Nanocrystalline diamond thin films on titanium-6 aluminum-4 vanadium alloy temporomandibular joint prosthesis simulants by microwave plasma chemical vapor depositionкод для вставкиСкачать
INFORMATION TO USERS This manuscript has been reproduced from the microfilm master. UMI films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from any type of computer printer. The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margins, and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send UMI a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion. Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand comer and continuing from left to right in equal sections with small overlaps. ProQuest Information and Learning 300 North Zeeb Road, Ann Arbor, Ml 48106-1346 USA 800-521-0600 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. NANOCRYSTALLINE DIAMOND THIN FILMS ON TI-6AL-4V TEMPOROMANDIBULAR JOINT PROSTHESIS SIMULANTS BY MICROWAVE PLASMA CHEMICAL VAPOR DEPOSITION by MARC DOUGLAS FRIES A DISSERTATION Submined to the graduate faculty of The University of Alabama, The University of Alabama in Huntsville, and The University of Alabama at Birmingham, in partial fulfillment of the requirements for the degree of Doctor of Philosophy BIRMINGHAM, ALABAMA 2002 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UMI Number 3078527 Copyright 2002 by Fries, Marc Douglas All rights reserved. UMI’ UMI Microform 3078527 Copyright 2003 by ProQuest Information and Learning Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. ProQuest Information and Learning Company 300 North Zeeb Road P.O. Box 1346 Ann Arbor, Ml 48106*1346 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Copyright by Marc Douglas Fries 2002 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ABSTRACT OF DISSERTATION GRADUATE SCHOOL, UNIVERSITY OF ALABAMA AT BIRMINGHAM Decree Ph.D. Program Materials Science_______________________________ Name of Candidate Marc Douglas Fries_____________________________________ Committee Chair Yogesh K. Vohra______________________________________ Title Nanocrvstalline Diamond Thin Films on Ti-6A1-4V Temporomandibular Joint Prosthesis Simulants bv Microwave Plasma Chemical Vapor Deposition_______ A course of research has been performed to assess the suitability of nanocrystal line diamond (NCD) films on Ti-6A1-4V alloy as wear-resistant coatings in biomedical implant use. A series of temporomandibular (TMJ) joint condyle simulants were pol ished and acid-passivated as per ASTM F86 standard for surface preparation of implants. A 3-nm-thick coating of NCD film was deposited by microwave plasma chemical vapor deposition (MPCVD) over the hemispherical articulation surfaces of the simulants. Plasma chemistry conditions were measured and monitored by optical emission spectros copy (OES), using hydrogen as a relative standard. The films consist of diamond grains around 20 nm in diameter embedded in an amorphous carbon matrix, free of any detectable film stress gradient. Hardness averages 65 GPa and modulus measures 600 GPa at a depth of 250 nm into the film surface. A diffuse film/substrate boundary produces a minimal film adhesion toughness (Tc) of 158 J/m2. The mean RMS roughness is 14.6±4.2 nm, with an average peak roughness of 82.6±65.9 nm. Examination of the surface morphology reveals a porous, dendritic surface. iii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Wear testing resulted in two failed condylar coatings out of three tests. No mac roscopic delamination was found on any sample, but micron-scale film pieces broke away, exposing the substrate. Electrochemical corrosion testing shows a seven-fold reduction in corrosion rate with the application of an NCD coating as opposed to polished, passivated Ti-6A1-4V, producing a corrosion rate comparable to wrought Co-Cr-Mo. In vivo biocompatibility testing indicates that implanted NCD films did not elicit an immune response in the rabbit model, and osteointegration was apparent for both compact and trabecular bone on both NCD film and bare Ti-6A1-4V. Overall, NCD thin film material is reasonably smooth, biocompatible, and very well adhered. Wear testing indicates that this material is unacceptable for use in demand ing TMJ applications without improvements to wear resistance behavior. Identified problems include high surface roughness due to an inadequate seeding procedure and a porous film surface. It is believed that these problems can be solved by future research, in which case NCD thin films should prove to be well-suited as wear resistant coatings in biomedical applications. iv Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. DEDICATION For my sister Alissa, who left us far too soon. I love you. I miss you. v Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ACKNOWLEDGMENTS As the old saw puts it, nothing worth doing is ever easy. Weil, nothing easy is done alone either, and that’s part of what makes it worth doing. There are a great many people who have made this work possible, and they are truly a "great” many. My wife Kate deserves to have her name up there on the title page with mine, for caring for me body and soul while my mind was systematically sizzling neurons trying to comprehend this mess. With this particular hurdle completed, I am very much looking forward to carrying on with our lives together. I also owe my littermates a debt of gratitude for their unwavering support and dedication. Jeff, Denise, Alissa, and littermate-in-law Sami have each provided encour agement and the occasional "fattening cookie" when I needed it. Spousal units Patti and Robert have also given me invaluable support and encouragement and I thank you both. Aunt Cheryl and Betsy Adams also deserve a healthy measure of thanks for their support and encouragement over the years. I must also thank my mom, Karen Weir, for her obviously pivotal role in this endeavor. Thank you for not leaving me at Disneyland that time I went back inside for ice cream, and for countless other occasions when the nurturing instinct was put to the test. Thank you for passing on all the curiosity, wisdom, and persistence that have made this work possible. vi Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Many thanks are due to my committee members, Drs. Yogesh Vohra, Renato Camata, Michael George, Gregg Janowski, William Lacefield, Jr., and Drs. Advincula and Venugopalan as well. I also greatly appreciate the contributions of magi cian/machinist Jerry Sewell, Drs. Reed Patterson and Neal Chesnut, as well as Paul Baker, Nenad Velisavljevic, Rebecca O Connor Davis, and Aaron Catledge. The contri butions of Drs. Joel Ager m , LBL; John Hutchinson, Harvard U.; and Earl Ada, UAH also made a substantial impact on this work and I am grateful. Thanks also to the faculty and staff of both the UAB Physics and Materials Science and Engineering departments. Dr. Robin Griffin deserves credit for her slightly maniacal and highly effective teaching style, as well as all her assistance. Thanks are due also to another excellent teacher, Cheri Moss. I especially appreciate Dr. Barry Andrews and Letitia Hayes for their understanding and support during a difficult time. I’m certain that I am forgetting people, and I apologize. You should all know that a part of this work is yours in one way or another. Thank you all, ladies and gents. This work was performed with the support of the NASA/Alabama Space Grant Consortium Fellowship program and the National Institute of Dental and Craniofacial Research (NIDCR), National Institutes of Health (NIH), under grant number 1ROIDE13952-A1. vii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. TABLE OF CONTENTS Page ABSTRACT......................................................................................................................iii DEDICATION................................................................................................................... v ACKNOWLEDGEMENTS.............................................................................................. vi LIST OF TABLES............................................................................................................. x LIST OF FIGURES........................................................................................................... xi LIST OF ABBREVIATIONS......................................................................................... xiv CHAPTER 1 INTRODUCTION...............................................................................................1 Problem Description................................................................................... 2 Formal Hypothesis Statement................................................................... 10 2 DEPOSITION................................................................................................... 11 Description of the Deposition Process...................................................... 11 . Plasma Chemistry Considerations............................................................. 18 Relative Plasma Species Concentration by OES Spectroscopy................................................................................ 20 Effects of Methane and Nitrogen Flow Rates on Deposition Conditions...................................................................23 Plasma Physics Considerations.................................................................27 Effects of Chamber Pressure on Deposition Conditions.................................................................................... 30 Effects of Microwave Power on Deposition Conditions.................................................................................... 29 Effects of Plasma Temperature Distribution on Film Growth and Morphology...................................................... 34 Experimental Method for NCD Deposition on Condyle Simulants...................................................................................................37 viii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. TABLE OF CONTENTS (Continued) Page CHAPTER 3 CHARACTERIZATION................................................................................. 41 Introduction.............................................................................................. 41 Film Phase Composition by Raman Spectroscopy...................................43 Sample Phase Composition and Film Grain Size by X-Ray Diffraction (XRD).................................................................................... 47 Elemental Composition Depth Profile by X-Ray Photoelectron Spectroscopy (XPS)..................................................................................51 Morphology of Diamond Film Surface by Scanning Electron Microscope (SEM)....................................................................................55 Film Surface Roughness by Diamond-Stylus Profilometry...................... 60 Film Hardness and Modulus by Nanoindentation..................................... 63 Film Adhesion Strength Analysis by Indentation Testing........................ 66 Assessment of Characterization Findings.................................................70 4 PERFORMANCE ANALYSIS....................................................................................75 Introduction.............................................................................................. 75 Electrochemical Analysis of NCD Biocompatibility................................ 76 Polarization Resistance Determination by Electrochemical Impedance Spectroscopy (EIS)..................................................... 76 Corrosion Behavior Analysis by Potentiodynamic Polarization....................................................................................80 Assessment of Electrochemistry Findings.................................... 84 Wear Resistance Assessment of NCD-Coated TMJ by Mandibular Movement Simulator.................................................................................84 In Vivo Biocompatibility Assessment..................................................... 101 5 CONCLUSIONS.........................................................................................................104 6 REFERENCES........................................................................................................... 110 APPENDIX A COMPENDIUM OF LOW PRESSURE OES TRENDS FOR WAVEMAT 6-KW MPCVD SYSTEM..............................................116 B IACUC FORM FOR IN VIVO BIOCOMPATIBILITY STUDY.................. 137 ix Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LIST OF TABLES Table Page 1 Mean Film Grain Size and XRD Data........................................................................49 2 TMJ Simulant Surface Roughness............................................................................. 61 3 Interface Toughness and Relevant Values..................................................................69 4 Electrochemical Impedance Spectroscopy Data.........................................................79 5 Potentiodynamic Polarization Data.............................................................................83 x Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LIST OF FIGURES Figure Page 1 Diagram showing TMJ anatomy.................................................................................. 2 2 Diagram showing articulation of the TMJ................................................................... 3 3 Typical, commercially available TMJ prosthesis........................................................ 4 4 A total hip prosthesis showing the effects of osteolytic degradation due to wear particle generation.................................................................................... 6 5 Diagram of TMJ condyle simulants with dimensions................................................. 9 6 Wavemat 6-kW MPCVD system................................................................................12 7 CN and C2 peak intensities versus time......................................................................14 8 Typical apparent sample temperature curve over deposition period by two-color infrared pyrometry......................................................................................16 9 Diagrammatic explanation for growth of diamond from plasma.................................18 10 Typical OES spectrum............................................................................................... 20 11 Relationship between the CN/Ha ratio and CH4 and N2 flow rates at 20 Torr and 1.00 kW microwave power, as normalized to H a values at 5% CH4 ................... 25 12 Relationship between the C2/H 01. ratio and CH* and N2 flow rates at 50 Torr and 1.00 kW microwave power, as normalized to H a values at 5% CHt................... 26 13 Calculated plasma temperature distribution by Hassouni et al................................... 28 14 Plasma temperature distribution with change in pressure...........................................29 15 Ha peak area versus microwave power......................................................................30 16 Hp/Ha ratio versus microwave power........................................................................31 xi Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LIST OF FIGURES (Continued) Figure Page 17 C2/HCXratio versus microwave power..........................................................................32 18 CN/Ha ratio versus microwave power........................................................................32 19 Diagram of Tp dependence upon microwave power....................................................34 20 Images of partially coalesced NCD film...................................................................... 35 21 Raman spectra comparison......................................................................................... 43 22 Detail of deconvoluted Raman spectrum.....................................................................44 23 Phonon shift versus angle around simulant radius of curvature..................................46 24 XRD spectra from wear test disk and condyle simulant..............................................48 25 Elemental concentration depth profile by XPS............................................................ 52 26 SEM image of wear test disk surface........................................................................... 56 27 "Analog" simulation of film growth showing effect of seed size distribution............. 57 28 SEM images of simulant film surface.......................................................................... 59 29 Simulant and wear test disk hardness by nanoindentation...........................................64 30 Simulant and wear test disk hardness by nanoindentation...........................................65 31 SEM image of Brale C indent..................................................................................... 67 32 Collected EIS data for polished, passivated Ti-6A1-4V and NCD-coated samples.........................................................................................................................77 33 Potentiodynamic polarization data from polished, passivated Ti-6A1-4V and NCD-coated samples............................................................................................ 81 34 Mandibular movement simulator (MMS)....................................................................85 35 SEM image of wear track on TMJ 5 condyle.............................................................. 88 36 BSE image of same field as Figure 35......................................................................... 88 xii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LIST OF FIGURES (Continued) Figure Page 37 SEM image of wear track on the disk opposing the TMJ 5 condyle........................... 89 38 BSE image of same field as Figure 37........................................................................ 89 39 Wear track edge on TMJ 5.......................................................................................... 90 40 Wear track edge on test disk opposing TMJ 5 ............................................................ 90 41 SEM image of wear track on TMJ 6 condyle............................................................. 93 42 BSE image of same field as Figure 41........................................................................ 93 43 SEM image of wear track on the disk opposing the TMJ 6 condyle........................... 94 44 BSE image of same field as Figure 43........................................................................ 94 45 Wear track edge on TMJ 6 .......................................................................................... 95 46 Wear track edge on test disk opposing TMJ 6 ............................................................ 95 47 SEM image of wear track on TMJ 7 condyle............................................................. 98 48 BSE image of same field as Figure 47........................................................................ 98 49 SEM image of wear track on the disk opposing the TMJ 7 condyle........................... 99 50 BSE image of same field as Figure 49........................................................................ 99 51 Wear track edge on TMJ 7 ........................................................................................ 100 52 Wear track edge on test disk opposing TMJ 7 ...........................................................100 53 Cross section image of implanted NCD-coated disk.................................................102 xiii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LIST OF ABBREVIATIONS AC Alternating current Bcc Body-centered cubic BEN Bias-enhanced nucleation BSE Backscattered electron detector CSM Continuous stiffness mode CR Corrosion rate E bd Breakdown potential EIS Electrochemical impedance spectroscopy E ocp Open circuit potential Epp Primary passivation potential fee Face-centered cubic FDA Food and Drug Administration FP Microwave forward power FWHM Full width at half maximum rc Interface toughness hep Hexagonal close-packed HFCVD Hot filament chemical vapor deposition Ic o r r Corrosion current ISE Indentation size effects IACUC Institutional Animal Care and Use Committee Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LIST OF ABBREVIATIONS (Continued) LRI Laser reflectance interferometry MMS Mandibular movement simulator MPCVD Microwave plasma chemical vapor deposition NCD Nanocrystalline diamond OES Optical emission spectroscopy PACVD Plasma-assisted chemical vapor deposition PAR Princeton Applied Research PMMA Poly-methylmethacrylate PMT Photo multiplier tube RMS Root mean squared seem Standard cubic centimeter per second SEM Scanning electron microscope Te Electron temperature Ti-6A1-4V Titanium - 6% Aluminum - 4% Vanadium TMJ Temporomandibular joint TP Plasma temperature (ion temperature) Ts Substrate temperature UAB University of Alabama at Birmingham UHMWPE Ultra-high molecular weight polyethylene XPS X-ray photoelectron spectroscopy XRD X-ray diffraction Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 1 INTRODUCTION "Don’t quote Latin. Say what you need to say, and then sit down." -Arthur Wellesley, Duke of Wellington Yes, the title is wordy. Describing this work in a single sentence has lead to a rather unwieldy title, but only out of necessity. The research it describes is the develop ment of a coating of a hard, wear-resistant layer of nanocrystalline diamond (NCD) film on simulated condyles from temporomandibular joint (TMJ) implants. It is important that the title include mention of the Ti-6A1-4V alloy ^ - that the simulated condyles are composed of since most metallic TMJ condyles in use today are composed of Co-CrM o .3-5 The scope of this research only incorporates deposition on the condylar portion of the implant, and only on simulated implants. The use of "simulated" in this context means that TMJ condyle-shaped components were machined from bulk Ti-6A1-4V, polished, and chemically treated as though they were going to be implanted in a patient. It does not imply that this is a strictly computational investigation, as a more conven tional use of the term "simulated" may imply. The method of NCD deposition chosen for this work is microwave plasma chemical vapor deposition (MPCVD), and so the title of this work coalesces from an array of descriptive terms - "Nanocrystalline diamond deposition on Ti-6A1-4V temporomandibular joint prosthesis simulants by microwave plasma chemical vapor deposition.” It is hoped that this research will eventually lead 1 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 to a substantial improvement in the TMJ implants currently available in terms of both the service lifetimes of current implant designs and in a broader range of temporomandibular medical conditions that can be successfully treated. Problem Description The human TMJ, as seen in Figure 1, has proven to be difficult to replace with a reliable artificial implant A ? This is due in part to the sliding/rotating articulation of this joint,^ which requires a component geometry that is more complex than those found in most other artificial joints (see Figure 2). In addition, TMJ implants are small and reside in close proximity to the eye, the ear, various nerves, and the brain, with very little bone Fossa Eminence Disc Condyle Ear Canal Masseter Muscle Figure 1. Diagram showing TMJ anatomy. Note the position o f the condyle relative to the fossa eminence which it articulates against, and the proximity o f the TMJjoint to the ear canal. Adapted from reference 7. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3 Figure 2. Diagram showing articulation o f the TMJ. Notice the sliding/rotating articulation o f the condyle and how it dislocates from the fossa eminence. From reference 7. available for fixation.^. 10 TMJ implant service lifetimes have historically been short compared to other common joint replacements as a result, and implants have even been the subject of product recall A 11 Attempts to describe some of the more detailed aspects of TMJ disorders and their treatments run afoul of a very basic shortcoming - very little basic science has been performed on such fundamental issues as the articulation mechanics of the temporoman dibular joint and disorder causes or progression. For example, the vast preponderance of TMJ disorder patients are w o m e n , 12 yet the reasons why are unclear. The basic structure of the TM joint is also poorly understood, as little research has been performed on typical TMJ loading forces, or on anatomical differences between men, women, and various races. As TMJ researcher Dr. Regina Landesberg, DMD, Ph.D., stated, "We need to know what is normal [in the TMJ] before we can say what is abnormal." 11 Perhaps the best example of the dearth of information is found in the first marketed TMJ implants, which were manufactured by Vitek Inc. and Dow Coming. Dow Coming's Silastic sheeting was a bilayer plastic implant designed to replace the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 Figure 3. Typical, commercially available TMJ prosthesis. The condylar compo nent is labeled "A " and is fitted here with a polymer articulating component. Note the complex geometry o f the fossa component "B" as compared to the hemispherical condyle. This image has been altered to remove patient identification labels. TMJ disk that normally cushions the movement of the joint. This product was placed on the market prior to 1976, before the FDA was granted the authority to regulate medical devices. The Vitek component was a similar device marketed in 1983 that was granted FDA approval based on its similarity to the Dow Coming device.** After receiving numerous complaints o f implant fragmentation and delamination, the FDA recalled all Vitek components in 1991, and Dow Coming suspended production o f Silastic compo nents in 1993.6’ 1^ Between 1984 and 1998, over 75% of the FDA's Medical Device Reporting adverse event reports cited malfunctions or patient injuries involving either Dow Coming or Vitek components.** Patients cited severe pain around the ear and jaw, limited jaw movement, bone degeneration, joint noise, and vision and hearing problems among the problems associated with implant failure.** It is believed that the root cause of Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5 the failure of these devices is simple lack of understanding of the magnitude of forces present in TMJ articulation.^ Sadly, this lack of fundamental understanding persists today. 11 A serious effort is currently underway in the dental research community to develop a complete under standing of temporomandibular disorders and their genesis, 13-16 but research is still at a formative stage. For this reason, discussion of the medical issues surrounding TMJ implants will be based on general implant research in this work. Discussions of the relevance of wear behavior, electrochemical corrosion, and other relevant issues will be based on research performed mostly on other commonly replaced joints such as the hip and knee. Failure of joint prostheses typically occurs by one of three general modes. Sur geon error in placement of the prosthesis can lead to excessive wear, poor fixation, or patient discomfort, causing early failure of the device. This mode is typically a short term failure, with implant lifespan measured from weeks to a few years. The second typical failure mode is postoperative infection. Treatment of the infection typically requires removal of the prosthesis, immobilization of the joint, and treatment with antibiotics until the infection has dissipated. Good operative sanitation procedures are the obvious remedy for this mode of failure. The third failure mode is due to osteoly s is ^ of surrounding bone due to wear particle generation, such as that pictured in Figure 4. Surface roughening of articulating implant components leads to increased wear panicle generation from the polymer implant components that implant condyles articulate a g a in s t. *8’ 19 Polymer panicles, in turn, cause osteolysis in the bone proximal to the implant.20’ 21 The implant then loosens, leading to improper joint aniculation, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6 Figure 4. A total hip prosthesis showing the effects o f osteolytic degradation due to wear particle generation. The femoral head component marked "A" articulates against a UHMWPE liner in the acetabular cup labeled "B”. Wear particles gener ated by a corrosion-roughened articulation surface have migrated to implant-bone boundaries. The bone labeled "C" is healthy tissue with normal density. The darker areas marked "D" are regions o f bone resorption brought on by immune reactions to wear particles. Adapted from reference 17. progressively severe implant wear, and accelerating bone loss. Wear debris has also been shown to spur the growth of scar tissue proximal to the joint, causing decreased joint mobility and patient d i s c o m f o r t . 2 2 , 2 3 Currently the generally prescribed course of action for osteolytic degradation is revision surgery, which entails replacement of the implant or the use of a different restorative treatment.^ This third, long-term failure mode is dependent upon implant design and materials composition and is the primary consideration in this body of research. Since long-term implant failure has its roots in the wear behavior properties of the articulating components of the implant, an improvement of wear behavior should result in a longer service lifetime for the implant and greater comfort and utility for the patient. It is hoped that a significant improvement can be Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7 achieved by coating implant articulation surfaces with a very hard, smooth layer of nanocrystalline diamond. NCD film is a two-phase system composed of nanometer-scale crystalline dia mond crystals with an intergranular phase of amorphous carbon. These films tend to exhibit exceptional values for hardness, smoothness, wear resistance, and low roughness. Hardness values typically range from 5 0 to 100 GPa, where the latter value is comparable with natural single-crystal diamond. K. Miyoshi has measured the dynamic coefficients of friction of NCD films similar to those produced in this body of research in both dry nitrogen and humid air (at 4 0 % relative h u m i d i t y ) . 2 4 , 25 His findings of 0 . 0 3 5 and 0 . 0 3 0 in dry nitrogen and humid air, respectively, are very low values, especially in the pres ence of humidity. The typical metal/polymer couples used in implants today exhibit dynamic coefficients of friction in serum of between 0.05 and 0.12,26 although the different media used in these measurements do not allow for absolute comparison. Miyoshi’s paper also reports very low values for the wear factor of NCD film, where wear factor is the volume of material removed per unit load and unit sliding distance. In dry nitrogen, NCD film exhibited a wear factor of 1.6 x 10‘7 mm3/Nm, and 1.8 x 10'7 mm3/Nm in humid air. 24 * 25 Similarly, C. Machado et al report that NCD films exhibit excellent wear resistance as compared to bare Ti-6AI-4V when tested on a mandibular simulation d e v i c e . 2 7 Finally, NCD films produced in this research reproducibly exhibit root mean squared (RMS) roughness values around lOnm. The exceptional attributes of NCD films suggest that they are well suited to application as wear-resistant layers for implant articulating surfaces, and so this research was undertaken to investigate the feasibility of using NCD films to improve biomedical implants. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Currently, a wide range of human joints can be replaced with prosthetic devices. Hips, knees, shoulders, temporomandibular joints and fingers can all be replaced with fully articulating versions of the natural joints. Other joints, such as ankles and vertebra, can be treated through means that do not actually replace the bone itself. For those prostheses that replace the original joint, the articulating surfaces are commonly com posed of a metal condylar surface that articulates against a polymer, such as ultra-high molecular weight polyethylene ( U H M W P E ) . 2 8 Ti-6A1-4V was once commonly used to manufacture implant condyles, as it exhibits excellent mechanical properties^’ 21 and supert) resistance to electrochemical c o r r o s i o n . 2 9 , 30 ft has been shown, however, that Ti-6A1-4V exhibits poor wear resistance in vivo, and the resulting titanium particles subsequently cause bone resorption proximal to the i m p l a n t . 2 0 , 21 por this reason metallic implant condyles are made almost exclusively of C o - C r - M o ^ * ^ alloy today. Some patients, however, are especially sensitive to the nickel found in Co-Cr-Mo, and so a small number of implants are fashioned out of Ti-6A1-4V for these patients. NCD deposition on Ti-6AJ-4V alloy has been studied extensively and is known to exhibit good hardness and low r o u g h n e s s . 31-34 Since Ti-6A1-4V is acceptable as an articulating component material in all terms but wear resistance, coating Ti-6A1-4V implant condyle components with NCD may result in a component that is acceptable for use as an implant in terms of wear resistance and biocompatibility. Out of all the possible choices in implants, temporomandibular joint implants were chosen as the focus of this study primarily because of their small size. A TMJ implant condyle is suitably small compared to the plasma volume such that it does not noticeably distort the plasma in the MPCVD device used here. Such distortion would Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 9 11mm 9mm Figure 5. Diagram o f TMJ condyle simulants with dimensions. The simulant on the left is in the polished passivated state prior to deposition. The simulant in the center is shown post-deposition with a 3-pm coating o f NCDfilm. The diagram to the right shows TMJ simulant dimensions, as they were measuredfrom a commercially avail able implant. Notice the radius o f curvature o f 7.5 mm. cause tuning problems in the MPCVD microwave resonance cavity and would likely result in a non-uniform NCD coating. Other important factors include the positive impact that an improvement in TMJ treatment procedures would have on the sufferers of tem poromandibular disorders. It has been estimated that more than 10 million people suffer from TMJ-related disorder symptoms in the United States alone. ^ The current state of TMJ implant design and materials technology*** 1U 12 provides an opportunity to make a significant positive impact upon the health, comfort, and well-being of a great many TMJ disorder sufferers around the world. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 10 Formal Hypothesis Statement The complex geometry of implant articulating surfaces can be coated with a welladhered thin film of nanocrystalline diamond by microwave plasma chemical vapor deposition. The NCD film will prove to be sufficiently wear resistant, smooth, reliably adhered, and biocompatible for long-term use as the articulating surface of an implant articulation surface. Ti-6A1-4V is chosen as a substrate material for its proven compati bility with the MPCVD diamond deposition process and for it’s excellent material proper ties in all pertinent categories except for wear resistance. The addition of a thin layer of NCD will improve the wear resistance of Ti-6AI-4V to a point where the system is acceptable for use as an articulating-surface implant material. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 2 DEPOSITION "The most exciting phrase to hear in science, the one that heralds new discoveries, is not, "Eureka!" (I found it!), but, "That’s funny..."" • Isaac Asimov Description of the Deposition Process All deposition carried out in this research was performed in either a custom-built 1.2-kW MPCVD system at the University of Alabama at Birmingham (UABp5 or in a Wavemat Corp. 6 kW MPCVD device shown in Figure 6. Both devices operate by heating the sample surface by direct contact with the plasma, as opposed to other types of CVD devices, such as plasma-assisted chemical vapor deposition (PACVD) systems, that use a colder plasma and either a hot filament or stage heater as a supplementary heat source. This is an important distinction, as PACVD devices generally exhibit lower mean plasma temperature values than MPCVD systems. The higher mean plasma temperature in MPCVD devices equates to accelerated growth or growth of harder films as well as the production of a thicker TiC interlayer due to increased surface diffusion by the higher-energy carbon s p e c i e s ^ , a thicker TiC layer, in turn, presumably generates greater adhesion in the resulting film than that produced in a PACVD system. This is the primary justification for the use of a MPCVD for coating implants, since it is critical to maintain film adhesion throughout the lifespan of the component. As a point of clarifica tion, the term "plasma temperature" (Tp) as used here refers to the temperature o f the ions and neutral particles in the plasma, as opposed to electron temperature (Te). 11 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 12 [ t u p li TrwMnw — M M dM nt Figure 6 . Wavemat 6 -kW MPCVD system. Notice the black microwave resonance chamber in the image on the left, and the microwave source mounted above the deposition apparatus. The inset image shows the plasma during NCD deposition on a TMJ simulant. In the diagram on the right, notice the simulant on its translation stage and the comparative size difference between the plasma and the TMJ simulant. MPCVD devices generate a plasma at low pressure (typically 20-150 Torr) at the local maxima within a microwave resonance cavity. Both o f the devices described in this research utilize a 2.45-GHz microwave power supply. The greater power output capabil ity and larger deposition chamber size of the 6 kW system allow deposition over surfaces up to 10 cm diameter, where the 1.2 kW system is restricted to samples 2.54 cm or less in size. Another major difference with the Wavemat device is that this machine uses a quartz bell jar to isolate the low-pressure plasma environment from the resonance cavity. The sample sits on a small (2.54 cm), vertically translating molybdenum stage in the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 13 center of a primary fixed stage 12.7 cm in diameter. When fully recessed, a flat sample is several centimeters away from the hottest portion of the plasma so that the deposition conditions can be set up and stabilized without subjecting the sample to a prolonged period of hydrogen etching. The Wavemat 6kW MPCVD system used for most of this research is controlled by a personal computer running custom software written in the LabVIEW programming language. The software monitors the system during deposition and automatically shuts down the deposition process in the event of an alarm condition. Sample temperature is stored versus time in a data file over the course of the entire deposition. Repeated experiments have shown that diamond film growth is highly irregular or nonexistent for as-polished Ti-6A1-4V samples in the 6 kW system. Seeding the surface alleviates this problem by ensuring that diamond nuclei larger than the critical nucleus size (r*) are distributed over the sample surface. In order to seed the substrate with diamond n u c l e i ^ , each sample is ultrasonically agitated in a 5.0% w/w I -2-pm dia mond/water solution for 50 minutes to scratch the sample surface. The sample is then cleaned under a stream of water for 2 minutes and any powder deposits are manually removed. The sample is then rubbed against a wet polishing cloth loaded with 1-2-pm diamond powder for I minute, and rinsed under running water for about 30 seconds. Even coverage of diamond particles is not certain when using this method, and this fact will play an important part in deposited film morphology, as shall be discussed later. The sample is then loaded onto the translation stage in the fully recessed position. The chamber is pumped down to minimum pressure. Hydrogen flow is started, and once chamber pressure exceeds 0.5 Torr microwave emission is initiated at the minimum value Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 14 CN and C2 Peak Intensity vs. Time Normaiiad to Matching Max Values CN >» « c ® c 0 200 600 400 Time (s) 800 Figure 7. CN and C2 peak intensities versus time. Raw peak intensity o f CN (387.5 nm) and C2 (515.5 nm) peaks were collected versus time atfixed PMT voltage. Note the very rapid appearance o f CN and relatively slow C2 appearance. o f 0.6 kW for this device. Plasma ignition is usually instantaneous. Pressure and power are raised proportionally until the intended operating conditions are achieved. To begin deposition, the sample is raised into the plasma until the sample center measures 750±1°C as measured using a Mikron M77LS two-color optical pyrometer. At this point, methane and nitrogen are admitted into the chamber, and the presence of plasma species more massive than hydrogen drives up the sample temperature to »800°C by momentum transfer from the heavy plasma species to the sample surface. The following several minutes are characterized by diamond nuclei growth at the sample surface as the presence o f carbon-containing plasma species (excluding CN) feeds diamond growth through a process not folly understood by the scientific community at this time.35, 37-39 within 2 minutes, CN reaches a maximum concentration that is greater than the equilibrium Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 15 concentration found as little as 2 additional minutes later (see Figure 7), and it continues to decrease slightly over a period of several hours. Cj concentration builds as an expo nential function towards an equilibrium value that is established in approximately 15 minutes. Once the growing diamond nuclei reach a size large enough for internal reflection in the 1.2- and 1.4-pm infrared range detected by the optical pyrometer, the apparent sample temperature climbs sharply. This phase is readily evident in Ts vs. t data and typically begins 6 to 10 minutes after CH» and N2 admission. It is probably reasonable to assume that the film is near coalescence at this point, although uniformity in nuclei distribution is uncertain when ultrasonic seeding is used^6 to pre-treat the sample. For purposes of characterization of the interlayer region, it is assumed that carbon diffusion into the bulk Ti alloy effectively ceases at this point, since once the sample surface is covered with NCD film, further carbon diffusion would require the dissociation of diamond in the film. This reaction is assumed to be energetically unfavorable, although diffusion of titanium through the intergranular carbon phase is likely to continue throughout the deposition period. The Ts vs. t data continues to show periodic oscillations for the remainder of the deposition period, or until the amplitude of the oscillations decreases beyond resolution. Typical Ts vs. t data is found in Figure 8. The oscillations arise from internal reflection within the growing film, cycling through periods of constructive and destructive interfer ence.^ Depending on the infrared clarity of the film, the apparent sample temperature has been seen to oscillate by more than ±150°C from the true sample temperature. Film growth has been compared to the oscillations seen in Ts vs. t and in laser reflectance Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 16 TMJ 6 Optical Pyrometry Data 500/106/7.95 seem HjCH JH2 950 900 850 O 800 O tn I- 750 700 650 600 0 1000 2000 3000 4000 5000 6000 Time (s) Figure 8 . Typical apparent sample temperature curve over deposition period by twocolor infrared pyrometry. Point "A " indicates the start o f CH4 and N2 flow with a rise in sample temperature. Period "B" is the initial nucleation period. Oscillations are seen in film growth period "C" and the sample cools in period "D" after CH4 and N2 flow stops. interferometry (LRI),40 and in the case of the Mikron M77LS pyrometer used in this research 2 */3 interference fringes in the Ts vs. t data equate to ~1 pm of film growth at the spot where the pyrometer is focused. Using the optical pyrometry data in place of a reflected laser beam in the case of LRI results in fewer oscillations during film growth due to the shorter wavelength of infrared light relative to, for example, a He-Ne laser. However, the geometry o f the 6 kW MPCVD would restrict a reflected laser beam to a very shallow reflected angle, and the beam would be severely attenuated by fragmenta tion protection screens on the windows of the 6 kW system. In practice, measurement of Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 17 film growth by optical pyrometry is simple, reliable, and provides Him thickness meas urements in the ~100-nm range. Typically the amplitude of the oscillations dampens over time. Since the rate of amplitude decay is ostensibly a function of the Him intergranular phase volume fraction, the possibility exists that Ts vs. t amplitude decay may provide an in situ approximation technique for predicting the hardness and modulus and, by extension, the adhesion strength of an NCD film during the deposition process. This possibility is beyond the scope of this research and is left to future work. Once the film has grown to the desired thickness, methane and nitrogen flows are stopped and the total flow time is recorded. The sample is monotonically cooled by a custom software routine that lowers the chamber pressure and microwave power consis tently over a prescribed period. The cooldown period is typically 12 minutes, simply because this value has produced the fewest delamination events in past experiments. Once pressure and power are minimized, the microwave generator is shut off, and so the plasma is extinguished, H2 flow is stopped, and both water and air cooling to the MPCVD system are stopped so that the sample is allowed to cool by radiative and conductive cooling along with the entire deposition chamber. This step prevents the rapid cooling that would occur if cooling water continued to flow with no microwave heat source present. After a further cooling period of approximately IS minutes, the deposition chamber is slowly returned to atmospheric pressure, and the sample is re moved from the chamber. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 18 tE Sofid Diamond Reaction Coordinate Figure 9. Diagrammatic explanationfo r growth ofdiamondfrom plasma. The energy o f a reaction is found here as the difference between the energy o f the reactants (Er) and the energy o f the products (Ep). The energy required to drive the reaction to completion is the difference between the threshold energy (EJ and Er. For a plasma, shown as the upper curve, the mean energy states o f both the reactants and products are significantly higher than those found at room temperature and pressure. E, does not change, so the effective energy barrier to formation o f diamond is lower. All energies are arbitrary and this figure is shown as a diagram only. Pbsma Chemistry Considerations The microstructure and mechanical properties o f deposited NCD films are strongly dependent upon the chemical composition o f the plasma that the film condenses from. Controlling the chemical composition of the film defines values for adhesion, roughness, hardness, continuity, and thickness distribution. The features of the plasma that must be controlled in order to manage film properties include chamber pressure, microwave power, and gas flow rates. Additionally, sample temperature (Ts) must be maintained at 750°C in hydrogen plasma such that Ts ~ 800°C after methane, nitrogen, and/or other gases are admitted to the chamber. No consensus is established in the scientific community as to the formal growth mechanism for diamond films. Diamond researcher Dr. John Angus most clearly sum marized the situation with his comment, "The critical research problem [in diamond film Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 19 growth] remains the lack of understanding of the molecular processes taking place during the nucleation and growth of diamond from the vapor."41 Rather than trying to explain all the different theories of diamond growth here, this work will present a generalized explanation of the process. Formation of diamond is an endothermic process with a large activation energy (see Figure 9). Although graphite is the stable ground state for solid carbon at atmospheric pressure and room temperature, the energy threshold for the decomposition of diamond to graphite is great enough that the reaction does not normally occur. The process of creating diamond from a carbon source requires the same reaction to occur in reverse. The large energy threshold between diamond and graphite states must be overcome, and then diamond persists as a metastable state. There are two routes to accomplish this reaction: high pressure and high temperature, or low-pressure vapor synthesis. For the low-pressure vapor synthesis used here, the reaction is greatly simpli fied in terms of energy of formation simply by raising the mean energy state of the vapor. As a plasma, the gaseous carbon source exists in a high energy density environment. The result of this, energetically speaking, is that the initial and final states of the carbon are much more energetic relative to the fixed threshold energy for the gaseous carbon to diamond transition. A great deal of ambiguity exists as to exactly which mechanism supplies the necessary energy to overcome the energy threshold to diamond formation, but the final result is metastable diamond that possesses a severe energy threshold to decomposition at room temperature and pressure. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 20 TMJ 6 E OES Spectrum CN co Ha iCH c B c 350 400 450 500 550 600 650 700 Wavelength (nm) Figure 10. Typical OES spectrum. This spectrum was collected during NCD deposi tion on a TMJ simulant. Visible are CN, Ci, CH, //?. and the H a and HfiBalmer lines. C? is measured at 515.5 nm, where it is most intense fo r MPCVD deposition condi tions. Relative Plasma Species Concentration by OES Spectroscopy In this research, the primary tool used to monitor the plasma during deposition is optical emission spectroscopy (OES).42-46 This method simply gathers light from the excited plasma and separates it by wavelength. The peaks in the resulting spectrum correspond to chemical species present in the plasma. Unlike most spectrometry systems, OES does not use a standardized source of photons but instead collects light emitted from the plasma itself. In the system used in this research, an inverted, uncoated Leitz lOx Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 21 microscope objective is used to focus the emitted light onto the end of a fiber optic cable. The cable channels the light into an Acton Research SpectroPro -500i monochromator, where it is divided by wavelength and detected using an Acton Research PD-439 photo multiplier tube (PMT) as shown in Figure 10. A personal computer with an Acton Research SC 1 Spectra Card data acquisition card and its software display the collected data and control the monochromator. The standard OES spectrum parameters used in this research include a range of 600 data points collected from 370 to 670 nm with a signal integration time of 70 ms and a 1200 line/mm grating. PMT tube voltage is varied such that the most intense peak recorded has a peak value of around 60,000 counts, since the acquisition card memory cannot handle more than 65,000 counts. The spectra are exported to a spreadsheet that divides the spectrum into small segments near each promi nent peak, subtracts the background from that peak, and calculates the peak area by a simple bin-counting method. The peaks monitored include the Balmer hydrogen emis sion lines of Ha at 656.3 nm and HP at 486.1 nm, the CH A2A-X2n transition at 431.4 nm, the Cz A3n g-X3n u (Swan) system at 516.5 nm, and the B2I-X 2I violet CN system at 388.3 nm.43 A new method has been produced by this research to express OES spectrum in formation in a relative fashion. In order to compare spectra between experiments and relate thin film properties to plasma chemistry conditions, each OES spectral peak is expressed as an intensity ratio relative to the H a peak collected in the same spectrum. The hydrogen flow rate is kept at a constant 500 standard cubic centimeters per second (seem) for all experiments. Presumably, all measurements of chemical species in each experiment are comparable, as they are measured against a species with reasonably well- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. understood b e h a v i o r . 4 2 , 4 3 , 4 5 Ha intensity will vary with chamber pressure, micro wave power, and other variables, but the emission intensities of all chemical species monitored will also vary in accordance to changes in these parameters. See Appendix A for a series of graphs on H a emission in various MPCVD conditions. H a intensity is strongly dependent upon OES operating parameters and deposition chamber geometry, but it is still acceptable to use it as a basis for measuring plasma chemical conditions since any variance in H a intensity behavior will be comparable in all experiments performed in the same regime and on the same deposition device. Whether or not this method can be used for comparison of experiments between MPCVD systems with different chamber geometries or between different CVD systems is an open question at this time. Most likely, since this system is dependent upon details of the OES system such as the type of lens used, the type of fiber optic cable used, and the dimensions of the deposition chamber, each set of deposition conditions characterized using this method will be specific to the machine or class of machines that the measurements are taken on. Since the scope of this project covers experiments performed on a single MPCVD device within a relatively narrow set of deposition conditions the Ha standardization method is an acceptable means of plasma chemistry measurement. This method has proven to be near real time, reliable, and easy to implement. Other researchers have used argon actinometry^? to collect standardized spectra for quantitative OES species measurement, but it has been shown^^ that addition of argon to the deposition process has a measurable effect on the grain size and roughness of the resulting diamond film. As a result, actinometry is not recommended for methane concentrations greater than 6%,4? which precludes the use of actinometry in this study. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 23 One drawback to the OES system used in this research is its lack of spatial resolution, as light is gathered from the entire deposition chamber. As a result, OES spectral lines are of low resolution compared to OES systems that use some form of focusing a p p a r a t u s , and the spectra contain information from comparatively cold gas far from the substrate. As a result, no conclusive data can be obtained on the chemical composition or the plasma temperature at the growth surface itself. Since the local plasma environment at the substrate will vary with that of the entire plasma, the collected OES spectra are still relevant, even if they are more generalized than may be optimal. Effects o f Methane and Nitrogen Flow Rates on Deposition Conditions Of the plasma species detectable with the OES system used, Cz and CN are given dominant consideration. It has been understood for quite some time that C2 concentration is closely tied to diamond growth r a t e , 48 and several researchers have made an excellent case for the idea that C 2 is in fact the principal diamond growth s p e c i e s . 3 9 , 4 6 ,4 9 When predicting the film qualities expected to deposit from a given plasma, a third important factor to consider is microwave power. This last factor will be discussed in the plasma physics section that follows this section. A calibration experiment was performed using the 6-kW MPCVD system to assess trends in CN/Ha, C2/Ha, and other OES ratios with respect to chamber pressure, microwave power, and methane and nitrogen flow rates. Calibration was performed for constant pressure and power with variable CH4 and N2 flow rates over three pressure regimes, and for constant flow rates and pressure with variable power. Since this work focuses on diamond deposition at low pressures, three pressure regimes, of 20, 35, and 50 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 24 Torr, were chosen for consideration. For flow rate variation, methane flow was varied from 5 to 25% of hydrogen flow in steps of 5%, and nitrogen was varied between 0 and 9% of methane flow in steps of 3%. Hydrogen flow was maintained at a constant 500 seem through the entire calibration process, and microwave power was fixed at 1.00 kW except where otherwise noted. For each flow rate step, the system was allowed to equilibrate for 5 minutes after gas flow adjustment. An OES scan was collected of 600 points between 370 and 670 nm, with integration time of 70 ms/point, PMT voltage (Vpmt) varied for maximum resolution, and detected current range fixed at IE-6 A. OES ratios collected during calibration were plotted in a 3D graph and fit by the least squares method using a bivariate quadratic equation. The fit yields a six-term equation that describes the behavior of each OES ratio for a fixed pressure and power, in terms of methane and nitrogen flow rates over a range of values typically seen in deposition. Comparing the same data collected at different pressures yields a picture of OES ratio behavior with respect to pressure for the same span of flow rates, with fixed microwave power. Concurrently with the above data collection, H a raw intensity was collected. OES scans of 60 points each were collected between 653.5 and 659.5 nm, with an inte gration time of 70 ms/point and PMT voltage fixed at -500V. The area under the H a line was calculated by measured peak area after background subtraction, and graphs of raw H a intensity versus CHt and flow were calculated for each pressure regime. The resulting graphs show practically no change at 35 and 50 Torr, but a significant decline at higher flow rates in the 20 Torr regime. As a result, graphs of CN/Ha and C:/Ha nor malized to H a intensity at 5% CH4 flow were also prepared, to remove the effect of Ha Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 25 Normaized CN/Ha Ratio vs. %CH4and %N2 P = 20 Ton-, FP = I.OOkW Normafind to Ha V/Mues at 5% CH4 Z = 0.00802000 - 0.0186Y2 - 601 e V ♦ 0298Y - 0.000151X ♦ 0.331 Figure 11. Relationship between the CN/Ha ratio and CH4 and fyflo w rates at 20 Torr and LOO kW microwave power, as normalized to H a values at 5% CH+ Note the dependence o f CN/Ha upon %N2 over CH4 flow at this pressure. Dark-colored circles indicate data points that lie below thefit curve, while light-colored circles lie above it. line depression from the graphs and so more accurately show the behavior of the species in the OES ratio numerator. A complete compendium o f the graphical data obtained in this calibration is listed in Appendix A. For the purposes of this discussion, results will be discussed primarily for the important CN/Ha and C2 /Ha ratios. Results of the fixed power, variable CH4, and N2 flow rate calibration over the three pressure regimes show that CN/Ha varies strongly with %N2 over all pressure regimes, but is less effected by CH4 flow at 20 Torr (see Figure 11). Increasing the amount of available molecular nitrogen also increases the amount of available atomic nitrogen, and the CN/Ha ratio behavior reflects this relationship. For low-pressure Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 26 Normaized C ^ia Rabo vs. %CH4and %N2 P * 50 T«r, FP * I.OOkW NomRad to Ha VMun at 5% CH4 Z = 0.00305(XY) ♦ 0.000255Y2 + 0000178X2- 00360Y + 0.10CX -0.00254 Figure 12. Relationship between the C /H a ratio and CH 4 and N2 flow rates at 50 Torr and 1.00 kW microwave power, as normalized to H a values at 5% CH4. Note the independence o f C2/H a ratiofrom N2 flow at this pressure. conditions where both atomic nitrogen and atomic carbon are scarce, atomic nitrogen formation is seen as the CN formation rate-limiting reaction. Also evident, however, is the fact that for pressures greater than about 35 Torr, CN formation is equally dependent upon both N2 and CH4 flow rates. The C2/H a ratio reveals practically no dependence upon N2 flow at all, and it varies roughly linearly with increasing CH4 , flow as seen in Figure 12. This behavior is consistent with a simple combination reaction dependent only on the concentration of a single specie,50 whereby the product concentration will vary directly with the reactant. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 27 Plasma Physics Considerations The effects of pressure and microwave power on calibrated OES ratios will be discussed here, as a continuation from the previous chapter. Since microwave power essentially relates to the temperature distribution of the plasma, this very important factor will be considered here as well. Plasma temperature plays a pivotal role in the nucleation and growth of diamond films, and so upon the morphology, hardness, and modulus of the films. Effects o f Chamber Pressure on Deposition Conditions Changes in pressure affect OES ratios mostly through changes in magnitude of the existing trend. Since both CN and Ci form by combination reactions, pressure plays an important role in their concentration. With greater pressure, the mean free path of reactants decreases, and so the likelihood of reactant interaction increases. Since there are more reactant interactions with greater pressure, the equilibrium concentration of these two species increases with increasing pressure. The most interesting feature of the series of CN/Ha graphs with varying pressure is the lack of change. For all three pressure regimes, the graph is virtually identical. See Appendix A for the complete set of OES ratio graphs. The indication here is that the concentration of CN is at its equilibrium value in all instances; in other words, CN saturates very quickly. Furthermore, this indicates that the formation of CN is spontane ous and very likely exothermic. CN formation is apparently more energetically favorable than the other reactions for which atomic carbon production is a limiting step, and so CN formation should dampen the production of such other species. It is interesting to note Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 28 3000 I Figure 13. Calculated plasma temperature distribution by Hassouni et al. The image to the left represents plasma ion temperature distribution fo r a MPCVD reactor similar to the Wavemat device at 0.60 kW microwave power and 18.75 Torr pressure. The graph to the right shows the temperature distribution through the hottest part o f the plasma. Plasma temperature will follow a similar function across the sample surface, albeit to a lesser extent. From reference 51. here that the Ci/H a fit does not vary significantly with %N2 , which states that C 2 and CN formation are not mutually dependent even at low pressure. The slope of the C2 /HGC fit increases linearly with increasing pressure. One minor fluctuation is found at 20 Torr, where the C2/HCI fit deviates slightly from planarity at high CH4 and N 2 flow rates. This is due to competition between CN and C2 formation for the relatively small proportion of atomic C available at low pressure. The indication is that CN production is favored, as C iM a decreases with increasing %N2 at 25% CH». The physical change implemented in the plasma through change in pressure is ac tually a nonlinear redistribution of plasma temperature, which is somewhat surprising given the roughly linear change in OES ratios that is seen. K. Hassouni et ai^l have exhaustively modeled the physics of MPCVD reactors in terms of absorbed power, Te, Tp, and other factors. Figure 13 shows the Tp distribution for an MPCVD system very similar to the Wavemat device used here at a lower microwave power than that typically Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 29 Figure 14. Plasma temperature distribution with change in pressure. Notice that the area under the curves is roughly equal, but the sample center is exposed to much higher plasma temperature with increasing pressure. This figure is for diagrammatic pur poses only. used for NCD deposition (0.6 kW versus 0.8 to i.O kW). Of particular interest is the distribution of temperature across the sample surface, which for optimal conditions would be constant. Figure 13 also shows a cross section of Tp as shown through the hottest portion of the plasma. The distribution of plasma temperature at the sample surface can be expected to mirror that seen in the cross section through the plasma, although to a lesser extent. If we assume that the absorbed energy does not significantly change along with changes in pressure, and that microwave power remains constant, then plasma tempera ture will change in the manner shown in Figure 14. Notice that the physical size of the plasma decreases with increasing pressure, and that the center of the plasma becomes hotter. Also important is the increasingly sharp temperature gradient across the sample surface with increasing pressure. Both of these features are very important, and they will be discussed further in the section on temperature distribution. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 30 Ha Peak Area vs. FP Vpur Fixed at-500V P = 35Torr, CMJH2 = 0.15, Nj/CH4 = 0.045 250000 200000 a S < J£ aa s 150000 100000 X Y =158874X -78831 50000 0.6 0.8 1 1.2 1.4 1.6 1.8 2 FP (kW) Figure 15. Ha peak area versus microwa\>e power. This graph shows the linear relationship between raw H a peak area and microwave forward power (FP) fo r fixed CH4 and N2 flow rates, fixed pressure, and fixed PMT voltage. Effects o f Microwave Power on Deposition Conditions The second series of calibration experiments fixed the pressure and flow rates at 35 Torr and 15% CH4 and 4.5% Ni. These conditions define the midpoint of the parame ter "map" defined in the constant-power calibration. Microwave power was then varied through 0.6 to 2.0 kW in steps of 0.2 kW. The system was allowed to equilibrate for five minutes after each change in microwave power, and then a standard OES spectrum like that described earlier was collected. Reflected power rose from 0.01 to 0.03 kW through the course of this measurement, but chamber tuning was not adjusted, so as to preserve the plasma temperature distribution throughout the experiment. The data collected from this trial were plotted in 2D graphs of OES ratios (or raw peak intensity in the case of Ha) and fit using a least squares routine and the equation best suited to each graph. Raw H a peak data were collected using the same conditions described for raw H a measurement earlier. VPMTwas fixed at -500V for all measurements. The result Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 31 Hp/Ha vs. FP P = 35 Torr. CH/H2 = 0.15. Nj /CH4 = 0.045 0.35 0.3 0 33 °-2 5 S. 0.2 a £ 0.15 CS» 1 0.1 0.05 0.6 Y = 0.0748X2 • 0232X + 0.431 0.8 1 1.2 1.4 1.6 1.8 2 FP (kW) Figure 16. Hfi/Ha ratio versus microwave power. This data indicates the lack o f a clear relationship between Hp/Ha and microwave power despite claims that this ratio is an indicator o f plasma temperature. For the Wavemat system andfor this OES setup, this ratio is not useful as a measurement o f plasma temperature. (Figure 15) is a linear increase in H a intensity with microwave power. This result indicates that plasma temperature increases linearly with power. This validates the assumption that power absorption is constant with change in power over this range of values, as a drop in absorption efficiency would manifest itself as a negative deviation from linearity at high power values. Some researchers have claimed that the H(VHa ratio is directly related to plasma temperature,45 yet the data collected here and presented in Figure 16 seem to invalidate that claim. However, when you consider that the OES system does not pick out a loca tion in the plasma to measure but rather measures the entire chamber, ambient gas included, the results make sense. The H0/Ha ratio actually decreases with increasing power, yet this result is most likely due to vagueness in the collection scheme rather than a true physical phenomenon. This data serves as a reminder that with all data observed Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 32 Cj/Havs.FP P = 35 Torr. CI-yHj * 0.15. Nj/CH4 = 0.045 O « £C 1.5 a o* 0.5 Y = 0.394X2 - 0.402X + 1.01 0.6 0.8 1 1.2 1.4 1.6 2 1 .8 FP (kW) Figure 17. C2/HQ ratio versus microwave power. This relationship is governed by a second-orderfunction as given by the equation above. CN/Ha vs. FP P = 35 Torr, CH4/H2 = 0.15, Nj /CH4 = 0.045 O ffl 0C a 5 z o Y= 4.21 X2 - 6.60X + 3.35 0.6 0.8 1 1.2 1.4 1.6 1.8 2 FP (kW) Figure 18. CN/Ha ratio versus microwave power. Note that this secondorderfunction is more sensitive to microwave power than C2/H 0 L Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 33 here the observer must consider the fact that the data represents the entire volume of the plasma. The Cj/Ha ratio increases as a second-order function with power, as seen in Figure 17. If the rate-limiting step of C2 is assumed to be combination of excited-state atomic carbon, then the reaction will proceed as a function of the mean free path of the reactants. For this case, the mean free path is a volume, and so the number of excitedstate atomic carbon species that any given atomic carbon species can react with will lie inside a sphere centered on the carbon atom and defined by the mean free path (i.e., by the temperature) and the excited state lifetime of the species. 50 The result that arises from all this is a nonlinear increase in the combination reaction rate with increasing power, as seen in this graph. Because the formation of CN is also dependent upon a combination reaction, the CN/Ha ratio also varies as a second-order reaction with increasing power, as seen in Figure 18. The more energetically favored formation of CN over Ci is expressed in the much greater increase of CN relative to Ci, with increasing plasma temperature in the plasma core. The physical change in plasma temperature distribution with change in micro wave power follows a different pattern than change in pressure. Pressure change in volves a constant power input spread between different particle densities. By contrast, changing the power changes the plasma temperature at all points in the plasm a. 50. 51 The plasma expands radially outward in a near-linear fashion with increasing power, and the temperature distribution increases at all points as long as the microwave coupling mode does not change geometry.51 The result is a change in plasma temperatures similar Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 34 FP, t D l i f t c f r m Pit Center Figure 19. Diagram ofTp dependence upon microwave power. Note that the plasma expands as more power is applied, with a corresponding increase in maxi mum plasma temperature. This graph is diagrammatic only, and does not represent actual data. to that shown in Figure 19. Note that, for a sample surface in contact with the plasma, the temperature gradient outward from the plasma center can be expected to maintain a similar slope as power changes. Effects o f Plasma Temperature Distribution on Film Growth and Morphology Changes in the plasma temperature have a profound effect on the morphology and composition of deposited films. Capillarity theory holds that growth rate ( R ) is directly proportional to and the critical nucleus size (r*) is inversely proportional to vapor-phase temperature, ^2 so the spatial distribution of plasma temperature in contact with the sample will have a direct impact on the growth of the film in terms of nucleation density and growth rate.52 The result of this interplay of factors is a critical "zone" of plasma temperature where coalesced NCD film growth occurs. Since this zone is based on plasma temperature (if growth species concentration is neglected), areas that are favor able and unfavorable to coalesced diamond Him growth will coexist as annular zones Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 35 Figure 20. Images o f partially coalesced NCDfilm. The upper left picture is taken underfluorescent light and shows refractionfringes. The "film" appears to cover the entire surface. The upper right image is o f the same field under incandescent lighting nearly parallel with the sample surface. Dark areas are complete film, and light areas are fields o f particles much like those seen in the bottom SEM image o f a tEfferent sample at higher magnification (note the scale markers). spreading radially outward from the center of the plasma. Figure 20 shows a sample with visible zones of complete and incomplete film coalescence, even though the entire sample surface is coated with NCD thin "film.” Three zones exist in a plasma, as defined by plasma temperature distribution. The first, innermost zone, is a region where high Tp defines an r* value that is equal to or greater than the mean seed crystal diameter. This assessment relies on the assumption that only seed crystals are responsible for nuclei formation, but even if that assumption is Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 36 invalid it works well to explain the process. All nuclei with radii less than r* are absorbed into the plasma, and if r* is large enough the number of nuclei that survive to grow will not cover the entire surface (coalescence). Limitations in the OES system used in this work do not allow measurement of Tp in regions where this occurs but, generally speak ing, pressures of greater than 120 Torr at ~1.00 kW or power greater than 1.50 kW at lower pressures will locally heat the plasma enough to prevent coalescence as described here. This broad statement relies on factors such as sample geometry and seed nucleus size and should not be literally interpreted. The center zone, a sort of "Goldilocks zone" for NCD films, is the region where r* is less than the mean seed crystal size and R is high enough for the growing nuclei to coalesce into a complete film. In Figure 20, the center image lit from the side shows a dark annular band. That band is a smooth, coalesced film that is not scattering the incident light. The zone in the center is composed of small grains of noncoalesced film material, and they scatter the incident beam to the camera and so appear brighter. The outermost zone, again illustrated in the center image in Figure 20, also does not coalesce into a complete film, but this is due to low growth rate despite a low value of r*. It can be safely assumed that the vapor phase of depositing species is directly propor tional to the growth r a t e . 5 2 This outermost zone is the result of a relatively cold plasma that contains a concentration of growth species that is too low to support film growth. An interesting footnote is that the Raman spectra collected from this zone and the innermost zone are often very similar in appearance. This finding illustrates that film growth processes in the center zone are not suspended, but rather the film etching processes Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 37 occur at rates equal to or greater than film deposition, leading to a dynamic flux of film material growth that does not coalesce. The "trick," then, is to optimize plasma chemistry and physical conditions such that optimal conditions for coalesced film growth with the desired properties occurs over the entire area of the substrate. This interplay of factors is extraordinarily complex, but empirical data produces useful trends. Generally speaking, high microwave power results in large-grained diamond film. Low power (typically less than 1.00 kW) usually produces nanometer-scale diamond grains. High pressure results in faster growth rates but severely restricts the radial size of a deposited film, and the film will contain a greater disparity of grain size and surface morphology in a radial fashion than a film grown under similar conditions at lower pressure. Increasing the CN/Ha ratio generally reduces the diamond grain size, but at ratios greater than about 2.0, the 1550 cm'1Raman band becomes significant and the film hardness drops appreciably. Increasing the C^/Ha ratio generally increases the growth rate, which for very high values (greater than =1 pm/h) results in decreased film hardness. Film grain size is maximized for moderate (~ 1.5) ratios of C2/H0 Cat high pressure (100 Torr or higher). Diamond grain size decreases with decreasing microwave power and increasing CN/Ha, although for low pressures (=50 Torr or less) and high C 2/Ha ratios all films contain nanometer-scale diamond grains. Experimental Method for NCD Deposition on Condyle Simulants A series of TMJ condyle simulants were prepared in such a manner that they closely resemble commercially available implant condyles in dimension, and their surface preparation follows methods prescribed for actual implants. Concurrently a 2.54- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 38 cm-diameter by 2-mm-thick Ti-6A1-4V disk was coated using the same deposition conditions as the condyles in order to serve as an articulating surface during wear testing of the condyle simulants. Three Ti-6A1-4V condyle simulants were machined from Ti6AI-4V rod stock using physical dimensions measured from a standard commercial implant condyle. The simulants were polished to very low roughness (see Table II in Chapter 3) on a motorized rotary sample holder, and the wear test disk was polished against a motorized polishing wheel. Surface roughness was measured using a diamond stylus profilometer, as will be discussed in the next chapter. The simulants were pas sivated by oxidative acid immersion as per ASTM F86.53 a NCD film was deposited to a thickness of 3 pm on each simulant and the wear test disk, using the method described above for NCD deposition using the 6-kW MPCVD system. The wear test disk was placed on a flat molybdenum translation stage, and a few allowances were made for the geometry of the simulants during deposition. A custom-made molybdenum translation stage that held the simulants in a 3 mm-deep recessed well was fabricated and used for these trials. The simulants were tall enough that they could not be retracted below the level of the primary stage, so pressure and microwave power settings were chosen that maintained a temperature at the condyle tip below the 420°C lower cutoff of the optical pyrometer prior to sample height adjustment. Other than those adaptations, NCD deposi tion was performed by the same method as is standard for flat samples. The optical pyrometer was focused on the tips of the condyles and the center of the wear test disk for sample temperature measurement throughout the experiment. Deposition was performed using the standard hydrogen flow rate of 300 seem, CH» flow rate of 106 seem (21.2% of Hi flow rate), and nitrogen flow rate of 7.95 seem Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 39 (7.25% of CH» flow rate). The nitrogen flow rate, and subsequently the CN/Ha ratio, was higher than in previous TMJ simulant deposition trials carried out during develop ment of the process since an unknown fault in the MPCVD system resulted in a minimum vacuum of 0.050 Torr for the device. This value is roughly five times that previously achieved, and films deposited with this high base pressure repeatedly resulted in delami nated nanocrystalline films. Although it cannot be tested with the existing equipment, it is believed that the brittle films produced with high base pressure are the result of oxygen contamination and the preferential etching of the intergranular film phase that accompa nies it. Other r e s e a r c h e r s ^ have found that diamond film growth is minutely sensitive to oxygen concentration during deposition. The higher nitrogen flow rate used for deposi tion here seems to counteract oxygen etching by promoting the growth of the intergranu lar phase through an unusually large nitrogen addition. Over the course of three deposition procedures for the TMJ simulants, the ob served values for Ci/Ha and CN/Ha were 1.13±0.01 and 2.39+0.01, respectively. Deposition time was 89±8 minutes, giving an average growth rate of 2.0±0.2 pm/h. Pressure was maintained at 34 Torr, and microwave emission was consistently 0.73 kW for each trial. Deposition continued until each sample was coated to a thickness of 3 pm, as measured at the tip of the condyle using optical pyrometry apparent temperature oscillations as described p r e v i o u s l y . 5 5 CH* and N2 were simultaneously stopped, and the sample was monotonically cooled over a 12 minute period. These same deposition conditions were repeated for the flat plate used in wear testing. The parameters used here were optimized in a series of experiments on both flat plates and TMJ simulants in terms Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 40 of Him coalescence and adhesion, and preliminary hardness and modulus measurements taken from films deposited over a range of conditions. It is interesting to note that the growth rate seen on the TMJ simulants is 3.3 times that seen on flat plates in the same deposition conditions. The only difference between these two situations is the geometry of the sample, and so this must be the causative factor for the greater growth rate. Presumably, the geometry of the TMJ simulant must allow more diffusion and possibly convection within the plasma than the flat plate does, thereby promoting faster carbon mass transport to the growing film surface. The elec tromagnetic field generated from the hemispherical TMJ simulant may also play a role, although both hypotheses are perhaps best left to theorists to investigate and, at any rate, they lie outside the scope of this project. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 3 CHARACTERIZATION "Men show their characters in nothing more clearly than in what they think laughable. ” - Johann Wolfgang von Goethe Introduction The articulating surfaces of a prosthetic joint must articulate under cyclic load hundreds or thousands of times per day, in a chemical environment resembling warm seawater, without maintenance or external lubrication for decades on end. In order to be useful as wear resistant coatings for biomedical implants, deposited films must be able to stand up to this punishing application. They must be hard to stand up to erosive wear; smooth to prevent particulate generation from the mating component they articulate against; and well-adhered, since film delamination would effectively destroy the implant. A series of methods have been used to quantify the physical properties of diamond films and determine their suitability as an implant material. In order to assess the feasibility of applying NCD thin films to implant applica tions, the implant simulants must be tested under conditions that the final product would endure. Several main components must be tested. Primarily, NCD thin film deposition must be demonstrated over a curved substrate geometry similar to those found in im plants. Although commercial products are available that employ microcrystalline dia mond and diamond-like films over complex geometries such as cutting tools by hot- 41 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 42 filament chemical vapor deposition (HFCVD),31 to date MPCVD has not been used to deposit NCD thin films over samples closely resembling implants components. Also, the deposited film must meet a set of requirements based on the demands of implant use. First, the film must be smooth so as to minimize particulate generation during articulation in actual use. Second, the film must be reasonably hard so as to stand up to both sliding wear and potential third-body wear in use. In addition to simple hardness measurement, wear testing must be performed in a testing device relevant to the specific application. Third, the NCD itself must be tested for biocompatibility. Previous efforts at employing diamond and diamond-like thin films for biomedical purposes have included qualitative efforts to examine biocompatibility, 56 but no ASTM-standardized testing has yet been performed on the specific diamond film utilized in this research. Concurrently with characterization activities performed to quantitatively examine the requirements listed above, there must be quantitative analysis measures to understand the morphology of the NCD thin films so that the mechanisms at work can be better understood and optimized for future studies. The films must be quantitatively examined in terms of such features as diamond grain size, phase composition, interface composition, and general morphology. Pursuant to this goal the films will be characterized by Raman spectroscopy, diamond stylus profilometry, scanning electron microscopy (SEM), X-ray diffraction (XRD), Xray photoelectron spectroscopy (XPS), and by indentation to measure adhesion strength.32 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 43 Raman Spectra Ar* Laser 1340 cm Film Flake 2* C/3 c TMJ TMJ TMJ Disk (D c 1000 1200 1400 1600 1800 Wavenumber (cm*1) Figure 21. Raman spectra comparison. The five Raman spectra here are taken from a stress-free NCD flake, the center o f the wear test disk, and the tips o f the three TMJ simulants. Note the overall similarity and typical nanocrystalline diamond features at 1150,1340, and 1500 cm'1. Also note the zone center phonon line fo r the stress-free flake is found at 1332 cm'1, not 1340 cm' 1 like the other spectra. Film Phase Composition and Stress Distribution by Raman Spectroscopy Raman spectroscopy yields semi-quantitative compositional information about the phases present in a Raman-active material by measurement of energy loss in reflected laser light due to phonon excitation by the incident beam. This method is simple and non-destructive, and yields information about the phases present in a material. The Raman spectrometer used for this work is a Dilor XY Modular Spectrometer with a 514.2 nm Ar+ laser operated at 300 mW output power. The laser beam is directed through a microscope with a lOOx objective that focuses the beam onto the target area. The re flected beam is collected through the microscope and directed to an OMAvision CCD Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 44 Figure 22. Detail o f deconvolved Raman spectrum. Note the peak at 1342.5 cm' 1 (with the vertical line) used to inferfilm stress in this work. detector cooled by liquid nitrogen. The collected spectrum is displayed and analyzed using Dilor Easy4 software. Spectra collected from the tips of each TMJ condyle simulant (Figure 21) show the paired broad features at 11SO cm'1and around 1500 cm'1that are associated with the intergranular amorphous phase of NCD films.^? The shifted, broadened peak associated with the zone center phonon of diamond is found around 1340 cm'1. The additional feature at about 1SSO cm'1arises from an unknown source, although experiments here at UAB have shown that this peak appears in proportion with increasing CN/Ha ratio in the plasma from which it was deposited. AU spectra collected from the TMJ simulants and from the wear test disk exhibited virtually identical spectra, qualitatively indicating good compositional consistency between the samples. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 45 Other researchers have found that the diamond zone-center phonon shift can be used to make quantitative measurement of diamond film s t r e s s . 5 8 Unfortunately, the equations Ager III and Drory derived for stress measurement are only valid in the case of microcrystalline diamond films. These films contain a large volume fraction of bulk diamond that can freely propagate phonons over a comparatively long, undisturbed crystal lattice relative to the very small diamond lattices found in nanocrystalline dia mond films. The result is a pair of discrete peaks shifted to higher energies than the unstressed, triply degenerate phonon peak seen at 1332 cm'1in unstressed single-crystal diamond. In the case of nanocrystalline diamond, phonon confinement and peak broad ening effects arising from nanometer-scale diamond grains and a large volume fraction of amorphous material give rise to a broad feature around 1350 cm'1that defies reliable deconvolution. The edge of the feature itself, around 1340 cm 1, however, can be reliably fit to a Lorentzian peak (see Figure 22). Since this feature does arise from the diamond zone-center p h o n o n , 57,58 the difference between the deconvolved position of the 1340cm'1peak and the natural zone-center phonon position of 1332 cm'1is an expression of film stress. Included in Figure 21 is a Raman spectrum taken from a NCD film flake presumably free of biaxial stress. The diamond zone-center phonon for this spectrum is found at 1332 cm'1, which verifies the notion that shift of the 1340 cm'1 feature is due to latent film stress. Since the theoretical basis for this peak shift has not been mathemati cally derived, nor is it likely to be for the foreseeable future, this work will use the raw peak shift value as a function of film stress. While no absolute film stress magnitude can be found in this manner, any trend in film stress should be expressed in peak shift values. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 46 Diamond Zone Center Phonon Shift vs. Angie Along Radius of TMJ Condyle Tip ® — 11.5 ■ ® fe 0 O TMJ 5 • TMJ 6 □ TMJ 7 o 11 [ ! N «5 1 I 1°-5II 10 ■ 9.5 30 0 10 20 40 Angle Along Radius of TMJ Condyle Tip (Degrees) Figure 23. Phonon shift versus angle around simulant radius o f curvature. This graph shows that no clear stress concentration exists in the NCD films around the radius o f curvature o f the TMJ simulants. Data points have been stacked where equal fo r clarity. Inset image in lower right shows points on condyle simulants where Raman spectra were collected. Each TMJ simulant was placed on a Melles Griot Inc. 0-45° rocking cradle such that the tip of the simulant remained fixed in space as the cradle was translated. Raman spectra were collected at the tip of the simulant and then in 10° steps up to 40° off of vertical along the radius of curvature of the simulant wear surface. Each spectrum was then deconvoluted following background subtraction using Dilor LabSpec software, and the difference between the measured =1340 cm'1peak and 1332 cm'1was calculated. If a significant stress distribution were present in the film along the radius of curvature of the simulant, then the peak shift ought to show a marked trend. This is important because if there were a film stress gradient across the sample then the non-uniform wear an implant will experience in actual use may be more likely to delaminate the coating than if the stress distribution is uniform. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 47 When the peak shift values are plotted versus angle, no clear trend emerges (Figure 23). A linear plot of the data results in a slope of less than 0.01, while the stan dard deviation of the entire data set is 0.58. These results indicate that no clear stress distribution gradient exists along the curvature of the TMJ simulants. Sample Phase Composition and Film Grain Size by X-Ray Diffraction (XRD) XRD can be configured for thin film analysis to provide qualitative phase compo sition data within about 1 mm of a sample surface. Incident beam angle 6 is restricted to 1-5°, and 20-scanning mode is used to retain a scan geometry whereby the incident x-ray beam is fixed relative to the sample at a grazing angle of incidence. Resulting scan data shows the presence of hexagonal close-packed (hep) titanium, facecentered cubic (fee) TiC, and diamond (111) and (220) peaks in NCD-deposited samples. No graphite (002) basal-plane diffraction can be detected in any sample in this body of work. This data is consistent with a nanocrystalline diamond film adhered to titanium with a significant layer of TiC between the film and the bulk substrate. Of interest is the comparative breadth of the diamond (111) peak relative to the other peaks in the spectrum. Peak broadening generally arises from one or more of three main sources at room temperature: divergence and scatter in the x-ray beam due to equipment optics, material strain, and small grain size. Peak broadening due to equip ment optics can be reliably measured by comparing a measured peak width to the width of a peak from a strain-free material with large grains (Bs). For this study the width of the (111) peak of an annealed disk of polycrystalline silicon was measured by XRD under the same measurement geometry and conditions as the NCD samples. For Gaussian Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 48 XRD Spectra Tm IDWi >> 1at c TMJ • 20 30 40 50 60 70 80 °20 Figure 24. XRD spectra from wear test disk and condyle simulant. Predominant phases include a-Ti (labeled Ti), TiC, and diamond. A minor 0-Ti contribution is seen in the wear test disk. peaks, broadening due to material properties only (B) is found by using the following formula: (i) Where Bm is the measured value of full width at half maximum intensity (FWHM) for the peak under investigation. Williamson and Hall^9 derived an equation for measuring both the grain size and material strain contributions to peak broadening from the value for B. Their treatment assumed that the material is under uniform triaxial stress, however, and for the case of an NCD thin film under biaxial compressive stress their assumptions do not apply. Williamson and Hall built their treatment atop the older equation by Scherrer 60 that assumes that no strain is present in the film. While this assumption is not true for thin films, it is commonly used to report thin film grain size and may be more Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 49 Table I: Mean Film Grain Size and XRD Data Wear Test Disk TMJ Simulant Diamond (111) 43.90 43.87 Peak Position (Degrees) FWHM (Degrees) 0.60 0.84 Integral Breadth (Degrees) 0.68 0.84 Mean Grain Size (nm) 30.4 21.1 accurate than Williamson and Hall’s triaxial stress model. The Scherrer equation is as follows: kA t = --------5 cos 6 (2 ) Where k is a constant that has been found to vary between 0.89 and 1.39, but is com monly given the value k=0.9.^1 X represents the x-ray wavelength used for the meas urement, in this case the Cu Ka emission at 0.154 nm. 0 is the angle in radians between the sample stage and the detector for the peak under investigation, which is 0.384 for the diamond (111) peak at 26=44°. t is the mean grain diameter in nm, or in Angstroms if that measurement is applied to X. XRD scan data were collected from the wear test disk and a representative TMJ simulant following the wear testing described in Chapter 4. For the wear test disk, data were collected through a range from 20=20° to 20=80°. The scan step size was set at 0.040°, the time per step was 4 s/step, and the incident beam angle 0 was fixed at 5°. For the TMJ simulant, the scan step size was set to 0.020°, incident angle 6 was lowered to 1 °, and time per step was set to 15 s/step due to the small sample volume present at the tip of the condyle. Additionally, the TMJ simulant was raised 100 pm higher than normal on the sample stage in order to collect data from a larger volume than just the very tip of the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 50 condyle. This adjustment is responsible for the small diamond peaks relative to the other peaks in the spectrum, since data was collected from what amounts to a ring of film material around a disk of substrate at the condyle tip. Collected data are presented in Table I, including the peak position and the FWHM value for the diamond (111) peak. Applying Equation 2 to the collected data yields a mean grain size of 30.4 nm for the wear test disk and 20.9 nm for the TMJ simulant. These values are in accordance with expectations, since the much faster growth rate seen during deposition on the TMJ simulants should produce a smaller grain size than the slower growth seen with the wear test disk. It is reasonable to expect that the smaller-grained film on the TMJ simulants should exhibit higher yield strength than the wear test disk, as per the Hall-Petch re la tio n sh ip .6 2 ,63 The collected spectra contain the expected phases of crystalline diamond, a titanium, and TiC. Also of note is the lack of graphite in the films, as seen in the lack of the basal plane (002) graphite diffraction peak at 20=26.2°. The wear test disk is seen to contain a detectable amount of 3 titanium. Prior to the wear test, no 3 titanium was seen in identical coated samples prepared from the same sheet of Ti-6A1-4V. Since the wear test did not expose the sample to temperatures greater than the ~850°C 3 transus (with allowance made for dissolved vanadium), the most likely explanation is that the 3 tita nium is the product of shear-induced martensitic transformation of small amounts of titanium at the wear test sites on the disk. No 3 titanium was seen in the TMJ simulant, but it is possible that the x-ray beam did not sample the test location on the simulant. This finding is not of crucial importance, but it stands to affirm the notion that bare Ti6A1-4V possesses poor wear properties.62 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 51 Elemental Composition Depth Profile by X-Ray Photoelectron Spectroscopy (XPS) A sample was prepared specifically for XPS depth profiling by growing an NCD film on a 16-mm by 1.5-mm disk of Ti-6A1-4V that had been passivated as per ASTM F86^3. The film was grown to a thickness of =400 nm in order to facilitate depth-profile sputtering under conditions of 1.00 kW microwave power; 35.0 Torr; and flow rates of 500 seem Hi, 100 seem CH4 , and 7 seem N2. Mean C2/Ha and CN/Ha ratios were 1.38±0.02 and 3.09±0.01, respectively, over the course of 23 minutes of deposition time. The abbreviated deposition time used may reduce the thickness of the interface region relative to a film deposited to a more typical thickness of 3 pm. However, since the continued growth of the interface region after film coalescence would require the ener getically unlikely breakdown of diamond from the film in order to progress, it is probably reasonable to assume that the interface present for a 3-pm-thick film would not be drastically thicker from that measured here. The device used for this measurement is a Kratos AXIS 165 Multi-technique Electron Spectrometer. The electron energy analyzer is a 165-mm mean radius concentric hemispherical analyzer equipped with an 8 -channel detector and a monochromatic A1 x-ray source. The sample was etched in 4-minute increments using a 4 kV Ar+ ion beam with a surface current density of 1.4 pA between XPS scanning steps. Each scan step included oxygen Is, vanadium 2p, titanium 2p, nitrogen Is, bulk carbon, carbidic carbon, and aluminum 2p specific scans. The meas urement was taken until the measured atomic concentration of bulk titanium reached 50%. Data is presented as atomic concentration of each species with respect to etch time. It must be noted that etch time does not exactly correlate with depth into the sample, as Ar* sputtering yield changes for each material encountered by the beam. Other sources Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 52 100 - DLC 80- c £ a OLC 6 0- C 8c O 0 .« E 1 40 - »- 0 5000 10000 15000 20000 25000 30000 Etch Time, sec Figure 25. Elemental concentration depth profile by XPS. Note that the three predominant elemtents in the interface region on the right side o f the graph include film carbon (labeled "DLC" here), Ti, and carbidic carbon. Notice that carbidic carbon maintatins a concentration roughly half that ofTi, indicating the presence o f TiC and metallic Ti. This profile represents a diffuse film/substrate boundary. o f error include surface layer mixing under bombardment of the Ar+ beam and stray XPS signal gathered from the edges of the sputtered zone as a result of the Gaussian shape of the sputtering beam. It should be pointed out that the three most prominent curves (titanium, bulk carbon, and carbidic carbon) all become noisier with increasing depth though the interface region as a result o f these effects. The collected data show the classic pattern for a diffuse interlayer boundary, with bulk carbon decreasing in proportion to TiC and metallic Ti with increasing depth. The proportion of carbidic carbon in the sample decreases, starting at the point where bulk Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 53 titanium overtakes bulk carbon as the primary species detected. The ratio of titanium to carbidic carbon is roughly 2:1 from the onset of the detection of titanium up to the point where it makes up about 30% of the total volume, indicating that metallic titanium is present along with TiC throughout the interface region. Oxygen is present in the interface region at a mean value of 3.0% for the last 8160 s of sputtering time. This finding indicates that the titanium surface is not completely etched bare during exposure to the hydrogen plasma during film deposition. It cannot be said from this data whether oxygen exists as a distinct TixOy phase or as a solute in the interface, although Perry et ap3 reports the presence of amorphous, non-stoichiometric titanium oxides at diamond film fracture surfaces. The fact that no titanium oxides are found in the XRD spectrum of this diamond film suggests that the oxides are amorphous, although the relatively small quantity present may fall below the detection limit of the XRD device. It is worth noting that nitrogen is detected in only trace quantity in the NCD film. This finding indicates that no CN is added to the growing diamond film during deposition and, by extension, that the carbon bound up in the CN molecule does not significantly participate in film growth. This data gives insight into the mechanism by which grain size is reduced in diamond films grown in the presence of nitrogen. The absence of nitrogen in the film indicates that grain size decreases as a result of some phenomenon other than incorporation of nitrogen as a substitutional solute. Other researchers have proposed that abstraction of terminal hydrogen by CN from the growth surface drives reconstruction of the growing crystallites.^ The possibility also exists that momentum transfer from comparatively massive C 2 and CN species in the plasma drive physical Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 54 reconstruction of the growth surface, and so to a certain extent crystal size scales in versely to plasma CN concentration, as shown by Catledge and V o h r a . 6 5 Since Zhou et a ]6 6 h a v e demonstrated that diamond thin films can be grown in the absence of hydrogen the second alternative appears to be the more likely of the two. Aluminum and vanadium were also measured as a function of film depth. Alumi num is seen to increase in concentration up to about 4.5%, and vanadium reaches about 1.6% in the last 4080 s of sputtering time. Normal alloy concentrations are 6% A1 and 4% V and the alloy elements make up -70% of the concentration in the last 4080 s of sputtering. By normalizing the detected concentrations to the "alloy" volume fraction of 0.7, aluminum concentration is about 6.4% of the alloy, and vanadium makes up 2.3%. Aluminum is enriched relative to the bulk titanium alloy, and vanadium is depleted. This finding is at odds with the findings of Perry et a l , 3 3 who reported complete depletion of both aluminum and vanadium at the interfacial layer. In the original Ti-6A1-4V alloy, aluminum performs as an alpha-phase stabilizer, and vanadium stabilizes the martensitic bcc b e t a - p h a s e . 6 2 Since the layer nearest the surface is entirely hep alpha phase as measured by XRD, the relative abundance of aluminum and scarcity of vanadium is reasonable. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 55 Morphology of Diamond Film Surface by Scanning Electron Microscope (SEM) Given that the wear behavior of materials for use in implant articulating surfaces is vitally important, the surface morphology of those films plays a critical role. Dowson et al were able to show that a very small increase in the roughness of the metal compo nent in an implant results in a very large increase in the production of wear particles from the polymer component of the implant.^? It is assumed here that the optimal bearing surface should have low values for both the mean and standard deviation of surface roughness measurements and be composed of compact material. Both long-wave and short-wave roughnesses are assumed to be detrimental, as are surface porosity, inclu sions, or protrusions of any sort. Following the NCD deposition procedure, all three TMJ simulants and the 2.54 cm-diameter disk that opposed the simulants in wear testing were examined by SEM for qualitative examination of surface properties. A Philips SEM 515 electron microscope was used for image collection, utilizing an Everhart-Thomley secondary electron detector and a solid-state, four-quadrant, inline backscattered electron d e t e c t o r . 6 8 The electron source was a LaB6 crystal with accelerating voltage (VACc) of 30 kV. Images were collected using Spirit software from Princeton Gamma-Tech. All samples were prepared by sonication in acetone for 20 minutes followed by a rinse in methanol, and they were mounted using adhesive, conductive carbon-impregnated tabs on 2.54 cm-diameter aluminum stubs. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 26. SEM image o f wear test disk surface. Note the scale marker in the lower right comer and the micron-scale moundfeatures labeled "A”here. See the text fo r description. Figure 26 shows the wear test disk as imaged at SOOOx using the secondary elec tron detector. Note the micron marker for scale. This field contains more surface fea tures than the overall film does on average and was chosen to illustrate the three major distinguishing features noted in the image. The mounded structure labeled "A" in the image stands proud from the surface, and Raman spectroscopy reveals that it and those like it are composed of the same material as the bulk film. Note that these features are commonly several microns in diameter, and their roughly hemispherical shape means they can stand several microns above the surrounding film. Periodically, small areas of a darker material can be seen on the surface, labeled "8" in the image. Raman spectros Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. copy reveals that these are areas of relatively low diamond content and higher amorphous carbon than the bulk film. Consequently, these areas should be softer than the surround ing film. Also of note is the fact that each of the mound features labeled "A" is associ ated with a band of the softer material, labeled "C". These features are expected to degrade the wear resistance of the material, as the mound features will accelerate particu late production, and the softer portions of film are prone to pitting.-^ R. Messier has proposed a m odel^’ 70 that seems to explain film mound features (see Figure 27). In his work with "analog" simulations, or simulations involving very large quantities of atoms, he developed a cone growth model for thin film growth. The model assumes that thin film growth begins from a distribution of starting points (nuclei) at a surface and progresses in a geometrically programmed manner. As a result, the modeled thin film growth is seen to grow as thin cones forming a common surface. As the model progresses and the simulated thin film "grows," selection rules define which cones will continue to evolve after intersection with neighboring cones. When it is $uj 3 St o 50.0 100.0 150.0 200.0 Figure 27. "Analog" simulation offilm growth showing effect o f seed size distribution. R. Messier et al produced simulations o f thin film growth showing expected morpholo gies o f thin film grown from randomly distributed, identical nuclei (left) andfilm grown from a Gaussian distribution o f particle size (right). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 58 assumed that all starting nuclei have identical properties, the final surface is uniform to within a small perturbation within the width of each cone. If the nuclei are given a Gaussian distribution of initial properties, however, the resulting model features a flat surface punctuated by the occasional large mound feature. The result is a modeled film that is remarkably similar in appearance to the Him seen in Figure 26. The diamond powder used to seed the sample prior to NCD deposition is composed of a Gaussian size distribution of diamond particles with a maximum size between 1 and 2 Jim in diameter, so the deposited dim morphology is in excellent agreement with Messier’s model. The problem with the mound features on the film, then, arises from insufficiencies in the seeding procedure. Figure 28 contains SEM images of the tips of the three TMJ simulants used in this research following film deposition and imaged under the same conditions as those used for the wear test disk above. The difference in morphology between the TMJ simulants and the wear test disk is quite striking, even though deposition conditions were identical for all samples (see Chapter 2). The films on the TMJ simulants contain a large propor tion of surface porosity and the overall appearance is that of a dendritic structure. The likely cause of this morphology is the very high growth rate seen for the TMJ simulants, which was 3.3x that of the flat wear disk. The resulting surface morphology is not optimal for wear resistance, since the thin dendrites are prone to fragmentation under load. Given the average dendrite diameter of around 1 Jim, it is probably likely that particles in the range of 10s to 100s of nanometers in size will be produced by the break down of the dendrites under load. Additionally, the very large dendrite surface area with many crevices present will degrade the electrochemical corrosion resistance of the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 59 Figure 28. SEM images o f simulant film surface. These images are from TMJ 5 (top), 6 (middle), and 7 (bottom) at the tip o f each condyle simulant. Note the dendritic, porous appearance o f each surface. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 60 film relative to a flat, smooth surface. A possible fix for this problem is simply lowering the Ci/Ha ratio of the plasma by V 3 .3 to match the growth rate to that of a disk and thus potentially producing a smooth film free of dendrites. However, since the Him nucleation rate ( N ) is directly proportional to the vapor pressure of depositing species,52 decreas ing C2 would effectively decrease the nucleation density and may result in non-coalesced film growth. This problem is not trivial, and future research should include a course of research to optimize Him properties over non-planar surfaces such as TMJ condyles. Overall, neither the disk surface nor the TMJ simulant surface is well-suited to wear resistance, and the corrosion properties of the simulant surface are likely to be poor relative to a flat surface. Recommendations for future work include revamping the seeding procedure to minimize nuclei size distribution and possibly lowering the Ci/Ha ratio for TMJ deposition to promote the development of a smooth film surface. Film Surface Roughness by Diamond-Stylus Profilometry A Tencor Alpha-Step 500 diamond stylus profilometer was used to measure sur face roughness both before and after deposition on each TMJ simulant and on the wear test disk. Findings from the wear test disk will be presented in the wear testing section in Chapter 4. The vertical measurement range was restrained at 13 pm, which effectively restricted surface roughness measurement to the very tip of the condyles where the curvature is minimal. Scan length was set at 1000 pm, and scan speed was 100 pm/s for a scan time of 10 s each. Horizontal resolution of the scans was set at 1.00 pm, vertical resolution was 0.1 nm, the applied stylus force was 2.3 mg, and a long-range cutoff filter Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 61 TMJ 5 TMJ 6 TMJ 7 Overall Table II: TMJ Simulant Surface Roughness Pre-Deposition Rg(nm) RP(nm) r a (nm) Stdev Stdev Mean Mean Stdev Mean 2.4 8.9 4.4 5.9 35.0 21.9 4.1 0.8 5.5 1.2 3.7 14.5 0.7 5.0 6.5 1.5 19.0 11.8 5.0 7.0 2.4 22.8 12£ L3 TMJ 5 TMJ 6 TMJ 7 Overall Post-Deposition Rg(nm) Ra (nm) Stdev Mean Mean Stdev 14.1 3.0 0.7 9.8 11.4 2.8 8.3 1.2 2.4 11.9 18.3 6.9 14.6 4.2 10.0 1.4 TMJ 5 TMJ 6 TMJ 7 Mean Change Rp (nm) Stdev Mean 73.0 63.6 52.2 47.5 86.7 122.6 82.6 65.9 Percent Change from Original Value 1 1RP Rq Stdev Mean Mean Stdev Stdev Mean 66 -71 58 -32 109 190 50 107 133 102 260 1184 138 182 360 243 545 635 100 10 110 79 262 429 of 25 pm was applied to the collected data. The long-wave cutoff is intended to remove the curvature of the condyle from consideration in long-wave roughness values. Each scan was analyzed in terms of the average roughness of the scan (RA), the root mean squared roughness (Rq), and peak roughness (Rp). Average roughness is defined as the mean value of all high and low deviations from an overall mean value of the surface line. Rq is the root mean squared value associated with RA, and the two values typically agree closely. Rp is the peak height of the scan, or simply the height of the highest spot as measured from the mean. Five scans were collected for all samples, and the average Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. values are reported in Table n. All questionable data in this data set were tested using Dixon’s Q-test^l and were found to be acceptable, at the 95% confidence level. First of all, the deposition process increased the Ra value by twice its pre-deposition value overall, with a mild increase in the standard deviation. This finding is in agreement with the =2x increase in Rq , with a subsequent increase in standard deviation. Taken together, these overall roughness measurements indicate that the surface roughness increases two fold during the deposition process, yet remains within 15 nm. Peak roughness increases significantly following deposition, and the standard de viation of that measurement also increases substantially. This indicates that randomly scattered, substantially prominent peaks are generated during the deposition procedure. The number of larger features on the surface is both larger and more irregular than those found prior to deposition. The prime candidate as the cause of this effect is the seeding procedure, which scatters solid diamond particles 1 to 2 microns in diameter across the sample surface. Since the total film thickness only amounts to 3 microns, it is reasonable to assume that much of the peak roughness of the deposited film is the result of diamond seeds. These seeds are assimilated into the growing film, and the irregularities formed in the NCD film as it grows over and around these solid lumps of material likely generate the features up to =80 nm tall found on the completed film. These findings indicate that film surface roughness could be significantly improved through refinement of the seeding technique. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 63 Film Hardness and Modulus by Nanoindentation Nanoindentation is essentially the nanometer-scale analogue of traditional inden tation testing for determining the hardness of a material. In place of a weight-driven constant load driving a macroscopic indenter tip, a nanoindenter uses a piezomechanical actuator to drive a Berkovich three-sided diamond indenter into the sample. Indenter tips are typically only around 100 nm in diameter, and so the volume of material under test is on the order of cubic microns or less. In place of measuring the hardness through peak indentation load as is found in macro-indentation methods such as Knoop, Rockwell, and Brale, the MTS Nanolndentor XP nanoindenter used in this research measures hardness and modulus hundreds of times during an indent loading cycle. Nanoindentation was performed using continuous stiffness mode (CSM) meas urement of hardness and modulus optimized for thin films. Since the films used in this study are very hard, indent depths were restrained to 500 nm to prevent damage to the indenter tip. This depth was chosen because indentation size effects (ISE)^- tend to dissipate for NCD films beyond a depth of about 150 nm, leaving 350 nm of film thick ness from which to derive useful hardness and modulus data. X- and y-axis surface find distance was restricted to 5 pm to allow for the curvature of the condyle surface, but the default value of 50 pm was retained for the flat wear test disk. A starting point was chosen on each simulant condyle near the extreme tip of the sample so that each indentwas approximately normal to the sample surface. For the simulants, the distance between indents was restricted to 5 pm, and 50 pm was used for the disk. All indents were performed following microscope-to-indenter calibration and CSM indentation of a fused silica standard to ensure that tip geometry constants were correct. The Poisson’s ratio of Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 64 rm rm m p ifm M n M n m n •T M J AMragal ♦ P i t t _______ 0 50 100 150 200 260 300 OtaptacamwM lata Surface (fan) 350 400 450 500 Figure 29. Simulant and wear test disk hardness by nanoindentation. Notice the low hardness values near the film surface, and the roughly 20-GPa difference between disk and simulantfilm hardness. the film is set as 0.1, since no measurement of this value is known to exist. NCD films are not as hard as natural diamond, which exhibits a Poisson's ratio of 0.07. This value was rounded up one decimal place to serve as an estimated value for NCD films. The three TMJ simulants are labeled here as TMJ 5,6, and 7. All three simulants were indented five times, and in each case four of the five indents yielded usable data. Hardness and modulus data for the wear test disk were calculated from three indents that yielded useful information. Figure 29 is a graph o f measured hardness versus indentation depth for the three simulant and one wear test sample sets. Figure 30 expresses modulus Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 65 »— — •——— ra n avooiaua py lawovioannDon 1400 • D isk ■TMJ A trag el 1200 1000 a 000 e. I i j | 600 400 200 0 0 50 1G0 160 200 250 300 350 400 460 500 DhutacMnmit Irto Surfac* (mm) Figure 30. Simulant and wear test disk hardness by nanoindentation. The decline in modulus for the wear test disk film may be partially due to sample geometry. data in the same format. The data indicates that the wear test disk surface is harder by about 20 GPa for the first 200 nm o f thickness. It is reasonable to state that the TMJ simulants articulated against a surface that was 20 GPa harder than their own surfaces during the wear test described in Chapter 4. The peak hardness value for the wear disk is about 80 GPa, or 80% o f the hardness o f natural diamond. The peak average value for the TMJ simulants is about 65 GPa, but that value is measured at around 250 nm into the film. Normally, data from the first -150 nm of film is suspect due to ISE effects but all three TMJ samples show relatively little scatter in this regime, so it is taken here as a real phenomenon. The slope of the curve is shallower than would be expected for a very Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 66 hard film, and there is relatively little deviation from the curve between the three sam ples. As seen in the SEM data earlier in this chapter, all three TMJ simulants exhibit a dendritic surface layer with lower hardness than a coherent film. It is surmised from this nanoindentation data that the lower boundary of the dendritic layer is seen in the hardness data as a hardness curve inflection change, and so the dendritic layer is about 125 nm thick. This observation is in agreement with the SEM imagery collected following the wear test (seen in Chapter 4), which shows a coherent, smooth Him surface in the wear tracks once the dendritic surface layer is wom away. Modulus data indicates that the surface of the disk possesses a higher value than the average value from the simulants over the first 100 nm of thickness, but this meas urement is subject to uncertainty due to ISE effects J - For thicknesses greater than about 250 nm, the disk actually possesses a lower average modulus than the TMJ simulant average. This effect may be the result of sample geometry. All disks coated using the Wavemat MPCVD system end up warped upward in a radially symmetric fashion with the highest point at the disk center. The magnitude is approximately in the 500 jim range. The sample is fixed to a mount during nanoindentation testing, but the possibility exists that the disk is free to flex through a few tens of microns while tested. This factor likely introduces an artificially low modulus value into the wear test disk measurement. Film Adhesion Strength Analysis by Indentation Testing An article by M.D. Drory and J.W. Hutchinson describes a method for determin ing energy-based interface toughness (fc) for diamond films on titanium substrates by Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 31. SEM image o f Brale C indent. Notice the film material inside the indent (inset), which appears to be adhered despite extreme plastic deformation at the indent site. in d e n ta tio n .^ , 73 Following their method, the disk used in wear testing was indented and measured for film adhesion toughness. The disk used for wear testing was indented ten times, using a Brale C conical in denter tip under a load of 150 kg. One of the ten indents resulted in a circular delamina tion around the indent. The delaminated indent was imaged using a Philips SEM 515 under the same imagery conditions described in the SEM section above (see Figure 31). The radii of the indent and the delaminated area were measured by applying Pythagoras’ Theorem to lines between x,y pixel coordinates collected from around the radii and the center of the indent on a digital image. Five values were taken for each radius, which were averaged to return the diameters of the indent and delamination. An R/a value of Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 68 2.7 was calculated from the radius of the delaminated region (R) and the radius of the indent (a). The elastic energy per unit area for a thin film (Qo) is given as:^2 a, = 4 ^ (3) Where i) is Poisson’s ratio, a is latent film stress, t is film thickness, and E is the film Young’s modulus. Thickness is known to be 3 pm, and the Young’s modulus used here is 600 GPa, as taken from nanoindentation of the wear disk (see the nanoindentation section in this chapter). The value of Young’s modulus is an estimate, since modulus decreases with thickness through the film possibly due to substrate effects. The average value recorded for the wear test at 500 nm into the surface is ==600 GPa, and it is assumed here that this value is reasonable for the material at the film/substrate interface where delami nation occurs. A value of 0.1 was used as the Poisson’s ratio of nanocrystalline diamond. Since no method exists to accurately measure latent stress in NCD diamond, a value of 7 GPa (compressive) was taken from measurements of microcrystalline diamond thin films on Ti-6A1-4V a l l o y ^ S . The value for Qo calculated in this manner is 221 J/m2. This value represents the latent elastic strain present in the undisturbed NCD film following deposition. In a freely propagating crack through a stressed film, some quantity of latent energy will be released by the relaxation of the system. This value is the energy release rate (G), and the crack will arrest when G equals the Tc. To find Tc, G is normalized to a value of G that represents only the residual stress in the film if a free edge were intro duced (Go);32 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 69 Table III: Interface Toughness and __________ Relevant Values___________ R/a 2.7 fi, 221 J/m2 o/E 0.014 G0 121 J/m2 G/G,_______ 13 _________ rc 158 J/m2 (4) For this work, Go is calculated as 121 J/m2. Drory and Hutchinson calculated curves of G/Go by a finite element technique, and consulting these curves produces a G/Go value of =1.3. For the Go value above, G equals 158 J/m2. Given the assumption that delamina tion occurs with a narrow annular strip of film material trailing the crack tip, G = rc and so the interface toughness of the NCD film on the wear test disk is 158 J/m2. All relevant measurements and calculations are listed in Table m. By comparison with the measured interface toughness, the brittle interface of sili cate glass on copper^ possesses a Tc value of 2 J/m2, AI2O3 thin films on c o p p er^ have a Tc value of 150 J/m2, and microcrystalline diamond Film on Ti-6A1-4V exhibits a Tc value^ of 44 J/m2. It is worth restating that the value of Tc calculated here for the wear test disk represents the toughness of the only indent out of ten that delaminated, so in reality the actual film toughness is even higher than the calculated value. Equation 4 reveals an interesting insight into the nature and history of diamond thin film deposition. Historically, inadequate film adhesion has been cited as one of the major reasons why diamond thin film products have not been commercialized. Equation 4 shows that G, and thus Tc, is proportional to twice the Young’s modulus of the film, which typically scales with film hardness. In reading the literature, it appears that most Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 70 diamond researchers sought to maximize film hardness and modulus, and so adhesion was compromised. The Young’s modulus values of 500 to 600 GPa seen here may represent a critical value in terms of film adhesion for the NCD/Ti-6AI-4V system, and it definitely plays a critical role in establishing the excellent adhesion properties of the material. Assessment of Characterization Findings The physical, chemical, and morphological attributes of the NCD films deposited on TMJ simulants have been characterized. The results of these examinations provide a complete description of the composition of NCD thin films as deposited on the flat wear test disk and the TMJ condyle simulants as they are produced using the equipment and methods described in the deposition section. As confirmed by Raman spectroscopy, the NCD films produced in this experi ment exhibit a phase composition comparable to that of NCD films reported previ ously.^ The =1550 cm' 1 feature of unknown origin is somewhat more intense than that seen elsewhere, but this is in accordance with previous experiments showing a relation ship between high nitrogen flow rates and the ~1550 cm ' 1 feature. Since a widely ac cepted explanation for the origin of this peak does not exist, no reliable conclusions can be drawn from this finding. Measurements of the diamond zone-center phonon shift show that no appreciable stress gradient exists in the thin film along the radius of curva ture of the TMJ simulants. This finding is beneficial for the indented application of the NCD film in its role as a protective film, as any pre-existing stress gradient would likely make the film more prone to spallation and delamination under load. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 71 XRD data indicates that the films found on both the wear test disk and the TMJ simulants bear a typical composition for NCD films. Crystalline diamond, a titanium, and TiC were found on each sample. The diamond peaks were considerably broadened, and application of the Scherrer equation to the FWHM values of those peaks indicates that the wear test disk bears a mean diamond crystallite size of 30.4 nm, and the TMJ simulant measured contains diamond crystallites with a mean grain size of 20.9 nm. These values are in line with expectations given the much faster growth rate of the films on the TMJ simulants, which would dampen grain size overall. From this finding, it is believed that the film on the simulants will exhibit a higher yield stress than the film on the wear test. Also, 0 titanium was detected in the wear test disk but not in the TMJ simulant. If the 0 titanium is a result of local shear induced in the wear test, then the geometry of the test may have allowed similar 0 titanium to go unnoticed in the TMJ simulant. This finding has no real ramifications for the properties of the film or system in general, as it is already known that Ti-6A1-4V exhibits poor wear p r o p e r t i e s . ^ XPS depth profiling of a NCD-coated Ti-6A1-4V sample revealed a thick diffu sion boundary between the diamond film and the bulk metal. A diffuse boundary is usually associated with well-adhered f i l m s , 52 and indeed adhesion testing indicates that these NCD films are very well adhered. No nitrogen was incorporated in the film to within 1% atomic concentration, even though a large proportion of CN was present in the plasma. Oxygen is found within the diffuse boundary region, but only to a few percent, and XRD does not detect a discrete Ti0 2 phase in deposited samples. Vanadium and aluminum are found in the diffuse boundary region at concentrations that are within the expected limits. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 72 Film morphology as examined by SEM indicates that the films are poorly suited to biomedical wear-related applications. Raised mound features composed of film material are common on the disk, and the films deposited on the simulants bear a den dritic structure at the surface with a large porosity volume fraction. In the case of the disk, the raised mounds can be expected to generate an unacceptable volume of wear debris during articulation.26 The dendritic nature of the films coating the TMJ simulants is expected to initially generate a large volume of wear debris in the 10 - to 1000 -nm range as the brittle dendrites crumble under shearing load. Nanoindentation data from the TMJ films implies that the dendrites give way to a coherent film at a depth of about 150 nm into the surface. Once that volume of material is wom away, SEM indicates that a smooth wear surface is available for articulation. However, the large volume of wear debris generated by breakdown of the dendrites should continue to degrade the films by third-body w ear.26 it is believed that the mound features found on the disk are the result of the Gaussian distribution of diamond seeds left on the surface by the ultra sonic/mechanical abrasion seeding method used here. This opens the possibility that a method that produces a more even distribution of seed sizes, such as bias-enhanced nucleation ( B E N ) 7 5 may alleviate this shortcoming. Preventing the formation of den drites at the surface of the TMJ simulants is a more complicated problem that will require a course of research on its own. Stylus profilometry shows that the NCD coating process increases short-wave roughness by a factor of two overall, and that peak roughness increases by a very signifi cant 260% relative to the initial value. It is thought that the seeding process contributes strongly to this discrepancy, given that the seed diamonds are very large in proportion to Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 73 the film crystallites and to the thickness of the Him itself. Optimization of the seeding technique may reduce film roughness and thereby improve the wear properties of the coated material. Nanoindentation data indicates that the TMJ simulants exhibit a mean hardness of about 65 GPa. The data also shows that the surface is softer than the bulk film, although this finding is subject to some uncertainty due to ISE e f f e c t s ^ . Taken with the SEM images collected after the wear tests (in Chapter 4), however, the cross section of the film becomes clear. The outermost =125 nm of film surface is composed of a dendritic structure composed of film material that crumbles easily under load. The bulk film past that point is composed of solid film material with hardness of around 65 GPa and modulus of around 600 GPa. The wear test disk is at least 20 GPa harder than the TMJ simulants at all points near the surface. This finding is critical for understanding the results of the wear test presented in the next chapter. The modulus values recorded for the wear test disk are surprisingly low, and are likely artificial due to flex of the sample while testing was underway. Attempts to measure film adhesion strength by conical Brale indentation as per the method of Drory and H u tc h in so n 3 2 ,73 resulted in only a single delamination event out of ten indents. Adhesion measurements taken from that delamination are considered to be a low value for the true adhesion strength of the film for this reason. Values calculated from measurements of the delaminated zone around the indent include 221 J/m2 for the elastic energy per unit area in the adhered film, and the interface toughness value Tc equals 158 J/m2. This value compares favorably with other ceramic/metal Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 74 systems and is at least 3.5x the value reported for microcrystalline diamond films on Ti6A1-4V.32 Overall, then, the story of this material is a NCD thin film containing nanometerscale diamond grains with a hardness value of 63±10 GPa and a modulus of 670±200 GPa at the surface. The interface between the thin film and the sample is a diffuse boundary with an adhesion toughness of more than 138 J/m2, which indicates that the film is very well adhered to the bulk alloy. The surface of the film, however, is not as smooth as would be optimal for wear resistance properties as a biomedical device. It is believed that improvements to the seeding process and further research to control the surface properties of NCD films deposited on curved surfaces may overcome this prob lem. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 4 PERFORMANCE ANALYSIS "For a successful technology, reality must take precedence over public re lations, fo r Nature cannot be fooled. " - Richard Feynman Introduction Performance analysis differs from characterization in that the measurements taken here are functions of the interactions of the NCD/Ti-6A1-4V with its environment, as opposed to measuring the properties of the material itself. The following measurements are descriptions of how the material system will behave in its projected application and so are especially important for the biomedical aspect of this study. In order to be useful as an implant material, the NCD condyle system must exhibit acceptable biocompatibility and wear resistance performance. Biocompatibility, as defined by Williams,76 is "the quality of not having toxic or injurious effects on biological systems." In the case of implants, wear debris generation^, 20, 2 1 , 7 7 ,7 8 ^ surface c o r r o s i o n 2 9 , 7 9 -8 1 ^ the primary contributors to degradation of the biological system. The following tests assess the electrochemical corrosion and wear resistance properties of NCD coatings used in this research. 75 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 76 Electrochemical Analysis Assessment o f NCD Biocompatibility Two methods are used here to assess the electrochemical behavior of NCD films in relevant media. Electrochemical impedance spectroscopy (EIS) is used to examine the resistance and capacitance qualities of the deposited film, and potentiodynamic polariza tion is used to measure the corrosion rate, relative nobility, and passivation behavior of a film under applied voltage. Both methods used in concert provide a thorough description of the corrosion behavior of the system. Polarization Resistance Determination by EIS EIS uses a low-amplitude AC waveform to determine the polarization resistance for a sample surface, which is a fundamental physical property of the protective layer. The system is a three-electrode system; the sample makes up the working electrode and is cycled in voltage potential relative to a passive, Iarge-area counter electrode. A reference electrode is used to sample the electromagnetic field emanating from the sample in order to measure the voltage and current across the face of the sample. The reference electrode is a closed system that interacts with the larger system via a salt bridge, and the electrode itself is usually a calomel or Ag/Ag+ electrode. The cyclic potential is varied in fre quency, and impedance data of the system is collected. The collected data can be ex pressed in a number of graph types, but for the purposes of this experiment the Nyquist plot format will be used, where the imaginary component of film impedance is graphed vs. the real component of impedance. When plotted in this manner, the y=0 intercepts correspond to resistance values, and the maximum value along each curve corresponds to Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 77 the maximum angular frequency ((Um a x ) of the measured impedance. Film capacitance is found by the following equation: Where Ri is the second y=0 intercept in the semicircular data fit. The initial y=0 intercept for a single semicircle is R q , or the inherent resistance of the experimental setup. Subsequently, Ri minus the value of the previous intercept constitutes the polarization resistance value for each material/electrolyte interface in the sample system. These resistance values are measures of the polarization resistance of the material, which are functions of the work required to perform a charge transfer reaction at the sample surface. Large values of polarization resistance indicate a material that resists 1 .6 e6 Passivated Ti-6AUV 8 e5 NCD-Coated TI-6AI-4V 4 e5 0 1e 6 3 e6 2 e6 4 e6 5 e6 REAL Figure 32. Collected EIS data fo r polished, passivated Ti-6Al-4V and NCD-coated samples. Note the single curvature fo r the bare metal and the pair o f curves present in the NCD-coated sample data. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 78 electrochemical interaction with the environment, while materials with low polarization resistance values tend to corrode freely or at least conduct electricity across the mate rial/electrolyte interface. Six 7-mm-diameter, 1-mm-thick cylindrical samples of Ti-6A1-4V alloy were ma chined and polished to approximately 20 nm RMS surface roughness. All six samples were acid-passivated as per ASTM F86^3 standard for oxidative passivation of Ti-6A14V for implant purposes. Three samples were then coated with nanocrystalline diamond thin film by MPCVD in a custom 1.2-kW system built by Y.K. Vohra and T.G. M c C a u Ie y 3 5 at UAB. Gas flow rates were 500 seem H 2 , 88 seem CH4 and 8.8 seem N2.65 OES data were collected but were unreliable, due to the presence of some signal attenuating agent, possibly a C-N-based polymer, on the inside of the MPCVD system windows. NCD films were deposited to a thickness of 3 pm. A Princeton Applied Research (PAR) potentiostat/galvanostat 263A and PAR Model 5210 lock-in amplifier running PAR PowerSuite software were used to collect data for this experiment. The test was performed in de-aerated Hank’s balanced salt solution at body temperature, and the working electrode was opposed to a pair of large area platinum counter electrodes and a Ag/AgCl reference electrode. Impedance data was collected over a frequency range of 1 mHz to 100 kHz and a excitation voltage of 5mV peak-to-peak for each sample. All data are represented with respect to the reference electrode. The data from^all three of the control and all three of the experimental samples were each averaged to produce a Nyquist plot of the two data sets. The curves produced in this manner were manually fit using a compass due to limitations in the software. Measured y-intercept points were used to calculate polarization resistance (Rp) values. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 79 Table IV: Electrochemical Impedance Spectroscopy Data 5.00E+06 9.42E+06 C (F) 2.12E-14 2.07E+06 1.24E+07 3.14E+06 7.85E+06 1.54E-13 1.03E-14 Rp (Q ) Passivated Ti-6AI-4V NCD-Coated Ti-6AI-4V 1 st Curve 2nd Curve g>m a x and capacitance values were calculated using measured values of (Umax for the two data sets. Table IV contains values for Rp, (Um a x . and C calculated from the data. The Nyquist plot data collected in this trial (Figure 31) show a pair of impedance curves for the NCD-coated sample. The first curve provides Rp and (Um a x values lower than those seen in the control sample, which is essentially a measurement of the proper ties of the adhered TiOj protective oxide layer. The second curve yields Rp and (Um a x values far in excess of those seen in the protective oxide. Apparently, the second curve is a measurement of the properties of the diamond phase present in the film, which would explain the extreme property values. The first curve is not unexpected, as several re searchers report that the film should be fairly c o n d u c t i v e . 8 2 , 8 3 When the literature is consulted, however, some confusion is introduced into the findings. Sakharova et al^4 and van de Lagemaat et al&5 have performed similar EIS studies for microcrystalline and homoepitaxial growth, respectively. Both groups report finding two impedance curves very similar to the data presented here. Interpretation becomes difficult, then, since if the first impedance curve is the result of a second phase, then no such curve should be seen in van de Lagemaat et al’s data using homoepitaxially-grown diamond. Two possibilities arise - either the homoepitaxial diamond contained grain boundaries that give rise to the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 80 first impedance curve seen in their data, or the curve results from an anomalous dielectric relaxation, as Sakharova asserts.^ In any case, the EIS data collected here is consistent with the known behavior of the NCD diamond film: a relatively unreactive diamond phase is asserted in the data, and yet the film is electrically conductive, as seen by the ability to easily image the film using SEM without a sputtered conductive layer. Potentiodynamic polarization data also assert that the film is passive and relatively noble, and yet electrically conductive to some extent. Corrosion Behavior Analysis by Potentiodynamic Polarization Potentiodynamic polarization is used to quantitatively express the corrosion behavior of a substance as a function of applied voltage. The equipment used consists of an electrochemical cell with the sample mounted in a polymer mount such that a known surface area is exposed to the electrolyte, two large-area counter electrodes, a reference electrode with a salt bridge extending to a point a few millimeters away from the sample, and a source of Ni to de-gas the electrolyte. A potentiostat varies the sample potential relative to the counter electrodes over a prescribed range of voltage, and current across the sample is measured through the reference electrode. The collected data is expressed as a graph of potential between the sample and the counter electrode (i.e., cell potential) versus the log of the current across the sample times the log of the exposed sample area. A second electrode, commonly a calomel or Ag/Ag+electrode, is used to sample the electromagnetic field at the sample surface and so measure the potential between the sample and the reference electrode. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 81 Several points along the potentiodynamic polarization curve indicate changes in the electrochemical behavior of the sample surface and are important in characterizing the behavior of the sample. The point where the sample shifts between cathodic behavior and anodic behavior relative to the counter electrode is described as the open circuit potential ( E o c p )> since no current flows between the sample surface and the counter electrode. E ocp is also known as the corrosion potential or E c o r r . but here it will be referred to as E o c p - The current density associated with E o c p is the corrosion current Ic o r r . which is important in calculating the corrosion rate of the sample. The curve shape that accompanies E o c p >s commonly referred to as the Tafel region. Potentiodynamic polarization measurements are commonly begun with the sample potential slightly more 1600 jO E. 800.0 4000 NCO-CoatedTI-6AMV Passivated TI-6AMV T -6 Figure 33. Potentiodynamic polarization data from polished, passivated Ti-6Al-4V and NCD-coated samples. Notice the higher, more noble open-circuit potential o f the coated samples and lower associated current density. Also note the lack o f any pronounced oxidation or passivation behaviorfo r either sample set up to 1600 mV. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 82 negative than E ocp. since no phenomena of interest for this test occur below this value. The open circuit potential is a measure of how noble the sample is relative to platinum; the more noble the sample, the more current necessary to begin electrochemical break down of the sample. At a slightly more positive current density than the open circuit potential, the sample current density typically reaches a maximum known as the primary passivation potential (Epp). This is the point where a protective, adherent oxide forms and current density drops with increasing cell potential. Cell potential typically drops from Epp for a range of values determined by how adherent and protective the oxide coating is, up to a point where the oxide or other protective coating begins to break down. Current density increases at this point, which is known as the breakdown potential (E bd )Six samples identical to those used in the EIS experiment were prepared and likewise tested in de-aerated Hank’s balanced salt solution at body temperature. Each sample was subjected to potentiodynamic polarization using a PAR potentiostat/galvanostat Model 273 running PerkinElmer SoftCorr III software standardized to ASTM G5 8 6 The working electrodes were opposed to a pair of large area platinum counter electrodes and a Ag/AgCl reference electrode. The working electrode potential was cycled from at least 100 mV more negative than the open circuit potential up to 1600 mV relative to the reference electrode. E ocp and I c o r r values were measured using Tafel fits of each trial around the open circuit potential. Corrosion rate (CR) data was calcu lated for Ti-6A1-4V using a derivative of Faraday's Law: C R = Ico-«« W A VpF Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (6 ) 83 Table V: Potentiodynamic Polarization Data Corrosion Rate IcORR (mm/yr) (nA/cm2) Passivated Ti-6A1-4V 449.2E-6 +/- 269.9E-6 27.06+/-16.26 3.81+/- 5.29 NCD-Coated Ti-6A1-4V 63.2E-6 +/- 87.8E-6 94+/- 0.79 24.8E-6 +/- 13.1E-6 Wrought Co-Cr-Mo Eocp (mV) -93.4 +/-19.2 153.6+/-10.1 -171 +/- 4.7 Where A is area, W is atomic weight, V is valency, p is density, and F is Faraday’s constant (96,500 C/mol). Table V shows the calculated values of CR, Ic o r r , and E o c p for the coated and control groups. Ti-6AI-4V is considered an excellent material for biomedical applications in terms of corrosion resistance due to its inherent protective oxide 7^, 81, 87 xhis feature is typified in Figure 33, where the uncoated material shows no sign of electrochemical breakdown behavior all the way up to a potential of 1600 mV, indicating a material with a stable, well-adhered oxide coating that protects the bulk material from chemical attack. We see that, with the addition of the NCD coating, the material becomes more noble, as seen in the more positive open circuit potential ( E o c p ) of the coated samples. From this data it is seen that the addition of a NCD film decreases the CR of the system approxi mately seven-fold. A one-way ANOVA test performed at P=0.05 indicates that the mean CR values for NCD-coated Ti-6A1-4V and wrought Co-Cr-Mo are statistically indistin guishable. The mean CR of wrought Co-Cr-Mo was taken from a separate experiment performed with the same apparatus and under the same conditions as this e x p e r i m e n t . 8 8 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 84 Assessment o f Electrochemistry Findings The data indicates that, from an electrochemical corrosion standpoint, NCDcoated Ti-6A1-4V is a major improvement upon passivated Ti-6A1-4V to the point that it could serve as a replacement for wrought Co-Cr-Mo as an implant material. The addition of an NCD coating lowers the CR of the system seven-fold over polished, passivated Ti6A1-4V and remains well adhered up to 1600 mV relative to the reference electrode. EIS data further reveals the behavior of the coating system by showing the dual-curve Nyquist behavior typically seen in diamond films, with a maximum polarization resistance greater than that of the native titanium oxide. Wear Resistance Assessment of NCD-Coated TMJ by Mandibular Movement Simulator The three NCD-coated TMJ simulants described in the Deposition section were evaluated for their wear resistance properties using a custom-built mandibular movement simulator (MMS), shown in Figure 34.89 This device is designed to accurately repro duce the normal articulation of a human jaw and is adapted to mount TMJ simulants in place of the natural TMJ anatomy. The MMS is a fiberglass reproduction of an actual human skull and so it closely mimics the mechanical movement and anatomical structure of a typical patient. The test is performed using a cyclic chewing motion of the mandible, with maximum chewing forces measured at the teeth of 200 N during testing. MMC testing was performed by Dr. Camillo Machado of the UAB Prosthodontics & Biomate rials Department. For this test, each of the three NCD-coated TMJ simulants was mounted in the anatomically correct temporomandibular position, opposite the 2.34 cmdiameter NCD-coated Ti-6A1-4V disk described earlier. Each TMJ simulant was sub Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 34. Mandibular movement simulator (MMS). The MMS (left image) is based on a composite replica o f a human skull to replicate the natural motion o f the TMJ. The image on the right shows detail o f a TMJ condyle simulant and wear test disk in position and prior to fixation. jected to 500,000 loaded cycles in the MMS, which is equivalent to 4.4 yr of typical use.The conditions of this test are considered to be more extreme than those experienced by an actual implant because in the test the TMJ articulates against a flat plate in a point wear mode. An actual implant would articulate against a conformal fossa component that would increase the articulating surface area, thereby decreasing the pressure at any point of the fllm in contact. Eventually, an NCD-coated fossa component will be fabricated to produce a more realistic TMJ couple, and this wear test will be reproduced on that couple. Coating a fossa component is a research project in its own right, with its own set o f challenges. This test, however, will provide wear data that is critical for the develop ment o f a fossa component that is optimized for articulation against an NCD-coated condyle. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 86 Concurrently with stylus profilometry measurements of the TMJ condyle simu lants, the surface roughness of the wear test disk was measured. The measured RAwas 10.8±3.3nm, Rq was 15.3±5.3 nm, and Rp measured 55.5±35.9 nm over five measure ments. Ra for the disk was within 8% of the mean value of RAfor the simulants. Like wise, Rq was within 4.8% of the mean value for the simulants, and Rp showed a 33% improvement over the mean value for the simulants. This last value is somewhat surpris ing considering the mound stuctures seen on the disk surface but not on the TMJ simu lants in SEM imagery (see Chapter 3). Two of the three Rp values for the disk are more than twice the average value of the other three, so apparently only two profilometry scans out of five encountered the mound features. This seems to be an instance where the average value of these measurements does not adequately describe the true condition of the material. Following wear testing on the MMS, each TMJ condyle and the disk were imaged using SEM under conditions identical to those described in the SEM imagery section in Chapter 3. Images were collected at lOOx of the wear test track on the TMJ simulant and the matching track on the wear test disk using both the secondary electron and then the backscattered electron detectors. Images were then taken of the wear track edge at 1490x for both the TMJ simulant and the disk. The images are presented with each condyle grouped with the wear track it articulated against on the disk. These images tell the story not only of the wear test, but also of the surface morphology of both the simulants and the disk. Figures 35,41, and 47 show the wear test track on the TMJ simulants. The lightcolored circle in the center of the images is the wear track itself, which is composed of Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 87 three concentric zones for Figures 35 and 41. Figure 47 does not contain zones "A" and "B" because the film on this sample did not wear through. The innermost zone (marked "A" on the images) is bare Ti-6A1-4V where the film has worn through and exposed the substrate. Notice the large scratches in this region, on the order of several microns wide. The next zone away from the center of the wear track, labeled "B" in this image, is the TiC-containing interface layer between the film and the bulk alloy. This region resisted some of the more severe scratching due to the comparative hardness of TiC relative to Ti6A1-4V. The film material, "C", is the next zone away from the center of the wear track. Note that the film has resisted gross delamination, and indeed examination of all of the wear tracks produces the same observation. The film obviously failed, but the failure mode was grinding wear and not large-scale spallation or delamination. Of particular interest are the areas in region "C" with a polished appearance, most notable in Figures 41 and 47. The polished areas are especially interesting because they imply that, for film surfaces where the mating surfaces are smooth enough not to gouge each other, the film self-polishes to a surface most likely conducive to good wear behavior. Note also the large, micron-scale scratches present in the film all the way around the wear track. These scratches indicate that there was prolonged contact between high points on the wear test disk and the film surface surrounding the wear track. These scratches are particularly striking evidence of the severe damage done to the NCD film on the condylar compo nents by the mounds present on the wear disk film surface. Further visual evidence of this damage is found in the feature labeled "D." The scratches at these locations were caused during either loading or unloading of the TMJ simulant into the wear-testing device, even though the simulant and disk surfaces must have made only a single pass Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 88 D 1 DO | j r n Figure 35. SEM image o f wear track on TMJ 5 condyle. The film has eroded, exposing the Ti alloy. Notice the scratches around the periphery o f the wear track due to moundfeatures on the test disk, as well as the loading or unloading scar in the lower left comer. Note also that the film has not delaminated. 100 pm Figure 36. BSE image o f same field as Figure 35. The lightcolored region is exposed titanium alloy. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 37. SEM image o f wear track on the disk opposing the TMJ 5 condyle. The wear track is larger than that seen on the condyle, but most o f the wear is concentrated at location "E". Wear debris has accumulated around the track. 1 00 p m Figure 38. BSE image o f same field as Figure 37. Although the film is thin at the wear track, the substrate is not exposed. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 90 Figure 39. Wear track edge on TMJ 5. Notice the standing film edge at upper left. The predominantly TiC interlayer region is seen in the center o f the image, and bulk Ti alloy is in the lower right. Note the gouging of the alloy. IBHH ■ ■ 1 Figure 40. Wear track edge on test disk opposing TMJ 5. Note the protruding mound features labeled "F. ” Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 91 across each other. All three images, and most notably Figure 35, contain small particles at the film surface. Closer examination using the SEM revealed that these objects do not appear to be attached to the surface. They are apparently debris that were generated in the wear test and came to rest on the film surface. Much of this debris is on the scale of microns in diameter. Figures 36,42, and 48 are complimentary images to the images described above. These images were collected from the same field as Figures 35,41, and 47 using the backscattered electron detector (BSE). Since the image formed using the BSE detector is based on the atomic number of the elements in the image,168 these images are essentially maps showing all the locations where the NCD film is missing from the surface. Figures 42 and 48 clearly show the areas of the wear track where the film has been removed. Figure 48 shows that the film in the wear track is thinner than the surrounding film, as displayed by the more massive titanium showing through the film there. This image shows conclusively that the film is thinner due to wear but that the substrate is not exposed. Figures 37,43, and 49 are lOOx images of the disk at each wear track. On each image, the label "E" indicates the center of the wear track as determined visually. One interesting aspect of the three wear tracks is the range of sizes they cover. Figures 37 and 43 feature wear tracks that are much larger than that seen in Figure 49. Since Figure 49 is also the couple to the only TMJ simulant that did not wear through the film, it is believed that contact between the bare alloy in the simulant and the film surface on the disk caused enough friction to translate the disk in its mount and enlarge the wear test track on the disk. It is likely that similar movement would occur in worn NCD-coated Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 92 implants in actual use, and such translation can be expected to contribute to loosening of the implant. Also of note is the relative lack of mound features in the wear tracks. Apparently the mound features either broke off or were ground away. In either event significant wear debris was generated. Figures 38,44, and SO are the same fields as Figures 37,43, and 49, respectively, but these images were collected using the BSE detector. Figures 38 and 50 indicate that neither of these films wore completely through to the substrate. The meandering track seen in Figure 43 reveals a single spallation event where the substrate is exposed. Overall, the wear tracks on the wear test disk exhibit less damage than that seen on the TMJ condyle simulants. There are two reasons for this. One, the wear track covers a larger area on two of the three tracks on the disk and so the total wear inflicted on any one area is less than on the simulant. Secondly, and perhaps more importantly, nanoin dentation data (see Chapter 3) indicates that the surface of the wear test disk is harder than the surfaces of the TMJ simulants. During the test, the majority of wear occurs on the softer TMJ simulant surface. This has important implications for design of a coated fossa component. In an actual implant, the fossa component contacts the condylar component over a larger fossa surface area than is represented in the disk used in this test. If the TMJ simulant were the harder of the two surfaces, then the majority of the wear while in actual use should occur on the fossa component. This may be beneficial, since it implies that simpler methods than MPCVD can be used for coating the fossa component. It may be possible to coat the condylar component with an optimized NCD film while the fossa is coated with a slightly softer material like tetrahedral amorphous carbon (ta-C). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 93 Figure 41. SEM image o f wear track on TMJ 6 condyle. Note the similarities to Figure 35, with the exception of a zone ("C") o f polishedfilm around the bare alloy. Film erosion is not as severe in this test as that seen fo r TMJ 5. 1 LHl [ j m Figure 42. BSE image o f same field as Figure 41. The lightcolored region is exposed titanium alloy. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 94 Figure 43. SEM image o f wear track on the disk opposing the TMJ 6 condyle. The wear track appears to have migrated during the test, as the track is not linear. Most o f the wear is concentrated at point "E." Figure 44. BSE image o f same field as Figure 43. A small piece offilm has delaminatedfrom the center o f the track. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 95 Figure 45. Wear track edge on TMJ 6 . The standing film edge is easily visible. Close inspection reveals a network o f cracks on the film surface. Large scratches seen here are apparently secondary damage suffered while removing the sample. Figure 46. Wear track edge on test disk opposing TMJ 6 . Note the protruding mound features labeled "F." Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 96 ta-C can be applied at low deposition temperatures relative to MPCVD and would be easier to implement on the complex fossa geometry. Figures 39,45, and 51 are 1490x images of the edge of the wear track area for each TMJ condyle simulant. Figure 51 shows a reasonably smooth articulation surface and a set of secondary damage scratches incurred during removal of the condyle from the MMS following the test. Figures 39 and 45 show galling,162 scratches sustained during unloading of the condyle from its mount after testing, and other damage to the exposed alloy surface. Of particular interest is the edge of the NCD film, which does not show glassy fracture or typical erosive wear but rather the abrupt shedding of blocks of mate rial from the exposed film edge. This mode of failure is especially deleterious in terms of biocompatibility because it generates micron-scale particles that are too large to be engulfed by phagocytes and expelled from the patient. Generation of larger particles such as these typically leads to the formation of scar tissue around the joint and a wider foreign-body reaction by the immune system. Large-scale wear debris generation is implicated in osteolytic degradation proximal to the implant with its attending discomfort and loss of implant f i x a t i o n . *8,20, 21,77 The final set of images, Figures 40,46, and 52, are 1490x images of the edge of the wear tracks on the wear test disk. Notice the mound features that are visible in cross section, having been worn down (labeled "F" on the images). These features are further evidence that these large surface features do considerable damage to the condyles. Barely visible in these three figures is a pattern of lighter-colored spots in the polished sections of the film. These spots may be vestiges of the original diamond seeds, in which Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 97 case the random scatter of the spots across the surface illustrates the inadequacy of the current seeding technique in terms of seeding uniformity. The images collected following the wear test show considerable wear to both sur faces of the articulation couple. The NCD coating on the condyle simulants failed on two of the three tests, and there is evidence of substantial wear debris generation. Further more, the end-stage breakdown of the film material seems to include the generation of micron-scale particles by crumbling at a free edge. The wear behavior of this material as it exists here indicates that it is not recommended for use in biomedical applications. A pair of caveats must be considered at this point. For one, much of the damage to the film surface seen in the previous images appears to result or be exacerbated by the mounds of film material present on the flat plate. It may be that these mounds can be significantly reduced or even eliminated through refinement of the sample seeding method. Secondly, the wear test used here likely represents more severe conditions than an actual implant would experience in use. This supposition is based on the notion that the MMS test is essentially a pin-on-disk test due to the condyle-on-flat plate geometry. An actual implant would articulate against a conformal fossa component, and so local stresses applied to the film would be lower by comparison. If both of these issues are adequately addressed by future research, it is reasonable to expect that NCD films on TMJ condylar components will perform adequately or even exceptionally well for biomedical applications. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 98 D C 1 0 0 (i m Figure 47. SEM image o f wear track on TMJ 7 condyle. The film is intact. Note the polished region o f the track itself in contrast to the porous surface of the rest o f the film. Figure 48. BSE image o f same field as Figure 47. The film is thin in the center of the track, but the substrate is not exposed. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 99 E 100 pm Figure 49. SEM image o f wear track on the disk opposing the TMJ 7 condyle. The wear track is small compared to the other two samples and relatively uniform in shape. 100 pm Figure 50. BSE image o f same field as Figure 49. The film is thin but the substrate is not exposed. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 51. Wear track edge on TMJ 7. Dark scratches seen here appear to be damage incurred during sample removal. The track is quite smooth in appearance. Figure 52. Wear track edge on test disk opposing TMJ 7. Note the protruding moundfeatures labeled "F. ” Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 101 In vivo Biocompatibility Assessment J. Lemons et al of the UAB Prosthodontics & Biomaterials department have performed a study to assess the biocompatibility of NCD-coated Ti-6A1-4V alloy in vivo using animal m odels.^ The work was performed concurrently with the other research described here, and the author interacted regularly with the researchers in the in vivo work and maintained an up-to-date understanding of the research while it was underway. This work is important to the TMJ study in that it shows quite clearly that the animal models used here respond favorably to implantation of NCD materials on Ti-6A1-4V. These results show that NCD materials exhibit excellent biocompatibility and so are well suited as an implant material. Eight 7-mm-diameter, 2-mm-thick disks of Ti-6A1-4V alloy were coated with NCD to a thickness of 3 pm using the custom 1.2-kW MPCVD system at U A B . 3 5 The samples were seeded ultrasonically using 1-2-pm diamond p o w d e r ^ b and coated using flow rates of 500 seem Hi, 88 seem CH», and 8.8 seem Ni. The uncoated side of each sample was then carefully polished to a mirror finish. Four New Zealand white rabbits were surgically implanted with the prepared samples, two to each rabbit. One sample was implanted into the tibial proximal condyle and one sample into the femoral distal condyle of the same leg on each rabbit. The rabbits were examined clinically and radiographically to ensure proper placement of the implant, and the rabbits were euthanized at times up to 8 weeks. The femoral and tibial condyles with implanted samples were excised and sectioned, five per sample, by the Exakt system. The Exakt system is a standard, commercially available process whereby the biological material is dehydrated with a series of alcohols and infused with xylenes to Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 102 Figure 53. Cross section image o f implanted NCD-coated disk. Living bone tissue has grown across both the NCD and Ti-6Al-4Vsurfaces. Also important is the lack of immune response in the marrow tissue adjoining the implant. improve monomer penetration into the sample. The sample is then soaked in poly methylmethacrylate (PMMA) under vacuum at reduced temperature to allow the methylmethacrylate monomer to saturate the sample. The sample is then warmed to room temperature to allow the polymer to set, cut into thin sections with a diamond-blade band saw, and polished on a polishing wheel. The sections were then observed and photo graphed by optical microscope. The retrieved samples showed no foreign body reactions, fibrous granulation tissue, or abnormal characteristics within the bone/marrow spaces. Both compact and trabecular bone are found directly associated with both the titanium alloy surface and the NCD-coated surface without the presence of fibrous tissue at the interface. Results were consistent throughout the retrieved samples. These findings show that NCD thin films perform at least as well as Ti-6A1-4V in terms of general and histological biocompatibil Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 103 ity. Given the excellent biocompatibility properties of Ti-6A1-4V,91» 92 this finding identifies NCD thin film as a material well suited to long-term use in the human body. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 5 CONCLUSIONS 7 hate quotation. Tell me what you know." - Ralph Waldo Emerson Nanocrystalline diamond films deposited by MPCVD and monitored using fixedH2 OES monitoring have been applied to a clinically relevant sample geometry and tested under a variety of conditions chosen to test the suitability of the material for biomedical implant applications. The deposited NCD films have been extensively characterized in terms of phase composition, diamond grain size, elemental composition and distribution, morphology, roughness, hardness, modulus, and film adhesion. NCD films have been tested to measure their tribological properties, electrochemical behavior, wear resistance, and finally their biocompatibility. What emerges is a coherent picture of the nature and performance of NCD films in biological settings, and an assessment of their suitability to this application. The deposited films are nanocrystalline in nature, with a significant intergranular amorphous carbon phase as is normally seen in this material. When deposited on the hemispherical wear surface of a TMJ condyle simulant, there is no evidence of apprecia ble stress gradients in the film along the radius of the condyle. The magnitude of the latent stress present in the film is not available, due to the inadequacy of modem testing techniques and theoretical understanding to quantify stress for this film material. How ever, we do know from indentation testing that the adhesion toughness of this 104 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 105 material is greater than 158 J/m2, which is represents an interface more than four times tougher than that found in microcrystalline diamond films on Ti-6A1-4V.32 Adhesion is vitally important for biomedical applications, as any significant film delamination would quickly destroy an implant. In terms of adhesion, these films are very well suited to use in biomedical implant articulation surfaces. NCD film roughness data represents a mixed bag of findings. On longer scales, it is seen from profilometry data that the simulant surface increases significantly following NCD thin-film deposition. The surface seeding procedure used in this work is identified as the culprit. It must be pointed out that the large mound features on the flat disk were seen to cause tremendous damage during the wear test, illustrating the necessity for low film roughness on all length scales. On shorter scales, film roughness is in the 10-nm range, which is probably acceptable, but given the strong dependence of particulate generation rate on film surface roughness**? any implant product produced with NCD thin films would benefit from refinement in this area. Overall, this research identifies ultrasonic sample seeding as a technique in need of improvement or replacement. Wear testing results for NCD thin films illustrate the poor wear behavior of this material. In two of the three tests the film failed, exposing the titanium surface. The porous film surface found on the simulants wore away in all areas where it was in contact with the disk, but the surface exposed in this process was quite smooth and appeared to present a viable bearing surface. Once a critical thickness was reached, however, the film failed in a crumbling manner. Micron-scale blocks of film were removed from the wear track at this point, and the titanium substrate was placed in direct contact with the articu lation couple. The samples that reached this point suffered severe wear-related damage, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 106 and if they were actually in use as implants the patient could expect to be harmed by the generated debris. The wear test contains several shortcomings that must be addressed. For one, the flat wear test disk made a less than ideal articulation surface. Actual TMJ implants are designed to maximize wear track area by articulating the condylar component against a semi-conformal fossa component. The geometry of the test used here was closer to a pinon-disk geometry, which represents more extreme wear conditions than an actual implant would face. In this regard, a more reasonable wear test will have to await the develop ment of a more realistic fossa component, complete with suitable hard coating. A second consideration is that of the difference in film hardness between the TMJ simulants and the wear test disk, which exceeded 20 GPa near the film surface. It is seen in the SEM imagery of the wear test tracks that the disk suffered less damage than the condyle simulants, partially due to this discrepancy. In an optimized implant, the condyle would possess a harder film than the fossa, since the wear track of the condylar component will most likely be smaller than the adjoining track on the fossa. If these two concerns were adequately addressed, this wear test would become a more relevant absolute test of wear resistance for this material and TMJ implants in general. Taken together, the electrochemical and in vivo studies performed here illustrate the superlative biocompatibility properties of NCD thin films. Electrochemical testing reveals a seven-fold decrease in the corrosion rate of NCD-coated Ti-6A1-4V as opposed to polished, passivated Ti-6A1-4V commonly used in biomedical applications. These tests were performed in a relevant biochemical media and represent the behavior of the film under the extreme chemical conditions found in the human body. Overall, the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 107 properties of NCD-coated Ti-6A1-4V are excellent and show this material to be at least as electrochemically stable as Co-Cr-Mo. The results of the in vivo study further demon strate this behavior. NCD coatings display biochemical and histological behavior that is at least as good as Ti-6A1-4V, which is known to possess superior biocompatibility behavior. It can be stated that intact NCD coatings exhibit excellent biocompatibility properties and in this respect are well suited to use in biomedical applications. The formal hypothesis statement for this work specifies four criteria that the NCD film must meet satisfactorily in order for this material to be relevant as a wear-resistant coating for implant applications. The films must be wear-resistant, smooth, reliably adhered, and biocompatible. In two of those areas, namely adhesion and biocompatibil ity, NCD thin films prove to possess exceptional qualities. Of the remaining two, surface smoothness was problematic. Deposition on a flat coupon illustrated the random growth of micron-scale mound features across the surface, and profilometry of the condyle simulants indicated a degree of long-wave roughness that requires remediation. The remaining criterion, wear resistance, highlights the predominant shortcoming found in this material as produced in the manner described in this work. The dendritic morphol ogy of the TMJ simulant surface fractures easily under load, causing significant wear debris generation. This very hard wear debris would serve to drastically shorten the lifespan of the film in actual use, not to mention the biological trauma it would spawn in tissues proximal to an implant. Furthermore, once the film wears down to a critical thickness it tends to crumble in micron-scale blocks that damage the wear couple further, expose and then grind against the substrate surface, and inspire deleterious biological Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 108 reactions. On this basis, this material should not be utilized for wear-resistance applica tions in biomedical implants. On the basis of the evidence presented here, it is believed that the shortcomings highlighted above can be overcome in future work. As far as sample seeding goes, the BEN7^ technique may provide a relatively simple fix to this problem. Sheldon et al7^ have shown that film nuclei can be produced with exceptional uniformity in both size and distribution. The 6 kW MPCVD will require minor modification to bias the sample, and a course of research should be undertaken to quantify and optimize the nuclei size and distribution using this method. This work is the logical next step in NCD deposition for biomedical applications. Once the seeding problem has been overcome, the next step should be understanding and optimizing NCD film growth over the hemispherical condylar surface. This research has shown that control of plasma chemistry is not sufficient to control film morphology, but that substrate geometry must also be taken into consideration. A course of research should be performed to investigate film surface morphology with respect to varying methane flow rates, and with respect to nitrogen flow rates in the event that the results are inconclusive. A series of condyle simulants should be coated for several steps of methane flow, and the resulting films compared, primarily by nanoindentation for hardness and modulus data, and by SEM for surface morphology. It is suggested that methane flow be varied from 32 seem (V3 3 of 106 seem used in this work) up to some value near 106 seem, with nitrogen flow fixed at 7.5% of methane flow and other deposi tion parameters constant. Decreased Ci/Ha will result in lower growth rates, but concep tually should eliminate the porosity problem seen in this work. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 109 In order to gain a fundamental understanding of NCD film growth and morphol ogy under these conditions, a course of work should be undertaken to examine film growth on curved surfaces at different times in the growth process and different points along the curvature of the condyle. In this work, film growth behavior was largely inferred from the complete film and would benefit from closer study. Research into the temporal and spatial distribution of film growth over the curvature of the implant would be both novel and insightful. This research has accomplished the material goal of investigation of the properties of NCD films with respect to a biomedical application. This research is also important in that it illustrates both the benefits and difficulties of this system so as to direct the course of future research. There is both a great deal of work yet to be done, and a great deal of promise for this technology. It is the fervent hope of this researcher that the scientific understanding of this material system is advanced from this point, so that the many patients of dehabilitating joint disorders will again enjoy the liberty of natural movement. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LIST OF REFERENCES ASTM FI36-02 Standard Specification for Wrought Titanium-6 Aluminum-4 Vanadium ELI (Extra Low Interstitial) Alloy fo r Surgical Implant Applications (UNS R56401), Vol. 13.01 (ASTM International, West Conshohocken, PA, 2002). ASTM FI472-02 Standard Specification for Wrought Titanium-6 Aluminum-4 Vanadium Alloy for Surgical Implant Applications (UNS R56400), Vol. 13.01 (ASTM International, West Conshohocken, PA, 2002). ASTM F75-01 Standard Specification for Cobalt-28 Chromium-6 Molybdenum Alloy Castings and Casting Alloy for Surgical Implants (UNS R30075), Vol. 13.01 (ASTM International, West Conshohocken, PA, 2001). ASTM F799-02 Standard Specification for Cobalt-28 Chromium-6 Molybdenum Alloy Forgings for Surgical Implants (UNS R31537, R31538, R31539), Vol. 13.01 (ASTM International, West Conshohocken, PA, 2002). ASTM FI537-00 Standard Specification fo r Wrought Cobalt-28 Chromium-6 Molybdenum Alloy for Surgical Implants (UNS R31537, UNS R31538, and UNS R31539), Vol. 13.01 (ASTM International, West Conshohocken, PA, 2000). D.W. Feigal, "TMJ Implants - A Consumer Informational Update," http://www.fda.gov/cdrh/consumer/tmjupdate.html (accessed 12 Oct 2002). TMD: Temporomandibular disorders, (NIH Publication No. 94-3487, Bethesda, MD, 1994). T. Gillespy in and M. Richardson, "TMJ Anatomy," cpmcnet.columbia.edu/dept/dental/Dental_Educational_Software/TMJ/TMJAnatomy.htm 1(accessed 20 Aug 2002). M.G. Fontenot, in Temporomandibular Disorders and Related Pain Conditions, Progress in Pain Research and Management, Vol. 4, edited by Sessle B.J., Dionne R.A. (IASP Press, Seattle, 1995) S. Saha, C. Campbell, A. Sarma, and R. Christiansen, Critical Reviews in Biomedical Engineering 28 (2000) p. 399. D. Christensen, TMJ Science 1 (2001) p. 9. 110 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. I ll B. Hall and P. Sheridan, "NIH Panel Makes Recommendations for Temporomandibular Disorders," http://www.hhs.gov/news/press/1996pres/960501.html (accessed 09/22/02). A. Catic and M. Naeije, J. Oral Rehabil. 26 (1999) p. 661. A. Isberg and P.L. Westesson, Oral Surg. Oral med. Oral Pathol. Oral Radiol Endod. (1998) p. 152. 8 6 Y.J. Chen, L.M. Gallo, and S. Palla, in Inti Assoc, for Dental Research (IADR, San Diego, CA, 2002). D.B. Gossi, L.M. Gallo and S. Palla, in Inti Assoc, for Dental Research (IADR, San Diego, CA, 2002). C.R. Wheeless, "Wheeless’Textbook of Orthopaedics," http://ortho-u.net/(accessed 09/20/02). M.J. Jasty, W.E. Floyd, A.L. Schiller, S.R. Goldring, and W.H. Harris, J. Bone Joint Surg. Am. 6 8 (1986) p. 912. R. Lappalainen, H. Heinonen, A. Anttila, and S. Santavirta, Diamond and Related Materials 7 (1998) p. 482. A.V. Lombardi, T.H. Mallory, B.K. Vaughn and P. Drouillard, J. Bone Joint Surg. Am. 71 (1989) p. 1337. H.J. Agins, N.W. Alcock, M. Bansal, E.A. Salvati, P.D. Wilson, P.M. Pellici, and P.G. Builogh, J. Bone Joint. Surg. Am. 70 (1988) p. 347. H.G. Wilert and M. Semlitsch, J. Biomed. Mater. Res. 11 (1977) p. 157. H.G. Wilert and M. Semlitsch, Sulzer Tech. Rev. 2 (1975) p. 119. R.L.C. Wu, A.K. Rai, A. Garscadden, P. Kee, H.D. Desai, and K. Miyoshi, J. Appl. Phxs. 72 (1992) p. 110. K. Miyoshi, R.L.C. Wu, A. Garscadden, P.N. Barnes, and H.E. Jackson, J. Appl. Phxs. 74 (1993) p. 4446. A.S. Litsky and M. Spector, in Orthopedic Basic Science, edited by S.R. Simon (Am. Acad, of Ortho. Surg., New York, NY, 1994) p. 447. C. Machado, W. Lacefield, S. Catledge and Y.K. Vohra, in Proceedings o f 28th Annual Meeting (Society for Biomaterials, Tampa, Florida, 2002) p. 104. ASTM F648-00 Standard Specification for Ultra-High-Molecular-Weight Polyethylene Powder and Fabricated Form for Surgical Implants, Vol. 13.01 (ASTM International, West Conshohocken, PA, 2000). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 112 A Review of Corrosion Behavior of Surgical Implant Alloys, (Princeton Applied Research, Oak Ridge, TN, 1999). R. Venugopalan, M.A. George, J.J. Weimer, and L.C. Lucas, Biomaterials 20 (1999) p. 1709. S.-S. Park and J.-Y. Lee, J. Appl. Phys. 69 (1990) p. 2618. M.D. Drory and J.W. Hutchinson, Science 263 (1994) p. 1753. S.S. Perry, J.W. Ager HI, G.A. Somoijai, R.J. McClelland, and M.D. Drory, J. Appl. Phys. 74 (1993) p. 7542. M.L. Terranova, M. Rossi, and G. Vitali, J. Appl. Phys. 80 (1996) p. 3552. T.G. McCauley, Synthesis and Characterization of Metastable Phases o f Carbon, Dissertation (University of Alabama at Birmingham, Birmingham,AL, 1996). C. Tang and D.C. Ingram, J. Appl. Phys. 78 (1995) p. 5745. M.N.R. Ashford, P.W. May, and C.A. Rego, Chemical Society Reviews 23 (1994) p. 21. L.R. Martin, American Ceramic Society Bulletin 71 (1992) p. 1419. D.M. Gruen, MRS Bulletin (1998) p. 32. B.R. Stoner, B.E. Williams, S.D. Wolter, K. Nishimura, and J.T. Glass, J. Mater. Res. 7 (1992) p. 257. J.C. Angus, F. Buck, M. Sunkara, T. Groth, C. Hayman, and R. Gat, MRS Bulletin 39 (1989) p. 38. Y. Shigesato, R.E. Boekenhauer, and B.W. Sheldon, Appl. Phys. Lett. 63 (1993) p. 314. T. Vandevalde, M. Nesladek, C. Quaeyhaegens, and L. Stals, Thin Solid Films 290*291 (1996) p. 143. T. Vandevalde, M. Nesladek, C. Quaeyhaegens, and L. Stals, Thin Solid Films 308-309 (1997) p. 154. T. Vandevalde, T.D. Wu, C. Quaeyhaegens, J. Vlekken, M. DDlieslaeger, and L. Stals, Thin Solid Films 340 (1999) p. 159. A.N. Goyette, J.E. Lawler, L.W. Anderson, D.M. Gruen, T.G. McCauley, D. Zhou, and A.R. Krauss, J. Phys. D: Appl. Phys. 31 (1998) p. 1975. A. Gicquel, M. Chenevier, K. Hassouni, A. Tserepi, and M. Dubus, J. Appl. Phvs. 83 (1998) p. 7504. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 113 F.G. Celii and J.E. Butler, Annu. Rev. Phys. Chem. 42 (1991) p. 643. R.J.H. Klein-Douwel, J.J.L. Spaanjaars and J.J. ter Meulen, in Applications o f Diamond Films and Related Materials: Third International Conference (1995) p. 413 P. Atkins, Physical Chemistry (W.H. Freeman and Co., New York, NY, 1994). K. Hassouni, T.A. Grotjohn, and A. Gicquel, J. Appl. Phys. 86 (1999) p. 134. M. Ohring, Materials Science o f Thin Films: Deposition and Structure (Academic Press, San Diego, CA, 2002). ASTM F86-01 Standard Practice for Surface Preparation and Marking o f Metallic Surgical Implants, Vol. 13.01 (ASTM International, West Conshocken, PA, 2001). Y. Liou, A. Inspektor, R. Weimer, D. Knight, and R. Messier, J. Mater. Res. 5 (1990) p. 2305. Z.L. Akkerman, Y. Song, Z. Yin, and F.W. Smith, Appl. Phys. Lett. 72 (1998) p. 903. S. Mitura, A. Mitura, P. Niedzielski, and P. Couvrat, Chaos, Solitons & Fractals 10 (1999) p. 2165. M. Yoshikawa, Y. Mori, H. Obata, M. Maegawa, G. Katagiri, H. Ishida, and A. Ishitani, Appl. Phys. Lett. 67 (1995) p. 694. J.W. Ager HI and M.D. Drory, Physical Review B 48 (1993) p. 2601. G.K. Williamson and W.H. Hall, Acta Metallurgica 1 (1953) p. 22. P. Scherrer, mNachrichten Gesellschafts Wissenschaftlich (Gottingen, 1918) p. 96 C. Suryanarayana and M.G. Norton, X-Ray Diffraction: A Practical Approach (Plenum Press, New York, NY, 1998). W.F. Smith, Structure and Properties of Engineering Alloys (McGraw-Hill, Inc., New York, NY, 1981). J.P. Hall and Z.Z. Petch, Can. J. Metallurgy 26 (1954) p. 254. A. Afzal, C.A. Rego, W. Ahmed, and R.I. Cherry, Diam. Relat. Mater. 7 (1998) p. 1033. S.A. Catledge and Y.K. Vohra, J. Appl. Phys. 86 (1999) p. 698. D. Zhou, T.G. McCauley, L.C. Qin, A.R. Krauss, and D.M. Gruen, J. Appl. Phys. 83 (1998) p. 540. D. Dowson, S. Taheri, and N.C. Wallbridge, Wear 119 (1987) p. 177. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 114 J.I. Goldstein, D.E. Newbury, P. Echlin, D.C. Joy, A.D. Romig Jr, C.E. Lyman, C. Fiori, and E. Lifshin, Scanning Electron Microscopy and X-Ray Microanalysis (Plenum Press, New York, NY, 1992). B. Yang, B.L. Waldon, R. Messier, and W.B. White, SPIE 821 (1987) p. 68. R. Messier, J. Vac. Sci. Technol. A4 (1986) p. 490. J.C. Miller and J.N. Miller, Statistics for Analytical Chemistry (Ellis Horwood PTR Prentice Hall, New York, NY, 1993). J. Gong, J. Wu, and Z. Guan, Journal o f the European Ceramic Society 19 (1999) p. 2625. M.D. Drory and J.W. Hutchinson, Proc. Roy. Soc. A 452 (1996) p. 2319. R.O. Ritchie, R.M. Cannon, B.J. Dalgleigh, R.H. Dauskardt, and J.M. McNaney, Materials Science and Engineering A 166 (1993) p. 221. B.W. Sheldon, R. Csencsits, J. Rankin, R.E. Boekenhauer, and Y. Shigesato, J. Appl. Phys. 75 (1994) p. 5001. D.F. Williams, The Williams Dictionary o f Biomaterials (Liverpool University Press, Liverpool, UK, 1999). J.T. Dearborn and W.H. Harris, J. Bone Joint Surg. Am. 82 (2000) p. 1146. S.T. Woolson and M.G. Murphy, J. Bone Joint Surg. Am. 77 (1995) p. 1311. M. Pourbaix, Biomat. 5 (1984) p. 122. A.F. vonRecum, Handbook of Biomaterials Evaluation (Taylor & Francis Inc., Philadelphia, PA, 1999). P. Kovacs, Trans. Soc. Biomat. 19 (1993) p. 272. M. Iwaki, S. Sato, K. Takahashi, and H. Sakairi, Nuclear Instruments and Methods 209/210 (1983) p. 1129. G.M. Swain, A.B. Anderson, and J.C. Angus, MRS Bulletin (1998) p. 56. A. Sakharova, L. Nyikos, and Y. Pleskov, Electrochemica Acta 37 (1992) p. 973. J. van de Lagemaat, D. Vanmaekelbergh, and J.J. Kelly, Journal o f Electroanalytical Chemistry 475 (1999) p. 139. ASTM G5-94(1999)el Standard Reference Test Method for Making Potentiostatic and Potentiodynamic Anodic Polarization Measurements, Vol. 03.02 (ASTM International, West Conshohocken, PA, 1999). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 115 R.J. Solar, S.R. Pollack, and E. Korostoff, J. Biomed. Mater. Res. 13 (1979) p. 217. S. Woodard, A Novel Metalloceramic Coating for Reduced UHMWPE Wear in Biomedical Devices, Master’s Thesis (University of Alabama at Birmingham, Birmingham, AL, 2002). C. Machado, M.C. Castaneda, J. Diaz, P. Pineros and L. Suarez, (unpublished manuscript). J. Lemons, P.J. Louis, and P. Beck, (unpublished). G.S. Leventhal, J. Bone Joint Surg. 33A (1951) p. 473. D.M. Brunette, Titanium in Medicine (Springer, Berlin, Germany, 2001). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. APPENDIX A COMPENDIUM OF LOW PRESSURE OES TRENDS FOR WAVEMAT 6-KW MPCVD SYSTEM Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 117 Ha Peak Area vs. %CH4 and %N. Fixed a t -500V P = 20 Torr, FP = 1 OOkW 100000 < D U- < <0 <d CL a 80000 60000 40000 20000 x o \° %CH< Z = 29.6(XY) - 1 5 .5 ^ + 121X2 + 354Y- 4910X + 94600 Figure A 1. Raw H a peak area versus CH4 and N2 flow rates at 20 Torr. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 118 Ha Peak Area vs. %CH4 and %N, Fixed a t -500V P = 35 Torr, FP = 1.00kW 1111 (0 0 80000 11 i t CO 0 Q. 25 40000 20000 X % c^ R r H/ u 250 Z = 40.3(XY) - 5 .0 3 ^ + 63.7X2 + 328Y- 2680X + 103000 Figure A2. Raw H a peak area versus CH4 and N2 flow rates at 35 Torr. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 119 Ha Peak Area vs. %CH4 and %N: VpM,- Fixed at -500V P = 50 Torr, FP = 1 OOkW 1 11111 as o> CO © Q. d 80000 60000 40000 20000 X %CH< * * T H/ ° 250 Z = 24.7(XY) - 18.3Y2+ 44.0X2 + 495Y - 1720X + 112000 Figure A3. Raw H a peak area versus CH4 and N2 flow rates at 50 Torr. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 120 Normalized CJHa Ratio vs. %CH4 and %N2 P = 20 Torr, F P = 1.00kW Normalized to H a V alues a t 5% CH4 2 0\« ' %CH K R T h Z = -0.00142(XY) - 0.00192Y2 - 0.000668X2 + 0.0168Y + 0.0614X - 0.150 Figure A 4. C /H a ratio versus CH4 and N2 flow rates at 20 Torr. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 121 Normalized CJHa Ratio vs. %CH4 and %N2 P = 35 Torr, FP = 1.00kW Normalized to H a V alues at 5% CH4 % cH W * T H. Z = -0.00103(XY) - 0.00248Y2 - 8.79 e‘5X2 + 0.0157Y + 0.0635X - 0.00639 Figure A5. C /H a ratio versus CH4 and N2 flow rates at 35 Torr. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 122 Normalized CJHa Ratio vs. %CH4 and %N2 P = 50 Torr, FP = 1 OOkW Normalized to Ha Values at 5% CH4 /oCH< tv r t h Z = 0.00305(XY) + 0.000255Y2 + 0.000178X2- 0.0360Y + 0.100X -0.00254 Figure A6. CJH a ratio versus CHi and N2 flow rates at 50 Torr. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 123 Normalized CH/Ha Ratio vs. %CH4 and %N2 P = 20 Torr, F P = 1.00kW Normalized to H a Values a t 5% CH4 0.3 o \° Z = -0.00078(XY) - 0.0015Y2 - 0.00023X2 + 0.011Y + 0.013X + 0.25 Figure A 7. CH/Ha ratio versus CH4 and N2 flow rates at 20 Torr. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. tiO VS- /®c r ^A Nom*'**6 AO ^5 20 %CH 4 W RT H, ♦ 0 -2 ^ . 0.00° ^ * w .0.00060VZ+ 6 7 = - 9 .0 e 'H * O g „ » io v e r s u s c h / H « rfl" ° ch, ^ + 0 .0 0 ° ^ _ „ i 35 T orr NiJ1oW Figure**- , Wo ^ . — ^ o e O ^ ^ ' 0" " eP,0dU" " Pt0W 125 Normalized C H /H a Ratio vs. %CH4 and %N2 P = 50 Torr, F P = 1.00kW Normalized to H a Values a t 5% CH4 2 0.3 Z = 0.00015(XY) - 5.8 e‘5Y2 - 0.00020X2 - 0.0022Y + 0.0065X + 0.11 Figure A 9. CH/Ha ratio versus CH4 and N2 flow rates at 50 Torr. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 126 Normalized C N /H a Ratio vs. %CH4 and %N2 P = 20 Torr, F P = 1.00kW Normalized to Ha Values at 5% CH4 Z = 0.00802(XY) - 0.0186Y2- 6 . 0 1 ^ + 0.298Y - 0.000151X+ 0.331 Figure A 10. CN/Ha ratio versus CH4 and N2 flow rates at 20 Torr. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 127 Normalized CN/Ha Ratio vs. %CH4 and %N2 P = 35 Torr, FP = 1.00kW Normalized to H a V alues a t 5% CH4 * h t h Z = 0.00602(XY) - 0.00851Y2 -0.000584X2 + 0.160Y + 0.0143X + 0.124 Figure A ll. CN/Ha ratio versus CH4 and N2 flow rates at 35 Torr. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 128 Normalized CN/Ha Ratio vs. %CH4 and %N2 P = 50 Torr, FP = 1 OOkW Normalized to Ha V alues a t 5% CH4 o\° /o C H < w * r H Z = 0.00825(XY) + 0.00187Y2- 0.00115X2+ 0.0272Y + 0.0317X- 0.0252 Figure A12. CN/Ha ratio versus CH4 and N2 flow rates at 50 Torr. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 129 Ha Peak Area vs. FP VPMr Fixed at -500V P = 35Torr, CH4/H2 = 0.15, Nj/CH* = 0.045 250000 200000 < 150000 (0 <D a x 100000 Y = 1 5 8 8 7 4 X -78831 50000 0.6 0.8 1 1.2 1.4 1.6 1.8 FP (kW) Figure A13. Raw H a peak area with respect to microwave forward power (FP). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 130 Cg/Ha vs. FP P = 35 Torr, CH4/H2 = 0.15, iy C H 4 = 0.045 2 R atio 1.5 Cg/Ha 1 0 .5 Y = 0 .3 9 4 X 2 - 0 .4 0 2 X + 1.01 0 0.6 0.8 1 1.2 1.4 1.6 1.8 F P (kW ) Figure A14. Cj/Ha ratio with respect to microwave forward power (FP). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 131 CN/Ha vs. FP P « 35 Torr, CH4/H2 = 0.15, lyC H * = 0.045 o to cc 8 X z o Y= 4.21 X2 -6.60X + 3.35 0.6 0.8 1 1.2 1.4 1.6 1.8 FP (kW) Figure A15. CN/Ha ratio with respect to microwave forward power (FP). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 132 CH/Ha vs. FP P = 35 Torr, CH4/H2 = 0.15, iy C H 4 = 0.045 0 .4 CH/Ha Ratio 0 .3 0.2 0.1 Y = 0.181X2 - 0.607X + 0.646 0.6 0.8 1 1.2 1.4 1.6 1.8 FP (kW) Figure A16. CH/Ha ratio with respect to microwave forward power (FP). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 133 CH+/Ha vs. FP P = 35 Torr, CH4/H2 = 0.15, Ng/CH* = 0.045 0.12 r 0.1 o £CO 0.08<P o 0C S 0.06 g 0.04 0.02 Y = -0 .0 0 1 6 1 X + 0 .0 8 7 3 0.6 0.8 1 1.2 1.4 1.6 1.8 F P (kW ) Figure A17. CH+/Ha ratio with respect to microwave forward power (FP). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 134 Hp/Ha vs. FP P = 35 Torr, CH4/H2 = 0.15, N^CH* = 0.045 0.3 5 0 .3 o CO cr 0.2 5 0.2 8 5 0.15 x 0.1 CO. 0.05 0.6 Y = 0.0748X 2 - 0 .2 3 2 X + 0.431 0.8 1 1.2 1.4 1.6 1.8 F P (kW) Figure A18. Ha/Ha ratio with respect to microwave forward power (FP). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 135 CN/C2 v s . FP P = 35 Torr, CH4/H2 = 0.15, IV C H 4 = 0.045 O c5 CC CM O z o Y = - 8.25X4+ 42.1 X3 - 74.6X2 + 55.4X -13.7 0.6 0.8 1 1 .2 1.4 1.6 1.8 FP (kW) Figure A19. CN/Cj ratio with respect to microwave forward power (FP). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 136 CN/CH vs. FP P = 35 Torr, CH4/H2 = 0.15, N.>/CH4 = 0.045 40 ~ ou I 20 ta CL O Z o Y = - 51.6X4+ 267X3 - 465X2 + 339X - 85.2 -10 0.6 0.8 1 1.2 1.4 1.6 1.8 FP (kW) Figure A20. CN/CH ratio with respect to microwave forward power (FP). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 APPENDIX B IACUC FORM FOR IN VIVO BIOCOMPATIBILITY STUDY Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 138 | H M I T H E u n iv e r s it y o f I d W A L A B A M A AT BIRMINGHAM Office of the Provost NOTICE OF APPROVAL DATE: March 25,2002 TO: Yogesh Vohra, Ph.D. CH 3871170 FAX: 934*8042 FROM: Clinton J. Grubbs. Ph.D.. Chairman C'A'n' Institutional Animal Care and Use Committee SUBJECT: Nanocrystalline Coatings for Dental TMJImplants (NIH) 020105515 On January 15,2002, the University of Alabama at Birmingham Institutional Animal Care and Use Committee (IACUC) reviewed the animal use proposed in the above referenced application. It approved the use of the following species and numbers of animals: Species Rabbits Rats Use Category B A Number in Category 64 72 Animal use is scheduled for review one year from January 15,2002. Approval from the IACUC must be obtained before implementing any changes or modifications in the approved animal use. Please keep this record for your files, and forward the attached letter to the appropriate granting agency. Refer to Animal Protocol Number (APN) 020105515 when ordering animals or in any correspondence with the IACUC or Animal Resources Program (ARP) offices regarding this study, if you have concerns or questions regarding this notice, please call the IACUC office at 934*7692. IwafSuMenal Animal Care and Uaa Commlttse B10 Vdkar Halt 1717 7th Avanua South 206.S34.78S2 • Fax 205.S34.11S8 iacucfiuab.edu www.uab.adu/lacuc The University of Alabama at amtngham Mafflng Addraaa: VHB10 1530 3RD AVES BIRMINGHAM AL 35204*0010 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. GRADUATE SCHOOL UNIVERSITY OF ALABAMA AT BIRMINGHAM DISSERTATION APPROVAL FORM DOCTOR OF PHILOSOPHY Name of Candidate Marc Douglas Fries__________________________ Graduate Program Materials Science___________________________ Title of Dissertation Nanocrvstalline Diamond Thin Films on TI-6AL-4V _________________ Temporomandibular Joint Prothesis Simulants bv _________________ Microwave Plasma Chemical Vapor Deposition I certify that I have read this document and examined the student regarding its content. In my opinion, this dissertation conforms to acceptable standards of scholarly presentation and is adequate in scope and quality, and the attainments of this student are such that he may be recommended for the degree of Doctor of Philosophy. Dissertation Committee: Name Yoeesh K. Vohra Signature j Chair Renato P. Camata Michael George Gregg M. Janowski A William Lacefield. Jr. Director of Graduate Program Dean, UAB Graduate School _ Date e»2 / *6 / p i O O 3 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.