|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
1 EFFECTS OF NANOPARTICLES ON THE WEAR RESISTANCE OF POLYTETRAFLU OROETHYLENE By DAVID LAWRENCE BURRIS A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2007
2 2007 David Lawrence Burris
3 This document is dedicated to my parents, Kare n and Larry Burris, who encouraged me to find my own path
4 ACKNOWLEDGMENTS I thank the National Science Foundation, W .L. Gore and the Air Force Office of Scientific Research for financial support of this research. I thank Jim Hanrahan of W.L. Gore for his personal support. I thank DuPont for a steady supply of Teflon 7C molding resin. I also owe a debt of gratitude to the Major Analytical Instrumentation Center (MAIC) at the University of Florida for use of a variety of instruments including the SEM, TEM, microtome, SWLI and Nanoindenter. I thank Jerry Bourne and Kerry Sieban of MAIC for all of their personal help and expertise. I thank Josh Lowitz, Catherine Sa ntos and Renee Duncan for their experimental contributions. In addition, I thank Professors Terry Blanchet, Linda Schadler, Scott Perry, Susan Sinnott and Simon Philpot for invaluable collabo rations. These studies would not have been possible without their efforts. I thank my parents, Karen and Larry for consta nt love and encouragement. They always went the extra mile to afford me every opportun ity. I thank my fiance Jade for love and support in every part of my life. I thank the Tribology Labor atory for thoughtful input on experimentation, help solving tough problems, hard work and friendship. Lastly, I thank Greg Sawyer, my graduate advisor, to whom I owe more than I can possibly repay. He is a teacher in every sense of the word. He has provided me with educational, prof essional and personal support whenever I needed it, and has selflessly mentored me fo r the past six years.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4 LIST OF TABLES................................................................................................................. ..........8 LIST OF FIGURES.......................................................................................................................10 ABSTRACT...................................................................................................................................15 CHAP TER 1 INTRODUCTION..................................................................................................................17 Lubrication in Mechanical Design..........................................................................................17 Introduction to Tribometry..................................................................................................... 18 Solid Lubricants......................................................................................................................19 Polymeric Composites..................................................................................................... 20 Polymeric Nanocomposites.............................................................................................22 2 POLYTETRAFLUOROETHYLENE AS A SOLID LUBRICANT ..................................... 25 Neat Polytetrafluoroethylene..................................................................................................25 PTFE Composites................................................................................................................ ...27 PTFE Tribological Nanocomposites....................................................................................... 29 3 IDENTIFICATION OF WEAR RESIST ANCE MECHANISMS: MOTIVATION F OR CURRENT STUDIES............................................................................................................ 36 Investigations of Transfer Films.............................................................................................37 Transfer Film Morphology..............................................................................................40 Transfer Film Composition.............................................................................................46 Transfer Film Chemistry.................................................................................................50 Role of Transfer Films..................................................................................................... 53 Investigations of Internal Interfaces....................................................................................... 54 Investigations of PTFE Phase and Morphology..................................................................... 58 Summary.................................................................................................................................73 4 EXPERIMENTAL METHODS.............................................................................................74 Materials.................................................................................................................................74 Powder Blending....................................................................................................................75 Sample Preparation.................................................................................................................76 Characterization......................................................................................................................77 Morphology and Dispersion............................................................................................77 Thermal Properties.......................................................................................................... 78
6 Mechanical Properties..................................................................................................... 80 Tribological Properties.................................................................................................... 83 Experimental Uncertainty for Friction Coefficients ................................................. 87 Experimental Uncertainty of Wear Rate.................................................................. 90 5 DESIGN OF EXPERIMENTS............................................................................................... 94 Effects of Mechanical Blending.............................................................................................94 Processing Temperature and Crystalline Morphology ...........................................................95 Mechanical Processing on Nanoparticle Dispersion ..............................................................98 Nanoparticles on the Melt Behavior of PTFE......................................................................100 Filler Dispersion on Nanocomposite Properties................................................................... 102 Filler Material on Nanoc om posite Properties....................................................................... 103 6 RESULTS.............................................................................................................................104 Mechanical Blending............................................................................................................ 104 Characterization of Particle Morphology......................................................................104 Thermal Characterization.............................................................................................. 107 Processing Temperature and Crystalline Morphology .........................................................112 Thermal Characterization.............................................................................................. 112 Mechanical Properties Characterization........................................................................ 115 Tribological Characterization........................................................................................ 121 Mechanical Processing on Nanoparticle Dispersion ............................................................127 Nanoparticles on the Melt Behavior of PTFE......................................................................131 Filler Dispersion on Nanocomposite Properties................................................................... 137 Thermal Characterization.............................................................................................. 137 Mechanical Characterization......................................................................................... 143 Tribological Characterization........................................................................................ 152 Filler Material on Nanoc om posite Properties....................................................................... 158 Nanoparticle Dispersion................................................................................................158 Thermal Characterization of Powder Ensembles.......................................................... 160 Thermal Characterization of Nanocomposites.............................................................. 163 Mechanical Characteriza tion of Nanocom posites......................................................... 166 Tribological Characteriz ation of Nanocom posites........................................................171 7 DISCUSSION.......................................................................................................................175 Fibrillation............................................................................................................................175 Crystalline Morphology and Crystallinity............................................................................ 176 Mechanical Properties.......................................................................................................... 177 Dispersion..................................................................................................................... ........177 Crack Arrestment, Debris Generation and Transfer ............................................................. 179 Closing Remarks...................................................................................................................187 8 CONCLUSIONS.................................................................................................................. 189
7 LIST OF REFERENCES.............................................................................................................190 BIOGRAPHICAL SKETCH.......................................................................................................200
8 LIST OF TABLES Table page 5-1. Blending treatments of neat PTFE to simu late the effects of nanoparticle dispersion on the polym er.................................................................................................................... ....95 5-2. Experimental matrix investigating the eff ects of sinter temperature o n the tribological, thermal and mechanical properties of unfilled virgin PTFE.............................................. 98 5-3. Experimental matrix examining the eff ect of loading and dispersion technique on the powder dispersion of nanopart icles and PTFE. Nanoparticles are 80 nm alpha phase alumina........................................................................................................................ .....100 5-4. Experimental matrix for experiments studying the influence of nanoparticles on the therm al response of the PTFE powder. All samples are jet-milled identically.............. 101 5-5. Experimental matrix investigating the e ffects of nanoparticle loading on the therm al, tribological and mechanical properties of the compression molded nanocomposite...... 102 5-6. Experimental matrix for experiments studying the influence of nanoparticles on the therm al, mechanical and tribological response of the PTFE nanocomposites................. 103 6-1. Blending treatments of neat PTFE to simu late the effects of nanoparticle dispersion on the polym er. ................................................................................................................. ..111 6-2. Experimental matrix investigating the eff ects of sinter te mperatur e on the tribological, thermal and mechanical properties of unfilled virgin PTFE. The uncertainty on melt temperature measurem ents is 0.1C.................................................................................126 6-3. Estimated nanoparticle density results of SEM observation of nanoparticle decorated PTFE powder f ollowing mixing via hand-mixing and jet-milling for 1wt%, 2wt% and 5 wt% 80 nm alumina in PTFE............................................................................. 130 6-4. Results of DSC of powder ensemb les following m ixing via hand-mixing and jetmilling for 1wt%, 2wt% and 5 wt% 80 nm alumina in PTFE. ................................... 137 6-5. Results of DSC of hand-mixed and jet-milled 1wt%, 2wt% and 5 wt% 80 nm alum ina-PTFE compression molded nanocomposites..................................................... 141 6-6. Results of mechanical testing of handm ixed and jet-milled 1wt%, 2wt% and 5 wt% 80 nm alumina-PTFE compression molded nanocomposites........................................... 150 6-7. Results of tribological testing of ha nd-m ixed and jet-milled 1wt%, 2wt% and 5 wt% 80 nm alumina-PTFE compression molded nanocomposites........................................... 155 6-8. Results of DSC of jet-milled 12.5 wt% 80 nm phase and 44 nm phase aluminaPTFE powder ensembles.................................................................................................. 163
9 6-9. Quantitative results of DSC for jet-milled 12.5 wt% 80 nm phase and 44 nm phase alumina-PTFE compression molded nanocomposites........................................... 165 6-10. Results of mechanical testing of jet-milled 12.5 wt% 80 nm phase and 44 nm phase alumina-PTFE compression molded nanocomposites........................................... 169 6-11. Results of tribological testing of handmixed and jetmilled 1wt%, 2wt% and 5 wt% 80 nm alumina-PTFE compression molded nanocomposites...................................... 174
10 LIST OF FIGURES Figure page 1-1. Wear rate plotted versus friction coeffi cient for various solid lubricating polym eric composites, unfilled polymers, and polymer blends.......................................................... 21 2-1. Wear rate versus filler loading for so m e of the PTFE based microcomposite systems found in the tribology literature......................................................................................... 28 2-2. Representation of the matrix and filler pa rticles used in com partmentalization modeling... 33 2-3. Plot of required filler volume fraction plotted vs. the dim ensionless diameter..................... 33 3-1. Surfaces used to study roughness effects on PTFE nanocom posite transfer and wear.......... 38 3-2. Wear rate plotted versus counte rface surface roughness, Rq for 5 wt % 40 nm phase and 80 nm phase alumina-PTFE nanocomposites.......................................................... 39 3-3. Wear rate plotted versus the maximum tran s fer film thickness as measured with optical interferometry................................................................................................................. ...42 3-4. An AFM image of the transfer film produ ced through unidirectiona l sliding of P TFE on a polished silicon wafer..................................................................................................... 43 3-5. Microtribometry friction results for the cr ossed cylinder oriented P TFE transfer film tests.......................................................................................................................... ..........44 3-6. Aluminum atomic content (%) in the transf er films of virgin and fluorinated nanocomposites plotted versus the atomic content as prepared in the bulk...................... 48 3-7. Friction coefficient plotted versus oxygen content as m easured using X-Ray photoelectron spectroscopy................................................................................................49 3-8. Transfer film thickness, oxygen content an d friction coefficient plotted versus track position over half of the wear tr ack for a 1% fluorinated sam ple...................................... 50 3-9. Comparison of the core level C 1s spectra of unfilled PTFE to that of the nanocomposite transfer film.............................................................................................. 51 3-10. Transmission electron images of a) 44 nm b) 40 nm and c) 80 nm particles used in this study............................................................................................................. ...55 3-11. Wear rate plotted versus alumina lo ading for alum ina-PTFE nanocomposites with varying nanoparticles agai nst a lapped counterface........................................................... 56 3-12. Tribological results of test with untreated and fluor inated silane treated 40 nm phase alumina-PTFE nanocomposites......................................................................................... 57
11 3-13. Normalized properties pl otted versus tem perature for variable temperature studies available in the literature.................................................................................................... 61 3-14. SEM images of the worn surface of a 5% 80nm phase alum ina-PTFE nanocomposite.................................................................................................................. .63 3-15. SEM images of the worn surface of a 5% 80nm phase alum ina-PTFE nanocomposite after being fractur ed at room temperature................................................ 63 3-16. X-Ray diffraction for neat PTFE, a low wear nanocomposite and the sam e nanocomposite after a 400C heat treatment..................................................................... 65 3-17. Differential scanning calorimetry (DSC ) of neat PTFE, a low wear nanocom posite and the same nanocomposite af ter a 400C heat treatment............................................... 65 3-18. Atomic force microscopy of the crys talline morphology of a 1 wt% 40 nm aluminaPTFE nanocomposite......................................................................................................... 68 3-19. Wear volume plotted versus sliding distance for a 1 wt% 40 nm alumina-PTFE nanocomposite before and after a heat treatment to 400C............................................... 70 3-20. Wear rate and counterface temperature vers us slid ing distance for variable temperature tribology testing of a wear re sistant PTFE nanocomposite............................................... 71 3-21. Normalized DSC power plotted versus temperature. .........................................................72 4-1. Secondary electron im ages of Teflon 7C as received from DuPont. .....................................74 4-2. Schematic representations of the com ponents comprising the TA instruments Q20 differential scanning calorimeter (DSC)............................................................................ 79 4-3. MTS 858 Mini Bionix II load from used for characterizing m echanical properties.............81 4-4. Tribometer used for friction and wear testing. ...................................................................... 84 4-5. Scanning white light interferometry meas urem ent of a representa tive lapped stainless steel (304) counterface....................................................................................................... 85 4-6. Model representation of m isaligned measurement axes........................................................ 88 5-1. Differential Scanning calorimetry therm ogram of as-received virgin PTFE during a heat/cool/heat cycle............................................................................................................96 6-1. SEM images of virgin PTFE following vary ing m echanical treatments typically used in particle dispersion............................................................................................................ 105 6-2. Estimated particle size results of virgin PTFE following varying mechanical treatm ents..................................................................................................................... ....106
12 6-3. Differential scanning calorimetry of PTFE powders with varying m echanical history. ... 108 6-4. Quantified results of differential scanni ng calorim etry of PTFE powders with varying mechanical history........................................................................................................... 110 6-5. Differential scanning calorimetry of PT FE s amples compression molded with varying sintering hold temperatures.............................................................................................. 113 6-6. Results of differential scanning calorimet ry of PTFE sa mples compression molded with varying sintering hold temperatures.................................................................................114 6-7. Results of mechanical testing of PT FE sa mples compression molded with varying sintering hold temperatures.............................................................................................. 116 6-8. Quantified results of mechanical tes ting of PTFE sa mples compression molded with varying sintering hold temperatures.................................................................................116 6-9. Backscattered electron images of the fract u re surfaces of PTFE with varying sintering temperatures.....................................................................................................................119 6-10. Secondary electron images of the fractur e surfaces of PTFE with varying sintering tem peratures.....................................................................................................................120 6-11. Tribological results of w ear testing PTFE sam ples with varying sintering temperature...122 6-12. Optical images of the transfer film s of PTFE sam ples with varying sintering temperatures following test interrupti ons at various sliding distances............................123 6-13. Stylus profile measurements across the transfer film s of PTFE samples of varying sintering temperature following 26 00 m of sliding at 6.3 MPa. .................................... 124 6-14. Wear rate and friction coefficient plotted versus sintering tem perature........................... 125 6-15. Representative SEM images of nanoparticle dispersions ................................................. 128 6-16. Estimated nanoparticle density plotted ve rsus alum ina loading for hand-mixed and jetmilled powder samples....................................................................................................130 6-17. Thermograms from differential sca nning calorim etry of alumina-PTFE powder ensembles blended by hand and by jet-milling................................................................ 132 6-18. Quantified results from differential s canning calorim etry of alumina-PTFE powder ensembles blended by hand and by jet-milling................................................................ 136 6-19. Heat flow plotted versus tem perature from differentia l scanning calorimetry of handmixed and jet-milled 0, 1, 2 and 5 wt% alumina-PTFE nanocomposites compression molded at 362C..............................................................................................................139
13 6-20. Heat flow plotted versus tem perature from differential scanning calorimetry of repeat specimens of hand-mixed and jet-milled 0, 1, 2 and 5 wt% alumina-PTFE nanocomposites compression molded at 362C. ............................................................ 139 6-21. Quantified results of differential scanni ng calorim etry of hand-mixed and jet-milled 0, 1, 2 and 5 wt% alumina-PTFE nanocompos ites compression molded at 362C............. 142 6-22. Optical micrographs of the micr structure of the nanocomposites at 20X m agnification.................................................................................................................. .145 6-23. Results from analysis of opaque micros tructural dom ains within the nanocomposites.... 146 6-24. Results from mechanical characteriz ation of 80 nm alpha phase alumina-PTFE nanocomposites................................................................................................................ 148 6-25. Scanning electron microscopy of the fracture surfaces.....................................................151 6-26. Results of tribology experiments of hand-m ixed and jet-milled 0, 1, 2 and 5 wt% alumina-PTFE nanocomposites compression molded at 362C...................................... 153 6-27. Wear rate versus alumina nanoparticle wt% for hand-m ixed and jet-milled 0, 1, 2 and 5 wt% alumina-PTFE nanocomposites compression molded at 362C.......................... 155 6-28. Analyses of wear tracks following th e tribological experim e nts with hand-mixed nanocomposites................................................................................................................ 156 6-29. Analyses of wear tracks following th e tribo logical experiments with jet-milled nanocomposites................................................................................................................ 157 6-30. Powder dispersions for 12.5 wt% alumina nanoparticles in PTFE................................... 159 6-31. DSC heat flow plotted versus temp erature for heating and cooling of powder ensem bles of unfilled PTFE, 12.5 wt% phase alumina nanoparticle in PTFE and12.5 wt% phase alumina in PTFE.......................................................................160 6-32. Quantified results of differe ntial scanning calorim etry of a and phase alumina nanoparticles dispersed in PTFE plotted versus filler wt%............................................. 162 6-33. DSC heat flow plotted versus temper ature for heating and cooling of com pression molded samples of unfilled PTFE (x2) and nanocomposites of 12.5 wt% and phase alumina-PTFE........................................................................................................ 164 6-34. Quantified DSC results plotted vers us filler wt% from unfilled PTFE and nanocomposites of 12.5 wt% and phase alumina-PTFE....................................... 166 6-35. Stress plotted versus e ngineering strain for 12.5 wt% and ph ase alumina-PTFE nanocomposites................................................................................................................ 167
14 6-36. Quantified results from m echanical tests of 12.5 wt% and phase alumina-PTFE nanocomposites................................................................................................................ 168 6-37. SEM images of the fracture surf aces of alu mina-PTFE nanocomposites......................... 170 6-38. Friction coefficient and wear rate plotted versus s liding distance..................................... 171 6-39. Stylus profilometric measurements of the surfaces of transfer film s................................ 172 6-40. Wear rate plotted versus sliding distance for jet-milled 12.5 wt% and phase alum ina-PTFE nanocomposites....................................................................................... 173 6-41. Wear rate plotted versus maximum transfer film thickness for alumina-PTFE nanocomposites of various particle phase, size shape and loading................................. 173 7-1. Optical images of polished sections of 12.5 wt% a) and b) phase alum ina-PTFE nanocomposites................................................................................................................ 178 7-2. Wear rate versus sliding distance for jet-milled a phase alumina-PTFE nanocomposites with backscattered electron im ages of w ear surfaces at corresponding wear rates.........181 7-3. Worn volume versus sliding distance for the 2 and 5 wt% phase alum ina nanocomposites................................................................................................................ 182 7-4. Wear rate versus sliding distance for jet-milled a phase alumina-PTFE nanocomposites.. 183 7-5. Abrasion rate to the counter face plotted versus filler wt%.. ............................................... 185 7-6. Hypothesized model of the w ear of effective nanocom posites........................................... 187
15 Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy EFFECTS OF NANOPARTICLES ON THE WEAR RESISTANCE OF POLYTETRAFLU OROETHYLENE By David L. Burris December 2007 Chair: W. Gregory Sawyer Major: Mechanical Engineering Solid lubricants comprise an important cla ss of materials and find use in applications where the use of more traditional lubricati on techniques is undesi rable or precluded. Polytetrafluoroethylene (PTFE) is reputed as having the lowest friction of any bulk polymer and is used to lubricate a wide vari ety of systems from armor pierci ng bullets to frying pans, but high wear rates limit its application. The use of nanof illers has proven to be an effective means for reducing the wear of PTFE w ithout introducing detrimental e ffects on its other beneficial properties; microfillers often increase abrasion a nd friction, and reduce mechanical integrity and chemical resistance. Past studies have been us ed to identify several potential wear resistance mechanisms of PTFE nanocomposites: 1) bonding a nd strength at the fill er/matrix interface, 2) dispersion and mechanical effects of load suppor t and crack deflection, 3) morphological effects of nanoparticles on the matrix and 4) fibrillati on and toughening, 5) transfer film coverage, 6) transfer film orientation and 7) chemical degradation. It is found that for wear resistance to increase by more than two orders of magnitude at trace nanoparticle loadings, the filler must activate a synergism of wear resistance mechanisms. These studies suggest that the nanofillers interact with the PTFE and possi bly lead to a finer scale lame llar structure. The resulting
16 mechanical properties preclude easy crack propagation through the material, which results in the regulation of debris size. The small size of the debris makes removal from the interface difficult and as a result, the debris are tr ansferred to both the counterface and pin. The transferred debris deform, and over time, form protective transfer f ilms. These thin and protective transfer films are largely responsible for the additional 100X in wear resistance over traditional PTFE composites.
17 CHAPTER 1 INTRODUCTION Lubrication in Mechanical Design Tribology is the study of surfaces in relative motion, and lubri cation is one of the most critical and underestimated aspect s of mechanical design. Nearly all moving mechanical systems rely on lubrication for motion. In the best case, a poor understanding of tribology leads to inefficiency; in the worst case, it leads to ca tastrophic system failure. Materials do not have intrinsic friction coefficients or wear rates. These parameters are syst em dependent and can be strongly affected by speed, pressu re, temperature and environmental pressure and composition. Modern designs are subject to increasingly dynamic and harsh environments with space tribology being a classic example. The engineer must not only design tribological systems that can endure such harsh environments, but forecast the system response to extreme changes in the operating environment. Engineers often regard materials in terms of bulk properties. In tribology, surfaces dominate the phenomena under study. Atoms in the bulk of a material are surrounded by and bonded to neighboring atoms, and as a result, they are in a low energy state. At the surface, atoms are exposed and in a higher energy state. Macroscopically, this is manifested as surface energy. When two clean surfaces come into contact, there is a strong tendency of the system to enter a lower energy state by forming bonds at the contacting surface protuberances or asperities . Thus, one of, if not the major challenge in tribology is to keep solid surfaces separated. This is traditionally achieved through the use of a lubricant whose pr imary objective is to separate solid surfaces, preventing direct asperity contact and reduc ing the likelihood for seizure. In fluid lubricated systems, surfaces can be separated by a hydrodynamic film. In grease lubricated systems, the grease has low surface en ergy relative to the surface and a boundary layer
18 naturally forms to reduce direct contact of the su rfaces. Because sliding in these systems occurs through shearing of low strength films, friction coe fficients and wear rates are characteristically low . Introduction to Tribometry Friction coefficients and wear rates are the most commonly discussed and quantified parameters in tribology. Friction coefficients can dictate required motor torques and loads, and wear can lead to debris generation, binding, slop a nd limited life. Because of the important role of each in design, they are the primary metric s of performance in tribological systems. A friction coefficient, is defi ned as a ratio of the force that resists sliding to the normal force. A tribometer is a device used to measure friction coefficients. While there is no standard tribometry test, experimental set ups generally utilize similar desi gn philosophies. In its simplest form, a flat sample is slid against the flat surf ace of a much larger and harder block of material called the counterface. This results in an approxi mately uniform pressure distribution within the sample. In many cases, the counterface material and surface finish are important factors in system performance. Upon slidin g, the frictional and normal forces are measured or inferred at the specimen simultaneously. A detailed uncertain ty analysis of the me asurement of friction coefficient on a similar pin-on-flat tribometer was performed by Schmitz et al. , and illustrates the metrology challenges associated with such a seemingly simple measurement. Wear rate, k, is defined as the volume of ma terial removed per unit of normal load per unit distance of sliding, with typical units being mm3/(Nm). Values of wear rates can vary by many orders of magnitude depending on the bearing ma terials, environment and lubricant. Since volumetric measurements can not practically cover this range, the test length often becomes a function of the wear rate being measured (loads are usually held cons tant). Reporting the
19 experimental uncertainty is necessary to indicate th e quality of measurement, and it is especially important when wear rates are low. Calcula tions of normal load a nd sliding distance with associated uncertainties are fa irly straightforward, but measurements of volume loss often require more careful consideration. For materi als that do not uptake or outgas, material mass measurements are typically made because dimens ional distortions due to elasticity, plasticity, creep and thermal fluctuations can confound dimensional measurements of wear. Density can be calculated by making an initial sample mass m easurement with dimensional measurements or with another direct measurement of volume. Schmitz et al.  performed a detailed uncertainty analysis of wear rate measurement for a pin-on-flat tribometer. Solid Lubricants In an increas ing number of applications the use of traditional fluid or boundary lubricants is undesirable or even precluded; these applications often necessitate the us e of solid lubricants. A solid lubricant is typically a low strength mate rial that promotes low friction sliding without the requirement of an external lubricant. In ma ny cases, a solid lubricant is used as a low cost, environmentally friendly alternativ e to fluid and grease based syst ems, eliminating the need for fluids, reservoirs, pumps, filters and maintenance . In some cases, solid lubricants play a supplementary role. Thin, solid lubricant films are used on foil air bearings and engine valve trains to protect the components during start-up and shut-down wh en speeds are insufficient for aeroor hydrodynamic lubrication. Fluid and greas e lubricated systems are highly sensitive to contamination, and many rely on solid lubri cant seals to keep di rt, debris and other contamination out of the tribological interface. In areas like the food industry, solid lubricants are used for conveyor bushings to prevent contamin ation of the product by th e lubricant. In the integrated circuits industry, very chem ically resistant so lid lubricants (e.g. PTFE) must be used
20 for valve seats and bushings in the transport of caustic etchants. Many aerospace applications require solid lubricants to endur e an extreme range of harsh environmental conditions [5-7]. These environmental challenges can include salt spray, sand, radiation, vacuum and cryogenic temperatures. Polymeric Composites Polym eric solid lubricants comprise an im portant segment of the lubrication field, providing robust lubrication under a wide range of conditions, but with poor performance relative to fluid and grease lubri cants. Figure 1 is a graph of w ear rate plotted versus friction coefficient for various unfilled polymers, polymer blends, and polymer composites used in tribology studies [8-19]. While tribological performance does not have a single unique definition, broadly speaking, solid lubricants with low wear rate s and low friction coefficients are desirable. For practical purposes a designer might include constant performance guidelines (Figure 1 illustrates how such guidelines might be used) whose slopes depend on the relative importance of friction coefficient and wear rate fo r a specific application (note: wear rates are on a log scale). High performance engineering po lymers like Polyethereth erketone (PEEK) and Polyimide (PI) have good wear resistance but hi gh friction coefficients, while low friction materials like Polytetrafluoroethylene (PTFE) us ually have prohibitively high wear rates. In general, neat polymers lack the tribological perf ormance required for most applications; there are many examples of polymer composites in tribology.
21 Figure 1-1. Wear rate plotted vers us friction coefficient for vari ous solid lubricating polymeric composites, unfilled polymers, and polymer blends. The target region is the lower left hand corner, a region of ultra low wear rate and friction coefficient. The data points are labeled with the constituents and listed as a-p: a) PTFE/PEEK composite Lu et al. 1995, b) Si3N4/PEEK nanocomposite Wang et al. 1996, c) PA6/HDPE blend Palabiyik et al. 2000, d) PTFE/PEEK composite Wang et al. 2000, e) ZnO/PTFE nanocomposite Li et al. 2001, f) FEP/PTFE composite Menzel et al. 2002, g) CNT/PTFE nanocomposite Chen et al. 2003, h) Al2O3/PTFE nanocomposite Sawyer et al. 2003, j) Al2O3/PTFE nanocomposite and unfilled PTFE Burris and Sawyer 2005, k) epoxy/ePTFE composite McCook et al. 2005, m) Al2O3/PTFE nanocomposite Burris and Sawyer 2006, o) PEEK/PTFE co mposite and unfilled PEEK Burris and Sawyer 2006, p) unfilled PI unpublis hed, V = 50.8 mm/s, P = 6.25 MPa, reciprocating pin-on-disk tribometer. One philosophy of material desi gn in tribology is to improve the frictional behavior of a wear resistant polymer. For example, additi ons of PTFE to PEEK have been found to significantly reduce friction coefficients; this ofte n results in reduced wear. In this particular composite a soft PTFE film is preferentially draw n from the composite to separate the surfaces, protecting the relatively soft polym eric material from direct asperity contact, and providing a low
22 shear friction reducing film to accommodate the slid ing motion; this is call ed a transfer film. Transfer films are exceedingly important to the success of a solid lubricant, both protecting the bulk and reducing the sliding fo rces at the interface . The opposite method is also employed where hard particle and fiber fillers are used to reduce the wear of a low friction high wear materi al like PTFE, often at the expense of friction coefficient. There are significant efforts dedi cated to the research and development of low friction, low wear solid lubricants with traditional particle and fiber fillers, many of which have successfully transferred to application. Friedrich et al.  and Zhang  reviewed the state of the art of polymer composites in trib ology in 1995 and 1998, respectively. Polymeric Nanocomposites In tribology, one of the drawbacks of the trad itional hard m icron-sized particle and fiber fillers frequently used to reinforce polymers is that they tend to abrade the counterface. Abrasion prevents the formation of a protective transfer film, increases the friction coefficient and counterface roughness, and leads to third body wear of the composite. Nanopa rticles (defined as particles with a characteristic dimension less than 100 nm) have the pot ential to reduce the abrasion that leads to these cascading and problem atic events. Because nanoparticles are of the same size scale as counterface aspe rities, they may polish the highest asperities and promote the development of tribologically favorable transfer films. Once formed, transfer films shield the composite from direct asperity contact and damage . Another benefit of nanoparticles is that at low loadings (< 5%), nanocomposites can have tremendous particle number densities and interfacial surface areas. Consequently, nanoparticles have great potential for impacting a number of phys ical properties at low filler loadings. Siegel et al. found that with about 2% (volume) alumina nanoparticles, the tensile strain to failure
23 of PMMA improved by 400%, and Ng et al.  found the scratch resistance of a TiO2-Epoxy nanocomposite to be superior to both unfilled an d micro-filled Epoxy. Du ring strain dependent Raman spectroscopy measurements of multi-wa lled carbon-nanotube (MWCNT) filled Lexan polycarbonate, Eitan et al.  found that load was transferred to the nanotubes. They also found that an epoxide surface treatment of the na notubes improved the load transfer through the interface, highlighting the role of the interface on mechanical properties. Nanofillers can not only improve material properties through mechanical transfer, but they can also influence the crystallization and morphology of the polymeric matrix which can further alter physical properties. Many authors have observed the dire ct effects of the nanopa rticles on the matrix through changes in the glass transition and degr adation temperatures of the polymer matrices [26-29]. Clearly, nanoparticles can influence th e crystallinity, morphology and behavior of the polymer itself and the potential for multifuncti onality in these nanostructured materials is substantial. Detailed studies of matrix prope rties are needed but generally lacking in the tribology literature. Within the past decade there have been a number of tribological studies conducted to investigate the role of nanoparticles in pol ymer nanocomposites [8, 10, 11, 16-19, 30-34]. Early studies by Wang et al.  used nanoparticles in a PEEK matrix. The nanoparticles were dispersed in the PEEK powder by ul trasonication in an alcohol bat h. In an initial study, <50 nm Si3N4 was found to be effective in reducing the wear rate and friction coefficient of PEEK. The improvements in tribological performance were mostly attributed to the vast improvements observed in the quality of the transfer films. A follow-up study l ooked directly at the effects of particle size and shape on the tribological be havior of the composite . Micron-scale whiskers, microparticles and nanoparticles of SiC were used with 5% loading in PEEK. The
24 whiskers were effective in reducing the wear of PEEK (~33%) but friction was only reduced ~8%. The microparticles were effective in redu cing the friction coefficient (~33%) but wear rate was only reduced by ~9%. The nanoparticles eff ectively reduced both with a reduction in wear rate of ~44% and a reduction in friction coe fficient of ~50%. In a later size study involving nanometer ZrO2 from 10 nm to 100 nm in a PEEK matr ix, it was found that for approximately 2% loading, both friction coeffici ent and wear rate increased monot onically with increased filler size (improved performance of PEEK with various loadings of SiO2 nanoparticles was also found) [32, 33]. In each of these studies, thin uniform transfer films accompanied reduced wear rates and friction coefficients. In 2000, Schwartz and Bahadur published a study that examined the influence of alumina nanoparticles on the tribological behavior of pol yphenylene sulfide (PPS) . Powders were dispersed with what is described as an electric mixer. A 2X reduction of wear was observed for a 2% filled nanocomposite. They found good co rrelation between the bond strength of the transfer film and the wear rate of the composite and concluded that the role of the filler was to anchor the transfer film. They attributed the increased w ear rates at loadings above 2% to abrasion of the transfer film by nanoparticle aggregates.
25 CHAPTER 2 POLYTETRAFLUOROETHYL ENE AS A SOLID LUBRICANT Neat Polytetrafluoroethylene Tribology in extrem e environments is one of the primary driving forces for the development of novel solid lubricants with impr oved performance. In space environments, for example, the tribological components are subjected to near perfect vacuum, intense radiation, atomic oxygen and a wide thermal range. Propertie s like vapor pressure, chemical inertness and thermal stability are critical in these applications and preclude the use of traditional fluid and grease lubricants. Polytetraflu oroethylene (PTFE) is a unique polymer; not only is it widely regarded to have the lowest fr iction coefficients of any bulk pol ymer, but it also has a low vapor pressure, is chemically inert and has one of the largest operational th ermal ranges of any bulk polymer. It is uniquely suitable for a variety of extreme environment tribological applications, and is the solid lubricant matrix material under inves tigation in this study. PTFE typically consists of 20,000-200,000 mers, or repeating units, of tetrafluoroethylene (C2F4) in a helix configuration. The carbon-fluorine bond is very strong, and although the carbon backbone is only single bonded, it is loca ted within a fluorine encasement, which effectively shields it from chemical attack. Th e unique physical propertie s of PTFE are derived from high chemical stability and smooth linear morphology of the PTFE molecule. The tribological properties of PTFE have been studied for more than 50 years. McLaren and Tabor found that the friction of PTFE beha ved as if governed by viscoelastic effects, increasing with increased speed and decreased te mperature . Makinson and Tabor found that thin transfer films were developed on the c ounterface during sliding with PTFE . In addition, they found evidence that the film was st rongly adhered to the counterface contrary to the conventional wisdom that low friction of PT FE was the result of poor adhesion. From these
26 results, it was concluded that the tribological interface consisted of self-mated PTFE surfaces. This conclusion led to the hypothesis that motion occurred through the shea ring of crystallites past one another in a lamellar fashion similar to the shearing of a deck of cards. Pooley and Tabor found frictional anisotropy with high fricti on against the chain direction and low friction with the chain direction and conc luded that low friction was due to the smooth molecular profile with high radial stiffness and low ax ial resistance to sliding . Despite the beneficial fricti onal characteristics of PTFE fo r tribological application, its wear rate is significantly higher than many other polymers; this high rate of wear has prohibited its use in many applications and is largely res ponsible for the limited PTFE research in the area of solid lubrication. During low speed sliding (< 10mm/s), PTFE has a low friction coefficient (between =0.03 and =0.1) and mo derate wear resistance (10-5 mm3/Nm). Makinson and Tabor  found that as the sliding sp eed increased to above 10 mm/s at room temperature, a transition from mild to severe wear (10-5 mm3/Nm to 10-3 mm3/Nm) accompanied increased friction. As speed is increased from an original condition of low friction and moderate wear, they conjectured that the stresses requ ired for sliding exceeded the stress required to cause failure at boundaries between crystalline domains in the sinter ed material; this leads to larger debris and increased wear rates. Tanaka proposed a simila r model with failure occurring at boundaries of the characteristic banded stru cture of PTFE . Blanchet and Kennedy  studied this severe wear transition at several temperatures and found an increase in the transition speed to accompany increased temperature. When the w ear rate, k, was plotted versus the friction coefficient, the transition to severe wear o ccurred at = 0.1 in each case. These results are consistent with the proposed tr ansition mechanism of Makinson and Tabor  and suggest that the severe wear transition is a response to the st ress state and thus the friction coefficient, while
27 the friction coefficient is a function of both speed and temperature. Recent studies suggest that the temperature dependence of the friction coefficient may be due to thermally activated barriers to sliding. Several samples from the studies of Blanchet and Kennedy were microtomed perpendicular to the wear surface in the direction of sliding after mild and severe wear had taken place. Cracks were found to propagate in the di rection of sliding beneath a layer of worked material at subsurface depths consistent with observed debris thicknesses for severe wear samples. No such cracks were found in mild wear samples. They believed that defects in the sintered material acted as initiated cracks. Wh en speeds are low, the kinetic friction coefficient at the tribological interf ace is low, and the static friction coe fficient between internal crack faces is sufficient to fully support the surface tracti ons. However, when the kinetic coefficient of friction at the tribological interf ace increases with increased sliding speed and exceeds the static coefficient of friction ( ~ 0.1) at internal PTFE/PTFE interf aces, the crack tips must support shear. This leads to a progressi ve delamination wear process sim ilar to that described in Suhs delamination theory of wear . PTFE Composites For decades, fillers have been successfully used to reduce the wear of PTFE. In Figure 21, wear rate is plotted versus filler wt% for testing of some representative PTFE composites found in the literature [12, 14, 41-43]. Despite being tested with varying c onfigurations, testers, methods, pressures, speeds and fillers, there is a systematic trend of d ecreased wear rate with increased loading up to 50 wt%.
28 Figure 2-1. Wear rate versus f iller loading for some of the PT FE based microcomposite systems found in the tribology literature: a) Li et al. 2000 graphite b) Bahadur and Gong 1992 graphite c) Lu et al. 1995 PEEK d) Burroughs et al. 1999 B2O3 e) Menzel and Blanchet 2002 irradiated FEP. N eat PTFE has a high wear rate (k~10-3 mm3/Nm) at speeds above 10 mm/s, while comp osites typically approach a moderate wear rate (k~10-5 mm3/Nm) as filler loading increases above 10 wt%. The wear reducing mechanism of fillers in PTFE based composites remains a topic of debate. Lancaster  proposed th at the hard wear-resistant filler s, especially those with a high aspect ratio, preferentially suppor t the load and reduce the wear of PTFE in the composite. Sung and Suh  found that vertically oriented fibe rs were most effective in reducing wear, but suggested that the critical role of the filler was to arrest crack propagation, rather than to support the load. Tanaka et al.  suggested that the f iller prevented the initial transfer of the PTFE to the counterface, and thus prevented transfer wear Briscoe  noted the formation of a thin, well adhered transfer film for a high density poly ethylene composite and hypot hesized that fillers provide augmented transfer film adhesion, and th us reduced transfer w ear by slowing transfer film removal and the requisite replenishment. Using X-Ray photoelectron spectroscopy (XPS), Gong et al.  found that the wear rate of PTFE was independent of chemical bonding with the
29 counterface, and concluded that cohesive failure within the PTFE must govern its wear rate. Blanchet et al.  had similar findings with XPS analys is of PTFE and PTFE composites in dry sliding, and concluded that the we ar reducing role of the filler is to slow primary removal of material from the bulk by arres ting crack propagation rather than slowing secondary removal of material from the counterface via increased transfer film adhesi on. Bahadur and Tabor  and Blanchet and Kennedy  saw dir ect relationships between wear rate, debris size and the ease with which debris are expelled from the contact and concluded that the fillers interrupt the formation of the larger debris that form during severe wear of PTFE. PTFE Tribological Nanocomposites Various fibe r and particle fillers have succe ssfully reduced the wear of PTFE by several orders of magnitude, but they also increase frict ion coefficients and abrade favorable transfer films and the counterface, both of which limit the effectiveness of the filler. Additionally, the high filler loadings (~20%) needed for significant wear reductions have detrimental effects on the beneficial frictional, thermal and chemical properties that make PTFE so attractive to designers. The use of nanoparticles has the potential to eliminate many of the limitations of traditional fillers in a PTFE matrix. Low load ings of nanoparticles have imparted impressive improvements in mechanical properties such as st rength, modulus and strain to failure to other polymeric matrices. They have also been f ound to reduce abrasion and promote transfer film development. Despite the success of micro-fillers in abating severe wear of PTFE and the demonstrated benefits of nanoparticles on the properties of other polymer matr ices, there was a sentiment in the field that nanoscopic fillers were ineffec tive in reducing the wear of PTFE. This was
30 primarily based on a study by Tanaka and Kawa kami  that showed inferior wear performance of sub-micron TiO2-PTFE composites to PTFE composites with larger sized fillers of other materials. As a result of these findings, it remains widely accepted that nanofillers cannot provide improvements in the wear resist ance of PTFE because they are readily swept away within the matrix as debris by relatively large asperities. In 2001, Li et al.  filled PTFE with 15 wt% nano-s cale ZnO, and found a two order of magnitude reduction in wear while retaining a low coefficient of friction. This study not only established that nanofillers could be as effective as microparticle s in reducing the wear of PTFE at lower loadings, but it also de monstrated that low friction coefficients could be retained upon loading. Uniform, well-adhered transfer films were observed for low wear composites and no signs of abrasion to the count erface were observed. Chen et al.  created a PTFE nanocomposite with single-walled carbon-nanotubes and found that friction coefficient was reduced slightly and wear resi stance was improved by more than two orders of magnitude over unfilled PTFE. Sawyer et al.  made nanocomposites of PTFE with 38 nm Al2O3 and found a 600x reduction in wear with 20 wt% filler concentration. Wear was reduced monotonically as filler concentration was increased to 20 wt%. The most important result from these initial exploratory PTFE nanocomposite studies was a 10 X improvement in wear resistance at 0.4 wt% nanoparticle loading; in the microcomposites l iterature, negligible reductions in wear are observed with less than 5 wt% microparticle loadi ng. Initial rules of mixtures and preferential load support models of wear resistance were inadequate to predict the su ccess of nanofillers at low loadings, resulting in an impetus to form ulate new models for wear resistance in these unique materials to facilitate future material design.
31 Early investigations of the dominant wear reduction mechanisms in PTFE nanocomposites focused on strengthening and toughening of th e matrix and the transfer films. Li et al.  used secondary electron microscopy (SEM) to study cr oss sections of unfille d and nanofilled PTFE. The neat PTFE had many fibers drawn from th e bulk while the nanocomposite did not. They suggested that the nanoparticles effectively prevented the destru ction of the banded structure. They also found thick, patchy transfer films fo rmed by unfilled PTFE, while thin, tenacious transfer films were formed by the wear resi stant nanocomposite. It was offered that the nanoparticles help bond the transfer film to the counterface which promotes low wear by protecting the soft composite from direct asperity damage. Chen et al.  also found evidence to suggest that the nanotubes prevented destruction of the crystalline structure of the PTFE. The high aspect ratio fillers were thought to reinforce the matrix by intertwining with PTFE crystals. In addition, they hypothesized that the nanotubes may provide addi tional self lubrication after breaking off from the composite duri ng wear. In the study by Sawyer et al. , SEM revealed that the PTFE particles were decorated by the nanoscopic alumina during a powder blending process that preceded compression molding. The resulting structure after molding was cellular with thin regions of highly concentrated alum ina rich material surrounding micrometer sized domains of nominally unfilled PTFE. These con centrated regions were hypothesized to act as barriers to crack propagation, reduc ing the delamination wear of PTFE. Further, it was offered that with increasing filler concentration, the number, size and possibly strength of the compartmentalizing regions increased. In 2006, Burris proposed a simple, delamination-based wear model for PTFE nanocomposites that assumed that the severe wear mode of PTFE is one in which cracks preexist or initiate and propagate to failure . Once these cracks encounter resistance (filler),
32 they are arrested or are turned toward the surf ace to generate a wear particle. Fillers were therefore assumed to play a crack arres ting role that was described as damage compartmentalization. For simplification, it was assumed that the rate of in itiation is constant so that each wear particle represents an initiation point. Therefore, the smaller the wear particle at each initiation point, the lower the rate of wear (this is the basis of the wear model presented by Bahadur and Tabor . A few things should be kept in mind with this model: 1) the rate of initiation is probably faster in a filled system due to the addition of imperfections at the part icle/matrix interfaces, 2) only cracks at depths of the same order of magnitude as the compartmentalized length can effectively be liberated as debris because of the surrounding material, and 3) the wear volume scales by R3, where R is the characteristic compartment (matrix particle) radi us. Point 2) likely counteracts the false assumption that the initiation rate is consta nt, and point 3) suggests that small matrix particles are de sirable for reducing wear. The PTFE used during processing was a gran ular compression molding resin and it was presumed that cracks are arrested most e fficiently when each matrix particle is compartmentalized by a monolayer of filler; th is corresponds to the least amount of filler required to effectively arrest a crack propagating through any matrix particle. At filler loadings less than the critical loading, the probability of arresting each crack is diminished. At loadings much greater than the critical loading, there is insufficient matrix available to effectively bind all of the particles and the mechanic al properties of the composite ra pidly deteriorate with loading. The model system is shown schematically in Figure 2-2.
33 Figure 2-2. Representation of the matrix and filler particles used in compartmentalization modeling. The model matrix particle and fill er particles are treate d as spheres. The volume fraction for complete coverage of the matrix by filler is solved for using various simplifying assumptions. Figure 2-3. Plot of required f iller volume fraction plotted vs. the dimensionless diameter. Nearly an order of magnitude reduction in required filler content is achieved by an order of magnitude reduction in filler diameter. Figure 2-3 shows the volume fraction of filler required for complete damage compartmentalization as a function of the relative size of filler to matrix particles. This graph can be used as a simple tool for composite design, and is instruct ive in considering the
34 advantages of nanocomposites over microcomposite s; by reducing the filler diameter by an order of magnitude, the required filler content is also re duced by an order of magnitude. If the matrix and filler particles are of the same size (D*=1), the simplified model gives a required 80 vol% filler (this is underestimated by 15% due to the assumption that th e filler is much smaller than the matrix). If the filler is 1/100 the size of the matrix (D*=0.01), the model predicts a required 3.8 vol% filler (1.9% underestimated). For D* = 0.001, as is typical for nanocomposites, 0.4 vol% filler is required for monolayer cove rage. Hence, very effective damage compartmentalization in PTFE should be possibl e at very low filler volume fractions. Though the model is oversimplified, it has the potential to capture a wide range of effects not accounted for using rules of mixtures, including m echanical crack deflec tion, morphological and crystallization effects in the matrix due to the presence of the nanoparticles. In 2005, Burris and Sawyer used a highly energe tic jet-mill dispersion technique to create PTFE nanocomposites with 80 nm particles of alpha phase alumina . The nanocomposites were about an order of magnitude more wear resi stant than the state of the art of the time, but most striking was the fact that the 3000X impr ovement in wear resistan ce occurred with only 0.5% alumina loading; prior to this study, wear reductions were on the order of 100X and were only found at nanoparticle loadings above 5 wt%. In 2006, McElwain directly stud ied the effects of size on the tribological properties of 5 wt% PTFE composites using the same alpha phase alumina . Particles were dispersed using a high speed dry powder shear mi xing technique. They found that 40 and 80 nm nanocomposites were on the order of 104 more wear resistant, while 1, 2, 5 and 20 m composites were on the order of 102 more wear resistant than neat PTFE. These results suggested a transition in behavior from the nanoscale to the microscale as opposed to the continuous behavior predicted
35 by the damage compartmentalization model. Intere stingly, despite the similarity in performance of nanocomposites at 5 wt%, McEl wain found negligible improveme nts at 1 wt%, while Burris and Sawyer retained high wear resistance. Th is result implicated th e powder blending (jetmilling) as a crucial part of the processing. Following McElwains size study, additional experiments were conducted to study the cause of the reduced microcomposite wear resi stance. They created hybrid composites with nanoparticles and microparticles and found that the presence of the mi croparticles in the composite disabled low wear sliding of the nanocomposites by providing an additional wear pathway that was otherwise unavailable. The micr oparticles abraded the transfer film and led to the abrasion and transfer wear proces s typical of microcomposites.
36 CHAPTER 3 IDENTIFICATION OF WEAR RESIST ANCE MECHANISMS: MOTIVATION F OR CURRENT STUDIES In general, the results from the PTFE nanoc omposites tribology liter ature are striking. Contrary to early suggestions th at nanoparticles would be ineffective fillers in reducing wear of PTFE, the use of nanoparticles in PTFE has been very successful with 1,000X improvements in wear resistance occurring with as little as 1wt% (0.5% by volume) nanoscale filler. There is however, a clear lack of unders tanding of the fundamental m echanisms of wear and wear resistance in these materials, which makes PTFE nanocomposite design an exercise of trial and error. It has been demonstrated that the presence of micron-scale abrasives can increase wear rates by orders of magnitude due to a disruption of transfer film development . However, studies showing impressive impr ovements at low nanoparticle load ings are contrasted by studies demonstrating equally unimpressive improvements; significant differences in the tribological properties of nanocomposites in th e literature occur with seemingly subtle differences in the materials and processing methods. Many undefine d variables arise from study to study, and too often, only qualitative descriptors of transfer films, debris morphology, mechanical properties and most importantly, nanoparticle dispersion are used. To date, it is unclear whether the key factors driving the physical properties of PTFE nanocomposites have been identified. Previous studies suggest that th e wear rates of these systems are complex and coupled, possibly involving crack deflection, filler/matrix interactions, regulation of debris size and debris/counterface interactions, but there is a cu rrent need for more quantitat ive measurements to enable identification of relevant wear resistance mech anisms. Given the scope of this area, a broad exploratory effort was nece ssary to direct the current research initiative.
37 Transfer films have been shown to be a criti cal part of the solid lubricant tribo-system, providing a low shear interface for sliding and forming a protective layer over counterface asperities. In the material science literature, it is well understood that the filler/matrix interface can have a dramatic influence on a number of critical properties [25, 53-60]. In nanocomposites, the number and area of these interf aces are inherently large, and these regions can have far-field effects on matrix crystallinity, phase and morphology. These thr ee areas were systematically studied by Burris et al. ; the results are outlined in the following sections. The first quantitatively examine the morphological, tribolog ical, compositional and chemical properties of the transfer films and address their influen ces on the tribo-system. The following section examines the nature of the matr ix/filler interface and its effect on the tribology of the system. The final section discusses the phase and morphol ogy changes in the PTFE that occur as a result of nanoparticle inclusion, and the effects of these changes on the wear resistance of the nanocomposite. Investigations of Transfer Films Contrary to early suggestions that nanoparticles could not appreciably im prove the wear resistance of PTFE, it has been shown that nanofi llers can be far superior to microfillers with a transition in the dominant wear reduction mechanis m likely occurring at a particle size on the order of 100 nm. Preliminary evidence suggests that reduced counterface abrasion, reduced third body wear and retention of prot ective transfer films are pr imarily responsible for the improvements in wear resistance. Thin, uniform transfer film s consistently accompany wear resistance in the tribological nanocompos ites literature [8, 9, 11, 17, 19, 30-33, 51], but quantitative measurements of these films are lacki ng. Some authors suggest that wear resistance is due to the transfer film prot ecting the composite while others o ffer that the films are formed as a consequence of low wear. It is currently unclear why and how these films form, how they
38 facilitate wear resistance, if they are composed primarily of the PTFE, the filler or the composite, and if chemical reactions are involved. Burris and Sawyer  conducted a study with 5 wt% phase and phase aluminaPTFE nanocomposites against various rough counterf aces to study the effect of asperity size on the transfer and wear of different PTFE na nocomposites. The surfaces were made using different standard finishing techniques and inte rferometry measurements of these surfaces are shown in Figure 3-1. Figure 3-1. Surfaces used to study roughness effect s on PTFE nanocomposite transfer and wear: a) electro-polished Rq (root mean squa red roughness) = 80nm, b) lapped Rq = 160nm, c) wet-sanded Rq = 390nm, d) dry-sanded Rq = 580nm. Note that the lay of the wet-sanded surface is or iented in the direction of sliding; it is smoother in the direction of sliding than against it.
39 Figure 3-2. Wear rate plotted versus count erface surface roughness, Rq for 5 wt% 40 nm phase and 80 nm phase alumina-PTFE nanocomposites. The two phases of alumina filler produce wear rates that differ by 100X on average with different surface sensitivities. Wear rates for these co mposites are plotted vers us counterface roughness Rq in Figure 3-2. The different phases of alumina were found to re sult in widely different tribological properties despite identical processing a nd testing. Wear rates of alumina nanocomposites increased monotonically from 50-300x10-6 mm3/Nm with increased surface roughness. Additionally, wear debris were relatively large and transfer films thick and discontinuous. Wear rates of alumina nanocomposites did not correlate wi th roughness and were significantly lower than those of the nanocomposites ranging from 0.8-10x10-6 mm3/Nm. Wear rates from tests conducted on counterfaces without predominan t orientation were equivalen tly low despite roughness ranging from 80 580 nm Rq. Transfer films on these surfaces were all thin and uniform. Testing against the oriented wet-sanded surface on the other hand increased th e wear rate of the
40 nanocomposites by an order-of-magnitude. A repeat at this condition confirmed the validity of the result. Transfer films on the wet-sanded surface were incomplete, thick and banded in the direction of sliding. Transfer Film Morphology An examination of the data collected th roughout the test reveals an additional key difference between and phase alumina nanocomposites. The nanocomposites reached steady state almost immediately, while the nanocomposites had a significant transient period of moderate wear followed by a tr ansition to a lower steady state wear rate. Despite the relative insensitivity of steady state wear rates to counterface roughness for nanocomposites, the transient wear rate (during transfer film deve lopment) increased monotonically with increased roughness. Additionally, the total volume remove d during the transient portion of the test increases with increased roughness. These results s uggest that as material is removed from the sample and deposited onto the counterface, more of the asperities become covered by a transfer film and the wear rate is reduced. Larger asper ities require more material to transfer before steady state is reached, but at steady state, abrasi on is insignificant and wear rate is independent of roughness. The orientation of the wet-sande d surface likely disrupted the formation of a stable transfer film, resulting in comparable transient and steady state wear rates. It can be concluded that the presence of a protective transf er film is necessary for low wear of PTFE nanocomposites. It is also interesting to not e that when neither composite was sufficiently protected by transfer films, either during the tr ansient region or against the wet-sanded surface, the alumina nanocomposites outperformed the nanocomposites. This suggests a difference in the wear mechanisms, which likely governs the ability of the composite to form
41 protective films during sliding. Qualitatively, tran sfer films were found to increase in thickness and discontinuity with in creasing wear rate. Thin, uniform transfer films and fine debris consistently accompany wear resistance in these studies and in the nanocomposites tribology lit erature. Global relationships between wear rates and transfer films were st udied by quantitatively measuring transfer films of widely varying PTFE-based tribo-systems using either scanning white-light interferometry or mapping stylus profilometry. These systems include 5 wt% alumina-PTFE composites with and particle phases, 40 nm, 80 nm and 0.5 m particle sizes, and counterfaces of polished, lapped, wetsanded and dry-sanded surface finishes. Wear rate is plotted as a function of maximum transfer film thickness in Figure 3-3. Despite varying part icle phase size and surface finish, wear rate is approximately proportional to the maximum thickne ss of the transfer film cubed. Not only do thicker films imply larger debris, but it is sugge sted that thick transfer films are more easily removed by the passing pin and as a conseque nce need more rapid replenishment. It is well known that under ce rtain low speed sliding conditions, PTFE deposits very thin and oriented transfer films [ 36, 37, 62-65]. The orientation produces a model sliding condition where chain entanglement is minimized and pure ax ial sliding of PTFE ch ains past one another results in the very low friction coeffi cients observed under these conditions ( = 0.03-0.07). It is hypothesized that the role of the filler is to reduce gross damage to PTFE which promotes the formation of thin, aligned PTFE films under seve re sliding conditions and enables low wear of the nanocomposite.
42 Figure 3-3. Wear rate plotted versus the maximu m transfer film thickness as measured with optical interferometry. This data includes results of 5 wt% 44 nm 80 nm and 0.5 m composites against the best and wors t performing counterfaces, and 5 wt% 80 nm and 0.5 m composites and unfilled PTFE against polished surfaces. Wear rate is proportional to the maximum transfer film thickness cubed. The friction and wear properties of the films themselves were measured using microtribometry to test the hypothesis that th in, aligned films of unfilled PTFE are wear resistant. Model films of neat PTFE were depos ited onto a thin steel foil with a sliding velocity of 254 m/s for 1000 reciprocation cycles at 25C under 6.3 MPa of normal pressure. Atomic force microscopy was used to estimate an averag e film thickness of 50 nm; an AFM image of a PTFE transfer film formed during low speed, uni directional sliding on bare polished silicon is shown in Figure 3-4. It is eviden t that such films are highly aligned in the sliding direction and on the nanometer size scale.
43 Figure 3-4. An AFM image of the transfer film produced through unidirectiona l sliding of PTFE on a polished silicon wafer. The white arrow inside the PTFE transfer film indicates the sliding direction. The si ngle line profile (sho wn in white on the trimetric view) portrays a relatively smooth film; features within the image clearly document the fibrillated and oriented nature of the transfer film. Topographic data measured across the film edge indicate a transfer film th ickness at this location of substantially less than 10 nm. AFM studies were conducted by Professor Scott Perrys group at the University of Florida. After creation, the film covered foils were cut into rectangular samples for testing. Custom designed sample mounts fixed opposing foils into a crossed-cylinder geometry. This geometry reduces misalignment sensitivity, minimizes edge effects and helps reduce pressures to values more typical of those found in macro-scale testing. Parallel (chains oriented in the direction of sliding) and perpendicular (chain s oriented against the direction of sliding) aligned films were tested to study the hypothesized tribological anisotropy of aligned PTFE films. Normal and friction forces were continuously measur ed at the stationary pin, while a 600 m reciprocation displacement was imposed on the counterface. Tests with an averag e sliding speed of 100 m/s and a normal load of 500 mN were conducted ov er 250 sliding cycles. The contact patch was estimated with ex-situ optical observation to be 200 m in diameter; this translates into an average pressure of 15 MPa. The results of the microtribometry experiments are shown in Figure 3-5.
44 Figure 3-5. Microtribometry friction results for the crossed cyli nder oriented PTFE transfer film tests. Friction coefficient is examined versus reciprocation cycle for a) parallel and c) perpendicular configuration. The evol ution of friction coefficient along the reciprocation track is also plotted for both the b) parallel and d) perpendicular configurations. The perpendicular alignment of the films leads to rapid failure of the films. In line with the hypothesis that orientation in the sliding dire ction facilitates low friction and wear, perpendicular alignment of the films led to complete failure of the film (denoted by >0.2) in about 10 cycles, while parallel aligned f ilms were at least 10X more wear resistant.
45 Despite having similar average values of friction coefficient for the first few passes, differences can be seen in the positionally resolved fricti on data on the right of Fi gure 3-5. Examining the first pass, the parallel sample has a steady friction loop, whil e the perpendicula r friction loop has significant scatter. The mechanism of moti on accommodation appears more damaging in the case of the perpendicularly aligned films, and th e tendency of these films to reorient into the direction of sliding is likely re sponsible for the erratic friction and wear behavior. The parallel aligned films have much lower wear presumably as a result of the stable orientation. A hybrid configuration, which consisted of a perpendicular top film and a parallel bottom film, was created to test the hypothesis that sliding preferentially occurs at parallel interfaces in samples with different possible sliding orientations. The frictional behavior was nearly identical to that of the parallel configuration which suggests that the parallel interface is the preferred interface for sliding. Since protective transfer film s are necessary to reduce wear of these nanocomposites against counterface asperities, the wear rate of the transfer film places a lower limit on the wear rate of the composite. An estimate of wear rate for the parallel aligned films was calculated to determine whether model films of unfilled PTFE could possibly support low wear sliding. The contact area on the top foil (pin) and bottom foil (counterface) are 0.033 and 0.12 mm2, respectively, so failure of the top film should occu r first, followed by dire ct asperity contact and rapid deterioration of the bottom film. Failure of the top film in parallel alignment occurred after approximately 150 cycles. The wear ra te in this case is calculated as, 63 5 235010 210 15/1501.210/ x mm mm kx NmmcyclesxmcycleNm Eq. 3-1 Despite the superiority of the parallel aligned transfer film estimation of the wear rate reveals that rates are still orde rs of magnitude higher than those found for many low wear PTFE
46 nanocomposites (~10-5 versus 10-7 mm3/Nm). It can be concluded that even model thin and aligned transfer films of PTFE are incapable of supporting low wear sliding. The films formed by sliding of a low wear nanocomposite must ther efore be comprised of composite material or some more wear resistant variant of PTFE. Similar experiments were conducted for transf er films of a low w ear 10 wt% PEEK filled PTFE composite. In each configuration, the com posite film had lower and more stable friction coefficients for the duration of 1000 cycle test s with no obvious signs of wear in post test analysis. Clearly, the compositions and chemistries of these fi lms are additional factors that require quantification for a more complete understanding of these nanocomposite systems. Transfer Film Composition Though X-Ray photoelectron spectro scopic (XPS) chemical and compositional analyses of unfilled polymer and micro-composite transf er films have been conducted by several investigators [48, 49, 66], it is unclear how comp osition and chemistry evolve in the transfer films of PTFE nanocomposites or how this evolution influences tribological phenomena observed during testing. XPS was used to test the hypothesis that transf er films consist of composite material. PTFE nanocomposites were created with 1/8%, 1/2 % and 1% (by volume) loadings of 40 nm alumina nanoparticles. In a control set, the particle s were untreated and in a second set of samples, the nanoparticles were treat ed with a fluorinated silane. The treatment was hypothesized to improve dispersibility a nd compatibility with the matrix. Treated (fluorinated) and untreated samples were compression molded, machined and tested. The tribological experiments were conducted on a linear r eciprocating tribomet er. Photoelectron spectra of the C, O, F, Al and Si regions we re collected using a PHI 5700 Xray Photoelectron spectrometer equipped with a monochromatic Al K X-ray source (h =1486.7 eV) incident at
47 90 relative to the axis of a hemispherical an alyzer. The spectrometer was operated at high resolution with a pass energy of 23.5eV, a photoelectron take off angle of 45 from the surface normal and an analyzer spot diam eter of 1.1mm. All spectra were collected at room temperature with a base pressure of 1 x 10 -9 torr. Electron binding energies were calibrated with respect to the C1s line at 291.6eV (C-F). Even at very low loadings, alumina was found to transfer to the counterface with the PTFE (trace amounts of Si were also found in fluorinated samples). The atomic fraction of aluminum in the transfer films is plotted versus the alum inum content in the bulk in Figure 3-6. Aluminum in the transfer films of fluorinat ed samples (filled circles) was found in direct proportion to the loading in the bulk sample. The untreated 1/8% sample had an unexpect edly high amount of aluminum in the film, possibly due to poor di spersion or agglomeration during processing. Alternatively, aluminum could have accumulated in the transfer film with wear of the sample -this sample had 50X higher wear than any of the other samples -Blanchet and Han [67, 68] have previously described this mechanism as one of preferential rem oval of PTFE from the system which leaves the in terface rich with filler.
48 Figure 3-6. Aluminum atomic content (%) in th e transfer films of virgin and fluorinated nanocomposites plotted versus the atomic content as prepared in the bulk. Measurements were made in the center of each wear track. In low wear samples the aluminum content in the transfer film is linearly proportional to the aluminum content in the bulk sample. Evidence of filler accumulation was observed for the less wear resistant sample. XPS st udies were led by Professor Scott Perrys group at the University of Florida. Similar XPS analyses of transfer films fo rmed from both types of alumina particles revealed a higher proportion of oxygen than w ould otherwise be predicted based on the aluminum present, suggesting tribo-chemical oxidation of the PTFE during extended sliding. Furthermore, these measurements demonstrate a correlation between friction coefficient and the oxygen content of the transfer film (Figure 3-7). It is unclear whether an increase in friction of the fully formed transfer film drives the oxidatio n or if oxidation itself, occurring in the creation of the transfer films, leads to the higher friction coefficients.
49 Figure 3-7. Friction coefficient plotted versus oxygen content as measured using X-Ray photoelectron spectroscopy. The symbol u denotes untreated alumina and f denoted fluorinated alumina. XPS studies were led by Professor Scott Perrys group at the University of Florida. The transfer films formed from PTFE na nocomposites containing fluorinated alumina exhibited frictional behavior uncha racteristic of PTFE; friction coe fficients were lowest at the reversals (ends of the wear track ) and highest in the center. Based on this behavior and the oxygen-friction coefficient correlation presente d above, it was predicte d that oxygen content decreases toward the ends of th e wear track where friction was found to be lowest. Figure 3-8 shows friction coefficient, transfer film thic kness and oxygen content as functions of the pin center track position over half of the track for a 2 wt% fluorinated 40 nm phase alumina-PTFE nanocomposite. XPS was used to determine oxygen content in the films and three dimensional mapping stylus profilometry was used to map film thickness; the thickness envelope shown in Figure 3-8 reflects averaged results of mapping stylus profilometric measurements of the transfer
50 film plus and minus one standard deviation. Both oxygen content and thickness correlate well with the friction coefficien t along the wear track for th e final cycle of sliding. Figure 3-8. Transfer film thickne ss, oxygen content and friction coefficient plotted versus track position over half of the wear track for a 1% fluorinated sample. The transfer film thickness envelope represents the mean plus and minus one standard deviation. The position dependent friction coefficient corr elates well with both oxygen content and thickness. Transfer Film Chemistry The oxidation of PTFE is initially surpri sing given its known chemical inertness. However, very wear resistant materials produce ve ry thin transfer film s that are exposed to prolonged frictional energy dissipation at the in terface. Over extended sliding distances, sufficient energy can be absorbed by these thin la yers to initiate even low probability chemical events. XPS analysis of a 5 wt% 80 nm alumina transfer film after sliding with a wear rate of 10-7 mm3/Nm further probed the tribo-chemical degrad ation of low wear PTFE. The comparison of the core level C 1s spectra of unfilled PTFE to that of the nanocomposite transfer film shown in Figure 3-9 demonstrates clear evidence of a ch emical transformation of PTFE in the process of transfer film deposition and wear.
51 Figure 3-9. Comparison of the core level C 1s spectra of unfilled PTFE to that of the nanocomposite transfer film. The appear ance of a new peak at 288 eV provides evidence of a chemical transformation in the PTFE that occurs during wear. XPS studies were led by Professor Scott Perrys group at the University of Florida. The data confirm a reduction in C-F intensity at 292 eV, consistent with de-fluorination of the transfer film and the measured reduction in th e F 1s integrated intens ity (data not shown). This change is accompanied by the appearance of new C species giving rise to intensity at 288 eV and 284 eV. While the relative changes in the 292 eV and 284 eV regions could be rationalized through the adsorpti on/deposition of adventitious carbon during sliding or sample preparation, the 288 eV feature can be correlated with the relative wear resistance of filled PTFE transfer films. Several investigators have inte ntionally degraded PTFE usi ng various techniques including gamma ray, electron beam, and ultraviolet irradia tion and have correlated chemical degradation with changes in mechanical proper ties [69-72]. For example, Zhao et al.  noted a rapid fluorine loss on the surface of PT FE when exposed to vacuum u ltraviolet radiation as well as increased optical absorbance due to carbon exposure which resulted in a brownish appearance of the sample. Similar brown discolorations are of ten found in low wear tr ansfer films. Lappan et
52 al. [74, 75] and Oshima et al.  used infrared spectroscopy to identify various reaction products from the degradation, noting processes involvi ng defluorination and chain scission which produce terminal and branched CF3 groups, carbon-carbon double bonds and branched (crosslinked) carbon structures. COF and COOH have also been observed. High speed magic angle spinning nuclear magnetic resona nce (HS MAS NMR) spectroscopy measurements of degraded PTFE by Katoh et al.  and Fuchs et al.  provided evidence of cross-linking in degraded PTFE. Oshima et al.  used the HS MAS NMR data to identify chemical structures of CF2, CF (cross-linked) and =CF (double bonded carbon) in XPS spectra. These degraded PTFE structures showed very similar spectroscopic signatures to the 288 eV peak found in wear resistant transfer films. Blanchet et al. [14, 79, 80] previously conducted tribol ogy experiments on irradiated PTFE and FEP composites and demonstrated that the wear resistance of each was improved by several orders-of-magnitude with 30 Mrad doses of el ectron irradiation. More recently, an analogous study was conducted to further explore the influen ce of degradation on the tribological properties of PTFE. A commercial chemical etch was us ed to emulate tribo-chemical degradation by stripping fluorine from the surface of an unfilled PTFE wear sample. Intensity at 288 eV was verified in the XPS spectrum of the etched surf ace. The tribological properties of the degraded PTFE surface coating were measured for linear re ciprocating sliding at 6.3 MPa and 50 mm/s. The degraded PTFE was found to be 100X more wear resistant than unfilled PTFE, verifying that conjugated PTFE offers the possibility of in creased wear resistance. This period of wear resistance was followed by a sharp increase in wear and a 10% reduction in friction coefficient as the more wear resistant degraded surface wore th rough to the virgin PTFE beneath. The rate of material consumption due to wear here was greater than the rate of tribo-chemical degradation,
53 and it is likely that in these nanocomposite systems, the degradation mechanism supplements the other more dominant mechanisms th at enable degradation to occur. Role of Transfer Films It has been shown that transfer film morphology, composition and chemistry all play important roles in determining the wear rate of th e films and thus the tribo-system. It has also been shown that certain surface characteristics can destabilize the transfer film which disables low wear sliding. It is clear that the presence of a high quality transfer film is a necessary condition for low wear sliding of PTFE nanocompos ites; it is unclear wh ether it is a sufficient condition for low wear. To te st this hypothesis, an alumin a-PTFE nanocomposite known to ordinarily produce poor quality transfer films (5wt% 44 nm alumina) was tested upon a predeposited transfer film formed under u ltra-low wear sliding conditions (k<10-8 mm3/Nm ). It was found that the composite ha d the same wear rate (7x10-5 mm3/Nm) whether it was tested against a wear resistant transfer film or a fresh counterface. Desp ite the presence of an ultra-low wear transfer film, thick platel ets indicative of delamination wear were deposited on top of the pre-existing transfer film. The abra sive wear that is reduced by the transfer film appears to have a negligible influence on the w ear rate of this nanocomposite due to the severity of its delamination wear. In-situ optical microscopy of the wear tr ack revealed that the transferred material is very unstable, moving appreciably af ter each cycle. Clearly the mechanics of the nanocomposite itself dominate the properties of th is system and govern the development of the transfer film. These results sugge st that while thin transfer film s are required for low wear, they are more likely a consequence of the low wear debris morphology than the source of wear resistance.
54 Investigations of Internal Interfaces The critica l role of transfer films in enab ling low wear of PTFE nanocomposites has been demonstrated, but bulk properties of the com posite seem to dictate the initiation and development of the films as well as the ability of the composite to achieve low wear against high quality transfer films. Bahadur and Tabor  and Blanchet a nd Kennedy  noted a trend of decreased wear debris size with decreased wear ra te and suggested that th e primary role of the filler was to reduce the si ze of the wear debris. Because the wear rate is proportional to volume, which is proportional to the cube of a characteris tic diameter of the debr is, reducing debris size inherently reduces the wear rate, promotes engagement of debris with the surface and improves transfer film stability. Burris and Sawyer  hypothesized that the size and shape of the debris during run-in were critical in the development of these transfer films, and that the film morphology observed during low wear was a conseque nce of low wear rather than the cause. They envisioned a wear model proposed by Blan chet and Kennedy , where the cracks that lead to the destructive delamination in PTFE were effectively arrested by the filler, resulting in reduced debris size and stable transfer film formation. The strength of the interface would have a critical influence on such a system. Often, nanoparticles and polymer matrices are inert by design to limit environmental sensitivity of the tribological response. This inertness limits chemical interaction at the filler/matrix interface and can lead to inherent weakness. Wagner and Vaia  articulated the importance of the bonding at the in terface for a nanotube reinforced polymer. They calculated interfacial shear strength to be 3 MPa with only van der Waals interactions present, and in excess of 100 MPa with only 1% covalent bonding of th e carbon atoms. Many investigators have successfully improved interfacial bonding in compos ites with surface coatings that compatibilize the filler with the matrix. He et al.  found improved mechanical properties and dispersion
55 when the nanoparticles were plasma-modified, and Eitan et al.  found improved load transfer via strain dependent Raman spectroscopy and impr oved bulk mechanical properties of a treated MWCNT filled polycarbonate over the nanocompos ite with untreated nanotubes. While the studies of Burris and Sawyer  showed a str ong dependence of wear rate on alumina phase suggesting the potential importance of these in ternal interfaces in tr ibology, it was unclear whether observed differences were due to particle phase or size since both phase and size varied in the experiments. Recently, a series of experiments were conducted on PTFE nanocomposites with smaller phase alumina to evaluate the potentia l size effect. Figure 3-10 shows TEM images of the 44 nm 40 nm and 80 nm phase nanoparticles. The phase particles have an irregular plate-li ke morphology while the particles are spherical. Figure 3-10. Transmission electron images of a) 44 nm b) 40 nm and c) 80 nm particles.
56 Figure 3-11. Wear rate plotted versus alumin a loading for alumina-PTFE nanocomposites with varying nanoparticles against a lapped counterface: a) 44 nm phase alumina, b) 40 nm phase alumina, c) 80 nm phase alumina. Error bars represent the combined standard uncertainty in the wear rate. Tribological experiments were conducted on na nocomposites of these particles at various loadings against lapped 304 stainless steel counte rfaces in standard laboratory conditions at 50 mm/s and 6.3 MPa of normal pressure. Wear rate is plotted versus alumina loading in Figure 311. The wear rates of the 40 and 80 nm alumina-PTFE nanocomposites are insensitive to size and loading in the range from -5% filler and these partic les provide additional 100-10,00X improvements in wear resist ance over comparably loaded phase alumina-PTFE nanocomposites. It can be concl uded that the differences in wear rates are not attributable to particle size, and although these dispersions have not yet been characterized, the difference is thought not to be due to disp ersibility since the 40 nm phase particles have higher specific surface area and were observed to agglomerate subs tantially more than the other particles during
57 nanoparticle imaging. Though the w ear reduction mechanism of the phase particles remains unclear, it appears to be related to the nature of the interface. It was hypothesized that additi onal gains in tribological performance could be achieved by compatibilizing the nanoparticles with the ma trix. A nanoparticle surface fluorination was thought to not only provide compatibility with the matrix, but the decrease in nanoparticle surface energy due to the fluorination was thought to aid dispersion; poor dispersibility was suspected as the source of th e unusual scatter in the 40 nm nanocomposites. The 40 nm phase particles were chemically treated with 3,3,3 Trifluoropropyl Trimethyoxysilane. Infrared absorption spectroscopy confirmed the presen ce of the fluorinated groups and thermal gravimetric analysis was used to estimate the mass fraction at 3%. Nanocomposites with filler loadings of 1/8, 1/2 and 1 % untreated and fluorinated nanopart icles were tested to investigate the effects of the interface treatme nt on the tribological properties of the nanocomposite. Wear volume is plotted versus the slid ing distance in Figure 3-12a. Figure 3-12. Tribological result s of test with untreated and fluorinated silane treated 40 nm phase alumina-PTFE nanocomposites: a) worn volume plotted versus sliding distance b) wear rate plotted versus filler loading Linear reciprocation experiments were conducted against lapped 304 stainless steel counterfaces under standard laboratory conditions with a normal load of 250 N.
58 In general, the fluorinated samples were very well behaved in comparison to the untreated samples. Transients are steeper and longer las ting with more material removed for less filled samples. The untreated alumina nanocomposite s behaved erratically by comparison. Steady state wear rates are plotted versus filler loading in Figure 3-12b. The large sample to sample variation of the untreated nanoc omposites is clear with the low wt% samples having nearly an order-of-magnitude difference in we ar rates of identically prepared samples. The wear rates of the functional nanocomposites differ by less than 2X from 1/8% to 1% and decreases linearly with increasing filler loading; variations in wear rates of 2X or less for PTFE nanocomposites are rare even for sample to sample variations. With the addition of 1/8% functional nanoparticles, the wear resistance of PTFE was improved by over 5,000X. It is unclear whet her the variations in wear rates were dominated by dispersion, interface strength or both. Investigations of PTFE Phase and Morphology Previous studies have shown that subtle changes such as nanopartic le shape or phase can have dram atic effects on the wear rate of the PTFE nanocomposite. A previous hypothesis was that these factors increased the particle/matrix interface strength, but, filler particle surfaces can also affect the local phase, morphology and mobility of the polymer chains at the interface. This layer of affected polymer in th e vicinity of the particle is known as the interfacial region, interaction zone and interphase. It will be referr ed to here as an interfacial region. Due to its characteristic nanometer size scale, it is ofte n appropriately neglected in microcomposites. However, the interfacial region can often be comp arable in thickness to nanoparticle fillers and can therefore dominate the properties of a nanoc omposite. In amorphous polymers, the primary impact is on the chain morphology and mobility. Sternstein et al. [59, 60] did a systematic study of the particle/matrix interface strength effects on the rheology of nanocomposites and proposed a theory for reinforcement and nonlinear viscoela sticity of polymer nanocomposites that is based
59 on the trapped entanglements of chains near th e polymer/matrix interface and the consequent farfield effects on other polymer ch ains. Maiti and Bhowmick  used atomic force microscopy (AFM) to measure interfacial thickness, and found thickness of the region to increase with increased filler/matrix compatibility. Eitan et al.  observed a similar relationship for treated and untreated nanotubes in a polycarbonate matr ix. Additionally, it was found that fracture occurred preferentially within the matrix itself rather than at the filler/matrix interface, suggesting that the interface and interfacial re gion can both be stronger than the unfilled polymer. The manifestations of interface eff ects in amorphous polymers have been observed by numerous researchers and incl ude changes in glass transiti on temperature, modulus, toughness and rheology [25, 27-29, 56, 59, 60, 81]. In semi-c rystalline polymers, nanoparticles can alter the crystalline phase [ 82], morphology [83, 84] and the degree of crystallinity [85, 86] which can extend the influence of the nanoparticles. Such changes in crystalline morphology can have significant influences on mechani cal behavior, but it is unclear whether such effects are present in PTFE nanocomposites or to what extent tribologi cal properties are altered due to these effects. The role of crystallinity and crystal phas e is explored in the following paragraphs. PTFE is a semi-crystalline polymer known to have a complex molecular organization with three phases existing near room temperature at am bient pressure. Phase II [87-92] is typically stable below 19C, and is characterized by a triclinic unit cell w ith a=b=0.559 nm and =119.3. The molecular conformation is described by Clark  as a non-commensurable 13/6 helix. As the temperature increases above 19C, phase IV becomes stable, the molecules untwist to a possibly commensurable 15/7 helix and the unit cell becomes hexagonal w ith a 1% increase in the lattice parameter (a=b=0.566 nm ) ; macroscopically, this results in an increase in volume . Additionally, Kimmig et al.  noted a rapid increase in th e density of coherent helical
60 reversal defects possibly associated with th e onset of helix commensurability. As the temperature increases above 30C, the hexagona l unit cell becomes distorted , the Bragg reflection at 2 = 42 broadens  and molecular disorder increases [85, 93, 96] as intramolecular forces begin to dominate intermolecu lar forces . It has been suggested that the observed disorder is due to axial and rota tional oscillations of molecules becoming more pronounced while helical reversal defects incr ease in length and become incoherent. While it has been shown that the degree of crystallinity has a minimal role on the wear rate of PTFE , phase and temperature have both been found to have dramatic influences on the mechanical and tribological propert ies of PTFE. Flom and Porile noted a dramatic effect of the phase of PTFE on its tribological properties . They performed sliding experiments with selfmated PTFE at speeds of 11 and 1890 mm/s and found an abrupt and reversible increase in the friction coefficient as the bac kground temperature increased above a threshold value near room temperature in both cases. They hypothesized that the increase was associated with the phase transition from II to IV at 19C. Steijn  also found increased friction coefficients as the temperature increased from 19 to 30C despite the more global trend of reduced friction coefficient with increased temperature. Ma kinson and Tabor  found evidence of strong adhesion to the counterface and proposed a lamella r effect of intercrystalline shear governed by van der Waals interactions between polymer chains. Studies by Steijn , Tanaka et al. , Blanchet and Kennedy , McCook et al.  and Burris et al.  supported this hypothesis finding frictional responses of PTFE to be therma lly activated and consiste nt with van der Waals interactions. More generally, Joyce et al. , Brown et al. [103-106] found similar dependencies of various mechanical properties to temperature in the range from 200-400K. The
61 measured property in each of these studies has been normalized by the room temperature measurement and is plotted versus temperature in Figure 3-13. Figure 3-13. Normalized propertie s plotted versus temperature fo r variable temperature studies available in the literature. The normalized property is define d as the ratio of the value at temperature to the room temperature valu e. Various tribological and mechanical experiments suggest that deformation of PTFE is a thermally activated process. Tanaka et al. [38, 46] focused on the characteristic bands of PTFE that contain both crystalline and amorphous material In a study investigating the w ear mechanisms of PTFE, they found insensitivity of wear rate to crystallinity, but found that with certain processing conditions, the characteristic size of the banded structure coul d be reduced with a subsequent reduction in the wear rate. In varying the background temperature and sliding speed of PTFE during testing, a transition to low wear rate was found to occu r as temperature increa sed past a critical temperature. At the 50 mm/s sliding sp eed used here, the data from Tanaka et al.  indicates a
62 transition temperature near the IV to I phase transition of PTFE (30C). A recent study of the effects of PTFE phase on toughness by Brown and Da ttelbaum  helps explain this result. Increased fracture toughness was found for phase I (T>30C) over phases II (T<19C) and IV (19C
63 Figure 3-14. SEM images of the worn surface of a 5% 80nm phase alumina-PTFE nanocomposite: a) low magnification, b) hi gh magnification. The mudflat cracking is a characteristic that is repeatedly observed for these wear resistant PTFE nanocomposites. Wear debris appears to be on the order of 1 m, while the cracking patterns encompass 10s of micrometers of material. The liberation of large wear debris appears to be inhibited by fibrils spanning the cracks. Figure 3-15. SEM images of the worn surface of a 5% 80nm phase alumina-PTFE nanocomposite after being fractured at room temperature: a) low magnification, b) high magnification. Fibrils completely span a 150 m crack.
64 It is clear from the literatur e that the phase and morphology of PTFE are important in determining toughness and ease of fibrillati on [103, 104]. It is also well known that nanoparticles can stabilize metastable crystalline phases in polymers and change the crystalline morphology . Therefore, one of the mechanis ms that could explain the dramatic changes in wear rate at such small particle loadings is an effect of particle loading on the polymer phase and morphology. The following studies were conducted on a wear resistant 1 wt% 40 nm a phase alumina-PTFE nanocomposite to explore the eff ects of crystalline structure, morphology and phase on the tribology of PTFE. In order to determine the role of morphol ogy in the nanofilled PTFE, the melting behavior was monitored using differential scanning calorimetry (DSC) and the structure examined using X-Ray diffraction (XRD). Fi gure 3-16 shows the XRD results of PTFE and a PTFE nanocomposite. The nanocomposite has a full width half maximum (FWHM) at the main diffraction peak (18 degrees) that is twice that of the unfilled PTFE with a larger amorphous background. This implies that the nanoparticle s have interrupted th e lamellar crystalline structure. There is also a slight shift in the pe ak value suggesting an inte rruption of the unit cell. The high temperature DSC results, shown in Figure 3-17, support this result. The neat polymer and the untreated nanocomposite have the trad itional 327C melt temperature of melt-processed PTFE. However, the nanocomposite has a much la rger melt peak at a higher temperature of about 345C; this temperature is consistent with the melt temperat ure of virgin PTFE before melt processing. The higher melt temperature is often attributed to the la rger, more perfect crystals of the virgin resin [95, 107, 108]. Despite being processed under identical conditions (360C, 4 MPa), the unfilled PTFE has melt characteristics indicative of melt processed PTFE, while the nanocomposite retains the melt characteristics of virgin resin. The nanoparticles may impede the
65 mobility of PTFE molecules, re quiring greater amounts of thermal energy before reorganization of the PTFE to the lower or der gel state can occur. Figure 3-16. X-Ray diffraction for neat PTFE, a low wear nanocomposite and the same nanocomposite after a 400C heat treatm ent: a) the main reflections at 2 =18 and b) the phase sensitive region from 30 < 2 < 45. XRD studies were led by Professor Linda Schadlers group at Rensselaer Polytechnic Institute. Figure 3-17. Differential scan ning calorimetry (DSC) of neat PTFE, a low wear nanocomposite and the same nanocomposite after a 400C heat treatment: a) in the phase sensitive, low temperature region and b) in the hi gh temperature region during PTFE melting. Phase II typically exists at temperatures below 20C, IV between 20C and 30C, and I above 30C. DCS studies were led by Professor Linda Schadlers group at Rensselaer Polytechnic Institute.
66 One hypothesis for wear resistance of these syst ems is that the virgin structure facilitates fibrillation of the nanocomposites du ring wear. Another is that th e mechanical destruction that occurs during jet-mill processing results in a fi brillated structure that is stabilized by the nanoparticles to temperatures above the process te mperature. In either case, it is hypothesized that the nanoparticles stabilize otherwise unstable structures to elevated temperatures. To test this hypothesis, the sample was characterized fo llowing a 400C heat treatment designed to anneal the crystalline structure induced during the origin al processing. After the heat treatment, the XRD peak at 18 degrees has sharpened and returned to the same position as the unfilled PTFE (Figure 3-16). In the region most sensitive to the phases present, the 1 wt% nanofilled resin has the lowest peak height to amorphous b ackground ratio (1.5 to 1) compared to the neat resin (3 to 1) and the heat treated 1 wt% nanofil led resin (about 2 to 1) As the proportion of phase I increases, the peaks (especially the peak at 42 degrees) decrease in intensity compared to the background. Because these samples are a mixt ure of at least 2 phases, it is difficult to ascertain exactly what structure is present. These results, however, imply that the 1 wt% nanofilled resin has the most phase I and the least phase IV of the materials shown and that the heat treatment to 400C leads to the most phase IV. This is also supported by the DSC data at low temperature shown in Figure 3-17. The data show that the 1 wt% nanofilled sample has much less of a phase I to IV tr ansition than the unfilled PTFE, while the heat treated composite shows a significant I-IV transiti on. Therefore, one would expect more phase I in the 1 wt% nanofilled composite and the least in the heat treated 1 wt% nanofilled composite. In addition, the transformation temperature from I-IV is lo wer for the 1 wt% nanofilled sample further supporting the argument that it has the most phase I present. Finally, at room temperature the IV-II transition has begun in the heat treated sample; Brown a nd Dattelbaum found that phase II
67 had the lowest toughness . This would also explain the significantly higher ordering in this sample seen during XRD measurement. Both XRD and DSC provide evidence that more of the tough phase I is present in the untreated nanocomposite than either the heat treated nanocomposite or the unfilled PTFE at room temperature. Atomic force microscopy was used to probe the PTFE morphologies in the nanocomposite before and after heat treatment. The AFM measurements shown in Figure 3-18 support the features observed in XRD and DS C. Clear differences can be seen before and after heat treatment with the nanocomposite before h eat treatment having much thinner and more organized lamellae than it had after heat treatment. Before heat treatment, the lamellae are well aligned and appear to have the folded-ribbon mor phology that is often cited for virgin PTFE [95, 107, 108], while after heat treatment, lamellae are thick, tangled and packed into substantially larger characteristic regions. Marega et al.  suggest that the reduc tion in the IV-I transition peak and the shifts in the phase transition temperature to lower temperatures (DSC) are indicative of increased molecular disorder and thinned lamellae. XRD and AFM measurements of the control nanocomposite show evidence of increased molecular disorder, and thinned lamellae, respectively. These characteristics may explain the stability of phase I for this sample at lower temperatures, and the high degree of la mellar alignment over the field of view supports the hypothesis that the nanoparticles stabilized the virgin morphology during processing.
68 Figure 3-18. Atomic force microscopy of the crystalline morphology of a 1 wt% 40 nm aluminaPTFE nanocomposite a) before heat treatment, b) after heat treatment, c) before heat treatment higher magnification and d) after heat treatment higher magnification. Asprepared, the nanocomposite has very small lamellae with finer-scale, more ordered packing than after heat treatment. Regions c) and d) are outl ined in a) and b), respectively. AFM studies were led by Prof essor Scott Perrys gr oup at the University of Florida.
69 To test the hypothesis that the effects of the nanopartic les on the morphology of PTFE reduce wear, tribological experi ments were conducted on the same nanocomposite before and after heat treatment. The control expe riment was conducted on the 1 wt% 40 nm alumina filled PTFE sample in linear reciprocation with 50% relative humidity at a temperature of 25C. The sliding speed and normal pressure were 50.8 mm/s and 6.3 MPa, respectively. After standard processing, the nanocomposite was tested for approximately 80 km. The sample was then removed and heat treated. A new counter face was used, the sample was faced, removing only the material necessary to make it flat, and the experiment was continued. After the heat treatment, the friction coefficient was reduced by about 10%, wear rate increased by 100X and transfer film thickness and discont inuity increased. The tribologi cal results are shown in Figure 3-19. The tribological benefits of the crystal line effects of the nanopa rticles on the PTFE are vast. The nanoparticle phase, chemistry, size, loading and dispersion ar e nominally identical before and after the heat treatment, so the increase in wear rate is attributed only to these effects on the crystalline structure of the PTFE.
70 Figure 3-19. Wear volume plotted versus sliding di stance for a 1 wt% 40 nm alumina-PTFE nanocomposite before and after a heat tr eatment to 400C. The microstructural benefits of the nanoparticles are lost after heat treatment and the wear rate increases by 100X. Optical images of the transfer film before and after heat treatment are shown on the right. A variable temperature tribolog ical experiment was conducted to test the hypothesis that a transition to phase II results in reduced toughness and increased wear. An additional nanocomposite test specimen was cut from the original low wear 1 wt% 40 nm phase aluminaPTFE puck described earlier. The sample was run-in on a fresh counterface for 2500 m of sliding with 6.3 MPa of pressure at 40C The wear rate during run-in was 6x10-7 mm3/Nm; this is consistent with the run-in observed in the orig inal experiment. Forecasting from the original experiment, the wear rate after a 2500m run-in is near steady state and on the order of k~10-7 mm3/Nm. After an additional 25 m of sliding, th e temperature was continuously decreased over the next 30 m to a target of 10C; the dew point temperature was 8C. Because thermal expansions during the test dominate volumetric measurements, in situ displacement measurement techniques could not be used. In or der to capture the transition from low to high wear, a videoscope was used to track debris generation at the reversal of each cycle.
71 The results of the varied temperature experime nt are shown in Figure 3-20 with wear rate and counterface temperature plotted as functions of sliding distance. The wear rate curve was estimated from mass loss measurements and in-situ observation of the transition. The data shown here was recorded after the initial run-in period. For the first 47 m of sliding, no observable debris was liberated fr om the contact indicating retenti on of the low wear state. At 47 m, the temperature was 14C and the first we ar fragment appeared. This was followed by rapid deterioration of performance until a steady state wear rate of k ~ 5x10-4 mm3/Nm was reached at 52m and about 12C. Figure 3-20. Wear rate and counterface temper ature versus sliding distance for variable temperature tribology testing of a wear resi stant PTFE nanocomposite. Wear rate is an estimate based on mass measurements and in-situ observation of the transition. Normal pressure and sliding speed were 6.3 MPa and 50 mm/s, respectively. An abrupt and drastic change in the wear mechanism occurs as the temperature is reduced below about 14C. It was hypothesized that the wear resistance of the nanocomposite was due to a disruption of the crystalline structure by the nanoparticles that resulted in increased disorder, stabilization of
72 the higher toughness phase to lower temperatures and fibrillation unde r stress. The results from DSC of this sample are plotted versus temperat ure in Figure 3-21. The transition to high wear occurs when most of the sample has transitione d to the low toughness phase II. In addition to being more brittle, Brown et al.  found phase II to be stronge r and stiffer than phases I and IV. It is possible that as temp erature is dropped and the material begins to transition to phase II, deformation preferentially occurs in the remainin g phase IV material. At a critical point, enough of the material has been convert ed such that phase II must cont ribute to motion accommodation. Eventually, crack initiation, propa gation and gross failure occur. The fact that unfilled PTFE does not exhibit high wear resistance at comparab le locations on the phase diagram suggests that facilitation of fibrillation is one of the critical wear resistance mechanisms in PTFE nanocomposites. The breakdown of wear perfor mance coinciding with transition to phase II suggests that the beneficial fibril related toughening mechan ism that results from the nanoparticle filler, is disabled in phase II. Figure 3-21. Normalized DSC power plotted vers us temperature. The transition to high wear occurs at a critical point after the low toughness phase II transition temperature.
73 Summary The contributions of nanoparticles span well beyond the traditional rules of mixtures models that adequately describe many of the commercial PTFE microcomposites, with nanoparticle dispersion, surface morphology and chemistry likely driving the fundamental wear resistance mechanisms that synergistically c ouple to produce low wear. Recent experiments indicate that tenacious, high quality transfer films are necessary but not sufficient for wear resistance, and likely result from a pre-existi ng condition of wear resi stance. In addition, mechanical effects such as crack pinning and deflection are unlikely sour ces of the substantial improvements in wear resistance observed at lo w nanoparticle loadings. The filler/matrix interface appears to have a dramatic impact on the mobility of the polymer, and the resulting effects on polymer morphology and phase likely drive the tribological properties. In one experiment, an annealing heat treatment erased many of the morphological signatures of the low wear PTFE nanocomposite and increased wear by 100X despite the fact that critical parameters such as nanoparticle loading, dispersion, size, shape and phase remained nominally unchanged. These findings strongly suggest that the nanoparticle effects on the matrix phase and morphology dominate other wear resistance mechanisms in PTFE, but many unanswered questions remain regarding the exact nature of this particular w ear resistance mechanism. The goal of the present study is to answer some of these basic remaining questions. Thermal, mechanical and tribological properties of neat PTFE are characte rized as functions of mechanical and thermal history to eliminate the effects of the nanopa rticles. Nanoparticle dispersion, thermal, mechanical and tribological characterization of PTFE nanocomposites follow. Mechanical and thermal histories are removed as variables to study the direct eff ects of the dispersion technique, nanoparticle loading, dispersion and phase.
74 CHAPTER 4 EXPERIMENTAL METHODS Materials DuPont 7C PTFE is used as the m atrix materi al in these studies; it is a virgin, compression moldable PTFE resin. It has a molecular weight that ranges from approximately 106-107 Da and an average manufacturers claimed particle size of 35 m. As received, the particles are agglomerated into aggregates on the 100 m 10 mm size-scale. Scanning electron images of a representative agglomeration are shown in Figure 4-1. Figure 4-1. Secondary electron images of Teflon 7C as received from DuPont. Agglomerations range from 100 m 10 mm. The average individual pa rticles size is reported to be 35 m. This value is consistent with m easurements made using scanning electron microscopy (SEM). The primary filler selected for use in this study is Al2O3, or alumina. Alumina is very hard, wear resistant, chemically stable, and wide ly available as laborator y and commercial grade materials with particle diameters ranging from tens of nanometers to millimeters. Variables related to the alumina in this study are filler loading (volume %), average particle size (APS) and crystalline phase. The crystalline phases used in this study are 70% : 30% (referred to as ) and 99% alpha (referred to as ), and have spherical and irregular particle shapes, respectively.
75 Powder Blending The sam e unique properties that make PTFE an attractive solid lubricant also make PTFE nanocomposites inherently difficult to process. The most effective dispersion techniques, namely liquid dispersion/in-situ polymerization and high shear ra te melt mixing are precluded for PTFE. As a result, various reputably inefficient dry powder blending techniques are commonly used; the most typical techniques are examined here. One method used here is absent from the litera ture and will be referred to as a shear sieve (a common powder sugar sifter). This technique is studied here because it is low energy and nondestructive, but has the potential to fibrillate PTFE. Th e shear sieve consists of a hemispherically shaped 20 mesh sieve through whic h the powers are classified. A rotating rod interferes with and stretches the sieve; the pow ders separating the two elements are compressed and sheared as the bar scrapes across the sieve. Agglomerates are broken and powders are sheared into a collection bin. This process is repeated until all of th e powder has been classified. The second blending method studied here is referred to as a rotary shear mixer; this apparatus was described previously by McElwain . Briefly, a mixing arm is rotated in one direction. At the end of the arm is the mixing cup which is mounted at an angle that is about 30 off axis from the arm. This cup is rotated in the opposite direction about its own axis of revolution. This motion path creates high accele ration gradients and rapid mixing. Powders in this study are mixed at an arm speed of 1500 RPM for 30 seconds. The third method under study is ultrasoni cation [10, 11, 17, 18, 32, 33]. Powders are placed in an ultrasonic ethanol ba th for 30 minutes. The forth technique under investigation is blade mixing ; the blade mixer used here is essentially a high spee d, high power blender. Powders were placed in the blade mixer and a blade speed of 30,000 RPM was maintained for 40 seconds.
76 The fifth blending method is jet-milling; this technique has been described in detail previously by Sawyer et al. [8, 16, 19]. Briefly, the powders are fed into an alumina lined grinding chamber with a high speed circumferent ial flow of dry air. Upon entry into the chamber, powders are impacted at high speed, a gglomerates are broken, and hard nanoparticles are imbedded into softer matrix particles. Centri fugal forces drive the heaviest particles to the periphery for further grinding while small particle s are preferentially forced into the collection chamber in the center. Sample Preparation After blending, powders are sintered using co m pression molding. The powder ensemble is placed in a mold and inserted into a compressi on molding station. Two PID controlled heating platens are mounted to the pressu re platens of a hydraulic laborator y press. The 7C PTFE used in these studies has a reported maximum initial melt temperature of 352 C, and enters a viscous gel state rather than a liquid st ate. As the temperature is incr eased, the viscosity decreases until degradation begins at a temperature above ~425 C. Following cooling and recrystallization, subsequent melting occurs at 327 C. Since the material is under pressure in the mold during processing, the actual temperature re quired for melting is likely a bit higher. The samples in this study are first cold pressed at 60 MPa for 15 minutes to evacuate the air fr om the powder. After cold pressing, the pressu re is reduced to 4 MPa where it remains for the duration of the schedule; this hold pressure evacuates poten tial volatiles from the powder en semble. Next, the temperature is increased to 380 C at a rate of 2 C/min. Upon reaching 380 C, the temperature drops to 362 C where it is held for 3 hours. The temperature spike to 380 C occurs due to the tuning of the PID controller (overshoot), and ensure s initial melt is reached by the entire part. Once the hold
77 schedule is complete, the mold is cooled at 2 C/min. The compression molded puck is machined to final sample dimensions using computer numerical control. Dimensional measurements are made with a digital micrometer. This measurement has a reported repeatability of 1 m resulting in a Type B uncertainty of 5 m (5X the reported repeatability). Sample mass is measured on a precision analytical ba lance having a Type B uncertainty of 50 g. The density of the sample is calculated following dimensional and mass measurements using the following formula. m lhw Eq. 4-1 where m is the mass, l is the length, h is the height and w is the width. According to the law of propagation of uncertainty, the square of the uncertainty of th e measurand (density in this case) is equal to the sum of the sensitivities of each measurement. The sensitivity of a measurement is defined as the square of the product of partial derivative of the measurand with respect to the measurement and the uncertainty of the measurem ent. The square of the uncertainty in the density is calculated as follows, 2222 22222 2221 ()cccccmmm Uumuluwuh lwh lwhlwhlwh Eq. 4-2 Characterization Morphology and Dispersion A Tescan Vega XMU variable pressure scanni ng electron microscope (SEM) was used to characterize PTFE morphology and na noparticle dispersion in this st udy. This microscope has a differentially pumped column that enables ba ckscattered and secondary electron imaging at pressures ranging from 10-2 to 103 Pa. This high pressure ca pability is essential for the characterization of the nonconducting polymeric tr ibology samples used in this study. At low
78 pressures or high vacuum, the nonconductive nature of these samples results in a high concentration of charge and heat near the pr obe during electron impingement. Thin conductive coatings are typically applied to the surface to combat the issues of heating and charging, but such coatings are invasive and destructive, potentially introducing misleading artifacts and rendering the tribological sample us eless following observation. Thermal Properties Thermal characterization was conducted using a TA instruments Q20 differential scanning calorimeter (DSC). A schematic representation of this system is shown in Figure 4-2. The sample bases are heated and the temperature difference between the sample and the reference sample is measured. The heat flux is obtained using the following formula, T q R Eq. 4-3 where q is the power conducted between the two samples, T is the temperature difference and R is the thermal resistance between the samples. The reference pan is empty and the sample pan contains the polymeric sample. Because the sample pan has a higher thermal capacity (due to the additional mass and thermal capacity of the sa mple), the reference pan temperature increases more rapidly causing a temperature difference during heating and resulting in conductive heat flow to the sample pan. This power or temper ature difference is approx imately constant due to constant differences in heat capacity until glass, phase or state transitions occur. A glass transition results in a change in heat capacity wh ich is detected as a change in the baseline power or the temperature difference between the samples during heating or cooling. Phase and state transitions require heat to rearrange the material into the more thermodynamically stable configuration before temperature can increase. Th ese transitions are detected as peaks or wells in the thermogram.
79 Figure 4-2. Schematic representations of the components comprising the TA instruments Q20 differential scanning calorime ter (DSC); a) the component s responsible for heating and cooling the sample, b) the sensor and c) the sensor with the sample and reference sitting on the respective sample platforms. Schematics from www.tainstruments.com. The heart of the design is the thermal cell wh ich contains the samples. The two sample pans are designed with thicker and significantly flatter bases than those of traditional DSC pans to reduce thermal contact resistance. In addition to the structural improvements of the pans, repeatability is improved with a self-locking pa n crimping press that en sures pans are crimped with consistent force. These crimped pans rest on polar ends of the monolithic constantan sample holder. Directly beneath each of the tw o sample mounts are area thermocouples. These area thermocouples are more accurate than trad itional point thermocouples. An additional thermocouple is located near th e center of the conducting material between the samples for a more accurate measurement of the transition te mperatures over the more typically reported furnace temperature. Mounted below the sample holder is a silver base which conducts heat from the furnace below to the samples. The furnace is mounted above a cooling flange via
80 nickel rods that conduct heat away form th e furnace. The axial design greatly reduces temperature gradients in the thermal cell. This DSC has a manufacturer reported absolute uncertainty in the measurement of temperature of 0.1C. The uncertainty in th e measurement of energy is reported to be 1% relative to the measured value. The instrument is calibrated using a standard indium sample and heats of fusion are reported as the Joules of energy of a given transition per gram of the sample (J/g). The uncertainty in the measurement of heat of fusion is therefore dependent upon the uncertainty in the mass measurement. In this case, the mass measurement uncertainty is 50 g. According to the law of propagation of uncertainty the square of the uncertainty in the measurement of heats of fusion is calculated as, 22 22 2 21 0.010.00005cE UHE mm Eq. 4-4 where H is the heat of fusion, m is the mass in grams and E is the heat of the transition in Joules. Considering a typical heat of fusion for PTFE of 50 J/g and a sample mass of 0.005 g, the mass and energy measurements contribute equa lly to provide an uncertainty of 0.7 J/g. Mechanical Properties Mechanical characterization was conducted us ing the MTS 858 Mini Bionix II load frame shown schematically in Figure 4-3a. The design is symmetric with a 1,000 lb load cell located at the bottom of the load frame and a steel hydraulica lly controlled ram locate d in the center of a height adjustable cross-beam provides the deformation displacements and loads. A Linear Variable Differential Transformer (LVDT) provi des closed-loop feedback for the hydraulic ram and displacement data for strain measurements.
81 Figure 4-3. A) MTS 858 Mini Bionix II load from used for characterizing mechanical properties. A uniaxial load cell measures tensile force on the sample and has an approximate Type B measurement uncertainty of 1N. A Linear Variable Differential Transformer is mounted coaxially with the servo-hydraulic ram for displacement measurement. The approximate Type B uncer tainty in displacement is 0.02 mm. B) Sample and sample dimensions. Samples for mechanical testing are created using the standard processing procedures outlined in the sample preparation section above. Following compression molding, the polymeric samples are machined to the shape shown in Figure 4-3b using computer numeric
82 control (CNC) for consistent sample dimensions The sample is gripped and the pulling force can only be achieved through frictio n at the clamp interface. Assu ming a friction coefficient of 0.1 and a maximum clamp stress equal to the yi eld stress, a free-body di agram reveals that the clamp face area must be five times the tensile cro ss-section. Thus a dogbone shape is required to prevent slip. The final dimensions provide a factor of safety of two to ensure that the clamp does not slip during the test. The corners of the dogbone are eased such that the shape provides a well-defined tensile section in the center of the sample where strain can occur without significant stress concentration. The quantities of interest during mechanical te sting are stresses and strains in the tensile section. The calculations of these quantities are as follows, F hw Eq. 4-5 0 00ll l ll Eq. 4-6 where is the stress, F is the axial force, h is the sample height, w is the sample width, is the engineering strain, l is the length du ring loading and l0 is the original sample length. The uncertainties in the measurements of and depend on the uncertainties in the measurements of F, h, w, l and l0 through the law of propagation of uncertainty. The quantities h and w are measured using a digital micrometer with a resolu tion of 1m. Thus, the Type B uncertainty in the measurement is 5 m. However, the geometri c irregularities in the sa mples were found to be on the order of 10 m. The uncer tainties in these measurements are therefore conservatively taken as 20m. The uncertainty in the force m easurement is calculated by comparing the force measurements of various standard masses against those from scales with known uncertainties. The uncertainty is conservatively estimated to be ~1 N. The displacement (l) uncertainty was
83 calculated using the same method of comparison w ith measurements of known uncertainty. This value is estimated to be approximately 20m. The length l0 was prescribed by the machining gcode but was difficult to measure accurately due to the radius at the corner. The accuracy with which the CNC milling machine can hold a dimension over this length has been found to be approximately 40 m. Therefore, the uncertainty conservatively ascribed to this dimension is 200m. Using the law of propagation of uncertainty the squares of the uncertainties in stresses and strains are, 222 2222 221 ()()()()cFF UuFuhuw hwhwhw Eq. 4-7 22 222 0 2 001 ()()()cl Uulul ll Eq. 4-8 Each term on the right of these equations can be divided by the combined standard uncertainty squared to give the sensitivity of the measura nd uncertainty to the measurement uncertainty in question. For example, the second term on the right of Eq. 4-8 divided by the left hand side of Eq. 4-8 is the sensitivity of strain to the uncertainty in l0. Despite the relatively large uncertainty in the value of l0 over l it accounts for less than 20% of the un certainty in strain calculations. Tribological Properties A laboratory designed linear reci procating pin-on-flat tribomet er, shown schematically in Figure 4-4, is used to test the wear and friction of the samples. This testing apparatus and the uncertainties associated with the experimental measurements are described in detail in Schmitz et al. [3, 4]. Although open to the air, the entire a pparatus is located inside a class 10,000 softwalled cleanroom (for reduced particulate abrasion) with conditioned laboratory air at 25C and from 25-50% relative humidity. In all cases, the normal load is Fn = 250 N, reciprocation length
84 is S = 25.4 mm (50.8 mm/cycle), sl iding speed is V = 50.8 mm/s, a pparent contact area is A = 40 mm2 and the nominal contact pressure is P0 = 6.3 MPa. In general, th e length of the test depends on the wear resistance of the material. Figure 4-4. Tribometer used for friction and wear testing. A pin of the bulk test material is pressed flat against a linearly reciprocati ng counterface. A 6-channel load cell reacts all of the loads incurred by the sample. The conditions are constant throughout the test i.e. as wear occurs. Tribological systems depend strongly on a number of factors. One very important factor in the study of soft solid lubricants is the tr ibological counter-surface or counterface; surface metrology is a scientific field that focuses on the accurate measurement, characterization and tribological exploration of such surfaces. A wide variety of stat istical parameters are used to describe these surfaces. A common feature to most tribological counterfaces is a negative skew. Negative skew is a statistical term used to describe distributions with long tails on the negative side of the mean line. Negatively skewed surfaces are those that have valleys that are deeper
85 than the heights of the surface protuberances, or asperities. This is important in tribology because the high asperities abrade the solid lubri cant while the deep valleys have little effect aside from potentially providing lu bricant reservoirs and anchoring beneficial transfer films. Figure 4-5. Scanning white light interferometry measurement of a representative lapped stainless steel (304) counterface. The su rface has a negative skew and the average and standard deviation of root mean squa red roughness (Rq) were found to be 161 nm and 35 nm, respectively. AISI 304 stainless steel is used as the counterface material; it is corrosion resistant and relatively soft (measured hardness of 87 kg/mm2 Rockwell B) which allows for easy detection of abrasion from the filler. New counterfaces are used for each experiment, so efforts were made to ensure that each nominally identical counterface sample was created under the same conditions. Five random samples were measured over a 230 x 300 m area using a scanning white light
86 interferometer. A representative measurement is shown in Figure 4-5. The surface map reveals a relatively flat surface cut by a number of deep scratches. The negative skewness is evident from the height histogram next to the map. Surface profiles are also shown in the x and y directions with the vertical axes magnified 65X. From the five random measurements, the average and standard deviation of the root mean squared roughness (Rq) were found to be 161 nm and 35 nm, respectively. Prior to testing, the counterface and composite pin are cleaned with methanol, and dried with a laboratory wipe. The composite samples are mounted directly to a 6-channel load cell which reacts all of the forces and moments on the pin sample. The counterface is mounted to a linear reciprocating stage beneat h the pin sample. A pneumatic cylinder is used to apply a normal force, which is continuously reacted a nd measured by the load cell, and computer controlled using an electro-pneumatic valve. A linear thruster isolates the pneumatic cylinder from frictional loads. Four 1 inch diameter r ods located within an aluminum housing are guided by linear bearings and provide high stiffness in the transverse dire ction. An LVDT mounted to the thruster monitors pin displacement. A stepper motor is controlled within custom data acquisition software. The stepper motor rotates the ball screw that drives the linear table. Another LVDT continuously measures the table position. Instantaneous data is collected for normal load, friction force, table position and pin position at 1000 Hz over 1 cycle at specified intervals. Data for one cycle is extracted using the positional LVDT. The instantaneous data are also averaged over two cycles and saved at a specified interval that depends on the length of the test.
87 Experimental Uncertainty for Friction Coefficients The primary function of this tribometer is to obtain values of friction coefficients and wear rates for different tribological systems of interest Neither friction coefficient nor wear rate can be directly measured, so they must be calculated based on other measurable quantities. The uncertainties associated with the calculation of friction coefficient and wear rate based on this tribometer were analyzed by Schmitz et al.. The friction coefficient is defined as, f nF F Eq. 4-9 where Ff and Fn are the friction and normal forces, respec tively. Often, the measurement axes, X and Y, are used to calculate an approximate friction coefficient, 'X YF F Eq. 4-10 where the measurement axes are not necessarily aligned with the normal and frictional axes. These misalignments inevitably arise from kinema tic chains of imperfect machining, assembly and compliance. These misalignments can cause substantial errors in the calculation of friction coefficient if unaddressed. Figure 4-6 illustrates these effects. The measurement axes, X and Y are assumed to be orthogonal and biased from the normal (N) and frictional (F) axes by an angle,
88 Figure 4-6. Model representation of misaligned measurement axes. The measurement axes are assumed orthogonal and are rotated about the normal and frictional axes by an angle, The measured force responses to th e tribological interactions are, cos()sin()cos()sin()cos()sin()xfn nnnFfffff Eq. 4-11 cos()sin()cos()sin()cos()sin()Ynfn nnFfffff Eq. 4-12 resulting in, cos()sin()cos()sin() cos()sin()cos()sin()n X Ynf F Ff Eq. 4-13 Defining an error fraction E as, 22 2'cos()sin()sin()(1)(1) 1 cos()sin()(cos()sin()) E Eq. 4-14 The misalignment angle, can be found by monitoring the force response to a stationary loaded pin sample. Schmitz et al.  finds to be approximately 2 using this method on the tribometer used in this study. For a friction coefficient of = 0.1, the correspondi ng measurement error is 35%. Thus, the measured friction coefficient is extremely sensitive to these misalignments. One could conceivably apply a coordinate transformation to force data with the measured value of
89 to calculate the true value of but the angular measurement would still vary slightly from test to test, making this type of analysis very cumbersome. Instead, if a reversal technique is used, the error in the approximate friction coefficien t calculations is largely eliminated. The aforementioned analysis will apply to the forward direction, so, 'cos()sin() cos()sin()f Eq. 4-15 Upon reversal, Fxr=fncos( )+fnsin( )) and Fyr=fncos( )+ sin( ), and 'cos()sin() cos()sin()r Eq. 4-16 By subtracting r from f and dividing by two (average of absolute values), we obtain a new value of the approximate friction coefficient 2222cos()sin()cos()sin() cos()sin()cos()sin() 2cos()cos() Eq. 4-16 With a friction coefficient of 0.1 and a maximu m angular misalignment of 4, this reversal technique introduces a bias in the friction coefficient of 0.0005; i.e. the measured friction coefficient is 0.0005 greater than th e true friction coeffici ent. Without reversals, the bias would be 0.0711, or about 350X the bias with reversals. The angular misalignments are not measured for each experiment and the bias in friction coe fficient measurements are therefore treated as single ended uncertainties which are added to the experimental uncertainty in the measured friction coefficient The measured friction coefficient has uncerta inties associated with the measurements of the forces Fx and Fy. The squared combined standard unc ertainty in the measured friction coefficient is calculated using Eq. 4-17.
90 2 2 222 21 (')()()yx ccxcy yF UuFuF FF Eq. 4-17 The uncertainties in the normal force and friction force directions are approximately 2N and 1N, respectively, and are dominated by time dependent thermal drift. With a normal load of 250N, the uncertainty in the measured friction coefficient is not particularly sensitive to friction coefficient, and ranges from 0.0040 to 0.0044 with values of measured friction coefficient ranging from = 0.05 to = 0.25. The uncertainty in the force dominates the angular uncertainty and it is thus appropria te and conservative to prescribe error bars with a magnitude of 0.005 to friction coefficient data. Experimental Uncertainty of Wear Rate The volume of material lost dur ing the wear process is genera lly proportional to the normal load and the sliding distance by a wear rate, k. The wear rate of a material can be used to determine component life and is defined as, nV k FD Eq. 4-18 Where V is the volume lost, Fn is the normal load and D is the sliding distance. Often, wear volume is calculated by ma king displacement measurements of a given cross section or by measuring the mass before and after a test. In many polymer systems, specifically, for wear resistant polymers, creep and thermal expansion can become significant po rtions of the total volume calculation. Mass loss measurements become difficult in situations where environmental uptake is expected. Since PTFE is known to have a high creep rate and coefficient of thermal expansion, as well as low water uptake and outga ssing, mass measurements are used to quantify wear in these studies. The volume lost is then, m V where 123 im LLL. The change in
91 mass is m, is the density, mi is the initial mass, and L1, L2 and L3 are the lengths of the rectangular solid sample. The sliding distance is2 DSN where S is the reciprocation length (for 1/2 cycle) and N is the number of reciprocation cycles. The wear rate can be expressed in terms of the measured quantities as, 1232inmLLL k mFSN Eq. 4-19 The law of propagation of uncertainty can be applied to Eq. 3-11 to find the combined standard uncertainty or the e xpected dispersion of values obt ained for the wear rate. The sensitivity of each measurement is calculated by ta king the partial derivative of the measurand, k, with respect to the measurement. Each sensitiv ity term is then squared and multiplied by the square of the uncertainty in that measurement. These contributions are added to find the square of the combined standard uncer tainty of the wear rate, as 3 2 2 3 2 2 2 2 1 2 2 1 2 2 2Lu L K Lu L K Lu L K mu m K Kuc Su S K mu m K Fu F Ki i n N 2 2 2 2 2 2 Eq. 4-20 Evaluating the partial derivatives gives, 2 2 2 31 1 2 2 32 2 2 321 22 2 2 Lu SNmF LmL Lu SNmF LmL mu SNmF LLL Kuin in in c n in inFu SNmF LLmL Lu SNmF LmL2 2 2 321 3 2 2 212 2 Su NSmF LLmL mu SNmF LLmLin i in 2 2 2 321 2 2 2 3212 2 Eq. 4-21 where nominal values of the measurements are used in numerical calculation. Determining the uncertainty of each measurement requires Type A or Type B evaluation consisting of either statistical methods or engineeri ng judgment, respectively. A deta iled analysis can be found in
92 Schmitz et al.. Briefly, the uncertainties of the sample length and mass measurements are taken conservatively as five times the manufactur ers specified repeatab ility; 0.005 mm and 0.05 mg respectively. The uncertainty u(m) is calculated using the law of propagation of uncertainty ifmmm Eq. 4-22 Taking the partial derivative of this fu nction with respect to the measurements, mi and mf gives, 2 2 22 2222211ifif ifmm umumumumum mm Eq. 4-23 Since the initial and final masses are approxima tely the same, it is reasonable to assume that the uncertainties in these measurem ents are equal, and Eq. 3-15 becomes, 222iumum Eq. 4-24 and, 2iumum Eq. 4-25 The uncertainty in the normal load due to time dependent fluctuations was found to overwhelm angular misalignment errors and short term scatter, and the uncertainty is taken conservatively as u(Fn)=2N, twice the observed thermal drift of the load cell and electronics. The uncertainty in the number of cycles is zero and the uncertainty in the reciprocation length is u(S )= 0.2mm which results from a combination of an assumed angular misalignment of 2 and one standard deviation of meas urements for commanded motions. Many materials exhibit initial w ear transients, which preclude the use of single point mass measurements. Often, a least squares regression of the steady data is used to obtain a more representative value for th e wear rate of a material. A modifi ed numerical approach to the above uncertainty analysis is required for such regre ssions . Interrupted measurements are made
93 periodically during each test a nd are used to distinguish the steady region of wear from the transient region. A Monte Carlo si mulation uses the uncertainties (UcV) and Uc(Fn*D)) and nominal values of measurement to calculate the average regression slope and standard deviation of the slopes from 1,000 simulated data sets. Th e regression represents the wear rate and the standard deviation of the slope re presents the uncertainty in wear rate. The uncertainty intervals on wear rate data represent the experime ntal uncertainty in the measurement.
94 CHAPTER 5 DESIGN OF EXPERIMENTS Effects of Mechanical Blending Various techniques, including centrifugal mi xing, ultrasonication, blade mixing and jetmilling, have been used throughout the nanocomposite literature to disperse nanoparticles. Jetmilling has proven to be a particularly successful t echnique, especially at low loadings, but it is unclear if this success is a product of mechanical ly induced morphological changes or simply the result of improved dispersion. One hypothesis is that the high energy of the jet-milling technique results in a fibrillated structur e that promotes wear resistance when stabilized by nanoparticles during sintering. Fibr ils of PTFE are oriented and extr emely strong compared to the bulk material and are often the source of improved st rength, stiffness, toughne ss and creep resistance [13, 103, 105, 109, 110]. In this experiment, the morphological and stru ctural changes induced by the mechanical history of the sample are studie d to test the hypothesis th at the success of the jet-mill is due to fibrillation induced during processing. Neat PTFE particles are treated with one of five mechanical blending techniques, which include, elastic sieve classification, centrifuga l mixing, ultrasonication, blade mixing and jetmilling. The experimental treatment and the treatment operating conditions are shown in Table 5-1. The first goal of this study is to charac terize the thermal characteristics of the virgin powder. The second goal is to establish the effects of mechanical mixing on the structure and morphology of the PTFE. This study is critical in determining whether the fibrils observed for wear resistant materials are formed during mech anical processing or wear events. Scanning electron microscopy (SEM) is used to evaluate the topographical structural effects of the mechanical treatments. Particle reduction, deformation and fibri llation are of interest. The melt behavior of PTFE is very sensit ive to the nature of the crysta lline structure and morphology, and
95 as a result, it is also reputably sensitive to m echanical history. Differe ntial Scanning Calorimetry (DSC) is used to study the ther mal response of the powders here to provide insight into the resulting crystalline structure af ter each processing technique. Melt peaks, crystallinity and matrix mobility are of interest. These comp limentary techniques can provide structural information from the nanoscale to the microscale, and provide a much more complete understanding of the effects of th ese dispersion techniques on the ma trix itself than is currently available. Table 5-1. Blending treatments of neat PTFE to simulate the effects of nanoparticle dispersion on the polymer. The morphological and ther mal characteristics of the powders are studied following treatment. Sample Powder treatment Conditions 1 None Control; virgin as-received powder 2 None Control; virgin as-received powder 3 None Control; virgin as-received powder 4 None Control; virgin as-received powder 5 None Control; virgin as-received powder 6 Shear sieve 1 pass through 10 mesh sieve 7 Ultrasonication Ultrason ic agitation; 30 minutes 8 Hauschild rotary mixer 1500 RPM for 30 seconds 9 Blade mixer 30,000 RPM for 40 seconds 10 Jet-mill 150 psi; 3 passes; ~5g/min Processing Temperature and Crystalline Morphology The melt behavior of PTFE is very unique, a nd the characteristics of its thermal behavior are widely known to depend strong ly on the thermal and mechanical history of the sample [94, 95, 111, 112]. This is due to the strong sensitivity of its crystalline stru cture to both, and the strong relationship between crystall ine structure and thermal response. In general, the more ordered the crystalline structur e, the higher the melt temperatur e [95, 107, 108]. Virgin PTFE from the reactor has large regions of high order, and has a report ed melt temperature of 353C. Following melting and recrystalliza tion, PTFE is relatively di sordered and has the more
96 commonly cited melt temperature of 327C. Figur e 5-1 shows differential scanning calorimetry (DSC) thermograms for virgin PT FE powder. During the first heat, the melt temperature of the powder is 343C with a large melt peak indicativ e of high crystallinity (~75% using a heat of fusion of 82 J/g for perfectly crystalline PTFE ). Following melt, the sample was cooled and the material crystallized. During the sec ond heat, melting occurs at a temperature of 327C with less consumed energy indicating lower crystallinity (~25%). In a preliminary thermal characterization study of a 1 wt% PTFE na nocomposite, a large melt peak of 340C was observed even though the sample was processed usi ng standard processing with a spike to 380C and a hold at 362C. This result suggested that the nanoparticles may inhibit PTFE mobility at these temperatures, and thus stabilize the ordered virgin morphology of the PTFE. Figure 5-1. Differential Scanni ng calorimetry thermogram of as-received virgin PTFE during a heat/cool/heat cycle. The first melt is i ndicative of virgin PT FE, while the second melt occurs after the first melt and recrystallization. Following melt and recrystallization, both the melt temperature and crystallinity are lower.
97 A working hypothesis for wear resistance in PTFE nanocomposites is that the nanoparticles stabilize th e virgin folded ribbon morphology of the PTFE. This structure may facilitate fibrillation during deformation which provides an efficient deformation mechanism and a strengthened running surface in the sliding di rection [103-105, 114-117]. Based on this wear reduction mechanism, virgin PTFE is hypothesized to have improved wear resistance over melt processed PTFE. This study invest igates the thermal, tribological and mechanical characteristics of PTFE as a function of processing temperat ure to test the hypothe sis that the virgin morphology facilitates fibrillation and promotes wear resistance in the absence of nanoparticles. Because the PTFE particles in th e sample require melting for effective sintering and mechanical integrity, the morphology of the PTFE cannot be separated from the cohesion and strength of the sample. Various temperatures in the vicinity of the melt temperature were employed in order to achieve the best possible bala nce of inter-particle cohesion and virgin morphology. At intermediate temperatures, low melt structures may melt to form cohesive particle boundaries while high temperature structur es remain un-melted to promot e fibrillation during wear. Virgin PTFE was compression molded at five experimental temperatures. The control temperature is 362C; at this test condition, th e sample should fully melt and recrystallize into a more disordered, less crystalline structure with a melt peak near 327C. This fully sintered PTFE is known to have a high wear rate on the order of k=10-3 mm3/Nm under typical conditions [8, 9]. Additional temperatur es of 300C, 327C, 345C and 353C were used to study the thermal, tribological and mechanical characteristic s of virgin PTFE as a function of the sintering temperature. Virgin powders were used to fill a mold and the sample was compacted at 80 MPa for 15 minutes. A temperature ramp of 2C/minute was used to heat the sample at a light hold
98 pressure of 4 MPa. The sample is held at the process temperature for 3 hours and is then cooled at the same rate. The experimental matrix for this study is shown in Table 5-2. Tr ibological experiments were conducted on a linear recipr ocating tribometer at 50.8 mm/ s and 6.3 MPa of pressure. Mechanical tensile tests were conducted with a strain ramp of 1%/minute, and thermal measurements were carried out using differen tial scanning calorimetry and a ramp rate of 10C/minute. Table 5-2. Experimental matrix investiga ting the effects of sint er temperature on the tribological, thermal and mechanical properties of unfilled virgin PTFE. Sample Sintering temperature (C) Experiments 11 300 Thermal and tribological 12 327 Thermal and tribological 13 345 Thermal and tribological 14 353 Thermal and tribological 15 362 Thermal and tribological 16 300 Thermal and mechanical 17 327 Thermal and mechanical 18 345 Thermal and mechanical 19 353 Thermal and mechanical 20 362 Thermal and mechanical Mechanical Processing on Nanoparticle Dispersion The processing of PTFE and PTFE composites is inherently difficult. The same unique thermal and chemical properties that make PT FE an attractive extreme environment solid lubricant also make PTFE nanocom posites inherently difficult to process. The most effective polymer matrix dispersion tec hniques, namely liquid dispersion/in-situ polymerization and high shear-rate melt-mixing [118-120], are precluded for PTFE. As a resu lt, filler materials must be dispersed into dry PTFE powders with any of a number of dry powder blending techniques. This method cannot create a uniform disp ersion. In a best case scenar io, fillers and matrix particle agglomerations are disbanded, perfectly disperse d and equally distributed to decorate each PTFE
99 particle. The limit of uniformity, in this case, is the largest particle bei ng dispersed. This is important in nanocomposite synthe sis since the size-scale of the ch aracteristic repeating unit in the dispersion is orders of magnitude larger than the filler particle. In addition to these inherent dispersion and processing lim itations of PTFE, nanoparticles are notoriously difficult to disperse with dry powder blending tec hniques. The potential benefits of the nanoparticle fillers result from large interfacial areas an d number densities for a given volume of material. This also makes the surface forces large compared to the body forces which results in agglomeration. Dr y powder blending techniques are not particularly effective in disbanding nanoparticle agglomerates because they primarily rely on accelerations and inertias for particle dispersion. In a worst case scen ario, agglomerations are not disbanded, and the composite mimics the dispersion char acteristics of a microcomposite. As a result of the processing difficulties of PTFE nanocomposites, dispersion and the role of dispersion on the tribological properties of the nanocomposite ar e among the largest of uncertainties in PTFE nanocomposites tribol ogy. The polymer nanocomposites tribology literature is full of examples where poor performance is likely the result of poor dispersion, but due to the paucity of dispersion ch aracterization, this relationship has not yet been established [8, 10, 11, 16, 17, 19, 31-34, 46, 121]. In parallel studies by Burris et al. and McElwain  similar performance was obtained at 5 wt% loading of 80 nm phase alumina mixed using a jetmill and a hauschild mixer, respectively. However, while Burris et al.retained high wear resistance at 1 wt% loading, McElwain found a 0.8wt% nanocomposites to be 5,000X less wear resistant. The hauschild mixer is suspected to have poor dispersion ability at low loadings. Dispersion is also the factor whic h contributes most to the difficulty in interpreting the state of the field. This is mostly due to a lack of dispersion characterization and a lack of standard
100 processing techniques. In additi on, important factors such as bl ade speed and time of dispersion are seldom mentioned. In this study, nanoparticle disp ersions are examined. The experimental matrix is shown in Table 5-3 and includes three nanoparticle loading conditions (1, 2 and 5 wt%) and two nanoparticle dispersion techniques (none and jetmilling). In each case the neat PTFE powders are jet-milled prior to nanoparticle inclusion to se parate dispersion effects from blending effects. Following jet-milling, the appropriate weights of the constituents are combined and mixed for 1 minute by hand to obtain gross homogeneity. The control sample is untreated and the test samples are further dispersed using a jet-mill. Following blending, the samples are studied using scanning electron microscopy to interrogate the effects of the dispersion treatments on the nanoparticle dispersions. Table 5-3. Experimental matrix examining the effect of loading and di spersion technique on the powder dispersion of nanopart icles and PTFE. Nanoparticles are 80 nm alpha phase alumina. sample wt% Dispersion treatment test 21 1 Hand-mixed Dispersion characterization 22 1 Jet-milled Dispersion characterization 23 2 Hand-mixed Dispersion characterization 24 2 Jet-milled Dispersion characterization 25 5 Hand-mixed Dispersion characterization 26 5 Jet-milled Dispersion characterization Nanoparticles on the Melt Behavior of PTFE A standing hypothesis for the wear resist ance in PTFE nanocomposites is that the nanoparticles stabilize the virgin PTFE structure to temperat ures that would otherwise destroy this structure. This structure is thought to promote wear resistance by facilitating an energyabsorbing fibril-forming deformation-mechanism during sliding. This hypothesis is largely based on a thermal and tribological study of one 1 wt% wear resi stant alumina-PTFE
101 nanocomposite. In this study, a very wear resi stant nanocomposite had a DSC thermogram with a large high temperature melt peak indicative of th e virgin PTFE structure. It was hypothesized that the nanoparticle inclusion resulted in rete ntion of the virgin structure which promoted fibrillation and wear resistance. To test this hypothesis, the na nocomposite was heat treated to 400C to destroy the virgin stru cture. Following heat treatm ent, the thermogram reflected melted and recrystallized PTFE a nd wear resistance was lost. Table 5-4. Experimental matrix for experime nts studying the influence of nanoparticles on the thermal response of the PTFE powder. All samples are jet-milled identically. sample wt% Dispersion treatment test 21 1 Hand-mixed Thermal analysis 22 1 Jet-milled Thermal analysis 23 2 Hand-mixed Thermal analysis 24 2 Jet-milled Thermal analysis 25 5 Hand-mixed Thermal analysis 26 5 Jet-milled Thermal analysis It is clear that the morphological characteristi cs of the PTFE itself has a significant impact on the tribological properties of th e material since nanoparticle mate rial, size, shape, loading and dispersion were nominally constant before and after heat treatment. It is unclear what role the nanoparticles play in the stabi lization of the virgin morphology. It is possible that the nanoparticles retard melting to higher temperature, in which case the polymer would not have melted during compression molding. This is unlikely since the melt temperature of the nanocomposite sample was found to be no higher than that of virgin PTFE using differential scanning calorimetry. Another possibility is th at the polymer melts dur ing processing, but the high nanoparticle density and surface area limit polymer mobility durin g melt and nucleate recrystallization upon co oling. In this study, differential scanning calorimetry is used to study the thermal behaviors of PTFE and PTFE nanocom posite powder ensembles. The experimental matrix for this test schedule is shown in Table 5-4.
102 Filler Dispersion on Nanocomposite Properties An essential part of understa nding the tribological and de formation mechanisms of a material is a thorough understandi ng of the material under investiga tion, but this combination of material science and tribology is almost complete ly absent from the literature . In this study, the thermal, tribological and mechanical properties of compression molded samples are investigated. The powder ensembles are compressi on molded into test pucks using the standard compression molding procedure outlined in the sa mple preparation section. The experimental matrix describing this study on compression molded nanocomposites is shown in Table 5-5. Table 5-5. Experimental matrix investigating the effects of na noparticle loading on the thermal, tribological and mechanical properties of the compression molded nanocomposite. Nanoparticles are 80 nm alpha phase alumina. PTFE powders are jet-milled prior to nanoparticle inclusion to e liminate PTFE agglomeration and size as variables. sample wt% Dispersion treatment Test 27 1 Hand-mixed Thermal, tribological 28 1 Jet-milling Thermal, tribological 29 2 Hand-mixed Thermal, tribological 30 2 Jet-milling Thermal, tribological 31 5 Hand-mixed Thermal, tribological 32 5 Jet-milling Thermal, tribological 33 1 Hand-mixed Thermal, mechanical 34 1 Jet-milling Thermal, mechanical 35 2 Hand-mixed Thermal, mechanical 36 2 Jet-milling Thermal, mechanical 37 5 Hand-mixed Thermal, mechanical 38 5 Jet-milling Thermal, mechanical Tribological properties are characterized using the linear reciprocating tribometer discussed in Chapter 4 using a sliding speed of 50 mm/s, a normal pressure of 6.3 MPa and a track length of 25 mm. The length of the test will be a function of the wear resistance of the sample and will persist until the sample loses ~50 mm3 of material or accumulates ~750,000 sliding cycles. An additional compression molded puck is processed for mechanical characterization. Tensile tests are conducted on each sample at 1%/min until the sample can no
103 longer support load. Thermal samples are extr acted from each compression molded puck prior to machining. Filler Material on Nanocomposite Properties A study by Burris and Sawyer found that the phase of the alumina nanoparticles was much more effective than the phase . A more recent study has shown that this is also true when filler particle size is held constant. These results are unusual and suggest that the shape or chemistry dominate the dispersion or the effects of the particles on polymer structure, mobility, nucleation or recrystallization, and as a consequence, dominates the tribological properties of the composite. In this study, the influences of partic le phase are investigated in terms of dispersion, thermal effects on the matrix, tribological properties and mechanical properties. Following dispersion and thermal studies of the powders, the ensembles are compression molded for thermal, tribological and mech anical characterizati on of the processed nanocomposites. Thermal studies on compression molded samples utilize a heat-cool-heat from -10-400C to study the phase transitions, glass tr ansition and melt behavior of the pressuresintered and free-sintered material The experimental matrix is s hown in Table 5-6. Tribological and mechanical characterizations are conducted as described in Chapter 4. Table 5-6. Experimental matrix for experime nts studying the influence of nanoparticles on the thermal, mechanical and tribological re sponse of the PTFE nanocomposites. All samples are jet-milled and compression mold ed according to the procedures outlined in Chapter 4. Sample Loading (wt%) APS (nm) Phase 15 0 N/A N/A 39 12.5 80 40 12.5 44
104 CHAPTER 6 RESULTS Mechanical Blending Characterization of Particle Morphology Representative images from SEM analysis of the PTFE particles following varying mechanical blending treatments are shown in Figure 6-1. Box plots for the estimated mean particle sizes, standard devi ations and upper and lower quartiles are shown in Figure 6-2. Pseudo-quantitative estimates of the particle sizes from these studies are listed in Table 6-1. The manufacturer reported average particle size (APS) of the Teflon 7C compression molding resin is 35 m, but SEM analysis of the as-rece ived powders (Figure 61a) suggests an average particle size of approximately 20 m. The vast majority of the particles are round with a characteristic size near 20 m, but there are also particles th at are highly elongated with the long dimension being on the 50-100 m size scale and the short dimension being on the 1-5 m size scale. The distribution is positively skewed and contains only a small fraction of particles with a characteristic dimension less than ~15 m. Similar analyses of the sieved, ultrasonicated rotary mixed and blade mixed powders in Figure 6-1(b-e) reveal pa rticle distributions and morphologies that are nomi nally identical to the virgin powders to within the detection limits of the SEM surveying method employed. Each powder predominantly contains round 20 m particles intermixed with smaller round particles and larger elongated particles. PTFE Teflon 7C is reputably di fficult to grind due to the small size and high toughness of the resi n; this is evidenced by the l ack of influence the various blending techniques had on particle size and morphology. The blender in particular was
105 expected to effect both given th e severity of the operation with steel grinding blades rotating through the powders at 30,000 RPM, but even th is technique had a negligible effect. Figure 6-1. SEM images of virgin PTFE followi ng varying mechanical treatments typically used in particle dispersion: a) untreated, b) sieved, c) ultrasonicated, d) rotary mixed, e) blade mixed and f) jet-milled.
106 This lack of influence is contrasted by gross effects of the jet-milling operation to both the particle size distribution and the particle morphology. Jet-milled powders have a substantially smaller mean near 5m and do not contain any of the highly elongated structures possessed by the other powders. Despite resisting any size reductions from th e other grinding operations in the study, nearly all of the PT FE particles in the as-received powders were ground to the original diameter on average. This equates to a 4X increase in particle surface area and a 64X increase in the number of particles, both of which may have an impact on the melt and crystallization characteristics of the resin as well as any compartmentalization effects in a tribological nanocomposite [8, 16, 51]. One of the most important features of the jet-milled powders, however, is the complete lack of a fibril lated morphology. It can be concluded that the PTFE fibrils observed in the running surfaces of very wear resistant PTFE nanocomposites are products of deformation processes during sliding a nd are not created during powder blending. Figure 6-2. Estimated particle size results of virgin PTFE following varying mechanical treatments: a) untreated, b) sieved, c) ultras onicated, d) rotary mixed, e) blade mixed and f) jet-milled. Averages, standard deviations and quartiles are estimated from SEM surveying.
107 Thermal Characterization Measurements of the thermal characteristics of the powders were made to interrogate the thermal implications of the mechanical proce ssing and the resulting PTFE morphologies. The resulting thermograms from DSC measurement of PTFE powders following the mechanical treatments under investigation are shown in Figure 6-3. Five independent control samples with no mechanical treatment were created and tested to evaluate material variability, sample preparation variability and measurement variabi lity. These five independent experiments are shown in blue in Figure 6-3. The measuremen ts fall very close to one another and indicate excellent repeatability in th e material, sample preparati on and thermal measurement. Measurements of melt temperatures and enthalpi es from these thermograms are tabulated in Table 6-1 and the result s are shown graphically in Figure 6-4. A mean first melt peak for untreated, as-received PTFE powders of 342.76C was calculated. The standard deviation in the m easurements was 0.08C. Using the t statistic (because of the small sample size) with 4 degrees of freedom, the limits of the true population mean can be calculated for a desired confidence percentage. For 99% confidence in this case, the t statistic is 3.747. From the central limit th eorem, the standard deviation for the distribution of mean values from random sampling can be calculated as 0.08 0.036 5ys nC Eq. 6-1 The lower limit on the true mean for 99% confidence is, 3.7470.08 342.76 342.63 5ts y n C Eq. 6-2 i.e. there is less than 1% probability that the 5 random samples used in these experiments came from a population with a mean of 342.63C or less.
108 Figure 6-3. Differential scanning calorimetry of PTFE powders with varying mechanical history. Jet-milling provided the only significant de viation from the thermal behavior of virgin PTFE. Once the lower limit for the true untreated population mean has been quantified, the probability that the other samples did or did not belong to the same population can be computed. The values of first melt for the sieved and rota ry mixed samples lie in between the upper and lower limits for the true population mean for virgin powders, and it cann ot be concluded that these powders have different first melts. The sonicated and blade mixed samples were 0.13 and 1 below the lower limit for the mean. Since ~16% of the data lies greater than 1 below the mean, it cannot be concluded that either of these powders have a different first melt temperature than the untreated powder. The first melt temp erature of the jet-milled powder however, is 22below the lower limit for the mean first melt temperature of virgin PTFE. Less than 3x10-5 % of the data lies above 5, so there is essentially zero prob ability that the jet-milled PTFE
109 sample came from the untreated distribution and it can be concluded that its first melt temperature of the PTFE has been reduced due to jet-milling with near ly 100% confidence. From the data collected, none of the other opera tions had an effect on size or ordering of the polymer, but the lower melt temperature of th e jet-milled sample suggests that the crystals have been reduced in size and/or reduced to a less ordered morphology. This implication is consistent with SEM observation of the powders which shows significant reductions in the sizes of nominally single crysta l particles in the resin. The lowe ring of the melt temperature following jet-milling suggests that the crystals have become smaller and less ordered. The same analysis can be conducted for the fi rst and second enthalpies of formation and the second melt. For the first heat of formation, there is significantly more scatter for nominally identical samples than there was for the first melt te mperature. This is partially attributable to the 1% uncertainty in energy and the uncertainty of mass measurements, but it also likely reflects a true variation in crystallinity within the powder. The lower limit for the mean first enthalpy of formation of the untreated resin is 70.4 J/g with a standard deviati on of 1.2 J/g. Each of the treated, non-jet-milled samples lie within the lim its of the true population mean for untreated powders. The jet-milled powder had a heat of formation that was 1below this lower limit, but with only 84% confidence, there is insufficient ev idence to conclude that the two samples have different population means. A lack of change in the heat of formation is not completely surprising; the particle bulk does not appear to be appreciably defo rmed with the vast majority of the mechanical destruction occurring on and near the particles surfaces which constitute only a very small fraction of the total mass contributing to the signal.
110 Figure 6-4. Quantified results of differential scanning calorimetry of PTFE powders with varying mechanical history: a) first melt peak temperature, b) firs t heat of fusion, c) second melt peak temperature and d) second he at of fusion. Treatments are labeled as follows: 1) untreated, 2) sieved, 3) ultras onicated, 4) rotary mixed, 5) blade mixed and 6) jet-milled. For the second melt, a lowe r limit of 327.57C for the virgin population mean is calculated. In this case all of the samples have melt temper atures far enough below the lower mean limit to conclude with better than 97% co nfidence that melt temperatures are below that of the untreated control sample. Sieved, rota ry blended and blade mixed samples are ~2 below the
111 untreated lower limit, while sonicated and jet-milled samples are ~5 below. Despite having statistically different second melt temperatures than the untreated samples, these samples had, at most, a 0.2C melt temperature difference as opposed to the 2C difference for the jet-milled sample on first melt. It is unclear what, if any im plications this slight temperature difference has, especially when considering the 0.1C uncertainty in the measurement. Table 6-1. Blending treatments of neat PTFE to simulate the effects of nanoparticle dispersion on the polymer. Sizes are reported usi ng lower quartile, mean and upper quartile, respectively. Five independent samples of the unprocessed PTFE were tested for control statistics on thermal behavior. Uncertainty on temperature data is 0.1C. Morphology Thermal properties Size (m) Tm1(C) Hm1(J/g) U(Hm1) Tm2(C) Hm2(J/g) U(Hm2) 1 untreated 15,20,30 342.88 71.6 1.0 327.63 16.6 0.2 2 untreated 15,20,30 342.80 70.7 1.0 327.67 15.8 0.2 3 untreated 15,20,30 342.74 72.8 1.0 327.62 16.1 0.2 4 untreated 15,20,30 342.66 73.3 1.0 327.61 16.5 0.2 5 untreated 15,20,30 342.72 73.3 1.0 327.59 16.5 0.2 1-5 mean 1-5 20 3 342.76 0.08 72.4 1.2 1.0 327.62 0.03 16.3 0.3 0.2 6 sieved 15,20,30 342.64 73.1 1.0 327.51 16.4 0.2 7 sonicated 15,20,30 342.62 72.1 1.0 327.41 16.5 0.2 8 rotary 15,20,30 342.68 72.2 1.0 327.50 16.4 0.2 9 blade 15,20,30 342.55 72.6 1.0 327.49 16.5 0.2 10 jet-mill 1,3,7 340.89 69.1 1.0 327.43 19.1 0.3 An upper limit of 16.8 J/g is computed for th e true population mean of the second heat of fusion of untreated virgin PTFE resi n. The scatter in this data is significantly smaller than it was for the first heat of fusion. This is likely due to the carefully contro lled cooling rate following melt and its regulating effect on the crystallinity of the samples. Each of the non-jet-mill treated samples had heats of fusion that are between th e population mean limits for 99% confidence. The jet-milled sample has a heat of fusion that is nearly 8 above the upper limit of the untreated
112 PTFE for 99% confidence. It can be conclude d with essentially 100% confidence that the melt recrystallized jet-milled sample has a greater crystallinity than the untreated resin. The jet-milled powder had approximately 50 times the number of particles for a given mass. This may have increased the number of nucleation sites to promote increased crystallinity. Processing Temperature and Crystalline Morphology Thermal Characterization Previous studies have shown that successf ul melt-processed PTFE nanocomposites shared the thermal and X-Ray diffraction characteristics of virgin PTFE prior to melting. Naturally this observation led to a hypothesis th at the role of the nanoparticles was to stabilize the virgin morphology of PTFE during melt pro cessing. It remains unclear if this morphological signature alone accounts for the wear resistance of the low wt% nanocomposites, or if the nanoparticles lead to wear resistance through other mechanisms. Here, PTFE test specimens were prepared at varying hold temperature to investigate the role of PTFE mo rphology in the absence of the nanoparticles. Following compression molding, the tribologi cal samples were machined from the compression molded puck, with the wear surface being located at th e axial center of the cylinder. DSC samples were taken from a location of the puck that is within 2 mm of the wear surface. The resulting DSC thermograms are shown for each of the five samples in Figure 6-5. The evolution of the thermal behavior is clearly seen in Figure 6-5a. The 300C, 327C and 345C samples exhibit similar melt behavior to the virgin powder; thus, very little if any of the sample ever melted during compression molding. At higher temperatures, only the low ordered crystals in the lower temperature regime of the melt curve enter the gel or melt state during processing. These melted regions recrystallize and exhibit the lower melt temperature of the lower ordered recrys tallized PTFE. The effects of this partial melting are clearly visible for the
113 345C, 353C and 362C samples with the amount of recrystallized material at the ~327C melt temperature increasing with in creased hold temperature. It is interesting to note that the samples with hold temperatures that are signific antly higher than the me lt temperature of the material do not fully melt. This is clearly the case at the 362C hold temperature. The thermocouple measured temperature at a location very near the test area and despite being heated 20C-30C over the melt peak, a small fraction of the material did not melt. During compression molding, the sample is under pressure. This pressu re likely inhibits mobility of the polymer, and additional thermal energy is necessary to induce melting. Figure 6-5. Differential sca nning calorimetry of PTFE samples compression molded with varying sintering hold temp eratures; a) first melt following compression molding, b) second melt following recrystallization at a co oling rate of 10C/min. The first melt curves show a gradual transition from th e virgin morphology having a melt peak of ~343C and the melt crystallized morphology having a melt peak of ~327C. Following recrystallization from 400C at 10C/min, the thermal characteristics are nearly identical. Melt peaks and enthalpies are plotted versus hol d temperature in Figure 6-6. These results are tabulated in Table 6-2. The melt peak remains nominally unchanged as more and more melting occurs until the magnitude of the 327C me lt peak height exceeds that at 344C. The 362C sample was the only sample where the low temperature peak height exceeds the high
114 temperature peak height. Once melting begins, th ere appears to be an approximately linear decrease in crystallinity (enthalpy) with increased peak temperature. Figure 6-6. Results of differential scanning ca lorimetry of PTFE samples compression molded with varying sintering hold temperatures; a) first peak melt temperature, b) first heat of fusion, c) second peak melt temperature following recrystalliz ation at a cooling rate of 10C/min, d) second heat of fusion following recrystallizati on at a cooling rate of 10C/min. There is a gradual tran sition from the virgin morphology to the recrystallized morphology as more melting occurs. These DSC results show that the morphology of the PTFE can be controlled using hold temperature variations in the ab sence of nanoparticles. This st udy is not entirely fair since the nanocomposites are heated and held at 362C. Th e consequence of the low hold temperature is the potential for improper sintering and reduced continuity between the particles. However, the
115 surfaces of the particles have lower ordering than the internal bulk and are therefore thought to melt preferentially. Thus, the 345C and 353C samples should offer a strong opportunity to study samples with virgin bulk morphology and strong interparticle c onnectivity. Mechanical Properties Characterization The virgin morphology is thought to promote w ear resistance by facilitating fibrillation during deformation and therefor e increasing the energy required to remove material during sliding. The mechanical properties of these sa mples were studied to investigate the role of fibrillation and toughness on the wear properties of the sample Stress is plotted versus engineering strain in Figure 6-7 for the hold temperatures of 327C, 345C, 353C and 362C. The sample held at 300C was not resilient enough to endure machining and broke during fixturing. The 345C, 353C and 362C samples had similar moduli on th e order of 300 MPa, with the 362C sample being slightly less stiff. This may be due to the lower crystallinity of the 362C sample. While the 362C and 353C samples followed a continuous trend until exceeding the ultimate stress, the 345C sample appears to have experienced several local rupture events before failing and slowly losing load carrying cap acity from 2.5-5% strain. The 327C sample had a much lower initial elastic m odulus and stiffened slightly with strain before failing at 1.2% strain. The early failure and low stiffness is lik ely related to a lack of coherence from improper sintering and easy crack propagation It is interesting that the lower hold temperature samples had slower stress decay after failure. This could be an indication of low stress fibrillation events occurring to a greater extent for materi als with the virgin morphology. Ultimate stress and engineering strain are plotted versus hold temp erature in Figures 6-8a and 6-8b, respectively. There is a clear trend of increased ultimate stress, strain and toughne ss with increased hold temperature.
116 Figure 6-7. Results of mechanical testing of PTFE samples compression molded with varying sintering hold temperatures. The extension rate was 1mm/min which corresponds to a strain rate of approximately 7%/min. There is a significant increase in ultimate strength, strain and toughne ss as the sintering temper ature increases. The 300C sample broke during machining and could not be tested. Figure 6-8. Quantified results of mechanical te sting of PTFE samples compression molded with varying sintering hold temp eratures; a) ultimate stress plotted versus sintering temperature, and b) ultimate strain vers us sintering temperature. There is a significant increase in ultimate strength, strain and toughness as the sintering temperature increases.
117 These fracture surfaces were studied using SE M to gain insight in to the mechanical properties observed during testing. Figure 69 shows backscattered el ectron images of the fractured surfaces at 300X. The 372C surface is very smooth in comparison to the other surfaces. The protrusions from the other surfaces represent areas of concen trated strain at the cusp of failure. Brown and Dattelbaum illustrated three basic modes of failure for PTFE: 1) brittle fracture, 2) microvoid coalescence and 3) fibril formation, alignment and extension . The smooth surface of the 327C sample suggests a brittle fracture mode, but the stress-strain behavior resembled ductile behavior with a slow transition from maximum load carrying capacity to no load carrying cap acity. There are a number of protrusions scattered sparsely across the surface of this sample. This sample likely experienced brittle fracture at 1.2% elongation across the majority of the surface leaving a low area fraction of PTFE in tact for fibrillation and load transfer during the remainde r of the test. The higher temperature samples failed via microvoid nucleation and coalescence; the protruding mate rial reflects the material on the periphery of the microvoids. The surface topography of the 345C sample had a higher number of protrusions standing proud from th e surface, but a number of cracks and step elevation changes (one shown) were observed. Brittle fractures over poorly sintered sections likely account for the step decreases in load. The 353C and 362C samples appear similar but close inspection reveals a higher content of smooth areas and less prominent protrusions on the surface of the 352C sample. A larger concentration of weak areas resulted in a larger number of nucleated microvoids and a smaller area in betw een for load support and deformation. This resulted in lower ultimate stress and strain. Secondary electron images of these surfaces, sh own in Figure 6-10, were taken at 600X to highlight the fibrils and other more detailed as pects of each surface. In accordance with the
118 hypothesis that the virgin morphology facilitates fibrillation, the lower temperature samples were observed to have more and fine r fibrils pulled from the surfac e during fracture than higher temperature samples. Very fine fibrils were pulled from the surface of the 327C sample, though these features are difficult to resolve in the image. Bundles of fibrils are observed to have been drawn in large numbers from th e surface of the 345C sample. Far fewer but oriented fibril bundles are observed on the surface of the 353C sample, and even fewer are observed on the surface of the 362C sample. The SEM observati on suggests that finer and less frequent the characteristic deformation structure on the surf ace, the more concentrated the load, the more localized the stress and the lower the load carrying capacity of the sample. The virgin morphology was hypothesized to fa cilitate fibrillation and im prove tribological properties through damage compartmentalization and im proved toughness. While the low sintering temperatures were shown to facilita te fine-scale fibrillation, they also lead to lower area fractions for fibrillation, a general lack of coherence in the sample and diminished mechanical properties including toughness. The lack of coheren ce may contribute to low wear however by compartmentalizing damage within smaller volumes of material and therefore limiting rates of wear and promoting transfer f ilm initiation and growth.
119 Figure 6-9. Backscattered elec tron images of the fracture su rfaces of PTFE with varying sintering temperatures; a) 327 C, b) 345C, c) 353C and d) 362C. With increasing sintering temperature, stress a nd strain to failure, there is an increased area fraction of elongated material protruding from the surface. This elongated mate rial is thought to dominate the ductile mode of failure (m icrovoid coalescence), while the area not covered by elongated protrusions likely e xperienced brittle fracture from poor interparticle cohesion.
120 Figure 6-10. Secondary electron images of the fr acture surfaces of PTFE with varying sintering temperatures; a) 327C, b) 345C, c) 353C and d) 362C. Fibrils were drawn from each of the surfaces during fracture. With in creasing sintering temperature, stress and strain to failure, the fibril bundles tend to be thicker and denser, suggesting that a much larger fraction of the material wa s involved in the support of the load.
121 Tribological Characterization Average friction coefficients and wear ra tes for samples created with varying hold temperatures are listed in Tabl e 6-2. Friction coefficient a nd volume loss are plotted versus sliding distance in Figures 6-11a and 6-11b, resp ectively. In each case the friction coefficient decreased with increased sliding distance. This is typical of PTFE; as it runs in, transfer films are formed on the counterface and the near surface region of the polymer becomes more oriented. Optical images of the transfer film s throughout testing are shown in Figure 6-12. The first frictional data point was taken largely out of the influence of transfer films. The initial friction coefficient varied from = 0.135 to = 0.155. Interestingly, this initial friction coefficient was lowest for th e fully sintered 362C sample, but there was no trend with temperature as the 353C sample had the highest in itial friction coefficient. Statistically, these differences are significant, but they are likely driven by the unknown deterministic influences of the chemical and topological natures of the counterface and pin rather than the bulk pin morphology. In general, the friction coefficient decreased with increased sliding distance in a manner consistent with transfer film formation and surfac e orientation. Slight deviation from this trend was observed for the 345C sample. The frictio n coefficient increases sharply following a gradual decrease near 50m of sliding. A dramatic decrease follows with an abrupt increase occurring just before the third test interruption at 1600m of slidi ng. Upon restarting the test, the friction coefficient again decrease d, reaching near steady-state at a friction coefficient of about = 0.105 before the forth test interruption at 2600m Qualitatively, the transfer film morphology of the 345C sample after 2600 m of sliding is st riated in the sliding direction with a banded structure having a characteristic length of about 100 m, while many of the other images capture the more typical plate-like transfer mor phology. The surface topographies were also
122 quantitatively studied using a stylus profilometer. Surface measurements acr oss the transfer film of each sample are shown for 2600 m of sliding in Figure 6-13. Both the 327C and the 345C transfer films cover a significantly high frac tion of the counterface and both are thick (~30-40 m) compared to the 353C and 362C samples (5 -15m). The 327C sample also had striations in the sliding direction that are on the order of 10 m high, while the low friction 345 C transfer film was striated with a height of 30 m Both striated transfer films appeared to have formed on top of pre-deposited pl ate-like transfer with a characteri stic height of ~20 m. During the period of striated transfer and low friction for the 345C sample, a transition to low wear also occurred (Figure 6-11b). Despite th e fact that the 345 C was able to form a striated transfer film which led to low friction and wear, the underlyi ng transfer was unstable and the sample was unable to retain low friction and wear for a significant proportion of the total slidi ng distance. Figure 6-11. Tribological resu lts of wear testing PTFE samples with varying sintering temperature; a) friction coefficient plotted versus sliding distan ce and b) wear volume loss plotted versus sliding distance. The normal load was 250 N and the normal pressure was 6.3 MPa. The sliding speed was 50 mm/s and the reciprocation length was 25.4 mm. The tribologica l properties are largely i ndependent of the sintering temperature, crystalline mor phology, strength and toughness.
123 Figure 6-12. Optical images of the transfer films of PTFE samples with varying sintering temperatures following test interruptions at various sliding distances. The transfer morphologies range from a striated tran sfer morphology aligned into the sliding direction to the thick plate-like transfer typically cited for PTFE.
124 Figure 6-13. Stylus profile measurements across the transfer films of PT FE samples of varying sintering temperature following 26 00 m of sliding at 6.3 MPa. The transfer films of the 327C and 345C samples were significan tly thicker than those of the 352C and 362C samples. The large striations (~100m wide by ~20 m tall) of the 345 C sample coincided with a period of low friction and low wear sliding. The uncertainty in the measurement is on the order of 10nm in the vertical direction and 1 m in the horizontal. The average friction coefficients and steady wear rates are plotte d versus sintering temperature in Figure 6-14. There is a 40% di fference in wear rate from 345C to 353 C. Though significantly larger than th e uncertainty and a seemingly substantial difference, such differences are found with nominally identical samples of unfilled PTFE The variation in
125 friction coefficient is also small compared to variations that are typically found. Therefore the tribological differences observed here are insignificant despite the extreme differences in the crystalline morphology and in the mechanical st rength and elongation to failure. While the 345C sample morphology did promote a striated transfer morphology an d significantly lower friction and wear for short duratio ns, even these differences were negligible when compared to samples with trace (<1%) loadings of certain na nofillers in prior studies. The effects of the nanoparticles likely extend beyond mechanical lo ad support and stabiliza tion of the virgin morphology. Figure 6-14. Wear rate and fric tion coefficient plotted versus si ntering temperature. Uncertainty confidence intervals on wear ra te data are much smaller than the data points. The boxes on friction coefficient data represent the experimental uncertainty while the error bars represent the standard deviation throughout the test.
126 Table 6-2. Experimental matrix investiga ting the effects of sint er temperature on the tribological, thermal and mechanical pr operties of unfilled virgin PTFE. The uncertainty on melt temperature measurements is 0.1C. Temperature (C) 300 327 345 353 362 Tm1 (C) 343.75 343.97 343.89 344.60 328.90 Hm1 (J/g) 73.4 73.9 74.2 54.1 24.8 Uc(Hm1) 0.8 0.9 0.9 0.7 0.3 Tm2 (C) 328.04 328.01 327.89 328.01 327.99 Hm2 (J/g) 22.3 22.5 22.2 22.0 22.3 Uc(Hm2)0.3 0.3 0.3 0.3 0.3 (mg/mm3)2.230 2.210 2.221 2.150 2.107 Uc() (mg/mm3)0.012 0.011 0.011 0.010 0.010 u (MPa) N/A 1.17 3.98 6.24 7.67 Ucu )N/A 0.08 0.07 0.10 0.10 u (%) N/A 1.14 2.40 2.51 4.49 Ucu)N/A 0.01 0.01 0.01 0.01 N/A 0.116 0.116 0.119 0.119 () N/A 0.007 0.011 0.007 0.004 Uc() N/A 0.005 0.005 0.005 0.005 k x106 (mm3/Nm) N/A 508 418 593 493 U(kx106) N/A 3 2 3 2
127 Mechanical Processing on Nanoparticle Dispersion The beneficial thermal and chemical propert ies that make PTFE an attractive extreme environment solid lubricant also make it a particularly difficult matrix to work. The available techniques consist of dry powder blending and powder ultrasonication in an aqueous bath. Neither of these techniques is used fo r most other polymeric nanocomposites as in-situ polymerization and melt mixing have proven to be far superior nanoparticle dispersion techniques. One hypothesis for the succe ss of prior nanocomposites alumina-PTFE nanocomposites processed using a jet-mill powder blending technique is that the high amount of mechanical energy used in this process results in superior disbandi ng of the nanoparticle agglomerates which leads to improved nanoparticle disp ersion. It is thought that the benefits of polymeric nanocomposite can only be realized on ce the nanoparticles have been effectively dispersed. In this study, the effects of nanoparticle agglomerate disb anding and dispersion are under investigation. Three alumina nanoparticle lo adings, namely, 1wt%, 2wt% and 5wt%, were mixed by hand. Half of each batch was then blende d using the jet-mill. It should be recalled that the jet-mill was the only technique energetic enough to alter the thermal behavior, size and shape of the virgin powder. All of th e PTFE in the study was jet-milled prior to nanoparticle inclusion to remove jet-milling effects as variables. The hand-mixing technique was used as a very low energy, worst-case extreme to counter the high energy jet-milling technique. Following powder blending, the powders were obs erved using high vacuum seconda ry electron microscopy. Ten random samples of each powder condition were obser ved, and the numbers of alumina particles visible on the much larger PTFE particles were noted.
128 Figure 6-15. Representative SEM images of a) 5 wt% 80 nm phase alumina hand-mixed with PTFE (secondary electrons), b) 5 wt% 80 nm phase alumina jet-milled with PTFE (secondary electrons), c) neat jet-milled PTFE (secondary electrons), d) hypothesized nanoparticle agglomerate (secondar y electrons), e) 5 wt% 80 nm phase alumina hand-mixed with PTFE (backscatte red electrons), f) 5 wt% 80 nm phase alumina jet-milled with PTFE (backscattered electrons)
129 Representative images of the hand-mixed and jet-milled powder dispersions at 5wt% loading are shown in Figure 6-15a-b. These images can be contrasted to the inset image of neat PTFE shown in 6-15c. In the images of th e nanoparticle-PTFE powder ensembles (Figures 615a and 6-15b), submicron domains with high brightness were easily observed with a 50 m field of view. Spectroscopy was not conducted to determine the composition of these domains, however, comparable regions were never observed on neat PTFE powders and the densities increased with increased nanoparticle loading for a given blending c ondition. These facts suggest that these bright submic ron regions are alumina nanopartic les. Under this assumption, these images suggest a large discrepancy in the number of particles decorating the PTFE after hand-mixing and jet-milling. In the case of the hand-mixed samples, 2-15 m particles with nodular surfaces were observed with the density in creasing with increased nanoparticle loading. The appearance of these partic les is consistent with sec ondary electron images of the agglomerations in the as-received 40 nm phase alumina powder of McElwain . Backscattered electron imaging of these powders show bright contra st for these regions. Given the higher atomic weight of Al and large interaction volume (~1 m) for backscattered electrons, these results suggest that the nodular regi ons are nanoparticle agglomerates. Lower magnification backscattered electron imaging of the powders, illustrated by representative images in Figures 6-15e-f, revealed the pres ence of larger and more spatially frequent agglomerates in the hand-mixed powders. Observations of dispersions of individual nanoparticles (using secondary electron imagi ng) and nanoparticle agglomerates (using backscattered electron imaging) qualitatively suggest that jet-milling effectively disbanded agglomerates and dispersed nanoparticles while hand-mixing did not.
130 Figure 6-16. Estimated nanoparticle density plot ted versus alumina loading for hand-mixed and jet-milled powder samples. The error bars re present the standard deviation in the ten random measurements. The center line is the calculated average and the box represents plus and minus one standard de viation of the distri bution of averages calculated from 10 random samples from the population (i.e. the average of the population lies within the box with 68% confidence). Table 6-3. Estimated nanoparticle density resu lts of SEM observation of nanoparticle decorated PTFE powder following mixing via hand-mixing and jet-milling for 1wt%, 2wt% and 5 wt% 80 nm alumina in PTFE. Ten random samples were taken from each condition. The mean and standard deviati on of the ten measurements are tabulated. Sample Mixing technique Alumina wt% particle density (particles/m2) particle density (particles/m2) 1 Hand-mixed 1 5.85x10-4 3.91 x10-4 2 Hand-mixed 2 8.18 x10-4 6.41 x10-4 3 Hand-mixed 5 35.3 x10-4 13.4 x10-4 4 Jet-milled 1 21.5 x10-4 13.3 x10-4 5 Jet-milled 2 64.2 x10-4 24.3 x10-4 6 Jet-milled 5 299 x10-4 98.8 x10-4 The estimated nanoparticle densities from SE M observations of the powders are tabulated in Table 6-3. A comparator pl ot of the estimated nanoparticle density (number of particles divided by the area of coverage) versus alumina wt% is shown in Figure 6-16; nodular particles
131 or nanoparticle agglomerates are only counted once. While there are clearly substantial errors associated with the calculation of number density with this tec hnique, relative comparisons are thought to be fair. The error bars represen t the standard deviation of the measurement and the box represents the standard deviation of the calculated mean. The error bars are theref ore about three standard deviations out, and the population mean lies within those error bars with 99.7% confidence. It can therefore be concluded that there are substa ntially more nanoparticle s (approaching an order of magnitude) decorating the surfaces of PTFE powders following jet-milling than following hand-mixing. Based on the number of hypothesize d agglomerations observed during surveying of the hand-mixed powders, the difference is attr ibutable to the relativ ely poor disbanding of nanoparticle agglomerates in the case of the hand-m ixed powders. In addition, the jet-mill is an open system and it is thought that many of the fine nanoparticles may be lost during the jetmilling procedure, while the closed-system in the hand-mixing procedure ensures an average loading of 5 wt%. Sawyer et al.  used thermal gravimetric analysis to estimate the loss of 60% of the nanoparticles during jet-milling. It is clear from this study that the high energy of the jet-mill helps break-up notoriously sticky nanop article agglomerations and leads to a much higher density of nanoparticle coverage despite potentially having a lower alumina weight fraction. Nanoparticles on the Melt Behavior of PTFE In the previous study it was shown that the number density of jet-milled powders is significantly higher than that of hand-mixed powders for a given nanoparticle loading. One of the potential wear resistance mechanisms is th e alteration of the crys talline structure and morphology of the PTFE by the nanoparticle incl usions. It is hypothesi zed that, because the nanoparticles are of the same-size scale as the polymer lamellae, they may nucleate
132 crystallization and promote a fine r crystalline structure than is otherwise possible. The size reduction of the characteristic ba nded structure of PTFE was pr eviously cited by Tanaka as a wear resistance mechanism. In Tanakas experi ments, the cooling rate following sintering was used to alter the size of the bande d structure, not nanoparticles. Figure 6-17. Thermograms from differential scanning calorimetry of alumina-PTFE powder ensembles blended by hand and by jet-milling. These curves are compared to jetmilled unfilled PTFE powders. All of the PTFE powders in this study were et-milled prior to nanoparticle inclusion. In this study, the thermal characteristics of the alumina-PTFE powder ensembles are studied using differential scanning calorimetry at a heating and cooling rate of 10C/minute. Each powder was heated to 400C, equilibrated, c ooled to 250C, equilibrated and reheated to 400C. With this experiment, the first melt, cr ystallization and second melt of the powders are studied in the absence of other thermal or mech anical histories. The thermograms from these
133 measurements are shown in Figure 6-17. The quantified results from these measurements are tabulated in Table 6-4 and are plotted versus filler wt% in Figure 6-18. First consider Figure 6-18a, where the first pe ak melt temperature is plotted versus the filler wt%. The jet-milled powder ensembles behaved in a manner similar to the jet-milled PTFE, although with a slightly higher melt temper ature after nanoparticle inclusion. The handmixed powder ensembles had a significantly high er first melt temperature regardless of the loading. In fact, these powders had a first me lt temperature similar to the values found for virgin, sieved, ultrasonicated, rotary mixed a nd blade mixed PTFE at around 342.5C. It is possible that the large nanoparticle agglomerations tend to increase the melt te mperature to a greater extent than the smaller singular nanopartic les in the jet-milled powder ensemble, but one would expect temperature to increase with alumin a content. Given that the melt temperature was reduced slightly with increased loading, a more probable cause for this is the hand mixing action and the lack of nanoparticle coverage. The hand mixing may have led to agglomeration of the relatively sticky PTFE. Because the nanopar ticles did not effectively decorate the PTFE surfaces, the particles may tend to stick during mixi ng. This would also lead to a reduced melt temperature with increased alumina content si nce the higher particle density would deter agglomeration of the polymer. The increase in the first melt of the jet-milled powders does appear to be statistically signi ficant. This phenomenon may be re lated to the original hypothesis that the nanoparticles stabilize the PTFE to slightly higher temperatures by immobilizing the contacting lamellae. However, the difference is much less than 1C and is insignificant in the context of the polymer processing. The heat of fusion of the first melt is shown in Figure 6-18b. There is a clear trend of decreased heat of fusion with increased loading. However, the mass of the sample included the
134 mass of the alumina which does not contribute to th e power signal. If the as-prepared alumina loading is considered, an iso-crystalline line can be constructed. This line represents the behavior of the nanocomposite if it contained the as-prepared mass of alumina and had the same polymer crystallinity as the unfilled polymer. Thus, we can use the iso-crystalline line to compare the crystallinity of the nanocomposite to th at of the neat polymer. After considering the iso-crystalline line, it is clear that the crysta llinities of the nanocomposites are comparable to those of the unfilled polymer. The 5wt% samples may have had slightly lower heats of fusion and thus crystallinity, but the di fference is not substantial eno ugh to draw any conclusions. The recrystallization behavior is shown in Fi gures 6-18c and 6-18d. The recrystallization temperatures (6-18c) of the nanocomposites fluctu ate about the value obtained for unfilled jetmilled PTFE in a non-systematic manner with magnit udes that are greater than the uncertainty in temperature. This scatter is insignificant. The uncertainty in power is rather large (~1%) and the recrystallization peak is broad. The peak temperature is determ ined as the location where peak power is measured. As a result, even though the measurement of temperature is much more accurate than the fluctuation, the fluctuati on of power on the broad curve tends to make determination of the peak temperature difficult an d increases the uncertainty of peak temperature in this case. Hence, despite the hypothesis that the nanoparticles may nucleate crystallization and promote a finer scale crystalline structure, evidence of this was not observed in the crystallization behavior of the nanocomposites. In the heat of recrystallization curves in Figure 6-18d, differences between the jet-milled and hand-mixed powders are found; the low wt % jet-milled powders behaved like the unfilled powder, while the hand-mixed powders had a lower heat of fusion. Once again, this difference is comparable to that found between the treated and untreated powders in the earlier thermal studies
135 of neat PTFE at about 2 J/g. This furthe r supports the hypothesis th at the hand mixing action may tend to agglomerate the PTFE leading to virgin-like characteristics. Both sets of powders showed a slight tendency to increase crystall inity with increased loading; the hand-mixed powders showed a greater tendency than the jet-milled powders. This may be a result of the higher nanoparticle density decreas ing the agglomeration tendency. The results from the second melt are shown in Figures 6-18e and 6-18f. The differences in the peak melt temperatures shown in Figure 6-18e are statistically insignif icant. This supports the hypothesis that the differences in the crysta llization temperatures ar e insignificant and are simply the result of the uncertainty in power meas urements and the inherently low sensitivity due to the broadness of the crystallization peak. As expected, the heats of fusion in Figure 6-18f almost replicate the heat of cr ystallization data in 6-18d.
136 Figure 6-18. Quantified results from differentia l scanning calorimetry of alumina-PTFE powder ensembles blended by hand and by jet-milling; a) first peak melt temperature plotted versus alumina wt%, b) first heat of fu sion plotted versus alumina wt%, c) peak crystallization temperature plotted versus alumina wt%, d) heat of fusion during crystallization plotted versus alumina wt%, e) second peak melt temperature plotted versus alumina wt% and f) second (melting) heat of fusion plotted versus alumina wt%. Error bars represent the experi mental uncertainty in each case.
137 Table 6-4. Results of DSC of powder ensembles following mixing via hand-mixing and jetmilling for 1wt%, 2wt% and 5 wt% 80 nm alumina in PTFE. The uncertainty of temperature measurements is 0.1C Jet-milled Hand-mixed Jet-milled Unfilled 1 wt% 2 wt% 5 wt% 1 wt% 2 wt% 5 wt% Tm1 (C) 340.9 342.3 342.1 342.0 341.2 341.3 341.2 Hm1 (J/g) 72.7 72.3 72.3 67.8 72.2 71.6 67.9 Uc(Hm1) 1.0 0.9 1.1 1.0 1.0 1.0 1.0 Tc1 (C) 314.7 315.4 315.0 315.4 314.4 314.9 314.4 Hc1 (J/g) 30.3 28.2 28.1 28.0 30.3 29.4 29.0 Uc(Hc1) 0.4 0.4 0.4 0.4 0.4 0.4 0.4 Tm2 (C) 327.4 327.6 327.4 327.4 327.5 327.5 327.4 Hm2 (J/g) 29.9 27.7 27.6 27.9 30.3 28.9 28.5 Uc(Hm2) 0.4 0.4 0.4 0.4 0.4 0.4 0.4 Filler Dispersion on Nanocomposite Properties Thermal Characterization In the prior investigation of the influe nces of nanoparticles on the melt and recrystallization behavior of PTFE, the nanoparticle s were found to have little effect. Deviations were found between the two dispersion techniqu es, namely, jet-milling and hand mixing, and were attributed to a tendency of the PTFE par ticles to re-agglomerate during hand mixing due to the poor nanoparticle decoration observed in the previous disper sion study. However, very different phase and melt characteristics were found previously for eff ective nanocomposites and unfilled PTFE, so the thermal characteristics of the processed nanocomposites were studied here to include the thermal and pressure hist ory of the compression molding process. The differential scanning calorimetry therm ograms for 1, 2 and 5 wt%, hand-mixed and jet-milled alumina-PTFE nanocomposites are show n from 280C to 380C in Figure 6-19. One repeat of each sample was conducted. Unfilled PT FE is also shown as the control sample. The unfilled PTFE processed at 362C has the traditio nal low temperature melt peak at about 327C, but it also has a subtle melt peak at the hi gh temperature melt of about 350C. The 2 wt%
138 samples had a prominent high temperature me lt peak, while the 1 and 5 wt% samples only showed the low temperature melt peak. The repeat tests verify that the un usual discontinuity in behavior at 2 wt% is repeatable with repeated sampli ng of the same compression molded specimen. In order to test whether this is tr uly a function of the load ing or a coincidence of varying processing conditions, a repeat specimen was comp ression molded under nominally identical conditions. The DSC results for the fi rst melt of the repeat specimens are shown in Figure 6-20. These results show only the low temperature melt peak for the 2 wt% jet-milled sample and suggest inconsistency in the proce ssing either due to disp ersion or the compression molding process. Such stark behavioral differences were not observed during the well controlled cooling and heating conditions of the recrys tallization and second melt in the DSC which further suggests that the behavioral differences observed in th e first melt were due to unknown deterministic variations from compression molding rather than di fferences in dispersions. The samples with the high melt peak likely had higher pressure s during processing whic h increased the melt temperature. Even during well-controlled crystallization conditions in the DSC, the nanocomposites had a tendency to have higher crys tallinities than the unfilled PTFE.
139 Figure 6-19. Heat flow plotted versus temper ature from differential scanning calorimetry of hand-mixed and jet-milled 0, 1, 2 and 5 wt% alumina-PTFE nanocomposites compression molded at 362C: a) first me lt, b) crystallization and second melt. Figure 6-20. Heat flow plotted versus temper ature from differential scanning calorimetry of repeat specimens of hand-mixed and jet-milled 0, 1, 2 and 5 wt% alumina-PTFE nanocomposites compression molded at 362C. Samples with nominally identical composition, dispersion and processing we re found to have significant thermal behavioral differences in some cases.
140 The quantified results with experimental uncer tainties for these studies are found in Table 6-5 and peak temperatures and heats of fusion are plotted versus filler wt% in Figure 6-21. Consider 6-21a in which the first melt peak temperature is plotted versus filler wt%. The peak melt temperature remains nominally unchanged for all of the samples except for the 2 wt% nanocomposites; these first melt peak differences are likely due to inconsistencies in the processing. The heat of fusion is plotted versus filler wt% in figure 6-21b; lines of constant heat of fusion have also been plotted. Overall, there is a strong trend of increa sed heat of fusion, and therefore, crystallinity as the filler loading incr eases. The lack of a high temperature melt peak for any of the 5 wt% samples suggests that the sa mple were fully sintered and the 30% increase in crystallinity of these samples suggests that the nanoparticles may have facilitated crystallization. The very high crystallinity of the 2 wt% samples is attributable mostly to incomplete sintering (indicated by the high melt peak temperatures). One jet-milled 2 wt% sample was fully sintered (low first melt temperat ure) and had increased cr ystallinity consistent with the trend from the 5 wt% samples. The recrystallization behavior is shown in Figures 6-21c and 6-21d. The effects of pressure and thermal history are attenuated after the first heat, so the re crystallization behavior mostly reflects the nanoparticle effects on recrystalliz ation of the polymer. It should be noted that the effects are not totally er ased since most of the incompletely sintered samples are entering first melt while others enter the second melt. There is essentially no difference between the recrystallization temperatures. The heats of fu sion for recrystallizati on in Figure 6-21d again support a trend of increasi ng crystallinity with incr easing filler content, but there is now lower crystallinity of the 2 wt% samples than for the ot her nanocomposites. This is attributed to the differing processing histories prior to the first melt of DSC. Th e second melt behaviors shown in
141 Figures 6-21e and 6-21f are expectedly quite simila r to the recrystallizatio n behaviors in Figures 6-21c and 6-21d. Table 6-5. Results of DSC of hand-mixed and jet-milled 1wt%, 2wt% and 5 wt% 80 nm alumina-PTFE compression molded nanocompos ites. The first two samples originate from the same compression molded sample used for tribological testing while the third sample (first melt only) originated from another compression molded sample used for mechanical testing. The uncertainty of temperature measurements is 0.1C. The uncertainty of enthalpy calculations is 0.4 J/g unless specified otherwise. Jet-milled Hand-mixed Jet-milled Unfilled 1 wt% 2 wt% 5 wt% 1 wt% 2 wt% 5 wt% 328.5 344.0 329.4 328.9 343.3 329.9 328.6 344.2 329.8 329.0 343.3 330.0 Tm1 (C) 328.9 329.4 344.2 328.9 329.7 329.7 329.5 33.4 53.5 41.8 35.7 63.0 43.0 30.8 49.1 42.3 38.2 62.7 40.2 Hm1 (J/g) 32.5 40.5 49.2 33.8 34.4 38.9 35.6 0.4 0.7 0.6 0.5 0.8 0.6 0.4 0.7 0.6 0.5 0.8 0.6 Uc(Hm1) 0.4 0.6 0.7 0.4 0.4 0.5 0.5 315.5 316.5 316.2 315.7 316.3 316.6 Tc1 (C) 316.1 315.6 316.6 315.8 315.5 316.4 316.3 31.4 28.9 35.2 32.4 31.9 35.33 Hc1 (J/g) 29.3 34.6 30.2 35.7 33.4 30.9 33.9 327.7 328.2 328.5 327.6 328.2 328.4 Tm2 (C) 328.0 327.7 328.2 328.9 327.6 328.3 328.6 31.0 29.4 36.2 31.1 30.9 34.0 Hm2 (J/g) 28.6 31.9 29.9 36.3 31.4 30.8 34.0
142 Figure 6-21. Quantified results of differential scanning calorimetry of hand-mixed and jet-milled 0, 1, 2 and 5 wt% alumina-PTFE nanocom posites compression molded at 362C: a) first peak melt temperature plotted versus fi ller wt%, b) first h eat of fusion plotted versus filler wt%, c) peak crystallization temperature plot ted versus filler wt%, d) heat of fusion for crystallization plotted versus filler wt%, e) second peak melt temperature plotted versus filler wt%, f) s econd heat of fusion plotted versus filler wt%. Error bars represent the experi mental uncertainty in each case.
143 Mechanical Characterization Wear rates and friction coefficients are ofte n strongly correlated to various mechanical properties of the materials involved . Since wear is a process of material removal, resilience and toughness often have the best correlation w ith low wear. In past studies, extraordinarily low wear rates of these nanocomposite materials were found to accompany extensive fibrillation of the PTFE under concentrated loadings . In this study, the mechanical properties of these nanocomposite materials are quantified using a calibra ted load frame to study the effects of the nanoparticle s on load support, elongation to failure and toughness of the polymer. Sections of the compression molded specimens were removed and polished prior to mechanical testing. These samples were studied under an optical microscope at 20X magnification. Figure 22 shows the results of optical microscopy. Despite the vast differences observed in the dispersions of the jet-milled and hand-mixed powders, the compression molded samples show the same unique microstructu re which has what appears to be ~10 m diameter inclusions that increase in propor tion to the loading. Studying just these images, one might presume that the inclusions are simply alumina agglomerations, but such a model is difficult to rationalize given the previous SE M observations of the powder ensembles, which showed that the hand-mixed samples contained alumina agglomerations on the ~2-15 m size scale with sparsely distributed singular nanoparticles, while jet-milled powders had less frequent and smaller alumina agglomerations (1-5m) with an order of magnitude hi gher density of singular nanoparticles decorating the PTFE pow der surfaces. It is possible, but unlikely, that the alumina sought itself and diffused through the melt resulting in substantially different dispersions before and after melting has occurred. Another possibility is that the domains ap pear opaque due to an alteration of the crystalline st ructure of the PTFE. This c ould occur as a result of a high
144 nanoparticle concentration in the vicinity of the PTFE particle. Such an explanation would be consistent with several observations: 1) the na noparticles were found in high concentration only on the large round PTFE particles, 2) the crystallinity of the PTFE tended to increase with nanoparticle loading, 3) The aver age domain size corresponds much more closely to the larger round PTFE particles than to the alumina agglom erations observed in electron microscopy of the powders. Mapping Raman spectroscopy was used to determine if alumina was more or less prevalent within these regions. The mapping region was divided in to 100 pixels, and 20 spectra were taken at each pixel. Standard samples of PTFE and alumina were examined and the peaks of each sample were established. The 20 spectra taken at each pixel were averaged, and the alumina peak was normalized to the defining PTFE peak. A contour plot of the relative alumina intensity is plotted for the area in the accompanyi ng optical micrograph in Figure 6-23(a-b). The results of Raman spectroscopy indicate that the op aque domains are alumina rich. In a parallel study, one of these domains was plucked fr om the matrix, smashed between two glass microscope slides and observed using SEM (Fi gures 6-23(c-f)). The bulk within the domain does contain individual nanopartic les but appears to be primarily comprised of the polymeric matrix. The cause and structure of these domains remain unclear, but preliminary results are consistent with the hypothesis th at they are nanoparticle rich regions of PTFE with altered crystalline structure.
145 Figure 6-22. Optical micrographs of the micrstructure of the nanocomposites at 20X magnification: a) 1 wt% jet-milled, b) 1 wt% hand-mixed, c) 2 wt% jet-milled, d) 2 wt% hand-mixed, e) 5 wt% jet-milled, f) 5 wt% hand-mixed.
146 Figure 6-23. Results from analysis of opaque microstructural domains within the nanocomposites: a) optical image of the area of Raman spectroscopic analysis, b) contour maps of the alumina peak normalized to the PTFE peak following Raman spectroscopic mapping, c) a single domain plucked from the nanocomposite, d) the domain smashed between glass microscope slides, e) 50/50 overlay of a secondary electron image and a backscattered electr on image of the smashed domain and f) 50/50 SE/BSE image of a region of the smas hed domain at higher magnification.
147 The results of mechanical te sting of the nanocomposites are shown in Figure 6-24. Stress is plotted versus engineering strain for jet-milled and hand-mixed nanocomposites in Figures 624a and 6-24b, respectively. Ultimate stress and strain are plotted versus filler wt% in Figures 624c and 6-24d, respectively. Similar ranges of ultimate stresses and st rains, namely, 8-16 Mpa and 250-600%, respectively, were obtained for both jet-milled and hand-mixed dispersion treatments. This result was unexpected given the original hypothesis that the dispersions drive the nanocomposite properties and that the hand-mixed dispersion represents a worst-case scenario while the jet-milling technique was t hought superior to other more traditional techniques. Even more unexpected was the fact that the hand-mi xed samples actually outperformed the jet-milled samples in terms of consistency in a qualitative sense. The poorest performing hand-mixed sample was the 2 wt% sample The bottom half of this sample yielded after about 20% strain and was unable to support enough tensile load to yield the rest of the sample. The cause of this selective failure is unclear but a thermal and/or pressure gradient is suspected to have cause incomplete sintering of only part of the sample. The DSC scan in Figure 6-20 from a section near the center of this sample supports the hypothe sis showing that only partial sintering occurred. Despite the weakness in this part of the sample, the total strain was nearly 300%, and the length corrected ~600% strain to failure is consistent with the rest of the hand-mixed field. The jet-milled samples were le ss consistent, showing no trend with loading or with the thermal behaviors shown in Figure 6-20. Despite these nanocomposites having a degr ee of inconsistency in their mechanical behaviors, all of the samples were strengt hened and toughened signi ficantly due to the nanoparticles with the least tough nanocomposite being over 100X tougher than unfilled PTFE sample processed at the same conditions. In add ition to the impressive gains in ultimate strain,
148 the stresses shown here reflect th e nominal pre-strained cross-sect ional areas of the samples and not the true area. Therefore, gi ven the amount of strain in the cross-section, the true stresses at failure were substantially higher. For example, the 2 wt% jet-milled sample had a final unstressed cross-sectional area of about 5 mm2 and the true stress carried by the sample at failure was approximately 40MPa. These stress and strain differences over neat PTFE are impressive considering the low loading of the nanoparticles and suggest that the improvements arise from a unique crystalline structure that re sults from the nanoparticle influe nce rather than a direct load carrying effect of th e nanoparticles. Figure 6-24. Results from mechanical charact erization of 80 nm alpha phase alumina-PTFE nanocomposites, a) stress versus strain for jet-milled nanocomposites, b) stress versus strain for hand-mixed nanocomposites, c) ultimate stress versus filler wt%, d) ultimate strain versus filler wt%.
149 Scanning electron microscopy images of the fr acture surfaces are shown in Figure 6-25. The alumina rich domains that appear in optical microscopy also appear bright in backscattered electron imaging. The domains have not been stra ined appreciably (or at all) during mechanical testing and do not appear to dir ectly contribute to the load carrying capacity of the composite. Various voids can be found where domains have been plucked out during fracture. Evidently, there is poor cohesion between the domains and the matrix as removal of the domains results in imperceptible damage to the surrounding matrix. Varying matrix fracture morphologies accompany the variations in mechanical behaviors. In all cases, the failure mechanis m is characterized by extensive fibrillation. The deformation behavior of the matrix is drastically different from that of the unfilled polymer, which suggests that nanoparticles inhabit and alte r the matrix in between the alum ina rich domains. The surfaces of the 1 wt% jet-milled and 5 wt% hand-mixed nanocomposites are smooth and suggest the least local fibril extension and rapid break at failure. It should be noted that the bright, smooth regions on the 1 wt% jet-milled sample represent locations where the fibrils were smashed flat from contact with the sample holder. The 2 wt% jet-milled and 1 wt% hand-mixed samples had the best combination of stress and strain at failure and demonstrate uniformly and substantially strained and elongated material at the surfaces. The 2 wt% handmixed sample shows extensive small-scale fibrillation at the fracture surface wh ich reflects ductile failure of highly strained material. Recall that the true strain of this sample was closer to 600% rather than the reported 270% due to yielding and straining of only about half of the sa mple. The 5 wt% jet-milled sample also demonstrated small-scale fibrillation Portions of this surface appear very smooth indicating brittle failure. Following failure of th is portion of the surface, only very small areas remained for load support resulting in gross fibrillation and gradual loss of load carrying
150 capacity. Fibrils are very strong and can endure massive elongations before failure. As a result, both samples showing small-scale fibrillation (2wt% hand-mixed and 5 wt% jet-milled) required 10-15% strain to completely lose load carrying capacity after failure while the other samples required less than 0.1%. Interestingly, Michler  sugge sts that particle inclusions (e.g. alumina rich domains in this case), with weak interfacial strength with the matrix, provide a toughness mechanism in polymers by creating a stress concentration that results in local yielding and fi brillation (crazing). Higher magnification imaging of this weak interf ace did confirm such local fibrillation, but this mechanism was surely swamped by the effects of the nanoparticles on the remaining fibrillated matrix. The same mechanism may have been ac tivated by the nanoparticles in the regions in between the domains. Michler also notes a th in-layer yielding mechan ism where very small lamellae exhibit large-scale deformation whic h manifests itself into >100X increases in toughness and elongation. This is opposed to the smaller toughness in creases found for the crazing mechanism that typically occurs at the particle inte rface. It is thought that the nanoparticles may reduce the PTFE lamellae size to that which facilitates the thin-layer yielding deformation mechanism. Evidence for such a mechanism is provided by the AFM imaging shown in Figure 3-18. Table 6-6. Results of mechanical testing of hand-mixed and jet-milled 1wt%, 2wt% and 5 wt% 80 nm alumina-PTFE compression molded nanocomposites. Jet-milled Hand-mixed Jet-milled Unfilled 1 wt% 2 wt% 5 wt% 1 wt% 2 wt% 5 wt% (MPa) 7.7 13.7 10.1 14.6 13.3 15.2 12.0 Uc() 0.1 0.1 0.1 0.2 0.1 0.2 0.1 (%) 4.49 594 271 437 285 557 257 Uc() 0.01 8 5 6 4 7 3
151 Figure 6-25. Scanning electron microscopy of th e fracture surfaces of a) 1 wt% jet-milled alumina-PTFE, b) 1 wt% hand-mixed alumina-PTFE, c) 2 wt% jet-milled aluminaPTFE, d) 2 wt% hand-mixed alumina-PTFE, e) 5 wt% jet-milled alumina-PTFE, f) 5 wt% hand-mixed alumina-PTFE.
152 Tribological Characterization The tribological properties of the 1, 2 and 5 wt% jet-milled and hand-mixed nanocomposites were tested using a reciprocating tribometer with a normal pressure of 6.3 MPa, a sliding speed of 50 mm/s and 50 mm of travel per cycle of sliding. The results of tribological testing are shown in Figure 6-26. First consider the friction coe fficients plotted versus sliding distance for the jet-milled nanocomposites in Figure 6-26a. The 1 wt% sample ran for 5 km before reaching the maximum allowable volume loss of 50 mm3; this is about an order of magnitude more sliding than could be endured by unfilled PTFE. The friction coefficient of this sample was fairly steady for the 5 km duration of the test at = 0.17. The 2 wt% sample had significantly more variation in the friction coefficient than did the 1 wt% sample. For the first 5 km, the friction coefficient was actually lower for the 2 wt% sample than for the 1 wt% sample despite having a higher concentration of the hard alumina filler. However, after 10 km of sliding, the friction coefficient began to increas e until reaching a steady state value of = 0.22. Throughout the test, the friction co efficient is observed to decreas e and then increase abruptly. These abrupt changes are due to the test inte rruptions for the mass loss measurements. The decrease in friction coefficient is likely due to a combination of cont aminant adsorption after separation of the contact, and reduced contact area from the pin and counterface running surfaces being re-assembled out of regist ry. At 5 wt%, the initial friction coefficient during run-in was again lower than for both the 1 and 2 wt% sample s, but quickly transitio ned to a higher steady state value of = 0.27. The friction coefficien ts of the hand-mixed samp les are plotted versus sliding distance in Figure 6-26b. The 1 wt% sa mple had a friction coefficient behavior very similar to that of the jet-milled 1 wt% sample; this sample ran nearly twice the sliding distance of the jet-milled sample and showed a decrease in friction coefficient to = 0.15 toward the end of the test. The 2 wt% sample had the highest average friction coefficient among hand-mixed
153 samples and both the 2 and 5 wt% samples had si gnificantly shorter tran sient periods of low friction than the jet-milled samples. In almost every case, the friction coefficient varied dramatically with sliding distance. This is unlike the behavior observed for unfilled PTFE; unfilled PTFE had a transient friction coefficient that started at approximately = 0.15 and decayed to around = 0.12 as sliding continued with shor t-range variations of about 0.01. Figure 6-26. Results of tribol ogy experiments of hand-mixed and jet-milled 0, 1, 2 and 5 wt% alumina-PTFE nanocomposites compression mold ed at 362C: a) friction coefficient versus sliding distance for jet-milled nanocomposites, b) friction coefficient versus sliding distance for hand-mixed nanocom posites, c) volume loss versus sliding distance for jet-milled nanocomposites, d) volume loss versus sliding distance for hand-mixed nanocomposites.
154 The results of wear volume measurements ar e plotted in Figure 6-26c and 6-26d for jetmilled and hand-mixed samples, respectively. In both cases, enormous changes in the friction and wear behavior occurred with increased filler loading. The largest range of behaviors occurred in the jet-milled samples; the 1 wt% sample wore 45 mm3 of material in 4 km of sliding, while the 5 wt% sample wore less than 5 mm3 in 40 km of sliding. The rate of wear of the 1 wt% samples continuously decreased as s liding distance increased, suggesting that the entire test occurred under effectively transient sliding conditions. The 2 wt% jet-milled sample also had a substantial tr ansient period of sliding where the w ear rate continuously decreased with increased sliding distance until a steady state wa s reached at 5 km of sliding. The 5 wt% jetmilled sample had little to no transient sliding before reaching steady conditions for the remainder of the test. In every case, the ha nd-mixed samples showed shorter and less severe transient periods of wear. Despite having si gnificant agglomerati on and relatively poor nanoparticle dispersion, the 2 and 5 wt% hand-mixed samples qui ckly reached low steady state wear rates comparable to those found for jet-mill ed samples. Following a brief period of low wear sliding, however, a change in the tribo-system caused an increa se in the wear rate of about 2-4X. Wear rate is plotted vers us filler wt% in Figur es 6-27. The wear rates of hand-mixed and jet-milled samples behaved similarly as a function of filler wt%. Wear rate appears to correlate with the alumina loading and density of the opaque, alumina rich domains in the material rather than the degree of nanoparticle a gglomeration and dispersion.
155 Figure 6-27. Wear rate versus alumina nanoparticle wt% for hand-mixed and jet-milled 0, 1, 2 and 5 wt% alumina-PTFE nanocomposites compression molded at 362C. Table 6-7. Results of tribological testing of hand-mixed and jet-milled 1wt%, 2wt% and 5 wt% 80 nm alumina-PTFE compression molded nanocomposites. The uncertainty in the friction coefficient measurements is less than Uc()=0.005. Jet-milled Hand-mixed Jet-milled Unfilled 1 wt% 2 wt% 5 wt% 1 wt% 2 wt% 5 wt% 0.119 0.165 0.276 0.243 0.173 0.169 0.239 () 0.004 0.010 0.027 0.029 0.005 0.029 0.053 k (x 10-7 mm3/Nm) 4930 189 7.43 5.27 423 2.54 1.64 Uc(k (x 10-7 mm3/Nm)) 20 4 0.05 0.03 3 0.1 0.02 Each transfer film was measured using 3D stylus profilometric mapping. The area measured by profilometry was examined using SEM; backscattered electron images are shown below the accompanying topography maps in Fi gures 6-28 and 6-29. It is clear from the profilometry that the nanocomposites were abrasive to the counterfaces. It is also clear that the
156 hand-mixed samples were significantly more abra sive to the counterfaces than were the jetmilled samples. This is likely the cause of the 2-4X higher steady state wear rates of the handmixed samples. The portions of the surfaces cove red by transfer films appear are represented by the dark regions of the surface in the backsca ttered images. Figure 6-28. Analyses of wear tracks following the tribological experiments with hand-mixed nanocomposites: a) stylus profilometric map of the wear track for the 1wt% composite, b) backscattered electron image of the area from profilometry for the 1wt% composite, c) stylus profilometric map of the wear track for the 2wt% composite, d) backscattered electron image of the area from profilometry for the 2wt% composite, e) stylus profilometric map of the wear track for the 5wt% composite, f) backscattered electron image of the area from profilometry for the 5wt% composite
157 Figure 6-29. Analyses of wear tracks following the tribological experiments with jet-milled nanocomposites: a) stylus profilometric map of the wear track for the 1wt% composite, b) backscattered electron image of the area from profilometry for the 1wt% composite, c) stylus profilometric map of the wear track for the 2wt% composite, d) backscattered electron image of the area from profilometry for the 2wt% composite, e) stylus profilometric map of the wear track for the 5wt% composite, f) backscattered electron image of the area from profilometry for the 5wt% composite.
158 Filler Material on Nanocomposite Properties Nanoparticle Dispersion It was hypothesized previously that the domin ant factor in the tribological success or failure of a PTFE nanocomposite is the nanopartic le dispersion. Nanoparticles are inherently difficult to disperse due to the large ratio of the surface forces to the inertial forces, and PTFE is a difficult matrix to disperse nanoparticles within due to its high melt viscosity, difficult polymerization and sticky nature. Hand-mi xed and jet-milled nanocomposites were created and tested with the intention that the hand-mixed dispersions would be as poor as one would expect in practice, while the jet-milled samples were thought to have a very good dispersion. SEM observations of the powders confirmed that the hand-mixed powders primarily consisted of agglomerated filler (with agglomerates being of the 2-15 m size scale), while jet-mill samples had few small agglomerations and an order of magnitude more individual nanoparticles decorating the PTFE surfaces. Unexpectedly, th e hand-mixed samples performed comparably to jet-milled samples in mechanical testing and tribological testing being orders of magnitude tougher and more wear resistant than unfilled PT FE. These results suggest that perhaps the shape or chemistry of the nanoparticles dominate the behavior of the nanocomposites rather than dispersion. In a previous st udy, it was found that if the phase alumina nanoparticles of the current investigation were treated with a fluorin ated silane coupling agent, the tribological properties of the nanocomposites improved. It was thought that a reduction in surface energy contributed to improved dispersi on which resulted in improved pr operties at lower loadings, but the latest results suggest that the differences may be due to some beneficial property of the treated surfaces. Previous studies also found that phase alumina nanoparticles result in poorer tribological performance when compared to the phase alumina particles. It is
159 hypothesized that the wear resistance mechanism of the nanoparticles is to alter the crystalline structure of the polymer. These changes affect the mechanical properties, which determine the size, morphology and number density of debris particles generated. The debris particles affect the formation and stability of the transfer film and the transfer film determines the tribological properties of the system. This investigation st udies the thermal, mechanical and tribological properties of 12.5 wt% jet-milled phase and phase alumina-PTFE nanocomposites to further explore this hypothesis. Figure 6-30 shows SEM images of the 12.5 wt% jet-milled alumina-PTFE powder ensembles. The phase particles used in this study (44 nm reported av erage particle size) are clearly smaller than the alpha phase alumina particles (80 nm repor ted average particle size), but the effects of particle size have previously been found to be small. Few small agglomerations were found in these powders and both dispersions are characterized by a large number of nanoparticles decorating the surface s of the PTFE particles. Th e same dispersion characteristics were observed for the jet-milled samples of the previous study which varied greatly from the agglomerated nature of the hand-mixed dispersion. Figure 6-30. Powder dispersi ons for 12.5 wt% alumina nanoparticles in PTFE: a) 40 nm phase alumina, b) 80 nm phase alumina.
160 Thermal Characterization of Powder Ensembles Thermal characterizations using DSC were conduct ed to explore the pote ntial effects of the nanoparticles on the mobility, melt, crystal nucleation and lamellae of PTFE. The results of these characterizations are shown in Figure 631. The neat PTFE began melting first, the phase alumina filled PTFE melted second and the phase alumina filled PTFE melted third. In general, the melt distributions of the filled samples were narrower. This may be an indication that as low order structures melt at low temperatures, the nanoparticles act to nucleate and recrystallize higher orde red structures synchronously. The recrystallization curve reflects the steeper melt curve, and the nanocomposite powde rs show lower crystallinity which likely reflects the lack of cont ribution of the 12.5 wt% of alumina filler. Figure 6-31. DSC heat flow pl otted versus temperature for heating and cooling of powder ensembles of unfilled PTFE, 12.5 wt% phase alumina nanoparticle in PTFE and12.5 wt% phase alumina in PTFE. The results of these thermal characterizati on experiments have been quantified and are plotted versus filler wt% in Figure 6-32 and are tabulated in Table 6-8. The first peak melt
161 temperature is plotted versus filler wt% in Fi gure 6-32a. The previous results for jet-milled powders are also included for the interpretation of trends. There is a clear trend of increasing melt temperature with increasing filler wt%. This is not due to a shift in th e curve, but rather to a reduction in the low temperature portion of the signal of neat PTFE. This may have to due with nucleation and recrystallization of the PTFE into higher order structures because of the nanofiller; determination of this effect will require modulated DSC. The effect occurs with both and phase nanofillers, but to a greater extent with the smaller phase alumina. The enthalpy of fusion for the first melt is plot ted versus filler wt% in Figure 6-32b. There is a clear trend of decreased crystallinity with incr eased filler. This decrease is consistent with the prepared loading of the inactive nanoparticles that are included in the initial mass measurement. The global crystallinity of the jetmilled PTFE is not expected to vary as a result of the nanoparticle inclusion. The jet-mill is an open system and based on previous thermal gravimetric analysis, it has long been thought that as many as 60% of the nanoparticle fines are preferentially lost during powder blending. Th ese results suggest that the loadings of the nanocomposites are equal to the loadings as prepared. The peak recrystallization temperature is pl otted versus filler wt% in Figure 6-32c. As with the first melt temperature, this is due to narr owing of the curve. It is unclear why the curve has become narrower, but it is pos sible that the nanoparticles prom ote an increase in the ordering of initially low order crystals. The enthalpy of recrystallization is plotted versus filler wt% in Figure 6-32d. Lower wt% nanocomposites show no change in crystallinity, but both 12.5 wt% samples showed a reduction in crystallinity of nearly 5%. Repeat samples of the virgin material were tested to determine variability and the result s showed a standard deviation of 0.3 J/g. The 12.5 wt% nanoparticle-PTFE powder ensembles had 4 J/g lower heats of fusion than the control
162 sample after recrystallization. These results su ggest that the PTFE powders recrystallize with lower crystallinity in the presence of the nanopart icles and in the absence of pressure. Figure 6-32. Quantified results of differential scanning calorimetry of a and phase alumina nanoparticles dispersed in PTFE plotted versus filler wt%: a) first peak melt temperature plotted versus filler wt%, b) first heat of fusion plotted versus filler wt%, c) peak crystallization temperature plotted versus filler wt%, d) heat of fusion for crystallization plotted versus filler wt%. Error bars represent the experimental uncertainty in each case.
163 Table 6-8. Results of DSC of jet-milled 12.5 wt% 80 nm phase and 44 nm phase aluminaPTFE powder ensembles.. Alumina phase loading 0 wt% 1 wt% 2 wt% 5 wt% 12.5 wt% 12.5 wt% Tm1 (C) 340.9 341.2 341.3 341.2 341.8 342.5 Hm1 (J/g) 72.7 72.2 71.6 67.9 64.0 66.5 U(Hm1) 1.0 1.0 1.0 1.0 0.9 0.9 Tm1 (C) 316.1 314.4 314.9 314.4 315.8 315.2 Hm1 (J/g) 26.6 30.3 29.4 29.0 23.7 23.0 Thermal Characterization of Nanocomposites The thermal characteristics of the compre ssion molded nanocomposites were studied to determine the role of the na noparticles under relevant compression molding conditions. The results from DSC measurements of neat PTFE, 12.5 wt% 80 nm phase alumina-PTFE and 12.5 wt% 44 nm phase alumina-PTFE nanocomposite are shown in Figure 6-33; the much larger melt peaks of the nanocomposite samples were unexpected consideri ng the opposite trend found from thermal characterization of the powders. A repeat sample of the neat PTFE was included in these measurements and showed nearly id entical behavior to the first sample. Both unfilled samples had a small high temperature melt peak near 350C suggesting that a small fraction of the material did not melt duri ng processing. The lack of any comparable high temperature melt peak for the nanocomposites indicat es that these samples were fully sintered. This is somewhat counterintuitive given the pow der DSC results which showed an increase in first melt temperature with nanoparticle loading. The heats of fusion were also significantly higher for the nanocomposites than for the neat samples contrary to the powder results which showed a small tendency to reduce crystallinity with nanoparticle loading. Following melting, the samples showed a comparable heat of fusion during recrystallization despite the relatively high loading of inactive filler.
164 Figure 6-33. DSC heat flow pl otted versus temperature for h eating and cooling of compression molded samples of unfilled PTFE (x2) and nanocomposites of 12.5 wt% and phase alumina-PTFE. The quantified results from DSC studies of the and phase alumina-PTFE nanocomposites are shown in Figure 6-34 and tabulated in Table 6-9. The first melt temperature is shown plotted versus filler wt % in Figure 6-34a. The outlying data points at 2 wt% were due to incomplete sintering of the samples. The other samples all show a slight tendency of increased melt temperature with in creased loading. The 12.5 wt% sample had the most dramatic increase in melt temperature due to a shif t of the melt curve; it is unclear whether the increased temperatures for phase samples are significant. The heat of fusion for the first melt is plotted versus filler wt% in Figure 6-34b. The unusually hi gh crystallinity of the 2wt% samples is due to incomplete sintering and theref ore can not be correlated with the rest of the data. For fully sintered samples, there is a clear trend of increased heat of fusion and crystallinity with filler loading. The phase alumina filled sample showed a greater increase
165 in crystallinity than did the phase alumina filled sample. This is consistent with the hypothesis that the nanoparticles act as nuclea tion sites for crystallization. The phase particles are smaller and therefore have a grea ter number of nucleation sites fo r a given loading. The recrystallization temperature is plotted versus filler wt% in Figure 6-34c. The phase nanocomposites did not show a dependence of recrystallization temperature on filler loading. The particles appear to have re duced the recrystallization temp erature by approximately 2C. Examination of 6-33 reveals that th ese particles caused a shift in th e recrystallization behavior to lower temperatures. The heat of fusion for recrystallization is plotted versus filler wt% in Figure 6-34d. Even after well controlled cooling conditions, the trend of increased crystallinity with increased nanoparticle loading; the crystallinity of the PTFE increased by 50% and 66% with 12.5 wt% and phase alumina, respectively. Table 6-9. Quantitative results of DSC for jet-milled 12.5 wt% 80 nm phase and 44 nm phase alumina-PTFE compression molded nanocomposites. Alumina phase loading 0 wt% 1 wt% 2 wt% 5 wt% 12.5 wt% 12.5 wt% 328.9 328.9 343.3 329.9 329.0 343.3 330.0 Tm1 (C) 328.9 329.7 329.7 329.5 329.5 330.9 32.5 35.7 63.0 43.0 38.2 62.7 40.2 Hm1 (J/g) 32.5 34.4 38.9 35.6 46.8 58.6 0.4 0.5 0.9 0.6 0.5 0.9 0.6 U(Hm1) 0.4 0.5 0.5 0.5 0.6 0.8 316.1 315.7 316.3 316.6 Tc1 (C) 316.1 315.5 316.4 316.3 315.6 314.4 29.3 32.4 31.9 35.3 Hc1 (J/g) 26.6 33.4 30.9 33.9 43.7 48.3 0.4 0.4 0.4 0.5 U(Hc1)0.4 0.4 0.4 0.4 0.6 0.7
166 Figure 6-34. Quantified DSC results plotted versus filler wt% from unfilled PTFE and nanocomposites of 12.5 wt% and phase alumina-PTFE: a) first peak melt temperature plotted versus filler wt%, b) first heat of fusion plotted versus filler wt%, c) peak crystallization temperature plotted versus filler wt%, d) heat of fusion for crystallization plotted versus filler wt%. Mechanical Characterization of Nanocomposites Mechanical testing of the 12.5 wt% and phase alumina-PTFE nanocomposites was carried out to investigate the potential roles of nanoparticle surface characteristics on strength, strain, toughness and fracture mechanisms of the polymer. Stress is plotted versus strain for these samples in Figure 6-35. The stress of the phase alumina filled sample increased with
167 strain consistent with an elastic modulus of ~330 MPa until failu re occurred at a stress of 5 MPa. The phase alumina filled sample had a modul us of ~140 MPa and an ultimate stress of approximately 1.4 MPa. Failure occurred at a st rain of only 1.5% for both samples. In both cases, failure was followed by a slight increase in stress and a slow loss of load carrying capacity until the samples became completely severed. Figure 6-35. Stress plotted versus engineering strain for 12.5 wt% and phase aluminaPTFE nanocomposites. Failure stress is plotted versus filler wt% in Fi gure 6-36a and failure st rain is plotted versus filler wt% in Figure 6-36b. The nanocomposites show an optimum in strength, strain and therefore, toughness at a nanoparticle loading of 2 wt% phase 80 nm alumina. Coincidentally, this corresponds to a tendency for incomplete sintering of the 2 wt% samples during compression molding. Previous modeling by Burris  s uggested that the volume fraction for complete monolayer coverage by the filler can be approximated as, *4 41fD D Eq. 6-3
168 where D* is Dfiller/Dmatrix. At monolayer coverage, the matr ix particles have very limited connectivity which likely limits mechanical perf ormance. Given 80nm filler and 3m matrix particles, monolayer coverage occurs at 10% by volume, or 20 wt%. Optimum mechanical properties occur here for ~10% monolayer coverage. Figure 6-36. Quantified results fr om mechanical tests of 12.5 wt% and phase aluminaPTFE nanocomposites: a) failure stress versus filler wt%, b) Failure strain versus filler wt%. Electron images of the fracture surfaces of 0, 1, 2, 5 and 12.5 wt% phase and 12.5 wt% phase alumina-PTFE nanocomposites are shown in Figure 6-37. The unfilled PTFE shows a corrugated surface structure. The majority of the load support and strain accommodation likely occurred via the protruding material on th e surface following mi crovoid nucleation and coalescence. The 1 wt% phase alumina filled sample has a relatively smooth and fibrillated surface profile. Domains of alumina rich material are identifiable as either bright inclusions or as dark regions where the inclusions have been removed. These domains are not affected by the applied stress and persist as whole entities within one of the halves of the fractured part. This sample was significantly stronger an d more elastic than the neat control sample and the entire cross section appears to have been uniformly stressed, stra ined and failed. The 2wt% phase
169 sample is far less smooth with cords of strained PTFE being evident and oriented across the surface. The alumina rich domains are clearly present. At 5 wt%, the surface morphology is clearly different with smaller, more distinct ropes of PTFE clearly decorating the fracture surface intermixed with loose fibrillated material and very smooth domains where brittle fracture likely occurred. The topography of the 12.5 wt% sample is very tortuous with comparatively large changes in elevation occurring across the surface. The bulk of the surface now appears to have a brittle fracture morphology with many thin fibrils protruding from the surface as a result of the highly localized stresses following failure of th e bulk of the surface. The domains are more difficult to distinguish due to the greater extent of the topographical cont rast. The 12.5 wt% phase alumina-PTFE sample has a very diffe rent fracture morphology than the 12.5 wt% phase alumina-PTFE sample despite their compar ably poor mechanical properties. The entire surface of this sample has a brittle fracture mor phology. Few very fine fibrils are also present on the surface, but they are too fi ne to distinguish at the scale s hown. Various other large-scale cracks on and within the fracture surface are clearly visible. The morphology of this sample apparently gives little resistance to crack initiation and propagation. Table 6-10. Results of mechanical testing of jet-milled 12.5 wt% 80 nm phase and 44 nm phase alumina-PTFE compression molded nanocomposites. Alumina phase loading 0 wt% 1 wt% 2 wt% 5 wt% 12.5 wt% 12.5 wt% (MPa) 7.7 13.3 15.2 12.0 5.0 1.4 Uc() 0.1 0.1 0.2 0.1 0.1 0.1 (%) 4.49 285 557 257 1.3 1.3 Uc() 0.01 4 7 3 0.1 0.1
170 Figure 6-37. SEM images of the fracture surf aces of alumina-PTFE nanocomposites: a) 0 wt% phase alumina-PTFE, b) 1 wt% phase alumina-PTFE, 2 wt% phase aluminaPTFE, 5 wt% phase alumina-PTFE, 12.5 wt% phase alumina-PTFE, 12.5 wt% phase alumina-PTFE.
171 Tribological Characterization of Nanocomposites Friction coefficient and wear volume loss are plotted versus sliding distance in Figures 638a and 6-38b for and phases of alumina, respectively. Quantitative results are given in Table 6-11. Stylus prof ilometry measurements of the transf er films after varying numbers of cycles are shown in Figure 6-39. Though SEM showed material tran sfer into surface scratches, the stylus profilometer could not distinguish transf er films from the counter faces for cycles 1 and 11. In a manner consistent with the lower wt% jet-milled samples of the previous study, the friction coefficient of the phase alumina filled sample was minimized during the transient period of sliding at = 0.13 after 20 m of sliding At steady state, the friction coefficient fluctuated between = 0.25 and = 0.30. The cause of the friction minimum and the relatively abrupt jump to a rather high fric tion coefficient are unclear. Surface profilometry of transfer film during steady state sliding shows very thin transfer morphol ogy with evidence of mild abrasion to the counterface after 1,111,111 sliding cycles. The extent of the abrasion is less at 12.5 wt% than it was at 1,2 or 5 wt% loading. This is counterintuitive based on a model including alumina as the only abrasive element. The lower abrasion at the higher loading must be due to the development of a more tenacious and protective transfer film. Figure 6-38. Friction coefficient and wear rate plotted versus sliding distance for a) jet-milled 12.5 wt% phase alumina-PTFE nanocomposites, b) jet-milled 12.5 wt% phase alumina-PTFE nanocomposites.
172 Figure 6-39. Stylus profilometric measurements of the surfaces of transfer films. Transfer film morphologies are plotted versus sliding cycle for alumina-PTFE nanocomposites with a) phase alumina filler, b) phase alumina filler. Burris and Sawyer  previously found a pow er law relationship between transfer film thickness and wear rate. The wear rate is plotted versus average transfer film thickness for these tests in Figure 6-41. This data reinforces the hypothesis that the wear rate is a strong function of the transfer film thickness, and therefore th e debris size and shape, and cohesion of the nanocomposite since these factors determine the film thickness.
173 Figure 6-40. Wear rate plotted versus sliding distance for jet-milled 12.5 wt% and phase alumina-PTFE nanocomposites. Figure 6-41. Wear rate plotte d versus maximum transfer film thickness for alumina-PTFE nanocomposites of various particle phase, size shape and loading.
174 Table 6-11. Results of tribological testing of hand-mixed and jet-milled 1wt%, 2wt% and 5 wt% 80 nm alumina-PTFE compression molded nanocomposites. The uncertainty in the friction coefficient measurements is less than Uc()=0.005. 12.5 wt% phase alumina 12.5 wt% phase alumina Cycles K U(k) () K U(k) () 1 7.2x10-4 4x10-4 N/A N/A 2.7x10-4 4x10-4 N/A N/A 11 2.0x10-4 4x10-5 0.212 0.003 3.9x10-4 4x10-5 0.202 0.002 111 3.3x10-5 5x10-6 0.170 0.030 1.1x10-4 6x10-6 0.204 0.001 1,111 4.3x10-7 4x10-7 0.205 0.033 6.3x10-5 2x10-6 0.215 0.006 11,111 5.1x10-7 8x10-8 0.236 0.034 1.3x10-6 4x10-7 0.179 0.032 111,111 1.6x10-7 1x10-8 0.287 0.010 2.5x10-5 7x10-7 0.169 0.010 1,111,111 6.9x10-8 4x10-9 0.270 0.013 N/A N/A N/A N/A
175 CHAPTER 7 DISCUSSION Polym er nanocomposite tribology is an exceed ingly difficult area in which to conduct fundamental research. Tribo-systems cont ain chemical, mechanic al and topographical interactions over a wide range of length and times scales, making tribology an inherently multidisciplinary area. The tribological inte rface is buried, inaccessible and dynamically evolving throughout the wear process. As a result, tribologists must often resort to forensic science to make inferences about the evolution of the system during sliding. In addition, polymer nanocomposite materials contain many co mplex materials science challenges. The study of the molecular, lamellar a nd crystalline structures and or ganizations of neat PTFE is arduous and the additional complexities introduced into this system with the inclusion of nanoparticles are vast. Myriad hyp otheses for wear resistance mechanisms of such systems have been proposed. The results from these studies he lp clarify the state of our understanding of the wear resistance mechanisms of PTFE na nocomposites by refuting some hypotheses and reinforcing others. Fibrillation The first studies in this work confirmed the hypothesis that the jet-mill dispersion technique is more energetic and destructive to the PTFE than other co mmon blending techniques by a statistically significant margin. Jet-milling not only disbands the agglomerations of PTFE, but it damages individual particles, reducing average size by about 75%. Contrary to a hypothesis that the jet-mill fibrilla ted the PTFE and led to stabilizat ion of a fibrillated structure via the nanoparticles during compression molding, none of the techniques were found to result in fibrillation of the PTFE. Rather, the PTFE appear ed to have been damage d in a brittle fracture mode and did not contain any more fibrillated materi al than the virgin material did. It is now
176 clear that nanoparticle stabilization of fibrils pr oduced during jet-milling is not a wear resistance mechanism. Crystalline Morphology and Crystallinity Preliminary investigations of the heat-treating effects on th e tribology of an originally wear resistant PTFE nanocomposite had a significan t impact on the way we think about these nanocomposite systems. These studies showed th at wear resistance coul d vary by orders of magnitude simply due to the crystalline structur e of the PTFE for nominally constant particle material, shape, size, loading, and dispersion. Before the heat treatment, the wear resistant nanocomposite showed thermal characteristics of the virgin PTFE morphology while the thermal characteristics of the high-wear heat-treated sample reflected melt processed PTFE. Based on these results, it was hypothesized that the role of the nanoparticles is to stabilize the virgin morphology during processing. Virgin crystals are highly ordered and were thought to promote fibrillation under stress. It was therefore conjectured that unfilled PTFE with the virgin morphology should be wear resistant. Samples were sintered at temperat ures that should melt and sinter the lower order particle boundaries without affecting the hi ghly ordered particle bulks. DSC showed that the processing led to a dual PTFE morphology, the degree of which varied with hold temperature, but none of the samp les had unique wear resistance. Thus, the nanoparticles do provide benefits aside from th e simple stabilization of the virgin PTFE morphology. Several nanocomposite samples in this study di d have the thermal signature of the virgin morphology, but this feature did not correlate with wear resistance. Rather, wear resistance correlated strongly with both alum ina loading and phase. The crystallinity of the nanocomposite samples increased dramatically with increas ed nanoparticle loadi ng suggesting that the
177 nanoparticles do in fact nucleate and promot e crystallization. However, while both phase alumina and phase alumina filler led to a similar increase in crystallinity, the samples had very different wear performance. Thus, the crystallinity did not universally correlate with wear resistance. There are many factor s related to crystalline struct ure and morphology that cannot be deciphered from these DSC measurements, includi ng lamellar size, thickness, organization and homogeneity. Such factors may play domina nt roles in the microstructure, deformation mechanisms and wear behavior of the PTFE. The AFM images shown in Figure 3-18 provide possible evidence of such a mechanism but clarif ication of these factors will likely require further studies using lamellae-scale probing techniques. Mechanical Properties It has been long hypothesized th at the most effective nanopar ticles are those that promote toughness and high strains to failure. While th e low wt% nanocomposites were found to have higher strength, two orders of magnitude highe r elongation to failure, and three orders of magnitude higher wear resistance than neat PTFE, the very wear resistant 12.5 wt% phase alumina filled sample had diminished strength and elongation to failure compared to the neat PTFE. In addition, while the mechanical properties of the 12.5 wt% and phase alumina nanocomposites were comparably poor, their wear resistances differed by orders of magnitude. The differing microstruc tural effects of the and phase nanoparticles were clearly illustrated by their fracture surfaces. The phase alumina produced a relatively smooth brittle fracture surface, while the phase nanoparticles led to a very fibri llated and tortuous fracture surface. Dispersion Nanoparticle dispersion is one of the most discussed but l east characterized factors in nanocomposite research. This is largely due to the difficulty in achieving good dispersion, the
178 difficulty in observing nanopartic les and characterizing dispersi on, and the idealization of dispersion as uniform in most mental models. In this study, the extremes of dispersion were studied in an attempt to more clearly define the role of dispersion in these unusual PTFE-based systems. SEM observation of hand-mixed powders showed that the nanoparticles were poorly dispersed and mostly resided w ithin agglomerates; very few a gglomerations were found within the jet-milled powders. Tribological testing of both sets of samples showed that hand mixed samples had comparable low wear rates to jet-milled samples. In addition, jet-milled phase alumina and phase alumina filled samples had very similar powder dispersions but very different tribological characteristics These results also suggest a secondary role of dispersion to the more primary effects of the filler. Figure 7-1. Optical images of polished sections of 12.5 wt% a) and b) phase aluminaPTFE nanocomposites. One constant between the effective phase alumina filled samples was the presence of ~520 m domains; these domains were not observed in the high wear phase alumina filled samples. Optical images of polished sections of these samples are shown in Figure 7-1. Raman
179 spectroscopy revealed that these regions were alumina rich. Fracture and SEM observations of individual domains suggest th at these domains are nanopartic le rich PTFE rather than nanoparticle agglomerates. It remains unclear what these domains are, why they appear within phase alumina filled systems and not within the phase alumina-PTFE nanocomposites. It is hypothesized that the su rface properties of the phase alumina promotes changes in the crystalline structure that do not occur with the phase alumina. The opacity of these regions is hypothesized to be the result of thes e crystalline change s to the PTFE. Crack Arrestment, Debris Generation and Transfer In 1984, Bahadur and Tabor hypothesized that the ro le of the fillers in PTFE, in general, is to regulate the size and shape of the debris . After numerous studies and hypotheses regarding the wear resistance mechanisms of fillers in PTFE, the observation that small wear debris accompany wear resistance of PTFE-bas ed materials is universally noted, making it one of, if not the only unrefuted hypothesized wear resistan ce mechanism in these materials. In 2005, following observations of unprecedented wear resistance, optimum surface texture, and strong correlation between transfer film thickness and wear ra te, Burris and Sawyer extended this hypothesis . It was suggested that the reduc tion of debris size by successful fillers not only reduces wear directly by reducing the volume of each particle, but it also facilitates engagement of the small debris into the counterface ther eby enabling the development of tenacious and protective transfer films. These well-adhered, wear resistant, and protective films have been found to be a necessary condition for low wear (<10-6 mm3/Nm) sliding. The results collected in these studies reaffirm this hypothesis. Blanchet and Han [67, 68] showed that effec tive wear-resistant micron-scale fillers tend to accumulate at the surface resulting in a continuously decreasing wear rate that is followed by low
180 wear at steady state. In a manner consistent with the filler accumulation of microcomposites, wear rate was found to decrea se with increased sliding distance for the effective ( phase alumina) nanocomposite systems in this study. W ear rate is plotted vers us sliding distance for the 12.5 wt% a phase alumina-PTFE nanocomposite in Figure 7-2. Backscattered electron images of wear su rfaces of various phase alumina-PTFE nanocomposites are shown connected to their corresponding wear rates for illustration of the evolution of the wear surface. The alumina rich domains were never observed on unwo rn, machined surfaces. Initial wear rates are similar to those found for unfilled PTFE at these te st conditions and are likely driven by asperity plowing of the nanocomposite. As sliding continues, domains begi n to populate th e wear surface and wear rates decrease. The dom ains reside within or on top of material that appears to be equivalent to the bulk. Optical microscopy of b) revealed that many domains were deposited over previously worn grooves on the surface sugges ting that they were first removed as debris and then back-transferred to the pin. After sufficient sliding distance, low wear rates (<10-6 mm3/Nm) accompany the presence of a coherent co ating that populates the remaining regions. Optically, this coating has a brown tint and the material has been identified as de-fluorinated and possibly conjugated PTFE 
181 Figure 7-2. Wear rate versus sliding di stance for jet-milled a phase alumina-PTFE nanocomposites with backscattered electron im ages of wear surfaces at corresponding wear rates: a) 2 wt% jet-milled (unworn), b) 1 wt% jet-milled (worn), c) 1 wt% handmixed (worn) and d) 2 wt% jet-milled (worn). This coating is referred to as a running film and is thought to provi de a protective barrier against wear. The achievement of low wear st eady state conditions likel y coincides with the initiation of this film. These running films are cracked with the orientation of the cracks running parallel and perpendicular to the sliding direction. The appearan ce of the cracking is consistent with the notion that the films are harder and mo re brittle than the bulk, and the cracks likely formed during local asperity contac ts. It is unclear if the brittl e properties are due to degradation
182 from frictional energy, or if brittle sections w ithin the bulk are preferentially fractured and transferred. Interestingly, many of the cracks begin and end at the film/domain interface but never penetrate the domains; it is unclear if the domains serve to arrest the cracks or if they initiate the cracks due to a weak interface. The worn 2 wt% jet-milled sample was retested against a fresh counterface to determine if the transfer film itself was wear resistant. The results of the original and follow-up tests are shown in Figure 7-3. Originally, the 2 wt% jet-milled sample lost nearly 20 mm3 of volume and required more than 5 km of sliding distance befo re achieving low-wear, steady-state sliding. The 5 wt% sample had a much shorter transient peri od. Despite having the same bulk properties, the presence of the running film immediately made the 2 wt% resistant to wear against an unprotected counterface. Figure 7-3. Worn volume versus sliding distance for the 2 and 5 wt% phase alumina nanocomposites. The 2 wt% nanocomposite wa s tested against a fresh counterface to test the hypothesis that the running films provide increased wear resistance in the absence of pre-existing transfer films.
183 Wear rate is Figure 7-4 plotted versus sliding distance for jet-milled phase aluminaPTFE nanocomposites. In each case, the wear ra te decreased with increased sliding distance consistent with accumulation of a wear resistant phase at the surface. In addition, the rate of decrease, increased with increasing alumina wt%, i.e. shorter sliding distance were required for higher wt% samples to achieve low wear slid ing (defined as a wear rate less than 10-6 mm3/Nm). The optical images of polished samples of 1, 2 and 5 wt% alumina (Figure 6-22) show that the number density of the bright domains correlates well with the nanoparticle loading. The strong decrease in the transient period of sliding with increased filler wt% and domain density suggests that the domain concentration in the bulk dictates the initial c ondition for the debris regulation process, the rate of accumulation and the initia tion of the coherent running film which drives wear resistance. Figure 7-4. Wear rate versus sliding di stance for jet-milled a phase alumina-PTFE nanocomposites.
184 Despite hypotheses that PTFE nanocomposites would be nonabrasive, the SEM and stylus images in Figures 6-28 and 6-29 show clear signs of counterface abrasion to relatively soft 304 stainless steel counterfaces. The counterface abrasion depth was measured using stylus profilometry and the abrasion rate s are plotted versus filler wt % in Figure 7-5. There is a counterintuitive trend of decrea sed counterface abrasion with in creased filler loading for jetmilled samples. This is most likely related to the number of passes by the abrasive pin before the protective surface films were formed. The 12. 5 wt% filled system evolved quickly due to a large initial concentration of th e crack arresting domains produci ng reduced particle size. The hand mixed samples were significantly more abrasive than the jet-milled samples despite evolving more quickly in general. This is fu rther evidence that the nanoparticles were highly agglomerated in these samples. These large abrasive aggregates likely tear easily through sections of the transfer film and counterface leaving fresh metal surfaces to further damage the pin. As a result, steady state wear rates of hand-mixed samples were 2-3 times higher than those of the jet-milled samples.
185 Figure 7-5. Abrasion rate to the counterface plotted versus filler wt%. Rates are calculated as an average value using post-test stylus profilometry measurements. The results and observations collected in th eses studies suggest the hypothesized wear resistance model illustrated schematically in Figure 7-6. It has been well documented that PTFE transfers immediately after contac ting foreign surfaces. At first contact of these tribo-systems, the asperity level interactions l ead to instant transfer of molecu lar-scale PTFE transfer films and potentially, to local cracking of the material. The exact cause of damage localization in these systems remains unclear, but nanoparticle alterati on of the lamellar struct ure of the PTFE may contribute by enabling extensive fibrillation around the crack. The alumina rich domains may also localize damage by interfering with crac k propagation. The compartmentalization of damage results in reduced debris size, engageme nt of debris into the counterface roughness and onto the worn pin surface. Because the domains are responsible for crack arrestment, the extent of the damage during the transien t period decreases as the density of the alumina rich domains
186 near the surface increases. These domains also app ear to transfer first, acting to initiate the transfer and running films. As ma terial is continuously worn, more of the domains are liberated. For a given amount of wear, the probability of lib erating a domain for transfer and film initiation increases as filler wt% increases. Thus, the volume of wear needed to initiate the films decreases with increased filler wt%. As the wear surface becomes rich in the domains, the composite wear debris, which likely consists of a nanoparticle altered, highly fi brillated and tough form of PTFE, populates the space in between the domains. Past XPS studies with these wear resistant running films have shown peaks consistent with PT FE defluorination, conjugation, bonding of carbon radicals with environmental speci es and potentially, cross-linking [ 61]. Other investigators have found that such degraded forms of PTFE are brittle [69-79]. These surfaces can become so wear resistant that the mechanical energy absorbed by this interface initiates degradation of even highly chemically resistant materials like PTFE During steady state, the dominant wear mechanism appears to be degradation, embrittlement and fracture of small portions of the running film.
187 Figure 7-6. Hypothesized model of the wear of effective nanocomposites. a) Damage is localized around the asperity contact resulting in small debris size that facilitates transfer film formation. b) The transfer film continues to develop as the pin surface accumulates worn domains. c) The accumula tion of domains at th e surface initiates the formation of the fibrillated running film and at steady state low wear results from interfacial sliding of protec tive transfer and r unning films. d) Following steady state sliding, separation reveals w ear resistant films populated with alumina rich domains and degraded PTFE. Closing Remarks The wear resistance mechanisms of PTFE remain unclear. The matrix itself is complex and poorly understood, the dispersions are difficult to characteri ze, the nanoscale particles are inherently difficult to characterize, and tribology is itself one of the more complex sciences in
188 which to conduct fundamental studies. However, th ese studies have provid ed valuable insights into the validity of previous hypotheses regarding dispersion, filler material, matrix crystallinity and morphology, and toughness. A dominant wear resistance mechanism appears to be the compartmentalization of damage to small regions n ear the real area of c ontact. If this hypothesis is accepted as the primary wear reduction mechanism of these materials, two questions naturally come to mind: 1) by what mechanism is debris size regulated and 2) how do the nanoparticles initiate this mechanism. SEM observations ha ve shown that cracks do not propagate through the alumina rich domains. The weak interface be tween the domains and the composite material likely initiates and deflects these cracks. The ma terial science dictating the formation of these domains remains an open question. It is very interesting that of two phase s of the same material, one promotes this mechanism while the other do es not. The current hypothesis is that the nanoparticle surfaces may promote a finer lamella r structure that dominates the deformation mechanisms of the polymer crystals. Perhaps most importantly, these studies have provided fruitful directions for future studies. Future studies must include nanoscale characterization and in-situ observation of the pin and the transfer film over time and over various length scales using electron a nd white-light optics, spectroscopy and interferometry. Tribometer designs for such studies are currently underway and should provide valuable insights into the nature of the transient wear process, the dominant factors determining the friction coefficients, the causes of counterface abrasion and transfer film formation, evoluti on and disruption.
189 CHAPTER 8 CONCLUSIONS 1. Nanoparticles were found to have dramatic effects on a number of physical properties of PTFE including, crystallinity, strength, elongation, toughness, wear resistance and friction coefficient. Crystallinity was increase d by more than 50% with 12.5 wt% alumina nanoparticles. With 1 wt% alumina nanopartic les, ultimate strength was improved by nearly a factor of two and elongation to failure wa s improved by 100X. At 12.5 wt% the samples became very brittle and had lower strength and elongation than neat PTFE. Wear resistance at 2 wt% was improved by over 1,000X. 2. The changes in crystallinity and toughness did not correlate we ll with the wear resistance of the sample. A very brittle, high crystall inity 12.5 wt% sample had comparable wear performance to a very tough, low crystallinity 2 wt% sample. 3. Dispersion and agglomeration of the nanopartic les had little effect on the wear performance of the nanocomposites. The agglomerated handmixed samples had smaller, less severe wear transients but were more abrasive to the c ounterface and had higher steady-state wear rates by a factor of 2-4. 4. Alumina phase dominated the wear resistan ce of the nanocomposite. Dispersion, thermal and mechanical properties were similar, but wear rates were well over 100X higher for the round alumina filled sample than for the irregular phase alumina filled sample.
190 LIST OF REFERENCES 1. Miyoshi, K. (1999), "Consi derations in Vacuum Tribol ogy (Adhesion, Friction, Wear, and Solid Lubrication in Vacuum)," Tribology International, 32, 11, pp 605-616. 2. Stachowiak, G. and Batchelor, A. (2001), Engineering Tribology. ButterworthHeinemann. 3. Schmitz, T., et al.(2005), "The Difficulty of Measuring Low Friction: Uncertainty Analysis for Friction Coefficient Measurements, "Journal of Tribology -Transactions of the ASME, 127, 3, pp 673-678. 4. Schmitz, T., et al.(2004), "Wear-Rate Uncertainty Analysis, "Journal of TribologyTransactions of the ASME, 126, 4, pp 802-808. 5. Murari, A. and Barzon, A. (2003), "Compari son of New PEEK (R) Seals with Traditional Helicoflex for Ultra High Vacuum Applications," Vacuum, 72, 3, pp 327-334. 6. Murari, A., Vinante, C., and Monari, M. (2002), "Comparison of PEEK and Vespel (R) SP1 Characteristics as Vacuum S eals for Fusion Applications," Vacuum, 65, 2, pp 137145. 7. Thomas, D. and Woolfenden, S. (2006), "Fur ther Developments of Biaxially Oriented PTFE Gasket Materials," Sealing Technology, 2006, 2, pp 11-13. 8. Burris, D. and Sawyer, W. (2005), "T ribological Sensitivity of PTFE/Alumina Nanocomposites to a Range of Traditional Surface Finishes," Tribology Transactions, 48, 2, pp 147-153. 9. Burris, D. and Sawyer, W. (2006), "A Low Friction and Ultra Low Wear Rate PEEK/PTFE Composite," Wear, 261, 3-4, pp 410-418. 10. Chen, W., et al.(2003), "Tribological Behavior of Carbon-Nanotube-Filled PTFE Composites," Tribology Letters, 15, 3, pp 275-278. 11. Li, F., et al.(2001), "The Friction and Wear Charac teristics of Nanometer ZnO Filled Polytetrafluoroethylene," Wear, 249, 10-11, pp 877-882. 12. Lu, Z. and Friedrich, K. (1995), "On S liding Friction and Wear of PEEK and Its Composites," Wear, 181, pp 624-631. 13. Mccook, N., et al.(2005), "Wear Resistant Solid Lubri cant Coating Made from PTFE and Epoxy," Tribology Letters, 18, 1, pp 119-124.
191 14. Menzel, B. and Blanchet, T. (2002), "Effect of Particle Size an d Volume Fraction of Irradiated FEP Filler on the Transfer Wear of PTFE," Lubrication Engineering, 58, 9, pp 29-35. 15. Palabiyik, M. and Bahadur S. (2000), "Mechanical a nd Tribological Properties of Polyamide 6 and High Density Polyethyl ene Polyblends w ith and without Compatibilizer," Wear, 246, 1-2, pp 149-158. 16. Sawyer, W., et al.(2003), "A Study on the Friction and Wear Behavior of PTFE Filled with Alumina Nanoparticles," Wear, 254, 5-6, pp 573-580. 17. Wang, Q., et al.(1996), "An Investigation of the Friction and Wear Properties of Nanometer Si3N4 Filled PEEK," Wear, 196, 1-2, pp 82-86. 18. Wang, Q., et al.(2000), "The Friction and Wear Characteristics of Nanometer Sic and Polytetrafluoroethylene Filled Polyetheretherketone," Wear, 243, 1-2, pp 140-146. 19. Burris, D. and Sawyer, W. (2006), "Imp roved Wear Resistance in Alumina-PTFE Nanocomposites with Irregular Shaped Nanoparticles," Wear, 260, 7-8, pp 915-918. 20. Bahadur, S. (2000), "The Development of Transfer Layers and Their Role in Polymer Tribology," Wear, 245, 1-2, pp 92-99. 21. Friedrich, K., Lu, Z., and Hager, A. (1995), "Recent Advances in Polymer Composites' Tribology," Wear, 190, 2, pp 139-144. 22. Zhang, S. (1998), "State-ofthe-Art of Polymer Tribology," Tribology International, 31, 1-3, pp 49-60. 23. Siegel, R., et al.(2001), "Mechanical Behavior of Polymer and Ceramic Matrix Nanocomposites," Scripta Materialia, 44, 8-9, pp 2061-2064. 24. Ng, C., Schadler, L., and Siegel, R. (1999) "Synthesis and Mechanical Properties of Tio2-Epoxy Nanocomposites," Nanostructured Materials, 12, 1-4, pp 507-510. 25. Eitan, A., et al.(2006), "Reinforcement Mechanisms in MWCNT-Filled Polycarbonate," Composites Science and Technology, 66, 9, pp 1162-1173. 26. Agag, T., Koga, T., and Takeichi, T. (2001), "Studies on Thermal and Mechanical Properties of Polyimide-Clay Nanocomposites," Polymer, 42, 8, pp 3399-3408. 27. Ash, B., Schadler, L., and Siegel, R. (2002), "Glass Tran sition Behavior of Alumina/Polymethylmethacrylate Nanocomposites," Materials Letters, 55, 1-2, pp 83-87. 28. Ratna, D., et al. (2006), "Poly(Ethylene Oxide)/Clay Nanocomposite: Thermomechanical Properties and Morphology," Polymer, 47, 11, pp 4068-4074.
192 29. Yasmin, A., et al.(2006), "Mechanical and Ther mal Behavior of Clay/Epoxy Nanocomposites," Composites science and Technology, 66, pp 2415-2422. 30. Bhimaraj, P., et al.(2005), "Effect of Matrix Mor phology on the Wear and Friction Behavior of Alumina Nanoparticle/Poly( Ethylene) Terephthalate Composites," Wear, 258, 9, pp 1437-1443. 31. Schwartz, C. and Bahadur, S. (2000), "Studies on the Tribological Behavior and Transfer Film-Counterface Bond Strength for Polyphe nylene Sulfide Filled with Nanoscale Alumina Particles," Wear, 237, 2, pp 261-273. 32. Wang, O., Xue, Q., and Shen, W. (1997), "The Friction and Wear Properties of Nanometre SiO2 Filled Polyetheretherketone," Tribology International, 30, 3, pp 193197. 33. Wang, Q., et al. (1996), "The Effect of Par ticle Size of Nanometer ZrO2 on the Tribological Beha viour of PEEK," Wear, 198, 1-2, pp 216-219. 34. Xue, Q. and Wang, Q. (1997), "Wear Mechanisms of Polyetheretherketone Composites Filled with Various Kinds of SiC," Wear, 213, 1-2, pp 54-58. 35. McLaren, K. and Tabor, D. (1963), "Visco-Elastic Properties a nd Friction of Solids Friction of Polymers Influence of Speed and Temperature," Nature, 197, 487, pp 856. 36. Makinson, K. and Tabor, D. (1964), "Friction + Transfer of Polytetrafluoroethylene," Nature, 201, 491, pp 464. 37. Pooley, C. and Tabor, D. (1972), "Friction and Molecular Structure Behavior of Some Thermoplastics," Proceedings of the Royal Society of London Series A-Mathematical and Physical Sciences, 329, 1578, pp 251. 38. Tanaka, K., Uchiyama, Y., and Toyooka S. (1973), "Mechanism of Wear of Polytetrafluoroethylene," Wear, 23, 2, pp 153-172. 39. Blanchet, T. and Kennedy, F. (1992), "Sliding Wear Mechanism of Polytetrafluoroethylene (PTFE) and PTFE Composites," Wear, 153, 1, pp 229-243. 40. Suh, N. (1973), "Delamination Theory of Wear," Wear, 25, 1, pp 111-124. 41. Bahadur, S. and Gong, D. (1992), "The Ac tion of Fillers in th e Modification of the Tribological Behavior of Polymers," Wear, 158, 1-2, pp 41-59. 42. Burroughs, B., Kim, J., and Blanchet, T. (1999), "Boric Acid Se lf-Lubrication of B2O3Filled Polymer Composites," Tribology Transactions, 42, 3, pp 592-600.
193 43. Li, F., et al.(2000), "The Tribological Behaviors of Copper-Coated Graphite Filled PTFE Composites," Wear, 237, 1, pp 33-38. 44. Lancaster.Jk (1972), "Polymer-Based Bear ing Materials Role of Fillers and Fiber Reinforcement in Wear," Wear, 22, 3, pp 412. 45. Sung, N. and Suh, N. (1979), "Effect of Fi ber Orientation on Fricti on and Wear of Fiber Reinforced Polymeric Composites," Wear, 53, 1, pp 129-141. 46. Tanaka, K. and Kawakami, S. (1982), "Effect of Various Fillers on the Friction and Wear of Polytetrafluoroethylene-Based Composites," Wear, 79, 2, pp 221-234. 47. Briscoe, B., Pogosian, A., and Tabor, D. (1974), "Friction and Wear of High-Density Polythene Action of Lead Ox ide and Copper Oxide Fillers," Wear, 27, 1, pp 19-34. 48. Gong, D., Xue, Q., and Wang, H. (1991), "ESCA Study on Tribochemi cal Characteristics of Filled PTFE," Wear, 148, 1, pp 161-169. 49. Blanchet, T., Kennedy, F., and Jayne, D. (1993), "XPS Analysis of the Effect of Fillers on PTFE Transfer Film Devel opment in Sliding Contacts," Tribology Transactions, 36, 4, pp 535-544. 50. Bahadur, S. and Tabor, D. (1984), "The Wear of Filled Polytetrafluoroethylene," Wear, 98, 1-3, pp 1-13. 51. Burris, D. (2006), "Wear Resistance Mechan isms in Polytetrafluoroethylene (PTFE) Based Tribological Nanocomposites," Master's Thesis, University of Florida, Gainesville, Florida, 52. McElwain, S. (2006), "Wear Resistant PT FE Composites Via Nano-Scale Particles," master's thesis, Rensselaer Polytechnic Institute, Troy, New York. 53. Reynaud, E., et al.(2001), "Nanofillers in Poly meric Matrix: A Study on Silica Reinforced PA6," Polymer, 42, 21, pp 8759-8768. 54. Sevostianov, I. and Kachanov, M. (2007), "Eff ect of Interphase Layers on the Overall Elastic and Conductive Properties of Matrix Composites. Applications to Nanosize Inclusion," International Journal of Solids and Structures, 44, 3-4, pp 1304-1315. 55. Wagner, D. and Vaia, R. (2004), "Na nocomposites: Issues at the Interface," Materials Today, pp 38-42. 56. Koerner, H., et al.(2005), "Deformation-Morphology Co rrelations in Electrically Conductive Carbon Nanotube Thermoplastic Polyurethane Nanocomposites," Polymer, 46, 12, pp 4405-4420.
194 57. He, P., et al.(2006), "Surface Modification and Ultras onication Effect on the Mechanical Properties of Carbon Nanofiber /Polycarbonate Composites," Composites Part A-Applied Science and Manufacturing, 37, 9, pp 1270-1275. 58. Maiti, M. and Bhowmick, A. (2006), "New Insights into Rubber-Clay Nanocomposites by AFM Imaging," Polymer, 47, 17, pp 6156-6166. 59. Sternstein, S., et al.(2005), "Reinforcement and Nonlinear Viscoelastic ity of Polymer Melts Containing Mixtures of Nanofillers," Rubber Chemistry and Technology, 78, 2, pp 258-270. 60. Zhu, A. and Sternstein, S. (2003), "Nonlinear Viscoelasticity of Nanofilled Polymers: Interfaces, Chain Statistics and Properties Recovery Kinetics," Composites Science and Technology, 63, 8, pp 1113-1126. 61. Burris, D.L., et al., Polytetrafluoroethylene Matrix Nanocomposites for Tribological Applications, in Tribology of Polymeric Nanocomposites, K. Friedrich, Editor. 2008. 62. Beamson, G., et al.(1996), "Characterization of PTFE on Silicon Wafer Tribological Transfer Films by XPS, Imaging XPS and AFM," Surface and Interface Analysis, 24, 3, pp 204. 63. Breiby, D., et al.(2005), "Structural Surprises in Friction-Deposited Films of Poly(Tetrafluoroethylene)," Macromolecules, 38, 6, pp 2383-2390. 64. Wheeler, D. (1981), "The Transfer of Polytetrafluoroethylene Studied by X-Ray Photoelectron-Spectroscopy," Wear, 66, 3, pp 355-365. 65. Wittmann, J. and Smith, P. (1991), "Highly Oriented Thin-Films of Poly(Tetrafluoroethylene) as a Substrat e for Oriented Grow th of Materials," Nature, 352, 6334, pp 414-417. 66. Gong, D., et al.(1990), "Effect of Tribochemical R eaction of Polytetrafluoroethylene Transferred Film with Substr ates on Its Wear Behavior," Wear, 137, 2, pp 267-273. 67. Blanchet, T. (1995), "A Model for Poly mer Composite Wear Behavior Including Preferential Load Support and Surface Accumulation of Filler Particulates," Tribology Transactions, 38, 4, pp 821-828. 68. Han, S. and Blanchet, T. (1997), "Experimental Evaluation of a Steady-State Model for the Wear of Particle-Filled Polymer Composite Materials," Journal of TribologyTransactions of the ASME, 119, 4, pp 694-699. 69. Abdou, S. and Mohamed, R. (2002), "Charact erization of Structural Modifications in Poly-Tetra-Fluoroethylene Induced by Electron Beam Irradiation," Journal of Physics and Chemistry of Solids, 63, 3, pp 393-398.
195 70. Briscoe, B., Mahgerefteh, H., and Suga, S. (2003), "The Effect of Gamma Irradiation on the Pressure Dependence of the Room Temperature Transition in PTFE," Polymer, 44, 3, pp 783-791. 71. Fayolle, B., Audouin, L., and Verdu, J. (2003), "Radiation Induced Embrittlement of PTFE," Polymer, 44, 9, pp 2773-2780. 72. Harling, O., Kohse, G., and Riley, K. (2002), "Irradiation Performance of Polytetrafluoroethylene (Teflon ) in a Mixed Fast Neutron and Gamma Radiation Field," Journal of Nuclear Materials, 304, 1, pp 83-85. 73. Zhao, X., et al.(2005), "An Experimental Study of Low Earth Orbit Atomic Oxygen and Ultraviolet Radiation Effects on a Spacecraf t Material Polytetrafluoroethylene," Polymer Degradation and Stability, 88, 2, pp 275-285. 74. Lappan, U., et al.(2002), "Number-Average Molecular Weight of Radiation-Degraded Poly(Tetrafluoroethylene). An End Group Analysis Based on Solid-State NMR and IR Spectroscopy," Polymer, 43, 16, pp 4325-4330. 75. Lappan, U., et al.(2001), "Radiation-Induced Bran ching and Crosslinking of Poly(Tetrafluoroethylene) (PTFE)," Nuclear Instruments and Methods in Physics Research Section B-Beam Intera ctions with Materials and Atoms, 185, pp 178-183. 76. Oshima, A., et al.(2001), "Chemical Structure and P hysical Properties of RadiationInduced Crosslinking of Po lytetrafluoroethylene," Radiation Physics and Chemistry, 62, 1, pp 39-45. 77. Katoh, E., et al.(1999), "Evidence for Radiati on Induced Crosslinking in Polytetrafluoroethylene by Means of High-Resolution Solid-State F-19 High-Speed MAS NMR," Radiation Physics and Chemistry, 54, 2, pp 165-171. 78. Fuchs, B. and Scheler, U. (2000), "Bra nching and Cross-Linking in Radiation-Modified Poly(Tetrafluoroethylene): A Solid-State NMR Investigation," Macromolecules, 33, 1, pp 120-124. 79. Blanchet, T. and Peng, Y. (1996), "Wear-Resistant Polytetrafluor oethylene Via Electron Irradiation," Lubrication Engineering, 52, 6, pp 489-495. 80. Blanchet, T. and Peng, Y. (1998), "Wear Resistant Irradiated FEP Unirradiated PTFE Composites," Wear, 214, 2, pp 186-191. 81. Yang, K., et al.(2006), "Morphology and Mechanical Properties of Polypropylene/Calcium Carbonate Nanocomposites," Materials Letters, 60, 6, pp 805809.
196 82. Kojima, Y., et al.(1993), "Mechanical-Properties of Nylon 6-Clay Hybrid," Journal of Materials Research, 8, 5, pp 1185-1189. 83. Xiao, Z., et al.(2006), "Probing the Use of Smal l-Angle Light Scattering for Characterizing Structure of Titanium Dioxide/Low-Density Polyethylene Nanocomposites," Journal of Polymer Science Part B-Polymer Physics, 44, 7, pp 10841095. 84. Yang, H., et al.(2007), "Crystal Growth in Alumin a/Poly(Ethylene Terephthalate) Nanocomposite Films," Journal of Polymer Science, Part B: Polymer Physics, 45, 7, pp 747-757. 85. Kuchta, F., et al.(1999), "Materials with Improved Properties from Polymer-CeramicNanocomposites" Mat Res Soc Symp Proc, 576, pp 363-368. 86. Petrovicova, E., et al.(2000), "Nylon 11/Silica Nanocompos ite Coatings Applied by the HVOF Process. I. Micros tructure and Morphology," Journal of Applied Polymer Science, 77, 8, pp 1684-1699. 87. Bunn, C. and Howells, E. (1954), "Structures of Molecules and Crystals of Fluorocarbons," Nature, 174, 4429, pp 549-551. 88. Clark, E. (1999), "The Molecular Conforma tions of Polytetrafluoroethylene: Forms II and IV," Polymer, 40, 16, pp 4659-4665. 89. Farmer, B. and Eby, R. (1981), "Energy Calc ulations of the Crysta l-Structure of the LowTemperature Phase (II) of Polytetrafluoroethylene," Polymer, 22, 11, pp 1487-1495. 90. Rigby, H. and Bunn, C. (1949), "A Room-Temperature Transition in Polytetrafluoroethylene," Nature, 164, 4170, pp 583-583. 91. Weeks, J., Clark, E., and Eby, R. (1981), "Crystal-Structure of the Low-Temperature Phase(II) of Polytetrafluoroethylene," Polymer, 22, 11, pp 1480-1486. 92. Weeks, J., Eby, R., and Clark, E. (1981), "D isorder in the CrystalStructures of Phase-1 and Phase-2 of Copolymers of Tetraf luoroethylene and Hexafluoropropylene," Polymer, 22, 11, pp 1496-1499. 93. Kimmig, M., Strobl, G., and Stuhn, B. (1994), "Chain Reorientation in Poly(Tetrafluoroethylene) by Mobile Twin-Helix Reversal Defects," Macromolecules, 27, 9, pp 2481-2495. 94. Tieyuan, F., et al.(1986), "Study of Factors Affecting Room Temperature Transition of Polytetrafluoroehtylene," Chinese Journal of Polymer Science, 2, pp 170-179.
197 95. Marega, C., et al.(2004), "Relationship between the Size of the Latex Beads and the Solid-Solid Phase Transitions in Emulsion Polymerized Poly(Tetrafluoroethylene)," Macromolecules, 37, 15, pp 5630-5637. 96. D'amore, M., et al.(2004), "Disordered Ch ain Conformations of Poly(Tetrafluoroethylene) in the High -Temperature Crystalline Form I," Macromolecules, 37, 25, pp 9473-9480. 97. Ting-Yung, H. and Eiss, N. (1983), "Effect s of Molecular Weight and Crystallinity on Wear of Polytetrafluoroethylene," Wear of Materials: Inter national Conference on Wear of Materials, pp 636-642. 98. Flom, D. and Porile, N. (1955), "F riction of Teflon Sliding on Teflon," Journal of Applied Physics, 26, 9, pp 1088-1092. 99. Steijn, R. (1968), "Sliding Experime nts with Polytetr afluoroethylene," ASLE Transactions, 11, 3, pp 235. 100. Mccook, N., et al.(2005), "Cryogenic Friction Behavior of PTFE Based Solid Lubricant Composites," Tribology Letters, 20, 2, pp 109-113. 101. Burris, D.L., Perry, S.S., and Sawyer, W.G. (2007), "Macroscopic Evidence of Thermally Activated Friction with Polytetrafluoroethylene," Tribology Letters, 27, 3, pp 323-328. 102. Joyce, J. (2003), "Fracture Toughness Evaluation of Polytetrafluoroethylene," Polymer Engineering and Science, 43, 10, pp 1702-1714. 103. Brown, E. and Dattelbaum, D. (2005), "The Role of Crystalline Phase on Fracture and Microstructure Evolution of Polytetrafluoroethylene (PTFE)," Polymer, 46, 9, pp 30563068. 104. Brown, E., et al.(2006), "The Effect of Crys tallinity on th e Fracture of Polytetrafluoroethylene (PTFE)," Materials Science and Engi neering C-Biomimetic and Supramolecular Systems, 26, 8, pp 1338-1343. 105. Rae, P. and Brown, E. (2005), "The Propert ies of Poly(Tetrafluoroethylene) (PTFE) in Tension," Polymer, 46, 19, pp 8128-8140. 106. Rae, P. and Dattelbaum, D. (2004), "The Pr operties of Poly (Tetra fluoroethylene) (PTFE) in Compression," Polymer, 45, 22, pp 7615-7625. 107. Pucciariello, R. and Villani, V. (2004), "M elting and Crystallization Behavior of Poly (Tetrafluoroethylene) by Temp erature Modulated Calorimetry," Polymer, 45, 6, pp 20312039.
198 108. Pucciariello, R. and Villani, V. (2004), "Phase Behavior at Low Temperature of Poly(Tetrafluoroethylene) by Temp erature-Modulated Calorimetry," Journal of Fluorine Chemistry, 125, 2, pp 293-302. 109. Huang, J. and Lee, W. (2001), "Sealing and Mechanical Behaviors of Expanded PTFE Gasket Sheets Characterized by Pvrc Room Temperature Tightness Tests," Materials Chemistry and Physics, 68, 1-3, pp 180-196. 110. Kurumada, K., et al.(1998), "Structure Generation in PTFE Porous Membranes Induced by the Uniaxial and Biaxial Stretching Operations," Journal of Membrane Science, 149, 1, pp 51-57. 111. Yang, J., et al.(2005), "Morphology Evolution in PTFE as a Function of Melt Time and Temperature I. High Molecular Weight Singleand Multi-Mol ecule Folded Chain Single Crystals and Band Structures," Chinese Journal of Polymer Science, 23, 2, pp 123135. 112. Yang, J., et al.(2005), "Morphology Evolution in Polyte trafluoroethylene as a Function of Melt Time and Temperature. Part II I. Effect of Prior Deformation," Polymer, 46, 20, pp 8723-8733. 113. Lau, S.F., Suzuki, H., and Wunderlich, B. (1984), "The Thermodynamic Properties of Polytetrafluoroethylene," Journal of Polymer Science Part B-Polymer Physics, 22, 3, pp 379-405. 114. Jie, M., et al.(1998), "Damage Evolution and Energy Dissipation of Polymers with Crazes," Theoretical and Applie d Fracture Mechanics, 28, 3, pp 165-174. 115. Kong, X., Chakravarthula, S.S., and Qiao, Y. (2006), "Evolution of Collective Damage in a Polyamide 6-Silicate Nanocomposite," International Journal of Solids and Structures, 43, 20, pp 5969-5980. 116. Marissen, R. (2000), "Craze Growth Mechanics," Polymer, 41, 3, pp 1119-1129. 117. Michler, G.H., Adhikari, R., and Henning, S. (2004), "Toug hness Enhancement of Nanostructured Amorphous and Semicrystalline Polymers," Macromolecular Symposia, 214, pp 47-71. 118. Dennis, H.R., et al .(2001), "Effect of Melt Processi ng Conditions on the Extent of Exfoliation in Organoclay-Based Nanocomposites," Polymer, 42, 23, pp 9513-9522. 119. Fornes, T.D., et al.(2001), "Nylon 6 Nanocomposites: The Effect of Matrix Molecular Weight," Polymer, 42, 25, pp 9929-9940.
199 120. Ma, D.L., et al.(2005), "Effect of Titania Nanopa rticles on the Morphology of Low Density Polyethylene," Journal of Polymer Science Part B-Polymer Physics, 43, 5, pp 488-497. 121. Burris, D.L., et al.(2007), "Polymeric Nanocomposites for Tribological Applications," Macromolecular Materials and Engineering, 292, 4, pp 387-402. 122. Briscoe, B. (1981), "Wear of Polyme rs an Essay on Fundamental-Aspects," Tribology International, 14, 4, pp 231-243.
200 BIOGRAPHICAL SKETCH David Burris graduated from Lemon Bay High Sc hool in Englewood, Florida, in June of 1998. He commenced his higher education at the University of Florida that summer. In the spring of 2002, he joined the Tribology Laboratory at the University of Florida where he began his studies of tribology. The following spring he received his bachelors degree and began his graduate work studying the tribol ogy of PTFE based solid lubrican ts. Upon receiving the degree of Doctor of Philosophy in mechanical engineer ing, David hopes to begin a research laboratory designed to study the underlying mechanisms of friction and wear at solid sliding interfaces.