IMPROVING MULTI FUNCTIONAL PROPERTIES IN POLYMER BASED NANO COMPOSITES BY INTERFACIAL ENHANCEMENT

Transcription

1 IMPROVING MULTI FUNCTIONAL PROPERTIES IN POLYMER BASED NANO COMPOSITES BY INTERFACIAL ENHANCEMENT A Dissertation Presented By Navid Tajaddod to The Department of Mechanical and Industrial Engineering in partial fulfillment of the requirements for the degree of Doctor of Philosophy In the field of Mechanical Engineering (Concentration: Materials Science) Northeastern University Boston, Massachusetts July, 2017

2

3

4 ACKNOWLEDGEMENTS Firstly, I would like to express my sincere gratitude to my advisor Professor Marilyn Minus for the continuous support of my Ph.D. study and related research, for his patience, motivation, and immense knowledge. Her guidance helped me in all the time of research and writing of this thesis. I could not have imagined having a better advisor and mentor for my Ph.D. study. Besides my advisor, I would like to thank the rest of my thesis committee: Professor Yiannis Levendis, Professor Sunho Choi, and Professor Hongli Zhu, for their insightful comments and encouragement, but also for the questions which incented me to widen my research from various perspectives. Thirdly, I also want to thank my labmates, Dr. Yiying Zhang, Dr. Kenan Song, Dr. Jiangsha Meng, Dr. Emily Green, Mr. Heng Li, Ms. Dilinazi Aishanjiang, and Ms. Aishah Ayaz, for their generous help and support on experiments, data analysis, modeling, as well as personal life. Finally, I would like to thank my family: my parents and to my sisters for supporting me spiritually throughout my life in general. Last but not least, my deepest gratitude goes to my wife Negar, for her support, encouragement, quiet patience, believing in my potentials and supporting me to become a better human being every day. iii

5 TABLE OF CONTENTS ACKNOWLEDGEMENTS LIST OF TABLES LIST OF FIGURES LIST OF SYMBOLS AND ABBREVIATIONS SUMMARY iii vii viii xiv xvii CHAPTER 1. Literature Review Interfacial and Interphase Structures in Polymer Nanocomposites (PNC) Polymer Nanocomposites (PNCs) The Interface and Interfacial Region The Importance of Interphase in PNCs Interphase Formation by Polymer Crystallization and Templated Orientation Polymer Crystallization Carbon Nanotube (CNT) Induced Polymer Crystallization and Interphase Boron Nitride (BN) Induced Polymer Crystallization and Interphase Polyacrylonitrile (PAN)/CNT Composite PAN Heat Treatment of PAN-Based Carbon Fibers PAN/CNT Composites as a Precursor for Carbon Fiber and Film Ultra-high Molecular Weight Polyethylene (UHMWPE)/BN Composite UHMWPE Thermal Conductivity of Polymer/BN Composites Thermal Conductivity of Polyethylene (PE)/BN 29 PART ONE 31 CHAPTER 2. Multi-functional Properties of PE/BN Composite by Interfacial Enhancement Introduction Experimental Section Materials Preparation of BN Platelets Fiber Formation Characterization Results and Discussion Matrix Structural Development Influencing Filler Exfoliation Polymer Orientation and Fibrillar Development Mechanical Characterization Thermal Conductivity Characterizations 47 iv

6 2.4 Conclusion 49 CHAPTER 3. A Study on Polyethylene and Polyethylene/Boron Nitride Gel Processing and Subsequent Spinning For Continuos Fiber Introduction Experimental Section Materials Preparation of BN Particles Formation of Gel Preparation of a Well-Dispressed PE-BN Gel Transferring Gel to the Spinning Syringe Proper Gel Spinning Parameters 59 CHAPTER 4. Gel Spinning of the PE Fibers Introduction Experimental Section Materials Preparation of BN Particles Gel Formation Fiber Spinning and Drawing Characterization Result and Discussion Morphological Observation of the Fiber and the Structural Development Characterization of BN Particles Polymer Orientation and Fibrillar Development Mechanical Characterization SAXS Characterization Thermal Analysis by DSC Thermal Conductivity Conclusion 87 PART TWO 89 CHAPTER 5. Low-Temperature Graphitic Formation Promoted by Confined Interphase Structures in Polyacrylonitrile/Carbon Nanotube Materials Introduction Experimental Section Materials Solution Processing and Film Fabrication Results and Discussion Precursor Films Film Stabilization Film Carbonization Graphitization Treatment Raman Spectroscopy Conclusion 117 v

7 CHAPTER 6. Effect of Low-Temperature Graphitized Structure on Electrical Properties of PAN/CNT Materials Introduction Experimental Section Materials Solution Processing and Film Fabrication Results and Discussion Structural Morphology of the Films WAXD Analysis of the Films Electrical Conductivity of the Films Raman Analysis of the Films EDX Analysis of the Films Conclusion 139 CHAPTER 7. Summary and Recommendation for Future Works Summaries Recommendations 141 APPENDIX A. Sample Prepration For Thermal Conductivity Mesurments 142 APPENDIX B. Electrical Conductivity of Carbon Films for Longer Carbonization Time 143 B.1 Introduction 143 B.2 Experimental Section 143 B.2.1 Materials 143 B.2.2 Solution Processing and Fiber Fabrication 143 B.3 Results and Discussion 145 B.3.1 Electrical Conductivity of the Films 145 B.3.2 WAXD Study of the Films 146 APPENDIX C. Fabrication of Carbon Fibers From Enhanced Interfacial Region Precursor 148 C.1 Introduction 148 C.2 Experimental Section 148 C.2.1 Materials 148 C.2.2 Solution Processing and Fiber Fabrication 148 C.2 Results and Discussion 151 REFERENCES 154 vi

8 LIST OF TABLES Table 1.1 Thermal properties of BNNs. 28 Table 1.2 Several BN/polymer composites and their room temperature thermal properties. 30 Table 2.1 Experimental mechanical properties for all fabricated fibers. 45 Table 4.1 Crystal structure properties of BN and nbn nano-platelets based on WAXD study Table 4.2 Herman s orientation factor (f) calculated using Wilchinsky s method, crystallinity calculated from WAXD (i.e., Xc,WAXD) results, DSC data (i.e., ΔHfiber, T1 and T2), and crystallinity calculated from DSC results (i.e., Xc,DSC) Table 4.3 Experimental and theoretical Young s modulus of the fibers. 82 Table 5.1 Effect of carbonization time and temperature on (002) plane crystal structure development in carbonized films. Table 5.2 Higher order WAXD peaks associated with the hexagonal and rhombohedral forms of layered graphite Table 6.1 Weight loss percentage of films after each heat treatment. 127 Table 6.2 The 2θ and (002) d-spacing values of carbonized films. 131 Table 6.3 Raman data of D-band and G-band of precursor and carbonized film at 1100 C for 5 min. 138 Table 6.4 EDX analysis of films. 139 Table B.1 2θ and (002) d-spacing values of carbonized films at 1100 C for 5, 20, 40, and 60 min. Table C.1 2θ and d-spacing information of the carbonized Control-PAN, Composite-1, Composite-2, GT, IM7, and T300 carbon fiber vii

9 LIST OF FIGURES Figure 1.1 Illustration showing the interface and interphase (i.e., interfacial region) areas within a PNC. Figure 1.2 Effect of (a) interface thickness (t) and (b) interface electrical conductivity (σs) on the electrical conductivity of CNT composites (σe). (f: percolation threshold) 41. Figure 1.3 A schematic of polymer (a) lamella (i.e., folded-chain), and (b) fibrillar (i.e., extended-chain) crystals Figure 1.4 Structure of (a) SWNT and (b) MWNT. 12 Figure 1.5 Scanning electron microscopy (SEM) image of nano hybrid shish kebab structure in a PE/CNT system. (Arrows shows the CNT) 12. Figure 1.6 SEM images of different PAN interphase structures around CNT. (a) Tubular coating, (b) Hybrid shish-kebab, (c) Solvent rich (blobs) 18. Figure 1.7 Atomic structures of (a) layered structure of hexagonal boron nitride (hbn), (b) a single BNNS and (c) a multi-wall BNNT. Figure 1.8 SEM images of (a) undrawn PE/BN nano-platelet composite fiber showed shish-kebab like structure (yellow arrows) as well as the region of PE crystallizing in the vicinity of BN nano-platelet (red dotted region). Full-atomistic molecular MD simulations snapshots of (b1) PE chains crystallized only on BN nano-platelet basal plane, and (b2) PE molecules crystallized around the BN nano-platelet 105. Figure 1.9 (a) Molecular structure of PAN, (b) PAN ladder structure 123, and (c) reactions occurring during the stabilization process 136. Figure 1.10 Structural changes/evolution by cross-linking during carbonization 142. Figure 1.11 WAXD curves for carbonized and graphitized PAN/SWNT film samples 145. Figure 1.12 HR-TEM images at low and high magnifications are provided for two different PAN/SWNT composites: (a1, a2, a3, and a4) PAN/SWNT-1 and (b1 and b2) PAN/ SWNT Figure 1.13 Orthorhombic unit cell crystal structure of PE viii

10 Figure 1.14 Effect of drawing on increasing the crystallinity, orientation, and interactions of chains in UHMWPE. Schematics of (a) undrawn UHMWPE and (b) drawn UHMWPE morphologies. Figure 2.1 SEM images of (a1) BN-1 platelets with the average diameter of ± 90.9 nm and (a2) BN-2 platelets with the average diameter of ± nm. Dashed circle and arrow show the typical measurement taken for each platelet. The averages and standard deviations for both BN batches are calculated based on at least 100 measurements. (b) WAXD intensity-versus-2θ profiles of BN-1 and BN-2 platelet samples and (c) Raman spectra of the processed BN-1 and BN-2, as well the as-received BN. Figure 2.2 (a) Graphical representation of the procedure for preparation of the UHMWPE and UHMWPE/BN fibers, (b) solution of control UHMWPE in xylene at 140 C and (c) solution of UHMWPE in xylene at 78 C (i.e., arrows point to regions where transparency decreases with temperature). Figure 2.3 SEM images of fibers microstructure. (a1) shish-kebab structure in undrawn UHMWPE fiber, and (a2) fibrillar/extended-chain (shish) structure in drawn UHMWPE fiber. (b1) Shish-kebab structure in undrawn UHMWPE/BN-1 fiber, (b2) fibrillar/extended-chain (shish) structure in drawn UHMWPE/BN-1 fiber, and (b3) magnified the image of the boxed region in b1 shows the UHMWPE lamellae crystals formed on BN-1 platelets (see arrows). (c1) Shish-kebab structure in undrawn UHMWPE/BN-2 fiber, (c2) fibrillar/extendedchain structure in drawn UHMWPE/BN-2 fiber, and (c3) magnified the image of the boxed region in b2 shows the UHMWPE lamellae crystals formed on BN-2 platelets (see arrows). Figure 2.4 Raman spectra of all fibers. The BN peak at 1368 cm -1 diminishes after hot-drawing for both composite fibers (i.e., UHMWPE/BN-1 and UHMWPE/BN-2) as shown in boxed region. Figure 2.5 2D-WAXD patterns of (a1) undrawn and (a2) drawn UHMWPE fibers, and (a3) the corresponding intensity-versus-2θ profiles. 2D- WAXD patterns of (b1) undrawn and (b2) drawn UHMWPE/BN-1 fibers, and (b3) the corresponding intensity-versus-2θ profiles. 2D- WAXD patterns of (c1) undrawn and (c2) drawn UHMWPE/BN-2 fibers, and (c3) intensity-versus-2θ profiles. Figure 2.6 The typical stress-strain curves for the (a) undrawn fibers (inset shows magnification for the low-strain region) and (b) drawn fibers. Figure 2.7 SEM images of the fiber cross-section. (a) prepared by polishing, (b) prepared by glass blade microtoming ix

11 Figure 3.1 Schematic representation of the different flask, heating, and stirring set-ups used during trials to make a PE and PE-BN gel. Figure 3.2 Homogenous clear gel made from PE (bubbles seen in the gel were removed after several hours) Figure 3.3 A well-dispersed milky colored PE-BN (10 wt% BN) gel in the round bottom flask Figure 3.4 A schematic illustrating the overall gel spinning processing method. 61 Figure 4.1 SEM images of control PE fibers microstructure. (a) shish-kebab structure in undrawn PE fiber, and (b) fibrillar/extended-chain (shish) structure in drawn PE fiber. Figure 4.2 SEM images of composite fibers microstructure. (a, b) shish-kebab structure in undrawn PE/5BN fiber, and (c) shish-kebab structure in undrawn PE/10BN fiber, (e, f) fibrillar/extended-chain (shish) structure in drawn PE/10BN fiber, (g) fracture region of PE/10BN after mechanical test, (h, i) aggregation regions of BN in PE/10BN fiber. Figure 4.3 SEM images of PE/1nBN fibers microstructure. (a) shish-kebab structure in undrawn PE/1nBN fiber (arrows show loose nbns), (b) aggregation of nbn particles in undrawn fibers, (c) fracture region after the the mechanical test (arrow show loose nbn particle in the structure), and (d) fibrillar/extended-chain (shish) structure in drawn fiber (arrows show nbn particles formed at the surface region of the fiber). Figure 4.4 FT-IR spectra of all the fibers. Two peaks related to BN particles at and cm -1 are observed for composite fibers. Figure 4.5 WAXD graph of BN and nbn particles. nbn shows higher intensity for (100) plane due to the exfoliations. Figure 4.6 (a) 2D-WAXD patterns of fibers, and (b) the corresponding intensity-versus-2θ profiles of the fibers. Figure 4.7 WAXD azimuthal intensity scans of undrawn (U) and drawn (D) (a) PE, (b) PE/1BN, (c) PE/5BN, (d) PE/10BN and (e) PE/1nBN fibers for the (110) reflection peaks. Figure 4.8 Mechanical properties of the fibers. (a) Young s modulus and (b) tensile strength of the fibers Figure 4.9 2D-SAXS patterns of all the fibers. 83 x

12 Figure D-SAXS intensity-versus-q curves for (a) undrawn and (b) drawn fiber samples. Figure 4.11 DSC graphs of (a) 1 st heating cycle and (b) 2 nd heating cycle of the fibers. Figure 5.1 Graphical representation for the heat treatment procedures used in this work for (a) stabilization, (b) carbonization, and (c) graphitization of the samples. Figure 5.2 SEM images of cross-sections of fabricated films: (a1) PT-1 film, (a2) zoom in area of the boxed region in a1 image; (b1) PT-2 film, (b2) zoom in area of boxed region shows CR-L, (b3) zoom in area of boxed region shows PR-L; (c1) SW-1 film, (c2) zoom in area of boxed region in c1 image, (d1) SW-2 film, (d2) zoom in area of boxed region shows CR-L, (d3) zoom in area of boxed region shows PR-L. Figure 5.3 WAXD spectra of the PT-BP, PT-1, PT-2, SW-BP, SW-1 and SW- 2 films. PT-1 and SW-1 films show a broader PAN (110) peak, PT-2 and SW-2 materials exhibit a sharper crystalline PAN (110) peak. Several broad peaks pertaining to the SWNT are also observed (see arrows). Figure 5.4 Magnified SEM images for a two-layered hpbp showing the (a) CR- L and (b) PR-L, regions. Figure 5.5 Cross-section schematic for the fabricated films. The interphase region is assumed to be larger around smaller diameter SWNT bundles due to prior studies 18, 220. Figure 5.6 DSC thermographs of PT-1 and PT-2 films stabilized at the heating rates of 1 C min -1. Figure 5.7 WAXD spectra for the stabilized PT-1, PT-2, SW-1 and SW-2 films. A broad peak between 2θ from 23 to 25 is related to PAN ladder structure 145. Figure 5.8 SEM images of cross-sections for (a1) stabilized PT-1 film, (a2) zoom in of boxed region in the PT-1 film; (b1) PT-2 film, (b2) zoom in of boxed region in the PR-L area in the PT-2 film, (b3) zoom in of boxed region in the CR-L area in the PT-2 film; (c1) stabilized SW-1 film, (c2) zoom in of boxed region in the SW-1 film, (d1) SW-2 film, (d2) zoom in of boxed region in the PR-L area in the SW-2 film, (d3) zoom in of boxed region in the CR-L area in the SW-2 film. Figure 5.9 WAXD spectra for the carbonized films (a) at temperatures ranging from 900 to 1100 C and (b) at 1500 C. (002) graphitic peak occurs at 2θ of ~26. AC is the amorphous carbon peak. Peaks 1,2,3,4 and xi

13 are related to higher order graphitic planes (100) of hexagonal form with ABAB stacking, (101) of hexagonal form with ACBACB stacking or (100) of rhombohedral form, (101) of hexagonal form, (102) of hexagonal form with ACBACB stacking or (110) of rhombohedral form and (102) of hexagonal form with ABAB stacking or (103) of hexagonal form with ACBACB stacking, respectively 145. Figure 5.10 SEM of carbonized films. (a) PT-1 and (b1) PT-2 films carbonized at 1000 C for 5 min, (b2) and (b3) zoom in of boxed region in (b1); (c) PT-1 and (d1) PT-2 films carbonized at 1100 C for 5 min, (d2) and (d3) zoom in of boxed region in (d1); (e) SW-1 and (f1) SW-2 films carbonized at 1100 C for 5 min, (e2) and (e3) zoom in of boxed region in (e1); and (g) PT-1 and (h1) PT-2 films carbonized at 1500 C for 5 min, (h2) and (h3) zoom in of boxed region in (b1). Figure 5.11 WAXD curves for the PT-1, PT-2, SW-1 and SW-2 films graphitized at 2100 C. Figure 5.12 SEM image of graphitized (a1) PT-1 film (previously carbonized at 1100 ºC for 5 min), (a2) zoom in of boxed area in a1; (b1) PT-2 film (previously carbonized at 1100 ºC for 5 min), (b2) zoom in of boxed region in CR-L area in b1, (b3) zoom in of boxed region in PR-L area in b1, (c1) PT-2 film (previously carbonized at 1000 ºC for 5 min), (c2) zoom in of boxed region in CR-L area in c1, (c3) zoom in of boxed region in PR-L area in c1; (d1) SW-2 film (previously carbonized at 1100 ºC for 5 min), (d2) zoom in of boxed region in CR-L area in d1, (d3) zoom in of boxed region in PR-L area in d1. Figure 5.13 The Raman spectra of PT-2 films carbonized at 1100 C for 5 min and 900 C for 20 min. Figure 6.1 Graphical representation for the heat-treatment procedures used in this work for (a) stabilization and (b) carbonization of the samples. Figure 6.2 SEM images of the precursor, carbonized at 900 C for 20 min, and carbonized at 1100 C for 5 min:(a) PT cross-section, (b) PT surface region, (c) PT-1 cross-section, (d) PT-1 surface region, (e) PT-2 crosssection and, (f) PT-2 surface region. Figure 6.3 WAXD spectra of the PT, PT-1, and PT-2 (a) precursor and (b) stabilized films Figure 6.4 WAXD spectra of the PT, PT-1, and PT-2 carbonized films. 130 Figure 6.5 Electrical conductivity data of the precursor and stabilized films. 132 Figure 6.6 Electrical conductivity data of carbonized films. 135 xii

14 Figure 6.7 Raman spectra of precursor and carbonized film at 1100 C for 5 min. (a) RBM and (b) D-band and G-band Figure A.1 Cross-sections of (a) undrawn PE and (b) drawn PE fibers in epoxy resin (arrows shows the fibers). Figure B.1 Electrical conductivity of the carbonized film at 1100 C and different time (i.e., 5, 20, 40 and 60 min). Figure B.2 WAXD spectra for the PT-2 films carbonized for 5, 20, 40, and 60 min. Figure C.1 A schematic of the holder that made to allow fibers hold the weight during the heat treatment procedures. Figure C.2 - WAXD spectra of the (a) Control-PAN, (b) Composite-1 and (c) Composite-2 carbon fibers. Figure C.3 2D-WAXD patterns of the (a) Control-PAN, (b) Composite-1, and (c) Composite-2 carbon fibers xiii

15 LIST OF SYMBOLS AND ABBREVIATIONS PNC polymer nanocomposite PE polyethylene BN boron nitride PAN polyacrylonitrile CNT carbon nanotube BNNT boron nitride nanotube BNNS boron nitride nano sheet IFSS interfacial shear strength PVA polyvinyl alcohol PBT poly(butylene terephthalate) ipp isotactic polypropylene PLLA poly(l-lactide) PCL poly(e-caprolactone) PE-b-PEO polyethylene-b-poly(ethylene oxide) BP buckypaper LCP liquid crystalline polymer SWNT single-wall carbon nanotube MWNT multi-wall carbon nanotube CVD chemical vapor deposition SEM scanning electron microscopy BNNs boron nitride nano structures B boron xiv

16 N nitrogen hbn hexagonal boron nitride MD molecular dynamics PVF polyvinyl formal PMMA poly(methyl methacrylate) PS polystyrene PVB poly(vinyl butyral) PEVA poly(ethylene vinyl alcohol) PBA 1-pyrenebutyric acid AA acrylic acid MAA methacrylic acid IA itaconic acid AM acrylamide MA methacrylate VGCNF vapor grown carbon nanofiber TEM transition electron microscopy HR-TEM high-resolution transition electron microscopy WAXD wide-angle x-ray diffraction UHMWPE ultra-high molecular weight polyethylene XRD X-ray diffraction DSC differential scanning calorimeter TDTR time-domain thermoreflectance Al aluminum FT-IR Fourier transform infrared spectroscopy ATR attenuated total reflectance xv

17 SAXS Small-Angle X-ray Scattering DMA dynamic mechanical analyzer FWHM full widths at half maximum hpbp hybrid polymer/cnt buckypaper DMF dimethylformamide S solvent NS non-solvent EDX energy-dispersive X-ray RBM radial breathing mode xvi

18 SUMMARY Polymer nanocomposites (PNCs) have become an area of increasing interest for study in the field of polymer science and technology since the rise of nanotechnology research. Despite the significant amount of progress being made towards producing high quality PNC materials, improvement in the mechanical, electrical, thermal and other functional properties still remain a challenge. To date, these properties are only a fraction of the expected theoretical values predicted for these materials. Development of interfacial regions between the filler and matrix within the composite has been found to be an important focus in terms of processing. Proper interfacial control and development may ensure excellent interaction and property transfer between the filler and polymer matrix in addition to improvement of multi-functional properties of PNCs. The property-structure importance for the existence of the interfacial and interphase region within PNCs is discussed in this thesis work. Two specific PNC systems are selected for study as part of this dissertation in order to understand the effect of interfacial region development on influencing multi-functional property trends. Polyethylene (PE)/boron nitride (BN) and polyacrylonitrile (PAN)/carbon nanotube (CNT) composites were selected to investigate their mechanical performance and thermal and electrical conductivity properties, respectively. For these systems it was found that the interfacial region structure is directly related to the enhancement of the subsequent multi-functional properties. xvii

19 xviii

20 CHAPTER 1. LITERATURE REVIEW 1.1 Interfacial and Interphase Structures in Polymer Nanocomposites (PNC) Polymer Nanocomposites (PNCs) Polymer nanocomposites (PNCs) have become an area of increasing interest for study in the field of polymer science and technology since the rise of nanotechnology research. For PNC materials, fillers with at least one dimension less than 100 nm are dispersed within the polymer matrix. These fillers may have different shapes, sizes, and properties. The decrease in the scale of fillers from micro to nano leads to two major factors (i) higher surface area per volume and (ii) smaller size leading to issues in the dispersion. Despite the significant amount of progress being made towards producing high-quality PNC materials, improvement in the mechanical, electrical, thermal and other functional properties still remain a challenge. To date, the measured properties of PNCs are only a fraction of the expected theoretical values predicted for these materials. One region of interest in PNC materials is the interfacial regime. This area is important toward ensuring proper interaction and property transfer between the filler and polymer matrix. The importance for the existence of this region in PNCs is discussed in subsequent sections The Interface and Interfacial Region The interface region is the two-dimensional boundary between the filler and the matrix. The interfacial region (i.e., interphase) is the inhomogeneous three-dimensional zones between the filler and polymer phase. In this region, the matrix properties are typically found to not match those of the bulk polymer phase in the composite 1. The 1

21 interface and interfacial regions (i.e., interphase) of polymer matrix composites are depicted in Figure 1.1. It has been shown that final characteristics of PNCs materials, such as mechanical, electrical, thermal, optical and magnetic properties, are related to matrix, filler and interface/interphase interaction between matrix and filler 2. At the interfacial region, the matrix can be influenced most directly by the filler features. For this reason, the interface/interphase region plays an important role on the final properties of PNCs 2. Figure 1.1 Illustration showing the interface and interphase (i.e., interfacial region) areas within a PNC. Processing conditions, as well as matrix and filler materials properties, affect the interface interactions and structures 1. Changing these parameters can be important for tuning the strength of the interaction between the bulk matrix, interphase region, and the filler. For example investigations of different systems, have shown at least three major interfacial bonding types between the matrix and fillers. These types include mechanical, 2

22 physical and chemical bonding 1. Mechanical bonding includes interlocking or gripping interactions between materials. Physical interaction involves low energy secondary, van der Waals, dipolar, or hydrogen bonding (i.e., 8-16 kj mol -1 ) between materials. Chemical bonding includes bonds with higher energy (i.e., ~ kj mol -1 ). This happens when there is a reaction at the interface and includes covalent, ionic, and metallic bonding The Importance of Interphase in PNCs Interfacial interactions between the polymer matrix and nano fillers increase due to the large surface area afforded by the small scale of the inclusion. Nano fillers such as carbon nanotube (CNT), graphene, boron nitride nanotube (BNNT) and boron nitride nano sheet (BNNS) all exhibit high aspect ratio and large surface area 3. Therefore, the interphase region around nano fillers tends to be predominated in the PNC and can be interconnected 4. Fundamental studies toward understanding the interphase regions in PNCs are still being developed. However, it is believed that the processing conditions and subsequent quality of interphase region have a great impact on mechanical, electrical, and thermal properties in PNCs 2. Effects of interphase on these properties are explained in the following subsections Effect of Interphase on the Mechanical Properties of PNCs Large aspect ratio, good dispersion, alignment and interfacial stress transfer are requirements for filler mechanical reinforcement in PNCs and govern the effective transfer of the external stress to the nano fillers. Equation (1.1) provides a theoretical relationship to analyze the effective load on the filler and matrix in the PNCs. 3

23 F f = E f V f F m E m 1 V (1.1) f F f, F m are loads carried by filler and matrix, respectively. E f and E m are Young s moduli of filler and matrix, respectively. V f is the volume fraction of filler. Because E f is much larger than E m, the load carried by filler is typically much higher than the matrix 5. In practical applications, strain within the matrix is higher than the filler 5. For this reason, a shear stress field around the filler will develop and the value of this field is maximized in the vicinity of the inclusion and decreases at distances farther away 6. The stress felt by the filler typically increases linearly with external stress on the PNC. At a critical shear stress value debonding occurs between the bulk matrix and filler or matrix-filler interphase region 6. This critical value is referred to as the interfacial shear strength (IFSS). Suggested values of IFSS for non-covalently bonded CNT-polymer composites is in the range from 50 to 100 MPa 5. Higher values such as 500 MPa 7 and 350 MPa 8 are reported for covalently bonded composites. Improvement in the IFSS properties has been found for composites exhibiting ordered or crystalline interphase formation around nano fillers 9. This ordered or crystalline interphase structure in PNCs is also found to be mechanically stronger than the amorphous structure because of the presence of fewer defects/disordered regions 10. Crystalline interphase formation due to nucleation and template growth of several polymeric matrices such as PE 11-16, nylon 6,6 13, polyvinyl alcohol (PVA) 17, poly acrylonitrile (PAN) 18, poly(butylene terephthalate) (PBT) 19-21, isotactic polypropylene (ipp) 22, poly(l-lactide) (PLLA) 22, poly(e-caprolactone) (PCL) 23, and polyethylene-b-poly(ethylene oxide) (PE- 4

24 b-peo) block copolymer 24 on CNT have been reported. The crystalline morphology makeup of the polymer has also been investigated for several of these ordered interphase regions It is believed that templated ordered crystal growth of a polymer on CNT nuclei in the PNC helps to improve the stress transfer of between matrix and filler 25, 29, 30. Therefore, for mechanical enhancement and achievement of complete stress transfer between the polymer matrix and nano filler, controlling the interphase development is critical. Fundamental understanding of interphase formation is still being developed. However, techniques like shear or flow crystallization have showed promise in terms of the formation of ordered interphase growth, and such processing techniques may even be incorporated in fabrication procedures for PNCs Effect of Interphase on the Electrical Properties of PNCs Electrically conductive PNCs have been investigated for potential application in several fields such as electrostatic material, electromagnetic shielding 32, artificial nerves 33, sensors 34, organic solar cells 35, light emitting diodes 36, bio resistive coating 37, and selective membranes 38. Electrical conductivity in PNCs is based on percolated pathways formed by conductive filler inclusions. According to percolation theory 39, conductive pathways form after a critical filler concentration (i.e., percolation threshold). The percolation threshold in the range from 0.24 to 1.35 vol% was calculated for rod-like filler with aspect ratio higher than Previous work has shown a model for effective electrical conductivity in CNT composites, where this property is dependent on CNT length, diameter, concentration, and 5

25 also interfacial regions 41. This work shows that both interfacial thickness and interfacial electrical conductivity have an effect on the electrical conductivity of the composite (Figure 1.2). According to this model increasing the interfacial layer thickness decreases the percolation threshold 41. In addition, larger interfacial layer thickness within a given threshold region also enhances the effective electrical conductivity of CNT composites 41. This model also shows that before the percolation threshold, effective electrical conductivity increases very slowly with interfacial thickness. However, after the percolation threshold, the effective electrical conductivity increases rapidly 41. Figure 1.2 Effect of (a) interface thickness (t) and (b) interface electrical conductivity (σs) on the electrical conductivity of CNT composites (σe). (f: percolation threshold) 41. The interfacial region also impacts the electrical properties in PNCs. For example, specific polymer interfaces that trap more carriers are assumed to reduce dielectric life times in materials 42. On the other hand improving interfacial interactions is necessary for high permittivity in PNCs with dielectric fillers 43. For PNCs with conductive fillers, interfacial regions with good insulating characteristics are required for achieving high 6

26 permittivity 43. In addition, CNT induced polymer chain alignment in CNT buckypaper (BP)/liquid crystalline polymer (LCP) composites has shown to improve both mechanical and electrical properties of the LCP matrix 44. CNT sheets or BPs are free standing films of entangled CNT 45 with variable electrical 46, field emission 47, and mechanical properties 46. It has been shown that nitric acid treatment of BPs helps to improve the film tensile strength from 10 to 74 MPa and Young s modulus from 0.8 to 5.0 GPa, while in-plane electrical conductivity decreased from to S m -1 due to degradation of the tubes 46. It can be concluded that to achieve good mechanical properties and maintain good electrical conductivity interfacial interaction plays a crucial role in the BPs. It has also been shown that surfactant and functionalizing groups may also negatively impact electrical and thermal properties 48. More recent results in this dissertation work have shown that polymer crystallization can be utilized to replace the use of surfactant and functionalizing groups on CNT, and subsequently improve the functional properties of polymer composites. For this reason, polymer crystallization in BP/polymers composites may improve the overall functional properties of PNCs Effect of Interphase on the Thermal Properties of PNCs Typically thermal conductivity of polymer/cnt materials is usually less than 1 W m -1 K -1. However, simple calculations using rule-of-mixture analysis and assuming the thermal conductivity of the CNT to be ~1000 W m -1 K -1 and the polymer matrix to be ~0.3 W m -1 K -1, results in the value of tens or even hundreds of W m -1 K -1 depending on filler loading are obtained. Several parameters including interfacial resistance, nano filler 7

27 distribution, dispersion and alignment 49 are found to play a role on the resultant thermal conductivity properties of PNCs. The electrical percolation threshold for CNT or graphene filled polymer composites is very low (i.e., 0.1 % volume fraction ). For this reason, at 1% volume fraction (typically used for PNCs), an electrically conductive network of CNT already exists 53. Therefore, heat resistance at the CNT and polymer interphase region is assumed as a major factor for differences between experimental and theoretical thermal conductivity data 54. The thermal resistance at interfacial boundary layers is called the Kapitza resistance 55. This resistance is proportional to temperature discontinuity across the interface and is defined by Equation (1.2). R K = T Q (1.2) R K is interfacial resistance, T is temperature discontinuity, and Q is power per unit area flowing across the interface 55. Interfacial thermal resistance can arise from two sources. First, from acoustic mismatch as a result of lack of existence of common vibration frequencies between the polymer matrix and nano filler 54. Second, from differences between surface wettability and imperfect physical contact between the polymer matrix and nano filler 54. Several attempts have been made to improve the interfacial heat transfer. Matching surface chemistry according to wetting properties of the interface 56 is one of these attempts. Chemical functionalization is also shown to decrease the interfacial resistance between polymer and CNT 57. However, chemical functionalization can reduce intrinsic 8

28 conductivity properties of CNT 57. Interfacial polymer crystallization can also be an alternative for improving interfacial heat transfer. In polymers, heat transfer cannot be dominated by free electrons. Therefore heat is conducted by phonons via transmission of vibrational or rotational energy of molecules. Phonon transfer is more efficient in a highly crystalline polymer phase. Therefore, highly crystalline polymers, where molecules are closer and more compact, show better heat transfer properties. For example, highly crystalline PE can have high thermal conductivity 58, 59. Theoretical studies predict thermal conductivity of ~350 W m -1 K along individual PE chains. Defects such as polymer chain ends, entanglement, voids, and impurities which can scatter phonons are reduced in the crystalline phase. For these reasons, the crystalline interphase around the nano filler is a promising way for transferring heat in PNCs. 1.2 Interphase Formation by Polymer Crystallization and Templated Orientation Polymer Crystallization Polymer crystals can form from melt, solution or during polymerization 61. Crystallization can affect mechanical, thermal, optical and chemical properties of polymers. Polymers with less chain regularity such as branching points or bulky side groups have shown lower degrees of crystallinity. Polymer crystallization begins with nucleation and formation of nuclei or crystal embryo, once the formed nuclei reach a critical size, crystal growth occurs 61, 62. Nucleation and growth can happen by rearrangement of chains into the highly periodic 3-D structure by either lamella (foldedchain) or fibrillar (extended-chain) geometries (Figure 1.3). Lamella crystals grow perpendicular to chain axis direction. In contrary, fibrillar crystals chains are aligned 9

29 parallel to the chain direction. To form nuclei in fibrillar crystallization, entangled and coiled chains must disentangle and undergo transitional diffusion 63. External forces like flow fields can accelerate the process of nucleation and growth for fibrillar crystals. High modulus and strength fibers have been made by introducing flow-fields in polymer solutions Study of these fibers has shown that the chain molecules are extended along the fiber axis. Several commercial fibers such as PE, nylon 6, nylon 6,6, PP, and polyethylene terephthalate have been made by fibrillar extended chain crystallization. Well-known shish-kebab structures are also made by fibrillar crystallization processes, where the extended chain crystal acts as the shish Controlled conditions of nucleation for fibrillar structures are still a challenge for a wide range of polymers. CNT with nano size dimensions has been shown as a promising nucleation site for the formation of fibrillar structures in polymers. Figure 1.3 A schematic of polymer (a) lamella (i.e., folded-chain), and (b) fibrillar (i.e., extended-chain) crystals. 10

30 1.2.2 Carbon Nanotube (CNT) Induced Polymer Crystallization and Interphase CNTs have exceptional electrical, thermal, optical, and transport properties 71. Single-wall CNT (SWNT) counterparts were identified by Iijima 72 and Bethune 73 in SWNTs are composed of carbon atoms and are like rolled graphitic sheet into a seamless tube with diameter s between 0.4 to 4 nm 74 (Figure 1.4a). Multi-wall CNTs (MWNTs) (Figure 1.4b) consist of multiple rolled layers (concentric tubes) of graphene with a diameter between 5 and 100 nm and a 0.34 nm spacing between layers 74. CNT length can be in the range of nanometers to millimeters. Therefore, they exhibit high aspect ratio due to a small diameter and a long length. Properties of CNTs are affected by their diameter and chirality. CNTs exhibit a thermal conductivity of up to 6000 W m -1 K -1 at room temperature 75. Large current densities of up to 100 MA cm -2 have also been predicted 76. Chemical vapor deposition (CVD) have the most common synthetic method for CNT growth. However, other techniques like arc-discharge 71, 72, 80 and laser ablation 81, 82 are used. 11

31 Figure 1.4 Structure of (a) SWNT and (b) MWNT. Large surface areas, long lengths, nanoscale diameters, and high aspect ratios have allowed CNT to act as promising crystal nucleating agents for polymers. As mentioned earlier, nucleation and templated growth of several polymeric matrixes such as PE 11-16, nylon 6,6 13, PVA 17, PAN 18, PBT 19-21, ipp 22, PLLA 22, PCL 23, and PE-b-PEO 24 on CNT have been reported. CNT can also aid alignment of polymer chains along their axis. PVA/SWNT fibers made under shear-flow conditions have been shown to exhibit templated interfacial polymer coating on SWNT increasing the mechanical properties of the composites 10. Tensile strength and Young s modulus as high as 4.9 GPa and 128 GPa were found, respectively 10. The CNT-induced hybrid shish-kebab structure of PE (Figure 1.5) and nylon-6,6 were made by solution crystallization 12, 14, 15, 83. By changing the crystallization conditions in PAN/CNT system, the different interphase regions were observed (i) tubular coating, (ii) hybrid shish-kebab and (iii) solvent rich blobs (Figure 1.6) 18. These studies all show the importance of polymer crystallization conditions on tailoring and fabricating different polymer interfacial regions for crystalline matrices. 12

32 Figure 1.5 Scanning electron microscopy (SEM) image of nano hybrid shish kebab structure in a PE/CNT system. (Arrows shows the CNT) 12. Figure 1.6 SEM images of different PAN interphase structures around CNT. (a) Tubular coating, (b) Hybrid shish-kebab, (c) Solvent rich (blobs) Boron Nitride (BN) Induced Polymer Crystallization and Interphase In recent years, boron nitride nano structures (BNNs) including boron nitride nanosheet (BNNS) 84 and BNNT 85 materials have been studied significantly, since they 13

33 are structural analogs of graphene 86 and CNT 71, respectively. In these materials, the carbon atoms are alternately substituted with either boron (B) or nitrogen (N) atoms (Figure 1.7). Due to these configurational changes, there are also significant differences in terms of the band gap structure, which makes these BNNs materials exhibit semiconducting or insulating properties (i.e., ranging from ev) 3, 87. BNNs are also white and their color. This may be useful as potential transparency and composite color can be an important factor in specific application areas. In addition, BNNs also exhibit excellent oxidative 88 and chemical stability 88, as well as mechanical 89, thermal 90, ultraviolet 91, and lubricating 92 properties. Another attractive feature of these boron-based materials is that they exhibit excellent thermal conductivity along the basal plane structure of the hexagonal lattice 93. Figure 1.7 Atomic structures of (a) layered structure of hexagonal boron nitride (hbn), (b) a single BNNS and (c) a multi-wall BNNT. As mentioned, polymer nucleation and templated crystallization on various graphitic substrates with different geometries, including flat surfaces of graphite 94-96, graphene 97, or graphene oxide nanosheets 11, 98, curved surfaces of carbon fibers , and highly curved surfaces of CNT materials 11, 25, 26, 28, 103, 104 has been observed. In addition to carbon and CNT structures, BN and BNNs can also act as nucleation sites for polymer 14

34 crystallization. Strong interphase interaction has been seen in flow induce crystallization of PE on BN nano-platelets. Full-atomistic molecular dynamics (MD) simulations have shown that large crystallization interaction areas exist for the PE chains on the basal plane or around the BN nano-platelets. This strong interaction can also help to exfoliate the BN nano-platelet upon drawing of fibers (Figure 1.8), where good interfacial interaction is promoted 105. However, studies of BN composites systems are still limited Beyond BN platelets, polymer nucleation and templated crystallization on BNNs is also an important area of study in order to improve the transfer of properties to the polymer matrix. BN and BNNs may have important contributions for both mechanical as well as thermal enhancement in PNCs. Figure 1.8 SEM images of (a) undrawn PE/BN nano-platelet composite fiber showed shish-kebab like structure (yellow arrows) as well as the region of PE crystallizing in the vicinity of BN nano-platelet (red dotted region). Full-atomistic molecular MD simulations snapshots of (b1) PE chains crystallized only on BN nano-platelet basal plane, and (b2) PE molecules crystallized around the BN nano-platelet

35 This dissertation is focused on the use of PNCs for multi-functional applications. To this end, the enhancement of electrical and thermal conductivity properties in PNCs will be studied from a fundamental point of view. As discussed, the inclusion of filler in PNCs, as well as interfacial region formation, is important toward tailoring and improving these properties. For this reason, two systems are chosen for the proposed effort. (1) PAN/CNT composites for subsequent carbon-carbon materials development/processing to examine structural effects on electrical properties; and (2) PE/BN composites to investigate similar effects on thermal properties in PNCs. The following sections provide a summary of preliminary studies that have been performed and pertain to these two systems. In addition, some of the structural advantages of these systems are also discussed to show the fundamental reasons for using these PNCs to investigate electrical and thermal enhancement for the systems. 1.3 Polyacrylonitrile (PAN)/CNT Composite PAN PAN with the linear formula (C3H3N)n (Figure 1.9a) is a synthetic thermoplastic, semi-crystalline organic polymer. About ~90% of carbon fiber production line has PAN as a precursor material, due to its relatively high properties and high carbon yield As compared to pitch-based carbon fibers, PAN-based carbon fibers 113, 114 have higher tensile and compressive strengths 115. The other carbon fiber precursor materials include cellulose and other polymers, such as polyester, polyamides, PE, PVA, poly(phenylene), and phenolic resin 111, 116. However, the final carbon content is too low for commercial viability at the current stage. 16

36 Copolymers of acrylonitrile with one or more comonomers are typically used as precursor PAN fibers for carbon fiber production. These copolymers act as structural inhomogeneities in order to increase precursor fiber drawability and to shorten the stabilization time 111, Typically, PAN copolymers contain 2% to 15% of acidic comonomers 111, 116, 117, such as acrylic acid (AA), methacrylic acid (MAA), itaconic acid (IA), acrylamide (AM) and methacrylate (MA) Heat Treatment of PAN-Based Carbon Fibers In order to achieve carbon fibers with proper properties (e.g., high strength and high modulus), understanding and utilizing proper heat treatment conditions is necessary. In general, three steps are required for conversion of PAN into the carbon fiber. (1) stabilization, (2) carbonization and (3) graphitization 121. Graphitization is the optional step used to increase structural perfection as well as modulus properties. In the following subsections, the importance of the carbon fiber processing steps will be explained in detail Stabilization Precursor stabilization is a crucial step toward producing stable chemical structures from PAN precursors. At this stage, molecules of PAN are converted to a thermally stable ladder structure (Figure 1.9b) 122 by crosslinking between the molecules 123 to avoid melting at high temperatures. Several chemical reactions happen during stabilization processes including cyclization, dehydrogenation, aromatization, oxidation, and crosslinking 124. CH2 and C N groups also disappear and C=C, C=N, and =C H groups forms

37 Stabilization temperatures are usually in the range of C (preferably at or below 270 C) 122, Stabilization at low temperature is slow and incomplete but at high temperature may result in overheating or degrading of the molecules. The optimum heating rate for stabilization is below 5 C min -1 (i.e., 1-3 C min -1 ) for avoiding shrinkage due to shortening of the PAN molecules 117, 134. Stabilization takes place under tension in the inert or oxidative atmosphere. The processes of dehydrogenation, cyclization, and oxidation are illustrated in Figure 1.9c. An oxidizing atmosphere is preferred because of higher carbon yield and better mechanical properties present in the final carbonized structure. Oxygen increases the activation energy but also acts as an initiator of activated centers for cyclization process. The density of the precursor increases during oxidative stabilization due to 8-12 wt% oxygen uptake 135. Oxidation is a diffusion controlled process and it takes several hours to be complete. However, for PAN copolymers containing 2% MA, this process only takes ~25 min 136. During dehydrogenation double bonds forms after elimination of water from oxidized PAN precursors 120. Dehydrogenation can happen before or after cyclization. The double bond formed during dehydrogenation is required in order to increase thermal stability and reduce chain scission during carbonization. Oxygen is required for dehydrogenation and this reaction does not happen in an inert atmosphere 134. The most important step in stabilization is the cyclization process. At this step, stable ladder polymers formed by the reaction of the nitrile groups in the precursor with adjacent groups 120 and triple bonds (C N) convert to double bonds (C=N) 137. The cyclization reaction can happen in the inert or oxidative atmosphere since oxygen is not involved in this process Carbonization 18

38 The next step for fabrication of carbon fibers is carbonization (Figure 1.10). Stabilized fibers are heated in an inert atmosphere 139, 140 from C 117 or even as high as 3000 C 129. Inert gases used are typically nitrogen or argon 139, 140. However carbonized fibers in nitrogen atmospheres have shown noticeable improvement in tensile and modulus 141. At early stages of carbonization (i.e., below 600 C) thermal pyrolysis 133 and intermolecular crosslinking takes place 117. Typically the heating rate is low (i.e., 5 C min -1 ) 133 and It allows for a low rate release of volatiles to minimize pores and surface irregularities 117. Higher heating rates are possible at a temperature ranging from C. At this stage, cyclized sections coalesce by cross-linking. Typically carbonization procedure time is on the order of an hour. Therefore the processing time is minimized (i.e., min) at a temperature more than 1000 C 117. Electrical conductivity and thermal conductivity can be as high as 50,000 S m -1 and 10 W m -1 K -1, respectively

39 Figure 1.9 (a) Molecular structure of PAN, (b) PAN ladder structure 123, and (c) reactions occurring during the stabilization process 136. Figure 1.10 Structural changes/evolution by cross-linking during carbonization

40 Graphitization Graphitization is an expensive and optional step after carbonization. The temperature range is from 1500 to 3000 C 117, 143 and this process happens in the inert atmosphere. For nitrogen environment graphitization occurs below 2000 C to avoid reaction of carbon with nitrogen. The graphitization process is on the order of minutes but low cooling rates are preferred. Graphitization increases the crystal size as well as the preferred orientation of graphitic structure and subsequent planes 117. After graphitization, 99 % of the precursor is converted to carbon structure 144. Electrical conductivity and thermal conductivity can be as high as 100,000 S m -1 and 50 W m -1 K -1, respectively for graphitized carbon fibers PAN/CNT Composites as a Precursor for Carbon Fiber and Film As mentioned, PAN is used as the dominant precursor material for carbon fiber production. Carbon fibers have a high fabrication cost. However, their properties are also lower than predicted values. For these reasons, fabrication of PAN/CNT carbon fibers or films seems to be promising 145. Recently several researchers have been demonstrated the fabrication of PAN/CNT nanocomposites fibers and films 26, , which show better mechanical, electrical and thermal properties in comparison with similar PAN fibers or films. Heat treated PAN/CNT structure was also shown to exhibit better properties as compared to control-pan materials. These preliminary studies show that PAN/CNT materials can be used as precursors for carbon fibers or films. In one study, carbon fiber modulus and strength improved by 49% and 64%, respectively, due to the addition of 1 wt% CNT to the PAN matrix 27. In another study, tensile strength and modulus of PAN/CNT improved by 138% by addition of the same amount of CNT to PAN 152. Carbon 21

41 films with a modulus of ~30 GPa were fabricated by heat treatment of PAN/MWNT and vapor grown carbon nanofiber (VGCNF) 153, 154. Stabilized and carbonized PAN/SWNT have shown specific capacitance as high as 250 F g Despite the improvement in mechanical or electrical properties, the addition of CNT influences structural changes within PAN during stabilization and/or carbonization. These structural changes include (i) conversion of PAN into a highly ordered ladder structure, (ii) better chain alignment and orientation, (iii) fewer chain terminations, (iv) longer conjugated segment at CNT interphase, and (v) formation of less β-amino nitrile groups formation 27, 156. PAN/CNT interphase regions also have shown the formation of the graphitic structure upon carbonization at an even low temperature (i.e., 1100 C) 27, 145, 149, This graphitic structure was investigated by Raman spectroscopy, X-ray spectroscopy and transition electron microscopy (TEM). Raman results show an increase in G-band peaks 27, wideangle X-ray diffraction (WAXD) results show broad carbonaceous peaks at ~26.0 (Figure 1.11) 27, 145 and high-resolution transition electron microscopy (HR-TEM) has been used to image the graphitic structure at the PAN/CNT interphase regions (Figure 1.12)

42 Figure 1.11 WAXD curves for carbonized and graphitized PAN/SWNT film samples 145. Interphase formation in PAN/CNT composites can decrease the onset of graphite phase formation in PAN matrix. PAN interphase regions can also affect electrical and mechanical properties of PAN/CNT composite as a precursor for carbon/carbon composite. Additional preliminary studies have done on the effect of this interphase feature on structural changes and electrical conductivity and are discussed in Chapter 3. 23

43 Figure 1.12 HR-TEM images at low and high magnifications are provided for two different PAN/SWNT composites: (a1, a2, a3, and a4) PAN/SWNT-1 and (b1 and b2) PAN/ SWNT Ultra-high Molecular Weight Polyethylene (UHMWPE)/BN Composite UHMWPE UHMWPE is PE - the simplest and most widely used thermoplastic polymer consisting of long hydrocarbon chains with a chemical formula of (C2H4)n and a molecular weight between two and six million. The most common crystal structure of UHMWPE is orthorhombic and its unit cell is shown in Figure Metastable monoclinic and 24

44 hexagonal crystal structures are also possible for UHMWPE. UHMWPE materials are tough and show excellent wear resistance properties. UHMWPE contains large molecules (i.e., chain), which interact together with weak van der Waals bonds. However increasing interaction along the length of these long chains increases the amount of effective secondary bonds and the inter-molecular strength. By stretching or drawing fibers made from UHMWPE, these weak secondary bonds can break and this allows the chains to further align in a parallel orientation toward obtaining a high level of crystallinity and molecular interaction (Figure 1.14). Figure 1.13 Orthorhombic unit cell crystal structure of PE

45 Figure 1.14 Effect of drawing on increasing the crystallinity, orientation, and interactions of chains in UHMWPE. Schematics of (a) undrawn UHMWPE and (b) drawn UHMWPE morphologies. Gel spinning solutions of UHMWPE at high temperatures coupled with further hot drawing is used for fabrication of highly oriented commercial fibers 161. These fibers consist of highly oriented molecules and exhibit a high level of crystallinity. For these reasons, gel-spun UHMWPE fibers show high mechanical strength and thermal conductivity 162, 163. Young s modulus and tensile strength of commercial gel-spun UHMWPE are ~113 GPa and ~3.4 GPa for Dyneema and GPa and 2.5 to 3.6 GPa for Spectra 164 fibers. Heat transfer in PE like many other polymers cannot be dominated by free electrons. Therefore, heat is conducted by phonons via transmission of vibrational or rotational energy of molecules. Phonon transfer is more efficient in the highly crystalline polymer, 26

46 where molecules are closer and more compact. Semi-crystalline PE illustrates low level of thermal conductivity but ordered highly crystalline PE can have high thermal conductivity 58, 59. Defects such as polymer chain end, entanglements, voids, and impurities can scatter phonons. Therefore, commercial micron size diameter (10 to 25 mm) fibers show lower thermal conductivity ranging from W m -1 K Decreasing the diameter also reduces the defects in the unit length of fiber reducing photon scattering. Recently nano size diameter (i.e., 131 nm) fiber with high thermal conductivity of 104 W m -1 K -1 has been reported 168. It is important to note that most polymer fiber exhibit typical thermal conductivity values below 1 W m -1 K -1. The high values reported for PE are related to the ability to control fiber crystallinity and orientation Thermal Properties of BNNs Thermal conductivity (κ) is a physical property of the materials ability to conduct heat. The differential form for thermal conductivity is described by Fourier s law for heat conduction, shown in Equation (1.3). q = κ T (1.3) q is the local heat flux density (i.e., the amount of energy that flows through a unit area per unit time (W m -2 )), κ is the thermal conductivity (W m -1 K -1 ) and T is the temperature gradient (K m -1 ). The negative sign shows that heat flows from a high to low temperature. In solids, heat transfers by phonons (atomic vibration) and electrons. But in non-metallic materials free movement of electrons is limited and heat transfer is dominated by phonons. 27

47 Several theoretical and experimental studies have been performed for measuring thermal properties of BN structures (Table 1.1). The in-plane thermal conductivity can be thousands of W m -1 K -1, while along other lattice planes only up to hundreds of W m -1 K - 1. Theoretical studies show that thermal conductivity of BNNS is simulated to be ~400 W m -1 K using molecular dynamics, and W m -1 K , 171 using the nonequilibrium Green s function method. BNNT are predicted up to 6000 W m -1 K , 173 but experimentally estimated to vary from 120 to 960 W m -1 K , 175. Experimental thermal conductivity results for hexagonal BN (hbn) along a basal plane at 290 K vary from 200 to 500 W m -1 K , 177. The thermal conductivity of 11-layer hbn was found to be 360 W m -1 K and of 40 nm diameter BNNT to be ~200 W m -1 K It is also reported that smaller diameter nanotubes show higher thermal conductivity 179. Thermal conductivity values of hbn and BNNs are in a wide range but it is high enough to use in the applications that require high thermal conductivity but electrically insulating conditions. Table 1.1 Thermal properties of BNNs. BNNT BNNS Thermal Conductivity (W m -1 K -1 ) (Theoretical) (Experimental) (Theoretical) 40 (Experimental) Thermal Stability Up to C Up to C Thermal Conductivity of Polymer/BN Composites As mentioned earlier in-plane thermal conductivity of BN is higher than along other crystal planes. Thus BNNs can have advantages due to higher thermal conductivity. For this reason, BNNs are assumed to be great candidates for thermally conductive and electrically insulating polymer composites. Several polymer matrix composite materials 28

48 have been made so far using hbn or BNNs as fillers. A list of these composite with their measured thermal conductivity, percentage and type of fillers are shown in Table Thermal Conductivity of Polyethylene (PE)/BN As mentioned earlier, the thermal conductivity of many polymeric materials and their composite are still below their predicted values. Both UHMWPE and BN are shown promising thermal transfer properties. Several factors affecting thermal properties of polymer composite including interfacial resistance, filler dispersion, and alignment. The large resistance to the heat transfer at the polymer and fillers boundary is responsible for the low experimental thermal conductivity properties exhibited by BN polymer composites. Highly crystalline structures of UHMWPE and also the good interfacial interaction between UWMWPE and BN can enhance the interfacial heat transfer between polymeric matric and the fillers in UHMWPE/BN composites. BNNs show higher thermal conductivity and the larger surface area in comparison to micron size BN. Incorporating BNNs in UHMWPE composite may positively affect the thermal properties of such composites. Preliminary results in the MINUS lab show that PE can form crystalline interphase on BN. Increasing ordered structure within PE while incorporating well dispersed BN as inclusions is a promising way to enhance thermal conductivity properties of PE fiber. Preliminary work done to achieve aforementioned structural condition and its effect on the thermal and mechanical properties are discussed in Chapter 2. 29

49 Table 1.2 Several BN/polymer composites and their room temperature thermal properties. Polymer BN type BN level (wt%) Thermal conductivity of pure polymer (W m -1 K -1 ) Thermal conductivity of composite (W m -1 K -1 ) Surface modification polyvinyl formal BNNT catechin 180 (PVF) PVA BNNT 1-3 ~ catechin 180 poly(methyl methacrylate) BNNT ethanol 181 (PMMA) polystyrene (PS) BNNT ethanol 181 poly(vinyl butyral) BNNT ethanol 181 (PVB) poly(ethylene vinyl alcohol) BNNT ethanol 181 (PEVA) epoxy BNNS 50 ~30 isopropyl alcohol 182 PVA BNNS 50 ~10 isopropyl alcohol 182 cellulose BNNS 5-50 ~ epoxy BNNTs BNNSs ~ pyrenebutyric acid (PBA) Ref

50 PART ONE PE/BN Composite Fibers 31

51 CHAPTER 2. MULTI-FUNCTIONAL PROPERTIES OF PE/BN COMPOSITE BY INTERFACIAL ENHANCEMENT 2.1 Introduction As mentioned in Chapter 1, due to their multi-functional properties, BN materials are potentially excellent candidates as fillers in polymer-based composites 180, 181. The use of BN materials in various polymer matrices toward mechanical enhancement has been previously studied For example, composites such as PVA/BN and polybenzimidazole/bn 186 show 40% and 27% increases in tensile modulus, respectively. However, other composites such as epoxy/bn 187, poly(methyl methacrylate)/bn 188 demonstrate lower levels of reinforcement. Ineffective reinforcement efficiency observed in polymer/bn composites has been related to poor stress transfer between the components due to weak interfacial interactions as well as poor filler dispersion and orientation 189. For this reason development of processing routes for BN/polymer composites that aim to aid the development of interfacial structure is important. In this Chapter, work pertaining to the use of a flow-crystallization procedure to fabricate UHMWPE/BN composite fibers is outlined. Based on previous studies, it has been shown that inducing polymer crystallization in the presence of nano-fillers can enhance polymer alignment and interfacial interactions within the resultant composite material 12, 18. For example, a similar solution-based flow-crystallization process was used previously to induce interfacial crystal growth of PE during composite fiber fabrication utilizing nano-carbons 12, 190. Based on these previous studies, it was shown that nano- 32

52 carbons can affect PE interfacial crystallization, and in doing so increase the polymer orientation as well as the PE/CNT interaction. This dissertation work also explores the potential of using this crystallization process to improve polymer-bn interactions within the composite fiber, which may play a role in the reinforcement efficiency. To understand the process-structure-property relationships in the fibers, the final composites were characterized using microscopy, spectroscopy, as well as thermal and mechanical analyses tools. The effects of crystalline morphology, orientation factor, the percentage of crystallinity, and lamellae thickness distribution were also investigated for their contribution to the composite properties. 2.2 Experimental Section Materials Hexagonal BN platelets (CAS , ~1 μm, 98%, ρ = 2.29 g cm -3 ) were purchased from Sigma-Aldrich. UHMWPE (Mw ~ g mol -1, ρ = g cm -3 ) was obtained from DSM Dyneema. Ethanol (CAS , Mw = g mol -1, 99.5%) and xylene (CAS , Mw = g mol -1, ρ = 0.86 g cm -3 ) solvents were obtained from Sigma-Aldrich and used as-received Preparation of BN Platelets The as-received BN platelets were dispersed via different routes in order to produce two BN filler batches (i.e., BN-1 and BN-2). Batch 1: BN-1 was prepared by sonicating 1500 mg of the BN platelets in 50 ml of ethanol (Fisher FS30 bath sonicator, frequency 43 khz; power, 150 W) for 24 h. The sonicated material was subsequently centrifuged (Sorvall 33

53 Legend Micro 21 Microcentrifuges, Ventilated, 120V 60Hz) at 3000 rpm for 90 min. To remove the ethanol the supernatant (~15 to 40 mg) of the centrifuged dispersion was filtered, collected, and dried at room temperature (25 C) for 24 h. The average platelet diameter for the BN-1 batch was measured to be ± 90.9 nm (Figure 2.1a1). Considering that BN-1 preparation utilized both sonication and centrifugation, a large initial weight of the BN particles was needed to yield a sufficient sample in the supernatant (i.e., ~15 to 40 mg per batch) for spinning. For this reason, BN-2, which is only treated by sonication does not require the same starting weight of BN particles. Instead, a weight more comparable to the final yield in BN-1 is chosen. Batch 2: BN-2 was prepared by sonicating 48 mg of BN platelet in 50 ml ethanol for 24 h. No centrifugation was used in the preparation of BN-2. The sonicated material was collected and dried using a similar procedure to that of BN-1. The average platelet diameter for the BN-2 batch was measured to be ± nm (Figure 2.1a2). It was estimated that due to the preparation differences for the BN batches, the use of centrifugation would lead to a more uniform distribution of platelet size, and this is reflected based on analysis for the average platelet diameter. For composites prepared using both BN batches, the platelets were subsequently added to xylene (concentration of 0.96 mg ml -1 ) and introduced to the crystallizing polymer solutions (see Section for a solution and composite preparation). As discussed, in general, BN-1 batches exhibit a smaller average diameter and a more uniform size distribution, as compared to BN-2 batches. However, both batches show similar platelet thickness (i.e., stack-size of BN sheets). Figure 2.1b shows the WAXD intensity (I(θ)) versus Bragg angle (2θ) curves for both BN-1 and BN-2 batches. Sharp diffraction 34

54 peaks at 2θ = 26.7 corresponding to the (002) inter-layer spacing between the BN sheets 191 are observed for both batches. The stack size (measured by WAXD) for the BN platelets in both BN-1 and BN-2 samples were 18.8 ± 0.05 nm and 18.2 ± 0.06 nm, respectively. Figure 2.1c provides the Raman spectra for as-received BN as well as processed BN-1 and BN-2 batches. All samples exhibit a peak at 1368 cm -1 corresponding to the characteristic E2g phonon mode for BN 192. Figure 2.1 SEM images of (a1) BN-1 platelets with the average diameter of ± 90.9 nm and (a2) BN-2 platelets with the average diameter of ± nm. Dashed circle and arrow show the typical measurement taken for each platelet. The averages and standard deviations for both BN batches are calculated based on at least 100 measurements. (b) WAXD intensity-versus-2θ profiles of BN-1 and BN-2 platelet samples and (c) Raman spectra of the processed BN-1 and BN-2, as well the asreceived BN. 35

55 2.2.3 Fiber Formation UHMWPE and UHMWPE/BN fibers were fabricated using a similar flowcrystallization approach reported previously 12. This method requires that the fibers form in a flowing dilute polymer solution at the crystallization temperature. The polymer dissolution and crystallization procedure are graphically outlined in Figure 2.2a, and described by processing steps (i) to (v). (i) UHMWPE is dissolved in xylene at a polymer concentration of 0.11 mg ml -1 at ~140 C using an overhead mechanical stirrer (Eurostar power-b IKA WERKE stirrer) equipped with a cylindrical stir bar at a speed of 650 RPM (Figure 2.2b). (ii) The polymer solution temperature is reduced to ~95 C while stirring is continued for injection of the BN platelet batches. This solution injection temperature was chosen based on several trials investigating temperatures between 90 to 110 C. These injection conditions were found most appropriate to yield consistent fibers during collection. (iii) A glass syringe equipped with a 22-gauge blunt tip needle was prepared with 0.8 ml of the freshly sonicated BN-1 or BN-2 dispersions (see Section 2.2.2) and introduced to the stirred polymer solutions at a rate of 2 ml min-1. (iv) Stirring is stopped at ~85 C for UHMWPE/BN-2, ~82 C for UHMWPE/BN-1, and ~78 C for control UHMWPE fibers, respectively, to aid crystallization and overall fiber formation. The temperatures used in each system were based on an abrupt observation of the change in solution transparency (i.e., transition from a transparent to cloudy appearance, shown in Figure 2.2c). (v) The final fibers were collected from the cylindrical stir bar after the solution is cooled to room temperature (~25 C). The resultant composite fibers contain a solid concentration of ~11 wt% BN to the UHMWPE solid content. Fibers were 36

56 subsequently hot-drawn at 130 C. The average draw ratio for composite fibers and control UHMWPE fibers was ~4 and ~12, respectively. Prior to characterization, the spun fibers were dried at room temperature (~25 C) in a vacuum oven for 24 hours. Figure 2.2 (a) Graphical representation of the procedure for preparation of the UHMWPE and UHMWPE/BN fibers, (b) solution of control UHMWPE in xylene at 140 C and (c) solution of UHMWPE in xylene at 78 C (i.e., arrows point to regions where transparency decreases with temperature) Characterization A Zeiss Supra 25 field emission SEM (operating voltage 5 kv) was used for image analysis of the fibers and BN platelets. All samples were coated using a Gatan high- 37

57 resolution ion beam coater with a thin chromium layer (15-20 nm) for imaging. WAXD was performed on a Rigaku RAPID II curved detector X-ray diffraction (XRD) system equipped with a 3 kw sealed tube source (voltage 40 kv and current 30 ma). WAXD curve fitting and analysis was performed using software s PDXL 2 (version ) and 2DP (version ). Raman spectroscopy was carried out using a Jobin Yvon LabRam HR800 equipped with a laser with a wavelength of 532 nm. Thermal tests were performed using a differential scanning calorimeter (DSC) (Q200, manufactured by TA Instruments), at a heating rate of 10 C min -1 from 25 to 300 C in nitrogen. Tensile tests were conducted using a dynamic mechanical analyzer (DMA) (RSA-G2 series, manufactured by TA Instruments) with the gauge length of 10 mm and the extension rate of 0.05 mm min -1. The number of tested samples ranged from 5 to 8 and the diameter of samples was calculated by a weight method (Equation (2.1)), where D, m, l, and ρ are the diameter (mm), mass (mg), length (mm) and density (g cm -3 ) of fibers, respectively. The density of fibers was calculated using Equation (2.2), where ρ BN, ρ UHMWPE, wt% BN, and wt% BUHMWPEN are densities and weight fractions of BN and UHMWPE, respectively. Densities of control UHMWPE and composite fibers are determined to be g cm -3 and g cm -3, respectively. This approach assumes that the fibers exhibit a circular cross-section. D = 2 m/lπρ (2.1) ρ c = ρ BN wt% BN + ρ PE wt% PE (2.2) Time-domain thermoreflectance (TDTR) was used to measure thermal conductivity. Fabricated fibers were embedded in epoxy resin. Cross-sections of the fibers were prepared for testing by polishing with alumina or by sectioning using microtoming. A nano-size 38

58 layer of aluminum (Al) was coated on the cross-sectioned area to perform the measurements. 2.3 Results and Discussion Matrix Structural Development Influencing Filler Exfoliation The microstructure for the fabricated fibers was investigated using SEM and representative images are shown in Figure 2.3. A well-known shish-kebab structure is observed for the undrawn UHMWPE, UHMWPE/BN-1, and UHMWPE/BN-2 fibers. This morphology is expected due to the processing approach used for fabrication of the fibers 12, 190, 193. For the undrawn composite fibers, the UHMWPE lamellae crystals were found to nucleate and grow on the BN platelets (see arrows in Figures 2.3b3 and 2.3c3). Comparatively, drawn UHMWPE, UHMWPE/BN-1, and UHMWPE/BN-2 fibers show the predominant fibrillar morphology due to the unfolding of UHMWPE chains from the kebab structures upon hot-drawing. In addition to chain extension, hot-drawing of the fibers also led to the exfoliation of the BN platelets due to the UHMWPE/BN interfacial interaction (i.e., lamellae growth on the platelet surface). The experimental analysis of this exfoliation process is discussed in the subsequent paragraphs. Prior computational analysis for this system suggests that increasing interfacial interaction between UHMWPE and BN platelets contributes to BN exfoliation 105. This work demonstrates experimentally that the undrawn composite fibers exhibit a sufficient amount of UHMWPE crystal lamellae nucleation and growth on BN particle surface. As mentioned, it has been shown by computational analysis that during hot-drawing the chain unfolding from these surfacegrown lamellae provides sufficient work to surpass the threshold of force for BN platelet 39

59 exfoliation 105. Structurally it was also observed that both undrawn and drawn composite fibers exhibit more voids than the control fibers. These voids may be due to the BN exfoliation process and can act as defects, resulting in a decrease in the fiber mechanical properties. Figure 2.3 SEM images of fibers microstructure. (a1) shish-kebab structure in undrawn UHMWPE fiber, and (a2) fibrillar/extended-chain (shish) structure in drawn UHMWPE fiber. (b1) Shish-kebab structure in undrawn UHMWPE/BN-1 fiber, (b2) fibrillar/extended-chain (shish) structure in drawn UHMWPE/BN-1 fiber, and (b3) magnified the image of the boxed region in b1 shows the UHMWPE lamellae crystals formed on BN-1 platelets (see arrows). (c1) Shish-kebab structure in undrawn UHMWPE/BN-2 fiber, (c2) fibrillar/extended-chain structure in drawn UHMWPE/BN-2 fiber, and (c3) magnified the image of the boxed region in b2 shows the UHMWPE lamellae crystals formed on BN-2 platelets (see arrows). Raman spectroscopy was used to characterize the BN platelets within the composite fibers (Figure 2.4). Hexagonal BN typically shows two Raman active mode peaks at approximately 51.8 and cm -1 representing the E2g symmetry vibrations 192. The low- 40

60 frequency peak (i.e., 51.6 cm -1 ) is due to sliding of whole planes with respect to one another, while the high-frequency peak (i.e., cm -1 ) is related to inter-sliding of the B and N atoms between the basal planes 192. It has also been observed that the intensity of higher frequency peak decreases with fewer platelet layers 194. This high-frequency peak is very predominant in the Raman spectra for the undrawn UHMWPE/BN-1 and UHMWPE/BN-2 composite fibers and observed at ~1368 cm -1. This peak intensity was observed to significantly diminish after hot-drawing for both the UHMWPE/BN-1 and UHMWPE/BN-2 composite fibers. The decrease in the relative intensity of BN Raman peak is a result of reducing the number of BN layers in the platelet stack, due to the exfoliation during hot-drawing of the composite fibers. These Raman results are also in accordance with WAXD data, which was used to analyze the change in BN platelet morphology within the fibers before and after hot-drawing (Figure 2.5). Both undrawn composite fibers show the (002) BN reflection (Figures 2.5b1 and 2.5c1) due to a predominant peak at 2θ ~ 26.7 (Figures 2.5b3 and 2.5c3) corresponding to the inter-layer spacing for the BN platelets. Comparatively, the drawn UHMWPE/BN-1 and UHMWPE/BN-2 fibers showed significant diminishment for the BN (002) reflection/peak intensity (Figures 2.5b2, 2.5b3, 2.5c2, and 2.5c3). This is another complementary indication of platelet exfoliation in both composite fibers upon hot-drawing. 41

61 Figure 2.4 Raman spectra of all fibers. The BN peak at 1368 cm -1 diminishes after hot-drawing for both composite fibers (i.e., UHMWPE/BN-1 and UHMWPE/BN-2) as shown in boxed region. 42

62 Figure 2.5 2D-WAXD patterns of (a1) undrawn and (a2) drawn UHMWPE fibers, and (a3) the corresponding intensity-versus-2θ profiles. 2D-WAXD patterns of (b1) undrawn and (b2) drawn UHMWPE/BN-1 fibers, and (b3) the corresponding intensity-versus-2θ profiles. 2D-WAXD patterns of (c1) undrawn and (c2) drawn UHMWPE/BN-2 fibers, and (c3) intensity-versus-2θ profiles. 43

63 Previous studies have shown that exfoliation of layered structures such as clay 195, CNC 196, as well as BN 105, 189 can be induced in various polymer matrices. Preliminary studies show that interfacial interactions between UHMWPE and BN can lead to almost complete exfoliation of the BN into individual sheets within the polymer matrix. The phenomenon observed in this work provides evidence for developing a processing approach to enhance the dispersion of layered-fillers in a polymer matrix. However, this requires excellent interfacial interaction between the polymer and filler. The use of shear crystallization promotes PE folded chain growth at the basal plane surface for BN Polymer Orientation and Fibrillar Development The WAXD data shown in Figure 2.5 also reveals micro- and nano-structural changes within the matrix for all fibers. The 2D-WAXD diffraction pattern of undrawn the UHMWPE fiber (Figure 2.5a1) shows full rings corresponding to the (110) and (200) reflections, which pertain to the PE crystal unit cell. Comparatively, the 2D-WAXD patterns of undrawn UHMWPE/BN-1 and UHMWPE/BN-2 fibers (Figures 2.5b1 and 2.5c1) exhibit asymmetric rings where higher intensity is observed along the equatorial direction. This indicates a more oriented chain structure with respect to the fiber axis is present in the as-spun composite fibers, and this may be influenced by the presence of BN platelets. Previous studies have also shown that the presence of rigid nano-fillers can help to induce chain alignment during flow-assisted (i.e., fibrillar crystallization) spinning processes 190. Post hot-drawing of all fibers led to improved chain orientation as evidenced by the transition in the diffraction patterns from rings to equatorial arcs (Figures 2.5a2, 2.5b2, and 2.5c2). This transition is more pronounced in the control fiber (Figure 2.5a2) as compared to composite (Figures 2.5b2 and 2.5c2), which is a result of the much lower 44

64 drawability of the samples. While the BN particles can provide a template for polymer orientation (i.e., Figures 2.5b1 and 2.5c1), the overall fiber uniformity is affected and drawing becomes more difficult with the high filler concentration. As mentioned, BN filler content is fairly high (i.e., 11 wt%). To this end, controlling filler distribution throughout the overall fiber is a challenge, and a lack of uniformity is observed for the composite samples (Figure 2.3), which will ultimately affect post-processing procedures like hotdrawing Mechanical Characterization Mechanical properties of all fibers are summarized in Table 2.1. The Young s moduli values of the undrawn UHMWPE fibers increased by 204% and 240% when loaded with BN-1 and BN-2, respectively. Hot-drawing of fibers increases Young s modulus of UHMWPE, UHMWPE/BN-1, and UHMWPE/BN-2 fibers by 1115%, 1031% and 1170%, respectively. The tensile strength values of the undrawn UHMWPE fibers increased by 371% and 632% with the addition of BN-1 and BN-2, respectively. In general, upon hotdrawing, the tensile strength was increased by 2584%, 479%, and 349% for control, UHMWPE/BN-1, and UHMWPE/BN-2 fibers, respectively. Table 2.1 Experimental mechanical properties for all fabricated fibers. Undrawn Drawn Sample E σ ε (GPa) (MPa) (%) UHMWPE 0.27 ± ± ± 94.1 UHMWPE/BN ± ± ± 60.5 UHMWPE/BN ± ± ± 35.9 UHMWPE 3.01± ± ± 17.4 UHMWPE/BN ± ± ± 3.2 UHMWPE/BN ± ± ±

65 Representative stress-strain curves for all fibers are shown in Figures 2.6a and 2.6b. The stress-strain curves for the undrawn fibers show the presence of work/strain hardening before fracture (Figure 2.6a). Drawn fibers show a curve more consistent with brittle fracture (Figure 2.6b). These curves provide some insight regarding the mechanical properties observed for each system. For the undrawn fibers, work-hardening is mostly observed in the composite as compared to the control fibers. This process is believed to be facilitated by the presence of the BN fillers. The BN-1 fibers show higher extension as compared to the BN-2 samples, which may be related to the ability of the polymer matrix to deform more readily in the presence of the smaller filler platelets. However, the degree of work-hardening is more pronounced in the BN-2 composites (i.e., larger average platelet size) leading to a higher fracture strength. The undrawn fibers also show a higher modulus as compared to the control fiber, which is related to the introduction of chain ordering and orientation in the composite fibers by the BN fillers (Figure 2.5). Figure 2.6 The typical stress-strain curves for the (a) undrawn fibers (inset shows magnification for the low-strain region) and (b) drawn fibers. The trend in mechanical behavior of the drawn fibers is related to the morphological changes occurring during the drawing process. In general, modulus values are found to be 46

66 larger for the composites (Table 2.1). The improvement in Young s modulus of drawn composite fibers with a draw ratio of 4 is larger than that of drawn control UHMWPE fiber with a draw ratio of 12. Orientation factor (f) determined from WAXD analysis is comparable for all fibers. Therefore, one possible reason for improvements in the measured Young s modulus may be indicated by the evidence for the exfoliation of BN platelets, which is due to strong filler-polymer interactions (i.e., exfoliation is driven by polymer matrix/chain sliding). This exfoliation process may also locally (i.e., at the filler-polymer interphase) regulate or confine the directional molecular alignment, which could also contribute to the increase of composite moduli. The BN-2 samples exhibit a higher modulus as compared to the BN-1 fibers. This may be related to the platelet size, which should provide more confinement to the polymers. The data also provide evidence that the exfoliation effect may be more pronounced in the BN-2 samples as compared to the BN-1 fibers. Comparison of the tensile strength measurements shows that the composites are on average lower than the control fibers (Table 2.1). However, in general, the strength measurements show a comparable range for all samples. Based on the aforementioned SEM analysis (Figure 2.3), the morphology of all fibers shows the presence of voids which are related to the spinning process used as well as filler presence. These defects will contribute to a decrease in the strength of the fibers. These voids are more pronounced in the composites, where additional contribution may occur due to the BN platelet exfoliation process Thermal Conductivity Characterizations TDTR was used to measure the thermal conductivity of PE composite fibers 167. Optical reflectivity of materials is known to vary with the change of the temperature. From 47

67 the interpretation of the reflectivity of materials, heat conductivity can be determined. In the TDTR method, once a material is heated up, the change in the reflectance of the surface can be utilized to derive the thermal properties. One advantage of this method is that it provides enough spatial resolution for thermal conductivity measurements of UHMWPE fibers. Sample preparation is of great importance to ensure proper measurement using this method. The surface roughness of the fiber cross-section should be minimized to 30 nm. To allow for measurement along the fiber length the samples are embedded and aligned in an epoxy mount using a paper holder. To measure thermal conductivity along fiber direction, fiber cross-sections were polished using an alumina powder (0.05 µm) to decrease the surface roughness. After polishing, an Al layer was sputter coated over the cross-sectional area (Figure 2.7a). Al is commonly used as the transducer layer for TDTR measurements due to its relatively high thermal conductivity, high thermoreflectance, large piezo-optical coefficients, and good adhesion to most materials. Initial results show that the measured thermal conductivity for undrawn PE/BN composite was ~11.9 W m -1 K -1. This initial value is very promising because it belongs to an undrawn fiber which does not show high orientation or improved crystalline structure. Furthermore, the cross-sectional roughness of this fiber is not at the required range for optical testing. To improve sample preparation a glass blade microtoming method was also used to decrease the sample roughness. SEM image of cross-section of this microtomed sample is shown in Figure 2.7b. The surface roughness appears to be smaller than that for the polished cross-section. However, further testing is required to quantify the roughness of this sample. Thermal conductivity for the PE and PE/BN samples is also ongoing, and careful attention is being given to the effect of structural make-up and its relationship to the fiber properties. 48

68 Figure 2.7 SEM images of the fiber cross-section. (a) prepared by polishing, (b) prepared by glass blade microtoming. Chapter 1 reflects the initial work performed to explore the potential for the use of BN as a filler material for UHMWPE. The use of fibrillar crystallization approaches as opposed to traditional fiber spinning routes was chosen here to explore whether BN fillers were a good candidate for templating polymer orientation and crystal nucleation. The results show that BN fillers are capable of influence the PE morphology. The UHMWPE and UHMWPE/BN fibers also show promising results in terms of mechanical and thermal conductivity properties. One drawback of this method is that continuous fibers may not be produced via fibrillar crystallization approaches. For this reason, the results from this study are used as a foundation to develop a route for continuous fiber production and to improve its properties. 2.4 Conclusion UHMWPE/BN composite fibers are fabricated using shear crystallization and subsequently drawn near the polymer melting point. Due to this processing approach, insitu exfoliation of BN in the polymer matrix occurred. Exfoliation of BN platelets is confirmed by WAXD and Raman spectroscopy analysis. SEM images illustrate that 49

69 crystallization of UHMWPE occurs on the surface of the BN platelets exhibiting good interfacial interaction. This polymer-bn interaction was found to contribute to exfoliation of BN platelets upon hot-drawing. These morphological changes resulted in improvement in mechanical properties for the composite fibers. Thermal property characterization for these samples is ongoing. 50

70 CHAPTER 3. A STUDY ON POLYETHYLENE AND POLYETHYLENE/BORON NITRIDE GEL PROCESSING AND SUBSEQUENT SPINNING FOR CONTINUOS FIBER 3.1 Introduction Chapter 2 provides some foundational knowledge that BN platelets are able to induce PE crystal growth and template polymer orientation. This previous work also shows that the interfacial interaction between PE and BN is strong and can lead to BN exfoliation and simultaneous polymer orientation during fiber hot-drawing. However, the previous method does not offer a processing route to allow for the continuous formation of fiber during spinning. To this end, this Chapter focuses on the developing a gel-spinning method for the polyethylene/bn fibers. Gel-spinning is a well-known fiber formation method for UHMWPE (i.e., Dyneema and Spectra). To utilize this approach in this work, the gel formation of UHMWPE containing BN must be tailored to allow for sufficient interaction between the filler and matrix during spinning. In addition, the introduction of filler interferes with the polymer-polymer chain junctions formed in the gel. Therefore, the polymer concentration and filler content must also be considered. Studies regarding the conditions for improving the composite gels and spinning process are outlined here. 51

71 3.2 Experimental Section Materials Hexagonal BN platelets (CAS , ~1 μm, 98%, ρ = 2.29 g cm -3 ) were purchased from Sigma-Aldrich. UHMWPE or PE (Mw ~ g mol -1, ρ = g cm - 3 ) was obtained from DSM Dyneema. 2-proponal (CAS , Mw = g mol -1, ρ = g cm -3 ), ethyl alcohol (CAS , Mw = g mol -1, ρ = g cm -3 ) and xylene (CAS , Mw = g mol -1, ρ = 0.86 g cm -3 ) solvents were obtained from Sigma-Aldrich and used as-received Preparation of BN Particles The as-received BN platelets were dispersed via different routes in order to produce two BN filler batches (i.e., BN and nbn). Batch 1: BN was prepared by sonicating BN platelets in xylene (Fisher FS30 bath sonicator, frequency 43 khz; power, 150 W) for 2 h. For preparing 1, 5, and 10 wt% composite fiber, 17, 89, and 187 mg of the BN platelets were sonicated in 100 ml xylene, respectively. This concentration is used to form 1, 5, and 10 wt% composite fibers subsequently named PE/1BN, PE/5BN, and PE/10BN. Batch 2: nbn was prepared by sonicating 1000 mg of BN platelet in 200 ml of 2-proponal for 24 h. The sonicated material was subsequently centrifuged (Sorvall Legend Micro 21 Microcentrifuges, Ventilated, 120V 60Hz) at 1000 rpm for 15 min. The supernatants were collected by using a pipette. Collected supernatant dried and sonicated in xylene for another 48 h. The prepared solution of nbn in xylene was stable at least for 24 h. For preparing 1 wt% composite fiber, 17 mg of the BN platelets in 100 ml xylene was used. The resultant fiber from this solution named PE/1nBN. Control fiber was named PE fiber. 52

72 3.2.3 Formation of Gel Schematics for the different routes that were tried for gel preparation are illustrated in Figure 3.1. In the first trial, PE dissolved in xylene in the flat bottom flask by using overhead stirrer. A hot plate was used to heat the solution of PE in xylene up to 138 C (Figure 3.1a). Different concentrations of PE in xylene were tried. By using an overhead stirrer, rod climbing was observed after gel formation. For this reason, the concentration of PE in xylene was decreased to minimize the effect of rod climbing. After finding the appropriate concentration to avoid rod climbing, the gel was formed for spinning. However, using this set-up, the lower region of the gel (which was close to hot plate surface) degraded (turning to yellowish) and stuck to the bottom of flask because that region was warmer than other portions of the flask. To solve this issue, a second set-up was considered where an oil bath was used for introducing uniform heat (shown in Figure 3.1b). By using the oil bath, different temperatures were used, as well as the different concentrations of PE in xylene. In this second set-up, degradation of PE at the bottom of the flask decreased but the issue did not fix completely. To reduce this lack of uniformity in the heating of the flask and subsequent degradation of the PE at the bottom of the flask, a third set-up was conceived. In this approach, a round bottom flask was used (Figure 3.1c) to decrease the area near the hot plate surface. However, by using this new setup rod climbing was more pronounced, even when using the lower concentration determined previously. For this reason, a magnetic stirrer was used. Initially magnetic stirring worked well, but became limited as a full gel was formed. To solve this issue an overhead stirrer was added at the time of gel formation (Figure 3.1d) to help the magnetic stirrer to rotate and mix the solution. Up to this point, 53

73 the combination of the magnetic and overhead stirrer worked well to decrease polymer degradation and increase the uniformity in the gel formation from PE/xylene solution. However, rod climbing was still an issue. When rod climbing formation started, room temperature xylene was added to the already heated solution of PE/xylene. The amount of xylene added was limited as to not exceed to the amount of solvent needed for the solution. For this reason, the concentration of PE was decreased to half of the original starting concentration. The solution was ultimately kept at this condition to fully dissolve all the PE and ensure a uniform gel. It should be mentioned that several experiments were performed to find the right temperature for the oil bath as well as, the appropriate rotation speed for the both magnetic and overhead stirrer at the different stages of gel formation. After completing these solution formation trials, the final condition and set-up to prepare a uniform gel was developed (shown in Figure 3.1e). (1) 1.7 gr of PE was added to 100 ml xylene at room temperature (25 C). Using a magnetic stirrer (700 rpm) PE was dispersed in xylene for 5 min before the hot plate was turned on. After initial dispersion the hotplate temperature was set at 220 C. PE started to swell due to the increase in temperature and after 1 hour, the solution temperature reached ~137 C and initial gel formation was observed. At this stage the magnetic stirrer did not rotate due to the viscosity of the formed gel. Therefore, overhead stirrer was added to the solution (125 rpm) and the rotation speed of the magnetic stirrer was decreased to a lower rate (225 rpm). (2) Within ~5 min of the changeover in stirring set-up, rod climbing started. 100 ml of room temperature xylene was added to the solution to immerse the PE gel that climbed around the overhead stirrer. (3) The rotation speed was gradually increased for both magnetic and overhead stirrer to 300 rpm and 500 rpm, respectively for another 2 hrs. During this time, 54

74 rod climbed decreased. At the end of this process a clear well-dispersed gel prepared (Figure 3.2). Figure 3.1 Schematic representation of the different flask, heating, and stirring setups used during trials to make a PE and PE-BN gel. 55

75 Figure 3.2 Homogenous clear gel made from PE (bubbles seen in the gel were removed after several hours) Preparation of a Well-Dispressed PE-BN Gel After finding the right parameters to make a homogenous clear gel of PE, several attempts were required to achieve a homogenous gel of PE-BN. In Chapter 2, BN platelets were injected into the PE solution at the onset of PE crystallization (i.e., temperature more than 90 C. The same technique was initially used here to increase the crystallization of the PE at the surface region of the BN particles. However, this technique (injection of BN in PE solution at 90 C) failed due to aggregation of BN particles within the PE gel. Injection at a higher temperature also failed for the similar reason. The PE concentration in xylene for gel spinning is relatively high as compared to the dilute concentration of PE in xylene used for the shear crystallization method (i.e., 850 mg ml -1 vs mg ml -1, respectively). 56

76 It was clear, from these studies, that it was not possible to make a uniform composite gel by using the previous injection method discussed in Chapter 2. Due to these dispersion issues, freshly sonicated BN at the proper concentration in xylene was added to the round bottom flask that is shown in Figure 3.1e. 1.7 gr of PE was added to this dispersion. Room temperature xylene was added to increase the xylene amount to 100 ml while the magnetic stirrer was set at the rotation speed of 700 rpm. Hot plate temperature was set at 220 C to reach to a solution temperature of ~138 C, where stirring continued to reach a homogenous gel. From this procedure a milky PE/BN gel containing relatively well-dispersed BN within the gel network of PE molecules at the desired concentration was obtained. This milky homogenous gel is shown in Figure 3.3 for a 10 wt% BN in PE gel. Figure 3.3 A well-dispersed milky colored PE-BN (10 wt% BN) gel in the round bottom flask. 57

77 3.2.5 Transferring Gel to the Spinning Syringe For gel spinning of the fibers, prepared gels are required to be transferred to a syringe. For this reason, a hot gel made using the aforementioned parameters from PE or PE-BN solutions should be moved to the syringe. During this transfer, about one-third of gel was wasted as it could not be removed from the flask. Once the remaining gel was transferred, the syringe was heated by using band heater. The temperature of the gel inside the syringe was maintained at ~110 C, while the band temperature was kept at 150 C. During this gel-transfer process several problems were observed. (A) First, during initial transfer some portions of the hot gel would pass through to the needle of the syringe and subsequently cool down rapidly within the needle. Since only the syringe is heated by the band, the needle portion could not get warm enough. For this reason, the cooled gel remained stuck in the needle area and no fiber could be formed. To solve this issue the syringe was preheated to 110 C (band heater temperature was 150 C) and a fiber instantly formed as the gel emerged from the syringe. (B) However, a second issue arose related to the overall quality of formed fiber. This quality is in terms of uniformity in diameter as well as continuity of formed fiber. (C) In addition to the fiber quality, use of the band heater was not convenient as it required the user to handle the hot syringe. Changes were made to the spinning process to fix these issues. (D) One final challenge observed was related to the concentration of the gel. As the PE and PE-BN gels cooled prior to spinning the gels contract forcing out some of the excess xylene solvent. For this reason, the concentration of the spinning dope was about half the desired concentration for the gel spinning method. This issue was also connected to the changes made during the solution/gel preparation stage 58

78 where more xylene added to avoid the rod climbing during gel formation process (Section 3.2.3). To solve these issues, the prepared gel was cooled down slowly to the room temperature over a 2 hrs period and left to equilibrate for an additional 15 hrs at room temperature. This cooled gel was subsequently added to the room temperature syringe. The syringe filled with gel was moved to the syringe pump (shown in Figure 3.4). A band heater was used to increase the gel temperature to the desired temperature for gel spinning. During this transition, more than half of the xylene solvent was rejected from the gel. In another observation as compared to the PE gels even more xylene was rejected from the gels which contained more BN particles. Due to the change in PE and BN concentrations within the gels upon rejection of the xylene, subsequent spinning parameters required refinement to ensure a gel fiber could be formed Proper Gel Spinning Parameters After finding the optimum parameters to prepare, transfer and warm up the gel, now finding the optimum gel spinning parameters became important and this study is discussed in the following sections Gel Spinning Temperature The melting temperature for UHMWPE is ~140 C, while the boiling temperature for xylene is also ~140 C. Based on the experiments performed and discussed in Chapter 2, crystallization of UHMWPE occurs at a temperature below 95 C. Based on this data, the spinning temperature should be within the range of 95 to 140 C. For this reason, three 59

79 temperatures were selected 100, 110 and 120 C, respectively. Spinning at 100 C was not successful since the viscosity of the solution/gel was too high. The heated gel did not have a proper flow to easily exit the needle and form a fiber. Spinning at 120 C also was not successful since the viscosity of the solution was now too low. The heated gel came out of syringe too easily making it difficult to control the rate of fiber spinning. The temperature of 110 C seemed to be most appropriate for the spinning the gels processed in these studies. However, even with this temperature the spinning rate needed to be optimized. To set the gel temperature at 110 C, heater band temperature was set at 150 C and after 1 hr gel temperature reached to 110 C Gel Spinning and Fiber Take-up Rate The schematic of the gel spinning process is shown in Figure 3.4. The optimum spinning rate was set at 1 ml min -1 and resultant fiber was collected on take-up wheel at the rate of 0.8 m min -1 after pulling out of ethanol bath. These rates were selected based on several experiments that have done to find out the optimum rate for both the spinning and taking-up the fibers. For spinning, higher rate resulted in thicker fibers near the needle exit. Also at very high rates the plunger was not able to move forward since the viscosity of the gel did not allow the movement of the plunger. Comparitively, a lower rate resulted in discontinues fiber formation likely due to the effect of gravity acting on the gel as it exited the needle. The fiber take-up rate was set based on the proper rate of the fiber spinning. For this purpose, it was found that higher rates led to fiber breakage and lower rates contributed to non-uniform fiber diameter. Therefore, it was found that the minimum rate which fiber did not break and showed a uniform diameter was 0.8 m min

80 Figure 3.4 A schematic illustrating the overall gel spinning processing method Coagulation Bath To form a fiber, the heated gel came out of needle with a proper spinning rate that was explained before. Then this hot gel was passed through a 3 mm air gap between the needle and the coagulation bath and was entered the coagulation bath after that (Figure 3.4). The coagulation bath length was ~400 mm. Based on the previous study, water was used as coagulation bath 197. After trying several experiments, it was found here that fiber collection was not possible when using water as a coagulation bath. Different spinning speeds and take-up rates were used, but it was not feasible to transfer the fiber to the takeup wheels since it was weak and would break even inside the water bath. The fiber weakness was likely due to incomplete coagulation occurring in water. To solve this issue, ethanol (ethyl alcohol) was used at room temperature. Using this solvent, a higher level of coagulation for these fiber occurred and it was found out that it is possible to transfer the fiber from coagulation bath to the take-up wheels Drying Processes for the As-spun Fibers It was found that the as-spun fibers would stick to each other on the take-up wheel. This is indicative of incomplete coagulation of the gel fibers. Based on the studies 61

81 performed here it was found that use of the lab scale spinning set-up would not allow for complete coagulation time to be achieved (i.e., limitation in bath length). Therefore, all spun fibers were moved to from one take-up wheel to a second allowing for the fibers to pass through the air while minimizing the contact regions of fibers. The fibers were then left in a fume hood for 24 hrs to dry. These dried as-spun fibers were further hot-drawn to increase the multifunctional properties of the fiber by increasing the order. 62

82 CHAPTER 4. GEL SPINNING OF THE PE FIBERS 4.1 Introduction As mentioned in Chapter 2, the fibers that have been made using the shear crystallization technique does not show uniform quality and cannot be produced continuously. Mechanical properties of the fibers were also not comparable with commercial fibers properties due to these procedure limitations. To solve these problems, the gel spinning technique was used to allow for scale-up of fiber fabrication, as well as to increase the overall properties of fibers by enhancing the uniformity of and chain orientation within the fibers. The efforts that have been made to find the optimized parameters for gel spinning of the UHMWPE were discussed in Chapter 3. In this Chapter, the morphology of the fibers that were gel-spun and hot-drawn using optimized parameters are characterized and the results are discussed. 4.2 Experimental Section Materials hbn platelets (CAS , ~1 μm, 98%, ρ = 2.29 g cm -3 ) were purchased from Sigma-Aldrich. UHMWPE or PE (Mw ~ g mol -1, ρ = g cm -3 ) was obtained from DSM Dyneema. 2-proponal (CAS , Mw = g mol -1, ρ = g cm -3 ), ethyl alcohol (CAS , Mw = g mol -1, ρ = g cm -3 ) and xylene (CAS , Mw = g mol -1, ρ = 0.86 g cm -3 ) solvents were obtained from Sigma-Aldrich and used as-received. 63

83 4.2.2 Preparation of BN Particles The as-received BN platelets were dispersed via different routes in order to produce two BN filler batches (i.e., BN and nbn). Batch 1: BN was prepared by sonicating BN platelets in xylene (Fisher FS30 bath sonicator, frequency 43 khz; power, 150 W) for 2 h. For preparing 1, 5, and 10 wt% composite fiber, 17, 89, and 187 mg of the BN platelets sonicated in 100 ml xylene. This concentration is used to form 1, 5, and 10 wt% fibers and named PE/1BN, PE/5BN, and PE/10BN. Batch 2: nbn was prepared by sonicating 1000 mg of BN platelet in 200 ml of 2-proponal for 24 h. The sonicated material was subsequently centrifuged (Sorvall Legend Micro 21 Microcentrifuges, Ventilated, 120V 60Hz) at 1000 rpm for 15 min. The supernatants were collected by using a pipette. Collected supernatant dried and sonicated in xylene for another 48 h. The prepared solution of nbn in xylene was stable at least for 24 h. For preparing 1 wt% composite fiber, 17 mg of the BN platelets in 100 ml xylene was used. The resultant fibers from this solution were named PE/1nBN, while control fibers were referred to as PE Gel Formation To prepare control and composite PE gels, 1.7 g of PE was added to the prepared solution of BN in xylene in a round bottom flask. The solution was heated up to 140 C in an oil bath. The solution was further stirred magnetically for about 1 hr at 700 rpm. After 1 hr temperature rises to 130 C and at this point the stirring speed was changed to ~200 rpm. After 5 min the viscosity of mixture increased. At this point, 100 ml room temperature xylene added to the solution. Rotation speed gradually increased to 350 rpm within 2 hrs. At this stage, the homogeneous gel of control or composite PE formed. After that, heat 64

84 removed and prepared gel cooled down to room temperature which subsequently used for fiber spinning Fiber Spinning and Drawing The schematic of the setup for gel spinning is shown in Figure 3.1. The prepared gel was collected from the round bottom flask. At this stage, excess xylene removed from the gel. The excess xylene was 70, 80, 80, 80, and 90 ml for PE, PE/1BN, PE/5BN, PE/10BN, and PE/1nBN gel respectively. Resultant gels were moved to 100 ml syringe and heated up to ~110 C within 1 hr. The solution then flowed through a capillary die of 1 mm diameter and a length of 30 mm. Upon exit the gel fiber was quenched into ethanol at ambient temperature. The spinning rate was set at 1 ml min -1 and the resultant as-spun fiber collected at the rate of 0.8 m min -1. These rates were elected based on several experiments to find out the optimum rate for both the spinning and collecting the fibers. For spinning, higher extrusion rates resulted in thicker fibers exiting the needle. In addition, at very high rate the plunger was not able to move forward due to the viscosity of the gel. Lower extrusion rates resulted in discontinues fiber formation. The fiber collection rate was set based on the proper extrusion rate of the fiber spinning. For this purpose, it was found out that higher take-up rate will break the fiber, while lower take-up rates led to thickening. The as-spun (undrawn) fibers were hot-drawn at 120 C, 130 C, and 135 C with a draw ratio of 5, 4, and 3, respectively. For PE/10BN fiber, the 3rd stage draw ratio was 2.5 since this fiber started to break at a higher draw ratio. For this reason, the final draw ratio was 60 for PE, PE/1BN, PE/5BN, and PE/1nBN fibers, and 50 for PE/10BN fiber. 65

85 4.2.5 Characterization A Zeiss Supra 25 field emission SEM (operating voltage 5 kv) was used for image analysis of the fibers and BN platelets. All samples were coated using a Gatan highresolution ion beam coater with a thin chromium layer (15-20 nm) for imaging. WAXD was performed on a Rigaku RAPID II curved detector XRD system equipped with a 3 kw sealed tube source (voltage 40 kv and current 30 ma). WAXD curve fitting and analysis was performed using software s PDXL 2 (version ) and 2DP (version ). The chemical structure on fiber surface was determined by scanning the surface of fiber assemblies using a Thermo Fisher Nicolet i50 Fourier transform infrared spectroscopy (FT- IR) at attenuated total reflectance (ATR) mode. The spectra were recorded in the manner of attenuated total reflectance. All spectra were run in 16 scans per spectrum in the range of cm -1. Small-Angle X-ray Scattering (SAXS) was performed using a Rigaku MicroMax-007HF system with CuKα radiation (operation voltage 40 kv and current 30 ma). Thermal tests were performed using a DSC (Q200, manufactured by TA Instruments), at 3 different cycles. First, a 1 st heating cycle with a rate of 10 C min -1 from 25 to 200 C, then cooling cycle at with a rate of 10 C min -1 from 200 to 25 C, and finally 2 nd heating cycle with a rate of 10 C min -1 from 25 to 200 C. All DSC cycles performed in nitrogen atmosphere. Tensile tests were conducted using a DMA (RSA-G2 series, manufactured by TA Instruments) with the gauge length of 10 mm and the extension rate of 0.07 mm min -1. The number of tested samples ranged from 5 to 10. The cross-sections of samples were measured using an optical microscope (Olympus BX-51). For this reason, fibers were mounted in epoxy resin and the cross-sections of fibers were prepared after polishing for the measurements. 66

86 4.3 Result and Discussion Morphological Observation of the Fiber and the Structural Development The microstructure for the fabricated fibers was investigated using SEM and representative images are shown in Figures 4.1, 4.2, and 4.3. A well-known shish-kebab structure is observed for the undrawn fibers and is shown in Figure 4.1a for PE, Figures 4.2a and 4.2b for PE/5BN, Figures 4.2c and 4.2d for PE/10BN, and Figure 4.4a for PE/1nBN. This morphology is expected due to the processing approach used for gel formation and fabrication of the fibers 12, 190, 193. For the undrawn PE/1BN, PE/5BN, and PE/10BN composite fibers, PE lamellae crystals were also found to nucleate and grow on the BN platelets (see arrows in Figures 4.2b, 4.2c, and 4.2d). Comparatively, drawn fibers show the predominant fibrillar morphology due to the unfolding of PE chains from the kebab structures upon hot-drawing (i.e., Figure 4.1b for PE fiber and Figures 4.2e and 4.2f for composite PE/10BN fiber). Figure 4.1 SEM images of control PE fibers microstructure. (a) shish-kebab structure in undrawn PE fiber, and (b) fibrillar/extended-chain (shish) structure in drawn PE fiber. 67

87 In general, a good interaction between PE molecules and BN particles was observed for PE/1BN, PE/5BN and PE/10BN for both undrawn (see arrows in Figures 4.2b, 4.2c, and 4.2d) and drawn fibers (see arrows in Figures 4.2e and 4.2f). Even after the mechanical test, SEM study of the fractured area shows a good interaction between PE and BN particles remains due to the lack of full delamination between the matrix and particles that is observed (i.e., PE/10BN fiber in Figure 4.2g). It was observed that by increasing the wt% of BN, some aggregation regions of BN in the fibers were formed (Figures 4.2h and 4.2f for PE/10BN). These regions will act as defects sites in the structure and will affect the mechanical properties. These defective regions are discussed later in the Chapter. These defect sites may also play a role in decreasing the final stage draw ratio for PE/10BN fibers. Structurally, it was also observed that both undrawn and drawn composite fibers exhibit more voids than the control fibers (Figures 4.1 and 4.2). These voids may be due to the BN exfoliation process and can act as defects, resulting in a decrease in the fiber mechanical properties. 68

88 Figure 4.2 SEM images of composite fibers microstructure. (a, b) shish-kebab structure in undrawn PE/5BN fiber, and (c) shish-kebab structure in undrawn PE/10BN fiber, (e, f) fibrillar/extended-chain (shish) structure in drawn PE/10BN fiber, (g) fracture region of PE/10BN after mechanical test, (h, i) aggregation regions of BN in PE/10BN fiber. For PE/1nBN fiber, crystallization of PE on the nbn particles was not observed (Figure 4.3a). Also, it was found that nbn particles prefers to stick to each other instead of dispersing throughout the fibers (Figure 4.3b). Fracture regions which were collected after mechanical tests also show poor interaction or crystallization between PE and nbn particles (Figure 4.3c). After spinning of PE/1nBN fibers, it was also observed that nbn 69

89 particles are ejecting from the PE/1nBN fiber. Therefore, these particles are found on the surface of the fibers. These particles are shown in Figure 4.3d for drawn PE/1nBN fibers. Figure 4.3 SEM images of PE/1nBN fibers microstructure. (a) shish-kebab structure in undrawn PE/1nBN fiber (arrows show loose nbns), (b) aggregation of nbn particles in undrawn fibers, (c) fracture region after the the mechanical test (arrow show loose nbn particle in the structure), and (d) fibrillar/extended-chain (shish) structure in drawn fiber (arrows show nbn particles formed at the surface region of the fiber) Characterization of BN Particles FT-IR was used to characterize the BN platelets within the composite fibers (Figure 4.4). There are four major peaks related to PE bonding at , , , and cm -1. Two peaks related to BN particles at and cm -1 are also observed for all composites fibers. By comparing PE/1nBN and other composite fibers, no evidence of other peaks observed. For this reason, no crystal change or functionalization 70

90 was predicted for nbn particles in comparison to BN particles. It was observed that by increasing the BN ratio, intensity of the related FT-IR peaks of them also increase. Figure 4.4 FT-IR spectra of all the fibers. Two peaks related to BN particles at and cm -1 are observed for composite fibers. WAXD graph of BN and nbn particles are shown in Figure 4.5 and their crystal properties are shown in Table 4.1. Based on the WAXD data, it is clear that intensity ratio of (002)/(100) crystal structure decreases for nbn in comparison to BN particles. The full widths at half maximum (FWHM) of (002) crystal plan of nbn is also wider than BN particles. These results show that nbn particles exfoliated and relative thickness of them is smaller than BN particles. The method that used here for nbn fabrication widely used for exfoliation of hexagonal BN particles and fabrication of BNNSs. 71

91 Figure 4.5 WAXD graph of BN and nbn particles. nbn shows higher intensity for (100) plane due to the exfoliations. Table 4.1 Crystal structure properties of BN and nbn nano-platelets based on WAXD study BN nbn Crystal 2θ d-spacing FWHM Size 2θ d-spacing FWHM Size Plane ( ) (nm) ( ) (nm) ( ) (nm) ( ) (nm) (002) (100) (101) Polymer Orientation and Fibrillar Development The WAXD data shown in Figure 4.6 also reveals micro- and nano-structural changes within the matrix for all fibers. The 2D-WAXD diffraction pattern of the fiber (Figure 4.6a) shows full rings corresponding to the (110) and (200) orthorhombic 72

92 reflections for undrawn fibers, which pertain to the PE crystal unit cell. Comparatively, the 2D-WAXD patterns of drawn fibers exhibit equatorial arc. Post hot-drawing of all fibers led to improved chain orientation as evidenced by the transition in the diffraction patterns from rings to equatorial arc. Despite previous research, no changes in the intensity in the WAXD peak for the BN particles were observed after drawing. For the BN particles prepared here, there was no extreme sonication processing used as compared to the previous research (Chapter 2), for this reason, exfoliation of BN after drawing may be less. For the nbn particles in PE/1nBN fibers, where harsh and long sonication were used, the intensity of (002) plane is not very visible in comparison to the BN peak for the PE/1BN fibers. Since the harsher sonication conditions render the nbn into few layers, no (002) peak was expected especially after drawing. 73

93 Figure 4.6 (a) 2D-WAXD patterns of fibers, and (b) the corresponding intensityversus-2θ profiles of the fibers. 74

94 Figure 4.7 WAXD azimuthal intensity scans of undrawn (U) and drawn (D) (a) PE, (b) PE/1BN, (c) PE/5BN, (d) PE/10BN and (e) PE/1nBN fibers for the (110) reflection peaks. PE chain orientation factor values (f) with respect to the fiber axis for each fibrillar structure were calculated using both the (110) (Figure 4.7) and (200) reflection peaks by Wilchinsky's method 198, 199 and results are shown in Table 4.2. On the basis of these 75

95 orientation results, composite fibers show lower degrees of orientation for as-spun fibers as compared to the control samples. These orientation data for the undrawn fibers show a different trend in comparison with previous data (Chapter 2). For those fibers formed by shear crystallization, orientation data of composite fiber was higher than control fiber due to effect of BN particles during the crystallization process. In general, the orientation factor data for the gel-spun are lower than those found previously. The gel-spun fibers are more homogenous in terms of appearance as compared to the fibers formed by shear crystallization. However, the microstructure may not be as well ordered as the previous fibers. After drawing, orientation of all the gel-spun fiber was increased significantly. As shown in the SEM images and 2D-WAXD data, upon drawing of these fibers, kebab structures diminish and highly ordered fibrillar structures form. The orientation for the drawn PE/1BN fibers is a little lower than drawn PE fibers. However, as the BN content increases, orientation also improved. This may be due to their effect in facilitating of chain slippage as well as some exfoliation of BN platelets during drawing. Orientation of undrawn PE/1nBN fibers are lower than all the fibers, but it is higher after drawing. For undrawn fibers, as shown in the SEM images (Figure 4.4) poor interaction of nbn, which decreases the crystallization as well as introduces defects, may be the reasons for the lower orientation. Higher orientation is may be due to enhancement of the slippage of PE molecules during how-drawing by nbn particles. The percent of crystalline polymer (X c.waxd ) in the matrix was also calculated using WAXD data and the results are shown in Table 4.2. X c.waxd of undrawn fibers are very similar. It appears that the addition of the BN platelets up to 5 wt% slightly enhance the crystallinity, but reduction in crystallinity was observed for higher percentage of BN (i.e., 76

96 10 wt%). In general, drawing of the fibers increases the crystallinity. However, the increase in the crystallinity of the fibers is less for composites with higher BN amount. This can be due to the effect of BN aggregations which may hinder the mobility of PE molecules during the drawing. As explained in the fiber spinning and drawing in the experimental section (Section 4.2.4), PE/10BN fibers was shown to have a lower draw ratio probably due to the same reason. Table 4.2 Herman s orientation factor (f) calculated using Wilchinsky s method, crystallinity calculated from WAXD (i.e., Xc,WAXD) results, DSC data (i.e., ΔHfiber, T1 and T2), and crystallinity calculated from DSC results (i.e., Xc,DSC). WAXD DSC Sample Orientation Xc,WAXD a) ΔHfiber b) factor (f) [%] [J g -1 ] [%] PE PE/1BN Undrawn PE/5BN PE/10BN PE/1nBN PE PE/1BN Drawn PE/5BN PE/10BN PE/1nBN Xc,DSC c) a) crystallinity measured from WAXD; b) heat of fusion of fibers; c) crystallinity measured from DSC Mechanical Characterization Mechanical properties of all fibers are shown in Figure 4.8. The Young s moduli values of the undrawn fibers increased by addition of BN particles (i.e., PE/1BN, PE/5BN, and PE/10 BN) (Figure 4.8a1). For PE/1nBN fibers, Young s moduli of undrawn fiber is less than the control PE fiber due to decrease in the orientation of the fibers as shown in Table 4.3. Hot-drawing of fibers increases the Young s modulus of all the fibers (Figure 4.8a2) which is related to the introduction of chain ordering and orientation in the 77

97 composite fibers during hot-drawing. Young s moduli of composites made using increasing BN loadings (i.e., PE/1BN, PE/5BN, and PE/10 BN) first increase and reach the maximum for the 5 wt% BN. However, the mechanical properties decreased at 10 wt% BN. For PE/1nBN drawn fibers, the modulus was lower than the control drawn PE fibers. This fiber shows better orientation in comparison to the other fibers, but at the same time the Young s modulus is lower than all the fibers. The mechanical behavior of the drawn fibers is related to the morphological changes occurring during the drawing process. In general, modulus values are found to be larger for the as-spun composites made using BN loading of PE/1BN, PE/5BN, and PE/10 BN, respectively (Figure 4.8). The improvement in Young s modulus of drawn composite fibers made from BN particles can be related to the higher orientation as well as exfoliation of BN particles during tensile test. Orientation factor (f) determined from WAXD shows a slight decrease for PE/1BN fibers and slightly increases by addition of more BNs. Among composites made from BN particles, PE/10BN has the lowest modulus probably due to lowest draw ratio between the fibers. (i.e., 50 for PE/10BN vs. 60 for other fibers). The Orientation factor of drawn PE/1nBN fiber is highest among the drawn fibers (0.95), but at the same time the Young s modulus is the lowest among the fibers. Therefore, one possible reason for improvements in the measured modulus may be indicated by the exfoliation of BN platelets upon tensile test, which is due to strong filler-polymer interactions (i.e., exfoliation is driven by polymer matrix/chain sliding). This exfoliation process may also locally (i.e., at the filler-polymer interphase) regulate or confine the directional molecular alignment, which could also contribute to the increase of composite moduli made from BN particles. nbn particles should contain fewer layers. For this reason, 78

98 possible exfoliation of these particles are less than BN particles which contains more layers. In general, the strength measurements show a comparable range for all undrawn samples. Comparison of the tensile strength measurements of drawn fiber shows that the PE/1BN and PE/5BN composites show higher tensile strength than the control fibers. By addition of BN (i.e., PE/10BN) tensile strength decreases and is lower than control PE fiber. PE/1nBN fiber shows the lowest tensile strength. Based on the SEM analysis (Figures 4.2 and 4.3), the morphology of all fibers show the presence of voids which are related to the spinning process used as well as filler presence. These defects will contribute to a decrease in the strength for the fibers. The lower mechanical properties of PE/1nBN fiber could be related to the method of preparation of nbn particles. IPA treatment in conjunction with the use of xylene may have changed the surface chemistry of nbn particles. As compared to the BN platelets treated with only xylene, those which included IPA showed a lack of interaction between the nbn and PE molecules during gel preparation and fiber spinning. For this reason, the nbn dispersion did not perform as well as the BN dispersion in the processing of these gel-spun fibers, and this is reflected in their mechanical properties (Figure 4.8) In general, the gel-spun fibers here are found to exhibit much better properties as compared to the fibers produced by shear-crystallization. By comparing the mechanical properties of gel spun fibers (Chapter 3) with shear crystallized fiber (Chapter 2), ~46 and ~6 fold improvement in E and ơ for control undrawn fibers was observed, respectively. These improvements show that the continuous gel spun fibers contain less defective features as compared to those produced through crystallization routes. However, both 79

99 studies show that the role of good interfacial interaction between the matrix and polymer is crucial for excellent stress transfer and mechanical properties. Figure 4.8 Mechanical properties of the fibers. (a) Young s modulus and (b) tensile strength of the fibers. For a polymer filled with aligned platelets like BN or nbn, the modulus and strength can be expressed by the modified rule of mixture and are shown in Table 4.3. E = (η le E BN E PE )V BN + E PE (4.1) And 80

100 σ = (η lσ σ BN σ PE )V BN + σ PE (4.2) Where, E BN, E PE, σ BN, σ PE are the Young s modulus and tensile strength of the BN platelets and PE matrix and η le and η lσ are the length efficiency factors which reflect the dependence of reinforcement on platelet length and increase from 0 and 1 with increasing platelet aspect ratio (length/thickness, l/t). η le is defined by Equation η le = 1 tanh(nl t ) ( nl t ) (4.3) with G PEV BN n = E BN (1 V BN ) (4.4) Where, G PE is PE shear modulus. Shear modulus can be estimated to be ~16 GPa from E PE = 2G PE (1 + ν PE ) where ν PE is the PE Poisson ratio (~0.5) 185, 202. E BN is 750 GPa 177. Based on the theoretical study by using rule of mixture analysis, a similar increasing trend in modulus is observed except for all fibers. Based on the experimental properties (Figure 4.8) a similar trend is observed with the exception of the PE/10BN. As observed in SEM images (Figure 4.2), aggregation of BNs in the PE/10BN fibers is found as well as the presence of voids in the structure. For this reason, the fiber properties decrease. The 81

101 orientation values are also lower than the other composite fibers, showing that the defects present also affect chain orientation. This also contributes to the lower experimental modulus values for the PE/10BN (Table 4.3). Comparatively both 1 and 5 wt% fibers show a higher modulus experimentally than is predicted. This increase in properties is likely due to the increased interfacial interaction observed in the fibers by SEM. It is also important to note that all theoretical analysis is done assuming perfect orientation and dispersion of the BN particles. For this reason, the experimental properties show here are very promising as these fibers still have considerable room for improvement in the BN dispersion and orientation. Table 4.3 Experimental and theoretical Young s modulus of the fibers SAXS Characterization E (GPa) (Experimental) E (GPa) (Theoretical) PE PE/1BN PE/5BN PE/10BN In addition to SEM and WAXD, SAXS measurements were also used to characterize the morphological changes in the fibers (Figure 4.9). Meridional lobes observed in the SAXS pattern are consistent with lamellae crystal development or kebabs, while equatorial streaks are consistent with the presence of fibrillar/extended-chain crystals referred to shish for the crystals formed in the fibers 199. The presence of both lobes and streaks in the SAXS patterns of undrawn fibers are consistent with the shish-kebab structures observed for all the undrawn (i.e., as-spun) fibers (Figures 4.1, 4.2, and 4.3). After drawing, the meridional lobes diminish and the equatorial streaks become more 82

102 pronounced for all the fibers. These structural changes shown by SAXS also support the observations found by both SEM (Figures 4.1, 4.2, and 4.3) and WAXD (Figure 4.6) measurements. Figure 4.9 2D-SAXS patterns of all the fibers. Overall SAXS intensity-versus-q profiles corresponding to kebab/lamellae distributions (Figures 4.10a and 4.10b) were obtained by integrating along the meridional direction. The curves for the undrawn fiber samples show broad scattering peaks (Figure 4.10a). The SAXS curves for the drawn samples show no distinct scattering peaks, which is due to the wide distribution of lamellae thicknesses present in these fibers. 83

103 Figure D-SAXS intensity-versus-q curves for (a) undrawn and (b) drawn fiber samples Thermal Analysis by DSC Thermal analysis of the fibers was investigated using DSC, where both curves and relative information pertaining to melting behavior and crystallinity are shown and listed in Figure 4.11 and Table 4.1, respectively. Figure 4.11a is 1 st DSC heating curve and Figure 4.11 b is the 2 nd DSC heating curve. The 1 st heating curve demonstrates the thermal history of fibers, while the 2 nd heating curve does not show the history of the fibers. Based on 1 st heating curve, undrawn fibers show a melting peak at ~135 C and below 140 C associated with melting of orthorhombic folded chain PE crystals (Figure 4.11a). DSC graphs for drawn fiber show three to four melting peaks. Melting peak that observed has been assigned to melting of unconstrained orthorhombic PE, constrained orthorhombic PE, and orthorhombic-hexagonal phase transition, followed by melting of the hexagonal phase. 193, 210, 211. By comparing the melting behavior of the fibers based on the 2 nd peak (Figure 4.11b), it is obvious that thermal history of the fibers from drawing is removed, especially 84

104 for the drawn samples, and both undrawn and drawn fibers show similar melting behavior with a melting peak at ~131 C for undrawn and ~132 C for drawn fibers. For the drawn PE/1BN and PE/1nBN fibers, the intensity of constrained orthorhombic is lower than drawn PE, PE/5BN and PE/10BN which shows less amount of this phase in the microstructure. Figure 4.11 DSC graphs of (a) 1 st heating cycle and (b) 2 nd heating cycle of the fibers. 85

105 Fiber crystallinity (X c.dsc ) was determined based on DSC measurements by using Equation 4.5, and shown in Table 4.2. H fiber is the heat of fusion measured for each sample (Table 4.2), and H o is estimated for a PE single crystal to be ~293 J g X c.dsc (%) = H fiber H o 100 (4.5) The 1 st heating curves are used for this analysis to ensure that the thermal processing history of all samples is comparable. In general, the crystallinity values calculated using DSC (X c.dsc ) are lower than the values calculated using WAXD (X c.waxd ). This difference is related to the presence of confined PE crystals within the fibers, which do not melt completely during the first heat cycle. Analysis of X c.waxd and X c.dsc values show that the undrawn composite fibers exhibit a higher degree of confinement as compared to the control fibers (Table 4.2). The presence of the rigid BN platelets is expected to increase confinement of the PE crystals. This is supported by the observation of PE nucleation at the surface of the BN platelets in the undrawn fibers (Figures 4.2b and 4.2c). Considering that the BN platelets becomes exfoliated during drawing, confinement becomes less pronounced as the single layer BN sheets exhibit less rigid properties. Comparison of the X c.waxd and X c.waxd values for the drawn fibers (Table 4.2), also shows that confinement effects are very similar for all samples. This is due to the structural conversion to a more fibrillated state as a result of drawing. The drawn composites also no longer show more pronounced confinement effects in comparison to 86

106 the control fibers considering that the exfoliation of the BN platelet stacks leads to the presence of more flexible BN sheets (i.e., much less rigid than the layered counterparts in the undrawn state). 4.4 Thermal Conductivity Thermal conductivity of PE/BN fibers can be predicted by using a rule-of-mixture approach. Thermal conductivity of BN platelet and undrawn UHMWPE fiber can assume to be ~40 W m -1 K -1 and ~1 W m -1 K -1, respectively (Chapter 1). By using the theoretical rule-of-mixture analysis, thermal conductivity for 1.2 % vol BN platelet loading the composite is predicted to be ~14.6 W m -1 K -1. This prediction is very much in line with the thermal conductivity measurements reported in Chapter 2 for undrawn PE/BN fiber. By assuming the value of thermal conductivity of drawn UHMWPE to be ~40 W m -1 K -1, for 0.5 % vol BN, thermal conductivity of PE/BN fibers is estimated to reach ~40 W m -1 K -1. This prediction is a relatively high value for a polymer composite material (Table 1.2). Based on this analysis, if the BN content is increased even higher thermal conductivity can be achieved. While experimental measurements were not taken for the gel-spun fibers, the samples were prepared for testing (see Appendix A). 4.5 Conclusion UHMWPE/BN composite fibers are fabricated using shear crystallization and subsequently drawn near the polymer melting point. Due to this processing approach insitu exfoliation of BN in the polymer matrix occurred. Exfoliation of BN platelets is confirmed by WAXD and Raman spectroscopy analysis. SEM images illustrate that crystallization of UHMWPE occurs on the surface of the BN platelets exhibiting good 87

107 interfacial interaction. This polymer-bn interaction was found to contribute to exfoliation of BN platelets upon hot-drawing. These morphological changes resulted in improvement in mechanical properties for the composite fibers. Thermal property characterization for these samples is ongoing. 88

108 PART TWO PAN/CNT Composite Films 89

109 CHAPTER 5. LOW-TEMPERATURE GRAPHITIC FORMATION PROMOTED BY CONFINED INTERPHASE STRUCTURES IN POLYACRYLONITRILE/CARBON NANOTUBE MATERIALS 5.1 Introduction PAN is widely used as a precursor for carbon fiber production , 116, 117, 124, 144, 213. Typically, a three-step heat treatment process is performed in order to convert PAN precursors to carbon fibers. These steps are: (i) thermal stabilization, (ii) carbonization, and (iii) optional graphitization. During stabilization between a temperature ranges of 200 to 300 C, a ladder structure will be formed due to cyclization of PAN molecules 124, 128. Carbonization processes increase the carbon content through formation of a nearamorphous microcrystal structure 129. Carbonization is usually performed within a temperature range from 1000 to 1700 C 116, 144. Optional graphitization will further enhance the ordering of the carbon structure by formation of graphite and this process occurs at temperatures greater than 2000 C and up to 3000 C 116, 144. By increasing the heat treatment temperatures and applying tension to the fibers during pyrolysis, ordering of the carbon structure increases 131, 214, 215. Semi-crystalline carbon is referred to turbostratic, while highly ordered carbon is graphitic. The inter-layer distance between turbostratic carbon planes varies and is on average >0.34 nm 131, 216, while the distance between perfect graphite planes is nm 217, 218. Conversion to this graphitic structural form in the material improves the mechanical properties of carbon fibers. 90

110 New research on carbon fiber development and processing have been of interest toward decreasing the gap between commercial and theoretical tensile strength properties for the materials 27, 157. An increase in tensile strength properties is expected to occur in the presence of oriented graphitic regions which can be increased throughout the fibers, while minimizing defective carbonaceous regions. There have been several recent studies focused on understanding the role of controlling the precursor polymer structure toward producing a specific variation of carbonaceous structure upon pyrolysis by addition of CNTs to the PAN matrix 27, 145, 156, 158, 219. CNTs have been shown to induce ordered interphase 18, 26 which can promote the ordering of the PAN ladder structure after stabilization 27, 156. This interphase structure leads to PAN conversion to graphitic structures at ~1100 C 27, 149, 158, 159, 156. However, while it has been shown that graphitic structure can form at 1100 C at the interphase, studies focused on understanding the mechanism behind this conversion process and subsequent structure are still needed. In this paper, PAN/CNT composites were prepared to purposely isolate interphase regions in the composite. By preparing such precursor materials, PAN conversion during carbonization heat treatment could be studied specifically. This work is performed to confirm the value of introducing ordered PAN interphase regions to the precursor to control graphitic forms in the final carbon material. This work also shows the role of polymer confinement effects in early carbonaceous and graphitic formation. 91

111 5.2 Experimental Section Materials The PAN used in this work is a poly(acrylonitrile-co-methacrylic acid) random copolymer with methacrylic acid content of 4 wt% (Mw ~513,000 g mol -1 ), obtained from Exlan Co. Japan. Two types of CNT materials were used, (i) PT (purified SWNT), ~94.5%, Continental Carbon Nanotechnologies, Inc.) and (ii) SW (SWNT, ~90%, SouthWest Nano Technologies, Inc., batch SG76) Solution Processing and Film Fabrication Dispersion Preparation The materials used in this study are PAN/CNT composite films. For hybrid polymer/cnt buckypaper (hpbp) fabrication, PAN powders were first dissolved in 90 C dimethylformamide (DMF) (obtained from Sigma Aldrich) at a concentration of 250 mg L -1. CNTs of equal amount were then dispersed in the polymer solution for 24 hrs via a bath sonicator (Fisher FS30, frequency 43 khz, power 150 W). For control BP fabrication, CNTs were dispersed in DMF using similar sonication conditions BP and hpbp Films Preparation After sonication, the PAN/CNT dispersion was subjected to a solution-based shear crystallization process by stirring the dispersion at 90 C. Simultaneously vacuum distillation was applied to remove half the volume of solvent at a controlled time and temperature. Subsequently, this overall system was cooled down to room temperature (~25 92

112 C). A non-solvent for PAN was added into dispersion at various solvent:non-solvent (S:NS) ratios. The two ratios chosen in this work were 7:1 and 1:2, respectively. This S:NS treatment during solution processing of the PAN/CNT films help to isolate regions, where PAN has enhanced interfacial interaction with the CNT. The isolation of this interphase region is important for studying subsequent carbonization process effects. The final PAN/CNT dispersions were then filtered through a nylon membrane (0.45 µm pore size obtained from Millipore) to form the PAN/CNT hpbp. The films made using PT-SWNT with S:NS ratios of 7:1 and 1:2, respectively, were named PT-1 and PT-2, accordingly. The films made with SW-SWNT were named SW-1 and SW-2, respectively at similar S:NS ratios of 7:1 and 1:2. A free-standing hpbp was removed from the filter paper after drying in a vacuum oven and further characterized to understand the effect of processing on the structure of the hybrid films. Control CNT BPs were also fabricated using a similar filtration process for comparison Heat-treatment Processes A Lindberg/Blue M Mini-Mite tube furnace equipped with a quartz tube (diameter 1 inch, obtained from Quartz Scientific Inc.) was used for heat-treatment to conduct both stabilization and carbonization procedures. Two flow meters were used to control the inlet and outlet gas flow through the furnace. The entire system was sealed and the outlet gas flow was directly released into a venting system. The film samples were held in compression using a customized sample holder throughout both heat treatment processes. For stabilization (in air), the temperature was (i) ramped up at 1 C min -1 from room temperature (~25 C) to 250 C or 300 C; (ii) maintained isothermally for 10 hrs; and (iii) gradually decreased to room temperature. For carbonization (in argon), the 93

113 temperature was (i) ramped up at 5 C min -1 from room temperature to 900, 1000 or 1100 C; (ii) maintained isothermal for 20 or 40 min at 900 C, 5 or 20 min at 1000 C, and 5 min at 1100 C; and (iii) gradually decreased to room temperature (~25 C). In both processes, the required air or argon gas flow was maintained constant at 3000 ccm until the chamber was cooled down to room temperature (~25 C). To confirm the existence of structural features the early onset graphite (i.e., formed between 900 to 1100 C), the pyrolysed composite films were further heat-treated at up to 2100 C for 40 min in vacuum using high temperature furnace (Red Devil vacuum furnace WEBB 124). Figure 5.1 summarizes the time-temperature profile for the stabilization, carbonization, and graphitization procedures Sample Characterization Morphology characterization was performed using a Zeiss Supra 25 field emission SEM (operating voltage 5 kv). All film samples were fractured and mounted to a 90 pin stub with the fractured end facing up for SEM observation. Precursor (non-carbonized) samples were coated with a thin gold/palladium layer (15-20 nm) for image purposes using a Gatan high-resolution ion beam coater. WAXD was performed on a Rigaku RAPID II equipped with a curved detector XRD system with a 3 kw sealed tube X-ray source (operating voltage 40 kv and current 30 ma). XRD curve fitting and analysis was performed using software PDXL 2 (version ) and 2DP (version ). Raman spectroscopy was conducted on a Jobin Yvon LabRam HR800 (laser wavelength 532 nm). DSC analysis was conducted by heating samples from 40 to 400 C at a heating rate of 1 C min -1. DSC curves were examined using the TA Instrument Universal Analysis software. 94

114 Figure 5.1 Graphical representation for the heat treatment procedures used in this work for (a) stabilization, (b) carbonization, and (c) graphitization of the samples. 95

115 5.3 Results and Discussion Precursor Films A non-solvent is added during filtration to process the hpbps in order to isolate PAN- CNT interactions. PT-1 and SW-1 hpbps are fabricated using a 7:1, S:NS ratio, while PT- 2 and SW-2 materials utilize a 1:2, S:NS ratio. It is expected that as the non-solvent increases, PAN-CNT isolated interactive regions also increase and these changes are observed in the composite morphology. Cross-sections of fabricated precursor PT-1 and PT-2 hpbp films are shown in Figures 5.2a and 5.2b. PT-1 film cross-sections show a heterogeneous structure, where CNTs are randomly distributed throughout the films. However, the PT-2 films show a two-layered structure; A PAN-rich layer (PR-L) which is dominated by PAN and contains a lower concentration of CNTs, and a CNT-rich layer (CR-L) that contains little polymer but a very high concentration of CNT. From a morphological point of view, the structure of the CR-L in the PT-2 sample is similar to the overall structure for the PT-1 film, while the PR-L is much denser and shows even distributions of the CNT. Although for both films that same amount of PAN and CNT is used, in general, the PT-2 film is thicker than the PT-1 samples. This thickness difference may be related to the loose and compact structures present in the CR-L and PR-L regions, respectively. As mentioned, hpbps from two CNT types were fabricated. Similar nonlayered and layered structures are observed for SW-1 and SW-2 films and are shown in Figures 5.2c and 5.2d. The SW-2 two-layered films also exhibit a PR-L and CR-L and film thickness trends are similar to those for the PT-based hpbps. 96

116 Figure 5.2 SEM images of cross-sections of fabricated films: (a1) PT-1 film, (a2) zoom in area of the boxed region in a1 image; (b1) PT-2 film, (b2) zoom in area of boxed region shows CR-L, (b3) zoom in area of boxed region shows PR-L; (c1) SW-1 film, (c2) zoom in area of boxed region in c1 image, (d1) SW-2 film, (d2) zoom in area of boxed region shows CR-L, (d3) zoom in area of boxed region shows PR-L. Based on electron microscopy analysis, the formation of the layered structures in the PT-2 and SW-2 hpbps were found to affect PAN morphology. It was also observed that during film processing (by filtration) the approximate PAN weight loss percentage for PT- 1, SW-1, PT-2 and SW-2 was found to be 27.3, 25.0, 11.1 and 10.0%, respectively. Therefore, in general, the concentration of PAN in the PT-2 and SW-2 films is slightly higher than for the PT-1 and SW-1 samples. Considering the presence of PAN in all films, WAXD analysis was performed in order to investigate the actual structural features of the PAN polymer in the composites. The WAXD micrographs for the PT-1, PT-2, SW-1 and SW-2 films are shown in Figure 5.3. It is observed that the PT-2 and SW-2 films show a narrow and predominate peak at 2θ of 16.7º associated with crystalline PAN (dotted box in Figure 5.3). The broad peak observed for the PT-1 and SW-1 films is consistent with a 97

117 more amorphous PAN morphology (dotted box in Figure 5.3). This WAXD data suggest that for the single layered films the PAN predominately retains an amorphous structure, whereas in the two-layered films the polymer is more crystalline. Based on the electron microscopy analysis, it is clear that PAN in the two-layered films is isolated in the PR-L, this WAXD data also suggests that the PAN in this region is highly ordered. As shown in Figure 5.4, the SWNT in the PR-L are more homogenously distributed as compared to the CR-L region. This WAXD analysis shows that the improved dispersion of CNT in the PR- L region allows for PAN to form more ordered domains. Previous research has shown that well-dispersed SWNT can promote interphase formation in PAN 18, 145. These PT-2 and SW-2 hpbps allow for the isolation of such ordered regions in the composite material, which enables a close study of how ordered interphase PAN affects heat-treatment toward turbostratic carbon and graphitic formations. Figure 5.3 WAXD spectra of the PT-BP, PT-1, PT-2, SW-BP, SW-1 and SW-2 films. PT-1 and SW-1 films show a broader PAN (110) peak, PT-2 and SW-2 materials exhibit a sharper crystalline PAN (110) peak. Several broad peaks pertaining to the SWNT are also observed (see arrows). 98

118 Figure 5.4 Magnified SEM images for a two-layered hpbp showing the (a) CR-L and (b) PR-L, regions. Based on both microscopy and WAXD analysis a cross-sectional schematic was developed to illustrate the significance of the layered structure in the hpbps (Figure 5.5). The schematic shows that the distribution and subsequent ordered (interphase) formation of PAN in the vicinity of the CNT differs for the one-layer films (PT-1 or SW-1) as compared to the two-layered films (PT-2 or SW-2). These structural features ultimately affect the overall PAN structural distribution within the precursor films, and this is expected to influence the formation of the carbonized structures upon heat-treatment. Isolation of the interphase regions is important for tracking the transformation of this region during carbonization, and its eventual role in mechanical and electrical properties of the overall composite. In addition, the different types of CNT used here (i.e., PT and SW) are found to slightly affect the ordered (interphase) structure formed (Figure 5.3), which will in turn influence carbonization

119 Based on the schematic in Figure 5.5, the one-layered films exhibit a large distribution of CNT bundle size with surrounding PAN matrix. While some interphase PAN may be present, the matrix is dominated by amorphous PAN as indicated by WAXD. The two-layer schematic show both CR-L and PR-L regions. It should be noted that for the film processing conditions used, the transition from the CR-L to PR-L layer is abrupt. The CR-L region is dominated by CNT of varying bundle size exhibiting an average of ~20.48 nm 220. The CR-L is also a looser structure were many voids are present and little PAN is found. Comparatively, the PR-L was found to exhibit much smaller average bundle size (~6.25 nm) with less variation, lower void content, and higher presence of PAN 220. Based on WAXD this polymer is also found to be ordered (crystalline), and this is associated with interphase PAN. Some disordered PAN matrix is also expected to be present. However, unlike the one-layered films (no CR-L or PR-L distinction), the ordered interphase PAN dominated based on WAXD analysis. By using this visual scheme, the heat-treatment of the one-layer and two-layered films are tracked as a function of morphology, in order to understand the effects of pyrolysis. 100

120 Figure 5.5 Cross-section schematic for the fabricated films. The interphase region is assumed to be larger around smaller diameter SWNT bundles due to prior studies 18, 220. Due to the structure of the PAN hpbps films produced here, DSC analysis was first performed to anticipate stabilization effects. DSC thermographs for the PT-1 and PT-2 films under air show that the heat evolved for the PT-1 films is higher than that of the PT- 2 samples (Figure 5.6). This suggests that the CNT in the PT-2 films interact with PAN in a way that hinders the PAN stabilization process. In other words, the PAN present in the PT-2 films exhibits higher thermal stability in comparison to the PAN in the PT-1 films. This effect may be related to the increase of ordered packing in PAN for the PT-2 films as compared to PT-1 (Figure 5.3). Previous work 27 has shown that with increasing presence of interfacial structure in PAN/CNT composites, due to compact crystal formation of PAN in this region, stabilization is less complete at similar heat treatment conditions used for the sample which shows no interphase formation. Therefore, in this work, the stabilization 101

121 temperatures are examined to promote complete processing (cyclization) at this stage. Completion of exothermic reaction is found to occur at ~300 C and higher, while the reaction begins at ~250 C and 300 C to examine the effect of PAN to ladder structure transformation. Completion of this transformation is important for subsequent carbonization steps. Figure 5.6 DSC thermographs of PT-1 and PT-2 films stabilized at the heating rates of 1 C min Film Stabilization WAXD data of the stabilized films at 250 and 300 C for both PT-1 and PT-2 samples are shown in Figure 5.7. The PAN peak at 2θ of 16.7º diminishes for all the films except the PT-2 sample stabilized at 250 C. This indicated that stabilization is incomplete for the PT-2 films as mentioned early, where DSC curves (Figure 5.6) show lower heat flow. This 102

122 is not the case for films stabilized at 300 C. WAXD graphs for the SW-1 and SW-2 films stabilized at 300 C are also shown in Figure 5.7. Similar to the PT films stabilized at 300 C, the PAN peak is not observed in these samples. For this reason, all samples were subsequently stabilized at 300 C before further carbonization and graphitization. The stabilization process converts PAN to a cyclized ladder and cross-linked structure. This new structure is observed in WAXD by the presence of a broad scattering peak between 2θ from 23 to 25 (Figure 5.7) 145. Figure 5.7 WAXD spectra for the stabilized PT-1, PT-2, SW-1 and SW-2 films. A broad peak between 2θ from 23 to 25 is related to PAN ladder structure 145. SEM images for the cross-sections of all stabilized films are shown in Figure 5.8. Like the precursor films, PT-1 shows a non-layer structure (Figure 5.8a) while PT-2 films maintain its bi-layer structure (Figure 5.8b). The thickness of PR-L in the PT-2 films is 103

123 decreased in comparison to the precursor sample. A similar trend is observed for SW-1 and SW-2 films (Figures 5.8c and 5.8d). It is expected that some weight loss will occur in the PAN matrix during heat-treatment and conversion to the cyclized form 117, 144. Figure 5.8 SEM images of cross-sections for (a1) stabilized PT-1 film, (a2) zoom in of boxed region in the PT-1 film; (b1) PT-2 film, (b2) zoom in of boxed region in the PR-L area in the PT-2 film, (b3) zoom in of boxed region in the CR-L area in the PT- 2 film; (c1) stabilized SW-1 film, (c2) zoom in of boxed region in the SW-1 film, (d1) SW-2 film, (d2) zoom in of boxed region in the PR-L area in the SW-2 film, (d3) zoom in of boxed region in the CR-L area in the SW-2 film Film Carbonization The stabilized films are carbonized at different temperatures and times (i.e., 900 C for 20 and 40 min, 1000 C for 5 and 20 min, 1100 C for 5 min, and 1500 C for 5 min). Previous work by this research group has shown that the presence of interphase PAN can lead to early onset of graphitic structures at 1100 C 145. In this work, the time and temperature effects on PAN to carbon evolution during carbonization is explored further. WAXD spectra for these carbonized films are shown in Figures 5.9a and 5.9b. The values 104

124 of d-spacing, FWHM, and crystal size for (002) graphitic peak at 2θ of ~26 are provided for all films in Table 5.1. Layered structure, as well as time and temperature dependence of heat treatment on the structure of the carbonized films, is also discussed in the following sub-sections. Figure 5.9 WAXD spectra for the carbonized films (a) at temperatures ranging from 900 to 1100 C and (b) at 1500 C. (002) graphitic peak occurs at 2θ of ~26. AC is the amorphous carbon peak. Peaks 1,2,3,4 and 5 are related to higher order graphitic planes (100) of hexagonal form with ABAB stacking, (101) of hexagonal form with ACBACB stacking or (100) of rhombohedral form, (101) of hexagonal form, (102) of hexagonal form with ACBACB stacking or (110) of rhombohedral form and (102) of hexagonal form with ABAB stacking or (103) of hexagonal form with ACBACB stacking, respectively

125 Table 5.1 Effect of carbonization time and temperature on (002) plane crystal structure development in carbonized films. PT SW Temperature ( C) Time (min) θ ( ) d-spacing (nm) FWHM ( ) Crystal size (nm) I (101)H/(100)R:I a) (101)H θ ( ) d-spacing (nm) FWHM ( ) Crystal size (nm) I (101)H/(100)R:I a) (101)H a) I: intensity of peak, H: hexagonal, R: rhombohedral. Layer(s) Effect of Layer Structure Dark boxes in Figures 5.9a and 5.9b show the (002) graphitic peak region at 2θ of ~26. For PT-1 and PT-2 films carbonized at 900 C for 20 min, 900 C for 40 min, 1000 C for 20 min, 1100 C for 5 min, and 1500 C for 5 min the (002) graphitic peak at 2θ of ~26 is visible. No graphitic peak is observed for films carbonized at 1000 C for 5 min. In general, the (002) graphitic peak for the PT-2 film is broader than for the PT-1 sample carbonized at similar conditions. At this condition, the FWHM for PT-2 film at 2θ of ~26 is 3.267, while for the PT-1 film it is The WAXD results indicate, based on the average 2θ, d-spacing, and crystal size parameters for the graphitic (002) plane, that the PT-1 films generally exhibit a more ordered carbon structure as compared to PT-2 samples. This result seems somewhat counterintuitive to what was observed for the precursor materials, where PT-2 films show more crystalline PAN. Previous studies also suggest that this ordered PAN helps to promote ordered carbon formation 145. However, the PT-1 films 106

126 (showing less interphase) exhibit higher degree of graphite ordering in terms of crystal size. This result may be due to two features. First, in general, PT-2 films are found to be thicker than PT-1 films. For this reason, less diffusion is anticipated in PT-2 film, which may delay the carbonization process. By comparing PT-2 films which were heat treated at the same temperature (i.e., 900 C) but for longer times (20 vs. 40 min), more complete carbonization was observed as time increases. This is shown by the presence of a more intense and narrower (002) graphitic peak. This result also confirms the role of thickness on the heat-treatment of the films. Second, WAXD and electron microscopy analysis of the precursor film indicate that the PT-2 film has a predominate presence of interphase regions in the PR-L, this region may not be very confined due to the lower content of CNT in this area. On the other hand, for PT-1 films, the PAN is more confined by surrounding rigid CNT even though there may be lower overall interphase content. Therefore, these carbonization results suggest that, while the presence of the PAN interphase is important for the formation of graphitic structure at low carbonization temperature, this region also requires some additional confinement of the PAN to allow for more ordered development early on. Previous work has shown the influence of precursor and confinement on the early onset of ordered carbonaceous species 145. Analyses of the carbonized SW-1 and SW-2 films show no obvious graphitic peak. Instead, the peak at 2θ of ~26 from WAXD is broad indicating a predominance of amorphous or turbostratic carbon (Figures 5.9a and 5.9b). This result confirms that the CNT type also plays a role not only to nucleate and promote formation of the interphase PAN but also to provide confinement during the carbonization process

127 Effect of Time and Temperature While the onset of graphitic phase transformation was found to occur at temperature as low as 900 C, sufficient time for the graphitization transition to occur is variable. Films carbonized at different temperatures and keeping time constant were compared to understand its effect on changes in crystal structure of the films. It was found that the graphitic peak intensity increased with temperature. These results suggest that a higher degree of graphitization occurs in films carbonized at lower temperature. For example, PT- 1 and PT-2 films carbonized at 900 C for 40 min show graphitic (002) peak, but films carbonized at higher temperature using less time (i.e., 1000 C for 5 min) do not show any graphitic (002) peak. By comparing films carbonized at 1100 C for 5 min and 1000 C for 20 min, it was found that crystal size is smaller and FWHM is broader for samples carbonized at higher temperature but less amount of time (i.e., 1100 C for 5 min). WAXD curves (Figures 5.9 and Table 5.2) also show higher order peaks related to the crystalline form of the graphitic structure formed. Peaks at 2θ of ~43.7 are related to the (101) crystal plane for hexagonal (i.e., with ACBACB stacking) or (100) of rhombohedral carbon structures. Peaks at 2θ of ~44.7 are related to (101) crystal plane of hexagonal form (i.e., with ABAB stacking). The intensity ratio between the peaks at 2θ of 43.7 and 44.7 is increased by carbonizing at higher temperature (Table 5.1). This implies that for these samples, carbonization at higher temperature increases the amount of (101) of hexagonal with ACBACB stacking or (100) of rhombohedral structures which causes an increase in interlayer distance between the graphitic structure. For this reason, carbonization at lower temperature (i.e., 900 C) but longer time (i.e., 40 min) may lead to a more ordered graphitic structure. The intensity ratio for the peaks at 2θ of 43.7 and

128 in the PT-2 film is higher than PT-1 at similar carbonization condition. This suggests that the presence of more confined PAN in PT-1 films, which decreases the formation of crystal planes with larger spacing between graphitic layers (i.e., ACBACB stacking of hexagonal and rhombohedral structure). Table 5.2 Higher order WAXD peaks associated with the hexagonal and rhombohedral forms of layered graphite 145. Peak Indexed planes for different crystal unit cells Experimental (ABAB) (ACBACB) 2θ ( ) Hexagonal Hexagonal Rhombohedral (1 0 0) (1 0 1) (1 0 0) (1 0 1) (1 0 2) (1 1 0) (1 0 2) (1 0 3) To investigate the microstructure of carbonized films, SEM images of PT films carbonized at 1000 C for 5 min, 1100 C for 5 min, and 1500 C for 5 min were examined and compared (Figure 5.10). The films all show evidence of shrinkage buckling and waviness in the PAN matrix during heat-treatment and conversion to carbon (which is expected). For the two-layered films, this effect is most pronounced considering that the PR-L would undergo more shrinkage as compared to the CR-L. As discussed, WAXD results show the presence of a (002) peak (i.e., graphitic formation). It is expected that this graphitic structure forms in the near vicinity of the CNT. Based on previous work 145, highresolution transmission electron microscopy analysis has shown graphitic formation near the CNT at low carbonization temperature of 1100 C. It is assumed for this work that the graphitic structure formed at low temperature (i.e., 900 to 1100 C) also forms most probably at the PAN-precursor/CNT interphase or in confined PAN regions. 109

129 Figure 5.10 SEM of carbonized films. (a) PT-1 and (b1) PT-2 films carbonized at 1000 C for 5 min, (b2) and (b3) zoom in of boxed region in (b1); (c) PT-1 and (d1) PT- 2 films carbonized at 1100 C for 5 min, (d2) and (d3) zoom in of boxed region in (d1); (e) SW-1 and (f1) SW-2 films carbonized at 1100 C for 5 min, (e2) and (e3) zoom in of boxed region in (e1); and (g) PT-1 and (h1) PT-2 films carbonized at 1500 C for 5 min, (h2) and (h3) zoom in of boxed region in (b1). Figure 5.10 b2 shows the CNTs at the cross-section of CR-L region for the film that carbonized at 1000 C for 5 min. As mentioned before, this film does not show graphitic peak. On the other hand, Figure 5.10 d2 shows the CR-L region of the film carbonized at 1100 C for 5 min. This film shows graphitic peak. Based on the thicker region around CNTs, it is obvious that graphitic layers were formed around CNTs by comparing to Figure 5.10 b2. Carbonization at higher temperature changed the CNTs structure but formation of the graphitic region is still visible (Figure 5.10 h2.) 110

130 5.3.4 Graphitization Treatment WAXD curves for the graphitized films are shown in Figure All graphitized films exhibit an increase for the (002) plane at 2θ of 26 in terms of crystal size and d- spacing. This result is expected for both SW and PT materials regardless of the initial presence of graphitic structure during carbonization since heat-treatment above 2000 C leads to graphite formation and perfection. At high heat-treatment temperatures, all turbostratic and amorphous carbon forms transform to graphite. In addition, any graphitic structures formed at lower temperature continue to heal (increase in crystalline perfection) leading to narrowing (decrease of FWHM) for the (002) peak. In general, the intensity of the (002) peak for the PT-2 and SW-2 films is higher than for the PT-1 and SW-1 samples. Considering the similar nature of the components present in both film types (1 versus 2), this change in intensity may be related to enhanced development of graphite structures in PR-L region at higher temperature. This development is more apparent after microscopy analysis (Figure 5.12). The average (002) d-spacing is nm and the crystal size is ~6.0 nm. Healing of the graphitic structures formed at low temperature (900 to 1100 C) during graphitization is evidenced by the changes in the peaks show in the WAXD curves (Figures 5.9 and 5.11). This is an important finding because it confirms that the ordered structure formed at low temperature is likely graphitic rather than a hard turbostratic phase 216, 221. This work shows the evidence that the PAN/CNT interphase, as well as CNT confinement of the matrix, are both key precursor morphology features which aid early formation of graphite. Hard turbostratic phases have previously been shown to form in PAN-precursor materials at lower temperatures with the use of pressure 221. Under such conditions, the carbon form appears to exhibit a graphitic structure at low temperatures

131 but then undergoes no additional structural changes during graphitization heat-treatment processes. For this reason, such order carbon is reformed to as hard turbostratic in nature. The opposite effect is observed here. The order carbon forms found as early as 900 C continue to undergo healing as graphitization heat treatment is applied, and this confirms the phase is graphitic. These results confirm that CNT can play a role in PAN to graphitic transformations at low temperature-time combinations (~900 C). Figure 5.11 WAXD curves for the PT-1, PT-2, SW-1 and SW-2 films graphitized at 2100 C. SEM images of graphitized PT-1, PT-2 and SW-2 films previously carbonized at 1100 C for 5 min and PT-2 previously carbonized at 1000 C for 5 min are shown in Figure The PR-L in PT-2 and SW-2 which was previously carbonized at 1100 C for 5 min both show a distinct layered structure (Figures 5.12b3 and 5.12d3) in comparison to the CR-L region and PT-1 film (non-layered) (Figures 5.12a2, 5.12b2, and 5.12d2). The 112

132 layered morphology observed for the graphitized films is consistent for graphitize materials. The fact that this morphology is isolated to the PR-L region suggests that the PAN-CNT interaction enhanced by the S:NS treatment within precursor material (Figure 5.4) is mostly responsible for the eventual graphitic presence. The WAXD data presented here provides information for the whole films. However, coupled with the SEM characterization, the analysis helps to define the regions for more prominent PAN to graphite transitions. The PT-2 film, which is carbonized for the same time but lower temperature (1000 C), does not show a layer structure in the PR-L region (Figure 5.12c3). This suggests that for low-temperature graphite formation, the heat-treatment temperaturetime conditions do play a role in the film morphology. The distinct observation of SWNT in the cross-sections of PT-1 and PT-2 regions of the PT-2 graphitized film is also not obvious. This makes sense as all the material is now graphitic in nature. Regardless, layered formation is most developed and continuous in nature for the PT-2 films. The PT-2 CR-L region shows a web-like SWNT morphology confirming the porous nature of this layer and the initial lack of precursor PAN matrix. 113

133 Figure 5.12 SEM image of graphitized (a1) PT-1 film (previously carbonized at 1100 ºC for 5 min), (a2) zoom in of boxed area in a1; (b1) PT-2 film (previously carbonized at 1100 ºC for 5 min), (b2) zoom in of boxed region in CR-L area in b1, (b3) zoom in of boxed region in PR-L area in b1, (c1) PT-2 film (previously carbonized at 1000 ºC for 5 min), (c2) zoom in of boxed region in CR-L area in c1, (c3) zoom in of boxed region in PR-L area in c1; (d1) SW-2 film (previously carbonized at 1100 ºC for 5 min), (d2) zoom in of boxed region in CR-L area in d1, (d3) zoom in of boxed region in PR-L area in d Raman Spectroscopy To further understand the role of PAN interphase to graphite transformation, Raman spectroscopy analysis was performed on the PT-2 films carbonized at 900 C for 114

134 20 min and 1100 C for 5 min (Figure 5.13). D-band intensity of the PR-L region is higher than the CR-L. This is likely related to the presence of the graphitic layer edge defects 222. Since there is more PAN in this PR-L region, carbonization transformations during pyrolysis likely occur here as compared to the CR-L regions. CNT structures within the overall films also undergo annealing during processing. For this reason, the CR-L is expected to exhibit a lower D-band intensity. In general, the D-band to G-band intensity ratio of the PR-L and CR-L regions for the films carbonized at 1100 C for 5 min and 900 C for 20 min are 2.5 and 2, respectively. The higher ratio of the PR-L and CR-L intensity for film carbonized at 1100 C for 5 min indicates more graphitic defects are present in the PR-L regions. This data is in accordance with WAXD data (Figure 5.8a). As mentioned, WAXD data showed that films carbonized at 1100 C for 5 min had more hexagonal structure with ACBACB stacking coupled with rhombohedral stacking in comparison to films treated at 900 C for 20 min. Films carbonized at 900 C for 20 min show more hexagonal ABAB stacking with little rhombohedral presence. The coupling of hexagonal and rhombohedral structures leads to more defects present in the graphitic planes, and this is also confirmed through the Raman analysis. Raman spectroscopy combined with the WAXD and microscopy analysis all confirm that the PR-L exhibits a distinctive graphitic layered morphology stemming from the initially ordered morphology found in the precursor PR-L. The formation of the twolayered precursor also shows the selectivity of PAN-CNT interactions and its influence on interphase growth. The analysis of these carbonized materials clearly shows the linkage between precursor morphology and final carbonized film structure. 115

135 Overall, these works show that CNT do play a role in indicating ordered PAN formation and this structure is important for forming order carbon at low temperatures during pyrolysis. The work also confirms that confinement of the polymer is also important since its inherent shrinkage is not helpful for carbon formation. When comparing both effects of interphase formation and confinement through analysis of the two-layered composite films, it is clear that interphase formation is most influential for early graphitic formation and subsequent healing during graphitization (Figure 5.12). The type of CNT used in these materials is also a key parameter and shows that polymer-nanotube interactions leading to precursor interphase formation are not equal. Therefore, multiple parameters should be carefully controlled during composite precursor formation, stabilization, carbonization, and graphitization, as these PAN/CNT materials are explored for use as the next-generation carbon precursor. The advantage of using these systems is the potential ability to control morphology throughout all processing steps, as shown in this work. 116

136 Figure 5.13 The Raman spectra of PT-2 films carbonized at 1100 C for 5 min and 900 C for 20 min. 5.4 Conclusion PAN/CNT films containing 50 wt% CNT were made by enhancing polymer crystallization conditions and processing parameters. Different PR-L and CR-L regions were made by changing S:NS ratios before filtration processing. WAXD analysis reveals that PR-L regions consist of highly ordered/crystalline PAN. All films were stabilized at 250 and 300 C, where 300 C was found to be most favorable. Stabilized films were carbonized at temperatures ranging from 900 to 1500 C. Carbonization times were also 117

137 viewed. WAXD and SEM analysis were used to understand the effect of stabilization temperature and carbonization conditions. It was found that graphitic structure formed at specific temperature/time conditions (i.e., 40 minutes at 900 C or 20 minutes at 1000 C). This low temperature graphitic structure can be directly correlated to the PAN interphase formation as well as PAN confinement. Both are observed to be equally effective in allowing for early onset of graphitic forms. High temperature graphitization confirms that graphite and not a hard turbostratic phase is formed in this study. High temperature treatment also confirms that interphase PAN is more important in inducing uniform layered graphitic forms throughout the material. Raman analysis confirms the presence of layered graphitic features in the PR-L as compared to the CR-L for the two-layered precursors. This showed the selectivity of PAN to only prefer interfacial interaction with smaller bundles. 118

138 CHAPTER 6. EFFECT OF LOW-TEMPERATURE GRAPHITIZED STRUCTURE ON ELECTRICAL PROPERTIES OF PAN/CNT MATERIALS 6.1 Introduction The variation of carbonaceous structures present in carbon fibers (i.e., amorphous, turbostratic, and graphitic carbon) all have a significant influence on their morphology and resultant electrical, mechanical, and thermal properties. Increasing the crystalline order (more graphitic) and orientation for the carbon form typically leads to improvement in the axial properties of the carbon fiber 27, 157. Therefore, there have been several recent studies focused on understanding the role of controlling precursor polymer structures to influence and restrict the variation of carbonaceous structures formed upon pyrolysis. Several research studies are focused on enhancing the properties of PAN since PAN is the main precursor for carbon fiber production , 116, 117, 144, 213. The addition of CNT was found to increase the order of PAN chain molecules and reluctant carbon fiber properties 145, 156, 158, 219, 27. Carbon fibers made using PAN/CNT precursors show graphitic structure formation at the PAN-CNT interface at a relatively low temperature (i.e., 1100 C) 27, 149, 156, 158, 159. Studies from this dissertation work have recently shown that as early as 900 C graphite formation is possible at the interface of PAN-CNT 223. This work is outlined in Chapter 5. Carbon structures show varying electrical conductivity properties. Therefore, carbon structure variations due to atomic bonding and defects affect electrical properties of the 119

139 fiber or carbonized material. For example, graphite with a sp 2 atom-bonding state is an electrical conductor, while a diamond tetrahedral structure with sp 3 atom-bonding state is insulator 224. The electrical conductivity of graphite within the basal plane ranges from 2000 to 4000 S cm -1, and normal to the basal plane this value is 3.33 S cm Electrical conductivity of amorphous carbon ranges from 12.5 to 20 S cm , 227. CNTs also show very good electrical conductivity properties. The electrical conductivity of individual SWNT was reported to be as high as S cm , 229, and individual MWNT was reported to as high as S cm , 231. Forming defects and nanotube bundles decrease the electrical conductivity of CNTs Bundles of SWNT show electrical conductivity of S cm , 234 while electrical conductivity of MWNT bundles are one to two degrees lower in magnitude 235, 236. BP made from parallel SWNT show electrical conductivity of 200 S cm The electrical conductivity of carbon fibers made from PAN polymer precursors is reported at 666 S cm or S cm , while electrical conductivity of carbon fibers made from pitch precursors is reported to range from 4000 to 4545 S cm Increase in the order of carbon structure (more graphitic) increases the electrical properties of carbon fibers 238. Disruption in planar π- conjugated structure of CNTs, by formation of sp 3 bonds, will affect electron transport in CNTs and therefore lead to a sharp decrease in the conductivity An increase in tensile strength properties is expected to occur if the presence of oriented graphitic regions can be increased throughout the fibers, while minimizing defective carbonaceous regions. While it has been shown that graphitic structure can form at 900 C at the interphase of PAN-CNT 223, studies focused on understanding the mechanism behind this conversion process and subsequent structure are still needed. In this 120

140 study, PAN/CNT composites were prepared to purposely isolate interphase regions in the composite 220, 223. By preparing such precursor materials, PAN conversion during carbonization heat-treatment could be studied specifically. This study shows that selective CNT can influence both early stage of graphitic formation as well as enhance properties after carbonization (i.e., electrical conductivity). While, it has been shown that different CNT has varying effects on the early onset of graphitic phase formation 145, 223, this study shows that, in one type of CNT, the diameter also affect the graphitic phase formation. Electrical conductivity enhancement in the reported carbonized films can be related to CNT diameter as well as CNT structure integrity and introduction of graphite after carbonization. This work is performed to confirm the value of introducing order PAN interphase regions to the precursor to control graphitic forms in the final carbon material. This work also shows the role of polymer confinement effects in early carbonaceous and graphitic formation. 6.2 Experimental Section Materials The PAN used in this work is a poly(acrylonitrile-co-methacrylic acid) random copolymer with methacrylic acid content of 4 wt% (Mw ~513,000 g mol -1 ), obtained from Exlan Co. Japan. CNT material used is PT (purified SWNT, ~94.5%, Continental Carbon Nanotechnologies, Inc.). DMF is obtained from Sigma Aldrich Solution Processing and Film Fabrication Dispersion Preparation 121

141 The materials used in this study are PAN/CNT composite films. For hpbp fabrication, PAN powders were first dissolved in 90 C DMF at a concentration of 250 mg L -1. CNTs of equal amount were then dispersed in the polymer solution for 24 hrs via a bath sonicator (Fisher FS30, frequency 43 khz, power 150 W). For control BP fabrication, CNTs were dispersed in DMF using similar sonication conditions BP and hpbp Films Preparation After sonication, the PAN/CNT dispersion was subjected to a solution-based shear crystallization process by stirring the dispersion at 90 C. Simultaneously vacuum distillation was applied to remove half the volume of solvent at a controlled time and temperature. Subsequently, this overall system was cooled down to room temperature (~25 C). A non-solvent for PAN was added into dispersion at various S:NS ratios. The two ratios chosen in this work were 7:1 and 1:2, respectively. This S:NS treatment during solution processing of the PAN/CNT films help to isolate regions, where PAN has enhanced interfacial interaction with the CNT. The isolation of this interphase region is important for studying subsequent carbonization process effects. The final PAN/CNT dispersions were then filtered through a nylon membrane (0.45 µm pore size obtained from Millipore) to form the PAN/CNT hpbp. The films made using PT-SWNT with S:NS ratios of 7:1 and 1:2, respectively, were named PT-1 and PT-2, accordingly. A free-standing hpbp was removed from the filter paper after drying in a vacuum oven and further characterized to understand the effect of processing on the structure of the hybrid films. Control CNT BPs were also fabricated using a similar filtration process for comparison and named PT. 122

142 Heat-treatment Processes A Lindberg/Blue M Mini-Mite tube furnace equipped with a quartz tube (diameter 1 inch, obtained from Quartz Scientific Inc.) was used for heat-treatment to conduct both stabilization and carbonization procedures. Two flow meters were used to control the inlet and outlet gas flow through the furnace. The entire system was sealed and the outlet gas flow was directly released into a venting system. The film samples were held in compression using a customized sample holder throughout both heat-treatment processes. For stabilization (in air), the temperature was (i) ramped up at 1 C min -1 from room temperature (~25 C) to 300 C; (ii) maintained isothermally for 10 hrs; and (iii) gradually decreased to room temperature. For carbonization (in argon), the temperature was (i) ramped up at 5 C min -1 from room temperature to 900 or 1100 C; (ii) maintained isothermal for 20 or 40 min at 900 C, and 5 min at 1100 C; and (iii) gradually decreased to room temperature (~25 C). In both processes, the required air or argon gas flow was maintained constant at 3000 ccm until the chamber was cooled down to room temperature (~25 C). Figure 6.1 summarizes the time-temperature profile for the stabilization and carbonization procedures Sample Characterization Morphology characterization was performed using a Zeiss Supra 25 field emission SEM (operating voltage 5 kv) houses s PGT SHARA Peltier cooled energy-dispersive X- ray (EDX) spectroscopy for elemental composition analysis. All film samples were fractured and mounted to a 90 pin stub with the fractured end facing up for SEM observation. Precursor (non-carbonized) samples were coated with a thin gold/palladium 123

143 layer (15-20 nm) for image purposes using a Gatan high-resolution ion beam coater. WAXD was performed on a Rigaku RAPID II equipped with a curved detector XRD system with a 3 kw sealed tube X-ray source (operating voltage 40 kv and current 30 ma). XRD curve fitting and analysis was performed using software PDXL 2 (version ) and 2DP (version ). Raman spectroscopy was conducted on a Jobin Yvon LabRam HR800 (laser wavelength 532 nm). The electrical conductivity measurements were carried out on a four-point probe instrument (Tektronix DMM 4050) and the thicknesses of films were measured using an optical microscope (Olympus BX 51). Figure 6.1 Graphical representation for the heat-treatment procedures used in this work for (a) stabilization and (b) carbonization of the samples. 6.3 Results and Discussion Structural Morphology of the Films For this study, PAN-CNT interactions are isolated by addition of a non-solvent during filtration process. PT-1 and PT-2 hpbps are fabricated using a 7:1 and 2:1, S:NS ratio, respectively. PT BP is fabricated to better understand the effect of CNT and PAN during heat-treatment of hpbps. SEM images of cross-sections and surface regions of 124

144 fabricated precursor, carbonized at 900 C for 20 min, and carbonized at 1100 C for 5 min of PT, PT-1, and PT-2 films are shown in Figure 6.2. PT films show the representative morphology for the inherent dispersion of CNTs. The cross-section and surface regions of the PT precursor films are shown in Figures 6.2a1 and 6.2b1. During the carbonization process, CNTs coalesce, shorten and straighten (Figures 6.2a2, 6.2a3, 6.2b2, and 6.2b3) this combination results in less overlap of the tubes and changes the dispersion quality such that defects are introduced. For this reason, the effective property contributions for the CNTs to the electrical conductivity is diminished as percolation pathways change 242. Film carbonized at 900 C shows coalescence and shortening (Figures 6.2a2 and 6.2b2). The coalescence and shortening increased for films carbonized at 1100 C (Figures 6.2a3, and 6.2b3). Due to the coalescence and shortening for the tubes, cross-sectional views of film carbonized at 1100 C show mostly carbon domains and the CNTs distribution and structure is not obvious. PT-1 precursor film cross-sections and surface regions show a heterogeneous structure, where the CNTs are randomly distributed throughout the film within the PAN matrix (Figure 6.2c1 and 6.2d1). PT-1 films carbonized at 900 C does not show coalescence and shortening like the PT film (Figures 6.2c2 and 6.2d2). PAN matrix and PAN-CNT interactions prevent these structural changes and tube damage similar to what is seen in PT precursor film. For this reason, the surrounding PAN matrix helps to maintain the integrity of CNTs. The PT-1 film carbonized at 1100 C (Figures 6.2c3 and 6.2d3) shows major structural changes in comparison to same films carbonized at 900 C (Figures 6.2c2 and 6.2d2). On one hand, coalescence and shortening of CNTs does occur. However, on the other hand, some additional carbon-like particles also form. Similar to the PT films. 125

145 cross-sectional views of the film carbonized at 1100 C exhibit carbon domains rather than showing a clear distinction of the CNTs distribution. PT-2 precursor films show a two-layered structure; a PAN-rich layer (PR-L) which is dominated by PAN and contains a lower concentration of CNTs (Figure 6.2e1 and 6.2f1), and a CNT-rich layer (CR-L) that contains little polymer but a very high concentration of CNT (Figure 6.2e4 and 6.2f4) 220, 223. From a morphological point of view, the structure of the CR-L in the PT-2 sample is similar to the overall structure of the PT-1 film, while the PR-L is much denser and shows even distributions of the CNT. The phase separated films were also introduced and discussed in Chapter 5. CR-L regions of the PT-2 film carbonized at 900 C (Figures 6.2e2 and 6.2f2) shows similar configuration to the PT-1 films. However, films carbonized at 1100 C shows a different trend. Compared to the PT and PT-1 films carbonized at 1100 C, CNTs dispersion in the PT-2 films are still visible. Observation of this domain of CNTs means that they maintain their structural integrity in these films. Also, fewer carbon domains are visible in the PT-2 films. This confirms that the CNTs in the PT and PT-1 films transition to carbon domains during heat treatment (Figures 6.2e3 and 6.2f3). This result also confirms that the interaction of PAN-CNT within CR-L region of PT-2 films is different than for the PT-1 film as expected 220, 223. As shown in Figures 6.2e5 and 6.2e6, after carbonization at 900 C and 1100 C, PR-L layer of PT-2 films shows a different morphology. Here the morphology shows an enhancement of graphitization in this region 223. The graphitization in this region also increases at higher temperature. Surface views of this region (Figures 6.2f5 and 6.2f6) also shows that the CNTs also hold their integrity as well as their network structure which may contribute to the film properties. 126

146 Figure 6.2 SEM images of the precursor, carbonized at 900 C for 20 min, and carbonized at 1100 C for 5 min:(a) PT cross-section, (b) PT surface region, (c) PT-1 cross-section, (d) PT-1 surface region, (e) PT-2 cross-section and, (f) PT-2 surface region. Table 6.1 Weight loss percentage of films after each heat treatment. Stabilization Carbonization at Carbonization Carbonization at 900 C/20 min at 900 C/40 min 1100 C/5 min PT PT PT WAXD Analysis of the Films WAXD analysis was performed in order to investigate the actual structural features of the films before and after heat-treatments. The WAXD micrographs graphs for the PT, PT-1 and PT-2 films are shown in Figure 6.3a. It is observed that the PT-2 film show a narrow and predominate peak at 2θ of 16.7º associated with crystalline PAN (dotted box 127

147 in Figure 6.3a). The broad peak observed for the PT-1 film is consistent with a more amorphous PAN morphology (dotted box in Figure 6.3a). This WAXD data suggest that for the single layered films the PAN retains an amorphous structure, whereas in the twolayered films it is more crystalline. Considering that the majority of PAN in the two-layered films are isolated in the PR-L, this data also suggests that the PAN in this region is highly ordered. PT film consists only CNTs and, therefore, no PAN peak observed. Figure 6.3b shows the WAXD spectra of PT, PT-1 and PT-2 films after stabilization heat-treatment. At this stage the PAN structure is converted to a stable ladder structure by this oxidation stabilization heat-treatment 124, 223. Therefore, the PAN peak is no longer observed after stabilization heat-treatment. Figure 6.3 WAXD spectra of the PT, PT-1, and PT-2 (a) precursor and (b) stabilized films. Carbonization heat-treatment is the next step in the carbon fiber production process to convert the ladder structure forms to turbostratic or graphite-like layer structures 144. Figure 6.4 shows the WAXD spectra of PT, PT-1 and PT-2 films after carbonization heattreatment at 900 C for 20 and 40 min and 1100 C for 5 min. It has been shown that by 128

148 selecting the right temperature and adequate time graphite structure can form at the interface of CNT and PAN 223. The (002) plane peaks at 2θ of ~26 in Figure 6.4 shows that graphite layers are formed in the composite fibers. The same peak is also seen for PT samples. However, it is a broad peak for PT samples carbonized at 900 C for 20 min. This peak could be related to the conversion of SWNTs to MWNTs or layer carbon structures by thermal coalescence of HiPco SWNTs during heat-treatment which has shown before for higher temperatures 242. The 10 hrs of stabilization heat-treatment in air used in this study may enhance the coalescence of SWNTs during carbonization. For this reason, coalescence of SWNTs happens at a relatively lower temperature in comparison with previous research 242. Iron oxide peaks appear at 1100 C. The peaks observed at 2θ of 35.71, and represent the peaks of (311), (400) and (531) diffractions of Fe3O4 243, respectively and 2θ of 35.71, and represent the peaks of (111), (200) and (220) diffractions of FeO 244, respectively. The (311) plane of Fe3O4 and (111) plane of FeO at 2θ of is also observed for PT films carbonized at 900 C for 40 min. The formation of iron oxide at a lower temperature for PT film suggests the shielding effect of PAN in the composite film which hinders the oxidation of iron particles. Table 6.2 shows the 2θ and (002) d-spacing values of carbonized films. Based on these data, 2θ and (002) d-spacing values of PT-1 and PT-2 are very close. Using EDX data (Table 6.4), diffusion of Fe particles is found to differ in these films. However, this diffusion likely occurs due to oxidation at higher temperature. This diffusion process may play a rule in the formation of a more perfect (i.e., with less defect) graphitic structure in the PT-2 films. This structure which contains less defects, also will be responsible for the increase in 129

149 electrical conductivity properties observed for the PT-2 films carbonized at 1100 C for 5 min (Figure 6.6). Additional discussion of the EDX data is found in Section Figure 6.4 WAXD spectra of the PT, PT-1, and PT-2 carbonized films. 130

150 Table 6.2 The 2θ and (002) d-spacing values of carbonized films. Carbonization Sample Condition PT PT-1 PT-2 Temperature ( C) Time (min) 2θ ( ) d (nm) Size (nm) 2θ ( ) d (nm) Size (nm) 2θ ( ) d (nm) Size (nm) Electrical Conductivity of the Films Electrical conductivity was measured and data are shown in Figure 6.5 for precursor and stabilized films and Figure 6.6 for carbonized film at different time and temperatures. Electrical conductivity of the PT, PT-1, PT-2 (CR-L), and PT-2 (PR-L) precursors films are , 22.06, 32.66, and 2.20 S cm -1, respectively. For PT and PT-1 electrical conductivity values of top and back surfaces are similar as expected since there are no layered structure like that observed for the PT-2 films. Comparatively, the backsurface region of the PT-2 films (PR-L) contains more PAN and shows the lower value (~1/15) in comparison to top layer surface region (CR-L) which contains more CNT. In general, electrical conductivity of precursor composite films are lower than the PT film due to lower electrical conductivity of PAN matrix. After stabilization, all films show an improvement in the electrical conductivity data. The electrical conductivity of the stabilized PT, PT-1, PT-2 (CR-L), and PT-2 (PR- L) films surface region was increased to , 85.60, 66.64, and S cm -1, respectively. The electrical conductivity of CNT films increase as stabilization in air (under oxygen) anneal the CNT 239 and likely degrade the amorphous carbon regions 245. Formation of a carboxyl group ( COOH) also increases the electrical conductivity of CNT film 246. Carboxylic groups are acceptor dopant groups which maintain the π conjugated 131

151 systems of CNT and therefore increase the conductivity 239, 247. For a composite film increase in electrical conductivity could be due to increase in electrical conductivity of CNT network as well as the conversion of PAN to the ladder structure. Figure 6.5 Electrical conductivity data of the precursor and stabilized films. After carbonization, a sharp drop in electrical conductivity is observed for the PT films (declined from to S cm -1 ). During carbonization, carbon and carboxyl group in CNTs are remove by forming CO and CO2 and disrupt the structure of CNTs. Disruption in the planar π-conjugated structure of CNTs, by the formation of sp 3 bonds, will affect electron transport in CNTs and therefore decreases the conductivity Straightening of CNTs and length and diameter shortening of them during carbonization (Figure 6.2) also decrease the electrical conductivity of PT film 248. These changes are also discussed later by Raman results in Figure 6.7. The electrical conductivity of PT film 132

152 slightly increases for the films that carbonized at a higher temperature (i.e., 900 vs C) or longer time (i.e., 20 vs. 40 min) as shown in Figure 6.6. For PT-1 films a sharp drop in electrical conductivity does not happen and values are only slightly lower in comparison to their stabilized films (declined from to S cm -1 ). These values are still higher than the electrical conductivity values of PT films. The shielding effect of the PAN matrix decreases the disruption in CNTs structure during heat treatment. Formation of some graphitic structure in the vicinity of CNTs due to the conversion of PAN to graphite also plays a role in the higher value of electrical conductivity of PT-1 films in comparison to PT film. Conversion of the PAN at the CNTs interface to graphite structure and decreasing of sp 3 bond formation increases the electron transport. Electrical conductivity of PT-1 film carbonized at 1100 C is lower than the film carbonized at 900 C. As shown in WAXD data in Figure 6.4, iron particles oxidation happens for the film carbonized at 1100 C. For this reason, electrical conductivity of film carbonized at 1100 C for 5 min is lower since iron oxide particles are less conductive than iron particles. The electrical conductivity values of carbonized PT-2 films are generally higher than PT and PT-1 films. Large improvement in the electrical conductivity is observed for PR-L region of the PT-2 after carbonization especially for the films carbonized at 1100 C for 5 min. That film shows ~66 and ~14 times improvement in comparison to the precursor and stabilized films, respectively. This improvement comes from the selective PAN-CNT area formed in the PR-L and showed previously 220. Raman results (discussed later) also confirm that PR-L region contains more SWNT in comparison to CR-L region or PT-1 films due to the selective method that used for these films. As shown before, electrical 133

153 conductivity of individual SWNT is two orders of magnitude higher than bundled SWNTs Electrical conductivity values of CR-L region also increased especially for the films carbonized at 1100 C for 5 min. For PT-2 films values of carbonized film for PR-L and CR-L layers are very close to each other. It could be due to the higher graphitic phase in PR-L region which provide a pass for electron transfer even when the measurement was done for CR-L region. As discussed previously, based on WAXD data in Figure 6.4, iron particles oxidize for the film carbonized at 1100 C. Despite PT-1 films, the electrical conductivity of PT-2 films carbonized at 1100 C was increased in comparison to films carbonized at 900 C. This increase is related to the movement of iron oxide particles to the interface of PR-L and CR- L region of the film. Iron oxide transfer from the surface decrease the defect in those regions and improve the graphitic layers of the region closer to the surface which will increase the electrical conductivity. EDS results (Table 6.4) confirm such movements and is explained later. 134

154 Figure 6.6 Electrical conductivity data of carbonized films Raman Analysis of the Films To better understand the changes in the structure of films, Raman measurement was conducted on precursor and the films carbonized at 1100 C for 5 min. Figure 6.7a gives Raman spectra of radial breathing mode (RBM) band and Figure 6.7b gives Raman spectra of D-band and G-bands. It has been shown that SWNT bundles shift the RBM, G-band, and D-band to the left for nm laser in comparison to individual SWNT 249. By comparing Raman result, it is seen that PR-L region acts selectivity and contains more individual (not bundled) SWNTs. D-band and G-band peak position for PT-2 (PR-L) is ~10 and ~6 cm -1 higher than PT, PT-1, and PT-2 (CR-L), respectively. This result is in accordance with our previous results 220. The separation of bundles in PT-2 films can later affect the electrical conductivity. PR-L region of PT-2 films shows higher electrical conductivity value due to this selective separation (Figure 6.6). 135

155 After carbonization, the intensity of RBM for PT samples decreases significantly. As shown in SEM images (Figure 6.2) this is due to changes in the structure of tubes during heat-treatments. During heat-treatment, CNTs coalesce, shorten, straighten and therefore lose some of the inherent tube properties. RBM peaks position is related to the radial of SWNTs 249, 250. In general, after carbonization, the position of RBM bands shifted to the left (blue shift). Which means enlargement of the tube diameter 249, 250. This finding is in accordance with a previous observation 251. By comparing IG-band/ID-band ratio, the degree of graphitization can be understood. For PT and PT-1 films this ratio decrease after carbonization at 1100 C, but for PT-2 films increases for PR-L and CR-L regions (Table 6.3). This result is in accordance with electrical conductivity data which shows an improvement in electrical conductivity for PT- 2 films in both layers. Despite the layered structure in the PT-2 films, IG-band/ID-band ratio of PT-1 and PT films decreases in comparison with precursor films which confirm less development of layers structure and decreases in the electrical conductivity. By further looking at G-band data of carbonized film we can see that the intensity of G-band peak is higher for carbonized PT-2 film (both layers) in comparison to PT-1 and PT films. Higher the intensity means metallic nanotube and lower means semi-conductive nanotubes. This also can be another reason of lower electrical conductivity for PT and PT- 1 films in comparison to PT-2 films. 136

156 Figure 6.7 Raman spectra of precursor and carbonized film at 1100 C for 5 min. (a) RBM and (b) D-band and G-band 137

157 Table 6.3 Raman data of D-band and G-band of precursor and carbonized film at 1100 C for 5 min. Precursor Carbonized at 1100 C for 5 min D-band G-band I G-band/I D-band D-band G-band I G-band/I D-band PT PT PT-2 (CR-L) PT-2 (PR-L) EDX Analysis of the Films EDX analysis for carbonized films is shown in Table 6.4. Based on this analysis the Fe diffusion is tracked and is observed to migrate toward the interfacial region in the two layered films. Both surface and cross-sectional analysis is used to understand this diffusion process. The data supports the evidence that Fe particles move toward the interface region in the films from both CR-L and PR-L layers. The reduced concentration of Fe in the PR-L region may aid the formation of highly ordered graphitic structures at higher temperature as the effect of Fe oxidation is reduced. This migration of Fe to the interface region may be one reason for the increase in the order of layered carbon in PT-2 film. To this end, there is an increase in electrical conductivity of PT-2 film in comparison to PT-1 film. 138

158 Table 6.4 EDX analysis of films. Surface Crosssection Precursor Carbonized at 1100 C for 5 min C (wt %) O (wt %) Fe (wt %) C (wt %) O (wt %) Fe (wt %) PT PT PT-2 (CR-L) PT-2 (PR-L) PT-2 (Layers Interface) N/A N/A N/A PT PT PT-2 (CR-L) PT-2 (PR-L) PT-2 (Layers Interface) N/A N/A N/A Conclusion PAN/CNT films containing 50 wt% CNT were made by enhancing polymer crystallization conditions and processing parameters. Different PR-L and CR-L regions were made by changing S:NS ratios before using a filtration process to form films. Raman analysis was used to study the distribution of CNTs and the effect of heat treatment on resultant changes in the structure. Raman analysis confirms that the PR-L regions contain smaller bundle size CNTs. Electrical conductivity of films also was measured. The electrical conductivity results suggest that PT-2 film carbonized at 1100 C show the best graphitic layered structure due to enhancement from interfacial region in PR-L region. EDX data describe the rule of diffusion of Fe particles during heat treatment and resultant improvement in electrical conductivity of the two-layered films due to improvement in layers structure by decreasing the defects. 139

159 CHAPTER 7. SUMMARY AND RECOMMENDATION FOR FUTURE WORKS 7.1 Summaries The property-structure importance for the existence of the interfacial and interphase region within PNCs is discussed in this thesis work. Proper interfacial control and development may ensure excellent interaction and property transfer between the filler and polymer matrix in addition to improvement of multi-functional properties of in PNCs. Two specific PNC systems, PE/BN and PAN/CNT composites, were selected to investigate their mechanical performance and thermal and electrical conductivity properties, respectively. For these systems, it was found that the interfacial region structure is directly related to the enhancement of the subsequent multi-functional properties. For PE/BN composites a good interfacial structure was introduced by using shear crystallization technique. By understanding the process for forming this enhanced interface, a gel-spinning process was designed to scale-up of fibers made from PE/BN. These fibers were subsequently studied and characterized to understand the relationship between the property changes and the presence of interfacial interactions. These results are outlined in Chapter 2 thru Chapter 4. For PAN/CNT composites stabilization, carbonization and graphitization heat treatments were used to understand the effect of enhanced interfacial region between PAN and CNT on carbon fiber production. The interface and interphase regions between PAN and the CNT were selectively isolated. Using these systems, the formation of graphitic in 140

160 these regions were tracked as function of temperature and time. In addition, some reaction effects are initially studied to understand their role in structural formation of carbon. These results show a direct linkage between ordered interphase structure and early graphite formation. These results are outlined in Chapter 5 and Chapter Recommendations Based on the studies performed in this work and the conclusions drawn from the results, the following recommendations are made for future research efforts that may extend upon the findings outlined in this dissertation work. Fabrication or use of an extrusion system for PE gel spinning will increase the control over the gel spinning process in comparison with the efforts that used here. Further computational and experimental study are required on nbn particles to understand the effect of interaction BNNS and PE. This study can be also extended to BNNT. Further development is needed to accurately measure the axial thermal conductivity properties for small-diameter composite fibers. Initial results for this work suggest that PE/BN composites are excellent candidates for thermally conductive materials. Heat treatment optimization studies can be extended to the PAN/CNT fibers that are made to incorporate enhanced interphase PAN-CNT regions (Appendix B) to ensure the formation of highly graphitic formation. Graphitic interphase in carbon fibers may lead to superior mechanical and electrical properties in comparison to commercially available carbon fibers. This propertystructure relationship requires further analysis. 141

161 APPENDIX A. SAMPLE PREPRATION FOR THERMAL CONDUCTIVITY MESURMENTS In order to prepare gel-spun samples for thermal conductivity measurement, bundle of fibers were mounted into the epoxy resin. To avoid heating of the fibers during mounting with epoxy, two-hour PELCO epoxy mount kit were used to cold mount the fibers. Later, the cross-sections of the fibers polished by 1, 0.3, and 0.05 μm alumina powder. The final cross-sections of the bundles for undrawn and drawn PE fibers are shown in Figures A.1a and A.1b, respectively. The cross-sections should work as mirror for laser beam in order to be able to measure the thermal conductivity of the fibers. For obtaining the least error in the measurement, the less roughness is necessary. Microtoming the crosssection may be a better option for samples if it is available. Figure A.1 Cross-sections of (a) undrawn PE and (b) drawn PE fibers in epoxy resin (arrows shows the fibers). 142

162 APPENDIX B. ELECTRICAL CONDUCTIVITY OF CARBON FILMS FOR LONGER CARBONIZATION TIME B.1 Introduction It was shown in Chapter 6 that the film carbonized for 5 min at a relatively higher temperature (i.e., 1100 C) shows higher electrical conductivity. In this Appendix films were carbonized for a longer duration times (i.e., 20, 40, and 60 min) at 1100 C to understand the effect of time on the film structure and subsequent electrical conductivity properties. The changes in the crystal structure also studied by using WAXD analysis. B.2 Experimental Section B.2.1 Materials The PAN used in this work is a poly(acrylonitrile-co-methacrylic acid) random copolymer with methacrylic acid content of 4 wt% (Mw ~513,000 g mol -1 ), obtained from Exlan Co. Japan. CNT material used is PT (purified SWNT, ~94.5%, Continental Carbon Nanotechnologies, Inc.). DMF is obtained from Sigma Aldrich. B.2.2 Solution Processing and Fiber Fabrication B Dispersion Preparation The materials used in this study are PAN/CNT composite films. For hpbp fabrication, PAN powders were first dissolved in 90 C DMF at a concentration of 250 mg 143

163 L -1. CNTs of equal amount were then dispersed in the polymer solution for 24 hrs via a bath sonicator (Fisher FS30, frequency 43 khz, power 150 W). B hpbp Films Preparation After sonication, the PAN/CNT dispersion was subjected to a solution-based shear crystallization process by stirring the dispersion at 90 C. Simultaneously vacuum distillation was applied to remove half the volume of solvent at a controlled time and temperature. Subsequently, this overall system was cooled down to room temperature (~25 C). A non-solvent for PAN was added into dispersion at a various S:NS ratios. Only PAN/CNT precursor films using a 1:2 ratio were chosen in this study. The final PAN/CNT dispersions were then filtered through a nylon membrane (0.45 µm pore size obtained from Millipore) to form the PAN/CNT hpbp. B Heat-treatment Processes A Lindberg/Blue M Mini-Mite tube furnace equipped with a quartz tube (diameter 1 inch, obtained from Quartz Scientific Inc.) was used for heat-treatment to conduct both stabilization and carbonization procedures. Two flow meters were used to control the inlet and outlet gas flow through the furnace. The entire system was sealed and the outlet gas flow was directly released into a venting system. The film samples were held in compression using a customized sample holder throughout both heat-treatment processes. For stabilization (in air), the temperature was (i) ramped up at 1 C min -1 from room temperature (~25 C) to 300 C; (ii) maintained isothermally for 10 hrs; and (iii) gradually decreased to room temperature. For carbonization (in argon), the temperature was (i) ramped up at 5 C min -1 from room temperature to 1100 C; (ii) maintained 144

164 isothermal for 5, 20, 40 or 60 min at 1100 C; and (iii) gradually decreased to room temperature (~25 C). In both processes, the required air or argon gas flow was maintained constant at 3000 ccm until the chamber was cooled down to room temperature (~25 C). B Sample Characterization WAXD was performed on a Rigaku RAPID II equipped with a curved detector XRD system with a 3 kw sealed tube X-ray source (operating voltage 40 kv and current 30 ma). XRD curve fitting and analysis was performed using software PDXL 2 (version ) and 2DP (version ). The electrical conductivity measurements were carried out on a four-point probe instrument (Tektronix DMM 4050) and the thicknesses of films were measured using an optical microscope (Olympus BX51). B.3 Results and Discussion B.3.1 Electrical Conductivity of the Films Electrical conductivity properties of films carbonized at 1100 C and various times (i.e., 5, 20, 40 and 60 min) are shown in Figure B.1. By increasing the carbonization time, electrical conductivity of layers increases. These increases are related to the increases in the order of the structure as mentioned in Chapter 5 and Chapter

165 Figure B.1 Electrical conductivity of the carbonized film at 1100 C and different time (i.e., 5, 20, 40 and 60 min). B.3.2 WAXD Study of the Films WAXD spectra of the films that carbonized at different time are shown in Figure B.2. The values of d-spacing and crystal size for (002) graphitic peak at 2θ of ~26 are provided for all films in Table B.1. By increasing the carbonization time, 2θ values are also increased. Therefore, d-spacing is decreasing. This data shows that increasing carbonization holding time enhances the graphitic and ordered structure formation. However, based on the electrical conductivity data and WAXD data, improvement in the ordered structure are not significant in comparison to the effect of the carbonization temperature (Sections and 6.3.3). 146

166 Figure B.2 WAXD spectra for the PT-2 films carbonized for 5, 20, 40, and 60 min. Table B.1 2θ and (002) d-spacing values of carbonized films at 1100 C for 5, 20, 40, and 60 min. 5 min 20 min 40 min 60 min 2θ d-spacing Size (nm)

167 APPENDIX C. FABRICATION OF CARBON FIBERS FROM ENHANCED INTERFACIAL REGION PRECURSOR C.1 Introduction In Chapter 5 and Chapter 6 of this dissertation, effect of different heat treatment condition on micro- and nano--structure of the films that were made from the precursors, which contained enhanced interfacial region, were studied. In this Appendix, precursor fibers made to enhance similar PAN-CNT interfacial regions were examined. The fibers were subjected to stabilization and carbonization heat treatment to comparing their structure with the available commercial carbon fibers by using WAXD data. C.2 Experimental Section C.2.1 Materials The PAN used in this work is a poly(acrylonitrile-co-methacrylic acid) random copolymer with methacrylic acid content of 4 wt% (Mw ~513,000 g mol -1 ), obtained from Exlan Co. Japan. CNT material used is PT (purified SWNT, ~94.5%, Continental Carbon Nanotechnologies, Inc.). DMF is obtained from Sigma Aldrich. C.2.2 Solution Processing and Fiber Fabrication C Dispersion Preparation (Performed by Heng Li) The materials used in this study are PAN/CNT composite fiber. For hybrid polymer/cnt fabrication, PAN powders were first dissolved in 90 C DMF at a 148

168 concentration of 250 mg L -1. CNTs of equal amount were then dispersed in the polymer solution for 24 hrs via a bath sonicator (Fisher FS30, frequency 43 khz, power 150 W). C PAN and PAN/CNT Fiber Preparation (Performed by Heng Li) After sonication, the PAN/CNT dispersion was subjected to a solution-based shear crystallization process by stirring the dispersion at 90 C. Simultaneously vacuum distillation was applied to remove half the volume of solvent at a controlled time and temperature. Subsequently, this overall system was cooled down to room temperature (~25 C). A non-solvent for PAN was added into dispersion at various S:NS ratios. Just 1:2 ratios were chosen in this work. This S:NS treatment during solution processing of the PAN/CNT fiber help to isolate regions, where PAN has enhanced interfacial interaction with the CNT. The isolation of this interphase region is important for studying subsequent carbonization process effects. The final PAN/CNT dispersions were then filtered through a nylon membrane (0.45 µm pore size obtained from Millipore) to form the PAN/CNT. The PAN solution used for fiber spinning is prepared in DMF at 90 C with weight concentration of 7% for control and composite fibers. In particular, the carbon filler is added in the form of paste after phase separation and used in composite fiber spinning process. Approximate CNT concentrations studied are (1) 0.29 wt% and (2) 1.17 wt% for Composite-1, Composite-2 fibers, respectively. The fiber spinning is conducted via flowassisted gel-spinning developed by MINUS lab. Both as-spun control and composite fibers are further coagulated in methanol bath overnight and then go through hot-drawing at 120 C and 180 C stages before characterizations. The final draw ratios for Control-PAN, Composite-1, and Composite-2 fibers are 21.65, 26.07, and 28.31, respectively. 149

169 C Heat treatment Processes A Lindberg/Blue M Mini-Mite tube furnace equipped with a quartz tube (diameter 1 inch, obtained from Quartz Scientific Inc.) was used for heat-treatment to conduct both stabilization and carbonization procedures. Two flow meters were used to control the inlet and outlet gas flow through the furnace. The entire system was sealed and the outlet gas flow was directly released into a venting system. For the heat treatment of these fibers a holder made from stainless steel and is shown in Figure C.1. This holder remained inside the tube of the furnace and allowed the fibers to hold a weight of ~12 mg. For stabilization (in air), the temperature was (i) ramped up at 1 C min -1 from room temperature (~25 C) to 300 C; (ii) maintained isothermally for 10 hrs; and (iii) gradually decreased to room temperature. For carbonization (in argon), the temperature was (i) ramped up at 5 C min -1 from room temperature to 1100 C; (ii) maintained isothermal for 5 min at 1100 C; and (iii) gradually decreased to room temperature (~25 C). In both processes, the required air or argon gas flow was maintained constant at 3000 ccm until the chamber was cooled down to room temperature (~25 C). Figure C.1 A schematic of the holder that made to allow fibers hold the weight during the heat treatment procedures. C Sample Characterization Morphology characterization was performed using a Zeiss Supra 25 field emission SEM (operating voltage 5 kv). All film samples were fractured and mounted to a 90 pin stub with the fractured end facing up for SEM observation. Precursor (non-carbonized) 150

170 samples were coated with a thin gold/palladium layer (15-20 nm) for image purposes using a Gatan high-resolution ion beam coater. WAXD was performed on a Rigaku RAPID II equipped with a curved detector XRD system with a 3 kw sealed tube X-ray source (operating voltage 40 kv and current 30 ma). XRD curve fitting and analysis was performed using software PDXL 2 (version ) and 2DP (version ). C.2 Results and Discussion WAXD spectra for the Control-PAN, Composite-1, Composite-2 carbon fibers. are shown in Figures C.2 and C.3. The values of d-spacing and crystal size for (002) graphitic peak at 2θ of ~26 are provided for all films as well as three different carbon fiber that was made from PAN precursor (i.e., GT, IM7, and T300) are shown in Table C.1. GT carbon fibers were made at Georgia Tech University and IM7 and T300 are commercial carbon fibers. These fibers are also subjected to higher heat treatments as compared to those used in this study. The Composite-1 and Composite-2 carbon fibers show better properties as compared to the control-pan carbon fiber. However, control-pan carbon fibers that were made in this study shows lower (002) crystal size and higher (002) d-spacing in comparison to GT, IM7, and T300 carbon fibers. Based on these data, it is possible that carbon fibers made from Composite-1 and Composite-2 could have comparable or potentially improved crystal structure than GT, IM7, and T300 carbon fibers. To get to this point the optimization of the precursor spinning processing should be further improved. By understanding the right heat treatment condition for these fibers, one could be able to make a carbon fiber with a superior crystal structure and therefore properties in comparison to the already available commercial carbon fibers. However, this dissertation shows that carbonization at 1100 C could be a right choice for carbon fiber heat treatment. Understanding the tension 151

171 applied during heat treatment of the fibers and improving the method that used here for adding this tension to fibers could also help to improve the overall structures of the fabricated carbon fibers. Figure C.2 - WAXD spectra of the (a) Control-PAN, (b) Composite-1 and (c) Composite-2 carbon fibers. Figure C.3 2D-WAXD patterns of the (a) Control-PAN, (b) Composite-1, and (c) Composite-2 carbon fibers. 152