Aspects of the Fracture Toughness of Carbon Nanotube...

Aspects of the Fracture Toughness of Carbon Nanotube Modified

Epoxy Polymer Composites

Vahid Mirjalili

Doctor of Philosophy

Department of Mechanical Engineering, Facutly of Engineering

McGill University

Montreal, Quebec, Canada

Nov. 25, 2010

A thesis submitted to McGill University in partial fulfillment of the requirements for a

doctoral degree

Copyright 2010 All rights reserved.

DEDICATION

To my family who always supported my dreams

iii

ABSTRACT

Epoxy resins used in fibre reinforced composites exhibit a brittle fracture behaviour,

because they show no sign of damage prior to a catastrophic failure. Rubbery materials

and micro-particles have been added to epoxy resins to improve their fracture toughness,

which reduces strength and elastic properties. In this research, carbon nanotubes (CNTs)

are investigated as a potential toughening agent for epoxy resins and carbon fibre

reinforced composites, which can also enhance strength and elastic properties. More

specifically, the toughening mechanisms of CNTs are investigated theoretically and

experimentally. The effect of aligned and randomly oriented carbon nanotubes (CNTs) on

the fracture toughness of polymers was modelled using Elastic Plastic Fracture Mechanics.

Toughening from CNT pull-out and rupture were considered, depending on the CNTs

critical length. The model was used to identify the effect of CNTs geometrical and

mechanical properties on the fracture toughness of CNT-modified epoxies. The modelling

results showed that a uniform dispersion and alignment of a high volume fraction of CNTs

normal to the crack growth plane would lead to the maximum fracture toughness

enhancement. To achieve a uniform dispersion, the effect of processing on the dispersion

of single walled and multi walled CNTs in epoxy resins was investigated. An instrumented

optical microscope with a hot stage was used to quantify the evolution of the CNT

dispersion during cure. The results showed that the reduction of the resin viscosity at

temperatures greater than 100 °C caused an irreversible re-agglomeration of the CNTs in

the matrix. The dispersion quality was then directly correlated to the fracture toughness

of the modified resin. It was shown that the fine tuning of the ratio of epoxy resin, curing

agent and CNT content was paramount to the improvement of the base resin fracture

toughness. For the epoxy resin (MY0510 from Hexcel), an improvement of 38% was

achieved with 0.3 wt.% of Single Walled CNT (SWNT). Finally, the CNT-modified epoxy

resin was used to manufacture carbon fibre laminates by resin film infusion and prepreg

technologies. The Mode I and Mode II delamination properties of the CNT-modified

composite increased by 140% and 127%, respectively. In contrast, this improvement was

not observed for the base CNT-modified polymers, used to manufacture the composite

laminates. A qualitative analysis of the fractured surface using a Scanning Electron

Microscope revealed a good dispersion in the composites samples, confirming the

importance of processing to harness the full potential of carbon nanotubes for toughening

polymer composites.

iv

RÉSUMÉ

Les résines époxy utilisées dans des composites à renforts fibreux ont en général un

comportement à rupture fragile qui peut conduire à une rupture catastrophique des

composites. Afin d’améliorer leur ténacité à la rupture, des matériaux caoutchouteux et

des microparticules sont ajoutés, au dépend de leurs propriétés mécaniques. Dans cette

recherche, des nanotubes de carbone (CNTs) ont été ajoutés à la résine époxy pour

améliorer sa ténacité. Plus spécifiquement, les mécanismes de résistance à la rupture des

nanotubes de carbone ont été étudiés de façon expérimentale et numérique. Tout

d’abord, l’effet de l’alignement des nanotubes de carbone (aligné ou aléatoire) sur la

résistance à la rupture a été modélisé en utilisant les lois de mécanique de la rupture

élastique et plastique. L’influence de la longueur critique des CNT sur les conditions de

rupture et sur les mécanismes de résistance à la rupture par arrachement des nanotubes à

été considérée. Le modèle développé a été ensuite utilisé pour identifier l’effet des

propriétés géométriques et mécaniques des nanotubes de carbone sur la ténacité à la

rupture des résines époxy modifiées. Les résultats montrent qu’une dispersion uniforme

ainsi qu’une orientation des nanotubes de carbone perpendiculairement à la direction de

propagation de la fissure conduisent à une amélioration de la ténacité de la résine. L’effet

du procédé de fabrication sur la dispersion des nanotubes de carbone à paroi simple et à

parois multiples a été également étudié expérimentalement. Une plaque chauffante

instrumentée avec un microscope optique a été utilisée pour quantifier la dispersion des

CNT pendant la polymérisation de la résine. Les résultats montrent qu’une réduction de la

viscosité de la résine à des températures supérieures à 100ºC cause une ré-agglomération

irréversible des CNT dans la matrice. La qualité de la dispersion a été ensuite corrélée à la

ténacité de la résine modifiée. La détermination d’un ratio optimum entre la résine époxy,

le catalyseur et la concentration de CNT est primordiale pour améliorer la ténacité de base

de la résine. Pour la résine époxy étudiée (MY0510 de Hexcel), une amélioration de 38% a

été obtenue avec 0.3% de CNT à paroi simple. Finalement, la résine modifiée avec les CNT

a été utilisée pour fabriquer des laminés avec des renforts de fibres de carbone par les

procédés d’infusion de résine et de préimprégnés. Les propriétés de délamination du

composite ont été augmentées d’un maximum de 140% (mode I) et 127% (mode II) par

rapport aux propriétés de base du composite. Cette amélioration n’a pas été observée

pour les échantillons de résine modifiée sans renfort. Une analyse qualitative de la surface

de cassure par microscope électronique à balayage (SEM) révèle une bonne dispersion des

CNT dans le composite. Ceci reconfirme l’importance du procédé de fabrication et de la

dispersion afin d’utiliser les nanotubes de carbone au maximum de leur potentiel pour

renforcer les composites à matrice polymère.

v

ACKNOWLEDGMENTS

I am grateful to my research supervisor Prof. Pascal Hubert, Associate Professor in the

Mechanical Engineering Department, for his support and guidance throughout my work.

His inspiring advice and criticism guided this thesis all the way to the end. This work would

have never been done without his great vision, experience, and insight.

I want to express my high gratitude to Dr. Benoit Simard and Dr. Yakienda Martinez-Rubi

from the Steacie Institute for Molecular Studies and Dr. Behnam Ashrafi, and Dr. Andrew

Johnston from the Institute for Aerospace Research of the National Research Council of

Canada in Ottawa for their valuable input and collaboration in the success of this project.

Also, our collaborators at Bombardier Aerospace, Dr. Abdelatif Atarsia, and at Nanoledge

Inc., Dr. Patrice Lucas, greatly contributed to the success of this project.

Many thanks and appreciation goes to my friends at Structures and Composite Materials

Laboratory who have helped, supported and made my time enjoyable during my PhD

studies at McGill University, specially Dr. Mousavand T., Mr. Lallemand M., Mr.

Ramachandramoorthy R., and Mr. Yourdkhani M. for their contribution in different parts

of the experimental work. My sincere gratitude goes to Ms. Khoun L. and Mr. Kratz J. for

their valuable inputs to the thesis. I am also thankful to Profs. Musa Kamal, Francois

Barthelat and Raynald Gauvin for giving me access to their laboratories at McGill

University.

Financial support from Fonds québécois de la recherche sur la nature et les technologies

and the Natural Sciences and Engineering Research Council is greatly appreciated.

Last, but not least, I am sincerely thankful to my family for their incredible support. I am

lucky to have a wonderful, loving, and supportive family; I want to thank them all.

vi

TABLE OF CONTENTS

DEDICATION .................................................................................................................................. ii

TABLE OF CONTENTS ................................................................................................................. vi

LIST OF TABLES .......................................................................................................................... viii

LIST OF FIGURES......................................................................................................................... ix

Chapter 1. Introduction .............................................................................................................. 13

1. 1. Summary ..................................................................................................................... 13

1. 2. Thesis structure ............................................................................................................ 15

Chapter 2. Literature Review ..................................................................................................... 17

2. 1. Summary ..................................................................................................................... 17

2. 2. Brittle polymer toughening ............................................................................................ 17

2. 2. 1. Rubber toughened epoxies .................................................................................... 18

2. 2. 2. Rigid-particle toughened epoxies............................................................................ 20

2. 2. 3. Fracture mechanisms in unfilled and filled epoxies ................................................. 24

2. 3. Carbon nanotubes and their potential as a reinforcement ............................................. 26

2. 3. 1. CNT processing challenges .................................................................................... 27

2. 3. 2. Toughening potentials of CNTs .............................................................................. 30

2. 3. 3. CNT toughening of composites .............................................................................. 34

2. 3. 4. Fracture mechanisms in nanocomposites ............................................................... 36

2. 4. Summary and Thesis Objectives .................................................................................. 37

Chapter 3. Modelling CNT Toughening Mechanisms ................................................................. 40

3. 1. Summary ..................................................................................................................... 40

3. 2. Introduction .................................................................................................................. 40

3. 3. Fracture Toughness Modelling ..................................................................................... 41

3. 3. 1. Bridging Effect of Randomly Oriented CNTs ........................................................... 44

3. 4. Summary and Discussions ........................................................................................... 51

Chapter 4. Fracture Toughness of Carbon Nanotube Reinforced Resins .................................... 53

4. 1. Summary ..................................................................................................................... 53

4. 2. Materials ...................................................................................................................... 53

4. 2. 1. SWNT resin system ............................................................................................... 53

4. 2. 2. MWNT resin system ............................................................................................... 54

4. 3. Experimental Procedures ............................................................................................. 55

4. 3. 1. Fracture toughness specimen dimensions .............................................................. 55

4. 3. 2. Specimen preparation ............................................................................................ 56

4. 3. 3. Fracture toughness measurement test setup .......................................................... 58

4. 3. 4. Hot stage: dispersion analysis ................................................................................ 59

vii

4. 3. 5. Shear stage: dispersion analysis ............................................................................ 60

4. 3. 6. Rheological analysis .............................................................................................. 61

4. 4. Results and Discussions .............................................................................................. 62

4. 4. 1. Hot-stage test results ............................................................................................. 62

4. 4. 2. Fracture toughness test results .............................................................................. 82

4. 5. Correlation between the model and the experimental results ........................................ 93

4. 6. Summary and Discussions ........................................................................................... 94

Chapter 5. Carbon Nanotube Modified Carbon Fibre Composites .............................................. 96

5. 1. Summary ..................................................................................................................... 96

5. 2. Materials ...................................................................................................................... 96

5. 2. 1. SWNT Modified Prepreg (SWNT composites) ........................................................ 96

5. 2. 2. MWNT Modified Resin Film (MWNT composites) ................................................... 97

5. 3. Experimental Procedures ............................................................................................. 97

5. 3. 1. Test Plan ............................................................................................................... 97

5. 3. 2. Specimen dimensions ............................................................................................ 98

5. 3. 3. Mode I Interlaminar Fracture Toughness ................................................................ 98

5. 3. 4. Mode II Interlaminar Fracture Toughness ............................................................... 99

5. 3. 5. Specimen preparation ............................................................................................ 99

5. 3. 6. Mode I interlaminar fracture toughness test and data analysis .............................. 103

5. 3. 7. Mode II interlaminar fracture toughness test and data analysis ............................. 105

5. 3. 8. SEM Image Analysis ............................................................................................ 107

5. 4. Results and Discussions ............................................................................................ 107

5. 4. 1. Resin Characterization ......................................................................................... 107

5. 4. 2. Hybrid Composite Characterization ...................................................................... 111

5. 5. Summary and Conclusions ......................................................................................... 124

Chapter 6. Conclusions and Contribution of the Thesis ............................................................ 126

Chapter 7. References ............................................................................................................ 130

viii

LIST OF TABLES

Table 2-1: Rubber toughened epoxy as bulk resin, adhesive film, and as matrix in carbon fibre

composites [54] ............................................................................................................................ 20

Table 2-2: Summary of the effect of particles on the mechanical properties of an epoxy resin

(experimental data) ...................................................................................................................... 22

Table 2-3: Mechanical properties of CNT, compared to other materials ........................................ 27

Table 2-4: Hybrid effect of silica nano particles in rubber toughened epoxy [98, 149] .................... 35

Table 3-1. Mechanical properties of Carbon fibre and CNTs ......................................................... 48

Table 3-2. Input values to Equation 7 - 9 for Figure 3-6 ................................................................ 48

Table 3-3. Effect of CNT - resin properties on the critical length and Jb, (NE=No Effect) ................ 51

Table 4-1: Summary of different types of SWNT used .................................................................. 54

Table 4-2: Summary of the SWNT + MY0510 formulations and cure cycle used ........................... 54

Table 4-3: Summary of the MWNT + bisphenol-A epoxy formulations ........................................... 55

Table 4-4: Summary of dispersion characterization tests for the SWNT system ............................ 81

Table 4-5: Summary of dispersion characterization tests for the MWNT system (0.3 wt.%) ........... 81

Table 4-6: Summary of the fracture toughness percentage change compared to the base resin.... 85

Table 4-7. MWNT system fracture toughness percentage change compared to the base resin ..... 86

Table 5-1. Specimen dimensions refer to figures .......................................................................... 98

Table 5-2. Percentage change of fracture toughness values in mode I after addition of CNTs ..... 117

Table 5-3. Percentage change of fracture toughness values in mode II after addition of CNTs .... 117

Table 5-4. Summary of fracture toughness improvement ............................................................ 124

ix

LIST OF FIGURES

Figure 1-1. Comparison between the fracture toughness of epoxies (yellow) with Al (gray), [7] .... 14

Figure 1-2. SEM image of CNT bundles bridging a crack opening [8] ............................................ 15

Figure 2-1. Schematic of crack growth in a rubber modified epoxy, including rubber stretching

before rupture [11] ....................................................................................................................... 19

Figure 2-2. Stress intensity factor for different fillers...................................................................... 22

Figure 2-3. Apparatus designed to obtain aligned specimen (a) plan and (b) side views ............... 23

Figure 2-4. Fracture toughness as a function of volume fraction for different alignment conditions

[65] .............................................................................................................................................. 23

Figure 2-5. Schematic of toughening mechanisms in particle filled polymers [12, 79]: 1. Crack

pinning and bowing, 2. Particle bridging, 3. Crack deflection and debonding, 4. Particle yielding

(plastic deformation), 5. Plastic zone at crazing, 6. Micro-cracking ................................................ 24

Figure 2-6. Crack pinning in modified resins [9] ............................................................................ 25

Figure 2-7. Crack deflection due to the existence of short fibres [84] ............................................. 26

Figure 2-8. Functionalization of CNTs [99] .................................................................................... 28

Figure 2-9. Schematic diagram (a) showing a typical calendaring component, [5] ......................... 29

Figure 2-10. Qualitative characterization of the MWNT dispersion [115] ........................................ 30

Figure 2-11. Effect of CNT on the fracture toughness (a) neat epoxy and (b) CNT reinforced [5] ... 31

Figure 2-12. Dispersing CNTs using a three-roll mill (calendaring) technique [5], (a) 50 m, (b) 20

m, (c) 10 m, (d) 5 m................................................................................................................ 32

Figure 2-13. Fracture toughness results; higher fracture toughness for higher gap settings [5] ...... 32

Figure 2-14. Effect of different types of CNT on the fracture toughness [4] .................................... 33

Figure 2-15. Hybrid effect of silica nano particle and rubber toughened epoxy [98] ....................... 36

Figure 2-16. Schematic description of fracture mechanisms of CNTs [4] ....................................... 37

Figure 3-1. Schematic description of CNTs toughening mechanisms [4], and J-integral contour ... 41

Figure 3-2. Possible CNT length distribution along the crack growth path ..................................... 44

Figure 3-3. CNT with an angle with respect to the crack growth plane ........................................ 45

Figure 3-4. f() as function of the angle ° .................................................................................... 47

Figure 3-5. Orientation of a nanotube in 3D space ........................................................................ 47

Figure 3-6. Effect of CNT-bridging on the fracture toughness of brittle resins as a function of the

average length of CNTs for a Single Walled CNT ......................................................................... 50

Figure 3-7. Steps to improve the toughness of brittle polymers by incorporating CNTs.................. 51

Figure 4-1. Dimensions of the fracture toughness specimen ......................................................... 56

Figure 4-2. Casting mould for fracture toughness specimen preparation ....................................... 57

Figure 4-3. Fullam tensile test fixture and the initial crack under optical microscope ...................... 58

Figure 4-4. Linkam Examina hot-stage setup ................................................................................ 60

x

Figure 4-5. Linkam optical shearing system, (a) closed, (b) opened, (c) schematic of shear stage

setup with the sample between the two quartz plates ................................................................... 61

Figure 4-6. (a) The AR 2000 Rheometer with disposable parallel plates installed, (b) Close-up of the

sample between two parallel plates .............................................................................................. 61

Figure 4-7. 0.3% wt. Unfunctionalized Laser SWNT system dispersion analysis – with no hardener

.................................................................................................................................................... 62

Figure 4-8. 0.3% wt. Anionic Laser SWNT system dispersion analysis– with no hardener ............. 63

Figure 4-9. 0.3% wt. Unfunctionalized Plasma SWNT system dispersion analysis – with no hardener

.................................................................................................................................................... 63

Figure 4-10. SWNT dispersion stability analysis for two types of hardener: DDS and Aradur......... 64

Figure 4-11. SWNT system dispersion analysis – 100:49 Resin to DDS ratio ................................ 65

Figure 4-12. SWNT system dispersion analysis – 100: 55 Resin to DDS ratio ............................... 65



Figure 4-15. SWNT system dispersion analysis – pre-heated to dissolve DDS and further mixed for

improved dispersion quality .......................................................................................................... 67

Figure 4-16. Dispersion quality evolution during the cure, MWNT system with TETA hardener ..... 67

Figure 4-17. Dispersion quality evolution during the cure, MWNT system with IPD hardener......... 68

Figure 4-18. Dispersion quality evolution during the cure, MWNT system with IPD/N3 hardener ... 68

Figure 4-19. Dispersion quality evolution during the cure, MWNT system with IPD/TETA hardener 69

Figure 4-20. Image processing steps, RGB to Grey to Black & White, for IPD/N3 system ............. 69

Figure 4-21. Image sequences from the hot-stage test setup for MWNT system with IPD ............. 70

Figure 4-22. Dispersion quantification results for the MWNT system with IPD, Af calculated from Eq.

4-5 ............................................................................................................................................... 71

Figure 4-23. Image sequences from the hot-stage test setup for MWNT system with IPD/N3 ........ 71

Figure 4-24. Dispersion quantification results for the MWNT system with IPD/N3, Af calculated from

Eq. 4-5 ......................................................................................................................................... 72

Figure 4-25. Typical rheology curve for MY0510/DDS/SWNT formulation. From room temperature

ramp (3 °C/min) to 250 °C with control variable of 12 % strain ........................................................... 73

Figure 4-26. Rheology results, comparing DDS vs. Aradur hardener. From room temperature ramp

(3 °C/min) to 140 °C hold up to gelation ......................................................................................... 74

Figure 4-27. MWNT system viscosity profile for TETA and IPD/TETA as hardener ....................... 75

Figure 4-28. MWNT system viscosity profile for IPD and IPD/N3 as hardener ............................... 75

Figure 4-29. CNT dispersion stability analysis using the Linkam shear-stage setup ...................... 77

Figure 4-30. Shear stage test result for MWNT system with no hardener, 5% strain ...................... 78

Figure 4-31. Shear stage test result for MWNT system with IPD, 5% strain ................................... 78

Figure 4-32. Shear stage test result for MWNT system with IPD, 10% strain ................................. 79

xi

Figure 4-33. Shear stage test result for MWNT system with IPD/N3, 5% strain ............................. 79

Figure 4-34. Shear stage test result for MWNT system with IPD/N3, 10% strain ........................... 79

Figure 4-35. Shear stage test result for MWNT system with IPD/TETA, 5% strain ......................... 80

Figure 4-36. Shear stage test result for MWNT system with IPD/TETA, 10% strain ....................... 80

Figure 4-37: Typical load-displacement curve for epoxy resin ....................................................... 82

Figure 4-38. Fracture toughness test results, MY0510 epoxy system with SWNT; Aradur (left), and

DDS (right) ................................................................................................................................... 83

Figure 4-39. Fracture toughness test results for MY0510 / 0.1% Anionic SWNT with different DDS:

MY0510 ratios, cure cycle # 1 ...................................................................................................... 84

Figure 4-40. Fracture toughness test results for MY0510 / 0.1% Anionic SWNT with different DDS:

MY0510 ratios, cure cycle # 2 ...................................................................................................... 84

Figure 4-41. Fracture toughness test results for MY0510 / SWNT system with different Anionic

SWNT wt.% (100:60 DDS ratio), Cure cycle # 1 ........................................................................... 86

Figure 4-42. Fracture toughness test results, bisphenol-A with 0.3% wt. MWNT and different types

of hardener .................................................................................................................................. 87

Figure 4-43. Fracture toughness test results, bisphenol-A with 0.3% wt. MWNT with different

IPD:TETA ratio ............................................................................................................................. 87

Figure 4-44. SEM analysis of the fractured surface of neat polymer (SWNT system) (MY: DDS ratio

100: 60) ....................................................................................................................................... 89

Figure 4-45. SEM analysis of the fractured surface of 0.1% SWNT modified polymer (MY: DDS ratio

100: 60) ....................................................................................................................................... 91

Figure 4-46. SEM analysis of the fractured surface of 0.3% MWNT modified polymer (Hardener

IPD/N3) ........................................................................................................................................ 92

Figure 4-47. The critical strain energy release rate for the results of Figure 4-41 ........................... 93

Figure 4-48. Bridging contribution, model vs. experiment .............................................................. 94

Figure 5-1. DCB specimen ........................................................................................................... 98

Figure 5-2. Mode II specimen ....................................................................................................... 99

Figure 5-3: Stacking procedure for the MWNT system ................................................................ 100

Figure 5-4: Panel size and Teflon insert location......................................................................... 100

Figure 5-5: Lay-up of the panels and bagging ............................................................................. 101

Figure 5-6: Bagging sequence.................................................................................................... 101

Figure 5-7: Cutting pattern for the DCB and ENF samples .......................................................... 102

Figure 5-8. Sample preparation process from a panel to Mode I and II specimens ...................... 102

Figure 5-9. Resin film sample preparation .................................................................................. 103

Figure 5-10. Fixture linking MTS testing system to Mode I DCB specimen tabs .......................... 103

Figure 5-11. Typical load – displacement curve for a Mode I fracture test of the resin film system

(2377-1) ..................................................................................................................................... 104

xii

Figure 5-12. Typical Mode I R-curve for the MWNT composites (2377)....................................... 105

Figure 5-13. Mode II fracture test fixture on MTS Insight setup ................................................... 106

Figure 5-14. Typical Load-displacement curve for NPC and PC Mode II tests ............................. 107

Figure 5-15. Fracture toughness of SWNT modified polymer ...................................................... 108

Figure 5-16. Fracture toughness of MWNT modified resin film .................................................... 108

Figure 5-17. SEM images of the fractured surface of the neat polymer samples (MWNT system) at

different magnifications .............................................................................................................. 109

Figure 5-18. SEM images of the fractured surface of the 2377 and 2378 MWNT system; images (a –

f) are for the 2377 sample (increased magnification from (a) to (f)); images (g – l) are for the 2378

specimen with increased magnification from (g) to (l) .................................................................. 111

Figure 5-19. Load-displacement curves neat and CNT modified DCB samples ........................... 114

Figure 5-20. R-curve values comparing neat vs. CNT modified DCB samples............................. 115

Figure 5-21. Average Mode I initiation and propagation values for neat and CNT modified samples

.................................................................................................................................................. 116

Figure 5-22. Average mode II interlaminar fracture toughness values ......................................... 118

Figure 5-23. SEM images of fractured DCB coupons; a) CNT pull-outs are highlighted by red arrows

and CNT peelings are shown by dotted black arrows; b-e) SEM of fractured mode II ENF coupons

at different magnifications .......................................................................................................... 120

Figure 5-24. SEM analysis of the delaminated surface of neat composite laminates ................... 121

Figure 5-25. SEM analysis of the delaminated surface of 2377 MWNT composite laminates....... 122

Figure 5-26. SEM analysis of the delaminated surface of 2378 MWNT composite laminates at

different magnification (magnified areas are highlighted by red squares) .................................... 123

Figure 6-1. Steps to improve the toughness of brittle polymers by incorporating CNTs................ 127

13

Chapter 1. Introduction

1. 1. Summary

Composite materials are increasingly used in aerospace, automotive and renewable

energy industries. This growth is mainly due to the higher strength-to-weight ratio offered

by composites, when compared to metals. A major component of these laminated fibre

reinforced composites is a polymer matrix that holds the fibres together. The most widely

used polymeric resins are thermoset epoxies, which provide a high modulus, but low

fracture toughness leading to catastrophic failure. Since fibres are mechanically stronger

than the matrix, [1], the matrix fracture toughness is the key material property that

controls damage initiation and growth in composites. As shown in Figure 1-1, fracture

toughness of epoxies is relatively low. To address this issue, there has been extensive

research on the toughening of epoxies using rubbery and/or thermoplastic micro-particles

as a toughening agent; which will be reviewed in Chapter 2. Unfortunately, this technique

has a major disadvantage: other mechanical properties such as modulus and ultimate

strength of polymers are deteriorated when these toughening agents are added.

Recently, studies on toughening of epoxy have incorporated Carbon Nanotubes (CNTs) as

a toughening agent into the epoxy systems. These nano-sized particles have shown

potential for toughness enhancement at low carbon nanotube (CNT) content, by

introducing several toughening mechanisms, such as CNT bridging (Figure 1-2), crack

pinning, and crack deflection. Another important aspect of adding CNTs to polymers is the

enhanced multi-functional properties of the final formulation, such as improved elastic

properties of the polymers, and also improved electrical and thermal conductivities [2].

However, addition of CNTs to polymers introduces new challenges in the processing of

nano-modified polymers. CNTs increase the viscosity of the base polymer, [3], and affect

the processing of these nano-modified polymers. Therefore, understanding the relation

between the processing parameters and final material properties, as well as developing

new processing techniques, are necessary to achieve maximum property enhancement for

CNT modified polymers. Another challenge in the processing of these formulations is the

agglomeration of CNTs due to their high aspect ratio (Length/Diameter>1000), which

deteriorates the dispersion quality. This agglomeration becomes even more problematic

14

during the curing process, because uniform dispersion quality is important for

homogeneous material properties.

In the past decade, the effect of carbon nanotubes as the reinforcement of the matrix in

composite materials has been theoretically and experimentally studied, [4-6]. However, a

detailed investigation of the dispersion evolution during the curing process and its effect

on the final performance of the CNT-modified formulations is still missing in the literature,

and will be addressed in this thesis. The qualitative analysis of the CNTs dispersion reveals

the main drivers of re-agglomeration of CNTs during the curing process.

Figure 1-1. Comparison between the fracture toughness of epoxies (yellow) with Al (gray), [7]

The research methodology is based on a theoretical modelling and then experimental

verification of the model. The goal of the modelling section is to identify toughening

mechanisms that make CNTs a unique toughening agent, i.e. CNT bridging. The model

highlights the governing parameters that maximize the toughening effects of CNTs. In the

experimental section, the fracture toughness values for two types of epoxy polymers are

15

calculated when Single Walled (SWNT) and Multi Walled CNTs (MWNT) are added to the

neat epoxy. The results are compared to the prediction of the model and potential

sources of discrepancy are further discussed. The nano-modified polymer is then used in

composite laminates to characterize their effect on the delamination fracture toughness.

Figure 1-2. SEM image of CNT bundles bridging a crack opening [8]

1. 2. Thesis structure

The thesis begins with a detailed review of literature on epoxy toughening techniques in

Chapter 2. Traditional methods of brittle polymer toughening are reviewed, followed by

the description of the toughening mechanisms that contribute to fracture toughness

improvements. Then, a summary of the literature review will highlight potential research

areas from which the research objectives will be defined.

Chapter 3 will present the CNT bridging model that can be used to identify the toughening

potential of CNTs. The most dominant toughening mechanisms in CNT modified polymers,

i.e. CNT bridging, is modelled as a function of physical and mechanical properties of CNTs.

In Chapter 4, the effect of low CNTs content (<1 wt. %) on toughening of two epoxy

systems is experimentally verified. Two types of CNT, i.e. Single Wall CNT (SWNT) and

Multi Wall CNT (MWNT) will be used for this experimental section. The Single-Edge

Notched Bending specimens will be used to obtain the fracture toughness measurements.

The verification of the source of CNT re-agglomeration during the curing process of the

16

nano-modified polymers will also be presented. The results are then correlated to the

results of fracture toughness tests.

In Chapter 5, the delamination resistance of CNT modified composites will be studied.

Traditionally, structural composite laminates consist of strong fibres and a polymer matrix.

In this chapter, CNTs were added to the polymer matrix, which were then used to

impregnate the fibre mat. This new system (fibre mat + CNTs + polymer) is a hybrid system

that showed major improvement in the delamination fracture toughness.

With Chapter 6, the thesis will conclude by highlighting the key findings of the research

and the novel contribution to this field. Finally, future work is discussed.

17

Chapter 2. Literature Review

2. 1. Summary

In this chapter, different toughening techniques of brittle polymers as well as composite

structures are reviewed. Special attention is given to epoxy as the base resin and to

carbon nanotube (CNT) as the toughening agent.

The literature review begins by studying the traditional approach in toughening brittle

polymers. Different parameters affecting the toughness of the modified resin are then

reviewed. Then, the fracture mechanisms which improve the toughness of modified resins

are explained. After studying micro-scale fracture mechanism, a brief introduction to

carbon nanotubes will follow. The following section reviews literature on the effect of

carbon nanotube as a filler for resins. The manufacturing challenges are then discussed,

and several experiments which studied the fracture toughness of CNT-modified resin are

presented. From these experiments possible nano-scale fracture mechanisms are

summarized. Then CNT reinforcement of composite laminates (hybrid systems) is

reviewed. The chapter concludes by summarizing the state of the art in CNT toughening

and identifying the thesis objective.

2. 2. Brittle polymer toughening

One of the main drawbacks of brittle polymers is their low fracture toughness. Therefore

toughening of brittle polymers has been studied in the past three decades [9-11]. Fibre-

reinforced composites are sensitive to cracks and lose much of their structural properties

when damaged. There are three important damage initiation modes in a laminated

composite, i.e. Matrix cracking, delamination and fibre fracture. The first two modes

depend to a large extent on the properties of the matrix [12].

Matrix cracks initiate in plies having tensile stress applied perpendicular to the fibres, [12,

13]. The interfacial crack growth and subsequent coalescence with cracks in adjacent off-

axis plies lead to the development of delamination. Delamination also initiates from zones

of high interlaminar stresses such as free edges, notches and other geometric

discontinuities. Delamination may also develop during the manufacturing process as a

18

result of incomplete curing, residual stresses or through the introduction of foreign

particles, or as a result of impact damage, [14, 15].

Different toughening techniques were used to improve the delamination properties of

composite laminates [16-23]. Most epoxy resins are brittle and have Mode I fracture

energies of about 80- 300 J/m2, either in bulk [24-27] or in the delamination mode in a

composite [28-31], or as an adhesive, [32-34]. This low fracture toughness values seriously

limits full potential of weight reductions offered by composites [35].

There are two main solutions to the problem of low fracture toughness of brittle

polymers, [12]: 1. use of thermoplastic resins instead of thermosetting systems which

provides up to an order of magnitude higher fracture toughness values [24, 36], 2. modify

the brittle thermosetting polymer by adding rubber, or inorganic micro-particles [16, 17,

37-43]. While the former provides a very tough system, the manufacturing process of the

thermoplastic resins is very challenging and expensive. Hence, modifying thermoset resins

by adding rubber or inorganic micro-particles becomes a more attractive alternative,

because of the easier processing of such system [12].

In unmodified epoxies, the mechanical properties of a cured part is a function of the

curing agent as well as the curing process [44]. An important aspect of the curing process

is how it affects the cross-linking between the epoxy molecules and the reactive groups on

each end of the curing agent. The density of the cross-linking directly affects the resin

properties. Lower cross-linking density improves the fracture toughness by allowing

elongation before rupture of epoxy network, whereas higher density of epoxy group

cross-links increase the Tg but lowers elongation to failure [12]. These effects need to be

studied in more detail while toughening agents are added to improve the mechanical

properties of the epoxy.

The rubber toughened epoxies (2-4 kJ/m2) are tougher than particle filled systems (0.5 - 1

kJ/m2), but, on the downside elastic properties as well as glass transition temperatures are

reduced [12]. These disadvantages open up the opportunity to explore novel nano-

particles, such as Carbon Nanotubes, which showed potential to improve not only elastic

and thermal properties of the resin but also its fracture toughness [4, 5, 45-48].

2. 2. 1. Rubber toughened epoxies

Addition of rubber to improve the fracture toughness of epoxy was initiated by the work

of McGarry in 1970 [49]. There have been many studies on the experimental and

19

analytical understanding of the effect of rubber on toughness of epoxies since then [11,

23, 50-56]. A short summary of some of the most interesting research is given here.

Kunz-Douglass modelled the effect of rubbery particles on the fracture toughness using an

energy-based approach and showed that rubbery particles were stretched as the crack

opens and failed by tearing at a critical elongation length. This fracture mechanism was

the basis of their analytical toughening model [11].

Figure 2-1. Schematic of crack growth in a rubber modified epoxy, including rubber stretching before rupture [11]

The type of rubber used as the toughening agent should meet two criteria [57]. First, the

compatibility of the rubber with the epoxy group, and second the dispersibility of the

rubber to produce a uniformly dispersed solution. The carboxyl terminated butadiene-

acrylonitrile (CTBN) is among the best candidates [57]. As reported [25], there is an

optimum volume fraction of rubber above which the rubber act as the dominant part and

the strength of the epoxy deteriorated dramatically. Kinloch et al. [18] showed that adding

15 wt% of Carboxyl Terminated Acrylonitrile Butadiene Rubber (CTBN) increased the

fracture toughness by a factor of four, with a small decrease in the modulus. The

maximum fracture energy of rubber-modified epoxy is approximately 30 times that of the

unmodified epoxy [12, 25].

Scott et al. showed that the fracture toughness of the modified polymer is dependent on

geometry of the specimen, i.e. bulk properties vs. adhesive or matrix material in

composite laminate, [54]. The summary of their result is shown in Table 2-1. The table

shows that the initiation fracture energies for the composite (unidirectional carbon fibre-

reinforced composite) made from unmodified resin and the adhesive are very similar to

that of the bulk resin. However, the initiation fracture energy for the composite with a

modified resin matrix shows only a modest increase, whereas the increase for the bulk

resin is considerable. By studying energy absorption of thin films with different thickness,

they [54] concluded that the size of the plastic zone plays a key role in lower fracture

toughness values of rubber toughened epoxy while used as thin films in composites.

20

Table 2-1: Rubber toughened epoxy as bulk resin, adhesive film, and as matrix in carbon fibre composites [54]

Rubber* content (%) Gc (kJ/m2)

Bulk resin Adhesive film Composites

0 0.33 0.28 0.28 3.2 1.4 0.33 0.37 6.2 2.2 1.35 0.36 9 3.2 1.5 0.49

* MY750 was toughened by CTBN as rubber

** adhesive film thickness ~ 200 m

Finally, since some rubber modified epoxies showed lower tensile strength and modulus,

as well as lower Tg values, inorganic particles have also been added to rubber-toughened

epoxies so that high toughness, high strength and high modulus may be obtained [58-62],

simultaneously. These modified polymers will be studied in the next section.

2. 2. 2. Rigid-particle toughened epoxies

Similar to the addition of rubber to toughen epoxy, adding rigid particles to the resin

system of a composite material can improve the resin mechanical properties, and hence

the composite structure [63]. Several inorganic fillers such as alumina, silica, barium

titanite and aluminum hydroxide have been investigated [63-71]. Two resin properties are

affected when particles are added to the resin: 1. resin viscosity and 2. glass transition

temperature; these two parameters affect processing of the composites and limits

variation of the type of toughening agent and its volume fraction. The mechanical

properties of the particulate filled epoxy resin are directly related to the properties of the

resin, filler, and the bonding condition. The filler volume fraction, particle size, aspect

ratio, modulus, and strength, as well as the resin-filler adhesion, and the toughness of the

resin are the main parameters governing the properties of these modified resins and has

been studied extensively in the literature [58, 59, 68, 69, 72, 73]. In the following

subsections, effect of each of these parameters will be studied on epoxy resin as one of

the most commonly used thermoset resins in industry.

2. 2. 2. 1. The effect of volume fraction of the filler

The effect of volume fraction of the filler on the mechanical properties of modified resins

has been studied in several articles, [20, 21]. Most of these studies are based on

experimental data. Moloney and Kausch, [20], reported the relation between the Young

21

modulus and the stress intensity factor as a function of filler volume fraction, based on

several experiments with different fillers. Increasing the volume fraction of the filler from

0 to 40% increased the elastic modulus of the composite from 3 GPa to 12, 15, and 22 GPa

for Silica, Alumina and Silicon carbide, respectively. They showed that for epoxy resins

with high glass transition, i.e. brittle resins, the fracture toughness values increased

linearly with the increase in the volume fraction of the filler up to 400%.

Spandoukis and Young [74] reported similar results. They showed that for epoxy resins

addition of high volume fraction of the filler (above 50%) increased the viscosity of the

resin so that the processing of the composite became impossible. For resins with volume

fraction lower than 30% sedimentation occurred. Therefore an optimum range for the

filler volume fraction is between 30-50%.

2. 2. 2. 2. Effect of filler particle size

Most of the literature, [20, 21, 74], showed that particle size did not affect the Young

modulus and stress intensity factor. Whereas, the strength of the resin decreased as the

filler size increased, due to the higher possibility of flaws within the particle. They

concluded that the smaller the filler size, the higher was the strength of the resin. On the

down side, decreasing the particle size increased the viscosity of the resin. Smaller

particles had a greater surface area, and thus the viscosity increased leading to complexity

in the manufacturing process of composites.

2. 2. 2. 3. Aspect Ratio

Aspect ratio is defined as the length divided by the diameter of the particles. The effect of

aspect ratio of the particles was studied by Moloney and Kausch [20] and more recently

by Fu et al. [75]. Figure 2-2 illustrates the effect of aspect ratio on the fracture toughness.

Short glass fibres, with an aspect ratio of about 15, are the toughest among those plotted

in the figure. Although higher aspect ratio of the fillers provides higher toughness, due to

the higher viscosity of the modified resin, their manufacturing process becomes very

difficult.

In a modelling work, Kelly [76] modelled the effect of aspect ratio of short fibres on

composite strength and toughness. They studied the effect of fibre length and aspect ratio

on the mechanical properties of composites. They showed that the work of fracture

increased with aspect ratio up to a critical size and then dropped. The reason was

22

explained as transition from fibre pull-out, improving the facture toughness, to fibre

failure.

Figure 2-2. Stress intensity factor for different fillers

Table 2-2 summarizes the effect of particle volume fraction, size, aspect ratio, resin

adhesion, and the matrix toughness on the mechanical properties of a typical epoxy resin.

Table 2-2: Summary of the effect of particles on the mechanical properties of an epoxy resin (experimental data)

Property Effect on composite

Modulus Toughness Strength

Particle volume fraction [20, 21] Increase Increase Constant

Particle size [20, 21, 77] Constant Constant Decrease

Particle aspect ratio (l/d) [76] --- Increase Increase

Matrix-particle adhesion [20] Constant Constant Increase

Matrix toughness [18, 38] Small

decrease Increase Decrease

2. 2. 2. 4. Effect of alignment of particles

Norman and Robertson [65] showed that alignment of the toughening particles normal to

the crack growth plane enhanced the fracture toughness of the base resin. They studied

the toughening effect of glassy particles with different alignment directions in a

photopolymerizable resin. The fracture toughness of both aligned and random particle

inside a resin system was improved. Using different alignment directions, they also

studied the contribution of different toughening mechanisms.

Short glass fibre

l/D= 15

l/D= 1

l/D= 4

l/D= 2-3

A187 treated glass beads

Silica & Alumina

Silicon carbide

23

The aligned-particles in the composites were prepared by suspending particles in a non-

conductive monomer, aligning them using an electric field, and polymerizing the

monomers while maintaining the alignment direction for the particles. The schematic of

the apparatus that they used to arrange the particles is shown in Figure 2-3.

Figure 2-3. Apparatus designed to obtain aligned specimen (a) plan and (b) side views

With their electric field setup, fracture toughness for different conditions were studied:

randomly aligned particles, and three other orientations depicted in Figure 2-4. The

results showed that the maximum improvement in the fracture toughness was obtained

for the particles that were aligned normal to the crack growth plane.

Figure 2-4. Fracture toughness as a function of volume fraction for different alignment conditions [65]

24

2. 2. 3. Fracture mechanisms in unfilled and filled epoxies

In this section, fracture mechanisms in neat and modified epoxies will be reviewed.

Several researchers have studied fracture mechanisms in brittle polymers [12, 17, 20, 78].

As reported by Moloney and Kausch [20], two types of crack propagation are observed for

brittle polymer systems:

1. Unstable, stick-slip propagation

2. Stable, continuous propagation

The addition of particles to the resin systems alters the unstable crack growth mode into

stable crack growth. Adding rubber to the resin system increases the fracture energy up to

sixty times the fracture energy of a neat resin system. Several mechanisms have been

proposed to explain the increased fracture properties. These mechanisms are [12, 54]:

1. Deformation of the rubber particles across the crack tip

2. Crazing of the matrix

3. Blunting of the crack tip

4. Absorption of energy by the matrix

The plastic zone size for rubber modified resins is considerably larger than an unmodified

resin, and hence larger energy absorption by the matrix. A summary and schematic of

these mechanisms is shown in Figure 2-5.

Figure 2-5. Schematic of toughening mechanisms in particle filled polymers [12, 79]: 1. Crack pinning and bowing, 2. Particle bridging, 3. Crack deflection and debonding, 4. Particle yielding

(plastic deformation), 5. Plastic zone at crazing, 6. Micro-cracking

The main toughening mechanisms in modified polymers are crack pinning and deflection

as reported in several papers [16, 68, 69, 80-85]. A schematic of crack pinning is shown in

25

Figure 2-6. This mechanism has been proposed by Lange [9]. Evans [10] modelled the

fracture energy increase of a bowed crack as a function of particle size and particle

spacing, i.e. r/c in Figure 2-6. He showed that the toughness increase due to crack pinning

and crack deflection is only a function of the geometry of the particles.

Figure 2-6. Crack pinning in modified resins [9]

Depending on the geometry of the particles, different toughening mechanisms have been

suggested and studied. One of the most important geometrical properties is the aspect

ratio of the filler which plays an important role in the fracture toughness improvement.

For fillers with higher aspect ratios, several energy consuming mechanisms have been

reported, such as:

1. Particle debonding, including fibre pull-out and fibre rupture [10, 37, 86, 87]

2. Matrix plastic deformation [88, 89]

3. Crack pinning [16, 80-83]

4. Crack deflection [68, 69, 84, 85]

A schematic of a crack deflection process, which consumes energy, is shown in Figure 2-7.

Two different aspect ratios of the fibre are shown on the figure. Higher aspect ratios of

the fibres force the crack front to move along a longer distance leading to more energy

consumption. Thus, longer aspect ratios result in higher fracture toughness

improvements.

26

Figure 2-7. Crack deflection due to the existence of short fibres [84]

In terms of modelling of the toughening mechanisms, one of the early works was

presented by Lange [9]. He modelled the crack pinning mechanism by estimating the

additional energy that is required to grow the crack due to the existence of rigid particles.

Evans [10, 84] built on Lange’s model and introduced correction factors to better predict

the experimental data. Rose [82] developed a model that contained both crack pinning

and particle bridging.

Kunz-Douglass et al. have modelled the dissipated energy during the crack growth in

rubber modified epoxy and experimentally verified their model [11]. Huang and Kinloch

[88] considered a more thorough model and verified it through several experiments. In a

recent publication, Zhao et al. [47] reviewed most of the recent modelling work in

polymer toughening.

2. 3. Carbon nanotubes and their potential as a reinforcement

Since their discovery, carbon nanotubes have been an attractive candidate for material

reinforcement [90-92]. Characterizing individual CNTs to find their mechanical properties

is more complex compared with other materials; the main reasons are,

1. CNTs aggregate into bundles of different diameters (10 to >1000 nm)

2. The length of the nanotubes vary within a wide range (1 to >1000 m)

3. The diameter of the nanotubes vary within a wide range (1 nm (SWNT) to >50 nm

(MWNT))

4. The morphology of the CNTs can vary greatly

5. Defects are probable both at the ends or sidewalls of the tubes

27

Nevertheless, several researchers managed to find the Young modulus and strength of

different types of CNTs. The most attractive methods of characterization include, [91]:

1. Micro-Raman spectroscopy

2. Thermal oscillations by TEM

3. Atomic-force microscope cantilever

Table 2-3 lists the mechanical properties of Single and Multi Walled CNTs, (SWNT and

MWNT), as well as carbon fibre as reported in [1, 93, 94]. According to the measured

mechanical properties, CNTs are strong candidates for different applications and most

importantly reinforcement of materials.

Table 2-3: Mechanical properties of CNT, compared to other materials

Reinforcement Diameter

(nm) Density (g/cm3)

Young’s Modulus (GPa)

Tensile Strength (GPa)

Failure strain (%)

MWNT [93] 10 – 40 1.8 – 2 800 20 – 40 2 – 12 SWNT [94] 0.6 – 3 1.4 – 1.8 1000 10 – 52 5 – 10

Carbon fibre [1] 10000 2 400 4 0.5 – 1

In the past decade, nano-sized fillers attracted researchers for epoxy modification, as

these fillers showed potential for simultaneous toughness, modulus and ductility

improvements [46-48, 95-98]. However, the potential of CNTs in multifunctional property

enhancement is limited by the challenges in the processing of CNT/polymer formulations.

2. 3. 1. CNT processing challenges

In order to effectively transfer the high mechanical properties of the carbon nanotubes to

an epoxy polymer, a good understanding of the manufacturing process is required. Due to

the high surface energy of CNTs, they tend to agglomerate in polymer solutions which

consequently affect the performance of the CNT modified polymer. Different

functionalization techniques and dispersing methods have been introduced [99] to

overcome the problem of agglomeration of CNTs in a resin polymer ,which will be

reviewed here.

28

2. 3. 1. 1. Functionalization

It has been reported that the best reinforcement of composites with CNTs, especially for

improved fracture toughness, requires a strong CNT − matrix interfacial bonding. The

higher interfacial bonding, the more resistance the resin attains against fibre pull-out.

Frankland et al. [100] showed that a great improvement can be achieved in mechanical

properties of a resin even if only 1% of the carbon atoms interact with the polymer

molecules. A better dispersion of the CNTs in the resin can be obtained through

functionalization.

Figure 2-8 shows the process of carboxylic functionalization of CNTs. As the first step, an

oxidative treatment of the nanotubes is used to develop carboxylic groups. The carboxyl

group enables the nanotube to create bond with the polymeric resin. In this step, since

the carbon nanotube cap is opened, the CNT properties are degraded. These carboxyl

groups would react with multifunctional amines and form active bonds with these amines

in the second step. In the third step, when the resin is added, the active amino functions

create bonds with the polymeric molecules of the resin.

Figure 2-8. Functionalization of CNTs [99]

It should be noted that a disadvantage of the functionalization process is the degradation

of the carbon nanotubes. This degradation deteriorates the mechanical properties of the

nanotubes.

2. 3. 1. 2. Dispersion

The importance of dispersion and their effect on mechanical properties of CNT modified

polymers has been discussed in many publications, [96, 99, 101-105]. Due to the high

29

surface area of CNTs, Van-der-Waals forces which exist between the CNTs leads to the

agglomeration of the nanotubes within the resin. As a result, during the mixing of the

nanotubes with a resin system, only few molecules of the polymer can penetrate between

the agglomerated nano-fillers and react with them. To achieve an effective reinforcement

by adding carbon nanotubes, CNTs should be dispersed uniformly into the resin.

Several methods have been proposed [96, 99] to disperse the nanotubes, such as chemical

treatment of CNTs, including, use of solvents [106], surfactant [107, 108], functionalization

[109, 110], polymer wrapping of CNTs [111, 112], and non-covalent bonding of polymer

chain to CNTs [113]. After this chemical treatment process, the mixing of the CNTs with

resin requires mechanical shear forces to further separate the agglomerated CNT inside

the resin solution. These techniques can be categorized as:

1. Sonication uses ultra-sonic devices to locally apply a high impact energy. Since this

impact energy introduces small shear forces, this method is more suitable for very

low viscosity resins, and for a small volume. Another problem with this method is

that the applied energy can rupture the CNTs and deteriorate the mechanical

properties of the CNTs. The best way to apply sonication is to disperse the CNTs in

an appropriate solvent with very low viscosity. The resin should then be added to

the mixture, while simultaneous heating can evaporate the solvent.

2. Stirring is a common method to disperse particles in liquid systems. A modified

propeller size and shape can be used to disperse nano-particles inside a resin.

3. Calendaring is another dispersing technique based on shear and tension stresses

between rollers of a three-role mill [5, 114]. Figure 2-9 shows the schematic of a

calendaring configuration. Different roller speed would apply a shear force

required to disperse the nano-particles, improving the dispersion quality of the

sample.

Figure 2-9. Schematic diagram (a) showing a typical calendaring component, [5]

30

A major challenge in understanding the effect of dispersion on the mechanical

performance of the material formulation is to quantify the state of dispersion. In a recent

study, Hamming et al. [103] experimentally verified the relation between the thermal

properties and the dispersion quality of a nano-modified PMMA. In order to quantify

dispersion, they defined a mean distance between the clusters of nano-particle

agglomerates. They showed an inverse relation between the Tg and the mean distance.

The lower the mean distance value, the higher is the Tg.

In another study, Fan et al. [115] characterized the dispersion of MWNT /vinyl ester

solution through an experimental setup shown in Figure 2-10. They benefited from

capillary force in fibre glass tows, and showed that the quality of MWNT dispersion is

proportional to the height h, the level to which the suspension rises in the fibre tow.

Andrews et al. [104] introduced the concept of dispersion index and correlated it to the

mixing time.

Figure 2-10. Qualitative characterization of the MWNT dispersion [115]

Most of the investigations on CNT dispersion were focused on the dispersion quality at

room temperature, and only recently, the effect of CNT dispersion during the curing

process have been reported and discussed, [116, 117].

2. 3. 2. Toughening potentials of CNTs

Several researchers have modelled the toughening potentials of CNTs in composite

laminates [118, 119]. Recently, the effect of CNTs as the filler of the resin system in

31

composite materials has been experimentally investigated [43, 46, 89, 94, 96, 101, 114,

118-124]. Figure 2-11 (a) and (b) show the fracture surface of a neat and CNT reinforced

epoxy resin. The CNT modified resin contains more river lines and a rougher surface that

consumed more energy during the crack propagation compared to the neat resin system.

As a result, nanocomposites are believed to show enhanced fracture toughness [5]. A very

important aspect of CNT toughening of polymers is the wide variation in the reported

enhancement of fracture toughness in literature. For example, researchers have reported

different results on the effect of 1% SWNT on PMMA, ranging from 6% increase in elastic

modulus to more than 50% for the same materials system [125, 126].

Figure 2-11. Effect of CNT on the fracture toughness (a) neat epoxy and (b) CNT reinforced [5]

The effect of several variables such as volume fraction, geometrical properties of CNTs

such as their size and aspect ratio, and their surface modification needs to be verified for

an effective CNT reinforcement of composites [103, 127]. The quality of dispersion [128-

130] and the interfacial adhesion between the CNTs and polymers chains [131-134] are

the most important composite processing parameters that need to be studied thoroughly.

2. 3. 2. 1. Effect of dispersion on fracture toughness

Several researchers have studied the effect of dispersion on the mechanical properties of

CNT modified polymers [5, 102, 103], but only very few of them were able to propose a

robust method to quantify dispersion [103-105, 115].

Thostenson and Chou [5] verified the effect of dispersion of CNTs, on the final fracture

toughness of the reinforced resin. Their setup for the manufacturing process is detailed in

32

Figure 2-9. Different gap settings resulted in different agglomeration contents, as shown

in Figure 2-12. Their results for the fracture toughness measurement are shown in Figure

2-13. At a relatively low CNT weight fraction content, they reported an improvement in

the fracture toughness.

Figure 2-12. Dispersing CNTs using a three-roll mill (calendaring) technique [5], (a) 50 m, (b) 20

m, (c) 10 m, (d) 5 m

Figure 2-13. Fracture toughness results; higher fracture toughness for higher gap settings [5]

0.4

0.6

0.8

1

1.2

1.4

0 1 2 3 4 5 6

K Ic

(M

Pa.

m1

/2)

Filler Content (wt.%)

Larger Gap

Smaller Gap

33

For those modified resins where the gap setting between the rollers of the three-roll mill

was larger (10 m), the overall fracture toughness was higher than those with smaller gap

(5m). Having a 10 m gap led to larger agglomeration of CNTs. They studied the fracture

surface to explain the improvement that they observed:

1. On the fracture surface, the nanotubes were pulled-out; this was a source of

energy dissipation due to the fibre pull-out and interfacial debonding.

2. Outside the area where nanotubes were agglomerated, the surface was smooth,

similar to neat resin.

3. Tail-like structures can be recognized where nanotubes were agglomerated. The

tail-like structures contributed to the crack deflection

For the 5 m gap setting, there was no tail-like structure. Some nanotube pull-outs were

observed for this configuration. As they discussed, having a smoother surface was a

possible explanation for the lower fracture toughness of the 5m gap setting. Whereas,

the 10 m gap setting contains both agglomerated and dispersed area. These two

features enable the modified resin to interact better with the crack front than the smaller

gap setting.

2. 3. 2. 2. Effect of different types of carbon nanotubes

Gonjy et al. [4] experimentally studied the effect of different types of CNTs, i.e. SWNT,

Double Wall Carbon Nanotube (DWNT), MWNT; on the fracture toughness of a CNT

modified resin. The result of their study is shown in Figure 2-14.

Figure 2-14. Effect of different types of CNT on the fracture toughness [4]

34

As it can be seen in the figure, there were no significant increase of fracture toughness

values when the filler content increased from 0.1% to 0.5%. The 0.3% was the optimum

filler content weight percentage. They proposed toughening mechanisms due to the

addition of CNTs including CNT pull-out, CNT rupture, telescopic pull-out in MWNTs,

bridging and debonding of the CNT walls from the surrounding polymer. There was no

modelling work to show the potential of each type of CNTs and to correlate the results to

the model.

2. 3. 3. CNT toughening of composites

Delamination is a major failure mechanism associated with the weaker interlaminar

property of composites that allow cracks to grow between the plies of a laminate. Since

fibres are mechanically stronger than the matrix [1], the matrix fracture toughness is the

key material property that controls damage initiation and growth in composites. Carbon

nanotubes have also been added to composite laminates to improve their mechanical,

thermal, and electrical properties and also provide a structural component with

multifunctional properties [96, 135-138]. Most of these efforts were to improve matrix

dominated properties, i.e. interlaminar reinforcement to improve delamination

resistance. Sensing of crack growth and health monitoring of composite structures is also

a very interesting potential application of CNTs [137, 139]. These studies showed an

increase in fracture toughness even at low-carbon nanotube (CNT) content.

There are two main techniques for the manufacturing of composite laminates modified

with CNTs: 1. CNT modification of matrix, and 2. CNT modification of fibre [135]. The

former has the advantage of being simple and also very similar to the traditional

processing methods of composites in the industry. The difficulties of this technique

include:

1. Filtering of CNTs during the impregnate of fibre mat [115, 140]

2. High viscosity of the resin system leading to major processing issues [3, 124, 141,

142]

CNT modification of fibres on the other hand has several advantages even though it is a

more complex processing method. This technique resolves the problem of dispersion and

aggregation of CNTs during the manufacturing. Also, CNTs are aligned perpendicular to

the fibres which is a desired direction to improve the delamination properties. Whereas

for CNT modified resin, direction of the CNTs in the composite is along the flow path

[135].

35

Several researchers modelled the potential of CNTs as a reinforcement particle to improve

the delamination resistance [118, 119, 138, 143-145]. In these works, two toughening

mechanisms were considered: the MWNT pull-out from the matrix and a sword-in-sheath

mechanism caused by the failure of the outermost layer of the MWNT. However, for long

CNTs embedded in a polymer, there is a critical length for CNT bridging that will define

other toughening mechanisms. By analogy with long fibre reinforced composites, the

nanotubes will pull-out if their length is below a critical value. For CNT having a length

higher than a critical value, there will be a combination of CNT pull-out and rupture [1].

There are several experimental research works that studied delamination resistance in

both mode 1 and mode 2 loading conditions [123, 140, 146-148]. Most of the

experimental works were with MWNT and only very few of them worked with SWNT

modified resin, [135].

A study by Kinloch et al. [98, 149] showed a synergistic effect when silica nano particles

were combined with rubber toughened epoxy. The result of their study is summarized in

Table 2-4. The silica nano particle which improved the fracture toughness of the base

epoxy by 400% became more effective in rubber toughened epoxy (same base epoxy). The

nano-modified rubber toughened epoxy was 200% tougher than rubber toughened epoxy

and 2200% tougher than the base epoxy. Figure 2-15 shows the synergistic effect of

rubber and nano particles as function of weight fraction of nano particle.

Table 2-4: Hybrid effect of silica nano particles in rubber toughened epoxy [98, 149]

Type of formulation Fracture toughness (J/m2) *

Ref.

Epoxy (Bis-phenol A)

103

[98, 149] Epoxy + rubber (ATBN)

1200

Epoxy + rubber (ATBN) + Surface modified SiO2

2300

Epoxy + Surface modified SiO2 460

* Maximum achieved

36

Figure 2-15. Hybrid effect of silica nano particle and rubber toughened epoxy [98]

2. 3. 4. Fracture mechanisms in nanocomposites

In general, the plastic zone size of a brittle resin is very small; adding CNTs as the filler of

the resin increase the size of the plastic deformation; hence, improvement in the fracture

toughness is achieved. The most important mechanisms leading to the enhancement of

the fracture toughness in nanocomposites are [4, 47, 48, 89, 99, 114]:

1. Localized inelastic matrix deformation and void nucleation

2. Particle-fibre debonding

3. Crack deflection

4. Crack pinning and bowing

5. Fibre pull-out

6. Crack-tip blunting

7. Particle-fibre deformation or failure at the crack-tip

Figure 2-16 shows a schematic of possible fracture mechanisms. A CNT incorporated in a

typical resin system is shown in Figure 2-16 (a). Depending on the interfacial bonding

between CNTs and polymer molecules different fracture mechanisms can be recognized.

In Figure 2-16 (b), a weak interfacial bonding leads to pull-out of the CNT from the resin.

Figure 2-16 (c) shows the case when the bonding is very strong, stronger than the fibre

strength, so that the CNT is ruptured before debonding from the resin. In the case of

strong bonding, especially for multi-walled CNTs, there is a possibility of outer shell

rupture of the CNTs and the pull-out of the inner shells, shown in Figure 2-16 (d). Figure

37

2-16 (e) illustrates the case when functionalized roots are strongly connected to the resin

system allowing partial debonding of the side walls of CNTs, and eventually CNT bridging

the crack.

Figure 2-16. Schematic description of fracture mechanisms of CNTs [4]

All of the proposed toughening mechanisms are based on analogies to micro-particle

fracture toughening mechanisms. However, their application to nano-scale fracture

mechanisms is questionable. As it is already explained for the effect of different types of

CNTs, i.e. SWNT, DWNT, and MWNT, there is no reliable explanation for fracture

behaviour of CNT-reinforced resins. The micro-scale fracture mechanism can help

researchers explain the fracture behaviour, but further research is required to understand

the fracture behaviour at the nano-scale.

2. 4. Summary and Thesis Objectives

CNTs can be used as reinforcing filler for polymer resins, similar to other particle-

reinforced resins. CNTs showed an improvement in mechanical properties of the resin at

very low volume fraction. However, a very important conclusion from the literature

regarding CNT toughening of polymers is the wide range of reported improvement. For

example, researchers have reported different results on the effect of 1% SWNT on PMMA,

ranging from 6% increase in elastic modulus to more than 50% for the same materials

system by another research group [125, 126]. Thus further research is required to achieve

the maximum toughening potential of CNTs through modelling and experiments.

38

In the light of above, the main objective of this work is to investigate the effect of CNTs on

brittle polymers as a toughening agent, through theoretical modelling and experimental

analysis. This will be achieved by focusing on the following aspects:

1. Modelling the CNTs toughening effect on polymers and composites (Chapter 3)

A more in depth modelling and experimental investigation of the effect of the physical

properties and processing parameters of CNTs on the final properties of the modified

resins is missing in the literature. There are only very few researchers who proposed

model for toughening potential of CNTs in polymers; therefore a detailed modelling

investigation of toughening mechanisms is needed to understand the key properties of

CNTs that mainly affect the fracture toughness improvement. This modelling would

help us identify the processing properties that need to be understood in order to

achieve major fracture toughness enhancement when CNTs are added.

2. Understanding the source of dispersion degradation during the processing of

polymeric composites. Achieving uniform dispersion of CNTs in polymeric

formulations (Chapter 4)

In terms of processing, achieving a strong interfacial bonding between CNTs and the

polymeric molecules of resin, as well as dispersing nanotubes uniformly into the resin

system are still major challenges. These two processing parameters need to be studied

and correlated to the final fracture toughness properties of composite structure.

According to the literature, a uniform dispersion is critical in achieving high quality

samples; however, a thorough understanding of the main sources of dispersion

degradation during the manufacturing of composite samples is missing in the

literature, particularly the effect of curing process on dispersion degradation.

Dispersion quality is significantly affected during the curing process. A series of test

will be performed to identify the main source of dispersion degradation during the

cure. This will be achieved by quantifying the dispersion quality during the cure and by

correlating the results to the rheological characteristics of the formulation. As a main

step in trying to understand source of dispersion degradation, a new image analysis

tool is presented to quantify dispersion.

3. Systematically fine tuning the formulations to understand the effect of polymer

processing parameters on microstructure development and the final mechanical

properties of the material (Chapter 4 and 5)

39

There is no clear relation between the dispersion quality of samples and the final

fracture properties of nano-modified formulations. Thus, through series of

experiments the effect of dispersion quality on the fracture toughness of CNT-

modified polymers and composites will be studied. Finally, CNT modified composite

laminates will be tested for their delamination properties.

4. Analysis of the fractured surface (using Scanning Electron Microscopy) to find the

direct effect of CNTs as a toughening agent and potentially identify new

toughening mechanisms. (Chapter 4 and 5)

Another aspect of CNT modified polymer is to understand the toughening mechanisms

when CNTs are added. A detailed investigation of fractured surface may potentially

identify new toughening mechanisms. In Chapters 4 and 5, we will study the fracture

surface of both polymers and composites containing CNTs.

40

Chapter 3. Modelling CNT Toughening Mechanisms

3. 1. Summary

In this chapter, the effect of aligned and randomly oriented carbon nanotube (CNT), with

respect to the crack growth plane, on the fracture toughness of polymers is modelled

using the Elastic Plastic Fracture Mechanics. According to a critical length, two dominant

toughening mechanisms for CNT-modified polymers are presented, i.e. CNT pull-out and

CNT rupture. The model is then used to identify the effect of CNTs geometrical and

mechanical properties on the enhancement of fracture toughness in CNT-modified

polymers. The key CNT properties are the radius, average length, ultimate strength,

elongation before failure, interfacial shear strength between CNTs and the polymer.

3. 2. Introduction

CNT reinforced resins can increase the composite ductility through different toughening

mechanisms such as CNT pull-out (i.e. CNT/matrix debonding), CNT bridging, and crack

deviation [4, 84]. Among the observed mechanisms, CNTs bridging is the only mechanism

that benefits from the high CNTs mechanical properties. It was shown that crack pinning

and crack deviation are mostly controlled by the shape of the reinforcing phase, [84].

More recently, the fracture toughness of Multi-walled CNT (MWNT) modified polymer

was modelled [118, 119]. In these works, two toughening mechanisms were considered:

the MWNT pull-out from the matrix and a sword-in-sheath mechanism caused by the

failure of the outermost layer of the MWNT. However, for long CNTs embedded in a

polymer, there is a critical length for CNT bridging that will define other toughening

mechanisms. By analogy with long fibre reinforced composites, the nanotubes will pull-

out if their length is below a critical value. For CNT having a length higher than a critical

value, there will be a combination of CNT pull-out and rupture [1].

In light of the above, the main objective of this chapter is the modelling of both the CNT

pull-out and the CNT rupture considering the CNT critical length. The proposed model also

addresses for the first time the effect of random CNT orientation in the polymer matrix

whereas previous modelling studies focused on perfectly aligned CNTs [118, 119]. There

has been no published model, which considers the effect of randomly distributed

nanotubes on fracture toughness enhancement. The CNT bridging was modelled using

41

Elastic Plastic Fracture Mechanics (EPFM), considering CNTs length, diameter, volume

fraction and alignment.

It should also be noted that even though other toughening mechanism such as crack

deviation exists in the CNT-modified polymers, CNTs bridging is the only mechanism that

benefits from the extraordinary mechanical properties of CNTs. While other toughening

mechanisms such as crack deviation does not benefit from the high mechanical properties

of CNTs and exist with other types of nano-reinforcements, such as nanoclays. Other types

of toughening, e.g. crack deviation, is a function of shape of the nano-particles, [84-86].

3. 3. Fracture Toughness Modelling

The addition of CNTs to the resin has two main effects: an increase of the neat resin

elastic modulus [150] and the introduction of toughening mechanisms such as CNT

bridging, [4, 99, 114]. Depending on the embedded length of CNTs in the resin, CNT

bridging can involve either CNT pull-out or CNT rupture. In both cases, the nanotubes

bridge the crack surfaces shielding the crack front from carrying the entire tensile load.

Hence, CNT pull-out and rupture are responsible for the nonlinear stress–strain behaviour

of modified resin systems.

Figure 3-1. Schematic description of CNTs toughening mechanisms [4], and J-integral contour

Due to the nonlinear behaviour of CNT-modified resin, Elastic Plastic Fracture Mechanics

(EPFM) is required to model the toughening mechanisms. Therefore, the J-integral

method along a closed contour around the crack tip was employed to model the effect of

42

the toughening mechanisms [151]. Previously, short fibre pull-out in composites was

modelled using the work of fracture method [76]. However, the derivation technique is

different from that of the J-integral method presented in this chapter. The approach of

this chapter is specifically different for the CNT rupture which considers the combined

effect of CNTs length, diameter, volume fraction and alignment for the first time. Figure

3-1 illustrates the J-integral path; JA is related to the remotely applied load, Jint. is the

intrinsic toughness of the resin, and Jb is the CNTs bridging effect.

According to Rice [151], the J-integral along a closed path can be decomposed and

rearranged as:

int.A bJ J J Equation 1

where Jint. is a function of the elastic energy release rate as a crack initiates. This part of

the J-integral can be found from experiments. However, as we are interested in increasing

the ductility of brittle resins, the main focus is the modelling of the CNT bridging effect.

Assuming that the CNTs are normal to the crack growth plane, two possible bridging

scenarios exist, i.e. pull-out and rupture. A critical length differentiates between the two

mechanisms and can be computed from a simple force balance on a single nanotube and

its interfacial bonding with the polymer chains. It can be found from,

c ul

r

Equation 2

where r is the nanotube radius, u is the nanotube ultimate strength, and is the

interfacial shear stress between the nanotube and the polymer. Pull-out will occur when

the embedded length of a CNT, l, is equal or smaller than a critical length, lc/2, (i.e. l ≤ lc/2).

Nanotubes will rupture when the embedded length is greater than the critical length, (l >

lc/2) [76].

The contribution of the bridging effect can be calculated using the definition of the J-

integral given by:

ii

uJ wdy T ds

x

Equation 3

where w is the strain energy density, Ti are the components of the traction vector defined

as the stresses acting normal to the contour, ui are the displacement vector components,

and ds is a length increment along the contour .

43

To model the contribution of the pull-out process, for CNTs with a volume fraction Vf and

an average length L, we assume that the nanotubes are embedded in the resin with an

equal length l, Figure 3-2(a). Similar to the Dugdale-Barenblatt strip yield model [152], the

CNTs exert a traction force on the crack faces. Therefore, the Dugdale-Barenblatt model

has been modified for CNT-modified polymer systems. Accordingly, for a relatively long

bridging zone, the first term in the J contour integral vanishes as dy = 0, Eq. 3, and the

contribution of the Jpull-out for CNTs can be expressed as,

2

00

( ) 22 ( ) ( )

ly y

pull out yy yy f f

u u x l lJ ds x dx V dl V

x x r r

Equation 4

where is the length of the pull-out zone, Figure 3-1, and 2×uy=l at the end of the pull-out

zone.

When l > lc/2, Jb is the energy release rate for the rupture of CNTs at the failure strain

(Jrupture). This contribution can be modelled as,

max

0( )

u

rupture f yyJ V u du Equation 5

where umax can be approximated as, umax≈ L×max, where L is the CNT total length and max

is the CNT elongation before failure. It is assumed that the wall of the nanotube is

detached from the resin, while the two ends are still attached. Thus, the value of Jrupture

represents the area under the stress versus displacement curve for rupturing a CNT.

Assuming a linear stress-strain curve for the CNTs, this area can be approximated as (u × L

× max)/2 and the energy associated with CNT failure is:

max

1

2rupture f uJ V L Equation 6

An average Jb is calculated for nanotubes with average length L, when the embedded

length varies. Figure 3-2 (b, c) show the possible CNTs distribution for L ≤ lc and L ≥ lc,

respectively.

44

(a) equally embedded length (b) (L ≤ lc) (c) (L ≥ lc)

Figure 3-2. Possible CNT length distribution along the crack growth path

For L ≤ lc, all CNTs pull-out with no rupture and the total contribution of Jb is given by:

/ 22

2

0( ) 1

( )/ 2 12

L

pull out f f

l r dl LJ V V

L r

for (L ≤ lc) Equation 7

For L ≥ lc, assuming that CNTs are fully dispersed, only Vf×lc/L portion of the CNTs pull-out,

and the rest of the CNTs (Vf×(1-lc/L)) rupture. It should be noted that according to the

value of lc for CNTs, the pull-out portion of the nanotubes can be very small. The total pull-

out contribution is,

/ 22

2

0( ) 1

( )/ 2 12

cl

f c c cpull out f

c

l r dlV l l lJ V

L l L r

for (L ≥ lc) Equation 8

and the nanotube rupture contribution is

max

1(1 )

2

crupture f u

lJ V L

L for (L ≥ lc) Equation 9

The total contribution of CNT bridging for the (L ≥ lc) is the sum of Eqs. 8 and 9.

These equations show that for an average length of CNTs from zero to the critical length,

the toughening contribution from nanotube pull-out is proportional to L2, Eq. 7, and for

longer CNTs, it is inversely proportional to L, Eq. 8. However, for the latter, nanotube

rupture significantly increases the total bridging effect.

3. 3. 1. Bridging Effect of Randomly Oriented CNTs

The main assumption in the previous section was that CNTs were normal to the crack

growth plane. In this section, the bridging effect of randomly oriented CNTs is estimated.

It is assumed that CNTs have uniform orientation distribution. To estimate the

contribution of randomly oriented CNTs, first the Jb for one nanotube with a known angle,

lc

L

l lc

L

l lc

L

l

45

, need to be found, Jb(). The stress carried by a nanotube oriented with an angle is

shown in Figure 3-3.

Figure 3-3. CNT with an angle with respect to the crack growth plane

Using the plane stress coordinate transformation, the transformation of the stress acting

on the crack is given by,

cos(2 )2 2

cos(2 )2 2

sin(2 )2

x

y

x y

Equation 10

In the calculation of the J-integral, for a relatively long bridging zone, the first term of the J

contour integral vanishes as dy = 0, and the traction vector can be found from:

x x x x y y

y x y x y y

T n n

T n n

Equation 11

Since, nx = 0 and ny = 1, the traction vector is then reduced to:

sin(2 )2

cos(2 )2 2

x

y

T

T

Equation 12

The stress, , acting on the nanotube, is a function of the nanotube radius, r, length of the

nanotube, l, and , the interfacial shear stress between the polymer and nanotube. The

stress was considered as a sinusoidal function of ; with = 0 at = 0, and =l/r at =

90°.

y’

x’

46

sin( )l

r

Equation 13

Based on Eqs. 11- 13 and the definition of the J-integral in Eq. 7, and knowing that

ds/dx=1, the energy required to pull-out a nanotube, i.e. bridging a crack with an angle

is given by:

cos sin

0 0( ) 2 2

l l yi xb f i f x y

uu uJ V T ds V T ds T ds

x x x

cos sin

0 0

2 22 3 3

sin( ) sin(2 ) (1 cos(2 ))

sin ( )(cos ( ) sin ( )) ( )

l l

f

f f

V l du l dvr

l lV V f

r r

Equation 14

In order to adapt the CNT bridging model, i.e. Eqs. 7 – 9, for the random orientation of

CNTs, Jb() in Equation 14 should be integrated over [0° – 90°], with a known orientation

distribution function, f(). But, the orientation angle in 3D space and the distribution

function are usually unknown. Therefore, to estimate the Jb() for randomly oriented

CNTs, a critical angle, c, is assumed which divide the bridging contribution of randomly

oriented nanotubes into those that do not contribute, f() = 0, 0°≤≤ c, and those that

are aligned to the crack growth plane, f() = 1, c ≤≤ 90°. The critical angle, c, is

estimated as follow:

2 2 290 90 90

0 0

2 290

0

90

0

0( ) ( ) (0) (1)

( ) (90 )

( ) 90

50

c

cf f f

f f c

c

c

l l lJ d V f d V d V d

r r r

l lV f d V

r r

f d

Equation 15

In Figure 3-4, the trigonometric part of Eq.15, f(), plotted as a function of . The area

under the curve, A1, is the integration of f()over [0° – 90°] and is equal to the area of the

dashed rectangle A2 in the figure corresponding to a critical angle, c=50°.

47

Figure 3-4. f() as function of the angle °

Thus, the CNT bridging effect for randomly oriented CNTs in a 3D space can be found by

multiplying Eqs. 7 – 9 by the probability of having CNTs oriented between 50° and 90°.

Figure 3-5. Orientation of a nanotube in 3D space

This probability is the area of a spherical cap having a height of L/2×(1-sin(50)), divided by

the area of a half sphere with diameter L, as shown in Figure 3-5. This ratio can be

calculated as:

2

2

2 ( 2) (1 sin(50))Probability 1 sin(50) 0.23

2 ( 2)

L

L

Equation 16

Thus, from Eq. 16, only 23% of randomly oriented CNTs will contribute to the bridging

process. By multiplying Eqs. (7)–(9) by 23%, the CNT bridging effect for randomly oriented

CNTs in a 3D space can be found.

°

A2

A1

f

L

48

Using the proposed bridging model, Eqs. 7 – 9, and Eq. 16, the contribution of CNT

bridging to the toughening effect of CNTs in brittle resins can be found.

Figure 3-6 depicts this contribution for aligned and randomly oriented CNTs as a function

of the average length of nanotubes. The figure illustrates different scenarios by varying

the parameters in Eqs. 7 – 9. Below the critical length, Eq. 2, there is only CNT pull-out; for

CNTs with average lengths higher than the critical length the bridging effect is the sum of

CNT pull-out and rupture contribution, Eq. 7 – 9. The CNT critical length is calculated from

values for the interfacial shear strength and tensile strength of nanotubes. For the former,

Barber et al. [153] measured a nanotube-polymer interfacial strength of 47 MPa from

experiments of MWCNT pull-out from a cured polyethylene-butane matrix. And for the

latter, typical values for the tensile strength and the failure strain of CNTs are given in

Table 3-1.

Table 3-1. Mechanical properties of Carbon fibre and CNTs

Reinforcement Diameter

(nm) Density (g/cm3)

Young’s Modulus (GPa)

Tensile Strength (GPa)

Failure strain (%)

MWNT [93] 10 – 40 1.8 – 2 800 20 – 40 2 – 12

SWNT [94] 0.6 – 3 1.4 – 1.8 1000 10 – 52 5 – 10

In Figure 3-6 (a), the CNTs radius is assumed to be 0.5 (nm), = 47 (MPa), u(CNT) = 40

(GPa), Vf = 3% and elongation to fracture of CNTs is assumed to be 10%. The values are

representative of SWNTs. In each of the following figures, from Figure 3-6(b) to (f), one

parameter will change and its effect on the CNT bridging is shown. Table 3-2 lists all the

input values to Eqs. 7 – 9 for Figure 3-6.

Table 3-2. Input values to Equation 7 - 9 for Figure 3-6

Figure 3-6 r Vf Ultimate Tensile

Strength (u) Elongation to

fracture () Interfacial Shear

Strength () (a) 0.5 nm 3% 40 GPa 10% 47 MPa

(b) 7.5 nm 3% 40 GPa 10% 47 MPa

(c) 7.5 nm 3% 20 GPa 10% 47 MPa

(d) 7.5 nm 3% 20 GPa 5% 47 MPa

(e) 7.5 nm 3% 20 GPa 5% 20 MPa

(f) 7.5 nm 20% 20 GPa 5% 47 MPa

Comparing Figure 3-6 (a) to (b), an increase in the SWNT radius can lead to a higher

contribution of bridging. MWNTs are weaker than the SWNTs [93, 94] , meaning that the

49

increase in radius comes at the expense of tensile strength of CNTs. Thus, to predict the

contribution of MWNTs, the ultimate strength was reduced to 20 GPa, Figure 3-6(c), and

in Figure 3-6(d), the elongation to fracture was reduced to 5%, to account for more brittle

MWNTs.

The interfacial bonding between the CNTs and polymer chains is the key factor in

determining the possible mechanical enhancement of nano-modified mixture. In order to

better understand the effect of the interfacial bonding on fracture toughness, in Figure

3-6(e), the interfacial shear stress was reduced to 20 MPa, demonstrating a weaker bond

strength between the nanotube and the polymer. The result shows that depending on the

average length of the nanotubes, a lower bond strength between CNTs and the polymer

chains may be favourable for enhanced toughness of the polymer.

Finally, in Figure 3-6(f) two cases of Vf = 3% and Vf = 20% were compared. At average

length of 10 m, aligned nanotubes with Vf = 3%, and randomly dispersed nanotubes with

Vf = 20% theoretically enhance the toughness by 0.17 and 0.2 (kJ/m2), respectively. It can

be concluded that the increase of nanotube volume fraction from 3% to 20% in a

randomly dispersed mixture of nanotube can have the same effect of the fracture

toughness enhancement as trying to align CNTs is the low volume fraction mixture.

50

(a) r = 0.5 nm (b) Increasing r to 7.5 nm

(c) Decreasing u to 20 GPa (d) Decreasing to 5%

(e) Decreasing to 20 MPa (f) Increasing Vf to 20% with =47 Mpa

Figure 3-6. Effect of CNT-bridging on the fracture toughness of brittle resins as a function of the average length of CNTs for a Single Walled CNT

J bri

dg

ing (

kJ/m

2)

vf =3%

vf =20%

J bri

dg

ing (

kJ/m

2)

J bri

dg

ing (

kJ/m

2)

Lavg.(m)

Aligned Random

Critical Length

Only pull-out Pull-out + rupture

Lavg.(m)

Lavg.(m) Lavg.(m)

Lavg.(m) Lavg.(m)

51

A typical value for the fracture toughness of epoxy resin at crack initiation is around 250

(J/m2). Based on the proposed model, CNTs have the potential to improve the fracture

toughness of brittle resins in mode-I fracture. The results also suggest that the CNT

rupture is the main toughening mechanism in CNT modified resins. Accordingly, to

improve the fracture toughness of CNT-modified polymers, long CNTs with high volume

fraction should be incorporated into the resin system, aligned perpendicular to the

fracture growth surface. Figure 3-7 shows the recommended steps to improve the

fracture toughness of CNT-modified resins.

Randomly oriented Alignment of CNTs Higher Vf Incorporation of longer

CNTs

Figure 3-7. Steps to improve the toughness of brittle polymers by incorporating CNTs

The effect of CNT/polymer physical and chemical properties on the critical length and the

bridging effect (toughening potential) predicted by the model is summarized in Table 3-3.

The table shows that the model presented in the theory section can be used to verify the

effect of different type of CNTs and the interfacial bonding on the toughness of a modified

resin.

Table 3-3. Effect of CNT - resin properties on the critical length and Jb, (NE=No Effect)

Diameter Vf Interfacial Shear Strength

(IFSS)

Ultimate Tensile

Strength

Elongation to

fracture

Critical length (lc) NE NE

Jpull-out NE

Jrupture

3. 4. Summary and Discussions

Based on the CNT bridging model presented in this chapter, aligning relatively long (~10

μm) carbon nanotubes perpendicular to the crack growth plane has great potential to

enhance the toughness of brittle polymers. This improvement is due to the toughening

mechanisms, mainly CNT bridging, that these nano particles introduce inside a polymer.

The model also shows that in most cases SWNTs were the best choice for toughening. To

52

be able to accurately predict the toughening potential of a nano-modified system, all the

parameters required for the model should be separately measured, e.g. single CNT pull-

out test to find interfacial shear strength, or TEM to find the average length and diameter

of CNTs. Therefore development of more characterization tools at nanoscale is required.

Another very important assumption of this modelling work is the perfect dispersion of

CNTs inside a polymer solution. However, in reality CNTs create bundles and agglomerates

inside the CNT modified solutions with lowers the mechanical properties of the nano fibre

compared to the individual CNTs. As a potential future modelling possibility, in most

modelling work, CNTs are considered as a single high performance element, however,

effect of CNT bundles and aggregates should be modelled to understand the effectiveness

of the CNTs.

Finally, toughness enhancement with CNTs requires increasing the volume fraction and

length of CNTs, and aligning them normal to the crack growth plane. These modifications

are bounded by the limitations and challenges in the processing of the CNT polymer

mixture. In the next chapter, the effect of CNTs on the fracture toughness of epoxy

polymers will be studied, and the results will be compared to the modelling work

presented in this chapter.

53

Chapter 4. Fracture Toughness of Carbon Nanotube Reinforced Resins

4. 1. Summary

In this chapter, the effect of Carbon Nanotubes, Single Wall (SWNT) and Multi Wall

(MWNT), on the fracture toughness of epoxy resins is studied. The effect of carbon

nanotube loading, functionalization, type of hardener and hardener-to-resin ratio is

considered. The experimental results are then compared to the predictions of the model

presented in Chapter 3. The dispersion stability during the curing process is then studied

as a function of cure temperature, cure rate, and type of hardener. A new image

processing approach is then introduced to quantify dispersion. Finally, the dispersion

quality is correlated with the measured fracture toughness.

4. 2. Materials

This chapter is focused on mode-I plane-strain fracture toughness of epoxy polymer

modified with SWNT and MWNT.

4. 2. 1. SWNT resin system

The SWNT used in this work were unfunctionalized and Anionic functionalized Single Wall

Carbon Nanotube (SWNT) supplied by the National Research Council Canada’s Steacie

Institute for Molecular Sciences (NRC-SIMS) located in Ottawa, Ontario, Canada. Two

types of nanotube synthesis technique were used, i.e. laser ablation technique leading to

higher quality and longer nanotubes [154], and plasma synthesized SWNTs which are

shorter and contained more impurities [155]. The formulations contained both

unfunctionalized SWNTs as well as negatively charged SWNTs (Anionic). Table 4-1

summarized the type of SWNTs that were used in this study.

The polymer used with the SWNT system was a standard aerospace grade epoxy, Araldite®

MY0510 epoxy, supplied by Huntsman. This epoxy was used with two different types of

hardener as the curing agent: 1. Aradur® HY976-1 which is 4, 4-Diaminodiphenyl Sulphone

54

(referred to as DDS), and 2. Aradur® 5200 which is an aromatic diamine (referred to as

Aradur).

Table 4-1: Summary of different types of SWNT used

SWNT

Synthesis technique Laser, Plasma CNT chemical treatments Unfunctionalized, Negatively charged (Anionic)

For the SWNT system, the procedure for SWNT-epoxy integration is detailed in [124]. Two

types of hardener were used to cure the resin. The hardeners were added before sample

preparation. For the DDS system, Huntsman’s recommended hardener to resin ratio was

49:100 for a gel time of 19 min at 180°C. For the Aradur 5200, the recommended mixing

ratio is 35:100 for a gel time of 9 min at 180°C.The following DDS to resin ratios, 100:49,

100:55, 100:60 and 100:67, were selected to find the optimum hardener to resin ratio for

the SWNT-modified MY0510. .

Two different cure cycles were used for DDS cured resins. Cure cycle # 1, recommended

by Huntsman, consisted of a 2 hrs hold at 130 °C with a 3 °C/min ramp rate, followed by a

2 hrs hold at 180 °C for a total cure time of 5 hrs. An additional hold at 130 was added to

gel the resin at a lower temperature and hence to stabilize the dispersion quality before

complete curing of the formulations. Cure cycle # 2 was a shorter single hold cycle with a 2

hrs hold at 200 °C with a 3 °C/min ramp rate.

For Aradur, cure cycle #3: 2 hours at 150°C and then 2 hours at 180°C was used. Table 4-2

summarizes the SWNT system formulations that were used in this chapter. The loading of

the CNTs was 0.3% unless mentioned otherwise.

Table 4-2: Summary of the SWNT + MY0510 formulations and cure cycle used

Hardener Resin: Hardener Ratio Cure cycle ( 3 °C/min ramp rate) Characterization

DDS 100:49, 100:55, 100:60, 100:67

# 1: 2 hrs at 130 °C, 2 hrs at 180 °C # 2: 2 hrs at 200 °C

Rheology, Dispersion, Fracture toughness

Aradur 100:35 # 3: 2 hrs at 150°C, 2 hrs at 180°C

4. 2. 2. MWNT resin system

The MWNTs used in this study were supplied by Baytubes®. Baytubes® are agglomerates

of MWNTs with low outer diameter, narrow diameter distribution and a high aspect ratio

(length-to-diameter ratio) of around 100 to 500. Their outer mean diameter is 13−16 nm.

Their length varies from 1−10 m. Baytubes® were produced based on chemical vapour

55

deposition and were functionalized in such a way that they develop some physical

interactions with the matrix (non-covalent linkage).

The MWNTs were mixed with bisphenol-A epoxy resin as described in [156]. Four types of

curing agent were used with the MWNT resin system:

1. isophorone diamine (IPD),

2. triethylenetriamine (TETA),

3. Mix of IPD with triteriamine, (N3),

4. Mix of IPD with TETA.

For IPD, the mixing ratio was 100 (resin): 23 (hardener) and the cure cycle was 2 hours

hold at 120 °C as recommended by manufacturer. TETA on the other hand cures at room

temperature and gels around 30 min. After a 24h curing at room temperature, a 2-hr post

curing at 150°C was recommended by the manufacturer. The mixing ratio for the TETA

and epoxy resin was 100 (resin): 14 (hardener). For the mix of IPD with N3, in which N3

acts as a catalyst to speed up the reaction, N3 was first mixed with IPD. N3 to IPD ratio

was 3:100 and the cure cycle was the same as for IPD. Finally, three formulations of IPD

with TETA with IPD:TETA ratios of 20/80, and 50/50, and 80/20 wt% were tested. The cure

cycle for these formulations was the same as the cure cycle for TETA. Table 4-3

summarizes the formulations used for the MWNT system. The loading for the MWNT

system was 0.3% unless otherwise specified.

Table 4-3: Summary of the MWNT + bisphenol-A epoxy formulations

Hardener Resin: Hardener Ratio Cure cycle (3 °C/min ramp rate) Characterization

IPD 100:23 2 hrs at 120 °C

Rheology, Dispersion, Fracture toughness

IPD/N3 100:3 (IPD:N3) 2 hrs at 120 °C

TETA 100:14 24 hrs at 25 °C, 2hrs at 150°C

IPD/TETA 80:20 (IPD:TETA) 50:50 20:80

24 hrs at 25 °C, 2hrs at 150°C

4. 3. Experimental Procedures

4. 3. 1. Fracture toughness specimen dimensions

The specimen dimensions were chosen according to the standard test methods for plain-

strain fracture toughness of plastic materials, ASTM D5045 – 91. The dimensions of the

samples were of special importance to measure geometry-independent values for fracture

toughness. To minimize material use, Single Edge Notch Bending (SENB) test was chosen

56

for 4-point bending test and the dimensions of the samples were chosen as

20mm×4mm×2mm with the notch depth of 2(mm) and a span of 16(mm). Specimen

dimensions are shown in Figure 4-1. The sample width, W, was W = 2×B, where B is the

specimen thickness. In both geometries the crack length, a, was selected such that 0.45 <

a/W < 0.55.

Figure 4-1. Dimensions of the fracture toughness specimen

In order for a result to be considered valid according to the ASTM, the following size

criteria must be satisfied:

2

, , ( ) 2.5 Q yB a W a K Equation 4-1

where KQ is the conditional or trial KIc value, and y is the yield stress of the material for the

temperature and loading rate of the test. The criteria require that B must be sufficient to

ensure plane strain and that (W − a) be sufficient to avoid excessive plasticity in the

ligament. According to the material datasheet for the epoxies that were used [157], KQ is

approximately 1 MPAm1/2 and y is 75 MPa. Hence, according to Equation 4-1, if the

dimensions of the samples are above 500 m, then the dimensions of the specimen is a

valid according to ASTM standard.

4. 3. 2. Specimen preparation

A mould was designed to produce 10 specimens at a time. Minimum amount of flow of

the resin is desirable to reduce the possibility of void formation while preparing the

samples. Hence, mould casting was chosen to produce the samples. The conceptual

design of the mould is depicted in Figure 4-2.

Thickness = 2mm

a = 2mm

S = 16mm

W = 4mm

57

(a) Conceptual design of the mould (b) Top Teflon insert (dimensions in mm)

(c) The picture of the mould in use

Figure 4-2. Casting mould for fracture toughness specimen preparation

The mould contained two Teflon inserts (shown in white color in Figure 4-2(a)) where the

resin was poured. Having channels on the top Teflon insert allowed easier sample

removal. The aluminum components (shown in grey) were used to close and seal the

mould. After the mould was closed, a 5 bar pressure was applied on one end of the

channels. The pressure forced the resin into the channels and minimized the size of any

possible voids. The mould was then heated according to the recommended cure cycle of

the resin (Table 4-2 and Table 4-3). The actual picture of the closed mould with applied

pressure is shown in Figure 4-2(c).

After removing the samples from the mould, the samples were notched to create the

initial crack, a. A sharp notch was first prepared with depth of 1.7 mm and width of 300

m using the Accutom model of Struers precision diamond saw. Subsequently, a natural

58

crack was initiated by sliding a fresh razor blade across the notch root with depth of

around 300 m.

4. 3. 3. Fracture toughness measurement test setup

A 100lb Fullam tensile fixture (Figure 4-3) was used to perform the tests under an Optical

Microscope (Olympus BX-51M). The samples were tested at room temperature ranging

from 23 – 26 °C.

Metrology was conducted for measuring specimen thickness at two locations and

specimen width at three locations using a micrometer. The crack length was measured

under the optical microscope prior to the fracture test on both ends of the crack front.

The specimen was installed in the test fixture and aligned visually such that the loading pin

was at the centre of specimen thickness and the specimen was not twisted. Also to ensure

a constant moment during the crack propagation, a 4-point bending fixture was used to

apply the load, and an equivalent 3-point bending force was used in the data reduction.

The data acquisition system was zeroed and started. A very slow loading rate of 0.005

mm/s was chosen to ensure a slow crack growth. The load-displacement curve for each

test was recorded for data reduction to find the fracture toughness values.

Figure 4-3. Fullam tensile test fixture and the initial crack under optical microscope

The plane-strain fracture toughness, KIC was calculated from the maximum load, Pmax as

follows:

1/2

max ( )ICK P BW f x Equation 4-2

59

where,

2

3/2

(1.99 (1 )(2.15 3.93 2.72 ))( ) 6

(1 2 )(1 )

x x x xf x x

x x

Equation 4-3

and,

Wx

a Equation 4-4

where W is the specimen width and B is the specimen thickness and a is the crack length.

The detail of the data reduction is given in [158]. In this thesis, all the polymer fracture

tests were performed under 4-point bending loading condition and an equivalent 3-point

bending load was used to calculate the stress intensity fracture toughness values.

4. 3. 4. Hot stage: dispersion analysis

One of the key parameters that affect the mechanical properties of CNT modified

polymers is the dispersion quality of the samples; not only at room temperature and

during the mixing of the CNTs with resin, but also after casting the samples prior to the

gelation point of the resin. A well dispersed sample is highly desirable to achieve proper

load transfer from the structure to its nano structure. Poor dispersion may deteriorate the

strength and the fracture toughness values of CNT modified resins compared with the

neat resin.

The effect of elevated temperature on the deterioration of the CNT dispersion quality was

noticed during the sample preparation. To verify the effect of temperature on dispersion

degradation prior to the gelation point of the resin mixtures, a systematic series of tests

was performed. The main focus of these tests was to understand the effect of the curing

process on the dispersion quality of a CNT-modified polymer.

A Linkam Examina Dynamic hot-stage was used (Figure 4-4) to monitor the changes in

dispersion quality during the curing process of CNT-modified polymers. The hot-stage is

designed to be used with an upright microscope, where the objective lens is above the

sample. The objective lens is isolated from the sample by the stage lid window which is a

fixed distance from the heating/cooling element. A magnification of 250× was used for all

the images.

For each test, 1 gram from the same formulation that was prepared for fracture toughness

specimens was used for the hot-stage test. A drop of formulation was placed between two

60

glass substrates. The thickness of resin film was approximately 200 microns. As shown in

Figure 4-4, the sample was then placed inside the hot-stage. The cure cycle was then

applied and the dispersion variation in the formulation was closely observed and recorded

with an Olympus optical microscope.

(a) Actual picture with an opened stage lid (b) Schematic of the setup (dimensions in mm)

Figure 4-4. Linkam Examina hot-stage setup

4. 3. 5. Shear stage: dispersion analysis

In order to understand the main source of dispersion degradation during the curing

process of CNT-modified resins, a shear stage was used. The Linkam Optical Shearing

System (Figure 4-5) allowed structural dynamics of complex fluids to be directly observed

via standard optical microscope while they were under controlled temperature and shear.

The shear stage used two highly polished quartz plates that were parallel to each other.

Each plate was in thermal contact with an independently controlled pure silver heater

utilising platinum resistors sensitive to 0.1°C. The bottom plate, on which the sample was

placed, operated in either oscillatory, steady or step shear modes. The gap between the

two plates can be precisely set from 5 to 2500 m.

Similar to the hot-stage test, for each formulation 3-5 grams of resin was placed between

the two quartz plate and then the distance between the two plates were set to 500 m.

The shear and cure cycle was then applied. The tests were stopped before the gelation

point of the formulation which was measured during viscosity tests. The detail parameters

of each test are given in the result section.

4.5

12.5

~1

3.5

Objective Lens

Stage lid insert0.2 mm thick lid window

Sample0.2 mm cover slip

Silver block heating element

Condenser Lens

61

(a) (b) (c)

Figure 4-5. Linkam optical shearing system, (a) closed, (b) opened, (c) schematic of shear stage setup with the sample between the two quartz plates

4. 3. 6. Rheological analysis

To better understand the results of the shear-stage and hot-stage, viscosity of the same

formulation from the same batch were measured using the TA Instrument AR2000

Rheometer (Figure 4-6).

(a) (b)

Figure 4-6. (a) The AR 2000 Rheometer with disposable parallel plates installed, (b) Close-up of the sample between two parallel plates

A disposable 25 mm parallel-plate setup was used. The rheological properties of the

polymers were determined in the oscillatory mode. The 25 mm disposable parallel plate

attachment and the plates themselves were first installed into the rheometer. The

rheometer was then calibrated. This consisted of mapping the air-bearing, calibrating the

system inertia and setting the zero-gap. The polymer formulation was then carefully

62

deposited on the lower plate. The gap between the two plates was in the range of 500 to

700 µm (volume of approximately 1 ml). The upper plate was then lowered until the edge

of the sample was parallel to that of the plates. The environmental test chamber doors

were closed and cure cycle was applied.

Dynamic temperature tests (oscillatory temperature ramp) were performed to observe

the variations in the viscosity profile of the resins with temperature. The experiments

were heated from room temperature to 250 °C at a ramp rate of 3 °C/min. The control

variable of 12 % strain with the sampling rate of 1 point every 10 seconds was used.

4. 4. Results and Discussions

4. 4. 1. Hot-stage test results

4. 4. 1. 1. SWNT system

The dispersion quality analysis results of the 0.3% SWNT mixed with MY0510 with no

hardener is shown in Figure 4-7 – Figure 4-9. Each figure shows the results of dispersion

quality at three different temperatures, i.e. 27 °C, 100 °C, and 200 °C. In Figure 4-7

(unfunctionalized SWNTs) and in Figure 4-8 (anionic SWNTs), the nanotubes were

synthesized using the laser ablation technique. SWNTs for the formulation in Figure 4-9

were synthesized using plasma technique. For all the three formulations, the heating rate

was 50 °C/min.

(a) 27 °C (b) 100 °C (c) 200 °C

Figure 4-7. 0.3% wt. Unfunctionalized Laser SWNT system dispersion analysis – with no hardener

100 μm 100 μm 100 μm

63

(a) 27 °C (b) 100 °C (c) 200 °C

Figure 4-8. 0.3% wt. Anionic Laser SWNT system dispersion analysis– with no hardener

(a) 27 °C (b) 100 °C (c) 200 °C

Figure 4-9. 0.3% wt. Unfunctionalized Plasma SWNT system dispersion analysis – with no hardener

As the temperature increases, the viscosity of polymers drops leading to less resistance

towards the agglomeration of the SWNTs, however, SWNT agglomeration was only

evident in Figure 4-9, were lower quality, unfunctionalized plasma SWNTs were used.

Addition of two types of hardeners, i.e. DDS and Aradur, to the 0.3% wt. SWNT system is

shown in Figure 4-10, where the state of the CNT dispersion at room temperature (25 °C)

was compared to 180 °C. At 25 °C, the nanotubes were well dispersed, however, when the

temperature was increased to 180 °C, the nanotubes started to agglomerates and the CNT

dispersion quality deteriorated significantly. The only sample that was relatively stable at

180 °C was the Laser SWNTs in the MY0510 epoxy with the Aradur hardener. The

agglomeration observed for the other cases was mainly caused by a combination of low

resin viscosity, thermal expansion and curing of the mixture. The cause of dispersion

degradation will be studied in detail in section 4. 4. 1. 4. Figure 4-10 also shows that the

temperature ramp rate is not the cause of dispersion degradation of the formulations,

since both temperature ramp rates (5 and 100 °C/min) resulted in CNT dispersion

degradation.

100 μm 100 μm 100 μm

100 μm 100 μm 100 μm

64

25 °C 180 °C

Unfunc. 0.3 % Laser DDS 5 °C/min

Unfunc. 0.3 % Laser DDS 100 °C/min

Unfunc. 0.3 % Laser Aradur 20 °C/min

Unfunc. 0.3 % Plasma Aradur 20 °C/min

Figure 4-10. SWNT dispersion stability analysis for two types of hardener: DDS and Aradur

The hot-stage results of different DDS to SWNT formulations are shown in Figure 4-11 to

Figure 4-14. The loading of the SWNT was 0.1 % wt. and they were negatively charged

(Anionic). Cure cycle # 1 (Table 4-2) was used to cure the samples.

An interesting observation of Figure 4-11 to Figure 4-14 was the dissolution of DDS in the

formulation. DDS is a solid aromatic amine hardener at room temperature which dissolved

in the MY0510 mixture at around 110 °C. Even though, the SWNTs and DDS were well

dispersed at room temperature, after the dissolution of DDS, the areas occupied by DDS

were replaced by resin only. An example of such area is highlighted in Figure 4-11 (c).

These regions would further grow as the temperature increased, resulting in further

degradation of the dispersion quality.

100 μm 100 μm

100 μm 100 μm

100 μm 100 μm

100 μm 100 μm

65

(a) 30 °C (b) 80 °C (c) 105 °C

(d) 109 °C (e) 111 °C (f) 130 °C

Figure 4-11. SWNT system dispersion analysis – 100:49 Resin to DDS ratio

(a) 30 °C (b) 80 °C (c) 105 °C

(d) 106 °C (e) 108 °C (f) 111 °C

Figure 4-12. SWNT system dispersion analysis – 100: 55 Resin to DDS ratio

100 μm 100 μm 100 μm

100 μm 100 μm 100 μm

100 μm 100 μm 100 μm

100 μm 100 μm 100 μm

66

(a) 30 °C (b) 80 °C (c) 100 °C

(d) 104 °C (e) 110 °C (f) 115 °C


(a) 25 °C (b) 70 °C (c) 100 °C

(d) 130 °C


As a solution to this problem, the DDS based SWNT systems were pre- heated to 100 °C to

dissolve the DDS particles. The pre-heated sample was then cooled down to 50 °C. An

additional 5-minute shear mixing was then applied to the sample with dissolved DDS to

improve the dispersion quality of the formulation. Cure cycle # 2 (Table 4-2) was then

applied. The results for pre-heated 0.1% SWNT / MY0510 with 100:60 resin to DDS ratio

with the additional shear mixing is shown in Figure 4-15. Comparing the results of Figure

100 μm 100 μm 100 μm

100 μm 100 μm 100 μm

100 μm 100 μm 100 μm

100 μm

67

4-13 with Figure 4-15, the additional pre-heating and mixing step clearly improved the

dispersion quality.

(a) 50 °C (b) 120 °C (c) 130 °C

Figure 4-15. SWNT system dispersion analysis – pre-heated to dissolve DDS and further mixed for improved dispersion quality

4. 4. 1. 2. MWNT system

For the MWNT system, the dispersion stability was studied by changing the curing agent

for the MWNT system. The result of the MWNT system with TETA as hardener is shown in

Figure 4-16. The sample was cured at room temperature for 24 hrs. The dispersion quality

stayed the same even after a 2-hr post-cure of the formulation. Since the gelation point

was occurred during the room temperature hold for 24 hrs, and MWNTs could not move

after the gelation point, the 2-hr post-cure had minimal effect on dispersion degradation

of the formulation.

(a) 25 °C (b) 25 °C after 24 hrs (c) after 2 hrs @ 150 °C

Figure 4-16. Dispersion quality evolution during the cure, MWNT system with TETA hardener

Since the dispersion quality remained constant for this formulation while the resin cured

at room temperature, it can be concluded that the chemical process of polymerization and

the 3D network formation during the curing process has minimal effect on dispersion

degradation. Viscosity drop and thermal expansion of the resin during the cure cycle for

the other formulations, i.e. IPD, IPD/N3, are the main drivers of dispersion degradation.

100 μm 100 μm 100 μm

100 μm 100 μm 100 μm

68

The result of hot-stage test for the IPD system is shown in Figure 4-17. For the IPD system

the dispersion degraded during the first heating cycle, as shown in Figure 4-17. The

dispersion degradation started around 50 – 55 °C and stayed the same after 75 °C.

(a) 25 °C (b) 55 °C (c) 75 °C

Figure 4-17. Dispersion quality evolution during the cure, MWNT system with IPD hardener

A faster reaction up to the gelation point for the mix of IPD/N3 showed better dispersion

stability (Figure 4-18). The faster the reaction, the shorter is the gelation time, leading to

less time for MWNTs to freely move and agglomerate. N3 was recommended by the

manufacturer to act as a catalyst to speed up the reaction. This improvement can be seen

when comparing the results of IPD/N3 dispersion analysis in Figure 4-18 with the results of

IPD formulation in Figure 4-17.

(a) 25 °C (b) 55 °C (c) 75 °C

Figure 4-18. Dispersion quality evolution during the cure, MWNT system with IPD/N3 hardener

The result of dispersion analysis for the mixture of 50% IPD and 50% TETA is shown in

Figure 4-19. In the first 30 minutes of the curing process, the MWNT agglomerated, but

they remained constant even after the post curing process. The dispersion stability of the

IPD/TETA formulations was better compared to the IPD samples, as it allowed curing the

formulation at room temperature.

The dispersion analysis results are necessary to understand the relation between

dispersion quality and the final fracture toughness of each formulation. However, it is very

important to identify the root causes of dispersion degradation in CNT-modified polymers,

which will be addressed in the next section by quantifying the dispersion test results.

100 μm 100 μm 100 μm

100 μm 100 μm 100 μm

69

(a) 25 °C (b) 25 °C after 30 min (c) 25 °C after 24 hours

Figure 4-19. Dispersion quality evolution during the cure, MWNT system with IPD/TETA hardener

4. 4. 1. 3. Quantifying dispersion: using Image Analysis

To quantify the dispersion analysis results, presented in the previous subsection, a Matlab

code was developed to process the Linkam hot-stage images (Appendix A.1. Matlab code

for image analysis). The quantification will help correlate the dispersion analysis with the

viscosity profile of the resin from the rheological analysis.

Figure 4-20 illustrates the steps of image processing. Figure 4-20 (a) and (d) show the

typical output images from the hot-stage experiments at room temperature and at an

elevated temperature.

(a) 25 °C (b) Grey scale image (c) BW, MWNT area 0.9

(d) 120 °C (e) Grey scale image (f) BW, MWNT area 0.41

Figure 4-20. Image processing steps, RGB to Grey to Black & White, for IPD/N3 system

100 μm 100 μm 100 μm

70

The Matlab code first converted the images to grey scale images (Figure 4-20 (b) and (e)),

and then the images were converted to a black and white format (Figure 4-20 (c) and (f));

black regions represented the CNT agglomerations, and white regions represented regions

that contains only resin. A fractional area, Af, was then calculated as:

f

Area of CNT agglomerates (black area)A

Total Area (black area + white area) Equation 4-5

The areas were calculated using “bwarea” function of Matlab.

Figure 4-21 and Figure 4-23 illustrate the sequences of images taken from the MWNT

system with IPD and IPD/N3 as hardener, respectively. The corresponding dispersion curve

(Af – Time) is shown in Figure 4-22 and Figure 4-24, respectively. In these figures, each

point on the curve corresponds to an image taken from the hot-stage setup. In Figure

4-22, images 2_01, 2_04, and 2_07 of Figure 4-21 are highlighted. The drop in the

dispersion curve is consistent with the dispersion degradation seen on the images. Similar

pattern is seen In Figure 4-24, where the drop in the dispersion curve captured the

dispersion degradation shown in image N_07 to N_11 in Figure 4-23.

Figure 4-21. Image sequences from the hot-stage test setup for MWNT system with IPD

71

Figure 4-22. Dispersion quantification results for the MWNT system with IPD, Af calculated from Eq. 4-5

Figure 4-23. Image sequences from the hot-stage test setup for MWNT system with IPD/N3

0 500 1000 1500 2000 2500 3000 35000

0.2

0.4

0.6

0.8

1

Af

Time(sec)

Dispersion & Temp vs. Time

0 500 1000 1500 2000 2500 3000 35000

20

40

60

80

100

120

140

T(

C)

2_01

2_04

2_07

72

Figure 4-24. Dispersion quantification results for the MWNT system with IPD/N3, Af calculated from Eq. 4-5

The increase in the dispersion curve after the first drop (Figure 4-24) is due to the curing

of the resin and correspondingly darkening of the resin layer. The importance of the

dispersion curve can be noticed when compared to the rheology curve for the same

material system.

The dispersion quantification curves for other formulations are presented in the

Appendix.A.2.

4. 4. 1. 4. Sources of dispersion degradation

According to the results of the dispersion stability analysis for both SWNT and MWNT

system, dispersion has degraded in all the formulation that required a curing cycle at an

elevated temperature, whereas those that were cured at room temperature were the only

formulations with stable dispersion. These facts together with the results of samples with

no hardener (Figure 4-9) demonstrate that the reaction of the hardener with the polymer

formulation is not the major driver of dispersion degradation, but rather an elevated

temperature is the key factor in dispersion degradation.

0 500 1000 1500 2000 2500 3000 35000

0.2

0.4

0.6

0.8

1

Af

Time(sec)


0 500 1000 1500 2000 2500 3000 35000

20

40

60

80

100

120

140

T(

C)

73

As temperature increases the viscosity of the resin drops and the CNTs can move freely. At

the same time, due to the thermal expansion of the resin, internal shear forces apply. The

combinational effects of these physical phenomena are studied in this section through

series of rheological and shear tests.

Viscosity drop during the cure cycle. As shown in the dispersion degradation analysis for

the SWNT system (Figure 4-10), the dispersion degradation started as the temperature

increased and then stabilized when the minimum viscosity reached. This result is

consistent with the viscosity profile of the SWNT system (Figure 4-25).

As can be seen in Figure 4-10, in the DDS system all the formulations became unstable at

around 105 °C to 110 °C, and then stabilized; this is the temperature range that the

minimum viscosity of the formulations were reached and stayed constant up to the

gelation point.

Figure 4-25. Typical rheology curve for MY0510/DDS/SWNT formulation. From room temperature ramp (3 °C/min) to 250 °C with control variable of 12 % strain

In Figure 4-26, the rheological behaviour of the DDS system is compared with the Aradur

system. As can been clearly seen, the formulations containing DDS had higher viscosity

values at room temperature compared to the Aradur formulations. The main difference

0.1

1

10

100

1000

10000

20 40 60 80 100 120 140 160 180 200

Vis

cosi

ty (P

a.s)

Temperature °C

MY0510/DDS SWNT/MY0510/DDS

Minimum Viscosity Region

74

was at elevated temperatures, where there was a noticeable drop in the viscosity of the

DDS system. In contrast, the viscosity of Aradur formulations remained relatively

unchanged. This difference in viscosity drop can explain the higher dispersion degradation

of the DDS formulations.

Figure 4-26. Rheology results, comparing DDS vs. Aradur hardener. From room temperature ramp (3 °C/min) to 140 °C hold up to gelation

Similar to the SWNT system, for the MWNT formulations the results of the TETA and IPD

dispersion quality analysis can be correlated to the rheological analysis of each

formulation. Comparing the temperature at which that dispersion quality starts to

degrade with the viscosity profile demonstrates the relation between the viscosity drop

and dispersion degradation as temperature increases.

The results of the rheology tests on the MWNT system are shown in Figure 4-27 and

Figure 4-28. The sample for each rheology test was taken from the same batch that was

prepared for the dispersion test. In the case of TETA and IPD/TETA there is no drop in the

viscosity profile, and consistently the dispersion remained stable during the gelation and

curing of the formulation. Also, as shown in Figure 4-27, TETA had a higher viscosity

compared to the mix of IPD/TETA. The dispersion analysis also confirms the higher

0.1

1

10

100

1000

20 40 60 80 100 120 140

Vis

cosi

ty (P

a.s)

Temperature °C

DDS_Plasma Aradur_Laser Aradur_Plasma

75

stability of the MWNT system cured with TETA as opposed to the formulation cured with

the mix of TETA and IPD.

Figure 4-27. MWNT system viscosity profile for TETA and IPD/TETA as hardener

Figure 4-28. MWNT system viscosity profile for IPD and IPD/N3 as hardener

0

30

60

90

120

1

10

100

1000

10000

0 2000 4000 6000 8000 10000

Tem

per

atu

re °

C

Vis

cosi

ty P

a.s

Time (s)

20

30

40

50

60

70

80

90

100

110

120

1

10

100

1000

10000

0 200 400 600 800 1000

Tem

p (

°C)

Vis

cosi

ty (

Pa.

s)

Time (s)

IPD

IPD/N3

TETA IPD 20% TETA 80%

IPD 50% TETA 50%

IPD 80% TETA 20%

76

On the other hand, for the MWNT formulation with IPD as the hardener, where the curing

process started at room temperature with a temperature ramp to 120 °C, the dispersion

degraded during the first 500 seconds of the experiment (Figure 4-22). According to Figure

4-28 for the IPD sample, this period corresponded to the portion of the viscosity curve

where the viscosity dropped and stabilized before the gelation started. A similar pattern

was observed for the IPD/N3 sample. However, since the viscosity of the MWNT

formulation for the IPD/N3 hardener was higher than the IPD hardener, a more stable

dispersion was observed from the hot-stage test. It should be noted that all the test

parameters for the dispersion and rheology tests were exactly the same.

Applied shear

As discussed earlier, applied shear due to the thermal expansion of the resin is another

key driver of dispersion degradation. In order to understand the relation between applied

shear and dispersion degradation, a series of shear-stage tests were performed. The setup

for these tests was kept the same as for the rheology and hot-stage tests.

SWNT system. The results of the shear stage test for the 0.3 % plasma synthesized Aradur

system are shown in Figure 4-29. The tests were performed under oscillatory mode of the

shear stage (freq.= 0.1 Hz). The gap between the two glass substrates of the shear stage

was set to be 500 μm. Two different types of tests were performed: 1. shear only, at a

constant temperature, and, 2. combined shear and temperature ramp.

The results showed that when there was only shear acting on the thin film and no

temperature profile was applied, no major dispersion degradation appeared. On the other

hand, at an elevated temperature the dispersion quality deteriorated, as expected from

the hot-stage test results. The comparison between the hot-stage results (Figure 4-10) and

the shear-stage results of Figure 4-29, shows that applied shear at an elevated

temperature worsens the dispersion deterioration caused by elevated temperature.

MWNT system. Four main formulations were studied for the MWNT system, including

MWNT system with no hardener, with IPD, IPD/N3, and IPD/TETA as hardener. The gap

between the parallel plates was set to 700 m similar to the rheology test on the MWNT

system. Also, the frequency was set to 1 Hz.

77

Only Shear at 25 °C Shear and Temperature 20 °C/min, hold at 100 °C

0.3 % Plasma Aradur Oscillation 0.1 Hz, Gap 500 μm

0.3 % Plasma Aradur Oscillation 0.1 Hz, Gap 500 μm

Figure 4-29. CNT dispersion stability analysis using the Linkam shear-stage setup

The result of the shear stage test for the MWNT system with no hardener is shown in

Figure 4-30. From image (a) to (c) the formulation was held at 30 °C for 30 min, after

which a temperature ramp at 5 °C/min was followed. Even though, the dispersion

degraded after 30 min at 30 °C, the degradation was very minimal. As soon as the

temperature increased, i.e. viscosity dropped and resin started to expand, the degradation

became more noticeable. It can be concluded that an elevated temperature (reduced

viscosity) played a more dominant role on the degradation of the MWNT dispersion, while

the applied shear helped.

25 °C

50 °C

75 °C

100 °C

t=0

t=45 s

t=90 s

t=135 s

100 μm 100 μm

100 μm 100 μm

100 μm 100 μm

100 μm 100 μm

78

(a) 30 °C (b) at 30 °C after 15 min (c) at 30 °C after 30 min

(d) 60 °C (e) 90 °C (f) 120 °C

Figure 4-30. Shear stage test result for MWNT system with no hardener, 5% strain

The shear stage results for the MWNT system with hardener are shown in Figure 4-31 to

Figure 4-36. For the formulations with hardener, two strain rates were tested, i.e. 5% and

10%.

Since the shear stage needed to be operated at low-viscosity to ensure the safety of the

equipment, for the IPD and IPD/N3 systems, the sample were heated up to 80 °C and then

kept at that temperature for 5 min to maintain the formulation at its lowest viscosity. For

these two formulations, the results show that an increase in the strain rate worsen the

dispersion stability at higher temperature. For the IPD/N3 system, the effect of strain rate

was more noticeable; at 10% strain rate the quality of dispersion dramatically degraded.

(a) 30 °C (b) 55 °C (c) 80 °C

Figure 4-31. Shear stage test result for MWNT system with IPD, 5% strain

100 μm 100 μm 100 μm

100 μm 100 μm 100 μm

100 μm 100 μm 100 μm

79

(a) 30 °C (b) 55 °C (c) 80 °C

Figure 4-32. Shear stage test result for MWNT system with IPD, 10% strain

(a) 30 °C (b) 55 °C (c) 80 °C

Figure 4-33. Shear stage test result for MWNT system with IPD/N3, 5% strain

(a) 30 °C (b) 55 °C (c) 80 °C

Figure 4-34. Shear stage test result for MWNT system with IPD/N3, 10% strain

For the IPD/TETA system, which contained 50% IPD formulation mixed with 50% wt. TETA

formulation, the shear was applied at 30 °C for 10 minutes. The results are shown in

Figure 4-35 for 5% applied strain and in Figure 4-36 for 10% applied strain. As expected,

since the test temperature for this formulation was at 30 °C, no major dispersion

deterioration occurred.

100 μm 100 μm 100 μm

100 μm 100 μm 100 μm

100 μm 100 μm 100 μm

80

(a) 30 °C (b) 30 °C after 5 min (c) 30 °C after 10 min

Figure 4-35. Shear stage test result for MWNT system with IPD/TETA, 5% strain

(a) 30 °C (b) 30 °C after 5 min (c) 30 °C after 10 min

Figure 4-36. Shear stage test result for MWNT system with IPD/TETA, 10% strain

Table 4-4 and Table 4-5 summarize the dispersion characterization tests for the SWNT and

MWNT systems, respectively. Within the SWNT formulations, only three samples resulted

in good dispersion, two of which did not contain a curing agent. The formulation with

good dispersion quality was the Unfunctionalized Laser SWNT in MY0510 (0.3 wt.%). Also,

the processing technique of dissolving DDS in MY and remixing the formulation resulted in

relatively good dispersion quality.

For the MWNT system, all the curing agents that gelled at room temperature resulted in

good dispersion quality. For the IPD system, the higher temperature, required to cure the

resin, deteriorated the dispersion quality. However, the IPD/N3 system led to a relatively

good dispersion quality.

Based on the results of dispersion stability analysis, an increase in temperature results in a

reduced viscosity of the formulation, expansion of the resin, and at the same time,

initiates the curing process. The combination of these processing parameters allowed the

CNTs to move freely before the gelation of the resin and to re-agglomerate because of

their high surface tension.

100 μm 100 μm 100 μm

100 μm 100 μm 100 μm

81

According to the shear stage test results, applied shear during the curing process can

further deteriorate the dispersion quality; however, the drop in the viscosity as

temperature increased during the cure, proved to be the dominant processing parameter

that caused dispersion degradation.

Table 4-4: Summary of dispersion characterization tests for the SWNT system

Fixed parameters Variable parameters Temperature

Cycle Viscosity

drop Cure Dispersion

No hardener 0.3% wt.

50 °C/min to 200°C

Unfunctionalized Laser

Yes Yes No Good

Anionic Laser Yes Yes No Good Unfunctionalized

Plasma Yes Yes No Poor

DDS 0.3% wt.


5 °C/min to 180 °C Yes Yes Yes Poor

100 °C/min to 180 °C Yes Yes Yes Poor

Aradur 0.3% wt.

5 °C/min to 180 °C


Yes Yes Yes Good

Unfunctionalized Plasma

Yes Yes Yes Poor

DDS 0.1% wt. Anionic

MY0510:DDS 100:49, 55, 60, 67

Yes Yes Yes

Poor (The higher the DDS ratio, the

better the dispersion)

0.1% wt. Anionic

MY0510:DDS 100: 60

DDS dissolved and remixed at high

temperature Yes Yes Yes Average

Table 4-5: Summary of dispersion characterization tests for the MWNT system (0.3 wt.%)

Variable parameters

Characteristics Temperature

Cycle Viscosity

drop Cure Dispersion

TETA Cure at room No No Yes Good IPD Cure at high temp Yes Yes Yes Poor

IPD/N3 Fast Cure at High

temperature Yes Yes Yes OK

IPD/TETA Cure at room No No Yes Good

82

4. 4. 2. Fracture toughness test results

An example of a typical load versus displacement curve is shown in Figure 4-37. All

samples including CNT-modified polymers showed a brittle fracture with no improvement

in the ductility of the samples. Therefore, according to ASTM D 5045, the maximum load

was used to find KIc.

Figure 4-37: Typical load-displacement curve for epoxy resin

In Subsections 4. 4. 2. 1. and 4. 4. 2. 2. , the fracture toughness values of the SWNT and

MWNT polymer systems are presented.

4. 4. 2. 1. SWNT system

Effect of different SWNT weight fraction and different types of hardener

Figure 4-38 shows the KIc for two types of hardeners, i.e. Aradur and DDS with different

types and loading of SWNTs with cure cycle # 1 (Table 4-2). The results of the fracture

tests show that the DDS hardener led to higher fracture toughness compared to Aradur.

This was expected as DDS was specifically recommended by the manufacturer as the

optimum curing agent to achieve a toughened MY0510 epoxy system. In general, the

addition of carbon nanotube to the MY0510/DDS system deteriorated the fracture

toughness values. However, for the MY0510/Aradur, the addition of the laser SWNTs

slightly improved toughness by 8%. This result is consistent with the CNT dispersion

0

5

10

15

20

25

0 0.05 0.1 0.15 0.2 0.25 0.3

Loa

d (

N)

Crosshead displacement (mm)

83

stability analysis results shown in Figure 4-29, where the Aradur system containing

Unfunctionalized Laser SWNT showed relatively more stable dispersion quality during the

cure compared to DDS system. The laser SWNT dispersion with the Aradur as the

hardener was more stable compared with the other samples. Similarly, the reduction in

fracture toughness, for the plasma SWNT with Aradur and also the DDS system can be

interpreted as a result of poor CNT dispersion quality of the nanotubes (Figure 4-10). The

increase in the SWNT weight fraction with the MY0510/DDS decreased the fracture

toughness. This result is clearly in contradiction with the modelling work presented in the

Chapter 3.

Figure 4-38. Fracture toughness test results, MY0510 epoxy system with SWNT; Aradur (left), and DDS (right)

Effect of DDS to Resin Ratio

The fracture toughness characterization results of the DDS with different MY0510 to DDS

ratios are shown in Figure 4-39 and Figure 4-40. The former was cured according to cure

cycle # 1 and the latter was cured according to cure cycle # 2. Those columns in white

represent the fracture toughness values of the neat resin for each ratio.

The results shows that among all the samples, the neat samples with 67:100 DDS to resin

ratio yielded the highest average fracture toughness, however, the error bars are

relatively large for these specimens. Also, there is large variation in the fracture toughness

of the neat resin cured with different cure cycle.

1.48 1.381.14 1.28

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

Neat Plasma 0.4% Plasma 1.6% Laser 0.4%

0.72 0.780.57

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

Neat Laser 0.3% Plasma 0.3%

Kic

(M

Pa√

m)

DDS Aradur

84

Figure 4-39. Fracture toughness test results for MY0510 / 0.1% Anionic SWNT with different DDS: MY0510 ratios, cure cycle # 1

Figure 4-40. Fracture toughness test results for MY0510 / 0.1% Anionic SWNT with different DDS: MY0510 ratios, cure cycle # 2

1.19 1.10 1.031.24

1.091.31

1.51 1.42

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

Kic

(M

Pa(

m0.

5))

49:100 55:100 60:100 67:100

NEAT MY0510 SWNT

0.981.19

0.83 0.891.01

1.28

0.991.19 1.13 1.18

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

Kic

(M

Pa(

m0.

5))

49:100 55:100 60:100 (1) 67:100

NEAT MY0510 SWNT

60:100 (2)

85

The longer cure cycle (cycle # 1) resulted in higher fracture toughness values for the base

neat resins at all hardener to resin ratios, whereas cure cycle # 2 yielded lower base KIc.

However, addition of SWNTs in cure cycle # 2 consistently increased the modified polymer

fracture toughness. Table 4-6 summarizes the percentage change in fracture toughness

values to the base neat resin in each hardener to resin ratio.

Table 4-6: Summary of the fracture toughness percentage change compared to the base resin

Resin : Hardener ratio 100:49 100:55 100:60 100:67

Cure cycle # 1 - 8 % 20 % 20 % - 6 %

Cure Cycle # 2 22 % 7 % Batch 1 Batch 2

4 % 27 % 20 %

For the cure cycle # 1, SWNTs only improved the fracture toughness of the system at

100:55 and 100:60 ratios, whereas in cycle # 2, addition of the SWNTs at all hardener to

resin ratios improved the fracture toughness. The 100:60 proved to be the optimum ratio

as the results were reproducible. Another aspect that can be concluded from the results is

the large scatter that exists in the result for each sample due to the brittle nature of the

fracture.

When the optimum hardener-to-resin ratio was identified, the effect of different weight

percentages of SWNTs was studied. For the 100:60 MY to DDS ratio, the result of different

SWNT wt.% is shown in Figure 4-41. As can be seen in the figure, a consistent fracture

toughness enhancement was achieved by increasing the weight fraction of SWNT in the

base polymer. By addition of only 0.3% SWNT, 38% fracture property enhancement was

achieved.

Another very important aspect of this improvement is the reduced scattering of the

results. For brittle resin systems, due to the sensitivity of the specimen to micro-scale

defects the error bars for fracture tests are usually very large. The error bars on Figure

4-41 was considerably reduced as the weight fraction of SWNTs was increased. Smaller

error bars can be considered as an improved crack growth behaviour: from unstable, rapid

growth to a more stable and consistent crack propagation.

86

Figure 4-41. Fracture toughness test results for MY0510 / SWNT system with different Anionic SWNT wt.% (100:60 DDS ratio), Cure cycle # 1

4. 4. 2. 2. MWNT system

Effect of hardener type

The results of the fracture toughness test for the MWNT system is shown in Figure 4-42

and Figure 4-43. The former includes the results of IPD, TETA, and the mix of IPD/N3, and

the latter shows the result of the mix of IPD and TETA comprising of IPD:TETA ratios of

20/80, and 50/50, and 80/20 wt%.

Addition of the MWNTs to all the base formulation slightly improved the fracture

toughness values, except for the case of the manufacturer’s recommended hardener, IPD.

The percentage changes in the fracture toughness values of MWNT modified polymer

compared to the base formulation are summarized in Table 4-7.

Table 4-7. MWNT system fracture toughness percentage change compared to the base resin

Hardener IPD TETA IPD/N3 IPD(20)

TETA(80) IPD(50)

TETA(50) IPD(80)

TETA(20)

Fracture toughness % change

- 11 % 12 % 17 % 4 % 5 % 9 %

1.091.31 1.38

1.51

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

Neat 0.1% 0.2% 0.3%

Kic

(M

Pa(

m0.

5))

87

Figure 4-42. Fracture toughness test results, bisphenol-A with 0.3% wt. MWNT and different types of hardener

Figure 4-43. Fracture toughness test results, bisphenol-A with 0.3% wt. MWNT with different IPD:TETA ratio

1.020.91

0.76 0.85

1.131.32

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

Kic

(M

Pa(

m0.

5))

IPD TETA IPD/N3

NEAT bisphenol A MWNT

0.76 0.79 0.79 0.83 0.79 0.86

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

Kic

(M

Pa(

m0.

5))

20:80 50:50 80:20

NEAT bisphenol A MWNT

88

The idea behind using a room temperature cure was to maintain the dispersion quality at

room temperature and to lock the MWNTs up to the gelation point of the resin. For the

samples containing TETA as the curing agent, the dispersion quality during the curing

process is shown in Figure 4-16. The dispersion quality remained constant leading to

improved fracture toughness values. Whereas in the IPD system, the dispersion degraded

during the first heating cycle, as shown in Figure 4-17, and consequently resulted in

reduced fracture toughness for MWNT-modified formulations. The dispersion degradation

started around 50 – 55 °C and stayed the same after 75 °C.

Nevertheless the improvement in the fracture toughness values of the MWNT system with

TETA as hardener, the toughness values for the base formulation is 25% lower than the

fracture toughness of the MWNT system with IPD. Therefore, a mix of IPD with TETA was

considered to evaluate the potential of mixing two hardeners. As shown in Figure 4-43,

none of these formulations resulted in a higher fracture toughness improvement when

compared with the MWNT system with TETA.

Even though, there is an improvement for most of the samples in the average fracture

toughness values, the high error bars should be noted for each set. The high variation in

the fracture toughness values can be related to the brittleness of the resin.

The results clearly showed that the CNT-modified resins should be regarded as a new

material formulation. These formulations should be optimized regardless of the

recommended hardener and curing procedure by the manufacturer.

4. 4. 2. 3. SEM analysis of the fractured surfaces for nano-modified polymers

SWNT modified polymer.

The SEM images of the fractured surface, taken by a Hitachi SU-8000 Cold Field Emission

SEM, are shown in Figure 4-44 for the neat MY0510. This is the base sample that yielded

highest fracture toughness improvement with MY to DDS ratio of 100:60 (Table 4-6). As

can be seen in the SEM images, the fracture surface was very smooth and showed no

toughening feature. There were only on very few cavities (Figure 4-44 (c, d)) that could be

due to the impurities that entered the mould during the sample preparation.

89

Figure 4-44. SEM analysis of the fractured surface of neat polymer (SWNT system) (MY: DDS ratio 100: 60)

However, when 0.1 wt.% of SWNTs were added to the neat sample, the morphology of

the surface changed, which resulted in fracture toughness improvement. This change of

morphology is shown in SEM images of the fracture surface at different magnifications in

Figure 4-45. The major difference between the neat and SWNT modified specimens was

the roughness of the surface. The roughness of the SWNT modified samples were

considerably increased leading to higher energy consumption to grow the crack.

Another observation from the lower magnifications images (Figure 4-45(a-c)) was a

relatively good dispersion of the SWNTs on the fracture surface. Even though there were

agglomerations of the SWNTs (SWNT islands), the islands were well dispersed.

Also, there were several toughening mechanisms observable at higher magnifications

(Figure 4-45 (d-i)) such as crack pinning and crack deviation (highlighted on Figure 4-45

(d)). The crack was pinned when reached a SWNT island, and then deviated to the two

sides of the SWNT island.

b a

c d

90

SWNT pull-out was another toughening mechanism observed in Figure 4-45 (f-i). SWNTs

with different orientation angle with respect of the crack growth plane were pulled out,

and contributed to the increased toughness values. These pulled out SWNTs as well as the

cavities that were caused by the pull-out process are highlighted in the images with the

black arrows.

a

b c

d e

91

Figure 4-45. SEM analysis of the fractured surface of 0.1% SWNT modified polymer (MY: DDS ratio 100: 60)

MWNT modified polymer.

A similar trend was observed for the MWNT system. The result of SEM analysis for those

samples that yielded the highest fracture toughness values (MWNT with IPD/N3) is shown

in Figure 4-46. The fractured surface was very rough, and there were several MWNT

islands on the surface. By zooming into those islands, Figure 4-46 (e, f), the MWNT pull-

out were also apparent. Other than MWNT pull out, several MWNTs were peel off the

surface, as highlighted in image (e) with black arrows. These are MWNTs with orientation

angel parallel to the crack growth plane.

f g

h i

92

Figure 4-46. SEM analysis of the fractured surface of 0.3% MWNT modified polymer (Hardener IPD/N3)

c

a

d

b

e f

93

4. 5. Correlation between the model and the experimental results

In this section, the experimental fracture toughness values are compared to the modelling

predictions. This comparison was done for the MY0510 / SWNT system with different wt.

% of Anionic SWNT with 100:60 DDS ratio, Figure 4-41. Since the model predicts the

increased energy resulted from bridging contribution, the stress intensity factor, K ic is

transferred into the critical strain energy release rate, Gic using [158],

2 2(1 ) icic

KG

E

Equation 4-6

where = 0.3 and is the passion ratio, and E = 3.4 GPa and is the elastic modulus. Figure

4-47 shows the results of the critical strain energy release rate, Gic, for the specimens in

Figure 4-41. By subtracting the GIc values for the SWNT modified specimens from the neat

resin sample in Figure 4-47, the toughening contribution of SWNTs can be estimated

(Figure 4-48(a)).

Figure 4-47. The critical strain energy release rate for the results of Figure 4-41

Figure 4-48(a) illustrates the change in the strain energy release rate, Gic, after the

addition of SWNTs predicted by the model and verified experimentally. The predicted CNT

bridging from the model is shown in Figure 4-48(b). The figure shows the bridging

contribution estimated based on the model for randomly dispersed CNTs. As the input for

the model, the interfacial bonding of 47 (MPa) [153], and u(CNT) = 40 (GPa) was used.

The elongation to fracture of SWNTs was assumed to be 10%. The dashed line represents

373.73

524.66 564.05671.25

0

200

400

600

800

Neat 0.1% 0.2% 0.3%

Neat SWNT modified

GIc

(J/m

2)

94

0.1% volume fraction of CNTs, and the solid line represents a higher volume fraction of Vf

= 0.3%. Also, the solid line represent a larger SWNT radius, r=10 nm, whereas for the

dashed line, r=1 nm.

The model (Figure 4-48(b)) predicts potential of up to 250 J/m2 fracture toughness

enhancement via CNT bridging for an average SWNT length of 200 m. On the other hand,

the maximum enhancement from the experiment was approximately 300(J/m2). The

difference can be explained by considering that the model only predicted the contribution

of CNT bridging. However, other toughening mechanisms such as crack pinning and crack

deviation exist due to the addition of CNTs. Also, it should be noted that to achieve more

accurate prediction from the model, each of the input parameters should be measured for

the SWNT and the epoxy system that was used in this study.

(a) Toughening contribution of SWNTs experimental versus modelling predictions

(b) Effect of bridging based on modelling

Figure 4-48. Bridging contribution, model vs. experiment

4. 6. Summary and Discussions

This chapter presented the effect of processing parameter, such as temperature and cure

cycle on the dispersion quality of CNT modified epoxies. The fracture toughness

characterization of SWNT and MWNT modified epoxies were then presented and

correlated to the dispersion quality of the formulations. The results showed the

importance of good dispersion quality on the fracture toughness improvement. Also,

curing at elevated temperatures resulted in viscosity drop and caused dispersion quality

degradation, and consequently lowered the fracture toughness values. The results also

150.92190.32

297.52

90

250

0

100

200

300

400

500

0.1% 0.2% 0.3% 0.1% 0.3%

Experimnet Model

0

50

100

150

200

250

300

0 50 100 150 200

G

Ic (J

/m2)

J bri

dg

ing (

J/m

2)

=47 MPa u = 40 GPa

= 10 %

r = 1 nm, Vf = 0.1%

r = 10 nm Vf = 0.3%

L (m)

95

showed that CNT modified systems should be regarded as a new system and optimized

accordingly.

For the SWNT system, the results showed that the curing process played a key role in the

effectiveness of the SWNTs. Curing cycle was a major source of complexity in CNT-

modified polymers [159, 160], as it affects nano-scale polymerization of the monomers.

Future research on nano-scale curing process monitoring can lighten up potential source

of dispersion degradation. Both type of curing agent and hardener to resin ratios were

studied. Comparing a solid powder hardener, i.e. DDS, with the liquid curing agent, i.e.

Aradur, DDS dissolved in the resin system at 100 °C and consequently resulted in resin rich

locations (locally no CNTs) that affected the dispersion quality. As the DDS hardener to

resin ratio increased, the effect of SWNTs on fracture toughness improved. However,

addition of more hardener above the 100:60 ratio had a negative effect.

For the MWNT system, different types of hardener were studied. It can be concluded that

the chemical process of polymerization during the curing process is not the main driver of

dispersion degradation since the dispersion quality remained constant when the resin was

cured at room temperature. Viscosity drop and thermal expansion of the resin were the

drivers of dispersion degradation. The most effective solution to solve the problem of

dispersion degradation during the cure was proved to be curing at low temperature up to

the gelation point. This process improved the CNT dispersion stability, however, on the

downside, this strategy resulted in an overall low fracture toughness values. According to

the fracture toughness characterization, even though we achieved up to 27%

improvement by addition of CNTs to the base polymer, due to the large variation in the

fracture toughness values of the formulations, extensive research is still required to

achieve major improvement.

Finally, the SEM analysis of the fractured surfaces showed several toughening mechanisms

that were contributing to the increased fracture toughness. These mechanisms, include

CNT pull-out, CNT peel-off, and crack pinning and deviation when reaching agglomerated

CNTs (CNT islands).

96

Chapter 5. Carbon Nanotube Modified Carbon Fibre Composites

5. 1. Summary

In this chapter, the effect of Carbon Nanotubes as a reinforcement of laminated

composites will be studied. In Chapter 4, the fracture toughness of CNT modified polymers

has been studied, and key processing issues have been discussed. It is of great interest to

further explore the effect CNTs by incorporating CNT modified resin in composite

structures. Two CNT-modified epoxies were used to manufacture carbon fibre laminates

by resin film infusion and prepreg technologies.

5. 2. Materials

5. 2. 1. SWNT Modified Prepreg (SWNT composites)

The same material preparation procedure as explained in Chapter 4.2. for SWNT system

was used to prepare the base resin for composite laminate preparation. The loading of

SWNT in the formulation was 0.1 wt. %.

SWNTs were produced at the NRC Steacie Institute for Molecular Sciences (NRC-SIMS)

using a double-laser method reported previously, [161]. The nanotubes were then

negatively charged (Anionic) and were mixed with an epoxy polymer (A proprietary epoxy

system optimized for toughness and consisting of 4 standard bi-functional epoxy resins, 2

catalysts, one plasticizer and one hardener was used for the SWNT composites). The

formulation was then shipped to Newport Adhesive Co. in order to prepare carbon fibre

prepreg using a drum winder. Unidirectional carbon fibre prepregs with a thickness of

0.156 mm was then prepared for both non-modified (baseline) and SWNT-modified resins.

The details of the prepreg manufacturing and the neat resin formulation are protected by

Newport.

97

5. 2. 2. MWNT Modified Resin Film (MWNT composites)

For the MWNT composites, the material was provided by Nanoledge Inc. The base epoxy

resin was a hot-melt resin (solid form at room temperature), which was a blend of liquid,

semi-solid and solid bisphenol A. The curing agent was dicyandiamide with a hardener-to-

resin ratio of 5:100.

The MWNTs were the same as those used in Chapter 4 and were commercially available

from Bayer [162]. Two different formulations were studied:

1. MWNT composite (2377): this formulation only contains 0.3 wt. % MWNTs mixed

with the resin. The MWNTs were functionalized to develop physical interaction

with the matrix with no covalent linkage.

2. MWNT + nano-filler composites (2378): this formulation was prepared in a similar

fashion to the 2377 MWNT composite system. However, additional proprietary

soft nano-fillers (4 wt. %) were added to the mixture to absorb energy as crack

propagates.

After the preparation of the resin formulation, thin resin films (semi-solid at room

temperature) were manufactured using a three roll coater. The areal weight of the resin

films were 225 gram per square meter with a thickness of 205 µm. These two resin film

formulation were then used to impregnate the carbon fibres manufactured by Jb martin,

(TC-18-N).

5. 3. Experimental Procedures

5. 3. 1. Test Plan

This investigation focused on the two principal modes of delamination growth, Mode I and

Mode II. Specimens were tested under pure Mode I (interlaminar tension) and pure Mode

II (interlaminar shear) according to the ASTM D5528-01 standard. In order to understand

the effect of CNTs on the hybrid composite system, the fracture toughness of the polymer

used in each composite system was characterized under 3 point bending as explained in

Section 4.3 prior to testing the composites delamination properties. A total of five samples

were tested in Mode I, Mode II and for the modified resin fracture toughness.

98

5. 3. 2. Specimen dimensions

Rectangular double cantilever beam (DCB) specimens and rectangular end-notched

flexure (ENF) specimens were used for Mode I and Mode II fracture tests, respectively.

The specimens were cut from panels according to the ASTM D5528-01 standard [163]. The

lengths of the initial delaminations and the key specimen dimensions are presented in

Table 5-1. A 10-micron Teflon film was inserted at the mid-plane of all specimens to create

an initial delamination. All specimens were 20 mm wide (w) and 4.4 mm thick (t).

Table 5-1. Specimen dimensions refer to figures

Specimen Type

Average Length

Length of Teflon Initial Delamination

Average Width

Average Thickness

l (mm) Insert lTeflon (mm) Length ao (mm) w (mm) t (mm)

Mode I (DCB) 140 60 50 20 4.5

Mode II (ENF) 170 70 20-40* 20 4.5

*Three delamination lengths were considered for each specimen (20 mm, 30 mm and 40 mm).

5. 3. 3. Mode I Interlaminar Fracture Toughness

A schematic of a Mode I interlaminar fracture toughness specimen is shown in Figure 5-1.

The Double Cantilever Beam (DCB) specimens were rectangular, uniform thickness

laminated composites, with a non-adhesive Teflon insert on the mid-plane that served as

a delamination initiator. Opening forces were applied to the Mode I DCB specimens

through loading tabs that were fixed onto the initially delaminated end, as shown in the

figure. The tabs spanned the entire width of the specimen, and a steel pin linked them to

a loading fixture of the testing equipment (MTS insight). The initial delamination length,

ao, was measured from the center of the pinhole to the end of the Teflon insert.

Figure 5-1. DCB specimen

Loading block

Teflon insert

Adhesive tape

Length (l)

Initial delamination length (ao)

Thickness (t)

Width (b)

Length of block

Length of Teflon insert (lTeflon)

99

5. 3. 4. Mode II Interlaminar Fracture Toughness

Mode II specimens were subjected to three-point bending loads and were simply

supported by two rollers. The specimens were rectangular beams (Figure 5-2). The initial

delamination length, ao, was measured from the center of the supporting roller on the

delaminated end to the end of the Teflon insert.

Figure 5-2. Mode II specimen

5. 3. 5. Specimen preparation

For the interlaminar fracture toughness characterization of CNT modified composites,

panels of Carbon Fibre composites were manufactured and specimen were cut from the

panel according the required dimensions. For each formulation, two panels were made: 1.

Base carbon fibre laminate, 2. CNT-modified carbon fibre laminate.

Prior to the Mode I and Mode II interlaminar fracture tests, the fracture toughness of the

polymer resins that were used to impregnate the fibres were experimentally measured.

The polymer resin samples for fracture toughness tests were manufactured according to

the described procedure in Chapter 4. However, for the resin film system, the procedure is

explained in detail in subsection 5. 3. 5. 2.

5. 3. 5. 1. Composite Panel Manufacturing

Laminate Preparation. For the SWNT prepreg system, a panel of 11”×6.5”× 0.148”

(280x165×3.8 mm3) was manufactured for each material. The thickness of prepreg sheets

was 0.156 mm (0.0062”). As a result, a panel was manufactured from 24 layers of

unidirectional prepregs with a Teflon film of 0.0005×2.5” ×11” inserted between the 12th

and 13th layers according to Figure 5-4.

Length (l) = 170 mm

Teflon insert = 70 mm

Thickness (t) = 4. 5 mm

Width (b) = 20 mm

100

For the MWNT resin film infusion system, 15”×5”× 0.177” (380×130×4.5 mm3) panels were

manufactured. For this purpose 8 layers of Fibre/resin film were stacked as shown in

Figure 5-3.

Figure 5-3: Stacking procedure for the MWNT system

A Teflon thin film was placed in the mid-plane of the laminates. The size of the panels is

shown in Figure 5-4 and hatched areas represent the Teflon insert. The detail of the

laminate preparation is shown in Figure 5-5. Each panel was then vacuum bagged

according to strategy given in Figure 5-6.

Figure 5-4: Panel size and Teflon insert location

380 mm

130 mm

70 mm 50-55 mm

Fibre direction

101

Lay-up Teflon insert Bagging Autoclave ready

Figure 5-5: Lay-up of the panels and bagging

1. Carbon Fibre / Resin 2. Sealing compound 3. Fibreglass bleeder

4. Release film 5. Breather 6. Bagging film 7. Vacuum port

Figure 5-6: Bagging sequence

The samples were cured in an autoclave at 100 psi (applied at the beginning of the cure

cycle). The cure cycle was 2 hrs hold at 130 °C and 2 hrs hold at 200 °C with the ramp rate

of 3 °C/min.

Trimming and Cutting

From the panels, DCB specimens were cut according to the ASTM D5528-01 standard.

After the cure, the panel edges were trimmed to remove excess material. 6 DCB

specimens and 6 ENF specimens were obtained from each panel (Figure 5-7). A water-

cool-diamond table saw was used to cut the specimens and to trim the plates.

102

Figure 5-7: Cutting pattern for the DCB and ENF samples

Once the specimens were cut, their edges were polished in order to accurately locate the

end of the Teflon insert (the delamination tip). The dimensions of the specimens were

then measured, and loading tabs were bonded onto the ends of DCB specimens with a

double-sided tape adhesive (Metlbond 1113). A fixture was used to help position the tabs

(Figure 5-8(b)). The fixture with specimens was then placed in an oven at 100°C for 120

min to allow the tape adhesive to cure. This cure cycle was selected to ensure that the

temperature would be below the glass transition temperature of the resin in the

specimens, which was approximately 140°C. The evolution of the sample preparation

process is illustrated in Figure 5-8.

(a) A composite Panel (b) Loading-tab positioning fixture (c) Mode I specimens

Figure 5-8. Sample preparation process from a panel to Mode I and II specimens

5. 3. 5. 2. Resin Film Sample Preparation

In order to prepare fracture toughness samples for the SWNT-modified polymer, the

sample preparation procedure as described in detail in Section 4.3 was followed, because

Mode2-4

Mode2-5

360 mm

110 mm

20 mm

130 - 135 mm

Mode2-3

Mode2-2

Mode2-1

Mode1-4

Mode1-5

Mode1-3

Mode1-2

Mode1-1

103

the resin was in liquid form at room temperature. For the MWNT system, the resin films

were in solid form at room temperature; therefore, 16 layers of resin films were stacked

and placed inside the Teflon part of a mould as shown in Figure 5-9. The pressure was

applied at room temperature and the samples were cured. The cure cycle was 2 hrs hold

at 130 °C and 2 hrs hold at 200 °C with the ramp rate of 2.5 °C/min. Once the samples

were cured, the specimens for fracture toughness were cut from the plate (60x40×2.5

mm3) according to the dimensions discussed in Section 4.3.

Figure 5-9. Resin film sample preparation

5. 3. 6. Mode I interlaminar fracture toughness test and data analysis

The test specimens were stored and tested at Standard Laboratory Atmosphere of 23±3°C

and 50±10% humidity. Opening forces were applied to the Mode I DCB specimens through

loading tabs that were fixed onto the initially delaminated end. A steel pin linked these

tabs to a fixture on the testing system, as shown in Figure 5-10.

Mode I test setup Specimen connected to testing fixture Crack growth monitoring

Figure 5-10. Fixture linking MTS testing system to Mode I DCB specimen tabs

Mode I tests were performed in displacement control at a loading rate of 0.5 mm/min and

unloading rate of 25 mm/min. The crosshead displacement and the corresponding

104

reaction force exerted by the specimens were captured at 2 second intervals with a data-

acquisition software (MTS TestWorks 4). Load and displacement were then related to

delamination length as measured with a ruler on the specimen edges. A typical load

versus displacement graph is presented in Figure 5-11.

Figure 5-11. Typical load – displacement curve for a Mode I fracture test of the resin film system (2377-1)

Delamination initiates when the initial portion of the load – displacement curve deviates

from linearity. This critical load, represented by a green circle on Figure 5-11, was used to

generate the value for initiation fracture toughness. Delamination continued to grow in an

instantaneous and unstable manner, which translated into a saw-tooth relationship

between load and crosshead displacement. Delamination growth occurred at the top of

the saw-tooth, as indicated by red triangles, where the material instantaneously released

strain energy. Incremental delamination growth was on average 1-5 mm in length, and the

delamination was allowed to grow for 55 mm. It was therefore possible to capture over

ten distinct delamination growth increments for all specimens. A value for crosshead

displacement (δ), load (P) and delamination length (a) is known at each point represented

by the circle and triangles on Figure 5-11. A value for fracture toughness may therefore be

associated to each of these points, using specimen geometry and the data reduction

techniques presented in ASTM D5528-01. Three data reduction method for calculating GIc

values are described in the ASTM standard: a modified beam theory (MBT), a compliance

calibration method (CC), and a modified compliance calibration method (MCC). The detail

of each method is given in Appendix B.

0

10

20

30

40

50

60

0 5 10 15 20 25 30 35 40

Load (N)

Crosshead Displacement (mm)

Loading

Propagation points

Initiation point

105

A resistance to delamination curve (R-curve) was then obtained by plotting all the fracture

toughness values onto one graph, i.e. GIc values as function of delamination length. A

typical Mode I R-curve is presented in Figure 5-12.

For each material system, 5 specimens were tested. The average value of the G Ic initiation

for the CNT modified specimens was compared to the average GIc initiation value of the

neat resin specimens. The GIc propagation value for each material system was calculated

as the average of the propagation values (red triangles) for the 5 specimens. This G Ic

propagation value was then used to verify the effect of CNT modification of composite

panels.

Figure 5-12. Typical Mode I R-curve for the MWNT composites (2377)

5. 3. 7. Mode II interlaminar fracture toughness test and data analysis

Bending forces were applied to the Mode II End-Notched Flexure (ENF) specimens through

a three-point bending setup, as shown in Figure 5-13. Two Mode II interlaminar fracture

toughness values were calculated for each sample: 1. Non-precracked (NPC) toughness

which is an interlaminar fracture toughness value that is determined from the pre-

implanted Teflon insert, and 2. Precracked (PC) toughness which is determined after the

delamination advanced from the pre-implanted Teflon insert. Delamination growth was

highly unstable in Mode II, therefore, only initiation values for fracture toughness could be

obtained.

100

300

500

700

900

1100

45 55 65 75 85 95 105

GIC

Delamination Length (mm)

Propagation fracture toughness Initiation fracture toughness

106

Figure 5-13. Mode II fracture test fixture on MTS Insight setup

The first fracture test was performed considering the end of the Teflon insert as the

delamination tip (non-precracked fracture test). The initial delamination length was set to

30 mm (a0) from the crack tip. Displacement was applied to the specimen until a drop in

load occurred, and the specimen was then unloaded. The end of this delamination

became the new tip. The specimen was then repositioned such that the distance between

the new tip and the center of the support roller on the delaminated end was equal to the

original initial delamination length of 30 mm. The test was restarted with this new

configuration (precracked fracture test). Through data reduction, two candidate values

for initiation toughness were obtained in both configurations, for a total of four. The

lowest value that passed a qualification process was conserved and used as the GIIc.

Typical load-displacement curves for non-precracked and precracked fracture test are

shown in Figure 5-14.

The two fracture tests (non-precracked and precracked fracture tests) were both

preceded by two compliance calibration tests. The objective was to quantify the

compliance of each configuration which is used to find the Mode II interlaminar fracture

toughness. The detail of data reduction to find the mode II interlaminar fracture

toughness is given in Appendix B.

107

Figure 5-14. Typical Load-displacement curve for NPC and PC Mode II tests

5. 3. 8. SEM Image Analysis

The SEM images of the fractured surface were taken with a Hitachi SU-8000 Cold Field

Emission SEM. For low volume fraction CNT-modified polymers that are non-conductive, a

low accelerating voltage is required to eliminate surface charging of the samples that

distort the SEM images. This ultra-high-resolution SEM is optimized for nanostructure

characterization at low accelerating voltage (< 1kV). Low-kV SEM allows imaging of

insulating samples without coating the samples with conductive metal (e.g. Gold) that can

deteriorate surface morphology.

5. 4. Results and Discussions

In this section, the results of resin characterization are presented, followed by results of

composite samples characterization. It is important to understand the effect of CNTs on

the two resin systems and then correlate the results to the hybrid composite systems.

5. 4. 1. Resin Characterization

5. 4. 1. 1. Fracture toughness

The results of mode I fracture toughness test for the two polymer systems are shown in

Figure 5-15 and Figure 5-16. These tests are the results of plain strain fracture toughness

0

50

100

150

200

250

300

350

400

450

0 0.5 1 1.5 2 2.5

Load

(N)

Displacement (mm)

NPC

108

test according to the ASTM D5045, under the fullam 3-point bending fixture. All the

dimension and sample preparation procedure is similar to the procedure outlined in

Section 4.3.

Figure 5-15. Fracture toughness of SWNT modified polymer

Figure 5-16. Fracture toughness of MWNT modified resin film

For the SWNT system, addition of the CNTs reduced the critical stress intensity factor, K ic

by 12%. For the MWNT system, both 2377 and 2378 samples improved the fracture

toughness of the base resin by 17% and 5%, respectively. The 2377 demonstrated the

highest fracture toughness values.

2.271.99

0

0.5

1

1.5

2

2.5

3

Neat Anionic SWNT 0.1%

Kic

(M

Pa(

m0.

5))

1.722.01

1.81

0

0.5

1

1.5

2

2.5

3

NEAT 2377 2378

Kic

(M

Pa

(m0

.5))

109

In order to understand better the results, the fractured surfaces of the samples were

studied under the SEM, to identify potential toughening mechanisms.

5. 4. 1. 2. Fractography

The SEM images of the fractured surface taken with the Hitachi SU-8000 Cold Field

Emission SEM are shown in Figure 5-17 for the neat polymer of the MWNT system. The

images show the smooth fractured surface at different magnifications. As can be seen on

the images, there were river lines on the surface confirming a brittle fracture, but aside

from the river lines, the fracture surface was very smooth and shiny, with no specific

feature.

Figure 5-17. SEM images of the fractured surface of the neat polymer samples (MWNT system) at different magnifications

For the MWNT modified specimens, a rough surface with several toughening features on

the surface, such as crack pinning and CNT pull out was observed. Figure 5-18 compares

the fracture surface of the 2377 specimens with the 2378 specimens at different

a b

c d

110

magnification levels. In general, both samples contained river lines indicating a brittle

fracture mechanism. However, the surface of the 2377 specimen was rougher than the

2378 and more CNT pull-out was observed for the 2377 sample. These features on the

SEM images confirm the higher fracture toughness value for the 2377 samples as shown in

Figure 5-16. Also, as highlighted in Figure 5-18(d) (the red square) in some areas, the

agglomerated MWNTs caused a local polymer failure creating a concave surface. This

process introduced a new energy dissipation mechanism for fracture toughening.

Figure 5-18 (f, l) can also be used to find the diameter of the MWNTs in the nano-modified

resin film. This diameter can be approximated to be around 40-50 nm.

2377 2378

a g

b h

c i

d j

111

Figure 5-18. SEM images of the fractured surface of the 2377 and 2378 MWNT system; images (a – f) are for the 2377 sample (increased magnification from (a) to (f)); images (g – l) are for the

2378 specimen with increased magnification from (g) to (l)

5. 4. 2. Hybrid Composite Characterization

5. 4. 2. 1. Mode I delamination properties

Typical load-displacement curves comparing neat and CNT modified DCB samples, for both

SWNT and MWNT systems, are shown in Figure 5-19. Both neat and CNT modified

samples demonstrated a linear load–displacement relation up to the crack initiation point.

However, the CNT modified samples in both SWNT and MWNT systems sustained a higher

initiation load. The load-displacement data were used to generate the resistance curves

shown in Figure 5-20. The results of GIc initiation and propagation for all the specimens

were then averaged and reported as the Mode I interlaminar fracture toughness, shown in

Figure 5-21.

In the case of SWNT system, the initiation value increased by 3% compared with the neat

DCB samples, and the average propagation value increased by 13%. For the case of MWNT

resin film DCB samples, the 2377 sample showed a 33% and 48% increase in the initiation

and propagation Gic, respectively, compared to the base laminate. For the 2378 samples,

Gic initiation and propagation values were increased by 143% and 106%, respectively. The

MWNT system contained two types of CNT modification. 2377 contained only MWNT

whereas the 2378 contained MWNT as well as a proprietary thermoplastic toughener. The

thermoplastic toughener (the chemistry of which is protected under a provisional patent

e k

f l

112

by Nanoledge Inc) was chemically treated to improve the fracture toughness. For the

MWNT system, these improvements clearly showed major toughening contributions. A

similar increase of 13% in mode I fracture toughness was reported by Romhany and

Szebenyi [164] for a MWNT loading of 0.3 wt%. In another work, Karapappas et al. [147]

demonstrated that 1 wt% loading of MWNT can result in 60% improvement; however, for

small loading of 0.1 wt.% a slight reduction in both mode I and mode II fracture toughness

values was reported.

An important observation from the Mode I delamination results (Gic) is in the

effectiveness of CNTs when compared to the fracture toughness of the CNT-modified resin

(SENB samples). Addition of the SWNT to the base resin decreased the fracture toughness

of the polymer by 10%. In contrast, the addition of SWNT increased the initiation Gic by

3%. In the case of MWNT system, the polymer toughness for the 2377 and 2378 increased

by 17% and 5%, whereas for the composite DCB samples, the initiation Gic increased by

33% and 143%, respectively. This trend can be explained by looking at the source of

energy dissipation as the crack grows, i.e. fibre/matrix debonding [152]. In the DCB

samples, the nature of the interaction among the CNTs, the polymer and the fibre is

different from the interaction between the resin and CNTs in the SENB samples [165, 166].

Also, the carbon fabric acts as a network that limits the movement of CNTs during the cure

process leading to a more uniform CNT dispersion in DCB specimens.

Based on the fracture toughness values of the polymers (SENB tests, (Figure 5-15 and

Figure 5-16)), the higher the base polymer fracture toughness values, the higher was the

initiation Gic values of the DCB specimen. The base epoxy polymer in the SWNT was

tougher when compared with the MWNT system. Consequently, the initiation Gic for the

SWNT system was higher than the initiation Gic for the MWNT system as shown in Figure

5-21. Bradley and Cohen showed that the composite fracture toughness is a function of

the base resin fracture toughness [167]. Therefore, for the initiation Gic when there were

no active toughening mechanisms, the resin properties were dominant in the fracture

toughness of the DCB samples [12, 167]. It should also be noted that the standard loading

rates and the crack tip geometry are different between the SENB sample and DCB samples

[168, 169]. The crack tip geometry influences GIc results particularly for crack initiation.

The SENB specimens contained a sharp, natural pre-crack whereas DCB specimens

contained a Teflon insert (crack initiator) which may have a blunting effect which may

raise GIc for crack initiation.

For both SWNT and MWNT modified composites, the average mode I interlaminar

initiation toughness values were lower than the propagation values. The reason for the

higher propagation values was that the first incremental delamination started from the

113

end of the Teflon insert (crack initiator), whereas, when the crack propagated, toughening

mechanisms such as fibre/CNT bridging kicked in requiring higher energy to further grow

the crack [170, 171]. These toughening mechanisms gave rise to the apparent fracture

toughness values [172]. Also, since the percentage increase in the propagation values was

larger than the initiation values, it can be concluded that there were toughening

mechanisms due to the addition of CNTs.

Unlike the SWNT system, the MWNT system exhibited a rising r-curve which was a sign of

CNT-bridging and other toughening mechanisms such as crack pinning and bowing. The

MWNT system demonstrated higher delamination toughness improvement compared

with the SWNT system. This behaviour could be explained by differences in the

manufacturing process leading to variations in part quality. Also, the base epoxy resin and

the fabric used in the manufacturing of the DCB samples were different between the two

systems. The results imply better interaction between the fibre and the resin in the

MWNT system. Also, in the SWNT prepreg system, the crack propagation within the

laminates was incremental with no sharp decrease in the load values, whereas, the MWNT

system demonstrated a saw-tooth load-displacement curve.

It should be noted that even though other toughening mechanisms such as crack

deviation exists in the CNT-modified polymer composites, CNT bridging is the main

mechanism that benefits from the mechanical properties of CNTs. Other types of

toughening, e.g. crack deviation, is a function of shape of the nano-particles [10, 84-86],

and does not benefit from the high mechanical properties of CNTs and exist with other

types of nano-reinforcements, such as nanoclays.

In summary, for Mode I delamination results, addition of the CNTs in both SWNT and

MWNT systems improved the initiation and propagation GIc. CNTs were pulled out from

the matrix polymer and contributed to the toughening mechanisms, resulting in the

higher interlaminar propagation toughness values. Other CNT toughening mechanisms

may also contribute to the increase of the fracture toughness, such as, crack deviation and

crack pinning which will be discussed in the Fractography section (Section 5.4.2.3).

114

(a) SWNT prepreg system

(b) MWNT resin film infusion system

Figure 5-19. Load-displacement curves neat and CNT modified DCB samples

0

10

20

30

40

50

60

70

80

90

0 5 10 15 20 25 30 35

Load

(N)


Neat

SWNT

0

10

20

30

40

50

60

70

80

90

0 5 10 15 20 25 30 35

Loa

d (N

)


Neat

2377

2378

115



Figure 5-20. R-curve values comparing neat vs. CNT modified DCB samples

100

150

200

250

300

350

400

450

500

45 55 65 75 85 95 105

GIC

(J/

m2)


0.1% SWNT

Neat

0

200

400

600

800

1000

1200

1400

1600

1800

45 55 65 75 85 95 105

GIc

(J/m

2)


Neat

2377

2378

116



Figure 5-21. Average Mode I initiation and propagation values for neat and CNT modified samples

314.42 322.88342.83387.11

0

50

100

150

200

250

300

350

400

450

Neat 0.1% wt. SWNT

GIc

(J/

m2)

Initiation Propagation

171.26 227.92

417.53424.22

629.72

876.19

0

200

400

600

800

1000

1200

Neat-M1 2377-M1 2378-M1

GIc

(J/

m2 )

Gic Initiation Gic Propagation

117

Table 5-2 summarizes the percentage change in the mode I values after addition of CNTs.

Table 5-2. Percentage change of fracture toughness values in mode I after addition of CNTs

Percentage Increase with respect to neat resin

SWNT Prepreg MWNT Resin Film

2377 2378

KIc - 12% 17 % 5 %

GIc Initiation 3 % 33 % 143 %

GIc Propagation 13 % 48 % 106 %

5. 4. 2. 2. Mode II delamination properties

The objective of the non-precracked (NPC) and the precracked (PC) fracture tests was to

capture delamination initiation, since delamination growth was highly unstable in Mode II

[173]. According to Appendix B data analysis procedure, precracked and non-precracked

Mode II delamination results were calculated for both SWNT and MWNT systems. The

results of 3 specimens for each system are averaged and shown in Figure 5-22.

As seen in Figure 5-22, in both systems addition of CNTs increased the mode II

interlaminar fracture toughness. This increased release rate energy may be attributed to

CNTs bridging at the delamination site. The SWNT system demonstrated higher NPC

values compared to PC values, whereas for the MWNT system, the PC values were 2% –

12% higher than the NPC values. This can be explained by different thickness of the Teflon

inserts as well as different processing parameter that has been used to manufacture

samples. Table 5-3 summarizes the percentage change in mode II interlaminar fracture

toughness for both systems after adding CNTs.

Table 5-3. Percentage change of fracture toughness values in mode II after addition of CNTs

Percentage increase with respect to neat resin

SWNT Prepreg MWNT Resin Film

2377 2378

GIIc NPC 12 % 23 % 127 %

GIIc PC 27 % 13 % 108 %

For the MWNT system, it is inferred that the 2378 sample with the CNT and the nanofiller,

performs better than the 2377 samples. Even though the 2378 samples with CNT and

nanofiller performed better than the 2377 samples, it should also be noted that the

standard deviations from the average results was larger.

118



Figure 5-22. Average mode II interlaminar fracture toughness values

1794.992017.55

1104.131407.05

0

500

1000

1500

2000

2500

Neat 0.1% SWNT

G II

c (J

/m2)

Non Precracked Precracked

486598

1102

542 614

1127

0

200

400

600

800

1000

1200

1400

Neat 2377 2378

G II

c (J

/m2 )

Non Precracked Precracked

119

5. 4. 2. 3. Fractography

The Hitachi SU-8000 Cold Field Emission SEM was used to further study the fracture

surface of the composite laminates after delamination tests.

SWNT Composite.

Figure 5-23 (a) shows an SEM image of a fractured DCB coupon failed under the mode I

delamination test. The bridging effects can be seen for larger bundles of SWNTs.

According to this image, the nanotubes contributed to the increase of fracture toughness

failure by crack bridging through two main mechanisms: 1. SWNT pull-out for CNTs

oriented with an angle with respect to the crack growth plane, (red arrows), 2. SWNT peel-

off for CNTs parallel to the crack plane.

In Figure 5-23 (b, c), SEM images of the Mode II fracture surfaces for neat and SWNT-

modified samples are shown, respectively. A comparison between the two SEM images

demonstrated that there were more hackles (the dominant toughening mechanisms in

Mode II) for the nanocomposites. Figure 5-23 (d, e) shows higher magnification SEM

images of fractured samples.

According to the literature, Mode II delaminations are caused by two dominant

toughening mechanisms; micro-cracks and hackles which are both microscopic matrix

failure modes [23, 39, 45, 174-176]. Also, under mode II loadings, fibre bridging is less

dominant. Unlike DCB specimens that exhibited continuous crack growth along the

fibre/matrix interface, ENF specimens showed discontinuous crack growth by micro-crack

coalescence leading to the creation of hackles. According to the SEM image analysis

(Figure 5-23 (b, c)), SWNT bundles acted as rigid fillers arresting the crack propagation.

The arrested cracks created more hackles in the matrix rich interface areas. So, it is

realistic to anticipate a higher amount of hackles present at the fracture surface of the

SWNT modified composite laminates as compared to that of the base composite

laminates.

120

Figure 5-23. SEM images of fractured DCB coupons; a) CNT pull-outs are highlighted by red arrows and CNT peelings are shown by dotted black arrows; b-e) SEM of fractured mode II ENF

coupons at different magnifications

MWNT Composite.

The results of the Mode I fracture surface of the composite laminates with no MWNT at

different magnifications are shown in Figure 5-24. Different locations on the surface at

a

b c

d e

121

x2k magnification are shown in Figure 5-24 (a, b). The former was taken at a location

where the resin was detached from the fibres, and the later was caused by crack growth in

a resin-rich region. Also, in images (a) and (b), there were signs of brittle fracture (river

lines). Also, the fractured surfaces between two river lines were very smooth with no

roughness. This smoothness of the surface was explored by further zooming into the

regions as shown in images (c) with x10K and (d) with x50k magnifications.

Figure 5-24. SEM analysis of the delaminated surface of neat composite laminates

For the MWNT-modified composites, the fracture surface was considerably rougher

compared with the neat resin laminates. The results for the 2377 and 2378 MWNT

laminates are shown in Figure 5-25 and Figure 5-26, respectively. For both formulations,

an important observation was the interaction of the resin film with the fibre fabric. As can

be seen in Figure 5-25(a) and Figure 5-26(a), the fibres were covered with a layer of

modified resin, with relatively well-dispersed MWNTs. This stronger interaction was the

major difference between the neat composite laminates and the MWNT modified ones,

leading to higher delamination resistance. For both MWNT modified formulations (2377

a b

c d

122

and 2378) MWNTs were agglomerated into CNT-rich islands. However, for the 2377

formulation, between the MWNT agglomerated islands, the fracture surface was smooth,

whereas for the 2378 formulation, the surface was considerably rough. Also, for the 2377

formulation, on each MWNT rich island, several MWNTs were pulled out as shown in

Figure 5-25 (e, f), which clearly contribute to higher energy consumption and

consequently, higher resistivity to crack growth.

Figure 5-25. SEM analysis of the delaminated surface of 2377 MWNT composite laminates

a b

c d

e f

123

Figure 5-26. SEM analysis of the delaminated surface of 2378 MWNT composite laminates at different magnification (magnified areas are highlighted by red squares)

For the 2378 formulation, on the other hand, the areas between the MWNT islands were

rough, due to the plasticizer added to improve the delamination properties, Figure 5-26 (c,

d). Otherwise, the same toughening mechanisms existed as the 2377 formulation.

a b

c d

e f

124

5. 5. Summary and Conclusions

In this chapter, the effect of Carbon Nanotubes (both SWNT and MWNT) on the

delamination properties of composite laminates was investigated. Mode I and Mode II

tests were conducted to verify the potential of adding CNTs to enhance delamination

properties of laminates processed with two methods: 1. CNT-modified prepreg

lamination, and 2. Resin film/fibre mat layup. Table 5-4 summarize the result of the

experiments presented in this chapter.

Table 5-4. Summary of fracture toughness improvement

Percentage Increase with respect to neat resin SWNT

Prepreg MWNT Resin Film

2377 2378

Polymer Stress Intensity KIc - 12% 17 % 5 %

Composite

Mode I GIc Initiation 3 % 33 % 143 %

GIc Propagation 13 % 48 % 106 %

Mode II GIIc NPC 12 % 23 % 127 %

GIIc PC 27 % 13 % 108 %

The CNT-modified polymers showed deteriorated toughness properties (in case of SWNT

system), or a relatively minor improvement (MWNT system). In contrary, the laminated

composite demonstrated a major improvement of delamination properties as a result of

adding CNTs. The following reasons might justify this behaviour:

1. The fibre mat in the composite samples acted as a network preventing the

movement of CNT bundles and consequently improved dispersion quality

compared to the polymer samples.

2. During the end-notch fracture toughness test of resins, the crack front has the

freedom to propagate in any direction along the specimen width. As a result, the

crack could progress along the path with lowest resistance against the crack

growth. This path might include impurities of CNTs with weak interfacial

interaction with the surrounding polymer chains. However, for the case of mode I

and mode II interlaminar fracture toughness tests, the crack growth was only

limited to the 7 μm-thick interlaminar layer. As a result, CNTs in the interlaminar

region could more effectively bridge the crack front.

The scanning electron microscopy gave additional information about the morphology of

the fracture surface and toughening mechanisms. The toughening mechanisms included

CNT pull-out, crack pinning, crack deviation, and CNT peel-off. By comparing the

morphology of the polymer based samples with those of composite laminates, it can be

125

concluded that there were more toughening features on the surface of delaminated

composites.

Finally, as stated in Chapter 4, manufacturing processes plays a key role in the

effectiveness of CNT modification of polymers. The results of this chapter also confirm the

same dependence for composites laminate processing.

126

Chapter 6. Conclusions and Contribution of the Thesis

In this thesis, the effect of CNTs as a toughening agent on polymer based composite was

modelled and experimentally investigated. The contributions of this work are summarized

as follows:

1) An analytical CNT-bridging model identified the key CNT physical and mechanical

properties that affect the toughness of a brittle matrix. Adding CNTs to polymers results

in several toughening mechanisms that enhances fracture toughness, i.e. CNT bridging,

crack pinning, and crack deviation. However, only the Carbon Nanotube bridging

mechanism benefits from the extraordinary strength of CNTs, whereas other toughening

mechanisms are only a function of the geometry of the toughening agent.

By analogy with long fibre reinforced composites, the nanotubes will pull-out if their

length is below a critical length. For CNT having a length higher than a critical value, there

will be a combination of CNT pull-out and rupture, [1]. Modelling of both the CNT pull-out

and the CNT rupture was presented. The proposed model also addressed the effect of

random CNT orientation in the polymer matrix for the first time, whereas previous

modelling studies focused on perfectly aligned CNTs, [118, 119].

The model is useful in identifying the effect of different geometrical and mechanical

properties of CNTs on the final fracture toughness of CNT-modified formulations. Based

on the CNT bridging model, aligning relatively long (>10μm) carbon nanotubes

perpendicular to the crack growth plane has great potential to enhance the toughness of

brittle polymers. Also, toughness enhancement with CNTs requires increasing the volume

fraction and length of CNTs, and aligning them normal to the crack growth plane. The

model also shows that in most cases, SWNTs were the better choice for polymer

toughening compared to MWNTs. Figure 6-1 summarizes the conclusions of the modelling

work. The proposed steps to improve the toughness are bounded by the limitations and

challenges in the processing of the CNT polymer mixtures. A very important assumption of

this modelling work is the perfect dispersion of CNTs inside the polymer solution.

However, in reality CNTs create bundles and agglomerate inside the CNT modified

solutions. CNT bundles and agglomerates have lower mechanical properties, when

127

compared with the individual CNTs. Therefore, it is critical to understand the effect of

processing on the fracture toughness of CNT-modified polymers.

Randomly oriented Alignment of CNTs Higher Vf Incorporation of longer

CNTs

Figure 6-1. Steps to improve the toughness of brittle polymers by incorporating CNTs

2) The evolution of CNT dispersion during the curing process was observed and

investigated. Even though the CNTs were well dispersed at room temperature, when the

curing process occurred at high temperature (>100 °C), the dispersion quality

deteriorated. An image analysis tool was used to quantify the dispersion deterioration to

identify the root cause of dispersion degradation. The results of dispersion analyses were

correlated to the viscosity curves obtained from rheological analyses. It was shown that

the chemical process of polymerization (curing process) was not the main driver of

dispersion degradation, since the dispersion quality remained constant when polymer

systems were cured at room temperature. However, a drop in viscosity and the thermal

expansion of the resin at high temperature (>100 °C) proved to be the main drivers of

dispersion degradation. Also, the dispersion analysis showed a direct correlation between

dispersion quality and the fracture toughness enhancement of CNTs in modified polymers.

This was a major contribution, as this approach resulted in identifying the optimized

SWNT formulation. The research started with a 23% reduction in the fracture toughness

when a low CNT content was added to the polymer matrix, and through a systematic

series of dispersion and fracture tests, we achieved 38% improvement in stress intensity

fracture toughness values.

3) The addition of very low CNT content (<1 wt.%) to epoxy was systematically studied

and resulted in the improvement of fracture toughness (up to 38%). This was done by

fine- tuning the formulations to achieve uniform dispersion of CNTs in the polymer. The

fine tuning process included changing of the type of curing agent or the ratio of polymer

to curing agent. Two types of CNTs were considered: SWNT (in MY0510 Epoxy) and MWNT

(in bisphenol-A epoxy).

128

For the SWNT system, the effect of the type of curing agent and the hardener – to – resin

ratio were studied. A solid powder hardener, i.e. DDS, was compared to a liquid curing

agent, i.e. Aradur. DDS resulted in higher fracture toughness values compared to Aradur.

DDS dissolved in the resin system at 100 °C and consequently resulted in resin-rich

locations (locally no CNTs), which affected the dispersion quality. As the DDS hardener to

resin ratio increased, the effect of SWNTs on fracture toughness improved. However,

addition of more hardener above the 100:60 ratio had a negative effect. For this system, a

maximum of 38% improvement of fracture toughness was achieved. For the MWNT

system, different types of hardener were studied. The most effective solution to solve the

problem of dispersion degradation during the cure was to gel the resins at low

temperatures. This process improved the CNT dispersion stability.

The SEM analysis of the fracture surfaces showed several toughening mechanisms that

contributed to the increased fracture toughness. These mechanisms include CNT pull-out,

CNT peel-off, crack pinning, and crack deviation from the agglomerated CNTs (CNT

islands).

4) The CNTs were incorporated into the polymer matrix of two types of Carbon Fibre

Composite Laminates and up to 140% improvement in delamination properties was

achieved. Both Mode I and Mode II tests were conducted to verify the potential of adding

CNTs to enhance delamination properties of laminated composites. Two manufacturing

techniques were considered: 1. CNT-modified prepreg lamination, and 2. Resin film/fibre

fabric layup. The CNT-modified polymers showed deteriorated toughness properties in

case of SWNT system (-12%), and a relatively minor improvement for the MWNT system

(17%). On the other hand, the laminated composites demonstrated a major improvement

of delamination properties as a result of adding CNTs (up to 140%).

This different behaviour in the fracture toughness properties of polymers vs. composite

laminates was explained by studying the fracture surface through Scanning Electron

Microscopy. The morphology of the fracture surface demonstrated a better dispersion

quality and more toughening features on the surface of the delaminated composites

compared with that of the polymer. Carbon fibres seem to act as a network, preventing

CNTs to move freely and re-agglomerate, leading to uniform dispersion of CNTs in

laminated composites.

129

Future works

The following subjects could be potential topics for further research in the future:

1. Effect of CNT bundles and aggregates should be modelled to understand the actual

effect of the CNTs, since in reality, CNTs agglomerated due to their high surface

tension. Most of the current modelling work on CNT-modified polymers in terms of

structural/property relation considered CNTs as a single high performance

material, however, CNTs agglomerated and created bundles, as confirmed by the

SEM images. These bundles or agglomerated CNTs have different physical and

mechanical properties than individual CNTs.

2. The dispersion degradation analysis presented in this work was performed under

optical microscope. This level of magnification is essential for understanding of

CNT agglomeration at micro-level. CNT dispersion analysis at nano-scale (using

SEM) would help clarify nano-scale behaviour of these formulations during the

curing process. A major challenge however is the design of a robust test setup that

can perform the test under SEM.

3. An important downside of CNT-modified polymers is the large variation in the

mechanical characterization results of different batches with the same

formulation. The results of the fracture toughness characterization were very

sensitive to the sample preparation procedure. More detailed research on the root

causes of such variations can be used to develop standards for material processing

and to pave the way for the industrial application of these material systems.

4. Investigating other base polymer materials as well as other types of nano-fillers

can help better understand the mechanisms leading to improved properties. As an

example, the choice of a very ductile polymer (such as a thermoplastic polymer), in

which cracks propagate at a very stable manner, will allow observation of the

toughening mechanism during the crack growth. This crack growth monitoring will

be important to identify novel toughening mechanisms.

According to the results of this research for the CNT-modified composites, further

research is required to effectively manufacture composite laminates with mechanical and

multifunctional properties that harness the full potential of Carbon Nanotubes as polymer

reinforcement.

130

Chapter 7. References

[1] Hull, D., 1981, An Introduction to Composite Materials, Cambridge University Press.

[2] Bauhofer, W., and Kovacs, J. Z., 2009, "A Review and Analysis of Electrical Percolation in Carbon Nanotube Polymer Composites," Composites Science and Technology, 69(10), pp. 1486-1498.

[3] Potschke, P., Fornes, T. D., and Paul, D. R., 2002, "Rheological Behavior of Multiwalled Carbon Nanotube/Polycarbonate Composites," Polymer, 43(11), pp. 3247-3255.

[4] Gojny, F. H., Wichmann, M. H. G., Fiedler, B., and Schulte, K., 2005, "Influence of Different Carbon Nanotubes on the Mechanical Properties of Epoxy Matrix Composites - a Comparative Study," Composites Science and Technology, 65(15-16), pp. 2300-2313.

[5] Thostenson, E. T., and Chou, T.-W., 2006, "Processing-Structure-Multi-Functional Property Relationship in Carbon Nanotube/Epoxy Composites," Carbon, 44(14), pp. 3022-3029.

[6] Seyhan, A. T., Tanoglu, M., and Schulte, K., 2009, "Tensile Mechanical Behavior and Fracture Toughness of Mwcnt and Dwcnt Modified Vinyl-Ester/Polyester Hybrid Nanocomposites Produced by 3-Roll Milling," Materials Science and Engineering: A, 523(1-2), pp. 85-92.

[7] Ashby, M. F., Shercliff, H., Cebon, D., and Books24x, I., 2007, Materials Engineering, Science, Processing and Design,

[8] Ajayan, P. M., Schadler, L. S., Giannaris, C., and Rubio, A., 2000, "Single-Walled Carbon Nanotube–Polymer Composites: Strength and Weakness," Advanced Materials, 12(10), pp. 750-753.

[9] Lange, F. F., 1970 "Interaction of a Crack Front with a Second-Phase Dispersion " PHIL MAG., 22(179), pp. 983-992.

[10] Evans, A. G., 1972, "Strength of Brittle Materials Containing Second Phase Dispersions," Philosophical Magazine, 26(6), pp. 1327-1344.

[11] Kunz-Douglass, S., Beaumont, P. W. R., and Ashby, M. F., 1980, "A Model for the Toughness of Epoxy-Rubber Particulate Composites," Journal of Materials Science, 15(5), pp. 1109-1123.

[12] Garg, A. C., and Mai, Y.-W., 1988, "Failure Mechanisms in Toughened Epoxy Resins--a Review," Composites Science and Technology, 31(3), pp. 179-223.

[13] Garg, A., 1985, "Fracture Behavior of Cross-Ply Graphite/ Epoxy Laminates," Engineering Fracture Mechanics, 22(6), pp. 1035-1048.

[14] Lammerant, L., and Verpoest, I., 1996, "Modelling of the Interaction between Matrix Cracks and Delaminations During Impact of Composite Plates," Composites Science and Technology, 56(10), pp. 1171-1178.

131

[15] Khashaba, U. A., "Delamination in Drilling Gfr-Thermoset Composites," Composite Structures, 63(3-4), pp. 313-327.

[16] Green, D. J., Nicholson, P. S., and Embury, J. D., 1979, "Fracture of a Brittle Particulate Composite," Journal of Materials Science, 14(7), pp. 1657-1661.

[17] Kinloch, A. J., Shaw, S. J., and Hunston, D. L., 1982, "Crack Propagation in a Rubber-Toughened Epoxy," Cambridge, MA, Engl, pp. 29-1.

[18] Kinloch, A. J., Young, R. J., Spanoudakis, J., and Maxwell, D., 1982, "Failure Mechanisms in Brittle Particle-Filled Epoxy Resins," Cambridge, England.

[19] Mcmeeking, R. M., and Evans, A. G., 1982, "Mechanics of Transformation-Toughening in Brittle Materials," Proc. Journal of the American Ceramic Society, 65, pp. 242-246.

[20] Moloney, A. C., Kausch, H. H., Kaiser, T., and Beer, H. R., 1987, "Parameters Determining the Strength and Toughness of Particulate Filled Epoxide Resins," Journal of Materials Science, 22(2), pp. 381-393.

[21] Bandyopadhyay, S., 1990, "Review of the Microscopic and Macroscopic Aspects of Fracture of Unmodified and Modified Epoxy Resins," Materials Science and Engineering A, 125(2), pp. 157-184.

[22] Kim, J.-K., and Mai, Y.-W., 1991, "High Strength, High Fracture Toughness Fibre Composites with Interface Control--a Review," Composites Science and Technology, 41(4), pp. 333-378.

[23] Sue, H. J., Jones, R. E., and Garcia-Meitin, E. I., 1993, "Fracture Behaviour of Model Toughened Composites under Mode I and Mode Ii Delaminations," Journal of Materials Science, 28(23), pp. 6381-6391.

[24] Johnston, N. J., 1984, "Synthesis and Toughness Properties of Resins and Composites," Proc. NASA Conference Publication, pp. 75-95.

[25] Bascom, W. D., Cottington, R. L., Jones, R. L., and Peyser, P., 1975, "The Fracture of Epoxy- and Elastomer-Modified Epoxy Polymers in Bulk and as Adhesives," Journal of Applied Polymer Science, 19(9), pp. 2545-2562.

[26] Bascom, W., Ting, R., Moulton, R., Riew, C., and Siebert, A., 1981, "The Fracture of an Epoxy Polymer Containing Elastomeric Modifiers," Journal of Materials Science, 16(10), pp. 2657-2664.

[27] Mai, Y. W., Atkins, A. G., Selby, K., and Miller, L. E., 1975, "On the Velocity-Dependent Fracture Toughness of Epoxy Resins," Journal of Materials Science, 10(11), pp. 2000-2003.

[28] Koutsky, J., River, B., Ebewele, R., Luce, M., and Han, K. S., 1983, "Evaluation of Fracture Energies of Composites Using Cantilever Beam Techniques," Proc. Organic Coatings and Applied Polymer Science Proceedings, 48, pp. 822-825.

[29] Garg, A., and Ishai, O., 1985, "Hygrothermal Influence on Delamination Behavior of Graphite/Epoxy Laminates," Engineering Fracture Mechanics, 22(3), pp. 413-427.

132

[30] Keary, P. E., Ilcewicz, L. B., Shaar, C., and Trostle, J., 1985, "Mode I Interlaminar Fracture Toughness of Composites Using Slender Double Cantilevered Beam Specimens," Journal of Composite Materials, pp. 154-177.

[31] Whitney, J. M., Browning, C. E., and Hoogsteden, W., 1982, "A Double Cantilever Beam Test for Characterizing Mode I Delamination of Composite Materials," Journal of Reinforced Plastics and Composites, pp. 297-313.

[32] Gledhill, R. A., Kinloch, A. J., Yamini, S., and Young, R. J., 1978, "Relationship between Mechanical Properties of and Crack Progogation in Epoxy Resin Adhesives," Polymer, 19(5), pp. 574-582.

[33] Gledhill, R. A., and Kinloch, A. J., 1976, "Failure Criterion for the Fracture of Structural Adhesive Joints," Polymer, 17(8), pp. 727-731.

[34] Mai, Y. W., 1975, "Quasi-Static Adhesive Fracture," The Journal of Adhesion, 7(2), pp. 141 - 153.

[35] Rhodes, M. D., 1980, "Damage Tolerance Research on Composite Compression Panel," NASA-CP 2142 ), pp. 107–143.

[36] Lhymn, C., and Schultz, J. M., 1987, "Strength and Toughness of Fibre-Reinforced Thermoplastics: Effect of Temperature and Loading Rate," Composites, 18(4), pp. 287-292.

[37] Green, D. J., Nicholson, P. S., and Embury, J. D., 1978, "Crack Shape Study in a Brittle, Non-Bonded, Particulate Composite," Solid Wastes Management Refuse Removal Journal, 3B(pp. 941-948.

[38] Pearson, R. A., and Yee, A. F., 1983, "Effect on Crosslink Density on the Toughening Mechanism of Elastomer Modified Epoxies," Washington, DC, USA, 49, pp. 316-320.

[39] Lee, J. J., Lim, J. O., and Huh, J. S., 2000, "Mode Ii Interlaminar Fracture Behavior of Carbon Bead-Filled Epoxy/Glass Fiber Hybrid Composite," Polymer Composites, 21(2), pp. 343-352.

[40] Mimura, K., Ito, H., and Fujioka, H., 2001, "Toughening of Epoxy Resin Modified with in Situ Polymerized Thermoplastic Polymers," Polymer, 42(22), pp. 9223-9233.

[41] Baughman, R. H., Zakhidov, A. A., and De Heer, W. A., 2002, "Carbon Nanotubes - the Route toward Applications," Science, 297(5582), pp. 787-792.

[42] Qiao, Y., 2003, "Fracture Toughness of Composite Materials Reinforced by Debondable Particulates," Scripta Materialia, 49(6), pp. 491-496.

[43] Gorga, R. E., and Cohen, R. E., 2004, "Toughness Enhancements in Poly(Methyl Methacrylate) by Addition of Oriented Multiwall Carbon Nanotubes," Journal of Polymer Science, Part B: Polymer Physics, 42(14), pp. 2690-2702.

[44] Morgan, R., 1985, Epoxy Resins and Composites I.

[45] Wetzel, B., Rosso, P., Haupert, F., and Friedrich, K., 2006, "Epoxy Nanocomposites - Fracture and Toughening Mechanisms," Engineering Fracture Mechanics, 73(16), pp. 2375-2398.

133

[46] Zhao, Q., and Hoa, S. V., 2007, "Toughening Mechanism of Epoxy Resins with Micro/Nano Particles," Journal of Composite Materials, 41(2), pp. 201-219.

[47] Zhao, S., Schadler, L. S., Duncan, R., Hillborg, H., and Auletta, T., 2008, "Mechanisms Leading to Improved Mechanical Performance in Nanoscale Alumina Filled Epoxy," Composites Science and Technology, 68(14), pp. 2965-2975.

[48] Zhao, S., Schadler, L. S., Hillborg, H., and Auletta, T., 2008, "Improvements and Mechanisms of Fracture and Fatigue Properties of Well-Dispersed Alumina/Epoxy Nanocomposites," Composites Science and Technology, 68(14), pp. 2976-2982.

[49] Mcgarry, F. J., 1970, "Building Design with Fibre Reinforced Materials," Proceedings of the Royal Society of London. A. Mathematical and Physical Sciences, 319(1536), pp. 59-68.

[50] Kinloch, A. J., Shaw, S. J., Tod, D. A., and Hunston, D. L., 1983, "Deformation and Fracture Behaviour of a Rubber-Toughened Epoxy: 1. Microstructure and Fracture Studies," Polymer, 24(10), pp. 1341-1354.

[51] Bandyopadhyay, S., 1984, "Crack Propagation Studies of Bulk Polymeric Materials in the Scanning Electron Microscope," Journal of Materials Science Letters, 3(1), pp. 39-43.

[52] Drake, R., and Siebert, A. R., 1975, "Elastomer-Modified Epoxy Resins for Structural Applications," SAMPE Q, 6(4), pp. 11-21.

[53] Kinloch, A. J., Shaw, S. J., and Hunston, D. L., 1983, "Deformation and Fracture Behaviour of a Rubber-Toughened Epoxy: 2. Failure Criteria," Polymer, 24(10), pp. 1355-1363.

[54] Scott, J. M., and Phillips, D. C., 1975, "Carbon Fibre Composites with Rubber Toughened Matrices," Journal of Materials Science, 10(4), pp. 551-562.

[55] Liu, L., and Wagner, H. D., 2005, "Rubbery and Glassy Epoxy Resins Reinforced with Carbon Nanotubes," Composites Science and Technology, 65(11-12), pp. 1861-1868.

[56] González, I., Eguiazábal, J. I., and Nazábal, J., 2006, "Rubber-Toughened Polyamide 6/Clay Nanocomposites," Composites Science and Technology, 66(11-12), pp. 1833-1843.

[57] Kunz, S. C., Sayre, J. A., and Assink, R. A., 1982, "Morphology and Toughness Characterization of Epoxy Resins Modified with Amine and Carboxyl Terminated Rubbers," Polymer, 23(13), pp. 1897-1906.

[58] Maxwell, D., Young, R. J., and Kinloch, A. J., 1984, "Hybrid Particulate-Filled Epoxy-Polymers," Journal of Materials Science Letters, 3(1), pp. 9-12.

[59] Kinloch, A. J., Maxwell, D., and Young, R. J., 1985, "Micromechanisms of Crack Propagation in Hybrid-Particulate Composites," Journal of Materials Science Letters, 4(10), pp. 1276-1279.

[60] Kinloch, A. J., Maxwell, D. L., and Young, R. J., 1985, "The Fracture of Hybrid-Particulate Composites," Journal of Materials Science, 20(11), pp. 4169-4184.

[61] Young, R. J., Maxwell, D. L., and Kinloch, A. J., 1986, "The Deformation of Hybrid-Particulate Composites," Journal of Materials Science, 21(2), pp. 380-388.

134

[62] Tse, M.-F., 1985, "Physical Properties of Barium Titanate-Filled Rubber-Modified Epoxies," Journal of Applied Polymer Science, 30(9), pp. 3625-3647.

[63] Moloney, A. C., Kausch, H. H., and Stieger, H. R., 1983, "The Fracture of Particulate-Filled Epoxide Resins," Journal of Materials Science, 18(1), pp. 208-216.

[64] Young, R. J., and Beaumont, P. W. R., 1975, "Failure of Brittle Polymers by Slow Crack Growth," Journal of Materials Science, 10(8), pp. 1343-1350.

[65] Norman, D. A., and Robertson, R. E., 2003, "Rigid-Particle Toughening of Glassy Polymers," Polymer, 44(8), pp. 2351-2362.

[66] Gadkaree, K. P., and Salee, G., 1983, "Fatigue Crack Propagation in Composites with Spherical Fillers - 1," Polymer Composites, 4(1), pp. 19-25.

[67] Griffiths, R., and Holloway, D., 1970, "The Fracture Energy of Some Epoxy Resin Materials," Journal of Materials Science, 5(4), pp. 302-307.

[68] Spanoudakis, J., and Young, R., 1984, "Crack Propagation in a Glass Particle-Filled Epoxy Resin," Journal of Materials Science, 19(2), pp. 473-486.

[69] Spanoudakis, J., and Young, R., 1984, "Crack Propagation in a Glass Particle-Filled Epoxy Resin," Journal of Materials Science, 19(2), pp. 487-496.

[70] Moloney, A. C., and Kausch, H. H., 1985, "Direct Observations of Fracture Mechanisms in Epoxide Resins," Journal of Materials Science Letters, 4(3), pp. 289-292.

[71] Evans, A. G., 1972, "The Strength of Brittle Materials Containing Second Phase Dispersions," Philosophical Magazine, 26(6), pp. 1327 - 1344.

[72] Moloney, A. C., Kausch, H. H., and Stieger, H. R., 1984, "The Fracture of Particulate-Filled Epoxide Resins," Journal of Materials Science, 19(4), pp. 1125-1130.

[73] Broutman, L. J., and Sahu, S., 1971, "The Effect of Interfacial Bonding on the Toughness of Glass Filled Polymers," Materials Science and Engineering, 8(2), pp. 98-107.

[74] Spanoudakis, J., and Young, R. J., 1984, "Crack Propagation in a Glass Particle-Filled Epoxy Resin," Journal of Materials Science, 19(2), pp. 487-496.

[75] Fu, S.-Y., Feng, X.-Q., Lauke, B., and Mai, Y.-W., 2008, "Effects of Particle Size, Particle/Matrix Interface Adhesion and Particle Loading on Mechanical Properties of Particulate-Polymer Composites," Composites Part B: Engineering, 39(6), pp. 933-961.

[76] Kelly, A., 1970, "Interface Effects and the Work of Fracture of a Vibrous Composite," Proceedings of the Royal Society of London. Series A, Mathematical and Physical Sciences, 319(1536), pp. 95-116.

[77] Spanoudakis, J., and Young, R. J., 1984, "Crack Propagation in a Glass Particle-Filled Epoxy Resin," Journal of Materials Science, 19(2), pp. 473-486.

[78] Yamini, S., and Young, R. J., 1980, "Mechanical Properties of Epoxy Resins Em Dash 1. Mechanisms of Plastic Deformation," Journal of Materials Science, 15(7), pp. 1814-1822.

135

[79] Pearson, R. A., and Yee, A. F., 1993, "Toughening Mechanisms in Thermoplastic-Modified Epoxies: 1. Modification Using Poly(Phenylene Oxide)," Polymer, 34(17), pp. 3658-3670.

[80] Evans, A. G., 1972, "The Strength of Brittle Materials Containing Second Phase Dispersions," Philosophical Magazine, 26(6), pp. 1327-1344.

[81] Lange, F. F., 1970, "The Interaction of a Crack Front with a Second-Phase Dispersion," Philosophical Magazine, 22(179), pp. 983-992.

[82] Rose, L. R. F., 1987, "Toughening Due to Crack-Front Interaction with a Second-Phase Dispersion," Mechanics of Materials, 6(1), pp. 11-15.

[83] Ahlquist, C. N., 1975, "On the Interaction of Cleavage Cracks with Second Phase Particles," Acta Metallurgica, 23(2), pp. 239-243.

[84] Faber, K. T., and Evans, A. G., 1983, "Crack Deflection Processes--I. Theory," Acta Metallurgica, 31(4), pp. 565-576.

[85] Faber, K. T., and Evans, A. G., 1983, "Crack Deflection Processes--Ii. Experiment," Acta Metallurgica, 31(4), pp. 577-584.

[86] Evans, A. G., Williams, S., and Beaumont, P. W. R., 1985, "On the Toughness of Particulate Filled Polymers," Journal of Materials Science, 20(10), pp. 3668-3674.

[87] Hutchinson, J. W., and Jensen, H. M., 1990, "Models of Fiber Debonding and Pullout in Brittle Composites with Friction," Mechanics of Materials, 9(2), pp. 139-163.

[88] Huang, Y., and Kinloch, A. J., 1992, "Modelling of the Toughening Mechanisms in Rubber-Modified Epoxy Polymers," Journal of Materials Science, 27(10), pp. 2763-2769.

[89] Johnsen, B. B., Kinloch, A. J., Mohammed, R. D., Taylor, A. C., and Sprenger, S., 2007, "Toughening Mechanisms of Nanoparticle-Modified Epoxy Polymers," Polymer, 48(2), pp. 530-541.

[90] Iijima, S., 1991, "Helical Microtubules of Graphitic Carbon," Nature, 354(6348), pp. 56.

[91] Hussain, F., Hojjati, M., Okamoto, M., and Gorga, R. E., 2006, "Review Article: Polymer-Matrix Nanocomposites, Processing, Manufacturing, and Application: An Overview," Journal of Composite Materials, 40(17), pp. 1511-1575.

[92] Yokozeki, T., Iwahori, Y., and Ishiwata, S., 2007, "Matrix Cracking Behaviors in Carbon Fiber/Epoxy Laminates Filled with Cup-Stacked Carbon Nanotubes (Cscnts)," Composites Part A: Applied Science and Manufacturing, 38(3), pp. 917-924.

[93] Yu, M.-F., Lourie, O., Dyer, M. J., Moloni, K., Kelly, T. F., and Ruoff, R. S., 2000, "Strength and Breaking Mechanism of Multiwalled Carbon Nanotubes under Tensile Load," Science, 287(5453), pp. 637-640.

[94] Coleman, J. N., Khan, U., Blau, W. J., and Gun'ko, Y. K., 2006, "Small but Strong: A Review of the Mechanical Properties of Carbon Nanotube-Polymer Composites," Carbon, 44(9), pp. 1624-1652.

136

[95] Zhao, Q., Wood, J. R., and Wagner, H. D., 2001, "Stress Fields around Defects and Fibers in a Polymer Using Carbon Nanotubes as Sensors," Applied Physics Letters, 78(12), pp. 1748-1750.

[96] Rana, S., Alagirusamy, R., and Joshi, M., 2009, "A Review on Carbon Epoxy Nanocomposites," Journal of Reinforced Plastics and Composites, 28(4), pp. 461-487.

[97] Naous, W., Yu, X.-Y., Zhang, Q.-X., Naito, K., and Kagawa, Y., 2006, "Morphology, Tensile Properties, and Fracture Toughness of Epoxy/Al2o3 Nanocomposites," Journal of Polymer Science Part B: Polymer Physics, 44(10), pp. 1466-1473.

[98] Kinloch, A., Mohammed, R., Taylor, A., Eger, C., Sprenger, S., and Egan, D., 2005, "The Effect of Silica Nano Particles and Rubber Particles on the Toughness of Multiphase Thermosetting Epoxy Polymers," Journal of Materials Science, 40(18), pp. 5083-5086.

[99] Fiedler, B., Gojny, F. H., Wichmann, M. H. G., Nolte, M. C. M., and Schulte, K., 2006, "Fundamental Aspects of Nano-Reinforced Composites," Composites Science and Technology, 66(16), pp. 3115-3125.

[100] Frankland, S. J. V., Caglar, A., Brenner, D. W., and Griebel, M., 2002, "Molecular Simulation of the Influence of Chemical Cross-Links on the Shear Strength of Carbon Nanotube-Polymer Interfaces," Journal of Physical Chemistry B, 106(12), pp. 3046-3048.

[101] Suhr, J., and Koratkar, N., 2008, "Energy Dissipation in Carbon Nanotube Composites: A Review," Journal of Materials Science, 43(13), pp. 4370-4382.

[102] Ma, P.-C., Siddiqui, N. A., Marom, G., and Kim, J.-K., "Dispersion and Functionalization of Carbon Nanotubes for Polymer-Based Nanocomposites: A Review," Composites Part A: Applied Science and Manufacturing, In Press, Accepted Manuscript(

[103] Hamming, L. M., Qiao, R., Messersmith, P. B., and Catherine Brinson, L., 2009, "Effects of Dispersion and Interfacial Modification on the Macroscale Properties of Tio2 Polymer-Matrix Nanocomposites," Composites Science and Technology, 69(11-12), pp. 1880-1886.

[104] Andrews, R., Jacques, D., Minot, M., and Rantell, T., 2002, "Fabrication of Carbon Multiwall Nanotube/Polymer Composites by Shear Mixing," Macromolecular Materials and Engineering, 287(6), pp. 395-403.

[105] Prashantha, K., Soulestin, J., Lacrampe, M. F., and Krawczak, P., 2009, "Present Status and Key Challenges of Carbon Nanotubes Reinforced Polyolefins: A Review on Nanocomposites Manufacturing and Performance Issues," Polymers and Polymer Composites, 17(4), pp. 205-245.

[106] Ausman, K. D., Piner, R., Lourie, O., Ruoff, R. S., and Korobov, M., 2000, "Organic Solvent Dispersions of Single-Walled Carbon Nanotubes: Toward Solutions of Pristine Nanotubes," Journal of Physical Chemistry B, 104(38), pp. 8911-8915.

[107] Gong, X., Liu, J., Baskaran, S., Voise, R. D., and Young, J. S., 2000, "Surfactant-Assisted Processing of Carbon Nanotube/Polymer Composites," Chemistry of Materials, 12(4), pp. 1049-1052.

[108] Paredes, J. I., and Burghard, M., 2004, "Dispersions of Individual Single-Walled Carbon Nanotubes of High Length," Langmuir, 20(12), pp. 5149-5152.

137

[109] Felten, A., Bittencourt, C., Pireaux, J. J., Van Lier, G., and Charlier, J. C., 2005, "Radio-Frequency Plasma Functionalization of Carbon Nanotubes Surface O2, Nh3, and Cf4 Treatments," Journal of Applied Physics, 98(7), pp. 1-9.

[110] Utegulov, Z. N., Mast, D. B., He, P., Shi, D., and Gilland, R. F., 2005, "Functionalization of Single-Walled Carbon Nanotubes Using Isotropic Plasma Treatment: Resonant Raman Spectroscopy Study," Journal of Applied Physics, 97(10), pp. 1-4.

[111] Riggs, J. E., Guo, Z., Carroll, D. L., and Sun, Y. P., 2000, "Strong Luminescence of Solubilized Carbon Nanotubes [2]," Journal of the American Chemical Society, 122(24), pp. 5879-5880.

[112] Star, A., Stoddart, J. F., Steuerman, D., Diehl, M., Boukai, A., Wong, E. W., Yang, X., Chung, S. W., Choi, H., and Heath, J. R., 2001, "Preparation and Properties of Polymer- Wrapped Single-Walled Carbon Nanotubes," Angewandte Chemie - International Edition, 40(9), pp. 1721-1725.

[113] Chen, R. J., Zhang, Y., Wang, D., and Dai, H., 2001, "Noncovalent Sidewall Functionalization of Single-Walled Carbon Nanotubes for Protein Immobilization [11]," Journal of the American Chemical Society, 123(16), pp. 3838-3839.

[114] Gojny, F. H., Wichmann, M. H. G., Kopke, U., Fiedler, B., and Schulte, K., 2004, "Carbon Nanotube-Reinforced Epoxy-Composites: Enhanced Stiffness and Fracture Toughness at Low Nanotube Content," Composites Science and Technology, 64(15), pp. 2363-2371.

[115] Fan, Z., Hsiao, K.-T., and Advani, S. G., 2004, "Experimental Investigation of Dispersion During Flow of Multi-Walled Carbon Nanotube/Polymer Suspension in Fibrous Porous Media," Carbon, 42(4), pp. 871-876.

[116] Mirjalili, V., and Hubert, P., 2009, "Effect of Carbon Nanotube Dispersion on the Fracture Toughness of Polymers," Proc. ICCM 17, Edinburgh.

[117] Ma, P.-C., Mo, S.-Y., Tang, B.-Z., and Kim, J.-K., 2010, "Dispersion, Interfacial Interaction and Re-Agglomeration of Functionalized Carbon Nanotubes in Epoxy Composites," Carbon, 48(6), pp. 1824-1834.

[118] Blanco, J., Garcia, E. J., Guzman De Villoria, R., and Wardle, B. L., 2009, "Limiting Mechanisms of Mode I Interlaminar Toughening of Composites Reinforced with Aligned Carbon Nanotubes," Journal of Composite Materials, 43(8), pp. 825-841.

[119] Liyong Tong, Xiannian Sun, and Ping Tan, 2008, "Effect of Long Multi-Walled Carbon Nanotubes on Delamination Toughness of Laminated Composites," Journal of Composite Materials, 42(1), pp. 5-23.

[120] Satapathy, B. K., Weidisch, R., Po?Tschke, P., and Janke, A., 2007, "Tough-to-Brittle Transition in Multiwalled Carbon Nanotube (Mwnt)/Polycarbonate Nanocomposites," Composites Science and Technology, 67(5), pp. 867-879.

[121] Satyanarayana, N., Rajan, K., Sinha, S., and Shen, L., 2007, "Carbon Nanotube Reinforced Polyimide Thin-Film for High Wear Durability," Tribology Letters, 27(2), pp. 181-188.

138

[122] Li, C., Thostenson, E. T., and Chou, T.-W., 2008, "Sensors and Actuators Based on Carbon Nanotubes and Their Composites: A Review," Composites Science and Technology, 68(6), pp. 1227-1249.

[123] Godara, A., Mezzo, L., Luizi, F., Warrier, A., Lomov, S. V., Van Vuure, A. W., Gorbatikh, L., Moldenaers, P., and Verpoest, I., 2009, "Influence of Carbon Nanotube Reinforcement on the Processing and the Mechanical Behaviour of Carbon Fiber/Epoxy Composites," Carbon, 47(12), pp. 2914-2923.

[124] Hubert, P., Ashrafi, B., Adhikari, K., Meredith, J., Vengallatore, S., Guan, J., and Simard, B., 2009, "Synthesis and Characterization of Carbon Nanotube-Reinforced Epoxy: Correlation between Viscosity and Elastic Modulus," Composites Science and Technology, 69(14), pp. 2274-2280.

[125] Haggenmueller, R., Gommans, H. H., Rinzler, A. G., Fischer, J. E., and Winey, K. I., 2000, "Aligned Single-Wall Carbon Nanotubes in Composites by Melt Processing Methods," Chemical Physics Letters, 330(3-4), pp. 219-225.

[126] Ramanathan, T., Liu, H., and Brinson, L. C., 2005, "Functionalized Swnt/Polymer Nanocomposites for Dramatic Property Improvement," Journal of Polymer Science, Part B: Polymer Physics, 43(17), pp. 2269-2279.

[127] Tjong, S. C., 2006, "Structural and Mechanical Properties of Polymer Nanocomposites," Materials Science and Engineering: R: Reports, 53(3-4), pp. 73-197.

[128] Krishnamoorti, R., 2007, "Strategies for Dispersing Nanoparticles in Polymers," MRS Bulletin, 32(4), pp. 341-347.

[129] Balazs, A. C., Emrick, T., and Russell, T. P., 2006, "Nanoparticle Polymer Composites: Where Two Small Worlds Meet," Science, 314(5802), pp. 1107-1110.

[130] Schadler, L. S., Kumar, S. K., Benicewicz, B. C., Lewis, S. L., and Harton, S. E., 2007, "Designed Interfaces in Polymer Nanocomposites: A Fundamental Viewpoint," MRS Bulletin, 32(4), pp. 335-340.

[131] Eitan, A., Fisher, F. T., Andrews, R., Brinson, L. C., and Schadler, L. S., 2006, "Reinforcement Mechanisms in Mwcnt-Filled Polycarbonate," Composites Science and Technology, 66(9), pp. 1162-1173.

[132] Smith, N. A., Antoun, G. G., Ellis, A. B., and Crone, W. C., 2004, "Improved Adhesion between Nickel-Titanium Shape Memory Alloy and a Polymer Matrix Via Silane Coupling Agents," Composites Part A: Applied Science and Manufacturing, 35(11), pp. 1307-1312.

[133] Pukanszky, B., 2005, "Interfaces and Interphases in Multicomponent Materials: Past, Present, Future," European Polymer Journal, 41(4), pp. 645-662.

[134] Rong, M. Z., Zhang, M. Q., and Ruan, W. H., 2006, "Surface Modification of Nanoscale Fillers for Improving Properties of Polymer Nanocomposites: A Review," Materials Science and Technology, 22(pp. 787-796.

[135] Qian, H., Greenhalgh, E. S., Shaffer, M. S. P., and Bismarck, A., 2010, "Carbon Nanotube-Based Hierarchical Composites: A Review," Journal of Materials Chemistry, 20(23), pp. 4751-4762.

139

[136] Hussain, M., Nakahira, A., and Niihara, K., 1996, "Mechanical Property Improvement of Carbon Fiber Reinforced Epoxy Composites by Al2o3 Filler Dispersion," Materials Letters, 26(3), pp. 185-191.

[137] Chou, T.-W., Gao, L., Thostenson, E. T., Zhang, Z., and Byun, J.-H., 2010, "An Assessment of the Science and Technology of Carbon Nanotube-Based Fibers and Composites," Composites Science and Technology, 70(1), pp. 1-19.

[138] Wicks, S. S., De Villoria, R. G., and Wardle, B. L., 2009, "Interlaminar and Intralaminar Reinforcement of Composite Laminates with Aligned Carbon Nanotubes," Composites Science and Technology, 70(1), pp. 20-28.

[139] Zhang, W., Sakalkar, V., and Koratkar, N., 2007, "In Situ Health Monitoring and Repair in Composites Using Carbon Nanotube Additives," Applied Physics Letters, 91(13), pp. 133102-3.

[140] Sadeghian, R., Gangireddy, S., Minaie, B., and Hsiao, K.-T., 2006, "Manufacturing Carbon Nanofibers Toughened Polyester/Glass Fiber Composites Using Vacuum Assisted Resin Transfer Molding for Enhancing the Mode-I Delamination Resistance," Composites Part A: Applied Science and Manufacturing, 37(10), pp. 1787-1795.

[141] Gojny, F. H., Wichmann, M. H. G., Fiedler, B., Bauhofer, W., and Schulte, K., 2005, "Influence of Nano-Modification on the Mechanical and Electrical Properties of Conventional Fibre-Reinforced Composites," Composites Part A: Applied Science and Manufacturing, 36(11), pp. 1525-1535.

[142] Kinloch, I. A., Roberts, S. A., and Windle, A. H., 2002, "A Rheological Study of Concentrated Aqueous Nanotube Dispersions," Polymer, 43(26), pp. 7483-7491.

[143] Wicks, S. S., De Villoria, R. G., and Wardle, B. L., "Interlaminar and Intralaminar Reinforcement of Composite Laminates with Aligned Carbon Nanotubes," Composites Science and Technology, 70(1), pp. 20-28.

[144] Liyong Tong, Xiannian Sun, and Ping Tan, 2008, "Effect of Long Multi-Walled Carbon Nanotubes on Delamination Toughness of Laminated Composites," Journal of Composite Materials, pp. 5-23.

[145] Guz, I. A., Rodger, A. A., Guz, A. N., and Rushchitsky, J. J., 2008, "Predicting the Properties of Micro- and Nanocomposites: From the Microwhiskers to the Bristled Nano-Centipedes," Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, 366(1871), pp. 1827-1833.

[146] Wichmann, M. H. G., Sumfleth, J., Gojny, F. H., Quaresimin, M., Fiedler, B., and Schulte, K., 2006, "Glass-Fibre-Reinforced Composites with Enhanced Mechanical and Electrical Properties - Benefits and Limitations of a Nanoparticle Modified Matrix," Engineering Fracture Mechanics, 73(16), pp. 2346-2359.

[147] Karapappas, P., Vavouliotis, A., Tsotra, P., Kostopoulos, V., and Paipetis, A., 2009, "Enhanced Fracture Properties of Carbon Reinforced Composites by the Addition of Multi-Wall Carbon Nanotubes," Journal of Composite Materials, pp. 977-985.

140

[148] Yokozeki, T., Iwahori, Y., Ishiwata, S., and Enomoto, K., 2007, "Mechanical Properties of Cfrp Laminates Manufactured from Unidirectional Prepregs Using Cscnt-Dispersed Epoxy," Composites Part A: Applied Science and Manufacturing, 38(10), pp. 2121-2130.

[149] Kinloch, A. J., Lee, J. H., Taylor, A. C., Sprenger, S., Eger, C., and Egan, D., 2003, "Toughening Structural Adhesives Via Nano- and Micro-Phase Inclusions," The Journal of Adhesion, 79(8), pp. 867-873.

[150] Ashrafi, B., and Hubert, P., 2006, "Modeling the Elastic Properties of Carbon Nanotube Array/Polymer Composites," Composites Science and Technology, 66(3-4), pp. 387-396.

[151] Saxena, A., 1998, Nonlinear Fracture Mechanics for Engineers CRC Press.

[152] Anderson, T. L., 1994, Fracture Mechanics: Fundamentals and Applications, CRC Press, Florida.

[153] Barber, A. H., Cohen, S. R., and Wagner, H. D., 2003, "Measurement of Carbon Nanotube-Polymer Interfacial Strength," Applied Physics Letters, 82(23), pp. 4140-4142.

[154] Guo, T., Nikolaev, P., Thess, A., Colbert, D. T., and Smalley, R. E., 1995, "Catalytic Growth of Single-Walled Manotubes by Laser Vaporization," Chemical Physics Letters, 243(1-2), pp. 49-54.

[155] Ren, Z. F., Huang, Z. P., Xu, J. W., Wang, J. H., Bush, P., Siegal, M. P., and Provencio, P. N., 1998, "Synthesis of Large Arrays of Well-Aligned Carbon Nanotubes on Glass," Science, 282(5391), pp. 1105-1107.

[156] Nanoledge, 2010, "Processing Methods for A1d1 Bisphenol a Epoxy Resin Filled with Carbon Nanotubes."

[157] Huntsman, C., 2003, Araldite® My0510 Material Data Sheet.

[158] Astm, 2007, "Astm D5045 - 99: Standard Test Methods for Plane-Strain Fracture Toughness and Strain Energy Release Rate of Plastic Materials."

[159] Qiu, J., and Wang, S., "Reaction Kinetics of Functionalized Carbon Nanotubes Reinforced Polymer Composites," Materials Chemistry and Physics, 121(1-2), pp. 295-301.

[160] Xie, H., Liu, B., Yuan, Z., Shen, J., and Cheng, R., 2004, "Cure Kinetics of Carbon Nanotube/Tetrafunctional Epoxy Nanocomposites by Isothermal Differential Scanning Calorimetry," Journal of Polymer Science Part B: Polymer Physics, 42(20), pp. 3701-3712.

[161] Kingston, C. T., Jakubek, Z. J., Denommee, S., and Simard, B., 2004, "Efficient Laser Synthesis of Single-Walled Carbon Nanotubes through Laser Heating of the Condensing Vaporization Plume," Carbon, 42(8-9), pp. 1657-1664.

[162] Bayer, M., 2006, Baytubes Development Product C150 P.

[163] Astm, 2007, "Astm D5528 - 01 Standard Test Method for Mode I Interlaminar Fracture Toughness of Unidirectional Fiber-Reinforced Polymer Matrix Composites."

[164] Romhany, G., and Szebanyi, G., 2009, "Interlaminar Crack Propagation in Mwcnt/Fiber Reinforced Hybrid Composites," Express Polymer Letters, 3(3), pp. 145-151.

141

[165] Jordan, W. M., and Bradley, W. L., 1988, "The Relationship between Resin Mechanical Properties and Mode I Delamination Fracture Toughness," Journal of Materials Science Letters, 7(12), pp. 1362-1364.

[166] Siddiqui, N. A., Woo, R. S. C., Kim, J.-K., Leung, C. C. K., and Munir, A., 2007, "Mode I Interlaminar Fracture Behavior and Mechanical Properties of Cfrps with Nanoclay-Filled Epoxy Matrix," Composites Part A: Applied Science and Manufacturing, 38(2), pp. 449-460.

[167] Bradley, W. L., and Cohen, R. N., 1985, "Matrix Deformation and Fracture in Graphite-Reinforced Epoxies," Proc. ASTM Special Technical Publication, W. S. Johnson, ed. Pittsburgh, PA, USA, pp. 389-410.

[168] Compston, P., Jar, P. Y. B., and Takahashi, K., 2000, "The Use of Crack Opening Displacement Rate to Assess Matrix-to-Composite Mode I Toughness Transfer," Journal of Materials Science Letters, 19(1), pp. 17-19.

[169] Compston, P., Jar, P. Y. B., Burchill, P. J., and Takahashi, K., 2002, "The Transfer of Matrix Toughness to Composite Mode I Interlaminar Fracture Toughness in Glass-Fibre/Vinyl Ester Composites," Applied Composite Materials, 9(5), pp. 291-314.

[170] Karger-Kocsis, J., 2001, "Stick-Slip Type Crack Growth During Instrumented High-Speed Impact of Hdpe and Hdpe/Selar Discontinuous Laminar Microlayer Composites," Journal of Macromolecular Science, Part B: Physics, 40(3), pp. 343 - 353.

[171] Webb, T. W., and Aifantis, E. C., "Crack Growth Resistance Curves and Stick-Slip Fracture Instabilities," Mechanics Research Communications, 24(2), pp. 123-130.

[172] Tugrul Seyhan, A., Tanoglu, M., and Schulte, K., 2008, "Mode I and Mode Ii Fracture Toughness of E-Glass Non-Crimp Fabric/Carbon Nanotube (Cnt) Modified Polymer Based Composites," Engineering Fracture Mechanics, 75(18), pp. 5151-5162.

[173] Cowley, K. D., and Beaumont, P. W. R., 1997, "The Interlaminar and Intralaminar Fracture Toughness of Carbon-Fibre/Polymer Composites: The Effect of Temperature," Composites Science and Technology, 57(11), pp. 1433-1444.

[174] Russell, A. J., 1987, "Micromechanisms of Interlaminar Fracture and Fatigue," Polymer Composites, 8(5), pp. 342-351.

[175] Kim, H. S., and Ma, P., 1998, "Mode Ii Fracture Mechanisms of Pbt-Modified Brittle Epoxies," Journal of Applied Polymer Science, 69(2), pp. 405-415.

[176] Lee, S., 1997, "Mode Ii Delamination Failure Mechanisms of Polymer Matrix Composites," Journal of Materials Science, 32(5), pp. 1287-1295.

[177] Davidson, B. D., and Sun, X., 2006, "Geometry and Data Reductions for a Standardized End Notched Flexure Test for Unidirectional Composites," Journal of ASTM International, 3(9).

142

Appendix A

The Matlab code for image analysis is presented in Appendix A. 1. In Appendix A. 2. the

results of dispersion quantification curves are given for all the formulations used in

Chapter 4.

A. 1. Matlab code for image analysis

p = which('N_01.JPG'); filelist = dir([fileparts(p) filesep 'N_**.JPG']); fileNames = filelistname'; loop=size(fileNames); fprintf('\n') fprintf('This program quantifies the dispersion degradation. Please enter

information about the cure cycle.\n\n') start_temp = input('Please input initial temperature in degree C: '); end_temp = input('Please input final temperature in degree C: '); ramp_rate = input('Please input ramp rate in degree C/min: '); image_interval = input('Please input imaging interval in seconds: ');

time(1) = 0; temp(1)= start_temp;

for j= 1:(loop(1)-1) time(j+1)=time(j)+image_interval; if temp(j)>= end_temp; temp(j+1)= end_temp; else temp(j+1)= temp(j)+ (ramp_rate)*image_interval/60; if temp(j+1)>= end_temp; temp(j+1)= end_temp; else temp(j+1); end end %if end

for i= 1:loop(1) rgb(:,:,:,i) = imread(filenames(i)); gray(:,:,:,i) = rgb2gray(rgb(:,:,:,i)); gray_imdouble(:,:,:,i) = im2double(gray(:,:,:,i)); gray_norm(:,:,:,i) = ((gray_imdouble(:,:,:,i) -

(min(min(gray_imdouble(:,:,:,i)))))/ max(max(gray_imdouble(:,:,:,i)))); gray_aver(i) = (mean(mean(gray_norm(:,:,:,i)))); level(i) = graythresh(gray_norm(:,:,:,i)); end

gray_average = mean(mean(gray_aver)) levelavg = mean(mean(level))

for i= 1:loop(1) bw(:,:,:,i) = im2bw(gray_norm(:,:,:,i), gray_average);

143

[m n]=size(bw(:,:,:,i)); resin = bwarea(bw(:,:,:,i)); total = m*n; vfCNT(i) = 1-(resin/total); end

for i= 1:loop(1) vf(i) = vfCNT(i)/max(max(vfCNT)); end

figure,

[AX,H1,H2] = plotyy(time,vf,time,temp,'plot');

set(get(AX(1),'Ylabel'),'String','Af') set(get(AX(2),'Ylabel'),'String','T(\circC)')

set(AX(2),'Ylim',[0 end_temp+20],'YTick',[0:20:end_temp+20]) set(AX(1),'Ylim', [0 1],'YTick',[0:0.2:1])

set(gca,'box','off')

xlabel('Time(sec)') title('Dispersion & Temp vs. Time')

set(H1,'LineStyle','-','marker','.') set(H2,'LineStyle','-')

144

A. 2. Dispersion Quantification results

SWNT System

DDS 100:49 5C/min unfunctionalized 0.3%

0 200 400 600 800 1000 1200 1400 1600 18000

0.2

0.4

0.6

0.8

1

Af

Time(sec)


0 200 400 600 800 1000 1200 1400 1600 18000

20

40

60

80

100

120

140

160

180

200

220

T(

C)

MY3_24

145

DDS 100:49 unfunctionalized SWNT 100 c/min

0 50 100 150 200 250 300 350 400 4500

0.2

0.4

0.6

0.8

1

Af

Time(sec)


0 50 100 150 200 250 300 350 400 4500

20

40

60

80

100

120

140

160

180

200

220

T(

C)

146

MWNT system: No hardener

0 500 1000 1500 2000 2500 3000 35000

0.2

0.4

0.6

0.8

1

Af

Time(sec)


0 500 1000 1500 2000 2500 3000 35000

20

40

60

80

100

120

140

T(

C)

147

TETA

0 1 2 3 4 5 6 7 8

x 104

0

0.2

0.4

0.6

0.8

1

Af

Time(sec)


0 1 2 3 4 5 6 7 8

x 104

0

20

40

T(

C)

148

IPD:TETA (20:80)

0 500 1000 1500 2000 2500 3000 35000

0.2

0.4

0.6

0.8

1

Af

Time(sec)


0 500 1000 1500 2000 2500 3000 35000

20

40

T(

C)

149

IPD:TETA 50:50

0 1 2 3 4 5 6 7 8 9

x 104

0

0.2

0.4

0.6

0.8

1

Af

Time(sec)


0 1 2 3 4 5 6 7 8 9

x 104

0

20

40

T(

C)

150

Appendix B

The data reduction techniques used to find the Mode I and II delamination properties in

the Chapter 5 is presented in this Appendix B.

Mode I data analysis

3 data reduction techniques for Mode I interlaminar fracture toughness measurements

are given in the ASTM standard (ASTM D5528-01), i.e. Modified Beam Theory, Compliance

Calibration, and Modified Compliance Calibration. The strain energy release rate equation

corresponding to each method is given in Table B-1. Fracture toughness was computed

with all three equations at every point. The lowest values at every point were retained.

The lowest overall value stemmed from the first point (green circle) and was considered as

the initiation fracture toughness value. The subsequent points (red triangles) generated

the propagation fracture toughness values. The constants in each equation can be found

based on Figure B-1.

Table B-1. Data reduction techniques to obtain fracture toughness.

Modified Beam Theory Compliance Calibration

Modified Compliance Calibration

GI

MBT CC MCC

Figure B-1. Constants used in the fracture toughness data reduction presented in Table B-1

C1/3

aΔ

Log C

Log a

n = Δy/Δx

Δy

Δx

a/h

C1/3

A1 = Δy/Δx

Δy

Δx

151

Mode II data analysis

The objective of the non-precracked and the precracked fracture tests was only to capture

delamination initiation, since delamination growth was highly unstable in Mode II [173].

For the first non-precracked compliance calibration analysis, with a0=20mm, the load and

extension were first adjusted to start from zero to eliminate the initial non-linearity in the

values. The max load value was then noted and the load versus displacement graph was

plotted. From the slope of a curve fitted to the linear region the compliance of the sample

was calculated. The same procedure was followed for the second non-precracked

compliance calibration analysis, with a0=40mm.

For the fracture test where the unloading data were also recorded, the load versus

displacement graph was obtained for both loading and unloading. The compliance for

loading and unloading was calculated as C0 and Cu. The non-precracked compliance

calibration coefficients of A and m were found using the slope and the intercept of the C0

versus a03 curve. Using these values the acalc was calculated using Equation B-1. The Pmax

for the fracture test was found and the candidate toughness was calculated from Equation

B-3. Finally, the candidate toughness was validated from Equation B-4, and if it was in the

range of 15 ≤ %GQ ≤ 35, the candidate toughness was accepted, if not the results from this

test are discarded.

Given that the results were approved, the data from the precracked compliance

calibration tests and the fracture test was analyzed following the same procedure as the

non-precracked. The precracked critical mode II strain energy release rate, GIIc was

obtained.

Compliance calibration. Prior to the non-precracked and the precracked fracture,

compliance calibration tests were performed. In both the non-precracked and the

precracked fracture configurations, for the first compliance calibration test, the specimen

was positioned such that the distance between the center of the support roller on the

delaminated end and the delamination tip was a0 - 10 mm (20 mm). For the second

compliance calibration test, this distance was set to a0 + 10 mm (40 mm). For all

compliance calibration tests, specimens were loaded to 50% of the expected critical load

at that particular delamination length (i.e., load that would initiate delamination growth).

The width of the sample B was recorded at the three contact locations of the three rollers

when the specimen is tested. Also, the thickness of the specimen, 2h, was recorded at six

locations, two at the same three locations where the width of the specimen was recorded,

one on right and one on the left side. The variation in thickness should be less than 0.1mm

and the variation in width should be less than 0.75 mm.

152

The edges of the specimen were coated with a white paint to locate the delamination tip

and mark the compliance calibration markings as shown in Figure B-2. The Teflon insert tip

was marked and the compliance calibration marking was performed at a0 from the Teflon

insert tip, and at 10mm on either side of this mark.

Figure B-2. ENF Test Fixture and Specimen [177]

The specimen was then placed on the rollers such that its longitudinal direction was

perpendicular to the loading rollers. The compliance calibration and the fracture tests

were conducted at a displacement rate of 0.5mm/min in loading and 1.6mm/min in

unloading. During the two compliance calibration tests, the peak loads were 50% of the

expected value of the critical fracture load at that particular length. These loads were

approximated based on previous experiments on similar material system.

Experimentally, the specimen was first tested at the non-precracked compliance

calibration of 20mm and 40mm, prior to the non-precracked fracture test at 30mm. For

the non-precracked fracture test, the sample was placed at a=a0=30mm and the specimen

was loaded until the delamination advances from the Teflon tip, which was noted by a

drop in the load versus displacement curve. From the unloading data of the non-

precracked fracture test, the value of acalc was calculated, using:

Equation B-1

13

ucalc

C Aa

m

153

where Cu is the compliance of the non-precracked test unloading line. A and m were

determined using a linear least-squares regression of the 3 NPC compliances versus the

crack length, determined using:

Equation B-2

where A is the intercept and m is the slope. The three compliances were those from the

two CC tests (at a0 -10 mm, a0 +10 mm) and the NPC fracture test. Once the acalc was

calculated, it was marked as the new precracked crack tip as shown in Figure B-3 and new

precracked compliance calibration marking is placed at 30mm and at 10mm to either side

of this mark. Then the specimen was tested in a similar manner to the non-precracked

test, but this time using the precracked compliance calibration marking. At first the

specimen compliance was calibrated at 20mm from the precracked crack tip mark and

then at 40mm. Finally the specimen was placed at the centre of the precracked

compliance calibration mark and loaded until the delamination advances by noticing a

drop on the load versus displacement curve.

Figure B-3. Schematic of the configuration when the same sample is used for the NPC and PC test

The specimen fracture toughness for both non-precracked and the pre-cracked tests were

determined using:

Equation B-3

where m is the compliance calibration coefficient found using the method mentioned

above, Pmax is the maximum load from the fracture test, a0 is the crack length of the

fracture test and B is the average specimen length, calculated from the three values

recorded initially.

3C A ma

2 23

2

Max oQ

mP aG

B

154

The validation of the non-precracked and precracked tests was determined using the %GQ

achieved during the compliance calibration calculated using:

Equation B-4

where Pj is the peak load value achieved during the compliance calibration. The test was

accepted only when the 15 ≤ % GQ ≤ 35, otherwise the results from the test is rejected.

2

2

100% ; 1,2

j j

Q

Max o

P aG Max j

P a

Date post:	14-Sep-2018
Category:	Documents
Upload:	nguyenthuan
View:	216 times
Download:	0 times

Aspects of the Fracture Toughness of Carbon Nanotube...

Documents