relation between toughness and molecular coupling at cross
TRANSCRIPT
University of Wollongong Thesis Collections
University of Wollongong Thesis Collection
University of Wollongong Year
Relation between toughness and
molecular coupling at cross-linked
polymer/solid interfaces
Michaela TymichovaUniversity of Wollongong
Tymichova, Michaela, Relation between toughness and molecular coupling at cross-linkedpolymer/solid interfaces, PhD thesis, School of Mechanical, Materials and Mechatronics,University of Wollongong, 2005. http://ro.uow.edu.au/theses/463
This paper is posted at Research Online.
http://ro.uow.edu.au/theses/463
RELATION BETWEEN TOUGHNESS AND MOLECULAR
COUPLING AT CROSS-LINKED POLYMER/SOLID INTERFACES
A thesis submitted in fulfillment of the
requirements for the award of the degree
Doctor of Philosophy
from
UNIVERSITY OF WOLLONGONG
by
MICHAELA TYMICHOVA
FACULTY OF ENGINEERING
2005
i
CERTIFICATION
I, Michaela Tymichova, declare that this thesis, submitted in fulfilment of the requirements
for the award of Doctor of Philosophy, in the Faculty of Engineering, University of
Wollongong, is wholly my own work unless otherwise referenced or acknowledged. The
document has not been submitted for qualifications at any other academic institution.
Michaela Tymichova
19 August 2005
iii
Table of Contents Page
Certification i
Table of Contents iii
List of Tables vi
List of Figures vii
Abstract xi
Acknowledgements xv
1. INTRODUCTION 1
1.1 Introduction 3
1.2 References 7
2. SYNTHESIS AND CHARACTERISATION OF POLYMERIC SYSTEMS 9
2.1 Functionalisation of Polystyrene Chains 11
2.1.1 “Living” Free Radical Polymerisation Techniques 11
2.1.1.1 Nitroxy-mediated Free Radical Polymerisation 12
2.1.1.2 Atom Transfer Radical Polymerisation 14
2.1.2 Chemical Modification of Polystyrene Chain Polymer Analysis 16
2.1.2.1 Bromination of Polystyrene 16
2.1.2.2 Conversion of Brominated Polystyrene 18
2.2 Polymer Analysis 20
2.2.1 Gel Permeation Chromatography 20
2.2.2 Nuclear Magnetic Resonance Spectroscopy 24
2.3 Results and Discussion 26
2.3.1 Copolymers prepared by NMP 26
2.3.2 Copolymers prepared by ATRP 26
2.3.3 Analysis of Brominated Polystyrene
by 13C-NMR Spectroscopy 27
2.3.4 Monitoring Process of Converting PS to PS-NH2
iv
by 1H-NMR Spectroscopy 30
2.4 References 35
3. POLYMER CHAINS AS COUPLING AGENTS 37
3.1 Coupling Agents 39
3.1.1 Commercially-available Coupling Agents 39
3.1.2 Mechanism of Coupling through Silane Coupling Agents 43
3.1.3 Nature of Bonding with Silane Coupling Agents 46
3.1.4 Self-Assembled Monolayers 47
3.1.5 Mixed Monolayers 49
3.2 Adsorption of Polymer Chains 51
3.2.1 Competitive Adsorption of Polymer Chains 53
3.2.2 Physisorption vs. Chemisorption 56
3.3 Tethered Polymer Chains 56
3.3.1 Grafting to vs. Grafting from Techniques 58
3.4 Techniques for Synthesis of Grafted Polymer Chains 59
3.4.1 Grafting of Polymer Brushes using NMP 60
3.4.2 Grafting of Polymer Brushes using ATRP 62
3.4.3 Grafting of Polystyrene chains and their Functionalisation 63
3.5 Analysis of Grafted Surfaces 65
3.5.1 Ellipsometry 65
3.5.2 Surface Energy by Contact Angle Measurement 68
3.6 Results and Discussion 71
3.6.1 NMP Results 71
3.6.2 ATRP Results 73
3.6.3 Bromination of PS-SiClMe2 and Deposition of Mixed
PSBr0.1-SiClMe2/PS-SiClMe2 monolayers 77
3.7 References 85
v
4. INTERFACIAL THOUGHNESS MEASUREMENTS 89
4.1 Mechanism of Adhesion 91
4.1.1 Donor-Acceptor Interactions 92
4.2 Interfacial Fracture Toughness 94
4.2.1 Modes of Fracture 95
4.3 Thermosets 97
4.4 System Studied 99
4.4.1 Mechanism of Cross-linking 99
4.4.2 Characterisation of Epoxy System 102
4.5 Fracture Mechanism of Thermosets 103
4.6 Interfacial Toughness Measurements 104
4.6.1 Asymmetric Double Cantilever Test 107
4.7 Tailoring the Interfacial Toughness using
Polymeric Coupling Agents 108
4.8 ADCT Experiments and Results 111
4.9 References 121
5. DISCUSSION 123
5.1 Discussion 125
5.2 References 135
6. CONCLUSION 137
6.1 Conclusion 139
6.2 Suggestions for Future Work 141
6.3 References 143
APPENDICES 145
vi
List of Tables
Page
II-1: Copolymerisation conditions 15
II-2: Examples of packaging materials for high resolution GPC 23
III-1: Examples of silanes with different non-hydrolysable groups 42
III-2: Surface Tension of Test Liquids 70
III-3: PS and PS/PHEMA brushes 71
III-4: PS brush on Si wafer prepared by NMP 73
III-5: Grafting of PS chains 74
III-6: PS brush on Si wafer 74
III-7: Thickness of PHEMA-TMS/PS brush 76
III-8: Mixed PSBr(0.1)SiClMe2 and PS-SiClMe2 layers (Exp. 1) 78
III-9: Mixed PSBr(0.1)SiClMe2 and PS-SiClMe2 layers (Exp. 2) 79
III-10: Mixed PSBr(0.1)SiClMe2 and PS-SiClMe2 layers (Exp. 3) 80
III-11: Mixed PSBr(0.1)SiClMe2 and PS-SiClMe2 layers (Exp. 4) 80
III-12: Change in thickness of the polymer layer after Heck’s reaction (Exp. 1) 82
III-13: Change in thickness of the polymer layer after Heck’s reaction (Exp. 2) 82
III-14: Change in thickness of the polymer layer after Heck’s reaction (Exp. 3) 82
III-15: Change in thickness of the polymer layer after Heck’s reaction (Exp. 4) 83
IV-1: Cure regime of DGEBA/DMP system 102
IV-2: Gc and sample parameters 111
vii
List of Figures
Page
Figure 2-1: PS/PS(NH2), PS/PHEMA and PS/PGMA copolymers. 11
Figure 2-2: 1-phenyl-1-(2’,2’,6’,6’-tetramethyl-1’-piperidinyloxy)ethane. 12
Figure 2-3: NMP mechanism. 13
Figure 2-4: Bromination of polystyrene. 17
Figure 2-5: Formation of C-N bond using Heck reaction. 18
Figure 2-6: Introduction of -NH2 groups along the PS chains. 19
Figure 2-7: Schematic example of a chromatogram. Xi is an amount
of material eluted, Mi represents different fractions. 21
Figure 2-8: An example of a calibration curve [www.sdk.co.jp]. 22
Figure 2-9a: 13C-NMR spectrum of pure PS-SiClMe2. 27
Figure 2-9b: 13C-NMR spectrum of partially brominated PS(Br)-SiClMe2. 28
Figure 2-9c: 13C-NMR spectrum of partially brominated PS(Br)-SiClMe2. 28
Figure 2-9d: 13C-NMR spectrum of partially brominated PS(Br)-SiClMe2. 29
Figure 2-10: Conversion of PS to PS(NH2) . 30
Figure 2-11a: 1H-NMR spectrum of PS (MW=200K). 31
Figure 2-11b: 1H-NMR spectrum of PS(Br)x (x = 0.1). 31
Figure 2-11c: 1H-NMR spectrum of PS-N(Si(Me3)2). 32
Figure 2-11d: 1H-NMR spectrum of PS-NH2. 32
Figure 2-12a: 1H-NMR spectrum of PS-NH2. 33
Figure 2-12b: 1H-NMR spectrum of PS-NH2 with D2O added. 33
Figure 3-1: Schematics of coupling mechanism using coupling agents. 40
Figure 3-2: Functionalities of silane hydrolysable groups . 41
Figure 3-3: Examples of short and long chain silane molecules. 42
Figure 3-4: Hydrolysis of a silane coupling agent with three methoxy
end-functional groups. 44
Figure 3-5: Formation of oligomers and their condensation to the substrate. 44
viii
Figure 3-6: a) H-bonding to the substrate, b) consequent covalent bond
formation upon curing. 44
Figure 3-7: Bonding of γ-APS to silicon substrate. 46
Figure 3-8: Schematics of the silanation reaction. 48
Figure 3-9: Conformations of adsorbed polymer chains. 51
Figure 3-10: General adsorption isotherm of a polymer with a high affinity
for a substrate. 52
Figure 3-11: Pancake, mushroom and brush conformations. 57
Figure 3-12: Chemisorption of surface active initiators to Si substrate. 60
Figure 3-13: a) deposition of mixed initiators; b) polymer brush. 61
Figure 3-14: a) Deposition of γ-APS and n-BTMS on Si wafer
b) ATRP agent attachment. 62
Figure 3-15: Chemically adsorbed PSBr and PS chains on Si wafer. 63
Figure 3-16: Heck reaction using LiN(SiMe3)2. 63
Figure 3-17: Catalysts for Heck reaction: a) Pd(dba)2 , b) P(t-Bu)3. 64
Figure 3-18: Conversion of the PSBr to the silane protected PS. 64
Figure 3-19: Conversion of the silane terminated PS to PS-NH2. 64
Figure 3-20: Single-film model. ϕ is an angle of incident and reflected beam,
n and k are the real and imaginary parts of a refractive index
of a substrate and the film, t is the film thickness. 65
Figure 3-21: Schematics of an analysing section of AutoEl-II Ellipsometer. 66
Figure 3-22: Double-film model. ϕ is an angle of incident and reflected beam,
n and k are the real and imaginary parts of a refractive index
of a substrate and the film, t is the film thickness. 67
Figure 3-23: a) So > 0, b) So < 0. Θ is an equilibrium contact angle. 69
Figure 3-24: γ-APS thickness vs. γ-APS concentration deposited on Si wafer. 75
Figure 3-25: Bonding of Si to epoxy resin via PS-NH2 chains. 84
Figure 4-1: Adsorption of basic PMMA onto acidic silica from basic, neutral
and acidic solvents. 93
Figure 4-2: Modes of loading: a) mode I, b) mode II, c) mode III . 96
ix
Figure 4-3: Schematics of a) stable and b) stick/slip mode of crack propagation. 98
Figure 4-4: Structures of epoxy and diamino curing agents. 100
Figure 4-5: Mechanism of Cross-linking between di-epoxy resin
and diamino hardener system. 101
Figure 4-6: Tg of various cure regimes of DGEBA/DMP measured by DMA. 103
Figure 4-7: Schematics of the “sandwich” structure for testing
PS/glass interface. 105
Figure 4-8: Schematics of the “sandwich” structure for testing
PS/silicon interface. 106
Figure 4-9: Asymmetric double cantilever beam specimen; Gc = f (E, h, ∆2, a4). 108
Figure 4-10: Photographs of an ACDT specimen. 109
Figure 4-11: Monolayer thickness of the modified Si surfaces (Exp 1). 112
Figure 4-12: Surface energy of the modified Si surfaces (Exp 1). 112
Figure 4-13: Interfacial fracture toughness measurements using ADCT (Exp 1). 113
Figure 4-14: Monolayer thickness of the modified Si surfaces (Exp 2). 114
Figure 4-15: Surface energy of the modified Si surfaces (Exp 2). 114
Figure 4-16: Interfacial fracture toughness measurements using ADCT (Exp 2). 115
Figure 4-17: Monolayer thickness of the modified Si surfaces (Exp 3). 116
Figure 4-18: Interfacial fracture toughness measurements using ADCT (Exp 3). 116
Figure 4-19: Monolayer thickness of the modified Si surfaces (Exp 4). 117
Figure 4-20: Surface energy of the modified Si surfaces (Exp 4). 118
Figure 4-21: Interfacial fracture toughness measurements using ADCT (Exp 4). 118
xi
Abstract
The relationship between the interfacial fracture toughness (Gc) and molecular
coupling between epoxy polymers and silicon wafers was studied using the asymmetric
double cantilever test (ADCT). In order to investigate the molecular coupling, the coupling
molecules had to be applied along the polymer/substrate interface in various concentrations.
The grafting from technique by means of the “living” free radical polymerisation
techniques, namely nitroxy-mediated “living” free polymerisation (NMP) and atom transfer
radical polymerisation (ATRP), were used to prepare suitable coupling molecules.
Unfortunately, these techniques did not produce satisfactory results, and the new route of
grafting to was investigated. This method involved chemical attachment of brominated
polystyrene chains (PS) to the silicon substrate. The bromine functional groups of the
brominated polystyrene (PS(Br)) were then further converted to amino (-NH2) groups using
the Heck reaction, producing PS(NH2).
The conversion method was first tested using the free (unattached) polystyrene
(MW = 200K) which was brominated (molar fraction of brominated units, x = 0.1). The
conversion from PS to PS(Br) and then to PS(NH2) was monitored by proton and carbon
nuclear magnetic resonance techniques (1H-NMR and 13C-NMR) and gel permeation
chromatography.
For the grafting to experiments, monochlorosilane end-functionalised polystyrene
(PS-SiClMe2) (MW = 8000) was used. Various ratios of PS and PS(Br) ranging from 0 to
20% PS(Br) were deposited onto silicon substrates. Applying the Heck reaction, Br groups
were converted to NH2, producing surfaces with different concentrations of amino groups
which were expected to react with the epoxy polymer.
ADCT was adapted to investigate the dependence of the interfacial fracture
toughness on the degree of coupling. The dependence of Gc on the concentration of the
coupling molecules was not directly confirmed. This was attributed mainly to the
xii
challenges in deposition of the polymer chains and the final conversion to PS(NH2).
However, the differences in Gc values between experiments suggested that Gc’s measured
by ADCT reflect the differences in surface properties of the polymer modified surfaces, and
therefore this technique is suitable for interfacial adhesion measurements between epoxy
polymers and solid substrate.
xiii
I dedicate this thesis to all people who dare to be different, to all who yearn for
achieving something that seems impossible but they do it anyway. Sometimes, the results
are not what we expected or were hoping for. It does not matter. The important thing is that
the seeds were planted.
xv
Acknowledgements
I’d like to take the opportunity to express my gratitude to many people who helped
me on my journey. Many thanks go to my supervisors prof. Hugh Brown and Dr. Chris
Lukey. To Hugh, for believing in my skills and capabilities to entrust this project to me. To
Chris, for the endless discussions, and also for his kindness and being there for me
whenever I was close to giving up. I would also like to express my appreciation to the
following people: prof. A. Whittaker (Centre for Magnetic Resonance, QUT), Dr. W. Lie
and Ms. S. Chapman (University of Wollongong) for the NMR analysis; prof. S. Pyne
(University of Wollongong) for his expert advice in chemical synthesis and analysis; Dr. C.
Hawker (University of California Santa Barbara, USA) and Dr. M. Whittaker (PolymerRat)
for their assistance with NMP and ATRP techniques; and Prof. S. X. Dou and Dr. K.
Konstantinov (University of Wollongong) for providing a space for the air sensitive
chemical procedures.
I am very grateful to Robert Oslanec, my dearest friend and life companion, for
sharing his experience but also for sometimes not saying anything and letting me figure it
out for myself. Special gratitude goes to Lorelle Pollard for her cheerful face and caring.
And at last but not at least, I’d like to thank all my colleagues, Sandra Cram, Nathan Jones,
Damien Jinks, Donna Capararo, Wanda Melfo, Daniel McCubbery, Haider K. Habboby,
Siu W. Wai, Nicolas L. Baut, Dominic Phelan, Kirsten Carpenter, and Mark Reed for their
sharing and comradeship.
This research was funded by ARC grant (Adhesion Fundamental Award).
CHAPTER 1
Introduction
3
1.1 Introduction
Cross-linked glassy polymers (unsaturated polyesters, epoxy resins, phenolic
resins, etc.) are materials widely used as adhesives, coatings, composites and
encapsulants. Some examples of epoxy resins are bisphenol-A-glycidyl ether oligomers,
novolacs, aliphatic glycidyl ethers and cycloaliphatic epoxies. Upon curing, these
polymers form insoluble 3-D networks. Various curing agents are used, including
aliphatic amines, aliphatic amine polyamides, aromatic amines, cyclic polyacid
anhydrides, dicyandiamides and others. In this study, a monomeric bisphenol-A-
glycidyl ether (DGEBA) and diamino methylpentane cross-linker (DMP) are used (see
Chapter 4). Epoxy resins have excellent resistance to many chemicals and high
temperatures. Because of these properties, they are widely used as adhesives for metals,
ceramics, glass, concrete and wood. They are also used in high performance coatings,
water resistant paints, solvent free paints, electrical insulators and fibreglass
composites. Interfacial adhesion is an important property in all these applications, so
improving the adhesion properties of these polymers would greatly increase their
applicability in various situations.
One of the important applications of interfacial adhesion between epoxy resins
and inorganic materials is in the manufacture of printed circuit boards (PCB). PCB
technology is based on glass reinforced epoxy laminates. The processing involves
mechanical and chemical treatments (drilling, degreasing, drying, copper plating,
rinsing, stripping, etching, immersion in molten tin, cleaning and humidity testing)1.
When water or moisture penetrates into the interfacial regions it causes delamination
and damages the boards. It has been shown, that adding an appropriate silane, as a
replacement for chrome treatments, creates bonds that withstand boiling water
conditions and therefore improves the interfacial adhesion strength between the glass
fibres and epoxy1.
To improve the adhesion strength between various materials, adhesion
promoters, also called coupling agents, are often used. Organofunctional silanes are one
4
of the most versatile coupling agents, used to bond organic and inorganic materials.
They are hybrids of silica and organic materials (see Chapter 3). There are a large
number of different silanes available, and they have to be carefully chosen for a
particular application. The best interfacial bond strength is usually achieved by
choosing a silane with the organofunctional group that has maximum reactivity with a
particular resin during cure.
Finding the optimal types of silanes and methods of their application to the
various surfaces has been well researched1-10. However, it is not known how much
coupling agent is required in order to achieve the desired adhesion strength. This is
because of the lack of understanding of the mechanism of interfacial adhesion between
resin/inorganic substrate on a molecular level. To the author’s best knowledge, there is
little understanding about the amount of a chemical reaction (concentration of coupling
molecules along the interface) needed to achieve a particular adhesion strength between
a solid substrate and a cross-linked polymer. This applies to other epoxy/inorganic
material applications also, including structural laminates, adhesives and coatings.
The aim of this study was to investigate the mechanism of coupling, and the
relationship between interfacial toughness and the areal density of covalently bonded
coupling molecules. In thermoplastics (polystyrene, polymethylmethacrylate, etc.),
coupling across an interface is mainly due to chain entanglements on both sides of the
interface. In cross-linked systems, strong adhesion is more likely to be achieved by a
chemical reaction between the thermoset and the coupling molecules. In order to gain a
quantitative understanding of the effects of the coupling agents on the interfacial
strength between a solid inorganic substrate and a cross-linked polymer (DGEBA), a
system had to be designed where only covalent bonds would contribute to the adhesion
strength.
A considerable amount of work was dedicated to finding macromolecules of a
suitable, well-defined size and structure, which would serve as a chemical bridge
between the substrate and the polymer. The first logical choice would be to use silane
5
coupling agents because of their ability to react strongly both with the inorganic
substrate and the cross-linked polymer. However, because of their single organic
functionality and undesirable behaviour during deposition (bonding of the functional
group to the surface - see Chapter 3), non-polymeric silanes were not suitable for this
study. Multiple organic functionality on each chain gives a better chance for each
molecule to react with the polymer. This requirement is an essential part of the
quantitative study of interfacial toughness.
Several avenues were explored in order to prepare suitable coupling molecules
with multiple functionalities. Initially, it was thought that dendrimers or hyperbranched
polymers, extensively studied by Tomalia et. al.11,12, might be used. The advantage of
these macromolecules is that their size, shape, surface chemistry and topology can be
controlled during the synthesis. The main disadvantage is that the preparation of these
molecules in a controlled manner is quite demanding.
Another avenue was to grow polymer chains with particular functional groups
directly from the substrate, referred to as “grafting from” method. This can be achieved
by controlled “living” free radical polymerisation techniques (CRP). These synthetic
routes have the advantage of producing polymers with well-defined molar mass and
structure. In the last several years, a technique known as nitroxy-mediated “living” free
radical polymerisation has been studied by Hawker to produce well-defined
macromolecules called polymer brushes13. Another CRP technique investigated in this
study was atom transfer radical polymerisation (ATRP). Compared to nitroxy-mediated
living free radical polymerisation, this technique allows a wider variety of monomers to
be co-polymerised.
An alternative way to prepare the multifunctional polymer coupling molecules is
a method of “grafting to”, where chemically modified polymer chains are chemically
attached to the substrate. A commercial silane-terminated polystyrene of very low
polydispersity was chemically modified to produce a polystyrene-based coupling agent
with multiple amino groups along the chain14-17. This procedure was tested using free
6
polymer chains. Each step of the chemical modification was monitored using 13C-NMR
and 1H-NMR techniques (Chapter 2). This last approach was the most successful in
preparing silicon oxide surfaces modified with various amounts of coupling agent with
amino functionalities, and therefore it was used for the adhesion studies of the
interfacial fracture toughness between epoxy (DGEBA) and silicon oxide.
It is known that polystyrene/silicon oxide and polystyrene/epoxy resin adhesion
strength is very weak18. By choosing this system and developing a coupling agent based
on polystyrene, the effect of hydrogen bonding, dipolar and acid-base interactions, that
otherwise contribute to the interfacial toughness, were minimised. Silane end-functional
polystyrene chains were chemically modified by attaching randomly distributed amino
groups. These molecules were used as coupling agents for studying the interfacial
toughness between a silicon substrate and epoxy resins. The multiple functionality of
the coupling molecules ensured that each molecule reacts with the epoxy resin during
the curing procedure.
In order to understand the effect of molecular coupling on adhesion toughness,
the degree of coupling across the interface has to be controlled and evaluated. The
amount of coupling can be calculated from the amount of coupling agent deposited onto
a substrate. In this work, ellipsometry was used to determine the layer thickness of the
coupling molecules. From the thickness of the layers, the chain areal density can be
calculated.
The relationship between interfacial toughness and molecular coupling in a
cross-linked system was studied by applying the methods of fracture mechanics. One
end of the silane end-functional aminopolystyrene chains was covalently attached to the
silicon wafer, and the other end is glued to a pre-cured epoxy beam using the same
epoxy resin (see Chapter 4). The Asymmetric double cantilever test (ADCT),
previously used by Smith and Brown18-21 to study interfacial toughness of non-
crosslinked polymers, was modified and applied to measure the interfacial fracture
toughness. In this study, some characteristics of the testing samples were altered in
7
order to overcome difficulties associated with brittleness of the silicon wafers, and to
ensure that the crack propagated along the silicon oxide/polymer interface.
1.2 References
1 E. P. Plueddemann, Silane Coupling Agents (Plenum Press, 1982). 2 B. Arkles; Vol. 7 (Petrarch Systems, Inc., 1977). 3 J. B. Brzoska, J. B. Azouz, and F. Rondelez, Langmuir 10, 4367-4373 (1994). 4 J. Duchet, J.-F. Gerard, J. P. Chapel, and B. Chabert, Composite Interfaces 8, 177-
187 (2001). 5 J. Duchet, J. P. Chapel, B. Chabert, and J. F. Gerard, Macromolecules 31, 8264-
8272 (1998). 6 J. Duchet, B. Chabert, J. P. Chapel, J. F. Gerard, J. M. Chovelon, and N. Jaffrezic
Renault, Langmuir 13, 2271-2278 (1997). 7 C. R. Kessel and S. Granick, Langmuir 7, 532-538 (1991). 8 A. J. Kinloch, Structural Adhesives: Developments in Resins and Primers (Elsevier
Applied Science Publishers, London). 9 M. E. McGovern, K. M. R. Kallury, and M. Thompson, Langmuir 10, 3607-3614
(1994). 10 P. Silberzan, L. Leger, D. Ausserre, and J. J. Benattar, Langmuir 7, 1647-1651
(1991). 11 D. A. Tomalia, Scientific American May, 62-66 (1995). 12 D. A. Tomalia, A. M. Naylor, and W. A. Goddard III, Angew. Chem. Int. Ed. Engl.
29, 138-175 (1990). 13 C. J. Hawker, Acc. Chem. Res. 30, 373-382 (1997). 14 R. P. Kambour and J. T. Bendler, Macromolecules 19, 2679-2682 (1986). 15 S. Lee, M. Jorgensen, and J. F. Hartwig, Organic Letters 3, 2729-2732 (2001). 16 V. C. Gibson and W. Reed, in Europian Patent Application (UK, 1998), p. 1-9. 17 K. Suzuki, K. Yamaguchi, A. Hirao, and S. Nakahama, Macromolecules 22, 2607-
2611 (1989).
8
18 J. W. Smith, E. J. Kramer, F. Xiao, C.-Y. Hui, W. Reichert, and H. R. Brown, J. of
Mat. Sci. 28, 4234-4244 (1993). 19 J. W. Smith, E. J. Kramer, and P. J. Mills, J. Polymer Sci.:Part B: Polymer Physics
32, 1731-1744 (1994). 20 H. R. Brown, Macromolecules 24, 2752-2756 (1991). 21 H. R. Brown, K. Char, V. R. Deline, and P. F. Green, Macromolecules 26, 4155-
4163 (1993).
CHAPTER 2
Synthesis and Characterisation
of Polymeric Systems
11
2.1 Functionalisation of Polystyrene Chains
Physical and chemical properties of polymer chains can be tailored by
introducing various functional groups along the chain. This can be achieved either by
chemical modification of an existing chain or by a synthetic route of copolymerisation.
Traditionally, copolymers are prepared by free radical or ionic polymerisation
techniques. When polymers with well-controlled properties are desired, “living” free-
radial polymerisation techniques can be applied.
2.1.1 “Living” Free Radical Polymerisation Techniques
“Living” free radical polymerisation techniques have been developed in the last
decade to synthesise molecules with well defined structure, physical and chemical
properties. Polymers with controlled molecular mass, desired functionality and low
polydispersity can be prepared without using the demanding reaction conditions of
anionic and cationic polymerisation techniques.
There are currently three types of “Living” free radical polymerisation
techniques: nitroxy-mediated free radical polymerisation (NMP), atom transfer radical
polymerisation (ATRP) and reversible addition fragmentation chain transfer (RAFT).
All of them follow the same principles, but use different types of the initiators and
therefore the mechanism varies. The NMP and ATRP are techniques used in the current
study, and are briefly described below. Styrene/4-aminostyrene (PS/PS(NH2)),
styrene/2-hydroxyethyl methacrylate (PS/PHEMA) and styrene/glycidyl methacrylate
(PS/PGMA) were copolymerised (Fig. 2-1) in order to create copolymers specifically
designed to react with the DGEBA/DMP system through their functional groups (-NH2,
-OH and –OCHCH2).
Figure 2-1: PS/PS(NH2), PS/PHEMA and PS/PGMA copolymers.
12
2.1.1.1 Nitroxy-mediated Free Radical Polymerisation
Nitroxy-mediated free radical polymerisation (NMP) uses TEMPO (2.2.6.6-
tetramethylpiperidinyloxy)-based unimolecular initiators (Fig. 2-2) with a thermally
unstable alkoxyamine (C-ON) bond to reversibly react with the growing polymer chain
end. The unimolecular initiator undergoes thermal fragmentation or homolysis to give a
stable nitroxide and a polymer radical. When the polymer radical reacts with the
monomer added to a solution, the result is the polymer radical with an increased degree
of polymerisation through monomer addition. The extended polymeric radical then
recombines with the nitroxide to form the inactive molecule, and the cycle repeats (Fig.
2-3)1.
Figure 2-2: 1-phenyl-1-(2’,2’,6’,6’-tetramethyl-1’-piperidinyloxy)ethane.
The nitroxide free radical reacts with the polymeric radical but it does not
initiate the growth of any other polymer chains. Because the concentration of these
radical chain ends is low there is a decreased possibility of side reactions (termination,
disproportionation and combination) compared to the traditional free radical
polymerisation techniques. This behaviour leads to the possibilities of preparing a wide
range of molecules. For example styrene based homopolymers (polystyrene and poly(4-
hydroxystyrene))2,3, random and block copolymers (PS/PMMA, PS/PHEMA)1, complex
macromolecular architectures4 (grafts, dendrimers, hyperbranched polymers) and other
unique macromolecular structures (dendritic-linear block copolymers) 5. All these novel
molecular structures are finding applications in surface science, adhesion, coating
technologies, microelectronics and biotechnology. A great advantage of this technique
is its simple, undemanding reaction conditions with no need for laborious reagent
purification.
13
Figure 2-3: NMP mechanism.
Controlled synthesis of random copolymers of polystyrene with poly(2-
hydroxyethyl) methacrylate (PS/PHEMA), poly(glycidoxy) methacrylate PS/PGMA
and aminopolystyrene (PS/PS(NH2)) were carried out according to the procedures
published by Hawker et. al.4. A mixture of the monomers (90% styrene, 10% other
monomer) and the unimolecular initiator (Fig. 2-2) were placed in a round bottom flask
sealed with a septum, placed in an oil bath and heated at 125oC for 24 hour under an
inert atmosphere. The resulting copolymers, which were either solid or very viscous,
were dissolved in CH2Cl2 and precipitated into a large amount of methanol, then dried
in an oven at 60oC under vacuum. This process of dissolving, precipitating and drying
was repeated 3 times to remove any unreacted monomers and solvent. The molecular
mass of the polymer is given by the molar ratio of monomers to the unimolecular
initiator. In the present work, 500:1 and 200:1 molar ratio of monomer to initiator were
used. The polymers were analysed by GPC and NMR.
14
2.1.1.2 Atom Transfer Radical Polymerisation
While NMP proved to be most successful for synthesis of styrene based
polymers, atom transfer radical polymerisation (ATRP) is also suitable for
copolymerisation of other monomers like methacrylates, methacrylamides and
acrylonitriles. In ATRP, the active species (radicals*) are generated by a reversible
redox process catalysed by a transition metal complex (Mtz−Y/Ligand). The oxidation
state of the metal complex changes by one and the halogen atom (X) is removed from
the dormant species Pn−X to give a carbon centred Pn* radical and the metal complex
X−Mtz+1−Y/Ligand6. This carbon-halogen cleavage is a reversible homolytic process.
Polymer chains grow by addition of monomers to the intermediate polymer radicals Pn*.
The mechanism of growing the polymer chains is shown below:
Pn−X + Mtz−Y/Ligand ↔ Pn* + X−Mtz+1−Y/ Ligand
Pn* + Monomer → Pn−Pm
Cu-based ATRP, which was used in the present study, is the most versatile
technique in the field of “living” polymerisation7. A wide range of polymers such as
acrylates, methacrylates, styrenes, acrylonitriles, acrylamides and vinylpyridines and
their copolymers have been synthesised using Cu-based catalyst and ligands with
amine, pyridine and imine structures6-10.
ATRP polymerization procedures were based on the work of Matyjaszewski7.
Styrene (St) was homopolymerized and co-polymerized with 2-(trimethyl(silyloxy)ethyl
methacrylate (HEMA-TMS) and glycidyl methacrylate (GMA). Ethyl-2-
bromoisobutyrate and CuBr were used as an initiator and a catalyst respectively. Two
different ligands were used, 2,2’-bipyridine (Bipy) and tris(2-
dimethylaminoethyl)amine (Me6-TREN). The advantage of Me6-TREN, compared to
bipyridine, is that it is homogeneous. A complex is created with CuBr which is soluble
in the polymerization solution. This is very important in the case of grafted copolymers
where the system is already heterogenous because of the solid substrate. All materials
were supplied by Aldrich and used without future purification unless otherwise
indicated.
15
Monomer(s), initiator, catalyst, ligand and solvent were mixed in appropriate
ratios and placed in the vials sealed with a septum. Where Bipy was used, the ratio of
monomers (M) : initiator (I) : catalyst (C) : ligand (L) was 200 : 1 : 1 : 2.5. When Me6-
TREN was used the ratio M : I : C : L was 200 : 1 : 1 : 1. The solution was degassed for
approximately half an hour with argon. The vials were then placed in an oil bath set
either to 90oC or 110oC. The polymerisation conditions are shown in Table II-1. After a
certain period of time, as shown in the Table II-1, the vials were removed from the oil
bath and cooled. The formed polymers were diluted in chloroform (1:1) and passed
through a small column filled with alumina to remove the catalyst. The polymers were
then precipitated in methanol and dried in an oven at room temperature under vacuum.
Table II-1: Copolymerisation conditions
Monomers Monomer
ratio
Ligand Solvent Temp Time
GMA/St 1/9 Me6-TREN toluene 110oC 4hrs
2/8 Bipy bulk 90oC ON*
HEMA-TMS/St 1/9 Me6-TREN toluene 110oC 9hrs
2/8 Bipy bulk 90oC ON*
* overnight
16
2.1.2 Chemical Modification of Polystyrene Chain
Many difficulties were associated with the preparation of styrene/aminostyrene
copolymer with an end-functional group from monomers using the methods described
in chapters 2.1.1.1 and 2.1.1.2. Due to the reactivity of aminostyrene monomer, the
amino group had to be protected by reacting with di-t-butylcarbonate (Boc2) prior to
copolymerisation (this procedure is not described in this thesis but can be found
elsewhere11). The side reactions of the Boc2 protecting group, incompatibility between
the monomers and inability to find a suitable solvent that was compatible with
monomers, homopolymers and copolymer caused further difficulties in purification,
handling and utilisation of the resultant mixture.
This led to a search for a method for the modification of existing end-
functionalised polystyrene chains. There is a well-established method that involves the
bromination of phenyl rings12. This relatively simple and quantitative method, described
below, allows the introduction of bromine functional groups along the chain that can be
further converted to the desired functionality.
2.1.2.1 Bromination of Polystyrene
Polystyrene (PS)/ brominated polystyrene (PS(Br)x) blends have been used in
many studies, including the miscibility and phase separation of polymer-polymer
blends12, polymer adsorption to solid substrates13 and surface segregation14. The Flory-
Huggins parameter that governs the interactions between polymers can be varied easily,
which makes the PS/ PS(Brx) blend an ideal system for such studies.
The bromination reaction is relatively easy and does not change the
polydispersity of a given polymer12, which makes it a perfect starting point for chemical
modification and/or functionalisation of PS chains. Bromination of phenyl rings is an
example of electrophilic aromatic substitution. Bartulin15 showed that under the
specified conditions in the absence of light the bromination occurs exclusively on the
aromatic ring at the para position (Fig. 2-4).
17
Figure 2-4: Bromination of polystyrene.
In the current project, the synthesis of PS(Br)x was carried out following the
procedure of Kambour and Bendler12. Polystyrene of MW = 200,000 g/mol and
polydispersity 1.06 was purchased from Polysciences, Inc. and used without further
treatment. Polymer was dissolved in nitrobenzene at a concentration of 7wt%. The
polymer solution was placed in a test tube equipped with a magnetic stirrer and closed
with a rubber septum. The tube was covered by aluminium foil to prevent the light-
catalysed free radical reactions that would produce backbone bromination. The desired
amount of bromine was added to the polymer solution and left under stirring for 24
hours at room temperature. Excess bromine was then neutralised by addition of a few
drops of 1-pentene. The quenched reaction mixture was then slowly poured into
methanol with high stirring. The precipitated polymer was filtered, washed with more
methanol, dried in a vacuum oven at 80oC and dissolved in toluene. This process was
repeated three times to remove any residual nitrobenzene and other contaminants.
The amount of bromine was calculated from the following equation:
xMMmEm Bro
PS
PSBr 2
= ,
where E is the excess factor for bromine stoichiometric amount (E = 1.20, i.e. 20%
excess), PSm is the mass of polystyrene, oPSM is the molar mass of polystyrene repeat
unit, 2BrM is the molar mass of Br2 molecule and x is the desired mole fraction of 4-
bromostyrene units in the chain.
18
As found in the previous work of Genzer14 and Oslanec13, the experimentally
determined bromination levels (by Quantitative elemental analysis) agreed well with the
degrees of bromination calculated from the formulae shown above.
2.1.2.2 Conversion of Brominated Polystyrene
The second step was to introduce amino groups along the PS chain. A method
published by Lee16 for converting aryl halides to the anilines using lithium
bis(trimethylsilyl)amide (LiN(SiMe3)2 or LiHMDS, Aldrich) and palladium-based
catalysts was adapted. This method is based on the Heck reaction, the palladium-
catalysed arylation and alkenylation of alkenes, discovered by Heck17 in 70's. This
reaction is frequently applied in the metal-catalysed C-C bond forming process. In this
study as well as in the study conducted by Lee at al.16, the Heck mechanism was used to
form C-N bond (Fig. 2-5).
Figure 2-5: Formation of C-N bond using Heck reaction.
19
Figure 2-6: Introduction of –NH2 groups along the PS chains.
This procedure was used to introduce amino groups along the polystyrene chain
(Fig. 2-6) as follows: Brominated polystyrene was dissolved in dry toluene and oxygen
was removed by bubbling nitrogen through the solution for 1/2 hour in a vial closed by
a septum. In an argon filled glove box, LiHMDS (20% solution in THF) and catalytic
amounts (0.05 mmol - 0.002mmol) of tert-butyl phosphine (P(t-Bu)3, 10wt% in hexane)
and bis(dibenzylideneacetone)palladium (Pd(dba)2, powder) were added to the polymer
solution (all chemicals supplied by Aldrich). The mixture was stirred for 24 hours.
Several drops of 1M methanolic HCl were then added and the mixture was stirred for
several minutes to convert the amino silane group to ammonium chloride (-NH3+Cl-).
Then the polymer was poured into a large excess of diethyl ether and isolated by
filtration. The solid was redispersed in THF/MeOH (9/1) solution and 1M methanolic
KOH was added18 to produce the free amino group (-NH2). The modified polymer was
precipitated into distilled water containing a few drops of triethylamine19, filtered and
dried in the oven at 60oC under the vacuum. The resulting aminopolystyrene was
analysed by GPC and NMR.
20
2.2 Polymer Analysis
Spectroscopic and chromatographic techniques are the main techniques for
polymer characterisations. In this study, Gel Permeation Chromatography (GPC) and
Nuclear Magnetic Resonance (NMR) techniques were used for structural and chemical
analyses. GPC was used to determine molar mass of modified polystyrene, to assure
that the chemical reactions involved caused no chain degradation. NMR was used to
monitor the product of each step during the chemical modification procedure.
2.2.1 Gel Permeation Chromatography
Gel Permeation Chromatography (GPC), also called size exclusion
chromatography (SEC), is a standard technique used for molar mass analysis of
macromolecules. The main advantages are that the analyses are relatively fast, only
small amounts of material are needed, and the equipment is fully automated. Another
important advantage of GPC over other methods used for molar mass analysis
(osmometry, light scattering, viscosity measurement etc.) is that it also gives molar
mass distributions. This is used for example in fractional GPC, where polydisperse
substances are separated into monodisperse fractions of different molar mass.
GPC is a special case of the liquid chromatography process, where the
stationary phase (a porous cross-linked gel) is packed in a chromatographic column and
swollen by a mobile phase (eluent). The gel consists of spherical beads with a very
small diameter and a narrow distribution size. The size of the pores should be of the
same order of magnitude as a size of particles to be separated. A dilute solution of the
polymer, dissolved in an appropriate solvent, is flushed through the stationary phase
where the molecules are separated according to their size. There must be no interactions
between the eluent and the stationary phase of the column. The separation is a diffusion
driven process. The accessibility of the pores by diffusion of different molecules is a
function of both molecular size and pore size. Retention and elution of particles in the
column is given by their hydrodynamic volume (equations 1 and 2), which can be
related to molar mass through the intrinsic viscosity of the material. Smaller molecules
penetrate deeper into the pores of the gel and therefore are eluted after the larger
21
molecules. Molecules whose average hydrodynamic radius is larger than that of the gel
pores cannot be separated.
Hydrodynamic volume Vh = (4/3)πRh3 , (1)
where Rh is
Hydrodynamic radius Rh = kbT/6πηD, (2) where kb = 13.81 x 10-24 J/K is Boltzmann's constant T [K] is temperature η is intrinsic viscosity D [m2/s] is diffusion coefficient
GPC is a relative method. Polymer standards with known molar mass and low
polydispersities are used to calibrate the instrument. The data are collected in the form
of an elution chromatogram (Fig. 2-7), which is converted into molar mass distribution
via a calibration curve. The profile of the elution curve is determined by the molar mass
distribution of the sample. The data from the chromatogram are analysed using
computer software to give number-average molar mass (Mn), weight-average molar
mass (Mw), Z-average molar mass (Mz) and polydispersity (PD).
Figure 2-7: Schematic example of a chromatogram. Xi is an amount of material eluted,
Mi represents different fractions.
Mi Time, Volume
X
Mn Mw
Mz
22
The precise correlation between elution volume and molecular size cannot be
calculated. Therefore, each column must be calibrated with polymer standards of known
molar mass (Fig. 2-8).
Figure 2-8: An example of a calibration curve [www.sdk.co.jp].
The operating range lies in the linear region that can be expressed as:
Log M = a + bVe ,
where a and b are system-specific constants.
If the calibration standards are not available, a universal calibration curve can be
used (Appendix A)20. This universal curve is based on the fact that the structure of the
polymers does not affect Ve, and that Ve is governed only by the product of the intrinsic
viscosity [η] and M for a given set of columns. Therefore:
[η]1 × M1 = [η]2 × M2 (Ve,1 = Ve,2)
It has been confirmed with different polymers that at a certain elution volume Ve, the
product ([η] x M) is constant21. The intrinsic viscosity [η] relates to molar mass M
through the empirical Mark-Houwink relationship:
[η] = KMa ,
where K and a are tabulated constants.
23
Knowledge of the two coefficients K and a is necessary in order to apply the universal
calibration method.
The GPC system consists of a chromatograph, a detector and a data acquisition
system. The liquid chromatograph is composed of a solvent reservoir with an in-line
filter, a high pressure pumping system with adjustable flow rate, an injector for sample
introduction, and a set of columns. A range of different packing materials is available.
Some examples are shown in Table II-2. The choice of columns is very important
because the gel type and its packing density determine the separation efficiency.
Usually, several columns with different pore diameter ranges are used to achieve a good
separation. The most common GPC detector is the differential refractometer that
provides a signal proportional to polymer concentration. The main advantage of this
type of detector is its non-specificity and quasi-universal application. However, it is
very sensitive to temperature variations so that the temperature has to be carefully
stabilised20. The data acquisition system converts the information from the
chromatogram to the average molar mass and molar mass distributions using the
calibration curve.
Table II-2: Examples of packaging materials for high resolution GPC 21
GPC analyses were performed using a SHIMADZU LC-10AT VP Liquid
Chromatograph equipped with 2 Waters polystyrene-packed Styragel columns (HR4
and HR2 4.6 x 300 mm) connected in series, and connected to a differential
refractometer mass detector. THF was used as the eluent at a flow rate 0.3mL/min, and
the instrument was calibrated using polystyrene standards from Polymer Laboratories.
24
Polymer solutions (1-2 mg/ml) in THF were prepared and filtered into GPC vials using
Millex 0.20 µm PTFE syringe filters.
2.2.2 Nuclear Magnetic Resonance Spectroscopy
NMR is a form of absorption spectroscopy that gives information about the
number of magnetically distinct atoms in a molecule. When a sample is exposed to a
magnetic field it can absorb electromagnetic radiation in the radio frequency region at
frequencies that are characteristic to the molecules in the sample.
Many nuclei have a spinning charge that generates a magnetic dipole along their
axis. 1H, 12C, 16O, 13C, 15N, 19F and 31P are examples of such nuclei. If such a nucleus is
placed in an external magnetic field, its spin aligns itself either along with or against the
applied magnetic field. In the magnetic field, the nucleus absorbs the radio waves of a
specific frequency, which causes it to flip from the lower to the higher energy state with
the opposite spin22. In other words, when the frequency of the external magnetic field
equals the frequency of the electric field generated by the spinning proton, then these
two fields can couple and the energy is absorbed causing a spin change. This is called
resonance. The energy absorbed equals the amount of energy between these two energy
states. A 1H NMR spectrum is a plot of the applied magnetic field strength as a function
of the intensity of the absorption by the individual protons in a molecule.
The position of the peaks also depends on the immediate electronic environment
of the absorbing proton. Chemically equivalent protons exhibit the same chemical shift
(δ). Protons with different chemical environment have different chemical shifts. This is
because the electrons in the molecule shield the proton slightly from the external
magnetic field. A reference signal of tetramethylsilane (TMS) is often used to determine
the chemical shifts in an analysed molecule. The chemical shift of the protons in TMS is
arbitrarily set to zero. TMS is used as a reference standard because it has 12 chemically
equivalent hydrogens which give a distinct peak.
The relative intensities (area under the peak) of signals are proportional to the
number of protons contributing to each signal. Each proton is also subjected to the
25
magnetic field generated by the adjacent protons. This phenomenon is referred to as
spin-spin coupling, and causes signal splitting. If a proton has n chemically non-
equivalent protons, its 1H NMR signal will be split into n + 1 peaks. Chemically
equivalent protons do not cause any splitting.
Another nucleus commonly used in NMR spectroscopy is 13C. Its natural
abundance is very low (only 1.1%) and its magnetic moment is only 1/4 that of the
proton, therefore the sensitivity of NMR detection is much lower that in the case of
proton NMR. Development of Fourier transform instruments made it possible to detect 13C and made 13C NMR spectroscopy a useful technique for determining the structure of
organic molecules and polymers.
Dissolving polymers in NMR solvents produces solutions of a high viscosity due
to the chain entanglements. The long-range motions of dissolved polymeric molecules
might be slow but the local segmental motions are usually rapid23. Therefore NMR is a
useful technique for probing the molecular structure of polymers and other
macromolecules.
In this study, NMR spectroscopy was used to monitor each step of the chemical
modification procedure of polystyrene (PS) to amino-polystyrene (PS(NH2)). 1H NMR
and 13C NMR spectra were collected using a Varian Mercury-300 MHz NMR
spectrometer. CDCl3 containing 0.03% of tetramethylsilane (TMS) as a reference signal
was used as the solvent. For 1H NMR, solution concentrations were 10-20 mg/ml, and
for 13C NMR the concentration was higher.
26
2.3 Results and Discussion
Three different paths to prepare copolymers with desired functionalities were
explored in this work - NMP, ATRP and the chemical modification of an existing
polymer chain. Even though all three methods were more less successful in preparing
the free copolymers, NMP and ATRP did not produce the satisfactory results when
applied to the technique of grafting these copolymers from the silicon surfaces.
Therefore, only a brief description of these results is given. On the other hand, the
method of chemical modification of the polystyrene chain was applied successfully in a
grafting to technique (Section 3.3) and therefore is discussed in more detail.
2.3.1 Copolymers prepared by NMP
The PS monomer, PS/PHEMA and PS/PS(NH2) copolymers prepared by NMP
(see Section 2.1.1.1) were analysed using GPC and NMR. GPC analysis, including
molecular weight determination, and NMR spectra of PS/PHEMA and PS/PS(NH2) can
be found in Appendix B. Copolymerisation of styrene/4-aminostyrene turned out to be
very complicated due to a limited solubility of 4-aminostyrene in styrene monomer.
Also, the molecular mass of the resulting copolymer was much lower then the
theoretical molecular mass calculated from the monomer/initiator ratio. Somewhat
better results were achieved by copolymerisation of BOC protected 4-aminostyrene
with styrene shown in Appendix C.
2.3.2 Copolymers prepared by ATRP
The PHEMA-TMS/PS and PGMA/PS copolymers were analysed using GPC
and NMR. For PGMA/PS copolymers, the molecular mass ranged from 4K to 54K with
polydispersities (PD) between 1.12 – 1.62. The molar mass of PHEMA/PS copolymers
varied from 10K to 37K with PD between 1.18 – 1.27. A summary of results and
reaction conditions can be found in Appendix D.
27
X = 0
2.3.3 Analysis of Brominated Polystyrene by 13C-NMR Spectroscopy
The bromination of polystyrene was monitored by 13C-NMR spectroscopy. The
starting material was polystyrene with dimethylchlorosilane end-functional group (PS-
SiClMe2, Polysciences) of molecular mass MW = 8000. Several polystyrene samples
with an increasing degree of bromination (x = 0.1, 0.3 and 0.5) were prepared following
the procedure described in Section 2.1. The 13C NMR chemical shifts of pure
polystyrene and partially brominated polystyrenes are shown in Figures 2-9(a-d).
Figure 2-9a shows 13C NMR (CDCl3) δ 145 (C1, carbon atom number 1), δ 128
(C2,3,5,6), δ 125 (C4), δ 42-46 (C7), δ 40 (C8). A peak around δ 119.2 (C4-Br) starts to
emerge with x = 0.1 (Fig. 2-9b). With a higher degree of bromination the intensity of
this peak increases (Fig. 2-9c,d). These results are in agreement with the findings of
Farrall and Fréchet24 in their study of halogenated polystyrenes. The samples were
analysed by Prof. Andrew Whittaker from CMR at the University of Queensland. The
13C solution-state NMR spectra were acquired on a Bruker DRX500 spectrometer,
operating at 500.13 MHz and 125.77 MHz at 1H and 13C, respectively. Samples were
dissolved in deuterated chloroform to a concentration of 5-10 wt.%. Spectra were
acquired using inverse gated decoupling, with a sweep width of 27.8 kHz and
acquisition size of 64k to make a total acquisition time of 1.18 s. The 90o pulse time was
9.2 µs. The recycle delay was 15 s, and for each sample 2048 scans were co-added to
improve the signal-to-noise ratio.
Figure 2-9a: 13C-NMR spectrum of pure PS-SiClMe2.
HC C
H21
2
3
45
6
78
x
28
X = 0.1
X = 0.3
Figure 2-9b: 13C-NMR spectrum of partially brominated PS(Br)-SiClMe2.
Figure 2-9c: 13C-NMR spectrum of partially brominated PS(Br)-SiClMe2.
HC C
H2
Br
HC C
H21
2
3
45
6
78
x y
29
X = 0.5
Figure 2-9d: 13C-NMR spectrum of partially brominated PS(Br)-SiClMe2.
The relaxation time of polymer chains in NMR analysis takes much longer than
for typical low molecular mass organic molecules, and this affects the final resolution.
In this study, a reasonable resolution was required in order to compare the various
degrees of bromination, therefore, short PS chains [MW=8000] were used.
In the following study of the conversion of PS(Br) to PS(NH2), 1H-NMR
spectroscopy was used to detect the presence of the amino groups. This was a
qualitative rather than quantitative study, therefore longer PS chains [MW=200,000]
were used. The use of the long chains has a great advantage in relation to the isolation
(precipitation) of the final product. Precipitation of short polymer chains is more
challenging, usually requiring special techniques such as decanting or centrifuging.
30
HC C
H2
Br NMe3Si
NH2
(I) (II) (III)
SiMe3
(IV)
n1
2
3
4
6
5
7 8
n n n
9
9
2.3.4 Monitoring Conversion of PS(Br) to PS(NH2) by 1H-NMR Spectroscopy
In the following experiment, the process of converting polystyrene to
aminostyrene, described in Section 2.1 was monitored by 1H-NMR spectroscopy. The
schematic of the conversion process is shown in Figure 2-10. The starting material was
a monodisperse polystyrene standard of MW = 200K, obtained from Polysciences Inc.
Figure 2-10: Conversion of PS to PS(NH2).
The 1H-NMR spectrum of pure PS (MW=200,000) is shown in Figure 2-11a.
Figure 2-11b shows the change in a shape of an aromatic polystyrene peak (δ = 6.05–
6.65 ppm) due to bromination in the para position (x = 0.1). In Figure 2-11c, the
broadening of the TMS peak (δ = 0 ppm) is due to the presence of the trimethylsilyl
amino group along the PS chain. Also, there is a new peak around δ = 3.4 ppm which
would suggest a partial conversion of the bis(trimethylsilyl) amine group (-N(SiMe3)2)
to the amino group during the precipitation of the polymer into methanol. The full
conversion of PS-N(SiMe3)2 to PS(NH2) is demonstrated in Figure 2-11d. The spectrum
shows the disappearance of the trimethylsilyl group (δ = 0 ppm) and the presence of the
amino group (δ = 3.4 ppm). The final PS(NH2) was only partially soluble when re-
dissolved in THF for further precipitation, but it dissolved completely in DMF. There
was no difference observed in the 1H-NMR spectra of these separate fractions.
31
Figures 2-12a,b show the effect of D2O addition, and confirm the presence of
the amino group. A few drops of D2O were added to an NMR tube containing a sample
of PS(NH2) dissolved in CDCl3, and the solution was shaken. In the presence of D2O,
the hydrogen atoms are exchanged for deuterium atoms, and as a result, the –NH2 peak
of δ = 3.4 ppm disappears from the 1H-NMR spectrum. This is an indirect method used
to confirm the presence of amino groups in the organic molecules25. The 1H-NMR
spectra were collected using Varian Mercury VX 300 MHz NMR spectrometer.
Figure 2-11a: 1H-NMR spectrum of PS (MW=200K).
Figure 2-11b: 1H-NMR spectrum of PS(Br)x (x = 0.1).
32
Figure 2-11c: 1H-NMR spectrum of PS-N(Si(Me3)2).
Figure 2-11d: 1H-NMR spectrum of PS(NH2).
33
Figure 2-12a: 1H-NMR spectrum of PS(NH2).
Figure 2-12b: 1H-NMR spectrum of PS(NH2) with D2O added.
34
Thus the conversion of polystyrene to PS(NH2) was confirmed by NMR
spectroscopy techniques. The 13C NMR and 1H NMR spectra show each step of the
conversion process. For polymers, the T1 relaxation time is very long compared to that
for smaller organic molecules. Therefore, a dead time set for each analysis needs to be
long in order to obtain quantitative analysis with a sufficient resolution. In the present
case, only qualitative analyses were required. The 13C NMR spectra of brominated
polystyrenes show a clear trend of increasing intensity of the 119.2 ppm peak with an
increasing degree of bromination along the polystyrene chain. The position of this peak
is in agreement with the findings of Farrall and Fréchet24 in their work on halogenated
polystyrenes. The authors showed that the peak δ = 119.8 ppm corresponds to the
hydrogen atom in a para position. Therefore the presence of this peak is a confirmation
that the bromination occurs in para position. The conversion of PS(Brx) to PS-
N(SiMe3)2), the disappearance of bis(trimethylsilyl) group in the next step, and the final
formation of the amino group are evident in the 1H NMR spectra.
35
2.4 References
1 C. J. Hawker, Acc. Chem. Res. 30, 373-382 (1997). 2 C. J. Hawker, G. G. Barclay, A. Orellana, J. Dao, and W. Devonport,
Macromolecules 29, 5245 (1996). 3 G. G. Barclay, C. J. Hawker, H. Ito, A. Orellana, P. R. L. Malenfant, and R. F.
Sinta, Macromolecules 31, 1024-1031 (1998). 4 M. Husseman, E. E. Malmström, M. McNamara, M. Mate, D. Mecerreyes, D. G.
Benoit, J. L. Hedrick, P. Mansky, E. Huang, T. P. Russell, and C. J. Hawker,
Macromolecules 32, 1424-1431 (1999). 5 E. E. Malmström and C. J. Hawker, Macromol. Chem. Phys. 199, 923-935
(1998). 6 K. Matyjaszewski and J. Xia, Chem. Rev. 101, 2921-2990 (2001). 7 K. Matyjaszewski, B. Gobelt, H. Paik, and C. P. Horwitz, Macromolecules 34,
430-440 (2001). 8 S. Perrier, D. Berthier, I. Willoughby, D. Batt-Coutrott, and D. M. Haddleton,
Macromolecules 35 (2002). 9 X. Zhang, J. Xia, and K. Matyjaszewski, Macromolecules 31, 5167 (1998). 10 X. Zhang and K. Matyjaszewski, Macromolecules 32, 1763 (1999). 11 V. L. Covolan, G. Ruggeri, and E. Chiellini, J. Poly. Sci., Part A: Poly. Chem.
38, 2918 (2000). 12 R. P. Kambour and J. T. Bendler, Macromolecules 19, 2679-2682 (1986). 13 R. Oslanec, Thesis, UPENN, 1997. 14 J. Genzer, Thesis, UPENN, 1996. 15 J. Bartulin, G. Cardenas, and H. Maturana, Hydrometallurgy 8, 137 (1982). 16 S. Lee, M. Jorgensen, and J. F. Hartwig, Organic Letters 3, 2729-2732 (2001). 17 R. F. Heck and J. P. Nolley, J. Org. Chem. 37, 2320 (1972). 18 V. C. Gibson and W. Reed, in European Patent Application (UK, 1998), p. 1-9. 19 K. Suzuki, K. Yamaguchi, A. Hirao, and S. Nakahama, Macromolecules 22,
2607-2611 (1989). 20 J. L. Viovy and L. Lesec, Advances in Polymer Science, Vol. 114 (Springer-
Verlag Berlin Heidelberg, Berlin, 1994). 21 E. Schroder, G. Muller, and K.-F. Arndt, Polymer characterization (Hanser
Publishers, Munich, 1989).
36
22 R. M. Silverstein, G. C. Bassler, and T. C. Morril, in Spectrometric
Identification of Organic Compounds (John Wiley & Sons, Inc., New York,
1991). 23 J. E. Mark, Physical Properties of Polymers Handbook (AIP Press, New York,
1996). 24 J. M. Farral and J. M. J. Frechet, Macromolecules 12, 426-428 (1979). 25 S. Pyne.
CHAPTER 3
Polymer Chains
as Coupling Agents
39
3.1 Coupling Agents
Coupling agents are the adhesion promoters used to improve the adhesion between
organic polymers and inorganic surfaces, and to maintain this improved adhesion for a
period of time in various environmental conditions, including moisture or water and heat.
Glass fibre–resin composites are the classical examples (Fig. 3-1) of the use of such
materials. Coupling agents are often used as primers, applied as 0.1 – 10 µm thick layers
during different stages of the pretreatments to improve the performance of the bonded
components. The main roles of the primer are to establish a strong initial bond, and to
increase the service life of the adhesive joint by protecting the substrate from hydration and
corrosion. The result is a strong and moisture resistant interfacial bond. Coupling agents
can be also used as a surface finish (coating) or a surface modifier. The thickness of these
layers varies from one to several monolayers. They are also used as an additive.
Organosilanes, organozirconates, organozirco-aluminates and chrome complexes are some
examples of coupling agents used in adhesives technology1. Zirconate-based primers are
used for pretreatment of polyolefin films prior to printing. Zircoaluminates are applied as
primers for coating fillers in resins and polymers. Chrome complexes are used for example
for increasing the strength and durability of aluminium/polyethylene interfaces.
3.1.1 Commercially-available Coupling Agents
The most commonly used coupling agents are organosilanes. They are primarily
used to promote adhesion between metals, oxides, glass and polymers (resins, coatings,
adhesives etc.). Organosilanes are hybrids of silica and organic polymers, making them
ideal candidates for the bonding together of organic resins and mineral surfaces. Their
history goes back to the 1940’s when they were developed to pretreat glass fibres for fibre
composites, to increase the stability of the glass/polymer interface and to improve the water
resistance of these composite materials1. The general structure can be written as RnSiX(4-n)2.
R is a nonhydrolysable organofunctional group that reacts with organic resins and
40
polymers. X is a hydrolysable group which reacts with an inorganic substrate. Some of the
widely used silanes are amino-functional, epoxy-functional, mercapto-functional, carboxyl-
functional, hydroxyl-functional and vinyl-functional silanes. Epoxy-functional and
carboxyl-functional silanes are excellent coupling agents for thermosets like epoxies,
phenolics, melamines and urethanes3.
Figure 3-1: Schematics of coupling mechanism using coupling agents.
Commercially available silanes can be classified according to the type (-chloro, -
methoxy, -ethoxy) and functionality (mono-, di-, tri-) of the hydrolysable group X (Fig. 3-
2), non-hydrolysable group R (amino-, epoxy-, vinyl-, chloro) (Table III-1) and the chain
length (Fig. 3-3). The choice of silane depends on the particular application. The reactivity
of the hydrolysable group decreases in the order: chlorosilane > methoxysilane >
ethoxysilane. Monofuctional silanes yield a single monolayer, but the layer tends to lack
long term hydrolytic stability. In contrast, trifunctional silanes have high hydrolytic
stability but have a tendency to form multilayers. Silanes with two hydrolysable groups
form more flexible interfaces than trifunctional silanes. The chain length has a significant
effect on the way the silane molecules self-organise on the substrate. The longer the chain
the greater the degree of self-assembly or orientation due to the van der Waals forces acting
between the neighbouring chains. Park and Kim4 investigated the surface energies of four
Silicon oxide
EPOXY RESIN
41
different n-alkyl monochlorosilanes: trimethyl chlorosilane (C1), n-butyldimethyl
chlorosilane (C4), n-octyldimethyl chlorosilane (C8) and n-octadecyldimethyl chlorosilane
(C18). By using the dynamic contact angle measurement technique, they evaluated the
surface energies of these silanes deposited on glass surfaces. They also found that C1 and
C18 formed well organised layers whereas C4 and C8 formed more disordered layers. As
the alkyl chain length decreased, the surface molecules became more liquid-like with lower
surface density and coverage.
Figure 3-2: Functionalities of silane hydrolysable groups.
ClSi
Cl
Cl
H3COSi
OCH3
OCH3
EtOSi
OEt
OEt
ClSi
CH3
Cl
H3CSi
Cl
CH3
-trichlorosilane -trimethoxysilane -triethoxysilane
-dichlorosilane -monochlorosilane
42
Table III-1: Examples of silanes with different non-hydrolysable groups.
Cl(H3CO)3SiH2CH2C
H2C CHSi(OC2H5)3
H2NCH2CH2CH2Si(OC2H5)3
CH2O(CH2)3Si(OCH3)3
O
CH3CH2CH2CH2SiCl3
3-Aminopropyltriethoxysilane
3-Glycidoxypropyltrimethoxysilane
Vinyltriethoxysilane
4-Chlorophenylethyltrimethoxysilane
Butyltrichlorosilane
Figure 3-3: Examples of short and long chain silane molecules.
Organosilanes can be applied from solution (aqueous or non-aqueous), from vapour
phase or by plasma polymerisation. More details about these methods, including their
advantages and disadvantages can be found elsewhere5-10.
ClSi
Cl
CH2
Cl
CH3
ClSi
Cl
Cl
CH2
CH2
CH3
(OTS)
3
16
Butyltrichlorosilane Octadecyltrichlorosilane(n-BTMS)
43
3.1.2 Mechanism of coupling through silane coupling agents
Several theories have been advanced to explain the mechanism of adhesion between
organic polymers and hydrophilic inorganic surfaces3. They include Chemical Bonding,
Surface Wetting, Deformable Layers, Restrained Layer, and Reversible Hydrolysis, each of
which is briefly described below:
The Chemical Bonding Theory
As noted above, coupling agents contain hydrolysable end-functional groups (X)
that can react with a substrate. These are typically alkoxy (MeOSi≡), acyloxy (AcOSi≡),
amino (Me2NSi≡) or chloro (Cl3Si≡) groups. The surfaces of glass or silicon wafers with a
native silicon oxide layer have silanol groups (Si-OH) which are available for covalent
bonding with the end-functional groups of coupling agents.
According to Arkles2, this reaction involves 4 steps. Firstly, hydrolysis of X groups
attached to Si (Fig. 3-4). Water needed for hydrolysis is either added or is present on the
substrate surface. The reactive silanol groups are formed which leads to condensation of
the hydrolysed molecules into oligomers (Fig. 3-5). These oligomers bond to surface
silanol groups via hydrogen bonding (Fig. 3-6a). In the last step, which occurs during the
drying or curing process, covalent bonding with the substrate is achieved (Fig. 3-6b). If
the substrate is silica or glass, siloxane interfacial bonds (Si-O-Si) are formed. However, if
the substrate is metallic, the interfacial Si-O-metal bonds are susceptible to hydrolysis11.
The other end of the coupling agent has at least one different functional group that
reacts with a laminating resin during the curing process. This way, the coupling agent forms
a chemical bridge between two materials creating a strong interfacial bond (Fig. 3-1).
44
Figure 3-4: Hydrolysis of a silane coupling agent with three methoxy end-functional
groups.
Figure 3-5: Formation of oligomers and their condensation to the substrate.
Figure 3-6: a) H-bonding to the substrate, b) consequent covalent bond formation upon
curing.
MeOSi
OMe
R
OMeH2O
HOSi
OH
R
OH
HOSi
OH
R
OH
HOSi
OH
R
OH
HOSi
OH
R
OH
- 2H2OHO Si O
R
OH
Si O
R
OH
Si
R
OH
OH
OH OH OH+
HO Si O
R
O
Si O
R
O
Si
R
OH
OH H H H H H
O O O
HO Si O
R
O
Si O
R
O
Si
R
OH
O
- 2H2O
H
H O
b
a
45
Surface Wetting Theory
According to this theory, adhesion between the lower-energy organic resins and the
high-energy mineral surfaces is strong as the system has a tendency to achieve as low a
surface energy as possible. Complete wetting is necessary to achieve this interfacial
strength. Low viscosity silane coupling agents can be applied to assist the wetting of the
substrate. It was found however, that once the silanes bond to the substrate it is their
reactivity rather than their surface energy that has the greatest effect on the final interfacial
strength between the resin and the surface12,13. In other words, the reactivity of the silane
coupling agent with the polymer resin has greater significance than the polarity or
wettability of the treated mineral surface3.
Deformable Layer Theory
Due to a difference in thermal coefficients between the resin and the mineral
substrate, the resulting differential thermal shrinkage causes stress build-up along the
interface upon cooling. According to the deformable layer theory, when a silane coupling
agents is applied, a flexible interfacial layer is formed which may relieve some of this stress
without bond rupture3. The maximum toughness of such composites is achieved by this
deformable layer at the interface.
Restrained Layer Theory
This theory supposes that an interfacial region is formed with a modulus value
between that of the resin and the substrate14. It was suggested that when the silane coupling
agent is applied it tightens up the polymeric resin at the interface, which helps to achieve
maximum bonding strength and resistance to hydrolytic debonding. The maximum
chemical resistance of a composite is obtained with a restrained layer at the interface.
Reversible Hydrolysis Mechanism
This theory suggests that the bonds of the coupling agents to the substrate break and
reform reversibly in order to release the interfacial stresses15.
46
3.1.3 Nature of Bonding with Silane Coupling Agents
The silane coupling agent forms oxane bonds M-O-Si with various mineral surfaces
(M) such as oxides of silicon, aluminium, zirconium, tin, titanium, nickel, iron, boron and
carbon2. There is no obvious reason to expect a great improvement in water resistance
because some of these oxane bonds, as in the case of Si, Fe and Al, are hydrolysable3.
However, it was experimentally shown that stable bonds are formed with the oxides of Si,
Al, Zr, Sn, Ti, and Ni. Less stable bonds were formed with the oxides of B, Fe and C.
Alkali metal oxides and carbonates do not form stable oxane bonds2. More information on
bonding of the silane coupling agents to different surfaces can be found elsewhere16-20.
From a chemical point of view, the commercially available γ-
Aminopropyltriethoxysilane (γ-APS) would be a logical candidate for the bonding of a
silicon substrate and an epoxy resin because of its amino functionality. However, the nature
of aminosilanes is rather complicated. It has been shown that there is a strong hydrogen
bonding between silanol groups Si-OH of the substrate and the amine groups21 (Fig. 3-7).
Kanan et. al.22 showed, using infrared spectroscopy, that each molecule of (3-
aminopropyl)dimethylethoxysilane deposited on the silica powders from the vapour phase
formed surface bonds with two silanol sites. One Si-O-Si bond was formed through the
dimethylethoxysilane end-group, the second was a H-bond with the amine. As a result,
there were no free amino groups dangling from the silica surface. In this study, the fact that
not all amino groups might be available for chemical reaction with the epoxy resin would
result in incomplete information about the reactivity of the modified silicon surfaces.
Figure 3-7: Bonding of γ-APS to silicon substrate.
SiO
H2C
SiOO O O
H2C CH2
CH2
NH2
CH2
H2C
H2N
Si Si Si SiOH
OHSi
Sisilica
47
3.1.4 Self-Assembled Monolayers
There are two mainstream methods for the deposition of the organic molecules onto
a solid substrate in order to form organised monolayers: the Langmuir-Blodgett technique,
and the technique of self-assembled monolayers (SAM). The Langmuir-Blodgett technique
is a well established method for the preparation of physisorbed monolayers on solid
substrates. These monolayers are linked to the surface via physical bonds, and therefore are
not very stable when exposed to certain chemical environments. Recently this technique
was used by Devaux et al23,24 to form an initiator monolayer chemisorbed on the surface as
a precursor for TEMPO-mediated polymerisation of polystyrene brushes on an AFM tip24.
Sagiv25 developed a technique for preparation of SAM on planar surfaces by
adsorption of amphipathic molecules26. These monolayers are stronger and more
chemically resistant than Langmuir-Blodgett monolayers because they are attached to the
substrate via a chemical bond. As an example, the SAM were constructed on gold surfaces
via a thiol group27 and on silica surfaces via a trichlorosilane group28. The most typical
substrate for studying SAM is a silicon wafer with a native silicon oxide layer. The typical
thickness of this layer is around 18Å, and the surface has smoothness on an atomic level.
This enables surface property modification using SAM without any roughness effects.
Trichlorosilane coupling agents can be used on any silica-type surfaces such as glass and
metallic oxides. The widely assumed mechanism of the silanation reactions is shown in
Figure 3-8.
The silica surfaces are hydrated, and the silanol groups are available for chemical
reaction with the trichlorosilane functional groups. The suggested steps of the silanation
reaction are 7:
1. Physisorption - the polar trichlorosilane molecules are strongly attracted to the
hydrolysed silica surface.
2. Hydrolysis – at the surface, the silane groups are hydrolysed and they form
hydrogen bonds to the silanol groups on the silica substrate.
48
3. Condensation – water is eliminated and each chain is anchored to the substrate and
to the adjacent chains, creating a network.
Figure 3-8: Schematics of the silanation reaction7.
49
Such layers can be very stable and well-oriented. There are several important
factors influencing the final quality of the SAM such as temperature that affects the rate of
reaction as well as the solubility parameter of the trichloromolecules in a particular solvent.
Chain length, and a presence of an end group, have a profound effect on the surface quality
and the final composition of the resulting monolayer.
Sagiv suggests, in his paper on organised monolayers prepared by adsorption of
octadecyltrichlorosilane (OTS)25, that it is very unlikely that every OTS molecule would
form covalent bonds both with the substrate and the adjacent OTS molecules within one
monolayer. The more probable scenario is that some OTS molecules are covalently bonded
to the substrate and form H-bonds to the neighbouring molecules. The resulting monolayer
would be a combination of both chemically and physically adsorbed molecules, creating a
stable 2-D network of closely packed molecules. Therefore in the case of OTS, such
monolayers can be prepared on many types of polar surfaces even if density and
distribution of the surface OH groups vary. To prepare well-defined planar monolayers, it is
important to use anhydrous solutions. If there is more water present than just a surface
monolayer, the trifunctional silanes tend to condense in the bulk and form thick polymeric
films on the surface.
3.1.5 Mixed Monolayers
The studies reported in this thesis involved the adsorption of mixed monolayers
onto silicon wafers. In his work on adsorption of mixed monolayers, Sagiv25 describes the
properties of SAM prepared by deposition of long-chain saturated fatty acids, long-chain
substituted cyanine dyes and OTS molecules mixed at different combinations and ratios.
Cleaned glass slides were dipped in the solutions of these molecules in a mixture of
chloroform, tetrachloromethane and n-hexane. Deposition time ranged from a few seconds
to 10 minutes, at which point the layers were saturated. The saturation was confirmed by
absorption spectroscopy. The monolayer composition vs. solution composition was also
studied in the same work. By plotting the molar ratio of OTS/dye in a monolayer (RM) vs.
50
the molar ratio of OTS/dye in a solution (RS), Sagiv showed that RM values are
substantially smaller than the corresponding RS. For example, for RS = 50 the
corresponding value of RM = 16. However, when the same solution of OTS-dye was
applied to an aluminium substrate, RM was found to be larger than the corresponding RS 29.
Regarding the reproducibility of the resulting layers, the author devoted great care to the
reproducible cleaning of the glass surfaces. However, there was still a large variation in the
RM values reported even with glass slides from the same cleaning batch. This is a relevant
issue because the same was found in studies described in this thesis of mixed monolayers
deposited on silicon wafers. Sagiv concluded that the deposition of mixed monolayers is a
very complex and dynamic process. The final composition and properties of the final
monolayer depend on the interactions of the molecules in solution, and on the interactions
of the adsorbed molecules with a particular substrate. Although it is possible to prepare
mixed monolayers in a one step procedure, not any arbitrary monolayer composition might
be achievable. As the monolayers consist of both physisorbed and chemisorbed molecules,
their final composition is usually different from the solution concentration, and the
composition is a specific function of the particular solute-substrate system and therefore
difficult to predict.
51
3.2 Adsorption of Polymer Chains
In a dilute solution, the polymer chains are swollen and have a conformation of
random coils. As the chains adsorb at a surface, their conformation changes in order to
minimise the free energy of the surface. The excluded volume interactions cause the
random coils to expand. Jenkel and Rumbach30 first proposed three types of chain
conformations: trains (all segments are in contact with the surface), loops (no contact with
the substrate) and tails (free chain ends) (Fig. 3-9). Neutron scattering and neutron
reflectivity techniques have been used to determine the profiles of physisorbed
homopolymers, block copolymers and end-functionalised polymer chains. If there is a low
concentration of adsorbed polymers, the chains flatten in order to decrease the surface
energy. As the coverage and molar mass of chains increases, the length of loops and tails
increases. A structure with a large number of tails with approximately equal length is called
a polymer brush. The brush can be both physisorbed or chemisorbed, depending on the
nature of the functional groups.
Figure 3-9: Conformations of adsorbed polymer chains.
When polymer chains are deposited onto a solid impenetrable substrate, the polymer
chains undergo three different regimes in a solution. First, when the polymer chains are a
large distance from the surface in a very dilute solution, the chains are swollen, it is
assumed they are not interacting, and the distance between the molecules is greater than the
Tail
Loop
Train
Loop
Tail
52
Radius of gyration Rg of a single molecule. This concept is described in Section 3.3. Jones
and Richards found the concentration gradient to be non-uniform in this regime31. As the
chains approach the surface, they become less swollen and begin to overlap. This is the
regime known as a semi-dilute solution, where the chains interact strongly with each other
and the concentration fluctuations are much smaller. When the molecules adsorb onto the
surface, they are in the regime of a high concentration solution where there are no
fluctuations in the spatial concentration31.
A quantity characterising the adsorption process is called the surface excess z*.
(1)
where φ(z) is the volume fraction of polymers as a function of distance from the substrate,
and φ(∞) is the bulk volume fraction of polymer. In dilute solution, z* corresponds to the
adsorbed amount of polymer. A plot of the equilibrium amount of adsorbed polymer (z*)
vs. bulk solution concentration (φ(∞)) is called an adsorption isotherm. A typical example
of a high affinity adsorption isotherm is shown in Figure 3-10.
Figure 3-10: General adsorption isotherm of a polymer with a high affinity for a substrate.
Polymer segment/solvent molecule, polymer segment/surface and solvent/surface
interactions determine the amount of polymer adsorbed at the surface. In a good solvent the
total amount of polymer adsorbed increases with molar mass (M) for polymers of low M.
z*
φ(∞)
∫∞
∞−=0
* )()(( dzzz φφ
53
For polymers with high M the adsorbed amount reaches a plateau. There is no plateau in
adsorption from theta solvents and the amount of polymers increases continuously with M.
As a result, the adsorbed amounts in theta solvents are usually higher than in good
solvents32.
As noted above, the adsorption of polymer chains on a solid impenetrable surface involves
the loss in conformational and translational entropies due to the localisation of the chains
on the surface. The process might also involve a displacement of small molecules (solvent
molecules for example) from the surface. According to polymer adsorption theory32, a
polymer segment will adsorb if its adsorption (interaction) energy ( au2 ) is more negative
than that of the solvent molecule ( au1 ). A dimensionless interaction energy parameter χs for
the adsorption of polymer from solvent is defined as:
(2)
where au1 is the adsorption energy of solvent and au2 is the adsorption energy of a polymer
segment. The polymer adsorbs if χs is positive. There is a critical value of χs that a long
polymer chain has to exceed in order to adsorb. Below this critical value the entropy pulls
polymer chains away from the surface. This is due to the conformational and translational
entropy loss caused by a localisation of the polymer segment at the solid surface.
3.2.1 Competitive Adsorption of Polymer Chains
When more than one polymeric species competes for surface sites, this is referred to
as competitive adsorption. There are two main reasons for different polymers to compete:
1. Entropy effect: simultaneous adsorption of chains of broad range of molecular mass
2. Enthalpy effect: polymer chains contain segments of different adsorption energies
kTuu aa /)( 21s −=χ
54
In the first case, it has been found33 that longer polymer chains adsorb
preferentially, displacing shorter ones from the surface. This displacement is due to the
increased entropy by exchanging one long chain for a number of short chains. Hence free
energy is reduced, which thus favours adsorption of longer chains. This entropy tendency to
adsorb the larger chains is valid regardless of the polymer shape (linear, branched, flexible
etc). As a result, when polymer samples of high polydispersity are used the high molar
mass species adsorb selectively. In the second case, preferential adsorption occurs when
one of the polymers has a higher surface affinity, and therefore a more negative segmental
adsorption energy parameter. The au2 of a particular polymer with respect to the solvent can
be calculated using the self-consistent-field method (SCF). Enthalpy also acts when the
large amount of small molecules (monomers, solvent molecules etc.) are attracted to the
surface more strongly than the polymer segments which causes polymer desorption30. In
both cases, there might be a dynamic process of adsorption and desorption of some
molecules until the equilibrium is reached. The rate of exchange depends on the
temperature, molar mass and structure of adsorbing and desorbing chains 33. Adsorption-
desorption kinetics can be thought of as occurring in four steps: 1. diffusion to a surface 2.
adsorption onto a surface 3. surface rearrangements (conformational reorganisation) and 4.
displacement of polymer chains by other species. The first two steps are rapid whereas the
third and fourth steps are slow and history dependent33.
The field of competitive adsorption of various polymer species was extensively
researched by Frantz and Granick33-35. In studies of the enthalpic effects in competitive
adsorption of hydrogenated polystyrene (hPS), deuterated polystyrene (dPS) and carboxylic
acid terminated polystyrene (hPS-COOH) onto silicon oxide34, the authors used infrared
spectroscopy to determine the surface excess of these polymers. It was found that the
chemical structure of the polymer chains can outweigh the entropic tendency towards a
preferential adsorption of longer chains. In a non-competitive adsorption study of PS of
various M it was found that for fractions with M smaller then 115 000 g/mol the adsorption
was rapid, reaching a plateau within 10 minutes. Larger molecules reached equilibrium
within 2 hours. In a competitive adsorption experiment of a 1 : 1 mixture (by weight) of
hPS (12K) and dPS (12.5K), dPS displaced hPS almost completely within 15 minutes.
55
When higher molecular masses (> 500K) were used, hPS was not detectable on the surface.
However, when the molecular mass of hPS (12K) was larger than that of dPS (6K), the
protonated hPS adsorbed selectively. For polymers with higher M the equilibration time
was considerably longer than for samples of low M, usually around 2 hours. FTIR-ATR
measurements suggest that initially the species of low M occupy the surface. Then, as the
adsorption proceeds, the higher M chains start to displace the lower M chains until all or
almost all of the surface area is occupied by high M species. The authors also explored the
effect of polar chain end-functionality on kinetics of adsorption. In a competitive adsorption
using a 1:1 mixture of hPS-COOH (7K) and dPS (6K), carboxyl terminated chains
segregated to the silicon oxide surface almost completely, with almost no trace of dPS.
Even when the M of hPS-COOH was 2x higher than M of dPS, hPS-COOH still segregated
to the surface almost completely. Only when the mismatch between M of dPS (550K) and
M of hPS-COOH (7K) was very large, dPS adsorbed preferentially but the surface excess
of hPS-COOH was still significant, at 25% of the surface excess measured during a non-
competitive adsorption. The magnitude of the differential sticking enthalpy was calculated
to be 0.03 kT for dPS vs. hPS and 6.4 kT for hPS-COOH vs. hPS. These results showed
that even a small difference in chemical structure can have a major effect on surface
segregation (adsorption). In other words, even a small preferential adsorption per polymer
segment causes large preferential adsorption of the whole chain due to enthalpy.
In addition to the adsorption – desorption kinetic studies of various polymer species,
Schneider and Granick35 showed the effect of solvent quality on the interaction energy
parameter χs, defined as the difference in adsorption energy between a polymer segment
and a solvent molecule. In cyclohexane, which is a theta solvent for PS, the difference in χs
for hPS and dPS is equal to 0.03 kT. In CCl4, a good solvent, the difference in χs for hPS
and dPS is much smaller, around 2 x 10-3 kT.
56
3.2.2 Physisorption vs. Chemisorption
When the nature of the interactions between the polymer molecules and the
substrate is physical (H-bonds, Van der Waals, dispersion and dipolar forces), the
adsorption process is called physisorption. Examples of this process are the adsorption of
PS, PS-OH and PS-COOH onto a silicon substrate36. Bond strength of the adsorbed chains
can be expressed in terms of a surface sticking energy per monomer ε with units of [kBT].
In the case of hydrogen bonding, ε ~ 4 kBT which demonstrates that strong physical bonds
are quite common37. However, these bonds are reversible under certain conditions such as
exposure to a good solvent, displacement by other polymers or molecules in a process of
competitive adsorption. They can also be thermally unstable and prone to dewetting which
makes the surface layer non-homogeneous. When a chemical bond is formed with the
substrate, the process is called chemisorption. As shown before, silanes form chemical
bonds with solid surfaces and therefore silane-terminated polymer chains chemisorb38. The
adsorption of silane end-functionalised polymers with other functionality introduced along
the chain, for example bromine, therefore involves both physisorption and chemisorption
processes.
3.3 Tethered Polymer Chains
When free polymer chains are dissolved in a theta solvent, they assume the
Gaussian coil configuration, and their dimensions can be expressed in terms of the radius of
gyration Rg31:
Rg ~ aN 1/2 (3)
where N is a number of the repeat units in the polymer chain and a is a statistical segment
length. When the solvent is good and excluded volume interactions are taken into account,
where the individual polymer segments exclude other segments from the space pervaded by
the polymer segment, then the above expression becomes:
57
Rg ~ aN ν (4)
where ν is an excluded volume exponent. For dilute solution ν = 0.588.
When the polymer chains are tethered by one end to a solid surface, their
dimensions change. If the grafting density σ of polymer chains adsorbed to a solid surface
is low (average distance between grafting points > Rg), the polymers chains are isolated
from their neighbors and they adopt either ‘pancake’ or ‘mushroom’ conformation. The
pancake conformation is expected when the polymer chains have a high affinity for the
surface39. In that case the enthalpic polymer/chain interactions are stronger than the entropy
and the chains lie flat on the surface (Fig. 3-11a). When the polymer’s affinity to the
surface is very low, the polymer chains adopt the mushroom conformation (Fig. 3-11b).
Dimensions of these ‘mushrooms’ are similar to the radius of gyration of the free chains (Rg
~ N 3/5). As the grafting density σ increases, the neighboring chains start to overlap. The
excluded volume interactions cause chains to stretch away from the surface until the
entropy starts to dominate due to the changes in the chain configuration. The resulting
arrangement of polymer chains is called a polymer ‘brush’ (Fig. 3-11c). The equilibrium
thickness h of a polymer brush varies linearly with the degree of polymerisation40:
h ~ Naσ 1/3 (5)
where a is the statistical segment length and σ is the grafting density.
Figure 3-11: Pancake, mushroom and brush conformations.
b) Mushroom c) Brush a) Pancake
58
3.3.1 Grafting to vs. Grafting from Techniques
As mentioned previously, the polymer chains can be attached to a solid surface by
physisorption or chemisorption. Physisorption is a reversible process that involves physical
interactions, Van der Waals and hydrogen bonding, whereas chemisorption involves a
formation of the covalent bonds. There are two approaches to formation of the covalent
bonds: grafting to and grafting from. The grafting to technique involves a reaction of the
end-functionalised polymer chains with a suitable substrate. This method can be relatively
simple, but does not yield high density brush structures because the chains must diffuse
through the already-adsorbed polymer layer to reach the reactive sites on the surface. This
adsorption barrier increases as the thickness of the adsorbed layer increases. As a result, the
grafing to procedure produces a low density brush of low thickness41 as only small amounts
of the polymer can be deposited. For example, Zhao et al42 investigated ultra thin layers of
Si(OH)3-terminated polystyrene prepared by spin coating onto silicon wafers using AFM.
Geoghegan et al43 prepared triethoxysilane-PS brushes to study the kinetics of brush
penetration into a rubber network. Clarke et al38 studied the structure of grafted PS layers in
various polymers matrices. Various aspects of the grafting to mechanism have been
reviewed extensively elsewhere44-48.
Many researchers focused on developing various techniques for forming a brush-
substrate system using the grafting from approach. The main advantage of this technique is
that it is possible to prepare thick brushes with a grafting density much higher than that
possible using the grafting to technique. The basic principle of all these methods is the
deposition of surface initiators onto the surface, and then growing the polymer chains from
these sites using cationic, anionic or “living” free radical polymerisation techniques.
59
3.4 Techniques for Synthesis of Grafted Polymer Chains
Research focusing on tailoring the surface properties of inorganic materials using
chemically attached polymer brushes boomed in the 1990’s and it is still a growing field.
The methods of preparing the polymer brushes of various physical and chemical properties
in situ evolved dramatically. Initial research involved a multiple step process for generating
surface-imobilised initiator monolayers using azo compounds for the free radical
polymerisation of vinyl polymers41. This multiple step process, however, led to low
grafting densities of the surface attached initiator, incomplete conversions and side
reactions. As quantitative analysis of organic monolayers is very challenging41, therefore it
became impossible, without knowing the exact composition of the initiator layers, to
understand the mechanism of the subsequent polymerisation. Also, there was no simple
way to detach the grafted polymer chains from the surface for further analysis. Therefore
another approach was taken.
Prucker and Rühe41 developed a procedure for the attachment of a complete initiator
in a single reaction step, using azo initiators containing a chlorosilane group. These
initiators consist of an anchoring group (mono-, di- or trifunctional chlorosilane) which
links the initiator to the surface (silica gel), the initiator itself, and a cleavable group that
allows detachment of the polymer chains after polymerisation for further analysis. A use of
di- and trichlorosilane anchoring groups, however, resulted in crosslinked and disordered
silane multilayers, and the graft densities varied from experiment to experiment.
Structurally well organised monolayers were achieved by using the monochlorosilane
anchoring group. The grafting densities of this monofunctional initiator were varied by
varying the silane/silica ratio during the deposition procedure. Polystyrene chains were
grown from the surface-imobilised initiator by free radical polymerisation of styrene. The
polystyrene chains were then degrafted by a transesterification reaction carried out with
methanol/p-toluenesulfonic acid in toluene, and analysed by GPC and light scattering
techniques. Silica gel was used in this study because it has a higher surface area than planar
silicon wafer, and therefore a larger amount of material was available for analysis. The
60
major weakness of this method is the thermal instability of azo-based initiators. Another
drawback is that use of a traditional free radical polymerisation technique is suitable for
preparing homopolymer brushes only. It doesn’t allow the formation of polymer brushes
with accurately controlled structure, or copolymer brushes. These disadvantages can be
overcome by using “living” free radical polymerisation techniques as shown below.
3.4.1 Grafting of Polymer Brushes using NMP
The “living” free radical polymerisation techniques (NMP and ATRP), described in
Section 2.1, can be used for growing polymer brushes from solid substrates. First, the
surface active initiators are applied to the surface (Fig. 3-12) and then polymer brushes are
grown from these active sites (Fig. 3-13). Choice of monomers and their ratio to a bulk
initiator determines the final structure and properties of the grafted brush. This method has
been explored by Hawker et. al.49 in their work on the controlled synthesis of polymer
brushes.
Figure 3-12: Chemisorption of surface active initiators to Si substrate.
61
The aim of the present study was to use these brushes as coupling agents between a
cross-linked polymer (epoxy) and silicon substrates. Therefore it was necessary to vary the
concentration of the active sites on the substrate and the area density of polymer chains
grown from the substrate. An in-active initiator was synthesised by adapting the procedure
for the synthesis of an active initiator published elsewhere50. The schematics for both the
active and the in-active initiator synthesis can be found in Appendix E. These active and
inactive initiators were mixed at different ratios and chemisorbed to the silicon wafers from
the anhydrous toluene solutions. The polystyrene based chains were grown under NMP
conditions from the active sites (Fig.3-13).
Figure 3-13: a) deposition of mixed initiators; b) polymer brush.
Si Si Si Si Si
X X X
Active Initiator
Si Si Si Si Si
X X X
Polymer brush
a)
b)
62
3.4.2 Grafting of Polymer Brushes using ATRP
In this work, the approach was to attach to a silicon wafer a molecule with an active
functionality suitable for reaction with an ATRP agent. In this case γ-aminopropyl
trimethoxysilane (γ-APS) was used. The main advantage of this approach is that γ-APS is a
commercially available reagent, widely used in the industry, for example for bonding steel
to epoxy. In order to change the concentration of this active initiator, it was mixed with n-
butyltrimethoxysilane (n-BTMS) at different ratios (Fig.3-14a). This silane has a terminal
methyl group, and therefore is not able to attach itself to the ATRP agent. 2-
Bromoisobutyryl bromide was reacted with the surface bound γ-APS, forming a surface
attached ATRP initiators (Fig. 3-14b) from which the polymer brushes were grown. The
preparative conditions for polymerisation can be found in Table II-1, Section 2.1.1.2.
Figure 3-14: a) Deposition of γ-APS and n-BTMS on Si wafer
SiH3CO
OOCH3
H3C
SiH3CO
OOCH3
H2N
SiH3CO
OOCH3
H3Ca)
SiH3CO
OOCH3
H2N
Br
O
Br
+
SiH3CO
OOCH3
HN O
Brb)
63
b) ATRP agent attachment.
3.4.3 Grafting of Polystyrene Chains and their Functionalisation
Some of these procedures were reported in Section 2.1 where the process of the
functionalisation of free polystyrene was described. In this section, the process of “grafting
to” of the brominated silane, end-functionalised, polystyrene and the consequent conversion
to amino-functional polystyrene is described.
Brominated, silane-terminated, polystyrene [PSBr0.1-SiClMe2 (MW = 8000, x =
0.1)], mixed with PS-SiClMe2 in various ratios, was deposited onto silicon wafers from
toluene solution, forming the mixed monolayer (Fig. 3-15).
Figure 3-15: Chemically adsorbed PSBr and PS chains on Si wafer. The conversion of the bromine to the amino functional group is performed by
adapting a palladium-catalysed method of converting aryl halides to anilines using lithium
bis(trimethylsilyl)amide (LiN(SiMe3)2) (Fig. 3-16). The reaction is catalysed by
bis(dibenzylideneacetone)palladium Pd(dba)2 and tri-t-butylphosphine P(t-Bu)3 (Fig. 3-17)
and is known as the Heck reaction. Details of the preparation can be found in section 3.6.3.
Br
Br
Br
Br
BrBr
X NH2
R NMe3Si SiMe3
LiR+
1. Pd(dba)2/P(t-Bu)3 (1:1)
2. HCl, neutralisation
X = Br or Cl
64
Figure 3-16: Heck reaction using LiN(SiMe3)2.
a) b)
Figure 3-17: Catalysts for Heck reaction: a) Pd(dba)2 , b) P(t-Bu)3.
The conversion procedure was carried out in two steps. In the first step, the bromine
group was converted to the bis(trimethylsilyl)amide protective group (Fig. 3-18). The
second step consists of de-protection of -N(SiMe3)2, and conversion to –NH2 by
acidification and neutralisation (Fig. 3-19).
Figure 3-18: Conversion of the PSBr to the silane protected PS.
Figure 3-19: Conversion of the silane terminated PS to PS-NH2.
O
OPd P
C
C C
CH3 CH3H3C
CH3
CH3CH3
H3CH3C
H3C
NSiMe3Me3Si
Br
Li
NMe3Si SiMe3
nn
+Pd(dba)2/ P(t-Bu)3
NSiMe3Me3Si
NH3+Cl- NH2
NaOH
n
HCl
n n
65
3.5 Analysis of Grafted Surfaces
3.5.1 Ellipsometry
Ellipsometry is an optical technique used to analyse transparent thin films. It has
been used mainly in semiconductor research and fabrication to determine properties of
layer stacks of thin films and the interfaces between the layers. Ellipsometry is also utilised
by researchers from other fields such as biology, medicine and polymer coatings. The main
advantage of ellipsometry is that it is a non-destructive, highly sensitive technique with a
measurement range from monolayers to micrometers. The upper limit of film thickness
amenable to ellipsometric measurement is determined by the film uniformity and
homogeneity. The precision of the measurement is also given by the optical properties of
both substrate and the transparent film. To achieve the best possible resolution (units of
angstroms) the sample should be optically flat, non-scattering, highly reflecting, have
uniform film refractive index (homogeneous and isotropic), large differences between film
and substrate refractive indices, and the film should be transparent.
Figure 3-20: Single-film model. ϕ is the angle of the incident and reflected beams, n and k
are the real and imaginary parts of the refractive index of the substrate and the film, t is the
film thickness.
φ φ
λ
φ2φ2
Substrate: n3, k3
Film: n2, k2, t
Ambient
66
A highly collimated beam of linearly polarised light generated by a laser travels
through the transparent film/films to a substrate where it is reflected and analysed by a
photodetector. The case of a single-film model is shown in Figure 3-20. From the measured
difference between the states of polarisation of the incident and reflected beam, various
properties of the reflecting surface can be computed. Single wavelength ellipsometry can
measure only a film thickness and refractive index whereas Spectroscopic ellipsometers
(multiple wavelength beam) can analyse complex multilayered structures, interface
roughness, inhomogeneous and anisotropic layers.
Figure 3-21: Schematic of the analysing section of the AutoEl-II Ellipsometer.
A Rudolph Research AutoEl-II Ellipsometer was used to measure the average
thickness of the functionalised polystyrene chain deposited on Si wafers. The schematic
setup is shown in Figure 3-21. The AutoEl-II is a nulling type Ellipsometer. It is equipped
with a helium-neon laser generating collimated, monochromatic light at a wavelength 632.8
nm. The incident beam angle was set to 70o. A double-film model was used for the analysis
(Fig. 3-22).
Autocollimator
Eyepiece
Laser head
Polariser
Sample
Analyser
Adjustable stageCompensator slide
67
Figure 3-22: Double-film model. ϕ is the angle of the incident and reflected beams, n and
k are the real and imaginary parts of the refractive index of the substrate and the film, t is
the film thickness.
For a single film model and when the refractive index n of an analysed layer is
known, the equations below can be used to determine the film thickness t:
where: pR and sR = total reflection coefficients
pr12 , sr12 = the Fresnel reflection coefficients for the ambient medium-film
interface parallel and perpendicular to plane of incidence
pr23 , sr23 = the Fresnel reflection coefficients for the film-substrate
interface parallel and perpendicular to plane of incidence
φ φ
λ
φ1φ1
φ2
Substrate: n3, k3
Film: n2, k2, t
Film: n1, k1, t
Ambient
)cos(2
)exp)(exp1()exp1)(exp(exptan
22
22312
22312
22312
22312
φλπβ
ββ
ββ
nt
rrrrrrrr
RR
jssjpp
jssjpp
s
pi
=
+++++==Ψ −−
−−∆
68
In this study, the single film model was used for the thickness calculations. The
silicon oxide layer thickness on the silicon wafers was determined prior the surface
chemical modification. These thickness values were included in a software for calculating
the deposited polymer layer thickness. The refractive index of polystyrene n = 1.592 was
used for both pure and mixed polymer monolayers.
3.5.2 Surface Energy by Contact Angle Measurement
In the field of adhesion and adhesives technology, knowledge of the surface energy
of a given material can be used to predict its surface properties and its interactions with
other materials. Contact angle measurements provide a very sensitive tool for examination
of surface chemistry and surface homogeneity, which are used to monitor surface
treatments and surface cleanliness as well as surface wettability. The main advantages of
this technique are its simplicity, rapidity of analysis and inexpensive equipment.
A surface free energy of solids γ is a measure of the attractive intermolecular forces
between a surface layer and a liquid phase. As a drop of a testing liquid is placed on top of
a solid surface, it either spreads or retracts, depending on the interfacial energy between the
solid and the liquid. The initial spreading coefficient So can be defined in terms of the
difference between the solid surface energy (γsv) and the liquid surface tension (γlv) together
with the interfacial tension (γsl) as51:
So = γsv - γlv - γsl (6)
The liquid spreads spontaneously on the solid when So > 0 (Fig. 3-23a). When So < 0 the
liquid wets the surface only partially (Fig. 3-23b).
69
Figure 3-23: a) So > 0, b) So < 0. Θ is an equilibrium contact angle.
The Young equation relates the equilibrium contact angle Θ to all three components, γsv, γsl
and γlv:
γsv = γsl + γlv cos Θ (7)
When the solid surface energy is higher than the liquid surface tension, then the surface free
energy γsv of the substrate resulting from adsorption of vapour from the liquid may be
considerably lower than the surface free energy in vacuum (γs). In order to balance this
discrepancy, an equilibrium spreading pressure πe is introduced52:
γsv = γs - πe (8)
πe can be neglected in the case where a high surface energy liquid wets a low surface
energy substrate. In the present study, water and diiodomethane were used as the testing
liquids for polymer coated surfaces. These surfaces generally have lower surface energies
and therefore πe was neglected in further calculations.
The intermolecular forces present when two dissimilar materials are brought into contact
are expressed in the Dupré equation as the work of adhesion WA 52:
γsl = γsv + γlv - WA (9)
When the dispersive and polar forces are considered, then according to the theory of
fractional polarity53:
Θγsl γsv
γlv
Θ
γsl γsv
γlv
a) b)
70
γ = γd + γp (10)
where γd and γp are the dispersive and polar components of the surface energy. The polar
component of surface energy includes dipole forces, induction forces and hydrogen
bonding54.
Using a geometric mean assumption, for a liquid drop forming a contact angle (θ) on a
solid surface, the combined Owens-Wendt/Young equations 55,56 state that:
(11)
At least 2 testing liquids of different polarities are used for the contact angle measurement
when this method is used. Dispersive (γlvd) and polar (γlv
p) surface tensions for many testing
liquids are well known. The values for water and diiodomethane (the test liquids used in
this work) are shown in Table III-2.
A Ramé-Hart Model 100 Goniometer system was used in this study to measure static
contact angles using the testing liquids mentioned above.
Table III-2: Surface Tension of Test Liquids
Liquid γld (mJ/m2) γl
p (mJ/m2) γl (mJ/m2)
Water 21.8 51.0 72.8 Diiodomethane 48.3 2.5 50.8
Another method used by researchers to determine the composition of thin polymeric
films is X-ray Photoelectron Spectroscopy (XPS)57. In this study, however, XPS studies did
not provide satisfactory results due to lack of sensitivity.
+=Θ+
dlv
plvP
svdsvd
lv
lv
γγγγ
γγ
2)cos1(
71
3.6 Results and Discussion
3.6.1 NMP Results
The procedures for the synthesis of polystyrene brushes and their derivatives can be
found in work of Hawker et. al.49. Identical conditions were employed in the present work.
The molecular mass of the non-grafted (free) polymer generated during the synthesis due to
the addition of “free” alkoxyamine initiator was analysed by GPC. The results for PS and
PS/PHEMA brushes are shown in Table III-3.
Table III-3: PS and PS/PHEMA brushes
Monomers M1/ M2 M/Init MWcalc MW tbrush [Å] σ [# chains/m2]
Styrene 1/0 500/1 47K 26K 83 2.0x1017
Styrene/HEMA 9/1 200/1 20K 3K 51 8.5x1017
M1/ M2 ………molar ratio of monomers in a reaction solution
M/Init………..molar ratio of all monomers to initiator molecules
MWcalc ………theoretical molecular mass (MW = M x MWaverage + MTEMPO) x conversion
MW………….molar mass determined by GPC
tbrush………….thickness of polymer brush measured by Ellipsometry
σ……………..areal chain density
The areal chain density σ has been calculated from the following equation:
MWNt
AA××== ρσ # , (12)
where A is the sample area (typically 1 cm2)
72
ρ is polymer density (ρPS = 1.05 g/cm3)
t is the measured thickness of the polymer layer
NA is Avogadro’s number (6.023x1023)
MW is the molecular mass of the polymer
# is number of polymer chains
Hawker49 reported that the relationship between brush thickness and molecular mass
of free polymer is almost linear. Therefore it can be assumed that the molecular mass of the
covalently attached polymer chains is consistent with the molecular mass of the free
polymer. In comparison, the chain density σ = 5x1017 chains/m2 of the polystyrene brush
grown from silicon wafer reported by Hawker et. al49 is somewhat higher than the value
found in the present study. The chain density of PS/PHEMA brush was found to be slightly
higher, implying a more dense brush.
Table III-3 shows poor agreement between the calculated MW and the measured
MW for Styrene/HEMA copolymers. The refractive index detector used in this study is
very useful for determining molar mass of homopolymers. Introducing copolymer might
affect an elution volume Ve and therefore resulting MW would deviate more from the
calculated value. In this study, however, there was only 10 mole% of HEMA incorporated
into the copolymer and therefore it is more likely that the large difference between the
calculated and measured MW was caused by incomplete conversion during the
copolymerisation process resulting in copolymers with a short chain length.
The aim of the next experiment was to obtain PS brushes of various densities. The
mixtures with different ratios between the active and in-active initiators were prepared. The
initiators were deposited onto Si wafers from anhydrous toluene solutions and the surfaces
were analysed by ellipsometry and contact angle goniometry. The results are summarised in
Table III-4. The initiator synthesised at the University of Wollongong (UOW) produced
layers of the same thickness as the initiator supplied by IBM Almaden Research Centre but
the surface energy was lower, especially the polar component γinitiator (polar). The polystyrene
brushes were grown by NMP from these surface bound initiators. The chains grown from
73
the UOW prepared initiators produced the thickness and areal density of almost one
magnitude higher than the chains grown from the IBM prepared initiators. The reason for
the differences is not clear.
Table III-4: PS brush on Si wafer prepared by NMP
Active Init
[%]
tinitiator
[Å]
γinitiator(total)
[mJ/m2]
γinitiator (polar)
[mJ/m2]
tPS
[Å]
σ
[# chains/m2]
0 70.6 46.4 5.3 15.4 3.7x1016
10 a 14.2 66.6 25.2 23.8 5.7x1016
20 a 11.0 63.4 26.4 9.3 2.2x1016
100 a 16.6 56.2 22.2 10.8 2.6x1016
100 b 16.2 40.2 10.9 94.8 2.3x1017
a active initiator prepared at IBM Almaden Research Centre b active initiator synthesised at UOW
Even though the brush synthesis was successful, the chain densities were lower than
in the previous experiment and the results show that there is a poor control over the
polymer chain density.
3.6.2 ATRP Results
For grafting of PS, several mixtures of various ratios of γ-APS/n-BTMS in
anhydrous toluene were prepared and deposited onto the silicon wafers at room temperature
overnight. The reaction conditions are summarised in Table III-5. After Soxhlet extraction
to remove any non-reacted polymer, the final polymer layer was analysed by Ellipsometry.
The “free” polymer was analysed by GPC. As shown in Table III-6, the thickness and the
chain density increases with increasing γ-APS concentration. Samples with 100% γ-APS
concentration are an exception.
74
Table III-5: Grafting of PS chains
Sample % APS M:I:C:L Ligand Temp [oC] Solvent
I 10 200:1:1:2.5 Bipy 110 bulk
II 25 200:1:1:2.5 Bipy 110 bulk
III 50 200:1:1:2.5 Bipy 110 bulk
IV 100 200:1:1:2.5 Bipy 110 bulk
IM 10 200:1:1:1 Me6-TREN 110 bulk
IIM 25 200:1:1:1 Me6-TREN 110 bulk
IIIM 50 200:1:1:1 Me6-TREN 110 bulk
IVM 100 200:1:1:1 Me6-TREN 110 bulk
Table III-6: PS brush on Si wafer
Sample Thickness [Å] MW PD σ [# chains/m2]
I 77 8609 1.18 5.7x1016
II 261 9850 1.24 1.7x1017
III 701 9948 1.18 4.5x1017
IV 289 9753 1.22 1.8x1017
IM 101 21306 1.14 3.0x1016
IIM 251 26260 1.16 6.0x1016
IIIM 561 27759 1.13 1.3x1017
IVM 450 28458 1.16 1.0x1017
75
Because of inconsistency in PS growth, the coupling agent itself was investigated.
From the thickness measurements of the deposited γ-APS/n-BTMS layers (Fig. 3-24), it is
clear that for the lower concentrations of γ-APS the increase in the layer thickness is very
low. For the higher concentrations, however, the thickness increases dramatically due to a
thick multilayer formation. This corresponds to the findings of Prucker and Rühe41. The
authors report that the tri-functional silanes exhibit a tendency towards the formation of
surface attached networks. In addition, the amino groups can have strong interactions with
the appropriate sites at the surface. If such layers are used for additional surface reactions,
the structure of the resulting monolayer is difficult to analyse and reproduce41.
Figure 3-24: γ-APS thickness vs. γ-APS concentration deposited on Si wafer.
In the following experiment, poly(2-(trimethyl(silyloxy)ethyl methacrylate)
(PHEMA-TMS) and PS copolymer brushes were prepared in the same way as described
above for pure PS brushes (HEMA-TMS/St ratio was 1/9, Me6-TREN was used as ligand,
reaction was carried out in a bulk, at 110oC for 4 hours). The concentration of γ-APS and
the ATRP reaction time were varied. The final layers were analysed by Ellipsometry. The
results are summarised in Table III-7.
APS/BTMS
0
100
200
300
400
0 20 40 60 80 100
[%] APS
Thi
ckne
ss [Å
]
76
Table III-7: Thickness of PHEMA-TMS/PS brush
γ-APS [%] Time [h] Thickness [Å]
5 4 149
10 4 126
25 4 123
50 4 112
50 2 95
50 4 93
50 6 88
50 8 95
Neither γ-APS concentration or reaction time produced the expected increase in
thickness (and therefore increased chain density). As mentioned previously, it is very
challenging to measure quantitatively the composition of the silane layer. Therefore it is
uncertain if these results are due to the lack of control over the silane deposition or the
ATRP process itself.
77
3.6.3 Bromination of PS-SiClMe2 and Deposition of Mixed PSBr0.1-SiClMe2/PS-
SiClMe2 monolayers
Silane terminated polystyrene (MW=8000) was dissolved in nitrobenzene (7 wt%)
and put into a test tube wrapped in aluminium foil. The calculated amount of bromine was
added and the solution was stirred in dark for 24 hour. The brominated polymer was
precipitated into MeOH. Due to the low molecular mass of the initial polystyrene, it was
not possible to filter the precipitated polymer in the traditional way using filter paper,
therefore the polymer was left to settle at the bottom of a beaker, the solvent was decanted
and the remainder was left to evaporate. The polymer was dried at 75oC under a vacuum
overnight. This step was repeated at least twice.
Silicon wafers were cut into 1.5 x 7.5 cm pieces using a diamond scriber and a
scalpel. Each wafer was cleaned using chloroform followed by piranha (H2SO4/H2O2)
solution in an ultrasonic bath. The wafers were washed with tap water, de-ionised water,
acetone and toluene and then dried with a nitrogen flow. They were then exposed to
ultraviolet ozone cleaning system (UVO) treatment for 1 hour, to remove the last traces of
organic contaminants, and used for the solution depositions immediately.
Various ratios of PSBr(0.1)SiClMe2/PS-SiClMe2 were dissolved in dry toluene to
form 1% solution. Cleaned Si wafers were dipped in the above polymer solutions and
stirred over night. The next day the wafers were baked in the oven at 120oC for several
hours and then cleaned by Soxhlet extraction for at least 12 hours to remove any physically
adsorbed polymer chains from the surfaces. Surface energy (γ) and thickness (t) of
chemically adsorbed mixed polymer monolayers were determined using contact angle and
ellipsometry measurements. The contact angles were measured using water and
diiodomethane.
The brominated polystyrene chains were deposited onto the silicon wafers before
any further chemical modification, in order to “preserve” the reactivity of the polymer’s
78
silane end-functional group and also to prevent the reaction of the amino groups with the
silicon substrate.
Experiment 1
The 1% solutions of PSBr(0.1)SiClMe2/PS-SiClMe2 (1, 2.5, 5 and 10 wt% of
PSBr(0.1)SiClMe2) in toluene were deposited on Si wafers. The ellipsometry and contact
angle measurements (CAM) of the resulting layers are summarised in Table III-8.
Table III-8: Mixed PSBr(0.1)SiClMe2 and PS-SiClMe2 layers (Exp. 1)
PSBr [%] t [Å] σ [# chains/m2] γtot [mJ/m2] γdisp [mJ/m2] γpolar [mJ/m2] 1 9.1 7.2x1016 41.98 35.24 6.73 1 8.6 6.1x1016 41.88 35.05 6.83
2.5 9.6 7.6x1016 41.89 35.54 6.34 2.5 9.0 7.1x1016 41.72 33.81 7.9 5 7.3 5.8x1016 44.52 31.22 13.29 5 7.4 5.9x1016 43.41 30.81 12.6
10 6.4 5.1x1016 42.72 29.92 12.8 10 6.4 5.1x1016 43.38 29.08 14.3
The thickness of the PSBr(0.1)SiClMe2/PS-SiClMe2 layers was in the range 6Å – 10Å, which
corresponds to a chain density ranging from 5.1x1016/m2 to 7.6x1016/m2. These values are
in the same order of magnitude as the polystyrene brushes of similar molecular mass
prepared by ATRP (Section 3.6.2). The total surface energy of these layers was relatively
uniform, but the polar component of the surface energy increased with an increasing
amount of brominated polystyrene in the layer. The value of 14.3 mJ/m2 for 10%
PSBr(0.1)SiClMe2 represents a 100% increase compared to 6.7 mJ/m2 for 1%
PSBr(0.1)SiClMe2.
79
Experiment 2
The previous experiment was repeated for different ratios of PSBr(0.1)SiClMe2/PS-SiClMe2
2, 4, 6, 8, 10, 13, 16, 20 wt% of PSBr(0.1)SiClMe2). Results of the Ellipsometry
measurements and CAM are summarised in Table III-9.
Table III-9: Mixed PSBr(0.1)SiClMe2 and PS-SiClMe2 layers (Exp. 2)
PSBr [%] t [Å] σ [# chains/m2] γtot [mJ/m2] γdisp [mJ/m2] γpolar [mJ/m2]2 21.9 1.7x1017 43.98 42.86 1.12 4 13.3 1.1x1017 43.38 42.3 1.08 6 21.9 1.7x1017 43.69 42.78 0.91 8 23.3 1.8x1017 43.01 42.18 0.83
10 19.7 1.6x1017 42.58 40.71 1.88 13 15.8 1.3x1017 43.02 40.9 2.12 16 21.8 1.7x1017 42.91 41.14 1.77 20 24.4 1.9x1017 42.57 41.24 1.34
In this experiment, the values of the polar surface energy are scattered and lower than in the
previous experiment. The thickness of the PSBr(0.1)SiClMe2/PS-SiClMe2 layers was in the
range 13Å – 25Å, which corresponds to a chain density ranging from 1.1x1017/m2 to
1.9x1017/m2.
80
Experiment 3
1% toluene solutions of 2, 4, 6, 8, 10, 12, 16, 20 wt% of PSBr(0.1)SiClMe2/ PS-SiClMe2
were prepared and deposited on Si wafers. Table III-10 shows the thickness and the areal
chain densities of the resulting layers. The surface energies were not analysed during this
experiment.
Table III-10: Mixed PSBr(0.1)SiClMe2 and PS-SiClMe2 layers (Exp. 3)
PSBr [%] t [Å] σ [# chains/m2] 2 21.0 1.7x1017 4 15.7 1.3x1017 6 15.3 1.2x1017 8 9.6 7.6x1016
10 16.0 1.3x1017 13 18.6 1.5x1017 16 13.3 1.1x1017 20 14.7 1.2x1017
Experiment 4
This experiment was run under the same conditions as Experiment 3. Table III-11 shows
the thickness and surface energies of the resulting layers.
Table III-11: Mixed PSBr(0.1)SiClMe2 and PS-SiClMe2 layers (Exp. 4)
PSBr [%] t [Å] σ [# chains/m2] γtot [mJ/m2] γdisp [mJ/m2] γpolar [mJ/m2] 2 14.7 1.2x1017 44.83 31.56 13.27 4 16.9 1.3x1017 41.8 36.76 5.04 6 13.0 1.0x1017 44.93 32.43 12.5 8 11.3 8.9x1016 43.87 33.47 10.37
10 13.1 1.0x1017 41.58 36.04 5.54 13 11.3 8.9x1016 45.04 32.51 12.53 16 11.2 8.9x1016 45.89 31.77 14.12 20 15.4 1.2x1017 41.37 37.27 4.1
81
The PSBr and PS chains used in this study had the same length, therefore it was not
clear if the thickness of the grafted layers would change with an increasing concentration of
PSBr chains in the grafted layer. The trend of a decreasing thickness with an increasing
PSBr concentration could be related to the attractive forces between the silica surface and
bromine atoms on the PSBr chains. As mentioned earlier, due to these attractions, the
polymer chains have a greater tendency to lie flat on the surface. This flat conformation
prevents other chains from coming into a contact with the surface. As a result of this
behaviour, it would be expected that the thickness and the chain density would decrease
with increasing PSBr concentration. The results, however, do not support this expectation,
and the thickness varied slightly from experiment to experiment showing no trend.
The next step was the conversion of the bromine to the amino- functional group
using Heck’s procedure. The modified Si wafers were put into containers with sealable lids
and placed in a glove box filled with argon. About 30 mL of dry toluene was added,
followed by LiN(SiMe3)2, P(t-Bu)3 and Pd(dba)2. For 1.0 mmol of brominated units,
1.1mmol of LiN(SiMe3)2 and 0.05 - 0.002mmol of P(t-Bu)3 and Pd(dba)2 were used. The
samples were stirred for 24 hours under an inert atmosphere, and then they were washed
with fresh toluene and dried under a nitrogen flow.
To de-protect the silane-protected amino group, the samples were placed into
containers with THF/MeOH (9/1) solution. A few drops of 1M methanolic HCl were added
to the solutions to convert the silane group to the -NH3+Cl- salt. Then the samples were
washed with diethyl ether, dried with N2 flow and placed into fresh THF/MeOH (9/1)
solution. 1M methanolic NaOH was added to convert -NH3+Cl- groups to -NH2 groups. The
samples were washed with water containing triethylamine, dried in the oven at 110oC and
stored in a desiccator under an inert atmosphere. The thickness of the final polymer layers
was measured by ellipsometry.
The thicknesses measured by ellipsometry before and after Heck reaction are
summarised in Tables III-(12-15). With an exception of Experiment 1 (Table III-12), the
82
monolayer thickness was dramatically reduced after the final conversion step, when the
monolayers were exposed to acidic and then basic environments.
Table III-12: Change in thickness of the polymer layer after Heck reaction (Exp 1)
PSBr [%]
Thickness after Soxhlet extract
[Å]
Thickness after HCl and NaOH
[Å]
[%] retained
1 8.9 9.3 98 2.5 9.3 9.3 98 5 7.4 12.5 144
10 6.4 9.3 125
Note: Thicknesses shown in this table are the averages values.
Table III-13: Change in thickness of the polymer layer after Heck reaction (Exp 2)
PSBr [%]
Thickness after Soxhlet extract
[Å]
Thickness after HCl and NaOH
[Å]
[%] retained
2 21.9 9.4 43 4 13.3 7.6 57 6 21.9 6.8 31 8 23.3 5.8 25
10 19.7 8.8 45 13 15.8 8.6 54 16 21.8 8.4 39 20 24.4 7.2 30
Table III-14: Change in thickness of the polymer layer after Heck reaction (Exp 3)
PSBr [%]
Thickness after Soxhlet extract
[Å]
Thickness after HCl and NaOH
[Å]
[%] retained
2 21 5.4 26 4 15.7 6.4 41 6 15.3 4.2 28 8 9.6 4.8 50
10 16 3.8 24 12 18.6 3.4 18 16 13.3 3.8 29 20 14.7 3.2 22
83
Table III-15: Change in thickness of the polymer layer after Heck reaction (Exp 4)
PSBr [%]
Thickness after Soxhlet extract
[Å]
Thickness after HCl and NaOH
[Å]
[%] retained
2 14.7 5.8 40 4 16.9 5.6 33 6 13 7 54 8 11.3 6.4 57
10 13.1 5.8 44 12 11.3 5.8 51 16 11.2 6.4 57 20 15.4 6.2 40
A control test was performed in order to investigate the cause of the polymer
monolayer reduction. Pure PS-SiClMe2 was deposited on a Si wafer from 1% toluene
solution under the same conditions as for the mixed monolayers, as described in the
previous experiments. The sample was subjected to Soxhlet extraction for 3 days to remove
any free polymer chains. The average monolayer thickness, as measured by ellipsometry,
was 40±1Å. Then, the sample was placed in pure toluene and 3 drops of 1M HCl in MeOH
were added while stirring. After the exposure to this acidic environment, the average
thickness was 39±1Å. The sample was put back into the toluene solution and an additional
3 drops of 1M HCl in MeOH were added. After removing the sample, the thickness was
measured again and it was found to be 38±1Å. From these observations, it is clear that the
sample containing only PS-SiClMe2 retained its original monolayer thickness which means
that no chains were detached from the Si surface.
In contrast to the single monolayer, the mixed monolayer thickness was reduced
after the samples were exposed to the acidic conditions. This reduction appeared to be a
completely random process varying from experiment to experiment even though the
experimental conditions were kept the same. The randomness of the washing off of the
polymer chains was apparent within each experiment as well, showing no trend with an
increasing concentration of the amino groups. Possible explanations for the washing off of
84
the polymer chains during this last step of conversion to amino- groups, along with its
implications for the adhesion studies, are discussed in Chapter 5.
These grafted chains were further used for coupling with diglycidylether of
bisphenol A (DGEBA). It is assumed that the amino groups are randomly distributed along
the polystyrene chain. These amino groups are available for the reaction with two
component epoxy resins where one of the components is the amine hardener.
In this study, DGEBA was used as the pre-polymer and either 1,5-diamino-1-
methylpentane (DMP) or Jeffamine D230 were used as curing agents (Chapter 4; Fig.4.4).
Figure 3-25: Bonding of Si to epoxy resin via PS-NH2 chains.
Upon curing, which is a process involving heating, epoxy groups react with amino
groups forming a highly cross-linked network. The unreacted DGEBA/DMP mixture was
used as a glue to bond cured DGEBA/DMP blocks to the polymer modified silicon wafers.
Upon curing, the amino groups of the functionalised polystyrene react with the epoxy
groups in the glue and a permanent bond between the polymer and the epoxy resin is
formed (Fig. 3-25).
O
Si
O
Si
O
Si
NH2
NH2H2N
H2NNH2
H2N HN
H2N NH2
NH2
NH2
H2N
O OHC OH
CH2
DGEBA/DMP
85
3.7 References
1 A. J. Kinloch, Adhesion and Adhesives-Science and Technology (Springer, 1987). 2 B. Arkles; Vol. 7 (Petrarch Systems, Inc., 1977). 3 E. P. Plueddemann, Silane Coupling Agents (Plenum Press, 1982). 4 J. M. Park and J. H. Kim, J. of Colloid & Int. Sci. 168, 103-110 (1994). 5 M. E. McGovern, K. M. R. Kallury, and M. Thompson, Langmuir 10, 3607-3614
(1994). 6 J. B. Brzoska, J. B. Azouz, and F. Rondelez, Langmuir 10, 4367-4373 (1994). 7 P. Silberzan, L. Leger, D. Ausserre, and J. J. Benattar, Langmuir 7, 1647-1651
(1991). 8 C. R. Kessel and S. Granick, Langmuir 7, 532-538 (1991). 9 X. Xiao, J. Hu, D. H. Charych, and M. Salmeron, Langmuir 12, 235-237 (1996). 10 P. K. C., R. W. Johnson, and T. A. Desai, Surf. and Coatings Tech. 154, 253-261
(2002). 11 E. P. Plueddemann, Int. J. Adhes. 1, 305 (1981). 12 W. D. Bascom, J. Colloid Interface Sci. 27, 789 (1968). 13 L. H. Lee, J. Colloid Interface Sci. 27, 751 (1968). 14 C. A. Kumins and J. Roteman, J. Polymer Sci. 1, 527 (1963). 15 A. J. Kinloch, Structural Adhesives: Developments in Resins and Primers (Elsevier
Applied Science Publishers, London). 16 J. S. Quinton and P. C. Dastoor, Appl. Surf. Sci. 152, 131 (1999). 17 M. R. Turner, E. Duguet, and C. Labruere, Surf. and Int. Analysis 25, 917 (1997). 18 R. Hild, C. David, H. U. Muller, B. Vokel, D. R. Kayser, and M. Grunze, Langmuir
14, 342 (1998). 19 A. Kuznetsova, E. A. Wovcho, and J. T. Yates, Langmuir 13, 5322 (1997). 20 B. L. Kropman, D. H. Blank, and H. Rogalla, Langmuir 16, 1469 (2000). 21 H. Ishida, C. Chiang, and J. L. Koenig, Polymer 23, 251-257 (1982). 22 S. M. Kanan, W. T. Y. Tze, and C. P. Tripp, Langmuir 18, 6623-6627 (2002). 23 C. Devaux, J. P. Chapel, E. Beyou, and P. Chaumont, Eur. Phys. J. E 7, 345 (2002). 24 C. Devaux and J. P. Chapel, The Eur. Phys. J. E 10, 77-81 (2003).
86
25 J. Sagiv, J. Am. Chem. Soc. 102, 92-98 (1980). 26 in An amphipathic (or amphiphilic) molecule is a compound having both a
hydrophilic and a hydrophobic end. Some examples of these molecules are soaps,
detergents and some coupling agents. Due to their unique structure, they can
organised themselves either into the structures called micelles or to form SAM as
mentioned above. 27 C. D. Bain and G. M. Whitesides, J. Am. Chem. Soc. 111, 7164 (1989). 28 R. Maoz and J. Sagiv, J. Colloid Interface Sci. 100, 465 (1984). 29 E. E. Polymeropoulos and J. Sagiv, J. Chem. Phys. 69, 1836 (1978). 30 E. Jenkel and B. Rumbach, Electrochem. 55, 612 (1951). 31 R. A. L. Jones and R. W. Richards, Polymers at Surfaces and Interfaces (1999). 32 G. J. Fleer and M. A. Cohen Stuart, Polymers at Interfaces (Chapman & Hall,
London, 1993). 33 P. Frantz and S. Granick, Physical Review Letters 66, 899-902 (1991). 34 P. Frantz, D. C. Leonhardt, and S. Granick, Macromolecules 24, 1868-1875 (1991). 35 H. M. Schneider and S. Granick, Macomolecules 25, 5054-5059 (1992). 36 D. R. Iyengar and T. J. McCarthy, Macromolecules 23, 4344-4346 (1990). 37 B. O'Shaughnessy and D. Vavylonis, Eur. Phys. J. E 11, 213-230 (2003). 38 C. J. Clarke, R. A. L. Jones, J. L. Edwards, K. R. Shull, and J. Penfold,
Macromolecules 28, 2024-2049 (1995). 39 M. A. Plunkett, Thesis, Royal Institute of Technology, 2002. 40 B. Zhao and W. J. Brittain, Prog. Polym. Sci. 25, 667-710 (2000). 41 O. Prucker and J. Rühe, Macromolecules 31, 592-601 (1998). 42 W. Zhao, G. Kraush, M. H. Rafailovich, and J. Sokolov, Macromolecules 27, 2933-
2935 (1994). 43 M. Geoghegan, C. J. Clarke, F. Boue, A. Menellw, T. Russ, and D. G. Bucknall,
Macomolecules 32, 5106-5114 (1999). 44 K. Ebata, K. Furukawa, and N. J. Matsumoto, J. Am. Chem. Soc. 120, 7367 (1998). 45 X. Yang, J. Shi, S. Johnson, and B. Swanson, Langmuir 14, 1505 (1998). 46 D. E. Bergbreiter, J. G. Franchina, and K. Kabza, Macomolecules 32, 4993 (1999).
87
47 V. Koutos, E. M. Van der Vegte, E. Pelletier, A. Stamouli, and G. Hadziioannou,
Macromolecules 30, 4719 (1997). 48 V. Koutos and E. M. Van der Vegte, Macromolecules 32, 1233 (1999). 49 M. Husseman, E. E. Malmström, M. McNamara, M. Mate, D. Mecerreyes, D. G.
Benoit, J. L. Hedrick, P. Mansky, E. Huang, T. P. Russell, and C. J. Hawker,
Macromolecules 32, 1424-1431 (1999). 50 J. Dao, D. G. Benoit, and C. J. Hawker, J. of Polym. Sci.: Part A: Polymer
Chemistry 36, 2161-2167 (1998). 51 M. E. Schrader, in Contact angle, Wettability and Adhesion, edited by K. L. Mittal
(Koninklijke Wohrmann BV, Zeist, 1993). 52 A. V. Pocius, Adhesion and Adhesives Technology (Hanser, 1997). 53 F. M. Fowkes, J. Phys. Chem. 67, 2538 (1963). 54 R. Menescal, R. West, and C. Murray, Macromolecules 24, 329 (1991). 55 H. Kobayashi and M. J. Owen, Macromolecules 23, 4929-4933 (1990). 56 J. H. Clint and A. C. Wicks, J. Adhes. Adhesives 21, 267 (2001). 57 K. M. R. Kallury, M. Thompson, C. P. Tripp, and M. L. Hair, Langmuir 8, 947-954
(1992).
CHAPTER 4
Interfacial Toughness
Measurements
91
4.1 Mechanism of Adhesion
Four main mechanisms of polymer adhesion have been suggested: mechanical
interlocking, diffusion, electrostatic forces and adsorption1.
Mechanical Interlocking
In this mechanism, also referred to as mechanical keying, the adhesive fills the
irregularities in the surface, and upon solidification the adhesive and the surface interlock.
Mercury amalgam for filling tooth cavities is a typical example of this adhesion
mechanism.
Diffusion Mechanism
This mechanism is based on interdiffusion of polymer chains across the interface.
Interdiffusion is possible when polymers are mutually soluble and the polymer segments
have sufficient mobility, which occurs at temperatures above the glass transition (Tg).
When the temperature of a polymer is above Tg, the free volume that exists between the
segments is large enough for other polymer chains to diffuse through the material. For
simple organic materials, solubility can be expressed in terms of solubility parameters (δ),
which are tabulated2. Two materials are soluble in each other when the values of δ are
similar. This is valid for any polymer/polymer or polymer/solvent system. Interdiffusion
usually does not occur when one of the polymers is highly cross-linked, crystalline, below
Tg or if the δ values of the two components are too dissimilar.
Electrostatic Forces
Particles with the same polarity repel each other whereas particles with opposite
polarities attract each other. Electrostatic forces are the second strongest molecular
interactions after covalent bonding3. They are responsible for the formation of ionic bonds
between atoms and molecules which bear a charge, as in a case of ionic crystals.
92
Adsorption Mechanism
When a good molecular contact is established between two materials, bonds may
form due to interatomic and intermolecular forces acting along the interface. Depending on
the nature of the interactions, the following interfacial bonds are formed: a) strong primary
bonds, due to chemisorption, including ionic, covalent and metallic bonds; b) weak
secondary bonds, due to physisorption, including van der Waals (dipole-dipole interaction
and dispersion forces) and hydrogen bonds; c) donor-acceptor interactions (acid-base
interactions) with an intermediate strength.
4.1.1 Donor-Acceptor Interactions
According to the Lewis nomenclature, a Lewis acid is an electron acceptor and a
Lewis base is an electron donor. Partially halogenated ethylene based polymers (PVC),
acrylic acid copolymers, solvents such as dichloromethane (CH2Cl2) and chloroform
(CHCl3), and surfaces of silica and Fe2O3 type are Lewis acids. PMMA, PS,
polycarbonates, polyimides, aromatic solvents (benzene and tetrahydofuran), and surfaces
of Al2O3 type are Lewis bases. Amines, alcohols and polyamides are the examples of both
electron acceptors and donors.
Acceptor-donor interactions can have a profound effect on adhesion through
changing adsorption characteristics, as demonstrated in Figure 4-1. This figure, showing
PMMA deposited on a silica surface from various solvents1, demonstrates that with an
increase of both basicity and acidity of solvent the amount of PMMA adsorbed on the silica
surface decreases. This is because basic solvents compete with basic polymer for the acidic
SiOH groups and, on the other hand, acidic solvents compete with acidic silica substrate for
the basic polymer1.
The system used in the present study consists of basic PS, PS(Br) which is expected
to be slightly more basic than PS (in terms of Lewis classification) due to the bromine atom
93
in the para position, a basic solvent (toluene) and acidic silica. Both PS and PS(Br) compete
with toluene for SiOH sites on silica. It can be assumed that more basic PS(Br) adsorbs
preferentially. This is also supported by the fact that toluene is a good solvent for PS
whereas the solubility of PS(Br) in toluene decreases with an increasing degree of
bromination4.
Figure 4-1: Adsorption of basic PMMA onto acidic silica from basic, neutral
and acidic solvents1.
The interaction parameters χ for PS/toluene and PS(Br)/toluene systems were calculated
according to following equation2:
2)(3.0 jii
RTV δδχ −+= (1)
where Vi is the molar volume of polymer, δi and δj are the tabulated solubility parameters of
polymer and solvent. Calculated interaction parameters for PS and PS(Br) are χPS = 0.35
and χPS(Br) = 0.36, showing not a big difference. It has to be noted that the solubility
parameters are semi-empirical values and therefore the χ’s are only estimates.
94
4.2 Interfacial Fracture Toughness
Fracture occurs when a sufficient amount of energy is released by the growth of a
crack to decrease the overall energy of the system. The energy released comes from the
stored elastic (potential) energy and depends on the loading system. The interfacial fracture
energy Gc, also called the critical strain-energy release rate, can be defined as the energy
required per unit area to propagate a crack. It depends on the rate of crack propagation and
the temperature. Fracture energy can be divided into two components1: intrinsic adhesive
fracture energy Go and energy lossΨ.
Gc = Go + Ψ (1)
Go is the energy required to fracture the bonds in order to propagate a crack through a unit
area. Ψ is the energy dissipated in viscoelastic and plastic deformations during the crack
propagation. When viscoelastic and plastic energy losses (Ψ) are negligible, then Go is a
direct measure of interatomic and intermolecular bonding forces along the interface. When
only secondary bonds are present and the crack propagates through a homogeneous
material, then:
Go = 2γ (2)
where γ is the surface free energy of the material.
In the case of fracture of an adhesive joint, the crack may propagate either in the
materials forming the joint or along the interface. Taking this behaviour into consideration,
Go can be expressed as:
Go = iGo(interfacial) + aGo(adhesive) + sGo(substrate) (3)
where Go(interfacial) is the intrinsic fracture energy of interfacial failure, Go(adhesive) is
the intrinsic fracture energy of cohesive failure in the adhesive, and Go(substrate) is the
intrinsic fracture energy of cohesive failure in the substrate. The pronumerals i, a and s are
95
the area fractions (i + a + s = 1). When fracture occurs only in the interface then Go =
Go(interfacial).
Stress that acts on the material around the crack tip whilst the chemical bonds and physical
bonds are still intact can be expressed as the stress intensity factor (K). Fracture occurs
when K exceeds a critical value Kc, called critical stress intensity factor or fracture
toughness. Gc and Kc are related material properties.
4.2.1 Modes of Fracture
During fracture, new surfaces are created within the fractured material. Fracture can
occur under various conditions defined as modes of fracture5:
Tensile fracture
A direct mechanical load is applied to a material, and the stress is increased continuously
until the material breaks. This type of fracture is uncontrolled and catastrophic.
Fatigue
An alternating stress is applied to a material, simulating loading cycles. After a certain
number of these loading cycles the material fractures. The stress applied is usually
considerably smaller than that required for a tensile fracture.
Creep fracture
Also called static fatigue, this type of fracture occurs when a constant load is applied for a
certain amount of time. The longer the time allowed, the smaller the stress required to
fracture the material. Creep fracture is also a function of temperature and environmental
conditions.
96
Wear/Abrasion
This is a special case of fracture, where small particles of the material are broken off the
surface. An example of this fracture would be the wear of rubber.
Environmental Stress Cracking
Material fractures under small stresses when exposed to a chemically active environment as
in the case of corrosion in metals, glass and polymers. In polymers, the cracking can also be
caused by swelling in water. In this case, water does not react chemically with the polymer
but causes a stress build up within the material.
Crack propagation
Crack propagation occurs in all of the above fracture modes. Depending on how the
stress is applied to a crack, there are three different modes of loading – mode I, II and III6.
Mode I is a cleavage or tensile opening mode, mode II is an inplane-shear mode and mode
III is an antiplane-shear mode (Fig. 4-2).
Figure 4-2: Modes of loading6: a) mode I, b) mode II, c) mode III.
97
4.3 Thermosets
Thermosets are cross-linked polymers forming a random three-dimensional network
during the curing process. Epoxy resins, unsaturated polyesters, phenol-formaldehyde and
amino resins are typical examples of thermosets. These materials have to be molded into a
desired shape during polymerisation and cannot be re-molded after the solidification
because they do not melt or flow at high temperatures. High modulus and resistance to
solvents makes them ideal for applications such as high strength adhesives and matrix
materials for fibre-reinforced composites.
Epoxy resins are the toughest thermosets commercially available. Fracture
toughness Gc is usually in a range 100-300 J/m2. Critical stress intensity factor KIc,
commonly increases with an increasing degree of cure, which is achieved either by higher
curing temperature or longer cure times6[and ref. there in].
There are three modes of crack propagation in epoxy resins6:
Stable brittle propagation
This mechanism is typical for fully cured polymers tested at low temperatures. The
fractured surfaces exhibit no features and Gc is low.
Stable ductile propagation
This crack propagation mechanism occurs in under-cured resins at high temperatures.
Fracture toughness is generally high. Fractured surfaces have a ridged structure.
Unstable brittle propagation
The crack propagates by a stick/slip mode. In this mode the crack jumps upon initiation and
then stops. The fracture surface shows crack arrest lines or broad bands. At the arrest zone,
98
the crack deviates from its original trajectory and the fractured surfaces tend to interlock.
This interlocking along with the plastic deformation around the crack tip results in a
blunting of the crack tip. Initially, the crack grows very slowly then it slips rapidly and
stops when the applied KI has decreased. KI is defined in section 4.6.
It was shown that resins with a high yield stress exhibit continuous crack propagation
whereas resins with low yield stress undergo stick/slip crack propagation (Fig. 4-3) 6.
Figure 4-3: Schematics of a) stable and b) stick/slip mode of crack propagation6.
99
4.4 System Studied
The kinetics and overall reaction of diglycidylether of bisphenol A (DGEBA) epoxy
resin cured with amino hardeners has been well characterised by several researchers using
differential scanning calorimetry (DSC), dynamic mechanical analysis (DMA), near infra-
red spectroscopy and solid-state nuclear magnetic resonance7-11. DGEBA fully cross-linked
with aliphatic amines has a relatively low Tg but it is well above room temperature (see
results section).
In the present study, monomeric DGEBA [DER resin 332] from Fluka (MWav = 340
g/mol) was cross-linked with two different curing agents: 1,5-diamino-3-methylpentane
(DMP) [DYTEK A amine] from Sigma-Aldrich and Jeffamine D230 from Aldrich. DMP
was chosen as an analogy to hexamethylene diamine (HMDA) which is known to form
highly cross-linked networks12. DMP is a liquid at room temperature and therefore easier to
mix with DGEBA than HMDA which is crystalline at room temperature. In addition,
HMDA had a tendency to exude from DGEBA during the curing process and to form
“blooms” on the surface. DMP showed no sign of such segregation. Jeffamines are curing
agents used to form systems with a very low Tg ranging from around -55oC to 85oC.
Chemical structures of DGEBA, DMP and D230 are shown in Figure 4-4.
4.4.1 Mechanism of Cross-linking
Epoxy resins contain reactive oxirane groups capable of reacting with various
chemical groups (hydroxyl, carboxyl, amino and cyano) via nucleophilic substitution. In
this study, aliphatic primary diamines (DMP and D230) were used as curing agents (Fig. 4-
4). In the initial step, the primary amine adds its active hydrogen to the epoxy group. The
resulting secondary amine then reacts with another epoxy group. Because there are two
active epoxy groups on each DGEBA molecule and two amino groups on both DMP and
D230 molecules, a three-dimensional network is formed upon curing (Fig. 4-5).
100
Figure 4-4: Structures of epoxy and diamino curing agents.
C
CH3
CH3
O
H2C
CHH2C
O OCH2
HCCH2
O
Monomeric diglycidylether of bisphenol A (DGEBA)
H2N NH2
CH3
1,5-diamino-methylpentane (DMP)
H2NHC C
H2O
CH3
CH2
HC NH2
CH3
x
General formula of Jeffamine
101
Figure 4-5: Mechanism of Cross-linking between di-epoxy resin and diamino hardener
system.
H2CHC
O
HC CH2
O
NH2H2N+
HC
HC
OHOH
CH2
CH2
NH
NH
NH2H2N
H2CHC
O
HC CH2
O
HC
HC
OHOH
CH2H2C
N
CH2
N
H2C
N
CH2
CH2
N
H2C
H2C
***
*
*
*
2
102
4.4.2 Characterisation of Epoxy System
The DGEBA/DMP system was analysed using a Perkin-Elmer DMA 7e in three
point bend mode under the following conditions: temperature scan rate 10oC/min, load 100
mN, dynamic load 80 mN, frequency 1Hz and nitrogen purge 20cc/min. The ratio of loss
modulus to storage modulus (E’’/E’ known as tan δ) was measured. The temperature when
tan δ starts to increase rapidly was taken as an analogue to the Tg measured by DSC. This
work was done in collaboration with Nathan Jones (UOW). The samples of various cure
regimes were tested by DMA (Table IV-1). Figure 4-6 shows that the values of Tg do not
change substantially when the time of the post-cure regime is varied. The average Tg value
is about 118oC. The Tg of Sample A (cured at 40oC for 24 hour) was near 105oC which is
much higher than expected. Typically, Tg of cured epoxies tends to be from 10oC to 15oC
above the cure temperature. In this case, Tg was more than 60oC above the cure
temperature. This could be caused by an exothermic reaction during the cure. Such
behaviour was observed when HMDA was used as a crosslinker. The system had to be
heated during mixing to keep HMDA in a liquid form. HMDA and DMP are structurally
very similar and therefore it is possible that there was an exothermic reaction during cure
that caused such a large increase in Tg.
DGEBA/D230 was characterised by Gaillard13. Reported Tg measured by Differential
Temperature Analysis (DTA) at heating rate 10oC/min was 82 ± 3oC.
Table IV-1: Cure regimes of DGEBA/DMP system
Sample ID Pre-cure regime Post-cure regime
A 24h @ 40oC
B 24h @ 40oC 24h @130oC
C 24h @ 40oC 36h @130oC
D 24h @ 40oC 48h @130oC
E 24h @ 40oC 60h @130oC
F 24h @ 40oC 72h @130oC
103
Figure 4-6: Tg of various cure regimes of DGEBA/DMP measured by DMA.
4.5 The Fracture Mechanism of Thermosets
It has been suggested that thermosets (cross-linked polymers) undergo brittle
fracture via a localised shear yielding mechanism6. When a crack is initiated, the energy
stored in the material starts to dissipate around the crack tip. As this amount of energy
increases, due to an increasing applied force, the growing crack becomes more and more
blunt and the toughness of the material in a vicinity of the crack tip increases. As the force
increases, the material around the crack tip starts to yield until fracture occurs. This fracture
can be either brittle or ductile. This mechanism is different from crazing which occurs
predominantly in thermoplastic (non-crosslinked) polymers, where the volume of the
crazed material increases. During the process of crack propagation, features called crazes
are formed. They are planar crack-like defects (microcracks) bridged by nanometer size
fibrils. These fibrils are load-bearing. They break down by either chain scission or dis-
entanglement14-16. The crazing mechanism has been well described elsewhere14-17.
A B C D E F100
105
110
115
120
125 Tg by Tan delta Onset [oC]
Tem
pera
ture
[o C]
Sample ID
104
Both mechanisms involve plastic deformation around the crack tip. During crazing
the volume of a material increases whereas during shear yielding the volume is constant.
Both crazing and yielding contribute to brittle fracture. It has been shown6 [and ref. there in] that
cross-linking of polymer chains inhibits crazing and leads to yielding. Thermosets are a
typical example of materials exhibiting such behaviour. When the polymer is tough and
capable of absorbing a large amount of energy via plastic deformation before it fails, the
homogeneous shear yielding results in ductile fracture. When shear yielding is localised
around the crack tip, the crack grows by chain pull-out or scission resulting in brittle
fracture.
4.6 Interfacial Toughness Measurements
Design for the interfacial toughness measurements used in this study was based on
work by Smith et al18,19. The authors studied adhesion between polymer/non-polymer
interfaces using an asymmetric double cantilever test (ADCT), where a specimen is
“sandwiched” between two glass beams. The aim of their work was to promote adhesion
between PS/glass and PS/Si, by modifying the glass and silicon surfaces with PS-PMMA
and PS-PVP block copolymers. The asymmetric double cantilever beam specimen is shown
in Figure 4-7. When a razor blade is inserted into the sample between the polystyrene and
the glass, as indicated in Figure 4-7, both mode I and II (tensile and shear) stress patterns
are generated which affect the propagating crack. This is due to asymmetry of the beams
caused by different beam stiffnesses.
A property combining both tensile and shear stress components is known as mode
mixity, Ψ. It can be expressed in terms of KII and KI (mode II and I) stress intensity factors
(Eq. 4). When Ψ is very small, the crack propagates along the weakest interface. By
adjusting thicknesses of the beams, the crack can be driven into a desired interface. The
modes of crack propagation I and II are described in section 4.2.
I
II
KK1tan−=Ψ (4)
105
Figure 4-7: Schematics of the “sandwich” structure for testing the PS/glass interface19.
Interfacial fracture toughness Gc depends strongly on the ratio of the beam thicknesses
h1/h2, where h1 is the thickness of the top beam and h2 is the thickness of the bottom beam.
Smith et al18 found that when the aspect ratio (h1/h2) > 0.86, the sample fails at high Gc and
there is high scatter in the data. The crack tends to oscillate between the PS layer and the
PS/epoxy interface. When (h1/h2) = 0.71, the crack is driven along the PS/glass interface
and Gc reaches a plateau. For (h1/h2) < 0.71 the thinner upper glass beam breaks. Smith et
all 18 reported that adding copolymers increased Gc from 1 J/m2 (unmodified PS/glass) up
to 25 J/m2.
Smith18 applied the same sample geometry to test the PS/silicon interface. The only
difference was that the silicon had to be affixed to a rigid inflexible foundation because
silicon is very brittle and fractures upon small displacement in bending. The sample
geometry is shown in Figure 4-8. In some cases, the authors were able to measure the
interfacial toughness even though it would be expected that the shear component of the
stress would drive the crack towards the more compliant polymer side19. A razor blade was
inserted into the sample and left in place for 24 hours. Initially, the crack propagated into
the polymer but the energy stored in the bent beam turned the crack back to the interface.
106
Figure 4-8: Schematics of the “sandwich” structure for testing PS/silicon interface18.
Gc of the PS/silicon interface was determined to be around 1 J/m2. Addition of PS-PMMA
block copolymer increased Gc up to 44 J/m2. The interfacial toughness was found to be
independent of the epoxy thickness but increased with increasing thickness of the block
copolymer.
Creton et al16 used the ADCT to study polymer/polymer interfaces reinforced with
block copolymers. The authors observed four different mechanisms of interfacial failure: a)
chain pullout when chains were shorter than their entanglement length; b) small plastic
deformation around the crack tip when the chain density (σ) (equation 12 in Section 3.6)
was low (low Gc); c) craze breakdown of the interface with higher σ; and d) chain scission
when the interface was very strong due to high σ. The authors concluded that Gc increased
with increasing areal density of the copolymer chains.
The mechanisms of the chain pull-out and scission were also reported by Norton et
al20 in a study of adhesion between epoxy/PS interfaces modified with PS-COOH chains
end-anchored to the epoxy. The authors found that Gc increased with increasing PS-COOH
107
length up to N = 838. Beyond this length, Gc started to decline. Other applications of
ADCT can be found elsewhere17,21-25.
4.6.1 Asymmetric Double Cantilever Test
The Asymmetric Double Cantilever Test (ADCT) is a special form of the double
cantilever beam specimen used for testing adhesion strength of adhesive joints6. Due to the
existence of a bi-material interface, deformation induced by both tensile and shear stress
components contribute to the interfacial toughness.
The applied interfacial fracture toughness is a function of the Young’s modulus of
both the epoxy E1 and the silicon wafer E2, the thicknesses of the beams h1 and h2, the crack
opening displacement ∆ (which equals the thickness of the razorblade), and the measured
crack length a.
As reported in the work by Smith18, the asymmetric double cantilever beam
geometry makes it possible to direct the crack along the preferred interface. Gc can be
calculated from an expression based on a model of a cantilever beam on an elastic
foundation derived by Kanninen26:
(5)
ahC 1
1 64.01+= ahC 2
2 64.01+=
Fracture energy is determined from the crack length a, which is measured during the test.
++∆= 3
1322
32
311
31
322
22
311
4
322
311
2
83
ChEChEChEChE
ahEhEGc
108
In this study, the razor blade was inserted between the epoxy and the silicon wafer
modified with a polymer coupling agent. A schematics of the system is shown in Figure 4-
9. The crack propagates by pushing the razor blade further along the interface.
Figure 4-9: Asymmetric double cantilever beam specimen; Gc = f (E, h, ∆2, a4).
As shown in Figure 4-9, the sample consists of two dissimilar beams. The
asymmetry arises not only from the beams having different thicknesses but also from
differences in the material properties of these beams. As a result, there are both tensile and
shear components of the stress acting ahead of the crack tip (mode I and II). The
propagating crack tends to deviate towards the more compliant material.
4.7 Tailoring the Interfacial Toughness using Polymeric Coupling Agents
In a similar manner to other studies where the block copolymers were used to
improve adhesion between two dissimilar materials 16-24, in this investigation the mixtures
of functionalised (active) PS and plain (inactive) PS chains were deposited onto silicon
substrates via the silane end-functional groups. Functionalisation of PS was achieved by
bromination, as described in section 2.1.2, and further conversion to amino groups as
described in section 3.6.3. The mechanism of depositing the mixed monolayers of PSBr0.1-
SiClMe2 (active polymer) and PS-SiClMe2 (inactive polymer) was explained in section
Razor blade Epoxy
Silicon wafer
Coupling agent
h1
h2
∆
Crack length a
109
3.4.2. By increasing the ratio of active/inactive polymer chains chemically anchored to the
silicon surface, the areal density σ of the active chains and therefore the number of
functional groups available for reaction was increased. A random distribution of functional
groups along each functionalised polymer chain was assumed.
When a large amount of epoxy is cast directly onto a glass or a silicon surface and
cured, the stresses created by the epoxy shrinking and by the difference in the thermal
expansion coefficients between the epoxy and the substrate are large enough to “strip” the
top of the glass surface or break the silicon wafer. When the block epoxy is pre-cured and
only a thin layer of the same or similar epoxy is used to glue the epoxy to the substrate,
these stresses are minimised. The pre-cured DGEBA/DMP blocks were cut into 7mm x
5mm x 75mm (H x W x L) beams27 and polished. The modified silicon substrates 15mm x
75mm (W x L) were glued to pre-cured epoxy beams using DGEBA/DMP or
DGEBA/D230 as an adhesive. One end of silicon wafer was covered by PTFE foil for an
easier crack initiation. The samples were pre-cured at 40oC, removed from the oven, and
the Si/epoxy interface was pre-cracked by inserting a razor blade. The razor blade was
removed and the samples were put back into the oven, cured for 3h at 80oC and post-cured
for 3h at 150oC 13. DGEBA and DMP were mixed at 2.1 : 1 molar ratio, DGEBA and D230
were mixed at 2 : 1 molar ratio. After curing, Gc was measured by ADCT. The photographs
of prepared samples are shown (Fig. 4-10).
Figure 4-10: Photograph of an ADCT specimen.
Propagated crack Si wafer
Epoxy beam
Razor blade
110
A razor blade was inserted into the pre-cracked interface and pushed forward until
the crack started to propagate. The razor blade was left inserted in a sample overnight to
allow enough time for the crack to equilibrate. The next day the crack length a was
measured, looking through the top of the epoxy beam, using an optical microscope. The
razor blade was then pushed further until the crack started to propagate again, and the
process was repeated as many times as possible. The limitations were the sample length
and/or premature fracture of the silicon beam.
111
4.8 ADCT Experiments
Experiment 1: ADCT of the epoxy/Si beam samples with mixed (PSBr/PS)-SiCl(Me)2
monolayers deposited on Si wafers
Mixtures of 1, 2.5, 5 and 10 wt % PSBr0.1-SiClMe2 in PS-SiClMe2 were dissolved in
toluene at a total concentration of 1 wt % polymer. Cleaned silicon wafers were dipped into
these solutions and stirred overnight using an orbital stirrer. Then the samples were
removed and cleaned using toluene by Soxhlet extraction overnight to remove any
physisorbed chains. After extraction, the samples were dried under a nitrogen flow and
stored in a desiccator. Thickness and surface energy of the modified silicon surfaces were
measured by ellipsometry and contact angle measurement. The results are shown in Figures
4-11 and 4-12. After conversion the bromine groups to the amino groups by the Heck
reaction, the layer thickness was found to be either identical or slightly higher. The small
increase in the thickness is probably caused by some impurities remaining on the surfaces
after the Heck reaction.
The DGEBA/DMP blocks were prepared and glued to the modified silicon wafers by
methods mentioned above. Both DMP and D230 were used as curing agents for DGEBA to
glue a DGEBA/DMP block to the modified silicon wafers. Gc values, calculated using
equation 5, are shown in Figure 4-13. The sample parameters are shown in Table IV-2.
Note that the BS(Br) [%] in all the plots below relates to the initial concentrations of PSBr
chains in the toluene deposition solutions. Note also that 2 different cross-linkers were
used.
Table IV-2: Gc and sample parameters
Sample [%]
∆ [m]
a [m]
h1(top) [m]
h2(bottom) [m]
E1(top) [N/m2]
E2(bottom) [N/m2]
Gc(average) [J/m2]
5 Jeff 1.00E-04 3.80E-03 7.00E-03 3.70E-04 3.00E+09 1.30E+11 73 10 Jeff 1.00E-04 3.40E-03 7.00E-03 3.70E-04 3.00E+09 1.30E+11 101 2.5 DMP 1.00E-04 3.90E-03 7.00E-03 3.70E-04 3.00E+09 1.30E+11 103 5 DMP 1.00E-04 3.40E-03 7.00E-03 3.70E-04 3.00E+09 1.30E+11 143 10 DMP 1.00E-04 3.20E-03 7.00E-03 3.70E-04 3.00E+09 1.30E+11 230
112
Figure 4-11: Monolayer thickness of the modified Si surfaces (Exp 1).
Figure 4-12: Surface energy of the modified Si surfaces (Exp 1).
0 2 4 6 8 100
2
4
6
8
10
12
14
16
18
20
Original thickness Thickness after Heck
Mixed PS(Br) and PS monolayersTh
ickn
ess
[Å]
PS(Br) [%]
0 2 4 6 8 100
5
10
15
20
25
30
35
40
45
50
Surface energies of PS(Br)/PS monolayers on Si wafers
γ [m
J/m
2 ]
PS(Br) [%]
γ Total γ Disp γ Polar
113
Figure 4-13: Interfacial fracture toughness measurements using ADCT (Exp 1).
Experiment 2: ADCT of the epoxy/Si beam samples with mixed (PSBr/PS)-SiCl(Me)2
monolayers deposited on Si wafers using DGEBA/DMP as a glue layer
Toluene solutions (1 wt %) of PSBr0.1-SiClMe2 (2, 4, 6, 8, 10, 13, 16 and 20%) in
PS-SiClMe2 were prepared. The same procedures as in the experiment 1 were used to
prepare the polymer modified Si surfaces for ADCT. DGEBA/DMP was used as the glue
layer. Thickness and surface energy of the mixed polymer monolayers are shown in
Figures 4-14 and 4-15. Figure 4-16 shows the variation of interfacial toughness with
increasing amount of the active polymer on the silicon surface. Note that the red line is only
a guide for the eye.
0 1 2 3 4 5 6 7 8 9 10 110
50
100
150
200
250
300
350
400
DGEBA/DMP and DGEBA/D230 glue
Gc [
J/m
2 ]
PS(Br) [%]
DGEBA/DMP DGEBA/Jeff
114
Figure 4-14: Monolayer thickness of the modified Si surfaces (Exp 2).
Figure 4-15: Surface energy of the modified Si surfaces (Exp 2).
0 2 4 6 8 10 12 14 16 18 20 2202468
1012141618202224262830
Original thickness Thickness after Heck
Mixed PS(Br) and PS monolayersTh
ickn
ess
[Å]
PS(Br) [%]
0 2 4 6 8 10 12 14 16 18 20 220
5
10
15
20
25
30
35
40
45
50
Surface energies of PS(Br)/PS monolayers on Si wafers
γ [m
J/m
2 ]
PS(Br) [%]
polar disp total
115
Figure 4-16: Interfacial fracture toughness measurements using ADCT (Exp 2).
Experiment 3: ADCT of the epoxy/Si beam samples with mixed (PSBr/PS)-SiCl(Me)2
monolayers deposited on Si wafers using DGEBA/D230 as a glue layer
Experiment 2 was repeated using fresh solutions of PSBr0.1-SiClMe2 (2, 4, 6, 8, 10,
12, 16 and 20%) in PS-SiClMe2. Thickness of the mixed polymer monolayers after the
Soxhlet extraction and after the Heck reaction is shown in Figure 4-17. DGEBA/D230 was
used as the glue layer. The resulting plot of Gc vs. solution concentration of PSBr0.1-
SiClMe2 is shown in Figure 4-18. The points that are not showing error bars were obtained
from only one measurement before the silicon substrate fractured. The surface energies of
these samples were not measured.
0 2 4 6 8 10 12 14 16 18 20 22
-10123456789
10111213
DGEBA/DMP layer
Gc [
J/m
2 ]
PS(Br) [%]
116
Figure 4-17: Monolayer thickness of the modified Si surfaces (Exp 3).
Figure 4-18: Interfacial fracture toughness measurements using ADCT (Exp 3).
0 2 4 6 8 10 12 14 16 18 20 22
0
10
20
30
40
50DGEBA/D230 layer
Gc [
J/m
2 ]
PSBr [%]
0 2 4 6 8 10 12 14 16 18 20 222
4
6
8
10
12
14
16
18
20
22
Mixed PS(Br) and PS monolayersTh
ickn
ess
[Å]
PS(Br) [%]
Original thickness THickness after Heck
117
Experiment 4: ADCT of the epoxy/Si beam samples with mixed (PSBr/PS)-SiCl(Me)2
monolayers deposited on Si wafers using DGEBA/D230 as a glue layer
The same conditions and procedures were used as in experiment 3, using fresh
solutions. Figures 4-19 and 4-20 show surface analysis of the mixed polymer layer. Figure
4-20 shows Gc measurements.
The control sample CS1 was prepared by depositing pure PS-SiClMe2 (inactive PS)
on the Si substrate. The sample was then glued to the DGEBA/DMP beam with
DGEBA/D230 glue and tested by ADCT as described before. Gc results were plotted along
with the results for the mixed polymer layers. The red line in Figure 4-20 serves only as a
guide.
Figure 4-19: Monolayer thickness of the modified Si surfaces (Exp 4).
0 2 4 6 8 10 12 14 16 18 20 220
2
4
6
8
10
12
14
16
18
20 Original thickness Thickness after Heck
Mixed PS(Br) and PS monolayers
Thic
knes
s [Å
]
PS(Br) [%]
118
Figure 4-20: Surface energy of the modified Si surfaces (Exp 4).
Figure 4-21: Interfacial fracture toughness measurements using ADCT (Exp 4).
0 2 4 6 8 10 12 14 16 18 20 22
5
10
15
20
25
30
35
40
45
50
Surface energies of PS(Br)/PS monolayers on Si wafersγ
[mJ/
m2 ]
PS(Br) [%]
Disp Polar Total
0 2 4 6 8 10 12 14 16 18 20 220
20
40
60
80
100
120
Mixed polymer layer Control sample
DGEBA/D230 Layer
Gc [
J/m
2 ]
PS(Br) [%]
119
4.8.1 Discussion of Gc
In Experiment 1, the thickness of the mixed polymer layers was found to be quite
uniform, averaging around 8Å. The polar component of surface energy shows a slightly
increasing trend with an increasing concentration of brominated PS. Several samples broke
before any reasonable crack lengths could be measured. The remaining samples show
increasing Gc with increasing concentration of PS(Brx). The slope is steeper, and the values
generally higher, for the samples where DGEBA/DMP glue was used. The average
interfacial toughness values of these samples were calculated to be between 100 and 230
J/m2. In the case of DGEBA/D230, the values were between 75 and 100 J/m2.
In Experiment 2, the thickness data are more scattered, oscillating around 20Å. The surface
energies were uniform across the range of concentrations, and the polar surface energy was
very low. Interfacial fracture toughness of these epoxy/Si interfaces was very low
averaging between 2 and 8 J/m2. The general trend is a slight increase in Gc with increasing
concentration of PS(Brx).
In experiment 3 (D230), The Gc values are very low as well, averaging between 5
and 20 J/m2. After an initial scatter, the Gc values level off and oscillate around 15 J/m2.
In Experiment 4, the average thickness was around 14Å and γpolar are scattered
between 2 and 15 mJ/m2. Gc values were between 70 to 95 J/m2 and remained steady across
the range of concentrations. The interfacial toughness of the control sample CS1
(epoxy/PS) was calculated to be 10 J/m2. These results show that adding the PS(Brx) and
converting it to PS(NH2) increased the interfacial toughness by factor of about four.
Some correlation between γpolar and Gc was observed. The possible meaning of this
observation will be further discussed in Chapter 5.
Two different cross-linkers, DMP and D230, were used to glue DGEBA/DMP
blocks to the polymer modified silicon surfaces, in order to investigate any effect of
120
different degree of cross-linking and different Tg on the characteristics of the crack
propagation. Due to high scatter between the experiments, no conclusions could be made.
The only observation made was that the samples prepared with DGEBA/D230 had a higher
success rate, meaning that there was less breaking of the silicon wafers. This behaviour
would suggest that the more flexible DGEBA/D230 glue relieves more stress that is
generated during the crack initiation process and therefore prevents breaking of the silicon.
The stiffer DGEBA/DMP glue, however, does not have as a high capacity to relieve the
stress, and therefore breaking of the silicon is more frequent.
The expression for Gc calculation, derived by Kanninen (equation 5), assumes both
equal width and length of the beams. In this study, the length of the beams was identical but
the width was different. The silicon beam was wider than the epoxy beam to prevent
breaking of the silicon. This change in the sample geometry would introduce an error in the
final Gc values. It was shown28 that this error is only small and therefore it was neglected in
this study. Another factor that affects the final fracture toughness is the mismatch between
the thermal coefficient of the epoxy and the silicon materials. The detailed description of
this subject can be found elsewhere28. For all experiments conducted in this study, the
sample geometry was kept constant. Therefore it was assumed that the errors caused by the
different sample dimensions and the thermal mismatch would contribute equally to Gc of
each sample and therefore were not taken into the account.
The surface energy measurements were not conducted after the Heck reaction
because of the risks of contamination and deactivation of amino functional groups.
121
4.9 References
1 A. J. Kinloch, Adhesion and Adhesives-Science and Techology (Springer, 1987). 2 J. E. Mark, Physical Properties of Polymers Handbook (AIP Press, New York,
1996). 3 A. V. Pocius, Adhesion and Adhesives Technology (Hanser, 1997). 4 R. Oslanec, (University of Wollongong, 2005). 5 E. H. Andrews, Fracture in Polymers (Oliver & Boyd, Edinburgh and London,
1968). 6 A. J. Kinloch and R. J. Young, Fracture Behaviour of Polymers (Applied Science
Publishers, London and New York, 1983). 7 B. G. Min and Z. H. Stachurski, Polymer 34, 4488 (1993). 8 B. G. Min, Z. H. Stachurski, and J. H. Hodgkin, Polymer 34, 4908 (1993). 9 B. G. Min, Z. H. Stachurski, J. H. Hodgkin, and G. R. Heath, Polymer 34, 3620
(1993). 10 M. E. Merritt, L. Heux, J. L. Halary, and J. Shaefer, Macromolecules 30, 6760-6763
(1997). 11 L. Heux, J. L. Halary, F. Laupretre, and L. Monnerie, Polymer 38, 1767 (1997). 12 L. Heux, F. Laupretre, J. L. Halary, and L. Monnerie, Polymer 39, 1269 (1998). 13 S. Gaillard, “Internship report,” (2001). 14 C.-Y. Hui, A. Ruina, C. Creton, and E. J. Kramer, Macromolecules 25, 3948-3955
(1992). 15 F. Xiao and W. A. Curtin, Macromolecules 28, 1654-1660 (1995). 16 C. Creton, E. J. Kramer, C.-Y. Hui, and H. R. Brown, Macromolecules 25, 3075-
3088 (1992). 17 H. R. Brown, Macromolecules 24, 2752-2756 (1991). 18 J. W. Smith, E. J. Kramer, F. Xiao, C.-Y. Hui, W. Reichert, and H. R. Brown, J. of
Mat. Sci. 28, 4234-4244 (1993). 19 J. W. Smith, E. J. Kramer, and P. J. Mills, J. Polymer Sci.:Part B: Polymer Physics
32, 1731-1744 (1994).
122
20 L. J. Norton, V. Smigolova, M. U. Pralle, A. Hubenko, K. H. Dai, E. J. Kramer, S.
Hahn, C. Berglund, and B. DeKoven, Macromolecules 28, 1999-2008 (1995). 21 J. Washiyama, C. Creton, and E. J. Kramer, Macromolecules 26, 6011-6020 (1993). 22 F. Xiao, C.-Y. Hui, and E. J. Kramer, J. of Mat. Sci. 28, 5620-5629 (1993). 23 K. Char, H. R. Brown, and V. R. Deline, Macromolecules 26, 4164-4171 (1993). 24 H. R. Brown, K. Char, V. R. Deline, and P. F. Green, Macromolecules 26, 4155-
4163 (1993). 25 J. Duchet, J. P. Chapel, and B. Chabert, Macromolecules 31, 8264-8272 (1998). 26 M. F. Kanninen, International Journal of Fracture 9, 83-92 (1973). 27 J. Benkoski, edited by P. communicacion (2003). 28 J. J. Bekonski, E. J. Kramer, H. Yim, M. S. Kent, and J. Hall, Langmuir 20, 3246-
3258 (2004).
CHAPTER 5
Discussion
125
5.1 Discussion
The results for interfacial fracture toughness (Gc) can be looked at as indirect
measurements of the interfacial properties. The initial analysis of Gc vs. initial
concentration of PSBr (Chapter 4) would suggest that within the range of concentrations of
the active PS(NH2) (2 – 20wt%) grafted to Si wafers as mixed monolayers of PS/PS(NH2),
there is a saturation in Gc values. The reproducibility between the experiments is poor,
however, and therefore it is not possible to draw any specific conclusions about the value of
the interfacial fracture toughness of these Si/epoxy interfaces. Possible explanations for this
behaviour are discussed below.
Based on the research of interfacial toughness between two immiscible polymers
and polymer/solid substrate, described in the introduction1-9, it would be expected that Gc
values would increase with increasing concentration of the active chains. In this study, the
interface containing only PS (no PS(NH2)) chains was significantly weaker than the
interfaces with PS(NH2) chains. Within each experiment, all non-zero PS(NH2)
concentrations gave essentially the same Gc. Between the experiments, however, Gc. varied
significantly. Possible reasons for this behaviour are discussed below.
Even though the silane coupling agents are widely used to improve water resistance
of various surfaces and interfaces, Plueddemann10 suggests that the siloxane bonds between
Si and silane coupling agents are not resistant to hydrolysis. This means that the siloxane
covalent bonds can be disrupted by water. The activation energy for this reaction is 98.6
kJ/mol10. The addition of benzoic acid during the hydrolysis process decreases this
activation energy to 25 kJ/mol, which is energy comparable to the strength of a hydrogen
bond. During the process of bonding of the silane coupling agents to the mineral surfaces
(M), there is a continuous process of bonding and de-bonding due to a reversible reaction
with water10:
M−O−Si− + H2O ⇔ M−O−H + H−O−Si−
126
Adding water to the system would drive the reaction to the right. Efficient adhesion can be
achieved when the equilibrium conditions are shifted to the left, towards the formation of
oxane bonds. This can be done by a) maximising the concentration of initial siloxane
(M−O−Si−) bonds, b) minimising the penetration of water to the interface and c) applying
polymeric material that “holds” silane molecules at the surface. The last case was
demonstrated in a study of water-assisted crack propagation between (3-
glycidoxypropyl)trimethoxysilane coated silicon wafers and DGEBA11. The authors
observed almost no effect of pH (varied from 7 to 11) on the interfacial fracture toughness.
In this study, the number of siloxane bonds at the interface was maximised by
treating the silicon wafers with hot piranha solution during the cleaning procedure. This
process is known to create a high concentration of Si-OH sites that are available for
reaction with silanes12. As mentioned in Section 3.6.3, when PS-SiClMe2 was deposited on
the silicon wafers, the polymer layers were stable even when exposed to acidic and basic
conditions. It was confirmed by ellipsometry that after adding 3 drops of 1M HCl in MeOH
solution to a PS-SiClMe2/silicon sample immersed in 25mL of THF/MeOH (9/1) and
stirred for 25 minutes, the PS layer retained its original thickness (~ 40Å). However, when
the mixed monolayers were deposited on silicon wafers, the polymer chains were washed
off during the last step of conversion when exposed to acidic and basic environments
(except the initial experiment 1). It is known that both acids and bases are catalysts for
hydrolysis and re-formation of the oxane bonds10. The ellipsometry measurements showed
that the thickness of the mixed PS(Br)/PS layers before the conversion was lower than that
of the plain PS-SiClMe2. Therefore the areal density of the mixed polymer chains (σ) was
lower than that of the plain PS chains. This might be due to the attraction of the polar
bromine groups of PS(Br) chains to the silicon oxide surface, which causes the chains to lay
flat on the surface and therefore prevent other chains from accessing the surface. The lower
σ might allow for the acid and the base to reach the surface and hydrolyse the siloxane
bonds. In this way, the NH2 groups of the converted polymers may act as conductors for the
acid and the base.
127
This washing off of the grafted chains could perhaps be prevented by using polymer
chains with tri-functional silane end-group instead of mono-functional silane end-group.
The mono-functional silane terminated polymer was ideal for producing a monolayer.
However, as discussed in Chapter 3, mono-functional silane does not produce stable layers.
The more stable layers are formed in case of tri-functional silanes through cross-linking
reactions of adjacent silanol groups. This cross-linking produces a layer that is more
resistant to hydrolysis13.
The disruption of the siloxane bonds found in this study is in agreement with the
results of Duchet et al14, who studied the stability of monofunctional alkylsilane layers with
Cl or N(CH3)2 hydrolysable groups, deposited on quartz slides and on nanometer size silica
particles, either from toluene solution or by vapour method. The samples were exposed to
0.1M KOH solution for various periods of time. The silane layer degradation was observed
by contact angle measurements.
In research done on polyester laminates10, single silanes were compared with mixed
silanes as treatment on glass microbeads in polyester casting. Methacrylate silane and
diamine-functional silane were mixed in a ratio 9/1. The authors found that the initial
adhesion improved by adding the diamine silane, however, after 24h of boiling in water, the
flexural strength decreased compared to the pure methacrylate silane. The explanation was
that the unreacted amine groups were hydrophilic enough to allow water to the interface
region.
Examples of hydrolysis of silane coupling agents can also be found in some
practical applications, such as glass-reinforced composites. Plueddemann10 studied the
flexural strength of glass-cloth-epoxy and glass-cloth-polyester composites after exposure
to hot aqueous solution of pH ranging from 2 to 10. In this case, silanes were used as
adhesion promoters. He found that the best retention of mechanical properties was in acid
or alkaline water. This situation is different to that described in the present study, however,
because the silane coupling agents are fixed between the composite layers so even if the
siloxane bonds are dis-lodged due to hydrolysis they reform upon heating. In the present
128
study, the samples are immersed in a THF/MeOH (9/1) solution during hydrolysis. There is
nothing to “hold” the polymer layers down to the silicon wafer, and hence the chains are
washed off into solution. This process of “washing off” of the chains is completely random
and varies from experiment to experiment.
Several analytical techniques were used by other researchers to examine the grafted
densities of silane layers. Duchet et al15 used solid state 29Si NMR spectroscopy, gas
chromatography and elemental atomic analysis to determine the grafting density of the
silane chains. When solid state 29Si NMR spectroscopy was used, the authors reported only
semi-quantitative results with accuracy about 4-5%. For long polymer chains, the accuracy
of the measurement would be expected to be even lower. In order to carry out gas
chromatographic analysis, the silane molecules were cleaved from the surface by
hydrofluoric acid etches. In the last method, an elemental atomic analysis, the grafting
density was determined from the atomic percentage of carbon. The disadvantage of this
method is that the carbons from the solvent molecules adsorbed at the surface are included
in the percentage and as a result, σ was overestimated. The authors reported that σ
decreased with an increasing length of silane chains because the bonded molecules
sterically hinder the reaction of additional molecules. Another technique used to analyse the
silane monolayers was transmission infrared spectroscopy, where the silane chains were
deposited on Aerosil silica16. The main reason for using silica particles was that the grafting
density is higher on highly curved surface of silica than on flat surfaces. This is due to
already mentioned steric effects of the chains which are more pronounced in case of the flat
surfaces. Also, the amount of grafted chains to analyse is much greater in cases of silica
particles where the surface area is significantly larger than in planar samples.
As shown above, the study of the grafting densities of the uniform silane
monolayers produces rather semi-quantitative results. To perform a quantitative analysis of
the mixed polymer monolayers chemically attached to a solid substrate is even more
challenging because the concentrations of the chains with particular atoms attached is very
low. Such analysis would be extremely demanding on the resolution of the available
surface analysis techniques. In this study, several samples with mixed polymer monolayers
129
were examined by grazing angle IR and XPS techniques, but no satisfactory results were
obtained.
For simplicity, it was assumed in this work, that the deposition of polymer chains
from solutions of different ratios between PS and PS(Br) chains would yield monolayers of
similar surface composition. Unfortunately, due to the washing off of the chains, this initial
information about the monolayer composition was lost. To perform a quantitative analysis
on the remaining partial monolayer on silicon wafer would be extremely challenging for the
reasons described above. The samples were therefore used for the asymmetric double
cantilever test (ADCT) assuming that both types of polymer chains were still present on the
surface.
In the study of adhesion between glass and polyethylene (PE), Duchet et al14
explored different regimes of the interfacial failure. Chlorosilane-terminated PE with
different chain lengths was grafted to the glass substrates and joined to a thick PE film. The
second glass slide was used to reinforce the PE film. The mechanism of failure depends on
the chain length and σ. For short chains with molecular mass below the entanglement
molecular mass, and low σ, the interface fails by chain pull-out. The Gc of such interfaces
was reported to be between 1 and 6 J/m2 14. For long chains and low σ, the interface fails
by chain scission. In both regimes, Gc varies linearly with σ (Gc ~ σ N2)17. In case of higher
σ, crazing involving chain disentanglement occurs, and Gc varies with σ2 (Gc ~ σ2)18.
In this study, the regime of long chains and low grafting density is applicable.
Based on the adhesion studies between the thermoplastic polymers and a solid substrate1-9,
it would be expected that the interfacial fracture toughness would increase with increasing
areal chain density. However, such behaviour was not observed. There appeared to be an
increasing trend in the Experiment 1 (Figure 4-13) when DMP was used as a cross-linker.
However, the plot contains only three measurement data and the error bars are large. For
this reason, the possible increase in Gc was not taken into further consideration. From the
dependence of Gc on initial solution concentration of the grafted mixed polymers layers
(PS/PSBr), it can be concluded that within the studied concentration range (0 – 20%) of the
130
active PS(NH2) chains the interfacial fracture toughness reached a plateau. Fracture
toughness of epoxies is usually in the range 100-300 J/m2. For the specific system of
DGEBA/HMDA (hexamethylene diamine), the reported Gc initiation value is around 570
J/m2 19. HMDA has a very similar structure and glass transition temperature (Tg) to the
DMP cross-linker used in this study (Chapter 4). Except for Experiment 1 (Section 4.7), the
highest Gc value measured was 95 J/m2 which is well below the fracture toughness of
epoxy. Therefore this plateau would be caused by the crack propagating along the interface
rather than within the bulk epoxy.
When the ADCT was performed on the sample with plain PS chains (i.e. no amine
functionality), the razor blade was easily inserted and the crack propagated in a continuous
manner without stopping. This behaviour is different from the samples with mixed
monolayers, where after a difficult crack initiation the crack would jump and then stop. In
this case, an increased force had to be applied in order to reinitiate the crack propagation.
This corresponded to the stick/slip propagation mechanism described in Chapter 4. Since
both plain PS and mixed PS/PSBr monolayer samples were prepared under the same
conditions, including the epoxy cure regime, this difference in crack propagation behaviour
can be related to the differences in the interfacial properties. This is clear evidence of a
higher Gc when PS(NH2) chains are present at the surface. Whether this is due to reaction
of the remaining chains with the epoxy, or if it is due to an increase in the surface
roughness, is not clear. The reason may be a combination of both.
It is not obvious why the samples from the initial experiment (Exp. 1) exhibited
different behaviour compared to the later experiments (Chapter 4). The samples were
prepared in the same way and exposed to same reaction conditions. The only difference
was that the silicon wafers used were sourced from two different suppliers. Silicon wafers
used in the initial experiment were supplied by Silicon Inc. (20Å native oxide layer),
whereas for the subsequent experiments, silicon wafers from Compart Technologies (17Å
native oxide layer) were used. The wafers were cleaned using the same procedure prior to
use. Wafers from both suppliers had the same dopant type and concentration, and
orientation 100. When control experiments were performed with mixed monolayers
131
deposited on both wafer types, the polymer chains were washed off under acidic conditions
from both wafers (Tables III-(13-15)). Considering these results, the difference in the
behaviour between the first initial experiment and the rest of the experiments is very
unlikely to be due to using different Si wafers.
A possible explanation for this behaviour might be in the testing technique itself. It
was shown that when bulk epoxy was tested in tensile mode (I), the stress intensity factor
(KIci) for crack initiation (i) varies with the degree of cure19. The same was observed for
GIci20. Clearly, there was a difference between Kci and Gci values for the crack initiation,
and those for the crack arrest (a), Kca and Gca 19. The evidence for different behaviour of Gci
and Gca was also given in a study of Mijovic et al20. The authors observed that for the
DGEBA/DETA (diethylene triamine) system, the GIci (initiation) increased with increasing
content of curing agent up to a certain concentration and then dropped slightly, whereas
GIca (arrest) changed only slightly. In another study19, the effect of the testing rate on the
fracture toughness was measured. The authors found that as the testing speed increased, KIci
decreased, but KIca remained constant. A variation of KIci with testing temperature was also
reported19. It was found that at low temperatures (below 0oC) the crack propagated in a
continuous manner. As the temperature increased, the crack propagation became unstable,
and KIci increased whereas KIca remained approximately constant.
This behaviour would suggest that GIca is less sensitive to the polymer chemistry
and testing conditions than GIci. On the other hand, in a study of the effect of various post-
cure times on the fracture energy of bulk epoxy20, the values of fracture energies of
initiation and arrest followed the same pattern – higher values of GIci and GIca were
observed within the initial stages of post cure. Later, GIci and GIca dropped slightly and
became nearly constant. The only difference between the initiation and arrest values was
that GIca was lower than GIci.
In this study, the crack in the epoxy/modified SiOx interface propagated in an
unstable stick/slip manner. The ADCT test was carried out manually and the crack length
was measured at the point where the crack arrested. This indicates that the Gc values
132
obtained from these measurements are actually Gca (arrest) values rather than Gci
(initiation). Gca measurements, as described above, are not sensitive to many factors such as
changing testing properties and the properties of the material. It is possible to speculate that
in this study, the GIca values, obtained from the ADCT, are not sensitive to the variation in
concentration of PS(NH2) chains at the interface, and therefore no trend in Gc was
observed. However, considering the fact that the measured GIc varied from batch to batch,
even though there is no trend observed within each batch, the test gives some indication
about the interfacial fracture energy. The large scatter in Gc values is typical for Gca
measurements. The possibility that the scatter in the data is due to the surface
contaminations from de-ionised water during the Si surface preparation (prior to the
monolayer deposition) is highly unlikely, as the UVO removes any residual traces of
organics from the surface. The samples were immersed into the polymeric solutions
immediately after the UVO cleanse.
The difference in the crack propagation behaviour between the PS and PS(NH2)
samples was obvious even though the ADCT was done by hand. As mentioned before, for
PS modified Si wafers, the crack propagated between the polymer layer and the epoxy very
easily in a stable continuous manner. The re-initiation of the crack was very easy. In
comparison, to re-initiate the crack in the PS(NH2) samples was difficult and often led to
breaking of the silicon.
The above observations led to the conclusion that the lack of variation in the
interfacial fracture toughness is more likely caused by the behaviour of the polymer chains
on the silicon surface. It was shown that the PS chains remained on the silicon surface even
after exposure to acidic and basic environments (Section 3.6.3), whereas PS(NH2) chains
were washed off during the step of converting –NH3+Cl- functional groups to -NH2 (Tables
III-13 to III-15). The amount of chains washed off varied randomly from batch to batch,
and therefore Gc results can not be related to the surface concentration of PS(NH2) chains.
The only conclusion that can be drawn at this stage is that the ADCT test measured Gca
values and that these values represent the fracture energy of the interfaces.
133
The main disadvantage of ADCT performed by hand, as used in this study, is that
the tensile/shear loading ratio (mixity) varies easily as the razor blade is pressed in. This
could be prevented by building a motor driven tool for performing the razor blade test,
similar to the one used for example by Smith4,7,21. This would allow for better control of the
speed and force applied, which is not possible when the test is done by hand. As a result the
mixity would be constant and the crack propagation cold be studied more objectively.
Comparing the results from the Experiments 1, 2 and 4 (see Section 4.8), the
correlation between the polar component of surface energy (γpolar) of the chemisorbed
PS/PS(Br) chains (before the conversion to PS/PS(NH2)) and the interfacial fracture energy
(Gc) between the epoxy and PS/PS(NH2) was found. Even though the variation of γpolar did
not reflect various concentrations of PS/PS(Br) chains in the original solutions, it varied
from experiment to experiment, suggesting different surface properties. The surfaces with
lower γpolar, around 3 mJ/m2, produced Gc ranging from 2 to 7 J/m2 (Experiment 2). The
surfaces with higher γpolar, around 10 mJ/m2, gave Gc ranging from 50 to 225 J/m2
(Experiment 1) and 70 to 95 J/m2 (Experiment 4). There might not be any direct connection
between γpolar and Gc but two possible hypotheses outlined below suggest some correlation:
1) After some polymer chains were washed off in the final conversion step,
there was still a sufficient amount of chains remaining on the surface with
-NH2 groups that reacted with the epoxy during cure. In such case, Gc would
reflect the variations in the interfacial chemistry. If γpolar corresponds to the
amount of the PS(Br) chains, it could be assumed that surfaces with a higher
γpolar had higher concentrations of PS(NH2) chains and therefore more links
across the Si/epoxy interface. Hence, Gc would increase with an increasing
γpolar due to the increasing concentration of the amino-functional polystyrene
chains.
2) As in scenario 1, the higher γpolar would suggest higher concentration of
PS(Br) and therefore PS(NH2) chains after the conversion. The hydrophilic
134
NH2 groups attracted water and therefore due to the hydrolysis of the
siloxane bonds, a larger amount of chains were washed off of the silicon
surface. This process could leave silicon wafers with only patches of
polymer. The areas of bare silicon would contribute to the strong interfacial
adhesion due to a good adhesion between epoxy and silicon oxide. In
addition, the increased surface roughness also contributes to stronger
adhesion.
The ellipsometry measurements showed that the final thickness of the polymer
chains was very similar for all three experiments, ranging from 5Å to 8Å. Such a small
variation in polymer thickness is within the instrumental error so it can be neglected.
Therefore the thickness measurements do not suggest differences in the structure of the
layers. Comparing the Experiments 1 and 2, the average thickness of the final polymer
layers was about 8Å. The γpolar and Gc values varied substantially however, suggesting that
the interfacial toughness is a measure of the amount of PS(NH2) between the silicon/epoxy
interface.
Further systematic studies including the characterisation of the deposited polymer
chains are necessary to confirm these observations. These will be outlined in the next
chapter.
135
5.2 References
1 K. Char, H. R. Brown, and V. R. Deline, Macromolecules 26, 4164-4171 (1993). 2 C. Creton, E. J. Kramer, C.-Y. Hui, and H. R. Brown, Macromolecules 25, 3075-
3088 (1992). 3 H. R. Brown, Macromolecules 24, 2752-2756 (1991). 4 J. W. Smith, E. J. Kramer, and P. J. Mills, J. Polymer Sci.:Part B: Polymer Physics
32, 1731-1744 (1994). 5 K. K. Weon and e. al., J. of Modern Physics B: Condensed Matter Physics,
Statistical Physics, Applied Physics 17, 1814-1820 (2003). 6 J. Washiyama, C. Creton, and E. J. Kramer, Macromolecules 26, 6011-6020 (1993). 7 J. W. Smith, E. J. Kramer, F. Xiao, C.-Y. Hui, W. Reichert, and H. R. Brown, J. of
Mat. Sci. 28, 4234-4244 (1993). 8 L. J. Norton, V. Smigolova, M. U. Pralle, A. Hubenko, K. H. Dai, E. J. Kramer, S.
Hahn, C. Berglund, and B. DeKoven, Macromolecules 28, 1999-2008 (1995). 9 Y. Sha, C.-Y. Hui, E. J. Kramer, S. F. Hahn, and C. A. Berglund, Macromolecules
29, 4728 (1996). 10 E. P. Plueddemann, Silane Coupling Agents (Plenum Press, 1982). 11 H. Yim, M. S. Kent, D. R. Tallant, M. J. Garcia, and J. Majewski, Langmuir 21,
4382-4392 (2005). 12 J. Duchet, J.-F. Gerard, J. P. Chapel, and B. Chabert, Composite Interfaces 8, 177-
187 (2001). 13 B. Arkles; Vol. 7 (Petrarch Systems, Inc., 1977). 14 J. Duchet, J. P. Chapel, B. Chabert, and J. F. Gerard, Macromolecules 31, 8264-
8272 (1998). 15 J. Duchet, B. Chabert, J. P. Chapel, J. F. Gerard, J. M. Chovelon, and N. Jaffrezic
Renault, Langmuir 13, 2271-2278 (1997). 16 C. P. Tripp and M. L. Hair, Langmuir 7, 923-927 (1991). 17 P. G. De Gennes, J. Phys. (Paris) 68, 1049 (1990). 18 H. R. Brown, Annu. Rev. Mater. Sci. 21, 463 (1991).
136
19 A. J. Kinloch and R. J. Young, Fracture Behaviour of Polymers (Applied Science
Publishers, Londong and New York, 1983). 20 J. Mijovic and J. A. Koutsky, Polymer 20, 1095-1107 (1979). 21 J. J. Benkoski, E. J. Kramer, H. Yim, M. S. Kent, and J. Hall, Langmuir 20, 3246-
3258 (2004).
CHAPTER 6
Conclusion
139
6.1 Conclusion
Various methods for the chemical modification of silicon surfaces by polymeric
coupling agents were investigated. These included grafting from (NMP and ATRP) and
grafting to techniques. Using the NMP technique, polystyrene (PS) chains were
successfully grown from silicon wafers. However, this method for growing copolymers
with various functionalities was found to be unsatisfactory. The copolymers obtained had
very short chain lengths, and the molar mass could not be satisfactorily controlled. ATRP
syntheses were successful for the preparation of free copolymers, however attempting to
grow polymers from the silicon surfaces did not produce molecules of sufficient length.
To overcome these difficulties, novel polymeric coupling agents were synthesised
by a chemical modification of the existing silane end-functional polystyrene chains with
low polydispersity. The PS-SiClMe2 chains were brominated (molar fraction of brominated
units, x = 0.1) and chemically grafted to the cleaned Si wafers as mixtures of PSBr-
SiClMe2/PS-SiClMe2 at various ratios ranging from 2 to 20% of brominated polystyrene.
The Br groups along the PS chains were then converted to NH2 groups by a chemical
procedure based on the Heck reaction. The conversion from PS to PS(Br) and subsequently
to PS(NH2) was carefully monitored by 13C-NMR and 1H-NMR techniques using free
(unattached) PS. The presence of the NH2 group was qualitatively confirmed by 1H-NMR,
which showed a peak around 3.4 ppm, corresponding to the amino group. In the final
conversion step of NH4+Cl- groups to NH2 groups, some polymer chains were washed off
the surface. It was concluded that because of the presence of hydrophilic NH2 groups, the
siloxane bond was hydrolysed. Because the polymer chains were swollen in a good solvent,
some of them were washed away.
The relationship between the interfacial fracture toughness and molecular coupling
was studied using the asymmetric double cantilever test (ADCT). The crack between the
epoxy and PS(NH2)/PS modified Si wafers was found to propagate is a stick/slip manner.
No significant variation of interfacial fracture toughness (Gc) with an increasing
140
concentration of the reactive PS(NH2) chains was observed. Some explanations for this
unexpected behaviour were suggested.
The ADCT was used previously by other researchers to study systems where
continuous crack propagation was more typical. Therefore, it was speculated that due to the
stick/slip behaviour observed, the Gc values measured were Gca (arrest) rather than Gci
(initiation). From the literature, it is known that GIca is less sensitive to the variations in
testing conditions (temperature, speed) than GIci therefore it was speculated that in this
study, the GIca values do not reflect the actual interfacial properties.
Another reason for not seeing any trend in Gc with an increasing concentration of
PS(NH2) chains within each experiment was concluded to be due to the washing off of the
chains caused by the hydrolysis of the siloxane bonds.
Even though there was no trend in the interfacial fracture toughness with increasing
concentration of amino-functional polystyrene chain concentration, Gc obtained by the
ADCT gives a clear indication of the presence of PS(NH2) chain links between the epoxy
and the silicon substrates because Gc of the samples with the plain PS chains was lower
than Gc of the samples with the mixed PS/PS(NH2) chains. In addition, a correlation
between the polar component of the surface energy (γpolar) of the initial PS/PS(Br) layers
and Gc was observed. It was hypothesised that even though these properties might not be
connected, the results demonstrate the increase of Gc with increasing γpolar.
141
6.2 Suggestions for Future Work
The relationship between the interfacial toughness of cross-linked polymers and
molecular coupling to a solid substrate is a field of great significance. In this work, more
light was shed on the topic, but additional research is needed in order to gain more
understanding of what is happening in the mixed polymer monolayers during the deposition
and the chemical modification procedures.
As described in Chapter 5, the characterisation of mixed polymer layers grafted to
Si surfaces is very challenging because of the low concentration of the amino groups in the
polymer monolayer. One possible solution could be to deposit a mixture of deuterated
polystyrene (dPS) and PS(Br), and analyse the resulting monolayers of dPS/PS(NH2) by
neutron reflectivity (NR), a method sensitive to the difference between deuterium and
hydrogen atoms. Also, the initial concentration of PS(Br) in a monolayer could be detected
by x-ray reflectivity (XR). This analytical method is sensitive to the difference in the
number of electrons, and therefore it would be sensitive to the presence of electron rich
bromine molecules. Both NR and XR along with attenuated total reflection infrared
spectroscopy, were used by other researchers in a study of hydrothermal degradation of
trimethoxysilane films deposited on Si wafers1. Nevertheless, because of the very low
concentrations of the functionalised polymer chains in the present study, the analysis would
be very challenging.
If the lower grafting density is the main reason for washing off of the chains, it
would be possible to achieve the higher σ by depositing the plain PS-SiCl2 on the silicon
surface first and than carrying out the bromination procedure. In order to carry out the
reaction quantitatively, free polymer would also have to be added to the bromination
solution in order to overcome the problem of a very low concentration of chemisorbed PS
chains. Because bromine is a relatively large molecule, the steric hindrance might cause a
difference between the degree of bromination of the free polymer and that of the
chemisorbed chains. The content of bromine in the chemisorbed polymer layer would
142
therefore have to be analysed. As mentioned above, X-ray reflectivity might be a suitable
technique for such analysis because of its sensitivity to electron rich bromine molecules. By
using this method of depositing PS chains first and then carrying out the bromination
procedure, all chains would be brominated. The only factor that could be varied in such
experiments is the degree of bromination of each polymer chain. In this study, however, the
concentration of brominated chains needed to be varied. Therefore, this approach would not
produce the surfaces required for the adhesion studies.
Also, growing functionalised chains using the “living” free polymerisation
techniques, including ATRP and reversible addition-fragmentation chain transfer (RAFT)
could be reassessed. The main advantage of these methods is that the functional groups are
incorporated into the growing polymer chains during the synthesis. This eliminates any
need for an additional chemical modification that might disrupt the siloxane bonds. The
Grafting from approach used in these methods usually yields polymer brushes of a higher
areal density and therefore the resulting monolayers are more resistant towards various
chemical environments. Some very recent progress was made in the field of molecular
brushes grafted from solid surfaces by ATRP2. The authors investigated the initiation
efficiency of n-butylacrylate brushes grown from poly(2-(2-bromopropionyl)oxyethyl
methacrylate) macroinitiator. They found that in comparison to the linear polymerisation of
the free polymer chains, the initiation in the grafting from polymerisation was not
quantitative at low conversion. The authors attributed this behaviour to the congested
environment. This could explain the lack of control reported in the present study, where the
molar mass of the polymer brushes grown by ATRP from the silicon surfaces was much
lower than the molar mass of the free polymer chains of the same composition. The authors
were able to enhance the initiation efficiency by increasing the rate of deactivation of the
growing species or decreasing the rate of propagation by increasing the concentration of the
catalyst (CuBr2) or by reducing the concentration of the monomer. It should be noted that
the polymer chains were grown from polymeric substrates and not from inorganic
substrates, as used in the present study.
143
Utilising the procedures from these recent studies1,2 would allow for growing the
mixed polymer layers from the silicon substrate in a controlled manner which would then
open the door for more systematic studies and understanding in the field of the interfacial
toughness between the cross-linked polymers and solid inorganic substrates.
Very recent references indicate that conditions under which surface initiated ATRP
are curried out are different to conditions for bulk ATRP3,4. At the time of writing these
thesis this information was not known.
6.3 References
1 H. Yim, M. S. Kent, D. R. Tallant, M. J. Garcia, and J. Majewski, Langmuir 21,
4382-4392 (2005). 2 J. J. Bekonski, E. J. Kramer, H. Yim, M. S. Kent, and J. Hall, Langmuir 20, 3246-
3258 (2004). 3 C. Xu, T. Wu, Y. Mei, C. M. Drain, J. D. Batteas, and K. L. Beers, Langmuir 21,
11136-11140 (2005). 4 J. Pietrasik, B. Cusick, T. Kowalewski, and K. Matyjaszewski, Polymer Preprints
46, 335-336 (2005).
Appendices
147
APPENDIX A: Universal calibration curve for a GPC column 1
1 M. P. Stevens, Polymer Chemistry (Oxford University Press, New York,
1990).
148
APPENDIX B: GPC and 1H-NMR results of PS and its copolymers
prepared by NMP
1H-NMR of PS synthesised by NMP using TEMPO
GPC of PS synthesised by NMP using TEMPO
149
1H-NMR of PHEMA/PS synthesised by NMP using TEMPO
GPC of PHEMA/PS synthesised by NMP using TEMPO
150
1H-NMR of PS/PS(NH2) synthesised by NMP using TEMPO
1H-NMR of PS/PS(NH2) synthesised by NMP using TEMPO D2O was added to an NMR solution to observe a disappearance of NH2 peak around 3.5 ppm.
1H-NMR of monomeric aminostyrene
151
GPC of PS/PS(NH2) synthesised by NMP using TEMPO
152
APPENDIX C: 1H-NMR spectra of PS(NH2)(Boc)/PS copolymer
1H-NMR of PS(NH2)(Boc)/PS synthesised by NMP using TEMPO
1H-NMR of synthesised Boc-Aminostyrene monomer
1H-NMR of commercial Aminostyrene monomer
153
APPENDIX D: Results for the copolymers prepared by ATRP
154
APPENDIX E: Schematics for synthesis of an active and an in-active
NMP initiators
Active Initiator
In-Active Initiator
ClCl
O
N
O
ON
HR2SiCl
H2PtCl6
O ON
SiR
ClR
NaH
OH
Alkoxyamine precursor
Surface-active alkoxyamine
NO+
TEMPO
Chloromethyl- styrene
Br
NaH
HO
O
HR2SiCl
H2PtCl6
O
Si R
Cl
R