PLACE IN RETURN BOX to remove this checkout from your record. To AVOID FINES return on or before date due. MAY BE RECALLED with earlier due date if requested. DATE DUE DATE DUE DATE DUE 5/08 K:IProj/Acc&Pres/CIRCIDateDuo.indd HYBRID HIE} DEVELOPME .\”I EH’ERIMENTATH in pa HYBRID HIERARCHICAL BIO-BASED MATERIALS: DEVELOPMENT AND CHARACTERIZATION THROUGH EXPERIMENTATION AND (I)MPUTATIONAL SIMULATIONS By Mahmoodul Haq A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requilements forthe degree of DOCTOR OF PHILOSOPHY CIVIL ENGINEERING 2009 ammo H DEVELOPME' mmm'rx Emiromnufly fnc m: by harnessing the s 2: 25m» scale) minforcer “2:3 rains. Biobascd cox momma] appefl a use of male m ”3 Combmed With bio] I. m- ._ _ v...‘|_= T‘ ABSTRACT HYBRID HIE RARCHICAL BIO-BASED MATERIALS: DEVELOPMENT AND CHARACTERIZATION THROUGH EXPERIMENTATION AND COMPUTATIONAL SIMULATIONS By Mahmoodul Haq Environmentally friendly bio-based composites with improved properties can be obtained by harnessing the synergy offered by hybrid constituents such as multiscale (nano- and micro- scale) reinforcement in bio-based resins composed of blends of synthetic and natural resins. Bio-based composites have recently gained much attention due to their low cost, environmental appeal and their potential to compete with synthetic composites. The advantage of mulciscale reinforcement is that it offers synergy at various length scales, and when combined with bio-based resins provide stiffness-toughness balance, improved thermal and banier properties, and increased environmental appeal to the resulting composites. Moreover, these hybrid nnterials are milorable in performance and in environmental impact. While the use of different concepts of multiscale reinforcement has been studied for synthetic composites, the study of multiphase/multiscale reinforcements for developing new types of sustainable nnterials is limited. The research summarized in this dissertation focused on development of multiscale reinforced bio-based composites and the effort to understand and exploit the synergy of its constituents through experimental characterization and computational simulations. Bio-based composites consisting of petroleum-based resin (unsaturated polyester), natural or bio-resin (epoxidized soybean and linseed oils), natural fibers (industrial hemp), and nanosilicate (nanoclay) inclusions were developed. The work followed the “nnterials by 25:: pirate b? : meats} explore :"cxi‘s oz bio res in 3:. at comm- balance 5 gm smog-Jr an 2:225: tougl'urss. '1 $3sz can be bros; 23: A mLfiIi'scale :tzrnmi to the d5 3;: m 3330 Scale 39.225: ””3531 w. in: problem to dcte design” philosophy by incorporating an integrated experimental and computational approach to strategically explore the design possibilities and limits. Experiments demonstrated that the drawbacks of bio-resin addition, which lowers stiffness, strength and increases permeability, can be counter-balanced through nanoclay reinforcement. Bio-resin addition yields benefits in impact strength and ductility. Conversely, nanoclay enhances stiffness but seems to decrease toughness. Thus, the traditionally seen opposite measures of stiffness and toughness can be brought to an efficient balance through the combination of bio-resin and nanoclay. A multiscale computational approach, namely a multi-FE based approach, was implemented to the developed materials to extrapolate the experimental matrix, to provide insight into nano-scale behavior beyond measurements and to hopefully serve as a tool for computational design of hybrid materials. Additionally, an enhanced RVE for modeling the three-phase rmterial was determined by solving a topology optimization based material layout problem to determine the distribution of bio- resin, thereby allowing modeling the nanostructure in greater detail and closer to reality. Overall, eco-friendly, tailorable, cost-effective and rmrltiscale reinforced bio-based composites were successfully developed. The improved multifaceted features possible for these sustainable bio-based materials are likely to increase their appeal for use in transportation and housing structural applications. Additionally, it is believed that the approach of understanding complex materials by integrating simulations and experiments, as attempted in this work, holds great promise, and a similar methodology can be applied for other types of hierarchical mterials, thereby providing guidance in designing those materials. Capyright ° Mahmoodul Haq 2009 To mypmwm' 4714’ my (1/529, firtéezicoflzzhtzowpmyem mafia“, gaze/472m and 51.7me A( Praise b6 10 the 1”: lIL‘ $131k to menu 5:35:21 ivfipnmn‘ and h“ Tn... gm fun for me Tl‘ mm m heli‘d me m 75;: Ere been 1051 withOUI W Hill to thank ihc me 1 Lip. Dr. Romld Harichand: Wcomnrnts. Additic Dim; Elm for sharing it 4.31M encouragement a lwuld lie to think .\ “:15 (Enter for their help fell. Siarosh Ravanbal'sh is I would also like r. :5. 3" lpineleCtron micms In) uldlflsetothanlti) "c h I “’Ollld also like *- L ‘ WmlflaHtMSU ACKNOWLEDGEMENTS Praise be to the LORD, The Most Beneficent, The Most Merciful It is difficult to overstate my gratitude to my advisor, Dr. Rigoberto Burguefio. His enthusiasm, inspiration, and his efforts to explain things clearly and simply, helped to make research great fun for me. Throughout my dissertation, his encouragement, sound advice and great vision helped me navigate through the difficult times during my PhD studies. I would have been lost without him. I wish to thank the members of PhD committee, Dr. Armr Mohanty, Dr. Manjusri M'ma, Dr. Ronald Harichandran, and Dr. Alejandro Diaz for their valuable time and constructive comments. Additionally, very special thanks are due to Dr. Amar Mohanty and Dr. Manjusri Misra for sharing with me their expertise in biocomposite rmterials, and ofr their guidance, encouragement and support throughout this research. I would like to thank Mr. Michael Rich and the staff of Composite Materials and Structures Center for their help in manufacturing and material testing. Also, I would like to thank Mr. Siavosh Ravanbaksh and Mr. Joseph Nguyen for their help at Civil Engineering Laboratories. I would also like to thank the staff, at MSU’s Center of Advanced Microscopy for their help in electron microscopy imaging. I would like to thank Dr. Hiroaki Miyagawa for his initial help in processing and understanding of materials. I thank Janelle Musch, Richard McGary and Lindsey Diggelmann for their help in manufacturing of biocomposite materials at different stages of this research. I would also like to thank my colleagues for their friendship and support throughout my stay at MSU. lastly, and most importantly, I would like to thank my family, specially my parents and mywife Sarah, for their patience, sacrifices and support. To them, I dedicate this thesis. vi ~'='=.'.ll E0 m}. k ‘5'... r gal 0““. (I :azrei ire most- to a T23 study W1 3.53429o66. Additi: at": were obtained f: Emirate School I am grateful to my brother Zia for having confidence in me and supporting me when it mattered the most: to chase mydream. This study was rminly funded by the National Science Foundation under grant (MS-0409666. Additionally, financial support in the form of assistanships and fellowship awards were obtained from the Department of Civil and Environmental Engineering, and the Graduate School at Michigan State University. All contributions are gratefully acknowledged. LIST Of TABLES ........... LET Of HGL'RES ......... CliLPTfR l. BRODL'CIION........... 1.1 Shannon ........ ....... 1.‘ theatres ................ 3 Baromti, 15.! Bit based compo- 132 Ei‘mccnrnt ' .r 135 Hind Biocompc 15.4 liaising and Sim 135 Stimuli: methoc L‘- llf‘i‘rod / Approach 1.4.1 Exterinmtation - ”1.1.3 Slflmlanons - COD. * ‘C‘PC ----.-...m........ “E'~"xization.............. 5 Task. and Figure-5m r :3 , .. Lemmas...“ . " COO-nu... Glam; WHORE \T OF ‘ . B ‘1 Wm-.................l( ; Lr~arron..........§: femur Method 4.1 liar-HIS In; ' A :1 1Twang“Rubles 52 So ECO?“ ml is anon DC 33 , . rg‘v 1-5 5 ml Want 1 R5) am 4“ Q» ”~............ ”~62 In; ”“855ng TABLE OF (DNTENTS LIST OF TABLES LIST OF FIGJRES CHAPTER 1. INTRODUCTION xiii xiv 1.1 Motivation 1.2 Objectives 13 Background 1.3.1 Bio based composite materials and Biocomposites 1.3.2 Enhancement through Nano-reinforcement 8r. Nanocomposite processing ........... 7 1.3.3 Hybrid Biocomposites from Bio-based Resins 8: Nanoclays 10 1.3.4 Modeling and simulation of micro/nano reinforced polymers 11 1.3.5 Multiscale methodology 14 1.4 Method / Approach 16 1.4.1 Experimentation - Processing 86 Characterization 17 1.4.2 Simulations - Computational Models 8: Multiscale Modeling 17 15 Scope 18 1.6 Organization 20 1.7 Tables and Figures 21 1.8 References 24 CHAPTER 2. DEVELOPMENT OF BIO-BASED CLAY NAN OCOMPOSITES 28 2.1 Abstract 78 2.2 Introduction 28 2.3 Experimental Methods 34 2.3.1 Materials 34 2.3.2 Testing 8?. Characterization 35 2.4 Processing Techniques 35 2.5 Processing Variables 39 2.5.1 Styrene Content in Resin System 39 2.5.2 Sonication Energy 39 2.5.3 Material Content and Nomenclature 40 2.6 Results 41 2.6.1 Overall Processing Efficiency 41 2.6.2 Tensile Properties 42 2.6.3 Morphology of Nanoclay Inclusions 45 2.6.4 Fractographic Observations 46 2.7 Discussion 47 2.8 Conclusions 49 2.9 Tables and Figures 51 2.10 References ‘59 Willi}. . ‘ E‘lOBfiED POLYMER N A 3.; .‘li-‘Sfl'iWa-mm ,. limits-Leon .............. ....... 33' Erpc‘nxmal Mahods 35.1 limb .................. 352 Extramural Matrix .1 355 PohnrrNanocomnss 35.4 icing 5i Gummizs 3.4.1 lens'lc Tests 6: Prom 3.4.” om.- .Wchanical .5 3.45 llrrzml Mechanical A 3.4.4 litter Absorption '1 3.1.5 Ersgv Absorbed per 1 3‘5 Transmissionlilettron 3‘ lmognphic Obsen a 5Dsnssinn Eh CK. rlmswm m‘”“°-”n Coo-ca idnosledgements .......... ' labesandfrgurts ........... 33 Rd- motes...” :‘jr Minn :::E " Emm’l‘tmaj \lflll’l CHAPTER 3. BIO-BASED POLYMER NANOCOMPOSITES: I - UPE / EMS 62 3.1 Abstract 62 3.2 Introduction 62 3.3 Experimental Methods 66 3.3.1 Mterials 66 3.3.2 Experimental Matrix and Nomenclature 66 3.3.3 Polymer Nanocomposite Processing 67 3.3.4 Testing 8: Characterization 69 3.4 Results 71 3.4.1 Tensile Tests 8?. Properties 71 3.4.2 Dynamic Mechanical Analysis 72 3.4.3 Thermal Mechanical Analysis 72 3.4.4 Moisture Absorption Tests 73 3.4.5 Energy Absorbed per Unit Volume (Toughness) 74 3.4.6 Transmission Electron Microscopy 75 3.4.7 Fractographic Observations (Scanning Electron Microscopy) 76 3.5 Discussion 77 3.6 Conclusions 80 3.7 Acknowledgements 81 3.8 Tables and Figures 82 3.9 References 90 CHAPTER 4. BIO-BASED POLYMER NANOCOMPOSITES: II - UPE / EML 92 4.1 Abstract 92 4.2 Introduction 92 4.3 Experimental Methods 96 4.3.1 Materials 96 4.3.2 Experimental matrix and nomenclature 97 4.3.3 Polymer nanocomposite processing 97 4.3.4 Testing and characterization 98 4.4 Results 99 4.4.1 Tensile tests and properties 99 4.4.2 Thermal properties 102 4.4.3 Moisture absorption 104 4.4.4 Impact strength ' 106 4.4.5 Nanoclay dispersion and morphology 108 4.5 Discussion 109 4.5.1 Properties and synergistic behavior 110 4.5.2 Perforrmnce limits and optimized material design 112 4.6 Conclusions 113 4.7 Tables and Figures 115 4.8 References , 122 . MERE. i1. ALW WW“ :‘ mtk‘fion-ww'“ 5:, Experimental meth 55.1 Bismark w~j°~° 532 Nanompoml 55.3 streaming“ 5.4 Thing 8% character :5 Results 55.1 Tensile modulus 352 Ebngationatfail 55.} Thermal propertic 55.4 Nanoclaydispersi 555 larrresuriace c 52' Dsussion................ 53 Comlosions .............. 33 Tables and Figures... OO WOO...- ERTI f {meon “MES ( CHAPTER 5. MULTISCALE HYBRID BIOCOMPOSITES: I - HF / UPE / EMS ..................... 125 5.1 Abstract 125 5.2 Introduction 125 5.3 Experimental methods 128 5.3.1 Materials 128 5.3.2 Nanocomposite processing -129 5.3.3 Manufacturing of hybrid biocomposite plates 130 5.4 Testing 8?. characterization 131 5.5 Results 132 5.5.1 Tensile modulus and ultirrrate tensile strength 132 5.5.2 Elongation at failure and notched Izod impact strengths 133 5.5.3 Thermal properties and moisture absorption 134 5.5.4 Nanoclay dispersion and exfoliation. 135 5.5.5 Fracture surface observations. 135 5.6 Discussion 137 5.7 Conclusions 140 5.8 Tables and Figures 141 5.9 References 146 CHAPTER 6. MULTISCALE HYBRID BIOCOMPOSITES: II - HF / UPE / EML .................... 148 6.1 Abstract 148 6.2 Introduction 148 6.3 Experimental methods 152 6.3.1 Materials 152 6.3.2 Experimental rmtrix and nomenclature 153 6.3.3 Polymer nanocomposite processing 153 6.3.4 Manufacturing of hybrid biocomposite plates 154 6.3.5 Testing and characterization 155 6.4 Results 156 6.4.1 Material characterization study 156 6.5 Discussion 165 6.6 Conclusions 170 6.7 Tables and Figures 172 6.8 References 181 CHAPTER 7. ANALYTICAL AND NUMERICAL MODELING OF MECHANICAL AND BARRIER PROPERTIES OF CLAY/ POLYMER NAN OCOMPOSITES .............. 184 7.1 Introduction - _ 184 7.2 Micromechanical Modeling of PNC’s 188 7.2.1 Mori Tanaka Estimates 188 7.3 Unit Cell Methods / FE - RVE based method 198 7.3.1 Representative Volume Element (RVE)- Selection and Boundary Conditions ...199 7.3.2 RVE Analyses 702 7.3.3 Results of RVE Analyses 204 7.4 Comparison of Tensile Modulus Results with Theoretical and FE predictions. ..... .205 X 'i, Theortucal Models i FE-RLE based or: .' Litre Difusion Tl hbismrt diffusiv 1'.” Moisture diffusn' 1'5 font-rm diffus iv. '3 Grandson of Diff '3 Damion of Dem 'ii Cortland ...me iii Til-ES 5! Figures ...... 0“ .. Madame“--- GilPTliRS. E-BASED MLLTISC'U. lj Alisha? L’ imitation .............-- lhln kl'cl Approach is; .V "rials ............... til Qtpttatioml Ho. hi Public Boundary H 17%?th Macrosth 1w ' . uplementatror ll“ llzhlevelFE Model ; lilo-kid FE Model 1‘ Ill-Ems and Discussior ii. obasecl Pohmfl : .- r 3.0., I 1": 74.) “mo“ Of WJd l 7.44 ”million of Tar lij e‘leld'ofiated, 1. 74] Cited: E’dOllated‘i '4: Embed, i 7.5 Theoretical Models for Diffusion / Heat Transfer Problems 205 7.6 FE-RVE based Diffusion Modeling and Comparison with Experiments .............. .208 7.7 Moisture Diffusion Results 709 7.7.1 Moisture diffusivity through neat resins (no clay) 710 7.7.2 Moisture diffusivity in virgin UPE reinforced with nanoclay 710 7.7.3 Moisture diffusivity in UPE/EML blends reinforced with nanoclay ................... .211 7.8 Comparison of Diffusion Results: Experiment and Theory 712 7.9 Discussion of Diffusion Results 713 7.10 Conclusion 716 ' 7.11 Tables 8CFigures 218 7.12 References 741 CHAPTER 8. FE-BASED MULTISCALE HOMOGENIZATION’ 243 8.1 Abstract 743 8.2 Introduction 244 8.3 Multilevel Approach 749 8.3.1 Materials 749 8.3.2 Computational Hornogenization Hypotheses 750 8.3.3 Periodic Boundary Conditions (PBC) 251 8.3.4 Coupling Macrostructure and Mcrostructure 752 8.3.5 FE Implementation 754 8.4 Multilevel FE Model of bio-based Polymer/ Clay Composites 755 8.4.1 Macromesh, RVE (micromesh), Material Models and Computational Aspects...256 8.5 Multi-level FE Model of Hybrid Biocomposites: 260 8.6 Results and Discussion 761 8.6.1 Bio-based Polymer Nanocomposites 762 8.6.2 Hybrid Bio-based Composites. 767 8.7 Conclusion 768 8.8 Appendix-Derivation of Tangential Stiffness Matrix 269 8.9 Tables 8: Figures 775 8.10 References 786 CHAPTER 9. MODELING OF THREE-PHASE BIO-BASED NAN OCOMPOSITES: DETERMINING BIO-RESIN DISTRIBUTION USING AN OPTIMIZATION- BASED MATERIAL DESIGN PROBLEM 788 9.1 Abstract: 288 9.2 Introduction: 290 9.3 Materials Design / Layout as an Optimization Problem 793 9.4 Case Studies: 797 9.4.1 Introduction to the models/ cases: 797 9.4.2 Description of models 798 9.4.3 Determination of Target Properties 298 9.4.4 Case-1: Exfoliated, 10% bio-resin, 10% available for layout optimization ........... 300 9.4.5 Case-2: Exfoliated, 20% bio-resin, 20% available for layout optimization ........... 301 9.4.6 Case-3: Exfoliated, 20% bio-resin, 10% available for layout optimization ........... 301 9.4.7 Case-4: Exfoliated, 10% bio-resin, Initial Value Varies Linearly 302 xi r .3 C3555: lntertalatt‘fl 4.9 CasefizlntercahIc’C :4" Caseklmercrla l C15e~8:lmert‘al.1‘. $.11: Damion of the 15. 11.2% simulations: " Cortision.................. . thirtiedgernents..- ?i Telesmdfigures..-” " Ferrite 3.1" LETTER 13 \ 1311M AM) (I).\ CU dlmntw ...... ... Reseath Findings ........ 1.;1 Processing of Bio 1:12 Sends on Biobasr 131.3 li-brid Bio based “.14 Amhtital and Co 3315 litterial lasouL. “3.317 Closure ....... ._ Pfe'SOphLa] CODClusi 21515119124611» .WhlSCB-le Comri lfllegrafion of E; 1 ”Elm?“ WC rosa use. ........ O...“ ..... 9.4.8 Case-5: Intercalated, 10% bio-resin, 10% available for layout optimization ........ .302 9.4.9 Case-6:1ntercalated, 20% bio-resin, 20% available for layout optimization ........ .303 9.4.10 Case-7:1ntercalated, 20% bio-resin, 10% available for layout optimization ....304 9.4.11 Case-8:1ntercalated, 10% bio-resin, Initial Value varies linearly 304 9.4.12 Discussion of the case-studies and Idealized RVEs 305 9.5 Multi-FE simulations: Tensile behavior using idealized RVEs 308 9.6 Conclusion 308 9.7 Acknowledgements 310 9.8 Tables and Figures 311 9.9 References 324 CHAPTER 10. SUMMARY AND CONCLUSIONS 326 10.1 Summary: 326 10.2 Research Findings 327 10.2.1 Processing of Bio-based Nanocomposites 327 10.2.2 Study on Bio-based Polymer Nanocomposites 328 10.2.3 Hybrid Bio-based Composites 330 10.2.4 Analytical and Computational Modeling 331 10.2.5 Material layout 332 10.2.6 Multi-scaleyo simulations / Multi-FEA 334 10.2.7 Closure 336 10.3 Philosophical Conclusions 337 103.1 Sustainability 337 10.3.2 Multi-scale Computations 338 10.3.3 Integration of Experimental results and Computational simulations ................ 338 10.3.4 Electron Microsc0py / Measurements at lower Scale 339 10.4 Research Needs 339 10.4.1 Atomistic Simulations, Transient and Non-linear Properties: 339 10.4.2 Statistical Considerations in Experimental Results 340 10.4.3 True Bio-degradable “Green” Materials 340 10.4.4 True Integration of Experiments and Computational Simulations ................... 341 10.4.5 Structural Application of Biocomposites and Large-scale Testing .................... 341 10.5 Research Impact 342 Ti'eI-l. Gammon of prot‘ '75:.- -. r'irritnental result: _;,_, Entrimental results Tie 31. Experimental matri {bani ENS COUICIIIS ..... 3:5: 4-1. Experimental matri. :layani ENS contents :1 Xi. BDCOmPOSKC mic“ ;te€>1.Bi0tomposite materi ‘3‘.“ I 1 a I ..es. Measured properties 1': 21.13am along diffs: M.“ ”...II“..... Ii; ’3‘) ' - " ~-lmponam issues limi LI '3. Ek‘mmm Min.) m m Elli Contents 5(brnparison of dif. {UPE eXpressed m perm egalorbem Ybanier PM W :6 Compamon of e) @1970“th faCIo 3.1} 0n florsfat . maIOVErh PfaCtors ; 39911 lDesc nption + LIST OF TABLES Table 2-1. Comparison of processing techniques. 51 Table 2-2. Experimental results for various processes at sonication energy of 60 k] .............. 52 Table 2-3. Experimental results for various processes at sonication energy of 300 k] ............ 52 Table 3-1. Experimental matrix showing specimen identification numbers and variation in clay and EMS contents 82 Table 4-1. Experimental matrix showing specimen identification numbers and variation in clay and EMS contents 115 Table 5-1. Biocomposite material identification and composition. 141 Table 6-1. Biocomposite material properties, Composition and Identification 172 Table 6-2. Measured properties of biocomposite material systems 173 Table 7-1. Modulus along different filler orientations (fibers and platelets) - Mori - Tang: Table 7.2. Important issues limitingabilityto model the nanocomposites [7.3] .................. .218 Table 7-3. Experimental rmtrix showing specimen identification numbers and variation in clay and EML contents 219 Table 7-4. Diffusivity coefficients of UPE/EML nanoclay composites. 4719 Table 7-5. Comparison of diffusivity coefficients of various nanocomposites with virgin UPE, expressed in percentage. Highlighted region shows nanocomposites that had equal or better barrier properties than virgin UPE. 219 Table 7-6. Comparison of experimental results and theoretical predictions for various particle overlap factors, at nanoclay content of 2.5 wt.% 720 Table 7-7. Comparison of experimental results and theoretical predictions for various particle overlap factors, at nanoclay content of 5 wt.% 220 Table 9- 1. Model Description for the material design layouts studied in this work .............. 311 Images i .- "-11 lSchemati: shorting 3121-1 Computation mate 10.1 51.43; ................ rrzs Schemati: of m1 r: 14. Rierarthical Modeii z51': 1-5.0.'erall approach of 7;:52-2. Schemtic deseriptic 2-1 tears modulus an: 13?- 3-3. Tensile modulm and 3’33 lilenfl failure straim 2.51mi. in... m... re: 36. E11 mic mg... plus Dill {353d ll 63 brig Mp Li DRESS C wnh 6: P13: 3 3‘3: 2-7, 13th LIST OF FIGURES Irmges in this dissertation are presented in color. Figure 1-1. Schermtic showing multiscale components and phase transitions 21 Figure 1-2. Computation materials - Length, time scales, modeling methods and associated tools [1-40]. 21 Figure 1-3. Schematic of multiscale FE approach (Adapted from [1-101) 22 Figure 1-4. Hierarchical Modeling and Evaluation Techniques 22 Figure 1-5. Overall approach of multiscale modeling for load bearing biocomposites ........... 23 Figure 2-1. Schematic description of various processing techniques. 53 Figure 2-2. Tensile modulus and strength at sonication energy of 60 k]. 53 Figure 2—3. Tensile modulus and strength at sonication energy of 300 k] 54 Figure 2-4. Tensile failure strains at sonication energy of 60 k] 54 Figure 2-5. Tensile failure strains at sonication energy of 300 k] 55 Figure 2-6. TEM micrographs showing clay dispersion and exfoliation for nanocomposites processed at 60 k], a) process B, b) process C with sonication energy applied in 3 steps, c) process C with 60 k] applied continuously. Encircled regions reveal agglomerated stack of clay particles . 56 Figure 2-7. TEM micrographs showing clay dispersion and exfoliation for nanocomposites processed at 300 k], a) process B, b) process D, c) High magnification of an intercalated particle from process D nanocomposite. 5.7 Figure 2-8. SEM innges showing tensile fracture surface morphologies for the nanocomposites from various processes: a) neat UPE, b) process B at 60 k], c) process C with sonication energy applied in 3 steps, 60 k], d) process C at 60 k] (continuous), e) process B at 300 k], and, f) process D at 300 k] 58 Figure 3-1. Processing technique of nanoclay reinforced EMS-UPE blends 83 Figure 3-2. Experimental tensile modulus of bio-blend resin systems with varying clay contents 83 xiv yr: 3333.33.1an “ 1.35 d Tensile strains a‘ .33: 35. Ultimate 1805115 Fr: 36. Literate tensile 1 it? 37. Storage Mo X233:omites........... " a 2:: >8. linear coefficient “r: 39. ibismre absoq rests: absorption aft: 73:: 31:. Moisture absorp is}. ill. Absorbed energ mtg clay contents a... :5 ‘7 ' ' mailed and intertahtr alum'bli‘lié’lltmgni 31:“ Of intercalated pat "s‘: 313 Figure 3-3. Experimental tensile modulus and Poisson’s ratio of neat resin (no clay) systerm 84 Figure 3-4. Tensile strains at failure of bio-blend resin systems with varying clay contents...84 Figure 3-5. Ultimate tensile strengths of bio-blend resin systems with varying clay contents 85 Figure 3—6. Ultimate tensile strengths and failure strains of neat resin (no clay) systerm ....... 85 Figure 3-7. Storage Modulus and Glass Transition Temperature of Bio-blend Nanocomposites 86 Figure 3-8. Linear coefficient of thermal expansion above and below 7; 86 Figure 3—9. Moisture absorption of neat resins (no clay). The inset plot compares the moisture absorption after steadystate has been achieved 87 Figure 3-10. Moisture absorption of bio-blend resin systems with varying clay contents ...... 87 Figure 3-11. Absorbed energy per unit volume (toughness) of bio-blend resin systems with varying clay contents 88 Figure 3-12. Bright-field TEM micrographs revealing homogenous dispersion with partially exfoliated and intercalated clay particles in UPE matrix. a) Low magnification, scale bar - 1 tun, b) High magnification, scale bar — 50 nm, approximately 3 to 4 particles per gallery of intercalated particle. 88 Figure 3-13. SEM micrographs of tensile failure surfaces, a) Neat UPE without inclusions, scale bar - 50 um, b) 10% bio-blend [EMS] in UPE with 1.5 wt.%. clay inclusions, scale bar - 50 um. 39 Figure 4-1. Processing of nanoclay reinforced bio-based (UPE/EML) resins. ...................... 116 Figure 4—2. Experimental tensile modulus of bio-based polymer systerm with varying nanoclay content. 116 Figure 4—3. Experimental ultimate tensile stresses for various bio-based polymer/ clay nanocomposites 117 Figure 4—4. Experimental tensile test elongations at failure for various bio-based polymer/ clay composites. 117 Figure 4-5. Variation of CIE below 7; with varying bio-resin (EML) and nanoclay content. 118 Figure 4-6. Variation of CIE above 7; with varying bio-resin (EML) and nanoclay content. 118 . . l :53: 47.132.106.10 0f $153 is: #8. Blair-.11? diffusivr 352: +9. Experimental 3::on in Studj '2': +13. Variation of 120 concenmion ---..-....... 55:5 411. TEM mitmgr iii/CHE» sell disper i}.’2."32.”25 - well dept little = 131ml, c) 133 C.- =firl, and d) ICC/C/E fraill'ensile modulus an 13:1: ' l i ' i ' >3. Lnear coefftcrent i W“... Douro-c... . 53:531Bright-field TEM I itcrated and intercalated 411111 b) high 3 , . c'rtrttc gtleryof mtercalariciiibparti ”3:5 35. SEM mic Err and Hating Figure 4—7. Variation of glass transition temperature (7;) with varying EML and nanoclay content. 119 Figure 4—8. Moisture diffusivity of neat resins (no clay). 119 Figure 4—9. Experimental diffusivity coefficients for all bio-based polymer/ clay nanocomposites in study. 120 Figure 4—10. Variation of Izod Impact strength with varying bio-resin (EML) and nanoclay concentration. 120 Figure 4-11. TEM micrographs showing degree of dispersion and morphology: a) 100/ 0/ 2.5- well dispersed, partially exfoliated and intercalated (scale - 1pm), b)70/30/2.5 - well dispersed, but higher degree of intercalation relative to 100/ 0/ 2.5 (scale - 1pm), c) 100/ 0/ 5 - Well dispersed, partially exfoliated and intercalated (scale gngum), and d) 100/ 0/ 5.0 - high magnification of an intercalated gallery (scale - 50 121 Figure 5-1. Tensile modulus and ultimate tensile strengths. 142 Figure 5-2. Impact strengths from notched Izod tests and tensile strains at failure. ............. 142 Figure 5-3. Linear coefficient of therrml expansion (CTE) and moisture absorption (MA). 143 Figure 5-4. Bright-field TEM micrographs revealing homogenous dispersion with partially exfoliated and intercalated clay particles in UPE matrix. a) low magnification, scale bar - 1 pm, b) high rmgnification, scale bar - 50 run, approximately 3 to 4 particles per gallery of intercalated particle 143 Figure 5-5. SEM micrographs of tensile fracture surfaces showing interfacial gaps between fiber and rmtrix. (a) biocomposite B: 10 % EMS and no nanoclay in UPE, (b) biocomposite C: 20 % EMS and no nanoclay in UPE, (c) biocomposite E: 10 % EMS and 1.5 wt. % nanoclay in UPE, and (d) representative fracture surface showing both pullout (circled region) and fracture (boxed region) of fibers, scale bar - 100 pm. Images (a) to (c): scale bar =10 pm. 144 Figure 5-6. SEM micrographs showing matrix region in tensile fracture surface of biocomposites. Micrographs (a), (b), (c) and (1 represent biocomposites A, B, D, E, respectively. All images have a magnification scale bar of 10 pm. 145 Figure 6-1. Schematic of processing technique of nanoclay reinforced UPE + bio-resin blends 174 Figure 6-2. Manufacturing of nanoclay reinforced biocomposites 174 Egg: :3. manual tens 7.7;: 7:4. Experimental ullir 3:: {+5. qurinrnul elon 3;: :5. Input strengths 0 F535 $7. lbisture diffuisivi: is: ~S. Suman of diffusi £15 69. EM microom Womposizc resin in 31th:? exfoliated and in? ligh Wagfification ima WW: I (no, 3/5. 35:55:15 Llensile fracture S mmg fiber- pullout (er .5) b) timed regior rgiom and fiber matrix an: anal gap: c) Bioco: Buomposirel (11V 3/ 3. iv = 13 uni)................... TI? ill llhm mglOnS Of :flJ'ansne D (73/331,, "r" v 3; l3) phaSC' 'SepamIeC BWomposiu- I (73/ ‘P ‘3. l-Ph}sical ‘ {:me RSV ....”3] repmSCIl ........... ‘7.) 7 - . . . .larutlon of [Ongrtt Harmon 1:: 1mm I \N“... 0...... .... ...-.... "31.4 ;m'l3m1lon of I‘Ongli; Figure 6-3. Experimental tensile moduhrs of biocomposite. 175 Figure 6-4. Experimental ultirmte tensile strengths of biocomposites 175 Figure 6-5. Experimental elongations at tensile failure of biocomposites 176 Figure 6-6. Impact strengths of biocomposites from notched Izod tests 176 Figure 6-7. Moisture diffuisivity of neat resin (no nanoclay) biocomposites. ......................... 177 Figure 6-8. Summary of diffusion coefficients of various biocomposites. 177 Figure 6-9. TEM micrographs showing nanoclay dispersion and morphology. a) Nanocomposite resin in Biocomposite E (100/ 0/ 2.5) showing good dispersion with partially exfoliated and intercalated morphology m a resin system (scale - 1pm), and b) magnification irmge of an intercalated gallery in nanocomposite resin of Brocomposrte I (100/ 0/ 5 .0) (scale- 50 um) 178 Figure 6-10. Tensile fracture surface analysis: a) Generic low magnification failure surface showing fiber-pull-out (encircled) and fiber fracture (arrows) regions, (scale bar - 200 um), b) Magnified regions showing fiber-pullout (encircled), fiber fracture (boxed) regions and fiber-matrix interfacial gap (arrows), (scale bar - 50 um). Fiber-matrix interfacial gap: c) Biocomposite A (100/ 0/ 0), d) Biocomposite D (70/30/0), e) Biocomposite I (100/ 0/ 5.0), and f) Biocomposite L (70/ 30/ 5.0). (Innges c to e, Scale bar - 10 pm) 179 Figure 6-11. Matrix regions of Tensile fracture surfaces. a) Biocomposite A (100/ 0/ 0) b) Biocomposite D (70/30/0), well blended UPE/EML region, c) Biocomposite D (70/30/0), phase-separated, bio-resin enriched region, (I) Biocomposite I (100/ 0/ 5 .0), and f) Biocomposite I (70/ 30/ 5.0) 180 Figure 7-1. Physical representations, coordinate systems, Mori—Tanalra model - filler orientations [7-3] 771 Figure 7-2 Variation of Longitudinal Modulus with Aspect ratio. 771 Figure 7-3. Variation in length and subsequently aspect ratio, across a dish-like platelet [7-3]. 772 Figure 7-4. Variation of Longitudinal Modulus (Eu) with variation in j; and (L/gl [7-7]......222 Figure 7-5. Variation of Transverse Moduhrs (E22) with variation in f; and (L/g) [7-7] ........ .223 Figure 7-6. Variation of In-Plane Shear Modulus (G12) with variation in f, and (L/gl [7-7]..223 Figure 7-7. Variation of Out of—Plane Shear Modulus (G2,) with variation in j; and (L/tl [7-7] 724 F5318. Variation of Pane Mum-o...- Ft': '49. Variation of Major Fri-1:, mason of Ion: incrifn inclusions. (U 75:11. Variation of Trar Enlist inclusions. (U 2:37.11 Variation of Long Pals: his: inclusions. (U "7" ”-13. Variation of Tran F25: his: inclusions, (U 75;: 7-14. Variation of F... nxisl W] Mama...“ $27-15. Variation of E1 1/ E 2:37-16. Kinematic Periodi. 7§:.~17.Scl16matic Represe T1: 7:18. Particle Periodic inch R“: dlmemion «- , 1;: a w R: “lung Stress co F5317} 727m .5.) ,‘ 5 Ll‘ ‘3 uh- Figure 7-8. Variation of Plane Strain Bulk Modulus (K23) with variation in f, and (L/tl [7-7]. 774 Figure 7-9. Variation of Major Poisson’s Ratio (v12) with variation in f, and (L/;) [7-7] ...... .225 Figure 7-10. Variation of Longitudinal Modulus (E ,1) with variation in f, and (L/;)- Hui-Shia FiberLike inclusions, (Us - 10, 50) [7-7]. m Figure 7-11. Variation of Transverse Modulus (E ,2) with variation in f; and (L/tl- Hui-Shia FiberLike inclusions, (U t - 10, 50) [7-7]. 226 Figure 7-12. Variation of Longitudinal Modulus (E ,1) with variation in j; and (L/tl- Hui-Shia Flake Like inclusions, (Us - 0.01, 0.1) [7-7]. 276 Figure 7-13. Variation of Transverse Modulus (E 22) with variation in 1; and (L/;)- I-Iui-Shia Flake Like inclusions, (U t - 0.01, 0.1) [7-7] 977 Figure 7-14. Variation of En with respect to particle vohrrne fraction( j; ) - Halpin-Tsai model [7-7] 777 Figure 7-15. Variation of Ell/Em. Halpin -Tsai model [7- 1] 778 Figure 7-16. Kinermtic Periodic Boundary Conditions 778 Figure 7-17. Schematic Representation of Particle Periodicity in RVE 779 Figure 7-18. Particle Periodicity. Circled regions show the inclusions of the truncated particles. RVE dimension of 500 x1000 nm 729 Figure 7- 19. a) Sample RVE with 2% wt. clay concentration, and, b) equivalent FE model 730 Figure 7-20. Resulting stress contours (Sll-Von-Mises) from RVE FE analyses ................. 231 Figure 7-21. Partly exfoliated and partly intercalated RVE model 7‘51 Figure 7-22. RVE with Mesh thickness - 1 element per clay particle (RED) 2'52 Figure 7-23. RVE with interface, Mesh thickness =- 1 element per clay Particle (RED) and Interface (green) shown above 732 Figure 7-24. RVE with Interface, Mesh thickness - 4 elements per Clay Particle and Interface (md) 732 Figure 7—25. Comparison of RVE Analyses with theory, Variation of Longitudinal Stiffness with varying EP/Em 23 3 - ' I is“ ill). Common Of R :71 1 Sun's oll Interface] “232128. Suitiof Cl'E thro .1139 .Cornparison of T: ms for 133% Matt Ll . Ditlusion modeli: ins son of water frorr 3&3“:me c) tortuou.< zm'Erpennrntal results Ft533131 Dfimisity plots ( triennial data and Sr. mitts c0: ffrients (D 3;"? ”'31 DIllUSMIy plots toms, Slml‘OlS indica‘ 222.733 mfinm plc EQF‘ERL with \‘mmg mkcxponential fit. If“? #74. Dlfimivm, pk '. 2'.me “11h vm'lrfig ”BUB exponential ill: Figure 7-26. Comparison of RVE Analyses with theory, Variation of Longitudinal Stiffness with varying aspect ratios 7‘53 Figure 7-27. Study of Interface properties through RVE Analyses 7‘54 Figure 7-28. Study of GTE through RVE analyses 734 Figure 7-29. Comparison of Tensile Modulus from theory, FEA - UCM and Experimental Results for 100% Neat UPE with varying clay content 7‘55 Figure 7-30. Diffusion modeling using FE- based RVEs. a) concentration contours showing diffusion of water from bottom to top, b) tortuous flow path simulation, low magnification, c) tortuous flow path simulation, high rmgnification, and d) Simulation and Experimental results comparison. 736 Figure 7-31. Drffusrvrty plots (M t / M 00 vs.J;/d) of neat resins (no clay). Symbols indicate experimental data and solid lines the exponential fit. Inset shows, the initial slope / diffusivity coefficients (D x 10'12 mz/s) as a function of bio-resin content ”7 Figure 7.32. Diffusivity plots (M, /Mco vs.Ji/d) of UPE (no EML) with varying clay contents. Symbols indicate experimental data and solid lines the exponential fit. ....... .237 Figure 7.33. Diffusivity plots (M, /Mw vs.J?/d) of UPE/EML blend containing 10%EML, with varying clay contents. Symbols indicate experimental data and solid lines the exponential fit. 7‘58 Figure 7-34. Diffusivity plots (Mt/MGo wag/d) of UPE/EML blend containing 20%EML, with varying clay contents. Symbols indicate experimental data and solid lines the exponential fit. 738 Figure 7-35. Diffusivity plots (M t / M 00 vs. J; / d) of UPE/EML blend containing 30% EML, with varying clay contents. Symbols indicate experimental data and solid lines the exponential fit. 7‘59 Figure 7-36. Summary of diffusivity coefficients of all bio-based nanoclay composites in this 239 Figure 7-37. Comparison of experimental results and theoretical predictions for various particle overlap factors, conesponding to clay content of 2 .5 wt. % 740 Figure 7—38. Comparison of experimental results and theoretical predictions for various particle overlap factors, corresponding to clay content of 5 wt. %. 740 Figure 8-1. A schematic of a typical 2D periodic RVE used in the current multiscale approach. Adapted from [8-15] 1 775 a.” 11 Sci-erratic of Iter: 33:: iv}. Funthan of mu: if aeration porm, iiplacexrnts {1‘}. Mac are armed to nucroscr 3;:r $4.1) A FIE-RVE sh m 53% exfoliate. :1?th shoving int; ff based RVE model..... 35:: $5. The: phase RH: 11% nanoclay (black), ( nghciogyRVE c) ml is: 3—6. lime phase RVE 32% nanoclay (black), ( trenchgrRVE c) mulr rial/..Enerirrental temi mm in to the experi Figure 8—2. Schenntic of Iterative Computational Procedure. Adapted from [8-19] ............. 275 Figure 8-3. Flowchart of multi-level FE approach. Deformation gradient at any macroscopic FE integration point, chm is transferred to microFEA models as vertex .ip displacements {ui}. Macroscopic stresses 03"“,on , and tangential stiffness 4Smacro , are returned to rmcroscoPic integration points from microscopic nrodels. .................. 276 Figure 8-4. a) A FE-RVE showing 100% exfoliated nanoclay morphology, b) A FE-RVE showing 50% exfoliated and 50% intercalated nanoclay particles, c) actual TEM micrograph showing intercalated gallery, and d) zoom in of the intercalated gallery in FE based RVE model 777 Figure 8-5. Three phase RVEs for materials with 10% EML (red) in UPE (green) with 2.5 wt.% nanoclay (black), (a) Single platelet idealized RVE, b) multi-particle, exfoliated morphology RVE c) multi—particle, intercalated morphology RVE 778 Figure 8—6. Three phase RVEs for rmterials with 20% EML (red) in UPE (green) with 2.5 wt.% nanoclay (black), (a) Single platelet idealized RVE, b) multi-particle, exfoliated morphology RVE c) multi-particle, intercalated morphology RVE 779 Figure 8-7. Experimental tensile modulus for neat bioblends (no clay). The dotted line is a sigrnoidal fit to the experimental data. 780 Figure 8-8. Schematic of multi—FE model used for fiber-reinforced hybrid composites......280 Figure 8-9. Parameters considered to corrrpare simulations and average experimental response ' 781 Figure 8—10. Tensile stress-strain response of two-phase RVEs: a) Neat (no clay) resins, b) Nanocomposite with 2.5 wt.% clay. ' 782 Figure 8-11. Comparison of tensile response from multi-FE simulations and experiments for biocomposites with 10% EML and 2.5wt.% nanoclay in UPE 7R3 Figure 8-12 Comparison of tensile response from multi-FE simulations and experiments for biocomposites with 20% EML and 2.5 wt.% nanoclay in UPE 783 Figure 8-13. Comparison of tensile response from multi-FE simulations using two-phase and three-phase RVEs. All rrrodels have intercalated morphologies 784 Figure 8-14. Comparison of tensile response from multi-FE simulations and experiments for virgin UPE (0% bio-resin and 0% nanoclay) biocomposites 784 Figure 8-15. Comparison of tensile response from multi-FE simulations and experiments for hybrid composites with 20% EML and 2.5 wt.% nanoclay 985 XX 1.": Q1 835: Cells: 3) EXlOlJ i win raresents the. rum: cirzrém blends of bro- res fitting dime of obiea 32.12422, b) matenal L1} sap S13: _u.J l 3::29-lfase2, Erioliared m m; change of due: cem' ,b) material hj raw ...r :‘TI' 94. Casel, Exfoliated male for material desigr limo”: W of de eaten 5151333, and c) ma “$2295. CasH, EXfOliarc d [T 52013;; Change Of Objec iflmimts’ bl D15- Lnburior 3330!] Step 53’ and C) m mi and Como, b) ”3”“ Step 230. ...; .......... Figure 9-1. Base Cells: a) Erdoliated Morphology, b) Intercalated Morphology. The white region represents the nanoclay and the blue region represents the design domain containing blends of bio-resin and petro-resin “512 Figure 9-2. Case-1, Exfoliated morphology with 10% bio-resin. a) Evolution of design, plots showing change of objective function, magnitude of design variable vector, and constraints, b) material layout at iteration step 50, and c) material layout at iteration step zoo 1113 Figure 9-3. Case-2, Exfoliated morphology with 20% bio-resin. a) Evolution of design, plots showing change of objective function, magnitude of design variable vector, and constraints, b) material layout at iteration step 50, and c) material layout at iteration step 200 314 Figure 9-4. Case-3, Exfoliated morphology with 20% bio-resin content and only 10% available for material design . a) Evolution of design, plots showing change of objective function, magnitude of design variable vector, and constraints, b) rmterial layout at iteration step 50, and c) material layout at iteration step 200. 315 Figure 9-5. Case-4, Exfoliated morphologywith 10% bio-resin. a) Evolution of design, plots showing change of objective function, magnitude of design variable vector, and constraints, b) Distribution of initial value of the design variable, c) material layout at iteration step 50, and c) material layout at iteration step 200. 316 Figure 9-6. Case-5, Intercalated morphology with 10% bio-resin content a) Evolution of design, plots showing change of objective function, rmgnitude of design variable vector, and constraints, b) material layout at iteration step 50, and c) material layout at iteration step 200. 317 Figure 9-7. Case-6, Intercalated morphology with 20% bio—resin content and all 20% available for material design. a) Evolution of design, plots showing change of objective function, rmgnitude of design variable vector, and constraints, b) material layout at iteration step 50, and c) material layout at iteration step 200. 318 Figure 9-8. Case-7, Intercalated morphology with 20% bio-resin content and only 10% available for material design . a) Evolution of design, plots showing change of objective function, nngnitude of design variable vector, and constraints, b) material layout at iteration step 50, and c) material layout at iteration step 200. 319 Figure 9-9. Case-8, Intercalated morphology with 10% bio-resin. a) Evolution of design, plots showing change of objective function, magnitude of design variable vector, and constraints, b) Distribution of initial value of the design variable, c) material layout at iteration step 50, and c) material layout at iteration step 200. 320 Figure 9-10. Simplified, idealized RVEs. The green, red and black regions represent petro- resin (UPE), bio-resin (EML) and nanoclay respectively. a) 10% EML, b) 20% EML322 53329-11. (amnion of it: implies WES sirh exp Figure 9-11. Comparison of tensile response from multi-FE simulations using idealized three-phase RVEs with experimental data, (a) 90/ 10/2.5, and (b) 80/ 20/ 2.5 .............. 323 (leprrl Inflation ” .l/fii’woi’m .i Emiromflulll' friendi 3+5 warms the S?” m scale) WWW” aim Biobascd come It. tri'nmrrnul Appfll mi wage of mrlriscale RN03 tr: :ombhed with bioba‘ 11:1" and barrier propertie losses. lbreover, these mutual impart While 1} it Md for synthetic con mammma Fem focused on (level (‘7 Jr 53.10 Wdersund and ex; Chapter 1. Introduction 1! 1!! otiwtz'wz Environmentally friendly bio-based composites with improved properties can be obtained by harnessing the synergy offered by hybrid constituents such as multiscale (nano- and micro- scale) reinforcement in bio-based resins composed of blends of synthetic and natural resins. Bio-based composites have recently gained much attention due to their low cost, environmental appeal and their potential to compete with synthetic composites. The advantage of multiscale reinforcement is that it offers synergy at various length scales, and when combined with bio-based resins provide stiffness-toughness balance, improved thermal and barrier properties, and increased environmental appeal to the resulting composites. Moreover, these hybrid materials are tailorable in performance and in environmental impact. While the use of different concepts of nrultiscale reinforcement has been studied for synthetic composites, the study of multiphase/multiscale reinforcements for developing new types of sustainable rmterials is limited. The research sumrmrized in this dissertation focused on developrrrent of multiscale reinforced bio-based composites and the effort to understand and exploit the synergy of its constituents through experimental characterization and computational simulations. Environmental concerns related to the use of synthetic, or petroleum-based, polymer matrix composites has propelled the development of composite materials based on natural or renewable sources [1-111-2]. Biocomposites, composed of natural fibers in synthetic or natural polymer matrices have recently gained much attention due to their low cost, mflmental friendliness, and their potential to compete with synthetic composites [1-311- 4] Nonetheless, the use of bio-based composites has been limited due to their lower M and rhenmphysi :“Tmal smrural materiai recess and perfomunce ar fart-imbued resin with Earl resin improve toughne 1:215: in roughness compro Sins and roughness are c Tiilforan efficient compc iizaormbfl The blOCOmPOSiIfi use 1"tunes in different p; 33551 into three length sea r235 + Why lnClusiom) in ii unmade (51mm mechanical and thermo-physical properties compared to synthetic composites and conventional structural materials [1-5]. A promising compromise between environmental friendliness and performance are bio-based resins, or bio-blends, obtained by replacing part of a petroleum-based resin with natural bio-resin. In addition to higher natural content, bio- based resins inrprove toughness of the resulting resin blend [1-6]-[1-8]. However, this increase in toughness compromises stiffness, barrier and thermal properties [1-6}[1-8]. Stiffness and toughness are opposing performance parameters and a proper balance is required for an efficient composite. One way to attain this balance is the addition of layered silicates, or nanoclays. The biocomposite used in this study is a hybrid heterogeneous material with heterogeneities in different relative dimension scales containing four major components: petro- resin, bio resin, natural fibers, and nano-clay inclusions. 'Ihis biocomposite can be classified into three length scales (see Figure 1-1), namely: a) nanoscale (resin system: resin blends + nanoclay inclusions), b) microscale (resin system + natural fiber reinforcement), and, c) rnacroscale (structural component). The effective properties depend on the size, shape, properties and spatial distribution of the heterogeneities (in our case, nanoclay, short fibersandbio-resin). m ' in f 'w' ' inf il k 'Alr‘m- («114' we ‘13 , m II‘ C uma-o 0f V0 _u_-.‘ f H - d ILL‘ W- Hence, the work followed the “materials by design” philosophy by incorporating an integrated experimental and computational approach to strategically explore the design possibilities and limits. Figure 1-1 provides a schermtic of the overall rrrultiscale methodology carried out in this ‘Vork Apart from experimental characterization at nano and micro-scales, the work focused on developing computational (finite elerrrent) models at each scale. The models were raised ash in: they mm an of material behavior were scales and the res main and scanning elec were scales. Representatir :1“: at each scale. The res mam models and e: 1'23 and Wmnfll rtsu EL: mm inaCfi‘tm'rcie mammal 1001 to ohm 13;, and verify the mOdel initial.“ POSSible. lli-e biobased resin 1 maid M9330 and b flaunt; lhe dlSmb . uti TUStOPY, bUt 5U Earned In order to model lhf Lh-I'CC- phASc [has “:43 t *Mfl H 3.131 developed such that they resemble the microstructure at their respective scales. This enables simulation of material behavior close to reality and understanding the phenomena occurring at respective scales and the resulting material behavior. Electron microscopy tools such as transmission and scanning electron microscopy were used to model clay and fibers at their respective scales. Representative volume elements (RVEs) were modeled or obtained from literature at each scale. The results of the computational models at each scale were compared with theoretical models and experimental results. Discrepancies between the computational modek and experimental results may occur due to various assumptions rmde in the model and experimental inaccuracies. Nonetheless, in order to use the respective models as a computational tool to obtain the next hierarchical level properties, it becomes essential to validate and verify the model with experiments, such that accurate transfer of information to the next level is possible. The bio-based resin system used in this study has blends of petro—polymer (UPE, unsaturated polyester) and bio-polymer (derivatives of soybean and linseed oil) reinforced with nanoclay. The distribution of nanoclay in the polymer matrix could be observed using electron microscopy, but similar distribution of bio-resin in primary petro-resin cannot be observed. In order to model the hybrid composite to great detail, it is essent'nl to realistically model the three-phase material, and hence the distribution of bio-resin was sought. A material Layout problem was designed and implemented in this work. This was aimed to develop an enhanced three phase RVE by using experimental data along with topology Optimization in an attempt to provide insight to the distribution of bio-resin in the RVE, and not to accurately solve the rmterial design problem. The resulting material layouts were simplified into idealized models and used in multiscale sirrrulations. In this wont, a direct micro—1133C“, finite element based approach, also referred as multi—level FE approach was k 3:3 [1311.13 m ”M n-‘amSlItS “5 SImUhICd rel-M H' llr WWW pm I fired mommies x am: and ham“ pmpen ...th *‘lf Properties “ I mamms Will that t 3172533331 modek to P” mm and any macro malt namely a multiFE l Impzlate the experiment; 2323321115 and to hopefuli file}. an enhanced R\ “is a topolOgy optlIDingiti .; l J. 3&45 «m, theme allouine 323 . M11, ccofriendlv "@‘l’il- g. TR Surcessqu-I ~65: mm“ ”Able biobaseci \r" iron and hOUSin QII-«aih of undemdm g C W m Th” “(”12 holc it: in used [1-9Il-10]. The tensile response of hybrid biocomposites and bio-based nanocorrrposites was simulated using multi-FE scheme and compared with experimental data. The experimental part of this research resulted in novel processing techniques of bio-based nanocorrrposites and comprehensive characterization of thermo-physical, mechanical and barrier properties of bio-based clay nanocomposites and biocomposites. Additionally, optimal material concentrations that provided ease of processing with good balance in multiple properties were obtained. The objective was to perform a limited amount of experiments such that the experimental information can be used to develop computational models to predict the behavior of materials for any combination of constituents and any macroscale structural component. A multiscale computational approach, namely a multi-FE based approach, was implemented to the developed materials to extrapolate the experimental matrix, to provide insight into nano-scale behavior beyond measurements and to hopefully serve as a tool for computational design of hybrid materials. Additionally, an enhanced RVE for modeling the three-phase material was determined by solving a topology optimization based rmterial layout problem to determine the distribution of bio-resin, thereby allowing modeling the nanostructure in greater detail and closer to reality. Overall, eco-friendly, tailorable, cost-effective and multiscale reinforced bio-based composites were successfully developed. The improved multifaceted features possible for these sustainable bio-based materials are likely to increase their appeal for use in transportation and housing structural applications. Additionally, it is believed that the aPPanch of understanding complex materials by integrating simulations and experiments, as WMd in this work, holds great promise, and a similar methodology can be applied for other types of hierarchical materials, thereby providing guidance in designing those rmterials. if? 02.31.35,?! lh: goal of this work w “Timing an integrated exp: 3335.3 mssihihies and hm." I I I 'I W; $21330 [hf mlmunl ‘ a" :r-hcds 10 manufacturr 111' R2 remixes of the work C. Mime T0 crime, cmironment f ' To use 5mm, 'of mate hashahaced or sum, .6 Medals and multiscai behased materials. ' To develop and use It thy nanocomposites “Wis for pmduci: I TO W}: Cl” cr " . W the ML. 12 Oé/E’ctz'w: The goal of this work was to develop eco-friendly, hybrid bio-based composites by incorporating an integrated experimental and computational approach to strategically explore the design possibilities and limits, such that the synergy offered by these materials could be exploited to the maximum. Moreover, as a part of the study efficient processing techniques and methods to manufacture nanocomposites and biocorrrposites were investigated. Experimental data will provide input to validate models and the development of RVEs. The overall objectives of the work can be smnmarized as follows: I gobal objective: To obtain / use a rational approach in development of cost- effective, environment friendly, multifunctional hybrid hierarchical biocomposites. ' To use saggy of materials at various Length scales, such that the resulting composite has bflefi gr smrigr mommies than the ELISE. ' base material. The synergy of rmterials and multiscale reinforcement provides multifrmctionality to the resulting bio-based materials. ' To Qvelgp and use relame' b: efficient pflessing techniques to synthesize bio-based clay nanocomposites and bioconrposites. Also, to study the feasibility of these techniques for producing large structural components ' To perform by emrimeng at each scale. The experiments will not only help in mm the resulting material properties but will provigk the 1m ' / bounds in processing and effects of constituents on resulting parameter(s). W ' nwillro'de' informatinfr vel' tinal lsI ' To Maud use v_alr;datec_l computational models at each scale through key experiments and to eliminate the cosrlytrial and error approach. 5 #1 253 x-mteh' transfer? W mghiks‘tl finite elemcr J £50701;er Er 3.1 Bio based compos ite Classical fiber reinf or raga: apphcauons, but ha I: “3365 at low weights : its. nth is carbon, or glzu risers. However, cmiron Isiah on nomenew. in: RP composites : 2417 Sher COmPOSiIes of b ' To Wham, such that irrforrmfion is mm W through computational models at various length scales. A multi—level finite element technique has been used. 173 flac/égmxmd 1.3.1 Bio based composite unterials and Biocomposites Classical fiber reinforced polymer (FRP) composites were initially targeted for aerospace applications, but have transitioned into numerous other fields where high strength and stiffness at low weights are required. In general, high strength and modulus synthetic fibers, such as carbon, or glass, are mixed with petroleum based resins, such as epoxies and polyesters. However, environmental concerns, such as biodegradability, recycling issues and dependability on non-renewable petroleum reserves, have limited the wide spread use of synthetic FRP composites and have propelled the development of alternatives such as natural fiber composites or biocomposites [1- 111-211-511-11H1-13] Biocomposites are comprised of mm (e.g., flax, hemp, jute, cellulose, kenaf, cotton, coir, bamboo etc.) and mm (unsaturated polyester, epoxy, poly urethane, phenolics, etc.) or W (CPOXidized vegetable oils, polyoles, rmleinated triglycerides, animal fats, etc.)[1-2}[1-4]. The use of only bio-resins has been limited due to perforrmnce concerns such as low mechanical and thermo-physical properties [1-611-711- 14]. Bio-based resins, or bio-blends, obtained by replacing/ blending part of the petroleum based resin with natural bio-resin have been developed by our group and it has been found that they can increase the toughness of the composites [1-611-7]. However, the increase in toughness due to the increase in bio-resin content is generally achieved by sacrificing M11888 [1-611-15}[1—17], banier [1-16] and thermal [1-1611-17] properties. Stiffness and k 3:355 3“ opposing perion‘.‘ zrfimbixomposite. One - reel :uaes or MDOClA}S i or is diseased earlier, ens var; Sher composites (bloc: m and automorive i: imagines for load bearing mi. nonniifonn fiber 5 31mm. Work in our grou; «.7er Delaware [1'31 I Ifillt overtome nirh efficie: Eritrean: panels was ioc Intention] materials as well a 35 mm Smeepuhiliry aimed Studies did r toughness are opposing performance parameters and a proper balance is required to develop an efficient biocomposite. One of the methods of attaining this balance with the addition of layered silicates or nanoclays for nano-reinforcement. As discussed earlier, environmental concerns have relatively increased the dermnd of natural fiber composites (biocomposites). Initial use has been in flooring, siding, roofing, fumiture and automotive interiors [1-111-311-18]. Nonetheless, consideration of biocomposites for load bearing applications has been limited due to their lower stiffness and strength, non-uniform fiber sizes, high moisture absorption and low stability at high temperature. Work in our group [1—511-1111-1211-1311—19] and similar work by researchers at university of Delaware [1—3Il-4] Ins shown that stiffness shortcomings of biocorrrposites could be overcome with efficient hybrid and cellular designs, and the performance of cellular biocomposite panels was found to be as efficient as building flooring systems from conventional nnterials as well as E-gbss/ polyester panels. Nonetheless, the issues of thermal and moisture susceptibility of biocorrrposites are still a concern. Moreover, the aforementioned studies did not include bio-resin or nanoclays in the manufacture of biocomposites. The incorporation of nanoclays and natural fibers in the resin systems PrOVide reinforcements to resin systems at two different scales. Similar multiscale reinforcernent has been studied for synthetic fibers for mechanical and electrical properties [1‘20}[1-22]. Conversely, the use of nanoclays and natural fiber reinforcement in resin blends of petroleum and bio-resins is very limited. 1.3.2 Enhancement through Nam-reinforcement 8c Nanocomposite processing MP Wee .- A 4 gm Points reinforced w; merino in nodulus, the ..ra oranxlajs consisr o: mhlarge surface areas ; sire: capacities of chi- emanation Mill 3 hOSI :5me gaps 0f lilo-Lilli!“ 1' testes. W on the lung fr‘mPchm‘ the PlALeleLs 333215 Emit“ lhf platelet ~41 exfoliated or separat: :emel W} the CkiOliang In: mpms’ it 1135 beer. “‘ ‘m 1. resume flirts in Polling“ nith xiii h Pm“ et al 2x03973155 Apart from mixes Muncfiomfiw firmmr flamttiabila 32:39,; - N«ea . 7‘»; e35 Polymers reinforced with layered silicates or nanoclays have shown to exhibit enhancements in modulus, themnl and banier properties at low concentrations. layered silicates or nanoclays consist of stacks of sheet like platelets with thickness of ~1 nm and extremely large surface areas and aspect ratios. The large surface areas and high cation exchange capacities of clay platelets enable the use of nanoclay with proper functionalizations with a host of polymers enabling the opportunity to circumvent the performance gaps of moisture and thermal dimensional instability of bio-fiber polymer composites. Depending on the functionalization of the clay and the compatibility of the chy with the host polymer, the platelets in the stacks are separated to enable the polymer chains to penetrate between the platelets. Depending on the degree of penetration of the polymer Chains, exfoliated or separated, and intercalated or stacked b) clay morphologies are obtained. While the exfoliated morphology may produce enhancerrrent in modulus and barrier properties, it has been found that intercalated composites produce better gains in tonghness for the resulting nanoconrposites. The enhancerrrent of mechanical and banier Properties in polymers with the addition of small concentrations of nanoclays is well CStablished. Le Baron et a1 [1—23] provide a good review on polymer clay hybrid nanocomposites. Apart from the increase in stiffness, the inclusion of nanoclay particles it”Induces multifunctionality to the resulting nanocomposite resin system by enhancing ban-fiel- properties, flammability and ablation resistance. As a part of this study, experimental cthterization of bio-based clay nanoconrposites was performed. Unsaturated polyester (LEE) was used as the primary petroleum based resin. Two bio-resins, namely epom'dized My! soyate (rails) and epoxidized methyl linseedate (EML) were blended with UPE. 'I‘ne r; and Ni“ 0f ‘he flicmw m PfOVlLlC’C Palmer clay nano inflation, nelt COmpOl tsiien used in the ind falter; of pohmers. 50h r: em to produce nar EISL‘Cl SBl’T generally in! ii“ 5055111 and the nanocla 5mal oi the solvent to EPW553331 it is ener teem removal Moreover ‘ details and results of the experimental characterization for EMS and EML based nanocomposites are provided in Chapter 3 and Chapter 4 respectively. Polymer clay nanocomposites have been synthesized commonly by in-situ polymerization, melt compounding and solvent based techniques. Melt compounding has been widely used in the industry but has been questioned as it raises issues of thermal degradation of polymers. Solvent based processing techniques (SBPT) have been reported in the literature to produce nanocomposites with clay platelets that are well dispersed and exfoliated SBPT generally involves sonication (mechanical stirring and ultrasonic agitation) of the solvent and the nanoclay, mixing of the polymer with the sonicated solution, followed by removal of the solvent to obtain the desired nanocomposites. One of the drawbacks of this process is that it is energy inefficient, as large amounts of energy is required in the solvent removal. Moreover solvent residues in the polymer may adversely affect the properties of the resulting nanocomposites. Also, excessive energy spent in solvent removal in bio based resins nray lead to phase separation of the resin blend and affect the properties 0f the resulting nanocomposite. Thus, efficient processing techniques are required to obtain bio based clay nanocomposites in an energyefficient and robust manner. As a part of this study, an investigation was performed to evaluate the various SOIVcnt based processing techniques for CPNCs so as to identify a time, energy and cost eff eClive approach. Polymer clay nanoconrposites were processed using SBPT with blends of Mlle-based unsaturated polyester (UPE) and bio-resin (EMS, epoxidized methyl soyate) with acetone as the solvent. The SBPT became infeasible at higher concentrations of EMS and Clay as it led to issues in solvent (acetone) removal. Four different techniques were Sl'l-«ltzziiied and the effects of energy spent on sonnication, styrene content in UPE and 9 . f . r message [5..th o I end atttmastes were 5 3" 11.11% the Process teases were found to F 33.33? nanocompogm fintianal composites PPS-tied in G‘iapwrz 3 53 “Fond Biocompc ll“! bio'bfied M its: Wm COrnb'med wh ital: biocoml’osites. . mi 1.1. As distrused r R‘r - «a $81835 pm‘ide ref processing technique on tensile, impact and barrier properties of the resulting nanocomposites were studied. Acetone (solvent) removal and related issues were overcome by modifying the processing techniques and eliminating the use of acetone. Two novel processes were found to produce balanced mechanical and barrier properties for bio-based UPE/ clay nanocomposites and hold promise for the development of bio-based multifunctional composites. A detailed explanation of the processing study and its outcomes are provided in Chapter 2 and reference [1-24]. 1.3.3 Hybrid Biocomposites from Bio-based Resins 8: Nanoclays The bio-based nanoconrposite resins synthesized and characterized as discussed above were combined with natural hemp fiber reinforcement to develop novel hybrid multiscale biocomposites. A brief introduction of generic biocorrrposites was provided in Section 1.1. As discussed earlier, the incorporation of nanoclays and natural fibers in the resin systems provide reinforcements to resin systems at two different scales. Similar Inultiscale reinforcement has been studied for synthetic fibers for mechanical and electrical properties [1-20}[1—22]. Meanwhile, the development of bioconrposites with the use of multiscale reinforcement such as nanoclays and natural fiber reinforcement in gsin blends of W is m to this work, and to the author’s knowledge such Work is not remixed elsewhere. The bio-based nanocomposite resins synthesized from L11’13/ EMS and UPE/EML combinations with nanoclay were experimentally characterized f or Various material properties such as tensile properties, impact strength, coefficient of them expansion, moisture absorption/ diffusivity, and tensile failure surface characteristics. me details and results of such characterization studies are provided in Chapter 5 and quapter 6. 10 or Modeling and Simul‘ Toe erhancemem gai ares etc.) is experimentallj scale to predict the over: gim-fl‘ersystems. It is cor genes of a pohmer-filler this of the filler, nodu than of the homogenizec :iiezrire. Various theorie mes md limitations. Os tn: Tm approaches, Variatic 1.3.4 Modeling and simulation of micro/nano reinforced polymers The enhancement gained from nanoscale fillers (e.g. nanotubes, silica, layered silicates, etc.) is experimentally well established. Many analytical and numerical models are available to predict the overall effective/ homogenized properties of these heterogeneous polymer-filler systems. It is commonly accepted that the overall effective physico-mechanical properties of a polymer-filler material depends on the type of the filler, its aspect ratio, modulus of the filler, modulus of the matrix, concentration of the filler, etc. Hence, prediCtion of the homogenized properties of heterogeneous materials is widely documented in literature. Various theories and models have been proposed, each with their own advantages and limitations. Overall, homogenization techniques can be classified into [1-25]: mean field approaches, variational bound methods, periodic rrricrofield approaches / unit cell methods, embedded cell approaches, windowing approaches, rules of mixtures and semi- el'npirical formulae such as Halpin-Tsai methods, etc. The simplest method of homogenization is the rule of mixtures wherein the overall properties are calculated as an average over the respective properties of the constituents WCighed by their volume fractions. Clearly, this method takes into account only one microstructural property, the volume fraction, and is applicable for simple geometries and linear nraterial properties [1-2311-25]. A more sophisticated method is the self-consistent or eff eChive-medium theories developed by Eshelby [1-26]. Eshelby’s theory was further deVCIOped by various researchers including Hashin [1-27], Hashin and Shtrikrnan [1-28], Hill [1 ‘29], and others [1-9]. Effective overall properties are obtained as an analytical solution of a bC>laxzrdaryvalue problem for a spherical/ ellipsoidal particle in an infinite rrratrix. In order to £1193 into account for the particle-particle interactions, Mon-Tanaka [1-25] modified the 11 L—m-n'r “Ufmr ‘m.'.n.;uum.'—'1LH #3. nelson theor.“ ML 5.515 of small“ ”ml grits. Emfldfd. or modified wan-13mm in iflier c men-2511.331 Another mathematical 53ers3tssan et al [1-31] an EL. 3 the ratio of characteri isel in an astnrpitotic err 1‘53 are then used to 117525321 methods, tum . 1% depends on the sele 35). The rut should b 11mm and Small e :tmading RVE'S are the Quill Or 11de al ”I? We Eshelby inclusion theory. Mori-Tanaka estimates are commonly used to estimate elastic properties of particulate composites and are found to agree reasonably with experimental results. Extended, or modified, Mon-Tanaka schemes have been proposed that also take into account variations in fiber orientations and fiber lengths using probability distribution functions [1-2511-30]. Another mathematical approach is the asymptotic horrrogenization method reported by Bensoussan et al. [1-31] and Sanchez [1-32]. In this method “natural length parameter,” which is the ratio of characteristic size of the heterogeneities to a measure of macrostructure, is used in an asymptotic expansion of the displacement and stress fields. Variational principles are then used to create a link between the scales [1-911-25]. Among the computational methods, unit cell methods have become widely used. The success of these methods depends on the selection of the unit cell or the representative volume element (RVE). The RVE should be selected such that it is large enough to represent the microstructure and small enough to allow efficient computational modeling. The Corresponding RVE’s are then analyzed with proper loading and boundary conditions via analydml or numerical methods. Unit cell methods allow modeling of micro-scale geometries in great detail (specifically in finite element based unit cell methods) and enable inVCStigation of the influence of different geometrical features, interfaces, interparticle interactions, etc., on the overall material response. However, since these approaches forEmulate the macroscopic constitutive relations based on the behavior of a single RVE Subjeczted to a given loading history, they are in fact successful only for small deformations [1 - 91 1251. Unlike most conventional fillers, the modeling of nanoclays, or layered silicates, pose new and unique modeling issues such as: a) particle size, b) hierarchical modeling, and c) 12 ¥ gifts and gill“? PmP" “36 graft} in prediaing homo fig: scale '5 in the order ataxia, depending on maverick}: my be in r1 17:5: in the order of micn mummies CXlllbll both 11'. momma] mode? aim the limintion in moc 133*st The concepts 133cm mOdfils can 1 lmibffi have med 10 Add, ram Sheng CI al [1-33 2‘ ileum) mlCrOrmch‘m mi Mic“ mOdel using m ”flush“ size. 311}an 1. ' . A algrfifl'fnmned l S ‘ him 1 interface and gallery properties [1-33]. Analytical micromechanical models have proved to be successful in predicting homogenized properties of conventional composites where the filler length scale is in the order of tens of microns or larger [1-33]. In nanoclay reinforced composites, depending on the processing, the clay morphology varies significantly. Exfoliated clays may be in range of nanometers and highly intercalated clay agglomerations my be in the order of micrometers. Moreover, it is commonly agreed that polymer/ clay nanocomposites exhibit both exfoliated and intercalated (hierarchical) morphologies. Cleariy, most micromechanical models cannot handle this complexity [1-33]. More importantly, it shows the limitation in modeling hierarchical morphologies that can vary by orders of magnitudes. The concepts of “rmtrix” and “particle” that are well-defined in micromechanical models can no longer be directly applied to polymer-clay nanocomposites. Researchers have tried to address this issue by defining an “effective particle” or a “pseudo inclusion.” Sheng et al. [1-33] employ an “effective particle” in analytical and numerical (finite element) micromechanical models. Similarly, Brune and Bicerano [1-34] developed a pseudo inclusion model using the Halpin-Tsai empirical model by substituting the effective Pseudo inclusion size. Miyagawa et al. [1-6] used the same concept in Mari-Tanaka models. T'he aforementioned studies with the concept of “effective particle” have shown to model the hierarchical nature of polymer/ clay nanocomposites successfully and have reported good agreement with experimental data [1-611-3311-34]. The final and most important issue in nlodeling of polymer/ clay nanocomposites is the particle-polymer interface. Depending on die Chemical compatibility, or functionalization, of the nanoclay and polymer the resulting htet'facial properties will vary considerably [1-35]. Additionally in intercalated clay Irlc>II>hologies, wherein the clay platelets are in stacks (called galleries), the properties of the "hteziar in the galleries is not fullyunderstood [1-33]. 13 ll]? abOVC n‘fntl acting of such comp 3,3111 address these iss ridge. nozrlinear mf‘er hienrthical mt filtered in the work N 5:: 2110mm rmhoc 2:1 {1-911-1311-36} me- [1-1211-3611-37 iii-reel methods [1-38: Rmc‘d‘? modelin: The above mentioned issues specific to polymer/ clay nanocomposites make modeling of such composites challenging. Finite element (FE) based modeling and simulation address these issues and enable modeling of interfaces, different particle sizes and morphologies, non-linear material properties and complex loading [1-25]. Due to the complex hierarchical nature of polymer/ clay nanocomposites multiscale models are considered in this work. Not limited to polymer/ clay nanocomposites, multiscale models or direct micro—mm methods have been used for homogenization of different heterogeneous rmterials [l-9Il-IOIl-36Hl-39]. These models have also been termed as “multiscale methods [1-1011-3611-371” “FE2 methods [l-38],” “integrated methods [1-39],” and “multilevel methods [1-38]”, and are briefly described in the following section. 1.3-5 Multiscale methodology Multiscale modeling covers a wide range of simulations and modeling techniques. Specifically, multiscale modeling can be coined as any modeling scheme spanning both length and time scales associated with analyses that describe material behavior (Figure 1-2,[1- 40]). The nmltiscale, computational modeling of materials has a prinnry challenge: hierarchical modeling of materials; namely, the accurate prediction of physical/ chemical Properties and behavior from nanoscale to macroscale without loss of intrinsic stmctural information. Ideally, a multiscale modeling scheme would be the one that links all the scales f 10111 nano to macro. In other words, it should connect the discrete molecular structure with the bulk continuous structure. Nevertheless, it has been found that the literature terms any Meling scheme that links any two scales a multiscale modeling scheme; although it may not take into account the quantum or atomistic behavior of the material. For instance, a n1()<31‘~'==1ing scheme that integrates a micromechanics (microstructure) and structural 14 1:531:13 limm/ glob-1D mc Missed 35 3 mulLiSCllC I In 1113 Study, the ehel 111mm and that of Cl femur. These aspeCIS ar rmhn'hr hm This . is wait and hence a multi *3: 3 studied Neverthel: mien behm'or from ate 1:113 scheme to study all lanous homogenim ‘rr- . . 55thOfriogemzatron i mos“OPE honogenec mm and Stress-Strain $35.: I mechanics (macro/ global) model to obtain the behavior of a material or structure would still be referred as a multiscale method, although it failed to propagate properties from the atomistic scale. In this study, the chemical compatibility or functionalization between the petro—resin with bio-resin and that of clay with resin system is governed by atomistic and molecuhr interactions. These aspects are generally studied using tools in computational modeling such as molecular dynamics. This aspect of atomistic/ molecular modeling is beyond the scope of this work and hence a multi-level modeling that linls different scales within the continuum regime is studied. Nevertheless, it has been shown that it is possible to integrate material simulation behavior from atomistic scale. It is clear that this would be the ideal multiscale modeling scheme to studyall material parameters. W Various homogenization techniques exist and have been used widely in the past. The process of homogenization involves finding the average properties at a lower scale due to the rmcroscopic homogeneous deformations. In other words, it involves finding elastic Constants and stress-strain (constitutive) relations that relate the two scales. For any generalized continua, constitutive equations are very difficult to write, and in complex Inaten'ajs, it is difficult to fit experimental data. Instead of trying to build differential systems to eStalblish a stress-strain relation at a macroscale, a multi-level finite element computation does not require any constitutive equations to be written at the macroscopic scale; all non- lirlea-rites are obtained from separate FE analyses at lower hierarchical (micro/nano) scale [1- 911‘ 10] [1-2511-36]. Hence, FE-based homogenization would be used in this research. Litemtlne prescribes that FE based models are constructed using the following three main C .. OI-:.s‘3lrtra.=:nts [1-911-36]: (1) modeling of the mechanical behavior at the lower scale (the 15 ¥ km“ “‘1" ‘ 53,60,311] 5min), and 1 mg the microrneehanit llhscale FE mo arm at he macroscale ; rial hehasior. In this sn 1:: used by Kouznetsova am point of the Semen '5 performed. La." conditions are d. 2m point is derived Em Winnie. The con 3113.3 derived from the R RVE), (2) localization rule which determines the local solutions inside the unit cell (for any given overall strain), and (3) homogenization rule giving the macroscopic stress tensor, knowing the micromechanical stress state. Multiscale FE modeling is an iterative procedure that assumes homogenized behavior at the macroscale and enables the incorporation of large deformations and arbitrary material behavior. In this study, the multi-level FE approach as developed by Smit [1-10] and later used by Kouznetsova [1-9] and Breuls [1-36] is proposed. In this approach, to each integration point of the macroscopic mesh an RVE is assigned and a separate FE computation is performed. From a macroscopic deformation tensor (Fmacm), appropriate boundary conditions are derived to be imposed on the RVE. The macroscopic stress at integration point is derived from the RVE by averaging the resulting RVE stress field over the RVE volume. The consistent macroscopic stiffness matrix at macroscopic integration point is derived from the RVE stiffness matrix. When using this multi-level approach there is no need to specify macroscopic constitutive behavior, instead the behavior at the mermcopic integration point it is derived from associated microstructure. Breuls [1-36] PIOVides a good explanation and derivation of the complete multi-level modeling scheme. A SChenntic representation of the multi-level FE scheme is provided in Figure 1-3. In this “’Ol‘k multi-level FE methodology was used to predict the tensile behavior of bio-based n"=‘110Composites and hybrid biocomposites and the results were compared with experimental bellilvior. The details and results are provided in Chapter 8. Z 4 zlletéod/Appmacé i The goal of this work was to develop eco-friendly, hybrid bio—based composites by c""C’I'IDOrating an mtegrated expenmental and computational approach to strategically explore 16 Mpg possibilities and l 33.221131]. neural mode 5:: moments are a key ..et mo 92 major I; ‘g._.._1_s - Computations Buttress Lisle m acco 1.4.1 Erperirmntation - al Development of c051 of biobassd pohmer bl Use of G’NCs for bit {l Gammon of C 112 Simuhtiom _ Com; il R‘vEs/m cells that “1501:: biomgim 5L will“ accePuble [ole b) The degree of m. mmm result, 1 Momd cl The M1160” of E the design possibilities and limits. As discussed earlier, the approach will integrate multiscale computation, mater'nl modeling and optimization and is schematically shown in Figure 1-4. Since experiments are a key part of modeling at each scale, the overall research study can be classified into E major tasks: Exmrimenm'o - Processing and characterization, and My - Computational Models, material layout optimization and nrultiscale modeling. Each of these tasks was accomplished with subdivision of tasls as follows: 1.4.1 Experimentation - Processing 8: Characterization: a) Development of cost and time efficient processing techniques for the development of bio-based polymer clay nanocomposites (Chapter 2). b) Use of CPNCs for biocomposite plates using hand layup and VARTM. (2) Characterization of CPNCs and biocomposites( Chapter 3 - Chapter 6). 1.4.2 Simulations - Computational Models 8t Multiscale Modeling W a) RVEs/ unit cells that can accurately represent the nano-reinforced polymers with or without bio-resins such that the RVE can accurately provide material parameters within acceptable tolerance to experimental results were developed. b) The degree of intercalation of nanoclay by validating virgin UPE results with experimental results (Number of layers in intercalated gallery obtained by TEM) was performed. C) The distribution of bio-resin was determined through a material layout optimization problem and validation by comparing to experimental results W 17 ,1 Amplified compuh1 pmpmics are 01113 recliner/clay R\lis 1115.1 biocomposites like" (Manuals 1qu Amlnlevsl finite e 11:. through proper RW: 327:: 8). A schematic d. ti: prodded in Figure 1. 3350321111 figure 1'5(C) 1 Timed due to m and co: its more of hybrid b used to IDOdel mechanic a) A simplified computational model (RVE) similar to that of nanoscale, wherein resin properties are obtained from lower-scale analyses of appropriate bio-blend polymer/ clay RVEs was developed and implemented in nrultiscale analyses for hybrid biocomposites. A multi-level finite element approach that integrated the three scales, from nano to macro, through proper RVEs and phenomenological models at each scale was impelemented (Chapter 8). A schenntic description of the overall computational approach used in this Work is provided in Figure 1-5 . The large scale biocomposites manufacturing and simulation as shown in Figure 1-5(c) was initially proposed to be a part of the work, but was later removed due to time and cost point of view. Nevertheless, the approach presented up to the tensile response of hybrid biocomposites validates the approach and is believed that it can extended to model mechanical behavior of large scale structural components. 1)- fiape The study aimed at developing cost-effective environmental friendly, multifunctional hybrid hietarchical biocomposites using a rational approach that eliminates costly trial and error experiments, by incorporating an integrated experimental and computational approach to St31"r7itegically explore the design possrbilities and limits. The efficiency of the hybrid biocorliposites is a result of the synergy of the different constituents at various length scales. The Work aimed at utilizing this synergistic effect to its full extent by identifying key ”hauler—en, phenomena and behavior of the various constituents that make the resultant bl°°°nlposita such that only the beneficial effect of each of the constituent is revealed and the detrimental effects are suppressed, thereby leading to a superior/ balanced 18 idiom] hybrid bioco m lezgh scales provided 353.301 constituents on r firm of efficient com The study aimed a' :Igezhed mmscopic be mm] scales. Ar the m the-i mead the constituti mpg reeks of detailed n retgenehies at the mno/ rr The proposed Stud} cu: bearded to thermal and d II31735{IE-Dome mechanh ”inept Properties by p: 3311 3 WC” ' tessful mod: | ‘ «Firm 0f novel, effici: mam 0f hiobased nl m mmbk, emimml ’ 31% h 9%an i: rm" - M13011 . 51min ”Ethodolg; hm medals them} are. ”A; P“..- |I multifunctional hybrid biocomposite. Experimental characterization these materials at various length scales provided vital information on bounds/ limits of processing and limits on effects of constituents on resultant properties. Most importantly, the experiments aid the development of efficient computational models. The study aimed at implementing a computational scheme that models a homogenized macroscopic behavior taking into considerations the heterogeneities at lower dimensional scales. At the macroscopic scale, no constitutive/ phenomenological model is defined; instead the constitutive behavior at macroscopic integration points is determined by averaging results of detailed modeling of the lower-scale material structure. The influence of heterogeneities at the nano/ micro scale was studied. The proposed study currently focused on mechanical behavior. A similar approach can be extended to thermal and diffusion analyses. The approach will increase the understanding of micro/nanoscale mechanisms governing various parameters, thereby enabling tailoring of macrosc0pic properties by proper synergy and distribution of materials at respective scales. Overall, a successful modeling scheme, as the one used in this research could enable dev"-31<>1:>I:nent of novel, efficient, cost effective multiftmctional materials. On the other hand, developrnent of bio-based materials and their applications for structural applications would lead to Sustainable, environmental friendly, bio-degradable and cost effective materials, while benefiting the agricultural industry. The success of this computational scheme will enable Prediction of macroscopic responses for various material compositions and macroscopic Shapes '- Similar methodology could be applied for other types of nanoparticles and hieré“"311an rmterials thereby providing guidance in designing such hybrid hierarchical untenals 19 35 Ogrrrrxrm . I}: disertztion has l‘i mama to the we . Omar reports the techriques for develop ' Canterl and Cup mnocomposites with ' Gapter 5 and Orr; biocomposites produ. ml Orrpter 4, respect ' Chip?! 7 pmiides a meChalice] and barrie- ' (hams reports {hi rdhihn'd biocompjl ' OW" 9 provides Emblem thar aimed ‘ of bbresin in a man. mm” and expel ' FM} Gupta ”COWndarjonS' 16 Organization The dissertation has been organized into ten chapters. The first chapter provides an introduction to the work with a brief background, scope and objectives of this work. Chapter 2 reports the study on development of novel solvent based processing techniques for development of bio-based nanocomposites. Chapter 3 and Chapter 4 report the experimental characterization of bio-based nanocomposites with EMS and EML bio-resins, respectively. Chapter 5 and Chapter 6 provide the experimental characterization of hybrid biocomposites produced from the nanocomposite resins as reported in anter 3 and Chapter 4, respectively. Gaapter 7 provides an overview of numerical and analytical models to predict the mechanical and barrier properties of bio-based nanocorrrposites. Chapter 8 reports the multiscale simulations of tensile response of nanocomposites and hybrid biocomposites using a multi-level FE technique. Orapter 9 provides an overview and results of a material layout optimization problem that aimed at determining an enhanced RVE, that obtains the distribution of bio-resin in a nanocomposite RVE by minimizing the error baween homogenized parameters and experimental target properties. Finally, Grapter 10 provides concluding remarks, research needs and recommendations. 20 .7 Nil" III/fig”?! scale Sample Elm 1-1. Schemaii 10'“ grammar 1.7 7454’: and [1371025 Mesa-Scale Micro-Scale Polymer + Layered Aluminosilicates Natural Fibers + Nano-reinforoed Resin Macro-Scale Sample Figure 1—1. Schematic showing rmrltiscale components and phase transitions W ¢ Multiscale ¢ Computational Modeling Mechanics Computational Chemistry MQDELINEJQQLS. Quantum Mechanics H‘Discrete Molecular Structure-——>l<—Continuous Material Structure—H Structural Mechanics Nanomechanics [Micromechanics 10"“ 10“ 10‘ 10° l I | | Mimi 10'15 10'" 10" L | | I F‘ lgxlre 1—2. Computation materials - Length, time scales, modeling methods and associated tools [1-40]. 21 ............. FE program With Integration fit macroscopic mesh oin t Loop (homogenized) p Compute 4 amacro ’ S macro FE Program with BCs ofRVE Compute microscopic R VE Macroscopic stress / OWE , “5;” mesh stram fields as hetero eneous Input Figure 1-3. Schematic of multiscale FE approach (Adapted from [1-10]). NAno _SCALE PW?“ W) .tmeso* (Rooln Syswn + micro-HMO nano-inclusions) l Experimental Evaluation 0! ‘ Large Scale I Structures Experimental Evaluation of prototypes: (Plates + SCALE Coupons) ATOMIC SCALE r-Oll-----, I ‘ ........... Experimental Evaluation of prototypes: (Coupons) 1 Evaluation of thermo+mechnnical and Physclal Properties like Stiffness. Toughness, CTE. banier properties 8. comparison with computational and theoretical models Figure 1-4. Hierarchical Modeling and Evaluation Techniques 22 [ ~ Nanoparticle l morphology validation usir V primary resrr _ E E E .- = -_ _ E a _ E .— I In > r: ‘o’ 2 E Q E rn b r — _ — _ _ — - —— E_- _ a — = NANO Nanoparticle — — morphology —' - ' validation using E —- E (a) rima resin P ry — :1 55 EB E : : Material Layout, _ _ __ E Distribution of Secondary _ resin (Bio-resin) E E _ Integration Point, Unique RVE assigned I O O O O C E O I I O 0 o MESO/ 4: . . . . . . MICRO 5 O 0 O O 0 O 'U a , . ,_ , a WWW (b) :‘s‘VS‘Q FIBER o- I? seem MACRO (c) Section of a load bearing ‘ biocomposite macro-structure Figure 1-5. Overall approach of multiscale modeling for load bearing biocomposites 23 5.! hirer! or \lchnn' AK. Misra .rlr . - . Biocomposrtes: A: 176/"37:134. :11 Mel], Riedel U. A: :33} Donnell A0, Dweih resin. Composites 341 More 35337001 RP. 3 a novel nodified 457-146. ill Rogerio R will SUS'tAinable cellui. resin for housingl' 2035, 13:139-149. iii ll'fvlgawa H, MDhar: from blends of Journal Of Polmj i7:- lifagawr H, Mohr:i bohased urisatxl Industrial and E r" ill litigant H, Mela" thermophysical Innocomposites. 1.8 fleflwflce: [1-1]. Mohanty AK, Misra M, Hinrichsen G. Biofibers, Biodegradable Polymers and Biocomposites: An Overview. Macromolecular Materials and Engineering. 2000; 27 6/ 277 :1-24. [1-2]. Nickel], Riedel U. Activities in biocomposites. Materials Today. 2003, 6:44-48. [1—3]. Donnell A0, Dweib MA, Wool RP. Natural fiber composites with plant oil based resin. Composites Science and Technology. 2004, 64: 1135-1145. [1-4]. Morye SS, Wool RP. Mechanical properties of glass/flax hybrid composites based on a novel modified soyabean oil matrix material. Polymer Composites. 2005, 26: 407-146. [1-5]. Burguefio R., Quagliata MJ, Mehta GM, Mohanty AK, Misra M, Drzal LT. Sustainable cellular biocorrrposites from natural fibers and unsaturated polyester resin for housing panel applications. Journal of Polymers and Environment. 2005, 13:139-149. [1-6]. Miyagawa H, Mohanty AK, Burguefio R, Drzal LT, Misra M Novel biobased resins from blends of functionalized soyabean oil and unsaturated polyester resin. Journal of Polymer Science : Part B: Polymer Physics. 2007, 45:698-704. [1-7]. Mpgawa H, Mohanty AK, Burguefio R, Drzal LT, Misra M Development of biobased unsaturated polyester containing functionalized vegetable oils. Industrial and Engineering Chemistry Research. 2006, 45:1014-1018. [1-81- Myagawa H, Mohanty AK, Burguefio R, Drzal LT, Misra M Oraracterization and thermophysical properties of unsaturated polyester-layered silicate nanocomposites. Journal of Nanoscience and Nanotechnology. 2006, 6:464-471. [1'91 Kouznetsova V, Brekelmans WAM, Baaijens FPT. An approach to micro-macro modeling of heterogeneous materials. Computation Mechanics. 2001, 27 :37-48. [1-101- Smit RJM, Brekelrmns WAM, Meijer HEH Prediction of the mechanical behavior of nonlinear heterogeneous systems by multi- level finite element modeling. Computational methods in Applied Mechanical Engineering. 1998, 155: 181-192. I 1‘111- Bllrguefio R, Quagliata M], Mohanty AK, Mehta GM, Drzal LT, Misra M. Hybrid biofiber-based composites for structural cellular plates. Composites Part A: Applied Science and Manufacturing. 2005, 36:581-593. [1-121' Burguefio R, Quagliata MJ, Mohanty AK, Mehta GM, Drzal LT, Misra M Hierarchical cellular designs for load bearing biocomposite beams and plates. Material Science and Engineering A. 2005, 390: 178-187. 24 f1; 83:;an R ngxm . learns, natural?! * applied Science {1.11}\lohann AK W383 orginoclav and bi A_\TEC Confers :1-‘5, llshra S, lfisra .‘vl. Performance of Composites Scien {ran lllaq M, Burguen pointer/layered Knish AI, not... toughness of can “mes- louma'; :ill- l‘kldo TK Aim J ESlV’CS IOJOm i-El . ‘ Coin}. PH Velma: ' ll 1)“ [l- 13]. Burguefip R, Qua-gliata M], Mohanty AK, Mehta GM, Drzal LT, Msra M Load be fiber composite cellular beams and panels. Composites part A: Applied Science and Manufacturing. 2004, 35: 645—656. [1-14]. Mohanty AK, Miyagawa H, Burguefio R, Misra M Biobased nanocomposites from organoclay and blends of unsaturated polyester and functionalized vegetable oil. ANI'EC, Conference Proceedings. 2005, 4:52-56. [1-15]. Nfishra S, Misra M, Tripathy SS, Nayak SK, Mohanty AK. Studies on Mechanical Performance of Bio fiber / Glass reinforced polyester hybrid composites. Composites Science and Technology. 2003, 63:1377-1385. [1-16]. I-th, Burguefio R, Mohanty AK, Misra M Bio-based unsaturated polyester/ layered silicate nanocomposites: characterization and therrno-physical properties. Composites Part A. 2009; 40:540—547. [1-17]. Ratna D. Mechanical properties and morphology of epoxidized soyabean oil— modified epoxy resin. Polymer IntemationaL 2001; 50:179-184. [1-18]. Shengton III HW, Wool RP, Hu B, Donnell AO, Bonnaille L, Can E, Cnapas R, Hong C. An All-Natural composite Material Roof System for Residetial Construction. Proceedings of International Conference on Advances of Building Technolgoy, Hong Kong, Cnina, December 2002. [1-19]. Quagliata M]. Development and Characterization of Biocomposite Cellular Bearm and Plates for Load-Bearing Components. Masters Thesis, Michigan State University, East Lansing, MI; 2003. [1- ”20] Kinloch A], Mohammed RD, Taylor AC, Sprenger S, Egan D. The interlaminar toughness of carbon fibre reinforced plastic composites using ‘hybrid-toughened’ rmtrices. Journal of Material Science. 2006, 41:5043- 5046. [1‘23- I-Isiao TK, Alms J, Advani SG. Use of epoxy/multiwalled carbon nanotubes as adhesives to join graphite fiber reinforced polymer composites. Nanotechnology. 2003, 14:791-793. [1‘22]- Gojny FH, Wrchmann MHG, Fiedler B, Bauhofer W, Schulte K. Influence of nano- modification on the mechanical and electrical properties of conventional fiber- reinforced polymers. Composites Part A. 2005, 36:1525-1535. [1‘231- Le Baron PC, Wang Z, Pinnavaia 'I]. Polymer - layered silicate nanocmposites: an overview. Applied Clay Science. 1999, 15:11-29 [1‘241- Haq M, Burguefio R, Mohanty AK, Misra M Processing Techniques for Bio-based Unsaturated-Polyester/ Clay Nanocomposites: Tensile Properties, Efficiency, and Limits. Composites - Part A 2009; 40:394-403. [1‘25]- Bohm, H]. A Short Introduction to Basic Aspects of Continuum Mechanics.” CDL- FMD Report, 3 (1998). Vienna Institute of Technology. 25 wflfibfllflfi fikmzmc 11.23; Hashin 2- shill“ nnltiphélse r mnm» :39; ill R A self-COl and Physics < [153} Dachlbauer D Composites localization Material 33: :1-31} Bensoussan A. Smrtures, 1' :2-32] Samba-Palmer menses, 1*. 1333-; Sheng N, Bow micromechar clayparticle. :34} anc DD, Dice and compre exfoliation a1 3353. norm EK, Mo Surface POllmmh- TechnolOg'u' 35 Beds RGM, Se cell dc Clement appr i lam (let Sluis O ftfngeneor MECh-‘mics ol ill. level F .. , Ch; ' belts boc ior of k APPHedMecl [1-26]. Eshelby JD. The determination of the field of an ellipsoidal inclusion and related. Proceedings of the Royal Soceity of London, Part A. 1957, 241:376-296. [1-27]. Hashin Z. The elastic modulii of heterogenous materials. Journal of Applied Mechanics. 1962, 29:143-150. [1-28]. Hashin Z, Shrikman S. A variational approach to the theory of elastic behavior of multiphase rmterials. Journal of the Mechanics and Physics of Solids. 1963, 11:127-140, [1-29]. Hill R. A self-consistent mechanics of composite rmterials. Journal of the Mechanics and Physics of Solids. 1965, 13: 213-222. [1-30]. Duschlbauer D, Bohm H], Peterrnann HE. Computational Sirmrlation of Composites Reinforced by planar random fibers: Homogenization and Localization by Unit Cell and Mean Field Approaches. Journal of Composite Materials 2006, 40:2217-2234. [1—31]. Bensoussan A, Lionis ]L, Papanicolau G. Asymptotic Analysis for Periodic Structures, 1978. North Holland, Amsterdam. [1-32]. Sanchez-Palencia E. Non-homogeneous Media and Vibration Theory. Lecture Notes in Physics, 1980. Springer-Vedag, Berlin. [1-33]. Sheng N, Boyce MC, Parks DM, Rutledge CC, Abes JI, Cohen RE. Multiscale micromechanical modeling of polymer/ clay nanocomposites and the effective clay particle. Polymer 2004, 45:487-506. [1-34]. Brune DD, Bicerano J. Micromechanics of nanocorrrposites: Comparison of tensile and compressive elastic modulii, and prediction of effects of incomplete exfoliation and imperfect alignment on modulus. Polymer 2002, 43:369-387. [1-35]. Drown EK, Mohanty AK, Pamlekar Y, Hasija D, Harte BR, Misra, M, Kurian Iv. The surface characteristics of organoclays and their effect on the properties of poly(trimethylene terephtahalate) nanocomposites. Composites Science and Technology, 2007, 67:3168-3175. [1'361- Breuls RGM, Sengers BG, Oornens CW], Bouten CVC, Baaijens FPT. Predicting local cell deforrmtions in engineered tissue constructs: A multi-level finite element approach. Transactions of the ASME 2002; 124:198-207. [1‘37]. Van der Sluis O, Schreurs PJG, Brekelrmns WAM, Meijer HEH. Overall behavior of heterogeneous elastoviscoplastic materials: effect of microstructural modeling. Mechanics of Materials 2000, 32:449-462. [ 1~38]. Feyel F, Graboche J'L FEZ multiscale approach for modeling the elastoviscoplastic behvior of long fiber SiC/Tr composite rmterials. Computational Methods in Applied Mechanical engineering. 2000, 183:309-330. 26 33;; lhtsul K. Tera solids With 636. 1142', Gates, T5. C materials: Conrosne [1- 39]. Matsui K, Terada K, Yuge K. Two-scale finite element analysis of heterogeneous solids with periodic microstructures. Computers and Structures 2004, 82:593- 606. [1-40]. Gates, T.S., Odegarrl, GM, Frankland, S.].V., and Clancy, T.C., “Computational materials: Nhrltiscale modeling and simulation of nanostructured rmterials” Composites Science and Technology. 65 (2005) 2416-2434. 27 (hapterl. Develc ill ffl'iUl’f lei-lead clay/polymer zli glorified {Ifll'l}‘l : Farr lllei‘lods were attention of high Find using 6ko silence, and sonicatil n“ :orzpogfie m. minim of hng MD Added Alter sc :r‘ C150helm. and hi "re ltd ‘0 enhanced 1.7 [rm/mm BlOl‘mcd YESir mm based resins if- A - ' mm DOW-’1. value limes for Mole failed from an em erg. _ on blow)“ 3‘s. leg. . life/n will": R \l Emilia}, \qnocomp. 1 Chapter 2. Development of Bio-has ed Clay Nanocomposites 2.1 Aértmct Bio-based clay/ polymer nanocorrrposites using blends of styrene-based unsaturated polyester and epoxidized methyl soyate were manufactured using solvent-based processing techniques. Four methods were evaluated to assess limitations related to solvent removal and incorporation of high clay and bio-resin content. Nanocomposite characterization was performed using electron microscopy and tensile tests. Solvent type, bio-resin addition sequence, and sonication energy were the key parameters governing processing efficiency and composite quality. Processes with bio-resin added after solvent removal show promise in incorporation of high bio-resin and nanoclay contents. Use of acetone as a solvent with bio-resin added after solvent removal led to nanocomposites with good nanoclay dispersion and exfoliation, and high tensile modulus. Direct sonication in the base resin diluted with Styrene led to enhanced and balanced gains in stiffness and toughness. 2.? [film/actor: Bio-based resin systems obtained as blends of fimctionalized vegetable oils and Petmleurn based resins or vegetable oil based resin systems can be reinforced with nanoclays ‘0 Obtain novel, value added applications for natural polymers [2-1]. The development of altennauzives for petroleum derived materials win. plant-based renewable materials has been propelled from an environmental viewpoint [2-1]-[2-4]. Mohanty et al. [2-2] provide a good overview on bio-polyrners, bio-fibers and bio-based composites, biocorrrposites. Bio-based \ 113% M., Burguefio, R., Mohanty, A.K., and Misra, M, “Processing Techniques for Bio-based Unsaturated- 40%”;an Nanocomposites: Tensile Properties, Efficiency, and Limits,” Composites- Part A, 2009; 28 m lined 15 a CO} 5': as Wilda-7' CO] strap-5136i VIM :1: seem leading 1 11222553 SCIiOUSlV 3 rm ol the res amt and a prop 53.1.15 has Show tl "tents of rigidit} a :r-iication [2-1] On this addition of b: I‘iESSllIg Ol nanoclay franc-thy reinforcec ”Eton and thermal all “@33ng issn Mil LeCliniqlles f Pmleumbase. $9.110 impair mUltifl ”less but improve] Wes IZ-sl m, m- enemy,- has llfl W'mmdlmto a( resins defined as a combination of pendemrlnsai resins as primary constituent and mum! bio resins as secondary constituent, have shown to improve the toughness of the resulting resin system [2-512-6]. This increase in toughness is due to the reduction in cross-link density in the system, leading to increased plastic deformations [2-7]. However, this increase in toughness seriously affects the modulus [2-5}[2-7], thermal [2-612-7] and barrier [2-6] properties of the resulting polymer. Stiffness and toughness are opposing performance parameters and a proper balance is required to develop an efficient biocomposite. Moreover, research has shown that use of plant-oil based polymeric materials do not show adequate properties of rigidity and strength for load-bearing applications by themselves and require modification [2-1]. One approach pursued by the authors is to recover the property losses from the addition of bio-resin by the addition of layered silicates or nanoclays. Although the processing of nanoclay reinforced polymers is well established, the solvent-based processing of nanoclay reinforced bio-based resin systems brings about new issues, such as phase separation and thermal degradation of the polymer constituents [2-6]. This work addresses these processing issues and summarizes the development of suitable solvent-based PTOCCSSing techniques for efficient production of nanoclay reinforced bio-based polymers. Petroleum—based polymers reinforced with layered silicates, or nanoclays, have been shown to impart multifrmctionalityto the resulting polymers with enhancements not only on Stiffness but improvements in thermal, barrier, flammability and ablation resistance PmPCIties [2-8]. The reinforcement of soy [2-112-912-10] and com [2-11] based bio-resins With nanoclays has yielded similar results. In spite of the noted enhancements, stiffness in“movement due to addition of nanoclay also increases the brittleness of the resulting Polymel's [2-6]. Moreover, petroleum-based resins such as unsaturated polyesters (UPEs), Which are commonly used due to their low cost, ease of handling and good balance of 29 opmkw”h gm oi “lb peUU term [3.3 3613-6 Infill: film has l mnhsralchmaf of ill with nanocla.‘ trig bioresin and n mtg bioresin to: files. Partial recon tries were achi mnposites thus tens) [2.112.512 Wilmernith incre lit commonly lli M. ,. ' ~ malty reinforced mechanical, electrical, chemical and fire resistance properties, are inherently brittle in nature [2-1212-13]. As far as bio-resins, their mechanical and thermal properties are far lower than required for structural applications. Research has shown that the blending of functionalized soybean oil with petro-based resins (UPE) can increase toughness of the petro-based thermoset [2-212-512-612-13]. The substitution of non-renewable synthetic resins with renewable resins has been proposed and studied considerably [2-512-13}[2-17]. The thermophysical characterization of bio-based resin systems using functionalized soybean oil and UPE with nanoclay has been performed by our group [2-6] by studying the effect of varying bio-resin and nanoclay content on tensile, irrrpact, moisture and tensile properties. Increasing bio-resin content was shown to improve ductility and toughness but reduces stiffness. Partial recovery of lost stiffness and inrprovernents in thenml and barrier properties were achieved with nanoclay reinforcement. The resulting bio-based nanocomposites thus showed multifunctional properties with improved mechanical (toughness) [2-112-512-612-1512—16], thermal [2- 1] and barrier properties [2-6] than the base polymer with increased environmental appeal. It is commonly agreed that the enhancement of physical and mechanical properties 0f nEllloclay reinforced polymers is dependent on the distribution and morphology of nanoclay in the matrix. In order to obtain maximum benefits from nanoclays, exfoliated mm"PhOIOgies are expected. However, in reality, a combination of intercalated and exfoliated morphologies is typically obtained [2. 1812- 19]. The degree of exfoliation and intercalation of nanoclay depends on the processing technique and compatibility (functionalization) of nanoday with the host polymer system. The four main processes currently used for the p Wu of nanoclay reinforced polymer systerm are exfoliation-adsorption, in-situ I3013’mfil’ization, melt intercalation, and template synthesis [2-812-20]. The exfoliation- 30 3;;an Pm“CS5 is ml?" and Doll-“r glee or solution b all sired 10 C fnfiimlmlom 50l‘ :3be CXiOllAIfi risers soluble. Vi lenient The pro lie gillneris then a he mnoeomposite r Considerable ll: llm' Ol SOlW {Zillion-1912.2 Loan and Dubo E“ mil-“g proper am 18 a soh. file”Invites. Th. if M exfoli— mimmiloslfis gym Helm Narrow: in , ‘ "9” °‘ “one " in mock.“ a Inter adsorption process is a solvent-based process in which a solvent compatible with borh nanoclayand polymer is used. Hence, this process is also referred to as the common solvent nrethod or solution based processing technique (SBPT). SBPT is a widely used technique and is reported to consistently give exfoliated morphologies provided that proper clay functionalization, solvent and blending conditions are met [2-21]. In this process, the nanoclay is exfoliated or separated into individual sheets using a solvent in which the polymer is soluble. When using an adequate solvent, the nanoclay can be easily dispersed in the solvent. This process is facilitated by stirring and the application of ultrasonic energy. The polymer is then added to the mixture for adsorption to the delaminated nanoclay sheets. The nanocomposite resin system is obtained upon solvent removal [2-20]. Considerable work on nanocomposites from solvent based processing techniques with a variety of solvents has been reported in the literature for both conventional polymers [2- 1212-1812-1912-2112-23H2-26] and bio-based polymers [2-112-612-912-1012—1112—27]. Alexandre and Dubois [2-20] and Le Baron et aL [2-22] provide a good review on synthesis and resulting properties of a wide range of polymers. Miyagawa et al. [2-1212-25] used acetone as a solvent in synthesizing epoxy/ clay [2-25] and polyesrer/ clay [2-12] nanocomposites. The resulting nanocomposites were well dispersed and clay morphologies were partially exfoliated and intercalated. Hutchinson et al. [2-24] compared epoxy/ clay nanocOmposites synthesized by simple direct mixing in virgin resin, and by using acetone as a 501Vent. Nanocorrrposites processed with acetone had better dispersion than those PrePaI'Cd by simple mixing. Burgentzle et al. [2-23] evaluated the parameters controlling the disPeISion of organophillic nanoclays in various organic solvents by studying the interaction between nanoclays and solvent molecules at various scales. The interactions at nanoscale We“: Studied by interlayer spacing, at microscale by studying the rheological behavior of clay 31 :yenSiJ-I‘S, and i need M high i:- W1 been“ essential for dr that: blending V it obtained when missed clay narlc ministers on t ratehtoralumjn eT3317?!V‘i‘gt'table o imimom d5? reed clay m, ifivlbfnzene. f m?““‘Jl°§l3$ Lin suspensions, and at macroscale by analyzing swelling of nanoclay in solvents. It was observed that high surface energy solvents were effective for gelation at the microscale, and that proper balance between the hydrophilic and hydrophobic parts of solvent molecules was essential for dispersion of organophilic clays. Morgan and I-hrris [2-21] investigated how solvent blending with and without high energy mixing dispersed two types of clay in polystyrene using chlorobenzene as a solvent. They found that a high degree of exfoliation was obtained when organo-montmorillonite clay and high energy sonication were used. For bio-based clay nanocomposites, Miyagawa et al. [2-27] studied the effect of clay and alumina nano-whiskers on the mechanical properties of epoxy/ vegetable oil blends. In each case, the nanoclay or alumina whiskers were sonicated in acetone for 2 hours and then mixed with the epoxy/ vegetable oil blend. The resulting nanoclay composites were reported to have homogeneous dispersion and exfoliated morphologies. Lu and Laka [2-11] synthesized bio— based clay nanocomposites by cationic polymerization of com-oil with styrene and diVinyl-benzene. The resulting nanocomposites were reported to have intercalated mlOl'phologies. Lin et al. [2-9], Uyarm et al. [2-1], and Song et al. [2-10] synthesized ePOXidized soybean oil/ clay nanocomposites by direct sonication in soybean oil. Research in our group [2-6] synthesized chy nanocomposites on blends of unsalurated polyester and soybean oil. While good dispersion and exfoliation of the nanoclay has been achieved, the standard SBPT restricts the incorporation of higher concentrations of clay and bio-resin content. One of the drawbacks of SBPT is the amount of energy required for solvent removal. The use of solvent based processing for bio-based resin systerm brings about problems of phase separation and thermal degradation of polymers [2-6]. Increase of Clay content requires an increase of the solvent solution for proper dispersion and eltl:oliation. This requires a large amount of energy to remove the solvent, which can cause 32 15:21 Wm mLMme’lEd mm] 50mm? 56:: s a need for C? The choice emmwe“ xfam’lflg large- m“. 353W ”35 filaments and if 2%; m the aim ‘ é eitrtechniqucs ‘ In his work. Serbprmm of mm trzsseg methods I '34 palm/clay r. Err ed polymer in Four differen InitiCr-pcd b} my gym J $33: ‘ mm at failure i thermal degradation of the polymers. Moreover, the research in our group aims in using nanoclay-reinforced bio-based resin systems for rmnufacturing large scale load-bearing structural components, and therefore large quantities of resin systems are required. Hence, there is a need for efficient processing that can also handle large quantities. The choice of using SBPT to develop bio-based nanocomposites is two fold. First, the aim was to develop nanoclay reinforced thermoset resin systems that could be used in manufacturing large-scale bio-based structural components (such as sandwich panels) using vacuum assisted resin transfer molding (VART'M). And secondly, this work presents new developments and improvements to the recommendations from an earlier study using SBPT [6]. Hence, the aim was to develop improvements on the SBPT and a study on the feasibility of other techniques was not performed as it was beyond the scope of the research aim. In this work, solvent based processing techniques were studied for efficient development of nanoclay-reinforced bio-based resin systems, by evaluating different Processing methods that can maximize the amount of bio-resin and nanoclay content. Bio- based polymer/ clay nanocomposites were processed using SBPT for blends of styrene based unsaturated polyester (UPE) and epoxidized methyl soyate (EMS) with acetone as the solVent. Four different processing techniques were studied by modifying the baseline SBPT devdoped by our group [2-612-12]. Onlytensile properties, namely: modulus, strength and elongations at failure were studied. Comprehensive thermophysicaL mechanical and barrier characterization of nanocomposite resins with soybased resins are reported elsewhere [2-6] and similar workwith linseed oil based resins is in progress and will be forthcoming in future Conhnunication. The effects of sonication energy and styrene content were also studied. Transmission electron microscopywas used to study the degree of dispersion and exfoliation of the nanoclay. Scanning electron microscopywas used to study the features of the tensile 33 3m surfaces. A cor 1:51: amount of bio patted 7; viewer/J. u four different I Z'td sith mncx momsies. In Lb amber: used test 13.] Materials llwmin comp 35, fiihholl Inc, P {I'mzed Ethyl sova lim- n - l. infill “m ”‘50 by Wei NW“ The m mm and 153 Parts Ti? 7 .l.i~.‘ lows. POT all P. :1 “31753 Wulua ' ~som fracture surfaces. A comparison of the four processes with respect to processing time, feasible amount of bio-resin and clay content, drawbacks and overall efficiency is also provided. 2.3 fxpenkzefltd/Metéodt Four different processes were studied for the development of bio-based resin blends reinforced with nanoclay inclusions. Tensile properties were studied for the resulting nanocomposites. In the following sections, details on materials, description of processes, nomenclature used, testing and characterization details are provided. 2.3.1 Materials The nnin component was ortho unsaturated polyester resin (UPE, Polylite" 32570- 00, Reichhold Inc., NC), which contains 33.5 wt.% styrene. A bio-based modifier, epoxidized methyl soyate (EMS, Vikoflex' 7010, Arkema Inc, PA) replaced parts of UPE. The nanoclay used in this work was Cloisite 30B° (Southern Clay Products, Inc. TX). The resin system (mixture of UPE, EMS and organoclay) was processed with cobalt naphthenate (Sign); Aldrich, MO) as a promoter and 2—butanone peroxide (Sigma Aldrich) as an initiator. A cOnstant ratio by weight of the resin system to the promoter and initiator was utilized to Cure all samples. The mixing ratio was 100 parts by weight of the resin system to 0.03 parts pr'OIIIOter and 1.50 parts initiator. Samples were cured at 100 °C for 2 hours, followed by 160 0C for 2 hours. For all processes in this study, the sonication energywas applied using Cole - Partner 750 W ultra-sonicator with a 25 mm diameter solid probe. 34 m testingw Tensile rests £3»; was 5 min/n Te firsion and n '2 emission elc 21rd using a diamc 5: iii: A E01. 1: relight field ima fiibn scanning clef zenith a thin gold 3'4 Pmesszizg 7‘ Mb in 0‘. ”mil application 1131:?! Mlfll 2.3.2 Testing 8: Characterization Tensile tests were performed according to ASTM D638 standards. The rate of loading was 5 mm/ min. Six tensile specimens were tested for each nanocorrrposite system. The dispersion and morphology of nanoclay inclusions in the nanocomposite were observed with transmission electron microscopy (TEM). Ultramicrotomy at room temperature was canied using a diamond knife with an included angle of 4° to produce sections around 70 nm thick. A JEOL 100CX TEM with 1236 filament with 120 kV acceleration was used to obtain bright field irmges. The tensile failure surfaces were observed with a JEOL 6400 field emission scanning electron microscope (SEND at 10 kV acceleration voltage. Specimens were coated with a thin gold film prior to SEM observations. 2. 4 [amassing Dede/yea: Research in our group has been aimed at manufacturing of bio-based composites for Structural applications using these nano-reinforced bio-based resins through vacuiun assisted resin transfer molding (VARTM). As a result, in order to estimate the overall efficiency of the processing technique(s), the quantities of the resins used in this study were the same as that Would be required during the manufacturing of large-scale biocomposite components. A Single batch processed 500 g of resin (UPE +EMS). As discussed earlier, the bio-resin cionizent was 5 wt.% and clay content was 3 wt.%. Hence, for a single batch, 475 g of UPE, 25 g of bio-resin (EMS) and 15.464 g of nanoclay were used. The processing study reported herein is based on modifications to a baseline solvent based processing technique used by 0111- group to develop polymer-clay nanocomposites [6112]. The baseline technique (Process 4K) Vvas rrrodified with the aim of increasing the feasible amounts of bio-resin and nanoclay While also increasing the overall efficiency of the process, thereby leading to four novel 35 iiiauragnetic sti are the qiiantitj Ilr acetone + nan scarier probe f0 tied and bOIh m (UPE + E.\ 725 El 55 °C until a! approaches. A detailed description of the baseline process (Process A) is explained in the following. For all other processing techniques the modifications made to Process A are provided: M; This is the baseline processing technique. To a 2000 ml beaker equipped with a nngnetic stir bar, nanoclay (3 wt.%) and acetone (750 ml) were added. From trial studies, the quantity of acetone was determined to be 50 ml acetone per gram of nanoclay. The acetone + nanoclay solution was continuously stined, while it was sonicated using a sonication probe for the desired time or energy. After sonication, the ultrasonic probe was removed and both the UPE (475 g) and EMS (25 g) resins were added. The resulting mixture (UPE + EMS + acetone + nanoclay) was continuously stined while placed on a hot plate at 55 °C until approximately 75% of the acetone was removed (3 hours). The remaining acetone was removed by vacuum extraction at 55 °C for 24 hours. A drawback of this method is that during the acetone removal process, the styrene present in the UPE is also removed. The amount of lost styrene was re-introduced to the blend after completing the acetone removal process. The processed solution was cooled to room temperature and blended with the initiator and promoter followed by curing. The amounts of initiator and Pmrnoter, and the curing temperature are provided in Section 2.3.1. A flow chart explaining ProCess A is shown in Figure 2-1. At higher concentrations of EMS and nanoclay two problerm were encountered in Pro(:ess A: 1) the resin seemed to cross-link during the acetone removal process, and 2) Phase separation between EMS and UPE was observed, leading to instability at high vacuum DreSsures during the acetone extraction process. These problems were not observed at nanoclay loading less than 1.5 wt.% and EMS contents of less than 10°/o. Hence, Process B 36 rat“; to: bl m .i.» 1253 M5111 Pa? 35 I] {fit 5 mml'ec has B soled d E present is Eli ff 111311111 Oi bio-l The proble it mm ermtti :1 Sets of excess item Llr prop fibril to elimix M? P we run and blC evolved by making slight modifications to Process A to increase the nanoclay and bio-resin (EMS) content. W: In this process (Figure 2- 1) the bio-resin is not added until all of the acetone is removed. After acetone removal, the EMS is added along with the lost styrene. Process B solved the problems of phase separation and gelling during acetone removal and enabled incorporation of higher clay contents. Nonetheless, since only neat UPE is present during the acetone removal process, at high bio-resin and nanoclay contents, the amount of UPE present is relatively srmll, thereby making the resin system highly viscous and limiting the amount of bio-resin content that can be added. The problems associated with in bio-nanocomposite processing are mainly due to the solvent extraction step. The common solvent used in this work was acetone. In addition to issues of excessive energy and time required for solvent removal, any residual acetone hampers the properties of the resulting resin system Processes C and D were thus developed to eliminate the use of acetone. mLQ Process C (Figure 2-1) consists of sonicating the nanoclay directly in the UPE resin and bio-resin is added after sonication. Due to the viscosity of the UPE resin System, the temperature of the solution increases rapidly even at very low sonication enefgies. Thus, sonication energy is applied in steps of 20 k] and the resin system is allowed to Cool down to room temperature before the next step is applied. The process of cooling and applying energy in steps does not allow high sonication energies to be achieved by this pro<:ess. Additionally, at high clay contents, the increased viscosity of resin system can da-‘IJage the sonication equipment. As a result, Process D (Figure 2- 1) was developed to re(111cc the resin system viscosity. 37 ELL—5D: hi this p ”3311;15qu solution feel} issues of residual 1.55 sum is removed marinade hours re sob'em (Le, acetor stresses A and B. Nonetl it nanocomposites sutth film but the use c ilml'Ol such processes m Process El “Finite for removal of ' 9, . ml)I m atttone B RI "-3? Ltt mt usmg a , ML): In this process the resin system viscosity is reduced by adding 50 wt.% additional styrene solution to the system. Styrene is an inherent component of neat UPE and thereby issues of residual solvent in the resin system are eliminated. After sonication, the excess styrene is removed by mechanical stirring and application of constant heat (~55 °C) for approximately 6 hours. Processes C and D eliminate the use of vacuum to remove the foreign solvent (i.e., acetone), thereby improving the time and energy efficiency relative to processes A and B. Nonetheless, as shown later, it was found in microscopic observations that nanocomposites synthesized with acetone as solvent produced better dispersion and exfoliation, but the use of vacuum and heat in acetone removal was limiting overall efficiency of such processes. As a result, Process E was developed. m: Process E (Figure 2-1) was developed with the goal of eliminating the vacuum use for removal of acetone in processes A and B. Process E is similar to process B, except that acetone is removed only through heating on a hot plate while mechanically Stirring - i.e., not using a vacuum Acetone removal is monitored by measuring the weight of the sonicated nanoclay/ resin solution until no further decrease in weight is observed, ilid-icating total removal of solvent. Any lost styrene and the target bio-resin content is added at the end. In this work, after solvent removal, irrespective of the solvent, care was taken to Inaintain same processing conditions (amounts of promoter and initiator, curing tirrre etc.,) for all processes, thereby minimizing the effects introduced due to varying cure conditions. PkIlce, the properties of the resulting nanocomposites and the nanoclay morphology from various processes can be directly compared. 38 5} Processing ”an 25.1 Stern: Content i. The Ll’E used int all with EMS reduce seer-d that prenuture c real styrene content lit if Stall some content “are led to the level or. 151 sonicationEDC'gl It is commonly agn ti edition of the YES“ 2‘“: nanoclay Wicks 3““ item. Moreover, using rates title also achie‘ liaison energy, hvels o press-rig techniques SIUt Z)’ Pmcerrirzg ”new; 2.5.1 Styrene Content in Resin System The UPE used in this study had a styrene content of 33.5 wt.%. Partial replacement of UPE with EMS reduces the overall styrene content of the resulting resin system It is suspected that premature curing during the acetone removal step was due to the reduction in overall styrene content. Hence, the effect of styrene content was studied by: a) maintaining the overall styrene content at 33.5 wt.% for the blended resin system, and, b) keeping the styrene level to the level originally present in the UPE part of the blended system 2.5.2 Sonication Energy It is commonly agreed that increased sonication time, or energy, improves dispersion and exfoliation of the resulting nanocomposites. Nevertheless, excessive energy rmy break the nanoclay particles and reduce their aspect ratio, which reduces the reinforcement efficiency. Moreover, using optimum sonication energy would be essential to obtain desired Properties while also achieving cost, time and energy efficiency. To assess the influence of SOllicanon energy, levels of 60 k] and 300 k] were studied. Processes B and C were initial Pmeessing techniques studied with 60 k] of energy. Electron microscoPy observations sl”‘OVved that 60 k] of energy was insufficient for adequate dispersion of the nanoclay Particles. Hence processes D and B were sonicated onlywith 300 k] of energy. A sonication etlel‘gy of 300 k] was later used for process B, but was infeasible for process C due to mcl‘Eased viscosity of the resin systerrr, application of energy in steps and cooling after each Step. 39 53 Material Content In order to com; genes must have simil 3:33 of tarianons in W11. Neverthe m nanoclay and bior El nttlerate amount ml} and it s believed 2.5.3 Material Content and Nomenclature In order to compare the various processing techniques in this study, each of the processes must have similar amounts of nanoclay and bio-resin. Detailed presentation of the effects of variations in clay and bio-resin content is beyond the scope of this communication. Nevertheless, work by our group on characterizing nanocomposites with various nanoclay and bio-resin contents obtained from Process A have already been reported [2-6]. A moderate amount of clay content (3 wt.%) was selected for all resin systems in this study and it is believed that conclusions of the study can be extended to other clay concentrations. Similarly, all processes were done with 5 wt.% bio-resin content. This level was chosen to avoid the problem of phase separation, as it was suspected that this problem can initiate at EMS contents beyond 10 wt.%. The following nomenclature is used to define the various parameters in this study: “Pram ID/Ody Grime/Sunbeam Brag/Styrene Weijrt D.” The styrene weight ID indicates whether additional styrene was added to rmintain 33.5 wt.% styrene content in the total resin system If additional styrene was added, the Styrene Weight ID is denoted by “A”, else by “B.” For example, “B/ 3/ 300/ A” denotes a resin system obtained from Process B containing 3 wt.% of nanoclay processed with 300k] of sonication energy, and containing additional styrene to maintain styrene concentration in the total resin system at 33.5 wt.%. In Process C, due to high viscosity issues, the sonication energy of 60 k] was applied in two ways: a) steps of 20 k] and b) continuous application of 60 k]. An asterisk (*) is marked next to 60 k] to indicate continuous application of 60 k] of sonication energy. Since the nanoclay resin systems contain 5% EMS, corresponding baseline neat resins with no nanoclay were processed with and without additional styrene. For neat resins, the following nomenclature is used: “NEAT-95/5-A”. The term NEAT 40 mmfllflochy 5F Fire is term indicates the ,5 film?! A comparfion Ol amopic charaaerizali nest, it results are C la: hair totem Since tht reinitialize the effect tried from them Resul ill Overall Processin Table 2-1 proside: it; lhe overall efficienc thinning lacrors: proc nasty and cooling iss ’i‘liml The factors . “sling [5&1qu (Sea indicates no nanoclay is present and the resin system contains 95%UPE and 5% bio-resin. The last term indicates the styrene weight ID as explained earlier. 2. (2’ fiery/2:; A comparison of the various processing techniques, tensile test results and microscopic characterization of the nanocomposites are provided in this section. For each process, the results are classified with respect to the amount of sonication energy supplied to the resin system. Since the nanoclay and EMS content was the same for all processes, the results indicate the effect of the processing technique and the nanoclay morphologies obtained from them Results are also compared with the properties of neat resin systems. 2.6.1 Overall Processing Efl'iciency Table 2-1 provides a comparison of the processing techniques evaluated in this study. The overall efficiency of the processing system was determined bytaking into account the following factors: processing time, relative energy demand, phase separation problems, viscosity and cooling issues, and most importantly the resulting material (i.e., tensile properties). The factors influencing the time and relative energy demand depend on processing technique (Section 3). The ”relative energy demand” qualitatively compares the energy required in processing and includes the applied sonication energy and the energy applied (heat + stirring) during solvent removal. The relative energy demand parameter is qualitative and is intended for relative comparison of the various processes. Tensile properties of the resulting bio-nanocomposites are provided in following section. Overall, it was observed that processes B and D had good efficiency relative to other processes. 41 15.2 Tensile Propcfl The gunman c ration energies of me content on tensi it processing techniqt aspired separate ls in titties are comparec T915116 modulu 2.6.2 Tensile Properties The summary of tensile test results is provided in Table 2-2 and Table 2-3 for sonication energies of 60 k] and 300 k], respectively. Effects of sonication energy and styrene content on tensile properties are provided in this section. For relative comparison of the processing techniques, the resin systems in which styrene weight was maintained are compared separately from those in which styrene content was not maintained. The tensile properties are compared to those of neat UPE. I 412mg filed/m Tensile modulus results for the various processes are provided in Figure 2-2 and Figure 2-3 for sonication energies of 60 k] and 300 k], respectively. Process C used only 60 K] of applied energy due to viscosity issues. The asterisk (*) in process C in Figure 2-2 indicates that the sonication energy was applied continuously and not in steps. It was observed that on average, nanocomposites from resin systems with additional styrene provided higher tensile modulus by about 20 % relative to their counterpart resin systems with similar bio-resin and clay contents but no additional styrene. This is attributed to increased cross-linking provided by styrene and reduction of the overall bio-resin content. With respect to neat resins, it was observed that the addition of bio-resin (5 wt.%) reduced the tensile modulus by an average of 4 % for resin systems in which styrene weight was rmintained and by 13 °/o for resin systems with no additional styrene. Nanocomposites from process B had tensile modulus values that were, on average 38 % and 50 % higher than neat UPE resin for sonication energies of 60 k] and 300 k]. Similarly, Process D led to nanocorrrposites with average tensile modulus 16 % higher than neat UPE resin for 300 k] of sonication energy. Process C showed an average modulus increase of 15 % when 60 k] of sonication energy was applied continuously, but showed 42 355 which showed .' 3‘; kj of sonication er "C21 .5 5 him? 72’ tit/{L Crime tensile Edith 2-3 for son 2'3"“ led from all prt It whom attribute th the hhfih'eb' brittle ; ‘P‘afll Ciion [0 dm rm Increases, the 5 little change in modulus when the same energywas applied in steps. Process E was the only process which showed a decrease in tensile modulus of about 50 °/o relative to neat UPE at 300 k] of sonication energy. This is rminly attributed to the residual acetone in the resin system. WW Ultirmte tensile stress results for the various processes are provided in Figure 2-2 and Figure 2-3 for sonication energies of 60 k] and 300 k], respectively. Nanocomposites synthesized from all processes had lower ultirmte tensile stresses than the neat resin systems. The authors attribute this to the stress concentrations created by the nanoclay reinforcement in the relatively brittle polymer matrix which in turn lead to lower tensile strength and lower ductility. While there is no general consensus on this point of view, computational studies by a parallel effort to this work [2-29] have shown evidence to this mechanism. As nanoclay content increases, the stiffness and brittleness of the resulting resin systerm increases thus reducing the ultimate tensile strength and ductility [2-6]. Moreover, the addition of bio-resin also reduces the ultimate tensile strength. For example, neat bio-based resins (no nanoclay) showed a decrease of in ultimate tensile strengtln by approximately 5 % for resin systems in which styrene weight was maintained and a reduction of about 14 % for resin systems with no additional styrene. Nanocomposites from processes B and E had the lowest tensile strengths, approximately54 % and 7 O % lower than the neat UPE resin. This is attributed to the thermal degradation of the resin system due to prolonged exposure to heat during the acetone removal process. Nanocomposites synthesized with direct sonication processes C and D had tensile strengths greater than those rmde with processes B and E, yet with approximately 40 % and 44 % lower strength than neat UPE, respectively. 43 . '/ ,2, c,’ ' Cfl' ' 'n‘ff‘wfilt 1 4"“ I}: elongations at :an 24 and Figure 3 fleeing, the addition of m to make the $516 res: B had the lowesr d n; H 95, in ultimate tens i2: process C, the resin mirmteb 8 % highe iii in Steps. Thus, 3757'!!! Ships seems to W The elongations at failure for nanocomposites from various processes are provided in Figure 2-4 and Figure 2-5 for sonication energies of 60 k] and 300 k] respectively. For neat resins, the addition of EMS increases the ductility. Nevertheless, the additional styrene seems to make the system more brittle and reduces the ductility. Nanocomposites from process B had the lowest ductility, followed by process C, with an average reduction of 47 °/o and 44 %, in ultimate tensile strain respectively relative to neat UPE. For nanocomposites from process C, the resin systems in which sonication energy was applied continuously, had approximately 8 % higher failure strains than the resin systems processed with energy applied in steps. Thus, the altemating cooling and heating required to apply sonication energy in steps seems to affect the behavior of the resulting resin system. Nanocomposites from process E, which can lead to residual acetone showed poor stiffness properties but increased ductility rehtive to nanocomposites from other processes. In summary, the tensile modulus due to addition of nanoclay increased for all processes, except process E (attributed to residual acetone), with Process B showing the largest improvement. Processes that led to high tensile modulus had lower ductility and strengths. Processes C and D had a balance of stiffness and toughness properties. Increased sonication energy lead to higher tensile modulus, which is attributed to better dispersion and exfoliation of clay particles. While various processes lead to different properties for the resulting nanocomposites, they are all considered viable and stable. Thus, the processing techniques could be selected depending on the application and desired tensile properties. ‘I. 53 lhlorpholoiiy 0f N Figure 2'6 and summits {mm van rhinon relative to 0315‘ grinsng. Process B is St tilt the degree 0f ”dc m similar .‘iuxonposites from Pn m showed relatively t Fret Mb). This was grates C, nanocompos Sliced relatively bent Stallion energy appli ”titty ikith increase “‘51 apphirtg sonica 236% before the 5." A . “mp “1 LemPeratur it": ' - ‘31 s(mutton me . 0C0 ”is“ C (Fing 2- 2330:0 “33 mgu Stem 2.6.3 Morphology of Nanoclay Inclusions Figure 2-6 and Figure 2-7 show transmission electron micrographs for nanocomposites from various processes at 60 k] and 300 k] of energy, respectively. The degree of dispersion and exfoliation was found to increase with increasing sonication energy. Process B (Figure 2-6a and Figure 2-7a) led to the highest degree of dispersion and exfoliation relative to other processes. This is due to the use of solvent for dispersion during processing. Process B is similar to process E, except for the solvent removal technique. As a result, the degree of exfoliation and dispersion of nanocomposites from process E a W similar to that obtained from process B, and thus not provided. Nanocomposites from process C, in which nanoclay was directly sonicated in the neat UPE resin showed relatively poor dispersion and agglomeration of clay particles (encircled in Figure 2-6b). This was due to the high viscosity of the resin system. Nevertheless, for process C, nanocomposites obtained from continuous application of sonication energy showed relatively better dispersion than nanocomposites from similar process with sonication energy applied in steps (Figure 2-6c). This could be due to the decrease in viscosity with increase in temperature during sonication, thus enabling better dispersion. When applying sonication energy in steps the resin system is cooled back to room temperature before the next energy step is applied. This increases the system viscosity due to the drop in temperature and affects the dispersion of nanoclay particles. Process D was also a direct sonication method, but the resin system was diluted with styrene, thereby reducing the viscosity. Nanocomposites from process D showed better dispersion than those from process C (Figure 2-7b). Also, the degree of exfoliation was considerably improved in nanocomposites resulting from process D relative to those obtained from process C. It seerm that agglomerated clay particles resulting from process C (encircled in Figure 2-6) 45 game into pmfl tiniest?“ " 333535 B {Quoted $333111 aid) the med from ProCf mm and exit nrahed nanocla} 1&4 Fractograpl The roughn: grgerties and critic; imbued to bri mtonposites [3- hrnfinc the degr Wilmposites fro] L‘Ule imam surf.- fiefiafiom, Slmpl 33518111 “lib mix. it for SC k] and 3 PERIOD, n folio WOmlDOSites. I] improve into partially intercalated particles (Figure 2-7c) when using process D. Overall, TEM micrographs revealed better dispersion and exfoliation for the nanocomposites from process B followed by those from Process D. The enhanced nanoclay morphology is consistent with the relatively superior tensile modulus results of the nanocomposites obtained from processes B and D. The micrographs confirm a well known fact that good dispersion and exfoliation are beneficial for stiffness improvement (seen in B) while intercalated nanoclays are better for toughness improvement (seen in D). 2.6.4 Fractographic Observations The roughness of the fracture surface has generally been associated to fracture properties and critical strain energy release rates [2-28]. A smooth featureless fracture surface is attributed to brittle failures and rougher fracture surfaces are attributed to tougher nanocomposites [2-2712—28]. Tensile fracture surfaces were evaluated to qualitatively determine the degree of dispersion and the improved fracture properties of resulting nanocomposites from different processing methods. Figure 2-8a shows the relatively smooth tensile fracture surface of a neat resin with 5% EMS content. As discussed for the TEM observations, samples from process B had the best dispersion of clay particles. This is consistent with relatively rougher fracture surfaces as observed in Figure 2-8b and Figure 2-8e for 60 k] and 300 k] respectively. If higher sonication energy produces better nanoclay dispersion, it follows that it should also lead to rougher surfaces in the resulting nanocomposites. This is supported by images in (Figure 2-8c and Figure 2-8d) for nanocomposites from process C that show rorrgher fracture surfaces than neat UPE but smoother than those obtained from Process B. There was no significant difference in fracture surface roughness due to variations in the way that sonication energy was applied: 46 93.355; surfaces than all warrior: Processing pla}S :rxessng technique sht tr: shouli also be effic heard that the solven fr intent The use of a 3'Phh’ilts. Neverthele 311 COSI and energy. N writs of resulting n hr process E, which h. 5110 {the anry resin La 12 r8511] Syslfm, Tn tare 0f StiHntSS and d continuous (Figure 2-8d) and in steps (Figure 2-8c). Nanocomposites from process D had rougher surfaces than all other samples except those from process B. .7. 7 Bait/(rub): Processing plays a vital role in the resulting properties of nanocomposites. The processing technique should not only produce nanocomposites with enhanced properties, but should also be efficient with respect to energy, cost and time. In this study, it was observed that the solvent played an important role in determining the overall efficiency of the system. The use of acetone as a solvent produced better dispersion and exfoliation of clay particles. Nevertheless, extraction of solvent foreign to the resin system is inefficient in time, cost and energy. Moreover, the residual foreign solvent produces adverse effects on properties of resulting nanoconrposites. This was observed in nanocomposites obtained from process E, which had poor tensile properties. Therefore, a solvent that is an inherent part of the prirmry resin is prefened. In this study, the UPE resin contained 33.5 wt.% styrene and hence, styrene monomer was used to dilute the solution and sonicate directly into the resin system. The nanoconrposites produced by such process (D) showed a good balance of stiffness and ductility. The elimination of solvent was achieved in process C, which involved direct sonication of nanoclay in the resin system and no solvent extraction. It required the least amount of time and energy. Nevertheless, viscosity related issues, such as system cooling, application of sonication energy in steps, and excess load on the sonication probes hamper the feasibility of this process. Moreover, the desired degree of exfoliation may not be obtained due to the high viscosity and insufficient sonication energy. Overall, viscosity 47 issues, poor dispersion and exfoliation and moderate improvements in the resulting properties limit the efficiency of Process C Tensile testing revealed superior tensile modulus, relative to neat UPE, for nanocomposites from process B, followed by those from process D. Samples obtained from processes C had only a moderate increase in tensile properties relative to neat UPE. Process E samples had a reduction in tensile modulus and this was attributed to the residual acetone. Regarding ductility or elongations at failure, of the resulting nanocomposites, those from process B had the least ductility, while those from processes C, D and E had moderate levels of ductility. Similarly, samples from process B had the lowest tensile Strength, while those from process D had the highest. Overall, the effects of bio-resin and nanoclay complement each other, producing composites with a good stiffness-toughness balance [2-6]. Yet, in order to take the benefits of the hybridization of nanoclay and bio-resin blends, the processing technique must enable incorporation of high contents of nanoclay and bio-resin. The TEM micrographs showed that the nanocomposites from process B had better degrees of dispersion and exfoliation followed by those from process D. The nanocomposites obtained from direct sonication (process C), had poor dispersion and agglomerated clay particles. The samples from process E were similar to those from process B, but had poor tensile properties due to the effects of residual acetone. Similarly, the fractographic observations of tensile failure surfaces using SEM showed rougher surfaces in samples with well dispersed nanoclay. The increased surface area from the rough surface can be associated with the higher energy necessary for crack propagation. Since the tougher clay particles deviate the crack propagation front around them, thus creating new surfaces. Thus, proper dispersion of clay particles would lead to rougher surfaces. Consequently, if fracture 48 tire roughness slated though P1 Overall. [ht :tremg mnoc 13} deterrents while the properties. P. is." D used acetone 21.: the use of p i at»; afieas the Press D h ideal f0 mintnts saith a r m aPl’llthflOrls en mommies. surface roughness is indicative of better nanoclay dispersion, the best dispersion was obtained through process B followed by process D. Overall, the study indicated that processes B and D are desirable methods for fabricating nanoclay reinforced bio-based resins. Process B leads to the best tensile modulus enhancements while process D allows obtaining nanoconrposites with a good balance of tensile properties. Process D was more time and energy efficient than process B. Processes B and D used acetone and styrene as solvents, respectively. Solvent extraction is the major step limiting the use of process B as it is time and energy inefficient. Moreover, residual acetone, if any, affects the properties of the resulting nanoconrposites. Process B is ideal for laboratory scale quantities and in applications wherein high modulus is the principal criteria. Process D is ideal for producing larger quantities of resin systerm such as for large structural components with a reasonable balance of properties. Depending on the desired properties and applications either of the processes could be used to develop bio-based clay nanocomposites. 2. 6’ Cont/what Bio-based resin systems from partial substitution of petroleum-based resins (pnnnry) with natural polymers (secondary) provide environmental friendliness, cost-effectiveness and improved toughness. The drawbacls from the addition of bio-resins to the base polymer have been shown to be recoverable through nanoclay reinforcement. Solvent-based processing of nanoclay reinforced bio-based resin systerm brings about new challenges, such as phase separation, thennal degradation of the resin system and limitations on the rmximum feasible bio-resin and nanoclay content. This study was aimed at finding an efficient solvent-based processing technique for the synthesis of nanoclay reinforced bio- 49 based rain 555ml" 93mm energy. ; 73' he techriqdfs ' preemies of the it an after solvent 1 he polemer and ti may resin) as at reduced the tin weblarge amou 1Feetprocessing tirr Clerc}: Two PITX mm Process E hem and exfo mm 0f direct sc mess elimfiums Eh TEL? mflocompo Elm-mm 0f Stiff r :f' Of these pro EMIOmPOSiLeS. This ‘e‘l . i “hated ”Emile : .hhal Cmdmj. 3.11 mp afill/01- idmm Elmnpomes Can of based resin systems by assessing and comparing four evolving methods. The effects of sonication energy, processing time, solvent type and the associated processing issues for each of the techniques were assessed by evaluating the nanoclay morphology and the tensile properties of the resulting polymer nanocomposites. It was found that the addition of bio- resin after solvent removal avoids problerm associated with phase separation between the base polymer and the bio-resin additive. Direct sonication and use of styrene (inherent part of prirmry resin) as a solvent resolved the drawbacks of foreign solvent (acetone) residue and reduced the time required for solvent removal. Processes that enable incorporation of relatively large amount of bio-resin and nanoclay content with minimal processing problems, lower processing time and desirable tensile properties were considered to have good overall efficiency. Two processes, namely processes B and D, were found to have good overall efficiency. Process B consists of using acetone as a solvent and led to the best nanoclay dispersion and exfoliation, resulting in samples with high tensile modulus. Process D consists of direct sonication of nanoclay in the resin system diluted with styrene. This process eliminates the use of a foreign solvent, thereby reducing processing time, and the resulting nanocomposites showed a good balance of tensile properties, namely a balanced improvement of stiffness and toughness. Depending on desired properties and applications, either of these processes is deemed suitable for effective production of bio-based nanocomposites. This study used electron microscopy for characterizing qualitative features and evaluated tensile properties only. A more comprehensive thermo-mechanical and physical characterization, along with processing constraints, would clearly be needed to develop and/ or identify a processing technique that maximizes the benefits that bio-based nanocomposites can offer. 50 :9 7:11:42; Jl’h’f Precessing Pranerers .7. 9 Ykéée: arm/[2311785 Table 2-1. Comparison of processing techniques. Process ID Processing Parameters A B C D E Remy high high moderate/low moderate High Time high high low moderate moderate/ high Problems 3 yes no no no no Max. nanoclayb 1.5 5 3 5 5 Max. bio-resin b 10 20 30 30 20 Cooling need c no no yes none/ moderate no Vacuum use yes yes no no no Probe issues none none high moderate/ none none Tensile N/A ' modulus 1 modulus modul a: 1 modul ro rties high , ow us ow us P PC low strength strength balance Efficiency poor moderate/ good moderate good poor/ moderate 3‘ Problems of phase separation and curing during processing; b In wt.%; c High viscosity requires application of energy in steps and cooling to room temp. 51 03/6ch 0’3.-’6C*/B star UPE stares/M NEAT-9553 \ l: Tensih Modulu lifll‘erage, 0: SL- Tabh 2-3. Ex; \ Process ID B\ 1’13/330/ A h’3/ 33; / B D/ 3/ 333/ A D-""3/3Cr:/ B E"'3/333/13 E4"3/333/3 Limit Modules, 1 .5ng6, a: StarK Table 2-2. Experimental results for various processes at sonication energy of 60 k] E, GPa UIS, MPa EF, % Process ID i 0' '55 a' f 0' B/ 3/ 60/ A 4.88 1.06 20.17 3.12 0.74 0.25 B/ 3/ 60/ B 4.02 1.04 17.61 5.78 0.74 0.03 C/ 3/ 60/ B 3.50 0.26 22.70 3.33 0.73 0.16 C/ 3/ 60*/ B 4.05 0.51 26.93 2.80 0.79 0.06 NEAT UPE 3.53 0.43 43.86 1.03 1.41 0.01 NEAT-95/5-A 3.40 0.25 41.87 7.83 1.41 0.32 NEAT-95/5-B 3.08 0.81 37.52 7.91 1.59 0.39 E: Tensile Modulus, UIS: Ultimate Tensile Strength, EF: Elongation at Failure 1? : Average, 0': Standard Deviation Table 2-3. Experimental results for various processes at sonication energy of 300 k]. E, GPa UIS, MPa EF, % Process ID 7c 0‘ 3? 0' 35 0' B/ 3/ 300/ A 5.32 2.08 11.54 3.30 0.44 0.07 B/ 3/ 300/ B 4.95 1.32 9.94 321 0.64 0.22 D/ 3/ 300/ A 4.10 0.67 24.41 2.10 0.84 0.14 D/ 3/ 300/ B 3.42 0.14 26.82 5.62 1.14 0.11 E/ 3/ 300/ A 1.75 0.51 12.51 1.76 1.06 0.33 E/ 3/ 300/ B 1.16 0.23 9.05 2.89 1.29 0.36 E: Tensile Modulus, UPS: Ultimate Tensile Strength, EF: Elongation at Failure )7 : Average, 0': Standard deviation 52 .MMMMM. I It. M mmmm ADD-EDGE) an camera-.- 0 a wet-WH must m; an (m e m are ”ea" mm o-mm-.m cm Figure 2-1. Schematic description of various processing techniques. — Tensile Modulus L 50 6 - — UTS #50 5. ~40 4. 3_ -3o 2- -20 1- >10 0- -0 Processing Technique Tensile Modulus (GPa) Ultimate Tensile Strength (MPa) Bf3l80lA BI3I60IB CI3/60IB Cl3l60'lB NEAT UPE NEAT-95/5-A NEAT-95I5-B Figure 2-2. Tensile modulus and strength at sonication energy of 60 k]. figs: 522...... £28 22.2: m m m m w o mamfimfisz <.mfim.h10%) and nanoclay (>15 wt.%) were possible, b) to perform a detailed characterization of an array of nano-reinforced bio-based polymer systems to 94 oi} the effects of mm! combinauor lehiior. Moreover, rtr LPE with E.) nee-titre) [417] fou; light in EMSbaset this ShOuld perfop In this Work LPE/EMS nanocorzl We made. Hm, an Emfilled from a dc: 33i1°foEML and i com 0f nanoclaj m On the eliecr “limit lhfi ldemif 1| study the effects of constituents on the synergistic behavior, and c) to find optimized material combinations that have ease of processing along with enhanced synergistic behavior. Moreover, studies by Miyagawa et. al [4-514—17] on neat resin (no clay) blends, using UPE with EMS (eporddized methyl soyate) [+5] and EML (epoxidized methyl linseedate) [4-17] found that stiffness loss due to addition of functionalized vegetable oil was higher in EMS-based resin systems. This suggests that nanocomposites from UPE/EML blends should perform better than those from UPE/ EMS blends. In this work, the shortcomings and recommendations from the above-mentioned UPE/ EMS nanocomposites study [4-6] were taken into consideration and improvements were made. First, an EMI. bio-resin was used. Secondly, an improved manufacturing method identified from a detailed processing study [4-25], and which allowed incorporation of up to 30 wt.% EML and 5.0 wt.% was used. Since it was possible to incorporate relatively large amounts of nanoclay and EML contents, this study was also aimed at finding performance limits on the effects of EML and nanoclay on the resulting nanocomposites, thereby allowing the identification of optimized material combinations that would result in a balance of properties along with ease of processing. Experimental evaluation of tensile properties (modulus, strengths and failure strains), irrpact strengths, glass transition temperature (T9, linear coefficient of thermal expansion and moisture diffusivity properties was perfonned. The degree of nanoclay dispersion and morphology of the different bio-based nanocomposites was assessed using transmission electron microscopy. The effect of bio resin and nanoclay on the surface morphology of resulting tensile fracture was studied using seaming electron microscopy and is reported in an earlier work [4—6], and for brevity purposes not provided here. Results indicate synergistic behavior of bio- resin blends and 95 mach) Willi Cl (jot-balanced u U my»: Erperimer intonation minions. ln tl arrestingarep 43.1 Materia lire mai resin (UPE, Po: hissed modi “PM Up to motion tl 333% epoxj it 9% mfthyi 13. Claire 333° (5 a Who r m 2‘bmlnone mi“ Slitem to Bertram} Emma Welt Cl nanoclay with complete to partial recovery of mrost properties. Material combinations that show balanced multifunctional properties along with ease of processing were also identified. 43 Expenmenm/fl/etéoafl' Experimental determination of thermo-physical properties and material characterization were performed on bio-based polymer blends reinforced with nanoclay inclusions. In the following sections, details on materials, processing, parameters studied, and testing are provided. 4.3.1 Materials The main component of bio-based polymer systems was ortho unsaturated polyester resin (UPE, Polylite' 32570-00, Reichhold Inc., NC), which contains 33.5 wt.% styrene. A bio-based modifier, epoxidized methyl linseedate (EML, Vikoflex° 9010, Arkema Inc., PA) replaced up to 30 wt.% of UPE. EMI. is a mixture of methyl esters of fatty acid compositions that construct the linseed oil. The detailed composition is 40-50 wt. % methyl linolenate epoxy, 24-26 wt. % methyl oleate epoxy, 17-22 wt. % methyl linoleate epoxy, 4-7 wt. % methyl palmitate, and 2-5% wt.% methyl stearate. The nanoclay used in this work was Cloisite 30B° (Southem Clay Products, Inc., TX). The resin system (mixture of UPE, EML and nanoclay) was processed with cobalt naphthenate (Sigma-Aldrich, MO) as a promoter and 2-butanone peroxide (Sigma-Aldrich) as an initiator. A constant ratio by weight of the resin system to the promoter and initiator was utilized to cure all samples. The mixing ratio was 100 parts by weight of the resin system to 0.03 part promoter and 1.50 part initiator. Samples were cured at 100 °C for 2 h, followed by 160 °C for 2 h. 96 4.32 Elmer“cnta The mount TEE” resin COmPO‘ of four neat “Sin (n: :fiiorted With nano minted b)’ “Fling used to describe the 4.3.2 Experimental matrix and nomenclature The amount of bio-resin (epoxidized methyl linseedate, EML) that replaced the primary resin component UPE, was varied from 0% to 30%, in increments of 10°/o. A total of four neat resin (no clay) systems were obtained. Each of these resin systems were then reinforced with nanoclay inclusions of 2.5 wt.% and 5.0 wt.%. Twelve polymer systems were evaluated byvarying clay and EML contents as summarized in Table 4—1. The nomenclature used to describe the polymer systems is: “A/fl/C’ , where “1’ refers to the amount of UPE and “A" represents the amount of EML as a percentage of the resin system, and “ 6’ refers to the weight fraction of nanoclay inclusions. For example, Specimen ID 7 in Table 4-1 is referred to as 80/ 20/ 2.5, indicating that the resin system has 80 parts of UPE and 20 parts of EML and is reinforced with 2.5 wt.% of nanoclay platelets. The resin system corresponding to 0% EML content and 0% nanoclay corresponds to virgin UPE and is considered the baseline material system and was used for comparison. 4.3.3 Polymer nanocomposite processing The technique used for processing the nanoclay reinforced bio-based resin systems follows the findings from our group’s study on solvent-based processing techniques for bio- based clay/ polymer nanocomposites [4-25]. The technique found to be most efficient consists of sonicating the nanoclay in acetone to an energy level of 300 k], using a solution concentration of approximately 50 l of acetone to 1 kg of clay, while it is constantly stined After sonication, only the UPE solution is added. The acetone + nanoclay + UPE solution is mixed continuously on a hot plate at approximately 55 °C to remove a majority of the acetone. The remaining acetone is removed by vacuum extraction at 55 °C for 24 h. During the acetone removal process the styrene present in the UPE is also removed. Thus, after 97 mat “mold the 503m is cooled I foiimd bl' curing ‘A 45.4 Testing and Tench tests geedof s rum/min at study (fable t centsponding to 1b. 550, it was suspetl llieover, the rese statements using filled in the then imitation numb “melon above al 153mm] analysis mm temper-mm high“ the te Gilliam 0f theml Mamba 0f the L‘Sltd Per Cm Stud Moimmle a't Wm that a1 1' 'r man SO ._ acetone removal the bio-resin, EML, is added along with the lost styrene. The processed solution is cooled to room temperature and blended with the initiator and promoter followed by curing. A flow chart depicting the process is shown in Figure 4—1. 4.3.4 Testing and characterization Tensile tests were performed according to ASTM D638 standards with a testing speed of 5 mm/ min. Six tensile specimens were tested for each case of polymer system in the study (Table 4-1). Experimental thermal tests were performed only on key designs corresponding to the extreme bio-resin (0 and 30%) and nanoclay (0 and 5 wt.%) contents. Also, it was suspected that the phase separation starts beyond 10% bio-resin content. Moreover, the research in our group aims at manufacturing large scale load bearing components using 10% bio-resin content. Hence, samples with 10°/o EML were also included in the thermal studies. Overall, themnal tests were performed on six specimens with identification numbers 1, 2, 4, 9, 10 and 12 (Table 4-1). The linear coefficient of thermal expansion above and below the glass transition temperature was obtained by themno- mechanical analysis using a TA Instruments TMA 2940. The samples were heated from room temperature to 140 °C at a rate of 4°C/ min. Strain and temperature was measured throughout the test. The linear slope of strain-temperature curve is reported as the coefficient of thermal expansion. The glass transition temperature was obtained from the intersection of the two linear portions of the strain-temperature curve. Two specimens were tested per case studied. Moisture absorption testing was performed using a Fisher Scientific Versa bath°-138 equipment that allows imrrnersion of samples in a distilled water bath with temperature maintained at 50 °C. The nanocomposites samples used for moisture measurements were 98 mam bars and Wm“ were watt 3:7 link the diffusio a emun: flat and pa {Tmthmx llhfor the fun 3 1 titted. Moisture d plat for various blC mired Izod rests inclusions in resultir c.) ’7 (D Q— 97 O“ — rectangular bars and had average dimensions of 75.0 mm x 12.5 mm x 3.0 mm. The specimens were coated with impervious two-part epoxy on all edges to eliminate edge effects and limit the diffusion only through the thickness of the sample. All samples were polished to ensure flat and parallel surfaces. The polished samples were placed in a vacuum oven at 80 °C for 24 h to remove residual moisture. The weight of the samples was measured every 12 h for the first 3 days, followed by once every week until steady state (equilibrium) was achieved. Moisture diffusivity coefficients were obtained from the moisture-gain versus time plots for various bio-based nanocomposites. Impact strength was obtained by performing notched Izod tests as per ASTM D256. The dispersion and morphology of nanoclay inclusions in resulting polymer systems was assessed with transmission electron microscopy (TEM). Ultramicrotomy at room temperature was carried using a d'amond knife with an included angle of 4° to produce sections approximately 70 nm thick. A IEOL 100C'X TEM with LaB6 filament with 120 kV acceleration was used to obtain bright field images. 4. 4 [Pare/z; The polymer system comesponding to 0% EML and 0% nanoclay conesponds to virgin UPE and is considered the baseline material system. All results in the following sections are compared to this baseline. 4.4.1 Tensile tests and properties MW Tensile modulus for virgin (UPE) and bio-based polymer systems with varying clay contents are provided in Figure 4—2. A dashed horizontal line comesponding to the modulus of the baseline UPE is also shown to highlight the relative perfonrance of the nanoclay reinforced bio-based polymer systems. As expected the trend of average tensile modulus 99 m masts redL (M) The average End 33% EM]. 5 me in average 1‘ radii rt°lo nanoci 5 ad 55%, respecr Similarly, f0 hath 25 at% and? 1%ng by25°/o. F0 51% nanoclay Ed 1. The redunion in av 33% for near pOhr 33%.]131’5 indicatej he}; recover the 10 33% EML the av 55mm Effect c m prodded b}, absened L.“ The “hima b , ‘ Suing-h of Eh ‘73 ObSewed [ha] values suggests reduction in average tensile modulus values with addition of bio- resin (EML). The average reduction was approximately in the range of 10, 50 and 65% forthe 10, 20 and 30% EML systems, respectively. The addition of nanoclay inclusions revealed an increase in average tensile modulus values. For virgin UPE (no EML), the addition of 2.5 and 5.0 wt.% nanoclay revealed improvements in the average tensile modulus in the range of 25 and 55%, respectively. Similarly, for UPE/EML blends with 10% EML content, the addition of nanoclay (both 2.5 wt.% and 5.0 wt.%) seems to have improved the tensile modulus values on an average by 25%. For UPE/EML blends with 20% EML content, the addition of 2.5 and 5.0 wt.% nanoclay led to an approximate reduction of 20% in average tensile modulus values. The reduction in average tensile modulus value due to the addition of 20% EML was around 50% for meat polymers and the enhancement due to 5 wt.% nanoclay was in the range of 30%. This indicates a partial recovery and that the amount of nanoclay was insufficient to fully recover the lost properties due to the addition of EML. Similarly, for bio-blends with 30% EML the average improvement due to nanoclay addition was insignificant. The detrimental effect of bio-resin (EML) on stiffness at 30% EML content was larger than the gains provided by nanoclay reinforcement and hence no significant improvement was observed. l E . 'l l The ultimate tensile strengths (UTS) of the developed bio-based nanocomposites are plotted in Figure 4-3. The trend of average UTS values suggests that the addition of bio- resin (EML) reduces the ultimate tensile strength of the composite. This is attributed to the low strength of the bio-resin and the reduced cross-link density of the bio-blend polymer. It was observed that the addition of EML reduced the average UI‘S values by approximately 40 100 aim) for pebmc is observed that 112 The is attributed to agn UPE (no bio mitten in the aver till 13%, 23% and file) revealed a n merely. For E) teem, as the “it respect to th Wing the U18 Whl'seerm to and 60% for polymer systems with 20 and 30% EML content, respectively. Additionally, it ws observed that nanoclay inclusions also reduced the average ultimate tensile strengths. This is attributed to the embrittlement of the polymer due to the addition of nanoclay. For virgin UPE (no bio-resin), the addition nanoclay (both 2.5 wt.% and 5.0 wt.%) revealed a reduction in the average UTS values by approximately 50 to 60%. Similarly for resin systems with 10%, 20% and 30% EML contents, the addition of nanoclay (both 2.5 wt.% and 5.0 wt.%) revealed a reduction in average UTS values by approximately 55, 70 and 80%, respectively. For EML contents up to 10%, the reduction of UTS due to EML content was insignificant, as the average UTS values of 10% EML composites had onlya minor variation with respect to the neat UPE nanocomposites. EML plays a more significant role in reducing the U13 for its content is beyond 10%,. The combination of bio-resin and nanoclay seems to produce a detrimental effect that significantly reduces the UI'S of the polymer nanocomposite. mm The ultimate tensile strains (elongation at tensile test failure) for the tested bio-based nanocomposite systems are shown in Figure 4-4. The overall trend considering the average tensile failure strain values suggests that the addition of bio-resin makes the resulting composites more ductile and hence increased elongations at failure were observed. The average value of the ultimate tensile strain for the neat (no nanoclay) 20% EML polymer system was lower than its 10% EML counterpart. This deviation from the trend (increase in EML, increase in tensile strains) is attributed to the presence of shrinkage cracks in the test specimen, which occur while curing and were found to be prominent only at this concentration (80/20/0). Nonetheless, the overall trend indicates increase in the elongation at break due to addition of bio-resin. 101 The trend of average tensile strain values suggests that the addition of nanoclay reduced the elongations at failure. For virgin UPE (no EML), the addition of 2.5 and 5.0 wt.% nanoclay reduced the average failure elongations by a range of approximately 50 to 75%. For nanoclay reinforced UPE/EML blends (considering both 2.5 wt.% and 5.0 wt.%), the addition of nanoclay seems to reduce the average tensile failure strains by approximately 50 and 35% for 10% and 20% EML contents, respectively. For UPE/ EML/ clay blends with 30% EML content, the average tensile strain values were observed to be equal or better than the baseline virgin UPE composite, indicating complete recovery of the lost ducitility due to addition of nanoclay by addition of 30% EML. 4.4.2 Thermal properties The linear coefficient of themnal expansion (CIE) above and below the glass transition temperature (T; are provided in Figure 4-5 and Figure 4-6, respectively for the key polymer systems, namely specimen numbers: 1, 2, 4, 9, 10 and 12 (see Table 4-1). For structural applications, the CIE above Tg is not of much importance, as the load carrying capacity is considerably reduced past 7; As a result, only the CIE results below T8 are discussed next. Nonetheless, CIE values above and below Tg follow similar trends and hence the discussions of CIE trends below 7; are valid for CIE values above T: The trend of the average values of CIE below Tg suggests that the addition of bio— resin (EML) increased the average values of ‘CTE. The average CIE values below Tg increased by approximately 10 and 15% for polymer systems with 10 and 30 % EML, respectively. The addition of 5 wt.% clay to virgin UPE (no EML) seems to reduce the 102 Ihfi‘ffi Willi lo 3 minutely 13% l relation in thermal were pamlhrecove The variario earthy content is 1 . . gas nansrnon tern; imam (EML) cor in; ' TL&wn m the m: dim"Till-‘Ongis an and [+171 Addie WE (3% EML) a, I?“ i" The Pohmer 71hr W10 the. blend “Th “0 Ham. LPE. Overall, it se l0 .- ' ' m33.1011 of blO-r average CI'E value by approximately 20%. The addition of 5 wt.% clay to bio-based polymers with 10 and 30% EML content reveals average CIE values below T8 of approximately 10% higher than baseline UPE, respectively. Overall trends suggest that the reduction in thermal properties (represented by the increased CIE) due to addition of EML were partiallyrecovered by the addition of nanoclay. The variation of average Tg values for nanocomposites with varying EML and nanoclay content is shown in Figure 4—7. The T8 of virgin UPE was 98.5 °C The average glass transition temperature values were observed to decrease proportionally with increasing bio-resin (EML) content. The average Tg values for neat polymers (no clay) revealed average reduction in the range of 5—10% with the addition of 10 to 30% EML, respectively. The decrease of 7; is attributed to the reduction in cross-link density of the bio-based polymer blend [4-17]. Addition of 5 wt.% nanoclay revealed increase in average Tg values of both neat UPE (0% EML) and the 10% EML bio-based polymers, on an average by approximately 10°/o. The polymer system with 5 wt.% nanoclay and 30% EML content had an average Tg value similar to that of the baseline UPE. This is an improvement from similar neat resin blend with no nanoclay, which had approximately10% lower average Tg value than baseline UPE. Overall, it seems that the addition of 5 wt.% nanoclay helped recover the Tg lost due to addition of bio-resin (EML). 103 P. “if" u ' l t 1 a .db 4.4.3 Moisture absorption Moisture absorption properties for the nanoclay reinforced bio-based composites were obtained by water immersion tests where the weight gain at any given time (M) was measured until steady state was achieved. Due to thickness variations in the specimens the amount of moisture absorbed by each was different. Thus, instead of simply assessing the amount of moisture absorbed, the speed of moisture absorption in polymers and nanocomposites was quantitatively compared by determining their diffusion coefficient [4~ 26]. Hence, moisture diffusivity coefficient was used as the parameter to assess the effect of bio-resin (EML) and nanoclay content on the different bio-based polymer nanocomposites. The diffusivity coefficient, D, was computed from the initial slope of the moisture gain, Mt/M,Jo versustime («fl/d) as: 2 -1 Mi». D_16[ JE/d J (+1) whereMtis the mass gainat anytime LMmisthe maximummass gainat equilibrium/steady state, and dis the thickness of the specimen. Figure 4—8 shows the plots of moisture gain versus time for meat (no clay) polymer systems. The experimental data is shown in symbols and exponential fits (average R2 value = 0.98) are superimposed as solid lines. The diffusivity coefficients were obtained by substituting the initial slope from the curves in Figure 4—8 into Equation (4— 1). The variation of the diffusivity coefficients D, with increasing bio-resin (EML) content are shown as an inset in Figure 4—8. The diffusion coefficients indicate the effect of bio-resin (EML) content in virgin UPE. The diffusivity coefficient of virgin UPE was found to be 3.67x10'12 mZ/s. 104 ‘H r “H h“ at. . K 1.5%. I 'i )0 ill. "Fm-t kru‘vl 36125 W; N g =0 Pritchard [4-27] reports a value of 3.0x10'12 mz/s for ortho-unsaturated polyester. The slightly higher value obtained in this work may be due to the differences in the resins and test conditions. The average values of the diffusivity coefficients show an increase of approximately 70, 115 and 250% with the addition of 10, 20 and 30% EML. This is essentially a linear increase with EML content, as confirmed by a linear regression analysis with an R2 value of 0.96. The diffusivity coefficients of all the nanoclay-reinforced bio-based nanocomposites in this study are summarized in Figure 4-9. The study of average values of diffusion coefficients reveals that addition of nanoclay improved the banier properties, observed as a reduction in the average diffusion coefficient values. A reduction in average values of diffusion coefficients of approximately 35 and 20% was observed due to addition of 2.5 wt.% nanoclayto neat UPE and 10°/o EMU UPE blend, respectively. For resin system with 20% EML and 2.5 wt.% nanoclay, the average diffusivity coefficients were similar to baseline UPE composite. Similar to resin systems with 2.5 wt.% nanoclay, a reduction in average diffusivity coefficients of approximately 70, 45 and 30% was observed due to 5 wt.% nanoclay inclusions in polymer systems with 0, 10, and 20% EML content, respectively. The average diffusion coefficient values for polymer systems with 30% EML reinforced with 5 wt.% nanoclay were approximately 130% higher than the value for the baseline UPE composite. Nonetheless, this is an improvement of approximately 120% considering that the average diffusivity coefficient for the neat UPE/EML blend (no clay) with 30% EML was 250% more than that of the baseline UPE. Thus, the loss in barrier properties (increased diffusivity) due to the adding bio-resin (EML) was recovered by the addition of nanoclay for the bio-based polymer systems using up to 20% EML. For bio-based polymers with 30% EML the detrimental effect of the bio-resin on diffusion properties was greater than the 105 rs an. enhancement provided by nanoclay reinforcement and hence only partial recovery of the degraded propertywas possible. 4.4.4 Impact strength The relative comparison of toughness for the various bio-based polymer nanocomposites in this study was performed by notched Izod impact tests. The impact strength test results are shown inFigure 4—10. It can be observed that toughness of the nanocomposites increased with increasing EML content and decreased with increasing clay content. Research has shown that the addition of nanoclay inclusions can enhance the toughness of nanocomposites, but it depends on the morphology of the clay platelets [+28]. At low nanoclay contents well-exfoliated clay morphology leads to increased toughness properties; while at higher loading partially exfoliated and interacted clay morphologies have better toughness performance [4-28]. The overall trend considering only the average values of the impact strengths suggests that the addition of bio-resin increased the impact strength, and thereby the toughness, of the resulting polymer systems. It was observed that the average Izod impact strength values increased in the range of 2 - 10°/o for EML contents of 10 - 30%. Additionally, there seenn to be little improvement in impact strengths on bio-based polymers with 20 and 30% EML contents. The addition of nanoclay seerm to increase the stiffness and reduce the toughness (impact strength) of the polymer nancomposites. For virgin UPE (no EML), the average impact strengths seem to reduce by approxirmtely 5% with addition of nanoclay (both 2.5 and 5.0 wt%). In generaL it was expected that the average impact strength values of nanocomposites with 5.0 wt.% nanoclay would be lower than those with 2.5 wt.% loading, since the increase in nanoclay content generally found to 106 reduces toughness. Nevertheless, the average impact strength values of the 2.5 and 5.0 wt.% bio-based polymer/ clay nanocomposites were reasonably similar. This could be due to the nanoclay morphology. Since a similar processing technique (equal sonication energy) was used for all clay contents, increased clay concentration (in this case 5 wt.%) would result in more intercalated particles. It is commonly accepted that intercalated particles increase toughness [4—28]. As a result, the impact strength lost due to the increased clay content seems to be balanced by the enhancement due to the presence of intercalated galleries. Such a balance may not occur for all clay contents, but seems to coincide for the processing technique used in this study and for the specific nanoclay loading cases considered. Overall, the average impact strength of nanoclayLreinforced bio-based polymer systems was 4 and 2% lower than that of the baseline UPE for 10 and 20% EML content, respectively. The average impact strength of nanoclay reinforced bio-based polymers containing 30% EML was lower than its counterparts with 10 and 20% EML contents. A reason behind this result could be the processing method. Polymer systems with high bio-resin content have a relatively lower amount of UPE and higher clay content. This requires higher amount of solvent for processing, leading to prolonged exposure to heat for solvent removal. This may finally cause thermal degradation of the polymer and thus a decrease in strength. The improvement in average impact strength values due to addition of 20% EML in UPE was around 10% while addition of 5 wt.% clay in neat UPE reduced the average impact strength values by 5%. Thus, the combination of 20% EML and 5 wt.% nanoclay seems to provide a synergistic stiffness-toughness balance in the performance of the bio-based polymer nanocomposite. 107 4.4.5 Nanoclay dispersion and morphology Figure 4—11 shows the bright field TEM (transmission electron microscopy) micrographs of representative nanoclay bio-based reinforced polymer systems. Overall, the morphologies displayed in the images indicate excellent nanoclay dispersion with a combination of exfoliated and intercalated arrangement. As expected, nanocomposites with low nanoclay concentrations (2.5 wt.%) revealed a better degree of exfoliation (Figure 4-11a), while those with higher loading (5 wt.%) revealed an increased degree of intercahtion.( Figure 4-11 (1). It was also observed that the degree of intercalation seemed to increase with bio-resin content (compare Figure 4-11 a and Figure 4—11b). One of the reasons for such increased intercalation may be due to the lower amount of UPE available during the processing of bio-nanocomposites with high bio-resin contents, as bio-resin is added at the final step of process (Figure 4-1). The roughness of the fracture surface has generally been associated to fracture properties and critical strain energy release rates. A smooth featureless fracture surface is attributed to brittle failures, and rougher fracture surfaces are attributed to tougher nanocomposites [4-29]. Thus, the effect of bio-resin and nanoclay loading on fracture toughness can be supported by studying the fracture surfaces from the tensile tests through scanning electron microscopy (SEM) irmging. This has been shown and discussed by the authors in a study of bio-based polymer/ clay nancomposites using epoxidized methyl soyate (EMS) as the bio-resin [2-6]. Although that prior study used EMS instead of EML, the SEM images had sirrrilar features and are thus not repeated here. Nonetheless, the knowledge gained from that study is still relevant here. Overall, it has been shown [2-6] that fracture surfaces roughness increased with increasing bio-resin and increasing clay content. This suggests that the combination of bio-resin and clay will provide tougher composites. 108 rk \J‘ t“: 5W?! f». Nevertheless, further understanding of micro-structural parameters, such as crack propagation mechanisms and interface studies, is still required to relate fracture and toughness measures to surface morphologies. 4. )' Daria/55¢): Experimental characterization of bio-based nanocomposites obtained from blends of unsaturated polyester (UPE, primary petro-based resin) and epoxidized methyl linseedate (EML, secondary bio-resin), and reinforced with layered silicates (Cloisite 30BO , nanoclay) revealed a wide variety of improvements in multiple properties. As expected, the experimental data revealed scatter/ variations in measured parameters. These variations were specifically larger for failure-dependent pararrreters, such as tensile failure strains, tensile strengths and impact strengths. Detailed statistical analyses taking into account these variations should be performed to quantitatively and precisely obtain the effects of constituents, namely EMS and nanoclay. In this work, only the average values were considered and hence a qualitative effect and overall trends based on average values could be obtained. Overall, it was observed that the combination of bio-based resin and nanoclay result in synergistic behavior of constituents producing bio-based nanocomposites with properties similar or superior than the virgin UPE (0% EML, 0% nanoclay). An earlier study by the authors [2-6] could incorporate only 10% bio-resin and 1.5 wt.% nanoclay due to processing limitations. This study used an improved processing technique that enabled incorporation of high amounts of bio- resin (30% EML) and nanoclay (5 wt.%) content, thereby increasing the environmental appeal of the resulting composites. A test array of 12 material compositions allowed studying effects of individual constituents on various material properties and finding limits to the synergistic behavior offered by these materials. 109 .J. DYE Pic SW U," ...-'L. 4.5.1 Properties and synergistic behavior Direct tensile tests revealed a decrease in tensile modulus due to the addition of EML, and a recovery of lost stiffness with the addition of nanoclay. For polymer systems with EML contents up to 10 wt.%, the loss of stiffness due to EML addition was not only recovered but enhanced by nanoclay addition. Significant enhancement in tensile modulus was observed with nanoclay addition for resin systems with 20% EML. However, full recovery of the losses from EML addition was not possible. Finally, the detrimental effect of EML on stiffness properties at 30% EML contents was higher than the enhancement provided by nanoclay and hence limited improvement was observed at this high level of EML content. The ultirrrate tensile strength (UTS) results revealed a reduction in UIS due to addition of both EML and nanoclay. The reduction in UIS due to EML addition was found to be minimal for EML contents up to 10 wt.%, as the UIS results were found to be similar to neat UPE counterparts. This may be due to good crosslink density and absence of phase separation between UPE and EML. For EML contents higher than 10% a reduction in UIS was observed with increasing EML content. Additionally, nanoclay inclusions were found to reduce the UIS. Nanoclay addition improves the stiffness but also the brittleness of the resulting nanocomposites, thereby reducing toughness. The authors attribute this to the stress concentrations created by the stiff nanoclay sheets in the brittle polymer matrix, which in turn leads to lower tensile strength and lower ductility. While there is no general consensus on this point of view, computational studies bya parallel effort to this work [4-30] have shown evidence to this rnechanisrrr. 110 {“3 :«T F]. The ductility (tensile failure strains) and toughness of nanoclay reinforced bio-based composites increased with addition of bio-resin and reduced with increasing nanoclay content. This increase in ductility and toughness is due to the reduction in cross-link density in bio-based polymer, leading to increased plastic deformations [2-7]. Results of tensile strengths and failure strains showed a large variation. This may be due to several factors, including processing issues such as the effect of residual acetone, EML-UPE phase interaction, nanoclayLmatrix interaction, etc. Nonetheless, it was observed that the increase in brittleness due to nanoclay inclusions could be balanced bythe increase in ductility due to EML addition. Stiffness and toughness are opposing performance parameters and a proper balance is required for an efficient composite. The combination of bio-resin and nanoclay provides a synergistic effect producing bio-based nanocomposites with good stiffness- toughness balance. The advantage of hybrid combinations (UPE + EML + nanoclay) is not limited to mechanical performance, as similar enhancements were observed in thermal and moisture properties. The thermal and banier properties of the polymer systems in the study showed trends similar to the tensile elastic modulus results. The addition of EML increased the moisture diffusivity of the resuhing bio-based composite. However, results indicate that the adverse effects of EML blends on rrroisture absorption can be recovered through the enhanced barrier properties provided by the nanoclay inclusions in bio-based polymer system with up to 20% EML content. Similarly, a reduction in themral properties with the addition of EML was observed, by considering increased CIE values and a decrease in T3. In this case, the thermal property lost (represented by an increase in CIE) due to the addition of EML was partially recovered by the nanoclay loading. Similarly, with respect to glass transition temperature, the degraded thermal property (indicated by a reduction in 7;) 1 1 1 i. “L. EL hr. due to blending of 30% EML was essentially recovered (~2% lower relative to the baseline UPE) by the addition of 5.0 wt.% nanoclay. While the thermal properties (specifically CIE) of bio-based polymers where only partially recovered with nanoclay reinforcement; moisture diffusion properties showed complete recovery for EML contents up to 20%. This could be explained by considering the physical aspects of the transient phenomena. In the diffusion process the clay platelets are impervious barriers to perrneant molecules, thereby forcing a tortuous diffusion path that improves moisture barrier properties. Use of a similar analogyto CIE suggests thatthe expansion ofthe polymerdue to heatis higherthan restraining action provided by the nanoclay, which results in the nanoclay platelets moving along with the polymer under temperature expansion. Thus, although the nanoclay platelets provide resistance to thermal-induced strains, the effect is not as pronounced as its role in curtailing moisture diffusion. Hence only partial recovery in CIE was possible for the bio-based polymers. 4.5.2 Performance limits and optimized material design The test nntrix allowed studying the effect of EML and nanoclay on the resulting properties for EML and nanoclay contents up to 30 and 5 wt.%, respectively. Optimal nnterial combination(s) that result in a balance of multiple properties, along with ease in processing were thus identified. In general, nanoclay reinforcement had little or no effect on bio-based polymers with 30% EML content. This suggests that the detrimental effects of EML blending were beyond the recovery offered by the nanoclay. Complete recovery by nanoclay reinforcement was possible in most properties for bio-based polymers with EML contents up to 10°/o. Similarly, partial recovery of rrrost properties was possible bio-based polymers with EML contents of 20%. It should be noted that aforementioned comments are 112 made in a general sense. For instance, moisture diffusivity was fully recovered in 20% EML nanocomposites reinforced with 5 wt.% clay. Yet such recovery was not possible for other properties. Overall, bio-based nanoconrposites with up to 20% EML content and reinforced with 2.5% chy (90/10/25 and 80/20/25) were identified as optimized rmterial compositions as they show promise in their processing ease and balanced properties. Overall, results from the experimental characterization of UPE/ EMl/ clay nanocomposites revealed synergistic behavior of constituents with improvements in multiple properties; wherein the decrease in some properties due to the addition of bio-resin or clay was completely or partially recovered by the synergistic effect of the hybrid material system The effect of EML or nanoclay on the resulting thermo-mechanical properties was found to compliment the other, such that the detrimental effect on performance due an individual constituent was overcome by the enhancement provided by the other. Hence, the hypothesis of this work to develop eco-friendly bio-based nanocomposites by recovering the properties lost due to bio- resin addition by nanoclay inclusions was found to be feasible, efficient and promising. 4! 6 Carri/51'0”: Results from this study indicate that synergistic behavior of material constituents can be used to produce novel bio-based nanocomposites with properties similar or better than the virgin base petroleum-based polymer. In this work, blends of unsaturated polyester (UPE, petro-resin) and epoxidized methyl linseedate (EML, bio-resin) were reinforced with nanoclay inclusions. Improved processing enabled incorporation of up to 5 wt.% nanoclay and 30% EML in virgin UPE, thereby considerably improving the environmental appeal of the nanocomposite. Detailed experimental characterization and the effect of rmterial 113 constituent concentrations on the resulting properties were studied. Material combinations that produced overall balanced or enhanced properties, including proper stiffness-toughness balance, were identified. The degradation of thenno-mechanical properties (e.g., stiffness, themml, diffusion) due to addition of EML were completely and partially recovered for EML contents of 10 and 20%, respectively; indicating excellent synergistic behavior of the constituents. However, such synergy was not observed for bio-based nanocomrposites with 30% EML content, as they showed little or no improvement in most properties due to nanoclay reinforcement. This indicates that the adverse effects of EML were beyond the enhancements offered by nanoclay reinforcement. Including ease of processing along with the resulting balanced or enhanced properties, bio-based polymers with EML contents of 10 and 20% reinforced with 2.5 wt.% nanoclay were identified as optimal designs. The synergistic behavior of the showcased bio-based polymer nanocomposites shows promise for optimized high-performance materhls with an environmentally conscious appeal. 114 4 7 711540: 4714/ Egg/ow Table 4-1. Experimental matrix showing specimen identification numbers and variation in clay and EMS contents Amount of EML replacing UPE in Clay UPE-EML blend. (as % of UPE) Content (wt.%) 0 10 20 30 0.0 1 2 3 4 2.5 5 6 7 8 5.0 9 10 11 12 115 1 S NNICATION l 1leth Organoclay + Acetone UPE + S onnicate d + Energy: 300“ Clay / Acetone Solution I J l + ENE w ADDITION A E 0 EM VAL Add bio-resin (EML) & Heat & vaccqm lost amount of Styrene extraction @ 55 C for 24 hours E- RE MIXIN 9! !BE fl Cool, Mix with promoter , _ Process blended at and Initiator curing temperature J Figure 4—1. Processing of nanoclay reinforced bio—based (U PE/ EML) resins. 0.0 wt.% Nanoclay 7 , - 2.5 wt.% Nanoclay m 5.0 wt.% Nanoclay Baseline UPE Tensile Modulus (GPa) h 0 1o 20 30 Bio-resin [ EML] content (%) Figure 4-2. Experimental tensile modulus of bio—based polymer systems with varying nanoclay content. 116 50 : -O- 0.0 wt.% Nanoclay A -- + 2.5 wt.% Nanoclay g ' -I- 5.0 wt.% Nanoclay :5 ‘“" g $— : _‘ ~- g 30 . U) 3 03 . 5 2° 1 p. 3 g . E 10 J D l o I l I I 0 10 20 30 Bio-resin [ EML ] content Figure 4-3. Experimental ultimate tensile stresses for various bio-based polymer/ clay nanocomposites. 3.5 .‘ -O- 0.0 wt.% Nanoclay I -V— 2.5 wt.% Nanoclay 3 0 I -I- 5.0 wt.% Nanoclay arsfi 2415 1:5{ Elongation at Failure (%) 1rr{ .a- 1 J °°5'- — orri . . . . 0 1o 20 30 BIO-resin [ EML] content Figure 4-4. Experimental tensile test elongations at failure for various bio-based polymer/ clay composites. 117 130 g) ' - 0.0 wt.% Nanoclay \ : 5.0 wt.% Nanoclay 5:, 120 -. c . .9 j 2 110 - a a a. . X _ I-l-l . TV 100 - E I L- o . is . .._ 90 - o . a c u 2 . o 80 - E . o o u o . 70 - 1O 30 Bio-resln [ EML ] content (%) Figure 4-5. Variation of CTE below T g with varying bio-resin (EML) and nanoclay content. 200 ‘ - _ 0.0 wt.% Nanoclay 190 I 5.0 wt.% Nanoclay Coefflclent of Thermal Expansion (um I °C) 0 1 0 30 Bio-resin [ EML] content (%) Figure 4—6. Variation of CTE above Tg with varying bio-resin (EML) and nanoclay content. 118 120 : - 0.0 wt.% Nanoclay . 5.0 wt.% Nanoclay 90- 803 Glass Transition Temperature ( °C) 70- 0 10 30 Bio-resin [ EML ] content (%) Figure 4-7. Variation of glass transition temperature (T5) with varying EML and nanoclay content. 1 0 ' . ‘ u-v v o 0.8 - . ' 12 — 1001010 M 0-6 ‘_ — 9011010 1 _ ‘ V ' 2 9 — 80I20I0 Mm . "E — 7013010 0.4 - '23 6 ' x u o 3 - 0.2 - -0- 1001010 - + 9011010 0- -I— 80/20/0 0 1o 20 3o 0 o - + 70’30’0 EML content (%) o 100 200 300 400 500 600 700 800 fi/d , («E/mm) Figure 4-8. Moisture diffusivity of neat resins (no clay). 119 14 : — 0.0 wt.% Nanoclay I - 2.5 wt.% Nanoclay 12 I m 5.0 wt.% Nanoclay 10 { Baseline D, Diffusivity Coefficient, x 10'12 mzls 0 10 20 30 Bio-resin [EML] content (%) Figure 4-9. Experimental diffusivity coefficients for all bio-based polymer/ clay nanocomposites in study. 15 I — 0.0 wt.% Nanoclay I - 2.5 wt.% Nanoclay m 5.0 wt.% Nanoclay Izod Impact Strengths (Jlm) 10 20 Bio-resin [ EML] content (%) Figure 4-10. Variation of Izod Impact strength with varying bio—resin (EML) and nanoclay concentration. 120 Figure 4-11. TEM micrographs showing degree of dispersion and morphology: a) 100/0/2.5- well dispersed, partially exfoliated and intercalated (scale = 1pm), b)70/30/2.5 — well dispersed, but higher degree of intercalation relative to 100/0/2.5 (scale = 1pm), c) 100/ 0/ 5 — Well dispersed, partially exfoliated and intercalated (scale = 211m), and d) 100/0/5.0 - high magnification of an intercalated gallery (scale = 50 nm). 121 4 (5’ X‘s/972,163; [4-1]. Uyarm H, Kuwabara M, Tsujimoto T, Nakano M, Usuki A, Kobayashi S. Green nanocomposites from renewable resources: Plant oil-clay hybrid materials. Chem Nhter 2003; 152492-2494. [4-2]. Mohanty AK, Misra M, Himichsen G. Biofibers, Biodegradable polymers and biocomrposites: an overview. Macromol Mater Eng 2000; 276/ 2771 24. [4—3]. Burguefio R, Quagliata MJ, Mohanty AK, Mehta GM, Drzal LT, Misra M Load bearing natural fiber composite cellular beams and panels. Composites Part A 2004; 35:645-656. [4—4]. Nickel J, Riedel U. Activities in biocomrposites. Mater Today 2003; 6:44-48. [4-5]. Miyagawa H, Mohanty AK, Burgueno R, Drzal LT, Misra M Novel bio-based resins from blends of functionalized soyabean oil and unsaturated polyester resin. J Polym Sci, Part B: Polym Phys 2007; 45:698-704. [4—6]. Haq M, Burgueno R, Mohanty AK, Misra M. Bio-based unsaturated polyester/ layered silicate nanocomposites: characterization and thermo-physical properties. Composites Part A 2009; 40:540-547. [4—7]. Rama D. Mechanical properties and morphology of epoxidized soyabean or]- modified epoxy resin. Polym Int 2001; 50:179-184. [4— 8]. Pinnavaia -TJ, Beall GW. Polymer-Clay Nanocomrposites, John Wiley 8?. Sons, Ltd., 2001. [4—9]. Liu Z, Erhan SZ, Xu J. Preparation, characterization and mechanical properties of epoxidized soybean oil/ clay nanoconrposites. Polymer 2005; 46:101 19- 10127 . [4-10]. Song B, Chen W, Liu Z, Erhan S. Compressive properties of epoxidized soybean oil/ clay nanocomposites. IntJ Plast 2006; 22:1549- 1568. [4-11]. Lu B, Laka RC Bio-Based nanocomposites from corn oil and functionalized organoclay prepared by cationic polymerization. Macromol Mater Eng 2007; 292:863-872. [4—12]. Alexandre M, Dubois P. Polymer-layered silicate nancomposites: preparation, properties and uses of a new class of materials. Mater Sci Eng R 2000; 28:1-63. [4- 13]. Le Baron PC, Wang Z, Pinnavaia, TJ. Polymer- layered silicate nanocomposites: an overview. Appl Clay Sci 1999; 15:11-29. [4- 14]. Hussain F, Hojjati M, Okamoto M, Gorga RE. Polymer-matrix Nanocomposites, Processing, Manufacturing and Application: An Overview. J Comp Mats 2006; 17. 1511- 1575. 122 16] '1 I L‘ a 1. 1‘4 l1 '4 5“ L720} II): Ls- [4— 15]. Miyagawa H, Mohanty AK, Burgueno R, Drzal LT, Misra M Characterization and thermophysical properties of unsaturated polyester-layered silicate nanocomrposites. J Nanosci Nanotechnol 2006; 6:464—471. [4-16]. Bakar M, Djaider F. Effect of plasticizers content on the mechanical properties of unsaturated polyester resin.J Thermoplast Compos Mater 2007; 20:53-64. [4-17]. Miyagawa H, Mohanty AK, Burguefio R, Drzal LT, Misra M Development of biobased unsaturated polyester containing functionalized vegetable oils. Ind Eng Chem Res. 2006; 45:1014- 1018. [4-18]. Mohanty AK, Miyagawa H, Burguefio R, Nfisra M Biobased nanocomrposites from organoclay and blends of unsaturated polyester and functionalized vegetable oil. ANI'EC, Conference Proceedings. 2005; 4: 52-56. [4— 19]. Ratna D, Banthia AK. Epoxidized soybean oil toughened epoxy adhesive. J Adhes Sci Technol 2000; 14:15-25. [4-20]. Miyagawa H, Mohanty A, Misra M, Drzal LT. Thermo-physical and impact properties of epoxy containing eporn'dized linseed oil, 1: Anhydride cured epoxy. Macromrol Mater Eng 2004; 289 6:2-9 635. [4—21]. Miyagawa H, Mohanty A, Misra M, Drzal LT. Themno-physical and impact properties of epoxy containing epoxidized linseed oil, 2: Amine cured epoxy. Macromol bilater Eng 2004; 289:636-641. [4-22]. Haq M, Burguefio R, Mohanty AK, Misra M Hybrid bio-based composites from blends of unsaturated polyester and soybean oil reinforced with nanoclay and natural fibers. Compos Sci Technol 2008; 68(15-16):3344-3351. [4-23]. Shabeer A, Chandrashekhara K, Schuman T, Phan N, Nguyen T. Mechanical, thermal and flammability properties of pultruded soy -based nanocomposites. SAMPE, Conference Proceedings. 2007; 52:71/ 1-71/ 12. [4—24]. Shabeer A, Sundaraman S, Cnandrashekhara K, Dharani LR. Physicochemical properties and fracture behavior of soyLbased resin. J Appl Polym Sci 2007; 105(2):656-663. [4—25]. Haq M, Burguefio R, Mohanty AK, Misra M Processing Techniques for Bio-based Unsaturated-Polyester/ Clay Nanocomposites: Tensile Properties, Efficiency, and Limits. Composites - Part A 2009; 40:394—403. [4-26]. Vlasveld DPN, Groenewold J, Bersee HEN, Picken SJ. Moisture absorption in polyamide-6 silicate nanocomrposites and rts influence on mechanical properties. Polymer 2005; 46. 12567-12576. [4—27]. Pritchard, G., Reinforced Plastics Durability. CRC Press, FL, USA. 1999. 123 [4-28]. Nair SV, Goettler LA, Lysek, B.A. Toughness of nanoscale and multiscale polyarnide -6,6 composites. Polym Eng Sci. 2002; 42:1872-1882. [4—29]. Wang L, Wang K, Oren L, Zhang Y, He C. Preparation, morphology and thermal/mechanical properties of epoxy/ clay nanoclay composite. Composites PartA. 2006; 37: 1890-1896. [4—30]. Haq M, Burguefio R. Modeling and simulation of bio-based polymer/ clay nanocomrposites through a multilevel FE approach. In: Proceedings of Sixth Intemational conference on Computation of Shell 8: Spatial Structures. New York, May, 2008. SessionT-Z-E. 124 i] 9 a L rIi: . alumni-la i2 Chapter 5. Multiscale Hybrid Biocorrrposites: I - HF / UPE / EMS 1 I. I Abner! Hybrid bio-based composites that exploit the synergy between natural fibers (industrial hemp) in a nano-reinforced bio-based polymer can lead to improved properties while maintaining environmental appeal. Bio-based resins obtained by partial substitution of unsaturated polyester (UPE) with epoxidized soybean oil (EMS) increase toughness but compromise stiffness and hygro-thermal properties. Reinforcement of the bio-based resin with nanoclays permits to retain stiffness without sacrificing toughness, while also improving barrier and thermal properties. Characterization of different hybrid composites verified this synergistic behavior in which systems with 10 % EMS and 1.5 wt. % nanoclay retained the original stiffness, strain to failure, and hygrothennal properties of the original resin while improving toughness. Optimum designs that maximize the synergy of the constituents are thus possible and the presented results provide an initial benchmark to identify such balance, and thus increase the potential applications of bio-based comrposites. )? 2 lflmdycabrz Environmental concemrs related to the use of synthetic, or petroleum-based, polymer matrix composites has propelled the development of comrposite materials based on manual or renewable sources [5-115-2]. Biocomposites, composed of natural fibers in synthetic or natural polymer matrices have recently gained mruch attention due to their low cost, environmental friendliness, and their potential to compete with synthetic composites [5-3]- ‘ Haq, M, Burguer'io, R, Mohanty, .ur, and Misra, M., “Hybrid Bio-based Composites from blends of Unsaturated Polyester and Soy Bean Oil Reinforced with Nanoclay and Natural Fibers,” Composites Science and Technology. zoos; 68:3344-3351 125 [5-7]. Nonetheless, the use of bio-based composites has been limited due to their lower mechanical and themno-physical properties compared to synthetic composites and conventional structural materials [5-615-7]. A promising compromise between environmental friendliness and perforrrance are bio-based resins, or bio-blends, obtained by replacing part of a petroleum-based resin with natural bio-resin. In addition to higher natural content, bio—based resins improve toughness of the resulting resin blend [5-815-9]. However, this increase in toughness compromises stiffness, barrier and themal properties [5- 815-9]. Stiffness and toughness are opposing performance parameters and a proper balance is required for an efficient composite. One wayto attain this balance is the addition of layered silicates, or nanoclays. Polymers reinforced with nanoclays have been shown to exhibit enhancements in mechanical, thermal and barrier properties at low concentrations [5- 10]. Nanoclays consist of stacls of sheet like silica platelets with thickness of ~1 nm and extremely large surface areas and aspect ratios. The enhancement of mechanical and barrier properties in polymers with addition of small concentrations of nanoclays is well reported, and a good review on polymer clay hybrid nanocomrposites is provided by Le Baron et al. [5-10]. The reinforcement of bio-based polymers with nanoclays has been shown to recover the decrease in stiffness, themnal and barrier properties due to the addition of bio-resin and can lead to improved stiffness-toughness balance [5-1115-12]. Thus, the synergy between nanoclay reinforcement in a bio-blend produces a polymer nanocomposite with similar or enhanced properties than the virgin polymer and holds great promise for use in wide applications. The true value of layered silicate nanocomrposites is, however, not solely the enhancement of the neat resin but rather the value-added properties it provides to a fiber- reinforced composite. Thus, in a biocomposite, natural fibers must remain as the main 126 lu- ~ ' 1, at: OI fl] 1- a, H“ reinforcement for stiffness and strength. Natural fibers such as flax, industrial hemp, jute and kenaf have been found to have specific strengths comparable to E-glass; and the elastic modulus and specific modulus of manual fiber composites has been found to be comparable or even superior to E-glass composites [5- 1]. The incorporation of nanoclays and natural fibers in resin systems thus provides reinforcements to resin systems at two scales. The nanoclay enhances the bio-based polymer system in stiffness and hygrothennal properties, while the natural fibers provide the main stiffness and strength. In addition, the enhanced barrier properties of the mano-reirrforced resin retard moisture from reaching the natural fibers, thereby providing a synergetic effect between scales for an efficient bio-based composite. Similar mrultiscale reinforcement concepts have been studied for synthetic fibers to enhance mechanical and electrical properties [5-13}[5-16]. Tsai and Wu [5-13] reinforced glass/ epoxy composites with nanoclay and found a reduction in longitudinal tensile strengths and mode I fracture toughness with increasing nanoclay content. Kinloch et al. [5- 14] report improvement in the interlarninar toughness of a carbon-fiber composite with multiphase toughened epoxy matrix. The epoxy matrix was reinforced with nanosilica and microsized nrbber particles. Kirrloch et al. [5-15] report that this multiphase toughened epoxy matrix provides increase in toughness without significant loss in mrodulus and thermal properties. Gojny et al. [516] reinforced the epoxy matrix of glass fiber composites with carbon black and carbon nanotubes and report improvements in electrical conductivity and interlaminar shear strength. Thus, while different concepts of multiscale reinforcement have been studied for synthetic composites, the studyof multiscale/ multiphase reinforcements on bio-based composites is limited. This work provides an overview of a study aimed at developing and characterizing novel multiscale hybrid biocomposites from bio-based polymer blends of unsaturated 127 polyester (UPE) and epoxidized methyl soyate (EMS) reinforced with nanoclay and chopped industrial (unprocessed) hemp. An optimum or synergistic balance of the various constituents in the resulting hybrid biocomposites is essential to achieve increased environmental friendliness and proper stiffness-toughness balance, along with the other tailorable features that they offer, such as enhanced and/ or controlled moisture and themal properties. This work aims at demonstrating that this synergistic balance can be achieved. Material systems that varied the bio- resin (EMS) content in UPE and the concentration of nanoclay while maintaining a constant amount of natural fibers were fabricated by compression molding. Mechanical and thermophysical properties were evaluated through standard tests. The degree of dispersion and exfoliation of nanoclay in the resins systems was characterized using transmission electron microscopy and the features of fracture surfaces were characterized using scanning electron microscopy. The results provide a benchmark to identify the synergistic balance of the constituents such that efficient, novel and multifunctional hybrid biocomposites can be designed. )1] ExpenheflM/metéodr The developed hybrid bio-based composites were compared in terms of basic thermo-physical properties and qualitative material distribution features. Details on materials, processing, parameters studied and testing are provided in this section. 5.3.1 Materials Industrial unprocessed raw hemp (Stemrergy Inc, London, Ontario, Canada) was used as natural fiber reinforcement. The fibers were hand chopped to an average length of 25 mm. The primary component of the resin blend was ortho unsaturated polyester resin (UPE , Polylite' 32570-00, Reichhold Inc., NC), which contains 33.5 wt. % styrene. The secondary 128 component that replaced parts of the UPE was a bio-based modifier: epoxidized methyl soyate (EMS, Vikoflex' 7010, Arkema Inc, PA). EMS is a mixture of methyl ester of fatty acid compositions that construct soybean oil. The detailed composition is 5-11% methyl epoxy linolenate, 43-56% methyl epoxy linoleate, 24-26% methyl epoxy oleate, and 7-11% methyl epoxy palmirate. The nanoclay used to reinforce the resin system was Cloisite 30B° (Southem Clay Products, Inc. TX). The nanoclay was chosen by trial and emor to be compatible with EMS and no detailed study on surfaces and interfaces of the constituents was perfomned. Compatibility is assumed to be due to the carbonyl groups of UPE/ EMS and free hydroxyl group of organic modifier part (methyl tallow bis-Z-hydroxyethyl quaternary ammonium) in the organoclay Cloisite 30B' .The resin blend was processed with cobalt naphthenate (Sigma Aldrich, MO) as a promoter and 2-butanone peroxide (Sigma Aldrich) as an initiator. A constant mirdng ratio of 0.03 parts promoter and 1.50 parts initiator to 100 parts of resin system was used for all resin systems. Table 1 provides the nomenclature, composition, concentration of constituents and densities of the five biocomposite systems studied. Biocomposite A, rmamufactured of only virgin UPE (no EMS and mo nanoclay) is considered the benchmark biocomposite and hence, the performance of all biocomposites is compared with this benchmark composite. 5.3.2 Nanocomposite processing The resin system (UPE +EMS +nanochy) had varying amounts of nanoclay and EMS content that replaced UPE. For nanoclay reinforced resin systems, the processing technique is similar to that reported in references [5-1115—12]. Nanoclaywas sonicated in acetone using a solution concentration of approximately 50 liters of acetone to 1 kilogram of clay, while it was constantly stirred by a magnetic stirrer. The amount of energy spent on sonication was 129 maintained at 30 kJ for all resin systems. The UPE and EMS were then added to the acetone-clay solution and mixed with a magnetic stiner on a hot plate for 3 to 4 hours to remove the majority of acetone. The remaining acetone was removed by vacuum extraction at approximately 55 °C for 12-24 hours. During the acetone removal process, along with acetone, the styrene present in resin system was also removed. The lost styrene was thus added after the acetone removal process. During vacuum extraction of the acetone, and at higher concentrations of EMS and clay, processing issues such as phase separation in UPE/ EMS blend and cross linking were observed. The above mentioned problems were not observed at nanoclay contents of 1.5 wt. % or less and bio-blends of 10% EMS or less. Hence, nanoclay reinforced resins with 20 % EMS was not feasible in this study. Studies by our group on bio-based nanocomposite processing have shown that the aforementioned issues can be overcome by modifications in the above process [5-17]. The styrene content in the UPE used in this studywas 33.5 wt. %. The replacement of parts of UPE with bio-resins reduced the overall styrene content of the resulting resin blend. It was suspected that premature curing of the resin blend during acetone removal was due to the reduced styrene content. Hence, styrene was added to the bio-based blend to maintain the overall styrene content at 33.5 wt. % for the resin system. The processed polymer nanocomposite (UPE + EMS + nanoclay) was used as the matrix for natural ffiaers in hybrid biocomposite material systems. 5.3.3 Manufacturing of hybrid biocomposite plates Flat biocomposite plates for five material systems described in Table 5-1 were manufactured using compression molding. Before use, the fibers were dried in vacuum oven at 80 °C and 100 kPa of pressure for 12 hours. The nano-reinforced resin system (UPE + 130 B! for if, irreverent Id! ‘VI‘IOEFS ’liJ ..1_ al.: EMS + nanoclay) was mixed with the promoter and initiator. The fibers were then impregnated with the resin system by hand until the material was consistent. The impregnated fibers were then placed in a frame mold. Care was taken to evenly distribute the fiber material in the mold to ensure a uniform sample, since natural fibers have a tendencyto clump and tangle together. The frame mold was placed between two steel plates covered with Teflon paper. The sample was then cured in a press under 550 kPa of pressure at 100 °C for 2 hours, followed by 150 °C for 2 hours. )3 4 Drab? 6- cédracth'zm‘z'wz Testing and characterization of the hybrid bio-based composites was performed on samples taken from the compression molded plates. The tensile properties (modulus, strength and elongation at failure) were measured on six to eight samples as per ASTM D638, while impact strength was measured on five samples as per ASTM D256. The coefficient of thermal expansion was measured on two to three samples by themno- mechanical analysis using TA instruments TMA 2940 analyzer. The specimens were heated from room temperature to 140 °C at a rate of 4 °C/ minute. The strain and temperature of the samples were measured throughout the test. The coefficient of thermal expansion was taken as the linear slope of the strain-temperature curve. The glass transition temperature was obtained from the intersection of the two linear portions of the strain-temperature curve. Moisture absorption testing was perfomned by storing the samples in an environmental chamber at 30 °C and 90 % relative humidity. Two rectangular samples of approximate dimensions: 63.5 mm x 12.7 mm x 3.2 mm were used for each biocomposite. The weight increase of the samples was measured until steady state was reached (~50 days). Moisture absorption was evaluated as the percent weight gained at steady state. The 131 dispersion and morphology of clay inclusions in the resin systems was assessed with transmission electron microscopy (TEM). AJEOL 100CX TEM with LaB6 filament and 120 kV acceleration was used to obtain bright field images from sections approximately 70 mm thick. Tensile failure surfaces of biocomposites were pre-coated with a thin gold film and observed in a JEOL 6400 field emission scanning electron microscope (SEM) at 10 kV acceleration voltage. )3 )' Kerri/z; The addition of EMS or nanoclay may have a beneficial or detrimental effect depending on the parameter studied. All results in the following sections are thus relative to benchmark biocomposite A, consisting of 100 % UPE (0 % EMS and 0 % nanoclay). 5.5.1 Tensile modulus and ultimate tensile strength Results for tensile modulus and strength are provided in Figure 5—1. The overall trend considering only the average values suggests that the addition of bio-resin in UPE reduced tensile modulus and ultimate tensile strength of reslting biocomposites, which is attributed to the low stiffness of the bio-resin. The average values of tensile modulus for biocomposites B (10 % EMS) and C (20 % EMS), were reduced by approximately 5 % and 40 %, respectively. Similarly, for biocomposites B and C, the average tensile strengths values decreased around 5 % and 20 %, respectively. Tensile modulus is primarily dependent on the micro-scale natural fiber reinforcement and hence a major increase in tensile modulus due to 1.5 wt. % of nanoclaywas not expected. Nevertheless, biocomposite D, which had 1.5 wt. % nanoclay in UPE (no EMS) showed an increase in the range of 5% in the average tensile modulus. A similar increase for biocomposite materials with 10 % EMS (B, E) was expected. Nevertheless, it seems that the reduction in tensile modulus due to addition of 10 % EMS 132 was more than the recovery provided by adding 1.5 wt. % nanoclay. Regarding strength, the addition of nanoclay in the resin blends showed a decrease in the average values of ultimate tensile strength in the range of 20 % and 25 % for biocomposites D and B, respectively. This decrease may be attributed to the embrittlement of the resin system. It can be seen from Figure 5-1 that there is large variance in test results. The variance could be due to many reasons including processing, presence of voids and irmpurities, improper fiber distribution, variation in fiber lengths, distribution of nanoclay etc. Detailed statistical analyses taking into account these variations should be performed to quantitatively and precisely obtain the effects of constituents, namely EMS and nanoclay. In this work, onlythe average values were considered and hence a qualitative effect and overall trends based on average values could be obtained. Overall, the addition of bio-resin reduces the tensile modulus and strength of the composite. The addition of nanoclay leads to a recovery in modulus but reduces the tensile strength. 5.5.2 Elongation at failure and notched Izod impact strengths Figure 5—2 summarizes the results of elongations at failure from tensile tests and notched Izod impact strengths. The overall trend considering only the average values suggests that the addition of EMS increased the toughness of the biocomposites, which was observed experimentally as increase in average values of elongations at failure and impact strengths. For resin blends with no nanoclay inclusions an average increase in ultimate strain values of approximately 30 % and 65 % was observed for biocomposites B (10 % EMS) and C (20 % EMS), respectively. Similarly, the average impact strength values increased by approximately 10 % and 15 % for biocomposites B and C, respectively. The addition of nanoclay seems to make the resin blends more brittle and hence reduces the ductility, 133 FL. 5.11. .. correspondingly the elongations at failure and impact strengths. The average values of ultimate strains and impact strengths showed an approximate reduction of around 25 and 20% in ultimate tensile strains and irrpact strengths, respectively for biocomposite D due to the addition of 1.5 wt. % clay in meat UPE. Biocomposite E, which has a combination of 10% EMS and 1.5 wt. % nanoclay showed a synergistic effect. The overall performance of biocomposite E was very similar to baseline composite A, with the average values of ultimate strains and the impact strengths within 5% of those of baseline bicomposite A. This suggests that the combination of bio-resin and nanoclay can balance individual deficiencies and the resulting hybrid composites will have the beneficial effects of each of the constituents resulting in good stiffness-toughness balance. 5.5.3 Thermal properties and moisture absorption Figure 5—3 summarizes the results of coefficient of thermal expansion (CIE) and moisture absorption (MA) tests; The overall trend considering the average values suggests that both C775 and MA increased with increasing bio-resin (EMS) content and decreased with the addition of nanoclay. The increase in average CTE values due to addition of 10 % EMS (biocomposite B) and 20 % EMS (biocomposite Q was around 10 % and 25 %, respectively. The average glass transition temperature (Tel value of virgin UPE was approximately 98 °C. The average Tg values of hybrid biocomposites D (100/0/15) and E (90/10/1.5) were 102 °C and 84 °C, respectively. The T8, of hybrid biocomposites was dominated by the properties of the resin system, and as such, the effect of nanoclay on the T8 of bio-based resins is reported in reference [5-12]. Similar to CTE, the increase in average moisture absorption value for biocomposite C was around 20 %. The addition of 1.5 wt. % 134 nanoclay reduced the average values of MA of biocomposite D by approximately 10 %. Biocomposite E, with 10 % EMS and 1.5 wt. % nanoclay, showed results similar to benchmark biocomposite A. Overall, it was observed that the reduction of average thermal and banier properties due to the addition of bio-resin were recovered by addition of nanoclay. 5.5.4 Nanoclay dispersion and exfoliation. Transmission electron microscopy (TEM) was used to observe the dispersion and morphology of nanoclay in the resin system (no fibers). Figure 5-4 shows the bright field TEM micrographs of UPE/ clay nanocomposites. It was observed that the nanoclay platelets were well distributed with partially exfoliated and intercalated morphologies. Figure 5-4a shows a general view of clay platelets distributed in UPE matrix and, Figure 5-4b shows an intercalated gallery containing three to four platelets. Similar micrographs were obtained for bio-based matrices (EMS blends) as the same amount of sonication energy was spent on all resin systems. Improved nanoclay exfoliation can be achieved with increased sonication energy. Nonetheless, excessive sonication energy/ time can break the clay platelets and reduce their aspect ratios, leading to inferior composite properties. A study on the effect of processing techniques, with varying sonication energies and solvents, on the resulting morphology and properties of bio-based polymer nanocomposites is reported in reference [5—17]. 5.5.5 Fracture surface observations. Scanning electron microscopy (SEM) was used to observe the features of tensile fracture surfaces and interfacial fiber-matrix. This approach has been used by others to evaluate the stiffness strength and toughness performance of biocomposites [2-4]-[2-6]. 135 Three types of failure mechanisms have been identified in literature for natural fiber reinforced polymers: matrix failure, fiber fracture, and fiber-matrix interfacial failure [2-6]. A weak interface or improper compatibility between fiber and matrix may lead to fiber pull-out instead of fracture, and may reduce the resulting mechanical properties. In this study, combinations of these failures were observed depending on the composition of the biocomposite. Figure 5-5d shows a representative SEM micrograph corresponding to biocomposite E with 10 % EMS in UPE and 1.5 wt. % nanoclay. The encircled region shows fiber pull-out and the boxed region shows fiber fracture. It was observed that the interfacial gaps around pulled-out fibers increased with increasing bio-resin content, suggesting weaker adhesion characteristics between fiber and the bio-based matrix. The interfacial characteristics of pullout failures were thus studied. Figure 5-5a and Figure 5-5b show that the interfacial gap increases with increasing bio-resin contents for 10 % EMS (biocomposite B) and 20 % EMS (biocomposite C), respectively. The interface gap for biocomposite E, with 10 % EMS and 1.5 wt. % clay revealed a similar interface gap as its counterpart neat resin biocomposite B, with 10 % EMS and no clay (Figure 5-5c). This would indicate that nanoclay reinforcement does not seem to alter the adhesion features between fiber and matrix. Tensile test results support the hypotheses that weaker interface leads to a more pronounced pull-out phenomenon, as reduction in tensile modulus and strength were observed with an increase of bio-resin. Fiber pull-out enables dissipation of more energy along the interface, which is consistent with higher impact strength and ductility values obtained when increasing bio-resin content. The final study in tensile fracture morphologies was the study of the matrix surface. Figure 5-6 shows matrix morphologies in the tensile fracture surfaces of biocomposites A, B, D and E (see Table 5-1 for nomenclature). The roughness of a fracture surface has generally 136 been associated with fracture properties and critical strain energy release rates [5-18]. Smooth featureless fracture surface is attributed to brittle failures and rougher fracture surfaces are attributed to tougher nanocomposites [5- 18]. SEM micrographs for biocomposites A (Figure 5-6a) and B (Figure 5-6b), with no nanoclay, revealed rm... surfaces that were relatively smooth and featureless. The roughness of the fracture surfaces increased with increasing bio-resin and nanoclay content. For neat resins (no clay), an increase in bio-resin content resulted in increased surface roughness in the fomn of reduced spacing of the fractal planes. Increase in surface roughness due to addition of clay (Figure 5-6c and Figure 5-6d) was more obvious. The increase in surface roughness due to combination of bio-resin and nanoclay suggests resulting hybrid biocomposites are tougher than benchmark biocomposite A. Research has shown that changes in fracture morphologies suggest different toughening mechanism at low and high clay loading [5-18]. It is difficult to quantitatively relate the fracture surface to toughness of the composites [5-18]. This requires better understanding of micro-structural parameters, crack propagation mechanisms and interface studies to relate fracture and toughness measures to the fracture surface morphologies. )3 6 Dita/55$” The characterization of hybrid biocomposites made of natural fibers, petroleum based resin (UPE), bio-resin (EMS) and nanoclay inclusions (Cloisite 30B") showed a variety of multifunctional properties. As expected, the experimental data revealed scatter/ variations in measured parameters. These variations were specifically larger for failure-dependent parameters, such as tensile failure strains, tensile strengths and impact strengths. Detailed statistical analyses taking into account these variations should be performed to quantitatively and precisely obtain the effects of constituents, namely EMS and 137 nanoclay. In this work only the average values were considered and hence a qualitative effect and overall trends based on average values could be obtained. It was observed that the combination of layered silicates and bio-based resin systems provide biocomposites with similar or better properties than the benchmark biocomposite manufactured from natural fibers and UPE. Mechanical testing revealed a decrease in stiffness-related parameters such as tensile modulus and ultimate tensile stress, and an increase in toughness related parameters, such as ductility (elongation at failure) and impact strengths, with increasing bio- resin content. The addition of nanoclay increased stiffness but reduced toughness of the composites. The irrprovement in toughness of the resulting biocomposites was thus compromised with the addition of nanoclay. The advantages of combining EMS and nanoclay provides a synergistic effect that is not limited to only achieving a stiffness- toughness balance for the resulting bio-based composites, as similar enhancements were observed in barrier and thermal properties. The addition of EMS increased moisture absorption and the coefficient of thermal expansion of the resulting biocomposite. However, results indicate that the adverse effects of bio-resin on permeability and themnal properties can be recovered with the enhancement provided by the nanoclay. The moisture absorption results also show that the biocomposite material system absorb less moisture than the natural fibers alone. Biocomposite A (benchmark) absorbed approximately 3 % moisture, while the hemp fibers alone are reported to absorb 8 % moisture [5-7]. The matrix thus serves as a banier, reducing the moistrne absorbed by the natural fibers. In a bio-based polymer, bio-resin is more penneable than its synthetic counterpart. The incorporation of nanoclay inclusions introduces a tortuous path for moisture movement thereby enhancing banier properties of the rratrix and enabling the recovery of the negative effects from bio-resin addition. 138 The enhancements provided by nanoclay were more pronounced in UPE biocomposites than UPE/ EMS biocomposites. This indicates that the negative effect from EMS addition was larger than the benefit provided by nanoclay and raises questions on the compatibility of EMS and nanoclay. The authors believe that the compatibility was adequate. In a parallel numerical study, material layout optimization of the tlrree-phase system (UPE, EMS, and nanoclay) was camied out to identify their distribution by matching the numerically homogenized tensile modulus to experimental data. Results indicate that the bio- resim tends to accumulate around the nanoclay platelet, suggesting affinity between them. This deposition of the more complaint EMS around the clay platelet affects the stress transfer and hence supports the lower degree of enhancements in mechanical properties. Transmission electron microscopy on the matrix revealed partially exfoliated and intercalated galleries. Improvements in nanocomposite processing can enable better exfoliation thereby leading to better stiffness and banier properties [5-17]. Meanwhile, an intercalated morphology gives better toughness properties [5-17]. This suggests that a balance of exfoliation and intercalation is best suited for stiffness/ toughness balance of the nanocomposite. However, since mechanical properties are governed by macro-fibers, exfoliated nanoclay morphology is preferred as it enhances banier and themnal properties. Scanning electron microscopy lead to interesting observations related to the toughness and the adhesion between the natural fibers and the nanoclay reinforced bio- based resin. While the effects of nanoclay and bio-resin additives on fiber adhesion are not fully understood, the interfacial properties can be improved by surface treatment of the fibers. Industrial raw (unprocessed) fibers were used in this work to characterize an initial benchmark towards a synergistic balance of the constituents. In this way, a lower performance bound for this type of hybrid bio-based composite was obtained. 139 Improvements in nanocomposite processing, incorporation of higher amounts of nanoclay and bio-resin along with the use of “engineered,” or treated, fibers would lead to enviromnentally friendly composites with competitive themno-mechanical properties. )I 7 60726.4(:sz Results from this study indicate that novel multiscale hybrid bio—based composites can be obtained from a combination of industrial hemp and blends of unsaturated polyester with epoxidized methyl soyate and nanoclay inclusions. The properties of the resulting biocomposites are tailorable and dictated by the amount and distribution of the constituents. Experimental characterization studies showed that the addition of bio-resin lowers mechanical parameters, such as stiffness and ultimate tensile stress, but increases toughness parameters, such as impact strengths and ductility. The addition of nanoclay enhances stiffness but seems to decrease toughness. Thus, the study shows that a proper stiffness/ toughness balance can be obtained by controlling the amount of bio-resin and nanoclay content. Moreover, the multiphase hybrid biocomposites have multifunctional properties, such as improved banier and themnal properties. Incorporation of higher concentrations of bio-resins and nanoclay along with improvements in processing will enable maximizing the multifrmctional properties that the hybrid biocomposites offer. The study shows that bio-based composites with proper stiffness/ toughness balance can be obtained while preserving environmental friendliness and cost effectiveness. The improved multifaceted features possible for these sustainable bio-based materials are likely to increase their appeal for use in transportation and housing structural applications. 140 )3 6’ fiéé’: afla’f'zgww Table 5—1. Biocomposite material identification and composition. Composition Fiber Fractions Density 5 . n (%) (%) (g/cc-l pecrme. Weight Volume . . Identification UPE EMS (Jay Fraction Fraction Resrn Comp csrte W/ Zr p... pr A 100 0 O 21 18 1.177 1.257 B 90 10 0 21 17 1.161 1.234 C 80 20 0 21 17 1.157 1.157 D 100 0 1.5 22 18 1.196 1.245 E 90 10 1.5 21 17 1.178 1.201 Densityof Fiber (pf) - 1.48 g/cc [7]. The nomenclature is also referred as: [UPE / EMS / Clay/ Wj/ Vf], For example, specirren ID “D” can be referred as [100 / 0 /1.5 / 22 /18]. 141 50 8 - Tensile Modulus ; - Ultimate Tensile Strength _ Tensile Modulus (GPa) Ultimate Tensile Strength (MPa) Composite Type Figure 5-1. Tensile modulus and ultimate tensile strengths. 60 1.2 I - Impact Strength : _ - Elongation at Break - 50 - - 1.0 - s E ‘ _ v x 40 - - 0.8 a” " - - s: a l E . . m 5 3°: _- 0.6 a m j g ‘6 - _ =- a 20 .- — 0.4 § E - r g . . 1.1.1 10: in 0 J ion A B C D E Composite Type Figure 5-2. Impact strengths from notched Izod tests and tensile strains at failure. 142 O 0 § -CTE * -MA .11. [~rr17 § 0 1 N O .3 Moisture Absorption. MA (%) ..L O .|.. Coefficient of Thermal Expansion, CTE (pm/mm°C) 0) O O 1... A B c D E Composite Type Figure 5-3. Linear coefficient of thermal expansion (CTE) and moisture absorption (MA). Figure 5-4. Bright-field TEM micrographs revealing homogenous dispersion with partially exfoliated and intercalated clay particles in UPE matrix. a) low magnification, scale bar = 1 pm, b) high magnification, scale bar = 50 nm, approximately 3 to 4 particles per gallery of intercalated particle 143 Figure 5—5. SEM micrographs of tensile fracture surfaces showing interfacial gaps between fiber and rmtrix. (a) biocomposite B: 10 % EMS and no nanoclayin UPE, (b) biocomposite C: 20 % EMS and no nanoclayin UPE, (c) biocomposite E: 10 °/o EMS and 1.5 wt. % nanoclay in UPE, and (d) representative fracture surface showing both pull-out (circled region) and fracture (boxed region) of fibers, scale bar = 100 um. Images (3) to (c): scale bar =10 pm. 144 Figure 5-6. SEM micrographs showing matrix region in tensile fracture surface of biocomposites. Micrographs (a), (b), (c) and (:1 represent biocomposites A, B, D, E, respectively. All images have a magnification scale bar of 10 pm. 145 )2 9 Xcflmtzce: [5-11 [5-2]. [5- 3]. [5-4]. [5-5]. [5- 6} [5-7]- [5- 8]. [5-9]. [5- 10]. [5-11]. [5- 12]. [5- 13]. Mohanty, A.K, Misra M, Hinrichsen, G. Biofibers, Biodegradable Polymers and Biocomposites: An Overview. Macromol Mater Eng 2000; 276/277:1-24. Nickel J, Riedel U. Activities in biocomposites. Mater Today 2003; 6:44-48. O’Donnell A, Dweib MA, Wool RP. Natural fiber composites with plant oil based resin. Compos Sci Technol 2004, 64:1135-1145. Shibata M, Ozawa K, Teramoto N, Yosomiya R, Takeishi H. Biocomposites made from short abaca fiber and biodegradable polyesters. Macromol Mater Eng 2003; 288:35-43. Morye SS, Wool RP. Mechanical properties of glass /flax hybrid composites based on a novel modified soyabean oil matrix material. Polym Campos 2005, 26:407- 146. Bodros E, Pillin I, Montrelay N, Baley C. Could biopolymers reinforced by randomly scattered flax fibers be used in structural applications. Compos Sci Techn012007; 67:462-470. Burguefio R, Quagliata, MJ, Mohanty AK, Mehta GM, Drzal LT, Misra M Load bearing natural fiber composite cellular beams and panels. Composites: A 2004; 35:645-656. Miyagawa H, Mohanty AK, Burgueno R, D1721 LT, Misra M Novel bio-based resins from blends of functionalized soyabean oil and unsaturated polyester resin. J Polym Sci, Part B: Polym Phys 2007; 45: 698-704 Miyagawa H, Mohanty AK, Burgueno R, Drzal LT, Misra M Development of bio- based unsaturated polyester containing functionalized vegetable oils. Ind Eng Chem Res 2006; 45:1014—1018. Le Baron PC, Wang Z, Pinnavaia T], Polymer - layered silicate nanocomposites: an overview. App] Gay Sci. 1999; 15:11-29 Miyagawa H, Mohanty AK, Burgueno R, Drzal LT, Misra M Characterization and thennophysical properties of unsaturated polyester- layered silicate nanocomposites] Nanosci Nanotech 2006; 6:464-471. Haq M, Burgueno R, Mohanty AK, Misra M Bio-based unsaturated polyester/ layered silicate nanocomposites: characterization and thermo-physical properties. Composites: A, Communicated. Tsai JL, Wu MD. Organoclay effect on mechanical responses of glass/ epoxy nanocomposites] Compos Mater 2008; 42553-568. 146 [5-14]. Kinloch A], Mohammed RD, Taylor AC, Sprenger S, Egan D. The interlaminar toughness of carbon fibre reinforced plastic composites using ‘hybrid-toughened’ matrices] Mater Sci 2006; 41:5043-5046. [5-15]. Kinloch A], Mohammed RD, Taylor AC, Sprenger S, Egan D. The effect of silica nano particles and nibber particles on the toughness of multiphase thermosetting epoxy polymers. J Mater Sci 2005; 405083-5086. [5-16]. GojnyFH, Wichmann MHG, Fiedler B, Bauhofer W, Schulte K. Influence of nano- modification on the mechanical and electrical properties of conventional fiber- reinforced polymers. Composites: A 2005; 36:1525-1535. [5-17]. Haq M, Burgueno R, Mohanty AK, Misra M Processing techniques for bio-based unsaturated-polyester/ clay nanocomposites: tensile properties, efficiency and limits. Polymer, Communicated. [5-18] Wang L, Wang K, Chen L, Zhang Y, He C. Preparation, morphology and thennal/ mechanical properties of epoxy/ clay nanoclhy composite. Composites: A 2006; 37:1890-1896. 147 Chapter 6. Multiscale Hybrid Biocomposites: II - HF / UPE / EML1 6 / 155/ma Environmentally friendly bio-based composites with enhancements in multiple properties can be obtained by harnessing the synergy offered by hybrid constituents such as multiscale (nano- and micro- scale) reinforcement in bio-based polymers composed of blends of synthetic and natural resins. Multiscale reinforcement offers synergy of constituents at various length scales, and when combined with bio-based resins provide stiffness-toughness balance, improved thermal and barrier properties, and increased environmental appeal to the resulting composites. Bio-based composites consisting of unsaturated polyester (UPE, petroleum-based resin), epoxidized methyl linseedate (EML, bio-resin), natural fibers (industrial hemp), and nanoclay inclusions were developed. The effects of EML and nanoclay content, and optimal material compositions that maximize synergy of resulting biocomposites were studied through experimental characterization of 12 biocomposites with EML and nanoclay contents of upto 30% and 5 wt.%, respectively. Results show synergistic behavior of resulting biocomposites with the effects of bio-resin addition being complimented by nanoclay, and vice versa. Moreover, these hybrid materials are tailorable in perfomnnce and in environmental impact, and show potential for wide applications. 62 lam/«(17'0” Hybrid bio-based composites consisting of multiscale reinforcements, namely natural fibers and nanoclay embedded in blends of petroleum based (unsaturated polyester) and l Haq, M, Burgueno, R, Mohanty, A.K., and Misra, M, “Multifunctional Biocomposites from Hybrids of Hemp fibers, Nanoclay, Linseed oil and Unsaturated Polyester.” Draft - to be submitted to Journal of Biobased Materials and Bioenergy. 148 19.1. . . .-...E 1.»... salt: " I 1- vegetable oil based resins (derivatives of vegetable oils, e.g soybean, linseed oil etc.), were found to produce composites that exhibit synergistic behavior, including stiffness-toughness balance, improvements in thermal and banier properties, and increased environmental appeal [6-1]. Biocomposites, composed of natural fibers in synthetic or natural polymer matrices have recently gained much attention due to their low cost, environmental friendliness, and their potential to compete with synthetic composites [6-2}[6-6]. Nonetheless, the use of bio-based composites has been limited due to their lower mechanical and thermophysical properties compared to synthetic composites and conventional structural rmterials [6-516-6]. Nanoclay reinforcement in matrices of biocomposites enhances the thermal and banier properties, and prevents moisture from reaching the natural fibers, thereby help maintain the integrity of the biocomposites and enhance its ' durability. The advantage of multiscale reinforcement is not limited to barrier properties, but when combined with bio-based resins provide stiffness-toughness balance, increase environmental appeal and provide rmrlti-property improvements of resulting biocomposites [at]. Environmental concerns related to the use of synthetic, or petroleum-based, polymer rmtrix composites has propelled the development of composite rmterials based on natural or renewable sources [6-816-9]. Bio-based resins, or bio-blends, obtained by replacing part of a petroleum-based resin with natural bio-resin have been found to be a promising compromise between environmental friendliness and performance [6-1], as the use of all natural bio-resins has been limited due to perfonnance concerns [6-7]. In addition to higher natural content, bio-based resins improve toughness of the resulting resin blend [6-1] [6-8I6- 10]. However, this increase in touglmess compromises stiffness, barrier and thermal properties [6-8I6-10]. Stiffness and toughness are opposing performance parameters and a 149 proper balance is required for an efficient composite. Hence, one way pursued by the authors to obtain stiffness-toughness balance and recover the lost barrier and thermal properties due to bio-resin addition is by addition of layered silicates (nanoclay). The reinforcement of petroleum-based polymers with nanoclay has been shown to impart multifunctionality to the resulting polymers with enhancements not only on stiffness but improvements in themnal, barrier, flamnnbility and ablation resistance properties [6-11]. Polymer-clay nanocomposites are well understood and considerable literature and review articles [6-11}[6-14] exist on this topic. The advantage of nanoclay reinforcement in bio- based polyrners is that it enables the recovery of the properties lost due to bio-resin addition, and produce bio-based nanocomposites with improved stiffness-toughness balance [6-1516- 16]. Thus, the synergy between nanoclay reinforcement in a bio-blend produces a polymer nanocomposite with similar or enhanced properties than the virgin polymer and holds great promise for use in wide applications [6-16]. The application of nanoclay reinforced bio-based polymers to fiber reinforced composites provides value added properties to resulting biocomposites [6-1]. Thus, in a biocomposite, natural fibers remain the main reinforcement for stiffness and strength, while the nanoclay enhances the resin properties thereby improving the transient properties [6-1]. The incorporation of nanoclay and natural fibers in resin systems thus provides reinforcements to resin systems at two scales. Although, different concepts of multiscale reinforcement have been studied for synthetic composites [617}[6-20], similar studies on multiscale/ multiphase reinforcements of bio-based composites are limited [6-1]. An initial attempt to use the multiscale reinforcements for biocomposites and to evaluate the synergy of the constituents was performed on bio-based resins containing blends of unsaturated polyester (UPE) and epoxidized methyl soyate (EMS), reinforced with 150 hemp fibers and nanoclay [6— 1]. The study incorporated up to 10°/o of EMS, 1.5 wt.% of nanoclay and an average vohrme fraction of 17% of short hemp fibers. The study showed promising results but was limited by the nanocomposite processing that allowed relatively less amounts of bio-resin (10°/o) and nanoclay (1.5 wt.%). In order to exploit the synergy offered by these materials and improve the environmental appeaL higher amounts of bio- resin and nanoclay were recommended [6-1]. Also, a studyon blends of neat resins (no clay), by replacing parts of UPE with EMS [6-16] and EML [521] suggests UPE/EML composites perform relatively better than similar composites from UPE/ EMS blends. In this work, the shortcomings and recommendations from the UPE/ EMS nanocomposites study [6-16] were addressed and improvements were made. Firstly, the bio- resin used in this work is EML instead of EMS, as it has been reported to perform better [6- 21]. Secondly, an improved manufacturing method identified from a detailed processing study[6-22], and which allowed incorporation of up to 30 wt.% EML and 5.0 wt.% nanoclay was used. Thirdly, the average volume fraction of mammal fibers (hemp fibers) used in this work (27%) was higher than that from UPE/ EMS study (17%) [6—1]. Finally, the possibility of incorporating relatively large amounts of nanoclay and bio-resin allowed detailed experimental characterization on 12 biocomposites, unlike UPE/ EMS study wherein only five biocomposite systems were studied. Such detailed characterization allowed studying the effects of constituent materials on resulting biocomposite properties and enabled finding Optimized material designs that maximize the synergy of these constituents. Biocomposites with material systems that varied the bio-resin (EML) content in UPE and the concentration of nanoclay while maintaining a constant amount of natural fibers were fabricated by compression molding. Mechanical and thermo-physical properties were evaluated through standard tests. The degree nanoclay dispersion and morphology in the resin systems was 151 characterized using transmission electron microscopy and the features of fracture surfaces were characterized using scanning electron microscopy. Results show synergistic behavior of constituents in the biocomposite with improved stiffness-toughness balance and enhancements in multiple properties of resulting environmental friendly and cost effective biocomposites. The resulting biocomposites are tailorable borh in performance and in environmental impact and have potential for wide applications. 6.7 15' xpenbzefltd/ metéoafr Experimental detennination of thenno-physical properties and material characterization was performed on biocomposite plates made from bio-based resin blends and reinforced with nanoclay inclusions and natural fibers. In the following sections, details on materials, processing, parameters studied, and testing are provided. 6.3.1 Materials Industrial unprocessed raw hemp (Stemergy, Ontario, Canada) was used as natural fiber reinforcement. The fibers were hand chopped to an average length of 25mm. The primary component of the resin blend was ortho unsaturated polyester resin (UPE, Polylite" 32570—00, Reichhold Inc., NC), which contains 33.5 wt.% styrene. The secondary component that replaced parts of UPE was a bio-based modifier, epoxidized methyl linseedate (EML, Vikoflex" 9010, Arkema Inc, PA). . EML is a mixture of methyl esters of fatty acid compositions that construct the linseed oil. The detailed composition is 40-50 wt % methyl linolenate epoxy, 24—26 wt % methyl oleate epoxy, 17-22 wt % methyl linoleate epoxy, 4—7 wt °/o methyl palmitate, and 2-5% methyl stearate. The nanoclay used to reinforce the resin system was Cloisite 30B° (Southern Clay Products, Inc. TX). The resin blend was processed with cobalt naphthenate (Sigma Aldrich, MO) as a promoter and 2-butanone 152 peroxide (Sigma Aldrich) as an initiator. A constant ratio by weight of the resin system to the promoter and initiator was utilized to cure all of the biocomposites. The mixing ratio was 100 parts by weight of the resin system to 0.03 parts promoter and 1.50 part initiator. 6.3.2 Experimental matrix and nomenclature All biocomposite material systems were composed of manual fibers and a set of UPE-EML bio-based resin blends. The amount of bio-resin (epoxidized methyl linseedate, EML) that replaced the primary resin component UPE, was varied from 0% to 30%, in increments of 10%. A total of 4 meat resin (no clay) systems were obtained. Each of these resin systems were then reinforced with nanoclay inchrsions of 0 wt.%, 2.5 wt.% and 5.0 wt.%. A total of 12 resin systems were obtained which were then reinforced with natural hemp fibers. A summary of the experimental matrix inchrding nomenclature, composition, concentration of constituents and densities of the 12 biocomposite material systems studied are provided in Table 3-1. 6.3.3 Polymer nanocomposite processing The technique used for processing the nanoclay reinforced bio-based resin systems follows the findings from our group’s study on solvent-based processing techniques for bio- based clay/ polymer nanocomposites [6-22]. The technique found to be most efficient consists of sonicating the nanoclay in acetone to an energy level of 300 k], using a solution concentration of approximately 50 l of acetone to 1 kg of clay, while it is constantly stined. After sonication, only the UPE solution is added. The acetone + nanoclay + UPE solution is mixed continuously on a hot plate at approximately 55 °C to remove a majority of the acetone. The remaining acetone is removed by vacuum extraction at 55 °C for 24 h. During the acetone removal process the styrene present in the UPE is also removed. Thus, after 153 acetone removal the bio-resin, EML, is added along with the lost styrene. The processed solution is cooled to room temperature and blended with the initiator and promorer followed by curing. A flow chart depicting the process is shown in Figure 6-1. The processed resin system with UPE, EML and nanoclay was used in the manufacturing of biocomposite plates. 6.3.4 Manufacturing of hybrid biocomposite plates Flat biocomposite plates for the twelve biocomposite material systems as described in Table 3-1 were manufactured using compression molding. The natural fibers were chopped to an average length of approximatelyZS mm. These fibers were washed in distilled water and dried at room temperature for 12 hours. The fibers were then dried in a vacuum oven at 80°C and 100 kPa of pressure for at least 12 hours before compression molding. The resin system obtained from nanocomposite processing as described in earlier section was mixed with cobalt naphthenate (Sigma Aldrich, MO) as a promoter (0.03% by weight of resin blend) and 2-butanone peroxide (Sigma Aldrich) as an initiator (1.5 wt.% of resin blend). The measured quantity of fibers were then impregnated with resin system (UPE + EML + nanoclay + initiator + promoter) by hand mixing until the material was consistent (by visual evaluation). The impregnated fibers were then placed in a frame mold. Care was taken to evenly distribute the fiber material in the mold to ensure a uniform sample since natural fibers have a tendency to clump and tangle together when mixed. The frame mold was placed between two steel plates covered with Teflon paper. The sample was then cured in a press under 550 kPa of pressure for a total of 4 hours using a time-temperature profile of 100 °C for 2 hours, followed by 150 °C for 2 hours. An overview of biocomposite manufacturing is provided inFigure 6—2. 154 6.3.5 Testing and characterization Characterization of the biocomposite material systems was done through ASTM testing of coupon samples taken from the compression molded plates. The tensile properties (modulus, strength and elongation at failure) were measured on 6 to 8 samples as per ASTM D638. The impact strength was measured on 5 samples using notched Izod tests as per ASTM D256. Moisture absorption testing was performed using Fisher Scientific Versa bath'-138 equipment that allowed immersion of samples in a distilled water bath with temperature maintained at 50 °C. The samples used for moisture measurements were rectangular bars and had average dimensions of 60.0 mm x 12.5 x 3.0 mm. The specimens were coated with impervious two-part epoxy on all edges to eliminate edge effects and limit the diffusion only through the thickness of the sample. All samples had flat and parallel surfaces as obtained from compression molded plate. The samples were placed in a vacuum oven at 80 °C for 24 hours to remove any residual moisture. The initial dry weights were recorded and the samples were immersed in water. The weight of the samples was measured every 12 hours for the first 3 days, followed by once every week until steady state (equilibrium, no further increase in weight of sample) was achieved. Moisture diffusivity coefficients were obtained from the moisture gain versus time plots for various biocomposites. The dispersion and morphology of clay inchrsions in the resin systems was assessed with transmission electron microscopy (TEM). A JEOL 100CX TEM with 12.36 filament and 120 kV acceleration was used to obtain bright field images from sections approximately 100 nm thick. Tensile failure surfaces of biocomposites were pre-coated with a thin gold film and observed in a JEOL 6400 field emission scanning electron microscope (SEM) at 10 kV acceleration voltage. 155 6 4 fiery/1:: The incorporation of EML and nanoclay may have a beneficial or detrimental effect on the resulting biocomposite properties, depending on the property considered. Since the amount of natural fibers were relatively similar (Weight fraction of 412%), the effects of bio-resin (EML) and nanoclay could be directly compared. All results in the following section are thus compared to benchmark biocomposite A, containing 100% UPE (0% EML and 0% nanoclay). Also, in the relative comparison in the following results, the biocomposite under consideration is provided in parentheses (See Table 3-1 for nomenclature). 6.4.1 Material characterization study (a) Tatsfle mnldus: Results for the tensile modulus are provided in Figure 6-3. Considering only the average values, the overall trend suggests that the modulus of biocomposite materials decreased with increasing bio-resin content. The average tensile modulus of biocomrposites B (10% EML), C (20% EML) and D (30% EML) was reduced approximately by 5, 30 and 50%, respectively. The addition of nanoclay enhances the properties of the resin system thereby increasing the overall composite properties. A major enhancement in modulus due to addition of nanoclay was not observed in our previous work as only 1.5 wt.% of nanoclay could be processed [6-1]. In this work, nanoclay content of up to 5 wt.% was used and hence relatively, a more pronounced effect of nanoclaywas observed. The sole effect of nanoclay was studied by comparing virgin UPE (no EML) biocomposites reinforced with nanoclay (E and I) with benchmark biocomposite. The average tensile modulus values improved approximately by 20% (E) and 30% (I) for nanoclay contents of 2.5 wt.% and 5.0 wt.%, respectively. 156 Similarly, for biocomposites containing 2.5 wt.% nanoclay, the biocomposite F (10°/o EML) had an average tensile modulus value approximately 5% higher than benchmark bicomposite. Similarly biocomposites with 20% (G) and 30% (H) EML contents showed reduction in average tensile modulus values of approximately 10% (G) and 45% (H), respectively. Finally, for biocomposites containing 5.0 wt.% nanoclay, biocomposites with 10% EML blend 0) showed an enhancement in average tensile moduhrs values of approximately of 15% 0), while addition of 20% (K) and 30% (L) EML contents showed reduction in average tensile modulus vahres by approximately 5% (K) and 30% (L), respectively. The biocomposites containing 10°/o EML and reinforced with nanoclay, revealed not only a complete recovery, but improvement in lost stiffness due to EML addition. Overall trend considering the average values of tensile modulus suggests that the addition of bio- resin (EML) reduces the tensile modulus and addition of nanoclay improves the tensile modulus. (b) Ullfimte mfle sbagnb: Results for the ultimate tensile strength are summarized in Figure 6-4. Similar to modulus, the sole effect of nanoclay on UIS was studied by comparing virgin UPE (no EML) biocomposites reinforced with nanoclay (E and I) with benchmark biocomposite. The biocomposite E, had average tensile strengths similar to baseline biocomposite A. Nevertheless, biocomposite I, showed an average reduction of approximately 20%. The effect of EML was studied by comparing neat resin (no clay) biocomposites. The biocomposites B (10% EML) and C (20% EML) showed an improvement in average UI'S vahres by approximately 20% (B) and 30% (C), respectively. Similarly for biocomposites reinforced with 2.5 wt.% nanoclay, improvements in average UTS values of approximately 20% (both F and G) was observed. Also, for biocomposites reinforced with 5.0 wt.% nanoclay, improvement in average UTS values of approximately 157 10% (I) and 20% (K) were observed. Finally, biocomposites with 30% EML content revealed reduction in average UIS values relative to counterparts with 10°/o and 20% EML contents but were higher than the baseline biocomposite (A). Biocomposites with 30% EML, namely, D (0% nanoclay) had average UIS values higher by approximately 10%, while biocomposites H (2.5 wt.% nanoclay) and L (5.0 wt.% nanoclay), on average had UIS values similar to benchmark biocomposite. Overall, the addition of nanoclay seems to reduce the UIS values of resulting biocomposites while addition of bio-resin (EML) seems to enhance the UIS values. P Nevertheless, reduction in average UIS values was observed for biocomposites with EML contents beyond 20%. Finally, all hybrid biocomposites that had both EML and nanoclay, showed average UIS values similar or higher than baseline biocomposite (A), indicating that the lost UIS properties due to nanoclay addition were recovered by EML addition, suggesting a synergistic behavior of the hybrid constituents. (c) 1510311201 at failure: The summary of tensile elongation at failure results is provided in Figure 6-5. The overall trend considering the average values of tensile failure strains suggests that the EML addition increases the ductility of the resulting biocomposite. For neat resin (no nanoclay) biocomposites, increase in average tensile failure elongation values of approximately 45% (B), 110% (C), and 150% (D) was observed. The nanoclay addition seems to make the resin systems more brittle, thereby reducing the ductility and comesponding tensile failure elongations. A reduction in average tensile failure elongations of approximately 20% was observed for nanoclay reinforced neat UPE (no EML) biocomposites (E and I). For biocomposites containing 2.5 wt.% nanoclay, the average tensile faihire elongations improved by approximately 20% (F), 55% (G) and 130% (H). 158 Similarly for biocomposites containing 5.0 wt.% nanoclay, the average tensile failure elongations increased by approximately 15% (I), 75% (K) and 90% (L). Overall nanoclay addition seems to make the resin system brittle and reduces the elongations at tensile failure of resulting biocomposites. The loss of ductility due to the addition of nanoclay seems to not only be recovered but improved by addition of EML in all biocomposites in this study. (d) NW Izod impact smell): Figure 6-6 summarizes the results from the notched Izod impact strength testing. The overall trend considering the average impact strength values suggests that impact strengths of resulting biocomposites increased with increasing EML content and decreased with the nanoclay addition. The increase in average impact strength values of neat resin (no clay) biocomposites due to addition of EML was approximately 70% (B), 105% (C) and 125% (D). The addition of nanoclay reduced the average impact strengths, as observed in virgin UPE (no EML) biocomposites reinforced with nanoclay. The impact strengths of biocomposites E and I, on an average reduced by approximately 20% due to addition of nanoclay. For biocomposites containing 2.5 wt.%, EML addition revealed enhancements in average impact strength values by approximately 65% (F), 90% (G) and 100%(I-D. Similarly for biocomposites containing 5.0 wt.% nanoclay, EML addition revealed increase in average impact strength values by approximately 60% (I), 80% (K) and 95% (L). Overall, the loss in impact strengths (toughness) due to addition of nanoclay seems to not only be fully recovered but enhancements were obtained by addition of EML content. A summary of the experimental tensile and impact strength results are provided in Table 6-2. 159 (1) Moisture abapa'm (MA): Moisture absorption properties for the biocomposites were obtained by water immersion tests where the weight gain at any given time (M) was measured until steady state was achieved. Due to thickness variations in the specimens the amount of moisture absorbed by each was different. Thus, instead of simply assessing the amount of mroisture absorbed, the speed of moisture absorption in polymers and nanocomposites was quantitatively compared by detemnining their diffusion coefficient [6- 231 Hence, moisture diffusivity coefficient was used as the pararreter to assess the effect of bio-resin (EML) and nanoclay content on the different bio-based polymer nanocomposites. The diffusivity coefficient, D, was computed from the initial slope of the moisture gain, Mt/Ill,3o versustime(-\/;/d)as: (fr 1) 7: M , / M w 2 = 12 [TM l where M, is the mass gain at anytime t,M,, is the maximum mass gain at equilibrium/steady state, and d is the thickness of the specimen. Figure 6-7 shows the plots of moisture gain versus time for meat (no clay) resin systems. The experimental data is shown in symbols and the exponential fits are superimposed as solid line. The diffusivity coefficients were obtained by substituting the initial slope from the curves in Figure 6-7 into Equation (6-1). The variation of the diffusivity coefficients D, with increasing EML content are shown as an inset in Figure 6-7. Due to the absence of nanoclay, the diffusion coefficients obtained show the effect of blending bio-resin (EML) in virgin UPE. The overall trend considering the average values of diffusivity coefficients of biocomposites suggests increase in diffusivitywith increasing EML 160 content. For neat resins (no nanoclay), the average diffusivity coefficient values increased by approximately 20% (B), 85% (C) and 200% (D). The addition of nanoclay seems to improve the banier properties, thereby reducing the diffusion coefficient values. For neat UPE (no EML) biocomposites, the addition of 2.5 wt.% and 5.0 wt.% nanoclay show reduction in average diffusion coefficient values by approximately 25% (E) and 35% (I), respectively. It is clear that the increase in moisture diffusivity coefficients due to addition of EML is higher than the reduction offered by nanoclay. For biocomposites containing 2.5 wt.% nanoclay, the average diffusivity coefficients values were higher by approximately 5% (F), 65% (G) and 190% (H) for EML contents of 10% (F), 20%(G), and 30%(I-D, respectively. The barrier properties lost due to addition of EML were only partially recovered by addition of 2.5 wt.% nanoclay. Nevertheless, for biocomposite J (10% EML, 5.0 wt.% nanoclay), average diffusion coefficient values were reduced by approximately 5%, indicating recoveryof loast barrier properties. Similarly, for biocomposites containing 5 wt.% nanoclay, the addition of 20% and 30% EML contents revealed increase in average diffusion coefficient values by approximately 20% (K) and 160% (L), respectively, thereby offering only partial recovery. Considerable enhancement in moisture diffusion properties were observed due to the addition of nanoclay. Nevertheless, the detrimental effect of EML was more than the enhancement provided by nanoclay. As a result a partial recovery was obtained in all biocomposites with EML contents more than 10°/o. Full recovery of lost banier properties due to EML addition was observed only in biocomposite J(10%EML, 5 wt.% nanoclay). The diffusivity coefficients of all biocomposites in this studyare summarized in Figure 6-8. (g) Narnia}! dsperszbr 47d afliwbr Transmission electron microscopy (TEM) was used in this study to observe the dispersion and morphology of nanoclay in the polymer system (no 161 fibers). Figure 6-9 (a) and (b) shows the bright field TEM micrographs of clay / UPE composites. Figure 6-9 (a) shows a general distribution of nanoclay particles in UPE matrix. It was observed that the clay platelets were well dismibuted and had partially exfoliated and intercalated morphologies. Figure 6-9 (b) shows a high magnification micrograph of an intercalated chy gallery in a UPE matrix containing 5 wt.% nanoclay. Similar micrographs were observed for UPE/EML blends reinforced with nanoclay. In order to obtain better exfoliation, additional energy must be spent in sonication of clay along with improvements in the functionalization of the nanoclay and the polymer. (h) Fm smface clam. Scanning electron microscopy (SEND was used to observe the features of tensile fracture surfaces and fiber-matrix interfaces. Such an approach has been used by others to study the fiber-distribution, stiffness and toughness performance of biocomposites [6-3I6-4I6-5].Three types of failure mechanisms have been identified in literature for natural fiber reinforced polymers: matrix failure, fiber fracture and fiber-matrix interfacial failure [6-5]. Similar failure morphologies were observed in the tensile fracture surfaces of the biocomposites in this study. The natural fibers used for all biocomposites in this study had same fiber qrantity, average lengths, surface characteristics and manufacturing procedures. Hence, the observed distribution of fibers was similar for all biocomposites. Figure 6-10 (a) and (b) show epresentative tensile fracture surfaces from biocomposite A. Figure 6-10 (a) is a low magnification (scale == 200 um) fracture surface showing the distribution of fibers and the various failure types. The fiber pull-out failure is observed by cavities left in the matrix (encircled regions) and fiber fracture is observed by loose fiber ends in the fracture surface (arrow regions). Similarly, Figure 6-10 (b) shows the failure morphologies such as pull-out 162 (encircled), fiber fracture (boxed) and fiber-matrix interface (arrow). Also, chimping / bunching of some fibers was observed. This may be due to two reasons: a) natural fibers have an inherent tendency to clump together, and b) the fibers used were not surface- treated, and hence (may have relatively poor adhesion characteristics with the polymer. Improvements of surface characteristics may avoid clumping of fibers and lead to better enhancement in material properties. Many studies have been performed to study the effect of surface treatrrents and a good review of the same is provided by Mohanty et. al [6-24]. SEM micrographs have also been used to observe interfacial perfomance or pullout characteristics of the fiber [6-3]. A weak interface, weaker matrix, improper compatibility between manual fiber and the matrix and/ or improper adhesion characteristics may lead to fiber pull-out instead of fracture and may reduce the resulting mechanical properties. In short random fiber composites, a fracture surface may generally show a combination of fiber pull-out, fiber fractures and fracture of matrix regions. The fiber pull-out is observed in tensile fracture surfaces by three main characteristics: a) observation of cavitywith shape and diameter of the fiber (indicating fiber pull—out, b) existence of long pulled-out fibers, which may have pulled out of the other half of the tensile specimen not viewed in SEM, and c) the interface gap - the gap between the fiber and the matrix in pulled-out fibers. It is generally possible to see the first two factors, (a) and (b) in all short-fiber composite tensile fracture surfaces. The pull-out strength of the fibers depends mainly on the embedment length of the fibers. In all randomly oriented short frber composites, it is possible that at the cross section of failure, there are some fibers that do not have enough embedment length and hence may pull out of the resin. This embedment length depends mainly on fiber-matrix adhesion property. Hence surface treatments may enhance the adhesion property and reduce the embedment length and corresponding pull-out type failure. A well embedded fiber may have 163 adequate bonding and lead to tensile fracture of the fiber. In general, SEM micrographs of tensile fracture surface will hence show a combination of the above mention failure mechanisms. In an earlier study by our group the interfacial gap (gap between the fiber and matrix) was studied for varying contents of bio-resin and nanoclay [6-1]. A sirmilar study was perfonred in this work by comparing the interfacial characteristics of specific biocomposites in the experimental matrix. The biocomposites with extreme values of bio-resin and nanoclay content within the experimental matrix under consideration (A, D, I and K) were selected. Figure 6-10 (c) shows the interfacial gap for the benchmark biocomposite A with no clay and no EML. Figure 6-10 (d) shows the increase in interfacial gap due to addition of 30% EML (D). The interface gap for biocomposite (I) with 0% EML and 5 Wt.% clay is shown in Figure 6-10 (e) and was observed to be similar or better (smaller interfacial gap) than benchmark biocomposite (A). A larger interface gap would have been expected if the nanoclay used and its morphology was having an adverse effect on interfacial properties. Also, the addition of nanoclay improves the overall matrix properties, thereby enabling fiber fracture instead of fiber pull-out and hence reduction in fiber-matrix interfacial gap. Finally, the biocomposite (L) with 30%EML and 5.0 wt.% nanoclay showed increase in interfacial gap relative to its counterpart biocomposite (I) and the benchmark composite (A). The study of interface and the effect of clay on fiber-matrix adhesion are not fully established and are beyond the scope of this work and hence no definite conclusions can be drawn at this time. The experimental tensile test results support this weaker interface and pull-out phenomena, as reduction in tensile modulus and strength was observed. The pull-out phenomena enables dissipation of more energy along the interface and hence higher impact strengths and higher ductilitywas observed for increasing bio-resin content. 164 The final study in tensile fracture morphologies was the study of matrix in biocomposites. The roughness of fracture surface has generally been associated fracture properties and critical strain energy release rates. A smooth featureless fracture surface is attributed to brittle failures and rougher fracture surfaces are attributed to tougher nanocomposites [6-25]. The matrix region in the fracture surface for biocomposite A (0% EML, 0% nanoclay), was found to be relatively smrooth and featureless (Figure 6-11 a). Figure 6-11 (b) shows the well-blended (UPE/EML) matrix region of biocomposite D (30%EMI., 0% clay). Figure 6-11 (c) shows the phase-separated, EML enriched rmatrix region of biocomposite D. Such phase separation was observed only in biocomposite (D) and not it Other biocomposites in this study. Overall, the roughness of the fractrue surfaces increased with increasing bio-resin (EML) content. Figure 6—11 (d) and (e) shows the matrix region of biocomposites with 5 wt.% nanoclay containing 0% EML (I) and 30% EML (L), respectively. The addition of nanoclay also increased the surface roughness of the matrix region. This suggests that the combination of EML blend and clay will provide tougher composites. Research has shown that change in fracture morphologies suggest different toughening mechanism at low and high clay loading. It is difficult to quantitatively relate the fracture surface to toughness of the composites [6-25]. This requires better understanding of micro-Structural parameter, crack propagation mechanisms and interface studies to relate the fracture and touglmess to the surface morphologies. of} Deters/072 Experimental characterization of hybrid biocomposites containing multiscale reinforcements of nanoclay and natural fibers (unprocessed hemp) in bio-based resins (blends of petroleum based UPE and bio- resim, EML) reveal synergistic behavior and 165 multifunctional properties. As expected, the experimental data revealed scatter/ variations in measured parameters. These variations were specifically larger for failure-dependent parameters, such as tensile failure strains, tensile strengths and impact strengths. Detailed statistical analyses taldng into account these variations should be perfonred to quantitatively and precisely obtain the effects of constituents, namely EMS and nanoclay. In this work, only the average values were considered and hence a qualitative effect and overall trends based on average values could be obtained. Overall, it was observed that hybrid combiratiom of layered silicates and bio-based resin systems provide biocomposites with similar or better properties than the baseline biocomposite containing natural fibers and virgin UPE (0% EML, 0% nanoclay). This work was an improvement of an earlier study [6-1] that had limitations on the amounts of bio-resin (EMS, 10%) and nanoclay (1.5 wt.%) used. The study [6-1] showed promise with respect to synergistic behavior of biocomposites and recommended incorporation of high amounts of bio- resin and nanoclay to fully exploit the benefits offered by these biocomposites. This work used improved processing that allowed incorporation of 30% of bio-resin (EML) and 5 wt.% nanoclay, thereby allowing to find limits to the synergistic perfomrance of these materials. Mechanical testing revealed enhancement in toughness related parameters such as impact strengths and ductility, and reduction of stiffness related parameters such as tensile modulus, with addition of bio-resin. Similarly, the nanoclay addition increased stiffness but reduced toughness of resulting biocomposites. An interesting observation was made in tensile strength results: the study on meat resins (no reinforcements) revealed decrease in tensile strengths with addition of bio-resin in UPE, which was attributed to the inherent weak nature of the bio-resin [6-16]. A similar trend was expected when such bio-based resins were used for biocomposites. Nevertheless, the 166 biocomposites evaluated in this work showed increase in tensile strengths for EML contents up to 20%, and them a drop in tensile strengths for EML contents of 30%. This mend was. consistent for nanoclay reinforced resin systems (Figure 6-4). Additonally, the nanoclay reinforced biocomposites had lower strengths than respective neat resin (no clay) counterparts. Similar reduction in tensile strengths was also observed for nanoclay reinforced (no natural fibers) resins. The authors believe that the failure initiates in the flaws and imperfections in the matrix and thus attribute the reduction in tensile strengths and ductility of nanoclay reinforced bio-based composites to the stress concentrations created by the nanoclay reinforcement in the relatively brittle polymer matrix. While there is no general consensus on this point of view, computational studies by a parallel effort to this work [6-26] have shown evidence to this mechanism It should be noted that most mechanical properties are governed primarily by the macro-reinforcement, (hemp fibers). Also, the fiber content in all the biocomposites was similar and the increase in tensile strengths due to increase in EML contents (up to 20%) suggests better synergy of the constituents. The addition of bio-resin makes the resin system ductile and less brittle. Such compliance offered by bio-resin addition allows the natural fibers to take rrrore stresses and hence the improvement in strengths. Finally, the decrease in tensile strengths at 30% EML content is attributed to phase separation of UPE and EML, thereby creating EMI. rich regions that lead to premature failure due to their inherent weak mature. Moreover, large variations in tensile strength values were observed which maybe due to many factors including manufacturing defects like fiber-distribution or voids, irmpurities such as the wood-like core rraterial present in the untreated natural fibers which may lead to premature tensile faihrre and cause scatter in the results. These defects and irmpurities may act as crack initiators, or lead to poor fiber-matrix adhesion. In spite of the scatter, 167 consistent trends (Figure 6-4) were observed with variations of nanoclay and bio-resin addition. Overall, mechanical characterization revealed that addition of EML enhanced toughness but reduced stiffness, while nanoclay improved stiffness but compromised toughness of the resulting biocomposites. The combination of nanoclay and EML provides a synergistic effect such that the detrimental effects of each constituent are complimented by the Other and beneficial properties are revealed in resulting biocomposites. The advantages of hybrid combinations of EML and nanoclay and the synergy they offer is not limited to stiffness-toughness balance but similar enhancements were observed in banier properties. The study on moisture diffusion of biocomposites revealed that addition of EML increased moisture diffusivity while excellent barrier properties enhancements provided by nanoclay inclusions helped in recovering the barrier properties lost due to EML addition. However, the increase in moisture diffusivity due to addition of EML was higher than the barrier resistance offered by nanoclay. Hence, complete recovery was possible only in biocomposites with 10% EML contents with 5 wt.% nanoclay. Nevertheless, nanoclay inclusions resulted in partial recovery in all other biocomposites. Also, it was observed that biocomposite material systems absorb less mroisture than natural fibers alone. Similar results were reported in an earlier study [6-1]. The matrix serves as a barrier for the moisture reaching the natural fibers. Hence, the merit of using nanoclayin the biocomposites is that it improves the barrier properties of the matrix by preventing/ delaying the moisture from reaching the manual fibers by introducing a tortuous path for moisture mrovenent, thereby help maintaining integrity of biocomposites and improving its durability. The improved processing and incorporation of high amounts of EML (30%) and nanoclay (5 wt.%) allowed to obtain limits to synergistic performance and the effects of each of the constituents on the resulting properties of the hybrid biocomposites. It was found 168 that biocomposites with EML contents beyond 30% showed least synergy among constituents. Although partial recovery was observed in most properties, the loss in properties due to the addition of EML was beyond the recovery provided by nanoclay. Biocomposites with 10% EML contents and nanoclay showed complete recovery with superior properties than the baseline biocomposites. Finally, biocomposites with 20% EML and nanoclay showed balanced properties, with properties similar or lower than baseline biocomposites. Considering ease of nanocomposites processing and synergistic performance of resulting biocomposite, EML contents Of 10% and 20% EML reinforced with nanoclay (2.5 wt.% and 5.0 wt.%) have been identified as Optimized material layouts. The nanocomposites processing technique used in this work were found to produce nanoclay platelets with a combination Of exfoliated and intercalated mrorphologies with excellent dispersion. Exfoliated morphologies lead to better stiffness and barrier properties, while intercalated morphology gives better toughness properties. A proper balance Of exfoliation and intercalation would be suggested for the desired stiffness — toughness balance. However, in biocomposites, and in general fiber reinforced composites, since mechanical properties are mostly govemed by fibers, exfoliated nanoclay morphologies are preferred for non-mechanical enhancements such as barrier and thermal properties. The study of tensile fracture surfaces of biocomposites revealed interesting observations on toughness and adhesion between manual fibers and hemp. The addition Of bio-resin (EML) led to increased fiber pull-outs and increased fiber-matrix interfacial gap, resulting in increased toughness and ductility. The addition of nanoclay and its effect on fiber-matrix adhesion is beyond the scope Of this work. Overall, the fracture surface Observations support the experimental Observation of synergistic behavior and the resulting stiffness- toughness balance due to the combination of EML and nanoclay. Meanwhile, the interfacial 169 all”: . properties and resulting properties of biocomposites can be improved by surface treatment Of the fibers [6-1]. Unprocessed industrial raw fibers were used in this work to characterize the resulting biocomposites such that lower perfomance bounds were Obtained. The use of “engineered,” or treated, fibers with nanoclay reinforced bio-resins would lead to novel environmentally friendly hybrid biocomposites with competitive and improvements in multiple properties. 6. o’ Chard/31'0”: Results from this study indicate that novel environmentally friendly bio-based composites with improvements in multiple properties can be obtained from combination Of multiscale reinforcements Of nanoclay and industrial hemp . fibers in resin blends Of unsaturated polyester (UPE) and epoxidized methyl linseedate(EML). Experimental characterization revealed addition Of EML increases toughness parameters like impact strengths and ductility, but reduces stiffness. Addition Of nanoclay improves stiffness but decreases toughness. Thus synergistic behavior of the hybrid biocomposite constituents resulted in excellent stiffness-toughness balance. Similar synergistic behavior was observed in moisture barrier properties where barrier properties lost due to bio-resin addition of EML were recovered by nanoclay. The incorporation of large amounts Of bio- resin (30%) and nanoclay (5 wt.%) enabled finding Optimized material layouts that maximize the synergy of these materials and also find the effect of the constituents resulting properties. It was found that biocomposites with EML contents beyond 30% showed least synergy, and properties lost due to bio-resin addition were not reasonably recovered by nanoclay inclusions. Biocomposites with 10°/o EML contents and nanoclay showed complete recovery with superior properties than the baseline biocomposites. Finally, biocomposites with 20% EML 170 and nanoclay showed balanced properties, with properties similar or slightly lower than baseline biocomposites. Considering ease of nanocomposite processing and synergistic perfomrance of resulting biocomposite, EML contents of 10°/o and 20% EML reinforced with nanoclay (2.5 wt.% and 5.0 wt.%) were identified as optimized material layouts. The mechanical properties of biocomposites are govemed mostly by manual fibers while nanoclay inclusions improve matrix properties, thereby enhancing transient properties like moisture diffusion and themral properties. Thus the use of multiscale reinforcements provides enhancements and synergistic behavior at different scales resulting in efficient bio based materials. Finally, these materials are tailorable in performance and environmental impact by controlling the constituent concentrations. Such tailorable nature along with the use of “engineered,” or treated, fibers with nanoclay reinforced bio-resins would lead to novel, environmentally friendly hybrid biocomposites with enhanced and competitive properties, thereby increasing the potential applications of such composites. 171 6 7 ”£46 and 133m.— Table 6-1. Biocomposite material properties, Composition and Identification Composition Fiber Fractions Density SPecimen (%) We' ht(%)VOlumre (g/cc.) Identification UPE EML (Jay Fracltgr'on Fraction Resin Composite W/ V} pm pt A 100 O 0.0 33 29 1.221 1.255 B 90 10 0.0 29 24 1.194 1.239 C 80 20 0.0 33 28 1.167 1.161 D 70 30 0.0 32 27 1.140 1.227 E 100 O 2.5 33 29 1.234 1.229 F 90 10 2.5 32 28 1.207 1.214 G 80 20 2.5 32 27 1.180 1.209 H 70 30 2.5 32 27 1.153 1.194 I 100 O 5.0 29 25 1.247 1.277 J 90 10 5.0 32 28 1.220 1.227 K 80 20 5.0 31 27 1.194 1.214 L 70 30 5.0 32 27 1.167 1.179 Densityof Fiber (pf) a«1.48 g/cc. The nomenclature is also referred as: [UPE / EML / Gay / Wf / Vf ], For example, specimen ID “6” can be referred as [90 / 10 / 2.5 / 32 / 28]. 172 Table 6-2. Measured properties of biocomposite material systems Material mum: / EML/ Gay/ W, / V,] MOE (GPa) UIS (MPa) EF (%) IS O/m) x 0' HI 0" RI HI A—[lOO / o / 0.0 / 33 / 29] B-[90 / 10 / 0.0 / 29 / 24] C[80 / 20 / 0.0 / 33 / 28] D-[70/30/0.0/32/27] E-[IOO / o / 2.5 / 33 / 29] F-[9O / 10 / 2.5 / 32 / 23] G-[80/20/25/32/27] H-[70 / 3o / 2.5 / 32 / 27] I-[lOO / o / 5.0 / 29 / 25] J-[9O / 10 / 5.0 / 32 / 28] K-[80 / 20 / 5.0 / 31 / 27] L-[70 /3o / 5.0 / 32 / 27] 6.11 5.75 4.26 3.18 7.45 6.55 5.57 3.45 7.95 6.88 5.70 4.43 1.52 0.50 0.53 0.61 0.34 2.23 1.17 0.60 1.93 0.66 0.72 0.17 19.85 25.53 28.45 22.37 20.55 24.47 25.27 19.53 16.40 21.89 25.31 20.34 2.12 2.41 3.21 4.49 1.04 1.85 1.50 2.12 3.17 3.24 1.79 2.15 0.37 0.54 0.77 0.92 0.28 0.45 0.58 0.85 0.32 0.42 0.64 0.71 0.02 0.08 0.08 0.14 0.05 0.08 0.03 0.18 0.14 0.08 0.09 0.07 26.11 44.73 53.60 58.22 21.02 43.48 49.94 52.67 20.24 42.12 46.52 50.92 5.50 6.33 11.08 3.28 5.08 4.79 5.29 19.82 3.14 7.53 5.57 2.49 E : Tensile modulus, on: IS: Impact Strength Ultimate tensile strength, an: Elongation at failure, 173 ONNICATI N Organoclay + Acetone + Energy= 300kJ C BIO + STYRENE ADDITION Add bio-resin (EML) & Llost amount of Styrene {r R -CURE MIX N Cool, Mix with promoter and Initiator MIXING UPE + Sonnicated Clay [Acetone Solution r ’ 1 ACETONE REM VAL Heat&vaccum extraction @ 55°C for 24 hours J CQRE Process blended at curing temperature Figure 6-1. Schematic of processing technique Of nanoclay reinforced UPE + bio-resin f FIB - P RATI 1. Chopped to an average length of 25 mm 2. Washed in distilled water 3. Drying: Room temperature for 12 h. and in Vacuum @ 80°C and 100 kPa for 12 h. M J r W Curing at a constant pressure of 550 kPa and temperature profile of 100°C for 2 h. L followed by 150 °C for 2 h. blends PRE— ESlN Mle D The prepared nanocomposite resin is cooled and mixed with promoter and initiator J Vlv F W Fl ER-RE IN MIXIN Measured quantity of dry fibers hand-mixed with nanocomposite resin until uniform consistency M J I 1 ELACEMENT Impregnated fibers placed in steel picture-frame mold and covered with teflon coated steel plates M J Figure 6-2. Manufacturing of nanoclay reinforced biocomposites Tenslle Modulus (GPa) Ultimate Tensile Strengths (MPa) - 0.0 wt.% Nanoclay 10 ‘ - 2.5 wt.% Nanoclay - 5.0 wt.% Nanoclay 0 10 20 30 Bio-resin [EML ] content (%) Figure 6-3. Experimental tensile modulus of biocomposite. 3 -.- 0.0 wt.% Nanoclay 32 . —v— 2.5 wt.% Nanoclay ] -I- 5.0 wt.% Nanoclay 23 5 24 -' 20 5 16 3 12 - . . . . 0 10 20 30 Bio-resin [EML ] content Figure 6-4. Experimental ultimate tensile strengths of biocomposites 175' : - 0.0 wt.% Nanoclay 1-0 '_ - 2.5 wt.% Nanoclay . - 5.0 wt.% Nanoclay Elongation at Failure (%) 0 10 20 30 Bio-resin [ EML] content (%) Figure 6-5. Experirnemtal elongations at tensile failure of biocomposites 70 _' + 0.0 wt.% Nanoclay - -"- 2.5 wt.% Nanoclay -I- 5.0 wt.% Nanoclay so{ 50{ 30{ Izod Impact Strengths (Jlm) 20f 1o ‘ . I . . o 10 20 30 Bio-resin [ EML] content Figure 6-6. Impact strengths of biocomposites from notched Izod tests 176 1-0 - +1oo/om I + 90/1010 . -l— 80l20/0 03 _ + 70/3010 ‘ - 100IOI0 M 0-5 j a .— gone/o r . a — 80l20l0 . u‘ — 70/30/0 Mm . .5 6 0.4 - '23 . x 4 _ . d . 2 4 0.2 - J I 0 o 10 20 so EML content % o.o ......U. 0 100 200 300 400 500 w/i/d , (JE/mm) Figure 6—7. Moisture diffuisivity of neat resin (no nanoclay) biocomposites. : 0.0 wt.% Nanoclay _ - 2.5 wt.% Nanoclay . 20:29 5.0 wt.% Nanoclay D, Diffusivity Coefficient, x 10'12 mzls 10 20 30 Bio-resin [EML] content (%) Figure 6-8. Summary of diffusion coefficients of various biocomposites. 177 50 nm Figure 6-9. TEM micrographs showing nanoclay dispersion and morphology. a) Nanocomposite resin in Biocomposite E (100/0/25) showing good dispersion with partially exfoliated and intercalated morphology in a resin system (scale = 1pm), and b) High magnification image of an intercalated galleryin nanocomposite resin of Biocomposite I (100/0/5.o) (scale = so mm). 178 «Z. f 7.00 um 10 um 10 um Figure 6—10. Tensile fracture surface analysis: a) Generic low magnification failure surface showing fiber-pullout (encircled) and fiber fracture (arrows) regions, (scale bar = 200 um), b) Magnified regions showing fiber-pullout (encircled), fiber fracture (boxed) regions and fiber- matrix interfacial gap (arrows), (scale bar = 50 pm). Fiber-matrix interfacial gap: c) Biocomposite A (100/0/0), d) Biocomposite D (70/30/0), e) Biocomposite I (100/0/5.0), and f) Biocomposite L (70/30/5.0). (Images c to e, Scale bar = 10 pm). 179 Figure 6-11. Matrix regions of Tensile fracture surfaces: a) Biocomposite A (100/0/0), b) Biocomposite D (70/30/0), well blended UPE/EML region, c) Biocomposite D (70/30/0), phase-separated, bio-resin enriched region, d) Biocomposite I (100/0/5.0), and f) Biocomposite I (70/30/5.0). 180 66’ Reference: [6- 1]. [6-2]. [6-3]. [6- 41 [6- 5]. [6- 61 [6-71- [6- 31 [6-9]. I-hq M, Burguefio R, Mohanty AK. Misra, M Hybrid Bio-based Composites from blends of Unsaturated Polyester and Soy Bean Oil Reinforced with Nanoclay and Natural Fibers,” Compos Sci Technol 2008, 68:3344-3351. O’Donnell A, Dweib MA, Wool RP. Natural fiber composites with plant oil based resin. Compos Sci Techn012004, 64:1135-1145. Shibata M, Ozawa K, Terarnoto N, Yosomiya R, Takeishi H. Biocomposites made from short abaca fiber and biodegradable polyesters. Macromol Mater Eng 2003; 288:35-43. Morye SS, Wool RP. Mechanical properties of glass /flax hybrid composites based on a novel modified soybean oil matrix material. Polym Compos 2005, 26:407- 146. Bodros E, Pillin I, Montrelay N, Baley C. Could biopolymers reinforced by randomly scattered flax fibers be used in structural applications. Compos Sci Technol 2007; 67:462-470. Burgueno R, Quagliata, MJ, Mohanty AK, Mehta GM, Drzal LT, Misra M Load bearing natural fiber composite cellular bearm and panels. Composites: A 2004; 35:645-656. Miyagawa H, Mohanty AK, Burgueno R, Drzal LT, Misra M Novel bio-based resins from blends of functionalized soyabean oil and unsaturated polyester resin. J Polym Sci, Part B: Polym Phys 2007; 45: 698-704 Mohanty, A.K, Misra M, I-linrichsen, G. Biofibers, Biodegradable Polymers and Biocomposites: An Overview. Macromol Mater Eng 2000; 276/277:1-24. Nickel], Riedel U. Activities in biocomposites. Mater Today 2003; 6:44-48. [6- 10]. Myagawa H, Mohanty AK, Burgueno R, Drzal LT, Misra M Development of bio- based unsaturated polyester containing functionalized vegetable oils. Ind Eng Chem Res 2006; 45:1014—1018. [6-11]. Pinnavaia 'I], Beall GW. Polymer-Clay Nanocomposites, John Wiley 8: Sons, Ltd., 2001. [6-12]. Le Baron PC, Wang Z, Pinnavaia 'I], Polymer - layered silicate nanocomposites: an overview. Appl Gay Sci. 1999; 15:11-29 [6-13]. Alexandre M, Dubois P. Polymer-layered silicate nancomposites: preparation, properties and uses of a new class of materials. Nhter Sci Eng R 2000; 28:1-63. 181 [6-14]. Hussain F, Hojjati M, Okarnoto M, Gorga RE. Polymer-matrix Nanocomposites, Processing, Manufacturing and Application: An Overview. J Comp Mats 2006; 17: 1511-1575. [6-15]. Myagawa H, Mohanty AK, Burgueno R, Drzal LT, Misra M Characterization and thermophysical properties of unsaturated polyester-layered silicate nanocomposites] Nanosci Nanotech 2006; 6:464-471. [6-16]. HaqM, Burgueno R, Mohanty AK, Misra M Bio-based unsaturated polyester/layered silicate nanocomposites: characterization and thermophysical properties. Composites: A 2009; 40:540-547. [6- 171 Tsai JL, Wu MD. Organoclay effect on mechanical responses of glass/ epoxy nanocomposites] Compos Mater 2008; 42:553-568. [6-18]. Kinloch AJ, Mohammed RD, Taylor AC, Sprenger S, Egan D. The interlarninar toughness of carbon fibre reinforced plastic composites using ‘hybrid-toughened’ rmtrices. J Mater Sci 2006; 415043-5046. [6- 19]. Kinloch AJ, Mohammed RD, Taylor AC, Sprenger S, Egan D. The effect of silica nano particles and rubber particles on the toughness of multiphase thermosetting epoxy polymers. J Mater Sci 2005; 405083-5086. [6-20]. Gojny FH, Wichmann MHG, Fiedler B, Bauhofer W, Schulte K. Influence of nano- rnodification on the mechanical and electrical properties of conventional fiber- reinforced polymers. Composites: A 2005; 36:1525- 1535. [6-21]. Haq M, Burguefio R, Mohanty AK, Misra M Development and Thennophysical Characterization of Clay Nanocomposites from Blends of Linseed oil and Unsaturated Polyester. Composites: A Communicated. [6-22]. Haq M, Burguefio R, Mohanty AK, Misra M Processing techniques for bio-based unsaturated—polyester/ clay nanocomposites: tensile properties, efficiency and limits. Composites: A 2009; 40:394-403. [6-23]. Vlasveld, D.P.N., Groenewold], Bersee, HEN” Picken, S..J Moisture absorption in polyamide-6 silicate nanocomposites and its influence on mechanical properties. Polymer 2005; 46: 12567-12576. [c241 Mohanty AK, Misra M, Drzal LT. Surface Modifications of Naurral Fibers and Performance of the Resulting Biocomposites: An Overview. Campos Interfaces 2001; 8:313-343. [6-25]. Wang L, Wang K, Chen L, Zhang _Y, He C. Preparation, morphology and thermal/mechanical properties of epoxy/ clay nanoclay composite. Composites: A 2006; 37:1890- 1896 [6-26]. I-hq M, Burgrrefio R. Modeling and simulation of bio-based polymer/ clay nanocomposites through a multilevel FE approach. InzProceedings of Sixth 182 International conference on Computation of Shell 8?. Spatial Structures. New York, May, 2008. SessionT-Z-E. 183 Chapter 7. Analytical and Numerical Modeling of Mechanical and Barrier Properties of Clay/ Polymer Nanocomposites 7.] lewd/letter; Polymeric materials are often reinforced by stiff fillers to improve mechanical properties. The efficiency of reinforcement depends on the filler aspect ratio, the filler mechanical properties and the adhesion between the rmtrix and the filler [7-1]. Nanocomposites consisting of highly anisotropic clay platelets (layered alumina silicates, LAS) dispersed in polymeric matrix material are of interest for many important industrial applications [7-2]. Clays, or LAS, are composed of extrernelythin (~1 rim) sheet like platelets that possess very large specific surface areas (SSA) and high aspect ratios. In addition, these platelets have an exceptionally high modulus in tension relative to most polymers and even marryother types of fillers. Such properties enable the platelets, when dispersed in a polymer matrix, to carry a significant component of the applied load. Because of the high surface area and the small inter-particle distances, the platelets can, in principle, significantly alter the properties of the polymer matrix [7-3]. Research has shown that at very low concentrations (~5°/o by weight) the modulus has been increased by 100%. Considerable research has been performed on polymer/ clay nanocomposites and apart from an increase in modulus, it has been observed that clay nanocomposites exhibit multifunctionality by exhibiting enhancement in many Other physical properties including thermal, barrier, flammability resistance and tribological perfomance. In order to efficiently exploit all the benefits that clay/ polymer nanocorrrposites can offer it is important to understand the challenges that are faced relative to conventional fillers. The challenges include: 184 a) Appropriate dispersion of reinforcements in the resin systems b) Interfacial issues: bonding, modeling, prediction of properties in the interfacial regime, interactions and functionalizations etc, and c) Obtaining nano—structural influences and representative information about the clay structure property relationship and nano-mechanical mechanisms. Issues include stress transfer mechanisms and theoretical predictions to validate or explain the experimental results taking into considerations possible mechanisms at the nano- scale. As discussed earlier, one of the major attractions of clay in polymers is the high SSA that provides exceptional enhancement of mechano-physical properties due at lower concentrations. At the same time, high SSA can induce undesirable effects. It introduces strong attractive forces between the clay particles, leading to excessive agglomeration. Hence, it becomes essential to separate them for the resin to penetrate/ infiltrate between them. On the contrary, if the clay particles are not completely exfoliated, it may lead to intercalated morphologies that have lower aspect ratios and less SSA relative to an exfoliated case. Due to the lower aspect ratios and lower SSA, intercalated clay particles have relatively less ability to effectively transfer stresses and hence lead to lower mechanical properties. Hence the role of dispersion is essential to obtain proper SSA and high aspect ratios that increase the efficiency of the system. In this work, the dispersion is performed by using an ultra-sonnicator. The next main issue in clay based composites is the interfacial adhesion between the matrix and clay particles. Sufficient stress transfer needs to take place to efficiently exploit the potential of clay as structural reinforcement. The interfacial bonding can be improved by proper functionalization of the clay surface. The choice of resin system and its compatibility 185 with the nanoinclusion (in this case clay) is very essential to obtain the benefits offered by the nano-dimension. Research has shown that the surface modification of clay surface provides compatibilitywith resin systems and helps in efficient interfacial strengths [7-4]. The third main issue is the need nano-mechanical models or theories that can address the outstanding issues of the polymer - clay interface. Although several theories have been developed, it is important to efficiently relate the nano-mechanical mechanisms to the experimental results. This would help in optimizing mechanical performance and the development of tailorable materials. The polymer nanocomposites used in this research have four rrnjor components: polyester based resin, bio-based resin, natural fibers, and nano-clay inclusions. Bio-based resin is included as it enhances the toughness of the resulting composite. More importantly, it is added as a replacement to the conventional petroleum based resins so as to make the resulting composites more eco-friendly. The purpose of this research is the development of load bearing components made of such biocomposites. In order to achieve this goal, it becomes essential to have a fundamental understanding of each component, its arrangement, proportion, and its effect on each of the mechanical parameter (stiffness, hygro-thermal properties, flammability, etc.) considered. Hence, it is required to find an optimal balance between the components that exhibit the best collective properties while hiding their intrinsic bad behavior. This understanding can be made possible by analytical and computational modeling. As discussed earlier, the interfacial region plays an important role in the enhancement of the mechanical and physical properties of the resulting composite. The addition of bio-resin is suspected to introduce a hyer around the nano-inclusions thereby reducing the efficiency of the stress transfer and hence a reduction in the overall mechanical 186 properties. It is thus essential to have proper modeling techniques to efficiently model, design and Optimize such materials. In this study the modeling of biocomposites can be classified into two main categories: a) modeling of polymer nanocomposites (clay + resin system and no natural fibers), and b) modeling of natural fiber reinforced biocomposites. Prediction of the mechanical properties of discontinuous fiber/ flake composite materials has been a subject of extensive study. As discussed earlier, the modeling of heterogeneous materials (in our case, resin system and clay particles) involves rmny complexities that should take into account the morphologies of the inclusion and resin system. The models should also take into account the various physico-chemical mechanism at the various scales. Numerous micromechanical models have been proposed to predict the elastic constants of discontinuous fiber/ flake composites. These models generally depend on parameters including particle/ matrix stiffness ratio zip/Em , particle volume fraction f), particle aspect ratio M, and orientation. 5; and in denote the elastic modulii of the particle and the nutrix, respectively. The prediction of homogenized properties of heterogeneous naterials has been widely discussed in literature, and good overview is provided by Bohm [7- 5]. Various theories and models have been proposed, and each of these models has its own advantages and limitations. Overall, the homogenization techniques can be classified into [7- 5]: mean field approaches (MFA), variational bound methods, periodic microfield approaches (PMA’s) / unit cell methods (UCM), embedded cell approaches (ECA), windowing approaches (WA), rules of mixtures and empirical formulae, such as Halpin-Tsai In this chapter, section 7.2 will focus on rrricrornechanical modeling of PNC‘s using MFA’s and empirical methods. Section 7.3 will focus on modeling of these PNCs with UCM 187 or Finite Element homogenization schemes, RVE studies etc. lastly, Section 7.4 will corrrpare some experimental results 7.2 Mkmmeeézmz'ee/ 11! ode/zhg off/VC} As discussed earlier, it is essential to obtain the overall properties of a heterogeneous material. Various homogenization techniques exist and have been used widely in the past. The process of homogenization involves finding the average properties at a lower scale due to the rmcroscopic homogeneous defonrntions. In other words, it involves finding elastic constants and stress-strain (constitutive) relations that relate the two scales. Homogenization procedures aim at finding a volume element’s (local) response to prescribed mechanical loads (macrosc0pic). The resulting homogenized behavior is idealized as statistically isotropic or statistically transversely isotropic. The classification of different models is based on the way they analyze / treat the local microscopic responses and then relate it with the nncroscopic behavior. The most commonly used analytical models are Halpin-Tsai and Mori Tanaka (Tandon—Weng and Hui-Shia) models. In this section, these models are briefly compared and discussed. These theories/ models differ in regard to their treatment of the filler geometry; however, they show analogous responses to how composite modulus responds to the filler properties. 7.2.1 Mori Tanaka Estimates Mon-Tanaka (M-T) estimates are popular micromechanical modeling techniques quite successful for conventional composites based on the mean field approach (MFA), namely the Eshelby Equivalent Inclusion method [7-6]. M-T average stress theory was derived on the principles of Eshelby’s inclusion model for predicting an elastic field in and 188 around an ellipsoidal particle in an infinite matrix [7-3]. To account for finite filler concentrations, Mori and Tanaka, however, considered a non-dilute composite consisting of many identical spheroidal particles that cause the matrix to experience an average stress different from that of the applied stress. To satisfy equilibrium conditions the volume average over the entire composite was forced to equal the applied stress [7-3]. One way for achieving the accountability of inter-particle interaction/ stresses consists of approximating the stresses acting on an inclusion, which may be viewed as the perturbation stresses caused by the presence of other inclusions superimposed on the applied far field stress, by an appropriate average rmtrix stress. Effective field theories of this type are generically called Mon-Tanaka (M-T) methods or Equivalent Inclusion Average Stress (BIAS) approaches [7- 5]. Although M-T model is successful, it has some concerns regarding its application to nanocomposites. The first is that although the MT method is predominantly an energy method, it is based on the Eshelby equivalent inclusion method which assumes a dilute suspension of particles; that is, the particles do not interact. Thus, particle interactions are accounted for only at high volume fractions, but are effectively ignored for low volume fractions. For clay polymer nanocomposites (CPNC)s, although reinforcement volume fractions are very low, the large number of particles with high aspect ratio suggests that particle interactions will have a significant influence on mechanical properties. Another concern is that the assumption of a completely random flake orientation suggests that the nano-composite is isotropic, which is not often the case [7-6]. The Tandon - Weng model and Hui-Shia models are the two most commonly cited methods in the literature which are based on the Mon-Tanaka estimates and are briefly discussed in the following sections. 189 .L—mI—‘e-Il had-rut " . Tm - ngMald As discussed earlier, Tandon-Weng used the average stress assumption and Eshleby’s solution to derive complete analytical solutions for the elastic moduli of an isotropic matrix filled with aligned spheroidal inclusions [7-31 With the inclusions aligned along the x; direction, there are five independent elastic constants associated with the transversely isotropic composites. These are the longitudinal elastic modulus E“, the transverse elastic modulus E22 , the in-plane shear modulus 012, the out of plane shear modulus 023, and the plane strain bulk modulus K 23 ,. Their results for the five elastic constants are as follows [77]: a) LagirubulEksu'cModdrc (E11) E“ = A (7-1) Em A + ¢f (Al + 2va2) h = 2A (7.2) Em 2A + f p[—2v,,,A3 + (1 — vm)A4 +(1+ vm)A5A] % =1 + G fp (7-3) "’ W + 2(1 -fp)S1212 190 a OdfpkneSbeeraMw (623) g2; =1+ G f” (8-4) "’ ——m—+2(1—f )52323 G, —G,,, P £2; _ (l + vm )(l - 2vm) (7'5) Km l-vm(1+ 2V12) +fp {2(V12 "Vm)A3 +[1-Vm(1+ 2"rzllAarl/A flMajoraberfiazPa'ssm’s ratio rfthe anpasite (V12) The major direction Poisson’s ratio V12 of the composite is not an independent modulus; it is related to the Other elastic constants by: vlzzzfll—E“ 1 + 1 (7-6) E22 4 023 K23 After finding E11, E22 and 623 from the above equations, V12 is directly related to K 23 and hence Equation (7-5) is implicit and can be solved by a simple iterative procedure. In the above equations, f p is the filler volume fraction, V", is the poisson’s ratio of the matrix, V12 is the major Poisson’s ratio of the composite and A1, A2 , A3 , A4, A5, and A are the functions of the Eshelby’s tensor and the properties of the filler. It is obvious that in the above equations when f p is equal to zero the effective modulus of the composite reduce to that of the untrix and when f p =1 each quantity reduces to that of the inclusions. 191 Equations (7- 1) and (7-2) are dependent upon the shape of the filler, e.g., fiber-like versus disk like ellipsoids, which are accounted for in Eshelby’s tensor. Moreover, Equation (7-2) is used instead of Equation (7-1) when predicting modulus parallel to either major axis of a disk-like spheroid. Figure 7-1 shows the Mari-Tanaka physical representation of glass fibers and disk-like platelets and Table 7-1 lists the different equations used to calculate composite modulii along the three principal orthogonal directions. Hui-Shia’s model was derived from the Tandon-Weng model. Although Tandon- Weng model provides exact solution, it is rarely used in practice, perhaps due to its complexity [7-8]. Hui and Shia [7-8] developed simplified expressions for the effective moduli of unidirectionally aligned two—phase composites from Tandon and Weng’s exact solution by nnking the assumptions that the Poisson’s ratios of the inclusion and the matrix are the same and are equal to 0.5. Of the five independent elastic constants, only E11, E22 are given here; since these are the most complex equations. For other constants, the same equations of the Tandon-Weng model are used as theyare alreadyvery simple. All modulii depend on a geometrical parameter g , which depends on the aspect ratio a . The geometrical parameter g is given by: F f l ——a——3 a(0:2 —1)2 —cosh"l(a)] 6121 2 _13 c g =i (a ) - , (7'7) J—f —a(1—a2)5—cos_l(a):l aSI L(l—a2)—2— e 192 It can be shown thatg(a =1) = 2/3. The case of a >>~1 corresponds to fiber like inclusions whereas a -<-10% EML). A particle overlap factor 7 = 0 , suggests that no overlap occurs, indicating that through paths consisting of only polymer exists in between the particles. Although such morphology does not exist in reality, it seems that the presence of bio-resin allows the permeant molecules to travel faster. Moreover, the presence of intercalated morphologies, considerably reduces the particle overlap and tortuous path. The combination of intercalation and the presence of a faster permeant path due to the presence of bio-resin seems like a through path analogous to a no particle overlap. This is supported by the good agreement between theoretical predictions and experimental results for bio-based 215 nanocomposites. Moreover, for particle overlap factor7 = 0 , the model predictions were higher than experimental results for all nanocomposites except neat UPE with 2.5 wt.%. The theoretical model used does not hold for EML contents of 30% or more. The Van Es model [7-20] is not intended to model nanoclay reinforced bio-blends. Nevertheless, understanding the tortuous path and the effect of bio-resin addition on the penneant path can help in understanding diffusion phenomena taking place in the bio-based nanocomposites. Overall, the diffusion study showed promise in the use of environmental friendly bio-based resin systems reinforced with nanoclay. The loss in barrier properties due to incorporation of bio-resin were not only recovered but enhanced by addition of nanoclay. Moreover, the results are indicative that the banier properties of resulting nanocomposites can be tailored by controlling the concentrations of bio-resin and nanoclay content. The ability to recover the lost barrier properties due to bio- resin addition by incorporating nanoclay can allow the use of higher amounts of bio-resin, making the resulting nanocomposites more environmental friendly without sacrificing performance. 7.10 Concétsz'on Modification of polymers at the nanoscale by addition of nano-inclusions has been shown to enhance mechanical properties and provide multifunctionality to the resulting polymer nanocomposites . However, quantitative evaluation of the mechanisms leading to these property enhancements as well as those that limit the performance of polymer nanocomposites is still limited due to experimental constraints. Computational simulation seems to be a good way to fill this gap. However, the simulation approach must be able to account for features of these materials at the nano- and micro- scales. 216 1&3 IV»?! .43. ... v.. ark. . ..1.,.L— Mechanics-based models have proved to be successful in predicting homogenized properties where the filler length scale is in the order of tens of microns or larger. In nanoclay reinforced composites, depending on the processing, the clay morphology varies significantly. Exfoliated clays may be in range of nanometers and highly intercalated clay agglomerations my be in the order of micrometers. Moreover, it is commonly agreed that polymer clay nanocomposites exhibit both exfoliated hierarchical morphologies. Clearly, mechanics-based models cannot handle this complexity. The final and most important issue in modeling of polymer/ clay nanocorrrposites is the particle-polymer interface. Depending on the chemical compatibility / frmctionalization of the nanoclay and polymer, the resulting interfacial properties will vary considerably. Additionally, in intercalated clay morphologies the properties of the material in galleries is not fully understood. The above mentioned issues specific to polymer/ clay nanocomposites make modeling of such composites challenging. Finite element (FE) based modeling and simulation can address some of these issues and enable modeling of interfaces, different particle sizes and morphologies, nonlinear rmterial properties and complex loading. In this chapter, the analytical models were implemented and finite element models were developed to model polymer clay nanocomposites. Simulations were compared with experimental results for mechanical and diffusion properties. The accuracy of the simulations and the analytical model depends upon the detailed modeling of the microstructure. The models with least assumptions and most realistic modeling, as expected yielded better agreement with experiments. 217 Z]! Taéé’: é figure: Table 7-1. Modulus along different filler orientations (fibers and platelets) - Mori — Tanaka Table 7-2. Important issues limiting ability to model the nanocomposites [7-3] ISSUE THEORY EXPERIMENTAL Filler Shape and Size Uniform shape I Non-uniform shape Constant dimensions I Distnbution of lengths and thickness I Imperfect exfoliation Filler Orientation Unidirectional I Some degree of misalignment Filler Interface Filler and matrix are I Imperfect bonding well bonded between filler and __ matrix Filler Modulus Assumes filler modulus I Filler is anisotropic is same in all directions Matrix considerations Assumes matrix is I Polymer chain isotropic orientation I Presence of polymer crystallites Filler concentration No particle - particle I Particle — particle Effects interactions interactions and Ignores changes in agglomerations viscosity I Oranges in viscosity No particle can alter morphology agglomeration during injection molding I Oranges in crystalline morphology 218 Table 7-3. Experimental matrix showing specimen identification numbers and variation in clay and EML contents Clay Amount of EML replacing UPE in Content UPE-EML blend. (as °/o of UPE) (wt%) 0 1o 20 30 0.0 1 2 3 4 2.5 5 6 7 8 5.0 9 10 11 12 Table 7-4. Diffusivity coefficients of UPE/EML nanoclay composites. Diffusivity Coefficients, D x 10'12 mzls Clay Amount of EML replacing UPE in Content UPE-EML blend. (as % of UPE) (wt.%) 0 10 20 30 0.0 3.666 6.350 7.965 12.781 2.5 2.458 3.019 3.714 12.341 5.0 1.050 2.071 2.522 8.385 Table 7-5. Comparison of diffusivity coefficients of various nanocomposites with virgin UPE, expressed in percentage. Highlighted region shows nanocomposites that had equal or better barrier properties than virgin UPE. Percentage increase of Diffusivity Coefficients, D relative to virgin UPE Clay Amount of EML replacing UPE in Content UPE-EML blend. (as % of UPE) (wt.%) 0 10 20 30 0.0 o 73 117 249 2.5 -33 -18 1 237 5.0 -71 .44 -31 129 219 Table 7-6. Comparison of experimental results and theoretical predictions for various particle overlap factors, at nanoclay content of 2.5 wt.% Diffusivity Coeficients, D x 10'” mz/s Clay Content = 2.5 wt.%, aspect ratio=130 Amount of EML replacing UPE in UPE-EML blend (as % of UPE) 0 10 20 30 Experiments 2.458 3.019 3.714 12.341 Theory, 7 a 0.25 1.583 2.742 3.439 5.519 Theory, 7 =- - 0.50 1.505 2.608 3.271 5.249 Theory, 7 - = 0.00 1.872 3242 4.067 6.526 Table 7-7. Comparison of experimental results and theoretical predictions for various particle overlap factors, at nanoclay content of 5 wt.% Diffusivity Coefficients, D x 10'12 m2/s Clay Content = 5.0 wt.%, aspect ratio =130 Amount of EML replacing UPE in UPE-EML blend (as % of UPE) 0 10 20 30 Experiments 1.050 2.071 2.522 8.385 Theory, 7 = 0.25 0.912 1.580 1.982 3.180 Theory, 7 = = 0.50 0.837 1.450 1.818 2.918 Theory, 7 - .. 0.00 1.248 2.162 2.712 4.352 220 H-T and M—T model H-T model M-T model 0. »1 — prolate shaped / (1 « 1 — oblate shaped / disc—like fiber-like Figure 7-1. Physical representations, coordinate systems, Mod-Tanaka model — filler orientations [7-3] 3.5 —0— Halpln-Tsal - (fp =0.01) -----v- Halpin-Tsai - (fp =0.02) 3.0 _ —I— Halpin-Tsai -(fp =0.03) E11IEm 1 100 Variation of Aspect Ratio (Log scale) 1000 Figure 7—2 Variation of Longitudinal Modulus with Aspect ratio. 221 Figure 7-3. Variation in length and subsequently aspect ratio, across a dish—like platelet [7-3]. 30 I + Tandon-Weng, L/t = 0.01 I + Tandon-Weng. M = 1 25 j -I— Tandon-Weng, lJt = 10 j + Tandon-Weng, lJt = 25 20 _- + Tandon-Weng, th= 100 E . llJ . \ 15 .. ‘- . ‘- . Lu . 10 '- 5% 0.0 0.2 0.4 0.6 0.8 1.0 Volume fraction (%) Figure 7—4. Variation of Longitudinal Modulus (E11) with variation in f}, and (L/t) [7—7]. 222 30 I + Tandon-Weng, Lit = 0.1 I + Tandon-Weng, th = 0.2 l 25 j -I— Tandon-Weng, th = 0.5 ‘ + Tandon-Weng, th = 1 20 _ + Tandon-Weng, th = 100 15{ E22 / Em 1o{ I u r v v I . 0.0 0.2 0.4 0.6 0.8 1.0 Volume fraction (%) Figure 7—5. Variation of Transverse Modulus (E22) with variation in j} and (L/z) [7-7] 30 I + Tandon-Weng, th = 0.1 I + Tandon-Weng. L/t = 0.2 25 j -I— Tandon-Weng, th = 0.5 ' + Tandon-Weng, L/t = 1 20 _ + Tandon-Weng, th = 100 155 G12/Gm 10f ' . 0.0 0.2 0.4 0.6 0.8 1.0 Volume fraction (%) Figure 7—6. Variation of In-Plane Shear Modulus (G12) with variation in j}, and (LI/t) [7-7] 223 30 I + Tandon-Weng, L/t = 0.1 I + Tandon-Weng, th = 0.2 25 j —I— Tandon-Weng, M = 0.5 ‘ + Tandon-Weng, L/t = 1 20 _- + Tandon-Weng, th = 100 623 I Gm 0.0 0.2 0.4 0.6 0.8 1.0 Volume fraction (%) Figure 7-7. Variation of Out of—Plane Shear Modulus (G23) with variation in f}, and (L/t) [7'7] + Tandon-Weng, th = 0.1 14 _' + Tandon-Weng, th = 0.2 -I— Tandon-Weng, L/t = 0.5 12 : + Tandon-Weng, th= 1 + Tandon-Weng, M = 100 K23/Km 0.0 0.2 0.4 0.6 0.8 1.0 Volume fraction (%) Figure 7-8. Variation of Plane Strain Bulk Modulus (K3) With variation in J; and (L/ t) [7- 224 0.36 . W 0'34 _ M + T-W, fp = 0.0 032 _: + T-W, fp = 0.1 : + T-W, fp = 0.3 0.30 f + T-W, fp = 0.5 2 + T-W, fp = 0.7 £3 0.28 I -O- T-W, fP =1.0 > 2 V V 7W 0.26 I Tandon-Weng (T-W) 0.24 -: 0.22 E 0.20 : WHO-HM. 0,13‘....................rfi..fi_r 0 10 20 30 40 50 Volume fraction (%) Figure 7—9. Variation of Major Poisson’s Ratio (v19 with variation in 1;, and (L/ t) [7—7] 14 -o— Hul-Shia (th=10) —v— Hui-Shia (th=50) 12 105 E11IEm . v 0.0 0.1 0.2 0.3 0.4 0.5 Volume fraction (%) Figure 7-10. Variation of Longitudinal Modulus (E,,) with variation in lit and (LI/z} Hui- Shia Fiber Like inclusions, (L/ t = 10, 50) [7—7]. 225 2.6 , : -o— Hui-Shia (Lit=10) 2.4 -j + Hui-Shia (th=50) 2.2 - 2.0 E 1.8 { d 1.6: 1 4 1.4: E22/Em 1.25 1.0 0.8-fivvitvvr...... 0.0 0.1 0.2 0.3 0.4 0.5 Volume fraction (%) l I I I I I I I rfifi Figure 7-11. Variation of Transverse Modulus (E22) with variation in f}; and (L/t)- Hui- Shia Fiber Like inclusions, (L/ t = 10, 50) [7-7]. 12 . . -o- Hui-Shia (LIt=0.01) 1 :+ Hul-Shla(LIt=0.1) o E11/Em O O.....,.... 0.0 0.1 0.2 0.3 0.4 0.5 Volume fraction (%) Figure 7—12. Variation of Longitudinal Modulus (E71) with variation in 1;, and (L/t)- Hui- Shia Flake Like inclusions, (L/ t = 0.01, 0.1) [7-7]. ' I I I I I I I I I I I I 226 ] -o- Hul-Shla(Ut=0.01) + Hul-Shla (Ut=0.1) E22/Em I T I I I l I I I' f j 0.....,....,.... 0.0 0.1 0.2 0.3 0.4 0.5 Volume fraction (%) Figure 7-13. Variation of Transverse Modulus (I322) with variation in [a and (L/t)- Hui- Shia Flake Like inclusions, (L/ t = 0.01, 0.1) [7-7] 3.5 I + Halpln-Tsal (th = 20) . + Halpin-Teal (th = 40) 3.0 I —I— Halpin-Tsai (th = 60) - + Halpin-Teal (th = 80) I + Halpin-Tsai (Lit = 100) 2.5 r / I / 2.0 ' I 1.5 2‘ / 0.00 0.02 0.04 0.06 0.08 0.10 4 E11IEm I I j’ r T T I Volume fraction (%) Figure 7-14. Variation of En with respect to particle volume fraction(fp) — Halpin-Tsai model [7-7] 227 E11IEm j + Halpin-Tsai (fp =0.01) - + Halpin-Tsai (fp =0.02) 5 .‘ + Halpin-Tsai (fp = 0.03) 4; i '1 J 3 a I #4 2 - 1........IT-...,. 0 100 200 300 400 Particle I matrix Stiffness Ratio (EplEm) Figure 7-15. Variation of B11 /Em. Halpin —Tsai model {7.1} Figure 7-16. Kinematic Periodic Boundary Conditions 228 a) Truncated inclusion at the edge of RVE b) Applying the particle periodicity by including the truncated particle on the other edge Figure 7-17. Schematic Representation of Particle Periodicity in RVE : t l , ' I V“ ' I ' I 7.7, l i > . 51 , ,_ 2, 00 0.001 0.000 0.003 0.004 0.005 0.005 0.007 0000 0.009 0.01 Max-1013‘! Figure 7-18. Particle Periodicity. Circled regions show the inclusions of the truncated particles. RVE dimension of 500 x1000 nm 229 Mort-m §§§§§§§ t'é so" 0‘; ""3: 7-. , , o immanmananmmomttm Martin]?! (a) ‘.’II‘\I!]llll’lZ l Hulli'l: lllll lllllt'll lhlrt‘ l'.‘\|’.l I (b) Figure 7-19. a) Sample RVE with 2% wt. clay concentration, and, b) equivalent FE model 230 ‘.'N'\'/[IIIF1. l lltll) r "llis‘l iLIlmllIntlulmn'nlrui. pl-r’llili Il."'lllll lllrlvll llllll Figure 7-20. Resulting stress contours (Sl l-Von—Mises) from RVE FE analyses Figure 7—21. Partly exfoliated and partly intercalated RVE model 231 Figure 7-22. RVE with Mesh thickness = 1 element per clay particle (RED) Figure 7-23. RVE with interface, Mesh thickness 2 1 element per clay Particle (RED) and Interface (green) shown above Figure 7-24. RVE with Interface, Mesh thickness: 4 elements per Clay Particle and Interface (red) 232 2.8 2.6 - A 2.4 - 2.2 .. 2.0 - EnlEm a 1.8 - #0 1.6 - 1.4 - / -o- fp=0.015; FEA (RVE) -I- fp=0.015; M-T estimate 1.2 - ... fp=0.025 FEA (RVE) -0- fp=0.025 MT- estimate 1.0 t r u t u 0 100 200 300 400 500 600 Variation of EplEm Figure 7-25. Comparison of RVE Analyses with theory, Variation of Longitudinal Stiffness with varying Ep/Em 2.6 2.4 - 2.2 ~ 2.0 - Ills ‘3: 1.8 - |.I.I 1.0 - 1.4 - —o— fp=0.010; FEA (RVE) 1 2 _ -l- fp=0.010; M-T estimate ' -d- fp=0.020 FEA (RVE) + fp=0.020 MT- estimate 1.0 I l l 0 100 200 300 400 Variation of Lit Figure 7—26. Comparison of RVE Analyses with theory, Variation of Longitudinal Stiffness with varying aspect ratios 233 CTE (composite) I CTE (matrix) E11, Longitudinal modulus 2.5 1.5 1 .2000 0.6000 0.2000 0.0000 0.5 -— — o L r " “i ' n r i " 1 20 40 60 80 100 120 E11 - Interface Modulus as % of Matrix modulus Figure 7-27. Study of Interface properties through RVE Analyses 1.0000 IK ~ ~—— o.eooo -_ fl +CTE - dir11JI 0.4000 - \\ CLAY CTE I MATRIX CTE I T l F 0 0.02 0.04 0.06 0.08 0.1 Volume fraction (fp) Figure 7-28. Study of CTE through RVE analyses 234 I - Finite Element Anal is Mate'h' ‘1°°% UPE _ - Experimental Data ys Nano Inclusion: Cloislte SOB 5 I - Theory- Halpin Tsai (Modified) Tensile Modulus (Gpa) U 0.5 1.0 Clay Concentration (weight %) Figure 7-29. Comparison of Tensile Modulus from theory, FEA — UCM and Experimental Results for 100% Neat UPE with varying clay content 235 - 100% Exfoliated - 50% intercalated - 100% Intercalated - Experimental Data D1 0010/0 = 3.67 x 10'12 m2/s D, Diffusivity Coefficient, x 10'12 m2ls 2.5 wt.% 5.0 wt.% Clay Content Figure 7-30. Diffusion modeling using FE—based RVES. a) concentration contours showing diffusion of water from bottom to top, b) tortuous flow path simulation, low magnification, c) tortuous flow path simulation, high magnification, and d) Simulation and Experimental results comparison. 236 1 2 —o— 100/0/0 ' . + 90/10/0 ; + 80/20/0 - + 70/30/0 1.0 - I . O 0.8-I . ‘ 12 Temp= 50 “C M, . M 0.6— 9 D - v . I D 5 0.4 - 3 0.2 - M, _ 4 ’Dt i .. 71d 2 0 . o 10 20 so 0.0 ,r,1 ..............T 0 100 200 300 400 500 600 700 300 t°-5/d , (s°-5/mm) Figure 7—31. Diffusivity plots (M , / M ,0 vs. x/I/d ) of neat resins (no clay). Symbols indicate experimental data and solid lines the exponential fit. Inset shows, the initial slope / diffusivity coefficients (D x 10'12 mZ/s) as a function of bio-resin content 1.2 1.0 i v o v e I : ' I 0.8 - . V M . —' 0.6 - Temp= 50 °C Mm ' I v M , _ D! 0.4 - _ 2 . v M, ml I . I 02‘. + 100/0/00 + 100/0/2.5 + 100/0/50 0.0 ' I ' I ' ' ' ' I ' ' ' ' I ' ' ' ' I ' ' ' ' I ' ' ' l ' ' ' ' I ' ' ' ' o 100 200 300 400 500 600 700 500 t°-5/d , (s°-5Imm) Figure 7—32. Diffusivity plots (M , / M a, vs. JI/d ) of UPE (no EML) with varying clay contents. Symbols indicate experimental data and solid lines the exponential fit. 237 ill '1 ELLE-J . 1.2 1.05 I ‘ C j o 0.8-- - V M I —* 0.6 - Temp= 50 0C Mm : o 2 M, =4 Dt ' 0.4 " M ml 2 I) I I 02: -0— 90/10/00 + 90/10125 . + 90/10/5.o 0.0 ....... o 100 200 300 400 500 600 700 000 t°‘5/d , (so'slmm) Figure 7-33. Diffusivity plots (M, /M., vs. «ff/d ) of UPE/EML blend containing 10%EML, with varying clay contents. Symbols indicate experimental data and solid lines the exponential fit. 1.2 1.0-I I ‘ V _ O 0.6-I M V = o l 0_6_ Temp 50 C M. : M, _ Dt _' —— ’—2 0.4 _ I Mo nfl . I 02: —o— 80i20/0.0 + 80/20/2.5 . + 80/20/5.0 0.0 ..........r,-..,fi.... ......m 0 100 200 300 400 500 600 700 800 t°'5ld , (SM/mm) Figure 7—34. Diffusivity plots (M, / M a, vs. JI/d) of UPE/EML blend containing 20%EML, with varying clay contents. Symbols indicate experimental data and solid lines the exponential fit. 238 1.2 1.03 p- -- -‘ A _ v- - - V 0.8-I ' 3 ' T — 50 °C M 0.6 - emp— , . M., . l M, 4 ]Dt . _ 2 0.4 . Ma 7111 0-2 ‘ -o- 70/30/00 4 + 70/30/2.5 ‘ + 70/30/50 0.0 ..rfi..,..............,.............. o 100 200 300 400 500 600 700 000 t°'5ld , (so'slmm) Figure 7—35. Diffusivity plots (M, / M D vs. 1” / d ) of UPE/EML blend containing 30% EML, with varying clay contents. Symbols indicate experimental data and solid lines the exponential fit. 3 - 0.001% I - 2.5 wt.% 12: - 5.0 wt.% Baseline UPE D. Diffusivity Coefficient, x 10'12 mzls O) Bio-resin [EML] content (%) Figure 7-36. Summary of diffusivity coefficients of all bio-based nanoclay composites in this study 239 n, Diffusivity Coefficient, x 10'12 mzls O 10 20 Bio-resin [EML] content (%) Figure 7-37. Comparison of experimental results and theoretical predictions for various particle overlap factors, corresponding to clay content of 2.5 wt. %. I - Expt. D, Diffusivity Coefficient, x 10'12 mzls -5 0 5 10 15 20 25 Bio-resin [EML] content (%) Figure 7—38. Comparison of experimental results and theoretical predictions for various particle overlap factors, corresponding to clay content of 5 wt. °/o. 240 Z 12 Reference; [74} [7-2] [7-3] [7-4l [7-5] [7-61 [7-7l [7-31 [7-9l [7-10] [7-11] [7-12] [7-13] Sheng N, Boyce MC, Parks DM, Rutledge GC, Abes JI, Cohen RE. Multiscale micromechanical modeling of polymer/ clay nanocomposites and the effective clay particle. Polymer 2004, 45:487-506. Brune DA, Bicerano J. Micromechanics of nanocomposites: comparison of tensile and compressive elastic modulii and prediction of incomplete exfoliation and imperfect alignment moduhrs. Polymer. 2002, 43:369-387. Fomes TD, Paul DR. Modeling properties of nylon6 / clay nanocomposites using composite theories. Polymer. 2003, 44:4993 - 5013. Le Baron PC, Wang Z, Pinnavaia 'IJ. Polymer - layered silicate nanocomposites: an overview. Applied Clay Science 1999; 15:11-29 Bohm H]. A short introduction to basic aspects of continuum micromechanics, A CDL-FMD Report 3-1998, Vienna Institute of Technology, Italy (1998). Fertig HI RS, Gamich MR. Influence of constituent properties and microstructural parameters on the tensile modulus of a polymer / clay nanocomposite. Composite Science and Technology. 2004, 64:2577-2588. Tandon GP, Weng G]. The effect of aspect ratio of inclusions on the elastic properties of unidirectionally aligned composites. Polymer composites. 1984, 5:327-333. Hui CY Shia D. Simple Formulae for the Effective Modulii of Unidirectional Aligned Composites. Polymer Engineering and Science. 1998, 38:774-782. Halpin JC Stiffness and expansion estimates for oriented short fiber composites. Journal of Composite Materials. 1969, 3:732-735. Haq M, Burgueno R, Mohanty AK, Misra M Bio-based unsaturated polyester/ layered silicate nanocomposites: characterization and thermophysical properties.” Composites - Part A. 2009; 40:540-547Gusev AA. Representative volume element size for elastic composites: a numerical study. Journal of Mechanics and .Physics of Solids 1997, 45:1449- 1459. Feyel F. A multilevel finite element method (FBI) to describe the response of highly nonlinear strucutures using generalized continua. Computational Methods in Applied Mechanical Engineering. 2003, 192: 3233-3244. Smit RJM, Brekelmans WAM, Meijer HEH. Prediction of the mechanical behavior of nonlinear heterogeneous systems by multi—level finite element modeling. Computational methods in Applied Mechanical Engineering. 1998, 155: 181-192. Olson GB. Computational design of hierarchically structured materials. Science. 1997, 277:1237—42. 241 [7-141 [7-151 [7-16l [7-17] l7-13l [7-19l [7-20] [7-211 [7-22] [7-231 [7-24] [7-25] [7-261 Kim JK, Hu C Woo RSC Sham ML. Moisture barrier characteristics of organoclay- epoxy nanocomposites. Corrrposites Science and Technology. 2005; 65:805-813. Vlasveld DPN, Groenewold J, Bersee HEN, Picken SJ. Moisture absorption in polyarnide-6 silicate nanocorrrposites and its influence on mechanical properties. Polymer 2005; 46:12567-12576. Rana HT, Gupta RK, GangaRao HVS, Sridhar LN. Measurement of moisture diffusivity through layered-silicate nanocomposites. Materials, Interfaces, and electrochemical phenomena, AIOrE Journal 2005; 51:3249-3256. Gorassi G, Tortora M, Vittoria V, Galli G, Oriellini E. Transport and mechanical properties of blends of poly (e-caprolactone) and a modified rnontrnorillonite poly (e-caprolactone) nanocorrrposites. Journal of Polymer Science; Part B: Polymer Physics. 2002; 40:1118-1124. Pourabas B, Raeesi V. Preparation of ABS/montmorillonite nanocomposite using a solvent / non solvent method. Polymer. 2005; 46: 5533-5540. Luduefia IN, Alvarez VA, Vazques A. Processing and rrricrostructure of PCI/ clay nanocomposites. Materials Science and Engineering A. 2007; 460-461:121-129. Van Es MA, Polymer clay nanocomposites. The importance of particle dimensions. PhD Thesis. Delft Institute of Technology; 2001. Haq M, Burgueno R, Mohanty AK, Misra M Processing techniques for bio-based unsaturated-polyester/ clay nanocorrrposites: tensile properties, efficiency, and limits. Composites. A, 2009, 40:540-547. Drozdov AD, Orristiansen J deC, Gupta RK, Shah AP. Model for Anormlous Moisture Diffusion through a Polymer-Gay Nanocorrrposite. Journal of Polymer Science: Part B: Polymer Physics. 2003; 41:476-492. Oren X, Papathanasiou TD. Barrier Properties of Flake-Filled Membranes: Review and Numerical Evaluation. Journal of Plastic Film and Sheeting. 2007; 23:319- 346. Nrelsen LE. Models for the Permeability of Filled Polymer Systerm. Journal of Macromolecular Science, Part A. 1967; 1(5):929-942. Pritchard G. Reinforced Plastics Durability. CRC Press, FL, USA. 1999. Haq M, Burgueno R, Mohanty AK, Misra M Hybrid Bio-based Composites from blends of Unsaturated Polyester and Soy Bean Oil Reinforced with Nanoclay and Natural Fibers. Composites Science and Technology. 2008, 68:3344-3351. 242 Chapter 8. FE-based Multiscale Homogenization"2 J? I Aértmct Multi-level finite element (FE) methods have gained importance for modeling heterogeneous materials as they offer the ability to model microgeometries in detail and permit incorporating nonlinear material behavior. A multi-level FE computation does not require any constitutive equations to be written at the macroscopic scale as all non-linearites are obtained from separate FE analyses at lower hierarchical (micro/nano) scales. In this work, a multi-level FE algorithm was implemented to model the tensile response of hybrid bio-based materials. A two-level (nano-micro) multi FE scherrre was used to predict the tensile response of two-phase (nanoclay + matrix) and three-phase (nanoclay + perm-resin + bio-resin) bio-based nanocomposites. Also, a three-level multi-FE (nano-micro-macro) algorithm was implemented to predict the tensile behavior of biocomposites (natural fiber + nanoclay + petro-resin + bio-resin). The nanocomposite geometry was modeled as realistic as possible using a PE based representative volume elerrrent (RVE) taking into account random nanoclay distribution and hierarchical nanoclay morphologies. The three- level multi- FE scheme modeling the hybrid fiber-reinforced composite was aimed to show the robustness of the multi-FE approach and not necessarily to model the material with high accuracy, and hence a simplified RVE was used. The multi-FE simulations agreed very well with the average experimental response in the initial stiffness regirrre and seem to deviate 1 Partial results from this chapter were presented in a conference proceeding: Haq, M and Burguefio, R. “Modeling and simulation of bio-based polymer/ clay nanocorrrposites through a multilevel FE approach,” Computation of Shell 8c Spatial Structures, Spanning Nano to Mega, IASS-IACM 2008. Ithaca, New York, USA May 2008. 2 Complete results to be submitted to Journal of Composites Science and Technology. 243 away near ultimate, for nanocomposiets (two-level) and hybrid corrrposites (three-level). The RVES that depict the rrraterial more realistically, namely intercalated morphologies and three- phase idealized RVES, performed better than exfoliated and two-phase RVES. One of the rrrain drawback of the multi-FE algorithm is that it basically is a sophisticated homogenization/ averaging scheme, which can account for non- linearities but cannot model damage. Once damage occurs the periodicity of the RVE ceases to exist and violates the assumptions of multi-FE method. Also, stress-concentration and numerical anomalies (if any) are averaged out and hence the scheme may over-predict the response. Overall, multi- FE algorithm hold great promise in modeling different length scales within the continuum regime. 6? 2 lam/action Modification of polymers at the nanoscale, by addition of nano-inclusions has been shown to enhance rrrechanical properties and provide multifunctionality to the resulting polymer nanocomposites [8-1]. However, quantitative evaluation of the mechanisms leading to these property enhancements as well as those that limit the performance of polymer nanocomposites is still limited due to experimental restrictions. Computational simulation seems to be a good wayto fill this gap. However, the simulation aiming to determine overall macro-scale properties must be able to account for features of these naterials at the nanoscale. The enhancement gained from nanoscale fillers (e.g. nanotubes, silica, layered silicates, etc.) is experimentally well established. Many analytical and numerical models are available to predict the overall effective/homogenized properties of these heterogeneous polymer-filler systems. It is commonly accepted that the overall effective physico-mechanical 244 properties of a polymer-filler material depends on the type of the filler, its aspect ratio, modulus of the filler, modulus of the matrix, concentration of the filler, etc. Hence, prediction of the homogenized properties of heterogeneous materials is widely documented in literature. Various theories and models have been proposed, each with their own advantages and limitations. Overall, homogenization techniques can be classified into [8-2]: mean field approaches, variational bormd methods, periodic rrricrofield approaches / unit cell methods, embedded cell approaches, windowing approaches, rules of mixtures and semi- empirical formulae such as Halpin-Tsai methods, etc. The simplest method of homogenization is the rule of mixtures wherein the overall properties are calculated as an average over the respective properties of the constituents weighed by their volume fractions. Clearly, this method takes into account only one microstructural property, the volume fraction, and is applicable for simple geometries and linear material properties [8-118—2]. A more sophisticated method is the self-consistent or effective-medium theories developed by Eshelby [8-3]. Eshelby’s theory was further developed by various researchers including Hashin [8-4], Hashin and Shtrikrnan [8-5], Hill [8-6], and others [87]. Effective overall properties are obtained as an analytical solution of a boundary value problem for a spherical/ ellipsoidal particle in an infinite matrix. In order to take into account for the particle-particle interactions, Mon-Tanaka [8-2] modified the Eshelby inclusion theory. Mon-Tanaka estimates are commonly used to estimate elastic properties of particulate composites and are found to agree reasonably with experimental results. Extended, or modified, Mari-Tanaka schemes have been proposed that also take into account variations in fiber orientations and fiber lengths using probability distribution functions [8-218- 8] . 245 Another mathematical approach is the asymptotic homogenization medrod reported by Bensoussan et al. [8-9] and Sanchez [8-10]. In this rrrethod “natural length parameter,” which is the ratio of characteristic size of the heterogeneities to a measure of rmcrostructure, is used in an asymptOtic expansion of the displacement and stress fields. Variational principles are then used to create a link between the scales [8-218-7]. Among the computational methods, unit cell methods have become widely used. The success of these methods depends on the selection of the unit cell or the representative volume element (RVE). The RVE should be selected such that it is large enough to represent the microstructure and small enough to allow efficient computational modeling. The corresponding RVE’s are then analyzed with proper loading and boundary conditions via analytical or numerical methods. Unit cell methods allow modeling of micro-scale geometries in great detail (specifically in finite element based unit cell methods) and enable investigation of the influence of different geometrical features, interfaces, interparticle interactions, etc., on the overall rmterial response. However, since these approaches fom'rulate the macroscopic constitutive relations based on the behavior of a single RVE subjected to a given loading history, they are in fact successful only for srmll deforrmtions [3:213:71- Unlike most conventional fillers, the modeling of nanoclays, or layered silicates, pose new and unique modeling issues such as: a) particle size, b) hierarchical modeling, and c) interface and gallery properties [8-11]. Analytical micromechanical models have proved to be successful in predicting homogenized properties of conventional composites where the filler length scale is in the order of tens of microns or larger [8-11]. In nanoclay reinforced composites, depending on the processing, the clay morphology varies significantly. Exfoliated clays rmy be in range of nanometers and highly intercalated clay agglomerations 246 may be in the order of micrometers. Moreover, it is commonly agreed that polymer/ clay nanocomposites exhibit both exfoliated and intercalated (hierarchical) morphologies. Clearly, most micromechanical models cannot handle this complexity [8-11]. More importantly, it shows the limitation in modeling hierarchical rrrorphologies that can vary by orders of magnitudes. The concepts of “matrix” and “particle” that are well-defined in micromechanical models can no longer be directly applied to polymer-clay nanocomposites. Researchers have tried to address this issue by defining an “effective particle” or a “pseudo inclusion.” Sheng et al. [8-11] employ an “effective particle” in analytical and numerical (finite elenrent) micromechanical models. Similarly, Brune and Bicerano [8-12] developed a pseudo inclusion model using the Halpin-Tsai empirical model by substituting the effective pseudo inclusion size. Miyagawa et al. [8-13] used the same concept in Mari-Tanaka models. The aforementioned studies with the concept of “effective particle” have shown to model the hierarchical nature of polymer/ clay nanocomposites successfully and have reported good agreement with experimental data [8-1118-1218-13]. The final and most important issue in modeling of polymer/ clay nanocomposites is the particle-polymer interface. Depending on the chemical compatibility, or functionalization, of the nanoclay and polymer the resulting interfacial properties will vary considerably [8-14]. Additionally in intercalated clay morphologies, wherein the clay platelets are in stacks (called galleries), the properties of the material in the galleries is not fullyunderstood [8-11]. The above mentioned issues specific to polymer/ clay nanocorrrposites rmke modeling of such corrrposites challenging. Finite element (FE) based modeling and simulation address these issues and enable rrrodeling of interfaces, different particle sizes and morphologies, non-linear nraterial properties and complex loading [8-2]. Due to the complex hierarchical nature of polymer/ clay nanocomposites multiscale models are considered in this 247 work. Not limited to polymer/ clay nanocomposites, rrrultiscale models or direct micro- macro mahods have been used for homogenization of different heterogeneous materials[8- 7I8-15}[8-19]. These models have also been termed as “multiscale methods[8-1518-16I8- 17],” “FE2 methods [8-18],” “integrated methods[8-19],” and “multi-level methods[8- 18]”. These methods do not use a material model for the macrostructure, but rather estimate the relevant nncroscopic stress-strain relationship by performing numerical homogenization on W (RVEs) that define lower scale heterogeneities. The link between the macro and micro behavior is then obtained by volume averaging [8-718-1518- 16]. Thus, computational multi-level approaches using finite element procedures offer the ability to model micro- and nano-scale geometries in greater detail and incorporate non- linear material behavior with minimal assumptions of inter-particle behavior or macro-scale constitutive response. In the multi-level FE approach presented here, a unique RVE is assigned to each macroscopic finite element integration point. The macroscopic defonmtion gradient at the respective integration point is applied as boundary conditions to the associated RVE. The tangential stiffness matrix is also derived from the RVE using a computational homogenization procedure as presented by Breuls et al. [8-15]. . This multi-level approach has been validated for a perforated sheet of hypoelastic material by Smit et a1 [8-16], bending of porous aluminum by Kouznetsova et al. [8-7] and study of local tissue defonmtions by Breuls et al.[8-15]. In this study, two types of composites, namely: a) nanocomposites, consisting of petro-resin (unsaturated polyester, UPE), bio-resin (epoxidized methyl linseedate, EML) and nanoclay (Cloisite 30 B" ), and b) hybrid biocomposites, consisting of micro-sized (~25 mm long) industrial hemp fibers and nanoclay in blends of petro- (UPE) and bio-resin (EML), were modeled using the multi-level FE approach. The non-linear 248 rmterial behavior of bio—based polymers used in this work is defined in the form of hyperelastic models (obtained from experimental data) at the nanoscale RVES. The tensile behavior of nanoclay reinforced bio-based polymer systems and fiber reinforced hybrid biocomposites is evaluated using this multi-level finite-element approach and its performance is compared with experimental data. 6?] Mir/1714’ w/Appmacfi 8.3.1 Materials The nanocomposite under consideration is an environmentally friendly material that aims at replacing parts of a petroleum based polymer (unsaturated polyester, UPE) with natural bio-resin (epoxidized linseed oil, EML) [8-20]. The resulting bio-based resin system has been shown to attain increased toughness but its stiffness is reduced [8-20]. The lost stiffness was improved by addition of nanoscale layered silicates (nanoclay), which are sheet- like layers approximately 1 nm thick and with aspect ratios of 50—1000. Depending on the processing technique, the nanoclay in the nanocomposite exists as exfoliated (well separated), intercalated (stacks of layers with equal spacing) and a combination of both. The FE based RVE for rnicrostructures in this study modeled these morphologies based on observations from transmission electron microscopy (TEM) irmges of manufactured nanocomposites, which indicated common occurrence of three platelets stacked in an intercalated gallery(see Figure 8-4c). This morphology was thus used in the RVEs(Figure 8-4b, d). Two types of composites, namely: a) nanocomposites, consisting of petro-resin (unsaturated polyester, UPE), bio-resin (epoxidized methyl linseedate, EML) and nanoclay (Cloisite 30 BQ ), and b) hybrid biocomposites, consisting of micro-sized (~25 mm long) industrial hemp fibers and nanoclay in blends of petro- (UPE) and bio-resin (EML), were 249 modeled using the multi-level FE approach. Additionally, the nanocomposites were modeled as two—phase material (nanoclay + effective rmtr'ur) and three-phase material (nanoclay + petro-resin + bio—resin). Experimental data is based on bio-based clay/ polymer nanocomposite (Chapter 4), and hybrid biocomposites (Chapter 6), with 20% bio-resin (EML) and 2.5 wt. % nanoclay. The material layout study (Chapter 9) also resulted in idealized RVE for nanocomposite with 10°/o EML and 2.5 wt.% nanoclay, and was hence used as an additional comparison for the nanocomposite example. This material composition was not used for hybrid biocomposites. Also, for the hybrid biocomposites, the idealized RVE (Chapter 9) with three-phase materials was used to describe the properties of the matrix. The properties of the natural fiber were back-calculated from the experimental data using rules of mixtures. Experimental tensile tests were performed according to ASTM D638 standards (See Orapter 4 and Chapter 6 for details). 8.3.2 Computational Homogenization Hypotheses The material considered in this work is a polymer nanocomposite consisting of layered silicates (nanoclay) embedded in a polymer matrix. The material is considered to be macroscopically homogeneous, such that continuum mechanics theory is applicable at that scale. At the microscale the nnterial is heterogeneous and consists of distinguishable components, namely nanoclay platelets with varying spatial distribution Also, the microscopic length scale considered is much larger than the molecular dimensions, such that continuum approach is applicable to all components at that scale, but much srmller than the characteristic length of macroscopic sample considered, and hence periodicity of the microstructure is acceptable [8-7]. 250 The assumption of local periodicity follows the work of Breuls et al. [8-15] and Kouznetsova [8-7]. Instead of assuming global periodicity of the microstructure, wherein the whole rmcroscopic material consists of spatially repeated unit cells, a more realistic assumption of local periodicity is followed. The microstructure can have different morphologies corresponding to different macroscopic points, while it repeats itself in the vicinity of the each individual macroscopic point [8-718-15] . This assumption allows us to model the random nanoclay morphology in a more realistic manner. The repetitive microstructural deformations suggest that the macroscopic stresses and strains around a rmcroscopic point can be obtained by averaging microstructural stresses and strains in a small representative part of the microstructure assigned to the macroscopic material point. In this work, a two-dimensional (2D) representative volume element (RVE) is used to model the microstructure. The use of periodic constraints and stress/ strain averaging for 2D models used in this work follow the derivation provided by Breuls et al. [8-15] and are briefly presented in the following section. 8.3.3 Periodic Boundary Conditions (PBC) A two-dimensional (2D) representative volume element defining the local microstructure of an assigned macroscopic material point is considered (Figure 8-1). For a 2D RVE, throughout the deformation processes, the PBCs indicate: a) shapes of any two opposite edges remain identical, and b) stress vectors on opposite edges are equal and Opposite in direction to satisfy stress continuity. In Smit et al. [8-16], appmpriate derivations of the boundary conditions are presented and were later used by Breuls et al. [8-15] and Kouznetsova [8-7]. The final expressions: 5514 = 5323 - 552 + 531 (3'1) 251 5543 = )712 - 551 + 554 (8'2) where 53,-}- is the displacement vector for any material point on the corresponding boundaryry- and j",- is the displacement vector for corresponding vertex z.’ In addition, for vertex 3, a constraint equation is formulated, such that the vertex 3 displacements are tied with the other vertices [8-17]: J73 " 552 = 534 — 531 (3'3) The rigid bodymotions can be eliminated by enforcing 53,- = 0 , for eitheri 6 {1,2,4}. In this work, displacements on vertex 1 were enforced to be zero, i.e. 531 = 0 . From the periodicity ‘ Equations (8-1) to (8-3) it can be observed that variables in , 5323 31,552 and 554 are independent, where as 5314 , 5343 , 533 are tied (dependent). Hence it is sufficient to prescribe displacements on the three vertices 531, )72 and 534 [8-15]. 8.3.4 Coupling Macrosu'ucture and Microstructure W According to the homogenization assumption stated above, the local equals the averaged deformation gradient macroscopic deformation gradient tensor Fmacm tensor over the RVE volume (F RVE)- The volume averaging is performed over the initial volume (V0 ), since the deformation tensor is computed from the undeformed RVE configuration [8-15]: Fmacro = FRVE = —1' I(§Oj;)dV0 (8'4) 0 yoeVo 252 where (€053) denOtes the defonrration tensor at any point 550 of the undeformed RVE. Equation (8-4) also states that the strain energy in the macroscopic point equals the strain energy of the corresponding RVE. The volume integral in Equation (8-4) can be transformed into a surface integral using Gauss divergence theorem, and can be simplified for a 2D RVE. For implementation in finite element program it is written as [8-15]: {1‘}er =[u1 “2 “4]= [FRVE 'Ilb’or 3’02 3’04] (35) where u,- denotes the displacement of RVE vertex point t.’ Using the equality condition from Equation (8-4), the RVE vertex point displacements due to macroscopic loads can be obtained directly fromF macro [gag Similar to the deformation assumption, the stress at any macroscopic material point equals the volume averaged RVE stresses. Stress averaging is done over the current volume (P) of the RVE [8-15]. 1 .. Umacro : URVE = 7 lig-(y) dV (8'6) y e V Using Gauss divergence theorem, and simplifying for a 2D RVE, the average stresses can be obtained as [8-15]: “RVE =-Syml(”4 3417+ (3’2‘ 3’1)le (3'7) Equation (8-7) relates the vertex forces (fr) acting on vertex point z' to the vertex displacements y,- Since 0' mm, =0'RVE , Equation (8-7) is taken as a constitutive equation at the corresponding macroscopic material point [8-7I8-15]. Overall, the macroscopic defonnation gradients at any macroscopic material point are transferred to a 2D RVE as 253 vertex displacements, and the RVE averaged stresses are utilized as the resulting stress at the respective material point (Figure 8-3). 8.3.5 FE Implementation The rmcroscopic homogenized tensile behavior is of interest and is decoupled with the micro structural heterogeneities in a computational sense by using two separate finite element models at each scale (Figure 8-3)[8-7I8-15]. These two models are linked using the concepts of local periodicity of the microstructure and stress/ strain averaging as described in 8.3 .4. The macroscopic body is discretized into finite elements. A RVE, representing the heterogeneous microstructure, is modeled in detail as a separate FE model and assigned to each integration point in the rmcroscopic finite element mesh. It should be noted that a unique RVE can be assigned to each integration point, depending on the heterogeneity and morphology at that point. A step by step explanation of the computational approach to multi-level FE problem is given in the following ' The external rmcroscopic loads are applied in an incremental rmnner and the macroscopic FE problem is solved with an incremental procedure. The solution of the macroscopic FE problem yields macroscopic deformation gradients at respective integration points (ip), F macro,ip ' i For each macroscopic integration point, the deformation gradient RVE macro,ip vertex displacements {uiRVE’ corresponding to vertices 1, 2 and 4 are computed using Equation (8-5). These vertex displacements combined with the periodic boundary 254 constraints complete the set of boundary conditions required to solve the micro/nano- FE model [8-15]. I Each RVE is solved using a separate incremental iterative procedure. After convergence, . . . . . 4 the rmcroscoprc stresses, 0' , and macroscopic tangential Stiffness matrix, S , macro macro are returned to the respective integration point. The macroscopic stress,0' , is macro obtained using Equation (8-7) and 45’“ is obtained from the variation of the averaged RVE stress using a computational homogenization approach and is explained briefly in the Appendix (Section 8.8) to this chapter. The RVE solution also provides the local stresses and strains inside the RVE, thereby enabling to study the effect of heterogeneities in detail. I The rmcroscopic stresses assigned to each macroscopic integration point are used to compute the internal nodal forces. If these internal forces are in balance with the externally applied macroscopic loads, convergence has occurred and the next load step is applied. If convergence has not yet been reached, the macroscopic iteration continues with updated macroscopic deformations. A schematic explaining the iterative multi-level FE procedure is provided in Figure 8-2. A detailed description of microscopic morphologies, micro and macro FE models, and numerical aspects are explained in the following section. 63 4 zI/tz/tzlkw/FE Mode/ofh’o-éwea’floéimer/C%y Chin/05:26: The tensile behavior of nanoclay reinforced bio-based polymer composite was simulated using the multi-level FE approach. Two-dimensional plane strain RVES were used 255 at the nanoscale to model the nanocomposites. The success of the computational scheme depends on detailed modeling such that the FE based RVEs represent the microstructure in a realistic rmnner. I-knce, actual transmission electron microscopy imaging was used to observe the nanoclay morphology, which revealed random dispersion of nanoclay with a combination of exfoliated and intercalated particles. The intercalated galleries had around 3 silicate layers in each gallery (Figure 8—4c and Figure 8-4d). As a result, two types of nanoclay morphologies were studied: a) completely exfoliated (Figure 8-4a), and b) partially exfoliated (50%) and partially intercalated (50%) nanoclay particles (Figure 8-4b). A unique RVE, with different random clay distnbrnions, was assigned to each rmcroscopic integration point, with the intent of better representing the actual microstructure. Ideally, the rmtrix consists of blends of bio-resin and petro-resin. In this study, two RVES were used to model the bio- based polymer nanocomposites. One containing two phases namely nanoclay and rmtrix; while the other consists of three-phases, namely: nanoclay, petro-resin and bio-resin. A brief description of the models is provided in the following. 8.4.1 Macromesh, RVE (micromesh), Material Models and Computatioml Aspects The uncroscopic tensile behavior of polymer/ clay nanocomposites is of interest. The multi-level approach used in this work, and presented earlier, can be applied to any number of macroscopic elements under arbitrary loads. Yet, for simplicity, in this study the macro mesh was assumed to represent a region within the gauge length of a tensile coupon with dimensions of 1 mm x 1 mm (Figure 8-3), and hence assumed to be subjected to pure tension. This could then be modeled by a single plane strain quadrilateral element with four integration points. 256 11.1mm Why; The nanoclay sheets were modeled as a linear elastic material (E-170 GPa [8- 2118-22D with exfoliated sheets having aspect ratios of 130. Electron microscopy revealed partially exfoliates and partially intercalated galleries, with approximately three platelets per gallery and inter-platelet spacing of 3 nm. Figure 8-4c and Figure 8-4d show the TEM micrograph of an intercalated gallery and its corresponding model in an RVE, respectively. Writes; a) Neat UPE: Tensile tests showed that neat UPE (no clay and no bio-resin) had essentiallya linear elastic response. Addition of bio-resin rrnkes the resulting bio-based polymer more ductile and non-linear, while addition of nanoclay makes the resulting nanocomposite stiffer and less ductile. Neat UPE was thus modeled as a linear elastic material (E - 3.65 GPa [8- 201). W The initial attempt of multi-FE algorithm was applied using two-phase models wherein the bio-based matrix was assumed to be a homogeneous, effective rrratrix. In this case, the neat bio-based polymer (UPE/EML blend without nanoclay) was modeled as a hyperelastic material using an Ogden strain energy potential with parameters determined from experimental data. Figure 8-4a and Figure 8-4b show sample two-phase RVES for exfoliated and intercalated morphologies, respectively. 9) Three-pm; math (Mb); ;|; UPE :t bio-gsin), The experimental characterization of bio-based nanocomposites revealed reduction in stiffness parameters due to addition of bio- resin [8-20]. It was hypothesized that bio-resin addition affects the stress-transfer between the stiff nanoclay particle and the matrix. Additionally visual observation of manufactured nanocomposites and scanning electron images of tensile fracture surfaces revealed phase- separation of petro-resin and bio-resin. The distribution of the bio-resin around a nanoclay 257 platelet could not be observed experimentally and hence an adhoc approach was adopted to detennine the bio-resin distribution (Chapter 9). In order to model the rmterial realistically at the lower/nanoscale, three phase RVES having distinct clay, petro—resin and bio-resin phases were used. The three phase RVES were obtained from the material layout Optimization study, as discussed in Chapter 9. The three-phase idealized RVES were used in the multi-FE algorithms. Since the RVES developed through the rrnterial layout procedure (Chapter 9) have only one clay particle, the concept of an idealized RVES was used in a random way to better represent the nanoclay distribution. Figure 8-5 shows: a) the single platelet idealized RVE, (b) an RVE containing randomly distributed platelets with exfoliated platelets and c) an RVE with random intercalated clay morphology for three-phase material modeling for compositions of 10% EML and 2.5 wt.% nanoclay. Similar models for 20% EML and 2.5 wt.% nanoclay are shown in Figure 8-6. The next challenge was the determination of properties of pure bio-resin (100% EML), which are not available, as blends of only up to 30% EML content in UPE could be nnnufactured. Additionally, the properties of 100% bio-resins found in literature cannot be used as they do not match the chemical corrrpositons of the bio-resin used in this work. Nevertheless, the experimental tensile modulus data studying the effect of EML concentration in UPE revealed a sigmoidal trend. Figure 8—7 shows the tensile modulus data and the sigrnoidal fit (R2 value .. 1). This suggests that beyond 30% bio-resin content, the reduction in tensile properties would be minirmL Hence, experimental data of bio-based resin with 30% EML content was used to model pure bio-resin. It should be noted that this assumption may lead to an over-estimation of properties as the pure bio-resin properties will be lower than 30% bio-blend. The commercial FE program Abaqus" [8-23] allows fitting non-linear hyperelastic models for any given experimental data. For the provided data, a hyper-elastic Marlow model was found 258 to be the most stable fit for all strains and hence was used to model the bio-resin. The model was validated by corrrparing the simulated response with experirrrental data for a neat RVE (no nanoclay), and, as expected, was found to be an exact match. W The selection of an RVE plays a vital role in the effectively modeling the heterogeneous nraterial. The RVE should be selected such that it is large enough to represent the microstructure realistically and small enough to allow efficient computations. The bio-based nanocomposites in this work were modeled in two ways: a) two-phase RVES (nanoclay + effective rmtrix), and b) three-phase RVES (nanoclay + petro-resin +bio-resin). All RVES were modeled using plane strain elements. The two-phase RVES had dimensions of 500 nm x 250 nm and each element having a size of lnm x 1nrn.. For a clay content of 2.5 wt.%, and exfoliated clay aspect ratios of 130, approximately 13 particles were randomly dispersed in the RVE. For three-phase models idealized RVES were obtained from a nnterial layout study (Chapter 9). In order to facilitate the computational speed, the RVE size was reduced to 500 nm x125 nm with element size of 1 nm x1 nrrr. These RVES resulted in 7 particles for a clay content of 2.5 wt.%. This reduction in size was found to have no influence on resulting tensile predictions. A custom Matlabo algorithm was used to create RVES with randomly distributed clay platelets and generate the input file for its analysis using the general purpose FE program ABAQUS. [s23]. The macro FE analysis was performed by a custom FE code in Matlab0 and the nanoscale RVE FE analysis was performed in ABAQUSo [8-23]. A schematic of the implemented multi-level FE approach with an exfoliated RVE is shown in Figure 8-3. 259 c9. )' Mix/tale w/ if z}! 049/ of 159/hid flz'ocomporz'tw: The nocroscopic tensile behavior of hybrid biocomposites was modeled sirrrilarly to the nanocomposites, as discussed in the earlier section. A simplified model assumed to be within the gage length of the tensile coupon was used at the macroscopic scale, and a unique lower-scale RVE was assigned to each uncroscopic integration point. The macroscopic model consists of nine plane strain quadrilateral element with four integration points each (Figure 8-8). The dimension of the fiber element was assigned as half the diameter of the raw industrial hemp fiber [8-24]. The dimensions of the matrix elements were assigned such that it satisfied the volume fraction of fiber/ matrix, of the material test used for validation. The material under consideration has an average volume fraction of natural fibers of 21% with 20% EML content in UPE and 2.5 wt.% nanoclay. As discussed earlier, in the multi-level FE approach no constitutive relationship is defined at the macroscale, instead all nonlinearities are obtained from the lower scale. The material properties of the nanocorrrposite models were described in the earlier section. In order to facilitate computational speed, a single particle RVE with three-phase material (Figure 8-8) was used. The material properties of natural hemp fiber were back-calculated from experimental results (Chapter 6), and using rule of mixtures. The average elastic modulus of the fibers was computed to be 12.5 GPa with a standard deviation of 3.0 GPa. The modulus of hemp fiber has been reported to be 70 GPa [8-24] whereas the average back-calculated modulus value was only 12.5 GPa. The lower value of computed modulus was expected as the experiment data represents short fibers while reported modulus uses aligned/ single hemp fiber [8-24]. The efficiency of short fibers is considerably lower than aligned fully bonded fiber composites thereby supports the lower computed values of the 260 modulus. Also, it should be noted that the modulus of natural fibers generally have significant scatter due to varying fiber quality. Finally, the hemp fiber was modeled as a linear elastic material with a modulus of 12.5 GPa. (53 6 Rest/lb“ dfldDziaan'o/z Tensile behavior of bio-based nanocomposites (UPE +EML +clay) and hybrid biocomposites (hemp fiber+UPE +EML+clay) was simulated through the multi-FE approach and the sirmrlations were compared with experimental data. The comparison of the predictions and the experiments was performed by studying the following parameters: a) initial stiffness, b) strain deviation parameter (8d ), and c) stresses at average experimental ultirmte strain. These parameters are schematically represented in Figure 8-9 and are explained in the following. The strain deviation parameter (8d ), or the strain level at which simulations starts to significantly deviate from the experimental response (Ed) is computed as: 6' ed =———"’~‘“”t (%) (8-8) gu,exp where ad’stm is the value of the strain at which the start of significant deviation of simulation and average experimental response was observed, and Eu is the average ,exp experimental ultirmte tensile strain. Additionally, O'u,exp , which is the average experimental stress comesponding tog“,exp , is compared with the stress-predictions from multi-FE models (Ume) corresponding torB‘u,exp . The FE sirmrlations in this study do not model damage and hence the simulations were terminated ate“,exp . 261 8.6.1 Bio-based Polymer Nanocomposites W Figure 8-10 compares the simulated and experimental tensile stress-strain responses for two-phase RVES. A nomenclature of A(°/o)/B(°/o)/Cth.%) is used to indicate the concentration of UPE/ bio-resin/ clay content. The high correlation between simulated and experimental responses for neat resins (no nanolay) in Figure 8-10a simply shows that the assumptions of infonnation transfer (deformation gradient and effective stiffness) between the micro and macro levels is adequate. Two material compositions namely 100/0/25 and 90/10/2.5 were studied. The 100/0/25 material composition represents neat UPE (no bio- resin) reinforced with nanoclay, and has a linear average experimental response (Figure 8-10b). Since both UPE and nanoclay were modeled as linear elastic, the simulated response was also linear elastic. The 1min purpose of performing this analysis was to study the effect of nanoclay morphology. It was observed (Figure 8-10b) that models with partially intercalated and exfoliated morphologies exactly rmtched average experimental response. Additionally, transmission electron imaging revealed that the actual microstructure had partially exfoliated and intercalated nanoclay morphology. This supports the good agreement of the simulations from partially intercalated model with experimental data, conversely indicating the need to realistically model the microstructure. Similar to material composition 100/0/25, the models with exfoliated and partially intercalated morphologies were studied for material composition 80/20/25. As described in earlier section, the bio-based matrix was modeled as a non-linear hyperelastic rmterial. The sirmrlations for 80/20/25 material compositions from both exfoliated and partially intercalated models seem to capture the non-linear behavior (Figure 8-10b). Both models 262 (exfoliated and partially intercalated) match the average experimental data in initial stiffness. The exfoliated model had a strain deviation value of 35% while the partially intercalated model had a strain deviation value of 50%. This indicates better agreement of the partially intercalated model with average experimental data, as similariy observed with rmterial composition of 100/0/25. At 8“ the exfoliated model prediction exceeds the average ,exp ultimate experimental stress (O'ufixp) by approximately 45%, while the partially intercalated model exceeds it by only 35%. Additionally, the experimental ultimate stress values (amexp) have a deviation of around 10%. Overall, it was found that the multi-FE simulations agree well with the average experimental response in initial stiffness, and the intercalated RVE models gave better predictions compared to the models with exfoliated morphology. W The single particle idealized RVE obtained from the material layout study (Chapter 9) was used along with random distribution in exfoliated and intercalated morphologies. Additionally, since two idealized RVES for 10°/o and 20% EML contents were obtained from the nnterial layout study, both RVES were used in the multi-FE simulations. Hence, for a given material composition, three models were studied, namely: a) single-particle idealized RVE, b) RVE with multi-particles and exfoliated morphology, and c) RVE with multi- particles and intercalated morphology. Figure 8-11 and Figure 8-12 compare the average experimental response of the multi-FE simulated tensile response of the aforementioned three models for bio-resin contents of 10% and 20% EML, respectively. The comparison of multi-FE simulations for material composition 90/10/25 and the average experirrrental response is provided in Figure 8-11. The initial stiffness of all three 263 models (single-particle idealized RVE, multi-particle +exfoliated and multi-particle + intercalated) rmtched the average experimental data. The simulations from the single-particle idealized model and the multi-particle intercalated model were found be very similar. The exfoliated platelets model had a strain deviation value around 20%, while the intercalated and single-particle idealized models had strain deviation values of 60%. This suggests that the single-particle idealized and intercalated platelet models performed better than the exfoliated platelet model Finally, the stresses at Emexp for the intercalated platelets model were approximately 30% higher than 0‘“ For the single-particle idealized RVE and the ,exp ' intercalated platelets rrrodel, the predictions exceeded 0'“ by around 15%. It should be ,exp noted that the 0' value has a deviation of around 15%. The above-mentioned u,exp comparisons were rrrade only with respect to the average experimental data and the deviations were not considered This suggests that in spite of the sirmrlations appearing to deviate slightly from the average experimental response at ultimate, consideration of the deviation in the data would indicate reasonable agreement with experiments. Additionally, the rmterial compositions with 10°/o bio-content are relatively brittle, which is what is thought to lead to the larger deviations near ultimate. Considering these variations and the assurrrptions made in the idealized RVE, it can be concluded that, overall, the sirmrlations perform well and that the results reasonably agree with the average experimental response. Specifically, the models matched the average experimental response in initial stiffness, and both single-particle idealized RVE and the intercalated platelets models gave better predictions relative to the exfoliated platelet model. The comparison of multi-FE simulations for material composition 80/20/25 and average the experimental response is provided in Figure 8-12. . The initial stiffness of the 264 three models (single-particle idealized RVE, multi-particle + exfoliated and multi-particle + intercalated) matched the average experimental data. The strain deviation values were approxirmtely 15, 40, and 70% for the exfoliated, intercalated and single-platelet idealized models, respectively. The model predicted stress values at Emexp were approximately 62, 50 and 15% higher than 0'“ These results suggest that the single-particle idealized RVE ,exp - performed the best, followed by the intercalated model. The exfoliated model did not show good agreement beyond the initial stiffness and strain deviation values of 15%. Additionally, the single-particle model had strain deviation values of 70% or more, suggesting that the model predicted the response reasonably well for more than 70% of the average experimental response. Finally, dump value has a deviation of around 10% and the single- particle model exceeded it by only 15%. The reasons attributed for the single-particle idealized RVE performing better than the multi-particle models are discussed in the following section. The above comparisons were made only with respect to the average experimental data and the deviations were not considered. Considering the data deviations and the assumptions nude in the idealized RVE, it can be concluded that, overall, the simulations performed well and that the results agree reasonably well with the average experimental response. Specifically, the models matched the average experimental response in initial stiffness, and both the single-particle idealized RVE and the intercalated platelets models gave better predictions compared to the exfoliated platelets model. Figure 8—13 shows the comparison of the simulations using two-phase and three- phase RVES. Only models with intercalated morphology and 20% EML content were compared, as intercalated morphologies were found to agree better with experiments. 265 Additionally, since the initial stiffness from all models rmtched the experimental response reasonably well, only the stresses at 6' are compared. The two-phase and three -phase u,exp models exceed 0' by 35% and 50%, respectively. It appears that for multi-platelet u,exp RVES, the two-phase model performed better than the three-phase RVE. However, this observation should be dealt with carefully. The model is as good as the assumptions at the lower scale. In the two-phase model the non-linear behavior of the bio-based matrix was modeled as a homogenized hyperelastic material fitted to the experimental data. In the three- phase models, UPE was assumed to be linear elastic and an upper-estimate of bio-resin material properties was assumed (refer to Section 8.4.1). Improved accuracy in material properties for each constituent would improve the performance of three-phase models, which additionally have more versatility by not requiring the fitting of a material model to the homogenized bio-based polymer. The three-phase single-particle model had better agreement with the average experimental response with strain deviation values beyond 70% and stresses at amexp exceeding amexp by approximately 15%. For three-phase models, the multi-particle intercalated model would be expected to perform better than the single-particle model, as it seemingly models the material more realistically. Nevertheless, the single-particle idealized RVE was obtained by Hatching elastic experimental properties with the effective properties from the model (see Chapter 9). The material layout study in Chapter 9 was not performed with multiple platelets, as it would require large models that would be computationally too expensive to perform the material layout optimization. Additionally, the multi-particle distribution obtained as an extension of idealized RVES violates the periodicity assumption used in the actual development of such idealized single particle RVE. For the same reason, 266 the single-particle idealized RVE performed better than other models, as it was obtained by fitting experimental data. 8.6.2 Hybrid Bio-based Composites. The multi-level FE approach used to model the bio-based nanocomposites as described in earlier sections of this chapter was extended to model hybrid biocomposites (hemp fiber + UPE + EML + nanoclay. The objective of modeling hybrid biocomposites in this study was mainly to show the feasibility and robustness of the multi-level FE scheme and not to accurately predict the response. Hence, simplified bio-based nanocomposite RVES (with linear elastic UPE, hyperelastic discrete EML and linear elastic nanoclay) were used along with a simplified segment of a linear elastic fiber (hemp). The comparison of multi-FE simulations and average experimental data was evahrated similar to the bio-based nanocomposites (see Section 8.6.1). Figure 8-14 compares the tensile response from multi- FE simulations and experiments for virgin UPE biocorrrposites (0% EML, 0% clay). The response is linear elastic as all the properties were assumed to be linear elastic. As expected, the initial stiffness matched the experiments. The strain deviation value was approximately 25%. The model predicted stress value at 5,, was approximately 40% higher than ,exp dump . Figure 8-15 compares the tensile-response of a hybrid biocomposite with 20% EML and 2.5 wt.% nanoclay. Similar to a virgin UPE biocomposite, the predicted initial stiffness agrees well with the experimental data. The strain deviation value was approximately 35%. The model predicted stress value at £u,exp was approximately 30% higher than dump . The models that had 20% bio-resin used a non-linear rrraterial model for bio-resin, and seem to agree average experimental response for up to 35%, while similar 267 linear elastic model (100/0/0) agreed only up to 25% of respective 8,, This suggests ,exp ° that material non— linearity helps in capturing the non— linear behavior, but is not the only source of non- linearity. The deviation of the simulation from the experimental data is due to unny factors, including material non-linearity, random distribution of the short hemp fibers, interaction of the short-fibers, fiber pull-out, straightening of the curved fibers, interfacial properties, etc. Hence, relying purely on material non-linearitywill still be unable to capture the non-linear response accurately, particularly near ultimate response. Nonetheless, properly designed multiscale FE sirmrlations can shed light in understanding the aforementioned complexities and sources of non-linearity. Again, the accuracy of the model is as good as the assunrptions and modeling of the actual phenomena. If all sources of non-linearities are properly modeled at the lower-scale, these multi-level schemes hold great promise for use in hierarchical rrraterials. J? 7 C‘wzcéaz’on A multi-level finite element approach was implemented and evaluated for simulating the tensile behavior of both bio-based-polymer/ clay nanocomposites and bioconrposites (fiber+bio-blend +nanoclay). The approach derives macro-scale constitutive relations from numerically homogenized micro/nanoscale behavior. Two-phase and three-phase models, each having a) a single-particle idealized RVE, b) an exfoliated multi-platelet RVE, and c) intercalated multi-platelet RVE were studied. All simulations were in good agreement with the initial stiffness. The intercalated platelet models performed better than the exfoliated platelet models (at ultimate). The idealized single-particle model had the best perforrmnce of all models with good agreement up to 70% of the ultimate strain observed from average experimental response. Also, considering the deviations in the experimental data, the stresses 268 predicted by the single-particle idealized RVE, corresponding to the average experirrrental ultimate tensile strain, were within 5% of the average experimental ultimate tensile stresses. Overall, the simulated responses agreed reasonably well with experimental results and were able to capture the nonlinear response. The models can be improved by more accurate definition of the rmterial properties at the micro- and nano-scale level. The approach holds great promise in understanding the behavior of polymers reinforced with nano-particles and, in general, the behavior of hybrid hierarchical materials. :5? 6’ Appefla’zk-Defl'wabfl offlflngtzk/flflzers Meme The objective of the appendix is to briefly report the derivation of the tangential stiffness rmtrix used in the multi-level FE scheme used in this work. The rrnterial that follows is directly from the work of Breuls et al. [8-15]. The effective macroscopic behavior for the problem at hand is computed in the form of the tangential stiffness matrix, which provides the relationship between infinitesimal stress and incremental strain variations as follows: 4 . S macro ° 60‘ macro _ 65 macro (3- 9) where, 4Smacm is the fourth order macroscopic tangential stiffness matrix, consisting of three terms, which are obtained from the linearization of the geometrical nonlinear deformation tensor, the volume ration factor and the constitutive equation. Linearization of the constitutive equation yields to the extra stress tenrr, 45”,", This extra stress term, 450“,“ , is obtained by a computational homogenization procedure as follows: 269 The extra-stress term 4S extra to the macroscopic tangential stiffness matrhr relates the macroscopic stress variations to strain variations 60 :45 : (6 a (8- 10) macro extra wherein x is the position vector of the macroscopic integration point. The macroscopic stress tensor at any integration point equals the averaged stress tensor in the corresponding RVE, as per Equation (8-7): am... =0... =—sym[(7— 7)7.+ (7— 7)7.] (8-7) in which, A is the current RVE area, Z , the local position vector of RVE vertex point z; and 7; is the local force vector acting in vertex point 1: Taking the variation of Equation (8- 7) yields: 631m: =—0Rm%+%syml56’_;‘3’i):i+5lg‘371)72l+ ‘1' J k J iSyMlej- y1)5f4 + "yll5f2l (8-11) Parts I, H and III in the above equation are re-written in fomrs of deforrmtion tensordF . Also, in order to store the three components of the three parts in the r.h.s of the above equation, the following are used: ] ' ~11 ' ~111 50RVE, xx 50RVE,xx -50'RVE,xxT 645811 5N1” ~1 URI/E.» . ~11 RVE,» . ~111 URI/E.» . 50' = 60' = 60' = , (8-12) RVE 6w; . RVE 6"” ~11 ’ RVE 50111 ~11 ~111 50RVE,yx_ __0'5 URVEJx 50' O'RVE ,yx_ 270 atrial-Jill‘! I'll-rifi'lu“ r. . . Similarly, we create vectors 6F and F _1 to store the components of the deformation tensor. "an,“ F; r a? F“ 5F: ”’ ; F“1 = {y (8-13) 6F F 1 xy 10’ -1 PF)“ .. .Fyx _ PART - I Part1 can be re-written as: — 5A - _ — O'RVE —A— = —O'RVEtr((5F - F 1) (8-14) Equation (8-14) can be elaborated to EIRVE = B167? (8-15) with ' —1 -r —1 -1‘ — xx-Fxx -0n°F}y —0'n-Fyx “Gn'ny ..1 -l —1 —l B]: —0'yy-Fxx -0'yy-Fyyl —0'yy-Fyx -0'yy-ny (8-16) ..1 — -l -l - ,0,an —0'xy.Fyy ‘ny'Fyx -0-xy.ny -1 -1 —1 —1 _—0'yx~Fxx —0‘yx-Fyy —0'yx-Fyx 70w°ny_ The components 0",] are the components of the macroscopic stress tensor Umacm as obtained from Equation (8-7). PART - II To rewrite partII interrns of 6F the followingisused: ~ ~0 5y1=6F'yi (8'17) ~o in which y ,- is the vertex position of the RVE in the reference configuration. 271 Then, 65£IVE = 3116}? (8'18) with F P1 0 P2 0 _ 0 P4 0 P3 1 l 1 1 311 - — -— — — (8'19) 2 P3 2 P2 2 P4 2 P1 ' l .1. .1. l __ 2 P3 2 P2 2 P4 2 P1- 1 and P1 = :14'[f2,x (Vise —y1(2x)+f4,x()'2,x ’3’in )] i.- P2 = ilfz. (yiy -yfy)+ f4. (yiy -yify)] p3 = %[f2,y(yg,x _yl),x)+ f4,y(yg,x —ylo,x)] p4 = :14'lf2.y(yg.y 7yliy)+ f4.x()’2.y 734211)] (8- 20) In which fix is the 1' component of the vertex force, acting on the vertex 1; you denotes the 2' component of the reference position of point 1' PART - III Part III is reformulated to: isms—mam (8-21) 1 _ _. _ _ __ _ _ ... _. =fl01§f1+6f1y1+y25f2+5f2y2+y45f4+5f4y4l Byrmking use of 272 671 = —674 — 672 (8-22) Equation 8-22 is written in matrix form as: panama-nan — yl,x 0 y2,x 0 y4,x O i :1”: O yl y 0 y2 y 0 y4 y l’y , , , s23 =1 1 1 1 1 .1. 1 . “Ye: ‘ ) A 2yl,y 2ylgt 2y2,y 2y2,x 2y4,y 2y4,x 5.2,}; l. .1. l .1. .1_ 1 (Y... -2 yl,y 2 J’1,x 2 3’2,y 2 3’2; 2 J’4,y 2 3’4,x_ a t E z b 4,},- The vertex force variations 5 are rewritten in terrrrs of displacement variations as: J = K R VEafvertex (8'24) The reduced RVE stiffness matrix K RVE is found by partitioning the RVE stiffness matrix: [Kw mm KP” KPP all? 5P wherein 511p refers to the iterative displacements of prescribed vertices 1, 2, and 4, and 5a, denotes the iterative displacements of the remaining nodes in the RVE. The vertex force variation JP is expressed in terms of prescribed displacements as: a}, = KRyEfilp (8'26) Km = K pp — K ”Kr-JK,? (3-27) The displacement (ill, is written in terms of 6F 273 yf. 0 yify 0 0 yfy 0 no. 0 O 0 O O y2,y O y2,x ° 0 ° 0 y4,x y4,y O 0 _ O y4,y O y4,x _ 15 Hence, 5011,14,- : CK ”3677 (8-29) Combining the tenm from part I to III yields: 551R” = [BI "l" B]! + CKRVFDW (8’30) Finally, Wisrewritteninterms of (as; using 6F = GH (8-31) with '17,, o F), 0 (V as.) _ O F yy 0 F x}, . _ (V 5x); G — , H — _ _ (8-32) 1;, o 1",, 0 (var. L O F yx 0 F (e 3;); Hence, the matrix representation of the extra stress term contribution 4S extra , to the macroscopic tangential stiffness rmtrix S macro can be written as: 50......= [3, +B,, +CKRVED] GH (3-33) 274 £9 Déde: (9 152311735 4 F43 F23 Figure 8-1. A schematic of a typical 2D periodic RVE used in the current multiscale approach. Adapted from [8-15]. Input Data L Compute External Forcefl r <> \ <\\0\j: mg intercalated gallery, and d) zoom in of the intercalated gallery in FE based RVE 51’1 w model. 277 (a) (b) (C) Figure 8—5. Three phase RVES for materials with 10% EML (red) in UPE (green) with 2.5 Wt.°/o nanoclay (black), (a) Single platelet idealized RVE, b) multi—particle, exfoliated morphology RV E c) multi—particle, intercalated morphology RVE 278 (a) (b) (C) Figure 8-6. Three phase RV Es for materials with 20% EML (red) in UPE (green) with 2.5 wt.% nanoclay (black), (a) Single platelet idealized RVE, b) multi-particle, exfoliated morphology RV E c) multi—particle, intercalated morphology RVE 279 4.5 0 Experimental Data ___ - - 2= 4.0- SingIdalfit,R 1 A 3.5- Ef“\\\ E \ 9" 30 \ 3 \ 3 'U 2.5- \\ O E \\ 2 20- \f .5 . c \ 0 \\ ..- 1.5" \\\ E 1.0- 0.5 l l l I 0 10 20 30 Bio-resin (EML) content Figure 8-7. Experimental tensile modulus for neat bio-blends (no clay). The dotted line is a sigrnoidal fit to the experimental data. l. iWi Integration Point, Unique nanoscale RVE, three-phase RVE aulgned Three phase nanocomposite RVE Figure 8‘8. Schematic of multi—FE model used for fiber-reinforced hybrid composites. 280 0' _ 8d,start u,exp --- 8d — (%) O'maxf ---______-_____-___-_____-_-_5_11_1191_&t191_}_, 8d , start -____-______._______.._ ....4 >8 u,exp Figure 8-9. Parameters considered to compare simulations and average experimental response 281 42 ‘ 365 30% E : a .24: 10 I a d g 18': a) q 12{ I —e— FEA-100l0/0 6 .' —'v— Experimental Data-100mm : + FEA-BOIZOIO j + Experimental Deta-BOIZOIO 0 ....1. .................. .Tj. .I. r 0.000 0.002 0.004 0.000 0.008 0.010 0.012 0.014 0.010 Strain (mm/mm) (a) 24 203 16_' ./ E u a ; '/ g 1 / (D / 3‘. / . —e— FEA-100IOI2.5-Exfoliated 3 ' / + FEA-100IOI2.5-intercalated- 50%,Ns3 - —l- Experimental Data - 100IOI2.5 4: -e— FEA- 80I20I2.5-Exfoliated « + FEA- 80I20I2.5-lntercalated-50%,N=3 j / —e— Experimental Data 4010/25 0 .........,..r..... ...r,..fi..... ..... 0.000 0.001 0.002 0.003 Figure 8- 1 O. Tensile stress-strain response of two—phase RVES: a) Neat (no clay) resins, b) 0.004 0.005 0.006 0.007 0.008 0.009 Strain (mm/mm) (b) Nanocomposite with 2.5 wt.% clay. 282 20* Stress (MPa) + Experimental Data + Exfoliated, multi-particles -I- Intercalated . multi-particles -.— Idealized RVE - single particle " 1 V I V I I Y I V I T j T r I r I 0.000 0.001 0.002 0.003 0.004 0.005 Strain (mm/mm) Figure 8-11. Comparison of tensile response from multi-FE simulations and experiments for biocomposites with 10% EML and 2.5wt.% nanoclay in UPE 25 20 r g 15 ~ 10 g . tn 10: / 5 . % -.- Experimental Data 1 ‘T' Exfoliated, multi-particles : + Intercalated , multi-particles . -.- Idealized RVE -single particle 0 ....................1....rr+.,.....r 0.000 0.001 0.002 0.003 0.004 0.005 0.006 0.007 Strain (mm/mm) Figure 8‘1 2 Comparison of tensile response from multi-FE simulations and experiments for biocomposites with 20% EML and 2.5 wt.% nanoclay in UPE 283 20‘ 15- Strese (MPa) + Experimental Data + 2-Phase RVE. multi-particles . -.- 3-Phase RVE, multi-particles - + 3-Phase RVE, single particle o ,1..- ..... I 0.000 0.001 0.002 0.003 0.004 0.005 0.006 0.007 Strain (mmlmm) Figure 8-13. Comparison of tensile response from multi-FE simulations using two-phase and three-phase RVES. All models have intercalated morphologies Stress (MPa) 3 10 - 5.‘ + Experimental Data . + Muiti-FE simulation 0 ......r..#.-......,.. 0.000 0.001 0.002 0.003 0.004 Strain (mm/mm) Figure 8-14. Comparison of tensile response from multi-FE simulations and experiments for virgin UPE (0% bio-resin and 0% nanoclay) biocomposites 284 35 30 j 25 fl 1 20- I ‘ Stress (MPa) 15{ 10{ -.— Experimental Data + Multi-FE simulation o, ..... 0.000 0.001 0.002 0.003 0.004 0.005 0.006 Strain (mm/mm) Figure 8-15. Comparison of tensile response from multi-FE simulations and experiments for hybrid composites with 20% EML and 2.5 wt.% nanoclay. 285 (2 10 Ker/{37322665 [8-1]. Le Baron PC, Wang Z, Pinnavaia T]. Polymer - layered silicate nanocomposites: an overview. Applied Clay Science 1999; 15:11-29 [8-2]. Bohm, HJ. A Short Introduction to Basic Aspects of Continuum Mechanics.” CDL-FMD Report, 3 (1998). Vienna Institute of Technology. [8-3]. Eshelby JD. The determination of the field of an ellipsoidal inclusion and related. Proceedings of the Royal Soceity of London, Part A. 1957, 241:376-296. [8-4]. Hashin Z. The elastic modulii of heterogenous materials. Journal of Applied Mechanics. 1962, 29:143-150. [8-5]. Hashin Z, Shrikman S. A variational approach to the theory of elastic behavior of multiphase naterials. Journal of the Mechanics and Physics of Solids. 1963, 11:127-140, [8-6]. Hill R. A self-consistent mechanics of composite materials. Journal of the Mechanics and Physics of Solids. 1965, 13: 213-222. [8-7]. Kouznetsova V, Brekelrmns WAM, Baaijens FPT. An approach to micro- macro modeling of heterogeneous materials. Computational Mechanics 2001; 27:37- 48. [8-8]. Duschlbauer D, Bohm HJ, Petermann HE. Computational Simulation of Composites Reinforced by planar random fibers: Homogenization and Localization by Unit Cell and Mean Field Approaches. Journal of Composite Materials 2006, 402217-2234. [8-9]. Bensoussan A, Lionis JL, Papanicolau G. Asymptotic Analysis for Periodic Structures, 1978. North Holland, Amsterdam. [8-10]. Sanchez-Palencia E. Non-homogeneous Media and Vibration Theory. Lecture Notes in Physics, 1980. Springer-Vedag, Berlin. [8-11]. Sheng N, Boyce MC Parks DM, Rutledge GC, Abes JI, Cohen RE. Multiscale micromechanical modeling of polymer/ clay nanocomposites and the effective clay particle. Polymer 2004, 45:487-506. [8-12]. Brune DD, Bicerano J. Micromechanics of nanocorrrposites: Comparison of tensile and compressive elastic modulii, and prediction of effects of incomplete exfoliation and imperfect alignment on modulus. Polymer 2002, 43:369-3 87. [& 13]. Miyagawa H, Rich MJ, Drzal LT. Amine-cured epoxy/ clay nanocomposites. H. Effect of the nanoclay aspect ratio. Journal of Polymer Science PartB: Polymer Physics, 424391-4400. 286 [8-14]. Drown EK, Mohanty AK, Parulekar Y, Hasija D, Harte BR, Misra, M, Kurian JV. The surface characteristics of organoclays and their effect on the properties of poly(trimethylene terephtahalate) nanocomposites. Composites Science and Technology, 2007, 67:3168-3175. [8-15]. Breuls RGM, Sengers BG, Oomens CWJ, Bouten CV C, Baaijens FPT. Predicting local cell defonmtions in engineered tissue constructs: A multi-level finite element approach. Transactions of the ASME 2002; 124:198-207. [8-16]. Smit RJM, Brekelmans WAM, Meijer HEH. Prediction of the mechanical behavior of nonlinear heterogeneous systems by multi-level finite element modeling. Computational methods in Applied Mechanical Engineering 1998, 155:181-192. [8-17]. Van der Sluis O, Schreurs PJG, Brekelmans WAM, Meijer HEH. Overall behavior of heterogeneous elastoviscoplastic materials: effect of microstructural modeling. Mechanics of Materials 2000, 32:449-462. [8-18]. Feyel F, Chaboche JL. FE2 multiscale approach for modeling the elastoviscoplastic behvior of long fiber SiC/ Ti composite materials. Computational Methods in Applied Mechanical engineering. 2000, 183:309-330. [8-19]. Matsui K, Terada K, Yuge K. Two-scale finite element analysis of heterogeneous solids with periodic rnicrostructures. Computers and Structures 2004, 82:593-606. [8-20]. Haq M, Burgueno R, Mohanty AK, Misra M Bio-based polymer nanocomposites from UPE/EML blends and nanoclay: Development, experimental characterization and limits to synergistic performance,” Composites: Part A. Communicated. [8-21]. Miyagawa H, Misra M, Drzal LT, Mohanty AK. Biobased Epoxy/ Layered Silicate Nanocomposites: Thermophysical PrOperties and Fracture Behavior Evaluation. Journal of Polymers and the Environment. 2005, 13:87-96. [8-22]. Luo J], Daniel IM. Graracterization and modeling of mechanical behavior of polymer/ clay nanocomposites. Composites Science and Technology. 2003, 63:1607- 1616. [8-23]. ABAQUS Documentation (Version 6.8) Dassault Systémes Sirmrlia Corp., Providence, RI, USA, 2009. [8-24]. Mohanty AK., Misra M, Drzal LT., Natural Fibers, Bio-polymers, and Biocomposites, 2005. CRC, Taylor 86 Francis Group. 287 1"“ _ _‘l"_,_..._" I _I.‘ Chapter 9. Modeling of Three-phase Bio-based Nanocomposites: Determining Bio-resin Distribution using an Optimization- based Material Design Problem 2 I Aéitraa' Environmentally friendly composites with material properties that can compete with conventional composites can be obtained by reinforcing blends of petroleum- and vegetable- oil-based polymers (bio-resin) with layered silicates. Such hybrid combinations have been found to produce composites that exhibit synergistic behavior with properties that are superior or similar to the primary petroleum-based polymer while the addition of bio-resin addition reduces stiffness and banier properties of the resulting composites. The lost properties have been shown to be recoverable by the addition of nanoparticles (nanoclayfi. Nevertheless, the enhancement efficiency provided by the nanoclay is reduced by the presence of bio-resin. It was hypothesized that the bio-resin forms an interface around the clay particle affecting the stress transfer between the reinforcing clay particle and the polymer rmtrix, thereby supporting the experimental reduction of stiffness due to bio-resin addition. The distribution of bio-resin depends on its concentration and compatibility with other constituents. In order to provide a quantitative insight on the effect of bio-resin on resulting properties, the distribution of bio-resin was sought. The available experimental data on thermo—elastic parameters was used along with topology optimization to determine the distribution of bio-resin and develop an enhanced RVE for modeling the three-phase material. The formulation of the problem followed a topology optimization method in which the objective function was minimization of the error between the model predictions and the 288 prescribed experimental thermo-elastic parameters subjected to constraints on bio-resin ' concentration. The effective properties were determined through a numerical homogenization method based on finite elerrrent discretization of the base cell. The optimization problem was solved using the method of moving asyrrrptotes. Two-dimensional linear elasticity (small strains), perfect bonding between material phases, uniform temperature distribution and constant material properties were assunred. This chapter presents the challenges of modeling polymer-blend nanocomposites, the approach taken to address the problem, the forrmrlation of a three-phase rmterial layout problem, a parametric study of the optimization problem, case-studies with experimental data and the resulting material layouts for various bio-resin contents. Simplified RVES were developed from the resulting rrraterial layout study and were used in multi-FE analyses to predict the nncroscopic response. The multi-level FE simulated tensile response using the simplified RVES from this study matched the average experimental data in initial stiffness. Additionally, the idealized RVE corresponding to 20% bio-resin content and 2.5 wt.% nanoclay, agreed reasonably with experimental data for up to 70% of the average experimental ultimate strains, and deviated from average experimental ultimate stress marginally (~5%), while the two-phase model deviated by approximately 35%. The response of this model was better than corresponding two-phase model which deviated considerably (at ultimate) by approximately 45%. This emphasizes the need to model materials as realistic as possible. Overall, the approach of validating computational rrrodels with experiments shows promise in developing models rationally, and thereby confidence in using such models in multiscale simulations linking the nanostructure with macroscopic properties. 289 If; 22 [nova/action: Novel eco-friendly bio-based nanocomposites obtained from reinforcing layered silicates (nanoclays) in blends of petroleum and natural polymers (bio-resin) have been shown to exhibit properties that are superior or similar to base petroleum-polymer [9-1}[9- 3]. The addition of bio-resin has shown to reduce stiffness and banier properties. The distribution of nanoclay can be observed experimentally through electron microsc0py. However, similar observations on distribution of petro/ bio blends are infeasrble. In order to understand and model these materials efficiently, the distribution of bio/petro resin is sought. The material layout problem used in this work was aimed to develop an enhanced three-phase RVE by using experimental data along with topology optimization in an attempt to provide insight to the distribution of bio-resin in the RVE, and not necessarily to accurately solve the material design problem. The motivation for this work follows from the desire to improve the modeling of hybrid multiscale eco-friendly composites. Environmental concerns related to the use of petroleum-based polymer matrix composites have propelled the development of composite materials based on natural or renewable sources. Bio-based resins, or bio-blends, obtained by replacing part of a petroleum-based resin with natural bio-resin increase environmental appeal along with but reduces stiffness and barrier properties of the resulting composites. A recent study [9-1}[9-3] has shown that the property degradations due to bio-resin addition can be recovered by the addition of nanoparticles (nanoclay) [9-1}[9-3]. Nevertheless, the enhancement efficiency provided by the nanoclay is reduced by the presence of bio-resin. It is hyp0thesized that the bio-resin forms an interface around the clay particle affecting the stress transfer, thereby supporting the observed reduction of stiffness due to bio-resin addition. The distribution of bio-resin depends on its concentration and compatibility with 290 l-r 4 the host polymer [9-4]. In order to provide a quantitative insight on the effect of bio-resin on the resulting properties, the distribution of bio-resin is sought. Experimental data was used along with a t0pology optimization method to determine the rrrostl likely distribution of bio-resin and develop an enhanced RVE for the multiscale modeling of the three-phase nnterial. The use of topology optimization to solve a material design problem was introduced by Sigmund [9-5}[9-8], whose goal was to find a periodic microstructure in which the mixture of two-distinct isotropic materials results in a composite rmterial whose overall homogenized elastic properties nntch prescribed target values. The results of such rmterial design problems is usually a spat'nl layout of a mixture that is optimized for prescribe objectives. Sigmund [9-8] applied the material design problem to obtain microstructures with extrerrral properties such as negative and zero poisson ratios. Diaz and Benard [9-9] used a similar approach to design rmterials with prescribed elastic properties using polygonal cells. In this work, the rmterial design approach is used to determine the distribution of bio-resin (weak material) and petro-resin in a base cell containing a fixed stiff particle (clay). Ideally, it is a two-rmterial spatial layout problem, wherein instead of a void, a weaker bio-resin material is present and the solid material is replaced with petro-resin. Experimental data for various constituent concentrations (petro- and bio-resin) is available [9-3] (also refer Chapter 4). The objective is to determine the distribution of weak material (bio-resin) in a base-cell such that the effective properties of the base cell rmtch the prescribed parameters obtained experimentally. The use of a single property (say elastic tensor) as a prescribed parameter will yield a material layout in the base cell whose effective properties would rmtch that parameter. Using this material layout to obtain effective properties other than the prescribed parameter will most likely yield incorrect results. For the 291 problem at hand, the material clearly has a unique bio-resin distribution for which various properties were experimentally measured. In other words using a single prescribed parameter to obtain the bio-resin distribution is not realistic. Hence, multiple material properties (elastic and thermal) are prescribed with the aim of obtaining a unique bio-resin distribution. Similar use of multiple prescribed pararnaters in a material design studywas used by Guest et al. [9- 1019-11] to obtain microstructures that maximize stiffness and permeability. The topology based material design problem has also been temred as “inverse homogenization problem [9-9].” The assumption for such problems is that base cell is periodic and that the effective properties of the mixture can be obtained by analyzing only the respective base cell. The optimization problem and determination of updated design variables and sensitivities was solved by the method of moving asynrptotes (MMA)[9- 1219- 13]. It should be emphasized that the work attempted here aims at providing an insight to the distribution of the weaker bio-resin and not necessarily at accurately solving the material design problem. The material under consideration is a nanocorrrposite bio-based material containg unsaturated polyester (UPE, primary perm-resin), epoxidized methyl linseedate (EML, bio- resin) and nanoclay. The nanoclay concentration was kept constant at 2.5 wt.% and two concentrations of bio-resin, namely, 10°/o and 20% were considered. Additionally, models with a single exfoliated nanoclay platelet and a single intercalated gallery were studied. The RVE size was determined by fixing the length of the RVE as twice the aspect ratio of nanoclay platelet, and the width was determined to satisfy the volume fractions. The resulting bio-resin distribution was simplified to develop idealized base cells, or representative volume elements (RVE). These idealized RVES were then used in a multi- 292 level FE scheme (as presented in Chapter 8) to simulate the tensile response of bio-based nanocomposites and the performance of the developed three-phase RVES were corrrpared with existing two-phase RVES and experimental data. The formulation used, examples studied, the determination of target properties and results are provided in the following sections. As expected, results show affinity of the weaker rmterial with the stiff nanoclay platelets. Additionally, multi-FE simulated tensile response from developed idealized RVES performed better than two-phase RVES (see Chapter 8). Also, the simulated tensile responses from idealized RVES were found to agree with average experimental response in initial stiffness and beyond 50-70% of the average experimental ultimate strain. Detailed results and discussions of multi—FE sirmrlations are provided in Chapter 8. Overall, idealized RVES developed from this studywere found to agree reasonably well with experiments. 2.3 Matefléé Deng/z / £01011! 4: 472 Opabzz'zatzbfl Pméé’m In typical material design problem the base cell Y is discretized using four node plane stress elements. The material property within an element is assumed to be constant and can vary from element to element. The target property tensors, in this case the elastic tensor (E‘) and therrml strain tensor (0.") are given and the goal is to find the vector of design variables , p ={ , p2,... pN} that results in effective tensors (BH and a”) that match the target properties as close as possible [9-9]. The elastic and therrml strain tensors (E, a) of the materials studied in this case are given as (E1,a1) , (152,612) and (E3,a3)corresponding to petro-resin (Material-1, M1), clay inchrsion(Material-2, M2) and bio-resin (Material-3, M3), respectively. In this work, the distribution of M2 (clay) is fixed. Also, pe , conesponds to the material property of either M1 (petro-resin) or M3 (bio-resin). The value of 293 pe corresponding to M1 and M3 is constrained to lie in the interval (0,11 The problem is formulated such that for any element the value of pe corresponding to 1 will lead to M3 (bio-resin). Equation (9- 1) describes the assignment of material properties to the elements in the base cell. In Equation (9-1), the value of “a” is unity for all elements, except for those with clay inclusions (MZ), where “a” is zero. Ee = (1 "' “>152 + alPeEs +(1— Pe )Erl (9-1) ae = (1 ”0% + a[pea3 +(1“Pe)arl The material design problem is formulated as a optimization problem where a prescribe amount of material is provided and the deviation from the target tensors, namely weighted mean squares, is minimized [9-9]. Theproblemtobesolvedisztofindp={p, p2,... pN}suchthat: . 1 6 .. 1 3 . mm ¢(p)=— WEZ(E;I —E,)2 +—Wa Zia}! -a,)2 2 [=1 2 [=1 N sub ZAepe =VBIOAmat (9‘2) e=l 05pmn Spa 31 where E I is a compact notation to denOte the six independent terms of an elastic tensor in two dimensions. Similarly, a I denotes the three independent terms of the therrml strain tensors. WE and Wa are scalar non-negative weight factors to emphasize importance to elastic and thermal properties, respectively. A6 is the area of element e, Am, is the total area occupied by the matrix elements (excluding clay), V B 10 is the volume fraction of bio- resin, and pm is a prescribed lower bound for pe , which was taken as 0.001 in this work. 294 The overall formulation of the problem as expressed in Equation (9-2), the effective thermo-elastic properties, determination of sensitivities were adapted from reference [9-8] and are provided as follows: a) figmenized promrties: For any microstructure which is represented by a periodic repetition of a base cell, the effective properties are obtained such that for any test strain (88) applied to the base cell, the average strain energy matches the strain energy in the base cell when subjected to sarrre field [9-9]. The final equations for effective/ homogenized properties are [9-8]: EH H=—£(35 -g )E(ag —gf)dY i,j=1,2,3 =._E(a‘ -5 C‘(I‘))E (gg—gf)dr i,j=1,2,3 (9-3) i=E.-,-( ”Wt, where 86 are linearly independent test strains, 8i and £C(I_')i are fluctuation strains due to applied mechanical and therrml test strains, E: is the effective elastic tensor, ,6; is the effective thermal stress tensor and a; is the effective thermal strain tensor. [ll Sens' . . l l . Thejdetailed derivations of the sensitivities are provided in reference [9—8] and are summarized in the following. The gradient of the objective function 0 required for the numerical solution of Equation (9-2) was obtained as: V p¢=(EH —E*)2§—I:+(aH —a*)aaa—H (9-4) p p 295 . .r sire--- VE:- 3' - ,. 1401451.!" Tin-I ”In Pills“ The sensitivities of effective elastic tensor are given by: E = ‘pr-l Ilsa — Mat -s")dA (9-5) 6p e e Similar to the elastic tensor, the sensitivities of the effective thenml stress tensor are given bx- afl” ppp“ .- c- - - —=-— a —s 1" 81—5} A 9—6 a. A.l( ma. )1 () The sensitivities of effective thermal strain tensor are obtained by differentiating Equation (9- 3) as follows: a__aH _6lE:fl l ‘ WlEHll apH 6” (3-7) 1 _ 9%};pr “sweeter-8% The optimization problem at hand was solved using the method of moving asyrnptotes (MMA). The equality in the constraint equation as given in Equation (9-2) was changed into two inequality constraints such that an upper and lower limit to the bio-resin content could be defined. I-knce, the two inequality constraints are as follows: N sub ZAepe 2 0951/3104,“ (9-8) e=l N sub ZAepe S. 1.10.1/310Ama, (9-9) e=1 This suggests that the amount of bio-resin can vary between 95% and 110% of the actual value of the bio-resin content as prescribed in the experiments/ target properties. This tolerance in the amount of bio-resin was provided as a means to indirectlytake into account 296 ‘filf'ha' the variations in experimental data, since only the average properties are used in prescribing target properties. 2 4 Care Ira/125: 9.4.1 Introduction to the models/ cases: The examples solved in this study aimed at modeling varying bio-resin content (10°/o and 20%) and varying clay morphology (exfoliated and intercalated). Additionally, it was observed that phase-separation between petro- and bio-resin at bio-resin contents starts at bio-resin contents beyond 10°/o. This indicates that at bio-resin contents below 10% there is sufficient cross-linking between the bio- and petro-resins, and the rrraterial essentially behaves in a homogenized rmnner and not as discrete rmterial distribution as assumed in this study. This would also suggests that for base cells with 20% bio-resin content only 10% is available for discrete distribution and the other 10°/o is cross-linked with the petro-resin in a true homogenized blend. The properties of this homogenized blend was lower than pure perm-polymer and hence reduced properties must be taken for the cross-linked blend in such cases. To address this issue, a base cell with exfoliated and intercalated morphology containing 20% bio-resin was optimized such that only 10°/o of bio-resin was available for discrete distribution while the remaining 10°/o was assumed to be cross-linked with petro- resin. Finally, for a multiple parameter objective function like the one used in this work, it has been found that providing uniform initial value of design variable may yield unsatisfactory results and it is thus suggested that the initial value of design variable be distributed linearly decreasing away from the centroid of the base cell [9-10]. Hence, two models containing 10°/o bio-resin contents were modeled with the aforementioned distribution of initial values. A summary of the cases studied are provided in Table 9-1. 297 9.4.2 Description of models The models studied in this work can be classified into two types namely: exfoliated models (Figure 9-1a) and intercalated models (Figure 9-1b). The white region in Figure 9-1 shows the fixed location of the clay platelet and the blue region is the design dormin (matrix) to consist of bio-resin and petro-resin. The RVE dimensions were taken such that the length of the RVE is twice the aspect ratio of the exfoliated platelet. The width of the RVE was computed to satisfy the clay/ matrix concentration. In this study a constant clay content of 2.5 wt.% is considered. For models with exfoliated morphologies, the RVE had dimensions of 260 nm x 35 nrrr. Similarly, RVES for the intercalated rrrorphologyhad dimensions of 260 nm x 105 nm Both models are discretized with square elements having dimensions of 1 nm x 1 nm. Hence the exfoliated and intercalted models have 9100 and 27300 elements. The nanoclay platelets were placed in the center of RVE thereby yielding symmetric results. 9.4.3 Determination of Target Properties The experimental properties avaihble for various bio-based polymer nanocomposites included tensile modulus and coefficient of thermal expansions for various material concentrations. Also, the Poisson’s ratios for neat resins (no clay) were available. Additionally, the effect of bio-resin on the aforementioned properties is known. As it can be seen, the available experimental data is insufficient to exactly define all six independent values of the constitutive tensor (E *). Hence, certain assumptions were nrade in defining the target properties. The effective properties of the model considering the matrix was comprised of all UPE (no bio-resin) was first obtained (E51135 and ang). Secondly, the matrix properties 298 h 2.1.. zit-21:1 . . "ti of neat bio-blends (in accordance to the petro/ bio ratios) were obtained. The average decrease in tensile modulus (y) and average increase in CTE (,6) due to the addition of a known concentration of bio-resin was known from experiments. It was assumed that the bio—resin affects the properties in the longitudinal direction of the clay reinforcement. Along the transverse direction the properties of bio-blends was assumed. The prediction of target properties is illustrated with the example of Case-1, namely erdoliated rrrorphologywith 10% EML content. The homogenized properties of the rrrodel with matrix properties equal to neat UPE resin (Ii-3.65 GPa, v=0.3805, or =100.1 rim/°C) are: "5.531 1.634 0.0007 ”76.103l 551.1,: 4.327 0.000 ; a5”: 107.578 urn/°C (9-10) _ sym 1.342_ _ 0.000 _ The homogenized properties of the model with matrix properties equal to a neat UPE/EML blend with 10°/o EML (E =3.18 GPa, v=0.3923, a =105.0 tun/°C) are: 74.950 1.485 0.0007 77.9707 553,10: 3.812 0.000; 0:53,,0: 113.83 rim/°C (9-11) _ sym 1.159_ _ 0.000 _ Therefore, the target propeties were defined from the combination of Equation (9- 10) and (9-11) as follows ”7(5531) 7(1.634) 0.000“ 'fl(76.103)‘ E*= 3.812 0.000 ; a*= 113.83 um/°C (9-12) _ sym 1.159_ _ 0.000 _ As explained earlier, the effect of nanoclay on transverse direction properties is assumed negligible, and hence the reduction in the elastic tensor tenns involving the 1- 299 direction due to addition of bio-resin was applied to the terms obtained from the UPE homogenized properties. The target properties for other cases in this study were similarly determined and are reportes later along with the results of the respective cases. The values of y for 10% and 20% bio-resin contents were 0.97 and 0.64, respectively. Similarly, the values of [3 for 10% and 20% bio-resin contents were 1.076 and 1.151, respectively. The description of the study cases and their results are only provided in the following. The discussions of results of the various cases follow the results (Section 9.4.12). 9.4.4 Case- 1: Exfoliated, 10% bio-resin, 10% available for layout optimization This case models an exfoliated clay morphology with 10% bio-resin content. The initial value of the design variables was distributed uniformly to all matrix elements. Figure 9-2a shows the histories of the objective function, magnitude of the design variable vector, and the values of constraints, as the design evolved. Figure 9-2b and Figure 9-2c show the bio-resin distribution at iteration steps of 50 and 200, respectively. The target properties are given in Equation (9-13) and the effective properties upon convergence are provided in Equation (9-14). The discussion of the results is provided in Section 9.4.12. ”5.365 1.585 0.000“ "81.887“ 13*: 3.812 0.000; a*= 113.729 um/°C (9-13) L sym 1.159_ _ 0.000 _ "5.323 1.494 0.0001 ”76.189' 15” = 3.806 0.000 ; aH = 116.534 um/°C (9-14) _ sym 1.162_ __—2.177_ 300 9.4.5 Case-2: Exfoliated, 20% bio-resin, 20% available for layout optimization This case models an exfoliated clay morphology with 20% bio-resin content. The initial value of the design variables was distributed uniformly to all matrix elements. Figure 9-3a shows the histories of the objective function, magnitude of the design variable vector, and the values of constraints. Figure 9-3b and Figure 9-3c show the bio-resin distribution at iteration steps 50 and 200, respectively. The target properties are given in Equation (9-15) and the effective properties at convergence are provided in Equation (9-16). The discussion of the results is provided in Section 9.4.12. ”3.540 1.0455 0.0001 "87.4881 13*: 2.321 0.000, a*= 122.270 um/°C (9-15) _ sym 0.668_ _ 0.000 _ ”3.840 1.153 0.000‘ "94.190“ EH = 3.407 0.000 ; aH = 119.960 urn/°C (9-16) _ sym O.987_ _—O.272_ 9.4.6 Case-3: Exfoliated, 20% bio-resin, 10°/o available for layout optimization This case models an exfoliated clay morphology with 20% bio-resin content. It is assumed that 10°/o of the bio-resin in cross-linked with petro-resin (and thus well homogenized as a single matrix material), and only 10°/o is available for discrete material design. The initial value of the design variables was distributed uniformly to all matrix elements. Figure 9-4a shows the histories of the objective function, the magnitude of the design variable vector, and the constraints as the design evolved. Figure 9-4b and Figure 9-4c show the bio-resin distribution at iteration steps 50 and 200, respectively. The target properties are given in Equation (9-15) and the effective properties at convergence are provided in Equation (9- 17). The discussion of the results is provided in Section 9.4.12. 301 "3.840 1.153 0.0001 "94.190‘ EH = 3.407 0.000 ; aH = 119.960 tun/°C (9-17) _ sym O.987_ _— O.272_ 9.4.7 Case-4: Exfoliated, 10% bio-resin, Initial Value Varies Linearly This case models an exfoliated morphology with 10°/o bio-resin content. The initial value of the design variables was varied linearly, decaying away from the centroid of the RVE. This was done as it has been reported to improve convergence performance. Figure 9-5a shows the histories of the objective function, magnitude of the design variable vector, and the values of constraints, as the design evolved. Figure 9-5b shows the distribution of initial value of design variable. Figure 9-5c and Figure 9-5d show the bio—resin distribution at iteration steps 50 and 200, respectively. The target properties for the 10% bio-resin materials are given in Equation (9-13) and the effective properties upon convergence are provided in Equation (9-18). The discussion of the results is provided in Section 9.4.12. "5.345 1.502 0.000“ ”75.399“ 5” = 3.818 0.000 ; aH = 115.360 rim/°C (9-18) __ sym 1.153_ _—1.4561_ 9.4.8 Case-5: Intercalated, 10°/o bio-resin, 10°/o available for layout optimization This case models an intercalated clay morphology with 10°/o bio-resin content. The initial value of the design variables was distributed uniformly to all matrix elements. Figure 9-6a shows the histories of the objective ftmction, magnitude of the design variable vector, and the values of constraints, as the design evolved. Figure 9-6b and Figure 9-6c show the bio-resin distribution at iteration steps 50 and 200, respectively. The target properties are 302 provided in Equation (9-19) and the effective properties upon convergence are provided in Equation (9—20). The discussion of the results is provided in Section 9.4.12. 5*: l— 5.018 1.591 0.0008 3.813 0.000 ; sym 1.159_ _4.947 1.468 0.000- 3.817 0.000 - sym 1.182_ b — 111.120 ' 87.644l _ 0.000 ' 84.002 ' 1 16.090 _—0.237 um/°C (9-19) um/°C (9-20) 9.4.9 Case- 6: Intercalated, 20% bio- resin, 20% available for layout optimization This case models an intercalated clay morphology with 20% bio-resin content. The initial value of the design variables was distributed uniformly to all rmtrix elements. Figure 9-7a shows the histories of the objective function, rmgnitude of the design variable vector, and the values of constraints, as the design evolved. Figure 9-7b and Figure 9-7c show the bio-resin distribution at iteration steps 50 and 200, respectively. The target properties are provided in Equation (9-21) and the effective properties upon convergence are provided in Equation (9-22). The discussion of the results is provided in Section 9.4.12. ”3.311 1.191 0.0001 E*= 3.255 0.000 ; Lsym 0.967_ "3.804 1.191 0.0001 EH : 3.255 0.000 - sym 0.967 L— _ 303 “93.753“ 118.23 _ 0.000 _ ’ 97.9861 120.220 Hm/°C _ 0.046 (921) pm/°C (9-22) 9.4.10 Case- 7: Intercalated, 20% bio-resin, 10°/o available for layout optinrization This case models an intercalated clay morphology with 20% bio-resin content. It is assumed that 10% of the bio-resin in cross-linked with petro-resin (and behaving in a homogenized rmnner), and only 10% is available for discrete material design. The inrtial value of the design variables was distributed uniforrnlyto all the matrix elements. Figure 9-8a shows the histories of the objective function, rmgnitude of the design variable vector, and the values of constraints, as the design evolved. Figure 9-8b and Figure 9-8c show the bio- resin distribution at iteration steps 50 and 200 respectively. The target properties are given in Equation (9-21) and the effective properties upon convergence are provided in Equation (9- 23). The discussion of the results is provided in Section 9.4.12. “3.818 1.301 0.000“ '95807‘ EH: 3.358 0.000 ; aH : 116.590 rim/or: (9-23) _ sym 1.025_ _ 0.254 _ 9.4.11 Case-8: Intercalated, 10°/o bio-resin, Initial Value varies linearly This case models an intercalated clay morphology with 10°/o bio-resin content. The initial value of the design variables was varied linearly, decaying away from the centroid of the RVE. Figure 9-9a shows the histories of the objective function, magnitude of the design variable vector, and the values of constraints, as the design evolved. Figure 9-9b shows the distribution of initial value of design variable. Figure 9-9c and Figure 9-9d show the bio-resin distribution at iteration steps 50 and 200 respectively. The target properties for 10°/o bio- resin materials are provided in Equation (9-19) and the effective properties upon convergence are provided in Equation (9-24). The discussion of the results is provided in the following section. 304 . I'I‘I "5.345 1.502 0.0001 "75.3991 EH : 3.818 0.000 ; aH : 115.360 rim/oc (9-24) _ sym 1.153_ _—1.4561_ 9.4.12 Discussion of the case-studies and Idealized RVES The material design based optimization studywas performed on eight cases that had varying amounts of bio-resin in UPE, exfoliated and intercalated nanoclay morphology and different intial values for design variables. It was observed that for all the cases studied that the objective function reached convergence at approximately after 50 iterations. Although srmll variations continued to occur after 50 iterations, the change in the values of objective function and the resulting effective properties was minimal. Overall, the effective elastic tensor agreed reasonably with prescribed target properties for all cases (For e.g., E11 was within 0 - 14% of target property). The same agreement was not observed for the themral strain tensor. One reason could be that the values of themral strain tensors are in the order of 10" lower than the individual elastic tensor properties, and hence the sensitivities due to themnl properties may be having a minimal influence on the objective function and thereby resulting rmterial layout. Guest et al. [9-10] report similar problems with a bi-objective problem wherein the maximize stiffness and permeability. They suggest that the weight factors be modified to address these issues. In this work, models that varied the weight factors of elastic and tensor parts of the objective function, and models that had the elastic and therrrral parts of objective function normalized with respective target properties were studied. Results similar to the one reported here were obtained. Nevertheless, additional analyses should be perfomred to address the non-zero parameter in the tlrerrml strain tensor (ag ). Diaz and Benard [9-9] used a formulation wherein they scaled the matrial properties 305 'urr‘ Jr J“ and also include weight factors for each of the individual components of the elastic tensor. It is possible that such approaches would allow better control of target properties and thereby the resulting rmterial layouts. The models with 20% bio- resin contents had effective properties whose objective function did not fully reach global mirrima. The problem at hand is an error-minimization problem and hence the global minima should occur at zero, indicating a perfect rmtch of target and effective properties. It should be noted that the target properties used in this work had rmny assumptions. This could be one of the reasons for the objective ftmction failing to reach the global minimum. For any optimization problem, the effect of parameters governing the solution strategy, in this case MMA algorithm, such as step size, penalty parameters, and filtering of sensitivities rrrust be studied. Such a study was carried out in this work for the case of exfoliated morphology and 10°/o bio-resin content. For brevity purposes detailed results of the study are not provided, but a brief summary is given in the following. It was found that a penalty parameter (p) of 5 allowed faster convergence of results. Values greater than p-5, led to solution convergence away from the global mirrirna. Similarly, a step size of 0.75 led to faster convergence, but in some cases caused the solution to diverge. Overall, the default values of the MMA algorithm [9-12] with the penalty parameter value of 3 and step size value of 0.5 perfomned reasonably well and were thus used in this work. As mentioned earlier, the objective of the study was not to accurately solve the material design problem but to provide an insight into the distribution of bio-resin that would allow the deve10pment of enhanced three-phase RVES for multi-FE simulations. Considering the variations in experimental results and the assumptions involved in determining the target properties, it is difficult to make strong conclusions on the resulting 306 ‘F material layout. Nevertheless, as expected it was observed that the weaker bio-resin gravitated to accumulate around the clay platelet. It was also noted that for 10% bio-resin content the bio-resin elements were distributed along the periphery of the clay platelet, while for 20% bio-resin layout was accumulated at the clay particle ends. The reduction in elastic properties for the 10% bio-resin content was minimal (7 =0.97) relative to 20% bio-resin content (7 -0.64). Considering the load transfer between the clay platelet and the surrounding matrix suggests that in order to reduce the stress-transfer efficiency, the weaker rrraterial should affect the interface between the stiff material and the rrntrix. This notion supports results obtained for models with 10% bio-resin contents. One would expect similar results for models with 20% bio-resin content. Nevertheless, it was found in these cases that the weaker material accumulated in pocket type regions around the ends of clay platelet. When stress transfer along l-direction occurs, these bio-resin rich areas at clayends will provide considerable compliance and hamper the efficiency of the clay platelet. In spite of the assumptions and shortcomings in convergence of solution, the study revealed two unique bio-resin distributrion for 10% and 20% bio-resin contents. The resulting material layouts were simplified and idealized RVES were proposed/ developed for these two bio-resin contents. Figure 9-10a and Figure 9-10b show the simplified idealized RVES developed from this study. Overall, the material design based optimization approach is a very powerful technique, and holds great promise in understanding phenomena that cannot be observed or experimentally measured. Nevertheless, the accuracy of the technique is as good as the assumptions and hence care should be taken in drawing conclusions, considering the random nature of the material, the simplicity of the models (single platelet and gallery) and the lack of accurate target properties. 307 2} Alix/2‘11}! szhzzktz'ow: Emile éeflaw‘ortazhg idealized/1? 14.": The idealized three-phase RVES (Figure 9-10) obtained from the optimization based material layout design study were used in multi-level FE algorithms to predict the tensile behavior of bio-based nanocomposites. Details of the multi-FE approach, the material models used and comparison of the two—phase and three-phase RVES along with random clay distribution are provided in Chapter 8. Figure 9-11a and Figure 9-11b provide the comparison of tensile response from the multi-FE simulation predictions and experiments for bio-based polymer/ clay nanocorrrposites with 10°/o and 20% EML contents respectively. Overall, the enhanced three-phase RVES obtained from the outcome of the material layout study reported here was found to capture the non-linear response better and rmtch the average experimental data very well. For instance, the multi-FE simulated tensile response from the idealized RVE for 20% bio-resin (Figure 9-10b) agreed with average experimental response for up to 70% of average experimental ultimate strain, and the stress prediction from the sirrrulation at average experimental ultimate strain deviated within 5% of average experimental ultimate stress. This was expected as the accuracy of the rrrodel increases when the material is modeled as realistic as possible. Detailed results of the multi-FE sirmrlations are provided in Chapter 8. 2 6 Conclusion Novel eco-friendly bio-based nanocomposites obtained from reinforcing layered silicates (nanoclays) in blends of petroleum and natural polymers (bio-resin) have been shown to exhibit properties that are superior or similar to the base petroleum-polymer. The addition of bio-resin reduces stiffness and barrier properties, which can b recovered by the use of nanoclay reinforcement. The distribution of nanoclay can be observed experimentally 308 through electron rrricorscopy, but similar observations on the distribution of petro and bio- resin constituents is infeasible. To understand and model these materials efficiently, the distribution of bio/petro resin was sought. The rmterial layout problem presented in this work was aimed to develop an enhanced RVE for multi-level computational simulation. The was the use of topology optimization along with experimental data. The goal was to provide to provide insight to the distribution of bio-resin in the RVE, and not necessarily to accurately solve the material design problem. As with any optimization problem, the accuracy of the rmterial design topology optimization problem used in this work, which minimizes the error between the experimental target properties and homogenized model predictions, depends on the input parameters. Any assumptions or inconect target properties will clearly influence the outcome of the optimization problem, and the obtained solution is specific to the problem considered and cannot be generalized to other material compositions. Solution of the material layout problem revealed affinity of the weak bio-resin material to the stiff nanoclay platelet. This was expected as bio-resin addition reduced mechanical properties, suggesting that it affected the stress-transfer between the stiff inclusion and the matrix. Additionally, polymer chemistry literature indicates that frmctionalization/ chemical affinity of the bio- resirr can have a strong influence in attracting the nanoclay [9-4], thereby supporting the results from the study. Idealized RVES for 10% and 20% EML contents with 2.5 wt.% nanoclay, with exfol'nted and intercalated morphologies, were deduced from material layout studies and used in multi-level FE predictions of the tensile behavior of bio-based nanocomposites. The tensile response predictions using the idealized RVES developed from this study captured the non- linear response and agreed with average experimental response in initial stiffness. Also, in case of idealized RVE comesponding to 20% bio-resin content, the simulations agreed with average 309 experimental response for up to 70% of average experimental ultimate strain, and the stress prediction from the simulation at average experimental ultimate strain deviated within 5% of average experimental ultimate stress. Overall, the nnterial design based optimization approach is a very powerful technique and holds great promise in designing unique materials, and in understanding phenomena that cannot be observed or measured experimentally. 2 7 Ace/20 wkafgemeflo’ This work was supported by the National Science Foundation under grant CMS- 0409666. The implementation of the method of moving asymptotes used here was provided by Prof. Krister Svanberg from the Department of Mathematics at KTH in Stockholm. We thank Prof. Svanberg for allowing us to use his program. Authors are thankful to Dr. Alejandro Diaz, Professor of Mechanical Engineering, Michigan State University, for his help in understanding the theory and formulation of the problem during this study. Authors are also thankful to Dr. Amar K Mohanty and Dr. Manju Misra, School of Packaging, Michigan State University, for their expertise in chemistry and processing of these materials. 310 2 (Y 745%: 4,2116};ng Table 9- 1. Model Description for the material design layouts studied in this work Weight Bio- resin content Initial Factors Volume fraction Value of # Model Nanoclay design Name Morphology WE W A Ac trial Available .for Variable optimization ,, pim' 1 Case-1 exfoliated 1 1 10 10 0.1 2 Case-2 exfoliated 1 1 20 20 0.2 3 Case-3 exfoliated 1 1 20 10 0.1 4 Case-4 exfoliated 1 1 10 10 1‘93?“ vanatron 5 Case-5 intercalated 1 1 10 10 0.1 6 Case-6 intercalated 1 1 20 20 0.2 7 Case-7 intercalated 1 1 20 10 0.1 8 (3:868 intercalated 1 1 10 10 1496?“ vanatron "' Uniform distribution of pim- assumed except for case-4 and case-8 wherein linear variation of initialvalue ofdesignvariable withvalue decreasingas theelement is awayfromthe centroid ofthe RVE. 311 ‘ --. (a) Figure 9—1. Base Cells: 2) Exfoliated Morphology, b) Intercalated Morphology. The white region represents the nanoclay and the blue region represents the design domain containing blends of bio—resin and petro-resin 312 A D V 0.045 30 5 0.040 i 25- g 0.035 0. 0.030 g 0.025 — 20' 3. ct — 0.020 — 3 15- '6 0.015 _g 0.010 101 g 0.005 , 0.000 . . . . . . . . 5 . . . . . - . - - 020406080100120140160180200 020400080100120140160180200 Iteratlon Iteratlon 0.0003 41.0526, :3 l a 3 '°-°9°4 3 0.0527 ‘.' .- g 0.0905 g 41,0523 g '°-°°°° g 0.0529 0 % 0.0907 13 4.0530 - O 5 41-0903 3 0.0531 . _ > 4.00004 ‘>' 0.0532 7 43.0910 . . - . - - - 410533 . . . . . . . Iteration mmuon (b) Iteration # 50 (c) Iteration # 200 1 0.5 0 Figure 9-2. Case-1, Exfoliated morphology with 10°/o bio-resin. 21) Evolution of design, plots showing change of objective function, magnitude of design variable vector, and constraints, b) material layout at iteration step 50, and c) material layout at iteration step 200. 313 A 3 22 S 2.0 ‘ E 1.0 u 15 35‘ g . g u 0. 30 E 12 ‘6 25- 3 1.0 - 20 g 0.3 ’ 1 0.5 . . . . . . . 15 . . . 0 20 40 50 50 100 120 140 100 150 200 0 20 4o 50 30 100 120 140 150 100 200 Iteration Iteration 0.0500 0.0525 .. at 0.0520 F 3 0.0092« 3 0.0530 ‘: ‘r 'E .00.“. g 0.0532 g g 0.0534 5 0.0000 0 0.0535 0 o '5 ‘5 0.0530 0 4.0898 3 3 _ 0.0540 I l > 0.0900 > '0-05‘2 . 0.0544 . . . 020406080100120140160180200 020400000100120140160180200 Iteration Iteratlon b) Iteration # 50 (c) Iteration # 200 1 E D 0.5 0 Figure 9-3. Case-2, Exfoliated morphology with 20% bio—resin. a) Evolution of design, plots showing change of objective function, magnitude of design variable vector, and constraints, b) material layout at iteration step 50, and c) material layout at iteration step 200. 314 A h) v 1.5 35 C 1'0; 1.5 30‘ . g 1.4 u. 251 g 1.3 o 12 q, 20- 3 11 — g - 15- g 1.0 T: ‘01 > 0.9 ' 0.3 . . . . . . . . . 5 . . . . . . . 0 20 4o 50 30 100 120 140 100 130 200 o 20 40 00 30 100 120 140 100 130 200 Iteration iteration 0.0526 A 0.0093- ,. 0.0523 .. N 3 3 0.0530 'r 0.oeoo« ... c:- g @3532 3 0.0002- g 0.0534 3 5 0.0535 g '0' ‘ 3 0.0535 0 0 g .o_m. _3 0.0540 3 ; 0.0542 '°'°’°°‘ 0.0544 020400000100120140100100200 ozoaocoaowououommzoo Iteration Iteration b) Iteration # 50 1 0.5 0 (c) Iteration # 200 1 0.5 0 Figure 9-4. Case—3, Exfoliated morphology with 20% bio-resin content and only 10°/o available for material design . a) Evolution of design, plots showing change of objective function, magnitude of design variable vector, and constraints, b) material layout at iteration step 50, and c) material layout at iteration step 200. 315 A D v 0.035 35 C 0 0.0301 3 30' t 5 0.025- n. 25 . 5 0.020- _ E a 20- 3 0.015- -- ° 15 '6 0.010- § 10‘ g 0.005- , 0.000 . . . . . . . . . 5 . . . . . . . . 0 20 40 00 00100120140100100 200 0 20 40 00 30100120140100100200 Iteration iteration 0.0525 0.0903 ... .. 0.0525 :9 0.0904- 5 ‘ ._. ._- 0.0521 ‘E 0.0905 ‘E E E 0.0523 g "090° g 0.0529 % 0.0907 g 0.0530 § 41-0908 3 0.0531 u _ > 0.09094 '>° 0.0532 0.0910 0.0533 20400080100120140160180200 20400000100120140160180200 iteration iteration (b) Initial Value of Design Variable ..an... , nous-» 0.2 w a “I" 1 U 0.1 ' ' 0 (c) Iteration # 50 (c) Iteration # 200 0.5 0 Figure 9-5. Case-4, Exfoliated morphology with 10°/o bio-resin. a) Evolution of design, plots showing change of objective function, magnitude of design variable vector, and constraints, b) Distribution of initial value of the design variable, c) material layout at iteration step 50, and c) material layout at iteration step 200. 316 A N V 0.05 45 c 1 o I E 0.04« ‘° C 3 ll. % 0.03 _ . e. 3 _ o 0.021 25 ‘8 3 001 l i ' , go . > 0.00 15 . . . . . . . . 020400030100120140100130200 020400030100120140150130200 iteration Iteration 0.05200 0.09072 A 0.05262 3 ‘3: 3 0.09074 .. 0.05204 1; 0.09070 1; 0.05250 2 0.09075 E 0.05200 1: 0.09090 2 0.05270 0 8 0.09052 0 0.05272 .- ‘- 0 4.09034 2 43.05274 7!: 0.09009 7:: ”-0527“ > 0.09000 > 0.05278 0% . . . 0.05230 . 020400050100120140160130200 020400030100120140100100200 Iteration Iteration b) Iteration # 50 (c) Iteration # 200 0 Figure 9-6. Case-5, Intercalated morphology with 10% bio—resin content a) Evolution of design, plots showing change of objective function, magnitude of design variable vector, and constraints, b) material layout at iteration step 50, and (2) material layout at iteration step 200. 317 A 3 2.0 00 S 1.3 . ' 70 ‘2 1.3 3 II. g 1.4 so . E 1.2 e. 3‘ '_ . o 1.0 50 ‘6 03 2° ' 40- ; 0.6 i i 0.4 . . 30 . . . . . . . 0 20 40 so so 100 120 140 100 190 200 o 20 40 so 30 100 120 140 160 130 200 iteration Iteration 0.09050 ;_~ 0.0905 g 3 3 0.05271 .. 0.09060 .. é a E 4.09035 3 43.0528‘ = 0.09070 g 8 0 0.0529- .5 0.09075 '6 O 7% 0.09090 _3 0.0530 > 0.09035 > 0.09090 . 0.0531 . . . . . . . o 20 40 so so 100 120 140 150 190 200 0 20 40 so 30 100 120 140 160 130 20¢ Iteration Iteration b) Iteration # 50 (c) Iteration # 200 0 Figure 9—7. Case-6, Intercalated morphology with 20% bio—resin content and all 20% available for material design. a) Evolution of design, plots showing change of objective function, magnitude of design variable vector, and constraints, b) material layout at iteration step 50, and c) material layout at iteration step 200. 318 A 3 1.5 00 5 1.41 .. 5°_ _ E 13 3 IL 3 12 ‘0 . 15 11 q % _ o 1.0 30 - '6 09 § ' ,0. g 03 . 0.7 . . . 10 . . . . 0 20 40 30 30 100 120 140 130 130 200 o 20 40 00 30 100 120 140 190 130 200 Iteration Iteration A 0.09012 5; E 0.09074 3 0‘0”“ '7 009073 '- E . 15 0.05270 < E 0.09073 g E 009080 2 0.05275- 0 O 2 0.09032 0 . 5230 - 2 0.09034 E 0° 3 2 3 0.09035 a 0.05235 > > 0.09033 0.05m . . . 0.05290 . . 0 20 40 30 30 100 120 140 100 130 200 0 2° 4° 60 80 100 120 140 160 180 200 Iteration iteration b) Iteration # 50 1 0.8 0. 0. 0.2 0 c) Iteration # 200 0.8 0.6 0.4 0.2 Figure 9 8. Case—7,1ntercalated morphology with 20% bio— resin content and only 10% available for material design . a) Evolution of design, plots showing change of objective function, magnitude of design variable vector, and constraints, b) material layout at iteration step 50, and c) material layout at iteration step 200. A 319 ’3 v 0.040 50 g 0.035' 45- I a 0.030 40 IL 3 0.025 _ 35- } a 8 0.020 — 30- ‘5 0.015 25- i I , - > 0010 ’ 20‘ 0.005 . . . 15 . . . . . . 0 20 40 00 00 100 120 140 100 100 200 0 20 4o 00 00 100 120 140 100 100 200 Iteration Iteration 7 0.05200 0.090 0 . A i A 0.05202 .— N e 2 0.05204 E 0.09075< Te. 0.05200 E E 0.05203 " 0.05270 0.090001 3 13 0 0.05272 ‘- E 2 0.05274 ,2. ‘°'°°°°5 ,3, 0.05270 > > 0.05270 0m90 0.05230 . . . 0 20 40 00 30 100 120 140 100100200 0 20 40 60 501001201401601802M Iteration Iteration (b) Initial Value of Design Variable Figure 9—9. Case-8, Intercalated morphology with 10% bio-resin. a) Evolution of design, plots showing change of objective function, magnitude of design variable vector, and constraints, b) Distribution of initial value of the design variable, c) material layout at iteration step 50, and c) material layout at iteration step 200. 320 Figure 9—9 continued. . . (c) Iteration # 50 1 0.3 0.0 i 4 0.4 02 (c) Iteration # 200 321 (a) Figure 9-10. Simplified, idealized RVES. The green, red and black regions represent petro— resin (UPE), bio-resin (EML) and nanoclay respectively. a) 10% EML, b) 20% EML 322 207 15‘ 10- Stress (MPa) -.- Experimental Data + idealized RVE - single particie (b) 18 I I I I r f 0.001 0.002 0.003 0.004 0.005 Strain (mmlmm) 10 14 12 i 10 Stress (MPa) / / + Experimental Data + Idealized RVE - single particle 0.000 I l I I I I I I I I I l' 1 I I I I I I I I ' I I I I ' I I I I I 0.001 0.002 0.003 0.004 0.005 0.006 0.007 Strain (mmlmm) Figure 9-11. Comparison of tensile response from multi—FE simulations using idealized three-phase RVES with experimental data, (a) 90/10/2.S, and (b) 80/20/2.5 323 29 X49727)“: [9-1]. Haq M, Burguefio R, Mohanty AK, Misra M “Processing techniques for bio—based unsaturated-polyester/ clay nanocomposites: tensile properties, efficiency and limits.” Composites: A 2009; 40:394-403. [9-2]. Haq M, Burguefio R, Mohanty AK, Misra M “Bio-based unsaturated polyester/ layered silicate nanocomposites: characterization and thermophysical properties.” Composites: A 2009; 40:540-547. [9-3]. Haq M, Burgueno R, Mohanty AK, Misra M “Development and 'Ihermophysical rization of Clay Nanocomposites from Blends of Linseed oil and Unsaturated Polyester.” Composites: A, Communicated. [9-4]. Drown BK, Mohanty AK, Pamlekar Y, Hasija D, Harte BR, Misra, M, Kurian JV. The surface characteristics of organoclays and their effect on the properties of poly(trimethylene terephtahalate) nanocomposites. Composites Science and Technology, 2007, 67:3168-3175. [9-5]. Sigmund 0., "Design of material structures using topology optimization." PhD. Thesis, Department of Solid Mechanics, Technical University of Denmark 1994. [9-6]. Sigmund 0., ”Materials with prescribed constitutive parameters: an inverse homogenization problem” International Journal of Solids and Structures, 1994; 31(17)2313-2339. [9-7]. Sigmund 0., "Tailoring materials with prescribed constitutive parameters.” Mechanics of Materials, 1995;20:351-368. [9-8]. Sigmund O. and Torquato 8., "Design of Materials with Extreme Thermal Expansion using a Three-Phase Topology Optimization Method," Journal of the Mechanics and Physics of Solids, 1997; 45(6):1037- 1067. [9-9]. Diaz, A, and Benard, A., "Designing materials with prescribed elastic properties using polygonal cells," International Journal for Numerical Methods in Engineering, 2003; 57:301- 314. [9-10]. Guest, JK. and Prevost, J, H, "Optimizing multifunctional materials: Design of microstructures for rmximized stiffness and fluid penneability,” International Joumal of Solids and Structures, 2006; 43 :7028-7047. [9-11]. Guest, JK. and Prevost, J, H, "Design of maximum permeability structures.” Computer Methods in Applied Mechanics and Engineering, 2007; 196:1006— 1017. [9-12]. Svanberg K., "The method of moving asymptotes." International Journal for Numerical Methods in Engineering, 1987; 24:359 -373. 324 [9-13]. Kruijf, N., Zhoue, S., Li, 0., and Mai, Y.W., "Topological design of structures and composite nnterials with rmrltiobjectives,” International Journal of Solids and Structures, 2007; 44:7 092-7109. [9-14]. Zeng, Q.H.,Yua, A.B., and Lub, G.Q., "Multiscale modeling and simulation of polymer nanocomposites," Progress in Polymer Science, 2008; 33:191-269. [9-15]. Sigmund 0., "A 99 line t0pology optimization code written in Matlab,” Structural and Multidisciplinary Optimization, 2001; 21:120-127. [9-16]. Bendsoe MP., ”Optirml shape design as a rmterial distribution problem," Structural Optimization 1939; 1:193-202. 325 Chapter 10. Summary and Conclusions 10. l Symmdoa' Environmental concerns related to the use of synthetic, or petroleum based, polymer matrix composites has propelled the development of composite nnterials based on natural or renewable sources. Biocomposites, composed of natural fibers in synthetic or natural polymer matrices have recently gained much attention due to their low cost, eco-friendliness, and their potential to compete with synthetic composites. Nonetheless, the use of bio-based composites has been limited due to their lower mechanical and thermophysical properties compared to synthetic composites and conventional structural rmterials. In this work, hybrid bio-based composites consisting of multiscale reinforcements, namely natural fibers and nanoclay embedded in blends of petroleum based (unsaturated polyester) and vegetable oil based resins (derivatives of soybean and linseed oils), were developed through integrated experiments and computational simulations. Experimental studies included development of novel processing techniques for bio-based polymer nanocomposites and detailed characterization of physical, thermal, mechanical and banier properties of both nanocomposites and biocomposites. Computational simulations that take into account the detailed morphologies at various hierarchical length scales were developed and validated/ integrated with experiments. Electron microscopy aided the development of realistic computational models (RVE, representative volume elements). The distribution of bio—resin in nanocomposites could not be observed experimentally (microscopy), hence, the available experimental data was used along with topology optimization to determine the distribution of the bio-resin and develop an enhanced RVE for modeling the three-phase 326 material. A multi-level finite element approach was implemented to link different length scales. The multi FE approach was used to evaluate the tensile response of clay+bio-based nanocomposites and micro-fiber+clay+bio-based composites. Overall, coo-friendly, tailorable, cost-effective and multiscale reinforced bio-based composites were successfully developed through the integration of experiments and computational simulations. It is believed that the approach of integrating simulations and experiments, as attempted in this work, holds great promise and similar methodology can be applied for other types of hierarchical materials, thereby providing guidance in designing those materials. [0.2 fleremrfi Firm/19w 10.2.1 Processing of Bio-based Nanocomposites I Processing plays a vital role in the resulting properties of both nanocomposites and biocomposites. ' In order to exploit the benefits/ synergy offered by the hybrid hierarchical bio-based _ materials studied in this work, incorporation of large amounts of bio-resin and nanoclay is needed, but such incorporation is controlled by the processing limitations of bio-based nanocomposite resins. Also, the overall properties of the bio-based composites are highly dependable on efficient processing of bio-based nanocomposite resins. I Solvent-based processing of nanoclay reinforced bio-based resin systems brings about new challenges, such as phase separation, thermal degradation of the resin system and limitations on the maximum feasible bio-resin and nanoclay content. 327 10.2.2 Two novel efficient solvent based processing techniques were developed through a study that compared four different techniques. The effects of sonication energy, processing time, solvent type and the associated processing issues for each of the techniques were assessed by evaluating the nanoclay morphology and the tensile properties of the resulting polymer nanocomposites. Processes that enable incorporation of relatively large amount of bio-resin and nanoclay content with minimal processing problems, lower processing time and desirable tensile properties were considered to have good overall efficiency. In particular, two processes were found to have good overall efficiency. One of them (Process B) consists of using acetone as a solvent and led to the best nanoclay dispersion and exfoliation, resulting in samples with high tensile modulus. The second one (Process D) consists of direct sonication of nanoclay in the resin system diluted with styrene. This process eliminates the use of a foreign solvent, thereby reducing processing time, and the resulting nanocomposites showed a good balance of tensile properties, namely a balanced improvement of stiffness and toughness. Depending on desired properties and applications, either of these processes is deemed suitable for effective production of bio-based nanocomposites. Study on Bio-based Polymer Nanocomposites Bio-based resin systerm from partial substitution of petroleum based resins (primary) with natural polymers (secondary) provide environmental friendliness, cost- effectiveness and improved toughness. Experimental characterization of thermophysical, mechanical, barrier properties and microscopic observations were studied on bio-based nanocomposites with 328 unsaturated polyester (UPE, primary resin, petro-based) and two types of bio-resins, namely epoxidized methyl soyate (EMS, soybean oil derivative) and epoxidized methyl linseedate (EML, linseed oil derivative), reinforced with nanochy. It was observed that the combination of nanoclays and bio-resins in UPE resin systems lead to composites with similar or better properties than the original virgin UPE resin system. The drawbacks from the addition of bio-resins to the base polymer were shown to be m through nanoclay reinforcement and vice versa. The addition of bio-resin increases toughness but reduces stiffness of the resin system. Similarly, the addition of nanoclay increases stiffness along with brittleness of the resulting nanocomposites, i.e., a reduction in toughness and ductility. The studies showed that a proper stiffness-mm balance can be obtained and can be 9.10124 by controlling the amount of bio-resin and nanoclay contents. Additionally, a comparative study of neat resins (no clay), from blends of EMS and EML in UPE revealed similar tensile properties for b0th resin blends. Nevertheless, considering ease of processing, EML resin is recommended for use. Also, EML composites have relatively better transparencythan EMS composites. If the transparency/translucency of resulting nanocomposites is an indication of phase-separation, EMS composites seemed to have had a higher degree of phase- separation (less transparent), also indicating that EML composites have relatively better compatibilitywith primary petro-resin UPE. 329 10.2.3 Hybrid Bio-based Composites: Hybrid bio-based composites that exploit the my between natural fibers (industrial hemp) in a nanoreinforced (nanoclay) bio-based polymer (unsaturated polyester + linseed/ soybean oil based resins) can lead to improvements in multiple properties while maintaining environmental appeal and cost-effectiveness. The properties of the resulting biocomposites are tailorable and seem controllable by the amount and distribution of the constituents. Experimental characterization studies showed that the addition of bio-resin lowers mechanical parameters, such as stiffness and ultimate tensile stress, but increases toughness parameters, such as impact strengths and ductility. The addition of nanoclay enhances stiffness but seems to decrease toughness. Thus, the study shows that a mstiifmssfimhmbalang can be obtained by controlling the amount of bio-resin and nanoclay content. The enhancement in stiffness due to nanoclay addition was found to be less pronounced in biocomposites than in nanocomposites, as the hemp fibers control the stiffness of the biocomposites The synergistic behavior is not limited to stiffness-toughness balance, and similar balance was observed in thermal and barrier properties, wherein the detrimental properties of bio-resin were recovered by nanoclay addition and vice versa. The multiphase nature of the hybrid bio-based composites revealed multi-functional attributes, such as improved barrier and thermal properties along with improvements in mechanical properties. The multiscale reinforcement provided enhancements at different length scales. The addition of nanoclay improved the properties of the polymer matrix, thereby 330 10.2.4 improving the barrier (thermal + moisture movement) properties, while the micro- sized hemp fibers govern rrrost macroscopic mechanical properties of resulting biocomposites. Tensile fracture surfaces (scanning electron microscopy) revealed an increase in the interfacial gap (gap between the hemp fiber and the nntrix), and increased amount of pulled-out fibers, with an increase in bio-resin content. This indicates an adverse effect of bio-resin on interfacial properties. Increased pull-out of fibers indicates higher dissipation of energy, thereby higher ductility and improved toughness as observed in experimental results. Amlytical and Computational Modeling The success of any modeling to predict the overall properties of a heterogeneous material depends on the representative vohrme element (RVE) or unit cell. The RVE should be selected such that it is large enough to represent the microstructure and small enough to allow efficient computations. Unlike most conventional nanoparticles/ fillers, nanoclays pose unique modeling issues such as a) particle size, b) hierarchicaVintercalated morphology and c) interface and inter-gallery properties. Exfoliated clays may have thicknesses in range of nanometers and highly intercalated clay agglomerations rmy have thicknesses in the order of micrometers. Moreover, it is commonly agreed that polymer clay nanocomposites exhibit both exfoliated hierarchical morphologies. Hence most analytical models that have assumptions of uniform properties with a constant aspect ratio for all particles fail to model nanoclay composites efficiently. 331 ‘m'. '1“. 1. ."h 10.2.5 Finite element based RVEs allow detailed/ realistic modeling of the material including different particle shapes and morphologies, and non- linear material properties with complex loading, and were thus used to successfully model the hybrid material in this study. Finite element RVEs used in this work modeled the unterial as realistically as possible and took into account the random distribution (due to processing) and morphologies (exfoliated, intercalated and both) of nanoclay phtelets. A comparison of FE based RVEs and analytical models (Mod-Tanaka estirmtes) with experimental data revealed better agreement of FE based RVEs with experiments than analytical models. This is attributed to the more realistic modeling of the material. Material layout The distribution of nanoclay in the polymer rmtrix could be observed using electron microscopy, but the distribution of bio-resin in the primary petro-resin cannot be observed by this method. In order to model the hybrid composite in greater detail, it is essential to realistically model the three-phase material, and hence the distribution of bio-resin was sought. The rmterial layout problem used in this work was aimed to develop an enhanced three-phase RVE by using experirrrental data along with topology optimization in an attempt to provide insight to the distribution of bio-resin in the RVE. Thus the goal was not to accurately solve the material design problem. As with any optimization problem, the accuracy of the material design topology Optimization problem used in this work, which minimizes the error between the 332 experimental target properties and homogenized model predictions, depends on the input parameters. Any assumptions or incorrect target properties will influence the outcome of the optimization problem, and is specific to the problem considered and cannot be generalized to other material compositions. The bio-resin distribution resulting from a mater'nl layout problem with any single target parameter (e.g., modulus) rmy not yield the same distribution from any other parameter (e.g., thermal expansion). Nevertheless, in reality the material should have a unique distribution that matches all experimental parameters. As a result, a material distribution/ design problem that uses multiple target parameters, like the one used in this work increases the confidence of the result. Optimization algorithms rely on many input variables, such as step size, penalty parameter, initial value of the design variable, etc. These variables are specific to any given problem and need to be changed for every problem. Solution of the material layout problem revealed affinity of the weak bio-resin material to the stiff nanoclay platelet. This was expected as bio-resin addition reduced mechanical properties, suggesting that it affected the stress-transfer between the stiff inclusion and the matrix. Additionally, in order to incorporate higher amounts of bio- resin and nanoclay, the knowledge from chemists reveal that the bio- resin may have functionalization/ chemical affinity with nanoclay thereby supporting the results from the Study. Idealized RVEs for 10% and 20% EML contents with 2.5 wt.% nanoclay, with exfoliated and intercalated morphologies were deduced from material layout studies and used in multi-FE prediction of tensile behavior of bio-based nanocomposites. 333 10.2.6 Multi-scale simulations / Multi-FEA Multiscale modeling is a term that covers a wide range of simulations and modeling techniques. Ideally, a multiscale modeling scheme would be the one that links all the scales from atomistic level to macro/ structural level. In other words, it should connect the discrete molecular structure with the bulk continuous structure. At the same time, any modeling scheme that links anytwo scales is commonly considered a multiscale modeling scheme, hence making the use of this descriptive analysis term subjective. The most important aspect of multiscale modeling is the accurate prediction of physical/ chemical properties and behavior from a lower scale (e.g., nano or micro scale) to a larger (or rmcro scale) without loss of intrinsic information. In other words, accurate mfir of effective properties from one length scale to other (also termed as hand-shaking) governs the success of the computational simulation. A multi-level finite element (FE) computation does not require any constitutive equations to be written at the rmcroscopic scale; all non-linearites are obtained from separate FE analyses at lower hierarchical (micro/nano) scale. In this work, a multi-level FE algorithm was implemented to model the hybrid bio- based materials. A two-level (nano-micro) multi-FE algorithm was used to predict the tensile response of two-phase (nanoclay + matrix) and three-phase (nanoclay + petro-resin + bio-resin) bio-based nanocomposites. Also, a three-level multi-FE (nano-micro-macro) algorithm was implemented to predict the tensile behavior of bicornposites (natural fiber + nanoclay + petro-resin + bio-resin). For bio-based nanocomposites, the multi-level FE predictions of all the models were in good agreement with the initial stiffness, obtained from average experimental data. 334 The two-phase intercalated models rmtched the average experimental response for up to 50% of average experimental ultirmte strain value, while this value was only 35% for exfoliated model. Also, the stress prediction corresponding to average experimental ultimate strain deviated from experimental stress by 45% and 35% for exfoliated and intercalated models, respectively. Overall, models with intercalated clay models showed better performance than exfoliated clay models. This was expected as the accuracy of the model increases when the rmterial is modeled as realistic as possible. The three-phase idealized single-particle model had the best perfonnance of all models with good agreement up to 70% of the ultimate strain observed from average experimental response. Abo, considering the deviations in the experimental data, the stresses predicted by the single-particle idealized RVE, corresponding to the average experimental ultirmte tensile strain, were within 5% of the average experimental ultimate tensile stresses. Overall, the sirmrlated responses agreed reasonably well with experimental results and were able to capture the nonlinear response. The multi-FE predictions of tensile response of hybrid biocomposites matched average experimental response with respect to initial stiffness. The virgin UPE biocomposite deviated from average experimental response at strain levels of 25% of ultimate strains, while this value was 35% for hybrid biocomposite (20% EML 86 2.5 wt.% nanoclay) using idealized three-phase RVE. This suggests that material non- linearity helps in capturing the non- linear behavior, but is not the only source of non—linearity. The deviation of the simulation from the experimental data is due to many factors, including material non-linearity, random distribution of the short hemp fibers, 335 10.2.7 interaction of the short-fibers, fiber pull-out, straightening of the curved fibers, interfacial properties, etc. Hence, relying purely on material non-linearity will still be unable to capture the non-linear response accurately, particularly near ultimate response. Nonetheless, properly designed multiscale FE simulations can shed light in understanding the aforementioned complexities and sources of non-linearity A simplified RVE was used in this work, with the objective of illustrating the promise of the multi-level computational scheme. A more realistic RVE with enhanced material properties and modeling of the reinforcement morphology would enable accurate modeling of such rmterials. One of the rnaindrawbacks of multi-FE algorithmis thatitis basicallyamore sophisticated homogenization/ averaging scheme, but cannot model damage. Once the damage occurs the periodicity of the RVE ceases to exist and the assumptions of the multi-FE method are violated. Also, stress-concentrations and numerical anomalies (if any) get averaged out and hence the scheme may over-predict the actual response. Overall, multi-FE algorithms hold great promise in modeling different length scale within the continuum regime. Closure In order to understand hybrid materials as the ones dealt with in this dissertation, inchrding the interactions between the polymer blends and the effect of nano- inclusions and micro-fibers on the overall properties, a good understanding of chemistry of such materials combined with atomistic simulations and proper experiments at the respective scales (including nano/ lower scale) is essential. 336 I The current work focused on mechanical behavior. A similar approach can be extended to therrml and diffusion reseonse. The success of this computational scheme will enable prediction of macroscopic responses for various material compositions and macroscopic shapes. A similar methodology could be applied for other types of nanoparticles and hierarchical materials thereby providing guidance in designing those rmterials. I Overall, the study used an integrated experimental and sirmrlations based approach to develop and understand hybrid, hierarchical, tailorable and eco-friendly materials. The study showed that bio-based composites with proper stiffness/ toughness balance can be obtained while preserving environmental friendliness and cost effectiveness. The improved multifaceted features possible for these sustainable bio- based materials are likely to increase their appeal for use in transportation and housing structural applications. [0.3 PfizL/ompéz'mr/ Chat/1111b”; 10.3.1 Sustainability One of the most common terminologies used in the literature for eco-friendly rmterials is “sustainablility”. The term “sustainable” is subjective and at times misleading. Oxford dictionary defines sustainable as “avoiding depletion of natural resources.” Although materials with plant-based origin or eco-friendly appeal may be used to develop “green” rmterials, it is not necessary a sustainable alternative by itself. The amount of energy utilized to produce these rmterials, or the effort spent on making such rmterials feasible may circumvent the benefits offered by their natural origin. Ideally, whenever sustainability is 337 defined, it should be combined with life-cycle assessment and energy efficiency of production and disposal. 10.3.2 Multi-scale Computations The advancements in measurement sciences and availability of computational power rmke multiscale sirmrlations feasible. Nevertheless, the use of these multiscale computational schemes to determine simple elastic properties might be an unnecessary effort Nevertheless, such powerful computational schemes can be used to understand complex materials, time dependent properties, non- linear behavior, modeling damage and phenomena that cannot be measured experimentally. In a similar note, most computational schemes have inherent assumptions in their theory and their numerical models, and thereby the resulting accuracy is dependent on these assumptions. For instance, multiscale schemes that rely on local periodicity cannot model material damage. Once damage occurs, the assumption of local periodicity is lost, and the scheme is invalid. Overall, the success of the computational scheme depends on how realistic/ detailed the models are, the assumptions/ simplifications made in the theory and the parameters under consideration. 10.3.3 Integration of Experimental results and Computational simulations The use of experimental data to validate computational models and the use of such validated models to predict properties of complex materials for constituent concentrations to which experimental data is not available, is one of the major advantages of this approach Nevertheless, the use of average properties to validate the models is questionable, especially when there is a large deviation in the parameter studied. In such cases, a probabilistic approach needs to be taken, and a relatively larger number of simulations need to be performed. 338 10.3.4 Electron Microscopy / Measurements at Lower Scale The use of electron microscopy or in general measurements at lower scale is very subjective and care should be taken in its interpretation. For instance, an electron micrograph studying nano-particle distribution in a polymer may show good dispersion in one experimental slide and completely agglomerated (i.e., bad) dispersion on another. In such cases, the overall dispersion of the particulates is a subjective call. A better understanding of the equipment, its working technique and sample preparation should be taken into consideration before conclusions from such sensitive observations are made. For instance, in transmission electron microscopy, the rmterial that is kept under observation has a size of few microns with a thickness of approximately 100 nm. Considering the macroscopic size, the randomness in processing techniques, and varied observations in different rmterial slides, drawing strong conchrsions on the rmcroscopic material from observations at that scale can be tricky and subjective and should be handled with care. 10. 4 Reyedrré Need? 10.4.1 Atomistic Simulations, Transient and Non-linear Properties: Atomistic sirmrlations (molecular dynamics) can be used to understand the phenomena for which experimental data is not available. For instance, a study on the interaction of polymer chains of petro-resin and bio-resin, bio-resin and bio-resin, and their interaction with nanoclay will shed light on the polymer blends and influence better chemistry and processing of the materials. Such lmowledge will allow better design of these materials and incorporation of more bio-resin. Additionally, the use of robust multi-scale computations should be used to understand the influence of lower scale heterogeneities on macroscoPic time-dependent, transient properties such as therrml and banier properties. 339 Similarly it is essential to model the progression of damage of these composites. It is believed that damage initiates at the lower/ atomic scales and proper use of robust multiscale computational schemes will help understand failure initiation and better design and development of complex materials. 10.4.2 Statistical Considerations in Experimental Results As expected, the experimental data revealed scatter/ variations in measured parameters. These variations were specifically larger for failure-dependent parameters, such as tensile failure strains, tensile strengths and impact strengths. In this work, only the average values were considered and hence a qualitative effect and overall trends based on average values could be obtained. Detailed statistical analyses taking into account these variations should be performed to quantitatively and precisely obtain the effects of constituents, namely bio-resin and nanoclay. 10.4.3 True Bio-degradable “Green” Materials The work reported uses blends of petroleum resins and bio-resins along with natural fibers. The use of all bio-resins was limited due to performance concems. Recent advancements in the areas of bio-polymers and material science have shown promise in use of 100% bio-resins that are also bio-degradable, with improved performance. Also, the natural fibers used in this work were untreated. This was done so that a lower limit on performance could be obtained. For instance, a combination of “engineered” (or treated) fibers with 100% bio-resin and use of bio-based nano-particulates (such as cellulose whiskers, clay platelets, silica) would produce efficient “green” rmterials. Due to the rapid 340 advancements in material science and chemistry (functionalizations), high-performance and “green” rrnterhl are becoming a reality. 10.4.4 True Integration of Experiments and Computational Simulations The computational materials based approach of developing materials requires proper integration of experiments and computational simulations. This indicates that computational models at each scale should essentiallybe compared/ matched with experiments at respective scale. In such a case, material properties for computational models at nano-scale or any lower scale require experiments at that scale. Bulk rmterial properties cannot be used for local/ lower scale models. For instance, nncroscopic properties of a polymer nanocomposite cannot be used in a computational model that studies the effect of nanoparticle. Instead experiments such as nano-indentation or atomic force microscopy (AFM) will produce more realistic local properties. Additionally, it is believed that AFM allows observation of soft and hard regions within a polymer. Ideally, such observations could enable obtaining the actual bio-resin distribution, thereby enabling better modeling of three-phase nanocomposites. Also, such observations could be used to validate models, similar to the ones developed in this study that determine the bio-resin layout using topology optimization. Overall, true integration of experiments and simulation requires measurements and sirmrlations at respective scales. 10.4.5 Structural Application of Biocomposites and Large-scale Testing The bio-based materials used in this work show great promise in use for structural applications. One of the nnin concerns of use of these materials for structural applications has been durability. The addition of nanoclay in the resin system delays the moisture reaching the natural fibers and thereby is believed to improve the durability of resulting 341 structural component. Also, their load-bearing efficiency can be improved if they are used in sandwich/hollow-core designs. Finally, large scale manufacturing and testing of such biocomposite rmterials is essential to prove their feasibility. 10. I A’emmré Impact The approach attempted in this work, namely the integration of experiments and computational simulations has promise in the understanding of micro/nanoscale mechanisrm governing various parameters, thereby enabling tailoring of rmcroscopic properties by proper synergy and distribution of materials at respective scales. This methodology is not limited to hybrid bio-based materials used in this work, but could be applied for other nanoparticle types and hierarchical materials. Ideally, computational models validated with selective experiments would allow prediction of properties for rmterial compositions to which experimental data is not available. In the cunent study, experimentally validated computational models were developed at nanoscale, and were used to predict the overall macroscopic response. At the same time, in a true sense, these validated models were not implemented in detemrining properties for unlmown material compositions. It is believed that such predictions are direct extension to the current work. Ideally, such predictions would eliminate costly trial and error experiments and enable better understanding and consequently help in design of complex materials. Overall, a successful modeling scheme, as the one used in this research could enable development of novel, efficient, cost effective multifunctional materials. This technique is powerful and can be used in diverse applications such as: a) modeling of bones and development of prosthetics, b) design of - textiles, c) development of alloys and materials from a combination of heterogeneities etc. On the other hand, development of bio-based materials and their 342 applications for structural applications would lead to sustainable, environmental friendly, bio-degradable and cost effective rrraterials. 343