Fibre-Reinforced Composites with Nacre-Inspired Interphase

Home , Nacre

Fibre-Reinforced Composites with Nacre-Inspired Interphase:

A Route Towards High Performance Toughened Hierarchical

Composites

François De Luca

A Thesis submitted in fulfilment of the requirements for the degree of Doctor of Philosophy

and the Diploma of Imperial College London

Chemical Engineering Department

Imperial College London

From June 2013 to December 2016

Abstract

Conventional fibre-reinforced polymer composite materials are well known for their high strength, stiffness, low weight and chemical resistance but composites do fail catastrophically, in a brittle manner, with little prior warning. When a fibre breaks in tension, shear stresses transfer load previously carried by the broken fibre to neighbouring fibres through the matrix, leading to local stress concentrations. As tensile loading continues, fibre breaks accumulate in the composite, eventually leading to the formation of a critical cluster, which triggers the failure of the composite. The aim of this research was to develop a novel hierarchical composite architecture consisting of fibres decorated with a nanostructured coating embedded in a matrix. A high performance and tough nanostructured composite interphase, inspired by nacre, should provide additional toughness in tension. A

Layer-by-Layer assembly method was used to assemble inorganic nanometre-wide platelets and a polyelectrolyte into a well-organised nanostructure, mimicking the “brick-and-mortar” architecture of nacre, which was developed and characterised. The nanostructure was successfully deposited around conventional reinforcing-fibres, such as carbon and glass fibres, and allowed for absorption of the energy arising from fibre breaks and substantial increase in debonding toughness in single fibre composite models. Impregnated fibre bundle composites were manufactured and tested in tension, which exhibited an increased tensile strength, strain to failure and work of fracture when the nanostructured composite interphase was incorporated. This work was part of the HiPerDuCT programme grant, collaboration between the departments of Aeronautics, Chemical Engineering, Chemistry and Mechanical

Engineering of Imperial College London and the University of Bristol.

Declaration of Originality

This manuscript is a description of the work achieved by the author in the Department of Chemical Engineering of Imperial College London between June 2013 to December 2016, under the supervision of Prof. Alexander Bismarck and Prof. Milo Shaffer. Except where acknowledged, the material presented is the original work of the author and no part of it has been submitted for a degree at Imperial College London or any other university.

The copyright of this thesis rests with the author and is made available under a

Creative Commons Attribution Non-Commercial No Derivatives licence. Researchers are free to copy, distribute or transmit the thesis on the condition that they attribute it, that they do not use it for commercial purposes and that they do not alter, transform or build upon it.

For any reuse or redistribution, researchers must make clear to others the licence terms of this work.

Acknowledgments

First of all, my sincere gratitude goes to my supervisors, Prof. Milo Shaffer and Prof.

Alexander Bismarck for entrusting a cutting-edge research PhD project to me and continuously providing me with guidance, inspiration and excellence. I am also grateful for their precious time they dedicated to follow my research, answer my concerns and transfer their true knowledge and passion.

I would like to thank Jonny Blaker and Robert Menzel for their expertise and the successful transfer of the research project to me. They have been a great support during my first year to reach my independence and help me learn the skills required to pursuit a PhD.

I am grateful to Joshua Elsdon for designing, building and maintaining the dipping robot throughout the entire research project.

My keen appreciation goes to Adam Clancy for helping me modifying the surface of glass and carbon fibres and Noelia Rubio Carrero for characterising the surface of oxidised carbon fibres by XPS.

I am indebted to Mahmoud Ardakani for the quality of his SEM training and regular help as well as his assistance on TEM and Cati Ware for her help on the preparation of a

TEM sample using the FIB technique. Also, I would like to thank Sergio Sernicola for his availability and conducting in-situ SEM nanoindentation and Richard Sweeney for his assistance on XRD rocking curve acquisition.

I would like to thank my officemates, Konstanze Seidler, Hele Diao, Tomi Herceg, and Rosminah Shamsuddin for their support, availability, helpful discussion and consideration. A special thank goes to David Anthony for his in depth discussion, availability to assist me on composite testing and valuable help and advice.

All my colleges from the PaCE and NanoHAC groups deserves sincere thanks for their support, friendliness and enthusiasm to share their experience and knowledge. So I would like to thank Henry Maples, Wonjun Lee, Stephen Hodge, Hin Chun Yau, Min Tang,

Koon Yang, Hanna Leese, Edward White, Foivos Markoulidis, Mustafa Bayazit, Jonathan

Weiner, Jonathan Davison, Cynthia Sheng Hu, Martina De Marco, Heather Au, Aaron

Thong, Chris Roberts, Alice Leung, Eileen Brandley, and Sandy Fisher. I would like thank you, Eero Kontturi and Katri Kontturi, for your kindness and interesting discussions.

Special thanks go to Gael Grail, Derrick Fam, Robert Woodward, Hugo De Luca and

Marcel Lorenz for their friendship and time spent together outside work, which made my life and experience in London a real pleasure. Also, I am very grateful to my family and friends, who showed constant support throughout my PhD. Finally, I would like to thank the people from the Chemical Engineering department I had lunch with daily.

Lastly, but not the least, I am very grateful to all the members of the HiPerDuCT programme grant team; Prof. Michael Wisnom, Prof. Paul Robinson, Prof. Kevin Potter,

Soraia Pimenta, Gergely Czel, HaNa Yu, Jakub Rycerz, James Finley, James Serginson,

James Trevarthen, Jingjing Sun, Joel Henry, Jonathan Fuller, Marco Longana, Meisam

Jalalvand, Mohammad Fotouhi, Omar Bacarreza, Putu Suwarta, Stefano Del Rosso, Thomas

Pozegic, Xun Wu, for all the discussions and advice during the whole PhD research project. I would also like to thank the UK Engineering and Physical Sciences Research Council

(EPSRC) (grant EO/I02946X/1) and Alexander Bismarck’s F-account for funding.

Contents

List of Publications/Patent ...... 12 List of Figures ...... 14 List of Tables ...... 25 List of Abbreviations ...... 26

1. Introduction ...... 27 1.1 Motivation ...... 27 1.2 Objectives ...... 29 1.3 Approach ...... 31 1.4 Structure of the thesis ...... 35

2. Literature Review...... 37 2.1 Fibre-reinforced polymer composites for load bearing applications ...... 37 2.1.1 Fundamentals of fibre-reinforced composites ...... 37 2.1.2 Role of fibre-reinforced composite interfaces ...... 39 2.1.3 Catastrophic failure of fibre-reinforced composites ...... 40 2.2 Engineered fibre-reinforced composite interphases for energy absorption ...... 41 2.2.1 Energy absorbing fibre polymer coatings ...... 42 2.2.2 Roughened interphases for interfacial mechanical frictions ...... 44 2.2.3 Incorporation of nanofillers in interphase to toughen composites ...... 45 2.2.4 Nanostructured interphases for toughened composite ...... 50 2.3 “Brick-and-mortar” structure of nacre ...... 53 2.3.1 Hierarchical structure of nacre ...... 53 2.3.2 Toughening mechanisms occurring in nacre ...... 55 2.3.3 Tensile and shear responses of nacre ...... 60 2.3.4 Scaling behaviour of nacre “brick-and-mortar” structure ...... 62 2.3.5 Flaw tolerance of platelets ...... 64 2.4 Nacre-inspired structures...... 66 2.4.1 Assembly methods of nacre-inspired structures ...... 67

2.4.2 Micrometre lengthscale nacre-inspired structures ...... 69 2.4.3 Nanometre lengthscale nacre-inspired structures ...... 73 2.4.4 Hierarchical, fibre and ternary nacre-inspired structures ...... 78 2.5 Summary ...... 84

3. Materials, Methods & Characterisation ...... 87 3.1 Materials for the development of hierarchical composites with nanostructured interphases ...... 87 3.1.1 Polyelectrolyte and platelets for nanostructured coating/interphase assembly .. 88 3.1.2 Reinforcing fibres for hierarchical composites with nanostructured interphases ...... 89 3.1.3 Polymer resins for the manufacture of hierarchical model and bundle composites with nanostructured interphases ...... 90 3.2 Methods to develop, manufacture and test hierarchical composites with nanostructured interphases ...... 90

3.2.1 Synthesis of Mg2-Al-CO3 LDH platelets ...... 90 3.2.2 Layer-by-Layer assembly procedure to assemble nanostructured coatings ...... 91 3.2.2.1 Layer-by-Layer deposition of nanostructured coatings on flat glass substrates ...... 91 3.2.2.2 Deposition of nanostructured coatings on bundles of glass or carbon fibres . 93 3.2.3 Carbon fibre surface treatment to increase charge density ...... 95 3.2.4 Fibre bundle composite preparation ...... 97 3.3 Characterisation of nanostructured coating/interphase, single fibre composite models and fibre bundle composites ...... 98 3.3.1 Determination of surface charge density of LDH platelets by ζ -potential measurements ...... 99 3.3.2 Determination of surface charge density of glass and carbon fibres by streaming ζ -potential measurements ...... 100 3.3.3 X-Ray diffraction techniques for the determination of LDH platelets and

(LDH/PSS)n coatings morphology ...... 102

3.3.4 UV-Vis spectroscopy to control the deposition of (LDH/PSS)n multilayer coatings on quartz slides ...... 104 3.3.5 Morphology of monolayers and multilayer coatings ...... 105 3.3.5.1 Scanning electron microscopy of monolayers and multilayer coatings ...... 105

3.3.5.2 Transmission electron microscopy of LDH platelets and (LDH/PSS)n coating cross-section ...... 106

3.3.6 Elemental composition of (LDH/PSS)n coating cross-section by energy dispersive X-ray spectroscopy ...... 107 3.3.7 Composition of carbon fibre surfaces determined by X-ray photoelectron spectroscopy ...... 108 3.3.8 LDH monolayer topography determined by atomic force microscopy ...... 109

3.3.9 Determination of organic content of (LDH/PSS)n coatings by thermogravimetric analysis...... 109

3.3.10 Determination of the mechanical properties of (LDH/PSS)n coatings by nanoindentation ...... 110 3.3.11 Nanostructured interphase mechanical characterisation techniques ...... 114 3.3.11.1 Determination of the interfacial properties and debonding behaviour of the nanostructured interphase by single fibre pull-out tests ...... 114 3.3.11.2 Determination of the interfacial properties and stress concentration of the nanostructured interphase by single fibre fragmentation tests ...... 117 3.3.12 Tensile properties of fibre bundle composites ...... 120

4. Development of Planar “Brick-and-Mortar” Nanostructured Coatings ...... 123 4.1 Introduction ...... 124 4.2 LDH platelet reinforcements ...... 124 4.2.1 Width distribution of LDH platelets ...... 125 4.2.2 LDH platelet structure and anisotropy ...... 127 4.2.3 LDH platelet surface charge ...... 128 4.3 Layer-by-Layer assembly of nanostructured coatings ...... 130 4.3.1 LDH and PSS monolayer deposition on flat substrates ...... 130

4.3.2 (LDH/PSS)n multilayer deposition on flat substrates ...... 133

4.4 Morphology of nanostructured (LDH/PSS)n coatings ...... 138

4.4.1 Organic content of nanostructured (LDH/PSS)n coatings ...... 139

4.4.2 Packing of LDH platelets in nanostructured (LDH/PSS)n coatings...... 142

4.4.3 LDH platelet alignment in (LDH/PSS)n coatings ...... 144

4.5 Mechanical properties of nanostructured (LDH/PSS)n coatings ...... 146

4.5.1 Mechanical properties of nanostructured (LDH/PSS)n coatings by shallow nanoindentation ...... 146

4.5.2 Plasticity of nanostructured (LDH/PSS)n coatings determined by deep nanoindentation ...... 149

4.5.3 Deformation mechanism of nanostructured (LDH/PSS)n coatings characterised by SEM in-situ nanoindentation ...... 152

4.5.4 Crack deflection in nanostructured (LDH/PSS)n coatings ...... 155

4.6 Towards the addition of a level of hierarchy in (LDH/PSS)n nanostructured coatings 156 4.7 Summary ...... 158

5. Transfer of Nanostructured Coatings onto Glass Fibres and their Properties as Composite Interphases ...... 161 5.1 Introduction ...... 162 5.2 Glass fibre surface charge density ...... 163

5.3 Adapting the Layer-by-Layer assembly of multilayer (LDH/PSS)n coatings around fibres in bundles ...... 164 5.3.1 LDH monolayer deposition on glass fibre bundles...... 165

5.3.2 Deposition of (LDH/PSS)n multilayer coatings on glass fibre bundles ...... 166 5.3.3 Morphology of nanostructured coatings deposited on glass fibre bundles ...... 167 5.3.4 Mechanical properties of nanostructured coatings deposited on glass fibre bundles ...... 169 5.4 Nanostructured interphase properties of single glass fibre model composites ...... 172 5.4.1 Determination of the impact of the nanostructured interphase on glass fibre debonding behaviour by single fibre pull-out ...... 172 5.4.2 Effect of the nanostructured interphase on interfacial shear strength and stress concentration in single glass fibre composite model fragmentation ...... 176 5.5 Summary ...... 179

6. Transfer of Nanostructured Coatings onto Carbon Fibres and their Properties as Composite Interphases ...... 181 6.1 Carbon fibre surface treatment to improve surface charge density ...... 182

6.2 Layer-by-Layer deposition of PDDA/(PSS/LDH)n coatings on oxidised carbon fibre bundles ...... 188 6.3 Properties of a nanostructured interphase in single carbon fibre model composites ... 190 6.3.1 Determination of the impact of the nanostructured interphase on carbon fibre debonding behaviour by single fibre pull-out ...... 190 6.3.2 Effect of a nanostructured interphase on interfacial shear strength and stress concentration in single carbon fibre composite model fragmentation ...... 193 6.4 Summary ...... 197

7. Hierarchical Composites Containing a Nanostructured Interphase ...... 199 7.1 Introduction ...... 200 7.2.Wettability of a nanostructured coatings by the epoxy resin ...... 200 7.3 Preparation of impregnated bundle composites ...... 201 7.4 Tensile properties of glass fibre bundle composites ...... 203 7.5 Tensile properties of carbon fibre bundle composites ...... 208 7.6 Summary ...... 212

8. Conclusions and Further Work ...... 214 8.1 Summary of achievements ...... 214 8.2 Implications ...... 217 8.3 Way forward ...... 218

Bibliography ...... 222

Appendix 1: Platelet Dimensions on Fibres...... 242 Appendix 2: FIB Sectioning of Nanostructured Coating ...... 245 Appendix 3: AE Tensile Grip ...... 250 Appendix 4: Quaternary-Amine Terminated Fibre Surfaces ...... 252 Appendix 5: Impact of PDDA Precursor Layer on Nanostructured Coating Adhesion ...... 257 Appendix 6: Impact of Organic Phase Proportion on Nanostructured Interphase Mechanical Properties...... 259

Appendix 7: PDDA/(PSS/LDH)n Transfer onto As-Received Carbon Fibres...... 261

List of Publications/Patent

Journal Papers

Francois De Luca, Robert Menzel, Jonny Blaker, John Birkbeck, Alexander Bismarck, Milo

S. P. Shaffer. Nacre-nanomimetics: Strong, Stiff and Plastic. ACS Appl. Mater. & Interfaces,

2015, 7, 26783-26791.

Robert Woodward, Francois De Luca, Aled D. Roberts, Alexander Bismarck. High Surface

Area, Emulsion-Templated Carbon Foams by Activation of PolyHIPEs Derived from

Pickering Emulsions. Materials, 2016, 9 (9), 776.

Francois De Luca, Giorgio Sernicola, Milo S. P. Shaffer, Alexander Bismarck. “Brick-and-

Mortar” Nanostructured Interphase for Fiber-Reinforced Polymer Composites. Ready for submission, ACS Appl. Mater. & Interfaces, expected 2017.

Francois De Luca, Adam Clancy, Milo S. P. Shaffer, Alexander Bismarck. Hierarchical

Carbon Fiber-Reinforced Polymer Composites Inspired by Nature. In preparation. Expected

2017.

Conference Papers & Oral Presentations

Francois De Luca, Robert Menzel, Jonny Blaker, John Birkbeck, Alexander Bismarck, Milo

S. P. Shaffer. Self-assembled “Brick-and-Mortar” Nanostructure Inspired by Nature: A Route

Towards High Mechanical Performance Nanocomposites. 20th International Conference on

Composite Materials, 19-24/07/2015, Copenhagen, Denmark.

Francois De Luca, Robert Menzel, Jonny Blaker, John Birkbeck, Alexander Bismarck, Milo

S. P. Shaffer. Anisotropic Nanostructures Inspired by Nature for Energy Absorbing

Composite Interfaces. 17th European Conference on Composite Materials, 27-30/06/2016,

Munich, Germany.

Poster presentations

Britannia Vondrasek, Francois De Luca, Robert Menzel, Alexander Bismarck, Milo S. P.

Shaffer. Titania Nanoplatelet Reinforced Films via Layer-by-Layer Deposition. MRS Spring

2014 Meeting & Exhibit, San Francisco, USA.

Francois De Luca, Robert Menzel, Jonny Blaker, John Birkbeck, Alexander Bismarck, Milo

S. P. Shaffer. Anisotropic Nanostructures Inspired by Nature for Energy Absorbing

Composite Interfaces. 17th European Conference on Composite Materials, 27-30/06/2016,

Munich, Germany.

Patent

Francois De Luca, Milo S. P. Shaffer, Alexander Bismarck. Nacre-like decorated fibre for hierarchical ductile composite. Patent application number: 1621494.2. Imperial Innovations.

Application filling date: 16/12/2016.

List of Figures

Figure 1.1: Schematic of conventional fibre-reinforced composite failure mechanism in tension. A fibre break leads to local stress transfer to the neighbouring fibres through matrix shearing (A). After breakage the fibre slips within the matrix over a short length, while further fragments may correlate, giving rise to local stress concentrations (B). As the tensile load increases, fibre fragments occur randomly throughout the material due to stochastic distribution of fibre strength (C), which eventually triggers catastrophic failure through the formation of a critical cluster (D)...... 28 Figure 1.2: Schematic of the expected toughening mechanisms occurring in the nanostructured interphase of hierarchical composites and the associated tensile response of a hierarchical composite. Energy absorption and dissipation along fibre length by crack deflection in the anisotropic nanostructured interphase (i) and progressively arrested uncorrelated fibre slippages through strain hardening of the interphase in shear (ii) (A). Tensile response of conventional and hierarchical fibre-reinforced polymer composites: formation of the critical cluster of fibre breaks triggering the sudden failure of conventional composite (i), stress dissipation in the nanostructured interphase leading to uncorrelated fibre fragments (ii) and subsequent stable fibre slippage (iii) (B)...... 31 Figure 1.3: An analytical approach for the adaptation of a “brick-and-mortar” nanostructured coating around a fibre. Arrangement of platelets around the circumference of a fibre as a function of the number of platelets, deposited over a polymer precursor layer. Wp, dpol, ∆dpol, dr and rf are the platelet width, polymer layer thickness, polymer layer fluctuation, platelet tangential deviation and fibre radius, respectively...... 33 Figure 1.4: Suitable platelet dimensions to enable conformal LbL deposition of the desired “brick-and-mortar” nanostructure on reinforcing fibres. Platelet dimension range defined by the maximum platelet width (wp,max) as a function of the fibre diameter (df)and the minimum platelet width (wp,min) for a platelet aspect ratio, s = 10, and an inorganic phase of at least 90 vol.% ...... 34 Figure 2.1: Formation of a critical cluster of fibre breaks triggering the catastrophic failure of the composite. X-ray radiograph of a quartz-epoxy composite when a critical cluster of fibre breaks forms (A) and triggers sudden failure of a composite (B). Adapted from Aroush et al.33 ...... 41 Figure 2.2: Schematics showing methods to deposit graphene oxide nanosheets onto reinforcing fibres. Electrophoretic deposition of GO nanosheets onto carbon fibres (A), dip coating carbon fibres with GO nanosheets (B) and direct grafting of GO nanosheets onto carbon and glass fibres (C and D, respectively). Adapted from references.91,92,94,97 46 Figure 2.3: Scanning electron microscopy images of composite interphases containing graphene oxide nanosheets and graphite nanoplatelets. Various concentrations of GO nanosheets and graphite nanoplatelets deposited onto carbon fibres through addition to sizing solution (A-D and E-F, respectively). Monolayers and aggregates of GO

nanosheets deposited onto carbon fibres through electrophoretic deposition (G and H, respectively). Single graphene oxide nanosheets grafted on carbon fibres (I). Adapted from references.88,91,94,95 ...... 49 Figure 2.4: Macro-micro finite element simulation of hierarchical composites with nanostructured interphases. Schematic of the structure of hierarchical fibre-reinforced composite with nanosheet reinforcements (A). Top and lateral view of simulated interphases as well as calculated crack path of aligned exfoliated nanosheets, aligned clusters, randomly orientated nanosheets and randomly distributed clusters (B, C, D and E, respectively). Adapted from Dai et al.102 ...... 51 Figure 2.5: Crack deflection in a layered interphase of ceramic matrix composites. Transmission electron microscopy images of (PyC-SiC)n multilayer interphase (A) and boron nitride layered crystal structure interphase (B). Adapted from references.105,106 ... 52 Figure 2.6: Representation of the hierarchical structure of natural nacre at various length scales. Abalone shell (A), inner nacreous aragonite layer “N” and prismatic calcite layer “P” (B), growth bands (C), aragonite platelets organised in columns (D-F) and aragonite platelet roughness arising from aragonite nanograins and asperities (G). Copied from Luz et al.114 ...... 54 Figure 2.7: Schematic of natural nacre deformation and mechanical response as a function of platelet aspect ratio. Fracture mechanisms of “brick-and-mortar” structures containing platelets with post-critical, s > sc, (platelet fracture) and sub-critical, s < sc, (platelet pull-out) aspect ratio according to the shear lag model (A). Associated variation of tensile strength as a function of the aspect ratio of the reinforcements (B). Adapted from Wang et al.13 ...... 57 Figure 2.8: Models of toughening mechanisms occurring in natural nacre over multiple length scales. Mineral bridging, nanoasperities and waviness at the platelet length scale during platelet pull-out, providing interlocking and crack deflection. Rotation and rearrangement of nanograins at the platelet surface length scale (B). Unfolding of molecular sacrificial domains in the organic matrix (C). Adapted from Kakisawa et al.148 ...... 58 Figure 2.9: Crack propagation within natural nacre. Optical microscope images of a crack propagating in nacre and the frontal zone ahead of the crack as evidenced by whitening (A). Schematic of the propagation of a crack in nacre, presenting crack bridging and deflection in the material as well as platelet pull-outs in the frontal zone (B). Copied from Bekah et al.5 ...... 59 Figure 2.10: Tensile and shear responses of natural nacre. Experimental tensile stress- strain curve and associated schematic of deformation mechanism of nacre (A and B, respectively). Experimental shear stress-strain curve and associated schematic deformation mechanism of nacre – platelet waviness leading to a resistance to sliding and, therefore, lateral expansion (C and D, respectively). Copied from Espinosa et al.16 61 Figure 2.11: Toughness amplification in “brick-and-mortar” structures. Effect of platelet size (t) on the toughness (J), with a fixed aspect ratio (s = 10) and platelet volume fraction (φ = 0.9). Adapted from reference.5 ...... 64 Figure 2.12: Most commonly used methods to produce nacre-inspired composites. Vacuum-assisted filtration (A), vacuum evaporation (B), solution casting (C), electrophoretic deposition (D) and layer-by-layer assembly (E). Adapted from references.165,173,183,184,190 ...... 67

Figure 2.13: Layer-by-layer deposition of multilayer coating. Schematic of the dipping of a substrate into polyelectrolyte solutions (A) and the associated Layer-by-Layer assembly of a multilayer coating (B). Copied from Decher et al.193 ...... 69 Figure 2.14: Scanning electron microscopy images of nacre-inspired composite structures at the natural micrometre length scale. Composites containing 11 and 60 vol.% of Al2O3 platelets in chitosan (A and B, respectively). Composites made of LDH/PVA with 2 and 4 wt.% of PVA (C and D, respectively). CaCO3/gelatine composite film containing 95 wt.% and 76 wt.% of CaCO3 (E and F, respectively). Adapted from references.14,187,211 ...... 70 Figure 2.15: Tensile response of nacre-inspired composites at the natural micrometre length scale. Stress-strain curves of composites containing 0 to 15 vol.% of Al2O3 platelets (Vp) in chitosan (A) and stress-strain curves of composites made of LDH/PVA containing 96.9, 69.7, 23.2 and 14.1 wt.% of LDH (“4”, “3”, “2” and “1 wt.% PVA/LDH”, respectively) (B). Adapted from references.14,211 ...... 72 Figure 2.16: Scanning electron microscopy images of layered nanocomposites. Cross- sections of clay nanosheet/PVA layered nanocomposites with different platelet aspect ratio (25, 140, 260, 750 and 3500) and polymer content from about 55 to 25 wt.% (A, B, C and D, respectively). Cross-sections of graphene oxide/sodium alginate layered nanocomposites with a graphene oxide nanosheet proportion of 0, 18.6, 56.3 and 100 vol.% (E, F, G and H, respectively). Copied from references.180,184 ...... 74 Figure 2.17: Tensile response of layered nanocomposites at the nanometre length scale. Stress-strain curves of layered nanocomposite containing synthetic clay nanosheets with varying aspect ratio (A). Stress-strain curves of layered nanocomposite made of GO, reduced GO, GO (95 wt.%)/chitosan, reduced GO (95 wt.%)/chitosan and chitosan films (curves 1, 2 3, 4 and 5, respectively) (B). Adapted from references. 174,184 ...... 76 Figure 2.18: Summary of the different interfacial interactions utilised in layered nanocomposites containing graphene oxide nanosheet reinforcements. Non-covalent bonding and covalent bonding interactions between graphene oxide nanosheets and polymer binder as well as synergistic interactions. Copied from Zhang et al.160 ...... 77 Figure 2.19: Manufacture of multi-level hierarchical “brick-and-mortar” nanocomposites. Processing steps to making hierarchical structure of TiO2 particles encapsulated in PMMA in a radical emulsion polymerisation step and a second polymer matrix (PVB) (A). Scanning electron microscopy images of a PVB-coated agglomerate with 2 levels of hierarchy (B) and of the interface between first-level particles and PVB interlayer at different magnifications (C). Copied from Brandt et al.231 ...... 79 Figure 2.20: Morphology of nacre-inspired spun fibres. Low (50 μm scale bar) and high (2.5 μm scale bar) magnification scanning electron microscopy images of belt-like shape spun (A and D, respectively), spun/twisted (B and E, respectively) and further twisted (C and F, respectively) nacre-inspired fibres made of graphene oxide nanosheets incorporated into PVA. Copied from Zhang et al.238 ...... 81 Figure 2.21: Deformation mechanisms in tension of ternary layered nanocomposites. Stress-strain curves of binary and ternary layered nanocomposites (A). Scanning electron microscopy images of fractured ternary layered nanocomposites and illustrative schematic of the fracture mechanism for clay/cellulose nano-fibril/chitosan , GO/clay/PVA and GO/MoS2/polyurethane (B, C and D, respectively). Copied from references.173,191,235,237 ...... 83

Figure 2.22: Mechanical properties of nacre-inspired layered nanocomposites. Summary of the tensile strength and toughness of graphene oxide-based layered nanocomposites (left) and schematic of the synergistic effects and ternary structures of layered nanocomposite (right). Copied from references.160,224 ...... 84 Figure 3.1: Structure of LDH platelet. Schematic representation of the cationic brucite-like layered structure of LDH platelet...... 89 Figure 3.2: Layer-by-Layer assembly on a flat slide. Schematic representation of the LbL procedure (A) and the dipping robot used to deposit (LDH/PSS)n multilayers onto a flat substrate (B)...... 93 Figure 3.3: Hydroxylation of glass fibre surfaces. Schematic illustration of the removal of organic sizing and hydroxylation of the surface of glass fibres by piranha treatment. .... 93

Figure 3.4: Layer-by-Layer assembly of PDDA/(PSS/LDH)n multilayer coating on a bundle of fibres. Schematic representation of the LbL procedure (A) and photograph of a bundle of glass fibre dipped in a PDDA solution (B) – the black arrow pinpoints a fibre bundle dipped in a PSS solution...... 95 Figure 3.5: Oxygen low pressure plasma treatment of carbon fibre surfaces. Photograph of the setup of oxidation of carbon fibres via oxygen low pressure plasma treatment. ... 96 Figure 3.6: Oxidation and deprotonation of carbon fibre surfaces. Schematic illustration of surface oxidation of carbon fibres via oxygen low pressure plasma followed by further oxidation in KMnO4 and associated negatively charged surface at pH 10...... 97 Figure 3.7: Schematic of a bundle composite mounted on a paper template for tensile testing. The bundle is placed onto double-sided sticky tape (1) and then glued to the template with an epoxy resin (2). One end of the template end-tab is longer than the other in order to place an acoustic sensor in direct contact with the template (3) ...... 98 Figure 3.8: ζ-potential. Representation of the electrical double-layer at the particle surface and the concept of ζ-potential...... 100 Figure 3.9: Streaming ζ-potential. Schematic of the formation of a streaming ζ-potential ∆μs at the fibre/electrolyte interface via the creation of a streaming current (A) and leaking current (B)...... 101 Figure 3.10: Streaming ζ-potential measurement. Adjustable cylindrical cell of the SurPASS apparatus filled with hydroxylated glass fibres...... 102 Figure 3.11: Schematic of XRD rocking curve acquisition in (003) diffraction conditions of LDH platelets. LDH platelets deposited onto a flat glass substrate in relation to the Ψ, Φ and θ angles...... 104 Figure 3.12: Coating nanoindentation. Schematic of thin coating nanoindendation procedure...... 111 Figure 3.13: Nanoindentation parameters. Schematic illustration of an indent and the associated load-displacement nanoindentation curve. Copied from Oliver and Pharr.271,272 ...... 112 Figure 3.14: Nanoindenation load-displacement curve. Representation of the deformation areas of a nanoindentation load-displacement curve...... 113 Figure 3.15: SEM in-situ nanoindentation of nanostructured coating. Nanoindentation of a nanostructured coating deposited on a flat glass slide (A) and deposited on a glass fibre (B)...... 114 17

Figure 3.16: Representation of single fibre pull-out preparation and testing. Single fibre embedding unit (A) and in-situ pull-out setup (B). Typical load-displacement (F – S) curves of a brittle and ductile failure (A and B, respectively) - interface loading, interface failure (complete debonding) and fibre extraction steps of the pull-out process (1, 2 and 3 respectively). Fmax, ld and le are the maximum load applied to the interface, the fibre debonding length and embedded length, respectively...... 115 Figure 3.17: Mechanism of single fibre pull-out with nanostructured interphase. Schematic representation of the expected sliding behaviour of a nanostructured coated fibre, partly embedded in epoxy during a pull-out test...... 117 Figure 3.18: Single fibre fragmentation test. Schematic illustration of single fibre fragmentation in a matrix with corresponding level of stress. Copied from Tripathi et al.276 ...... 118 Figure 3.19: Stress concentration at fibre break. Optical micrographs of a fibre fragment and subsequent fibre slippage in matrix in a standard (top) and cross-polarised light (bottom). Copied from Feih et al.22 ...... 119 Figure 3.20: Strain measurement of bundle composite tested in tension via video gauge. Images of paint-marked bundle composite specimen mounted in tensile grip (A) and video gauge equipped with a macro lens (B)...... 121 Figure 3.21: Tensile grip. Tensile grip mounted with an AE sensor to test small bundle composites (refer to Appendix.3 for dimensions)...... 122

Figure 4.1: Mg2-Al-CO3 LDH platelet dimensions. TEM images of LDH-1 (A and D), LDH-2 (B and E) and LDH-3 (C and G) platelets at low and high magnification, respectively, and width distribution of the LDH platelets from the various synthesis (H)...... 126

Figure 4.2: Anisotropic crystalline structure of Mg2-Al-CO3 LDH platelets. XRD diffractogramms of different synthesised LDH platelets (A), high resolution TEM image of a LDH platelet cross-section presenting the (003) planes parallel to the platelet surface (B) and (003) XRD diffraction peak of LDH platelets (C). Marked peaks, distinctive of a LDH structure, correspond to Joint Committee on Powder Diffraction Standard (JCPDS) file 22-700.279...... 128

Figure 4.3: LDH brucite-like cationic surface. ζ -potential of Mg2-Al-CO3 LDH platelets in 5 mM KCl solution from pH 3 to 10 (A) and images of LDH suspension (B) immediately after synthesis (i) and after ageing for two months (ii)...... 129 Figure 4.4: Deposition of LDH and PSS monolayer by LbL deposition. SEM images of a LDH-1 layer containing a significant excess of platelets (A-B), a LDH-1 monolayer with few overlaps after cleaning (C-D), a quasi-perfect LDH-1 monolayer after further cleaning (E-F) and a (LDH-1/PSS) bilayer (G-H) deposited on a glass slide using LbL deposition...... 132 Figure 4.5:Topography of LDH-1 monolayer deposited on glass slide. AFM height profile (A) and 2 μm wide image (B) of LDH-1 monolayer...... 133 Figure 4.6: Absorbance features of PSS and LDH platelets. UV-Vis spectra of PSS solution (A) and LDH suspension (B)...... 134

absorbance at 225 nm of coated quartz slides as a function of the number of bilayer for both one step and two step rinsing LbL processes (B). Top-view SEM micrographs of quartz slides coated with (LDH/PSS)5 (C) and (LDH/PSS)15 (E) using one and two step rinsing, respectively and an image of LDH-1 platelet suspension at the end of each LbL deposition process (D)...... 135

Figure 4.8: (LDH-1/PSS)n/LDH-1 Layer-by-Layer deposited multilayer coatings on quartz slides. SEM cross-section micrographs of quartz slide coated with (LDH- 1/PSS)n/LDH-1 with n=10 (A), 50 (B), 90 (C) and 155 (D)...... 136

Figure 4.9: UV-Vis control of (LDH/PSS)n coatings produced by Layer-by-Layer assembly using different LDH inorganic platelets with different dimensions and size distributions. Plotted UV-Vis absorbance at 225 nm (A) and 350 nm (B) related to PSS and LDH content, respectively, from deconvoluted UV-Vis spectra...... 137

Figure 4.10: (LDH/PSS)n coating thickness uniformity deposited on a quartz slide. Photographs of about 0.5, 0.7 and 1 μm-thick (LDH-2/PSS)n coatings deposited on quartz slides with high uniformity. SEM cross-section micrograph of a quartz slide coated with about 1.5 μm-thick (LDH-2/PSS)75 coating...... 138

Figure 4.11: Qualitative comparison of organic content in 1 μm-thick (LDH/PSS)n coatings containing different LDH platelet dimensions and size distributions. UV- Vis spectra (A) and deconvoluted PSS absorption band (at 225 nm) (B) of 1 μm-thick coatings deposited on quartz slide with different LDH platelet reinforcements (LDH-1, LDH-2 and LDH-3)...... 140

Figure 4.12: Organic content of (LDH/PSS)n coatings produced using different LDH platelets. Thermal gravimetric analysis of (LDH/PSS)n coatings and their individual constituents in air...... 141 Figure 4.13: Comparison of organic content as determined by thermal gravimetric analysis and UV-Vis spectroscopy in the (LDH/PSS)n coatings containing LDH-1, LDH-2 and LDH-3 platelets. Organic content of (LDH/PSS)n coatings obtained from TGA (Figure 4.12) and UV-Vis absorbance (Figure 4.11.B) at 225 nm of ~ 1 μm-thick (LDH/PSS)n coatings normalised to (LDH-1/PSS)n coating values...... 142

Figure 4.14: (LDH/PSS)n/LDH coating morphology. SEM top surface and cross-section images of (LDH-1/PSS)155/LDH-1 (A and D, respectively), (LDH-2/PSS)50/LDH-2 (B and E, respectively) and (LDH-3/PSS)50/LDH-3 (C and F, respectively)...... 143

Figure 4.15: LDH platelet packing in (LDH-2/PSS)n/LDH-2 “brick-and-mortar” nanostructure. TEM images of the cross-section of (LDH-2/PSS)50/LDH-2 (A and C). Photograph of the investigated (LDH-2/PSS)50/LDH-2 coating deposited on glass slide (B). EDX composition map acquired during TEM cross-sectional imaging of (LDH- 2/PSS)50/LDH-2 nanostructure (D to G)...... 144

Figure 4.16: Platelet alignment in (LDH/PSS)n/LDH coatings as a function of platelet dimensions and size distribution. Three-dimensional rocking curves of (LDH- 1/PSS)200/LDH-1, (LDH-2/PSS)75/LDH-2 and (LDH-3/PSS)75/LDH-3 coatings acquired at 2θ=11.7° ( diffraction condition of (003) crystalline plane) for Ψ = [0; 80°] (graph step: 30°) and Φ = [0; 360°] (graph step: 90°)...... 145

Figure 4.17: Shallow nanoindentation of (LDH/PSS)n/LDH nanostructutred coatings. Load-displacement curves obtained from shallow nanoindentation of about 1.5 μm-thick (LDH/PSS)n/LDH coatings containing LDH-1, LDH-2 and LDH-3 platelet reinforcements (A, B and C, respectively)...... 148

Figure 4.18: Plastic deformation of (LDH-2/PSS)n/LDH-2 nanostructured coatings. Top-view SEM images of indents (white arrows evidence cracks initiated from the edge of the indent) made in a ~ 3.5 μm thick (LDH-2/PSS)n/LDH-2 coating at varying loads (A-G) and the associated load-displacement nanoindentation curves (H)...... 150

Figure 4.19: Plastic index of a (LDH-2/PSS)n/LDH-2 nanostructured coating. SEM top surface and cross-section images of ~ 3.5 μm thick (LDH-2/PSS)n/LDH-2 deposited on a glass slide (A and B, respectively). Calculated plastic index of (LDH-2/PSS)n/LDH-2 from nanoindentation as a function of the load applied to the coating and the maximum indentation depth (C)...... 152

Figure 4.20: Pile-up formation in (LDH/PSS)n/LDH nanostructured coatings during nanoindentation. SEM side-view images of in-situ indents made on (LDH- 1/PSS)200/LDH-1, (LDH-2/PSS)75/LDH-2 and (LDH-3/PSS)75/LDH-3 ~1.5 μm thick coatings at a depth of about 10 μm (A, B and C, respectively) (black arrows indicate multiple and subsequent displacement of materials while white arrows pinpoint coating failures within the pile-ups) and the associated SEM top-view images of (LDH- 2/PSS)n/LDH-2 and (LDH-3/PSS)n/LDH-3...... 153

Figure 4.21: Crack propagation in (LDH/PSS)n/LDH nanostructured coatings made of LDH-2 and LDH-3 platelet reinforcements. Top view SEM images of cracks initiated from indents made with a Berkovich tip into ~ 1.5 μm thick (LDH-2/PSS)75/LDH-2 and (LDH-3/PSS)75/LDH-3 coatings at low and high magnification (A-C and B-D, respectively)...... 156

Figure 4.22: Attempt to add another level of hierarchy to the (LDH-2/PSS)n/LDH-2 nanostructured coating. Schematic and top SEM view of as deposited (LDH- 2/PSS)n/LDH-2 (A), calcined (LDH-2/PSS)n/LDH-2 (B) and PSS-rehydrated (LDH- 2/PSS)n/LDH-2 (C) coatings. Nanoindentation load-displacement curves of as deposited and calcined/rehydrated (LDH-2/PSS)n/LDH-2 coatings (D)...... 158 Figure 5.1: ζ-potential = f (pH) of hydroxylated glass fibres. ζ-potential curves of as- received sized and piranha treated (desized) glass fibres measured in 5 mM KCl supporting electrolyte solution for pH 3-10...... 164 Figure 5.2: LDH monolayer deposited onto a bundle of hydroxylated glass fibres. SEM images of piranha treated glass fibres dipped in LDH suspension and subsequently dipped in an aqueous NaOH solution of pH 10, 12 and 6 times without stirring (A and B, respectively) and 1 time with stirring (C and D) – red arrow indicates an area with excess of LDH platelets and black arrow shows a bald spot on the fibre...... 166

Figure 5.3: Deposited thin LDH/(PSS/LDH)1 multilayer coatings on a bundle of glass fibres. SEM images of a bundle of coated glass fibres at low (A) and high magnification (B)...... 167

Figure 5.4: Morphology of PDDA/(PSS/LDH)n nanostructured coatings deposited around glass fibres. SEM images of PDDA/(PSS/LDH)n coatings deposited onto glass fibres with various thickness (n=25, 50 and 75). Low and high magnification top view (left and middle column, respectively) and side view (right column)...... 168 Figure 5.5: Nacre-nanomimetic coatings deposited around glass fibres. Top surface and cross-section SEM micrographs of glass fibres coated with PDDA/(PSS/LDH)n coating. n=12 (A), n=25 (B), n=50 (C) and n=75 (D). SEM micrographs of fibre top surface (i) and cross section (ii and iii)...... 169

Figure 5.6: Results of shallow in-situ SEM nanoindenation into PDDA/(PSS/LDH)75 nanostructured coatings deposited around glass fibres. SEM micrograph of indented coated glass fibres (A) and load-displacement nanoindenation curves obtained from PDDA/(PSS/LDH)75 deposited on both flat glass slide and glass fibre (B)...... 170

Figure 5.7: Deformation of the PDDA/(PSS/LDH)n nanostructured coating onto glass fibres. Load-displacement curves of indented coated glass fibre at different depth (A) and calculated plastic index (B) of coated deposited on glass fibre and glass slide...... 171

Figure 5.8: Single fibre pull-out curves F = (S) of bare and PDDA/(PSS/LDH)n coated glass fibres. Load-displacement curves of the pull-out tests (A) and the associated maximum force applied to the fibre plotted as a function of the fibre embedded area in the matrix (B). Interfacial shear strength and debonding length ratio were measured as a function of the thickness (number of (LDH/PSS) bilayers, n) of the coating (C and D, respectively)...... 173 Figure 5.9: Relation between apparent interfacial shear strength and fibre embedded length. Apparent IFSS = f (le) for dezised bare glass fibres and glass fibres coated with PDDA/(PSS/LDH)n...... 175 Figure 5.10: Effect of nanostructured interphase on stress distribution near fibre fragments in single glass fibre composite models. Optical micrographs of the fragmentation test of bare and PDDA/(PSS/LDH)25 coated glass fibres in epoxy using cross-polarised (A and B, respectively) and non-polarised light in transmission (C and D, respectively) - vertical arrows pinpoint fibre breaks and horizontal arrows show fragment lengths...... 177 Figure 5.11: Distribution of single fibre fragment lengths after fragmentation. Histogram (A) and cumulative (B) distribution for bare and PDDA/(PSS/LDH)25 coated glass fibres...... 178 Figure 6.1: Surface morphology of oxygen plasma treated carbon fibres. Top surface SEM micrographs of as-received unsized fibres (A and B) and unsized fibres treated for 30 s (C and D), 5 min (E and F) and 20 min (G and H)...... 183 Figure 6.2: ζ-potential = f (pH) of oxygen plasma treated carbon fibres. ζ-potential curves of as-received and treated unsized carbon fibres from pH 3 to pH 10, in 5 mM KCl. ... 184

Figure 6.3: Effect of liquid phase KMnO4 oxidation of O2 low-pressure plasma treated carbon fibres on their ζ-potential. ζ -potential = f (pH) of oxygen plasma treated carbon fibres before and after further oxidation in KMnO4 solution, from pH 3 to pH 10...... 185 Figure 6.4: Surface composition of as-received and oxidised unsized carbon fibres. XP survey spectra of as-received, 5 min O2 low pressure plasma treated and plasma treated carbon fibres further modified by oxidation in KMnO4...... 186 Figure 6.5: Oxygen O1s and carbon C1s high resolution spectra of as-received and oxidised unsized carbon fibres. Oxygen O1s and carbon C1s of as-received (A and D, respectively), 5 min O2 low pressure plasma treated (B and E, respectively) and plasma treated followed by further oxidation in KMnO4 (C and F, respectively)...... 187 Figure 6.6: Nanostructured coating deposited on oxidised carbon fibres. Top surface and cross-section SEM micrographs of oxidised carbon fibres coated with PDDA/(PSS/LDH)n, with n=12 (A and B, respectively), n=25 (C and D, respectively),

n=50 (E and F, respectively) and n=75 (G and H, respectively) - white arrows pinpoint a visible cross-section of the nanostructured coating...... 189

Figure 6.7: Single fibre pull-out curves F = (S) of bare and PDDA/(PSS/LDH)n coated oxidised carbon fibres. Load displacement curves obtained by single fibre pull-out tests (A) and the associated maximum force applied to the fibre as a function of the fibre embedded area in the matrix (B) – linear fitting were forced to the origin. Interfacial shear strength and debonding length ratio were measured as a function of the thickness of the coating (C and D, respectively)...... 191 Figure 6.8: Relation between apparent interfacial shear strength and fibre embedded length. Apparent IFSS = f (le) for bare carbon fibres and coated oxygen plasma treated carbon fibres coated with PDDA/(PSS/LDH)n...... 192 Figure 6.9: Effect of nanostructured interphase on stress concentration near a fibre fragment in single carbon fibre composite models. Optical images of in-situ fragmentation tests of bare (A) and PDDA/(PSS/LDH)25 coated (B) oxidised carbon fibres using cross-polarised light in transmission at a given strain (A and B, respectively) and at +1.2% (C and D, respectively) and +2.4% strain (E and F, respectively) – vertical arrows pinpoint fibre fragments and horizontal arrows and dashed line highlight progressive sliding of the fibre. Optical images of bare and PDDA/(PSS/LDH)25 coated oxidised carbon fibres using non-polarised light in transmission after fragmentation test (G and H, respectively) - horizontal arrows show fragment lengths ...... 195 Figure 6.10: Distribution of fibre fragment lengths after fragmentation of oxidised carbon single fibre composite with and without nanostructured interphase. Histogram (left) and cumulative (right) distribution for bare and PDDA/(PSS/LDH)25 coated oxidised carbon fibres...... 196 Figure 7.1: Droplets of an ultra-low viscosity epoxy resin resting on nanostructured coatings terminated by an LDH- and PSS- monolayer. Contact angle of an epoxy droplet formed onto the surface of a nanostructured coating terminated with a PSS PE monolayer (A) and an LDH platelet monolayer (B)...... 201 Figure 7.2: Hierarchical glass fibre bundle composite. SEM image of a PDDA/(PSS/LDH)25 coated glass fibre bundle composite (A) and SEM images of an epoxy impregnated bundle at low and high magnification (B and C, respectively)...... 202 Figure 7.3: Hierarchical carbon fibre bundle composite. SEM image of a PDDA/(PSS/LDH)25 coated fibre bundle (A) and SEM images of an epoxy impregnated bundle at low and high magnification (B and C, respectively)...... 203 Figure 7.4: Results of glass fibre bundle composite tensile tests. Tensile strain-stress curves synchronised with the cumulative distribution of acoustic emission events occurring in the bundle composites containing sized, hydroxylated and PDDA/(PSS/LDH)25 coated glass fibres...... 205 Figure 7.5: Tensile fracture of glass fibre bundle composites. High speed camera images of glass fibre bundle composites rupture in tension containing sized (A), hydroxylated (B) and coated glass fibres (C). SEM images at low and high magnification of sized, hydroxylated and coated fibre bundle composite fractured surfaces (i, ii and iii, respectively)...... 207 Figure 7.6: Results of carbon fibre bundle composite tensile tests. Tensile strain-stress curves synchronised with the cumulative distribution of acoustic emission events

occurring in the bundle composites containing sized, oxidised and PDDA/(PSS/LDH)25 coated carbon fibres...... 209 Figure 7.7: Tensile fracture of carbon fibre bundle composites. High speed camera images of carbon fibre bundle composite rupture in tension containing sized (A), oxidised (B) and oxidised carbon fibres coated with PDDA/(PSS/LDH)25 coating (C). SEM images at low and high magnification of sized and oxidised fibre bundle composite fractured surfaces (i, and ii, respectively) as well as coated composite fractured surfaces (iii and iv, respectively)...... 210 Figure A1.1: Analytical approach to the relation between fibre diameter and maximum platelet width. Wp,max=f(df) for varying aspect ratio obtained from equation (A1-2) (A) with a phase proportion constraint of 90:10 and Wp,max = f(df) obtained from equation (1- 5) at a fixed aspect ratio of 10 (B) with the minimum platelet width defined by the smallest polymer molecular layer thickness achievable of about 1.5 nm...... 243 Figure A2.1: Electron deposited Platinum coating. SEM image of Platinum coating deposited on PDDA/(PSS/LDH)50 specimen...... 247 Figure A2.2: FIB milling. SEM micrographs of the deposited platinum coating onto the specimen surface (A), milling of regular trenched on both sides of the platinum coating (B) and cutting of the lamella (C) with Ga+ beam...... 247 Figure A2.3: Transfer of specimen to TEM grid. SEM image of specimen attachment to a TEM grid...... 248 Figure A3.1: Tensile grip. Schematic of the tensile grip for small bundle composite tensile test...... 250 Figure A3.2: Acoustic sensor fitting. Schematic of the AE sensor holder of tensile grip. . 251 Figure A4.1: Quaternary-amine terminated glass fibre surfaces. Synthesis steps of quaternary-amine terminated glass fibre surface modification...... 253 Figure A4.2: Surface charge density of quaternary-amine terminated glass fibres. ζ- potential curves of as-received sized and quaternary-amine terminated glass fibres, in 5 mM KCl for pH 3 to 10...... 254 Figure A4.3: Quaternary-amine terminated carbon fibre surfaces. Synthesis steps of quaternary-terminated amine carbon fibre surface modification...... 255 Figure A4.4: Effect of quaternary-terminated amine modified carbon fibres on their surface charge density. ζ-potential curves of as-received and modified carbon fibres from pH 3 to pH 11...... 256 Figure A5.1: Nanostructured coating with and without PDDA precursor layer. SEM cross-section images of glass fibre coated with (LDH/PSS)n coating at low and high magnification (A and B, respectively) and coated with PDDA/(PSS/LDH)n coating (C) – black arrows indicate coating delamination...... 257 Figure A5.2: Effect of a PDDA precursor layer on the debonding length ratio of coated glass fibre in epoxy. Load-displacement curves of single fibre pull-out (A) and associated debonding length ratio (B) for bare desized glass fibres as well as desized glass fibres coated with LDH/(PSS/LDH)25 and PDDA/(PSS/LDH)25...... 258 Figure A6.1: Effect of LDH platelet dimensions on the debonding length ratio of PDDA/(PSS/LDH)12 coated glass fibres. Load-displacement curves of single fibre pull- out tests (A) and associated debonding length ratio (B) of bare desized glass fibres as

well as desized glass fibres coated with PDDA/(PSS/LDH)12 containing LDH-1 (“small”-LDH) platelets and LDH-2 (“optimised”-LDH) platelets...... 260 Figure A7.1: Surface charge density of as-received carbon fibres. ζ-potential curves of as- received sized and unsized carbon fibres from pH 3 to pH 11, in 5 mM KCl...... 261 Figure A7.2: LDH monolayer LbL deposition on sized and unsized as-received carbon fibres. Top surface SEM micrographs of as-received and LDH coated sized carbon fibres (A and B, respectively). Top surface SEM micrographs of as-received and LDH coated unsized carbon fibres (C and D, respectively)...... 262 Figure A7.3: Surface morphology of nanostructured coatings deposited on a bundle of unsized carbon fibres. Top surface SEM micrographs of carbon fibres coated with PDDA/(PSS/LDH)n coating. n=12 (A), n=25 (B), n=50 (C) and n=75 (D)...... 264 Figure A7.4: Cross-section of nanostructured coatings deposited on a bundle of unsized carbon fibres. Cross-section SEM micrographs of carbon fibres coated with PDDA/(PSS/LDH)n coating. n=12 (A), n=25 (B) and n=50 (C and D)...... 265

Figure A7.5: Single fibre pull-out tests of bare PDDA/(PSS/LDH)n coated unsized carbon fibres. Load displacement curves of single fibre pull-out tests (A) and the associated maximum force applied to the fibre as a function of the fibre embedded area in the epoxy matrix (B). Interfacial shear strength and debonding length ratio were measured as a function of the thickness of the coating (C and D, respectively)...... 267

List of Tables

Table 4.1: Summary of Mg2-Al-CO3 LDH platelets properties...... 129

Table 4.2: Summary of the different (LDH/PSS)n nanostructure morphological properties. 146

Table 4.3: Summary of the different (LDH/PSS)n nanostructure mechanical properties...... 155 Table 5.1: Interphase properties obtained from pull-out tests...... 175 Table 5.2: Summary of interphase properties in single glass fibre composite from fragmentation tests...... 179 Table 6.1: Surface composition of as-received and oxidised unsized carbon fibres...... 188 Table 6.2: Interphase properties obtained from pull-out tests...... 193 Table 6.3: Summary of interphase properties in single oxidised carbon fibre composite from fragmentation tests...... 197 Table 7.1: Summary of tensile properties of glass fibre bundle composites ...... 208 Table 7.2: Summary of tensile properties of carbon fibre bundle composites ...... 211

List of Abbreviations

AE Acoustic Emission AFM Atomic Force Microscopy CVD Chemical Vapour Deposition DLR Debonding Length Ratio EDX Energy Dispersive X-ray FIB Focused Ion Beam FRP Fibre-Reinforced Polymer IFSS Interfacial Shear Strength ILSS Interlaminar Shear Strength LbL Layer-by-Layer LDH Layered Double Hydroxide PDDA Poly(diallyldimethylammonium chloride) PE Polyelectrolyte PMMA Poly(methyl methacrylate) PSS Poly(sodium 4-styrenesulfonate) PVA Poly(vinyl acetate) SEM Scanning Electron Microscopy SFF Single Fibre Fragmentation SFPO Single Fibre Pull-out TEM Transmission Electron Microscopy TGA Thermogravimetric Analysis XRD X-ray Diffraction XPS X-ray Photoelectron Spectroscopy

Introduction

1.1 Motivation

Failure of fibre-reinforced polymer (FRP) composites in tension is caused by progressive fibre fragmentation, which eventually leads to a localised critical number of accumulated breaks such that the stress concentration cannot be carried by the neighbouring fibres. As a consequence, the material breaks catastrophically without prior warning

(Figure 1.1). As one of the potential approaches to overcome inherent brittleness, composite architecture may be modified at the fibre/matrix interface, to provide a route to dissipate a substantial fraction of the energy arising from fibre breaks and promote a more gradual failure of the whole material through plastic deformation. To produce composites which exhibit a ductile failure, mechanisms are needed to allow for stress concentration dissipation/redistribution and plasticity. Crack deflection at the fibre/matrix interface and fibre sliding during debonding may avoid excessive stress concentrations in neighbouring

fibres and absorb the energy of the fibre break. However, fibre sliding must be gradually arrested to avoid damage localisation and to spread the increased load over a greater length of neighbouring fibres.

The hypothesis is that a nanostructured interphase can provide crack deflection and strain hardening in shear. The idea draws on nature’s ability to form planar and anisotropic nanocomposites in nacre, which displays these characteristics in shear.1 The

“brick-and-mortar” structure of nacre is made of stiff, brittle inorganic platelets, well-organised in a three-dimensional manner and glued together by a soft and viscoelastic organic phase. This material exhibits excellent mechanical properties, such as high stiffness, strength and toughness combined with strain hardening in shear and the ability to arrest a propagating crack. The volume fraction of inorganic platelets in nacre is about 95%, providing high stiffness and strength to the material, while the remaining 5% of the soft matrix phase allows the structure to deform plastically leading to high toughness

1.2 Objectives

The goal of the PhD was to apply the toughening mechanisms found in natural nacre, such as crack deflection at the platelet interface2 and strain hardening in shear caused by platelet sliding and subsequent interlocking,3 to curved fibre/matrix interfaces in FRP composites (Figure 1.2.A). Hence, the absorption of energy arising from a fibre break through crack deflection at the platelet interface along the length of the fibre can potentially isolate fragmented fibres in a composite and, therefore, delay the formation of critically sized clusters of broken fibres. Therefore, higher strength and strain to failure of composites should be achieved by avoiding the correlation of fibre breaks (Figure 1.2.B). The subsequent stable and progressively arrested sliding of the fibres, at each uncorrelated break, should provide for more plastic deformation in conventional FRP composite, leading to a non-linear stress-strain curve (Figure 1.2.B).

Typical nacre platelets, both in natural and artificial nacre, are in the order of 5 to

10 μm long.4 Thus, these flat structures cannot form a coherent layered coating around fibres with diameters of the same length scale. The working assumption, based on existing models, is that nacre-like properties can be replicated, independently of absolute scale, as long as the aspect ratio of the reinforcing platelets is maintained (at about 10) and the relative thickness of the soft organic layers remains about 5 to 10% of the thickness of the platelet.5,6 A polymer layer, defined by the dimensions of polymer molecules, is around 1-2 nm, thus platelets should ideally be approximately 10 to 20 nm thick and about 100 to 200 nm wide. At this scale, the curvature of the fibre is not significant, and hence the layered structure can be accommodated.

This study draws on existing literature mimicking the structure of planar nacre.7,8 Many attempts have been carried out to produce an artificial nacre structure with high strength, stiffness and toughness, at different length scales, using various materials and deposition methods.9 Among these deposition methods, only a few can be adapted to deposit artificial nacre around fibre surfaces, such as electrophoretic deposition10 and Layer-by-Layer (LbL) assembly.11 The LbL assembly method was selected because it is possible to deposit monolayers of both organic and inorganic materials with good thickness control, whereas deposition of organic layers by electrophoretic deposition remains challenging. In addition,

LbL assembly is self-limiting and can be used to coat any substrate geometry, which makes it appropriate to coat fibre bundles. LbL is based on electrostatic attraction of oppositely charged colloids, in this case inorganic platelets and organic polyelectrolyte (PE) polymer molecules; sequential dipping of a charged substrate into solutions containing these charged particles may lead to the deposition of a well-organised nanostructured coating, similar to the

“brick-and-mortar” structure of nacre but at the nanometre length scale. Each coating should have good interlayer adhesion thanks to electrostatic and van der Waals interactions. Even though the LbL deposition of a micrometre thick coating can take several hours, one advantage is that the generally slow routes to planar nacre are accelerated and amplified by the potential parallel coating of multiple fibres and, rendering enable manufacturing of fibre reinforced composite materials with a nanostructured interphase.

Figure 1.2: Schematic of the expected toughening mechanisms occurring in the nanostructured interphase of hierarchical composites and the associated tensile response of a hierarchical composite. Energy absorption and dissipation along fibre length by crack deflection in the anisotropic nanostructured interphase (i) and progressively arrested uncorrelated fibre slippages through strain hardening of the interphase in shear (ii) (A). Tensile response of conventional and hierarchical fibre-reinforced polymer composites: formation of the critical cluster of fibre breaks triggering the sudden failure of conventional composite (i), stress dissipation in the nanostructured interphase leading to uncorrelated fibre fragments (ii) and subsequent stable fibre slippage (iii) (B).

1.3 Approach

The concept of a nanostructured composite interphase requires a specific structure of the coating, involving a dense, layered packing of hard platelets in a soft matrix, such that the platelets pull-out (causing crack deflection), whilst being sufficiently close to interact with each other as they slide. The coating must also conform to the surface of the fibre to an approximate accuracy at least better than half the thickness of the polymer layer (∆dpol).

Hard inorganic platelets with hexagonal shape are suitable for the deposition of monolayers with a high degree of packing.12 Their anisotropy needs to be carefully tuned to ensure platelet pull-out rather than fracture; the aspect ratio (s) should be as large as possible,

13,14 whilst still lower than the critical value sc as defined below:

σp sc = τy (1-1)

where σp and τy are the platelet tensile stress and interface yield shear strength, respectively.

However, the critical aspect ratio sc of many platelets cannot be calculated as no bulk material properties are available. Therefore, an aspect ratio around 10 is likely to be a suitable starting point; aspect ratios around this value are commonly relevant for systems involving pull-out in a range of contexts including the design of composites15 and biological structures.16

Depending on the fibre diameter, a minimum number of platelets tangentially deposited around the circumference of the fibre can be estimated and subsequently related to a maximum platelet width. An analytical approach, based on the deviation of tangentially deposited platelets around a fibre, as a function of the number of platelets in the circumference of the fibre, can be used to estimate the maximum platelet width (Figure 1.3).

Figure 1.3: An analytical approach for the adaptation of a “brick-and-mortar” nanostructured coating around a fibre. Arrangement of platelets around the circumference of a fibre as a function of the number of platelets, deposited over a polymer precursor layer. Wp, dpol, ∆dpol, dr and rf are the platelet width, polymer layer thickness, polymer layer fluctuation, platelet tangential deviation and fibre radius, respectively.

The tangential deviation of the platelets from the fibre surface can be expressed as follows:

dr = l − rf (1-2)

where l and rf are the deviated length of the platelet to the centre of the fibre and fibre radius, respectively (Figure 1.3). The deviated length of the platelet to the centre of the fibre (l) can be expressed as a function of the number of platelets n as described below:

rf l = π cos( ) (1-3) n

For a large number of platelets, the fibre perimeter (Pf = 2·π·rf) is approximately the sum of the width of the platelets (wp) on its surface, leading to the following expression.

2π ∙ rf w = (1-4) p n

The maximum platelet width leading to a conformal coating on a polymer precursor layer can be associated with a platelet tangential deviation of half the thickness of the polymer layer (∆dpol = dr), similar to the fluctuation of the polymer layer thickness:

rf wp,max = 2rf ∙ arccos( ) (1-5) rf + ∆dpol

where wp,max is the maximum platelet width. Assuming a polymer layer around 1.5 nm thick, the maximum platelet width satisfying good conformation of the coating to the fibre can be determined for any reinforcing fibre diameter (Figure 1.4).

Figure 1.4: Suitable platelet dimensions to enable conformal LbL deposition of the desired “brick-and-mortar” nanostructure on reinforcing fibres. Platelet dimension range defined by the maximum platelet width (wp,max) as a function of the fibre diameter (df) and the minimum platelet width (wp,min) for a platelet aspect ratio, s = 10, and an inorganic phase of at least 90 vol.%

The desired volume ratio of inorganic to organic phase in each bilayer of the structure should be around 90:10 (90 vol.% of platelets), implying a polymer layer thickness 9 times thinner than the platelet thickness (the smaller amount of polymer present between two adjacent platelets is neglected compared to that in between two layers). The latter constraint defines the minimum platelet width, as a function of the polymer layer thickness dpol and platelet aspect ratio s, as follows:

wp,min = 9sdpol (1-6)

Typically, the diameter of major reinforcing fibres is in the range of 5 to 15 μm (5 to

10 μm and 10 to 15 μm for carbon and glass fibres, respectively). Therefore, at a given aspect ratio of 10, there is a narrow range of acceptable platelet dimensions that are most suitable to coat reinforcing fibres. For a typical polymer layer of a 1.5 nm (LbL self-assembly), maximum platelet dimensions of about 135 and 13.5 nm and 215 and 21.5 nm, in width and thickness, should be used to enable coating fibres with diameters of 6 and 15 μm, respectively; the minimum dimensions of 135 and 13.5 nm in width and thickness

(Figure 1.4), limiting the use of reinforcing fibres to a diameter of at least 6 μm. The influence of thick polymer layers, which might be accessible by other assembly methods, on these constraints is discussed in the Appendix.1. It is important to note that the window of suitable platelet sizes for the successful deposition of nanostructured coatings onto major reinforcing fibres, is very narrow and, therefore, the structure must be designed carefully.

1.4 Structure of the thesis

The thesis manuscript consists of eight chapters as outlined below:

 Chapter 1 outlines the motivation, main objectives and approach of the work

undertaken during the PhD research to develop a new generation of hierarchical

composites.

 Chapter 2 gives an overview of the context, challenges and work achieved by other

researchers, and provides the state-of-the-art in current, relevant scientific literature.

 Chapter 3 describes the methods used to reach the objectives of developing a ductile

hierarchical composite, along with the materials and characterisation techniques used.

 Chapter 4 presents the experimental work carried out to produce that nanostructured

coating, inspired by the “brick-and-mortar” structure of natural nacre, but scaled

down by more than one order of magnitude.

 Chapter 5 presents the experimental work transferring the nanostructured coating onto

the surfaces of glass fibres and the results on single fibre interface properties.

 Chapter 6, similarly to chapter 5, describes the experimental work carried out to

transfer the nanostructured coating onto carbon fibres, including the use of

as-received and surface treated carbon fibres and the effect on single fibre interface

properties.

 Chapter 7 presents the manufacturing and tensile testing of small bundle composites

with and without predetermined optimum interphase design, to investigate the effect

of the nanostructured interphase on the tensile properties of the composites

 Chapter 8 summarises the work produced during the PhD and draws conclusions

about the potential impact in the field of FRP composite materials. It also provides

guidance regarding additional work to be conducted in order to further investigate the

potential of this new generation of toughened hierarchical composites with

nanostructured interphases.

Literature Review

2.1 Fibre-reinforced polymer composites for load bearing applications

2.1.1 Fundamentals of fibre-reinforced composites

A composite material is the combination of at least two different constituents exhibiting properties by far exceeding those of the constituents with a clear separation called interface or interphase.17,18 Fibres in FRP composites are called reinforcements as they present superior mechanical properties over the resin in which they are individually embedded. When unidirectional fibres, such as carbon or glass fibres, are impregnated with a matrix, composite materials with high strength and stiffness are obtained. While fibres can carry the load applied to the composite thanks to their high strength and stiffness, the matrix is used to protect them, hold them in place and bind them together and, therefore, transferring

the stress throughout the whole material. Hence, the combination of the two constituents results in a material with high mechanical properties and low density, with better performance than the constituents alone.

High performance FRP composite materials, such as carbon fibre-reinforced epoxy resin, have been widely investigated and used over the past few decades, especially in the field of aeronautics, aerospace, oil industry, in marine structure and sporting goods. The materials exhibit high strength, stiffness and low density while being chemically inert, making them suitable for load bearing purposes and, therefore, structure applications. The carbon fibres are commonly made through continuous thermal decomposition of organic precursor fibres, such as PAN or RAYON, into fibrous materials with high carbon content.19

On the other hand, glass fibres are also widely used as fibre reinforcements for polymers, but have a lower strength and stiffness than carbon fibres but higher strain to failure, at a lower price.

FRP composite materials can be made using two different types of polymer matrices, namely, thermoplastic and thermoset resins. The shape of the thermoplastic resins can be modified via temperature change through melting and solidification, whereas that of thermosetting composites is permanently defined due to permanent crosslinking of the polymer chains. The latter are commonly stiffer and stronger that the former, as a result of the crosslinked molecular network, and also exhibit higher resistance to heat and chemicals but lower toughness

2.1.2 Role of fibre-reinforced composite interfaces

In unidirectional FRP composite materials, the load is mainly carried by the stiff fibres, while the matrix, in shear, assures for a good load transfer to the fibres in the perpendicular direction to the applied load. The interface between fibre and matrix was shown to control the properties of the composite.20 The higher the interfacial shear strength

(IFSS), the higher the tensile strength of composites. It is the fibre/matrix interface that ensures the load transfer after a fibre break, through shearing of the matrix if the interface is sufficiently strong. As a fibre break occurs in the matrix, the matrix bridges the break through shearing and, therefore, transfers the load to the neighbouring fibres along a length that is defined by the IFSS of the fibre/matrix interface. The transfer of the load generated by a fibre break, to a neighbouring fibre, is described by the shear lag model.21 The fibre/matrix interface can be studied using single fibre tests, such as fragmentation22 and pull out23 tests

(refer to Chapter.3), providing information regarding stress distribution, interface deformation, IFSS and pull-out friction.

Direct contact between manufactured carbon fibres and polymer matrices leads to poor interfacial adhesion due to hydrophobicity and chemical inertness of the carbon.

Therefore, commercially, fibres are treated to generate oxy-carbonated functional groups, which in turn form covalent bonds with the polymer resin.24-26 A coating called “sizing”, consisting of a solution or emulsion of polymeric components, is added onto the surface of the treated fibres. The sizing is used to protect the fibres from wear and abrasive damage during transport, handling and composite manufacturing,27,28 and can also improve fibre wettability with the matrix.29 Whilst the sizing is usually cured onto the fibre, it can also dissolve and, therefore, diffuse in the matrix, which results in better fibre/matrix interfacial adhesion and stress transfer, depending on their compatibility.30 A strong bond between fibres 39

and matrix results in high stiffness and strength, whereas a weak interface will promote energy absorption, suitable for impact conditions. Various methods to modify the surface of fibres and incorporate nanofillers31 at the fibre/matrix interface have been investigated over the past few decades, as a route to further improve the mechanical properties of the composites.

2.1.3 Catastrophic failure of fibre-reinforced composites

During progressive loading of unidirectional FRP composites in tension, fibre fragmentation occurs throughout the whole material at locations determined by the stochastic strength distribution of the fibres. As a fibre breaks, the load that was previously carried by the fibre is transferred to the neighbouring fibres through the matrix, which locally increases the stress concentration in these neighbouring fibres. As mentioned previously, and depending on the length along which the stresses are transferred, the material can redistribute the load as long as the stress concentration does not locally reach a critical point. If the tensile load increases further, more fibres break in the composite, eventually causing clusters of fibre breaks to form32-34, which is associated with a critical stress concentration that cannot be carried by the remaining fibres. The critical stress concentration leads to a sudden and catastrophic failure of the material. The formation of a critical cluster of breaks and the related catastrophic failure was observed on small composite specimens using synchrotron

X-ray microtomography33,35 (Figure 2.1). The catastrophic failure is controlled by the stress redistribution around the fibre breaks, which strongly depends on the matrix properties.36-39

Figure 2.1: Formation of a critical cluster of fibre breaks triggering the catastrophic failure of the composite. X-ray radiograph of a quartz-epoxy composite when a critical cluster of fibre breaks forms (A) and triggers sudden failure of a composite (B). Adapted from Aroush et al.33

During the failure process of the composite, fibres debond from the matrix and are subsequently pulled out. Mechanical pull-out frictions dissipate energy, increasing composite toughness. However, the sudden and brittle failure of FRP composite materials significantly limits the amount of dissipated energy within the material. Therefore, in addition to improving the strength of the fibre/matrix interface, a lot of effort has been invested in increasing the toughness of composites by designing and tuning the interphase region, between the fibre and the matrix.

2.2 Engineered fibre-reinforced composite interphases for energy absorption

An interphase is defined as a finite layer in between a fibre and a matrix with distinct properties, and was first introduced in the 1970s.40 A three-phase composite system containing a fibre reinforcement, a polymer matrix and an interphase can potentially be designed to achieve high strength and toughness materials. Engineering the interphase of composites was widely investigated over the past five decades.30,31,41 An explanation for the improvement in composite toughness, while maintaining high strength, relies on two

mechanisms, namely, toughening and adhesion enhancement.42 The toughening mechanisms are, interfacial debonding,43 post-debonding friction,44 fibre pull-out21,45 as well as stress redistribution after breakage46,47 and stress relief in the interphase.48,49 On the other hand, improved adhesion can be achieved by enhancing the interaction between fibre/matrix via the introduction of functional groups into the interphase. However, nowadays, improvement of composite toughness is mainly achieved by addition of toughening agents50,51 into the matrix rather than through interphase design, as composite toughening via interphase engineering,31 such as polymer coating41 and intermittent bonding,52-54 often leads to a reduction in composite strength.

Engineering the interphase is quite important since the onset of mechanical damage often occurs in this region. A strong interphase leads to high composite strength and stiffness with brittle failure and low energy absorption, but a weak interphase enables multiple delamination and high energy absorption at the cost of a reduction in the mechanical properties of the composite.31 Therefore, the interphase needs to be tailored to balance these motivations. An ideal optimised interphase would allow for crack deflection, fibre pull-out and excellent stress transfer with a high IFSS, significantly increasing the amount of energy dissipated in composites while assuring for high mechanical performances.

2.2.1 Energy absorbing fibre polymer coatings

Different types of polymer coatings of varying thickness have been investigated to tune the interphase of composites, aiming to improve both toughness and strength of the materials. Fibre coating methods,41 such as applying a polymer interlayer, is one of the most successful methods to achieve high strength and toughness composites by mean of interface

control.55 For instance, locally weakening the interface between the fibres and the matrix was shown to increase toughness of the composite via fibre delamination and subsequent pull-out, but significantly reduces the mechanical properties of the composites. The properties of fibre/matrix interface were intermittently modified by the deposition of discontinuous coatings;52-54,56 an advancing crack tip triggers debonding at the weak interface whereas strong interface blunt the running crack as described by Cook and Gordon.57 If enough strong interfaces are achieved in the composite to pick up the rule of mixture strength from the fibres,52,53 the rest of the interface volume can be used to increase energy absorption via long debonding and subsequent fibre pull-out. A theoretical study carried out by Broutman et al.58 shows the influence of the polymer interlayer modulus on the strength and energy absorption capability of the composite material, indicating the optimal coating modulus to be about one-tenth of that of the matrix. It was concluded that the interphase should remain as thin as possible to avoid any reduction of fibre volume fraction, which does affect the composite strength and stiffness. Also, the interphase should allow for weak bonding at the interface with high frictional bonding. The toughness of composite materials was shown to be enhanced by the use of an energy absorbing interphase,59-66 such as a fibre coating allowing for frictional sliding, plastic deformation, crack blunting and crack deflection. In this case, the amount of energy absorbed during frictional sliding of the fibres was found to be larger than that absorbed during fibre debonding. For instance, a rubbery polymer coating led to an increase in composite toughness by up to 100% with some loss in flexural strength.64

Similarly, colloidal rubber particles were deposited onto glass fibres to form homogenous layers, resulting in about 600 % improvement in impact toughness.63

More recently and among the diverse methods to deposit polymer coatings on fibres, plasma polymerisation was found to be one of the most effective methods to improve both strength and toughness of composites67,68. Plasma polymerisation was shown to lead to the

deposition of pinhole-free coatings with good homogeneity. Differently to sharp fibre/matrix interface produced by plasma treatment, few tens of nanometres thick plasma-polymerised coatings placed at the fibre/matrix interface can efficiently distribute the stress and, therefore, improve both the toughness and strength of composites. Up to 6.5 times increase in interlaminar shear strength (ILSS) of unidirectional glass fibre/polyester composites was reported with a 0.1 μm thick plasma polymerised coating deposited from a tetravinylsilane/O2 gas mixture.68

2.2.2 Roughened interphases for interfacial mechanical frictions

Increasing the surface fibre roughness, by surface treatment, were undertaken to promote interlocking between the fibres and matrix.69-71 For instance, surface etching with acid, was shown to cause deep ridges and introduce perforations, thereby, increasing mechanical interlocking and energy absorption.69 The interfacial adhesion and composite mechanical properties was found to be increased by fibre surface roughness, particularly at the nanometre length scale.72 Addition of nanoparticles into the interphase/sizing can enhance the surface roughness and increase the local modulus and shear strength.31 As a consequence of the increased surface roughness, a high level of friction occurs, which promote mechanical interlocking causing toughening.

Grafting of carbon nanotubes onto fibre surfaces, for instance via chemical vapour deposition (CVD), has also been intensively investigating over the past two decades to promote fibre interlocking.73 These nano-reinforcements, radially grafted to the fibres, lead to significant increase in IFSS, which depends on their alignment and length.74 Up to 150 % improvement in IFSS of grafted silica fibres embedded in a poly(methyl methacrylate)

(PMMA) matrix was reported.75 On the other hand, 50 % increase in mode-I fracture toughness was achieved for grafted carbon fibre/epoxy composites.76 However, the grafting of carbon nanotube onto glass and carbon fibre surfaces is also accompanied, in most cases, with a 15 to 55 % decrease in tensile strength of the fibres, due to fibre degradation caused by catalyst particles penetration.77

2.2.3 Incorporation of nanofillers in interphase to toughen composites

Crack propagating in interphases containing nanoreinforcements were found to deflect at the nanofiller interface as the particles act as obstacles.31,78,79 The more anisotropic the particle, the more deflected the crack. Therefore, anisotropic particles, when placed in the interphase, can significantly increase the toughness of the composite. Among those

30,80 81 nanoparticles, the most commonly used are carbon nanotubes, carbon nanofibres, SiO2 nanoparticles,82-84 clay nanosheets85,86 and graphene oxide (GO) nanosheets.87-95 However,

GO and clay nanosheets seem to be the most suitable nanoreinforcements to increase the toughness of FRP composite materials, thanks to their high anisotropic shape. These nanosheets can easily be added to the interphase by a simple fibre dipping method or direct grafting.91,96,97 The use of the latter methods eliminates any potential fibre damage occurring with other techniques, such as CVD (Figure 2.2). Over the past few years, interphases containing nanosheets have been used, which improved fibre/matrix interfacial and composite mechanical properties.88,93,94,97

A B

C D

Figure 2.2: Schematics showing methods to deposit graphene oxide nanosheets onto reinforcing fibres. Electrophoretic deposition of GO nanosheets onto carbon fibres (A), dip coating carbon fibres with GO nanosheets (B) and direct grafting of GO nanosheets onto carbon and glass fibres (C and D, respectively). Adapted from references.91,92,94,97

Among the different dipping methods, carbon or glass fibres can be immerged into an electrophoretic bath to coat them with GO nanosheets87,92,95 or graphite nanoplatelets89

(Figure 2.2.A). In such setups, a tow of fibres is used as the deposition electrode while the

GO nanosheets are well-dispersed in the bath. Unreduced GO nanosheets can easily be dispersed into aqueous media, at both acidic and basic pH, since they present a negatively charge surface (negative ζ-potential). Ultrasonicated ozone-modified GO have also been used to achieve stable dispersion for EPD.98 Various EPD parameters can be carefully controlled in order to adjust the kinetic and deposition yield of GO nanosheets on fibre surfaces, such as

EPD voltage, suspension pH, particle concentration and deposition time.99 By varying the voltage between the electrodes, different concentrations of GO nanosheets were coated onto fibre surfaces, starting from homogenous monolayers to large aggregates95 (Figure 2.3.G-H).

From a general point of view, increasing voltage was shown to lead to wrinkling and,

therefore, high coating roughness, porosity and non-uniform deposition,100 while low voltage produces high quality coatings. The EPD deposition of GO nanosheets on carbon fibre surfaces led to strong mechanical interlocking between the fibres and the matrix in a carbon fibre/epoxy composite, which in turn increased the toughness and strength of the interfacial region around the fibres with an improvement of up to 56 % reported in ILSS.87 Good adhesion of the GO interlayer was confirmed by atomic force microscopy (AFM), leading to failure between the GO and matrix rather than the GO/fibre interface.95

On the other hand, direct dispersion of GO nanosheets or graphite nanoplatelets into a solution containing sizing material was used to dip coat fibres88,94 (Figure 2.2.B). Varying the concentration of platelets in the sizing solution produced an interphase containing different proportions of platelets, from 1 to 10 wt.%. Yet again, homogenous dispersion of GO nanosheets was produced at low concentrations of GO nanosheets in sizing mixture as well as large nanosheets aggregates at higher concentrations (Figure 2.3.A-D). The incorporation of

5 wt.% of GO nanosheets in fibre sizing allowed for an improvement of 36 %, 13 % and 31

% in IFSS, ILSS and tensile strength of carbon fibre/epoxy composites, respectively, as compared to commercially sized carbon fibres.88 The improvement in the mechanical properties of the composites was explained by crack-tip bridging, allowing for successful redistribution of the stresses around the surface crack.

A dense monolayer of poly(ethylene imine)-functionalised silica nanoparticles was electrostatically deposited with good uniformity on glass fibres, via direct dipping, leading to an increase in IFSS, mechanical interlocking and fracture toughness.101 The improved interlocking and fracture toughness were found to be dependent on the dimensions of the silica nanoparticles. An improvement in IFSS of 35 % in glass fibre/epoxy composites, along with mechanical interlocking and higher fracture toughness were achieved with nanoparticles with a diameter greater of 26 nm. Smaller particles were found to be too small to increase

fibre roughness, while larger particles were associated with a poor ratio of particle-fibre contact to height, leading to premature debonding and poor stress transfer, reducing the IFSS of the composite.

Direct covalent-grafting of GO nanosheets onto glass or carbon fibres was also investigated, which improved the interfacial adhesion between fibre and matrix91,96,97 (Figure

2.2.C-D). Single GO nanosheets were grafted to fibre surfaces, avoiding overlaps and aggregates (Figure 2.3.I), leading to improvement in IFSS91 and ILSS97 of about 36 and

16 %, respectively, without compromising the tensile strength of the carbon fibres. The covalently grafted GO nanosheets onto fibre surfaces appeared wrinkled and roughened, leading the large plastic deformation at the GO/matrix interface, responsible for the improvement in the mechanical properties of the composites.

Figure 2.3: Scanning electron microscopy images of composite interphases containing graphene oxide nanosheets and graphite nanoplatelets. Various concentrations of GO nanosheets and graphite nanoplatelets deposited onto carbon fibres through addition to sizing solution (A-D and E-F, respectively). Monolayers and aggregates of GO nanosheets deposited onto carbon fibres through electrophoretic deposition (G and H, respectively). Single graphene oxide nanosheets grafted on carbon fibres (I). Adapted from references.88,91,94,95

The use of nanosheets in the interphase was found to improve the toughness and strength of the fibre/matrix interface, providing that no nanosheets aggregates formed.

However, the heterogeneous distribution of the nanoreinforcements did not lead to a well-defined, layered hybrid nanostructure, which could potentially allow for multiple crack deflection and strain hardening in shear. More understandings about the impact of the morphology of nanostructured interphases on the composite properties are needed.

2.2.4 Nanostructured interphases for toughened composite

A numerical study of nanosheets-reinforced FRP composites predicted that aligned and exfoliated nanosheets placed in the interphase composites could increase fatigue lifetimes compared to composites with clustered, aligned or randomly orientated nanosheets in the interphase.102 Nanosheets with a high degree of alignment around the fibres were predicted to lead to a stronger crack path deviation compared to a randomly orientated ones

(Figure 2.4).102 These nanostructured interphases are therefore expected to change the composite fatigue degradation process,103 extending their lifetime and damage tolerance.

Crack deviation and blocking, as well as nanoscale crack bridging can occur, which does lead to a dramatic change in crack initiation and propagation, significantly impacting the toughness of a composite.103 These toughening mechanisms, occurring at the nanometre lengthscale within the composite interphase, still need to be experimentally investigated. A hybrid layered nanostructured interphase, made of individualised and aligned anisotropic reinforcements, represents a promising route to increase the toughness of composites materials without compromising their stiffness and strength. Nevertheless, in order to achieve an effective nanostructure around the circumference of a fibre, the dimensions of the anisotropic reinforcements need to be adapted to the diameter of the fibres (refer to

Chapter.1). Hence, micrometre wide exfoliated nanosheets that have been intensively investigated, are not appropriate to design a suitable nanostructured interphase around conventional reinforcing-fibres.

A B

Figure 2.4: Macro-micro finite element simulation of hierarchical composites with nanostructured interphases. Schematic of the structure of hierarchical fibre-reinforced composite with nanosheet reinforcements (A). Top and lateral view of simulated interphases as well as calculated crack path of aligned exfoliated nanosheets, aligned clusters, randomly orientated nanosheets and randomly distributed clusters (B, C, D and E, respectively). Adapted from Dai et al.102

Similarly, the approach of engineering the interphase of ceramic matrix composites, investigated since the 1990s, follows the same logic.104,105 In such composites, the strain to failure of the matrix is lower than that of the fibres, leading to microcracking in the matrix. It has been found that the composites could be toughened by controlling the interphase, especially by using a layered interphase, leading to crack deflection and stress relaxation.104

An anisotropic layered structure, parallel to the surface of the fibres, was found to be the most effective way to improve toughness. A multilayer interphase105-107 made of pyrocarbon layers or a layered crystal structure interphase106 made of boron nitride, both showed the ability to deflect matrix cracks (Figure 2.5). Modelling also supported the experimental findings, promoting the use of a highly anisotropic tough interphase, which is applicable to any type of

composite material.108 However, the dimensions and geometry of the latter nanostructured interphase are not suitable for the proposed project and do not provide any strain hardening in shear.

Figure 2.5: Crack deflection in a layered interphase of ceramic matrix composites. Transmission electron microscopy images of (PyC-SiC)n multilayer interphase (A) and boron nitride layered crystal structure interphase (B). Adapted from references.105,106

From a general point of view, both modelling and experimental work agree on the potential of a well-organised layered nanostructured interphase providing a promising route to increase the toughness of FRP composite materials without sacrificing other mechanical performance. The assembly of rigid anisotropic reinforcements in the interphase in a homogenous manner, without overlap or formation of aggregates appears to be promising for the design of composites with improved toughness. The deflection of cracks at the reinforcement interface can potentially increase the absorption of energy in the materials.

Therefore, the anisotropy of the reinforcements needs to be optimised while allowing for a

conformal deposition in the circumference of a fibre. In addition, minimising the spacing between parallel reinforcements (high reinforcement volume fraction) in the interphase could also lead to progressive interlocking during pull-out, potentially providing strain hardening in shear and, therefore, stable fibre sliding.

The latter findings shed light on the importance of creating an interphase with a

“brick-and-mortar” structure made of aligned platelets, packed together and separated by a thin layer of a polymer matrix. Therefore, the classic “brick-and-mortar” structure of natural nacre, well-known for its toughening mechanisms including crack deflection and strain hardening through platelet sliding/interlocking, seems to be suitable for the design of an energy absorbing interphase for FRP composites.

2.3 “Brick-and-mortar” structure of nacre

2.3.1 Hierarchical structure of nacre

Nacre is one of the strongest natural composites and part of a two-layer system used by seashells as protection from their environment.1 Like many natural composites, the structure of nacre, found in the inner part of some mollusk shells, is a complex three-dimensional architecture, organised over multiple hierarchical levels109 (Figure 2.6).

The structure of nacre has been studied and reviewed by many researchers.1,13,16,109-120 Its hybrid structure is composed of 95 % of brittle inorganic aragonite (CaCO3) building blocks, around 200 to 900 nm thick and 5 to 8 μm wide (aspect ratio from 7 to 15).4 These blocks are organised in a “brick-and-mortar” structure, “glued” together by a soft chitin-containing

organic framework121-124. The organic layer is around 20-30 nm thick,116 which makes up the remaining 5 % of the structure. As a result of the bio-mineralisation of nacre, 10 to

50-nm-wide mineral bridges form in the organic phase joining platelets together.112,116,124-129

Similarly, asperities with a diameter of few tens of nanometres and a height of about 10 nm are also present on the surface of the aragonite platelets.16,116,118,130-132 At a smaller length scale, the aragonite platelets were found to be composed of cobble-like polygonal nanograins with a diameter of about 30 nm,118,133-135 which add a level of hierarchy to the

“brick-and-mortar” structure of nacre.

Figure 2.6: Representation of the hierarchical structure of natural nacre at various length scales. Abalone shell (A), inner nacreous aragonite layer “N” and prismatic calcite layer “P” (B), growth bands (C), aragonite platelets organised in columns (D-F) and aragonite platelet roughness arising from aragonite nanograins and asperities (G). Copied from Luz et al.114

Essentially made of brittle aragonite building blocks, one would assume that nacre would exhibit brittle behaviour. However, the combination of a small fraction of organic

phase, coupled with the specific “brick-and-mortar” architecture and arrangement at different length scales, leads to astonishing mechanical properties and progressive failure. The overall performance of nacre at the macroscale arises from mechanisms operating synergistically over several lengthscales during deformation, the combination allows for relatively high stiffness (~ 60 GPa),5,120,136-140 strength (~ 140 MPa)5,120,136-140 and toughness

(~ 1.24 kJ/m2).5,120,136-141 The stiffness and strength of the aragonite platelet are about

100 GPa and 300 MPa, respectively.137 Toughness of nacre is three orders of magnitudes higher than that of the aragonite platelets alone, due to multiple toughening mechanisms,133,135 making nacre effective at stress redistribution,5,120,139 energy absorption,5,120,141,142 flaw-tolerance,134,140,143,144 crack deflection,2 crack bridging,120,141,144 and strain hardening.16,131

2.3.2 Toughening mechanisms occurring in nacre

Whilst the high strength and stiffness result from the high content of aligned aragonite platelets, the primary toughening of nacre is produced by the pull-out of the few micrometre wide platelets from the organic phase in which they are embedded. The pull-out/sliding of the platelets110 occurs initially through organic matrix deformation, more specifically via platelet interface rupture rather than platelet failure due to a limited platelet aspect ratio, about 10.

Hence, the tensile strength of the platelets is higher than the shear yield strength of the interface. A critical aspect ratio of the platelets can be determined, at which the failure mode of the “brick-and-mortar” structure is altered, from platelet pull-out to platelet failure

(Figure 2.7).

Because of the “brick-and-mortar” organisation of nacre, the balance between tensile strength of the platelets (σp) and the yield shear strength of the interface (τy) determines the mode of failure of the material. The shear lag model13,14 can be used, according to which the applied load is transferred to the inorganic platelets through shear stresses in the biopolymer matrix. In this model, for a perfect bonding between the ductile organic and rigid inorganic phases (interfacial shear yield strength (τy) higher than the organic matrix shear strength

(τm)), the ultimate tensile strength (σc) of the hybrid nanocomposite can be calculated as follows:

σc = Vp ∙ σp + (1 − Vp ) ∙ σm (2-1)

where, Vp and σm are the volume fraction of platelet and tensile strength of the soft organic matrix, respectively. A critical aspect ratio (sc) of the platelet, determining the mode of failure of the composite and, therefore, its strength and toughness, can be calculated:

σp sc = (2-2) τy

Two different failure modes of nacre can occur depending on the aspect ratio of the platelets, leading to different strength and toughness of the composite:

 If s > sc, the platelets break leading to a brittle failure (low toughness) of the

composite.

 If s < sc, platelet sliding and subsequently pull-out of the organic matrix is promoted,

leading to high fracture toughness. The matrix yields before the platelets break, thus

toughening mechanisms such as platelet pull-out and matrix plastic flow occur before

the complete failure of the composite.

Figure 2.7: Schematic of natural nacre deformation and mechanical response as a function of platelet aspect ratio. Fracture mechanisms of “brick-and-mortar” structures containing platelets with post-critical, s > sc, (platelet fracture) and sub-critical, s < sc, (platelet pull-out) aspect ratio according to the shear lag model (A). Associated variation of tensile strength as a function of the aspect ratio of the reinforcements (B). Adapted from Wang et al.13

The subsequent sliding of the platelets involves frictional interactions between the platelets as well as plastic deformation of the organic phase, but is arrested by progressive interlocking, leading to strain hardening.16,130 Platelet interlocking are likely to act as active sites for crack deflection. Although the exact mechanism of platelet interlocking is still unclear in natural systems, wedging,1,145,146 asperities,16,116,130,131 mineral bridges,124-126 nanograin rotation,133 platelet waviness1,145 and negative Poisson’s ratio146,147 have all been proposed (Figure 2.8.A). In addition to platelet interlocking, at a smaller lengthscale, substantial deformation of the organic layers (providing cohesion) via unfolding and breakage of domains and platelet nanograins rearrangement enable for a large amount of energy to be dissipated throughout the material.117,123,148-150 Hence, multiple toughening mechanisms, at various lengthscales, are expected to occur synergistically (Figure 2.8).

A B C

Figure 2.8: Models of toughening mechanisms occurring in natural nacre over multiple length scales. Mineral bridging, nanoasperities and waviness at the platelet length scale during platelet pull-out, providing interlocking and crack deflection. Rotation and rearrangement of nanograins at the platelet surface length scale (B). Unfolding of molecular sacrificial domains in the organic matrix (C). Adapted from Kakisawa et al.148

The formation of a large frontal process zone after crack initiation in nacre was experimentally evidenced by many researchers.145,151 It was shown that cracks develop into an inelastic zone leading to irreversible strain and band dilation in the structure of nacre, providing substantial energy absorption and spreading. Barthelat et al.5 showed that a

“process zone parameter”, directly related to the toughness of the material, can be expressed as followed:

2 1 1 ∅ Gi ∙ Ji α = ∙ ∙ ∙ 2 (2-3) 4 t 1 − ∅ τs where Φ is the platelet fraction in the process zone and 푡 the platelet thickness. Gi, Ji and τs are the shear modulus, toughness and shear strength of nacre, respectively. The higher the unit volume of material in tension, the higher the toughness of the material.138 Thus, one can expect a denser process zone when reducing the thickness of the platelets and, therefore, more toughening to occur in the materials.5,138

A B

Figure 2.9: Crack propagation within natural nacre. Optical microscope images of a crack propagating in nacre and the frontal zone ahead of the crack as evidenced by whitening (A). Schematic of the propagation of a crack in nacre, presenting crack bridging and deflection in the material as well as platelet pull-outs in the frontal zone (B). Copied from Bekah et al.5

When a crack initiates from a defect within nacre structure, multiple deflections occur at the building block interfaces2 as well as crack bridging. As the crack propagates, the platelets are pulled-out ahead of the crack tip, providing resistance to further propagation.

Ahead of the crack, a great number of platelet sliding occurs in a zone called the frontal zone associated with inelastic deformation of the interfaces, caused by the rearrangement of organic molecules (Figure 2.9). Hence, a toughness of about 0.3 kJ·m-2 was measured in nacre,1,137,141,152-154 20 to 30 times higher than that of pure aragonite. In this sense, interlocking of the platelets hinders the propagation of a crack, redistributing the stresses in the material and allowing for energy dissipation through deformation of the organic network.

The platelet pull-out and subsequent interlocking is considered to be the main source of toughness in natural nacre.2,130,137,155

2.3.3 Tensile and shear responses of nacre

The sophisticated microstructure of nacre, composed of several hierarchical length scales, is key to significant amplification of the properties of its individual weak constituents.109 The stiffness and strength of nacre is due to the presence of a high volume fraction of rigid platelets with high anisotropy, which permits from efficient load transfer through the matrix layer. A large number of theoretical and numerical models5,6,138-140,143,156 predict further improvement in the stiffness and strength by increasing the aspect ratio of the platelets, while retaining the pull-out failure mode of the material143. Because the degree of hydration of nacre has a significant impact on the mechanical behaviour of the organic phase,137 dry and wet nacre specimens were tested in tension.1,130,151 The stress-strain curves reveal a linear elastic response of dry nacre up to a failure stress between 95 and 135 MPa, at which point a sudden brittle failure occurs. This behaviour is similar to the failure of aragonite ceramic. On the other hand, wet nacre exhibits a linear elastic region and subsequently a large inelastic deformation occurring at a stress of about 70 MPa throughout the whole composite. As the strain further increases, the material becomes harder, indicating strain hardening due to platelet interlocking and breaking of organic sub-units, as mentioned previously. The ultimate strain to failure, in tension, was measured to be about 1 % but could locally reach values of about 1.5 to 2 % (Figure 2.10.A).

Figure 2.10: Tensile and shear responses of natural nacre. Experimental tensile stress-strain curve and associated schematic of deformation mechanism of nacre (A and B, respectively). Experimental shear stress-strain curve and associated schematic deformation mechanism of nacre – platelet waviness leading to a resistance to sliding and, therefore, lateral expansion (C and D, respectively). Copied from Espinosa et al.16

Nacre’s response in shear is especially relevant for the proposed project1

(Figure 2.10). The shear modulus of the material was measured to be 10 and 14 GPa for wet and dry conditions, respectively. After an initial elastic deformation, both wet and dry nacre exhibited large inelastic deformation up to a failure shear strain of about 15 % and 8 %, respectively.1 The stress at which wet nacre started to deform inelastically, triggering platelet sliding, was about 20 MPa and progressively increased to about 50 MPa as strain hardening

occurred in the material. In dry nacre, strain hardening caused by platelet interlocking was also observed, but at higher stresses, between 55 and 70 MPa. The reduction in stresses at which the platelets slide over one another in wet nacre as compared to dry nacre, can be explained by the hydrated organic phase acting as a lubricant and limiting friction in the composite. These stresses are significantly lower than the strength of the platelets, which indicates that shearing of the interfaces is the main deformation mechanism in shear

(interlocks, friction and plastic deformation of the organic phase). Much more strain hardening was produced in shear than in tension, along with a more homogenous strain distribution throughout the material.1

2.3.4 Scaling behaviour of nacre “brick-and-mortar” structure

For the proposed project, the “brick-and-mortar” structure of nacre must be scaled down by more than one order of magnitude in order to achieve conformal coatings around reinforcing-fibres such as glass and carbon fibres (Chapter.1). The origin of high elastic modulus and strength of nacre, arising from a high volume fraction of aligned stiff inclusions with high aspect ratio, is well-understood and can be reproduced at any lengthscales.6 Models predict that the toughening mechanisms of nacre should be scale invariant,5,6 as long as the aspect ratio of the platelets remains below the critical value and that classic geometry and phase proportions of nacre are maintained.

Toughening behaviour are not straightforward and has been analytically studied over the past few years.5,138 Toughness models, including the combination of bridging and process zone toughening, enabled guidelines for the design of artificial nacre structures. While the aspect ratio should be high, but still below a critical value (< sc) promoting pull-out, and the

platelet volume fraction at least 90 vol.% to activate toughening mechanisms, the amount of toughness can be amplified by the use of small platelets (larger process zone toughening).

The overall toughness J and toughness generated by bridging J0 in “brick-and-mortar” structures are given by:

J0 J = (2-4) 1 − α where α is the “process zone parameter” (refer to 2.3.2). A decrease in platelet size would lead to an increase in process zone toughening and, therefore, to higher overall toughness amplification (Figure 2.11). In addition, nanometres thick platelet can be much stronger than mciroplatelets, allowing the use of larger aspect ratio and, therefore, higher composite performance. However, it is impossible to produce homogenous nanostructures containing nanosheets (~ 1 nm thick) with high platelet volume fraction as this would imply a polymer interlayer much thinner than a nanometre. Even though higher strength, modulus and toughness could be produce using small platelet, the design of well-organised nanostructures containing nanometres thick platelets still remain a challenge.

Figure 2.11: Toughness amplification in “brick-and-mortar” structures. Effect of platelet size (t) on the toughness (J), with a fixed aspect ratio (s = 10) and platelet volume fraction (φ = 0.9). Adapted from reference.5

2.3.5 Flaw tolerance of platelets

Defect-free materials should be able to carry stress up to their theoretical strength,143 however real materials, such as aragonite platelets in nacre, exhibit defects. The actual strength of the platelets σp is lower than the theoretical strength σth and can be calculated using the Griffith criterion:

σp = α ∙ Ep ∙ ψ (2-5) with,

γ ψ = (2-6) Ep ∙ t

where t, Ep and γ are the thickness, elastic modulus and the surface energy of the platelet and

α a parameter depending on the crack geometry. The thickness t* at which the material reaches a strength equal to the theoretical strength, and hence becomes flaw-tolerant, is defined by:

γ ∙ E ∗ 2 p t ≈ α ∙ 2 (2-7) σth

According to equation 2.5 and equation 2.6, thinner platelets benefit from higher strength and, therefore, allow for higher anisotropy (sc) when part of a “brick-and-mortar” structure (refer to equation 2.2). Therefore, for a half-cracked platelet (α approximately equals to √휋), the critical aspect ratio sc of platelets at which both constituents fail at the same time can be expressed as a function of the platelet thickness:143

1 π ∙ Ep ∙ γ sc = ∙ (2-8) τy t where 휏푦 is the interfacial shear yield strength. The thickness of the aragonite platelets in nacre was measured to be about 200 to 900 nm, higher than the critical thickness of about

30 nm, at which the material becomes flaw-tolerant and maximises its strength.143 The large thickness of the aragonite platelets may be a result of constrained optimisation of the biocomposite, limited by the minimum size of single biopolymer molecule. On the other hand, the biomineralisation process of aragonite platelet could be limiting their minimum thickness. However, the platelets can be seen as nanocomposites made of nanograins of aragonite,134 smaller than the critical thickness, which then could explain why nacre is so flaw-tolerant.

Several models provide guidance towards the design of artificial nacre, shedding light on the potential to improve toughness5,6,138,157 and stress distribution140 by decreasing the thickness of the platelets, but also increasing the strength6,14 and stiffness5,6,138,140,143 of nacre

mimics by using a large aspect ratio. The aspect ratio of the platelets has been identified as a critical parameter.5,6,138 High stiffness and high strength are achieved by increasing platelet anisotropy. Although the characteristic toughening mechanisms of nacre are expected to be scale invariant5 the model predict that the best nacre mimics should be achieved with small platelets at the optimum critical aspect ratio leading to platelet pull-out.6 Over the past decade, many attempts have been carried out to design and manufacture composites inspired by the structure of nacre at different lengthscales.7,9,158,159

2.4 Nacre-inspired structures

Artificial “brick-and-mortar” structures, inspired by natural nacre, have been assembled at different length scales and using various methods, over the past two decades

(reviews7,159-161). The mechanical properties and deformation mechanisms of natural nacre have motivated many researchers to develop structures exhibiting the combination of high toughness and strength, a long-lasting challenge in the field of load-bearing materials.

However, the design of “brick-and-mortar” structures with suitable phase proportions, geometry, aspect ratio and degree alignment is not a trivial task. A broad variety of techniques to assemble inorganic blocks and organic polymer into a “brick-and-mortar” architecture have been investigated, leading to varying composite morphologies and mechanical properties.

2.4.1 Assembly methods of nacre-inspired structures

Many methods9 have been used to mimic the structure of natural nacre at the same length scale as well as at nanometre length scale. Among all the methods used to develop nacre-inspired composites, LbL assembly,14,162-169 vacuum filtration,170-180 electrophoretic deposition,181-183 solution-casting164,184-186 and vacuum evaporation187-191 are the most commonly used and most promising methods (Figure 2.12), especially at the nanometre length scale.

A D

Figure 2.12: Most commonly used methods to produce nacre-inspired composites. Vacuum-assisted filtration (A), vacuum evaporation (B), solution casting (C), electrophoretic deposition (D) and layer-by-layer assembly (E). Adapted from references.165,173,183,184,190

The LbL assembly192,193 is particularly successful and interesting for the proposed research as it enables the deposition of particles onto any charged substrate geometry, such as the surface of a fibre, with a good monolayer control. The LbL assembly is based on

electrostatic attraction of charged particles on oppositely charged surfaces, upon dipping in particles suspensions or PE solutions. The electrostatic repulsion forces between charged particles allows for the deposition of monolayer without excess particles192-194 (Figure 2.13).

The LbL assembly approach is appealing as it is a cheap process that allows good control of the thickness of the coating at the nanometre scale, suitable for the sequential assembly of PE polymers and charged inorganic nanoparticles.158 High affinity and compatibility is achieved in the composite due to strong electrostatic and van der Waals interactions at the inorganic/organic interface. Furthermore, multilayer structure deposition can be achieved on charged, three-dimensional substrates with all kinds of geometry, as the absorption of particles is self-limiting and occurs uniformly. The LbL assembly has been used to produce thin coatings167,195-206 a few hundred nanometres thick as well as free-standing films14,162,163,165,167,168,205-208 a few microns thick, containing a nano- or micro-organised layered structures inspired by that of natural nacre.

2.4.2 Micrometre lengthscale nacre-inspired structures

14,169,209,210 Various types of micrometre-wide platelets, such as Al2O3 and Layered

Double Hydroxide (LDH)165,211 platelets, combined with chitosan, poly(vinyl acetate) (PVA),

PMMA or polyimide polymer resin (Figure 2.14), were used to produce nacre-inspired composites at the same micrometre length scale. More recently, composite structures made of

187,201 CaCO3 platelets embedded in gelatine or poly(vinyl pyrrolidone) have also been investigated. In all micrometre lengthscale composite structures, the aspect ratio of the platelets was selected to be below the critical value promoting platelet pull-out, typically

211 14 about 30 to 40 for the LDH and Al2O3 platelets.

Figure 2.14: Scanning electron microscopy images of nacre-inspired composite structures at the natural micrometre length scale. Composites containing 11 and 60 vol.% of Al2O3 platelets in chitosan (A and B, respectively). Composites made of LDH/PVA with 2 and 4 wt.% of PVA (C and D, respectively). CaCO3/gelatine composite film containing 95 wt.% and 76 wt.% of CaCO3 (E and F, respectively). Adapted from references.14,187,211

Nacre-inspired structures, at the same length scale as natural nacre have been manufactured but are generally limited to relatively low platelet concentrations, roughly about 15 to 20 wt.%. (Figure 2.14.A-B). Due to the difficulty of preparing a well-organised

structure at higher loading, the tensile properties start to degrade as a result of platelet misalignment and clustering in the composite.14,209,210 For instance, Bonderer et al.14 produced Al2O3-chitosan composites by bottom-up colloidal assembly and suggested that swelling of the polymer phase during dipping could be responsible for platelet misalignment and incorporation of voids at high inorganic content. The best modulus, strength and strain to failure of about 9.6 GPa, 315 Mpa and 22 %, respectively, were reported for composites containing 15 vol.% of Al2O3 platelets. Non linearity in the stress-strain curve was observed with a yield strain and yield strength of about 3.9 % and 275 MPa, respectively

(Figure 2.15.A). Transmission electron microscopy (TEM) investigation of composite fractured surface confirmed the pull-out mode of the Al2O3 platelets, responsible for the non-linearity in the stress-strain curve and toughening.209

Even though most of the micrometre length scale nacre-inspired structures possess a low inorganic phase content, recent work has reported a structure made of PVA and LDH platelets (Figure 2.14.C-D), which was produced by a similar bottom-up assembly with an inorganic phase as high as 69.7 wt.%.211 The tensile properties of this nacre mimic were measured about 169.4 MPa and 11.16 GPa for its strength and modulus, respectively. The strength of the composite is slightly superior to that of natural nacre, while its tensile modulus remains significantly lower. The stress-strain shows that the composite broke after elastic deformation without substantial plastic deformation (Figure 2.15.B). Due to the high content of inorganic platelet, voids were observed in the composite cross-section, which could have led to local high stress concentration. The relatively high strength of nacre-inspired composites were explained by the fact that the appropriate platelet aspect ratio (19 to 37) led to platelet pull-out as well as the good hydrogen bonding interaction between the platelets and the matrix. The interfacial interaction between the platelets and the polymer was improved by functionalisation of the platelets with a hydrophobic amine terminated silane,

which also allowed the platelets to remain at the air/water interface in a very well defined two-dimensional orientation during assembly.14,165,169,211 Nacre-inspired structures made of

CaCO3 platelets have also been recently investigated, these were synthesised by vacuum-evaporation-induced self-assembly187 or polymer-mediated minineral growth combined with LbL assembly.201 Various inorganic phase proportions were produced in a gelatine matrix (proteins derived from bovine skin), revealing progressive disorder in the structure with increasing platelet content (Figure 2.14.E-F). However, neither the phase alignment nor the mechanical properties of the composites was reported.

A B

Figure 2.15: Tensile response of nacre-inspired composites at the natural micrometre length scale. Stress-strain curves of composites containing 0 to 15 vol.% of Al2O3 platelets (Vp) in chitosan (A) and stress-strain curves of composites made of LDH/PVA containing 96.9, 69.7, 23.2 and 14.1 wt.% of LDH (“4”, “3”, “2” and “1 wt.% PVA/LDH”, respectively) (B). Adapted from references.14,211

Generally, the use of micrometre-wide platelets with an aspect ratio of 20 to 40 only allows for good alignment of the inorganic building blocks in an organic matrix at high organic phase content,210 which cannot reproduce the platelet interlocking mechanism observed in nacre. As a consequence, the “brick-and-mortar” structure of nacre could not be successfully reproduced, although the composites failed under platelet pull-out leading to

toughness, strength and stiffness in the range of [4 – 9 MJ·m-3], [100 – 300 MPa] and

[7 - 12 GPa], respectively. One can assume that a high volume fraction of inorganic platelets

(90 vol.%) is a requirement to obtain a high fracture toughness (dissipation and spreading of crack energy within a tortuous path) by activation of toughening mechanisms, such as crack deflection and platelet jamming.

2.4.3 Nanometre lengthscale nacre-inspired structures

The “brick-and-mortar” structure of nacre has also inspired many researchers to design a wide range of nanostructures with unusual mechanical properties, such as the combination of high toughness and high strength. Highly anisotropic nanoreinforcements, such as clay,162,163,170-172,175,184,185,189,191,196,205,212-222 graphene oxide160,164,169,173,174,176-

180,186,188,190,222-227 or LDH228 nanosheets have been investigated for the development of nanocomposites (Figure 2.16), using various aspect ratios184 and interface interactions.160

Whilst often termed nacre mimetics, the phrase is generally misleading because the aspect ratio of the inorganic nanoreinforcements and phase proportions are very different from those found in natural nacre. Hence, layered composite seems to be a more appropriate term to define these nanostructures.

Figure 2.16: Scanning electron microscopy images of layered nanocomposites. Cross-sections of clay nanosheet/PVA layered nanocomposites with different platelet aspect ratio (25, 140, 260, 750 and 3500) and polymer content from about 55 to 25 wt.% (A, B, C and D, respectively). Cross-sections of graphene oxide/sodium alginate layered nanocomposites with a graphene oxide nanosheet proportion of 0, 18.6, 56.3 and 100 vol.% (E, F, G and H, respectively). Copied from references.180,184

The use of various nanosheets aspect ratios allowed to understand the relation between the morphology of layered nanocomposites and their mechanical properties.184 The degree of organisation of the nanocomposites was shown to increase with the anisotropy of the nanosheets, accompanied with a reduction in the polymer phase proportion

(Figure 2.16.A-D). While most of the layered nanocomposites contain a polymer phase proportion of at least 30 wt.% due to the limitation that the organic layer thickness is similar to that of the nanosheet reinforcement (~ 1 nm), nanocomposites made of GO nanosheets with an inorganic phase proportion from 0 to 100 vol.% were produced180 (Figure 2.16.E-H) to investigate the impact of the inorganic content onto the morphology and mechanical properties of the nanocomposites. The layered nanocomposites showed a good GO dispersion and alignment for GO volume fraction of up to a volume fraction of 25.5 %. For higher volume fractions, nanocomposites tend to present significant nanosheet overlap (direct contact) and inhomogeneous exfoliation. At a nanosheet content higher than 70 wt.%, the structures become influenced by tactoid formation,189 limiting the assembly into a

homogenous layered arrangement to low reinforcement volume fractions. Further increase in

GO nanosheet concentration led to wavy heterogeneous structures.

It was shown by Zhang et al.160 that functional groups, on the surface of the GO nanosheets, allowed for strong interaction with the polymer binder through hydrogen bonding, leading to increase in strength and stiffness. Similarly, clay nanosheets have been combined with various soft polymer matrixes, tuning the interfacial interaction between the two phases by covalent or ionic cross-linking, leading to increased mechanical properties.163,171,172,208,229 Among all the different types of interaction between the nanosheets and the matrix, covalent bonding appeared to be the most successful route to increase the tensile strength of the nanocomposites, to values comparable to that of natural nacre.160 Most of layered nanocomposites, with an inorganic content about 70 wt.%, present a strength and toughness in the range of [100 – 200 MPa] and [0.1 – 4 MJ·m-3], respectively, as compared to

200 MPa and 2.6 MJ·m-3 for natural nacre. While the strength of the nanocomposite is similar to that of natural nacre, the toughness is usually lower. The low fraction of nanosheets and heterogeneous nanostructure of the layered nanocomposite limit its mechanical performance.

The high aspect ratio of the nanosheets ( > 500) is responsible for the high strength of the composite as well as its brittle failure, as a result of platelet fracture mode (Figure 2.17.A). A reduction of the aspect ratio of the nanosheets, down to 25, below the critical value associated with the strength of the platelets, was shown to lead to inelastic deformation accompanied with large reduction in strength (Figure 2.17.A).184

A B

Figure 2.17: Tensile response of layered nanocomposites at the nanometre length scale. Stress-strain curves of layered nanocomposite containing synthetic clay nanosheets with varying aspect ratio (A). Stress-strain curves of layered nanocomposite made of GO, reduced GO, GO (95 wt.%)/chitosan, reduced GO (95 wt.%)/chitosan and chitosan films (curves 1, 2 3, 4 and 5, respectively) (B). Adapted from references. 174,184

A synergistic effect174,190,224,230 of different interfacial interactions between the two phases of the nancomposites revealed further improvement in both strength and toughness, to values higher than those of natural nacre, in the range of [300 - 500 MPa] and

[2 - 18 MJ·m-3], respectively. Even though the homogeneity of the inorganic phase in these nanocomposites is still limited, the highest nanocomposite mechanical properties were obtained for layered nanocomposite containing an inorganic phase proportion similar to that of natural nacre (94.4 wt.%), at which both strength and toughness were optimised. A tensile strength of 526.7 MPa and a toughness of 17.7 MJ·m-3 were obtained for these nanocomposites, utilising the synergistic effect of hydrogen and covalent bonding between chitosan and GO nanosheets,174 accompanied with a brittle failure (Figure 2.17.B).

Figure 2.18: Summary of the different interfacial interactions utilised in layered nanocomposites containing graphene oxide nanosheet reinforcements. Non-covalent bonding and covalent bonding interactions between graphene oxide nanosheets and polymer binder as well as synergistic interactions. Copied from Zhang et al.160

The layered nanocomposite structures inspired by nacre exhibit the possibility to design materials with the combination of high toughness and strength. The mechanical properties of these nanocomposites were found to reach and sometime overcome those of natural nacre. However, these nanostructures developed with nanosheets are not equivalent the structure of nacre. The dimensions of the reinforcements, namely, their thickness and aspect ratio, do not enable for the retention of the classic “brick-and-mortar” structure

(geometry and phase proportions) of nacre. Hence, the typical toughening mechanisms of nacre such as platelet interlocking were not reproduced. In these systems, the platelet thickness is excessively reduced to less than or similar to the polymer binder thickness, yielding hybrids with a significantly greater organic content than nacre or heterogeneous structures.

2.4.4 Hierarchical, fibre and ternary nacre-inspired structures

The complex architecture of natural nacre spread over multiple length scales has motivated researchers to design advanced materials with multiple level of hierarchies.231

Other features of natural nacre, such as asperities in aragonite platelets leading to increased resistance during platelet sliding and chitin fibrils acting as secondary toughening phase, inspired the use of roughened platelets232,233 and the design of ternary nacre-inspired structures,169,173,176,191,221,224,234-237 respectively. Finally, in addition to layered nanocomposites, the structure of natural nacre also inspired the design of new fibres238-241 for use in tough and multifunctional composites.

Similar to natural nacre, a two level of hierarchy composite structure was produced using a two-step polymer-coating process followed by directional consolidation, enabling large-scale composite production (Figure 2.19.A). The obtained hierarchical structure mimic displays a “brick-and-mortar” structure (Figure 2.19.B) with improved mechanical properties as compared to a 1-level hierarchical composite made of the same materials. These improvements in the mechanical properties of the structure, by increasing the number of level of hierarchies, was also predicted by a numerical model.242 The composite manufacturing method shows the possibility to process any ceramic and polymer materials with controllable phase proportions, allowing to investigate hierarchical structure-property relationships.

B C

Figure 2.19: Manufacture of multi-level hierarchical “brick-and-mortar” nanocomposites. Processing steps to making hierarchical structure of TiO2 particles encapsulated in PMMA in a radical emulsion polymerisation step and a second polymer matrix (PVB) (A). Scanning electron microscopy images of a PVB-coated agglomerate with 2 levels of hierarchy (B) and of the interface between first-level particles and PVB interlayer at different magnifications (C). Copied from Brandt et al.231

Similarly to planar nanocomposites made of GO nanosheets, a layered arrangement inspired by the structure of natural nacre was also reproduced within the fibre with a diameter of a few tens of micrometres (Figure 2.20) using a wet-spinning assembly method of polymer-grafted GO or a polymer and GO dispersion. While planar layered nanocomposites are limited to centimetre-scale films, nacre-inspired fibres were produced continuously,

opening-up possibilities for manufacturing of high toughness nanocomposites and potentially tough fibre-reinforced materials. After fibre spinning, the fibres with a belt-shape, were twisted into circular shape with spring-like coils, reducing their diameter from about 225 to

110 μm (Figure 2.20.A-C). At the nanoscale, the organisation of the GO nanosheets embedded in PVA was found to exhibit a layered nanostructure similar to planar nanocomposites made from GO nanosheets, yet with additional circular geometry

(Figure 2.20.D-F). As observed for planar layered nanocomposites, the homogenous fibre nanostructure was characterised by a high polymer content of about 70 wt.% due to the limited thickness of the inorganic nanosheets as compared to the thickness of the organic layer. The mechanical properties of the nacre-inspired fibre, such as its strength and toughness, were measured about 270 MPa and 460 MJ·m-3, respectively, which are still low compared to other types of reinforcing fibres, limited by the properties of the layered nanocomposite structure. On the other hand, the proposed project, based on the adaptation of a nacre-like nanostructure onto the surface of a fibre, could potentially combined the high mechanical performance of reinforcing fibres, such as carbon fibres, with interfacial toughening mechanisms.

Figure 2.20: Morphology of nacre-inspired spun fibres. Low (50 μm scale bar) and high (2.5 μm scale bar) magnification scanning electron microscopy images of belt-like shape spun (A and D, respectively), spun/twisted (B and E, respectively) and further twisted (C and F, respectively) nacre-inspired fibres made of graphene oxide nanosheets incorporated into PVA. Copied from Zhang et al.238

Since 2014, nanosheet reinforcements, such as graphene oxide or clay nanosheets, have been combined with other types of reinforcements, with either 1-dimensional (1D) or

2-dimensional (2D) geometry to develop ternary layered nanocomposite structures, leading to synergistic effects. The incorporation of secondary 1D reinforcements in addition to 2D nanosheets leads to improvements in both strength and toughness of the nanocomposites.191,234,236,237,243 Additional 1D materials in layered nanocomposites were found to bridge propagating cracks and provide further resistance to platelet sliding.191

Among the different 1D materials used to design ternary nanocomposites, nanofribillated cellulose was found to cause crack bridging and undergoes pull-out followed by fracture in tension.191 A strain hardening response of the ternary nancomposite materials was observed as a result of the activation of multiple nanosheet sliding sites caused by the incorporation of

1D reinforcements (Figure 2.21.A). Furthermore, 1D reinforcements were also found to interlock (Figure 2.21.B), leading to increased toughness and strength of the ternary

nanocomposites,237 as compared to binary layered nanocomposites, from about 0.75 to

2.75 MJ·m-3 and 80 to 120 MPa, respectively The secondary 1D reinforcement placed between two layers of GO nanosheets were shown to lead to mechanical friction, resistance to platelet sliding, crack bridging and crack deflection throughout the nanocomposite.

Therefore, the fracture surfaces of these nanocomposites indicate nanosheet pull-out, significant crack deflection, multiple crack propagation, crack branching, wavy cracks, interlocked and bridged 1D reinforcement. Similarly, the combination of two different 2D reinforcements, such as GO and clay nanosheets235 (Figure 2.21.C) or GO nanosheets and molybdenum disulphide sheets173 (Figure 2.21.D), was found to increase both toughness and strength of the ternary nanocomposite through synergistic effects.

Figure 2.21: Deformation mechanisms in tension of ternary layered nanocomposites. Stress-strain curves of binary and ternary layered nanocomposites (A). Scanning electron microscopy images of fractured ternary layered nanocomposites and illustrative schematic of the fracture mechanism for clay/cellulose nano-fibril/chitosan , GO/clay/PVA and GO/MoS2/polyurethane (B, C and D, respectively). Copied from references.173,191,235,237

Synergistic effects caused by ternary layered nanocomposites, or the utilisation of various interfacial interactions together, result in a further increase of the mechanical properties of the nanocomposites, possessing both increased strength and toughness

(Figure 2.22). The best mechanical properties of layered nanocomposite (Figure 2.17.B) arise from a synergistic effect of multiple interfacial interactions.174 On the other hand, synergistic effect of multiple building blocks has the advantage of improving the fatigue behaviour of nanocomposites.191 Further improvement in the mechanical properties of the ternary nanocomposites should be possible by combining both synergistic effects, namely, by using various interface interactions and building blocks simultaneously in the nanostructure.

2.5 Summary

To summarise, FRP composite materials are very attractive because of their excellent mechanical properties at low weight, making them an ideal candidates for load bearing applications. However, they suffer from sudden and catastrophic failure caused by the formation of a critical cluster of fibre breaks, leading to high local stress concentrations at low strains to failure. The development of FRP composites with both high strength and toughness is a long-lasting challenge. Mechanisms to promote inelastic deformation and

stress dissipation in composites are therefore needed. Engineering the interphase of the FRP composite is one of the approaches to increase the toughness of composites while maintaining high stiffness and strength. Toughening mechanisms are required to occur in the interphase, while allowing good load transfer through strong interfacial interaction between the matrix and the fibres. Experimental and modelling evidence point to the possibility to design an anisotropic interphase made of individual and ordered rigid, 2D reinforcements, all separated by a soft polymer binder, similar to the “brick-and-mortar” structure of natural nacre. Crack deflection at the platelet interface and strain hardening of the structure in shear, through platelet sliding and subsequent interlocking, can potentially dissipate the energy arising from fibre fragmentation and allow progressive locking of fibre slippage in composites, respectively. This new concept requires a nacre-like interphase. Luckily, the mechanical properties of nacre in planar forms have already attracted considerable research effort, at various lengthscales. At the nanometre length, suitable for the proposed research, an amplification of the toughness is expected through the formation of a denser process zone.

Many researchers attempted to develop nanocomposites inspired by the layered arrangement of natural nacre at the nanometre scale, using nanosheet reinforcements combined with various types of soft polymer binders. However, the use of inorganic nanosheets could not reproduce the “brick-and-mortar” structure of nacre due to their large aspect ratio and limited thickness. The mechanical properties of these nanocomposites were determined to be as high as those of natural nacre, even better for layered nanocomposites with synergistic effects arising from either multiple types of interface interactions or building blocks. Nevertheless, the true structure of nacre has not been truly reproduced as the thickness of the inorganic reinforcement was in most cases similar to that of the polymer layer, leading to an excess of organic phase or the formation of tactoid structures. Indeed, the dimensions and aspect ratio of the nanosheets did not lead to well-defined platelet monolayers due to particle stacks and

misalignment. On the other hand, well-aligned micrometre-wide platelets could not be stacked to produce a sufficiently high volume fraction. Ideally the nacre mimic should be made of small nanometre length scale reinforcements to promote robustness, flaw tolerance and amplification of toughness. The performance of the layered nanocomposites has merely been investigated in tension, while the response in shear is still unknown. Hence, it is of great importance to reproduce the structure of nacre with the correct phase proportion, geometry and aspect ratio at the nanometre length scale to adapt the toughening mechanism occurring in natural nacre in shear in the interphase of FRP composites.

Materials, Methods & Characterisation

This chapter presents the materials used to develop and design a new generation of hierarchical composites with a nanostructued interphase. The methods used to synthesise nanoplatelets, assemble them into a nanostructured coating and eventually manufacturing of hierarchical composites are presented, along with all characterisation techniques utilised over the different length scales of the hierarchical material.

3.1 Materials for the development of hierarchical composites with nanostructured interphases

The developed hierarchical composite structure spreads over multiple length scales, namely, a nanostructured interphase and micrometre-wide reinforcing fibres, all contained in a resin matrix. Rigid platelets and soft polymers were required to produce a nacre mimic, and more especially, its toughening mechanisms based on the pull-out of the inorganic inclusions.

Therefore, polyelectrolytes were selected and combined with charged inorganic platelets in order to use a Layer-by-Layer (LbL) assembly method based electrostatic attraction forces.

Any type of hard anisotropic platelets could be considered as reinforcing blocks to mimic nacre via a LbL assembly, as long as they exhibit a high surface charge density. The materials used to create the overall structure are described in the following section.

3.1.1 Polyelectrolyte and platelets for nanostructured coating/interphase assembly

[Mg2Al(OH)6]CO3.nH2O LDH platelets synthesised in the laboratory with varying dimensions and size distribution were used as two-dimensional reinforcement in the

“brick-and-mortar” structure of the nanostructured coating/interphase. LDH platelets were selected for their well-defined hexagonal shape,244,245 tuneable dimensions,244,246 in the suitable range to coat typical reinforcing fibres (as presented in Figure 1.4) and high surface charge density,246 which make them interesting candidates for the deposition of self-assembled layers with a high degree of packing. The general formula of the composition

2+ 3+ 3+ 2+ of LDH platelets is [M 1-xM x(OH)2](Ax.nH2O), where M and M are divalent and trivalent metal ions and A- is an anion.247 Therefore, the structure of the layered inorganic platelet consists of a stacking sequence of cationic brucite-like layers containing M2+ and M3+ separated by water molecules and the anion that balances the positive charge (Figure 3.1).

2+ 3+ The cationic and anionic layers, in [Mg2Al(OH)6]CO3.nH2O LDH, contain Mg , Al and

2- CO3 , H2O, respectively. The metal hydroxides link within octahedral or tetrahedral positions

248,249 and form sheets stacked in a hexagonal space group (P63/mcm).

A PE consisting of repeating units containing ionic groups, was used as a soft matrix to glue the platelets together. Poly(sodium 4-styrenesulfonate) (PSS, Mw 70,000 30 wt.% in

H2O, Sigma-Aldrich) and poly(diallyldimethylammonium chloride) (PDDA, Aldrich,

Mw 100,000-200,000, 20 wt.% in H2O) in aqueous solution were selected as polyanion and polycation, respectively.

Figure 3.1: Structure of LDH platelet. Schematic representation of the cationic brucite-like layered structure of LDH platelet.

3.1.2 Reinforcing fibres for hierarchical composites with nanostructured interphases

Typical reinforcing fibres with a diameter in the range of 5 to 15 μm were chosen to carry out the proposed research. Glass fibres (AGY 933 S2, 3k) with a diameter of about

9 μm were used to investigate the transfer of a planar nanostructured coating from glass slides

(3” x 1” microscope slides, WVR stores) to curved fibre surfaces. In addition, commonly used high performance fibres, such as sized (HexTow, AS4 GP, 3k) and unsized (HexTow

AS4, 12k) carbon fibres with an average diameter of about 7 μm, were also used to create a nanostructured interphase and subsequently manufacture hierarchical composites. Both fibre types exhibit a diameter in the range highlighted in Chapter 1 (Figure 1.4). While carbon fibres were directly used either unsized or sized, sized glass fibres were used both as-received or were chemically desized. In addition, surface modified unsized carbon fibres and desized glass fibres were also used. The methods used to chemically desize the glass fibres and modify the surface of both carbon and glass fibres are described in detail later.

3.1.3 Polymer resins for the manufacture of hierarchical model and bundle composites with nanostructured interphases

A two-part transparent epoxy resin (Loctite, Double Bubble, IDH-1303596) with rapid curing at room temperature (less than 5 min) was used to investigate the mechanical properties of single fibre composite models. Rapid curing at room temperature was selected to avoid dehydration of the coating but also enables quick manufacturing test specimens. On the other hand, ultra-low viscosity (1010-1070 cP at 20°C) epoxy resin (PRIME 20ULV,

Gurit) with a curing temperature of 50°C was used to impregnate bundles of fibres in order to manufacture cylindrical bundle composites with fibre volume fraction of about 50 %. An ultra-low viscosity resin was chosen to ensure full resin impregnation of the fibre bundles simply by direct dripping of a mixture containing the epoxy resin and a hardener. About 2 ml of the mixture was dripped over each 5 cm long bundle. Owing to the low viscosity character of the epoxy resin, the excess of epoxy drained easily down to the bottom of the bundles by gravity.

3.2 Methods to develop, manufacture and test hierarchical composites with nanostructured interphases

3.2.1 Synthesis of Mg2-Al-CO3 LDH platelets

Mg2-Al-CO3 LDH platelets with varying dimensions and size distributions were synthesised via coprecipitation followed by hydrothermal treatment246 in the laboratory. A

10 ml metal salt solution containing 2 mM of Mg(NO3)2.6H2O (99%, Sigma-Aldrich) and

1 mM Al(NO3)3.9H2O (98%, Sigma-Aldrich) was added to a 40 ml basic solution containing

6 mM NaOH (VWR) and 0.6 mM Na2CO3 (VWR), in less than 5 s, under vigorous stirring

(750 rpm), followed by further stirring (750 rpm) at room temperature for 20 min. The mixture was then centrifuged at 15,000 rpm for 15 min to retrieve the LDH slurry.

Subsequently, the slurry was washed twice by re-dispersion in deionised water followed by bath sonication (75 W) for 5 min and finally centrifugation at 15,000 rpm for 15 min. After washing, the slurry was dispersed in 25 ml deionized water (0.4 wt.%) by bath sonication and placed in an autoclave for hydrothermal treatment at 100°C for 4 h or 72 h and at 125°C for

72 h to produce platelets with different widths and polydispersities. The LDH solution was used within the first month after the synthesis to avoid possible re-aggregation; the quality of the dispersion appeared stable over this timeframe by simply controlling the transparency of the solution by eye.

3.2.2 Layer-by-Layer assembly procedure to assemble nanostructured coatings

3.2.2.1 Layer-by-Layer deposition of nanostructured coatings on flat glass substrates

LbL assembly is based on the alternating deposition of oppositely charged particles and/or charged polymeric chains through electrostatic attractions. LbL was used to deposit and optimise nanostructured “brick-and-mortar” structures on quartz (75 mm x 25 mm, UQG

Optics) and glass slides (3” x 1” microscope slides, WVR stores). LDH platelet suspensions and PSS PE solutions were used to alternatively deposit monolayers of LDH and PSS on a slide. Prior to LbL deposition, the flat substrate was cleaned in piranha solution, a 3:1 mixture of sulphuric acid (95%, VWR) and hydrogen peroxide (50 wt.% in water, VWR), to eliminate

impurities and organic particles present on its surface.250 Quartz and glass slides were immersed in the piranha solution for about an hour at about 100°C and subsequently rinsed in deionised water multiple times. After treatment, the substrate was stored in a sealed jar filled with deionised water for a maximum of two weeks. After hydrothermal treatment, the LDH dispersion in water (25 ml) was further diluted with deionised water (20 ml) to obtain a LDH dispersion with a concentration of about 0.3 wt.% with a pH of 10. PSS (3.35 ml) was added to deionised water (1 L) to form an aqueous PE solution with a concentration of 0.1 wt.%.

The pH of the PSS solution was then adjusted to 10 by the addition of 0.1 M NaOH. In order to form a monolayer of aligned LDH platelets, the negatively charged substrate (glass or quartz) was dipped into a dispersion containing 0.3 wt.% positively charged LDH at pH 10

(as synthesized) for 10 min. The slide was subsequently rinsed by immersion in water at adjusted pH 10 (by addition of NaOH 0.1 M) for 2 min after LDH deposition. The rinsing step was carried out to remove excess particles loosely attached to the substrate as well as particles weakly associated to the surface/meniscus after each deposition step. To form

(LDH/PSS) bilayers and (LDH/PSS)n multilayers, the charged glass substrate was alternately dipped into the LDH (0.3 wt.%) dispersion and PSS solution (0.1 wt.%) for 10 min each, interspersed by 2 min rinsing steps in deionised water at an adjusted pH 10 in between each deposition step (Figure 3.2.A). The pH was kept constant at 10 throughout the entire process by addition of NaOH 0.1 M. To deposit thick (LDH/PSS)n multilayer coatings (n > 15), a home-made automatic dipping robot was used (Figure 3.2.B). The dipping robot was controlled using a Python script. The procedure was exactly the same as described above. The substrate dipping and removing rates were set to about 0.4 cm.s-1. After deposition of the last layer, the coating was rinsed and allowed to dry at room temperature overnight before characterisation.

Figure 3.2: Layer-by-Layer assembly on a flat slide. Schematic representation of the LbL procedure (A) and the dipping robot used to deposit (LDH/PSS)n multilayers onto a flat substrate (B).

3.2.2.2 Deposition of nanostructured coatings on bundles of glass or carbon fibres

Similarly to the deposition of nanostructured coating on a flat substrate, the LbL assembly was also adapted to the deposition onto the surface of glass and carbon fibres.

Because a charged surface is required to initiate LbL deposition of the nanostructured onto fibres, piranha treated glass fibres (hydroxylated), as well as as-received and oxidised unsized carbon fibres, with a negatively charged surface at pH 10 were used. The piranha treatment of the glass fibres was carried out similarly to that of glass slide, allowing for a removal of the sizing as well as a hydroxylation251 of the surface creating negatively charged sites on the fibre surfaces at pH 10 (Figure 3.3).

Figure 3.3: Hydroxylation of glass fibre surfaces. Schematic illustration of the removal of organic sizing and hydroxylation of the surface of glass fibres by piranha treatment.

A bundle of fibres (up to few hundred) were kept vertically aligned while attached to a piece of metal mounted onto the head of the robotic dipping robot (Figure 3.4.B). Both a direct deposition of a first positively-charged monolayer of LDH and the deposition of a

PDDA PE precursor layer were investigated prior to assembling (PSS/LDH)n coatings, in order to optimise the adhesion of the coating to the surface of the fibres. Similarly to the LbL deposition on a flat substrate, the pH was kept constant at 10 throughout the entire process.

Various deposition conditions, including moderate magnetic stirring (300 rpm) of all solutions or no stirring, as well as different rinsing conditions, were investigated to successfully transfer the nanostructured coating onto all individual fibres in the fibre bundle.

Coating deposition was carried out with the same home-made automatic dipping robot as for coating deposition on flat slides with dipping and removing rates fixed at about 0.4 cm·s-1

(Figure 3.4.A). After deposition of the last layer, the coated fibres were rinsed and allowed to dry at room temperature overnight before characterisation.

3.2.3 Carbon fibre surface treatment to increase charge density

In order to optimise the interaction between the nanostructured coating and the fibres and, therefore, to promote adhesion of the interphase, the surface charge density of carbon fibres was investigated and adjusted by suitable fibre surface treatment. Unsized carbon fibres were also surface treated in low pressure plasma (Plasma System Pico, 90179, Diener

Electronic, Germany) under a flow of oxygen (50 sccm) with varying exposure times (30 s to

20 min) (Figure 3.5). Plasma oxidation of carbon fibres252-256 leads to formation of various solid carbon oxides, such as carboxyl and hydroxyl groups on the surface of the fibres as well as an increase in surface roughness.

Figure 3.5: Oxygen low pressure plasma treatment of carbon fibre surfaces. Photograph of the setup of oxidation of carbon fibres via oxygen low pressure plasma treatment.

Further modification of the surface of carbon fibres after plasma treatment was carried out by dipping the treated carbon fibres into a 0.1M KMnO4 solution overnight, followed by rinsing in water, in order to convert the hydroxyl groups present on the surface of the modified carbon fibres into carboxyl groups, which are more likely to deprotonate

(Figure 3.6). The deprotonation of the oxygen-containing groups depends strongly on the pH of the solution in which the fibres are dipped. Negative charges are expected to arise (through deprotonation) at high pH, as can be determined by streaming potential measurements.255,257

Figure 3.6: Oxidation and deprotonation of carbon fibre surfaces. Schematic illustration of surface oxidation of carbon fibres via oxygen low pressure plasma followed by further oxidation in

KMnO4 and associated negatively charged surface at pH 10.

3.2.4 Fibre bundle composite preparation

Small bundle composites containing about 100 to 300 carbon fibres or 400 glass fibres, made of commercially sized, unsized/desized/modified control fibres and fibres coated with a nanostructured interphase, were manufactured and tested in tension in order to investigate the potential improvement in the strain to failure and toughness of the composites.

In order to produce 25 mm-long unidirectional cylindrical fibre bundle composite specimens, few hundreds of fibres were held together vertically in a paper cup held by a metal clip at the bottom and subsequently impregnated with an ultra-high viscosity epoxy resin (Prime, ULV20). An epoxy/hardener mixture, with a mixing weight ratio of 100:19, was rigorously mixed and degased for 30 min in a vacuum prior impregnation of fibre bundles. The epoxy/hardener mixture (about 2 ml) was dripped onto each vertically-placed

5 cm-long bundle contained in the paper cup in order to ensure full impregnation. The excess

resin self-drained under gravity. The curing of the bundles was performed at room temperature for 24 h followed by 16 h in an oven at 50°C. After curing the bundles were transferred onto a paper template (Figure 3.7). The top surface of the paper template end-tab was covered with another piece of paper with epoxy in between (Araldite 2011, two-component epoxy resin adhesive), to glue the bundle to the paper template.

Figure 3.7: Schematic of a bundle composite mounted on a paper template for tensile testing. The bundle is placed onto double-sided sticky tape (1) and then glued to the template with an epoxy resin (2). One end of the template end-tab is longer than the other in order to place an acoustic sensor in direct contact with the template (3)

3.3 Characterisation of nanostructured coating/interphase, single fibre composite models and fibre bundle composites

Various characterisation methods were used to successfully develop a nanostructure inspired by the “brick-and-mortar” structure of nacre, which was subsequently transferred onto the surface of fibres. These fibres coated with the nanostructure coating were used to produce single fibre composite models, which were characterised. Eventually small, impregnated fibre bundle composites, were manufactured and tested.

3.3.1 Determination of surface charge density of LDH platelets by ζ -potential measurements

A high surface charge density of nanoparticles allows for stable colloidal suspensions through electrostatic repulsion forces, which can be controlled by solution composition. The stability of a colloidal suspension can be related to the ζ-potential, which is the potential at the plane of shear between a surface and the surrounding electrolyte solution, measured as a function of the pH258 of the suspension. As a rule of thumb, if the ζ-potential exceeds a value of about |30 mV|, particles are able to repel one another due to electrostatic forces and, therefore, form a stable suspension. When placed in an electrolyte, an electrochemical double layer forms around the particles, which can be divided in two regions; the first region close to the surface is called the “Stern” layer, ions contained in this layer are strongly bound to the particle surface. Further away from the surface into the electrolyte, in the “diffuse” layer, the ions are more weakly attracted and are able to move. When a movement is induced between the particles and the electrolyte, for instance by applying an electric field, the diffuse part is sheared off and a potential at the boundary at which the separation between the two layers can be measured which is called the ζ-potential (Figure 3.8). Electrophoretic mobility measurements (Zeta PALS, Brookhaven) were performed to determine the ζ-potential of

Mg2-Al-CO3 LDH platelets in suspension in a 5 mM KCl aqueous solution as a function of pH (from 4 to 10). The electrophoretic mobility μe measured by electrophoretic light scattering was converted into ζ-potential, using the Henry equation:259

2 εrs.ε0 μ = ∙ ∙ f(ka) ∙ ζ (3-1) e 3 η

where εrs and ε0 are the relative permittivity of the electrolyte (80.10) and vacuum

(8.85.10-2 F·m-1), f(ka) the Henry’s function and η the dynamic viscosity of the electrolyte

(8.90.10-4 Pa.s).

Figure 3.8: ζ-potential. Representation of the electrical double-layer at the particle surface and the concept of ζ-potential.

3.3.2 Determination of surface charge density of glass and carbon fibres by streaming ζ -potential measurements

The surface charge density of fibres can be measured using the streaming potential method.255,257 The method is based on forcing an electrolyte solution to flow through a packed fibre bed (Figure 3.9). The electrolyte flows through the fibre bed, giving rise to a

“streaming” current, which generates a potential difference across the packed fibres.

100

Subsequently, the created potential hinders the mechanical transfer of charge leading to back current by ion diffusion. Once the two currents reach equilibrium at an applied pressure, the measured potential is the streaming potential ∆μs and can be converted into the ζ-potential using the Helmholtz-Smoluchowski equation260:

∆μs η L 1 ζ = . . . (3-2) ∆p εrs ε0 A R where Δp is the applied pressure, η the dynamic viscosity of the electrolyte, εrs and ε0 are the relative permittivity of the electrolyte and vacuum, respectively, R is the electrical resistance of the medium and L and A are the dimensions of the tube in which the fibres are packed, namely, the length and cross-section area, respectively. The slope obtained from the dependency of the streaming potential ∆μs and the applied pressure ∆p, in the tube, is proportional to the ζ-potential of the fibres.

Figure 3.9: Streaming ζ-potential. Schematic of the formation of a streaming ζ-potential ∆μs at the fibre/electrolyte interface via the creation of a streaming current (A) and leaking current (B).

101

ζ -potential measurements were carried out using a SurPASS apparatus (Anton Paar,

Austria) over the pH range of 3 to 10. The pH of a 5 mM KCl electrolyte solution was adjusted by automatic titration of 0.05 M KOH and 0.05 M HCl solution, in pH steps of about

0.3. The streaming current was measured using an adjustable cylindrical cell in which approximately 0.5 g of glass or carbon fibres were inserted (Figure 3.10).

Figure 3.10: Streaming ζ-potential measurement. Adjustable cylindrical cell of the SurPASS apparatus filled with hydroxylated glass fibres.

3.3.3 X-Ray diffraction techniques for the determination of LDH platelets and (LDH/PSS)n coatings morphology

Crystallographic structure and thickness of the synthesised LDH platelets were investigated using X-ray diffraction (XRD). The average spacing between crystalline planes can be calculated from a diffracted photon of an incident X-ray beam, using Bragg’s law.261

Indeed, the wavelength of the diffracted photon is directly related to the interplanar distance

(dhkl) associated with a specific angle θ formed between the incident beam and the diffracting

102

planes (Bragg angle). Therefore, each crystallographic plane of LDH nanoplatelets diffracts at a different angle θ.

The thickness of the anisotropic crystalline LDH platelets was determined using the

Scherrer equation262 while analysing the diffraction peak of their (003) crystalline planes, parallel to the platelet surface:

K ∙ λ t = (3-3) 003 β ∙ cosθ where t003 is the mean thickness of the (003) crystalline domain (here, the platelet thickness),

K a dimensionless shape factor equals to 0.89, λ the X-ray beam wavelength (1.5418 Å) , β the full width at half maximum of the diffraction peak studied and θ is the Bragg angle (5.9°).

A XRD equipment (X’Pert PRO, PANalytical) with a Cu-Kα X-ray source (1.5418 Å) was used to characterise LDH powders.

Alignment of the LDH platelets when deposited on a flat substrate such as a glass or quartz slide was investigated by acquiring three-dimensional rocking curves (Figure 3.11).

The θ angle was fixed at 5.9° corresponding to the diffraction of the (003) crystalline plane of the platelets, which is expected to be parallel to the surface of the substrate in case of a well-ordered, deposited LDH monolayer. The tilt angle Ψ was increased from 0 to 80° in 5° steps. For each value of Ψ, the diffraction intensity was recorded during a full revolution around (Φ angle) the sample with a step of 5° to investigate the misalignment of the platelets in all directions. The full width at half maximum of the obtained rocking curves was measured, which corresponds to a misalignment value of the platelet as compared to the flatness of the substrate. Similarly to the 2θ diffractogramms, the rocking curve were obtained on a XRD machine (X’Pert PRO, PANalytical) equipment with a Cu-Kα X-ray source (1.5418 Å).

103

Figure 3.11: Schematic of XRD rocking curve acquisition in (003) diffraction conditions of LDH platelets. LDH platelets deposited onto a flat glass substrate in relation to the Ψ, Φ and θ angles.

3.3.4 UV-Vis spectroscopy to control the deposition of

(LDH/PSS)n multilayer coatings on quartz slides

The energy carried by a photon can be absorbed by a molecule if the molecule possesses an electronic transition equal to the energy of the photon. The absorption of the radiation brings an electron into an excited state from a ground state. The absorption A depends on the wavelength of the radiation and is a function of the coefficient of molecular extinction ε, the molecular concentration C of the molecule and the path length of light d as described by the Beer-Lambert law below:263

A = ε ∙ c ∙ d (3-4)

On the other hand, nanoparticles such as LDH nanoplatelets scatter the light. The scatter depends on the size, shape and aggregation state of the nanoparticles.

104

Therefore, UV-Vis spectra were acquired over a wavelength range of 800 to 200 nm

(Lambda 35, PerkinElmer) for samples containing different numbers of (LDH/PSS) bilayers.

Coated quartz slides were placed in the UV-Vis spectrophotometer with a bare quartz slide used as reference. The absorption at 225 nm caused by the phenyl group of PSS, as well as the level of scatter at 350 nm, were plotted as a function of the number of bilayers in the deposited multilayer coating to confirm repeatable LbL deposition of each bilayer. Reference spectra were acquired on cleaned quartz substrates without any coating. The deconvolution of the absorption peak from the scatter related to PSS from the scatter enabled to individually fit the absorption features as a function of the number of deposited bilayers.

3.3.5 Morphology of monolayers and multilayer coatings

Electron microscopy enables characterisation and observations of materials below the micrometre length scale, providing structural and elemental information depending on the microscopy technique used. An incident electron beam with high energy, in the order of keV, interacts with a volume of material near the impact surface leading the emission of electrons and photons.264 The analysis of the emitted electrons and photons allows for a diverse set of microscopy characterisation methods.

3.3.5.1 Scanning electron microscopy of monolayers and multilayer coatings

The morphology of LDH platelet monolayers, (LDH/PSS)n multilayer coatings deposited on flat slides and PDDA/(PSS/LDH)n deposited on fibre bundles, model

105

hierarchical bundle composites as well as nanoindented nanostructured coating was carried out using scanning electron microscopy (SEM). The technique relies on scans of an aligned and focused electron beam onto conductive samples, generating secondary electrons, backscattered electrons and X-rays.265 The penetration of the electron beam into the sample is in the order of a micrometre, depending on the energy of the beam as well as the properties of the specimen analysed. The analysis of the secondary electron arising from the surface of the specimen after beam impact leads to topographical information with a resolution down to the nanometre length scale. Imaging of the samples was performed on an SEM (LEO Gemini

1525 FEGSEM). Due to the non-conductive nature of the LDH and PSS monolayers, a thin layer of gold (about 10 nm) was sputter coated (Emitech K575X, Peltier cooled sputter coater, Emitech Products Inc.) on top of each sample, prior to imaging. SEM was used to image the top surface and cross-section of the coatings, operating at 5 keV with an aperture of 30 μm.

3.3.5.2 Transmission electron microscopy of LDH

platelets and (LDH/PSS)n coating cross-section

TEM uses a high energy electron beam, on the order of 100 keV, focused on a thin specimen (less than 100 nm) allowing for the analysis of both inelasticcally and elastically scattered electrons transmitted through the samples.264 Morphological information with resolution at the atomic scale can be obtained.

The dimensions and size distribution of the LDH platelets were measured by analysing TEM images (JEOL, 2000FX). LDH platelets were deposited on a TEM grid

(Holey Carbon Films on 200 Mesh Copper Grids, Agar Scientific) by directly placing a drop

106

of a 0.1 wt.% suspension onto the grid followed by drying overnight at room temperature. A high-resolution TEM (2100 FX, JEOL) equipped with an Energy dispersive X-ray (EDX) spectrometer was used to image and characterise the composition of the nanostructured coating cross-section.

The observation of a specimen cross-section within a TEM requires the preparation of a thin slice of material with a thickness below 100 nm. Focused ion beam (FIB) is a good method to prepare such specimen.266 Based on the emission of a high energy gallium ion

(Ga+) beam, similar to an electron beam used in a SEM, the irradiated specimen is sputtered leading to removal of material. The FIB technique is deemed destructive. The more energetic the beam, the more material is removed from the samples. A sample cross-section, less than

100 nm-thick, was prepared by milling, using a gallium ion beam on a dual-beam FIB

(Nanolab 600, Helios) (refer to Appendix.2)

3.3.6 Elemental composition of (LDH/PSS)n coating cross-section by energy dispersive X-ray spectroscopy

EDX spectroscopy enables the analysis of the elemental composition of a sample and is commonly used in electron microscopes such as SEM and TEM.267 The electron beam of the microscope interacts with the surface of a specimen leading to the emission of X-rays through relaxation of electrons within the first nanometres of the sample surface. The emitted

X-rays are directly related to the energy levels of the elements present in the sample. Each element emits characteristic X-rays depending on its possible energy transitions. An EDX spectrometer was used in the TEM (up to 15 keV) to investigate element distribution in the nanostructured coating cross-section.

107

3.3.7 Composition of carbon fibre surfaces determined by X-ray photoelectron spectroscopy

Specimen surface chemistry can be quantitatively characterised using X-ray photoelectron spectroscopy (XPS). The technique is surface-sensitive and provides information of the elemental composition of the specimen surface.268 Irradiation with a focused X-ray beam generates photo-emitted electrons with a kinetic energy that depends on the binding energy to the atoms in the sample surface. Change in the oxidation state of the atoms gives rise to small shifts in the peak position (different binding energy) of the spectrum, called “chemical shifts”, which can be observed and interpreted.

The surface of carbon fibres (rinsed in deionised water and dried overnight under vacuum) before and after modification were analysed using a K-alpha+ XPS spectrometer

(ThermoFisher Scientific) equipped with a MXR3 Al Kα monochromated X-ray source

(hʋ = 1486.6 eV). X-ray gun power was set to 72 W (6 mA and 12 kV). Charge compensation was achieved with the FG03 flood gun using a combination of low energy electrons and ion flood source. Argon etching of the samples was done using the standard EX06 Argon ion source using 500 V accelerating voltage and 1 µA ion gun current. Survey scans were acquired using 200 eV pass energy, 1 eV step size and 100 ms dwell times. All high resolution spectra (C1s and O1s) were acquired using 20 eV pass energy, 0.1 eV step size and

1 s dwell times. Samples were prepared by pressing the fibres onto double side sticky copper based tape. Pressure during the acquisition of the XP spectra was (< 10-8 mbar). Casa XPS software (version 2.3.16) was used to process the data. A combination of Gaussian (70%) and

Lorentzian (30%) was used. All XPS spectra were charge corrected by referencing the fitted contribution of C-C graphite like carbon in the C1s signal 284.5 eV.

108

3.3.8 LDH monolayer topography determined by atomic force microscopy

AFM is a scanning probe microscopy providing topographic information and mechanical information from phase shift of a sample at the nanometre level.269 A tip is used to scan the surface of a sample in contact, tapping, and non-contact mode. Tapping mode is a dynamic mode based on the oscillation of a cantilever near its resonance frequency in the vicinity of the sample surface. The tip movements (height change) are recorded by the mean of a laser focusing onto the tip, providing surface morphology. The morphology of a first monolayer of LDH platelets deposited onto a glass slide was investigated by coating height profile measurement using AFM (Multimode VIII, Bruker) in tapping mode with a standard non-coated tapping mode probes (Windsor Scientific, UK). The micrographs were processed in Nanoscope Analysis (v1.40 Bruker).

3.3.9 Determination of organic content of (LDH/PSS)n coatings by thermogravimetric analysis

Thermogravimetric analysis (TGA) is commonly used to identify the composition of a sample through mass loss measurement during heating in a controlled gas atmosphere.270 The percentage of initial sample mass is monitored as a function of temperature. The amount of

PSS PE in the coatings was determined using TGA (Q500, TA Instruments), by combustion, leaving the LDH inorganic residues. TGA was performed from 30 °C to 900 °C at a rate of

5 °C/min in air (60 ml/min) for all coatings, stored in laboratory conditions for a week after deposition. The weighting precision of the equipment was about ± 0.01%, with a resolution of

0.1 μg. Sample weights of 1.5 to 2.5 mg were used. In addition to the (LDH/PSS)n/LDH

109

coatings, individual constituents of the multilayer such as PSS and LDH powders by direct casting of PSS solution and LDH suspension on a glass slide followed by water evaporation.

The PSS and LDH powders were analysed using the TGA same conditions as for coatings, i.e. after storage in laboratory conditions for a week, without any dehydration step. The weight proportion of organic PSS was determined using the rule of mixture, assuming that the contributions to the hybrid are simply additive.

3.3.10 Determination of the mechanical properties of

(LDH/PSS)n coatings by nanoindentation

Nanoindentation is a method widely used in the field of thin coatings for the measurement of hardness (H) and elastic modulus (E) along with quantification of viscoelasticity and plasticity.271,272 Adapted from the Vicker’s test273 of bulk materials, the method enables near-surface characterisation of thin coatings over a small volume. During a test, a flat coating is indented by a tip, with defined shape. The test can be divided in three differents steps, namely, loading, creep at constant maximum load and unloading of the specimen while the system is recording the applied load as function of the depth displacement

(Figure 3.12). The profile and shape of the tip is used to determine both hardness and elastic modulus271. The contact area of the tip with the specimen is described by the area function

A(hc), correlating the contact area as a function of the contact depth hc. The area function is pre-calibrated for the particular tip used, by indentation of a fused silica specimen with known properties.

110

Figure 3.12: Coating nanoindentation. Schematic of thin coating nanoindendation procedure.

The hardness of the tested specimen is directly measured from the load P applied divided by the area function A at the associated displacement depth:

P H = (3-5) A

Furthermore, the load-displacement curve obtained during nanoindentation can be used to determine the modulus of the tested specimen (Figure 3.13). The early part of the unloading segment of the curve can be linearly fitted and associated to the elastic modulus E using the equation:

1 √π S E = . . (3-6) β 2 A(hc) where β is a dimensionless parameter and S the slope of the unloading segment of the load-displacement curve. The contact depth hc is measured as the permanent indent left after indentation through the relation:

Pmax h = h − c max S (3-7)

Where Pmax and hmax are the maximum load applied to the coating and the depth at maximum load.

111

Figure 3.13: Nanoindentation parameters. Schematic illustration of an indent and the associated load-displacement nanoindentation curve. Copied from Oliver and Pharr.271,272

The mechanical properties of the (LDH/PSS)n/LDH coatings were investigated using nanoindentation (Nanotest NTX, Micro Materials) equipped with a Berkovich tip with a pyramidal shape. The coatings were indented 36 times with a spacing of 100 μm between the indents along both the X and Y axis at room temperature. Indent depth was limited to the first

15 % of the coating thickness to avoid any substrate effect while applying a load of 500 μN within 30 s. Viscoelasticity of the coatings caused by the deformation of the polymer phase was investigated by holding the indentation maximum load for about 30 s allowing the coating to creep. Unloading of the coatings was carried out within 30 s. Plastic deformation of (LDH-2/PSS)n/LDH was investigated by nanoindentation while loading a 3.5 μm-thick coating with a cube-corner tip at varying loads (0.25 to 5 mN). The index of plasticity152 ξ was obtained from the equation below:

A1 ξ = (3-8) A1 + A2 where A1 and A2 are the areas of the plastic deformation and viscoelastic recovery under the load-displacement curve, respectively (Figure 3.14).

112

Figure 3.14: Nanoindenation load-displacement curve. Representation of the deformation areas of a nanoindentation load-displacement curve.

Nanoindentation in-situ SEM (Auriga, Carl Zeiss) was also carried out to visualise the deformation mechanism occurring in a 1.5 μm-thick flat coating upon mechanical loading between 150 to 500 mN using a cube-corner tip (Figure 3.15.A). The use of such indenter stage allows to record of the nanoindentation and, therefore, for the observation of the formation of pile-ups and crack propagation in the vicinity of the indent. In addition, the same technique was used to measure the mechanical properties of nanostructured coatings deposited onto fibres (Figure 3.15.B). The use of SEM visualisation of the sample during testing enables for an accurate centring of the tip above a selected fibre. In order to increase the resolution of the experiment, while being limited (to measure of E and H) to a maximum indentation depth of about 150 nm (15% of the coating thickness), a Berkovich tip was mounted onto the indenter. Load-control mode of the indenter allowed for a constant loading of the coated fibre within 30 s. Similarly to nanoindention on a flat coating, the maximum load was held for 30 s and the unloading of the specimen occurred within 30 s, in order to assess whether the transfer of the coating from a planar substrate to the fibre surfaces was successful. A maximum load from 0.3 to 1.0 mN was applied to the coated fibre. Indentation

113

depth of up to 250 nm was reached to investigate the plasticity of the coating deposited on fibres. In order to avoid compliance or motion during indentation, sliver paint was applied to a bundle of coated fibres to keep them together, partly embedded. An area clear of silver paint was selected to carry out the tests.

Figure 3.15: SEM in-situ nanoindentation of nanostructured coating. Nanoindentation of a nanostructured coating deposited on a flat glass slide (A) and deposited on a glass fibre (B).

3.3.11 Nanostructured interphase mechanical characterisation techniques

3.3.11.1 Determination of the interfacial properties and debonding behaviour of the nanostructured interphase by single fibre pull-out tests

The single fibre pull-out (SFPO) test is based on the extraction of a fibre partly embedded in a matrix by applying a force parallel to the fibre.23 A single fibre, taped onto a washer, is dipped vertically into a liquid polymer resin contained into the cavity of a screw, using fine vertical adjustment screw (Figure 3.16.A). Thermoplastic and thermoset resins can be melted and cured, respectively, in the cavity of the screw, using a heating unit. Once partially embedded in a solidified matrix, the single fibre is transfer to an home-made

114

pull-out setup design and produced by Gerhard Kalinja, Ruediger Sernow and Martina

Bistritz at Numdesanstalt für Materialforschung und –prüfung (BAM) (Figure 3.16.B). The free-end of the fibre (about 0.5 to 1 cm long) is glued to a needle using an epoxy adhesive, connected to a piezo-translator, while the screw containing the partially embedded fibre is screwed into the piezo-force meter unit of the apparatus (Figure 3.16.B). A force is applied to the single fibre/matrix interface and measured with a resolution as low as 0.1 mN, at a constant displacement rate of 1 μm.s-1. The free distance between the matrix surface and the end of the needle is about few tens of micrometres. During the test, the applied load F is recorded as a function of the displacement S, while the pull-out of the fibre can be followed in-situ using an optical microscope (Stereo, SMZ-140 Series, Motic).

115

The pull-out test can be divided in three consecutive steps23,274 (Figure 3.16.C): First, the load applied to the fibre increases linearly due to elastic deformation of the fibre/matrix interface. Secondly, the interface reaches its maximum load carrying capability leading to either a sudden full debonding of the fibre or initiation of the debonding associated with plastic deformation. Subsequently, the load either decreases suddenly (brittle failure)

(Figure 3.16.C) or progressively levels off (ductile failure) (Figure 3.16.D) before load drop caused by full debonding. Finally, the last step consists of the pulling-out of the debonded fibre from the matrix, which is accompanied with fibre pull-out friction against the matrix, occurring until complete removal of the fibre from the matrix. Therefore, the maximum load required to fully debond of the fibre Fmax is related to the apparent IFSS of the interface as it can easily be converted into a strength:274,275

Fmax Fmax τ = = (3-9) Ae πdfle where Ae is the embedded area of the fibre in the matrix, and df and le are the diameter and the embedded length of the fibre, respectively. In order to obtain an accurate measurement of the apparent IFSS, several tests were carried out and Fmax was plotted as a function of Ae.

Consequently, the data were fitted to a linear curve, whose slope is directly related to IFSS of the fibre/matrix interface.

Fibres coated with nanostructured coating are expected to exhibit progressive and ductile interfacial failure (Figure 3.17) caused by platelet sliding/interlocking in shear (strain hardening character of the nanostructure) with long interfacial sliding prior to full debonding.

Furthermore, the ability of the fibre to slide before full debonding was estimated by the debonding length ratio (DLR):

ld DLR = (3-10) le

116

where ld and le are the fibre debonding length and the embedded length, respectively. ld was measured as the length along, which the fibre can slide prior to full debonding, which includes both elastic and plastic displacement of the interphase.

Figure 3.17: Mechanism of single fibre pull-out with nanostructured interphase. Schematic representation of the expected sliding behaviour of a nanostructured coated fibre, partly embedded in epoxy during a pull-out test.

3.3.11.2 Determination of the interfacial properties and stress concentration of the nanostructured interphase by single fibre fragmentation tests

The single fibre fragmentation (SFF) test was developed from the early work of Kelly and Tyson,21 who studied the breaking of tungsten fibres into multiple segments in a copper matrix. The SFF test can be applied to a lot of different types of fibres, embedded in various matrices, to assess the interfacial properties of the embedded fibres.276 The fragmentation of an axially aligned single fibre in tension can be observed in-situ using an optical microscope.

The stronger the interface between the fibre and the matrix, the shorter the critical fragment

117

length reached at full saturation as, once fractured, the fibre still carries load through interfacial interaction with the matrix. Indeed, the tensile stresses are transferred to the fibre fragments through the IFSS (τ) of the matrix/fibre interface. The fibre continues to fragment into shorter fragments until the stresses transferred to the fibre fragments are lower than the tensile strength of the fibre at the critical fragment length. The saturation stage, at which the fibre stops fragmenting, is then associated to the critical fragment length lc, such that the stress does not allow for further fibre breakage to occur (Figure 3.18).

Figure 3.18: Single fibre fragmentation test. Schematic illustration of single fibre fragmentation in a matrix with corresponding level of stress. Copied from Tripathi et al.276

The critical fragment length lc at saturation is directly related to the fibre strength σf, diameter df and IFSS τ by the Kelly-Tyson relation:

σf ∙ df τ = (3-11) 2lc 4 lc = l (3-12) 3

118

where l is the mean fibre fragment length at saturation.277

Figure 3.19: Stress concentration at fibre break. Optical micrographs of a fibre fragment and subsequent fibre slippage in matrix in a standard (top) and cross-polarised light (bottom). Copied from Feih et al.22

Furthermore, stress accumulation at the fibre/matrix interface can be visualised using a birefrengent matrix and cross-polarised light (Figure 3.19). SFF tests specimens were prepared by casting an epoxy (Loctite, double bubble two-part epoxy, IDH-1033596) film onto a large glass slide on which 5 aligned fibres were taped, about 200 to 300 μm above the surface of the slide on a double-sided sticky tape. The epoxy casting solution was prepared by mixing. A 30 wt.% epoxy-part solution (2.5 ml) and a 30 wt.% hardener-part solution

(2.5 ml) in acetone was casted twice on the slide in order to obtain a 500 μm-thick film after evaporation of the acetone overnight at room temperature in a fume hood. After evaporation of the acetone, the film was gently peeled of the substrate using a scalpel and tweezers and subsequently cut into a dog-bone shape while centring the fibre in the gauge length, using a punch press (ZCP020, Zwick testing machines Ltd, Hereforshire, UK) equipped with dog-bone die (type 5B, ISO 527-2). The specimens were kept at room temperature for at least two weeks prior to testing, to ensure epoxy curing and solvent evaporation. The specimens were 40 mm long and 7 mm wide with a gauge dimension of 15 mm by 2.5 mm region with a progressive increase in the specimen width between end-tab and gauge length from 2.5 to

7 mm along a length of 2.5 mm. 5 specimens per type of fibre were manufactured to obtain at

119

least 100 fibre fragments in order to measure the mean fragment length. The specimen was mounted on a small tensile tester (Linkam Scientific Instruments, TST350) equipped with a

200 N load cell. Initiation of fibre fragmentation and up to full saturation of the fibre fragmentation at a strain of about 25 %, at a cross-head speed of 15 μm·s-1, was tracked using camera mounted on an optical microscope (Leica DM2500) equipped with a DFC295 camera

(Leica Application Suite v4.0.0, Leica ∞/1.1 HI PLAN 40x/0.50). Stress concentration transferred to the birenfrengent epoxy resin in the vicinity of a fibre fragment as function of strain was observed in-situ under the microscope using cross polarised light in transmission to investigate the ability of the nanostructured interphase to dissipate and spread the stress arising from a fibre break.

3.3.12 Tensile properties of fibre bundle composites

Tensile tests of the manufactured bundle composites were carried out using an Instron machine equipped with a 1 kN load cell. A displacement rate of the grips of 0.5 mm·min-1 was applied during the test until rupture of the specimen. The strain was measured using painted marks tracked with a video gauge (iMETRUM MG223B PoE E0022522 from iMETRUM Ltd) equipped with a macro lens (iMETRUM material lens 233093) with a magnification 0.193 and focal length 309 mm, as shown in Figure 3.20.

120

Figure 3.20: Strain measurement of bundle composite tested in tension via video gauge. Images of paint-marked bundle composite specimen mounted in tensile grip (A) and video gauge equipped with a macro lens (B).

In addition, acoustic emissions (AE) were recorded as a way to track fracture events such as fibre breaks and composite fractures occurring in a composite specimen during tensile testing. Due to the relatively small size of the bundles, a PICO miniature acoustic sensor (20 kHz to 500 kHz) (Mistras Group Inc., Physical Acoustics Corporation, USA) was used and placed in direct contact with the paper template in the end-tab area using a coupling agent gel (Sonagel W, Sonatest, UK). The AE sensor was used in conjunction with 1283 USB

AE interface and AEwin software. A specific bottom grip (Figure 3.21), made of aluminium, was designed and manufactured to maintain the sensor in contact with the specimen (refer to

Appendix.3). The bottom grip allowed the AE sensor to be mounted and maintained in place next to the sample while being suited for a connection to a 1 kN load cell with an ambient background noise of about 65 dB. Therefore, all AE events with amplitude below 65 dB were discarded. The acoustic emission parameters were set to 100 ms, 200 ms, 200 ms and 10 ms for peak definition time, hit definition time, hit lockout time and time drive rate, respectively.

5 kN spring-loaded wedge grips (Instron) were used as top grip. A High speed camera

(Phantom v12.1) with a focal length of 100 mm (Zeiss Makro-Planar T* ZE lens) was used to visualise the fracture behaviour for the composites.

121

Figure 3.21: Tensile grip. Tensile grip mounted with an AE sensor to test small bundle composites (refer to Appendix.3 for dimensions).

122

Development of Planar “Brick-and-

Mortar” Nanostructured Coatings

A new anisotropic nanostructured interphase replacing commercial sizing for fibre-reinforced composite materials, which would enable for additional toughening throughout the whole material, was designed, developed and characterised. The structure of natural nacre109 is of great interest as it allows for crack deflection120 at the platelet interface along with platelet interlocking,16 providing toughening and strain hardening in shear, two fibre/matrix interface characteristics that may improve the properties of unidirectional FRP composites in tension. However, the structure of nacre cannot be conformally deposited on the surface of typical reinforcing fibres with diameters of few micrometres. Therefore, an anisotropic layered nanostructure similar to that found in natural nacre, made of inorganic platelet with dimensions fitting the curvature of a typical reinforcing fibre (refer to Chapter.1) was designed to reproduce the deformation mechanisms of nacre. The classic “brick-and- mortar” structure of nacre was scaled down by more than an order of magnitude while retaining the characteristic geometry, aspect ratios and relative proportions. Different embodiments, resulting from the assembly of platelet reinforcements with varying dimensions, were studied to shed light on the fundamental structural and scaling behaviour of

123

nacre nanomimics. Planar nacre nanomimics were first investigated to facilitate the characterisation of their structures before transferring the most promising embodiment onto the surface of reinforcing fibres (refer to Chapter.5). A LbL assembly method was selected to deposit nacre-nanomimetic coatings onto a flat charged substrate because it offers thickness control at the nanometre length scale193 and the potential to coat multiple charged fibres in a parallel manner.

4.1 Introduction

The main objectives were to develop a nanostructured coating with a strain hardening response in shear along with the ability to deflect a crack while allowing for substantial energy absorption. An LbL assembly was used to alternatively deposit monolayers of

Mg2-Al-CO3 LDH platelets of varying dimensions and size distributions alternating with monolayers of PSS PE as an organic soft matrix. The use of well-defined discrete LDH nanoplatelets, with an aspect ratio similar to the aragonite platelets in nacre, offers an alternative to nacre nanomimics with high inorganic content. An intermediate absolute platelet thickness in the range 10-20 nm, can be combined with a simple synthetic polymer

‘mortar’ around ten times thinner than the natural biopolymer, such that the correct dimensional ratios and phase proportions can be retained.

4.2 LDH platelet reinforcements

LDH platelets were selected as the nano-reinforcement because of their anisotropy and tunable dimensions via modification of their synthesis conditions, as well as for their

124

positively charged surface, which enables their assembly onto a charged substrate using LbL assembly. In addition, LDH platelets were chosen for their hexagonal geometry, which is similar to aragonite platelets in natural nacre, potentially leading to a high degree of packing.

[Mg2Al(OH)6]CO3.nH2O LDH platelets were synthesized by co-precipitation, followed by hydrothermal treatment246 at various conditions to produce platelet with varying dimensions and polydispersities. The hydrothermal treatment converts the amorphous LDH slurry, obtained after co-precipitation of the metal salt with base, into individual crystalline particles.

The different structural and surface properties of the synthesised platelets are discussed in the following sections.

4.2.1 Width distribution of LDH platelets

Platelets with different dimensions and size distributions were synthesised via various hydrothermal treatments (different temperature and time) in order to investigate the relation between the structure of the nacre-nanomimetic coatings and their mechanical properties.

Three different hydrothermal conditions were used to synthesize LDH platelets with various dimensions and polydispersity, while maintaining the same hexagonal geometry, aspect ratio, composition and surface charge density. An increase in the hydrothermal treatment time was shown to lead to larger platelets, while an increase in temperature led to broader polydispersity. According to TEM images (Figure 4.1.A-G), the products of the hydrothermal treatment formed stable dispersions of hexagonal platelets in aqueous solution, regardless of synthesis conditions, as required for successful monolayer deposition by LbL assembly. The average diameter and diameter distributions of the synthesized LDH platelets were determined from TEM images (Figure 4.1.H). Hydrothermal treatment at 100 °C for 4 h and

125

72 h, produced narrow particle size distributions, with average width of ~50 nm (LDH-1) and

~130 nm (LDH-2), respectively. After 72 h at 125 °C, the mean diameter remained around

~130 nm but the polydispersity increased significantly (LDH-3).

Following the co-precipitation of the base with the metal LDH precursor solutions, hydrothermal treatments longer than 72 h at temperature exceeding 125 °C led to re-aggregation of the platelets and produced micrometre-wide LDH aggregates in suspension.

126

4.2.2 LDH platelet structure and anisotropy

246,278 XRD confirmed the formation of phase-pure Mg2-Al-CO3 LDH without any trace of crystalline contaminant (Figure 4.2.A). All the platelet samples exhibit the same diffraction peaks with a d003 basal d-spacing of about 7.6 Å (in agreement with TEM observations of Figure 4.2.B). The diffraction peak at 2θ = 11.7° is characteristic of the (003) crystallographic plane parallel to the surface of the anisotropic platelets (Figure 4.2.B).

Sharper (003) diffraction peaks were measured for platelets treated for longer times and at higher temperatures, indicating a greater platelet thickness as a result of platelet growth

(Figure 4.2.C). The thickness of the LDH-1, LDH-2 and LDH-3 platelets was determined using the Scherrer broadening of the (003) XRD peak, obtained from LDH powders. The mean thickness of LDH-1, LDH-2 and LDH-3 platelets was measured to be 8.6, 13.6 and

15.8 nm, respectively.

127

A (003) LDH-1 B (003) 2 nm (006) (009) (012)(015)(018) (110)(113)

LDH-2

C (003) LDH-1

0.41° Intensity FWHM 11.68° LDH-3 LDH-2

0.28°

FWHM

11.63° Intensity LDH-3 0.25° FWHM 0 10 20 30 40 50 60 70 11.68°  11.0 11.5 12.0 12.5 2 (°) 2(°)

4.2.3 LDH platelet surface charge

In addition to their tunable dimensions, LDH platelets were also selected for their charge density. The platelets have a brucite-like cationic surface, which was characterized by a positive ζ-potential246 over the whole pH range, and found to be higher than +30 mV in

5 mM KCl (Figure 4.3.A).

128

A B

LDH 50

-potential (mV)  30 i ii 3 4 5 6 7 8 9 10

pH @ [KCl] = 5 mM

The high charge density of the LDH platelet surface enables for good suspension stability through inter-platelet electrostatic repulsion (Figure 4.3.B). Charged LDH platelets in water (diluted to 0.3 wt%, at pH 10) formed stable colloids with agglomeration visible to the naked eye only after several months of storage at room temperature. Individual and charged LDH platelets in suspension are particularly suitable for the use in LbL assembly of a nanostructure. A summary of all the properties of the different LDH platelets used to assemble nanostructures are presented in Table 4.1.

Table 4.1: Summary of Mg2-Al-CO3 LDH platelets properties.

ζ-pot. Synthesised Temp. Time Conc. Aspect Thickness Width pH 10 platelets / °C / h / wt.% ratio / nm / nm / mV LDH-1 100 4 ~ 6 ~ 8.6 49 ± 17 LDH-2 100 72 0.4 > +30 ~ 10 ~ 13.6 131 ± 44 LDH-3 125 72 ~ 8 ~ 15.8 130 ± 117 (Treatment conditions)

129

4.3 Layer-by-Layer assembly of nanostructured coatings

The quality of LDH platelet and PSS monolayer deposition was first investigated in order to establish a successful LbL dipping procedure to eventually assemble (LDH/PSS)n multilayer coatings. After which, the deposition of a multilayer coating was monitored by

UV-Vis spectroscopy and electron microscopy to confirm the homogeneity of the coating thickness.

4.3.1 LDH and PSS monolayer deposition on flat substrates

A single LDH-1 platelet layer was deposited onto a glass slide by direct dipping into a

0.3 wt.% LDH-1 suspension for 10 min. SEM shows that the glass was fully coated with a significant amount of excess platelets, loosely attached to the substrate (Figure 4.4.A-B).

Therefore, the slide was dipped into a water solution at pH 10, for 2 min, to remove most of the excess platelets, although few overlaps were still be observed (Figure 4.4.C-D). By dividing the rinsing step into 4 subsequent 30 s dippings of the glass slide, further removal of the excess of platelets was achieved, leading to a near perfect monolayer of LDH-1 with full coverage (Figure 4.4.E-F). LDH-1 platelets were found to lay-down flat on the substrate, leading to a high degree of packing, optimising their electrostatic interactions with the substrate. At the nanometre length scale, the platelets must easily attach-detach in order to optimise their interaction with the substrate, leading to greater LbL versatility. Subsequently, a PSS monolayer was deposited onto the LDH-1 monolayer coated glass slide following the same dipping procedure, namely, a 10 min dip into the PSS solution, followed by a 2 min rising step (divided into 4 dips), in an aqueous NaOH pH 10 solution. Full coverage of the

130

LDH monolayer with PSS PE was observed (Figure 4.4.G-H). Low molecular weight

(70, 000) PSS PE solution was used so as to minimise the organic layer thickness and reach a inorganic/organic thickness ratio of about 9 (refer to Chapter.1).

131

A B

Excess of LDH Excess 1 μm 250 nm

C D

LDH overlaps 1 μm 250 nm

E F

100 nm LDH monolayer 250 nm

G H ) (LDH/PSS 1 μm 250 nm

Figure 4.4: Deposition of LDH and PSS monolayer by LbL deposition. SEM images of a LDH-1 layer containing a significant excess of platelets (A-B), a LDH-1 monolayer with few overlaps after cleaning (C-D), a quasi-perfect LDH-1 monolayer after further cleaning (E-F) and a (LDH-1/PSS) bilayer (G-H) deposited on a glass slide using LbL deposition.

132

AFM height profile investigation of the deposited LDH-1 monolayer on a glass slide

(Figure 4.5) revealed a high degree of packing and consistent platelet dimensions, in agreement with the previous TEM observations (refer to 4.2.1 and 4.2.2). The monolayer was found to contain platelets a few tens of nanometres wide, a thickness of about 7 nm and, therefore, an aspect ratio of about 8.

A B 20 LDH-1 monolayer

0 7.7nm

Height(nm) -10

63nm 67nm

-20 0 100 200 300 400 Lateral Displacement (nm)

Figure 4.5:Topography of LDH-1 monolayer deposited on glass slide. AFM height profile (A) and 2 μm wide image (B) of LDH-1 monolayer.

4.3.2 (LDH/PSS)n multilayer deposition on flat substrates

UV-Vis spectroscopy was used as a way to control the deposition of a (LDH/PSS)n multilayer coating onto a quartz slide (Figure 4.6). An absorption peak at a wavelength of about 225 nm was observed from PSS solutions along with a second peak visible at about

260 nm at relatively high concentration. On the other hand, no absorption peaks were observed for the LDH suspension. However, the level of scatter at a fixed wavelength range

133

was found to be directly related to the amount of LDH platelets concentration in the suspension. To separate the components the absorbance band at 225 nm was related to the phenyl group of PSS,280 whilst the broad scattering feature at a fixed wavelength of 350 nm

(away from the absorption peak of PSS) was attributed to the platelets.

Figure 4.6: Absorbance features of PSS and LDH platelets. UV-Vis spectra of PSS solution (A) and LDH suspension (B).

The absorption at 225 nm was used to monitor the deposition of (LDH/PSS)n multilayer coatings on quartz slides, repeating the method developed for the deposition of

LDH platelet and PSS monolayers. The (LDH/PSS)n multilayer coatings deposited after a

2 min rinsing step (4 dips) exhibited a rather non-uniform deposition for a number of bilayers, n, greater than 5 (Figure 4.7). This non-uniformity of the coating was attributed to the progressive contamination of the LDH solution by PSS after the deposition of a few bilayers and vice-versa. The removal of the excess PSS and LDH platelets in a single rinsing step rapidly led to significant contamination of the rinsing water and, therefore, it was likely that loosely attached PSS molecules and LDH platelets were eventually transferred into the

LDH suspension and PSS solution, respectively, resulting in the salting out and subsequent

134

agglomeration of the PSS molecules and LDH platelets (Figure 4.7.D). SEM of the

(LDH/PSS)5 coatings showed rough LDH platelet layers, with randomly oriented platelets and small agglomerates (Figure 4.7.C). By plotting the absorption of the slide at 225 nm as a function of the number of deposited (LDH/PSS) bilayers, a non-linear increase in the absorbance was found, indicative of a non-consistent LbL bilayer (Figure 4.7.B).

A 0.4 2x(PSS/LDH) 1 C 2x(PSS/LDH) 2 0.3 2x(PSS/LDH) 3 2x(PSS/LDH) 4 2x(PSS/LDH) 0.2 5

0.1 500 nm Absorbance (a.u)

0.0 D 200 250 300 350 Wavelength (nm)

B 0.4 One rinsing step Two rising steps 0.3 E

0.2

0.1

0.0 500 nm Absorbance(a.u) 225 nm at 0 5 10 15 20 25 30 Number of bilayer (PSS/LDH)

Figure 4.7: (LDH/PSS)n multilayer coatings deposition onto quartz slides, monitored by UV-Vis spectroscopy. UV-Vis spectra of coated quartz slides with increasing number of (LDH/PSS) bilayers using a single rinsing step in between LbL deposition (A). Plotted absorbance at 225 nm of coated quartz slides as a function of the number of bilayer for both one step and two step rinsing LbL processes (B). Top-view SEM micrographs of quartz slides coated with (LDH/PSS)5 (C) and (LDH/PSS)15 (E) using one and two step rinsing, respectively and an image of LDH-1 platelet suspension at the end of each LbL deposition process (D).

135

By dividing the single rinsing step (4 dips) into two separate steps (two aqueous

NaOH pH 10 tubes per deposition – 2 dips per tube), contamination of the LDH solution with

PSS molecules was avoided, leading to well-organised nanostructure assembly. Indeed, the alignment of the LDH-1 platelets was preserved after the deposition of (LDH/PSS)15/LDH, as evidenced by SEM (Figure 4.7.E). The intensity of the absorption band related to the phenyl group of PSS at 225 nm revealed a linear increase with increasing (LDH/PSS) bilayers, indicating the successful deposition of multilayer coatings (Figure 4.7.B).

A B

1 μm 1 μm

C D

1 μm 1 μm

Figure 4.8: (LDH-1/PSS)n/LDH-1 Layer-by-Layer deposited multilayer coatings on quartz slides. SEM cross-section micrographs of quartz slide coated with (LDH-1/PSS)n/LDH-1 with n=10 (A), 50 (B), 90 (C) and 155 (D).

While the previous coatings were obtained by manual dipping, regular (LDH/PSS)n multilayer coatings (n > 15) were formed using an automated dipping robot. The thickness of the coatings was investigated by cross-section SEM images of the coatings, which revealed a linear increase of the coating thickness as a function of the number of bilayer deposited

136

(Figure 4.8). All types of LDH platelets (LDH-1, LDH-2 and LDH-3) were used to deposit coatings onto quartz slides, which was monitored by UV-Vis spectroscopy (Figure 4.9).

A B 0.20 1.25 (LDH/PSS) (LDH/PSS) n n 1.00 0.15

0.75 0.10

0.50 LDH-1 0.05 LDH-1 0.25 LDH-2 LDH-2 LDH-3 LDH-3

Absorbance at 225 nm Absorbance (a.u) 0.00 at 350 nm Absorbance (a.u) 0.00 0 50 100 150 200 250 300 0 50 100 150 200 250 300 Number of bilayers (n) Number of bilayers (n)

While the absorbance of Figure 4.7.B was directly plotted from UV-Vis absorbance of unprocessed spectra, here, the scattering and the specific absorption band at 225 nm related to the LDH nanoplatelets and the phenyl group of PSS, respectively, were deconvoluted. By plotting the absorbance of the deconvoluted band at 225 nm as well as the level of scatter at

350 nm as a function of the number of deposited bilayers, linear fits were obtained for both constituents, implying a repeatable deposition of each monolayer and, therefore, successful

LbL multilayer assembly (Figure 4.9). Moreover, the consistent appearance of the optical interference colour generated by the different coatings indicated excellent uniformity across a range of different samples (Figure 4.10.A-C). A coating thickness of about 0.4, 0.8 and

1.2 μm was measured for a number of (LDH-2/PSS) bilayers of 25, 50 and 75, respectively, which corresponds to a bilayer thickness of roughly 16 nm. SEM images of the coated quartz slide cross-section show very uniform thickness (about 1.5 μm) of the coating over the slide

(Figure 4.10.D).

137

A B C

1 cm

D 5 μm

4.4 Morphology of nanostructured (LDH/PSS)n coatings

The morphology and composition of the nanostructures, prepared using different synthesised LDH platelets and PSS, was characterised using UV-Vis spectroscopy, electron microscopy, thermal analysis and X-rays. The phase proportions, platelet packing and platelet alignment were investigated in order to later relate the composition and structure of the coatings to their mechanical properties.

138

4.4.1 Organic content of nanostructured (LDH/PSS)n coatings

LDH-1, LDH-2 and LDH-3 platelets were used to deposit (LDH/PSS)n multilayer coatings with a thickness of roughly 1 μm, yielding samples (LDH-1/PSS)155, (LDH-2/PSS)50 and (LDH-3/PSS)50. Due to the small dimensions of LDH-1 platelets, a greater number of

(LDH-1/PSS) bilayer deposition cycles was required to create a coating with a thickness of about 1 μm as compared to those produced using LDH-2 and LDH-3 platelets. By UV/Vis, comparable levels of scatter were measured at a wavelength of 350 nm, indicating a similar content of platelets, though different amounts of PSS were present in the coatings

(Figure 4.11.A). Indeed, a much stronger absorption by PSS was observed for the

(LDH-1/PSS)155 coating, attributed to the greater number of PSS layers than in the

(LDH-2/PSS)50 and (LDH-3/PSS)50 coatings, as evidenced by the deconvoluted spectra

(Figure 4.11.B). It is expected that each PSS layer was always the same thickness, limited by its charge screening ability, regardless of the LDH platelet thickness. However, by dividing the maximum absorbance at 225 nm of the deconvoluted spectra by the number of deposited

PSS layers, different values where obtained for the three coatings. A 7.7.10-3, 5.5.10-3 and

7.0.10-3 value was calculated for the 1 μm thick coating containing LDH-1, LDH-2 and

LDH-3 platelets, respectively, indicating that less PSS per deposition cycle was deposited when assembled with LDH-2 platelets. The reduced amount of PSS deposited over LDH-2 platelet monolayers as compared to that deposited over LDH-3 platelets, after each LbL cycle, can be attributed to a greater degree of platelet arrangement (narrow size-distribution), leading to smoother LDH monolayer surfaces (further discussed in 4.4.3). Similarly, the slope of Absorbance = f (n) in Figure 4.9 associated to the amount of PSS deposited by LbL cycle in each coating, follows the same trend (Table 4.2). Therefore, the organic content in

139

the embodiments is not only related to the platelet thickness but also to the packing efficiency of the platelets. UV-Vis spectra qualitatively show that use of thicker LDH-2 and LDH-3 platelets, produced thicker (LDH/PSS)n coatings with a smaller organic content.

A B

2.5 1.25 (LDH-1/PSS) ~1um-thick ~1um-thick 155 (LDH-1/PSS) (LDH-2/PSS) 2.0 155 1.00 50 (LDH-2/PSS) (LDH-3/PSS) 50 50 (LDH-3/PSS) 0.75 1.5 50

1.0 0.50

0.5 0.25 Absorbance (a.u)

0.00 0.0 225 250 275 300 325 350 Deconvoluted absorbance (a.u) 200 220 240 260 Wavelength (nm) Wavelength (nm)

Figure 4.11: Qualitative comparison of organic content in 1 μm-thick (LDH/PSS)n coatings containing different LDH platelet dimensions and size distributions. UV-Vis spectra (A) and deconvoluted PSS absorption band (at 225 nm) (B) of 1 μm-thick coatings deposited on quartz slide with different LDH platelet reinforcements (LDH-1, LDH-2 and LDH-3).

The PSS content in the (LDH/PSS)n coatings was also quantified using TGA

(Figure 4.12). All coatings and their individual constituents were analysed assuming that the contributions to the hybrid are simply additive. Using a rule of mixture, a decrease in the organic content from about 40 to 10 % was measured with increasing dimensions of the incorporated LDH platelets (LDH-1 and LDH-2, respectively) (Table 4.2). The

(LDH-2/PSS)50 coating had the highest inorganic content with an inorganic phase proportion of 88.4 wt.%, approaching that of natural nacre (~ 95 wt.%). The two other coatings

[(LDH-1/PSS)155 and (LDH-3/PSS)50] had an inorganic content of 57.3 and 83.3%, respectively (Table 4.2).

140

Figure 4.12: Organic content of (LDH/PSS)n coatings produced using different LDH platelets. Thermal gravimetric analysis of (LDH/PSS)n coatings and their individual constituents in air.

The latter results are in good agreement with the qualitative results obtained from

UV-Vis spectroscopy. Indeed, a direct comparison of the maximum UV-Vis absorbance at

225 nm and the organic content measured from TGA of (LDH-2/PSS)n and (LDH-3/PSS)n, normalised to the values obtained for (LDH-1/PSS)n, shows a good correlation (Figure 4.13).

Some discrepancies arise from difference in thickness of the coatings analysed by UV-Vis

(roughly 1 μm).

141

n 0.5 TGA (org. wt.%) UV-Vis (abs. at 225 nm) 0.4

0.3

0.2

0.1

Normalised (LDH-1/PSS) to 0.0 (LDH-2/PSS) (LDH-3/PSS) n n

Figure 4.13: Comparison of organic content as determined by thermal gravimetric analysis and UV-Vis spectroscopy in the (LDH/PSS)n coatings containing LDH-1, LDH-2 and LDH-3 platelets. Organic content of (LDH/PSS)n coatings obtained from TGA (Figure 4.12) and UV-Vis absorbance (Figure 4.11.B) at 225 nm of ~ 1 μm-thick (LDH/PSS)n coatings normalised to (LDH-1/PSS)n coating values.

4.4.2 Packing of LDH platelets in nanostructured

(LDH/PSS)n coatings

LDH platelet arrangement and packing in the different nanostructures was investigated by electron microscopy, using coatings with a thickness up to about 1 μm. Thick

(LDH/PSS)n/LDH coatings, regardless of the type of LDH platelets used to produce them, retained a uniform, densely-packed final layer as evidenced by SEM (Figure 4.14.A-C); cross-sections (Figure 4.14.D-F) show homogenous coatings with regular thicknesses of roughly 1 μm [(LDH-1/PSS)155, (LDH-2/PSS)50 and (LDH-3/PSS)50]. Top-surface images of the different coatings evidence a high degree of platelet packing. The high degree of packing in the nanostructures was achieved through the flat arrangement of anisotropic LDH platelets, forced by their high surface density. However, the narrow size-distribution of LDH-1 and

142

LDH-2 platelets seems to allow for a better degree of packing as compared to polydispersed

LDH-3 platelets, due to a better match in adjacent platelet thickness.

A B C

500 nm 500 nm 500 nm LDH-1 LDH-2 LDH-3 D E F

1 μm 1 μm 1 μm

In addition to the low organic content, the (LDH-2/PSS)50 coating has a high degree of packing with a “brick-and-mortar” appearance as exemplified by TEM cross-sections, mapped using EDX to distinguish the LDH platelets from the PSS (Figure 4.15), similar to the random platelet-distribution found in the “sheet nacre” structure found in bivalves109

(Nucula nitidosa). The carbon atoms contained in the PSS layer are well distributed around the platelets, which themselves have a uniform composition of Mg/Al/O.

143

4.4.3 LDH platelet alignment in (LDH/PSS)n coatings

The assembly of (LDH/PSS)n with different dimensions and size distributions of the

LDH platelets led to varying degrees of platelet alignment within the nanostructure. The degree of alignment was quantified using three-dimensional XRD rocking curves, acquired at a fixed 2θ = 11.7°, corresponding to the (003) crystal plane reflection parallel to the LDH platelet surface. Misalignment values were obtained from the Full Width at Half Maximum

(FWHM) of the rocking curves (Figure 4.16). The LDH platelets in the (LDH-2/PSS)50 coating showed an encouragingly high degree of alignment (±8°). Greater LDH polydispersity and thinner platelets reduced the alignment quality of the LDH platelets in

(LDH-3/PSS)50 (±17°) and (LDH-1/PSS)155 (±20°) coatings, respectively (Table 4.2). The narrower platelet size-distribution, the better degree of platelet alignment. However, the

144

thickness of the platelet also needs to be significantly larger than that of the polymer layer in order to minimise any misalignment caused by the roughness of the polymer layer.

Figure 4.16: Platelet alignment in (LDH/PSS)n/LDH coatings as a function of platelet dimensions and size distribution. Three-dimensional rocking curves of (LDH-1/PSS)200/LDH-1, (LDH-2/PSS)75/LDH-2 and (LDH-3/PSS)75/LDH-3 coatings acquired at 2θ=11.7° ( diffraction condition of (003) crystalline plane) for Ψ = [0; 80°] (graph step: 30°) and Φ = [0; 360°] (graph step: 90°).

In dense films, poorer alignment correlated with an increased organic phase content, as was in fact confirmed by TGA and UV-Vis spectroscopy in the case of the (LDH-1/PSS)n and (LDH-3/PSS)n coatings. Larger PSS content deposited per LbL cycle can be attributed to the significant misalignment of relatively small platelets. Similarly, the higher organic phase content in (LDH-3/PSS)n is most likely due to the presence of platelet mismatch and, therefore, a rougher monolayer, due to a larger polydispersity in platelet dimensions (refer to

4.2.1).

The most promising nanostructure inspired by the “brick-and-mortar” structure of nacre is the coating consisting of (LDH-2/PSS)n, with similar phase proportions, geometry and aspect ratio. It is interesting to correlate the morphology of each nanostructure to their mechanical response to understand the deformation mechanism occurring in the coating and, therefore, shed light on fundamental scaling behaviour of nacre nanomimics.

145

Table 4.2: Summary of the different (LDH/PSS)n nanostructure morphological properties.

LDH LDH LDH dep. rate PSS dep. rate (LDH/PSS) misalignment n / wt.% / abs.n-1 / abs.n-1 / ° LDH-1 57.3 ± 20 4.84.10-4 3.49.10-2 LDH-2 88.4 ± 8 1.21.10-3 2.79.10-2 LDH-3 83.3 ± 17 1.91.10-3 3.67.10-2

4.5 Mechanical properties of nanostructured (LDH/PSS)n coatings

Natural nacre exhibits great mechanical properties, such as high stiffness, strength and toughness while being made of materials having mediocre mechanical properties. The outstanding mechanical properties of nacre arise from its well-organised “brick-and-mortar” structure with specific phase proportions and aspect ratio. Here, we attempted to correlate the structure of our nanostructured coatings with their mechanical properties and deformation mechanisms, while investigating the impact of scaling down the structure by more than an order of magnitude. Shallow nanoindentation, deep nanoindentation and in-situ SEM nanoindentation using various tips was carried out.

4.5.1 Mechanical properties of nanostructured

(LDH/PSS)n coatings by shallow nanoindentation

Elastic modulus and hardness of the nanostructured coatings were determined from the unloading segment of the load-displacement nanoindentation curves (Figure 4.17) using the Oliver and Pharr method,271 while the plastic and elastic work were assessed from the area under the corresponding segments of the load-displacement curves. The indentation depth was kept below 15 % of the coating thickness to avoid any substrate effect. The indent

146

dimensions were in the order of 1-2 μm wide and 100-200 nm deep, leading to the indentation of at least 10 platelets along the width and through the thickness of the indent, which is representative of the material overall architecture. (LDH-2/PSS)n had the highest modulus (65.8 ± 3.2 GPa) of all samples and indeed the highest value reported to date for any artificial nacre; the result highlights the importance of producing nacre nanocomposites with high inorganic content and a well-organized uniform nanostructure. Despite the similarly high platelet content, (LDH-3/PSS)n possessed a lower modulus with greater scatter due to the heterogeneity in the structure; the lower modulus of (LDH-1/PSS)n is due to the higher organic phase content. Similar trends were observed in hardness, again the (LDH-2/PSS)n coating performed best (2.34 ± 0.18 GPa). These properties approach that of natural nacre according to most recent nanoindentation data,281 an elastic modulus and hardness of about E

= 73.62 ± 8.21 GPa and H = 3.32 ± 0.91 GPa, respectively, were reported for natural nacre in its dry state. Elastic modulus of natural nacre in the range of 40 – 60 MPa, in wet conditions, were reported.16,115

147

0.6 A (LDH-1/PSS) /LDH-1 n 0.5

0.4

0.3

0.2 Load (mN)

0.1 W W plastic elastic 0.0 0 30 60 90 120 150 180 Depth (nm)

0.6 B (LDH-2/PSS) /LDH-2 n 0.5

0.4

0.3

0.2 Load (mN)

0.1 W W plastic elastic 0.0 0 30 60 90 120 Depth (nm)

0.6 (LDH-3/PSS) /LDH-3 C n 0.5

0.4 Pop-ins

0.3

0.2 Load (mN)

0.1 W W plastic elastic 0.0 0 30 60 90 120 150 180 Depth (nm)

The shape of the load-displacement curves relates to the deformation mechanisms occurring in the coatings. The (LDH-1/PSS)n coatings show constant resistance to mechanical load as predominantly the organic matrix flows plastically. (LDH-3/PSS)n exhibits a high 148

initial resistance to loading and subsequent softening evidenced by load drops (“pop-in”) generated due to the presence of disorganised or unevenly sized inorganic platelets. For comparison, indentation of a geological aragonite monocrystal revealed the presence of

“pop-ins” caused by stress accumulation underneath the indenter tip, triggering catastrophic failure of the material; in contrast, natural nacre exhibits viscoelastic properties buffering stress concentrations within the platelets and yielding at a constant stress level.134 The polydispersed randomly orientated platelets (LDH-3) do not allow for in-plane sliding of the platelets over one another as observed in natural nacre, even at high inorganic content

(83.3 wt.%). Instead, high local stress concentrations arise at the interface between differently-sized platelets upon loading. Increasing the load applied to the coating eventually triggers failure of the jammed platelets, causing “pop-ins” in the loading segment of the load-displacement curve. In contrast, the (LDH-2/PSS)n coatings strain harden, an effect which can be attributed to in-plane sliding of the well-arranged LDH platelets and subsequent progressive platelet interlocking. This strain hardening phenomenon of (LDH-2/PSS)n coatings is similar to that of natural nacre,109 although the exact mechanism of platelet interlocking is unclear; in natural systems, wedging,145,151 asperities,130,131 mineral bridges,124,126 nanograin rotation133 and negative Poisson’s ratio146 have been proposed.

4.5.2 Plasticity of nanostructured (LDH/PSS)n coatings determined by deep nanoindentation

Indentations into a 3.5 μm-thick (LDH-2/PSS)n coating at varying loads using a sharp cube-corner tip were also carried out to attempt to assess the fracture toughness and plasticity of the coating, while keeping the indentation depth lower than the coating thickness

(Figure 4.18). In principle, fracture toughness can be estimated from the length of cracks

149

triggered from the corners of indentations.282 However, since the lengths of the cracks in the

(LDH/PSS)n systems are short compared to the indent size, quantitative analysis is difficult

(Figure 4.18.E-G). A minimum load of about 3 mN was required to initiate a crack from the edge of the indent with a length of about 100 nm, while the indent imprint width was about two orders of magnitude higher qualitatively, indicating a large fracture toughness of the

(LDH-2/PSS)n coating.

150

In order to estimate the energy absorption within the nanostructured coating, the

152 plastic index was measured for (LDH-2/PSS)n (Figure 4.19). A 3.5 μm thick coating was loaded with a sharp cube-corner tip at various loads to measure its index of plasticity, based on the areas of plastic deformation and viscoelastic recovery region under the load-displacement curve. The (LDH-2/PSS)n coating exhibits a plastic index levelling off at a value of about 0.95 for a depth superior to 1 μm; a sufficiently high number of LDH/PSS interfaces as compared to the volume of indented platelets needs to plastically deform in order to evaluate a stable bulk properties of the nanostructure. Deeper indents are required to obtain reliable plasticity index as compared to elastic modulus and hardness, which is a result of a large plastic field beneath the indentation tip. The plastic index of the “brick-and-mortar” nanostructure is therefore significantly higher than that of natural nacre,152 of 0.78. Others reported nanoindentation load-displacement curves for natural nacre have a lower proportion

131,283 of plastic deformation than the (LDH-2/PSS)n coating. The plasticity of the well-ordered

(LDH-2/PSS)n suggests that platelet sliding and subsequent interlocking in the vicinity of the crack tip are important enablers of energy dissipation through local deformation and crack deflection. (LDH-3/PSS)n has a similar volume density of interface, but is not sufficiently ordered to allow for controlled platelet sliding leading to strain hardening. The amplification of plasticity in the nacre nanomimic can be explained by an increase in the absolute volume of interface per unit volume, leading to larger platelet sliding/interlocking site density and subsequently more plasticity.

151

A B ~ 3.5 μmB

1 μm

C 5 1.0 (LDH-2/PSS) /LDH-2 n m) 4 

0.9 index Plastic 3

2 0.8

1 Indendation depth ( 0 0.7 0 1 2 3 4 5

Load (mN)

4.5.3 Deformation mechanism of nanostructured

(LDH/PSS)n coatings characterised by SEM in-situ nanoindentation

SEM in-situ nanoindentation of the three (LDH/PSS)n nanostructured coatings revealed different deformation mechanisms occurring in the coatings, which are influenced by the degree of alignment, volume fraction and the order of the LDH platelets (Figure 4.20).

152

2 μm

5 μm

2 μm

5 μm

2 μm

Figure 4.20: Pile-up formation in (LDH/PSS)n/LDH nanostructured coatings during nanoindentation. SEM side-view images of in-situ indents made on (LDH-1/PSS)200/LDH-1, (LDH-2/PSS)75/LDH-2 and (LDH-3/PSS)75/LDH-3 ~1.5 μm thick coatings at a depth of about 10 μm (A, B and C, respectively) (black arrows indicate multiple and subsequent displacement of materials while white arrows pinpoint coating failures within the pile-ups) and the associated SEM top-view images of (LDH-2/PSS)n/LDH-2 and (LDH-3/PSS)n/LDH-3.

While quantitative values were extracted from the displacement-load curves obtained from shallow indents (< 15 % of coating thickness) to avoid substrate effects, deeper indentations to a depth of about 10 μm were also performed in order to image the

153

deformation mechanisms occurring in the vicinity of the indent in-situ (Figure 4.20.A-C). In all cases, the nanocomposite coatings were pushed aside and piled-up during deep indentation. (LDH-1/PSS)n, with its high organic phase content, exhibited large and even pileups at the edges of the indent with a blunted appearance, typical for a viscoelastic polymer, which was also observed for wet natural nacre.117 The formation of smooth pileups indicates low friction within the coating and, therefore, no platelet interlocking. On the other hand, deep indentation into both (LDH-2/PSS)n and (LDH-3/PSS)n, with their high inorganic phase contents, produced shorter, steeper pileups, indicating more friction in the coating during material rearrangement. This phenomenon has also been observed for dry natural nacre,117 which exhibits more friction resistance at the platelet interface and greater strain

1 hardening. These pileups show distinct features compared to the (LDH-1/PSS)n indents, which can be attributed to platelet interlocking and coating fracture in (LDH-2/PSS)n and

(LDH-3/PSS)n, respectively. Progressive hardening occurs in the coating containing LDH-2 platelets; the first wave of material pushed aside by the tip is arrested, leading to the initiation of secondary propagating waves within the pileups. The top-view of the (LDH-2/PSS)n coatings after indentation (Figure 4.20.E), confirms multiple sites of platelets sliding within the pileups; a similar view of (LDH-3/PSS)n shows a more brittle deformation with cracks appearing in the hybrid material (Figure 4.20.F) presumably caused by stress accumulation at heterogeneities within the structure. Indeed, (LDH-2/PSS)n exhibits a higher ratio of plastic work (yielding) to elastic work compared to (LDH-3/PSS)n, while maintaining a better integrity after deformation. Sliding and interlocking of a large number of platelets, within the well-organized (LDH-2/PSS)n nanostructure, can provide a mechanism for substantial plastic deformation at high loadings, leading to a large work of deformation, combined with high modulus and hardness. The mechanical properties of the different coatings deposited on flat substrates are summarised in Table 4.3.

154

Table 4.3: Summary of the different (LDH/PSS)n nanostructure mechanical properties.

W /W Elastic modulus Hardness plastic elastic Platelet behavior (LDH/PSS)n / GPa / GPa (shallow indent) in pileups LDH-1 27.68 ± 0.98 1.35 ± 0.06 2.57 ± 0.05 Flow in matrix LDH-2 65.79 ± 3.16 2.34 ± 0.18 3.44 ± 0.13 Progressive interlocking LDH-3 35.30 ± 1.88 1.07 ± 0.12 3.20 ± 0.08 Failure

4.5.4 Crack deflection in nanostructured (LDH/PSS)n coatings

The much shorter cracks observed in (LDH-2/PSS)n compared to (LDH-3/PSS)n

(7 and 70 μm, respectively), qualitatively indicate a much higher fracture toughness for the

(LDH-2/PSS)n coating (Figure 4.21). The short cracks initiated in (LDH-2/PSS)n coatings

9,52 displayed significant deflections similarly to natural nacre, whereas in (LDH-3/PSS)n, the unevenly-sized platelet interfaces caused obvious local coating failures.

155

A B

1 μm 1 μm LDH-2 LDH-3 C D

200 nm 200 nm

4.6 Towards the addition of a level of hierarchy in

(LDH/PSS)n nanostructured coatings

As described in Chapter.2, natural nacre contains multiple levels of hierarchy since the aragonite platelets are made of aragonite nanograins,118 themselves embedded in an organic soft matrix, providing large plastic deformation through toughening mechanisms, such as nanograin rotation.133 One of the most exploited properties of LDH platelets is the possibility to transform their structure into porous mixed metal oxides through calcination at moderate temperature and to reconstruct the initial structure by rehydration.284,285 During calcination, the porogen (anion interlayer) is selectively removed, which offers the possibility to rehydrate the platelet with an ionic polymer, intercalated between the cationic layers.286

156

The incorporation of the guest polymer can be done by various methods.287 An attempt was therefore made to introduce PSS into the structure of LDH platelets as well as between the platelets, providing an additional level of hierarchy in the nansotructure. (LDH-2/PSS)n was calcined (at 450°C for 4 h) followed by rehydration in a PSS solution (0.1 wt.% for 6 days under nitrogen), as a route to thread the platelet with the soft PSS PE reincorporate PSS at the platelet interface.

SEM images of the as-deposited and calcined nanostructured coatings showed that some porosity arises in the coating after calcination along with noticeable disorganisation of the platelets (Figure 4.22.A-B). Rehydration of the coating in a PSS solution led to the incorporation of PSS in the structure, as shown by SEM images (Figure 4.22.C). However, shallow nanoindentation tests of the as-deposited and calcined/rehydrated coating revealed a loss in plasticity, evidenced by the significant reduction in the plastic deformation of the load-displacement curve. The stiffness of the coating was also slightly reduced, which may be due to a partly infiltration of the PSS PE throughout the coating. On the other hand, the hardness measured was higher than that of as-deposited (LDH-2/PSS)n coating. One can therefore attribute the latter mechanical properties of the calcined/rehydrated coating to poor

PSS infiltration at the LDH interface and potential LDH sintering during calcination, preventing platelets from sliding. Hence, the removal of the soft polymer phase between the platelets evidences its crucial role in bringing plasticity to the nanostructure. It is likely that some PSS re-infiltrated the surface of the coating during rehydration while CO2 contained in air restructured deeper platelets. Therefore, the addition of an additional level of hierarchy to the nanostructure was not successful; the additional toughening observed in natural nacre at the nanometre length scale was not implemented to the nanomimic.

157

A B

200 nm 200 nm

0.6

0.5 D (LDH/PSS) /LDH C 75 0.4 Calcined/rehydrated

0.3

0.2 Load (mN)

0.1 200 nm 0.0 0 20 40 60 80 100 Depth (nm)

4.7 Summary

The synthesis of LDH platelets by a hybrid co-precipitation/hydrothermal method produced stable colloidal suspensions of well-defined hexagonal platelets with an interesting intermediate thickness between single crystal layers and conventional nacre platelets. These nanoplatelets proved suitable for successful LbL assembly of an ordered and dense layered nanostructure, with significantly improved quality compared to structures prepared from thicker platelets using dip coating techniques. The use of platelets with a narrow size

158

distribution and small absolute size allowed for their self-assembly into dense films with high inorganic content (~90 wt.%) and a large number of layers with platelet misalignment as low as 8°. The smaller size of platelets allowed for a degree of reversibility during the self-assembly process, encouraging greater order and better packing. This system echoes the

“brick-and-mortar” structure of nacre, with similar proportions and aspect ratios, but uniformly scaled down by more than one order of magnitude. The minimum thickness of the adsorbed polymer layer (~1.5 nm), which does act as the soft interface, in turn defines the minimum platelet thickness (~13 nm) which allows for a high inorganic content. Thus, the current system based on platelets of this thickness may be close to optimum.

The known toughening mechanisms of nacre, such as platelet sliding and interlocking, as well as three dimensional crack deflection, were also found to occur in this reduced length scale embodiment. The observation of these coordinated mechanisms both confirms and requires the successful preparation of well-ordered films with the correct inorganic-organic composition; controls using more polydispersed or incorrectly sized platelets do not generate the required architecture or phenomenology. The best coatings possessed an elastic modulus and hardness close to that of natural nacre and yet allowed for large substantial plastic deformations to occur in the material, upon loading. This combination of high strength and stiffness along with plastic deformation is a long standing goal of nanocomposite materials.

The reduction in scale of the “brick-and-mortar” structure allows for an increase in the absolute interface density, potentially leading to a greater toughness, whilst retaining the strain hardening mechanisms of nacre. These robust well-arranged bioinspired hybrid nanocomposites offer opportunities to manufacture lightweight coatings with excellent mechanical performance. The use of nanoscale platelets is of interest as they become insensitive to pre-existing flaws, maximizing their strength and, therefore, allowing an increase of their critical aspect ratio.143 Moderate increases in lateral platelet size, whilst

159

maintaining the optimum thickness and avoiding platelet fracture, would offer improved all round mechanical performance for the design of high performance nanocomposites.5

The transfer of the optimised nanostructured coating onto curved fibre surfaces along with its impact on the response of single fibre model composites was then undertaken as a route to create toughening mechanisms within FRP composites and, therefore, increase its strength and strain to failure (refer to Chapter.1).

160

Transfer of Nanostructured Coatings onto

Glass Fibres and their Properties as

Composite Interphases

When a fibre breaks in tension within a FRP composite, local debonding may occur leading to short slippage of the broken fibre. Shear stresses transfer load previously carried by the broken fibre to neighbouring fibres.21 In addition to load redistribution, a specific stress concentration due to the localised stress field can occur around the broken fibre.288 As loading continues, fibre breaks accumulate in the composite, eventually forming a critical cluster.33 When the number of break in neighbouring fibres increases, the arising stress concentration eventually triggers a sudden failure of the whole composite, which typically occurs at relatively low strains to failure. The optimised nanostructured coating

[(LDH-2/PSS)n] (refer to Chapter.4) was transferred onto the surface of glass fibres using the same LbL approach in order to create a composite interphase capable of delaying the

161

formation of critical clusters of fibre breaks and to allow for plastic deformation at the fibre/matrix interface through stable fibre slippage. Crack deflection within the anisotropic coating is expected to effectively dissipate the stresses arising after a fibre break, preventing the early formation of critical clusters of broken fibres, while platelet interlocking within the nanostructured coating in shear leading to strain hardening should provide stable and progressively arrest fibre slippage after fragmentation. The LbL process was adapted to enable the deposition of a (LDH-2/PSS)n coating, with the retention of the desired, ordered nanostructure, onto glass fibre bundles. The produced coated glass fibres were mechanically tested by nanoindentation to confirm the transfer of the coating onto the fibres. Single fibre model composites were tested to assess the potential of the nanostructured composite interphase to improve the tensile behaviour of FRP composites.

5.1 Introduction

The dimensions of the LDH platelets were chosen from Chapter 4, in order to produce a (LDH/PSS)n nanostructured interphase with similar toughening mechanisms to those occuring in natural nacre. The dimensions of the platelets were carefully chosen to allow for conformal assembly of the nanostructure on the surface of the fibres (refer to Chapter.1).

Therefore, 130 nm wide LDH platelets with a narrow size distribution and a thickness of

13 nm were assembled with PSS around glass fibres, which allows for suitable phase proportions, geometry and a high degree of packing.

The surface of as-received glass fibres was first modified to produce a high surface charge density in the deposition medium, required for successful LbL assembly. Even though the deposition of micrometre thick coatings by LbL assembly is time consuming, the

162

self-limiting nature of the LbL process allows for the possibility to coat multiple substrate surfaces simultaneously, providing the ability to coat a large number of fibres in a parallel and continuous manner. The amplification in the amount of material deposited around multiple fibres makes the LbL approach promising and easily adaptable to current fibre sizing/modification processes. The parallel LbL deposition of LDH monolayers, (LDH/PSS) bilayers and (LDH/PSS)n multilayer coatings on fibres bundles was adapted to successfully transfer the optimised (LDH/PSS)n coating process. Finally, single fibre composite tests were used to study the interfacial response of the nanostructured interphase. The transfer of the toughening mechanisms, occurring in the planar coating, ino the interphase, was investigated

5.2 Glass fibre surface charge density

Based on electrostatic attraction forces between charged particles and oppositely charged substrates, the LbL deposition process requires a substrate with a high surface charge density for assembly and adhesion. Since the deposited fibre coating is subjected to shear loading during mechanical loading of composites, a good adhesion of the coating to the fibre surface is crucial. Hydroxylated glass and quartz slides were successfully used as substrate for the assembly of planar (LDH/PSS)n coatings (refer to Chapter.4). Hence, in order to deposit a nanostructured coating on the surface of glass fibres, highly negatively charged hydroxylated glass fibres were required.

The commercial sizing present on the surface of glass fibres was chemically removed by cleaning in piranha solution similarly to the treatment of glass slide substrates. This process also resulted in the formation of hydroxylated glass fibre surfaces and, therefore, a highly negatively charged fibre surface (> |50 mV|) at high pH (Figure 5.1), rendering the

163

fibres suitable for LbL deposition. According to the technical information provided by the manufacturer (AGY), the fibre properties should not be altered by the low pH piranha treatment (refer to Chapter.3), they only start dissolving at high pH (>11). The dissociable silanol surface functional groups present on the surface of hydroxylated glass fibres led to a shift of the isolelectric point (i.e.p.) of the ζ -pH curve towards slightly higher pH as well as more negative ζ–potential values at pH higher than 7, as compared to sized glass fibres. A

ζ-potential of about -53 mV was measured at pH 10 for piranha treated glass fibres, more negative than that of the as-received sized fibres (about -39 mV).

40 As received (sized) Hydroxylated (desized) 20

-20

-40 potential (mV)  -60 iep iep 2 4 6 8 10

pH @ [KCl] = 5 mM Figure 5.1: ζ-potential = f (pH) of hydroxylated glass fibres. ζ-potential curves of as-received sized and piranha treated (desized) glass fibres measured in 5 mM KCl supporting electrolyte solution for pH 3-10.

5.3 Adapting the Layer-by-Layer assembly of multilayer

(LDH/PSS)n coatings around fibres in bundles

Since LbL assembly is self-limiting at each stage, there is an intriguing possibility to coat a large number of fibres in parallel. To test this idea, the LbL deposition of the

164

nanostructured coating [(LDH-2/PSS)n] onto a bundle of glass fibres (with an average diameter of 9 μm) was attempted, starting with an initial LDH monolayer, followed by the deposition of (LDH/PSS)n multilayer coatings.

5.3.1 LDH monolayer deposition on glass fibre bundles

A bundle of hydroxylated glass fibres was dipped into a LDH suspension at pH 10 for

10 min and subsequently rinsed in an aqueous solution of NaOH of pH 10, for 2 min

(Figure 5.2). Multiple rinsing methods were investigated to successfully remove the excess of loosely attached LDH platelets and obtain good monolayer coverage without platelet overlap.

Hence, after the initial LDH deposition step, the bundle of glass fibres was then dipped 12 or

6 times in an aqueous NaOH solution of pH 10 (with a total dipping time of 2 min). The surfaces of the fibres after rinsing were investigated by SEM, which revealed a significant excess of LDH platelets even after 12 dips along with some bald spots, resulting from contact between fibres in the LDH suspension (Figure 5.2.A-B). The LDH excess regions and bald patches were avoided by moderately stirring the LDH suspension and rinsing solutions during dipping. SEM images of the resulting fibres indicated a coating with full coverage of the fibre surface (Figure 5.2.C-D). In fact, one dip in an aqueous NaOH solution of pH 10, while stirring, led to a homogenous LDH monolayer with no apparent platelet overlap.

165

A B

1 μm 1 μm

C D

1 μm 500 nm

Figure 5.2: LDH monolayer deposited onto a bundle of hydroxylated glass fibres. SEM images of piranha treated glass fibres dipped in LDH suspension and subsequently dipped in an aqueous NaOH solution of pH 10, 12 and 6 times without stirring (A and B, respectively) and 1 time with stirring (C and D) – red arrow indicates an area with excess of LDH platelets and black arrow shows a bald spot on the fibre.

Even though one rinsing step in an aqueous NaOH solution appeared to be sufficient for the removal of excess particles, the LbL process was kept unchanged as compared to the deposition of the coating onto flat substrates, i.e dipping the bundle two times in two aqueous

NaOH solutions (a total of 4 dips) to avoid contamination of the solutions (refer to 4.3.2), but with moderate stirring applied to all solutions and suspensions.

5.3.2 Deposition of (LDH/PSS)n multilayer coatings on glass fibre bundles

166

The deposition of LDH/(PSS/LDH)n coatings onto bundles of hydroxylated glass fibres was then investigated using the modified LbL deposition process. SEM images show the successful parallel deposition of the coating onto all fibres in the bundle with full coverage and good homogeneity (Figure 5.3).

Figure 5.3: Deposited thin LDH/(PSS/LDH)1 multilayer coatings on a bundle of glass fibres. SEM images of a bundle of coated glass fibres at low (A) and high magnification (B).

5.3.3 Morphology of nanostructured coatings deposited on glass fibre bundles

The following coatings were deposited over a PDDA precursor layer, which shown to promote better adhesion between the fibre substrate and coating (Appendix.5). The top and cross-section morphology of the PDDA/(PSS/LDH)n coatings was investigated by electron microscopy. The platelet alignment, packing and phase proportions were compared to those of coatings deposited onto flat substrates.

SEM images focusing on the surface at high magnification as well as at the edge of the coated fibres allowed for a rough assessment of the degree of alignment of the LDH platelets (Figure 5.4). The platelets contained in the last deposited coating layer appeared to be lay down flat onto the surface of the fibres, even in the PDDA/(PSS/LDH)75 coating. A

167

relatively smooth top surface was observed while focusing on the edge of the fibres at high magnification. High magnification SEM images of the coating surface show a good deposition of the last layer of LDH platelets, which appears to be similar to the 2-dimensional coating described in Chapter.4.

All fibres were successfully coated with a homogeneous coating all around the surface as confirmed by SEM investigations of fibre cross-sections (Figure 5.5). SEM images of coating cross-sections revealed a linear increase of the coating thickness with increasing number of (PSS/LDH)n bilayers deposited, indicating the reproducible deposition of PSS and

LDH monolayers onto the fibres. The measured coating thickness ranged from about 200 nm to about 1.2 μm, as the number of bilayers increased from 12 to 75. Similar deposition rates

168

as for nacre-nanomimetic coatings deposited onto flat glass slides were obtained: about

16 nm per (PSS/LDH) bilayer.

Figure 5.5: Nacre-nanomimetic coatings deposited around glass fibres. Top surface and cross-section SEM micrographs of glass fibres coated with PDDA/(PSS/LDH)n coating. n=12 (A), n=25 (B), n=50 (C) and n=75 (D). SEM micrographs of fibre top surface (i) and cross section (ii and iii).

5.3.4 Mechanical properties of nanostructured coatings deposited on glass fibre bundles

It was previously established that the mechanical properties of the coating strongly depends on its composition and the arrangement of the reinforcing platelets in the matrix

(refer to Chapter.4). Therefore, SEM in-situ nanoindentation was carried out on fibres coated with a thick PDDA/(PSS/LDH)75 coating and compared to the results obtained for the nanostructured coating deposited onto the flat glass substrate. Beside their geometry, the only difference between the two coatings is the presence of the PDDA precursor layer in the case of the coating deposited in fibres, improving coating adhesion to the surface of the fibre.

However, the elastic and plastic field applied to the coatings by nanoindentation were shallow

169

and confined to the surface of the coatings, which allowed for a direct comparison of their mechanical response.

A B On fibre PDDA/(PSS/LDH) On slide 75 0.6

0.4

Load (mN) 0.2

2 μm

0 25 50 75 100 125 Depth (nm)

The elastic modulus and hardness of the coating was measured from shallow indents with a maximum depth of about 100 nm, which is less than 10 % of the coating thickness, therefore avoiding any substrate effect (Figure 5.6). The indent widths (< 1 μm) were small compared to the coated fibre diameter (about 11.5 μm), implying a curvature of ± 8°, similar to the degree of platelet misalignment measured in the nanostructured coating deposition on a flat substrate (refer to Chapter.4). The coating can be considered to be flat at the scale of the measurement. Similar load-displacement curves as acquired for the nanostructured coating deposited on a flat glass substrate were also measured for the nanostructured glass fibre coatings. The elastic modulus and hardness determined using the Oliver and Pharr method,271 were 65.0 ± 8.2 GPa and 2.3 ± 0.7 GPa, respectively. The load-displacement curve as well as elastic modulus and hardness were very similar to the coating deposited onto a flat substrate

(65.8 ± 3.2 GPa and 2.3 ± 0.2 GPa – refer to 4.5.1). The similar mechanical properties and deposition rate of the coating deposited onto glass fibres as compared with the planar coating,

170

confirm the successful transfer of the anisotropic nanostructure onto curved fibre substrates, under retention of the desired, ordered structure.

A B 1.2 1.00 0.3 mN Deposited on fibre 1.0 0.5 mN 0.95 Deposited on slide 0.75 mN 0.8 1.0 mN 0.90

0.6 0.85

0.4 0.80 Load (mN) Plastic Index 0.2 0.75

0.0 0.70 0.00 0.05 0.10 0.15 0.20 0.25 50 100 150 200 250 300 Displacement (m) Nanoindentation depth (nm)

Deeper nanoindendation to a depth of about 20 % of the coating thickness deposited were performed in order to investigate the strain hardening response of the coating as well as its plasticity (Figure 5.7). The loading segment of the load-displacement curves obtained by nanoindentation exhibits some curvature, especially at high mechanical load, which is a signature of strain hardening. The plastic index152 of the coating, the ratio of the plastic deformation area of the curve divided by the overall area including both plastic deformation and elastic recovery, was found similar to that of coatings deposited on flat glass slide substrates. Similar to previous observations, the plastic index was found to increase with the nanoindentation depth. The maximum depth reached during nanoindentation was around

250 nm, which is still low compared to the depth at which the plastic index of the coating is supposed to level off at a value of 0.95 (refer to Chapter.4). Hence, the latter observations again confirmed the good reproduction of the coating on glass fibre surfaces.

171

5.4 Nanostructured interphase properties of single glass fibre model composites

5.4.1 Determination of the impact of the nanostructured interphase on glass fibre debonding behaviour by single fibre pull-out

Glass fibres coated with a nanostuctured coating were embedded in an epoxy resin in order to investigate the properties of the nanostructured composite interphase in shear.

Pull-out tests of single fibres were carried out using an established method (refer to

Chapter.3). A room temperature curing epoxy resin was selected to avoid any dehydration of the nanostructured coating. The strength of the matrix/fibre interface275 as well as the effect of the strain hardening behaviour of the coating, in shear, on fibre debonding and slippage was studied. The stable slippage length of the fibres was investigated as a ratio (“debonding length ratio”, DLR) of the length over which the fibre slips (debonding length, ld) prior to full debonding over the fibre embedded length le.

The effect of the nanostructured coating thickness on the properties of the interphase was investigated by SFPO tests (Figure 5.8). All load-displacement pull-out curves of the different single fibre systems present the same elastic loading segment followed by plastic deformation, which becomes more significant in the case of coated fibres. The plastic deformation occurring in the composite interphase in shear is a result of platelet sliding/interlocking while pulling-out of the PSS organic phase, allowing for stable fibre slippage (Figure 5.8.A).

172

PDDA/(PSS/LDH) coated glass fibres A Embbeded length ~ 150 m B n 0.3 Bare 0.2 n=12 Bare n=25 0.2 n=50 n=75 0.1 (N)

max 0.0 F 0.1 0.2 n=12 0.0 0 2 4 6 8 0.1 -3 -2 Embedded area (10 .mm ) C 0.0 50 0.2 n=25 40 30 0.1

0.0 (MPa) IFSS 10 Load, F (N) Load,F 0.2 n=50 0 Bare n=12 n=25 n=50 n=75 0.1 D

0.5 ) e /l

0.0 d 0.2 n=75 0.4 0.3

0.1 0.2 0.0 0 25 50 75 100 125 150 0.1

Debonding lengthratio (l 0.0 Displacement, S (m) Bare n=12 n=25 n=50 n=75

A high IFSS is of great importance, as it enables stress transfer between the fibres, ensuring high stiffness and strength of the composite. Furthermore, plastic deformation of the interphase in shear and a strain hardening response at relatively high mechanical load is desired, requiring a sufficiently high IFSS. IFSS of the different single fibre systems was determined (taking into account the increase in fibre diameter with the deposited coating) by plotting the maximum force required to fully debond the fibre as a function of the embedded

173

fibre area in the resin and subsequently measuring the slope of the linear fit (Figure 5.8.B).

All the coated fibres present slightly higher IFSS compared to bare desized glass fibres, confirming a good load transfer between the fibre and the matrix in presence of the nanostructured interphase (Figure 5.8.C). An increase in IFSS of about 20 %, 25 %, 44% and

30 % compared to the bare fibres was observed for PDDA/(PSS/LDH)12,

PDDA/(PSS/LDH)25, PDDA/(PSS/LDH)50 and PDDA/(PSS/LDH)75 coated fibres, respectively (Table 5.1). The increase of the IFSS may be a result of rougher fibre surface after coating deposition. Thus, long and stable slippage of the coated fibres can occur while assuring efficient stress transfer between the fibre and the matrix through plastic deformation of the composite interphase in shear (curvature in loading segment of the load-displacement curves).

In addition to the capability of the coated fibres to transfer the load, glass fibres coated with a PDDA/(PSS/LDH)n coating were found to possess greater DLR than the bare fibres (Figure 5.8.D). The use of coatings, consisting of 25 (PSS/LDH) bilayers was found to provide the longest stable slippage of the fibres, with a value of 0.41 (Table 5.1). One can attribute the improvement in fibre sliding to a greater number of platelet sliding/interlocking sites through the thickness of the nanostructured composite interphase. Further increase in coating thickness to about 0.8 and 1.2 μm (50 and 75 bilayers, respectively), resulted in a decrease of the DLR, though it still remained larger than that of the bare fibres. The reduction in the DLR at high coating thickness may be due to a decrease in the radial clamping force acting on the fibre through the coating, which is generated through shrinkage of the epoxy resin during curing, due to the radial compliance of the coating. The use of small platelets

(LDH-1) in the interphase, leading to high organic content and, therefore, no strain hardening in shear, was shown to provide no stable fibre slippage during fibre pull-out (refer to

Appendix.6). It was concluded that the extension of the debonding length, through plastic

174

deformation of the nanostructured interphase, is the result of strain hardening caused by platelets sliding/interlocking.

100 Bare fibre PDDA/(PSS/LDH) n=12 n 80 n=25 n=50 n=75 60

20 Apparent IFSS (MPa) ApparentIFSS

0 0 50 100 150 200 250 Embedded length (m)

Figure 5.9: Relation between apparent interfacial shear strength and fibre embedded length. Apparent IFSS = f (le) for dezised bare glass fibres and glass fibres coated with PDDA/(PSS/LDH)n.

The apparent IFSS was determined for all specimens and plotted as a function of the fibre embedded length, as a route to investigate the type of interfacial failure occurring in each fibre/matrix system.23 Therefore, while a brittle failure is associated with a decrease in the apparent IFSS with increasing fibre embedded length, on the other hand the apparent

IFSS of ductile interfaces is independent of the embedded fibre length. Hence, a ductile interfacial failure was observed for all single glass fibre systems (Figure 5.9).

Table 5.1: Interphase properties obtained from pull-out tests.

Coating thickness IFSS Debonding length ratio Interphase / μm / MPa / a.u. No coating (control) - 31.1 ± 2.7 0.14 ± 0.01

PDDA/(PSS/LDH)12 ~ 0.2 37.2 ± 1.9 0.27 ± 0.05

PDDA/(PSS/LDH)25 ~ 0.4 38.7 ± 2.9 0.40 ± 0.05

PDDA/(PSS/LDH)50 ~ 0.8 44.6 ± 1.6 0.27 ± 0.05

PDDA/(PSS/LDH)75 ~ 1.2 40.3 ± 1.6 0.22 ± 0.02

175

Since the fibres are expected to fragment multiple times within FRP composites during loading and, therefore, slip while redistributing the release energy, the improved debonding behaviour of the coated fibres is of greater importance. Substantial energy can be dissipated within the composite by increasing the degree of fibre fragmentation, leading to an increase of the overall toughness of the material. Hence, it is important that the fibres fragment without local stress concentrations that cause the formation of clusters of fibre breaks, leading to sudden catastrophic composite failure.

5.4.2 Effect of the nanostructured interphase on interfacial shear strength and stress concentration in single glass fibre composite model fragmentation

SFF tests were carried out in-situ, under an optical microscope, in order to investigate the distribution and dissipation of stresses arising from fibre breaks via birefringence pattern of the fibre/matrix interface,22 and to determine the IFSS of the fibre/matrix interface.

Birefringence patterns arising from stresses at the interface between the fibre and the matrix revealed significant differences in stress distribution and absorption in the vicinity of a fibre fragment when comparing bare and PDDA/(PSS/LDH)25 coated fibres embedded in epoxy

(Figure 5.10.A-B). While the fragmentation of bare fibres in epoxy led to an intense stress field near the fibre breaks (Figure 5.10.A), the nanostructured interphase provided the ability to reduce and spread stress concentrations arising from fibre breaks along the fibre length, as indicated by birefringence patterns (Figure 5.10.B). The dissipation of stress concentration may have occurred through crack deflection within the anisotropic interphase and progressive debonding slippage of the fibre.

176

Figure 5.10: Effect of nanostructured interphase on stress distribution near fibre fragments in single glass fibre composite models. Optical micrographs of the fragmentation test of bare and PDDA/(PSS/LDH)25 coated glass fibres in epoxy using cross-polarised (A and B, respectively) and non-polarised light in transmission (C and D, respectively) - vertical arrows pinpoint fibre breaks and horizontal arrows show fragment lengths.

The apparent IFSS of the most promising single fibre model composites obtained by pull-out tests (long fibre slippage), namely, PDDA/(PSS/LDH)25 coated fibres in epoxy, were investigated and compared to bare single glass fibre model composites at fragmentation saturation (specimens strained up to about 25 %). The fibre fragment lengths determined after fragmentation saturation (Figure 5.10.C-D) of fibre breaks are shown in histograms and as cumulative distribution (Figure 5.11). The distribution of the fibre fragment lengths of fibres coated with a PDDA/(PSS/LDH)25 nanostructured coating appeared to be slightly shifted towards smaller values, as a result of a stronger interface. A mean fragment length l

(arithmetic mean) of 335 ± 22 and 260 ± 17 μm was determined for bare and coated fibres, respectively.

177

A 1.0 Desized glass fibres B 30 Bare PDDA/(PSS/LDH) coated 0.8 25

20 0.6

0.4 10 0.2 Desized glass fibres Percentage (%) Bare

Cumulative distribution (PDDA/(PSS/LDH) coated 25 0 0.0 0 200 400 600 800 1000 1200 0 200 400 600 800 1000 1200 Fragment length (m) Fragment length (m) Figure 5.11: Distribution of single fibre fragment lengths after fragmentation. Histogram (A) and cumulative (B) distribution for bare and PDDA/(PSS/LDH)25 coated glass fibres.

The improvement in IFSS of the coated fibre/epoxy interface was assessed from the

21 fibre critical fragment length lc, using the Kelly-Tyson model. Since the tensile strength of the oxidised bare carbon fibres and coated carbon fibres should not differ, one can determine the increase in IFSS of coated fibres by directly comparing the aspect ratio of the critical fragments (Table 5.2), using a fibre diameter of 9 μm and critical fragment lengths of

447 ± 29 and 347 ± 22 μm for bare and coated fibres, respectively (refer to 3.3.11.2). The

IFSS increased by about 29 % when using PDDA/(PSS/LDH)25 coated glass fibres rather than bare glass fibres. The increase in IFSS is in good agreement with the results obtained from SFPO tests (Table 5.2). Quantitative IFSS values were obtained from single fibre fragmentation, using a fibre tensile strength of 3.45 GPa (provided by the manufacturer,

AGY). Hence, an IFSS of 44.2 ± 2.3 and 34.5 ± 2.1 MPa was measured for

PDDA/(PSS/LDH)25 coated and bare fibres, respectively. Hence, the nanostructured interphase was found to enable good stress transfer as well as stress dissipation along the fibre length.

178

Table 5.2: Summary of interphase properties in single glass fibre composite from fragmentation tests.

l l Aspect ratio IFSS (SFF) IFSS (SFPO) Interphase c / μm / μm / a.u / MPa / MPa No coating (control) 335 ± 22 447 ± 29 50 ± 3 34.5 ± 2.1 31.1 ± 2.7

PDDA/(PSS/LDH)25 260 ± 17 347 ± 22 39 ± 2 44.2 ± 2.3 38.7 ± 2.9

5.5 Summary

The ordered anisotropic nanostructure developed on a planar substrate (refer to

Chapter.4) was successfully transferred onto fibre surfaces. The dimensions of the platelets allowed for a conformal deposition of the coating onto 9 μm diameter fibres. The LbL assembly revealed the possibility to coat a large number of fibres simultaneously, making the process promising for continuous deposition on fibre tows. The desired nanostructure was retained as evidenced by the mechanical characterisation of the coating deposited onto the fibres. When used as an interphase in single fibre composite models, the nanostructure was shown to well adhere to the surface of the fibre when deposited over a soft PE precursor layer, allowing for efficient stress transfer. The toughening mechanisms occurring in the planar nanostructure were found to be beneficial when happening in the interphase. Crack deflection at the nanostructured interphase at the platelet soft polymer interface may be the mechanism responsible for stress absorption and spreading along the fibre length, potentially reducing the coupling/accumulation effect of fibre breaks. Futhermore, the strain hardening character of the interphase in shear, caused by platelet sliding/interlocking, allowed for plastic deformation of the nanostructured composite interphase and larger stable fibre slippage. The combination of the two phenomena shed light on promising nanostructured composite systems to produce a new generation of hierarchical FRP composites. The described nanostructured interphase presents opportunities as a new size, which may improve

179

mechanical performance of FRP composites by reducing fibre break accumulation and allowing for extended strain to failure by allowing for longer and stable fibre sliding during the debonding process of the fibres. The LbL process used to coat glass fibres can easily be adapted to manufacture small composites as relatively large bundles can be simultaneously coated (refer to Chapter.7).

180

Transfer of Nanostructured Coatings onto

Carbon Fibres and their Properties as

Composite Interphases

The planar nanostructured coating (refer to Chapter.4) was successfully transferred onto glass fibres, showing improvement in fibre debonding behaviour and interphase stress absorption of single fibre composite models (refer to Chapter.5). However, since the best structural reinforcements are carbon fibres, it is of great interest to investigate them as a substrate for the nanostructured interphase. Similarly to the previous chapter, the coating was applied onto charged fibre surfaces by LbL, followed by single fibre composite model testing. The emphasis remains to assess and increase the progressive sliding of the fibre during debonding from the surrounding matrix, whilst maintaining efficient stress transfer, and to dissipate stress arising from a fibre break.

181

6.1 Carbon fibre surface treatment to improve surface charge density

The LbL deposition process developed to coat bundles of glass fibres (refer to

Chapter.5), was also used to coat bundles of carbon fibres with the same motivation (as described in Chapter.1). However, modification of the surface of carbon fibres was required to ensure full fibre coverage with the desired coating and good adhesion. As-received carbon fibres have a limited charge density (ζ-potential about -20 mV), leading to poor interaction with the deposited coating (refer to Appendix.7). The most successful route to modify the surface of unsized carbon fibres and, therefore, to increase the electrostatic attraction forces between the PDDA precursor layer (ensuring sufficient coating adhesion) and the fibre, was achieved by surface oxidation. Surface treatment of carbon fibres, achieved through a wide range of methods, was found to be critical for composite performance as poor interfacial adhesion due to the matrix is caused by the hydrophobicity and chemical inertness of carbon fibre surfaces.25 Oxygen-containing groups on the surface of the carbon fibres can deprotonate, particularly carboxyl groups, leading to a negatively-charged surface at pH 10, which can be used during LbL deposition. As an alternative, quaternary-terminated amine groups, were introduced as a route to create a positively charged carbon fibre surface.

However, only a very low charge density was achieved on carbon fibre surfaces using this route (refer to Appendix.4).

The unsized carbon fibre surfaces were treated in low pressure oxygen plasma in order to retain the mechanical properties of the fibres while oxidising their surface. Three different exposure times to the plasma at an oxygen flow rate of 50 sccm were investigated, namely 30 s, 5 min and 20 min.

182

A B C D

2 μm 500 nm 2 μm 500 nm

E F G H

2 μm 500 nm 2 μm 500 nm

Figure 6.1: Surface morphology of oxygen plasma treated carbon fibres. Top surface SEM micrographs of as-received unsized fibres (A and B) and unsized fibres treated for 30 s (C and D), 5 min (E and F) and 20 min (G and H).

Qualitative comparison of carbon fibre surface roughness before and after treatment was undertaken by SEM (Figure 6.1). The crenulations present on the surface of the carbon fibres, which are a result of their manufacturing process, become more prominent after treatment, implying plasma etching. When treated for up to 5 min in O2 plasma, the roughness appears to increase only slightly, but drastically at longer exposure times, namely, after 20 min.

183

-20

-40

As received -potential (mV) -60  Treated 30 s Treated 5 min Treated 20 min -80 4 6 8 10 pH @ [KCl] = 5 mM

Figure 6.2: ζ-potential = f (pH) of oxygen plasma treated carbon fibres. ζ-potential curves of as-received and treated unsized carbon fibres from pH 3 to pH 10, in 5 mM KCl.

Oxidation, including plasma oxidation, of carbon fibres is well-known255 to increase the acidity of carbon fibres and, therefore, shift the i.e.p. towards lower pH. However, the i.e.p. of both untreated and treated carbon fibres was below pH 3 (Figure 6.2), so no change was observed. A negative value of the ζ-potential of all fibres over the entire pH range was measured, characteristic for some oxidised carbon fibres.255 Nevertheless, the plasma treatment resulted in more negative ζ-potential due to the introduction of more oxygen-containing groups and, therefore, more acidic carbon fibre surfaces.255 Hence, the

ζ-potential of plasma treated carbon fibres at pH 10, i.e. the pH used for the LbL assembly of the nanostructured coating, was found to be significantly more negative (-45 mV) than that of the as-received carbon fibres (Figure 6.2). The improvement in the absolute value of the

ζ-potential of the treated fibres does not seem to increase significantly with increasing plasma exposure times exceeding 5 min. Therefore, in order to minimize the surface roughness of the carbon fibre while allowing for high surface charge density, carbon fibres treated for 5 min

184

were used for further studies. High roughness LbL substrates should be avoided, in order to ensure good initial monolayer deposition of few nanometres thick platelets and preserve the platelet alignment throughout the thickness of the coating.

-20

-40

-potential (mV)

 -60

Plasma treated Plasma treated / Soaked in KMnO 4 -80 4 6 8 10

pH @ [KCl] = 5 mM

Moreover, an additional oxidation step was introduced by immersing the O2 plasma treated fibres (for 5 min) into a solution of KMnO4 (0.1 M) overnight. This additional liquid phase oxidation was expected to increase the number of carboxylic groups on their surface, through conversion of hydroxyl groups. At pH 10, a ζ-potential of -58 mV was measured as compared to -19 mV for as-received carbon fibres, which represents an improvement of over

40 % as compared to solely plasma treated fibres (Figure 6.3). The high ζ-potential measured for the modified carbon fibres is similar to that of hydroxylated glass fibres (-53 mV) previously investigated (refer to 5.2), suitable for LbL deposition (> |50 mV|).

185

As received C 1s

O 1s N 1s Si 2p

Plasma treated

Intensity (a.u) Plasma treated/soaked in KMnO 4

800 600 400 200 0 Binding Energy (eV)

Figure 6.4: Surface composition of as-received and oxidised unsized carbon fibres. XP survey spectra of as-received, 5 min O2 low pressure plasma treated and plasma treated carbon fibres further modified by oxidation in KMnO4.

The surface composition of as-received carbon fibres as well as fibres treated for

5 min was investigated by XPS (Figure 6.4). The O1s peak (Figure 6.5) varied with the incorporation of additional oxygen-containing groups after oxidation, related to more negative ζ-potential values. The low pressure oxygen plasma treatment increased the oxygen

186

content from 9.1 at.% to 11 at.%, with a further increase to 15.8 at.% with KMnO4 treatment

(Table 6.1). The relative proportion of C=O carbonyl groups appears to be slightly reduced, from 67.3 to 54.5 at.%, after oxygen plasma treatment, according to O1s high resolution spectra (Figure 6.5.A-C), but increased again to about 65.9 % after the KMnO4 treatment

(Table 6.1). In addition, the peak associated to carboxylic group of C1s high resolution spectrum significantly increased after further oxidation of the carbon fibres in KMnO4

(Figure 6.5.F), while the π-π* peak (shake-up satellite) disappears, suggesting additional oxygen groups on the surface of carbon fibres. It can therefore be concluded that the oxidation of the carbon fibre in KMnO4 leads to more oxygen-containing groups on their surface, with a large proportion of carboxylic groups. The O1s/C1s atomic ratio was found to be increased by a factor of 2 for carbon fibres further oxidised in KMnO4 as compared to as-received carbon fibres.

As received Plasma treated Plasma treated/KMnO4

O 1s I C=O (I) A B C C-O (II) Background Enveloppe

II Intensity (a.u) Intensity

535 530 535 530 535 530

Binding energy (eV) Binding Energy (eV) Binding Energy (eV) C 1s I C-C (I) D E F C-O (II) C=O (III) O=C-O (IV) Background

Enveloppe Intensity (a.u) Intensity IV III II

290 285 290 285 290 285 Binding energy (eV) Binding Energy (eV) Binding Energy (eV) Figure 6.5: Oxygen O1s and carbon C1s high resolution spectra of as-received and oxidised unsized carbon fibres. Oxygen O1s and carbon C1s of as-received (A and D, respectively), 5 min O2 low pressure plasma treated (B and E, respectively) and plasma treated followed by further oxidation in KMnO4 (C and F, respectively).

187

Other reports show that the O1s/C1s atomic ratio can increase by a factor 1.5 to 1.8 and about 2.0 for electrochemically oxidised and plasma oxidised PAN-based carbon fibres,289 respectively, as compared to untreated fibres. Significant increase in the proportion of carboxylic groups present on the surface of carbon fibre was measured after electrochemical oxidation289 while the main product of oxygen plasma was found to

290 be -C-OR groups. Hence, the combination of low O2 pressure plasma and further oxidation in KMnO4 enabled for the formation of a large number of oxygen-containing groups, particularly the production of carboxyl groups. The increase in both the proportion of oxygen-containing groups and the number of carboxylic groups on carbon fibre surfaces is responsible for the significant improvement in surface charge density, resulting from proton dissociation at high pH (Figure 6.3).

Table 6.1: Surface composition of as-received and oxidised unsized carbon fibres.

O 1s C 1s O C N C-C * C=O C-O C=O COOR Carbon fibre C=C π - π

at.% at.% at.% C-O at.% at.% at.% at. % at. % at.% As-received 9.1 87.5 2.3 67.3 32.7 93.6 2.7 1.7 2.0 Plasma 11.0 86.7 1.9 54.5 45.5 88.9 4.2 2.1 4.7 Plasma + KMnO 4 15.8 78.5 1.7 65.9 34.1 96.4 1.6 2.0 0.0

6.2 Layer-by-Layer deposition of PDDA/(PSS/LDH)n coatings on oxidised carbon fibre bundles

The previous LbL deposition methodology developed for glass fibres (Chapter.5), was applied to oxidised PAN carbon fibres (AS4) with high surface charge density, leading to

188

LDH monolayers with good coverage and consistent (LDH/PSS)n multilayer coatings, similar to the deposition on glass fibres. The deposition and morphology of PDDA/(PSS/LDH)n coatings on oxidised carbon fibres, with varying thickness, was investigated (Figure 6.6). On the other hand, due to poor interaction with the surface of untreated carbon fibre (low surface charge density), thick coatings (> 0.5 μm) were found to easily peel off, leading poor mechanical properties of the interphase in shear measured by single fibre composite model tests (refer to Appendix.7). Homogeneous PDDA/(PSS/LDH)n coatings with consistent thickness were observed in fibre cross-section SEM images (Figure 6.6). A coating thickness of about 200, 400 and 800 nm were measured for oxidised carbon fibres coated with

PDDA/(PSS/LDH)12, PDDA/(PSS/LDH)25 and PDDA/(PSS/LDH)50, respectively, in agreement with the observations made on coated glass fibres. However, the

PDDA/(PSS/LDH)75 coating thickness was erratic along the fibre length, as a result of significant fibre surface roughness, which was also observed by top surface SEM observations; no thickness could be reliably determined.

A B E F

2 μm n=12 2 μm 2 μm n=50 2 μm C D G H 2 μm

2 μm n=25 2 μm 2 μm n=75 2 μm

Figure 6.6: Nanostructured coating deposited on oxidised carbon fibres. Top surface and cross-section SEM micrographs of oxidised carbon fibres coated with PDDA/(PSS/LDH)n, with n=12 (A and B, respectively), n=25 (C and D, respectively), n=50 (E and F, respectively) and n=75 (G and H, respectively) - white arrows pinpoint a visible cross-section of the nanostructured coating.

189

6.3 Properties of a nanostructured interphase in single carbon fibre model composites

6.3.1 Determination of the impact of the nanostructured interphase on carbon fibre debonding behaviour by single fibre pull-out

The debonding behaviour of all coated fibres as well as the bare oxidised carbon fibres were tested using SFPO tests in epoxy. The load-displacement curves of the pull-out tests show that all coated fibres have a linear elastic loading segment followed by plastic deformation, indicative of shear deformation of the nanostructured interphase (Figure 6.7.A).

On the other hand, oxidised bare carbon fibres exhibit elastic loading of the interface followed by sudden debonding without obvious plastic deformation. The IFSS of the different single fibre composite models was obtained from the slope of Fmax = f (Ae) curves

(Figure 6.7.B). The interface of PDDA/(PSS/LDH)12, 25 coated fibres with the epoxy matrix was much stronger than that of bare fibres (32.4 MPa), with IFSS, of about 56.8 and

59.5 MPa, respectively (Figure 6.7.C). A progressive reduction in the IFSS was observed for thicker coatings, namely, PDDA/(PSS/LDH)50, 75. The IFSS of the latter systems were slightly higher than that of bare fibres (Table 6.2). The reduction in IFSS for

PDDA/(PSS/LDH)50, 75, as observed for single glass fibre systems (refer to 5.4.1), is likely to be related to a decrease in fibre radial clamping force (caused by epoxy during curing), as a result of a larger volume of coating relaxation via radial compliance.

190

A B 0.3 Bare 0.2 n=12 n=0 n=25 n=50 0.2 n=75

0.1 (N)

max 0.0 F 0.1 0.2 n=12 0.0 0 1 2 3 4 5 6 0.1 Embedded area (10-3.mm-2) C 0.0 0.2 n=25 60

0.1 40 0.0 IFSS (MPa) IFSS Load, F (N) Load,F 0.2 n=50 20

0.1 0 D Bare n=12 n=25 n=50 n=75

0.25 0.0 ) e /l 0.2 d n=75 0.20

0.1 0.15

0.0 0.10 0 40 80 120 0.05

Displacement, S (m) Debonding lengthratio (l 0.00 Bare n=12 n=25 n=50 n=75

The DLR, along which the fibre progressively slides prior to full debonding, was determined to be higher for coated oxidised carbon fibres (Figure 6.7.D). A maximum DLR of 0.17 was measured for PDDA/(PSS/LDH)25 coated fibres compared to 0.08 for bare carbon fibres (Table 6.2), which is in line with the results obtained from coated glass fibres

(refer to 5.4.1), and a result of plastic deformation of the interphase in shear. However, the improvement in debonding length is not as large as for glass fibres. The DLR was increased

191

by about 95 % as compared to bare carbon fibres, while the level of frictional interactions with the matrix after debonding was significantly reduced (Figure 6.7.A). A similar reduction of frictional forces was also observed for coated glass fibres (Chapter.5) and coated as-received (non-oxidised) carbon fibres (Appendix.7), confirming the delamination behaviour of the PDDA/(PSS/LDH)n coatings.

125 PDDA/(PSS/LDH) Bare fibre n n=12 100 n=25 n=50 n=75 75

25 Apparent IFSS (MPa) ApparentIFSS 0 50 100 150 200 250

Embedded length (m) Figure 6.8: Relation between apparent interfacial shear strength and fibre embedded length. Apparent IFSS = f (le) for bare carbon fibres and coated oxygen plasma treated carbon fibres coated with PDDA/(PSS/LDH)n.

In order to investigate the type of interface failure of the different fibre systems, the apparent IFSS was plotted as a function of the fibre embedded length (Figure 6.8). As observed for single glass fibre systems, the apparent IFSS does not seem to be dependent to the embedded length of the fibre in epoxy. Therefore, a ductile interfacial failure of model composites was deduced from the independent trend of the apparent IFSS with the fibre

23 embedded length. All systems except for fibres coated with PDDA/(PSS/LDH)12, which seems to present a dependence of the apparent IFSS with the fibre embedded length

192

(progressive decrease). However, a large scatter in the apparent IFSS, especially at relatively short embedded length (≤ 50 μm), can be responsible for this tendency.

Table 6.2: Interphase properties obtained from pull-out tests.

Coating thickness IFSS Debonding length ratio Interphase / μm / MPa / a.u No coating (control) - 32.4 ± 3.2 0.09 ± 0.01

PDDA/(PSS/LDH)12 ~ 0.2 56.8 ± 5.5 0.16 ± 0.01

PDDA/(PSS/LDH)25 ~ 0.4 59.5 ± 3.9 0.17 ± 0.02

PDDA/(PSS/LDH)50 ~ 0.8 40.2 ± 2.2 0.14 ± 0.01

PDDA/(PSS/LDH)75 Not determined 37.4 ± 2.7 0.14 ± 0.01

Providing the formation of a large number of uncorrelated fibre breaks in a composite containing PDDA/(PSS/LDH)25 coated fibres in tension, improvement in both IFSS and DLR could provide the possibility to increase the amount of energy absorption through fibre debonding and stable slippage.

6.3.2 Effect of a nanostructured interphase on interfacial shear strength and stress concentration in single carbon fibre composite model fragmentation

The ability of the nanostructured interphase to dissipate energy released by fibre breakage, through crack deflection at the platelet interface and along the fibre length, was investigated by in-situ SFF tests. The IFSS of the fibre/matrix interface and the effect of introducing the nanostructured interphase was also assessed and compared to data obtained from SFPO tests. Optical microscopy using cross-polarised light in transmission was carried out to observe the formation, accumulation and dissipation of stress in the vicinity of a fibre

193

break. Similar to the behaviour found for bare and coated glass fibres (refer to 5.4.2 ), two different stress fields were observed during fragmentation of single carbon fibres; bare oxidised carbon fibres exhibit an intense and large stress field building up near a fibre break, while PDDA/(PSS/LDH)25 coated oxidised carbon fibres present a less intense stress field, which seems to spread along the length of the fibre (Figure 6.9.A-B). Continuous imaging of the stress field, whilst increasing macroscopic strain at the interface, provided evidence of the progressive propagation of the stress during sliding of the coated fibre against the matrix, accompanied with a noticeable reduction of the stress field intensity (light intensity of birefringence pattern). On the other hand, the stress field intensity observed for fragmented bare fibre does not seem to change (Figure 6.9.C-F). Following the breakage of the fibre, part of the energy release is absorbed within the nanostructured “brick-and-mortar”, potentially through crack deflection and bridging, followed by further stress relief via progressive fibre debonding (stable slippage).

194

Figure 6.9: Effect of nanostructured interphase on stress concentration near a fibre fragment in single carbon fibre composite models. Optical images of in-situ fragmentation tests of bare (A) and PDDA/(PSS/LDH)25 coated (B) oxidised carbon fibres using cross-polarised light in transmission at a given strain (A and B, respectively) and at +1.2% (C and D, respectively) and +2.4% strain (E and F, respectively) – vertical arrows pinpoint fibre fragments and horizontal arrows and dashed line highlight progressive sliding of the fibre. Optical images of bare and PDDA/(PSS/LDH)25 coated oxidised carbon fibres using non-polarised light in transmission after fragmentation test (G and H, respectively) - horizontal arrows show fragment lengths.

Analysis of the fragment lengths of the fragmented single fibre specimens, at saturation (macroscopic strain of about 25%), was carried out to determine the IFSS. The fragment lengths were measured to determine the mean fragment length l (arithmetic mean) of each single fibre composite model. A mean fragment length of 212 ± 11 and 137 ± 9 μm was measured for oxidised bare and coated oxidised carbon fibres, respectively. By plotting the histogram and cumulative distribution of the fragment lengths of coated and oxidised bare fibres, it becomes clear that the fragment lengths of the coated fibres shifted towards shorter

195

lengths (Figure 6.10). A critical fragment length (lc) of 283 ± 14 and 182 ± 12 μm was deducted using the Kelly-Tyson model21 (Table 6.3).

A B 50 1.00 Oxidised carbon fibres Bare 40 PDDA/(PSS/LDH) coated 25 0.75

30 0.50

0.25

Percentage (%) Oxidised carbon fibres 10 Bare

Cumulative distribution PDDA/(PSS/LDH) coated 0.00 25 0 0 200 400 600 0 200 400 600 Fragment length (m) Fragment length (m) Figure 6.10: Distribution of fibre fragment lengths after fragmentation of oxidised carbon single fibre composite with and without nanostructured interphase. Histogram (left) and cumulative (right) distribution for bare and PDDA/(PSS/LDH)25 coated oxidised carbon fibres.

Assuming a similar tensile strength for both fibres, the improvement in IFSS of the coated fibres was measured by comparing the aspect ratio of the critical fragments. The aspect ratio was measured using a fibre diameter of 7 μm and critical fragment lengths of 283 and 182 μm, for bare and coated fibres, respectively (Table 6.3). This comparison indicates an increase of about 54 % in IFSS in the case of the composite with the nanostructured interphase, which is in good agreement with the IFSS measured by SFPO tests (Table 6.3).

Quantitative IFSS values obtained by SFF tests were measured, using a tensile strength of

4.413 GPa (provided by the manufacturer, Hexcel). IFSS values of 84.8 ± 6.6 and

55.2 ± 2.8 MPa was measured for PDDA/(PSS/LDH)25 coated and bare fibres, respectively.

196

Table 6.3: Summary of interphase properties in single oxidised carbon fibre composite from fragmentation tests.

l l Aspect ratio IFSS (SFF) IFSS (SFPO) Interphase c / μm / μm / a.u / MPa / MPa

Bare fibre 212 ± 11 283 ± 14 40 ± 2 55.2 ± 2.8 32.4 ± 3.2

PDDA/(PSS/LDH)25 137 ± 9 182 ± 12 26 ± 2 84.8 ± 6.6 59.5 ± 3.9

6.4 Summary

Coating the surface of 7 μm-wide carbon fibre tows with a nanostructured coating was successfully achieved using a LbL dipping process. Low pressure plasma oxidation followed by further oxidation in KMnO4 of the carbon fibre surface was required to promote good adhesion of the coating to the fibres. A poor surface charge density of the as-received unsized carbon fibres did not allow for sufficient interaction between the fibre and the coating leading to poor IFSS and subsequently no improvement in debonding energy as measured by

SFPO tests (refer to Appendix.7). A significant improvement in the surface charge density of the fibres, as measured by the ζ-potential, which increased from -19 mV to -58 mV at pH 10, was produced by oxidation.

Once a good quality coating was achieved, a combination of higher DLR, through stable fibre slippage, and improved IFSS was observed, with values of about 0.17 and

59.5 MPa, respectively, higher than those obtained for bare oxidised fibres (0.09 and

32.4 MPa, respectively). The increased IFSS of fibres coated with a nanostructured coating was also confirmed by SFF tests. Stress dissipation in the anisotropic composite interphase created by the nanostructured coating along the fibre length was observed during in-situ fibre fragmentation, in the single fibre model composites, potentially hindering the formation of clusters of correlated fibre breaks responsible for the sudden tensile failure of composites.

197

Similarly to hydroxylated glass fibres, plasma oxidised carbon fibres coated with nanostructured coatings exhibit a poor level of friction with the matrix during the extraction part of the pull-out test, caused by the delamination of the coating from the fibre during the interfacial failure process. However, at composite level, the fibre debonding process is predominant. One can expect that with the incorporation of the developed nanostructured interphase in a composite, a large number of uncorrelated broken fibre fragments should occur in the materials, leading to more stable slippage of broken fibres, hence greater strength and strain to failure

The most promising mechanical properties were obtained for oxidised carbon fibres coated with PDDA/(PSS/LDH)25, similarly to glass fibres. Hence, the manufacture and testing of small composites made of hydroxylated glass and oxidised carbon fibres coated with PDDA/(PSS/LDH)25 is of great interest. The impact of the nanostructured interphase on the tensile properties of unidirectional hierarchical composites will be described in the following chapter.

198

Hierarchical Composites Containing a

Nanostructured Interphase

Conventional FRP composites suffer from the formation of critical clusters of correlated fibre breaks, leading to high local stress concentrations and subsequently sudden composite failure. Here, an optimised anisotropic layered nanostructure inspired by nacre was deposited around hydroxylated glass and oxidised carbon fibres, which were used to manufacture model unidirectional composites. The nanostructured composite interphase should provide mechanisms for plasticity and hindering the formation of critical clusters of broken fibres. The LbL assembly of the nanostructured interface on multiple fibres simultaneously, allowed the manufacture of small bundle composites containing a few hundred fibres. The effect of the dissipation of stress concentrations arising from fibre breaks observed in single fibre composite models, was investigated as a route to avoid the formation of critical clusters and, therefore, for the production of composites with higher tensile strength, strain to failure and toughness. Potential non-linearity in the strain-stress curves

199

(plasticity) of the bundle composites by slippage of uncorrelated fibre fragments was investigated.

7.1 Introduction

Aligned, commercially-sized fibres, modified (hydroxylated or oxidised) and

PDDA/(PSS/LDH)25 coated fibres were impregnated with an epoxy matrix to manufacture small unidirectional fibre bundle composites, using coated glass (Chapter.5) and carbon

(Chapter.6) systems. The effect of the nanostructured composite interphase on the tensile properties of the composite, as compared to composites containing sized commercial fibres, was investigated. First, impregnated fibre tow composites were manufactured and subsequently characterised in terms of fibre volume fraction and morphology. Tensile tests of the composites were performed and monitored using a video gauge and AE along with high speed imaging to investigate the fracture behaviour of the composites.

7.2. Wettability of a nanostructured coatings by the epoxy resin

To successfully impregnate a bundle of coated glass and carbon fibres whilst achieving a high fibre volume fraction in the composite, it is important that the fibres coated with a nanostructured coating are wetted by the epoxy resin. Flat micrometre-thick nanostructured coatings (Chapter.4) with LDH- and PSS-terminated monolayers were used in order to identify, which interface is wetted best by the liquid, uncured epoxy resin, allowing to manufacture of bundle composites. A similar contact angle of the epoxy resin on the

200

surface of the two coatings was observed, about 22° and 23° for the LDH- and

PSS-terminated coatings, respectively, confirming good wettability of the epoxy resin over the coating (Figure 7.1). For a matter of consistency, PDDA/(PSS/LDH)25 was deposited onto glass and carbon fibres prior to epoxy impregnation to use the same interphase system, which was also studied in the previous chapters of this thesis.

A PSS surface B LDH surface

23 22

Figure 7.1: Droplets of an ultra-low viscosity epoxy resin resting on nanostructured coatings terminated by an LDH- and PSS- monolayer. Contact angle of an epoxy droplet formed onto the surface of a nanostructured coating terminated with a PSS PE monolayer (A) and an LDH platelet monolayer (B).

7.3 Preparation of impregnated bundle composites

Bundles consisting of around 400 commercially-sized, hydroxylated and coated glass fibres were resin impregnated by directly dripping a hardener/epoxy mixture onto the bundle held vertically (refer to 3.2.4). Following the LbL deposition on fibre bundles, and because of the self-limiting character of the deposition, all fibres present a uniform coating along the entire length of the immersed bundle in LbL solution/suspension (Figure 7.2.A). After curing, the bundles were cross-sectioned using a scalpel at room temperature and investigated under

201

an SEM to assess the shape of the bundle composite and the degree of resin impregnation

(Figure 7.2.B-C). The bundle composites had a circular cross-section and the fibres were well packed. The bundle composites had a fibre volume fraction varying in the range of about 50 to 60 %. The 400 nm thick nanostructured coating deposited around the fibres could not be resolved when the modified fibres were imaged in the resin matrix. A few cracks and delaminated fibres were observed, likely caused by the stress applied to the coating when cutting the composites using the scalpel.

Figure 7.2: Hierarchical glass fibre bundle composite. SEM image of a PDDA/(PSS/LDH)25 coated glass fibre bundle composite (A) and SEM images of an epoxy impregnated bundle at low and high magnification (B and C, respectively).

Carbon fibre bundle composites were similarly manufactured by direct impregnation of the fibres (about 100 to 300 fibres per bundles) with the ultra-low viscosity epoxy resin and subsequently cross-sectioned and imaged using SEM (Figure 7.3.B-C). Similarly to glass fibre bundles, the LbL assembly method allowed for the deposition of a homogenous coating along the length of the carbon fibres (Figure 7.3.A). Yet again, in order to assess the effect of the nanostructured interphase on the tensile properties of the composites, commercially sized, oxidised and coated carbon fibres were used as controls.

SEM investigation of impregnated bundles revealed high degree of fibre packing in all type of bundle composites, with a volume fraction in the range of 50 to 60 %. The variation in the volume fraction of fibre was not important in the evaluation of the tensile properties of

202

the composites, as the stresses were determined on the cross-sectional area of the fibres

(load-bearing materials). All bundles used to prepare the composites present some fibre twists, leading to local misalignment and direct fibre contact. However, the incorporation the nanostructured coating onto the surface of the fibres should avoid the direct contact between fibres, which can be beneficial for the mechanical performance of the composite.

Figure 7.3: Hierarchical carbon fibre bundle composite. SEM image of a PDDA/(PSS/LDH)25 coated fibre bundle (A) and SEM images of an epoxy impregnated bundle at low and high magnification (B and C, respectively).

7.4 Tensile properties of glass fibre bundle composites

The glass fibre bundle composites were tested in tension to assess whether strength and toughness improved whilst stiffness was maintained. The expected improvements should be achieved by isolation of the fibre fragments, avoiding the formation of critical clusters and extended fibre debonding slippage. In order to follow the fragmentation process, AE were recorded during the tensile tests to track the occurrence of fibre breaks. The test was filmed using a high-speed camera to visualise the fracture types occurring in the composites.

The strain-stress curves of the different bundle composites were overlapped with cumulative acoustic emission event curves (Figure 7.4). Because the same fibres, treated and untreated, were used to produce bundles, all bundle composites have a similar elastic

203

modulus. However, the strain to failure and strength of the bundle composites was found to vary significantly. The highest strain to failure was reached by bundle composites containing commercially sized glass fibres. Bundle composites containing desized glass fibres possessed lower strain to failure and strength, as a result of weaker matrix/fibre interfacial shear strength (no sizing). Even though PDDA/(PSS/LDH)25 coated glass fibres had a higher IFSS than that of bare hydroxylated glass fibres (refer to 5.4), the composites containing such fibres possessed the lowest strain to failure and strength. The early failure of the bundle composites containing coated glass fibres may be a result of the degradation of the

PDDA/(PSS/LDH)25 coating in tension, at a strain between 2.5 and 3 % (see below).

204

4000 Coated Hydroxylated 3000 Sized

2000

1000 Stress (MPa)

0 Coated

40 Hydroxylated 30

0 Sized

Cumulative Acoustic Event (counts)

0 1 2 3 4 5 Strain (%)

Figure 7.4: Results of glass fibre bundle composite tensile tests. Tensile strain-stress curves synchronised with the cumulative distribution of acoustic emission events occurring in the bundle composites containing sized, hydroxylated and PDDA/(PSS/LDH)25 coated glass fibres.

205

The first detectable AE events (recorded in the same environment background) occurred at a strain of about 2.3% for all modified glass fibres used in the different bundle composites. However, depending on the surface modification of fibres, the accumulation of acoustic events followed different trends. Only very few events were detected in the coated fibre bundles failing early, while more acoustic events occurred in composites containing hydroxylated and sized fibres, accumulating until the rupture of the bundle composite at a strain of 3.57 ± 0.12 and 4.10 ± 0.17 %, respectively. In sized fibre bundle composites, exhibiting the highest tensile properties, the highest number of acoustic events occurred. It is likely that the fracture of such composites was triggered by the formation of critical clusters of fibre breaks. It is therefore of interest to observe the type of fractures occurring in the composites to explain why the bundles made of coated fibres fail after the occurrence of a reduced number of fibre breaks.

The use of high-speed camera during tensile tests enabled the acquisition of pictures of the composite at failure (Figure 7.5). Bundle composites made of sized glass fibres exhibit a very localised and brittle fracture without obvious delamination of fibres, as usually observed in typical FRP composites, caused by the formation of a critical cluster of fibre breaks (Figure 7.5.A). On the other hand, bundle composites containing hydroxylated glass fibres exhibit failure with fibre delamination, indicating poor fibre/matrix interaction

(Figure 7.5.B). Glass fibre bundle composites containing fibres coated with

PDDA/(PSS/LDH)25, failed similar to hydroxylated fibre bundle composites, but exhibited even longer fibre delaminations, throughout the entire gauge length of the specimens

(Figure 7.5.C). Even though the IFSS of coated fibres was higher than that of hydroxylated fibres in epoxy, as measured by pull-out and fragmentation tests (refer to 5.4), greater delamination of the fibre/coating interface occurred in composite bundle tests.

206

A B C

(ii) (iii) (i)

500 μm 500 μm 500 μm i ii iii

50 μm 10 μm 10 μm

Figure 7.5: Tensile fracture of glass fibre bundle composites. High speed camera images of glass fibre bundle composites rupture in tension containing sized (A), hydroxylated (B) and coated glass fibres (C). SEM images at low and high magnification of sized, hydroxylated and coated fibre bundle composite fractured surfaces (i, ii and iii, respectively).

SEM images of the fractured specimens confirmed the brittle and localised failure of bundle composites containing sized fibres (Figure 7.5.i) and fibre delamination in hydroxylated fibre bundle composites (Figure 7.5.ii). For the coated fibres composites, long fibre delaminations with smooth and clear surfaces was observed, which confirmed the observed failure of the coating in tension (Figure 7.5.iii) at a strain of 2.5 to 3 %. A composite strain to failure of about 2.7 % was measured for bundles containing coated fibres, which represent a reduction of about 37 % and 24 % compared to composites containing commercially sized glass fibres and hydroxylated control fibres, respectively (Table 7.1). The strength of the bundle composites was also reduced to 2.53 GPa compared to 3.57 and

3.48 GPa for composites containing sized and hydroxylated fibres, respectively.

207

Table 7.1: Summary of tensile properties of glass fibre bundle composites

Strength Strain to failure Type of GF / GPa / % Sized 3.57 ± 0.15 4.10 ± 0.17 Hydroxylated 3.48 ± 0.10 3.57 ± 0.12 Coated 2.53 ± 0.08 2.71 ± 0.09

A reduction in the strain to failure and the tensile strength of the composites containing fibres coated with the nanostructured coating was observed due to limited performance of the coating in tension (limited stain to failure). The introduction of the nanostructured interphase into glass fibre bundle composite did not result in an improved toughness and strength of the materials. Investigation of the tensile properties of the coating is therefore of great interest, as it might help to explain the behaviour of the coating during tensile tests.

7.5 Tensile properties of carbon fibre bundle composites

In contrast to the glass fibre system, the strain to failure of the composites containing the

PDDA/(PSS/LDH)25 coated carbon fibres increased, as compared to the control composites

(Figure 7.6). As expected, all bundle composites had a similar stress-strain behaviour but different accumulation of AE trends. The AE in the carbon fibre bundle composites, tested under the same conditions as glass fibre bundle composites, were found less pronounced from the background. Prior to the final fracture of sized and oxidised fibre bundle composites, the only events detected were associated with high energy release. However, additional acoustic events, prior to final failure, were detected for coated fibre bundles. The latter observation is indicative of a higher number of fibre breaks in the coated fibre bundle composites, as compared to sized and oxidised fibre bundles.

208

5000 Sized Oxidised 4000 Coated

3000

2000

Stress (MPa) 1000

0 Coated

12 Oxidised 8

0 Sized

Cumulative Acoustic Event (counts)

0.0 0.5 1.0 1.5 2.0 2.5 Strain (%)

Figure 7.6: Results of carbon fibre bundle composite tensile tests. Tensile strain-stress curves synchronised with the cumulative distribution of acoustic emission events occurring in the bundle composites containing sized, oxidised and PDDA/(PSS/LDH)25 coated carbon fibres.

209

The fracture behaviour of the different carbon fibre bundle composites in tension was investigated (Figure 7.7). Coated carbon fibre bundle composites exhibit a more staggered fracture (Figure 7.7.C) indicating multiple fracture sites, as compared to sized and oxidised fibre bundles. Bundle composites containing sized and oxidised carbon fibres show localized rupture (Figure 7.7.A-B) similar to that of sized glass fibre bundle composites.

A B C

(i) (iii) 500 μm (ii) iii

(iv)

501 μ0 mμm

500 μm 500 μm 100 μm i ii iv

50 μm 10 μm 50 μm

Figure 7.7: Tensile fracture of carbon fibre bundle composites. High speed camera images of carbon fibre bundle composite rupture in tension containing sized (A), oxidised (B) and oxidised carbon fibres coated with PDDA/(PSS/LDH)25 coating (C). SEM images at low and high magnification of sized and oxidised fibre bundle composite fractured surfaces (i, and ii, respectively) as well as coated composite fractured surfaces (iii and iv, respectively).

210

The fractured composite surfaces, investigated by SEM, confirmed the brittle and localised failure of bundle composites containing sized fibres and oxidised (Figure 7.7.i-ii); some delaminations occurred along the fibres in the case of the bundle composites containing oxidised fibres due to poorer fibre/matrix interaction (no sizing). For the coated fibre bundle composites, multiple fracture sites were observed, such as the partially fractured section containing fibre delaminations (Figure 7.7.iii) and fully fractured brittle section

(Figure 7.7.iv), which may have eventually correlated at the point of final rupture. High magnification SEM images of failed coated fibre bundle composites show fibre delaminations in the partially fractured section characterised by rough and smooth surface regions, which indicates that the coating remained partly adhered to the fibre until fracture.

A strain to failure of about 2.12 % was measured for bundle composites containing coated carbon fibres. An improvement of about 30 and 42 % in the strain to failure was measured for the coated carbon fibre bundle composites as compared to commercially sized and the oxidised fibre bundle composites, respectively (Table 7.2). The tensile strength of the composite containing the PDDA/(PSS/LDH)25 nanostructured interphase was determined to be 4.61 GPa, which represents an improvement of 14 and 42 % as compared to commercially sized and the oxidised carbon fibre bundle composites, respectively.

Table 7.2: Summary of tensile properties of carbon fibre bundle composites

Strength Strain to failure Type of CF / GPa / % Sized 4.05 ± 0.91 1.63 ± 0.26 Unsized 3.19 ± 0.23 1.49 ± 0.25 Coated 4.61 ± 0.37 2.12 ± 0.32

The use of a nanostructured interphase in high performance carbon fibre reinforced composites resulted in significant improvements of the tensile response of these composites.

211

The composites were found to absorb more energy until failure because of the increased in strength and strain to failure while the elastic modulus remained unaffected. Pseudo-isolated fibre fragments may have occurred because of crack deflection (energy absorption) in the interphase, delaying the formation of critical fibre clusters and, therefore, resulting in an increased strength and strain to failure of the composite. However, no non-linearity of the strain-stress curves was observed due to limited plastic deformation of the interphase in shear

(fibre slippage), still restricted by the number of fibre breaks. One can expect, providing a higher density of uncorrelated/pseudo-isolated fibre breaks in the composite (more or large critical clusters), large plasticity to occur through the activation of fibre slippages.

7.6 Summary

Small cylindrical fibre-reinforced bundle composites with nanostructured interphase were manufactured using an ultra-low viscosity epoxy resin cured at room temperature in order to avoid any dehydration of the interphase. Sized fibres as well as oxidised or hydroxylated control, were used to manufacture small composites in order to provide a baseline for the influence of the nanostructured interphase. Good impregnation of the fibres was achieved with a high fibre volume fraction of about 50 to 60 % for both glass and carbon fibre bundle composites.

Tensile tests of glass fibre bundle composites with the use of AE and high speed camera revealed premature degradation of the PDDA/(PSS/LDH)25 coating leading to long delamination of the coated glass fibres at relatively low strain (< 3 %) and after the occurrence of very few detectable fibre failure events. The premature failure of the bundle composites led to a reduced strength and strain to failure, significantly reducing the work of fracture of the glass fibre composite. The strain to failure of the coating, in tension, may not

212

be high enough to ensure its integrity and performance past a strain of about 2.5 to 3 %.

Investigation of the coating properties in tension is required to further understand and overcome the current limitation of the composite system.

Tensile tests of carbon fibre bundle composites demonstrated improved tensile properties for bundle composites containing carbon fibres coated with the

PDDA/(PSS/LDH)25 nanostructured coating. A strain to failure of about 2.12 % was achieved compared to 1.63 % for the bundle composites containing commercially available sized fibres along with a higher strength, leading to a significant improvement in the work of fracture of the composite. The use of a nanostructured interphase may have enabled larger and/or more critical clusters of fibre breaks to form, leading to higher strain the failure and strength. A further increase in the dimension of the critical cluster should allow for further improvement in strength and strain to failure. Plastic deformation might occur in the composite through stable fibre slippage at fibre break sites, but would require the formation of a greater density of isolated fibre breaks in the composite via better isolation of fragments. One can assume that a thicker PDDA/(PSS/LDH)n nanostructured interphase (n > 25) may allow for more energy dissipation along the length of the fibre, reducing fragment interactions.

213

Conclusions and Further Work

8.1 Summary of achievements

Limited by the formation of a critical cluster of fibre breaks and the associated high local stress concentration triggering failure in tension, fibre-reinforced composites suffer from brittle failure at relatively low strain. Engineering the fibre/matrix interface of composites can significantly impact the amount of energy absorbed during the fracture process but usually at the cost of a reduction in the mechanical performance of the material.

Strain hardening in shear at fibre/matrix interfaces along with mechanisms to dissipate and spread stress along the length of the fibres was investigated as a promising approach to improve both the mechanical performance and toughness of the material in tension. Strain hardening in shear of the matrix/fibre interface was expected to allow for stable fibre slippage while crack deflection at the interface would enable energy absorption. The latter mechanisms, observed in natural nacre and occurring via the sliding and subsequent

214

interlocking of its inorganic inclusions, have been reproduced at the nanometre length scale to design a nanostructured coating conformal to the curvature of typical reinforcing fibres.

Therefore, for the first time, a nanostructured composite interphase inspired by the “brick- and-mortar” structure of natural nacre was developed, mimicking some of the toughening mechanism of nacre but at a smaller length scale. The reproduction of the mechanical properties, deformation mechanisms and toughening mechanisms of nacre, in a 0.4 μm thick coating, were found to significantly improve the debonding behaviour of glass and carbon fibres in single fibre composite models. When incorporated in the interphase of small unidirectional continuous bundle composites, an improvement in both strength and strain-to- failure was achieved without compromising the elastic modulus of the material. These improvements were attributed to stress absorption within the coating, arising from fibre break, hindering the formation of critical clusters of fibre breaks responsible for the sudden failure of the whole composite. The different achievements, step by step, in the design and development of such nanostructured composites interphase are described below:

 The design of an anisotropic nanostructured coating inspired by the

“brick-and-mortar” structure of natural nacre was successfully reproduced at the

nanometre length scale while retaining the classic geometry and phase proportions.

The nanomimetics produced by LbL assembly of quasi-monodispersed LDH platelets

with soft PSS PE present mechanical properties similar to those of natural nacre. The

classic deformation behaviours of natural nacre were also reproduced in the synthetic

embodiment, namely, crack deflection and strain hardening in shear, caused by

platelet sliding and subsequent platelet interlocking. In addition, the reproduction of

the “brick-and-mortar” structure at a smaller length scale than that of natural nacre led

to an increase in the absolute interface volume per unit volume, leading to denser

215

platelet sliding/interlocking volume and subsequently allowed for more energy to be

dissipated within the nanostructure as compared to natural nacre.

 The LbL assembly of the nanostructured coating was successfully adapted to coat

both glass fibres and carbon fibres, after modification to increase their surface charge

density. The surface charge density of glass fibre and carbon fibres was optimised for

LbL deposition via a piranha and oxidation treatment, respectively. The dimensions of

the platelets were carefully selected to allow for a conformal deposition of the coating

with the desired inorganic:organic phase proportion of 90:10 around fibre with a

diameter as small as 6 μm. LbL is usually used to produce planar coatings or free-

standing films, which are non-scalable. However, the self-limiting LbL process can

coat a large surface of materials at once. The time consuming LbL deposition of the

nanostructured coating was accelerated by parallel deposition on multiple fibres,

similarly to current fibre sizing deposition methods, offering a promising route to

manufacture hierarchical composites. The toughening mechanisms occurring in

planar, nanostructured coatings were also found to take place on the surface of the

fibres via single fibre composite mechanical tests. As a result, an improvement in

IFSS and DLR of the fibres embedded in an epoxy matrix was measured when coated

with a 0.4 μm thick PDDA/(PSS/LDH)25 nanostructured coating. In-situ

fragmentation of the coated fibres showed that a nanostructured interphase enabled to

dissipate and spread the stress concentration arising from fibre fragments, while

ensuring high stress transfer.

 Small hierarchical bundle composites were manufactured, using the optimal 0.4 μm

thick PDDA/(PSS/LDH)25 nanostructured coating, and tested in tension. While

216

premature failure of the coated glass fibre bundles occurred due to tensile failure and

subsequent delamination of the coating at strain of about 2.5 to 3 %, coated carbon

fibre bundles however showed very promising results with a significant increase in

the energy absorbed to fracture the composite. Higher strain to failure (below the

critical tensile strain of the coating) and strength was achieved along with an increase

in the number of detected acoustic events, evidencing of a greater number of fibre

breaks occurring in the composite prior to final failure. It is therefore likely that the

formation of the critical cluster was delayed.

Improvement in the tensile response of small carbon fibre composites was achieved by developing a nacre-nanomimetic coating exhibiting some of nacre in-plane toughening mechanisms and, therefore, implementing these mechanisms at the fibre/matrix interface. The resulting coating, when loaded in shear, was showed to provide toughening mechanisms enabling stress dissipation and strain hardening. Although the structure of nacre is well- known for performing in shear, its strain-to-failure of about 1.5 % in tension, which should also be observed in the nanomimic, is currently limiting the application of the nanostructured interphase concept in unidirectional continuous composites.

8.2 Implications

The work and achievement presented in the thesis shed light on the impact of a nanostructured interphase providing a potential mechanism for enhanced mechanical performance of FRP composite materials. Toughening mechanisms occurring in biomaterials were reproduced at the nanometre length scale and transferred to load bearing composite

217

material. Even though various types of interphases have been developed to engineer the interfaces of composite materials over the past few decades, the proposed nanostructured interphase sets a new standard and demonstrates unprecedented benefits for the design of energy absorbing composite interphase. A progressive debonding of the coated fibres through plastic deformation of the interphase in shear while ensuring efficient load transfer, as compared to control uncoated fibres, was achieved with an optimal nanostructured coating thickness of 0.4 μm. A stable debonding length and interfacial shear strength up to twice that of control carbon fibres (bare fibres) were measured. Stress relaxation and plastic deformation in the vicinity of fibre breaks with maintained fibre/matrix adhesion can therefore occur simultaneously.

The deposition of an optimised nanostructured coating on bundles of reinforcing-fibres, with a diameter as low as 6 μm, using the LbL approach developed, offers the possibility to manufacture tows of coated fibres in a continuous manner. One can imagine a simple sequence of dipping bath containing PSS, LDH platelets and rinsing H2O to build up the “brick-and-mortar” nanostructure on carbon fibres that were previously surface oxidised, similar to current sized deposition method. Tuneable nanostructures with a potentially scalable manufacturing process, encourages the production of larger scale composite materials for the investigation of their mechanical response in different loading modes, such as impact, fatigue, damping, etc.

8.3 Way forward

The potential of hierarchical composites with a well-defined nanostructured interphase is considerable. The first step to develop the work in this thesis will be to confirm

218

the presumed delayed critical cluster by direct imaging of the fibre breaks, for example, via

X-ray synchrotron microtomography33,35 during an in-situ tensile test or by observing the fragments after chemically removing the epoxy resin post-test. In addition, since the tensile properties of the coating seem to be limiting the performance of the current carbon fibre hierarchical composite in tension, it is of great interest to quantify and, therefore improve these properties by tuning the interactions between the platelets and the organic binder of the nanomimetic coating.

No non-linearity in the stress-strain curves was detected suggesting that the dimensions of the critical cluster in the current composite architecture still limits the number of fibre breaks in the material, which restricts the number of stable fibre slippage sites. For macroscopic plasticity, fragmentation must reach a certain length density such that all fibres slide. To increase the size of the critical cluster, the interphase must be optimised fully in terms of thickness, tensile properties and adhesion to fibre. It has been demonstrated that an improvement in the interfacial adhesion of the coating to the fibre can lead to better coating performance while loaded in shear (refer to Chapter.6 and Appendix.7). Since the interface between the fibre and the coating is limiting the current fibre/coating/matrix system, as evidenced by SEM images of pulled-out fibres, further improvement could be achieved by covalently bonding the coating to the surface of the fibres.

Alternatively, the use of short fibres291,292 decorated with the optimised

PDDA/(PSS/LDH)25 nanostructured coating should promote stable sliding of the fibre in an epoxy matrix without requiring fragmentation, allowing plasticity in a high performance discontinuous composite. The implementation of toughening mechanisms, inspired by

“brick-and-mortar” the structure of natural nacre such as crack deflection and platelet sliding/interlocking, in the interphase of continuous unidirectional fibre-reinforced composite, has been beneficial in term of strength and strain-to-failure. Nevertheless, one

219

could think of adapting additional toughening features observed in natural nacre into the interphase of fibre-reinforced composites, such as platelet surface roughness, mineral bridging or nanograin rotation, to further improve the performance of the composite.

The coating is expected to behave differently at low and high temperature as a result of a change in the properties of the PSS PE matrix, as well as the proportion of moisture contained in the coating. The behaviour of the hierarchical composite in tension at different temperature and relative humidity should be studied. An alternative matrix, not moisture sensitive, should be considered instead of PSS PE.

Modification of the LbL dipping system and procedure should be undertaken to scale up the process and manufacture larger specimens in order to test the composites in various modes. Since the nanostructured interphase can significantly increase the ability of the composite to absorb and spread stress concentrations in shear, hierarchical composite specimens should present improvements in terms of fatigue, resistance to impact and damping. On the other, due to the anisotropic aspect of the “brick-and-mortar” coating, no improvement in compression is expected to be achieved. Similarly, providing the formation of enough fibre fragment in the hierarchical composite in tension and, therefore, the activation of a large number of fibre slippage sites leading to non-linearity, the composite architecture could also be a suitable candidate to improve the notch sensitivity of composite as high pseudo-ductility was shown to reduce the stress concentration in open-hole specimens.293

The concept of a “brick-and-mortar” coating could also be applied to laminate architecture, by the incorporation of “brick-and-mortar” interleave layers in between carbon fibre plies, to dissipate energy arising from the fracture of individual plies, while ensuring for good load transfer between the load-bearing elements. The latter concept would probably

220

require the assembly of bigger platelets while using the same LbL method by direct dipping of plies in LbL solutions. Similarly, the addition of the “brick-and-mortar” coating could also be beneficial for z-pins to improve their debonding behaviour and avoid their delamination during composite curing cycle, which is known to occur through shrinkage of the host laminate.294

221

Bibliography

1 Barthelat, F., Tang, H., Zavattieri, P., Li, C. & Espinosa, H. On the mechanics of mother-of- pearl: A key feature in the material hierarchical structure. Journal of the Mechanics and Physics of Solids 55, 306-337, (2007). 2 Wang, R. Z., Wen, H. B., Cui, F. Z., Zhang, H. B. & Li, H. D. Observations of damage morphologies in nacre deformation and fracture. Journal of Materials Science 30, 2299-2304 (1995). 3 Katti, K. S., Katti, D. R., Pradhan, S. M. & Bhosle, A. Platelet interlocks are the key to toughness and strength in nacre. Journal of Materials Research 20, 1097-1100, (2011). 4 Fleischli, F. D., Dietiker, M., Borgia, C. & Spolenak, R. The influence of internal length scales on mechanical properties in natural nanocomposites: a comparative study on inner layers of seashells. Acta biomaterialia 4, 1694-1706, (2008). 5 Bekah, S., Rabiei, R. & Barthelat, F. Structure, scaling, and performance of natural micro- and nanocomposites. BioNanoScience 1, 53-61, (2011). 6 Barthelat, F. Designing nacre-like materials for simultaneous stiffness, strength and toughness: Optimum materials, composition, microstructure and size. Journal of the Mechanics and Physics of Solids 73, 22-37, (2014). 7 Yao, H. B., Ge, J., Mao, L. B., Yan, Y. X. & Yu, S. H. 25th anniversary article: Artificial carbonate nanocrystals and layered structural nanocomposites inspired by nacre: synthesis, fabrication and applications. Adv Mater 26, 163-187, (2014). 8 Zhang, C., Mcadams, D. A. & Grunlan, J. C. Nano/micro‐manufacturing of bioinspired materials: a review of methods to mimic natural structures. Advanced Materials (2016). 9 Corni, I. et al. A review of experimental techniques to produce a nacre-like structure. Bioinspiration & biomimetics 7, 031001, (2012). 10 Sarkar, P. & Nicholson, P. S. Electrophoretic deposition (EPD): mechanisms, kinetics, and application to ceramics. Journal of the American Ceramic Society 79, 1987-2002 (1996). 11 Kotov, N. A., Dekany, I. & Fendler, J. H. Layer-by-layer self-assembly of polyelectrolyte- semiconductor nanoparticle composite films. The Journal of Physical Chemistry 99, 13065- 13069 (1995). 12 Ma, R., Osada, M., Hu, L. & Sasaki, T. Self-assembled nanofilm of monodisperse cobalt hydroxide hexagonal platelets: topotactic conversion into oxide and resistive switching. Chemistry of Materials 22, 6341-6346 (2010).

222

13 Wang, J., Cheng, Q. & Tang, Z. Layered nanocomposites inspired by the structure and mechanical properties of nacre. Chemical Society reviews 41, 1111-1129, (2012). 14 Bonderer, L. J., Studart, A. R. & Gauckler, L. J. Bioinspired design and assembly of platelet reinforced polymer films. Science 319, 1069-1073, (2008). 15 Pimenta, S. & Robinson, P. An analytical shear-lag model for composites with ‘brick-and- mortar’ architecture considering non-linear matrix response and failure. Composites Science and Technology 104, 111-124, (2014). 16 Espinosa, H. D., Rim, J. E., Barthelat, F. & Buehler, M. J. Merger of structure and material in nacre and bone – Perspectives on de novo biomimetic materials. Progress in Materials Science 54, 1059-1100, (2009). 17 Matthews, F. L. & Rawlings, R. D. Composite materials: engineering and science. (Elsevier, 1999). 18 Hull, D. & Clyne, T. An introduction to composite materials. (Cambridge university press, 1996). 19 Bahl, O., Shen, Z., Lavin, J. G. & Ross, R. A. Manufacture of carbon fibers. Carbon fibers 3, 1 (1998). 20 Proctor, B. Composite materials I. Fibre reinforced composite materials. An introductory review. Faraday Special Discussions of the Chemical Society 2, 63-76 (1972). 21 Kelly, A. & Tyson, a. W. Tensile properties of fibre-reinforced metals: copper/tungsten and copper/molybdenum. Journal of the Mechanics and Physics of Solids 13, (1965). 22 Feih, S., Wonsyld, K., Minzari, D., Westermann, P. & Lilholt, H. Testing procedure for the single fiber fragmentation test. (2004). 23 Meretz, S., Auersch, W., Marotzke, C., Schulz, E. & Hampe, A. Investigation of morphology- dependent fracture behaviour with the single-fibre pull-out test. Composites Science and Technology 48, 285-290 (1992). 24 Dai, Z. et al. Chemical interaction between carbon fibers and surface sizing. Journal of Applied Polymer Science 124, 2127-2132, (2012). 25 Ho, C., Kin, K., Qian, H. & Bismarck, A. Carbon fiber: surface properties. Wiley Encyclopedia of Composites (2012). 26 Hoecker, F. & Karger‐Kocsis, J. Surface energetics of carbon fibers and its effects on the mechanical performance of CF/EP composites. Journal of applied polymer science 59, 139- 153 (1996). 27 Thomason, J. & Adzima, L. Sizing up the interphase: an insider's guide to the science of sizing. Composites Part A: Applied Science and Manufacturing 32, 313-321 (2001). 28 Weitzsacker, C. L., Bellamy, M. & Sherwood, P. M. Studies of the effect of size on carbon fiber surfaces. Journal of Vacuum Science & Technology A 12, 2392-2397 (1994).

223

29 Zhang, X., Kang, S. & Liu, Z. Latent curing agent modified epoxy sizing agent for high modulus carbon fiber. Open Materials Science Journal 5, 104-108 (2011). 30 Sharma, M. et al. Carbon fiber surfaces and composite interphases. Composites Science and Technology 102, 35-50, (2014). 31 Karger-Kocsis, J., Mahmood, H. & Pegoretti, A. Recent advances in fiber/matrix interphase engineering for polymer composites. Progress in Materials Science 73, 1-43, (2015). 32 Scott, A. E. et al. Accumulation of fibres breaks occuring in an unidirectional carbon/epoxy composite. Comparison between a multi-Scale modelling and high resolution computed tomogarphy. 16th International Conference on Composite Structures ICCM (2011). 33 Aroush, D. et al. A study of fracture of unidirectional composites using in situ high-resolution synchrotron X-ray microtomography. Composites Science and Technology 66, 1348-1353, (2006). 34 Scott, A. E., Mavrogordato, M., Wright, P., Sinclair, I. & Spearing, S. M. In situ fibre fracture measurement in carbon–epoxy laminates using high resolution computed tomography. Composites Science and Technology 71, 1471-1477, (2011). 35 Morton, H., Reed, P., Scott, A. E., Sinclair, I. & Spearing, S. M. In situ tensile fibre failure analysis by synchrotron radiation computed tomography. The 19th International Conference on Composite Materials (2013). 36 Nairn, J. A. A variational mechanics analysis of the stresses around breaks in embedded fibers. Mechanics of Materials 13, 131-154 (1992). 37 Beyerlein, I. J. & Phoenix, S. L. Stress concentrations around multiple fiber breaks in an elastic matrix with local yielding or debonding using quadratic influence superposition. J. Mech. Phys. Solids 44, 1997-2039 (1996). 38 Accorsi, M. L., Pegoretti, A. & Dibenedetto, A. T. Dynamic analysis of fibre breakage in single- and multiple-fibre composites. Journal of Materials science 31, 4181-4187 (1996). 39 Swolfs, Y. et al. Stress concentrations in an impregnated fibre bundle with random fibre packing. Composites Science and Technology 74, 113-120, (2013). 40 Papanicolaou, G. C., Paipetis, S. A. & Theocaris, P. S. The concept of boundary interphase in composite mechanics. Colloid & Polymer Science 256, 625-630 (1978). 41 Labronici, M. & Ishida, H. Toughening composites by fiber coating: a review. Composite Interfaces 2, 199-234, (1994). 42 Rhee, H. & Bell, J. Effects of reactive and non‐reactive fiber coatings upon performance of graphite/epoxy composites. Polymer composites 12, 213-225 (1991). 43 Outwater, J. & Murphy, L. Paper 11C, 24th Annual Technical Conference of Composites. (1969). 44 Kelly, A. Interface effects and the work of fracture of a fibrous composiet. Proc. Roy. Soc- London 319, 95-116 (1970).

224

45 Kim, J.-K., Baillie, C. & Mai, Y.-W. Interfacial debonding and fibre pull-out stresses. Journal of materials science 27, 3143-3154 (1992). 46 Piggott, M. Theoretical estimation of fracture toughness of fibrous composites. Journal of Materials Science 5, 669-675 (1970). 47 Fitz-Randolph, J., Phillips, D., Beaumont, P. & Tetelman, A. The fracture energy and acoustic emission of a boron-epoxy composite. Journal of Materials Science 7, 289-294 (1972). 48 Marom, G. & Arridge, C. Stress concentrations and transverse modes of failure in composites with a soft fibre-matrix interlayer. Materials Science and Engineering 23, 23-32 (1976). 49 Matonis, V. & Small, N. A macroscopic analysis of composites containing layered spherical inclusions. Polymer Engineering & Science 9, 90-99 (1969). 50 Awang Ngah, S. & Taylor, A. C. Toughening performance of glass fibre composites with core–shell rubber and silica nanoparticle modified matrices. Composites Part A: Applied Science and Manufacturing 80, 292-303, (2016). 51 Ozdemir, N. G., Zhang, T., Hadavinia, H., Aspin, I. & Scarpa, F. Glass fibre reinforced polymer composites toughened with acrylonitrile butadiene nanorubber. Composites Part B: Engineering 88, 182-188, (2016). 52 Atkins, A. Intermittent bonding for high toughness/high strength composites. Journal of Materials Science 10, 819-832 (1975). 53 Marston, T., Atkins, A. & Felbeck, D. Interfacial fracture energy and the toughness of composites. Journal of Materials Science 9, 447-455 (1974). 54 Atkins, A. & Mai, Y. Effect of water and ice on strength and fracture toughness of intermittently bonded boron-epoxy composites. Journal of Materials Science 11, 2297-2306 (1976). 55 Kim, J. K. & Mai, Y. W. High strength, high fracture toughness fibre composites with interface control a review. Composite Science and Technology 41, 333-378 (1991). 56 Atkins, A. Imparting strength and toughness to brittle composites. (1974). 57 Cook, J., Gordong, J.E., Evans, C. C. & Marsh, D. M. A mechanism for the control of crack propagation in all-brittle systems. Proc. Roy. Soc-London 282, 508 (1964). 58 Broutman, L. J. & Agarwal, B. D. A theoretical study of the effect of an interfacial layer on the properties of composites. Polymer Engineering & Science 14, 581-588 (1974). 59 Williams, J. H. & Kousiounelos, P. N. Thermoplastic fibre coatings enhance composite strength and toughness. Applied Science 11, 83-88 (1978). 60 Subramanian, R. & Crasto, A. Electrodeposition of a polymer interphase in carbon‐fiber composites. Polymer composites 7, 201-218 (1986). 61 Crasto, A. & Subramanian, R. The influence of the interphase on composite properties: poly (ethylene‐co‐acrylic acid) and poly (methyl vinyl ether‐co‐maleic anhydride) electrodeposited on graphite fibers. Polymer Composites 9, 78-92 (1988).

225

62 Peiffer, D. G. Impact Strength of thick-interlayer composites. Journal of Applied Polymer Science 24, 1451-1455 (1979). 63 Peiffer, D. G. & Nielsen, L. E. Preparation and mechanical properties of thick interlayer composites. Journal of Applied Polymer Science 23, 2253-2264 (1979). 64 Hancox, N. L. & Wells, H. The effect of fibre surface coatings on the mechanical properties of CFRP. Applied Science 10, 9-22 (1977). 65 Evans, A. G. & Zok, F. W. The physics and mechanics of fibre-reinforced brittle matrix composites. Journal of materials Science 29, 3857-3896 (1994). 66 Evans, A. G., Zok, F. W. & Davis, J. The role of interfaces in fiber-reinforced brittle matrix composites. Composites Science and technology 42, 3-24 (1991). 67 Marks, D. & Jones, F. Plasma polymerised coatings for engineered interfaces for enhanced composite performance. Composites Part A: Applied Science and Manufacturing 33, 1293- 1302 (2002). 68 Cech, V. Plasma-polymerized organosilicones as engineered interlayers in glass fiber/polyester composites. Composite Interfaces 14, 321-334 (2007). 69 Bijwe, J. & Sharma, M. Carbon fabric‐reinforced polymer composites and parameters controlling tribological performance. Wear of advanced materials, 1-60 (2011). 70 Tiwari, S., Bijwe, J. & Panier, S. Tribological studies on polyetherimide composites based on carbon fabric with optimized oxidation treatment. Wear 271, 2252-2260 (2011). 71 Jang, J. & Yang, H. The effect of surface treatment on the performance improvement of carbon fiber/polybenzoxazine composites. Journal of materials science 35, 2297-2303 (2000). 72 Gao, S.-L., Mäder, E. & Zhandarov, S. F. Carbon fibers and composites with epoxy resins: topography, fractography and interphases. Carbon 42, 515-529 (2004). 73 Qian, H., Greenhalgh, E. S., Shaffer, M. S. & Bismarck, A. Carbon nanotube-based hierarchical composites: a review. Journal of Materials Chemistry 20, 4751-4762 (2010). 74 Sager, R. et al. Effect of carbon nanotubes on the interfacial shear strength of T650 carbon fiber in an epoxy matrix. Composites Science and Technology 69, 898-904 (2009). 75 Qian, H., Bismarck, A., Greenhalgh, E. S. & Shaffer, M. S. Carbon nanotube grafted silica fibres: characterising the interface at the single fibre level. Composites Science and Technology 70, 393-399 (2010). 76 Kepple, K., Sanborn, G., Lacasse, P., Gruenberg, K. & Ready, W. Improved fracture toughness of carbon fiber composite functionalized with multi walled carbon nanotubes. Carbon 46, 2026-2033 (2008). 77 Qian, H., Bismarck, A., Greenhalgh, E. S. & Shaffer, M. S. Synthesis and characterisation of carbon nanotubes grown on silica fibres by injection CVD. Carbon 48, 277-286 (2010).

226

78 Rafiee, M. A. et al. Fracture and fatigue in graphene nanocomposites. Small 6, 179-183 (2010). 79 Faber, K. T. & Evans, A. G. Crack deflection processes -I. Acta. Metall. 31 (1983). 80 Lubineau, G. & Rahaman, A. A review of strategies for improving the degradation properties of laminated continuous-fiber/epoxy composites with carbon-based nanoreinforcements. Carbon 50, 2377-2395, (2012). 81 Schaefer, J. D., Rodriguez, A. J., Guzman, M. E., Lim, C.-S. & Minaie, B. Effects of electrophoretically deposited carbon nanofibers on the interface of single carbon fibers embedded in epoxy matrix. Carbon 49, 2750-2759, (2011). 82 Qin, W., Vautard, F., Askeland, P., Yu, J. & Drzal, L. T. Incorporation of silicon dioxide nanoparticles at the carbon fiber-epoxy matrix interphase and its effect on composite mechanical properties. Polymer Composites (2015). 83 Qin, W., Vautard, F., Askeland, P., Yu, J. & Drzal, L. Modifying the carbon fiber–epoxy matrix interphase with silicon dioxide nanoparticles. RSC Adv. 5, 2457-2465, (2015). 84 Yang, Y., Lu, C., Su, X. & Wang, X. Effects of emulsion sizing with nano-SiO2 on interfacial properties of carbon fibers/epoxy composites. Journal of Materials Science 42, 6347-6352, (2007). 85 Gao, S. L., Mäder, E. & Plonka, R. Nanostructured coatings of glass fibers: improvement of alkali resistance and mechanical properties. Acta Materialia 55, 1043-1052, d (2007). 86 Tang, G. et al. Study on the interfacial behavior of clay-coated carbon fiber-reinforced PEI composites. Polymer-Plastics Technology and Engineering 51, 861-865, (2012). 87 Jiang, J. et al. Preparation of graphene oxide coatings onto carbon fibers by electrophoretic deposition for enhancing interfacial strength in carbon fiber composites. Journal of The Electrochemical Society 163, D133-D139, (2016). 88 Zhang, X. et al. Interfacial microstructure and properties of carbon fiber composites modified with graphene oxide. ACS applied materials & interfaces 4, 1543-1552, (2012). 89 Park, J. K., Do, I.-H., Askeland, P. & Drzal, L. T. Electrodeposition of exfoliated graphite nanoplatelets onto carbon fibers and properties of their epoxy composites. Composites Science and Technology 68, 1734-1741, (2008). 90 Monfared Zanjani, J. S., Okan, B. S., Menceloglu, Y. Z. & Yildiz, M. Nano-engineered design and manufacturing of high-performance epoxy matrix composites with carbon fiber/selectively integrated graphene as multi-scale reinforcements. RSC Adv. 6, 9495-9506, (2016). 91 Zhang, R. L. et al. Directly grafting graphene oxide onto carbon fiber and the effect on the mechanical properties of carbon fiber composites. Materials & Design 93, 364-369, (2016).

227

92 Deng, C. et al. Influence of graphene oxide coatings on carbon fiber by ultrasonically assisted electrophoretic deposition on its composite interfacial property. Surface and Coatings Technology 272, 176-181, (2015). 93 Chen, L. et al. A design of gradient interphase reinforced by silanized graphene oxide and its effect on carbon fiber/epoxy interface. Materials Chemistry and Physics 145, 186-196, (2014). 94 Qin, W., Vautard, F., Drzal, L. T. & Yu, J. Mechanical and electrical properties of carbon fiber composites with incorporation of graphene nanoplatelets at the fiber–matrix interphase. Composites Part B: Engineering 69, 335-341, (2015). 95 Mahmood, H., Tripathi, M., Pugno, N. & Pegoretti, A. Enhancement of interfacial adhesion in glass fiber/epoxy composites by electrophoretic deposition of graphene oxide on glass fibers. Composites Science and Technology 126, 149-157, (2016). 96 Chen, L. et al. Grafting of silane and graphene oxide onto PBO fibers: multifunctional interphase for fiber/polymer matrix composites with simultaneously improved interfacial and atomic oxygen resistant properties. Composites Science and Technology 106, 32-38, (2015). 97 Chen, J. et al. Modifying glass fibers with graphene oxide: towards high-performance polymer composites. Composites Science and Technology 97, 41-45, (2014). 98 Rider, A., An, Q., Thostenson, E. & Brack, N. Ultrasonicated-ozone modification of exfoliated graphite for stable aqueous graphitic nanoplatelet dispersions. Nanotechnology 25, 495607 (2014). 99 Diba, M., Fam, D. W., Boccaccini, A. R. & Shaffer, M. S. Electrophoretic deposition of graphene-related materials: a review of the fundamentals. Progress in Materials Science 82, 83-117 (2016). 100 Liang, H., Bu, Y., Zhang, J., Cao, Z. & Liang, A. Graphene oxide film as solid lubricant. ACS applied materials & interfaces 5, 6369-6375 (2013). 101 Rutz, B. H. & Berg, J. C. Electrostatic deposition of silica nanoparticles between E-glass fibers and an epoxy resin. Journal of Applied Polymer Science 132, 41516, (2015).

102 Dai, G. & Mishnaevsky, L. Fatigue of multiscale composites with secondary nanoplatelet reinforcement: 3D computational analysis. Composites Science and Technology 91, 71-81, (2014). 103 Mishnaevsky, L. Nanostructured interfaces for enhancing mechanical properties of composites: Computational micromechanical studies. Composites Part B: Engineering 68, 75-84, (2015). 104 Naslain, R. Fibre-matrix interphases and interfaces in ceramic matrix composites processed by CVI. Composite Interfaces 1, 253-286 (1993).

228

105 Naslain, R. R. The design of the fibre-matrix interfacial zone in ceramic matrix composites. Composites Part A: Applied Science and Manufacturing 29, 1145-1155 (1998). 106 Naslain, R. R., Pailler, R. J. F. & Lamon, J. L. Single‐and multilayered interphases in SiC/SiC composites exposed to severe environmental conditions: an overview. International Journal of Applied Ceramic Technology 7, 263-275 (2010). 107 Gao, Y., Zhang, L., Wang, X., Chen, J. & Cheng, H. Interfacial characterization of an oxide fiber-reinforced silica matrix composite containing a single layer pyrocarbon interphase. Ceramics International 42, 6504-6509 (2016). 108 Carrère, N., Martin, E. & Lamon, J. The influence of the interphase and associated interfaces on the deflection of matrix cracks in ceramic matrix composites. Composites Part A: Applied Science and Manufacturing 31, 1179-1190 (2000). 109 Sun, J. & Bhushan, B. Hierarchical structure and mechanical properties of nacre: a review. RSC Advances 2, 7617-7632, (2012). 110 Mayer, G. & Sarikaya, M. Rigid biological composite materials: structural examples for biomimetic design. Experimental Mechanics, 395-403 (2002). 111 Barthelat, F., Rim, J. E. & Espinosa, H. A review on the structure and mechanical properties of mollusk shells - perspectives on synthetic biomimetic materials. Applied Scanning Probe Methods 13, 17-44 (2009). 112 Lopez, M. I., Meza Martinez, P. E. & Meyers, M. A. Organic interlamellar layers, mesolayers and mineral nanobridges: contribution to strength in abalone (Haliotis rufescence) nacre. Acta biomaterialia 10, 2056-2064, (2014). 113 Barthelat, F. Biomimetics for next generation materials. Philosophical transactions. Series A, Mathematical, physical, and engineering sciences 365, 2907-2919, (2007). 114 Luz, G. M. & Mano, J. F. Biomimetic design of materials and biomaterials inspired by the structure of nacre. Philosophical transactions. Series A, Mathematical, physical, and engineering sciences 367, 1587-1605, (2009). 115 Meyers, M. A., Chen, P.-Y., Lin, A. Y.-M. & Seki, Y. Biological materials: Structure and mechanical properties. Progress in Materials Science 53, 1-206, (2008). 116 Meyers, M. A., Lin, A. Y., Chen, P. Y. & Muyco, J. Mechanical strength of abalone nacre: role of the soft organic layer. Journal of the mechanical behavior of biomedical materials 1, 76-85, (2008). 117 Bezares, J., Peng, Z., Asaro, R. J. & Zhu, Q. Macromolecular structure and viscoelastic response of the organic framework of nacre in haliotis rufescens: a perspective and overview. Theoret. Appl. Mech. 38, 75–106 (2011). 118 Li, X., Chang, W., Chao, Y., Wang, R. & Chang, M. Nanoscale structure and mechanical characterisation of a natural composite material: the shell red abalone. Nano letters 4, 613- 617 (2004).

229

119 Mayer, G. Rigid biological systems as models for synthetic composites. Science 310, 1144- 1147, (2005). 120 Wang, R. & Gupta, H. S. Deformation and fracture mechanisms of bone and nacre. Annual Review of Materials Research 41, 41-73, (2011). 121 Xu, Z.-H., Yang, Y., Huang, Z. & Li, X. Elastic modulus of biopolymer matrix in nacre measured using coupled atomic force microscopy bending and inverse finite element techniques. Materials Science and Engineering: C 31, 1852-1856, (2011). 122 Bezares, J., Asaro, R. J. & Hawley, M. Macromolecular structure of the organic framework of nacre in Haliotis rufescens: implications for growth and mechanical behavior. Journal of structural biology 163, 61-75, (2008). 123 Sumitomo, T., Kakisawa, H., Owaki, Y. & Kagawa, Y. In situ transmission electron microscopy observation of reversible deformation in nacre organic matrix. Journal of Materials Research 23, 1466-1471, (2011). 124 Song, F., Soh, A. K. & Bai, Y. L. Structural and mechanical properties of the organic matrix layers of nacre. Biomaterials 24, 3623-3631, (2003). 125 Song, F. & Bai, Y. Mineral bridges of nacre and its effects. Acta Mechanica Sinica 17, 251- 257 (2001). 126 Song, F. & Bai, Y. L. Effects of nanostructures on the fracture strength of the interfaces in nacre. J. Master. Res. 18, 1741-1744 (2003). 127 Song, F., Zhang, X. H. & Bai, Y. L. Microstructure and characteristics in the organic matrix layers of nacre. J. Master. Res. 17, 1567-1570 (2002). 128 Lin, A. Y., Chen, P. Y. & Meyers, M. A. The growth of nacre in the abalone shell. Acta biomaterialia 4, 131-138, (2008). 129 Gries, K. et al. Correlation of the orientation of stacked aragonite platelets in nacre and their connection via mineral bridges. Ultramicroscopy 109, 230-236, (2009). 130 Wang, R. Z., Suo, Z., Evans, A. G., Yao, N. & Aksay, I. A. Deformation mechanisms in nacre. J. Master. Res. 16, 9, 2485-2493 (2001). 131 Barthelat, F., Li, C. M., Comi, C. & Espinosa, H. D. Mechanical properties of nacre constituents and their impact on mechanical performance. Journal of Materials Research 21, 1977-1986, (2006). 132 Evans, A. G. et al. Model for the robust mechanical behavior of nacre. Journal of Materials Research 16, 2475-2484 (2001). 133 Li, X., Xu, Z.-H. & Wang, R. In situ observation of nanograin rotation and deformation in nacre. Nano letters 6, 2301-2304 (2006). 134 Huang, Z. & Li, X. Origin of flaw-tolerance in nacre. Scientific reports 3, 1693, (2013). 135 Rousseau, M. et al. Multiscale structure of sheet nacre. Biomaterials 26, 6254-6262, (2005).

230

136 Currey, J. D. Mechanical properties of mother of pearl in tension. Proceedings of the Royal Society B: Biological Sciences 196, 443-463, (1977). 137 Jackson, A. P., Vincent, J. F. V. & Turner, R. M. The mechanical design of nacre. Proc. R. Soc. Lond. B 234, 415-440 (1988). 138 Barthelat, F. & Rabiei, R. Toughness amplification in natural composites. Journal of the Mechanics and Physics of Solids 59, 829-840, (2011). 139 Barthelat, F., Dastjerdi, A. K. & Rabiei, R. An improved failure criterion for biological and engineered staggered composites. Journal of the Royal Society, Interface / the Royal Society 10, 20120849, (2013). 140 Gao, H., Ji, B., Buehler, M. J. & Yao, H. Flaw tolerant bulk and surface nanostructures of biological systems. Tech Science Press 1, 37-52 (2004). 141 Shao, Y., Zhao, H.-P., Feng, X.-Q. & Gao, H. Discontinuous crack-bridging model for fracture toughness analysis of nacre. Journal of the Mechanics and Physics of Solids 60, 1400-1419, (2012). 142 Mayer, G. New toughening concepts for ceramic composites from rigid natural materials. Journal of the mechanical behavior of biomedical materials 4, 670-681, (2011). 143 Gao, H., Ji, B., Jager, I. L., Arzt, E. & Fratzl, P. Materials become insensitive to flaws at nanoscale: lessons from nature. Proceedings of the National Academy of Sciences of the United States of America 100, 5597-5600, (2003). 144 Shao, Y., Zhao, H. P. & Feng, X. Q. On flaw tolerance of nacre: a theoretical study. Journal of the Royal Society, Interface / the Royal Society 11, 20131016, (2014). 145 Barthelat, F. Nacre from mollusk shells: a model for high-performance structural materials. Bioinspiration & biomimetics 5, 1-8, (2010). 146 Espinosa, H. D. et al. Tablet-level origin of toughening in abalone shells and translation to synthetic composite materials. Nature communications 2, 173, (2011). 147 Song, F., Zhou, J., Xu, X., Xu, Y. & Bai, Y. Effect of a negative poisson ratio in the tension of ceramics. Physical Review Letters 100, (2008). 148 Kakisawa, H. & Sumitomo, T. The toughening mechanism of nacre and structural materials inspired by nacre. Science and Technology of Advanced Materials 12, 064710, (2011). 149 Smith, B. L. et al. Molecular mechanistic origin of the toughness of natural adhesives, fibres and composites. Nature 399 (1999). 150 Mohanty, B., Katti, K. S. & Katti, D. R. Experimental investigation of nanomechanics of the mineral-protein interface in nacre. Mechanics Research Communications 35, 17-23, (2008). 151 Barthelat, F. & Espinosa, H. D. An experimental investigation of deformation and fracture of nacre–mother of pearl. Experimental Mechanics 47, 311-324, (2007). 152 Sun, J.-y. & Tong, J. Fracture toughness properties of three different biomaterials measured by nanoindentation. Journal of Bionic Engineering 4, 11-17, (2007).

231

153 Kamat, S., Su, X., Ballarini, R. & Heuer, A. Structural basis for the fracture toughness of the shell of the conch Strombus gigas. Nature 405, 1036-1040 (2000). 154 Tang, H., Barthelat, F. & Espinosa, H. An elasto-viscoplastic interface model for investigating the constitutive behavior of nacre. Journal of the Mechanics and Physics of Solids 55, 1410-1438 (2007). 155 Bouville, F. et al. Strong, tough and stiff bioinspired ceramics from brittle constituents. Nature materials 13, 508-514, (2014). 156 Begley, M. R. et al. Micromechanical models to guide the development of synthetic ‘brick and mortar’ composites. Journal of the Mechanics and Physics of Solids 60, 1545-1560, (2012). 157 Rim, J. E., Zavattieri, P., Juster, A. & Espinosa, H. D. Dimensional analysis and parametric studies for designing artificial nacre. Journal of the mechanical behavior of biomedical materials 4, 190-211, (2011). 158 Podsiadlo, P., Shim, B. S. & Kotov, N. A. Polymer/clay and polymer/carbon nanotube hybrid organic–inorganic multilayered composites made by sequential layering of nanometer scale films. Coordination Chemistry Reviews 253, 2835-2851, (2009). 159 Yao, H. B., Fang, H. Y., Wang, X. H. & Yu, S. H. Hierarchical assembly of micro-/nano- building blocks: bio-inspired rigid structural functional materials. Chemical Society reviews 40, 3764-3785, (2011). 160 Zhang, Y. et al. Graphene-based artificial nacre nanocomposites. Chemical Society reviews, (2016). 161 Cheng, Q., Jiang, L. & Tang, Z. Bioinspired layered materials with superior mechanical performance. Acc Chem Res 47, 1256-1266, (2014). 162 Tang, Z., Kotov, N. A., Magonov, S. & Ozturk, B. Nanostructured artificial nacre. Nature materials 2, 413-418, (2003). 163 Podsiadlo, P. et al. Ultrastrong and stiff layered polymer nanocomposites. Science 318, 80-83, (2007). 164 Li, Y. Q., Yu, T., Yang, T. Y., Zheng, L. X. & Liao, K. Bio-inspired nacre-like composite films based on graphene with superior mechanical, electrical, and biocompatible properties. Adv Mater 24, 3426-3431, (2012). 165 Yao, H. B., Fang, H. Y., Tan, Z. H., Wu, L. H. & Yu, S. H. Biologically inspired, strong, transparent, and functional layered organic-inorganic hybrid films. Angewandte Chemie 49, 2140-2145, (2010). 166 Han, J. et al. Biomimetic design and assembly of organic-inorganic composite films with simultaneously enhanced strength and toughness. Chemical communications 47, 5274-5276, (2011).

232

167 Podsiadlo, P. et al. Exponential growth of LbL films with incorporated inorganic sheets. Nano letters 8, 1762-1770 (2008). 168 Podsiadlo, P., Tang, Z., Shim, P. & Kotov, N. A. Counterintuitive effect of molecular strength and role of molecular rigidity on mechanical properties of Layer-by-Layer assembled nanocomposites. Nano letters 7, 1224-1231 (2007). 169 Wang, J., Qiao, J., Wang, J., Zhu, Y. & Jiang, L. Bioinspired hierarchical alumina-graphene oxide-poly(vinyl alcohol) artificial nacre with optimized strength and toughness. ACS applied materials & interfaces 7, 9281-9286, (2015). 170 Yao, H. B., Tan, Z. H., Fang, H. Y. & Yu, S. H. Artificial nacre-like bionanocomposite films from the self-assembly of chitosan-montmorillonite hybrid building blocks. Angewandte Chemie 49, 10127-10131, (2010). 171 Walther, A. et al. Large-area, lightweight and thick biomimetic composites with superior material properties via fast, economic, and green pathways. Nano Lett 10, 2742-2748, (2010). 172 Walther, A. et al. Supramolecular control of stiffness and strength in lightweight high- performance nacre-mimetic paper with fire-shielding properties. Angewandte Chemie 49, 6448-6453, (2010). 173 Wan, S. et al. Synergistic toughening of graphene oxide/molybdenum disulfide/thermoplastic Polyurethane ternary artificial nacre. ACS Nano 9, 708-714 (2015). 174 Wan, S. et al. Use of Synergistic interactions to fabricate strong, tough, and conductive artificial nacre based on graphene oxide and chitosan. ACS Nano 9, 9830-9836 (2015). 175 Hao, W. et al. Tough and strong nacre-like composites from hyperbranched poly(amido amine) and clay nanosheets cross-linked by genipin. RSC Adv. 6, 1415-1421, (2016). 176 Zhao, H., Yue, Y., Zhang, Y., Li, L. & Guo, L. Ternary artificial nacre reinforced by ultrathin amorphous alumina with exceptional mechanical properties. Adv Mater, (2016). 177 Wu, Y. et al. Synergistic toughening of bioinspired artificial nacre by polystyrene grafted graphene oxide. RSC Adv. 5, 28085-28091, (2015). 178 Lian, M. et al. Gelatin-assisted fabrication of graphene-based nacre with high strength, toughness, and electrical conductivity. Carbon 89, 279-289, (2015). 179 Yan, L. et al. Enhanced multifunctional properties of graphene nanocomposites with nacre- like structures. Advanced Engineering Materials 17, 523-531, (2015). 180 Chen, K., Shi, B., Yue, Y., Qi, J. & Guo, L. Binary synergy strengthening and toughening of bio-inspired nacre-like graphene oxide/sodium alginate composite paper. ACS Nano 9, 8165- 8175 (2015). 181 Long, B., Wang, C., Lin, W., Huang, Y. & Sun, J. Polyacrylamide-clay nacre-like nanocomposites prepared by electrophoretic deposition. Composites Science and Technology 67, 2770-2774, (2007).

233

182 Wang, C.-A., Long, B., Lin, W., Huang, Y. & Sun, J. Poly(amic acid)–clay nacrelike composites prepared by electrophoretic deposition. Journal of Materials Research 23, 1706- 1712, (2011). 183 Lin, W., Wang, C., Le, H., Long, B. & Huang, Y. Special assembly of laminated nanocomposite that mimics nacre. Materials Science and Engineering: C 28, 1031-1037, (2008). 184 Das, P. et al. Nacre-mimetics with synthetic nanoclays up to ultrahigh aspect ratios. Nature communications 6, 5967, (2015). 185 Makiniemi, R. O. et al. Conducting, self-assembled, nacre-mimetic polymer/clay nanocomposites. ACS applied materials & interfaces 7, 15681-15685, (2015). 186 Tan, Z., Zhang, M., Li, C., Yu, S. & Shi, G. A general route to robust nacre-like graphene oxide films. ACS applied materials & interfaces 7, 15010-15016, (2015). 187 Li, X. Q. & Zeng, H. C. Calcium carbonate nanotablets: bridging artificial to natural nacre. Adv Mater 24, 6277-6282, (2012). 188 Cheng, Q., Wu, M., Li, M., Jiang, L. & Tang, Z. Ultratough artificial nacre based on conjugated cross-linked graphene oxide. Angewandte Chemie 52, 3750-3755, (2013). 189 Wang, J., Cheng, Q., Lin, L., Chen, L. & Jiang, L. Understanding the relationship of performance with nanofiller content in the biomimetic layered nanocomposites. Nanoscale 5, 6356-6362, (2013). 190 Cui, W. et al. A Strong integrated strength and toughness artificial nacre based on dopamine cross-linked graphene oxide. ACS Nano 8, 9511-9517, (2014). 191 Wang, J., Cheng, Q., Lin, L. & Jiang, L. Synergistic toughening of bioinspired poly(vinyl alcohol )- clay - nanofibrillar cellulose artificial nacre. ACS Nano 3, 2739-2745 (2014). 192 kotov, N. A. Ordered layered assemblies of nanoparticles. Mrs bull. 26, 992-997 (2001). 193 Decher, G. Fuzzy nanoassemblies: Toward layered polymeric multicomposites. Science 277, 1232-1237, (1997). 194 Fendler, J. H. Preparation and utilization of self-assembled ultrathin films composed of polyelectrolytes, nanoparticles and nanoplatelets. Croatica Chemica Acta 71, 1127-1137 (1998). 195 Burghard, Z. et al. Nanomechanical properties of bioinspired organic–inorganic composite Films. Advanced Materials 19, 970-974, (2007). 196 Kotov, N. A. et al. Mechanism of and defect formation in the self-assembly of polymeric polycation-montmorillonite ultrathin films. J. Am. Chem. Soc. 119, 6821-6832 (1996). 197 Zlotnikov, I., Gotman, I., Burghard, Z., Bill, J. & Gutmanas, E. Y. Synthesis and mechanical behavior of bioinspired ZrO2–organic nacre-like laminar nanocomposites. Colloids and Surfaces A: Physicochemical and Engineering Aspects 361, 138-142, (2010).

234

198 Conga, H. et al. Multilayered CaCO3/block-copolymer materials via amorphous precursor to crystal transformation. Colloids and Surfaces A: Physicochemical and Engineering Aspects 354, 279-283 (2010). 199 Couto, D. S., Alves, N. M. & Mano, J. F. Nanostructured multilayer coatings combining chitosan with bioactive glass nanoparticles. Journal of Nanoscience and Nanotechnology 9, 1741-1748, (2009). 200 Yeom, B., Kim, S., Cho, J., Hahn, J. & Char, K. Effect of interfacial adhesion on the mechanical properties of organic/inorganic hybrid nanolaminates. The Journal of Adhesion 82, 447-468, (2006). 201 Finnemore, A. et al. Biomimetic layer-by-layer assembly of artificial nacre. Nature communications 3, 966, (2012). 202 Burghard, Z., Zini, L., Srot, V., Bellina, P. & Bill, J. Toughening through nature-adapted nanoscale design. Nano letters 9, 4103-4108 (2009). 203 Xu, L. P. et al. Nacre-inspired design of mechanical stable coating with underwater superoleophobicity. ACS Nano 7, 5077-5083 (2013). 204 Vertlib, V. et al. Fast assembly of bio-inspired nanocomposite films. Journal of Materials Research 23, 1026-1035, (2011). 205 Podsiadlo, P. et al. Layer-by-Layer assembly of nacre-like nanostructured composites with antimicrobial properties. Langmuir: the ACS journal of surfaces and colloids 21, 11915- 11921 (2005). 206 Wei, H., Ma, N., Shi, F., Wang, Z. & Wang, X. Artificial nacre by alternating preparation of Layer-by-Layer polymer films and CaCO3 Strata. Chem. Mater. 19, 1974-1978, (2007). 207 Podsiadlo, P., Liu, Z., Paterson, D., Messersmith, P. B. & Kotov, N. A. Fusion of seashell nacre and marine bioadhesive analogs: high-strength nanocomposite by Layer-by-Layer assembly of clay and L-3,4-dihydroxyphenylalanine polymer. Advanced Materials 19, 949- 955, (2007). 208 Podsiadlo, P. et al. Can nature’s design be improved upon? High strength, transparent nacre- like nanocomposites with double network of Ssacrificial cross links. J. Phys. Chem. B 112, 14359–14363 (2008). 209 Woltersdorf, J., Pippel, E., Bonderer, L. J. & Studart, A. R. Strong and ductile platelet reinforced polymer films inspired by nature: microstructure and mechanical properties. 210 Mirkhalaf, M. & Barthelat, F. Nacre-like materials using a simple doctor blading technique: fabrication, testing and modeling. Journal of the mechanical behavior of biomedical materials 56, 23-33, (2015). 211 Shu, Y., Yin, P., Liang, B., Wang, H. & Guo, L. Bioinspired design and assembly of layered double hydroxide/poly(vinyl alcohol) film with high mechanical performance. ACS applied materials & interfaces 6, 15154-15161, (2014).

235

212 Das, P. et al. Facile access to large-scale, self-assembled, nacre-inspired, high-performance materials with tunable nanoscale periodicities. ACS applied materials & interfaces 5, 3738- 3747, (2013). 213 Wu, Q. et al. Facile and universal superhydrophobic modification to fabricate waterborne, multifunctional nacre-mimetic films with excellent stability. ACS applied materials & interfaces 6, 20597-20602, (2014). 214 Zhu, B. et al. Hierarchical Nacre mimetics with synergistic mechanical properties by control of molecular interactions in self-healing polymers. Angewandte Chemie International Edition 54, 8653-8657, (2015). 215 Verho, T. et al. Hydration and dynamic state of nanoconfined polymer layers govern toughness in nacre-mimetic nanocomposites. Adv Mater 25, 5055-5059, (2013). 216 Patro, T. U. & Wagner, H. D. Layer-by-layer assembled PVA/Laponite multilayer free- standing films and their mechanical and thermal properties. Nanotechnology 22, 455706, (2011). 217 Sehaqui, H., Kochumalayil, J., Liu, A., Zimmermann, T. & Berglund, L. A. Multifunctional nanoclay hybrids of high toughness, thermal, and barrier performances. ACS applied materials & interfaces 5, 7613-7620, (2013). 218 Wang, S.-s. et al. Nacre-inspired green artificial bionanocomposite films from the layerby- layer assembly of montmorillonite and chitosan. Chinese Journal of Polymer Science 32, 675- 680, (2014). 219 Yao, H.-B. et al. Gold Nanoparticle functionalized artificial nacre: facile in situ growth of nanoparticles on montmorillonite nanosheets, self-assembly, and their multiple properties. ASC Nano 6, 8250-8260 (2012). 220 Segad, M., Jönsson, B. & Cabane, B. Tactoid formation in montmorillonite. The Journal of Physical Chemistry C 116, 25425-25433, (2012). 221 Carosio, F., Cuttica, F., Medina, L. & Berglund, L. A. Clay nanopaper as multifunctional brick and mortar fire protection coating—Wood case study. Materials & Design 93, 357-363, (2016). 222 Zhu, B., Noack, M., Merindol, R., Barner-Kowollik, C. & Walther, A. Light-adaptive supramolecular nacre-mimetic nanocomposites. Nano Lett 16, 5176-5182, (2016). 223 Xiong, D.-B. et al. Graphene-and-copper artificial nacre fabricated by a preform impregnation process: Bioinspired strategy for strengthening-toughening of metal matrix composite. ACS Nano 9, 6934-6943 (2015). 224 Cheng, Q., Duan, J., Zhang, Q. & Jiang, L. Learning from nature: constructing integrated graphene-based artificial nacre. ACS Nano 9, 2231-2234 (2015).

236

225 Zhao, X. et al. Alternate multilayer films of poly(vinyl alcohol) and exfoliated graphene oxide fabricated via a facial Layer-by-Layer assembly. Macromolecules 43, 9411-9416, (2010). 226 Putz, K. W., Compton, O. C., Palmeri, M. J., Nguyen, S. T. & Brinson, L. C. High-nanofiller- content graphene oxide-polymer nanocomposites via vacuum-assisted self-assembly. Advanced Functional Materials 20, 3322-3329, (2010). 227 Hu, X. et al. A novel wet-spinning method of manufacturing continuous bio-inspired composites based on graphene oxide and sodium alginate. Nano Research, (2016). 228 Shu, Y. et al. Bioinspired nacre-like heparin/layered double hydroxide film with superior mechanical, fire-shielding, and UV-blocking properties. Industrial & Engineering Chemistry Research 53, 3820-3826, (2014). 229 Das, P. & Walther, A. Ionic supramolecular bonds preserve mechanical properties and enable synergetic performance at high humidity in water-borne, self-assembled nacre-mimetics. Nanoscale 5, 9348-9356, (2013). 230 Hu, K., Gupta, M. K., Kulkarni, D. D. & Tsukruk, V. V. Ultra-robust graphene oxide-silk fibroin nanocomposite membranes. Adv Mater 25, 2301-2307, (2013). 231 Brandt, K., Wolff, M. F., Salikov, V., Heinrich, S. & Schneider, G. A. A novel method for a multi-level hierarchical composite with brick-and-mortar structure. Scientific reports 3, 2322, (2013). 232 Niebel, T. P., Carnelli, D., Binelli, M. R., Libanori, R. & Studart, A. R. Hierarchically roughened microplatelets enhance the strength and ductility of nacre-inspired composites. Journal of the mechanical behavior of biomedical materials 60, 367-377, (2016). 233 Xia, S. et al. Nanoasperity: structure origin of nacre-inspired nanocomposites. ACS Nano 9, 2167-2172 (2015). 234 Gong, S. et al. Integrated ternary bioinspired nanocomposites via synergistic toughening of reduced graphene oxide and double-walled carbon nanotubes. ACS Nano 9, 11568-11573 (2015). 235 Ming, P. et al. Nacre-inspired integrated nanocomposites with fire retardant properties by graphene oxide and montmorillonite. J. Mater. Chem. A 3, 21194-21200, (2015). 236 Duan, J. et al. Bioinspired ternary artificial nacre nanocomposites based on reduced graphene oxide and nanofibrillar cellulose. ACS applied materials & interfaces 8, 10545-10550, (2016). 237 Yan, Y.-X., Yao, H.-B. & Yu, S.-H. Nacre-like ternary hybrid flms with enhanced mechanical properties by interlocked nanofiber design. Advanced Materials Interfaces, (2016). 238 Zhang, J. et al. Multiscale deformations lead to high toughness and circularly polarized emission in helical nacre-like fibres. Nature communications 7, 10701, (2016). 239 Hu, X., Xu, Z., Liu, Z. & Gao, C. Liquid crystal self-templating approach to ultrastrong and tough biomimic composites. Scientific reports 3, 2374, (2013).

237

240 Hu, X., Xu, Z. & Gao, C. Multifunctional, supramolecular, continuous artificial nacre fibres. Scientific reports 2, 767, (2012). 241 Xu, Z. & Gao, C. Graphene in macroscopic order: liquid crystals and wet-spun fibers. Acc Chem Res 47, 1267-1276, (2014). 242 Sen, D. & Buehler, M. J. Structural hierarchies define toughness and defect-tolerance despite simple and mechanically inferior brittle building blocks. Scientific reports 1, 35, (2011). 243 Liu, A. & Berglund, L. Clay Nanopaper composites of nacre-like structure based on montmorrilonite and cellulose nanofibers — Improvements due to chitosan addition. Carbohydrate Polymers 87, (2012). 244 Okamoto, K., Iyi, N. & Sasaki, T. Factors affecting the crystal size of the MgAl-LDH (layered double hydroxide) prepared by using ammonia-releasing reagents. Applied Clay Science 37, 23-31, (2007). 245 Arizaga, G., Satyanarayana, K. & Wypych, F. Layered hydroxide salts: synthesis, properties and potential applications. Solid State Ionics 178, 1143-1162, (2007). 246 Xu, Z. P., Stevenson, G., Lu, C.-Q. & Lu, G. Q. Dispersion and size control of layered double hydroxide nanoparticles in aqueous solutions. J. Phys. Chem. B 110, 16923-16929 (2006). 247 Nalawade, P., Aware, B., Kadam, J. V. & Hirlekar, R. S. Layered double hydroxides: a review. Journal of Scientific & Industrial Research 68, 267-272 (2009). 248 Zhitova, E. S. et al. Crystal chemistry of natural layered double hydroxides. 3. The crystal structure of Mg, Al-disordered quintinite-2H. Mineralogical Magazine 74, 841-848, (2010). 249 Rao, M. M., Reddy, B. R., Jayalakshmi, M., Jaya, V. S. & Sridhar, B. Hydrothermal synthesis of Mg–Al hydrotalcites by urea hydrolysis. Materials Research Bulletin 40, 347-359, (2005). 250 Li, Z. & Ruckenstein, E. Conductive surface via graft polymerization of aniline on a modified glass surface. Synthetic metals 129, 73-83 (2002). 251 Mutua, F. N., Lin, P., Koech, J. K. & Wang, Y. Surface modification of hollow glass microspheres. Materials Sciences and Applications 3, (2012). 252 Pittman, C. U., Jiang, W., He, G. R. & Gardner, S. D. Oxygen plasma and isobutylene plasma treatments of carbon fibers: determination of surface functionality and effects on composite properties. Carbon 36, 25-37 (1998). 253 Bismarck, A., Kumru, M. E. & Springer, J. Influence of oxygen plasma treatment of PAN- based carbon fibers on their electrokinetic and wetting properties. Journal of Colloid and Inteface Science 210, 60-72 (1999). 254 Oyama, H. T. & Wightman, J. P. Surface characterization of PVP-sized and oxygen plasma- treated carbon fibers. 26, 39-55 (1998).

238

255 Garcia, A. B., Cuesta, A., Montes-Moran, M. A., Martinez-Alonso, A. & Tascon, J. M. D. Zeta potential as a tool to characterize plasma oxidation of carbon fibers. Journal of Colloid and Inteface Science 192, 363-367 (1997). 256 Erden, S. et al. Continuous atmospheric plasma oxidation of carbon fibres: influence on the fibre surface and bulk properties and adhesion to polyamide 12. Plasma Chemistry and Plasma Processing 30, 471-487, (2010). 257 Elimelech, M., Chen, W. H. & Waypa, J. J. Measuring the zeta (electrokinetic) potential of reverse osmosis membranes by a streaming potential analyzer. Desalinisation 95, 269-286 (1994). 258 Potential, Z. A complete course in 5 minutes. Zeta-Meter Inc (2012). 259 Clogston, J. D. & Patri, A. K. Zeta potential measurement. Characterization of nanoparticles intended for drug delivery, 63-70 (2011). 260 Jacobasch, H.-J., Bauböck, G. & Schurz, J. Problems and results of zeta-potential measurements on fibers. Colloid and Polymer Science 263, 3-24 (1985). 261 Bragg, W. H. & Bragg, W. L. The reflection of X-rays by crystals. Proceedings of the Royal Society of London. Series A, Containing Papers of a Mathematical and Physical Character 88, 428-438 (1913). 262 Drits, V., Srodon, J. & Eberl, D. XRD measurement of mean crystallite thickness of illite and illite/smectite: reappraisal of the Kubler index and the Scherrer equation. Clays and clay minerals 45, 461-475 (1997). 263 Perkampus, H.-H., Grinter, H.-C. & Threlfall, T. UV-VIS Spectroscopy and its Applications. (Springer, 1992). 264 Williams, D. B. & Carter, C. B. The Transmission electron microscopy, 3-17 (Springer, 1996). 265 Reimer, L. Scanning electron microscopy: physics of image formation and microanalysis (IOP Publishing, 2000). 266 Giannuzzi, L. & Stevie, F. A review of focused ion beam milling techniques for TEM specimen preparation. Micron 30, 197-204 (1999). 267 Bell, D. & Garratt-Reed, A. Energy dispersive X-ray analysis in the electron microscope. Vol. 49 (Garland Science, 2003). 268 Hollander, J. M. & Jolly, W. L. X-ray photoelectron spectroscopy. Accounts of chemical research 3, 193-200 (1970). 269 Rugar, D. & Hansma, P. Atomic force microscopy. Physics today 43, 23-30 (1990). 270 Haines, P. J. Principles of thermal analysis and calorimetry. (Royal society of chemistry, 2002).

239

271 Oliver, W. C. & Pharr, G. M. An improved technique for determining hardness and elastic modulus using load and displacement sensing indentation experiments. J. Mater. Res. 7, 1564-1583 (1992). 272 Oliver, W. C. & Pharr, G. M. Measurement of hardness and elastic modulus by instrumented indentation: advances in understanding and refinements to methodology. Journal of Materials Research 19, 3-20 (2004). 273 Giannakopoulos, A., Larsson, P.-L. & Vestergaard, R. Analysis of Vickers indentation. International journal of solids and structures 31, 2679-2708 (1994). 274 Ramanathana, T., Bismarck, A., Schulza, E. & Subramanianb, K. The use of a single-fibre pull-out test to investigate the influence of acidic and basic surface groups on carbon fibres on the adhesion to poly(phenylene sulfide) and matrix-morphology-dependent fracture behaviour. Composites Science and Technology 61, 1703-1710 (2001). 275 Shamsuddin, S. R., Ho, K. K. C., Ng, P., Lee, A. F. & Bismarck, A. Synergy of matrix and fibre modification on adhesion between carbon fibres and poly(vinylidene fluoride). Composites Science and Technology 72, 56-64, (2011). 276 Tripathi , D. & Jones, F. R. Single fibre fragmentation test for assessing adhesion in fibre reinforced composites. Journal of Materials Science 33, 1-16 (1998). 277 Ohsawa, T., Nakayama, A., Miwa, M. & Hasegawa, A. Temperature dependence of critical fiber length for glass fiber‐reinforced thermosetting resins. Journal of Applied Polymer Science 22, 3203-3212 (1978). 278 Mousty, C., Kaftan, O., Prevot, V. & Forano, C. Alkaline phosphatase biosensors based on layered double hydroxides matrices: role of LDH composition. Sensors and Actuators B: Chemical 133, 442-448, (2008). 279 Balcomb, B., Singh, M. & Singh, S. Synthesis and characterization of layered double hydroxides and their potential as nonviral gene delivery vehicles. ChemistryOpen 4, 137-145 (2015). 280 Unsworth, N. K. et al. Comparison of dimethyl sulfoxide treated highly conductive poly(3,4- ethylenedioxythiophene):poly(styrenesulfonate) electrodes for use in indium tin oxide-free organic electronic photovoltaic devices. Organic Electronics 15, 2624-2631, (2014). 281 Sun, J. et al. Quasi-static and dynamic nanoindentation of some selected biomaterials. Journal of Bionic Engineering 11, 144-150, (2014). 282 Ostojic, P. & McPherson, R. A review of indentation fracture theory: its development, principles and limitations. International Journal of Fracture 33, 297-312 (1986). 283 Li, X., W., C., Chao, Y. J., Wang, R. & Chang, M. Nanoscale structural and mechanical characterization of a natural nanocomposite material: the shell of red abalone. Nano letters 4, 613-617 (2004).

240

284 Han, J., Dou, Y., Wei, M., Evans, D. G. & Duan, X. Erasable nanoporous antireflection coatings based on the reconstruction effect of layered double hydroxides. Angewandte Chemie 49, 2171-2174, (2010). 285 Dou, Y. et al. Fabrication of MMO–TiO2 one-dimensional photonic crystal and its application as a colorimetric sensor. Journal of Materials Chemistry 22, 14001, (2012). 286 Plank, J., Dai, Z. & Andres, P. R. Preparation and characterization of new Ca–Al– polycarboxylate layered double hydroxides. Materials Letters 60, 3614-3617, (2006). 287 Leroux, F. & Besse, J.-P. Polymer interleaved layered double hydroxide: a new emerging class of nanocomposites. Chemistry of Materials 13, 3507-3515 (2001). 288 Paipetis, A. & Galiotis, C. Effect of fibre sizing on the stress transfer efficiency in carbon/epoxy model composites. Composites Part A: Applied Science and Manufacturing 27, 755-767 (1996). 289 Yue, Z. R., Jiang, W., Wang, L., Gardner, S. D. & Pittman, C. U. Surface characterization of electrochemically oxidized carbon fibers. Carbon 37, 1785-1796, (1999). 290 Xie, Y. & Sherwood, P. M. X-ray photoelectron-spectroscopic studies of carbon fiber surfaces. Part IX: the effect of microwave plasma treatment on carbon fiber surfaces. Applied Spectroscopy 43, 1153-1158 (1989). 291 Wisnom, M. R. Mechanism to create high performance pseudo-ductile composites. IOP Conference Series: Materials Science and Engineering 139, (2016). 292 Yu, H., Longana, M. L., Swolfs, Y., Wisnom, M. R. & Potter, K. D. Hybrid effet of carbon/glass composites as a function of the strength distriubtion of aligned short carbon fibres. Proceedings ECCM-17, 17th European Conference on Composite Materials (2015). 293 Jalalvand, M., Czél, G. & Wisnom, M. R. Reducing the notch sensitivity of quasi-isotropic layups using thin-ply hybrid laminates. American Society of Composites-30th Technical Conference (2015).

294 Mouritz, A. Review of z-pinned composite laminates. Composites Part A: applied science and manufacturing 38, 2383-2397 (2007).

241

Appendix 1: Platelet Dimensions on Fibres

The aspect ratio (s) of inorganic platelets can be increased, up to a value smaller than the critical aspect ratio (sc), in order to optimise the mechanical properties of the coating.

Providing the use of an assembly method allowing for the deposition of a polymer layer with tuneable thickness (dpol), the half polymer layer thickness (∆dpol) to achieve the desired inorganic:organic thickness ratio of 9:1 can therefore be expressed as a function of the platelet width (wp) as below:

wp ∆d = (A1-1) pol 18. s

By combining Equation (1-5) and Equation (A1-1) together, we obtain the following expression of the maximum platelet width as function of the fibre radius rf:

rf wp,max = 2rf ∙ arccos( w ) r + p,max (A1-2) f 18 ∙ s

242

Figure A1.1: Analytical approach to the relation between fibre diameter and maximum platelet width. Wp,max=f(df) for varying aspect ratio obtained from equation (A1-2) (A) with a phase proportion constraint of 90:10 and Wp,max = f(df) obtained from equation (1-5) at a fixed aspect ratio of 10 (B) with the minimum platelet width defined by the smallest polymer molecular layer thickness achievable of about 1.5 nm.

An analytical resolution of equation (A1-2) leads to a direct relation between the diameter of the fibre (df) and the maximum platelet width, while taking into account the phase proportion requirement (90:10), as a function of the aspect ratio (s) of the platelets

(Figure A1.1.A). A decrease in fibre diameter as well as an increase in platelet anisotropy (s) shifts the platelet maximum width towards smaller values. Based on thinnest polymer layer of about 1.5 nm, it is not possible to access the coating structure with the desired phase proportion below a fibre diameter of 6 μm (Figure A1.1.B). The more lightly shaded region in Figure A1.1B highlights the range of platelet dimensions that can be used if thicker polymer layers can be used (Equation (A1-2)). For larger fibre diameters (> 15 μm), assembly methods allowing for the deposition of thicker polymer layers may be particularly useful to enable the use of larger platelet dimensions. In addition, improvements in the intrinsic strength of the platelets (for example using a different composition) would allow improved interphase performance by enabling the use of higher aspect ratio as defined in

243

Equation (1-1), although they may be applicable only to larger fibre diameters (as outlines in

Figure A1.1.A).

Larger fibres relax the maximum size constraint, and may allow larger platelets to be used effectively. However, the window remains narrow as typically reinforcing fibres are rarely larger than tens of microns in diameter.

244

Appendix 2: FIB Sectioning of

Nanostructured Coating

Materials

 LDH-2/(PSS/LDH-2)50 deposited on hydroxylated glass slide (3 x 1 inches, WVR stores).  TEM copper grids (copper 3-post grip, Omniprobe).

Equipment

 Focus ion beam instrument (dual-beam FIB Nanolab 600, Helios).  Sputter coating machine (EMITECH K575X, Peltier Cooled).

Section process

Sample preparation

1. After LbL deposition on glass slide, let the slide dry overnight at room temperature. 2. Cut a small piece (square about 1 cm * 1 cm) off the coated slide using a diamond tip, while avoiding contact with the coated surface of the slide. 3. Attach the specimen on a SEM stub cover with a carbon tape. 4. Pre-coat the specimen with a thin layer of chromium (about 10 nm).

System startup

1. Vent the FIB/SEM system and open the chamber. 2. Mount the specimen onto the sample holder and place it in the chamber. 3. Close the chamber and start the vacuum pump.

245

4. Once the vacuum is reached, use the SEM beam to selection a suitable area of the specimen using a magnification as low as x200 (5 keV, 0.17nA, 300 ns dwell time). 5. Increase magnification and focus on an obvious feature of the specimen. 6. Lock the working distance and change the Z position to 4 mm.

Electron deposited Platinum coating (Figure A2.1)

1. Platinum should be deposited at 0° using the electron beam (horizontal position). 2. Start Platinum (Pt) gun heating. 3. Set the dimensions of the rectangle to be coated to 10 μm (X) * 2 μm (Y) * 5 μm (Z). 4. Change the current beam to a value of 1.4 nA. 5. The deposition takes about 5 minutes. 6. After deposition, change the beam current back to 0.17nA.

Gallium deposited Platinum coating

1. Tilt the stage to 52°. 2. Image the specimen using FIB beam with the following parameters: 30 keV, 1 μs, 93pA. 3. Bring the rectangle to the centre of the image using beam shift. 4. Set the dimensions of the rectangle to be coated to 17 μm (X) * 2.5 μm (Y) * 2 μm (Z). 5. Change the current beam to a value of 0.46 nA. 6. Switch back to SEM imaging to control the deposition.

246

10 μm

Figure A2.1: Electron deposited Platinum coating. SEM image of Platinum coating deposited on PDDA/(PSS/LDH)50 specimen.

Milling (Figure A2.2)

1. Mill two parallel and regular trenches (25 μm * 10 μm * 6 μm) on either side of the platinum strap with a current of 21 nA. 2. Clean the cross section at Z = 3 μm to produce a lamella with uniform thickness at a current of 6.5 nA, while tilting the specimen by an angle of ± 3°. 3. Tilt the specimen to an angle of 7°. 4. Cut the lamella using a current of 2.8 nA at Z = 10 μm.

A B C

10 μm 10 μm 10 μm Figure A2.2: FIB milling. SEM micrographs of the deposited platinum coating onto the specimen surface (A), milling of regular trenched on both sides of the platinum coating (B) and cutting of the lamella (C) with Ga+ beam.

Transfer specimen to TEM grid

1. Adjust the tilt to 0°.

247

2. Centre the region of interest and go to the lowest magnification on both beam (SEM and FIB). 3. Insert OmniProbe needle. 4. Select live imaging on FIB and SEM with the following parameters: 93 pA, 30 kV (FIB) and 0.17 nA, 5 kV (SEM). 5. Ensure contact between the needle and the lamella using X and Y axis with the SEM and X and Z axis with the FIB. 6. Insert gun and deposit Platinum at a current of 93 pA and Z = 0.5 μm. 7. Cut the third section of the lamella at a current of 2.8 nA and Z = 10 μm. 8. Gently raise the needle to lift up the lamella by increasing Z. 9. Enter coordinates of the TEM grid. 10. Ensure contact between the needle and the grid. 11. Weld Platinum at Z = 1 μm and a current of 93 pA. 12. Cut off the needle at a current of 2.8 nA and Z = 10 μm.

20 μm

Figure A2.3: Transfer of specimen to TEM grid. SEM image of specimen attachment to a TEM grid.

Specimen thinning

1. Use continuous SEM images while milling by selecting a beam current about 4 times that of FIB current. 2. Mill both side of the specimen to a thickness of about 1.4 μm using a tilt angle of ± 1.5°, a current of 0.46 nA, Z= 2 μm and 30 kV.

248

3. Mill both side of the specimen to a thickness of about 700 nm using a tilt angle of ± 1.2°, a current of 93 pA, Z= 2 μm and 30 kV. 4. Clean the specimen cross-section using a tilt angle of ± 2.5°, a current of 47 pA, Z = 2-5 μm and 5 kV. 5. Thin the specimen further using a tilt angle of ± 7°, a current of 28 pA, Z= 20 μm and 2 kV.

System shutdown

1. Switch off the two beams. 2. Vent the chamber and remove the sample holder. 3. Close the chamber and start the vacuum pump.

249

Appendix 3: AE Tensile Grip

Tensile grip (main part)

Figure A3.1: Tensile grip. Schematic of the tensile grip for small bundle composite tensile test.

250

Acoustic emission sensor holder

Figure A3.2: Acoustic sensor fitting. Schematic of the AE sensor holder of tensile grip.

251

Appendix 4: Quaternary-Amine

Terminated Fibre Surfaces

Glass fibre surface treatment

In order to produce glass fibres with a high positive surface charge density, the surface of glass fibres were quaternary-amine terminated (Figure A4.1). Following piranha treatment, glass fibres (about 2 g) were dried under vacuum in the presence of P2O5 and transferred to a nitrogen glovebox. N,N-dimethyl-3-(trimethoxysilyl)-1-propanamine (95 %,

Fluorochem) (10 ml) and iodobutane (99 %, Sigma-Aldrich) (15 ml) were separately degassed by freeze-pump-thaw and transferred to the glovebox. The glass fibres were submerged in the neat methoxysilane at room temperature for 48 h before decanting off the residual methoxysilane. Neat iodobutane was added dropwise to the glass fibres until the fibres were submerged and the mixture was gently agitated with a glass rod. After 48 h, the iodoethane was decanted and the fibres were rinsed with excess acetonitrile before drying in ambient conditions.

252

Figure A4.1: Quaternary-amine terminated glass fibre surfaces. Synthesis steps of quaternary-amine terminated glass fibre surface modification.

Surface charge density of treated glass fibres

At pH lower than 5, the ζ-potential of the quaternary-amine terminated glass fibres was measured higher than +60 mV with a maximum of about +75 mV reached at pH 3

(Figure A4.2). The ζ-potential at pH 10 (pH used for the LbL deposition of the nanostructured coating) was measured about –57 mV, similar to hydroxylated glass fibres.

Even though better adhesion of the coating could be achieved at pH 3, in order to maintain the LbL process developed to deposit nanostructured coating on glass surfaces, pH 10 was used for all solutions and suspensions to coat fibres. Therefore, hydroxylated rather than quaternary-amine terminated glass fibres were used, in subsequent experiments, for simplicity.

253

80 60 40 20

0 -20 -potential (mV)  -40 As received (sized) -60 Quaternary-amine terminated 2 4 6 8 10 pH @ [KCl] = 5 mM

Figure A4.2: Surface charge density of quaternary-amine terminated glass fibres. ζ-potential curves of as-received sized and quaternary-amine terminated glass fibres, in 5 mM KCl for pH 3 to 10.

Carbon fibre surface treatment

Similarly to the glass fibres, carbon fibres were also quaternary-amine terminated to increase their surface charge density (Figure A4.3). Unsized carbon fibres (about 2 g) were placed in a two neck 500 mL round bottom flask fitted with a suba seal and reflux condenser, ensuring the fibre rested above the bottom of the flask to facilitate stirring, and were flushed with argon. O-dichlorobenzene (99 %, Sigma-Aldrich) (150 ml) and anhydrous acetonitrile

(99.8 % Sigma-Aldrich) (100 ml) was added to submerge the fibres completely and were sparged with argon for 30 min. N,N-dimethyl-1,4-phenylenediamine (98 %, Sigma-Aldrich)

(0.7 g) was added and the mixture heated to 50°C and stirred until a single dark purple phase was seen. Amyl nitrite (97 %, stabilised, Fischer Scientific UK) (1.34 ml) was added to the solution stirred at 50°C for 24 h followed by further stirring at room temperature for another

254

24 h. The solution was decanted and the fibres were rinsed with excess acetonitrile before a solution of iodoenthane (25 ml) and anhydrous DCM (dried over silica column before use)

(200 ml) was added and stirred at room temperature overnight. The solution was decanted and the fibres were rinsed with excess acetonitrile before drying in ambient conditions.

Figure A4.3: Quaternary-amine terminated carbon fibre surfaces. Synthesis steps of quaternary-terminated amine carbon fibre surface modification.

Surface charge density of treated carbon fibres

Another approach than fibre oxidisation, to improve the charge density of carbon fibres, relies on positively charged surface via quaternary amine-termination surface modification. Unfortunately, the initial acidic surface of the carbon fibres did not allow for a successful change in charge density. Indeed, a maximum positive ζ-potential value of about

255

+6 mV was reached at pH 3 (Figure A4.4), limiting the use of the modified carbon fibres for

LbL deposition.

-20 -potential (mV)  As received (unsized) Quarternary amine-termianted -40 4 6 8 10 pH @ [KCl] = 5 mM

Figure A4.4: Effect of quaternary-terminated amine modified carbon fibres on their surface charge density. ζ-potential curves of as-received and modified carbon fibres from pH 3 to pH 11.

256

Appendix 5: Impact of PDDA Precursor

Layer on Nanostructured Coating Adhesion

Nanostructured coating adhesion

The repeated deposition of (LDH/PSS)n bilayers onto glass fibre bundles initially appeared promising. However, during cross-section SEM sample preparation, the coating failed near the cut section (Figure A5.1.A-B), peeling off the fibre surface, as a result of a brittle failure at the rigid platelet/fibre surface interface. In order to improve the adhesion of the coating onto the fibre, an initial PDDA PE layer was deposited prior to coating deposition. Soft cationic PE layer such as PDDA can directly be deposited onto the negatively-charged surface of hydroxylated glass fibres. With the addition of a PDDA soft interlayer, the delamination of the coating was no longer observed (Figure A5.1.C).

Figure A5.1: Nanostructured coating with and without PDDA precursor layer. SEM cross-section images of glass fibre coated with (LDH/PSS)n coating at low and high magnification (A and B, respectively) and coated with PDDA/(PSS/LDH)n coating (C) – black arrows indicate coating delamination.

257

Nanostructured interphase properties

The direct deposition of the first LDH platelet monolayer on the glass fibre leads to an early delamination of the coating. The consequence of such poor interaction between the coating and the fibre results in limited debonding extension and reduced IFSS (Figure A5.2).

The PDDA/(PSS/LDH)25 and LDH/(PSS/LDH)25 coated glass fibres exhibit a different DLR values. When deposited on top of a PDDA precursor layer, the nanostructured interphase allows for a DLR of about 0.40 as compared to 0.26 for direct contact between the rigid LDH monolayer and the fibre surface. A complying soft anionic PDDA PE layer allowed for better interaction between coating and the fibre surface and, therefore, for better interaction and higher interfacial mechanical loading of the fibre/matrix interface. The deposition of a PDDA precursor layer prior to the (PSS/LDH)n multilayer coating was then required to thoroughly promote and investigate the potential plastic deformation of the interphase in shear during

SFPO test.

A B 0.16 0.5 ) e /l

n=25 d n=25 0.4 0.12 First layer 0.3 Bare 0.08

LDH PDDA 0.2 Load (N) 0.04 0.1

0.00 Debonding lengthratio (l 0.0 0 25 50 75 100 125 150 Bare LDH PDDA Displacement (um) Figure A5.2: Effect of a PDDA precursor layer on the debonding length ratio of coated glass fibre in epoxy. Load-displacement curves of single fibre pull-out (A) and associated debonding length ratio (B) for bare desized glass fibres as well as desized glass fibres coated with LDH/(PSS/LDH)25 and PDDA/(PSS/LDH)25.

258

Appendix 6: Impact of Organic Phase

Proportion on Nanostructured Interphase

Mechanical Properties

Nanostructured interphase properties

The use of “small” platelets (LDH-1) in the coating, which is associated with a high organic content (about 40 wt.%) and, hence, no strain hardening behaviour in shear (refer to

4.5.3), led to a reduction in the DLR (Figure A6.1). When coated with

PDDA/(PSS/LDH-1)12, the fibre slips over a similar length to that of bare fibre. A debonding length ratio of 0.13 and 0.12 was measured for coating and bare fibres, respectively. On the other hand, when well-dimensioned LDH platelets (LDH-2) are used to produce a nanostructured interphase with strain hardening behaviour in shear (organic content of about

10 wt.%), the fibres slip over a much greater length, leading to a DLR of 0.26, twice that of the bare fibres. The latter observation highlights the fact that an interphase providing strain hardening in shear is the key to enable stable slippage of the fibre. Stable fibre slippage is therefore only achieved with the correct nanostructure phase proportions.

259

A B 0.20 ) e

PDDA(PSS/LDH) /l

12 d PDDA/(PSS/LDH) 0.3 12 Bare 0.15 LDH-1 LDH-2 0.2 0.10

Load (N) 0.1 0.05

0.00 Debonding lengthratio (l 0.0 0 20 40 60 80 100 120 140 Bare LDH-1 LDH-2 Displ (um) Figure A6.1: Effect of LDH platelet dimensions on the debonding length ratio of PDDA/(PSS/LDH)12 coated glass fibres. Load-displacement curves of single fibre pull-out tests (A) and associated debonding length ratio (B) of bare desized glass fibres as well as desized glass fibres coated with PDDA/(PSS/LDH)12 containing LDH-1 (“small”-LDH) platelets and LDH-2 (“optimised”-LDH) platelets.

260

Appendix 7: PDDA/(PSS/LDH)n Transfer onto As-Received Carbon Fibres

LDH monolayer deposition

Both commercially-sized and unsized as-received carbon fibres, which possess a slightly negatively charged surface characterised with a ζ-potential of about -20 mV at pH 10

(Figure A7.1), were coated with a LDH monolayer (Figure A7.2) to assess the possibility to assemble PDDA/(PSS/LDH)n multilayer coatings.

0 As received Sized Unsized -10

-20 -potential (mV) 

-30 4 6 8 10 pH @ [KCl] = 5 mM

Figure A7.1: Surface charge density of as-received carbon fibres. ζ-potential curves of as-received sized and unsized carbon fibres from pH 3 to pH 11, in 5 mM KCl.

261

Because sized fibres exhibit a rougher surface than unsized carbon fibres, due to the presence of unevenly spread commercial sizing, an irregular LDH monolayer was deposited with bald regions (Figure A7.2.A-B). On the other hand, the smoother surface of unsized fibres did allow for the deposition of a satisfying LDH monolayer with good coverage and no apparent platelet overlap (Figure A7.2.C-D). Therefore, unsized carbon fibres were therefore investigated for the deposition of (LDH/PSS)n multilayer coatings over a PDDA precursor layer in order to assemble repeatable LDH and PSS monolayers with no overlap of platelet nor bald spots.

A B

10 μm 2 μm

C D

2 μm 2 μm

Figure A7.2: LDH monolayer LbL deposition on sized and unsized as-received carbon fibres. Top surface SEM micrographs of as-received and LDH coated sized carbon fibres (A and B, respectively). Top surface SEM micrographs of as-received and LDH coated unsized carbon fibres (C and D, respectively).

262

(LDH/PSS)n multilayer coating deposition

Nanostructured coatings with varying thickness were deposition on as-received unsized carbon fibres using the adapted LbL deposition procedure on fibre bundles. The morphology of the coating top surface and cross-section was carried out with SEM.

PDDA/(LDH/PSS)n coatings with 12, 25, 50 and 75 (PSS/LDH) bilayers were deposited on bundles of unsized carbon fibres containing a few hundred fibres (Figure A7.3). The morphology of the coating top surface was investigated by SEM, which revealed a good deposition of the coatings, consisting of a number of (PSS/LDH) bilayers as high as 50. Some apparent roughness was observed for thicker coating deposition, such as

PDDA/(PSS/LDH)75. The surface roughness observed for fibres coated with thick nanostructured coatings can be the result of a progressive disorder during LbL asembly caused by a poor interaction with the initial charge density of the carbon fibre surface.

263

A B

2 μm n=12 2 μm n=25 C D

2 μm n=50 2 μm n=75

Figure A7.3: Surface morphology of nanostructured coatings deposited on a bundle of unsized carbon fibres. Top surface SEM micrographs of carbon fibres coated with PDDA/(PSS/LDH)n coating. n=12 (A), n=25 (B), n=50 (C) and n=75 (D).

The thickness of PDDA/(PSS/LDH)n coatings made of up to 25 bilayers was found consistent with the dimensions of the LDH platelets and in line with the previous results obtained from coated glass fibres and glass slides (Figure A7.4.A-B). An uniform coating thickness of about 200 and 400 nm was measured on unsized carbon fibres after the deposition of PDDA/(PSS/LDH)12 and PDDA/(PSS/LDH)25, respectively. Thicker coatings, such as PDDA/(PSS/LDH)50, were found to easily peel off the surface of the carbon

(Figure A7.4.C-D), which evidences a poor interaction between the fibre and the coating. The coating was found partly removed in the vicinity of the cut-section of the fibres, most likely caused by the use of a scalpel. The thickness of the removed coating was roughly estimated at about a micrometer, which is in good agreement with a repeatable LbL deposition of

(LDH/PSS) bilayers.

264

A n=12 B n=25

500 nm 500 nm

C n=50 D n=50

10 μm 2 μm

Figure A7.4: Cross-section of nanostructured coatings deposited on a bundle of unsized carbon fibres. Cross-section SEM micrographs of carbon fibres coated with PDDA/(PSS/LDH)n coating. n=12 (A), n=25 (B) and n=50 (C and D).

Hence, bare unsized carbon fibres as well as unsized carbon fibres coated with 12, 25 and 50 (PSS/LDH) bilayers, over a PDDA precursor layer, were mechanically investigated using single fibre composite models and, more especially, SFPO tests. The morphology of the

PDDA/(PSS/LDH)75 coating deposited onto the fibres was not deemed satisfactory to merit further characterisation.

Interphase properties of model composites

The interfacial properties between the carbon fibres and an epoxy resin cured at room temperature were tested, along with the response of the nanostructured coating in shear, via

SFPO tests. Bare unsized carbon fibres show elastic and short plastic loading segment of their

265

interface with epoxy, followed by sudden debonding and a high level of friction during fibre extraction from the matrix (Figure A7.5.A). In contrast, similarly to coated glass fibres, coated carbon fibres exhibit larger plastic deformation of the interface during loading, which is the result of the deformation of the coating in shear, as well as a low level of friction with the matrix when being extracted. The load-displacement curves of the PDDA/(PSS/LDH)12 and PDDA/(PSS/LDH)75 exhibit very different behaviours. The former presents an elastic loading of the interface to a load comparable to that of bare fibre followed by plastic deformation causing fibre sliding, as desired. On the other hand, the thicker system shows a progressive decrease in the load, due to an unwanted delamination from the fibre surface.

266

A B 0.25 Bare 0.15 n=12 0.20 n=0 n=25 n=50 0.10 0.15

(N)

max 0.10

0.05 F 0.00 0.05

0.15 0.00 n=12 0 1 2 3 4 5 6 0.10 Embedded area (10-3.mm2)

C 0.05 50 40 0.00 0.15 30

n=25 20

0.10 (MPa) IFSS Load, F (N) Load,F

10 0.05 0 Bare n=12 n=25 n=50 0.00 D 0.15

) 0.5 n=50 e /l d 0.10 0.4

0.05 0.3

0.00 0.2 0 25 50 75 100 125 150 0.1

Displacement, S (m) Debonding lengthratio (l 0.0 Bare n=12 n=25 n=50

The IFSS of the different fibre/matrix systems were measured by plotting the maximum load carried by the interface at full fibre debonding as a function of the fibre embedded area in the matrix (Figure A7.5.B). A similar IFSS was measured for the

PDDA/(PSS/LDH)12 coated carbon fibres as compared to bare carbon fibres (Figure A7.5.C).

Further increase in the thickness of the nanostructured coating deposited on the carbon fibres led to a progressive decrease in the IFSS, correlating with the previous SEM observations; the thicker the coating, the weaker the interaction with the carbon fibre surface.

267

The length along with the fibre tends to slide during the debonding phase of the pull-out test appears to increase with the thickness of the coating (Figure A7.5.D). However, the reduction in IFSS might be partially responsible for the high debonding length value.

Indeed, the debonding length seems to be more attributed to a progressive removal of the weakly bonded coating rather than actual platelet sliding/interlocking in shear as the load reached at interfacial failure is very low. Nevertheless, the fibres coated with

PDDA/(PSS/LDH)12, exhibiting a high IFSS, which represent a true improvement with stable slippage of the fibre through plastic deformation of the coating in shear.

The interaction between the coating and the unsized carbon fibre surface does not seem to be strong enough to support the full shearing of the nanostructured interphase and the associated progressive fibre sliding at high mechanical load. Modification of the carbon fibre surface in order to increase their charge density was then required to improve the mechanical properties of the nanostructured interphase via stronger interactions with the deposited nanostructured coating.

268