stability and transformation of particulate mgo-based nanocomposites stabilit¤t und

Stability and Transformation of

Particulate MgO-based Nanocomposites

Stabilität und Transformation von

MgO-Partikel-basierten Nanokompositen

Der Technischen Fakultät der

Friedrich-Alexander-Universität Erlangen-Nürnberg

zur Erlangung des Doktorgrades

DOKTOR-INGENIEUR

vorgelegt von

Amirreza Gheisi

aus Schiraz

Als Dissertation genehmigt von der Technischen Fakultät

der Friedrich-Alexander-Universität Erlangen-Nürnberg

Tag der mündlichen Prüfung : 08.10.2015

Vorsitzende des Promotionsorgans: Prof. Dr. Peter Greil

Gutachter: Prof. Dr. Oliver Diwald

Prof. Dr. Martin Hartmann

Abstract

i

Abstract

The stability of nanostructures and the preservation of their chemical and

physical properties in different chemical environments are critical for their proper

implementation in devices. However, the metastability and high surface and interface

energies specific to nanomaterials are found to impose property changes during

material synthesis or post-synthesis processes. In particular, for metastable metal

oxide nanoparticles, post-synthesis processes, storage and aging can lead to their

transformation and change of their characteristics such as crystal structure, chemical

composition, morphology and size. Transformation through either chemical reactions

or physical processes may also enable the modification of as-synthesized

nanomaterials into derivatives of desired composition and structure. The study of

metastability and transformation in metal oxide nanostructures is challenging and

many issues are still unknown. For such studies it is crucial to have model systems

as starting materials with defined properties. In this work some exemplary types of

vapor phase grown metal oxide nanoparticles are used. These materials constitute a

promising class of model nanostructures because of their narrow particle-size

distribution and defined chemical composition, crystal structure and morphology.

For this work three material systems with different levels of metastability are

produced, namely i) MgO and ZnO nanoparticles, ii) Zn-Mg-O and Fe-Mg-O

nanoparticles and iii) MgO nanocubes with surface adsorbed SixOyClz moieties.

Corresponding characterization work has been performed with focus on the influence

of different thermal treatments and chemical composition of the surrounding

atmosphere. The experiments addressing issues like a) the dependence of the

photoluminescence (PL) properties of vapor phase grown MgO and ZnO

nanoparticles on the surface composition as well as the composition of the

surrounding continuous phase, b) annealing induced changes including phase

separation in metastable ternary nanocomposite particles of Zn-Mg-O and Fe-Mg-O

systems, and c) the metastability of MgO nanocubes that were functionalized with

SixOyClz moieties and their transformation into magnesium oxychloride

Mg3(OH)5Cl∙4H2O fibers in the ambient.

Abstract

ii

a) In the first part of this study, the photoluminescence properties of two

prototypical metal oxide nanoparticle systems, MgO as an ionic insulator

and ZnO as an ionic semiconductor, have been discussed and compared.

Gaseous oxygen suppresses PL emission at MgO nanoparticles due to

deactivation of surface excitons, while it enhances the PL emission of

semiconducting ZnO nanoparticles showing adsorption dependent band

bending. This study shows also that annealing induced changes in surface

composition critically influence the luminescence.

b) Control over the distribution of chemical components and the identification

of synergistic effects inside composites are challenging tasks in the

functionalization of mixed-metal oxide nanoparticles. Annealing protocols in

vacuum or O2 atmosphere are very efficient for the purification of the

particles, i.e. for the removal of carbon based contaminants originating from

precursors. In ternary metal oxide systems, differences in ionic radii/ or

charge can induce the segregation of admixed cations to the particle

surface. In the second part of this work, these effects have been utilized to

produce two types of composite transition metal-magnesium oxide

nanoparticles. Zn-Mg-O nanocomposite particles have been prepared by

flame spray pyrolysis (FSP) in cooperation with Prof. Lutz Mädler and Dr.

Huanjun Zhang from University of Bremen. Based on structural properties

and chemical composition of resulting nanoparticles, optical absorption and

photoluminescence emission properties are discussed as a result of bulk

and surface excitons. For the Zn-Mg-O solid solutions, annealing induced

surface-segregation of Zn2+ ions depletes the MgO-specific PL emission

from particle surfaces. Moreover, it is found that the surface hydroxyls play

a significant role as protecting groups against oxygen as a PL quencher at

the solid-gas interface.

Metal-organic chemical vapor synthesis (MO-CVS) is used for the

generation of nanoparticles of the ternary Fe-Mg-O system. The effect of

annealing on composition, structure and optical properties of particle

powders of different compositions are explored. The types of

nanocomposites discussed range from solid solutions of Fe3+ ions in the

Abstract

iii

periclase structure of MgO exhibiting superparamagnetic properties to phase

separated components of periclase MgO and magnesioferrite MgFe2O4

phases with antiferromagnetic behavior. Moreover, absence of MgO-specific

PL emission on annealed Fe-Mg-O nanoparticles points to the effective

segregation of Fe3+ ions into the surface of the composites.

c) Finally, the phenomenon of spontaneous growth of magnesium oxychloride

Mg3(OH)5Cl∙4H2O needles in air and at room temperature has been

discovered and explored in great detail. It is found that MgO/SixOyClz

material system, formed during surface functionalization of MgO nanocubes

with SiCl4 and O2, is highly metastable upon contact with water vapor. The

results underline the critical impact of parameters such as SiCl4 adsorption

temperature, nature of the surrounding atmosphere, and MgO particle size

and dispersion degree on the aforementioned process of fiber growth.

Kurzfassung

v

Kurzfassung

Die Stabilität von Nanostrukturen und der Erhalt ihrer chemischen und

physikalischen Eigenschaften in chemisch verschiedener Umgebung sind von

entscheidender Bedeutung für ihre Anwendung in Bauteilen. Jedoch führen

Metastabilität und hohe Oberflächen- bzw. Grenzflächenenergie, welche spezifisch

für Nanomaterialien sind, zur Änderung von bestimmten Eigenschaften während der

Synthese und Nachbehandlung. Im Speziellen können Nachbehandlung, Lagerung,

und Alterung für metastabile Metalloxid Nanopartikel zur Umwandlung oder

Änderung ihrer charakteristischen Eigenschaften wie Kristallstruktur, chemische

Zusammensetzung, Morphologie und Größe führen. Transformation durch entweder

chemische Reaktionen oder physikalische Prozesse können aber auch zu

gewünschten Änderungen der Zusammensetzung und Struktur führen. Die

Untersuchung von Metastabilität und Umwandlung in Metalloxid Nanostrukturen stellt

eine große Herausforderung dar und viele Aspekte sind noch unbekannt. Für solche

Untersuchungen ist es entscheidend Modellsysteme mit bekannten Eigenschaften zu

haben. In dieser Arbeit wurden einige dieser modellhaften und in der Gasphase

hergestellten Metalloxid Nanopartikel genutzt. Diese Materialien stellen aufgrund

ihrer engen Partikelgrößenverteilung, ihren definierten chemischen Eigenschaften,

sowie ihrer Kristallstruktur und Morphologie eine vielversprechende Klasse an

modellhaften Nanostrukturen dar.

Für diese Arbeit wurden drei Materialsysteme verschiedener Stufen an

Metastabilität produziert, namentlich i) MgO und ZnO Nanopartikeln, ii) Zn-Mg-O und

Fe-Mg-O Nanopartikeln und iii) MgO Nanowürfel mit auf der Oberfläche adsorbierten

SixOyClz Clustern. Entsprechende Charakterisierungsarbeit wurde mit Fokus auf den

Einfluss verschiedener thermischer Behandlungen und der chemischen

Zusammensetzung der umgebenden Atmosphäre geleistet. Die Experimente

adressieren Themen wie a) Abhängigkeit der Photolumineszenz-Eigenschaften der in

der Gasphase hergestellten MgO- und ZnO-Nanopartikeln, welche die

Oberflächenzusammensetzung, sowie die Zusammensetzung der umgebender

Phase betreffen, b) durch thermische Behandlungen induzierte Änderungen wie

Phasentrennung in ternären Nanokompositpartikeln wie Zn-Mg-O und Fe-Mg-O und

c) die Metastabilität von mit SixOyClz Clustern funktionalisierten MgO Nanowürfeln

Kurzfassung

vi

und deren Transformation in Magnesiumoxychlorid-Fasern (Mg3(OH)5Cl∙4H2O) unter

Umgebungsbedienungen.

a) Im ersten Abschnitt dieser Studie werden die Photolumineszenz-

eigenschaften von zwei prototypischen Metalloxidnanopartikelsystemen,

MgO als ionischer Isolator und ZnO als ionischer Halbleiter, diskutiert und

anschließend miteinander verglichen. Gasförmiger Sauerstoff verringert die

PL Emission der MgO Nanopartikeln aufgrund von Deaktivierung von

Oberflächenexzitonen, wohingegen es die PL Emission der halbleitenden

ZnO Nanopartikeln, die ein adsorptionsabhängiges band bending zeigen

erhöht. Diese Studie zeigt, dass durch thermische Nachbehandlung

induzierte Veränderungen der Oberflächenzusammensetzung kritischen

Einfluss auf die Lumineszenz haben.

b) Kontrolle über die Verteilung von chemischen Bestandteilen und die

Identifizierung von synergistischen Effekten innerhalb der Komposite stellen

für die Funktionalisierung von misch-metalloxidischen Nanopartikeln eine

Herausforderung dar. Thermische Aktivierungsverfahren im Vakuum oder

unter Sauerstoffatmosphäre sind sehr effizient zur Reinigung der Partikeln,

z.B. um Kohlenstoff-basierte Verunreinigungen, die durch Präkursoren

entstanden sind zu entfernen. In ternären Metalloxid Systemen können

unterschiedliche Ionenradien oder -ladungen die Segregation von

zugegebenen Kationen zur Partikeloberfläche hin induzieren. Im zweiten

Teil der Arbeit, wurden diese Effekte verwendet um zwei Typen der

Übergangsmetall-Magnesiumoxid Nanopartikeln zu produzieren. Zn-Mg-O

Nanokompositpartikeln wurden durch Flame Spray Pyrolysis (FSP) in

Kooperation mit Prof. Lutz Mädler und Dr. Huanjun Zhang der Universität

Bremen produziert. Basierend auf strukturellen Eigenschaften und der

chemischen Zusammensetzung der entstandenen Nanopartikeln wurden die

optischen Absorptions- und Photolumineszenzemissionseigenschaften,

welche durch Bulk und Oberflächen Exzitonen hervorgerufen werden,

diskutiert. Oberflächen Segregation von Zn²+ Ionen aufgrund thermischer

Aktivierung von Zn-Mg-O Mischkristallen verringert die MgO spezifischen PL

Emissionen der Partikeloberfläche. Außerdem wurde herausgefunden, dass

Kurzfassung

vii

Hydroxylgruppen auf der Oberfläche eine wichtige Rolle beim Schutz gegen

Sauerstoff als PL-Quencher in der Feststoff-Gas Grenzfläche spielen.

Metall-organische chemische Gasphasensynthese (MO-CVS) wurde für die

Herstellung von Nanopartikeln des ternären Fe-Mg-O Systems verwendet.

Der Einfluss der thermischen Behandlung auf Zusammensetzung, Struktur

und optische Eigenschaften der Nanopartikeln verschiedener

Zusammensetzungen wurde erforscht. Die Typen der diskutierten

Nanokomposite erstreckt sich vom Mischkristall der Fe3+ Ionen in der

Periklasstruktur des MgO, welches superparamagnetische Eigenschaften

zeigt, bis hin zu separierten Bestandteilen von Periklas-MgO und

Magnesioferrit MgFe2O4 Phasen mit antiferromagnetischen Verhalten.

Außerdem weist ein Mangel an MgO-spezifischen PL Emissionen von

thermisch behandelten Fe-Mg-O Nanopartikeln auf eine effektive Fe3+ Ionen-

Segregation zur Oberfläche der Komposite auf.

c) Zuletzt wurden Phänomene von spontan wachsenden

Magensiumoxychlorid-Fasern (Mg3(OH)5Cl∙4H2O) an Luft und bei

Raumtemperatur entdeckt und im genaueren Detail erforscht. Es wurde

erwiesen, dass das MgO/SixOyClz Materialsystem, welches während der

Oberflächenfunktionalisierung von MgO Nanowürfeln mit SiCl4 und O2

geformt wurde, in Kontakt mit Wasserdampf sehr metastabil ist. Die

Ergebnisse betonen den kritischen Einfluss von Parametern wie z.B. die

Adsorptionstemperatur von SiCl4, die Art der umgebenden Atmosphäre, die

MgO Partikelgröße und den Grad der Dispersion auf den oben genannten

Prozess des Nanofaserwachstums.

Publikationen

ix

Publikationen

Aus Gründen der wissenschaftlichen Priorität wurden einige Ergebnisse dieser

Dissertation bereits veröffentlicht:

(1) Gheisi, A. Sternig, A. Redhammer, G. Diwald, O. Thin water films and

magnesium hydroxide fiber growth. RSC Advances 5, 82564-82569, (2015).

(2) Gheisi, A. Neygandhi, C. Sternig, A. Carrasco, E. Marbach, H. Thomele, D.

Diwald, O. O2 Adsorption dependent photoluminescence emission from metal

oxide nanoparticles. Physical Chememistry Chemical Physics 16, 23922-23929

(2014).

(3) Gheisi, A. Sternig, A. Rangus, M. Redhammer, G. Hartmann, M. Diwald, O.

Spontaneous growth of magnesium hydroxide fibers at ambient conditions.

Crystal Growth and Design 14 (9), 4236–4239 (2014).

(4) Zhang, H. Gheisi, A. Sternig, A. Müller, K. Schowalter, M. Rosenauer, A.

Diwald, O. Mädler, L. Bulk and surface excitons in alloyed and phase-separated

ZnO-MgO particulate systems. ACS Applied Materials & Interfaces 4 (5), 2490-

2497 (2012).

Acknowledgements

xi

Acknowledgements

The achievements and completion of this thesis would not have been possible

without the support of many people. I would like to acknowledge everyone who has

helped me along the way. First and foremost, I would like to thank Prof. Dr. Oliver

Diwald for giving me the opportunity to undertake my thesis in his group. I am

thankful for all of the hours Prof. Diwald spent giving me feedback and guidance in

my research. Prof. Diwald has been a mentor and he taught me priceless lessons in

both science and life. Furthermore, I want to thank Prof. Diwald for many valuable

advices and discussions regarding the writing part of this work.

It has also been a privilege to have Dr. Andreas Sternig as my colleague. He

has been a great teacher and a good friend. I am very thankful for all of his scientific

and technical advices and supports throughout my research. I also want to thank Dr.

Sternig for the TEM measurements and analysis provided by him in the course of

works presented in chapters 4, 7 and 8.

I would like to thank Prof. Dr. Johannes Bernardi from the University Center for

Transmission Electron Microscopy, Vienna University of Technology for the TEM

measurements and analysis which have been performed for the works presented in

chapters 6 and 10.

I am very grateful to Prof. Dr. Martin Hartmann and Dr. Mojca Rangus from

Erlangen Catalysis Resource Center, University of Erlangen-Nürnberg for solid state

MAS NMR experiments and the data evaluation presented in chapter 7. Especially, I

would like to thank Prof. Hartmann for reviewing this thesis.

I also want to thank Dr. Hubertus Marbach and Dr. Esther Carrasco from the

Chair of Physical Chemistry II, University of Erlangen-Nürnberg for collaborative work

and Auger Electron Spectroscopy (AES) measurements and analysis presented in

chapter 4.

Acknowledgements

xii

I thank Prof. Lutz Mädler and Dr. Huanjun Zhang from Department of

Production Engineering, University of Bremen for providing samples synthesized by

flame spray pyrolysis and collaborative work presented in chapter 5.

It is a pleasure to acknowledge collaborators from Department of Materials

Science and Physics, University of Salzburg. I thank Prof. Dr. Werner Lottermoser

and Mag. Gerold Tippelt for the Mößbauer measurements and the data evaluation

presented in Chapter 6. I also thank Prof. Dr. Günther Redhammer for the analysis of

X-ray diffraction patterns presented in chapters 6, 7 and 8.

I am also grateful for the opportunity to get to know and work with wonderful

group of colleagues from the working group and from the Institute of Particle

Technology. Especially I owe a huge debt of gratitude to colleagues who helped me

in my research and who assisted me in performing experiments; Dr. Stefan

Baumann, Dr. Michael Elser, Dr. Nicolas Siedl, Chris Neygandhi, Daniel Thomele,

Johannes Schneider, Paula Hoppe and Irina Merschmann.

I would also like to express my appreciation to the Austrian Science Fund

(FWF, project I-312) and the German Research Foundation (DFG, project 1613/ 2-1)

for the financial support of this work.

Finally, my sincere gratitude goes to my parents for all their love and

encouragement throughout my life and my studies. And, my deepest gratitude goes

to my loving, supportive and patient wife Mehrnoush whose faithful encouragement

during the last stages of my PhD is so appreciated.

Table of Contents

xiii

Table of Contents

1 Introduction ........................................................................................................ 1

1.1 Stability of Metal Oxide Nanoparticles ............................................................. 1

1.2 MgO Nanoparticles as Probes for Interfacial Changes .................................... 3

1.3 Zn/Fe-Mg-O Metal Oxide Nanocomposites ...................................................... 5

1.4 Scope and Structure of this Work .................................................................... 8

2 Spectroscopic Techniques .............................................................................. 11

2.1 UV-Vis Diffuse Reflectance Spectroscopy ..................................................... 11

2.2 Optical Absorption Spectra of Transition Metal Oxides .................................. 14

2.3 Photoluminescence Spectroscopy ................................................................. 18

3 Experimental Details ........................................................................................ 21

3.1 Sample Preparation ....................................................................................... 21

3.2 UV-Vis Diffuse Reflectance Spectroscopy ..................................................... 28

3.3 Photoluminescence Spectroscopy ................................................................. 30

4 O2 Adsorption Dependent Photoluminescence Emission from Metal Oxide

Nanoparticles ................................................................................................... 31

4.1 Abstract .......................................................................................................... 31

4.2 Introduction .................................................................................................... 31

4.3 Experimental Section ..................................................................................... 33

4.4 Results ........................................................................................................... 36

4.5 Discussion ..................................................................................................... 43

4.6 Conclusions ................................................................................................... 47

4.7 Supporting Information ................................................................................... 49

5 Bulk and Surface Excitons in Alloyed and Phase-Separated ZnO-MgO

Particulate Systems ......................................................................................... 55

5.1 Abstract .......................................................................................................... 55

5.2 Introduction .................................................................................................... 55


5.4 Results and Discussion .................................................................................. 60

5.5 Conclusions ................................................................................................... 73

5.6 Supporting Information ................................................................................... 75

Table of Contents

xiv

6 Fe-Mg-O Nanocomposite Particle Systems: Controlled Synthesis and the

Influence of Annealing on Composition, Structure and Optical Properties 79

6.1 Abstract .......................................................................................................... 79

6.2 Introduction .................................................................................................... 80


6.4 Results ........................................................................................................... 88

6.5 Discussion ................................................................................................... 106

6.6 Conclusions ................................................................................................. 109

6.7 Supporting Information ................................................................................. 111

7 Spontaneous Growth of Magnesium Hydroxide Fibers at Ambient

Conditions ...................................................................................................... 115

7.1 Abstract ........................................................................................................ 115

7.2 Introduction .................................................................................................. 115

7.3 Results and Discussion ................................................................................ 116

7.4 Conclusion ................................................................................................... 120


8 Vapor phase based and Water Film mediated Growth of Magnesium

Oxychloride Fibers ......................................................................................... 127

8.1 Abstract ........................................................................................................ 127

8.2 Introduction .................................................................................................. 127

8.3 Experimental Section ................................................................................... 129

8.4 Results and Discussion ................................................................................ 131

8.5 Conclusion ................................................................................................... 144


9 Summary ......................................................................................................... 151

10 Appendix I: Impact of Annealing Processes on MgO Nanoparticles Size

Distribution and Morphology ........................................................................ 155

10.1 MgO nanoparticle changes influenced by annealing conditions .................. 155

10.2 Mass Spectroscopy during Standard Annealing of MgO ............................. 161

10.3 Annealing Dwell Time Effect ....................................................................... 166

10.4 Summary and Conclusion ........................................................................... 172

Table of Contents

xv

11 Bibliography ................................................................................................... 175

1 Introduction

1

1 Introduction

1.1 Stability of Metal Oxide Nanoparticles

The study of nanomaterials1 is currently an active area of research in a broad

variety of scientific and technological fields, from physics, chemistry and materials

engineering to medical, electronic and computer sciences. Among different

nanomaterials, pure and mixed metal oxide nanoparticles are the most common

materials in the large number of technical applications, like catalysts, sensors,

constituents of microelectronic devices and optical coatings.2,3 The detailed

understanding of property changes of metal oxide nanoparticles, i.e. their stability, is

not only crucial for improvement of their performances but also for eventual design of

new materials. The fundamental knowledge of their chemical, structural and surface

characteristics is necessary but, on the other hand, is directly linked to the

experimental conditions during the synthesis of the nanoparticles as well as to the

stability and transformations during post-synthesis processing.

Due to the high surface-to-volume ratio, surface structure and composition and

resulting surface reactivity dominate metal oxide nanoparticles properties. Different

intrinsic (ion vacancies and interstitial atoms) and extrinsic (impurities) surface

defects may contribute significantly to the nanoparticles overall properties.4,5 For

instance, surface defects such as oxygen vacancies affect the photoluminescence

properties of nanoparticles like ZnO or surface adsorbed hydroxyl groups influence

the dispersion stability of metal oxide nanoparticles in various liquid media.6,7,8

However, the high dependence of properties on surface and interface implies

challenges with regard to materials control when exposed to different and sometimes

less defined environmental phases. For example, nanoparticles handled in ambient

air are automatically exposed to water and residual gas atmosphere that can result in

significant changes of the optical and structural properties.

1 Nanomaterials are defined as substances with any external dimention in the size range from

approximately 1 nm to 100 nm (nanoscale), or having internal or surface structure in the nanoscale.1

1 Introduction

2

Many characteristics of the metal oxide nanoparticles, such as surface

electronic states or melting temperatures, relate to their high surface-to-volume ratios

as well as to the prevailing existence of non-equilibrium phases, residual stresses,

large portion of defects and interfaces.4,9,10 In the case of metastable nanoparticles,

variation of parameters, such as temperature, pressure, time, radiation and chemical

environment can induce property changes during materials handling, storage and

processing. For instance, thermal activation of the nanoparticles at different

processing steps can lead to increasing of ion diffusion, alloying and phase

separation. Moreover, formation or annihilation of bulk and surface defects and the

associated gain or loss of the related properties can occur at elevated

temperatures.11–13

Thus, stability in terms of physical, chemical and mechanical behavior is of

great importance for effective application of metal oxide nanoparticles. Numerous

stability studies concerned with various aspects of metal oxide nanoparticles such as

chemical composition, shape, size, recrystallization, segregation and phase

separation have been carried out.14–19 In fact, stability is a broad issue that can be

discussed based on the behavior of nanoparticle ensembles in different thermal,

chemical, mechanical and atmospheric conditions.

In this work some exemplary types of pure and composite metal oxide

nanoparticles have been subject to different thermal treatment procedures. For these

nanoparticles annealing induced property changes, i.e. thermal stability during their

processing, play a significant role in the control of their size, shape, composition and

crystal structure. Indeed, many of the metal oxide nanoparticles are solids at

non-equilibrium state, but to understand their thermal stability at different

temperatures it is necessary to know about the thermodynamic phase equilibrium

and related phenomena, such as phase transformation temperatures at specified

pressures, miscibility of components and phase separation. However, it is known that

the thermodynamic properties of nanomaterials differ from those of bulk

materials.20-22 This means that for nanomaterials, the deviation of the phase

transformation processes from bulk materials must be considered. Because of the

wide range of possible metal oxide nanoparticle systems and thermal treatment

1 Introduction

3

conditions, there is no comprehensive study that can describe the kinetic and

thermodynamic properties of such systems. Moreover, thermal stability cannot be

understood and discussed without considering the bulk and surface impurities and

defects, processing time and the nature of the surrounding continuous phase, i.e.

vacuum, gas or liquid.

The well-facetted and highly dispersed MgO nanocubes obtained by chemical

vapor synthesis (CVS) and subsequent thermal annealing, represent an ideal metal

oxide for related studies.23–28 Because of the high thermal stability, MgO

nanoparticles can be treated at high temperatures in order to achieve dehydroxylated

nanoparticle surfaces. This generates a particularly well-suited starting situation to

track surface-related property changes of particulate MgO-based nanoparticles by

spectroscopy.

The aim of the present work is to characterize the influence of annealing and

constituents of surrounding atmosphere on the stability and transformation of the

MgO-based pure and composite nanoparticles. For this purpose, the variations in

physico-chemical properties such as chemical composition, crystal structure, size,

shape and optical properties have been investigated.

1.2 MgO Nanoparticles as Probes for Interfacial Changes

Magnesium oxide as an ionic insulator material (Ebg ≈ 7.6 eV) with rock salt

crystal structure is one of the most representative metal oxides. Particularly, in

nano-size regime the highly dispersed MgO nanoparticles, with cubic morphology,

offer high amount of low coordinated surface active sites such as corners and edges

(Figure 1.1). These surface features make MgO nanoparticles suitable probes for

surface and interface related studies both theoretically23–25,29 and experimentally26–28.

The optical characterization of the surface and interface of MgO nanoparticles

provides an opportunity to establish a correlation between topological surface

features and the corresponding chemical and spectroscopic properties.

1 Introduction

4

Figure 1.1: Schematic representation of topological surface features such as low

coordinated corners (3C), edges (4C). Adapted from reference [30].

Low-coordinated surface anions, which are considered to be chemically reactive

sites, give rise to specific optical transitions in the UV range, which can be assigned

to the formation of surface excitons.29 For surface-dehydroxylated MgO nanocubes

with average particle size of about 6 nm, UV-Vis diffuse reflectance spectra show two

absorption bands at energies far below the bulk absorption threshold of MgO

(7.6 eV). The absorption bands at energies of 5.2 eV (λ = 240 nm) and 4.6 eV

(λ = 240 nm) have been attributed to low-coordinated edges and corners,

respectively.31 Photoluminescence studies have been done to find a connection

between the nature of emission and absorption sites on MgO surfaces. It is shown

that for surface-dehydroxylated MgO nanocubes in vacuum (p < 10-5 mbar), the

photoexcitation of the edge and corner sites results in emission bands at 3.4 eV

(λ = 370 nm) and 3.2 eV (λ = 380 nm), respectively.32 These exclusively surface-

related spectroscopic fingerprints of MgO are the basis for the investigation of the

surface- and structure-property stability of MgO-based nanocomposites in different

chemical and thermal conditions applied in this work.

1 Introduction

5

1.3 Zn/Fe-Mg-O Metal Oxide Nanocomposites

The combination of two metals in an oxide matrix can produce composite

materials with novel structural and chemical properties that rely on synergistic

combination of single-component properties. Nanoparticles of metal oxide composite

materials are increasingly important in many areas of chemistry, physics and material

science.33–36 Among these material systems, nanocomposites containing transition

metals and magnesium oxides have attracted considerable attention because of

particularly their potential for optoelectronic, magnetic and catalytic applications.37–40

Admixture of transition metals (TM) at low concentrations to MgO yields

substitution of magnesium ions with TM cations within the rock-salt structure of

MgO.12,41,42 Annealing of MgO-based nanocomposites with concentrations above the

solubility limit of the admixed TMs in MgO is expected to lead to segregation of the

admixed cations to the particle surface potentially upon phase separation.11,12,43 The

segregation is derived from differences in ionic radii and/or charge of the transition

metal and magnesium ions and occurs to reduce free energy of the system.44–46

Beyond the consideration of ionic radii and charge, the relative electronegativity of

transition metal cations plays a role as well.47 As all transition metals are less

electropositive than Mg, the TM2+-O2- bond is less ionic. This additionally drives the

segregation of TM2+ ions from the bulk into sites at the surface of the nanoparticles,

where the electrostatic potential is weaker. However, the spatially controlled

distribution and arrangement of the transition metal ions in host matrix is highly

challenging and depend on chemical nature of the dopant, chemical environment,

processing temperature and time.12,48

In this work we have generated composite nanoparticles of the Zn-Mg-O and

Fe-Mg-O system to investigate the annealing induced structural and optical property

changes and, consequently, to identify the arrangement of the components in

particular regions of the solid.

1 Introduction

6

1.3.1 Zn-Mg-O Nanocomposites

Magnesium oxide (Ebg ≈ 7.6 eV) and zinc oxide (Ebg ≈ 3.4 eV) belong to

different group of materials with regard to their crystallographic and electronic

structures. The admixture of narrow bandgap semiconductor ZnO to the wide

bandgap insulator MgO provides means to generate tunable bandgap nanomaterials

for use in many fields including optoelectronics, display devices, gas sensors and

photocatalysis to name a few.35,49–52 Most of the related studies on such particle

systems are dedicated to ZnO wurtzite structures with small amounts of MgO

admixed, whereas considerably fewer studies report periclase MgO doped with ZnO.

Due to their defined shape and tunable optical and chemical surface properties such

composites provide novel building blocks for the growth of cubic ZnO

nanostructures.53,54

Different synthesis methods including the sequential implantation of Zn+ and O

-

ions in single crystalline MgO,55 the calcination of polymer/metal salt complexes56 or

reactive electron beam evaporation57 have been employed for the generation of cubic

ZnO structures. However, detailed synthesis and characterization studies on well-

dispersed powders of ZnxMg1−xO nanocrystals are scarce, due to the lack of

appropriate preparation techniques and reliable surface characterization approaches.

In a recent work, that is done in our group, it has been demonstrated that

monocrystalline ZnxMg1−xO nanocomposites with exceptional regular cubic shape

can be obtained using chemical vapor synthesis (CVS) and subsequent vacuum

annealing.12 Such particles are described as insulating cubes which have scaffolds of

a semiconducting component. However, the utilized approach does not allow the

production of nanocomposites with Zn2+ concentrations above 12 at.%.

For the realization of ternary Zn-Mg-O mixed metal oxide with higher Zn2+

concentrations, Zn needs to be added in a more effective and controllable way to the

nanocomposite system. As an alternative approach, in this work, in cooperation with

Prof. Lutz Mädler and Dr. Huanjun Zhang from University of Bremen flame spray

pyrolysis process has been employed to produce nanocomposite samples with

higher Zn2+ concentrations. The composition and surface electronic structure of the

1 Introduction

7

produced samples are explored based on the influence of annealing induced

segregation, phase separation and surface dehydroxylation effects.

1.3.2 Fe-Mg-O Nanocomposites

Nanocomposites comprising iron as transition metal embedded in magnesium

oxide matrix have attracted increasing attention for their excellent catalytic, magnetic

and optical properties.40,58 In technological applications they are used in magnetic

devices39, photovoltaics38, pigments59 and in the fabrication of different types of

catalysts, like catalysts for DeSOx processes60 and carbon nanostructures

synthesis61.

Many methods have been developed in order to produce Fe-Mg-O

nanostructures. Examples are in melt doping of MgO with iron and subsequent high

temperature oxidation62, high temperature annealing of powder mixtures of Fe2O3

and MgO63, combustion of iron and magnesium nitrates in urea solution64, sol-gel

approaches followed by high temperature calcination59, spray pyrolysis38, sputtering65

and hydrothermal procedure66.

The structural, chemical and surface/interface properties of such nanoparticles

critically depend on preparation techniques and their post synthesis treatments. Most

of the wet methods like sol-gel or hydrothermal synthesis have intrinsic

disadvantages with regard to their use for more detailed studies or advanced

applications. Reasons for that are undesirable bulk and surface impurities originating

from synthesis, poor control over cation concentration and broad particle size

distribution.

In this work we employed for the first time a modified metal organic chemical

vapor synthesis (MO-CVS) approach for the growth of Fe-doped magnesium oxide

nanocrystals. This method and subsequent annealing in controlled gas atmosphere

provide condition to produce Fe-Mg-O nanocomposite samples with distinct

morphology, defined distribution of size and desired composition. For the obtained

nanocomposites, structural-, and optical-property changes are explored. These

1 Introduction

8

property changes relate to the different admixed iron concentrations as well as

annealing induced surface segregation and phase separation effects.

1.4 Scope and Structure of this Work

The focus of this work is on the experimental investigation of changes in metal

oxide nanoparticles that are induced by i) annealing and ii) controlled changes of the

surrounding gas phase. In this regard, we extensively explored variations in optical

absorption, photoluminescence emission, bulk and/or surface composition, particle

morphology, size and crystal structure. Different metal oxides and composites

thereof, namely MgO, ZnO, Zn-Mg-O, Fe-Mg-O and MgO/SixClyOz have been

synthesized and investigated in great detail.

The first part is primarily concerned with the effect of surface composition and

oxygen adsorption on the photoluminescence (PL) properties of vapor phase grown

ZnO and MgO nanoparticles (Chapter 4). Annealing induced optical property

changes of these prototypical metal oxides will be discussed in terms of changes in

the surface composition.

The second part deals with the admixture of two types of transition-metals

(Zn2+ and Fe3+) into the MgO matrix using different gas phase synthesis approaches.

The objective of this part was to obtain the information about the surface

functionalization of the MgO by the admixture of a second metal ion component. For

the nanoparticles of the ternary Zn-Mg-O system the correlation between the surface

states and their PL emission properties are discussed (Chapter 5). For Fe-Mg-O

nanocomposites a new gas phase synthesis approach has been established and

used to explore a new class of materials where the effect of annealing on

composition, structure, morphology and optical properties can be studied in detail

(Chapter 6).

In the third part of this work MgO nanoparticles have been exposed to SiCl4 and

O2 in an attempt to cover MgO nanocubes surfaces with SiOx shells. As a result, in

the ambient MgO nanocubes with surface adsorbed SixClyOz moieties transform into

1 Introduction

9

Mg3(OH)5Cl∙4H2O nanofibers (Chapter 7). This part will shed light on the chemical

and physical factors leading to formation of MgO/SixClyOz system and the growth

mechanism of Mg3(OH)5Cl∙4H2O nanofibers (Chapter 8). Moreover, the thermal

stability of the obtained fibers is discussed.

Material’s characterization in terms of particle size distribution, crystal structure,

morphology, and specific surface area is carried out by the help of scanning electron

microscopy (SEM), transmission electron microscopy (TEM), powder X-ray diffraction

(XRD) analysis and nitrogen sorption. The main elemental analysis techniques

employed in this work for compositional analysis of nanomaterials, are inductively

coupled plasma optical emission spectroscopy (ICP-OES) and energy dispersive

X-ray (EDX) spectroscopy. Additionally, UV-Vis diffuse reflectance spectroscopy and

photoluminescence spectroscopy are used to characterize the optical properties of

the nanoparticle ensembles.

2 Spectroscopic Techniques

11


2.1 UV-Vis Diffuse Reflectance Spectroscopy

2.1.1 Basics

Ultraviolet-Visible (UV-Vis) spectroscopy is used to obtain absorbance spectra

of a wide range of sample types such as liquids, gases and solids. The absorption

behavior of solutions containing low amounts (c < 0.01 M) of dissolved molecules can

be described by Beer-Lambert law (Equation 2.1). This law states that the

concentration of a solute in a dilute solution is directly proportional to the sample

absorbance.

A = ɛ c d Equation 2.1

Where A is absorbance (no unit), ɛ the molar absorptivity (L mol-1 cm-1), c

concentration of the compounds or molecules in solution (mol L-1) and d is the path

length of the sample (cm). The term Absorption refers to the physical process of

absorbing electromagnetic radiation. In absorption, the energy of electromagnetic

radiation is absorbed and excites electrons from the ground state to the excited state

of the compound.

For solid samples the incident beam is not only absorbed or transmitted but

may be also reflected (Figure 2.1). The ratio of the reflected to the incident beam is

defined as the amount of reflectance and can be measured by reflectance

spectroscopy.

Figure 2.1: The interaction of light with solid samples.


12

The data provided by this technique is used to describe the optical absorption

behavior and electronic structures of the solid powder samples.

2.1.2 Reflectance Measurement of a Powder Sample

In order to determine the relative amount of radiation reflected from a sample at

any wavelength, reflectance spectroscopy is used. Reflectance consists of two

contributions:

1) Specular reflectance

2) Diffuse reflectance

The mirror-like reflection which is described by the law of reflection is called

specular reflectance. Interaction of electromagnetic radiation with electric field of

particles and molecules can also results in reflection of radiation in all directions. This

is called electromagnetic scattering and refers to diffuse reflectance.

Most samples display a combination of specular and diffuse reflectance.

Depending on the equipment used, it is possible to measure specular reflectance,

diffuse reflectance or total reflectance separately. Diffuse reflectance spectroscopy

concerns the diffusely reflected radiation collected and detected by the detector from

an irradiated sample.

If the wavelength of the incident beam is significantly larger than the surface

variations (roughness) or particle size (d, the diameter of the particle) diffuse

reflectance dominates (i.e. when λ >> d). In nanoparticle powders the radiation

scatters many times within the particle ensemble, which is known as multiple

scattering.67 Several theories have been developed to describe multiple scattering of

light.68 Among these, theory of Kubelka-Munk provides a way to transform

reflectance spectrum into absorbance spectrum and characterize the absorption and

scattering of the irradiated sample layer.


13

2.1.3 The Kubelka-Munk Theory

The theory developed by Kubelka and Munk in 1931 describes diffusely

reflected radiation based on absorption (K) and scattering (S) coefficients of a

continuum medium capable of both scattering and absorbing radiation.68 The theory

describes reflectance properties by differential equations for parallel infinitesimal

small layers of thickness dx. For a layer of thickness dx of small-particle ensembles

irradiated in the x direction (Figure 2.2) the downward flux (I) is decreased by

absorption (KIdx) and scattering (SIdx) processes. In turn, the upward flux (J) is

increased by scattering (SJdx) and decreased by absorption (KJdx).

Figure 2.2: Model for the Kubelka-Munk analysis of reflectance data.

The following differential equations can be derived for fluxes I and J:

dI / dx = - ( K + S ) I + SJ Equation 2.2

dJ / dx = ( K+ S ) J – SI Equation 2.3

In the case of an infinitely thick sample (which is true for the nanoparticle

powder samples investigated in this work) the reflectance is independent of thickness

and is given by R∞ (= Rsample / Rstandard). Under this condition the solution of these

equations is simplified and Kubelka-Munk equation at any wavelength becomes:

F(R∞) ≡ K/S = (1-R∞)2/(2 R∞) Equation 2.4

where F(R∞) is named remission or Kubelka-Munk function. According to this

equation the reflectance at any wavelength is a function of K/S ratio and not their

absolute values. The Kubelka-Munk theory is based on several assumptions and can

be applied only under following experimental conditions :69

i. The sample has high diffusion and low absorption.

ii. The sample thickness is sufficient to avoid loss of radiation by transmission.


14

iii. The absorption intensity is very weak (F(R∞) ≤ 1).

iv. Photoluminescence is absent.

Therefore the model holds when the particle sizes are comparable or smaller

than the wavelength of the incident light. By this assumption the scattering coefficient

shows little variations with wavelength over the range of interest and Kubelka-Munk

theory can be successfully used to convert the diffuse reflectance spectrum into a

spectrum that is proportional to an absorbance spectrum.

2.2 Optical Absorption Spectra of Transition Metal Oxides

2.2.1 Basic Concepts

Transition metals both in atomic or ionic form have partially filled d orbitals.70

This definition includes most of the elements in group 3 to 12 in the periodic table.

Table 2.1 shows the elements in the first row of these groups.

Table 2.1: 1st row of elements in group 3 to 12 in the periodic table.

group 3 4 5 6 7 8 9 10 11 12

element Sc Ti V Cr Mn Fe Co Ni Cu Zn

atomic no. 21 22 23 24 25 26 27 28 29 30

valence configuration

d1s2 d2s2 d3s2 d5s1 d5s2 d6s2 d7s2 d8s2 d10s1 d10s2

The oxidation states of the transition metals depend on their state of

complexation and how they exist in corresponding coordination compounds.

Coordination compounds are Lewis acid (electron acceptor) -base (electron donor)

complexes where the transition metal atom or ion acts as a Lewis acid and forms

coordinate bonds with the ligands. Most of the ligands are anions or neutral

molecules (e.g., O2-, OH-, CO, NH3) and the nature of metal-ligand bond can range

from covalent to ionic.71

The arrangement of the d orbitals and their energy differences in the above

mentioned coordination compounds is reflected in their absorption spectra.72 There

are two widely used models to describe the electronic structure of these

complexes:73,74


15

1) Crystal-field theory

2) Ligand-field theory

Crystal-field theory is based on the analysis of transition metal ions in solids and

only applies to ions in crystals. In this model the ligands are considered as point

negative charges which repel the d orbital electrons of the metal ion. The crystal-field

theory explains the splitting of the d orbitals into different energies and relates the

optical absorption spectra to this splitting. The splitting energy is called ligand-field

splitting parameter Δ.

In crystal-field theory the overlap of ligands and metal atom orbitals is not

considered. Ligand-field theory takes molecular orbital theory into account and

provides a more complete framework to interpret spectroscopic data and the origin of

d orbitals splitting. Both theories use the ligand-field splitting parameter Δ, to

correlate the spectroscopic properties. Ligand-field theory gives a more complete

description of the electronic structure of the complexes. However, for the transition

metal ions in crystals the crystal-field theory can be used to interpret the energies

and intensities of electronic transitions in a more straightforward manner.72

For transition metal complexes octahedral and tetrahedral coordinations

describe the most important ligand arrangements. According to crystal-field theory

characterizing an octahedral crystal field, ligands are placed as six point negative

charges in an octahedral array (Figure 2.3). These charges interact with the d orbitals

of central metal ion and split them into a lower-energy triply degenerate set (t2g) and

a higher-energy doubly degenerate set (eg).


16

Figure 2.3: The orientation of the five d orbitals arising from the ligands of an

octahedral complex: the degenerate eg and t2g orbitals are separated by ligand field

splitting parameter, Δo.72

Another possible arrangement of ligands in a transition metal complex is

tetrahedral coordination. A tetrahedral crystal field splits the d orbitals into two sets of

e and t2 orbitals where e orbitals have lower energy than t2 orbitals (Figure 2.4).

Figure 2.4: The orientation of the five d orbitals arising from the ligands of a

tetrahedral complex: the degenerate e and t2 orbitals are separated by ligand field

splitting parameter, ΔT.72


17

An octahedral complex has six metal-ligand bonding interactions. The complex

with this arrangement has lower energy than a tetrahedral complex with just four

metal-ligand bonding interactions. Because of this lower energy the transition metal

complexes present more octahedral than tetrahedral coordination in their

structures.75

2.2.2 Absorption Spectra

The magnitude of ligand field splitting relates to the energy of electronic

transitions and corresponds to absorption of UV-Visible spectrum. However, because

of the presence of electron-electron repulsions within the metal orbitals the

absorption energies are not directly determined by ligand-field splitting. One of the

outcomes of the electron-electron repulsion is that a single transition can lead to

different absorption bands.72

For the electron configuration of 3d5 the electronic ground state of Fe3+ free ion

is described as 6S. 6S is the spectroscopic term symbol describing the degenerate

states in a 3d5 metal. The superscript 6 denotes a sextet state when each electron

occupies an individual d orbital. In a free atom all 5 d orbitals are degenerate but in a

complex not all of them are degenerate and beside electron-electron repulsion the

energy differences between t2g and eg orbitals must be taken into account. When

Fe3+ ion is surrounded by ligands, in octahedral coordination, t2g and eg d orbitals of

Fe3+ ion are separated. The separated orbitals written as t2g3 eg

2 configuration give

rise to

6A1 as the complex ground state. 6A

1 is the spectroscopic term for the d orbital

electrons in octahedral structure of the complex and relates to the spectroscopic term

of Fe3+ free ion in its ground state.

The absorption spectra of transition metal complexes result from electronic

transitions within the d-orbitals (ligand-field transitions) and charge transfer from

metal to ligand or vice-versa. All the ligand field transitions occur between the ground

state of the complex (e.g. 6A1

for Fe3+ in octahedral complex) and states which arise

from different possible electronic configurations of t2g and eg. Charge transfer bands

are related to electronic transitions between ligand and metal orbitals.


18

Quantitative interpretation of electronic absorption spectra of transition metal

complexes is possible by extracting the value of ligand field splitting parameter and

its correlation to related electronic transitions. This is done by using energy level

diagrams known as Tanabe-Sugano diagrams (T-S) diagrams76 (Figure 2.5).

Figure 2.5: Schematic of Tanabe-Sugano diagram for high-spin Fe3+ oxides.77

In T-S diagrams energies of the various electronic states E of each electron

configuration and the ligand field splitting parameters Δ are expressed in units of

Racah parameter B. E/B is plotted vs. Δ/B taking the ground state term of free ion as

zero.72 The Racah parameter B measures the electron-electron repulsion among

electrons in d orbitals.73 Using these diagrams it is possible to identify absorption

bands in UV-Vis spectra and calculate ligand field splitting parameter of related

electronic transitions.77

2.3 Photoluminescence Spectroscopy

2.3.1 Basic Principles

Photoluminescence is the spontaneous emission of light from a molecule or a

solid under optical excitation. Photoluminescence spectroscopy is a nondestructive

and extremely sensitive technique to probe electronic states of samples.78 When the


19

energy of incident light on a material is sufficient, the photons are absorbed and the

electrons are excited. Excited electrons return to their ground states by releasing the

excess energy following a radiative or nonradiative process (relaxation). In the case

of radiative relaxation the emitted light is called photoluminescence (PL).79

The photoluminescence can be divided into two principle types of transitions:

1) Fluorescence

2) Phosphorescence

Fluorescence and phosphorescence result from internal transitions before

relaxation to a ground state. The difference between them is best explained

considering spin multiplicities. The photon emitted in a transition between states of

the same spin multiplicity (e.g., singlet-singlet or triplet-triplet transitions) is called

fluorescence and the photon emitted in transition between states with different spin

multiplicity (e.g., triplet-singlet transition) is called phosphorescence.80 Fluorescence

has very short lifetimes of about 10-8 s to 10-4 s, whereas phosphorescence usually

has much longer life times in the range of 10-4 s to minutes.

Photoluminescence spectroscopy is widely used in the characterization of

surfaces because of its sensitivity to different electronic states located near surfaces

and interfaces. 78 The PL emission energy and intensity provides information about

the quality of surfaces and interfaces.

2.3.2 Fluorescence Quenching

In radiative transition the part of emission energy which is not dissipated by the

vibrational relaxations may still be reduced by other processes. For solid-gas

interfaces the PL intensity further can be decreased depending on the ability of

surrounding lattice or the surrounding gas-phase molecules to accept electronic

energies. This reduction in intensity is called quenching and can occur by different

mechanisms, such as dynamic and static quenching.


20

Dynamic quenching happens by deactivation of excited-state luminophore upon

contact with molecules in surrounding atmosphere. The molecules are called

quencher and are chemically not altered. A wide variety of molecules can act as

dynamic or collisional quenchers. Molecular oxygen is one of the examples with

widely spaced vibrational levels which can accept large quantum of electronic energy

and quench the fluorescence. In static quenching the luminophores and quenchers

form nonluminescent complexes and quenching occurs in the ground state.81

3 Experimental Details

21


3.1 Sample Preparation

The list of samples produced for this work is given in Table 3.1. The Table also

indicates the synthesis methods used for the production of the samples and the

chapters in which the related results are discussed.

Table 3.1: List of metal oxide nanoparticle samples produced for this work, their

synthesis methods and related chapters.

Sample Synthesis Method Chapter

MgO Chemical vapor synthesis 4, 6, 7, 8, 10

Flame spray pyrolysis 5

ZnO Metal organic chemical vapor synthesis 4

Zn-Mg-O Flame spray pyrolysis 5

Fe-Mg-O Metal organic chemical vapor synthesis 6

In this chapter only the synthesis details of the MgO and ZnO samples

produced by chemical vapor synthesis (CVS) and Fe-Mg-O samples produced by

metal organic chemical vapor synthesis (MO-CVS) method are presented. The

experimental details of the samples produced by other methods like flame spray

pyrolysis (FSP) (MgO or Zn-Mg-O in chapter 5) or water-assisted transformation

(chapters 7 and 8) are given in related chapters.

3.1.1 Chemical Vapor Synthesis

Chemical vapor synthesis (CVS) is a so called “bottom-up” method to produce

nanoparticles from precursors in the gas phase. The method is called “MO-CVS”

when a metal organic compound is used as precursor. This technique allows the

production of pure and mixed crystalline metal oxide nanoparticles.


22

3.1.1.1 MgO Synthesis

The CVS apparatus used for the production of MgO nanoparticle samples

consists of two quartz glass tubes which are mounted concentrically inside a

cylindrical furnace (Figure 3.1). The inner glass tube contains two ceramic ships with

Mg metal pieces (1 g of Mg in each ship). Heating of the furnace to a defined

temperature (T = 913 K) allows adjusting the Mg-vapor pressure. An argon gas flow

(QAr = 1000 sccm)1 is led through the inner tube to transport evaporated metal atoms

to the reaction zone where Mg and Ar gas mixture meets oxygen (QO2 = 900 sccm)

coming from the outer quartz glass tube. The exothermic oxidation reaction leads to a

bright stable Mg combustion flame in the reactor and MgO nanoparticles are formed

as a result of homogenous nucleation in the gas phase. A rotary vane pump keeps

the reactor system at a constant low pressure (p = 50 ± 2 mbar). Thanks to

continuous pumping, the residence time of nuclei within the flame remains short

enough to prevent substantial coarsening and coalescence. The MgO nanoparticles

are deposited downstream in a stainless steel net that is kept at the room

temperature.

Figure 3.1: Schematic diagram of the CVS reactor setup that is used for the

production of MgO nanoparticles. (MFC: mass flow controller; T: local operation

temperature; P: pressure gauge)

A by-pass system allows avoiding particle collection during uncontrolled

process conditions, i.e. heating and cooling phase. Therefore the product is only

collected at steady state process condition when the bypass is closed and the main

1 Q = volumetric flow rate


23

way is opened. To finish the particle collection the furnace is turned off and cooled

with pressurized air, which is blown through the space between furnace and outer

glass tube.

3.1.1.2 ZnO Synthesis

ZnO nanoparticles were synthesized by the means of a two-hot-zone MO-CVS

reactor (Figure 3.2). The reactor system employed for this purpose consists of one

quartz glass tube, which is placed inside a heating coil (first heating zone) followed

by a ceramic tube furnace (second heating zone). The first heating zone of the tube

hosts a ceramic ship with 1 g of zinc acetate dihydrate (Zn(CH3COO)2·2H2O) powder

(≥ 99.0%, Sigma-Aldrich), which is heated to T1 = 523 K to sublimate the precursor.

An oxygen gas flow (QO2 = 650 sccm) is mixed with the precursor in the gas phase

and transports the metal organic vapor to the second heating zone of the tube where

the tube furnace provides a temperature T2 = 1073 K inside the glass tube. At this

position of the tube, zinc precursor is decomposed and ZnO nanoparticles are formed

as a result of oxidation and homogeneous nucleation in the gas phase.

Figure 3.2: Schematic diagram of the MO-CVS reactor setup that is used for the

production of ZnO nanoparticles. (MFC: mass flow controller; T: local operation

temperatures; P: pressure gauge)

Continuous pumping keeps the residence time of nuclei within the second

heating zone of the tube short enough to prevent substantial coarsening and

coalescence. The ZnO nanoparticles are deposited downstream in a stainless steel

net that is kept at the room temperature. The total pressure (p = 15 ± 2 mbar), as well


24

as flow rate and temperatures in the reactor are kept constant during the time of

nanoparticle collection.

3.1.1.3 Fe-Mg-O Nanocomposite Synthesis

For the production of Fe-Mg-O nanoparticles we developed a new MO-CVS

apparatus, which provides control over the concentration of iron in FeMgO

nanocomposite samples. The two-hot-zone reactor system (Figure 3.3) employed for

this purpose consists of two quartz glass tubes, which are mounted concentrically

inside a heating coil (first heating zone) followed by a ceramic tube furnace (second

heating zone).

Figure 3.3: Schematic diagram of the MO-CVS reactor setup that is used for the

production of Fe-Mg-O nanoparticles. (MFC: mass flow controller; T: local operation

temperatures; P: pressure gauge)

In the first heating zone the inner glass tube hosts a ceramic ship with 1 g of

iron (III) acetylacetonate (Fe(C5H7O2)3) powder (≥ 99.9%, Sigma-Aldrich), which is

heated to temperature T1 = 353 K, 363 K or 373 K to sublimate the iron precursor at

adjustable evaporation rates. An argon gas flow (QAr = 1200 sccm) is led through the

inner tube to transport the metal organic vapor to the second heating zone where the

furnace provides a temperature T2 = 913 K. At this position a ceramic ship with 1 g

Mg metal grains is positioned inside the inner tube. Here, the magnesium is

evaporated and mixed with iron precursor vapor. The vapor mixture is then

transported by the argon gas flow to the end of the inner glass tube. There the argon,

magnesium vapor and iron precursor vapor mixture meets oxygen (QO2 = 1200 sccm)

which is flowing through the outer glass tube. At this position of the reactor, the


25

exothermic oxidation reaction leads to a stable Mg combustion flame which

decomposes the iron precursor and Fe-Mg-O nanoparticles are formed as a result of

homogenous nucleation in the gas phase. Because of continuous pumping, the

residence time of nuclei within the flame remains short enough to prevent substantial

coarsening and coalescence. The MgO nanoparticles are deposited downstream in a

stainless steel net that is kept at the room temperature. The total pressure (p = 70 ± 2

mbar), as well as argon and oxygen flow rates and T1 and T2 heating zones

temperatures are kept constant during the time of nanoparticle collection.

3.1.2 Thermal Annealing

After production, the nanoparticle powders were transferred from the reactor

into quartz glass cells (Figure 3.4). This step has to be performed via short time

(~15 min) of contact to air. The cells provide the condition for sample processing and

measurements in defined gas atmosphere and are used for: i) thermal activation of

the sample, ii) optical spectroscopy experiments and iii) sample storage.

Figure 3.4: Quartz glass cells used for thermal activation of the sample, optical

spectroscopic experiments and sample storage.

In the CVS apparatus the samples are produced at reduced pressures and in

controlled gas atmosphere. However, transfer of the samples from the closed reactor

into the quartz glass cell that occurs via short exposure to the ambient and

incomplete precursor decomposition in the case of MO-CVS result in different types

of contaminations. In order to guarantee decontaminated particle surfaces and


26

remove organic residues the activation of the sample by thermal annealing in defined

gas atmosphere is necessary. This post-synthesis treatment can be carried out

before experiments aiming at the exploration of surface/interface properties of the

nanoparticles. The procedure of thermal annealing applied depends on sample

contamination. The here employed thermal annealing procedures are of two types:

1) annealing under vacuum condition and 2) annealing in the presence of oxygen.

Different modifications of these procedures are used by changing the final

temperature or the dwell times at each temperature.

3.1.2.1 Thermal Annealing in Vacuum

In a typical standard annealing program the cell containing as-synthesized

powder sample was first evacuated (p < 10-5 mbar) at room temperature and then

heated to T = 1173 K in 100 K-steps according to the program described in Table 3.2

Just shortly after reaching T = 1173 K sample was cooled down to T = 1123 K and

contacted by 10 mbar of oxygen for 10 minutes to remove organic contaminants.

After evacuation of the oxygen, temperature was again increased to T = 1173 K and

it was kept for 60 minutes at pressures less than 5·10-6 mbar. This thermal treatment

takes 6 to 8 hours (depending on vacuum pumps used and amount of annealed

MgO) and offers perfect conditions to have a clean metal oxide surface.

Table 3.2: Standard thermal annealing program used for the activation of MgO

nanoparticle samples.

temperature (K) rate (K /min) pressure (mbar) dwell time

(min)

373 5 < 1.0 · 10-5 * 0

473 5 < 1.0 · 10-5 * 0

573 5 < 1.0 · 10-5 * 0

673 10 < 1.0 · 10-5 * 0

773 10 < 1.0 · 10-5 * 0

873 10 < 1.0 · 10-5 * 0

973 10 < 1.0 · 10-5 * 0

1073 10 < 1.0 · 10-5 * 0

1173 10 < 1.0 · 10-5 * 0

1123 30 O2 pressure =10 10

1123 30 < 1.0 · 10-5 * 0

1173 30 < 5.0 · 10-6 60

*. Base pressure; before increasing to the next temperature, base pressure must be

reached.


27

3.1.2.2 Thermal Annealing in the Presence of Oxygen

This post synthesis procedure is used for the samples produced with MO-CVS

reactors. High oxygen pressure was applied to remove carbon based contaminants

originating from synthesis. A modification of this procedure is outlined in Figure 3.5.

First the cell containing a nanoparticle powder was evacuated down to a pressure of

p < 10-5 mbar at room temperature.

Figure 3.5: Temperature profile (solid line, left ordinate scale) and applied oxygen

pressure (dashed line, right ordinate scale) during annealing treatment in the

presence of oxygen. (r: heating rate; td: dwell time).

The respective sample was then heated to T = 373 K at a rate = 2.5 K min-1,

held at this temperature for 15 min (dwell time, td) and then again subjected to

evacuation to p < 10-5 mbar. Further annealing steps are carried out at p = 650 mbar

of oxygen. The powder was stepwise heated in oxygen atmosphere to T = 473 K

(r = 5 K min-1, td = 15 min), T = 673 K (r = 10 K min-1, td = 30 min), T = 873 K

(r = 10 K min-1, td = 180 min) and T= 1173 K (r = 10 K min-1, td = 60 min).


28

After each annealing step the sample was cooled to room temperature

(cooling time ≈ 30 min) followed by an evacuation (≈ 15 min) to a base pressure of

p < 10-5 mbar in order to remove water and CO2 as oxidation products.

3.2 UV-Vis Diffuse Reflectance Spectroscopy

UV-Vis diffuse reflectance measurements were carried out with a Perkin Elmer

Lambda 950 spectrometer by using an integrating sphere. In Figure 3.6 the

integrating sphere module of the diffuse reflectance setup is shown.

Figure 3.6: The optical design of the PerkinElmer integrating sphere module.

Adapted from Perkin Elmer Lambda 950 product catalogue.

3.2.1 Integrating Sphere

In this work an integrating sphere is used instead of the standard detection

module for UV/Vis absorption spectroscopy. The sample cuvette is placed behind the

sphere in sample holder. The beam reflected by the sample is reflected onto the

internal reflective surface of the sphere before reaching the detectors inside the

sphere. The sphere's internal surface is made of the polymer Spectralon®, which

offers high degree of diffuse reflectance approaching 100%. The use of Spectralon®

restricts the spectral range to 200 nm - 2500 nm.

The spectrophotometer is equipped with a double-beam sphere made of

Spectralon® with a diameter of 150 mm. The surface area of the ports (entrance or

exit passages for light beams) corresponds to 2.5% of the internal reflective surface

for the specified sphere. The detectors inside the sphere (a photomultiplier for the


29

visible range and a PbS for the NIR) are protected against direct reflectance by

Spectralon® baffles. These baffles are essential for ensuring the accuracy of the

measurements. When measuring reflectance, a specular flap can be used to exclude

specular reflectance so that only diffuse reflectance is measured. For the

measurements done in this work the specular flap was removed to exclude specular

reflectance caused by the quartz cuvette.

3.2.2 Measurement

Before doing measurements the baseline was recorded using a standard

spectralon sample. Spectralon is a fluoropolymer, which has the highest diffuse

reflectance of any known material or coating over the ultraviolet, visible, and near-

infrared regions of the spectrum. This was performed to have a nominal 100%

reflectance collection as baseline. Then the measurement of the samples was carried

out from λ = 200 nm to 2000 nm. The cuvette containing the particles was located in

the sample holder when the entire quartz glass cell (Figure 3.4 left) was in a dark

environment. Figure 3.7 shows that higher reflectivity values are recorded under

vacuum conditions due to luminescence emission from the sample. The figure clearly

reveals how the presence of O2 is decreasing the fluorescence intensity of MgO in

comparison to measurement in vacuum of 5·10-6 mbar. To prevent

photoluminescence emission by the sample, the cell was filled with 10 mbar O2 as a

PL quencher.

Figure 3.7: Reflectance spectra of MgO samples measured in vacuum p < 5·10-6

mbar (a) and in the presence of 10 mbar O2 (b) in order to omit luminescence effects.

The spectra are recorded at T = 298 K.

http://en.wikipedia.org/wiki/Fluoropolymer

http://en.wikipedia.org/wiki/Diffuse_reflection

http://en.wikipedia.org/wiki/Diffuse_reflection

http://en.wikipedia.org/wiki/Ultraviolet

http://en.wikipedia.org/wiki/Visible_spectrum

http://en.wikipedia.org/wiki/Near-infrared

http://en.wikipedia.org/wiki/Near-infrared


30

3.3 Photoluminescence Spectroscopy

For photoluminescence (PL) measurements a Fluorolog®-3 Model FL3-22

Luminescence Spectrometer was used. The double-grating excitation and emission

spectrometers of the instrument offer excellent performance in resolution, sensitivity,

and stray-light rejection. This system is perfect for highly scattering samples as

powders investigated in this work. The spectrometer is comprised of the following

components:

1) Excitation source which is a 450 W continuous wave xenon lamp

2) Double-grating monochromator in excitation and emission sides

3) Photomultiplier tube (PMT) as detector

3.3.1 Long-Pass Filters

Long-pass filters attenuate wavelengths shorter than a cut-on wavelength and

transmit longer wavelengths. For PL emission spectra acquisition carried out in this

work long-pass filters are used for two different cut-on wavelengths of λ = 295 nm

and λ = 395 nm on the emission side. The filters eliminate contributions of the first-

and second-order of the excitation light.

4 O2 Adsorption Dependent Photoluminescence Emission

31


from Metal Oxide Nanoparticles

4.1 Abstract

Optical properties of metal oxide nanoparticles are subject to synthesis related

defects and impurities. Using photoluminescence spectroscopy and UV diffuse

reflectance in conjunction with Auger electron spectroscopic surface analysis we

investigated the effect of surface composition and oxygen adsorption on the

photoluminescence properties of vapor phase grown ZnO and MgO nanoparticles.

On hydroxylated MgO nanoparticles as a reference system, intense

photoluminescence features exclusively originate from surface excitons, the radiative

deactivation of which results in collisional quenching in an O2 atmosphere.

Conversely, on as-prepared ZnO nanoparticles a broad yellow emission feature

centered at hvEm = 2.1 eV exhibits an O2 induced intensity increase. Attributed to

oxygen interstitials as recombination centers this enhancement effect originates from

adsorbate-induced band bending, which is pertinent to the photoluminescence active

region of the nanoparticles. Annealing induced trends in the optical properties of the

two prototypical metal oxide nanoparticle systems, ZnO and MgO, are explained by

changes in the surface composition and underline that particle surface and interface

changes that result from handling and processing of nanoparticles critically affect

luminescence.

4.2 Introduction

Defect engineering belongs to the important challenges in the development of

functional particle systems. Moreover, control over defect populations in a particle

powder is required to endow it with new and desirable properties. The corresponding

approaches are particularly demanding since the generation, stabilization or

annihilation of defects with functional or unwanted properties needs to be monitored

along the entire process chain. This spans a wide range beginning with the

production of particle powders using a variety of different synthesis routes, to particle

processing and integration into the device and, ultimately, to device operation which


32

corresponds to the exposure of materials to electric current, heat, radiation or

mechanical stress.

Motivated by the photoelectronic properties of ZnO nanomaterials and the rich

spectrum of related applications, there is a continuously growing number of

publications related to the research topic ‘‘ZnO nanoparticles and photoluminescence

properties’’. Recent research activities address the relationship between different

defect types and photoluminescence emission features.82–93 Although it is well

established that surfaces and interfaces dominate and control the properties of

nanomaterials,94 the chemical and physical nature of ZnO nanoparticles remains

unspecified in most cases. Considering the fact that minor changes in synthesis,

handling and processing can alter the surface properties of the particles, a

particularly unsatisfactory situation is created where an increasing number of

publications report discrepant results for the same nanoparticulate material.

In this paper we demonstrate that vapor phase grown ZnO nanoparticles, which

were grown under oxygen rich conditions, show bright photoluminescence emission

that is linked to oxygen interstitials in the near surface region. Characteristic

adsorbate-induced changes in band bending of semiconducting metal oxides

substantially enhance photoluminescence emission in O2 atmosphere. A powder of

ionic MgO nanoparticles with purely surface dependent optical properties, on the

other hand, represents a well-suited reference system in order to assess the impact

of the nanoparticle surfaces on photoluminescence intensity changes. This study

involves a semiconducting and an insulating oxide and underlines the determining

influence of sample history and, concomitantly, the nature of the nanoparticle

surfaces with regard to their photoluminescence properties. Moreover, it clearly

shows that differences in the nature and composition of the surrounding continuous

phase (vacuum, gas or solution) may lead to substantial variations in observed

photoluminescence intensities.

Bare and well-facetted nanoparticle surfaces that are free from any type of

surface adsorbate certainly represent an ideal starting point for related studies.

However, required sample activation approaches typically involve high temperature

treatment under vacuum. Under such conditions, metal oxide nanomaterials typically


33

undergo substantial particle coarsening and coalescence due to their limited thermal

stability. As a result, most nanoparticles deteriorate and transform into ill-shaped

microcrystalline materials displaying very complex surface and interface

characteristics.5 The aim of the present study is to document the influence of oxygen

in the surrounding continuous phase on the properties of as-synthesized metal oxide

particles in comparison to those which were subjected to moderate annealing.

4.3 Experimental Section

4.3.1 Material Synthesis

For the production of MgO and ZnO nanoparticles we used chemical vapor

synthesis (CVS). The details of the MgO production technique are given

elsewhere.95,96 ZnO nanoparticles were synthesized by means of a two-hot-zone

CVS reactor (see Figure S4.1 in Supporting Information) The reactor system

employed for this purpose consists of one quartz glass tube, which is placed inside a

heating coil (first zone) followed by a ceramic tube furnace (second zone). The first

zone of the tube hosts a ceramic ship with zinc acetate dihydrate powder (≥ 99.0%,

Sigma-Aldrich), which is heated to T = 523 K to sublimate the precursor. An oxygen

stream (650 sccm) is mixed with the precursor in the gas phase and transports the

metal organic vapor to the second zone of the tube where the furnace provides

T = 1073 K inside the glass tube. At this zone, the precursor is decomposed and ZnO

nanoparticles form as a result of oxidation and homogeneous nucleation in the gas

phase. The total pressure (15 ± 2 mbar) as well as the flow rate and temperatures in

the reactor are kept constant during the time of nanoparticle collection.

4.3.2 Annealing

After production and short contact time with air, the nanoparticle powders were

transferred into quartz glass cells, within which thermal sample activation was

performed. Annealing treatment is used according to a defined procedure

(Supporting Information, Figure S4.2) for dehydration, dehydroxylation and removal

of carbon based surface contaminants. At the beginning, the cell containing

nanoparticle powder was evacuated to p < 10-5 mbar at room temperature. The


34

respective sample was then heated to T = 373 K at a rate (r) of 2.5 K min-1, held at

this temperature for 15 min (dwell time, td) and then again evacuated to p < 10-5

mbar. Further annealing steps were carried out at p = 650 mbar of oxygen. The

powder was stepwise heated in an oxygen atmosphere to T = 473 K (r = 2.5 K min-1,

td = 15min) and T = 673K (r = 5 K min-1, td = 60 min). Before each new annealing step

fresh oxygen was added. After each annealing step the sample was cooled to room

temperature (cooling time ≈ 30 min) followed by an evacuation to a base pressure of

p < 10-5 mbar in order to remove water and CO2 as oxidation products (Supporting

Information, Figure S4.2).

4.3.3 Structure and Morphology

X-ray diffraction (XRD) measurements were performed on a Bruker AXS D8

Advance diffractometer using Cu Ka radiation (λ = 154 pm). For transmission

electron microscopic measurements with a Phillips CM300 UT TEM operated at

300 kV, small amounts of the metal oxide powders were cast on a carbon grid just by

immersing the sample grid into the dry powder.

4.3.4 Spectroscopy

The Auger electron spectroscopy (AES) measurements have been performed in

a UHV chamber equipped with a Leo Gemini electron column and a hemispherical

electron energy analyzer (Omicron Nanotechnology/NanoSAM). During the

measurements the base pressure is below 3∙10-10 mbar. Auger spectra were acquired

at primary electron beam energy of 5 kV and a beam current of 3 nA in the constant

retardation ratio mode. The AES are acquired in the direct form, i.e. as N vs. E. Three

or four surveys are acquired in different areas of each sample. Each spectrum has

been taken in a specific area of the surface only once and the beam is blanked

afterwards before moving the sample stage to a different location. A scanned area

mode is used for the AES acquisition in order to spread the electron dose (a

maximum value of 1.2 x 10-2 C/cm2 for Mg) over a whole area of 117 x 88 μm2 and

thus to effectively reduce the local electron dose. This procedure guarantees

minimization of unwanted electron beam induced effects and to achieve spatially

averaged chemical information of the provided samples.97,98 Apart from the


35

corresponding metal and oxygen elements, carbon has been detected in the samples

exposed to ambient air. The relative concentrations of each element cA can be

obtained by using the following formula:

where IA, IB and IC correspond to the peak areas for the three elements of interest

(O, C, Zn or Mg) or to the peak-to-peak intensities in the derivative spectrum. The

selected peaks for the oxides correspond to the main transitions KL2,3L2,3 for O and

Mg and L3M4,5M4,5 for ZnO.99,100 SA, SB and SC refer to the corresponding sensitivity

factors which were taken from reference 101. For proof of consistency the

concentration estimates have been derived from both the original spectra (via area

analysis) as well as from the correspondingly derived ones (via peak heights). The

peak areas were determined via linear background subtraction. The intensities (areas

or peak-to-peak heights) of the different elements in each survey were estimated and

the corresponding averaged values were determined and used to calculate the

concentrations according to the formula given above.

UV diffuse reflectance spectra were acquired at room temperature using quartz

glass cells with a Perkin-Elmer Lambda 950 spectrophotometer, equipped with an

integrating sphere. The reflectance spectra were converted into absorption spectra

using the Kubelka Munk transform procedure. For photoluminescence (PL)

measurements, a Fluorolog®-3 Model FL3-22 spectrometer with a continuous wave

450 watt Xenon arc lamp was used for excitation. The double-grating excitation and

emission spectrometers of the instrument offer excellent performance in resolution,

sensitivity, and stray-light rejection. This system is well-suited for strongly scattering

samples such as the nanoparticle powders investigated in this work. PL spectroscopy

was performed at room temperature using a quartz glass cell that guarantees

vacuum conditions better than 5∙10-6 mbar.


36

4.4 Results

4.4.1 Structure and Particle Size

ZnO and MgO nanoparticle powders were characterized both directly after

synthesis and after controlled annealing at T = 673 K (SI, Figure S4.3). Figure 4.1

shows representative X-ray diffraction patterns (XRD) for ZnO and reveals that all

diffraction peaks are consistent with those of the wurtzite phase. No additional

crystalline phases were observed. Annealing of ZnO nanoparticles results in a

narrowing of the diffraction features widths, which points to a volume increase of the

coherently scattering solid and, therefore, to particle coarsening.

Figure 4.1: X-ray diffraction patterns of ZnO nanoparticles (a) after synthesis; (b)

after annealing in oxygen (T = 673 K, p = 650 mbar O2). Vertical lines correspond to

the standard XRD pattern of wurtzite ZnO (JCPDS No. 36-1451).

From the full width at the half maximum (FWHM) of the diffraction peaks102,103

the average crystallite sizes were calculated to be 6 ± 1 nm and 10 ± 1 nm for ZnO

after synthesis and after annealing treatment, respectively (details of the procedure

are found in the Supporting Information). Figure 4.2 shows TEM images of the metal

oxide nanoparticles after synthesis (Figure 4.2 a and c) and after annealing (Figure

4.2 a and d). All samples are relatively homogeneous in terms of particle size and

shape (see also Figure 4.3).


37

The powders of vapor phase grown nanoparticles can be characterized as

ensembles of loosely agglomerated, more (MgO) or less (ZnO) facetted

nanoparticles. TEM images also show comparable particle sizes for both types of

samples.

Figure 4.2: TEM images of (a) ZnO nanoparticles after synthesis; (b) ZnO

nanoparticles after annealing in oxygen (T = 673 K, p = 650 mbar O2); (c) MgO

nanoparticles after synthesis and (d) MgO nanoparticles after annealing in oxygen

(T = 673 K, p = 650 mbar O2).

After vacuum annealing at T = 673 K the particle size distribution in the ZnO

nanoparticle powder remains narrow and peaks at an average particle size of 10 nm.

This value is perfectly consistent with the average crystallite domain size (see

above). MgO nanoparticles, on the other hand, exhibit a higher thermal stability. Their

average crystallite domain size of d = 6 ± 1 nm does not change upon annealing (SI,

Figure S4.4). While MgO is known for its low sinterability even at high temperatures,

ZnO exhibits a high sinterability with high grain growth even at lower temperatures.

These well-established phenomena arise from differences in the interface energetics


38

between the two metal oxides.104 Supply of thermal energy to the metal oxide

nanoparticle ensemble initiates ion diffusion and allows the grains to coarsen and to

reorganize towards thermodynamic equilibrium. As observed for vapor phase grown

ZnO nanoparticles, annealing temperatures higher than T = 673 K can induce

significant particle size disproportionation effects (Figure 4.3).5

Figure 4.3: Cumulative particle size distribution plots for ZnO nanoparticle after gas

phase synthesis and after annealing to T = 673 and T = 873 K under vacuum and a

dry oxygen atmosphere.

4.4.2 Surface Composition

After vacuum annealing to T = 673 K and at p < 10-5 mbar metal oxide

nanoparticles remain partially hydroxylated and may also retain residual surface

carbonates or adsorbed CO2. Previous photoelectron spectroscopy measurements

on MgO nanoparticles revealed that – irrespective of vacuum annealing procedures,

which were applied ex situ and prior to the measurement – MgO nanoparticles

instantaneously adsorb carbon dioxide and other carbonaceous species from the

air.105 On the highly dispersed MgO with enhanced surface basicity as compared to

ZnO this process occurs within a few minutes of exposure to air and the surface

contamination level is typically in the range between 20 to 30%. Systematic FT-IR

studies on highly dispersed MgO samples reveal that only after vacuum annealing

induced desorption at temperatures higher than T = 873 K, the concentration of

hydrogen bonded neighboring surface OH groups becomes negligibly small.106–109


39

Since the present study starts with as-synthesized nanoparticles we expect the

coexistence of non-crystalline surface hydroxides, physisorbed water as well as

surface hydroxyls. After annealing to T = 673 K, i.e. above temperature of Mg(OH)2

conversion into MgO, only chemisorbed hydroxyls remain.106

The contaminants on ZnO nanoparticle powders mostly originate from the

synthesis process, i.e. the thermal decomposition of zinc acetate dihydrate

[Zn(CH3COO)2•2H2O].110 Macroscopically, the lightly ochre color of the as-

synthesized ZnO nanoparticle powder (SI, Figure S4.3) changes upon annealing to

T = 673 K to white color. This change in color indicates that the organic remnants

from the precursor material became decomposed and eliminated during annealing.

Attenuated total reflection (ATR)-IR spectroscopic measurements of ZnO

nanoparticle powders in air (spectra shown in Figure S4.5) reveal weak absorption

related to surface hydroxyls and carboxylates and substantiate this point. A recent

FT-IR study on ZnO nanoparticles111 which were derived from similar precursor

materials revealed that IR active and surface adsorbed CO2 112,113 exhibit thermal

stabilities up to T = 1073 K.

We characterized the surface composition and stoichiometry of the ZnO

nanoparticles with Auger electron spectroscopy in more detail. The estimated

concentrations of as-synthesized ZnO nanoparticles as well as annealed ZnO

nanoparticle powders are provided in Table 4.1.

Table 4.1: Surface composition of ZnO nanoparticles as determined by Auger

electron spectroscopy

Sample Relative concentration

Zn O C

ZnO as-synthesized 0.38 0.57 0.05

ZnO annealed 0.43 0.51 0.05

A relative carbon concentration of 0.05 was found in both cases. The results in

Table 4.1 suggest that the oxygen content exceeds that of Zn in both samples and is

higher in the as-synthesized sample. Comparison of the Auger spectra (not shown)

with those in the literature reveals a good agreement in energy and position of the


40

respective features. For a reliable determination of the ZnO

nanoparticles’ stoichiometry we used the atomic ratio of O to Zn (i.e. the ratio of O

intensity to Zn intensity corrected by their respective sensitivity factors,

(IO/SO)/(IZn/SZn)) and compared these values with data reported for different faces of

atomically clean ZnO single crystals114 (SI, Table S4.1). As a conclusion, the

annealed ZnO nanoparticle samples were found to correspond to a stoichiometric

compound, while the as-synthesized ZnO contains excess oxygen above the

stoichiometric composition.

4.4.3 Optical Properties

4.4.3.1 UV-Vis Diffuse Reflectance Spectroscopy

Before as well as after annealing the MgO nanoparticle powders adopt a white

slightly bluish opalescent color. The related UV-Vis diffuse reflectance spectra show

absorption features at energies above hv = 5.0 eV and hv = 5.4 eV for vacuum

annealed and as-synthesized MgO nanoparticle powders, respectively. The shift in

the absorption threshold is attributed to the annealing induced transformation of

photoluminescence inactive surface hydroxides – present on the as-synthesized

nanoparticle sample – into the oxide as well as to the desorption of surface adsorbed

water from the particle surface. A fraction of low coordinated surface elements

become uncovered and addressable by light excitation with sub band gap energy.26

In comparison to MgO as an insulator – the band gap E = 7.8 eV corresponds

to a wavelength below λ = 200 nm – the band gap of ZnO (E = 3.4 eV) corresponds

to the absorption threshold at λ = 370 nm and is clearly observable in the spectra of

ZnO (Figure 4.4 a after synthesis and Figure 4.4 c after annealing).


41

Figure 4.4: UV-Vis diffuse reflectance spectra of (a) ZnO after synthesis; (b) MgO

after synthesis; (c) ZnO after annealing in oxygen (T = 673 K, p = 650 mbar O2);

(d) MgO after annealing in oxygen (T = 673 K, p = 650 mbar O2). The spectra were

acquired at T = 298 K and in the presence of 10 mbar O2 in order to omit

luminescence effects. The dashed vertical lines indicate the excitation energies

selected for the photoluminescence emission measurements.

4.4.3.2 Photoluminescence Spectroscopy

We chose the excitation energies for the two metal oxides on the basis of the

most intense PL emission intensities and used hvExc = 4.6 eV (λExc = 270 nm) as an

excitation energy for ZnO and hvExc = 5.2 eV (λExc = 240 nm) as an excitation energy

for MgO (Figure 4.4). Whereas the respective excitation energy for ZnO exceeds the

optical band gap, the 5.2 eV excitation light used for MgO exclusively addresses

localized defect states related to coordinatively unsaturated surface ions or

interfaces.30,115–118 Figure 4.5 compares the PL emission spectra obtained on ZnO


42

under vacuum (p < 10-5) and in an oxygen atmosphere (p = 10 mbar O2). Figure 4.5 a

corresponds to a ZnO powder sample after synthesis. Spectrum acquisition under

vacuum reveals a weak UV luminescence band centered at hvEm = 3.3 eV

(λEm = 376 nm).

Figure 4.5: Photoluminescence (PL) spectra of (a) ZnO after synthesis; (b) ZnO after

annealing in oxygen (T = 673 K, p = 650 mbar O2); upon excitation a hvExc = 4.6 eV,

the respective PL emission spectra are recorded either under vacuum (p < 10-5 mbar)

or in an O2 atmosphere (10 mbar) at T = 298 K.

In addition, there is a broad emission feature with a maximum at hvEm = 2.1 eV

(λEm = 590 nm) and of only small intensity. In an oxygen atmosphere the intensity is

enhanced, while the band in the UV region becomes completely extinguished. Figure

4.5 b shows the spectra for annealed ZnO nanoparticles. Both spectra –irrespective

of whether they were acquired under vacuum or in an oxygen atmosphere – do not

show any UV emission band. In comparison to the as-synthesized sample, they show

a red-shifted visible band centered at hvEm = 2 eV (λEm = 630 nm) of by a factor of 3-4

reduced intensity.

As the MgO nanoparticles have been subjected to the same annealing

procedure, their surfaces remain hydroxylated. Despite the fact that after synthesis

most coordinatively unsaturated surface elements remain covered by adsorbates, the

respective particle system already shows substantial photoluminescence (Figure

4.6 a), such as the band centered at hvEm = 3.2 eV (λEm = 390 nm). Vacuum

annealing at T = 673 K leads to an intensity enhancement by a factor of about five. In


43

contrast to the PL emission effects observed for ZnO, the MgO related bands are

perfectly quenched in the presence of gaseous oxygen (see blue in Figure 4.6).

Figure 4.6: Photoluminescence (PL) spectra of (a) MgO nanoparticles after

synthesis; (b) MgO nanoparticles after annealing in oxygen (T = 673 K, p = 650 mbar

O2). Samples are excited at hvExc = 5.2 eV and acquired at T = 298 K either under

vacuum (black line, p < 10-5 mbar) or in an O2 atmosphere (blue line, 10 mbar).

4.5 Discussion

The effect of O2 adsorption on the photoluminescence (PL) emission properties

of two prototypical metal oxides with high and comparable surface-to-volume ratios is

in the focus of this study. As a first important conclusion, the nature of the

surrounding gas phase has a substantial effect on the photo-excited states formed

inside a nanoparticle powder. Moreover, oxygen adsorption effects can disclose

valuable information about the location of underlying defects.119,120 Table 4.2

summarizes the two opposite trends observed for O2 adsorption on the PL emission

intensity of insulating MgO nanoparticles, on the one hand, and semiconducting ZnO

nanoparticles, on the other hand.

Table 4.2: O2 pressure dependent photoluminescence emission trends in metal

oxide nanoparticle systems

hvExc / eV hvEm/ eV Intensity change with O2 partial pressure Ref.

ZnO 4.6 2.1 ↑ This study

MgO 5.2 3.2 ↓ 115


44

The ZnO photoluminescence emission at hvEm = 2.1 eV (λEm = 600 nm)

corresponds to yellow light and – when observed on ZnO nanostructures – is

attributed to deep trap states related to oxygen interstitials.84,92,121,122 DFT stability

diagrams indicate that oxygen interstitials represent stable defects in ZnO structures

under oxygen rich conditions.85,122 A recent STM luminescence spectroscopy study

on atomically clean ZnO thin films92 has revealed that an emission feature at

hvEm = 2.1 eV (λEm = 595 nm) is subject to the growth conditions in the presence of

excess of atomic oxygen.

We observe a positive PL intensity dependence on the presence of gaseous

oxygen (Figure 4.5 a). The O2 adsorption effect proves the underlying defect’s

location in the near surface region, i.e. in the range of the depletion layer of the

semiconductor particle, where adsorbate induced surface potential changes and

band bending become active.120 While molecular oxygen acts for surface excited

states of MgO as a photoluminescence quencher (see below), the absence of PL

quenching in the case of ZnO nanoparticles rules out that the underlying excitonic

transition is localized at distinct defects direct at the nanoparticle surface.

Materials specific issues need to be included into the discussion of the optical

ZnO nanoparticle properties: the PL emission effect is strongest on the

as-synthesized sample derived from the thermal decomposition of Zn acetate

dihydrate.123 On such samples, AES analysis points to a surface contamination with

carbon that roughly corresponds to 5% (Table 4.1). As the nanoparticles emerge

from the oxidative decomposition of an oxygen rich Zn precursor, we assume that

they exhibit a surplus in oxygen. Taking into account the measurement uncertainty of

the quantitative Auger analysis, the as-synthesized ZnO particles are in fact enriched

in oxygen above the stoichiometric composition (SI, Table S4.1), whereas those

which – prior to the AES analysis – had been annealed to T = 673 K correspond to a

stoichiometric compound. The oxygen excess may also stem from oxygen containing

adsorbates such as surface carbonates or carboxylates. The photoluminescence

fingerprint, i.e. the emission band in the yellow light range, however, points to oxygen

interstitials inside the lattice of the vapor phase grown nanoparticles. Annealing to

T = 673 K initiates particle coarsening and, concomitantly, leads to crystallite domain

size increase from 6 ± 1 nm to 10 ± 1 nm (Figure 4.2 and Figure 4.3). This reveals


45

that ion mobility and mass transfer are significant at these temperatures. We

therefore attribute the PL emission intensity decrease in Figure 4.5 b to the thermally

induced annihilation of growth related lattice interstitials. MgO nanoparticles are

completely different in this respect. As an ionic insulator and a nonreducible metal

oxide we can expect a stoichiometric compound exhibiting a substantially higher

thermal stability with regards to sintering and in comparison to ZnO (Figure 4.1,

Figure 4.2, Figure 4.3 and Figure S4.2).

In ZnO photogenerated electrons and holes either recombine radiatively by a

direct band-to-band recombination mechanism producing the UV emission band

(Figure 4.5 a) or via a trap assisted mechanism upon emission of photons having

less energy than the optical band gap.

The effect of oxygen on the intensity of ZnO nanoparticles can be consistently

explained by band bending on ionic semiconductors.120 Adsorption of respective

acceptor molecules (Figure 4.7) enhances surface band bending and drives

photogenerated holes into the surface near region.121 We suggest that subsequent

hole trapping at oxygen interstitials as deep trap states enforces their recombination

with photogenerated electrons yielding yellow photoluminescence emission.

For powders of entirely dehydroxylated MgO nanocubes, it is well-established

that two absorption bands – far below the bulk absorption threshold of MgO

(7.8 eV) – are associated with corner (4.6 eV) and edge sites (5.2 eV).118 In addition,

two closely spaced photoemission bands at 3.2 eV and 3.4 eV are linked to the

photoexcitation of corners and edges, respectively.124,125


46

Figure 4.7: Schematics of the energy band diagram (not to scale) for ZnO

nanoparticles after synthesis containing oxygen interstitials. The vertical arrows in

downward direction depict the radiative recombination process that produces the PL

emission band at 2.1 eV (λ = 600 nm, Figure 4.5). O2 adsorption (right panel)

enhances surface band bending and drives photogenerated holes into the surface

near region. Subsequent hole trapping by deep trap states such as oxygen

interstitials Oi enforces trap assisted recombination and yields yellow

photoluminescence emission.

As an entirely new insight from this study, hydroxylated MgO particle systems,

which were investigated right after synthesis and prior to any activation treatment

under vacuum, show appreciable surface dependent photoluminescence. Consistent

with the effect of hydroxyls on the electronic structure of MgO surface elements,126

the respective emission feature with a maximum at hvEm = 3.2 eV is red-shifted in

comparison to those observed on adsorbate free particle surfaces. While in the case

of ZnO where the PL emission originating from surface near region is only indirectly

affected by the adsorption of electron acceptors or donors, the excitonic properties of

MgO nanoparticles involve localized excited states and are exclusively surface

related.

The ZnO precursor used for this study, i.e. Zn(CH3COO)2·2H2O, is widely

employed for ZnO nanoparticle synthesis in the gas phase as well as in

solution.127-133 The majority of these studies lack information on synthesis related

remnants and their potential influence on the optical material properties. This study

also involves a detailed AES analysis and shows that at least a surface fraction of 5%


47

related adsorbates survives high vacuum treatment at T = 673 K. While these

contaminants do not seem to affect the properties of the photoluminescing

nanoparticles, molecular oxygen in the gas phase does. It must be concluded that

every change in its sticking properties as well as in its concentration in the

surrounding continuous phase critically affects the photoluminescence emission

yield, irrespective of whether the surface excited state is directly located at the

surface (MgO) or in the depletion layer underneath the surface of the semiconductor

(ZnO).

Photoluminescence is a well-suited spectroscopic technique to perform

adsorption studies on semiconducting metal oxide particle systems.134–136 Thus, a

systematic investigation of synthesis related additives and contaminants belongs to

the characterization challenges associated with the identification of defects and other

material specific factors that determine the photoelectronic properties of ZnO

nanostructures. Irrespective of whether one deals with semiconducting or insulating

metal oxide nanoparticles, a reliable discussion of their photoluminescence

properties always requires us to address the nature and composition of the particle

interfaces as well as the composition of the surrounding continuous phase. This is

particularly important for defect engineering and for the stabilization of derived

functional properties in nanomaterials where synthesis and processing matters.

4.6 Conclusions

This work compares the photoluminescence of two prototypical metal oxide

nanoparticle systems with identical surface-to volume ratios, ZnO as an ionic

semiconductor and MgO as an ionic insulator. As inferred from Auger electron

analysis and photoluminescence emission, the growth of ZnO nanoparticles under

oxygen rich conditions generates oxygen interstitials which act as deep traps and

assist in radiative charge carrier recombination upon emission of yellow light

(hvEm = 2.1 eV). The presence of oxygen in the surrounding continuous phase was

found to have a critical and – for the two metal oxides – opposite influence on the

measured photoluminescence intensities. While it quenches photoemission from

surface excited states on MgO nanoparticles, it substantially enhances the emission

intensity on ZnO because of adsorbate-induced band bending across the


48

semiconductor interface. This study underlines that resolving controversies in the

reported optical properties of technically relevant metal oxide nanoparticles requires

a more complete documentation of structural and compositional properties of bulk

and interfaces and – in particular – the address of their photoluminescence

determining surface property changes.


49

4.7 Supporting Information

4.7.1 Schematic of the Reactor

Figure S4.1: Schematic of the reactor used for production of ZnO nanoparticles.

4.7.2 Annealing Treatment Procedure

Figure S4.2: Temperature profile (solid line, left ordinate scale) and applied oxygen

pressure (dashed line, right ordinate scale) during annealing treatment.


50

4.7.3 Photos of ZnO and MgO Powders

Figure S4.3: Photos taken from powder samples after synthesis: a) ZnO, b) MgO;

and after annealing in oxygen (T = 673 K, p = 650 mbar O2): c) ZnO, d) MgO.

4.7.4 Analysis of the XRD Reflexes Widths

Pseudo-Voigt functions were used to determine the full-width at half maximum

(FWHM) of the main reflexes and the average crystallite sizes were calculated with

the Scherrer equation102:

D = Kλ / [W cos(θ)]

K is a constant which depends on the particle morphology and varies from 0.89 to

1.39 rad. Here K = 1 was used, which corresponds to an average volume of the

apparent size D independently of a particular morphology.103 λ is the wavelength of

CuKα radiation (in nm), W is the full width at half-maximum (FWHM in radian), and θ

is the diffraction angle (deg.).


51

4.7.5 Powder XRD of MgO Nanoparticles

Figure S4.4: Powder XRD data of MgO nanoparticles after synthesis and after

annealing to T = 673 K.

4.7.6 FT-IR Spectroscopy

The infrared spectra of the samples were measured by the attenuation total

reflection (ATR) technique using a Varian FTS-3100 spectrometer. The FT-IR

experiments were performed under ambient conditions. A small amount of sample

powder was casted on the ATR crystal. A total of 64 scans were accumulated for

each spectrum to obtain a reasonable signal to noise ratio with a spectral resolution

of 4 cm-1.

Figure S4.5: Infrared spectra of a) Zinc acetate dihydrate (precursor used to

synthesize ZnO); b) ZnO nanoparticles after synthesis; c) ZnO nanoparticles after

subsequent annealing in oxygen (T = 673 K, p = 650 mbar O2). All the spectra were

recorded in air and at room temperature.


52

ATR-Infrared spectroscopy measurements were performed to investigate the

potential presence of synthesis related organic remnants or surface adsorbed

species which originate from oxidation of the precursor during annealing. An IR

spectrum of the precursor clearly shows absorption peaks between 1110 cm-1 and

1730 cm-1 that are attributed to the fingerprint of the stretching modes of the acetate

groups (COOH).137 In addition to the dihydrate of the precursor salt, also water from

ambient adsorbs on the sample powder and results in a broad absorption feature

between 2600 cm-1 and 3550 cm-1. It is noticeable that for the sample after synthesis

all the absorption band intensities decrease upon decomposition and oxidation

reaction of the precursor at 1073 K which corresponds to the temperature of the flow

reactor during ZnO synthesis. All features are completely removed from ZnO sample

in the course of annealing. Since the ATR-IR studies were carried out in the ambient

but shortly after synthesis and annealing treatment, the above results indicate that

the density of the acetate and hydroxyl groups remarkably decreases by synthesis

and post-treatment conditions applied.

4.7.7 Surface Composition Analysis by Auger Electron Spectroscopy

Before quantification and consistent with reference 101, all the spectra in this

study have been divided by the kinetic energy E. Since the intensity/ energy

response function (IERF) for AES instruments operated in the constant retardation

ratio is proportional to E∙Δ(E) ,i.e. proportional to the product of the kinetic energy by

the detector efficiency Δ(E) [see for example 101], an additional correction to E-1 is

necessary. This has been checked by using it with AES of a MgO sample previously

annealed at around T = 1073 K in the UHV system (to avoid the presence of

contaminations) and following the evaluation procedure with this stoichiometric

compound. A minor correction over the factor E-1 was necessary in the O region of

E-0.985 and none in the Mg, validating essentially the procedure. These small

corrections were included in the evaluations for ZnO in Table S4.1 though.


53

Table S4.1 compares the atomic O/ Zn ratios for the ZnO nanoparticle characterized

in this study with results obtained on different atomically clean ZnO single crystals

faces reported in reference.114,138

4.7.8 Photoluminescence spectroscopy

Figure S4.6: Photoluminescence (PL) spectra of a) ZnO after synthesis; b) ZnO after

annealing in oxygen (T = 673 K, p = 650 mbar O2); upon excitation a hvExc = 5.2 eV,

the respective PL emission spectra are recorded either under vacuum (p < 10-5 mbar)

or in an O2 atmosphere (10 mbar) at T = 298 K.

Table S4.1

sample atomic ratio

O/Zn

single crystal ZnO (from Ref.114)

Prism Face 1.15

Zinc Face 1.05

Oxygen Face 1.25

annealed ZnO Derivative analysis 1.29 ± 0.06

Area analysis 1.19 ± 0.14

as-synthesized ZnO Derivative analysis 1.41 ± 0.04

Area analysis 1.50 ± 0.12


54

Figure S4.6 shows the PL emission spectra of ZnO sample obtained upon

excitation with hvExc = 5.2 eV (λExc = 240 nm). Figure S4.6 a and b indicate for ZnO

that excitation at hvExc = 5.2 eV produces qualitatively the same emission feature as

excitation at hvExc = 4.6 eV (λExc = 270 nm).

Figure S4.7: Photoluminescence (PL) spectra of a) MgO after synthesis; b) MgO

after annealing in oxygen (T = 673 K, p = 650 mbar O2); upon excitation a hvExc = 4.6

eV, the respective PL emission spectra are recorded either under vacuum (p < 10-5

mbar) or in an O2 atmosphere (10 mbar) at T = 298 K.

Figure S4.7 a reveals the emission spectrum for well dispersed MgO

nanoparticles just after synthesis. Whereas fully dehydroxylated MgO nanocubes

excitation at hvExc = 4.6 eV (λExc=270 nm) leads to corner excitation and produces

one PL emission band centered at hvEm= 3.3 eV (λEm = 380 nm), here one excitation

process contributes to the photoluminescence emission. Subsequent annealing

treatment (T = 673 K and p = 650 mbar O2) on MgO changes the shape of spectra

due to partial dehydroxylation of the sample surface and a band centered at

λEm = 400 nm is present (Figure S4.7 b). Figure S4.7 a and b both prove that unlike

ZnO, PL emission of MgO is quenched in O2 atmosphere.

5 Bulk and Surface Excitons in Alloyed and Phase-Separated ZnO−MgO

55

5 Bulk and Surface Excitons in Alloyed and Phase-

Separated ZnO-MgO Particulate Systems

5.1 Abstract

The rational design of composite nanoparticles with desired optical and

electronic properties requires the detailed analysis of surface and bulk contributions

to the respective overall function. We use flame spray pyrolysis (FSP) to generate

nanoparticles of the ternary Zn-Mg-O system the compositions of which range from

solid solutions of Zn2+ ions in periclase MgO to phase separated particle mixtures

which consist of periclase (cubic) MgO and wurtzite (hexagonal) ZnO phases. The

structure and composition of the composite ZnxMg1−xO (0 ≤ x ≤ 0.3) particles are

investigated using X-ray diffraction and high-resolution transmission electron

microscopy, whereas UV diffuse reflectance and photoluminescence (PL)

spectroscopy are used for the investigation of their optical properties. Vacuum

annealing has been carried out to track the effects of stepwise elimination of surface

adsorbates on the photoexcitation and PL emission properties. We demonstrate that

for Zn0.1Mg0.9O particles, the admixed ZnO suppresses the MgO specific surface

excitons and produces a PL emission band at λEm = 470 nm. Although gaseous

oxygen partially reduces the emission intensity of hydroxylated particles, it leads to

entire quenching in completely dehydroxylated samples after vacuum annealing at

T = 1173 K. Consequently, surface hydroxyls at the solid-gas interface play a

significant role as protecting groups against the PL-quenching effects of O2. The

obtained results are relevant for the characterization of ZnO-based devices as well as

for other metal oxide materials where the impact of the surface composition on the

photoelectronic properties is usually neglected.

5.2 Introduction

Composite nanoparticle systems are increasingly important for applications that

rely on the controlled synergistic combination of single-component properties. To

understand their property changes during synthesis, processing and under operation

conditions in a device, the rational development of functional nanomaterials requires


56

careful evaluation of the physical properties in the bulk and at the interfaces.

Magnesium oxide and zinc oxide are very distinct materials with regard to their

crystallographic and electronic structures. The photoexcitation of highly dispersed

MgO and other alkaline earth oxides, for instance, leads to the generation of surface

excitons that depend on the chemical composition of the particle surface, and

therefore, can be employed as surface probes.26,32 On the other hand, the

semiconductor ZnO (bandgap Ebg ≈ 3.4 eV) with its bulk luminescence properties has

attracted great attention because of its potential for UV light-emitting diodes (LEDs),

lasers and various other optoelectronic applications.84,139,140 Moreover, admixing ZnO

into MgO (Ebg ≈ 7.6 eV) provides means for engineering the bulk electronic

structure.35,49,50 Most of related studies on such particle systems are dedicated to

ZnO wurtzite structures with small amounts of MgO admixed, whereas considerably

fewer studies report periclase MgO doped with ZnO. A promising aspect of such

composites is related to the metastable cubic ZnO phase53 that is expected to be

compatible for p-type doping.141 Different synthesis methods including the sequential

implantation of Zn+ and O+ ions in single-crystalline MgO,55 the calcination of

polymer/metal salt complexes56 or reactive electron beam evaporation142 have been

employed for the generation of cubic ZnO structures. But detailed synthesis and

characterization studies on powders of well−dispersed ZnxMg1−xO nanocrystals are

scarce, because of the lack of appropriate preparation techniques and reliable

surface characterization approaches, which are crucial to understand the material

properties related to the gas−solid interface.

In a previous work we have shown that monocrystalline ZnxMg1−xO (x denotes

the molar ratio between Zn and (Zn+Mg)) nanocubes of exceptional regular cubic

shape and edge lengths below 25 nm can be produced by chemical vapor synthesis

(CVS).12 In line with ab initio calculations, the annealing induced Zn2+ segregation

into low coordinated surface sites of MgO nanocubes was tracked with UV diffuse

reflectance, FT−IR and PL spectroscopy. We observed completely new PL emission

features which were perfectly quenched upon exposure to gaseous oxygen,

demonstrating that corresponding excitation and subsequent radiative deactivation

processes are directly linked to excitons formed at the surface of the composite

nanocubes. Higher ZnO loadings with concentrations above the solubility limit of Zn2+

in MgO are expected to lead to particle systems where annealing induced phase


57

separation143 would generate highly dispersed ZnO deposits in contact with thermally

stable MgO based support particles. Such an approach would provide particulate

model systems where a systematic comparison between photoexcited surface states

on MgO as well as electronic transitions induced in ZnO can provide substantial

insights into stability and surface electronic structure of pure and composite ZnO

nanomaterials. However, as a major shortcoming of the direct combustion of Zn and

Mg vapors (CVS), an upper concentration limit of approximately 12 at % ZnO results

from the fact that during the combustion process Zn vapor cools down the flame and

in this way prevents the reproducible metal combustion over longer periods of time.

For the realization of composite ZnO/MgO systems with higher Zn2+ concentrations,

longer production times and higher yields, Zn vapor needs to be added in a more

effective and controllable way to the combustion zone. As an alternative approach

the versatile flame spray pyrolysis process can be scaled up to production

rates of kg h−1.144–146 ZnO and MgO nanoparticles were successfully produced with

flame spray pyrolysis in the past. However, there exists only one report on the flame

spray synthesis of ZnxMg1−xO nanoparticles which focuses on Mg−doping of ZnO and

is limited to the description of basic powder properties.147

This paper has three major objectives: first, we want to explore the potential of

flame spray pyrolysis (FSP) with respect to the generation of particles of the ternary

Zn-Mg-O system the compositions of which range from solid solutions of Zn2+ ions in

periclase MgO12 to mixtures where phase separation into a MgO rich periclase and a

ZnO rich wurtzite phase occurs. Second, we want to study the influence of annealing

induced surface dehydroxylation and removal of surface contaminants on the surface

electronic structure of ZnO based particle systems. As shown in previous papers32,148

the in depth characterization of interfacial effects requires treatment and

measurements of nanoparticle powders in defined gas atmospheres such as high

vacuum (p < 1∙10−5 mbar) or oxygen atmosphere. The third objective of this study

aims at the identification of potential synergistic effects between ZnO and MgO that

originate from surface and bulk doping, as well as from segregation and phase

separation effects.


58


5.3.1 Chemicals

Anhydrous magnesium(II) acetylacetonate (Strem Chemicals, 98%) and zinc(II)

naphthenate (Strem Chemicals, 65% in mineral spirits in naphthenic acid) were used

as received as the doped MgO. A mixture of xylene (BDH, ≥ 98.5%) and ethanol

absolute (RDH, ≥ 99.8%) was used as solvents for preparing the precursor solutions.

To increase the Mg(II) concentration in the precursor solutions, 2-ethylhexanoic acid

(RDH, ≥ 99%) was used to dissolve the Mg(II) salt.

5.3.2 FSP Synthesis

Zinc-doped MgO (ZnxMg1−xO) powders were synthesized in a flame spray

reactor. Detailed description of the setup can be found elsewhere.149 For the

preparation of the precursor solutions, 56.8 g of Mg(II) acetylacetonate was dissolved

in a mixture of 2-ethylhexanoic acid (167.2 mL), ethanol absolute (100.0 mL) and

certain amount of xylene, to make a final volume of 500.0 mL, corresponding to an

Mg(II) concentration of 0.5 mol/L. The Zn(II) solution (0.5 mol/L) was made by diluting

81.7 g of Zn(II) naphthenate (in naphthenic acid as received) with xylene to a final

volume of 250.0 mL. The Mg(II) and Zn(II) solutions were mixed at chosen volume

ratio to achieve the desired Zn(II) doping levels while keeping the total metal

concentration of 0.5 mol/L in the precursor solutions.

During flame spray pyrolysis experiments, the mixed precursor solution was fed

by a syringe pump (KD Scientific) at a fixed flow rate of 5.0 mL/min. The liquid was

dispersed into fine droplets by O2 gas (5.0 L/min, 1.5 bar) at the spray nozzle exit.

The spray was ignited by a CH4/O2 (1.5 and 3.2 L/min, respectively) supporting flame

to form a self−sustained flame. The flow rates of all gases were controlled by

calibrated mass flow controllers (Bronkhorst High-Tech). The generated ZnxMg1−xO

particles were directed by a vacuum pump and collected on a water-cooled glass

fiber filter (Whatman GF−6) placed 450 mm above the nozzle. After being collected

from the filter surface, the powders were sieved to remove the glass fibers.


59

5.3.3 XRD, N2 Physisorption, HRTEM, EDS, and HAADF−STEM

Characterizations

Powder X-ray diffraction analysis was carried out on a PANalytical X’pert

diffractometer with Cu Kα radiation source. Silicon zero−background plates were

used as the sample holder. The scanning program covers a 2θ range between 15°

and 140° at a step size of 0.03°. Rietveld refinements of the XRD patterns were

conducted using commercial refinement software (Bruker AXS Topas 4.2) to analyze

the crystal structure and quantify the phase concentration of the products. The

Brunauer−Emmett−Teller specific surface area (BET SSA) of the as−synthesized

ZnxMg1−xO powder was measured through N2 physisorption at liquid nitrogen

temperature using a Quantachrome Nova analyzer. The samples were degassed at

T = 473 K for 4 h before dosing with N2.

High-resolution transmission electron microscopy (HRTEM) was conducted on

an FEI Titan 80/300 electron microscope operated at 300 kV to investigate the

morphology and the microscopic elemental composition of the as−synthesized

powders. In particular, energy dispersive X-ray spectra (EDS) were collected from the

samples in the scanning transmission electron microscopy (STEM) mode, in which

Z-contrast imaging was performed with a high-angle annular dark-field (HAADF)

detector. The EDS spectra were used to analyze the spatial distribution of Mg, Zn,

and O elements on the length scale of a few hundred nanometers. Electron

diffraction was also performed to study the crystal structure of the samples. The TEM

samples were prepared by dispersing the as-synthesized particles in isopropanol

using an ultrasonic bath. Several drops of the suspension were dropped onto carbon-

coated copper grids, which were then subjected to vacuum storage before TEM

experiments.

5.3.4 Vacuum Annealing and Optical Investigations

To investigate the dependence of their optical properties on the surrounding

atmosphere, we transferred the powder samples into quartz glass cells, within which

thermal activation of the powders and spectroscopic measurements were performed

in high vacuum (at base pressure < 5∙10−6 mbar unless otherwise specified). A

typical procedure for dehydration, dehydroxylation and the removal of carbon−based


60

surface contaminants is as follows: the as-synthesized powders are heated to

T = 1123 K in high vacuum at a rate of 5 K min−1 and are then brought into contact

with 10 mbar O2 at this temperature. Subsequently, the sample−cell temperature is

raised to T = 1173 K and is kept at this temperature for 30 min before being cooled

down to room temperature.

UV diffuse reflectance spectra were acquired in the presence of 10 mbar of O2

using a Perkin-Elmer Lambda 950 spectrophotometer equipped with an integrating

sphere and then converted to absorption spectra using the Kubelka−Munk

transformation. PL spectra were measured on a Horiba Jobin Yvon Fluorolog−3

system (FL3−22) using a continuous wave 450 W xenon arc lamp for excitation. The

spectrometer is equipped with a double monochromator in emission and excitation to

guarantee optimal stray-light rejection. After vacuum annealing, the processed

powders were further characterized by N2 physisorption and TEM.

5.4 Results and Discussion

5.4.1 Characterization Results

5.4.1.1 Powder XRD

The powder XRD patterns of the as-synthesized ZnXMg1−XO samples are

shown in Figure 5.1. The XRD pattern of the as-synthesized MgO shows a periclase

structure with the most intense diffraction peak (002) at 42.96 . The relative positions

of other weaker peaks of this sample also match the periclase space group (fm3 m),

although with some small shifts in 2θ. Rietveld refinement results in an average

crystallite size of 9.9 nm (Table 5.1). With Zn doping at a loading of 10% (x = 0.1),

the XRD pattern is rather similar to that of pure MgO. No signature of other crystalline

phases including wurtzite ZnO can be identified, indicating a solid solution of

Mg-Zn-O with Zn being well blended into the lattice structure of periclase MgO during

the flame spray pyrolysis process. Rietveld refinement results in an average

crystallite size of 10.5 nm, slightly larger than that of the pure MgO sample.


61

Figure 5.1: Powder XRD patterns and Rietveld refinement results from FSP-made

ZnXMg1-XO samples. The raw XRD data, the refined patterns and their differences

are presented in black, red and blue colors, respectively. Diffraction peaks for cubic

MgO and hexagonal ZnO are also shown as bars to indicate their 2θ positions.

When the Zn loading is further increased (x = 0.3), diffraction peaks start to

emerge at 31.93°, 34.14°, 47.44°, 56.89°, and 68.04°, corresponding to (010), (002),

(012), (110), and (112) diffractions of wurtzite ZnO. The peak at 36.66° should be

contributed by both ZnO (011) and MgO (111) diffractions, if we compare the

intensity here with those at x = 1 and x = 0.9. These results show that at a high Zn

loading (30%), a certain amount of ZnO is phase-separated from periclase MgO to

form wurtzite structures. Rietveld refinement gives an average periclase crystallite

size of 10.4 nm and wurtzite crystallite size of 9.8 nm. Quantitatively, a wurtzite

concentration of 5.9% in weight percentage was estimated. Therefore, the majority of

Zn is blended into the periclase MgO lattices. The nominal formula for this solid

solution is calculated as Zn0.27Mg0.73O, corresponding to a wurtzite-to-periclase molar

ratio of 1:25 (calculation details in the Supporting Information).

It is also observed that as the Zn loading increases, the diffraction peaks of the

periclase phase slightly shift to the lower angle. Based on Bragg’s law (nλ = 2d sin θ),

the d-spacing and hence the lattice constant are expected to increase. As shown in

Table 5.1, the a value increases from 4.22 Å for pure MgO to 4.24 Å for Zn0.27Mg0.73O

accordingly. This lattice expansion can be explained by the slightly larger ionic radius


62

of Zn2+ (0.74 Å) than that of Mg2+ (0.72 Å) in the periclase phase.150 Such a trend

further supports that Zn2+ has been incorporated into the Mg2+ sites in the cubic

lattice as the Zn loading increases, leading to a monotonous, although small,

expansion of the unit cell. On the other hand, the lattice constants of the segregated

wurtzite phase at x = 0.3 were also found to differ slightly from those of pure

FSP−synthesized ZnO (XRD and Rietveld refinement data not shown here). The

lattice constants in the a and c-axis of the separated wurtzite phase are 3.24 Å and

5.22 Å, respectively, indicating a slight compression in the a-axis and expansion in

the c−axis, as compared to the refined 3.25 Å (a) and 5.21 Å (c) of the pure ZnO

sample. Such a change is consistent with the reported trend for Mg-doped ZnO thin

films.151,152

Table 5.1: Summary of the results from powder XRD and N2 physisorption analyses

on ZnXMg1-XO.

a Calculated from Rietveld refinement of the powder XRD patterns. b Calculated using

the formula dBET = 6/(ρSSA), where ρ is the molar-average density of the samples

based on the densities of cubic ZnXMg1-XO and hexagonal ZnO, g m−3. c The ZnO

crystallite sizes are not calculated because no diffraction peaks corresponding to the

wurtzite phase were observed from the powder XRD profiles.

Although the wurtzite-structured Mg-Zn-O solid solutions (mostly epitaxially

grown thin films) have been actively investigated during the past decade,151,153–155

their counterparts, i.e., the cubic-structured solid solutions, have been much less

addressed, especially in the powder or particle form. The solubility limit of ZnO in the

FSP-made samples measured here, 27% (molar percentage), closely matches that

predicted by MgO-ZnO phase diagram as reported by Segnit and Holland.141 In their

samples

after synthesis after vacuum annealing

dXRD (nm)a SSA

(m2/g)

dBETb

(nm)

aMgOa

(Å)

dXRD (nm)a SSA

(m2/g) dBET

b (nm) MgO

phase ZnO

phase MgO

phase ZnO

phase

MgO 9.9 N/Ac 190 8.9 4.22 10.6 N/Ac 150 11.2

Zn0.1Mg0.9O 10.5 N/Ac 172 8.9 4.22 11.9 N/Ac 110 13.9

Zn0.3Mg0.7O 10.4 9.8 143 9.1 4.24 49.1 54.0 17 76.1


63

work, a solubility limit of 28.8% for ZnO at 1000 °C in the cubic Mg-Zn-O solid

solution was estimated.

5.4.1.2 N2 Physisorption

The results of N2 physisorption (7-point BET) analysis on the FSP−made

ZnXMg1−XO powders are also shown in Table 5.1. As the Zn loading increases from 0

to 30% (molar), the specific surface area decreases from 190 m2/g to 143 m2/g. On

the basis of the formula dBET = 6/(ρSSA), the average particle size is estimated to be

8.9 nm, 8.9 nm, and 9.1 nm for 0%, 10%, and 30% Zn-doped MgO samples,

respectively. These results indicate that introducing Zn into MgO does not

significantly change the average particle size. The particle sizes estimated by N2

physisorption here are close to those calculated from Rietveld refinement, indicating

single−crystalline particles formed.

5.4.1.3 TEM/Electron Diffraction

The morphology of the as-synthesized ZnXMg1−XO particles was studied using

transmission electron microscopy, as shown in Figure 5.2. The low− resolution TEM

images (Figure 5.2 a, c and e) show comparable particle sizes at various Zn

loadings. Individual particles are interconnected to form aggregates, which is a typical

feature for flame-synthesized materials. HRTEM imaging (Figure 5.2 b, d and f)

reveals the highly crystalline nature of the as-synthesized particles, regardless of the

Zn loading. Sharp edges of the single particles are visible with no evidence of an

amorphous covering layer. The crystallite size in each sample is about 10 nm, which

closely matches the results from Rietveld refinement of the XRD patterns and the

BET analysis. The observation again suggests that these nanoparticles are

individually single− crystalline. Electron diffraction probing (insets in Figure 5.2 a, c

and e) shows for a Zn loading of 10% a similar set of rings (Figure 5.2 g and h),

whereas for the 30% Zn loading, new rings occur belonging to the diffraction patterns

of wurtzite ZnO. The electron diffraction patterns corroborate the powder XRD

results.


64

Figure 5.2: Results of TEM characterization on the ZnXMg1−XO samples. The insets

in a, c, and e show the electron diffraction patterns of each sample.

5.4.1.4 EDS and HAADF−STEM

Energy-dispersive X-ray spectroscopy in TEM mode was used to analyze the

chemical compositions of the ZnXMg1−XO particle aggregates. The Zn loadings

detected by conventional EDS probing are 9.5% and 25.6% for the nominal loadings

of 10% and 30%, respectively, suggesting a good fidelity of the product composition

to the precursor solution design. Furthermore, by conducting EDS analysis under the

HAADF−STEM mode, we expect to investigate the distribution of the Mg, Zn and O

elements on the submicrometer length scale while retaining a resolution of a few

nanometers, in order to identify the events of ZnO phase separation. Using this

mode, as the electron beam scans the sample, the EDS spectra were collected at a

series of locations and the total intensity of the collected X-ray intensity is

decomposed according to the Mg, Zn and O energy windows to give element-specific


65

photon intensities along the scanned path on the length scale of hundreds of

nanometers. We combine EDS with HAADF−STEM imaging because it is difficult, on

this length scale, to precisely reveal the spatial distribution of different elements by

Z-contrast imaging alone, considering the powder nature of the samples and their

uneven thickness. The results are shown in Figure 5.3.

Figure 5.3: Results of EDS analysis under the STEM–HAADF mode. a, c and e)

STEM–HAADF images of MgO, Zn0.1Mg0.9O and Zn0.3Mg0.7O, respectively; b, d and

f) Spatial variation of the EDS signal intensities of different elements in MgO,

Zn0.1Mg0.9O and Zn0.3Mg0.7O, respectively. For each sample, a typical EDS spectrum

was recorded first and used to define the energy windows for different elements,

which were then used to decompose the total counts of photons to element–specific

intensities.


66

For the pure MgO sample (Figure 5.3 a and b), the signal intensity of Mg goes

in phase with that of O on a length scale of 80 nm. At 10% Zn loading (Figure 5.3 c

and d), the change of the intensity in the Mg signal is still in phase with that of the O

signal; in addition, the measured Zn signal is also in phase with those of Mg and O,

suggesting a homogeneous blending of Zn into the MgO lattice. Such results of

microscopic compositional analysis are consistent with the XRD analysis results,

which show no noticeable phase separation of ZnO at 10% of Zn. When the Zn

loading increases to 30%, the patterns of the EDS signals bear more structures, as

shown in Figure 5.3 e and f. The scanning distance is about 500 nm. In contrast to

the synchronized trends in Figure 5.3 b and d, in certain regions in Figure 5.3 f the

Mg and Zn signals show out−of−phase trends of variation. For instance, there are

regions where the intensity of Mg drops significantly while the Zn intensity keeps

almost constant; in certain other regions, the intensities of the Mg and Zn signals vary

in opposite directions. Such information indicates that at 30% Zn loading, the

distribution of Zn within the MgO matrix is inhomogeneous. The results are in good

agreement with the powder XRD data from which ZnO phase separation is identified.

Furthermore, it should be within the relatively Zn-abundant regions (e.g., the

rightmost part of the highlighted “Zn-rich” region in Figure 5.3 f) that the segregation

of wurtzite ZnO is expected to occur. Note that the trends found in Figure 5.3 are

typical and the elemental distribution results are representative of all samples.

5.4.1.5 Vacuum Annealing, Structural Changes, and Optical Properties

In comparison to the UV diffuse reflectance spectra of the MgO sample (Figure

5.4 a), the spectra of the as-synthesized Zn0.1Mg0.9O and Zn0.3Mg0.7O samples show

Zn2+- induced changes in the absorption properties (Figure 5.4 b and c, respectively).

These changes are either characterized by a red-shifted absorption band

(Zn0.1Mg0.9O, Figure 5.4 b) or an absorption threshold at 3.4 eV which is consistent

with the band gap of ZnO (Figure 5.4 c).140 This latter observation of the Zn0.3Mg0.7O

sample is supported by the XRD data (Figure 5.1, Table 5.1) and EDS analysis under

the STEM mode (Figure 5.3 e and f). Phase separation into an MgO−rich periclase

phase and a ZnO−rich wurtzite phase should have occurred during the flame

synthesis where the particles are exposed to flame temperatures of above 2000 K.

While vacuum annealing at T = 1173 K does not affect the absorption edge of


67

Zn0.3Mg0.7O (gray and black curves in Figure 5.4 c), the absorption−edge shift

observed for the Zn0.1Mg0.9O sample (Figure 5.4 b) points to a modification of the

electronic structure that originates from the annealing-induced Zn2+ segregation into

the particle surface.12

Figure 5.4: UV diffuse reflectance spectra of (a) MgO, (b) Zn0.1Mg0.9O and (c)

Zn0.3Mg0.7O before (black) and after (grey) vacuum annealing at T = 1173 K for 30

minutes. The spectra were acquired at room temperature and in the presence of

10 mbar O2.

For a clearer comparison, the UV diffuse reflectance spectra of the annealed

powders are plotted in Figure 5.5 a in conjunction with their TEM images shown in

Figure 5.5 b−d. Vacuum annealing of MgO leads to the optical absorption behavior

(Figure 5.5 a) and the cubic shape (Figure 5.5 b) that are characteristic of MgO

nanocubes with an average edge length of 10 nm. Both observations are consistent

with the properties of CVS−grown MgO powders.25,31 The sharpness of the cubic

morphology is not preserved upon admixing of Zn2+ (Figure 5.5 c and d).


68

Figure 5.5: (a) UV diffuse reflectance spectra and (b–d) TEM images of MgO,

Zn0.1Mg0.9O and Zn0.3Mg0.7O nanoparticles after vacuum annealing at T = 1173 K for

30 minutes (the spectra were acquired at room temperature in the presence of

10 mbar O2).

Powder XRD analysis and Rietveld refinement results show that vacuum

annealing causes different degrees of crystallite-size increase in periclase and

wurtzite domains (Table 5.1). The average crystallite sizes of MgO and Zn0.1Mg0.9O

(both possessing only the periclase phase) increase by 7% and 14%, respectively. In

the case of Zn0.3Mg0.7O, the crystallite size of the periclase domain grows by a factor

of 4.7, whereas the crystallite size of the wurtzite phase increases by a factor of 5.5.

TEM imaging results, as shown in Figure 5.5 d and Figure S5.1 (Supporting

Information), are in good agreement with the crystallite sizes estimated by XRD

analysis. Through high resolution TEM studies on the vacuum annealed Zn0.3Mg0.7O

sample, we identified the phase-separated wurtzite domains with characteristic d-

spacing of 0.247 nm of the (110) planes (see Figure S5.1 b in the Supporting


69

Information), which are in contrast to the periclase domains with the d-spacing of

0.212 nm of the (002) planes (see Figure S5.1 d in the Supporting Information).

These observations are consistent with the powder XRD patterns of the vacuum

annealed Zn0.3Mg0.7O sample (see Figure S5.1 in the Supporting Information). Such

trends appear to indicate that Zn2+ ions inside the periclase lattice facilitate the

crystallite growth, possibly by enhancing the mobility of the ions and/or ion vacancies.

Meanwhile, such effects should favor the preferential segregation of Zn2+ ions into

low coordinated surface sites.12 Additional EDS analysis under the STEM−HAADF

mode rules out the possibility of annealing−induced compositional changes such as

the depletion of ZnO into the gas phase.

Based on the UV diffuse reflectance data (Figure 5.5 a), we chose 270 nm and

340 nm as the excitation wavelengths for probing the PL emission properties of these

powders.31 For dehydroxylated MgO nanocubes, λExc = 270 nm (hvExc = 4.6 eV)

selectively excites the corner sites, whereas a wavelength of λExc = 340 nm

(hvExc = 3.6 eV) should not produce photoexcited states in MgO. The photon energy

of 3.6 eV; however, exceeds the band gap of ZnO and can be used to probe

ZnO-specific electronic transitions. The as-synthesized MgO sample shows a broad

PL emission feature with a maximum at λEm = 470 nm (Figure 5.6, black lines); the

shape of the spectrum and the position of the maximum intensity are independent of

the excitation wavelength. Admixing of 10% (atomic) Zn leads to a broadening of the

emission feature and a blue shift of its maximum to λEm = 440 nm; again, the position

of the maximum emission is independent of the excitation wavelength (Figure 5.6,

gray lines). The PL emission intensity of both MgO and Zn0.1Mg0.9O decreases upon

changing the excitation wavelength from λExc = 270 to λExc = 340 nm. A narrow and

so far unexplained emission feature at λEm = 680 nm (hvEm = 1.8 eV) is observed as

well.

For Zn0.3Mg0.7O, both the shape of the PL emission spectrum and the position

of its maximum depend on the excitation energy: with λExc = 270 nm, an emission

band at λEm = 560 nm with a shoulder at 430 nm suggests that the overall spectrum

is composed of at least two contributions (Figure 5.6 a, dashed line). An excitation

wavelength of λExc = 340 nm produces an emission band at λEm = 590 nm (Figure

5.6 b, dashed line), the intensity of which is comparable to the one being subjected to


70

excitation with λExc =270 nm. All spectra in Figure 5.6 are insensitive to the presence

of O2 in the gas phase, i.e., no evidence is found for the energy transfer between the

photoexcited particles and O2 (previously established as a PL quencher31,140) from

the gas phase.

Figure 5.6: PL emission spectra of as-synthesized ZnXMg1−XO nanoparticles

acquired at p < 10–5 mbar and T = 298 K using excitation wavelengths of (a) 270 and

(b) 340 nm, respectively. The emission spectra are not affected by the presence of

10 mbar O2.

To investigate the optical properties of the partially hydroxylated samples, the

samples were subjected to annealing at T = 873 K prior to spectroscopic

measurements. Previous studies on MgO nanoparticles have shown that vacuum

annealing in combination with O2 treatment is effective for the elimination of carbon

based surface contaminants;26,32 meanwhile, the concentration of hydrogen−bonded

neighboring surface OH groups becomes negligibly low and only the bands

pertaining to isolated surface hydroxyl groups remain.25,107 As a result, the emission

with a maximum at λEm = 560 nm which was previously observed on as−synthesized

Zn0.3Mg0.7O (Figure 5.6 a) vanishes after being annealed at T = 873 K (Figure 5.7 a

and Figure S5.2 a in the Supporting Information). Pure MgO powders show a strong

surface dependent PL emission feature with a maximum at 470 nm (Figure 5.7). This

band is quenched by O2 from the gas phase, which is attributed to the protonated


71

corner sites of partially hydroxylated MgO nanocubes.25 This observation is in good

agreement with the findings on CVS−grown MgO powders. Interestingly, Zn2+

admixed into MgO decreases the intensity of the PL emission. This effect is attributed

to the annealing-induced segregation of Zn2+ ions into the edge and corner sites in

the periclase domain, thus inducing the depletion of MgO-specific surface excitons.12

As outlined in the Supporting Information (Figure S5.2 b) molecular oxygen does not

entirely quench the PL emission band at λEm = 440 nm (hvEm = 2.8 eV). This is

different from dehydroxylated particle surfaces and indicates that surface hydroxyls

counteract energy transfer between photoexcited surface states and molecular

oxygen.156

Figure 5.7: PL emission spectra of ZnXMg1−XO nanoparticles after vacuum annealing

at T = 873 (a) and T = 1173 K (b) (The dwell time at each temperature is 30 minutes;

the spectra were acquired at < 10–5 mbar and T = 298 K using λExc=270 nm). The PL

emission of MgO is effectively quenched upon the addition of 10 mbar O2 irrespective

of the annealing temperature. The PL emission of the Zn0.1Mg0.9O sample is

quenched by 10 mbar O2 only after annealing at T = 1173 K.

The intensity loss of the green emission from the partially hydroxylated samples

may have its origin in the annealing−induced annihilation of intrinsic bulk− or surface

defects including the surface hydroxyls.12,82,157,159 To address the latter possibility, we

performed a control experiment by exposing the partially dehydroxylated Zn0.3Mg0.7O


72

sample to air for 24 h (see Figure S5.2 a in the Supporting Information).

Subsequently, after being pumped down to <1∙10−6 mbar, the sample did not present

a restoration of the PL emission observed before annealing. Apparently, simple

surface hydroxylation via contact with moisture in the air is insufficient to restore the

ZnO specific PL behavior. In case of the MgO samples, a PL emission at hvEm = 2.7

eV (Figure 5.7 a, black line) grows with the increase in annealing temperature while

the band maximum shifts to 3.2 eV (Figure 5.7 b, black line). This trend is consistent

with the stepwise dehydroxylation of MgO nanocube surfaces upon generation of

bare metal oxide particle surfaces where surface excitons can selectively form at

corners and edges.31

Figure 5.8: Schematic summary of the correlation between the surface states of the

ZnXMg1−XO particles and their PL emission properties. High vacuum annealing leads

to dehydroxylated MgO and Zn-Mg-O surfaces and morphological transformation of

the particles. The size of the flashes qualitatively denotes the relative emission

intensity that changes upon annealing.

Figure 5.8 schematically summarizes the observed trends in the photoelectronic

properties of pure and Zn2+ admixed MgO cleaning. Starting with particles the


73

surfaces of which are covered with adsorbates that are generated from the synthesis

process (Figure 5.8, left), a PL emission at λEm = 590 nm is observed only for the

Zn0.3Mg0.7O sample with a separated wurtzite phase (bottom of Figure 5.8). The

corresponding excitation energy is below the threshold energy required to address

low−coordinated surface sites on MgO particles. The independence of the PL

emission on the surrounding gas atmosphere, in particular from molecular oxygen as

a PL quencher, indicates that the underlying process corresponds to the radiative

deactivation of bulk excitons.

Annealing at T = 873 K irreversibly extinguishes this ZnO specific PL feature

and facilitates the formation of surface excitons on MgO particles (Figure 5.7).

Moreover, such treatment not only removes carbon−based surface contaminants and

eliminates the surface hydroxyls, but also favors ion mobility inside the lattice,

therefore allowing for partial reorganization of the particles (Figure 5.8, right). While

the MgO particles transform into cubes and essentially retain their high dispersion

(Table 5.1), the admixing of Zn2+ ions reduces the thermal stability of the host

component and leads to coarsened and morphologically less sharply defined

particles. PL emission measurements clearly demonstrate that MgO specific surface

excitons are strongly depleted by annealing-induced Zn2+ segregation into low

coordinated surface sites where surface excitons are formed and/or can radiatively

deactivate. A new PL emission band with its maximum at λEm = 520 nm is observed,

which should originate from the Zn2+-decorated MgO particle surface and is related to

the chemical composition at the interface.

5.5 Conclusions

It has been demonstrated that flame spray pyrolysis is a valuable synthesis

approach for the generation of ternary Zn-Mg-O nanoparticles. Powder XRD, Rietveld

refinement and HRTEM−EDS collaboratively show that admixing 10% (atomic) of Zn

into MgO leads to Zn0.1Mg0.9O solid solutions, whereas admixing 30% of Zn gives

rise to the phase separation into an MgO-rich periclase phase and a ZnO-rich

wurtzite phase. Such structural differences critically affect the optical properties of the

particles, as revealed by UV diffuse reflectance and PL emission measurements:


74

(1) The as-synthesized MgO nanoparticles show PL characteristics that are

related to surface excitons associated with the corner sites of MgO

nanocubes.31,140

(2) Phase-separated wurtzite ZnO inside the as-synthesized Zn0.3Mg0.7O

particles gives rise to a green PL emission at λEm = 590 nm. This process is

not quenched by molecular oxygen in the gas phase. Removal of adsorbed

water, hydroxyls, and possible carbonaceous contaminants by vacuum

annealing and oxidation at T = 873 K leads to the extinction of this emission

process. Subsequent surface hydroxylation via contact with air moisture

does not restore this feature.

(3) Upon vacuum annealing, surface-segregation of Zn2+ ions causes depletion

of MgO-specific PL emission. While exposure to gaseous O2 partially

reduces the PL emission intensity of hydroxylated Zn0.1Mg0.9O particles, it

quenches entirely the emission in dehydroxylated particles. Consequently,

hydroxyls can be regarded as protecting groups against gaseous oxygen as

a PL quencher at the solid-gas interface.

The present study clearly underlines the importance of sorting out the bulk and

interface contributions to the overall optical performances of composite oxide

nanomaterials that can be further engineered as components for optoelectronic

applications.


75


5.6.1 Calculation of the Chemical Formula of the Periclase–Phase Solid

Solution

From Rietveld refinement, a weight percentage of the wurtzite phase of 5.9% is

estimated. We assume the wurtzite phase consists of only ZnO, while the periclase

phase contains both Zn and Mg. Starting with 0.3 mol of ZnO and 0.7 mol of MgO as

determined by the precursor solution design, the mass of wurtzite ZnO is:

(0.3 × 81.41 + 0.7 × 40.30) × 5.9% = 3.11(g)

The rest of the total mass in the periclase phase:

(0.3 × 81.41 + 0.7 × 40.30) × 94.1% = 49.53 (g)

The molar ratio between Zn and Mg in the periclase phase is:

((0.3 × 81.41 – 3.11) / 81.41) / 0.7 = 0.26/0.7

The formula of the periclase phase compound is:

Zn0.26Mg0.7O0.96 or Zn0.27Mg0.73O

The formula weight of the periclase phase compound is then 51.39 g mol–1.

Therefore, the molar ratio between the wurtzite and the periclase phase is:

((3.11/81.41) / (49.53/51.39)) = 1/25


76

5.6.2 TEM and XRD of the Vacuum Annealed Zn0.3Mg0.7O Sample

Figure S5.1: TEM images and XRD patterns showing different crystalline domains in

the vacuum annealed Zn0.3Mg0.7O sample. The measured distances of 0.247 nm (b)

and 0.212 nm (d) correspond to the (011) and (002) d–spacing in the hexagonal

(a and b) and cubic (c and d) domains, respectively. The (011) and (002) planes

correspond to the strongest reflections in the XRD patterns of the wurtzite (solid

circles) and the periclase (empty triangles) phases.


77

5.6.3 Photoluminescence Spectra

Figure S5.2: a) Effects of annealing and exposure to air/ O2 atmosphere on the PL

emission properties of the Zn0.3Mg0.7O sample. The annealing time is 30 minutes.

b) Room temperature photoluminescence emission spectra (λExc = 340 nm) of

hydroxylated Zn0.1Mg0.9O nanoparticle powders after vacuum annealing and

oxidation at T = 873 K. Different to samples with entirely dehydroxylated particle

surfaces molecular oxygen does not entirely quench the PL emission band at

hvEm = 2.8 eV which indicates that surface hydroxyls counteract energy transfer

between photoexcited surface states and molecular oxygen.

6 Fe-Mg-O Nanocomposite Particle Systems

79

6 Fe-Mg-O Nanocomposite Particle Systems: Controlled

Synthesis and the Influence of Annealing on

Composition, Structure and Optical Properties

6.1 Abstract

Fe-Mg-O nanocomposite particles have been prepared by chemical vapor

synthesis (CVS) and subsequent annealing in controlled gas atmosphere. The

composition of the composite Fe-Mg-O (Fe/(Fe+Mg) = 1, 6 or 9 at.%) particles is

investigated by Inductively Coupled Plasma-Optical Emission Spectroscopy (ICP-

OES) and EnergyDispersive X-ray (EDX) spectroscopy. The structure of the obtained

nanoparticles has been characterized by X-ray diffraction and transmission electron

microscopy, whereas Mößbauer spectroscopy is used to investigate the valence

state of Fe ions and magnetic properties of the samples. The results demonstrate

that the Fe-Mg-O composite can reveal superparamagnetic properties in case of

particle systems to be characterized as solid solutions of Fe3+ ions in periclase MgO

or antiferromagnetic behavior in case of phase separated particle mixtures containing

periclase MgO and magnesioferrite MgFe2O4 phases.

Fe-Mg-O nanoparticle powders are further explored by UV-Vis diffuse

reflectance and photoluminescence (PL) spectroscopy to investigate their optical

properties. The absorption spectra of samples consist of several absorption bands in

the energy range 2.6 eV ≤ hv ≤ 5.6 eV. These are attributed to different ligand field

and ligand-to-metal charge transfer electronic transitions. The absence of MgO

specific PL emission originating from surface excitons suggests annealing induced

surface segregation of Fe3+ ions for Fe-Mg-O nanocomposite particles. The results

show that the trends in the optical properties are subject to different iron

concentrations and annealing induced changes.


80

6.2 Introduction

The Fe-Mg-O system as one of the prevailing mineral systems in Earth’s lower

mantle has gained research interest over a long time because of its importance in

understanding physical and chemical properties of the Earth.159,160 Moreover, certain

phases of the Fe-Mg-O system like Magnesioferrite (MgFe2O4) exhibit optical,

chemical and magnetic properties with potential for applications as photovoltaic

devices38, gas sensors162, magnetic devices39, catalysis40 and pigments59. Many

studies discuss the behavior of natural Fe-Mg-O systems under different conditions

of temperature and pressure. Different preparative techniques have been used and

proposed to yield desired Fe-Mg-O solid phases with specific structural, electronic

and magnetic properties.

The methods that have been employed for the fabrication of Fe-Mg-O

nanostructures are very diverse. Examples are in melt doping of MgO with iron and

subsequent high temperature oxidation62, high temperature annealing of powder

mixtures of Fe2O3 and MgO63, combustion of iron and magnesium nitrates in urea

solution64, sol-gel approaches followed by high temperature calcination59, spray

pyrolysis38, sputtering162 and hydrothermal procedures66.

Structural studies on Fe-Mg-O systems revealed that the admixture of iron to

magnesium oxide can give rise to ferric (Fe3+) or ferrous (Fe2+) states of iron in the

mixture. The valence state of the iron ions in the mixture are subject to heat

treatment.162 Investigations reveal for both valence states, that the iron ions replace

magnesium ions in the cationic sublattice of the magnesium oxide structure. In

octahedral sites inside the cubic lattice the ionic radius of Fe2+ (r = 0.078 nm) is

different from that of Fe3+ (r = 0.064 nm).150 Compared with the ionic radius of Mg2+

ion (r = 0.072 nm) the divalent iron ion is slightly larger and the trivalent iron ion is

slightly smaller when substituting magnesium.


81

Figure 6.1: Schematic diagram of Fe2+ and Fe3+ substituting Mg2+ in MgO crystal

lattice.

Upon substitution of Mg2+ ions in the cationic sub lattice, Fe2+ ion donates two

valence electrons to its neighboring oxygen atoms and induces strain in the

surrounding lattice (Figure 6.1 left).163 Several studies reveal that Fe2+ ions tend to

cluster in order to decrease the strain they exert on the lattice.164–166

Upon replacement of Mg2+ by Fe3+ ions, the positive excess charge of Fe3+ can

be compensated by formation of Mg2+ vacancies.62 For every two Fe3+ ions there is

one Mg2+ vacancy which guarantees charge compensation (Figure 6.1 right). The

Fe3+ ions are reported to be either isolated in sites of octahedral symmetry or

clustered in tetrahedral sites.64,168

For Fe-Mg-O composite systems heat treatment can induce phase changes and

phase separation. This depends on total iron concentration, oxygen partial pressure

applied during annealing and annealing temperature. Phase diagrams of the ternary

Fe-Mg-O system have been reported for different atmospheres and at high

temperatures, typically for temperatures above T = 1273 K.168,169 These phase

diagrams reveal coexisting phases for Fe concentrations above Fe3+ solubility limit in

MgO lattice at XFe = 0.3 at.% and at T = 1273 K.41,42 A typical phase diagram of the

Fe-Mg-O system is shown in Figure 6.2. It can be seen that at T = 1273 K in the

samples with 0 at.% < XFe < 64 at.% both phases of magnesiowustite ((Mg,Fe)O) and

magnesioferrite (MgFe2O4) coexist. At the same temperature the samples with an

iron concentration of 64 at.% < XFe < 70 at.% only consist of magnesioferrite phase.

As another important point it is important to note the coexistence between the

magnesioferrite and hematite (Fe2O3) phase for samples with iron concentrations XFe


82

> 70 at.% and at T = 1273 K. It should be mentioned that phase diagram in Figure

6.2 does not specify the relative distributions of Fe2+/Fe3+ ions at a given composition

and temperature.

Figure 6.2: Phase diagram in the system Fe-Mg-O in air. After Phillips et al.42,168

XFe = nFe/(nFe+nMg). X: molar ratio; n: number of moles.

The magnesioferrite phase with the chemical formula of MgFe2O4 has a cubic

spinel structure. The spinel structure is based on fcc packing of O2- ions and has the

general formula of AB2O4 where A is a divalent and B a trivalent cation. In unit cell of

the normal spinel (Figure 6.3 left) having 64 tetrahedral and 32 octahedral interstices

the A ions occupy 8 of the tetrahedral and the B ions 16 of the octahedral interstices.

In the inverse spinel structure (Figure 6.3 right) A and B ions are equally distributed

over these 16 octahedral interstices while the remaining number of B ions occupy the

8 tetrahedral interstices. The deviation from these two types of cation distributions in

octahedral and tetrahedral sites is called inversion. In MgFe2O4 the distribution of A

and B cations corresponds to an intermediate state between normal and inverse

spinel structure. In any case the degree of inversion sensitively depends on the

thermal history of the sample.170


83

Figure 6.3: Schematic diagram of the normal (left) and inverse (right) spinel structure

with the general formula of A(II)B2(III)O4. In the normal spinel 8 of tetrahedral and 16

of octahedral sites are occupied by divalent (A) and trivalent (B) ions, respectively. In

inverse spinel divalent (A) ions swap with half of the trivalent (B) ions. Adapted from

reference [171].

While the above mentioned information which stems from equilibrium

thermodynamics is very valuable for a first orientation, the situation may be different

for nanomaterials which were synthesized under non equilibrium conditions.

Recently, nanoparticles of ternary Fe-Mg-O system like MgFe2O4 have received a lot

of attention because they can exhibit novel magnetic properties that are different from

their bulk counterparts.58 The structural, chemical and surface/interface properties of

such nanoparticles critically depend on preparation techniques and their post

synthesis treatments. Most of the wet methods like sol-gel or hydrothermal synthesis

have intrinsic disadvantages with regard to their use for more detailed studies or

advanced applications. Reasons for that are undesirable bulk and surface impurities

originating from synthesis, poor control of cation concentration and broad particle

size distribution.

This work aims to study the potential of metal organic chemical vapor synthesis

(MO-CVS) as a method to synthesize Fe-Mg-O nanoparticle composites using a Fe

(III) organic precursor. In the first part we explored the production of Fe-Mg-O

nanoparticles and established a robust process which provides control over iron

concentration and compositional homogeneity. In addition the effect of post-synthesis

thermal treatment on the as synthesized samples with different iron concentrations

was investigated in order to track the impact of thermal processing on structure and


84

compositional changes. In the third part we explored the role of total iron

concentration and annealing treatment on the optical properties of Fe-Mg-O mixed

oxides.


6.3.1 Material Synthesis

Fe-Mg-O nanoparticles were produced via metal organic chemical vapor

synthesis (MO-CVS). The details of the MgO preparation, which is based on

controlled combustion of metal vapor with oxygen under reduced pressure is given

elsewhere.95,115 Here we used for the first time a modified MO-CVS procedure, which

provides control over the concentration of iron in Fe-Mg-O nanocomposite samples.

The two-hot-zone reactor system (Figure 6.4) employed for this purpose consists of

two quartz glass tubes, which are mounted concentrically inside a heating coil (first

heating zone with T1 as operation temperature) followed by a ceramic tube furnace

(second heating zone with T2 as operation temperature).

Figure 6.4: Schematic diagram of the reactor setup used for production of Fe-Mg-O

nanoparticles. (MFC: mass flow controller; T: local operation temperatures;

P: pressure gauge)

In the first heating zone the inner glass tube hosts a ceramic ship with

iron (III) acetylacetonate (Fe(C5H7O2)3) powder (≥ 99.9%, Sigma-Aldrich) (1 g), which

is heated to temperature T1 = 353 K or 363 K or 373 K to sublimate the iron


85

precursor and to adjust the evaporation rates. An argon gas flow (QAr = 1200 sccm)1

is led through the inner tube to transport the metal organic vapor to the second

heating zone where the furnace provides a temperature T2 = 913 K. At this position a

ceramic ship with Mg metal grains (1 g) is positioned inside the inner tube. Here, the

magnesium is evaporated and the resulting metal vapor becomes mixed with the

gaseous iron precursor. The vapor mixture is then transported by the argon gas flow

to the end of the inner glass tube. There the argon, magnesium vapor and iron

precursor vapor mixture meets oxygen (QO2 = 1200 sccm) which is flowing through

the outer glass tube. At this reactor position, the exothermic oxidation reaction

(2Mg + O2 → 2MgO) leads to a stable Mg combustion flame which decomposes the

iron precursor. As a result of homogenous nucleation and particle growth Fe-Mg-O

nanoparticles are formed. Because of continuous pumping, the residence time of

nuclei within the flame is so short that it prevents substantial particle coarsening and

coalescence. The MgO nanoparticles are deposited in a stainless steel net that is

kept downstream at the room temperature. The total pressure (p = 70 ± 2 mbar), as

well as argon and oxygen flow rates and T1 and T2 heating zones temperatures are

kept constant during nanoparticle collection.

6.3.2 Annealing

Annealing treatment was used for dehydration, dehydroxylation of the particle

surfaces and for removal of carbon based contaminants originating from either

synthesis or short contact of the nanoparticle powders with air. This post-synthesis

treatment is performed according to an optimized procedure2 which is outlined in

Figure 6.5.

1 Q = volumetric flow rate

2 This optimized procedure is selected on the basis of previous exploratory annealing procedure

evaluations which also included the compositional analysis done on Fe-Mg-O samples. For this

purpose measurement of local concentration at different positions of each sample is carried out by

EDX. The investigations revealed that i) removal of remnant carbon species (from incomplete

precursor decomposition in the reactor) requires annealing at oxygen pressures of p = 650 mbar and

ii) samples with homogeneous distribution of iron and magnesium can be retained when the samples

are kept at T= 873 K for more than 2h.


86


pressure (dashed line, right ordinate scale) during annealing treatment.

At the beginning, the cell containing a nanoparticle powder was evacuated

down to a pressure of p < 10-5 mbar at room temperature. The respective sample

was then heated to T = 373 K at a rate = 2.5 K min-1, held at this temperature for

15 min (dwell time, td) and then again subjected to evacuation to p < 10-5 mbar.

Further annealing steps are carried out at p = 650 mbar of oxygen. The powder was

stepwise heated in oxygen atmosphere to T = 473 K (r = 5 K min-1, td = 15 min),

T = 673 K (r = 10 K min-1, td = 30 min), T = 873 K (r = 10 K min-1, td = 180 min) and

T = 1173 K (r = 10 K min-1, td = 60 min).

After each annealing step the sample was cooled to room temperature

(cooling time ≈ 30 min) followed by an evacuation (≈ 15 min) to a base pressure of



X-ray diffraction (XRD) measurements were performed on a Bruker AXS D8

Advance diffractometer using Cu Kα radiation (λ = 154 pm). For electron microscopic

measurements, small amounts of the metal oxide powders were cast on a carbon


87

grid for investigation. Scanning electron microscopy (SEM) measurements were

performed on a Zeiss Gemini Ultra 55 microscope operating at 20 kV equipped with

an EDX (energy-dispersive X-ray emission) detector which allows for local

composition analysis. The transmission electron microscopy (TEM) investigations

were performed on TECNAI F20 analytical microscope equipped with a field emission

gun and an S-twin objective lens.

6.3.2 Spectroscopy

The metal composition of the Fe-Mg-O nanoparticles was analyzed by

inductively coupled plasma - optical emission spectroscopy (ICP-OES), in a

PerkinElmer OptimaTM 8300. For ICP-OES measurements the powder samples were

dissolved in 5 vol.% HNO3. EDX spectroscopy was employed to characterize the

local composition at different sample positions and to investigate the compositional

homogeneity in the Fe-Mg-O nanoparticle agglomerates. The local resolution of

chemical information for EDX analysis is limited to the penetration depth and

scattering of the primary electrons and, therefore, to the volume where characteristic

X-rays are emitted of the sample. With acceleration voltages of 20 kV the sampled

volume in EDX studies are in the order of few μm3 for bulk Fe-Mg-O samples.

Therefore the change in the ratio between Fe and Mg of each sample is tracked by

EDX over more than 50 positions1 with spatial resolution of approximately 10 μm.

For Mössbauer spectroscopy the powder sample was embedded in a copper

ring and mounted in a conventional Mössbauer apparatus (Halder Elektronik GmbH)

with constant acceleration (time mode procedure). The absorber was exposed to a

nominal 50 mCi (1.8 GBq) 57Co source in Rh (Wissel Ltd). The transmitted intensities

were stored in a multichannel analyzer with 1024 channels (Halder Elektronik

GmbH). In order to improve statistics, the obtained spectra were folded to 510

channels. For a good resolution and a good peak / background ratio high counting

rates (generally half a million) were chosen. Before and after a sequence with

identical velocity adjustments a -iron spectrum was recorded in order to determine

1 Figure S6.1 in Supporting Information provides an example of how the local distributions of Fe

and Mg ratios were measured for many local spots of the samples.


88

the calibration factor - the velocity scale was subsequently recalculated from

channels to mm/s.

UV-Vis diffuse reflectance spectra were acquired at room temperature using

quartz glass cells with a Perkin-Elmer Lambda 950 spectrophotometer, equipped with

an integrating sphere. The reflectance spectra were converted into absorption

spectra using the Kubelka Munk transform procedure. For photoluminescence (PL)

measurements, a Fluorolog®-3 Model FL3-22 spectrometer with a continuous wave

450 watt Xenon arc lamp was used. PL spectroscopy was performed at room

temperature using quartz glass cell that guarantee vacuum condition better than

5·10-6 mbar.

6.4 Results

Part I: Composition, Structure and Morphology

6.4.1 Nanoparticle Composition

ICP-OES and EDX analysis are used to investigate the integral Fe

concentration as well as the compositional homogeneity (local concentration) of Fe in

the nanocomposite samples, respectively. The integral concentration of a given

element is its total amount in a volume of sample which is used for elemental

analysis. The local concentration is the amount of element in a certain position on the

length scale of tens of µm of the sample. The concentration values of each element

in atomic percent can be determined from ICP-OES and EDX results. Based on

these values the relative iron concentration, i.e. Fe/(Fe+Mg), is calculated and plotted

as XFe. Table 6.1 shows the integral XFe values for Fe-Mg-O samples produced by

using different operation temperatures T1 in the first heating zone of the

reactor (Figure 6.4). The concentrations are measured after controlled annealing of

each sample at T = 873 K and T = 1173 K (Figure 6.5).


89

Table 6.1: Relative iron concentrations determined by ICP-OES. The results are

shown for Fe-Mg-O samples produced by using different operation temperatures T1

in the first heating zone of the reactor and after controlled annealing of each sample

to T = 873 K and T = 1173 K. XFe = nFe/(nFe+nMg). X: molar ratio; n: number of moles.

The analysis of the Fe-Mg-O samples by ICP-OES (Table 6.1) clearly shows

that setting the temperature T1 and, thus, the sublimation rate of the iron precursor

determines the integral Fe concentration. Table 6.1 shows how by increasing T1 from

T = 353 K to T = 373 K the integral Fe concentration increases from XFe = 1 at.% to

XFe = 10 at.%. Variations in concentration between samples annealed at T = 873 K

and T = 1173 K are attributed to preparation and sampling errors.

The Fe-Mg-O samples produced with T1 = 353 K, 363 K and 373 K will be

called (1%) Fe-Mg-O, (6%) Fe-Mg-O and (9%) Fe-Mg-O based on their integral

concentrations.The integral concentrations obtained from ICP-OES give no

information about the compositional homogeneity of Fe and Mg in nanoparticle

samples; therefore the local concentrations were investigated by EDX. The

compositional homogeneity is defined as the probability of having uniform

composition (in terms of XFe) at different specified sample positions throughout the

nanocomposite powder. This is described by composition distribution function which

is obtained from cumulative XFe values (Figure 6.6).

Figure 6.6 shows composition distribution function of (1%) Fe-Mg-O, (6%)

Fe-Mg-O and (9%) Fe-Mg-O samples. For all samples after annealing at T = 873 K

narrow distribution functions with median values consistent with the integral values

were observed. Figure 6.6 b reveals how annealing to T = 1173 K affects the

composition distribution function. For the samples with higher Fe content, i.e. (6%)

Fe-Mg-O and (9%) Fe-Mg-O the determined functions are clearly broadened.

sample name (1%) Fe-Mg-O (6%) Fe-Mg-O (9%) Fe-Mg-O

temperature T1 in reactor first zone 353 K 363 K 373 K

annealing temperature 873 K 1173 K 873 K 1173 K 873 K 1173 K

XFe / at.% ± 10 % 1.6 1.4 6.2 5.5 9.4 9.7


90

However, for the (1%) Fe-Mg-O sample the composition distribution function remains

as narrow as after annealing to T = 873 K.

Figure 6.6: Composition distribution functions of Fe-Mg-O samples produced with

different iron precursor temperatures (T1) obtained by local EDX measurements; a)

samples annealed to T = 873 K; b) samples annealed to T = 1173 K. The minimum,

maximum and median values of each distribution curve are indicated.

XFe = nFe/(nFe+nMg). X: molar ratio; n: number of moles.

6.4.2 Mößbauer Spectroscopy

Mössbauer spectroscopy has been used as an effective method to determine

the valence state of the Fe ions involved, symmetry / distortion of the surrounding

charge distribution and presence or absence of magnetic ordering in the Fe-Mg-O

samples described above. As a first generally important result which applies for all

samples (see Figure S6.2 and Table S6.4 in Supporting Information) the Mössbauer

spectra reveal the presence of Fe3+ ions in a distorted octahedral environment. No

Fe2+ related phase has been detected. Since Fe3+ was employed as a precursor and

synthesis and post-synthesis treatment was performed in oxidizing environment this

result is not surprising.

The Mössbauer spectrum of the (1%) Fe-Mg-O sample annealed at T = 873 K

reveals no distinct absorptions (Figure S6.2). In contrast, the Mössbauer spectra of

the (6%) Fe-Mg-O and (9%) Fe-Mg-O samples annealed at T = 873 K reveal


91

doublets which are indicative of superparamagnetic phases of Fe3+. Each of these

spectra is fitted with two doublets whose isomer shifts () and quadrupole splittings

(QS) are consistent with those of Fe3+ in 6 fold coordination.41,64,172,173 However,

different QS values of the two fitted doublets of each spectrum suggest variations in

the degree of distortion in the octahedral sites.

The samples with different concentrations were further annealed at T = 1173 K

and after cooling to room temperature subjected to Mössbauer spectroscopic

measurements. The spectrum of the (1%) Fe-Mg-O sample annealed at T = 1173 K

reveals two doublets (Figure S6.2 and Table S6.4). Similar to the spectra of

(6%) Fe-Mg-O and (9%) Fe-Mg-O samples annealed at T = 873 K, the spectrum of

the (1%) Fe-Mg-O sample annealed at T = 1173 K indicates the existence of Fe3+ in

a 6 fold coordination. However, the site distribution of the two sorts of Fe3+ ions

reveals a ratio of approximately 2/3 to 1 which means random occupation.

The Mössbauer spectra of the (6%) Fe-Mg-O and (9%) Fe-Mg-O samples

annealed at T = 1173 K display complicated patterns and can be deconvoluted into

three sextets and one doublet (Figure S6.2 and Table S6.4). The distribution of the

hyperfine fields (Table S6.4) and the absence of Fe2+ are indicative of the spinel type

compound magnesioferrite MgFe2O4 with antiferromagnetic behavior.41,172 The

doublet is ascribed to a superparamagnetic phase which coexists with the

magnetically ordered phase. Both spectra clearly indicate the separation of samples

into a magnetically ordered phase with antiferromagnetic behavior and a

superparamagnetic phase.


6.4.3.1 SEM

SEM images of the undoped MgO nanocubes and the Fe-Mg-O nanocomposite

which were annealed to temperatures of T= 873 K and T = 1173 K are shown in

Figure 6.7. It can be seen that the samples annealed to temperature of T= 873 K

correspond to fluffy and loosely bound powders (Figure 6.7 a, c, e and g). Annealing

of the samples to T = 1173 K leads to more compact powders (Figure


92

6.7 b, d, f and h). These compact powders consist of particulate objects in the size

range of tens of nanometers (insets in Figure 6.7 f and h).

Figure 6.7: SEM images of the MgO and Fe-Mg-O samples after annealing to

T = 873 K (left) and T = 1173 K (right): a and b) MgO; c and d) (1%) Fe-Mg-O and;

e and f) (6%) Fe-Mg-O; g and h) (9%) Fe-Mg-O.

The SEM images of the samples annealed to T = 1173 K show Fe-Mg-O

nanocomposite powders are more coarse grained compared to the undoped MgO

nanoparticle powders. With increasing iron content of the nanocomposites, the trend


93

towards a compact and coarse grained powder increases. This is observed in the

SEM images of the (6%) Fe-Mg-O and (9%) Fe-Mg-O samples (Figure 6.7 f and h)

annealed to T = 1173 K.

6.4.3.2 TEM

TEM images of the undoped MgO sample annealed to temperatures of

T = 873 K (Figure 6.8 a) and T = 1173 K (Figure 6.8 b) show agglomerates consisting

of nanoparticles with cubic morphology. For the undoped MgO annealed at T = 873 K

rounded cubes as a result of multiple stepped edges can be observed in some

regions of the sample (Figure 6.8 a, inset). Increasing the annealing temperature of

the MgO samples to T = 1173 K does not change the particle size distribution of the

particle ensemble (Figure 6.9). However, a small increase in median value (~5 nm)

can be seen.

Figure 6.8: TEM images of the MgO samples after annealing to T = 873 K (a) and

T = 1173 K (b).


94

Figure 6.9: Cumulative particle size distribution plots for undoped MgO (blue) and

Fe-Mg-O samples after annealing to T = 873 K (a) and T = 1173 K (b). The

distribution was obtained by measuring the size of 300 to 400 particles at different

sample regions from TEM images.

Figure 6.10 a and b reveal a variety of morphological features of the primary

particles for the (1%) Fe-Mg-O sample annealed to temperature of T = 873 K. The

images indicate that particles with less regular shapes coexist with more cubic

crystallites having multiple stepped surfaces.

Figure 6.10 c and d indicate that annealing of the (1%) Fe-Mg-O sample at

T = 1173 K lead to growth of some nanocubes to larger dimensions (~50 nm to

80 nm) however the majority of the particles remain smaller in size. TEM, moreover,

reveals that by annealing at T = 1173 K, the particles adopt cubic morphology. This

development is similar to that of undoped MgO particles; however, the particle size

distributions of (1%) Fe-Mg-O and undoped MgO samples are different.


95

Figure 6.10: TEM images of the (1%) Fe-Mg-O samples after annealing to T = 873 K

(a and b) and T = 1173 K (c and d).

From the detailed analysis of many TEM images taken from different sample

regions it can be concluded that (1%) Fe-Mg-O sample retains its narrow size

distribution with an average particle size of about 7 nm (= median) after annealing to

T = 873 K (Figure 6.9 a). Compared to undoped MgO the lower average particle size

must be attributed to the presence of iron in the lattice of MgO which seems to limit

crystal growth in the temperature range of 300 K ≤ T ≤ 873 K. However, annealing of

the sample at T = 1173 K significantly increases particle growth and leads to broad

particle size distribution (Figure 6.9 b).


96

Figure 6.11 and Figure 6.12 represent the TEM images of (6%) Fe-Mg-O and

(9%) Fe-Mg-O samples after oxidative heat treatment. For both samples which were

annealed to T = 873 K the uniformity in particle size is similar. This is more clearly

reflected in the cumulative particle size distribution curves of the samples with the

average particle size of about 11 nm (Figure 6.9 a). However, particles exhibit

several morphological features such as cubic (or cuboid), semi-hexagonal or rounded

shapes (insets in Figure 6.11 a and Figure 6.12 a).


(a) and T = 1173 K (b).

Annealing of the (6%) Fe-Mg-O and (9%) Fe-Mg-O samples at T = 1173 K

leads to significant particle growth (Figure 6.11 b and Figure 6.12 b). Large crystal

domains are surrounded by much smaller particles. Unlike the samples which were

annealed to T = 873, the samples annealed to T = 1173 K exhibit significantly

broadened particle size distributions (Figure 6.9 b).


97


(a) and T = 1173 K (b).

From the Fe and Mg composition distributions (Figure 6.6) and particle size

distributions (Figure 6.9) of the samples we can conclude at this point:

For the (1%) Fe-Mg-O, the compositional homogeneity i.e. the composition

distribution function is retained as a result of increasing the annealing temperature

from T = 873 K to T = 1173 K. However annealing of this sample to temperature of

T = 1173 K broadens the particle size distribution. From this we conclude that

annealing to temperature of T = 1173 K drives the crystal growth at some sample

regions. Different from particle size distribution, the cubic shape of the particles is

retained when the (1%) Fe-Mg-O sample is annealed at T = 1173 K.

In contrast to the thermal stability of the (1%) Fe-Mg-O sample, annealing of the

(6%) Fe-Mg-O and (9%) Fe-Mg-O samples to T = 1173 K significantly changes their

composition distribution function. The (6%) Fe-Mg-O as well as (9%) Fe-Mg-O

samples behave in the same way. Annealing to T = 1173 K broadens particle size

distribution of (6%) Fe-Mg-O and (9%) Fe-Mg-O samples. This crystal growth at

some regions of these samples is exemplified by TEM observations. Particles of

these samples beside cubic shape exhibit some other morphological features like

hexagonal or rounded shapes (insets in Figure 6.11 and Figure 6.12).


98

6.4.3.3 XRD

The XRD patterns of undoped MgO and Fe-Mg-O samples are presented in

Figure 6.13 and reveal the effect of annealing treatment. For all samples annealed to

T = 873 K the peak positions match the diffraction lines of MgO periclase phase

(JCPDS card # 45-0946) (Figure 6.13 a). No additional diffraction features which

would point to iron segregation and phase separation of ferric or ferrous phase are

observed. This proves phase homogeneity throughout all powder samples which are

annealed up to temperature of T = 873 K.

Figure 6.13: X-ray diffraction (XRD) patterns of MgO and Fe-Mg-O powder samples

after annealing to T = 873 K (a) and T = 1173 K (b). 1) MgO; 2) (1%) Fe-Mg-O; 3)

(6%) Fe-Mg-O; 4) (9%) Fe-Mg-O. Vertical lines correspond to the standard XRD

patterns of periclase MgO (solid lines, JCPDS card # 45-0946) and spinel

magnesioferrite (MgFe2O4) (dashed lines, JCPDS card # 36-0398).

The diffraction patterns of the samples annealed to T = 1173 K reveal that MgO

(Figure 6.13 b-1) and (1%) Fe-Mg-O (Figure 6.13 b-2) retain their periclase phase

without emergence of a second crystalline phase. After annealing to T = 1173 K the

XRD patterns further reveal additional diffraction lines for (6%) Fe-Mg-O and

(9%) Fe-Mg-O samples. The most intense peaks are attributed to (220), (311), (422)

and (511) planes of the magnesioferrite spinel phase (JCPDS card # 36-0398).


99

The XRD patterns in Figure 6.13 b reveal narrowing of the diffraction features

widths of the periclase phase. This points to volume increase of the coherently

scattering solid and, therefore, annealing induced particle coarsening which is in

good agreement with particle growth deduced from TEM images. The average

crystalline domain sizes of periclase and magnesioferrite phases of the samples were

measured by Scherrer equation and from the reflex broadening (Table 6.2). For the

samples annealed to temperature of T = 873 K, the calculated average crystalline

domain sizes of periclase phase adopt similar values. The (1%) Fe-Mg-O sample,

however, shows slightly smaller domain size. The average crystalline domain size

increases by annealing of the samples from T = 873 K to T = 1173 K. For both

periclase and magnesioferrite phases this effect is smaller for the (9%) Fe-Mg-O

sample than (1%) Fe-Mg-O and the (6%) Fe-Mg-O samples.

Table 6.2: Average crystalline domain sizes of periclase phase and magnesioferrite

phase determined by Scherrer equation and from full width at half maximum (FWHM)

of the diffraction peaks.

sample annealed at T = 873 K annealed at T = 1173 K

periclase phase, average domain size / nm

MgO 11 ± 1 17 ± 1

(1%) Fe-Mg-O 9 ± 1 34 ± 1

(6%) Fe-Mg-O 12 ± 1 35 ± 1

(9%) Fe-Mg-O 12 ± 1 25 ± 1

magnesioferrite phase, average domain size / nm

(1%) Fe-Mg-O - -

(6%) Fe-Mg-O - 33 ± 1

(9%) Fe-Mg-O - 27 ± 1

Part II: Optical Spectroscopy

In the following discussion we will concentrate on optical absorption and

photoluminescence spectra of (1%) Fe-Mg-O and (9%) Fe-Mg-O samples.


100

6.4.4 UV-Vis Diffuse Reflectance Spectroscopy

The UV-Vis diffuse reflectance spectra were acquired at room temperature in

the presence of 10 mbar O2 to avoid possible photoluminescence. After data

acquisition they are converted to absorption spectra via the Kubelka-Munk transform

procedure (Figure 6.14).

Figure 6.14: Room temperature UV-Vis diffuse reflectance spectra of MgO and

Fe-Mg-O powder samples after oxidative heat treatment: a) (1%) Fe-Mg-O sample

annealed to T = 873 K; b) (9%) Fe-Mg-O sample annealed to T = 873 K; c)

(1%) Fe-Mg-O sample annealed to T = 1173 K; d) (9%) Fe-Mg-O sample annealed to

T = 1173 K. The sum curve and the individual resolved absorption bands are shown

as colored lines. The spectra are measured at T = 298 K and in the presence of

10 mbar O2.

Undoped MgO nanoparticle powders show absorption band with maxima at

hv = 5.7 eV (λ = 220 nm) for samples annealed up to T = 873 K and T = 1173 K.

This absorption band is attributed to the electronic excitation of 4-coordinated anions

located in edges of MgO nanocubes.26 Unlike MgO the optical absorption spectra of

Fe-Mg-O samples reveal additional overlapping absorption bands in the UV-Vis


101

range. The absorption spectra can be characterized as a superposition of absorption

bands with different band maxima. The absorption spectra can be deconvoluted into

several bands via Gaussian curve fitting using OriginPro 9.0 peak analyzer. The

results of the curve-fitting are shown in Figure 6.14 and Table 6.3.

Table 6.3: Curve fitting results related to spectra shown in Figure 6.14. FWHM = full

width at half maximum height of the fitted band.

sample annealed to

T = band

center max / eV

FWHM integral area /

%

(1%) Fe-Mg-O 873 K A 2.6 0.38 1.0

B 3.1 0.33 1.7

C 3.3 0.40 2.7

D 4.3 0.90 47.4

E 5.0 0.60 8.5

F 5.6 0.90 38.7

(1%) Fe-Mg-O 1173 K A 2.6 0.38 3.7

B 3.1 0.33 4.0

C 3.3 0.40 6.3

D 4.3 0.90 49.2

E 5.0 0.60 18.0

F 5.6 0.90 18.9

(9%) Fe-Mg-O 873 K A 2.6 0.42 7.6

B 3.1 0.50 9.9

C 3.3 0.75 23.7

D 4.3 1.00 34.9

E 5.0 0.70 11.8

F 5.6 0.90 12.2

(9%) Fe-Mg-O 1173 K A 2.6 0.38 6.0

B 3.1 0.33 6.8

C 3.3 0.40 12.8

D 4.3 0.90 38.3

E 5.0 0.60 14.0

F 5.6 0.90 22.2

Absorption spectrum of (1%) Fe-Mg-O sample annealed at T = 873 K is shown

in Figure 6.14 a. The spectrum consists of 6 absorption bands in the UV-Vis range.

The spectrum shows a well-defined band of low intensity centered on hv = 2.6 eV

(λ = 480 nm) (A) and two slope changes which suggest the presence of two bands

centered near hv = 3.1 eV (λ = 400 nm) (B) and hv = 3.3 eV (λ = 370 nm) (C). Two

intense and well-defined bands at hv = 4.3 eV (λ = 290 nm) (D) and hv = 5.6 eV

(λ = 222 nm) (F) are also present.


102

Figure 6.14 b shows the spectrum of (9%) Fe-Mg-O sample annealed to

T = 873 K. The fitted spectrum in Figure 6.14 b is composed of 6 bands around

hv = 2.6 eV (λ = 480 nm) (A), hv = 3.1 eV (λ = 400 nm) (B), hv = 3.3 eV (λ = 370 nm)

(C), hv = 4.3 eV (λ = 290 nm) (D), hv = 5.0 eV (λ = 250 nm) (E) and hv = 5.6 eV

(λ = 222 nm) (F) (Table 6.3).

Figure 6.14 c and d show the absorption spectra of Fe-Mg-O samples after

annealing to T = 1173 K. Comparison between Figure 6.14 a and c reveals a

decrease in relative intensity of the absorption bands D and C centered at

hv = 4.3 eV (λ = 290 nm) and hv = 3.3 eV (λ = 370 nm), respectively, while the

relative intensities of bands A, B and C remain essentially constant. Increase of the

annealing temperature from T = 873 K to T = 1173 K, additionally, leads to a change

in relative intensity of the absorption bands D and F centered at hv = 4.3 eV

(λ = 290 nm) and hv = 5.6 eV (λ = 222 nm), respectively.

In comparison to (1%) Fe-Mg-O sample the absorption spectrum of

(9%) Fe-Mg-O sample (Figure 6.14 d) undergoes more significant changes upon

annealing to T = 1173 K.

Figure 6.14 further shows absorption spectra of MgO samples annealed to

T = 873 K and T = 1173 K and reveals one intense band at hv = 5.7 eV (λ = 220 nm).

Finally, it is important to note that all spectra shown along with Figure 6.14 can be

consistently deconvoluted by one set of bands, A to F, with constant band-maxima

and widths (Table 6.3).

6.4.4.1 Electronic Absorption Spectra of Fe3+ Oxides

To clarify the origin of the spectral bands it is worthwhile to first discuss Fe

related optical absorption spectra in minerals. Absorption bands in the UV-Vis range

correspond to electronic transitions within the 3d5 atomic orbitals of Fe3+ cations.

There are essentially three types of electronic transitions: 1) Fe3+ ligand field

transitions, 2) transitions due to excitation of magnetically coupled adjacent Fe3+

cations, and 3) ligand-to-metal charge transfer (LMCT) transitions.77,174–176


103

Based on the ligand field theory for an ion with the electron configuration of 3 d5

the electronic ground state of Fe3+ free ion is described by the term 6S. When Fe3+

ion is surrounded by ligands in octahedral coordination, the d atomic orbitals of Fe3+

ion are split into two groups of orbitals expressed as t2g and eg which are separated

by ligand field splitting energy. The separated orbitals written as (t2g)3(eg)

2

configuration characterize the 6A1 state as the ground state.

The absorption bands in the spectra of Fe3+ containing minerals which are

associated with Fe3+ ligand field transitions basically originate from excitations of the

6A1 (6S) ground state to higher energy states (4E,

4A1 (4G), 4E (4D), 4T1 (4P),…).

Simultaneous excitation of two adjacent Fe3+ cations gives rise to additional bands in

the visible region. LMCT transitions occur at energies higher than ligand field

transitions and produce absorption bands below near UV region.77 The following

band assignment is based on work by Sherman and Waite77 about electronic spectra

of Fe3+ oxides.

6.4.4.2 Assignment of the absorption bands in Fe-Mg-O spectra

The absorption band A centered around hv = 2.6 eV (λ = 480 nm) (Figure 6.14

and Table 6.3) is assigned to the 6A1 +6A1

→ 4T1 (4G) +4T1 (

4G) excitation of Fe3+-Fe3+

pair. Absorption bands B centered at hv = 3.1 eV (λ = 400 nm), C at hv = 3.3 eV

(λ = 370 nm) and D at hv = 4.3 eV (λ = 290 nm) are attributed to the

6A1 → 4E, 6A1 (

4G), 6A1 → 4E (4D) and 6A1 → 4T1 (4P) ligand field transitions of Fe3+,

respectively. The additional absorption bands of E centered at hv = 5.0 eV

(λ = 250 nm) and F at hv = 5.6 eV (λ = 222 nm) are attributed to ligand-to-metal

charge transfer transitions.

6.4.5 Photoluminescence Spectroscopy

The photoluminescence spectra of undoped MgO and Fe-Mg-O powder

samples annealed to T = 873 K and T = 1173 K are given in Figure 6.15, Figure 6.16

and in supporting information Figure S6.3. The spectra recorded in vacuum show that


104

both emission band position and intensity change with Fe ion concentration,

annealing temperature and excitation energy.

Figure 6.15: Room temperature photoluminescence (PL) spectra of MgO and

Fe-Mg-O powder samples after annealing to T = 873 K (a) and T = 1173 K (b). PL

emission spectra are recorded in vacuum (p < 10-5 mbar) and at T = 298 K using an

excitation energy hvExc = 5.2 eV (λExc = 240 nm).

Figure 6.15 a shows the PL emission spectra of samples annealed to T = 873 K

for excitation energy hvExc = 5.2 eV (λExc = 240 nm). The spectra reveal for MgO a

strong emission band centered at hvEm = 3.2 eV (λEm = 390 nm), for (1%) Fe-Mg-O a

weak and broad emission feature with maxima around hvEm = 3 eV (λEm = 410 nm)

and in the case of the (9%) Fe-Mg-O sample the absence of any emission. Sample

annealing to T = 1173 K enhances the MgO specific emission. The PL emission

features of Fe-Mg-O samples however are perfectly quenched after annealing to

T = 1173 K.


105

Figure 6.16: Room temperature photoluminescence (PL) spectra of MgO and


emission spectra are recorded in vacuum (p < 10-5 mbar) and at T= 298 K using an


The PL emission intensity of MgO and Fe-Mg-O samples decreases upon

changing the excitation energy from hvExc = 5.2 eV (λExc = 240 nm) to hvExc = 4.6 eV

(λExc = 270 nm). MgO samples annealed to T = 873 K and T = 1173 K show broad PL

emission feature with maximum at hvEm = 2.9 eV (λEm = 430 nm) (Figure 6.16); the

position of the maximum intensity is independent of the annealing temperature. The

PL spectrum of the (1%) Fe-Mg-O sample annealed to T = 873 K reveals a broad

emission feature with maximum at hvEm = 3.2 eV (λEm = 380 nm). Annealing to

T = 1173 K leads to blue shift of the PL emission maximum to hvEm = 4 eV

(λEm = 310 nm) (Figure 6.16 b). For the (9%) Fe-Mg-O samples the excitation energy

of 4.6 eV (λExc = 270 nm) produces no measurable PL emission (Figure

6.16 a and b).

PL measurements in oxygen atmosphere (p = 10 mbar O2) (spectra not shown

here) results in PL emission quenching of all bands observed for MgO and for

(1%) Fe-Mg-O samples.


106

6.5 Discussion

In the present study the Fe-Mg-O nanocomposite particles of adjustable

composition and synthesized by the MO-CVS approach are characterized. The effect

of oxidative post-synthesis heat treatment on the compositional homogeneity of the

samples, the particle size distribution and the optical absorption properties is

investigated.

The Fe/(Fe+Mg) concentrations as determined by ICP-OES clearly shows

increasing iron concentrations with increasing evaporation temperatures of the iron

precursor in the MO-CVS reactor (Table 6.1). We considered samples with three

different iron concentrations for local elemental characterization by EDX. Although

the EDX results are not representative for the entire powder sample they illustrate

how uniformly iron is distributed throughout the particle ensemble. As a result,

compositional homogeneity changes with the annealing temperature. The

composition distribution function of the (6%) Fe-Mg-O and (9%) Fe-Mg-O samples

changes above an annealing temperature of T = 873 K. However, for the

(1%) Fe-Mg-O sample increase of the annealing temperature from T = 873 K to

T = 1173 K does not change the composition distribution function.

Mößbauer spectroscopy points to the exclusive existence of Fe3+ ions in the

Fe-Mg-O nanocomposite samples, irrespective from the total iron concentration and

annealing temperature. Addition of iron to the magnesium oxide lattice leads to

substitution of Mg2+ ions by Fe3+ ions. For all sample types which were annealed at

T = 873 K Mößbauer spectroscopy reveals the presence of Fe3+ ions in the 6 fold

coordination state. This is also true for the (1%) Fe-Mg-O sample annealed to

T = 1173 K. For the (6%) Fe-Mg-O and (9%) Fe-Mg-O samples annealed to

T = 1173 K, however, Mößbauer spectroscopy points to the presence of a

magnesioferrite phase which adopts the spinel structure with Fe3+ ions which are

distributed between the octahedral and tetrahedral interstices.

From the particle size distributions as determined by TEM analysis it is found

that the Fe-Mg-O particle ensembles annealed at T = 873 K exhibit average particle

sizes smaller than that of undoped MgO. We assume that up to this temperature the


107

incorporation of Fe3+ ions in the magnesium oxide lattice has a stabilizing effect. The

origin for stabilizing effect observed at T = 873 K is presently unresolved. By

annealing of the samples to higher temperature this stabilization effect is lost and

broadened particle size distributions with much larger particles are observed after

annealing to T = 1173 K (Figure 6.9). Except for (1%) Fe-Mg-O nanoparticles which

retain the cubic shape, most of the large particles formed at T = 1173 K adopt less

regular morphologies.

The Fe3+ ions in the lattice of MgO produce cation vacancies in order to

maintain the charge balance.62 Admixture of a larger amount of Fe3+ ions to the

lattice favors vacancy concentration enhancement and facilitates ion diffusion and

thus, sintering.

The structural characterization done by XRD in parallel to TEM analysis

confirms crystallite domain growth for all samples upon annealing temperature

increase. For (6%) Fe-Mg-O and (9%) Fe-Mg-O samples which were annealed to

T = 1173 K we observe phase separation in addition to particle growth. In line with

Mößbauer spectroscopy, the XRD patterns of the (6%) Fe-Mg-O and (9%) Fe-Mg-O

samples indicate the presence of magnesioferrite and periclase phases.

The influence of annealing on crystal growth and phase separation of Fe-Mg-O

samples is sketched in Figure 6.17. The magnesium flame in reactor (T ≈ 2000 K)

during oxidation reaction generates a non-equilibrium solid with high concentration of

iron (> 1 at.%) in the as-synthesized samples. This leads to formation of metastable

Fe-Mg-O nanocomposites with Fe3+ ions accommodated in different sites of the MgO

lattice. Consistent with the low solubility limit of Fe3+ in MgO lattice (XFe = 0.3 at.% at

T = 1273 K), annealing of the metastable particles to T = 873 K results in the surface

segregation of Fe3+ ions. Ion diffusion enhanced by the Mg2+ vacancies clearly favors

the crystal growth. Annealing of (1%) Fe-Mg-O sample to T = 1173 K further

enhances the ion mobility which would favor the thermodynamically driven

transformation into cubic particles. For samples with higher total iron concentrations

this leads to the phase separation into magnesioferrite and periclase phases.

.


108

Figure 6.17: Schematic illustrating the process of Fe3+ concentration dependent

crystal growth and phase separation upon annealing.

During phase separation several intermediate Fe-Mg-O structures are expected

to form. An exact identification of intermediate phases, however, is rather difficult at

this stage and would require more experiments and characterization work that is out

of scope of current work.

Optical properties of Fe-Mg-O nanocomposite samples are investigated using

UV-Vis diffuse reflectance and photoluminescence spectroscopy. The optical

absorption spectra of the Fe-Mg-O samples were deconvoluted into different bands

(Table 6.3). On the basis of band maxima they are assigned to electronic transitions

in different ligand fields, to the simultaneous excitation of magnetically coupled

adjacent Fe3+ cations or to ligand-to-metal charge transfer (LMCT) transitions. In the

light of the ligand field theory variations in absorption band intensities can be

explained by essentially two factors: 1) Fe3+ concentration and 2) annealing

temperature.


109

Different to the absorption properties of (1%) Fe-Mg-O samples, absorption

spectra of annealed (9%) Fe-Mg-O samples show increased absorption band

intensities in the energy range of UV-Visible spectra. Ligand field transitions are in

principle spin forbidden but magnetic coupling of neighbor Fe ions can break this

condition.77 In particular, the intensity increase of band A in spectra of samples with

higher iron concentration (Figure 6.14 b & d) is attributed to an enhanced degree of

Fe3+- Fe3+ coupling as compared to samples with lower iron concentration.

As inferred from compositional and structural analysis, supply of higher thermal

energy to the Fe-Mg-O samples during annealing increases Fe3+ions diffusion in the

crystal lattice and affects the ion distribution in the structure. This produces phase

separation in samples with iron concentrations above 1 at.% and affects the average

Fe-to-Fe distance as well. Moreover, annealing to T = 1173 K expands metal-oxygen

(M-O) interatomic distance in oxide materials containing Fe and Mg ions.177 Changes

in the Fe-to-Fe distance and M-O bond length leave a critical influence on magnetic

coupling of the neighboring Fe ions.175 A combination of all these effects may explain

the Fe3+ ion concentration dependent variation of absorption band intensities

observed (Figure 6.14).

In comparison with MgO nanoparticles which exhibit intense PL emission

effects which exclusively originate from the excitation of surface low coordinated

sites, admixture of Fe3+ to MgO quenches PL emission. Related trends in PL

emission with increase of Fe3+ concentration in the nanocomposite samples (Figure

6.15 and Figure 6.16) are linked to surface segregated Fe3+ ions as static PL

quencher species. On this basis we conclude that annealing induces surface

segregation of Fe3+ ions for the Fe-Mg-O nanocomposites.

6.6 Conclusions

Fe-Mg-O nanocomposite particles were produced by metal organic chemical

vapor synthesis using iron (III) acetylacetonate as metal precursor. This gives a very

good control over the iron concentration and provides the opportunity to study

composition and stability of metastable ternary nanocomposites of Fe-Mg-O systems.


110

Mößbauer spectroscopy proves the existence of iron ions in their trivalent state. This

technique also reveals the formation of superparamagnetic nanoparticles in the case

of (1%) Fe-Mg-O samples annealed at T = 873 K and annealed at T = 1173 K. For

(6%) Fe-Mg-O and (9%) Fe-Mg-O samples it could have been found that annealing

to T = 1173 K leads to the separation into a periclase phase and a magnesioferrite

with antiferromagnetic property. Mößbauer results are in perfect agreement with

results from X-ray diffraction and electron microscopy and point to annealing induced

particle growth. Crystal growth and phase separation is mediated by cation vacancies

which emerge from the charge imbalance of Fe3+ ions in the lattice. This enhances

ion mobility and ultimately leads to phase separation and spinel formation. An

explanation for this process is based on the non-equilibrium nature of the here

described nanoparticles: chemical vapor synthesis generates Fe-Mg-O

nanocomposites with incorporated Fe3+ ions above solubility limit in MgO lattice

(XFe = 0.3 at.% at T = 1273 K). Annealing to T = 873 K and T = 1173 K induces

crystal growth and surface segregation of Fe3+ ions. This is driven by

thermodynamics in combination with Fe3+ induced increase of the ion vacancy

concentration in MgO lattice which facilitates ion diffusion. By annealing of the

samples to temperatures above T = 873 K, excess Fe3+ ions aggregate and some of

them move into the tetrahedral interstices. This leads to phase separation and to

precipitation of magnesioferrite spinel.

This study underlines for Fe-Mg-O nanocomposites that differences in ionic

charge drives the annealing induced particle growth and segregation of the admixed

Fe3+ cations. Knowledge about synthesis and controlled processing of such material

systems enables the production of nanocomposites with defined compositional,

structural and optical properties. These nanomaterials may find applications in areas

such as catalysis, magnetic devices and optoelectronics.


111


6.7.1 Example of Local Concentration Measurement

Figure S6.1: Local concentration determination for a Fe-Mg-O sample as measured

by energy dispersive X-ray (EDX) spectroscopy. The figure reveals data of

composition analysis made of 5 different sample positions (A, B, C, …) and spots

(colored) to measure the relative Fe and Mg concentrations and their spatial

distribution over a larger area of the casted nanoparticles.


112

6.7.2 Mößbauer Spectroscopy

Figure S6.2: The Mößbauer spectra of the Fe-Mg-O samples measured at room

temperature after annealing to T = 873 K and T = 1173 K. a & d) (1%) Fe-Mg-O;

b & e) (6%) Fe-Mg-O; c & f) (9%) Fe-Mg-O


113

Table S6.4: The high fine parameters of the fitted curves to the Mößbauer spectra of

the Fe-Mg-O samples.

sample annealed at

T = Ssp

/ mm s

-1

QS / mm s

-1

H(0) / T

/ mm s

-1

A / %

2

(1%) Fe-Mg-O 873 K - - - - - -

(6%) Fe-Mg-O 873 K 1 0.3(3) 1.0(x) - 0.5(6) 49.7 0.35

2 0.3(3) 0.5(x) - 0.4(1) 50.3

(9%) Fe-Mg-O 873 K 1 0.3(x) 1.0(x) - 0.3(x) 46.0 0.41

2 0.3(x) 0.5(1) - 0.3(x) 54.0

(1%) Fe-Mg-O 1173 K 1 0.302(5) 0.95(3) - 0.54(4) 63.0 0.35

2 0.34(2) 0.6(1) - 0.358(2) 37.0

(6%) Fe-Mg-O 1173 K 1 0.30(4) 0.9(8) - 0.630(2) 19.2 1.31

2 0.27(7) 0.02(2) 46.04(5) 0.506(6) 35.2

3 0.25(1) -0.02(5) 43.8(1) 0.446(0) 23.1

4 0.28(1) -0.02(5) 40.82(9) 0.716(0) 20.6

(9%) Fe-Mg-O 1173 K 1 0.28(3) 0.8(7) - 0.694(2) 15.4 3.14

2 0.27(5) 0.01(2) 45.45(3) 0.538(4) 37.5

3 0.263(9) -0.01(4) 42.77(7) 0.486(0) 25.1

4 0.273(8) -0.03(4) 39.04(7) 0.800(0) 20.6

In most cases, we obtained quite well resolved spectra which were fitted with a

conventional self-made refinement routine using Lorentzian line shapes and different

sets of Mössbauer parameters corresponding to the non-equivalent crystallographic

sites involved. The parameters displayed in Table S6.4 have the following meaning:

Isomer shift relative to α-iron [mm/s]; Half width of the lines [mm/s]; Quadrupole

splitting QS = 1/2 eQVzz1+ɳ2/3 [mm/s] where Vzz = z-component of the electric field

gradient efg, Q = nuclear quadrupole moment, ɳ = asymmetry parameter (Vxx-Vyy)/Vzz

with |Vzz| ≥ |Vyy|≥|Vxx|; Internal magnetic field H(0) [T]; Amount of area [%] of the

relevant subspectrum – a measure for the site occupation of the corresponding Fe

ion.

The convergence of the iteration and the 2-value of the final fit served as an

indication of the quality of the refinement. The rather low values of 2 can be

considered as being rather good. The values are given with errors in round brackets;

where the latter exceeds the value, an “x” is given instead. The samples are listed in

the same order as the corresponding spectra of Figure S6.2, the underlying

subspectra are denoted as Ssp.; contributions lower than 2% (amount of area) are

left off.


114

6.7.3 PL Spectra

Figure S6.3: Room temperature photoluminescence (PL) spectra of MgO and


emission spectra are recorded in vacuum (p < 10-5 mbar) and at T= 298 K using an


7 Spontaneous Growth of Magnesium Hydroxide Fibers at Ambient Conditions

115

7 Spontaneous Growth of Magnesium Hydroxide Fibers at

Ambient Conditions

7.1 Abstract

Spontaneous transformation paths of nanomaterials point to guiding principles

for synthesis. We describe the room temperature transformation of MgO nanocubes

into Mg3(OH)5Cl·4H2O nanofibers in air and investigated the underlying formation

mechanism using electron microscopy, X-ray diffraction, and solid-state NMR

spectroscopy. Upon contact with water vapor, the magnesium hydroxide needles

were found to grow out of agglomerates of highly dispersed MgO nanocubes with

preadsorbed SiCl4. Corresponding one-dimensional nanostructures do not form on

low surface area materials. The presented growth approach is potentially extendable

to other hydrolyzable metal oxides at ultrafine dispersion.

7.2 Introduction

Spontaneous transformations of nanostructures are important key to their

chemical synthesis and application for two reasons: first, the underlying mechanisms

may provide guiding principles for the synthesis and controlled spatial arrangement of

anisotropic nanostructures.178–181 Second, knowledge about the transformation

behavior of nanomaterials in the environment is needed in order to reliably assess

the potential risk to biological systems. Depending on the nature of particle surfaces

and their interactions with gases in the atmosphere, disperse systems can undergo

completely changed interfacial chemistries with altered reaction networks as

compared to bulk materials.182

MgO nanocubes are excellent model systems for surface chemistry studies on

unsupported nanomaterials.25,118 At the same time, they are well-suited building

blocks for pure and composite nanostructures.143,183 We investigated different

approaches for SiCl4 adsorption on related particle ensembles (Supporting

Information and Chapter 8, Section 8.4.1) and discovered an unprecedented

transformation process at ambient conditions.


116


The following combined electron microscopy and diffraction study was

performed to explore the underlying mechanism. A dry powder of MgO nanoparticles

which was casted on the SEM grid can be characterized as agglomerates exhibiting

a fine-grained particulate structure.184 The high dispersion results from the MgO

nanocrystals of cubic morphology and an average edge length of approximately

6 nm. MgO nanocubes exhibit a periclase structure and are phase pure (Figure 7.1

d1). The relatively broad diffraction features are related to the limited crystalline

domain sizes (d ≈ 6 nm).

Figure 7.1: (a) SEM image of a MgO nanocubes agglomerate after room

temperature storage in humid air (p(H2O) = 32 mbar); (b) SEM and TEM images of

MgO nanocubes after contact with SiCl4 and subsequent room temperature storage

in vacuum (p < 10−5 mbar). Additional TEM images are shown in the Supporting

Information. (c) EM images of SiCl4 contacted MgO nanocubes after room

temperature storage in humid air; (d) X-ray diffraction patterns of (d1) MgO

nanocubes, (d2) MgO nanocubes after storage in humid air; (d3) MgO after SiCl4/O2

contact and storage in a vacuum (p < 10−5 mbar), and (d4) SiCl4 contacted MgO

nanocubes kept in humid air. In all experiments, the room temperature storage time

was 14 days.


117

After 3 days of sample contact with water-saturated air (p(H2O) = 32 mbar) at

room temperature, MgO nanocubes were quantitatively converted in to the brucite

phase (Figure 7.1 d2) according to

MgO + H2O MgO(OH)2 Equation 7.1

From the widths of the diffraction features, we determined for the crystallite

dimensions x001 and x110 coherence lengths of 5 and 10 nm, respectively, using the

Scherrer equation. This is in good agreement with the plate like morphology reported

of brucite nanocrystals.185,186 In the case of MgO which was previously contacted with

SiCl4 and O2 and exposed to water vapor thereafter, the material was converted into

a solid that is characterized by an entirely different X-ray diffraction (XRD) pattern

(Figure 7.1 d4). Phase analysis on the basis of pattern indexing and lattice parameter

refinement led to the compound Mg3(OH)5Cl·4H2O, thereby closely matching the

PDF database card #7-420.13 The refined cell parameters and atomic coordinates

were determined by Rietveld refinement of the powder XRD and are summarized in

Table S7.1, Table S7.2 and Table S7.3 (Supporting Information).

The transformation of MgO into Mg(OH)2 and Mg3(OH)5Cl·4H2O needles can

also be tracked by 25Mg NMR spectroscopy. A narrow resonance at 26.0 ppm as

observed for MgO nanocubes reveals a well-ordered Mg environment without defects

and impurities (Figure 7.2 a). Exposure to water vapor triggers the reaction described

in Equation 7.1, and the 25Mg NMR spectrum shows broad quadrupolar resonance

line (δ = 13 ppm, CQ = 3.0 MHz, ηQ = 0) characteristic for Mg(OH)2 (Figure 7.2

b).187,188 The spectrum also indicates a residual small fraction of MgO. The single

magnesium resonance observed for the chlorinated MgO powder which was stored in

a vacuum (Figure 7.2 c) indicates that the MgO nanocubes remain essentially

unaltered at this stage of sample treatment.

Exposure of chlorinated MgO nanocubes to water vapor initiates growth of

Mg3(OH)5Cl·4H2O needles. The 25Mg NMR spectrum of this new phase (Figure 7.2 d)

shows a signal at 0 ppm. Since 25Mg is a quadrupolar nucleus, its line shape also

gives some information about the orientation of the nucleus in the local electric field.

Narrow and symmetrical NMR resonance line of the Mg3(OH)5Cl·4H2O reflects a


118

symmetric environment of Mg in the MgO6 octahedra. In addition, the spectrum

confirms that Mg has been quantitatively incorporated into Mg3(OH)5Cl·4H2O.

Figure 7.2: Solid state single-pulse 25Mg MAS NMR spectra of MgO nanocubes

(a) before and (b) after exposure to H2O vapor (p(H2O) = 32 mbar) in a closed

sample chamber. The spectrum of the MgO powder sample which was contacted

with SiCl4/O2 and stored in a vacuum thereafter is shown as trace (c). Trace (d)

corresponds to the spectrum of Mg3(OH)5Cl·4H2O.

We carried out control experiments on commercially available low surface area

MgO powders (Aldrich #529699) with average domain sizes of dXRD = 35 nm and

found that - irrespective from whether the oxide was contacted with SiCl4/ O2 prior to

H2O vapor exposure or not - the powders became only superficially transformed into

the hydroxide of the brucite modification (with crystallite dimensions x001 and x110 of

28 and 25 nm, respectively. Further details in Chapter 8, Section 8.6.5). The absence

of needle growth as concluded from the corresponding XRD and transmission

electron microscopy (TEM) data clearly underlines the critical importance of high

surface-to-volume ratios for the here reported transformation process.

The associated increase in surface reactivity makes chlorinated MgO

nanocubes to physicochemically dynamic materials in environmental media.


119

Humidity, i.e., the presence of water molecules in the gas phase, is key to the growth

of Mg3(OH)5Cl·4H2O needles. In the presence of water vapor, the hydrolyzable Mg-O

surface elements break. In a way similar to the hydrolysis and polycondensation

reactions of typical sol−gel reactions, the broken Mg-O bond leads to the formation of

Mg2+ and OH− ions to form magnesium hydroxide, which spreads and forms thin

ribbon-like films on top of the particle agglomerates. The transformation is driven by

the high energy of highly dispersed chlorinated MgO/ Clx particles and SixClyOz

moieties (related experimental evidence see Supporting Information) as educts and

the thermodynamic stability of the newly formed compound Mg3(OH)5Cl·4H2O. Figure

7.3 illustrates the growth mechanism.

Figure 7.3: Schematic illustrating the growth mechanism of magnesium oxychloride

fibers from MgO nanocubes kept in H2O vapor environment. (1) MgO nanocubes; (2)

MgO nanocubes covered with SixClyOz layers (experimental NMR evidence in

Supporting Information); (3) Mg3(OH)5Cl·4H2O fibers start to grow out from the grains

of chlorinated and agglomerated nanocubes (see corresponding TEM image in the

inset and in the Supporting Information) as a result of contact with water vapor.

(4) After 14 days, the formation of hedgehog like structures is completed.

We assume that a thin liquid layer forms on the surface of the chlorinated

nanocube agglomerates and provides a new reaction medium upon contact with

water vapor. The local supersaturation of dissolved ions leads to the precipitation of

Mg3(OH)5Cl·4H2O seeds that anisotropically grow out of the reaction pool (Figure


120

7.3) with [010] as the growth direction and the (101) plane being subject to fast

material addition and growth. Further needle growth is sustained by ion transport

inside the aqueous film and across the concentration gradient.

7.4 Conclusion

Studies on the interplay between hydration state, microstructure, and

mechanical properties of oxyhydrates revealed the growth of Mg3(OH)5Cl·4H2O

needles in aqueous pastes of MgO/ MgCl2 mixtures.189–191 More recently material

chemists reported the solution synthesis of oxychloride nanorods upon contact of

nanocrystalline MgO with concentrated aqueous magnesium chloride solution.192 As

a completely novel effect, the here reported transformation process does occur in air

and leads to 1-D hydroxides with interfacial regions that are susceptible to ion

exchange. The overall process points to a simple and direct way to spatially control

needle growth via arrangement of MgO nanocube ensembles (Figure 7.1 a) as

seeding regions. Moreover, constitutional water at the surface as well as inside the

bulk structure provides additional functionality for the deposition of catalyst

particles,193 for ion insertion and exchange reactions that are critical for materials

design in energy conversion technology194–196 as well as for the development of

structural materials.191,196 From a different perspective, these results underline the

substantial metastability of nanomaterials and that many unintended transformations

may easily occur in both environmental and biological systems.


121


7.5.1 MgO Exposure to SiCl4

SiCl4 was cleaned employing the freeze-pump-thaw method prior to the

adsorption on MgO nanocubes. The base pressure of the cell containing MgO

nanocube powder, which was kept at room temperature, was less than 10-5 mbar.

The cell was immersed into liquid nitrogen in order to keep a constant temperature of

T = 77 K (low temperature adsorption = LTA). Typically, during one cycle, 200 mg of

MgO nanocube powder was exposed to SiCl4 gas with a pressure of 300 mbar for 5

minutes and subjected to subsequent oxidation step in O2 atmosphere (p = 700

mbar) for another period of 5 minutes. The numbers of SiCl4 and O2 molecules

provided in each reaction step correspond to 3x1021 for both gases (T= 298 K).

During subsequent evacuation to a base pressure of p < 10-4 mbar gaseous

reactants that did not adsorb on the particle surface were removed and the powder

sample was then subjected to a new cycle with fresh reactants. At the end of 6th cycle

evacuation to the base pressure is done at T = 298 K. The results presented in this

study were obtained on MgO nanocube powders which were subjected to 6 cycles.

7.5.2 Water assisted Needle Growth

The growth of needle-like structures results from the contact of chlorinated MgO

nanocubes (see part 7.5.1) with water vapor at room temperature. To keep this

process under controlled conditions, the chlorinated MgO powder samples were kept

under constant water vapor pressure (32 mbar H2O) inside a closed chamber (Figure

S7.1) at T = 298 K.

Figure S7.1: Experimental set up for controlled exposure of chlorinated MgO

nanocube powders to water vapor.


122

7.5.3 Analysis of the X-ray Diffraction Pattern

Indexing of the powder diffraction pattern of the sample contacted with SiCl4

and exposed to water vapour thereafter was successfully done using EXPO2013198

leading to a primitive monoclinic unit cell, space group P2/m. The determined lattice

parameters closely match those of Mg3(OH)5Cl·4H2O.198 Taking the atomic model of

Sugimoto et al. the crystal structure of Mg3(OH)5Cl*4H2O was refined using the

Rietveld method with FULLPROF.199 Thereby the atomic parameters of Mg, O and Cl

were allowed to vary freely, while the H-positions where fixed to the values given in

reference [198]. The isotropic atomic displacement parameters were grouped

together for the oxygen atoms and the Mg-atoms, but allowed the refine freely. It was

noted by Sugimoto et al. (2007) that the free Cl atom shows a mixed occupation with

0.5 OH2 + 0.5 Cl- This was confirmed in our refinement however with a slightly

different occupation of this site with 0.62(1) O + 0.38(1) Cl-.

Figure S7.2: Rietveld refinement of the XRD pattern shown in Figure 7.1 d4. The

upper tick marks correspond to Mg3(OH)5Cl·4H2O (F5-Phase) while the lower ones

are from the 2.0 wt.% impurity of tridymite SiO2. The difference plot between

observed and calculated intensities is shown in the lowest part of the Figure,

mismatch around 12°, 21° and 37° 2Theta is due to distinct internal micro-strain of

the sample which could not be modeled perfectly.


123

Table S7.1: Refined structural parameters for Mg3(OH)5Cl·4H2O (F-Phase) as

extracted from Rietveld refinements.

Space group P2/m Temperature 298 K a (Å) 9.6553(6) b (Å) 3.15207(19) c (Å) 8.3087(6)

(°) 90.00000

(°) 114.022(4)

(°) 90.00000

Unit cell volume (Å3) 230.97(3) Radiation Cu K1,2

Table S7.2: Refined atomic coordinates

ID x/a y/b z/c Uiso

Mg1 0.8045(7) 0 0.6981(7) 0.0234(18) Mg2 ½ ½ ½ 0.0234(18) O1 0.7258(9) ½ 0.5171(13) 0.0087(16) O2 0.5833(9) ½ 0.6813(9) 0.0087(16) O3 1.0218(8) ½ 0.6879(9) 0.0087(16) O4 0.8714(7) ½ 0.8821(10) 0.0087(16) Cl1a 0.6792(6) ½ 0.1104(7) 0.0087(16)

a the occupation of this site is 0.62(1) O + 0.38(1) Cl

Uiso = isotropic atomic displacement parameter

Table S7.3: Fixed atomic coordinates as taken from Sugimoto et al. 198

H1 0.73240 ½ 0.41200 0.02533 H2 0.58620 ½ 0.79360 0.02533 H3 1.10570 -0.18750 0.75870 0.02533 H4 1.04960 0.18750 0.60780 0.02533 H5 0.80000 ½ 0.95340 0.02533 H6 0.96590 ½ 0.96880 0.02533 H7 0.76790 ½ 0.08650 0.02533 H8 0.58200 ½ 0.01860 0.02533


124

Figure S7.3: Graphical representation of the crystal structure of Mg3(OH)5Cl·4H2O

(F5-Phase) in a projection onto the a-c plane.

7.5.4 Solid State NMR Experiments

Solid-state 25Mg and 29Si MAS (magic-angle spinning) NMR (nuclear

magnetic resonance) spectra were recorded on a 500 MHz Agilent NMR system

using Agilent 3.2 mm and 6 mm T3 HXY MAS Solids Probes. The 29Si and 25Mg

Larmor frequencies amounted to 99.30 MHz and 30.60 MHz, respectively. Single

pulse NMR spectra repetition delays were 5 s and 20 s, the number of scans

accumulated was in the order of magnitude of 10,000 and 1000. All the samples

were spun at 5 kHz during measurement.

20 0 -20 -40 -60 -80 -100 -120

(b)

/ ppm

(a)

29Si sp MAS NMR

Figure S7.4: 29Si MAS NMR spectra of the MgO nanocubes after exposure to SiCl4

and O2 (a) and storage in vacuum (b) which indicates the presence of short

polychlorosiloxane chains.


125

29Si single pulse MAS NMR spectrum (Figure S7.4) of MgO immediately after

chlorination shows only the presence of SiCl4 species (-21 ppm). However, after

storing the chlorinated sample in vacuum for two weeks, also peaks attributed to

SixClyOz compounds are observed. The much higher relative intensities of the -SiCl3

(-45 ppm) resonance compared to the -SiCl2- (-71 ppm) peak indicate the presence

of short chlorosiloxane chains, predominantly Cl3Si-O-SiCl3 and Cl3Si-O-SiCl2-O-

SiCl3.

7.5.5 SiCl4/ O2 Contacted MgO Nanocubes: Additional TEM Images

Figure S7.5: TEM images of MgO nanocubes after exposure to SiCl4/ O2 and

storage in vacuum. The structures which are in focus show clear-cut edges specific

to MgO nanocubes as the starting material.

8 Growth of Magnesium Oxychloride Fibers

127

8 Vapor phase based and Water Film mediated Growth of

Magnesium Oxychloride Fibers

8.1 Abstract

Highly dispersed metal oxides that are covered with thin films of water can give

rise to the spontaneous and spatially controllable growth of oxide and hydroxide

fibers in the ambient. Understanding the underlying formation mechanism is needed

for the exploration of related microstructure evolution and represents a critical

requirement for the rational development of industrial binders as well as for ceramic

precursors. This work examines the parameters of MgO nanocube functionalization

with oxychlorides, using SiCl4 as a water free chlorine ion source, and explores the

subsequent transformation of obtained composites into magnesium oxychloride

Mg3(OH)5Cl*4H2O fibers, which correspond to the main component of the oxychloride

cement phase. Specifically we show how the temperature of the functionalization

process as well as the materials’ level of dispersion determine the reaction pathway

to either obtain MgCl2·6 H2O and Mg(OH)2 or magnesium oxychloride fibers. Lessons

to be learned from this unique route to synthesize Mg3(OH)5Cl*4H2O nanofibers upon

water vapor contact can be applied to a variety microstructural evolution processes

that involve metal oxide nanoparticles in combination with superficial water which

acts both as a reactant as well as reaction medium for the hydration process.

8.2 Introduction

The implementation of nanostructured metal oxides and hydroxides into

functional and structural materials requires knowledge about their stability in

changing chemical environments. These will never be fully understood without the

examination of specific aspects of that complexity. Tailored particle systems with

interface properties that are accessible to experimental methods are indispensable

model systems for this purpose.13 This is also true for ceramics or cement-bonded

materials, where structure, texture and reactivity of the different constituent mineral

and mineral-like phases determine the functionality and stability of the composite. As

a nano- and microporous multicomponent system, cement is chemically highly


128

reactive. Various aspects or properties related to chemical interfaces are, however,

still not understood.200,201,202,203

Oxychloride (Sorel) cement corresponds to a fast curing cement for patching

the surface of freeways as well as other applications such as stucco, flame retardant

coatings and molded cement objects. Besides its good binding ability with inorganic

and organic compounds, the superior thermal, chemical and mechanical stability

makes it a high performance material which draws continuously further

attention.204,205 Sorel cement is typically produced by mixing an aqueous MgCl2 brine

solution with MgO powder.206,198 There is general agreement in the literature that this

cement emerges on the basis of a complex reaction network, which involves water,

MgO and MgCl2 admixed in specific proportions as starting materials189,191 but

considerable controversy as to what chemical reaction is responsible for the setting

reaction.

Relevant to Sorel cement formation, we recently discovered a spontaneous

formation process of magnesium oxychloride (Mg3(OH)5Cl*4H2O) fibers which grow

out of MgO-based nanoparticle agglomerates.207 This process occurs under ambient

conditions and starts from MgO nanocubes that are covered with SixOyClz moieties.

With a presumably important effect on the entire transformation process the impact of

different materials parameters, such as the required degree of dispersion, but also

process parameters, like the temperature for SiCl4 and O2 adsorption have remained

unexplored. In this study, we investigated their influence in detail to gain insights into

the underlying growth mechanism, to identify parameters for fiber growth optimization

and, ultimately, to control microstructure development. Moreover, an important

aspect related to the spontaneously forming needle-like shaped magnesium

oxychlorides relates to their stability towards annealing induced dehydration and their

potential to restore the parental oxide inside novel microstructures.

This Chapter is structured as follows: in the first part we provide a detailed

description of the MgO functionalization process which provides the reactive

precursor material leading to spontaneous growth of magnesium oxychloride fibers.

In particular, the challenge of achieving compositional homogeneity over the entire


129

nanoparticle ensemble is at the focus of this part. In the second part, insights about

the SiCl4 adsorption temperature, the effect of powder dispersion and surface area

enabled us to delineate the growth mechanism in the ambient, which involves thin

liquid water films as reaction medium. In the third part of this paper we addressed the

decomposition behavior of the magnesium oxychloride (Mg3(OH)5Cl*4H2O) fibers in

order to provide enabling knowledge for applications that operate at elevated

temperatures such as high temperature coatings, architectural materials or abrasive

tools.208,209,210,211


8.3.1 Chemical Vapor Synthesis of MgO

MgO nanoparticles were produced via a chemical vapor synthesis (CVS)

technique which is based on controlled combustion of metal vapor within a flow

reactor system.212,143 For further materials processing the samples are transferred

through ambient air into a quartz glass tube. In order to guarantee bare particle

surfaces as well as to achieve cubically shaped nanocrystals, thermal sample

activation via vacuum annealing is employed. This procedure leads full

dehydroxylation of sample surfaces as proved by FT-IR spectroscopy.107 A typical

procedure utilized for dehydration, dehydroxylation and removal of carbon−based

surface contaminants is as follows: as-synthesized powders are heated to T = 1123

K in high vacuum with a rate of 5 K min−1 and then are brought into contact with 10

mbar O2 at this temperature. Subsequently, the temperature is raised to T = 1173 K

and, at pressures p < 5∙10-6 mbar, kept at this temperature for 1 h before being

cooled down to room temperature.213 Beside CVS MgO we also employed

commercial nanocrystalline MgO (529699 Aldrich) for reference experiments (Figure

S8.1 and Figure S8.2).

As revealed by Transmission Electron Microscopy CVS MgO consists of highly

dispersed monocrystalline nanocubes with a high portion of edge and corner features

(Supporting Information, Figure S8.1 a). In contrast to MgO nanocubes, commercial

MgO can be characterized as an assembly of less regular shaped particles

(Supporting Information, Figure S8.1 b). Whereas the Scanning Electron Microscopy


130

(SEM) images of CVS MgO nanoparticles reveal a fine-grained material with an open

and porous secondary structure (Supporting Information, Figure S8.1 c), micrographs

related to commercial MgO particles reveal their more coarse grained nature.

Respective particles do not exhibit any characteristic and prevailing shape.

XRD confirms for both types of MgO powders the exclusive presence of

crystalline particles with Periclase structure (Supporting Information, Figure S8.2).

The average crystallite sizes, as determined from reflex broadening using the

Scherrer equation102,103, correspond to 6 ± 1 nm and 35 ± 1 nm for CVS MgO and

commercial MgO, respectively. While the measured specific surface area of CVS

MgO, which corresponds to SBET = 300 ± 20 m2·g-1, is perfectly consistent with the

XRD derived crystallite domain size, the situation is different for commercially

available MgO: in comparison to a XRD derived value of SXRD = 50 m2·g-1 we

measured with N2 sorption analysis a value of SBET = 12 ± 1 m2·g-1 which is lower by

a factor of five. This discrepancy points to polycrystalline grains exhibiting a large

fraction of intergranular solid interface area that is not susceptible to N2 adsorption.

Both CVS and commercial MgO were processed equally prior to functionalization and

growth studies described in this work.

8.3.2 MgO Surface Functionalization via SiCl4/ O2 Exposure

MgO is exposed to SiCl4 and O2 in a cyclic process. The schematic of the setup

used for this process is shown in Figure 8.1. Prior to adsorption SiCl4 is cleaned

employing the freeze-pump-thaw method. The base pressure of the sample tube at

room temperature (T = 298 K) is less than p = 10-5 mbar. For each cycle, the lower

part of the glass tube with the MgO nanocubes powder (adsorbent) was immersed

into a liquid nitrogen bath in order to keep T = 77 K as a constant temperature of

adsorption. Samples which were functionalized in this way will be designated in the

following as low temperature adsorption (LTA) samples.


131

Figure 8.1: Schematic illustrating the setup used for the MgO exposure to SiCl4/O2.


8.4.1 Surface Functionalization of MgO Particles and Spontaneous

Fiber Growth

Critical for functionalization as well as for the below described transformation

process is a - in terms of composition - homogeneous surface coverage of the MgO

particle ensembles with silicon tetrachloride (SiCl4) which acts here as a carbon free

source for chlorine.214 We performed respective gas phase functionalization by

SiCl4/ O2 admission at T = 298 K (Room temperature adsorption, RTA) or T = 77 K

(Low temperature adsorption, LTA). On this basis we identified the following

sequence of reaction steps as a particularly robust approach (Figure 8.2): during the

first cycle, approximately 200 mg of MgO nanocube powder is exposed to SiCl4 vapor

- with a vapor pressure of 300 mbar at room temperature - for 5 minutes and after

pumping subjected to O2 atmosphere (p = 700 mbar) for another period of 5 minutes.


132

Figure 8.2: Time-dependent development of SiCl4 and O2 partial pressures

describing the functionalization cycles employed for MgO nanocubes at T = 298 K or

at T = 77 K.

Gaseous reactants that did not adsorb on the particle surface are removed

upon subsequent pumping to a base pressure of p < 10-4 mbar and new reactant

molecules are supplied to the powder sample in the course of each subsequent

cycle. The results presented in this study were obtained on MgO nanocube powders

which were subjected to 6-cycles of room (RTA) or low temperature (LTA)

adsorption. After completion of the 6th cycle evacuation to the base pressure of

p < 10-4 mbar was completed at T = 298 K. Estimated surface coverage for CVS and

commercial MgO are listed in Table S8.1 of the Supporting Information.

For comparison, we performed experiments where SiCl4 was added to

dehydroxylated MgO nanocube powders (Figure 8.3 a) either at T = 77 K (LTA,

Figure 8.3 b) or at T = 298 K (RTA, Figure 8.3 c) and found that entirely different

reaction paths do occur dependent on the temperature of adsorption. At room

temperature, a strongly exothermic and poorly controllable process gives rise to a

sample heterogeneity which is visible to the eye (see Figure 8.3 c). Different powder

regions with particle agglomerates of varying quality and optical appearance are

formed as a result of an exothermic reaction. Parts of the originally fluffy MgO

nanocube powder (sample shown in Figure 8.3 a) becomes transformed into

compact white flakes in some regions of the sample (c2 in Figure 8.3). Both TEM

analysis (lower right panel in Figure 8.3) and powder X-ray diffraction (lower left

panel in Figure 8.3) reveal that only a fraction of the starting material becomes

converted into the substantially larger grains. These correspond to a so far unknown


133

intermediate which, within the time of the XRD measurement (< 5 hours), transforms

into a crystalline MgCl2·6H2O phase (Bischofite, c2 in Figure 8.3). The retained

second fraction inside the inhomogeneous powder agglomerates consists of MgO

nanocube with periclase structure (c1 in Figure 8.3).

Figure 8.3: Digital micrograph in the middle of the upper panel (a) shows a white and

opalescent MgO nanocube powder. The image on the left hand (b) shows a

homogeneous white and compact powder which results from MgO exposure to

SiCl4/O2 at T = 77 K (Figure 8.2). The inhomogeneous appearance of sample in the

right (c) originating from the same sample treatment at T = 298 K results from the

coexistence of two main fractions: c1) agglomerates of unreacted MgO nanocubes

agglomerates and c2) more compact and deeply white powder parts with a XRD

pattern specific to MgCl2·6H2O grains. The energy profile in the middle panel

underlines that adsorption at T = 77 K (b) inhibits the strongly exothermic reaction

that transforms MgO into Forsterite and further intermediates.


134

For specimen isolated from the more compact regions of the powder sample

(Figure 8.3 c2), phase analysis revealed magnesium dichlorhydrate MgCl2·6H2O

(Bischofite) as the predominant phase (69%) together with Forsterite (28%) and

periclase (3%) as minor fractions (Supporting Information, Figure S8.3). The sample

heterogeneity develops directly after SiCl4/ O2 adsorption and, thus, before the

reaction cell’s (Figure 8.1) vacuum was broken for sample transfer for materials

characterization. Water from the ambient, which is unavoidable in the present XRD

measurement set up, leads to the transformation of a so far unknown precursor

phase into magnesium dichlorhydrate (Figure 8.3 c2).

TEM images and XRD powder patterns related to fractions c1 and c2 (Figure

8.3) clearly reveal the difference between the starting material (MgO) and

MgCl2·6H2O originating from the strongly exothermic reaction between SiCl4 and

MgO, respectively.

On the other hand, earlier powder XRD and solid state 29Si MAS NMR studies

revealed that the LTA sample consists of MgO nanocubes which are covered with

surface adsorbed chlorosiloxane species SixClyOz.207 Despite their metastability and

high reactivity in the ambient (see below), the chemical composition as well the

composites’s nano- / microstructure can be retained over weeks by storage in

vacuum. Thus, for the here described functionalization approach, the temperature

during adsorption is the determining factor. In case of the room temperature

adsorption (RTA) experiment the thermal energy in conjunction with the heat of

reaction released allows the system to overcome the activation barrier towards

formation of forsterite and another, so far unidentified intermediate phase which in air

quickly transforms into MgCl2·6H2O (middle panel in Figure 8.3). Because of the

limited sample homogeneity of room temperature contacted MgO powder (RTA

Figure 8.3 c) we only continued with LTA samples on the investigation of reactivity

under ambient conditions.

We compared the reactivity of MgO nanocubes that were functionalized with

SixClyOz (LTA) towards water vapor with that of nanocubes with bare surfaces. A

sketch of the performed experiments together with the characteristic microstructures

obtained in the course of these experiments is provided in Figure 8.4. A closed


135

reaction system was used for keeping the different samples in water saturated air at

T = 298 K and for a specified time.

Figure 8.4: Schematic illustrating the H2O vapor phase induced transformation of

CVS MgO (a) into different microstructures, such as Mg(OH)2 nanosheet

agglomerates (b) or magnesium oxychloride fiber ensembles (c) (out of SixClyOz

functionalized MgO).

The MgO particles are arranged in particle agglomerates (Figure 8.4 a).

Imaging with Scanning Electron Microscopy (SEM) reveals on these agglomerates no

characteristic morphological features in the µm size range. Room temperature

exposure of bare MgO particle surfaces to water vapor produces thin lamellar

features (e.g. see arrow in the upper right panel of Figure 8.4) which, as will be

demonstrated below, correspond to Mg(OH)2 nanosheets. A completely new

microstructural situation, however, originates from water contact of chlorosiloxane


136

covered MgO nanocube agglomerates for exposure times t ≥ 3 days1 (lower panel in

Figure 8.4): assemblies of needle-like shaped crystals, which have grown radially out

of the individual agglomerates are observed. We also checked for related

microstructural transformation behavior on commercially available nanocrystalline low

surface area MgO material (SBET = 12 m²·g-1) which - prior to storage in water vapor

saturated air - had been subjected to functionalization cycles identical to those

applied to CVS MgO nanocubes (Supporting Information, Table S8.1). Analysis of the

XRD pattern (Supporting Information, Figure S8.4 a) and the SEM images

(Supporting Information, Figure S8.4 b) points to the exclusive formation of crystalline

Mg(OH)2 and the absence of fiber growth or additional crystallographic phases.

The microstructural changes observed for both pure and surface functionalized

MgO nanocubes (bottom of Figure 8.4) are in line with crystallographic structure

changes (Figure 8.5).

Figure 8.5: Left panel: (a) powder XRD pattern of a MgO powder after storage in

water-saturated air (p(H2O) = 32 mbar), (b) a chlorinated CVS MgO powder after

storage in vacuum (p < 10-4 mbar) and (c) a chlorinated CVS MgO powder after

storage in water-saturated air (p(H2O) = 32 mbar). The diffraction patterns are

consistent with a) the brucite phase Mg(OH)2; b) Periclase MgO and c) with the F5

1 Analysis of the XRD patterns reveal that chlorinated MgO nanocubes completely convert into

Mg3(OH)5·Cl∙4H2O fibers after 3 days storage in water-saturated air. TEM images show that water-

contacted samples consist of non-transformed MgO-related agglomerates in addition to fibers.

Absence of any XRD pattern that can be related to the observed non-transformed parts suggests their

low amount or their amorphous nature. However, TEM images reveal minor amount of non-

transformed agglomerates in samples stored for 14 days in water-saturated air compared to those

stored for 3 days.


137

phase of magnesium oxychloride (Mg3(OH)5·Cl∙4H2O). The right panel contains a

graphical representation of the corresponding crystal structure adapted from

reference [198].

Water vapor converts MgO nanocubes into Mg(OH)2 (a in left panel of Figure

8.5). However, the interaction between chlorosiloxane covered MgO nanocubes and

H2O vapor gives rise to the F5 phase Mg3(OH)5·Cl*4 H2O as concluded from the

XRD powder pattern ( Figure 8.5 c).207,215 A graphical representation of this structure

is provided in the right panel of Figure 8.5. It is important to note that that in the

course of all the experiments performed we have never obtained any evidence for

magnesium oxychloride carbonatation which would result from CO2 uptake from the

gas phase.215

For complementation of materials characterization we analyzed the LTA

samples at the different stages of their transformation with Transmission Electron

Microscopy (Figure 8.6).

Figure 8.6: Representative TEM images of a) chlorinated MgO nanocubes; b)

chlorinated MgO nanocubes after storage in vacuum (p < 10-4 mbar, 14 days) and c)

after 14 days of contact of chlorinated MgO nanocubes with water vapor (p(H2O)=32

mbar).


138

The granular contrast in Figure 8.6 a and b indicates a high degree of

crystallinity of the particles after low temperature contact with SiCl4/ O2 (Figure 8.2).

Displaying sharp edges and corners their cubic habitus is essentially retained.

Furthermore, consistent with the crystallite domain size, the particle sizes, as

determined by TEM analysis, remain in the range below 10 nm. Irrespective from the

retained primary particle properties, the agglomerates have become more compact

(upper micrographs of Figure 8.6 a and b) in comparison to MgO nanocube powders

before functionalization (Supporting Information, Figure S8.1).216,213

From microscopy and XRD data in conjunction with the NMR evidence for

siloxane formation on MgO surfaces 207 we infer that SiCl4/ O2 adsorption (Figure 8.2)

does not affect the primary particle properties but decreases their average distance

and by that increases the powder density. This explains the changes in the scattering

properties and, thus, the altered macroscopic optical appearance as revealed by the

digital images in Figure 8.3 b. The larger Mg3(OH)5·Cl*4H2O fibers (Figure 8.6) are

arranged as bundles (inset in Figure 8.6 c) with diameters in the range between 100

and 300 nm. After 14 days of H2O exposure we did not evidence residues of MgO

nanocubes which previously had served as Mg2+ ion source for magnesium

oxychloride fiber growth.

8.4.2 A Gas Phase Transformation Process with Thin Water Films as

Liquid Reaction Medium

Magnesium oxychloride fibers grow in aqueous MgO/ MgCl2 mixtures and give

rise to characteristic microstructures which are found in Sorel cement

phases.190,189,217,192 In the course of this study we identified a transformation process

that is based on the interaction between a water vapor saturated gas phase and a

highly dispersed metal oxide nanoparticle powder which was previously processed in

a water free environment. It is important to note at this point, that thin-film water is

ubiquitous in humid environments.218 Under such conditions nanomaterials become

instantaneously covered by water with film thicknesses ranging between one

molecular layer to few nanometers.219 These films provide a reaction medium220,221

with an essentially unexplored surface chemistry.219,218 The here described

transformation process as well as the unexpected reactivity of the chlorinated MgO


139

nanocube ensemble is attributed to the presence of water multilayers as a confined

liquid solvent medium for ions (Figure 8.7). Concentration gradients related to the

Mg2+, OH- and Cl- ions inside these water films enable ion transport along the

anisotropically growing structures. This sustains the growth process via diffusion and

material precipitation at the top part of the growing fibers.

Figure 8.7: Scheme illustrating the mechanism of fiber growth with [010] as the

growth direction and the (101) plane being subject to fast material addition and

growth. Contact with water vapor leads to the instantaneous formation of condensed

water films which sustain ion transport necessary to add new Mg3(OH)5Cl.4H2O units

to the top of the growing crystalline fibers.

We used SiCl4 as a water free chlorine ion source 214 and coadsorbed O2 at low

temperatures. Upon subsequent sample warming to room temperature under

dynamic vacuum conditions, this surface mixture transforms a fraction of activated

SiCl4/ O2 molecules into chlorosiloxane species that remain on the MgO nanocube

surfaces. Corresponding materials’ state is metastable and undergoes the

spontaneous transformation towards Mg3(OH)5Cl*4H2O fibers at room temperature

(Figure 8.4 and Figure 8.7). Comparison of the two types of nanocrystalline MgO

samples, the CVS MgO nanocube powder (Supporting Information, Figure S8.1 a

and b) and commercially available MgO (Supporting Information, Figure S8.1 c and

d), revealed the critical impact of the degree of dispersion on the transformation

process (Figure 8.4 and Figure 8.7). On CVS MgO, where reactive SiCl4/O2

adsorption (Figure 8.2) can occur, the estimated number of monolayer equivalents


140

corresponds to approximately 1 and is about 28 times higher for commercially

available nanocrystalline MgO (Supporting Information, Table S8.1). For the latter

type of material the dispersion is too low to enable effective intermixture between

SiCl4/ O2 derived adsorbates, on the one hand, and activating MgO surface sites, on

the other. As a consequence, SixClyOz production is negligible. The evacuation step

performed during warm up and in the last step of each cycle (Figure 8.2) removes all

physisorbed and weakly bound SiCl4/ O2 molecules. In the subsequent step,

exposure of the low surface area material towards H2O (Figure 8.4) leads to the

quantitative transformation of MgO into brucite (Supporting Information, Figure S8.4).

Thus, the concentration of surface activated chlorsiloxanes SixClyOz is insufficient to

engage the reaction path towards magnesium oxychloride needle formation. An

additional explanation relates to the complex reaction network which also requires

Mg-O dissolution to enrich the liquid water film with Mg2+ ions. Size dependent

dissolution effects 213 as well as microstructural parameters which hamper mass

transfer and eliminate the chance for magnesium oxychloride growth in case of the

commercially available starting material may play an additional role.

Silicon originating from SiCl4 is neither part of the crystalline product nor does it

contribute to any other crystalline phase as investigated by XRD measurements. We

expect that after fiber growth amorphous SiOx remains in the base region of the fiber

assemblies.

8.4.3 Vacuum Annealing Induced Decomposition of Oxychloride Fibers

The emergence of anisotropic magnesium oxychloride prompts the question

whether their thermally induced dehydration can be used to obtain nanocrystalline

MgO in more organized microstructures. For this reason, the annealing induced

decomposition of the Mg3(OH)5Cl*4H2O fibers was studied by powder X-ray

diffraction and electron microscopy. Vacuum annealing was applied to the

magnesium oxychloride fibers to temperatures above that of Mg(OH)2 decomposition,

i.e. at T = 673 K, T = 873 K and T = 1173 K.


141

Figure 8.8: Powder XRD patterns of a) Mg3(OH)5Cl*4H2O as starting material and of

the same material after 30 minutes of vacuum annealing up to the temperatures

specified b) T = 673 K, c) T = 873 K and d) T = 1173 K. The XRD pattern in b

contains diffraction peaks attributed to low hydrates of MgOHCl, and MgCl2*xH2O as

well as the signature which arises from the MgO periclase phase.222 Patterns c and d

reveal the predominant abundance of MgO periclase in addition to low intensity

peaks that are consistent with the Forsterite phase (magnesium silicate).

In the course of annealing to T = 673 K the F5 phase Mg3(OH)5Cl*4H2O

decomposes into periclase (MgO, 73 %) admixed to components containing lower

hydrates of MgOHCl or MgCl2*xH2O223 (Figure 8.8). The conversion into MgO seems

to be nearly completed after annealing to T = 873 K (Figure 8.8 c). From the width of

the MgO specific diffraction peaks we derived an average crystallite domain size of

dXRD = 35 nm which remains in the regime of nanocrystalline materials. Our results

are consistent with previous reports on the thermal decomposition of MgOHCl into

MgO and HCl gas at temperature above T = 750 K.223

MgOHCl MgO + HCl Equation 8.1

Figure 8.8 c reveals the additional presence of low intensity peaks related to

bischofite (10%) and to the Forsterite phase of magnesium silicate (Mg2SiO4, 17%).

The latter contribution indicates that residual silicon has become partially


142

reincorporated into the crystalline solid. The retransformation to MgO seems to be

completed after annealing at T = 1173 K i.e. at a temperature at which MgO particle

surfaces become dehydroxylated under high vacuum conditions.107 Representative

SEM images in Figure 8.9 shows the related microstructural changes.

Figure 8.9: SEM images of Mg3(OH)5Cl*4H2O samples isolated after vacuum

annealing at a) T = 673 K, b) T = 873 K, and c) T = 1173 K for 30 minutes in each

case.

Figure 8.9 a reveals the morphology changes of the Mg3(OH)5Cl*4H2O sample

after annealing at T = 673 K. The fiber bundles become more compact and at various

places we observed the detaching of thin fiber components.


143

Although the chemical constituents of the original fibers are partially

decomposed, these images show particles with needle- and ribbon-like morphology

which are different from those typically observed for MgO or Forsterite.213 From the

SEM images taken from the sample that was annealed at T = 873 K it becomes

apparent that the decomposition product is characterized by particulate segregates or

deposits around the original fiber (Figure 8.9 b). Moreover, the overall product

morphology of the sample which was previously annealed to T = 1173 K (Figure 8.9

c) strongly resembles the materials situation related to the sample after annealing to

T = 873 K. Some larger crystallites (> 100 nm) with hexagonal or less regular shapes

and with sharp edges can be identified as well (Figure 8.10).

Figure 8.10: TEM images of particulate products after thermal decomposition of

Mg3(OH)5Cl*4H2O via vacuum annealing to T =1173 K for 30 minutes.

The cartoon of Figure 8.11 summarizes the chemical transformation of MgO

nanocube powders into magnesium oxychloride fiber assemblies (1-3) followed by

their annealing induced pseudomorphic transformation into MgO nanocrystal based

fibers (3-4).


144

Figure 8.11: Schematic illustrating the organization of MgO nanoparticles into fibers

via the different stages of magnesium oxychloride fiber growth (1 → 3) and

subsequent vacuum annealing (3 → 4).

The structural and in particular the microstructural development represents an

entirely new way to generate MgO based nanostructures which are organized in

microstructures and exhibit a substantially enhanced level of spatial organization.

This opens an interesting promising route to generate nanocrystalline model systems

for surface chemistry studies that will contribute to our understanding of mechanical

property evolution in mineral binders. Moreover, we believe that such model systems

will be extremely useful in completely different fields such as the materials

development for adsorption/ desorption cycles and catalysis.

8.5 Conclusion

Agglomerated MgO nanoparticles can be functionalized with surface

oxychlorides via the gas phase using SiCl4 as chlorine source. This produces

metastable high energy materials which, as a result of H2O vapor exposure, undergo

spontaneous room temperature conversion into magnesium oxychloride fibers. Via

the detailed exploration of the process parameters that are critical for the formation of

magnesium oxychloride fibers we also demonstrate the great potential of CVS MgO

particles as a model system for the elucidation of growth of ceramic anisotropic

nanostructures, the associated heterogeneous chemistry and, ultimately, the

evolution of resulting microstructures. Links between materials’ dispersion and

functionalization temperature, on the one hand, and resulting hydroxide and

oxychloride structures, on the other, were established. An important aspect relates to

the localization of the microstructure evolution observed, i.e. the base of


145

Mg3(OH)5Cl*4H2O fiber growth. Spatially confined reaction media both in the µm

range (agglomerates of functionalized MgO nanocubes) as well as in the range of

few nanometers (thin liquid films as media for ion transport and reaction) were shown

to determine the transformation and microstructure evolution. Finally we addressed

the annealing induced re-transformation of Mg3(OH)5Cl*4H2O needles into spatially

organized metal oxide nano- and microstructures and demonstrate a novel way of

assembling metal oxide nanoparticles well-organized microstructures such as fibers.


146


8.6.1 Electron Microscopy

Figure S8.1: TEM images of CVS MgO (a) and commercial MgO (b) nanoparticles

as reference material. SEM images of CVS MgO (c) and commercial MgO (d)

nanoparticles. All images were taken after oxidation (T = 1123 K, p = 10 mbar O2)

and vacuum annealing (T = 1173 K, p < 5∙10-6 mbar).

Transmission Electron Microscopy clearly shows that CVS MgO is

characterized by agglomerates of highly dispersed monocrystalline nanocubes with a

high portion of edge and corner features (Figure S8.1 a). In contrast to MgO

nanocubes, commercial MgO can be characterized as an assembly of less regular

shaped particles (Figure S8.1 b).


147

8.6.2 Powder X-ray Diffraction

Figure S8.2: Powder XRD patterns of CVS MgO and commercial MgO nanoparticle

powders after oxidation (T = 1123 K, p = 10 mbar O2) and vacuum annealing

(T = 1173 K, p < 5 ·10-6 mbar). The diffractograms clearly reveal that both particle

types adopt crystal structures of the cubic phase. (Diffraction peaks indicating 2θ

positions for the periclase structure are indicated as bars.)

8.6.3 Estimated Surface Coverages of SiCl4/ O2 on Nanocrystalline MgO

Samples

Table S8.1: MgO sample weight, surface area and active sites available during the

exposure experiment. Pressure, number of molecules, monolayer equivalent (MLE)

and incident molecular flux of SiCl4 and O2 provided during one cycle.

*. According to a previous study the estimated number of active sites (corners and

edges) corresponds to roughly 3% of the total surface area available.

SiCl4/ cycle O2 / cycle

weight (mg)

surface area (m

2)

Est. number of active

sites* (m

-2)

pressure (mbar) at

RT

number of provided

molecules

MLE (%)

incident molec.

flux ( m

-2 s

-1)

pressure (mbar) at

RT

incident molecular

flux (m

-2 s

-1)

CVS MgO 200 60 1.8 300 3 × 1021

118 1·1027

700 6·1027

commercial MgO

200 2.5 << 1 300 3 × 1021

2832 1·1027

700 6·1027


148

Table S8.1 lists surface area of MgO powder samples, the estimated number of

active surface sites, number of molecules supplied at a given pressure, monolayer

equivalents (MLE, i.e. molecules provided to achieve a monolayer coverage at a

sticking coefficient of S = 1) and incident molecular flux of SiCl4 and O2 per cycle.

The gas properties are measured at T = 298 K considering the volume and other

conditions of gas reservoir before exposure.

8.6.4 XRD Phase Analysis of Reaction Products after contact of CVS

MgO with SiCl4 at RT

The quantitative phase analysis – with a good quality of refinement (e.s.d. of

wt% ~1 %) - is demonstrated in Figure S8.3.

Figure S8.3: XRD pattern and results of the phase analysis for the same powder

indicated in Figure 8.3 as fraction c2 and obtained after CVS MgO exposure to

SiCl4/O2 at room temperature.


149

8.6.5 Control Experiment with Commercial MgO

Figure S8.4: Commercial MgO after exposure to SiCl4/O2 and subsequent exposure

to water-saturated air (p(H2O) = 32 mbar): a) Powder XRD pattern revealing the

exclusive existence of brucite Mg(OH)2; b) SEM image.

9 Summary

151

9 Summary

The impact of different processing parameters in conjunction with nature and

composition of surrounding atmosphere on the metal oxide nanoparticle properties

optical absorption, photoluminescence emission, chemical composition, crystal

structure, morphology and size have been investigated. Several vapor phase grown

metal oxide nanoparticles and composite thereof have been synthesized and

explored using different experimental methods.

MgO and ZnO were synthesized by chemical vapor synthesis (CVS) and metal

organic-chemical vapor synthesis (MO-CVS) methods, respectively. An opposite

effect of O2 adsorption on the photoluminescence (PL) emission properties of these

two prototypical metal oxides has been observed. While molecular oxygen acts for

surface excited states of MgO as a PL quencher, it enhances the PL emission at

hvEm = 2.1 eV for ZnO nanoparticles. Stronger enhancement in PL emission was

found for as-synthesized ZnO nanoparticles as compared to the annealed ZnO

nanoparticles. Auger electron spectroscopic analysis reveals the existence of oxygen

interstitials in the surface near region of as-synthesized ZnO and the annihilation of

these defects by annealing. The enhancement in PL emission of as-synthesized

ZnO is explained by band bending effects on ionic semiconductors. For ZnO

nanoparticles gaseous oxygen affects the surface band bending by producing an

electric field near the surface that separates the electron-hole pair and promotes hole

transport to the surface. Subsequent hole trapping by oxygen interstitials as deep

trap states enforces their recombination with photogenerated electrons yielding

yellow PL light.

Particles of the ternary Zn-Mg-O system were prepared by the means of flame

spray pyrolysis (FSP) and subsequent vacuum annealing. Solid solution of Zn2+ in

periclase MgO is obtained for the nanoparticles with Zn2+ concentration of 10 at.%.

Higher Zn2+ concentrations of 30 at.% leads to mixtures where phase separation into

an MgO rich periclase and a ZnO rich wurtzite phase results from thermal treatment.

While gaseous oxygen partially reduces the PL emission of the hydroxylated

(as-synthesized) nanoparticles containing 10 at.% Zn2+, it entirely quenches the

emission in dehydroxylated particles after vacuum annealing. It is found that the PL

9 Summary

152

emission observed at hvEm = 2.1 eV for the as-synthesized Zn-Mg-O particles with

separated wurtzite phase is not quenched by gaseous oxygen but it is depleted after

vacuum annealing. Moreover, hydroxyls were found to act as protecting groups

against gaseous oxygen as a PL quencher at the solid−gas interface.

Using MO-CVS and subsequent annealing procedures in oxygen atmosphere,

Fe-Mg-O nanocomposite particles of different compositions (Fe/(Fe+Mg) = 1, 6 or

9 at.%) are synthesized. Fe-Mg-O nanoparticles that are annealed up to T = 873 K

are characterized as solid solutions of Fe3+ in periclase MgO which can reveal

superparamagnetic properties. Annealing of (1%) Fe-Mg-O samples up to T = 1173 K

leads to coarsening of the cubic particles that show superparamagnetic properties.

Annealing of (6% and 9%) Fe-Mg-O up to T = 1173 K leads to phase separated

particle mixtures containing periclase MgO and magnesioferrite MgFe2O4 phases that

are characterized as particles with antiferromagnetic behavior. The results point to

annealing induced crystal growth and phase separation which is facilitated by cation

vacancies emerging from the charge imbalance of Fe3+ ions in the Mg2+ lattice. In

addition, the optical absorption and PL emission properties of Fe-Mg-O particles are

investigated using UV-Vis diffuse reflectance and PL spectroscopy. The absorption

bands in the energy range 2.6 eV ≤ hv ≤ 5.6 eV are observed and attributed to

different ligand field and ligand-to-metal charge transfer electronic transitions. It is

found that the admixture of Fe3+ to MgO quenches the MgO specific PL emission

originating from surface excitons. This points to annealing induced surface

segregation of Fe3+ ions in Fe-Mg-O nanocomposites.

In the last part of this work, MgO is exposed to SiCl4 and O2 via the vapor phase

in a cyclic process. It is found that different reaction paths can occur depending on

adsorption temperature. While at room temperature adsorption (T = 298 K) of SiCl4 a

strong exothermic process leads to sample heterogeneity, at low temperature

(T = 77 K) adsorption homogeneous surface coverage of the MgO particle

ensembles with SiCl4 occurs. The low temperature adsorption leads to MgO

nanocubes which are covered with surface adsorbed chlorosiloxane SixClyOz species

as revealed by microscopy and XRD data in conjunction with the NMR studies. The

obtained MgO-SixClyOz material system is highly metastable and transforms

spontaneously into magnesium oxychloride Mg3(OH)5Cl∙4H2O needle-like structures

9 Summary

153

upon contact with water vapor. The absence of needle formation in the case of low

surface area MgO reveals the importance of high materials dispersion for the here

reported spontaneous phase transformation. This study also suggests that

condensed water layer on the growing fibers provides an ion transport and reaction

medium for their directional growth. Moreover, it is demonstrated that annealing

retransforms magnesium oxychlorides into the parental magnesium oxide particles

assembled in well-ordered fiber-like microstructures.

The here presented results have shown several annealing induced property

changes in metastable metal oxide nanoparticles. On the other hand it also became

clear that the chemical composition of the surrounding continuous phase strongly

influences the properties of the vapor phase grown metal oxide nanoparticles during

post-synthesis processing, aging and, ultimately, measurement. Moreover, aside

from the progress in the functionalization of MgO nanocubes with oxychlorides, the

growth of magnesium oxychloride Mg3(OH)5Cl∙4H2O fibers via water vapor contact

introduces a new approach for the fabrication of oxide and hydroxide fibers in the

ambient. In general, the results underline the necessity of considering sample history

and nature of the surrounding atmosphere in characterization of the metal oxide

nanoparticles. Related insights advance the understanding of the process-induced

property changes that give rise to differences in the functional performance of

identical materials that are treated differently or to formation of completely new type

of nanomaterials.

10 Appendix I: Impact of Annealing Processes on MgO Nanoparticles

155

10 Appendix I: Impact of Annealing Processes on MgO

Nanoparticles Size Distribution and Morphology

This chapter will discuss the impact of post-synthesis annealing processes on

particle morphology and size distribution of MgO nanoparticle samples produced by

chemical vapor synthesis (CVS). The synthesis approach is already described in

detail in section 3.1.1.1. The motivation for these studies comes from the unexpected

crystal growth which has been observed for MgO nanoparticles annealed in the

presence of residual gas atmosphere and high pressure of oxygen (pO2 = 650 mbar),

compared to those annealed under vacuum condition. This chapter is divided into

three parts. First, the structural and morphological characterization results which

show the differences in MgO nanoparticle samples annealed by two different

annealing programs will be discussed. Second, mass spectroscopic analysis done on

MgO samples annealed under vacuum condition will be shown at different evacuation

and annealing steps. Third, the crystal growth trend will be presented which is

investigated by considering different dwell times used in annealing programs.

10.1 MgO nanoparticle changes influenced by annealing conditions

Throughout this work it has been seen that vapor phase grown MgO

nanoparticle properties like particle size and morphology are influenced by annealing

conditions. In this context annealing condition refers to the presence or absence of

residual gases namely H2O, O2 and COx. Two employed annealing programs are:

i) annealing in vacuum condition named standard program and ii) annealing in the

presence of oxygen and residual gases named program number 1 (P1).

10.1.1 Standard Annealing

In this typical standard annealing program the cell containing as-synthesized

powder sample was first evacuated (p < 10-5 mbar) at room temperature and then

heated to T = 1173 K in 100 K-steps according to the program described in Table

10.1 and Figure 10.1.


156

Table 10.1: Standard vacuum annealing program used for the activation of

CVS-grown MgO.

temperature (K) rate (K /min) pressure (mbar) dwell time† (min)

373 5 < 1.0 · 10-5 * 0

473 5 < 1.0 · 10-5 * 0

573 5 < 1.0 · 10-5 * 0

673 10 < 1.0 · 10-5 * 0

773 10 < 1.0 · 10-5 * 0

873 10 < 1.0 · 10-5 * 0

973 10 < 1.0 · 10-5 * 0

1073 10 < 1.0 · 10-5 * 0

1173 10 < 1.0 · 10-5 * 0

1123 30 O2 pressure = 10 10

1123 30 < 1.0 · 10-5 * 0

1173 30 < 5.0 · 10-6 60

*. Base pressure; before increasing to the next temperature, base pressure must be

reached. †. Dwell time is the programed time that system remains at each

temperature after reaching base pressure.

Just shortly after reaching T = 1173 K sample was cooled down to T = 1123 K

and contacted by 10 mbar of oxygen for 10 minutes to remove organic contaminants.

After evacuation of the oxygen, temperature was again increased to T = 1173 K and

it was kept for 60 minutes at pressures less than 5·10-6 mbar. This thermal treatment

takes 6 to 8 hours (depending on vacuum pumps used and amount of annealed

MgO) and offers perfect conditions to have a clean metal oxide surface as proved by

FT-IR spectroscopy.107


157

Figure 10.1: Temperature profile (black line, left ordinate scale) and pressure profile

(blue line, right ordinate scale) during standard annealing. At T = 1123 K the samples

were oxidized with 10 mbar O2 for 10 minutes.

10.1.2 P1 Annealing

In this annealing method, outlined in Figure 10.2, at the beginning the cell

containing MgO nanoparticles powder was evacuated (p < 10-5 mbar) at room

temperature. The respective sample was then heated to T= 373 K at a rate of

2.5 K min-1, held at this temperature for 15 min (dwell time, td) and then was again

evacuated to p < 10-5 mbar. Further annealing steps are carried out at p = 650 mbar

of oxygen. The powder was stepwise heated in oxygen atmosphere to T = 473 K

(r = 5 K min-1, td = 15 min), T = 673 K (r = 10 K min-1, td = 30 min), T = 873 K (r = 10 K

min-1, td = 180 min) and T = 1173 K (r = 10 K min-1, td = 60 min).


158


pressure (dashed line, right ordinate scale) during P1 annealing procedure.

(r: heating rate; td: dwell time). Cooling times are not shown.

After each annealing step the sample was cooled down to room temperature

(cooling time ≈ 30 min) followed by an evacuation (≈ 15 min) to base pressure of


10.1.3 Results

10.1.3.1 Scanning Electron Microscopy

SEM images of the MgO nanoparticle powders after standard (Figure 10.3 a)

and P1 (Figure 10.3 b) annealing programs show visible differences in agglomerate

appearance. The SEM images of standard annealed sample correspond to fluffy and

loosely bond powders. The SEM images of P1 annealed sample reveal more

compact powders consisting of particulate objects in the size range of tens of

nanometers.


159

Figure 10.3: SEM images of MgO nanoparticle powder samples a) after standard

annealing; b) after P1 annealing. The SEM images show powders of sample after P1

annealing are coarser grained compared to samples after standard annealing.

10.1.3.2 Transmission Electron Microscopy

The TEM images of the MgO nanoparticle powder samples annealed by

standard method show high dispersion (Figure 10.4 a and b) with narrow particle size

distribution and average particle size of about 4 nm (Figure 10.5 a). The small

particle sizes as well as absence of sintering and solid-solid interfaces in this sample

lead to strong charging effects during transmission electron analysis. Movement of

particle agglomerates under electron beam made high resolution imaging very

difficult. Although the shape of the particles at this low resolution is not completely

clear but mostly particles with cubic morphology can be seen. Inspection of TEM

images of P1 annealed MgO sample (Figure 10.4 c and d) reveals that in spite of

significant increase in average particle size to about 18 nm, the particle size

distribution remains narrow (Figure 10.5 b). The P1 annealed MgO samples reveal

more compact powders with interconnected cubic particles compared to standard

annealed samples.


160

Figure 10.4: TEM images of MgO nanoparticle powder samples a and b) after

standard annealing; c and d) after P1 annealing. The TEM images show powders of

samples after P1 annealing have interconnected cubic particles and are more

compact compared to samples after standard annealing.

Figure 10.5: Cumulative particle size distribution plots for MgO nanoparticle powder

samples a) standard annealed; b) P1 annealed. The distribution is obtained from the

edge length of particles at different regions of each sample based on TEM images

(about 300 particles measured for each distribution).


161

10.1.3.3 X-ray Diffraction

X-ray diffraction (XRD) patterns of MgO nanocubes confirm that all samples

possess periclase cubic structure (JCPDS card # 45-0946) irrespective of annealing

method applied (Figure 10.6). The XRD pattern of the P1 annealed sample reveal

narrowing of the periclase phase diffraction features widths, which points to volume

increase of the coherently scattering solid and, therefore, to particle coarsening by

annealing in oxygen. This is in good agreement with particle growth deduced from

TEM images. Using Scherrer equation and from the full width at half maximum

(FWHM) of the MgO diffraction peaks the average crystalline domain sizes can be

calculated as 5 ± 1 nm and 16 ± 1 nm for standard and P1 annealed samples,

respectively.

Figure 10.6: X-ray diffraction patterns of MgO nanoparticle powder samples

a) standard annealed; b) P1 annealed. Vertical lines correspond to the standard XRD

pattern of periclase MgO (JCPDS card # 45-0946).

10.2 Mass Spectroscopy during Standard Annealing of MgO

Mass spectrometer was used to qualitatively track the composition of the

residual gas atmosphere and desorbed species (especially H2O) during cell

evacuation and annealing of MgO. For this purpose 0.2 g of MgO nanoparticle

powder produced by chemical vapor synthesis was evacuated and standard

annealed using the setup shown in Figure 10.7.


162

Figure 10.7: Schematic diagram of the vacuum annealing setup equipped with mass

spectrometer.

The experimental steps are as below:

1) Prior to evacuation, the as synthesized MgO powder sample was kept in air

for 1 hour to allow complete water adsorption from humidity

2) The sample was slowly evacuated by gradually opening the valve (sample

valve) between sample glass cell and vacuum chamber while only the rotary

vacuum pump was connected. The opening of the valve took almost 50

minutes. During this time mass spectrometer was working and the metering

valve was half opened at a fixed value of 500. This amount of opening in

metering valve guarantees the pressure less than 10-4 mbar which is needed

for the safe operation of Faraday detector in mass spectrometer analyzer.

3) After completely opening of the sample valve, turbo molecular pump was

connected.

4) The sample was annealed stepwise following the standard annealing

program until reaching to p < 5·10-6 mbar at T = 873 K. The sample was

then cooled down overnight.

5) The day after the sample was directly annealed to T = 973 K, T = 1073 K

and T = 1173 K until reaching to p < 5·10-6 at T = 1173 K when it was cooled

down afterwards.


163

10.2.1 Mass Spectroscopy during Initial Evacuation

Mass spectroscopy during sample valve opening at room temperature

(T = 297 K) shows the trend view of the relative pressure (or relative abundance)

changes of desorbed species (Figure 10.8). Figure 10.8 also contains the changes of

the total pressure in mass spectrometer chamber measured by cold cathode gauge.

The trend graphs as well as mass spectra at different times reveal that at the

beginning the majority of detected species consist of N2 and O2 (Figure 10.9 a)

whereas at the last minutes mainly OH and H2O are detected (Figure 10.9 b). It

should be mentioned that very small amounts (i.e. low partial pressure as recorded

by MS) of other masses which may are the fragment ions (e.g. CO2 fragments) are

also recorded by mass spectrometer.

Figure 10.8: Trend graph of relative pressure changes over time of the sample valve

opening. Partial pressures (link ordinate scale) are recorded by mass spectrometer

and total pressure (right ordinate scale) is measured by cold cathode pressure gauge

installed in the MS chamber. The fluctuation (peaks) of the curves arises from

stepwise opening of the sample valve. The related mass spectra of the time slots

indicated by vertical dashed lines are shown in Figure 10.9.


164

Figure 10.9: Mass spectra related to time slots indicated in Figure 10.8 by vertical

dashed lines. a) at t = 42.18 min; b) at t = 50.62 min.

10.2.2 Mass Spectroscopy during Vacuum Annealing

The temperature and vacuum chamber pressure profiles during vacuum

annealing are illustrated in Figure 10.10. The figure indicates how the total pressure

in the vacuum chamber increases due to heating induced desorption of molecules as

well as the time necessary to reach again a pressure less then p<10-5 mbar. It is

observed that at around T = 573 K the total pressure reaches the maximum value.

This temperature relates to maximum water and hydroxyl desorption as revealed

from mass spectra. It is further found out that the dwell times are highly dependent on

the vacuum capacity of the pumps used.


165

Figure 10.10: Temperature (black line, left ordinate scale) and vacuum chamber

pressure (blue line, right ordinate scale) profiles during standard annealing. Unlike

standard annealing there is no oxidation step at T= 1123 K. The related mass spectra

of the time slots indicated by vertical dashed lines are shown in Figure 10.11.

To qualitatively compare the main desorbed species during the standard

annealing, for some selected time slots, (four vertical dashed lines indicated in Figure

10.10) the mass spectra recorded by mass spectrometer are shown in Figure 10.11.


166

Figure 10.11: Mass spectra of the desorbed species during standard annealing

related to the time slots indicated in the Figure 10.10.

10.3 Annealing Dwell Time Effect

In this part the influence of different annealing dwell times on MgO crystal

growth and particle morphology is described. The effect of different dwell times at

T = 1173 K is discussed for annealing in vacuum condition as well as annealing in

the presence of oxygen and residual gases. Changes in crystal size are investigated


167

by analysis of XRD pattern peak broadening and particle size distributions derived

from TEM micrographs. Changes in morphology are also detected with TEM

micrographs.

10.3.1 Description of applied annealing programs

The vacuum annealing employed corresponds to standard annealing program

which is described in section10.1.1 but extended to variable dwell times at

T = 1173 K (Figure 10.12). 3 dwell times used for annealing of the sample at T =

1173 K are 1h, 7h and 10h. Due to experimental reasons the sample is cooled down

to room temperature over night after reaching T = 1173 K for the first time.

Figure 10.12: Temperature profile (black line, left ordinate scale) and pressure profile

(blue line, right ordinate scale) during vacuum annealing. At T = 1123 K the samples

were oxidized with 10 mbar O2 for 10 minutes. At T = 1173 K the MgO samples were

annealed with different dwell times of 1h, 7h and 10h to investigate crystal shape

and domain size changes. td = dwell time.

An annealing program named P2 has been employed to study the influence of

variable dwell times in the presence of oxygen and residual gases. 3 dwell times


168

used for P2 annealing of the sample at T = 1173 K are 1h, 7h and 10h. P2 annealing

program is described in Figure 10.13 and carried out as follows: First at room

temperature the sample is evacuated to a pressure of p < 10-5 mbar to remove

surrounding gas atmosphere. Mass spectroscopy results (Figure 10.9 and Figure

10.11) suggest that at this stage the sample`s surface remains covered with

adsorbed water molecules. After initial evacuation a constant oxygen pressure (p =

650 mbar) is applied and sample is heated to T = 873 K (r = 10 K∙min-1) and kept at

this temperature for 30 minutes. Subsequently, the sample is cooled down to room

temperature and the supernatant gases are removed by evacuating to p < 10-5 mbar.

Prior to the next heating step to T = 1173 K (r = 10 K∙min-1) replacement of residual

gas atmosphere with fresh oxygen (p = 650 mbar) is performed. At T = 1173 K

identical dwell times as vacuum annealing are applied. Finally the sample is cooled

down to room temperature and annealing program is finished.

Figure 10.13: Temperature profile (black line, left ordinate scale) and applied oxygen

pressure (blue line, right ordinate scale) during P2 annealing. At T = 1173 K the MgO

samples were annealed with different dwell times of 1h, 7h and 10h. td = dwell time.


169

10.3.2 Results

The effect of P2 annealing in the presence of oxygen and residual gases on

MgO particle size and morphology in comparison to standard annealing in vacuum

condition was studied. The analysis was done by means of TEM micrographs and x-

ray diffraction and can be summarized as follows:

Figure 10.14: TEM images of the MgO powder samples after a) 1h, b) 7h and c) 10h

standard annealing in vacuum at T = 1173 K.

TEM micrographs reveal that long dwell times up to 10h at T = 1173 K does not

influence the cubic shape of MgO nanoparticles by standard annealing (Figure

10.14). Compared to standard annealing, P2 annealing in the presence of oxygen


170

and residual gases lead to dramatic change in the shape of particles for all dwell

times used (Figure 10.15).

Figure 10.15: TEM images of the MgO powder samples after a) 1h, b) 7h and c) 10h

P2 annealing at T = 1173 K. The annealing is performed in the presence of oxygen

and residual gases.

In the case of P2 annealing all particles are more or less rounded in shape and

almost no cubes were observed (Figure 10.15). Moreover, high amount of sintering

can be seen between single crystals. The TEM images reveal steps and terraces on

the particle surfaces.

On the basis of the TEM micrographs, particle size distributions (PSD) were

derived (Figure 10.16 left). For MgO nanoparticles annealed with dwell times of 1h,

7h and 10h, standard annealing leads to almost identical PSDs having median


171

values of around 4 to 5 nm. In the case of P2 annealed nanoparticles, the PSDs are

broadened and shifted to bigger sizes with median values in the range of 20 nm

(td = 1h) and 25 nm (td = 7h and 10h). For P2 annealing, increase of dwell time from

1h to 7h leads to more particle growth, compared to standard annealing (Figure

10.16).

Figure 10.16: Left panel: Cumulative particle size distributions (PSD) of the annealed

MgO samples showing particle growth in the case of P2 annealing. Right panel:

Median values of the cumulative PSDs in dependence of dwell time at T = 1173 K.

Colored areas represent the borders of the respective distributions, showing a

significant broadening in case of P2 annealing.

XRD patterns for the respective samples confirm the trend in crystal growth

which has been seen from the results of particles size distributions. A significant

narrowing of the reflexes in case of P2 annealing is observed, implying an increase in

crystallite domain size. The XRD results are depicted in Figure 10.17. Using Scherrer

equation and from the full width at half maximum of the most pronounced reflexes

(related to 200 and 220 planes), an increase in crystallite domain size for the P2

annealed samples can be observed (Figure 10.18). Furthermore, with increasing the

dwell time at T = 1173 K, the crystallite sizes of P2 annealed samples seem to grow,

whereas no significant crystal growth can be seen for the standard annealed

samples.


172

Figure 10.17: X-ray diffraction (XRD) patterns of MgO powder samples a) standard

annealed; b) P2 annealed. The patterns attribute to the samples annealed with

different dwell times of 1h, 7h and 10h at T = 1173 K. Vertical lines correspond to the

standard XRD pattern of periclase MgO (JCPDS card # 45-0946).

Figure 10.18: Crystallite dimensions x200 and x220 of standard vacuum annealed and

P2 annealed samples with different dwell times of 1h, 7h and 10h at T = 1173 K.

10.4 Summary and Conclusion

From the results discussed in this appendix, it became clear that annealing in

the presence of oxygen and residual gases has a big influence on the particle size

and morphology of the vapor phase grown MgO samples. The MgO particles can

grow to sizes in the range of 20 nm by using P1 (Figure 10.2) or P2 (Figure 10.13)

annealing programs. Unlike P1 and standard annealing methods which lead to


173

particles with cubic shape (Figure 10.4 and Figure 10.14), P2 annealing gives rise to

particles with non-cubic shape (Figure 10.15).

The reason for different effects of these annealing programs can be found in

differences in process conditions. As described in Figure 10.1 standard annealing is

carried out under dynamic vacuum condition which means that all desorbing species,

especially water, are pumped away instantly. Influences of the residual gases are

therefore negligible. In P1 annealing the first heating step is performed under the

same condition as standard annealing. This means that P1 annealing up to T = 373 K

is done under dynamic vacuum followed by evacuation to p < 10-5 mbar. Moreover,

before each heating step in P1 annealing, residual gases are replaced by fresh

oxygen. This leads to an efficient removal of adsorbed water species from the particle

surfaces, the same as standard annealing. In contrast, P2 annealing procedure

provides the condition for only one residual gas replacement step which is after

T = 873 K. Therefore in P1 annealing up to T = 873 K the presence of residual water

must be considered.

The results suggest that annealing in the absence of water allows the (100)

surfaces become the most stable MgO surfaces whereas annealing in the presence

of adsorbed water destabilizes the (100) planes and allows other surfaces like (111)

become stabilized. As a result, MgO particles with cubic shape are obtained by

annealing in the water free condition and particles with non-cubic shape are obtained

by annealing in the presence of residual water. More detailed investigations are

required in order to gain in-depth understanding of the residual water induced

changes of MgO nanocubes during annealing. Such insight will improve the control

over particle size and morphology which determine the chemical and physical

properties of MgO nanoparticles like their surface activity and optical properties.

11 Bibliography

175

11 Bibliography

[1] International organization for standardization. Terms describing categories of nanostructured material. ISO/TS 80004-4:2011(en). Available at www.iso.org (Last seen: October 2015).

[2] Oksam, G. Metal oxide nanoparticles: synthesis, characterization and application, Journal of Sol-Gel Science and Technology 37, 161-164 (2006).

[3] Hahn, Y.-B. & Umar, A. Metal oxide nanostructures and their applications, American Scientific Publishers (2010).

[4] Pacchioni, G. & Freund, H. Electron transfer at oxide surfaces. The MgO paradigm: From defects to ultrathin films, Chemical Reviews 113, 4035–4072 (2012).

[5] Jupille, J. & Thornton, G. Defects at oxide surfaces, Springer International Publishing (2015).

[6] Wang, Z. G. Zu, X. T. Zhu, S. & Wang, L. M. Green luminescence originates from surface defects in ZnO nanoparticles, Physica E: Low-dimensional systems and nanostructures 35, 199–202 (2006).

[7] Faure, B. et al. Dispersion and surface functionalization of oxide nanoparticles for transparent photocatalytic and UV-protecting coatings and sunscreens, Science and Technology of Advanced Materials 14, 23001 (2013).

[8] Kango, S. et al. Surface modification of inorganic nanoparticles for development of organic–inorganic nanocomposites, Topical Issue on Polymer Hybrids 38, 1232–1261 (2013).

[9] Reed, J. & Ceder, G. Role of electronic structure in the susceptibility of metastable transition-metal oxide structures to transformation, Chemical Reviews 104, 4513–4534 (2004).

[10] Scaramuzza, S. Agnoli, S. & Amendola, V. Metastable alloy nanoparticles, metal-oxide nanocrescents and nanoshells generated by laser ablation in liquid solution: influence of the chemical environment on structure and composition, Physical Chemistry Chemical Physics (2015).

[11] Berger, T. Schuh, J. Sterrer, M. Diwald, O. & Knözinger, E. Lithium ion induced surface reactivity changes on MgO nanoparticles, Journal of Catalysis 247, 61–67 (2007).

[12] Stankic, S. Sternig, A. Finocchi, F. Bernardi, J. & Diwald, O. Zinc oxide scaffolds on MgO nanocubes, Nanotechnology 21 (2010).

[13] Berger, T. & Diwald, O. Defects in metal oxide nanoparticle powders, in defects at oxide surfaces by J. Jupille and G. Thornton, Springer International Publishing (2015).

[14] Mowbray, D. J. et al. Trends in metal oxide stability for nanorods, nanotubes, and surfaces, Journal of Physical Chemistry C 115, 2244–2252 (2011).

11 Bibliography

176

[15] Leite, E. R. et al. Development of metal oxide nanoparticles with high stability against particle growth using a metastable solid solution, Advanced Materials 14, 905 (2002).

[16] Riss, A. Elser, M. J. Bernardi, J. & Diwald, O. Stability and photoelectronic properties of layered titanate nanostructures, Journal of the American Chemical Society 131, 6198–6206 (2009).

[17] Vayssieres, L. On the thermodynamic stability of metal oxide nanoparticles in aqueous solutions, International Journal of Nanotechnology 2, 411–439 (2005).

[18] Zhu, J. et al. Two-step synthesis of Fe2O3 and Co3O4 nanoparticles. Towards a general method for synthesizing nanocrystalline metal oxides with high surface area and thermal stability, RSC Advances 2, 121–124 (2012).

[19] Korotcenkov, G. & Cho, B. K. Instability of metal oxide-based conductometric gas sensors and approaches to stability improvement, Sensors and Actuators B: Chemical 156, 527–538 (2011).

[20] Hornyak, G. L. Introduction to nanoscience & nanotechnology, CRC Press (2009).

[21] Jiang, Q. & Yang, C. C. Size effect on the phase stability of nanostructures, Current Nanoscience 4, 179-200 (2008).

[22] Wilde, G. Nanostructured Materials. 1st ed., Elsevier (2009).

[23] McKenna, K. P. Sushko, P. V. & Shluger, A. L. Inside powders: A theoretical model of interfaces between MgO nanocrystallites, Journal of the American Chememical Society 129, 8600–8608 (2007).

[24] Sushko, P. V. Gavartin, J. L. & Shluger, A. L. Electronic properties of structural defects at the MgO(001) surface, Journal of Physical Chemistry B 106, 2269-2276 (2002).

[25] Mueller, M. et al. Effect of protons on the optical properties of oxide nanostructures, Journal of the American Chememical Society 129, 12491–12496 (2007).

[26] Spoto, G. et al. Carbon monoxide MgO from dispersed solids to single crystals: A review and new advances, Progress in Surface Science 76, 71–146 (2004).

[27] Cornu, D. et al. Influence of natural adsorbates of magnesium oxide on its reactivity in basic catalysis, Physical Chemistry Chemical Physics 15, 19870–19878 (2013).

[28] Stankic, S. et al. Novel optical surface properties of Ca2+-doped MgO nanocrystals, Nano Letters 5, 1889–1893 (2005).

[29] A L Shluger et al. Self-trapping holes and excitons in the bulk and on the (100) surfaces of MgO, Journal of Physics: Condensed Matter 3, 8027 (1991).

[30] Shluger, A. L. Sushko, P. V. Kantorovich, L. N. Spectroscopy of low-coordinated surface sites: Theoretical study of MgO, Physical Review B 59, 2417 (1999).

11 Bibliography

177

[31] Stankic, S. et al. Size-dependent optical properties of MgO nanocubes, Angewandte Chemie International Edition 44, 4917–4920 (2005).

[32] Sternig, A. et al. Photoluminescent nanoparticle surfaces: The potential of alkaline earth oxides for optical applications, Advanced Materials 20, 4840–4844 (2008).

[33] Heifets, E. Eglitis, R. Kotomin, E. Maier, J. & Borstel, G. Ab initio modeling of surface structure for SrTiO3 perovskite crystals, Physical Review B 64, 235417 (2001).

[34] Rodriguez, J. A. Electronic and chemical properties of mixed-metal oxides: basic principles for the design of DeNOx and DeSOx catalysts, Metallic Oxides: filling the gap between catalysis and surface science 85, 177–192 (2003).

[35] Lin, S. S. et al. Phosphorus doped Zn1-xMgxO nanowire arrays, Nano Letters 9, 3877–3882 (2009).

[36] Gaskov, A. & Rumyantseva, M. In sensors for environment, health and security, edited by M.-I. Baraton, Springer, 3-30 (2009).

[37] Sun, M. Liu, H. Liu, Y. Qu, J. & Li, J. Graphene-based transition metal oxide nanocomposites for the oxygen reduction reaction, Nanoscale (2014).

[38] Ingler Jr, W. B. & Khan, S. U. M. Photoresponse of spray pyrolytically synthesized magnesium-doped iron (III) oxide (p-Fe2O3) thin films under solar simulated light illumination, Thin Solid Films 461, 301–308 (2004).

[39] Orna, J. et al. FeO/MgO/Fe heteroepitaxial structures for magnetic tunnel junctions, IEEE Transactions 44, 2862–2864 (2008).

[40] Rakmak, N. Wiyaratn, W. Bunyakan, C. & Chungsiriporn, J. Synthesis of Fe/MgO nano-crystal catalysts by sol–gel method for hydrogen sulfide removal, Chemical Engineering Journal 162, 84–90 (2010).

[41] Bhide, V. G. & Tambe, B. R. Investigation of the MgO∶Fe system using the mössbauer effect, Journal of Materials Science 4, 955-961 (1969).

[42] Groves, G. W. & FINE, M. E. Solid solution and precipitation hardening in

Mg‐Fe‐O alloys, Journal of Applied Physics 35, 3587–3593 (1964).

[43] Myrach, P. et al. Temperature-dependent morphology, magnetic and optical properties of Li-doped MgO, ChemCatChem 2, 854–862 (2010).

[44] Lee, W. Han, J. W. Chen, Y. Cai, Z. & Yildiz, B. Cation size mismatch and charge interactions drive dopant segregation at the surfaces of manganite perovskites, Journal of the American Chemical Society 135, 7909–7925 (2013).

[45] Battaile, C. C. Najafabadi, R. & Srolovitz, D. J. Simulation of segregation to free surfaces in cubic oxides, Journal of the American Ceramic Society 78, 3195–3200 (1995).

[46] Wynblatt, P. Rohrer, G. S. & Papillon, F. Grain boundary segregation in oxide ceramics, Ceramic and Metal Interfaces 23, 2841–2848 (2003).

[47] Noguera, C. Physics and chemistry at oxide surfaces, Cambridge University Press (1996).

11 Bibliography

178

[48] Skvortsova V. Trinkler, L. Transition metal ions luminescence in neutron irradiated magnesium oxide, IOP Conference Series: Materials Science and Engineering 15, 012055 (2010).

[49] Özgür, Ü. et al. A comprehensive review of ZnO materials and devices, Journal of Applied Physics 98, 41301 (2005).

[50] Takeuchi, I. et al. Monolithic multichannel ultraviolet detector arrays and continuous phase evolution in MgxZn1−xO composition spreads, Journal of Applied Physics 94, 7336–7340 (2003).

[51] Diler, E. Rioual, S. Lescop, B. Thierry, D. & Rouvellou, B. Stability of ZnMgO oxide in a weak alkaline solution, Thin Solid Films 520, 2819–2823 (2012).

[52] Maemoto, T. and Ichiba, N . and Ishii, H. and Sasa, S. and Inoue, M. Structural and optical properties of ZnMgO thin films grown by pulsed laser deposition using ZnO-MgO multiple targets, Journal of Physics: Conference Series 59, 670 (2007).

[53] Bates, C. H. White, W. B. & Roy, R. New high-pressure polymorph of zinc oxide, Science 137, 993 (1962).

[54] Pan, X. et al. Fabrication of p-type ZnMgO films via pulsed laser deposition method by using Li as dopant source, Applied Surface Science 253, 6060–6062 (2007).

[55] Eijt, S. W. H. Roode, J. de, Schut, H. Kooi, B. J. & Hosson, J. T. M. de. Formation and stability of rocksalt ZnO nanocrystals in MgO, Applied Physics Letters 91 (2007).

[56] Lu, G. Lieberwirth, I. & Wegner, G. A General polymer-based process to prepare mixed metal oxides: The Case of Zn1-xMgxO nanoparticles, Journal of the American Chemical Society 128, 15445–15450 (2006).

[57] Chen, J. Shen, W. Z. Chen, N. B. Qiu, D. J. and Wu, H. Z. The study of composition non-uniformity in ternary MgxZn1− xO thin films, Journal of Physics: Condensed Matter 15, L475 (2003).

[58] Biasi, R. S. de, Cardoso, L. H. G. Campos, J. B. de, Sanchez, D. R. & Cunha, J. B. M. d. Influence of annealing on magnesioferrite nanoparticles synthesized by a sol-gel/combustion method, Materials Research 12, 225–227 (2009).

[59] Candeia, R. A. et al. MgFe2O4 pigment obtained at low temperature, Materials Research Bulletin 41, 183–190 (2006).

[60] Rodriguez, J. A. et al. Reaction of SO2 with pure and metal-doped MgO: Basic principles for the cleavage of S–O bonds, The Journal of Chemical Physics 115, 10914–10926 (2001).

[61] Palizdar, M. Ahgababazadeh, R. Mirhabibi, A. Brydson, R. & Pilehvari, S. Investigation of Fe/MgO catalyst support precursors for the chemical vapour deposition growth of carbon nanotubes, Journal of Nanoscience and Nanotechnology 11, 5345–5351 (2011).

[62] Yager, T. A. & Klngery, W. D. The kinetics of clustering reactions in iron-doped MgO, Journal of Materials Science 16, 483-488 (1981).

11 Bibliography

179

[63] Hansson, R. Hayes, P. C. & Jak, E. Phase equilibria at sub-solidus conditions in the Fe-Mg-Zn-O system in air, Scandinavian Journal of Metallurgy 33, 355–361 (2004).

[64] Coquay, P. Grave, E. de, Peigney, A. Vandenberghe, R. E. & Laurent, C. Carbon nanotubes by a CVD method. Part I: Synthesis and characterization of the (Mg, Fe)O catalysts, Journal of Physical Chemistry B 106, 13186–13198 (2002).

[65] Kotnala, R. K. et al. Influence of annealing on humidity response of RF sputtered nanocrystalline MgFe2O4 thin films, Thin Solid Films 519, 6135–6139 (2011).

[66] Mahmoud, H. R. El-Molla, S. A. & Saif, M. Improvement of physicochemical properties of Fe2O3/MgO nanomaterials by hydrothermal treatment for dye removal from industrial wastewater, Powder Technology 249, 225–233 (2013).

[67] Bohren, C. F. & Huffman, D. R. Absorption and scattering of light by small particles, Wiley (1998).

[68] Christy, A. A. Kvalheim, O. M. & Velapoldi, R. A. Quantitative analysis in diffuse reflectance spectrometry: A modified Kubelka-Munk equation. Infrared and raman spectroscopy, Vibrational Spectroscopy 9, 19–27 (1995).

[69] Jackson, S. D. Metal oxide catalysis, Wiley (2009).

[70] Nič, M. Jirát, J. Košata, B. Jenkins, A. & McNaught, A. IUPAC compendium of chemical terminology (2009).

[71] Nakamura, D. Kurita, Y. Ito, K. & Kubo, M. Covalency of metal-ligand bonds in potassium hexahaloplatinates (IV) studied by the pure quadrupole resonance of halogens, Journal of the American Chemical Society 82, 5783–5787 (1960).

[72] Atkins, P. W. Inorganic chemistry. 5th edition, Oxford University Press (2010).

[73] Morin, F. J. Oxides of the 3d transition metals, Bell System Technical Journal

37, 1047‐1084 (1958).

[74] Förster, H. in Characterization I, edited by H. Karge & J. Weitkamp, Springer, pp. 337-426 (2004).

[75] Lakshmi, S. Endo, T. & Siva, G. in Advanced aspects of spectroscopy, edited by M. Akhyar Farrukh (2012).

[76] Tanabe, Y. & Sugano, S. On the absorption spectra of complex ions. I, Journal of Physical Society 9, 753–766 (1954).

[77] Sherman, D. & Waite T David. Electronic spectra of Fe3+ oxides and oxide hydroxides in the near IR to near UV. American Mineralogist Journal, 1262–1269 (1985).

[78] Gfroerer, T. H. in Encyclopedia of analytical chemistry, John Wiley & Sons Ltd. (2000).

[79] Andrew R. Barron (ed.). Physical methods in chemistry and nano science (2014).

[80] Anpo, M. & Che, M. in Advances in catalysis, edited by Werner, B. C. G. a. H. K. and Haag, O. 119–257, Academic Press (1999).

11 Bibliography

180

[81] Lakowicz, J. R. Principles of fluorescence spectroscopy. 3rd edition, Springer (2006).

[82] van Dijken, A. Meulenkamp, E. A. Vanmaekelbergh, D. & Meijerink, A. Identification of the transition responsible for the visible emission in ZnO using quantum size effects, Journal of Luminescence 90, 123–128 (2000).

[83] Ischenko, V. et al. Zinc oxide nanoparticles with defects, Advanced Functional Materials 15, 1945–1954 (2005).

[84] Djurisić, A. B. & Leung, Y. H. Optical properties of ZnO nanostructures, Small 2, 944–961 (2006).

[85] Janotti, A. & Van de Walle, Chris G. Native point defects in ZnO, Physical Review B 76, 165202 (2007).

[86] Brauer, G. et al. Characterization of ZnO nanostructures. A challenge to positron annihilation spectroscopy and other methods, Physica Status Solidi (c) 6, 2556–2560 (2009).

[87] Wang, D. et al. Positron annihilation study of the interfacial defects in ZnO nanocrystals. Correlation with ferromagnetism, Journal of Applied Physics 107, 23524 (2010).

[88] Klingshirn, C. et al. 65 years of ZnO research - old and very recent results, Physica Status Solidi (b) 247, 1424–1447 (2010).

[89] Knutsen, K. E. et al. Zinc vacancy and oxygen interstitial in ZnO revealed by sequential annealing and electron irradiation, Physical Review B 86, 121203 (2012).

[90] Zhang, H. et al. Bulk and surface excitons in alloyed and phase-separated ZnO–MgO particulate systems, ACS Applied Materials & Interfaces 4, 2490–2497 (2012).

[91] Drouilly, C. et al. Role of oxygen vacancies in the basicity of ZnO. From the model methylbutynol conversion to the ethanol transformation application, Applied Catalysis A: General 453, 121–129 (2013).

[92] Stavale, F. Nilius, N. & Freund, H.-J. STM luminescence spectroscopy of intrinsic defects in ZnO(0001) thin films, Journal of Physical Chemistry Letters 4, 3972–3976 (2013).

[93] Stavale, F. Pascua, L. Nilius, N. & Freund, H.-J. Luminescence properties of nitrogen-doped ZnO, Journal of Physical Chemistry C 118, 13693–13696 (2014).

[94] Baer, D. R. et al. Surface characterization of nanomaterials and nanoparticles: Important needs and challenging opportunities, Journal of Vacuum Science & Technology A 31, 50820 (2013).

[95] Knoezinger, E. Diwald, O. & Sterrer, M. Chemical vapour deposition - a new approach to reactive surface defects of uniform geometry on high surface area magnesium oxide, Journal of Molecular Catalysis A Chemical 162, 83–95 (2000).

[96] Siedl, N. Koller, D. Sternig, A. K. Thomele, D. & diwald, O. Photoluminescence quenching in compressed MgO nanoparticle systems, Physical Chemistry Chemical Physics 16, 8339–8345 (2014).

11 Bibliography

181

[97] Walz, M.-M. Vollnhals, F. Schirmer, M. Steinrück, H.-P. & Marbach, H. Generation of clean iron nanocrystals on an ultra-thin SiO(x) film on Si(001), Physical Chemistry Chemical Physics 13, 17333–17338 (2011).

[98] Vittadini, A. et al. Defects in oxygen-depleted titanate nanostructures, Langmuir 28, 7851–7858 (2012).

[99] Bassett, P. J. Gallon, T. E. Prutton, M. & Matthew, J. High resolution auger electron spectrum of MgO(100), Surface Science 33, 213–218 (1972).

[100] Onsgaard, J. Barlow, S. M. & Gallon, T. E. Clean ZnO surfaces and oxygen adsorption on the (0001) surface studied by electron spectroscopy, Journal of Physics C: Solid State Phys. 12, 925–942 (1979).

[101] Mroczkowski, S. Calculated Auger yields and sensitivity factors for KLL–NOO transitions with 1–10 kV primary beams, Journal of Vacuum Science & Technology A 3, 1860 (1985).

[102] Weidenthaler, C. Pitfalls in the characterization of nanoporous and nanosized materials, Nanoscale 3, 792–810 (2011).

[103] Weibel, A. Bouchet, R. Boulc, F. & Knauth, P. The big problem of small particles. A comparison of methods for determination of particle size in nanocrystalline anatase powders, Chemistry of Materials 17, 2378–2385 (2005).

[104] Castro, R. H. R. Tôrres, R. B. Pereira, G. J. & Gouvêa, D. Interface energy measurement of MgO and ZnO. Understanding the thermodynamic stability of nanoparticles, Chemistry of Materials 22, 2502–2509 (2010).

[105] Sternig, A. diwald, O. Gross, S. & Sushko, P. V. Surface decoration of MgO nanocubes with Sulfur oxides: Experiment and theory, Journal of Physical Chemistry C 117, 7727–7735 (2013).

[106] Knözinger, E. Jacob, K.-H. Singh, S. & Hofmann, P. Hydroxyl groups as IR active surface probes on MgO crystallites, Surface Science 290, 388–402 (1993).

[107] Diwald, O. Sterrer, M. & Knözinger, E. Site selective hydroxylation of the MgO surface, Physical Chemistry Chemical Physics 4, 2811–2817 (2002).

[108] Chizallet, C. et al. Identification of the OH groups responsible for kinetic basicity on MgO surfaces by 1H MAS NMR, Journal of Catalysis 268, 175–179 (2009).

[109] Chizallet, C. et al. Role of hydroxyl groups in the basic reactivity of MgO. A theoretical and experimental study, Oil & Gas Science and Technology - Rev. IFP 61, 479–488 (2006).

[110] This is reasoned by the fact, that in the relevant temperature range of the CVS reactor system, i.e. in the range between 700 and 1100 K, Zn exhibits a substantially smaller vapor pressure than Mg in order to be directly combusted at reduced oxygen pressures.

[111] Hlaing Oo, W. M. McCluskey, M. D. Lalonde, A. D. & Norton, M. G. Infrared spectroscopy of ZnO nanoparticles containing CO2 impurities, Applied Physics Letters 86, 73111 (2005).

11 Bibliography

182

[112] Noei, H. Wöll, C. Muhler, M. & Wang, Y. Activation of carbon dioxide on ZnO nanoparticles studied by vibrational spectroscopy, Journal of Physical Chemistry C 115, 908–914 (2011).

[113] Buchholz, M. Weidler, P. G. Bebensee, F. Nefedov, A. & Wöll, C. Carbon dioxide adsorption on a ZnO(1010) substrate studied by infrared reflection absorption spectroscopy, Physical Chemistry Chemical Physics 16, 1672–1678 (2014).

[114] Weisz, S. Z. Calibration of oxygen Auger signal from single-crystal ZnO surfaces, Journal of Vacuum Science & Technology A 5, 302 (1987).

[115] Siedl, N. Koller, D. Sternig, A. K. Thomele, D. & diwald, O. Photoluminescence quenching in compressed MgO nanoparticle systems, Physical Chemistry Chemical Physics 16, 8339–8345 (2014).

[116] Shluger, A. L. Sushko, P. V. & Kantorovich, L. N. Spectroscopy of low-coordinated surface sites: Theoretical study of MgO, Physical Review B 59, 2417–2430 (1999).

[117] Beck, K. M. et al. Energy and site selectivity in O-atom photodesorption from nanostructured MgO, Surface Science 602, 1968–1973 (2008).

[118] Sternig, A. Stankic, S. Müller, M. Siedl, N. & diwald, O. Surface exciton separation in photoexcited MgO nanocube powders, Nanoscale 4, 7494–7500 (2012).

[119] Stevanovic, A. & Yates, J. T. Probe of NH3 and CO adsorption on the very outermost surface of a porous TiO2 adsorbent using photoluminescence spectroscopy, Langmuir 28, 5652–5659 (2012).

[120] Zhang, Z. & Yates, J. T. Band bending in semiconductors: chemical and physical consequences at surfaces and interfaces, Chemical Reviews 112, 5520–5551 (2012).

[121] Sun, W.-C. Yeh, Y.-C. Ko, C.-T. He, H. & Chen, M.-J. Improved characteristics of near-band-edge and deep-level emissions from ZnO nanorod arrays by atomic-layer-deposited Al2O3 and ZnO shell layers, Nanoscale Research Letters 6, 556 (2011).

[122] Janotti, A. & Van de Walle, Chris G. LDA + U and hybrid functional calculations for defects in ZnO, SnO2, and TiO2, physica status solidi (b) 248, 799–804 (2011).

[123] Hlaing Oo, W. M. McCluskey, M. D. Lalonde, A. D. & Norton, M. G. Infrared spectroscopy of ZnO nanoparticles containing CO2 impurities, Applied Physics Letters 86, 73111 (2005).

[124] Stankic, S. Bernardi, J. diwald, O. & Knözinger, E. Optical surface properties and morphology of MgO and CaO nanocrystals, Journal of Physical Chemistry B 110, 13866–13871 (2006).

[125] Sternig, A. Mueller, M. McCallum, M. Bernardi, J. & Diwald, O. BaO clusters on MgO nanocubes: A quantitative analysis of optical-powder properties, Small 6, 582–588 (2010).

11 Bibliography

183

[126] Müller, M. et al. Effect of protons on the optical properties of oxide nanostructures, Journal of the American Chemical Society 129, 12491–12496 (2007).

[127] Mar, G. L. Timbrell, P. Y. & Lamb, R. N. Factors influencing the chemical vapor deposition of oriented ZnO films using zinc acetate, Chemistry of Materials 7, 1890–1896 (1995).

[128] Spanhel, L. Colloidal ZnO nanostructures and functional coatings. A survey, Journal of Sol-Gel Science & Technology 39, 7–24 (2006).

[129] Li, Z. Luan, Y. Mu, T. & Chen, G. Unusual nanostructured ZnO particles from an ionic liquid precursor, Chemical Communications, 1258–1260 (2009).

[130] Ghosh, M. et al. Role of ambient air on photoluminescence and electrical conductivity of assembly of ZnO nanoparticles, Journal of Applied Physics 110, 54309 (2011).

[131] Hu, X. Masuda, Y. Ohji, T. Saito, N. & Kato, K. Low-temperature fabrication of bunch-shaped ZnO nanowires using a sodium hydroxide aqueous solution, Journal of Nanoscience & Nanotechnology 11, 10935–10939 (2011).

[132] Yue, S. Yan, Z. Shi, Y. & Ran, G. Synthesis of zinc oxide nanotubes within ultrathin anodic aluminum oxide membrane by sol–gel method, Materials Letters 98, 246–249 (2013).

[133] Kołodziejczak-Radzimska, A. & Jesionowski, T. Zinc oxide—From synthesis to application. A review, Materials 7, 2833–2881 (2014).

[134] Idriss, H. & Barteau, M. A. Photoluminescence from zinc oxide powder to probe absorption and reaction of oxygen, carbon monoxide, hydrogen, formic acid, and methanol, Journal of Physical Chemistry 96, 3382–3388 (1992).

[135] Idriss, H. Application of luminescence techniques to probe surface–adsorbate interactions on oxide single crystals, Journal of Vaccum Science & Technology A 11, 209 (1993).

[136] Idriss, H. & Barteau, M. A. in Impact of Surface Science on Catalysis, 261–331, Elsevier (2000).

[137] Xia, Y. Xiong, Y. Lim, B. & Skrabalak, S. E. Shape-controlled synthesis of metal nanocrystals: simple chemistry meets complex physics?, Angewandte Chemie International Edition English 48, 60–103 (2009).

[138] Briggs, D. & Grant, J. T. Surface analysis by Auger and x-ray photoelectron spectroscopy, IM Publications (2003).

[139] Klingshirn, C. ZnO: Material, Physics and Applications, ChemPhysChem 8, 782–803 (2007).

[140] Woll, C. The chemistry and physics of zinc oxide surfaces, Progress in Surface Science 82, 55–120 (2007).

[141] Segnit, E. R. & Holland, A. E. The system MgO-ZnO-SiO2, Journal of the American Ceramic Society 48, 409–413 (1965).

[142] Chen, J. Shen, W. Z. Chen, N. B. Qiu, D. J. & Wu, H. Z. The study of composition non-uniformity in ternary MgxZn1-x O thin films, Journal of Physics: Condensed Matter 15, L475-L482 (2003).

11 Bibliography

184

[143] Sternig, A. et al. Phase Separation at the Nanoscale. Structural properties of BaO segregates on MgO-based nanoparticles, Journal of Physical Chemistry C 115, 15853–15861 (2011).

[144] Mueller, R. Mädler, L. & Pratsinis, S. E. Nanoparticle synthesis at high production rates by flame spray pyrolysis, Chemical Engineering Science 58, 1969–1976 (2003).

[145] Tani, T. Watanabe, N. Takatori, K. & Pratsinis, S. E. Morphology of oxide particles made by the emulsion combustion method, Journal of the American Ceramic Society 86, 898–904 (2003).

[146] Tani, T. Mädler, L. & Pratsinis, S. E. Journal of Nanoparticle Research 4, 337–343 (2002).

[147] Hendy, S. C. & Brown, I. W. M. Advanced materials and nanotechnology. Proceedings of the international conference (AMN-4), Dunedin, New Zealand, 8-12 February 2009 (American Institute of Physics, Melville, N.Y. 2009).

[148] Spoto, G. et al. Carbon monoxide MgO from dispersed solids to single crystals: a review and new advances, Progress in Surface Science 76, 71–146 (2004).

[149] Mädler, L. et al. Direct formation of highly porous gas-sensing films by in situ thermophoretic deposition of flame-made Pt/SnO2 nanoparticles, Sensors and Actuators B: Chemical 114, 283–295 (2006).

[150] Stokłosa, A. & Laskowska, B. Ionic radii in mixed-valence and nonstoichiometric metal oxides and their polymorphic forms, Journal of Chemical Crystallography 38, 913–925 (2008).

[151] Ohtomo, A. et al. MgxZn1-xO as a II–VI widegap semiconductor alloy, Applied Physics Letters 72, 2466 (1998).

[152] Kim, Y.-I. Page, K. & Seshadri, R. Synchrotron x-ray study of polycrystalline wurtzite Zn1-xMgxO (0 ≤ x ≤ 0.15). Evolution of crystal structure and polarization, Applied Physics Letters 90, 101904 (2007).

[153] Choopun, S. et al. Realization of band gap above 5.0 eV in metastable cubic-phase MgxZn1-xO alloy films, Applied Physics Letters 80, 1529 (2002).

[154] Park, W. I. Yi, G.-C. & Jang, H. M. Metalorganic vapor-phase epitaxial growth and photoluminescent properties of Zn1-xMgxO (0 ≤ x ≤ 0.49) thin films, Applied Physics Letters 79, 2022 (2001).

[155] Wang, Y. S. Thomas, P. J. & O'Brien, P. Optical properties of ZnO nanocrystals doped with Cd, Mg, Mn, and Fe ions, Journal of Physical Chemistry B 110, 21412–21415 (2006).

[156] Stankic, S. Bernardi, J. Diwald, O. & Knoezinger, E. Optical surface properties and morphology of MgO and CaO nanocrystals, Journal of Physical Chemistry B 110, 13866–13871 (2006).

[157] Vanmaekelbergh, D. & van Vugt, L. K. ZnO nanowire lasers, Nanoscale 3, 2783–2800 (2011).

[158] John, T. T. Priolkar, K. R. Bessière, A. Sarode, P. R. & Viana, B. Effect of [OH-] linkages on luminescent properties of ZnO nanoparticles, Journal of Physical Chemistry C 115, 18070–18075 (2011).

11 Bibliography

185

[159] Madi, K. Forest, S. Cordier, P. & Boussuge, M. Numerical study of creep in two-phase aggregates with a large rheology contrast: Implications for the lower mantle, Earth and Planetary Science Letters 237, 223–238 (2005).

[160] Larico, R. Justo, J. F. & Assali, L. V. C. Iron in magnesium oxide at high pressures: a first principles theoretical investigation, Physica Status Solidi B 250, 750–754 (2013).

[161] Kotnala, R. K. et al. Influence of annealing on humidity response of RF sputtered nanocrystalline MgFe2O4 thin films, Thin Solid Films 519, 6135–6139 (2011).

[162] Sonder, E. Stratton, T. G. & Weeks, R. A. Kinetics of Fe2+ oxidation and Fe3+ reduction in MgO single crystals, Journal of Chemical Physics 70, 4603–4608 (1979).

[163] Larico, R. Justo, J. F. & Assali, L. V. C. Iron in magnesium oxide at high pressures: a first principles theoretical investigation, Physica Status Solidi B 250, 750–754 (2013).

[164] Carles, V. & Rousset, A. Elaboration and characterization of submicronic magnesio-wüstite (Mg1−xFex)O powders, Solid State Ionics 83, 309–321 (1996).

[165] Jing, J. & Campbell, S. J. Mixed hyperfine interactions and defect structure in magnesiowüstite Fe−Mg−O, Hyperfine Interact 68, 283-286 (1992).

[166] Woods, K. N. & Fine, M. E. Nucleation and growth of magnesioferrite in MgO containing 0.9% Fe3+, Journal of the American Ceramic Society 52, 186–188 (1969).

[167] Skinner, A. R. PhD thesis: Electron paramagnetic resonance of some 3d ions in magnesium oxide (1986).

[168] PHILLIPS, B. SÕMIYA, S. & MUAN, A. Melting relations of magnesium oxide-Iron oxide mixtures in air, Journal of the American Ceramic Society 44, 167–169 (1961).

[169] Jung, I.-H. Decterov, S. A. & Pelton, A. D. Critical thermodynamic evaluation and optimization of the Fe–Mg–O system, Journal of Physics and Chemistry of Solids 65, 1683–1695 (2004).

[170] Inglis, A. D. Clustering in iron-doped magnesium oxide (1981).

[171] Lee, D. K. et al. A facile synthesis of multi metal-doped rectangular ZnO nanocrystals using a nanocrystalline metal-organic framework template, Nanoscale 6, 10995–11001 (2014).

[172] Stoch, P. Mössbauer spectroscopy study of the crystallization of vitrified waste inceneration ashes, Physics of Non-Crystalline Solids 10 345–346, 153–156 (2004).

[173] Dyar, M. D. Agresti, D. G. Schaefer, M. W. Grant, C. A. & Sklute, E. C. Mössbauer spectroscopy of earth and planetary materials, Annual Review Of Earth And Planetary Sciences 34, 83–125 (2006).

[174] Smith, G. Evidence for absorption by exchange-coupled Fe2+-Fe3+ pairs in the near infra-red spectra of minerals, Physics & Chemistry of Minerals 3, 375-383 (1978).

11 Bibliography

186

[175] Scheinost, A. C. Chavernas, A. Barron, V. & Torrent, J. Use and limitations of second-derivative diffuse reflectance spectroscopy in the visible to near-infrared range to identify and quantify Fe oxide minerals in soils, Clays and Clay Minerals 46, 528–536 (1998).

[176] Taran, M. Koch-Müller, M. & Langer, K. Electronic absorption spectroscopy of natural (Fe2+, Fe3+)-bearing spinels of spinel s.s.-hercynite and gahnite-hercynite solid solutions at different temperatures and high-pressures, Physics & Chemistry of Minerals 32, 175-188 (2005).

[177] Hazen, R. M. & Prewitt, C. T. Effects of temperature and pressure on interatomic distances in oxygen-based minerals, American Mineralogist 62, 309–315 (1977).

[178] Meldrum, F. C. & Cölfen, H. Controlling mineral morphologies and structures in biological and synthetic systems, Chemical Reviews 108, 4332–4432 (2008).

[179] Patzke, G. R. Zhou, Y. Kontic, R. & Conrad, F. Oxide nanomaterials: synthetic developments, mechanistic studies, and technological innovations, Angewandte Chemie International Edition English 50, 826–859 (2011).

[180] Chang, S.-Y. et al. Spontaneous growth of one-dimensional nanostructures from films in ambient atmosphere at room temperature. ZnO and TiO2, Journal of Materials Chemistry 21, 4264 (2011).

[181] Peng, X. Jin, J. Ericsson, E. M. & Ichinose, I. General method for ultrathin free-standing films of nanofibrous composite materials, Journal of the American Chemical Society 129, 8625–8633 (2007).

[182] Finlayson-Pitts, B. J. Reactions at surfaces in the atmosphere: integration of experiments and theory as necessary (but not necessarily sufficient) for predicting the physical chemistry of aerosols, Physical Chemistry Chemical Physics 11, 7760–7779 (2009).

[183] Sternig, A. Müller, M. McCallum, M. Bernardi, J. & Diwald, O. BaO clusters on MgO nanocubes: a quantitative analysis of optical-powder properties, Small 6, 582–588 (2010).

[184] McKenna, K. P. Koler, D. Sternig, A. Siedl, N. Govind, N. Sushko, P. V. Diwald, O. Optical properties of nanocrystal interfaces in compressed MgO nanopowders, ACS Nano 5, 3003-3009 (2011).

[185] Pang, H. Ning, G. Gong, W. Ye, J. & Lin, Y. Direct synthesis of hexagonal Mg(OH)2 nanoplates from natural brucite without dissolution procedure, Chemical Communications (Camb.) 47, 6317–6319 (2011).

[186] Ding, Y. et al. Nanoscale magnesium hydroxide and magnesium oxide powders. Control over size, shape, and structure via hydrothermal synthesis, Chemistry of Materials 13, 435–440 (2001).

[187] MacKenzie, K. J. D. & Meinhold, R. H. Thermal decomposition of brucite, Mg(OH)2: a 25Mg MAS NMR study, Thermochimica Acta 230, 339–343 (1993).

[188] Pallister, P. J. Moudrakovski, I. L. & Ripmeester, J. A. Mg ultra-high field solid state NMR spectroscopy and first principles calculations of magnesium compounds, Physical Chemistry Chemical Physics 11, 11487–11500 (2009).

11 Bibliography

187

[189] Matkovic, B. Popovic, S. Rogic, V. Zunic, T. & Y, J. F. Reaction products in magnesium oxychloride cement pastes. System MgO-MgCl2-H2O, Journal of the American Ceramic Society 60, 504–507 (1977).

[190] Montle, J. F. & Mayhan, K. G. Magnesium oxychloride as a fire retardant material, Journal of Fire and Flammability/Fire Retardant Chemistry Supplement, 243–254 (1974).

[191] Chau, C. K. & Li, Z. Microstructures of magnesium oxychloride Sorel cement, Advances in Cement Research 20, 85–92 (2008).

[192] Jeevanandam, P. Mulukutla, R. S. Yang, Z. Kwen, H. & Klabunde, K. J. Nanocrystals to nanorods. A precursor approach for the synthesis of magnesium hydroxide nanorods from magnesium oxychloride nanorods starting from nanocrystalline magnesium oxide, Chemistry of Materials 19, 5395–5403 (2007).

[193] Brown, M. A. Carrasco, E. Sterrer, M. & Freund, H.-J. Enhanced stability of gold clusters supported on hydroxylated MgO(001) surfaces, Journal of the American Chemical Society 132, 4064–4065 (2010).

[194] Goodenough, J. B. Evolution of strategies for modern rechargeable batteries, Accounts of Chemical Research 46, 1053–1061 (2012).

[195] Fabbri, E. Pergolesi, D. & Traversa, E. Materials challenges toward proton-conducting oxide fuel cells: a critical review, Chemical Society Reviews 39, 4355–4369 (2010).

[196] Matkovic, B. & Young, J. F. Microstructure of magnesium oxychloride cements, Nature-Physical Science 246, 79–80 (1973).

[197] Altomare, A. et al. A kit of tools for phasing crystal structures from powder data, Journal of Applied Crystallography 46, 1231–1235 (2013).

[198] Sugimoto, K. Dinnebier, R. E. & Schlecht, T. Structure determination of Mg3(OH)5Cl.4H2O (F5 phase) from laboratory powder diffraction data and its impact on the analysis of problematic magnesia floors, Acta Crystallographica B 63, 805–811 (2007).

[199] Rodríguez-Carvajal, J. Recent advances in magnetic structure determination by neutron powder diffraction, Physica B: Condensed Matter 192, 55–69 (1993).

[200] Artioli, G. & Bullard, J. W. Cement hydration. The role of adsorption and crystal growth, Crystal Research and Technology 48, 903–918 (2013).

[201] Weber, J. Bayer, K. & Pintér, F. in Historic Mortars, edited by J. Válek, J. J. Hughes & C. J. W. P. Groot, Springer, 89–103 (2012).

[202] Krauss Juillerat, F. Gonzenbach, U. T. Elser, P. Studart, A. R. & Gauckler, L. J. Microstructural control of self-setting particle-stabilized ceramic foams, Journal of the American Ceramic Society 94, 77–83 (2011).

[203] Holzer, L. Muench, B. Wegmann, M. Gasser, P. & Flatt, R. J. FIB-nanotomography of particulate systems—part I. particle shape and topology of interfaces, Journal of the American Ceramic Society 89, 2577–2585 (2006).

[204] Misra, A. K. & Mathur, R. Magnesium oxychloride cement concrete, Bulletin of Materials Science 30, 239–246 (2007).

11 Bibliography

188

[205] Tan, Y. Liu, Y. Zhao, Z. Paxton, J. Z. & Grover, L. M. Synthesis and in vitro degradation of a novel magnesium oxychloride cement, Journal of Biomedical Materials Research A 103, 194–202 (2015).

[206] Sorel, S. T. On a new magnesium cement, Comptes Rendus hebdomadires des Seances de Academie des sciences 65, 102-4 (1867).

[207] Gheisi, A. et al. Spontaneous growth of magnesium hydroxide fibers at ambient conditions, Crystal Growth & Design 14, 4236–4239 (2014).

[208] Castellar, M. D. de, Lorente, J. C. Traveria, A. & Tura, J. M. Cracks in sorel's cement polishing bricks as a result of magnesium oxychloride carbonatation, Cement and Concrete Research 26, 1199–1202 (1996).

[209] Alonso, C. & Fernandez, L. Dehydration and rehydration processes of cement paste exposed to high temperature environments, Journal of Materials Science 39, 3015–3024 (2004).

[210] Dinnebier, R. E. Freyer, D. Bette, S. & Oestreich, M. 9Mg(OH)2·MgCl2·4H2O, a high temperature phase of the magnesia binder system, Inorganic Chemistry 49, 9770–9776 (2010).

[211] Li, G. Yu, Y. Li, J. Wang, Y. & Liu, H. Experimental study on urban refuse/magnesium oxychloride cement compound floor tile, Cement and Concrete Research 33, 1663–1668 (2003).

[212] Sternig, A. Koller, D. Siedl, N. diwald, O. & McKenna, K. Exciton formation at Solid–Solid interfaces. A systematic sxperimental and ab Initio study on compressed MgO nanopowders, Journal of Physical Chemistry C 116, 10103–10112 (2012).

[213] Baumann, S. O. et al. Size effects in MgO cube dissolution, Langmuir 31, 2770–2776 (2015).

[214] Different to HCl, SiCl4 can be applied in water free form via the gas phase and - also important for the highly sensitive surface reaction reported here - exhibits a lower reactivity than TiCl4.

[215] Runčevski, T. Dinnebier, R. E. & Freyer, D. Dehydration of the sorel cement phase 3Mg(OH)2·MgCl 2 ·8H2O studied by in situ synchrotron X-ray powder diffraction and thermal analyses, Zeitschrift für anorganische und allgemeine chemie 640, 100–105 (2014).

[216] Sutter, E. A. & Sutter, P. W. Size-dependent phase diagram of nanoscale alloy drops used in vapor−liquid−solid growth of semiconductor nanowires, ACS Nano 4, 4943–4947 (2010).

[217] Chau, C. K. & Li, Z. Microstructures of magnesium oxychloride sorel cement, Advances in Cement Research 20, 85–92 (2008).

[218] Newberg, J. T. McIntire, T. M. & Hemminger, J. C. Reaction of bromide with bromate in thin-film water, Journal of Physical Chemistry A 114, 9480–9485 (2010).

[219] Garde, S. & Schlossman, M. L. Water at functional interfaces, MRS Bulletin 39, 1051–1053 (2014).

11 Bibliography

189

[220] Zhang, H. et al. A new vapor-phase hydrothermal method to concurrently grow ZnO nanotube and nanorod array films on different sides of a zinc foil substrate, Chemistry 18, 5165–5169 (2012).

[221] Liu, P. et al. A facile vapor-phase hydrothermal method for direct growth of titanate nanotubes on a titanium substrate via a distinctive nanosheet roll-up mechanism, Journal of the American Chemical Society 133, 19032–19035 (2011).

[222] Up to now we did not succeed in an unambiguous identification of the intermediate phases such as 3Mg(OH)2*MgCl2*5.4 and 4.6 H2O. See also reference [215].

[223] Kashani-Nejad, S. et al. MgOHCl thermal decomposition kinetics, Metallurgical and Materials Transactions B 36, 153-157 (2005).

stability and transformation of particulate mgo-based nanocomposites stabilit¤t und

Documents