-5 with a monolayer (red) of COF-5 The differences between the TDOS of bulk and monolayer structures are indicated in green The Fermi level EF is shifted to zero
Figure 7 Partial density of states of carbon (left) boron (center) and oxygen (right) atoms of COFs built from type-I connectors and different linkers The vertical dashed line in each figure indicates the Fermi level EF
116
It is of interest to see how the properties of COFs change depending on their composition of different
secondary building units that is for different connectors and linkers PDOS of COFs built from type-I
connectors and different linkers are plotted in Figure 7 The PDOS of carbon atoms is compared with
that of graphite (AA-stacking) while PDOS of boron and oxygen atoms are compared with that of
REF-I a structure which is composed solely of connector building blocks Going from top to bottom
of the plots the number of carbon atoms per unit cell increases It can be seen that this causes a
decrease of the band gap Figure 8 shows the PDOS of COFs built from type-a linkers and different
connectors where the COFs are arranged in the increasing number of carbon atoms in their unit cells
from top to the bottom Again the C PDOS is compared with that of graphite while both REF-I and
REF- III are taken in comparison to O and B PDOS The observed relation between number of carbon
atoms and band gap is verified
Figure 8 Partial density of states of carbon (left) boron (center) and oxygen (right) atoms of COFs built from type-a linkers The vertical dashed line in each figure indicates the Fermi level EF
Conclusion
In summary we have designed 2D COFs of various topologies by connecting building blocks of
different connectivity and performed DFTB calculations to understand their structural energetic and
electronic properties We have studied each COF in high-symmetry AA and AB as well as low-
symmetry inclined and serrated stacking forms The optimized COF structures exhibit different
interlayer separations and band gaps in different kinds of layer stackings however the density of
states of a single layer is not significantly different from that of a multilayer The alternate shifted
layers in AB serrated and inclined stackings cause less repulsive orbital interactions within the layers
which result in shorter interlayer separation compared to AA stacking All the studied COFs show
117
semiconductor-like band gaps We also have observed that larger number of carbon atoms in the
unit cells in COFs causes smaller band gaps and vice versa Energetic studies reveal that the studied
structures are stable however notable difference in the layer stacking is observed from
experimental observations This result is also supported by simulated XRDs
Methods
We have optimized the atomic positions and the lattice parameters for all studied COFs All
calculations were carried out at the Density Functional Tight-Binding (DFTB) [2021] level of theory
DFTB is based on a second-order expansion of the KohnndashSham total energy in the Density-Functional
Theory (DFT) with respect to charge density fluctuations This can be considered as a non-orthogonal
tight-binding method parameterized from DFT which does not require large amounts of empirical
parameters however maintains all the qualities of DFT The main idea behind this method is to
describe the Hamiltonian eigenstates with an atomic-like basis set and replace the Hamiltonian with
a parameterized Hamiltonian matrix whose elements depend only on the inter-nuclear distances and
orbital symmetries [21] While the Hamiltonian matrix elements are calculated using atomic
reference densities the remaining terms to the KohnndashSham energy are parameterized from DFT
reference calculations of a few reference molecules per atom pair The accuracy is very much
improved by the self-consistent charge (SCC) extension Two computational codes were used
deMonNano code [22] and DFTB+ code [23] The first code has dispersion correction [24]
implemented to account for weak interactions and was used to obtain the layered bulk structure of
COFs and their formation energies The performance for interlayer interactions has been tested
previously for graphite [24] The second code which can perform calculations using k-points was
used to calculate the electronic properties (band structure and density of states) Band gaps have
been calculated as an additional stability indicator While these quantities are typically strongly
underestimated in standard LDA- and GGA-DFT calculations they are typically in the correct range
within the DFTB method For validation of our method we have calculated some of the structures
using Density Functional Theory (DFT) as implemented in ADF code [2526]
Periodic boundary conditions were used to represent frameworks of the crystalline solid state The
conjugatendashgradient scheme was chosen for the geometry optimization The atomic force tolerance of
3 times 10minus4 eVAring was applied The optimization using Γ-point approximation was performed with the
deMon-Nano code on 2times2times4 supercells Some of the monolayers were also optimized using the
DFTB+ code on elementary unit cells in order to validate the calculations within the Γ-point
approximation The number of k-points has been determined by reaching convergence for the total
energy as a function of k-points according to the scheme proposed by Monkhorst and Pack [27]
118
Band structures were computed along lines between high symmetry points of the Brillouin zone with
50 k-points each along each line XRD patterns have been simulated using Mercury software [2829]
We have also performed first-principles DFT calculations at the PBE [30] DZP [31] level to support
our results quantitatively For simplicity we have used finite structures instead of bulk crystals
Supporting Information
Table S1 The calculated bond lengths and angles of all studied COFs Corresponding experimental values found from the literature are shown in brackets
COF Building
Blocks
C-B B-O O-C OBO
COF-1 I-a 1498 (1600) 1393 (1509) 1203 (1200)
COF-1M I-b 1497 1393 1203
COF-2M I-c 1497 1392 1203
COF-3M I-d 1496 1392 1201
PPy-COF I-e 1498 1393 1202 (1190)
COF-5 II-a 1496 (1530) 1399 (1367) 1443 (1384) 1135dagger (1139)
COF-10 II-b 1495 (1553) 1399 (1465) 1443 (1385) 1135dagger (1120)
COF-8 II-c 1495 (1548) 1399 (1479) 1443 1135dagger
COF-6 II-d 1496 1399 1443 1135dagger
TP COF II-e 1496 (1542) 1399 (1396) 1444 (1377) 1135dagger (1182)
COF-4M III-a 1496 1398 1449 1135dagger
COF-5M III-b 1496 1398 1449 1136dagger
COF-6M III-c 1496 1399 1451 1134dagger
COF-7M III-d 1496 1398 1449 1136dagger
TP COF-1M III-e 1496 1398 1450 1136dagger
COF-8M IV-a 1496 1398 1445 1131dagger
COF-9M IV-b 1495 1398 1444 1131dagger
119
COF-10M IV-c 1495 1391 1418 1126dagger
COF-11M IV-d 1498 1399 1450 1134dagger
TP COF-2M IV-e 1499 1399 1447 1134dagger
B3O3 connectivity dagger C2B2O connectivity
It can be noticed that the bond lengths of experimentally known COF-1 is much larger compared to
our optimized bond lengths as well as that of other synthesized COFs
Figure S2 Calculated C-C bond lengths in connectors and linkers Hydrogen atoms are omitted for clarity
Table S3 The calculated unit cell parameters a interlayer distance d Aring+ and mass density ρ gcm-3
] for serrated (Sa serrated armchair Sz serrated zigzag) and inclined (Ia inclined armchair Iz inclined zigzag) stacked COFs
COF Building
Blocks
a d ρ
Sa Sz Ia Iz Sa Sz Ia Iz
COF-1 I-a 1502 343 343 097 097
COF-1M I-b 2241 341 342 069 069
COF-2M I-c 1492 340 339 097 097
COF-3M I-d 0747 341 342 157 156
PPy-COF I-e 2232 341 341 086 086
120
COF-5 II-a 3014 342 342 341 340 057 057 058 058
COF-10 II-b 3758 341 341 342 340 046 046 046 046
COF-8 II-c 2251 341 341 342 342 073 073 072 072
COF-6 II-d 1505 342 341 340 340 105 106 106 106
TP COF II-e 3750 342 341 342 342 052 052 052 052
COF-4M III-a 2171 344 344 345 344 074 074 074 074
COF-5M III-b 2915 343 342 343 343 056 056 056 056
COF-6M III-c 1833 341 341 342 341 084 084 084 084
COF-7M III-d 1083 344 343 340 344 131 131 132 131
TP COF-1M III-e 2905 343 342 343 342 066 067 066 066
COF-8M IV-a 1748 341 341 342 342 142 142 142 142
COF-9M IV-b 2176 341 341 341 342 119 119 119 119
COF-10M IV-c 2254 340 340 340 340 128 128 128 128
COF-11M IV-d 1512 341 341 340 340 171 171 171 171
TP COF-2M IV-e 2173 340 340 340 340 137 137 137 137
REF-I I 0773 349 345 148 15
REF-III III 1445 348 349 106 106
Table S4 The calculated energies [kJ mol-1
] per bond formed between building blocks for serrated (Sa serrated armchair Sz serrated zigzag) and inclined (Ia inclined armchair Iz inclined zigzag) stacked COFs Ecb ndash condensation energy Esb ndash stacking energy and Efb ndash COF formation energy (Efb = Ecb + Esb) The calculated band gaps ∆ eV+ are given as well
COF Sa Sz Ia Iz
Esb Efb Δ Esb Efb Δ Esb Efb Δ Esb Efb Δ
-1 -2810 -1904 36 -2786 -1880 36
-1M -4426 -3477 30 -4389 -3440 30
-2M -5967 -5011 30 -5833 -4877 30
121
-3M -2667 -1904 40 -2591 -1828 40
PPy- -5916 -5058 26 -5865 -5007 26
-5 -3052 -2841 27 -3051 -2840 26 -3275 -3064 26 -3297 -3086 26
-10 -3868 -3551 26 -3869 -3552 25 -3830 -3513 26 -4293 -3976 25
-8 -4615 -4352 27 -4614 -4351 27 -4571 -4308 27 -4573 -4310 27
-6 -2978 -2793 30 -2978 -2793 30 -3117 -2932 30 -3090 -2905 30
TP -4557 -4326 26 -4562 -4331 26 -4511 -4280 26 -4511 -4280 26
-4M -1811 -1844 28 -1813 -1846 28 -1769 -1802 28 -1880 -1913 28
-5M -2626 -2619 27 -2630 -2623 26 -2597 -2590 26 -2600 -2593 26
-6M -3375 -3361 28 -3381 -3367 28 -3487 -3473 28 -3349 -3335 28
-7M -1724 -1894 32 -1726 -1896 31 -2333 -2503 32 -1866 -2036 31
TP -1M -3327 -3341 26 -3334 -3348 26 -3340 -3354 26 -3353 -3367 26
-8M -2897 -3684 21 -2895 -3682 21 -2971 -3758 21 -2983 -3770 21
-9M -3732 -4568 20 -3733 -4569 20 -3684 -4520 20 -3706 -4542 20
-10M -4512 -5459 21 -4513 -5460 21 -4491 -5438 21 -4495 -5442 21
-11M -2831 -3234 24 -2834 -3237 24 -2816 -3219 24 -2830 -3233 24
TP -2M -4500 -4470 20 -4505 -4475 20 -4495 -4465 20 -4496 -4466 20
122
Appendix E
Stability and electronic properties of 3D covalent organic frameworks
Binit Lukose Agnieszka Kuc and Thomas Heine
Prepared for publication
Abstract Covalent Organic Frameworks (COFs) are a class of covalently linked crystalline nanoporous
materials versatile for nanoelectronic and storage applications 3D COFs in particular have very
large pores and low-mass densities Extensive theoretical studies of their energetic and mechanical
stability as well as their electronic properties are discussed in this paper All studied 3D COFs are
energetically stable materials though mechanical stability does not exceed 20 GPa Electronically all
COFs are semiconductors with band gaps depending on the number of sp2 carbon atoms present in
the linkers similar to 3D MOF family
Introduction
Covalent Organic Frameworks (COFs)1-3 comprise an emerging class of crystalline materials that
combines organic functionality with nanoporosity COFs have organic subunits stitched together by
covalent entities including boron carbon nitrogen oxygen or silicon atoms to form periodic
frameworks with the faces and edges of molecular subunits exposed to pores Hence their
applications can range from organic electronics to catalysis to gas storage and sieving4-7 The
properties of COFs extensively depend on their molecular constituents and thus can be tuned by
rational chemical design and synthesis289 Step by step reversible condensation reactions pave the
123
way to accomplish this target Such a reticular approach allows predicting the resulting materials and
leads to long-range ordered crystal structures
Since their first mentioning in the literature13 COFs are under theoretical investigation10-17 mainly for
gas storage applications Methods such as doping18-24 and organic linker functionalization2526 have
been suggested to improve their storage capacities In addition to the moderate pore size and
internal surface area COFs have the privileges of a low-weight material as they are made of light
elements Hence their gravimetric adsorption capacity is remarkably high10 The lowest mass density
ever reported for any crystalline material is that for COF-108 (018 gcm-3) Also their stronger
covalent bonds compared to related Metal-Organic Frameworks (MOFs)27 endorse thermodynamic
strength These genuine qualities of COFs make them attractive for hydrogen storage investigations
Crystallization of linked organic molecules into 2D and 3D forms was achieved in the years 20051 and
20073 respectively by the research group of O M Yaghi Several COFs have been synthesized since
then and some of the 2D COFs has been proven good for electronic or photovoltaic applications428-33
However the growth in this area appears to be slow compared to rapidly developing MOFs albeit
the promising H2 adsorption measurements53435 and a few synthetic improvements736-42
COF-102 and COF-103 were synthesized3 by the self-condensation of the non-planar tetra(4-
dihydroxyborylphenyl)methane (TBPM) and tetra(4-dihydroxyborylphenyl)silane (TBPS) respectively
(see Fig 1 for the building blocks and COF geometries) The co-condensation of these compounds
with triangular hexahydroxytriphenylene (HHTP) results in COF-108 and COF-105 respectively with
different topologies Yaghi et al3 have reported the formation of highly symmetric topologies ctn
(carbon nitride dI 34 ) and bor (boracite mP 34 ) when tetrahedral building blocks are linked
together with triangular ones The topology names were adopted from reticular chemistry structure
resource (RCSR)43 Simulated Powder X-ray diffraction in comparison with experimental powder
spectrum suggested ctn topology for COF-102 103 and 105 and bor topology for COF-108 The
condensation of tert-butylsilanetriol (TBST) and TBPM leads to the formation of COF-20244 This was
reported as ctn topology The COF reactants and schematic diagrams of ctn and bor topologies are
given in Figure1 The assembly of tetrahedral and linear units is very likely to result in a diamond-like
form COF-300 was formed according to the condensation of tetrahedral tetra-(4-anilyl)methane
(TAM) and linear terephthaldehyde (TA)45 However the synthesized structure was 5-fold
interpenetrated dia-c5 topology43
In this work we present theoretical studies of 3D COFs using density functional based methods to
explore their structural electronic energetic and mechanical properties Our previous studies on 2D
COFs4647 had resulted in questioning the stability of eclipsed arrangement of layers (AA stacking) and
124
suggesting energetically more stable serrated and inclined packing In this paper we attempt to
explore the stability and electronic properties of the experimentally known 3D COFs namely COF-
102 103 105 108 202 and 300 In the present studies we follow the reticular assembly of the
molecular units that form low-weight 3D COFs Considering the structural distinction from MOFs
COFs lack the versatility of metal ingredients what results in diverse properties48 A collective study is
then carried out to understand the characteristics and drawbacks of COFs
Figure 1 (Upper panel) Building blocks for the construction of 3D COFs (Lower panel) lsquoctnrsquo and lsquoborrsquo
networks formed by linking tetrahedral and triangular building units
Methods
COF structures were fully optimized using Self-consistent Charge -Density Functional based Tight-
Binding (SCC-DFTB) level of theory4950 Two computational codes were used deMonNano51 and
125
DFTB+52 The first code which has dispersion correction53 implemented to account for weak
interactions was used for the geometry optimization and stability calculations The second code
which can perform calculations using k-point sampling was used to calculate the electronic
properties (band structure and density of states) The number of k-points has been determined by
reaching convergence for the total energy as a function of k-points according to the scheme
proposed by Monkhorst and Pack54 Periodic boundary conditions were used to represent
frameworks of the crystalline solid state Conjugatendashgradient scheme was chosen for the geometry
optimization Atomic force tolerance of 3 times 10minus4 eVAring was applied The optimization using Γ-point
approximation was performed on rectangular supercells containing more than 1000 atoms For
validation of our method we have calculated energetic stability using Density Functional Theory (DFT)
at the PBE55DZP56 level as implemented in ADF code5758 using cluster models The cluster models
contain finite number of building units and correspond to the bulk topology of the COFs XRD
patterns have been simulated using Mercury software5960
In this work we continued to use the traditional nomenclature of the experimentally known COFs All
of the structures have the same tetrahedral blocks and differ only in the central sp3 atom (carbon or
silicon) that is included in our nomenclature
Bulk modulus (B) of a solid at absolute zero can be calculated as
(1) B = 2
2
dV
EdV
where V and E are the volume and energy respectively
Owing to the dehydration reactions we have calculated the formation (condensation) energy of each
COF formed from monomers (building blocks) as follows
(2) EF = Etot + n EH2Otot ndash (m1 Ebb1
tot + m2 Ebb2tot)
where Etot -- total energy of the COF EH2Otot -- total energy of water molecule Ebb1
tot and Ebb2tot -- total
energies of interacting building blocks n m1 m2 -- stoichiometry numbers
Results and Discussions
Structure and Stability
We have optimized the atomic positions and cell dimensions of the COFs in the experimentally
determined topologies Cell parameters in comparison with experimental values are given in Table 1
The calculated bond lengths are C-B = 150 C-C = 139-144 (COF-300) C(sp3)-C = 156 C-O = 142 B-
126
O = 140 Si-C = 188 Si-O = 186 C-N = 131-136 Aring These values agree within 6 error with the
experimental values34445
Mass densities (see Table 1) of COFs range between 02 to 05 g cm-3 and are smaller for the silicon at
the tetrahedral centers This implies that the presence of silicon atoms impacts more on the cell
volume than on the total mass That means the replacement of sp3 C with Si in COF-108 can change
its mass density to a slightly lower value To our best knowledge among all the natural or
synthesized crystals COF-108 has the lowest mass-weight
In order to validate our optimized structures we have simulated X-ray Diffraction of each COF and
compared them with the available experimental spectra (see Figure2) Almost all of the simulated
XRDs have excellent correlation in the peak positions with the experiment Only COF-300 shows
somehow significant differences in the intensities These differences may be attributed to the
presence of guest molecules in the synthesized COF-30045
Table 1 The calculated cell parameters [Aring] mass density ρ gcm-3
+ band gap Δ eV+ bulk modulus B GPa+
and formation energy EF [kJ mol-1
] for all the studied 3D COFs Experimental values are given in brackets
along with the cell parameters HOMO-LUMO gaps [eV] of the building blocks are also given in brackets
along with the band gaps
Structure Building
Blocks
Cell
parameters
ρ Δ B EF
COF-102 (C) TBPM 2707 (2717) 043 39 (43) 206 -14995
COF-103 (Si) TBPS 2817 (2825) 039 38 (41) 139 -14547
COF-105-C TBPM HHTP 4337 019 33 (43 34) 80 -18080
COF-105 (Si) TBPS HHTP 4444 (4489) 018 32 (41 34) 79 -17055
COF-108 (C) TBPM HHTP 2838 (2840) 017 32 (43 34) 37 -17983
COF-108-Si TBPS HHTP 2920 016 30 (41 34) 29 -18038
COF-202 (C) TBPM TBST 3047 (3010) 050 42 (43 139) 143 -7954
COF-202-Si TBPS TBST 3157 046 39 (41 139) 153 -7632
COF-300 (C) TAM TA 2932 925 049 23 (41 26) 144 -40286
127
(2828 1008)
COF-300-Si TAS TA 3039 959 044 23 (40 26) 133 -39930
tetra-(4-anilyl)silane
Figure 2 Simulated XRDs are compared with the experimental patterns from the literature COF-300
exhibits some differences between the simulated and experimental XRDs while others show reasonably
good match
The bulk modulus (B) is a measure of mechanical stability of the materials The calculated values of B
are given in Table 1 Compared to the force-field based calculation of bulk modulus by Schmid et
al61 these values are larger The bulk modulus of COF-105 and COF-108 are relatively small
compared with other COFs Considering that the two COFs differ only in the topology it may be
concluded that ctn nets are mechanically more stable compared to bor COFs with carbon atoms in
the center are mechanically more stable than those with silicon atoms Bulk modulus of COF-102
103 COF-202 and interpenetrated COF-300 are higher than many of the isoreticular MOFs62 and
comparable to IRMOF-6 (1241 GPa) MOF-5 (1537 GPa)63 bNon-interpenetrated COF-300 (single
framework dia-a topology43) has much lower bulk modulus of only 317 GPa
Calculated formation energies (Table 1) are given in terms of reaction energies according to Eq 2
Condensation of monomers to form bulk 3D structures is exothermic in all the cases supporting
reticular approach The presence of C or Si at the vertex center does not show any particular trend in
the formation energies We have calculated the formation energy of non-interpenetrated COF-300
(dia-a) to be -39346 kJ mol-1 which is comparable to the interpenetrated cases For a quantitative
comparison we have performed DFT calculations at the PBEDZP level as implemented in ADF code
on finite structures The obtained formation energies for COF-105-C COF-105 (Si) COF-108 (C) COF-
108-Si are -9944 -9968 -1245 -12436 respectively They are in a reasonable agreement with the
128
DFTB results A comparison between COF-105 and COF-108 suggests that bor nets are energetically
more favored than ctn nets
Electronic Properties
Band gaps (Δ) calculated for the 3D COFs are in the range of 23-42 eV (see Table 1) which show
their semiconducting nature similar to the hexagonal 2D COFs47 and MOFs62 The band gap
decreases with the increase of conjugated rings in the unit cell relative to the number of other atoms
Si atoms in comparison with carbon atoms at the tetrahedral centers give a relatively smaller Δ This
is evident from the partial density of states of C (sp3) in COF-102 and Si in COF-103 plotted in Figure3
Carbon atoms define the band edges of the PDOS (see Figure4) The band gaps of COF-105 and COF-
108 differ only slightly (see Figure5) which means that the band gap is nearly independent of the
topology This is because for each atom the coordination number and the neighboring atoms remain
the same in both ctn and bor networks albeit the topology difference DOS of non-interpenetrated
(dia-a) and 5-fold interpenetrated (dia-c5) COF-300 are plotted in Figure6 It may be concluded from
their negligible differences that interpenetration does not alter the DOS of a framework We have
shown similar results for 2D COFs47
Figure 3 Partial density of states (DOS) of sp3 C in COF-102 (top) and Si in COF-103 (bottom) The latter is
inverted for comparison The Fermi level EF is shifted to zero
129
Figure 4 Partial and total density of states of COF-103 The Fermi level EF is shifted to zero
Figure 5 Density of states of COF-108 (bor) and COF-105 (ctn) The two COFs differ only in the topology
130
Figure 6 Density of states of non-interpenetrated (dia-c1) and 5-interpenetrated (dia-c5) COF-300
We also have calculated HOMO-LUMO gaps of the molecular building blocks (see Table 1) In
comparison band gaps of COFs are slightly smaller than the smallest of the HOMO-LUMO gaps of the
building units
Conclusion
In summary we have calculated energetic mechanical and electronic properties of all the known 3D
COF using Density Functional Tight-Binding method Energetically formation of 3D COFs is favorable
supporting the reticular chemistry approach Mechanical stability is in line with other frameworks
materials eg MOFs and bulk modulus does not exceed 20 GPa Also all COFs are semiconducting
with band gaps ranging from 2 to 4 eV Band gaps are analogues to the HOMO-LUMO gaps of the
molecular building units We believe that this extensive study will define the place of COFs in the
broad area of nanoporous materials and the information obtained from the work will help to
strategically develop or modify porous materials for the targeted applications
131
Appendix F
Structure electronic structure and hydrogen adsorption capacity of porous aromatic frameworks
Binit Lukose Mohammad Wahiduzzaman Agnieszka Kuc and Thomas Heine
Prepared for publication
Abstract
Diamond-like porous aromatic frameworks (PAFs) form a new family of materials that contain only
carbon and hydrogen atoms within their frameworks These structures have very low mass densities
large surface area and high porosity Density-functional based calculations indicate that crystalline
PAFs have mechanical stability and properties similar to those of Covalent Organic Frameworks Their
exceptional structural properties and stability make PAFs interesting materials for hydrogen storage
Our theoretical investigations show exceptionally high hydrogen uptake at room temperature that
can reach 7 wt a value exceeding those of well-known metal- and covalent-organic frameworks
(MOFs and COFs)
Introduction
Porous materials have been widely investigated in the fields of materials science and technology due
to their applications in many important fields such as catalysis gas storage and separation template
materials molecular sensors etc1-3 Designing porous materials is nowadays a simple and effective
strategy following the approach of reticular chemistry4 where predefined building blocks are used to
132
predict and synthesize a topological organization in an extended crystal structure The most famous
and striking examples of such materials are Metal- and Covalent-Organic Frameworks (MOFs and
COFs)56 These new nanoporous materials have many advantages high porosity and large surface
areas lowest mass densities known for crystalline materials easy functionalization of building blocks
and good adsorption properties
Gas storage and separation by physical adsorption are very important applications of such
nanoporous materials and have been major subjects of science in the last two decades These
applications are based on certain physical properties namely presence of permanent large surface
area and suitable enthalpy of adsorption between the host framework and guest molecules
Attempts to produce materials with large internal surface area have been successful and some of the
notable accomplishments include the synthesis of MOF-2107 (BET surface area of 6240 m2 g-1 and
Langmuir surface area of 10400 m2 g-1) NU-1008 (BET surface area 6143 m2 g-1) or 3D COFs9 (BET
surface area 4210 m2 g-1 for COF-103)
More recently a new family of porous materials emerged So-called porous-aromatic frameworks
(PAFs) have surface areas comparable to MOFs eg PAF-110 has a BET surface area of 5640 m2 g-1 and
Langmuir surface area of 7100 m2 g-1 Since they are composed of carbon and hydrogen only they
have several advantages over frameworks containing heavy elements MOFs with coordination bonds
often suffer from low thermal and hydrothermal stability what might limit their applications on the
industrial scale The coordination bonds can be substituted with stronger covalent bonds as it was
realized in the case of COFs6 however this lowers significantly their surface areas comparing with
MOFs
Besides its exceptional surface area PAF-110 has high thermal and hydrothermal stability and
appears to be a good candidate for hydrogen and carbon dioxide storage applications PAFs have
topology of diamond with C-C bonds replaced by a number of phenyl rings (see Figure 1)
Experimentally obtained PAF-110 and PAF-1111 have eg one and four phenyl rings respectively
connected by tetragonal centers (carbon tetrahedral nodes) The simulated and experimental
hydrogen uptake capacities of such PAFs exceed the DOE target12
In this paper we have studied structural electronic and adsorption properties of PAFs using Density
Functional based Tight-Binding (DFTB) method1314 and Quantized Liquid Density Functional Theory
(QLDFT)15 We have found exceptionally high storage capacities at room temperature what makes
PAFs very good materials for hydrogen storage devices beyond well-known MOFs and COFs We have
compared our results to widely-used classical Grand Canonical Monte-Carlo (GCMC) calculations
reported in the literature We have also studied other properties of these materials such as
133
structural energetic electronic and mechanical We explored the structural variance of diamond
topology by individually placing a selection of organic linkers between carbon nodes This generally
changes surface area mass density and isosteric heat of adsorption what is reflected in the
adsorption isotherms
Methods
Using a conjugate-gradient method we have fully optimized the crystal structures (atomic positions
and unit cell parameters) of PAFs The calculations were performed using Dispersion-corrected Self-
consistent Charge density-functional based tight-binding (DFTB) method as implemented in the
deMonNano code1617 Periodic boundary conditions were applied to a (3x3x3) supercell thus
representing frameworks of the crystalline solid state Electronic density of states (DOS) have been
calculated using the DFTB+ code18 with k-point sampling where the k space was determined by
reaching convergence for the total energy according to the scheme proposed by Monkhorst and
Pack19
Hydrogen adsorption calculations have been carried out using QLDFT within the LIE-0 approximation
thus including many-body interparticle interactions and quantum effects implicitly through the
excess functional The PAF-H2 non-bonded interactions were represented by the classical Morse
atomic-pair potential Force field parameters were taken from Han et al20 who originally developed
them for studying hydrogen adsorption in COFs that incorporate similar building blocks as PAFs The
authors adopted the approach to the shapeless particle approximation of QLDFT These PAF-H2
parameters have been determined by fitting first-principles calculations at the second-order Moslashllerndash
Plesset (MP2) level of theory using the quadruple-ζ QZVPP basis set with correction for the basis set
superposition error (BSSE) QLDFT calculations use a grid spacing of 05 Bohr radii and a potential
cutoff of 5000 K
Results and Discussion
Design and Structure of PAFs
We have designed a set of PAFs by replacing C-C bonds in the diamond lattice by a set of organic
linkers phenyl biphenyl pyrene para-terphenyl perylenetetracarboxylic acid (PTCDA)
diphenylacetylene (DPA) and quaterphenyl (see Figure 1) We label the respective crystal structures
as PAF-phnl (PAF-301 in Ref 12) PAF-biph (PAF-302 in Ref 12) PAF-pyrn PAF-ptph(PAF-303 in Ref
12) PAF-PTCDA PAF-DPA and PAF-qtph (PAF-304 in Ref 12) respectively Such a design of
frameworks should result in materials with high stability due to the parent diamond-topology and
pure covalent bonding of the network The selected linkers differ in their length width and the
134
number of aromatic rings These should play an important role for hydrogen adsorption properties
aromatic systems exhibit large polarizabilities and interact with H2 molecules via London dispersion
forces Long linkers introduce high pore volume and low mas-weight to the network while wide
linkers offer large internal surface area and high heat of adsorption Hence long linkers are of
advantage for high gravimetric capacity while wide linkers enhance volumetric capacity Proper
optimization of the linker size should result in a perfect candidate for hydrogen storage applications
Figure 1 a) Schematic diagram of the topology of PAFs blue spheres and grey pillars represent carbon
tetragonal centers and organic linkers respectively b) Linkers used for the design of PAFs i) phenyl ii)
biphenyl iii) pyrene iv) DPA v) para-Terphenyl vi) PTCDA and vii) quaterphenyl
Selected structural and mechanical properties of the investigated PAF structures are given in Table 1
Frameworks created with the above mentioned linkers have mass densities that range from 085 g
cm-3 (for phenyl) to 01 g cm-3 (for quaterphenyl) The lowest mass-density obtained for a crystal
structure was for COF-108 with = 0171 g cm-39 Our results show that PAF-DPA and PAF-ptph have
mass densities close to the one of COF-108 while PAF-qtph with = 01 g cm-3 exhibits the lowest
for all the PAFs investigated in this study
While the large cell size and the small mass density of PAF-qtph are an advantage for high
gravimetric hydrogen storage capacity other PAFs (eg PAF-pyrn with wide linker) would
compromise gravimetric for high volumetric capacity As both of them are important for practical
applications a balance between them is crucial
Table 1 Calculated cell parameters a (a=b=c α=β=γ=60⁰) mass densities () formation energies (EForm) band
gaps () bulk moduli (B) and H2 accessible free volume and surface area of PAFs studied in the present work
In parenthesis are given HOMU-LUMO gaps of the corresponding saturated linkers
PAFs
a
(Aring)
ρ
(g cm-3)
EForm
(kJ mol-1)
Δ
(eV)
B
(GPa)
H2 accessible
free volume
H2 accessible
surface area
135
() (m2 g-1)
PAF-phnl 97 085 -121 47 (55) 360 35 2398
PAF-biphl 167 032 -122 36 (40) 132 73 5697
PAF-pyrn 166 042 -124 26 (28) 192 66 5090
PAF-DPA 210 019 -122 35 (37) 87 84 7240
PAF-ptph 237 016 -119 32 (33) 56 86 6735
PAF-PTCDA 236 024 -122 18 (19) 95 81 5576
PAF-qtphl 308 010 -119 29 (30) 35 91 7275
Energetic and Mechanical Properties
We have investigated energetic stability of PAFs by calculating their formation energies We regarded
the formation of PAFs as dehydrogenation reaction between saturated linkers and CH4 molecules
For a unit cell containing n carbon nodes and m organic linkers the formation energy (EForm) is given
by
( )
where Ecell EL and
are the total energies of the unit cell saturated linkers CH4 and H2
molecules respectively This excludes the inherent stability of linkers and represents the energy for
coordinating linkers with sp3 carbon atoms during diamond-topology formation All the formation
energies calculated in the present work are given in Table 1 Negative values indicate that the
formation of PAFs is exothermic The values per formula unit do not deviate significantly for different
PAF sizes and shapes
Although diamond is the hardest known material insertion of longer linkers diminishes its
mechanical strength to some extent In order to study the mechanical stability of PAFs we have
calculated their bulk moduli which is the second derivative of the cell energy with respect to the cell
volume (see Table 1) The highest value that we have obtained is for PAF-phnl with B = 36 GPa This is
over ten times smaller than the bulk modulus of pure diamond 514 GPa (calculated at the DFTB
level)21 and 442 GPa (experimental value)22 Bulk modulus should scale as 1V (V - volume) if all
bonds have the same strength We have plotted such a function for PAFs and other framework
136
materials23 in Figure 2 As PAFs have various types of CmdashC bonds there are minor deviations from
the perfect trend
Figure 2 Calculated bulk modulus B as function of the volume per atom PAFs are highlighted in blue and
compared with other framework materials Red curve denotes the fitting to 1V function (V - volume)
The stability of PAF-phnl is similar to that of Cu-BTC MOF and much higher than for other MOFs such
as MOF-5 -177 and -20524 and 3D COFs23 Subsequent addition of phenyl rings decreases B the
lowest value of 35 GPa has been found for PAF-qtph which is in the same order as for COF-108 In
general all the studied PAFs except for PAF-phnl have bulk moduli in line with 3D COFs25 When the
organic linker is extended in the width (cf PAF-biph and PAF-pyrn) the mechanical stability increases
Electronic Properties
All PAF structures are semiconductors similar to COFs The band gaps of PAFs range from 18 to 47
eV (see Table 1) Interesting trends may be observed from the band gaps of this iso-reticular series
In comparison with diamond which has a band gap of 55 eV it may be understood that subsequent
insertion of benzene groups between carbon atoms in diamond decreases the band gap This is easily
understood as the sp3 responsible for the semiconducting character become far apart with large
number of π-electrons in between which tend to close the gap More importantly the values of
band gaps are directly related to the HOMO-LUMO gaps of the corresponding saturated linkers
which are just slightly larger Also more extended linkers (cf PAF-biph and PAF-pyrn or PAF-ptph and
PAF-PTCDA) reduce the band gap
In general conjugated rings reduce the band gap more effectively than CmdashC bond networks (cf PAF-
DPA PAF-ptph or PAF-qtphl) The extreme cases for this behaviour are graphene which is metallic
137
and diamond which is an insulator Overall the band gap may be tuned by placing suitable linkers in
the diamond network Similar results have been reported for MOFs2627
We have calculated the electronic density of states (DOS) for all the PAF structures Figure 3 a shows
carbon DOS of PAFs arranged in such a way that the band gaps decrease going from the top to the
bottom of the figure PAFs with larger band gaps have larger population of energy states at the top of
valence band and bottom of conduction band whereas for linkers with smaller band gaps the
distribution of energy states is rather spread In order to study the impact of sp3 carbon atoms in the
DOS we plotted their contributions against the total carbon DOS in the cases of PAF-phnl and PAF-
pyrn as examples (see Figure 3 b and c) In both cases the sp3 carbons exhibit no energy states in the
band gap and in the close vicinity of band edges
Figure 3 (a) Carbon density of states of all calculated PAFs From top to bottom the value of band gap
decreases (b) and (c) projected DOS on sp3 C atoms of PAF-phnl and PAF-pyrn respectively The vertical
dashed line indicates Fermi level EF
Hydrogen Adsorption Properties
One of the potential applications of PAFs is hydrogen adsorption We have calculated the gravimetric
and volumetric capacities and analyzed them to understand the contributions of the linkers on the
138
hydrogen uptake in different range of temperatures and pressures H2 accessible free volume and
surface area of the PAFs are given in Table 1 In order to assess the excess adsorption capacity the
free pore volume is necessary In our simulation the free pore volume is defined to be that where
the H2-host interaction energy is less than 5000K This covers all sites where H2 is attracted by the
host structure and excludes the repulsion area close to the framework The solvent accessible surface
areas of the PAFs were determined by rolling a probe molecule along the van der Waals spheres of
the atoms A probe molecule of radius 148 Aring was used which is consistent with the Lennard-Jones
sphere of hydrogen and commonly used in various H2 molecular simulations28
Very large surface areas per mass unit have been observed for all PAFs except for PAF-phnl PAF-DPA
and PAF-qtph for example have 7240 and 7274 m2 g-1 H2 accessible surface area respectively For
comparison highly porous MOF-210 and NU-100 give values of 6240 and 6143 m2 g-1 BET surface
areas respectively determined from the experimental adsorption isotherms78 It is worth
mentioning that longer linkers expand the pore and increase the surface area per unit volume and
unit mass Wider linkers provide a higher surface area per unit volume however they possess larger
mass density and hence the surface area per unit mass gets lower
Figure 4 shows the gravimetric and volumetric total and excess adsorption isotherms of PAFs at 77K
The results show that the gravimetric (total and excess) H2 uptake is dependent on the linker length
The longest linker in the PAF-qtph structure gives the total and excess gravimetric capacity of 32 and
128 wt respectively which is larger than for highly porous crystalline 3D COFs29 These numbers
are calculated for the limit pressure of 100 bar For the shortest linker in PAF-phnl we have obtained
only around 25 wt for the same pressure Using grand canonical Monte-Carlo methods (GCMC)
Lan et al12 have predicted total and excess adsorption capacities of PAF-qtph of 224 and 84 wt
respectively The deviations in results are attributed to the differences in both methods where
different force fields are used to describe atom-atom interactions
The volumetric adsorption capacities lower with the size of the linker and the largest uptake we have
found for the PAF-pyrn (46 and 35 kg m-3 total and excess respectively) while very low values were
found for PAF-qtph (30 and 145 kg m-3 total and excess respectively) at 35-bar pressure It could be
predicted that for PAF-phnl the volumetric adsorption reaches the larges values however due to its
very compact crystal structure it reaches saturation at the low-pressure region and does not exceed
30 kg m-3 for the total uptake From the excess adsorption isotherms however PAF-phnl is the best
adsorbing system for low-pressure application as it gets saturated at 11 bar by adsorbing 266 kg m-3
of H2 Generally we have observed that PAFs with lower pore sizes exhibited saturation of volumetric
capacities at lower pressures
139
Figure 4 Calculated H2 uptake of PAFs at 77 K (a)-(b) gravimetric and (c)-(d) volumetric total (upper panel)
and excess (lower panel) respectively
We have also calculated the adsorption performance of PAFs at room temperature The gravimetric
total adsorption isotherms are shown in Figure 5 Similar to the results obtained for T=77 K PAF-
qtph exhibits the highest total gravimetric capacity at room temperature which is as high as 732 wt
at 100 bar Lan et al12 have predicted the uptake of 653 wt at 100 bar using GCMC simulations
These results exceed the 2015 DOE target of 55 wt 3031 On the other hand at more applicable
pressure of 5 bar the uptake reaches only 05 wt This suggests however that the delivery amount
(approximately the difference between the uptakes at 100 and 5 bar) is still more than the DOE
target Phenyl linker at room temperature does not exceed the gravimetric uptake of 1 wt at 100
bar
Figure 5 Calculated gravimetric total (left) and excess (right) H2 uptake of PAFs at 300 K
140
At 300K excess gravimetric capacity is rather low and does not exceed 075 wt even for large
pressure (see Figure 5)
Effects of interpenetration
Very often frameworks with large pores crystallize as interpenetrated lattices In most cases this is
an undesired fact due to reduction of the pore size and free volume For instance COF-300 which
has diamond topology was found to have 5-interpenetrated frameworks32
We have studied the effect of interpenetration in the case of PAF-qtph which has the largest pore
volume among the materials in this study Without any steric hindrance PAF-qtph may be
interpenetrated up to the order of four The two three and four interpenetrated networks are
named as PAF-qtph-2 PAF-qtph-3 and PAF-qtph-4 respectively and in each case the interpenetrated
structures possess translational symmetry Table 2 shows calculated mass densities and H2 accessible
free volume and surface area of non-interpenetrated and n-fold penetrated PAF-qtph Obviously the
mass density increases by one unit for each degree of interpenetration PAF-qtph has 91 of its
volume accessible for H2 which means that 9 of volume is occupied by the skeleton of the PAF
Hence each degree of interpenetration decreases the free volume by 9 H2 accessible surface area
per unit mass remains the same up to 3-fold interpenetration However PAF-qtph-4 results in much
less accessibility for H2
Table 2 Calculated mass densities () and H2 accessible free volume and surface area of non-interpenetrated
and n-fold interpenetrated PAF-qtph where n = 2 3 4
PAF
(g cm-3)
H2 accessible
free volume ()
H2 accessible
surface area
(m2 g-1)
PAF-qtph 010 91 7275
PAF-qtph-2 020 82 7275
PAF-qtph-3 030 73 7275
PAF-qtph-4 040 64 5998
Figure 6 shows the excess and total adsorption isotherms (gravimetric and volumetric) of non-
interpenetrated and interpenetrated PAF-qtph at 77 K With the increasing degree of
141
interpenetration saturation occurs at lower values of pressure This is due to the smaller pore size
resulted from the high-fold interpenetration The excess gravimetric capacity decreases by 18 - 2 wt
per each n-fold interpenetration a result of the decreasing free space around the skeleton It is to be
noted that the difference between the total and the excess adsorption capacities of PAF-qtph is quite
large however it decreases less for interpenetrated structures This is because the interpenetrated
frameworks occupy the free space that otherwise would be taken by hydrogen to increase the total
capacity but not the excess
Figure 6 Calculated H2 uptake at 77 K of non-interpenetrated and n-fold interpenetrated PAF-qtph with n = 2
3 4 (a)-(b) gravimetric and (c)-(d) volumetric total (lower panel) and excess (upper panel) respectively
On the other hand the interpenetrated frameworks have larger volumetric capacity what is easily
understandable due to the volume reduction Significant increase in excess volumetric capacity has
been obtained for PAF-qtph-2 in comparison with PAF-qtph whereas only slightly lower increase was
obtained for PAF-qtph-3 in comparison with PAF-qtph-2 (see Figure 6) PAF-qtph-4 exhibited even
lower increase compared to PAF-qtph-3 suggesting that only a certain degree of interpenetration is
appreciable Although non-interpenetrated PAF has a large amount of H2 gas in the bulk phase due
to the high pore volume its total volumetric uptake does not exceed that of the interpenetrated
PAFs
Almost linear gravimetric isotherms were obtained at room temperature for all studied PAFs
including the interpenetrated cases (see Figure 7) Large decrease of gravimetric capacity was noted
142
when the PAF was interpenetrated from around 7 wt down to 2 wt for 4-fold interpenetrated
PAF-qtph The excess gravimetric capacity however is the same independent of the n-fold
interpenetration and maximum of 06 wt was found for 100-bar pressure (see Figure 7)
Figure 7 Calculated gravimetric total (left) and excess (right) H2 uptake of non-interpenetrated and n-fold
interpenetrated PAF-qtph (n = 2 3 4) at 300 K
Conclusions
Following diamond-topology we have designed a set of porous aromatic frameworks PAFs by
replacing CmdashC bonds by various organic linkers what resulted in remarkably high surface area and
pore volume
Exothermic formation energies of PAFs calculated from the dehydrogenation of the linkers with CH4
indicate strong coordination of linkers in the diamond topology The studied PAFs exhibit bulk moduli
that are much smaller than diamond however in the same order as other porous frameworks such
as MOFs and COFs PAFs are semi-conductors with their band gaps dependent on the HOMO-LUMO
gaps of the linking molecules
Using quantized liquid density functional theory which takes into account inter-particle interactions
and quantum effects we have predicted hydrogen uptake capacities in PAFs At room temperature
and 100 bar the predicted uptake in PAF-qtph was 732 wt which exceeded the 2015 DOE target
At the same conditions other PAFs showed significantly lower uptakes At 77 K the PAFs with similar
pore sizes exhibited similar gravimetric uptake capacities whereas smaller pore size and larger
number of aromatic rings result in higher volumetric capacities In smaller pores saturation of excess
capacities occurred at lower pressures Interpenetrated frameworks largely decrease the amount of
hydrogen gas in the pores and increase the weight of the material however they are predicted to
have high volumetric capacities
143
References
(1) Eddaoudi M Moler D B Li H L Chen B L Reineke T M OKeeffe M Yaghi O M
Accounts of Chemical Research 2001 34 319
(2) Seo J S Whang D Lee H Jun S I Oh J Jeon Y J Kim K Nature 2000 404 982
(3) Ferey G Mellot-Draznieks C Serre C Millange F Accounts of Chemical Research 2005 38
217
(4) Yaghi O M OKeeffe M Ockwig N W Chae H K Eddaoudi M Kim J Nature 2003 423
705
(5) Eddaoudi M Kim J Rosi N Vodak D Wachter J OKeeffe M Yaghi O M Science 2002
295 469
(6) Cote A P Benin A I Ockwig N W OKeeffe M Matzger A J Yaghi O M Science 2005
310 1166
(7) Furukawa H Ko N Go Y B Aratani N Choi S B Choi E Yazaydin A O Snurr R Q
OKeeffe M Kim J Yaghi O M Science 2010 329 424
(8) Farha O K Yazaydin A O Eryazici I Malliakas C D Hauser B G Kanatzidis M G
Nguyen S T Snurr R Q Hupp J T Nature Chemistry 2010 2 944
(9) El-Kaderi H M Hunt J R Mendoza-Cortes J L Cote A P Taylor R E OKeeffe M Yaghi
O M Science 2007 316 268
(10) Ben T Ren H Ma S Cao D Lan J Jing X Wang W Xu J Deng F Simmons J M Qiu
S Zhu G Angewandte Chemie-International Edition 2009 48 9457
(11) Yuan Y Sun F Ren H Jing X Wang W Ma H Zhao H Zhu G Journal of Materials
Chemistry 2011 21 13498
(12) Lan J Cao D Wang W Ben T Zhu G Journal of Physical Chemistry Letters 2010 1 978
(13) Oliveira A F Seifert G Heine T Duarte H A Journal of the Brazilian Chemical Society
2009 20 1193
(14) Seifert G Porezag D Frauenheim T International Journal of Quantum Chemistry 1996 58
185
(15) Patchkovskii S Heine T Physical Review E 2009 80
(16) Heine T Rapacioli M Patchkovskii S Frenzel J Koester A M Calaminici P Escalante S
Duarte H A Flores R Geudtner G Goursot A Reveles J U Vela A Salahub D R deMon-nano edn ed
deMon 2009
(17) Zhechkov L Heine T Patchkovskii S Seifert G Duarte H A Journal of Chemical Theory
and Computation 2005 1 841
(18) BCCMS Bremen DFTB+ - Density Functional based Tight binding (and more)
(19) Monkhorst H J Pack J D Physical Review B 1976 13 5188
(20) Han S S Furukawa H Yaghi O M Goddard III W A Journal of the American Chemical
Society 2008 130 11580
(21) Kuc A Seifert G Physical Review B 2006 74
(22) Cohen M L Physical Review B 1985 32 7988
(23) Lukose B Kuc A Heine T manuscript in preparation 2012
(24) Lukose B Supronowicz B Petkov P S Frenzel J Kuc A Seifert G Vayssilov G N
Heine T physica status solidi (b) 2011
(25) Amirjalayer S Snurr R Q Schmid R Journal of Physical Chemistry C 2012 116 4921
(26) Gascon J Hernandez-Alonso M D Almeida A R van Klink G P M Kapteijn F Mul G
Chemsuschem 2008 1 981
(27) Kuc A Enyashin A Seifert G Journal of Physical Chemistry B 2007 111 8179
(28) Dueren T Millange F Ferey G Walton K S Snurr R Q Journal of Physical Chemistry C
2007 111 15350
(29) Furukawa H Yaghi O M Journal of the American Chemical Society 2009 131 8875
144
(30) US DOE Office of Energy Efficiency and Renewable Energy and The FreedomCAR and
Fuel Partnership 2009
httpwww1eereenergygovhydrogenandfuelcellsstoragepdfstargets_onboard_hydro_storage_explanatio
npdf
(31) US DOE USCAR Shell BP ConocoPhillips Chevron Exxon-Mobil T F a F P Multi-Year
Research Development and Demonstration Plan 2009
httpwww1eereenergygovhydrogenandfuelcellsmypppdfsstoragepdf
(32) Uribe-Romo F J Hunt J R Furukawa H Klock C OKeeffe M Yaghi O M Journal of the
American Chemical Society 2009 131 4570
145
Appendix G
A novel series of isoreticular metal organic frameworks realizing metastable structures by liquid phase epitaxy
Jinxuan Liu Binit Lukose Osama Shekhah Hasan Kemal Arslan Peter Weidler Hartmut
Gliemann Stefan Braumlse Sylvain Grosjean Adelheid Godt Xinliang Feng Klaus Muumlllen Ioan-
Bogdan Magdau Thomas Heine and Christof Woumlll
Prepared for publication
Highly crystalline molecular scaffolds with nm-sized pores offer an attractive basis for the fabrication
of novel materials The scaffolds alone already exhibit attractive properties eg the safe storage of
small molecules like H2 CO2 and CH41-4 In addition to the simple release of stored particles changes
in the optical and electronic properties of these nanomaterials upon loading their porous systems
with guest molecules offer interesting applications eg for sensors5 or electrochemistry6 For the
construction of new nanomaterials the voids within the framework of nanostructures may be loaded
with nm-sized objects such as inorganic clusters larger molecules and even small proteins a
process that holds great potential as for example in drug release7-8 or the design of novel battery
materials9 Furthermore meta-crystals may be prepared by loading metal nanoparticles into the
pores of a three-dimensional scaffold to provide materials with a number of attractive applications
ranging from plasmonics10-12 to fundamental investigations of the electronic structure and transport
properties of the meta-crystals13
146
In the last two decades numerous studies have shown that MOFs also termed porous coordination
polymers (PCPs) fulfill many of the aforementioned criteria Although originally developed for the
storage of hydrogen eg in tanks of automobiles MOFs have recently found more technologically
advanced applications as sensors5 matrices for drug release7-8 and as membranes for enantiomer
separation14 and for proton conductance in fuel cells15 Although pore-sizes available within MOFs1
are already sufficiently large to host metal-nanoclusters such as Au5516-17 the actual realization of
meta-crystals requires in addition to structural requirements a strategy for the controlled loading
of the nanoparticles (NPs) into the molecular framework Simply adding NPs to the solutions before
starting the solvothermal synthesis of MOFs has been reported18 but this procedure does not allow
for controlled loading The layer-by-layer growing process of SUFMOFs allows the uptake of
nanosized objects during synthesis including the fabrication of compositional gradients of different
NPs Those guest NPs can range from pure metal metal oxide or covalent clusters over one-
dimensional objects such as nanowires carbon or inorganic nanotubes to biological molecules such
as drugs or even small proteins If the loading happens during synthesis alternating layers of
different NPs can be realized
The introduction of procedures for epitaxial growth of MOFs on functionalized substrates19 was a
major step towards controlled functionalization of frameworks Epitaxial synthesis allowed the
preparation of compositional gradients20-21 and provided a new strategy to load nano-objects into
predefined pores
Unfortunately the LPE process has so far been only demonstrated for a fairly small number of
MOFs141922-23 none of which exhibits the required pore sizes In addition for the fabrication of meta-
crystals the architecture of the network should be sufficiently adjustable to realize pores of different
sizes There should also be a straightforward way to functionalize the framework itself in order to
tailor the interaction of the walls with the targeted guest objects Ideally such a strategy would be
based on an isoreticular series of MOFs in which the pore size can be determined by a choice from a
homologous series of ligands with different lengths1
Here we present a novel isoreticular MOF series with the required flexibility as regards pore-sizes
and functionalization which is well suited for the LPE process This class denoted as SURMOF-2 is
derived from MOF-2 one of the simplest framework architectures24-25 MOF-2 is based on paddle-
wheel (pw) units formed by attaching 4 dicarboxylate groups to Cu++ or Zn++-dimers yielding planar
sheets with 4-fold symmetry (see Fig 1) These planes are held together by strong
carboxylatemetal- bonds26 Conventional solvothermal synthesis yields stacks of pw-planes shifted
relative to each other with a corresponding reduction in symmetry to yield a P2 or C2 symmetry2427-
28
147
The relative shifts between the pw-planes can be avoided when using the recently developed liquid
phase epitaxy (LPE) method22 As demonstrated by the XRD-data shown in Fig 2 for a number of
different dicarboxylic acid ligands the LPE-process yields surface attached metal organic frameworks
(SURMOFs) with P4 symmetry29 except for the 26-NDC ligand which exhibits a P2 symmetry as a
result of the non-linearity of the carboxyl functional groups on the ligand The ligand PPDC
pentaphenyldicarboxylic acid with a length of 25 nm is to our knowledge the longest ligand which
has so far been used successfully for a MOF synthesis303030 A detailed analysis of the diffraction data
allowed us to propose the structures shown in Fig 1 This novel isoreticular series of MOFs hereafter
termed SURMOF-2 also consists of pw-planes which ndash in contrast to MOF-2 ndash are stacked directly
on-top of each other This absence of a lateral shift when stacking the pw-planes yields 1d-pores of
quadratic cross section with a diagonal up to 4 nm (see Fig 2) Importantly for the SURMOF-2 series
interpenetration is absent For many known isoreticular MOF series the formation of larger and
larger pores is limited by this phenomenon if the pores become too large a second or even a third
3d-lattice will be formed inside the first lattice yielding interpenetrated networks which exhibit the
expected larger lattice constant but with rather small pore sizes31-33 the hosting of NPs thus becomes
impossible For SURMOF-2 the particular topology with 1d-pores of quadratic cross section is not
compatible with the presence of a second interwoven network and as a result interpenetration is
suppressed
Although the solubility of some the long dicarboxylic acids ligands used for the SURMOF-2 fabrication
(TPDC QPDC and PPDC see Fig 1) is fairly small we encountered no problems in the LPE process
since already small concentrations of dicarboxylic acids are sufficient for the formation of a single
monolayer on the substrate the elementary step of the LPE synthesis process34-35 Even with the
longest dicarboxylic acid available to us PPDC growth occurred in a straightforward fashion and
optimization of the growth process was not necessary
The P4 structures proposed for the SURMOF-2 series except for the 26-NDC ligand differ markedly
from the bulk MOF-2 structures obtained from conventional solvothermal synthesis2427-28 To
understand this unexpected difference and in particular the absence of relative shifts between the
pw-planes in the proposed SURMOF-2 structure we performed extensive quantum chemical
calculations employing approximate density-functional theory (DFT) in this case London dispersion-
corrected self-consistent charge density-functional based tight-binding (DFTB)36-38 to a periodic
model of MOF-2 and its SURMOF derivatives
Periodic structure DFTB calculations (Fig S1) confirmed that the P2m structure proposed by Yaghi
et al for the bulk MOF-2 material24 is indeed the energetically favorable configuration for MOF-2
while the C2m structure that has also been reported for bulk MOF-227-28 is slightly higher in energy
148
(41 meV per formula unit see Table 1) The optimum interlayer distance between the pw-planes in
the P4 structure is 58 Aring in very good agreement with the experimental value of 56 Aring (as obtained
from the in-plane XRD-data see Fig S2) It is important to note that at that distance the linkers
cannot remain in rectangular position as in MOF-2 but are twisted in order to avoid steric hindrance
and to optimize linker-linker interactions
The P4mmm structure as proposed for SURMOF-2 is less stable by 200 meV per formula unit as
compared to the P2m configuration Although the crystal energy for the P4 stacking is substantially
smaller than for P2 packing simulated annealing (DFTB model using Born-Oppenheimer Molecular
Dynamics) carried out at 300K for 10 ps demonstrated that the P4mmm structure corresponds to a
local minimum This observation is confirmed by the calculation of the stacking energy of MOF-2
where a shifting of the layers is simulated along the P2-P4-C2 path (Figure 3) On the basis of these
calculations we thus propose that SURMOF-2 adopts this metastable P4 structure
In Table 1 the interlayer interaction energies are summarized They range from 590 meV per formula
unit with respect to fully dissociated MOF-2 layers for Cu2(bdc)2 and successively increase for longer
linkers of the same type reaching 145 eV per formula unit for Cu2(ppdc)2 indicating that the linkers
play an important role in the overall stabilization of SURMOF-2 with large pore sizes Even stronger
interlayer interactions are found for different linker topologies (PPDC) A detailed computational
analysis of the role of the individual building blocks of SURMOF-2 for the crystal growth and
stabilization will be published elsewhere
The flat well-defined organic surfaces used as the templating (or nucleating) substrate in the LPE
growth process provide a satisfying explanation for why SURMOF-2 grows with the highly
symmetrical P4 structure instead of adopting the energetically more favorable bulk P2 symmetry3439
The carboxylic acid groups exposed on the surface force the first layer of deposited Cu-dimers into a
coplanar arrangement which is not compatible with the bulk P2 (or C2) structure but rather
nucleates the P4 packing In turn the carboxylate-groups in the first layer of deposited dicarboxylic
acids provide the same constraints for the next layer (see Fig 1) As a result of the layer-by-layer
method employed for further SURMOF-2 growth the same boundary conditions apply for all
subsequent layers The organic surface thus acts as a seed structure to yield the metastable P4
packing not an unusual motif in epitaxial growth40
The calculations allow us to predict that it will be possible to grow SURMOF structures with even
larger linker molecules as used here The stacking will further stabilize the P4 symmetry as the
interaction energy per formula unit is more and more dominated by the linkers (Table 1) At present
149
we cannot predict the maximum possible size of the pores but linker PPDC (see Fig 2) is thus far
unmatched as a component in non-interpenetrated framework structures
Acknowledgement
We thank Ms Huumllsmann Bielefeld University for the preparation of P(EP)2DC Financial support by
Deutsche Forschungsgemeinschaft (DFG) within the Priority Program Metal-Organic Frameworks
(SPP 1362) is gratefully acknowledged
Methods
Computational Details
All structures were created using a preliminary version of our topological framework creator
software which allows the creation of topological network models in terms of secondary building
units and their replacement by individual molecules to create the coordinates of virtually any
framework material The generated starting coordinates including their corresponding lattice
parameters have been hierarchically fully optimized using the Universal Force Field (UFF)41 followed
by the dispersion-corrected self-consistent-charge density-functional based tight-binding (DFTB)
method36-3842-43 that we have recently validated against experimental structures of HKUST-1 MOF-5
MOF-177 DUT-6 and MIL-53(Al) as well as for their performance to compute the adsorption of
water and carbon monoxide37 For all calculations we employed the deMonNano software44444444
We have chosen periodic boundary conditions for all calculations and the repeated slab method has
been employed to compute the properties of the single layers in order to evaluate the stacking
energy A conjugate-gradient scheme was employed for geometry optimization of atomic
coordinates and cell dimensions The atomic force tolerance was set to 3 x 10-4 eVAring
The confined conditions at the SURMOF-substrate interface was modeled by fixing the corresponding
coordinate in the computer simulations All calculated structures have been substantiated by
simulated annealing calculations where a Born-Oppenheimer Molecular Dynamics trajectory at 300K
has been computed for 10 ps without geometry constrains All structures remained in P4 topology
Experimental methods
The SURMOF-2 samples were grown on Au substrates (100-nm Au5-nm Ti deposited on Si wafers)
using a high-throughput approach spray method45 The gold substrates were functionalized by self-
assembled monolayers SAMs of 16-mercaptohexadecanoic acid (MHDA) These substrates were
mounted on a sample holder and subsequently sprayed with 1 mM of Cu2 (CH3COO)4middotH2O ethanol
solution and ethanolic solution of SURMOF-2 organic linkers (01 mM of BDC 26-NDC BPDC and
150
saturated concentration of TPDC QPDC PPDC P(EP)2DC) at room temperature After a given
number of cycles the samples were characterized with X-ray diffraction (XRD)
Figure 1 Schematic representation of the synthesis and formation of the SURMOF-2 analogues
151
Figure 2 Out of plane XRD data of Cu-BDC Cu-26-NDC Cu-BPDC Cu-TPDC Cu-QPDC Cu-P(EP)2DC and Cu-PPDC (upper left) schematic representation (upper right) and proposed structures of SURMOF-2 analogues (lower panel) All the SURMOF-2 are grown on ndashCOOH terminated SAM surface using the LPE method
152
Figure 3 Energy for the relative shift of each layer in two different directions horizontal (P4 to P2) and diagonal (P4 to C2) at a fixed interlayer distance of 56 Aring Structures have been partially optimized and energies are given per formula unit with respect to fully dissociated planes
Supporting information
Synthesis of organic linkers
(1) para-terphenyldicarboxylic acid (TPDC)
To a solution of para-terphenyl (1500 g 651 mmol 1 eq) and oxalyl chloride (3350 mL 3900 mmol
6 eq) in carbon disulfide (30 mL) at 0degC (ice bath) was added aluminium chloride (1475 g 1106
mmol 17 eq) After 1h stirring additional aluminium chloride (0870 g 651 mmol 1 eq) was added
the ice bath was removed and the reaction mixture was stirred at room temperature for 24h The
mixture was poured into crushed ice and stirred until the dark-brown mixture turned to yellow
Carbon disulfide was removed under reduced pressure and the resulting aqueous suspension was
filtered The solid was washed with diluted hydrochloric acid then with diethyl ether and dried
under vacuum overnight with an oil bath at 50degC Pale yellow solid yield 2028 g (98)
(2) para-quaterphenyldicarboxylic acid (QPDC)
153
To a solution of para-quaterphenyl (1000 g 326 mmol 1 eq) and oxalyl chloride (1680 mL 1956
mmol 6 eq) in carbon disulfide (20 mL) at 0degC (ice bath) was added aluminium chloride (0740 g 555
mmol 17 eq) After 1h stirring additional aluminium chloride (0435 g 326 mmol 1 eq) was added
the ice bath was removed and the reaction mixture was stirred at room temperature for 24h The
mixture was poured into crushed ice and stirred until the dark-brown mixture turned to yellow
Carbon disulfide was removed under reduced pressure and the resulting aqueous suspension was
filtered The solid was washed with diluted hydrochloric acid then with diethyl ether and dried
under vacuum overnight with an oil bath at 50degC Yellow solid yield 1186 g (92)
(3) P(EP)2DC
The synthesis of the P(EP)2DC-linker has been described in Ref 46
(4) para-pentaphenly dicarboxylic acid (PPDC)
Scheme 1 Synthesis of para-pentaphenly dicarboxylic acid i) Pd(PPh3)4 Na2CO3 toluenedioxane H2O 85oC ii) MeOH 6M NaOH HCl
para-pentaphenly dicarboxylic acid (H2L) was synthesized via Suzuki coupling of 44rdquo-dibromo-p-
terphenyl and 4-methoxycarbonylphenylboronic acid (Scheme 1) 44rdquo-dibromo-p-terphenyl (776 mg
200 mmol) 4-methoxycarbonylphenylboronic acid (108 g 60 mmol) and Na3CO3 (170 g 160 mmol)
were added to a degassed mixture of toluene14-dioxane (50 mL 221) under Ar [Pd(PPh3)4] (116
mg 010 mmol) was added to the mixture and heated to 85 degC for 24 hours under Ar The reaction
mixture was cooled to room temperature The precipitate was collected by filtration washed with
water methanol and used for next reaction without further purification The final product H4L was
obtained by hydrolysis of the crude product under reflux in methanol (100 mL) overnight with 6M
aqueous NaOH (20 mL) followed by acidification with HCl (conc) After removal of methanol the
final product was collected by filtration as a grey solid (078 g 83 yield) 1H NMR (d6-DMSO
250MHz 298K) δ (ppm) 805 (d J = 75Hz 4H) 789-783 (m 12H) 750 (d J = 75Hz 4H) 13C NMR
cannot be clearly resolved due to poor solubility MALDI-TOF-MS (TCNQ as matrix) mz 47002
cacld 47051 Elemental analysis Calculated C 8169 H 471 Found C 8163 H 479
Br Br MeOOC B
OH
OH
+
COOMe
COOMe
COOH
COOH
i ii
154
Figure S1 MOF-2 in P4 (as reported) P2 and C2 symmetry
155
Figure S2 In-plane XRD data of Cu-BDC Cu-26-NDC Cu-BPDC Cu-TPDC Cu-QPDC and Cu-P(EP)2DC All the
SURMOF-2 are grown on ndashCOOH terminated SAM surface using the LPE method The position of (010) plane
represents the layer distance
Table 1 Stacking energy and geometries of P4 SURMOF-2 derivatives
Symmetry a= c b Stacking Energy
Cu2(bdc)2 C2 1119 50 -076
Cu2(bdc)2 P2 1119 54 -08
Cu2(bdc)2 P4 1119 58 -059
156
Cu2(ndc)2 P2 1335 56 -04
Cu2(bpdc)2 P4 1549 59 -068
Cu2(tpdc)2 P4 1984 59 -091
Cu2(qpdc)2 P4 2424 59 -121
Cu2(P(EP)2DC)2 P4 2512 52 -173
Cu2(ppdc)2 P4 2859 59 -145
Stacking energies (DFTB level in eV) of SURMOFs The energies were calculated within periodic
boundary conditions and are given per formula unit
References
1 Eddaoudi M et al Systematic design of pore size and functionality in isoreticular MOFs and
their application in methane storage Science 295 469-472 (2002)
2 Rosi N L et al Hydrogen storage in microporous metal-organic frameworks Science 300
1127-1129 (2003)
3 Rowsell J L C amp Yaghi O M Metal-organic frameworks a new class of porous materials
Microporous and Mesoporous Materials 73 3-14 (2004)
4 Wang B Cote A P Furukawa H OKeeffe M amp Yaghi O M Colossal cages in zeolitic
imidazolate frameworks as selective carbon dioxide reservoirs Nature 453 207-U206 (2008)
5 Lauren E Kreno et al Metal-Organic Framework Materials as Chemical Sensors Chemical
Reviews 112 1105-1124 (2012)
6 Dragasser A et al Redox mediation enabled by immobilised centres in the pores of a metal-
organic framework grown by liquid phase epitaxy Chemical Communications 48 663-665
(2012)
7 Horcajada P et al Metal-organic frameworks as efficient materials for drug delivery
Angewandte Chemie-International Edition 45 5974-5978 (2006)
8 Horcajada P et al Flexible porous metal-organic frameworks for a controlled drug delivery
Journal of the American Chemical Society 130 6774-6780 (2008)
9 Hurd J A et al Anhydrous proton conduction at 150 degrees C in a crystalline metal-organic
framework Nature Chemistry 1 705-710 (2009)
10 Kreno L E Hupp J T amp Van Duyne R P Metal-Organic Framework Thin Film for Enhanced
Localized Surface Plasmon Resonance Gas Sensing Analytical Chemistry 82 8042-8046
(2010)
11 Lu G et al Fabrication of Metal-Organic Framework-Containing Silica-Colloidal Crystals for
Vapor Sensing Advanced Materials 23 4449-4452 (2011)
157
12 Lu G amp Hupp J T Metal-Organic Frameworks as Sensors A ZIF-8 Based Fabry-Perot Device
as a Selective Sensor for Chemical Vapors and Gases Journal of the American Chemical
Society 132 7832-7833 (2010)
13 Mark D Allendorf Adam Schwartzberg Vitalie Stavila amp Talin A A Roadmap to
Implementing MetalndashOrganic Frameworks in Electronic Devices Challenges and Critical
Directions European Journal of Chemistry (2011)
14 Bo Liu et al Enantiopure Metal-Organic Framework Thin Films Oriented SURMOF Growth
and Enantioselective Adsorption Angewandte Chemie-International Edition 51 817-810
(2012)
15 Sadakiyo M Yamada T amp Kitagawa H Rational Designs for Highly Proton-Conductive
Metal-Organic Frameworks Journal of the American Chemical Society 131 9906-9907 (2009)
16 Ishida T Nagaoka M Akita T amp Haruta M Deposition of Gold Clusters on Porous
Coordination Polymers by Solid Grinding and Their Catalytic Activity in Aerobic Oxidation of
Alcohols Chemistry-a European Journal 14 8456-8460 (2008)
17 Esken D Turner S Lebedev O I Van Tendeloo G amp Fischer R A AuZIFs Stabilization
and Encapsulation of Cavity-Size Matching Gold Clusters inside Functionalized Zeolite
Imidazolate Frameworks ZIFs Chemistry of Materials 22 6393-6401 (2010)
18 Lohe M R et al Heating and separation using nanomagnet-functionalized metal-organic
frameworks Chemical Communications 47 3075-3077 (2011)
19 Shekhah O et al Step-by-step route for the synthesis of metal-organic frameworks Journal
of the American Chemical Society 129 15118-15119 (2007)
20 Furukawa S et al A block PCP crystal anisotropic hybridization of porous coordination
polymers by face-selective epitaxial growth Chemical Communications 5097-5099 (2009)
21 Shekhah O et al MOF-on-MOF heteroepitaxy perfectly oriented [Zn(2)(ndc)(2)(dabco)](n)
grown on [Cu(2)(ndc)(2)(dabco)](n) thin films Dalton Transactions 40 4954-4958 (2011)
22 Shekhah O et al Controlling interpenetration in metal-organic frameworks by liquid-phase
epitaxy Nature Materials 8 481-484 (2009)
23 Zacher D et al Liquid-Phase Epitaxy of Multicomponent Layer-Based Porous Coordination
Polymer Thin Films of [M(L)(P)05] Type Importance of Deposition Sequence on the Oriented
Growth Chemistry-a European Journal 17 1448-1455 (2011)
24 Li H Eddaoudi M Groy T L amp Yaghi O M Establishing microporosity in open metal-
organic frameworks Gas sorption isotherms for Zn(BDC) (BDC = 14-benzenedicarboxylate)
Journal of the American Chemical Society 120 8571-8572 (1998)
25 Mueller U et al Metal-organic frameworks - prospective industrial applications Journal of
Materials Chemistry 16 626-636 (2006)
158
26 Shekhah O Wang H Zacher D Fischer R A amp Woumlll C Growth Mechanism of Metal-
Organic Frameworks Insights into the Nucleation by Employing a Step-by-Step Route
Angewandte Chemie-International Edition 48 5038-5041 (2009)
27 Carson C G et al Synthesis and Structure Characterization of Copper Terephthalate Metal-
Organic Frameworks European Journal of Inorganic Chemistry 2338-2343 (2009)
28 Clausen H F Poulsen R D Bond A D Chevallier M A S amp Iversen B B Solvothermal
synthesis of new metal organic framework structures in the zinc-terephthalic acid-dimethyl
formamide system Journal of Solid State Chemistry 178 3342-3351 (2005)
29 Arslan H K et al Intercalation in Layered Metal-Organic Frameworks Reversible Inclusion of
an Extended pi-System Journal of the American Chemical Society 133 8158-8161 (2011)
30 The MOF with the largest pore size recorded so far MOF-200 (Furukawa H et al Ultrahigh
Porosity in Metal-Organic Frameworks Science 329 424-428 (2010)) used a (trivalent)
444-(benzene-135-triyl-tris(benzene-41-diyl))tribenzoate (BBC) ligand The carboxylic
acid-to carboxylic acid distance is 20 nm compared to 25 nm in case of PPDC The cage size
in MOF-200 amounts to 18 nm by 28 nm clearly smaller than the 1d-channels in the PPDC
SURMOF-2 that are 28 nm by 28 nm
31 Batten S R amp Robson R Interpenetrating nets Ordered periodic entanglement
Angewandte Chemie-International Edition 37 1460-1494 (1998)
32 Snurr R Q Hupp J T amp Nguyen S T Prospects for nanoporous metal-organic materials in
advanced separations processes Aiche Journal 50 1090-1095 (2004)
33 Yaghi O M A tale of two entanglements Nature Materials 6 92-93 (2007)
34 Shekhah O Liu J Fischer R A amp Woumlll C MOF thin films existing and future applications
Chemical Society Reviews 40 1081-1106 (2011)
35 Zacher D Shekhah O Woumlll C amp Fischer R A Thin films of metal-organic frameworks
Chemical Society Reviews 38 1418-1429 (2009)
36 Elstner M et al Self-consistent-charge density-functional tight-binding method for
simulations of complex materials properties Physical Review B 58 7260-7268 (1998)
37 Lukose B et al Structural properties of metal-organic frameworks within the density-
functional based tight-binding method Physica Status Solidi B-Basic Solid State Physics 249
335-342 (2012)
38 Zhechkov L Heine T Patchkovskii S Seifert G amp Duarte H A An efficient a Posteriori
treatment for dispersion interaction in density-functional-based tight binding Journal of
Chemical Theory and Computation 1 841-847 (2005)
159
39 Zacher D Schmid R Woumlll C amp Fischer R A Surface Chemistry of Metal-Organic
Frameworks at the Liquid-Solid Interface Angewandte Chemie-International Edition 50 176-
199 (2011)
40 Prinz G A Stabilization of bcc Co via Epitaxial Growth on GaAs Physical Review Letters 54
1051-1054 (1985)
41 Rappe A K Casewit C J Colwell K S Goddard W A amp Skiff W M UFF a full periodic
table force field for molecular mechanics and molecular dynamics simulations Journal of the
American Chemical Society 114 10024-10035 (1992)
42 Seifert G Porezag D amp Frauenheim T Calculations of molecules clusters and solids with a
simplified LCAO-DFT-LDA scheme International Journal of Quantum Chemistry 58 185-192
(1996)
43 Oliveira A F Seifert G Heine T amp Duarte H A Density-Functional Based Tight-Binding an
Approximate DFT Method Journal of the Brazilian Chemical Society 20 1193-1205 (2009)
44 deMonNano v 2009 (Bremen 2009)
45 Arslan H K et al High-Throughput Fabrication of Uniform and Homogenous MOF Coatings
Adv Funct Mater 21 4228-4231 (2011)
46 Schaate A et al Porous Interpenetrated Zirconium-Organic Frameworks (PIZOFs) A
Chemically Versatile Family of Metal-Organic Frameworks Chemistry-a European Journal 17
9320-9325 (2011)
160
Appendix H
Linker guided metastability in templated Metal-Organic Framework-2 derivatives (SURMOFs-2)
Binit Lukose Gabriel Merino Christof Woumlll and Thomas Heine
Prepared for publication
INTRODUCTION
The molecular assembly of metal-oxides and organic struts can provide a large number of network
topologies for Metal-Organic Frameworks (MOFs)1-3 This is attained by the differences in
connectivity and relative orientation of the assembling units Within each topology replacement of a
building unit by another of same connectivity but different size leads to what is known as isoreticular
alteration of pore size The structure of MOFs in principle can be formed into the requirement of
prominent applications such as gas storage4-8 sensing910 and catalysis11-13 The structural
divergence and the performance can be further increased by functionalizing the organic linkers1415
In MOFs linkers are essential in determining the topology as well as providing porosity A linker
typically contains single or multiple aromatic rings the orientation of which normally undergoes
lowest repulsion from the coordinated metal-oxide clusters providing the most stable symmetry for
the bulk material We encounter for the first time a situation that the orientation of the linker
provides metastability in an isoreticular series of MOFs Templated MOF-2 derivatives (surface-MOF-
2 or simply SURMOFs-2)16 show this extraordinary phenomenon While bulk MOF-2 is determined to
be stable in P217 or C218-20 symmetry we found the existence of SURMOFs-2 in P4 symmetry
161
(Figure 1)16 Our theoretical approach to this problem identifies the role of the linkers in providing
P4 geometry the status of a local energy-minimum
MOF-2 derivatives are two-dimensional lattice structures composed of dimetal-centered four-fold
coordinated paddlewheels and linear organic linkers Solvothermally synthesized archetypal MOF-2
had transition metal (Zn or Cu) centers in the connectors and benzenedicarboxylic acid linkers17 The
derivatives under our investigation contain paddlewheels with Cu dimers and benzenedicarboxylic
acid (BDC) naphthalenedicarboxylic acid (NDC) biphenyldicarboxylic acid (BPDC)
triphenyldicarboxylic acid (TPDC) quaterphenyldicarboxylic acid (QPDC) P(EP)2DC21 and
pentaphenyldicarboxylic acid (PPDC) linkers pertaining to isoreticular expansion of pore size16 The
four-fold connectivity of the Cu dimers together with linearity of the linkers gives rise to layers with
quadratic (square) topology The interlayer separation d is typically much more than that of
graphite or 2D COFS22-24 because all the atoms in one layer do not lie in one plane
In bulk form the nearest layers are shifted to each other either towards one of the four linkers
(P2)171920 or diagonally towards one of the four surrounding pores (C2)18-20 in effort to reduce
the repulsion of alike charges in the adjacent paddlewheels when they are in eclipsed form (P4)
(Figure 1) The metal-dimers often show high reactivity which results in attracting water or
appropriate solvents in their axial positions The stacking along the third axis is typically through
interlayer interactions and through hydrogen bonds established between the solvents or between
the solvents and oxygen atoms in the paddlewheel The lateral shift of layers is inevitable without
additional supports Coordination of pillar molecules such as 14-diazabicyclo[222]octane (dabco) or
bipyridine in the axis of metal dimers between the adjacent layers25 is a practical mean to avoid
layer-offset however with the change of MOF dimensionality
Figure 1 Monolayer of MOF-2 and three kinds of layer packing -P4 P2 and C2- of periodic MOF-2
162
Alternate to solvothermal synthesis a step-by-step route may be enabled for the synthesis of
MOFs26-28 It adopts liquid phase epitaxy (LPE) of MOF reactants on substrate-assisted self-assembled
monolayers This is achieved by alternate immersion of the template in metal and ligand precursors
for several cycles This allows controlled growth of MOFs on surfaces The latest outcomes of this
method are the surface-standing MOF-2 derivatives (SURMOF-2s) This isoreticular SURMOF series
has linkers of different lengths (as given above) The cell dimensions that correspond to the length of
the pore walls range from 1 to 28 nm The diagonal pore-width of one of the MOFs (PPDC) accounts
to 4 nm
After evacuation of solvents (water in the present case) X-ray diffraction patterns were taken in
directions both perpendicular (scans out of plane) and parallel (scans in-plane) to the substrate
surface The presence of only one peak -(010)- in the perpendicular scan implies that the layers
orient perpendicular to the surface and the corresponding angle gives the cell parameter a (or c) In
the latter case the peaks - (100) and (001) ndash are identical and this rules out the possibility of layer-
offset Hence the symmetry of the MOFs was determined to be P4 These peaks give rise to the cell
parameter c which is nothing but the interlayer distance d This value (56 Aring) is relatively small for
P4 geometry to have water molecules in the axial position of paddlewheels Presence of some water
molecules observed using Infrared Reflection Absorption Spectroscopy (IRRAS) might be located near
paddlewheels but exposed to the pore The new observations with the SURMOFs are very intriguing
in the context that P4 symmetry appears to be impossible for solvothermally produced MOF-2
We have primarily reported that the P4 geometry of SURMOF-2 exists as a local energy-minimum16
The verification was made using an approximate method of density functional theory (DFT) which is
London dispersion-corrected self-consistent charge density-functional based tight-binding (DFTB) In
the bulk form absolute energies of P2 and C2 are 210 and 170 meV respectively higher than P4 per
a formula unit which is equivalent to the unit cell For SURMOF the barrier for leaving P4 was nearly
50 meV per formula unit It requires further analysis to unravel the reasons for this unusual
metastability We therefore performed an extensive set of quantum chemical calculations on the
composition of the constituent building units The procedure involves defining SURMOF geometry
and analyzing the translations of individual layers
The major individual contributions to the total energy are the interaction between the paddlewheel
units (favouring P2) London dispersion interaction between tilted linkers (favouring P4) and energy
to tilt the linkers In the iso-reticular series of SURMOF-2 the differences arise only in the
163
contributions from the linkers Hence we performed an extensive study only on the smallest of all
linkers- BDC A scaling might be appropriate for other linkers
RESULTS AND DISCUSSION
In solvothermal synthesis of layered MOFs it is assumed that the nucleation process is associated
with the interaction between two connectors This is rationalized by the fact that two paddlewheels
show the strongest possible noncovalent interaction between the individual MOF building blocks
present in MOF-2 As the orientation of the paddlewheels is restricted by the linkers we studied the
stacking of adjacent layers by determining the attraction of two adjacent interacting paddlewheels
upon their respective offsets Thus we investigated the geometries corresponding to lateral
displacements between the known stable arrangements of the bulk structure ie P4-to-P2 and P4-
to-C2 (Figure 2) In the model calculation the two paddlewheels were shifted from D4h symmetry to
two directions that respectively correspond to P2 and C2 symmetries in the bulk The minima along
the two shifts are referred as min-I and min-II and are having C2h symmetry It is interesting to note
that the interaction is in all cases attractive If only the paddlewheels are studied the D4h
configuration where both axes are concentric can be interpreted as transition state between the
two minima configurations P2 or C2 Comparing the two minima min-I corresponding to MOF-2 in
P2 symmetry indicates the formation of two CuO bonds while for min-II the reactive Cu centers do
not participate in the interlayer bonding
Formation of the diagonally shifted C2 bulk geometry for Zn and Cu based MOF-2 as reported in the
literature18-20 possibly is due to the presence of large solvent molecules such as DMF that
coordinate to the free Cu centers the paddlewheels
Figure 2 Displacements of two paddlewheels from D4h symmetry (corresponding to P4) towards minima that correspond to P2 and C2 bulk symmetries
164
To gain further insight on type of interactions for the three paddlewheel arrangements as found in
the bulk (Figure 3) we performed the topological analysis of the electron density for each
structure2930 An inspection of the paddlewheel shows that the electron density of the Cu-O bond has
a (relatively small) value of 0042 au thus showing only weak character In the forms related to P4
and C2 bulk symmetries all (3-1) critical points between the two paddlewheels show very small
density values (0004 au and less) In the P2 structure it is apparent the formation of a four-
membered Cu-O-Cu-O ring across the paddlewheel Note that the Cu-O interactions inside the
paddlewheel weaken (from 0042 to 0031 au) and two intermolecular Cu-O interactions with a
density value of 0024 au stabilize the P2 orientation These links if formed during nucleation will
be regularly repeated during the MOF-2 formation in solvothermal synthesis and due to its strong
binding of -08 eV they are the main reason for the stabilization of the P2 phase If the paddlewheels
are packed in P4 symmetry there must be additional means of stabilization present and that may
only arise from the linkers Moreover one should note that in order to reach the P4 symmetry a
layer-by-layer growth process is required as otherwise the linkers will readily stack as in the P2 bulk
form
165
Figure 3 Topological analysis of the electron density of fully optimized P4 (top left) C2 (top right) and P4 (bottom) structures Atom (grey) (3-1) (red) and (3+1) (green) critical points along with their charge densities are shown
The optimized geometry of a single-layer MOF-2 is shown in Figure 1 Note that the phenyl rings of
the linkers are oriented perpendicular to the MOF plane Contrary to graphene or 2D COFs the more
complex structure of MOF-2 layers may become subject to change upon the interlayer interactions
This is in particular important for the P4 stacking as reported for SURMOF-2 The interaction energy
of two linkers and two benzene rings as oriented in the monolayer has been computed as function
of the interlayer distance (Figure 4) At the experimental interlayer distance of 56 Aring the linkers are
so close that they repel each other strongly and stacking the monolayer structure at the
experimental interlayer distance would introduce an energy penalty of 08 eV per linker
It would not be exotic if we assume that the anchoring of layers on the substrate plays an important
role in setting d A supporting argument is the fact that all the SURMOFs in the iso-reticular series
have the same d An additional point is that the comparatively wider linkers NDC and LM do not
create any difference in the interlayer distance
166
Figure 4 Energy as a function of interlayer distance [d] in case of two paddlewheels and two benzene molecules The energies are given with respect to the complete dissociation of the building blocks
The best adaptation for the linkers in a closer alignment to avoid the repulsion is to partially rotate
the phenyl rings This reduces the Pauli repulsion and at the same time increases the attractive
London dispersion between the linkers However the rotation is energetically penalized by 06 eV as
accordance with similar calculations found in the literature31 and is with the same order of Zn4O-
tetrahedron clusters of the IRMOFs3233
Figure 5 Energy as a function of rotation of linker between two saturated paddlewheels The dihedral angle O-C1-C2-C3 θ between the linker and the carboxylic part in the paddlewheel Values are with respect to the ground state θ = 0⁰
To model the conditions in SURMOF-2 we need to combine the intermolecular interaction of the
linkers with the barrier associated to the rotation of the linker between two paddlewheel units as
given in Figure 5 We may consider a system of three linkers placed as if they are in three adjacent
layers of bulk P4 SURMOF-2 The linkers are undergoing a rotation along their C2 axis that would be
aligned to the Cu-Cu axes in the paddlewheels (see system-I in Figure 6) The dissociation energy of
167
the system includes four times the repulsion from one adjacent linker If we neglect the interaction
between the end-linkers which is only -35 meV the system represents the situation in bulk P4 MOF-
2 The gain of intermolecular energy due to twisting the stacked linkers is partially compensated by
the energy penalty arising from rotation of the linker between the paddlewheels and the resulting
energy shows a minimum at 22deg (Figure 6)
Figure 6 Model of the linker orientation in SURMOF-2 The stabilization of a linker inbetween two layers is approximated in System-I the arrows indicating the rotation while the energy penalty due to breaking the p conjugation of Figure 5 is given in System-II The resulting energy is the black line showing a minimum at 22deg The energy in System-I is with respect to its dissociation limit
Each linker may undergo rotation either clockwise or anti-clockwise with equal preferences in the
local environment However there may be a global control over the preference of each linker The
most stable structure can be figured out from the total energies of each possible arrangement Since
there are only two choices for each linker it may orient either in same fashion or alternate fashion
along X and Y directions If we expect a regular pattern the total number of possibilities are only
three as shown in Figure 8 where rotation of linkers (violet lines) are marked with lsquo+rsquo signs in one of
its two sides which means that facing (outer)-edges of linkers are tilted towards these sides The
three orderings may be verbalized as follows
(i) projection of the facing edges of oppositely placed linkers are either within the square or outside
(P4nbm) (ii) projection of the facing edge of only one of the oppositely placed linkers is within the
square (P4mmm) (iii) projection of the facing edges of all four linkers are either within the square
or outside (P4nmm)
The adjacent layers follow the same linker-orientations The unit cells of (i) and (iii) are four times
bigger than (ii) The stacking energies calculated at the DFTB level suggest P4nbm as the most stable
168
geometry The energies are respectively -048 -032 and -029 eV per formula unit for P4nbm
P4mmm and P4nmm respectively Within P4nbm symmetry the linker edges undergo lowest
repulsion compared to others It is to be noted that only P4nbm has four-fold rotational symmetry
along Z-axis about the Cu-dimer in any paddlewheel
Figure 7 Three possible arrangements of tilted linkers in the periodic SURMOF-2 Paddlewheels are shown as red diamonds and the linkers as violet lines Facing (outer) edges of the linkers are rotated towards the side with + signs The symmetries and calculated stacking energies (per formula unit) of the arrangements are also given
To quantify the different stacking energies we performed periodic DFT calculations on the structure
of Cu2(bdc)2 MOF-2 As the most stable P4nbm geometry demands high computational resources in
each calculation we used P4mmm geometry which has four times less atoms in unit cell We
explored tight-packing with rotated linkers and loose-packing with un-rotated linkers Two energy-
minima have been found as shown in Figure 8a at 58 and 65 Aring corresponding to rotated and un-
rotated states of linkers respectively The latter is 40 meV more stable than the former which
means that SURMOF-2 exists in a metastable state along the perpendicular axis The relative shift of
adjacent layers in the direction of the lattice vector (or ) represents the transition from P2 to P4
and back to P2 symmetry We shifted the adjacent layers parallel to and calculated the relative
energies Since the shift of adjacent layers in P4mmm geometry is not symmetric for positive and
negative directions of averages of the energies of the shift in both directions are plotted (see
Figure 9b) P4 is a local minimum while P2 is the global minimum and the energy barrier separating
the two minima accounts to 23 meV per formula unit Hence the P4 geometry of SURMOF-2 can be
taken as metastable state of MOF-2
169
Figure 8 a) Energy as function of interlayer separation d and b) Energy as a function of relative shift of adjacent layers for periodic MOF-2 Energies are given in eV per formula unit
The role of linkers in creating a local minimum at P4 is now apparent However when undergoing the
transition from P4 to P2 the relative motions of the horizontal and vertical linkers are different from
each other Hence a qualitative study is essential to accurately determine the role of each building
block for holding SURMOFs-2 in a metastable state We thus resolved the shift of entire adjacent
layers with respect to each other into relative motions of individual building blocks The experimental
interlayer distance (56 Aring) has been fixed and the energy-profiles have been calculated using DFT
The scans include the shift of
i) a paddlewheel over other
ii) a horizontal linker over other
iii) a vertical linker over other
iv) a paddlewheel over a horizontal linker
v) a paddlewheel over a vertical linker
Here vertical and horizontal linkers are the linkers vertical and horizontal to the shift directions
respectively A complete picture of individual shifts of the MOF-2 SBU including their energy-profiles
is given in Figure 9 The shift of the vertical over the horizontal linker was found insignificant and was
omitted A note of warning is that the tilted vertical linker meets different neighborhoods when
shifted to the left and right However an average energy of these two shifts seems sensible because
the nearest vertical linker in the same layer of P4nbm geometry has opposite orientation This
averaging also makes sense in a case that alternate layers undergo shifting to the same direction
leading to a zigzag (or serrated) packing of layers in the bulk As is readily seen in Figure 9 the
formation of the Cu-O-Cu-O rings between the paddlewheels (see Figure 3) is the key reasons for the
layer shift in P2 MOF-2 A support is obtained from the shifting of the paddlewheel over the
170
horizontal linker (Shift IV) A major objection is made by the vertical linkers (Shift III) The total
interaction becomes repulsive when a vertical linker shifts with respect to the other for about 05 Aring
This may alter the tilt of the linker however a minimum is already established The vertical linkers of
a layer in a way are trapped between those of adjacent layers The shift to either side from P4 most
probably decreases the interlayer separation However this demands further rotation of the vertical
linkers in order to reduce their mutual Pauli repulsion Overall the sharp minimum at P4 may be
taken as an inheritance from the lower interlayer distance of P4 geometry fixed by the anchoring on
the substrate
Figure 9 Shift of individual building blocks in the adjacent layers correspond to the situation in bulk The energies of each shifted arrangement and their total energy are given in the graph
The total energy involved in the shifting of two building blocks (one building block over the other) is
equivalent to the energy per one building block when it feels shift from two neighbors Only the
vertical linker is sensitive to the shift-direction of the two neighbors However since averages were
taken as discussed earlier the total energy becomes independent of the direction Besides the
relative motion of the SBU facing each other in P4 symmetry (Shifts I II and III) for strong distortions
we also have to consider the interaction of adjacent linker-connector interactions as represented in
Shifts IV and V The former stabilizes P2 while the latter support P4 Overall the total energy for all
the shifts shows a local minimum at P4 (see figure 9) in agreement with the periodic calculation
shown in Figure 8 Indeed the relative energy between P2 and P4 stackings estimated by the
171
superposition of individual contributions is 43 meV in close agreement to the 40 meV obtained by
the periodic calculations
Our finite-component model successfully provides adequate information on the individual
contributions of building blocks Vertical linkers play crucial role in holding the SURMOF with P4
symmetry which was never achieved with solvothermally synthesized MOF-2 Paddlewheels are
held most responsible for lateral shift The vertical linkers winningly establish a local minimum at P4
for the SURMOF
Recently we have reported SURMOF-2 derivatives with very large pore sizes(Ref) This has been
achieved by increasing the length of the linker units In view of our analysis of the stacking and
stability of MOF-2 we can now safely state that it will be possible to generate SURMOF-2 derivatives
with even larger pores with pore sizes essentially limited by the availability of stiff long organic
linker molecules This is exemplified by a calculation of the stacking energy of a series of phenyl
oligomers that is SURMOF-2 derivatives incorporating chains with 1 to 10 benzene rings in the
linkers The stacking energies per formula unit are -041 -067 -091 -121 -145 -168 -192 -215
-237 and -26 eV respectively Each benzene ring adds more than 02 eV into the stacking energy per
formula unit This energy is due to the London dispersion interaction between the linkers in the
neighboring layers
The drop of SURMOFs into the local minimum perhaps is blended with some parameters related to
synthetic environments This was beyond the scope of this work however we suggest that studies of
the mechanism of substrate-assisted layer-by-layer deposition of metal and ligand precursors may
give some primary insights into it
CONCLUSION
We have analyzed the reason for the different stackings observed for MOF-2 In the traditional
solvothermal synthesis MOF-2 is likely to crystallize in P2 symmetry determined by the strong
interaction between the paddlewheel units The coordination of large solvent molecules to the free
metal centers may lead to MOF-2 in C2 symmetry In contrast the formation of SURMOFs using
Liquid Phase Epitaxy (LPE) allows the formation of high-symmetry P4 MOF-2 This structure requires
a structural modification in terms of the orientation of the linkers with respect to the free monolayer
and the stacking is stabilized by London dispersion interactions between the linkers Increasing the
linker length is a straightforward way for the linear expansion of pore size and according to our
computations the pore size is only limited by the availability of linker molecules showing the desired
length Thus we presented a rare situation in which the linkers guarantee the persistence of a series
of materials in an otherwise unachievable state
172
COMPUTATIONAL DETAILS
The Becke three-parameter hybrid method combined with Lee-Yang-Par correlational functional
(B3LYP)34-37 and augmented with a dispersion term38 as implemented in Turbomole39 has been used
for DFT calculations The copper atoms were described using the basis set associated with the Hay-
Wadt40 relativistic effective core potentials41 which include polarization f functions42 This basis set
was obtained from the EMSL Basis Set Library4344 Carbon oxygen and hydrogen atoms were
described using the Pople basis set 6-311G Geometry optimizations of the MOF fragments were
performed using internal redundant co-ordinates45 A triplet spin state was assigned for each Cu-
paddlewheel46
Periodic structure calculations at the BLYP47 level were performed using the ADF BAND 2012
code4849 Dispersion correction was implemented based on the work by Grimme50 Triple-zeta basis
set was used The crystalline state of MOFs was computationally described using periodic boundary
conditions Bader charge analysis of paddlewheel-pairs was also performed using ADF 2012 code
The analysis was made at the level of B3LYP-D using an all-electron triple-zeta basis set
The non-orthogonal tight-binding approximation to DFT called Density Functional Tight-Binding
(DFTB)51 is computationally feasible for unit cells of several hundreds of atoms Hence this method
was used for extensive calculations on periodic structures This method computes a transferable set
of parameters from DFT calculations of a few molecules per pair of atom types The more accurate
self-consistent charge (SCC) extension of DFTB52-54 has been used for the study of bulk MOFs Validity
of the Slater-Koster parameters of Cu-X (X = Cu C O H) pairs were reported before55 The
computational code deMonNano56 which has dispersion correction implemented57 was used
If not stated otherwise the energies of the periodic structure are given per a formula unit (FU) of the
MOF which includes one paddlewheel and two linkers (one vertical linker and one horizontal linker)
REFERENCES
(1) Eddaoudi M Moler D B Li H L Chen B L Reineke T M OKeeffe M Yaghi O M Accounts of
Chemical Research 2001 34 319
(2) Li H Eddaoudi M OKeeffe M Yaghi O M Nature 1999 402 276
(3) Rowsell J L C Yaghi O M Microporous and Mesoporous Materials 2004 73 3
(4) Eddaoudi M Li H L Yaghi O M Journal of the American Chemical Society 2000 122 1391
(5) Rowsell J L C Yaghi O M Angewandte Chemie-International Edition 2005 44 4670
173
(6) Suh M P Park H J Prasad T K Lim D-W Chemical Reviews 2012 112 782
(7) Murray L J Dinca M Long J R Chemical Society Reviews 2009 38 1294
(8) Rosi N L Eckert J Eddaoudi M Vodak D T Kim J OKeeffe M Yaghi O M Science 2003 300
1127
(9) Kreno L E Leong K Farha O K Allendorf M Van Duyne R P Hupp J T Chemical Reviews 2012
112 1105
(10) Achmann S Hagen G Kita J Malkowsky I M Kiener C Moos R Sensors 2009 9 1574
(11) Lee J Farha O K Roberts J Scheidt K A Nguyen S T Hupp J T Chemical Society Reviews 2009
38 1450
(12) Farrusseng D Aguado S Pinel C Angewandte Chemie-International Edition 2009 48 7502
(13) Corma A Garcia H Llabres i Xamena F X Chemical Reviews 2010 110 4606
(14) Rowsell J L C Millward A R Park K S Yaghi O M Journal of the American Chemical Society 2004
126 5666
(15) Deng H Doonan C J Furukawa H Ferreira R B Towne J Knobler C B Wang B Yaghi O M
Science 2010 327 846
(16) Liu J Lukose B Shekhah O Arslan H K Weidler P Gliemann H Braumlse S Grosjean S Godt A
Feng X Muumlllen K Magdau I-B Heine T Woumlll C submitted to Nature Chemistry 2012
(17) Li H Eddaoudi M Groy T L Yaghi O M Journal of the American Chemical Society 1998 120 8571
(18) Carson C G Hardcastle K Schwartz J Liu X Hoffmann C Gerhardt R A Tannenbaum R
European Journal of Inorganic Chemistry 2009 2338
(19) Clausen H F Poulsen R D Bond A D Chevallier M A S Iversen B B Journal of Solid State
Chemistry 2005 178 3342
(20) Edgar M Mitchell R Slawin A M Z Lightfoot P Wright P A Chemistry-a European Journal 2001
7 5168
(21) Schaate A Roy P Preusse T Lohmeier S J Godt A Behrens P Chemistry-a European Journal
2011 17 9320
(22) Cote A P Benin A I Ockwig N W OKeeffe M Matzger A J Yaghi O M Science 2005 310
1166
(23) Wan S Guo J Kim J Ihee H Jiang D Angewandte Chemie-International Edition 2008 47 8826
174
(24) Lukose B Kuc A Frenzel J Heine T Beilstein Journal of Nanotechnology 2010 1 60
(25) Kitagawa S Kitaura R Noro S Angewandte Chemie-International Edition 2004 43 2334
(26) Shekhah O Wang H Zacher D Fischer R A Woell C Angewandte Chemie-International Edition
2009 48 5038
(27) Shekhah O Wang H Kowarik S Schreiber F Paulus M Tolan M Sternemann C Evers F
Zacher D Fischer R A Woll C Journal of the American Chemical Society 2007 129 15118
(28) Zacher D Schmid R Woell C Fischer R A Angewandte Chemie-International Edition 2011 50 176
(29) Bader R F W Accounts of Chemical Research 1985 18 9
(30) Merino G Vela A Heine T Chemical Reviews 2005 105 3812
(31) Tafipolsky M Schmid R Journal of Chemical Theory and Computation 2009 5 2822
(32) Kuc A Enyashin A Seifert G Journal of Physical Chemistry B 2007 111 8179
(33) Winston E B Lowell P J Vacek J Chocholousova J Michl J Price J C Physical Chemistry
Chemical Physics 2008 10 5188
(34) Becke A D Journal of Chemical Physics 1993 98 5648
(35) Lee C T Yang W T Parr R G Physical Review B 1988 37 785
(36) Vosko S H Wilk L Nusair M Canadian Journal of Physics 1980 58 1200
(37) Stephens P J Devlin F J Chabalowski C F Frisch M J Journal of Physical Chemistry 1994 98
11623
(38) Civalleri B Zicovich-Wilson C M Valenzano L Ugliengo P Crystengcomm 2008 10 405
(39) University of Karlsruhe and Forschungszentrum Karlsruhe gGmbH - TURBOMOLE V63 2011 2007
(40) Wadt W R Hay P J Journal of Chemical Physics 1985 82 284
(41) Roy L E Hay P J Martin R L Journal of Chemical Theory and Computation 2008 4 1029
(42) Ehlers A W Bohme M Dapprich S Gobbi A Hollwarth A Jonas V Kohler K F Stegmann R
Veldkamp A Frenking G Chemical Physics Letters 1993 208 111
(43) Feller D Journal of Computational Chemistry 1996 17 1571
(44) Schuchardt K L Didier B T Elsethagen T Sun L Gurumoorthi V Chase J Li J Windus T L
Journal of Chemical Information and Modeling 2007 47 1045
175
(45) von Arnim M Ahlrichs R Journal of Chemical Physics 1999 111 9183
(46) St Petkov P Vayssilov G N Liu J Shekhah O Wang Y Woell C Heine T Chemphyschem 2012
13 2025
(47) Gill P M W Johnson B G Pople J A Frisch M J Chemical Physics Letters 1992 197 499
(48) SCM Amsterdam Density Functional 2012
(49) Velde G T Bickelhaupt F M Baerends E J Guerra C F Van Gisbergen S J A Snijders J G
Ziegler T Journal of Computational Chemistry 2001 22 931
(50) Grimme S Journal of Computational Chemistry 2006 27 1787
(51) Seifert G Porezag D Frauenheim T International Journal of Quantum Chemistry 1996 58 185
(52) Elstner M Porezag D Jungnickel G Elsner J Haugk M Frauenheim T Suhai S Seifert G
Physical Review B 1998 58 7260
(53) Frauenheim T Seifert G Elstner M Hajnal Z Jungnickel G Porezag D Suhai S Scholz R
Physica Status Solidi B-Basic Research 2000 217 41
(54) Oliveira A F Seifert G Heine T Duarte H A Journal of the Brazilian Chemical Society 2009 20
1193
(55) Lukose B Supronowicz B Petkov P S Frenzel J Kuc A Seifert G Vayssilov G N Heine T
physica status solidi (b) 2011
(56) Heine T Rapacioli M Patchkovskii S Frenzel J Koester A M Calaminici P Escalante S Duarte
H A Flores R Geudtner G Goursot A Reveles J U Vela A Salahub D R deMon-nano edn ed deMon
2009
(57) Zhechkov L Heine T Patchkovskii S Seifert G Duarte H A Journal of Chemical Theory and
Computation 2005 1 841
iii
I extend my thanks to the research groups that I visited during the PhD time Dr Sourav Pal Director of National Chemical Laboratory Pune and Dr V Subrahmanian Central Leather Research Institute Chennai deserve my gratitude for giving me the opportunity to visit and work with their group members Also I am very thankful to Prof D Sc Georgi N Vayssilov Faculty of Chemistry University of Sofia for the interesting collaboration and visit to his group The financial assistance during each stay is greatly acknowledged I also thank the members of the respective groups namely Dr Petko Petkov and his family who made the visit to Bulgaria very much entertaining
Prof Dr Lars Pettersson of University of Stockholm Dr Tzonka Mineva of CNRS Montpellier and all other members of the HYPOMAP research project are acknowledged for the scientific discussions exposures and promotions
I acknowledge several projects of Prof Dr Thomas Heine for the financial support of my work and travel the funding sources include the European Commission Deutsche Forschungsgemeinschaft (DFG) and the joint Bulgarian-German exchange program (DAAD)
I thank all the co-authors of my publications who have contributed their knowledge ideas and work to accomplish our scientific goals Without their efforts all those works would not have been complete
Members of Research III of SES at Jacobs University namely Robert Carsten Joumlrg Bogdan Meisam Niraj Mahesh Vinu Pinky Patrice Mehdi Sidhant and all professors postdocs and students in Nanofun center are thankfully mentioned here
A lot of my friends in the campus deserve my thanks Mahesh Mahendran Vinu Deepa Srikanth Rajesh Arumugam Prasad Dhananjay Sunil Tripti Raghu Suneetha Rami Susruta Niraj Abhishek Ashok Rakesh Sagar Rohan Naveen Yauhen Yannic Mila and Samira are thanked for the gatherings travels making funs and those cricket and volleyball evenings Some of them are specially thanked for the occasional ldquogahn bayrdquo parties I owe many thanks to Yauhen Srikanth and Prasad for being good flat-mates and having talks on any matters Srikanth and Prasad are thanked again for generously extending their cooking skills to me
I wish to thank everybody with whom I have shared experiences in life I am obliged to my MSc lecturer Dr Rajan K John whose dreams have inspired and driven me to research In particular his accomplishments in the George Sudarshan Center CMS College Kottayam have molded me to take up this career My previous research supervisors Prof S Lakshmibala and Prof V Balakrishnan of IIT Madras and Dr Anita Mehta of SNBNCBS Kolkata are also acknowledged for their important influences in my academic life Additionally all my teachers friends and well-wishers from neighborhood school college GS Center IIT-M and SN Bose center are thanked and acknowledged Members of St Antonyrsquos Parish Olassa are also thanked for the regards and encouragement
Jacobs University Bremen and its people have been amazing in all sorts of things I am glad that I have been a member of the University With my full heart I thank the university authority for all its facilities that were open for me I also thank Dr Svenja Frischholz Mr Peter Tsvetkov and Ms Kaija Gruumlenefeld in the administration departments for the timely helps
Lastly and most importantly I wish to thank my dearest ones for all the sacrifices and love My parents K P Lukose and Molly and my brother Anit deserve to be thanked They have always supported and encouraged me to do my best in all matters of life I also wish to thank my entire extended family for providing me a loving environment
iv
Abstract
Framework materials are extended structures that are built into destined nanoscale architectures
using molecular building units Reticular synthesis methods allow stitching of a large variety of
molecules into predicted networks Porosity is an obvious outcome of the stitching process These
materials are classified and named according to the chemical composition of the building blocks For
instance Metal-Organic Frameworks (MOFs) consists of metal-oxide centers that are linked together
by organic entities The stitching process is straight-forward so that there are already thousands of
them synthesized Controlled growth of MOFs on substrates leads to what is known as surface-MOFs
(SURMOFs) A low-weight metal-free version of MOFs is known as Covalent Organic Frameworks
(COFs) They consist of light elements such as boron oxygen silicon nitrogen carbon and hydrogen
atoms Diamond-like structures with a variety of linear organic linkers between tetrahedral nodes is
called Porous Aromatic Frameworks (PAFs)
The thesis is composed of computational studies of the above mentioned classes of materials The
significance of such studies lies in the insights that it gives about the structure-property relationships
Density Functional Theory (DFT) and its Tight-Binding approximate method (DFTB) were used in
order to perform extensive calculations on finite and periodic structures of several frameworks DFTB
provides an ab-initio base on periodic structure calculations of very large crystals which are typically
studied only using force-field methods The accuracy of this approximate method is validated prior to
reasoning
As the materials are energized from building units and coordination (or binding) stability vs
structure is discussed Energy of formation and mechanical strength are particularly calculated Using
dispersion corrected (DC)-Self Consistent Charge (SCC) DFTB we asserted that 2D COFs should have a
layer arrangement different from experimental suggestions Our arguments supported by simulated
PXRDs were later verified using higher level theories in the literature Another benchmark is giving an
insightful view on the recently reported difference in symmetries of two-dimensional MOFs and
SURMOFs The result of it shows an extra-ordinary role of linkers in SURMOFs providing
metastability
Electronic properties of all the materials and hydrogen adsorption capacities of PAFs are discussed
COFs PAFs and many of the MOFs are semiconductors HOMO-LUMO gaps of molecular units have
crucial influence in the band gaps of the solids Density of states (DOS) of solids corresponds to that
of cluster models Designed PAFs give a large choice of adsorption capacities one of them exceeds
the DOE (US) 2005 target Generally the studies covered under the scheme of the thesis illustrate
the structure stability and properties of framework materials
- Dedicated to my Family and Rajan sir
Table of Contents 1 Outline 1
2 Introduction 2
21 Nanoporous Materials 2
22 Reticular Chemistry 3
23 Metal-Organic Frameworks 5
24 Covalently-bound Organic Frameworks 8
3 Methodology and Validation 10
31 Methods and Codes 10
32 DFTB Validation 11
4 2D Covalent Organic Frameworks 13
41 Stacking 13
42 Concept of Reticular Chemistry 15
5 3D Frameworks 17
51 3D Covalent Organic Frameworks 17
52 Porous Aromatic Frameworks 18
6 New Building Concepts 20
61 Isoreticular Series of SURMOFs 20
62 Metastability of SURMOFs 21
7 Summary 23
71 Validation of Methods 23
72 Weak Interactions in 2D Materials 25
73 Structure-Property Relationships 27
List of Abbreviations 31
List of Figures 32
References 33
Appendix A Review of covalently-bound organic frameworks 37
Appendix B Properties of MOFs within DFTB 81
Appendix C Stacking of 2D COFs 96
Appendix D Reticular concepts applied to 2D COFs 105
Appendix E Properties of 3D COFs 122
Appendix F Properties of PAFs 131
Appendix G Isoreticular SURMOFs of varying pore sizes 145
Appendix H Metastability in 2D SURMOFs 160
1
1 Outline
I prepared this cumulative thesis as it is permitted by Jacobs University Bremen as all work has been
published in international peer-reviewed journals is submitted for publication or in a late
manuscript state in order to be submitted soon The list of articles contains three published papers
three submitted manuscripts and two manuscripts that are to be submitted The articles are given in
Appendices A-H in the order of their discussions Each appendix has one paper and its supporting
information
The chapters from ldquoIntroductionrdquo to ldquoNew Building Conceptsrdquo contain general discussions about the
articles and provide a red thread leading through the articles The discussions mainly circle around
the context and the content of the articles
The chapter ldquoIntroductionrdquo provides a general introduction to the topic and a review of the materials
discussed in this thesis As no comprehensive review on ldquoCovalently-bound Organic Frameworksrdquo is
available in the literature we prepared a manuscript (Appendix A) covering that subject The chapter
ldquoMethodology and Validationrdquo discusses one article on validation of DFTB method in Metal-Organic
Frameworks (Appendix B) Each consecutive chapter discusses two articles The chapter ldquo2D
Covalent Organic Frameworksrdquo covers the studies on layer arrangements (Appendix C) followed by
analysis on how reticular chemistry concepts can be used to design new materials (Appendix D) The
chapter ldquo3D Frameworksrdquo discusses the stability and properties of 3D COFs (Appendix E) and PAFs
(Appendix F) It also discusses hydrogen storage capacities of PAFs The chapter ldquoNew Building
Conceptsrdquo has coverage on 2D MOFs that are built on organic surfaces The first article in this chapter
describes the achievement of our collaborators in synthesizing a series of SURMOFs of varying pore
sizes supported by our calculations indicating their matastability Extensive calculations revealing the
role of linkers in creating the unprecedented difference of SURMOFs in comparison with the bulk
MOFs is described in another article
Details of the articles and references to the appendices are given in the respective places in each
chapter The chapter ldquoSummaryrdquo gives an overview of all the results and discussions It also discusses
some impacts of the publications and concludes the thesis Overall the studies bring into picture
different classes of materials and analyze their structural stabilities and properties
2
2 Introduction
21 Nanoporous Materials
The field of nanomaterials covers materials that have properties stemming from their nanoscale
dimensions (1-100 nm) In contrast to nanomaterials which define size scaling of bulk materials the
major determinant of nanoporous materials is their pores Nanoporous materials are defined as
porous materials with pore diameters less than 100 nm and are classified as micropores of less than
2 nm in diameter mesopores between 2 and 50 nm and macropores of greater than 50 nm They
have perfectly ordered voids to accommodate interact with and discriminate molecules leading to
prominent applications such as gas storage separation and sieving catalysis filtration and
sensoring1-4 They differ from the generally defined nanomaterials in the sense that their properties
are mostly determined by pore specifications rather than by bulk and surface scales Hence the
focus is onto the porous properties of the materials
Utilization of the pores for certain applications relies on certain parameters such as pore size pore
volume internal surface area and wall composition For example physical adsorption of gases is high
in a material with large surface area which implies significantly high storage in comparison to a tank
Porosity can be measured using some inert or simple gas adsorption measurements Distribution of
pore size can be sketched from the adsorptiondesorption isotherm
Nanoporous materials abound in nature Zeolites for instance many of them are available as mineals
have been used in petroleum industry as catalysts for decades The walls of human cells are
nanoporous membranes Other examples are clays aluminosilicate minerals and microporous
charcoals Zeolites are microporous materials from the aluminosilicate family and are often known as
molecular sieves due to their ability to selectively sort molecules primarily based on a size exclusion
principle A material with high carbon content (coal wood coconut shells etc) can be converted to
activated carbon (AC) by controlled thermal decomposition in a furnace The resultant product has
large surface area (~ 500 m2 g-1) Porous glass is a material produced from borosilicate glasses having
pore sizes usually in nm to μm range With increasing environmental concerns worldwide porous
materials have become a suitable choice for separation of polluting gases storage and transport of
energy carriers etc Synthesis of open structures has advanced this area5-7 especially since the
invention of MCM-41 (Mobil Composition of Matter No 41) by Mobil scientists89 Furthermore
there are many templating pathways in making nanoporous materials10-13 Currently it is possible to
engineer the internal geometry at molecular scales
3
For more than a decade chemists are able to synthesize extended structures from well-defined and
rigid molecular building units Such designed and controlled extensions provide porosity which can
be scaled and modified by selecting appropriate building blocks The first realization of this kind was
a class of materials known as Metal-Organic Frameworks (MOFs) in which metal-oxides are stitched
together by organic molecules Synthesis of molecules into predicted frameworks have led to the
emergence of a discipline called reticular chemistry14 The new design-based synthetic approaches
have produced large number of nanoporous materials in comparison to the discovery-based
synthetic chemistry
22 Reticular Chemistry
The discipline of designed (rational) synthesis of materials is known as reticular chemistry Desired
materials can be realized by starting with well-defined and rigid molecular building blocks that will
maintain their structural integrity throughout the construction process The extended structures
adopt high symmetry topologies The synthetic approach follows well-defined conditions which
provide general control over the character of solids In short it is the chemistry of linking molecular
building blocks by strong bonds into predetermined structures
The knowledge about how atoms organize themselves during synthesis is essential for the design
The principal possibilities rely on the starting compounds and the reaction mechanisms Porosity is
almost an inevitable outcome of reticular synthesis Solvents may be used in most cases as space-
filling agents and in cases of more than one possibility as structure-directing agents
Thousands of materials in large varieties have been synthesized using the reticular chemistry
principles Reticular Chemistry Structure Resource (RCSR) is a searchable database for nets a project
initiated by of Omar M Yaghi and Michael OrsquoKeeffe15 They define nets as the collection of vertices
and edges that form an irreducible network in which any two vertices are connected through at least
one continuous path of edges Building units are finite nets as in the case of a polyhedron Periodic
structures have one two or three-periodic nets An example case of 3D diamond net (dia) is shown in
Figure 1 Graph-theoretical aspects together with explicative terminology and definitions can be
found in the literature16-18
Figure 1 a) ldquoAdamantinerdquo unit of diamond structure and b) diamond (dia) net
4
In other words a framework can be deconstructed into one or more fundamental building blocks
each of them assigned by a vertex in the net The vertices are the branching points and edges are
joining them The realization of the net in space by representing the vertices and lattice parameters
by co-ordinates and a matrix respectively is called embedding of the net Underlying topology of an
extended structure is the structure of the net inherited from the crystal structure that is invariant
under different embeddings For example Cu-BTC MOF has paddlewheels and benzene rings as
fundamental blocks The MOF structure can be simplified into its underlying topology as shown in
Figure 2
Figure 2 CU-BTC MOF and the corresponding tbo net
Alternatively the topology of a framework can be defined using the convention of so-called
secondary building units (SBUs) The shapes of SBUs are defined by points of extension of the
fundamental building blocks SBUs are invariant for building units of identical connectivity Based on
the number of connections with the adjacent blocks (4 for paddlewheel and 3 for the benzene) SBUs
of Cu-BTC can be represented as shown in Figure 3a Assembly of SBUs following the network
topology and insertion of the fundamental building blocks give the crystal structure (see Figure 3b for
the extension of SBUs to the topology of Cu-BTC)
In addition to MOFs Metal-Organic Polyhedra (MOP)19 Zeolite Imidazolate Frameworks (ZIFs)20 and
Covalent-Organic Frameworks (COFs)2122 are developing classes of materials within reticular
chemistry the first members of all of them were synthesized by the group of Yaghi MOPs are nano-
sized polyhedra achieved by linking transition metal ions and either nitrogen or carboxylate donor
organic units ZIFs are aluminosilicate zeolites with replacements of tetrahedral Si (Al) and bridging
oxygen by transition metal ion and imidazolate link respectively COFs are extended organic
5
structures constructed solely from light elements (H B C and O) The materials synthesized under
the reticular scheme are largely crystalline
Figure 3 a) Fundamental building blocks and SBUs of Cu-BTC and b) extension of SBUs following
crystal structure
23 Metal-Organic Frameworks
MOFs are porous crystalline materials consisting of metal complexes (connectors) linked together by
rigid organic molecules (linkers) MOFs are analogues to a class of materials called coordination
polymers (CPs) However there are primary differences between them CPs are inorganic or
organometallic polymer structures containing metal ions linked by organic ligands A ligand is an
atom or a group of atoms that donate one or more lone pairs of electrons to the metal cations and
thereby participate in the formation of a coordination complex In MOFs typically metal-oxide
centers are used instead of single metal ions as they provide strong bonds with organic linkers This
provides not only high stability but also high directionality because multiple bonds are involved
6
between metal-centers and organic linkers Predictability lies in the pre-knowledge about the
connector-linker interactions Thus the reticular design of MOFs derives from the precise
coordination geometry between metal centers and the linkers Figure 4a shows a schematic diagram
of MOFs By varying the chemical composition of nodes and linkers thousands of different MOF
structures with a large variety in pore size and structure have been synthesized Figure 4b shows
MOF-5 that consists of zinc-oxide tetrahedrons and benzene linkers
Figure 4 a) Schematic diagram of the single pore of MOF b) An example of a simple cubic MOF in
which zinc-oxide tetrahedron are linked together by benzene linkers Atom colors blue ndash Zn red ndash
O grey ndash C white ndash H
The synthesis of MOFs differs from organic polymerizations in the degree of reversible bond
formation Reversibility allows detachment of incoherently matched monomers followed by their
attachment to form defect-free crystals Assembly of monomers occurs as single step hence
synthetic conditions are of high importance The strength of the metal-organic bond is an obstacle
for reversible bond formation however solvothermal techniques are found out to be a convenient
solution23 Solvothermal synthesis generally allows control over size and shape distribution Using
post-synthetic methods further changes on cavity sizes and chemical affinities can be made
Materials that are stable with open pores after removal of guest molecules are termed as open-
frameworks The stability of guest-free materials is usually examined by Powder X-ray diffraction
(PXRD) analysis Thermogravimetric analysis may be performed to inspect the thermal stability of the
material Elemental analysis can detail the elemental composition of the material Physical
techniques such as Fourier transform infrared (FTIR) spectra and nuclear magnetic resonance (NMR)
may be used to verify the condensation of monomers to the desired topology Porosity can be
evidenced from adsorption isotherms of gases or mercury porosimetry
7
The number of linkers that are allowed to bind to the metal-center and the orientations of the linkers
depend exclusively on the coordination preferences of the metal The organic linkers are typically
ditopic or polytopic They are essential in determining the topology and providing porosity Longer
linkers provide larger pore size A series of compounds with the same underlying topology and
different linker sizes are called iso-reticular series The structure of MOFs in principle can be formed
into the requirement of prominent applications such as gas storage gas separation sensing and
catalysis The structural divergence and performance can be further increased by functionalizing the
organic linkers Hence several attempts are on-going in purpose to come up with the best material
possible in each application
Interpenetration of frameworks is an inevitable outcome of large pore volume Interpenetrated nets
are periodic equivalents of mechanically interlocked molecules rotaxanes and catenanes Depending
on topology they are either maximally separated termed as interpenetration or minimally separated
termed as interweaving (see Figure 5) Interpenetration can be beneficial to some porous structures
protecting from collapse upon removal of solvents
Figure 5 Schematic diagram of interpenetrated and interwoven frameworks
Controlled growth of MOFs on organic surfaces is achieved by R A Fischer and C Woumlll24 The then
named SURMOFs allow to fabricate structures possibly not achievable by bulk synthesis The growth
is achieved through liquid phase epitaxy (LPE) of MOF reactants on the surface (nucleation site) A
step-by-step approach which is alternate immersion of the substrate in metal and ligand precursors
supplies control of the growth mechanism
8
Figure 6 Schematic diagram of SURMOF
24 Covalently-bound Organic Frameworks
As a result of the long-term struggle for complete covalent crystallization in the year 2005 Cocircte et
al21 synthesized porous crystalline extended 2D Covalent-Organic Frameworks (COFs) using
reticular concepts The success was followed by the design and synthesis of 3D COFs in the year
200722 By now there are about 50 COFs reported in the literature COFs are made entirely from
light elements and the building blocks are held together by strong covalent bonds Most of them
were formed by solvothermal condensation of boronic acids with hydroxyphenyl compounds
Tetragonal building blocks were used for the formation of 3D COFs Alternate synthetic methods
were also used for producing COFs COFs are generally studied for gas storage applications However
they have also shown potentialities in photonic and catalytic applications
Alternative synthesis methods paved the way to new covalently bound organic frameworks
Trimerization of polytriazine monomers in ionothermal conditions gives rise to Covalent Triazine
Frameworks (CTFs)25 Similarly coupling reactions of halogenated monomers produced Porous
Aromatic Frameworks (PAFs) Albeit the short-range order one of the PAFs showed very high surface
area (5600 m2 g-1) and gas uptake capacity26
Due to low weight the covalently-bound materials show very high gravimetric capacities
Suggestions such as metal-doping functionalization and geometry modifications can be found in the
literature for the general improvement of the functionalities There are also various studies of their
structure and properties
A review on the synthesis structure and applications of covalently bound organic frameworks has
been prepared for publication
Article 1 Covalently-bound organic frameworks
Binit Lukose Thomas Heine
9
See Appendix A for the article
My contributions include collecting data and preparing a preliminary manuscript
Figure 7 SBUs and topologies of 2D COFs
10
3 Methodology and Validation
31 Methods and Codes
The Density-Functional based Tight Binding (DFTB) method2728 is an approximate Kohn-Sham DFT29-31
scheme It avoids any empirical parameterization by calculating the Hamiltonian and overlap matrix
elements from DFT-derived local orbitals (atomic orbitals) and atomic potentials The Kohn-Sham
orbitals are represented with the linear combination of atomic orbitals (LCAO) scheme The matrix
elements of Hamiltonian are restricted to include only one- and two-center contributions Therefore
they can be calculated and tabulated in advance as functions of the distance between atomic pairs
The remaining contributions to the total energy that is the nucleus-nucleus repulsion and the
electronic double counting terms are grouped in the so-called repulsive potential This two-center
potential is fitted to results of DFT calculations of properly chosen reference systems The accuracy
and transferability can be improved with the self-consistent charge (SCC) extension to DFTB32 This
method is based on the second-order expansion of the Kohn-Sham total energy with respect to
charge density fluctuations which are estimated by Mulliken charge analysis In order to account for
London dispersion empirical dispersion energy can be added to the total energy3334 Various reviews
are available on DFTB notably the recent ones by Oliveira et al35 and by Seifert et al36
DFTB is implemented in a large number of computer codes For this work we employed the codes
deMonNano37 and DFTB+38 as they allow DFTB calculations on periodic and finite structures
Typically dispersion corrected (DC) SCC-DFTB calculations were carried out Periodic boundary
conditions were used to represent the crystalline frameworks and as the unit cells are large the
standard approach used the point approximation Electronic density of states (DOS) have been
calculated using the DFTB+ code using k-point sampling where the k space was determined by
reaching convergence for the total energy according to the scheme proposed by Monkhorst and
Pack39
For DFT calculations of finite and periodic structures ADF40 Turbomole41 and Crystal0942 were used
For studies of finite models of COFs the calculations were performed at PBEDZP level However for
extensive calculations on SURMOF-2 models we used B3LYP-D The copper atoms were described
using the basis set associated with the Hay-Wadt43 relativistic effective core potentials44 which
include polarization f functions45 Carbon oxygen and hydrogen atoms were described using the
Pople basis set 6-311G
Details of the individual calculations are given in the individual articles in the appendix of this thesis
11
32 DFTB Validation
Figure 8 Deconstructed building units their schematic diagrams and final geometries of HKUST-1
(Cu-BTC) MOF-5 MOF-177 MOF-205 and MIL-53
12
In the literature MOFs and COFs are largely studied for applications such as gas storage using
classical force field methods46-48 First principles based studies of several hundreds of atoms are
computationally expensive Hence they are generally limited to cluster models of the periodic
structures Contrarily DFTB paves the way to model periodic structures involving large numbers of
atoms Being an approximate method DFTB holds less accuracy hence comparisons to experimental
data or higher level methods should be performed for validation
As MOFs contain metal centers andor metal oxides as well as organic elements the DFTB
parameters for both heavy and light elements as well as their mixtures are required Thus we have
chosen MOFs such as Cu-BTC (HKUST-1) MOF-5 MOF-177 MOF-205 and MIL-53 as our model
structures which contain Cu Zn and Al atoms apart from C O and H The MOFs comprising three
common connector units copper paddlewheels (HKUST-1) zinc oxide Zn4O tetrahedron (MOF-5
MOF-177 DUT-6 (MOF-205)) and aluminum oxide AlO4(OH)2 octahedron (MIL-53) Figure 8 shows
the schematic diagram of the MOFs
The validation calculations have been published
Article 2 Structural properties of metal-organic frameworks within the density-functional based
tight-binding method
Binit Lukose Barbara Supronowicz Petko St Petkov Johannes Frenzel Agnieszka B Kuc Gotthard
Seifert Georgi N Vayssilov and Thomas Heine Phys Status Solidi B 2012 249 335ndash342 DOI
101002pssb201100634
See Appendix B for the article
In this article DFTB has been validated against full hybrid density-functional calculations for model
clusters against gradient corrected density-functional calculations for supercells and against
experiment Moreover the modular concept of MOF chemistry has been discussed on the basis of
their electronic properties Our results show that DC-SCC-DFTB predicts structural parameters with a
good accuracy (with less than 5 deviation even for adsorbed CO and H2O on HKUST-1) while
adsorption energies differ by 12 kJ mol-1 or less for CO and water compared to DFT benchmark
calculations
My contributions to this article include performing DFTB based calculations on the MOFs (HKUST-1
MOF-5 MOF-177 and MOF-205) that is optimizing the geometries simulating powder X-ray
diffraction patterns and calculating density of states and bulk modulus Additional involvement is in
comparing structural parameters such as bond lengths bond angles dihedral angles and bulk
modulus with experimental data or data derived from DFT calculations and preparing the manuscript
13
4 2D Covalent Organic Frameworks
41 Stacking
Condensation of planar boronic acids and hydroxyphenyl compounds leads to the formation of two-
dimensional covalent organic frameworks (2D COFs) The layers are held together by London
dispersion interactions The first synthesized COFs COF-1 and COF-5 were reported to have AB
(P63mmc staggered as in graphite) and AA (P6mmm eclipsed as in boron nitride) stackings
respectively (see Figure 9)21 The synthesis of several further 2D COFs then followed49-51 all of them
were inspected for only two stacking possibilities ndash P6mmm or P63mmc and it was reported that
they aggregate in P6mmm symmetry As framework materials possess framework charges the
interlayer interactions significantly include Coulomb interactions P6mmm symmetry is the face-to-
face arrangement where the overlap of the stacked structures is maximized (maximization of the
London dispersion energy) however atom types of alike charges are facing each other in the closest
possible way With the interlayer separation similar to that in graphite or boron nitride the Coulomb
repulsion should be high in such arrangements One should notice that in the example case of boron
nitride the facing atom types are different We therefore assumed that a stable stacking should
possess layer-offset
Figure 9 AA and AB layer stacks of hexagonal layers
We considered two symmetric directions for layer shift and studied their total energies (see Figure
10) The stable geometries are found to have a layer-shift of ~14 Aring a value corresponding to the
shift of AA to AB stacking in graphite and compatible with the bond length between two 2nd row
atoms This stability-supported stacking arrangement as revealed from our calculations was
14
supported by good agreement between simulated and experimental PXRD patterns Hence
independent of the elementary building blocks any 2D COF should expose a layer-offset Based on
the direction of the shift we proposed two stacking kinds serrated and inclined (Figure 10) In the
former the layer-offset is back and forth while in the latter the layer-offset followed single direction
As serrated and inclined stackings have no significant change in stacking energy our calculations
cannot predict the long-range stacking in the crystal However this problem is known from other
layered compounds and the ultimate answer is yet to be revealed by analyzing high-quality
crystalline phases at low temperature
Figure 10 Stacking energy Es vs shift of adjacent layers for COF-5 Different stacking possibilities
and their energies are also shown
We published our analysis of the stacking in 2D COFs
Article 3 The Structure of Layered Covalent-Organic Frameworks
Binit Lukose Agnieszka Kuc Thomas Heine Chem Eur J 2011 17 2388 ndash 2392 DOI
101002chem201001290
See Appendix C for the article
15
My contributions to this article include performing the shift calculations simulating XRDs and partly
preparing the manuscript The article has made certain impact in the literature Most of the 2D COFs
synthesized afterwards were inspected for their stacking stability The instability of AA stacking was
also suggested by the studies of elastic properties of COF-1 and COF-5 by Zhou et al52 The shear
modulus shows negative signs for the vertical alignment of COF layers while they are small but
positive for the offset of layers Detailed analysis performed by Dichtel53 et al using DFT was
confirming the instability of AA stacking and suggesting shifts of about 13 to 25 Aring
42 Concept of Reticular Chemistry
Reticular chemistry means that (functional) molecules can be stitched together to form regular
networks The structural integrity of these molecules we also speak of building blocks remains in the
crystal lattices Consequently also the electronic structure and hence the functionality of these
molecules should remain similar
2D COFs are formed by condensation of well-defined planar building blocks With the usage of linear
and triangular building blocks hexagonal networks are expected The properties of each COF may
differ due to its unique constituents However the extent of the relationship of the properties of
building blocks in and outside the framework has not been studied in the literature
Reticular chemistry allows the design of framework materials with pre-knowledge of starting
compounds and reaction mechanisms Reverse of this is finding reactants for a certain topology We
intended to propose some building units suitable to form layered structures (see Figure 11) The
building units obey the regulations of reticular chemistry and offer a variety of structural and
electronic parameters
Our strategic studies on a set of designed COFs have been published
Article 4 On the reticular construction concept of covalent organic frameworks
Binit Lukose Agnieszka Kuc Johannes Frenzel Thomas Heine Beilstein J Nanotechnol 2010 1
60ndash70 DOI103762bjnano18
See Appendix D for the article
16
Figure 11 Schematic diagram of different building units forming 2D COFs
Various hexagonal 2D COFs with different building blocks have been designed and investigated
Stability calculations indicated that all materials have the layer offset as reported in our earlier
work54 (see Appendix C) We show that the concept of reticular chemistry works as the Density-Of-
States (DOS) of the framework materials vary with the the DOS of the molecules involved in the
frameworks However the stacking does have some influence on the band gap
My contributions to this article include performing all the calculations and preparing the manuscript
17
5 3D Frameworks
51 3D Covalent Organic Frameworks
First 3D COFs were reported in 2007 by the group of Yaghi22 Although the number of 3D COFs
synthesized so far has not been crossed half a dozen they are of particular interest for their very low
mass density Similar to 2D COFs condensation of boronic acids and hydroxyphenyl compounds led
to the formation of COF-102 103 105 108 and 202 Usage of tetragonal boronic acids leads to the
formation of 3D networks (Figure 12) All of them possessed ctn topology while COF-108 which has
the same material composition as COF-105 crystallized in bor topology COF-300 which was formed
from tetragonal and linear building units possessed diamond topology and was five-fold
interpenetrated55 COF-108 has the lowest mass density of all materials known today Unlike most of
the MOFs the connectors in COFs are not metal oxide clusters but covalently bound molecular
moieties In the synthesized COFs the centers of the tetragonal blocks were either sp3 carbon or
silicon atoms
Schmid et al56 have analyzed the two different topologies and developed force field parameters for
COFs The mechanical stability of COFs was also reported However no further study that details the
inherent energetic stability and properties of COFs was found in the literature Using DFTB we
performed a collective study of all 3D COFs in their known topologies with C and Si centers
Figure 12 Schematic diagram of tetragonal and triangular building blocks forming lsquoctnrsquo or lsquoborrsquo
topologies
Our studies of3D COFs have been prepared for publication
Article 5 Stability and electronic properties of 3D covalent organic frameworks
Binit Lukose Agnieszka Kuc Thomas Heine
18
See Appendix E for the article
My contributions to this article include performing all the calculations and preparing the manuscript
We discussed the energetic and mechanical stability as well as the electronic properties of COFs in
the article All studied 3D COFs are energetically stable materials with formation energies of 76 ndash
403 kJ mol-1 The bulk modulus ranges from 29 to 206 GPa Electronically all COFs are
semiconductors with band gaps vary with the number of sp2 carbon atoms present in the linkers
similar to 3D MOFs
52 Porous Aromatic Frameworks
Porous Aromatic Frameworks (PAFs) are purely organic structures where linkers are connected by sp3
carbon atoms (see Appendix A) Tetragonal units were coupled together to form diamond-like
networks The synthesis follows typically a Yamamoto-type Ullmann cross-coupling reaction those
reactions are known to be much simpler to be carried out than the condensation reactions necessary
to form COFs However as the coupling reactions are irreversible a lower degree of crystallinity is
achieved and the materials formed were amorphous The first PAF was reported in 2009 and
showed a very high BET surface area of 5600 m2g-1 The geometry of PAF-1 was equal to diamond
with the C-C bonds replaced by biphenyl Subsequent PAFs may be synthesized with longer linkers
between carbon tetragonal nodes Nevertheless synthesis of a PAF with quaterphenyl linker
provided an amorphous material of very low surface area due to the short range order
Synthetic complexities aside the diamond-like PAFs are interesting for their special geometries from
the viewpoint of the theorist It is interesting to see to what extent they follow the properties of
diamond Additionally the large surface area and high polarizability of aromatic rings are auxiliary for
enhanced hydrogen storage Thus we have explored the possibilities of diamond topology by
inserting various organic linkers in place of C-C bonds (Figure 13)
Figure 13 Diamond structure and various organic linkers to build up PAFs
Our studies of PAFs have been prepared for publication
19
Article 6 Structure electronic structure and hydrogen adsorption capacity of porous aromatic
frameworks
Binit Lukose Mohammad Wahiduzzaman Agnieszka Kuc Thomas Heine
See Appendix F for the article
In this article we have discussed the correlations of properties with the structures Exothermic
formation energies of PAFs calculated from the dehydrogenation of the linkers with CH4 indicate the
strong coordination of linkers in the diamond topology The studied PAFs exhibited bulk moduli much
smaller than diamond however in line with related MOFs and COFs The PAFs are semi-conductors
with their band gaps decrease with the increasing number of benzene rings in the linkers
Using Quantized Liquid Density Functional Theory (QLDFT) for hydrogen57 a method to compute
hydrogen adsorption that takes into account inter-particle interactions and quantum effects we
predicted a large choice of hydrogen uptake capacities in PAFs At room temperature and 100 bar
the predicted uptake in PAF-qtph was 732 wt which exceeds the 2015 DOE (US) target We
further discussed the structural impacts on the adsorption capacities
My contributions to this article include designing the materials performing calculations of stability
and electronic properties describing the adsorption capacities and preparing the manuscript
20
6 New Building Concepts
61 Isoreticular Series of SURMOFs
The growth of MOFs on substrates using liquid phase epitaxy (LPE) opened up a controlled way to
construct frameworks This is achieved by alternate immersion of the substrates in metal and ligand
precursors for several cycles Such MOFs named as SURMOFs can avoid interpenetration because
the degeneracy is lifted58 and are suited for conventional applications This is an advantage as
solvothermally synthesized MOFs with larger pore size suffer from interpenetration and the large
pores are hence not accessible for guest species
MOF-259 is a simple 2D architecture based on paddlewheel units formed by attaching four
dicarboxylic groups to Cu2+ or Zn2+ dimers yielding planar sheets with 4-fold symmetry The
arrangement of MOF layers possessed P2 or C2 symmetry Recently the group of C Woumlll at KIT has
synthesized an isoreticular series of SURMOFs derived from MOF-2 which has a homogeneous series
of linkers with different lengths (Figure 14) XRD comparisons suggest that these MOFs possess P4
symmetry The cell dimensions that correspond to the length of the pore walls range from 1 to 28
nm The diagonal pore-width of one of the SURMOFs (PPDC) accounts to 4 nm The symmetry of
SURMOFs as determined to be different from bulk MOF-2 should be understood by means of theory
As collaborators we simulated the structures and inspected each stacking corresponding to the
symmetries in order to understand the difference
Figure 14 The building units and schematic diagram of the formation of iso-reticular SURMOF
series
21
This collaborated work has been submitted for publication
Article 7 A novel series of isoreticular metal organic frameworks realizing metastable structures
by liquid phase epitaxy
Jinxuan Liu Binit Lukose Osama Shekhah Hasan Kemal Arslan Peter Weidler Hartmut Gliemann
Stefan Braumlse Sylvain Grosjean Adelheid Godt Xinliang Feng Klaus Muumlllen Ioan-Bogdan Magdau
Thomas Heine Christof Woumlll
See Appendix G for the article
The main contribution of this article was the experimental proof backed up by our computer
simulations that SURMOFs with exceptionally large pore sizes of more than 4 nm can be prepared in
the laboratory and they offer the possibility to host large materials such as clusters nanoparticles or
small proteins The most important contribution of theory was to show that while MOF-2 in P2
symmetry shows the highest stability the P4 symmetry as obtained using LPE for SURMOF-2
corresponds to a local minimum
My contribution to this article includes performing and analyzing the calculations Our theoretical
study went significantly beyond and will be published as separate article (Appendix H)
62 Metastability of SURMOFs
Following the difference in stability of SURMOFs-2 in comparison with MOF-2 we identified the role
of linkers as a constituent of the framework that provides the metastability to SURMOFs-2 (Figure
15) In MOFs linkers are essential in determining the topology as well as providing porosity Linkers
typically contain single or multiple aromatic rings and are ditopic and polytopic The orientation of
them normally undergoes lowest repulsion from the coordinated metal-oxide clusters and provides
high stability In the case of 2D MOFs non-covalent interlayer interactions lead to the most stable
arrangements Paddlewheels are identified as the reasons for the stability of bulk MOF-2 as they
form metal-oxide bridges across the MOF layers In the case of SURMOFs-2 the linkers are rotated in
a tightly-packed environment of layers and each linker in bulk MOF-2 is rotated in such a way that
any attempt of shifting is obstructed by repulsion between the neighboring linkers The energy
barrier for leaving P4 increases with the length of organic linkers Moreover SURMOF-2 derivatives
with extremely large linkers are energetically stable due to the increased London dispersion
interaction between the layers in formula units Thus we encountered a rare situation in which the
linkers guarantee the persistence of a series of materials in an otherwise unachievable state
22
Figure 15 Energy diagram of the metastable P4 and stable P2 structures
Our results on the linker guided stability of SUMORs-2 have been prepared for publication
Article 8 Linker guided metastability in templated Metal-Organic Framework-2 derivatives
(SURMOFs-2)
Binit Lukose Gabriel Merino Christof Woumlll Thomas Heine
See Appendix H for the article
This article is based solely on my scientific contributions
23
7 Summary
Nanotechnology is the way of ingeniously controlling the building of small and large structures with
intricate properties it is the way of the future a way of precise controlled building with incidentally
environmental benignness built in by design - Roald Hoffmann Chemistry Nobel Laureate 1981
Currently it is possible to design new materials rather than discovering them by serendipity The
design and control of materials at the nanoscale requires precise understanding of the molecular
interactions processes and phenomena In the next level the characteristics and functionalities of
the materials which are inherent to the material composition and structure need to be studied The
understanding of the materials properties may be put into the design of new materials
Computational tools to a large extend provide insights into the structures and properties of the
materials They also help to convert primary insights into new designs and carry out stability analysis
Our theoretical studies of a variety of materials have provided some insights on their underlying
structures and properties The primary differences in the material compositions and skeletons
attributed a certain choice in properties The contents of the articles discussed in the thesis may be
summarized into the following three parts
71 Validation of Methods
Simulations of nanoporous materials typically include electronic structure calculations that describe
and predict properties of the materials Density functional theory (DFT)30 is the standard and widely-
used tool for the investigation of the electronic structure of solids and molecules Even the optical
properties can be studied through the time-dependent generalization of DFT MOFs and COFs have
several hundreds of atoms in their unit cells DFT becomes inappropriate for such large periodic
systems because of its necessity of immense computational time and power Molecular force field
calculations60 on the other hand lack transferable parameterization especially for transition metal
sites and are hence of limited use to cover the large number of materials to be studied Apparently
a non-orthogonal tight-binding approximation to DFT called density functional tight-binding
(DFTB)28323561 has proven to be computationally feasible and sufficiently accurate This method
computes parameters from DFT calculations of a few molecules per pair of atom types The
parameters are generally transferable to other molecules or solids With self-consistent charge (SCC)
extension DFTB has improved accuracy In order to account weak forces the London dispersion
energy can be calculated separately using empirical potentials and added to total energy Successful
realizations of DFTB include the studies of large-scale systems such as biomolecules62
24
supramolecular compounds63 surfaces64 solids65 liquids and alloys66 Being an approximate method
DFTB needs validation Often one compares DFTB results of selected reference systems with those
obtained with DFT
Before electronic structure calculations of framework materials can be carried out it is necessary to
compute the equilibrium configurations of the atoms Geometry optimization (or energy
minimization) employs a mathematical procedure of optimization to move atoms so as to reduce the
net forces on them to negligible values We adopted the conjugate gradient scheme for the
optimizations using DFTB A primary test for the validation of these optimizations is the comparison
of cell parameters bond lengths bond angles and dihedral angles with the corresponding known
numbers We compared the DFTB optimized MOF COF and PAF geometries with the experimentally
determined or DFT optimized geometries and found that the values agree within 6 error
The angles and intensities of X-ray diffraction patterns can produce three-dimensional information of
the density of electrons within a crystal This can provide a complete picture of atomic positions
chemical bonds and disorders if any Hence simulating powder X-ray diffraction (PXRD) patterns of
optimized geometries and comparing them with experimental patterns minimize errors in the crystal
model Our focus on this matter directed us to identify the layer-offsets of 2D COFs for the first time
In the case of 3D COFs excellent correlations were generally observed between experimental and
simulated patterns Slight differences in the intensities of some of them were due to the presence of
solvents in the crystals as they were reported in the experimental articles PAFs as experimentally
being amorphous do not possess XRD comparisons MOFs within DFTB optimization have
undergone some changes especially in the dihedral angles in comparison with experimental
suggestion or DFT optimization This was verified from the differences in the simulated PXRD
patterns Another interesting outcome of XRD comparison was the discovery of P4 symmetry of
templated MOF-2 derivatives (SURMOFs-2) made by Woumlll et al
Bulk moduli which may be expressed as the second derivative of energy with respect to the unit cell
volume can give a sense of mechanical stability Our calculations provide the following bulk moduli
for some materials MOF-5 1534 (1537)67 Cu-BTC 3466 (3517)68 COF-102 206 (170)69 COF-
103 139 (118)69 COF-105 79 (57)69 COF-108 37 (29)69 COF-202 153 (110)69 The data in the
parenthesis give corresponding values found in the literature calculated using force-field methods
The bulk moduli of MOFs are comparable with the results in the literature however COFs show
significant differences Albeit the differences in values each type of calculation shows the trend that
bulk modulus decreases with decreasing mas density increasing pore volume decreasing thickness
of pore walls and increasing distance between connection nodes
25
Formation of framework materials from condensation of reactants may be energetically modeled
COFs are formed from dehydration reactions of boronic acids and hydroxyphenyl compounds The
formation energy calculated from the energies of the products and reactants can indicate energetic
stability Both DFTB for periodic structures and DFT for finite structures predict exothermic formation
of 3D COFs Likewise for 2D COFs such as COF-1 and 5 the formation of monolayers is found to be
endothermic within both the periodic model calculation using DFTB and finite model calculation
using DFT The stacking of layers provides them stability
72 Weak Interactions in 2D Materials
AB stacked natural graphite and ABC stacked rhombohedral graphite are found to be in proportions
of 85 and 15 in nature respectively whereas AA stacking has been observed only in graphite
intercalation compounds On the other hand boron nitride (h-BN) the synthetic product of boric
acid and boron trioxide exists with AA stacking 2D COFs were believed to have high symmetric AA
(eclipsed ndash P6mmm symmetry) stacking as boron nitride (h-BN) An exception was COF-1 which was
considered to have AB (staggered ndash P63mmc) stacking as graphite The AA stacking maximizes the
attractive London dispersion interaction between the layers a dominating term of the stacking
energy At the same time AA stacking always suffers repulsive Coulomb force between the layers
due to the polarized connectors It should be noted that in boron nitride oppositely charged boron
atoms and nitrogen atoms are facing each other The Coulomb interaction has ruled out possible
interlayer eclipse between atoms of alike charges This gave us the notion that 2D COFs cannot
possess layer-eclipsing Based on these facts we have suggested a shift of ~14 Aring between adjacent
layers at which one or two of the edge atoms of the hexagonal rings are situated perpendicular to
the center of adjacent rings In other words some or all of the hexagonal rings in the pore walls
undergo staggering with that of adjacent layers These lattice types were found to be very stable
compared to AA or AB analogues AA stacking was found to have the highest Coulombic repulsion in
each COF Within an error limit the bulk COFs may be either serrated or inclined The interlayer
separations of all the COF stacks are in the range of 30 to 36 Aring and increase in the order AB
serratedinclined AA70 The motivation for proposing inclined stacking for COFs is that the
rhombohedral graphite (ABC stacking) is the inclined version of the layer-offset in natural graphite
(AB stacking) The instability of AA stacking was also suggested by the studies of elastic properties of
COF-1 and COF-5 by Zhou et al52 The shear modulus show negative signs for the vertical alignment of
COF layers while they are small but positive for the offset of layers
The change of stacking should be visible in their PXRD patterns because each space group has a
distinct set of symmetry imposed reflection conditions We simulated the XRD patterns of COFs in
their known and new configurations and on comparison with the experimental spectrum the new as
26
well as AA stacking showed good agreement5470 The slight layer-offsets are only able to make a few
additional peaks in the vicinity of existing peaks and some changes in relative intensities The
relatively broad experimental data covers nearly all the peaks of the simulated spectrum In other
words the broad experimental peaks are explainable with layer-offset
A detailed analysis on the interlayer stacking of HHTP-DPB COF made by Dichtel et al53 is very
complementary53 This work uses molecular mechanics extensively to define potential energy
surfaces of stacking of two layers from eclipsed to staggered covering all possible layer offsets Low
energy region was near to the eclipsed stacking which was then zoomed in to examine using DFT for
higher accuracy The far-eclipsed regions encounter no energy barriers with the near-eclipsed regions
which suggest the unlikeness of forming near-staggered layering Within DFT the eclipsed
geometries are at high energy compared to the slight offsets around AA (~ 13 ndash 25 Aring away) For a
hexagonal COF the energetically preferred offset could be in one of the six hexagonal sectors (or
circle sectors) of central angle 60 The preference of falling into one sector over the others is
arbitrary and may be determined by symmetry set by nature or external parameters Hence the
layering order in bulk could be serrated inclined spiral or purely by random The latter case better
known as turbostratic disorder is then more like a rule for 2D COFs rather than an exception caused
by external forces This also explains why the experimental PXRD patters have relatively broad peaks
The layer-offset not only change the internal pore structure but also affect interlayer exciton and
vertical charge transport in opto-electronic applications
About stacking stability the square COFs are expected not to be different from hexagonal COFs
because the local environment causing the shifts is nearly the same The DFTB based calculations
reported along with the synthesis of electron-transporting 2D-NiPc-BTDA COF supports this notion71
Two shifted configurations ndash at 07 and 14 Aring away from AA ndash appeared to be energetically preferred
over the AA stacking It appears that AA stacking is only possible for boron nitride-like structures
were adjacent layers have atoms with opposite charges in vertical direction
SURMOFs-2 a series of paddlewheel-based 2D MOFs possess a different symmetry than
solvothermally synthesized MOF-2 due to the differences in interlayer interactions Typically the
interaction between the paddlewheels favors P2 or C2 symmetry of bulk MOF-2 rather than the P4
symmetry of SURMOFs-2 Bader charge analysis reveals that electron distribution between the Cu-
paddlewheels is the lowest when they follow P4 symmetry This indicates the smallest possibility of
having a direct Cu-Cu bond between the paddlewheels In monolayer organic linkers undergo no
rotation with respect to metal dimers
27
X-ray diffraction of SURMOFs-2 made by Woumlll et al indicated P4 symmetry and a relatively small
interlayer separation This increases the repulsion between the linkers and enforces them to rotate
The rotations of each linker in a layer are controlled globally The rotated linkers in adjacent layers
increase London dispersion however a paddlewheel-led shift towards any side increases repulsion
thereby locking SURMOF-2 in P4 packing Hence with the special arrangement of rotated linkers the
linker-linker interaction overcomes the paddlewheel-paddlewheel interaction
P4 symmetry opens up the pores more clearly than P2 or C2 Additionally the metal sites that
typically host solvents are inaccessible Furthermore improvements such as functionalizing the linker
may be easily carried out
Based on our calculations on 2D materials we emphasize the role of van der Waals interactions in
determining the layer-to-layer arrangements The promise of reticular chemistry which is the
maintainability of structural integrity of the building blocks in the construction process is partly
broken in the case of SURMOFs-2 This is because due to repulsion the linkers undergo rotation with
respect to the carboxylic parts albeit keeping the topology
73 Structure-Property Relationships
We have studied a variety of materials from the classes of MOFs COFs and PAFs The structural
differences arise from the differences in the constituents andor their arrangements Properties in
general are interlinked with structural specifications Therefore it is beneficial to know the
relationship between the structural parameters and properties
The mass density is an intensive property of a material In the area of nanoporous materials a crystal
with low mass density has advantages in applications involving transport Definitely the mass density
decreases with increasing pore volume Still the number of atoms in the wall and their weights are
important factors The pore size does not relate directly to the atom counts The volume per atom
(inverse of atom density) another intensive property of a material obliquely gives porosity Figure
16 shows the variation of mass density with volume per atom (including the volume of the atom) for
MOFs COFs and PAFs MOFs show higher mass density compared to other materials of identical
atom density implying the presence of heavy elements It is to be noted that Cu-BTC has mass
density 184 times higher than COF-102 The presence of metals in the form of oxide clusters in MOFs
increases the mass density and decreases the volume per atom Note that the low-weighted MOF in
the discussion (MOF-205) is heavier than many of the PAFs and COFs The relatively high mass
density and small volume per atom of COF-300 is due to its 5-fold interpenetration whereas COF-202
has additional tert-butyl groups which do not contribute to the system shape but affect the mass
density and the volume per atom COF-102 and 103 have same topology but different centers (C and
28
Si respectively) The Si centers increases the cell volume and decreases mass density COF-105 (Si
centers) and COF-108 (C centers) additionally differ in their topologies (ctn and bor respectively) It
appears that bor topology expands the material compared to ctn PAFs hold the extreme cases PAF-
phnl has the highest atom and mass densities whereas PAF-qtph has the smallest atom and mass
densities
Figure 16 Mass density as a function of volume per atom for MOFs COFs and PAFs
The bulk modulus indicates the materialsrsquo resistance towards compression which should in principle
decrease with increasing porosity At the same time larger number of atoms making covalent
networks in unit volume should supply larger bulk moduli Thus differences in molecular contents
and architectures give rise to different bulk moduli It is interesting to see how the mechanical
stability of nanoporous materials is related with the atom density We have obtained a correlation
between bulk modulus (B) and volume per atom (VA) in the case of the studied MOFs COFs and PAFs
as follows
29
where k is a constant (see Figure 17) The inverse-volume correlation indicates that the materials
close to the fitting curve have average bond strengths (interaction energy between close atoms)
identical to each other independent of number of bonds bond order and branching Only Cu-BTC
COF-202 and COF-300 show significant differences from the fitted curve Cu-BTC has 17 times larger
bulk modulus compared to COF-102 of similar volume per atom which implies the substantially
higher strength of the bond network resulting from paddlewheel units and tbo topology
Interpenetration decreased the volume per atom however increased bulk modulus through
interlayer interactions in the case of COF-300 The tert-butyl groups in COF-202 do not transmit its
inherent stability to the COF significantly however decreases the volume per atom Comparison
between COF-102 (C centers) and COF-103 (Si centers) discloses that Si centers decrease the
mechanical stability Comparison between COF-105 (ctn) and COF-108 (bor) suggests that ctn
topology possess higher stability This indicates that local angular preferences can amend the
strength of the bulk material
Figure 17 Bulk modulus as a function of volume per atom for MOFs COFs and PAFs
Band gaps of all the studied materials show that they are semiconductors except of Cu-BTC which
has unpaired electrons at the Cu centers Our studies of the density of states (DOS) of bulk MOFs and
the cluster models that have finite numbers of connectors and linkers show that electronic structure
30
stays unchanged when going from cluster to periodic crystal In the case of 2D COFs the DOS of
monolayers and bulk materials were identical Interpenetrated COF-300 showed no difference in the
electronic structure in comparison with the non-interpenetrated structure Based on these results
we may reach into a premature conclusion that electronic structure of a solid is determined by its
constituent bonded network sufficiently large to include all its building units
HOMO-LUMO gap of the building units determine the band gap of a framework material We have
observed that PAFs nearly reproduced the HOMO-LUMO gaps of the linkers 3D COFs that are made
of more than one building unit show that the band gap is slightly smaller than the smallest of the
HOMO-LUMO gaps of the building units For example
TBPM (43 eV) + HHTP (34 eV) COF-105COF-108 (32 eV)
TBPM (43 eV) + TBST (139 eV) COF-202 (42 eV)
TAM (41 eV) + TA (26 eV) COF-300 (23 eV)
The compound names are taken from appendix E Additionally the band gaps decrease with
increasing number of conjugated rings similar to HOMO-LUMO gaps of the linkers
I believe that the studies in the thesis have helped to an extent to understand the structure
stability and properties of different classes of framework materials The benchmark structures we
studied have the essential features of the classes they represent Ab-initio based computational
studies of several periodic structures are exceptional and thus have its place in the literature
31
List of Abbreviations
ADF Amsterdam Density Functional code
BLYP Becke-Lee-Yang-Parr functional
B3LYP Becke 3-parameter Lee Yang and Parr functional
COF Covalent-Organic Framework
CP Coordination Polymer
CTF Covalent-Triazine Framework
DC Dispersion correction
DFT Density Functional Theory
DFTB Density Functional Tight-Binding
DOS Density of States
DOE (US) Department of Energy (United States)
DZP Double-Zeta Polarized basis set
GGA Generalized Gradient Approximation
LCAO Linear Combination of Atomic Orbitals
LPE Liquid Phase Epitaxy
MOF Metal-Organic Framework
PAF Porous Aromatic Framework
PBE Perdew-Burke-Ernzerhof functional
PXRD Powder X-ray Diffraction Pattern
QLDFT Quantized Liquid Density Functional Theory
RCSR Reticular Chemistry Structure Resource
SBU Secondary Building Unit
SCC Self-Consistent Charge
TZP Triple-Zeta Polarized basis set
SURMOF Surface-Metal-Organic Framework
32
List of Figures
Figure 1 ldquoAdamantinerdquo unit of diamond structure and diamond (dia) net 3
Figure 2 CU-BTC MOF and the corresponding tbo net 4
Figure 3 Fundamental building blocks and SBUs of Cu-BTC and extension of SBUs following crystal
structure 5
Figure 4 Schematic diagram of the single pore of MOF and An example of a simple cubic MOF in
which zinc-oxide tetrahedron are linked together by benzene linkers Atom colors blue ndash Zn red ndash O
grey ndash C white ndash H 6
Figure 5 Schematic diagram of interpenetrated and interwoven frameworks 7
Figure 6 Schematic diagram of SURMOF 8
Figure 7 SBUs and topologies of 2D COFs 9
Figure 8 Deconstructed building units their schematic representations and final geometries of
HKUST-1 (Cu-BTC) MOF-5 MOF-177 MOF-205 and MIL-53 11
Figure 9 AA and AB layer stacks of hexagonal layers 13
Figure 10 Stacking energy Es vs shift of adjacent layers for COF-5 Different stacking possibilities and
their energies are also shown 14
Figure 11 Schematic diagram of different building units forming 2D COFs 16
Figure 12 Schematic diagram of tetragonal and triangular building blocks forming lsquoctnrsquo or lsquoborrsquo
topologies 17
Figure 13 Diamond structure and various organic linkers to build up PAFs 18
Figure 14 The building units and schematic diagram of the formation of iso-reticular SURMOF series
20
Figure 15 Energy diagram of the metastable P4 and stable P2 structures 22
Figure 16 Mass density as a function of volume per atom for MOFs COFs and PAFs 28
Figure 17 Bulk modulus as a function of volume per atom for MOFs COFs and PAFs 29
33
References
(1) Morris R E Wheatley P S Angewandte Chemie-International Edition 2008 47 4966
(2) Li J-R Kuppler R J Zhou H-C Chemical Society Reviews 2009 38 1477
(3) Corma A Chemical Reviews 1997 97 2373
(4) Seo J S Whang D Lee H Jun S I Oh J Jeon Y J Kim K Nature 2000 404 982
(5) Lee J Kim J Hyeon T Advanced Materials 2006 18 2073
(6) Stein A Wang Z Fierke M A Advanced Materials 2009 21 265
(7) Velev O D Jede T A Lobo R F Lenhoff A M Nature 1997 389 447
(8) Beck J S Vartuli J C Roth W J Leonowicz M E Kresge C T Schmitt K D Chu C T
W Olson D H Sheppard E W McCullen S B Higgins J B Schlenker J L Journal of the
American Chemical Society 1992 114 10834
(9) Kresge C T Leonowicz M E Roth W J Vartuli J C Beck J S Nature 1992 359 710
(10) Ying J Y Mehnert C P Wong M S Angewandte Chemie-International Edition 1999 38
56
(11) Velev O D Kaler E W Advanced Materials 2000 12 531
(12) Stein A Microporous and Mesoporous Materials 2001 44 227
(13) Tanev P T Pinnavaia T J Science 1996 271 1267
(14) Yaghi O M OKeeffe M Ockwig N W Chae H K Eddaoudi M Kim J Nature 2003
423 705
(15) OKeeffe M Peskov M A Ramsden S J Yaghi O M Accounts of Chemical Research
2008 41 1782
(16) Delgado-Friedrichs O OKeeffe M Journal of Solid State Chemistry 2005 178 2480
(17) Delgado-Friedrichs O Foster M D OKeeffe M Proserpio D M Treacy M M J Yaghi
O M Journal of Solid State Chemistry 2005 178 2533
(18) OKeeffe M Yaghi O M Chemical Reviews 2012 112 675
(19) Tranchemontagne D J L Ni Z OKeeffe M Yaghi O M Angewandte Chemie-
International Edition 2008 47 5136
(20) Hayashi H Cote A P Furukawa H OKeeffe M Yaghi O M Nature Materials 2007 6
501
(21) Cote A P Benin A I Ockwig N W OKeeffe M Matzger A J Yaghi O M Science
2005 310 1166
(22) El-Kaderi H M Hunt J R Mendoza-Cortes J L Cote A P Taylor R E OKeeffe M
Yaghi O M Science 2007 316 268
(23) Rowsell J L C Yaghi O M Microporous and Mesoporous Materials 2004 73 3
(24) Hermes S Zacher D Baunemann A Woell C Fischer R A Chemistry of Materials
2007 19 2168
34
(25) Kuhn P Antonietti M Thomas A Angewandte Chemie-International Edition 2008 47
3450
(26) Ben T Ren H Ma S Cao D Lan J Jing X Wang W Xu J Deng F Simmons J M
Qiu S Zhu G Angewandte Chemie-International Edition 2009 48 9457
(27) Porezag D Frauenheim T Kohler T Seifert G Kaschner R Physical Review B 1995
51 12947
(28) Seifert G Porezag D Frauenheim T International Journal of Quantum Chemistry 1996
58 185
(29) Kohn W Sham L J Physical Review 1965 140 1133
(30) Parr R G Yang W Density-Functional Theory of Atoms and Molecules New York Oxford
University Press 1989
(31) Hohenberg P Kohn W Physical Review B 1964 136 B864
(32) Elstner M Porezag D Jungnickel G Elsner J Haugk M Frauenheim T Suhai S
Seifert G Physical Review B 1998 58 7260
(33) Zhechkov L Heine T Patchkovskii S Seifert G Duarte H A Journal of Chemical
Theory and Computation 2005 1 841
(34) Elstner M Hobza P Frauenheim T Suhai S Kaxiras E Journal of Chemical Physics
2001 114 5149
(35) Oliveira A F Seifert G Heine T Duarte H A Journal of the Brazilian Chemical Society
2009 20 1193
(36) Seifert G Joswig J-O Wiley Interdisciplinary Reviews-Computational Molecular Science
2012 2 456
(37) Heine T Rapacioli M Patchkovskii S Frenzel J Koester A M Calaminici P
Escalante S Duarte H A Flores R Geudtner G Goursot A Reveles J U Vela A Salahub D
R deMon deMon-nano edn deMon-nano 2009
(38) BCCMS B DFTB+ - Density Functional based Tight binding (and more)
(39) Monkhorst H J Pack J D Physical Review B 1976 13 5188
(40) SCM Amsterdam Density Functional 2012
(41) University of Karlsruhe and Forschungszentrum Karlsruhe gGmbH - TURBOMOLE V63
2011 2007
(42) Dovesi R Saunders V R Roetti C Orlando R Zicovich-Wilson C M Pascale F
Civalleri B Doll K Harrison N M Bush I J DrsquoArco P Llunell M CRYSTAL09 Users Manual
University of Torino Torino 2009 2009
(43) Wadt W R Hay P J Journal of Chemical Physics 1985 82 284
(44) Roy L E Hay P J Martin R L Journal of Chemical Theory and Computation 2008 4
1029
35
(45) Ehlers A W Bohme M Dapprich S Gobbi A Hollwarth A Jonas V Kohler K F
Stegmann R Veldkamp A Frenking G Chemical Physics Letters 1993 208 111
(46) Garberoglio G Skoulidas A I Johnson J K Journal of Physical Chemistry B 2005 109
13094
(47) Han S S Mendoza-Cortes J L Goddard W A III Chemical Society Reviews 2009 38
1460
(48) Getman R B Bae Y-S Wilmer C E Snurr R Q Chemical Reviews 2012 112 703
(49) Cote A P El-Kaderi H M Furukawa H Hunt J R Yaghi O M Journal of the American
Chemical Society 2007 129 12914
(50) Wan S Guo J Kim J Ihee H Jiang D Angewandte Chemie-International Edition 2008
47 8826
(51) Wan S Guo J Kim J Ihee H Jiang D Angewandte Chemie-International Edition 2009
48 5439
(52) Zhou W Wu H Yildirim T Chemical Physics Letters 2010 499 103
(53) Spitler E L Koo B T Novotney J L Colson J W Uribe-Romo F J Gutierrez G D
Clancy P Dichtel W R Journal of the American Chemical Society 2011 133 19416
(54) Lukose B Kuc A Heine T Chemistry-a European Journal 2011 17 2388
(55) Uribe-Romo F J Hunt J R Furukawa H Klock C OKeeffe M Yaghi O M Journal of
the American Chemical Society 2009 131 4570
(56) Schmid R Tafipolsky M Journal of the American Chemical Society 2008 130 12600
(57) Patchkovskii S Heine T Physical Review E 2009 80
(58) Shekhah O Wang H Paradinas M Ocal C Schuepbach B Terfort A Zacher D
Fischer R A Woell C Nature Materials 2009 8 481
(59) Li H Eddaoudi M Groy T L Yaghi O M Journal of the American Chemical Society
1998 120 8571
(60) Rappe A K Casewit C J Colwell K S Goddard W A Skiff W M Journal of the
American Chemical Society 1992 114 10024
(61) Frauenheim T Seifert G Elstner M Hajnal Z Jungnickel G Porezag D Suhai S
Scholz R Physica Status Solidi B-Basic Research 2000 217 41
(62) Elstner M Cui Q Munih P Kaxiras E Frauenheim T Karplus M Journal of
Computational Chemistry 2003 24 565
(63) Heine T Dos Santos H F Patchkovskii S Duarte H A Journal of Physical Chemistry A
2007 111 5648
(64) Sternberg M Zapol P Curtiss L A Molecular Physics 2005 103 1017
(65) Zhang C Zhang Z Wang S Li H Dong J Xing N Guo Y Li W Solid State
Communications 2007 142 477
36
(66) Munch W Kreuer K D Silvestri W Maier J Seifert G Solid State Ionics 2001 145
437
(67) Bahr D F Reid J A Mook W M Bauer C A Stumpf R Skulan A J Moody N R
Simmons B A Shindel M M Allendorf M D Physical Review B 2007 76
(68) Amirjalayer S Tafipolsky M Schmid R Journal of Physical Chemistry C 2011 115
15133
(69) Amirjalayer S Snurr R Q Schmid R Journal of Physical Chemistry C 2012 116 4921
(70) Lukose B Kuc A Frenzel J Heine T Beilstein Journal of Nanotechnology 2010 1 60
(71) Ding X Chen L Honsho Y Feng X Saenpawang O Guo J Saeki A Seki S Irle S
Nagase S Parasuk V Jiang D Journal of the American Chemical Society 2011 133 14510
37
Appendix A
Review Covalently-bound organic frameworks
Binit Lukose and Thomas Heine
To be submitted for publication after revision
Contents
1 Introduction
2 Synthetic achievements
21 Covalent Organic Frameoworks (COFs)
22 Covalent-Triazine Frameworks (CTFs)
23 Porous Aromatic Frameworks (PAFs)
24 Schemes for synthesis
25 List of materials
3 Studies of the underlying structure and properties of COFs
4 Applications
41 Gas storage
411 Porosity of COFs
412 Experimental measurements
413 Theoretical preidctions
414 Adsorption sites
415 Hydrogen storage by spillover
42 Diffusion and selectivity
43 Suggestions for improvement
431 Geometry modifications
432 Metal doping
433 Functionalization
5 Conclusions
6 List and pictures of chemical compounds
38
1 Introduction
Nanoporous materials have perfectly ordered voids to accommodate to interact with and to
discriminate molecules leading to prominent applications such as gas storage separation and sieving
catalysis filtration and sensoring1-4 They are defined as porous materials with pore diameters less
than 100 nm and are classified as micropores of less than 2 nm in diameter mesopores between 2
and 50 nm and macropores of greater than 50 nm The internal geometry that determines pore size
and surface area can be precisely engineered at molecular scales Reticular synthetic methods
suggested by Yaghi and co-workers5-7 can accomplish pre-designed frameworks The procedure is to
select rigid molecular building blocks prudently and assemble them into destined networks using
strong bonds
Several types of materials have been synthesized using reticular chemistry concepts One prominent
group which is much appreciated for its credentials is the Metal-Organic Frameworks (MOFs)8-13 in
which metal-oxide connectors and organic linkers are judiciously selected to form a large variety of
frameworks The immense potential of MOFs is facilitated by the fact that all building blocks are
inexpensive chemicals and that the synthesis can be carried out solvothermally The progress in MOF
synthesis has reached the point that some of the MOFs are commercially available Several MOFs of
ultrahigh porosity have also been successfully synthesized notably the non-interpenetrated IRMOF-
74-XI which has a pore size of 98 Aring Scientists are also able to synthesize MOFs from cheap edible
natural products14 Since the discovery of MOFs many other crystalline frameworks have been
synthesized using reticular chemistry such as Metal-Organic Polyhedra (MOP)15-19 Zeolite
Imidazolate Frameworks (ZIFs)20-22 and Covalent Organic Frameworks (COFs)23-29
COFs are composed of covalent building blocks made of boron carbon oxygen and hydrogen in
many cases also including nitrogen or silicon stitched together by organic subunits The atoms are
held together by strong covalent bonds Depending on the selection of building blocks the COFs may
form in two (2D) or three (3D) dimensions Planar building blocks are the constituents of 2D COFs
whereas for the formation of 3D COFs typically tetragonal building blocks are involved High
symmetric covalent linking as it is perceived in reticular chemistry was confirmed for the end
products5
Unlike the case of supramolecular assemblies the absence of noncovalent forces between the
molecular building units endorses exceptional rigidity and stability for COFs They are in general
thermally stable above 4000 C Also in COFs the stitching of molecular building units guarantees an
39
increased order and allows control over porosity and composition Without any metals or other
heavy atoms present they exhibit low mass densities This gives them advantages over MOFs in
various applications for example higher gravimetric capacities for gas storage3031 The lowest
density crystal known until today is COF-10824 In addition COFs exhibit permanent porosity with
specific surface areas that are comparable to MOFs and hence larger than in zeolites and porous
silicates
MOF and COF crystals possess long range order although COFs have been achieved so far only at the
μm scale Reversible solvothermal condensation reactions are credited for the high order of
crystallinity Other new framework materials such as Covalent Triazine Frameworks (CTFs) and
Porous Aromatic Frameworks (PAFs) differed in ways of synthesis The former was prepared by
ionothermal polymerization while the latter was produced by coupling reactions PAFs lack long
range order in the crystals as a result of the irreversible synthesis Nevertheless many of the
materials are promisingly good for applications In this review we intend to discuss the synthetic
achievements of COF CTFs and PAFs and studies on their structure properties and prominent
applications
For easy understanding of the atomic geometry of a material the reader is referred to Table 1 which
gives the names of the framework materials the reactants and the synthesis scheme Figure 5 shows
the reactants and Figure 4 shows the reaction schemes Example reactions are given in Figure 1-3
Abbreviations of each chemical compound are given in Section 6
2 Synthetic achievements
21 Covalent Organic Frameworks (COFs)
In the year 2005 Yaghi et al23 achieved the crystallization of covalent organic molecules in the form
of periodic extended layered frameworks The condensation of discrete molecules of different sizes
enabled the synthesis of micro- and mesoporous crystalline COFs27 The selection of molecules with 2
and 3 connection sites for synthesis led to the formation of hexagonal layered geometries32 Yaghi et
al have synthesized half a dozen three-dimensional crystalline COFs solvothermally using tetragonal
building blocks containing 4 connection sites since 2007242829 Figure 1 shows a few examples of 2D
and 3D COFs formed by the self-condensation of boronic acids such as BDBA TBPM and TBPS and co-
condensation of the same boronic acids with HHTP
40
Figure 1 Solvothermal condensation (dehydration) of monomers to form 2D and 3D COFs Atom colors are boron yellow oxygen red yellow (big) silicon grey carbon
Alternate synthetic procedures were also exploited for production and functionalization of COFs
Lavigne et al33 showed that the synthesis of COFs using polyboronate esters is facile and high-yielded
41
Boronate esters often contain multiple catechol moieties which are prone to oxidation and are
insoluble in organic solvents Dichtel et al3435 presented Lewis acid-catalysis procedures to form
boronate esters of catechols and arylboronic acids without subjecting to oxidation Cooper et al36
successfully utilized microwave heating techniques for rapid production (~200 times faster than
solvothermal methods) of both 2D and 3D COFs The syntheses of phthalocyanine3437 or porphyrin38
based square COFs have been reported in literature The latter was noticed for its time-dependent
crystal growth which also affects its pore parameters
Abel et al39 reported the ability to form COFs on metallic surfaces COF surfaces (SCOFs) have been
formed by molecular dehydration of boroxine and triphenylene on a silver surface Albeit with some
defects the materials showed high temperature stability allowing to proceed with functionalization
Lackinger et al40 realized the role of substrates on the formation of surface COFs from the surface-
generated radicals Halogen substituted polyaromatic monomers can be sublimed onto metal
substrates and ultimately turned into a COF after homolysis and intermolecular colligation
Chemically inert graphite surfaces cannot support the homolysis of covalent carbon-halogen bonds
and thus cannot initiate the subsequent association of radicals COF layers can be formed onto
Cu(111)40 and Au(111)41 surfaces but with lower crystal ordering Beton et al41 deposited the
monomers on annealed Au(111) surfaces which supplied surface mobility for the monomers and
subsequently increased porosity and surface coverage of the covalent networks Unlike the bulk form
at the surface the monomers were self-organized onto polygons with 4 to 8 edges Such a template
was able to organize a sublimed layer of C60 fullerenes the number of C60 molecules adsorbed in a
cavity was correlated to the size of the polygon
In 2009 Sassi et al42 studied the problem of crystal growth of COFs on substrates They calculated
the reaction mechanism of 14-benzenediboronic acid (BDBA) on Ag(111) surface self-condensation
of BDBA molecules in solvents leads to the formation of two-dimensional (the planar) stacked COF-1
For the surface COFs the study using Density Functional Theory reveals that there are neither
preferred adsorption sites for the molecules nor a charge transfer between the molecules and the
surface Hence the electronic structure of the molecules remains unchanged and the role of the
metal surface is limited in providing the planar nature for the polymer The free reaction enthalpy
(Gibbs free energy) difference of the reaction ∆G = ∆H ndash T∆S is approximated using the harmonic
approximation taking into account the geometrical restrictions of the metal surface and the entropic
contributions of the released water molecules As result the formation of SCOF-1 has been found to
be exothermic if more than six monomers are involved Ourdjini et al43 reported the polymerization
of 14-benzenediboronic acid (BDBA) on different substrates (Ag(111) Ag(100) Au(111) and Cu(111))
and at different source and substrate temperatures to follow how molecular flux and adsorption-
42
diffusion environments should be controlled for the formation of polymers with the smallest number
of structural defects High flux is essential to avoid H-bonded supramolecular assemblies of
molecules and the substrate temperature needs to be optimized to allow the best surface diffusion
and longest residential time of the reactants Achieving these two contradictory conditions together
is a limitation for some substrates eg for copper Silver was found to be the best substrate for
producing optimum quality polymers Controlling the growth parameters is important since
annealing cannot cure defects such as formation of pentagons heptagons and other non-hexagonal
shapes which involved strong covalent bonds
Ditchel et al44 successfully grew COF films on monolayer graphene supported by a substrate under
operationally simple solvothermal conditions The films have better crystallinity compared to COF
powders and can facilitate photoelectronic applications straightaway Zamora et al45 have achieved
exfoliation of COF layers by sonication The thin layered nanostructures have been isolated under
ambient conditions on surfaces and free-standing on carbon grids
A noted characteristic of COFs is its organic functionality Judiciously selected organic units ndash pyrene
and triphenylene (TP) ndash for the production of TP COF can not only harvest photons over a wide range
but also facilitate energy migration over the crystalline solid25 Polypyrene (PPy)-COF that consists of
a single aromatic entity ndash pyrene- is fluorescent upon the formation of excimers and hence undergo
exciton migration across the stacked layers26 A nickel(II)-phthalocyanine based layered square COF
that incorporates phenyl linkers and forms cofacially stacked H-aggregates was reported to absorb
photons over the solar spectrum and transport charges across the stacked layers37 COF-366 and
COF-66 showed excellent hole mobility of 81 and 30 cm2 V-1s-1 surpassing that of all polycrystalline
polymers known until today46 A first example of an electron-transporting 2D COF was reported
recently47 This square COF was built from the co-condensation of nickel(II) phthalocyanine and
electron-deficient benzothiadiazole With the nearly vertical arrangement of its layers it exhibits an
electron mobility of 06 cm2 V-1s-1 The enhanced absorbance is over a wide range of wavelengths up
to 1000 nm In addition it exhibits panchromatic photoconductivity and high IR sensitivity
Lavigne et al48 achieved the condensation of alkyl (methyl ethyl and propyl) substituted organic
building units with boronic acids forming functionalized COFs The alkyl groups contribute for higher
molar adsorption of H2 however the increased mass density of the functionalized COFs cause for
decreased gravimetric storage capacities Boronate ester-linked COFs are stable in organic solvents
however undergo rapid hydrolysis when exposed in water49 In particular the instability of COF-1
upon H2O adsorption shall be mentioned here50 Lanni et al claimed that alkylation could bring
hydrolytic stability into COFs49
43
Functionalization and pore surface engineering in 2D COFs can be improved if azide appended
building blocks are stitched together in click reactions with alkynes51 Control over the pore surface
is made possible by the use of both azide appended and bare organic building units the ratios of
which is matching with the final amount of functionalization in the pore walls The click reactions of
azide groups with alkynes within the pores lead to the formation of triazole-linked groups on the
pore surfaces This strategy also gives the relief of not condensing the already functionalized building
units Jiang et al have asserted eclipsed layering for most of the final COFs using powder X-ray
diffraction pattern Exceptions are the hexagonal functionalized COFs with relatively high azide-
content (ge 75) Their weak XRD signals suggest possible layer-offset or de-lamellation Although
functionalization on pore surfaces exhibits the nature of lowering the surface area the possibility to
add reactive groups within the pores can be utilized for gas-sorption selectivity The authors have
claimed a 16 fold selectivity of CO2 over N2 using 100 acetyl-rich COF-5 compared to the pure COF-5
The range of the click reaction approach is so wide that relatively large chromophores can be
accommodated in the pores thereby making COF-5 fluorescent
Functionalization of 3D COFs has just reported in 201252 Dichtel et al used a monomer-truncation
strategy where one of the four dihydroxyborylphenyl moieties of a tetrahedral building unit was
replaced by a dodecyl or allyl functional group Condensation of the truncated and the pure
tetrahedral blocks in a certain ratio created a functionalized 3D COF The degree of functionalization
was in proportion with the ratio The crystallinity was maintained except for high loading (gt 37) of
truncated monomers The pore volume and the surface area were decreased as a function of loading
level
A heterogeneous catalytic activity of COFs is achieved with an imine-linked palladium doped COF by
enabling Suzuki-Miyaura coupling reaction within the pores53 The COF-LZU1 has a layered geometry
that has the distance between nitrogen atoms in the neighboring linkers in a level (~37 Aring) sufficient
to accommodate metal ions Palladium was incorporated by a post-treatment of the COF PdCOF-
LZU1 was applied to catalyze Suzuki-Miyaura coupling reaction This coupling reaction is generally
used for the facile formation of C-C bonds54 The increased structural regularity and high insolubility
in water and organic solvents are adequate qualities of imine-linked COFs to perform as catalysts
Experiments with the above COF show a broad scope of the reactants excellent yields of the
products and easy recyclability of the catalyst
The comparatively higher thermal stability of COFs is often noted and is explainable with their strong
covalent bonds The reversible dehydrations for the formation of most of the COFs point to their
instability in the presence of water molecules This has been tested and verified for some layered
COFs with boronate esters49 Such COFs are stable as hosts for organometallic molecules55 COF-102
44
framework was found to be stable and robust even in the presence of highly reactive cobaltocenes
The highly stable ferrocenes appear to have an arrangement within the framework led by π-π
interactions
22 Covalent Triazine Frameworks (CTFs)
In the year 2008 Thomas et al56 synthesized a crystalline extended porous nanostructure by
trimerization of polytriazine monomers in ionothermal conditions With the catalytic support of ZnCl2
three aromatic nitrile groups can be trimerized to form a hexagonal C3N3 ring The resulting structure
known as Covalent Triazine-based Frameworks (CTFs) is similar to 2D COFs in geometry and atomic
composition (see Figure 2) Usage of larger non-planar monomers at higher amounts of ionic salts
however led to the formation of contorted structures Interestingly those structures showed
enhanced surface area and pore volume The trimerization of monomers that lack a linear
arrangement of nitrile groups ended up as organic polymer networks Later the same group
introduced mesoporous channels onto a CTF (CTF-2)57 by increasing temperature and ion content
The resulting structure however was amorphous It is concluded that the reaction parameters and
the amount of salt play a crucial role for tuning the porosity and controlling the order of the material
respectively58
Figure 2 Trimerization of nitrile monomers to form CTF-1 Atom colors blue N grey C white H
Ben et al59 followed ionothermal conditions for the targeted synthesis of a 3D framework using
tetraphenylmethane block and a triangular triazine ring The end product was amorphous thermally
stable up to 400 0C and could be applied for the selective sorption of benzene over cyclohexane On a
later synthetic study Zhang et al60 reported that this polymerization could be accomplished in short
45
reaction times under microwave enhanced conditions The template-free high temperature dynamic
polymerization reactions constitute irreversible carbonization reactions coupled with reversible
trimerization of nitriles The irreversibility encounter in these condensation reactions is responsible
for the production of frameworks as amorphous solids6162
An amorphous yet microporous CTF was found to be able to catalyze the oxidation of glycerol with
Pd nanoparticles that are dispersed in the framework63 This solid catalyst appears to be strong
against deactivation and selective toward glycerate compared to Pd supported activated carbon This
is attributed to the relatively high stability of Pd nanoparticles that benefit from the presence of
nitrogen functionalities in CTF An advancement on selective oxidation of methane to methanol at
low temperature is accomplished by using an amorphous bipyridyl-rich platinum-coordinated CTF as
catalyst64
23 Porous Aromatic Frameworks (PAFs)
a notably high surface area (BET surface area of 5600 m2g-1 Langmuir surface area of 7100 m2g-1)65
PAF-1 has been synthesized in a rather simple way using a nickel(0)-catalyzed Yamamoto-type66
Ullmann cross-coupling reaction67(see Figure 3) The material is thermally (up to 520 0C in air) and
hydrothermally (in boiled water for a minimum of seven days) stable The framework exposes all
faces and edges of the phenyl rings to the pores and hence the high surface area can be achieved
while its high stability is inherited from the parent diamond structure The synthesized material was
verified as disordered and amorphous yet with uniform pore diameters The excess H2 uptake
capacity of PAF-1 was measured as 70 wt at 48 bar and 77 K It can adsorb 1300 mg g-1 CO2 at 40
bar and room temperature PAF-1 was also tested for benzene and toluene adsorption
Figure 3 Coupling reaction using halogenated tetragonal blocks for the formation of diamond-like PAF-1 Atom colors red Br grey C white H
46
An attempt to synthesize a diamondoid with quaterphenyl as linker is described recently68 A
tetrahedral unit and a linear linker each of them has two phenyl rings were polymerized using the
Suzuki-Miyaura coupling reaction54 The product (PAF-11) however stayed behind theoretical
predictions and performed poorly pointing to its shortcomings such as short-range order distortion
and interpenetration of phenyl rings Nevertheless the amorphous solid displayed high thermal and
chemical stabilities proneness for adsorbing methanol over water and usability for eliminating
harmful aromatic molecules
PAF-569 is a two-dimensional hexagonal organic framework synthesized using the Yamamoto-type
Ullmann reaction This material is composed only of phenyl rings however lack long range order
because of the distortion of the phenyl rings and kinetics controlled irreversible coupling reaction It
retains a uniform pore diameter and provides high thermal and chemical stability despite its
amorphous nature PAF-5 was suggested for adsorbing organic pollutants at saturated vapour
pressure and room temperature
Co-condensation reactions with piperazine enabled nucleophilic substitution of cyanuric chloride to
form a covalently linked porous organic framework PAF-670 The synthesis in mild conditions yields a
product with uniform morphology and a certain degree of structural regularity Being nontoxic this
material was tested for drug delivery thereby stepping into biomedical applications of covalently
linked organic frameworks Primary tests suggest that PAF-6 could be promoted for drug release for
its permanent porosity and facile functionalization In comparison with MCM-4171 a widely tested
inorganic framework PAF-6 performed equally or even superiorly
24 Schemes for synthesis
The majority of the COFs were synthesized using solvothermal step-by-step condensation
(dehydration) reactions The method incorporates reversibility and is applicable for supplying long
range order in COF materials The reactants generally consist of boronic acids and hydroxy-
polyphenyl (catechol) compounds Extended porous networks are obtained when O-B or O-C bonds
are formed into 5 or 6-membered rings after dehydration (scheme 1) Dehydration may also be
carried out using microwave synthesis Alternately boronic acid can react with catechol acetonide in
presence of a Lewis acid catalyst The molecules link to each other after the liberation of acetone and
water (scheme 2) The utilization of phenyl-attached formyl and amino groups as building units
results in the formation of imines after dehydration (scheme 3) A desired arrangement of molecular
arrays on surfaces can be fulfilled by depositing planar molecules on metallic surfaces followed by
covalent linking using annealing
47
Isocyanate groups follow trimerization to form poly-isocyanurate rings (scheme 4) Cyclotrimerization
of monomers containing nitrile groups or cyanate groups is prone to generate C3N3 rings (scheme 5)
However the ionothermal synthesis of them resulted with amorphous materials Unique bond
formation is often not achieved throughout the material and thus the crystal lacks long-range order
Analogously coupling reactions such as Ullmann and Suzuki-Miyaura give rise to amorphous
products However they are adequate in producing C-C bonds when halogen-substituted compounds
are reacted (scheme 6) Nucleophilic substitution may also be used for framing networks in cases
like reactants are formed into nucleophiles and ions after release of their end-groups (scheme 7)
48
Figure 4 Different schemes reported for the synthesis of covalently-bound organic frameworks
49
25 List of synthesized materials
Table 1 List of synthesized materials in the literature their building blocks synthesis methods measured surface area [m
2 g
-1] pore volume [cm
3 g
-1] and pore size [Aring]
COF Names Reactants Synthesis Surface
Area
Pore
volume
Pore
size
COF-1 BDBA Solvothermal condensation235072
scheme 1
711 62850 032
03650
9
COF-5 BDBA HHTP Solvothermal condensation23
scheme 1
1590 0998 27
Microwave3673 scheme 1 2019
BDBA TPTA Lewis acid catalysis35 TPTA
COF-6 BTBA HHTP Solvothermal condensation27
scheme 1
980 (L) 032 64
COF-8 BTPA HHTP Solvothermal condensation27
scheme 1
1400 (L) 069 187
COF-10 BPDA HHTP Solvothermal condensation27
scheme 1
2080 (L) 144 341
BPDA TPTA Lewis acid catalysis35 scheme 2
COF-18Aring BTBA THB Facile dehydration3348 scheme 1 1263 069 18
COF-16Aring BTBA alkyl-THB
(alkyl = CH3)
Facile dehydration48 scheme 1 753 039 16
COF-14Aring BTBA alkyl-THB
(alkyl = C2H5)
Facile dehydration48 scheme 1 805 041 14
COF-11Aring BTBA alkyl-THB
(alkyl = C3H7)
Facile dehydration48 scheme 1 105 0052 11
50
SCOF-1 BDBA Substrate-assisted synthesis39
scheme 1
SCOF-2 BDBA HHTP Substrate-assisted synthesis39
scheme 1
TP COF PDBA HHTP Solvothermal condensation25
scheme 1
868 079 314
PPy-COF PDBA Solvothermal condensation26
scheme 1
923 188
TBB COF TBB (on Cu(111) and
Ag(110) surfaces)
Surface polymerisation40 scheme
6
TBPB COF TBB (on Au(111)
surface)
Surface polymerisation41 scheme
6
BTP COF BTPA THDMA Solvothermal condensation72
scheme 1
2000 163 40
HHTP-DPB COF DPB HHTP Solvothermal condensation73
scheme 1
930 47
PICU-A DMBPDC Cyclotrimerization74 scheme 4
PICU-B DCF Cyclotrimerization74 scheme 4
COF-LZU1 DAB TFB Solvothermal condensation53
scheme 3
410 054 12
PdCOF-LZU1 COF-LZU1 PdAc Facile post-treatment53 146 019 12
XN3-COF-5 X N3-BDBA (100-X)
BDBA HHTP
Solvothermal condensation51
scheme 1
2160
(X=5)
1865 (25)
1722 (50)
1641 (75)
1421
(100)
1184
1071
1016
0946
0835
295
276
259
258
227
51
XAcTrz-COF-5 XN3-COF-5 PAc Click reaction51 2000
(X=5)
1561 (25)
914 (50)
142 (75)
36 (100)
1481
0946
0638
0152
003
298
243
156
153
125
XBuTrz-COF-5 XN3-COF-5 HP Click reaction51
XPhTrz-COF-5 XN3-COF-5 PPP Click reaction51
XEsTrz-COF-5 XN3-COF-5 MP Click reaction51
XPyTrz-COF-5 XN3-COF-5 PyMP Click reaction51
COF-42 DETH TFB Solvothermal condensation75
scheme 3
710 031 23
COF-43 DETH TFPB Solvothermal condensation75
scheme 3
620 036 38
CTF-1 DCB Ionothermal trimerization56
scheme 5
791 040 12
CTF-2 DCN Ionothermal trimerization57
scheme 5
90 8
mp-CTF-2 2255 151 8
CTF (DCP) DCP Ionothermal trimerization64
scheme 5
1061 0934 14
K2[PtCl4]-CTF DCP K2[PtCl4] Trimerization (scheme 5) +
coordination64
Pt-CTF DCP Pt Trimerization (scheme 5) +
coordination64
PAF-5 TBB Yamamoto-type Ullmann cross-
coupling reaction69 scheme 6
1503 157 166
52
PAF-6 PA CA Nucleophilic substitution70
scheme 7
1827 118
Pc-PBBA COF PcTA BDBA Lewis acid catalysis34 scheme 2 506 (L) 0258 217
NiPc-COF OH-Pc-Ni BDBA Solvothermal condensation37
scheme 1
624 0485 190
XN3-NiPc-COF OH-Pc-Ni X N3-BDBA
(100-X) BDBA
Solvothermal condensation51
scheme 1
XEsTrz-NiPc-
COF
XN3-NiPc-COF MP Click reaction51
ZnP COF TDHB-ZnP THB Solvothermal condensation38
scheme 1
1742 1115 25
NiPc-PBBA COF NiPcTA BDBA Lewis acid catalysis35 scheme 2 776
2D-NiPc-BTDA
COF
OHPcNi BTDADA Solvothermal condensation47
scheme 1
877 22
ZnPc-Py COF OH-Pc-Zn PDBA Solvothermal condensation
scheme 1
420 31
ZnPc-DPB COF OH-Pc-Zn DPB Solvothermal condensation
scheme 1
485 31
ZnPc-NDI COF OH-Pc-Zn NDI Solvothermal condensation
scheme 1
490 31
ZnPc-PPE COF OH-Pc-Zn PPE Solvothermal condensation
scheme 1
440 34
COF-366 TAPP TA Solvothermal condensation46
scheme 3
735 032 12
COF-66 TBPP THAn Solvothermal condensation46
scheme 1
360 020 249
53
COF-102 TBPM Solvothermal condensation24
scheme 1
3472 135 115
Microwave36
scheme 1
2926
COF-102-C12 TBPM trunk-TBPM-R
(R=dodecyl)
Solvothermal condensation52
scheme 1
12
COF-102-allyl TBPM trunk-TBPM-R
(R=allyl)
Solvothermal condensation52
scheme 1
COF-103 TBPS Solvothermal condensation24
scheme 1
4210 166 125
COF-105 TBPM HHTP Solvothermal condensation24
scheme 1
COF-108 TBPM HHTP Solvothermal condensation24
scheme 1
COF-202 TBPM TBST Solvothermal condensation28
scheme 1
2690 109 11
COF-300 TAM TA Solvothermal condensaion29
scheme 3
1360 072 72
PAF-1 TkBPM-X (X=C) Yamamoto-type Ullmann cross-
coupling reaction65 scheme 6
5600
PAF-2 TCM cyclotrimerization59 scheme 5 891 054 106
PAF-3 TkBPM-X (X=Si) Yamamoto-type Ullmann cross-
coupling reaction76 scheme 6
2932 154 127
PAF-4 TkBPM-X (X=Ge) Yamamoto-type Ullmann cross-
coupling reaction76 scheme 6
2246 145 118
PAF-11 TkBPM-X (X=C) BPDA Suzuki coupling reaction68 704 0203 166
54
scheme 6
3 Studies of structure and properties of COFs
31 2D COFs
Following the literature of the years 2005-2010 2D COFs were believed to have high symmetric AA
(eclipsed ndash P6mmm symmetry) stacking as boron nitride (h-BN) [REF] An exception was COF-1
which was considered to have AB (staggered ndash P63mmc) stacking as graphite[REF] The AA stacking
maximizes the attractive London dispersion interaction between the layers an important
contribution to the stacking energy At the same time AA stacking always suffers repulsive Coulomb
force between the layers due to the polarized connectors as the distance between atoms exposing
the same charge is closest It should be noted that in boron nitride boron atoms serve as nearest
neighbors to nitrogen atoms in adjacent layers The Coulomb interaction has ruled out possible
interlayer eclipse between atoms of alike charges Based on these facts we modeled77 a set of 2D
COFs with different possible stacks Each layer was drifted with respect to the neighboring layers in
directions symmetric to the hexagonal pore and we calculated the total energies and their Coulombic
contributions The AA stacking version was found to have the highest Coulombic repulsion in each
COF The lattice types with the layers shifted to each other by ~14 Aring approximately the bond length
between two atoms in a 2D COF were found to be more stable compared to AA or AB In this low-
symmetry configuration the hexagonal rings in the pore walls undergo staggering with that of
adjacent layers Within an error limit the bulk COFs may be either serrated or inclined as shown in
Figure 5 The interlayer separations of all the COF stacks are in the range of 30 to 36 Aring and increase
in the order AB serratedinclined AA32 The motivation for proposing inclined stacking for COFs is
that the rhombohedral graphite (ABC stacking) is the inclined version of the layer-offset in natural
graphite (AB stacking) The instability of AA stacking was also suggested by the studies of elastic
properties of COF-1 and COF-5 by Zhou et al78 The shear modulus show negative signs for the
vertical alignment of COF layers while they are small but positive for the offset of layers
55
Figure 5 Different stacks of a 2D covalent organic framework COF-1 The atoms are shown as Van der Waals spheres
The different stacking modes should in principle be visible in their PXRD patterns as each space
group has a distinct set of symmetry imposed reflection conditions We simulated the XRD patterns
of COFs in their known and new configurations and on comparison with the experimental spectrum
the new as well as AA stacking showed good agreement3279 However the slight layer-offsets in
conjunction with the large number of atoms in the unit cells change the PXRD spectrum only by the
appearance of a few additional peaks all of them in the vicinity of existing signals and by changes in
relative intensities Unfortunately the low resolution of the experimental data does now allow
distinguishing between the different stackings as the broad signals cover all the peaks of the
simulated spectrum
A detailed analysis on the interlayer stacking of HHTP-DPB COF has been made by Dichtel et al73 is
very complementary73 This work uses molecular mechanics extensively to define potential energy
surfaces of stacking of two layers from eclipsed to staggered covering all possible layer offsets The
low energy region was near to the eclipsed stacking which was then zoomed in to examine using DFT
for higher accuracy The far-eclipsed regions encounter no energy barriers with the near-eclipsed
regions which suggest the unlikeness of forming near-staggered layering Within DFT the eclipsed
geometries are at high energy compared to the slight offsets around AA (~ 13 ndash 25 Aring away) For a
hexagonal COF the energetically preferred offset could be in one of the six hexagonal sectors (or
circle sectors) of central angle 60 The preference of falling into one sector over the others is
arbitrary and may be determined by symmetry set by nature or external parameters Hence the
layering order in bulk could be serrated inclined spiral or purely by random The latter case better
known as turbostratic disorder is then more like a rule for 2D COFs rather than an exception caused
by external forces This also explains why the experimental PXRD patters have relatively broad peaks
The layer-offset may not only change the internal pore structure but also affect interlayer exciton
and vertical charge transport in opto-electronic applications
56
Concerning the stacking stability the square 2D COFs are expected not to be different from
hexagonal COFs because the local environment causing the shifts is nearly the same DFTB based
calculations reported along with the synthesis of electron-transporting 2D-NiPc-BTDA COF supports
this notion47 Two shifted configurations ndash at 07 and 14 Aring away from AA ndash appeared to be
energetically preferred over the AA stacking It appears that AA stacking is only possible for boron
nitride-like structures were adjacent layers have atoms with opposite charges in vertical direction In
analogy to BN layers it might be possible to force AA stacking by the introduction of alkalis in
between the layers
32 3D COFs
3D COFs in general are composed of tetragonal and triangular building blocks So far that their
synthesis leads to two high symmetry topologies ndash ctn and bor - in high yields24 The two topologies
differ primarily in the twisting and bulging of their components at the molecular level The
thermodynamic preference of one topology over the other may result from the kinetic entropic and
solvent effects and the relative strain energies of the molecular components It is straight-forward to
state that the effects at the molecular level crucial crucial in the bulk state since transformation from
one net to the other is impossible without bond-breaking There has not been any detailed study on
this matter experimentally or theoretically
Schmid et al8182 have developed force-field parameters from first principles calculations using
Genetic Algorithm approach The parameters developed for cluster models of COF-102 can
reproduce the relative strain energies in sufficient accuracies and be extended to calculations on
periodic crystals The internal twists of aromatic rings within the tetragonal units are different for ctn
and bor nets which lead to stable S4 and unstable D2d symmetries respectively This together with
the bulging of triangular units in the bor net reasoned for energetic preference of ctn over bor for all
boron-based 3D COFs79
The connectivity-based atom contribution method (CBAC) developed by Zhong et al80 can
significantly reduce computational time needed for quantum chemical calculation for framework
charges when screening a large number of MOFs or COFs in terms of their adsorption properties The
basic assumption is that atom types with same bonding connectivity in different MOFsCOFs have
identical charges a statement that follows from the concept of reticular chemistry where the
properties of the molecular building blocks keep their properties in the bulk After assigning the
CBAC charges to the atom types slight adjustment (usually less than 3 ) is needed to make the
frameworks neutral Simulated adsorption isotherms of CO2 CO and N2 in MOFs and COFs show that
CBAC charges reproduce well the results of the QM charges8384 The CBAC method still requires a
57
well-parameterized force field in order to account correctly for adsorption isotherms as the second
important contribution to the host-guest interaction is the London dispersion energy between
framework and adsorbed moleculesG
The elastic constants calculated for the 3D COFs COF-102 and COF-108 show that COF-102 is nearly
five times stronger than COF-10878 While the bulk modulus (B) of COF-102 (B = 216 GPa) exceeds
that of known MOFs such as MOF-5 MOF-177 and MOF-205 (B = 153 101 107 GPa respectively81)
the least dense COF-108 is at the edge of structural collapse (B = 49 GPa) The calculations were
made using Density Functional Tight-Binding (DFTB) method Another report82 based on the same
level of theory possibly with a different parameter set however reveals lower bulk moduli for both
COFs 177 GPa for COF-102 and 005 GPa for COF-108 The bulk moduli for COF-103 and COF-105 are
110 and 33 GPa respectively Recently we have reported the structural properties of 3D COFs The
calculations were made using self-consistent charge (SCC) DFTB method The bulk modulus of each
COF is as follows COF-102 ndash 206 COF-103 ndash 139 COF-105 ndash 79 COF-108 ndash 37 COF-202 ndash 153 and
COF-300 ndash 144 GPa Force field calculations79 provide slightly different values COF-102 ndash 170 COF-
103 ndash 118 COF-105 ndash 57 COF-108 ndash 29 COF-202 ndash 110 GPa Albeit the differences in values each
type of calculation shows the trend that bulk modulus decreases with decreasing mas density and
increasing pore volume and distance between connection nodes One has to note that the high
mechanical stabilities of 3D COFs as suggested by the calculations correspond always to defect-free
crystals Theory is expected therefore to overestimate experimental mechanical stability for real
materials
COFs in general have semiconductor-like band gaps (~ 17 to 40 eV)32 Layer-offset from eclipsed
layering slightly increases the band gap However the Density-Of-States (DOS) of a monolayer is
similar to that of bulk COF The band gap is generally smaller for a COF with larger number conjugate
rings
The thermal influence on the crystal structure of 3D COFs is also intriguing as negative thermal
expansion (NTE) the contraction of crystal volume on heating has been reported for COF-10283 The
studies were performed using molecular dynamics with the force field parameters by Schmid et al84
However the thermal expansion coefficient α = -151 times 10-6 K-1 is much smaller compared to that of
some of the reported MOFs that also show NTE behavior85 The volume-contraction upon the
increase of temperature arises from the tilt of phenyl linkers between B3O3 rings and sp3 carbon
atom in TBPM (figure ) Later Schmid et alreported that COF-102 103 105 108 exhibit NTE
behavior both in ctn and bor topologies whereas COF-202 only in ctn topology79 A typical
application is the realization of controllable thermal expansion composites made of both negative
and positive thermal expansion materials
58
4 Applications
41 Gas storage
The success in the synthesis of COFs was certainly the result of a long-term struggle for complete
covalent crystallization The discovery of COFs coincided with the time when world-wide effort was
paid to develop new materials for gas storage in particular for the development hydrogen tanks for
mobile applications Materials made exclusively from light-weight atoms and with large surface
areas were obviously excellent candidates for these applications The gas storage capacity of porous
materials relies on the success of assembling gas molecules in minimum space This is achieved by
the interaction energy exerted by storage materials on the gas molecules Because the interactions
are noncovalent no significant activation is required for gas uptake and release and hence the so-
called physical adsorption or physisorption is reversible The other type of adsorption ndash chemical
adsorption or chemisorption ndash binds dissociated hydrogen covalently or interstitially at the risk of
losing reversibility As it requires the chemical modification of the host material chemisorption is not
a viable route for COFs and might become possible only through the introduction of reactive
components into the lattice The total amount of gas adsorbed in the pores gives rise to what is
referred to as lsquototal adsorption capacityrsquo This quantity is hard to be assessed experimentally as a
measurement is always subjected to influence of the materials surface and gas present in all parts of
the experimental setup Therefore the lsquoexcess adsorption capacityrsquo is reported in experiment Here
the gas stored in the free accessible volume is subtracted from the total adsorption In experiment
this volume includes the voids in the material as well as any empty space between the sample
crystals This volume is typically determined by measuring the uptake of inert gas molecules (He for
H2 adsorption N2 or Ar for adsorption studies of larger molecules) at room temperature with the
assumption that the host-guest interaction between the material and He can be neglected The
different definitions of adsorption is given in Figure 6
Typically experiments measure excess values and simulations provide total values Quantities of
adsorbed molecules are usually expressed as gravimetric and volumetric units The former gives the
amount per unit mass of the adsorbent while the latter gives the amount per unit volume of the
adsorbent Explicative definitions and terminologies related to gas adsorption can be found
elsewhere86
59
Figure 6 Hydrogen density profile in a pore A denotes the excess A+B the absolute and A+B+C the total adsorption The skeletal volume corresponds to the volume occupied by the framework atomsCourtesy of Dr Barbara Streppel and Dr Michael Hirscher MPI for intelligent systems Stuttgart Germany
411 Porosity of COFs
It is common to inspect the porosity of synthesized nanoporous materials using some inert or simple
gas adsorption measurements Distribution of pore size can be sketched from the
adsorptiondesorption isotherm N2 and Ar isotherms provide an approximation of the pore surface
area pore volume and pore size over the complete micro and mesopore size range Usually the
surface area is calculated using the Brunauer-Emmet-Teller (BET) equation or Langmuir equation
Total pore volume (volume of the pores in a pre-determined range of pore size) can be determined
from either the adsorption or desorption phase The Dubinin-Radushkevic method the T-plot
method or the Barrett-Joyner-Halenda (BJH) method may also be employed for calculating the pore
volume The pore size is often corroborated by fitting non-local density functional theory (NLDFT)
based cylindrical model to the isotherm90-93 In some cases the desorption isotherm is affected by
the pore network smaller pores with narrower channels remain filled when the pressure is lowered
This leads to hysteresis on the adsorption-desorption isotherms The different methods employed for
pore structure analysis are characteristic for micropore filling monolayer and multilayer formations
capillary condensation and capillary filling
For any adsorbate in order to form a layer on pore surface the temperature of the surface must
yield an energy (kBT) much less than adsorbate-surface interaction energy Additionally the absolute
value of the adsorbate-surface binding energy must be greater than the absolute value of the
adsorbate-adsorbate binding energy This avoids clustering of the adsorbate into its three-
dimensional phase
60
At high pressure the adsorption isotherm shows saturation which means that no more voids are left
for further occupation The isotherms show different behaviors characteristic of the pore structure of
the materials There are known classifications based on these differences type I II III IV and V For
COFs and the related materials discussed in this review type I II and IV have been observed (see
Figure 7) As more adsorbate is attracted to the surface after the first surface layer completion one
can expect a bend in the isotherm Type I implies monolayer formation which is typical of
microporosity If the surface sites have significantly different binding energies with the adsorbate
type II behavior is resulted In type IV each ldquokneerdquo where the isotherm bends over and the vapor
pressure corresponds to a different adsorption mechanisms (multiple layers different sites or pores)
and represents the formation of a new layer
Figure 7 Different types of adsorptions in Covalently-bound Organic Frameworks
Argon and nitrogen adsorption measurements at respectively 87 and 77 K exhibit type I isotherms
for COF-1 6 102 and 103 and type IV isotherms for COF-5 8 10 102 and 10387 Smaller pore
diameters of COF-1 and 6 (~ 9 Aring each) are characteristic of monolayer gas formation on the internal
pore surface The same reasons are responsible for the type I character of COF-102 and COF-103
(pore size = 12 Aring each) isotherms albeit with some unusual steps at low pressure The type IV
isotherms of COF-5 8 10 102 and 103 show formation of monolayers followed by their
multiplication within their relatively larger pores (sizes are approximately 27 16 32 12 and 12 Aring
respectively) Other COFs that exhibit type I isotherms for N2 adsorption are the 3D COFs COF-202 (11
Aring) COF-300 (72 Aring) and COF-102-C12 (12 Aring) the alkylated 2D COF series COF-18Aring COF-16Aring COF-14Aring
COF-11Aring (the nomenclature includes their pore sizes) and a 2D square COF NiPc COF (19 Aring)
Isotherms like Type-II were observed for PPy-COF (188 Aring) COF-366 (176 Aring) COF-66 (249 Aring) Pc-
PBBA COF (217 Aring) ZnP-COF (25 Aring) COF-LZU1 (12 Aring) PdCOF-LZU1 (12 Aring) and some of the azide-
appended COFs 50AcTrz-COF-5 (156 Aring) 75AcTrz-COF-5 (153 Aring) 100AcTrz-COF-5 (125 Aring)
50BuTrz-COF-5 (232 Aring) 50PhTrz-COF-5 (212 Aring) 50EsTrz-COF-5 (212 Aring) and 25PyTrz-COF-5
(221 Aring) The majority of the COFs with mesopores exhibit type-IV isotherms They are TP-COF (314
Aring) BTP-COF (40 Aring) COF-42 (23 Aring) COF-43 (38 Aring) HHTP-DPB COF (47 Aring) CTC-COF (226 Aring) NiPc-PBBA
COF ZnPc-Py-COF (31 Aring) ZnPc-DPB-COF (31 Aring) ZnPc-NDI-COF (31 Aring) ZnPc-PPE-COF (34 Aring) 5N3-
61
COF (295 Aring) 25N3-COF (276 Aring) 50N3-COF (259 Aring) 75N3-COF (258 Aring) 100N3-COF (227 Aring)
5AcTrz-COF-5 (298 Aring) 25AcTrz-COF-5 (243 Aring) 5PyTrz-COF-5 (289 Aring) and 10PyTrz-COF-5
(242 Aring)
The synthesized microporous amorphous materials CTF-1 (12 Aring) CTF-DCP (14 Aring) PAF-1 and PAF-2
(106 Aring) exhibit type-I isotherms with N2 adsorption PAF-3 (127 Aring) PAF-4(118 Aring) PAF-5 (157 Aring)
PAF-6 (118 Aring) and PAF-11 showed surprisingly not the typical type I behavior in spite of their
microporosity but type-II isotherms
Many of the mesoporous COFs showed small hysteresis in the adsorption-desorption isotherm
pointing the possibility of capillary condensation Hysteresis was observed for the amorphous
materials in both mirco and meso-pore range Various reasons have been attributed for the observed
hysteresis including the softness of the material and guest-host interactions
412 Gas adsorption experiments
Hydrogen uptake capacities of some boron-based COFs were measured by Yaghi et al87 The excess
gravimetric saturation capacities of COF-1 -5 -6 -8 -10 -102 and -103 at 77 K were found to be 148
358 226 350 392 724 and 705 mg g-1 respectively The saturation pressure increases with an
increase in pore size similar to MOFs88 Pore walls of 2D COFs consist only of edges of connectors
and linkers The fact that faces and edges are largely available for interactions with H2 in 3D
geometries is a reason for their enhanced capacity Total adsorption generally increases without
saturation upon pressure because the difference between the total and the excess capacities
corresponds to the situation in bulk (or in a tank) COFs show in particular good gravimetric
capacities because of their low mass density while volumetric capacities typically do not exceed
those of MOFs The gravimetric hydrogen storage capacities of COF-102 and -103 exceeds 10 wt at
a pressure of 100 bar
COF-18 and its alkylated versions COF-16 -14 and -1 have H2 excess gravimetric capacities 155 144
123 and 122 wt respectively at hellipK and hellipbar
Excess uptake of methane in COFs at room temperature and at 70 bar pressure is as follows COF-1
and -6 up to10 wt COF-5 -6 and -8 between 10 and 15 wt and COF-102 and -103 above 20
wt 87 Isosteric heat of adsorption Qst calculated by fitting the isotherms to virial expansion with
the same temperature are relatively weaker (lt 10 kJmol) for COF-102 and COF-103 at low
adsorption which implies weaker adsorbate-adsorbent interactions In comparison COF-1 and 6
exhibit twice the enthalpy however the overall hydrogen storage capacity was very limited due to
62
the smaller pore volume High Qst at zero-loading is not desirable because the primary excess amount
adsorbed at very low pressures cannot be desorbed practically89
COF-102 and 103 outperform 2D COFs for CO2 storage87 The saturation uptakes at room
temperature were 1200 and 1190 mg g-1 for COF-102 and 103 respectively
A type IV isotherm has been observed for ammonia adsorption in COF-10 inherent to its mesoporous
nature90 The material showed an uptake of 15 mol kg-1 at room temperature and 1 bar the highest
of any nanoporous materials The repeated adsorption-desorption cycles were reported to disrupt
the layer arrangement of this 2D COF Yaghi et al90 were also able to make a tablet of COF-10 crystal
which performed nearly up to the crystalline powder
Not many COFs have been experimentally studied for gas storage applications in spite of high
expectations This has to be understood together as a result of the powder-like polycrystallization of
COFs The enthalpy Qst at low-loading accounted to only 46 kJmol
The excess and total hydrogen uptake capacity of PAF-1 at 48 bar and 77 K reached 7 wt and 10
wt respectively65 PAF-3 and PAF-478_ENREF_73 follow the same diamond topology as PAF-1 the
difference accounts in the central atoms of the tetragonal blocks PAF-1 3 and 4 have C Si and Ge
atoms at the centers respectively At 1 bar the respective adsorption capacities were 166 207 and
150 wt The highest value being found for PAF-3 is attributed to its relatively high enthalpy (66 kJ
mol-1) compared to PAF-1 (54 kJ mol-1) and PAF- 4 (66 kJ mol-1) At high pressure the surface area is
a key factor and because of the lower BET surface area PAF-3 and PAF-4 performs poor At 60 bar
their respective adsorption capacities were 55 and 42 wt For PAF-11 hydrogen uptake was 103
wt at 1 bar68
Methane uptakes of PAF-1 3 and 4 at 1 bar are 13 19 and 13 wt respectively76 Their enthalpies
are respectively 14 15 and 232 kJ mol-1 This suggests that Ge moiety have strong interactions with
methane
CO2 adsorption in PAF-1 at 40 bar and room temperature was 1300 mg g-1 which was slightly more
than the capacity of COF-102 and 10365 PAF-1 3 and 4 at 1 atm pressure could store 48 80 and 51
wt respectively At 273 K and 1 atm they stored 91 153 and 107 wt respectively The storage
capacities at 273 K and 298 K lead to the following zero-loading isosteric enthalpies 156 192 162
kJ mol-1 for PAF-1 3 and 4 respectively The higher uptake of PAF-3 may also be explained by its
relatively higher surface area with CO2 molecules
The selective adsorption of greenhouse gases in PAFs was discussed by Ben et al76 At 273 K and 1
atm pressure PAF-1 selectively adsorbes CO2 and CH4 respectively 38 and 15 times more in
63
amounts than N2 PAF-3 showed extraordinary selectivity of 871 for CO2 over N2 and 301 for CH4
over N2 For PAF-4 the selectivity was 441 for CO2 over N2 and 151 for CH4 over N2 Generally the
uptake of N2 Ar H2 O2 are significantly lower compared to CO2 and NH4 which makes these PAFs
suitable for separating them
Harmful gases like benzene and toluene can be stored in PAF-1 in amounts 1306 mg g-1 (1674 mmol
g-1) and 1357 mg g-1 (1473 mmol g-1) respectively at saturated pressures and room temperature65
In PAF-11 their adsorptions were in amounts of 874 mg g-1 and 780 mg g-1 respectively68 PAF-2 was
accounted to a much lower benzene adsorption (138 mg g-1) at the same conditions59 Adsorption of
cyclohexane vapor in PAF-2 was only 7 mg g-1 While PAF-11 exhibited hydrophobicity it could
accommodate significant amount of methanol (654 mg g-1 or 204 mmol g-1) at room temperature
and saturated pressure68 PAF-5 has been tested for storing organic pollutants69 At room
temperature and saturated pressure PAF-5 adsorbed methanol benzene and toluene in amounts
6531 cm3 g-1 (292 mmol g-1) 26296 cm3 g-1 (117 mmol g-1) and 2582 cm3 g-1 (115 mmol g-1)
respectively Their adsorptions in PAF-5 were deviated from the typical type-I behavior Methanol
exhibited a large slope at the high pressure region corresponding to the relative size of it Zhu G et
al dedicated PAF-6 for biomedical applications With no toxicity observed for PAF-6 over a range of
concentrations adsorption of ibuprofen was measured in it PAF-6 showed an uptake of 350 mg g-1
The release of ibuprofen was also tested in simulated body fluid which showed a release rate of 50
in 5 hours 75 in 10 hours and 100 in almost 46 hours
413 Theoretical predictions
Grand canonical Monte Carlo (GCMC) is a widely used method for the simulation of gas uptakes in
nanoporous materials In GCMC the number of adsorbates and their motions are allowed to change
at constant volume temperature and chemical potential to equilibrate the adsorbate phase The
motions are random guided by Monte Carlo methods and the energies are calculated by force field
methods The details of it may be found in the literature91
Goddart III WA et al92 simulated the uptake capacities of boron-based COFs using force fields derived
from the interactions of H2 and COFs calculated at MP2 level of theory The saturated excess uptakes
of both COF-105 (at 80 bar) and 108 (at 100 bar) were accounted to 10 wt at 77 K which are more
than the uptakes measured in ultrahigh porous MOF-200 205 and 21093 The predictions for other
COFs include 38 wt for COF-1 (AB stacking) 34 wt for COF-5 (AA stacking) 88 wt for COF-102
and 91 wt for COF-103 At 01 bar COF-1 in AB stacking showed highest excess uptake (17 wt )
compared to other COFs in the present discussion Total uptake capacities of the COFs are in the
following order COF-108 (189 wt ) COF-105 (183 wt ) COF-103 (113 wt ) COF-102 (106
64
wt ) COF-5 (55 wt ) and COF-1 (38 wt ) It is worth noticing the higher gravimetric capacities of
COF-108 and 105 which were not measured experimentally They benefit from their lower mass and
higher pore volume In case of volumetric capacity COF-102 (404 gL excess) surpasses the others at
high pressures The total volumetric capacities of the COFs are 499 498 399 395 361 and 338
gL for COF-102 103 108 105 1 and 5 respectively Their additional calculations predicted benzene
rings as favorite locations for H2 molecules
Similar values and trends have been reported by Garberoglio G94 who also reasoned poor solid-fluid
interactions in a large part of free volume for the low volumetric capacity of COF-108 and 105 A
room temperature excess gravimetric uptake of approximately 08 wt was obtained for COF-108
and 105 Quantum diffraction effects were added to the interaction energies of H2 with itself and the
material which were calculated using universal force-field (UFF) With possible overestimation
Klontzas et al30 and Bassem et al95 have reported total gravimetric uptake of 45 and 417 wt
respectively at 300 K for the same COFs Among the boron-based 3D COFs COF-202 exhibited much
smaller uptake capacity at high pressures due to its smaller pore volume95 Lan et al96 predicted a
very small H2 uptake of 152 wt COF-202 at 300K and 100 bar Studies of Yang et al97 clarifies that
pore filling with H2 in 2D COFs is a gradual process rather than being in steps of layer formation
Garberoglio and Vallauri98 pointed out that the relatively higher mass density and lower surface area
per unit volume of 2D COFs are the reasons for their lower uptakes in comparison with 3D COFs The
surface area of hexagonal 2D COFs (AA stacking) averaged around 1000 m2 cm-3 while that of 3D
COFs were about 1500 m2 cm-3
Simulations of H2 adsorption in the perfect crystalline form of PAF-1 and PAF-11 also known as PAF-
302 and PAF-304 respectively are carried out by Zhu et al99 At 77 K H2 excess uptake of PAF-302 (7
wt ) is slightly higher than COF-102 (684 wt )87 and slightly lower than MOF-177 (740 wt )100 At
room temperature PAF-302 exhibited a poor uptake (221 wt at 100 bar) PAF-304 showed
excellent total uptakes 2238 wt at 77 K and 100 bar 653 wt at 298 K and 100 bar These are
highest among all nanoporous materials
Babarao and Jiang studied room-temperature CO2 adsorption in COFs101 At low pressures COFs with
smaller pore volume exhibit a steep rise in their isotherm while at high pressures COF-105 and 108
(82 and 96 mmol g-1 respectively at 30 bar) dominate the others COF-6 has the highest isosteric heat
of adsoption (328 kJmol) hence it made the first steep rise in the volumetric isotherm followed by
COF-8 102 103 10 105 and 108 Correlations of gravimetric and volumetric capacities with mass
density pore volume porosity and surface area have been excellently manifested in this article101
With increasing framework-density gravimetric uptake falls inversely while volumetric capacity
decreases linearly The former rises with free volume while the latter rises and then drops slightly
65
Both of them rise with porosity and surface area Choi et al102 predicted 45 times more CO2 uptake in
COF-108 (3D) compared to COF-5 (2D) Recently Schmid et al79 also have predicted CO2 adsorption
especially for 3D COFs The uptake of COF-202 was almost half of COF-102 and 103 at room
temperature Type IV isotherm has been found for CO2 adsorption in COF-8 and 10 at low
temperatures (lt 200 K)97 Yang and Zhong97 excluded capillary condensation and dissimilar
adsorption sites but multilayer formation as reasons of the stepped behavior (Yang and Zhong
explained this as a consequence of multilayer formation rather than a result of capillary
condensation or dissimilar adsorption sites)
Methane adsorption in 2D and 3D COFs was first calculated by Garberoglio et al Among COF-6 8 and
10 the former which has smaller pore size and higher binding energy with CH4 shows better
volumetric adsorption (~ 120 cm3cm3 at 20 bar) than the others at room temperature at low
pressure region (lt 25 bar) Larger pore size favors COF-10 for higher volumetric adsorption (~ 160
cm3cm3 at 70 bar) at high pressures COF-8 with intermediate pore size shows largest excess amount
of methane adsorbed upon saturation however still lower than two 3D COFs COF-102 and 103
show excess volumetric adsorption of about 180 cm3cm3 at 35 bar This is significantly higher than
the uptake of two other 3D COFs COF-105 and 108 Entropic effects which primarily arise from the
change of dimensionality when the bulk phase of adsorbate transforms to adsorbed phase are
crucial in adsorption Garberoglio94 shows that solid-fluid interaction potential in large part of volume
of COF-105 and 108 is not strong enough to overcome the entropic loss On the other hand these
two COFs exhibit relatively higher gravimetric uptake (720 and 670 cm3g respectively) Goddard III et
al89 predicted total methane uptakes in the following amounts 415 wt in COF108 405 wt in
COF-105 31 wt in COF-103 284 wt in COF-102 196 wt in COF-10(AA) 169 wt in COF-
5(AA) 159 wt in COF-8(AA) 123 wt in COF-6(AA) and 109 wt in COF-1(AB) Yang and Zhong97
have shown that even at low temperatures (100 K) the pore filling of COF-8 with CH4 is rather
gradual than by step-by-step whereas for COF-10 multilayer formation provides a stepped behavior
in the isotherm Alternately Goddard et al pointed out that layer formation coexists with pore-filling
at room temperature89
414 Adsorption sites
First principle calculations on cluster models are typically employed to find favorite adsorption sites
and binding energies of adsorbates within frameworks Benzene rings are found to be a usual
location for H2 where it prefers to orient in perpendicular fashion3086110 Other favorite locations
include B3O3 and C2O2B rings110111 where H2 orients horizontally on the face as well as on the
edge3086 The central ring of HHTP does not provide enough binding to H2 atoms because of small
amount of charges There are some differences in the adsorption energies and favorite sites
66
calculated at different levels of theory Overall the reported binding energies between H2 and any
COF fragment are less than 6 kJ mol-1 In comparison to MOFs COFs do not offer any stronger binding
energy with H2 The advantage of COFs is their low mass density Lan et al103 reported that CO2 is
more preferred to be adsorbed on B3O3 or C2O2B rings than benzene rings Choi et al102 reported that
the oxygen sites are the primary locations for CO2 and CO2 bound COFs are the second strong binding
sites
415 Hydrogen storage by spillover
Hydrogen spillover is an alternate way to store hydrogen113114 It involves dissociation of hydrogen
gas by supported metal catalysts subsequent migration of atomic hydrogen through the support
material and final adsorption in a sorbent Here surface-diffusion of H atoms over the support is
known as primary spillover Secondary spillover is the further diffusion to the sorbent Bridging the
metal part with the sorbent is a practice to enhance the uptake It increases the contact between the
source and receptor and reduces the energy barriers especially in the secondary spillover As the
final sorption is chemisorption surface area of the sorbent is more important than pore volume
Yang and Li measured H2 storage in COF-1 by spillover They used Pt supported on active carbon
(PtAC) as source for hydrogen dissociation Active carbon was the primary receptor and COF-1 the
secondary receptor COF-1 showed much low uptake 128 wt at 77 K and 1 atm 026 wt at 298
K and 100 bar In comparison to MOFs these are very low104 However they have found that the
uptake could be scaled up by a factor of 26 by bridging COF-1 with PtAC by applying carbonization
They also report that heat of adsorption between H and surface sites is more important than surface
area and pore volume in enhancing the net adsorption by spillover
Ganz et al theoretically predicted that the uptake capacities in COFs by hydrogen spillover can be
higher than the measured value116117 Based on ab initio quantum chemistry calculations and
counting the active storage sites they predicted 45 wt for COF-1 in AB stacking and 55 wt for
COF-5 in AA stacking at room temperature and 100 bar
42 Diffusion and Selectivity
Computed self-diffusion coefficients for H2 and CH4 adsorbates in 2D COFs (in AA stacking) are up to
one order of magnitude higher than that in 3D MOFs such as MOF-598 In comparison to nanotubes
the diffusion of H2 molecules in 2D COFs is one order of magnitude slower The differences in
diffusion coefficients are attributed to different pore structures Interactions of the corners of the
hexagonal pore with the fluid present potential barriers along the cylindrical pore axis Diffusion
occurs with jumps after long waiting The height of the barrier is still smaller than in MOFs
67
Homogeneous pore walls and absence of pore corners in nanotubes contribute much less
corrugation to the solid-fluid potential energy surface Increase of self-diffusivities of H2 and CH4 with
increase in pore size of 2D COFs is also shown by Keskin[Ref] Due to the smaller size of H2 its
diffusivity is higher than that of CH4 in any material For reasons of their relative size In a mixture of
the two the self-diffusivity of CH4 increases while that of H2 decreases
Molecular dynamics simulation studies of the diffusion of gas molecules within a 2D COF performed
by Krishna et al105 revealed its relatively lower binding energy of hydrogen molecules within the pore
walls compared to argon CO2 benzene ethane propane n-butane n-pentane and n-hexane
Binding energy prevents the molecules from diffusing through the pore channels They tested if
Knudsen diffusion is applicable to COFs Knudsen diffusion occurs when the molecules frequently
collide with the pore wall This generally happens when the mean free path is larger than the pore
diameter The simulations had been carried out in BTP-COF which accounts a pore diameter of 4 nm
It is found out that the ratio of zero-loading diffusivity with the Knudsen diffusivity has a significant
correlation with isosteric heat of adsorption the stronger the binding energy of the molecules with
the walls the lower the ratio Hydrogen being an exception among the investigated molecules
exhibited the ratio close to unity while others with their isosteric heat of adsorption higher than 10
kJmol-1 have a value close to 01 For a mixture of two gases with significant differences in binding
energies the ratio of self-diffusivities affirms high diffusion selectivity
Selectivity of CH4 over H2 in COF-5 6 and 10 is studied by Keskin106 Narrow pores support the
selective adsorption of CH4 over H2 however they suffer from entropic effects at high pressures
which favor H2 adsorption Liu et al107 have reported that the adsorption selectivity of COFs and
MOFs are in general similar up to 20 bar Delta loading or working capacity (usually expressed in
molkg) is an important term often used to show the economics of the selective adsorption This is
defined as the difference in loadings of the preferred gas at adsorption and desorption pressures
Adsorption is usually in the range 1 to 100 bar while desorption in 01 to 1 bar High selectivity and
high working capacity are preferential for practical use COF-6 has higher selectivity among the three
studied COFs but lower working capacity106 Best selectivity-working capacity combination is shown
by COF-5106 Nonetheless it is not better than CuBTC or CPO-27-X (X = Zn Mg)121122 Liu et al107
attributed the high isosteric heat of adsorption of COF-6 and Cu-BTC for their high adsorption
selectivity They also pointed out that the electrostatic contribution of framework charges in COFs
are smaller than in MOFs however needs to be taken into account
While CH4 is strongly adsorbed H2 undergoes faster diffusion Hence the product of adsorption
selectivity (gt1) and diffusion selectivity (gt0 lt1) which is membrane selectivity is smaller than
adsorption selectivity Keskin106 has compared the selectivity of COF-membranes with known
68
membranes The selectivity of COF-10 is similar to MOF-177 and IRMOFs COF-5 and 6 outperform
them It is to be noted that COF-5 is CH4 selective while COF-10 is H2 selective although their
topologies are similar CH4 permeability of COF membranes is much higher than several MOFs and
ZIFs COF-5 and 6 exhibited high membrane selectivity together with high membrane permeability
Some discussions on the adsorption and membrane based selectivity of COF-102 in comparison with
IRMOFs and Cu-BTC for CH4CO2H2 mixtures can be found in ref108 Adsorption selectivity of COF-6
and Cu-BTC in CH4H2 and CH4CO2 mixtures surpass other COFs and MOFs107 sdfalf
43 Suggestions for improvement
The level of achievement made by COFs and related materials yet do not practically meet the
practical requirements of several applications Hence thoughts for improvement primarily focused
on overcoming their limitations and enhancing characteristic parameters Some theoretical
suggestions for improved performances are mainly discussed here
431 Geometric modifications
Functionalities may be improved by utilizing the structural divergence of framework materials
Several examples may be found in the literature for MOFs etc Klontzas et al suggested replacement
of phenyl linkers in COF-102 by multiple aromatic moieties while keeping the topology unchanged to
increase surface area and pore volume109 They used biphenyl triphenyl naphthalene and pyrene
linkers in the new COFs All of them showed enhanced gravimetric capacity compared to the parent
COF One of the modified COF-102 with triphenyl linker showed 267 and 65 wt at 77 K and 300 K
respectively at 100 bar This kind of replacement may affect the adsorption of each adsorbate
differently leading to the alteration of the selective adsorption of one component over the other110
Following the reticular construction scheme we modeled some new 2D COFs by cross-linking some
of the familiar and unfamiliar building blocks in the COF literature32 This showed the structural
divergence of COFs however they exhibited structural and electronic properties analogues to other
2D COFs
Similar modifications in 2D and 3D COFs for high CO2 uptake were made by Choi et al102 DPABA
(diphenylacetylene-44ʹ-diboronic acid) and its tetrahedral version TBPEPM (tetra[4-(4-
dihydroxyborylphenyl)ethynyl]phenyl) in place of BDBA in COF-5 and TBPMTBPS in COF-108COF-
105 respectively were used for new COF designs The new 2D COF promised a saturated CO2 uptake
higher than in COF-102 because of its large pore volume The new 3D COFs were worth for an uptake
twice more than in COF-105 and 108
69
Goddard et al also designed about 14 new COFs by substituting the aromatic rings in the tetragonal
part (TBPM or TBPS) of COF-102 COF-103 COF-105 COF-108 and COF-202 with vinyl groups or alkyl-
functionalized extended or fused aromatic rings111 The new designs adopted their parent topology
and their structural stability was confirmed by analyzing molecular dynamics (MD) simulations at
room temperature Fused aromatic rings and vinyl groups provided relatively lower and the highest
zero-loading isosteric heat of adsorption (Qst) respectively however the room-temperature delivery
amounts of methane in the respective COFs crossed the DOE target at 35 bar111 In the latter
methane-methane interaction compensated Qst on high-loading
Lino M A et al112 theoretically inspected the ability of layered COFs to exhibit structural variants of
layered materials The hexagonal 2D COF layers may be rolled into the form of nanotubes (NT) or
may exist in the form of fullerenes with the inclusion of pentagons One could think of a formula unit
which resembles the carbon atoms in carbon nanotubes (CNT) or fullerenes The NTs in general can
have any chirality although the study included only armchair and zigzag NTs Density Functional
Theory based calculations reveal that COF-1 NTs of diameter greater than 15 Aring is energetically
favored This was concluded from the comparison of strain-energy per formula unit of the COF-1 NTs
with that of small diameter CNTs The strain energies of zigzag and armchair nanotubes show similar
quadratic functions of tube-diameter The Youngrsquos modulus of COF-1 (22) NT is found out to be 120
GPa This is much smaller compared to typical CNTs which have Youngrsquos modulus average around
1100 GPa113 Band gaps of COF layers remain unchanged when they roll up into nanotubes A COF-1-
fullerene built by scaling C60 molecule has large diameter and hence much less strain energy
compared to C60 Babarao and Jiang reported that the predicted CO2 adsorption capacity of COF-1 NT
is similar to that of CNTs101
Balance between mass density and surface area and hence high gravimetric and volumetric
capacities can be achieved by proper utilization of pore volume or proper distribution of pores Choi
et al theoretically predicted high gravimetric and volumetric capacities for H2 in a pillared COF114 A
pyridine molecule vertically placed above a boron atom in the boroxine ring of COF-1 layer is found
energetically favorable however with wrinkling of the layer115 Here nitrogen atom in pyridine forms
a covalent bond with the boron atom This pillaring increases the separation between the layers
exposes the buried faces of boroxine and aromatic rings to pores and hence increases surface area
and free volume Accessible surface area and free volume have been tripled and gravimetric and
volumetric capacities were calculated to be 10 wt and 60 g L-1 respectively at 77 K and 100 bar114
This volumetric capacity is higher than in NU-100116 and MOF-21093 which show ultrahigh surface
area
70
432 Metal doping
Miao Wu et al studied doping of COF-10 with Li and Ca atoms for hydrogen storage117 The metal
dopants transferred charges to substrate which in turn provided large polarization to hydrogen
molecules Similar observation has been made by Lan et al96 for Li atoms in COF-202 However they
showed the tendency to aggregate at high concentration Lan et al extensively studied doping of
positively charged alkali (Li Na and K) alkaline-earth (Be Mg Ca) and transition (Sc and Ti) metals in
COF-102 and 105 for enhanced CO2 capture103 They found that benzene rings and the three outer
rings of HHTP are the favorite sites for all the metal dopants Other interested locations are edges of
benzene rings and oxygen atoms B3O3 C2O2B rings and the central ring of HHTP were the unwanted
areas Lithium showed stability on the favorite locations while sodium and potassium tended to
cluster even at low concentrations Alkaline-earth metals only weakly interacted with the COFs
whereas transition metals chemisorbed in them which would possibly damage the substrate Lithium
is found out to be a good dopant for enhanced gas storage
Doping electropositive metals would be of advantage because they provide stronger binding with H2
(~ 4 kcalmol)103125133134 The effect of counterions in COFs is studied by Choi et al118 They found out
that although the binding energy of LiF is smaller than Li ion F counterion can hold one hydrogen
atom Additionally Li+ and Mg2+ ions preferred to be isolated than aggregated A step further
Goddard et al119 have recently shown that in presence of tetrahydrofuran (THF a solvent) an
electron is transferred from Li to the benzene ring whereas in the case of gas phase the electron
remained in the atom Additionally they suggested ways to remove solvents which are weakly
coordinated to the material Zhu et al110 suggested doping Li+ after replacement of hydrogen atom by
oxygen ion in COF-102 Another suggestion was the replacement of hydrogen atom by O--Li+ group
Both the cases exhibited higher CO2 adsorption than in the case of Li doping below 10 bar At 10 bar
the three cases exhibited almost the same uptake capacity Below 1 bar the doped Li+ provided
stronger interaction with quadrupolar CO2 in comparison with the substituted O--Li+ group The
differences at low pressures are attributed to the differences in the magnitude of the charge of Li
The doped Li+ remained to hold nearly +1 charge whereas the inclusion of O--Li+ to the framework
diminished the charge of Li+ to a moderate amount Doping Li atom added only relatively small
amount of charge to Li
Doping with Li+ could increase the H2 binding energies up to 28 kJmol118 With properly distributed
metal ion dopants the H2 uptake capacity in COF-108 is predicted to be 65 wt Cao et al also
predicted 684 and 673 wt of H2 uptake in Li-doped COF-105 and 108 respectively at room
temperature and 100 bar120 In Li-doped COF-202 the predicted uptake was 438 wt for the same
conditions96 Goddard et al119 observed that high performance of Li doped COFsMOFs at low
71
pressures (1 ndash 10 bar) is a disadvantage when calculating delivery uptake capacity Na doping could
overcome this difficulty as its heat of adsorption is lower than in Li doped materials Their predicted
delivery gravimetric capacities from 1 to 100 bar at room temperature for Li and Na doped COF-102
and 103 were higher than the 2010 DOE target of 45 wt at room temperature
Lan et al described that CO2 is polarized in presence of Li cation and the polarization increases when
Li cation is physisorbed in COF103 Their calculated uptake capacities of lithium-doped COF-102 and
COF-105 at low loading is about eight and four times higher than that of nondoped ones respectively
Also a very high uptake of 2266 mgg was predicted in the doped COF-105 at 40 bar Li-doped COF-
102 and COF-103 also showed higher uptakes of methane ndash almost double the amount ndash compared
to nondoped COFs at room temperature and low pressures (lt 50 bar)121 Binding energy between
doped Li cation and CH4 was calculated to be 571 kcalmol
Li-functionalized COF-105 realized by inserting alkoxide groups provided very high volumetric uptake
of H2 Klontzas et al122 replaced three hydrogen atoms in HHTP by oxygen atoms in order to achieve
the functionalization In spite of the increased weight due to the additional oxygen atoms the COF
exhibited gravimetric capacity of 6 wt at 300 K
Incorporation of lithium tetrazolide group into a new PAF having diamond topology and triphenyl
linkers for enhanced hydrogen adsorption is suggested by Sun et al123 The tetrazolide anion (CHN4-)
interacts with the lithium cation via ionic bonds The anion and the cation together can bind upto 14
hydrogen molecules Two lithium tetrazolide groups were introduced in the middle phenyl ring of
each triphenyl linker and GCMC calculations predicted 45 wt for the network at 233 K and 100 bar
With the intention of increasing the H2 binding energy Li et al chemically implanted metals in the
place of light atoms in COF-108124 They replaced the C2O2B rings in COF-108 by C2O2Al C2N2Al and
C2Mg2N and each new COF showed stability within DFT This kind of doping does not allow
aggregation of metal clusters Their GCMC simulations showed that the replacement strategy could
improve the hydrogen storage capacity at room temperature by a factor of up to three124 Zou et al
suggested that para-substitution of two carbon atoms in benzene rings by two boron atoms can
facilitate charge transfer between the rings and metal dopants125 Their work revealed that
dispersion of metals forbid any clustering and the doping could enhance the H2 uptake capacity
significantly
433 Functionalization
Functionalization of porous frameworks with ether groups for selective CO2 capture is suggested by
Babarao et al126 The electropositive C atom of CO2 interacts mainly with the electronegative O or N
72
atom of the functional groups They studied PAF-1 functionalized with ndashOCH3 ndashNH2 and ndashCH2OCH2ndash
groups in its biphenyl linkers The latter one which is the tetrahydrofuran-like ether-functionalized
PAF-1 showed high adsorption capacity for CO2 and high selectivity for CO2CH4 CO2N2 and CO2H2
mixtures at ambient conditions
5 Conclusions
Successful covalent crystallization resulted as a class of materials Covalent Organic Frameworks This
review portrays different synthetic schemes successful realizations and potential applications of
COFs and related materials The growth in this area is relatively slow and thus promotions are
needed in order to achieve its potential
6 List and pictures of chemical compounds
alkyl-THB Alkyl-1245-tetrahydroxybenzene
BDBA 14-benzenediboronic acid
BPDA 44ʹ-biphenyldiboronic acid
BTBA 135-benzene triboronic acid
BTDADA 14-benzothiadiazole diboronic acid
BTPA 135-benzenetris(4-phenylboronic acid)
CA Cyanuric acid
DAB 14-diaminobenzene
DCB 14-dicyanobenzene
DCF 27-diisocyanate fluorine
DCN 26-dicyanonaphthalene
DCP 26-dicyanopyridine
DETH 25-diethoxyterephthalohydrazole
DMBPDC 33ʹ-dimethoxy-44ʹ-biphenylene diisocyanate
DPB Diphenyl butadyenediboronic acid
73
HP 1-hexyne propiolate
HHTP 23671011-hexahydroxytriphenylene
MP Methyl propiolate
N3-BDBA Azide-appended benzenediboronic acid
NDI Naphthalenediimide diboronic acid
NiPcTA Nickel-phthalocyanice tetrakis(acetonide)
OH-Pc-Ni 2391016172324-octahydroxyphthalocyaninato)nickel(II)
OH-Pc-Zn 2391016172324-octahydroxyphthalocyaninato)zinc
PA Piperazine
Pac 2-propenyl acetate
PcTA Phthalocyanine tetra(acetonide)
PdAc Palladium acetate
PDBA Pyrenediboronic acid
PPE Phenylbis(phenylethynyl) diboronic acid
PPP 3-phenyl-1-propyne propiolate
PyMP (3α13α2-dihydropyren-1-yl)methyl propionate
TA Terephthaldehyde
TAM tetra-(4-anilyl)methane
TAPP Tetra(p-amino-phneyl)porphyrin
TBB 135-tris(4-bromophenyl)benzene
TBPM tetra(4-dihydroxyboryl-phenyl)methane
TBPP Tetra(p-boronic acid-phenyl)porphyrin
TBPS tetra(4-dihydroxyboryl-phenyl)silane
TBST tert-butylsilane triol
74
TCM Tetrakis(4-cyanophenyl)methane
TDHB-ZnP Zinc(II) 5101520-tetrakis(4-(dihydroxyboryl)phenyl)porphyrin
TFB 135-triformylbenzene
TFPB 135-tris-(4-formyl-phenyl)-benzene
THAn 2345-Tetrahydroxy anthracene
THB 1245-tetrahydroxybenzene
THDMA Polyol 2367-tetrahydroxy-910-dimethyl-anthracene
TkBPM Tetrakis(4-bromophenyl)methane
TPTA Triphenylene tris(acetonide)
trunc-TBPM-R R-functionalized truncated TBPM
75
Figure 8 Reactants of Covalently-bound Organic Frameworks
76
Figure 9 (Figure 8 continued)
(1) Morris R E Wheatley P S Angewandte Chemie-International Edition 2008 47 4966 (2) Li J-R Kuppler R J Zhou H-C Chemical Society Reviews 2009 38 1477 (3) Corma A Chemical Reviews 1997 97 2373 (4) Seo J S Whang D Lee H Jun S I Oh J Jeon Y J Kim K Nature 2000 404 982 (5) Yaghi O M OKeeffe M Ockwig N W Chae H K Eddaoudi M Kim J Nature 2003 423 705
77
(6) OKeeffe M Peskov M A Ramsden S J Yaghi O M Accounts of Chemical Research 2008 41 1782 (7) Ockwig N W Delgado-Friedrichs O OKeeffe M Yaghi O M Accounts of Chemical Research 2005 38 176 (8) Li H Eddaoudi M OKeeffe M Yaghi O M Nature 1999 402 276 (9) Chen B L Eddaoudi M Hyde S T OKeeffe M Yaghi O M Science 2001 291 1021 (10) Eddaoudi M Moler D B Li H L Chen B L Reineke T M OKeeffe M Yaghi O M Accounts of Chemical Research 2001 34 319 (11) Eddaoudi M Kim J Rosi N Vodak D Wachter J OKeeffe M Yaghi O M Science 2002 295 469 (12) Chae H K Siberio-Perez D Y Kim J Go Y Eddaoudi M Matzger A J OKeeffe M Yaghi O M Nature 2004 427 523 (13) Furukawa H Kim J Ockwig N W OKeeffe M Yaghi O M Journal of the American Chemical Society 2008 130 11650 (14) Smaldone R A Forgan R S Furukawa H Gassensmith J J Slawin A M Z Yaghi O M Stoddart J F Angewandte Chemie-International Edition 2010 49 8630 (15) Eddaoudi M Kim J Wachter J B Chae H K OKeeffe M Yaghi O M Journal of the American Chemical Society 2001 123 4368 (16) Sudik A C Millward A R Ockwig N W Cote A P Kim J Yaghi O M Journal of the American Chemical Society 2005 127 7110 (17) Sudik A C Cote A P Wong-Foy A G OKeeffe M Yaghi O M Angewandte Chemie-International Edition 2006 45 2528 (18) Tranchemontagne D J L Ni Z OKeeffe M Yaghi O M Angewandte Chemie-International Edition 2008 47 5136 (19) Lu Z Knobler C B Furukawa H Wang B Liu G Yaghi O M Journal of the American Chemical Society 2009 131 12532 (20) Park K S Ni Z Cote A P Choi J Y Huang R Uribe-Romo F J Chae H K OKeeffe M Yaghi O M Proceedings of the National Academy of Sciences of the United States of America 2006 103 10186 (21) Hayashi H Cote A P Furukawa H OKeeffe M Yaghi O M Nature Materials 2007 6 501 (22) Banerjee R Furukawa H Britt D Knobler C OKeeffe M Yaghi O M Journal of the American Chemical Society 2009 131 3875 (23) Cote A P Benin A I Ockwig N W OKeeffe M Matzger A J Yaghi O M Science 2005 310 1166 (24) El-Kaderi H M Hunt J R Mendoza-Cortes J L Cote A P Taylor R E OKeeffe M Yaghi O M Science 2007 316 268 (25) Wan S Guo J Kim J Ihee H Jiang D Angewandte Chemie-International Edition 2008 47 8826 (26) Wan S Guo J Kim J Ihee H Jiang D L Angewandte Chemie-International Edition 2009 48 5439 (27) Cote A P El-Kaderi H M Furukawa H Hunt J R Yaghi O M Journal of the American Chemical Society 2007 129 12914 (28) Hunt J R Doonan C J LeVangie J D Cote A P Yaghi O M Journal of the American Chemical Society 2008 130 11872 (29) Uribe-Romo F J Hunt J R Furukawa H Klock C OKeeffe M Yaghi O M Journal of the American Chemical Society 2009 131 4570 (30) Klontzas E Tylianakis E Froudakis G E Journal of Physical Chemistry C 2008 112 9095 (31) Tylianakis E Klontzas E Froudakis G E Nanotechnology 2009 20 (32) Lukose B Kuc A Frenzel J Heine T Beilstein Journal of Nanotechnology 2010 1 60
78
(33) Tilford R W Gemmill W R zur Loye H C Lavigne J J Chemistry of Materials 2006 18 5296 (34) Spitler E L Dichtel W R Nature Chemistry 2010 2 672 (35) Spitler E L Giovino M R White S L Dichtel W R Chemical Science 2011 2 1588 (36) Campbell N L Clowes R Ritchie L K Cooper A I Chemistry of Materials 2009 21 204 (37) Ding X Guo J Feng X Honsho Y Guo J Seki S Maitarad P Saeki A Nagase S Jiang D Angewandte Chemie-International Edition 2011 50 1289 (38) Feng X A Chen L Dong Y P Jiang D L Chemical Communications 2011 47 1979 (39) Zwaneveld N A A Pawlak R Abel M Catalin D Gigmes D Bertin D Porte L Journal of the American Chemical Society 2008 130 6678 (40) Gutzler R Walch H Eder G Kloft S Heckl W M Lackinger M Chemical Communications 2009 4456 (41) Blunt M O Russell J C Champness N R Beton P H Chemical Communications 2010 46 7157 (42) Sassi M Oison V Debierre J-M Humbel S Chemphyschem 2009 10 2480 (43) Ourdjini O Pawlak R Abel M Clair S Chen L Bergeon N Sassi M Oison V Debierre J-M Coratger R Porte L Physical Review B 2011 84 (44) Colson J W Woll A R Mukherjee A Levendorf M P Spitler E L Shields V B Spencer M G Park J Dichtel W R Science 2011 332 228 (45) Berlanga I Ruiz-Gonzalez M L Gonzalez-Calbet J M Fierro J L G Mas-Balleste R Zamora F Small 2011 7 1207 (46) Wan S Gandara F Asano A Furukawa H Saeki A Dey S K Liao L Ambrogio M W Botros Y Y Duan X Seki S Stoddart J F Yaghi O M Chemistry of Materials 2011 23 4094 (47) Ding X Chen L Honsho Y Feng X Saenpawang O Guo J Saeki A Seki S Irle S Nagase S Parasuk V Jiang D Journal of the American Chemical Society 2011 133 14510 (48) Tilford R W Mugavero S J Pellechia P J Lavigne J J Advanced Materials 2008 20 2741 (49) Lanni L M Tilford R W Bharathy M Lavigne J J Journal of the American Chemical Society 2011 133 13975 (50) Li Y Yang R T Aiche Journal 2008 54 269 (51) Nagai A Guo Z Feng X Jin S Chen X Ding X Jiang D Nature Communications 2011 2 (52) Bunck D N Dichtel W R Angewandte Chemie-International Edition 2012 51 1885 (53) Ding S-Y Gao J Wang Q Zhang Y Song W-G Su C-Y Wang W Journal of the American Chemical Society 2011 133 19816 (54) Miyaura N Suzuki A Chemical Reviews 1995 95 2457 (55) Kalidindi S B Yusenko K Fischer R A Chemical Communications 2011 47 8506 (56) Kuhn P Antonietti M Thomas A Angewandte Chemie-International Edition 2008 47 3450 (57) Bojdys M J Jeromenok J Thomas A Antonietti M Advanced Materials 2010 22 2202 (58) Kuhn P Forget A Su D Thomas A Antonietti M Journal of the American Chemical Society 2008 130 13333 (59) Ren H Ben T Wang E Jing X Xue M Liu B Cui Y Qiu S Zhu G Chemical Communications 2010 46 291 (60) Zhang W Li C Yuan Y-P Qiu L-G Xie A-J Shen Y-H Zhu J-F Journal of Materials Chemistry 2010 20 6413 (61) Trewin A Cooper A I Angewandte Chemie-International Edition 2010 49 1533 (62) Mastalerz M Angewandte Chemie-International Edition 2008 47 445
79
(63) Chan-Thaw C E Villa A Katekomol P Su D Thomas A Prati L Nano Letters 2010 10 537 (64) Palkovits R Antonietti M Kuhn P Thomas A Schueth F Angewandte Chemie-International Edition 2009 48 6909 (65) Ben T Ren H Ma S Q Cao D P Lan J H Jing X F Wang W C Xu J Deng F Simmons J M Qiu S L Zhu G S Angewandte Chemie-International Edition 2009 48 9457 (66) Yamamoto T Bulletin of the Chemical Society of Japan 1999 72 621 (67) Zhou G Baumgarten M Muellen K Journal of the American Chemical Society 2007 129 12211 (68) Yuan Y Sun F Ren H Jing X Wang W Ma H Zhao H Zhu G Journal of Materials Chemistry 2011 21 13498 (69) Ren H Ben T Sun F Guo M Jing X Ma H Cai K Qiu S Zhu G Journal of Materials Chemistry 2011 21 10348 (70) Zhao H Jin Z Su H Jing X Sun F Zhu G Chemical Communications 2011 47 6389 (71) Mortera R Fiorilli S Garrone E Verne E Onida B Chemical Engineering Journal 2010 156 184 (72) Dogru M Sonnauer A Gavryushin A Knochel P Bein T Chemical Communications 2011 47 1707 (73) Spitler E L Koo B T Novotney J L Colson J W Uribe-Romo F J Gutierrez G D Clancy P Dichtel W R Journal of the American Chemical Society 2011 133 19416 (74) Zhang Y Tan M Li H Zheng Y Gao S Zhang H Ying J Y Chemical Communications 2011 47 7365 (75) Uribe-Romo F J Doonan C J Furukawa H Oisaki K Yaghi O M Journal of the American Chemical Society 2011 133 11478 (76) Ben T Pei C Zhang D Xu J Deng F Jing X Qiu S Energy amp Environmental Science 2011 4 3991 (77) Lukose B Kuc A Heine T Chemistry-a European Journal 2011 17 2388 (78) Zhou W Wu H Yildirim T Chemical Physics Letters 2010 499 103 (79) Amirjalayer S Snurr R Q Schmid R Journal of Physical Chemistry C 2012 116 4921 (80) Xu Q Zhong C Journal of Physical Chemistry C 2010 114 5035 (81) Lukose B Supronowicz B St Petkov P Frenzel J Kuc A B Seifert G Vayssilov G N Heine T Physica Status Solidi B-Basic Solid State Physics 2012 249 335 (82) Assfour B Seifert G Chemical Physics Letters 2010 489 86 (83) Zhao L Zhong C L Journal of Physical Chemistry C 2009 113 16860 (84) Schmid R Tafipolsky M Journal of the American Chemical Society 2008 130 12600 (85) Han S S Goddard W A III Journal of Physical Chemistry C 2007 111 15185 (86) Suh M P Park H J Prasad T K Lim D-W Chemical Reviews 2012 112 782 (87) Furukawa H Yaghi O M Journal of the American Chemical Society 2009 131 8875 (88) Wong-Foy A G Matzger A J Yaghi O M Journal of the American Chemical Society 2006 128 3494 (89) Mendoza-Cortes J L Han S S Furukawa H Yaghi O M Goddard III W A Journal of Physical Chemistry A 2010 114 10824 (90) Doonan C J Tranchemontagne D J Glover T G Hunt J R Yaghi O M Nature Chemistry 2010 2 235 (91) Getman R B Bae Y-S Wilmer C E Snurr R Q Chemical Reviews 2012 112 703 (92) Han S S Furukawa H Yaghi O M Goddard III W A Journal of the American Chemical Society 2008 130 11580 (93) Furukawa H Ko N Go Y B Aratani N Choi S B Choi E Yazaydin A O Snurr R Q OKeeffe M Kim J Yaghi O M Science 2010 329 424 (94) Garberoglio G Langmuir 2007 23 12154 (95) Assfour B Seifert G Microporous and Mesoporous Materials 2010 133 59
80
(96) Lan J Cao D Wang W Journal of Physical Chemistry C 2010 114 3108 (97) Yang Q Zhong C Langmuir 2009 25 2302 (98) Garberoglio G Vallauri R Microporous and Mesoporous Materials 2008 116 540 (99) Lan J H Cao D P Wang W C Ben T Zhu G S Journal of Physical Chemistry Letters 2010 1 978 (100) Furukawa H Miller M A Yaghi O M Journal of Materials Chemistry 2007 17 3197 (101) Babarao R Jiang J Energy amp Environmental Science 2008 1 139 (102) Choi Y J Choi J H Choi K M Kang J K Journal of Materials Chemistry 2011 21 1073 (103) Lan J Cao D Wang W Smit B Acs Nano 2010 4 4225 (104) Wang L Yang R T Energy amp Environmental Science 2008 1 268 (105) Krishna R van Baten J M Industrial amp Engineering Chemistry Research 2011 50 7083 (106) Keskin S Journal of Physical Chemistry C 2012 116 1772 (107) Liu Y Liu D Yang Q Zhong C Mi J Industrial amp Engineering Chemistry Research 2010 49 2902 (108) Keskin S Sholl D S Langmuir 2009 25 11786 (109) Klontzas E Tylianakis E Froudakis G E Nano Letters 2010 10 452 (110) Zhu Y Zhou J Hu J Liu H Hu Y Chinese Journal of Chemical Engineering 2011 19 709 (111) Mendoza-Cortes J L Pascal T A Goddard W A III Journal of Physical Chemistry A 2011 115 13852 (112) Lino M A Lino A A Mazzoni M S C Chemical Physics Letters 2007 449 171 (113) Krishnan A Dujardin E Ebbesen T W Yianilos P N Treacy M M J Physical Review B 1998 58 14013 (114) Kim D Jung D H Kim K-H Guk H Han S S Choi K Choi S-H Journal of Physical Chemistry C 2012 116 1479 (115) Kim D Jung D H Choi S-H Kim J Choi K Journal of the Korean Physical Society 2008 52 1255 (116) Farha O K Yazaydin A O Eryazici I Malliakas C D Hauser B G Kanatzidis M G Nguyen S T Snurr R Q Hupp J T Nature Chemistry 2010 2 944 (117) Wu M M Wang Q Sun Q Jena P Kawazoe Y Journal of Chemical Physics 2010 133 (118) Choi Y J Lee J W Choi J H Kang J K Applied Physics Letters 2008 92 (119) Mendoza-Cortes J L Han S S Goddard W A III Journal of Physical Chemistry A 2012 116 1621 (120) Cao D Lan J Wang W Smit B Angewandte Chemie-International Edition 2009 48 4730 (121) Lan J H Cao D P Wang W C Langmuir 2010 26 220 (122) Klontzas E Tylianakis E Froudakis G E Journal of Physical Chemistry C 2009 113 21253 (123) Sun Y Ben T Wang L Qiu S Sun H Journal of Physical Chemistry Letters 2010 1 2753 (124) Li F Zhao J Johansson B Sun L International Journal of Hydrogen Energy 2010 35 266 (125) Zou X Zhou G Duan W Choi K Ihm J Journal of Physical Chemistry C 2010 114 13402 (126) Babarao R Dai S Jiang D-e Langmuir 2011 27 3451
81
Appendix B
Structural properties of metal-organic frameworks within the density-functional based tight-binding method
Binit Lukose Barbara Supronowicz Petko St Petkov Johannes Frenzel Agnieszka Kuc
Gotthard Seifert Georgi N Vayssilov and Thomas Heine
Phys Status Solidi B 2012 249 335ndash342
DOI 101002pssb201100634
Abstract Density-functional based tight-binding (DFTB) is a powerful method to describe large
molecules and materials Metal-organic frameworks (MOFs) materials with interesting catalytic
properties and with very large surface areas have been developed and have become commercially
available Unit cells of MOFs typically include hundreds of atoms which make the application of
standard density-functional methods computationally very expensive sometimes even unfeasible
The aim of this paper is to prepare and to validate the self-consistent charge-DFTB (SCC-DFTB)
method for MOFs containing Cu Zn and Al metal centers The method has been validated against
full hybrid density-functional calculations for model clusters against gradient corrected density-
functional calculations for supercells and against experiment Moreover the modular concept of
MOF chemistry has been discussed on the basis of their electronic properties We concentrate on
MOFs comprising three common connector units copper paddlewheels (HKUST-1) zinc oxide Zn4O
tetrahedron (MOF-5 MOF-177 DUT-6 (MOF-205)) and aluminum oxide AlO4(OH)2 octahedron (MIL-
53) We show that SCC-DFTB predicts structural parameters with a very good accuracy (with less than
82
5 deviation even for adsorbed CO and H2O on HKUST-1) while adsorption energies differ by 12 kJ
mol1 or less for CO and water compared to DFT benchmark calculations
1 Introduction In reticular or modular chemistry molecular building blocks are stitched together to
form regular frameworks [1] With this concept it became possible to construct framework
compounds with interesting structural and chemical composition most notably metal-organic
frameworks (MOFs) [1ndash10] and covalent organic frameworks (COFs) [11ndash17] The interest in MOFs
and COFs is not limited to chemistry these crystalline materials are also interesting for applications
in information storage [18] sieving of quantum liquids [19] hydrogen storage [20] and for fuel cell
membranes [21ndash23]
Covalent organic framework and MOF frameworks are composed by combining two types of building
blocks the so-called connectors typically coordinating in four to eight sites and linkers which have
typically 2 sometimes also three or four connecting sites Fig 1 illustrates a simplified representation
of the topology of connectors and linkers by using secondary building units (SBUs) (see Fig 1)
Figure 1 The connector and the linker of the frameworks CuBTC (HKUST-1) MOF-177 MOF-5 DUT-6 (MOF-205) and MIL-53 and their respective secondary building units (SBUs) The arrows indicate the connection points of the fragments Red oxygen green carbon white hydrogen yellow copperzinc and blue aluminum
Linkers are organic molecules with carboxylic acid groups at their connection sites which form
bonds to the connectors (typically in a solvothermal condensation reaction) They can carry
functional groups which can make them interesting for applications in catalysis [24] Connectors are
83
either metal organic units as for example the well-known Zn4O(CO2)6 unit of MOF-5 [8] the
Cu2(HCOO)4 paddle wheel unit of HKUST-1 [10] or covalent units incorporating boron oxide linking
units for COFs [11] As the building blocks of MOFs and COFs comprise typically some ten atoms unit
cells quickly get very large In particular if adsorption or dynamic processes in MOFs and COFs are of
interest (super)cells of some 1000 atoms need to be processed While standard organic force fields
show a reasonable performance for COFs [25] the creation of reliable force fields is not
straightforward for MOFs as transferable parameterization of the transition metal sites is an issue
even though progress has been achieved for selected materials [26 27] The difficulty to describe
transition metals in particular if they are catalytically active as in HKUST-1 limits the applicability of
molecular mechanics (MM) even for QMMM hybrid methods [28]
On the other hand the density-functional based tight-binding (DFTB) method with its self-consistent
charge (SCC) extension to improve performance for polar systems is a computationally feasible
alternative This non-orthogonal tight-binding approximation to density-functional theory (DFT)
which has been developed by Seifert Frauenheim and Elstner and their groups [29ndash32] (for a recent
review see Ref [30]) has been successfully applied to a large-scale systems such as biological
molecules [33ndash36] supramolecular systems [37 38] surfaces [39 40] liquids and alloys [41 42] and
solids [43] Being a quantum-mechanical method and hence allowing the description of breaking and
formation of chemical bonds the method showed outstanding performance in the description of
processes such as the mechanical manipulation of nanomaterials [44ndash46] It is remarkable that the
method performs well for systems containing heavier elements such as transition metals as this
domain cannot be covered so far with acceptable accuracy by traditional semi-empirical methods [47
48] DFTB covers today a large part of the elements of the periodic table and parameters and a
computer code are available from the DFTBorg website
Recently we have validated DFTB for its application for COFs [16 17] and have shown that structural
properties and formation energies of COFs are well described within DFTB Kuc et al [49] have
validated DFTB for substituted MOF-5 frameworks where connectors are always the Zn4O(CO2)6 unit
which has been combined with a large variety of organic linkers In this work we have revised the
DFTB parameters developed for materials science applications and validated them for HKUST-1 and
being far more challenging for the interaction of its catalytically active Cu sites with carbon
monoxide (CO) and water The Cu2(HCOO)4 paddle wheel units are electronically very intriguing on a
first note the electronic ground state of the isolated paddle wheel is an antiferromagnetic singlet
state which cannot be described by one Slater determinant and which is consequently not accessible
for KohnndashSham DFT However the energetically very close triplet state correctly describes structure
and electronic density of the system and also adsorption properties agree well with experiment [32
84
50 51] We therefore use DFT in the B3LYP hybrid functional representation as benchmark for DFTB
validations for HKUST-1 added by DFT-GGA calculations for periodic unit cellsWewill show that the
general transferability of the DFTB method will allow investigating structural electronic and in
particular dynamic properties
2 Computational details All calculations of the finite model and periodic crystal structures of MOFs
were carried out using the dispersion-corrected self-consistent density functional based tight-binding
(DC-SCC-DFTB in short DFTB) method [30 52ndash54] as implemented in the deMonNano code [55] Two
sets of parameters have been used the standard SCC-DFTB parameter set developed by Elstner et al
[53] for Zn-containing MOFs for Cu- and Al-containing materials we have extended our materials
science parameter set which has been developed originally for zeolite materials to include Cu For
this element we have used the standard procedure of parameter generation we have used the
minimal atomic valence basis for all atoms including polarization functions when needed Electrons
below the valence states were treated within the frozen-core approximation The matrix elements
were calculated using the local density approximation (LDA) while the short-range repulsive pair-
potential was fitted to the results from DFT generalized gradient approximation (GGA) calculations
For more details on DFTB parameter generation see Ref [30] Periodic boundary conditions were
used to represent frameworks of the crystalline solid state The conjugate-gradient scheme was
chosen for the geometry optimization An atomic force tolerance of 3x10-4 eV Aring-1 was applied
The reference DFT calculations of the equilibrium geometry of the investigated isolated MOF models
were performed employing the Becke three-parameter hybrid method combined with a LYP
correlation functional (B3LYP) [56 57] The basis set associated with the HayndashWadt [58] relativistic
effective core potentials proposed by Roy et al [59] was supplemented with polarization ffunctions
[60] and was employed for description of the electronic structure of Cu atoms The Pople 6-311G(d)
basis sets were applied for the H C and O atoms The calculations were performed with the
Gaussian09 program suite [61] For optimization of the periodic structure of CuBTC at DFT level the
electronic structure code Quickstep [62] which is a part of the CP2K package [63] was used
Quickstep is an implementation of the Gaussian plane wave method [64] which is based on the
KohnndashSham formulation of DFT
We have used the generalized gradient functional PBE [65] and the GoedeckerndashTeterndashHutter
pseudopotentials [64 66] in conjunction with double-z basis sets with polarization functions DZVP-
MOLOPT-SR-GTH basis sets obtained as described in Ref [67] but using two less diffuse primitives
Due to the large unit cell sampling of Brillouin zone was limited only to the G-point Convergence
85
criterion for the maximum force component was set to 45 x 10-3 Hartree Bohr1 The plane wave
basis with cutoff energy of 400 Ry was used throughout the simulations
The initial crystal structures were taken from experiment (powder X-ray diffraction (PXRD) data) The
cell parameters and the atomic positions were fully optimized using conjugate-gradient method at
the DFTB level For the reference DFT calculations only the atomic positions of experimental crystal
structures were minimized The cluster models were cut from the optimized structures and saturated
with hydrogen atoms (see Fig 2 for exemplary systems of Cu-BTC)
3 Results and discussion
31 Cu-BTC We have optimized the crystal structure of Cu-BTC (HKUST-1) [10] using cluster and the
periodic models The structural properties were compared to DFT results (see Table 1) The
geometries were obtained for the activated form (open metal sites) and in the presence of axial
water ligands (see Table 2) as the synthesized crystals often have H2O molecules attached to the
open metal sites of Cu (Fig 2b) We have also studied changes of the cluster model geometry in the
presence of CO (Fig 2c) and the results are also shown in Table 2 We have obtained good agreement
with experimental data as well as with DFT results
Figure 2 (a) Periodic and cluster models of Cu-BTC MOF as used in the computer simulations (b) cluster model with adsorbed H2O and (c) CO molecules
We have also simulated the XRD patterns of Cu-BTC which show very good structural agreement for
peak positions between the experimental and calculated structures The XRD pattern of DFT
optimized structure is nearly a copy of that of the experimental geometry However for DFTB
optimized patterns the relative intensities of some of the peaks notably that of the peaks at 2θ =
138 and 2θ = 158 are different from the experimental XRDs (see Fig 3) The differences in bond
angles between simulation and experiment may affect the sharpness of the signals and hence the
86
intensity To support this argument we have calculated the radial pair distribution function (g(r))
which details the probabilities of all atomndashatom distances in a given cell volume Figure S3 in the
Supporting Information shows g(r) of CundashCu CundashC CundashO CndashC and CndashO atom pairs of DFTB
optimized DFT optimized and experimentally modeled Cu-BTC unit cells The peaks of DFT as well as
DFTB optimized geometries are much broadened whereas the experimentally modeled geometry
has single peaks at the most predicted atomndashatom distances However it is to be noted that DFTB
optimized geometries give slightly shifted peaks compared to those of the DFT optimized geometry
They are broadened around the experimental values The distances between Cu and C atoms which
are not direct neighbors have the largest deviations from the experiment what indicates
shortcomings of the bond angles
Table 1 Selected bond lengths (Aring ) bond angles (⁰) and dihedral angles (⁰) of Cu-BTC optimized at DFTB and DFT level
Bond Type Cluster Model Periodic Model Exp
Cu-Cu 250 (257) 250 (250) 263 Cu-O 205 (198) 202 (198) 195 O-C 134 (133) 133-138 (128) 125
OCuO 836-971 (898) 892-907 (873-937)
891 896
Cu-O-O-Cu plusmn56 ˗ plusmn62 (plusmn02) plusmn04 (plusmn47 ˗ plusmn131) 0
O-C-C-C plusmn38 - plusmn50 (0) plusmn03 - plusmn05 (plusmn06 - plusmn105) 063
Cell paramet a=b=c=27283 (26343)
α=β=γ=90 (90) a=b=c=26343
α=β=γ=90
In detail the bond lengths and bond angles do not change significantly when going from the cluster
to the periodic model confirming the reticular chemistry approach The only exception is the OndashCundash
O bond angle that differs by 4ndash78 between the two systems at both levels of theory
87
Figure 3 Simulated XRD patterns of Cu-BTC MOF with the experimental unit cell parameters and optimized at the DFTB and DFT level of theory
The bond length between Cu atoms is slightly underestimated comparing with experimental (by
maximum 5) and DFT (maximum by 3) results while all other bond lengths are somewhat larger
at DFTB
All bond lengths stay unchanged or become longer in the presence of water molecules The most
striking example is the CundashCu bond where the change reaches 012ndash017 Aring compared with the
structure with activated Cu sites Also bond angles increase by up to 28 when the H2O is present
The CundashO bond length with the oxygen atom from the carboxylate groups is shorter than that with
the oxygen atoms from the water molecules indicating that the latter is only weakly bound to the
copper ions The OndashC distances (~133 Aring) correspond to values between that for the typical single
(142 Aring ) and double (122 Aring) oxygenndashcarbon bond The calculated CringndashCcarboxyl bond lengths of
146A deg indicate that these are typical single sp2ndashsp2 CndashC bonds while experimental findings give a
slightly longer value (15 Aring) for this MOF When comparing dihedral angles CundashOndashOndashCu and OndashCndashCndashC
of the clusters fairly large deviations between DFTB and DFT calculations and experiment are visible
due to the softer potential energy surface associated with these geometrical parameters In periodic
models however the agreement of DFT and DFTB with experiment and with each other is
significantly improved
The unit cell parameters with and without water molecules obtained at the DFTB level overestimate
the experimental data by less than 4 which gives a fairly good agreement if we take into account
high porosity of the material These lattice parameters increase only slightly from 27283 to 27323 Aring
in the presence of water
We have calculated the binding energies of H2O molecules attached to the Cu-metal sites within the
cluster model At the DFTB level Ebind of 40 kJ mol-1 was obtained in a good agreement with the DFT
results (52 kJ mol-1) as well as results from the literature (47 kJ mol-1 [50]) We have also calculated
88
the binding energy of CO which was twice smaller than that of H2O (165 and 28 kJ mol-1 for DFTB
and DFT respectively) suggesting much stronger binding of O atom (from H2O) than C atom (from
CO) While the bond lengths in the MOF structure do not change in the presence of either H2O or CO
the differences in the binding energy come from much longer bond distances (by around 07 Aring) for
CundashC than for CundashO in the presence of CO and water molecules respectively
Furthermore we have studied the electronic properties of periodic and cluster model Cu-BTC by
means of partial density-of-states (PDOS)We have compared especially the Cu-PDOS for s- p- and d-
orbitals (see Fig 4) The results show that the electronic structure stays unchanged when going from
the cluster models to the fully periodic crystal of Cu-BTC Since the copper atoms are of Cu(II) the d-
orbitals are just partially occupied what results in the metallic states at the Fermi level This is a very
interesting result as other ZnndashO-based MOFs are either semiconductors or insulators [49]
Table 2 Selected bond lengths (Aring) bond angles (˚) and dihedral angles (˚) of Cu-BTC optimized at DFTB and DFT levels with water molecules in the axial positions Cluster models are calculated using water and carbon monoxide molecules The adsorption energies Eads (kJ mol-1) of the guest molecules are given as well The DFT data are given in parenthesis
Bond Type Cluster Model +
H2O Periodic
Model+ H2O Cluster Model +
CO
Cu-Cu 267 (266) 262 (260) 250 (260)
Cu-O 205 (197-206) 210 (196-200) 206 (199)
O-C 134 (127) 133 (128) 134 (127)
OCuO 843-955 (889-905)
871-921 (842-930) 842-967 (896)
Cu-O-O-Cu plusmn59 - plusmn76 (plusmn06 ˗ plusmn10)
plusmn02 ˗ plusmn18 (plusmn40 ˗ plusmn136)
plusmn51 - plusmn58 (plusmn70)
O-C-C-C plusmn39 - plusmn53 (plusmn05 - plusmn11)
plusmn03 - plusmn05 (plusmn06 - plusmn105)
plusmn35 - plusmn43 (plusmn12)
Cu-O(H2O) Cu-C(CO) 237 (219) 244 (233-
255) 307 (239)
Eads -4045 (-5200) -1648
(-2800)
32 MOF-5 -177 DUT-6 (MOF-205) and MIL- 53 We have also studied the structural properties
of MOF structures with large surface areas using the SCC-DFTB method Very good agreement with
the experimental data shows that this method is applicable for MOFs of large structural diversity as
well as for coordination polymers based on the MOF-5 framework which has been reported earlier
[49] Tables 3 and 4 show selected bond lengths and bond angles of MOF-5 [68] MOF-177 [69] DUT-
6 (MOF-205) [70 71] and MIL-53 [72] respectively
89
MOFs-5 -177 and DUT-6 (MOF-205) are built of the same connector which is Zn4O(CO2)6
octahedron The difference is in the organic linker 14-benzenedicarboxylate (BDC) 135-
benzenetribenzoates (BTB) and BTB together with 26-naphtalenedicarboxylate (NDC) for MOF-5 -
177 andDUT-6 (MOF-205) respectively (see Fig 5)
Figure 4 DFTB calculations of PDOS of (top left) Cu in Cu-BTC and Zn in MOF-5 (top right) MOF-177 (bottom left) and DUT-6 (MOF-205) (bottom right) In each plot s-orbitals (top) p-orbitals (middle) and d-orbitals (bottom) are shown as computed from periodic (black) and cluster (red) models Cluster models are given in Fig S4
All three MOFs have different topologies due to the organic linkers where the number of
connections is varied or where two different linker types are present MOF-5 is the most simple and
it is the prototype in the isoreticular series (IRMOF-1) [68] it is a simple cubic topology with
threedimensional pores of the same size and the linkers have only two connection points In the
case of MOF-177 the linker is represented by a triangularSBU that means three connection points
are present This results in the topology with mixed (36)-connectivity [69] DUT-6 (MOF-205) has a
much more complicated topology due to two types of linkers The first one (NDC) has just two
90
connection points while the second is the same as in MOF-177 with three connection points One
ZnndashO-based connector is connected to two NDC and four BTB linkers The topology is very interesting
all rings of the underlying nets are fivefold and it forms a face-transitive tiling of dodectahedra and
tetrahedra with a ratio of 13 [73]
Figure 5 The crystal structures in the unit cell representations of (a) MOF-5 (b) MOF-177 (c) DUT-6 (MOF-205) and (d) MIL-53 red oxygen gray carbon white hydrogen and blue metal atom (Zn Al)
MIL-53 is a MOF structure with one-dimensional pores The framework is built up of corner-sharing
AlndashO-based octahedral clusters interconnected with BDC linkers The linkers have again two
connection points MIL-53 shows reversible structural changes dependent on the guest molecules
[74] It undergoes the so-called breathing mode depending on the temperature and the amount of
adsorbed molecules
In this case also the bond lengths and bond angles are slightly overestimated comparing with the
experimental structures but the error does not exceed 3
91
Table 3 Selected bond lengths (Aring) and bond angles (˚) of MOF-5 (424 atoms in unit cell) MOF-177 (808 atoms in unit cell) and DUT-6 (MOF-205) (546 atoms in unit cell) optimized at DFTB level The available experimental data is given in parenthesis [70-73+ Orsquo denotes the central O atom in the Zn-O-octahedron
Bond Type MOF-5 MOF-177 DUT-6
(MOF-205)
Zn-Zn 330 (317) 322-336 (306-330)
325-331 (318)
Zn-Orsquo 202 (194) 202 (193) 202 (194) Zn-O 204 (192) 202-206
(190-199) 202 205 (193)
O-C 128 (130) 128 (131) 128 (125) ZnOrsquoZn 1095 (1095) 1056-1124
(1055 1092) 107-1118 (1084 1100)
OZnO 1083 1108 (1061)
1048 1145 (981-1281)
1046-1112 (1062 1085)
Zn-O-O-Zn 0 (0 plusmn17 plusmn20) plusmn122 - plusmn347 (plusmn176 - plusmn374)
05 - plusmn62 (0 plusmn29)
O-C-C-C 00 (0 - plusmn24) (plusmn 35 - plusmn 336) (plusmn65 - plusmn374)
plusmn04 plusmn22 (0 plusmn174)
Cell paramet a=b=c=26472 (25832) α=β=γ=90 (90)
a=b=37872 (37072) c=3068 (30033) α=β=90 (90) γ=120 (120)
a=b=c=31013 (30353) α=β=γ=90 (90)
We have calculated PDOS for MOF-5 MOF-177 and DUT-6 (MOF-205) (see Fig 4) Their band gaps
calculated to be 40 35 and 31 eV respectively indicate that they are either semiconductors or
insulators We have calculated it for both periodic crystal structure and a cluster containing a metal-
oxide connector and all its carboxylate linkers
Table 4 Selected bond lengths (Aring) and bond angles (˚) of MIL-53 (152 atoms in unit cell) optimized at DFTB level
Bond Type DFTB Exp
Al-Al 341 331 Al-O 189 183-191 O-C 133 129 130 O-Al-O 884 912 886 898 Al-O-Al 1289 1249 Al-O-O-Al 0 0 O-C-C-O 06 03 Cell paramet a=1246
b=1732 c=1365 α=β=γ=90
a=1218 b=1713 c=1326 α=β=γ=90
4 Mechanical properties Due to the low-mass density the elastic constants of porous materials
are a very sensitive indicator of their mechanical stability For crystal structures like MOFs we have
92
studied these by means of bulk modulus (B) B can be calculated as a second derivative of energy
with respect to the volume of the crystal (here unit cell)
The result shows that CuBTC has bulk modulus of 3466 GPa what is in close agreement with
B=3517 GPa obtained using force-field calculations [73 74] and experiment [75] Bulk moduli for the
series of MOFs resulted in 1534 1010 and 1073 GPa for MOF-5 -177 and DUT-6 (MOF-205)
respectively For MOF-5 B=1537 GPa was reported from DFTGGA calculations using plane-waves
[76 77] The results show that larger linkers give mechanically less stable structures what might be
an issue for porous structures with larger voids For MIL-53 we have obtained the largest bulk
modulus of 5369 GPa keeping the angles of the pore fixed
5 Conclusions We have validated the DFTB method with SCC and London dispersion corrections for
various types of MOFs The method gives excellent geometrical parameters compared to experiment
and for small model systems also in comparison with DFT calculations Importantly this statement
holds not only for catalytically inactive MOFs based on the Zn4O(CO2)6 octahedron and organic linkers
which are important for gas adsorption and separation applications but also for catalytically active
HKUST-1 (CuBTC) This is remarkable as the Cu atoms are in the Cu2+ state in this framework DFTB
parameters have been generated and validated for Cu and the electronic structure contains one
unpaired electron per Cu atom in the unit cell which makes the electronic description technically
difficult but manageable within the SCC-DFTB method SCC-DFTB performs well for the frameworks
themselves as well as for adsorbed CO and water molecules
Partial density-of-states calculations for the transition metal centers reveal that the concept of
reticular chemistry ndash individual building units keep their electronic properties when being integrated
to the framework ndash is also valid for the MOFs studied in this work in agreement with a previous
study of COFs [16] The electronic properties computed using the cluster models and the periodic
structure contains the same features and hence cluster models are good models to study the
catalytic and adsorption properties of these materials This is in particular useful if local quantum
chemical high-level corrections shall be applied that require to use finite structures
We finally conclude that we have now a high-performing quantum method available to study various
classes of MOFs of unit cells up to 10000 atoms including structural characteristics the formation
and breaking of chemical and coordination bonds for the simulation of diffusion of adsorbate
molecules or lattice defects as well as electronic properties The parameters can be downloaded
from the DFTBorg website
93
References
[1] E A Tomic J Appl Polym Sci 9 3745 (1965)
2+ M Eddaoudi D B Moler H L Li B L Chen T M Reineke M OrsquoKeeffe and O M Yaghi Acc Chem Res
34 319 (2001)
[3] M Eddaoudi H L Li T Reineke M Fehr D Kelley T L Groy and O M Yaghi Top Catal 9 105 (1999)
[4] B F Hoskins and R Robson J Am Chem Soc 111 5962 (1989)
[5] K Schlichte T Kratzke and S Kaskel Microporous Mesoporous Mater 73 81 (2004) [6] J L C Rowsell A
R Millward K S Park and O M Yaghi J Am Chem Soc 126 5666 (2004)
7+ O M Yaghi M OrsquoKeeffe N W Ockwig H K Chae M Eddaoudi and J Kim Nature 423 705 (2003)
[8] M Eddaoudi J Kim N Rosi D Vodak J Wachter M OrsquoKeeffe and O M Yaghi Science 295 469 (2002)
9+ M OrsquoKeeffe M Eddaoudi H L Li T Reineke and O M Yaghi J Solid State Chem 152 3 (2000)
[10] S S Y Chui SM F Lo J P H Charmant A G Orpen and I D Williams Science 283 1148 (1999)
11+ A P Cote A I Benin N W Ockwig M OrsquoKeeffe A J Matzger and O M Yaghi Science 310 1166 (2005)
[12] A P Cote H M El-Kaderi H Furukawa J R Hunt and O M Yaghi J Am Chem Soc 129 12914 (2007)
[13] H M El-Kaderi J R Hunt J L Mendoza-Cortes A P Cote R E Taylor M OrsquoKeeffe and O M Yaghi
Science 316 268 (2007)
[14] S Wan J Guo J Kim H Ihlee and D L Jiang Angew Chem 120 8958 (2008)
[15] S Wan J Guo J Kim H Ihlee and D L Jiang Angew Chem 121 5547 (2009)
[16] B Lukose A Kuc J Frenzel and T Heine Beilstein J Nanotechnol 1 60 (2010)
[17] B Lukose A Kuc and T Heine Chem Eur J 17 2388 (2011)
[18] D S Marlin D G Cabrera D A Leigh and A M Z Slawin Angew Chem Int Ed 45 77 (2006)
[19] I Krkljus and M Hirscher Microporous Mesoporous Mater 142 725 (2011)
[20] M Hirscher Handbook of Hydrogen Storage (Wiley-VCH Weinheim 2010)
[21] H Kitagawa Nature Chem 1 689 (2009)
[22] M Sadakiyo T Yamada and H Kitagawa J Am Chem Soc 131 9906 (2009)
[23] T Yamada M Sadakiyo and H Kitagawa J Am Chem Soc 131 3144 (2009)
94
[24] K S Jeong Y B Go S M Shin S J Lee J Kim O M Yaghi and N Jeong Chem Sci 2 877 (2011)
[25] R Schmid and M Tafipolsky J Am Chem Soc 130 12600 (2008)
[26] M Tafipolsky S Amirjalayer and R Schmid J Comput Chem 28 1169 (2007)
[27] M Tafipolsky S Amirjalayer and R Schmid J Phys Chem C 114 14402 (2010)
[28] A Warshel and M Levitt J Mol Biol 103 227 (1976)
[29] G Seifert D Porezag and T Frauenheim Int J Quantum Chem 58 185 (1996)
[30] A F Oliveira G Seifert T Heine and H A Duarte J Braz Chem Soc 20 1193 (2009)
[31] D Porezag T Frauenheim T Koumlhler G Seifert and R Kaschner Phys Rev B 51 12947 (1995)
[32] T Frauenheim G Seifert M Elstner Z Hajnal G Jungnickel D Porezag S Suhai and R Scholz Phys
Status Solidi B 217 41 (2000)
[33] M Elstner Theor Chem Acc 116 316 (2006)
Supporting Information
Figure S4 Cluster models as used for the P-DOS calculations in Fig 4 MOF-5 MOF-177 and DUT-6 (MOF-205) (from left to right)
95
Figure S3 Radial pair distribution function (g(r)) of Cu-Cu Cu-C Cu-O C-C and C-O atom-pairs in a Cu-BTC MOF unit cell
96
Appendix C
The Structure of Layered Covalent-Organic Frameworks
Binit Lukose Agnieszka Kuc and Thomas Heine
Chem Eur J 2011 17 2388 ndash 2392
DOI 101002chem201001290
Abstract Covalent-Organic Frameworks (COFs) are a new family of 2D and 3D highly porous and
crystalline materials built of light elements such as boron oxygen and carbon For all 2D COFs an AA
stacking arrangement has been reported on the basis of experimental powder XRD patterns with the
exception of COF-1 (AB stacking) In this work we show that the stacking of 2D COFs is different as
originally suggested COF-1 COF-5 COF-6 and COF-8 are considerably more stable if their stacking
arrangement is either serrated or inclined and layers are shifted with respect to each other by ~14 Aring
compared with perfect AA stacking These structures are in agreement with to date experimental
data including the XRD patterns and lead to a larger surface area and stronger polarisation of the
pore surface
Introduction In 2005 Cocircteacute and co-workers introduced a new class of porous crystalline materials
Covalent Organic Frameworks (COFs)[1] where organic linker molecules are stitched together by
connectors covalent entities including boron and oxygen atoms to a regular framework These
materials have the genuine advantage that all framework bonds represent strong covalent
interactions and that they are composed of light-weight elements only which lead to a very low
mass density[2] These materials can be synthesized solvothermally in a condensation reaction and
97
are composed of inexpensive and non-toxic building blocks so their large-scale industrial production
appears to be possible
Figure 1 Calculated and experimental XRD patterns of all studied COFs (COF-1 top left COF-5 top right COF-6 bottom left COF-8 bottom right)
To date a number of different COF structures have been reported[1ndash3] From a topological
viewpoint we distinguish two- and three-dimensional COFs In two-dimensional (2D) COFs the
covalently bound framework is restricted to 2D layers The crystal is then similar as in graphite or
hexagonal boron nitride composed by a stack of layers which are not connected by covalent bonds
but held together primarily by London dispersion interactions
98
The COF structures have been analyzed by powder X-ray diffraction (PXRD) experiments The
topology of the layers is determined by the structure of the connector and linker molecules and
typically a hexagonal pattern is formed due to the 2m (D3h) symmetry of the connector moieties
The individual layers are then stacked and form a regular crystal lattice With one exception (COF-
1)[1] for all 2D-COFs AA (eclipsed P6mmm) interlayer stacking has been reported[3andashd f g] This
geometrical arrangement maximizes the proximity of the molecular entities and results in straight
channels orthogonal to the COF layers which are known from the literature[1 3a]
The connectors of 2D-COFs include oxygen and boron atoms which cause an appreciable polarization
The AA stacking arrangement maximizes the attractive London dispersion interaction between the
layers which is the dominating term of the stacking energy At the same time AA stacking always
results in a repulsive Coulomb force between the layers due to the polarized connectors It should be
noted that the eclipsed hexagonal boron nitride (h-BN)[4] has boron atoms in one layer serving as
nearest neighbours to nitrogen atoms in adjacent layers (AB stacking) The Coulomb interaction has
ruled out possible interlayer eclipse between atoms of alike charges In this article we aim at
studying the layer arrangements of solvent-free COFs based on the interlayer interactions and the
minimum variance Various lattice types have been considered all significantly more stable than the
reported AA stacked geometry In all 2D COFs we have studied the Coulomb interaction between the
layers leads to a modification of the stacking and shifts the layers by about one interatomic distance
(~14 Aring) with respect to each other (see Figure 1)
Breaking of the highly symmetric layer arrangement has been observed for COF-1 and COF-10 after
removal of guest molecules from pores leading to a turbostratic repacking of pore channels[1 5]
The authors have confirmed the disorder either by the changes in the PXRD spectrum taken before
and after adsorptionndashdesorption cycle[1] or by the changes in the adsorption behavior[5] The
disruption of layer arrangement is attributed to the force exerted by the guest molecules Formation
of poly disperse pores has also been verified[5] The lattice types that we discuss in this article on
the other hand are neither the result of the pressure from any external molecule in the pore nor
having more than one type of pores They are the resultant of minimum variance guided by Coulomb
and London dispersion interactions For the COF models under investigation perfect crystallinity has
been considered
Methods We have studied the experimentally known structures of COF-1 COF-5 COF-6 and COF-8
Structure calculations were carried out using the Dispersion-Corrected Self-Conistent-Charge
Density-Functional-Based Tight-Binding method (SCC-DFTB)[6] DFTB is based on a second-order
expansion of the KohnndashSham total energy in DFT with respect to charge density fluctuations This
does not require large amounts of empirical parameters however maintains all qualities of DFT The
99
accuracy is very much improved by the SCC extension The DFTB code (deMonNano[6a]) has
dispersion correction[6d] implemented to account for weak interactions and was used to obtain the
layered bulk structure of COFs and their formation energies The performance for interlayer
interactions has been tested previously for graphite[6d] All structures correspond to full geometry
optimizations (structure and unit cell) XRD patterns have been simulated using the Mercury
software[7] To allow best comparison with experiment for PXRD simulations we used the calculated
geometry of the layer and of the relative shifts between the layers but experimental interlayer
distances The electrostatic potential has been calculated using Crystal 09[8] at the DFTVWN level
with 6-31G basis set
Results and Discussion
In order to see the favorite stacking arrangement of the layers we have shifted every second layer in
two directions along zigzag and armchair vertices of the hexagonal pores (see Figure 2) The initial
stacking is AA (P6mmm) and the final zigzag shift gives AB stacking (P63mmc) Considering that the
attractive interlayer dispersion forces and repulsive interlayer orbital interactions are different we
have optimized the interlayer separation for each stacking Figure 2 show their total energies
calculated per formula unit that is per established bond between linkers and connectors with
reference to AA stacking for COF-5 and COF-1 In each direction the energy minimum is found close
to the AA stacking position shifted only by about one aromatic C-C bond length (~14 Aring) so that
either connector or linker parts become staggered with those in the adjacent layers leading to a
stacking similar to that of graphite for either connector (armchair shift) or linker (zigzag shift) For
COF-5 the staggering at the connector or linker depends whether the shift is armchair or zigzag
respectively while for COF-1 the zigzag shift alone can give staggering in both the aromatic and
boron-oxygen rings
The low-energy minima in the two directions are labeled following the common nomenclature as
zigzag and armchair respectively Both shifts result in a serrated stacking order (orthorhombic
Cmcm) which shows a significantly more attractive interlayer Coulomb interaction than AA stacking
(see Table 1) while most of the London dispersion attraction is maintained and consequently
stabilizes the material Still this configuration can be improved if we consider inclined stacking
(Figure 1) that is the lattice vector pointing out of the 2D plane is not any longer rectangular
reducing the lattice symmetry from Cmcm (orthorhombic) to C2m (monoclinic)
Besides we have calculated the monolayer formation energy (condensation energy Ecb) from the
total energies of the monolayer and of the individual building blocks and the stacking formation
energy from the total energies of the bulk structure and of the monolayer for four selected COFs The
100
COF building blocks are the same as those used for their synthesis[1 3a] (BDBA for COF-1 BDBA and
HHTP for COF-5 BTBA and HHTP for COF-6 BTPA and HHTP for COF-8) The final energies given per
formula unit that is per condensed bond are given in Table 1 The serrated and inclined stacking
structures are energetically more stable than AA and AB Interestingly within our computational
model zigzag and armchair shifting is energetically equivalent
Figure 2 Change of the total energy (per formula unit) when shifting COF-5 even layers in zigzag (top) and armchair (middle) directions and COF-1 in zigzag (bottom) direction The vector d shows the shift direction The low-energy minima in the two directions are labelled as zigzag and armchair respectively The equilibrium structures are shown as well
The formation energies of the individual COF structures are in agreement with full DFT calculations
We have calculated using ADF[9] the condensation energy of COF-1 and COF-5 using first-principles
DFT at the PBE[10]DZP[11] level to qualitatively support our results For simplicity we have used a
finite structure instead of the bulk crystal Their calculated Ecb energies are 77 and 15 kJmol-1
respectively for COF-1 and COF-5 These values support the endothermic nature of the condensation
101
reaction and are in excellent agreement with our DFTB results (91 and 21 kJmol-1 respectively see
Table 1)
The change of stacking should be visible in X-ray diffraction patterns because each space group has a
distinct set of symmetry imposed reflection conditions Powder XRD patterns from experiments are
available for all 2D COF structures[1 3andashd f g] We have simulated XRD patterns for AA AB serrated
Table 1 Calculated formation energies for the studied COF structures Ecb = condensation Esb = stacking energy EL = London dispersion energy Ee = electronic energy Energies are per condensed bond and are given in [kJ mol
-1+ lsquoarsquo and lsquozrsquo stand for armchair and zigzag respectively Note that the electronic energy Ee=Esb-EL
includes the electronic interlayer interaction and the formation energy of the monolayer and that the formation of the monolayers is endothermic
Structure Stacking Esb EL Ee
COF-5 AA -2968 -3060 092
AB -2548 -2618 070
serrated z -3051 -3073 022
serrated a -3052 -3073 021
inclined z -3297 -3045 -252
inclined a -3275 -3044 -231
Monolayer Ecb= 211
COF-1 AA -2683 -2739 056
AB -2186 -2131 -055
serrated z -2810 -2806 -004
inclined z -2784 -2788 004
Monolayer Ecb= 906
COF-6 AA -2881 -2963 082
AB -2127 -2146 019
serrated z -2978 -2996 018
serrated a -2978 -2995 017
inclined z -2946 -2975 029
inclined a -2954 -2974 021
Monolayer Ecb= 185
COF-8 AA -4488 -4617 129
102
AB -2477 -2506 029
serrated z -4614 -4646 032
serrated a -4615 -4647 032
inclined z -4578 -4612 035
inclined a -4561 -4591 030
Monolayer Ecb= 263
and inclined stacking forms for COF-5 COF-1 COF-6 and COF-8 (see Figure 1for their comparison
with experimental spectrum) For completeness we have studied XRD patterns of optimized COFs
using the experimentally determined[1 3a] interlayer separations this means we have kept the
layer geometry the same as the optimized structures and different stackings were obtained by
shifting adjacent layers accordingly
COF-1 was reported as having different powder spectra for samples before (as-synthesized) and after
removal of guest molecules with a particular mentioning about its layer shifting after removal We
have compared the two spectra with our simulated XRDs in order to see the stacking in the pure
form and how the stacking is changed at the presence of mesitylene guests Except that we have only
a vague comparison at angles (2θ) 11ndash15 the powder spectrum after guest removal is more similar
to the XRDs of serrated and inclined stacking structures A notable difference from AB is the absence
of high peaks at angles ~12 ~15 and ~19 This gives rise to the suggestion that COF-1 is not a
notable exception among the 2D COFs it follows the same topological trend as all other frameworks
of this class having one-dimensional (1D) pores as soon as the solvent is removed from the pores
This suggestion is supported by the experimental fact that the gas adsorption capacity of COF-1 is
only slightly lower than that of COF-6 for even as large gas molecules as methane and CO2[12] It is
not obvious how these molecules would enter an AB stacked COF-1 framework where the pores are
not connected by 1D channels (see Figure 1) Moreover our calculations suggest that the serrated
and inclined stackings are energetically favorable (see Table 1)
Indeed all the new stackings discussed in this article yield XRD patterns which are in agreement with
the currently available experimental data (see Figure 1) The inclined stackings have more peaks but
those are covered by the broad peaks in the experimental pattern The same is observed for the (002)
peaks of serrated stackings At present 2D COF synthesis methods are not yet capable to produce
crystals of sufficient quality to allow more detailed PXRD analysis in particular for the solvent-free
materials Microwave synthesis of COF-5 by rapid heating is reported[13] as much faster (~200 times)
compared with solvothermal methods however the structural details (XRD etc) remained
103
ambiguous We are confident that better crystals will be produced in future which will allow the
unambiguous determination of COF structures and can be compared to our simulations
Finally we want to emphasize that the suggested change of stacking is not only resulting in a
moderate change of geometry and XRD pattern The functional regions of COFs are their pores and
the pore geometry is significantly modified in our suggested structures compared to AA and AB
stackings First the inclined and serrated structures account for an increase of the surface area by 6
estimated for the interaction of the COFs with helium (3 for N2 5 for Ar 56 for Ne) Moreover
the shift between the layers exposes both boron and oxygen atoms to the pore surface leading to a
much stronger polarity than it can be expected for AA stacked COFs This difference is shown in
Figure 3 which plots the electrostatic potential of COF-5 in AA serrated and inclined stacking
geometries While the pore surface of AA stacked COF-5 shows a positively and negatively charged
stripes the other stacking arrangements show a much stronger alternation of charges indicating the
higher polarity of the surface The surface polarity can also be expressed in terms of Mulliken charges
of oxygen and boron atoms calculated at the DFT level which are COF-5 AA qB = 048 and qO = -048
COF-5 serrated qB = 047 and qO = -049 COF-5 inclined qB = 045 and qO = -048
Figure 3 Calculated electrostatic potential of COF-5 AA (left) serrated (middle) and inclined (right) stacking forms Blue - negative and red-positive values of the 005 isosurface
Conclusion We have studied 2D COF layer stackings energetically and found that energy minimum
structures have staggering of hexagonal rings at the connector or linker parts This can be achieved if
the bulk structure has either serrated or inclined order These newly proposed orders have their
simulated XRDs matching well with the available experimental powder spectrum Hence we claim
that all reported 2D COF geometries shall be re-examined carefully in experiment Due to the change
of the pore geometry and surface polarity the envisaged range of applications for 2D COFs might
change significantly We believe that these results are of utmost importance for the design of
functionalized COFs where functional groups are added to the pore surfaces
104
References
[1] A P Cocircteacute A I Benin N W Ockwig M OKeeffe A J Matzger O M Yaghi Science 2005 310 1166
[2] H M El-Kaderi J R Hunt J L Mendoza-Cortes A P Cote R E Taylor M OKeeffe O M Yaghi Science
2007 316 268
[3] a) A P Cocircteacute H M El-Kaderi H Furukawa J R Hunt O M Yaghi J Am Chem Soc 2007 129 12914 b) J
R Hunt C J Doonan J D LeVangie A P Cocircte O M Yaghi J Am Chem Soc 2008 130 11872 c) R W
Tilford W R Gemmill H C zur Loye J J Lavigne Chem Mater 2006 18 5296 d) R W Tilford S J Mugavero
P J Pellechia J J Lavigne Adv Mater 2008 20 2741 e) F J Uribe-Romo J R Hunt H Furukawa C Klock M
OKeeffe O M Yaghi J Am Chem Soc 2009 131 4570 f) S Wan J Guo J Kim H Ihee D L Jiang Angew
Chem 2008 120 8958 Angew Chem Int Ed 2008 47 8826 g) S Wan J Guo J Kim H Ihee D L Jiang
Angew Chem 2009 121 5547 Angew Chem Int Ed 2009 48 5439
[4] R T Paine C K Narula Chem Rev 1990 90 73
[5] C Doonan D Tranchemontagne T Glover J Hunt O Yaghi Nat Chem 2010 2 235
[6] a) T Heine M Rapacioli S Patchkovskii J Frenzel A M Koester P Calaminici S Escalante H A Duarte R
Flores G Geudtner A Goursot J U Reveles A Vela D R Salahub deMon deMonnano edn Mexico DF
Mexico and Bremen (Germany) 2009 b) A F Oliveira G Seifert T Heine H A Duarte J Braz Chem Soc
2009 20 1193 c) G Seifert D Porezag T Frauenheim Int J Quantum Chem 1996 58 185 d) L Zhechkov T
Heine S Patchkovskii G Seifert H A Duarte J Chem Theory Comput 2005 1 841
[7] a) httpwwwccdccamacukproductsmercury b) C F Macrae P R Edgington P McCabe E Pidcock
G P Shields R Taylor M Towler J Van de Streek J Appl Crystallogr 2006 39 453
[8] R Dovesi V R Saunders C Roetti R Orlando C M Zicovich- Wilson F Pascale B Civalleri K Doll N M
Harrison I J Bush P DrsquoArco M Llunell Crystal 102 ed
[9] a) httpwwwscmcom ADF200901 Amsterdam The Netherlands b) G te Velde F M Bickelhaupt S J
A van Gisbergen C Fonseca Guerra E J Baerends J G Snijders T Ziegler J Comput Chem 2001 22 931
[10] J P Perdew K Burke M Ernzerhof Phys Rev Lett 1996 77 3865
[11] E Van Lenthe E J Baerends J Comput Chem 2003 24 1142
[12] H Furukawa O M Yaghi J Am Chem Soc 2009 131 8875
[13] N L Campbell R Clowes L K Ritchie A I Cooper Chem Chem Mater 2009 21 204
105
Appendix D
On the reticular construction concept of covalent organic frameworks
Binit Lukose Agnieszka Kuc Johannes Frenzel and Thomas Heine
Beilstein J Nanotechnol 2010 1 60ndash70
DOI103762bjnano18
Abstract
The concept of reticular chemistry is investigated to explore the applicability of the formation of
Covalent Organic Frameworks (COFs) from their defined individual building blocks Thus we have
designed optimized and investigated a set of reported and hypothetical 2D COFs using Density
Functional Theory (DFT) and the related Density Functional based tight-binding (DFTB) method
Linear trigonal and hexagonal building blocks have been selected for designing hexagonal COF layers
High-symmetry AA and AB stackings are considered as well as low-symmetry serrated and inclined
stackings of the layers The latter ones are only slightly modified compared to the high-symmetry
forms but show higher energetic stability Experimental XRD patterns found in literature also
support stackings with highest formation energies All stacking forms vary in their interlayer
separations and band gaps however their electronic densities of states (DOS) are similar and not
significantly different from that of a monolayer The band gaps are found to be in the range of 17ndash
40 eV COFs built of building blocks with a greater number of aromatic rings have smaller band gaps
Introduction
In the past decade considerable research efforts have been expended on nanoporous materials due
to their excellent properties for many applications such as gas storage and sieving catalysis
106
selectivity sensoring and filtration [1] In 1994 Yaghi and co-workers introduced ways to synthesize
extended structures by design This new discipline is known as reticular chemistry [23] which uses
well-defined building blocks to create extended crystalline structures The feasibility of the building
block approach and the geometry preservation throughout the assembly process are the key factors
that lead to the design and synthesis of reticular structures
One of the first families of materials synthesized using reticular chemistry were the so-called Metal-
Organic Frameworks (MOFs) [4] They are composed of metal-oxide connectors which are covalently
bound to organic linkers The coordination versatility of constituent metal ions along with the
functional diversity of organic linker molecules has created immense possibilities The immense
potential of MOFs is facilitated by the fact that all building blocks are inexpensive chemicals and that
the synthesis can be carried out solvothermally MOFs are commercially available and the scale up of
production is continuing Since the discovery of MOFs many other crystalline frameworks have been
synthesized using reticular chemistry such as Metal-Organic Polyhedra (MOP) [5] Zeolite
Imidazolate Frameworks (ZIFs) [6] and Covalent Organic Frameworks (COFs) [7]
In 2005 Cocircteacute and co-workers introduced COF materials [7-14] where organic linker molecules are
stitched together by covalent entities including boron and oxygen atoms to form a regular
framework These materials have the distinct advantage that all framework bonds represent strong
covalent interactions and that they are composed of light-weight elements only which lead to a very
low mass density [7-9] These materials can be synthesized by wet-chemical methods by
condensation reactions and are composed of inexpensive and non-toxic building blocks so their
large-scale industrial application appears to be possible From a topological viewpoint we distinguish
two- and three-dimensional COFs In two-dimensional (2D) COFs the covalently bound framework is
restricted to layers The crystal is then similar as in graphite composed of a stack of layers which
are not connected by covalent bonds
COFs compared with MOFs have lower mass densities due to the absence of heavy atoms and
therefore might be better for many applications For example the gravimetric uptake of gases is
almost twice as large as that of MOFs with comparable surface areas [1516] Because COFs are fairly
new materials many of their properties and applications are still to be explored
Recently we have studied the structures of experimentally well-known 2D COFs [17] We have found
that commonly accepted 2D structures with AA and AB kinds of layering are energetically less stable
than inclined and serrated forms This is because AA stacking maximises the Coulomb repulsion due
to the close vicinity of charge carrying atoms alike (O B atoms) in neighbouring layers The serrated
and inclined forms are only slightly modified (layers are shifted with respect to each other by asymp14 Aring)
107
and experience less Coulomb forces between the layers compared to AA stacking This is equivalent
to the energetic preference of graphite for an AB (Bernal) over an AA form (simple hexagonal) if we
ignore the fact that interlayer ordering in serrated and inclined forms are not uniform everywhere A
known example of this is that in eclipsed hexagonal boron nitride (h-BN) boron atoms in one layer
serve as nearest neighbours to nitrogen atoms in adjacent layers (AB stacking) The Coulomb
interaction rules out possible interlayer eclipse between atoms with similar charges in this case
In the present work we aim to explore how far the concept of reticular chemistry is applicable to the
individual building units which define COFs For this purpose we have investigated a set of reported
and hypothetical 2D COFs theoretically by exploring their structural energetic and electronic
properties We have compared the properties of the isolated building blocks with those of the
extended crystal structures and have found that the properties of the building units are indeed
maintained after their assembly to a network
Results and Discussion
Structures and nomenclature
We have considered four connectors (IndashIV) and five linkers (andashe) for the systematic design of a
number of 2D COFs (Figure 1) Each COF was built from one type of connector and one type of linker
thus resulting in the design of 20 different structures Moreover we have considered two
hypothetical reference structures which are only built from connectors I and III (no linker is present)
REF-I and REF-III Properties of other COFs were compared with the properties of these two
structures Some of the studied COFs are already well known in the literature [781314] and we
continue to use their traditional nomenclature while hypothetical ones are labelled in the
chronological order with an M at the end which stands for modified
Figurel 1 The connector (IndashIV) and linker (andashe) units considered in this work The same nomenclature is used in the text Carbon ndash green oxygen ndashred boron ndash magenta hydrogen ndash white
108
Using the secondary building unit (SBU) approach we can distinguish the connectors between
trigonal [T] (connectors I II III) and hexagonal [H] (connector IV) and the linkers between linear [l]
(linkers a b e) and trigonal [t] (linkers c d) Topology of the layer is determined by the geometries
of the connector and linker molecules and typically a hexagonal pattern is formed due to the D3h
symmetry of the connector moieties Based on these topologies of the constituent building blocks
we have classified the studied COFs into four groups Tl Tt Hl and Ht (Figure 2) Hereafter we will
use this nomenclature to describe the COF topologies
Figurel 2 Topologies of 2D COFs considered in this work (from the left) Tl Tt Hl and Ht Red and blue blocks are secondary building units corresponding to connectors and linkers respectively
We have considered high-symmetry AA and AB kinds of stacking (hexagonal) and low-symmetry
serrated (orthorhombic) and inclined (monoclinic) kinds of stacking of the layers The latter two were
discussed in a previous work on 2D COFs [17] As an example the structure of COF-5 [7] in different
kinds of stacking of layers is shown in Figure 3 In eclipsed AA stacking atoms of adjacent layers lie
directly on top of each other whilst in staggered AB stacking three-connected vertices lie directly on
top of the geometric center of six-membered rings of neighbouring layers In both serrated and
inclined kinds of stacking the layers are shifted with respect to each other by approximately 14 Aring
resulting in hexagonal rings in the connector or linker being staggered with those in the adjacent
layers In serrated stacking alternate layers are eclipsed In inclined stacking layers lie shifted along
one direction and the lattice vector pointing out of the 2D plane is not rectangular For COFs made of
connector I due to the absence of five-membered C2O2B rings a zigzag shift leads to staggering in
both connector and linker parts For those made of other connectors staggering at the connector or
linker depends on whether the shift is armchair or zigzag respectively [17]
Structural properties
We have compared structural properties of isolated building blocks with those of the extended COF
structures Negligible differences have been found in the bond lengths and bond angles of the
building blocks and the corresponding crystal structures This indicates that the structural integrity of
the building blocks remains unchanged while forming covalent organic frameworks confirming the
109
principle of reticular chemistry In addition the CndashB BndashO and OndashC bond lengths are almost the same
when different COF structures are compared (see Table S1 in Supporting Information File 1) The
experimental bond lengths are asymp154 Aring for CndashB asymp138 Aring for OndashC and 137ndash148 Aring for BndashO However
in the case of COF-1 the experimental values are slightly larger (160 Aring for CndashB and 151 Aring for BndashO)
This could be the reason why our calculated bond lengths for COF-1 are much shorter than the
experimental values while all the other structures agree within 5 error The calculated CndashC bond
lengths vary in the range from 136ndash147 Aring (Figure S2 in Supporting Information File 1) and are the
same for the COFs and their constituent building blocks at the respective positions of the carbon
atoms In addition the reference structures REF-I and REF-III have direct BndashB bond lengths of 167 Aring
and 166 Aring respectively which is shorter by 014 Aring than a typical BndashB bond length The calculated
bond angles OBO in B3O3 and C2O2B rings are 120deg and 113deg respectively
Figure 3 Layer stackings considered AA AB serrated and inclined
Interlayer distances (d) which is the shortest distance between two layers (equivalent to c for AA
c2 for AB and serrated) are different in all kinds of stacking AB stacked 2D COFs have shorter
interlayer distances than the corresponding AA serrated and inclined stacked structures Among the
latter three AA stacked COFs have higher values for d because of the higher repulsive interlayer
orbital interactions resulting from the direct overlap of polarized alike atoms between the adjacent
layers This results in higher mass densities for AB stacked COF analogues Serrated and inclined
stacks have only slightly higher mass densities compared to AA The differences in mass densities for
all kinds of stacking are attributed to the differences in their interlayer separations The d values of
most of the COFs are larger than that of graphite in AA stacking but smaller in AB stacking
Cell parameters (a) and mass densities (ρ) of all the COFs constructed from the considered
connectors and linkers are shown in Table 1 (and in Table S3 in Supporting Information File 1) Mass
densities of all the COFs are much lower than that of graphite (227 gmiddotcmminus3) and diamond (350
gmiddotcmminus3) AAserratedinclined stacked COF-10s have the lowest mass densities (045046046
gmiddotcmminus3) which is lower than that of MOF-5 (059 gmiddotcmminus3) [4] and comparable to that of highly porous
MOF-177 (042 gmiddotcmminus3) [18]
110
In order to identify the stacking orders we have analyzed X-ray diffraction (XRD) patterns of the well-
known COFs (COF-10 TP COF PPy-COF see Figure 4) in all the above discussed stacking kinds The
change of stacking should be visible in XRDs because each space group has a distinct set of symmetry
imposed reflection conditions The XRD patterns of AA serrated and inclined stacking kinds which
differ within a slight shift of adjacent layers to specific directions are in agreement with the presently
available experimental data [81314] Their peaks are at the same angles as in the experimental
spectrum whereas AB stacking clearly shows differences The slight differences in the (001) angle
between each stacking resemble the differences in their interlayer separations The inclined
stackings have more peaks however they are covered by the broad peaks in the experimental
patterns Similar results for COF-1 COF-5 COF-6 and COF-8 have been discussed in our previous
work [17]
Figure 4 The calculated and experimental [81314] XRD patterns of PPy-COF (top) COF-10 (middle) and TP COF (bottom)
111
Table 1 The calculated unit cell parameter a [Aring] interlayer distance d [Aring] and mass density ρ [gmiddotcmminus3
] for AA and AB stacked COFs Note that the cell parameter a is the same for all stacking types Experimental data [719] is given in parentheses
COF Building
Blocks
a d ρ
AA AB AA AB
COF-1 I-a 1502 (15620) 351 313 (332) 094 106
COF-1M I-b 2241 349 304 068 078
COF-2M I-c 1492 347 312 095 106
COF-3M I-d 0747 349 327 153 164
PPy-COF I-e 2232 (22163) 349 (3421) 297 084 099
COF-5 II-a 3014 (30020) 347 (3460) 326 056 060
COF-10 II-b 3758 (37810) 347 (3476) 318 045 (045) 050
COF-8 II-c 2251 (22733) 346 (3476) 320 071 (070) 077
COF-6 II-d 1505 (15091) 346 (3599) 327 104 110
TP COF II-e 3750 (37541) 348 (3378) 320 051 056
COF-4M III-a 2171 350 318 073 080
COF-5M III-b 2915 350 318 055 061
COF-6M III-c 1833 345 318 083 090
COF-7M III-d 1083 350 330 129 136
TP COF-1M III-e 2905 349 310 065 074
COF-8M IV-a 1748 359 329 140 148
COF-9M IV-b 2176 349 330 117 174
COF-10M IV-c 2254 342 336 127 128
COF-11M IV-d 1512 346 338 168 172
TP COF-2M IV-e 2173 347 332 134 140
REF-I I 0773 359 336 144 148
REF-III III 1445 353 336 104 121
Graphite 247 343 335 220 227
112
Energetic stability
We have considered dehydration reactions the basis of experimental COF synthesis to calculate
formation energies of COFs in order to predict their energetic stability Molecular units 14-
phenylenediboronic acid (BDBA) 11rsquo-biphenyl]-44-diylboronic acid (BPDA) 5rsquo-(4-boronophenyl)-
11rsquo3rsquo1rdquo-terphenyl]-44rdquo-diboronic acid (BTPA) benzene-135-triyltriboronic acid (BTBA) and
pyrene-27-diylboronic acid (PDBA) were considered as linkers andashe respectively with -B(OH)2 groups
attached to each point of extension (Figure 5) Self-condensation of these building blocks result in
the formation of B3O3 rings and the resultant COFs are those made of connector I and the
corresponding linkers This process requires a release of three or six water molecules in case of t or l
topology respectively
Figure 5 The reactants participating in the formation of 2D COFs
Co-condensation of the above molecular units with compounds such as 23671011-
hexahydroxytriphenylene (HHTP) hexahydroxybenzene (HHB) and dodecahydroxycoronene (DHC)
(Figure 5) gives rise to COFs made of connectors II III and IV respectively and the corresponding
linkers Self-condensation of tetrahydroxydiborane (THDB) and co-condensation of HHB with THDB
result in the formation of the reference structures REF-I and REF-III respectively In relation to the
corresponding connectorlinker topologies these building blocks satisfy the following equations of
the co-condensation reaction for COF formation
2 2 3 COF 12 H O Tl T l (1)
113
2 1 1 COF 6 H O Tt T t (2)
2 1 3 COF 12 H O Hl H l (3)
2 1 2 COF 12 H O Ht H t (4)
with a stochiometry appropriate for one unit cell The number of covalent bonds formed between
the building blocks is equivalent to the number of released water molecules we refer to it as
ldquoformula unitrdquo and will give all energies in the following in kJmiddotmolminus1 per formula unit
Table 2 The calculated energies [kJ molminus1
] per bond formed between building blocks for AA and AB stacked COFs Ecb is the condensation energy Esb is the stacking energy and Efb is the COF formation energy (Efb = Ecb
+ Esb) The calculated band gaps Δ eV+ are given as well
COF Building
Blocks
Mono-
layer
AA AB
Ecb Esb Efb ∆ Esb Efb ∆
COF-1 I-a 906 -2683 -1777 33 -2187 -1280 36
COF-1M I-b 949 -4266 -3366 27 -2418 -1469 31
COF-2M I-c 956 -5727 -4771 28 -4734 -3778 30
COF-3M I-d 763 -2506 -1742 38 -2801 -2037 40
PPy-COF I-e 858 -5723 -4866 24 -3855 -2998 26
COF-5 II-a 211 -2968 -2756 24 -2548 -2337 28
COF-10 II-b 317 -3766 -3448 23 -1344 -1026 26
COF-8 II-c 263 -4488 -4224 25 -2477 -2213 28
COF-6 II-d 185 -2881 -2695 28 -2127 -1942 31
TP COF II-e 231 -4453 -4222 24 -1480 -1250 27
COF-4M III-a -033 -1730 -1763 26 -880 -913 26
COF-5M III-b 007 -2533 -2526 25 -972 -965 25
COF-6M III-c 014 -3231 -3217 26 -2134 -2120 28
114
COF-7M III-d -170 -1635 -1805 30 -1607 -1777 32
TP COF-1M III-e -014 -3226 -3240 24 -1277 -1291 24
COF-8M IV-a -787 -2756 -3543 18 -2680 -3467 21
COF-9M IV-b -836 -3577 -4414 17 -3003 -3839 21
COF-10M IV-c -947 -4297 -5244 18 -4192 -5140 22
COF-11M IV-d -403 -2684 -3087 21 -2833 -3236 24
TP COF-2M IV-e 030 -4345 -4315 18 -4117 -4087 21
We have calculated the condensation energy of a single COF layer formed from monomers (building
blocks hereafter called bb) according to the above reactions as follows
tot tot tot totcb m H2O 1 bb1 2 bb2E E n E ndash m E m E (5)
where Emtot ndash total energy of the monolayer EH2O
tot is the total energy of the water molecule Ebb1tot
and Ebb2tot are the total energies of interacting building blocks and n m1 m2 are the corresponding
stoichiometry numbers
We have also calculated the stacking energy Esb of layers
tot totsb nl s mE E n E (6)
where Enltot is the total energy of ns number of layers stacked in a COF Finally the COF formation
energy can be given as a sum of Ecb and Esb (see Table 1 and Table S1)
Electronic properties
All COFs including the reference structures are semiconductors with their band gaps lying between
17 eV and 40 eV (Table 2 and Table S4 in Supporting Information File 1) The largest band gaps are
of the reference structures while the lowest values are of COFs built from connector IV The band
gaps are different for different stacking kinds This difference can be attributed to the different
optimized interlayer distances Generally AB serrated and inclined stacked COFs have band gaps
comparable to or larger than that of their AA stacked analogues
115
We have calculated the electronic total density of states (TDOS) and the individual atomic
contributions (partial density of states PDOS) The energy state distributions of COFs and their
monolayers are studied and a comparison for COF-5 is shown in Figure 6 In all stacking kinds
negligible differences are found for the densities at the top of valence band and the bottom of
conduction band These slight differences suggest that the weak interaction between the layers or
the overlap of π-orbitals does not affect the electronic structure of COFs significantly Hence there is
almost no difference between the TDOS of AA AB serrated and inclined stacking kinds Therefore in
the following we discuss only the AA stacked structures
Figure 6 Total densities of states (DOS) (black) of AA (top left) AB (top right) serrated (bottom left) and inclined (bottom right) comparing stacked COF-5 with a monolayer (red) of COF-5 The differences between the TDOS of bulk and monolayer structures are indicated in green The Fermi level EF is shifted to zero
Figure 7 Partial density of states of carbon (left) boron (center) and oxygen (right) atoms of COFs built from type-I connectors and different linkers The vertical dashed line in each figure indicates the Fermi level EF
116
It is of interest to see how the properties of COFs change depending on their composition of different
secondary building units that is for different connectors and linkers PDOS of COFs built from type-I
connectors and different linkers are plotted in Figure 7 The PDOS of carbon atoms is compared with
that of graphite (AA-stacking) while PDOS of boron and oxygen atoms are compared with that of
REF-I a structure which is composed solely of connector building blocks Going from top to bottom
of the plots the number of carbon atoms per unit cell increases It can be seen that this causes a
decrease of the band gap Figure 8 shows the PDOS of COFs built from type-a linkers and different
connectors where the COFs are arranged in the increasing number of carbon atoms in their unit cells
from top to the bottom Again the C PDOS is compared with that of graphite while both REF-I and
REF- III are taken in comparison to O and B PDOS The observed relation between number of carbon
atoms and band gap is verified
Figure 8 Partial density of states of carbon (left) boron (center) and oxygen (right) atoms of COFs built from type-a linkers The vertical dashed line in each figure indicates the Fermi level EF
Conclusion
In summary we have designed 2D COFs of various topologies by connecting building blocks of
different connectivity and performed DFTB calculations to understand their structural energetic and
electronic properties We have studied each COF in high-symmetry AA and AB as well as low-
symmetry inclined and serrated stacking forms The optimized COF structures exhibit different
interlayer separations and band gaps in different kinds of layer stackings however the density of
states of a single layer is not significantly different from that of a multilayer The alternate shifted
layers in AB serrated and inclined stackings cause less repulsive orbital interactions within the layers
which result in shorter interlayer separation compared to AA stacking All the studied COFs show
117
semiconductor-like band gaps We also have observed that larger number of carbon atoms in the
unit cells in COFs causes smaller band gaps and vice versa Energetic studies reveal that the studied
structures are stable however notable difference in the layer stacking is observed from
experimental observations This result is also supported by simulated XRDs
Methods
We have optimized the atomic positions and the lattice parameters for all studied COFs All
calculations were carried out at the Density Functional Tight-Binding (DFTB) [2021] level of theory
DFTB is based on a second-order expansion of the KohnndashSham total energy in the Density-Functional
Theory (DFT) with respect to charge density fluctuations This can be considered as a non-orthogonal
tight-binding method parameterized from DFT which does not require large amounts of empirical
parameters however maintains all the qualities of DFT The main idea behind this method is to
describe the Hamiltonian eigenstates with an atomic-like basis set and replace the Hamiltonian with
a parameterized Hamiltonian matrix whose elements depend only on the inter-nuclear distances and
orbital symmetries [21] While the Hamiltonian matrix elements are calculated using atomic
reference densities the remaining terms to the KohnndashSham energy are parameterized from DFT
reference calculations of a few reference molecules per atom pair The accuracy is very much
improved by the self-consistent charge (SCC) extension Two computational codes were used
deMonNano code [22] and DFTB+ code [23] The first code has dispersion correction [24]
implemented to account for weak interactions and was used to obtain the layered bulk structure of
COFs and their formation energies The performance for interlayer interactions has been tested
previously for graphite [24] The second code which can perform calculations using k-points was
used to calculate the electronic properties (band structure and density of states) Band gaps have
been calculated as an additional stability indicator While these quantities are typically strongly
underestimated in standard LDA- and GGA-DFT calculations they are typically in the correct range
within the DFTB method For validation of our method we have calculated some of the structures
using Density Functional Theory (DFT) as implemented in ADF code [2526]
Periodic boundary conditions were used to represent frameworks of the crystalline solid state The
conjugatendashgradient scheme was chosen for the geometry optimization The atomic force tolerance of
3 times 10minus4 eVAring was applied The optimization using Γ-point approximation was performed with the
deMon-Nano code on 2times2times4 supercells Some of the monolayers were also optimized using the
DFTB+ code on elementary unit cells in order to validate the calculations within the Γ-point
approximation The number of k-points has been determined by reaching convergence for the total
energy as a function of k-points according to the scheme proposed by Monkhorst and Pack [27]
118
Band structures were computed along lines between high symmetry points of the Brillouin zone with
50 k-points each along each line XRD patterns have been simulated using Mercury software [2829]
We have also performed first-principles DFT calculations at the PBE [30] DZP [31] level to support
our results quantitatively For simplicity we have used finite structures instead of bulk crystals
Supporting Information
Table S1 The calculated bond lengths and angles of all studied COFs Corresponding experimental values found from the literature are shown in brackets
COF Building
Blocks
C-B B-O O-C OBO
COF-1 I-a 1498 (1600) 1393 (1509) 1203 (1200)
COF-1M I-b 1497 1393 1203
COF-2M I-c 1497 1392 1203
COF-3M I-d 1496 1392 1201
PPy-COF I-e 1498 1393 1202 (1190)
COF-5 II-a 1496 (1530) 1399 (1367) 1443 (1384) 1135dagger (1139)
COF-10 II-b 1495 (1553) 1399 (1465) 1443 (1385) 1135dagger (1120)
COF-8 II-c 1495 (1548) 1399 (1479) 1443 1135dagger
COF-6 II-d 1496 1399 1443 1135dagger
TP COF II-e 1496 (1542) 1399 (1396) 1444 (1377) 1135dagger (1182)
COF-4M III-a 1496 1398 1449 1135dagger
COF-5M III-b 1496 1398 1449 1136dagger
COF-6M III-c 1496 1399 1451 1134dagger
COF-7M III-d 1496 1398 1449 1136dagger
TP COF-1M III-e 1496 1398 1450 1136dagger
COF-8M IV-a 1496 1398 1445 1131dagger
COF-9M IV-b 1495 1398 1444 1131dagger
119
COF-10M IV-c 1495 1391 1418 1126dagger
COF-11M IV-d 1498 1399 1450 1134dagger
TP COF-2M IV-e 1499 1399 1447 1134dagger
B3O3 connectivity dagger C2B2O connectivity
It can be noticed that the bond lengths of experimentally known COF-1 is much larger compared to
our optimized bond lengths as well as that of other synthesized COFs
Figure S2 Calculated C-C bond lengths in connectors and linkers Hydrogen atoms are omitted for clarity
Table S3 The calculated unit cell parameters a interlayer distance d Aring+ and mass density ρ gcm-3
] for serrated (Sa serrated armchair Sz serrated zigzag) and inclined (Ia inclined armchair Iz inclined zigzag) stacked COFs
COF Building
Blocks
a d ρ
Sa Sz Ia Iz Sa Sz Ia Iz
COF-1 I-a 1502 343 343 097 097
COF-1M I-b 2241 341 342 069 069
COF-2M I-c 1492 340 339 097 097
COF-3M I-d 0747 341 342 157 156
PPy-COF I-e 2232 341 341 086 086
120
COF-5 II-a 3014 342 342 341 340 057 057 058 058
COF-10 II-b 3758 341 341 342 340 046 046 046 046
COF-8 II-c 2251 341 341 342 342 073 073 072 072
COF-6 II-d 1505 342 341 340 340 105 106 106 106
TP COF II-e 3750 342 341 342 342 052 052 052 052
COF-4M III-a 2171 344 344 345 344 074 074 074 074
COF-5M III-b 2915 343 342 343 343 056 056 056 056
COF-6M III-c 1833 341 341 342 341 084 084 084 084
COF-7M III-d 1083 344 343 340 344 131 131 132 131
TP COF-1M III-e 2905 343 342 343 342 066 067 066 066
COF-8M IV-a 1748 341 341 342 342 142 142 142 142
COF-9M IV-b 2176 341 341 341 342 119 119 119 119
COF-10M IV-c 2254 340 340 340 340 128 128 128 128
COF-11M IV-d 1512 341 341 340 340 171 171 171 171
TP COF-2M IV-e 2173 340 340 340 340 137 137 137 137
REF-I I 0773 349 345 148 15
REF-III III 1445 348 349 106 106
Table S4 The calculated energies [kJ mol-1
] per bond formed between building blocks for serrated (Sa serrated armchair Sz serrated zigzag) and inclined (Ia inclined armchair Iz inclined zigzag) stacked COFs Ecb ndash condensation energy Esb ndash stacking energy and Efb ndash COF formation energy (Efb = Ecb + Esb) The calculated band gaps ∆ eV+ are given as well
COF Sa Sz Ia Iz
Esb Efb Δ Esb Efb Δ Esb Efb Δ Esb Efb Δ
-1 -2810 -1904 36 -2786 -1880 36
-1M -4426 -3477 30 -4389 -3440 30
-2M -5967 -5011 30 -5833 -4877 30
121
-3M -2667 -1904 40 -2591 -1828 40
PPy- -5916 -5058 26 -5865 -5007 26
-5 -3052 -2841 27 -3051 -2840 26 -3275 -3064 26 -3297 -3086 26
-10 -3868 -3551 26 -3869 -3552 25 -3830 -3513 26 -4293 -3976 25
-8 -4615 -4352 27 -4614 -4351 27 -4571 -4308 27 -4573 -4310 27
-6 -2978 -2793 30 -2978 -2793 30 -3117 -2932 30 -3090 -2905 30
TP -4557 -4326 26 -4562 -4331 26 -4511 -4280 26 -4511 -4280 26
-4M -1811 -1844 28 -1813 -1846 28 -1769 -1802 28 -1880 -1913 28
-5M -2626 -2619 27 -2630 -2623 26 -2597 -2590 26 -2600 -2593 26
-6M -3375 -3361 28 -3381 -3367 28 -3487 -3473 28 -3349 -3335 28
-7M -1724 -1894 32 -1726 -1896 31 -2333 -2503 32 -1866 -2036 31
TP -1M -3327 -3341 26 -3334 -3348 26 -3340 -3354 26 -3353 -3367 26
-8M -2897 -3684 21 -2895 -3682 21 -2971 -3758 21 -2983 -3770 21
-9M -3732 -4568 20 -3733 -4569 20 -3684 -4520 20 -3706 -4542 20
-10M -4512 -5459 21 -4513 -5460 21 -4491 -5438 21 -4495 -5442 21
-11M -2831 -3234 24 -2834 -3237 24 -2816 -3219 24 -2830 -3233 24
TP -2M -4500 -4470 20 -4505 -4475 20 -4495 -4465 20 -4496 -4466 20
122
Appendix E
Stability and electronic properties of 3D covalent organic frameworks
Binit Lukose Agnieszka Kuc and Thomas Heine
Prepared for publication
Abstract Covalent Organic Frameworks (COFs) are a class of covalently linked crystalline nanoporous
materials versatile for nanoelectronic and storage applications 3D COFs in particular have very
large pores and low-mass densities Extensive theoretical studies of their energetic and mechanical
stability as well as their electronic properties are discussed in this paper All studied 3D COFs are
energetically stable materials though mechanical stability does not exceed 20 GPa Electronically all
COFs are semiconductors with band gaps depending on the number of sp2 carbon atoms present in
the linkers similar to 3D MOF family
Introduction
Covalent Organic Frameworks (COFs)1-3 comprise an emerging class of crystalline materials that
combines organic functionality with nanoporosity COFs have organic subunits stitched together by
covalent entities including boron carbon nitrogen oxygen or silicon atoms to form periodic
frameworks with the faces and edges of molecular subunits exposed to pores Hence their
applications can range from organic electronics to catalysis to gas storage and sieving4-7 The
properties of COFs extensively depend on their molecular constituents and thus can be tuned by
rational chemical design and synthesis289 Step by step reversible condensation reactions pave the
123
way to accomplish this target Such a reticular approach allows predicting the resulting materials and
leads to long-range ordered crystal structures
Since their first mentioning in the literature13 COFs are under theoretical investigation10-17 mainly for
gas storage applications Methods such as doping18-24 and organic linker functionalization2526 have
been suggested to improve their storage capacities In addition to the moderate pore size and
internal surface area COFs have the privileges of a low-weight material as they are made of light
elements Hence their gravimetric adsorption capacity is remarkably high10 The lowest mass density
ever reported for any crystalline material is that for COF-108 (018 gcm-3) Also their stronger
covalent bonds compared to related Metal-Organic Frameworks (MOFs)27 endorse thermodynamic
strength These genuine qualities of COFs make them attractive for hydrogen storage investigations
Crystallization of linked organic molecules into 2D and 3D forms was achieved in the years 20051 and
20073 respectively by the research group of O M Yaghi Several COFs have been synthesized since
then and some of the 2D COFs has been proven good for electronic or photovoltaic applications428-33
However the growth in this area appears to be slow compared to rapidly developing MOFs albeit
the promising H2 adsorption measurements53435 and a few synthetic improvements736-42
COF-102 and COF-103 were synthesized3 by the self-condensation of the non-planar tetra(4-
dihydroxyborylphenyl)methane (TBPM) and tetra(4-dihydroxyborylphenyl)silane (TBPS) respectively
(see Fig 1 for the building blocks and COF geometries) The co-condensation of these compounds
with triangular hexahydroxytriphenylene (HHTP) results in COF-108 and COF-105 respectively with
different topologies Yaghi et al3 have reported the formation of highly symmetric topologies ctn
(carbon nitride dI 34 ) and bor (boracite mP 34 ) when tetrahedral building blocks are linked
together with triangular ones The topology names were adopted from reticular chemistry structure
resource (RCSR)43 Simulated Powder X-ray diffraction in comparison with experimental powder
spectrum suggested ctn topology for COF-102 103 and 105 and bor topology for COF-108 The
condensation of tert-butylsilanetriol (TBST) and TBPM leads to the formation of COF-20244 This was
reported as ctn topology The COF reactants and schematic diagrams of ctn and bor topologies are
given in Figure1 The assembly of tetrahedral and linear units is very likely to result in a diamond-like
form COF-300 was formed according to the condensation of tetrahedral tetra-(4-anilyl)methane
(TAM) and linear terephthaldehyde (TA)45 However the synthesized structure was 5-fold
interpenetrated dia-c5 topology43
In this work we present theoretical studies of 3D COFs using density functional based methods to
explore their structural electronic energetic and mechanical properties Our previous studies on 2D
COFs4647 had resulted in questioning the stability of eclipsed arrangement of layers (AA stacking) and
124
suggesting energetically more stable serrated and inclined packing In this paper we attempt to
explore the stability and electronic properties of the experimentally known 3D COFs namely COF-
102 103 105 108 202 and 300 In the present studies we follow the reticular assembly of the
molecular units that form low-weight 3D COFs Considering the structural distinction from MOFs
COFs lack the versatility of metal ingredients what results in diverse properties48 A collective study is
then carried out to understand the characteristics and drawbacks of COFs
Figure 1 (Upper panel) Building blocks for the construction of 3D COFs (Lower panel) lsquoctnrsquo and lsquoborrsquo
networks formed by linking tetrahedral and triangular building units
Methods
COF structures were fully optimized using Self-consistent Charge -Density Functional based Tight-
Binding (SCC-DFTB) level of theory4950 Two computational codes were used deMonNano51 and
125
DFTB+52 The first code which has dispersion correction53 implemented to account for weak
interactions was used for the geometry optimization and stability calculations The second code
which can perform calculations using k-point sampling was used to calculate the electronic
properties (band structure and density of states) The number of k-points has been determined by
reaching convergence for the total energy as a function of k-points according to the scheme
proposed by Monkhorst and Pack54 Periodic boundary conditions were used to represent
frameworks of the crystalline solid state Conjugatendashgradient scheme was chosen for the geometry
optimization Atomic force tolerance of 3 times 10minus4 eVAring was applied The optimization using Γ-point
approximation was performed on rectangular supercells containing more than 1000 atoms For
validation of our method we have calculated energetic stability using Density Functional Theory (DFT)
at the PBE55DZP56 level as implemented in ADF code5758 using cluster models The cluster models
contain finite number of building units and correspond to the bulk topology of the COFs XRD
patterns have been simulated using Mercury software5960
In this work we continued to use the traditional nomenclature of the experimentally known COFs All
of the structures have the same tetrahedral blocks and differ only in the central sp3 atom (carbon or
silicon) that is included in our nomenclature
Bulk modulus (B) of a solid at absolute zero can be calculated as
(1) B = 2
2
dV
EdV
where V and E are the volume and energy respectively
Owing to the dehydration reactions we have calculated the formation (condensation) energy of each
COF formed from monomers (building blocks) as follows
(2) EF = Etot + n EH2Otot ndash (m1 Ebb1
tot + m2 Ebb2tot)
where Etot -- total energy of the COF EH2Otot -- total energy of water molecule Ebb1
tot and Ebb2tot -- total
energies of interacting building blocks n m1 m2 -- stoichiometry numbers
Results and Discussions
Structure and Stability
We have optimized the atomic positions and cell dimensions of the COFs in the experimentally
determined topologies Cell parameters in comparison with experimental values are given in Table 1
The calculated bond lengths are C-B = 150 C-C = 139-144 (COF-300) C(sp3)-C = 156 C-O = 142 B-
126
O = 140 Si-C = 188 Si-O = 186 C-N = 131-136 Aring These values agree within 6 error with the
experimental values34445
Mass densities (see Table 1) of COFs range between 02 to 05 g cm-3 and are smaller for the silicon at
the tetrahedral centers This implies that the presence of silicon atoms impacts more on the cell
volume than on the total mass That means the replacement of sp3 C with Si in COF-108 can change
its mass density to a slightly lower value To our best knowledge among all the natural or
synthesized crystals COF-108 has the lowest mass-weight
In order to validate our optimized structures we have simulated X-ray Diffraction of each COF and
compared them with the available experimental spectra (see Figure2) Almost all of the simulated
XRDs have excellent correlation in the peak positions with the experiment Only COF-300 shows
somehow significant differences in the intensities These differences may be attributed to the
presence of guest molecules in the synthesized COF-30045
Table 1 The calculated cell parameters [Aring] mass density ρ gcm-3
+ band gap Δ eV+ bulk modulus B GPa+
and formation energy EF [kJ mol-1
] for all the studied 3D COFs Experimental values are given in brackets
along with the cell parameters HOMO-LUMO gaps [eV] of the building blocks are also given in brackets
along with the band gaps
Structure Building
Blocks
Cell
parameters
ρ Δ B EF
COF-102 (C) TBPM 2707 (2717) 043 39 (43) 206 -14995
COF-103 (Si) TBPS 2817 (2825) 039 38 (41) 139 -14547
COF-105-C TBPM HHTP 4337 019 33 (43 34) 80 -18080
COF-105 (Si) TBPS HHTP 4444 (4489) 018 32 (41 34) 79 -17055
COF-108 (C) TBPM HHTP 2838 (2840) 017 32 (43 34) 37 -17983
COF-108-Si TBPS HHTP 2920 016 30 (41 34) 29 -18038
COF-202 (C) TBPM TBST 3047 (3010) 050 42 (43 139) 143 -7954
COF-202-Si TBPS TBST 3157 046 39 (41 139) 153 -7632
COF-300 (C) TAM TA 2932 925 049 23 (41 26) 144 -40286
127
(2828 1008)
COF-300-Si TAS TA 3039 959 044 23 (40 26) 133 -39930
tetra-(4-anilyl)silane
Figure 2 Simulated XRDs are compared with the experimental patterns from the literature COF-300
exhibits some differences between the simulated and experimental XRDs while others show reasonably
good match
The bulk modulus (B) is a measure of mechanical stability of the materials The calculated values of B
are given in Table 1 Compared to the force-field based calculation of bulk modulus by Schmid et
al61 these values are larger The bulk modulus of COF-105 and COF-108 are relatively small
compared with other COFs Considering that the two COFs differ only in the topology it may be
concluded that ctn nets are mechanically more stable compared to bor COFs with carbon atoms in
the center are mechanically more stable than those with silicon atoms Bulk modulus of COF-102
103 COF-202 and interpenetrated COF-300 are higher than many of the isoreticular MOFs62 and
comparable to IRMOF-6 (1241 GPa) MOF-5 (1537 GPa)63 bNon-interpenetrated COF-300 (single
framework dia-a topology43) has much lower bulk modulus of only 317 GPa
Calculated formation energies (Table 1) are given in terms of reaction energies according to Eq 2
Condensation of monomers to form bulk 3D structures is exothermic in all the cases supporting
reticular approach The presence of C or Si at the vertex center does not show any particular trend in
the formation energies We have calculated the formation energy of non-interpenetrated COF-300
(dia-a) to be -39346 kJ mol-1 which is comparable to the interpenetrated cases For a quantitative
comparison we have performed DFT calculations at the PBEDZP level as implemented in ADF code
on finite structures The obtained formation energies for COF-105-C COF-105 (Si) COF-108 (C) COF-
108-Si are -9944 -9968 -1245 -12436 respectively They are in a reasonable agreement with the
128
DFTB results A comparison between COF-105 and COF-108 suggests that bor nets are energetically
more favored than ctn nets
Electronic Properties
Band gaps (Δ) calculated for the 3D COFs are in the range of 23-42 eV (see Table 1) which show
their semiconducting nature similar to the hexagonal 2D COFs47 and MOFs62 The band gap
decreases with the increase of conjugated rings in the unit cell relative to the number of other atoms
Si atoms in comparison with carbon atoms at the tetrahedral centers give a relatively smaller Δ This
is evident from the partial density of states of C (sp3) in COF-102 and Si in COF-103 plotted in Figure3
Carbon atoms define the band edges of the PDOS (see Figure4) The band gaps of COF-105 and COF-
108 differ only slightly (see Figure5) which means that the band gap is nearly independent of the
topology This is because for each atom the coordination number and the neighboring atoms remain
the same in both ctn and bor networks albeit the topology difference DOS of non-interpenetrated
(dia-a) and 5-fold interpenetrated (dia-c5) COF-300 are plotted in Figure6 It may be concluded from
their negligible differences that interpenetration does not alter the DOS of a framework We have
shown similar results for 2D COFs47
Figure 3 Partial density of states (DOS) of sp3 C in COF-102 (top) and Si in COF-103 (bottom) The latter is
inverted for comparison The Fermi level EF is shifted to zero
129
Figure 4 Partial and total density of states of COF-103 The Fermi level EF is shifted to zero
Figure 5 Density of states of COF-108 (bor) and COF-105 (ctn) The two COFs differ only in the topology
130
Figure 6 Density of states of non-interpenetrated (dia-c1) and 5-interpenetrated (dia-c5) COF-300
We also have calculated HOMO-LUMO gaps of the molecular building blocks (see Table 1) In
comparison band gaps of COFs are slightly smaller than the smallest of the HOMO-LUMO gaps of the
building units
Conclusion
In summary we have calculated energetic mechanical and electronic properties of all the known 3D
COF using Density Functional Tight-Binding method Energetically formation of 3D COFs is favorable
supporting the reticular chemistry approach Mechanical stability is in line with other frameworks
materials eg MOFs and bulk modulus does not exceed 20 GPa Also all COFs are semiconducting
with band gaps ranging from 2 to 4 eV Band gaps are analogues to the HOMO-LUMO gaps of the
molecular building units We believe that this extensive study will define the place of COFs in the
broad area of nanoporous materials and the information obtained from the work will help to
strategically develop or modify porous materials for the targeted applications
131
Appendix F
Structure electronic structure and hydrogen adsorption capacity of porous aromatic frameworks
Binit Lukose Mohammad Wahiduzzaman Agnieszka Kuc and Thomas Heine
Prepared for publication
Abstract
Diamond-like porous aromatic frameworks (PAFs) form a new family of materials that contain only
carbon and hydrogen atoms within their frameworks These structures have very low mass densities
large surface area and high porosity Density-functional based calculations indicate that crystalline
PAFs have mechanical stability and properties similar to those of Covalent Organic Frameworks Their
exceptional structural properties and stability make PAFs interesting materials for hydrogen storage
Our theoretical investigations show exceptionally high hydrogen uptake at room temperature that
can reach 7 wt a value exceeding those of well-known metal- and covalent-organic frameworks
(MOFs and COFs)
Introduction
Porous materials have been widely investigated in the fields of materials science and technology due
to their applications in many important fields such as catalysis gas storage and separation template
materials molecular sensors etc1-3 Designing porous materials is nowadays a simple and effective
strategy following the approach of reticular chemistry4 where predefined building blocks are used to
132
predict and synthesize a topological organization in an extended crystal structure The most famous
and striking examples of such materials are Metal- and Covalent-Organic Frameworks (MOFs and
COFs)56 These new nanoporous materials have many advantages high porosity and large surface
areas lowest mass densities known for crystalline materials easy functionalization of building blocks
and good adsorption properties
Gas storage and separation by physical adsorption are very important applications of such
nanoporous materials and have been major subjects of science in the last two decades These
applications are based on certain physical properties namely presence of permanent large surface
area and suitable enthalpy of adsorption between the host framework and guest molecules
Attempts to produce materials with large internal surface area have been successful and some of the
notable accomplishments include the synthesis of MOF-2107 (BET surface area of 6240 m2 g-1 and
Langmuir surface area of 10400 m2 g-1) NU-1008 (BET surface area 6143 m2 g-1) or 3D COFs9 (BET
surface area 4210 m2 g-1 for COF-103)
More recently a new family of porous materials emerged So-called porous-aromatic frameworks
(PAFs) have surface areas comparable to MOFs eg PAF-110 has a BET surface area of 5640 m2 g-1 and
Langmuir surface area of 7100 m2 g-1 Since they are composed of carbon and hydrogen only they
have several advantages over frameworks containing heavy elements MOFs with coordination bonds
often suffer from low thermal and hydrothermal stability what might limit their applications on the
industrial scale The coordination bonds can be substituted with stronger covalent bonds as it was
realized in the case of COFs6 however this lowers significantly their surface areas comparing with
MOFs
Besides its exceptional surface area PAF-110 has high thermal and hydrothermal stability and
appears to be a good candidate for hydrogen and carbon dioxide storage applications PAFs have
topology of diamond with C-C bonds replaced by a number of phenyl rings (see Figure 1)
Experimentally obtained PAF-110 and PAF-1111 have eg one and four phenyl rings respectively
connected by tetragonal centers (carbon tetrahedral nodes) The simulated and experimental
hydrogen uptake capacities of such PAFs exceed the DOE target12
In this paper we have studied structural electronic and adsorption properties of PAFs using Density
Functional based Tight-Binding (DFTB) method1314 and Quantized Liquid Density Functional Theory
(QLDFT)15 We have found exceptionally high storage capacities at room temperature what makes
PAFs very good materials for hydrogen storage devices beyond well-known MOFs and COFs We have
compared our results to widely-used classical Grand Canonical Monte-Carlo (GCMC) calculations
reported in the literature We have also studied other properties of these materials such as
133
structural energetic electronic and mechanical We explored the structural variance of diamond
topology by individually placing a selection of organic linkers between carbon nodes This generally
changes surface area mass density and isosteric heat of adsorption what is reflected in the
adsorption isotherms
Methods
Using a conjugate-gradient method we have fully optimized the crystal structures (atomic positions
and unit cell parameters) of PAFs The calculations were performed using Dispersion-corrected Self-
consistent Charge density-functional based tight-binding (DFTB) method as implemented in the
deMonNano code1617 Periodic boundary conditions were applied to a (3x3x3) supercell thus
representing frameworks of the crystalline solid state Electronic density of states (DOS) have been
calculated using the DFTB+ code18 with k-point sampling where the k space was determined by
reaching convergence for the total energy according to the scheme proposed by Monkhorst and
Pack19
Hydrogen adsorption calculations have been carried out using QLDFT within the LIE-0 approximation
thus including many-body interparticle interactions and quantum effects implicitly through the
excess functional The PAF-H2 non-bonded interactions were represented by the classical Morse
atomic-pair potential Force field parameters were taken from Han et al20 who originally developed
them for studying hydrogen adsorption in COFs that incorporate similar building blocks as PAFs The
authors adopted the approach to the shapeless particle approximation of QLDFT These PAF-H2
parameters have been determined by fitting first-principles calculations at the second-order Moslashllerndash
Plesset (MP2) level of theory using the quadruple-ζ QZVPP basis set with correction for the basis set
superposition error (BSSE) QLDFT calculations use a grid spacing of 05 Bohr radii and a potential
cutoff of 5000 K
Results and Discussion
Design and Structure of PAFs
We have designed a set of PAFs by replacing C-C bonds in the diamond lattice by a set of organic
linkers phenyl biphenyl pyrene para-terphenyl perylenetetracarboxylic acid (PTCDA)
diphenylacetylene (DPA) and quaterphenyl (see Figure 1) We label the respective crystal structures
as PAF-phnl (PAF-301 in Ref 12) PAF-biph (PAF-302 in Ref 12) PAF-pyrn PAF-ptph(PAF-303 in Ref
12) PAF-PTCDA PAF-DPA and PAF-qtph (PAF-304 in Ref 12) respectively Such a design of
frameworks should result in materials with high stability due to the parent diamond-topology and
pure covalent bonding of the network The selected linkers differ in their length width and the
134
number of aromatic rings These should play an important role for hydrogen adsorption properties
aromatic systems exhibit large polarizabilities and interact with H2 molecules via London dispersion
forces Long linkers introduce high pore volume and low mas-weight to the network while wide
linkers offer large internal surface area and high heat of adsorption Hence long linkers are of
advantage for high gravimetric capacity while wide linkers enhance volumetric capacity Proper
optimization of the linker size should result in a perfect candidate for hydrogen storage applications
Figure 1 a) Schematic diagram of the topology of PAFs blue spheres and grey pillars represent carbon
tetragonal centers and organic linkers respectively b) Linkers used for the design of PAFs i) phenyl ii)
biphenyl iii) pyrene iv) DPA v) para-Terphenyl vi) PTCDA and vii) quaterphenyl
Selected structural and mechanical properties of the investigated PAF structures are given in Table 1
Frameworks created with the above mentioned linkers have mass densities that range from 085 g
cm-3 (for phenyl) to 01 g cm-3 (for quaterphenyl) The lowest mass-density obtained for a crystal
structure was for COF-108 with = 0171 g cm-39 Our results show that PAF-DPA and PAF-ptph have
mass densities close to the one of COF-108 while PAF-qtph with = 01 g cm-3 exhibits the lowest
for all the PAFs investigated in this study
While the large cell size and the small mass density of PAF-qtph are an advantage for high
gravimetric hydrogen storage capacity other PAFs (eg PAF-pyrn with wide linker) would
compromise gravimetric for high volumetric capacity As both of them are important for practical
applications a balance between them is crucial
Table 1 Calculated cell parameters a (a=b=c α=β=γ=60⁰) mass densities () formation energies (EForm) band
gaps () bulk moduli (B) and H2 accessible free volume and surface area of PAFs studied in the present work
In parenthesis are given HOMU-LUMO gaps of the corresponding saturated linkers
PAFs
a
(Aring)
ρ
(g cm-3)
EForm
(kJ mol-1)
Δ
(eV)
B
(GPa)
H2 accessible
free volume
H2 accessible
surface area
135
() (m2 g-1)
PAF-phnl 97 085 -121 47 (55) 360 35 2398
PAF-biphl 167 032 -122 36 (40) 132 73 5697
PAF-pyrn 166 042 -124 26 (28) 192 66 5090
PAF-DPA 210 019 -122 35 (37) 87 84 7240
PAF-ptph 237 016 -119 32 (33) 56 86 6735
PAF-PTCDA 236 024 -122 18 (19) 95 81 5576
PAF-qtphl 308 010 -119 29 (30) 35 91 7275
Energetic and Mechanical Properties
We have investigated energetic stability of PAFs by calculating their formation energies We regarded
the formation of PAFs as dehydrogenation reaction between saturated linkers and CH4 molecules
For a unit cell containing n carbon nodes and m organic linkers the formation energy (EForm) is given
by
( )
where Ecell EL and
are the total energies of the unit cell saturated linkers CH4 and H2
molecules respectively This excludes the inherent stability of linkers and represents the energy for
coordinating linkers with sp3 carbon atoms during diamond-topology formation All the formation
energies calculated in the present work are given in Table 1 Negative values indicate that the
formation of PAFs is exothermic The values per formula unit do not deviate significantly for different
PAF sizes and shapes
Although diamond is the hardest known material insertion of longer linkers diminishes its
mechanical strength to some extent In order to study the mechanical stability of PAFs we have
calculated their bulk moduli which is the second derivative of the cell energy with respect to the cell
volume (see Table 1) The highest value that we have obtained is for PAF-phnl with B = 36 GPa This is
over ten times smaller than the bulk modulus of pure diamond 514 GPa (calculated at the DFTB
level)21 and 442 GPa (experimental value)22 Bulk modulus should scale as 1V (V - volume) if all
bonds have the same strength We have plotted such a function for PAFs and other framework
136
materials23 in Figure 2 As PAFs have various types of CmdashC bonds there are minor deviations from
the perfect trend
Figure 2 Calculated bulk modulus B as function of the volume per atom PAFs are highlighted in blue and
compared with other framework materials Red curve denotes the fitting to 1V function (V - volume)
The stability of PAF-phnl is similar to that of Cu-BTC MOF and much higher than for other MOFs such
as MOF-5 -177 and -20524 and 3D COFs23 Subsequent addition of phenyl rings decreases B the
lowest value of 35 GPa has been found for PAF-qtph which is in the same order as for COF-108 In
general all the studied PAFs except for PAF-phnl have bulk moduli in line with 3D COFs25 When the
organic linker is extended in the width (cf PAF-biph and PAF-pyrn) the mechanical stability increases
Electronic Properties
All PAF structures are semiconductors similar to COFs The band gaps of PAFs range from 18 to 47
eV (see Table 1) Interesting trends may be observed from the band gaps of this iso-reticular series
In comparison with diamond which has a band gap of 55 eV it may be understood that subsequent
insertion of benzene groups between carbon atoms in diamond decreases the band gap This is easily
understood as the sp3 responsible for the semiconducting character become far apart with large
number of π-electrons in between which tend to close the gap More importantly the values of
band gaps are directly related to the HOMO-LUMO gaps of the corresponding saturated linkers
which are just slightly larger Also more extended linkers (cf PAF-biph and PAF-pyrn or PAF-ptph and
PAF-PTCDA) reduce the band gap
In general conjugated rings reduce the band gap more effectively than CmdashC bond networks (cf PAF-
DPA PAF-ptph or PAF-qtphl) The extreme cases for this behaviour are graphene which is metallic
137
and diamond which is an insulator Overall the band gap may be tuned by placing suitable linkers in
the diamond network Similar results have been reported for MOFs2627
We have calculated the electronic density of states (DOS) for all the PAF structures Figure 3 a shows
carbon DOS of PAFs arranged in such a way that the band gaps decrease going from the top to the
bottom of the figure PAFs with larger band gaps have larger population of energy states at the top of
valence band and bottom of conduction band whereas for linkers with smaller band gaps the
distribution of energy states is rather spread In order to study the impact of sp3 carbon atoms in the
DOS we plotted their contributions against the total carbon DOS in the cases of PAF-phnl and PAF-
pyrn as examples (see Figure 3 b and c) In both cases the sp3 carbons exhibit no energy states in the
band gap and in the close vicinity of band edges
Figure 3 (a) Carbon density of states of all calculated PAFs From top to bottom the value of band gap
decreases (b) and (c) projected DOS on sp3 C atoms of PAF-phnl and PAF-pyrn respectively The vertical
dashed line indicates Fermi level EF
Hydrogen Adsorption Properties
One of the potential applications of PAFs is hydrogen adsorption We have calculated the gravimetric
and volumetric capacities and analyzed them to understand the contributions of the linkers on the
138
hydrogen uptake in different range of temperatures and pressures H2 accessible free volume and
surface area of the PAFs are given in Table 1 In order to assess the excess adsorption capacity the
free pore volume is necessary In our simulation the free pore volume is defined to be that where
the H2-host interaction energy is less than 5000K This covers all sites where H2 is attracted by the
host structure and excludes the repulsion area close to the framework The solvent accessible surface
areas of the PAFs were determined by rolling a probe molecule along the van der Waals spheres of
the atoms A probe molecule of radius 148 Aring was used which is consistent with the Lennard-Jones
sphere of hydrogen and commonly used in various H2 molecular simulations28
Very large surface areas per mass unit have been observed for all PAFs except for PAF-phnl PAF-DPA
and PAF-qtph for example have 7240 and 7274 m2 g-1 H2 accessible surface area respectively For
comparison highly porous MOF-210 and NU-100 give values of 6240 and 6143 m2 g-1 BET surface
areas respectively determined from the experimental adsorption isotherms78 It is worth
mentioning that longer linkers expand the pore and increase the surface area per unit volume and
unit mass Wider linkers provide a higher surface area per unit volume however they possess larger
mass density and hence the surface area per unit mass gets lower
Figure 4 shows the gravimetric and volumetric total and excess adsorption isotherms of PAFs at 77K
The results show that the gravimetric (total and excess) H2 uptake is dependent on the linker length
The longest linker in the PAF-qtph structure gives the total and excess gravimetric capacity of 32 and
128 wt respectively which is larger than for highly porous crystalline 3D COFs29 These numbers
are calculated for the limit pressure of 100 bar For the shortest linker in PAF-phnl we have obtained
only around 25 wt for the same pressure Using grand canonical Monte-Carlo methods (GCMC)
Lan et al12 have predicted total and excess adsorption capacities of PAF-qtph of 224 and 84 wt
respectively The deviations in results are attributed to the differences in both methods where
different force fields are used to describe atom-atom interactions
The volumetric adsorption capacities lower with the size of the linker and the largest uptake we have
found for the PAF-pyrn (46 and 35 kg m-3 total and excess respectively) while very low values were
found for PAF-qtph (30 and 145 kg m-3 total and excess respectively) at 35-bar pressure It could be
predicted that for PAF-phnl the volumetric adsorption reaches the larges values however due to its
very compact crystal structure it reaches saturation at the low-pressure region and does not exceed
30 kg m-3 for the total uptake From the excess adsorption isotherms however PAF-phnl is the best
adsorbing system for low-pressure application as it gets saturated at 11 bar by adsorbing 266 kg m-3
of H2 Generally we have observed that PAFs with lower pore sizes exhibited saturation of volumetric
capacities at lower pressures
139
Figure 4 Calculated H2 uptake of PAFs at 77 K (a)-(b) gravimetric and (c)-(d) volumetric total (upper panel)
and excess (lower panel) respectively
We have also calculated the adsorption performance of PAFs at room temperature The gravimetric
total adsorption isotherms are shown in Figure 5 Similar to the results obtained for T=77 K PAF-
qtph exhibits the highest total gravimetric capacity at room temperature which is as high as 732 wt
at 100 bar Lan et al12 have predicted the uptake of 653 wt at 100 bar using GCMC simulations
These results exceed the 2015 DOE target of 55 wt 3031 On the other hand at more applicable
pressure of 5 bar the uptake reaches only 05 wt This suggests however that the delivery amount
(approximately the difference between the uptakes at 100 and 5 bar) is still more than the DOE
target Phenyl linker at room temperature does not exceed the gravimetric uptake of 1 wt at 100
bar
Figure 5 Calculated gravimetric total (left) and excess (right) H2 uptake of PAFs at 300 K
140
At 300K excess gravimetric capacity is rather low and does not exceed 075 wt even for large
pressure (see Figure 5)
Effects of interpenetration
Very often frameworks with large pores crystallize as interpenetrated lattices In most cases this is
an undesired fact due to reduction of the pore size and free volume For instance COF-300 which
has diamond topology was found to have 5-interpenetrated frameworks32
We have studied the effect of interpenetration in the case of PAF-qtph which has the largest pore
volume among the materials in this study Without any steric hindrance PAF-qtph may be
interpenetrated up to the order of four The two three and four interpenetrated networks are
named as PAF-qtph-2 PAF-qtph-3 and PAF-qtph-4 respectively and in each case the interpenetrated
structures possess translational symmetry Table 2 shows calculated mass densities and H2 accessible
free volume and surface area of non-interpenetrated and n-fold penetrated PAF-qtph Obviously the
mass density increases by one unit for each degree of interpenetration PAF-qtph has 91 of its
volume accessible for H2 which means that 9 of volume is occupied by the skeleton of the PAF
Hence each degree of interpenetration decreases the free volume by 9 H2 accessible surface area
per unit mass remains the same up to 3-fold interpenetration However PAF-qtph-4 results in much
less accessibility for H2
Table 2 Calculated mass densities () and H2 accessible free volume and surface area of non-interpenetrated
and n-fold interpenetrated PAF-qtph where n = 2 3 4
PAF
(g cm-3)
H2 accessible
free volume ()
H2 accessible
surface area
(m2 g-1)
PAF-qtph 010 91 7275
PAF-qtph-2 020 82 7275
PAF-qtph-3 030 73 7275
PAF-qtph-4 040 64 5998
Figure 6 shows the excess and total adsorption isotherms (gravimetric and volumetric) of non-
interpenetrated and interpenetrated PAF-qtph at 77 K With the increasing degree of
141
interpenetration saturation occurs at lower values of pressure This is due to the smaller pore size
resulted from the high-fold interpenetration The excess gravimetric capacity decreases by 18 - 2 wt
per each n-fold interpenetration a result of the decreasing free space around the skeleton It is to be
noted that the difference between the total and the excess adsorption capacities of PAF-qtph is quite
large however it decreases less for interpenetrated structures This is because the interpenetrated
frameworks occupy the free space that otherwise would be taken by hydrogen to increase the total
capacity but not the excess
Figure 6 Calculated H2 uptake at 77 K of non-interpenetrated and n-fold interpenetrated PAF-qtph with n = 2
3 4 (a)-(b) gravimetric and (c)-(d) volumetric total (lower panel) and excess (upper panel) respectively
On the other hand the interpenetrated frameworks have larger volumetric capacity what is easily
understandable due to the volume reduction Significant increase in excess volumetric capacity has
been obtained for PAF-qtph-2 in comparison with PAF-qtph whereas only slightly lower increase was
obtained for PAF-qtph-3 in comparison with PAF-qtph-2 (see Figure 6) PAF-qtph-4 exhibited even
lower increase compared to PAF-qtph-3 suggesting that only a certain degree of interpenetration is
appreciable Although non-interpenetrated PAF has a large amount of H2 gas in the bulk phase due
to the high pore volume its total volumetric uptake does not exceed that of the interpenetrated
PAFs
Almost linear gravimetric isotherms were obtained at room temperature for all studied PAFs
including the interpenetrated cases (see Figure 7) Large decrease of gravimetric capacity was noted
142
when the PAF was interpenetrated from around 7 wt down to 2 wt for 4-fold interpenetrated
PAF-qtph The excess gravimetric capacity however is the same independent of the n-fold
interpenetration and maximum of 06 wt was found for 100-bar pressure (see Figure 7)
Figure 7 Calculated gravimetric total (left) and excess (right) H2 uptake of non-interpenetrated and n-fold
interpenetrated PAF-qtph (n = 2 3 4) at 300 K
Conclusions
Following diamond-topology we have designed a set of porous aromatic frameworks PAFs by
replacing CmdashC bonds by various organic linkers what resulted in remarkably high surface area and
pore volume
Exothermic formation energies of PAFs calculated from the dehydrogenation of the linkers with CH4
indicate strong coordination of linkers in the diamond topology The studied PAFs exhibit bulk moduli
that are much smaller than diamond however in the same order as other porous frameworks such
as MOFs and COFs PAFs are semi-conductors with their band gaps dependent on the HOMO-LUMO
gaps of the linking molecules
Using quantized liquid density functional theory which takes into account inter-particle interactions
and quantum effects we have predicted hydrogen uptake capacities in PAFs At room temperature
and 100 bar the predicted uptake in PAF-qtph was 732 wt which exceeded the 2015 DOE target
At the same conditions other PAFs showed significantly lower uptakes At 77 K the PAFs with similar
pore sizes exhibited similar gravimetric uptake capacities whereas smaller pore size and larger
number of aromatic rings result in higher volumetric capacities In smaller pores saturation of excess
capacities occurred at lower pressures Interpenetrated frameworks largely decrease the amount of
hydrogen gas in the pores and increase the weight of the material however they are predicted to
have high volumetric capacities
143
References
(1) Eddaoudi M Moler D B Li H L Chen B L Reineke T M OKeeffe M Yaghi O M
Accounts of Chemical Research 2001 34 319
(2) Seo J S Whang D Lee H Jun S I Oh J Jeon Y J Kim K Nature 2000 404 982
(3) Ferey G Mellot-Draznieks C Serre C Millange F Accounts of Chemical Research 2005 38
217
(4) Yaghi O M OKeeffe M Ockwig N W Chae H K Eddaoudi M Kim J Nature 2003 423
705
(5) Eddaoudi M Kim J Rosi N Vodak D Wachter J OKeeffe M Yaghi O M Science 2002
295 469
(6) Cote A P Benin A I Ockwig N W OKeeffe M Matzger A J Yaghi O M Science 2005
310 1166
(7) Furukawa H Ko N Go Y B Aratani N Choi S B Choi E Yazaydin A O Snurr R Q
OKeeffe M Kim J Yaghi O M Science 2010 329 424
(8) Farha O K Yazaydin A O Eryazici I Malliakas C D Hauser B G Kanatzidis M G
Nguyen S T Snurr R Q Hupp J T Nature Chemistry 2010 2 944
(9) El-Kaderi H M Hunt J R Mendoza-Cortes J L Cote A P Taylor R E OKeeffe M Yaghi
O M Science 2007 316 268
(10) Ben T Ren H Ma S Cao D Lan J Jing X Wang W Xu J Deng F Simmons J M Qiu
S Zhu G Angewandte Chemie-International Edition 2009 48 9457
(11) Yuan Y Sun F Ren H Jing X Wang W Ma H Zhao H Zhu G Journal of Materials
Chemistry 2011 21 13498
(12) Lan J Cao D Wang W Ben T Zhu G Journal of Physical Chemistry Letters 2010 1 978
(13) Oliveira A F Seifert G Heine T Duarte H A Journal of the Brazilian Chemical Society
2009 20 1193
(14) Seifert G Porezag D Frauenheim T International Journal of Quantum Chemistry 1996 58
185
(15) Patchkovskii S Heine T Physical Review E 2009 80
(16) Heine T Rapacioli M Patchkovskii S Frenzel J Koester A M Calaminici P Escalante S
Duarte H A Flores R Geudtner G Goursot A Reveles J U Vela A Salahub D R deMon-nano edn ed
deMon 2009
(17) Zhechkov L Heine T Patchkovskii S Seifert G Duarte H A Journal of Chemical Theory
and Computation 2005 1 841
(18) BCCMS Bremen DFTB+ - Density Functional based Tight binding (and more)
(19) Monkhorst H J Pack J D Physical Review B 1976 13 5188
(20) Han S S Furukawa H Yaghi O M Goddard III W A Journal of the American Chemical
Society 2008 130 11580
(21) Kuc A Seifert G Physical Review B 2006 74
(22) Cohen M L Physical Review B 1985 32 7988
(23) Lukose B Kuc A Heine T manuscript in preparation 2012
(24) Lukose B Supronowicz B Petkov P S Frenzel J Kuc A Seifert G Vayssilov G N
Heine T physica status solidi (b) 2011
(25) Amirjalayer S Snurr R Q Schmid R Journal of Physical Chemistry C 2012 116 4921
(26) Gascon J Hernandez-Alonso M D Almeida A R van Klink G P M Kapteijn F Mul G
Chemsuschem 2008 1 981
(27) Kuc A Enyashin A Seifert G Journal of Physical Chemistry B 2007 111 8179
(28) Dueren T Millange F Ferey G Walton K S Snurr R Q Journal of Physical Chemistry C
2007 111 15350
(29) Furukawa H Yaghi O M Journal of the American Chemical Society 2009 131 8875
144
(30) US DOE Office of Energy Efficiency and Renewable Energy and The FreedomCAR and
Fuel Partnership 2009
httpwww1eereenergygovhydrogenandfuelcellsstoragepdfstargets_onboard_hydro_storage_explanatio
npdf
(31) US DOE USCAR Shell BP ConocoPhillips Chevron Exxon-Mobil T F a F P Multi-Year
Research Development and Demonstration Plan 2009
httpwww1eereenergygovhydrogenandfuelcellsmypppdfsstoragepdf
(32) Uribe-Romo F J Hunt J R Furukawa H Klock C OKeeffe M Yaghi O M Journal of the
American Chemical Society 2009 131 4570
145
Appendix G
A novel series of isoreticular metal organic frameworks realizing metastable structures by liquid phase epitaxy
Jinxuan Liu Binit Lukose Osama Shekhah Hasan Kemal Arslan Peter Weidler Hartmut
Gliemann Stefan Braumlse Sylvain Grosjean Adelheid Godt Xinliang Feng Klaus Muumlllen Ioan-
Bogdan Magdau Thomas Heine and Christof Woumlll
Prepared for publication
Highly crystalline molecular scaffolds with nm-sized pores offer an attractive basis for the fabrication
of novel materials The scaffolds alone already exhibit attractive properties eg the safe storage of
small molecules like H2 CO2 and CH41-4 In addition to the simple release of stored particles changes
in the optical and electronic properties of these nanomaterials upon loading their porous systems
with guest molecules offer interesting applications eg for sensors5 or electrochemistry6 For the
construction of new nanomaterials the voids within the framework of nanostructures may be loaded
with nm-sized objects such as inorganic clusters larger molecules and even small proteins a
process that holds great potential as for example in drug release7-8 or the design of novel battery
materials9 Furthermore meta-crystals may be prepared by loading metal nanoparticles into the
pores of a three-dimensional scaffold to provide materials with a number of attractive applications
ranging from plasmonics10-12 to fundamental investigations of the electronic structure and transport
properties of the meta-crystals13
146
In the last two decades numerous studies have shown that MOFs also termed porous coordination
polymers (PCPs) fulfill many of the aforementioned criteria Although originally developed for the
storage of hydrogen eg in tanks of automobiles MOFs have recently found more technologically
advanced applications as sensors5 matrices for drug release7-8 and as membranes for enantiomer
separation14 and for proton conductance in fuel cells15 Although pore-sizes available within MOFs1
are already sufficiently large to host metal-nanoclusters such as Au5516-17 the actual realization of
meta-crystals requires in addition to structural requirements a strategy for the controlled loading
of the nanoparticles (NPs) into the molecular framework Simply adding NPs to the solutions before
starting the solvothermal synthesis of MOFs has been reported18 but this procedure does not allow
for controlled loading The layer-by-layer growing process of SUFMOFs allows the uptake of
nanosized objects during synthesis including the fabrication of compositional gradients of different
NPs Those guest NPs can range from pure metal metal oxide or covalent clusters over one-
dimensional objects such as nanowires carbon or inorganic nanotubes to biological molecules such
as drugs or even small proteins If the loading happens during synthesis alternating layers of
different NPs can be realized
The introduction of procedures for epitaxial growth of MOFs on functionalized substrates19 was a
major step towards controlled functionalization of frameworks Epitaxial synthesis allowed the
preparation of compositional gradients20-21 and provided a new strategy to load nano-objects into
predefined pores
Unfortunately the LPE process has so far been only demonstrated for a fairly small number of
MOFs141922-23 none of which exhibits the required pore sizes In addition for the fabrication of meta-
crystals the architecture of the network should be sufficiently adjustable to realize pores of different
sizes There should also be a straightforward way to functionalize the framework itself in order to
tailor the interaction of the walls with the targeted guest objects Ideally such a strategy would be
based on an isoreticular series of MOFs in which the pore size can be determined by a choice from a
homologous series of ligands with different lengths1
Here we present a novel isoreticular MOF series with the required flexibility as regards pore-sizes
and functionalization which is well suited for the LPE process This class denoted as SURMOF-2 is
derived from MOF-2 one of the simplest framework architectures24-25 MOF-2 is based on paddle-
wheel (pw) units formed by attaching 4 dicarboxylate groups to Cu++ or Zn++-dimers yielding planar
sheets with 4-fold symmetry (see Fig 1) These planes are held together by strong
carboxylatemetal- bonds26 Conventional solvothermal synthesis yields stacks of pw-planes shifted
relative to each other with a corresponding reduction in symmetry to yield a P2 or C2 symmetry2427-
28
147
The relative shifts between the pw-planes can be avoided when using the recently developed liquid
phase epitaxy (LPE) method22 As demonstrated by the XRD-data shown in Fig 2 for a number of
different dicarboxylic acid ligands the LPE-process yields surface attached metal organic frameworks
(SURMOFs) with P4 symmetry29 except for the 26-NDC ligand which exhibits a P2 symmetry as a
result of the non-linearity of the carboxyl functional groups on the ligand The ligand PPDC
pentaphenyldicarboxylic acid with a length of 25 nm is to our knowledge the longest ligand which
has so far been used successfully for a MOF synthesis303030 A detailed analysis of the diffraction data
allowed us to propose the structures shown in Fig 1 This novel isoreticular series of MOFs hereafter
termed SURMOF-2 also consists of pw-planes which ndash in contrast to MOF-2 ndash are stacked directly
on-top of each other This absence of a lateral shift when stacking the pw-planes yields 1d-pores of
quadratic cross section with a diagonal up to 4 nm (see Fig 2) Importantly for the SURMOF-2 series
interpenetration is absent For many known isoreticular MOF series the formation of larger and
larger pores is limited by this phenomenon if the pores become too large a second or even a third
3d-lattice will be formed inside the first lattice yielding interpenetrated networks which exhibit the
expected larger lattice constant but with rather small pore sizes31-33 the hosting of NPs thus becomes
impossible For SURMOF-2 the particular topology with 1d-pores of quadratic cross section is not
compatible with the presence of a second interwoven network and as a result interpenetration is
suppressed
Although the solubility of some the long dicarboxylic acids ligands used for the SURMOF-2 fabrication
(TPDC QPDC and PPDC see Fig 1) is fairly small we encountered no problems in the LPE process
since already small concentrations of dicarboxylic acids are sufficient for the formation of a single
monolayer on the substrate the elementary step of the LPE synthesis process34-35 Even with the
longest dicarboxylic acid available to us PPDC growth occurred in a straightforward fashion and
optimization of the growth process was not necessary
The P4 structures proposed for the SURMOF-2 series except for the 26-NDC ligand differ markedly
from the bulk MOF-2 structures obtained from conventional solvothermal synthesis2427-28 To
understand this unexpected difference and in particular the absence of relative shifts between the
pw-planes in the proposed SURMOF-2 structure we performed extensive quantum chemical
calculations employing approximate density-functional theory (DFT) in this case London dispersion-
corrected self-consistent charge density-functional based tight-binding (DFTB)36-38 to a periodic
model of MOF-2 and its SURMOF derivatives
Periodic structure DFTB calculations (Fig S1) confirmed that the P2m structure proposed by Yaghi
et al for the bulk MOF-2 material24 is indeed the energetically favorable configuration for MOF-2
while the C2m structure that has also been reported for bulk MOF-227-28 is slightly higher in energy
148
(41 meV per formula unit see Table 1) The optimum interlayer distance between the pw-planes in
the P4 structure is 58 Aring in very good agreement with the experimental value of 56 Aring (as obtained
from the in-plane XRD-data see Fig S2) It is important to note that at that distance the linkers
cannot remain in rectangular position as in MOF-2 but are twisted in order to avoid steric hindrance
and to optimize linker-linker interactions
The P4mmm structure as proposed for SURMOF-2 is less stable by 200 meV per formula unit as
compared to the P2m configuration Although the crystal energy for the P4 stacking is substantially
smaller than for P2 packing simulated annealing (DFTB model using Born-Oppenheimer Molecular
Dynamics) carried out at 300K for 10 ps demonstrated that the P4mmm structure corresponds to a
local minimum This observation is confirmed by the calculation of the stacking energy of MOF-2
where a shifting of the layers is simulated along the P2-P4-C2 path (Figure 3) On the basis of these
calculations we thus propose that SURMOF-2 adopts this metastable P4 structure
In Table 1 the interlayer interaction energies are summarized They range from 590 meV per formula
unit with respect to fully dissociated MOF-2 layers for Cu2(bdc)2 and successively increase for longer
linkers of the same type reaching 145 eV per formula unit for Cu2(ppdc)2 indicating that the linkers
play an important role in the overall stabilization of SURMOF-2 with large pore sizes Even stronger
interlayer interactions are found for different linker topologies (PPDC) A detailed computational
analysis of the role of the individual building blocks of SURMOF-2 for the crystal growth and
stabilization will be published elsewhere
The flat well-defined organic surfaces used as the templating (or nucleating) substrate in the LPE
growth process provide a satisfying explanation for why SURMOF-2 grows with the highly
symmetrical P4 structure instead of adopting the energetically more favorable bulk P2 symmetry3439
The carboxylic acid groups exposed on the surface force the first layer of deposited Cu-dimers into a
coplanar arrangement which is not compatible with the bulk P2 (or C2) structure but rather
nucleates the P4 packing In turn the carboxylate-groups in the first layer of deposited dicarboxylic
acids provide the same constraints for the next layer (see Fig 1) As a result of the layer-by-layer
method employed for further SURMOF-2 growth the same boundary conditions apply for all
subsequent layers The organic surface thus acts as a seed structure to yield the metastable P4
packing not an unusual motif in epitaxial growth40
The calculations allow us to predict that it will be possible to grow SURMOF structures with even
larger linker molecules as used here The stacking will further stabilize the P4 symmetry as the
interaction energy per formula unit is more and more dominated by the linkers (Table 1) At present
149
we cannot predict the maximum possible size of the pores but linker PPDC (see Fig 2) is thus far
unmatched as a component in non-interpenetrated framework structures
Acknowledgement
We thank Ms Huumllsmann Bielefeld University for the preparation of P(EP)2DC Financial support by
Deutsche Forschungsgemeinschaft (DFG) within the Priority Program Metal-Organic Frameworks
(SPP 1362) is gratefully acknowledged
Methods
Computational Details
All structures were created using a preliminary version of our topological framework creator
software which allows the creation of topological network models in terms of secondary building
units and their replacement by individual molecules to create the coordinates of virtually any
framework material The generated starting coordinates including their corresponding lattice
parameters have been hierarchically fully optimized using the Universal Force Field (UFF)41 followed
by the dispersion-corrected self-consistent-charge density-functional based tight-binding (DFTB)
method36-3842-43 that we have recently validated against experimental structures of HKUST-1 MOF-5
MOF-177 DUT-6 and MIL-53(Al) as well as for their performance to compute the adsorption of
water and carbon monoxide37 For all calculations we employed the deMonNano software44444444
We have chosen periodic boundary conditions for all calculations and the repeated slab method has
been employed to compute the properties of the single layers in order to evaluate the stacking
energy A conjugate-gradient scheme was employed for geometry optimization of atomic
coordinates and cell dimensions The atomic force tolerance was set to 3 x 10-4 eVAring
The confined conditions at the SURMOF-substrate interface was modeled by fixing the corresponding
coordinate in the computer simulations All calculated structures have been substantiated by
simulated annealing calculations where a Born-Oppenheimer Molecular Dynamics trajectory at 300K
has been computed for 10 ps without geometry constrains All structures remained in P4 topology
Experimental methods
The SURMOF-2 samples were grown on Au substrates (100-nm Au5-nm Ti deposited on Si wafers)
using a high-throughput approach spray method45 The gold substrates were functionalized by self-
assembled monolayers SAMs of 16-mercaptohexadecanoic acid (MHDA) These substrates were
mounted on a sample holder and subsequently sprayed with 1 mM of Cu2 (CH3COO)4middotH2O ethanol
solution and ethanolic solution of SURMOF-2 organic linkers (01 mM of BDC 26-NDC BPDC and
150
saturated concentration of TPDC QPDC PPDC P(EP)2DC) at room temperature After a given
number of cycles the samples were characterized with X-ray diffraction (XRD)
Figure 1 Schematic representation of the synthesis and formation of the SURMOF-2 analogues
151
Figure 2 Out of plane XRD data of Cu-BDC Cu-26-NDC Cu-BPDC Cu-TPDC Cu-QPDC Cu-P(EP)2DC and Cu-PPDC (upper left) schematic representation (upper right) and proposed structures of SURMOF-2 analogues (lower panel) All the SURMOF-2 are grown on ndashCOOH terminated SAM surface using the LPE method
152
Figure 3 Energy for the relative shift of each layer in two different directions horizontal (P4 to P2) and diagonal (P4 to C2) at a fixed interlayer distance of 56 Aring Structures have been partially optimized and energies are given per formula unit with respect to fully dissociated planes
Supporting information
Synthesis of organic linkers
(1) para-terphenyldicarboxylic acid (TPDC)
To a solution of para-terphenyl (1500 g 651 mmol 1 eq) and oxalyl chloride (3350 mL 3900 mmol
6 eq) in carbon disulfide (30 mL) at 0degC (ice bath) was added aluminium chloride (1475 g 1106
mmol 17 eq) After 1h stirring additional aluminium chloride (0870 g 651 mmol 1 eq) was added
the ice bath was removed and the reaction mixture was stirred at room temperature for 24h The
mixture was poured into crushed ice and stirred until the dark-brown mixture turned to yellow
Carbon disulfide was removed under reduced pressure and the resulting aqueous suspension was
filtered The solid was washed with diluted hydrochloric acid then with diethyl ether and dried
under vacuum overnight with an oil bath at 50degC Pale yellow solid yield 2028 g (98)
(2) para-quaterphenyldicarboxylic acid (QPDC)
153
To a solution of para-quaterphenyl (1000 g 326 mmol 1 eq) and oxalyl chloride (1680 mL 1956
mmol 6 eq) in carbon disulfide (20 mL) at 0degC (ice bath) was added aluminium chloride (0740 g 555
mmol 17 eq) After 1h stirring additional aluminium chloride (0435 g 326 mmol 1 eq) was added
the ice bath was removed and the reaction mixture was stirred at room temperature for 24h The
mixture was poured into crushed ice and stirred until the dark-brown mixture turned to yellow
Carbon disulfide was removed under reduced pressure and the resulting aqueous suspension was
filtered The solid was washed with diluted hydrochloric acid then with diethyl ether and dried
under vacuum overnight with an oil bath at 50degC Yellow solid yield 1186 g (92)
(3) P(EP)2DC
The synthesis of the P(EP)2DC-linker has been described in Ref 46
(4) para-pentaphenly dicarboxylic acid (PPDC)
Scheme 1 Synthesis of para-pentaphenly dicarboxylic acid i) Pd(PPh3)4 Na2CO3 toluenedioxane H2O 85oC ii) MeOH 6M NaOH HCl
para-pentaphenly dicarboxylic acid (H2L) was synthesized via Suzuki coupling of 44rdquo-dibromo-p-
terphenyl and 4-methoxycarbonylphenylboronic acid (Scheme 1) 44rdquo-dibromo-p-terphenyl (776 mg
200 mmol) 4-methoxycarbonylphenylboronic acid (108 g 60 mmol) and Na3CO3 (170 g 160 mmol)
were added to a degassed mixture of toluene14-dioxane (50 mL 221) under Ar [Pd(PPh3)4] (116
mg 010 mmol) was added to the mixture and heated to 85 degC for 24 hours under Ar The reaction
mixture was cooled to room temperature The precipitate was collected by filtration washed with
water methanol and used for next reaction without further purification The final product H4L was
obtained by hydrolysis of the crude product under reflux in methanol (100 mL) overnight with 6M
aqueous NaOH (20 mL) followed by acidification with HCl (conc) After removal of methanol the
final product was collected by filtration as a grey solid (078 g 83 yield) 1H NMR (d6-DMSO
250MHz 298K) δ (ppm) 805 (d J = 75Hz 4H) 789-783 (m 12H) 750 (d J = 75Hz 4H) 13C NMR
cannot be clearly resolved due to poor solubility MALDI-TOF-MS (TCNQ as matrix) mz 47002
cacld 47051 Elemental analysis Calculated C 8169 H 471 Found C 8163 H 479
Br Br MeOOC B
OH
OH
+
COOMe
COOMe
COOH
COOH
i ii
154
Figure S1 MOF-2 in P4 (as reported) P2 and C2 symmetry
155
Figure S2 In-plane XRD data of Cu-BDC Cu-26-NDC Cu-BPDC Cu-TPDC Cu-QPDC and Cu-P(EP)2DC All the
SURMOF-2 are grown on ndashCOOH terminated SAM surface using the LPE method The position of (010) plane
represents the layer distance
Table 1 Stacking energy and geometries of P4 SURMOF-2 derivatives
Symmetry a= c b Stacking Energy
Cu2(bdc)2 C2 1119 50 -076
Cu2(bdc)2 P2 1119 54 -08
Cu2(bdc)2 P4 1119 58 -059
156
Cu2(ndc)2 P2 1335 56 -04
Cu2(bpdc)2 P4 1549 59 -068
Cu2(tpdc)2 P4 1984 59 -091
Cu2(qpdc)2 P4 2424 59 -121
Cu2(P(EP)2DC)2 P4 2512 52 -173
Cu2(ppdc)2 P4 2859 59 -145
Stacking energies (DFTB level in eV) of SURMOFs The energies were calculated within periodic
boundary conditions and are given per formula unit
References
1 Eddaoudi M et al Systematic design of pore size and functionality in isoreticular MOFs and
their application in methane storage Science 295 469-472 (2002)
2 Rosi N L et al Hydrogen storage in microporous metal-organic frameworks Science 300
1127-1129 (2003)
3 Rowsell J L C amp Yaghi O M Metal-organic frameworks a new class of porous materials
Microporous and Mesoporous Materials 73 3-14 (2004)
4 Wang B Cote A P Furukawa H OKeeffe M amp Yaghi O M Colossal cages in zeolitic
imidazolate frameworks as selective carbon dioxide reservoirs Nature 453 207-U206 (2008)
5 Lauren E Kreno et al Metal-Organic Framework Materials as Chemical Sensors Chemical
Reviews 112 1105-1124 (2012)
6 Dragasser A et al Redox mediation enabled by immobilised centres in the pores of a metal-
organic framework grown by liquid phase epitaxy Chemical Communications 48 663-665
(2012)
7 Horcajada P et al Metal-organic frameworks as efficient materials for drug delivery
Angewandte Chemie-International Edition 45 5974-5978 (2006)
8 Horcajada P et al Flexible porous metal-organic frameworks for a controlled drug delivery
Journal of the American Chemical Society 130 6774-6780 (2008)
9 Hurd J A et al Anhydrous proton conduction at 150 degrees C in a crystalline metal-organic
framework Nature Chemistry 1 705-710 (2009)
10 Kreno L E Hupp J T amp Van Duyne R P Metal-Organic Framework Thin Film for Enhanced
Localized Surface Plasmon Resonance Gas Sensing Analytical Chemistry 82 8042-8046
(2010)
11 Lu G et al Fabrication of Metal-Organic Framework-Containing Silica-Colloidal Crystals for
Vapor Sensing Advanced Materials 23 4449-4452 (2011)
157
12 Lu G amp Hupp J T Metal-Organic Frameworks as Sensors A ZIF-8 Based Fabry-Perot Device
as a Selective Sensor for Chemical Vapors and Gases Journal of the American Chemical
Society 132 7832-7833 (2010)
13 Mark D Allendorf Adam Schwartzberg Vitalie Stavila amp Talin A A Roadmap to
Implementing MetalndashOrganic Frameworks in Electronic Devices Challenges and Critical
Directions European Journal of Chemistry (2011)
14 Bo Liu et al Enantiopure Metal-Organic Framework Thin Films Oriented SURMOF Growth
and Enantioselective Adsorption Angewandte Chemie-International Edition 51 817-810
(2012)
15 Sadakiyo M Yamada T amp Kitagawa H Rational Designs for Highly Proton-Conductive
Metal-Organic Frameworks Journal of the American Chemical Society 131 9906-9907 (2009)
16 Ishida T Nagaoka M Akita T amp Haruta M Deposition of Gold Clusters on Porous
Coordination Polymers by Solid Grinding and Their Catalytic Activity in Aerobic Oxidation of
Alcohols Chemistry-a European Journal 14 8456-8460 (2008)
17 Esken D Turner S Lebedev O I Van Tendeloo G amp Fischer R A AuZIFs Stabilization
and Encapsulation of Cavity-Size Matching Gold Clusters inside Functionalized Zeolite
Imidazolate Frameworks ZIFs Chemistry of Materials 22 6393-6401 (2010)
18 Lohe M R et al Heating and separation using nanomagnet-functionalized metal-organic
frameworks Chemical Communications 47 3075-3077 (2011)
19 Shekhah O et al Step-by-step route for the synthesis of metal-organic frameworks Journal
of the American Chemical Society 129 15118-15119 (2007)
20 Furukawa S et al A block PCP crystal anisotropic hybridization of porous coordination
polymers by face-selective epitaxial growth Chemical Communications 5097-5099 (2009)
21 Shekhah O et al MOF-on-MOF heteroepitaxy perfectly oriented [Zn(2)(ndc)(2)(dabco)](n)
grown on [Cu(2)(ndc)(2)(dabco)](n) thin films Dalton Transactions 40 4954-4958 (2011)
22 Shekhah O et al Controlling interpenetration in metal-organic frameworks by liquid-phase
epitaxy Nature Materials 8 481-484 (2009)
23 Zacher D et al Liquid-Phase Epitaxy of Multicomponent Layer-Based Porous Coordination
Polymer Thin Films of [M(L)(P)05] Type Importance of Deposition Sequence on the Oriented
Growth Chemistry-a European Journal 17 1448-1455 (2011)
24 Li H Eddaoudi M Groy T L amp Yaghi O M Establishing microporosity in open metal-
organic frameworks Gas sorption isotherms for Zn(BDC) (BDC = 14-benzenedicarboxylate)
Journal of the American Chemical Society 120 8571-8572 (1998)
25 Mueller U et al Metal-organic frameworks - prospective industrial applications Journal of
Materials Chemistry 16 626-636 (2006)
158
26 Shekhah O Wang H Zacher D Fischer R A amp Woumlll C Growth Mechanism of Metal-
Organic Frameworks Insights into the Nucleation by Employing a Step-by-Step Route
Angewandte Chemie-International Edition 48 5038-5041 (2009)
27 Carson C G et al Synthesis and Structure Characterization of Copper Terephthalate Metal-
Organic Frameworks European Journal of Inorganic Chemistry 2338-2343 (2009)
28 Clausen H F Poulsen R D Bond A D Chevallier M A S amp Iversen B B Solvothermal
synthesis of new metal organic framework structures in the zinc-terephthalic acid-dimethyl
formamide system Journal of Solid State Chemistry 178 3342-3351 (2005)
29 Arslan H K et al Intercalation in Layered Metal-Organic Frameworks Reversible Inclusion of
an Extended pi-System Journal of the American Chemical Society 133 8158-8161 (2011)
30 The MOF with the largest pore size recorded so far MOF-200 (Furukawa H et al Ultrahigh
Porosity in Metal-Organic Frameworks Science 329 424-428 (2010)) used a (trivalent)
444-(benzene-135-triyl-tris(benzene-41-diyl))tribenzoate (BBC) ligand The carboxylic
acid-to carboxylic acid distance is 20 nm compared to 25 nm in case of PPDC The cage size
in MOF-200 amounts to 18 nm by 28 nm clearly smaller than the 1d-channels in the PPDC
SURMOF-2 that are 28 nm by 28 nm
31 Batten S R amp Robson R Interpenetrating nets Ordered periodic entanglement
Angewandte Chemie-International Edition 37 1460-1494 (1998)
32 Snurr R Q Hupp J T amp Nguyen S T Prospects for nanoporous metal-organic materials in
advanced separations processes Aiche Journal 50 1090-1095 (2004)
33 Yaghi O M A tale of two entanglements Nature Materials 6 92-93 (2007)
34 Shekhah O Liu J Fischer R A amp Woumlll C MOF thin films existing and future applications
Chemical Society Reviews 40 1081-1106 (2011)
35 Zacher D Shekhah O Woumlll C amp Fischer R A Thin films of metal-organic frameworks
Chemical Society Reviews 38 1418-1429 (2009)
36 Elstner M et al Self-consistent-charge density-functional tight-binding method for
simulations of complex materials properties Physical Review B 58 7260-7268 (1998)
37 Lukose B et al Structural properties of metal-organic frameworks within the density-
functional based tight-binding method Physica Status Solidi B-Basic Solid State Physics 249
335-342 (2012)
38 Zhechkov L Heine T Patchkovskii S Seifert G amp Duarte H A An efficient a Posteriori
treatment for dispersion interaction in density-functional-based tight binding Journal of
Chemical Theory and Computation 1 841-847 (2005)
159
39 Zacher D Schmid R Woumlll C amp Fischer R A Surface Chemistry of Metal-Organic
Frameworks at the Liquid-Solid Interface Angewandte Chemie-International Edition 50 176-
199 (2011)
40 Prinz G A Stabilization of bcc Co via Epitaxial Growth on GaAs Physical Review Letters 54
1051-1054 (1985)
41 Rappe A K Casewit C J Colwell K S Goddard W A amp Skiff W M UFF a full periodic
table force field for molecular mechanics and molecular dynamics simulations Journal of the
American Chemical Society 114 10024-10035 (1992)
42 Seifert G Porezag D amp Frauenheim T Calculations of molecules clusters and solids with a
simplified LCAO-DFT-LDA scheme International Journal of Quantum Chemistry 58 185-192
(1996)
43 Oliveira A F Seifert G Heine T amp Duarte H A Density-Functional Based Tight-Binding an
Approximate DFT Method Journal of the Brazilian Chemical Society 20 1193-1205 (2009)
44 deMonNano v 2009 (Bremen 2009)
45 Arslan H K et al High-Throughput Fabrication of Uniform and Homogenous MOF Coatings
Adv Funct Mater 21 4228-4231 (2011)
46 Schaate A et al Porous Interpenetrated Zirconium-Organic Frameworks (PIZOFs) A
Chemically Versatile Family of Metal-Organic Frameworks Chemistry-a European Journal 17
9320-9325 (2011)
160
Appendix H
Linker guided metastability in templated Metal-Organic Framework-2 derivatives (SURMOFs-2)
Binit Lukose Gabriel Merino Christof Woumlll and Thomas Heine
Prepared for publication
INTRODUCTION
The molecular assembly of metal-oxides and organic struts can provide a large number of network
topologies for Metal-Organic Frameworks (MOFs)1-3 This is attained by the differences in
connectivity and relative orientation of the assembling units Within each topology replacement of a
building unit by another of same connectivity but different size leads to what is known as isoreticular
alteration of pore size The structure of MOFs in principle can be formed into the requirement of
prominent applications such as gas storage4-8 sensing910 and catalysis11-13 The structural
divergence and the performance can be further increased by functionalizing the organic linkers1415
In MOFs linkers are essential in determining the topology as well as providing porosity A linker
typically contains single or multiple aromatic rings the orientation of which normally undergoes
lowest repulsion from the coordinated metal-oxide clusters providing the most stable symmetry for
the bulk material We encounter for the first time a situation that the orientation of the linker
provides metastability in an isoreticular series of MOFs Templated MOF-2 derivatives (surface-MOF-
2 or simply SURMOFs-2)16 show this extraordinary phenomenon While bulk MOF-2 is determined to
be stable in P217 or C218-20 symmetry we found the existence of SURMOFs-2 in P4 symmetry
161
(Figure 1)16 Our theoretical approach to this problem identifies the role of the linkers in providing
P4 geometry the status of a local energy-minimum
MOF-2 derivatives are two-dimensional lattice structures composed of dimetal-centered four-fold
coordinated paddlewheels and linear organic linkers Solvothermally synthesized archetypal MOF-2
had transition metal (Zn or Cu) centers in the connectors and benzenedicarboxylic acid linkers17 The
derivatives under our investigation contain paddlewheels with Cu dimers and benzenedicarboxylic
acid (BDC) naphthalenedicarboxylic acid (NDC) biphenyldicarboxylic acid (BPDC)
triphenyldicarboxylic acid (TPDC) quaterphenyldicarboxylic acid (QPDC) P(EP)2DC21 and
pentaphenyldicarboxylic acid (PPDC) linkers pertaining to isoreticular expansion of pore size16 The
four-fold connectivity of the Cu dimers together with linearity of the linkers gives rise to layers with
quadratic (square) topology The interlayer separation d is typically much more than that of
graphite or 2D COFS22-24 because all the atoms in one layer do not lie in one plane
In bulk form the nearest layers are shifted to each other either towards one of the four linkers
(P2)171920 or diagonally towards one of the four surrounding pores (C2)18-20 in effort to reduce
the repulsion of alike charges in the adjacent paddlewheels when they are in eclipsed form (P4)
(Figure 1) The metal-dimers often show high reactivity which results in attracting water or
appropriate solvents in their axial positions The stacking along the third axis is typically through
interlayer interactions and through hydrogen bonds established between the solvents or between
the solvents and oxygen atoms in the paddlewheel The lateral shift of layers is inevitable without
additional supports Coordination of pillar molecules such as 14-diazabicyclo[222]octane (dabco) or
bipyridine in the axis of metal dimers between the adjacent layers25 is a practical mean to avoid
layer-offset however with the change of MOF dimensionality
Figure 1 Monolayer of MOF-2 and three kinds of layer packing -P4 P2 and C2- of periodic MOF-2
162
Alternate to solvothermal synthesis a step-by-step route may be enabled for the synthesis of
MOFs26-28 It adopts liquid phase epitaxy (LPE) of MOF reactants on substrate-assisted self-assembled
monolayers This is achieved by alternate immersion of the template in metal and ligand precursors
for several cycles This allows controlled growth of MOFs on surfaces The latest outcomes of this
method are the surface-standing MOF-2 derivatives (SURMOF-2s) This isoreticular SURMOF series
has linkers of different lengths (as given above) The cell dimensions that correspond to the length of
the pore walls range from 1 to 28 nm The diagonal pore-width of one of the MOFs (PPDC) accounts
to 4 nm
After evacuation of solvents (water in the present case) X-ray diffraction patterns were taken in
directions both perpendicular (scans out of plane) and parallel (scans in-plane) to the substrate
surface The presence of only one peak -(010)- in the perpendicular scan implies that the layers
orient perpendicular to the surface and the corresponding angle gives the cell parameter a (or c) In
the latter case the peaks - (100) and (001) ndash are identical and this rules out the possibility of layer-
offset Hence the symmetry of the MOFs was determined to be P4 These peaks give rise to the cell
parameter c which is nothing but the interlayer distance d This value (56 Aring) is relatively small for
P4 geometry to have water molecules in the axial position of paddlewheels Presence of some water
molecules observed using Infrared Reflection Absorption Spectroscopy (IRRAS) might be located near
paddlewheels but exposed to the pore The new observations with the SURMOFs are very intriguing
in the context that P4 symmetry appears to be impossible for solvothermally produced MOF-2
We have primarily reported that the P4 geometry of SURMOF-2 exists as a local energy-minimum16
The verification was made using an approximate method of density functional theory (DFT) which is
London dispersion-corrected self-consistent charge density-functional based tight-binding (DFTB) In
the bulk form absolute energies of P2 and C2 are 210 and 170 meV respectively higher than P4 per
a formula unit which is equivalent to the unit cell For SURMOF the barrier for leaving P4 was nearly
50 meV per formula unit It requires further analysis to unravel the reasons for this unusual
metastability We therefore performed an extensive set of quantum chemical calculations on the
composition of the constituent building units The procedure involves defining SURMOF geometry
and analyzing the translations of individual layers
The major individual contributions to the total energy are the interaction between the paddlewheel
units (favouring P2) London dispersion interaction between tilted linkers (favouring P4) and energy
to tilt the linkers In the iso-reticular series of SURMOF-2 the differences arise only in the
163
contributions from the linkers Hence we performed an extensive study only on the smallest of all
linkers- BDC A scaling might be appropriate for other linkers
RESULTS AND DISCUSSION
In solvothermal synthesis of layered MOFs it is assumed that the nucleation process is associated
with the interaction between two connectors This is rationalized by the fact that two paddlewheels
show the strongest possible noncovalent interaction between the individual MOF building blocks
present in MOF-2 As the orientation of the paddlewheels is restricted by the linkers we studied the
stacking of adjacent layers by determining the attraction of two adjacent interacting paddlewheels
upon their respective offsets Thus we investigated the geometries corresponding to lateral
displacements between the known stable arrangements of the bulk structure ie P4-to-P2 and P4-
to-C2 (Figure 2) In the model calculation the two paddlewheels were shifted from D4h symmetry to
two directions that respectively correspond to P2 and C2 symmetries in the bulk The minima along
the two shifts are referred as min-I and min-II and are having C2h symmetry It is interesting to note
that the interaction is in all cases attractive If only the paddlewheels are studied the D4h
configuration where both axes are concentric can be interpreted as transition state between the
two minima configurations P2 or C2 Comparing the two minima min-I corresponding to MOF-2 in
P2 symmetry indicates the formation of two CuO bonds while for min-II the reactive Cu centers do
not participate in the interlayer bonding
Formation of the diagonally shifted C2 bulk geometry for Zn and Cu based MOF-2 as reported in the
literature18-20 possibly is due to the presence of large solvent molecules such as DMF that
coordinate to the free Cu centers the paddlewheels
Figure 2 Displacements of two paddlewheels from D4h symmetry (corresponding to P4) towards minima that correspond to P2 and C2 bulk symmetries
164
To gain further insight on type of interactions for the three paddlewheel arrangements as found in
the bulk (Figure 3) we performed the topological analysis of the electron density for each
structure2930 An inspection of the paddlewheel shows that the electron density of the Cu-O bond has
a (relatively small) value of 0042 au thus showing only weak character In the forms related to P4
and C2 bulk symmetries all (3-1) critical points between the two paddlewheels show very small
density values (0004 au and less) In the P2 structure it is apparent the formation of a four-
membered Cu-O-Cu-O ring across the paddlewheel Note that the Cu-O interactions inside the
paddlewheel weaken (from 0042 to 0031 au) and two intermolecular Cu-O interactions with a
density value of 0024 au stabilize the P2 orientation These links if formed during nucleation will
be regularly repeated during the MOF-2 formation in solvothermal synthesis and due to its strong
binding of -08 eV they are the main reason for the stabilization of the P2 phase If the paddlewheels
are packed in P4 symmetry there must be additional means of stabilization present and that may
only arise from the linkers Moreover one should note that in order to reach the P4 symmetry a
layer-by-layer growth process is required as otherwise the linkers will readily stack as in the P2 bulk
form
165
Figure 3 Topological analysis of the electron density of fully optimized P4 (top left) C2 (top right) and P4 (bottom) structures Atom (grey) (3-1) (red) and (3+1) (green) critical points along with their charge densities are shown
The optimized geometry of a single-layer MOF-2 is shown in Figure 1 Note that the phenyl rings of
the linkers are oriented perpendicular to the MOF plane Contrary to graphene or 2D COFs the more
complex structure of MOF-2 layers may become subject to change upon the interlayer interactions
This is in particular important for the P4 stacking as reported for SURMOF-2 The interaction energy
of two linkers and two benzene rings as oriented in the monolayer has been computed as function
of the interlayer distance (Figure 4) At the experimental interlayer distance of 56 Aring the linkers are
so close that they repel each other strongly and stacking the monolayer structure at the
experimental interlayer distance would introduce an energy penalty of 08 eV per linker
It would not be exotic if we assume that the anchoring of layers on the substrate plays an important
role in setting d A supporting argument is the fact that all the SURMOFs in the iso-reticular series
have the same d An additional point is that the comparatively wider linkers NDC and LM do not
create any difference in the interlayer distance
166
Figure 4 Energy as a function of interlayer distance [d] in case of two paddlewheels and two benzene molecules The energies are given with respect to the complete dissociation of the building blocks
The best adaptation for the linkers in a closer alignment to avoid the repulsion is to partially rotate
the phenyl rings This reduces the Pauli repulsion and at the same time increases the attractive
London dispersion between the linkers However the rotation is energetically penalized by 06 eV as
accordance with similar calculations found in the literature31 and is with the same order of Zn4O-
tetrahedron clusters of the IRMOFs3233
Figure 5 Energy as a function of rotation of linker between two saturated paddlewheels The dihedral angle O-C1-C2-C3 θ between the linker and the carboxylic part in the paddlewheel Values are with respect to the ground state θ = 0⁰
To model the conditions in SURMOF-2 we need to combine the intermolecular interaction of the
linkers with the barrier associated to the rotation of the linker between two paddlewheel units as
given in Figure 5 We may consider a system of three linkers placed as if they are in three adjacent
layers of bulk P4 SURMOF-2 The linkers are undergoing a rotation along their C2 axis that would be
aligned to the Cu-Cu axes in the paddlewheels (see system-I in Figure 6) The dissociation energy of
167
the system includes four times the repulsion from one adjacent linker If we neglect the interaction
between the end-linkers which is only -35 meV the system represents the situation in bulk P4 MOF-
2 The gain of intermolecular energy due to twisting the stacked linkers is partially compensated by
the energy penalty arising from rotation of the linker between the paddlewheels and the resulting
energy shows a minimum at 22deg (Figure 6)
Figure 6 Model of the linker orientation in SURMOF-2 The stabilization of a linker inbetween two layers is approximated in System-I the arrows indicating the rotation while the energy penalty due to breaking the p conjugation of Figure 5 is given in System-II The resulting energy is the black line showing a minimum at 22deg The energy in System-I is with respect to its dissociation limit
Each linker may undergo rotation either clockwise or anti-clockwise with equal preferences in the
local environment However there may be a global control over the preference of each linker The
most stable structure can be figured out from the total energies of each possible arrangement Since
there are only two choices for each linker it may orient either in same fashion or alternate fashion
along X and Y directions If we expect a regular pattern the total number of possibilities are only
three as shown in Figure 8 where rotation of linkers (violet lines) are marked with lsquo+rsquo signs in one of
its two sides which means that facing (outer)-edges of linkers are tilted towards these sides The
three orderings may be verbalized as follows
(i) projection of the facing edges of oppositely placed linkers are either within the square or outside
(P4nbm) (ii) projection of the facing edge of only one of the oppositely placed linkers is within the
square (P4mmm) (iii) projection of the facing edges of all four linkers are either within the square
or outside (P4nmm)
The adjacent layers follow the same linker-orientations The unit cells of (i) and (iii) are four times
bigger than (ii) The stacking energies calculated at the DFTB level suggest P4nbm as the most stable
168
geometry The energies are respectively -048 -032 and -029 eV per formula unit for P4nbm
P4mmm and P4nmm respectively Within P4nbm symmetry the linker edges undergo lowest
repulsion compared to others It is to be noted that only P4nbm has four-fold rotational symmetry
along Z-axis about the Cu-dimer in any paddlewheel
Figure 7 Three possible arrangements of tilted linkers in the periodic SURMOF-2 Paddlewheels are shown as red diamonds and the linkers as violet lines Facing (outer) edges of the linkers are rotated towards the side with + signs The symmetries and calculated stacking energies (per formula unit) of the arrangements are also given
To quantify the different stacking energies we performed periodic DFT calculations on the structure
of Cu2(bdc)2 MOF-2 As the most stable P4nbm geometry demands high computational resources in
each calculation we used P4mmm geometry which has four times less atoms in unit cell We
explored tight-packing with rotated linkers and loose-packing with un-rotated linkers Two energy-
minima have been found as shown in Figure 8a at 58 and 65 Aring corresponding to rotated and un-
rotated states of linkers respectively The latter is 40 meV more stable than the former which
means that SURMOF-2 exists in a metastable state along the perpendicular axis The relative shift of
adjacent layers in the direction of the lattice vector (or ) represents the transition from P2 to P4
and back to P2 symmetry We shifted the adjacent layers parallel to and calculated the relative
energies Since the shift of adjacent layers in P4mmm geometry is not symmetric for positive and
negative directions of averages of the energies of the shift in both directions are plotted (see
Figure 9b) P4 is a local minimum while P2 is the global minimum and the energy barrier separating
the two minima accounts to 23 meV per formula unit Hence the P4 geometry of SURMOF-2 can be
taken as metastable state of MOF-2
169
Figure 8 a) Energy as function of interlayer separation d and b) Energy as a function of relative shift of adjacent layers for periodic MOF-2 Energies are given in eV per formula unit
The role of linkers in creating a local minimum at P4 is now apparent However when undergoing the
transition from P4 to P2 the relative motions of the horizontal and vertical linkers are different from
each other Hence a qualitative study is essential to accurately determine the role of each building
block for holding SURMOFs-2 in a metastable state We thus resolved the shift of entire adjacent
layers with respect to each other into relative motions of individual building blocks The experimental
interlayer distance (56 Aring) has been fixed and the energy-profiles have been calculated using DFT
The scans include the shift of
i) a paddlewheel over other
ii) a horizontal linker over other
iii) a vertical linker over other
iv) a paddlewheel over a horizontal linker
v) a paddlewheel over a vertical linker
Here vertical and horizontal linkers are the linkers vertical and horizontal to the shift directions
respectively A complete picture of individual shifts of the MOF-2 SBU including their energy-profiles
is given in Figure 9 The shift of the vertical over the horizontal linker was found insignificant and was
omitted A note of warning is that the tilted vertical linker meets different neighborhoods when
shifted to the left and right However an average energy of these two shifts seems sensible because
the nearest vertical linker in the same layer of P4nbm geometry has opposite orientation This
averaging also makes sense in a case that alternate layers undergo shifting to the same direction
leading to a zigzag (or serrated) packing of layers in the bulk As is readily seen in Figure 9 the
formation of the Cu-O-Cu-O rings between the paddlewheels (see Figure 3) is the key reasons for the
layer shift in P2 MOF-2 A support is obtained from the shifting of the paddlewheel over the
170
horizontal linker (Shift IV) A major objection is made by the vertical linkers (Shift III) The total
interaction becomes repulsive when a vertical linker shifts with respect to the other for about 05 Aring
This may alter the tilt of the linker however a minimum is already established The vertical linkers of
a layer in a way are trapped between those of adjacent layers The shift to either side from P4 most
probably decreases the interlayer separation However this demands further rotation of the vertical
linkers in order to reduce their mutual Pauli repulsion Overall the sharp minimum at P4 may be
taken as an inheritance from the lower interlayer distance of P4 geometry fixed by the anchoring on
the substrate
Figure 9 Shift of individual building blocks in the adjacent layers correspond to the situation in bulk The energies of each shifted arrangement and their total energy are given in the graph
The total energy involved in the shifting of two building blocks (one building block over the other) is
equivalent to the energy per one building block when it feels shift from two neighbors Only the
vertical linker is sensitive to the shift-direction of the two neighbors However since averages were
taken as discussed earlier the total energy becomes independent of the direction Besides the
relative motion of the SBU facing each other in P4 symmetry (Shifts I II and III) for strong distortions
we also have to consider the interaction of adjacent linker-connector interactions as represented in
Shifts IV and V The former stabilizes P2 while the latter support P4 Overall the total energy for all
the shifts shows a local minimum at P4 (see figure 9) in agreement with the periodic calculation
shown in Figure 8 Indeed the relative energy between P2 and P4 stackings estimated by the
171
superposition of individual contributions is 43 meV in close agreement to the 40 meV obtained by
the periodic calculations
Our finite-component model successfully provides adequate information on the individual
contributions of building blocks Vertical linkers play crucial role in holding the SURMOF with P4
symmetry which was never achieved with solvothermally synthesized MOF-2 Paddlewheels are
held most responsible for lateral shift The vertical linkers winningly establish a local minimum at P4
for the SURMOF
Recently we have reported SURMOF-2 derivatives with very large pore sizes(Ref) This has been
achieved by increasing the length of the linker units In view of our analysis of the stacking and
stability of MOF-2 we can now safely state that it will be possible to generate SURMOF-2 derivatives
with even larger pores with pore sizes essentially limited by the availability of stiff long organic
linker molecules This is exemplified by a calculation of the stacking energy of a series of phenyl
oligomers that is SURMOF-2 derivatives incorporating chains with 1 to 10 benzene rings in the
linkers The stacking energies per formula unit are -041 -067 -091 -121 -145 -168 -192 -215
-237 and -26 eV respectively Each benzene ring adds more than 02 eV into the stacking energy per
formula unit This energy is due to the London dispersion interaction between the linkers in the
neighboring layers
The drop of SURMOFs into the local minimum perhaps is blended with some parameters related to
synthetic environments This was beyond the scope of this work however we suggest that studies of
the mechanism of substrate-assisted layer-by-layer deposition of metal and ligand precursors may
give some primary insights into it
CONCLUSION
We have analyzed the reason for the different stackings observed for MOF-2 In the traditional
solvothermal synthesis MOF-2 is likely to crystallize in P2 symmetry determined by the strong
interaction between the paddlewheel units The coordination of large solvent molecules to the free
metal centers may lead to MOF-2 in C2 symmetry In contrast the formation of SURMOFs using
Liquid Phase Epitaxy (LPE) allows the formation of high-symmetry P4 MOF-2 This structure requires
a structural modification in terms of the orientation of the linkers with respect to the free monolayer
and the stacking is stabilized by London dispersion interactions between the linkers Increasing the
linker length is a straightforward way for the linear expansion of pore size and according to our
computations the pore size is only limited by the availability of linker molecules showing the desired
length Thus we presented a rare situation in which the linkers guarantee the persistence of a series
of materials in an otherwise unachievable state
172
COMPUTATIONAL DETAILS
The Becke three-parameter hybrid method combined with Lee-Yang-Par correlational functional
(B3LYP)34-37 and augmented with a dispersion term38 as implemented in Turbomole39 has been used
for DFT calculations The copper atoms were described using the basis set associated with the Hay-
Wadt40 relativistic effective core potentials41 which include polarization f functions42 This basis set
was obtained from the EMSL Basis Set Library4344 Carbon oxygen and hydrogen atoms were
described using the Pople basis set 6-311G Geometry optimizations of the MOF fragments were
performed using internal redundant co-ordinates45 A triplet spin state was assigned for each Cu-
paddlewheel46
Periodic structure calculations at the BLYP47 level were performed using the ADF BAND 2012
code4849 Dispersion correction was implemented based on the work by Grimme50 Triple-zeta basis
set was used The crystalline state of MOFs was computationally described using periodic boundary
conditions Bader charge analysis of paddlewheel-pairs was also performed using ADF 2012 code
The analysis was made at the level of B3LYP-D using an all-electron triple-zeta basis set
The non-orthogonal tight-binding approximation to DFT called Density Functional Tight-Binding
(DFTB)51 is computationally feasible for unit cells of several hundreds of atoms Hence this method
was used for extensive calculations on periodic structures This method computes a transferable set
of parameters from DFT calculations of a few molecules per pair of atom types The more accurate
self-consistent charge (SCC) extension of DFTB52-54 has been used for the study of bulk MOFs Validity
of the Slater-Koster parameters of Cu-X (X = Cu C O H) pairs were reported before55 The
computational code deMonNano56 which has dispersion correction implemented57 was used
If not stated otherwise the energies of the periodic structure are given per a formula unit (FU) of the
MOF which includes one paddlewheel and two linkers (one vertical linker and one horizontal linker)
REFERENCES
(1) Eddaoudi M Moler D B Li H L Chen B L Reineke T M OKeeffe M Yaghi O M Accounts of
Chemical Research 2001 34 319
(2) Li H Eddaoudi M OKeeffe M Yaghi O M Nature 1999 402 276
(3) Rowsell J L C Yaghi O M Microporous and Mesoporous Materials 2004 73 3
(4) Eddaoudi M Li H L Yaghi O M Journal of the American Chemical Society 2000 122 1391
(5) Rowsell J L C Yaghi O M Angewandte Chemie-International Edition 2005 44 4670
173
(6) Suh M P Park H J Prasad T K Lim D-W Chemical Reviews 2012 112 782
(7) Murray L J Dinca M Long J R Chemical Society Reviews 2009 38 1294
(8) Rosi N L Eckert J Eddaoudi M Vodak D T Kim J OKeeffe M Yaghi O M Science 2003 300
1127
(9) Kreno L E Leong K Farha O K Allendorf M Van Duyne R P Hupp J T Chemical Reviews 2012
112 1105
(10) Achmann S Hagen G Kita J Malkowsky I M Kiener C Moos R Sensors 2009 9 1574
(11) Lee J Farha O K Roberts J Scheidt K A Nguyen S T Hupp J T Chemical Society Reviews 2009
38 1450
(12) Farrusseng D Aguado S Pinel C Angewandte Chemie-International Edition 2009 48 7502
(13) Corma A Garcia H Llabres i Xamena F X Chemical Reviews 2010 110 4606
(14) Rowsell J L C Millward A R Park K S Yaghi O M Journal of the American Chemical Society 2004
126 5666
(15) Deng H Doonan C J Furukawa H Ferreira R B Towne J Knobler C B Wang B Yaghi O M
Science 2010 327 846
(16) Liu J Lukose B Shekhah O Arslan H K Weidler P Gliemann H Braumlse S Grosjean S Godt A
Feng X Muumlllen K Magdau I-B Heine T Woumlll C submitted to Nature Chemistry 2012
(17) Li H Eddaoudi M Groy T L Yaghi O M Journal of the American Chemical Society 1998 120 8571
(18) Carson C G Hardcastle K Schwartz J Liu X Hoffmann C Gerhardt R A Tannenbaum R
European Journal of Inorganic Chemistry 2009 2338
(19) Clausen H F Poulsen R D Bond A D Chevallier M A S Iversen B B Journal of Solid State
Chemistry 2005 178 3342
(20) Edgar M Mitchell R Slawin A M Z Lightfoot P Wright P A Chemistry-a European Journal 2001
7 5168
(21) Schaate A Roy P Preusse T Lohmeier S J Godt A Behrens P Chemistry-a European Journal
2011 17 9320
(22) Cote A P Benin A I Ockwig N W OKeeffe M Matzger A J Yaghi O M Science 2005 310
1166
(23) Wan S Guo J Kim J Ihee H Jiang D Angewandte Chemie-International Edition 2008 47 8826
174
(24) Lukose B Kuc A Frenzel J Heine T Beilstein Journal of Nanotechnology 2010 1 60
(25) Kitagawa S Kitaura R Noro S Angewandte Chemie-International Edition 2004 43 2334
(26) Shekhah O Wang H Zacher D Fischer R A Woell C Angewandte Chemie-International Edition
2009 48 5038
(27) Shekhah O Wang H Kowarik S Schreiber F Paulus M Tolan M Sternemann C Evers F
Zacher D Fischer R A Woll C Journal of the American Chemical Society 2007 129 15118
(28) Zacher D Schmid R Woell C Fischer R A Angewandte Chemie-International Edition 2011 50 176
(29) Bader R F W Accounts of Chemical Research 1985 18 9
(30) Merino G Vela A Heine T Chemical Reviews 2005 105 3812
(31) Tafipolsky M Schmid R Journal of Chemical Theory and Computation 2009 5 2822
(32) Kuc A Enyashin A Seifert G Journal of Physical Chemistry B 2007 111 8179
(33) Winston E B Lowell P J Vacek J Chocholousova J Michl J Price J C Physical Chemistry
Chemical Physics 2008 10 5188
(34) Becke A D Journal of Chemical Physics 1993 98 5648
(35) Lee C T Yang W T Parr R G Physical Review B 1988 37 785
(36) Vosko S H Wilk L Nusair M Canadian Journal of Physics 1980 58 1200
(37) Stephens P J Devlin F J Chabalowski C F Frisch M J Journal of Physical Chemistry 1994 98
11623
(38) Civalleri B Zicovich-Wilson C M Valenzano L Ugliengo P Crystengcomm 2008 10 405
(39) University of Karlsruhe and Forschungszentrum Karlsruhe gGmbH - TURBOMOLE V63 2011 2007
(40) Wadt W R Hay P J Journal of Chemical Physics 1985 82 284
(41) Roy L E Hay P J Martin R L Journal of Chemical Theory and Computation 2008 4 1029
(42) Ehlers A W Bohme M Dapprich S Gobbi A Hollwarth A Jonas V Kohler K F Stegmann R
Veldkamp A Frenking G Chemical Physics Letters 1993 208 111
(43) Feller D Journal of Computational Chemistry 1996 17 1571
(44) Schuchardt K L Didier B T Elsethagen T Sun L Gurumoorthi V Chase J Li J Windus T L
Journal of Chemical Information and Modeling 2007 47 1045
175
(45) von Arnim M Ahlrichs R Journal of Chemical Physics 1999 111 9183
(46) St Petkov P Vayssilov G N Liu J Shekhah O Wang Y Woell C Heine T Chemphyschem 2012
13 2025
(47) Gill P M W Johnson B G Pople J A Frisch M J Chemical Physics Letters 1992 197 499
(48) SCM Amsterdam Density Functional 2012
(49) Velde G T Bickelhaupt F M Baerends E J Guerra C F Van Gisbergen S J A Snijders J G
Ziegler T Journal of Computational Chemistry 2001 22 931
(50) Grimme S Journal of Computational Chemistry 2006 27 1787
(51) Seifert G Porezag D Frauenheim T International Journal of Quantum Chemistry 1996 58 185
(52) Elstner M Porezag D Jungnickel G Elsner J Haugk M Frauenheim T Suhai S Seifert G
Physical Review B 1998 58 7260
(53) Frauenheim T Seifert G Elstner M Hajnal Z Jungnickel G Porezag D Suhai S Scholz R
Physica Status Solidi B-Basic Research 2000 217 41
(54) Oliveira A F Seifert G Heine T Duarte H A Journal of the Brazilian Chemical Society 2009 20
1193
(55) Lukose B Supronowicz B Petkov P S Frenzel J Kuc A Seifert G Vayssilov G N Heine T
physica status solidi (b) 2011
(56) Heine T Rapacioli M Patchkovskii S Frenzel J Koester A M Calaminici P Escalante S Duarte
H A Flores R Geudtner G Goursot A Reveles J U Vela A Salahub D R deMon-nano edn ed deMon
2009
(57) Zhechkov L Heine T Patchkovskii S Seifert G Duarte H A Journal of Chemical Theory and
Computation 2005 1 841
iv
Abstract
Framework materials are extended structures that are built into destined nanoscale architectures
using molecular building units Reticular synthesis methods allow stitching of a large variety of
molecules into predicted networks Porosity is an obvious outcome of the stitching process These
materials are classified and named according to the chemical composition of the building blocks For
instance Metal-Organic Frameworks (MOFs) consists of metal-oxide centers that are linked together
by organic entities The stitching process is straight-forward so that there are already thousands of
them synthesized Controlled growth of MOFs on substrates leads to what is known as surface-MOFs
(SURMOFs) A low-weight metal-free version of MOFs is known as Covalent Organic Frameworks
(COFs) They consist of light elements such as boron oxygen silicon nitrogen carbon and hydrogen
atoms Diamond-like structures with a variety of linear organic linkers between tetrahedral nodes is
called Porous Aromatic Frameworks (PAFs)
The thesis is composed of computational studies of the above mentioned classes of materials The
significance of such studies lies in the insights that it gives about the structure-property relationships
Density Functional Theory (DFT) and its Tight-Binding approximate method (DFTB) were used in
order to perform extensive calculations on finite and periodic structures of several frameworks DFTB
provides an ab-initio base on periodic structure calculations of very large crystals which are typically
studied only using force-field methods The accuracy of this approximate method is validated prior to
reasoning
As the materials are energized from building units and coordination (or binding) stability vs
structure is discussed Energy of formation and mechanical strength are particularly calculated Using
dispersion corrected (DC)-Self Consistent Charge (SCC) DFTB we asserted that 2D COFs should have a
layer arrangement different from experimental suggestions Our arguments supported by simulated
PXRDs were later verified using higher level theories in the literature Another benchmark is giving an
insightful view on the recently reported difference in symmetries of two-dimensional MOFs and
SURMOFs The result of it shows an extra-ordinary role of linkers in SURMOFs providing
metastability
Electronic properties of all the materials and hydrogen adsorption capacities of PAFs are discussed
COFs PAFs and many of the MOFs are semiconductors HOMO-LUMO gaps of molecular units have
crucial influence in the band gaps of the solids Density of states (DOS) of solids corresponds to that
of cluster models Designed PAFs give a large choice of adsorption capacities one of them exceeds
the DOE (US) 2005 target Generally the studies covered under the scheme of the thesis illustrate
the structure stability and properties of framework materials
- Dedicated to my Family and Rajan sir
Table of Contents 1 Outline 1
2 Introduction 2
21 Nanoporous Materials 2
22 Reticular Chemistry 3
23 Metal-Organic Frameworks 5
24 Covalently-bound Organic Frameworks 8
3 Methodology and Validation 10
31 Methods and Codes 10
32 DFTB Validation 11
4 2D Covalent Organic Frameworks 13
41 Stacking 13
42 Concept of Reticular Chemistry 15
5 3D Frameworks 17
51 3D Covalent Organic Frameworks 17
52 Porous Aromatic Frameworks 18
6 New Building Concepts 20
61 Isoreticular Series of SURMOFs 20
62 Metastability of SURMOFs 21
7 Summary 23
71 Validation of Methods 23
72 Weak Interactions in 2D Materials 25
73 Structure-Property Relationships 27
List of Abbreviations 31
List of Figures 32
References 33
Appendix A Review of covalently-bound organic frameworks 37
Appendix B Properties of MOFs within DFTB 81
Appendix C Stacking of 2D COFs 96
Appendix D Reticular concepts applied to 2D COFs 105
Appendix E Properties of 3D COFs 122
Appendix F Properties of PAFs 131
Appendix G Isoreticular SURMOFs of varying pore sizes 145
Appendix H Metastability in 2D SURMOFs 160
1
1 Outline
I prepared this cumulative thesis as it is permitted by Jacobs University Bremen as all work has been
published in international peer-reviewed journals is submitted for publication or in a late
manuscript state in order to be submitted soon The list of articles contains three published papers
three submitted manuscripts and two manuscripts that are to be submitted The articles are given in
Appendices A-H in the order of their discussions Each appendix has one paper and its supporting
information
The chapters from ldquoIntroductionrdquo to ldquoNew Building Conceptsrdquo contain general discussions about the
articles and provide a red thread leading through the articles The discussions mainly circle around
the context and the content of the articles
The chapter ldquoIntroductionrdquo provides a general introduction to the topic and a review of the materials
discussed in this thesis As no comprehensive review on ldquoCovalently-bound Organic Frameworksrdquo is
available in the literature we prepared a manuscript (Appendix A) covering that subject The chapter
ldquoMethodology and Validationrdquo discusses one article on validation of DFTB method in Metal-Organic
Frameworks (Appendix B) Each consecutive chapter discusses two articles The chapter ldquo2D
Covalent Organic Frameworksrdquo covers the studies on layer arrangements (Appendix C) followed by
analysis on how reticular chemistry concepts can be used to design new materials (Appendix D) The
chapter ldquo3D Frameworksrdquo discusses the stability and properties of 3D COFs (Appendix E) and PAFs
(Appendix F) It also discusses hydrogen storage capacities of PAFs The chapter ldquoNew Building
Conceptsrdquo has coverage on 2D MOFs that are built on organic surfaces The first article in this chapter
describes the achievement of our collaborators in synthesizing a series of SURMOFs of varying pore
sizes supported by our calculations indicating their matastability Extensive calculations revealing the
role of linkers in creating the unprecedented difference of SURMOFs in comparison with the bulk
MOFs is described in another article
Details of the articles and references to the appendices are given in the respective places in each
chapter The chapter ldquoSummaryrdquo gives an overview of all the results and discussions It also discusses
some impacts of the publications and concludes the thesis Overall the studies bring into picture
different classes of materials and analyze their structural stabilities and properties
2
2 Introduction
21 Nanoporous Materials
The field of nanomaterials covers materials that have properties stemming from their nanoscale
dimensions (1-100 nm) In contrast to nanomaterials which define size scaling of bulk materials the
major determinant of nanoporous materials is their pores Nanoporous materials are defined as
porous materials with pore diameters less than 100 nm and are classified as micropores of less than
2 nm in diameter mesopores between 2 and 50 nm and macropores of greater than 50 nm They
have perfectly ordered voids to accommodate interact with and discriminate molecules leading to
prominent applications such as gas storage separation and sieving catalysis filtration and
sensoring1-4 They differ from the generally defined nanomaterials in the sense that their properties
are mostly determined by pore specifications rather than by bulk and surface scales Hence the
focus is onto the porous properties of the materials
Utilization of the pores for certain applications relies on certain parameters such as pore size pore
volume internal surface area and wall composition For example physical adsorption of gases is high
in a material with large surface area which implies significantly high storage in comparison to a tank
Porosity can be measured using some inert or simple gas adsorption measurements Distribution of
pore size can be sketched from the adsorptiondesorption isotherm
Nanoporous materials abound in nature Zeolites for instance many of them are available as mineals
have been used in petroleum industry as catalysts for decades The walls of human cells are
nanoporous membranes Other examples are clays aluminosilicate minerals and microporous
charcoals Zeolites are microporous materials from the aluminosilicate family and are often known as
molecular sieves due to their ability to selectively sort molecules primarily based on a size exclusion
principle A material with high carbon content (coal wood coconut shells etc) can be converted to
activated carbon (AC) by controlled thermal decomposition in a furnace The resultant product has
large surface area (~ 500 m2 g-1) Porous glass is a material produced from borosilicate glasses having
pore sizes usually in nm to μm range With increasing environmental concerns worldwide porous
materials have become a suitable choice for separation of polluting gases storage and transport of
energy carriers etc Synthesis of open structures has advanced this area5-7 especially since the
invention of MCM-41 (Mobil Composition of Matter No 41) by Mobil scientists89 Furthermore
there are many templating pathways in making nanoporous materials10-13 Currently it is possible to
engineer the internal geometry at molecular scales
3
For more than a decade chemists are able to synthesize extended structures from well-defined and
rigid molecular building units Such designed and controlled extensions provide porosity which can
be scaled and modified by selecting appropriate building blocks The first realization of this kind was
a class of materials known as Metal-Organic Frameworks (MOFs) in which metal-oxides are stitched
together by organic molecules Synthesis of molecules into predicted frameworks have led to the
emergence of a discipline called reticular chemistry14 The new design-based synthetic approaches
have produced large number of nanoporous materials in comparison to the discovery-based
synthetic chemistry
22 Reticular Chemistry
The discipline of designed (rational) synthesis of materials is known as reticular chemistry Desired
materials can be realized by starting with well-defined and rigid molecular building blocks that will
maintain their structural integrity throughout the construction process The extended structures
adopt high symmetry topologies The synthetic approach follows well-defined conditions which
provide general control over the character of solids In short it is the chemistry of linking molecular
building blocks by strong bonds into predetermined structures
The knowledge about how atoms organize themselves during synthesis is essential for the design
The principal possibilities rely on the starting compounds and the reaction mechanisms Porosity is
almost an inevitable outcome of reticular synthesis Solvents may be used in most cases as space-
filling agents and in cases of more than one possibility as structure-directing agents
Thousands of materials in large varieties have been synthesized using the reticular chemistry
principles Reticular Chemistry Structure Resource (RCSR) is a searchable database for nets a project
initiated by of Omar M Yaghi and Michael OrsquoKeeffe15 They define nets as the collection of vertices
and edges that form an irreducible network in which any two vertices are connected through at least
one continuous path of edges Building units are finite nets as in the case of a polyhedron Periodic
structures have one two or three-periodic nets An example case of 3D diamond net (dia) is shown in
Figure 1 Graph-theoretical aspects together with explicative terminology and definitions can be
found in the literature16-18
Figure 1 a) ldquoAdamantinerdquo unit of diamond structure and b) diamond (dia) net
4
In other words a framework can be deconstructed into one or more fundamental building blocks
each of them assigned by a vertex in the net The vertices are the branching points and edges are
joining them The realization of the net in space by representing the vertices and lattice parameters
by co-ordinates and a matrix respectively is called embedding of the net Underlying topology of an
extended structure is the structure of the net inherited from the crystal structure that is invariant
under different embeddings For example Cu-BTC MOF has paddlewheels and benzene rings as
fundamental blocks The MOF structure can be simplified into its underlying topology as shown in
Figure 2
Figure 2 CU-BTC MOF and the corresponding tbo net
Alternatively the topology of a framework can be defined using the convention of so-called
secondary building units (SBUs) The shapes of SBUs are defined by points of extension of the
fundamental building blocks SBUs are invariant for building units of identical connectivity Based on
the number of connections with the adjacent blocks (4 for paddlewheel and 3 for the benzene) SBUs
of Cu-BTC can be represented as shown in Figure 3a Assembly of SBUs following the network
topology and insertion of the fundamental building blocks give the crystal structure (see Figure 3b for
the extension of SBUs to the topology of Cu-BTC)
In addition to MOFs Metal-Organic Polyhedra (MOP)19 Zeolite Imidazolate Frameworks (ZIFs)20 and
Covalent-Organic Frameworks (COFs)2122 are developing classes of materials within reticular
chemistry the first members of all of them were synthesized by the group of Yaghi MOPs are nano-
sized polyhedra achieved by linking transition metal ions and either nitrogen or carboxylate donor
organic units ZIFs are aluminosilicate zeolites with replacements of tetrahedral Si (Al) and bridging
oxygen by transition metal ion and imidazolate link respectively COFs are extended organic
5
structures constructed solely from light elements (H B C and O) The materials synthesized under
the reticular scheme are largely crystalline
Figure 3 a) Fundamental building blocks and SBUs of Cu-BTC and b) extension of SBUs following
crystal structure
23 Metal-Organic Frameworks
MOFs are porous crystalline materials consisting of metal complexes (connectors) linked together by
rigid organic molecules (linkers) MOFs are analogues to a class of materials called coordination
polymers (CPs) However there are primary differences between them CPs are inorganic or
organometallic polymer structures containing metal ions linked by organic ligands A ligand is an
atom or a group of atoms that donate one or more lone pairs of electrons to the metal cations and
thereby participate in the formation of a coordination complex In MOFs typically metal-oxide
centers are used instead of single metal ions as they provide strong bonds with organic linkers This
provides not only high stability but also high directionality because multiple bonds are involved
6
between metal-centers and organic linkers Predictability lies in the pre-knowledge about the
connector-linker interactions Thus the reticular design of MOFs derives from the precise
coordination geometry between metal centers and the linkers Figure 4a shows a schematic diagram
of MOFs By varying the chemical composition of nodes and linkers thousands of different MOF
structures with a large variety in pore size and structure have been synthesized Figure 4b shows
MOF-5 that consists of zinc-oxide tetrahedrons and benzene linkers
Figure 4 a) Schematic diagram of the single pore of MOF b) An example of a simple cubic MOF in
which zinc-oxide tetrahedron are linked together by benzene linkers Atom colors blue ndash Zn red ndash
O grey ndash C white ndash H
The synthesis of MOFs differs from organic polymerizations in the degree of reversible bond
formation Reversibility allows detachment of incoherently matched monomers followed by their
attachment to form defect-free crystals Assembly of monomers occurs as single step hence
synthetic conditions are of high importance The strength of the metal-organic bond is an obstacle
for reversible bond formation however solvothermal techniques are found out to be a convenient
solution23 Solvothermal synthesis generally allows control over size and shape distribution Using
post-synthetic methods further changes on cavity sizes and chemical affinities can be made
Materials that are stable with open pores after removal of guest molecules are termed as open-
frameworks The stability of guest-free materials is usually examined by Powder X-ray diffraction
(PXRD) analysis Thermogravimetric analysis may be performed to inspect the thermal stability of the
material Elemental analysis can detail the elemental composition of the material Physical
techniques such as Fourier transform infrared (FTIR) spectra and nuclear magnetic resonance (NMR)
may be used to verify the condensation of monomers to the desired topology Porosity can be
evidenced from adsorption isotherms of gases or mercury porosimetry
7
The number of linkers that are allowed to bind to the metal-center and the orientations of the linkers
depend exclusively on the coordination preferences of the metal The organic linkers are typically
ditopic or polytopic They are essential in determining the topology and providing porosity Longer
linkers provide larger pore size A series of compounds with the same underlying topology and
different linker sizes are called iso-reticular series The structure of MOFs in principle can be formed
into the requirement of prominent applications such as gas storage gas separation sensing and
catalysis The structural divergence and performance can be further increased by functionalizing the
organic linkers Hence several attempts are on-going in purpose to come up with the best material
possible in each application
Interpenetration of frameworks is an inevitable outcome of large pore volume Interpenetrated nets
are periodic equivalents of mechanically interlocked molecules rotaxanes and catenanes Depending
on topology they are either maximally separated termed as interpenetration or minimally separated
termed as interweaving (see Figure 5) Interpenetration can be beneficial to some porous structures
protecting from collapse upon removal of solvents
Figure 5 Schematic diagram of interpenetrated and interwoven frameworks
Controlled growth of MOFs on organic surfaces is achieved by R A Fischer and C Woumlll24 The then
named SURMOFs allow to fabricate structures possibly not achievable by bulk synthesis The growth
is achieved through liquid phase epitaxy (LPE) of MOF reactants on the surface (nucleation site) A
step-by-step approach which is alternate immersion of the substrate in metal and ligand precursors
supplies control of the growth mechanism
8
Figure 6 Schematic diagram of SURMOF
24 Covalently-bound Organic Frameworks
As a result of the long-term struggle for complete covalent crystallization in the year 2005 Cocircte et
al21 synthesized porous crystalline extended 2D Covalent-Organic Frameworks (COFs) using
reticular concepts The success was followed by the design and synthesis of 3D COFs in the year
200722 By now there are about 50 COFs reported in the literature COFs are made entirely from
light elements and the building blocks are held together by strong covalent bonds Most of them
were formed by solvothermal condensation of boronic acids with hydroxyphenyl compounds
Tetragonal building blocks were used for the formation of 3D COFs Alternate synthetic methods
were also used for producing COFs COFs are generally studied for gas storage applications However
they have also shown potentialities in photonic and catalytic applications
Alternative synthesis methods paved the way to new covalently bound organic frameworks
Trimerization of polytriazine monomers in ionothermal conditions gives rise to Covalent Triazine
Frameworks (CTFs)25 Similarly coupling reactions of halogenated monomers produced Porous
Aromatic Frameworks (PAFs) Albeit the short-range order one of the PAFs showed very high surface
area (5600 m2 g-1) and gas uptake capacity26
Due to low weight the covalently-bound materials show very high gravimetric capacities
Suggestions such as metal-doping functionalization and geometry modifications can be found in the
literature for the general improvement of the functionalities There are also various studies of their
structure and properties
A review on the synthesis structure and applications of covalently bound organic frameworks has
been prepared for publication
Article 1 Covalently-bound organic frameworks
Binit Lukose Thomas Heine
9
See Appendix A for the article
My contributions include collecting data and preparing a preliminary manuscript
Figure 7 SBUs and topologies of 2D COFs
10
3 Methodology and Validation
31 Methods and Codes
The Density-Functional based Tight Binding (DFTB) method2728 is an approximate Kohn-Sham DFT29-31
scheme It avoids any empirical parameterization by calculating the Hamiltonian and overlap matrix
elements from DFT-derived local orbitals (atomic orbitals) and atomic potentials The Kohn-Sham
orbitals are represented with the linear combination of atomic orbitals (LCAO) scheme The matrix
elements of Hamiltonian are restricted to include only one- and two-center contributions Therefore
they can be calculated and tabulated in advance as functions of the distance between atomic pairs
The remaining contributions to the total energy that is the nucleus-nucleus repulsion and the
electronic double counting terms are grouped in the so-called repulsive potential This two-center
potential is fitted to results of DFT calculations of properly chosen reference systems The accuracy
and transferability can be improved with the self-consistent charge (SCC) extension to DFTB32 This
method is based on the second-order expansion of the Kohn-Sham total energy with respect to
charge density fluctuations which are estimated by Mulliken charge analysis In order to account for
London dispersion empirical dispersion energy can be added to the total energy3334 Various reviews
are available on DFTB notably the recent ones by Oliveira et al35 and by Seifert et al36
DFTB is implemented in a large number of computer codes For this work we employed the codes
deMonNano37 and DFTB+38 as they allow DFTB calculations on periodic and finite structures
Typically dispersion corrected (DC) SCC-DFTB calculations were carried out Periodic boundary
conditions were used to represent the crystalline frameworks and as the unit cells are large the
standard approach used the point approximation Electronic density of states (DOS) have been
calculated using the DFTB+ code using k-point sampling where the k space was determined by
reaching convergence for the total energy according to the scheme proposed by Monkhorst and
Pack39
For DFT calculations of finite and periodic structures ADF40 Turbomole41 and Crystal0942 were used
For studies of finite models of COFs the calculations were performed at PBEDZP level However for
extensive calculations on SURMOF-2 models we used B3LYP-D The copper atoms were described
using the basis set associated with the Hay-Wadt43 relativistic effective core potentials44 which
include polarization f functions45 Carbon oxygen and hydrogen atoms were described using the
Pople basis set 6-311G
Details of the individual calculations are given in the individual articles in the appendix of this thesis
11
32 DFTB Validation
Figure 8 Deconstructed building units their schematic diagrams and final geometries of HKUST-1
(Cu-BTC) MOF-5 MOF-177 MOF-205 and MIL-53
12
In the literature MOFs and COFs are largely studied for applications such as gas storage using
classical force field methods46-48 First principles based studies of several hundreds of atoms are
computationally expensive Hence they are generally limited to cluster models of the periodic
structures Contrarily DFTB paves the way to model periodic structures involving large numbers of
atoms Being an approximate method DFTB holds less accuracy hence comparisons to experimental
data or higher level methods should be performed for validation
As MOFs contain metal centers andor metal oxides as well as organic elements the DFTB
parameters for both heavy and light elements as well as their mixtures are required Thus we have
chosen MOFs such as Cu-BTC (HKUST-1) MOF-5 MOF-177 MOF-205 and MIL-53 as our model
structures which contain Cu Zn and Al atoms apart from C O and H The MOFs comprising three
common connector units copper paddlewheels (HKUST-1) zinc oxide Zn4O tetrahedron (MOF-5
MOF-177 DUT-6 (MOF-205)) and aluminum oxide AlO4(OH)2 octahedron (MIL-53) Figure 8 shows
the schematic diagram of the MOFs
The validation calculations have been published
Article 2 Structural properties of metal-organic frameworks within the density-functional based
tight-binding method
Binit Lukose Barbara Supronowicz Petko St Petkov Johannes Frenzel Agnieszka B Kuc Gotthard
Seifert Georgi N Vayssilov and Thomas Heine Phys Status Solidi B 2012 249 335ndash342 DOI
101002pssb201100634
See Appendix B for the article
In this article DFTB has been validated against full hybrid density-functional calculations for model
clusters against gradient corrected density-functional calculations for supercells and against
experiment Moreover the modular concept of MOF chemistry has been discussed on the basis of
their electronic properties Our results show that DC-SCC-DFTB predicts structural parameters with a
good accuracy (with less than 5 deviation even for adsorbed CO and H2O on HKUST-1) while
adsorption energies differ by 12 kJ mol-1 or less for CO and water compared to DFT benchmark
calculations
My contributions to this article include performing DFTB based calculations on the MOFs (HKUST-1
MOF-5 MOF-177 and MOF-205) that is optimizing the geometries simulating powder X-ray
diffraction patterns and calculating density of states and bulk modulus Additional involvement is in
comparing structural parameters such as bond lengths bond angles dihedral angles and bulk
modulus with experimental data or data derived from DFT calculations and preparing the manuscript
13
4 2D Covalent Organic Frameworks
41 Stacking
Condensation of planar boronic acids and hydroxyphenyl compounds leads to the formation of two-
dimensional covalent organic frameworks (2D COFs) The layers are held together by London
dispersion interactions The first synthesized COFs COF-1 and COF-5 were reported to have AB
(P63mmc staggered as in graphite) and AA (P6mmm eclipsed as in boron nitride) stackings
respectively (see Figure 9)21 The synthesis of several further 2D COFs then followed49-51 all of them
were inspected for only two stacking possibilities ndash P6mmm or P63mmc and it was reported that
they aggregate in P6mmm symmetry As framework materials possess framework charges the
interlayer interactions significantly include Coulomb interactions P6mmm symmetry is the face-to-
face arrangement where the overlap of the stacked structures is maximized (maximization of the
London dispersion energy) however atom types of alike charges are facing each other in the closest
possible way With the interlayer separation similar to that in graphite or boron nitride the Coulomb
repulsion should be high in such arrangements One should notice that in the example case of boron
nitride the facing atom types are different We therefore assumed that a stable stacking should
possess layer-offset
Figure 9 AA and AB layer stacks of hexagonal layers
We considered two symmetric directions for layer shift and studied their total energies (see Figure
10) The stable geometries are found to have a layer-shift of ~14 Aring a value corresponding to the
shift of AA to AB stacking in graphite and compatible with the bond length between two 2nd row
atoms This stability-supported stacking arrangement as revealed from our calculations was
14
supported by good agreement between simulated and experimental PXRD patterns Hence
independent of the elementary building blocks any 2D COF should expose a layer-offset Based on
the direction of the shift we proposed two stacking kinds serrated and inclined (Figure 10) In the
former the layer-offset is back and forth while in the latter the layer-offset followed single direction
As serrated and inclined stackings have no significant change in stacking energy our calculations
cannot predict the long-range stacking in the crystal However this problem is known from other
layered compounds and the ultimate answer is yet to be revealed by analyzing high-quality
crystalline phases at low temperature
Figure 10 Stacking energy Es vs shift of adjacent layers for COF-5 Different stacking possibilities
and their energies are also shown
We published our analysis of the stacking in 2D COFs
Article 3 The Structure of Layered Covalent-Organic Frameworks
Binit Lukose Agnieszka Kuc Thomas Heine Chem Eur J 2011 17 2388 ndash 2392 DOI
101002chem201001290
See Appendix C for the article
15
My contributions to this article include performing the shift calculations simulating XRDs and partly
preparing the manuscript The article has made certain impact in the literature Most of the 2D COFs
synthesized afterwards were inspected for their stacking stability The instability of AA stacking was
also suggested by the studies of elastic properties of COF-1 and COF-5 by Zhou et al52 The shear
modulus shows negative signs for the vertical alignment of COF layers while they are small but
positive for the offset of layers Detailed analysis performed by Dichtel53 et al using DFT was
confirming the instability of AA stacking and suggesting shifts of about 13 to 25 Aring
42 Concept of Reticular Chemistry
Reticular chemistry means that (functional) molecules can be stitched together to form regular
networks The structural integrity of these molecules we also speak of building blocks remains in the
crystal lattices Consequently also the electronic structure and hence the functionality of these
molecules should remain similar
2D COFs are formed by condensation of well-defined planar building blocks With the usage of linear
and triangular building blocks hexagonal networks are expected The properties of each COF may
differ due to its unique constituents However the extent of the relationship of the properties of
building blocks in and outside the framework has not been studied in the literature
Reticular chemistry allows the design of framework materials with pre-knowledge of starting
compounds and reaction mechanisms Reverse of this is finding reactants for a certain topology We
intended to propose some building units suitable to form layered structures (see Figure 11) The
building units obey the regulations of reticular chemistry and offer a variety of structural and
electronic parameters
Our strategic studies on a set of designed COFs have been published
Article 4 On the reticular construction concept of covalent organic frameworks
Binit Lukose Agnieszka Kuc Johannes Frenzel Thomas Heine Beilstein J Nanotechnol 2010 1
60ndash70 DOI103762bjnano18
See Appendix D for the article
16
Figure 11 Schematic diagram of different building units forming 2D COFs
Various hexagonal 2D COFs with different building blocks have been designed and investigated
Stability calculations indicated that all materials have the layer offset as reported in our earlier
work54 (see Appendix C) We show that the concept of reticular chemistry works as the Density-Of-
States (DOS) of the framework materials vary with the the DOS of the molecules involved in the
frameworks However the stacking does have some influence on the band gap
My contributions to this article include performing all the calculations and preparing the manuscript
17
5 3D Frameworks
51 3D Covalent Organic Frameworks
First 3D COFs were reported in 2007 by the group of Yaghi22 Although the number of 3D COFs
synthesized so far has not been crossed half a dozen they are of particular interest for their very low
mass density Similar to 2D COFs condensation of boronic acids and hydroxyphenyl compounds led
to the formation of COF-102 103 105 108 and 202 Usage of tetragonal boronic acids leads to the
formation of 3D networks (Figure 12) All of them possessed ctn topology while COF-108 which has
the same material composition as COF-105 crystallized in bor topology COF-300 which was formed
from tetragonal and linear building units possessed diamond topology and was five-fold
interpenetrated55 COF-108 has the lowest mass density of all materials known today Unlike most of
the MOFs the connectors in COFs are not metal oxide clusters but covalently bound molecular
moieties In the synthesized COFs the centers of the tetragonal blocks were either sp3 carbon or
silicon atoms
Schmid et al56 have analyzed the two different topologies and developed force field parameters for
COFs The mechanical stability of COFs was also reported However no further study that details the
inherent energetic stability and properties of COFs was found in the literature Using DFTB we
performed a collective study of all 3D COFs in their known topologies with C and Si centers
Figure 12 Schematic diagram of tetragonal and triangular building blocks forming lsquoctnrsquo or lsquoborrsquo
topologies
Our studies of3D COFs have been prepared for publication
Article 5 Stability and electronic properties of 3D covalent organic frameworks
Binit Lukose Agnieszka Kuc Thomas Heine
18
See Appendix E for the article
My contributions to this article include performing all the calculations and preparing the manuscript
We discussed the energetic and mechanical stability as well as the electronic properties of COFs in
the article All studied 3D COFs are energetically stable materials with formation energies of 76 ndash
403 kJ mol-1 The bulk modulus ranges from 29 to 206 GPa Electronically all COFs are
semiconductors with band gaps vary with the number of sp2 carbon atoms present in the linkers
similar to 3D MOFs
52 Porous Aromatic Frameworks
Porous Aromatic Frameworks (PAFs) are purely organic structures where linkers are connected by sp3
carbon atoms (see Appendix A) Tetragonal units were coupled together to form diamond-like
networks The synthesis follows typically a Yamamoto-type Ullmann cross-coupling reaction those
reactions are known to be much simpler to be carried out than the condensation reactions necessary
to form COFs However as the coupling reactions are irreversible a lower degree of crystallinity is
achieved and the materials formed were amorphous The first PAF was reported in 2009 and
showed a very high BET surface area of 5600 m2g-1 The geometry of PAF-1 was equal to diamond
with the C-C bonds replaced by biphenyl Subsequent PAFs may be synthesized with longer linkers
between carbon tetragonal nodes Nevertheless synthesis of a PAF with quaterphenyl linker
provided an amorphous material of very low surface area due to the short range order
Synthetic complexities aside the diamond-like PAFs are interesting for their special geometries from
the viewpoint of the theorist It is interesting to see to what extent they follow the properties of
diamond Additionally the large surface area and high polarizability of aromatic rings are auxiliary for
enhanced hydrogen storage Thus we have explored the possibilities of diamond topology by
inserting various organic linkers in place of C-C bonds (Figure 13)
Figure 13 Diamond structure and various organic linkers to build up PAFs
Our studies of PAFs have been prepared for publication
19
Article 6 Structure electronic structure and hydrogen adsorption capacity of porous aromatic
frameworks
Binit Lukose Mohammad Wahiduzzaman Agnieszka Kuc Thomas Heine
See Appendix F for the article
In this article we have discussed the correlations of properties with the structures Exothermic
formation energies of PAFs calculated from the dehydrogenation of the linkers with CH4 indicate the
strong coordination of linkers in the diamond topology The studied PAFs exhibited bulk moduli much
smaller than diamond however in line with related MOFs and COFs The PAFs are semi-conductors
with their band gaps decrease with the increasing number of benzene rings in the linkers
Using Quantized Liquid Density Functional Theory (QLDFT) for hydrogen57 a method to compute
hydrogen adsorption that takes into account inter-particle interactions and quantum effects we
predicted a large choice of hydrogen uptake capacities in PAFs At room temperature and 100 bar
the predicted uptake in PAF-qtph was 732 wt which exceeds the 2015 DOE (US) target We
further discussed the structural impacts on the adsorption capacities
My contributions to this article include designing the materials performing calculations of stability
and electronic properties describing the adsorption capacities and preparing the manuscript
20
6 New Building Concepts
61 Isoreticular Series of SURMOFs
The growth of MOFs on substrates using liquid phase epitaxy (LPE) opened up a controlled way to
construct frameworks This is achieved by alternate immersion of the substrates in metal and ligand
precursors for several cycles Such MOFs named as SURMOFs can avoid interpenetration because
the degeneracy is lifted58 and are suited for conventional applications This is an advantage as
solvothermally synthesized MOFs with larger pore size suffer from interpenetration and the large
pores are hence not accessible for guest species
MOF-259 is a simple 2D architecture based on paddlewheel units formed by attaching four
dicarboxylic groups to Cu2+ or Zn2+ dimers yielding planar sheets with 4-fold symmetry The
arrangement of MOF layers possessed P2 or C2 symmetry Recently the group of C Woumlll at KIT has
synthesized an isoreticular series of SURMOFs derived from MOF-2 which has a homogeneous series
of linkers with different lengths (Figure 14) XRD comparisons suggest that these MOFs possess P4
symmetry The cell dimensions that correspond to the length of the pore walls range from 1 to 28
nm The diagonal pore-width of one of the SURMOFs (PPDC) accounts to 4 nm The symmetry of
SURMOFs as determined to be different from bulk MOF-2 should be understood by means of theory
As collaborators we simulated the structures and inspected each stacking corresponding to the
symmetries in order to understand the difference
Figure 14 The building units and schematic diagram of the formation of iso-reticular SURMOF
series
21
This collaborated work has been submitted for publication
Article 7 A novel series of isoreticular metal organic frameworks realizing metastable structures
by liquid phase epitaxy
Jinxuan Liu Binit Lukose Osama Shekhah Hasan Kemal Arslan Peter Weidler Hartmut Gliemann
Stefan Braumlse Sylvain Grosjean Adelheid Godt Xinliang Feng Klaus Muumlllen Ioan-Bogdan Magdau
Thomas Heine Christof Woumlll
See Appendix G for the article
The main contribution of this article was the experimental proof backed up by our computer
simulations that SURMOFs with exceptionally large pore sizes of more than 4 nm can be prepared in
the laboratory and they offer the possibility to host large materials such as clusters nanoparticles or
small proteins The most important contribution of theory was to show that while MOF-2 in P2
symmetry shows the highest stability the P4 symmetry as obtained using LPE for SURMOF-2
corresponds to a local minimum
My contribution to this article includes performing and analyzing the calculations Our theoretical
study went significantly beyond and will be published as separate article (Appendix H)
62 Metastability of SURMOFs
Following the difference in stability of SURMOFs-2 in comparison with MOF-2 we identified the role
of linkers as a constituent of the framework that provides the metastability to SURMOFs-2 (Figure
15) In MOFs linkers are essential in determining the topology as well as providing porosity Linkers
typically contain single or multiple aromatic rings and are ditopic and polytopic The orientation of
them normally undergoes lowest repulsion from the coordinated metal-oxide clusters and provides
high stability In the case of 2D MOFs non-covalent interlayer interactions lead to the most stable
arrangements Paddlewheels are identified as the reasons for the stability of bulk MOF-2 as they
form metal-oxide bridges across the MOF layers In the case of SURMOFs-2 the linkers are rotated in
a tightly-packed environment of layers and each linker in bulk MOF-2 is rotated in such a way that
any attempt of shifting is obstructed by repulsion between the neighboring linkers The energy
barrier for leaving P4 increases with the length of organic linkers Moreover SURMOF-2 derivatives
with extremely large linkers are energetically stable due to the increased London dispersion
interaction between the layers in formula units Thus we encountered a rare situation in which the
linkers guarantee the persistence of a series of materials in an otherwise unachievable state
22
Figure 15 Energy diagram of the metastable P4 and stable P2 structures
Our results on the linker guided stability of SUMORs-2 have been prepared for publication
Article 8 Linker guided metastability in templated Metal-Organic Framework-2 derivatives
(SURMOFs-2)
Binit Lukose Gabriel Merino Christof Woumlll Thomas Heine
See Appendix H for the article
This article is based solely on my scientific contributions
23
7 Summary
Nanotechnology is the way of ingeniously controlling the building of small and large structures with
intricate properties it is the way of the future a way of precise controlled building with incidentally
environmental benignness built in by design - Roald Hoffmann Chemistry Nobel Laureate 1981
Currently it is possible to design new materials rather than discovering them by serendipity The
design and control of materials at the nanoscale requires precise understanding of the molecular
interactions processes and phenomena In the next level the characteristics and functionalities of
the materials which are inherent to the material composition and structure need to be studied The
understanding of the materials properties may be put into the design of new materials
Computational tools to a large extend provide insights into the structures and properties of the
materials They also help to convert primary insights into new designs and carry out stability analysis
Our theoretical studies of a variety of materials have provided some insights on their underlying
structures and properties The primary differences in the material compositions and skeletons
attributed a certain choice in properties The contents of the articles discussed in the thesis may be
summarized into the following three parts
71 Validation of Methods
Simulations of nanoporous materials typically include electronic structure calculations that describe
and predict properties of the materials Density functional theory (DFT)30 is the standard and widely-
used tool for the investigation of the electronic structure of solids and molecules Even the optical
properties can be studied through the time-dependent generalization of DFT MOFs and COFs have
several hundreds of atoms in their unit cells DFT becomes inappropriate for such large periodic
systems because of its necessity of immense computational time and power Molecular force field
calculations60 on the other hand lack transferable parameterization especially for transition metal
sites and are hence of limited use to cover the large number of materials to be studied Apparently
a non-orthogonal tight-binding approximation to DFT called density functional tight-binding
(DFTB)28323561 has proven to be computationally feasible and sufficiently accurate This method
computes parameters from DFT calculations of a few molecules per pair of atom types The
parameters are generally transferable to other molecules or solids With self-consistent charge (SCC)
extension DFTB has improved accuracy In order to account weak forces the London dispersion
energy can be calculated separately using empirical potentials and added to total energy Successful
realizations of DFTB include the studies of large-scale systems such as biomolecules62
24
supramolecular compounds63 surfaces64 solids65 liquids and alloys66 Being an approximate method
DFTB needs validation Often one compares DFTB results of selected reference systems with those
obtained with DFT
Before electronic structure calculations of framework materials can be carried out it is necessary to
compute the equilibrium configurations of the atoms Geometry optimization (or energy
minimization) employs a mathematical procedure of optimization to move atoms so as to reduce the
net forces on them to negligible values We adopted the conjugate gradient scheme for the
optimizations using DFTB A primary test for the validation of these optimizations is the comparison
of cell parameters bond lengths bond angles and dihedral angles with the corresponding known
numbers We compared the DFTB optimized MOF COF and PAF geometries with the experimentally
determined or DFT optimized geometries and found that the values agree within 6 error
The angles and intensities of X-ray diffraction patterns can produce three-dimensional information of
the density of electrons within a crystal This can provide a complete picture of atomic positions
chemical bonds and disorders if any Hence simulating powder X-ray diffraction (PXRD) patterns of
optimized geometries and comparing them with experimental patterns minimize errors in the crystal
model Our focus on this matter directed us to identify the layer-offsets of 2D COFs for the first time
In the case of 3D COFs excellent correlations were generally observed between experimental and
simulated patterns Slight differences in the intensities of some of them were due to the presence of
solvents in the crystals as they were reported in the experimental articles PAFs as experimentally
being amorphous do not possess XRD comparisons MOFs within DFTB optimization have
undergone some changes especially in the dihedral angles in comparison with experimental
suggestion or DFT optimization This was verified from the differences in the simulated PXRD
patterns Another interesting outcome of XRD comparison was the discovery of P4 symmetry of
templated MOF-2 derivatives (SURMOFs-2) made by Woumlll et al
Bulk moduli which may be expressed as the second derivative of energy with respect to the unit cell
volume can give a sense of mechanical stability Our calculations provide the following bulk moduli
for some materials MOF-5 1534 (1537)67 Cu-BTC 3466 (3517)68 COF-102 206 (170)69 COF-
103 139 (118)69 COF-105 79 (57)69 COF-108 37 (29)69 COF-202 153 (110)69 The data in the
parenthesis give corresponding values found in the literature calculated using force-field methods
The bulk moduli of MOFs are comparable with the results in the literature however COFs show
significant differences Albeit the differences in values each type of calculation shows the trend that
bulk modulus decreases with decreasing mas density increasing pore volume decreasing thickness
of pore walls and increasing distance between connection nodes
25
Formation of framework materials from condensation of reactants may be energetically modeled
COFs are formed from dehydration reactions of boronic acids and hydroxyphenyl compounds The
formation energy calculated from the energies of the products and reactants can indicate energetic
stability Both DFTB for periodic structures and DFT for finite structures predict exothermic formation
of 3D COFs Likewise for 2D COFs such as COF-1 and 5 the formation of monolayers is found to be
endothermic within both the periodic model calculation using DFTB and finite model calculation
using DFT The stacking of layers provides them stability
72 Weak Interactions in 2D Materials
AB stacked natural graphite and ABC stacked rhombohedral graphite are found to be in proportions
of 85 and 15 in nature respectively whereas AA stacking has been observed only in graphite
intercalation compounds On the other hand boron nitride (h-BN) the synthetic product of boric
acid and boron trioxide exists with AA stacking 2D COFs were believed to have high symmetric AA
(eclipsed ndash P6mmm symmetry) stacking as boron nitride (h-BN) An exception was COF-1 which was
considered to have AB (staggered ndash P63mmc) stacking as graphite The AA stacking maximizes the
attractive London dispersion interaction between the layers a dominating term of the stacking
energy At the same time AA stacking always suffers repulsive Coulomb force between the layers
due to the polarized connectors It should be noted that in boron nitride oppositely charged boron
atoms and nitrogen atoms are facing each other The Coulomb interaction has ruled out possible
interlayer eclipse between atoms of alike charges This gave us the notion that 2D COFs cannot
possess layer-eclipsing Based on these facts we have suggested a shift of ~14 Aring between adjacent
layers at which one or two of the edge atoms of the hexagonal rings are situated perpendicular to
the center of adjacent rings In other words some or all of the hexagonal rings in the pore walls
undergo staggering with that of adjacent layers These lattice types were found to be very stable
compared to AA or AB analogues AA stacking was found to have the highest Coulombic repulsion in
each COF Within an error limit the bulk COFs may be either serrated or inclined The interlayer
separations of all the COF stacks are in the range of 30 to 36 Aring and increase in the order AB
serratedinclined AA70 The motivation for proposing inclined stacking for COFs is that the
rhombohedral graphite (ABC stacking) is the inclined version of the layer-offset in natural graphite
(AB stacking) The instability of AA stacking was also suggested by the studies of elastic properties of
COF-1 and COF-5 by Zhou et al52 The shear modulus show negative signs for the vertical alignment of
COF layers while they are small but positive for the offset of layers
The change of stacking should be visible in their PXRD patterns because each space group has a
distinct set of symmetry imposed reflection conditions We simulated the XRD patterns of COFs in
their known and new configurations and on comparison with the experimental spectrum the new as
26
well as AA stacking showed good agreement5470 The slight layer-offsets are only able to make a few
additional peaks in the vicinity of existing peaks and some changes in relative intensities The
relatively broad experimental data covers nearly all the peaks of the simulated spectrum In other
words the broad experimental peaks are explainable with layer-offset
A detailed analysis on the interlayer stacking of HHTP-DPB COF made by Dichtel et al53 is very
complementary53 This work uses molecular mechanics extensively to define potential energy
surfaces of stacking of two layers from eclipsed to staggered covering all possible layer offsets Low
energy region was near to the eclipsed stacking which was then zoomed in to examine using DFT for
higher accuracy The far-eclipsed regions encounter no energy barriers with the near-eclipsed regions
which suggest the unlikeness of forming near-staggered layering Within DFT the eclipsed
geometries are at high energy compared to the slight offsets around AA (~ 13 ndash 25 Aring away) For a
hexagonal COF the energetically preferred offset could be in one of the six hexagonal sectors (or
circle sectors) of central angle 60 The preference of falling into one sector over the others is
arbitrary and may be determined by symmetry set by nature or external parameters Hence the
layering order in bulk could be serrated inclined spiral or purely by random The latter case better
known as turbostratic disorder is then more like a rule for 2D COFs rather than an exception caused
by external forces This also explains why the experimental PXRD patters have relatively broad peaks
The layer-offset not only change the internal pore structure but also affect interlayer exciton and
vertical charge transport in opto-electronic applications
About stacking stability the square COFs are expected not to be different from hexagonal COFs
because the local environment causing the shifts is nearly the same The DFTB based calculations
reported along with the synthesis of electron-transporting 2D-NiPc-BTDA COF supports this notion71
Two shifted configurations ndash at 07 and 14 Aring away from AA ndash appeared to be energetically preferred
over the AA stacking It appears that AA stacking is only possible for boron nitride-like structures
were adjacent layers have atoms with opposite charges in vertical direction
SURMOFs-2 a series of paddlewheel-based 2D MOFs possess a different symmetry than
solvothermally synthesized MOF-2 due to the differences in interlayer interactions Typically the
interaction between the paddlewheels favors P2 or C2 symmetry of bulk MOF-2 rather than the P4
symmetry of SURMOFs-2 Bader charge analysis reveals that electron distribution between the Cu-
paddlewheels is the lowest when they follow P4 symmetry This indicates the smallest possibility of
having a direct Cu-Cu bond between the paddlewheels In monolayer organic linkers undergo no
rotation with respect to metal dimers
27
X-ray diffraction of SURMOFs-2 made by Woumlll et al indicated P4 symmetry and a relatively small
interlayer separation This increases the repulsion between the linkers and enforces them to rotate
The rotations of each linker in a layer are controlled globally The rotated linkers in adjacent layers
increase London dispersion however a paddlewheel-led shift towards any side increases repulsion
thereby locking SURMOF-2 in P4 packing Hence with the special arrangement of rotated linkers the
linker-linker interaction overcomes the paddlewheel-paddlewheel interaction
P4 symmetry opens up the pores more clearly than P2 or C2 Additionally the metal sites that
typically host solvents are inaccessible Furthermore improvements such as functionalizing the linker
may be easily carried out
Based on our calculations on 2D materials we emphasize the role of van der Waals interactions in
determining the layer-to-layer arrangements The promise of reticular chemistry which is the
maintainability of structural integrity of the building blocks in the construction process is partly
broken in the case of SURMOFs-2 This is because due to repulsion the linkers undergo rotation with
respect to the carboxylic parts albeit keeping the topology
73 Structure-Property Relationships
We have studied a variety of materials from the classes of MOFs COFs and PAFs The structural
differences arise from the differences in the constituents andor their arrangements Properties in
general are interlinked with structural specifications Therefore it is beneficial to know the
relationship between the structural parameters and properties
The mass density is an intensive property of a material In the area of nanoporous materials a crystal
with low mass density has advantages in applications involving transport Definitely the mass density
decreases with increasing pore volume Still the number of atoms in the wall and their weights are
important factors The pore size does not relate directly to the atom counts The volume per atom
(inverse of atom density) another intensive property of a material obliquely gives porosity Figure
16 shows the variation of mass density with volume per atom (including the volume of the atom) for
MOFs COFs and PAFs MOFs show higher mass density compared to other materials of identical
atom density implying the presence of heavy elements It is to be noted that Cu-BTC has mass
density 184 times higher than COF-102 The presence of metals in the form of oxide clusters in MOFs
increases the mass density and decreases the volume per atom Note that the low-weighted MOF in
the discussion (MOF-205) is heavier than many of the PAFs and COFs The relatively high mass
density and small volume per atom of COF-300 is due to its 5-fold interpenetration whereas COF-202
has additional tert-butyl groups which do not contribute to the system shape but affect the mass
density and the volume per atom COF-102 and 103 have same topology but different centers (C and
28
Si respectively) The Si centers increases the cell volume and decreases mass density COF-105 (Si
centers) and COF-108 (C centers) additionally differ in their topologies (ctn and bor respectively) It
appears that bor topology expands the material compared to ctn PAFs hold the extreme cases PAF-
phnl has the highest atom and mass densities whereas PAF-qtph has the smallest atom and mass
densities
Figure 16 Mass density as a function of volume per atom for MOFs COFs and PAFs
The bulk modulus indicates the materialsrsquo resistance towards compression which should in principle
decrease with increasing porosity At the same time larger number of atoms making covalent
networks in unit volume should supply larger bulk moduli Thus differences in molecular contents
and architectures give rise to different bulk moduli It is interesting to see how the mechanical
stability of nanoporous materials is related with the atom density We have obtained a correlation
between bulk modulus (B) and volume per atom (VA) in the case of the studied MOFs COFs and PAFs
as follows
29
where k is a constant (see Figure 17) The inverse-volume correlation indicates that the materials
close to the fitting curve have average bond strengths (interaction energy between close atoms)
identical to each other independent of number of bonds bond order and branching Only Cu-BTC
COF-202 and COF-300 show significant differences from the fitted curve Cu-BTC has 17 times larger
bulk modulus compared to COF-102 of similar volume per atom which implies the substantially
higher strength of the bond network resulting from paddlewheel units and tbo topology
Interpenetration decreased the volume per atom however increased bulk modulus through
interlayer interactions in the case of COF-300 The tert-butyl groups in COF-202 do not transmit its
inherent stability to the COF significantly however decreases the volume per atom Comparison
between COF-102 (C centers) and COF-103 (Si centers) discloses that Si centers decrease the
mechanical stability Comparison between COF-105 (ctn) and COF-108 (bor) suggests that ctn
topology possess higher stability This indicates that local angular preferences can amend the
strength of the bulk material
Figure 17 Bulk modulus as a function of volume per atom for MOFs COFs and PAFs
Band gaps of all the studied materials show that they are semiconductors except of Cu-BTC which
has unpaired electrons at the Cu centers Our studies of the density of states (DOS) of bulk MOFs and
the cluster models that have finite numbers of connectors and linkers show that electronic structure
30
stays unchanged when going from cluster to periodic crystal In the case of 2D COFs the DOS of
monolayers and bulk materials were identical Interpenetrated COF-300 showed no difference in the
electronic structure in comparison with the non-interpenetrated structure Based on these results
we may reach into a premature conclusion that electronic structure of a solid is determined by its
constituent bonded network sufficiently large to include all its building units
HOMO-LUMO gap of the building units determine the band gap of a framework material We have
observed that PAFs nearly reproduced the HOMO-LUMO gaps of the linkers 3D COFs that are made
of more than one building unit show that the band gap is slightly smaller than the smallest of the
HOMO-LUMO gaps of the building units For example
TBPM (43 eV) + HHTP (34 eV) COF-105COF-108 (32 eV)
TBPM (43 eV) + TBST (139 eV) COF-202 (42 eV)
TAM (41 eV) + TA (26 eV) COF-300 (23 eV)
The compound names are taken from appendix E Additionally the band gaps decrease with
increasing number of conjugated rings similar to HOMO-LUMO gaps of the linkers
I believe that the studies in the thesis have helped to an extent to understand the structure
stability and properties of different classes of framework materials The benchmark structures we
studied have the essential features of the classes they represent Ab-initio based computational
studies of several periodic structures are exceptional and thus have its place in the literature
31
List of Abbreviations
ADF Amsterdam Density Functional code
BLYP Becke-Lee-Yang-Parr functional
B3LYP Becke 3-parameter Lee Yang and Parr functional
COF Covalent-Organic Framework
CP Coordination Polymer
CTF Covalent-Triazine Framework
DC Dispersion correction
DFT Density Functional Theory
DFTB Density Functional Tight-Binding
DOS Density of States
DOE (US) Department of Energy (United States)
DZP Double-Zeta Polarized basis set
GGA Generalized Gradient Approximation
LCAO Linear Combination of Atomic Orbitals
LPE Liquid Phase Epitaxy
MOF Metal-Organic Framework
PAF Porous Aromatic Framework
PBE Perdew-Burke-Ernzerhof functional
PXRD Powder X-ray Diffraction Pattern
QLDFT Quantized Liquid Density Functional Theory
RCSR Reticular Chemistry Structure Resource
SBU Secondary Building Unit
SCC Self-Consistent Charge
TZP Triple-Zeta Polarized basis set
SURMOF Surface-Metal-Organic Framework
32
List of Figures
Figure 1 ldquoAdamantinerdquo unit of diamond structure and diamond (dia) net 3
Figure 2 CU-BTC MOF and the corresponding tbo net 4
Figure 3 Fundamental building blocks and SBUs of Cu-BTC and extension of SBUs following crystal
structure 5
Figure 4 Schematic diagram of the single pore of MOF and An example of a simple cubic MOF in
which zinc-oxide tetrahedron are linked together by benzene linkers Atom colors blue ndash Zn red ndash O
grey ndash C white ndash H 6
Figure 5 Schematic diagram of interpenetrated and interwoven frameworks 7
Figure 6 Schematic diagram of SURMOF 8
Figure 7 SBUs and topologies of 2D COFs 9
Figure 8 Deconstructed building units their schematic representations and final geometries of
HKUST-1 (Cu-BTC) MOF-5 MOF-177 MOF-205 and MIL-53 11
Figure 9 AA and AB layer stacks of hexagonal layers 13
Figure 10 Stacking energy Es vs shift of adjacent layers for COF-5 Different stacking possibilities and
their energies are also shown 14
Figure 11 Schematic diagram of different building units forming 2D COFs 16
Figure 12 Schematic diagram of tetragonal and triangular building blocks forming lsquoctnrsquo or lsquoborrsquo
topologies 17
Figure 13 Diamond structure and various organic linkers to build up PAFs 18
Figure 14 The building units and schematic diagram of the formation of iso-reticular SURMOF series
20
Figure 15 Energy diagram of the metastable P4 and stable P2 structures 22
Figure 16 Mass density as a function of volume per atom for MOFs COFs and PAFs 28
Figure 17 Bulk modulus as a function of volume per atom for MOFs COFs and PAFs 29
33
References
(1) Morris R E Wheatley P S Angewandte Chemie-International Edition 2008 47 4966
(2) Li J-R Kuppler R J Zhou H-C Chemical Society Reviews 2009 38 1477
(3) Corma A Chemical Reviews 1997 97 2373
(4) Seo J S Whang D Lee H Jun S I Oh J Jeon Y J Kim K Nature 2000 404 982
(5) Lee J Kim J Hyeon T Advanced Materials 2006 18 2073
(6) Stein A Wang Z Fierke M A Advanced Materials 2009 21 265
(7) Velev O D Jede T A Lobo R F Lenhoff A M Nature 1997 389 447
(8) Beck J S Vartuli J C Roth W J Leonowicz M E Kresge C T Schmitt K D Chu C T
W Olson D H Sheppard E W McCullen S B Higgins J B Schlenker J L Journal of the
American Chemical Society 1992 114 10834
(9) Kresge C T Leonowicz M E Roth W J Vartuli J C Beck J S Nature 1992 359 710
(10) Ying J Y Mehnert C P Wong M S Angewandte Chemie-International Edition 1999 38
56
(11) Velev O D Kaler E W Advanced Materials 2000 12 531
(12) Stein A Microporous and Mesoporous Materials 2001 44 227
(13) Tanev P T Pinnavaia T J Science 1996 271 1267
(14) Yaghi O M OKeeffe M Ockwig N W Chae H K Eddaoudi M Kim J Nature 2003
423 705
(15) OKeeffe M Peskov M A Ramsden S J Yaghi O M Accounts of Chemical Research
2008 41 1782
(16) Delgado-Friedrichs O OKeeffe M Journal of Solid State Chemistry 2005 178 2480
(17) Delgado-Friedrichs O Foster M D OKeeffe M Proserpio D M Treacy M M J Yaghi
O M Journal of Solid State Chemistry 2005 178 2533
(18) OKeeffe M Yaghi O M Chemical Reviews 2012 112 675
(19) Tranchemontagne D J L Ni Z OKeeffe M Yaghi O M Angewandte Chemie-
International Edition 2008 47 5136
(20) Hayashi H Cote A P Furukawa H OKeeffe M Yaghi O M Nature Materials 2007 6
501
(21) Cote A P Benin A I Ockwig N W OKeeffe M Matzger A J Yaghi O M Science
2005 310 1166
(22) El-Kaderi H M Hunt J R Mendoza-Cortes J L Cote A P Taylor R E OKeeffe M
Yaghi O M Science 2007 316 268
(23) Rowsell J L C Yaghi O M Microporous and Mesoporous Materials 2004 73 3
(24) Hermes S Zacher D Baunemann A Woell C Fischer R A Chemistry of Materials
2007 19 2168
34
(25) Kuhn P Antonietti M Thomas A Angewandte Chemie-International Edition 2008 47
3450
(26) Ben T Ren H Ma S Cao D Lan J Jing X Wang W Xu J Deng F Simmons J M
Qiu S Zhu G Angewandte Chemie-International Edition 2009 48 9457
(27) Porezag D Frauenheim T Kohler T Seifert G Kaschner R Physical Review B 1995
51 12947
(28) Seifert G Porezag D Frauenheim T International Journal of Quantum Chemistry 1996
58 185
(29) Kohn W Sham L J Physical Review 1965 140 1133
(30) Parr R G Yang W Density-Functional Theory of Atoms and Molecules New York Oxford
University Press 1989
(31) Hohenberg P Kohn W Physical Review B 1964 136 B864
(32) Elstner M Porezag D Jungnickel G Elsner J Haugk M Frauenheim T Suhai S
Seifert G Physical Review B 1998 58 7260
(33) Zhechkov L Heine T Patchkovskii S Seifert G Duarte H A Journal of Chemical
Theory and Computation 2005 1 841
(34) Elstner M Hobza P Frauenheim T Suhai S Kaxiras E Journal of Chemical Physics
2001 114 5149
(35) Oliveira A F Seifert G Heine T Duarte H A Journal of the Brazilian Chemical Society
2009 20 1193
(36) Seifert G Joswig J-O Wiley Interdisciplinary Reviews-Computational Molecular Science
2012 2 456
(37) Heine T Rapacioli M Patchkovskii S Frenzel J Koester A M Calaminici P
Escalante S Duarte H A Flores R Geudtner G Goursot A Reveles J U Vela A Salahub D
R deMon deMon-nano edn deMon-nano 2009
(38) BCCMS B DFTB+ - Density Functional based Tight binding (and more)
(39) Monkhorst H J Pack J D Physical Review B 1976 13 5188
(40) SCM Amsterdam Density Functional 2012
(41) University of Karlsruhe and Forschungszentrum Karlsruhe gGmbH - TURBOMOLE V63
2011 2007
(42) Dovesi R Saunders V R Roetti C Orlando R Zicovich-Wilson C M Pascale F
Civalleri B Doll K Harrison N M Bush I J DrsquoArco P Llunell M CRYSTAL09 Users Manual
University of Torino Torino 2009 2009
(43) Wadt W R Hay P J Journal of Chemical Physics 1985 82 284
(44) Roy L E Hay P J Martin R L Journal of Chemical Theory and Computation 2008 4
1029
35
(45) Ehlers A W Bohme M Dapprich S Gobbi A Hollwarth A Jonas V Kohler K F
Stegmann R Veldkamp A Frenking G Chemical Physics Letters 1993 208 111
(46) Garberoglio G Skoulidas A I Johnson J K Journal of Physical Chemistry B 2005 109
13094
(47) Han S S Mendoza-Cortes J L Goddard W A III Chemical Society Reviews 2009 38
1460
(48) Getman R B Bae Y-S Wilmer C E Snurr R Q Chemical Reviews 2012 112 703
(49) Cote A P El-Kaderi H M Furukawa H Hunt J R Yaghi O M Journal of the American
Chemical Society 2007 129 12914
(50) Wan S Guo J Kim J Ihee H Jiang D Angewandte Chemie-International Edition 2008
47 8826
(51) Wan S Guo J Kim J Ihee H Jiang D Angewandte Chemie-International Edition 2009
48 5439
(52) Zhou W Wu H Yildirim T Chemical Physics Letters 2010 499 103
(53) Spitler E L Koo B T Novotney J L Colson J W Uribe-Romo F J Gutierrez G D
Clancy P Dichtel W R Journal of the American Chemical Society 2011 133 19416
(54) Lukose B Kuc A Heine T Chemistry-a European Journal 2011 17 2388
(55) Uribe-Romo F J Hunt J R Furukawa H Klock C OKeeffe M Yaghi O M Journal of
the American Chemical Society 2009 131 4570
(56) Schmid R Tafipolsky M Journal of the American Chemical Society 2008 130 12600
(57) Patchkovskii S Heine T Physical Review E 2009 80
(58) Shekhah O Wang H Paradinas M Ocal C Schuepbach B Terfort A Zacher D
Fischer R A Woell C Nature Materials 2009 8 481
(59) Li H Eddaoudi M Groy T L Yaghi O M Journal of the American Chemical Society
1998 120 8571
(60) Rappe A K Casewit C J Colwell K S Goddard W A Skiff W M Journal of the
American Chemical Society 1992 114 10024
(61) Frauenheim T Seifert G Elstner M Hajnal Z Jungnickel G Porezag D Suhai S
Scholz R Physica Status Solidi B-Basic Research 2000 217 41
(62) Elstner M Cui Q Munih P Kaxiras E Frauenheim T Karplus M Journal of
Computational Chemistry 2003 24 565
(63) Heine T Dos Santos H F Patchkovskii S Duarte H A Journal of Physical Chemistry A
2007 111 5648
(64) Sternberg M Zapol P Curtiss L A Molecular Physics 2005 103 1017
(65) Zhang C Zhang Z Wang S Li H Dong J Xing N Guo Y Li W Solid State
Communications 2007 142 477
36
(66) Munch W Kreuer K D Silvestri W Maier J Seifert G Solid State Ionics 2001 145
437
(67) Bahr D F Reid J A Mook W M Bauer C A Stumpf R Skulan A J Moody N R
Simmons B A Shindel M M Allendorf M D Physical Review B 2007 76
(68) Amirjalayer S Tafipolsky M Schmid R Journal of Physical Chemistry C 2011 115
15133
(69) Amirjalayer S Snurr R Q Schmid R Journal of Physical Chemistry C 2012 116 4921
(70) Lukose B Kuc A Frenzel J Heine T Beilstein Journal of Nanotechnology 2010 1 60
(71) Ding X Chen L Honsho Y Feng X Saenpawang O Guo J Saeki A Seki S Irle S
Nagase S Parasuk V Jiang D Journal of the American Chemical Society 2011 133 14510
37
Appendix A
Review Covalently-bound organic frameworks
Binit Lukose and Thomas Heine
To be submitted for publication after revision
Contents
1 Introduction
2 Synthetic achievements
21 Covalent Organic Frameoworks (COFs)
22 Covalent-Triazine Frameworks (CTFs)
23 Porous Aromatic Frameworks (PAFs)
24 Schemes for synthesis
25 List of materials
3 Studies of the underlying structure and properties of COFs
4 Applications
41 Gas storage
411 Porosity of COFs
412 Experimental measurements
413 Theoretical preidctions
414 Adsorption sites
415 Hydrogen storage by spillover
42 Diffusion and selectivity
43 Suggestions for improvement
431 Geometry modifications
432 Metal doping
433 Functionalization
5 Conclusions
6 List and pictures of chemical compounds
38
1 Introduction
Nanoporous materials have perfectly ordered voids to accommodate to interact with and to
discriminate molecules leading to prominent applications such as gas storage separation and sieving
catalysis filtration and sensoring1-4 They are defined as porous materials with pore diameters less
than 100 nm and are classified as micropores of less than 2 nm in diameter mesopores between 2
and 50 nm and macropores of greater than 50 nm The internal geometry that determines pore size
and surface area can be precisely engineered at molecular scales Reticular synthetic methods
suggested by Yaghi and co-workers5-7 can accomplish pre-designed frameworks The procedure is to
select rigid molecular building blocks prudently and assemble them into destined networks using
strong bonds
Several types of materials have been synthesized using reticular chemistry concepts One prominent
group which is much appreciated for its credentials is the Metal-Organic Frameworks (MOFs)8-13 in
which metal-oxide connectors and organic linkers are judiciously selected to form a large variety of
frameworks The immense potential of MOFs is facilitated by the fact that all building blocks are
inexpensive chemicals and that the synthesis can be carried out solvothermally The progress in MOF
synthesis has reached the point that some of the MOFs are commercially available Several MOFs of
ultrahigh porosity have also been successfully synthesized notably the non-interpenetrated IRMOF-
74-XI which has a pore size of 98 Aring Scientists are also able to synthesize MOFs from cheap edible
natural products14 Since the discovery of MOFs many other crystalline frameworks have been
synthesized using reticular chemistry such as Metal-Organic Polyhedra (MOP)15-19 Zeolite
Imidazolate Frameworks (ZIFs)20-22 and Covalent Organic Frameworks (COFs)23-29
COFs are composed of covalent building blocks made of boron carbon oxygen and hydrogen in
many cases also including nitrogen or silicon stitched together by organic subunits The atoms are
held together by strong covalent bonds Depending on the selection of building blocks the COFs may
form in two (2D) or three (3D) dimensions Planar building blocks are the constituents of 2D COFs
whereas for the formation of 3D COFs typically tetragonal building blocks are involved High
symmetric covalent linking as it is perceived in reticular chemistry was confirmed for the end
products5
Unlike the case of supramolecular assemblies the absence of noncovalent forces between the
molecular building units endorses exceptional rigidity and stability for COFs They are in general
thermally stable above 4000 C Also in COFs the stitching of molecular building units guarantees an
39
increased order and allows control over porosity and composition Without any metals or other
heavy atoms present they exhibit low mass densities This gives them advantages over MOFs in
various applications for example higher gravimetric capacities for gas storage3031 The lowest
density crystal known until today is COF-10824 In addition COFs exhibit permanent porosity with
specific surface areas that are comparable to MOFs and hence larger than in zeolites and porous
silicates
MOF and COF crystals possess long range order although COFs have been achieved so far only at the
μm scale Reversible solvothermal condensation reactions are credited for the high order of
crystallinity Other new framework materials such as Covalent Triazine Frameworks (CTFs) and
Porous Aromatic Frameworks (PAFs) differed in ways of synthesis The former was prepared by
ionothermal polymerization while the latter was produced by coupling reactions PAFs lack long
range order in the crystals as a result of the irreversible synthesis Nevertheless many of the
materials are promisingly good for applications In this review we intend to discuss the synthetic
achievements of COF CTFs and PAFs and studies on their structure properties and prominent
applications
For easy understanding of the atomic geometry of a material the reader is referred to Table 1 which
gives the names of the framework materials the reactants and the synthesis scheme Figure 5 shows
the reactants and Figure 4 shows the reaction schemes Example reactions are given in Figure 1-3
Abbreviations of each chemical compound are given in Section 6
2 Synthetic achievements
21 Covalent Organic Frameworks (COFs)
In the year 2005 Yaghi et al23 achieved the crystallization of covalent organic molecules in the form
of periodic extended layered frameworks The condensation of discrete molecules of different sizes
enabled the synthesis of micro- and mesoporous crystalline COFs27 The selection of molecules with 2
and 3 connection sites for synthesis led to the formation of hexagonal layered geometries32 Yaghi et
al have synthesized half a dozen three-dimensional crystalline COFs solvothermally using tetragonal
building blocks containing 4 connection sites since 2007242829 Figure 1 shows a few examples of 2D
and 3D COFs formed by the self-condensation of boronic acids such as BDBA TBPM and TBPS and co-
condensation of the same boronic acids with HHTP
40
Figure 1 Solvothermal condensation (dehydration) of monomers to form 2D and 3D COFs Atom colors are boron yellow oxygen red yellow (big) silicon grey carbon
Alternate synthetic procedures were also exploited for production and functionalization of COFs
Lavigne et al33 showed that the synthesis of COFs using polyboronate esters is facile and high-yielded
41
Boronate esters often contain multiple catechol moieties which are prone to oxidation and are
insoluble in organic solvents Dichtel et al3435 presented Lewis acid-catalysis procedures to form
boronate esters of catechols and arylboronic acids without subjecting to oxidation Cooper et al36
successfully utilized microwave heating techniques for rapid production (~200 times faster than
solvothermal methods) of both 2D and 3D COFs The syntheses of phthalocyanine3437 or porphyrin38
based square COFs have been reported in literature The latter was noticed for its time-dependent
crystal growth which also affects its pore parameters
Abel et al39 reported the ability to form COFs on metallic surfaces COF surfaces (SCOFs) have been
formed by molecular dehydration of boroxine and triphenylene on a silver surface Albeit with some
defects the materials showed high temperature stability allowing to proceed with functionalization
Lackinger et al40 realized the role of substrates on the formation of surface COFs from the surface-
generated radicals Halogen substituted polyaromatic monomers can be sublimed onto metal
substrates and ultimately turned into a COF after homolysis and intermolecular colligation
Chemically inert graphite surfaces cannot support the homolysis of covalent carbon-halogen bonds
and thus cannot initiate the subsequent association of radicals COF layers can be formed onto
Cu(111)40 and Au(111)41 surfaces but with lower crystal ordering Beton et al41 deposited the
monomers on annealed Au(111) surfaces which supplied surface mobility for the monomers and
subsequently increased porosity and surface coverage of the covalent networks Unlike the bulk form
at the surface the monomers were self-organized onto polygons with 4 to 8 edges Such a template
was able to organize a sublimed layer of C60 fullerenes the number of C60 molecules adsorbed in a
cavity was correlated to the size of the polygon
In 2009 Sassi et al42 studied the problem of crystal growth of COFs on substrates They calculated
the reaction mechanism of 14-benzenediboronic acid (BDBA) on Ag(111) surface self-condensation
of BDBA molecules in solvents leads to the formation of two-dimensional (the planar) stacked COF-1
For the surface COFs the study using Density Functional Theory reveals that there are neither
preferred adsorption sites for the molecules nor a charge transfer between the molecules and the
surface Hence the electronic structure of the molecules remains unchanged and the role of the
metal surface is limited in providing the planar nature for the polymer The free reaction enthalpy
(Gibbs free energy) difference of the reaction ∆G = ∆H ndash T∆S is approximated using the harmonic
approximation taking into account the geometrical restrictions of the metal surface and the entropic
contributions of the released water molecules As result the formation of SCOF-1 has been found to
be exothermic if more than six monomers are involved Ourdjini et al43 reported the polymerization
of 14-benzenediboronic acid (BDBA) on different substrates (Ag(111) Ag(100) Au(111) and Cu(111))
and at different source and substrate temperatures to follow how molecular flux and adsorption-
42
diffusion environments should be controlled for the formation of polymers with the smallest number
of structural defects High flux is essential to avoid H-bonded supramolecular assemblies of
molecules and the substrate temperature needs to be optimized to allow the best surface diffusion
and longest residential time of the reactants Achieving these two contradictory conditions together
is a limitation for some substrates eg for copper Silver was found to be the best substrate for
producing optimum quality polymers Controlling the growth parameters is important since
annealing cannot cure defects such as formation of pentagons heptagons and other non-hexagonal
shapes which involved strong covalent bonds
Ditchel et al44 successfully grew COF films on monolayer graphene supported by a substrate under
operationally simple solvothermal conditions The films have better crystallinity compared to COF
powders and can facilitate photoelectronic applications straightaway Zamora et al45 have achieved
exfoliation of COF layers by sonication The thin layered nanostructures have been isolated under
ambient conditions on surfaces and free-standing on carbon grids
A noted characteristic of COFs is its organic functionality Judiciously selected organic units ndash pyrene
and triphenylene (TP) ndash for the production of TP COF can not only harvest photons over a wide range
but also facilitate energy migration over the crystalline solid25 Polypyrene (PPy)-COF that consists of
a single aromatic entity ndash pyrene- is fluorescent upon the formation of excimers and hence undergo
exciton migration across the stacked layers26 A nickel(II)-phthalocyanine based layered square COF
that incorporates phenyl linkers and forms cofacially stacked H-aggregates was reported to absorb
photons over the solar spectrum and transport charges across the stacked layers37 COF-366 and
COF-66 showed excellent hole mobility of 81 and 30 cm2 V-1s-1 surpassing that of all polycrystalline
polymers known until today46 A first example of an electron-transporting 2D COF was reported
recently47 This square COF was built from the co-condensation of nickel(II) phthalocyanine and
electron-deficient benzothiadiazole With the nearly vertical arrangement of its layers it exhibits an
electron mobility of 06 cm2 V-1s-1 The enhanced absorbance is over a wide range of wavelengths up
to 1000 nm In addition it exhibits panchromatic photoconductivity and high IR sensitivity
Lavigne et al48 achieved the condensation of alkyl (methyl ethyl and propyl) substituted organic
building units with boronic acids forming functionalized COFs The alkyl groups contribute for higher
molar adsorption of H2 however the increased mass density of the functionalized COFs cause for
decreased gravimetric storage capacities Boronate ester-linked COFs are stable in organic solvents
however undergo rapid hydrolysis when exposed in water49 In particular the instability of COF-1
upon H2O adsorption shall be mentioned here50 Lanni et al claimed that alkylation could bring
hydrolytic stability into COFs49
43
Functionalization and pore surface engineering in 2D COFs can be improved if azide appended
building blocks are stitched together in click reactions with alkynes51 Control over the pore surface
is made possible by the use of both azide appended and bare organic building units the ratios of
which is matching with the final amount of functionalization in the pore walls The click reactions of
azide groups with alkynes within the pores lead to the formation of triazole-linked groups on the
pore surfaces This strategy also gives the relief of not condensing the already functionalized building
units Jiang et al have asserted eclipsed layering for most of the final COFs using powder X-ray
diffraction pattern Exceptions are the hexagonal functionalized COFs with relatively high azide-
content (ge 75) Their weak XRD signals suggest possible layer-offset or de-lamellation Although
functionalization on pore surfaces exhibits the nature of lowering the surface area the possibility to
add reactive groups within the pores can be utilized for gas-sorption selectivity The authors have
claimed a 16 fold selectivity of CO2 over N2 using 100 acetyl-rich COF-5 compared to the pure COF-5
The range of the click reaction approach is so wide that relatively large chromophores can be
accommodated in the pores thereby making COF-5 fluorescent
Functionalization of 3D COFs has just reported in 201252 Dichtel et al used a monomer-truncation
strategy where one of the four dihydroxyborylphenyl moieties of a tetrahedral building unit was
replaced by a dodecyl or allyl functional group Condensation of the truncated and the pure
tetrahedral blocks in a certain ratio created a functionalized 3D COF The degree of functionalization
was in proportion with the ratio The crystallinity was maintained except for high loading (gt 37) of
truncated monomers The pore volume and the surface area were decreased as a function of loading
level
A heterogeneous catalytic activity of COFs is achieved with an imine-linked palladium doped COF by
enabling Suzuki-Miyaura coupling reaction within the pores53 The COF-LZU1 has a layered geometry
that has the distance between nitrogen atoms in the neighboring linkers in a level (~37 Aring) sufficient
to accommodate metal ions Palladium was incorporated by a post-treatment of the COF PdCOF-
LZU1 was applied to catalyze Suzuki-Miyaura coupling reaction This coupling reaction is generally
used for the facile formation of C-C bonds54 The increased structural regularity and high insolubility
in water and organic solvents are adequate qualities of imine-linked COFs to perform as catalysts
Experiments with the above COF show a broad scope of the reactants excellent yields of the
products and easy recyclability of the catalyst
The comparatively higher thermal stability of COFs is often noted and is explainable with their strong
covalent bonds The reversible dehydrations for the formation of most of the COFs point to their
instability in the presence of water molecules This has been tested and verified for some layered
COFs with boronate esters49 Such COFs are stable as hosts for organometallic molecules55 COF-102
44
framework was found to be stable and robust even in the presence of highly reactive cobaltocenes
The highly stable ferrocenes appear to have an arrangement within the framework led by π-π
interactions
22 Covalent Triazine Frameworks (CTFs)
In the year 2008 Thomas et al56 synthesized a crystalline extended porous nanostructure by
trimerization of polytriazine monomers in ionothermal conditions With the catalytic support of ZnCl2
three aromatic nitrile groups can be trimerized to form a hexagonal C3N3 ring The resulting structure
known as Covalent Triazine-based Frameworks (CTFs) is similar to 2D COFs in geometry and atomic
composition (see Figure 2) Usage of larger non-planar monomers at higher amounts of ionic salts
however led to the formation of contorted structures Interestingly those structures showed
enhanced surface area and pore volume The trimerization of monomers that lack a linear
arrangement of nitrile groups ended up as organic polymer networks Later the same group
introduced mesoporous channels onto a CTF (CTF-2)57 by increasing temperature and ion content
The resulting structure however was amorphous It is concluded that the reaction parameters and
the amount of salt play a crucial role for tuning the porosity and controlling the order of the material
respectively58
Figure 2 Trimerization of nitrile monomers to form CTF-1 Atom colors blue N grey C white H
Ben et al59 followed ionothermal conditions for the targeted synthesis of a 3D framework using
tetraphenylmethane block and a triangular triazine ring The end product was amorphous thermally
stable up to 400 0C and could be applied for the selective sorption of benzene over cyclohexane On a
later synthetic study Zhang et al60 reported that this polymerization could be accomplished in short
45
reaction times under microwave enhanced conditions The template-free high temperature dynamic
polymerization reactions constitute irreversible carbonization reactions coupled with reversible
trimerization of nitriles The irreversibility encounter in these condensation reactions is responsible
for the production of frameworks as amorphous solids6162
An amorphous yet microporous CTF was found to be able to catalyze the oxidation of glycerol with
Pd nanoparticles that are dispersed in the framework63 This solid catalyst appears to be strong
against deactivation and selective toward glycerate compared to Pd supported activated carbon This
is attributed to the relatively high stability of Pd nanoparticles that benefit from the presence of
nitrogen functionalities in CTF An advancement on selective oxidation of methane to methanol at
low temperature is accomplished by using an amorphous bipyridyl-rich platinum-coordinated CTF as
catalyst64
23 Porous Aromatic Frameworks (PAFs)
a notably high surface area (BET surface area of 5600 m2g-1 Langmuir surface area of 7100 m2g-1)65
PAF-1 has been synthesized in a rather simple way using a nickel(0)-catalyzed Yamamoto-type66
Ullmann cross-coupling reaction67(see Figure 3) The material is thermally (up to 520 0C in air) and
hydrothermally (in boiled water for a minimum of seven days) stable The framework exposes all
faces and edges of the phenyl rings to the pores and hence the high surface area can be achieved
while its high stability is inherited from the parent diamond structure The synthesized material was
verified as disordered and amorphous yet with uniform pore diameters The excess H2 uptake
capacity of PAF-1 was measured as 70 wt at 48 bar and 77 K It can adsorb 1300 mg g-1 CO2 at 40
bar and room temperature PAF-1 was also tested for benzene and toluene adsorption
Figure 3 Coupling reaction using halogenated tetragonal blocks for the formation of diamond-like PAF-1 Atom colors red Br grey C white H
46
An attempt to synthesize a diamondoid with quaterphenyl as linker is described recently68 A
tetrahedral unit and a linear linker each of them has two phenyl rings were polymerized using the
Suzuki-Miyaura coupling reaction54 The product (PAF-11) however stayed behind theoretical
predictions and performed poorly pointing to its shortcomings such as short-range order distortion
and interpenetration of phenyl rings Nevertheless the amorphous solid displayed high thermal and
chemical stabilities proneness for adsorbing methanol over water and usability for eliminating
harmful aromatic molecules
PAF-569 is a two-dimensional hexagonal organic framework synthesized using the Yamamoto-type
Ullmann reaction This material is composed only of phenyl rings however lack long range order
because of the distortion of the phenyl rings and kinetics controlled irreversible coupling reaction It
retains a uniform pore diameter and provides high thermal and chemical stability despite its
amorphous nature PAF-5 was suggested for adsorbing organic pollutants at saturated vapour
pressure and room temperature
Co-condensation reactions with piperazine enabled nucleophilic substitution of cyanuric chloride to
form a covalently linked porous organic framework PAF-670 The synthesis in mild conditions yields a
product with uniform morphology and a certain degree of structural regularity Being nontoxic this
material was tested for drug delivery thereby stepping into biomedical applications of covalently
linked organic frameworks Primary tests suggest that PAF-6 could be promoted for drug release for
its permanent porosity and facile functionalization In comparison with MCM-4171 a widely tested
inorganic framework PAF-6 performed equally or even superiorly
24 Schemes for synthesis
The majority of the COFs were synthesized using solvothermal step-by-step condensation
(dehydration) reactions The method incorporates reversibility and is applicable for supplying long
range order in COF materials The reactants generally consist of boronic acids and hydroxy-
polyphenyl (catechol) compounds Extended porous networks are obtained when O-B or O-C bonds
are formed into 5 or 6-membered rings after dehydration (scheme 1) Dehydration may also be
carried out using microwave synthesis Alternately boronic acid can react with catechol acetonide in
presence of a Lewis acid catalyst The molecules link to each other after the liberation of acetone and
water (scheme 2) The utilization of phenyl-attached formyl and amino groups as building units
results in the formation of imines after dehydration (scheme 3) A desired arrangement of molecular
arrays on surfaces can be fulfilled by depositing planar molecules on metallic surfaces followed by
covalent linking using annealing
47
Isocyanate groups follow trimerization to form poly-isocyanurate rings (scheme 4) Cyclotrimerization
of monomers containing nitrile groups or cyanate groups is prone to generate C3N3 rings (scheme 5)
However the ionothermal synthesis of them resulted with amorphous materials Unique bond
formation is often not achieved throughout the material and thus the crystal lacks long-range order
Analogously coupling reactions such as Ullmann and Suzuki-Miyaura give rise to amorphous
products However they are adequate in producing C-C bonds when halogen-substituted compounds
are reacted (scheme 6) Nucleophilic substitution may also be used for framing networks in cases
like reactants are formed into nucleophiles and ions after release of their end-groups (scheme 7)
48
Figure 4 Different schemes reported for the synthesis of covalently-bound organic frameworks
49
25 List of synthesized materials
Table 1 List of synthesized materials in the literature their building blocks synthesis methods measured surface area [m
2 g
-1] pore volume [cm
3 g
-1] and pore size [Aring]
COF Names Reactants Synthesis Surface
Area
Pore
volume
Pore
size
COF-1 BDBA Solvothermal condensation235072
scheme 1
711 62850 032
03650
9
COF-5 BDBA HHTP Solvothermal condensation23
scheme 1
1590 0998 27
Microwave3673 scheme 1 2019
BDBA TPTA Lewis acid catalysis35 TPTA
COF-6 BTBA HHTP Solvothermal condensation27
scheme 1
980 (L) 032 64
COF-8 BTPA HHTP Solvothermal condensation27
scheme 1
1400 (L) 069 187
COF-10 BPDA HHTP Solvothermal condensation27
scheme 1
2080 (L) 144 341
BPDA TPTA Lewis acid catalysis35 scheme 2
COF-18Aring BTBA THB Facile dehydration3348 scheme 1 1263 069 18
COF-16Aring BTBA alkyl-THB
(alkyl = CH3)
Facile dehydration48 scheme 1 753 039 16
COF-14Aring BTBA alkyl-THB
(alkyl = C2H5)
Facile dehydration48 scheme 1 805 041 14
COF-11Aring BTBA alkyl-THB
(alkyl = C3H7)
Facile dehydration48 scheme 1 105 0052 11
50
SCOF-1 BDBA Substrate-assisted synthesis39
scheme 1
SCOF-2 BDBA HHTP Substrate-assisted synthesis39
scheme 1
TP COF PDBA HHTP Solvothermal condensation25
scheme 1
868 079 314
PPy-COF PDBA Solvothermal condensation26
scheme 1
923 188
TBB COF TBB (on Cu(111) and
Ag(110) surfaces)
Surface polymerisation40 scheme
6
TBPB COF TBB (on Au(111)
surface)
Surface polymerisation41 scheme
6
BTP COF BTPA THDMA Solvothermal condensation72
scheme 1
2000 163 40
HHTP-DPB COF DPB HHTP Solvothermal condensation73
scheme 1
930 47
PICU-A DMBPDC Cyclotrimerization74 scheme 4
PICU-B DCF Cyclotrimerization74 scheme 4
COF-LZU1 DAB TFB Solvothermal condensation53
scheme 3
410 054 12
PdCOF-LZU1 COF-LZU1 PdAc Facile post-treatment53 146 019 12
XN3-COF-5 X N3-BDBA (100-X)
BDBA HHTP
Solvothermal condensation51
scheme 1
2160
(X=5)
1865 (25)
1722 (50)
1641 (75)
1421
(100)
1184
1071
1016
0946
0835
295
276
259
258
227
51
XAcTrz-COF-5 XN3-COF-5 PAc Click reaction51 2000
(X=5)
1561 (25)
914 (50)
142 (75)
36 (100)
1481
0946
0638
0152
003
298
243
156
153
125
XBuTrz-COF-5 XN3-COF-5 HP Click reaction51
XPhTrz-COF-5 XN3-COF-5 PPP Click reaction51
XEsTrz-COF-5 XN3-COF-5 MP Click reaction51
XPyTrz-COF-5 XN3-COF-5 PyMP Click reaction51
COF-42 DETH TFB Solvothermal condensation75
scheme 3
710 031 23
COF-43 DETH TFPB Solvothermal condensation75
scheme 3
620 036 38
CTF-1 DCB Ionothermal trimerization56
scheme 5
791 040 12
CTF-2 DCN Ionothermal trimerization57
scheme 5
90 8
mp-CTF-2 2255 151 8
CTF (DCP) DCP Ionothermal trimerization64
scheme 5
1061 0934 14
K2[PtCl4]-CTF DCP K2[PtCl4] Trimerization (scheme 5) +
coordination64
Pt-CTF DCP Pt Trimerization (scheme 5) +
coordination64
PAF-5 TBB Yamamoto-type Ullmann cross-
coupling reaction69 scheme 6
1503 157 166
52
PAF-6 PA CA Nucleophilic substitution70
scheme 7
1827 118
Pc-PBBA COF PcTA BDBA Lewis acid catalysis34 scheme 2 506 (L) 0258 217
NiPc-COF OH-Pc-Ni BDBA Solvothermal condensation37
scheme 1
624 0485 190
XN3-NiPc-COF OH-Pc-Ni X N3-BDBA
(100-X) BDBA
Solvothermal condensation51
scheme 1
XEsTrz-NiPc-
COF
XN3-NiPc-COF MP Click reaction51
ZnP COF TDHB-ZnP THB Solvothermal condensation38
scheme 1
1742 1115 25
NiPc-PBBA COF NiPcTA BDBA Lewis acid catalysis35 scheme 2 776
2D-NiPc-BTDA
COF
OHPcNi BTDADA Solvothermal condensation47
scheme 1
877 22
ZnPc-Py COF OH-Pc-Zn PDBA Solvothermal condensation
scheme 1
420 31
ZnPc-DPB COF OH-Pc-Zn DPB Solvothermal condensation
scheme 1
485 31
ZnPc-NDI COF OH-Pc-Zn NDI Solvothermal condensation
scheme 1
490 31
ZnPc-PPE COF OH-Pc-Zn PPE Solvothermal condensation
scheme 1
440 34
COF-366 TAPP TA Solvothermal condensation46
scheme 3
735 032 12
COF-66 TBPP THAn Solvothermal condensation46
scheme 1
360 020 249
53
COF-102 TBPM Solvothermal condensation24
scheme 1
3472 135 115
Microwave36
scheme 1
2926
COF-102-C12 TBPM trunk-TBPM-R
(R=dodecyl)
Solvothermal condensation52
scheme 1
12
COF-102-allyl TBPM trunk-TBPM-R
(R=allyl)
Solvothermal condensation52
scheme 1
COF-103 TBPS Solvothermal condensation24
scheme 1
4210 166 125
COF-105 TBPM HHTP Solvothermal condensation24
scheme 1
COF-108 TBPM HHTP Solvothermal condensation24
scheme 1
COF-202 TBPM TBST Solvothermal condensation28
scheme 1
2690 109 11
COF-300 TAM TA Solvothermal condensaion29
scheme 3
1360 072 72
PAF-1 TkBPM-X (X=C) Yamamoto-type Ullmann cross-
coupling reaction65 scheme 6
5600
PAF-2 TCM cyclotrimerization59 scheme 5 891 054 106
PAF-3 TkBPM-X (X=Si) Yamamoto-type Ullmann cross-
coupling reaction76 scheme 6
2932 154 127
PAF-4 TkBPM-X (X=Ge) Yamamoto-type Ullmann cross-
coupling reaction76 scheme 6
2246 145 118
PAF-11 TkBPM-X (X=C) BPDA Suzuki coupling reaction68 704 0203 166
54
scheme 6
3 Studies of structure and properties of COFs
31 2D COFs
Following the literature of the years 2005-2010 2D COFs were believed to have high symmetric AA
(eclipsed ndash P6mmm symmetry) stacking as boron nitride (h-BN) [REF] An exception was COF-1
which was considered to have AB (staggered ndash P63mmc) stacking as graphite[REF] The AA stacking
maximizes the attractive London dispersion interaction between the layers an important
contribution to the stacking energy At the same time AA stacking always suffers repulsive Coulomb
force between the layers due to the polarized connectors as the distance between atoms exposing
the same charge is closest It should be noted that in boron nitride boron atoms serve as nearest
neighbors to nitrogen atoms in adjacent layers The Coulomb interaction has ruled out possible
interlayer eclipse between atoms of alike charges Based on these facts we modeled77 a set of 2D
COFs with different possible stacks Each layer was drifted with respect to the neighboring layers in
directions symmetric to the hexagonal pore and we calculated the total energies and their Coulombic
contributions The AA stacking version was found to have the highest Coulombic repulsion in each
COF The lattice types with the layers shifted to each other by ~14 Aring approximately the bond length
between two atoms in a 2D COF were found to be more stable compared to AA or AB In this low-
symmetry configuration the hexagonal rings in the pore walls undergo staggering with that of
adjacent layers Within an error limit the bulk COFs may be either serrated or inclined as shown in
Figure 5 The interlayer separations of all the COF stacks are in the range of 30 to 36 Aring and increase
in the order AB serratedinclined AA32 The motivation for proposing inclined stacking for COFs is
that the rhombohedral graphite (ABC stacking) is the inclined version of the layer-offset in natural
graphite (AB stacking) The instability of AA stacking was also suggested by the studies of elastic
properties of COF-1 and COF-5 by Zhou et al78 The shear modulus show negative signs for the
vertical alignment of COF layers while they are small but positive for the offset of layers
55
Figure 5 Different stacks of a 2D covalent organic framework COF-1 The atoms are shown as Van der Waals spheres
The different stacking modes should in principle be visible in their PXRD patterns as each space
group has a distinct set of symmetry imposed reflection conditions We simulated the XRD patterns
of COFs in their known and new configurations and on comparison with the experimental spectrum
the new as well as AA stacking showed good agreement3279 However the slight layer-offsets in
conjunction with the large number of atoms in the unit cells change the PXRD spectrum only by the
appearance of a few additional peaks all of them in the vicinity of existing signals and by changes in
relative intensities Unfortunately the low resolution of the experimental data does now allow
distinguishing between the different stackings as the broad signals cover all the peaks of the
simulated spectrum
A detailed analysis on the interlayer stacking of HHTP-DPB COF has been made by Dichtel et al73 is
very complementary73 This work uses molecular mechanics extensively to define potential energy
surfaces of stacking of two layers from eclipsed to staggered covering all possible layer offsets The
low energy region was near to the eclipsed stacking which was then zoomed in to examine using DFT
for higher accuracy The far-eclipsed regions encounter no energy barriers with the near-eclipsed
regions which suggest the unlikeness of forming near-staggered layering Within DFT the eclipsed
geometries are at high energy compared to the slight offsets around AA (~ 13 ndash 25 Aring away) For a
hexagonal COF the energetically preferred offset could be in one of the six hexagonal sectors (or
circle sectors) of central angle 60 The preference of falling into one sector over the others is
arbitrary and may be determined by symmetry set by nature or external parameters Hence the
layering order in bulk could be serrated inclined spiral or purely by random The latter case better
known as turbostratic disorder is then more like a rule for 2D COFs rather than an exception caused
by external forces This also explains why the experimental PXRD patters have relatively broad peaks
The layer-offset may not only change the internal pore structure but also affect interlayer exciton
and vertical charge transport in opto-electronic applications
56
Concerning the stacking stability the square 2D COFs are expected not to be different from
hexagonal COFs because the local environment causing the shifts is nearly the same DFTB based
calculations reported along with the synthesis of electron-transporting 2D-NiPc-BTDA COF supports
this notion47 Two shifted configurations ndash at 07 and 14 Aring away from AA ndash appeared to be
energetically preferred over the AA stacking It appears that AA stacking is only possible for boron
nitride-like structures were adjacent layers have atoms with opposite charges in vertical direction In
analogy to BN layers it might be possible to force AA stacking by the introduction of alkalis in
between the layers
32 3D COFs
3D COFs in general are composed of tetragonal and triangular building blocks So far that their
synthesis leads to two high symmetry topologies ndash ctn and bor - in high yields24 The two topologies
differ primarily in the twisting and bulging of their components at the molecular level The
thermodynamic preference of one topology over the other may result from the kinetic entropic and
solvent effects and the relative strain energies of the molecular components It is straight-forward to
state that the effects at the molecular level crucial crucial in the bulk state since transformation from
one net to the other is impossible without bond-breaking There has not been any detailed study on
this matter experimentally or theoretically
Schmid et al8182 have developed force-field parameters from first principles calculations using
Genetic Algorithm approach The parameters developed for cluster models of COF-102 can
reproduce the relative strain energies in sufficient accuracies and be extended to calculations on
periodic crystals The internal twists of aromatic rings within the tetragonal units are different for ctn
and bor nets which lead to stable S4 and unstable D2d symmetries respectively This together with
the bulging of triangular units in the bor net reasoned for energetic preference of ctn over bor for all
boron-based 3D COFs79
The connectivity-based atom contribution method (CBAC) developed by Zhong et al80 can
significantly reduce computational time needed for quantum chemical calculation for framework
charges when screening a large number of MOFs or COFs in terms of their adsorption properties The
basic assumption is that atom types with same bonding connectivity in different MOFsCOFs have
identical charges a statement that follows from the concept of reticular chemistry where the
properties of the molecular building blocks keep their properties in the bulk After assigning the
CBAC charges to the atom types slight adjustment (usually less than 3 ) is needed to make the
frameworks neutral Simulated adsorption isotherms of CO2 CO and N2 in MOFs and COFs show that
CBAC charges reproduce well the results of the QM charges8384 The CBAC method still requires a
57
well-parameterized force field in order to account correctly for adsorption isotherms as the second
important contribution to the host-guest interaction is the London dispersion energy between
framework and adsorbed moleculesG
The elastic constants calculated for the 3D COFs COF-102 and COF-108 show that COF-102 is nearly
five times stronger than COF-10878 While the bulk modulus (B) of COF-102 (B = 216 GPa) exceeds
that of known MOFs such as MOF-5 MOF-177 and MOF-205 (B = 153 101 107 GPa respectively81)
the least dense COF-108 is at the edge of structural collapse (B = 49 GPa) The calculations were
made using Density Functional Tight-Binding (DFTB) method Another report82 based on the same
level of theory possibly with a different parameter set however reveals lower bulk moduli for both
COFs 177 GPa for COF-102 and 005 GPa for COF-108 The bulk moduli for COF-103 and COF-105 are
110 and 33 GPa respectively Recently we have reported the structural properties of 3D COFs The
calculations were made using self-consistent charge (SCC) DFTB method The bulk modulus of each
COF is as follows COF-102 ndash 206 COF-103 ndash 139 COF-105 ndash 79 COF-108 ndash 37 COF-202 ndash 153 and
COF-300 ndash 144 GPa Force field calculations79 provide slightly different values COF-102 ndash 170 COF-
103 ndash 118 COF-105 ndash 57 COF-108 ndash 29 COF-202 ndash 110 GPa Albeit the differences in values each
type of calculation shows the trend that bulk modulus decreases with decreasing mas density and
increasing pore volume and distance between connection nodes One has to note that the high
mechanical stabilities of 3D COFs as suggested by the calculations correspond always to defect-free
crystals Theory is expected therefore to overestimate experimental mechanical stability for real
materials
COFs in general have semiconductor-like band gaps (~ 17 to 40 eV)32 Layer-offset from eclipsed
layering slightly increases the band gap However the Density-Of-States (DOS) of a monolayer is
similar to that of bulk COF The band gap is generally smaller for a COF with larger number conjugate
rings
The thermal influence on the crystal structure of 3D COFs is also intriguing as negative thermal
expansion (NTE) the contraction of crystal volume on heating has been reported for COF-10283 The
studies were performed using molecular dynamics with the force field parameters by Schmid et al84
However the thermal expansion coefficient α = -151 times 10-6 K-1 is much smaller compared to that of
some of the reported MOFs that also show NTE behavior85 The volume-contraction upon the
increase of temperature arises from the tilt of phenyl linkers between B3O3 rings and sp3 carbon
atom in TBPM (figure ) Later Schmid et alreported that COF-102 103 105 108 exhibit NTE
behavior both in ctn and bor topologies whereas COF-202 only in ctn topology79 A typical
application is the realization of controllable thermal expansion composites made of both negative
and positive thermal expansion materials
58
4 Applications
41 Gas storage
The success in the synthesis of COFs was certainly the result of a long-term struggle for complete
covalent crystallization The discovery of COFs coincided with the time when world-wide effort was
paid to develop new materials for gas storage in particular for the development hydrogen tanks for
mobile applications Materials made exclusively from light-weight atoms and with large surface
areas were obviously excellent candidates for these applications The gas storage capacity of porous
materials relies on the success of assembling gas molecules in minimum space This is achieved by
the interaction energy exerted by storage materials on the gas molecules Because the interactions
are noncovalent no significant activation is required for gas uptake and release and hence the so-
called physical adsorption or physisorption is reversible The other type of adsorption ndash chemical
adsorption or chemisorption ndash binds dissociated hydrogen covalently or interstitially at the risk of
losing reversibility As it requires the chemical modification of the host material chemisorption is not
a viable route for COFs and might become possible only through the introduction of reactive
components into the lattice The total amount of gas adsorbed in the pores gives rise to what is
referred to as lsquototal adsorption capacityrsquo This quantity is hard to be assessed experimentally as a
measurement is always subjected to influence of the materials surface and gas present in all parts of
the experimental setup Therefore the lsquoexcess adsorption capacityrsquo is reported in experiment Here
the gas stored in the free accessible volume is subtracted from the total adsorption In experiment
this volume includes the voids in the material as well as any empty space between the sample
crystals This volume is typically determined by measuring the uptake of inert gas molecules (He for
H2 adsorption N2 or Ar for adsorption studies of larger molecules) at room temperature with the
assumption that the host-guest interaction between the material and He can be neglected The
different definitions of adsorption is given in Figure 6
Typically experiments measure excess values and simulations provide total values Quantities of
adsorbed molecules are usually expressed as gravimetric and volumetric units The former gives the
amount per unit mass of the adsorbent while the latter gives the amount per unit volume of the
adsorbent Explicative definitions and terminologies related to gas adsorption can be found
elsewhere86
59
Figure 6 Hydrogen density profile in a pore A denotes the excess A+B the absolute and A+B+C the total adsorption The skeletal volume corresponds to the volume occupied by the framework atomsCourtesy of Dr Barbara Streppel and Dr Michael Hirscher MPI for intelligent systems Stuttgart Germany
411 Porosity of COFs
It is common to inspect the porosity of synthesized nanoporous materials using some inert or simple
gas adsorption measurements Distribution of pore size can be sketched from the
adsorptiondesorption isotherm N2 and Ar isotherms provide an approximation of the pore surface
area pore volume and pore size over the complete micro and mesopore size range Usually the
surface area is calculated using the Brunauer-Emmet-Teller (BET) equation or Langmuir equation
Total pore volume (volume of the pores in a pre-determined range of pore size) can be determined
from either the adsorption or desorption phase The Dubinin-Radushkevic method the T-plot
method or the Barrett-Joyner-Halenda (BJH) method may also be employed for calculating the pore
volume The pore size is often corroborated by fitting non-local density functional theory (NLDFT)
based cylindrical model to the isotherm90-93 In some cases the desorption isotherm is affected by
the pore network smaller pores with narrower channels remain filled when the pressure is lowered
This leads to hysteresis on the adsorption-desorption isotherms The different methods employed for
pore structure analysis are characteristic for micropore filling monolayer and multilayer formations
capillary condensation and capillary filling
For any adsorbate in order to form a layer on pore surface the temperature of the surface must
yield an energy (kBT) much less than adsorbate-surface interaction energy Additionally the absolute
value of the adsorbate-surface binding energy must be greater than the absolute value of the
adsorbate-adsorbate binding energy This avoids clustering of the adsorbate into its three-
dimensional phase
60
At high pressure the adsorption isotherm shows saturation which means that no more voids are left
for further occupation The isotherms show different behaviors characteristic of the pore structure of
the materials There are known classifications based on these differences type I II III IV and V For
COFs and the related materials discussed in this review type I II and IV have been observed (see
Figure 7) As more adsorbate is attracted to the surface after the first surface layer completion one
can expect a bend in the isotherm Type I implies monolayer formation which is typical of
microporosity If the surface sites have significantly different binding energies with the adsorbate
type II behavior is resulted In type IV each ldquokneerdquo where the isotherm bends over and the vapor
pressure corresponds to a different adsorption mechanisms (multiple layers different sites or pores)
and represents the formation of a new layer
Figure 7 Different types of adsorptions in Covalently-bound Organic Frameworks
Argon and nitrogen adsorption measurements at respectively 87 and 77 K exhibit type I isotherms
for COF-1 6 102 and 103 and type IV isotherms for COF-5 8 10 102 and 10387 Smaller pore
diameters of COF-1 and 6 (~ 9 Aring each) are characteristic of monolayer gas formation on the internal
pore surface The same reasons are responsible for the type I character of COF-102 and COF-103
(pore size = 12 Aring each) isotherms albeit with some unusual steps at low pressure The type IV
isotherms of COF-5 8 10 102 and 103 show formation of monolayers followed by their
multiplication within their relatively larger pores (sizes are approximately 27 16 32 12 and 12 Aring
respectively) Other COFs that exhibit type I isotherms for N2 adsorption are the 3D COFs COF-202 (11
Aring) COF-300 (72 Aring) and COF-102-C12 (12 Aring) the alkylated 2D COF series COF-18Aring COF-16Aring COF-14Aring
COF-11Aring (the nomenclature includes their pore sizes) and a 2D square COF NiPc COF (19 Aring)
Isotherms like Type-II were observed for PPy-COF (188 Aring) COF-366 (176 Aring) COF-66 (249 Aring) Pc-
PBBA COF (217 Aring) ZnP-COF (25 Aring) COF-LZU1 (12 Aring) PdCOF-LZU1 (12 Aring) and some of the azide-
appended COFs 50AcTrz-COF-5 (156 Aring) 75AcTrz-COF-5 (153 Aring) 100AcTrz-COF-5 (125 Aring)
50BuTrz-COF-5 (232 Aring) 50PhTrz-COF-5 (212 Aring) 50EsTrz-COF-5 (212 Aring) and 25PyTrz-COF-5
(221 Aring) The majority of the COFs with mesopores exhibit type-IV isotherms They are TP-COF (314
Aring) BTP-COF (40 Aring) COF-42 (23 Aring) COF-43 (38 Aring) HHTP-DPB COF (47 Aring) CTC-COF (226 Aring) NiPc-PBBA
COF ZnPc-Py-COF (31 Aring) ZnPc-DPB-COF (31 Aring) ZnPc-NDI-COF (31 Aring) ZnPc-PPE-COF (34 Aring) 5N3-
61
COF (295 Aring) 25N3-COF (276 Aring) 50N3-COF (259 Aring) 75N3-COF (258 Aring) 100N3-COF (227 Aring)
5AcTrz-COF-5 (298 Aring) 25AcTrz-COF-5 (243 Aring) 5PyTrz-COF-5 (289 Aring) and 10PyTrz-COF-5
(242 Aring)
The synthesized microporous amorphous materials CTF-1 (12 Aring) CTF-DCP (14 Aring) PAF-1 and PAF-2
(106 Aring) exhibit type-I isotherms with N2 adsorption PAF-3 (127 Aring) PAF-4(118 Aring) PAF-5 (157 Aring)
PAF-6 (118 Aring) and PAF-11 showed surprisingly not the typical type I behavior in spite of their
microporosity but type-II isotherms
Many of the mesoporous COFs showed small hysteresis in the adsorption-desorption isotherm
pointing the possibility of capillary condensation Hysteresis was observed for the amorphous
materials in both mirco and meso-pore range Various reasons have been attributed for the observed
hysteresis including the softness of the material and guest-host interactions
412 Gas adsorption experiments
Hydrogen uptake capacities of some boron-based COFs were measured by Yaghi et al87 The excess
gravimetric saturation capacities of COF-1 -5 -6 -8 -10 -102 and -103 at 77 K were found to be 148
358 226 350 392 724 and 705 mg g-1 respectively The saturation pressure increases with an
increase in pore size similar to MOFs88 Pore walls of 2D COFs consist only of edges of connectors
and linkers The fact that faces and edges are largely available for interactions with H2 in 3D
geometries is a reason for their enhanced capacity Total adsorption generally increases without
saturation upon pressure because the difference between the total and the excess capacities
corresponds to the situation in bulk (or in a tank) COFs show in particular good gravimetric
capacities because of their low mass density while volumetric capacities typically do not exceed
those of MOFs The gravimetric hydrogen storage capacities of COF-102 and -103 exceeds 10 wt at
a pressure of 100 bar
COF-18 and its alkylated versions COF-16 -14 and -1 have H2 excess gravimetric capacities 155 144
123 and 122 wt respectively at hellipK and hellipbar
Excess uptake of methane in COFs at room temperature and at 70 bar pressure is as follows COF-1
and -6 up to10 wt COF-5 -6 and -8 between 10 and 15 wt and COF-102 and -103 above 20
wt 87 Isosteric heat of adsorption Qst calculated by fitting the isotherms to virial expansion with
the same temperature are relatively weaker (lt 10 kJmol) for COF-102 and COF-103 at low
adsorption which implies weaker adsorbate-adsorbent interactions In comparison COF-1 and 6
exhibit twice the enthalpy however the overall hydrogen storage capacity was very limited due to
62
the smaller pore volume High Qst at zero-loading is not desirable because the primary excess amount
adsorbed at very low pressures cannot be desorbed practically89
COF-102 and 103 outperform 2D COFs for CO2 storage87 The saturation uptakes at room
temperature were 1200 and 1190 mg g-1 for COF-102 and 103 respectively
A type IV isotherm has been observed for ammonia adsorption in COF-10 inherent to its mesoporous
nature90 The material showed an uptake of 15 mol kg-1 at room temperature and 1 bar the highest
of any nanoporous materials The repeated adsorption-desorption cycles were reported to disrupt
the layer arrangement of this 2D COF Yaghi et al90 were also able to make a tablet of COF-10 crystal
which performed nearly up to the crystalline powder
Not many COFs have been experimentally studied for gas storage applications in spite of high
expectations This has to be understood together as a result of the powder-like polycrystallization of
COFs The enthalpy Qst at low-loading accounted to only 46 kJmol
The excess and total hydrogen uptake capacity of PAF-1 at 48 bar and 77 K reached 7 wt and 10
wt respectively65 PAF-3 and PAF-478_ENREF_73 follow the same diamond topology as PAF-1 the
difference accounts in the central atoms of the tetragonal blocks PAF-1 3 and 4 have C Si and Ge
atoms at the centers respectively At 1 bar the respective adsorption capacities were 166 207 and
150 wt The highest value being found for PAF-3 is attributed to its relatively high enthalpy (66 kJ
mol-1) compared to PAF-1 (54 kJ mol-1) and PAF- 4 (66 kJ mol-1) At high pressure the surface area is
a key factor and because of the lower BET surface area PAF-3 and PAF-4 performs poor At 60 bar
their respective adsorption capacities were 55 and 42 wt For PAF-11 hydrogen uptake was 103
wt at 1 bar68
Methane uptakes of PAF-1 3 and 4 at 1 bar are 13 19 and 13 wt respectively76 Their enthalpies
are respectively 14 15 and 232 kJ mol-1 This suggests that Ge moiety have strong interactions with
methane
CO2 adsorption in PAF-1 at 40 bar and room temperature was 1300 mg g-1 which was slightly more
than the capacity of COF-102 and 10365 PAF-1 3 and 4 at 1 atm pressure could store 48 80 and 51
wt respectively At 273 K and 1 atm they stored 91 153 and 107 wt respectively The storage
capacities at 273 K and 298 K lead to the following zero-loading isosteric enthalpies 156 192 162
kJ mol-1 for PAF-1 3 and 4 respectively The higher uptake of PAF-3 may also be explained by its
relatively higher surface area with CO2 molecules
The selective adsorption of greenhouse gases in PAFs was discussed by Ben et al76 At 273 K and 1
atm pressure PAF-1 selectively adsorbes CO2 and CH4 respectively 38 and 15 times more in
63
amounts than N2 PAF-3 showed extraordinary selectivity of 871 for CO2 over N2 and 301 for CH4
over N2 For PAF-4 the selectivity was 441 for CO2 over N2 and 151 for CH4 over N2 Generally the
uptake of N2 Ar H2 O2 are significantly lower compared to CO2 and NH4 which makes these PAFs
suitable for separating them
Harmful gases like benzene and toluene can be stored in PAF-1 in amounts 1306 mg g-1 (1674 mmol
g-1) and 1357 mg g-1 (1473 mmol g-1) respectively at saturated pressures and room temperature65
In PAF-11 their adsorptions were in amounts of 874 mg g-1 and 780 mg g-1 respectively68 PAF-2 was
accounted to a much lower benzene adsorption (138 mg g-1) at the same conditions59 Adsorption of
cyclohexane vapor in PAF-2 was only 7 mg g-1 While PAF-11 exhibited hydrophobicity it could
accommodate significant amount of methanol (654 mg g-1 or 204 mmol g-1) at room temperature
and saturated pressure68 PAF-5 has been tested for storing organic pollutants69 At room
temperature and saturated pressure PAF-5 adsorbed methanol benzene and toluene in amounts
6531 cm3 g-1 (292 mmol g-1) 26296 cm3 g-1 (117 mmol g-1) and 2582 cm3 g-1 (115 mmol g-1)
respectively Their adsorptions in PAF-5 were deviated from the typical type-I behavior Methanol
exhibited a large slope at the high pressure region corresponding to the relative size of it Zhu G et
al dedicated PAF-6 for biomedical applications With no toxicity observed for PAF-6 over a range of
concentrations adsorption of ibuprofen was measured in it PAF-6 showed an uptake of 350 mg g-1
The release of ibuprofen was also tested in simulated body fluid which showed a release rate of 50
in 5 hours 75 in 10 hours and 100 in almost 46 hours
413 Theoretical predictions
Grand canonical Monte Carlo (GCMC) is a widely used method for the simulation of gas uptakes in
nanoporous materials In GCMC the number of adsorbates and their motions are allowed to change
at constant volume temperature and chemical potential to equilibrate the adsorbate phase The
motions are random guided by Monte Carlo methods and the energies are calculated by force field
methods The details of it may be found in the literature91
Goddart III WA et al92 simulated the uptake capacities of boron-based COFs using force fields derived
from the interactions of H2 and COFs calculated at MP2 level of theory The saturated excess uptakes
of both COF-105 (at 80 bar) and 108 (at 100 bar) were accounted to 10 wt at 77 K which are more
than the uptakes measured in ultrahigh porous MOF-200 205 and 21093 The predictions for other
COFs include 38 wt for COF-1 (AB stacking) 34 wt for COF-5 (AA stacking) 88 wt for COF-102
and 91 wt for COF-103 At 01 bar COF-1 in AB stacking showed highest excess uptake (17 wt )
compared to other COFs in the present discussion Total uptake capacities of the COFs are in the
following order COF-108 (189 wt ) COF-105 (183 wt ) COF-103 (113 wt ) COF-102 (106
64
wt ) COF-5 (55 wt ) and COF-1 (38 wt ) It is worth noticing the higher gravimetric capacities of
COF-108 and 105 which were not measured experimentally They benefit from their lower mass and
higher pore volume In case of volumetric capacity COF-102 (404 gL excess) surpasses the others at
high pressures The total volumetric capacities of the COFs are 499 498 399 395 361 and 338
gL for COF-102 103 108 105 1 and 5 respectively Their additional calculations predicted benzene
rings as favorite locations for H2 molecules
Similar values and trends have been reported by Garberoglio G94 who also reasoned poor solid-fluid
interactions in a large part of free volume for the low volumetric capacity of COF-108 and 105 A
room temperature excess gravimetric uptake of approximately 08 wt was obtained for COF-108
and 105 Quantum diffraction effects were added to the interaction energies of H2 with itself and the
material which were calculated using universal force-field (UFF) With possible overestimation
Klontzas et al30 and Bassem et al95 have reported total gravimetric uptake of 45 and 417 wt
respectively at 300 K for the same COFs Among the boron-based 3D COFs COF-202 exhibited much
smaller uptake capacity at high pressures due to its smaller pore volume95 Lan et al96 predicted a
very small H2 uptake of 152 wt COF-202 at 300K and 100 bar Studies of Yang et al97 clarifies that
pore filling with H2 in 2D COFs is a gradual process rather than being in steps of layer formation
Garberoglio and Vallauri98 pointed out that the relatively higher mass density and lower surface area
per unit volume of 2D COFs are the reasons for their lower uptakes in comparison with 3D COFs The
surface area of hexagonal 2D COFs (AA stacking) averaged around 1000 m2 cm-3 while that of 3D
COFs were about 1500 m2 cm-3
Simulations of H2 adsorption in the perfect crystalline form of PAF-1 and PAF-11 also known as PAF-
302 and PAF-304 respectively are carried out by Zhu et al99 At 77 K H2 excess uptake of PAF-302 (7
wt ) is slightly higher than COF-102 (684 wt )87 and slightly lower than MOF-177 (740 wt )100 At
room temperature PAF-302 exhibited a poor uptake (221 wt at 100 bar) PAF-304 showed
excellent total uptakes 2238 wt at 77 K and 100 bar 653 wt at 298 K and 100 bar These are
highest among all nanoporous materials
Babarao and Jiang studied room-temperature CO2 adsorption in COFs101 At low pressures COFs with
smaller pore volume exhibit a steep rise in their isotherm while at high pressures COF-105 and 108
(82 and 96 mmol g-1 respectively at 30 bar) dominate the others COF-6 has the highest isosteric heat
of adsoption (328 kJmol) hence it made the first steep rise in the volumetric isotherm followed by
COF-8 102 103 10 105 and 108 Correlations of gravimetric and volumetric capacities with mass
density pore volume porosity and surface area have been excellently manifested in this article101
With increasing framework-density gravimetric uptake falls inversely while volumetric capacity
decreases linearly The former rises with free volume while the latter rises and then drops slightly
65
Both of them rise with porosity and surface area Choi et al102 predicted 45 times more CO2 uptake in
COF-108 (3D) compared to COF-5 (2D) Recently Schmid et al79 also have predicted CO2 adsorption
especially for 3D COFs The uptake of COF-202 was almost half of COF-102 and 103 at room
temperature Type IV isotherm has been found for CO2 adsorption in COF-8 and 10 at low
temperatures (lt 200 K)97 Yang and Zhong97 excluded capillary condensation and dissimilar
adsorption sites but multilayer formation as reasons of the stepped behavior (Yang and Zhong
explained this as a consequence of multilayer formation rather than a result of capillary
condensation or dissimilar adsorption sites)
Methane adsorption in 2D and 3D COFs was first calculated by Garberoglio et al Among COF-6 8 and
10 the former which has smaller pore size and higher binding energy with CH4 shows better
volumetric adsorption (~ 120 cm3cm3 at 20 bar) than the others at room temperature at low
pressure region (lt 25 bar) Larger pore size favors COF-10 for higher volumetric adsorption (~ 160
cm3cm3 at 70 bar) at high pressures COF-8 with intermediate pore size shows largest excess amount
of methane adsorbed upon saturation however still lower than two 3D COFs COF-102 and 103
show excess volumetric adsorption of about 180 cm3cm3 at 35 bar This is significantly higher than
the uptake of two other 3D COFs COF-105 and 108 Entropic effects which primarily arise from the
change of dimensionality when the bulk phase of adsorbate transforms to adsorbed phase are
crucial in adsorption Garberoglio94 shows that solid-fluid interaction potential in large part of volume
of COF-105 and 108 is not strong enough to overcome the entropic loss On the other hand these
two COFs exhibit relatively higher gravimetric uptake (720 and 670 cm3g respectively) Goddard III et
al89 predicted total methane uptakes in the following amounts 415 wt in COF108 405 wt in
COF-105 31 wt in COF-103 284 wt in COF-102 196 wt in COF-10(AA) 169 wt in COF-
5(AA) 159 wt in COF-8(AA) 123 wt in COF-6(AA) and 109 wt in COF-1(AB) Yang and Zhong97
have shown that even at low temperatures (100 K) the pore filling of COF-8 with CH4 is rather
gradual than by step-by-step whereas for COF-10 multilayer formation provides a stepped behavior
in the isotherm Alternately Goddard et al pointed out that layer formation coexists with pore-filling
at room temperature89
414 Adsorption sites
First principle calculations on cluster models are typically employed to find favorite adsorption sites
and binding energies of adsorbates within frameworks Benzene rings are found to be a usual
location for H2 where it prefers to orient in perpendicular fashion3086110 Other favorite locations
include B3O3 and C2O2B rings110111 where H2 orients horizontally on the face as well as on the
edge3086 The central ring of HHTP does not provide enough binding to H2 atoms because of small
amount of charges There are some differences in the adsorption energies and favorite sites
66
calculated at different levels of theory Overall the reported binding energies between H2 and any
COF fragment are less than 6 kJ mol-1 In comparison to MOFs COFs do not offer any stronger binding
energy with H2 The advantage of COFs is their low mass density Lan et al103 reported that CO2 is
more preferred to be adsorbed on B3O3 or C2O2B rings than benzene rings Choi et al102 reported that
the oxygen sites are the primary locations for CO2 and CO2 bound COFs are the second strong binding
sites
415 Hydrogen storage by spillover
Hydrogen spillover is an alternate way to store hydrogen113114 It involves dissociation of hydrogen
gas by supported metal catalysts subsequent migration of atomic hydrogen through the support
material and final adsorption in a sorbent Here surface-diffusion of H atoms over the support is
known as primary spillover Secondary spillover is the further diffusion to the sorbent Bridging the
metal part with the sorbent is a practice to enhance the uptake It increases the contact between the
source and receptor and reduces the energy barriers especially in the secondary spillover As the
final sorption is chemisorption surface area of the sorbent is more important than pore volume
Yang and Li measured H2 storage in COF-1 by spillover They used Pt supported on active carbon
(PtAC) as source for hydrogen dissociation Active carbon was the primary receptor and COF-1 the
secondary receptor COF-1 showed much low uptake 128 wt at 77 K and 1 atm 026 wt at 298
K and 100 bar In comparison to MOFs these are very low104 However they have found that the
uptake could be scaled up by a factor of 26 by bridging COF-1 with PtAC by applying carbonization
They also report that heat of adsorption between H and surface sites is more important than surface
area and pore volume in enhancing the net adsorption by spillover
Ganz et al theoretically predicted that the uptake capacities in COFs by hydrogen spillover can be
higher than the measured value116117 Based on ab initio quantum chemistry calculations and
counting the active storage sites they predicted 45 wt for COF-1 in AB stacking and 55 wt for
COF-5 in AA stacking at room temperature and 100 bar
42 Diffusion and Selectivity
Computed self-diffusion coefficients for H2 and CH4 adsorbates in 2D COFs (in AA stacking) are up to
one order of magnitude higher than that in 3D MOFs such as MOF-598 In comparison to nanotubes
the diffusion of H2 molecules in 2D COFs is one order of magnitude slower The differences in
diffusion coefficients are attributed to different pore structures Interactions of the corners of the
hexagonal pore with the fluid present potential barriers along the cylindrical pore axis Diffusion
occurs with jumps after long waiting The height of the barrier is still smaller than in MOFs
67
Homogeneous pore walls and absence of pore corners in nanotubes contribute much less
corrugation to the solid-fluid potential energy surface Increase of self-diffusivities of H2 and CH4 with
increase in pore size of 2D COFs is also shown by Keskin[Ref] Due to the smaller size of H2 its
diffusivity is higher than that of CH4 in any material For reasons of their relative size In a mixture of
the two the self-diffusivity of CH4 increases while that of H2 decreases
Molecular dynamics simulation studies of the diffusion of gas molecules within a 2D COF performed
by Krishna et al105 revealed its relatively lower binding energy of hydrogen molecules within the pore
walls compared to argon CO2 benzene ethane propane n-butane n-pentane and n-hexane
Binding energy prevents the molecules from diffusing through the pore channels They tested if
Knudsen diffusion is applicable to COFs Knudsen diffusion occurs when the molecules frequently
collide with the pore wall This generally happens when the mean free path is larger than the pore
diameter The simulations had been carried out in BTP-COF which accounts a pore diameter of 4 nm
It is found out that the ratio of zero-loading diffusivity with the Knudsen diffusivity has a significant
correlation with isosteric heat of adsorption the stronger the binding energy of the molecules with
the walls the lower the ratio Hydrogen being an exception among the investigated molecules
exhibited the ratio close to unity while others with their isosteric heat of adsorption higher than 10
kJmol-1 have a value close to 01 For a mixture of two gases with significant differences in binding
energies the ratio of self-diffusivities affirms high diffusion selectivity
Selectivity of CH4 over H2 in COF-5 6 and 10 is studied by Keskin106 Narrow pores support the
selective adsorption of CH4 over H2 however they suffer from entropic effects at high pressures
which favor H2 adsorption Liu et al107 have reported that the adsorption selectivity of COFs and
MOFs are in general similar up to 20 bar Delta loading or working capacity (usually expressed in
molkg) is an important term often used to show the economics of the selective adsorption This is
defined as the difference in loadings of the preferred gas at adsorption and desorption pressures
Adsorption is usually in the range 1 to 100 bar while desorption in 01 to 1 bar High selectivity and
high working capacity are preferential for practical use COF-6 has higher selectivity among the three
studied COFs but lower working capacity106 Best selectivity-working capacity combination is shown
by COF-5106 Nonetheless it is not better than CuBTC or CPO-27-X (X = Zn Mg)121122 Liu et al107
attributed the high isosteric heat of adsorption of COF-6 and Cu-BTC for their high adsorption
selectivity They also pointed out that the electrostatic contribution of framework charges in COFs
are smaller than in MOFs however needs to be taken into account
While CH4 is strongly adsorbed H2 undergoes faster diffusion Hence the product of adsorption
selectivity (gt1) and diffusion selectivity (gt0 lt1) which is membrane selectivity is smaller than
adsorption selectivity Keskin106 has compared the selectivity of COF-membranes with known
68
membranes The selectivity of COF-10 is similar to MOF-177 and IRMOFs COF-5 and 6 outperform
them It is to be noted that COF-5 is CH4 selective while COF-10 is H2 selective although their
topologies are similar CH4 permeability of COF membranes is much higher than several MOFs and
ZIFs COF-5 and 6 exhibited high membrane selectivity together with high membrane permeability
Some discussions on the adsorption and membrane based selectivity of COF-102 in comparison with
IRMOFs and Cu-BTC for CH4CO2H2 mixtures can be found in ref108 Adsorption selectivity of COF-6
and Cu-BTC in CH4H2 and CH4CO2 mixtures surpass other COFs and MOFs107 sdfalf
43 Suggestions for improvement
The level of achievement made by COFs and related materials yet do not practically meet the
practical requirements of several applications Hence thoughts for improvement primarily focused
on overcoming their limitations and enhancing characteristic parameters Some theoretical
suggestions for improved performances are mainly discussed here
431 Geometric modifications
Functionalities may be improved by utilizing the structural divergence of framework materials
Several examples may be found in the literature for MOFs etc Klontzas et al suggested replacement
of phenyl linkers in COF-102 by multiple aromatic moieties while keeping the topology unchanged to
increase surface area and pore volume109 They used biphenyl triphenyl naphthalene and pyrene
linkers in the new COFs All of them showed enhanced gravimetric capacity compared to the parent
COF One of the modified COF-102 with triphenyl linker showed 267 and 65 wt at 77 K and 300 K
respectively at 100 bar This kind of replacement may affect the adsorption of each adsorbate
differently leading to the alteration of the selective adsorption of one component over the other110
Following the reticular construction scheme we modeled some new 2D COFs by cross-linking some
of the familiar and unfamiliar building blocks in the COF literature32 This showed the structural
divergence of COFs however they exhibited structural and electronic properties analogues to other
2D COFs
Similar modifications in 2D and 3D COFs for high CO2 uptake were made by Choi et al102 DPABA
(diphenylacetylene-44ʹ-diboronic acid) and its tetrahedral version TBPEPM (tetra[4-(4-
dihydroxyborylphenyl)ethynyl]phenyl) in place of BDBA in COF-5 and TBPMTBPS in COF-108COF-
105 respectively were used for new COF designs The new 2D COF promised a saturated CO2 uptake
higher than in COF-102 because of its large pore volume The new 3D COFs were worth for an uptake
twice more than in COF-105 and 108
69
Goddard et al also designed about 14 new COFs by substituting the aromatic rings in the tetragonal
part (TBPM or TBPS) of COF-102 COF-103 COF-105 COF-108 and COF-202 with vinyl groups or alkyl-
functionalized extended or fused aromatic rings111 The new designs adopted their parent topology
and their structural stability was confirmed by analyzing molecular dynamics (MD) simulations at
room temperature Fused aromatic rings and vinyl groups provided relatively lower and the highest
zero-loading isosteric heat of adsorption (Qst) respectively however the room-temperature delivery
amounts of methane in the respective COFs crossed the DOE target at 35 bar111 In the latter
methane-methane interaction compensated Qst on high-loading
Lino M A et al112 theoretically inspected the ability of layered COFs to exhibit structural variants of
layered materials The hexagonal 2D COF layers may be rolled into the form of nanotubes (NT) or
may exist in the form of fullerenes with the inclusion of pentagons One could think of a formula unit
which resembles the carbon atoms in carbon nanotubes (CNT) or fullerenes The NTs in general can
have any chirality although the study included only armchair and zigzag NTs Density Functional
Theory based calculations reveal that COF-1 NTs of diameter greater than 15 Aring is energetically
favored This was concluded from the comparison of strain-energy per formula unit of the COF-1 NTs
with that of small diameter CNTs The strain energies of zigzag and armchair nanotubes show similar
quadratic functions of tube-diameter The Youngrsquos modulus of COF-1 (22) NT is found out to be 120
GPa This is much smaller compared to typical CNTs which have Youngrsquos modulus average around
1100 GPa113 Band gaps of COF layers remain unchanged when they roll up into nanotubes A COF-1-
fullerene built by scaling C60 molecule has large diameter and hence much less strain energy
compared to C60 Babarao and Jiang reported that the predicted CO2 adsorption capacity of COF-1 NT
is similar to that of CNTs101
Balance between mass density and surface area and hence high gravimetric and volumetric
capacities can be achieved by proper utilization of pore volume or proper distribution of pores Choi
et al theoretically predicted high gravimetric and volumetric capacities for H2 in a pillared COF114 A
pyridine molecule vertically placed above a boron atom in the boroxine ring of COF-1 layer is found
energetically favorable however with wrinkling of the layer115 Here nitrogen atom in pyridine forms
a covalent bond with the boron atom This pillaring increases the separation between the layers
exposes the buried faces of boroxine and aromatic rings to pores and hence increases surface area
and free volume Accessible surface area and free volume have been tripled and gravimetric and
volumetric capacities were calculated to be 10 wt and 60 g L-1 respectively at 77 K and 100 bar114
This volumetric capacity is higher than in NU-100116 and MOF-21093 which show ultrahigh surface
area
70
432 Metal doping
Miao Wu et al studied doping of COF-10 with Li and Ca atoms for hydrogen storage117 The metal
dopants transferred charges to substrate which in turn provided large polarization to hydrogen
molecules Similar observation has been made by Lan et al96 for Li atoms in COF-202 However they
showed the tendency to aggregate at high concentration Lan et al extensively studied doping of
positively charged alkali (Li Na and K) alkaline-earth (Be Mg Ca) and transition (Sc and Ti) metals in
COF-102 and 105 for enhanced CO2 capture103 They found that benzene rings and the three outer
rings of HHTP are the favorite sites for all the metal dopants Other interested locations are edges of
benzene rings and oxygen atoms B3O3 C2O2B rings and the central ring of HHTP were the unwanted
areas Lithium showed stability on the favorite locations while sodium and potassium tended to
cluster even at low concentrations Alkaline-earth metals only weakly interacted with the COFs
whereas transition metals chemisorbed in them which would possibly damage the substrate Lithium
is found out to be a good dopant for enhanced gas storage
Doping electropositive metals would be of advantage because they provide stronger binding with H2
(~ 4 kcalmol)103125133134 The effect of counterions in COFs is studied by Choi et al118 They found out
that although the binding energy of LiF is smaller than Li ion F counterion can hold one hydrogen
atom Additionally Li+ and Mg2+ ions preferred to be isolated than aggregated A step further
Goddard et al119 have recently shown that in presence of tetrahydrofuran (THF a solvent) an
electron is transferred from Li to the benzene ring whereas in the case of gas phase the electron
remained in the atom Additionally they suggested ways to remove solvents which are weakly
coordinated to the material Zhu et al110 suggested doping Li+ after replacement of hydrogen atom by
oxygen ion in COF-102 Another suggestion was the replacement of hydrogen atom by O--Li+ group
Both the cases exhibited higher CO2 adsorption than in the case of Li doping below 10 bar At 10 bar
the three cases exhibited almost the same uptake capacity Below 1 bar the doped Li+ provided
stronger interaction with quadrupolar CO2 in comparison with the substituted O--Li+ group The
differences at low pressures are attributed to the differences in the magnitude of the charge of Li
The doped Li+ remained to hold nearly +1 charge whereas the inclusion of O--Li+ to the framework
diminished the charge of Li+ to a moderate amount Doping Li atom added only relatively small
amount of charge to Li
Doping with Li+ could increase the H2 binding energies up to 28 kJmol118 With properly distributed
metal ion dopants the H2 uptake capacity in COF-108 is predicted to be 65 wt Cao et al also
predicted 684 and 673 wt of H2 uptake in Li-doped COF-105 and 108 respectively at room
temperature and 100 bar120 In Li-doped COF-202 the predicted uptake was 438 wt for the same
conditions96 Goddard et al119 observed that high performance of Li doped COFsMOFs at low
71
pressures (1 ndash 10 bar) is a disadvantage when calculating delivery uptake capacity Na doping could
overcome this difficulty as its heat of adsorption is lower than in Li doped materials Their predicted
delivery gravimetric capacities from 1 to 100 bar at room temperature for Li and Na doped COF-102
and 103 were higher than the 2010 DOE target of 45 wt at room temperature
Lan et al described that CO2 is polarized in presence of Li cation and the polarization increases when
Li cation is physisorbed in COF103 Their calculated uptake capacities of lithium-doped COF-102 and
COF-105 at low loading is about eight and four times higher than that of nondoped ones respectively
Also a very high uptake of 2266 mgg was predicted in the doped COF-105 at 40 bar Li-doped COF-
102 and COF-103 also showed higher uptakes of methane ndash almost double the amount ndash compared
to nondoped COFs at room temperature and low pressures (lt 50 bar)121 Binding energy between
doped Li cation and CH4 was calculated to be 571 kcalmol
Li-functionalized COF-105 realized by inserting alkoxide groups provided very high volumetric uptake
of H2 Klontzas et al122 replaced three hydrogen atoms in HHTP by oxygen atoms in order to achieve
the functionalization In spite of the increased weight due to the additional oxygen atoms the COF
exhibited gravimetric capacity of 6 wt at 300 K
Incorporation of lithium tetrazolide group into a new PAF having diamond topology and triphenyl
linkers for enhanced hydrogen adsorption is suggested by Sun et al123 The tetrazolide anion (CHN4-)
interacts with the lithium cation via ionic bonds The anion and the cation together can bind upto 14
hydrogen molecules Two lithium tetrazolide groups were introduced in the middle phenyl ring of
each triphenyl linker and GCMC calculations predicted 45 wt for the network at 233 K and 100 bar
With the intention of increasing the H2 binding energy Li et al chemically implanted metals in the
place of light atoms in COF-108124 They replaced the C2O2B rings in COF-108 by C2O2Al C2N2Al and
C2Mg2N and each new COF showed stability within DFT This kind of doping does not allow
aggregation of metal clusters Their GCMC simulations showed that the replacement strategy could
improve the hydrogen storage capacity at room temperature by a factor of up to three124 Zou et al
suggested that para-substitution of two carbon atoms in benzene rings by two boron atoms can
facilitate charge transfer between the rings and metal dopants125 Their work revealed that
dispersion of metals forbid any clustering and the doping could enhance the H2 uptake capacity
significantly
433 Functionalization
Functionalization of porous frameworks with ether groups for selective CO2 capture is suggested by
Babarao et al126 The electropositive C atom of CO2 interacts mainly with the electronegative O or N
72
atom of the functional groups They studied PAF-1 functionalized with ndashOCH3 ndashNH2 and ndashCH2OCH2ndash
groups in its biphenyl linkers The latter one which is the tetrahydrofuran-like ether-functionalized
PAF-1 showed high adsorption capacity for CO2 and high selectivity for CO2CH4 CO2N2 and CO2H2
mixtures at ambient conditions
5 Conclusions
Successful covalent crystallization resulted as a class of materials Covalent Organic Frameworks This
review portrays different synthetic schemes successful realizations and potential applications of
COFs and related materials The growth in this area is relatively slow and thus promotions are
needed in order to achieve its potential
6 List and pictures of chemical compounds
alkyl-THB Alkyl-1245-tetrahydroxybenzene
BDBA 14-benzenediboronic acid
BPDA 44ʹ-biphenyldiboronic acid
BTBA 135-benzene triboronic acid
BTDADA 14-benzothiadiazole diboronic acid
BTPA 135-benzenetris(4-phenylboronic acid)
CA Cyanuric acid
DAB 14-diaminobenzene
DCB 14-dicyanobenzene
DCF 27-diisocyanate fluorine
DCN 26-dicyanonaphthalene
DCP 26-dicyanopyridine
DETH 25-diethoxyterephthalohydrazole
DMBPDC 33ʹ-dimethoxy-44ʹ-biphenylene diisocyanate
DPB Diphenyl butadyenediboronic acid
73
HP 1-hexyne propiolate
HHTP 23671011-hexahydroxytriphenylene
MP Methyl propiolate
N3-BDBA Azide-appended benzenediboronic acid
NDI Naphthalenediimide diboronic acid
NiPcTA Nickel-phthalocyanice tetrakis(acetonide)
OH-Pc-Ni 2391016172324-octahydroxyphthalocyaninato)nickel(II)
OH-Pc-Zn 2391016172324-octahydroxyphthalocyaninato)zinc
PA Piperazine
Pac 2-propenyl acetate
PcTA Phthalocyanine tetra(acetonide)
PdAc Palladium acetate
PDBA Pyrenediboronic acid
PPE Phenylbis(phenylethynyl) diboronic acid
PPP 3-phenyl-1-propyne propiolate
PyMP (3α13α2-dihydropyren-1-yl)methyl propionate
TA Terephthaldehyde
TAM tetra-(4-anilyl)methane
TAPP Tetra(p-amino-phneyl)porphyrin
TBB 135-tris(4-bromophenyl)benzene
TBPM tetra(4-dihydroxyboryl-phenyl)methane
TBPP Tetra(p-boronic acid-phenyl)porphyrin
TBPS tetra(4-dihydroxyboryl-phenyl)silane
TBST tert-butylsilane triol
74
TCM Tetrakis(4-cyanophenyl)methane
TDHB-ZnP Zinc(II) 5101520-tetrakis(4-(dihydroxyboryl)phenyl)porphyrin
TFB 135-triformylbenzene
TFPB 135-tris-(4-formyl-phenyl)-benzene
THAn 2345-Tetrahydroxy anthracene
THB 1245-tetrahydroxybenzene
THDMA Polyol 2367-tetrahydroxy-910-dimethyl-anthracene
TkBPM Tetrakis(4-bromophenyl)methane
TPTA Triphenylene tris(acetonide)
trunc-TBPM-R R-functionalized truncated TBPM
75
Figure 8 Reactants of Covalently-bound Organic Frameworks
76
Figure 9 (Figure 8 continued)
(1) Morris R E Wheatley P S Angewandte Chemie-International Edition 2008 47 4966 (2) Li J-R Kuppler R J Zhou H-C Chemical Society Reviews 2009 38 1477 (3) Corma A Chemical Reviews 1997 97 2373 (4) Seo J S Whang D Lee H Jun S I Oh J Jeon Y J Kim K Nature 2000 404 982 (5) Yaghi O M OKeeffe M Ockwig N W Chae H K Eddaoudi M Kim J Nature 2003 423 705
77
(6) OKeeffe M Peskov M A Ramsden S J Yaghi O M Accounts of Chemical Research 2008 41 1782 (7) Ockwig N W Delgado-Friedrichs O OKeeffe M Yaghi O M Accounts of Chemical Research 2005 38 176 (8) Li H Eddaoudi M OKeeffe M Yaghi O M Nature 1999 402 276 (9) Chen B L Eddaoudi M Hyde S T OKeeffe M Yaghi O M Science 2001 291 1021 (10) Eddaoudi M Moler D B Li H L Chen B L Reineke T M OKeeffe M Yaghi O M Accounts of Chemical Research 2001 34 319 (11) Eddaoudi M Kim J Rosi N Vodak D Wachter J OKeeffe M Yaghi O M Science 2002 295 469 (12) Chae H K Siberio-Perez D Y Kim J Go Y Eddaoudi M Matzger A J OKeeffe M Yaghi O M Nature 2004 427 523 (13) Furukawa H Kim J Ockwig N W OKeeffe M Yaghi O M Journal of the American Chemical Society 2008 130 11650 (14) Smaldone R A Forgan R S Furukawa H Gassensmith J J Slawin A M Z Yaghi O M Stoddart J F Angewandte Chemie-International Edition 2010 49 8630 (15) Eddaoudi M Kim J Wachter J B Chae H K OKeeffe M Yaghi O M Journal of the American Chemical Society 2001 123 4368 (16) Sudik A C Millward A R Ockwig N W Cote A P Kim J Yaghi O M Journal of the American Chemical Society 2005 127 7110 (17) Sudik A C Cote A P Wong-Foy A G OKeeffe M Yaghi O M Angewandte Chemie-International Edition 2006 45 2528 (18) Tranchemontagne D J L Ni Z OKeeffe M Yaghi O M Angewandte Chemie-International Edition 2008 47 5136 (19) Lu Z Knobler C B Furukawa H Wang B Liu G Yaghi O M Journal of the American Chemical Society 2009 131 12532 (20) Park K S Ni Z Cote A P Choi J Y Huang R Uribe-Romo F J Chae H K OKeeffe M Yaghi O M Proceedings of the National Academy of Sciences of the United States of America 2006 103 10186 (21) Hayashi H Cote A P Furukawa H OKeeffe M Yaghi O M Nature Materials 2007 6 501 (22) Banerjee R Furukawa H Britt D Knobler C OKeeffe M Yaghi O M Journal of the American Chemical Society 2009 131 3875 (23) Cote A P Benin A I Ockwig N W OKeeffe M Matzger A J Yaghi O M Science 2005 310 1166 (24) El-Kaderi H M Hunt J R Mendoza-Cortes J L Cote A P Taylor R E OKeeffe M Yaghi O M Science 2007 316 268 (25) Wan S Guo J Kim J Ihee H Jiang D Angewandte Chemie-International Edition 2008 47 8826 (26) Wan S Guo J Kim J Ihee H Jiang D L Angewandte Chemie-International Edition 2009 48 5439 (27) Cote A P El-Kaderi H M Furukawa H Hunt J R Yaghi O M Journal of the American Chemical Society 2007 129 12914 (28) Hunt J R Doonan C J LeVangie J D Cote A P Yaghi O M Journal of the American Chemical Society 2008 130 11872 (29) Uribe-Romo F J Hunt J R Furukawa H Klock C OKeeffe M Yaghi O M Journal of the American Chemical Society 2009 131 4570 (30) Klontzas E Tylianakis E Froudakis G E Journal of Physical Chemistry C 2008 112 9095 (31) Tylianakis E Klontzas E Froudakis G E Nanotechnology 2009 20 (32) Lukose B Kuc A Frenzel J Heine T Beilstein Journal of Nanotechnology 2010 1 60
78
(33) Tilford R W Gemmill W R zur Loye H C Lavigne J J Chemistry of Materials 2006 18 5296 (34) Spitler E L Dichtel W R Nature Chemistry 2010 2 672 (35) Spitler E L Giovino M R White S L Dichtel W R Chemical Science 2011 2 1588 (36) Campbell N L Clowes R Ritchie L K Cooper A I Chemistry of Materials 2009 21 204 (37) Ding X Guo J Feng X Honsho Y Guo J Seki S Maitarad P Saeki A Nagase S Jiang D Angewandte Chemie-International Edition 2011 50 1289 (38) Feng X A Chen L Dong Y P Jiang D L Chemical Communications 2011 47 1979 (39) Zwaneveld N A A Pawlak R Abel M Catalin D Gigmes D Bertin D Porte L Journal of the American Chemical Society 2008 130 6678 (40) Gutzler R Walch H Eder G Kloft S Heckl W M Lackinger M Chemical Communications 2009 4456 (41) Blunt M O Russell J C Champness N R Beton P H Chemical Communications 2010 46 7157 (42) Sassi M Oison V Debierre J-M Humbel S Chemphyschem 2009 10 2480 (43) Ourdjini O Pawlak R Abel M Clair S Chen L Bergeon N Sassi M Oison V Debierre J-M Coratger R Porte L Physical Review B 2011 84 (44) Colson J W Woll A R Mukherjee A Levendorf M P Spitler E L Shields V B Spencer M G Park J Dichtel W R Science 2011 332 228 (45) Berlanga I Ruiz-Gonzalez M L Gonzalez-Calbet J M Fierro J L G Mas-Balleste R Zamora F Small 2011 7 1207 (46) Wan S Gandara F Asano A Furukawa H Saeki A Dey S K Liao L Ambrogio M W Botros Y Y Duan X Seki S Stoddart J F Yaghi O M Chemistry of Materials 2011 23 4094 (47) Ding X Chen L Honsho Y Feng X Saenpawang O Guo J Saeki A Seki S Irle S Nagase S Parasuk V Jiang D Journal of the American Chemical Society 2011 133 14510 (48) Tilford R W Mugavero S J Pellechia P J Lavigne J J Advanced Materials 2008 20 2741 (49) Lanni L M Tilford R W Bharathy M Lavigne J J Journal of the American Chemical Society 2011 133 13975 (50) Li Y Yang R T Aiche Journal 2008 54 269 (51) Nagai A Guo Z Feng X Jin S Chen X Ding X Jiang D Nature Communications 2011 2 (52) Bunck D N Dichtel W R Angewandte Chemie-International Edition 2012 51 1885 (53) Ding S-Y Gao J Wang Q Zhang Y Song W-G Su C-Y Wang W Journal of the American Chemical Society 2011 133 19816 (54) Miyaura N Suzuki A Chemical Reviews 1995 95 2457 (55) Kalidindi S B Yusenko K Fischer R A Chemical Communications 2011 47 8506 (56) Kuhn P Antonietti M Thomas A Angewandte Chemie-International Edition 2008 47 3450 (57) Bojdys M J Jeromenok J Thomas A Antonietti M Advanced Materials 2010 22 2202 (58) Kuhn P Forget A Su D Thomas A Antonietti M Journal of the American Chemical Society 2008 130 13333 (59) Ren H Ben T Wang E Jing X Xue M Liu B Cui Y Qiu S Zhu G Chemical Communications 2010 46 291 (60) Zhang W Li C Yuan Y-P Qiu L-G Xie A-J Shen Y-H Zhu J-F Journal of Materials Chemistry 2010 20 6413 (61) Trewin A Cooper A I Angewandte Chemie-International Edition 2010 49 1533 (62) Mastalerz M Angewandte Chemie-International Edition 2008 47 445
79
(63) Chan-Thaw C E Villa A Katekomol P Su D Thomas A Prati L Nano Letters 2010 10 537 (64) Palkovits R Antonietti M Kuhn P Thomas A Schueth F Angewandte Chemie-International Edition 2009 48 6909 (65) Ben T Ren H Ma S Q Cao D P Lan J H Jing X F Wang W C Xu J Deng F Simmons J M Qiu S L Zhu G S Angewandte Chemie-International Edition 2009 48 9457 (66) Yamamoto T Bulletin of the Chemical Society of Japan 1999 72 621 (67) Zhou G Baumgarten M Muellen K Journal of the American Chemical Society 2007 129 12211 (68) Yuan Y Sun F Ren H Jing X Wang W Ma H Zhao H Zhu G Journal of Materials Chemistry 2011 21 13498 (69) Ren H Ben T Sun F Guo M Jing X Ma H Cai K Qiu S Zhu G Journal of Materials Chemistry 2011 21 10348 (70) Zhao H Jin Z Su H Jing X Sun F Zhu G Chemical Communications 2011 47 6389 (71) Mortera R Fiorilli S Garrone E Verne E Onida B Chemical Engineering Journal 2010 156 184 (72) Dogru M Sonnauer A Gavryushin A Knochel P Bein T Chemical Communications 2011 47 1707 (73) Spitler E L Koo B T Novotney J L Colson J W Uribe-Romo F J Gutierrez G D Clancy P Dichtel W R Journal of the American Chemical Society 2011 133 19416 (74) Zhang Y Tan M Li H Zheng Y Gao S Zhang H Ying J Y Chemical Communications 2011 47 7365 (75) Uribe-Romo F J Doonan C J Furukawa H Oisaki K Yaghi O M Journal of the American Chemical Society 2011 133 11478 (76) Ben T Pei C Zhang D Xu J Deng F Jing X Qiu S Energy amp Environmental Science 2011 4 3991 (77) Lukose B Kuc A Heine T Chemistry-a European Journal 2011 17 2388 (78) Zhou W Wu H Yildirim T Chemical Physics Letters 2010 499 103 (79) Amirjalayer S Snurr R Q Schmid R Journal of Physical Chemistry C 2012 116 4921 (80) Xu Q Zhong C Journal of Physical Chemistry C 2010 114 5035 (81) Lukose B Supronowicz B St Petkov P Frenzel J Kuc A B Seifert G Vayssilov G N Heine T Physica Status Solidi B-Basic Solid State Physics 2012 249 335 (82) Assfour B Seifert G Chemical Physics Letters 2010 489 86 (83) Zhao L Zhong C L Journal of Physical Chemistry C 2009 113 16860 (84) Schmid R Tafipolsky M Journal of the American Chemical Society 2008 130 12600 (85) Han S S Goddard W A III Journal of Physical Chemistry C 2007 111 15185 (86) Suh M P Park H J Prasad T K Lim D-W Chemical Reviews 2012 112 782 (87) Furukawa H Yaghi O M Journal of the American Chemical Society 2009 131 8875 (88) Wong-Foy A G Matzger A J Yaghi O M Journal of the American Chemical Society 2006 128 3494 (89) Mendoza-Cortes J L Han S S Furukawa H Yaghi O M Goddard III W A Journal of Physical Chemistry A 2010 114 10824 (90) Doonan C J Tranchemontagne D J Glover T G Hunt J R Yaghi O M Nature Chemistry 2010 2 235 (91) Getman R B Bae Y-S Wilmer C E Snurr R Q Chemical Reviews 2012 112 703 (92) Han S S Furukawa H Yaghi O M Goddard III W A Journal of the American Chemical Society 2008 130 11580 (93) Furukawa H Ko N Go Y B Aratani N Choi S B Choi E Yazaydin A O Snurr R Q OKeeffe M Kim J Yaghi O M Science 2010 329 424 (94) Garberoglio G Langmuir 2007 23 12154 (95) Assfour B Seifert G Microporous and Mesoporous Materials 2010 133 59
80
(96) Lan J Cao D Wang W Journal of Physical Chemistry C 2010 114 3108 (97) Yang Q Zhong C Langmuir 2009 25 2302 (98) Garberoglio G Vallauri R Microporous and Mesoporous Materials 2008 116 540 (99) Lan J H Cao D P Wang W C Ben T Zhu G S Journal of Physical Chemistry Letters 2010 1 978 (100) Furukawa H Miller M A Yaghi O M Journal of Materials Chemistry 2007 17 3197 (101) Babarao R Jiang J Energy amp Environmental Science 2008 1 139 (102) Choi Y J Choi J H Choi K M Kang J K Journal of Materials Chemistry 2011 21 1073 (103) Lan J Cao D Wang W Smit B Acs Nano 2010 4 4225 (104) Wang L Yang R T Energy amp Environmental Science 2008 1 268 (105) Krishna R van Baten J M Industrial amp Engineering Chemistry Research 2011 50 7083 (106) Keskin S Journal of Physical Chemistry C 2012 116 1772 (107) Liu Y Liu D Yang Q Zhong C Mi J Industrial amp Engineering Chemistry Research 2010 49 2902 (108) Keskin S Sholl D S Langmuir 2009 25 11786 (109) Klontzas E Tylianakis E Froudakis G E Nano Letters 2010 10 452 (110) Zhu Y Zhou J Hu J Liu H Hu Y Chinese Journal of Chemical Engineering 2011 19 709 (111) Mendoza-Cortes J L Pascal T A Goddard W A III Journal of Physical Chemistry A 2011 115 13852 (112) Lino M A Lino A A Mazzoni M S C Chemical Physics Letters 2007 449 171 (113) Krishnan A Dujardin E Ebbesen T W Yianilos P N Treacy M M J Physical Review B 1998 58 14013 (114) Kim D Jung D H Kim K-H Guk H Han S S Choi K Choi S-H Journal of Physical Chemistry C 2012 116 1479 (115) Kim D Jung D H Choi S-H Kim J Choi K Journal of the Korean Physical Society 2008 52 1255 (116) Farha O K Yazaydin A O Eryazici I Malliakas C D Hauser B G Kanatzidis M G Nguyen S T Snurr R Q Hupp J T Nature Chemistry 2010 2 944 (117) Wu M M Wang Q Sun Q Jena P Kawazoe Y Journal of Chemical Physics 2010 133 (118) Choi Y J Lee J W Choi J H Kang J K Applied Physics Letters 2008 92 (119) Mendoza-Cortes J L Han S S Goddard W A III Journal of Physical Chemistry A 2012 116 1621 (120) Cao D Lan J Wang W Smit B Angewandte Chemie-International Edition 2009 48 4730 (121) Lan J H Cao D P Wang W C Langmuir 2010 26 220 (122) Klontzas E Tylianakis E Froudakis G E Journal of Physical Chemistry C 2009 113 21253 (123) Sun Y Ben T Wang L Qiu S Sun H Journal of Physical Chemistry Letters 2010 1 2753 (124) Li F Zhao J Johansson B Sun L International Journal of Hydrogen Energy 2010 35 266 (125) Zou X Zhou G Duan W Choi K Ihm J Journal of Physical Chemistry C 2010 114 13402 (126) Babarao R Dai S Jiang D-e Langmuir 2011 27 3451
81
Appendix B
Structural properties of metal-organic frameworks within the density-functional based tight-binding method
Binit Lukose Barbara Supronowicz Petko St Petkov Johannes Frenzel Agnieszka Kuc
Gotthard Seifert Georgi N Vayssilov and Thomas Heine
Phys Status Solidi B 2012 249 335ndash342
DOI 101002pssb201100634
Abstract Density-functional based tight-binding (DFTB) is a powerful method to describe large
molecules and materials Metal-organic frameworks (MOFs) materials with interesting catalytic
properties and with very large surface areas have been developed and have become commercially
available Unit cells of MOFs typically include hundreds of atoms which make the application of
standard density-functional methods computationally very expensive sometimes even unfeasible
The aim of this paper is to prepare and to validate the self-consistent charge-DFTB (SCC-DFTB)
method for MOFs containing Cu Zn and Al metal centers The method has been validated against
full hybrid density-functional calculations for model clusters against gradient corrected density-
functional calculations for supercells and against experiment Moreover the modular concept of
MOF chemistry has been discussed on the basis of their electronic properties We concentrate on
MOFs comprising three common connector units copper paddlewheels (HKUST-1) zinc oxide Zn4O
tetrahedron (MOF-5 MOF-177 DUT-6 (MOF-205)) and aluminum oxide AlO4(OH)2 octahedron (MIL-
53) We show that SCC-DFTB predicts structural parameters with a very good accuracy (with less than
82
5 deviation even for adsorbed CO and H2O on HKUST-1) while adsorption energies differ by 12 kJ
mol1 or less for CO and water compared to DFT benchmark calculations
1 Introduction In reticular or modular chemistry molecular building blocks are stitched together to
form regular frameworks [1] With this concept it became possible to construct framework
compounds with interesting structural and chemical composition most notably metal-organic
frameworks (MOFs) [1ndash10] and covalent organic frameworks (COFs) [11ndash17] The interest in MOFs
and COFs is not limited to chemistry these crystalline materials are also interesting for applications
in information storage [18] sieving of quantum liquids [19] hydrogen storage [20] and for fuel cell
membranes [21ndash23]
Covalent organic framework and MOF frameworks are composed by combining two types of building
blocks the so-called connectors typically coordinating in four to eight sites and linkers which have
typically 2 sometimes also three or four connecting sites Fig 1 illustrates a simplified representation
of the topology of connectors and linkers by using secondary building units (SBUs) (see Fig 1)
Figure 1 The connector and the linker of the frameworks CuBTC (HKUST-1) MOF-177 MOF-5 DUT-6 (MOF-205) and MIL-53 and their respective secondary building units (SBUs) The arrows indicate the connection points of the fragments Red oxygen green carbon white hydrogen yellow copperzinc and blue aluminum
Linkers are organic molecules with carboxylic acid groups at their connection sites which form
bonds to the connectors (typically in a solvothermal condensation reaction) They can carry
functional groups which can make them interesting for applications in catalysis [24] Connectors are
83
either metal organic units as for example the well-known Zn4O(CO2)6 unit of MOF-5 [8] the
Cu2(HCOO)4 paddle wheel unit of HKUST-1 [10] or covalent units incorporating boron oxide linking
units for COFs [11] As the building blocks of MOFs and COFs comprise typically some ten atoms unit
cells quickly get very large In particular if adsorption or dynamic processes in MOFs and COFs are of
interest (super)cells of some 1000 atoms need to be processed While standard organic force fields
show a reasonable performance for COFs [25] the creation of reliable force fields is not
straightforward for MOFs as transferable parameterization of the transition metal sites is an issue
even though progress has been achieved for selected materials [26 27] The difficulty to describe
transition metals in particular if they are catalytically active as in HKUST-1 limits the applicability of
molecular mechanics (MM) even for QMMM hybrid methods [28]
On the other hand the density-functional based tight-binding (DFTB) method with its self-consistent
charge (SCC) extension to improve performance for polar systems is a computationally feasible
alternative This non-orthogonal tight-binding approximation to density-functional theory (DFT)
which has been developed by Seifert Frauenheim and Elstner and their groups [29ndash32] (for a recent
review see Ref [30]) has been successfully applied to a large-scale systems such as biological
molecules [33ndash36] supramolecular systems [37 38] surfaces [39 40] liquids and alloys [41 42] and
solids [43] Being a quantum-mechanical method and hence allowing the description of breaking and
formation of chemical bonds the method showed outstanding performance in the description of
processes such as the mechanical manipulation of nanomaterials [44ndash46] It is remarkable that the
method performs well for systems containing heavier elements such as transition metals as this
domain cannot be covered so far with acceptable accuracy by traditional semi-empirical methods [47
48] DFTB covers today a large part of the elements of the periodic table and parameters and a
computer code are available from the DFTBorg website
Recently we have validated DFTB for its application for COFs [16 17] and have shown that structural
properties and formation energies of COFs are well described within DFTB Kuc et al [49] have
validated DFTB for substituted MOF-5 frameworks where connectors are always the Zn4O(CO2)6 unit
which has been combined with a large variety of organic linkers In this work we have revised the
DFTB parameters developed for materials science applications and validated them for HKUST-1 and
being far more challenging for the interaction of its catalytically active Cu sites with carbon
monoxide (CO) and water The Cu2(HCOO)4 paddle wheel units are electronically very intriguing on a
first note the electronic ground state of the isolated paddle wheel is an antiferromagnetic singlet
state which cannot be described by one Slater determinant and which is consequently not accessible
for KohnndashSham DFT However the energetically very close triplet state correctly describes structure
and electronic density of the system and also adsorption properties agree well with experiment [32
84
50 51] We therefore use DFT in the B3LYP hybrid functional representation as benchmark for DFTB
validations for HKUST-1 added by DFT-GGA calculations for periodic unit cellsWewill show that the
general transferability of the DFTB method will allow investigating structural electronic and in
particular dynamic properties
2 Computational details All calculations of the finite model and periodic crystal structures of MOFs
were carried out using the dispersion-corrected self-consistent density functional based tight-binding
(DC-SCC-DFTB in short DFTB) method [30 52ndash54] as implemented in the deMonNano code [55] Two
sets of parameters have been used the standard SCC-DFTB parameter set developed by Elstner et al
[53] for Zn-containing MOFs for Cu- and Al-containing materials we have extended our materials
science parameter set which has been developed originally for zeolite materials to include Cu For
this element we have used the standard procedure of parameter generation we have used the
minimal atomic valence basis for all atoms including polarization functions when needed Electrons
below the valence states were treated within the frozen-core approximation The matrix elements
were calculated using the local density approximation (LDA) while the short-range repulsive pair-
potential was fitted to the results from DFT generalized gradient approximation (GGA) calculations
For more details on DFTB parameter generation see Ref [30] Periodic boundary conditions were
used to represent frameworks of the crystalline solid state The conjugate-gradient scheme was
chosen for the geometry optimization An atomic force tolerance of 3x10-4 eV Aring-1 was applied
The reference DFT calculations of the equilibrium geometry of the investigated isolated MOF models
were performed employing the Becke three-parameter hybrid method combined with a LYP
correlation functional (B3LYP) [56 57] The basis set associated with the HayndashWadt [58] relativistic
effective core potentials proposed by Roy et al [59] was supplemented with polarization ffunctions
[60] and was employed for description of the electronic structure of Cu atoms The Pople 6-311G(d)
basis sets were applied for the H C and O atoms The calculations were performed with the
Gaussian09 program suite [61] For optimization of the periodic structure of CuBTC at DFT level the
electronic structure code Quickstep [62] which is a part of the CP2K package [63] was used
Quickstep is an implementation of the Gaussian plane wave method [64] which is based on the
KohnndashSham formulation of DFT
We have used the generalized gradient functional PBE [65] and the GoedeckerndashTeterndashHutter
pseudopotentials [64 66] in conjunction with double-z basis sets with polarization functions DZVP-
MOLOPT-SR-GTH basis sets obtained as described in Ref [67] but using two less diffuse primitives
Due to the large unit cell sampling of Brillouin zone was limited only to the G-point Convergence
85
criterion for the maximum force component was set to 45 x 10-3 Hartree Bohr1 The plane wave
basis with cutoff energy of 400 Ry was used throughout the simulations
The initial crystal structures were taken from experiment (powder X-ray diffraction (PXRD) data) The
cell parameters and the atomic positions were fully optimized using conjugate-gradient method at
the DFTB level For the reference DFT calculations only the atomic positions of experimental crystal
structures were minimized The cluster models were cut from the optimized structures and saturated
with hydrogen atoms (see Fig 2 for exemplary systems of Cu-BTC)
3 Results and discussion
31 Cu-BTC We have optimized the crystal structure of Cu-BTC (HKUST-1) [10] using cluster and the
periodic models The structural properties were compared to DFT results (see Table 1) The
geometries were obtained for the activated form (open metal sites) and in the presence of axial
water ligands (see Table 2) as the synthesized crystals often have H2O molecules attached to the
open metal sites of Cu (Fig 2b) We have also studied changes of the cluster model geometry in the
presence of CO (Fig 2c) and the results are also shown in Table 2 We have obtained good agreement
with experimental data as well as with DFT results
Figure 2 (a) Periodic and cluster models of Cu-BTC MOF as used in the computer simulations (b) cluster model with adsorbed H2O and (c) CO molecules
We have also simulated the XRD patterns of Cu-BTC which show very good structural agreement for
peak positions between the experimental and calculated structures The XRD pattern of DFT
optimized structure is nearly a copy of that of the experimental geometry However for DFTB
optimized patterns the relative intensities of some of the peaks notably that of the peaks at 2θ =
138 and 2θ = 158 are different from the experimental XRDs (see Fig 3) The differences in bond
angles between simulation and experiment may affect the sharpness of the signals and hence the
86
intensity To support this argument we have calculated the radial pair distribution function (g(r))
which details the probabilities of all atomndashatom distances in a given cell volume Figure S3 in the
Supporting Information shows g(r) of CundashCu CundashC CundashO CndashC and CndashO atom pairs of DFTB
optimized DFT optimized and experimentally modeled Cu-BTC unit cells The peaks of DFT as well as
DFTB optimized geometries are much broadened whereas the experimentally modeled geometry
has single peaks at the most predicted atomndashatom distances However it is to be noted that DFTB
optimized geometries give slightly shifted peaks compared to those of the DFT optimized geometry
They are broadened around the experimental values The distances between Cu and C atoms which
are not direct neighbors have the largest deviations from the experiment what indicates
shortcomings of the bond angles
Table 1 Selected bond lengths (Aring ) bond angles (⁰) and dihedral angles (⁰) of Cu-BTC optimized at DFTB and DFT level
Bond Type Cluster Model Periodic Model Exp
Cu-Cu 250 (257) 250 (250) 263 Cu-O 205 (198) 202 (198) 195 O-C 134 (133) 133-138 (128) 125
OCuO 836-971 (898) 892-907 (873-937)
891 896
Cu-O-O-Cu plusmn56 ˗ plusmn62 (plusmn02) plusmn04 (plusmn47 ˗ plusmn131) 0
O-C-C-C plusmn38 - plusmn50 (0) plusmn03 - plusmn05 (plusmn06 - plusmn105) 063
Cell paramet a=b=c=27283 (26343)
α=β=γ=90 (90) a=b=c=26343
α=β=γ=90
In detail the bond lengths and bond angles do not change significantly when going from the cluster
to the periodic model confirming the reticular chemistry approach The only exception is the OndashCundash
O bond angle that differs by 4ndash78 between the two systems at both levels of theory
87
Figure 3 Simulated XRD patterns of Cu-BTC MOF with the experimental unit cell parameters and optimized at the DFTB and DFT level of theory
The bond length between Cu atoms is slightly underestimated comparing with experimental (by
maximum 5) and DFT (maximum by 3) results while all other bond lengths are somewhat larger
at DFTB
All bond lengths stay unchanged or become longer in the presence of water molecules The most
striking example is the CundashCu bond where the change reaches 012ndash017 Aring compared with the
structure with activated Cu sites Also bond angles increase by up to 28 when the H2O is present
The CundashO bond length with the oxygen atom from the carboxylate groups is shorter than that with
the oxygen atoms from the water molecules indicating that the latter is only weakly bound to the
copper ions The OndashC distances (~133 Aring) correspond to values between that for the typical single
(142 Aring ) and double (122 Aring) oxygenndashcarbon bond The calculated CringndashCcarboxyl bond lengths of
146A deg indicate that these are typical single sp2ndashsp2 CndashC bonds while experimental findings give a
slightly longer value (15 Aring) for this MOF When comparing dihedral angles CundashOndashOndashCu and OndashCndashCndashC
of the clusters fairly large deviations between DFTB and DFT calculations and experiment are visible
due to the softer potential energy surface associated with these geometrical parameters In periodic
models however the agreement of DFT and DFTB with experiment and with each other is
significantly improved
The unit cell parameters with and without water molecules obtained at the DFTB level overestimate
the experimental data by less than 4 which gives a fairly good agreement if we take into account
high porosity of the material These lattice parameters increase only slightly from 27283 to 27323 Aring
in the presence of water
We have calculated the binding energies of H2O molecules attached to the Cu-metal sites within the
cluster model At the DFTB level Ebind of 40 kJ mol-1 was obtained in a good agreement with the DFT
results (52 kJ mol-1) as well as results from the literature (47 kJ mol-1 [50]) We have also calculated
88
the binding energy of CO which was twice smaller than that of H2O (165 and 28 kJ mol-1 for DFTB
and DFT respectively) suggesting much stronger binding of O atom (from H2O) than C atom (from
CO) While the bond lengths in the MOF structure do not change in the presence of either H2O or CO
the differences in the binding energy come from much longer bond distances (by around 07 Aring) for
CundashC than for CundashO in the presence of CO and water molecules respectively
Furthermore we have studied the electronic properties of periodic and cluster model Cu-BTC by
means of partial density-of-states (PDOS)We have compared especially the Cu-PDOS for s- p- and d-
orbitals (see Fig 4) The results show that the electronic structure stays unchanged when going from
the cluster models to the fully periodic crystal of Cu-BTC Since the copper atoms are of Cu(II) the d-
orbitals are just partially occupied what results in the metallic states at the Fermi level This is a very
interesting result as other ZnndashO-based MOFs are either semiconductors or insulators [49]
Table 2 Selected bond lengths (Aring) bond angles (˚) and dihedral angles (˚) of Cu-BTC optimized at DFTB and DFT levels with water molecules in the axial positions Cluster models are calculated using water and carbon monoxide molecules The adsorption energies Eads (kJ mol-1) of the guest molecules are given as well The DFT data are given in parenthesis
Bond Type Cluster Model +
H2O Periodic
Model+ H2O Cluster Model +
CO
Cu-Cu 267 (266) 262 (260) 250 (260)
Cu-O 205 (197-206) 210 (196-200) 206 (199)
O-C 134 (127) 133 (128) 134 (127)
OCuO 843-955 (889-905)
871-921 (842-930) 842-967 (896)
Cu-O-O-Cu plusmn59 - plusmn76 (plusmn06 ˗ plusmn10)
plusmn02 ˗ plusmn18 (plusmn40 ˗ plusmn136)
plusmn51 - plusmn58 (plusmn70)
O-C-C-C plusmn39 - plusmn53 (plusmn05 - plusmn11)
plusmn03 - plusmn05 (plusmn06 - plusmn105)
plusmn35 - plusmn43 (plusmn12)
Cu-O(H2O) Cu-C(CO) 237 (219) 244 (233-
255) 307 (239)
Eads -4045 (-5200) -1648
(-2800)
32 MOF-5 -177 DUT-6 (MOF-205) and MIL- 53 We have also studied the structural properties
of MOF structures with large surface areas using the SCC-DFTB method Very good agreement with
the experimental data shows that this method is applicable for MOFs of large structural diversity as
well as for coordination polymers based on the MOF-5 framework which has been reported earlier
[49] Tables 3 and 4 show selected bond lengths and bond angles of MOF-5 [68] MOF-177 [69] DUT-
6 (MOF-205) [70 71] and MIL-53 [72] respectively
89
MOFs-5 -177 and DUT-6 (MOF-205) are built of the same connector which is Zn4O(CO2)6
octahedron The difference is in the organic linker 14-benzenedicarboxylate (BDC) 135-
benzenetribenzoates (BTB) and BTB together with 26-naphtalenedicarboxylate (NDC) for MOF-5 -
177 andDUT-6 (MOF-205) respectively (see Fig 5)
Figure 4 DFTB calculations of PDOS of (top left) Cu in Cu-BTC and Zn in MOF-5 (top right) MOF-177 (bottom left) and DUT-6 (MOF-205) (bottom right) In each plot s-orbitals (top) p-orbitals (middle) and d-orbitals (bottom) are shown as computed from periodic (black) and cluster (red) models Cluster models are given in Fig S4
All three MOFs have different topologies due to the organic linkers where the number of
connections is varied or where two different linker types are present MOF-5 is the most simple and
it is the prototype in the isoreticular series (IRMOF-1) [68] it is a simple cubic topology with
threedimensional pores of the same size and the linkers have only two connection points In the
case of MOF-177 the linker is represented by a triangularSBU that means three connection points
are present This results in the topology with mixed (36)-connectivity [69] DUT-6 (MOF-205) has a
much more complicated topology due to two types of linkers The first one (NDC) has just two
90
connection points while the second is the same as in MOF-177 with three connection points One
ZnndashO-based connector is connected to two NDC and four BTB linkers The topology is very interesting
all rings of the underlying nets are fivefold and it forms a face-transitive tiling of dodectahedra and
tetrahedra with a ratio of 13 [73]
Figure 5 The crystal structures in the unit cell representations of (a) MOF-5 (b) MOF-177 (c) DUT-6 (MOF-205) and (d) MIL-53 red oxygen gray carbon white hydrogen and blue metal atom (Zn Al)
MIL-53 is a MOF structure with one-dimensional pores The framework is built up of corner-sharing
AlndashO-based octahedral clusters interconnected with BDC linkers The linkers have again two
connection points MIL-53 shows reversible structural changes dependent on the guest molecules
[74] It undergoes the so-called breathing mode depending on the temperature and the amount of
adsorbed molecules
In this case also the bond lengths and bond angles are slightly overestimated comparing with the
experimental structures but the error does not exceed 3
91
Table 3 Selected bond lengths (Aring) and bond angles (˚) of MOF-5 (424 atoms in unit cell) MOF-177 (808 atoms in unit cell) and DUT-6 (MOF-205) (546 atoms in unit cell) optimized at DFTB level The available experimental data is given in parenthesis [70-73+ Orsquo denotes the central O atom in the Zn-O-octahedron
Bond Type MOF-5 MOF-177 DUT-6
(MOF-205)
Zn-Zn 330 (317) 322-336 (306-330)
325-331 (318)
Zn-Orsquo 202 (194) 202 (193) 202 (194) Zn-O 204 (192) 202-206
(190-199) 202 205 (193)
O-C 128 (130) 128 (131) 128 (125) ZnOrsquoZn 1095 (1095) 1056-1124
(1055 1092) 107-1118 (1084 1100)
OZnO 1083 1108 (1061)
1048 1145 (981-1281)
1046-1112 (1062 1085)
Zn-O-O-Zn 0 (0 plusmn17 plusmn20) plusmn122 - plusmn347 (plusmn176 - plusmn374)
05 - plusmn62 (0 plusmn29)
O-C-C-C 00 (0 - plusmn24) (plusmn 35 - plusmn 336) (plusmn65 - plusmn374)
plusmn04 plusmn22 (0 plusmn174)
Cell paramet a=b=c=26472 (25832) α=β=γ=90 (90)
a=b=37872 (37072) c=3068 (30033) α=β=90 (90) γ=120 (120)
a=b=c=31013 (30353) α=β=γ=90 (90)
We have calculated PDOS for MOF-5 MOF-177 and DUT-6 (MOF-205) (see Fig 4) Their band gaps
calculated to be 40 35 and 31 eV respectively indicate that they are either semiconductors or
insulators We have calculated it for both periodic crystal structure and a cluster containing a metal-
oxide connector and all its carboxylate linkers
Table 4 Selected bond lengths (Aring) and bond angles (˚) of MIL-53 (152 atoms in unit cell) optimized at DFTB level
Bond Type DFTB Exp
Al-Al 341 331 Al-O 189 183-191 O-C 133 129 130 O-Al-O 884 912 886 898 Al-O-Al 1289 1249 Al-O-O-Al 0 0 O-C-C-O 06 03 Cell paramet a=1246
b=1732 c=1365 α=β=γ=90
a=1218 b=1713 c=1326 α=β=γ=90
4 Mechanical properties Due to the low-mass density the elastic constants of porous materials
are a very sensitive indicator of their mechanical stability For crystal structures like MOFs we have
92
studied these by means of bulk modulus (B) B can be calculated as a second derivative of energy
with respect to the volume of the crystal (here unit cell)
The result shows that CuBTC has bulk modulus of 3466 GPa what is in close agreement with
B=3517 GPa obtained using force-field calculations [73 74] and experiment [75] Bulk moduli for the
series of MOFs resulted in 1534 1010 and 1073 GPa for MOF-5 -177 and DUT-6 (MOF-205)
respectively For MOF-5 B=1537 GPa was reported from DFTGGA calculations using plane-waves
[76 77] The results show that larger linkers give mechanically less stable structures what might be
an issue for porous structures with larger voids For MIL-53 we have obtained the largest bulk
modulus of 5369 GPa keeping the angles of the pore fixed
5 Conclusions We have validated the DFTB method with SCC and London dispersion corrections for
various types of MOFs The method gives excellent geometrical parameters compared to experiment
and for small model systems also in comparison with DFT calculations Importantly this statement
holds not only for catalytically inactive MOFs based on the Zn4O(CO2)6 octahedron and organic linkers
which are important for gas adsorption and separation applications but also for catalytically active
HKUST-1 (CuBTC) This is remarkable as the Cu atoms are in the Cu2+ state in this framework DFTB
parameters have been generated and validated for Cu and the electronic structure contains one
unpaired electron per Cu atom in the unit cell which makes the electronic description technically
difficult but manageable within the SCC-DFTB method SCC-DFTB performs well for the frameworks
themselves as well as for adsorbed CO and water molecules
Partial density-of-states calculations for the transition metal centers reveal that the concept of
reticular chemistry ndash individual building units keep their electronic properties when being integrated
to the framework ndash is also valid for the MOFs studied in this work in agreement with a previous
study of COFs [16] The electronic properties computed using the cluster models and the periodic
structure contains the same features and hence cluster models are good models to study the
catalytic and adsorption properties of these materials This is in particular useful if local quantum
chemical high-level corrections shall be applied that require to use finite structures
We finally conclude that we have now a high-performing quantum method available to study various
classes of MOFs of unit cells up to 10000 atoms including structural characteristics the formation
and breaking of chemical and coordination bonds for the simulation of diffusion of adsorbate
molecules or lattice defects as well as electronic properties The parameters can be downloaded
from the DFTBorg website
93
References
[1] E A Tomic J Appl Polym Sci 9 3745 (1965)
2+ M Eddaoudi D B Moler H L Li B L Chen T M Reineke M OrsquoKeeffe and O M Yaghi Acc Chem Res
34 319 (2001)
[3] M Eddaoudi H L Li T Reineke M Fehr D Kelley T L Groy and O M Yaghi Top Catal 9 105 (1999)
[4] B F Hoskins and R Robson J Am Chem Soc 111 5962 (1989)
[5] K Schlichte T Kratzke and S Kaskel Microporous Mesoporous Mater 73 81 (2004) [6] J L C Rowsell A
R Millward K S Park and O M Yaghi J Am Chem Soc 126 5666 (2004)
7+ O M Yaghi M OrsquoKeeffe N W Ockwig H K Chae M Eddaoudi and J Kim Nature 423 705 (2003)
[8] M Eddaoudi J Kim N Rosi D Vodak J Wachter M OrsquoKeeffe and O M Yaghi Science 295 469 (2002)
9+ M OrsquoKeeffe M Eddaoudi H L Li T Reineke and O M Yaghi J Solid State Chem 152 3 (2000)
[10] S S Y Chui SM F Lo J P H Charmant A G Orpen and I D Williams Science 283 1148 (1999)
11+ A P Cote A I Benin N W Ockwig M OrsquoKeeffe A J Matzger and O M Yaghi Science 310 1166 (2005)
[12] A P Cote H M El-Kaderi H Furukawa J R Hunt and O M Yaghi J Am Chem Soc 129 12914 (2007)
[13] H M El-Kaderi J R Hunt J L Mendoza-Cortes A P Cote R E Taylor M OrsquoKeeffe and O M Yaghi
Science 316 268 (2007)
[14] S Wan J Guo J Kim H Ihlee and D L Jiang Angew Chem 120 8958 (2008)
[15] S Wan J Guo J Kim H Ihlee and D L Jiang Angew Chem 121 5547 (2009)
[16] B Lukose A Kuc J Frenzel and T Heine Beilstein J Nanotechnol 1 60 (2010)
[17] B Lukose A Kuc and T Heine Chem Eur J 17 2388 (2011)
[18] D S Marlin D G Cabrera D A Leigh and A M Z Slawin Angew Chem Int Ed 45 77 (2006)
[19] I Krkljus and M Hirscher Microporous Mesoporous Mater 142 725 (2011)
[20] M Hirscher Handbook of Hydrogen Storage (Wiley-VCH Weinheim 2010)
[21] H Kitagawa Nature Chem 1 689 (2009)
[22] M Sadakiyo T Yamada and H Kitagawa J Am Chem Soc 131 9906 (2009)
[23] T Yamada M Sadakiyo and H Kitagawa J Am Chem Soc 131 3144 (2009)
94
[24] K S Jeong Y B Go S M Shin S J Lee J Kim O M Yaghi and N Jeong Chem Sci 2 877 (2011)
[25] R Schmid and M Tafipolsky J Am Chem Soc 130 12600 (2008)
[26] M Tafipolsky S Amirjalayer and R Schmid J Comput Chem 28 1169 (2007)
[27] M Tafipolsky S Amirjalayer and R Schmid J Phys Chem C 114 14402 (2010)
[28] A Warshel and M Levitt J Mol Biol 103 227 (1976)
[29] G Seifert D Porezag and T Frauenheim Int J Quantum Chem 58 185 (1996)
[30] A F Oliveira G Seifert T Heine and H A Duarte J Braz Chem Soc 20 1193 (2009)
[31] D Porezag T Frauenheim T Koumlhler G Seifert and R Kaschner Phys Rev B 51 12947 (1995)
[32] T Frauenheim G Seifert M Elstner Z Hajnal G Jungnickel D Porezag S Suhai and R Scholz Phys
Status Solidi B 217 41 (2000)
[33] M Elstner Theor Chem Acc 116 316 (2006)
Supporting Information
Figure S4 Cluster models as used for the P-DOS calculations in Fig 4 MOF-5 MOF-177 and DUT-6 (MOF-205) (from left to right)
95
Figure S3 Radial pair distribution function (g(r)) of Cu-Cu Cu-C Cu-O C-C and C-O atom-pairs in a Cu-BTC MOF unit cell
96
Appendix C
The Structure of Layered Covalent-Organic Frameworks
Binit Lukose Agnieszka Kuc and Thomas Heine
Chem Eur J 2011 17 2388 ndash 2392
DOI 101002chem201001290
Abstract Covalent-Organic Frameworks (COFs) are a new family of 2D and 3D highly porous and
crystalline materials built of light elements such as boron oxygen and carbon For all 2D COFs an AA
stacking arrangement has been reported on the basis of experimental powder XRD patterns with the
exception of COF-1 (AB stacking) In this work we show that the stacking of 2D COFs is different as
originally suggested COF-1 COF-5 COF-6 and COF-8 are considerably more stable if their stacking
arrangement is either serrated or inclined and layers are shifted with respect to each other by ~14 Aring
compared with perfect AA stacking These structures are in agreement with to date experimental
data including the XRD patterns and lead to a larger surface area and stronger polarisation of the
pore surface
Introduction In 2005 Cocircteacute and co-workers introduced a new class of porous crystalline materials
Covalent Organic Frameworks (COFs)[1] where organic linker molecules are stitched together by
connectors covalent entities including boron and oxygen atoms to a regular framework These
materials have the genuine advantage that all framework bonds represent strong covalent
interactions and that they are composed of light-weight elements only which lead to a very low
mass density[2] These materials can be synthesized solvothermally in a condensation reaction and
97
are composed of inexpensive and non-toxic building blocks so their large-scale industrial production
appears to be possible
Figure 1 Calculated and experimental XRD patterns of all studied COFs (COF-1 top left COF-5 top right COF-6 bottom left COF-8 bottom right)
To date a number of different COF structures have been reported[1ndash3] From a topological
viewpoint we distinguish two- and three-dimensional COFs In two-dimensional (2D) COFs the
covalently bound framework is restricted to 2D layers The crystal is then similar as in graphite or
hexagonal boron nitride composed by a stack of layers which are not connected by covalent bonds
but held together primarily by London dispersion interactions
98
The COF structures have been analyzed by powder X-ray diffraction (PXRD) experiments The
topology of the layers is determined by the structure of the connector and linker molecules and
typically a hexagonal pattern is formed due to the 2m (D3h) symmetry of the connector moieties
The individual layers are then stacked and form a regular crystal lattice With one exception (COF-
1)[1] for all 2D-COFs AA (eclipsed P6mmm) interlayer stacking has been reported[3andashd f g] This
geometrical arrangement maximizes the proximity of the molecular entities and results in straight
channels orthogonal to the COF layers which are known from the literature[1 3a]
The connectors of 2D-COFs include oxygen and boron atoms which cause an appreciable polarization
The AA stacking arrangement maximizes the attractive London dispersion interaction between the
layers which is the dominating term of the stacking energy At the same time AA stacking always
results in a repulsive Coulomb force between the layers due to the polarized connectors It should be
noted that the eclipsed hexagonal boron nitride (h-BN)[4] has boron atoms in one layer serving as
nearest neighbours to nitrogen atoms in adjacent layers (AB stacking) The Coulomb interaction has
ruled out possible interlayer eclipse between atoms of alike charges In this article we aim at
studying the layer arrangements of solvent-free COFs based on the interlayer interactions and the
minimum variance Various lattice types have been considered all significantly more stable than the
reported AA stacked geometry In all 2D COFs we have studied the Coulomb interaction between the
layers leads to a modification of the stacking and shifts the layers by about one interatomic distance
(~14 Aring) with respect to each other (see Figure 1)
Breaking of the highly symmetric layer arrangement has been observed for COF-1 and COF-10 after
removal of guest molecules from pores leading to a turbostratic repacking of pore channels[1 5]
The authors have confirmed the disorder either by the changes in the PXRD spectrum taken before
and after adsorptionndashdesorption cycle[1] or by the changes in the adsorption behavior[5] The
disruption of layer arrangement is attributed to the force exerted by the guest molecules Formation
of poly disperse pores has also been verified[5] The lattice types that we discuss in this article on
the other hand are neither the result of the pressure from any external molecule in the pore nor
having more than one type of pores They are the resultant of minimum variance guided by Coulomb
and London dispersion interactions For the COF models under investigation perfect crystallinity has
been considered
Methods We have studied the experimentally known structures of COF-1 COF-5 COF-6 and COF-8
Structure calculations were carried out using the Dispersion-Corrected Self-Conistent-Charge
Density-Functional-Based Tight-Binding method (SCC-DFTB)[6] DFTB is based on a second-order
expansion of the KohnndashSham total energy in DFT with respect to charge density fluctuations This
does not require large amounts of empirical parameters however maintains all qualities of DFT The
99
accuracy is very much improved by the SCC extension The DFTB code (deMonNano[6a]) has
dispersion correction[6d] implemented to account for weak interactions and was used to obtain the
layered bulk structure of COFs and their formation energies The performance for interlayer
interactions has been tested previously for graphite[6d] All structures correspond to full geometry
optimizations (structure and unit cell) XRD patterns have been simulated using the Mercury
software[7] To allow best comparison with experiment for PXRD simulations we used the calculated
geometry of the layer and of the relative shifts between the layers but experimental interlayer
distances The electrostatic potential has been calculated using Crystal 09[8] at the DFTVWN level
with 6-31G basis set
Results and Discussion
In order to see the favorite stacking arrangement of the layers we have shifted every second layer in
two directions along zigzag and armchair vertices of the hexagonal pores (see Figure 2) The initial
stacking is AA (P6mmm) and the final zigzag shift gives AB stacking (P63mmc) Considering that the
attractive interlayer dispersion forces and repulsive interlayer orbital interactions are different we
have optimized the interlayer separation for each stacking Figure 2 show their total energies
calculated per formula unit that is per established bond between linkers and connectors with
reference to AA stacking for COF-5 and COF-1 In each direction the energy minimum is found close
to the AA stacking position shifted only by about one aromatic C-C bond length (~14 Aring) so that
either connector or linker parts become staggered with those in the adjacent layers leading to a
stacking similar to that of graphite for either connector (armchair shift) or linker (zigzag shift) For
COF-5 the staggering at the connector or linker depends whether the shift is armchair or zigzag
respectively while for COF-1 the zigzag shift alone can give staggering in both the aromatic and
boron-oxygen rings
The low-energy minima in the two directions are labeled following the common nomenclature as
zigzag and armchair respectively Both shifts result in a serrated stacking order (orthorhombic
Cmcm) which shows a significantly more attractive interlayer Coulomb interaction than AA stacking
(see Table 1) while most of the London dispersion attraction is maintained and consequently
stabilizes the material Still this configuration can be improved if we consider inclined stacking
(Figure 1) that is the lattice vector pointing out of the 2D plane is not any longer rectangular
reducing the lattice symmetry from Cmcm (orthorhombic) to C2m (monoclinic)
Besides we have calculated the monolayer formation energy (condensation energy Ecb) from the
total energies of the monolayer and of the individual building blocks and the stacking formation
energy from the total energies of the bulk structure and of the monolayer for four selected COFs The
100
COF building blocks are the same as those used for their synthesis[1 3a] (BDBA for COF-1 BDBA and
HHTP for COF-5 BTBA and HHTP for COF-6 BTPA and HHTP for COF-8) The final energies given per
formula unit that is per condensed bond are given in Table 1 The serrated and inclined stacking
structures are energetically more stable than AA and AB Interestingly within our computational
model zigzag and armchair shifting is energetically equivalent
Figure 2 Change of the total energy (per formula unit) when shifting COF-5 even layers in zigzag (top) and armchair (middle) directions and COF-1 in zigzag (bottom) direction The vector d shows the shift direction The low-energy minima in the two directions are labelled as zigzag and armchair respectively The equilibrium structures are shown as well
The formation energies of the individual COF structures are in agreement with full DFT calculations
We have calculated using ADF[9] the condensation energy of COF-1 and COF-5 using first-principles
DFT at the PBE[10]DZP[11] level to qualitatively support our results For simplicity we have used a
finite structure instead of the bulk crystal Their calculated Ecb energies are 77 and 15 kJmol-1
respectively for COF-1 and COF-5 These values support the endothermic nature of the condensation
101
reaction and are in excellent agreement with our DFTB results (91 and 21 kJmol-1 respectively see
Table 1)
The change of stacking should be visible in X-ray diffraction patterns because each space group has a
distinct set of symmetry imposed reflection conditions Powder XRD patterns from experiments are
available for all 2D COF structures[1 3andashd f g] We have simulated XRD patterns for AA AB serrated
Table 1 Calculated formation energies for the studied COF structures Ecb = condensation Esb = stacking energy EL = London dispersion energy Ee = electronic energy Energies are per condensed bond and are given in [kJ mol
-1+ lsquoarsquo and lsquozrsquo stand for armchair and zigzag respectively Note that the electronic energy Ee=Esb-EL
includes the electronic interlayer interaction and the formation energy of the monolayer and that the formation of the monolayers is endothermic
Structure Stacking Esb EL Ee
COF-5 AA -2968 -3060 092
AB -2548 -2618 070
serrated z -3051 -3073 022
serrated a -3052 -3073 021
inclined z -3297 -3045 -252
inclined a -3275 -3044 -231
Monolayer Ecb= 211
COF-1 AA -2683 -2739 056
AB -2186 -2131 -055
serrated z -2810 -2806 -004
inclined z -2784 -2788 004
Monolayer Ecb= 906
COF-6 AA -2881 -2963 082
AB -2127 -2146 019
serrated z -2978 -2996 018
serrated a -2978 -2995 017
inclined z -2946 -2975 029
inclined a -2954 -2974 021
Monolayer Ecb= 185
COF-8 AA -4488 -4617 129
102
AB -2477 -2506 029
serrated z -4614 -4646 032
serrated a -4615 -4647 032
inclined z -4578 -4612 035
inclined a -4561 -4591 030
Monolayer Ecb= 263
and inclined stacking forms for COF-5 COF-1 COF-6 and COF-8 (see Figure 1for their comparison
with experimental spectrum) For completeness we have studied XRD patterns of optimized COFs
using the experimentally determined[1 3a] interlayer separations this means we have kept the
layer geometry the same as the optimized structures and different stackings were obtained by
shifting adjacent layers accordingly
COF-1 was reported as having different powder spectra for samples before (as-synthesized) and after
removal of guest molecules with a particular mentioning about its layer shifting after removal We
have compared the two spectra with our simulated XRDs in order to see the stacking in the pure
form and how the stacking is changed at the presence of mesitylene guests Except that we have only
a vague comparison at angles (2θ) 11ndash15 the powder spectrum after guest removal is more similar
to the XRDs of serrated and inclined stacking structures A notable difference from AB is the absence
of high peaks at angles ~12 ~15 and ~19 This gives rise to the suggestion that COF-1 is not a
notable exception among the 2D COFs it follows the same topological trend as all other frameworks
of this class having one-dimensional (1D) pores as soon as the solvent is removed from the pores
This suggestion is supported by the experimental fact that the gas adsorption capacity of COF-1 is
only slightly lower than that of COF-6 for even as large gas molecules as methane and CO2[12] It is
not obvious how these molecules would enter an AB stacked COF-1 framework where the pores are
not connected by 1D channels (see Figure 1) Moreover our calculations suggest that the serrated
and inclined stackings are energetically favorable (see Table 1)
Indeed all the new stackings discussed in this article yield XRD patterns which are in agreement with
the currently available experimental data (see Figure 1) The inclined stackings have more peaks but
those are covered by the broad peaks in the experimental pattern The same is observed for the (002)
peaks of serrated stackings At present 2D COF synthesis methods are not yet capable to produce
crystals of sufficient quality to allow more detailed PXRD analysis in particular for the solvent-free
materials Microwave synthesis of COF-5 by rapid heating is reported[13] as much faster (~200 times)
compared with solvothermal methods however the structural details (XRD etc) remained
103
ambiguous We are confident that better crystals will be produced in future which will allow the
unambiguous determination of COF structures and can be compared to our simulations
Finally we want to emphasize that the suggested change of stacking is not only resulting in a
moderate change of geometry and XRD pattern The functional regions of COFs are their pores and
the pore geometry is significantly modified in our suggested structures compared to AA and AB
stackings First the inclined and serrated structures account for an increase of the surface area by 6
estimated for the interaction of the COFs with helium (3 for N2 5 for Ar 56 for Ne) Moreover
the shift between the layers exposes both boron and oxygen atoms to the pore surface leading to a
much stronger polarity than it can be expected for AA stacked COFs This difference is shown in
Figure 3 which plots the electrostatic potential of COF-5 in AA serrated and inclined stacking
geometries While the pore surface of AA stacked COF-5 shows a positively and negatively charged
stripes the other stacking arrangements show a much stronger alternation of charges indicating the
higher polarity of the surface The surface polarity can also be expressed in terms of Mulliken charges
of oxygen and boron atoms calculated at the DFT level which are COF-5 AA qB = 048 and qO = -048
COF-5 serrated qB = 047 and qO = -049 COF-5 inclined qB = 045 and qO = -048
Figure 3 Calculated electrostatic potential of COF-5 AA (left) serrated (middle) and inclined (right) stacking forms Blue - negative and red-positive values of the 005 isosurface
Conclusion We have studied 2D COF layer stackings energetically and found that energy minimum
structures have staggering of hexagonal rings at the connector or linker parts This can be achieved if
the bulk structure has either serrated or inclined order These newly proposed orders have their
simulated XRDs matching well with the available experimental powder spectrum Hence we claim
that all reported 2D COF geometries shall be re-examined carefully in experiment Due to the change
of the pore geometry and surface polarity the envisaged range of applications for 2D COFs might
change significantly We believe that these results are of utmost importance for the design of
functionalized COFs where functional groups are added to the pore surfaces
104
References
[1] A P Cocircteacute A I Benin N W Ockwig M OKeeffe A J Matzger O M Yaghi Science 2005 310 1166
[2] H M El-Kaderi J R Hunt J L Mendoza-Cortes A P Cote R E Taylor M OKeeffe O M Yaghi Science
2007 316 268
[3] a) A P Cocircteacute H M El-Kaderi H Furukawa J R Hunt O M Yaghi J Am Chem Soc 2007 129 12914 b) J
R Hunt C J Doonan J D LeVangie A P Cocircte O M Yaghi J Am Chem Soc 2008 130 11872 c) R W
Tilford W R Gemmill H C zur Loye J J Lavigne Chem Mater 2006 18 5296 d) R W Tilford S J Mugavero
P J Pellechia J J Lavigne Adv Mater 2008 20 2741 e) F J Uribe-Romo J R Hunt H Furukawa C Klock M
OKeeffe O M Yaghi J Am Chem Soc 2009 131 4570 f) S Wan J Guo J Kim H Ihee D L Jiang Angew
Chem 2008 120 8958 Angew Chem Int Ed 2008 47 8826 g) S Wan J Guo J Kim H Ihee D L Jiang
Angew Chem 2009 121 5547 Angew Chem Int Ed 2009 48 5439
[4] R T Paine C K Narula Chem Rev 1990 90 73
[5] C Doonan D Tranchemontagne T Glover J Hunt O Yaghi Nat Chem 2010 2 235
[6] a) T Heine M Rapacioli S Patchkovskii J Frenzel A M Koester P Calaminici S Escalante H A Duarte R
Flores G Geudtner A Goursot J U Reveles A Vela D R Salahub deMon deMonnano edn Mexico DF
Mexico and Bremen (Germany) 2009 b) A F Oliveira G Seifert T Heine H A Duarte J Braz Chem Soc
2009 20 1193 c) G Seifert D Porezag T Frauenheim Int J Quantum Chem 1996 58 185 d) L Zhechkov T
Heine S Patchkovskii G Seifert H A Duarte J Chem Theory Comput 2005 1 841
[7] a) httpwwwccdccamacukproductsmercury b) C F Macrae P R Edgington P McCabe E Pidcock
G P Shields R Taylor M Towler J Van de Streek J Appl Crystallogr 2006 39 453
[8] R Dovesi V R Saunders C Roetti R Orlando C M Zicovich- Wilson F Pascale B Civalleri K Doll N M
Harrison I J Bush P DrsquoArco M Llunell Crystal 102 ed
[9] a) httpwwwscmcom ADF200901 Amsterdam The Netherlands b) G te Velde F M Bickelhaupt S J
A van Gisbergen C Fonseca Guerra E J Baerends J G Snijders T Ziegler J Comput Chem 2001 22 931
[10] J P Perdew K Burke M Ernzerhof Phys Rev Lett 1996 77 3865
[11] E Van Lenthe E J Baerends J Comput Chem 2003 24 1142
[12] H Furukawa O M Yaghi J Am Chem Soc 2009 131 8875
[13] N L Campbell R Clowes L K Ritchie A I Cooper Chem Chem Mater 2009 21 204
105
Appendix D
On the reticular construction concept of covalent organic frameworks
Binit Lukose Agnieszka Kuc Johannes Frenzel and Thomas Heine
Beilstein J Nanotechnol 2010 1 60ndash70
DOI103762bjnano18
Abstract
The concept of reticular chemistry is investigated to explore the applicability of the formation of
Covalent Organic Frameworks (COFs) from their defined individual building blocks Thus we have
designed optimized and investigated a set of reported and hypothetical 2D COFs using Density
Functional Theory (DFT) and the related Density Functional based tight-binding (DFTB) method
Linear trigonal and hexagonal building blocks have been selected for designing hexagonal COF layers
High-symmetry AA and AB stackings are considered as well as low-symmetry serrated and inclined
stackings of the layers The latter ones are only slightly modified compared to the high-symmetry
forms but show higher energetic stability Experimental XRD patterns found in literature also
support stackings with highest formation energies All stacking forms vary in their interlayer
separations and band gaps however their electronic densities of states (DOS) are similar and not
significantly different from that of a monolayer The band gaps are found to be in the range of 17ndash
40 eV COFs built of building blocks with a greater number of aromatic rings have smaller band gaps
Introduction
In the past decade considerable research efforts have been expended on nanoporous materials due
to their excellent properties for many applications such as gas storage and sieving catalysis
106
selectivity sensoring and filtration [1] In 1994 Yaghi and co-workers introduced ways to synthesize
extended structures by design This new discipline is known as reticular chemistry [23] which uses
well-defined building blocks to create extended crystalline structures The feasibility of the building
block approach and the geometry preservation throughout the assembly process are the key factors
that lead to the design and synthesis of reticular structures
One of the first families of materials synthesized using reticular chemistry were the so-called Metal-
Organic Frameworks (MOFs) [4] They are composed of metal-oxide connectors which are covalently
bound to organic linkers The coordination versatility of constituent metal ions along with the
functional diversity of organic linker molecules has created immense possibilities The immense
potential of MOFs is facilitated by the fact that all building blocks are inexpensive chemicals and that
the synthesis can be carried out solvothermally MOFs are commercially available and the scale up of
production is continuing Since the discovery of MOFs many other crystalline frameworks have been
synthesized using reticular chemistry such as Metal-Organic Polyhedra (MOP) [5] Zeolite
Imidazolate Frameworks (ZIFs) [6] and Covalent Organic Frameworks (COFs) [7]
In 2005 Cocircteacute and co-workers introduced COF materials [7-14] where organic linker molecules are
stitched together by covalent entities including boron and oxygen atoms to form a regular
framework These materials have the distinct advantage that all framework bonds represent strong
covalent interactions and that they are composed of light-weight elements only which lead to a very
low mass density [7-9] These materials can be synthesized by wet-chemical methods by
condensation reactions and are composed of inexpensive and non-toxic building blocks so their
large-scale industrial application appears to be possible From a topological viewpoint we distinguish
two- and three-dimensional COFs In two-dimensional (2D) COFs the covalently bound framework is
restricted to layers The crystal is then similar as in graphite composed of a stack of layers which
are not connected by covalent bonds
COFs compared with MOFs have lower mass densities due to the absence of heavy atoms and
therefore might be better for many applications For example the gravimetric uptake of gases is
almost twice as large as that of MOFs with comparable surface areas [1516] Because COFs are fairly
new materials many of their properties and applications are still to be explored
Recently we have studied the structures of experimentally well-known 2D COFs [17] We have found
that commonly accepted 2D structures with AA and AB kinds of layering are energetically less stable
than inclined and serrated forms This is because AA stacking maximises the Coulomb repulsion due
to the close vicinity of charge carrying atoms alike (O B atoms) in neighbouring layers The serrated
and inclined forms are only slightly modified (layers are shifted with respect to each other by asymp14 Aring)
107
and experience less Coulomb forces between the layers compared to AA stacking This is equivalent
to the energetic preference of graphite for an AB (Bernal) over an AA form (simple hexagonal) if we
ignore the fact that interlayer ordering in serrated and inclined forms are not uniform everywhere A
known example of this is that in eclipsed hexagonal boron nitride (h-BN) boron atoms in one layer
serve as nearest neighbours to nitrogen atoms in adjacent layers (AB stacking) The Coulomb
interaction rules out possible interlayer eclipse between atoms with similar charges in this case
In the present work we aim to explore how far the concept of reticular chemistry is applicable to the
individual building units which define COFs For this purpose we have investigated a set of reported
and hypothetical 2D COFs theoretically by exploring their structural energetic and electronic
properties We have compared the properties of the isolated building blocks with those of the
extended crystal structures and have found that the properties of the building units are indeed
maintained after their assembly to a network
Results and Discussion
Structures and nomenclature
We have considered four connectors (IndashIV) and five linkers (andashe) for the systematic design of a
number of 2D COFs (Figure 1) Each COF was built from one type of connector and one type of linker
thus resulting in the design of 20 different structures Moreover we have considered two
hypothetical reference structures which are only built from connectors I and III (no linker is present)
REF-I and REF-III Properties of other COFs were compared with the properties of these two
structures Some of the studied COFs are already well known in the literature [781314] and we
continue to use their traditional nomenclature while hypothetical ones are labelled in the
chronological order with an M at the end which stands for modified
Figurel 1 The connector (IndashIV) and linker (andashe) units considered in this work The same nomenclature is used in the text Carbon ndash green oxygen ndashred boron ndash magenta hydrogen ndash white
108
Using the secondary building unit (SBU) approach we can distinguish the connectors between
trigonal [T] (connectors I II III) and hexagonal [H] (connector IV) and the linkers between linear [l]
(linkers a b e) and trigonal [t] (linkers c d) Topology of the layer is determined by the geometries
of the connector and linker molecules and typically a hexagonal pattern is formed due to the D3h
symmetry of the connector moieties Based on these topologies of the constituent building blocks
we have classified the studied COFs into four groups Tl Tt Hl and Ht (Figure 2) Hereafter we will
use this nomenclature to describe the COF topologies
Figurel 2 Topologies of 2D COFs considered in this work (from the left) Tl Tt Hl and Ht Red and blue blocks are secondary building units corresponding to connectors and linkers respectively
We have considered high-symmetry AA and AB kinds of stacking (hexagonal) and low-symmetry
serrated (orthorhombic) and inclined (monoclinic) kinds of stacking of the layers The latter two were
discussed in a previous work on 2D COFs [17] As an example the structure of COF-5 [7] in different
kinds of stacking of layers is shown in Figure 3 In eclipsed AA stacking atoms of adjacent layers lie
directly on top of each other whilst in staggered AB stacking three-connected vertices lie directly on
top of the geometric center of six-membered rings of neighbouring layers In both serrated and
inclined kinds of stacking the layers are shifted with respect to each other by approximately 14 Aring
resulting in hexagonal rings in the connector or linker being staggered with those in the adjacent
layers In serrated stacking alternate layers are eclipsed In inclined stacking layers lie shifted along
one direction and the lattice vector pointing out of the 2D plane is not rectangular For COFs made of
connector I due to the absence of five-membered C2O2B rings a zigzag shift leads to staggering in
both connector and linker parts For those made of other connectors staggering at the connector or
linker depends on whether the shift is armchair or zigzag respectively [17]
Structural properties
We have compared structural properties of isolated building blocks with those of the extended COF
structures Negligible differences have been found in the bond lengths and bond angles of the
building blocks and the corresponding crystal structures This indicates that the structural integrity of
the building blocks remains unchanged while forming covalent organic frameworks confirming the
109
principle of reticular chemistry In addition the CndashB BndashO and OndashC bond lengths are almost the same
when different COF structures are compared (see Table S1 in Supporting Information File 1) The
experimental bond lengths are asymp154 Aring for CndashB asymp138 Aring for OndashC and 137ndash148 Aring for BndashO However
in the case of COF-1 the experimental values are slightly larger (160 Aring for CndashB and 151 Aring for BndashO)
This could be the reason why our calculated bond lengths for COF-1 are much shorter than the
experimental values while all the other structures agree within 5 error The calculated CndashC bond
lengths vary in the range from 136ndash147 Aring (Figure S2 in Supporting Information File 1) and are the
same for the COFs and their constituent building blocks at the respective positions of the carbon
atoms In addition the reference structures REF-I and REF-III have direct BndashB bond lengths of 167 Aring
and 166 Aring respectively which is shorter by 014 Aring than a typical BndashB bond length The calculated
bond angles OBO in B3O3 and C2O2B rings are 120deg and 113deg respectively
Figure 3 Layer stackings considered AA AB serrated and inclined
Interlayer distances (d) which is the shortest distance between two layers (equivalent to c for AA
c2 for AB and serrated) are different in all kinds of stacking AB stacked 2D COFs have shorter
interlayer distances than the corresponding AA serrated and inclined stacked structures Among the
latter three AA stacked COFs have higher values for d because of the higher repulsive interlayer
orbital interactions resulting from the direct overlap of polarized alike atoms between the adjacent
layers This results in higher mass densities for AB stacked COF analogues Serrated and inclined
stacks have only slightly higher mass densities compared to AA The differences in mass densities for
all kinds of stacking are attributed to the differences in their interlayer separations The d values of
most of the COFs are larger than that of graphite in AA stacking but smaller in AB stacking
Cell parameters (a) and mass densities (ρ) of all the COFs constructed from the considered
connectors and linkers are shown in Table 1 (and in Table S3 in Supporting Information File 1) Mass
densities of all the COFs are much lower than that of graphite (227 gmiddotcmminus3) and diamond (350
gmiddotcmminus3) AAserratedinclined stacked COF-10s have the lowest mass densities (045046046
gmiddotcmminus3) which is lower than that of MOF-5 (059 gmiddotcmminus3) [4] and comparable to that of highly porous
MOF-177 (042 gmiddotcmminus3) [18]
110
In order to identify the stacking orders we have analyzed X-ray diffraction (XRD) patterns of the well-
known COFs (COF-10 TP COF PPy-COF see Figure 4) in all the above discussed stacking kinds The
change of stacking should be visible in XRDs because each space group has a distinct set of symmetry
imposed reflection conditions The XRD patterns of AA serrated and inclined stacking kinds which
differ within a slight shift of adjacent layers to specific directions are in agreement with the presently
available experimental data [81314] Their peaks are at the same angles as in the experimental
spectrum whereas AB stacking clearly shows differences The slight differences in the (001) angle
between each stacking resemble the differences in their interlayer separations The inclined
stackings have more peaks however they are covered by the broad peaks in the experimental
patterns Similar results for COF-1 COF-5 COF-6 and COF-8 have been discussed in our previous
work [17]
Figure 4 The calculated and experimental [81314] XRD patterns of PPy-COF (top) COF-10 (middle) and TP COF (bottom)
111
Table 1 The calculated unit cell parameter a [Aring] interlayer distance d [Aring] and mass density ρ [gmiddotcmminus3
] for AA and AB stacked COFs Note that the cell parameter a is the same for all stacking types Experimental data [719] is given in parentheses
COF Building
Blocks
a d ρ
AA AB AA AB
COF-1 I-a 1502 (15620) 351 313 (332) 094 106
COF-1M I-b 2241 349 304 068 078
COF-2M I-c 1492 347 312 095 106
COF-3M I-d 0747 349 327 153 164
PPy-COF I-e 2232 (22163) 349 (3421) 297 084 099
COF-5 II-a 3014 (30020) 347 (3460) 326 056 060
COF-10 II-b 3758 (37810) 347 (3476) 318 045 (045) 050
COF-8 II-c 2251 (22733) 346 (3476) 320 071 (070) 077
COF-6 II-d 1505 (15091) 346 (3599) 327 104 110
TP COF II-e 3750 (37541) 348 (3378) 320 051 056
COF-4M III-a 2171 350 318 073 080
COF-5M III-b 2915 350 318 055 061
COF-6M III-c 1833 345 318 083 090
COF-7M III-d 1083 350 330 129 136
TP COF-1M III-e 2905 349 310 065 074
COF-8M IV-a 1748 359 329 140 148
COF-9M IV-b 2176 349 330 117 174
COF-10M IV-c 2254 342 336 127 128
COF-11M IV-d 1512 346 338 168 172
TP COF-2M IV-e 2173 347 332 134 140
REF-I I 0773 359 336 144 148
REF-III III 1445 353 336 104 121
Graphite 247 343 335 220 227
112
Energetic stability
We have considered dehydration reactions the basis of experimental COF synthesis to calculate
formation energies of COFs in order to predict their energetic stability Molecular units 14-
phenylenediboronic acid (BDBA) 11rsquo-biphenyl]-44-diylboronic acid (BPDA) 5rsquo-(4-boronophenyl)-
11rsquo3rsquo1rdquo-terphenyl]-44rdquo-diboronic acid (BTPA) benzene-135-triyltriboronic acid (BTBA) and
pyrene-27-diylboronic acid (PDBA) were considered as linkers andashe respectively with -B(OH)2 groups
attached to each point of extension (Figure 5) Self-condensation of these building blocks result in
the formation of B3O3 rings and the resultant COFs are those made of connector I and the
corresponding linkers This process requires a release of three or six water molecules in case of t or l
topology respectively
Figure 5 The reactants participating in the formation of 2D COFs
Co-condensation of the above molecular units with compounds such as 23671011-
hexahydroxytriphenylene (HHTP) hexahydroxybenzene (HHB) and dodecahydroxycoronene (DHC)
(Figure 5) gives rise to COFs made of connectors II III and IV respectively and the corresponding
linkers Self-condensation of tetrahydroxydiborane (THDB) and co-condensation of HHB with THDB
result in the formation of the reference structures REF-I and REF-III respectively In relation to the
corresponding connectorlinker topologies these building blocks satisfy the following equations of
the co-condensation reaction for COF formation
2 2 3 COF 12 H O Tl T l (1)
113
2 1 1 COF 6 H O Tt T t (2)
2 1 3 COF 12 H O Hl H l (3)
2 1 2 COF 12 H O Ht H t (4)
with a stochiometry appropriate for one unit cell The number of covalent bonds formed between
the building blocks is equivalent to the number of released water molecules we refer to it as
ldquoformula unitrdquo and will give all energies in the following in kJmiddotmolminus1 per formula unit
Table 2 The calculated energies [kJ molminus1
] per bond formed between building blocks for AA and AB stacked COFs Ecb is the condensation energy Esb is the stacking energy and Efb is the COF formation energy (Efb = Ecb
+ Esb) The calculated band gaps Δ eV+ are given as well
COF Building
Blocks
Mono-
layer
AA AB
Ecb Esb Efb ∆ Esb Efb ∆
COF-1 I-a 906 -2683 -1777 33 -2187 -1280 36
COF-1M I-b 949 -4266 -3366 27 -2418 -1469 31
COF-2M I-c 956 -5727 -4771 28 -4734 -3778 30
COF-3M I-d 763 -2506 -1742 38 -2801 -2037 40
PPy-COF I-e 858 -5723 -4866 24 -3855 -2998 26
COF-5 II-a 211 -2968 -2756 24 -2548 -2337 28
COF-10 II-b 317 -3766 -3448 23 -1344 -1026 26
COF-8 II-c 263 -4488 -4224 25 -2477 -2213 28
COF-6 II-d 185 -2881 -2695 28 -2127 -1942 31
TP COF II-e 231 -4453 -4222 24 -1480 -1250 27
COF-4M III-a -033 -1730 -1763 26 -880 -913 26
COF-5M III-b 007 -2533 -2526 25 -972 -965 25
COF-6M III-c 014 -3231 -3217 26 -2134 -2120 28
114
COF-7M III-d -170 -1635 -1805 30 -1607 -1777 32
TP COF-1M III-e -014 -3226 -3240 24 -1277 -1291 24
COF-8M IV-a -787 -2756 -3543 18 -2680 -3467 21
COF-9M IV-b -836 -3577 -4414 17 -3003 -3839 21
COF-10M IV-c -947 -4297 -5244 18 -4192 -5140 22
COF-11M IV-d -403 -2684 -3087 21 -2833 -3236 24
TP COF-2M IV-e 030 -4345 -4315 18 -4117 -4087 21
We have calculated the condensation energy of a single COF layer formed from monomers (building
blocks hereafter called bb) according to the above reactions as follows
tot tot tot totcb m H2O 1 bb1 2 bb2E E n E ndash m E m E (5)
where Emtot ndash total energy of the monolayer EH2O
tot is the total energy of the water molecule Ebb1tot
and Ebb2tot are the total energies of interacting building blocks and n m1 m2 are the corresponding
stoichiometry numbers
We have also calculated the stacking energy Esb of layers
tot totsb nl s mE E n E (6)
where Enltot is the total energy of ns number of layers stacked in a COF Finally the COF formation
energy can be given as a sum of Ecb and Esb (see Table 1 and Table S1)
Electronic properties
All COFs including the reference structures are semiconductors with their band gaps lying between
17 eV and 40 eV (Table 2 and Table S4 in Supporting Information File 1) The largest band gaps are
of the reference structures while the lowest values are of COFs built from connector IV The band
gaps are different for different stacking kinds This difference can be attributed to the different
optimized interlayer distances Generally AB serrated and inclined stacked COFs have band gaps
comparable to or larger than that of their AA stacked analogues
115
We have calculated the electronic total density of states (TDOS) and the individual atomic
contributions (partial density of states PDOS) The energy state distributions of COFs and their
monolayers are studied and a comparison for COF-5 is shown in Figure 6 In all stacking kinds
negligible differences are found for the densities at the top of valence band and the bottom of
conduction band These slight differences suggest that the weak interaction between the layers or
the overlap of π-orbitals does not affect the electronic structure of COFs significantly Hence there is
almost no difference between the TDOS of AA AB serrated and inclined stacking kinds Therefore in
the following we discuss only the AA stacked structures
Figure 6 Total densities of states (DOS) (black) of AA (top left) AB (top right) serrated (bottom left) and inclined (bottom right) comparing stacked COF-5 with a monolayer (red) of COF-5 The differences between the TDOS of bulk and monolayer structures are indicated in green The Fermi level EF is shifted to zero
Figure 7 Partial density of states of carbon (left) boron (center) and oxygen (right) atoms of COFs built from type-I connectors and different linkers The vertical dashed line in each figure indicates the Fermi level EF
116
It is of interest to see how the properties of COFs change depending on their composition of different
secondary building units that is for different connectors and linkers PDOS of COFs built from type-I
connectors and different linkers are plotted in Figure 7 The PDOS of carbon atoms is compared with
that of graphite (AA-stacking) while PDOS of boron and oxygen atoms are compared with that of
REF-I a structure which is composed solely of connector building blocks Going from top to bottom
of the plots the number of carbon atoms per unit cell increases It can be seen that this causes a
decrease of the band gap Figure 8 shows the PDOS of COFs built from type-a linkers and different
connectors where the COFs are arranged in the increasing number of carbon atoms in their unit cells
from top to the bottom Again the C PDOS is compared with that of graphite while both REF-I and
REF- III are taken in comparison to O and B PDOS The observed relation between number of carbon
atoms and band gap is verified
Figure 8 Partial density of states of carbon (left) boron (center) and oxygen (right) atoms of COFs built from type-a linkers The vertical dashed line in each figure indicates the Fermi level EF
Conclusion
In summary we have designed 2D COFs of various topologies by connecting building blocks of
different connectivity and performed DFTB calculations to understand their structural energetic and
electronic properties We have studied each COF in high-symmetry AA and AB as well as low-
symmetry inclined and serrated stacking forms The optimized COF structures exhibit different
interlayer separations and band gaps in different kinds of layer stackings however the density of
states of a single layer is not significantly different from that of a multilayer The alternate shifted
layers in AB serrated and inclined stackings cause less repulsive orbital interactions within the layers
which result in shorter interlayer separation compared to AA stacking All the studied COFs show
117
semiconductor-like band gaps We also have observed that larger number of carbon atoms in the
unit cells in COFs causes smaller band gaps and vice versa Energetic studies reveal that the studied
structures are stable however notable difference in the layer stacking is observed from
experimental observations This result is also supported by simulated XRDs
Methods
We have optimized the atomic positions and the lattice parameters for all studied COFs All
calculations were carried out at the Density Functional Tight-Binding (DFTB) [2021] level of theory
DFTB is based on a second-order expansion of the KohnndashSham total energy in the Density-Functional
Theory (DFT) with respect to charge density fluctuations This can be considered as a non-orthogonal
tight-binding method parameterized from DFT which does not require large amounts of empirical
parameters however maintains all the qualities of DFT The main idea behind this method is to
describe the Hamiltonian eigenstates with an atomic-like basis set and replace the Hamiltonian with
a parameterized Hamiltonian matrix whose elements depend only on the inter-nuclear distances and
orbital symmetries [21] While the Hamiltonian matrix elements are calculated using atomic
reference densities the remaining terms to the KohnndashSham energy are parameterized from DFT
reference calculations of a few reference molecules per atom pair The accuracy is very much
improved by the self-consistent charge (SCC) extension Two computational codes were used
deMonNano code [22] and DFTB+ code [23] The first code has dispersion correction [24]
implemented to account for weak interactions and was used to obtain the layered bulk structure of
COFs and their formation energies The performance for interlayer interactions has been tested
previously for graphite [24] The second code which can perform calculations using k-points was
used to calculate the electronic properties (band structure and density of states) Band gaps have
been calculated as an additional stability indicator While these quantities are typically strongly
underestimated in standard LDA- and GGA-DFT calculations they are typically in the correct range
within the DFTB method For validation of our method we have calculated some of the structures
using Density Functional Theory (DFT) as implemented in ADF code [2526]
Periodic boundary conditions were used to represent frameworks of the crystalline solid state The
conjugatendashgradient scheme was chosen for the geometry optimization The atomic force tolerance of
3 times 10minus4 eVAring was applied The optimization using Γ-point approximation was performed with the
deMon-Nano code on 2times2times4 supercells Some of the monolayers were also optimized using the
DFTB+ code on elementary unit cells in order to validate the calculations within the Γ-point
approximation The number of k-points has been determined by reaching convergence for the total
energy as a function of k-points according to the scheme proposed by Monkhorst and Pack [27]
118
Band structures were computed along lines between high symmetry points of the Brillouin zone with
50 k-points each along each line XRD patterns have been simulated using Mercury software [2829]
We have also performed first-principles DFT calculations at the PBE [30] DZP [31] level to support
our results quantitatively For simplicity we have used finite structures instead of bulk crystals
Supporting Information
Table S1 The calculated bond lengths and angles of all studied COFs Corresponding experimental values found from the literature are shown in brackets
COF Building
Blocks
C-B B-O O-C OBO
COF-1 I-a 1498 (1600) 1393 (1509) 1203 (1200)
COF-1M I-b 1497 1393 1203
COF-2M I-c 1497 1392 1203
COF-3M I-d 1496 1392 1201
PPy-COF I-e 1498 1393 1202 (1190)
COF-5 II-a 1496 (1530) 1399 (1367) 1443 (1384) 1135dagger (1139)
COF-10 II-b 1495 (1553) 1399 (1465) 1443 (1385) 1135dagger (1120)
COF-8 II-c 1495 (1548) 1399 (1479) 1443 1135dagger
COF-6 II-d 1496 1399 1443 1135dagger
TP COF II-e 1496 (1542) 1399 (1396) 1444 (1377) 1135dagger (1182)
COF-4M III-a 1496 1398 1449 1135dagger
COF-5M III-b 1496 1398 1449 1136dagger
COF-6M III-c 1496 1399 1451 1134dagger
COF-7M III-d 1496 1398 1449 1136dagger
TP COF-1M III-e 1496 1398 1450 1136dagger
COF-8M IV-a 1496 1398 1445 1131dagger
COF-9M IV-b 1495 1398 1444 1131dagger
119
COF-10M IV-c 1495 1391 1418 1126dagger
COF-11M IV-d 1498 1399 1450 1134dagger
TP COF-2M IV-e 1499 1399 1447 1134dagger
B3O3 connectivity dagger C2B2O connectivity
It can be noticed that the bond lengths of experimentally known COF-1 is much larger compared to
our optimized bond lengths as well as that of other synthesized COFs
Figure S2 Calculated C-C bond lengths in connectors and linkers Hydrogen atoms are omitted for clarity
Table S3 The calculated unit cell parameters a interlayer distance d Aring+ and mass density ρ gcm-3
] for serrated (Sa serrated armchair Sz serrated zigzag) and inclined (Ia inclined armchair Iz inclined zigzag) stacked COFs
COF Building
Blocks
a d ρ
Sa Sz Ia Iz Sa Sz Ia Iz
COF-1 I-a 1502 343 343 097 097
COF-1M I-b 2241 341 342 069 069
COF-2M I-c 1492 340 339 097 097
COF-3M I-d 0747 341 342 157 156
PPy-COF I-e 2232 341 341 086 086
120
COF-5 II-a 3014 342 342 341 340 057 057 058 058
COF-10 II-b 3758 341 341 342 340 046 046 046 046
COF-8 II-c 2251 341 341 342 342 073 073 072 072
COF-6 II-d 1505 342 341 340 340 105 106 106 106
TP COF II-e 3750 342 341 342 342 052 052 052 052
COF-4M III-a 2171 344 344 345 344 074 074 074 074
COF-5M III-b 2915 343 342 343 343 056 056 056 056
COF-6M III-c 1833 341 341 342 341 084 084 084 084
COF-7M III-d 1083 344 343 340 344 131 131 132 131
TP COF-1M III-e 2905 343 342 343 342 066 067 066 066
COF-8M IV-a 1748 341 341 342 342 142 142 142 142
COF-9M IV-b 2176 341 341 341 342 119 119 119 119
COF-10M IV-c 2254 340 340 340 340 128 128 128 128
COF-11M IV-d 1512 341 341 340 340 171 171 171 171
TP COF-2M IV-e 2173 340 340 340 340 137 137 137 137
REF-I I 0773 349 345 148 15
REF-III III 1445 348 349 106 106
Table S4 The calculated energies [kJ mol-1
] per bond formed between building blocks for serrated (Sa serrated armchair Sz serrated zigzag) and inclined (Ia inclined armchair Iz inclined zigzag) stacked COFs Ecb ndash condensation energy Esb ndash stacking energy and Efb ndash COF formation energy (Efb = Ecb + Esb) The calculated band gaps ∆ eV+ are given as well
COF Sa Sz Ia Iz
Esb Efb Δ Esb Efb Δ Esb Efb Δ Esb Efb Δ
-1 -2810 -1904 36 -2786 -1880 36
-1M -4426 -3477 30 -4389 -3440 30
-2M -5967 -5011 30 -5833 -4877 30
121
-3M -2667 -1904 40 -2591 -1828 40
PPy- -5916 -5058 26 -5865 -5007 26
-5 -3052 -2841 27 -3051 -2840 26 -3275 -3064 26 -3297 -3086 26
-10 -3868 -3551 26 -3869 -3552 25 -3830 -3513 26 -4293 -3976 25
-8 -4615 -4352 27 -4614 -4351 27 -4571 -4308 27 -4573 -4310 27
-6 -2978 -2793 30 -2978 -2793 30 -3117 -2932 30 -3090 -2905 30
TP -4557 -4326 26 -4562 -4331 26 -4511 -4280 26 -4511 -4280 26
-4M -1811 -1844 28 -1813 -1846 28 -1769 -1802 28 -1880 -1913 28
-5M -2626 -2619 27 -2630 -2623 26 -2597 -2590 26 -2600 -2593 26
-6M -3375 -3361 28 -3381 -3367 28 -3487 -3473 28 -3349 -3335 28
-7M -1724 -1894 32 -1726 -1896 31 -2333 -2503 32 -1866 -2036 31
TP -1M -3327 -3341 26 -3334 -3348 26 -3340 -3354 26 -3353 -3367 26
-8M -2897 -3684 21 -2895 -3682 21 -2971 -3758 21 -2983 -3770 21
-9M -3732 -4568 20 -3733 -4569 20 -3684 -4520 20 -3706 -4542 20
-10M -4512 -5459 21 -4513 -5460 21 -4491 -5438 21 -4495 -5442 21
-11M -2831 -3234 24 -2834 -3237 24 -2816 -3219 24 -2830 -3233 24
TP -2M -4500 -4470 20 -4505 -4475 20 -4495 -4465 20 -4496 -4466 20
122
Appendix E
Stability and electronic properties of 3D covalent organic frameworks
Binit Lukose Agnieszka Kuc and Thomas Heine
Prepared for publication
Abstract Covalent Organic Frameworks (COFs) are a class of covalently linked crystalline nanoporous
materials versatile for nanoelectronic and storage applications 3D COFs in particular have very
large pores and low-mass densities Extensive theoretical studies of their energetic and mechanical
stability as well as their electronic properties are discussed in this paper All studied 3D COFs are
energetically stable materials though mechanical stability does not exceed 20 GPa Electronically all
COFs are semiconductors with band gaps depending on the number of sp2 carbon atoms present in
the linkers similar to 3D MOF family
Introduction
Covalent Organic Frameworks (COFs)1-3 comprise an emerging class of crystalline materials that
combines organic functionality with nanoporosity COFs have organic subunits stitched together by
covalent entities including boron carbon nitrogen oxygen or silicon atoms to form periodic
frameworks with the faces and edges of molecular subunits exposed to pores Hence their
applications can range from organic electronics to catalysis to gas storage and sieving4-7 The
properties of COFs extensively depend on their molecular constituents and thus can be tuned by
rational chemical design and synthesis289 Step by step reversible condensation reactions pave the
123
way to accomplish this target Such a reticular approach allows predicting the resulting materials and
leads to long-range ordered crystal structures
Since their first mentioning in the literature13 COFs are under theoretical investigation10-17 mainly for
gas storage applications Methods such as doping18-24 and organic linker functionalization2526 have
been suggested to improve their storage capacities In addition to the moderate pore size and
internal surface area COFs have the privileges of a low-weight material as they are made of light
elements Hence their gravimetric adsorption capacity is remarkably high10 The lowest mass density
ever reported for any crystalline material is that for COF-108 (018 gcm-3) Also their stronger
covalent bonds compared to related Metal-Organic Frameworks (MOFs)27 endorse thermodynamic
strength These genuine qualities of COFs make them attractive for hydrogen storage investigations
Crystallization of linked organic molecules into 2D and 3D forms was achieved in the years 20051 and
20073 respectively by the research group of O M Yaghi Several COFs have been synthesized since
then and some of the 2D COFs has been proven good for electronic or photovoltaic applications428-33
However the growth in this area appears to be slow compared to rapidly developing MOFs albeit
the promising H2 adsorption measurements53435 and a few synthetic improvements736-42
COF-102 and COF-103 were synthesized3 by the self-condensation of the non-planar tetra(4-
dihydroxyborylphenyl)methane (TBPM) and tetra(4-dihydroxyborylphenyl)silane (TBPS) respectively
(see Fig 1 for the building blocks and COF geometries) The co-condensation of these compounds
with triangular hexahydroxytriphenylene (HHTP) results in COF-108 and COF-105 respectively with
different topologies Yaghi et al3 have reported the formation of highly symmetric topologies ctn
(carbon nitride dI 34 ) and bor (boracite mP 34 ) when tetrahedral building blocks are linked
together with triangular ones The topology names were adopted from reticular chemistry structure
resource (RCSR)43 Simulated Powder X-ray diffraction in comparison with experimental powder
spectrum suggested ctn topology for COF-102 103 and 105 and bor topology for COF-108 The
condensation of tert-butylsilanetriol (TBST) and TBPM leads to the formation of COF-20244 This was
reported as ctn topology The COF reactants and schematic diagrams of ctn and bor topologies are
given in Figure1 The assembly of tetrahedral and linear units is very likely to result in a diamond-like
form COF-300 was formed according to the condensation of tetrahedral tetra-(4-anilyl)methane
(TAM) and linear terephthaldehyde (TA)45 However the synthesized structure was 5-fold
interpenetrated dia-c5 topology43
In this work we present theoretical studies of 3D COFs using density functional based methods to
explore their structural electronic energetic and mechanical properties Our previous studies on 2D
COFs4647 had resulted in questioning the stability of eclipsed arrangement of layers (AA stacking) and
124
suggesting energetically more stable serrated and inclined packing In this paper we attempt to
explore the stability and electronic properties of the experimentally known 3D COFs namely COF-
102 103 105 108 202 and 300 In the present studies we follow the reticular assembly of the
molecular units that form low-weight 3D COFs Considering the structural distinction from MOFs
COFs lack the versatility of metal ingredients what results in diverse properties48 A collective study is
then carried out to understand the characteristics and drawbacks of COFs
Figure 1 (Upper panel) Building blocks for the construction of 3D COFs (Lower panel) lsquoctnrsquo and lsquoborrsquo
networks formed by linking tetrahedral and triangular building units
Methods
COF structures were fully optimized using Self-consistent Charge -Density Functional based Tight-
Binding (SCC-DFTB) level of theory4950 Two computational codes were used deMonNano51 and
125
DFTB+52 The first code which has dispersion correction53 implemented to account for weak
interactions was used for the geometry optimization and stability calculations The second code
which can perform calculations using k-point sampling was used to calculate the electronic
properties (band structure and density of states) The number of k-points has been determined by
reaching convergence for the total energy as a function of k-points according to the scheme
proposed by Monkhorst and Pack54 Periodic boundary conditions were used to represent
frameworks of the crystalline solid state Conjugatendashgradient scheme was chosen for the geometry
optimization Atomic force tolerance of 3 times 10minus4 eVAring was applied The optimization using Γ-point
approximation was performed on rectangular supercells containing more than 1000 atoms For
validation of our method we have calculated energetic stability using Density Functional Theory (DFT)
at the PBE55DZP56 level as implemented in ADF code5758 using cluster models The cluster models
contain finite number of building units and correspond to the bulk topology of the COFs XRD
patterns have been simulated using Mercury software5960
In this work we continued to use the traditional nomenclature of the experimentally known COFs All
of the structures have the same tetrahedral blocks and differ only in the central sp3 atom (carbon or
silicon) that is included in our nomenclature
Bulk modulus (B) of a solid at absolute zero can be calculated as
(1) B = 2
2
dV
EdV
where V and E are the volume and energy respectively
Owing to the dehydration reactions we have calculated the formation (condensation) energy of each
COF formed from monomers (building blocks) as follows
(2) EF = Etot + n EH2Otot ndash (m1 Ebb1
tot + m2 Ebb2tot)
where Etot -- total energy of the COF EH2Otot -- total energy of water molecule Ebb1
tot and Ebb2tot -- total
energies of interacting building blocks n m1 m2 -- stoichiometry numbers
Results and Discussions
Structure and Stability
We have optimized the atomic positions and cell dimensions of the COFs in the experimentally
determined topologies Cell parameters in comparison with experimental values are given in Table 1
The calculated bond lengths are C-B = 150 C-C = 139-144 (COF-300) C(sp3)-C = 156 C-O = 142 B-
126
O = 140 Si-C = 188 Si-O = 186 C-N = 131-136 Aring These values agree within 6 error with the
experimental values34445
Mass densities (see Table 1) of COFs range between 02 to 05 g cm-3 and are smaller for the silicon at
the tetrahedral centers This implies that the presence of silicon atoms impacts more on the cell
volume than on the total mass That means the replacement of sp3 C with Si in COF-108 can change
its mass density to a slightly lower value To our best knowledge among all the natural or
synthesized crystals COF-108 has the lowest mass-weight
In order to validate our optimized structures we have simulated X-ray Diffraction of each COF and
compared them with the available experimental spectra (see Figure2) Almost all of the simulated
XRDs have excellent correlation in the peak positions with the experiment Only COF-300 shows
somehow significant differences in the intensities These differences may be attributed to the
presence of guest molecules in the synthesized COF-30045
Table 1 The calculated cell parameters [Aring] mass density ρ gcm-3
+ band gap Δ eV+ bulk modulus B GPa+
and formation energy EF [kJ mol-1
] for all the studied 3D COFs Experimental values are given in brackets
along with the cell parameters HOMO-LUMO gaps [eV] of the building blocks are also given in brackets
along with the band gaps
Structure Building
Blocks
Cell
parameters
ρ Δ B EF
COF-102 (C) TBPM 2707 (2717) 043 39 (43) 206 -14995
COF-103 (Si) TBPS 2817 (2825) 039 38 (41) 139 -14547
COF-105-C TBPM HHTP 4337 019 33 (43 34) 80 -18080
COF-105 (Si) TBPS HHTP 4444 (4489) 018 32 (41 34) 79 -17055
COF-108 (C) TBPM HHTP 2838 (2840) 017 32 (43 34) 37 -17983
COF-108-Si TBPS HHTP 2920 016 30 (41 34) 29 -18038
COF-202 (C) TBPM TBST 3047 (3010) 050 42 (43 139) 143 -7954
COF-202-Si TBPS TBST 3157 046 39 (41 139) 153 -7632
COF-300 (C) TAM TA 2932 925 049 23 (41 26) 144 -40286
127
(2828 1008)
COF-300-Si TAS TA 3039 959 044 23 (40 26) 133 -39930
tetra-(4-anilyl)silane
Figure 2 Simulated XRDs are compared with the experimental patterns from the literature COF-300
exhibits some differences between the simulated and experimental XRDs while others show reasonably
good match
The bulk modulus (B) is a measure of mechanical stability of the materials The calculated values of B
are given in Table 1 Compared to the force-field based calculation of bulk modulus by Schmid et
al61 these values are larger The bulk modulus of COF-105 and COF-108 are relatively small
compared with other COFs Considering that the two COFs differ only in the topology it may be
concluded that ctn nets are mechanically more stable compared to bor COFs with carbon atoms in
the center are mechanically more stable than those with silicon atoms Bulk modulus of COF-102
103 COF-202 and interpenetrated COF-300 are higher than many of the isoreticular MOFs62 and
comparable to IRMOF-6 (1241 GPa) MOF-5 (1537 GPa)63 bNon-interpenetrated COF-300 (single
framework dia-a topology43) has much lower bulk modulus of only 317 GPa
Calculated formation energies (Table 1) are given in terms of reaction energies according to Eq 2
Condensation of monomers to form bulk 3D structures is exothermic in all the cases supporting
reticular approach The presence of C or Si at the vertex center does not show any particular trend in
the formation energies We have calculated the formation energy of non-interpenetrated COF-300
(dia-a) to be -39346 kJ mol-1 which is comparable to the interpenetrated cases For a quantitative
comparison we have performed DFT calculations at the PBEDZP level as implemented in ADF code
on finite structures The obtained formation energies for COF-105-C COF-105 (Si) COF-108 (C) COF-
108-Si are -9944 -9968 -1245 -12436 respectively They are in a reasonable agreement with the
128
DFTB results A comparison between COF-105 and COF-108 suggests that bor nets are energetically
more favored than ctn nets
Electronic Properties
Band gaps (Δ) calculated for the 3D COFs are in the range of 23-42 eV (see Table 1) which show
their semiconducting nature similar to the hexagonal 2D COFs47 and MOFs62 The band gap
decreases with the increase of conjugated rings in the unit cell relative to the number of other atoms
Si atoms in comparison with carbon atoms at the tetrahedral centers give a relatively smaller Δ This
is evident from the partial density of states of C (sp3) in COF-102 and Si in COF-103 plotted in Figure3
Carbon atoms define the band edges of the PDOS (see Figure4) The band gaps of COF-105 and COF-
108 differ only slightly (see Figure5) which means that the band gap is nearly independent of the
topology This is because for each atom the coordination number and the neighboring atoms remain
the same in both ctn and bor networks albeit the topology difference DOS of non-interpenetrated
(dia-a) and 5-fold interpenetrated (dia-c5) COF-300 are plotted in Figure6 It may be concluded from
their negligible differences that interpenetration does not alter the DOS of a framework We have
shown similar results for 2D COFs47
Figure 3 Partial density of states (DOS) of sp3 C in COF-102 (top) and Si in COF-103 (bottom) The latter is
inverted for comparison The Fermi level EF is shifted to zero
129
Figure 4 Partial and total density of states of COF-103 The Fermi level EF is shifted to zero
Figure 5 Density of states of COF-108 (bor) and COF-105 (ctn) The two COFs differ only in the topology
130
Figure 6 Density of states of non-interpenetrated (dia-c1) and 5-interpenetrated (dia-c5) COF-300
We also have calculated HOMO-LUMO gaps of the molecular building blocks (see Table 1) In
comparison band gaps of COFs are slightly smaller than the smallest of the HOMO-LUMO gaps of the
building units
Conclusion
In summary we have calculated energetic mechanical and electronic properties of all the known 3D
COF using Density Functional Tight-Binding method Energetically formation of 3D COFs is favorable
supporting the reticular chemistry approach Mechanical stability is in line with other frameworks
materials eg MOFs and bulk modulus does not exceed 20 GPa Also all COFs are semiconducting
with band gaps ranging from 2 to 4 eV Band gaps are analogues to the HOMO-LUMO gaps of the
molecular building units We believe that this extensive study will define the place of COFs in the
broad area of nanoporous materials and the information obtained from the work will help to
strategically develop or modify porous materials for the targeted applications
131
Appendix F
Structure electronic structure and hydrogen adsorption capacity of porous aromatic frameworks
Binit Lukose Mohammad Wahiduzzaman Agnieszka Kuc and Thomas Heine
Prepared for publication
Abstract
Diamond-like porous aromatic frameworks (PAFs) form a new family of materials that contain only
carbon and hydrogen atoms within their frameworks These structures have very low mass densities
large surface area and high porosity Density-functional based calculations indicate that crystalline
PAFs have mechanical stability and properties similar to those of Covalent Organic Frameworks Their
exceptional structural properties and stability make PAFs interesting materials for hydrogen storage
Our theoretical investigations show exceptionally high hydrogen uptake at room temperature that
can reach 7 wt a value exceeding those of well-known metal- and covalent-organic frameworks
(MOFs and COFs)
Introduction
Porous materials have been widely investigated in the fields of materials science and technology due
to their applications in many important fields such as catalysis gas storage and separation template
materials molecular sensors etc1-3 Designing porous materials is nowadays a simple and effective
strategy following the approach of reticular chemistry4 where predefined building blocks are used to
132
predict and synthesize a topological organization in an extended crystal structure The most famous
and striking examples of such materials are Metal- and Covalent-Organic Frameworks (MOFs and
COFs)56 These new nanoporous materials have many advantages high porosity and large surface
areas lowest mass densities known for crystalline materials easy functionalization of building blocks
and good adsorption properties
Gas storage and separation by physical adsorption are very important applications of such
nanoporous materials and have been major subjects of science in the last two decades These
applications are based on certain physical properties namely presence of permanent large surface
area and suitable enthalpy of adsorption between the host framework and guest molecules
Attempts to produce materials with large internal surface area have been successful and some of the
notable accomplishments include the synthesis of MOF-2107 (BET surface area of 6240 m2 g-1 and
Langmuir surface area of 10400 m2 g-1) NU-1008 (BET surface area 6143 m2 g-1) or 3D COFs9 (BET
surface area 4210 m2 g-1 for COF-103)
More recently a new family of porous materials emerged So-called porous-aromatic frameworks
(PAFs) have surface areas comparable to MOFs eg PAF-110 has a BET surface area of 5640 m2 g-1 and
Langmuir surface area of 7100 m2 g-1 Since they are composed of carbon and hydrogen only they
have several advantages over frameworks containing heavy elements MOFs with coordination bonds
often suffer from low thermal and hydrothermal stability what might limit their applications on the
industrial scale The coordination bonds can be substituted with stronger covalent bonds as it was
realized in the case of COFs6 however this lowers significantly their surface areas comparing with
MOFs
Besides its exceptional surface area PAF-110 has high thermal and hydrothermal stability and
appears to be a good candidate for hydrogen and carbon dioxide storage applications PAFs have
topology of diamond with C-C bonds replaced by a number of phenyl rings (see Figure 1)
Experimentally obtained PAF-110 and PAF-1111 have eg one and four phenyl rings respectively
connected by tetragonal centers (carbon tetrahedral nodes) The simulated and experimental
hydrogen uptake capacities of such PAFs exceed the DOE target12
In this paper we have studied structural electronic and adsorption properties of PAFs using Density
Functional based Tight-Binding (DFTB) method1314 and Quantized Liquid Density Functional Theory
(QLDFT)15 We have found exceptionally high storage capacities at room temperature what makes
PAFs very good materials for hydrogen storage devices beyond well-known MOFs and COFs We have
compared our results to widely-used classical Grand Canonical Monte-Carlo (GCMC) calculations
reported in the literature We have also studied other properties of these materials such as
133
structural energetic electronic and mechanical We explored the structural variance of diamond
topology by individually placing a selection of organic linkers between carbon nodes This generally
changes surface area mass density and isosteric heat of adsorption what is reflected in the
adsorption isotherms
Methods
Using a conjugate-gradient method we have fully optimized the crystal structures (atomic positions
and unit cell parameters) of PAFs The calculations were performed using Dispersion-corrected Self-
consistent Charge density-functional based tight-binding (DFTB) method as implemented in the
deMonNano code1617 Periodic boundary conditions were applied to a (3x3x3) supercell thus
representing frameworks of the crystalline solid state Electronic density of states (DOS) have been
calculated using the DFTB+ code18 with k-point sampling where the k space was determined by
reaching convergence for the total energy according to the scheme proposed by Monkhorst and
Pack19
Hydrogen adsorption calculations have been carried out using QLDFT within the LIE-0 approximation
thus including many-body interparticle interactions and quantum effects implicitly through the
excess functional The PAF-H2 non-bonded interactions were represented by the classical Morse
atomic-pair potential Force field parameters were taken from Han et al20 who originally developed
them for studying hydrogen adsorption in COFs that incorporate similar building blocks as PAFs The
authors adopted the approach to the shapeless particle approximation of QLDFT These PAF-H2
parameters have been determined by fitting first-principles calculations at the second-order Moslashllerndash
Plesset (MP2) level of theory using the quadruple-ζ QZVPP basis set with correction for the basis set
superposition error (BSSE) QLDFT calculations use a grid spacing of 05 Bohr radii and a potential
cutoff of 5000 K
Results and Discussion
Design and Structure of PAFs
We have designed a set of PAFs by replacing C-C bonds in the diamond lattice by a set of organic
linkers phenyl biphenyl pyrene para-terphenyl perylenetetracarboxylic acid (PTCDA)
diphenylacetylene (DPA) and quaterphenyl (see Figure 1) We label the respective crystal structures
as PAF-phnl (PAF-301 in Ref 12) PAF-biph (PAF-302 in Ref 12) PAF-pyrn PAF-ptph(PAF-303 in Ref
12) PAF-PTCDA PAF-DPA and PAF-qtph (PAF-304 in Ref 12) respectively Such a design of
frameworks should result in materials with high stability due to the parent diamond-topology and
pure covalent bonding of the network The selected linkers differ in their length width and the
134
number of aromatic rings These should play an important role for hydrogen adsorption properties
aromatic systems exhibit large polarizabilities and interact with H2 molecules via London dispersion
forces Long linkers introduce high pore volume and low mas-weight to the network while wide
linkers offer large internal surface area and high heat of adsorption Hence long linkers are of
advantage for high gravimetric capacity while wide linkers enhance volumetric capacity Proper
optimization of the linker size should result in a perfect candidate for hydrogen storage applications
Figure 1 a) Schematic diagram of the topology of PAFs blue spheres and grey pillars represent carbon
tetragonal centers and organic linkers respectively b) Linkers used for the design of PAFs i) phenyl ii)
biphenyl iii) pyrene iv) DPA v) para-Terphenyl vi) PTCDA and vii) quaterphenyl
Selected structural and mechanical properties of the investigated PAF structures are given in Table 1
Frameworks created with the above mentioned linkers have mass densities that range from 085 g
cm-3 (for phenyl) to 01 g cm-3 (for quaterphenyl) The lowest mass-density obtained for a crystal
structure was for COF-108 with = 0171 g cm-39 Our results show that PAF-DPA and PAF-ptph have
mass densities close to the one of COF-108 while PAF-qtph with = 01 g cm-3 exhibits the lowest
for all the PAFs investigated in this study
While the large cell size and the small mass density of PAF-qtph are an advantage for high
gravimetric hydrogen storage capacity other PAFs (eg PAF-pyrn with wide linker) would
compromise gravimetric for high volumetric capacity As both of them are important for practical
applications a balance between them is crucial
Table 1 Calculated cell parameters a (a=b=c α=β=γ=60⁰) mass densities () formation energies (EForm) band
gaps () bulk moduli (B) and H2 accessible free volume and surface area of PAFs studied in the present work
In parenthesis are given HOMU-LUMO gaps of the corresponding saturated linkers
PAFs
a
(Aring)
ρ
(g cm-3)
EForm
(kJ mol-1)
Δ
(eV)
B
(GPa)
H2 accessible
free volume
H2 accessible
surface area
135
() (m2 g-1)
PAF-phnl 97 085 -121 47 (55) 360 35 2398
PAF-biphl 167 032 -122 36 (40) 132 73 5697
PAF-pyrn 166 042 -124 26 (28) 192 66 5090
PAF-DPA 210 019 -122 35 (37) 87 84 7240
PAF-ptph 237 016 -119 32 (33) 56 86 6735
PAF-PTCDA 236 024 -122 18 (19) 95 81 5576
PAF-qtphl 308 010 -119 29 (30) 35 91 7275
Energetic and Mechanical Properties
We have investigated energetic stability of PAFs by calculating their formation energies We regarded
the formation of PAFs as dehydrogenation reaction between saturated linkers and CH4 molecules
For a unit cell containing n carbon nodes and m organic linkers the formation energy (EForm) is given
by
( )
where Ecell EL and
are the total energies of the unit cell saturated linkers CH4 and H2
molecules respectively This excludes the inherent stability of linkers and represents the energy for
coordinating linkers with sp3 carbon atoms during diamond-topology formation All the formation
energies calculated in the present work are given in Table 1 Negative values indicate that the
formation of PAFs is exothermic The values per formula unit do not deviate significantly for different
PAF sizes and shapes
Although diamond is the hardest known material insertion of longer linkers diminishes its
mechanical strength to some extent In order to study the mechanical stability of PAFs we have
calculated their bulk moduli which is the second derivative of the cell energy with respect to the cell
volume (see Table 1) The highest value that we have obtained is for PAF-phnl with B = 36 GPa This is
over ten times smaller than the bulk modulus of pure diamond 514 GPa (calculated at the DFTB
level)21 and 442 GPa (experimental value)22 Bulk modulus should scale as 1V (V - volume) if all
bonds have the same strength We have plotted such a function for PAFs and other framework
136
materials23 in Figure 2 As PAFs have various types of CmdashC bonds there are minor deviations from
the perfect trend
Figure 2 Calculated bulk modulus B as function of the volume per atom PAFs are highlighted in blue and
compared with other framework materials Red curve denotes the fitting to 1V function (V - volume)
The stability of PAF-phnl is similar to that of Cu-BTC MOF and much higher than for other MOFs such
as MOF-5 -177 and -20524 and 3D COFs23 Subsequent addition of phenyl rings decreases B the
lowest value of 35 GPa has been found for PAF-qtph which is in the same order as for COF-108 In
general all the studied PAFs except for PAF-phnl have bulk moduli in line with 3D COFs25 When the
organic linker is extended in the width (cf PAF-biph and PAF-pyrn) the mechanical stability increases
Electronic Properties
All PAF structures are semiconductors similar to COFs The band gaps of PAFs range from 18 to 47
eV (see Table 1) Interesting trends may be observed from the band gaps of this iso-reticular series
In comparison with diamond which has a band gap of 55 eV it may be understood that subsequent
insertion of benzene groups between carbon atoms in diamond decreases the band gap This is easily
understood as the sp3 responsible for the semiconducting character become far apart with large
number of π-electrons in between which tend to close the gap More importantly the values of
band gaps are directly related to the HOMO-LUMO gaps of the corresponding saturated linkers
which are just slightly larger Also more extended linkers (cf PAF-biph and PAF-pyrn or PAF-ptph and
PAF-PTCDA) reduce the band gap
In general conjugated rings reduce the band gap more effectively than CmdashC bond networks (cf PAF-
DPA PAF-ptph or PAF-qtphl) The extreme cases for this behaviour are graphene which is metallic
137
and diamond which is an insulator Overall the band gap may be tuned by placing suitable linkers in
the diamond network Similar results have been reported for MOFs2627
We have calculated the electronic density of states (DOS) for all the PAF structures Figure 3 a shows
carbon DOS of PAFs arranged in such a way that the band gaps decrease going from the top to the
bottom of the figure PAFs with larger band gaps have larger population of energy states at the top of
valence band and bottom of conduction band whereas for linkers with smaller band gaps the
distribution of energy states is rather spread In order to study the impact of sp3 carbon atoms in the
DOS we plotted their contributions against the total carbon DOS in the cases of PAF-phnl and PAF-
pyrn as examples (see Figure 3 b and c) In both cases the sp3 carbons exhibit no energy states in the
band gap and in the close vicinity of band edges
Figure 3 (a) Carbon density of states of all calculated PAFs From top to bottom the value of band gap
decreases (b) and (c) projected DOS on sp3 C atoms of PAF-phnl and PAF-pyrn respectively The vertical
dashed line indicates Fermi level EF
Hydrogen Adsorption Properties
One of the potential applications of PAFs is hydrogen adsorption We have calculated the gravimetric
and volumetric capacities and analyzed them to understand the contributions of the linkers on the
138
hydrogen uptake in different range of temperatures and pressures H2 accessible free volume and
surface area of the PAFs are given in Table 1 In order to assess the excess adsorption capacity the
free pore volume is necessary In our simulation the free pore volume is defined to be that where
the H2-host interaction energy is less than 5000K This covers all sites where H2 is attracted by the
host structure and excludes the repulsion area close to the framework The solvent accessible surface
areas of the PAFs were determined by rolling a probe molecule along the van der Waals spheres of
the atoms A probe molecule of radius 148 Aring was used which is consistent with the Lennard-Jones
sphere of hydrogen and commonly used in various H2 molecular simulations28
Very large surface areas per mass unit have been observed for all PAFs except for PAF-phnl PAF-DPA
and PAF-qtph for example have 7240 and 7274 m2 g-1 H2 accessible surface area respectively For
comparison highly porous MOF-210 and NU-100 give values of 6240 and 6143 m2 g-1 BET surface
areas respectively determined from the experimental adsorption isotherms78 It is worth
mentioning that longer linkers expand the pore and increase the surface area per unit volume and
unit mass Wider linkers provide a higher surface area per unit volume however they possess larger
mass density and hence the surface area per unit mass gets lower
Figure 4 shows the gravimetric and volumetric total and excess adsorption isotherms of PAFs at 77K
The results show that the gravimetric (total and excess) H2 uptake is dependent on the linker length
The longest linker in the PAF-qtph structure gives the total and excess gravimetric capacity of 32 and
128 wt respectively which is larger than for highly porous crystalline 3D COFs29 These numbers
are calculated for the limit pressure of 100 bar For the shortest linker in PAF-phnl we have obtained
only around 25 wt for the same pressure Using grand canonical Monte-Carlo methods (GCMC)
Lan et al12 have predicted total and excess adsorption capacities of PAF-qtph of 224 and 84 wt
respectively The deviations in results are attributed to the differences in both methods where
different force fields are used to describe atom-atom interactions
The volumetric adsorption capacities lower with the size of the linker and the largest uptake we have
found for the PAF-pyrn (46 and 35 kg m-3 total and excess respectively) while very low values were
found for PAF-qtph (30 and 145 kg m-3 total and excess respectively) at 35-bar pressure It could be
predicted that for PAF-phnl the volumetric adsorption reaches the larges values however due to its
very compact crystal structure it reaches saturation at the low-pressure region and does not exceed
30 kg m-3 for the total uptake From the excess adsorption isotherms however PAF-phnl is the best
adsorbing system for low-pressure application as it gets saturated at 11 bar by adsorbing 266 kg m-3
of H2 Generally we have observed that PAFs with lower pore sizes exhibited saturation of volumetric
capacities at lower pressures
139
Figure 4 Calculated H2 uptake of PAFs at 77 K (a)-(b) gravimetric and (c)-(d) volumetric total (upper panel)
and excess (lower panel) respectively
We have also calculated the adsorption performance of PAFs at room temperature The gravimetric
total adsorption isotherms are shown in Figure 5 Similar to the results obtained for T=77 K PAF-
qtph exhibits the highest total gravimetric capacity at room temperature which is as high as 732 wt
at 100 bar Lan et al12 have predicted the uptake of 653 wt at 100 bar using GCMC simulations
These results exceed the 2015 DOE target of 55 wt 3031 On the other hand at more applicable
pressure of 5 bar the uptake reaches only 05 wt This suggests however that the delivery amount
(approximately the difference between the uptakes at 100 and 5 bar) is still more than the DOE
target Phenyl linker at room temperature does not exceed the gravimetric uptake of 1 wt at 100
bar
Figure 5 Calculated gravimetric total (left) and excess (right) H2 uptake of PAFs at 300 K
140
At 300K excess gravimetric capacity is rather low and does not exceed 075 wt even for large
pressure (see Figure 5)
Effects of interpenetration
Very often frameworks with large pores crystallize as interpenetrated lattices In most cases this is
an undesired fact due to reduction of the pore size and free volume For instance COF-300 which
has diamond topology was found to have 5-interpenetrated frameworks32
We have studied the effect of interpenetration in the case of PAF-qtph which has the largest pore
volume among the materials in this study Without any steric hindrance PAF-qtph may be
interpenetrated up to the order of four The two three and four interpenetrated networks are
named as PAF-qtph-2 PAF-qtph-3 and PAF-qtph-4 respectively and in each case the interpenetrated
structures possess translational symmetry Table 2 shows calculated mass densities and H2 accessible
free volume and surface area of non-interpenetrated and n-fold penetrated PAF-qtph Obviously the
mass density increases by one unit for each degree of interpenetration PAF-qtph has 91 of its
volume accessible for H2 which means that 9 of volume is occupied by the skeleton of the PAF
Hence each degree of interpenetration decreases the free volume by 9 H2 accessible surface area
per unit mass remains the same up to 3-fold interpenetration However PAF-qtph-4 results in much
less accessibility for H2
Table 2 Calculated mass densities () and H2 accessible free volume and surface area of non-interpenetrated
and n-fold interpenetrated PAF-qtph where n = 2 3 4
PAF
(g cm-3)
H2 accessible
free volume ()
H2 accessible
surface area
(m2 g-1)
PAF-qtph 010 91 7275
PAF-qtph-2 020 82 7275
PAF-qtph-3 030 73 7275
PAF-qtph-4 040 64 5998
Figure 6 shows the excess and total adsorption isotherms (gravimetric and volumetric) of non-
interpenetrated and interpenetrated PAF-qtph at 77 K With the increasing degree of
141
interpenetration saturation occurs at lower values of pressure This is due to the smaller pore size
resulted from the high-fold interpenetration The excess gravimetric capacity decreases by 18 - 2 wt
per each n-fold interpenetration a result of the decreasing free space around the skeleton It is to be
noted that the difference between the total and the excess adsorption capacities of PAF-qtph is quite
large however it decreases less for interpenetrated structures This is because the interpenetrated
frameworks occupy the free space that otherwise would be taken by hydrogen to increase the total
capacity but not the excess
Figure 6 Calculated H2 uptake at 77 K of non-interpenetrated and n-fold interpenetrated PAF-qtph with n = 2
3 4 (a)-(b) gravimetric and (c)-(d) volumetric total (lower panel) and excess (upper panel) respectively
On the other hand the interpenetrated frameworks have larger volumetric capacity what is easily
understandable due to the volume reduction Significant increase in excess volumetric capacity has
been obtained for PAF-qtph-2 in comparison with PAF-qtph whereas only slightly lower increase was
obtained for PAF-qtph-3 in comparison with PAF-qtph-2 (see Figure 6) PAF-qtph-4 exhibited even
lower increase compared to PAF-qtph-3 suggesting that only a certain degree of interpenetration is
appreciable Although non-interpenetrated PAF has a large amount of H2 gas in the bulk phase due
to the high pore volume its total volumetric uptake does not exceed that of the interpenetrated
PAFs
Almost linear gravimetric isotherms were obtained at room temperature for all studied PAFs
including the interpenetrated cases (see Figure 7) Large decrease of gravimetric capacity was noted
142
when the PAF was interpenetrated from around 7 wt down to 2 wt for 4-fold interpenetrated
PAF-qtph The excess gravimetric capacity however is the same independent of the n-fold
interpenetration and maximum of 06 wt was found for 100-bar pressure (see Figure 7)
Figure 7 Calculated gravimetric total (left) and excess (right) H2 uptake of non-interpenetrated and n-fold
interpenetrated PAF-qtph (n = 2 3 4) at 300 K
Conclusions
Following diamond-topology we have designed a set of porous aromatic frameworks PAFs by
replacing CmdashC bonds by various organic linkers what resulted in remarkably high surface area and
pore volume
Exothermic formation energies of PAFs calculated from the dehydrogenation of the linkers with CH4
indicate strong coordination of linkers in the diamond topology The studied PAFs exhibit bulk moduli
that are much smaller than diamond however in the same order as other porous frameworks such
as MOFs and COFs PAFs are semi-conductors with their band gaps dependent on the HOMO-LUMO
gaps of the linking molecules
Using quantized liquid density functional theory which takes into account inter-particle interactions
and quantum effects we have predicted hydrogen uptake capacities in PAFs At room temperature
and 100 bar the predicted uptake in PAF-qtph was 732 wt which exceeded the 2015 DOE target
At the same conditions other PAFs showed significantly lower uptakes At 77 K the PAFs with similar
pore sizes exhibited similar gravimetric uptake capacities whereas smaller pore size and larger
number of aromatic rings result in higher volumetric capacities In smaller pores saturation of excess
capacities occurred at lower pressures Interpenetrated frameworks largely decrease the amount of
hydrogen gas in the pores and increase the weight of the material however they are predicted to
have high volumetric capacities
143
References
(1) Eddaoudi M Moler D B Li H L Chen B L Reineke T M OKeeffe M Yaghi O M
Accounts of Chemical Research 2001 34 319
(2) Seo J S Whang D Lee H Jun S I Oh J Jeon Y J Kim K Nature 2000 404 982
(3) Ferey G Mellot-Draznieks C Serre C Millange F Accounts of Chemical Research 2005 38
217
(4) Yaghi O M OKeeffe M Ockwig N W Chae H K Eddaoudi M Kim J Nature 2003 423
705
(5) Eddaoudi M Kim J Rosi N Vodak D Wachter J OKeeffe M Yaghi O M Science 2002
295 469
(6) Cote A P Benin A I Ockwig N W OKeeffe M Matzger A J Yaghi O M Science 2005
310 1166
(7) Furukawa H Ko N Go Y B Aratani N Choi S B Choi E Yazaydin A O Snurr R Q
OKeeffe M Kim J Yaghi O M Science 2010 329 424
(8) Farha O K Yazaydin A O Eryazici I Malliakas C D Hauser B G Kanatzidis M G
Nguyen S T Snurr R Q Hupp J T Nature Chemistry 2010 2 944
(9) El-Kaderi H M Hunt J R Mendoza-Cortes J L Cote A P Taylor R E OKeeffe M Yaghi
O M Science 2007 316 268
(10) Ben T Ren H Ma S Cao D Lan J Jing X Wang W Xu J Deng F Simmons J M Qiu
S Zhu G Angewandte Chemie-International Edition 2009 48 9457
(11) Yuan Y Sun F Ren H Jing X Wang W Ma H Zhao H Zhu G Journal of Materials
Chemistry 2011 21 13498
(12) Lan J Cao D Wang W Ben T Zhu G Journal of Physical Chemistry Letters 2010 1 978
(13) Oliveira A F Seifert G Heine T Duarte H A Journal of the Brazilian Chemical Society
2009 20 1193
(14) Seifert G Porezag D Frauenheim T International Journal of Quantum Chemistry 1996 58
185
(15) Patchkovskii S Heine T Physical Review E 2009 80
(16) Heine T Rapacioli M Patchkovskii S Frenzel J Koester A M Calaminici P Escalante S
Duarte H A Flores R Geudtner G Goursot A Reveles J U Vela A Salahub D R deMon-nano edn ed
deMon 2009
(17) Zhechkov L Heine T Patchkovskii S Seifert G Duarte H A Journal of Chemical Theory
and Computation 2005 1 841
(18) BCCMS Bremen DFTB+ - Density Functional based Tight binding (and more)
(19) Monkhorst H J Pack J D Physical Review B 1976 13 5188
(20) Han S S Furukawa H Yaghi O M Goddard III W A Journal of the American Chemical
Society 2008 130 11580
(21) Kuc A Seifert G Physical Review B 2006 74
(22) Cohen M L Physical Review B 1985 32 7988
(23) Lukose B Kuc A Heine T manuscript in preparation 2012
(24) Lukose B Supronowicz B Petkov P S Frenzel J Kuc A Seifert G Vayssilov G N
Heine T physica status solidi (b) 2011
(25) Amirjalayer S Snurr R Q Schmid R Journal of Physical Chemistry C 2012 116 4921
(26) Gascon J Hernandez-Alonso M D Almeida A R van Klink G P M Kapteijn F Mul G
Chemsuschem 2008 1 981
(27) Kuc A Enyashin A Seifert G Journal of Physical Chemistry B 2007 111 8179
(28) Dueren T Millange F Ferey G Walton K S Snurr R Q Journal of Physical Chemistry C
2007 111 15350
(29) Furukawa H Yaghi O M Journal of the American Chemical Society 2009 131 8875
144
(30) US DOE Office of Energy Efficiency and Renewable Energy and The FreedomCAR and
Fuel Partnership 2009
httpwww1eereenergygovhydrogenandfuelcellsstoragepdfstargets_onboard_hydro_storage_explanatio
npdf
(31) US DOE USCAR Shell BP ConocoPhillips Chevron Exxon-Mobil T F a F P Multi-Year
Research Development and Demonstration Plan 2009
httpwww1eereenergygovhydrogenandfuelcellsmypppdfsstoragepdf
(32) Uribe-Romo F J Hunt J R Furukawa H Klock C OKeeffe M Yaghi O M Journal of the
American Chemical Society 2009 131 4570
145
Appendix G
A novel series of isoreticular metal organic frameworks realizing metastable structures by liquid phase epitaxy
Jinxuan Liu Binit Lukose Osama Shekhah Hasan Kemal Arslan Peter Weidler Hartmut
Gliemann Stefan Braumlse Sylvain Grosjean Adelheid Godt Xinliang Feng Klaus Muumlllen Ioan-
Bogdan Magdau Thomas Heine and Christof Woumlll
Prepared for publication
Highly crystalline molecular scaffolds with nm-sized pores offer an attractive basis for the fabrication
of novel materials The scaffolds alone already exhibit attractive properties eg the safe storage of
small molecules like H2 CO2 and CH41-4 In addition to the simple release of stored particles changes
in the optical and electronic properties of these nanomaterials upon loading their porous systems
with guest molecules offer interesting applications eg for sensors5 or electrochemistry6 For the
construction of new nanomaterials the voids within the framework of nanostructures may be loaded
with nm-sized objects such as inorganic clusters larger molecules and even small proteins a
process that holds great potential as for example in drug release7-8 or the design of novel battery
materials9 Furthermore meta-crystals may be prepared by loading metal nanoparticles into the
pores of a three-dimensional scaffold to provide materials with a number of attractive applications
ranging from plasmonics10-12 to fundamental investigations of the electronic structure and transport
properties of the meta-crystals13
146
In the last two decades numerous studies have shown that MOFs also termed porous coordination
polymers (PCPs) fulfill many of the aforementioned criteria Although originally developed for the
storage of hydrogen eg in tanks of automobiles MOFs have recently found more technologically
advanced applications as sensors5 matrices for drug release7-8 and as membranes for enantiomer
separation14 and for proton conductance in fuel cells15 Although pore-sizes available within MOFs1
are already sufficiently large to host metal-nanoclusters such as Au5516-17 the actual realization of
meta-crystals requires in addition to structural requirements a strategy for the controlled loading
of the nanoparticles (NPs) into the molecular framework Simply adding NPs to the solutions before
starting the solvothermal synthesis of MOFs has been reported18 but this procedure does not allow
for controlled loading The layer-by-layer growing process of SUFMOFs allows the uptake of
nanosized objects during synthesis including the fabrication of compositional gradients of different
NPs Those guest NPs can range from pure metal metal oxide or covalent clusters over one-
dimensional objects such as nanowires carbon or inorganic nanotubes to biological molecules such
as drugs or even small proteins If the loading happens during synthesis alternating layers of
different NPs can be realized
The introduction of procedures for epitaxial growth of MOFs on functionalized substrates19 was a
major step towards controlled functionalization of frameworks Epitaxial synthesis allowed the
preparation of compositional gradients20-21 and provided a new strategy to load nano-objects into
predefined pores
Unfortunately the LPE process has so far been only demonstrated for a fairly small number of
MOFs141922-23 none of which exhibits the required pore sizes In addition for the fabrication of meta-
crystals the architecture of the network should be sufficiently adjustable to realize pores of different
sizes There should also be a straightforward way to functionalize the framework itself in order to
tailor the interaction of the walls with the targeted guest objects Ideally such a strategy would be
based on an isoreticular series of MOFs in which the pore size can be determined by a choice from a
homologous series of ligands with different lengths1
Here we present a novel isoreticular MOF series with the required flexibility as regards pore-sizes
and functionalization which is well suited for the LPE process This class denoted as SURMOF-2 is
derived from MOF-2 one of the simplest framework architectures24-25 MOF-2 is based on paddle-
wheel (pw) units formed by attaching 4 dicarboxylate groups to Cu++ or Zn++-dimers yielding planar
sheets with 4-fold symmetry (see Fig 1) These planes are held together by strong
carboxylatemetal- bonds26 Conventional solvothermal synthesis yields stacks of pw-planes shifted
relative to each other with a corresponding reduction in symmetry to yield a P2 or C2 symmetry2427-
28
147
The relative shifts between the pw-planes can be avoided when using the recently developed liquid
phase epitaxy (LPE) method22 As demonstrated by the XRD-data shown in Fig 2 for a number of
different dicarboxylic acid ligands the LPE-process yields surface attached metal organic frameworks
(SURMOFs) with P4 symmetry29 except for the 26-NDC ligand which exhibits a P2 symmetry as a
result of the non-linearity of the carboxyl functional groups on the ligand The ligand PPDC
pentaphenyldicarboxylic acid with a length of 25 nm is to our knowledge the longest ligand which
has so far been used successfully for a MOF synthesis303030 A detailed analysis of the diffraction data
allowed us to propose the structures shown in Fig 1 This novel isoreticular series of MOFs hereafter
termed SURMOF-2 also consists of pw-planes which ndash in contrast to MOF-2 ndash are stacked directly
on-top of each other This absence of a lateral shift when stacking the pw-planes yields 1d-pores of
quadratic cross section with a diagonal up to 4 nm (see Fig 2) Importantly for the SURMOF-2 series
interpenetration is absent For many known isoreticular MOF series the formation of larger and
larger pores is limited by this phenomenon if the pores become too large a second or even a third
3d-lattice will be formed inside the first lattice yielding interpenetrated networks which exhibit the
expected larger lattice constant but with rather small pore sizes31-33 the hosting of NPs thus becomes
impossible For SURMOF-2 the particular topology with 1d-pores of quadratic cross section is not
compatible with the presence of a second interwoven network and as a result interpenetration is
suppressed
Although the solubility of some the long dicarboxylic acids ligands used for the SURMOF-2 fabrication
(TPDC QPDC and PPDC see Fig 1) is fairly small we encountered no problems in the LPE process
since already small concentrations of dicarboxylic acids are sufficient for the formation of a single
monolayer on the substrate the elementary step of the LPE synthesis process34-35 Even with the
longest dicarboxylic acid available to us PPDC growth occurred in a straightforward fashion and
optimization of the growth process was not necessary
The P4 structures proposed for the SURMOF-2 series except for the 26-NDC ligand differ markedly
from the bulk MOF-2 structures obtained from conventional solvothermal synthesis2427-28 To
understand this unexpected difference and in particular the absence of relative shifts between the
pw-planes in the proposed SURMOF-2 structure we performed extensive quantum chemical
calculations employing approximate density-functional theory (DFT) in this case London dispersion-
corrected self-consistent charge density-functional based tight-binding (DFTB)36-38 to a periodic
model of MOF-2 and its SURMOF derivatives
Periodic structure DFTB calculations (Fig S1) confirmed that the P2m structure proposed by Yaghi
et al for the bulk MOF-2 material24 is indeed the energetically favorable configuration for MOF-2
while the C2m structure that has also been reported for bulk MOF-227-28 is slightly higher in energy
148
(41 meV per formula unit see Table 1) The optimum interlayer distance between the pw-planes in
the P4 structure is 58 Aring in very good agreement with the experimental value of 56 Aring (as obtained
from the in-plane XRD-data see Fig S2) It is important to note that at that distance the linkers
cannot remain in rectangular position as in MOF-2 but are twisted in order to avoid steric hindrance
and to optimize linker-linker interactions
The P4mmm structure as proposed for SURMOF-2 is less stable by 200 meV per formula unit as
compared to the P2m configuration Although the crystal energy for the P4 stacking is substantially
smaller than for P2 packing simulated annealing (DFTB model using Born-Oppenheimer Molecular
Dynamics) carried out at 300K for 10 ps demonstrated that the P4mmm structure corresponds to a
local minimum This observation is confirmed by the calculation of the stacking energy of MOF-2
where a shifting of the layers is simulated along the P2-P4-C2 path (Figure 3) On the basis of these
calculations we thus propose that SURMOF-2 adopts this metastable P4 structure
In Table 1 the interlayer interaction energies are summarized They range from 590 meV per formula
unit with respect to fully dissociated MOF-2 layers for Cu2(bdc)2 and successively increase for longer
linkers of the same type reaching 145 eV per formula unit for Cu2(ppdc)2 indicating that the linkers
play an important role in the overall stabilization of SURMOF-2 with large pore sizes Even stronger
interlayer interactions are found for different linker topologies (PPDC) A detailed computational
analysis of the role of the individual building blocks of SURMOF-2 for the crystal growth and
stabilization will be published elsewhere
The flat well-defined organic surfaces used as the templating (or nucleating) substrate in the LPE
growth process provide a satisfying explanation for why SURMOF-2 grows with the highly
symmetrical P4 structure instead of adopting the energetically more favorable bulk P2 symmetry3439
The carboxylic acid groups exposed on the surface force the first layer of deposited Cu-dimers into a
coplanar arrangement which is not compatible with the bulk P2 (or C2) structure but rather
nucleates the P4 packing In turn the carboxylate-groups in the first layer of deposited dicarboxylic
acids provide the same constraints for the next layer (see Fig 1) As a result of the layer-by-layer
method employed for further SURMOF-2 growth the same boundary conditions apply for all
subsequent layers The organic surface thus acts as a seed structure to yield the metastable P4
packing not an unusual motif in epitaxial growth40
The calculations allow us to predict that it will be possible to grow SURMOF structures with even
larger linker molecules as used here The stacking will further stabilize the P4 symmetry as the
interaction energy per formula unit is more and more dominated by the linkers (Table 1) At present
149
we cannot predict the maximum possible size of the pores but linker PPDC (see Fig 2) is thus far
unmatched as a component in non-interpenetrated framework structures
Acknowledgement
We thank Ms Huumllsmann Bielefeld University for the preparation of P(EP)2DC Financial support by
Deutsche Forschungsgemeinschaft (DFG) within the Priority Program Metal-Organic Frameworks
(SPP 1362) is gratefully acknowledged
Methods
Computational Details
All structures were created using a preliminary version of our topological framework creator
software which allows the creation of topological network models in terms of secondary building
units and their replacement by individual molecules to create the coordinates of virtually any
framework material The generated starting coordinates including their corresponding lattice
parameters have been hierarchically fully optimized using the Universal Force Field (UFF)41 followed
by the dispersion-corrected self-consistent-charge density-functional based tight-binding (DFTB)
method36-3842-43 that we have recently validated against experimental structures of HKUST-1 MOF-5
MOF-177 DUT-6 and MIL-53(Al) as well as for their performance to compute the adsorption of
water and carbon monoxide37 For all calculations we employed the deMonNano software44444444
We have chosen periodic boundary conditions for all calculations and the repeated slab method has
been employed to compute the properties of the single layers in order to evaluate the stacking
energy A conjugate-gradient scheme was employed for geometry optimization of atomic
coordinates and cell dimensions The atomic force tolerance was set to 3 x 10-4 eVAring
The confined conditions at the SURMOF-substrate interface was modeled by fixing the corresponding
coordinate in the computer simulations All calculated structures have been substantiated by
simulated annealing calculations where a Born-Oppenheimer Molecular Dynamics trajectory at 300K
has been computed for 10 ps without geometry constrains All structures remained in P4 topology
Experimental methods
The SURMOF-2 samples were grown on Au substrates (100-nm Au5-nm Ti deposited on Si wafers)
using a high-throughput approach spray method45 The gold substrates were functionalized by self-
assembled monolayers SAMs of 16-mercaptohexadecanoic acid (MHDA) These substrates were
mounted on a sample holder and subsequently sprayed with 1 mM of Cu2 (CH3COO)4middotH2O ethanol
solution and ethanolic solution of SURMOF-2 organic linkers (01 mM of BDC 26-NDC BPDC and
150
saturated concentration of TPDC QPDC PPDC P(EP)2DC) at room temperature After a given
number of cycles the samples were characterized with X-ray diffraction (XRD)
Figure 1 Schematic representation of the synthesis and formation of the SURMOF-2 analogues
151
Figure 2 Out of plane XRD data of Cu-BDC Cu-26-NDC Cu-BPDC Cu-TPDC Cu-QPDC Cu-P(EP)2DC and Cu-PPDC (upper left) schematic representation (upper right) and proposed structures of SURMOF-2 analogues (lower panel) All the SURMOF-2 are grown on ndashCOOH terminated SAM surface using the LPE method
152
Figure 3 Energy for the relative shift of each layer in two different directions horizontal (P4 to P2) and diagonal (P4 to C2) at a fixed interlayer distance of 56 Aring Structures have been partially optimized and energies are given per formula unit with respect to fully dissociated planes
Supporting information
Synthesis of organic linkers
(1) para-terphenyldicarboxylic acid (TPDC)
To a solution of para-terphenyl (1500 g 651 mmol 1 eq) and oxalyl chloride (3350 mL 3900 mmol
6 eq) in carbon disulfide (30 mL) at 0degC (ice bath) was added aluminium chloride (1475 g 1106
mmol 17 eq) After 1h stirring additional aluminium chloride (0870 g 651 mmol 1 eq) was added
the ice bath was removed and the reaction mixture was stirred at room temperature for 24h The
mixture was poured into crushed ice and stirred until the dark-brown mixture turned to yellow
Carbon disulfide was removed under reduced pressure and the resulting aqueous suspension was
filtered The solid was washed with diluted hydrochloric acid then with diethyl ether and dried
under vacuum overnight with an oil bath at 50degC Pale yellow solid yield 2028 g (98)
(2) para-quaterphenyldicarboxylic acid (QPDC)
153
To a solution of para-quaterphenyl (1000 g 326 mmol 1 eq) and oxalyl chloride (1680 mL 1956
mmol 6 eq) in carbon disulfide (20 mL) at 0degC (ice bath) was added aluminium chloride (0740 g 555
mmol 17 eq) After 1h stirring additional aluminium chloride (0435 g 326 mmol 1 eq) was added
the ice bath was removed and the reaction mixture was stirred at room temperature for 24h The
mixture was poured into crushed ice and stirred until the dark-brown mixture turned to yellow
Carbon disulfide was removed under reduced pressure and the resulting aqueous suspension was
filtered The solid was washed with diluted hydrochloric acid then with diethyl ether and dried
under vacuum overnight with an oil bath at 50degC Yellow solid yield 1186 g (92)
(3) P(EP)2DC
The synthesis of the P(EP)2DC-linker has been described in Ref 46
(4) para-pentaphenly dicarboxylic acid (PPDC)
Scheme 1 Synthesis of para-pentaphenly dicarboxylic acid i) Pd(PPh3)4 Na2CO3 toluenedioxane H2O 85oC ii) MeOH 6M NaOH HCl
para-pentaphenly dicarboxylic acid (H2L) was synthesized via Suzuki coupling of 44rdquo-dibromo-p-
terphenyl and 4-methoxycarbonylphenylboronic acid (Scheme 1) 44rdquo-dibromo-p-terphenyl (776 mg
200 mmol) 4-methoxycarbonylphenylboronic acid (108 g 60 mmol) and Na3CO3 (170 g 160 mmol)
were added to a degassed mixture of toluene14-dioxane (50 mL 221) under Ar [Pd(PPh3)4] (116
mg 010 mmol) was added to the mixture and heated to 85 degC for 24 hours under Ar The reaction
mixture was cooled to room temperature The precipitate was collected by filtration washed with
water methanol and used for next reaction without further purification The final product H4L was
obtained by hydrolysis of the crude product under reflux in methanol (100 mL) overnight with 6M
aqueous NaOH (20 mL) followed by acidification with HCl (conc) After removal of methanol the
final product was collected by filtration as a grey solid (078 g 83 yield) 1H NMR (d6-DMSO
250MHz 298K) δ (ppm) 805 (d J = 75Hz 4H) 789-783 (m 12H) 750 (d J = 75Hz 4H) 13C NMR
cannot be clearly resolved due to poor solubility MALDI-TOF-MS (TCNQ as matrix) mz 47002
cacld 47051 Elemental analysis Calculated C 8169 H 471 Found C 8163 H 479
Br Br MeOOC B
OH
OH
+
COOMe
COOMe
COOH
COOH
i ii
154
Figure S1 MOF-2 in P4 (as reported) P2 and C2 symmetry
155
Figure S2 In-plane XRD data of Cu-BDC Cu-26-NDC Cu-BPDC Cu-TPDC Cu-QPDC and Cu-P(EP)2DC All the
SURMOF-2 are grown on ndashCOOH terminated SAM surface using the LPE method The position of (010) plane
represents the layer distance
Table 1 Stacking energy and geometries of P4 SURMOF-2 derivatives
Symmetry a= c b Stacking Energy
Cu2(bdc)2 C2 1119 50 -076
Cu2(bdc)2 P2 1119 54 -08
Cu2(bdc)2 P4 1119 58 -059
156
Cu2(ndc)2 P2 1335 56 -04
Cu2(bpdc)2 P4 1549 59 -068
Cu2(tpdc)2 P4 1984 59 -091
Cu2(qpdc)2 P4 2424 59 -121
Cu2(P(EP)2DC)2 P4 2512 52 -173
Cu2(ppdc)2 P4 2859 59 -145
Stacking energies (DFTB level in eV) of SURMOFs The energies were calculated within periodic
boundary conditions and are given per formula unit
References
1 Eddaoudi M et al Systematic design of pore size and functionality in isoreticular MOFs and
their application in methane storage Science 295 469-472 (2002)
2 Rosi N L et al Hydrogen storage in microporous metal-organic frameworks Science 300
1127-1129 (2003)
3 Rowsell J L C amp Yaghi O M Metal-organic frameworks a new class of porous materials
Microporous and Mesoporous Materials 73 3-14 (2004)
4 Wang B Cote A P Furukawa H OKeeffe M amp Yaghi O M Colossal cages in zeolitic
imidazolate frameworks as selective carbon dioxide reservoirs Nature 453 207-U206 (2008)
5 Lauren E Kreno et al Metal-Organic Framework Materials as Chemical Sensors Chemical
Reviews 112 1105-1124 (2012)
6 Dragasser A et al Redox mediation enabled by immobilised centres in the pores of a metal-
organic framework grown by liquid phase epitaxy Chemical Communications 48 663-665
(2012)
7 Horcajada P et al Metal-organic frameworks as efficient materials for drug delivery
Angewandte Chemie-International Edition 45 5974-5978 (2006)
8 Horcajada P et al Flexible porous metal-organic frameworks for a controlled drug delivery
Journal of the American Chemical Society 130 6774-6780 (2008)
9 Hurd J A et al Anhydrous proton conduction at 150 degrees C in a crystalline metal-organic
framework Nature Chemistry 1 705-710 (2009)
10 Kreno L E Hupp J T amp Van Duyne R P Metal-Organic Framework Thin Film for Enhanced
Localized Surface Plasmon Resonance Gas Sensing Analytical Chemistry 82 8042-8046
(2010)
11 Lu G et al Fabrication of Metal-Organic Framework-Containing Silica-Colloidal Crystals for
Vapor Sensing Advanced Materials 23 4449-4452 (2011)
157
12 Lu G amp Hupp J T Metal-Organic Frameworks as Sensors A ZIF-8 Based Fabry-Perot Device
as a Selective Sensor for Chemical Vapors and Gases Journal of the American Chemical
Society 132 7832-7833 (2010)
13 Mark D Allendorf Adam Schwartzberg Vitalie Stavila amp Talin A A Roadmap to
Implementing MetalndashOrganic Frameworks in Electronic Devices Challenges and Critical
Directions European Journal of Chemistry (2011)
14 Bo Liu et al Enantiopure Metal-Organic Framework Thin Films Oriented SURMOF Growth
and Enantioselective Adsorption Angewandte Chemie-International Edition 51 817-810
(2012)
15 Sadakiyo M Yamada T amp Kitagawa H Rational Designs for Highly Proton-Conductive
Metal-Organic Frameworks Journal of the American Chemical Society 131 9906-9907 (2009)
16 Ishida T Nagaoka M Akita T amp Haruta M Deposition of Gold Clusters on Porous
Coordination Polymers by Solid Grinding and Their Catalytic Activity in Aerobic Oxidation of
Alcohols Chemistry-a European Journal 14 8456-8460 (2008)
17 Esken D Turner S Lebedev O I Van Tendeloo G amp Fischer R A AuZIFs Stabilization
and Encapsulation of Cavity-Size Matching Gold Clusters inside Functionalized Zeolite
Imidazolate Frameworks ZIFs Chemistry of Materials 22 6393-6401 (2010)
18 Lohe M R et al Heating and separation using nanomagnet-functionalized metal-organic
frameworks Chemical Communications 47 3075-3077 (2011)
19 Shekhah O et al Step-by-step route for the synthesis of metal-organic frameworks Journal
of the American Chemical Society 129 15118-15119 (2007)
20 Furukawa S et al A block PCP crystal anisotropic hybridization of porous coordination
polymers by face-selective epitaxial growth Chemical Communications 5097-5099 (2009)
21 Shekhah O et al MOF-on-MOF heteroepitaxy perfectly oriented [Zn(2)(ndc)(2)(dabco)](n)
grown on [Cu(2)(ndc)(2)(dabco)](n) thin films Dalton Transactions 40 4954-4958 (2011)
22 Shekhah O et al Controlling interpenetration in metal-organic frameworks by liquid-phase
epitaxy Nature Materials 8 481-484 (2009)
23 Zacher D et al Liquid-Phase Epitaxy of Multicomponent Layer-Based Porous Coordination
Polymer Thin Films of [M(L)(P)05] Type Importance of Deposition Sequence on the Oriented
Growth Chemistry-a European Journal 17 1448-1455 (2011)
24 Li H Eddaoudi M Groy T L amp Yaghi O M Establishing microporosity in open metal-
organic frameworks Gas sorption isotherms for Zn(BDC) (BDC = 14-benzenedicarboxylate)
Journal of the American Chemical Society 120 8571-8572 (1998)
25 Mueller U et al Metal-organic frameworks - prospective industrial applications Journal of
Materials Chemistry 16 626-636 (2006)
158
26 Shekhah O Wang H Zacher D Fischer R A amp Woumlll C Growth Mechanism of Metal-
Organic Frameworks Insights into the Nucleation by Employing a Step-by-Step Route
Angewandte Chemie-International Edition 48 5038-5041 (2009)
27 Carson C G et al Synthesis and Structure Characterization of Copper Terephthalate Metal-
Organic Frameworks European Journal of Inorganic Chemistry 2338-2343 (2009)
28 Clausen H F Poulsen R D Bond A D Chevallier M A S amp Iversen B B Solvothermal
synthesis of new metal organic framework structures in the zinc-terephthalic acid-dimethyl
formamide system Journal of Solid State Chemistry 178 3342-3351 (2005)
29 Arslan H K et al Intercalation in Layered Metal-Organic Frameworks Reversible Inclusion of
an Extended pi-System Journal of the American Chemical Society 133 8158-8161 (2011)
30 The MOF with the largest pore size recorded so far MOF-200 (Furukawa H et al Ultrahigh
Porosity in Metal-Organic Frameworks Science 329 424-428 (2010)) used a (trivalent)
444-(benzene-135-triyl-tris(benzene-41-diyl))tribenzoate (BBC) ligand The carboxylic
acid-to carboxylic acid distance is 20 nm compared to 25 nm in case of PPDC The cage size
in MOF-200 amounts to 18 nm by 28 nm clearly smaller than the 1d-channels in the PPDC
SURMOF-2 that are 28 nm by 28 nm
31 Batten S R amp Robson R Interpenetrating nets Ordered periodic entanglement
Angewandte Chemie-International Edition 37 1460-1494 (1998)
32 Snurr R Q Hupp J T amp Nguyen S T Prospects for nanoporous metal-organic materials in
advanced separations processes Aiche Journal 50 1090-1095 (2004)
33 Yaghi O M A tale of two entanglements Nature Materials 6 92-93 (2007)
34 Shekhah O Liu J Fischer R A amp Woumlll C MOF thin films existing and future applications
Chemical Society Reviews 40 1081-1106 (2011)
35 Zacher D Shekhah O Woumlll C amp Fischer R A Thin films of metal-organic frameworks
Chemical Society Reviews 38 1418-1429 (2009)
36 Elstner M et al Self-consistent-charge density-functional tight-binding method for
simulations of complex materials properties Physical Review B 58 7260-7268 (1998)
37 Lukose B et al Structural properties of metal-organic frameworks within the density-
functional based tight-binding method Physica Status Solidi B-Basic Solid State Physics 249
335-342 (2012)
38 Zhechkov L Heine T Patchkovskii S Seifert G amp Duarte H A An efficient a Posteriori
treatment for dispersion interaction in density-functional-based tight binding Journal of
Chemical Theory and Computation 1 841-847 (2005)
159
39 Zacher D Schmid R Woumlll C amp Fischer R A Surface Chemistry of Metal-Organic
Frameworks at the Liquid-Solid Interface Angewandte Chemie-International Edition 50 176-
199 (2011)
40 Prinz G A Stabilization of bcc Co via Epitaxial Growth on GaAs Physical Review Letters 54
1051-1054 (1985)
41 Rappe A K Casewit C J Colwell K S Goddard W A amp Skiff W M UFF a full periodic
table force field for molecular mechanics and molecular dynamics simulations Journal of the
American Chemical Society 114 10024-10035 (1992)
42 Seifert G Porezag D amp Frauenheim T Calculations of molecules clusters and solids with a
simplified LCAO-DFT-LDA scheme International Journal of Quantum Chemistry 58 185-192
(1996)
43 Oliveira A F Seifert G Heine T amp Duarte H A Density-Functional Based Tight-Binding an
Approximate DFT Method Journal of the Brazilian Chemical Society 20 1193-1205 (2009)
44 deMonNano v 2009 (Bremen 2009)
45 Arslan H K et al High-Throughput Fabrication of Uniform and Homogenous MOF Coatings
Adv Funct Mater 21 4228-4231 (2011)
46 Schaate A et al Porous Interpenetrated Zirconium-Organic Frameworks (PIZOFs) A
Chemically Versatile Family of Metal-Organic Frameworks Chemistry-a European Journal 17
9320-9325 (2011)
160
Appendix H
Linker guided metastability in templated Metal-Organic Framework-2 derivatives (SURMOFs-2)
Binit Lukose Gabriel Merino Christof Woumlll and Thomas Heine
Prepared for publication
INTRODUCTION
The molecular assembly of metal-oxides and organic struts can provide a large number of network
topologies for Metal-Organic Frameworks (MOFs)1-3 This is attained by the differences in
connectivity and relative orientation of the assembling units Within each topology replacement of a
building unit by another of same connectivity but different size leads to what is known as isoreticular
alteration of pore size The structure of MOFs in principle can be formed into the requirement of
prominent applications such as gas storage4-8 sensing910 and catalysis11-13 The structural
divergence and the performance can be further increased by functionalizing the organic linkers1415
In MOFs linkers are essential in determining the topology as well as providing porosity A linker
typically contains single or multiple aromatic rings the orientation of which normally undergoes
lowest repulsion from the coordinated metal-oxide clusters providing the most stable symmetry for
the bulk material We encounter for the first time a situation that the orientation of the linker
provides metastability in an isoreticular series of MOFs Templated MOF-2 derivatives (surface-MOF-
2 or simply SURMOFs-2)16 show this extraordinary phenomenon While bulk MOF-2 is determined to
be stable in P217 or C218-20 symmetry we found the existence of SURMOFs-2 in P4 symmetry
161
(Figure 1)16 Our theoretical approach to this problem identifies the role of the linkers in providing
P4 geometry the status of a local energy-minimum
MOF-2 derivatives are two-dimensional lattice structures composed of dimetal-centered four-fold
coordinated paddlewheels and linear organic linkers Solvothermally synthesized archetypal MOF-2
had transition metal (Zn or Cu) centers in the connectors and benzenedicarboxylic acid linkers17 The
derivatives under our investigation contain paddlewheels with Cu dimers and benzenedicarboxylic
acid (BDC) naphthalenedicarboxylic acid (NDC) biphenyldicarboxylic acid (BPDC)
triphenyldicarboxylic acid (TPDC) quaterphenyldicarboxylic acid (QPDC) P(EP)2DC21 and
pentaphenyldicarboxylic acid (PPDC) linkers pertaining to isoreticular expansion of pore size16 The
four-fold connectivity of the Cu dimers together with linearity of the linkers gives rise to layers with
quadratic (square) topology The interlayer separation d is typically much more than that of
graphite or 2D COFS22-24 because all the atoms in one layer do not lie in one plane
In bulk form the nearest layers are shifted to each other either towards one of the four linkers
(P2)171920 or diagonally towards one of the four surrounding pores (C2)18-20 in effort to reduce
the repulsion of alike charges in the adjacent paddlewheels when they are in eclipsed form (P4)
(Figure 1) The metal-dimers often show high reactivity which results in attracting water or
appropriate solvents in their axial positions The stacking along the third axis is typically through
interlayer interactions and through hydrogen bonds established between the solvents or between
the solvents and oxygen atoms in the paddlewheel The lateral shift of layers is inevitable without
additional supports Coordination of pillar molecules such as 14-diazabicyclo[222]octane (dabco) or
bipyridine in the axis of metal dimers between the adjacent layers25 is a practical mean to avoid
layer-offset however with the change of MOF dimensionality
Figure 1 Monolayer of MOF-2 and three kinds of layer packing -P4 P2 and C2- of periodic MOF-2
162
Alternate to solvothermal synthesis a step-by-step route may be enabled for the synthesis of
MOFs26-28 It adopts liquid phase epitaxy (LPE) of MOF reactants on substrate-assisted self-assembled
monolayers This is achieved by alternate immersion of the template in metal and ligand precursors
for several cycles This allows controlled growth of MOFs on surfaces The latest outcomes of this
method are the surface-standing MOF-2 derivatives (SURMOF-2s) This isoreticular SURMOF series
has linkers of different lengths (as given above) The cell dimensions that correspond to the length of
the pore walls range from 1 to 28 nm The diagonal pore-width of one of the MOFs (PPDC) accounts
to 4 nm
After evacuation of solvents (water in the present case) X-ray diffraction patterns were taken in
directions both perpendicular (scans out of plane) and parallel (scans in-plane) to the substrate
surface The presence of only one peak -(010)- in the perpendicular scan implies that the layers
orient perpendicular to the surface and the corresponding angle gives the cell parameter a (or c) In
the latter case the peaks - (100) and (001) ndash are identical and this rules out the possibility of layer-
offset Hence the symmetry of the MOFs was determined to be P4 These peaks give rise to the cell
parameter c which is nothing but the interlayer distance d This value (56 Aring) is relatively small for
P4 geometry to have water molecules in the axial position of paddlewheels Presence of some water
molecules observed using Infrared Reflection Absorption Spectroscopy (IRRAS) might be located near
paddlewheels but exposed to the pore The new observations with the SURMOFs are very intriguing
in the context that P4 symmetry appears to be impossible for solvothermally produced MOF-2
We have primarily reported that the P4 geometry of SURMOF-2 exists as a local energy-minimum16
The verification was made using an approximate method of density functional theory (DFT) which is
London dispersion-corrected self-consistent charge density-functional based tight-binding (DFTB) In
the bulk form absolute energies of P2 and C2 are 210 and 170 meV respectively higher than P4 per
a formula unit which is equivalent to the unit cell For SURMOF the barrier for leaving P4 was nearly
50 meV per formula unit It requires further analysis to unravel the reasons for this unusual
metastability We therefore performed an extensive set of quantum chemical calculations on the
composition of the constituent building units The procedure involves defining SURMOF geometry
and analyzing the translations of individual layers
The major individual contributions to the total energy are the interaction between the paddlewheel
units (favouring P2) London dispersion interaction between tilted linkers (favouring P4) and energy
to tilt the linkers In the iso-reticular series of SURMOF-2 the differences arise only in the
163
contributions from the linkers Hence we performed an extensive study only on the smallest of all
linkers- BDC A scaling might be appropriate for other linkers
RESULTS AND DISCUSSION
In solvothermal synthesis of layered MOFs it is assumed that the nucleation process is associated
with the interaction between two connectors This is rationalized by the fact that two paddlewheels
show the strongest possible noncovalent interaction between the individual MOF building blocks
present in MOF-2 As the orientation of the paddlewheels is restricted by the linkers we studied the
stacking of adjacent layers by determining the attraction of two adjacent interacting paddlewheels
upon their respective offsets Thus we investigated the geometries corresponding to lateral
displacements between the known stable arrangements of the bulk structure ie P4-to-P2 and P4-
to-C2 (Figure 2) In the model calculation the two paddlewheels were shifted from D4h symmetry to
two directions that respectively correspond to P2 and C2 symmetries in the bulk The minima along
the two shifts are referred as min-I and min-II and are having C2h symmetry It is interesting to note
that the interaction is in all cases attractive If only the paddlewheels are studied the D4h
configuration where both axes are concentric can be interpreted as transition state between the
two minima configurations P2 or C2 Comparing the two minima min-I corresponding to MOF-2 in
P2 symmetry indicates the formation of two CuO bonds while for min-II the reactive Cu centers do
not participate in the interlayer bonding
Formation of the diagonally shifted C2 bulk geometry for Zn and Cu based MOF-2 as reported in the
literature18-20 possibly is due to the presence of large solvent molecules such as DMF that
coordinate to the free Cu centers the paddlewheels
Figure 2 Displacements of two paddlewheels from D4h symmetry (corresponding to P4) towards minima that correspond to P2 and C2 bulk symmetries
164
To gain further insight on type of interactions for the three paddlewheel arrangements as found in
the bulk (Figure 3) we performed the topological analysis of the electron density for each
structure2930 An inspection of the paddlewheel shows that the electron density of the Cu-O bond has
a (relatively small) value of 0042 au thus showing only weak character In the forms related to P4
and C2 bulk symmetries all (3-1) critical points between the two paddlewheels show very small
density values (0004 au and less) In the P2 structure it is apparent the formation of a four-
membered Cu-O-Cu-O ring across the paddlewheel Note that the Cu-O interactions inside the
paddlewheel weaken (from 0042 to 0031 au) and two intermolecular Cu-O interactions with a
density value of 0024 au stabilize the P2 orientation These links if formed during nucleation will
be regularly repeated during the MOF-2 formation in solvothermal synthesis and due to its strong
binding of -08 eV they are the main reason for the stabilization of the P2 phase If the paddlewheels
are packed in P4 symmetry there must be additional means of stabilization present and that may
only arise from the linkers Moreover one should note that in order to reach the P4 symmetry a
layer-by-layer growth process is required as otherwise the linkers will readily stack as in the P2 bulk
form
165
Figure 3 Topological analysis of the electron density of fully optimized P4 (top left) C2 (top right) and P4 (bottom) structures Atom (grey) (3-1) (red) and (3+1) (green) critical points along with their charge densities are shown
The optimized geometry of a single-layer MOF-2 is shown in Figure 1 Note that the phenyl rings of
the linkers are oriented perpendicular to the MOF plane Contrary to graphene or 2D COFs the more
complex structure of MOF-2 layers may become subject to change upon the interlayer interactions
This is in particular important for the P4 stacking as reported for SURMOF-2 The interaction energy
of two linkers and two benzene rings as oriented in the monolayer has been computed as function
of the interlayer distance (Figure 4) At the experimental interlayer distance of 56 Aring the linkers are
so close that they repel each other strongly and stacking the monolayer structure at the
experimental interlayer distance would introduce an energy penalty of 08 eV per linker
It would not be exotic if we assume that the anchoring of layers on the substrate plays an important
role in setting d A supporting argument is the fact that all the SURMOFs in the iso-reticular series
have the same d An additional point is that the comparatively wider linkers NDC and LM do not
create any difference in the interlayer distance
166
Figure 4 Energy as a function of interlayer distance [d] in case of two paddlewheels and two benzene molecules The energies are given with respect to the complete dissociation of the building blocks
The best adaptation for the linkers in a closer alignment to avoid the repulsion is to partially rotate
the phenyl rings This reduces the Pauli repulsion and at the same time increases the attractive
London dispersion between the linkers However the rotation is energetically penalized by 06 eV as
accordance with similar calculations found in the literature31 and is with the same order of Zn4O-
tetrahedron clusters of the IRMOFs3233
Figure 5 Energy as a function of rotation of linker between two saturated paddlewheels The dihedral angle O-C1-C2-C3 θ between the linker and the carboxylic part in the paddlewheel Values are with respect to the ground state θ = 0⁰
To model the conditions in SURMOF-2 we need to combine the intermolecular interaction of the
linkers with the barrier associated to the rotation of the linker between two paddlewheel units as
given in Figure 5 We may consider a system of three linkers placed as if they are in three adjacent
layers of bulk P4 SURMOF-2 The linkers are undergoing a rotation along their C2 axis that would be
aligned to the Cu-Cu axes in the paddlewheels (see system-I in Figure 6) The dissociation energy of
167
the system includes four times the repulsion from one adjacent linker If we neglect the interaction
between the end-linkers which is only -35 meV the system represents the situation in bulk P4 MOF-
2 The gain of intermolecular energy due to twisting the stacked linkers is partially compensated by
the energy penalty arising from rotation of the linker between the paddlewheels and the resulting
energy shows a minimum at 22deg (Figure 6)
Figure 6 Model of the linker orientation in SURMOF-2 The stabilization of a linker inbetween two layers is approximated in System-I the arrows indicating the rotation while the energy penalty due to breaking the p conjugation of Figure 5 is given in System-II The resulting energy is the black line showing a minimum at 22deg The energy in System-I is with respect to its dissociation limit
Each linker may undergo rotation either clockwise or anti-clockwise with equal preferences in the
local environment However there may be a global control over the preference of each linker The
most stable structure can be figured out from the total energies of each possible arrangement Since
there are only two choices for each linker it may orient either in same fashion or alternate fashion
along X and Y directions If we expect a regular pattern the total number of possibilities are only
three as shown in Figure 8 where rotation of linkers (violet lines) are marked with lsquo+rsquo signs in one of
its two sides which means that facing (outer)-edges of linkers are tilted towards these sides The
three orderings may be verbalized as follows
(i) projection of the facing edges of oppositely placed linkers are either within the square or outside
(P4nbm) (ii) projection of the facing edge of only one of the oppositely placed linkers is within the
square (P4mmm) (iii) projection of the facing edges of all four linkers are either within the square
or outside (P4nmm)
The adjacent layers follow the same linker-orientations The unit cells of (i) and (iii) are four times
bigger than (ii) The stacking energies calculated at the DFTB level suggest P4nbm as the most stable
168
geometry The energies are respectively -048 -032 and -029 eV per formula unit for P4nbm
P4mmm and P4nmm respectively Within P4nbm symmetry the linker edges undergo lowest
repulsion compared to others It is to be noted that only P4nbm has four-fold rotational symmetry
along Z-axis about the Cu-dimer in any paddlewheel
Figure 7 Three possible arrangements of tilted linkers in the periodic SURMOF-2 Paddlewheels are shown as red diamonds and the linkers as violet lines Facing (outer) edges of the linkers are rotated towards the side with + signs The symmetries and calculated stacking energies (per formula unit) of the arrangements are also given
To quantify the different stacking energies we performed periodic DFT calculations on the structure
of Cu2(bdc)2 MOF-2 As the most stable P4nbm geometry demands high computational resources in
each calculation we used P4mmm geometry which has four times less atoms in unit cell We
explored tight-packing with rotated linkers and loose-packing with un-rotated linkers Two energy-
minima have been found as shown in Figure 8a at 58 and 65 Aring corresponding to rotated and un-
rotated states of linkers respectively The latter is 40 meV more stable than the former which
means that SURMOF-2 exists in a metastable state along the perpendicular axis The relative shift of
adjacent layers in the direction of the lattice vector (or ) represents the transition from P2 to P4
and back to P2 symmetry We shifted the adjacent layers parallel to and calculated the relative
energies Since the shift of adjacent layers in P4mmm geometry is not symmetric for positive and
negative directions of averages of the energies of the shift in both directions are plotted (see
Figure 9b) P4 is a local minimum while P2 is the global minimum and the energy barrier separating
the two minima accounts to 23 meV per formula unit Hence the P4 geometry of SURMOF-2 can be
taken as metastable state of MOF-2
169
Figure 8 a) Energy as function of interlayer separation d and b) Energy as a function of relative shift of adjacent layers for periodic MOF-2 Energies are given in eV per formula unit
The role of linkers in creating a local minimum at P4 is now apparent However when undergoing the
transition from P4 to P2 the relative motions of the horizontal and vertical linkers are different from
each other Hence a qualitative study is essential to accurately determine the role of each building
block for holding SURMOFs-2 in a metastable state We thus resolved the shift of entire adjacent
layers with respect to each other into relative motions of individual building blocks The experimental
interlayer distance (56 Aring) has been fixed and the energy-profiles have been calculated using DFT
The scans include the shift of
i) a paddlewheel over other
ii) a horizontal linker over other
iii) a vertical linker over other
iv) a paddlewheel over a horizontal linker
v) a paddlewheel over a vertical linker
Here vertical and horizontal linkers are the linkers vertical and horizontal to the shift directions
respectively A complete picture of individual shifts of the MOF-2 SBU including their energy-profiles
is given in Figure 9 The shift of the vertical over the horizontal linker was found insignificant and was
omitted A note of warning is that the tilted vertical linker meets different neighborhoods when
shifted to the left and right However an average energy of these two shifts seems sensible because
the nearest vertical linker in the same layer of P4nbm geometry has opposite orientation This
averaging also makes sense in a case that alternate layers undergo shifting to the same direction
leading to a zigzag (or serrated) packing of layers in the bulk As is readily seen in Figure 9 the
formation of the Cu-O-Cu-O rings between the paddlewheels (see Figure 3) is the key reasons for the
layer shift in P2 MOF-2 A support is obtained from the shifting of the paddlewheel over the
170
horizontal linker (Shift IV) A major objection is made by the vertical linkers (Shift III) The total
interaction becomes repulsive when a vertical linker shifts with respect to the other for about 05 Aring
This may alter the tilt of the linker however a minimum is already established The vertical linkers of
a layer in a way are trapped between those of adjacent layers The shift to either side from P4 most
probably decreases the interlayer separation However this demands further rotation of the vertical
linkers in order to reduce their mutual Pauli repulsion Overall the sharp minimum at P4 may be
taken as an inheritance from the lower interlayer distance of P4 geometry fixed by the anchoring on
the substrate
Figure 9 Shift of individual building blocks in the adjacent layers correspond to the situation in bulk The energies of each shifted arrangement and their total energy are given in the graph
The total energy involved in the shifting of two building blocks (one building block over the other) is
equivalent to the energy per one building block when it feels shift from two neighbors Only the
vertical linker is sensitive to the shift-direction of the two neighbors However since averages were
taken as discussed earlier the total energy becomes independent of the direction Besides the
relative motion of the SBU facing each other in P4 symmetry (Shifts I II and III) for strong distortions
we also have to consider the interaction of adjacent linker-connector interactions as represented in
Shifts IV and V The former stabilizes P2 while the latter support P4 Overall the total energy for all
the shifts shows a local minimum at P4 (see figure 9) in agreement with the periodic calculation
shown in Figure 8 Indeed the relative energy between P2 and P4 stackings estimated by the
171
superposition of individual contributions is 43 meV in close agreement to the 40 meV obtained by
the periodic calculations
Our finite-component model successfully provides adequate information on the individual
contributions of building blocks Vertical linkers play crucial role in holding the SURMOF with P4
symmetry which was never achieved with solvothermally synthesized MOF-2 Paddlewheels are
held most responsible for lateral shift The vertical linkers winningly establish a local minimum at P4
for the SURMOF
Recently we have reported SURMOF-2 derivatives with very large pore sizes(Ref) This has been
achieved by increasing the length of the linker units In view of our analysis of the stacking and
stability of MOF-2 we can now safely state that it will be possible to generate SURMOF-2 derivatives
with even larger pores with pore sizes essentially limited by the availability of stiff long organic
linker molecules This is exemplified by a calculation of the stacking energy of a series of phenyl
oligomers that is SURMOF-2 derivatives incorporating chains with 1 to 10 benzene rings in the
linkers The stacking energies per formula unit are -041 -067 -091 -121 -145 -168 -192 -215
-237 and -26 eV respectively Each benzene ring adds more than 02 eV into the stacking energy per
formula unit This energy is due to the London dispersion interaction between the linkers in the
neighboring layers
The drop of SURMOFs into the local minimum perhaps is blended with some parameters related to
synthetic environments This was beyond the scope of this work however we suggest that studies of
the mechanism of substrate-assisted layer-by-layer deposition of metal and ligand precursors may
give some primary insights into it
CONCLUSION
We have analyzed the reason for the different stackings observed for MOF-2 In the traditional
solvothermal synthesis MOF-2 is likely to crystallize in P2 symmetry determined by the strong
interaction between the paddlewheel units The coordination of large solvent molecules to the free
metal centers may lead to MOF-2 in C2 symmetry In contrast the formation of SURMOFs using
Liquid Phase Epitaxy (LPE) allows the formation of high-symmetry P4 MOF-2 This structure requires
a structural modification in terms of the orientation of the linkers with respect to the free monolayer
and the stacking is stabilized by London dispersion interactions between the linkers Increasing the
linker length is a straightforward way for the linear expansion of pore size and according to our
computations the pore size is only limited by the availability of linker molecules showing the desired
length Thus we presented a rare situation in which the linkers guarantee the persistence of a series
of materials in an otherwise unachievable state
172
COMPUTATIONAL DETAILS
The Becke three-parameter hybrid method combined with Lee-Yang-Par correlational functional
(B3LYP)34-37 and augmented with a dispersion term38 as implemented in Turbomole39 has been used
for DFT calculations The copper atoms were described using the basis set associated with the Hay-
Wadt40 relativistic effective core potentials41 which include polarization f functions42 This basis set
was obtained from the EMSL Basis Set Library4344 Carbon oxygen and hydrogen atoms were
described using the Pople basis set 6-311G Geometry optimizations of the MOF fragments were
performed using internal redundant co-ordinates45 A triplet spin state was assigned for each Cu-
paddlewheel46
Periodic structure calculations at the BLYP47 level were performed using the ADF BAND 2012
code4849 Dispersion correction was implemented based on the work by Grimme50 Triple-zeta basis
set was used The crystalline state of MOFs was computationally described using periodic boundary
conditions Bader charge analysis of paddlewheel-pairs was also performed using ADF 2012 code
The analysis was made at the level of B3LYP-D using an all-electron triple-zeta basis set
The non-orthogonal tight-binding approximation to DFT called Density Functional Tight-Binding
(DFTB)51 is computationally feasible for unit cells of several hundreds of atoms Hence this method
was used for extensive calculations on periodic structures This method computes a transferable set
of parameters from DFT calculations of a few molecules per pair of atom types The more accurate
self-consistent charge (SCC) extension of DFTB52-54 has been used for the study of bulk MOFs Validity
of the Slater-Koster parameters of Cu-X (X = Cu C O H) pairs were reported before55 The
computational code deMonNano56 which has dispersion correction implemented57 was used
If not stated otherwise the energies of the periodic structure are given per a formula unit (FU) of the
MOF which includes one paddlewheel and two linkers (one vertical linker and one horizontal linker)
REFERENCES
(1) Eddaoudi M Moler D B Li H L Chen B L Reineke T M OKeeffe M Yaghi O M Accounts of
Chemical Research 2001 34 319
(2) Li H Eddaoudi M OKeeffe M Yaghi O M Nature 1999 402 276
(3) Rowsell J L C Yaghi O M Microporous and Mesoporous Materials 2004 73 3
(4) Eddaoudi M Li H L Yaghi O M Journal of the American Chemical Society 2000 122 1391
(5) Rowsell J L C Yaghi O M Angewandte Chemie-International Edition 2005 44 4670
173
(6) Suh M P Park H J Prasad T K Lim D-W Chemical Reviews 2012 112 782
(7) Murray L J Dinca M Long J R Chemical Society Reviews 2009 38 1294
(8) Rosi N L Eckert J Eddaoudi M Vodak D T Kim J OKeeffe M Yaghi O M Science 2003 300
1127
(9) Kreno L E Leong K Farha O K Allendorf M Van Duyne R P Hupp J T Chemical Reviews 2012
112 1105
(10) Achmann S Hagen G Kita J Malkowsky I M Kiener C Moos R Sensors 2009 9 1574
(11) Lee J Farha O K Roberts J Scheidt K A Nguyen S T Hupp J T Chemical Society Reviews 2009
38 1450
(12) Farrusseng D Aguado S Pinel C Angewandte Chemie-International Edition 2009 48 7502
(13) Corma A Garcia H Llabres i Xamena F X Chemical Reviews 2010 110 4606
(14) Rowsell J L C Millward A R Park K S Yaghi O M Journal of the American Chemical Society 2004
126 5666
(15) Deng H Doonan C J Furukawa H Ferreira R B Towne J Knobler C B Wang B Yaghi O M
Science 2010 327 846
(16) Liu J Lukose B Shekhah O Arslan H K Weidler P Gliemann H Braumlse S Grosjean S Godt A
Feng X Muumlllen K Magdau I-B Heine T Woumlll C submitted to Nature Chemistry 2012
(17) Li H Eddaoudi M Groy T L Yaghi O M Journal of the American Chemical Society 1998 120 8571
(18) Carson C G Hardcastle K Schwartz J Liu X Hoffmann C Gerhardt R A Tannenbaum R
European Journal of Inorganic Chemistry 2009 2338
(19) Clausen H F Poulsen R D Bond A D Chevallier M A S Iversen B B Journal of Solid State
Chemistry 2005 178 3342
(20) Edgar M Mitchell R Slawin A M Z Lightfoot P Wright P A Chemistry-a European Journal 2001
7 5168
(21) Schaate A Roy P Preusse T Lohmeier S J Godt A Behrens P Chemistry-a European Journal
2011 17 9320
(22) Cote A P Benin A I Ockwig N W OKeeffe M Matzger A J Yaghi O M Science 2005 310
1166
(23) Wan S Guo J Kim J Ihee H Jiang D Angewandte Chemie-International Edition 2008 47 8826
174
(24) Lukose B Kuc A Frenzel J Heine T Beilstein Journal of Nanotechnology 2010 1 60
(25) Kitagawa S Kitaura R Noro S Angewandte Chemie-International Edition 2004 43 2334
(26) Shekhah O Wang H Zacher D Fischer R A Woell C Angewandte Chemie-International Edition
2009 48 5038
(27) Shekhah O Wang H Kowarik S Schreiber F Paulus M Tolan M Sternemann C Evers F
Zacher D Fischer R A Woll C Journal of the American Chemical Society 2007 129 15118
(28) Zacher D Schmid R Woell C Fischer R A Angewandte Chemie-International Edition 2011 50 176
(29) Bader R F W Accounts of Chemical Research 1985 18 9
(30) Merino G Vela A Heine T Chemical Reviews 2005 105 3812
(31) Tafipolsky M Schmid R Journal of Chemical Theory and Computation 2009 5 2822
(32) Kuc A Enyashin A Seifert G Journal of Physical Chemistry B 2007 111 8179
(33) Winston E B Lowell P J Vacek J Chocholousova J Michl J Price J C Physical Chemistry
Chemical Physics 2008 10 5188
(34) Becke A D Journal of Chemical Physics 1993 98 5648
(35) Lee C T Yang W T Parr R G Physical Review B 1988 37 785
(36) Vosko S H Wilk L Nusair M Canadian Journal of Physics 1980 58 1200
(37) Stephens P J Devlin F J Chabalowski C F Frisch M J Journal of Physical Chemistry 1994 98
11623
(38) Civalleri B Zicovich-Wilson C M Valenzano L Ugliengo P Crystengcomm 2008 10 405
(39) University of Karlsruhe and Forschungszentrum Karlsruhe gGmbH - TURBOMOLE V63 2011 2007
(40) Wadt W R Hay P J Journal of Chemical Physics 1985 82 284
(41) Roy L E Hay P J Martin R L Journal of Chemical Theory and Computation 2008 4 1029
(42) Ehlers A W Bohme M Dapprich S Gobbi A Hollwarth A Jonas V Kohler K F Stegmann R
Veldkamp A Frenking G Chemical Physics Letters 1993 208 111
(43) Feller D Journal of Computational Chemistry 1996 17 1571
(44) Schuchardt K L Didier B T Elsethagen T Sun L Gurumoorthi V Chase J Li J Windus T L
Journal of Chemical Information and Modeling 2007 47 1045
175
(45) von Arnim M Ahlrichs R Journal of Chemical Physics 1999 111 9183
(46) St Petkov P Vayssilov G N Liu J Shekhah O Wang Y Woell C Heine T Chemphyschem 2012
13 2025
(47) Gill P M W Johnson B G Pople J A Frisch M J Chemical Physics Letters 1992 197 499
(48) SCM Amsterdam Density Functional 2012
(49) Velde G T Bickelhaupt F M Baerends E J Guerra C F Van Gisbergen S J A Snijders J G
Ziegler T Journal of Computational Chemistry 2001 22 931
(50) Grimme S Journal of Computational Chemistry 2006 27 1787
(51) Seifert G Porezag D Frauenheim T International Journal of Quantum Chemistry 1996 58 185
(52) Elstner M Porezag D Jungnickel G Elsner J Haugk M Frauenheim T Suhai S Seifert G
Physical Review B 1998 58 7260
(53) Frauenheim T Seifert G Elstner M Hajnal Z Jungnickel G Porezag D Suhai S Scholz R
Physica Status Solidi B-Basic Research 2000 217 41
(54) Oliveira A F Seifert G Heine T Duarte H A Journal of the Brazilian Chemical Society 2009 20
1193
(55) Lukose B Supronowicz B Petkov P S Frenzel J Kuc A Seifert G Vayssilov G N Heine T
physica status solidi (b) 2011
(56) Heine T Rapacioli M Patchkovskii S Frenzel J Koester A M Calaminici P Escalante S Duarte
H A Flores R Geudtner G Goursot A Reveles J U Vela A Salahub D R deMon-nano edn ed deMon
2009
(57) Zhechkov L Heine T Patchkovskii S Seifert G Duarte H A Journal of Chemical Theory and
Computation 2005 1 841
top related