Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
Bio-tribology of Plant Cell Walls: Measuring the
interactive forces between cell wall components
Grace Dolan
A thesis submitted for the degree of Doctor of Philosophy at
The University of Queensland in 2017
School of Chemical Engineering
i
Abstract
Plants have naturally evolved complex cell wall structures to give mechanical and friction
properties that facilitate plant growth and development. A biomimetic approach is employed to
gain insight into how plants are able to lubricate moving surfaces at multiple length scales. The
outcomes of this study advance the fundamental scientific understanding of plant cell wall biology.
New insights provided here may have significant implications for optimising the value of plant
material as a food source, biofuel precursor, and as a model for functional biomaterial design,
particularly for medical applications. A review of the plant cell wall structure, mechanics, and
extension processes is used to identify the forces and tribological contacts that are relevant for plant
growth. Plant cell walls are essentially hydrogel composites of cellulose fibres within a matrix of
biopolymers (e.g. hemicelluloses, pectin) and water. Plant tissue is comprised of a cluster of plant
cells where adjacent cell walls are mediated by a pectin rich middle lamella layer. Plant growth is
initiated by expansins which are proteins that disrupt cellulose fibre contact points, leading to the
extension of the cell wall matrix. As cells expand within the tissue structure, a sliding contact forms
between adjacent walls that are extending at different rates. The two critical length scales that are
considered here to influence plant growth are the contact between cellulose nano-fibres in the cell
wall matrix, and the sliding interface between adjacent cell walls.
A large component of this thesis is the development of techniques to mimic the two tribological
contacts; that is, fibre-fibre and cell-cell interactions. The sliding interface between two surfaces is
achieved using a rotational rheometer, which also allows in situ material characterisation of the
surfaces with pre-compression, relaxation, and oscillatory shear mechanical testing steps. The
interactive forces between nano-fibres are measured directly using an Atomic Force Microscope
(AFM) tip to laterally pull fibres out of a network. Bacterial cellulose networks are used as a model
system that is compatible with the developed techniques. Bacterial cellulose is a good model
because of the structural similarity to plant cellulose, and its ability to grow as a self-assembled
random fibre network with the shape and dimensions controlled by the vessel within which it is
grown. The lubricating role of individual cell wall components at the two tribological contacts is
examined; including arabinoxylan (AX), xyloglucan (XG), pectin, and expansins. This is achieved
ii
by growing composite bacterial cellulose networks with AX and XG, and by adding pectin or
expansin solutions to the liquid medium surrounding the pre-formed cellulose networks during
testing.
The unique aspect of this mechanical study is that the experimental results are analysed using
computation modelling. Appropriate models are used to simulate the behaviour of fibrous
assemblies at multiple length scales, and under conditions akin to the experimental set up. The
modelling component advances the interpretation of experimental results, down to the exact
contribution of individual cell wall components. One major discovery is that XG reduces the
adhesion between cellulose fibres at the nano-scale. The result is consistent with the macro-scale
measurement of static friction between two cellulose hydrogels, which is shown to be driven by
cellulose fibres interacting at the interface, and is reduced in the presence of XG. Expansins are
found to act in a similar way to XG in that they reduce the static friction between cellulose hydrogel
surfaces. Finally, pectin in solution acts as a viscous film that increases the separation between
hydrogel surfaces, and the reduced surface contact directly reduces the measured static friction.
In this thesis, I develop a suite of techniques (experimental and computational) that enable multi-
scale characterisation of soft matter systems comprising fibrous assemblies. The key outcome of
this work is the direct measurement of input parameters for the development of a multi-scale 3D
mechanical model of fibre networks. Such a model provides a predictive tool for enhancing our
understanding of the underlying principles that explain plant growth; which is hypothesised to be
driven by fibre-fibre interactions within the cell wall, and cell-cell friction.
iii
Declaration by author
This thesis is composed of my original work, and contains no material previously published or
written by another person except where due reference has been made in the text. I have clearly
stated the contribution by others to jointly-authored works that I have included in my thesis.
I have clearly stated the contribution of others to my thesis as a whole, including statistical
assistance, survey design, data analysis, significant technical procedures, professional editorial
advice, and any other original research work used or reported in my thesis. The content of my thesis
is the result of work I have carried out since the commencement of my research higher degree
candidature and does not include a substantial part of work that has been submitted to qualify for
the award of any other degree or diploma in any university or other tertiary institution. I have
clearly stated which parts of my thesis, if any, have been submitted to qualify for another award.
I acknowledge that an electronic copy of my thesis must be lodged with the University Library and,
subject to the policy and procedures of The University of Queensland, the thesis be made available
for research and study in accordance with the Copyright Act 1968 unless a period of embargo has
been approved by the Dean of the Graduate School.
I acknowledge that copyright of all material contained in my thesis resides with the copyright
holder(s) of that material. Where appropriate I have obtained copyright permission from the
copyright holder to reproduce material in this thesis.
iv
Publications during candidature
Peer-reviewed papers:
• G. K. Dolan, G. E. Yakubov, G. W. Greene, N. Amiralian, P. K. Annamalai, D. J. Martin
and J. R. Stokes, Langmuir, 2016, DOI: 10.1021/acs.langmuir.6b03467.
Publications included in this thesis
No publications included.
v
Contributions by others to the thesis
This thesis was completed after review and editing suggestions from my RHD supervisors;
Professor Jason Stokes and Dr Gleb Yakubov.
Some of the analysis in Chapter 4: Section 4.3.1 is completed using a published poroelastic
mechanical model, for which the corresponding MATLAB files in Appendix E are provided by Dr
Mauricio Bonilla.
The ComsolTM Multiphysics model described in Chapter 4: Section 4.2.3 is constructed by Dr
Mauricio Bonilla.
SEM and TEM images in Chapter 5 are provided by Dr Nasim Amiralian (Australian Institute of
Biotechnology and Nanotechnology (AIBN), The University of Queensland), Dr George Greene
(Deakin University), and the Centre for Microscopy and Microanalysis (CMM) and the University
of Queensland.
For Chapter 5 Dr George Greene (Deakin University) supplied SPEEK fibre samples and
suggestions for how to analysis their interactions, Professor Darren Martin’s research group (AIBN)
supplied PVA, CNC and CNF fibre samples.
Dr Gleb Yakubov assisted in the analysis of the two different pulling scenarios in Chapter 5:
Section 5.3.1 and the calculation of the Hamaker constants in Chapter 5: Section 5.3.3.
Dr Mauricio Bonilla prepared the ComsolTM Multiphysics model and helped with its
implementation in Chapter 6: Section 6.3.2.
The AFM images in Chapter 6: Section 6.3.1 were performed by Dr Gleb Yakubov.
The expansins and the cellulose hydrogels used in Chapter 7 are provided by Professor Daniel
Cosgrove’s research group at Pennsylvania State University (PSU), PA, USA. The results in
Section 7.3.1 and most of Section 7.3.2 are from experiments performed at PSU.
vi
Statement of parts of the thesis submitted to
qualify for the award of another degree
None.
vii
Acknowledgements
My most sincere thanks go to my RHD advisors, Professor Jason Stokes and Dr Gleb Yakubov,
who provided constant support, guidance, and mentorship throughout my PhD. They have been
very generous with their time and resources and I am forever grateful. I have learnt an immense
amount from their wealth of knowledge in terms of scientific insights and research skills.
I thank a number of people for their technical support with training on instruments and procedures.
Particularly Dr Dongjie Wang and Deirdre Mikkelsen for their help with preparing bacterial
cellulose materials, Dr Lien Chau who provided assistance with all of my fabrication needs, Dr
Elena Taran who trained me on the AFM and confocal microscope, Dr Michael Boehm who trained
me on the Rheometer.
I would like to thank Professor Jason Stokes Lab, The Microbiology Lab at the Centre for Nutrition
and Food Science (CNAFS), and the Australian National Fabrication Facility Queensland at the
University of Queensland for letting me use their equipment.
Thank you to Professor Daniel Cosgrove’s and Federico Harte’s research groups who supported by
lab visit at PSU.
Thank you to the members of Professor Jason Stokes’ research group, and the Australian Research
Council (ARC) Centre of Excellence (CoE) in Plant Cell Walls for their support and constructive
feedback throughout my PhD. In particular I would mention the consistent and helpful feedback
given by Professor Mike Gidley.
I acknowledge funding from the Australian Postgraduate Award, UQ Advantage Top-up
Scholarship, ARC CoE in Plant Cell Walls (managed by Professor Jason Stokes), and the Graduate
School International Travel Award (GSITA).
viii
Keywords
Cellulose, plant cell walls, tribology, hydrogels, nano-fibres, adhesion, mechanics, poroelastic,
modelling, multi-scale.
Australian and New Zealand Standard
Research Classifications (ANZSRC)
ANZSRC code: 091209, Polymers and Plastics, 40%
ANZSRC code: 090408, Rheology, 30%
ANZSRC code: 029901, Biological Physics, 30%
Fields of Research (FoR) Classification
FoR code: 0912, Material Engineering, 40%
FoR code: 0904, Chemical Engineering, 30%
FoR code: 0299, Other Physical Sciences, 30%
ix
Table of Contents
Abstract ................................................................................................................................................ i
List of Figures and Tables .............................................................................................................. xiii
List of Abbreviations .................................................................................................................... xviii
Introduction ........................................................................................................................................ 1
1.1 Motivation .................................................................................................................................. 1
1.2 Aims and Objectives .................................................................................................................. 2
1.1 Thesis Outline ............................................................................................................................. 4
Literature Review .............................................................................................................................. 6
2.1 Bio-tribology and Bio-lubrication .............................................................................................. 6
2.2 Plant Cell Wall Structure, Mechanics and Growth .................................................................... 9
2.2.1 Plant Cell Wall Architecture.............................................................................................. 10
2.2.2 Cell Wall Mechanics ......................................................................................................... 17
2.2.3 Plant Growth and Cell Wall Extension .............................................................................. 20
2.2.4 Tribological Contacts in Plant Cell Walls ......................................................................... 26
2.2.5 Bacterial Cellulose as a Model System for Plant Cell Walls ............................................ 27
2.3 Mechanical and Friction Properties of Hydrogels .................................................................... 30
2.3.1 Hydrogel Material Characterisation .................................................................................. 31
2.3.2 Relating Hydrogel Mechanics to Friction Response ......................................................... 33
2.3.3 Cellulose Hydrogel Mechanics and the Relevance to Plant Cell Walls ............................ 37
2.4 Measurement of Fibre-Fibre Interactions ................................................................................. 38
2.4.1 Fibre Network Models ....................................................................................................... 39
x
2.4.2 Fibre Mechanics ................................................................................................................ 40
2.4.3 Experimental Approaches for Fibre-Fibre Measurement .................................................. 44
2.5 Future Perspective: Scope and Goal of Thesis ......................................................................... 46
References for Chapter 2 ................................................................................................................ 49
Research Methodology .................................................................................................................... 55
3.1 Materials ................................................................................................................................... 56
3.1.1 Electrospun Fibres ............................................................................................................. 56
3.1.2 Cellulose Nano-fibres Extracted from a Plant Source ....................................................... 57
3.1.3 Bacterial Cellulose ............................................................................................................. 58
3.1.4 Pectin Solutions ................................................................................................................. 60
3.1.5 Bacterial Expansins ........................................................................................................... 61
3.2 Measurements ........................................................................................................................... 61
3.2.1 Tribo-rheological Technique in a Rotational Rheometer .................................................. 61
3.2.2 Dip-and-drag Technique in an AFM ................................................................................. 63
References for Chapter 3 ................................................................................................................ 67
Friction, lubrication, and in situ mechanics of poroelastic cellulose hydrogels ......................... 68
4.1 Introduction and Background ................................................................................................... 68
4.2 Experimental Section ............................................................................................................... 71
4.2.1 Physical characterisation of hydrogel mechanics and friction .......................................... 71
4.2.2 Modelling the hydrogel mechanics during compression-relaxation .................................. 74
4.2.3 Simulating the interface between hydrogels during compression-relaxation .................... 75
4.2.4 Pectin solution and viscosity measurements ..................................................................... 77
4.3 Results and Discussion ............................................................................................................. 78
4.3.1 Mechanical properties of poroelastic hydrogels ................................................................ 78
4.3.2 Contact area between poroelastic hydrogels ..................................................................... 85
4.3.3 Tribo-rheological response between hydrogels ................................................................. 88
xi
4.3.4 Influence of substrate mechanics on interfacial friction .................................................... 91
4.3.5 Influence of solvent viscosity on interfacial friction ......................................................... 94
4.3.6 Interfacial friction at the true contact area ......................................................................... 96
4.3.7 Stick-slip and stiction behaviour ....................................................................................... 99
4.4 Concluding Remarks .............................................................................................................. 105
References for Chapter 4 .............................................................................................................. 107
Method development for measuring the adhesive forces between individual nano-fibres ..... 109
5.1 Introduction and Background ................................................................................................. 109
5.2 Experimental Section ............................................................................................................. 111
5.2.1 Model fibre systems......................................................................................................... 111
5.2.2 Dip-and-drag technique ................................................................................................... 111
5.3 Results and Discussion ........................................................................................................... 113
5.3.1 Dip-and-drag Lateral Force Spectroscopy of SPEEK electrospun mats of varying network
density .................................................................................................................................. 113
5.3.2 Dip-and-drag Lateral Force Spectroscopy of PVA network ........................................... 122
5.3.3 Analysis of adhesive forces between fibres ..................................................................... 122
5.3.4 Dip-and-drag Lateral Force Spectroscopy of CNC and CNF networks in air and water 124
5.3 Concluding Remarks .............................................................................................................. 127
References for Chapter 5 .............................................................................................................. 128
Measuring the effect of hemicelluloses on the adhesive forces between cellulose fibres ......... 130
6.1 Introduction and Background ................................................................................................. 130
6.2 Experimental Section ............................................................................................................. 132
6.3 Results and Discussion ........................................................................................................... 132
6.3.1 Probing contacts between individual cellulose fibres ...................................................... 132
6.3.2 Simulating fibre-fibre detachment events ........................................................................ 136
6.3.3 The role of hemicellulose at contacts between cellulose fibres ....................................... 142
6.4 Concluding Remarks .............................................................................................................. 146
xii
References for Chapter 6 .............................................................................................................. 147
The effect of bacterial expansins on cellulose fibre interactions ............................................... 149
7.1 Introduction ............................................................................................................................ 149
7.2 Background on the ‘wall-loosening’ activity of expansins .................................................... 150
7.3 Experimental Section ............................................................................................................. 154
7.3.1 Pre-treatment of bacterial cellulose hydrogels ................................................................ 154
7.3.2 Mechanical assay of expansin activity using the tribo-rheological technique ................ 155
7.3.1 Mechanical assay of expansin activity using the Dip-and-drag technique ...................... 155
7.4 Results .................................................................................................................................... 155
7.4.1 The effect of bacterial expansins on the mechanics of bacterial cellulose hydrogels ..... 155
7.4.2 The effect of bacterial expansins on the friction response between pairs of bacterial
cellulose hydrogels ................................................................................................................... 158
7.4.1 The effect of bacterial expansins on adhesion between cellulose fibres ......................... 161
7.5 Discussion .............................................................................................................................. 163
7.6 Concluding Remarks .............................................................................................................. 164
References for Chapter 7 .............................................................................................................. 166
Concluding Remarks and Future Work ...................................................................................... 167
8.1 Concluding Remarks .............................................................................................................. 167
8.2 Recommendations for Future Work ....................................................................................... 169
Appendices ...................................................................................................................................... 172
Appendix A: Producing bacterial cellulose from Gluconacetobacter xylinus ............................. 172
Appendix B: Chromium Mask Fabrication .................................................................................. 178
Appendix C: SU-8 Master Fabrication ......................................................................................... 193
Appendix D: PDMS Microarray Fabrication ............................................................................... 196
Appendix E: MATLAB code for poroelastic mechanical model ................................................. 197
Appendix F: Text file of raw data for input into MATLAB code for the poroelastic mechanical
model ............................................................................................................................................ 204
xiii
Appendix G: Raw data from the compression-relaxation steps for hydrogels (cellulose, CAX,
CXG) at all CRs ........................................................................................................................... 206
Appendix H: Raw data from the tribo-rheological test for all hydrogels and solvents ................ 212
Appendix I: Cellulose concentration based on the G’ of the hydrogels for cellulose, CAX, and
CXG .............................................................................................................................................. 216
Appendix J: Friction curves with angular velocity for hydrogel pairs (cellulose, CAX, CXG) at all
CRs ............................................................................................................................................... 217
Appendix K: Force-distance data for SPEEK fibres mats............................................................ 223
Appendix L: MATLAB code for finding the peaks in lateral force-distance curves from the dip-
and-drag technique ....................................................................................................................... 226
Appendix M: MATLAB code for measure the slope before the peaks in the lateral force-distance
curves from the dip-and-drag technique ....................................................................................... 231
Appendix N: Solving 3 non-linear simultaneous equations in MATLAB ................................... 232
Appendix O: Interfacial yield stress and G’ for triplicate hydrogel pairs for each treatment
(control, YOAJ, WWY, RKKQ, D82) ......................................................................................... 233
Appendix P: Raw data from dip-and-drag experiments on Cellulose, CAX, and CXG networks
with and without expansins .......................................................................................................... 236
Appendix Q: Histogram of peak heights for Cellulose, CAX, and CXG networks with and
without expansins ........................................................................................................................ 239
xiv
List of Figures and Tables
Figure 2.1. Tribology at multiple length scales from full-film lubrication ...................................................... 7
Figure 2.2. Pictorial representation of the viscous flow equation in 2.1 ....................................................... 8
Figure 2.3. Schematic of compression between two opposing cartilage ...................................................... 9
Figure 2.4. Structure of the primary cell wall ............................................................................................... 10
Figure 2.5. Predicted 36-chain CEF model with hexagonal cross-section .................................................... 11
Figure 2.6. Common artefacts with AFM topography .................................................................................. 12
Figure 2.7. Cellulose fibril orientation relative to axis of elongation ........................................................... 13
Figure 2.8. Transmission electron micrographs of tungsten/tantalum/carbon........................................... 14
Figure 2.9. The original tethered network model ........................................................................................ 14
Figure 2.10. Distribution of pectin epitopes in the outer epidermal wall .................................................... 16
Figure 2.11. Mechanical measurement of isolated cell walls using ............................................................. 19
Figure 2.12. Time lapsed images of growing root ........................................................................................ 27
Figure 2.13. Schematic of cellulose biosynthesis by Gluconacetobacter xylinus ......................................... 28
Figure 2.14. Model of bacterial cellulose fibre based on SANS and SAXS data ........................................... 29
Figure 2.15. Bacterial Cellulose Hydrogel ..................................................................................................... 30
Figure 2.16. Typical stress-time curve for compression-relaxation ............................................................. 32
Figure 2.17. Gel-Gel friction experiment in a rotational rheometer ............................................................ 35
Figure 2.18. An aerial image and corresponding side-view ......................................................................... 36
Figure 2.19. Cellulose fibres suspended over grating .................................................................................. 41
Figure 2.20. Force-distance curve obtained near the middle ...................................................................... 42
Figure 2.21. Fibre suspended across a trench of distance L......................................................................... 42
Figure 2.22. Schematic diagram of an AFM cantilever dragging ................................................................. 43
Figure 2.23. Force-cantilever and stress-strain curves in the elastic ........................................................... 44
Figure 2.24. Experimental configurations for fibre-fibre measurements .................................................... 45
Figure 2.25. Two fibre free ends arranged in parallel and cross-cylinder .................................................... 46
Figure 3.1. Bacterial cellulose hydrogels adhered to emery paper ............................................................. 63
Figure 3.2. Test procedure for mechanical and friction characterisation .................................................... 63
Figure 3.3. Cellulose micro-hydrogels grown in PDMS micro-array ............................................................ 64
Figure 3.4. An image taken after adhering micro-gels to the glass .............................................................. 65
Figure 3.5. AFM image of glued micro-gel showing the edge ...................................................................... 65
Figure 4.1. Schematic showing the steps for characterising the mechanics................................................ 72
xv
Figure 4.2. Sensitivity analysis of the compression speed on the friction ................................................... 73
Figure 4.3. Sensitivity analysis of the rotation rate on the friction response .............................................. 73
Figure 4.4. Representation of a cellulose hydrogel disk showing that the .................................................. 74
Figure 4.5. (a) Equivalent simulation system to the double hydrogel contact ............................................ 76
Figure 4.6. Viscosity of pectin solutions (0.5, 1, 2, and 4 wt%) across the .................................................. 78
Figure 4.7. Compression-relaxation profiles of pairs of (a) cellulose ........................................................... 80
Figure 4.8. Compression profiles of cellulose, CAX, and CXG hydrogel ....................................................... 81
Figure 4.9. Two consecutive compressions of a single (a) Cellulose ............................................................ 83
Figure 4.10. Recovery of the weight of a single bacterial cellulose hydrogel .............................................. 84
Figure 4.11. Simulated film thickness at the centre, i.e. r = 0 (filled symbols) ............................................ 86
Figure 4.12. The variation in film thickness with radial position in the ....................................................... 87
Figure 4.13. Logarithmic plot of the fraction of the interface that is in contact ......................................... 87
Figure 4.14. Simulated film thickness at the centre of the interface, i.e. r = 0 ............................................ 88
Figure 4.15. Logarithmic plot of the fraction of the interface that is in contact ......................................... 89
Figure 4.16. Characteristic tribological responses for hydrogels tested in water........................................ 89
Figure 4.17. Stress-strain curve of single Cellulose hydrogel glued to both ................................................ 90
Figure 4.18. (a) Axial and (b) Radial modulus from biphasic modelling ....................................................... 93
Figure 4.19. Logarithmic plot of the interfacial yield stress against G’ ........................................................ 94
Figure 4.20. Linear relationships between G’ and axial modulus ................................................................ 94
Figure 4.21. Tribological response of pairs of cellulose hydrogels .............................................................. 95
Figure 4.22. Logarithmic plot of the interfacial yield stress against ............................................................ 96
Figure 4.23. The corrected interfacial shear stress (τc) plotted against ....................................................... 98
Figure 4.24. Interfacial yield stress versus the cellulose concentration ...................................................... 99
Figure 4.25. Characteristic friction behaviours during the constant rotation ........................................... 100
Figure 4.26. Characterisation of stick-slip behaviour showing the elastic ................................................. 101
Figure 4.27. (a) Shear stress over time for constant rotation (CAX ........................................................... 101
Figure 4.28. Shear stress () and angular velocity () measured over time .............................................. 102
Figure 4.29. The peak in angular velocity (m during the slip cycle ......................................................... 103
Figure 4.30. The slope of the slip cycle, ksl, plotted against the peak in .................................................... 104
Figure 4.31. Scatterplot of the G’ against viscosity for the different sliding .............................................. 104
Figure 4.32. Shear stress and angular velocity measured over time for .................................................... 105
Figure 5.1. Microscopy images of nano-fibrous networks. (a) SEM .......................................................... 112
Figure 5.2. The AFM tip is engaged with the substrate at a constant ....................................................... 113
xvi
Figure 5.3. Typical lateral force-distance curves for SPEEK samples ......................................................... 115
Figure 5.4. (a) and (b) show SEM images superimposed with proposed ................................................... 116
Figure 5.5. Analysis of example force distance curve for SPEEK sample ................................................... 117
Figure 5.6. SEM micrographs at 10 000 x magnification of electrospun ................................................... 118
Figure 5.7. (a) Illustrations of two possible scenarios for pulling a fibre ................................................... 121
Figure 5.8. (a) Representative force-distance curve for PVA and .............................................................. 122
Figure 5.9. Representative force-distance curves for (a) CNC in air .......................................................... 126
Figure 6.1. (a) and (b) AFM images of an air-dried cellulose network ....................................................... 133
Figure 6.2. Lateral force-distance curve showing a typical peak that ........................................................ 135
Figure 6.3. (a) Example force-distance curve for cellulose fibre network ................................................. 135
Figure 6.4. Force balance across a section of the fibre network ............................................................... 137
Figure 6.5. Simplified setup of the system depicted in Figure 6.3 ............................................................. 138
Figure 6.6. Predicted force curves for combinations of 2 different ........................................................... 139
Figure 6.7. Best surface fit describing the functional relationship ............................................................ 140
Figure 6.8. The slope, s, of the experimental force-distance curve ........................................................... 141
Figure 6.9. (a) Example force-distance curve for CAX network ................................................................. 143
Figure 6.10. SEM images of (a) cellulose, (b) CAX, and (c) CXG ................................................................. 143
Figure 6.11. Distribution of experimental values of the slope, s, for ......................................................... 144
Figure 6.12. Depiction of the contact between two cellulose fibres ......................................................... 145
Figure 7.1. (1) Expansin (red pacman) binds to cellulose fibres (brown lines) .......................................... 150
Figure 7.2. Protein structure of expansin, showing two distinct domains................................................. 151
Figure 7.3. Activities of selected variants of bacterial expansin protein ................................................... 151
Figure 7.4. Activities of variants of bacterial expansin protein on ............................................................. 152
Figure 7.5. Compression curves of (a) cellulose hydrogel pair treated ...................................................... 157
Figure 7.6. Shear stress over time for a constant rotation rate for the ..................................................... 159
Figure 7.7. Interfacial yield stress against G’ for pairs of bacterial ............................................................ 160
Figure 7.8. IYSR is the interfacial yield stress measured in the presence ................................................... 161
Figure 7.9. The pull-off force is calculated from the Jarzynski’s average .................................................. 163
Figure 8.1. Design for shape of cellulose networks grown in PDMS mould .............................................. 170
Figure G.1. Normal stress during compression-relaxation steps for .......................................................... 206
Figure G.2. Normal stress during compression-relaxation steps for .......................................................... 206
Figure G.3. Normal stress during compression-relaxation steps for .......................................................... 207
Figure G.4. Normal stress during compression-relaxation steps for .......................................................... 207
xvii
Figure G.5. Normal stress during compression-relaxation steps for .......................................................... 208
Figure G.6. Normal stress during compression-relaxation steps for .......................................................... 208
Figure G.7. Normal stress during compression-relaxation steps for .......................................................... 209
Figure G.8. Normal stress during compression-relaxation steps for .......................................................... 209
Figure G.9. Normal stress during compression-relaxation steps for .......................................................... 210
Figure G.10. Normal stress during compression-relaxation steps for ....................................................... 210
Figure G.11. Normal stress during compression-relaxation steps for ....................................................... 211
Figure G.12. Normal stress during compression-relaxation steps for ....................................................... 211
Figure H.1. Friction curves for Cellulose hydrogel pairs in water .............................................................. 212
Figure H.2. Friction curves for CAX hydrogel pairs in water ...................................................................... 212
Figure H.3. Friction curves for CXG hydrogel pairs in water ...................................................................... 213
Figure H.4. Friction curves for a pair of cellulose hydrogels in 0.5 wt% .................................................... 213
Figure H.5. Friction curves for a pair of cellulose hydrogels in 1 wt% ....................................................... 214
Figure H.6. Friction curves for a pair of cellulose hydrogels in 2 wt% ....................................................... 214
Figure H.7. Friction curves for a pair of cellulose hydrogels in 4 wt% ....................................................... 214
Figure J.1. Shear stress and angular velocity measured over time ............................................................ 217
Figure J.2. Shear stress and angular velocity measured over time ............................................................ 217
Figure J.3. Shear stress and angular velocity measured over time ............................................................ 218
Figure J.4. Shear stress and angular velocity measured over time ............................................................ 218
Figure J.5. Shear stress and angular velocity measured over time ............................................................ 219
Figure J.6. Shear stress and angular velocity measured over time ............................................................ 219
Figure J.7. Shear stress and angular velocity measured over time ............................................................ 220
Figure J.8. Shear stress and angular velocity measured over time ............................................................ 220
Figure J.9. Shear stress and angular velocity measured over time ............................................................ 221
Figure J.10. Shear stress and angular velocity measured over time .......................................................... 221
Figure J.11. Shear stress and angular velocity measured over time .......................................................... 222
Figure J.12. Shear stress and angular velocity measured over time .......................................................... 222
Figure K.1. Lateral force-distance curve of SPEEK sample A ...................................................................... 223
Figure K.2. Lateral force-distance curve of SPEEK sample B ...................................................................... 223
Figure K.3. Lateral force-distance curve of SPEEK sample C ...................................................................... 224
Figure K.4. Lateral force-distance curve of SPEEK sample D ...................................................................... 224
Figure K.5. Lateral force-distance curve of SPEEK sample E ...................................................................... 225
Figure O.1. Interfacial yield stress against G’ for 2 different ..................................................................... 233
Figure O.2. Interfacial yield stress against G’ for triplicate ........................................................................ 233
xviii
Figure O.3. Interfacial yield stress against G’ for triplicate ........................................................................ 234
Figure O.4. Interfacial yield stress against G’ for triplicate ........................................................................ 234
Figure O.5. Interfacial yield stress against G’ for triplicate ........................................................................ 235
Figure O.6. Interfacial yield stress against G’ for triplicate ........................................................................ 235
Figure P.1. Lateral deflection-distance curve for Cellulose network ......................................................... 236
Figure P.2. Lateral deflection-distance curve for Cellulose network ......................................................... 236
Figure P.3. Lateral deflection-distance curve for CAX network ................................................................. 237
Figure P.4. Lateral deflection-distance curve for CAX network ................................................................. 237
Figure P.5. Lateral deflection-distance curve for CXG network ................................................................. 238
Figure P.6. Lateral deflection-distance curve for CXG network ................................................................. 238
Figure Q.1. Histogram of peaks heights (n = 166) from lateral .................................................................. 239
Figure Q.2. Histogram of peaks heights (n = 200) from lateral .................................................................. 239
Figure Q.3. Histogram of peaks heights (n = 158) from lateral .................................................................. 240
Figure Q.4. Histogram of peaks heights (n = 183) from lateral .................................................................. 240
Figure Q.5. Histogram of peaks heights (n = 64) from lateral .................................................................... 241
Figure Q.6. Histogram of peaks heights (n = 103) from lateral .................................................................. 241
Table 2.1. A summary of key structural differences in cellulose fibres ....................................................... 30
Table 2.2. Summary of key findings from relevant studies on hydrogel ..................................................... 34
Table 3.1. Composition of 300 mL of Liquid HS medium ............................................................................. 59
Table 4.1. Material functions in the mechanical model ............................................................................... 74
Table 4.2. Mechanical parameters of pairs of cellulose and ........................................................................ 81
Table 4.3. Hysteresis areas for consecutive compressions of Cellulose ...................................................... 84
Table 4.4. The apparent linear viscoelastic moduli (G’, G’’ in kPa) of the ................................................... 85
Table 5.1. The exponentially averaged peak height values from a set ...................................................... 120
Table 5.2. The experimental and theoretical values of the Hamaker ........................................................ 125
Table 5.3. The Jarzynski’s average peak height value for CNC and CNF .................................................... 126
Table 6.1. Model parameters extracted from equations 6.5 - 6.7 using ................................................... 144
Table 7.1. The storage (G’) and loss (G”) moduli of hydrogel pairs that .................................................... 158
Table 7.2. The storage (G’) and loss (G”) moduli of hydrogel pairs that .................................................... 158
xix
List of Abbreviations
AX – Arabinoxylan
XG – Xyloglucan
DLVO – Derjaguin, Landau, Vervey, and Overbeek
AFM – Atomic Force Microscope
CEF – Cellulose Elementary Fibril
SANS – Small Angle Neutron Scattering
WAXS – Wide Angle X-ray Scattering
SAXS – Small Angle X-ray Scattering
NMR – Nuclear Magnetic Resonance
MD – Molecular Dynamics
HG – Homogalacturonan
RG – Rhamnogalacturonan
DE – Degree of Esterification
FEM – Finite Element Modelling
PME – Pectin Methyl Esterases
SEM – Scanning Electron Microscopy
XRD – X-ray Diffraction
SFA – Surface Force Apparatus
FT-IR – Fourier transform infrared
SPEEK – Sulfonated polyether ether ketone
PVA – Polyvinyl Alcohol
xx
CNF – Cellulose nanofibrils
CNC – Cellulose nanocrystals
TEM – Transmission Electron Microscopy
HS – Hestrin and Schramm
PDMS – Polydimethylsiloxane
RO – Reverse Osmosis
SAOS – Small Amplitude Oscillatory Shear
CR – Compression ratio
CZM – Cohesive Zone Model
PPBS – Putative Polysaccharide Binding Site
1
Chapter 1
Introduction
1.1 Motivation
Natural systems have evolved ways of controlling friction in water-based environments far
superior to engineered solutions which rely on the utility of oil-based lubricants. Animals
implement many self-lubricating systems such as observed in the gait of a snail, the passage
of food in the human digestive tract, and in synovial joints. These phenomena have been the
subject of biomimetic studies which aim to gain insight into the mechanistic details and
recreate these lubricating functions in man-made systems. Plants have received significantly
less attention among researchers for their equally remarkable tribological feats. Plant
tribology is readily observed in the self-cleaning ability of the lotus leaf which has surface
microstructures that control the spreadability of water droplets, and in the way plant roots
penetrate soil by shedding an external layer of cells which releases a water-based mucilage
material that lines the root channel. Plant materials have a desirable combination of high
strength and light weight material properties. Furthermore, plants are able to grow on long
time scales and respond to environmental changes (e.g. water pressure and wind) on short
time scales.
The major difference between plant and animal systems is that plant cells are surrounded by a
cell wall. Plant cell walls are remarkably complex composite materials that have a highly
controlled composition and architecture to facilitate a combination of high tensile strength for
resisting mechanical forces and extensibility for plant growth. During plant growth, turgor
pressure within cells is converted into mechanical energy that drives the cell wall matrix to
yield and extend as cellulose fibrils and surrounding polymers move relative to each other.
CHAPTER 1. INTRODUCTION Grace Dolan, PhD Thesis
2
Irreversible cell wall extension is typically initiated by the action of a wall protein called
expansin. A pectin-rich layer is sandwiched between neighbouring cell walls, facilitating the
coordinated expansion of adjoining cells by controlling friction and adhesion between the
surfaces. It is reasonable to consider that at multiple length scales within plant systems,
lubrication mechanisms are utilised for overcoming friction forces arising within and between
wall structures that would otherwise limit their growth and development. Evidence for these
critical tribological contacts and further details of cell wall architecture and plant growth
processes are reviewed extensively in Chapter 2.
Plant biotribology is a new, challenging and exciting area with a lot of potential for gaining
insight into natural lubrication and the fundamental science underpinning these mechanisms.
A major challenge in studying plant systems is experimentally achieving the length and time
scales relevant for plant growth processes. Based on advances in biotribology and
measurement capabilities, researchers are now in a position to tackle complex plant systems.
This will include developing techniques to mimic conditions akin to the growing plant cell
wall using model systems and controlled experimental parameters for a systematic approach.
The primary approach taken in this thesis is to investigate model fibrous systems with
material properties analogous to plant cell walls. I seek to apply novel approaches for
investigating gel-gel and fibre-fibre friction, as these are key tribological elements in the
context of plant growth. The unique aspect of this thesis is the use of computational
modelling to analyse the experimental results. In this way, key findings from the systematic
testing of model systems are used to generate universal relationships that apply to more
complex systems such as plant cell walls.
1.2 Aims and Objectives
A combined experimental and computational approach is used to gain insight into the
potential tribological interactions occurring within and between plant cells during growth and
mechanical deformation. This is divided in three more focussed research targets:
1. To consider the shear forces between plant cells by investigating the friction between
composite cellulose hydrogels as cell wall mimics.
➢ Develop a technique using model surfaces that represents the sliding contact
between adjacent cell walls and allows in situ mechanical characterisation.
CHAPTER 1. INTRODUCTION Grace Dolan, PhD Thesis
3
➢ Use a poroelastic mechanical model to analyse experimental results and determine
the material functions that directly influence the mechanical and friction
responses.
2. To consider critical forces between cellulose fibres and their contribution to network
mechanics under elastic and plastic deformations.
➢ Develop a technique to probe the interactive forces (elastic, adhesive, and friction)
between nanofibers in model fibre networks.
➢ Use complementary simulations to determine the material parameters that directly
influence the measured adhesion between fibres.
3. To probe the influence of non-cellulosic plant polysaccharides (arabinoxylan,
xyloglucan and pectin) and growth proteins (expansins) on fibre-fibre and gel-gel
level interactions and infer potential lubricating mechanisms of individual cell wall
components.
➢ Apply the developed technique for measuring gel-gel friction to pairs of
composite cellulose hydrogels incorporating arabinoxylan (AX) and xyloglucan
(XG). Use poroelastic mechanical modelling to precisely determine the effect of
AX and XG on the interaction between cellulose fibres at the interface.
➢ Apply the developed technique for measuring gel-gel friction to pairs of cellulose
hydrogels separated by a film of pectin solution with varying concentration. Using
the poroelastic mechanical model and squeeze film lubrication theory, the
influence of solvent viscosity on the interaction between cellulose fibres at the
interface is determined.
➢ Apply the developed technique for measuring gel-gel friction to pairs of cellulose
hydrogels in the presence of expansins to assess whether they affect the formation
of cellulose fibre contacts at the interface.
➢ Apply the developed technique for measuring fibre-fibre interactive forces to
composite cellulose fibre networks incorporating AX and XG to investigate the
effect of these polymers on the adhesion force at fibre contacts.
➢ Apply the developed technique for measuring fibre-fibre interactive forces to
cellulose fibre and composite networks in the presence of expansins to investigate
their effect on preformed fibre contacts.
CHAPTER 1. INTRODUCTION Grace Dolan, PhD Thesis
4
1.3 Thesis Outline
To address the aims and objectives listed above the thesis is broken down into the chapters
briefly described below.
Chapter 2 (Literature review): The structure and mechanics of the primary plant cell wall,
and the process of cell wall extension during plant growth are reviewed in detail. From the
combined evaluation of the cell wall structure and plant growth processes, the interface
between adjoining cell walls and the contact between individual cellulose fibres are identified
as important tribological contacts. In order to mimic these tribological contacts, experimental
techniques for measuring the friction between soft surfaces and measuring interactive forces
between nanofibers are reviewed. Due to the complexity of plant cell walls and the difficulty
performing controlled physical measurement on plant materials, the appropriateness of
bacterial cellulose as a model system is evaluated. Plant cellulose has a crystalline core
surrounded by a paracrystalline shell. The chemical and mechanical methods of extracting
plant cellulose disrupt the paracrystalline layer, which is the region that facilitates cellulose
interactions with the hemicellulose amorphous phase in plant cell walls. Bacterial cellulose
has a similar structure to plant cellulose in terms of the crystalline and paracrystalline phases
which makes it a good model for studying the interactions between cellulose and
hemicelluloses. In order to interpret the multi-scale mechanics of plant cell walls and
corresponding model systems, existing computational models that describe poroelastic
materials and fibre network assemblies are reviewed. Finally, a future perspective on this
field of research is presented.
Chapter 3 (Research Methodologies): In this section the methods for preparing bacterial
cellulose and composite hydrogels are outlined. The techniques that are developed for
measuring cell-cell and fibre-fibre interactions are summarised, with more specific protocol
included in the relevant results chapters.
Chapter 4 (Friction, lubrication and in situ mechanics of poroelastic cellulose hydrogels): A
rheo-tribological technique is developed using a rotational rheometer to measure the in situ
mechanics and friction response between pairs of poroelastic cellulose hydrogels. For the first
time, a poroelastic mechanical model is used to predict the area of contact at the interface as
pairs of hydrogels are compressed together prior to shearing. By accounting for the contact
area, the direct effect of cell wall components (AX, XG, and pectin) on the interactions
CHAPTER 1. INTRODUCTION Grace Dolan, PhD Thesis
5
between cellulose fibres at the interface is determined. The key finding from this chapter
leads to the hypothesis that XG reduces the adhesion between cellulose fibres, which is tested
in Chapter 6. This chapter gives insight into how the composition and assembly of plant cell
walls influence the shear forces experienced at the interface between adjacent cell walls.
Chapter 5 (Method development for measuring the adhesive forces between individual nano-
fibres): This chapter details the development of a novel ‘Dip-and-drag’ technique in the
Atomic Force Microscope (AFM) for measuring the adhesive force between individual nano-
fibres. Model electrospun fibres are used to validate the technique due to their cylindrical
shape with relatively consistent dimensions, the ability to control network density, and the
fact that adhesive forces between the fibres are dominated by DLVO interactions. The
experimentally determined adhesive force is compared to the theoretically predicted value to
validate the technique.
Chapter 6 (Measuring the effect of hemicelluloses on the adhesive forces between cellulose
fibres): The Dip-and-drag technique developed in Chapter 5 is applied the bacterial cellulose
networks to measure the adhesive force between individual fibres. Composite networks with
AX and XG are also tested to measure the adhesive force at cellulose fibre contacts that are
mediated by these molecules. Simulations of fibre detachment events in silico are included in
this study to determine the specific effect of AX and XG on the physical properties of fibre
contacts. The key finding from this chapter is that XG reduces the adhesion between cellulose
fibres which is consistent with the results from the rheo-tribological technique in Chapter 4.
Chapter 7 (The effect of bacterial expansins on cellulose fibre contacts): the rheo-tribological
(Chapter 4) and Dip-and-drag (Chapter 5) techniques developed in this thesis provide two
mechanical assays for measuring the interaction between cellulose fibres. Thus the techniques
are applied to investigate the action of expansins on cellulose fibre contacts. In this chapter
the activity of native and mutant bacterial expansins at contacts between bacterial cellulose
fibres is evaluated to gain insight into the potential mechanism of action of plant expansins
on cell walls during growth.
Chapter 8 (Concluding remarks and future work): Major findings from this thesis are
presented with their key implications in the context of plant cell wall mechanics and growth
processes. Suggestions for future work in this research area are listed.
6
Chapter 2
Literature Review
2.1 Bio-tribology and Bio-lubrication
Tribology concerns the friction, wear and lubrication of interacting surfaces in relative
motion, whether it is rolling, sliding, normal approach or separation of surfaces1. A rubbing
contact formed by two solids and separated by a fluid is a dynamic system. The gap between
surfaces depends on rubbing speed, surface adhesion and mechanics, rheological properties
of the fluid, and presence of surface bound/adsorbed molecular layers. The forces that
dominate friction depend on the length scales of the gap relative to surface roughness, and the
range of surface forces.
Figure 2.1 highlights the system parameters that are critical for describing the tribological
response across a generic fluid film thickness range. The right-hand limit represents the
presence of a very thin or discontinuous fluid film between moving surfaces, where the
contact is mediated by surface bound or adsorbed molecular layers. This is referred to as the
boundary lubrication regime which is characteristic for system conditions of low fluid
viscosity, small relative velocity, and/or high normal load2. In this regime the surfaces are
engaged into direct contact and friction depends on the effective area of interacting asperities
and the energy of adhesion between them. Surface-bound molecules may enhance lubrication
by generating a repulsive force through some physical means, or conversely may increase
friction and adhesive forces through attractive interactions2-4.
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
7
Figure 2.1. Tribology at multiple length scales from full-film lubrication (left hand side) to
surface contact (right hand side). Full film lubrication is dominated by the properties of the
lubricant, specifically its viscosity. In the absence of a fluid film the friction is dominated by
the adhesive interaction between surfaces that are in contact.
If the rubbing speed is high enough, a viscous liquid produces a hydrodynamic lift force
sufficient to fully separate the surfaces. This regime, called hydrodynamic lubrication, is
represented on the left-hand side of the scale in Figure 2.1. In the presence of a full liquid
film with thickness much larger than the size of asperities, the surfaces are completely
separated and do not interact. The lubrication behaviour in this regime is dominated by
rheological properties of the pressurised fluid film separating the moving surfaces2. In the
case of Newtonian liquids, the full film lubrication is predictable and can be fully described
by the lubricant bulk viscosity and geometry of the rubbing surfaces. Viscous forces for
Newtonian films are calculated according to equation 2.1, which is represented pictorially in
Figure 2.2. In between full film lubrication and boundary friction, a mixed lubrication regime
exists where the interfaces between opposing surfaces in relative motion are partially
lubricated by solvent whilst some asperity contacts occur2. Friction is minimised at fluid film
thicknesses corresponding to the transition between the mixed and hydrodynamic regime, and
is influenced by a combination of the surface and fluid film properties of the given system.
𝜏 = 𝜇𝑑𝑉
𝑑𝑧
2.1
Water is the natural medium for biological systems and their lubrication processes. Water is
an excellent lubricant because it has a low viscosity and stays fluid even down to molecularly
thin films5. However, water as a lubricant has poor load-bearing capacity due to its low
viscosity at high pressures. Hence it is squeezed out from the gap at high loads leading to
asperity contact and increased friction. Water-based lubrication is achieved in many
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
8
Figure 2.2. Pictorial representation of the viscous flow equation in 2.1; the stress, τ, required
to shear a Newtonian fluid between two surfaces depends on the viscosity of the fluid, µ, and
the fluid film thickness, dz.
biological systems through biological lubricant components, such as heavily glycosylated
proteins (e.g. mucins and aggrecans), and polysaccharides that maximise the load-bearing
capacity of water through steric and electrostatic repulsion between surface bound aqueous
films6. The locomotion of land-based limbless animals, such as snails, provides an interesting
example of the use of water together with sugar molecules to control friction and lubrication
at the continuous interaction between the animal and substrate7. Snails secrete a water based
mucus layer with yield stress fluid behaviour. Periodic pulses of muscular (elastic)
deformation generated by the foot lead to transitions between solid behaviour and fluid flow
of the mucus layer, thereby driving motion. The mucus contains dissolved polysaccharides
which increase the viscosity of water. A lubricant with higher viscosity generates a larger
fluid film thickness, which reduces surface asperity interaction.
Biological lubricant additives not only enhance lubrication behaviour when dissolved in
solution, but also when the molecules are adsorbed to surfaces6. For example, mucins protect
and lubricate the surfaces of epithelial tissues lining ducts and lumens within the human
body8, 9. This lubrication mechanism relies on the bottle brush structure of the molecule. The
protein backbone attaches to surfaces and the hydrophilic sugar chains immobilise large
amounts of water in the contact zone protecting the lining of ducts from abrasion6.
Sugar-based bottle-brush molecules forming hierarchical structures have also been implicated
in the complex biological lubrication mechanism exhibited in human joints6, 10. Articular
cartilage effectively demonstrated the ability of lubricating molecules to achieve low friction
and significant load-bearing capacity in aqueous media10-12. The high density of sugar chains
in adsorbed polymer brush layers on opposing surfaces generate long-ranged repulsive forces
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
9
of both steric and osmotic origin4, 6, 13. Mutual interpenetration of opposing polymer chains is
restricted to a narrow interfacial region in Figure 2.3, thus the interface is maintained as a
highly fluid layer and the system has a high load-bearing capacity13-16.
Figure 2.3. Schematic of compression between two opposing cartilage surfaces with
aggrecan layers (reproduced from Han et al.17).
The architecture of molecules that are implicated in the lubrication mechanisms of biological
systems, such as aggrecan, is studied to determine the features that are essential for the
observed low friction. Good lubricity is achieved when molecules are extensively hydrated
and the hydration layer has high fluidity during shearing to minimise energy dissipation10, 18.
Boundary lubrication between opposing surfaces with pre-adsorbed polymeric layers is
enhanced by solvent properties that promote a more extended and hydrated film, which
consequently has a higher thickness19. We can extend this knowledge to predict candidate
lubricating molecules in plant systems, based on the physical and chemical structure.
2.2 Plant Cell Wall Structure, Mechanics and Growth
Plant systems are rarely the subject of tribological studies despite the relative movement of
multi-scale structures associated with growth processes that would require sophisticated
lubricating mechanisms. Here I will connect the bodies of knowledge surrounding the cell
wall composition and the assembly into individual cells and tissue (cell aggregates), multi-
scale mechanics, and growth processes to identify tribological contacts that are an essential
part of the strategy by which plants grow and maintain mechanical strength.
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
10
2.2.1 Plant Cell Wall Architecture
The plant cell wall has crucial roles in providing structural strength, a barrier to pathogens,
plant growth, cell differentiation, signalling, and water transport20. Cell walls are highly
heterogeneous and complex structures that possess a remarkable combination of tensile
strength and extensibility important for plant growth21. The primary cell wall consists of a
cellulose-hemicellulose framework embedded in an inter-fibrillar matrix of pectic
polysaccharides in Figure 2.4, all interacting with an independent domain of structural
proteins22.
Figure 2.4. Structure of the primary cell wall (reproduced from Smith23).
Cellulose
The cellulose fibres are the main load-bearing elements of the plant cell wall structure due to
their high tensile strength21, 24. Cellulose has a hierarchical structure made up of parallel
linear β-1,4-linked glucan chain aggregates25. Cellulose Elementary Fibrils (CEFs) are the
units synthesised by cellulose synthase rosettes at the plasma membrane. CEFs associate
through their hydrophilic faces to form microfibrils, which may contain a single or multiple
CEFs, and microfibrils bundle together to form macrofibrils or cellulose ribbons26. The
Atomic Force Microscope (AFM) has been used to measure the height and consequently infer
the diameter of single cellulose microfibrils from partially hydrated primary cell wall samples
isolated from onion (Allium cepa L.) and Arabidopsis thaliana (L.) Heynh27. The microfibrils
were found to be 4-6 nm in diameter in the cell walls and ca. 3 nm in walls that have been
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
11
treated to remove pectic polysaccharides. These results indicate that microfibrils largely
contain a single cellulose crystallite surrounded by non-cellulosic polysaccharides and no, or
only a few, cellulose aggregates (macrofibrils) exist. Ding et al.26 probe the morphology of
CEFs in maize cell walls using the AFM and reveal a hexagonal shape (3.2 x 5.3 nm) in
cross-section. Based on this measurement the authors propose a 36-chain CEF model, that is,
36 β-1,4-linked glucan chains aggregate to form a crystalline hexagonal shape CEF as
pictured in Figure 2.5. The size estimate of CEFs from plant sources has been measured using
SANS28, WAXS28, 29, SAXS, and NMR29 gives a diameter of ca. 3 nm. Fernandes et al.28
propose a 24-chain rectangular model based on this diameter. The structural characterisation
of CEFs using AFM estimates a size slightly larger than that measured using a suite of
scattering techniques. This discrepancy could be due to certain limitations associated with
AFM topography measurements. The first common artefact in topography measurements is
profile broadening due to tip-sample convolution. The thickness of the AFM tip leads to an
overestimation of the sample width, as depicted in Figure 2.6a. Secondly, the sample height
can be underestimated due to elastic deformation of non-rigid samples. Whilst cellulose
microfibres have a high modulus, the rigidity of the substrate underneath and the potential
deformation along the length of the fibre may affect sample height measurements.
Additionally, the samples for the reported AFM measurements are dried, which may lead to
aggregation of microfibrils and non-cellulosic polymers and overestimation of the CEF size.
The non-cellulosic polymers coating the microfibrils were shown by Davies and Harris27 to
significantly increase the measured CEF size. Whereas, for scattering techniques, the non-
cellulosic polymers do not have a significant effect due to the different crystallinity compared
to the CEF. Considering this, I conclude the CEF model that is most supported by experiment
data is the 24-chain model.
Figure 2.5. Predicted 36-chain CEF model with hexagonal cross-section and dimensions of
5.3 x 3.2 nm (reproduced from Ding et al.26).
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
12
a b
Figure 2.6. Common artefacts with AFM topography. a) shows the effect of the tip width on
profile broadening which leads to an overestimate of the sample width. b) elastic deformation
of the sample leads to an underestimate of the sample height.
The validity of 18-, 24-, and 36-chain CEF models has been investigated through molecular
dynamics (MD) simulations, and it is argued that the 18- and 24-chain structure are more
viable models30. The spacing of the cellulose chains in the CEF structure that are predicted
from the simulations correlate with previously published WAXS31 and x-ray diffraction
data32. Furthermore, the crystalline conformation of the 18- and 24-chain models are in better
agreement with high temperature molecular dynamic simulations produced by Matthews et
al.33.
The cellulose microfibrils in the plant cell wall are deposited in a controlled orientation,
typically perpendicular to the axis of cell elongation (Figure 2.7), thereby resisting lateral
swelling and permitting longitudinal expansion25, 34. The orientation of cellulose microfibrils
varies between the inner and outer scales, from a transverse through random to longitudinal
orientation respectively35-40.
AFM imaging of onion epidermis shows that the cellulose microfibrils come into close
proximity with one another in short segments, obscured by matrix material, rather that direct
cross-linking between cellulose36. The surface of cellulose is described as amorphous with a
crystalline domain in the core28, 41. A paracrystalline state of cellulose exists with
intermediate mechanical properties between crystalline (high modulus) and amorphous (low
modulus) phases. The paracrystalline phase has a partially ordered structure that might be
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
13
relevant for explaining the surface chains on cellulose microfibrils that are the interfaces
between crystalline cellulose and semicrystalline hemicellulose phases in the cell wall41.
Plant cell wall models consistently show that non-cellulosic polysaccharides in the inter-
fibrillar matrix are associated with the surfaces of cellulose microfibrils, however there is no
clear consensus on the nature of these associations. The current models for the roles of non-
cellulosic components within the plant cell wall are discussed below.
Figure 2.7. Cellulose fibril orientation relative to axis of elongation
Hemicelluloses
Hemicelluloses are major matrix polysaccharides that interact with the cellulose network in
the plant cell wall. The most abundant hemicelluloses across plant species are xyloglucan
(XG) and arabinoxylan (AX). Early models of primary cell walls based on selective
enzymatic degradation of suspension-cultured sycamore cell walls show hydrogen bonds
interconnecting XG and cellulose42. The nature of this interaction with cellulose could be via
entanglement with the amorphous glucans on the fibril surface43, tightly bound layers around
the fibril surface 44, physical entrapment inside the microfibril during synthesis45, or covalent
via a transglycosylation reaction46. It has been suggested that XG chains of the appropriate
length may crosslink microfibrils creating load-bearing tethers that reinforce the cell wall and
prevent self-association and lateral separation of adjacent microfibrils21, 47. This original
tethered-network model is supported by transmission electron micrographs of cellulose/XG
composites as seen in Figure 2.8.
The validity of the widely accepted ‘tethered network’ model was recently revisited by
investigating biomechanical changes induced by substrate-specific endoglucanases48. It was
found that endoglucanases that hydrolyse both XG and cellulose are required to induce creep,
suggesting that there is a minor, relatively inaccessible XG compartment that may be
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
14
Figure 2.8. Transmission electron micrographs of tungsten/tantalum/carbon replicas of
cellulose (left) and cellulose/xyloglucan composite (right) (reproduced from Chanliaud et
al.49).
intertwined or otherwise complexed with cellulose as shown in Figure 2.9. It is proposed that
structurally important XG is located in limited regions of tight contact between cellulose
microfibrils and these structures determine cell wall mechanics and enzyme-induced creep.
One possible interaction scheme could involve lateral non-covalent bonding by a single XG
layer mediating adhesion between adjacent microfibrils. The XG at these junctions may be
entangled with disordered surface glucans on adjacent microfibrils. Another possible
interpretation of the interaction at the junction points is that XG penetrates deeply into
cellulose microfibrils, resulting in disordered regions which are similarly attracted to other
disordered regions on adjacent microfibrils.
Figure 2.9. The original tethered network model (Left) and the revised plant cell wall
architecture (Right); red rods = cellulose fibrils, black lines = xyloglucan molecules
(reproduced from Park and Cosgrove48).
Unlike the interaction between XG and cellulose, AX is suggested to form strong non-
covalent bonds with itself and cellulose fibres based on in vitro cellulose binding experiments
on the walls of barley aleurone cells (containing 85% arabinoxylan)50. However, it has also
been proposed that for monocot cell walls, arabinoxylan could be linked to cellulose through
a paracrystalline interaction which is comparable to XG interacting with disordered surface
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
15
glucans or penetrating cellulose fibrils creating disordered regions51. The network of AX
chains and cellulose fibres adds strength to the cell walls50.
Pectic Polysaccharides
Pectic polysaccharides typically consist of a complex family of acidic polysaccharides built
from several structural domains. Homogalacturonan (HG) is the most abundant pectic
polysaccharide, followed by rhamnogalacturonan I (RG I), and smaller amounts of
Rhamnogalacturonan II, Xylogalacturonan, Arabinan, and Arabinogalactan I. All essentially
consist of galacturonan backbones with or without various side chain additions. Cross-linking
between pectic polysaccharides and other cell wall components is implicated in three-
dimensional network formation in the primary cell wall and middle lamella layer separating
adjacent cells, and is essential for intercellular adhesion. Adhesion between adjacent cells in
plant tissue facilitates plant growth and development, and is important for maintaining
mechanical strength.
There are three proposed cross-links connecting pectic molecules: (i) The negatively charged
carboxyl groups on blocks of galacturonic acid residues of two unesterified HG chains can
form cross-links with Ca2+ ions52, 53. (ii) Dimers of RG II are formed through borate-diol ester
cross-links54. (iii) Pectic polysaccharides may also be linked to relatively hydrophobic cell
wall components via galacturonoyl ester bonds55. The presence of different kinds of cross-
links is supported by a study of cell-cell adhesion in sugar-beet root parenchyma which was
found to depend on both ester and Ca2+ cross-linked polymers56.
In addition to self-association, early models of the primary cell wall suggest that a single
pectic polysaccharide is indirectly associated with multiple microfibrils through covalent
interaction with xyloglucan such that the interfibrillar matrix of directly linked pectin and
xyloglucan mechanically behave as a single entity42. Several pectic polysaccharides have also
been shown to exhibit direct cellulose binding capacities, most likely through hydrogen
bonding with their neutral side chains57. It is proposed that the arabinan and galactan chains
coat the microfibrils to create a continuous cellulose-pectin network. This leads to
competitive binding between XG and pectin with cellulose57-59, and both interactions may
have a load-bearing role60.
Pectic polymers are synthesized and deposited in cell walls in a highly esterified form61.
Subsequently, wall-based pectin methyl esterases (PMEs) hydrolyse (or remove) methyl-ester
groups from the HG backbone, which changes the molecular structure and the way it interacts
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
16
in the cell wall. Wall-based PMEs spatially regulate matrix properties by producing HG with
blockwise and non-blockwise distributions of methyl esters at discrete micro-domains within
the cell wall62. The blockwise de-esterified HG epitope contains long stretches of unesterified
HG, and is present at the cell wall lining intercellular spaces and the region of the wall closest
to the plasma membrane. A large increase in electrostatic interaction between calcium ions
and free carboxylic acids occurs with the level of de-esterification63, and nine consecutive
galacturonic residues are necessary for association with calcium ions53. This suggests that the
pectin epitope with long stretches of unesterified HG will be calcium cross-linked. The
activity of PMEs is balanced by pectin methyl-esterase inhibitors (PMEIs) to affect the
mechanical properties of the middle lamella in a controlled way64.
During plant growth, separation of adjacent cells tends to occur at cell corner junctions,
pictured in Figure 2.10, generating high separation stresses that are dissipated to the corners
of the intercellular space that is formed65. Thus the distribution of the calcium cross-linked
epitope at the intercellular space is particularly important for resisting the separation stresses
that are concentrated at this location.
Figure 2.10. Distribution of pectin epitopes in the outer epidermal wall and inner tissue walls
of maize coleoptiles (reproduced from Schindler et al.66).
Both the non-blockwise de-esterified and partially methyl-esterified HG epitopes were
identified in the cell wall lining the intercellular spaces62. The non-blockwise de-esterified
HG epitope occurs in large amounts at the corners of the intercellular space, whereas partially
methyl-esterified domains have an increased abundance in the region of the wall closest to
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
17
the plasma membrane62. This is in agreement with other models of the distribution of pectin
epitopes, one of which is presented in Figure 2.10. In these models, unesterified pectins are
restricted to the corner junctions and esterified pectins are present in the middle lamella far
away from the corner junction and in the primary walls61, 66. In the growing regions of
hypocotyl mung bean seedlings, a high content of methylesterified pectins are observed,
whereas non-growing regions are characterised by unesterified pectins63. The composition of
pectin in different areas of the plant tissue and at different areas of growth and non-growth is
complex. A possible interpretation of the controlled distribution of pectin is that
methylesterified pectins that do not form calcium cross-links facilitate cell separation and
movement within the walls and are thus located in the growing regions. Whereas the cross-
linked pectin concentrated at the corner junctions and non-growing regions contributes to
adhesion. Cell separation during development occurs through polygalacutonase-mediated
cleavage of the HG backbone of DE pectin64.
2.2.2 Cell Wall Mechanics
If we reflect on our everyday experience with plant material and compare the texture of say a
lettuce and a carrot, we get some appreciation for the vastly different mechanical properties
between different plant tissues from different species. In fact, the range of mechanical
properties has been captured by measuring the Young’s modulus and compressive strength of
plant material. If a given stress is applied to a material and the resultant deformation is
recorded as a strain, the Young’s modulus is taken as the slope of the linear region of the
stress-strain curve. The compressive strength is a measure of the stress required to ‘break’ a
material under compression. Gibson67 reports that the Young’s modulus spans 5 orders of
magnitude and the compressive strength spans 3 orders of magnitude for plant materials from
parenchyma to the densest palm. This range of mechanical properties is attributed to the
composition and microstructure of the cell walls, tissue level structure, and turgor pressure
for plant material with intact membranes. Thus it is very informative to measure the
composition and mechanics of the plant cell wall, the mechanics of an isolated plant cell, and
the mechanics of clusters of cells (plant tissue) for a single plant species. This type of holistic
investigation gives significant insights into how individual components and hierarchical
structures contribute to material properties. These insights can be extended to understand
biological phenomena such as plant growth, and could assist in the design of advanced
functional materials.
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
18
Zamil et al.68 cut wall fragments from the middle region of a single cell of onion epidermal
peel and performs uniaxial tensile tests in vacuo on the dried sample to characterise the
behaviour in the major and minor growth directions. The modulus was of the order of 4 GPa
for both directions. In a subsequent publication, Zamil et al.69 present stress-strain curves for
never-dried cell wall fragments without maintaining continuous hydration. The average
modulus value was 1.5 GPa. Interestingly, when the samples are tested under continuous
hydration, the shape of the stress-strain curves changes showing a first linear region, plateau
zone, and second linear zone. The author attributes this behaviour to the biphasic nature of
the material. A similar non-linear shape in the stress-strain curve is observed for an onion
epidermis tissue sample subjected to uniaxial tension, although the plateau region in between
the two linear regions is less pronounced than for hydrated samples70, 71. The Young’s
modulus from the slope of the first linear region is of the order of 5 MPa for transverse
orientation and 20 MPa for longitudinal orientation71. Comparing the results for cell wall
fragments and tissue samples shows that mechanical properties can vary by two orders of
magnitude depending on the length scale, and further highlights the complexity of
characterising the hierarchical structure of plant material.
The final piece of the puzzle to achieve a complete picture or mechanical model of plant
material is the mechanics of a whole cell. This differs from cell wall fragments because it
considers the cell geometry, which is particularly important for understanding the relationship
between turgor pressure and wall stress (this is critical for cell wall extension and will be
discussed in a separate section below). Mechanical measurement of isolated single cells have
been achieved using compression testing by micromanipulation with a cylindrical probe72 or
pyramidal tip73, 74, as shown in Figure 2.11. Digiuni et al.73 and Bonilla et al.74 show that
individual plant cells behave in a viscoelastic manner that is driven by their internal
hydrostatic turgor pressure. Wang et al.72 also show a time-dependent mechanical response
where water plays a significant role that is dependent on the deformation rate.
A mesh-free particle method is used to simulate the mechanics of an individual plant cell
which is modelled as an interior liquid phase surrounded by a viscoelastic solid material75.
Van Liedekerke et al.75 investigate the relaxation mechanics of a cell by running a simulation
starting from the cell in a stretched state, removing the stretching force, and letting it relax as
the strain is monitored. The relaxation behaviour of an isolated cell shows a viscoelastic
response where the relaxation time depends on the viscosity of the interior fluid. Thus it is not
only the cell wall mechanics measured from tensile testing of wall mechanics that determine
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
19
the mechanics of plant cells. Van Liedekerke et al.75 also perform simulations on cell
aggregates,
a b
Figure 2.11. Mechanical measurement of isolated cell walls using a) compression testing by
micromanipulation with a cylindrical probe (reproduced from Wang et al.72) and b)
indentation experiments using a pyramidal AFM tip (reproduced from Bonilla et al.74).
looking at the influence of cell geometry and intercellular voids. The most important
observation from this simulation is that stresses are not uniformly distributed in the
aggregates. All previous attempts to directly measure the mechanics of plant materials
reinforce their time- and length-scale dependent and heterogeneous nature.
What I see as a gap in the literature is a multi-scale approach using model systems to build up
relationships between important material parameters. With the current method of measuring
the mechanics of cellular and sub-cellular structures from the same plant material, there is a
limitation in relating the results to a generalised multi-scale model due to confounding factors
such as heterogeneity in composition and cell geometry, the interfacial middle lamella layer
between neighbouring cells, and the distribution of wall stress from turgor pressure and other
external forces. The key advantages of a model system are that they are largely homogenous
and enable the influence of individual components to be investigated through controlling the
composition. For example, I can alter one material property in a systematic way and
investigate the effect on the mechanics at one specific length scale. Furthermore,
measurements from these well-characterised model systems can be interpreted using
complementary finite element modelling (FEM) to produce generalised relationships that can
predict the behaviour of any system that is described by the same constitutive equations (e.g.
anisotropic, poroelastic etc.). This approach enables the development of a multi-scale model
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
20
of plant material that will give insight into the specific contributions of cell wall components
on cell wall and tissue level mechanics.
2.2.3 Plant Growth and Cell Wall Extension
The slow, time-dependent, irreversible extension of the cell wall during plant growth occurs
due to cellulose microfibrils and associated matrix polysaccharides in relative sliding
motion20. The irreversible yielding of growing cells is suggested to be due to biochemical
loosening of the wall and not the viscoelastic property of the wall itself76. The stress
relaxation of the wall, that is, the structural rearrangement of the wall leading to a decay in
the wall stress, is the driving force for water uptake and expansion of the cell76 77. The time
when stress relaxation begins and the rate of stress relaxation are found to correlate well with
the amount of hemicellulose in the cell wall based on a stress-relaxation assay on lettuce
hypocotyl treated with gibberellic acid78. Based on a physical model representing the stress-
relaxation phenomenon in the cell wall simulated with Maxwell viscoelastic elements,
increasing the molecular weight of hemicelluloses shifts the stress relaxation time spectrum
of the cell wall to higher relaxation times77. A mathematical model of hemicellulose cross-
link dynamics in an expanding cell wall also shows that increasing the crosslink rest length
increases stress relaxation79. The relaxation of cell walls may also be affected by pectic
polysaccharides where pectate cross-links affect the wall porosity and possibly the
accessibility of primary wall relaxation proteins to their substrate80. The change in porosity is
also likely to affect the resistance to water movement into the cell after stress relaxation.
Additionally, the high mobility of pectin influenced by hydration81 could directly affect the
relaxation behaviour of the biphasic cell wall.
Cell expansion is anisotropic for the majority of plant systems, which means that it
preferentially expands along one axis. The origin of the anisotropy is due to a mismatch in
either wall mechanics or applied stress in the lateral and transverse directions82, 83. Cell wall
extensibility is higher (lower apparent Young’s modulus) transverse compared with parallel
to the net cellulose alignment, thus the direction of growth anisotropy and maximal expansion
rate are qualitatively specified by the mean fibre orientation37, 38, 68, 69, 71, 79, 82, 84-89. However,
the degree of orientation among cellulose fibres does not quantitatively correlate with the
degree of growth anisotropy37, 38, 82. Thus orientation of cellulose fibres is not solely
responsible for growth anisotropy.
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
21
Mathematical modelling of hemicellulose cross-link dynamics, incorporating enzyme
mediated cross-link dynamics, demonstrates that enzymes targeting cross-links are effective
in softening the wall in its pre-yield state and the pectin matrix determines the post-yield
extensibility79. This is supported by the use of heat inactivation to show protein dependence
of extension transverse to the net cellulose orientation, consistent with an important role of
xyloglucan endotransglucosylase activity on cross-links for extension90. An alternative
explanation of growth anisotropy is the distribution of PME activity such that the wall along
the growth direction is more elastic due to its richness in low degree of methyl esterification
pectin91. Selective weakening of the longitudinal walls as the origin of anisotropic growth
only works in the context of a tissue. For an isolated cell the longitudinal wall would bulge in
the absence of transversely orientated fibres. The critical role of pectin is also shown by
Braybrook and Peaucelle92 where de-methyl esterification of pectin alone is sufficient to
induce local tissue growth in the meristem of Arabidopsis thaliana. In summary, the initial
controlled deposition of cellulose combined with biochemical loosening of one or both of XG
and pectin initiates anisotropic extension. This is followed by passive reorientation of
cellulose microfibrils parallel to the axis of elongation which slows growth.
Expansins
The biochemical loosening of the cell wall is consistently linked to the wall protein expansin;
however, the mechanism of action is unclear. Expansins are proteins secreted in plant cell
walls during growth to unlock the network of wall polysaccharides and permit turgor-driven
cell enlargement93. An ‘acid growth’ phenomenon is observed where expansins induce stress
relaxation and extension of isolated cell walls in a pH-dependent manner such that plant cell
walls typically extend faster at lower pH25, 93, 94. When expansins are applied to living cells
they stimulate cell enlargement which indicates that the proteins function under normal
physiological conditions in the walls of living cells, and that expansins can be naturally
present in plant tissue in sub-saturating amounts95. The action of expansins is immediate and
only small relative amounts of the protein are required to induce wall extension, suggesting
that the mechanism relies on a few ‘sticky’ spots or entanglements at a given time where
expansins act to promote polysaccharide slippage93. NMR studies support this claim, showing
that expansins do not increase the general mobility of wall polysaccharides96. The action of
expansins leads to longer and thinner walls after extension, but the cell wall is not weakened
or structurally changed in the long term.
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
22
The cell wall strength is determined by the sum of the strength provided by each load-bearing
bond. Wall relaxation occurs through these bonds being broken and new bonds formed so
that the load is transferred and the strength of the wall remains the same, thus the action of
expansins must be reversible97. Consistently, expansins are reported to reversibly disrupt non-
covalent binding between wall polymers that facilitates polymer slippage under stress97-102.
The mechanistic detail for expansin action is yet incomplete; some of the key hypotheses
from literature are listed below. It is noted that over the time of these references the cell wall
model has changed, in particular the role of XG as tethers, and the structure of the expansin
protein has become better characterised.
‘…we suggest that expansins act as a sort of biochemical grease…expansins might facilitate
polymer slippage under stress by disrupting noncovalent binding between wall polymers…’99
‘By ‘unzipping’ the glucans that tether cellulose microfibrils together, expansins may induce
wall stress relaxation…’100
‘…expansins lubricate glucan-glucan interactions allowing (in some instances) entangled
polymers (or fibrils) to slide relative to one another…’102.
‘…expansins have separate hydrophobic and hydrophilic regions, and thus, they may act in
part as surfactants…’103
‘…BsEXLX1 [bacterial expansin] could move on the surface of cellulose and disrupt
hydrogen bonds by twisting glucan chains.’104
McQueenmason and Cosgrove98 compare the activity of expansins extracted from plant
material to that of urea, which weakens all the hydrogen bonds between fibres in paper
causing stress relaxation. Results show that expansin activity causes slower stress relaxation
compared to the native cell wall substrate treated with urea, and may involve progressive
disruption of bonds and translocation of the protein as the fibres slide apart. However the
small concentrations of expansins relative to the number of fibre contact points required for
extension and the apparently immediate response to the presence of expansin argue against
this interpretation.
Cosgrove93 previously explained the relatively minor amounts of expansin needed for wall
extension as relating to a similarly small amount of critical junction zones or ‘sticky’ spots
existing in the wall. At these ‘sticky’ spots, glucans that are attached to microfibril surfaces
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
23
interact with each other or adjacent microfibrils forming tethers that need to be disentangled
to promote polysaccharide slippage. Expansin movement may be confined to lateral diffusion
along the surface of the cellulose microfibrils93. This would enable the bound protein to scan
the microfibril surface, locally loosening its attachment to the matrix, and allowing chain
movement and stress relaxation. Yet a number of theoretical difficulties with this ‘sticky’
network model are raised by Thompson105. These discrepancies include: the cell wall is
physically measured to be stronger than what would be predicted by the model, the work
done in wall expansion is greater than the energy stored in hydrogen bonding between tethers
and microfibrils, hemicellulose tethers are unlikely to limit wall extension because bacterial
cellulose networks with hemicellulose are weaker than pure cellulose networks, calcium
chelators promote creep, and contraction and extension of cell wall material is inhibited under
conditions of reduced relative permittivity.
Attempts to determine the mechanistic details of the action of expansins include the use of
techniques assessing hydrolytic and enzymatic activity, binding specificity, mechanical
properties and substrate microstructure. Expansin proteins extracted from plant cell walls do
not cause a time-dependent weakening of the cell wall and the effect on mechanical
properties is reversible, indicating that action is not through a hydrolase-type mechanism98,
106. Bacterial expansins possess no hydrolytic activity against cellulosic and hemicellulosic
substrates99 107-110 but have a synergistic effect with cellulase107-110. Wall hydrolases may
work synergistically with expansins by reducing the matrix viscosity thereby sensitising the
wall to expansin’s action111.
Binding specificity has been studied using endogenous alpha-expansins in Arabidopsis
thaliana hypocotyls wild type and XG deficient mutants48. Expansins cause increased
extension under constant stress, as well as increased stress relaxation for both the native and
XG deficient samples, with the highest activity observed for the native hypocotyl. These
results suggest that the target of alpha-expansin is most likely the cellulose-XG matrix, but
hydrogen bonded contacts between other (1-4) beta-glucan can also serve as substrates.
Furthermore, McQueenmason and Cosgrove99 show that expansins extracted from wall
fragments bind weakly to crystalline cellulose, and binding increases for cellulose coated
with hemicellulose. They conclude that expansins bind to insoluble matrix polymers that are
tightly associated with cellulose.
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
24
Wang et al.112 show that bacterial expansin binding is highly specific to cellulose domains
enriched with XG which could be a consequence of XG changing the crystallinity at the
cellulose fibre surface. McQueenmason and Cosgrove 99 also report the increased binding of
plant expansins to amorphous regions formed after native cellulose was treated with 4M
NaOH to form the less crystalline cellulose II, which contains more extensive amorphous
regions. However there is evidence to suggest that binding capacities of BsEXLX1 are not
specifically related to the crystallinity of the cellulosic substrate where binding order did not
match crystallinity order for a series of substrates113. Contradicting this, Bunterngsook et
al.114 report that binding capacity of expansins from a different bacterial strain is affected by
the degree of crystallinity of the cellulosic substrate. Binding efficiency to amorphous
cellulose (PASC) was higher than Avicel. However the binding efficiency of arabinoxylan,
which is the most amorphous substrate, falls in the middle and suggests that binding
preference is based on both the types of polysaccharides and the crystallinity state of the
substrates. This could explain why Kim et al.113 show microcrystalline cellulose from a
bacterial source has higher binding efficiency compared to the less crystalline Avicel because
the different sources of cellulose may have different compositions of the crystalline
allomorphs.
The structure of the BsEXLX1 protein suggests that it binds to a single glucan chain rather
than a highly crystalline surface110. Therefore, the expansin may directly reconfigure
polysaccharides at the cellulose-matrix interface, leading to cell wall creep. By binding a
glucan that is part of the load-bearing network, the resulting distortion in its shape could drive
slippage at the junction if the wall is in tension110. Seki et al.109 find no evidence for expansin
activity transforming the crystal structure of cellulose macroscopically from X-ray diffraction
results; however this does not discount the possibility of very minor (i.e. single glucan chain)
structural changes. This shape distortion mechanism may support some of the ideas presented
in the conformational defect model by Lipchinsky115. Lipchinsky115 proposes that a region of
a cellulose chain on the surface of a microfibril is released due to the action of expansin. This
causes a conformation defect that can move toward microfibril-matrix interfaces driven by
stress gradients on the microfibril surface. The defect then causes the interface to deflect and
leads to the dissociation of matrix polysaccharides from celluloses microfibrils. This mode of
action was proposed because it is better able to reconcile the large effect of a small amount of
expansin on initiating wall extension, which involves polysaccharide slippage at multiple
entanglements. The continued motion of the conformational defect means that expansins are
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
25
indirectly participating in dissociating multiple entanglements in the microfibril-matrix
complex. The Lipchinsky model is further supported by MD simulations of the binding of
BsEXLX1 to a single glucan chain, which shows that the chain can assume a stable twisted
conformation.
The activity of expansins is typically assessed using the following mechanical assays:
• Creep test: monitoring the extension under a constant applied stress.
• Stress relaxation test: Applying a constant extension and monitoring how the wall
stress decays over time.
• Tensile test: Apply an increasing stress and observed at the relationship between the
stress and extension, as well as the stress at which there is a critical failure in the
sample (referred to as the ‘tensile’ strength.
The substrates used in the assays above include plant tissue, cellulose paper (Whatmann filter
paper), and bacterial cellulose hydrogels. There are challenges to using plant tissue as a
means of elucidating the mechanism of action due to the inability to separate the contributing
factors. Park and Cosgrove48 compare a native and XG-deficient plant tissue to investigate
the contribution of XG, but comment on the potential compensation of other cell wall
components such as AX. Kim et al.107 report that BsEXLX1 reduces the tensile strength of
filter paper, and SEM images show some microstructural changes where fibrils were less
interconnected or overlapped compared to the absence of BsEXLX1. An important question
is whether filter paper is a good model of the cellulose interactions in plant cell walls because
the cellulose has been mechanically and chemically treated and dried to make the paper
product. Finally, Whitney et al.102 show that extracted alpha-expansin increased creep on a
hydrated bacterial cellulose composite material containing XG, compared to no effect on
cellulose-only materials. However, for the same samples, uniaxial extension tests shown that
expansins have an effect on both cellulose and cellulose-XG samples. For me, this is a strong
indication that the mechanism of expansin action is strongly dependent on the distribution of
stress in the system. Furthermore, model systems such as bacterial cellulose hydrogels are
appropriate for controlling the direction and extent of deformation.
What is lacking from this work is an understanding of how to precisely interpret the changes
in macroscale mechanical behaviour (e.g. creep, stress relaxation, tensile strength) in terms of
the microstructural features of the system. For example, does an increase in extensibility of a
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
26
cellulose network relate to a decrease in the number of fibre-fibre contact points in the
system, or a decrease in the yield stress or adhesion at the contact points? I will build on the
use of bacterial cellulose as a model system to answer questions such as these by looking at
fibre-fibre and network-network level interactions in the presence of expansins, with
accompanying FEM simulations to uncover the specific physical changes that explain
observed mechanical changes.
2.2.4 Tribological Contacts in Plant Cell Walls
Based on this review of plant mechanics and growth, I identify two critical tribological
contacts. The first is the interface between adjacent cell walls in a plant tissue. In the review
of pectic polysaccharides above, the pectin rich middle lamella layer is postulated to allow
cell separation during plant growth and development. However, to my knowledge, the
presence of shear forces and the role of friction at this interface have not been explicitly
referenced. From a modelling perspective, the boundary conditions at the interface between
two cells include an assumption of no-slip. The simplest model implementation of this is a
shared wall, i.e. wall-wall interactions are infinitely stiff75, 116. Results from such models
describe plant cell tissue where the pectins in the lamellae are anchored to the rest of well to
some degree. However, there is some experimental evidence that the no-slip boundary
condition does not always hold. Kwiatkowska and Dumais117 and Uyttewaal et al.118 find that
neighbouring cells can exhibit different growth rates. This growth rate heterogeneity is
observed in a Figure 2.12, which shows images from a time-course study of elongating root
cells119. A comparison of the images taken 4 hours apart reveals an apparent relative motion
between the columns of elongating cells. The relative extension rates of adjacent cell walls
will generate a sliding interface. For this case, the interactions at the middle lamella layer,
and the tensile and adhesive forces that arise between the cell walls during growth have not
been modelled.
The second tribological contact that is critical for plant growth is the interface between
individual cellulose fibres, often mediated by other cell wall polymers. The process of wall
extension during plant growth is detailed in the previous section of this literature review
chapter. It is understood that the loosening of fibre contacts drives wall stress relaxation, and
consequently cell expansion. Thus, during growth there is a large potential for sliding
contacts between cellulose fibres and other matrix polymers.
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
27
For both inter-cellular and inter-fibrillar sliding contacts, the shear forces will depend on the
(cell wall or fibre) mechanics, the adhesive interaction, and the tribo-rheological behaviour of
polymers located at the interface. Bringing all of these physical processes together is critical
to the development of a multiscale model that is capable of describing the mechanics during
plant growth.
Figure 2.12. Time lapsed images of growing root (reproduced from Anderson et al.119).
Tracking the ‘end-to-end’ interfaces between cell walls, (marked with a dotted and solid line)
shows that there is relative motion between adjacent cells.
2.2.5 Bacterial Cellulose as a Model System for Plant Cell Walls
A significant number of studies looking at the mechanics of the plant cell wall have utilised
living plant tissue and isolated cells or cell wall fragments. Plant cell walls are complex
systems to work with for the number of reasons expressed in this section of the literature
review. I have so far indicated where a model system, specifically bacterial cellulose
hydrogels, can provide a means to systematically interrogate the role of microstructural
features on the overall mechanical properties. These findings will be used to gain insight into
the influence of individual cell wall components, and their assembly, on the mechanics of
plant cell walls.
Gluconacetobacter xylinus is a species of bacteria that produces cellulose in an analogous
fashion to plants, offering an alternative and more systematic approach to look at the
mechanics of the primary cell wall using bacterial cellulose composites as a model system49,
120, 121. Cellulose molecules synthesised in the bacterial cell are extruded out of pores in the
outer and cytoplasmic membrane (elementary fibril). Subsequently, elementary fibrils
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
28
aggregate to form microfibrils which form ribbon-shaped macrofibrils120-122. This process is
illustrated in Figure 2.13.
The average cross-section of hydrated bacterial cellulose ribbons is ca. 36nm123. The
hierarchical structure of bacterial cellulose has been investigated using small angle scattering
techniques with XRD and SEM123. The scattering results for BC ribbons are consistent with a
core-shell model. In this model the core is composed of cellulose crystallites and the outer
region (shell) consists of solvent accessible paracrystalline cellulose, as shown in Figure 2.14.
XRD results reveal that the number of cellulose chains contained in the elementary fibrils of
BC is significantly greater than that previously reported for plant cellulose where the
dimension for BC are 5nm x 8nm (126-144 cellulose chains) versus 3nm x 3nm (24 cellulose
chains) for plants123. Furthermore, bacterial cellulose aggregates to form a flat ribbon shaped
microfiber with a cross section of the order of 35nm123.
Figure 2.13. Schematic of cellulose biosynthesis by Gluconacetobacter xylinus (reproduced
from Mikkelsen and Gidley121).
Results from morphological characterisation of composite bacterial cellulose hydrogels with
arabinoxylan and xyloglucan show that both hemicelluloses interact with the surface of
cellulose ribbons123. XG incorporation reduces the cellulose crystallinity, CEF size but not
overall microfibril dimensions, and XG cross bridges between CEFs are also seen123, 124. This
is not observed for arabinoxylan which interacts with the surface of the ribbons through non-
specific adsorption123.
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
29
The method for producing pure bacterial cellulose networks or pellicles, an example of which
is shown in Figure 2.15, involves the fermentation of Gluconacetobacter xylinus in Hestrin
and Schramm liquid medium121. In the bacterial cellulose model system the composition of
Figure 2.14. Model of bacterial cellulose fibre based on SANS and SAXS data (reproduced
from Martinez-Sanz et al.123).
the extracellular environment into which the secreted cellulose is deposited can be controlled
to investigate the role of cell wall polymers on the molecular, microscopic and macroscopic
properties of the material121. Cellulose-based composites can be produced by reviving,
culturing and fermenting Gluconacetobacter xylinus in the presence of hemicelluloses and/or
pectins. For all compositions the population of the aerobic bacteria initially increases, whilst
producing a limited amount of cellulose within the entire liquid phase, until the dissolved
oxygen supply is exhausted120. Following this, bacteria populating the liquid-air interface
maintain their activity and a cellulose gel grows on top of the liquid phase. The hydrogels are
then washed with copious amounts of water to remove bacteria and polymers that are held
non-specifically, and stored in 0.02 wt% sodium azide solution which acts as a biocide121.
The hydrogels are used as a general model for studying the assembly, structure, and
properties of the plant cell wall. The bacterial cellulose model has a number of advantages
over isolated plant cell wall material including elimination of microheterogeneity and harsh
extraction conditions, as well as the ability to produce multi-centimeter samples for
mechanical measurements. The use of bacterial cellulose as a model for plant cell walls has
been validated through microscopic observation of similar architectures, similarity in
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
30
Figure 2.15.1 Bacterial Cellulose Hydrogel
expansin activity between hemicellulose/cellulose composites and plant tissue, and reduced
cellulose crystallinity in the presence of hemicelluloses, suggesting that they can access
microfibrils prior to aggregation into the final ribbon assembly121. The structural differences
between plant and bacterial cellulose fibres and networks that should be considered are
summarised in Table 2.1 below.
Table 2.1. A summary of key structural differences in cellulose fibres and cellulose networks
from bacterial and plant sources.
Bacterial Plant
CEF 5x8 nm (~135 cellulose
chains)
3x3 nm (24 cellulose chains)
Cross-section size of macro-
fibre (shape)
35 nm (flat ribbon) 3 nm (cylindrical)
Orientation of cellulose
fibres in a network
Random Highly oriented (parallel).
2.3 Mechanical and Friction Properties of Hydrogels
One of the primary motivations of this thesis is to elucidate potential interactions occurring
between plant cell walls that undergo wall stress relaxation during plant growth and other
deformations experienced at the cellular level. I consider that studying the physics of
hydrogel tribology is relevant because plant cell walls are essentially hydrogel composites of
cellulose fibrils within a matrix of biopolymers (e.g. hemicelluloses, pectin) and water.
During dynamic growth processes, the load-bearing cross-links in the wall structure are
biochemically loosened, which leads to wall stress relaxation that drives cell expansion76, 77.
As cells expand within the tissue structure, I consider there to be a sliding contact between
adjacent extending walls. Thus plant cell walls require modes of lubrication under
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
31
compression (static) and sliding (dynamic) conditions, which have not previously been
explored. Since it is challenging to investigate the mechanical response using plant cell walls
directly, model hydrogel systems will be used to probe the role of microstructure on their
tribological behaviour. In this section of the literature review I will appraise the current
approaches to characterising the mechanical and friction properties of hydrogels.
2.3.1 Hydrogel Material Characterisation
The term hydrogel is used to describe and interconnected polymer network that is solvated
with water. The unique aspect of hydrogels is that they have the potential to exhibit so-called
poroelasticity, whereby under deformation the mechanical response is not only a function of
the elastic polymer network, but also the movement of solvent through that network. The
effect of water on the mechanics and friction properties of hydrogels depends on the
relaxation rate and compression rate. This is best explained by looking at shape of a
compression relaxation curve, an example of which is drawn in Figure 2.16. During
compression, the normal stress increases with strain as the hydrogel is deformed. This
increasing stress is due to a combination of interstitial fluid pressurisation in deformation of
the elastic network. The balance between these two things depends on the ability of the
solvent to escape the network. Once the hydrogel is kept at a constant strain, normal stress
relaxation is observed. The built up interstitial fluid pressure drives fluid flow out of the
hydrogel, and leads to structural rearrangement of the hydrogel network as the normal stress
decays.
Hydrogels with low permeability (effectively synonymous with porosity) are defined as
having a relaxation rate << compression rate. When hydrogels with relative low permeability
are compressed the solvent does not have time to flow such that the effect of poroelasticity is
negligible and the system behaves as an incompressible solid. For these systems the
compression curve in Figure 2.16 is linear and there is negligible normal stress relaxation.
The hydrogel mechanics can be adequately characterised by applying linear contact
mechanics models to the linear part of the compression profile.
Hydrogels with high permeability are defined as having a relaxation rate >> compression
rate. Modelling the non-linear mechanical response of the hydrogel, as in Figure 2.16, is
complex and involves the interplay between poroelastic behaviour of the fluid and the elastic
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
32
response of the matrix. In essence, the mechanism of stress build-up and relaxation in
unconfined compression is summarised in three steps:
(i) lateral expansion of matrix and interstitial fluid pressurisation,
(ii) contraction of matrix and fluid flow out of gel due to pressure gradient,
(iii) normal stress relaxation due to fluid redistribution125.
Figure 2.16. Typical stress-time curve for compression-relaxation of a hydrogel material.
During compression the normal stress increases as the sample is deformed (increasing strain).
Once the compression is stopped and the sample is held at a constant strain the normal stress
relaxes due to solvent flow out of the hydrogel and structural rearrangement of the network.
The compression-relaxation behaviour of hydrogels with high permeability has been
modelled using transversely isotropic linear biphasic theory126, 127. This model has been
successfully applied to experimental compression-relaxation profiles to determine material
parameters such as axial and radial modulus and permeability of articular cartilage126, 128, 129
as well as bacterial cellulose hydrogels127. Stress relaxation rate sets an important distinction
between mechanical properties of hydrogels with high and low permeability. Attempts to
relate the friction response and mechanical properties of hydrogels will require an appropriate
definition of ‘modulus’.
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
33
2.3.2 Relating Hydrogel Mechanics to Friction Response
In Table 2.2 I present a digest of what I perceive to be the most important and relevant
experimental studies on hydrogel friction. For each study the table catalogues the technique,
hydrogel, experimental and material parameters that affect friction, and the relationship
between the measured substrate modulus and the coefficient of friction. From Table 2.2 it is
clear that there is no consistent quantitative scaling relationship between the mechanics and
friction of the hydrogel systems. It is noted that different studies characterise the mechanics
of the hydrogel using different definitions of modulus. For example, Zhang et al.130 measure
the elastic modulus from indentation experiments and report an inverse relationship with the
coefficient of friction, whereas Chang et al.131 measure the dynamic modulus from oscillatory
shear measurements and report a positive correlation with the coefficient of friction. Another
point to make after analysing the studies in Table 2.2 is that often the mechanics of the
hydrogel are not varied in a controlled way. Comparing the same two studies from above,
Zhang et al.130 control hydrogel modulus through freeze-thawing cycles and composition, and
Chang et al.131 induce changes in the gel state by changing solvent composition. Thus any
relationships drawn with the modulus of the hydrogels are confounded by the variation in one
or more other factors.
To investigate the relationship between hydrogel mechanics and friction I will use a bacterial
cellulose system, because it provides superior control of the modulus compared to the studies
in Table 2.2. Lopez-Sanchez et al.127 demonstrate that the mechanical properties of bacterial
cellulose hydrogels can by varied by compressing the hydrogel. I am mindful that controlling
the hydrogel mechanics in this way also effects the network structure, and in particular the
solids concentration of the hydrogel. However, Lopez-Sanchez et al.127 and Bonilla et al.132
provide extensive mechanical modelling to accompany the experimental compression-
relaxation results, which makes it possible to interpret the contribution of microstructural
changes to the mechanical response of the hydrogels.
Instruments commonly used for friction measurements such as tribometers (sphere-plane),
surface force apparatus (SFA) (cross-cylinder), and colloidal probe AFM (sphere-plane) have
clearly defined areas of contact, which is pivotal for interpreting the surface friction.
However, the way bacterial cellulose is produced (detailed in Chapter 3) means that the
hydrogels cannot be fabricated in cylindrical or spherical shapes. For disk-shaped bacterial
34
Table 2.2. Summary of key findings from relevant studies on hydrogel friction.
Reference Technique Hydrogel Parameters varied Mechanics–Friction Relationship
Zang et al.130 Friction: Hydrogel against cartilage surface in
microtribometer.
Mechanics: compressive elastic modulus measured
with flat-head cylinder indenter.
Poly(vinyl
alcohol)/Hydroxylapatite
composite.
Freeze-thawing cycles and
HA content which change
hydrogel modulus.
An inverse relationship between
coefficient of friction and modulus.
Baykal et
al.128
Friction: pin-on disk tester with hydrogel pin against
ceramic disk.
Mechanics: confined compression (i.e. impermeable
side walls and permeable top wall) and unconfined
compression (i.e. no side walls and impermeable top
and bottom walls).
Proprietary hydrogel. Velocity, load, lubricant
(water and bovine serum),
hydrogel modulus and
permeability.
Positive correlation between
coefficient of friction and hydrogel
modulus.
Kozbial et
al.133
Friction: Nano-tribometer with stainless steel ball
against hydrogel surface.
Mechanics: Compression testing in a Brookfield
Texture Analyser to determine Young’s modulus.
-carrageenan hydrogel. External load, cross-
linking density, velocity,
and environment (water or
air).
Increasing cross-linking density
increases friction force and is
proportional to Young’s modulus.
Kurokawa et
al.134
Friction: Parallel plate rheometer with gel disks
attached to both plates.
Mechanics: Elastic modulus from applying Hertz
model to force maps using AFM with a triangular
tip.
Cross-linked poly(N,N’-
dimethyl acrylamide)
Elastic modulus. Negligible effect of modulus on the
coefficient of friction due to the
friction response being dominated by
the presence of loose polymeric
chains at the surface of the hydrogel.
Kagata et
al.135
Friction: Parallel plate rheometer with gel disks
attached to both plates.
Mechanics: Dynamic modulus from oscillatory shear
test in the rheometer.
Physically cross-linked
poly(vinyl alcohol),
chemically cross-linked
poly(2-acrylomido-2-
methylpropane-sulphonic
acid sodium salt.
Velocity, normal stress,
temperature
An inverse relationship between
coefficient of friction and modulus
whereby modulus changes with
temperature.
Chang et
al.131
Friction: Parallel plate rheometer with gel disks
attached to both plates.
Mechanics: Dynamic modulus determined by small-
amplitude oscillatory shear measurements in the
rheometer.
Poly(N-
isoproylacrylamide)
Angular velocity, normal
load, different gel state
(swollen and collapsed)
induced by changes in
solvent composition.
Positive correlation between the
coefficient of friction and dynamic
modulus.
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
35
cellulose hydrogels, a parallel plate rheometer is the most readily available technique to probe
hydrogel-hydrogel friction.
To investigate the behaviour at the interface between pairs of hydrogels a rotational
rheometer with parallel plate geometries has been used as shown in Figure 2.17. Pairs of
hydrogels are attached to parallel plates and brought into compressive contact at a constant
normal stress, after the system has equilibrated one of the plates is rotated at a specific
angular velocity while the frictional torque is recorded over time131, 134-136. In previous studies
using this technique, the elastic modulus of hydrogels is measured external to the friction test
by indentation tests134, or in situ using small amplitude oscillatory shear before friction
testing131, 135. The ability to measure the mechanics of hydrogel in situ is a major advantage of
using a parallel plate rheometer to investigate hydrogel friction. This powerful technique has
not previously provided insight into the mechanics of plant cell walls because it is not well
suited to plant material.
Figure 2.17. Gel-Gel friction experiment in a rotational rheometer (reproduced from Chang
et al.131).
The crucial challenge with using parallel plate geometry for friction measurements is the
unknown contact area. This is highlighted explicitly by Yamamoto et al.137, who capture
images of the contact between polyacrylamide hydrogels and glass in water. As shown in
Figure 2.18, there are regions of contact and regions of trapped water. The degree of contact
was found to be dependent on velocity and contact pressure. These observations are
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
36
consistent with an early study by Roberts138, who experimentally visualised the interfacial
film between rubber and glass with optical interference measurements. Elastic deformation of
the rubber surface leads to ready entrapment of liquid at the interface. Adhesive forces
between the rubber and glass lead to contact at the point where the opposing surfaces are
closest. The contact area then spreads to trap pockets of fluid at the interface, that is, wetted
regions are surrounded by areas of solid-solid contact. A decrease in lubricant viscosity
resulted in the fluid pockets disappearing at a rate inversely propositional to the viscosity.
Squeeze film lubrication theory has been applied to study the behaviour of a fluid film at the
interface between hydrogel surfaces when they approach each other in the normal direction.
The behaviour of the film is typically described by a modified Reynolds equation. The load
carrying capacity, film thickness, and squeeze time are predicted for hydrogel systems with
varying permeability and elastic modulus139-141. I will use this modelling approach in
combination with the well-defined model for the mechanics of bacterial cellulose systems to
Figure 2.18. An aerial image and corresponding side-view illustration of the contact between
a hydrogel and glass surface. There is a region of trapped water at the interface that causes
the area of contact to be less than the area of the hydrogel. (Reproduced from Yamamoto et
al.137).
predict the contact area between parallel hydrogel disks bought into compression in a
solvated environment. Findings from this study will have significant impact on the ability to
accurately interpret the friction response, and relate it to the mechanics of hydrogels.
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
37
2.3.3 Cellulose Hydrogels Mechanics and the Relevance to Plant Cell Walls
Bacterial cellulose hydrogels are compressible poroelastic gels that relax to near zero loads
even after compression to high strains (>50%)127, 142. The mechanical behaviour of these
systems during compression-relaxation has been previously characterised from both
experimental127, 142 and modelling approaches127, 132.
Bacterial cellulose hydrogel properties are modulated by creating composite hydrogels with
arabinoxylan (AX) and xyloglucan (XG). The effect of hemicelluloses on the mechanical
properties of cellulosic hydrogels have been studied using uniaxial143, 144 and biaxial49 tensile
testing, and under compression127. From uni-axial testing it was observed that
cellulose/xyloglucan composites have a lower modulus (decreased stiffness and increased
extensibility) compared to the pure cellulose network143. Deformation of cellulose/xyloglucan
composites via equi-biaxial tension, mimicking the effect of turgor pressure on the cell wall,
shows increased compliancy and time-dependent creep behaviour compared to pure
cellulose49. It is suggested that during the biosynthesis of bacterial cellulose, the adsorption of
XG onto the cellulose fibril surface reduces the number of fibre contacts within a network 124.
A reduced number of fibre entanglements in the presence of XG is a common hypothesis for
the observed increase in extensibility of cellulose-xyloglucan composites under tension,
compared to cellulose hydrogels49, 143. Moreover, this supports the biological interpretation of
the role of XG in the tethered network model of the plant cell wall. The presence of
hemicellulose cross-links prevents microfibril aggregation, resulting in a decreased
mechanical strength that facilitates turgor-mediated cell expansion/extension25, 143. The
increase in extensibility due to the presence of arabinoxylan in the cellulose composite is
modest compared to xyloglucan144.
Under compressive strain the presence of arabinoxylan and xyloglucan in cellulose
composites increases the modulus compared to pure cellulose, although this effect is
dependent on compressive strain rate142. At higher compressive strain rates (>10µm/s) the
effect of arabinoxylan and xyloglucan are similar, however at slower rates (1µm/s) the
arabinoxylan has little effect compared to xyloglucan. These results are contrary to those
from uni- and bi-axial tensile testing because the surfaces applying compression are
impermeable, limiting water flow to the radial direction which has lower porosity than in the
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
38
axial direction. These mechanical studies again highlight the important role of water
movement.
The lubricating role of water and other polysaccharides at fibril-fibril contacts has been
investigated by characterising the fibril orientation and interaction in three bacterial cellulose
composites (cellulose C, cellulose/pectin CP, and cellulose xyloglucan CXG) for an applied
uniaxial strain using FT-IR and dynamic 2D FT-IR spectroscopies145. On the molecular level,
the reorientation of cellulose fibrils is in the direction of the applied strain. ‘The cellulose-
network reorientation depends on the composition of the matrix, including water content,
which lubricates the motion of macromolecules in the network’145. Enhanced reorientation of
cellulose fibrils in systems with higher water content supports the role of water in lubricating
fibrillar surfaces in molecular motion. Results from dynamic 2D FT-IR experiments suggest
that the independent response of cellulose and pectin to a small amplitude strain could be a
result of few fibril contacts in the composite. Whereas cross-linked cellulose-xyloglucan
domains are uniformly strained along the direction of stretch and the alignment results in
greater extensibility. Pure cellulose networks contain the highest density of mechanically
relevant contact points which are more easily disrupted upon elongation compared to CXG
composites.
The findings from Kacurakova et al.145 provide a starting point for a more detailed
investigation into the role of cell wall components on the mechanics of hydrogel systems.
This section of the literature review has emphasised the need for a combined experimental
and modelling approach to illuminate the exact microstructural features that explain the
mechanics of bacterial cellulose hydrogels. For example, to evaluate the hypothesis that XG
reduces the number of contacts between cellulose fibres in the network, the number of cross
links in the model network can be varied in silico. The outputs of the model can then be
compared to the mechanical measurements of bacterial cellulose hydrogels and composites
with XG. These findings will contribute new knowledge in terms of the lubricating
mechanisms of wall components at intercellular contacts during plant growth.
2.4 Measurement of Fibre-Fibre Interactions
The mechanical properties of a range of fibrous materials have been studied using
compression, uni- and bi-axial tensile testing and small-amplitude oscillatory shear49, 127, 143,
146-153. The results from these studies indicate that the mechanics of random fibre networks is
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
39
defined by the intrinsic mechanical properties of nanofibres, surface interactions between
fibres, the network microstructure, and number and nature of entanglements and/or cross-
links, as well as the solvent properties.
The strength of adhesion between cellulose fibres and the surrounding polymer matrix is a
key factor that determines the mechanical properties of cellulose composites, including
natural cellulose fibre systems such as plant cell walls. The remarkable load bearing capacity,
in combination with mechanical toughness, makes such materials stand out amongst other
biocompatible hydrogels for material applications. Currently, the most reliable information
regarding inter-fibre adhesion can be deduced indirectly from analysis of macroscopic
mechanical properties of composites. Inputs into mechanical models of fibrous assemblies
include the network structure, mechanics of the fibres and fibre junctions. A review of the
fibre network models, experimental approaches for measuring mechanics of single fibres and
interactions between individual fibres is detailed here.
2.4.1 Fibre Network Models
The development of structural models of fibre networks provides predictive capabilities for
design and evaluation, as well as enhancing understanding of the underlying principles
controlling deformation processes in natural systems. In order to model fibre networks,
specific assumptions around the mechanics of fibre-fibre contacts have to be implemented.
Fibre network models commonly treat contacts between fibres as rigid junctions, which can
approximate elastic network deformation (i.e. no sliding at fibre contacts)154-157. However,
this is only appropriate when fibre interactions are very strong. Constitutive equations are
constructed such that the stress-strain response depends on orientation, density and fibre
properties. The assembly of fibres in the network strongly influences the overall network
mechanics. This is evidenced when comparing the mechanics of fibres and fibre network of
bacterial cellulose. The Young’s modulus of individual cellulose fibres is of the order of 100
GPa158, and is orders of magnitude greater than the modulus of bacterial cellulose which is of
the order of 1 MPa49, 127, 143. Moreover, bacterial cellulose hydrogels incorporating XG have a
modulus of the order 0.1 MPa49, 127, 143. I expect that XG has negligible effect on the
individual cellulose fibre modulus, or at least not a sufficient effect to account for the order of
magnitude reduction in the modulus at the hydrogel level. Instead, it is likely that XG reduces
the hydrogel modulus through its effect on fibre orientation, number fibre contact points, and
the mechanics of fibre contacts.
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
40
Mechanical testing of cellulose hydrogels shows plastic deformation127, which suggests that
fibre-fibre contacts are not rigid. Pan and Carnaby159 present a mechanical model of a fibre
network that incorporates fibre bending and slippage at contact points. A force balance based
on the applied contact forces and the fibre orientation is implemented to determine if contact
points are sliding (inter-fibre friction) or non-slipping (deformation through bending of fibre
segments) in response to an applied shear stress. For viscoelastic fibre networks, Chatterjee160
applied an energy penalty to the breaking of each fibre-fibre contact and related it to the
stored elastic energy of deformation. Heyden and Gustafsson161 include a slip criterion for
bonds in a network mechanics model whereby bonds show non-linear stick-slip facture
behaviour and slip leads to the degradation of stiffness and strength properties of the bond.
Using a finite element method of analysis, a comparison is made between 3D network
simulations and experimental results for dry-shaped cellulose fibre materials. The modelling
and experimental results are compared in terms of homogenised mechanical properties such
as stress versus strain performance under tensile deformation, initial anisotropic 2D and 3D
stiffness properties, strength and fracture energy. Reasonable agreement was observed up to
peak tensile stress, after which considerations around strain localisation and heterogeneity in
the material take effect161.
I recognise the need to build models that can predict the yield point at fibre-fibre contacts,
based on the plastic deformation behaviour of fibre networks. This would provide important
insights into the loosening of cellulose fibre contacts in the cell wall during extension and
plant growth. Accurate knowledge on the fibre mechanics and adhesive potential between
fibres at network junctions will be required to build and validate such a model, and this
information can only obtained through experimentation.
2.4.2 Fibre Mechanics
The mechanical properties of cellulose fibres have primarily been measured with the AFM
using nano-indentation techniques. Single cellulose microfibrils are horizontally suspended
over a trench, and the Young’s modulus is measured using a modified nano-scale three-point
bending test where the fibre is deflected by the AFM cantilever with a known force at several
points along the fibre158. A bacterial cellulose suspension was dispersed by sonication, placed
on a silicon-nitride coated grating with a pitch of 3 µm, then spin coated to disperse the fibres
as a uniform monolayer as seen in Figure 2.19.
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
41
Figure 2.19. Cellulose fibres suspended over grating (reproduced from Guhados et al.158).
For mechanical testing, force spectra were obtained by recording cantilever deflection as a
function of vertical sample displacement as shown in Figure 2.20158. The Young’s modulus is
determined from the slope of the force spectra according to Equation 2.1 where k is the
cantilever spring constant, E is the Young’s modulus, I is the area moment of inertia (I =
(πwh3)/64 for width, w and height, h of a fibre with elliptical cross-section), a is the distance
from one end of the fibre, and L is the length of the fibre.
𝑑𝑦
𝑑𝑧= [1 +
𝑘
3𝐸𝐼(
𝑎(𝐿 − 𝑎)
𝐿)
3
]
−1
(2.1)
The measured Young’s modulus was 78+/-17 GPa for fibres with diameters ranging from 35
to 90 nm. This experimental procedure, depicted in Figure 2.21, looks at only a small vertical
deflection from which only the elastic modulus can be determined. A slight modification to
this configuration has been used to pull the fibre laterally, allowing for a larger deflection162.
Investigating the mechanics of fibres under large deformations is more informative in terms
of characterising the stretching and bending of the fibre. This level of detail is required for
building a network model that can predict plastic deformation behaviour which is expected to
involve large deformation of the fibres before slippage or detachment at contact points.
Instead of vertical deflection, the lateral deflection of an AFM cantilever can be used as a
force sensor for measuring the tensile properties of individual electrospun polymer nano-
fibres suspended across trenches162. The fibres were electrospun and glued onto a parallel bar
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
42
TEM grid where the spaces between the bars act as trenches that are deep enough to allow the
AFM cantilever to hook onto the fibre and pull it laterally as shown in Figure 2.22. Using the
lateral motion of the cantilever to probe the mechanical properties of a suspended nano-fibre
increases the range of motion significantly, compared to the vertical geometry, enabling the
breaking strength and elongation at break to be measured.
Figure 2.20. Force-distance curve obtained near the middle of the suspended fibre (Solid
curve: approach. Dashed curve: retraction) (reproduced from Guhados et al.158).
Figure 2.21. Fibre suspended across a trench of distance L, deformed by a vertical force F
applied at a distance a from one end of the fibre (Reproduced from Guhados et al.158).
This three-point testing configuration deforms the fibre via tensile stretching rather than
bending due to the low modulus and high aspect ratio of the tested fibre. This assumption is
validated by the non-linear force-cantilever travel curve in the elastic region as seen in Figure
2.23. Assuming a tensile stretching model, the corresponding stress-strain curve is derived
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
43
using Equations 2.2, 2.3 and 2.4. σ and ε are the fibre tensile stress and strain respectively, Fl
is the cantilever lateral force, A is the cross-sectional area of the fibre, x is the cantilever
travel from the initial point of contact with the fibre, θ is the angle between the stretched and
initial fibre position and l0 is half the initial length of the fibre. The Young’s modulus, yield
stress/strain, and breaking stress/strain of the test fibre are obtained from the stress-strain
curves. In this thesis I use the large range of motion of the cantilever in the lateral
configuration to probe the yielding and friction behaviour at cellulose fibril junctions as a
single fibre is pulled out of cellulose and composite networks.
𝜎 = 𝐹𝑙
2𝐴𝑠𝑖𝑛(𝜃)
(2.2)
𝜀 =(
𝑥sin (𝜃)
) − 𝑙0
𝑙0
(2.3)
tan(𝜃) =𝑥
𝑙0 (2.4)
Figure 2.22. Schematic diagram of an AFM cantilever dragging a fibre laterally (reproduced
from Gestos et al.162).
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
44
Figure 2.23. Force-cantilever and stress-strain curves in the elastic region for single test
nanofibre (reproduced from Gestos et al.162).
2.4.3 Experimental Approaches for Fibre-Fibre Measurement
Direct measurement of fibre adhesion has been achieved for sub-micron (>100 nm)
electrospun fibres through a variety of elaborate experiments. This includes taping
freestanding fibres to cardboard mounts163, 164 or gluing between two microspheres on an
AFM cantilever165 as shown in Figure 2.24. Two fibres are arranged orthogonal to each other
at the midpoint in a cross-cylinder configuration and pressed into contact for a given time.
The applied load and vertical displacement are monitored as the fibres are pulled apart at a
constant speed and the pull-off force is measured. Shi et al. (2010, 2012) show that for
polycaprolactone and Nylon 6 fibers, the pull-off force increases linearly with radius of the
fiber. Stachewicz et al.166 also measure the adhesion force in parallel configurations by
attaching two individual fibres to AFM tips at one end and bringing the free fibre ends into
contact. The measured pull-off force critically depends on the contact area which changes
with fibre configuration. Despite the free fibre ends in the Stachewicz et al.166 assembly, the
degrees of freedom are still restricted to only enable the two extreme fiber orientations;
parallel and cross-cylinder as shown in Figure 2.25.
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
45
Figure 2.24. Experimental configurations for fibre-fibre measurements (Reproduced from
Shi et al.163 and Wang et al.165).
It is challenging to use the methods described above to assess all the possible fibre
orientations between parallel and orthogonal, that are observed in randomly assembled
fibrous networks. Another limitation of these techniques is the inability to measure the
potential for fibres to interact via mechanical entanglement. Xing et al.167 show that there is a
significant effect on adhesion if fibers are permitted to wrap around each other. In their
experiment an electrospun polystyrene nanofiber is observed to wrap around a nanoparticle
attached to an AFM tip.
The experimental techniques describe above that look at the contact between single
electrospun fibres have not been applied to cellulose fibres due to the difficulty isolation and
handling individual fibres. Gutsmann et al.168 attempt to measure the mechanical properties
of a highly organised hierarchical assembly of collagen fibres. Force-extension curves are
collected when pulling parallel substructures out of the assembly using an AFM cantilever
however it could not be confirmed whether the rupture events observed were between the tip
and the molecule or between the molecules. Yan and Li169 use the AFM with an –OH
functionalized cantilever tip to measure the inter- fibre bonding properties , including pull-off
forces and work of adhesion, of wet wood pulp fibre surfaces that are solid, swollen and
micro-fibrillated. The technique is limited by the fact that the inter-fibre bonding properties
are controlled by surface deformability of the pulp fibre. There is large scope for developing
a novel method that can be applied to measure the mechanics and forces at contact between
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
46
individual cellulose fibres that eliminates the effect bulk deformation of the fibre assembly.
This would have major implications for understanding the role of interactions between
cellulose and other cell wall components including expansins on plant cell wall mechanics
during growth processes.
Figure 2.25. Two fibre free ends arranged in parallel and cross-cylinder contacts (reproduced
from Stachewicz et al.166).
2.5 Future Perspective: Scope and Goals of Thesis
A significant amount of literature is presented that investigates the structure of plant cell wall
materials, and the physical processes underlying plant growth. The tribological contacts that
are essential for facilitating growth include the deforming interface between adjacent cell
walls in the plant tissue, and the shearing of conjoined cellulose fibres and surrounding
matrix polymers in an extending cell wall. I postulate that certain lubrication mechanisms are
controlling the friction response at these contacts, but these have not yet been substantiated.
Mechanical measurements on plant tissue and isolated wall fragments have shown that the
materials are poroelastic. This highlights the scope for systematically exploring plant cell
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
47
wall mechanics using model hydrogel systems, which are defined by their poroelasticity. The
mechanics of bacterial cellulose and composite hydrogels have been investigated. However
the reported findings are mostly limited to experimental observations, with no way to
interpret the microstructural components that determine the macro-scale material properties.
A step-change in the approach to investigating hydrogel mechanics is provided through the
use of modelling by Lopez-Sanchez et al.127 and Bonilla et al.132. I will extend the use of a
combined experiment-modelling approach to interrogate the role of cell wall components and
their assembly on the mechanics and friction behaviour of bacterial cellulose hydrogels.
In previous work on articular cartilage and other hydrogel systems, researchers have drawn
correlations between the hydrogel modulus and the coefficient of friction. However, results
from these studies cannot be distilled down to any universal material function that relates
hydrogel mechanics and friction properties. I have identified an opportunity to use a
rotational rheometer to measure the friction response and in situ mechanical properties of
bacterial cellulose hydrogels. Coupled with the existing mechanical model127, 132 for the
hydrogel system, I will construct a generalised relationship between mechanics and friction
properties. This study will provide a compelling case for the potential role of cell wall
mechanics in lubricating the intercellular contacts during plant growth.
The second major goal of this thesis is to directly measure the adhesive energy stored at
cellulose fibre contacts in the absence of, or mediated by other non-cellulosic wall polymers.
This study will be revealing in terms of the way cellulose fibres linkages are formed, and the
structural role of individual cell wall components. A primary motivation for this work is the
need to have specific information on the mechanics of fibres contact in order to build a fibre
network model. The direct measurement technique and fibre network model will be
invaluable for gaining new insight into the action of expansins that loosens fibre contact and
drives cell wall extension.
This thesis has significant impact for the eventual development of a working mechanical
model that predicts the multi-scale material properties of plant tissue, and the physical
processes that underlie cell wall extension and plant growth. The key distinction between this
thesis and previous studies is the use of model systems to systematically probe individual
structural features, coupled with finite element modelling to interpret the effect of structural
features on the overall mechanical properties. I provide unique perspectives through the
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
48
application of an engineering and physical sciences toolkit to investigate plant biological
processes.
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
49
References for Chapter 2
1. G. W. Stachowiak and A. W. Batchelor, Engineering Tribology, 3rd Edition, 2005.
2. B. Bhushan, Introduction to Tribology, 2nd Edition., John Wiley & Sons , Ltd, UK,
2nd. edn., 2013.
3. J. P. Gong, G. Kagata, Y. Iwasaki and Y. Osada, Chinese Journal of Polymer Science,
2000, 18, 271-275.
4. J. Klein, E. Kumacheva, D. Mahalu, D. Perahia and L. J. Fetters, Nature, 1994, 370,
634-636.
5. J. Klein, U. Raviv, S. Perkin, N. Kampf, L. Chai and S. Giasson, J. Phys.-Condes.
Matter, 2004, 16, S5437-S5448.
6. S. Lee and N. D. Spencer, Science, 2008, 319, 575-576.
7. L. Mahadevan, S. Daniel and M. K. Chaudhury, Proceedings of the National
Academy of Sciences of the United States of America, 2004, 101, 23-26.
8. M. Andrianifahanana, N. Moniaux and S. K. Batra, Biochimica Et Biophysica Acta-
Reviews on Cancer, 2006, 1765, 189-222.
9. M. C. Rose, Am. J. Physiol., 1992, 263, L413-L429.
10. A. Dedinaite, Soft Matter, 2012, 8, 273-284.
11. S. M. T. Chan, C. P. Neu, G. DuRaine, K. Komvopoulos and A. H. Reddi,
Osteoarthritis and Cartilage, 2010, 18, 956-963.
12. T. Hardingham and M. Bayliss, Seminars in Arthritis and Rheumatism, 1990, 20, 12-
33.
13. M. T. Muller, X. P. Yan, S. W. Lee, S. S. Perry and N. D. Spencer, Macromolecules,
2005, 38, 5706-5713.
14. B. Zappone, M. Ruths, G. W. Greene, G. D. Jay and J. N. Israelachvili, Biophysical
Journal, 2007, 92, 1693-1708.
15. J. Klein, Annual Review of Materials Science, 1996, 26, 581-612.
16. J. Klein, Proceedings of the Institution of Mechanical Engineers Part J-Journal of
Engineering Tribology, 2006, 220, 691-710.
17. L. Han, A. J. Grodzinsky and C. Ortiz, in Annual Review of Materials Research, Vol
41, eds. D. R. Clarke and P. Fratzl, 2011, vol. 41, pp. 133-168.
18. M. Chen, W. H. Briscoe, S. P. Armes and J. Klein, Science, 2009, 323, 1698-1701.
19. L. Macakova, G. E. Yakubov, M. A. Plunkett and J. R. Stokes, Tribology
International, 2011, 44, 956-962.
20. D. J. Cosgrove, Nature Reviews Molecular Cell Biology, 2005, 6, 850-861.
21. S. Wolf, K. Hematy and H. Hoefte, in Annual Review of Plant Biology, Vol 63, ed. S.
S. Merchant, 2012, vol. 63, pp. 381-407.
22. N. C. Carpita and D. M. Gibeaut, Plant Journal, 1993, 3, 1-30.
23. L. G. Smith, Nature Reviews Molecular Cell Biology, 2001, 2, 33-39.
24. T. Hamann, Frontiers in plant science, 2012, 3, 77-77.
25. C. Somerville, S. Bauer, G. Brininstool, M. Facette, T. Hamann, J. Milne, E. Osborne,
A. Paredez, S. Persson, T. Raab, S. Vorwerk and H. Youngs, Science, 2004, 306,
2206-2211.
26. S.-Y. Ding, S. Zhao and Y. Zeng, Cellulose, 2014, 21, 863-871.
27. L. M. Davies and P. J. Harris, Planta, 2003, 217, 283-289.
28. A. N. Fernandes, L. H. Thomas, C. M. Altaner, P. Callow, V. T. Forsyth, D. C.
Apperley, C. J. Kennedy and M. C. Jarvis, Proceedings of the National Academy of
Sciences of the United States of America, 2011, 108, E1195-E1203.
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
50
29. C. J. Kennedy, G. J. Cameron, A. Sturcova, D. C. Apperley, C. Altaner, T. J. Wess
and M. C. Jarvis, Cellulose, 2007, 14, 235-246.
30. D. P. Oehme, M. T. Downton, M. S. Doblin, J. Wagner, M. J. Gidley and A. Bacic,
Plant Physiology, 2015, 168, 3-U654.
31. R. H. Newman, S. J. Hill and P. J. Harris, Plant Physiology, 2013, 163, 1558-1567.
32. Y. Nishiyama, Journal of Wood Science, 2009, 55, 241-249.
33. J. F. Matthews, M. Bergenstrahle, G. T. Beckham, M. E. Himmel, M. R. Nimlos, J.
W. Brady and M. F. Crowley, Journal of Physical Chemistry B, 2011, 115, 2155-
2166.
34. O. Lerouxel, D. M. Cavalier, A. H. Liepman and K. Keegstra, Current Opinion in
Plant Biology, 2006, 9, 621-630.
35. K. Kafle, X. Xi, C. M. Lee, B. R. Tittmann, D. J. Cosgrove, Y. B. Park and S. H.
Kim, Cellulose, 2014, 21, 1075-1086.
36. T. Zhang, S. Mahgsoudy-Louyeh, B. Tittmann and D. J. Cosgrove, Cellulose, 2014,
21, 853-862.
37. T. I. Baskin, in Annu. Rev. Cell Dev. Biol., Annual Reviews, Palo Alto, 2005, vol. 21,
pp. 203-222.
38. D. Suslov, J. P. Verbelen and K. Vissenberg, Journal of Experimental Botany, 2009,
60, 4175-4187.
39. C. T. Anderson, A. Carroll, L. Akhmetova and C. Somerville, Plant Physiology, 2010,
152, 787-796.
40. E. F. Crowell, M. Gonneau, S. Vernhettes and H. Hofte, C. R. Biol., 2010, 333, 320-
324.
41. K. Kulasinski, S. Keten, S. V. Churakov, D. Derome and J. Carmeliet, Cellulose,
2014, 21, 1103-1116.
42. K. Keegstra, K. W. Talmadge, W. D. Bauer and Albershe.P, Plant Physiology, 1973,
51, 188-196.
43. B. Zhao and H. J. Kwon, Journal of Adhesion Science and Technology, 2011, 25,
557-579.
44. L. D. Talbott and P. M. Ray, Plant Physiology, 1992, 98, 357-368.
45. K. Baba, Y. Sone, A. Misaki and T. Hayashi, Plant and Cell Physiology, 1994, 35,
439-444.
46. M. Hrmova, V. Farkas, J. Lahnstein and G. B. Fincher, Journal of Biological
Chemistry, 2007, 282, 12951-12962.
47. T. Fujino, Y. Sone, Y. Mitsuishi and T. Itoh, Plant and Cell Physiology, 2000, 41,
486-494.
48. Y. B. Park and D. J. Cosgrove, Plant Physiology, 2012, 158, 1933-1943.
49. E. Chanliaud, K. M. Burrows, G. Jeronimidis and M. J. Gidley, Planta, 2002, 215,
989-996.
50. M. McNeil, P. Albersheim, L. Taiz and R. L. Jones, Plant Physiology, 1975, 55, 64-
68.
51. D. Burke, P. Kaufman, M. McNeil and Albershe.P, Plant Physiology, 1974, 54, 109-
115.
52. I. Braccini, R. P. Grasso and S. Perez, Carbohydrate Research, 1999, 317, 119-130.
53. F. Liners, J. F. Thibault and P. Vancutsem, Plant Physiology, 1992, 99, 1099-1104.
54. T. Ishii, T. Matsunaga, P. Pellerin, M. A. O'Neill, A. Darvill and P. Albersheim,
Journal of Biological Chemistry, 1999, 274, 13098-13104.
55. J. A. Brown and S. C. Fry, Plant Physiology, 1993, 103, 993-999.
56. M. Marry, K. Roberts, S. J. Jopson, I. M. Huxham, M. C. Jarvis, J. Corsar, E.
Robertson and M. C. McCann, Physiologia Plantarum, 2006, 126, 243-256.
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
51
57. A. W. Zykwinska, M. C. J. Ralet, C. D. Garnier and J. F. J. Thibault, Plant
Physiology, 2005, 139, 397-407.
58. A. Zykwinska, J.-F. Thibault and M.-C. Ralet, Carbohydrate Polymers, 2008, 74,
957-961.
59. A. Zykwinska, J.-F. Thibault and M.-C. Ralet, Carbohydrate Polymers, 2008, 74, 23-
30.
60. A. Peaucelle, S. Braybrook and H. Hoefte, Frontiers in plant science, 2012, 3.
61. F. Liners and P. Vancutsem, Protoplasma, 1992, 170, 10-21.
62. W. G. T. Willats, C. Orfila, G. Limberg, H. C. Buchholt, G. van Alebeek, A. G. J.
Voragen, S. E. Marcus, T. Christensen, J. D. Mikkelsen, B. S. Murray and J. P. Knox,
Journal of Biological Chemistry, 2001, 276, 19404-19413.
63. R. Goldberg, C. Morvan and J. C. Roland, Plant and Cell Physiology, 1986, 27, 417-
429.
64. F. B. Daher and S. A. Braybrook, Frontiers in plant science, 2015, 6.
65. M. C. Jarvis, Plant Cell and Environment, 1998, 21, 1307-1310.
66. T. M. Schindler, R. Bergfeld, P. Vancutsem, D. Vonsengbusch and P. Schopfer,
Protoplasma, 1995, 188, 213-224.
67. L. J. Gibson, Journal of the Royal Society Interface, 2012, 9, 2749-2766.
68. M. S. Zamil, H. J. Yi, M. A. Haque and V. M. Puri, American Journal of Botany,
2013, 100, 1105-1115.
69. M. S. Zamil, H. J. Yi and V. M. Puri, Journal of Materials Science, 2015, 50, 6608-
6623.
70. P. M. Pieczywek and A. Zdunek, Journal of Food Engineering, 2014, 123, 50-59.
71. E. Vanstreels, A. C. Alamar, B. E. Verlinden, A. Enninghorst, J. K. A. Loodts, E.
Tijskens, H. Ramon and B. M. Nicolai, Postharvest Biology and Technology, 2005,
37, 163-173.
72. C. X. Wang, J. Pritchard and C. R. Thomas, J. Texture Stud., 2006, 37, 597-606.
73. S. Digiuni, A. Berne-Dedieu, C. Martinez-Torres, J. Szecsi, M. Bendahmane, A.
Arneodo and F. Argou, Biophysical Journal, 2015, 108, 2235-2248.
74. M. R. Bonilla, J. R. Stokes, M. J. Gidley and G. E. Yakubov, Soft Matter, 2015, 11,
1281-1292.
75. P. Van Liedekerke, P. Ghysels, E. Tijskens, G. Samaey, B. Smeedts, D. Roose and H.
Ramon, Phys. Biol., 2010, 7.
76. D. J. Cosgrove, Int. J. Plant Sci., 1993, 154, 10-21.
77. R. Yamamoto, J. Plant Res., 1996, 109, 75-84.
78. H. Kawamura, S. Kamisaka and Y. Masuda, Plant and Cell Physiology, 1976, 17, 23-
34.
79. R. J. Dyson, L. R. Band and O. E. Jensen, J. Theor. Biol., 2012, 307, 125-136.
80. D. J. Cosgrove, Annual Review of Plant Physiology and Plant Molecular Biology,
1999, 50, 391-417.
81. J. Harholt, A. Suttangkakul and H. V. Scheller, Plant Physiology, 2010, 153, 384-395.
82. T. I. Baskin and O. E. Jensen, Journal of Experimental Botany, 2013, 64, 4697-4707.
83. H. I. Wu, R. D. Spence, P. J. H. Sharpe and J. D. Goeschl, Plant Cell and
Environment, 1985, 8, 563-570.
84. L. Aouar, Y. Chebli and A. Geitmann, Sex. Plant Reprod., 2010, 23, 15-27.
85. H. Kha, S. C. Tuble, S. Kalyanasundaram and R. E. Williamson, Plant Physiology,
2010, 152, 774-786.
86. D. Suslov and J. P. Verbelen, Journal of Experimental Botany, 2006, 57, 2183-2192.
87. C. F. Wei, L. S. Lintilhac and P. M. Lintilhac, Planta, 2006, 223, 1058-1067.
88. H. J. Yi and V. M. Puri, Plant Physiology, 2012, 160, 1281-1292.
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
52
89. H. J. Yi and V. M. Puri, American Journal of Botany, 2014, 101, 244-254.
90. V. S. T. Van Sandt, D. Suslov, J. P. Verbelen and K. Vissenberg, Annals of Botany,
2007, 100, 1467-1473.
91. A. Peaucelle, R. Wightman and H. Hoefte, Curr. Biol., 2015, 25, 1746-1752.
92. S. A. Braybrook and A. Peaucelle, PLoS ONE, 2013, 8.
93. D. J. Cosgrove, Nature, 2000, 407, 321-326.
94. D. L. Rayle and R. E. Cleland, Plant Physiology, 1992, 99, 1271-1274.
95. B. M. Link and D. J. Cosgrove, Plant Physiology, 1998, 118, 907-916.
96. K. M. Fenwick, D. C. Apperley, D. J. Cosgrove and M. C. Jarvis, Phytochemistry,
1999, 51, 17-22.
97. A. Peaucelle, S. Braybrook and H. Hofte, Frontiers in plant science, 2012, 3.
98. S. McQueenmason and D. J. Cosgrove, Proceedings of the National Academy of
Sciences of the United States of America, 1994, 91, 6574-6578.
99. S. J. McQueenmason and D. J. Cosgrove, Plant Physiology, 1995, 107, 87-100.
100. D. J. Cosgrove, Bioessays, 1996, 18, 533-540.
101. D. J. Cosgrove, Current Opinion in Plant Biology, 2000, 3, 73-78.
102. S. E. C. Whitney, M. J. Gidley and S. J. McQueen-Mason, Plant Journal, 2000, 22,
327-334.
103. Q. Wang, L. Chen, H. Lin, D. B. Yu, Q. Shen, L. Wan and Y. H. Zhao, Enzyme and
Microbial Technology, 2016, 92, 67-75.
104. R. L. Silveira and M. S. Skaf, Physical Chemistry Chemical Physics, 2016, 18, 3510-
3521.
105. D. S. Thompson, Journal of Experimental Botany, 2005, 56, 2275-2285.
106. S. McQueenmason, D. M. Durachko and D. J. Cosgrove, Plant Cell, 1992, 4, 1425-
1433.
107. E. S. Kim, H. J. Lee, W.-G. Bang, I.-G. Choi and K. H. Kim, Biotechnology and
Bioengineering, 2009, 102, 1342-1353.
108. B. Bunterngsook, W. Mhuantong, V. Champreda, A. Thamchaipenet and L.
Eurwilaichitr, Bioresource Technology, 2014, 159, 64-71.
109. Y. Seki, Y. Kikuchi, R. Yoshimoto, K. Aburai, Y. Kanai, T. Ruike, K. Iwabata, R.
Goitsuka, F. Sugawara, M. Abe and K. Sakaguchi, Planta, 2015, 241, 83-93.
110. N. Georgelis, A. Tabuchi, N. Nikolaidis and D. J. Cosgrove, Journal of Biological
Chemistry, 2011, 286, 16814-16823.
111. M. W. Shieh and D. J. Cosgrove, J. Plant Res., 1998, 111, 149-157.
112. T. Wang, Y. B. Park, M. A. Caporini, M. Rosay, L. Zhong, D. J. Cosgrove and M.
Hong, Proceedings of the National Academy of Sciences of the United States of
America, 2013, 110, 16444-16449.
113. I. J. Kim, H.-J. Ko, T.-W. Kim, I.-G. Choi and K. H. Kim, Biotechnology and
Bioengineering, 2013, 110, 401-407.
114. B. Bunterngsook, L. Eurwilaichitr, A. Thamchaipenet and V. Champreda,
Bioresource Technology, 2015, 176, 129-135.
115. A. Lipchinsky, Acta Physiologiae Plantarum, 2013, 35, 3277-3284.
116. J. A. Fozard, M. Lucas, J. R. King and O. E. Jensen, Frontiers in plant science, 2013,
4.
117. D. Kwiatkowska and J. Dumais, Journal of Experimental Botany, 2003, 54, 1585-
1595.
118. M. Uyttewaal, A. Burian, K. Alim, B. T. Landrein, D. Borowska-Wykret, A. Dedieu,
A. Peaucelle, M. Ludynia, J. Traas, A. Boudaoud, D. Kwiatkowska and O. Hamant,
Cell, 2012, 149, 439-451.
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
53
119. C. T. Anderson, I. S. Wallace and C. R. Somerville, Proceedings of the National
Academy of Sciences of the United States of America, 2012, 109, 1329-1334.
120. M. Iguchi, S. Yamanaka and A. Budhiono, Journal of Materials Science, 2000, 35,
261-270.
121. D. Mikkelsen and M. J. Gidley, in Plant Cell Wall: Methods and Protocols, ed. Z. A.
Popper, Humana Press Inc, Totowa, 2011, vol. 715, pp. 197-208.
122. R. M. Brown, J. Macromol. Sci.-Pure Appl. Chem., 1996, A33, 1345-1373.
123. M. Martinez-Sanz, P. Lopez-Sanchez, M. J. Gidley and E. P. Gilbert, Cellulose, 2015,
22, 1541-1563.
124. J. Gu and J. M. Catchmark, Cellulose, 2014, 21, 275-289.
125. C. G. Armstrong, W. M. Lai and V. C. Mow, Journal of Biomechanical Engineering-
Transactions of the Asme, 1984, 106, 165-173.
126. B. Cohen, W. M. Lai and V. C. Mow, Journal of Biomechanical Engineering-
Transactions of the Asme, 1998, 120, 491-496.
127. P. Lopez-Sanchez, M. Rincon, D. Wang, S. Brulhart, J. R. Stokes and M. J. Gidley,
Biomacromolecules, 2014, 15, 2274-2284.
128. D. Baykal, R. J. Underwood, K. Mansmann, M. Marcolongo and S. M. Kurtz, Journal
of the Mechanical Behavior of Biomedical Materials, 2013, 28, 263-273.
129. M. A. Soltz and G. A. Ateshian, Journal of Biomechanical Engineering-Transactions
of the Asme, 2000, 122, 576-586.
130. D. Zhang, Y. Shen and S. Ge, Science in China Series E-Technological Sciences,
2009, 52, 2474-2480.
131. D. P. Chang, J. E. Dolbow and S. Zauscher, Langmuir, 2007, 23, 250-257.
132. M. R. Bonilla, P. Lopez-Sanchez, M. J. Gidley and J. R. Stokes, Acta Biomater.,
2016, 29, 149-160.
133. A. Kozbial and L. Li, Materials Science & Engineering C-Materials for Biological
Applications, 2014, 36, 173-179.
134. T. Kurokawa, J. P. Gong and Y. Osada, Macromolecules, 2002, 35, 8161-8166.
135. G. Kagata, J. P. Gong and Y. Osada, Journal of Physical Chemistry B, 2002, 106,
4596-4601.
136. J. P. Gong, G. Kagata and Y. Osada, Journal of Physical Chemistry B, 1999, 103,
6007-6014.
137. T. Yamamoto, T. Kurokawa, J. Ahmed, G. Kamita, S. Yashima, Y. Furukawa, Y. Ota,
H. Furukawa and J. P. Gong, Soft Matter, 2014, 10, 5589-5596.
138. A. D. Roberts, Journal of Physics D-Applied Physics, 1971, 4, 423-&.
139. N. M. Bujurke and R. B. Kudenatti, Appl. Math. Comput., 2006, 174, 1181-1195.
140. M. Yousfi, B. Bou-Said and J. Tichy, Lubr. Sci., 2015, 27, 505-522.
141. N. M. Bujurke, R. B. Kudenatti and V. B. Awati, Math. Biosci., 2007, 209, 76-89.
142. P. Lopez-Sanchez, J. Cersosimo, D. Wang, B. Flanagan, J. R. Stokes and M. J.
Gidley, PLoS ONE, 2015, 10.
143. S. E. C. Whitney, M. G. E. Gothard, J. T. Mitchell and M. J. Gidley, Plant
Physiology, 1999, 121, 657-663.
144. D. Mikkelsen, B. M. Flanagan, S. M. Wilson, A. Bacic and M. J. Gidley,
Biomacromolecules, 2015, 16, 1232-1239.
145. M. Kacurakova, A. C. Smith, M. J. Gidley and R. H. Wilson, Carbohydrate Research,
2002, 337, 1145-1153.
146. D. Poquillon, B. Viguier and E. Andrieu, Journal of Materials Science, 2005, 40,
5963-5970.
147. L. L. Ma, J. Y. Xu, P. A. Coulombe and D. Wirtz, Journal of Biological Chemistry,
1999, 274, 19145-19151.
CHAPTER 2. LITERATURE REVIEW Grace Dolan, PhD Thesis
54
148. J. F. Leterrier, J. Kas, J. Hartwig, R. Vegners and P. A. Janmey, Journal of Biological
Chemistry, 1996, 271, 15687-15694.
149. Y. Tseng, K. M. An, O. Esue and D. Wirtz, Journal of Biological Chemistry, 2004,
279, 1819-1826.
150. P. A. Janmey, U. Euteneuer, P. Traub and M. Schliwa, Journal of Cell Biology, 1991,
113, 155-160.
151. Y. Zou and Y. H. Zhang, Ann. Biomed. Eng., 2009, 37, 1572-1583.
152. R. Huang, W. Z. Li, X. X. Lv, Z. J. Lei, Y. Q. Bian, H. B. Deng, H. J. Wang, J. Q. Li
and X. Y. Li, Biomaterials, 2015, 53, 58-75.
153. L. Zhou, H. He, C. Jiang and S. He, Journal of Applied Polymer Science, 2015, 132.
154. T. Komori and M. Itoh, Text. Res. J., 1991, 61, 420-428.
155. M. A. Narter, S. K. Batra and D. R. Buchanan, Proc. R. Soc. A-Math. Phys. Eng. Sci.,
1999, 455, 3543-3563.
156. C. W. Wang, L. Berhan and A. M. Sastry, J. Eng. Mater. Technol.-Trans. ASME,
2000, 122, 450-459.
157. X. F. Wu and Y. A. Dzenis, Journal of Applied Physics, 2005, 98.
158. G. Guhados, W. K. Wan and J. L. Hutter, Langmuir, 2005, 21, 6642-6646.
159. N. Pan and G. A. Carnaby, Text. Res. J., 1989, 59, 285-292.
160. A. P. Chatterjee, Journal of Applied Physics, 2006, 100.
161. S. Heyden and P. J. Gustafsson, J. Pulp Pap. Sci., 1998, 24, 160-165.
162. A. Gestos, P. G. Whitten, G. M. Spinks and G. G. Wallace, Polymer Testing, 2013,
32, 655-664.
163. Q. Shi, K.-T. Wan, S.-C. Wong, P. Chen and T. A. Blackledge, Langmuir, 2010, 26,
14188-14193.
164. Q. Shi, S.-C. Wong, W. Ye, J. Hou, J. Zhao and J. Yin, Langmuir, 2012, 28, 4663-
4671.
165. X. Wang, J. F. Najem, S.-C. Wong and K.-t. Wan, Journal of Applied Physics, 2012,
111.
166. U. Stachewicz, F. Hang and A. H. Barber, Langmuir, 2014, 30, 6819-6825.
167. M. Xing, W. Zhong, X. L. Xu and D. Thomson, Langmuir, 2010, 26, 11809-11814.
168. T. Gutsmann, G. E. Fantner, J. H. Kindt, M. Venturoni, S. Danielsen and P. K.
Hansma, Biophysical Journal, 2004, 86, 3186-3193.
169. Y. a. L. Yan, K., Journal of Materials Science Research, 2013, 2, 23-33.
55
Chapter 3
Research Methods
The scope and goal statement from Chapter 1 Section 1.2 and Chapter 2 Section 2.5 is to
measure the interactive forces between cellulose fibres, and investigate the role of non-
cellulosic cell wall components at these fibre contacts. This thesis focuses on the direct
measurement of model systems to make inferences around the mechanics of plant cell walls
during growth and other deformations. From Chapter 2 Section 2.2.1, plant cellulose has a
crystalline core surrounded by a paracrystalline shell. The chemical and mechanical methods
of extracting plant cellulose disrupt the paracrytalline layer, which is the region that
facilitates cellulose interactions with the hemicellulose amorphous phase in plant cell walls.
Bacterial cellulose has a similar structure to plant cellulose in terms of the crystalline and
paracrystalline phases, and can be produced as a pure cellulose network or as a composite
with hemicelluloses (AX and XG). A tribo-rheological technique is developed in a rotational
rheometer in Chapter 4 to measure the shear forces between pairs of bacterial cellulose
hydrogels and composites that are brought into compressive contact whilst surrounded by
solvent. The influence of AX and XG on the interaction between cellulose fibres at the
interface is investigated. The other major cell wall component that is considered in this thesis
is pectin. Pectin is rich in the middle lamella layer and is hypothesised to play a role in
lubricating the contact between adjacent cell walls. This is tested in Chapter 4 by varying the
pectin concentration in the solvent as pairs of bacterial cellulose hydrogels are compressed
together, and measuring the shear forces at the interface. A dip-and-drag technique is
developed in an AFM to measure the adhesive force between individual nanofibres.
Electrospun polymer fibres and cellulose fibres extracted from plant sources are used to
validate the technique in Chapter 5. The influence of hemicelluloses on bacterial cellulose
fibre contacts is investigated using the dip-and-drag technique in Chapter 6. Finally, both the
CHAPTER 3. RESEARCH METHODOLOGY Grace Dolan, PhD Thesis
56
tribo-rheological and dip-and-drag techniques are used in Chapter 7 to assess the activity of
bacterial expansin on bacterial cellulose systems and elucidate the activity of plant expansins
on plant materials. Here, I detail the materials and measurements on bacterial cellulose
systems that are used in this thesis.
3.1 Materials
3.1.1 Electrospun Fibres
The dip-and-drag technique developed in Chapter 5 is validated by measuring the adhesive
force between electrospun polymer fibres. Polymer fibre adhesion is dominated by DLVO
interactions and thus the measured adhesive force can be compared to the theoretical
prediction from DLVO theory. The other advantage is that the fibre network density can be
varied easily through the electrospinning process to determine whether the bulk network
properties influence the measured adhesive force at single fibre contacts. Electrospun
Sulphonated Poly Ether Ether Ketone (SPEEK) fibre samples with varying fibre network
densities are supplied by Dr. George W. Greene from Deakin University. Electrospun
Polyvinyl Alcohol (PVA) fibres are supplied by Professor Darren Martin’s research group at
the Australian Institute for Bioengineering and Nanotechnology (AIBN), The University of
Queensland.
Electrospun Sulphonated Poly Ether Ether Ketone (SPEEK) fibres
SPEEK is prepared from Victrex polyether ether ketone 450PF (PEEK; Mw = 38,300)
following the protocols outlined by Huang et al.1. A solution of PEEK is prepared in
concentrated sulfuric acid (95-97 %) at room temperature under mechanical stirring at a
concentration ratio of 5/95 (w/v). Once complete dissolution of the PEEK is achieved, the
solution is sealed in a solution bottle and incubated in an oven at 36 °C for 15 hours. The
SPEEK is then precipitated from solution by the addition of deionised water from a Milli-Q
Advantage A10 system with a resistance of 18 Ω.cm at 25 °C. The precipitated SPEEK
powder is recovered using filtration and washed thoroughly with clean deionised water to
remove any residual acid. The recovered and washed SPEEK powder is then dried in a
vacuum oven at 50 °C for approximately 48 hours to remove any residual water.
For electrospinning, a solution of 15 wt % SPEEK is prepared in dimethylformamide on a
hotplate at 60 °C under mechanical stirring for 24 h. The SPEEK solution is loaded into a 1
CHAPTER 3. RESEARCH METHODOLOGY Grace Dolan, PhD Thesis
57
ml Hamilton syringe which is fixed with a blunted 20 G x 1½ in. stainless steel syringe
needle. The syringe is loaded onto a syringe pump and the syringe needle is connected to a
high voltage power supply (Gamma High Voltage Research, USA). A clean glass
microscope slide is mounted to a grounded collection plate using double sided tape and the
collection plate and slide are covered with aluminium foil. A hole is cut out of the centre of
the aluminium foil to expose an approximately 1 cm x 1 cm area of the underlying glass slide.
The collection plate is then positioned 17 cm away from the tip of the syringe needle. The
electrospinning is done by pumping the SPEEK solution through the syringe under a constant
flow rate of 0.12 ml/hr and an applied voltage of 20 kV to the syringe needle. The SPEEK
nanofibres are electrospun onto the substrate for a given time ranging from 1 to 5 minutes,
resulting in fibre matts with varying network densities. The SPEEK nanofibre network is then
imaged using a NeoScope JCM-5000 SEM (JEOL) and the average fibre diameter, 127±9.3
nm, is obtained directly from the SEM micrographs analysed using ImageJ software.
Electrospun Polyvinyl Alcohol (PVA) fibres
PVA polymer (molecular weight of 85-124 kg/mol -Sigma-Aldrich, Castle Hill, Australia) is
firstly dissolved in deionised water at 80 °C for 4 hours. The solution is left to stand,
unstirred, for a few minutes in order to degas before electrospinning. For the electrospinning
process, polymer solution is loaded into a 5 ml syringe and a positive electrode is clipped
onto the syringe needle with a 0.5 mm diameter. The flow rate of the PVA solution is 0.5
mm/hour, at an applied voltage of 22 KV and tip to collector distance of 13 cm. PVA solution
is electrospun horizontally onto the target. After electrospinning, the collected nanofibre mat
is dried in a vacuum oven at 60 °C for 8 hours. The morphology of PVA nanofibres is
investigated using Scanning Electron Microscope (JEOL JSM-6460LA). From SEM images,
the average diameter of 50 individual nanofibres is 163 ± 42 nm.
3.1.2 Cellulose Nano-fibres Extracted from a Plant Source
In Chapter 5, the dip-and-drag technique is also applied to cellulose nano-fibres that interact
via hydrogen bonding. Cellulose Nanofibrils (CNF) and Cellulose Nanocrystals (CNC) are
supplied by Professor Darren Martin’s research group at the Australian Institute for
Bioengineering and Nanotechnology (AIBN), The University of Queensland.
CHAPTER 3. RESEARCH METHODOLOGY Grace Dolan, PhD Thesis
58
Cellulose Nanofibres (CNF)
Cellulose nanofibres (CNF) from Triodia pungens, an Australian native grass, are produced
based on the method explained elsewhere2, 3. Briefly, water washed ground grass is
delignified using a 2 %(w/v) sodium hydroxide solution, followed by bleaching two times
with a 1 %(w/v) acidic solution of sodium chlorite. Cellulose nanofibrils are obtained by
passing a 0.3 %(w/w) dispersion of bleached pulp through the high pressure homogeniser
(Panda 2K NS1001L, GEA Niro Soavi S.p.A, Italy) at a pressure of 700 bar for two passes.
TEM images of nanofibrils are obtained using a JEOL 1011 TEM (JEOL Pty Ltd., Frenchs
Forest, Australia) at 100 KV. The dimensions of nanofibrils are measured using digital image
analysis (Image J) of several TEM images at the same magnification. Individual CNF
obtained from Triodia pungens have an average diameter of 4.5 ± 1.5 nm, and a length of
several microns.
Cellulose Nanocrystals (CNC)
Cellulose nanocrystals (CNC) of Triodia pungens are obtained by sulphuric acid hydrolysis
of bleached pulp. T.pungens bleached pulp is treated with 40 %(v/v) sulphuric acid at 45 °C
for 3 h, followed by centrifuging the hydrolysed fibre dispersion for 20 minutes at 4750 rpm.
The centrifugation step is repeated 4 times in order to remove excess acid and dissolved
extractable material. Hydrolysed fibres are dialysed against distilled water for a week, and
redispersed in deionised water using an ultrasonic probe (Q500 Sonicator, via QSonica,
Newtown, United States) at 25 % amplitude, a frequency of 20 kHz, and output energy of
500 W for 20 minutes. The diameter and length of CNC measured from TEM images are 3.5
± 0.8nm and 497 ± 106nm, respectively.
3.1.3 Bacterial Cellulose
Bacterial cellulose is considered to be a good model system for investigating cellulose-
hemicellulose interactions that are prevalent in the plant cell wall. The method for producing
cellulose hydrogels is based on the fermentation of Gluconacetobacter xylinus in liquid
media4, 5. The bacterium is aerobic and locates at the air interface. The synthesised cellulose
fibres form a self-assembled network that floats on the surface of the liquid fermentation
medium and takes the shape of the container in which it is grown. For the tribo-rheological
technique in Chapter 4, hydrogel disks are grown in a 41 mm diameter container. The
thickness of the hydrogel disks is somewhat controlled by the fermentation time. The
CHAPTER 3. RESEARCH METHODOLOGY Grace Dolan, PhD Thesis
59
hydrogels used in this thesis are measured using calipers to have an average thickness of
approximately 2.5 mm.
Gluconacetobacter xylinus (ATCC 53524 American Type Culture Collection, Manassas, VA,
U.S.A) is stored in Hestrin and Schramm (HS) agar medium at -80 °C. The composition of
HS medium is given in Table 3.1. The frozen strain is revived by incubating on HS agar
plates at 30 °C for 72 hours. A loop of bacteria is transferred to 20 mL of liquid HS medium
with 50 %w/v glucose solution. The liquid medium is adjusted to pH 5 with 0.1 M HCl,
before incubating under static conditions for a further 72 hours at 30 °C. The cellulose
network that forms on the surface of the medium contains trapped bacteria. An orbital
platform shaker (KS 260 IKA-Werke, Staufen, Germany) is used at 350 rpm for 5 minutes to
dislodge the bacteria into the liquid medium. This forms the primary inoculum, of which 1
mL is aliquoted into sterile containers (41 mm diameter) with 9 mL of HS liquid medium.
The containers are then incubated under static conditions for 72 hours at 30 °C before being
agitated on the orbital platform shaker (350 rpm, 5 mins). The samples are harvested from the
liquid medium using metal forceps, and the surface in contact with the medium is scraped
gently along the rim of the container to remove loosely attached cellulose strings. The
hydrogels undergo six rinse treatments in ice cold ultrapure water (resistivity 18.2 MΩ.cm at
25 °C from Milli-Q water purification system) under agitation. The rinsing is carried out on
the orbital platform shaker with 3x30 min and 3x10 min steps at 100 rpm, replacing the water
at each step. Finally, the hydrogels are stored in 0.02 wt% sodium azide to prevent
microbiological contamination and growth.
Table 3.1. Composition of 300 mL of Liquid HS medium
Ingredient Amount
Peptone 1.5 g
Yeast Extract 1.5 g
Na2HPO4.2H2O 1.014 g
Citric acid 0.345 g
Glucose (50 %) 12 mL
Water 288 mL
Composite hydrogels are produced by dissolving polysaccharides (XG or AX) into solution
(1 g/100 mL) at 85 °C stirring at 250 rpm overnight. The polysaccharide solution is added to
CHAPTER 3. RESEARCH METHODOLOGY Grace Dolan, PhD Thesis
60
a twice concentrated HS medium in a 1:1 ratio. Composites had a polysaccharide
incorporation of 41.2 % for the cellulose/AX composite (CAX) and 44.4 % for the
cellulose/XG (CXG). A comprehensive protocol for the production of cellulose and
composite hydrogels is provided in Appendix A.
Microarray moulds
To be compatible with the dip-and-drag technique, the dimensions of the bacterial cellulose
networks have to be much smaller than the hydrogels produced above. A
Polydimethylsiloxane (PDMS) microarray mould has been designed here to grow bacterial
cellulose micro-hydrogels within 50 µm diameter cylindrical wells, 50 µm in depth. The
design is etched onto a chromium mask which is then used for SU-8 master fabrication to
give a silicon wafer mould with an array of cylindrical pillars (50 µm diameter and height).
PDMS is cast onto the silicon wafer mould to create arrays of micro-wells that can be
inoculated with bacteria in liquid HS medium to grow cellulose micro-hydrogels. Detailed
procedures for the chromium mask, SU-8 fabrication, and PDMS casting are presented in
Appendices B – D.
To grow cellulose micro-hydrogels, the above procedure for fermenting Gluconacetobacter
xylinus is followed. The PDMS microarray is plasma treated on high for 38 seconds to make
the surface hydrophilic. The primary inoculum is pipetted onto the PDMS microarray, and
the hydrophilic nature of the surface promotes the spreading of inoculum and sedimentation
of bacteria into the individual wells. The surface of the microarray is blotted to remove
excess liquid medium, allowing micro-gels to grow as a thin layer on the surface of the
confined micro-wells. The micro-gels are harvested after 72 hrs incubation under static
conditions by washing the surface of the microarray with ice cold water. A comprehensive
protocol for producing cellulose micro-hydrogels is provided in Appendix A.
3.1.4 Pectin Solutions
The effect of pectin solutions on cellulose fibres interactions is tested using the tribo-
rheological technique in Chapter 4. Commercial pectin extracted from citrus peel with a
degree of esterification of 65 (GENU® pectin 150 USA-SAG type D slow set, CPKelco,
Atlanta, USA) was slowly added to reverse osmosis (RO) treated water with resistivity of
18.2 MΩcm (Satorius Stedim) at 0.5, 1, 2, 4 wt%. The pectin solutions are mixed for 2 hours
using an overhead stirrer, then transferred to a container and left overnight on a roller mixer.
CHAPTER 3. RESEARCH METHODOLOGY Grace Dolan, PhD Thesis
61
To measure the viscosity of the pectin solution an AR-G2 (TA instruments) with cone and
plate fixture (40 mm diameter, 2° angle) was used to perform a shear rate continuous ramp
between 1 to 1000 s-1 (5 points per decade, 30 s per point).
3.1.5 Bacterial Expansins
Bacterial expansins were supplied by Professor Daniel Cosgrove’s research group at
Pennyslvania State University (PSU), USA. Wild type bacterial expansin (EXLX1 or YOAJ)
and 3 protein variants designated WWY, RKKQ and D82, are included in this study. Details
of the mutations to each of the protein variants are discussed in Chapter 7: Section 7.1. The
method of treating bacterial cellulose substrates with expansins is explained in Chapter 7:
Section 7.2.
3.2 Measurements
3.2.1 Tribo-rheological Technique in a Rotational Rheometer
In Chapter 2: Section 2.3.2 I review the use of a rotational rheometer to measure the friction
between hydrogels. In the literature, hydrogel pairs are brought into compressive contact at a
given normal load and the friction is measured for a controlled angular velocity. The major
difference between the bacterial cellulose hydrogels studied in this thesis and the hydrogels
studied in literature is that the cellulose hydrogels have a much higher permeability and relax
to near-zero loads after compression through a poroelastic mechanism. I build upon the
measurement technique in literature to include a comprehensive in situ mechanical
characterisation of the hydrogel system, coupled with computation modelling of the hydrogel
poroelasticity and squeeze flow of the solvent at the interface during compression. The
experimental protocol for the tribo-rheological technique is detailed here. The sensitivity of
the results to certain experimental parameters is explored in Chapter 4: Section 4.2.1. The
poroelastic mechanical model is detailed in Chapter 4: Section 4.2.2, and the computational
model for simulating the squeeze flow of solvent during compression is detailed in Chapter
4: Section 4.2.3.
A Haake MARS III stress controlled rheometer (Thermo Fisher Scientific, Karlsruhe,
Germany) is used to measure the friction behaviour between pairs of bacterial cellulose
hydrogel disks (~41 mm diameter) produced according to the procedure in Section 3.1.3. All
measurements are performed using a 60 mm titanium parallel plate (customised in the
CHAPTER 3. RESEARCH METHODOLOGY Grace Dolan, PhD Thesis
62
RHEOWIN Job Manager software to have a diameter equivalent to the hydrogel) and a
titanium metal dish. Equivalent diameter circles of emery paper (P240/S85, 58 µm
roughness) are attached to the parallel geometries using double adhesive tape. The zero point
for the gap is set between the emery paper covered geometries. Pairs of hydrogels are fixed to
the emery paper, as illustrated in Figure 3.1, using cyanoacrylate adhesive (RS components,
NSW, Australia) and centred with the help of a stencil. The initial measurement gap is set at
8000 µm, that is, a distance of 8000 µm separates the two emery paper surface (not the
distance between the hydrogel surfaces). At this gap, the top geometry is below the edge of
the bottom dish but the hydrogels are not touching. The dish is then filled with water or
pectin solution to completely submerge the top geometry. The normal force is set to zero
before the top plate is lowered to a gap of 3000 µm at a constant axial ramp rate (33 µm/s),
following the previously published compression-relaxation procedure for bacterial cellulose
hydrogels6. After compression, the normal stress relaxes due to the poroelastic properties of
the samples and water transport out of the network6. The hydrogels are held at the constant
axial strain for 15 minutes to achieve an equilibrium normal stress. A small amplitude
oscillatory shear (SAOS) step at 1 Pa and 1 Hz is run for 60 seconds to determine the storage
and loss modulus (G’ and G”) of the system; these conditions are chosen to ensure that all
measurements are in the linear regime where the stress is linear with strain. After the
compression-relaxation and SAOS steps, a steady shear rate of 0.1 s-1 is applied for 2
minutes. This test procedure is illustrated in Figure 3.2. The torque is recorded and converted
to shear stress at the rim of the hydrogel using equation 3.1.
𝜏 =2𝑀
𝜋𝑅3
(3.1)
τ is the shear stress at the radius of the hydrogel, M is the torque, and R is the radius of the
hydrogel. The hydrogels are then further compressed in 500 µm steps and the above
procedure is repeated until a minimum gap of 1500 µm. The initial height (hi) of the hydrogel
pairs is taken to be 5000 µm, which is twice the average initial thickness of a single gel disk
prior to compression. The compression ratio (CR) is calculated in terms of the final height
after compression (hf) using equation 3.2.
𝐶𝑅 = ℎ𝑖 − ℎ𝑓
ℎ𝑖
(3.2)
CHAPTER 3. RESEARCH METHODOLOGY Grace Dolan, PhD Thesis
63
Figure 3.1. Bacterial cellulose hydrogels adhered to emery paper covered geometries.
Figure 3.2. Test procedure for mechanical and friction characterisation of poroelastic
hydrogels: (1) Compression at a constant axial rate to a set compressive strain, followed by a
15 minute relaxation step. (2) Small amplitude oscillatory shear of the relaxed system. (3)
Constant angular velocity step.
3.2.2 Dip-and-drag Technique in the AFM
Techniques in literature for directly measuring the adhesive force between nanofibers are
reviewed in Chapter 2: Section 2.4.3. Due to the difficulty of isolating individual bacterial
cellulose fibres, I develop a technique that can be applied directly to fibre networks. In
Chapter 5 I validate the technique using model electrospun fibres that are produced according
to Section 3.1.1. The specific details for applying the dip-and-drag technique to the model
electrospun fibre mats are provided in Chapter 5: Section 5.2.2. In Chapter 6 I use the
validated technique on bacterial cellulose and composite networks with AX and XG produced
according to Section 3.1.3. The experimental protocol for applying the dip-and-drag
technique to bacterial cellulose systems is provided here.
Cellulose micro-hydrogels grown within the confined geometries of a PDMS microarray are
shown in Figure 3.3. The gels are stained with calcofluor white and imaged in the confocal
CHAPTER 3. RESEARCH METHODOLOGY Grace Dolan, PhD Thesis
64
microscope with DAPI filter set for epifluorescence (wavelength 450-500 m). The array of
micro-hydrogels is placed face down onto a glass slide that has been plasma-treated on high
for 38 seconds. The plasma treatment removes contaminants and makes the surface
hydrophilic, which breaks the surface tension of the liquid in the micro-wells and draws to
micro-hydrogels onto the glass slide. The gels are left in contact with the substrate for
approximately 1 hour to allow this process to occur. The PDMS mould is then peeled off
leaving the micro-hydrogels deposited on the glass surface.
Figure 3.3. Cellulose micro-hydrogels grown in PDMS micro-array, stained with calcafluor
white and imaged with a confocal microscope using DAPI filter set for epifluoresence
(wavelength 450-500).
The micro-hydrogels are glued to the substrate using a JPK Nanowizard II AFM mounted on
an inverted optical microscope (JPK Instruments, Germany). The AFM is equipped with a
CellHesion® module that has a Z-piezoelectric translator range of 100 µm. First, a small
volume of 5 minute curing epoxy resin (UHU GmbH & Co. KG, Germany) (equal parts base
and curing agent) is deposited onto the substrate close to the micro-hydrogels. The droplet of
epoxy resin can be administered by hand using a fine glass rod with direction from
microscope. The AFM tip is then lowered into the epoxy resin droplet and subsequently
lowered onto the edge of the micro-gel. This is repeated for the opposite edge of the micro-
hydrogel so that it is adhered to the substrate at two points as illustrated in Figure 3.4. The
epoxy resin is given sufficient time to cure before the adhered micro-hydrogel is imaged.
The AFM was loaded with a stiff cantilever (HQ:NSC35/Cr-Au BS, Cantilever A) from
Mikromasch (Nano World AG, Germany) for imaging hydrogels in intermittent contact mode
in air. The imaging is performed at a scan rate of 2 Hz for a 60 x 60 µm scan size with 1024 x
1024 pixels. The set point and drive amplitudes are around 1 V and the drive frequency is
CHAPTER 3. RESEARCH METHODOLOGY Grace Dolan, PhD Thesis
65
Figure 3.4. An image taken after adhering micro-gels to the glass substrate, and a confocal
image showing the micro-gels attached at two points by epoxy-resin.
around 200 kHz. The image is used to identify the edge of the micro-hydrogel. The micro-
gels are then rehydrated by pipetting reverse osmosis (RO) treated water with resistivity of
18.2 MΩcm (Satorius Stedim) around the cantilever. The manipulation control function in the
AFM software is then used to trace a cantilever path over the image as shown in Figure 3.5.
The cantilever is translated laterally outwards from the edge of the micro-hydrogel in order to
pull fibres out of the network.
Figure 3.5. AFM image of glued micro-gel showing the edge of the network and a
superimposed trace of the cantilever path for lateral force measurements.
Lateral force measurements are taken with a set point vertical deflection of 300 nN and a
cantilever travel speed of 0.3 µm/s. A cantilever of high stiffness is used in order to apply a
high lateral force for separating fibre contacts. In order to hook onto the loose fibre loops that
occur at the edge of the micro-hydrogel, as seen in the inset of Figure 3.5, the AFM tip is
engaged with the substrate several microns outside of the identified edge and dragged away
CHAPTER 3. RESEARCH METHODOLOGY Grace Dolan, PhD Thesis
66
from the micro-hydrogel. Then the tip is disengaged from the surface and moved (without
touching the substrate) to the starting point of the subsequent trace which is incrementally
closer to the edge of the hydrogel. This “dip-and drag” procedure is repeated several times
until the first peaks in the lateral deflection curve are observed.
The vertical sensitivity of the cantilever is measured to be 3x10-7 N/V using the built-in
calibration manager in the JPK NanaWizard® software. For lateral calibration of the
cantilever, the Torsional Sader Method7 is used to find the torsional spring constant (5.16x10-
8 N m), and the lateral sensitivity (3.12x10-5 N/V) is calculated using a non-contact
calibration procedure8. To find the lateral sensitivity of the cantilever in water (2.35x10-5
N/V) the lateral sensitivity in air is multiplied by the ratio of the refractive index of air and
water. Analysis and interpretation of the lateral force distance curves is detailed in Chapter 5.
CHAPTER 3. RESEARCH METHODOLOGY Grace Dolan, PhD Thesis
67
References for Chapter 3
1. R. Y. M. Huang, P. H. Shao, C. M. Burns and X. Feng, Journal of Applied Polymer
Science, 2001, 82, 2651-2660.
2. N. Amiralian, P. K. Annamalai, P. Memmott and D. J. Martin, Cellulose, 2015, 22,
2483-2498.
3. N. Amiralian, P. K. Annamalai, P. Memmott, E. Taran, S. Schmidt and D. J. Martin,
RSC Adv., 2015, 5, 32124-32132.
4. E. Chanliaud and M. J. Gidley, Plant Journal, 1999, 20, 25-35.
5. D. Mikkelsen and M. J. Gidley, in Plant Cell Wall: Methods and Protocols, ed. Z. A.
Popper, Humana Press Inc, Totowa, 2011, vol. 715, pp. 197-208.
6. P. Lopez-Sanchez, M. Rincon, D. Wang, S. Brulhart, J. R. Stokes and M. J. Gidley,
Biomacromolecules, 2014, 15, 2274-2284.
7. C. P. Green, H. Lioe, J. P. Cleveland, R. Proksch, P. Mulvaney and J. E. Sader,
Review of Scientific Instruments, 2004, 75, 1988-1996.
8. K. Wagner, P. Cheng and D. Vezenov, Langmuir, 2011, 27, 4635-4644.
68
Chapter 4
Friction, lubrication and in situ
mechanics of poroelastic cellulose
hydrogels.
4.1 Introduction and Background
Naturally occurring lubrication mechanisms in biological systems have largely motivated the
study of aqueous lubrication. The systems commonly studied involve polymeric networks
and surface films with solvent present throughout the polymeric network and at the
tribological interface. Solvated interconnected polymer networks are typically referred to as a
gel, or a hydrogel in the case where the solvent is water. The unique aspect of gels is that
they have the potential to exhibit so-called poroelasticity, whereby under deformation the
mechanical response is not only a function of the elastic polymer network, but also the
movement of solvent through that network. The mechanical properties that determine
whether a full or partial solvent layer can be supported at the interface between gel surfaces
and/or gelled surface films is still not fully understood.
The most commonly used example of biological soft material where poroelastic effects are
potentially important to their tribological performance is articular cartilage. In this example,
under compressive forces, the interstitial solvent supports a significant proportion of the
normal load1. This load support from the solvent reduces the load on the polymeric matrix
that contributes to friction. For two surfaces in relative lateral motion, the coefficient of
friction is defined as the ratio of tangential force resisting the motion and an applied normal
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
69
load. Controlled normal load tests are challenging to perform for highly permeable hydrogels
because they compress under load as the interstitial solvent moves out of the network. When
compressing two poroelastic gels to a fixed gap, the normal force may relax to near-zero
values. However, provided there is contact between the gels, lateral movement under this
condition is expected to provide quantifiable information on adhesive forces between the gels
as an assessment of their tribological properties.
The primary motivation of this thesis is to elucidate potential interactions occurring between
plant cell walls that undergo wall stress relaxation during plant growth and other
deformations experienced at the cellular level. I consider that studying the physics of
hydrogel tribology is relevant because plant cell walls are essentially hydrogel composites of
cellulose fibrils within a matrix of biopolymers (e.g. hemicelluloses, pectin) and water.
During dynamic growth processes, the load-bearing cross-links in the wall structure are
biochemically loosened, which leads to wall stress relaxation that drives cell expansion2, 3.
As cells expand within the tissue structure, I consider there to be a sliding contact between
adjacent extending walls. Thus plant cell walls, like articular cartilage, require modes of
lubrication under compression (static) and sliding (dynamic) conditions, which have not
previously been explored. Since it is challenging to investigate the mechanical response using
plant cell walls directly, I use bacterial nano-fibrillar cellulose hydrogels to probe the role of
microstructure on their tribological behaviour. A mechanical study on the poroelasticity of
these bacterial nano-fibrillar cellulose hydrogels and composites with arabinoxylan (CAX)
and xyloglucan (CXG) shows that they are highly permeable and fully relax after
compression4.
The sliding friction behaviour between two hydrogels in a loaded contact is distinctly
different from non-porous solid materials. Hydrogel tribopairs have a significantly lower
friction coefficient compared to non-porous solids and they do not conform to Amonton’s
law for solid friction5. The non-linear relationship between frictional force and load has been
investigated for a variety of hydrogels using a rotational rheometer with parallel plate
geometries. Pairs of hydrogel disks are attached to the plates, and under constant compressive
load, one plate is rotated at a specific angular velocity while the torque is recorded over time6-
9. The mechanism of gel-gel friction is explained by Gong et al using a repulsion-adsorption
model10. Gong’s model states that between two repulsive surfaces, friction is determined by
the hydrodynamic lubrication of the solvent layer at the interface. For the case of two
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
70
attractive surfaces, friction is due to elastic deformation of the polymer chains that are
interacting at the interface.
Experimental tribology studies involving hydrogel substrates, and soft contacts generally,
find that the friction coefficient depends on the velocity and load condition, as well as
material properties including elastic modulus, permeability, relaxation rate, and residual force
after relaxation11-15. Therefore, in addition to chemical structure and surface properties of the
hydrogels that are explained through Gong’s repulsion-adsorption model, the mechanical
response to compression and relaxation is also likely to play a role in the tribological
behaviour. The compression-relaxation response of hydrogel materials is determined by the
interplay between the viscous behaviour of the solvent throughout the porous network and the
elastic response of the porous network.
In essence, the mechanism of stress build-up and its relaxation during unconfined
compression of gels can be summarised in three steps:
(i) lateral expansion of the matrix and interstitial fluid pressurisation,
(ii) contraction of the matrix in combination with fluid flow out of the gel due to a
pressure gradient,
(iii) normal stress relaxation of the gel due to fluid redistribution16.
The compression-relaxation behaviour of hydrogels has been modelled using transversely
isotropic linear biphasic theory17, 18. This model has been successfully applied to
experimental compression-relaxation profiles to determine material parameters such as axial
modulus, radial modulus, and permeability of articular cartilage11, 17, 19 and bacterial nano-
fibrillar cellulose hydrogels18.
Squeeze film lubrication theory has been applied to study the behaviour of a fluid film at the
interface between hydrogel surfaces when they approach each other in the normal direction.
The behaviour of the film is typically described by a modified Reynolds equation. The load
carrying capacity, film thickness, and squeeze time are predicted for hydrogel systems with
varying permeability and elastic modulus20-22. Carbone and Persson theoretically show that
for soft materials (e.g. rubber), the viscoelastic losses slow down the dewetting of an
interface. When the spreading velocity approaches zero, the interface has a dry zone in the
centre and a trapped liquid region around the rim23. This description of the interface is in
contrast to the experimentally visualised interfacial film between rubber and glass from
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
71
optical interference measurements24. Adhesive forces between the rubber and glass lead to
contact at the point where the opposing surfaces are closest. The contact area then spreads
and traps pockets of fluid at the interface, that is, wetted regions are surrounded by areas of
solid-solid contact. A decrease in lubricant viscosity resulted in the fluid pockets
disappearing at a rate inversely propositional to the viscosity. Yamamoto et al.25 visualise the
heterogeneous contact between polyacrylamide hydrogels and glass, and relate the
macroscopic contact area to the frictional stress measured for a range of sliding velocities.
Under the dynamic conditions of the friction experiments, increasing the velocity drives the
spreading process of trapped water.
In this chapter, I aim to uncover the underlying physics for the tribological contact between
poroelastic hydrogels. I use a tribo-rheological test procedure where pairs of cellulose
hydrogels are squeezed into contact in a series of compression-relaxation steps on a
rheometer. After the gels are pressed into contact and the relaxing normal force is allowed to
equilibrate, a constant rotation rate is applied to evaluate the apparent friction between
surfaces. I investigate the effect of substrate modulus and solvent viscosity on the tribological
response. Scaling relationships between the apparent static friction, substrate modulus, and
solvent viscosity are then interpreted using simulations that predict the interfacial contact area
between gels. Results presented here provide insight into the static and dynamic lubrication
modes for poroelastic hydrogels.
4.2 Experimental Section
4.2.1 Physical characterisation of hydrogel mechanics and friction
Bacterial cellulose hydrogels and composites with AX and XG are prepared according to the
method in Chapter 3: Section 3.1.3. The hydrogel mechanics and friction behaviour are
measured in a rotational rheometer and the steps are detailed in Chapter 3: Section 3.2.1. In
summary, hydrogel pairs attached to the parallel geometries are brought into compressive
contact at a given CR (refer to equation 3.2 in Chapter 3: Section 3.2.1). After the normal
force relaxes to an equilibrium value, a SAOS test is used to measure the G’ and G” of the
system. Finally the top plate is rotated at a controlled angular velocity to assess the friction
response of the hydrogel pair. These three steps are illustrated in Figure 4.1. The bottom
geometry is in fact a dish that holds either water or pectin solution, such that there is solvent
at the interface between hydrogels during the measurements.
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
72
Figure 4.1. Schematic showing the steps for characterising the mechanics and tribological
behaviour of hydrogel pairs.
Lopez-Sanchez et al.18 perform compression-relaxation steps on single bacterial cellulose
hydrogels and show that the network microstructure and mechanics are influenced by the
speed of compression. Here I test whether the friction response between a pair of cellulose
hydrogels is sensitive to the compression speed. Cellulose hydrogel pairs with water as a
solvent are compressed to a CR value of 0.6 at a compression speed of 10, 100, and 1000
m/s. The shear stress curves measured between the hydrogels at a controlled rotation rate of
1rad/s are presented in Figure 4.2. The compression-relaxation steps establish the contact
between the hydrogels prior to the rotation step. However, there is no substantial influence of
the compression speed on the shear stress measured at a constant angular velocity. The
mechanics during the compression-relaxation steps are used to predict the contact area at the
interface prior to shearing with the model described in Section 4.2.3. In this thesis, a
compression speed of 33 m/s is selected to be consistent with the previous work by Lopez-
Sanchez et al.18. This compression speed is kept constant for all future measurements to
ensure that its influence on the microstructure is controlled across all samples.
There are two potential contributions to the shear stress measured at the interface between
two hydrogels. The first is the adhesion between the hydrogel surfaces that are in contact,
which is proportional to the contact area. The second is the viscous stresses from any solvent
present at the interface between the hydrogels. The viscous stress, τv, depends on angular
rotation rate ω, based on Newton’s law of viscosity in equation 4.1, where a Newtonian fluid
with viscosity, η, is present between two parallel plates with radius R and separated by a
distance, H.
𝜏𝑣 = 𝜂𝜔𝑅
𝐻
(4.1)
According to equation 4.1, the friction stress increases with angular rotation rate. This
relationship applies to any pockets of solvent that may be present at the interface between the
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
73
Figure 4.2. Sensitivity analysis of the compression speed on the friction response of bacterial
cellulose hydrogel pairs tested in water at a CR of 0.6 and rotation rate of 1 rad/s. The curves
are representative examples of the compressions speeds, which are labelled in the legend.
hydrogels. Figure 4.3 shows the friction response at different angular rotation rates for
cellulose hydrogels in water. There is no substantial effect of the angular rotation rate on the
shear stress measured. Thus there is apparently negligible viscous stress generated by the
solvent at the interface. The measured shear stress must be dominated by the adhesion
between surfaces in contact at the interface, which is the basis of detailed analysis in the
coming sections of this chapter.
Figure 4.3. Sensitivity analysis of the rotation rate on the friction response of bacterial
cellulose hydrogel pairs in water, compressed at speed of 1000 m/s to a CR of 0.6. The
curves are representative examples of the rotation rates, which are labelled in the legend.
Shear Strain (-)
0.0 0.5 1.0 1.5 2.0
Sh
ea
r S
tre
ss (
kP
a)
0.0
0.1
0.2
0.3
10 m/s
100 m/s
1000 m/s
Shear Strain (-)
0.0 0.5 1.0 1.5 2.0
Sh
ea
r S
tre
ss (
kP
a)
0.0
0.1
0.2
0.3
0.01 rad/s0.1 rad/s1 rad/s
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
74
4.2.2 Modelling the hydrogel mechanics during compression-relaxation
The micromechanical behaviour of bacterial cellulose hydrogels has been modelled by
Lopez-Sanchez et al.18 using poroelastic theory. The hydrogel is treated as a linear
transversely isotropic material, with the plane of isotropy depicted in Figure 4.4. The material
parameters that describe the mechanics, and their description, are summarised in Table 4.1.
Figure 4.4. Representation of a cellulose hydrogel disk showing that the 1-2 plane is
transversely isotropic, that is, the material is isotropic in the radial direction. The image is
reproduced from Lopez-Sanchez et al.18.
Table 4.1. Material functions in the mechanical model.
Material Function Description
E1 Radial modulus or Young’s modulus in the 1-2 plane
ν21 Poisson’s ratio in the 1-2 plane
E3 Axial modulus or Young’s modulus perpendicular to the 1-2
plane
ν31 Poisson’s ratio perpendicular to the 1-2 plane
G31 Shear modulus perpendicular to the 1-2 plane
The results from the compression-relaxation experiment are recorded as the normal stress
measured with time. The sample is compressed for t0 seconds, followed by normal stress
relaxation at a constant axial strain. The normal stress predicted by the model during
compression and relaxation is given as equation 4.2 and 4.3, respectively18.
𝜎𝑛(𝑡) = 𝐸3𝜀0̇𝑡 + 𝐸1𝜀0̇𝑅
2
𝐶11𝑘∆3 {
1
8−∑
exp(−𝛼𝑛2𝐶11𝑘𝑡/𝑅
2)
𝛼𝑛2[𝛼𝑛2∆22 − ∆1/(1 + 𝜈21)
∞
𝑖=1
} ; 0 < 𝑡 < 𝑡0 (4.2)
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
75
𝜎𝑛(𝑡) = 𝐸3𝜀0̇𝑡0
+ 𝐸1𝜀0̇𝑅
2
𝐶11𝑘
×∆3 {∑exp(−𝛼𝑛
2𝐶11𝑘𝑡/𝑅2) − exp[−𝛼𝑛
2𝐶11𝑘(𝑡 − 𝑡0)/𝑅2]
𝛼𝑛2[𝛼𝑛2∆22 − ∆1/(1 + 𝜈21)
∞
𝑖=1
} ; 𝑡 > 𝑡0
(4.3)
𝜎𝑛(𝑡) is the normal stress at time t, 𝜀0̇ is the compression speed, R is the radius of the
hydrogel, C11 is given by equation 4.4, k is the ratio of the intrinsic permeability and fluid
viscosity. ∆1, ∆2, and ∆3 are given in equations 4.5, 4.6, and 4.7, respectively. 𝛼𝑛
corresponds to the roots of the transcendental equation 4.8, of which J1 and J0 are Bessel
functions of the first kind. The MATLAB files for solving the series of equations 4.2 to 4.8
are provided in Appendix E, and an example of the input data is given in Appendix F. The
model fit of experimental data is provided in Section 4.3.1. This model is referred to as the
‘poroelastic mechanical model’ throughout the chapter.
𝐶11 = 𝐸1(1 − 𝜈312𝐸1/𝐸3)/[(1 + 𝜈21)Δ1]
(4.4)
Δ1 = 1 − 𝜈21 − 𝜈312𝐸1/𝐸3
(4.5)
Δ2 = (1 − 𝜈312𝐸1/𝐸3)/(1 + 𝜈21)
(4.6)
Δ3 = Δ2/Δ1
(4.7)
𝐽1(𝑥) − (1 − 𝜈31
2𝐸1/𝐸31 − 𝜈21 − 2𝜈312𝐸1/𝐸3
) 𝑥𝐽0(𝑥) = 0 (4.8)
4.2.3 Simulating the interface between hydrogels during compression-
relaxation
The poroelastic cellulose hydrogels are surrounded by solvent and are brought into
compressive contact at a constant speed. Variation in the thickness h of the solvent film
during compression is predicted using the Reynolds equation, given by equation 4.9 where p
is the pressure in the solvent film and r the radial coordinate shown in Figure 4.5.
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
76
Figure 4.5. (a) Equivalent simulation system to the double hydrogel contact. The modulus of
the soft substrate must be corrected as described by Szeri26. (b) Pressure distribution on the
interstitial film at maximum compression.
1
𝑟(𝑟ℎ3
𝑑𝑝𝑓𝑖𝑙𝑚
𝑑𝑟) = 𝜋𝑟2
(4.9)
Under quasistatic conditions, the film pressure must be equilibrated by the axial stress at the
solid boundaries, 𝜎𝑔𝑒𝑙, which is in turn governed by the poroelastic constitutive equation
4.1017:
𝜎𝑔𝑒𝑙 =−𝑝𝑔𝑒𝑙𝐈 + 𝜎𝑠 (4.10)
I is the identity matrix, 𝑝𝑔𝑒𝑙 is the pressure of the fluid within the hydrogel pore space, and
𝜎𝑠is the stress tensor of the solid matrix, which is assumed to be linear elastic and
transversely isotropic. In such a case, the solid matrix is described by five parameters from
the poroelastic mechanical model above: E3, E1, 21, 31, and the permeability k.
The actual system reduces to an equivalent model system in which the top hydrogel is rigid
and the bottom substrate has an equivalent Young’s modulus26, E* = E/(1 - 2). For bacterial
cellulose, E1 is much larger than E3, and the axial force during compression is much larger
than the residual force after relaxation18, 27. Taking 31 0 and 21 0.518, 27, the equivalent
moduli are given by E21* = 1.3E21 and E31
* = E31. The current simulation assumes a perfectly
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
77
poroelastic sample. This assumption affects the transient evolution of the film thickness but
does not affect the final thickness upon relaxation, since that depends solely on the axial
modulus. In order to obtain the curves, typical values for bacterial cellulose were assumed:
E21 = 200 kPa, E31 = 10 kPa and k = 110-13 m2 18. The model also requires a pre-set initial
film thickness which we take to be 20 µm; however, order of magnitude changes in this value
do not significantly affect the equilibrium film thickness after compression-relaxation.
For purely elastic gels, equations (4.9) and (4.10) can be solved analytically. For poroelastic
gels, an accurate solution can only be reached numerically. To this end, a finite element
package Comsol™ Multiphysics is used, which enables the prediction of the normal force
and film thickness, and the sensitivity of these parameters to solvent viscosity and axial
modulus of the hydrogels. It is assumed that there is no mixing between the solvent and the
interstitial fluid in the hydrogel, either inside or outside the simulation box. This means that
solvent under pressure escapes the film boundaries and can be reabsorbed back into the film
without affecting the pectin concentration. For the simulation, a 2D axisymmetric geometry
is used with triangular elements as shown in Figure 4.5. The mesh size was refined until the
results were seen to be size independent. The simulation of the interface between poroelastic
surfaces is referred to as the ‘ComsolTM Multiphysics model’ throughout this chapter.
4.2.4 Pectin solution and viscosity measurements
The lubricating effect of solvent viscosity is tested by increasing the pectin concentration in
the solution surrounding the hydrogel pair. Pectin solutions (0.5, 1, 2, 4 wt%) are prepared
and their viscosity measured according to the procedure in Chapter 3: Section 3.1.4. The
viscosity curves are shown in Figure 4.6 below.
The binding of pectin to cellulose is largely through the neutral sugar side chains28.
Furthermore, the binding of pectin to cellulose is weak and reversible with water washing.
The crystallinity of bacterial cellulose and the assembly of cellulose fibres into a network are
not influenced by the presence of pectin in the fermentation medium28. The commercial
pectin used here is debranched and is therefore expected to act as a viscosifier and have
negligible interaction with the cellulose fibres in the hydrogel.
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
78
Figure 4.6. Viscosity of pectin solutions (0.5, 1, 2, and 4 wt%) across the range of shear rate
from 1 to 100 s-1.
4.3 Results and Discussion
4.3.1 Mechanical properties of poroelastic hydrogels
The measured mechanical responses of bacterial cellulose and composite hydrogels during
compression-relaxation steps are shown by the lines in Figure 4.7. During compression, the
top plate is lowered at a constant rate, so the time-axis is proportional to axial strain. For the
relaxation step, the axial strain is held constant over time. The SEM image inserts in Figure
4.7 show differences in network structure (images provided by Dr Patricia Lopez-Sanchez
and Dr Dongjie Wang, 2014). The curves are non-linear during compression and the normal
stress relaxes to near-zero loads as water moves out of the hydrogel; this response is
characteristic of poroelastic hydrogels. The experimental data is fitted with the poroelastic
mechanical model published by Lopez-Sanchez et al.18 and detailed in Section 4.2.2. The
experimental data and corresponding model fits are shown in Figure 4.7.
The poroelastic mechanical model assumes that the pair of hydrogels behaves as a continuous
system, that is, that the two hydrogels are in contact and no solvent is present between them.
This assumption is valid based on the results in Figure 4.3 in Section 4.2.1. The exact amount
of solvent present at the interface is investigated in Section 4.3.2, but the fit of the poroelastic
model to the experimental data in Figure 4.7 is considered to be good enough to approximate
the material properties of the hydrogels. The model does not track the data well during the
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
79
relaxation step, however the material properties are determined by the poroelastic model from
the linear slope during compression and the residual force after relaxation and at these points
the measured data and model fit are in good agreement.
The axial modulus, radial modulus, and permeability of the hydrogel are described in Table
4.1 in Section 4.2.2, and are obtained from fitting the experimental data with the poroelastic
model. The fitted values are presented in Table 4.2 and the raw data for the compression-
relaxation of cellulose, CAX, and CXG hydrogels at all CRs are included in Appendix G.
Figure 4.7 shows that for the same axial strain, the peak normal stresses on compression of
CXG and CAX are larger than cellulose hydrogels. The higher resistance to compression for
CXG and CAX is consistent with the lower permeability and higher axial modulus in Table
4.2. As the hydrogels are further compressed, water is squeezed out in the absence of lateral
expansion. This leads to an increase in cellulose concentration and changes in the network
structure. Fibre aggregation of compressed cellulose and composite hydrogels has been
shown when comparing SEM images before and after compression4, 18. The effect of fibre
aggregation is highlighted in Table 4.2 where the axial and radial moduli of all gel pairs
increase, while the permeability decreases with increasing CR. Results are not included for a
CR value of 0.4 because of the difficulty modelling the non-linearity of parameters during
this first compression step. Experimentally, the bacterial cellulose hydrogels are relatively
inhomogeneous in the axial direction. This arises because of the way the bacterial cellulose
grows on the surface of the liquid culture medium. The bacteria are aerobic and preferentially
locate at the air interface. As a result, the network is visibly denser at the air interface and
‘looser’ toward the surface that faces the liquid medium during fermentation. The denser
surfaces of the hydrogels are positioned at the interface during measurement. Upon
compression the network becomes more concentrated and the overall concentration gradient
within the hydrogel is reduced, which leads to much better agreement with the assumptions
of the poroelastic model.
During compression, there is an initial linear response up to a critical value of strain. This
linear region characterises the deformation that can be sustained without water moving out of
the network, and the system behaves elastically. The transition to non-linear behaviour occurs
when the forces applied during compression are sufficient to overcome the resistance to water
displacement, which is a function of the network permeability. In order to compare the
compression mechanics of the different hydrogels, the compression curves are normalised
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
80
a
b
c
Figure 4.7. Compression-relaxation profiles of pairs of (a) cellulose, (b) CAX, and (c) CXG
hydrogels showing experimental data (symbols) and the model fit (lines). Representative
curves are at a CR value of 0.7.
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
81
against the peak normal stress and plotted together in Figure 4.8. Non-linearity of the
cellulose stress-strain curve occurs at a much lower strain compared to CAX and CXG.
Lopez-Sanchez et al.18 report purely elastic behaviour of CXG and CAX composites at a
large compression rate (33 m/s), whereas Cellulose displayed elastic behaviour only up to
strains of ca 1%. This is in good agreement with the results in Figure 4.8 and suggests that for
the given compression rate, pairs of hydrogels behave similarly to a single hydrogel. The
highly non-linear response of cellulose hydrogels is consistent with ca. 4-fold higher
permeability compared to CAX and CXG. That is, water movement within the hydrogel has
more influence on the mechanics during the compression of Cellulose hydrogels compared to
CAX and CXG.
Table 4.2. Mechanical parameters of pairs of cellulose and composite hydrogels at different
CR
CR Axial modulus, E3
(kPa)
Radial modulus, E1
(kPa)
Permeability, k x 1010
m2
Cellulose CAX CXG Cellulose CAX CXG Cellulose CAX CXG
0.5 7.5 16 24 240 290 190 4.4 1.3 1.9
0.6 12 39 41 340 420 320 3.7 1.1 1.1
0.7 22 64 73 440 550 400 3.7 0.7 0.9
Figure 4.8. Compression profiles of cellulose, CAX, and CXG hydrogel pairs normalised for
peak normal force. Representative curves at CR 0.7.
Compressive Strain (%)
0 2 4 6 8 10 12 14 16 18
Norm
al S
tress/P
eak N
orm
al S
tress
0.0
0.2
0.4
0.6
0.8
1.0
CelluloseCAXCXG
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
82
The irreversible (within the allowed recovery time) water displacement during compression
can be quantified by looking at consecutive compressions. Cellulose, CAX, and CXG
hydrogels are compressed to 50% strain in a single step then unloaded for 30minutes before
repeating the compression. The areas between the consecutive compression curves in Figure
4.9 represent the mechanical energy that is not recovered after the first compression, and the
values are presented in Table 4.3. From these results the cellulose hydrogel has a 3-fold
higher irreversible deformation, which is attributed to greater water displacement and
concomitant fibre aggregation during compression.
The potential for fibre aggregation is further assessed by measuring the ability of a
compressed hydrogel to re-swell after the compression force is removed. This experiment
was completed by Kevin Setiadi who was a Masters student co-supervised by myself and Dr
Patricia Lopez-Sanchez. A single hydrogel (initial thickness ~2500 m) is compressed to
three different thicknesses; 2000 m, 1000 m, and 100 m. The compressed hydrogels are
then placed in a water dish. The hydrogel is removed after a given period of time using metal
forceps so as to not squeeze the material substantially. As the hydrogel is being removed it is
dragged along the edge of the dish to remove surface water as best as possible. The hydrogel
is then weighed before being returned to the water dish for another length of time. The weight
of the hydrogel is recorded over the cumulative time. The weights of the hydrogels
compressed to different thickness are compared to the uncompressed weight in Figure 4.10.
The results show that the hydrogels do not completely recover their water content after
compression, and the re-swelling ability decreases with increasing compression. This result
strongly suggests that cellulose fibres form new adhesive contacts and irreversibly change the
network microstructure, preventing complete water uptake which would be required for
restoring the original hydrogel weight after the compression force is removed. Increasing the
compression leads to an increase in the probability that cellulose fibres will adhere together.
The aim of this Chapter is to investigate the relationship between hydrogel mechanics and the
friction response. I therefore identify a key issue with characterising the mechanics using the
parameters in Table 4.2. The axial modulus is strongly influenced by the permeability of the
hydrogel. The presence of AX and XG restricts water movement during compression, which
leads to a higher axial modulus. However, only shear deformation is applied during the
friction measurement, during which fluid pressurisation is not as substantial compared to
compression. The radial modulus is expected to provide a more accurate representation of the
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
83
a
b
c
Figure 4.9. Two consecutive compressions of a single (a) Cellulose, (b) CAX, and (c) CXG
hydrogel to a CR of 0.6. The circles symbols are for the first compression and the square
symbols are second compression. The difference between the two curves for each hydrogel
shows the degree of plastic deformation.
Gap ( m)
1000 2000 3000 4000
Norm
al S
tress/P
eak N
orm
al S
tress
0.0
0.2
0.4
0.6
0.8
1.0
Gap ( m)
1000 2000 3000 4000
Norm
al S
tress/P
eak N
orm
al S
tress
0.0
0.2
0.4
0.6
0.8
1.0
Gap ( m)
1000 2000 3000 4000
Norm
al S
tress/P
eak N
orm
al S
tress
0.0
0.2
0.4
0.6
0.8
1.0
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
84
network’s mechanical response to shear deformation during the friction measurement. The
radial modulus is measured indirectly from the poroelastic model fitted to compression-
relaxation data. After the compression and normal stress relaxation steps, the hydrogel
mechanics are characterised by the G’ and G” from SAOS measurements, and the values are
presented in Table 4.4.
Table 4.3. Hysteresis areas for consecutive compressions of Cellulose, CAX, and CXG
hydrogel pairs
Hydrogel Pair Hysteresis Area (×10-4, N.m)
Cellulose 2.9
CAX 1.1
CXG 1.0
Figure 4.10. Recovery of the weight of a single bacterial cellulose hydrogel over time after
compression. The dotted line shows the original weight of an uncompressed hydrogel. The
three data sets are for cellulose hydrogels compressed to the different thicknesses that are
labelled (initial thickness ~ 2500m).
When comparing the moduli in Table 4.2 and 4.4 it is observed that the presence of AX and
XG substantially increases the axial modulus, has little effect on the radial modulus, and
substantially decreases the G’. AX and XG increase the resistance to water displacement
which is measured as a high axial modulus of the hydrogel. Bonilla et al.27 describe in detail,
the application of the poroelastic mechanical model to cellulose composite hydrogels and
states that ‘radial strains arising as a consequence of internal pressure gradients are
Time (s)
0 20 40 60 80 100 120
We
igh
t (g
)
0
1
2
3
4
5uncompressed
2000 m
1000 m
100 m
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
85
infinitesimal’. Thus the radial modulus in Table 4.2 is a measure of the mechanical response
of the cellulose fibre network to small deformations applied by fluid pressure within the
hydrogel. SAOS measurements apply a finite oscillating shear strain on the hydrogel, thus the
measured G’ reflects the hydrogel’s mechanical response to deformation on a much larger
length scale compared to the radial modulus. AX and XG do not appear to influence the
networks response to the small deformations which are measured by the radial modulus in
Table 4.2. However, the reduction in the measured G’ for CAX and CXG compared to
cellulose hydrogels in Table 4.3 suggests that AX and XG influence the networks response to
large deformation. The effect of AX and XG could be through one or a combination of the
network orientation, fibre contact density, and the nature of fibre contacts. The role of AX
and XG on fibre contacts is investigated in Chapter 6. This Chapter investigates whether any
relationships exist between the friction behaviour and the axial modulus, radial modulus, or
G’.
Table 4.4. The apparent linear viscoelastic moduli (G’, G’’ in kPa) of the hydrogels obtained
by performing SAOS immediately prior to tribological measurement.
CR Cellulose CAX CXG
G’ G” G’ G” G’ G”
0.4 5.0±0.0 0.6±0.0 1.3±0.2 0.2±0.0 0.8±0.0 0.1±0.0
0.5 11.0±0.1 1.4±0.1 2.2±0.0 0.4±0.0 1.4±0.0 0.2±0.0
0.6 23.7±0.1 2.9±0.3 3.8±0.1 0.7±0.0 2.4±0.0 0.4±0.0
0.7 73.9±1.3 11.3±1.6 8.3±0.2 1.6±0.1 5.3±0.0 0.9±0.0
4.3.2 Contact area between poroelastic hydrogels
Chapter 2: Section 2.3.1 highlights on of the key challenges in measuring the friction
between hydrogel disks in a rotational rheometer as the unknown contact area. The interface
between two hydrogels is visualised as a region of fluid in the centre, surrounded by an
annulus region of surface contact. I hypothesise that the contact area is a function of the axial
modulus of the hydrogel that is a measure of the ability of the surfaces to deform during
compression, and the viscosity of the solvent that determines its resistance to flow out of the
interface. The ComsolTM Multiphysics model described in Section 4.2.3 is used to simulate
the compression-relaxation of two poroelastic surfaces in a solvated environment (prior to
sliding). The model is used to predict the film thickness between surfaces with water as a
solvent, and the sensitivity to substrate axial modulus and solvent viscosity.
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
86
Due to the poroelasticity and the potential for substrate deformation during compression, as
well as squeeze flow of fluid out of the contact, the pressure at the interface varies with
spatial position and the compression stress or strain. The film thickness at the centre (r=0)
and edge (r=R) of the interface is plotted against time in Figure 4.11, where compression
takes place in the first 12 seconds and then the system relaxes. Across the entire interface, the
film thickness decreases with increasing compressive strain, which is proportional to time up
to 12 s. The model predicts that the film thickness between the hydrogels is initially larger at
the outer edge than in the centre. With increasing compression, the film thickness at the edge
decreases more rapidly than in the centre and becomes negative. Note, due to how the model
is formulated, film thickness values below zero merely indicate that the surfaces are in
contact and have deformed. The model predicts that a finite film thickness is maintained at
the centre of the interface, and the surfaces are in contact in the outer annular region. The
final film thickness is shown as a function of radial position together with a depiction of the
contact in Figure 4.12.
Figure 4.11. Simulated film thickness at the centre, i.e. r = 0 (filled symbols), and edge, i.e. r
= R (open symbols) of the interface against time. The hydrogels (axial modulus, E3 = 5kPa)
are compressed in water (viscosity 0.001 Pa.s) at a constant rate (33 µm/s) from 0 – 12 s
followed by relaxation at a constant CR.
The radial position when the film thickness equals zero in Figure 4.12 marks the start of
surface contact, which extends to the edge of the surface. Thus the annular surface contact
area can be determined. Figure 4.13 shows that the power law exponent is 0.3 for the
relationship between ‘contact area/total area’ and hydrogel axial modulus across the range of
Time (s)
0 10 20 30 40 50
Thic
kn
ess (
m)
-10
0
10
20
r = 0
r = R
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
87
5-30 kPa. The range of axial modulus in Figure 4.13 is of the order of the axial modulus in
Table 4.2. For hydrogels with a low axial modulus, the balance between squeeze flow forces
and substrate deformation drives substrate deformation, which leads to a larger wetted area in
the centre of the contact.
a
b
Figure 4.12. (a) The variation in film thickness with radial position in the interface at the
final time (t = 48s) from Figure 4.11. The graph shows a film of finite thickness at the centre
of the interface, and gel-gel contact beyond r = 15mm. (b) shows a depiction of the contact at
the interface between poroelastic hydrogels (not to scale).
Figure 4.13. Logarithmic plot of the fraction of the interface that is in contact, predicted from
the simulations, against the axial modulus of the hydrogels. The linear curve fit has a slope of
0.3.
Radial position (mm)
0 5 10 15 20
Film
Thic
kn
ess (
m)
0.0
0.5
1.0
1.5
2.0
Log (axial modulus)
3.5 4.0 4.5 5.0
Log
(C
on
tact
/ to
tal a
rea)
-0.3
-0.2
-0.1
0.0
0.1
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
88
The ComsolTM Multiphysics model is used to investigate the influence of solvent viscosity on
the contact area. The thickness of the solvent layer between poroelastic hydrogels during
compression is simulated for Newtonian solvents with viscosity values ranging from 0.001 to
0.5 Pas. The results from the simulations are used to predict the scaling relationship between
solvent viscosity and contact area. In Figure 4.14, the film thickness at the centre of the
interface is plotted against time for a solvent viscosity of 0.01 and 0.1 Pa.s. The important
observation is that the final film thickness, and consequently the area of the wetted region in
the centre, increases with increasing solvent viscosity. Figure 4.15 presents a logarithmic plot
of the fraction of the interface that is in contact, against the solvent viscosity. The model
shows that increasing the solvent viscosity reduces the contact area at the interface according
to a power law scaling with an exponent of -0.16.
Figure 4.14. Simulated film thickness at the centre of the interface, i.e. r = 0, against time.
The hydrogels (axial modulus, E3 = 5kPa) are compressed at a constant rate (33µm/s) from 0
– 12 s, followed by relaxation at a constant CR. The solvent viscosity (η) is labelled for each
curve.
4.3.3 Tribological response between hydrogels
The shear forces between cellulose hydrogels are measured after compressing two gels into
“contact” to different CR values and rotating the upper surface. The torque required to rotate
the upper disk at a constant velocity is converted to shear stress and plotted against shear
strain to evaluate the tribological behaviour. Figure 4.16 presents two characteristic
tribological responses; (4.16a) stick-slip and (4.16b) stiction, which are observed at the same
CR value of 0.6 for different hydrogels in water.
Time (s)
0 10 20 30 40 50
Thic
kness (
m)
0
5
10
15
20
25
30
= 0.1 Pa s
= 0.01 Pa s
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
89
Figure 4.15. Logarithmic plot of the fraction of the interface that is in contact, predicted from
the simulations, against the solvent viscosity. The linear curve fit has a slope of -0.16.
Figure 4.16. Characteristic tribological responses for hydrogels tested in water at CR 0.6.
stick-slip sliding (Cellulose) and stiction (CXG).
“Stick-slip” refers to the situation where the interface cycles between adhesive contact and
relative sliding, and is observed with a peak and a distinctive zig-zag pattern in the shear
stress-strain curve; this is observed for the cellulose hydrogels in water in Figure 4.16a.
Starting from rest, the measured shear stress is linear with strain up to a critical value,
hereafter referred to as an interfacial yield stress that I consider to be a measure of apparent
“static friction” between the hydrogels. The initial linear region at low strains is characteristic
Log (Viscosity)
-3.5 -3.0 -2.5 -2.0 -1.5 -1.0 -0.5
Log
(C
on
tact
/ to
tal a
rea)
-0.8
-0.7
-0.6
-0.5
-0.4
-0.3
-0.2
Strain (-)
0 1 2 3 4 5
Sh
ea
r S
tre
ss (
kP
a)
0.0
0.1
0.2
0.3
0.4
0.5Cellulose
CXG
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
90
of bulk elastic deformation. Due to the potential for cellulose fibre adhesion within a
network, shown by the results of consecutive compressions and re-swelling ability of
hydrogels in Figures 4.9 and 4.10, it is hypothesised that hydrogels brought into compressive
contact interact through adhesion between fibres at the surfaces. Furthermore, it is
hypothesised that fibre adhesion at the interface results in the pairs of hydrogels being
conjoint and hence behaving as a continuous bulk gel at low strains. This is supported by
comparing the mechanics of a single hydrogel and pair of hydrogels under shear deformation,
after being compressed to the same CR. The single hydrogel is attached to both rheometer
plates, whereas one hydrogel surface is attached to each rheometer plate for the hydrogel
pairs. The similarity of the initial linear slope of the stress-strain curves in Figure 4.17
indicates that the system’s shear modulus is the same, regardless of whether a single or pair
of hydrogels is being deformed.
Figure 4.17. Stress-strain curve of single Cellulose hydrogel glued to both rheometer plates
(line) and a pair of Cellulose hydrogels with one surface of each gel glued to a rheometer
plate (symbols). Samples are compressed to a CR of 0.5.
Hydrogel pairs can be treated as a continuous system up until the yield point. A yield point is
not observed for the single hydrogel within the range of stress tested. Thus the interfacial
yield stress is characteristic of the interface rather than failure of the adhesion of the
hydrogels to the rheometer plates, or failure of the hydrogels themselves. During the stick
cycle, the surfaces are adhered together and the shear stress increases with strain as the gels
undergo shear deformation. When the applied shear stress is sufficient to overcome the static
friction between the gels, the “slip” cycle is initiated and the surfaces move relative to each
Shear Strain (-)
0.001 0.01 0.1 1
Shear
Str
ess (
kP
a)
0.01
0.1
1
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
91
other. The rotating surface undergoes rapid acceleration-deceleration to try to maintain the set
rotation rate. During the slip cycle, the shear stress decreases with strain until it reaches the
kinetic friction stress and the surfaces adhere again.
Figure 4.16b shows the shear response for CXG composite hydrogels with water; the
measured shear stress increases linearly with strain up to the interfacial yield stress, which is
followed by a decrease to a relatively constant shear stress value required to maintain the
controlled rotation rate. This is characteristic of “stiction”, which is the situation where the
stress increases until a static friction is overcome, followed by a decrease to a steady “kinetic
friction” value29. A very weak dependence of shear stress on strain is apparent beyond the
peak stress. I consider that during shearing the hydrogel surfaces are prevented from coming
into adhesive contact again, which may be governed by the hydrodynamic drag of the solvent
phase or the restructuring of the hydrogel network at the interface between the two surfaces.
These two phenomena (stick-slip and stiction) have been observed for model hydrogel
systems and articular cartilage surfaces in sliding contacts at a range of length scales;
including those obtained using a similar tribo-rheological technique6, colloidal probe AFM6,
13, micro-tribometer12, Surface Force Apparatus (SFA)29 and a SFA-type experiment where
two glass tubes coated with a gel layer are in cross-cylinder orientation30. The fact that stick-
slip and stiction responses are observed between hydrogel surfaces in a range of experimental
set ups suggests that these friction phenomena are a material/system property rather than an
instrumental artefact, and that the measured interfacial yield stress is related to adhesive
contact forces between the hydrogel surfaces. The stick-slip and stiction behaviours will be
explored later in Section 4.3.7.
Sections 4.3.4 to 4.3.6 focus on the interfacial yield stress as a measure of the static friction
between hydrogel surfaces. The hydrogel tribology is considered in terms of the contribution
of any solvent layer present between the surfaces and/or interactions occurring at the
interface, which may also be a function of the mechanical properties of the substrates.
4.3.4 Influence of substrate mechanics on interfacial friction
Cellulose, CAX, and CXG hydrogel pairs measured at all CRs with water as the solvent
display either stick-slip or stiction behaviour; these are considered to be at least partially
adhesive contacts. Representative friction curves for the two phenomena are shown in Figure
4.16a and 4.16b. The friction curves for all samples (cellulose, CAX, CXG) at all CRs are
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
92
included in Appendix H. The accuracy of the recorded friction response is quantified by
repeating the controlled rotation rate step at a given CR three consecutive times and
comparing the interfacial yield stress, which is found to have an associated error of 16%. The
variation between different hydrogel pairs is quantified by measuring the interfacial yield
stress at the same CR for 3 different pairs of cellulose hydrogels and is found to be 14%.
The mechanical properties of bacterial cellulose hydrogels are shown to be time- and strain-
dependent due to poroelastic effects. A key objective of this study is to determine which
material properties determine the friction response between hydrogels. Relationships between
the interfacial yield stress and the axial modulus or radial modulus from the poroelastic
model in Table 4.2 are presented in Figure 4.18. There is a positive correlation between
interfacial yield stress and both the radial and axial modulus. From both graphs in Figure
4.18, AX and XG appear to reduce the interfacial yield stress compared to cellulose
hydrogels with equivalent axial or radial modulus. AX or XG either change the interaction
between hydrogel surfaces by reducing adhesion, or the axial and radial moduli do not
capture the mechanics of the hydrogel that influence the contact area at the interface.
In Figure 4.18, the data points for the composite hydrogels (CAX and CXG) are shifted
towards the right hand side of the x-axis, relative to the cellulose hydrogels. This could
suggest that the axial and radial modulus from poroelastic mechanical model overestimate the
material property that more generally determines the friction response of poroelastic
hydrogels. Water movement is more restricted in CAX and CXG compared to cellulose,
resulting in an increased modulus measured during compression. After compression the
systems relax to an equilibrium normal stress prior to the interfacial yield stress being
measured. Material characterisation of the fully relaxed networks is achieved with SAOS, and
the G’ of the hydrogel pairs from Table 4.4 is plotted against the interfacial yield stress in
Figure 4.19.
From Figure 4.19, the tribological response of the poroelastic hydrogels has a strong
dependency on the substrate’s G’, with a power law exponent of 0.8. The generality of the
relationship for all hydrogel compositions (Cellulose, CAX, and CXG) and compression
ratios confirms that the G’ measured after normal stress relaxation characterises the
mechanics of the system that determine tribological behaviour. Very few studies in literature
measure the G’ of hydrogels, instead relating the elastic modulus during compression to the
friction response. For the cellulose hydrogels used in this study, their high permeability
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
93
means that the normal stress relaxes to near-zero values when the system is at a constant axial
strain. Thus the system in not under load as it is being sheared.
a b
Figure 4.18. (a) Axial and (b) Radial modulus from biphasic modelling of compression-
relaxation data plotted with the interfacial yield stress. Dashed lines are the curve fits of
cellulose data with a power law exponent of 1.5 and 3.3 for axial and radial modulus,
respectively.
The CXG data in Figure 4.19 appears to have a slightly lower interfacial yield stress
compared to the CAX hydrogels at the same G’. This suggests that XG may have an effect on
the adhesive interactions between the hydrogels. However, we cannot assume that the contact
area between the gels is governed by their diameter because of the presence of a fluid film
during the compression. To assess the inherent adhesiveness between the gels, it is thus
necessary to deconvolute the influence on the interfacial yield stress from surface contact
area and substrate modulus. The ComsolTM Multiphysics model predicts the contact area for a
given axial modulus of the hydrogel in Figure 4.13. In Figure 4.20, the G’ from Table 4.4 is
plotted against the axial modulus from Table 4.2. The two moduli are proportional, with a
different proportionality constant for cellulose, CAX, and CXG. This means that the scaling
relationship in Figure 4.13 can be used to factor out the effect of contact area on the measured
friction response for hydrogels with different G’. The ComsolTM Multiphysics model also
shows a scaling relationship between solvent viscosity and the contact area. In the next
section I experimentally investigate the influence of solvent viscosity on the interfacial yield
stress between cellulose hydrogels.
Axial Modulus (kPa)
1 10 100
Inte
rfacia
l Y
ield
Str
ess (
kP
a)
0.001
0.01
0.1
1
10CelluloseCAXCXG
Radial Modulus (kPa)
100 1000
Inte
rfa
cia
l Y
ield
Str
ess (
kP
a)
0.001
0.01
0.1
1
10CelluloseCAXCXG
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
94
Figure 4.19. Logarithmic plot of the interfacial yield stress against G’ of cellulose and
composite hydrogels in water. The linear curve fit has a slope of 0.8.
Figure 4.20. Linear relationships between G’ and axial modulus, showing different
proportionality constants for Cellulose, CAX, and CXG.
4.3.5 Influence of solvent viscosity on interfacial friction
The viscosity of the solvent is varied experimentally using pectin in water. The pectin
solutions included in this study are Newtonian (within the measured shear rates) with a
viscosity range from 0.001 to 0.5 Pa.s and all the friction curves are provided in Appendix H.
Figure 4.21 presents the measured tribological response between cellulose hydrogels at a
Log(G')
2 3 4 5
Log(I
nte
rfacia
l Y
ield
Str
ess)
1
2
3
4CelluloseCAX CXG
Axial modulus (kPa)
0 10 20 30 40 50 60 70 80
G' (
kP
a)
0
20
40
60
80CelluloseCAX CXG
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
95
representative CR value of 0.5 with varying solvent viscosity. Except for the system with the
highest viscosity, stick-slip behaviour is prevalent, consistent with intimate contact between
the surfaces and high adhesion. At a critical viscosity there is a transition in tribological
response marked by a stress-strain curve that initially shows a stiction response which
evolves into stick-slip sliding at large strains. Further increasing the viscosity of the solvent
leads to stiction behaviour, where the shear stress is maintained at a roughly constant value at
high strains. In this case, I propose that the viscosity increases the interfacial separation and
leads to only a small region of contact at the interface.
Figure 4.21. Tribological response of pairs of cellulose hydrogels at CR 0.5 with solvent
viscosity in the range of 0.001 to 0.5 Pa.s.
A logarithmic plot of the interfacial yield stress against the solvent viscosity for cellulose
hydrogel pairs at all compression ratios is shown in Figure 4.22. The data set for each CR is
fitted with a power law model with set exponent of -0.16, which is the predicted exponent for
the relationship between solvent viscosity and contact area in Figure 4.15. Remarkably, this
exponent is entirely compatible with the experimental results for CR values of 0.5 to 0.7, and
suggests that the dependence of the measured interfacial yield stress on viscosity arises
because of the varying degree of contact between the hydrogels. The predicted power law
model does not fit the measured data set with a CR of 0.4.
At this CR value, the substrates have the lowest modulus and the smallest contact area. I thus
suggest this higher exponent arises because of a greater potential for viscous stresses to
influence the tribology, beyond the influence on contact area during squeeze flow. That is,
equation 4.1 in Section 4.2.1 has a non-negligible contribution to the measured shear stress. It
Strain (-)
0 1 2 3 4 5
Sh
ea
r S
tre
ss (
kP
a)
0.0
0.1
0.2
0.3
0.4
0.5
0.6 = 0.001 Pa.s
= 0.5 Pa.s
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
96
is worth reinforcing here that there is a higher degree of uncertainty in the fitting of the
poroelastic mechanical model to the experimental data for hydrogels at CR 0.4, which is
explain in Section 4.3.2. Furthermore, the assumption of the poroelastic mechanical model is
that the pairs of hydrogels behave as a continuous material, which is no longer valid if there
is a contribution of viscous stresses from the solvent. The results for CR 0.4 will be left out of
all further analysis.
Figure 4.22. Logarithmic plot of the interfacial yield stress against the lubricant viscosity for
cellulose hydrogels. The different symbols represent different CR, as listed in the legend. The
data set for a given CR is fitted with a slope of -0.16.
4.3.6 Interfacial friction at the true contact area
The findings from the ComsolTM Multiphysics model illustrate how the interfacial contact
between poroelastic hydrogels is modulated through the substrate mechanics and solvent
viscosity. The scaling relationships between the area of contact and substrate modulus or
solvent viscosity in Figures 4.13 and 4.15 respectively, provide an approach to factor out any
differences in the measured friction response that are due to differences in contact area.
Firstly, the measured G’ (G’m) is corrected for contact area according to equation 4.12. G’c is
the corrected G’, At is the total area, and Ac is the area of contact.
𝐺′𝑐 =𝐺′𝑚𝐴𝑡𝐴𝑐
(4.12)
Log(Viscosity)
-3 -2 -1 0
Log(I
nte
rfacia
l Y
ield
Str
ess)
0
1
2
3
4
0.40.50.60.7
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
97
From Figure 4.13, the area of contact as a fraction of the total area is related to G’ according
to equation 4.13.
𝐴𝑐𝐴𝑡
= 0.037𝐺′𝑐0.32
(4.13)
Combing equation 4.12 and 4.13 gives equation 4.14.
𝐺′𝑐 = (𝐺′𝑚0.037
)
11.32
(4.14)
Then the measured interfacial yield stress (τm) is corrected for contact area according to
equation 4.15. τc is the corrected interfacial yield stress and η is the solvent viscosity. The
exponents are taken from the relationships in Figure 4.13 and 4.15.
𝜏𝑐 =𝜏𝑚𝜂
0.16
𝐺′𝑐0.3
(4.15)
τc is plotted against G’c in Figure 4.23 for all hydrogels, solvent viscosities, and all
compressions except for CR = 0.4 which did not fit the model in Figure 4.22. Applying the
corrections in equation 4.14 and 4.15 accounts for the influence of viscosity on the contact
area during the compression process prior to shearing. The data for different pectin solutions
collapse onto the power law relationship that is fitted to the cellulose and CAX hydrogels in
water in Figure 4.23. The power law regression has an exponent of 2/3. The CXG data does
not appear to fit the regression well and reasons for this will be discussed separately.
I postulate that the dependence of the corrected interfacial yield stress on the substrate
modulus is due to increased adhesion between cellulose fibres at the contact area. In Section
4.3.1 I show that cellulose fibres within a network have the potential to adhere together
during compression, and the formation of new cellulose contacts in the network increases
with increasing compression. Therefore, when pairs of hydrogels are brought into
compressive contact, it is expected that cellulose fibres from opposing surfaces have the
potential to adhere together. The G’ of the hydrogel increases with CR because water is
squeezed out of the network and the solids concentration increases. Thus with increasing
compression, which translates to increasing G’, there is a higher probability of adhesive
contacts forming at the area of contact.
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
98
Figure 4.23. The corrected interfacial shear stress (τc) plotted against the corrected G’ (G’c)
for all hydrogels and all solvent viscosities. Symbols with a white cross inside represent the
cellulose hydrogels tested in water; all other black symbols are in pectin solutions with
different concentration. The grey circles are CAX and the open circles are CXG hydrogels in
water at all compression ratios. The solid line is a linear fit of the Cellulose and CAX
hydrogels in water, with a slope of 0.67.
Lopez-Sanchez et al.18 show a positive relationship between G’ and concentration for
bacterial cellulose hydrogels, their raw data is included in Appendix I and is used to predict
the cellulose concentration of hydrogel samples in this study from their G’. I interpolate
between the data points in Appendix I to find the cellulose concentration for the G’c of my
systems; values of G’c outside the range of the provided data are omitted. The τc is plotted
against cellulose concentration for cellulose, CAX and CXG hydrogels in Figure 4.24. For
cellulose and CAX, the τc scales linearly with cellulose concentration. This is consistent with
the proposed mechanism, where increasing the number of fibres at the interface
proportionally increases the adhesion between the surfaces.
Revisiting Table 2.2 from Chapter 2, a positive correlation between friction and hydrogel
modulus is observed for proprietary hydrogels 11, hydrogels with different crosslinking
density 14, and different gel states (swollen/collapsed) 6. The hydrogels tested in each study
have varying microstructure, which may lead to a different concentration of polymer at the
interface driving adhesive interactions. Inverse relationships between friction and hydrogel
modulus were observed with changing hydrogel composition 15 and temperature 8. For
example, increasing the content of a polymer that is rigid and not particularly adhesive could
Log (G'c)
3.0 3.5 4.0 4.5 5.0
Log
c
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4Cellulose, CR=0.5Cellulose, CR=0.6Cellulose, CR=0.7CAX, all CRCXG, all CRCellulose, CR=0.5, waterCellulose, CR=0.5, waterCellulose, CR=0.5, water
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
99
reduce the friction and simultaneously increase the hydrogel modulus. The temperature will
affect the kinetics of the polymers at the interface, and thus the rate at which they can form
adhesive bridges between the hydrogel surfaces that determine friction. Thus, the mechanism
I propose for the relationship in Figure 4.23 is consistent with the relationships observed for
other hydrogel studies in literature.
Figure 4.24. Interfacial yield stress versus the cellulose concentration calculated from the
measured G'. The power law regression fitted to the Cellulose and CAX data has an exponent
of 1.
From Figure 4.24, XG reduces the τc compared to cellulose and CAX hydrogels for
equivalent cellulose concentration. The presence of XG presumably reduces the adhesion
between cellulose fibres at the interface, whereas AX has no influence. An alternative
explanation is that the XG changes the microstructure in a manner that reduces the presence
of cellulose fibres at the interface, or orients the fibres such that they are less likely to come
into contact with another fibre from the opposing surface. The effect of AX and XG on
cellulose fibre-fibre adhesion will be measured directly in Chapter 6 and all further
discussion around the mechanism of interaction between cellulose, CAX and CXG can be
found there.
4.3.7 Stick-slip and stiction behaviour
The two characteristic tribological responses observed between hydrogels in this study are
introduced in Section 4.3.3 as stick-slip and stiction. Representative shear stress curves for
Cellulose Concentration (%)
1
c (
kP
a)
0.001
0.01
CelluloseCAXCXG
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
100
these two phenomena are presented in Figure 4.25. Cellulose and CAX hydrogels across the
range of CR in water show a relatively consistent zig zag pattern in the shear stress after the
interfacial yield stress (τm), indicative of stick-slip sliding. For CXG hydrogels across the
range of CR in water, and cellulose hydrogels at low CR and high solvent viscosity, stiction
behaviour is observed where the measured shear stress peaks at τm and then decreases to a
lower equilibrium value τk.
a b
Figure 4.25. Characteristic friction behaviours during the constant rotation rate step showing
static (m) and kinetic (k) shear stresses: (a) stick-slip sliding for Cellulose/water at CR 0.6,
and (b) stiction behaviour for CXG/water at CR 0.6.
The terms characterising stick-slip behaviour are labelled in Figure 4.26. Prior to τm the
surfaces are adhered together and the initial slope of the shear stress-strain curve is a measure
of the bulk shear modulus, G, of the system. If the surfaces are adhered together during
subsequent stick cycles, the slope during stick, kst, should also be a measure of the system’s
shear modulus. In Figure 4.27a, time is proportional to strain and the G is measured at low
strain values for comparison to the mechanics of the stick cycle, where similarly low strains
are applied. In Figure 4.27b, kst is plotted against G with a line fit of kst = G, which shows
that the two moduli are in fact equal for comparable deformation strains.
Klein31 investigated the frictional dissipation in stick-slip sliding between two surfaces,
where the top surface is subject to a lateral force via the attached spring being pulled at a
constant velocity. During the stick cycle the spring extends and applies and increasing force
on the surface until the adhesion is overcome and the top surface slips past the bottom one.
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
101
Klein31 highlights that the elastic energy stored in the spring during the stick cycle must be
dissipated during the slip cycle. This is true of a system that is controlled by a passive spring.
However, for my friction experiments the rheometer uses embedded feedback equations to
actively control the angular velocity at the set point value.
Figure 4.26. Characterisation of stick-slip behaviour showing the elastic modulus, kst, during
the stick cycle, the slope of the slip cycle, ksl, the energy of the stick and slip cycles, Est and
Esl, the static friction stress, τm, kinetic friction stress, τk, and the length of the stick, γst, and
slip, γsl, cycles.
a b
Figure 4.27. (a) Shear stress over time for constant rotation (CAX at CR 0.7 in water); the
shear modulus, G, is taken as the slope of the linear region at low strains and kst is the slope
of the stick cycle. The standard deviation of the kst measured for each peak in a given shear
stress-strain curve is approximately 15%. (b) Scatterplot of the slope of kst against G with a
line fit of kst = G. Data for cellulose and CAX hydrogels at all CR in water are included.
G (Pa/s)
10
kst
(Pa
/s)
10
CelluloseCAX kst = G
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
102
For the cellulose hydrogels in the rotational rheometer, when the applied shear stress is
greater than τm, the slip cycle is initiated and the rotating surface rapidly accelerates. This is
followed by a deceleration and corresponding decrease in the applied shear stress as the
systems adjusts to the set rotation rate. After deceleration the surfaces stick again because the
applied stress is now less than τm. τk is expected to be the point at which the top surface has
decelerated enough to allow the fibres and the interface to adhere again. This acceleration-
deceleration during the slip cycle can be observed in the measured angular velocity during the
friction test in Figure 4.28. The sharp peaks in angular velocity are aligned to the slip cycle,
where the surface accelerates along the first half and decelerates along the second half of the
slip length in Figure 4.28. The friction curves with corresponding angular velocity for
hydrogel pairs (cellulose, CAX, CXG) at all CRs are included in Appendix J.
Figure 4.28. Shear stress () and angular velocity () measured over time during constant
rotation rate step for cellulose hydrogels at a CR of 0.6 in water. Inset shows a close up of a
single angular velocity peak (m to show how it aligns to the slip cycle.
Typically the angular velocity at the beginning and end of the slip cycle is comparable, which
indicates that the slip length is determined by the ability of the instrument to bring the
rotation rate back to the set point value. The magnitude of the angular velocity during the
acceleration-deceleration is expected to be controlled by the torsion that was being applied at
the point of slipping, and the feedback system of the instrument for correcting the rotation
rate which is constant for all samples. From Figure 4.25a, τm is the measured shear stress
applied at the point of slipping. As expected, the peak angular velocity (m) during the slip
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
103
cycle is shown to increase with τm in Figure 4.29. The m then controls the slope of the slip
cycle, whereby the time taken to correct the angular velocity scales with the magnitude of the
increase in angular velocity above the set point value. The slope of the slip cycle, ksl, is
shown to increase with m in Figure 4.30. Thus the stick-slip response is due to the adhesive
nature of the interface that prevents the rheometer from being able to accurately control the
set point rotation rate.
It is anticipated that reducing the adhesion or contact area between the hydrogel surfaces will
eliminate the stick-slip response. The findings from the ComsolTM Multiphysics model in
Section 4.3.5 show that the contact area at the interface is reduced with decreasing G’ and
increasing solvent viscosity. A map of the G’ and solvent viscosity for the cellulose hydrogel
pairs across the range of CR and pectin concentrations is presented in Figure 4.31, with the
tribological behaviour labelled as either stiction or stick-slip. The clustering of the different
responses in Figure 4.31 follows the contact area model of adhesion. Stick-slip is observed
for a large contact area, whereas stiction behaviour is observed when the contact area at the
interface is low. The presence of a viscous layer at the interface after the τm is overcome is
proposed to prevent the cellulose fibres from coming into contact again. Thus the set rotation
rate can be maintained at a constant stress value τk.
Figure 4.29. The peak in angular velocity (m during the slip cycle over the torque applied
to the rotating surface at the point of slipping (m).
From Figure 4.24 in Section 4.3.6, the adhesion between surfaces is reduced due to the effect
of XG on the interaction between cellulose fibres at the interface. The stiction curve and
m (kPa)
0.01 0.1 1
m (
rad
/s)
0.01
0.1
1
CelluloseCAX
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
104
corresponding angular velocity for CXG at CR = 0.7 is presented in Figure 4.32. The angular
velocity in Figure 4.32 confirms that stiction occurs when the adhesion between surfaces is
sufficiently reduced such that the rheometer is able to maintain the set point rotation rate by
applying a constant stress τm.
Figure 4.30. The slope of the slip cycle, ksl, plotted against the peak in angular velocity (m)
during the slip cycle. The standard deviation of the ksl measured for each peak in a given
shear stress-strain curve is approximately 15%.
Figure 4.31. Scatterplot of the G’ against viscosity for the different sliding behaviours
labelled in the legend.
m (rad/s)
0.0 0.1 0.2 0.3 0.4
ksl (P
a/s
)
-140
-120
-100
-80
-60
-40
-20CelluloseCAX
Viscosity (Pa.s)
0.0001 0.001 0.01 0.1 1
G' (
kP
a)
1
10
100
StictionStick-Slip
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
105
Figure 4.32. Shear stress and angular velocity measured over time for a set rotation rate. This
curve is for CXG at CR = 0.7 but is representative of all friction curves that display a stiction
response.
4.4 Concluding remarks
Tribological interactions occurring between pairs of poroelastic cellulose hydrogels are
determined using a torsionally driven parallel plate rheometer. After compressing cellulose
gels into contact, the measured shear stress increases with strain to an interfacial yield stress,
which is followed by either a stick-slip pattern or a decrease in stress to a steady value to
indicate sliding friction between the gels. The measured interfacial yield stress is found to be
dependent on the hydrogel modulus and the viscosity of the fluid medium. However, this
observation doesn’t consider the influence of the true contact area between the gels. To
overcome this issue with the parallel plate geometry, compression-relaxation and SAOS are
performed prior to torsional sliding. When the results are combined with a computational
model that utilises a poroelastic model for the compressible substrates in association with
squeeze flow of Newtonian fluid between the gels, the area of contact between the gels is
predicted. Strong evidence is found for the presence of contact between the cellulose
hydrogels in an annular region at the outer edge of the disks, and an inner region of entrapped
solvent at the centre of the interface; in particular, the area of contact between the gels
following compression reduces with increasing solvent viscosity. This is in agreement with
work by Yamamoto et al.25 using a rotational rheometer to measure the friction between
hydrogel surfaces; direct visualisation of the contact reveals a region of trapped water in the
Time (s)
0 10 20 30 40
Shear
Str
ess (
, kP
a)
0.00
0.05
0.10
0.15
Angula
r velo
city (
, ra
d/s
)
0.00
0.01
0.02
0.03
0.04
0.05
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
106
centre surrounded by an area of surface contact. The authors are able to relate the measured
friction to the true contact area at the interface. In the present study, when factoring in true
contact area, the interfacial yield stress scales with the storage modulus of the hydrogels with
a power law exponent of 0.67. The scaling of interfacial yield stress with modulus is
associated with its direct relationship to the cellulose fibre concentration. I postulate that the
interfacial yield stress at the true contact area is an indicator of the static friction between
gels, and that it is a function of contact forces between cellulose fibres in opposing surfaces.
With the approach presented in this chapter, the surface interactions between hydrogel pairs
with different composition can be measured. The presence of xyloglucan is shown to reduce
the static friction between hydrogels, whereas arabinoxylan does not. The cellulose
hydrogels studied have a water-filled fibrous network that is akin to the microstructure of
plant cell walls. Thus results provided here give insights into the potential role of wall
components on cell-cell friction.
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
107
References for Chapter 4
1. E. D. Bonnevie, V. J. Baro, L. Wang and D. L. Burris, Journal of Biomechanics,
2012, 45, 1036-1041.
2. D. J. Cosgrove, Int. J. Plant Sci., 1993, 154, 10-21.
3. R. Yamamoto, J. Plant Res., 1996, 109, 75-84.
4. P. Lopez-Sanchez, J. Cersosimo, D. Wang, B. Flanagan, J. R. Stokes and M. J.
Gidley, PLoS ONE, 2015, 10.
5. J. P. Gong and Y. Osada, Prog. Polym. Sci., 2002, 27, 3-38.
6. D. P. Chang, J. E. Dolbow and S. Zauscher, Langmuir, 2007, 23, 250-257.
7. J. P. Gong, G. Kagata and Y. Osada, Journal of Physical Chemistry B, 1999, 103,
6007-6014.
8. G. Kagata, J. P. Gong and Y. Osada, Journal of Physical Chemistry B, 2002, 106,
4596-4601.
9. T. Kurokawa, J. P. Gong and Y. Osada, Macromolecules, 2002, 35, 8161-8166.
10. J. P. Gong, Soft Matter, 2006, 2, 544-552.
11. D. Baykal, R. J. Underwood, K. Mansmann, M. Marcolongo and S. M. Kurtz, Journal
of the Mechanical Behavior of Biomedical Materials, 2013, 28, 263-273.
12. K. Chen, D. Zhang, Z. Dai, S. Wang and S. Ge, Journal of Bionic Engineering, 2014,
11, 378-388.
13. A. C. Dunn, J. M. Uruena, Y. Huo, S. S. Perry, T. E. Angelini and W. G. Sawyer,
Tribology Letters, 2013, 49, 371-378.
14. A. Kozbial and L. Li, Materials Science & Engineering C-Materials for Biological
Applications, 2014, 36, 173-179.
15. D. Zhang, Y. Shen and S. Ge, Science in China Series E-Technological Sciences,
2009, 52, 2474-2480.
16. C. G. Armstrong, W. M. Lai and V. C. Mow, Journal of Biomechanical Engineering-
Transactions of the Asme, 1984, 106, 165-173.
17. B. Cohen, W. M. Lai and V. C. Mow, Journal of Biomechanical Engineering-
Transactions of the Asme, 1998, 120, 491-496.
18. P. Lopez-Sanchez, M. Rincon, D. Wang, S. Brulhart, J. R. Stokes and M. J. Gidley,
Biomacromolecules, 2014, 15, 2274-2284.
19. M. A. Soltz and G. A. Ateshian, Journal of Biomechanical Engineering-Transactions
of the Asme, 2000, 122, 576-586.
CHAPTER 4. RESULTS Grace Dolan, PhD Thesis
108
20. N. M. Bujurke and R. B. Kudenatti, Appl. Math. Comput., 2006, 174, 1181-1195.
21. M. Yousfi, B. Bou-Said and J. Tichy, Lubr. Sci., 2015, 27, 505-522.
22. N. M. Bujurke, R. B. Kudenatti and V. B. Awati, Math. Biosci., 2007, 209, 76-89.
23. G. Carbone and B. N. J. Persson, Journal of Chemical Physics, 2004, 121, 2246-2252.
24. A. D. Roberts, Journal of Physics D-Applied Physics, 1971, 4, 423-&.
25. T. Yamamoto, T. Kurokawa, J. Ahmed, G. Kamita, S. Yashima, Y. Furukawa, Y. Ota,
H. Furukawa and J. P. Gong, Soft Matter, 2014, 10, 5589-5596.
26. A. Z. Szeri, Fluid Film Lubrication: Theory and Design, Cambridge University Press,
Cambridge, United Kingdom, 1998.
27. M. R. Bonilla, P. Lopez-Sanchez, M. J. Gidley and J. R. Stokes, Acta Biomater.,
2016, 29, 149-160.
28. D. H. Lin, P. Lopez-Sanchez and M. J. Gidley, Food Hydrocolloids, 2016, 52, 57-68.
29. D. W. Lee, X. Banquy and J. N. Israelachvili, Proceedings of the National Academy
of Sciences of the United States of America, 2013, 110, E567-E574.
30. A. Suzuki, R. Ishii, Y. Yamakami and K. Nakano, Colloid and Polymer Science,
2011, 289, 561-568.
31. J. Klein, Physical Review Letters, 2007, 98.
32. J. Stiernstedt, H. Brumer, III, Q. Zhou, T. T. Teeri and M. W. Rutland,
Biomacromolecules, 2006, 7, 2147-2153.
109
Chapter 5
Method development for measuring
the adhesive forces between
individual nano-fibres
5.1 Introduction and Background
Inter-fibre adhesion is a key factor in the functionality of naturally occurring fibrous
assemblies, which includes plant cell walls. The ubiquity of fibrous structures in nature calls
for the development of techniques enabling the direct measurements of adhesion between
nanofibres. These techniques are set to play an important role in biomimetic design since
fibre-fibre interactions are of fundamental importance to the overall network mechanics and
material performance.
The mechanical properties of a range of fibrous materials have been studied using
compression, uni- and bi-axial tensile testing and small-amplitude oscillatory shear1-12. The
results from these studies indicate that the mechanics of random fibre networks is defined by
the intrinsic mechanical properties of nanofibres3, 7, surface interactions between fibres1, 9, 12,
the network microstructure1, 4, 5, 9-12, and number and nature of entanglements and/or cross-
links1, 4, 5, 7-12.
The development of structural models of fibre networks provides predictive capabilities for
design and evaluation, as well as enhancing understanding of the underlying principles
CHAPTER 5. RESULTS Grace Dolan, PhD Thesis
110
controlling natural systems. The common approach to account for the adhesive potential is to
simply treat contacts between fibres as rigid junctions13-16, which is only appropriate when
fibre interactions are very strong. To estimate the yield strain of a fibre networks, that is the
strain corresponding to the limits for linear and elastic deformations, Chatterjee17 applies an
energy penalty for breaking the fibre contact points that equals the stored elastic energy of
deformation. However, a reliable model for any particular system requires accurate
knowledge of the adhesive potential between fibres at network junctions, which requires
validation through experiments.
Current experimental approaches for direct measurement of fibre adhesion for sub-micron
(>100nm) electrospun fibres are reviewed in Chapter 2: Section 2.4.3. In summary, the
existing techniques require isolation and handling of individual fibres. The ends of an
individual fibre are attached to two points such that the length of the fibre is suspended. The
suspended midsections of two fibres are brought into contact in a cross-cylinder or parallel
configuration. A pull-off force is measured as the fibres are pulled apart at a constant
velocity. Here I develop a novel method for measuring the adhesive-detachment forces
between nano-fibres in situ, that is, within the fibre network. The technique captures the
native fibre contact configurations of self-assembled networks, and eliminates the need for
isolation and handling of nano-fibres. The method uses an AFM to perform what is referred
to as a ‘dip-and-drag’ test, which involves inserting an AFM tip into the network and
dragging it laterally. The forces measured are related to detachment events between fibres as
they are pulled apart from each other by the AFM tip. The technique is explained in detail in
Chapter 3: Section 3.2.2.
In the ideal scenario, the dip-and-drag technique pulls a single fibre and probes a single
contact zone to measure the detachment force. This cannot be practically and consistently
achieved, and often multiple fibres are pulled out of the network simultaneously leading to
large-scale network deformation. The fibre network density cannot be controlled for bacterial
cellulose systems, so model fibrous systems are first used to investigate the effect of network
deformation on fibre detachment. I validate the method using well-characterized electrospun
polymer fibres fabricated from sulphonated polyether ether ketone (SPEEK) and polyvinyl
alcohol (PVA). In these systems the adhesive forces are dominated by DLVO interactions,
making them suitable for testing the novel technique. I also test the ability of the technique to
measure hydrogen bonding interactions using fibre mats of Cellulose Nanocrystals (CNC)
and nano-fibrillated cellulose (CNF). The hydrogen bonding energy between the cellulose
CHAPTER 5. RESULTS Grace Dolan, PhD Thesis
111
fibres is different in water and air, and both environments are tested to confirm the ability of
the technique to measure these differences. This chapter is focussed on the rigorous
validation of the dip-and-drag method. In Chapter 6 the technique is applied to bacterial
cellulose systems to investigate the effect of individual cell wall components on cellulose
fibre interactions.
5.2 Experimental Section
5.2.1 Model fibre Systems
Fibre samples are prepared according to the fabrication procedures in Chapter 3: Sections
3.1.1 and 3.1.2. Images of the SPEEK, PVA, CNC and CNF fibres are presented in Figure
5.1. SPEEK fibres are electrospun directly only a glass substrate, whereas for PVA, CNC and
CNF samples, preformed fibres network are glued onto a glass substrate with 5-minute curing
epoxy resin (UHU GmbH & Co. KG, Germany) (equal parts base and curing agent). The
epoxy resin is applied delicately using a narrow glass rod. A section of the network
approximately 5mm x 5mm is glued around the edges leaving a small gap through which the
AFM tip can pull fibres out of the network.
5.2.2 Dip-and-drag technique
The dip-and-drag technique applied to bacterial cellulose micro-hydrogels is explained in
Chapter 3: Section 3.2.2. The sample principle is applied to the model fibre systems (SPEEK,
PVA, CNC, and CNF). Briefly, the JPK Nanowizard II AFM was mounted on an inverted
optical microscope (JPK Instruments, Germany). The AFM was loaded with a stiff cantilever
(HQ:NSC35/Cr-Au BS, Cantilever A) from Mikromasch (Nano World AG, Germany). The
networks were first imaged in intermittent contact mode in air to identify the exposed edge.
The imaging is performed at a scan rate of 2 Hz for a 60 x 60 µm scan size with 1024 x 1024
pixels. The set point and drive amplitudes are around 1 V and the drive frequency is around
200 kHz. Using manipulation control in contact mode, the path of the AFM tip was traced
over the image such that the tip was engaged around the exposed edge of the network for
PVA, or anywhere inside the homogeneous network for SPEEK samples, and then dragged
outward to measure fibre detachment events as illustrated in Figure 5.2.
CHAPTER 5. RESULTS Grace Dolan, PhD Thesis
112
Figure 5.1. Microscopy images of nano-fibrous networks. (a) SEM micrograph of
electrospun SPEEK nanofibre network. The inset shows a single SPEEK fibre with the
corresponding diameter measurement. (b) SEM micrograph of electrospun PVA nanofibre
network. (c) TEM image of Triodia pungens nanofibrils obtained via high pressure
homogenizer (CNF). (d) TEM image of Triodia pungens nanocrystal network obtained via
acid hydrolysis (CNC).
For lateral force measurements the set point vertical deflection was 300 nN and the AFM tip
travel speed was 0.3 µm/s. This lateral force measurement was repeated several times on
different parts of the network. The initial placement of the AFM tip inside the network is
random and in some cases may land on a fibre rather than the substrate. Force-distance curves
with an initial constant baseline force, where the lateral force is equal to the substrate friction
as depicted in Figure 5.2, are selected for analysis to ensure that a set point vertical force is
established between the AFM tip and the substrate. The test was repeated for a given SPEEK
CHAPTER 5. RESULTS Grace Dolan, PhD Thesis
113
Figure 5.2. The AFM tip is engaged with the substrate at a constant normal force and moved
in a lateral pulling direction. The initial lateral force is due to friction between the AFM tip
and substrate. When the tip engages with a fibre the lateral force increases due to fibre
deformation until a detachment event occurs at a fibre-fibre contact in the network.
fibre sample with aniline (Sigma) as a solvent. Aniline has a refractive index between that of
glass and SPEEK causing negative van der Waals interactions that substantially reduce the
adhesion between the fibres and substrate 18. By contrast, aniline has only a marginal effect
on adhesive forces between SPEEK fibres, because van der Waals interaction between
surfaces of the same material is always attractive 19. Subsequently, the results are compared
to those obtained from the measurement in air to confirm that the recorded forces are due to
fibre-fibre interactions and not influenced by fibre-substrate adhesion.
The lateral deflection data was recorded during the AFM tip trace and converted to lateral
force according to the procedure discussed in Chapter 3: Section 3.2.2. Measurements are
completed in ambient air for all samples as well as in water for CNC and CNF.
5.3 Results and Discussion
5.3.1 Dip-and-Drag Lateral Force Spectroscopy of SPEEK electrospun
mats of varying network density.
Figure 5.3 presents lateral force-distance measurements obtained for SPEEK fibre samples at
two extremes of network density (force-distance data for all SPEEK samples can be found in
CHAPTER 5. RESULTS Grace Dolan, PhD Thesis
114
Appendix K). The AFM tip is engaged with the substrate and translated laterally whilst
maintaining a constant normal force. During dragging, the baseline force is due to a
combination of friction between the tip and substrate, and elastic deformation of the network.
For low network density, it is suggested that the fibres are sufficiently far apart that a single
fibre is pulled without deforming large sections of the network. Thus the relatively constant
baseline force (designated ‘bf’) in Figure 5.3a is anticipated to be dominated by tip-substrate
friction. In contrast, the lateral force measured for the dense fibre network in Figure 5.3b is
steadily increasing, suggesting a baseline force that is dependent on degree of network
deformation. The baseline force in Figure 3b is initially constant, confirming that the AFM
tip first comes into contact with the substrate at the set point vertical force before contacting
fibres with lateral movement. The overall increase in lateral force is suggested to correspond
to the AFM tip dragging a number of fibres collectively, which leads to a large and
cumulative contribution of network deformation to the measured force.
In the force-distance profiles of SPEEK samples with low network density, consistent peaks
above the baseline force are observed (designated by * in Figure 5.3a). I propose that the
observed sharp increase in lateral force (above the baseline) corresponds to the AFM tip
engaging with a fibre and bringing it into tension. The abrupt decrease in lateral force is thus
associated with a detachment event at a fibre-fibre contact, such that the fibre is no longer in
tension and the signal returns back to the baseline value. Figure 5.3a indicates where I take
the height of the peak force (designated by ‘h’) to be a measure of the force required for
detachment at a fibre contact, which is akin to a ‘pull-off’ force between fibres. Figure 5.4a is
an SEM image of the SPEEK fibre substrate superimposed with a white line to represent a
potential 2 micron length pathway for the lateral movement of the AFM covered in Figure
5.3a. Whilst this particular image does not necessarily correspond to the section of the fibre
network that is measured, it provides an indication that the fibre detachment events occur in
line with the density of fibre interactions (designated by * in Figures 5.3a and 5.4a). The
image shows that it is plausible that 4 contacts are probed during measurement over the 2µm
lateral distance whilst maintaining a stable baseline force.
For high network density samples, a sharp drop in the lateral force is observed relative to the
deformation-dependent baseline force (Figure 5.3b). In this case, the measured force is
expected to be distributed across a number of fibre contact points. I suggest that when the
local force at a single contact point exceeds the adhesion force, fibre-fibre detachment occurs,
seen as a sharp drop in the measured lateral force. The peak height or pull-off force is
CHAPTER 5. RESULTS Grace Dolan, PhD Thesis
115
labelled ‘h’ in Figure 5.3b. In Figure 5.4b a 7µm trace is marked that corresponds to 7 fibre
contacts being disrupted. The fibres that are pulled during the trace have remaining
connections to the network leading to large scale deformation and an increasing baseline
force. This proposed trace is consistent with the observations in Figure 5.3b.
a
b
Figure 5.3. Typical lateral force-distance curves for SPEEK samples with (a) constant and
(b) increasing baseline force (bf). (*) denotes the peak events identified during data
processing. The peak height (h), calculated as the distance between the maximum and
subsequent minimum of the peak is labelled. The z-piezo position of the cantilever holder
relative to the substrate is plotted on the right-hand y-axis.
In Figure 5.3 there is an initial decrease in the z-piezo position of the cantilever (right axis in
Figure 5.3a and 5.3b) which corresponds to an overshoot of the vertical deflection before
CHAPTER 5. RESULTS Grace Dolan, PhD Thesis
116
settling to the set point with changing height. The overall drift in the height, particularly in
Figure 5.3b, is attributed to optical crosstalk between the photodetector signals related to
normal and torsional deflections of the cantilever. Due to rotational misalignment of the
photodetector during lateral force measurements, the vector representing the lateral shift of
the laser has a non-zero vertical shift 20. A substantial height change is observed in Figure
5.3b at a distance of around 6.5µm, which we suggest is due to fibre breakage. The frequency
of these types of events can be determined by histogram analysis of the population of peak
heights.
The entire set of force-displacement curves is analyzed using a semi-automated MATLAB
routine, for which the code is included in the Appendix L. The code identifies a peak if the
average force of x number of consecutive points on either side is less than the force at the
point of interest. The value of x is adjusted for the background noise frequency. The local
maximum and minimum of the identified peaks are found. The peak height is taken as the
distance between the maximum and subsequent minimum.
a
b
Figure 5.4. (a) and (b) show SEM images superimposed with proposed AFM tip traces
corresponding to the lateral distances measured for respective SPEEK samples in Figure 5.3
(a) and (b).
CHAPTER 5. RESULTS Grace Dolan, PhD Thesis
117
Figure 5.5a shows a representative force-distance curve for the lowest network density
SPEEK sample with the baseline force subtracted. The peak heights extracted from the set of
fibre pulling measurements are presented in a histogram in Figure 5.5b. The tail of the
distribution at large peak heights is attributed to the situations where the AFM tip cuts
through the fibres, which are shown to occur with low frequency.
Figure 5.5. Analysis of example force distance curve for SPEEK sample showing (a)
representative force-distance curve and (b) histogram of peak heights obtained from an entire
series of curves.
Peak heights correspond to fibre detachment events when a fibre under tension is released
from the network either by the fibre breaking or the adhesion between fibres at a contact zone
being overcome. SEM images of the different SPEEK samples after testing, labelled A
through to E, are shown in Figure 5.6. There is evidence of broken fibres, however some
broken fibres are also observed for SPEEK samples that have not been measured using the
dip-and-drag technique. Furthermore, the broken fibres seem to largely occur around the
globular structures, as seen in Figure 5.6, and could be an artefact of the electrospinning
process. Whilst fibre breakage may be occurring, the overall increasing baseline force in
Figure 5.3b suggests that for the most part fibres remain in tension. The experimental results
are compared to theoretical adhesion energies in the next section to support the interpretation
of the peak heights as a measure of the adhesion between fibres.
The shape of the distribution is largely influenced by the fibre network density and the
number of contacts that are in tension just prior to a detachment event occurring. To illustrate
this point, two scenarios of dragging a fibre are considered. As depicted in Figure 5.7a, either
the pulling force is applied at a single fibre contact or divided between two fibre contacts. For
CHAPTER 5. RESULTS Grace Dolan, PhD Thesis
118
the first scenario, the pulling force (FTOTAL) is equal to the adhesion force (Fadh) at the single
contact zone. For the second scenario the AFM tip can pull from any point along a fibre
connecting two contact zones, where the two distances are designated L1 and L2. The pulling
force stretches the fibre by the distance 2, and hence the strain applied to each segment of
the fibre is δ/L1 and δ/L2. The detachment at the weakest contact zone, either 1 or 2, occurs
when the resulting elastic force applied to the respective segment, 𝐹𝑖, is equal to the adhesive
force, 𝐹adh𝑖 as in equation 5.1.
𝐹𝑖 = 𝜋𝑅2𝐸𝛿𝑖
∗
𝐿𝑖= 𝐹adh𝑖
(5.1)
Figure 5.6. SEM micrographs at 10 000 x magnification of electrospun SPEEK nanofibre
networks with different electrospinning times labelled A through to E (scale bar is 1 µm). A2
is an SEM micrograph of substrate A at 2000 x magnification showing the breakages of
fibres around the globular-like structures.
CHAPTER 5. RESULTS Grace Dolan, PhD Thesis
119
R is fibre radius, E is Young’s modulus. Consequently, the strain at the pull-off, 𝛿𝑖∗ is given
by equation 5.2.
𝛿𝑖∗ =
𝐿𝑖
𝜋𝑟2𝐸𝐹adh𝑖
(5.2)
Here, the stretching force is assumed to be linear with deformation, and the pull-off force is
independent of elastic parameters of the fibre. The resulting force, FTOTAL, is the sum of the
forces applied to both segments of the fibre and is a function of the ratio of L1 and L2. The
expression for FTOTAL in equation 5.3 uses the Heaviside function as a convenient operator
which ‘selects’ whether the detachment occurs at contact 1 or 2. It can be seen from equation
5.3, that the total pulling force at the point of detachment is independent of elastic parameters
of the fibre and depends only on the ratio of fibre segment lengths, 𝑙 =𝐿1
𝐿2 , and the force of
adhesion (𝐹adh𝑖).
FTOTAL = 𝐹1 + 𝐹2 = [𝐻(𝛿2∗ − 𝛿1
∗) ∙ 𝐹adh1𝑙 + 𝐻(𝛿1
∗ − 𝛿2∗) ∙ 𝐹adh2
] ∙1 + 𝑙
𝑙
(5.3)
H() is the Heaviside function. Each detachment event depends on the three random variables
that contribute to the value of the force, Fadh1 , Fadh2 , and l. If Fadh1 and Fadh2 are assumed to
be part of the same distribution {𝐹adh} then FTOTAL is given by equation 5.4.
𝐹TOTAL = {𝐹adh} ∙ (1 + 𝑙) ∙ {1
𝑙, 𝑙 > 1
1, 𝑙 ≤ 1
(5.4)
Further, if a uniform distribution of the random variable l is assumed, then the ensemble
average force can be calculated using equation 5.5.
⟨𝐹TOTAL⟩ =1
𝑙∫ 𝐹TOTAL𝑑𝑙 ≈ 1.2773 ∙ {𝐹adh}
𝑙
0
(5.5)
Therefore, for the contact where the pulling force is divided between two fibres the measured
force is expected to have a similar distribution to the adhesive force of a single junction with
a correction coefficient of 1.3, as shown in Figure 5.7b. Amongst many peak force
distributions recorded, all of them feature a wide distribution of forces similar to the example
shown in Figure 5.5b. Due to the breadth of the distributions, both scenarios are captured
within the range of the most frequent event. Further refinement can be achieved by applying
CHAPTER 5. RESULTS Grace Dolan, PhD Thesis
120
Jarzynski’s averaging 21, 22 to the data according to equation 5.6. This exponential averaging
gives more weight to more frequent events that occur at low peak height values.
𝑒𝑥𝑝[−∆𝑄] = lim𝑛→∞
⟨𝑒𝑥𝑝[−ℎ𝑖]⟩𝑁 (5.6)
Q is the exponentially averaged peak height, <>N denotes arithmetic averaging of N peak
events, and hi represents the peak height of the ith event.
The average peak heights for SPEEK samples from the Jarzynski equality in equation 5.6 are
presented in Table 5.1. Remarkably, there is no dependence of the peak height on the network
density when comparing the SEM images in Figure 5.6 with the values in Table 5.1; in
particular samples D and E show the lowest fibre network density but not necessarily the
lowest peak height. This provides evidence that regardless of whether the baseline force is
deformation-dependent or not, the origin of the sharp decrease or peak in lateral force is the
same and likely to be a fibre-fibre detachment event. This supports the peak height
representing a ‘pull-off’ force between two fibres in contact. Elastic deformation of the fibre
network thus has little influence on the pull-off force, and the average for all samples is
8.4±1.4µN.
Table 5.1. The exponentially averaged peak height values from a set of force-distance curves
measured for SPEEK samples referenced A-E for increasing network density. Total number
of peak events is 280.
Sample
Reference
Peak Height
(µN)
A 6.8
B 6.9
C 11
D 7.3
E 10
To confirm that the measured detachment events are not due to fibre-substrate interactions,
the dip-and-drag technique is used to measure the forces for one of the SPEEK samples (E)
with aniline as a solvent. Aniline has a refractive index between glass and SPEEK, giving rise
to repulsive van der Waals interactions between the fibre and substrate whilst maintaining
attractive van der Waals interactions between fibres. There was found to be no statistically
significant difference between the population of peak heights measured in air and aniline with
CHAPTER 5. RESULTS Grace Dolan, PhD Thesis
121
a
b
Figure 5.7. (a) Illustrations of two possible scenarios for pulling a fibre and probing either 1
or 2 contact points. The insets show the SEM images of the SPEEK networks that may
illustrate the microscopic representation of such scenarios. (b) A theoretical model of a pull-
off experiment based on 10 000 junctions with normally distributed inter-fibre adhesive
energies. Open circles represent the frequency distribution of pull-off forces for Scenario 1.
Open diamonds represent the corresponding distribution of pull-off forces for Scenario 2,
calculated based on equation 5.3 and a uniform distribution of the parameter 𝑙 = 𝐿1 𝐿2⁄ ,
which corresponds to the random position of the AFM tip with respect to the contact points.
Solid lines are best fits using a Gaussian function. Dash lines mark the most frequent value of
pull-off force for each scenario. The dash-dot line marks the force value for the case of a
symmetric pull, i.e., l = 1
CHAPTER 5. RESULTS Grace Dolan, PhD Thesis
122
a p-value of 0.96. Thus the detachment events measured are attributed to the interactions
between fibres.
5.3.2 Dip-and-Drag Lateral Force Spectroscopy of PVA network
The developed technique is further tested on a network of electrospun PVA fibres with an
average diameter of 163 ± 42 nm. The force-distance curves for the PVA sample, an example
of which is in Figure 5.8a, shows a large contribution of network deformation to the baseline
force that is consistent with a highly dense network. The histogram of peak heights in Figure
5.8b is characterized by the Jarzyski’s averaged peak height of 6.1µN which I take as the
PVA pull-off force for further analysis.
Figure 5.8. (a) Representative force-distance curve for PVA and (b) histogram of peak height
values for the entire set of curves.
5.3.3 Analysis of adhesive forces between fibres
It is appropriate to consider electrospun polymer fibres in the network to be interacting
through van der Waals forces. The electrospinning procedure gives reasonably cylindrical
fibres and capillary forces are expected to be negligible for the contact angles created by two
cylinders. I examine whether the order of magnitude of the measured pull-off forces for
SPEEK and PVA fibres is comparable to the forces (FvdW) predicted from van der Waals
interaction energy between two cylindrical bodies in parallel configuration according to
equation 5.7 23. For a parallel configuration the length (L) of fibre-fibre contact needs to be
specified. The L used for the calculation is the contact length of two fibres interacting with an
orientation at the midpoint between parallel and orthogonal, that is, a 45° angle between
longitudinal axes of fibres.
CHAPTER 5. RESULTS Grace Dolan, PhD Thesis
123
𝐹vdW = 𝐴H𝐿√𝑅
16𝐷5
2⁄
(5.7)
AH is the Hamaker constant, R is the fibre radius and D is the separation distance which is
reported to be of the order of 1 Å for strong van der Waals interaction 23. There is uncertainty
around the values of the Hamaker constants for SPEEK and PVA. Therefore I will use
equation 5.7 to predict the Hamaker constant from the experimental pull-off forces 8.4 µN
and 6.1 µN as FvdW for SPEEK and PVA, respectively. Additionally, to test the effect of the
pulling configuration on the measured force, the experimental pull-off forces are adjusted by
applying the factor 1.3 in accordance with equation 5.5. The Hamaker constant for these
corrected values is again predicted from equation 5.7. The Hamaker constants computed this
way correspond to the scenario where the dragging force is predominantly divided between
two fibre contacts. I will then compare the experimentally determined values to the
theoretically calculated Hamaker constants using a number of different approaches.
The theoretically predicted Hamaker constants are calculated using two methods. Firstly, a
method based on extracting the surface energies from the wetting data is employed, which
gives 𝐴HSE = 5.81 x10-20 J and 10.5 x10-20 J for SPEEK and PVA, respectively 24, 25.
Secondly, a full calculation based on the Lifshitz’ theory of van der Waals forces is
performed 19, which gives 𝐴HL = 12.9 x10-20 J and 10.5 x10-20 J for SPEEK and PVA,
respectively. These calculations use the integrated form of the equation for the free energy of
interaction developed by Parsegian and Ninham 26. The permittivity spectra are calculated
using the method of Hough and White 27 in equation 5.8, which is based on the representation
of the material’s dielectric response 휀(𝑖) at the imaginary frequency (𝑖) as a sum of two
damped Lorentz oscillators 28.
휀(𝑖) = 1 +𝐶UV
(1 + (
𝜔UV⁄ ))
+𝐶IR
(1 + (
𝜔IR⁄ ))
(5.8)
All constants, CUV, UV, CIR, IR, are determined from the experimental data available in the
literature. The CUV and UV values are extracted from the Cauchy equation (equation 5.9) and
the refractive index spectra available for SPEEK 29 (CUV = 1.6, UV =1.24×1016 rad/s) and
PVA 30 (CUV = 1.5, UV =1.36×1016 rad/s).
CHAPTER 5. RESULTS Grace Dolan, PhD Thesis
124
𝑛2(𝜔) − 1 = 𝐶UV + (𝑛2(𝜔) − 1)𝜔2
𝜔UV2
(5.9)
The IR frequencies are evaluated based on FTIR spectra for SPEEK 31 and PVA 30. The
maximum absorption in the IR are found at 3451 cm-1 (IR = 6.50×1014 rad/s) and 2044 cm-1
(IR = 3.85×1014 rad/s) for SPEEK and PVA, respectively. The CIR is determined based on
the approximate equation 5.10 28.
𝐶IR = 휀0 − 𝐶UV − 1 (5.10)
0 is the static dielectric constant, which was determined from the electrical impedance
measurements for SPEEK (0 = 5, CIR = 2.4) 32 and PVA 33 (0 = 3, CIR = 0.5).
From Table 5.2, the AH values for PVA calculated using wetting data and the Lifshitz’ theory
show an excellent agreement, and highlight the fundamental equivalency of both approaches.
For SPEEK, the discrepancy is somewhat larger, which may be associated with changes in
the SPEEK material in the presence of water that impact the wetting studies 25. The
experimentally determined values of the Hamaker constant are found to be in a good
agreement with theoretical predictions. In particular, a very good agreement is observed for
SPEEK when 𝐴HL is compared to the experimental Hamaker constant calculated using
equation 5.7 and 5.5, i.e. for the scenario where the dragging force is divided between two
fibre contacts. In contrast, for PVA we observe a very good agreement for the case of a single
contact model, i.e. ⟨𝐹TOTAL⟩ ≈ {𝐹adh}. Although these values should be taken as an order of
magnitude approximation, it is plausible that the structure of the network may play a role in
determining the probability of pulling configurations. From Figure 5.1, fibres in the SPEEK
electrospun mat are more regularly aligned compared to PVA, where many fibres are being
curled and entangled into bundles containing more than two fibres. Thus, a more grid-like
configuration of SPEEK fibre mats may favor configurations that involve two-fibre contacts.
5.3.4 Dip-and-Drag Lateral Force Spectroscopy of CNC and CNF networks
in air and water
The dip-and-drag technique is also applied to CNC and CNF fibre mats. CNF has regular
hydroxyl groups which retain their hydrogen bonding potential, whereas the sulphuric acid
used for hydrolysis changes some hydroxyl groups to sulphate groups on the surface of CNC.
CHAPTER 5. RESULTS Grace Dolan, PhD Thesis
125
Table 5.2. The experimental and theoretical values of the Hamaker constant (𝐴H) for SPEEK
and PVA. The theoretical values are calculated using the surface energy approach (𝐴HSE), and
the full Lifshitz’ theory calculation using the integrated form of the equation for the free
energy of interaction developed by Parsegian and Ninham (𝐴HL) (see text for details of the
calculation). The experimental values are calculated using eq 7 and (i) the average
experimental values of pull-off forces (𝐴Hexp), (ii) the average experimental values of pull-off
forces adjusted using a correction factor from eq 5 that corresponds to the scenario where the
dragging force is divided between two fibre contacts (𝐴Hexp°).
𝑭𝐩𝐮𝐥𝐥−𝐨𝐟𝐟 [𝝁𝑵] 𝑨𝐇𝐒𝐄 [J 10-20] 𝑨𝐇𝐋
[J 10-20] 𝑨𝐇𝐞𝐱𝐩 [J 10-20] 𝑨𝐇𝐞𝐱𝐩° [J 10-20] L [nm]
SPEEK 8.4 5.8 12.9 16.07 12.6 331
PVA 6.1 10.5 10.5 9.4 7.4 362
The average diameter and length of individual CNC is 3.5±0.8 nm and 500±100 nm,
respectively, while the average diameter of CNF is 4.5±1.5 nm with a length of several
microns. In both cases the nanofibres agglomerate upon drying to give fibre diameters of the
order of 300nm. There are certain challenges with verifying the experimental pull-off force
from the dip-and-drag technique against theoretical predictions for hydrogen bonding. The
hydrogen bonding energy can be used to predict the adhesive energy between fibres if the
number of hydrogen bonding sites at the contact area is approximated. However, the
experimental pull-off force would need to be converted to an adhesive energy for
comparison. It is not clear how the fibres are separated during the dip-and-drag test, for
example, they could be peeled apart. Information about the mechanism of fibre detachment is
required to choose an appropriate model for converting the pull-off force to adhesion energy.
In Chapter 6: Section 6.3.2 I include simulations from a ComsolTM Multiphysics model to
relate the pull-off force and adhesion energy at the fibre contact. In this chapter, I assess
whether the dip-and-drag technique is able to measure pull-off forces between CNC and CNF
fibres that interact through hydrogen bonding. I can validate this by testing the fibre samples
in air and water and comparing how the experimental values scale with the hydrogen bonding
energies in air and water.
The force-distance curves for CNC and CNF in air, for which representative curves are in
Figure 5.9a and 5.9b, show large peaks in lateral force at the beginning of the AFM tip path.
The maximum lateral force is reached and the sharp drop is followed by stepped decreases in
the successive peaks. A similar profile is observed for CNC and CNF in water, shown in
CHAPTER 5. RESULTS Grace Dolan, PhD Thesis
126
Figure 5.9c and 5.9d, although the peaks appear smoother/rounder than in air. The curves are
analysed and the Jarzynski’s average of peak heights are presented in Table 5.3.
Figure 5.9. Representative force-distance curves for (a) CNC in air, (b) CNF in air, (c) CNC
in water, and (d) CNF in water.
Table 5.3. The Jarzynski’s average peak height value for CNC and CNF samples in air and
water. Sample size for each is approximately 50 peak events.
Sample Peak Height
(µN)
CNC in air 11
CNC in water 2.7
CNF in air 9.1
CNF in water 3.4
The peak heights, or pull-off forces, are comparable for CNF and CNC suggesting that the
slight changes to surface chemistry and the different aspect ratios of the fibres do not
significantly affect the interactions between fibres. The pull-off forces in air are larger than
those measured for SPEEK and PVA fibres. This is likely to be due to a combination of the
CHAPTER 5. RESULTS Grace Dolan, PhD Thesis
127
larger fibre diameter of the aggregated CNC and CNF compared to the electrospun fibres,
and the higher adhesion energy associated with hydrogen bonding compared to van der
Waals interactions which dominate for the polymer fibres. The differences between the pull-
off forces in air and water in Table 5.3 roughly scale with the difference between hydrogen
bonding energy in air (20kJ/mol) compared to water (6.6kJ/mol) 34. This finding strongly
supports the ability of the technique to measure the hydrogen bonding interactions between
fibres. The relationship between the pull-off force and adhesion energy for cellulose fibre
contacts will be investigated in Chapter 6.
5.4 Concluding Remarks
I have developed a novel technique for measuring the adhesion between individual nano-
fibres. Currently, there are technical challenges associated with isolating and handling
individual nano-fibres for measuring the adhesion between them in cross and parallel
configurations. In this study, an AFM cantilever tip is used to drag fibres out of a network
and measure the pull-off force. This technique has the unique advantage of working directly
with fibrous networks which inherently have a distribution of fibre diameters and
orientations. I consider the Jarzynski’s average of the distribution of detachment forces to
represent the pull-off force corresponding to the event where a single fibre contact is probed.
The experimental results are in good agreement with theoretical adhesion for the nanofibres
(electrospun SPEEK, electrospun PVA) which interact through van der Waals forces. CNC
and CNF fibre mats in air and water are used to demonstrate the applicability of the technique
for measuring the pull-off force for fibres interacting through hydrogen bonding. The
validated dip-and-drag technique will be used to measure the effect of important cell wall
components on the adhesive interactions between bacterial cellulose fibres. This will be done
in combination with a computational study that relates the measured pull-off force from the
dip-and drag technique to the adhesive energy at fibre contacts.
CHAPTER 5. RESULTS Grace Dolan, PhD Thesis
128
References for Chapter 5
1. E. Chanliaud, K. M. Burrows, G. Jeronimidis and M. J. Gidley, Planta, 2002, 215,
989-996.
2. R. Huang, W. Z. Li, X. X. Lv, Z. J. Lei, Y. Q. Bian, H. B. Deng, H. J. Wang, J. Q. Li
and X. Y. Li, Biomaterials, 2015, 53, 58-75.
3. P. A. Janmey, U. Euteneuer, P. Traub and M. Schliwa, Journal of Cell Biology, 1991,
113, 155-160.
4. J. F. Leterrier, J. Kas, J. Hartwig, R. Vegners and P. A. Janmey, Journal of Biological
Chemistry, 1996, 271, 15687-15694.
5. P. Lopez-Sanchez, M. Rincon, D. Wang, S. Brulhart, J. R. Stokes and M. J. Gidley,
Biomacromolecules, 2014, 15, 2274-2284.
6. L. L. Ma, J. Y. Xu, P. A. Coulombe and D. Wirtz, Journal of Biological Chemistry,
1999, 274, 19145-19151.
7. D. Poquillon, B. Viguier and E. Andrieu, Journal of Materials Science, 2005, 40,
5963-5970.
8. Y. Tseng, K. M. An, O. Esue and D. Wirtz, Journal of Biological Chemistry, 2004,
279, 1819-1826.
9. S. E. C. Whitney, M. G. E. Gothard, J. T. Mitchell and M. J. Gidley, Plant
Physiology, 1999, 121, 657-663.
10. L. Zhou, H. He, C. Jiang and S. He, Journal of Applied Polymer Science, 2015, 132.
11. Y. Zou and Y. H. Zhang, Ann. Biomed. Eng., 2009, 37, 1572-1583.
12. P. Lopez-Sanchez, J. Cersosimo, D. Wang, B. Flanagan, J. R. Stokes and M. J.
Gidley, PLoS ONE, 2015, 10.
13. T. Komori and M. Itoh, Textile Research Journal, 1991, 61, 420-428.
14. M. A. Narter, S. K. Batra and D. R. Buchanan, Proceedings of the Royal Society a-
Mathematical Physical and Engineering Sciences, 1999, 455, 3543-3563.
15. C. W. Wang, L. Berhan and A. M. Sastry, Journal of Engineering Materials and
Technology-Transactions of the Asme, 2000, 122, 450-459.
16. X. F. Wu and Y. A. Dzenis, Journal of Applied Physics, 2005, 98.
17. A. P. Chatterjee, Journal of Applied Physics, 2006, 100.
18. A. A. Feiler, L. Bergstrom and M. W. Rutland, Langmuir, 2008, 24, 2274-2276.
19. I. E. Dzyaloshinskii, E. M. Lifshitz and L. P. Pitaevskii, Advances in Physics, 1961,
10, 165-209.
CHAPTER 5. RESULTS Grace Dolan, PhD Thesis
129
20. M. Munz, Journal of Physics D-Applied Physics, 2010, 43.
21. C. Jarzynski, Physical Review Letters, 1997, 78, 2690-2693.
22. C. Jarzynski, Physical Review E, 1997, 56, 5018-5035.
23. J. N. Israelachvili, Intermolecular and Surface Forces, 3rd Edition, 2011.
24. J. Vial and A. Carre, International Journal of Adhesion and Adhesives, 1991, 11, 140-
143.
25. P. Kanakasabai, P. Vijay, A. P. Deshpande and S. Varughese, Journal of Power
Sources, 2011, 196, 946-955.
26. V. A. Parsegian and B. W. Ninham, Journal of Theoretical Biology, 1973, 38, 101-
109.
27. D. B. Hough and L. R. White, Advances in Colloid and Interface Science, 1980, 14,
3-41.
28. K. Cavalier and F. Larche, Colloids and Surfaces a-Physicochemical and Engineering
Aspects, 2006, 276, 143-145.
29. B. T. Koziara, K. Nijmeijer and N. E. Benes, Journal of Materials Science, 2015, 50,
3031-3040.
30. S. Mahendia, A. K. Tomar, R. P. Chahal, P. Goyal and S. Kumar, Journal of Physics
D-Applied Physics, 2011, 44.
31. J. M. Song, J. Shin, J. Y. Sohn and Y. C. Nho, Macromolecular Research, 2011, 19,
1082-1089.
32. S. Changkhamchom and A. Sirivat, Polymer Bulletin, 2010, 65, 265-281.
33. S. El-Sayed, K. H. Mahmoud, A. A. Fatah and A. Hassen, Physica B-Condensed
Matter, 2011, 406, 4068-4076.
34. S. Y. Sheu, D. Y. Yang, H. L. Selzle and E. W. Schlag, Proceedings of the National
Academy of Sciences of the United States of America, 2003, 100, 12683-12687.
130
Chapter 6
Measuring the effect of
hemicelluloses on the adhesive forces
between cellulose fibres
6.1 Introduction and Background
Based on the results in Figure 4.23 in Chapter 4: Section 4.3.6, I hypothesise that XG reduces
the adhesion between cellulose fibres, whereas AX has no effect. The main objective of this
chapter is to test this hypothesis. In Chapter 5 a dip-and-drag technique is developed to probe
the interactive forces at individual fibre contacts. The lateral force is recorded as an AFM tip
is dragged through a fibre network. Peaks in the recorded force-distance curves are
interpreted to represent fibre detachment events. Here, the dip-and-drag technique is applied
to bacterial cellulose networks and composites with AX and XG. These polymers interact
with the surface of cellulose fibres and mediate cellulose fibres contacts. The force-distance
curves obtained for cellulose and composite networks are compared, to determine the
influence of AX and XG. A model of fibre detachment is constructed in ComsolTM
Multiphysics to interpret and validate the experimental force-distance curves, and elucidate
the precise effect of AX and XG on the physical properties of cellulose fibre contacts.
Cellulose fibres are described as having a paracrystalline surface layer and a crystalline
domain in the core1, 2. The paracrystalline state has intermediate mechanical properties
between crystalline (high modulus) and amorphous (low modulus) phases. The partially
ordered structure of the paracrystalline surface layer is thought to permit an association
CHAPTER 6. RESULTS Grace Dolan, PhD Thesis
131
between the crystalline cellulose core and the semicrystalline hemicellulose in the cell wall1.
The hierarchical structure of bacterial cellulose has been investigated using small angle
scattering techniques with XRD and SEM3. Bacterial cellulose ribbons are approximated by a
core-shell model similar to plant cellulose where the outer region (shell) consists of solvent
accessible hydrated paracrystalline cellulose and the core is composed of cellulose
crystallites. Thus it is expected that the nature of interactions between bacterial cellulose and
hemicelluloses is representative of the associations in plant cell walls.
AFM imaging of onion epidermis shows that the cellulose fibres come into close proximity
with one another over short distances4. It is difficult to use this direct visualisation technique
to investigate the nature of interaction between cellulose fibres because the contact points are
obscured by matrix material. However, due to the prevalence of matrix material around the
contact points it is suspected that the polymers interact with cellulose and influence the nature
of the fibre contacts. Park and Cosgrove5 investigate the biomechanical changes in cucumber
and Arabidopsis hypocotyl walls induced by substrate-specific endoglucanases. They find
that endoglucanases that hydrolyse both XG and cellulose are required to induce creep. Park
and Cosgrove5 postulate that there is a minor, relatively inaccessible XG compartment that
may be intertwined or otherwise complexed with cellulose at fibre contact points. There are a
number of hypotheses for the nature of interaction between XG and cellulose, including;
entanglement with the amorphous glucans on the fibril surface5, 6, physical entrapment inside
the microfibril during synthesis5, 7, covalent bonding via a transglycosylation reaction8, or
non-covalent bonding5. In vitro cellulose binding experiments on the walls of barley aleurone
cells, containing 85% arabinoxylan (AX), indicate strong non-covalent bonds between the
AX chains themselves and with cellulose fibres9. XG and AX are the most abundant
hemicelluloses across plant species and their interaction with the surface of cellulose fibres is
investigated in this chapter.
Currently, the most reliable information regarding inter-fibre adhesion is inferred indirectly
from the analysis of macroscopic mechanical properties of cellulose networks. The
mechanical properties of bacterial cellulose and composite hydrogels (with AX and XG) have
been probed using small deformation oscillatory rheology tests and large deformation
uniaxial tensile testing10, equi-biaxial tensile testing11, and under compression12. From these
tests, the modulus of cellulose hydrogels and cellulose composites are measured to range
from 0.1 to 1 MPa10-12. The mechanical properties of fibre networks are vastly different to
individual bacterial cellulose fibres, which are reported in Chapter 2: Section 2.4.2 to have a
CHAPTER 6. RESULTS Grace Dolan, PhD Thesis
132
Young’s modulus of the order of 100 GPa13. From these multi-scale measurements, and
based on fibre network models reviewed in Chapter 2: Section 2.4.1, it is implicit that the
mechanical properties of cellulose-based composites are largely driven by interactions
between cellulose fibres, matrix polymers, and the assembly of these components into a
network.
The AFM has enabled direct measurement of the friction and adhesion forces between model
cellulose surfaces including pulp fibres14, 15, spherical cellulose particles16-18, and cellulose
thin films18-21. All of these attempts to measure cellulose friction and adhesion probe contacts
between large aggregates of cellulose fibres of the order of 10 m. This length scale is
relevant for biomaterials that use commercial sources of cellulose, but is not representative of
interactions between individual cellulose fibres that are important for plant cell wall
mechanics. Hence, the dip-and-drag technique provides new insights in to the interactions
between cellulose fibres, AX, and XG at individual fibre contacts.
6.2 Experimental Section
Bacterial cellulose and composite networks are grown in confined 50 µm wells and deposited
onto a glass slide for measurement. The complete steps for producing these substrates are
detailed in Chapter 3: Section 3.1.3. The dip-and-drag technique is explained in Chapter 3:
Section 3.2.2. All measurements are carried out in deionised water from a Milli-Q Advantage
A10 system with a resistance of 18 Ω.cm at 25 °C.
6.3 Results and Discussion
6.3.1 Probing contacts between individual cellulose fibres
The structure of cellulose fibres synthesised by Gluconacetobacter xylinus is hierarchical.
Firstly, cellulose chains synthesised and extruded out of the pores in the bacteria’s plasma
membrane. These cellulose chains assemble into protofibrils with a diameter of ca. 2-4nm
(Iguchi, Yamanaka et al. 2000). Subsequently, protofibrils aggregate into ribbon-shaped
fibres with dimensions of the order of tens of nm. The morphology of bacterial cellulose
ribbons and fibre contacts are show in Figure 6.1a and 6.1b. Using this bacterium, cellulose
can be produced to form a very thin 3D-network. The vertical dimension of the network is
required to be smaller than the height of the AFM tip so that the tip can penetrate the network
CHAPTER 6. RESULTS Grace Dolan, PhD Thesis
133
and contact the glass surface underneath to establish a baseline force during the dip-and-drag
experiments. Single fibre loops around the edge are pulled away from the network to probe
the interactive forces at fibre contact points, as seen in Figure 6.1c.
a
b
c
Figure 6.1. (a) and (b) AFM images of an air-dried cellulose network showing fibres and
fibre contacts. The scale on the left hand side is the vertical dimension, and the scale on the
right hand side is the horizontal dimension. (c) AFM image of the edge of cellulose network
showing a loose fibre loop that is pulled with the AFM tip. The arrow represents the desired
path of the AFM tip, where is engages with the glass substrate at a vertical force of 300nN
CHAPTER 6. RESULTS Grace Dolan, PhD Thesis
134
and is then dragged outward from the network to bring the fibre into tension and drive a fibre
detachment event. The scale on the right hand side is the verticle dimension and the
coordinates in a horizontal plane are marked on the left and bottom axes.
The dip-and-drag technique and analysis methods in Chapter 5: Section 5.3.1 are applied to
bacterial cellulose composite systems to determine the role of cell wall components on
cellulose fibre interactions. Peaks in the recorded lateral force-distance curves are observed in
a relatively consistent manner. I postulate that these peaks, an example of which is shown in
Figure 6.2, correspond to detachment events at contact points between individual cellulose
fibres. I interpret the observed sharp increase in force (above the noise) as corresponding to
the AFM tip engaging with a cellulose fibre in tension. This is followed by a detachment
event at a fibre contact point. Thus, the fibre being pulled by the AFM tip is no longer in
tension and the signal returns back to the moving baseline. For these systems, the baseline
force is not constant and is a function of the network mechanics, as described in Chapter 5:
Section 5.3.1. The lateral force-distance data for a set of experiments is processed to identify
peaks that represent the detachment at a fibre contact. An example force-distance curve for
the cellulose network is presented in Figure 6.3a. The peaks are denoted with ‘*’, and the
peak height for one of the detachment events is labelled ‘h’ in Figure 6.3a. A histogram of the
entire population of peak heights is included in Figure 6.3b.
In order for a detachment event to occur, the force applied directly at the contact must be
greater that then adhesive force between the fibres. To assist in interpreting these results, an
example of a force balance across a section of the network during a pulling experiment is
considered in Figure 6.4. The critical observation from this scenario is that the pulling
configuration is an important factor in determining which fibre contact is most likely to
detach. The AFM tip applies a force directly to the fibre that it is in contact with, and this
force gets divided between a number of fibres as you move further into the network. For
example, the 7 fibres at the bottom system boundary experience approximately a seventh of
the pulling force applied to the single fibre at the top system boundary. Thus, if the adhesive
force at all the fibres contacts is comparable, fibre detachment is most to likely occur at the
encircled contact in Figure 6.4 as it experiences the largest direct pulling force. In Figure 6.4
the pull-off force at the encircled contact is assumed to be equal to the pulling force measured
by the AFM tip at the point of detachment.
CHAPTER 6. RESULTS Grace Dolan, PhD Thesis
135
Figure 6.2. Lateral force-distance curve showing a typical peak that is representative of a
detachment event at a fibre contact point.
a
b
Figure 6.3. (a) Example force-distance curve for cellulose fibre network, * denotes the peaks
in the curve that represent detachment events at fibre contacts and h is an example of the peak
height. (b) Histogram of complete data set of peak heights (n = 100).
Peak Height (µN)
0 5 10 15 20 25 30 35
Count
0
5
10
15
20
25
CHAPTER 6. RESULTS Grace Dolan, PhD Thesis
136
Jarzynski’s equality in equation 6.1 is used to characterise the pull-off force required to
detach a single fibre contact, and gives a value of 4.12 µN.
𝑒𝑥𝑝[−∆𝑄] = lim𝑛→∞
⟨𝑒𝑥𝑝[−ℎ𝑖]⟩𝑁 (6.1)
Q is the exponentially averaged peak height, <>N denotes arithmetic averaging of N peak
events, and hi represents the peak height of the ith event.
6.3.2 Simulating fibre-fibre detachment events
In order to simulate the scenario portrayed in Figure 6.4, a simplified model is implemented
in ComsolTM Multiphysics using the beam mechanics interface. The model setup is depicted
in Figure 6.5. Contacts 1 and 2 in Figure 6.4 are assumed to be fixed in the simulation. The
cross section of the fibrils is assumed to be rectangular (30 nm width × 15 nm height) and the
fibril modulus is taken as 78 GPa13. The contact is modelled as a collection of ten beams
separated by 1 nm, each with an equilibrium length, 𝑙0. The mechanics of the contact are set
up to follow a simplified cohesive zone model (CZM) structure22, with the contact strength K
following equation 6.2. The CZM describes the cohesive forces that occur as material
elements are separated. During separation, the overall cohesive force reaches a maximum,
which subsequently reduces to zero when the surfaces are completely separate. This is
consistent with the experimentally measured peaks in the force-distance curves and thus
provides a good model of the detachment at fibre contacts.
𝐾 = 𝐾0𝐻(𝜀𝑐 − 𝜀) + 𝐾0𝑒−𝛼(𝜀−𝜀𝑐)𝐻(𝜀 − 𝜀𝑐) (6.2)
𝐾0 is the contact strength of the unstretched contact beams, 𝜀 is contact strain, 𝜀𝑐 is the
critical contact strain, and H(x) is the Heaviside function which takes the value of zero for x <
0 and unity for x ≥ 0. Hence, the contact beams weaken exponentially when 𝜀 > 𝜀𝑐 with a
decay constant . This is a departure from the usual formulation of the CZM, which is
written in terms of contact displacement rather than strain and assumes a weakening law for
𝐾 that leads to a finite detachment displacement. For the present system, incorporating a
finite detachment displacement adds an additional parameter which is highly dependent on
the type of polymer (AX or XG) mediating the fibre contact and cannot be easily extracted
from the experimental data. Expressing K in terms of strain separates the calculated force
from the influence of 𝑙0. What the modified version of the CZM in equation 6.3 physically
CHAPTER 6. RESULTS Grace Dolan, PhD Thesis
137
describes is that after the critical strain is reached, softening of the contact occurs and the
stiffness tends to zero at a rate governed by α. A small value of α leads to quick detachment;
this creates instability in the ComsolTM Multiphysics simulation because it means that there is
a fibre freely “floating” in space. Therefore, is fixed ( = 15) for numerical stability. It is
possible to set up a model in which complete fibre detachment occurs, but I am examining
the pull-off force and not detachment length so this refinement of the model is not required.
Figure 6.4. Force balance across a section of the fibre network to illustrate that the pulling
force recorded by the AFM tip is a good estimate of the force acting at the (encircled) fibre-
fibre contact, where the detachment event occurs. The dashed line marks the system
boundary over which the force balance is applied.
Parametric sweeps are performed over K0, 𝜀𝑐, and the ratio between beam lengths (b = L1/L2)
that are labelled in Figure 6.5. While apparently restrictive, the constructed scenario is
representative of the fact that only the encircled contact will, on average, detach upon pulling.
In reality, contacts 1 and 2 can translate in the pulling direction without detaching, which
increases the pulling distance at the point of detachment. However, I am investigating the
pull-off force which is found below to be independent of the structural factor b, and is
exclusively related to the parameters of the encircled contact in Figure 6.5. This means that
CHAPTER 6. RESULTS Grace Dolan, PhD Thesis
138
the force applied to the contact is constant as the fibres are being pulled apart, up until the
time that the contact is separated.
Figure 6.5. Simplified setup of the system depicted in Figure 6.3 implemented in ComsolTM
Multiphysics. Due to their large aspect ratio, the cellulose fibres can be modelled as ideal
beams. ComsolTM Multiphysics does not allow the use of its adhesion features in this kind of
system, so the contact is modelled as a collection of beams (which are shown in the inset) that
soften when a critical strain is reached. Contacts 1 and 2 in Figure 6.4 are assumed to be
fixed.
Some sample curves from the parametric sweeps are presented in Figure 6.6. The simulated
pulling force increases linearly with distance until a peak force is reached, beyond which the
pulling force decreases as the contact strength decays and the fibres are separated. This does
not follow the shape of the experimental peaks in force-displacement, for example in Figure
6.1d; however it is found that varying the decay constant α had no influence on the peak
pulling force, and the selected value of α = 15 simply allowed for stable simulations. The
CHAPTER 6. RESULTS Grace Dolan, PhD Thesis
139
peak pulling force is equivalent to the experimentally measured peak heights and is taken as
the pull-off force between fibres under the specific conditions. When comparing the
respective force-distance curves with black and grey symbols in Figure 6.6, it is clear that b
changes the initial linear slope but does not affect the pull-off force. This result is
fundamentally important because it confirms that, on average, the pull-off force is
independent of the system configuration. Moreover, comparing curves with the same K0 and
b but different c reveals that c does not modify the initial slope of the curve. Increasing c
leads to an increase in the pull-off force. The slope and the pull-off force increase with K0.
The curves in Figure 6.6 are sample curves to illustrate the effect of changing system
parameters. From these results, it can be concluded that the initial slope, s, of the force-
distance curve depends on the contact strength K0 and the structural factor b, and is
independent of the critical strain c. The peak force, Fpull-off, depends on the contact strength
K0 and the critical strain c, and is independent of the structural factor b. These conclusions
are summarised in equation 6.3 and equation 6.4.
𝑠 = 𝑓(𝐾0, 𝑏) (6.3)
𝐹𝑝𝑢𝑙𝑙−𝑜𝑓𝑓 = 𝑓(𝐾0, 𝑐) (6.4)
Figure 6.6. Predicted force curves for combinations of 2 different values of b, K0 and c. The
values of K0, c, and b are labelled in the legend.
Distance ( m)
0.0 0.1 0.2 0.3
Pulli
ng f
orc
e (
N)
0.00
0.04
0.08
0.12
0.16K0 = 10 MPa; c = 0.2; b = 0.5
K0 = 10 MPa; c = 0.2; b = 1
K0 = 10 MPa; c = 0.4; b = 0.5
K0 = 10 MPa; c = 0.4; b = 1
K0 = 20 MPa; c = 0.2; b = 0.5
K0 = 20 MPa; c = 0.2; b = 1
K0 = 20 MPa; c = 0.4; b = 0.5
K0 = 20 MPa; c = 0.4; b = 1
CHAPTER 6. RESULTS Grace Dolan, PhD Thesis
140
Figure 6.7 presents 3-D plots of the functions in equation 6.3 and equation 6.4, and the
equations of the best curve fits are given in equation 6.5 and 6.6, respectively.
𝑠 = 𝑒𝑥𝑝[436.51 − 0.017𝐾00.5 ln(𝐾0) + 1.05 ln(𝐾0) − 130.2𝑏 + 24.86𝑏 ln(𝑏)
− 434.64exp(−𝑏 11.67) + 25.95𝑏0.5 ln(𝑏) + 87.97𝑏0.5⁄ ]
(6.5)
𝐹𝑝𝑢𝑙𝑙−𝑜𝑓𝑓 = 10𝑒𝑥𝑝 [−5.21 −0.088
𝜀𝑐− 2.67 exp(−𝜀𝑐) − 6×10−4𝐾0
+ 6×10−3𝐾00.5 + ln(𝐾0)]
(6.6)
a
b
Figure 6.7. Best surface fit describing the functional relationship between a) slope s, contact
strength K0, and the structural parameter b; and b) peak force Fpull-off, contact strength K0, and
critical contact strain c.
The system is underdetermined with two equations, equation 6.5 and equation 6.6, and three
unknowns; K0, b, and εc. One way to circumvent this issue is by including a third equation for
the adhesion energy, which is approximated for cellulose fibres based on the theoretical
hydrogen bonding energy. The expression for adhesion energy, A, is given in equation 6.7.
The adhesion energy due to hydrogen bonding between cellulose surfaces is estimated from
information about the cellulose structure. Ding et al.23 use the AFM to directly image
different types of cell walls in maize. From the results, the authors propose a model of plant
cellulose fibres with 36-chain cellulose elementary fibres (CEFs) associated through
hydrophilic faces to create ribbon-like bundles. In this model, the CEFs are hexagonal shape
with four hydrophilic faces showing 8 cellulose chains involved in hydrogen bonding over a
height of 3.2nm. The repeating unit of cellulose, two glucose monomers, is approximately
CHAPTER 6. RESULTS Grace Dolan, PhD Thesis
141
1nm in length24, and provides sites for 4 hydrogen bonds25. For cellulose chains on the
surface of cellulose ribbons, 2 sites will be available for hydrogen bonding with an adjacent
ribbon at a contact point. Therefore an area of 3.2nm x 1nm has a potential for 16 hydrogen
bonds. In water, the hydrogen bond energy is 6.6kJ/mol26, giving a value of 0.055 J/m2 for A
between cellulose fibres. This value provides a good estimate for comparing the influence of
AX and XG at fibre contacts. The value for 𝑙0, between cellulose fibres is taken to be 0.3nm,
which is of the order of the distance between two water molecules.
𝐴 =𝐾0𝑙0𝑐
2
2
(6.7)
Now the system comprises 3 equations and 3 unknowns. The experimental Fpull-off for
cellulose is found above to be 4.12 µN. The slope of the experimental force-distance curves
prior to each Fpull-off is extracted using the MATLAB file in Appendix M. The distribution of
the experimental slope, s, for cellulose is plotted as a histogram in Figure 6.8. The most
frequent value of the slope is 0.36 µN/µm. These two values (s = 0.36 µN/µm and Fpull-off =
4.12 µN) are subbed into equations 6.5 and 6.6. Solving the simultaneous equations 6.5-6.7
using a MATLAB code in Appendix N gives the following values: K0 = 134 MPa, b = 1.45,
and εc = 1.65.
Figure 6.8. The slope, s, of the experimental force-distance curve is measured before each
peak event. The histogram presents the distribution of experimental values of s for cellulose.
(n = 100).
Slope, s ( N/ m)
0.0 0.4 0.8 1.2 1.6 2.0
Count
0
4
8
12
16
CHAPTER 6. RESULTS Grace Dolan, PhD Thesis
142
6.3.3 The role of hemicellulose at contacts between cellulose fibres
The role of hemicelluloses, AX and XG, on the interactions between cellulose fibres is
examined using the dip-and-drag technique on composite networks. The bacterial cellulose
was grown in liquid culture medium with the dissolved polysaccharide, which may influence
the assembly of individual fibres and the fibre network. Example force-distance curves and
histograms of the population of measured peak heights are presented in Figure 6.9 for CAX
and CXG networks. From Jarzynski’s equality, the values of Fpull-off are 3.04 µN and 1.79 µN
for CAX and CXG respectively. Comparing these values to the pull-off force for a cellulose
network (4.12 µN) indicates that AX reduces the adhesion energy between cellulose by 25%
(One way ANOVA, P-Value 0.005) and XG reduces the adhesion energy by more than 50%
(One way ANOVA, P-Value 0.001).
SEM images of the cellulose, CAX, and CXG networks in Figure 6.10 show that the presence
of hemicelluloses does not substantially change the network structure in terms of fibre
orientation and the length of fibre segments spanning two contact points. This means that
when applying the system of equations 6.5 to 6.7, the b calculated for cellulose is a good
estimate for CAX and CXG networks also.
Histograms of the experimental values of s for CAX and CXG are presented in Figure 6.11.
The slope for CXG (s = 0.15N/m) is notably less than CAX and cellulose, which is
reflected in the example force-distance curve in Figure 6.8c. From the distribution of slopes
for CAX in Figure 6.11a, the slope is taken to be the same as cellulose (s = 0.36N/m). The
experimental values for s and Fpull-off are input into equations 6.5 – 6.7 to extract K0, εc, and
A. The values are presented in Table 6.1.
From Table 6.1, K0 is lower for CXG compared to cellulose and CAX. This means that the
initial contact strength is reduced when a cellulose fibre contact is mediated by XG. From
Figure 6.10c, XG coats the surface of the cellulose fibres and is prevalent at fibre-fibre
contacts. It is reasonable to suggest that XG reduces K0 by forming a layer at the interface
between cellulose fibres that has a lower modulus (less crystalline) than the cellulose fibres
themselves. The value of εc is reduced in the presence of AX and relatively unchanged in the
presence of XG. Therefore, the strain required to initiate fibre detachment via a softening of
the contact strength is smallest for a cellulose fibre contact mediated by AX and largest for
CHAPTER 6. RESULTS Grace Dolan, PhD Thesis
143
a
b
c
d
Figure 6.9. (a) Example force-distance curve for CAX network, (b) histogram of peak
heights measured for CAX networks, (c) example force-distance curve for CXG network, (d)
histogram of peak heights measured for CXG networks.
a b C
Figure 6.10. SEM images of (a) cellulose, (b) CAX, and (c) CXG networks with a scale bar
of 1m. Images were provided by Dr. Patricia Lopez-Sanchez.
Lateral Distance (µm)
0 1 2 3 4 5 6
Late
ral F
orc
e (
µN
)
0
10
20
30
40
Peak Height (µN)
0 5 10 15 20 25 30 35
Cou
nt
0
5
10
15
20
25
Lateral Distance (µm)
0 1 2 3 4 5 6
Late
ral F
orc
e (
µN
)
0
10
20
30
40
Peak Height (µN)
0 5 10 15 20 25 30 35
Count
0
5
10
15
20
25
CHAPTER 6. RESULTS Grace Dolan, PhD Thesis
144
XG. εc is a function of the pulling distance and 𝑙0, and the effect of AX and XG structure on
𝑙0 is discussed below. The adhesion energy for CAX and CXG is substantially reduced
compared to cellulose for the same value of 𝑙0. If the adhesion energy is kept constant at the
value for a cellulose-cellulose contact (with𝑙0 = 0.3 nm), the 𝑙0for CAX and CXG is 0.6 nm
and 1 nm, respectively.
a
b
Figure 6.11. Distribution of experimental values of the slope, s, for (a) CAX and (b) CXG.
Table 6.1. Model parameters extracted from equations 6.5 - 6.7 using experimental values of
slope, s, and pull-off force, Fpull-off.
Cellulose CAX CXG
b 1.5 1. 5 1.5
K0 (MPa) 130 130 32
εc 1.7 1.2 1.8
A (J/m2)
𝑙0= 0.3nm 0.06 0.03 0.02
𝑙0= 0.6nm - 0.06 0.03
𝑙0= 1nm - - 0.06
Arabinoxylans consist of a linear chain backbone of β-D-xylopyranosyl residues linked
through (14) glycosidic linkages, with α-L-Arabinofuranosyl (Araf) residues attached to
some of the backbone27. The backbone of XG is chemically identical to cellulose with β-
(14)-linked D-glucopyranose residues28. In XG from Dicotyledons and onions, D-
xylopyranose (Xlyp) residues are attached to about 60-75% of the backbone. AX and XG
Slope, s ( N/ m)
0.0 0.4 0.8 1.2 1.6 2.0
Count
0
2
4
6
8
10
Slope, s ( N/ m)
0.0 0.4 0.8 1.2 1.6 2.0
Count
0
4
8
12
16
20
CHAPTER 6. RESULTS Grace Dolan, PhD Thesis
145
chains on the surface of cellulose fibres are thus expected to influence the contact length, 𝑙0.
At a minimum, one glucose residue (0.5 nm)28 and one Xlyp residue (0.5 nm)29 is expected to
be separating cellulose fibres mediated by XG, giving a value of 1nm for 𝑙0 which is in good
agreement with Table 6.1. The relative amount and sequence distribution of side chain
residues of AX molecules vary depending on the source27. At a minimum, a single Xlyp
residue is expected to separate cellulose fibres mediated by AX. This is consistent with the
value of 𝑙0 around 0.6nm in Table 6.1. The equilibrium lengths of the cellulose fibre contacts
mediated by the different polymers are illustrated in Figure 6.12. The results suggest that the
adhesion energy between the cellulose fibres remains roughly constant in each case, with the
role of AX and XG instead being to change the equilibrium contact length. The contact is
modelled in ComsolTM Multiphysics as effective springs where the spring constant decays
with strain. This is equivalent to the surface energy potential decaying with separation
distance. Thus the presence of AX and XG increases the separation distance and
consequently decreases the surface interaction between cellulose fibres.
The values of the pull-off force between individual cellulose fibres in the presence of AX and
XG are consistent with the macro-scale network mechanics. The cellulose and CAX networks
Figure 6.12. Depiction of the contact between two cellulose fibres and two cellulose fibres
mediated by arabinoxylan or xyloglucan with an estimate of the respective equilibrium
contact length, 𝑙0.
behave similarly under uni-axial tension30 whereas the CXG network displays decreased
stiffness and increased extensibility10. Gu et al.31 speculate that during the biosynthesis of
CHAPTER 6. RESULTS Grace Dolan, PhD Thesis
146
bacterial cellulose the adsorption of XG onto the cellulose surface reduces the number of
network entanglements. However, the results from the dip-and-drag technique strongly
suggest that XG changes the interaction between cellulose fibres and reduces the modulus of
the network by reducing the adhesive force between fibres. One possible explanation for the
reduced pull-off force between cellulose fibres in the presence of XG is that the
polysaccharide forms an amorphous layer sandwiched between the cellulose fibres that leads
to a reduced contact modulus; if this is the case, a smaller force is required to separate the
fibres. In Chapter 4: Section 4.3.6 I postulate that the static friction between two cellulose
hydrogel surfaces is driven by the adhesion between individual cellulose fibres at the
interface. The static friction between pairs of cellulose hydrogels is shown to be reduced by
approximately half in the presence of XG. This follows the close to 50% reduction in the
pull-off force between individual cellulose fibres in the presence of XG.
6.4 Concluding Remarks
The Dip-and-drag technique is successfully applied to extract information about the
interactive forces between cellulose fibres, and the influence of important cell wall
components at these contacts. The measured peaks in lateral force-distance curves are
interpreted as representing fibre-fibre detachment events, where the Jarzynski’s average peak
height is taken as a pull-off force. The sensitivity of this pull-off force to uncontrolled
variables in the fibre network is investigated using ComsolTM Multiphysics to simulate fibre-
fibre detachment based on a cohesive model. The variables include; network structure,
equilibrium contact length, contact strength, and critical strain. The model indicates that XG
reduces the pull-off force by decreasing the contact strength and increasing the critical strain.
This result is consistent with the increased extensibility of macro-scale CXG networks under
tensile testing, and the reduced static friction between CXG networks measured in Chapter 4:
Section 4.3.6. Based on the ComsolTM Multiphysics model and adhesion energy equation, the
presence of XG and AX increases the separation between cellulose fibres to distances that are
in good agreement with the structure (residue lengths) of the molecules. Insights into the
interactions between cellulose, AX, and XG are critical for building a 3-D mechanical model
of the plant cell wall structure and assembly.
CHAPTER 6. RESULTS Grace Dolan, PhD Thesis
147
References for Chapter 6
1. K. Kulasinski, S. Keten, S. V. Churakov, D. Derome and J. Carmeliet, Cellulose,
2014, 21, 1103-1116.
2. A. N. Fernandes, L. H. Thomas, C. M. Altaner, P. Callow, V. T. Forsyth, D. C.
Apperley, C. J. Kennedy and M. C. Jarvis, Proceedings of the National Academy of
Sciences of the United States of America, 2011, 108, E1195-E1203.
3. M. Martinez-Sanz, P. Lopez-Sanchez, M. J. Gidley and E. P. Gilbert, Cellulose, 2015,
22, 1541-1563.
4. T. Zhang, S. Mahgsoudy-Louyeh, B. Tittmann and D. J. Cosgrove, Cellulose, 2014,
21, 853-862.
5. Y. B. Park and D. J. Cosgrove, Plant Physiology, 2012, 158, 1933-1943.
6. B. Zhao and H. J. Kwon, Journal of Adhesion Science and Technology, 2011, 25,
557-579.
7. K. Baba, Y. Sone, A. Misaki and T. Hayashi, Plant and Cell Physiology, 1994, 35,
439-444.
8. M. Hrmova, V. Farkas, J. Lahnstein and G. B. Fincher, Journal of Biological
Chemistry, 2007, 282, 12951-12962.
9. M. McNeil, P. Albersheim, L. Taiz and R. L. Jones, Plant Physiology, 1975, 55, 64-
68.
10. S. E. C. Whitney, M. G. E. Gothard, J. T. Mitchell and M. J. Gidley, Plant
Physiology, 1999, 121, 657-663.
11. E. Chanliaud, K. M. Burrows, G. Jeronimidis and M. J. Gidley, Planta, 2002, 215,
989-996.
12. P. Lopez-Sanchez, M. Rincon, D. Wang, S. Brulhart, J. R. Stokes and M. J. Gidley,
Biomacromolecules, 2014, 15, 2274-2284.
13. G. Guhados, W. K. Wan and J. L. Hutter, Langmuir, 2005, 21, 6642-6646.
14. S. R. Andersson and A. Rasmuson, J. Pulp Pap. Sci., 1997, 23, J5-J11.
15. F. Huang, K. C. Li and A. Kulachenko, Journal of Materials Science, 2009, 44, 3770-
3776.
16. A. Carambassis and M. W. Rutland, Langmuir, 1999, 15, 5584-5590.
17. J. Stiernstedt, H. Brumer, III, Q. Zhou, T. T. Teeri and M. W. Rutland,
Biomacromolecules, 2006, 7, 2147-2153.
CHAPTER 6. RESULTS Grace Dolan, PhD Thesis
148
18. S. M. Notley, M. Eriksson, L. Wagberg, S. Beck and D. G. Gray, Langmuir, 2006, 22,
3154-3160.
19. R. Nigmatullin, R. Lovitt, C. Wright, M. Linder, T. Nakari-Setala and A. Gama,
Colloids and Surfaces B-Biointerfaces, 2004, 35, 125-135.
20. J. Stiernstedt, N. Nordgren, L. Wagberg, H. Brumer, D. G. Gray and M. W. Rutland,
Journal of Colloid and Interface Science, 2006, 303, 117-123.
21. S. Zauscher and D. J. Klingenberg, Colloids and Surfaces a-Physicochemical and
Engineering Aspects, 2001, 178, 213-229.
22. K. Park and G. H. Paulino, Applied Mechanics Reviews, 2011, 64.
23. S.-Y. Ding, S. Zhao and Y. Zeng, Cellulose, 2014, 21, 863-871.
24. C. Brett and K. Waldron, Physiology and biochemistry of plant cell walls, 1990.
25. V. L. Finkenstadt, T. L. Hendrixson and R. P. Millane, Journal of Carbohydrate
Chemistry, 1995, 14, 601-611.
26. S. Y. Sheu, D. Y. Yang, H. L. Selzle and E. W. Schlag, Proceedings of the National
Academy of Sciences of the United States of America, 2003, 100, 12683-12687.
27. M. S. Izydorczyk and J. E. Dexter, Food Research International, 2008, 41, 850-868.
28. S. C. Fry, Journal of Experimental Botany, 1989, 40, 1-11.
29. G. Dervilly-Pinel, J. F. Thibault and L. Saulnier, Carbohydrate Research, 2001, 330,
365-372.
30. D. Mikkelsen, B. M. Flanagan, S. M. Wilson, A. Bacic and M. J. Gidley,
Biomacromolecules, 2015, 16, 1232-1239.
31. J. Gu and J. M. Catchmark, Cellulose, 2014, 21, 275-289.
149
Chapter 7
The effect of bacterial expansins on
cellulose fibre interactions
7.1 Introduction
The key hypothesis in this thesis is that plant growth is influenced by the interactive forces
(including friction) occurring between cellulose fibres and between cells. Uncovering
potential mechanisms for these interactions provides a framework to include in multi-scale
mechanical models of the plant cell wall. A significant step towards building a multi-scale
mechanical model of plant growth is elucidating the activity of expansins. Expansin is a
protein that is implicated in initiating cell wall elongation. In Chapter 2: Section 2.2.3, I
review expansins in terms of their binding specificity and subsequent effect on the
mechanical and structural properties of plant cell walls and other cellulose substrates.
To date, the most probable mechanism of action of expansins is the disruption of cellulose
fibre contacts to promoting slippage; this is represented pictorially in Figure 7.1. Firstly,
expansins bind to cellulose and are purported to slide along the length of the cellulose fibre.
Secondly, a bound expansin is able to locally disrupt cellulose fibre contact. The final step is
that cellulose fibres and surrounding matrix polysaccharides are able to slide relative to each
other. Recent advances in our understanding of how expansins are able to achieve steps 1 and
2 in Figure 7.1 are discussed in Section 7.2. Due to the important role of expansins in cell
wall remodelling, they are required in almost all plant physiological development aspects
from germination to fruiting1. Marowa et al.1 state that “expansins, combined with other
(breeding) tools can be useful in manipulating many plant physiological aspects such as
CHAPTER 7. RESULTS Grace Dolan, PhD Thesis
150
germination, stem development, yield and plant’s ability to withstand biotic and abiotic
stress”. In order to use expansins for crop improvement, it is important to be able to predict
how the molecular interactions between expansins and the plant cell wall influence the cell
wall mechanics. Chapter 4 and Chapter 6 highlight the capacity of the tribo-rheological and
dip-and drag techniques to measure the interactions between cellulose fibres, AX, and XG.
The goal of this chapter is to use both techniques to investigate the effect of expansins on
cellulose fibre interactions, and at cellulose fibre contacts mediated by AX and XG.
Figure 7.1. (1) Expansin (red pacman) binds to cellulose fibres (brown lines) and slides along
the length of the fibres as shown by the arrow. (2) Expansins disrupt cellulose fibre contact
points. In Chapter 2: Section 2.2.3 there are a number of hypotheses for the action of
expansin that is depicted by the lightning bolt in this Figure. (3) Disruption at cellulose fibre
contacts leads to relative movement of fibres, as shown by the arrows, and elongation of the
cell wall.
7.2 Background on the ‘wall-loosening’ activity of expansins
In Figure 7.2, expansins are shown to consist of two compact domains (D1 and D2). There
are highly conserved surface residues spanning across both domains that make up what is
referred to as the putative polysaccharide binding site (PPBS)2. It is difficult to assess the
function of the two domains and the residues at the PPBS due to the poor heterologous
expression of plant expansins. Thankfully, bacterial expansin (YOAJ) has high structural
similarity to plant expansins and can be easily expressed in Escherichia coli to generate
protein variants (mutants)2. Key amino acid residues are modified by site-directed
mutagenesis to assess their role in the binding and wall loosening activity of the protein.
CHAPTER 7. RESULTS Grace Dolan, PhD Thesis
151
Georgelis et al.2 use two mechanical assays for expansin activity; the breaking force of filter
paper in Figure 7.3, and the creep rate of alkali–pre-treated wheat coleoptiles in Figure 7.4.
Figure 7.2. Protein structure of expansin, showing two distinct domains (D1 and D2), that is
interacting with a cellulose elementary fibril. Image is reproduced from Silveira and Skaf3.
Figure 7.3. Activities of selected variants of bacterial expansin protein on Whatman filter
paper measured as a breaking force. The breaking force of the control (buffer with no
expansin) and native expansin (labelled EXLX1 and equivalent to YOAJ) are marked with
horizontal lines for comparison. Reproduced from Georgelis et al.2.
The substantial decrease in the breaking force of filter paper when treated with YOAJ in
Figure 7.3 suggests that bare cellulose is a key target for wall loosening activity. The activity
of various mutant expansins in Figure 7.3 and 7.4 is found to be closely linked to their
CHAPTER 7. RESULTS Grace Dolan, PhD Thesis
152
measured binding affinity to cellulose substrates (cotton fibres, Avicel, and filter paper)2. Key
protein variants that assess the role of binding on wall loosening activity include:
• WWY (labelled triple in Figure 7.3 and W125A-W126A-Y157A in Figure 7.4) –
substitution of the three aromatic amino acid residues on D2 that are part of the PPBS.
• D82A – substitution of the Asp-82 residue on D1 which is part of the PPBS.
• RKKQ (labelled R173Q-K180Q-K183Q in Figure 7.4) – three positively charged
residues in D2 are replaced by neutral glutamine.
Figure 7.4. Activities of variants of bacterial expansin protein on alkali-pretreated wheat
coleoptile measured as a relative creep rate. The activity of the native expansin (YOAJ) is
marked with a horizontal line at relative creep rate of 1. Reproduced from Georgelis et al.2.
The WWY and D82A mutants show no activity for either of the assays in Figure 7.3 and 7.4.
On the other hand, the RKKQ mutant is shown to increase the creep rate of a cell wall
substrate. The replacement of 3 positive residues to neutral ones reduces the interaction
between the RKKQ expansin and pectin in the cell wall, and thus allows the expansin to
CHAPTER 7. RESULTS Grace Dolan, PhD Thesis
153
diffuse through the wall and bind to cellulose more freely2. This is supported by the results of
a binding assay for expansin variants in which 1, 2, or 3 positively charged residues are
replaced with neutral ones. The variants’ binding to wheat coleoptile cell walls decreased
proportionally to the number of positive charges that were replaced.
According to Georgelis et al.2, the YOAJ structure (Protein Data Bank code 3D30) reveals
that the side chains of the three residues mutated in WWY are twisted in a clockwise
direction, which suggests that D2 binds to a single glucan chain on the surface of the
cellulose fibre. The authors postulate that YOAJ “may induce wall creep when these residues
bind a glucan that is part of the load-bearing network in the cell wall, distorting its shape and
allowing physical slippage at the junction if the wall is in tension”2. This interpretation of the
structure of the PPBS is supported by Silveira and Skaf3 who use molecular dynamics
simulations to show that YOAJ can “hydrogen bond a free glucan chain in a twisted
conformation and that the twisting is chiefly induced by means of residue Asp82 located on
D2”. Furthermore, their results suggest that YOAJ can slide along the hydrophobic surface of
crystalline cellulose via D2 and simultaneously disrupt hydrogen bonds by twisting glucan
chains through D13.
The structure of the cellulose used for mechanical assays is critical because the action of
expansin is likely to be through the binding to a single glucan chain rather than the flat,
highly crystalline cellulose surface. Part of the amorphous or paracrytalline phases of
cellulose may be removed for substrates that have been mechanically or chemically treated,
such as Avicel and filter paper. Chapter 6: Section 6.1 compares the structure of plant and
bacterial cellulose fibres and highlight that bacterial cellulose is a good model system for
plant cellulose as it possesses the same paracrystalline surface layer surrounding the
crystalline core. Moreover, in this thesis I have developed two techniques that measure the
interaction between bacterial cellulose fibres. The first is the tribo-rheological technique that
is detailed in Chapter 3: Section 3.2.1 and measures the static friction between two cellulose
hydrogels that is driven by cellulose fibres interacting at the interface. The second is the Dip-
and-drag technique in Chapter 3: Section 3.2.2 which is applied to measure the pull-off force
between individual bacterial cellulose fibres and at contacts mediated by AX and XG. In this
chapter, both of the techniques are used as mechanical assays to gain insight into the activity
of bacterial expansins on bacterial cellulose substrates. The native expansin and the protein
variants denoted WWY, D82, and RKKQ, are included in this study for a comprehensive
CHAPTER 7. RESULTS Grace Dolan, PhD Thesis
154
structure-function analysis. Key findings from the bacterial system provide insights into the
mechanism of action of plant expansins during plant growth.
7.3 Experimental Section
7.3.1 Pre-treatment of bacterial cellulose hydrogels
Bacterial cellulose hydrogels are grown by the Cosgrove lab at Pennsylvania State University
using a similar procedure to that described in Chapter 3: Section 3.1, and cultivated in 12-
well plates to give 25mm diameter hydrogels. The harvested hydrogels are then boiled in
deionised water for 1 hour and rinsed repeatedly to ensure all bacteria is removed from the
cellulose network and is not available as a substrate for expansins. Due to the high water
content (> 99%), the average weight of the hydrogels is measured and taken, based on
density, as the volume of water (~ 2 mL). The native and variant expansins are stored as
concentrated solutions (order of 10 g/L) in 20mM Hepes buffer (pH 7.5). Therefore, the
large volume of water in the hydrogels means that an impractically small volume of the stock
solution is required to be added to give a working concentration of expansin in 20 mM Hepes
buffer (pH 7.5) of 200 g/mL. Pipetting a small volume of concentrated expansin solution
directly onto the hydrogel surface means that it would take a long time for the expansin to
diffuse and reach a constant concentration throughout the hydrogel. Therefore, the bacterial
cellulose hydrogels are compressed to remove a fraction of the water so that the expansin
solution can be diluted and added to the hydrogel in larger volumes. Four hydrogels were
placed between 2 perspex plates under a load of 70 g for 4 hours. The compressed hydrogels
weigh ~ 400 mg, so the volume of water in the hydrogels is ~0.4mL. The compressed
hydrogels are then placed in a 12-well plate, one hydrogel per well. The stock solutions of
expansin are accordingly diluted up to a volume of 1.6 mL (uncompressed – compressed
volume) to give a final concentration of expansin in 20mM Hepes buffer (pH 7.5) of 200
µg/mL once added to the compressed hydrogels. The larger volume of diluted expansin
solution is added to the wells to imbibe the hydrogel. The 12-well plate with the lid on is
wrapped in Parafilm to prevent evaporation, and left on a see-saw platform shaker in a 4°C
cold room overnight. The hydrogels are observed to re-swell to approximately 75% of the
initial hydrogel thickness (prior to compression) as the surrounding expansin solution is
absorbed.
CHAPTER 7. RESULTS Grace Dolan, PhD Thesis
155
7.3.2 Mechanical assay of expansin activity using the tribo-rheological
technique
The tribo-rheological technique is carried out using a Discovery Hybrid Rheometer HR-3
(TA Instruments) with a 25mm parallel plate and the bottom peltier plate. 25mm discs of
emery paper are attached to the plates with double sided tape and the hydrogels are glued to
the emry with cyanoacrylate glue. The procedure outline in Chapter 3: Section 3.2.1 is
followed, with the only modification being an axial ramp distance of 250 µm rather than 500
µm. For these experiments the hydrogels are not surrounded by water because the
concentration gradient would drive expansins to diffuse out of the hydrogel. However, during
compression fluid is squeezed out of the gels and surrounds the interface, and evaporation is
not expected to change the concentration of expansins within the measurement time. All
expansins (YOAJ, WWY, D82, and RKKQ) treatments are tested on different cellulose
hydrogel pairs in triplicate, and a control hydrogel pair treated with buffer and no expansin is
tested on each day of experimentation.
7.3.3 Mechanical assay of expansin activity using the Dip-and-drag
technique
The dip-and-drag technique is applied to bacterial cellulose networks as described in Chapter
3: Section 3.2.2. The only deviation from these detailed steps is that instead of water, the
measurements are carried out in either 20 mM Hepes buffer (pH 7.5) for the control, or 200
g/mL of YOAJ in 20 mM Hepes buffer (pH 7.5).
7.4 Results
7.4.1 The effect of bacterial expansins on the mechanics of bacterial
cellulose hydrogels
The first step of the tribo-rheological technique is to compress the hydrogel pair to a constant
axial strain, which is calculated as a compression ratio and denoted CR. The volume of the
hydrogel is decreased during compression, leading to an increase in the interstitial fluid
pressure that drives the flow of water out of the network. Thus the mechanics of the hydrogel
during compression is determined by the poroelasticity, or the interplay between the
interstitial fluid and the solid matrix as discussed in Chapter 4: Section 4.3.1. In Figures 7.5a
CHAPTER 7. RESULTS Grace Dolan, PhD Thesis
156
and 7.5b, the compressive mechanics of cellulose hydrogels treated with native (YOAJ) and
mutant (WWY, RKKQ, D82) expansins are presented with the control (buffer, no expansins)
hydrogels grown from the same batch of fermentation medium. In Figures 7.5c and 7.5d, the
measured normal stress during compression is divided by the peak value for a given sample
from Figures 7.5a and 7.5b so that the shape of the curves can be accurately compared. There
is no observable difference in the compression behaviour, which indicates that expansins
don’t influence the poroelasticity of the cellulose network. In Chapter 4: Section 4.3.2 I show
that the area of contact between hydrogels is determined by their axial modulus. From Figure
7.5, the axial modulus of all samples is the same, so the true contact area for all samples is the
same and the results can be directly compared to investigate the effects of native and mutant
expansins on the surface interactions. After the hydrogel pairs are brought into compressive
contact and the normal stress is allowed to relax to its equilibrium value, the viscoelastic
moduli of the system are measured using SAOS. The G’ and G” measured at 1 Pa and 1Hz
for different CR are presented in Tables 7.1 and 7.2. Table 7.1 includes the results for a
hydrogel pair treated with YOAJ expansin and an untreated hydrogel pair from the same
formation batch, compressed through a range of CRs. Table 7.2 includes the results for
hydrogel pairs treated with WWY, RKKQ, D82, and an untreated hydrogel pair from the
same formation batch, compressed through a range of CRs. Each treatment (YOAJ, WWY,
RKKQ, D82) is tested in triplicate and the results for only one replicate are included in
Tables 7.1 and 7.2. From the results in Tables 7.1 and 7.2 there is no substantial difference in
the G’ and G” between the hydrogels treated with expansins and the untreated controls, above
what could be attributed to variation between sample replicates.
In Chapter 4: Section 4.3.1 I show that compression leads to the formation of new fibre
contacts within the cellulose network. Furthermore, I show that G’ scales with cellulose
concentration, and postulate that an increase in the cellulose concentration directly increases
the probability that fibres in the network will come into contact and adhere together. From the
results in Table 7.1 and 7.2, expansins do not appear to influence G’ and therefore are not
expected to effect the cellulose network architecture. Note, the hydrogels are pre-compressed
under a fixed weight before re-swelling in expansin solution. Therefore, some cellulose fibre
contacts may be formed during the pre-compression step in the absence of expansin. After the
pre-compression step, expansins are added and the hydrogels are compressed again before the
G’ is measured. Expansin are shown to have negligible influence on the compression
CHAPTER 7. RESULTS Grace Dolan, PhD Thesis
157
mechanics in Figure 7.5. Thus it is expected that the cellulose network structure after the pre-
compression step dominates the G’ measured in Tables 7.1 and 7.2.
a
b
c
d
Figure 7.5. Compression curves of (a) cellulose hydrogel pair treated with the native (YOAJ)
expansin and, (b) cellulose hydrogel pairs treated with mutant expansins (WWY, RKKQ, and
D82). In (a) and (b) the compression curve for untreated hydrogels from the same batch of
fermentation medium is included as the control. (c) The compression curves in (a) are
normalised against their peak normal stress value. On the y-axis the measured normal stress
is divided by the peak value of the respective curve from (a). (d) The compression curves in
(b) are normalised against their peak normal stress value. On the y-axis the measured normal
stress is divided by the peak value of the respective curve from (b).
Time (s)
0 2 4 6 8
Norm
al S
tress (
kP
a)
0
2
4
6
ControlYOAJ
Time (s)
0 2 4 6 8
Norm
al S
tress (
kP
a)
0
2
4
6
Mutants ControlWWYRKKQD82
Time (s)
0 2 4 6 8
Norm
al S
tress/P
eak N
orm
al S
tress
0.0
0.2
0.4
0.6
0.8
1.0ControlYOAJ
Time (s)
0 2 4 6 8
No
rma
l S
tre
ss/P
ea
k N
orm
al S
tre
ss
0.0
0.2
0.4
0.6
0.8
1.0
Mutants ControlWWYRKKQD82
CHAPTER 7. RESULTS Grace Dolan, PhD Thesis
158
Table 7.1. The storage (G’) and loss (G”) moduli of hydrogel pairs that are untreated or
treated with YOAJ expansin are measured using SAOS at 1 Pa and 1 Hz. G’ and G” are
measured across a range of CR.
CR Control (no expansins) YOAJ
G’ G” G’ G”
0.08 380 ± 6 50± 0.4 310 ± 4 50 ± 0.2
0.15 420 ± 3 50± 0.7 340 ± 2 50 ± 0.5
0.23 430 ± 2 50± 0.6 390 ± 2 60 ± 0.7
0.31 480 ± 4 60± 0.5 470 ± 3 70 ± 0.6
0.38 550 ± 3 70± 1 590 ± 4 90 ± 1
0.46 680 ± 3 90± 0.7 770 ± 5 120 ± 1
0.54 910 ± 5 130± 3 1100 ± 7 180 ± 2
Table 7.2. The storage (G’) and loss (G”) moduli of hydrogel pairs that are untreated or
treated with expansin variants (WWY, RKKQ, D82) are measured using SAOS at 1 Pa and 1
Hz. G’ and G” are measured across a range of CR.
CR Control (no
expansins)
WWY RKKQ D82
G’ G” G’ G” G’ G” G’ G”
0.1 210 ± 3 40 ± 0.4
0.2 340 ± 3 50 ± 1 260 ± 3 40 ± 1 320 ± 6 50 ± 0.4
0.3 480 ± 3 80 ± 1 520 ± 3 80 ± 1 460 ± 4 90 ± 1 650 ± 7 100 ± 1
0.4 810 ± 5 130 ± 1 810 ± 4 120 ± 2 730 ± 5 140 ± 1 930 ± 9 140 ± 2
0.5 1400 ± 7 230 ± 2 1300 ± 9 190 ± 4 1000 ± 7 200 ± 2 1400 ± 12 200 ± 2
0.6 2500 ± 12 380 ± 6 2000 ± 15 380 ± 8 2300 ± 18 340 ± 7
7.4.2 The effect of bacterial expansins on the friction response between
pairs of bacterial cellulose hydrogels
After the compression and SAOS steps the hydrogel pairs are sheared at a constant rotation
rate. For approximately equivalent G’, which is a measure of cellulose concentration, the
shear stress curves in the presence of expansins are compared in Figure 7.6. The initial slope
of the curves show non-linear behaviour, but the shape of the curves for expansin treated
hydrogels is similar to the respective control. Thus the shear mechanics of the system appear
to be unaffected by expansins. The shapes of the curves also display either stick-slip or
CHAPTER 7. RESULTS Grace Dolan, PhD Thesis
159
stiction behaviour as discussed in Chapter 4: Sections 4.3.3 and 4.3.7. The stick-slip peaks
are less pronounced and consistent in Figure 7.6 compared to those in Figure 4.26a. In
Section 4.3.7, the parameters of the stick-slip peaks are shown to be related to the ability of
the rheometer to control the angular velocity around the set point. The Haake Mars III
rheometer used in Chapter 4 is a controlled stress machine, whereas the Discovery HR-3
rheometer used in this chapter can operate as a controlled stress or controlled strain machine.
The Discovery HR-3 machine in controlled strain mode is expected to be able to control the
angular velocity at the set point better than the Haake Mars III, which explains the less
pronounced stick-slip response in Figure 7.6 compared to Figure 4.26a. During the stick cycle
the cellulose fibres at the interface between the hydrogels are purported to adhere together
and during the slip cycle it is expected that the fibres are separated. Stiction behaviour is
observed for CXG composite hydrogels in Section 4.3.3, and is proposed to be due to the XG
coating the surface of the cellulose fibres reducing re-adhesion between cellulose fibres at the
interface under dynamic conditions. With this interpretation in mind, the results in Figure 7.6
indicate that the presence of expansins (YOAJ, RKKQ, and D82) reduces re-adhesion of
cellulose fibres at the interface during sliding.
a
b
Figure 7.6. Shear stress over time for a constant rotation rate for the (left) native and (right)
variant expansin proteins with the corresponding controls for samples with comparable G’
(cellulose concentration).
The interfacial yield stress is taken as the peak in the shear stress curve measured during the
constant rotation rate step of the tribo-rheological procedure, as in Chapter 4: Section 4.3.3.
In Figure 7.7, the interfacial yield stress is plotted against G’ for representative cellulose
Strain (-)
0.0 0.5 1.0 1.5
Shear
Str
ess (
Pa)
0
50
100
150
ControlYOAJ
Strain (-)
0.0 0.5 1.0 1.5
Shear
Str
ess (
Pa)
0
50
100
150
ControlWWYRKKQD82
CHAPTER 7. RESULTS Grace Dolan, PhD Thesis
160
hydrogels treated with YOAJ, WWY, RKKW, D82, and controls. The interfacial yield stress
and G’ for the triplicate samples of each treatment are included in Appendix O. The control
data is fitted with a power law regression with exponent 0.8, which is equivalent to the linear
scaling of the interfacial yield stress at the true contact area with cellulose concentration
based on Chapter 4: Figures 4.19 and 4.24. For an equivalent G’ (cellulose concentration),
the presence of expansins reduces the interfacial yield stress between hydrogels. This means
that expansins either reduce the probability that fibres at the interface form adhesive contacts,
or reduce the adhesive force between cellulose fibres interacting at the interface.
a
b
Figure 7.7. Interfacial yield stress against G’ for pairs of bacterial cellulose hydrogels treated
with (a) native (YOAJ) and (b) mutant expansins (WWY, RKKQ, D82). The data for the
respective control (buffer, no expansins) samples is included in each graph. Each data set is
for a single pair of hydrogels that is representative of the behaviour of the replicates. The line
in each graph is a power law regression fitted to the control data with a fixed exponent of 0.8.
The control data in Figure 7.7a is interpolated to approximate the interfacial yield stress at the
same discrete values of G’ as the YOAJ data. The measured interfacial yield stress (τm) for
YOAJ treated samples is then divided by the interfacial yield stress approximated for the
control (τcontrol) at the same G’, to give a relative interfacial yield stress (IYSR) as in Equation
7.1. This process is repeated for the WWY, RKKW, and D82 using the control data in Figure
7.7b. The IYSR values for each expansin are averaged across the triplicate samples and range
of CR, and are presented in Figure 7.8. All expansins appear to reduce the interfacial yield
stress compared to the respective control, as shown by an IYSR is less than 1. Based on a
One-way ANOVA test, YOAJ, RKKQ, and D82 give a statistically significant decrease in the
interfacial yield stress compared to the control (marked with an * in Figure 7.8). Importantly,
G' (kPa)
1
Inte
rfacia
l Y
ield
Str
ess (
Pa)
100
ControlYOAJ
G' (kPa)
0.1 1
Inte
rfa
cia
l Y
ield
Str
ess (
Pa
)
10
100
ControlWWYRKKQD82
CHAPTER 7. RESULTS Grace Dolan, PhD Thesis
161
the results presented in Figure 7.8 highlight that the activity of YOAJ, RKKQ, and D82 lead
to a decrease in the apparent static friction between cellulose hydrogels that is not due to
relative differences in the G’. Thus, these expansins are expected to influence the interfacial
properties by reducing the probability that fibre contacts form at the interface, or by reducing
the adhesion between fibre contacts at the interface. It is noted here that putatively effective
(YOAJ, RKKQ) and ineffective (D82) expansins both have an effect and there is no clear
pattern in their activity.
𝐼𝑌𝑆𝑅(𝐺′) =𝜏𝑚(𝐺
′)
𝜏𝑐𝑜𝑛𝑡𝑟𝑜𝑙(𝐺′)
(7.1)
Figure 7.8. IYSR is the interfacial yield stress measured in the presence of expansin (YOAJ,
WWY, RKKQ, D82) divided by the interfacial yield stress for the control sample at the same
G’, which is approximated by interpolating the measured control data. The results are
averaged across the triplicate samples and the range of CR and the standard deviation is
included as an error bar. * denotes the samples for which the measured interfacial yield stress
is statistically different from the control from a One-Way ANOVA test with a significance
value of p < 0.05.
7.4.3 The effect of bacterial expansins on adhesion between cellulose fibres
In Chapter 6: Section 6.3.3 the Dip-and-drag technique is used to measure the adhesion
between cellulose fibres mediated by XG. The presence of XG is found to reduce the
adhesion between cellulose fibres by approximately half. The same experiments are carried
out on cellulose and composite (CAX, CXG) networks in the presence of YOAJ expansin,
following the method in Section 7.2.3. The pull-off force between fibres is calculated from
CHAPTER 7. RESULTS Grace Dolan, PhD Thesis
162
the Jarzynski’s average of the distribution of peaks as described in Chapter 5: Section 5.3.1.
The pull-off force in the presence of YOAJ is divided by the value in the absence of
expansins to give a relative pull-off force, and is presented in Figure 7.9. Raw data from the
dip-and-drag experiments on Cellulose, CAX and CXG networks with and without YOAJ
expansin are included in Appendix P. The histograms of peak heights from the raw data for
Cellulose, CAX and CXG networks with and without YOAJ expansin are included in
Appendix Q. YOAJ expansin is observed to have no activity in reducing the adhesion
between cellulose fibres or a cellulose fibres contact mediated by AX, whereas a small
decrease in the pull-off force is observed in the presence of YOAJ for CXG. In Figure 7.9,
the present of YOAJ seems to increase the pull-off force measured for a cellulose network. It
is unclear why this would be the case and complex statistical analysis of the distribution of
peak heights may be more revealing than the Jarzynski’s averaging method from Chapter 5:
Section 5.3.1. However, changing the type of analysis would require validation using model
systems similar to the level of detail in Chapter 5 and this is outside of the scope of this
thesis. One-way ANOVA of the population of the different networks (cellulose, CAX, and
CXG) are compared in the presence and absence of YOAJ, and no statistical difference in the
means is observed (with significance value of 0.05).
From the results in Figure 7.9, there is no evidence that YOAJ expansin reduces the adhesion
between cellulose fibres. This supports the mechanism of action whereby YOAJ reduces the
probability that cellulose fibres form contacts. However, there are limitations with comparing
the results from the dip-and-drag technique to the tribo-rheological technique for measuring
the friction between hydrogels. The dip-and-drag technique investigates the effect of
expansin on cellulose fibre contacts in a preformed network, whereas the tribo-rheological
technique measures the influence of expansins on forming new cellulose contacts. Therefore,
expansins may influence the adhesion between contacts as they are forming but not after they
are already formed. Additionally, as part of the preparation of substrates for the dip-and-drag
technique, the cellulose network is air-dried before gluing to the glass substrate. The moisture
level in the network after air drying is unknown. Depending on the level of drying, the
structure of the normally hydrated paracrystalline surface layer of cellulose fibres may be
affected, consequently reducing the ability of expansins to bind to the fibres. Expansin
activity has been measured on filter paper substrates which are dried during processing,
although the cellulose material is also subjected to other chemical and mechanical processes
that are likely to change the nature of cellulose fibre contacts. This filter paper result suggests
CHAPTER 7. RESULTS Grace Dolan, PhD Thesis
163
that the paracrystalline surface layer may not be crucial for the binding of expansins to
cellulose fibres. Thus it is unclear why there is no observable effect of YOAJ on the pull-off
force between cellulose fibres measured using the dip-and-drag technique.
Figure 7.9. The pull-off force is calculated from the Jarzynski’s average of the population of
experimentally measured peak heights in the force-distance curves from the dip-and-drag
technique. The pull-off force in the presence of YOAJ is normalised against the value in the
absence of YOAJ for cellulose (C), CAX, and CXG networks.
7.5 Discussion
The WWY mutant does not bind to cellulose2 and has no significant effect on the interfacial
yield stress in Figure 7.8. Figure 7.8 indicates that D82 has the highest activity followed by
YOAJ and RKKQ, and all of these expansins reduce the interfacial yield stress such that it is
statistically different from the control. This is an unexpected result as YOAJ and RKKQ are
putatively effective and D82 is a putatively ineffective expansin, yet this does not appear as a
clear trend in these experimental outcomes. Due to the low concentration of expansins
required for the observed activity, it is improbable that there would be sufficient surface
coverage of expansins bound to cellulose to directly prevent fibre contact. Silveira et al.3
suggest that D2 allows the expansin protein to slide along cellulose fibres as D1 introduces a
structural defect in one of the glucan chains on the surface. D82 has a mutation on D1.
However, the functioning D2 may permit the expansin to rapidly move along the cellulose
fibres and reduce the probability that the fibres form contacts at the interface, thus reducing
the interfacial yield stress. When both D1 and D2 are functioning, as for YOAJ and RKKQ,
C CAX CXG
Rela
tive p
ull-
off forc
e
0.0
0.2
0.4
0.6
0.8
1.0
1.2
CHAPTER 7. RESULTS Grace Dolan, PhD Thesis
164
the surface glucan chain that has undergone a conformational change is expected to behave
similarly to a XG chain mediating a cellulose fibre contact. In Chapter 6: Section 6.3.3, XG
is shown to reduce the adhesion between cellulose fibres by reducing the contact strength and
increasing the separation between the cellulose fibres in contact. By inducing a 90-degree
twist in a surface glucan chain3, I propose that expansins increase the separation distance
between cellulose fibres in contact, compared to when the surface chain is in planar
conformation.
7.6 Concluding Remarks
The activity of native expansin (YOAJ) and protein variants (WWY, RKKQ, D82) on
bacterial cellulose hydrogels is measured using two mechanical assays: the tribo-rheological
technique in Chapter 4, and the dip-and-drag technique in Chapter 6. The WWY variant
shows no activity for either assays, which is attributed to a mutation of the binding site that
prevents binding to cellulose. YOAJ, RKKQ, and D82 reduce the surface interaction between
cellulose hydrogels measured using the tribo-rheological technique.
Previous mechanical assays of these expansin’s activity include creep tests and tensile tests2.
Based on these tests the expansins are described as having ‘loosening’ or ‘weakening’
activities, however the exact mechanistic detail is unclear. The advantage of the tribo-
rheological test as an assay is that, through the computation model in Chapter 4: Section
4.2.3, the measured results can be interpreted in terms of what is physically occurring at the
interface. The results from the tribo-rheological test lead to two clear hypotheses for the
activity of expansins.
(1) Prevent the formation of new contact between cellulose fibres.
(2) Reduce the adhesion between fibres that come into contact.
There are two mechanisms of action that could explain these 2 hypotheses. The first is that,
through D2, the expansins attach to cellulose fibres and move rapidly along the length of the
fibres such that there is a high probability that the expansin will get in the way of
approaching cellulose fibres. The second is that expansins attach to cellulose fibres and,
through D1, cause a structural defect in one of the surface glucan molecules. The
conformational change in the surface glucan is proposed to reduce the adhesion between
fibres by reducing the contact modulus and increasing the separation distance between
cellulose fibres. The results are largely consistent with the recent model of expansin action,
CHAPTER 7. RESULTS Grace Dolan, PhD Thesis
165
whereby expansins bind to cellulose and introduce a conformational change in a single
surface glucan chain along the fibre2-4. Structural similarity between plant and bacterial
sources of cellulose and expansins means that key learnings from this study provide insights
into the mechanism of action of plant expansins on cell walls during plant growth.
CHAPTER 7. RESULTS Grace Dolan, PhD Thesis
166
References for Chapter 7
1. P. Marowa, A. M. Ding and Y. Z. Kong, Plant Cell Reports, 2016, 35, 949-965.
2. N. Georgelis, A. Tabuchi, N. Nikolaidis and D. J. Cosgrove, Journal of Biological
Chemistry, 2011, 286, 16814-16823.
3. R. L. Silveira and M. S. Skaf, Physical Chemistry Chemical Physics, 2016, 18, 3510-
3521.
4. A. Lipchinsky, Acta Physiologiae Plantarum, 2013, 35, 3277-3284.
167
Chapter 8
Concluding Remarks and Future
Work
8.1 Concluding Remarks
The primary objective of this thesis is to investigate the individual contributions of plant cell
wall components and their assembly on the lubrication of moving surfaces at multiple length
scales during plant growth. Based on an extensive review of the literature on plant cell wall
architecture, mechanics, and growth processes, key tribological contacts that are considered
to be important for elongation growth are identified. The tribological contacts comprise of the
sliding interface between adjacent elongating cell walls that are typically mediated by a
pectin rich middle lamella layer, and the contact between cellulose fibres and other matrix
polymers that move relative to each other as cell walls extend. Using bacterial cellulose
hydrogels as cell wall analogues, two techniques are developed to probe these tribological
contacts.
A tribo-rheological technique is developed in a rotational rheometer and is used to measure
the in situ mechanical properties and friction behaviour at the interface between cellulose
hydrogels, which is akin to the contact between adjacent cell walls. Pairs of hydrogels are
brought into compressive contact in a solvated environment and then sheared. Due to the
poroelasticity of cellulose hydrogels, a pressure gradient develops along the interface upon
compression and leads to a region of fluid in the centre surrounded by an annulus of surface
contact. Thus, the shear forces measured at the interface are due to a combination of the
viscous forces of the solvent and the adhesion between the hydrogel surfaces that are in
CHAPTER 8. CONCLUSIONS Grace Dolan, PhD Thesis
168
contact. A computational simulation of the experiment is performed in order to predict the
thickness of the solvent film and the contact area at the interface. The contact area is sensitive
to the axial modulus of the hydrogels and the viscosity of the solvent. The interfacial yield
stress (or static friction) at the true contact area is shown to linearly scale with the
concentration of cellulose fibres at the interface. Thus the interfacial yield stress is a measure
of the interactive forces between cellulose fibres. The cellulose-XG composite shows a
reduced interfacial yield stress for the same cellulose concentration compared to the cellulose
and cellulose-AX composite hydrogels. This result indicates that XG reduces the adhesion
between cellulose fibres.
The potential lubricating role of pectin at the interface between sliding surfaces is
investigated using the tribo-rheological technique by compressing a viscous pectin film
between cellulose hydrogel pairs prior to shearing the surfaces. Increasing the pectin
concentration, and thus the solvent viscosity, is shown to reduce the interfacial yield stress by
directly reducing the contact area between the hydrogel surfaces. Pectin has a complex
concentration gradient and distribution of structures within the plant cell wall. A pectin rich
layer is typically observed at the interface between adjacent plant cell walls and the range of
structures is such that pectin can exist in solution or in a cross-linked gel with calcium. From
the results presented in this thesis it is inferred that pectin in solution decreases the contact
area and thus the shear forces between adjacent elongating cell walls. Enzymes that
modulated pectin structure in the middle lamella layer control the viscosity and can therefore
modulate the interfacial separation between neighbouring cell walls.
A dip-and-drag technique is developed using the AFM to directly measure the adhesive
interactions between individual cellulose nano-fibres in a network. The technique is
successfully validated using model electrospun polymer fibres by comparing experimental
adhesive forces to theoretical predictions based on DLVO interactions (Chapter 5: Section
5.3.3). The dip-and-drag method applied to bacterial cellulose and composite networks
reveals a reduced adhesive force at cellulose fibre contacts that are mediated by XG, whereas
AX has no effect. This is in agreement with the results from the tribo-rheological technique.
Fibre detachment simulations in silico show that XG reduces the contact strength, which
decreases the measured adhesion force. Furthermore, for a cellulose fibre contact mediated by
XG, the computational model predicts a separation distance between the fibres that matches
the length of the repeating unit in a XG molecule.
CHAPTER 8. CONCLUSIONS Grace Dolan, PhD Thesis
169
XG and AX interact differently at cellulose fibre contacts. Both polymers adsorb to cellulose
fibre surfaces, and it is expected that there is some competitive binding. XG reduces the
adhesion at fibre contacts, particularly under dynamic conditions when the cellulose fibres
are in relative motion as the cell wall extends. On the other hand, AX does not change the
adhesion between cellulose fibres but prevents XG from adsorbing to the fibre surface. The
relative synthesis of XG and AX and their incorporation into the cell wall structure can thus
modulate the overall mechanics of the plant cell wall.
Both the tribo-rheological and dip-and-drag techniques are used as mechanical assays for
assessing the activity of bacterial expansins on bacterial cellulose. Expansins have no
measurable effect on the adhesive interactions between cellulose fibres using the dip-and drag
technique. However, expansins reduce the interaction between cellulose fibres at the interface
between hydrogels. The implication of these findings is that expansins effect the formation of
new fibre contacts, but don’t act on pre-formed fibre contacts. However, the results from the
dip-and-drag technique are limited by the fact that the cellulose network is partially dried
during substrate preparation. Drying potentially changes the structure of cellulose fibre
contacts, making them inaccessible for expansin action. The results from the tribo-rheological
technique are consistent with the most recently proposed mechanism of action of expansins.
Expansins are purported to change the conformation of a single glucan chain on the surface
of a cellulose fibre. This surface chain appears to act similarly to a XG molecule mediating a
cellulose fibre contact based on the results from the tribo-rheological technique.
8.2 Recommendations for Future Work
In this thesis, the role of individual cell wall components on the interaction between cellulose
fibres has been investigated. The next step is to consider the influence of cell wall
components in combination. The relative concentrations of AX and XG in bacterial cellulose
model systems should be systematically varied to capture the complexity of the plant cell
wall composition and structure. Finally, the influence of gelled (calcium cross-linked) pectin
on the poroelastic mechanics of the cell wall and the interaction between cellulose fibres
should be investigated.
The dip-and-drag and tribo-rheological techniques developed in this thesis are shown to be
promising assays for the activity of bacterial expansins on bacterial cellulose substrates. This
is the key recommended area for future work, where a more comprehensive use of these
CHAPTER 8. CONCLUSIONS Grace Dolan, PhD Thesis
170
techniques may be revealing in terms of the exact mechanistic detail of expansin action.
Firstly, the dip-and-drag technique should be applied to never-dried cellulose networks. This
will involve refining the method of gluing the cellulose networks to the glass substrate. One
idea would be to grow networks in a shape that will allow sections of the network to be dried
and glued, whilst the rest of the network is kept hydrated. An example of this is illustrated in
Figure 8.1. Furthermore, advanced statistical analysis of the peak heights in the force-
distance curves collected in the absence and presence of expansins may be able to pick up
more subtle differences that are not detected with the Jarzynski’s equality averaging. Finally,
the dip-and-drag technique should be applied to bacterial cellulose networks that are grown in
the presence of expansins. This will provide further evidence for determining whether
expansins prevent cellulose fibre contacts or reduce the adhesion between cellulose fibre
contacts. The tribo-rheological technique gives a qualitative indication of the formation of
new fibre contacts within a hydrogel after compression (Chapter 4: Section 4.3.1). Thus if
sufficient volumes of expansins are available to add to uncompressed hydrogels, the effect of
expansins on the formation of new cellulose fibre contacts during compression can be
investigated.
Figure 8.1. Design for shape of cellulose networks grown in PDMS mould (of the same
shape). The circles on the ends can be dried and glued whilst the length of the rectangular
section in the middle is long enough to add a water droplet and keep hydrated. The size of the
circle should be large enough to manually glue to the glass substrate.
Key findings from this thesis and the results from the recommended future work should be
used to build a multiscale 3-D model of plant cell wall mechanics and growth. The
CHAPTER 8. CONCLUSIONS Grace Dolan, PhD Thesis
171
descriptive model should have the capability of predicting the mechanical response to
deformation, and the effect of changes in the wall composition and assembly. This would
include the activation of expansins, or enzymes that modulate pectic polysaccharide structure.
The direct visualisation of plant tissue during plant growth and other deformations may be
useful for model validation. For example, monitoring the thickness of the middle lamella
layer as the pectin structure is modified through enzyme action would be informative for
confirming the relationship between solvent viscosity and surface contact between adjacent
cell walls.
172
Appendices
Appendix A: Producing bacterial cellulose from
Gluconacetobacter xylinus
DAY 0
50% D-Glucose solution preparation
1. Dissolve 25 g D-Glucose into 50 mL Milli-Q water (adding the glucose little by little while it
is stirring, leave the solution stirring until totally dissolved)
In laminar flow, sterilize the 50% D-glucose solution by using 0.2 μm filter. Keep the sterilized 50%
D-glucose solution in the fridge.
Bacteria Revival
For 100mL of HS medium:
- Peptone = 0.5g
- Yeast Extract = 0.5g
- Na2HPO4.2H2O = 0.338g
- Citric acid = 0.115g
- Glucose (50%) = 4mL
- dH2O = 96mL
1. Prepare 100mL HS medium, add 1.5g agar, then autoclave.
2. After autoclaving, when the temperature is about 700C, add in 4 mL 50% D-glucose solution
into each HS-agar medium.
3. Pour a certain amount of liquid HS-agar medium into 90 mm petri dish and leave them on the
bench to make solid. 100 mL is good for 6 or 7 plates.
4. It will form a solid plate in around 30 min to 1 hr. Use 2 plates for revival and keep the rest 5
plates for subculture.
5. Prepare some ice in a small plastic beaker (from level 3). Take a whole tube of beads (where
we store the ATCC53524) from -80 °C freezer and place the tube into ice immediately.
6. In the laminar flow, use a needle to bring one bead out and put into the agar plate. Use loop to
spread the bead on one corner of the agar plate.
7. Change the side of the loop or change to a new loop every time when change corner (see
figure below).
APPENDICES Grace Dolan, PhD Thesis
173
8. Put the plate upside down and cultivate into 30 °C incubator for 48-72 hr.
***Make sure that the incubator fan is turned off.
***Revival plates last a month, only 2 subcultures from a revival plate is recommended or
you could subculture once from the first subculture (i.e. maximum of 2 passages).
Subculture (48 or 72hours after revival)
1. 9 am bring all the plate out from the fridge to warm up (either put them on the bench of the
laminar flow or in the incubator)
2. 10am take one loop of bacteria from the revival plate and spread it onto a new plate to finish
subculture. Repeat the same movement as revival.
DAY 1
Preparing HS Medium
For 300mL of medium:
- Peptone = 1.5g
- Yeast Extract = 1.5g
- Na2HPO4.2H2O = 1.014g
- Citric acid = 0.345g
- Glucose (50%) = 12mL
- dH2O = 288mL
**make 2x concentrated for composites
1. Weigh dry materials directly into a 500mL glass beaker on the open balance in the following
order: Na2HPO4.2H2O, citric acid, YE, peptone (make sure you replace the parafilm on the
containers of the last two materials). Note: you should wipe spatula with paper towel between
measuring different materials and you need to change spatula if the dry powder does not wipe
off completely.
2. Turn MilliQ water on and let it run until it reads 18.2. Using a measuring cylinder add
approximately 200 mL (or some volume that is less than the total volume needed) of milliQ
water to beaker containing the dry powders and a magnetic stirring bar. Turn the stirrer on as
the first bit of water is added. Mix until dissolved.
3. Calibrate pH meter if required. Wash pH probe with distilled water (make sure you do not get
water above the black line on the probe) and dry with kimwipes. Place probe in the medium
(should generally get a reading around the 6.2 mark). Add drops of 1M HCl to bring the pH
reading down to 5.
APPENDICES Grace Dolan, PhD Thesis
174
4. Add contents of the beaker to the measuring cylinder (can use another magnetic stirrer bar in
order to hold the stirrer inside the beaker to the bottom making it easier to pour out the
medium) and then add additional milliQ water to get the volume up to 300 mL. Pour the
contents of the measuring cylinder back into the beaker and put back on stirrer. Adjust the pH
so it’s 5 again (this should require no more than a few drops).
5. Pour the medium into a schott bottle and leave the lid so that it is not tight. Cover the lid with
alfoil, put a strip of autoclave tape on the bottle and label.
6. Place in autoclave.
Preparing AX solution
(AX solution is used to prepare a cellulose-AX composite network, the procedure is the same for XG
to prepare a cellulose-XG composite)
7. Prepare a bottle of water for autoclaving to be later used for dissolving AX. Rinse 500mL
Schott bottle and fill to the 400mL mark. Autoclave for 15min.
8. Need a 500mL Schott bottle autoclaved the day before (or a few days before). Take out the
autoclaved 500mL Schott bottle and allow to stand at room temperature so that it is not hot
when you are weighting the AX powder.
Labelling primary and scale up containers
9. Whilst autoclave is running start labelling containers required for primary inoculum and
scale-up. Label primary with: HS, 10, date of primary inoculum, initials, 1
0 on the lids. Label
scale ups with: HS, #no., date of scale up, initials, #no. on the lids.
Primary Inoculum
BC BC composite
4 x 70mL containers
filled with 10mL HS
medium
4 x 70mL containers filled
with 20mL HS medium
Scale up
o Cellulose hydrogel
70mL container (φ=41mm)
Primary Inoculum 1mL
HS medium 9mL
o Composite hydrogel (with AX or XG)
70mL container (φ=41mm)
Primary Inoculum 2mL
HS medium 18mL
Preparing AX solution continued.
10. For dissolving AX set up hot plate and stirrer with mineral oil in a beaker in the laminar flow
cabinet. Have the Bunsen burner on and wipe hands with ethanol to prevent contamination.
11. Temporarily move the balance in the cabinet.
12. Using an autoclaved spatula (wrapped in foil) measure of 3g AX (for 300mL) into a 500mL
Schott bottle.
APPENDICES Grace Dolan, PhD Thesis
175
13. Add autoclaved magnetic stirrer bar (wrapped in foil).
14. Set the hot plate to 850C with the temperature probe in the oil (in order to get solution to
800C).
15. Set the stirrer to 240rpm.
16. Add 300mL of the autoclaved water (autoclaved earlier this day so the water is still hot which
will help with the dissolving) in 10mL lots using a pipette. Add the first 3 lots and mix around
well by swirling the bottle (the powder will go gluggy). Increase the rpm and continue to
slowing add the water.
17. Leave the AX solution to dissolve overnight.
DAY 2
Aliquot medium
Cellulose hydrogel
1. Wipe outside of schott bottle containing medium with ethanol.
2. Add 12mL (2x6mL) glucose to HS medium.
3. Pipette 10mL of medium into primary containers.
4. Change pipette to 9mL with tip still on and pipette 9mL into all 70mL scale up containers.
5. Pipette 1000µL to all 5mL scale up containers followed by another 350µL (1.35mL total
volume).
Cellulose composite (with AX or XG)
1. Take AX solution off hotplate after it has dissolved overnight. Let cool after taking off the
hotplate (to speed up this process put the bottle in some water). Need to be able to touch the
AX solution to the skin on the inside of your arm comfortably.
2. Make sure you have autoclaved pipette tips – put tips in the dry oven after autoclaving to
remove any condensation.
3. Light Bunsen burner.
4. Pipette 12mL (2x6mL) glucose into each batch of HS medium (2x concentrated), swirl bottle
to mix around.
5. Split AX solution evenly between the 2 bottles of HS medium. Make sure you record how
much total volume you’ve added (will be greater than 150mL as the AX powder increases the
volume). Swirl around to mix.
6. Pipette/aliquot medium into primary and scale up containers that have been labelled.
7. **With pipette tip on you can only adjust the volume DOWN.
Put all scale up containers in the fridge.
Primary inoculum
1. With plastic loop take a loopful of bacteria off the subculture plate and stir into the medium
which has been aliquotted into the primary containers.
2. With the lids ajar place in the incubator.
DAY 5
**Take primary containers out of the incubator and scale up containers out of the fridge to get to
room temperature.
APPENDICES Grace Dolan, PhD Thesis
176
For cellulose micro-gels
Cleaning Microarrays
Place microarrays that have been bonded to petri dishes in the plasma cleaner and turn the vacuum on
for 15minutes.
1. Plasma clean for 1.5minutes on high.
2. Turn off for 1 minute (to let cool)
3. Plasma clean for another 1.5minutes on high.
4. Cover the surface of the microarrays with water to maintain hydrophilicity
5. With petri dish lids on and in a large 14cm petri dish with the lid on, transport to the
microbiology lab.
Scale up
1. Place the primary inoculum containers onto the shaker and shake for 5minutes on 350rpm.
2. Combine all the primary inoculum into 1 container to homogenise.
3. Pipette 200µL of primary inoculum onto the surface of the microarrays. Leave for 10 minutes
for the bacteria to sediment.
4. Pipette 150µL of primary inoculum into all 5mL scale up containers.
5. Pipette 1mL of primary inoculum into all 70mL scale up containers.
6. Put all containers in the incubator with lids ajar and record the time (they will need to
incubate for 72hours).
Blotting
1. For the microarrays, with a 50µL pipette remove excess liquid.
2. Fold up and Olympus lens cleaning tissue, hold briefly near the flame, then rub lightly on the
surface of the microarray.
3. Place 5mL milli-Q water in the bottom of 70mL sterile container (to create 100% RH).
4. Place the lids back on the petri dishes and drop into the containers.
5. Put all containers in the incubator with lids ajar and record the time (they will need to
incubate for 72hours).
DAY 8
Harvesting
1. Take containers out of the incubator and place them on the shaker (350rpm, 5mins).
2. Pour pellicles in 5mL containers into tea strainer over a glass beaker.
3. Invert tea strainer over a 1L beaker and pour 500mL of cold water over the pellicle so that
they fall into the beaker.
4. With fine tip tweezer slide the individual 41mm pellicles over the edge of the cups to remove
excess strands of cellulose etc. and place in a 1L beaker. Pour 500mL of cold water over the
pellicles.
5. Place the beaker on the shaker for 3x30min and 3x10min at 100rpm replacing the water each
time.
6. After all washes, drain off the water and place in batches in 70mL sterile containers then
cover with 0.2% sodium azide solution.
For the cellulose microgels
APPENDICES Grace Dolan, PhD Thesis
177
7. Wash the microarray 4 times by pipetting cold water on the surface, swirling around, and the
removing the water with the pipette. After the final washing step, cover the surface of the
microarray with 0.2% sodium azide solution.
APPENDICES Grace Dolan, PhD Thesis
178
Appendix B: Chromium Mask Fabrication
This standard operating procedure is provided by the Australian National Fabrication Facility,
Queensland (ANFF-Q).
EQUIPMENT Heidelberg uPG 101 laser direct writer
4 containers (large enough to hold 5” mask)
AZ 726 MIF Developer
Milli-Q water
Chrome etchant
Acetone
Isopropanol
Design file (format: GDSII, Gerber or DXF)
NOTE: This protocol uses a GDSII file as an example
Always use Writemode III – for other writemodes contact ANFF staff
LOADING MASK AND EXPOSING DESIGN 1. Make booking online (Raspberry).
2. Fill in EXPERIMENT IN PROGRESS sheet.
3. Copy design file into the following location on the Heidelberg computer and in the folder name
that corresponds to the file type (e.g. GDSII file in GDSII folder) :
C:\Himt\ExposureWizard\Designs\GDSII
4. Double click on file to open it in LayoutEditor software. Check the design for any error.
5. Start the uPG 101 Exposure Wizard using the link icon on the desktop:
6. WELCOME screen opens. Check that it reads “Writemode III – 20mm”. If not, immediately
contact ANFF staff. If it is Writemode III then wait for initialisation to complete then click Next.
APPENDICES Grace Dolan, PhD Thesis
179
7. Click Load Design and select design file.
APPENDICES Grace Dolan, PhD Thesis
180
8. GDSII OPTIONS screen opens. Magnification Factor is for scaling purposes – if no scaling is
required enter the value ‘1’. Select GDSII structure containing the layers (e.g. MainSymbol) and
tick the specific layer to expose. Click Create to load design file.
APPENDICES Grace Dolan, PhD Thesis
181
9. Once design file is loaded click Next. “Exp Time” is the approximate time take to expose the
design.
APPENDICES Grace Dolan, PhD Thesis
182
10. CHECK OPTIONS screen opens.
a. Expose Options:
APPENDICES Grace Dolan, PhD Thesis
183
Parameters are selected based on the substrate type and brand that is used. Currently, the values
to use are 9mW laser output power, 53% pixel pulse duration, and Energy Mode 1x1.
b. Design Options:
INVERTED - By default, the structures that were defined in the design are exposed on the
substrate (positive photoresist process). If the checkbox option Inverted is selected, all areas with
no structure defined are exposed instead. The limits of the exposure file are defined by the
outmost structures. If a larger area should be exposed around the structures, a frame had to be
defined by entering the required distance from the structures to the edge of the exposed area into
the text box. Maximum value for the frame is 10mm.
APPENDICES Grace Dolan, PhD Thesis
184
AUTOMATIC CENTERING – If Automatic Centering option is not selected, the design will be
positioned so that the origin of the mask drawing is exposed on the center of the substrate. If
Automatic Centering is selected, the wizard automatically shifts any design so that the
geometrical center is exposed on the center of the substrate, regardless of where the origin is
located. The option of automatic centering greys out when the origin of the design is the
geometrical center.
To keep things simple, we recommend including a border of 120mmx120mm around your design
and always select Automatic Centering. c. Once design options are selected, click on Show Control Panel.
11. CONTROLS screen opens. Press To Un/Load (1) - the stage moves to the loading position at
the front of the machine. Once the Loading Status window opens, immediately press Close (2) on
the CONTROLS screen.
WARNING: DO NOT press ANY other buttons on the CONTROLS screen as this may
accidentally causes the write head to come crashing down on the stage. Such damages will
be extremely costly and may take months to fix.
APPENDICES Grace Dolan, PhD Thesis
185
12. LOAD the substrate:
a. Wait until all movement has finished. Open the cover lid. Mount the substrate against the
alignment pins on the stage – make sure the resist coated side of the substrate is turned up. Do
NOT load the mask if the stage is not in this position.
b. Switch ON the vacuum with the vacuum knob. Check whether the plate is really held tight by
trying to move it slightly sideways. If plate is not held, switch off the vacuum and clean the plate
backside and chuck before trying again.
APPENDICES Grace Dolan, PhD Thesis
186
13. Return to OPTIONS screen and press Next.
14. MANUAL ALIGNMENT screen opens. Press Find Plate Center.
APPENDICES Grace Dolan, PhD Thesis
187
15. FIND PLATE CENTER screen opens. Tick fast mode and press Start.
APPENDICES Grace Dolan, PhD Thesis
188
16. Once FIND PLATE CENTRE finishes scanning the edges, press ACCEPT.
APPENDICES Grace Dolan, PhD Thesis
189
17. Return to MANUAL ALIGNMENT screen and press Next.
APPENDICES Grace Dolan, PhD Thesis
190
18. EXPOSE DESIGN screen opens. Tick Auto Unload after Exposure and press Expose.
APPENDICES Grace Dolan, PhD Thesis
191
DEVELOPING MASK Summary Table
* Note: Time may vary slightly depending on the chrome
etchant solution
A. Prepare 4 large containers. Separately pour the following solutions into the containers:
1. 200mL AZ 726 MIF Developer
2. 400mL Milli-Q water
3. 400mL Chrome etchant
4. 200mL Acetone
APPENDICES Grace Dolan, PhD Thesis
192
B. Place the mask in the container with AZ 726 MIF Developer for 45sec.
C. Immediately transfer the mask into the container with Milli-Q water and rinse for about 10secs. Remove the mask from the container and spray down with fresh Milli-Q water.
D. Dry the mask with a gentle stream of nitrogen.
E. Place the mask in the container with chrome etchant for 1min 15secs or until area to be
etched appears clear.
F. Immediately transfer the mask into the container with Milli-Q water and rinse for about
20secs. Remove the mask from the container and spray down with fresh Milli-Q water.
G. Dry the mask with a gentle stream of nitrogen and inspect the mask under the microscope to
check etching is complete. If not, return the mask into the container with chrome etchant for further etching – take care with time as over etching may occur.
H. Place the mask in the container with acetone for about 2mins with sonication to remove the
resist layer. If required, while the mask is in acetone, use a soft wipe to wipe the surface of the mask to assist with removal of resist debris – be gentle as scratching may occur.
I. The mask is rinsed with isopropanol and spray dried with a gentle stream of nitrogen.
J. Chrome etchant may be reused – pour it back into its bottle for storage.
K. Discard all other solutions in the corresponding waste containers provided.
APPENDICES Grace Dolan, PhD Thesis
193
Appendix C: SU-8 Master Fabrication
This standard operating procedure is provided by the Australian National Fabrication Facility,
Queensland (ANFF-Q).
MATERIALS AND EQUIPMENT Photo-mask
Silicon wafer
Acetone
Isopropanol (IPA)
Propylene glycol monomethyl ether acetate (PGMEA)
UV Light source
Hot plates
SU-8 Photo-resist (select type for desired feature height)
PROCEDURE 1. CLEAN the silicon wafer with acetone then IPA and dry with a stream of nitrogen.
2. DEHYDRATE the wafer on a hot-plate at 180 °C for 20 min.
3. Ti PRIMER TREATMENT (Optional)
a. Spin-coat Ti Primer on wafer (about 1mL per 4 inch wafer)
Step 1) 500 rpm 100 rpm/s 10s
Step 2) 3000 rpm 300 rpm/s 30s
b. Bake at 200 °C for 1 min and allow to cool down to room temperature.
4. Pour photo-resist onto the wafer (1 mL per inch of wafer diameter) and SPIN-COAT at the
appropriate speed for the required thickness:
* Recommended speed values are shown in the following table.
SPIN-COATING SPEED (RPM) FOR COMMON THICKNESSES
APPENDICES Grace Dolan, PhD Thesis
194
5. SOFT BAKE the wafer on a hot-plate through a series of step change in temperature (from 65
°C → 95 °C → 65 °C, see Summary Table). Always allow the wafer to cool down to room temperature before next step.
6. Place photo-mask onto the wafer and UV EXPOSE for applicable exposure dose (see
Summary Table). Make sure the printed side (i.e. for chrome mask, “bronze” side and not “silver”
side) of the mask is in direct contact with the photo-resist layer, and check UV intensity with radiometer before exposure.
7. POST BAKE the wafer on a hot-plate through a series of step change in temperature (from 65
°C → 95 °C → 65 °C, see Summary Table) to selectively cross-link the exposed portions.
Always allow the wafer to cool down to room temperature before next step. Extra baking step for
mask before development at 95 degrees for 2 mins
8. DEVELOPE the silicon wafer with PGMEA to remove any uncross-linked photo-resist – i.e.
until features are revealed. Strong agitation is recommended when developing high aspect ratio and thick film structures.
9. RINSE the PGMEA with IPA to stop the developing process. If streaks of milky white appear then re-rinse with PGMEA then IPA.
10. DRY the silicon wafer with a stream of nitrogen.
11. HARD BAKE the silicon wafer on a hot-plate at 120 °C for 20 min. This step is required for
DRIE (Deep Reactive Ion Etching) process.
APPENDICES Grace Dolan, PhD Thesis
195
SUMMARY TABLE (for Steps 5, 6 and 7)
CLEANING AND WASTE DISPOSAL 1. WIPE and CLEAN the spin-coater with acetone. Place all used wipes in the designated
container in the fumehood.
2. EMPTY WASTE SOLUTIONS into the appropriate waste containers:
BEFORE LEAVING CLEANROOM TURN-OFF hot-plates, spin-coater and UV light source. Leave UV light source on if you know
someone else will be using it in the same day.
APPENDICES Grace Dolan, PhD Thesis
196
Appendix D: PDMS Microarray Fabrication
Preparing PDMS
Using SYLGARD 184 Silicone Elastomer Kit prepare PDMS with base:binder ratio of 10:1
1. Tare weight of plastic cup.
2. Weigh 10g of base (record exact weight).
3. Tare balance again.
4. Weigh 1g of binder (record exact weight).
5. Mix well using a wooden stick.
6. Place PDMS under vacuum for 15minutes in order to degas (remove bubbles). Note: you can
spray dry with nitrogen to remove any extra bubbles remaining after 15minutes in the vacuum
desiccator.
Following this procedure will make enough PDMS to produce enough to fill the mould twice (i.e.
make 6 microarrays which can be cut into halves to fit the petri dishes, therefore 12 sets of microwells
in total)
Curing the PDMS in the mould
1. Using a plastic pipette/dropper, with a portion of the narrow end cut off, evenly spread the
PDMS into the mould.
2. Degas in the vacuum dessicator and spray dry with N2 to remove extra bubbles if required (or
you can use put in the plasma cleaner with the vacuum pump on to remove excess bubbles).
3. Place mould on the hotplate (whilst cold) and turn on to 800C. Leave for 20minutes to begin
curing.
4. Cut out of mould and repeat from step 1. With the remaining PDMS or put mould directly
into the 650C oven overnight.
Removing PDMS microarrays from the mould and binding to petri dishes
1. Cut through the PDMS around the individual silicon wafer moulds leaving a border of
approximately 3mm.
2. Trace around the edge inserting the tweezers to ensure the PDMS is cut all the way through.
3. Peel the PDMS off the mould and place onto 14cm petri dish with wells facing up.
4. Using tweezers push gently on the border of the PDMS so that it sticks to the petri dish
making it easier to cut.
5. With the scalpel, cut off the majority of the border (and cut the corners off if you need it to fit
in the circle area of the glass bottom petri dish).
6. Place the cut out microarrays on tape with the patterned side facing down onto the sticky side
of the tape.
7. Put the tape with the bottom of the microarrays facing up and the small petri dishes (without
lid) in the plasma cleaner.
8. Pump down for 15minutes (i.e. with the vacuum pump on before the oxygen is introduced).
9. Treat for 38second on high.
10. After treatment put the PDMS microarrays on the small petri dishes and press down to
remove air bubbles. Pull off tape and leave to bond overnight.
APPENDICES Grace Dolan, PhD Thesis
197
Appendix E: MATLAB code for poroelastic mechanical model
Linear_orthotropicF.m
close all clear all fclose('all'); global e0 t0 v31 E3 fdifer k_r2 ikest icomp Pinf2 ifast
A=importdata('C3000.txt','\t',1); %C3000.txt is the text file of the raw
data during the compression-relaxation steps for cellulose hydrogels
compressed to 3000microns B=A.data; t=B(:,1); %t=time in s tc=B(:,2); %tc=time in s for a certain step: compression or relaxation h=B(:,3); %h=gap between the rheometer plates P=B(:,4); %P=normal stress in Pa n=length(t); cycindex=zeros(30,2); paramcyc=zeros(30,7); tcyc=zeros(30,2); test=zeros(30,2); Pfit=0*P; tfit=0*t; extr=1; %1=Extrapolate relaxtation part through an exponential
decay (if relaxation is not complete) %0=Do not extrapolate ikest=1; %1=fit permeability %0=find permeability from intercept icomp=1; %1=fit compressive modulus %0=find compressive modulus from slope ifast=0; %1=fast compression
deltat=0.0; %Time between consecutive cycles
%Initial Model Parameters ramprate=33.333; %ramp rate um/s (positive for compression) h0=1996; %Initial sample thickness um r0=20e3; %Initial radius um v31=0.0; %Poisson ratio zr (=-err/ezz) v21=0.5; %Poisson ratio zphi (=-ephi/err) E1=1e5; %Young modulus rr, Pa tgf=0.1; %tramp/tporo nn=1;
if ifast==1 ikest=1; icomp=1; end
%Define lower and upper limits for fitting parameters lb(1)=1; ub(1)=1e7; %E1 lb(2)=0.1; ub(2)=0.99; %v21 ik=3; if icomp==1 ik=4; lb(3)=1; ub(3)=1e7; %E3
APPENDICES Grace Dolan, PhD Thesis
198
end
if ikest==1 lb(ik)=0.001/r0^2; ub(ik)=1e7/r0^2; %k_r2 end if ifast==1 fdifer0=-100; lb(5)=-1e8; ub(5)=0.0; %fdifer end
h00=h0; i1=1; tlast=0; cycle=0; while i1<n k=i1; nslope=0.4; %number of poins for determination of slope while k<n k=k+1; if tc(k)<1e-6 break end end cycle=cycle+1; disp(cycle) if cycle==13 disp(cycle) end
i2=k-1+floor(k/n); tdata=tc(i1:i2); Pdata=P(i1:i2)-P(i1); tcyc(cycle,1)=deltat+tlast; if cycle==1 tcyc(cycle,1)=0; end cycindex(cycle,1)=i1; cycindex(cycle,2)=i2; e0=ramprate/h00;
% Correct time data if ifast==0 count=1; for i=i1+1:i2 if tc(i)>tc(i-1) count=count+1; continue else ir=i; tcor=tc(i-1); t0=tcor; tcyc(cycle,2)=t0+deltat+tlast; if cycle==1 tcyc(cycle,2)=t0; end break end end tdata(count+1:end)=tc(ir:i2)+tcor; else
APPENDICES Grace Dolan, PhD Thesis
199
[fmax,count]=max(Pdata); t0=tdata(count); if Pdata(count+1)>0.95*Pdata(count) t0=(tdata(count)+tdata(count+1))/2; count=count+1; end tdata(count+1:end)=tdata(count+1:end)+deltat; ir=count+i1-1; end
% Determine the slope during compression nslope=round(nslope*count); tlin=tdata(count-nslope:count); Plin=Pdata(count-nslope:count); Pend=Plin(end); p = polyfit(tlin,Plin,1); E3=p(1)/e0; Pinf1=p(1)*t0; deltaP1=p(2);
% Extrapolate exponential decay if extr==1 Pdec=Pdata(count+1:end); tdec=tdata(count+1:end); ydecay=@(b,tt) b(1)+b(2)*exp(-b(3)*tt); b0=[Pdec(1)/2,Pdec(1)/2,0.01];
problem=createOptimProblem('lsqcurvefit','objective',ydecay,'xdata',tdec,..
. 'ydata',Pdec,'x0',b0,'lb',1e-6*b0,'ub',1e6*b0); ms=MultiStart; bdecay=run(ms,problem,12); ydecfit=ydecay(bdecay,tdec); plot(tdec,Pdec,'o',tdec,ydecfit,'r') Pinf2=bdecay(1); close else Pinf2=1/3*(Pdata(end)+Pdata(end-1)+Pdata(end-2)); end fdifer=Pinf2-Pinf1;
% Estimate the permeability k_r2=e0/8/deltaP1;
% Initial values beta0(1)=1.1*E1; beta0(2)=v21; ik=4; if ikest==1 beta0(ik)=k_r2; if k_r2<0 beta0(ik)=0.01*ub(ik); end end
if ifast==1 beta0(5)=fdifer0; end
APPENDICES Grace Dolan, PhD Thesis
200
if icomp==1 beta0(3)=1.1*E3; end
nsolver=nn*ik; if cycle==1
problem=createOptimProblem('lsqcurvefit','objective',@orthoF,'xdata',tdata,
... 'ydata',Pdata,'x0',beta0,'lb',lb,'ub',ub); ms=MultiStart; [beta,fval,exitflag]=run(ms,problem,nsolver); else
problem=createOptimProblem('lsqcurvefit','objective',@orthoF,'xdata',tdata,
... 'ydata',Pdata,'x0',beta0,'lb',lb,'ub',ub); ms=MultiStart; [beta,fval,exitflag]=run(ms,problem,nsolver);
%[beta,resnorm]=lsqcurvefit(@orthoF,beta0,tdata,Pdata,lb,ub); %[beta,fval,exitflag]=run(ms,problem,1); end if ikest==1 k_r2=beta(ik); end
if ifast==1 fdifer=beta(5); end if icomp==1 E3=beta(3); end yfit=orthoF(beta,tdata); Pfit(i1:i2)=yfit+P(i1)-P(1); tfit(i1:i2)=tdata+deltat+tlast; if cycle==1 tfit(i1:i2)=tdata; end
E1=beta(1); v21=beta(2); E1final=Pinf2/(e0*t0); paramcyc(cycle,2:4)=[E1,E3,v21];
delta1=1-v21-2*v31^2*E1/E3; C11=E1*(1-v31^2*E1/E3)/((1+v21)*delta1); k=1e-12*r0^2*k_r2; tg=1/C11/k_r2; paramcyc(cycle,5)=k; paramcyc(cycle,6)=-fdifer; paramcyc(cycle,7)=Pend; paramcyc(cycle,8)=E1final;
lb(1)=E1; ub(1)=1e9; %E1 ik=3; if icomp==1 ik=4; lb(3)=E3; ub(3)=1e9; %E3 end
APPENDICES Grace Dolan, PhD Thesis
201
tlast=tlast+tdata(end);
h00=h00-ramprate*t0; paramcyc(cycle,1)=(h0-h00)/h0; i1=i2+1; end
variab={'strain';'Er';'Ez';'v21';'k';'Aggreg. Force'}; nv=length(variab);
for i=1:nv BB{1,i}=variab{i,1}; end for i=2:cycle+1 for j=1:6 BB{i,j}=paramcyc(i-1,j); end end xlswrite('Fitting results',BB,1)
plot(tfit,Pfit,'LineWidth',3); xlabel('time(s)'); ylabel('P(Pa)'); hold on plot(tfit,P-P(1),'or','MarkerSize',3);
orthoF.m
function y=orthoF(beta,t)
global v31 t0 nt e0 fdifer icomp ikest E3 k_r2 ifast
y=0*t; nt=100; E33=E3; k_r22=k_r2; E1=beta(1); v21=beta(2); ik=3; if icomp==1 E33=beta(3); ik=4; end if ikest==1 k_r22=beta(ik); end if ifast==1 fdifer=beta(5); end
E3=E33; k_r2=k_r22; v312=v31^2; alpha = trascend(E1/E3,v312,v21); alpha=alpha.^2; delta1=1-v21-2*v312*E1/E3; delta2=(1-v312*E1/E3)/(1+v21); delta3=(1-2*v312)*delta2/delta1;
APPENDICES Grace Dolan, PhD Thesis
202
delta22=delta2^2; C11=delta2/delta1*E1; den=delta1/(1+v21); tg=1/(C11*k_r2);
[tm,ind]=min(abs(t-t0)); if tm>t0 ind=ind-1; end maxt=length(t); c1=E3*e0*t0; c2=E1*e0*tg*delta3;
for i=1:ind s=0; ti=t(i); for j=2:nt s=exp(-alpha(j)*ti/tg)/(alpha(j)*(delta22*alpha(j)-den))+s; end y(i)=c1*ti/t0+c2*(0.125-s); end k=0; ta=tg/alpha(2); for i=ind+1:maxt s=0; k=k+1; ti=t(i); ff=fdifer*(1-exp(-(ti-t0)/ta)); for j=2:nt s1=exp(-alpha(j)*ti/tg); s2=exp(-alpha(j)*(ti-t0)/tg); s=s+(s1-s2)/(alpha(j)*(delta22*alpha(j)-den)); end y(i)=c1-c2*s+ff; end
end
trascend.m
function alpha = trascend(E1_E3,v312,v21)
global nt
alpha=zeros(nt,1); alpha(1)=0; f=(1-v312*E1_E3)/(1-v21-2*v312*E1_E3);
fun=@(x)besselj(1,x)-x*besselj(0,x)*f;
a1=1e-5; a2=a1; s=1; for i=2:nt s1=fun(a1); while s>0 a2=a2+0.2; s2=fun(a2); s=s1*s2;
APPENDICES Grace Dolan, PhD Thesis
203
end x0=[a1,a2]; alpha0=fzero(fun,x0); a1=alpha0+1e-5; alpha(i)=alpha0; a2=a1; s=1; end
end
APPENDICES Grace Dolan, PhD Thesis
204
Appendix F: Text file of raw data for input into MATLAB code
for the poroelastic mechanical model
The raw data during the compression-relaxation steps is extracted from the Rheowin Data
Manager software.
- t in s, is the time throughout the compression and relaxation step.
- t_seg in s, which is labelled tc in the MATLAB code, is the time during a certain step
(i.e. compression or relaxation).
- h in mm, is the gap between the emry covered rheometer plates.
- P_in_ Pa, is the normal stress.
APPENDICES Grace Dolan, PhD Thesis
205
This data is copied into an text file and labelled C3000.txt for cellulose hydrogel compressed
APPENDICES Grace Dolan, PhD Thesis
206
Appendix G: Raw data from the compression-relaxation steps for
hydrogels (cellulose, CAX, CXG) at all CRs
Figure G.1. Normal stress during compression-relaxation steps for cellulose hydrogels at CR
0.4.
Figure G.2. Normal stress during compression-relaxation steps for cellulose hydrogels at CR
0.5.
Time (s)
0 20 40 60 80 100
Norm
al S
tress (
kP
a)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
Time (s)
0 20 40 60 80 100
Norm
al S
tress (
kP
a)
0.0
0.5
1.0
1.5
2.0
2.5
3.0
APPENDICES Grace Dolan, PhD Thesis
207
Figure G.3. Normal stress during compression-relaxation steps for cellulose hydrogels at CR
0.6.
Figure G.4. Normal stress during compression-relaxation steps for cellulose hydrogels at CR
0.7.
Time (s)
0 20 40 60 80 100
Norm
al S
tress (
kP
a)
0
1
2
3
4
5
Time (s)
0 20 40 60 80 100
Norm
al S
tress (
kP
a)
0
2
4
6
8
APPENDICES Grace Dolan, PhD Thesis
208
Figure G.5. Normal stress during compression-relaxation steps for CAX hydrogels at CR
0.4.
Figure G.6. Normal stress during compression-relaxation steps for CAX hydrogels at CR
0.5.
Time (s)
0 20 40 60 80 100
Norm
al S
tress (
kP
a)
0
1
2
3
4
Time (s)
0 20 40 60 80 100
Norm
al S
tress (
Pa)
0
2
4
6
8
APPENDICES Grace Dolan, PhD Thesis
209
Figure G.7. Normal stress during compression-relaxation steps for CAX hydrogels at CR
0.6.
Figure G.8. Normal stress during compression-relaxation steps for CAX hydrogels at CR
0.7.
Time (s)
0 20 40 60 80 100
Norm
al S
tre
ss (
kP
a)
0
2
4
6
8
10
12
14
16
Time (s)
0 20 40 60 80 100
Norm
al S
tre
ss (
kP
a)
0
5
10
15
20
25
30
APPENDICES Grace Dolan, PhD Thesis
210
Figure G.9. Normal stress during compression-relaxation steps for CXG hydrogels at CR
0.4.
Figure G.10. Normal stress during compression-relaxation steps for CXG hydrogels at CR
0.5.
Time (s)
0 20 40 60 80 100
Norm
al S
tress (
kP
a)
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
Time (s)
0 20 40 60 80 100
Norm
al S
tress (
kP
a)
0
2
4
6
8
APPENDICES Grace Dolan, PhD Thesis
211
Figure G.11. Normal stress during compression-relaxation steps for CXG hydrogels at CR
0.6.
Figure G.12. Normal stress during compression-relaxation steps for CXG hydrogels at CR
0.7.
Time (s)
0 20 40 60 80 100
Norm
al S
tre
ss (
kP
a)
0
2
4
6
8
10
12
14
16
Time (s)
0 20 40 60 80 100
Norm
al S
tress (
kP
a)
0
5
10
15
20
25
30
APPENDICES Grace Dolan, PhD Thesis
212
Appendix H: Raw data from the tribo-rheological test for all
hydrogels and solvents
Friction curves for Cellulose, CAX and CXG at all CR values in water.
Figure H.1. Friction curves for Cellulose hydrogel pairs in water at all compression ratios.
Figure H.2. Friction curves for CAX hydrogel pairs in water at all compression ratios.
Shear Strain (-)
0.001 0.01 0.1 1 10
Shear
Str
ess (
kP
a)
0.01
0.1
1
10
CR = 0.7
CR = 0.4
Shear Strain (-)
0.001 0.01 0.1 1 10
Shear
Str
ess (
kP
a)
0.01
0.1
1
10
CR = 0.7
CR = 0.4
APPENDICES Grace Dolan, PhD Thesis
213
Figure H.3. Friction curves for CXG hydrogel pairs in water at all compression ratios.
Friction curves for Cellulose hydrogel in all pectin solutions at all CR values.
Figure H.4. Friction curves for a pair of cellulose hydrogels in 0.5 wt% pectin solution at all
values of CR.
Shear Strain (-)
0.001 0.01 0.1 1 10
Shear
Str
ess (
kP
a)
0.01
0.1
1
10
CR = 0.7
CR = 0.4
Shear Strain (-)
0.001 0.01 0.1 1 10
Shear
Str
ess (
kP
a)
0.01
0.1
1
10
CR = 0.7
CR = 0.4
APPENDICES Grace Dolan, PhD Thesis
214
Figure H.5. Friction curves for a pair of cellulose hydrogels in 1 wt% pectin solution at all
values of CR.
Figure H.6. Friction curves for a pair of cellulose hydrogels in 2 wt% pectin solution at all
values of CR.
Shear Strain (-)
0.001 0.01 0.1 1 10
Shear
Str
ess (
kP
a)
0.01
0.1
1
10
CR = 0.7
CR = 0.4
Shear Strain (-)
0.001 0.01 0.1 1 10
Shear
Str
ess (
kP
a)
0.01
0.1
1
10
CR = 0.7
CR = 0.4
APPENDICES Grace Dolan, PhD Thesis
215
Figure H.7. Friction curves for a pair of cellulose hydrogels in 4 wt% pectin solution at all
values of CR.
Shear Strain (-)
0.001 0.01 0.1 1 10
Shear
Str
ess (
kP
a)
0.01
0.1
1
10
CR = 0.7
CR = 0.4
APPENDICES Grace Dolan, PhD Thesis
216
Appendix I: Cellulose concentration based on the G’ of the
hydrogels for cellulose, CAX, and CXG.
Cellulose CAX CXG
G' Cellulose concentration G' Cellulose concentration G' Cellulose concentration
Pa % Pa % Pa %
101.9 0.71221 159 0.68698 112.7016 1.24404
159 0.73679 387 0.72277 203.6787 1.3418
237 0.7634 723 0.7625 335.9278 1.45624
331.4 0.79171 1160 0.80686 485.7625 1.59203
435.5 0.82221 1580 0.85669 716.3231 1.75574
607 0.85514 2010 0.91309 882.8999 1.95697
761.8 0.89083 2530 0.97743 999.65 2.21031
974.8 0.92962 3120 1.05153 1260.495 2.53899
1098 0.97195 3770 1.13778 1713.081 2.98251
1246 1.01831 4410 1.23945 2645.768 3.61376
1455 1.06932 5100 1.36107 5130.178 4.58396
1787 1.12512 6050 1.50916
2244 1.18772 7500 1.6934
2669 1.2577 9680 1.92889
2960 1.33643 14800 2.24045
3600 1.42569
4409 1.52772
5246 1.64675
5905 1.7844
6733 1.94894
8173 2.14475
10370 2.3843
13150 2.68409
19030 3.06571
24860 3.57981
31250 4.30108
33720 5.38633
30450 7.20408
APPENDICES Grace Dolan, PhD Thesis
217
Appendix J: Friction curves with angular velocity for hydrogel
pairs (cellulose, CAX, CXG) at all CRs
Figure J.1. Shear stress and angular velocity measured over time during the constant rotation
rate step for cellulose hydrogels in water at CR 0.4.
Figure J.2. Shear stress and angular velocity measured over time during the constant rotation
rate step for cellulose hydrogels in water at CR 0.5.
Time (s)
0 5 10 15 20
Sh
ea
r S
tre
ss (
, kP
a)
0.00
0.05
0.10
0.15
0.20
An
gu
lar
ve
locity (
, ra
d/s
)
0.0001
0.001
0.01
0.1
1
Time (s)
0 5 10 15 20
Sh
ea
r S
tre
ss (
, kP
a)
0.00
0.05
0.10
0.15
0.20
0.25
0.30A
ng
ula
r ve
locity (
, ra
d/s
)
0.0001
0.001
0.01
0.1
1
APPENDICES Grace Dolan, PhD Thesis
218
Figure J.3. Shear stress and angular velocity measured over time during the constant rotation
rate step for cellulose hydrogels in water at CR 0.6.
Figure J.4. Shear stress and angular velocity measured over time during the constant rotation
rate step for cellulose hydrogels in water at CR 0.7.
Time (s)
0 10 20 30 40
Shear
Str
ess (
, kP
a)
0.0
0.1
0.2
0.3
0.4
0.5
Angula
r velo
city (
, ra
d/s
)
0.0001
0.001
0.01
0.1
1
Time (s)
0 20 40 60 80 100 120
Shear
Str
ess (
, kP
a)
0.0
0.5
1.0
1.5
Angula
r velo
city (
, ra
d/s
)
0.0001
0.001
0.01
0.1
1
APPENDICES Grace Dolan, PhD Thesis
219
Figure J.5. Shear stress and angular velocity measured over time during the constant rotation
rate step for CAX hydrogels in water at CR 0.4.
Figure J.6. Shear stress and angular velocity measured over time during the constant rotation
rate step for CAX hydrogels in water at CR 0.5.
Time (s)
0 2 4 6 8 10
Sh
ea
r S
tre
ss (
, kP
a)
0.000
0.015
0.030
0.045
0.060
An
gu
lar
ve
locity (
, ra
d/s
)
0.001
0.01
0.1
1
Time (s)
0 5 10 15 20
Shear
Str
ess (
, kP
a)
0.00
0.02
0.04
0.06
0.08
0.10
Angula
r velo
city (
, ra
d/s
)
0.001
0.01
0.1
1
APPENDICES Grace Dolan, PhD Thesis
220
Figure J.7. Shear stress and angular velocity measured over time during the constant rotation
rate step for CAX hydrogels in water at CR 0.6.
Figure J.8. Shear stress and angular velocity measured over time during the constant rotation
rate step for CAX hydrogels in water at CR 0.7.
Time (s)
0 5 10 15 20
Sh
ea
r S
tre
ss (
, kP
a)
0.00
0.05
0.10
0.15
An
gu
lar
ve
locity (
, ra
d/s
)
0.0001
0.001
0.01
0.1
1
Time (s)
0 10 20 30 40
Sh
ea
r S
tre
ss (
, kP
a)
0.00
0.05
0.10
0.15
0.20
0.25
An
gu
lar
ve
locity (
, ra
d/s
)
0.0001
0.001
0.01
0.1
1
APPENDICES Grace Dolan, PhD Thesis
221
Figure J.9. Shear stress and angular velocity measured over time during the constant rotation
rate step for CXG hydrogels in water at CR 0.4.
Figure J.10. Shear stress and angular velocity measured over time during the constant
rotation rate step for CXG hydrogels in water at CR 0.5.
Time (s)
0 5 10 15 20
Shear
Str
ess (
, kP
a)
0.00
0.01
0.02
0.03
Angula
r velo
city (
, ra
d/s
)
0.000
0.005
0.010
0.015
0.020
0.025
0.030
Time (s)
0 5 10 15 20
Sh
ea
r S
tre
ss (
, kP
a)
0.00
0.01
0.02
0.03
0.04
An
gu
lar
ve
locity (
, ra
d/s
)
0.00
0.01
0.02
0.03
0.04
APPENDICES Grace Dolan, PhD Thesis
222
Figure J.11. Shear stress and angular velocity measured over time during the constant
rotation rate step for CXG hydrogels in water at CR 0.6.
Figure J.12. Shear stress and angular velocity measured over time during the constant
rotation rate step for CXG hydrogels in water at CR 0.7.
Time (s)
0 5 10 15 20
Shear
Str
ess (
, kP
a)
0.00
0.01
0.02
0.03
0.04
0.05
0.06
Angula
r velo
city (
, ra
d/s
)
0.00
0.01
0.02
0.03
0.04
Time (s)
0 10 20 30 40
Shear
Str
ess (
, kP
a)
0.00
0.05
0.10
0.15
Angula
r velo
city (
, ra
d/s
)
0.00
0.01
0.02
0.03
0.04
0.05
APPENDICES Grace Dolan, PhD Thesis
223
Appendix K: Force-distance data for SPEEK fibres mats.
Figure K.1. Lateral force-distance curve of SPEEK sample A.
Figure K.2. Lateral force-distance curve of SPEEK sample B.
Lateral Distance ( m)
0 2 4 6 8 10 12 14
Late
ral F
orc
e (
N)
0
10
20
30
40
Lateral Distance ( m)
0 2 4 6 8 10 12 14
Late
ral F
orc
e (
N)
0
10
20
30
40
APPENDICES Grace Dolan, PhD Thesis
224
Figure K.3. Lateral force-distance curve of SPEEK sample C.
Figure K.4. Lateral force-distance curve of SPEEK sample D.
Lateral Distance ( m)
0 2 4 6 8 10 12 14
Late
ral F
orc
e (
N)
0
10
20
30
40
50
60
Lateral Distance ( m)
0 2 4 6 8 10 12 14
Late
ral F
orc
e (
N)
0
20
40
60
80
100
APPENDICES Grace Dolan, PhD Thesis
225
Figure K.5. Lateral force-distance curve of SPEEK sample E.
Lateral Distance ( m)
0 2 4 6 8 10 12 14 16
Late
ral F
orc
e (
N)
0
20
40
60
80
APPENDICES Grace Dolan, PhD Thesis
226
Appendix L: MATLAB code for finding the peaks in lateral
force-distance curves from the dip-and-drag technique
dist = load('NFC_air_distance_b.txt'); %this is for a single trace in a
curve i.e. the cantilever doesn't lift off during this data LatF = load ('NFC_air_Lateral Force_b.txt'); %this is for a single trace
in a curve i.e. the cantilever doesn't lift off during this data
%First step of peak identification. hold off %start new figure Peaks = zeros(size(LatF)); % Make another array to fill up with peaks in
Lateral Force data. for aa = 11:(numel(LatF)-11) % May need to change this range depending
on how many points you % specify on either side of the conditions
below. if ((LatF(aa-10) < LatF(aa)) && (LatF(aa+10) < LatF(aa))) % Identifies a
peak based on the average of % lateral forces a
certain number of points % either side being
less than the point of interest. % You can change
the number of points either side that % is in the
condition based on visual inspection. Peaks(aa) = LatF(aa); % Enter values that satify the conditions of being
a peak. else Peaks(aa) = 0; % Enters a zero if not a peak. %plot (dist,LatF) % Plot raw data. %hold on %scatter (dist,Peaks) % Plots peaks, points that aren't peaks fall on the
origin end end
% (dist,LatF) of local maximums. hold off % Start new graph. [row,col] = find (Peaks); % Find the location of nonzero values of
matrix. r=row; Peaks2 = zeros(numel(r),1); % Start a new matrix for nonzero values of
Peaks matrix. for bb = 1:numel(r); Peaks2(bb,1) = dist(r(bb)); % Distance corresponding to nonzero values of
Peaks matrix. Peaks2(bb,2) = Peaks(r(bb)); % Lateral force corresponding to nonzero
value of Peaks matrix. %plot (dist,LatF) % Plot raw data. %hold on %scatter (Peaks2(:,1),Peaks2(:,2)) % Plot (dist,LatF) of nonzero values
from Peaks matrix. end Peaksdist=(Peaks2(:,1)); % x-values of nonzero values from Peaks matrix. PeakLatF=(Peaks2(:,2)); % y-values of nonzero values from Peaks matrix.
APPENDICES Grace Dolan, PhD Thesis
227
GrpStartPts = zeros(numel(Peaksdist),1); % Make another array to fill up
with the end points of %groupings of peaks where only the %local maximum is an actual peak. for cc = 2:numel(Peaksdist) if Peaksdist(cc)-Peaksdist(cc-1)>0.03 %Conditions to make the end
point of a cluster of % data for which only the local % maximum is relavent. i.e. if % the separation between the % group end point and the next % group start point is at least % 0.1um GrpStartPts(cc) = Peaksdist(cc); % Enter points that satisfy the condition
above. else GrpStartPts(cc) = 0; % Enter zero for points that don't
satisfy the condition above. end end [row2,col2] = find (GrpStartPts); % Find nonzero values of GrpStartPts. r2 = row2; % Rows of nonzero values of
GrpStartPts. GrpStartPtsVal = zeros(numel(r2),1); % Start a new matrix for location
of GrpStartPts in Peaksdist matrix. for dd = 1:numel(r2) GrpStartPtsVal(dd)=GrpStartPts(r2(dd)); end
LocGrpStartPts = zeros(numel(GrpStartPtsVal),1); % Start a new matrix
for location of GrpStartPts in Peaksdist matrix. for ee = 1:numel(GrpStartPtsVal) LocGrpStartPts(ee)=find(Peaksdist==GrpStartPtsVal(ee)); end
LocalMaxLatF = zeros((numel(LocGrpStartPts)+1),1); %Start a new matrix for
local max within groups of peaks identified % by GrpStartPts. for ff=2:(numel(LocGrpStartPts)) LocalMaxLatF(1)= max(PeakLatF(1:(LocGrpStartPts(1)-1))); % First
group of peaks start from first row in matrix to the first % end point
which is the point before that specified in the % first row
of the GrpStartPts matrix. LocalMaxLatF(ff) = max(PeakLatF((LocGrpStartPts(ff-
1)):(LocGrpStartPts(ff)-1))); % range of the matrix of peaks
% bound by the GrpStartPts LocalMaxLatF(numel(LocGrpStartPts)+1)=
max(PeakLatF((LocGrpStartPts(numel(LocGrpStartPts))):numel(PeakLatF))); end
LocalMaxLoc2 = zeros(numel(LocalMaxLatF),2); % Start a new matrix for the
location of local maximums. for gg = 2:(numel(LocalMaxLatF)-1) LocalMaxLoc2(1,:)=find((PeakLatF(1:(LocGrpStartPts(1)-
1)))==LocalMaxLatF(1)); LocalMaxLoc2(gg,:)=((LocGrpStartPts(gg-1)-
1)+find(PeakLatF((LocGrpStartPts(gg-1)):(LocGrpStartPts(gg)-
1))==LocalMaxLatF(gg)));
APPENDICES Grace Dolan, PhD Thesis
228
LocalMaxLoc2(numel(LocalMaxLatF),:)=
(LocGrpStartPts(numel(LocGrpStartPts))-
1)+find(PeakLatF(LocGrpStartPts(numel(LocGrpStartPts)):numel(PeakLatF))==Lo
calMaxLatF(numel(LocalMaxLatF))); % need the first term on the RHS of the equation above because the location
given is the row from the selected region of % the matrix not the whole matrix so you need to correct for that. end
LocalMaxLoc= LocalMaxLoc2(:,1);
LocalMaxdist = zeros(numel(LocalMaxLoc),1); % Start a new matrix for the
distance corresponding to the local maximums. for hh=1:(numel(LocalMaxLoc)) LocalMaxdist(hh)= Peaksdist(LocalMaxLoc(hh)); end
%scatter(LocalMaxdist,LocalMaxLatF) % Plot the local maximums.
%First step of valley identification. hold off %start new figure Valleys = zeros(size(LatF)); % Make another array to fill up with valleys
in Lateral Force data. for kk = 11:(numel(LatF)-11) % May need to change this range depending
on how many points you % specify on either side of the conditions
below. if (LatF(kk-10) > LatF(kk)) && (LatF(kk+10) > LatF(kk)) % Identifies
valleys based on the value of % lateral force a
certain number of points % either side being
greater than the point of interest. % You can change
the number of points either side that % is in the
condition based on visual inspection. Valleys(kk) = LatF(kk); % Enter values that satify the conditions of
being a valley. else Valleys(kk) = 0; % Enters a zero if not a peak. %plot (dist,LatF) % Plot raw data. %hold on %scatter (dist,Valleys) % Plots peaks, points that aren't peaks fall on
the origin end end
% (dist,LatF) of local minimums. %hold off % Start new graph. [row3,col3] = find (Valleys); % Find the location of nonzero values of
matrix. r3=row3; Valleys2 = zeros(numel(r3),1); % Start a new matrix for nonzero values of
Valleys matrix. for ll = 1:numel(r3); Valleys2(ll,1) = dist(r3(ll)); % Distance corresponding to nonzero values
of Valleys matrix.
APPENDICES Grace Dolan, PhD Thesis
229
Valleys2(ll,2) = Valleys(r3(ll)); % Lateral force corresponding to
nonzero value of Valleys matrix. %plot (dist,LatF) % Plot raw data. %hold on %scatter (Valleys2(:,1),Valleys2(:,2)) % Plot (dist,LatF) of nonzero
values from Valleys matrix. end Valleysdist=(Valleys2(:,1)); % x-values of nonzero values from Valleys
matrix. ValleysLatF=(Valleys2(:,2)); % y-values of nonzero values from Valleys
matrix. GrpStartPts2 = zeros(numel(Valleysdist),1); % Make another array to fill up
with the end points of %groupings of valleys where only the %local minimum is an actual valley. GrpStartPts2(1) = Valleysdist(1); for mm = 2:numel(Valleysdist) if Valleysdist(mm)-Valleysdist(mm-1)>0.03 %Conditions to make the
start point of a cluster of % data for which only the local % minimum is relavent. i.e. if % the separation between the % group end point and the next % group start point is at least % 0.1um GrpStartPts2(mm) = Valleysdist(mm); % Enter points that satisfy the
condition above. else GrpStartPts2(mm) = 0; % Enter zero for points that don't
satisfy the condition above. end end [row4,col4] = find (GrpStartPts2); % Find nonzero values of GrpStartPts2. r4 = row4; % Rows of nonzero values of
GrpStartPts. GrpStartPtsVal2 = zeros(numel(r4),1); % Start a new matrix for location
of GrpStartPts2 in Valleysdist matrix. for nn = 1:numel(r4) GrpStartPtsVal2(nn)=GrpStartPts2(r4(nn)); end
LocGrpStartPts2 = zeros(numel(GrpStartPtsVal2),1); % Start a new matrix
for location of GrpStartPts in Valleysdist matrix. for oo = 1:numel(GrpStartPtsVal2) LocGrpStartPts2(oo)=find(Valleysdist==GrpStartPtsVal2(oo)); end
LocalMinLatF = zeros((numel(LocGrpStartPts2)),1); %Start a new matrix for
local min within groups of valleys identified % by GrpStartPts2. for pp=1:numel(LocGrpStartPts2)-1
LocalMinLatF(pp) =
min(ValleysLatF((LocGrpStartPts2(pp)):(LocGrpStartPts2(pp+1)-1))); % range
of the matrix of peaks
% bound by the GrpStartPts2. % Last grouping doesn't doesn't have an end point. make it bound by the % last point in the ValleysLatF matrix. LocalMinLatF(numel(LocGrpStartPts2))=
min(ValleysLatF((LocGrpStartPts2(numel(LocGrpStartPts2))):numel(ValleysLatF
)));
APPENDICES Grace Dolan, PhD Thesis
230
end
LocalMinLoc2 = zeros(numel(LocalMinLatF),2); % Start a new matrix for the
location of local minimums. for qq = 1:numel(LocalMinLatF)-1 LocalMinLoc2(qq,:)=(LocGrpStartPts2(qq)-
1)+find((ValleysLatF((LocGrpStartPts2(qq)):(LocGrpStartPts2(qq+1)-
1))==LocalMinLatF(qq))); % Need the first term on the RHS of the equation above because the
location given is the row from the selected region of % the matrix not the whole matrix so you need to correct for that. %LocalMinLoc2(numel(LocalMinLatF),:)=
LocGrpStartPts2(numel(LocGrpStartPts2)); % This last line is required because the number of minimums is 1 more % than the number of grouping start points. LocalMinLoc2(numel(LocalMinLatF),:)=
(LocGrpStartPts2(numel(LocGrpStartPts2))-
1)+find(ValleysLatF(LocGrpStartPts2(numel(LocGrpStartPts2)):numel(ValleysLa
tF))==LocalMinLatF(numel(LocalMinLatF))); end
LocalMinLoc= LocalMinLoc2(:,1);
LocalMindist = zeros(numel(LocalMinLoc),1); % Start a new matrix for the
distance corresponding to the local minimums. for rr=1:(numel(LocalMinLoc)) LocalMindist(rr)= Valleysdist(LocalMinLoc(rr)); end
%scatter(LocalMindist,LocalMinLatF) % Plot the local minimums.
APPENDICES Grace Dolan, PhD Thesis
231
Appendix M: MATLAB code for measure the slope before the
peaks in the lateral force-distance curves from the dip-and-drag
technique
% Slope before local maximums LocMaxdist = zeros(numel(LocalMaxdistreal),1); % Start a new matrix for
location of local maximum in raw data. for ii = 1:numel(LocalMaxdistreal); LocMaxdist(ii) = find(dist==LocalMaxdistreal(ii)); end linregbeforeMax = zeros(numel(LocalMaxdistreal),2); % Start a new matrix
for linear regression of the curve before local maximum. for jj = 1:numel(LocalMaxdistreal); linregbeforeMax(jj,:)=polyfit(dist((LocMaxdist(jj)-
30):LocMaxdist(jj)),LatF((LocMaxdist(jj)-30):(LocMaxdist(jj))),1); end Slopebeforemax=linregbeforeMax(:,1) LatFfit2 = zeros(31,numel(LocalMaxdistreal)); for jjj=1:numel(LocalMaxdistreal); LatFfit2(:,jjj) =
polyval(linregbeforeMax(jjj,:),(dist((LocMaxdist(jjj)-
30):(LocMaxdist(jjj))))); end
APPENDICES Grace Dolan, PhD Thesis
232
Appendix N: Solving 3 non-linear simultaneous equations in
MATLAB
function F = root3d(x) F(1)=exp(436.51-0.017*(x(2)).^0.5.*log(x(2))+1.05*log(x(2))-
130.2*1.45+24.86*1.45*log(1.45)-434.64*exp(-
1.45/11.67)+25.95*(1.45)^0.5.*log(1.45)+87.97*(1.45).^0.5)-0.15; F(2)=10*exp(-5.21-0.088./x(3)-2.67*exp(-x(3))-6*10^-4*x(2)+6*10^-
3*(x(2)).^0.5+log(x(2)))-1.79; F(3)=0.5*((x(2)*(3*10^-3)*(x(3))^2))-x(1);
fun = @root3d;
x0=[0.002,16,0.3];
x=fsolve(fun,x0)
APPENDICES Grace Dolan, PhD Thesis
233
Appendix O: Interfacial yield stress and G’ for triplicate hydrogel
pairs for each treatment (control, YOAJ, WWY, RKKQ, D82)
Figure O.1. Interfacial yield stress against G’ for 2 different (with different symbols)
hydrogel pairs that are untreated (from the same batch of cellulose hydrogels as those treated
with YOAJ in Figure O.2).
Figure O.2. Interfacial yield stress against G’ for triplicate (with different symbols) hydrogel
pairs that are treated with YOAJ.
G' (kPa)
0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Inte
rfa
cia
l Y
ield
Str
ess (
Pa)
20
40
60
80
100
120
140
G' (kPa)
0.2 0.4 0.6 0.8 1.0 1.2 1.4
Inte
rfa
cia
l Y
ield
Str
ess (
Pa)
0
20
40
60
80
100
APPENDICES Grace Dolan, PhD Thesis
234
Figure O.3. Interfacial yield stress against G’ for triplicate (with different symbols) hydrogel
pairs that are untreated (from the same batch of cellulose hydrogels as those treated with
WWY, RKKQ, D82 in Figure O.4 – O.6).
Figure O.4. Interfacial yield stress against G’ for triplicate (with different symbols) hydrogel
pairs that are treated with WWY.
G' (kPa)
0.0 0.4 0.8 1.2 1.6 2.0
Inte
rfa
cia
l Y
ield
Str
ess (
Pa)
0
20
40
60
80
100
120
140
G' (kPa)
0.0 0.5 1.0 1.5 2.0 2.5 3.0
Inte
rfa
cia
l Y
ield
Str
ess (
Pa)
0
20
40
60
80
100
120
140
160
180
APPENDICES Grace Dolan, PhD Thesis
235
Figure O.5. Interfacial yield stress against G’ for triplicate (with different symbols) hydrogel
pairs that are treated with RKKQ.
Figure O.6. Interfacial yield stress against G’ for triplicate (with different symbols) hydrogel
pairs that are treated with D82.
G' (kPa)
0 1 2 3 4
Inte
rfa
cia
l Y
ield
Str
ess (
Pa)
0
50
100
150
200
250
G' (kPa)
0.0 0.5 1.0 1.5 2.0 2.5
Inte
rfa
cia
l Y
ield
Str
ess (
Pa)
0
20
40
60
80
100
APPENDICES Grace Dolan, PhD Thesis
236
Appendix P: Raw data from dip-and-drag experiments on
Cellulose, CAX, and CXG networks with and without expansins
Figure P.1. Lateral deflection-distance curve for Cellulose network in 20 mM Hepes buffer
(pH 7.5).
Figure P.2. Lateral deflection-distance curve for Cellulose network in 20 mM Hepes buffer
(pH 7.5) with 200 g/mL of YOAJ expansin.
Distance ( m)
0 2 4 6 8 10
Late
ral D
eflection (
V)
0.0
0.2
0.4
0.6
0.8
1.0
Distance ( m)
0 2 4 6 8
La
tera
l D
efle
ctio
n (
V)
0.0
0.2
0.4
0.6
0.8
1.0
APPENDICES Grace Dolan, PhD Thesis
237
Figure P.3. Lateral deflection-distance curve for CAX network in 20 mM Hepes buffer (pH
7.5).
Figure P.4. Lateral deflection-distance curve for CAX network in 20 mM Hepes buffer (pH
7.5) with 200 g/mL of YOAJ expansin.
Distance ( m)
0 2 4 6 8 10 12
La
tera
l D
efle
ctio
n (
V)
0.0
0.5
1.0
1.5
2.0
Distance ( m)
0 2 4 6 8 10 12
Late
ral D
eflection (
V)
0
1
2
3
4
APPENDICES Grace Dolan, PhD Thesis
238
Figure P.5. Lateral deflection-distance curve for CXG network in 20 mM Hepes buffer (pH
7.5).
Figure P.6. Lateral deflection-distance curve for CXG network in 20 mM Hepes buffer (pH
7.5) with 200 g/mL of YOAJ expansin.
Distance ( m)
0 5 10 15 20
La
tera
l D
efle
ctio
n (
V)
0.0
0.5
1.0
1.5
2.0
2.5
Distance ( m)
0 5 10 15 20
La
tera
l D
efle
ctio
n (
V)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
APPENDICES Grace Dolan, PhD Thesis
239
Appendix Q: Histogram of peak heights for Cellulose, CAX, and
CXG networks with and without expansins
Figure Q.1. Histogram of peaks heights (n = 166) from lateral deflection-distance curves for
Cellulose networks in 20 mM Hepes buffer (pH 7.5).
Figure Q.2. Histogram of peaks heights (n = 200) from lateral deflection-distance curves for
Cellulose networks in 20 mM Hepes buffer (pH 7.5) with 200 g/mL of YOAJ expansin.
Peak Height (V)
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Cou
nt
0
20
40
60
80
100
Peak Height (V)
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Cou
nt
0
20
40
60
80
100
APPENDICES Grace Dolan, PhD Thesis
240
Figure Q.3. Histogram of peaks heights (n = 158) from lateral deflection-distance curves for
CAX networks in 20 mM Hepes buffer (pH 7.5).
Figure Q.4. Histogram of peaks heights (n = 183) from lateral deflection-distance curves for
CAX networks in 20 mM Hepes buffer (pH 7.5) with 200 g/mL of YOAJ expansin.
Peak Height (V)
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Cou
nt
0
10
20
30
40
50
60
Peak Height (V)
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Count
0
10
20
30
40
50
60
APPENDICES Grace Dolan, PhD Thesis
241
Figure Q.5. Histogram of peaks heights (n = 64) from lateral deflection-distance curves for
CXG networks in 20 mM Hepes buffer (pH 7.5).
Figure Q.6. Histogram of peaks heights (n = 103) from lateral deflection-distance curves for
CXG networks in 20 mM Hepes buffer (pH 7.5) with 200 g/mL of YOAJ expansin.
Peak Height (V)
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Cou
nt
0
10
20
30
40
50
Peak Height (V)
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Count
0
10
20
30
40
50