a 1\§z.u..\ 1.. ‘ . ‘IS .1; “Iii-Lat . , .I..O¢.i.fl‘ld‘l J . Dian. . gin-.1... I... :tflufiwhflhfl rs. 1 Want! Oath . Luv}: . .- h.) ‘5‘...— ”IE“: ‘ Bu... $1.. «(1.. . ‘ . .531: h... 2.... I‘k’fi-‘h fit. I Eon? in: .3.l.- § .. f... e... flung“ . a in; .‘I‘ .févi:‘t f5 - .4 wiurfiw San}... inns“... 5.. Jr. isflifiiqufir .42.... , iii.“ 3‘ .\ I A: . .fiufig .5. . in" f . . an. - "$.31 um :iyiéazz a 353.. . .9; 9 59.5.3. 2.. inl Inuit Ill. . I51 Eu ‘ .3§% «1.5.... 9‘ Laughs-nits. . . IDIIS‘GII‘ V3a‘l’u‘s-I 33’14‘ a 25223.33. . i... figufiu :3 tunic. :5 . swag; ... 2...: f skiifi . "11.1. I . 3:: .5. nitric!” and“): a 19‘ 315: u a... Ilnfiblt , 31"" ?ii or! . S a 1.7.. ¢&-5....¥:his.‘ I. .. 3.. s , ll»... Cliffhw‘) I331»? an"... X. . v .i 2.. 13....Lfiw‘: A. ‘13: ‘ .5: .2 3: «S. 553:3} ‘5: .Ial... 1: I 33:4. .5 v! a ltnvnn t- Ira»h.3r 163(2): tflth‘iv ‘35:... . lax} Ila-.0: oil). 9.. . [5% £25.... U. I‘o \.. P .-lol. .,..l;!. . E i. hum}. .l‘tl“ , 1. ‘ 3:, V J..— ‘Y ‘ 3:... .2 . .83 r. ..... to, .2... ‘ dA Full . .21, . ..§.. 1: . I .- ......,... .lr. 0| AI . , 1‘ MICHIGAN STATE UNIVERSITY LIBRARY 300? This is to certify that the dissertation entitled NOVEL TRIBOLOGICAL SYSTEMS USING SHAPE- MEMORY ALLOYS AND THIN FILMS presented by RSITY MICH 48824-1048 LIBRARIES I YIJUN ZHANG TEUNIVE MICHIGAN STA EAST LANSING. has been accepted towards fulfillment of the requirements for the Ph.D degree in Materials Science and //1 7&6 Date MSU is an Affirmative Action/Equal Opportunity Institution MICHIGAN STATE UNIVERSITY PLACE IN RETURN BOX to remove this checkout from your record. To AVOID FINES return on or before date due. MAY BE RECALLED with earlier due date if requested. DATE DUE DATE DUE DATE DUE 2/05 p:/ClRC/DaIeDue.indd-p.1 NOVEL TRIBOLOGICAL SYSTEMS USING SHAPE-MEMORY ALLOYS AND THIN FILMS By Yijun Zhang A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Chemical Engineering and Materials Science 2006 ABSTRACT NOVEL TRIBOLOGICAL SYSTEMS USING SHAPE-MEMORY ALLOYS AND THIN FILMS By Yijun Zhang Shape memory alloys and thin films are shown to have robust indentation-induced shape memory and superelastic effects. Loading conditions that are similar to indentations are very common in tribological systems. Therefore novel tribological systems that have better wear resistance and stronger coating to substrate adhesion can be engineered using indentation-induced shape memory and superelastic effects. By incorporating superelastic NiTi thin films as interlayers between chromium nitride (CrN) and diamond-like carbon (DLC) hard coatings and aluminum substrates, it is shown that the superelasticity can improve tribological performance and increase interfacial adhesion. The NiTi interlayers were sputter deposited onto 6061 T6 aluminum and M2 steel substrates. CrN and DLC coatings were deposited by unbalanced magnetron sputter deposition. Temperature scanning X-ray diffraction and nanoindentation were used to characterize NiTi interlayers. Temperature scanning wear and scratch tests showed that superelastic NiTi interlayers improved tribological performance on aluminum substrates significantly. The two-way shape memory effect under contact loading conditions is demonstrated for the first time, which could be used to make novel tribological systems. Spherical for the first time, which could be used to make novel tribological systems. Spherical indents in NiTi shape memory alloys and thin fihns had reversible depth changes that were driven by temperature cycling, afier thermomechanical cycling, or one-cycle slip-plasticity deformation training. Reversible surface topography was realized after the indents were planarized. Micro- and nano- scale circular surface protrusions arose from planarized spherical indents in bulk and thin film NiTi alloy; line surface protrusions appeared from planarized scratch tracks. Functional surfaces with reversible surface topography can potentially result in novel tribological systems with reversible friction coefficient. A three dimensional constitutive model was developed to describe shape memory effects with slip-plasticity. After implementing the model in a commercial finite element program (ABAQUS), spherical indentation-induced superelasticity in NiTi was simulated and the results were compared to the results of experiments, including measurement of load-displacement curves and recovery ratios. Shallow indents were shown to have complete recovery upon unloading; deep indents had partial recovery. Stress-strain distributions, and the extent of the indentation-induced phase transformation region were also analyzed. Acknowledgement I am much obliged to Prof. David.S.Grummon at Michigan State University and Dr. Yang-Tse Cheng at General Motors Corp. for their advice, help, encouragement and inspiration in the art and practice of scientific research, which are truly invaluable to me. It would have been impossible for me to complete even a minute piece of this work without their support. I give sincere appreciation to my committee members: Dr. Carl Boehlert, Dr. Martin A. Crimp, Dr. John Shaw and Dr. Thomas Pence for their teachings, expertise, suggestions and generous help. I am also gratefiil for the help received fiom : Anita Weiner, Mike Luktich, Erkan Konca, Jihui Yang, and Yue Qi at General Motors Corp.; Dr. Thomas R. Bieler, Dr. Yuchen Lan, Hairong J iang, F ei Ren, John Foltz, Kebin Low, and J iaming Zhang at Michigan State University Table of Contents List of Tables .................................................................................................................... viii List of Figures ..................................................................................................................... ix Chapter 1 Introduction ......................................................................................................... 1 Chapter 2 Literature Review .............................................................................................. 6 2.1 NiTi Shape Memory Alloys ............................................................................... 6 2.1.1 Historical Development of Shape Memory Alloys .................................... 6 2.1.2 Phases in NiTi Alloy .................................................................................. 7 2.1.3 The Shape Memory Effect ......................................................................... 9 2.1.4 Phase Transformation Temperatures ........................................................ 11 2.2 Sputter-Deposited NiTi Thin Films .................................................................. 13 2.2.1 Fabrication of NiTi Thin Films ................................................................ 13 2.2.2 Effect of Stress on Transformation: the Clausius-Clapeyron Equation 14 2.3 Superelastic Effect of NiTi Shape Memory Alloy ........................................... 17 2.3.1 Residual Stress in NiTi Thin Films .......................................................... 17 2.3.2 Temperature Dependence of Superelasticity ............................................ 19 2.3.3 Superelasticity and Phase Transformations .............................................. 20 2.3.4 Mechanical and Microstructural Behaviors in Superelasticity ................. 24 2.3.5 Effects of Thermo-Mechanical Treatments on Superelasticity ................ 25 2.4 The Two-way Shape memory Effect ................................................................ 27 2.4.1. Introduction .............................................................................................. 27 2.4.2. Training Methods and Mechanisms of TWSME ..................................... 28 2.4.3 Phase Transformation Temperatures of TWSME .................................... 33 2.5 Contact Mechanics and Tribological Study ..................................................... 36 2.5.1 Elastic Contact Problems .......................................................................... 36 2.5.2 Elastic-Plastic Contact Problems and Instrumented Indentation ............. 38 2.5.3 Indentation-induced Shape Memory Effect ............................................. 41 2.5.4 Wear and Scratch Damage ....................................................................... 42 2.5.5 Tribological Study on Bulk NiTi Shape Memory Alloy .......................... 46 2.5.6 Hard Coatings and Interlayers .................................................................. 48 2.6 Tribological Study of Hard Coatings on Soft Substrates ................................. 50 2.6.1 Experimental Research ............................................................................. 50 2.6.2 Theoretical Research ................................................................................ 53 2.6.3 Preliminary Study of CerNiTizAl Tribological System ......................... 57 2.7 Constitutive Modeling of Shape Memory Effect ............................................. 59 2.7.1 One dimensional Constitutive Model of SME ......................................... 59 2.7.2 One—dimensional Ivshin-Pence Model ..................................................... 61 2.7.3 Three Dimensional Constitutive Models of SME .................................... 63 2.8 Summary .......................................................................................................... 65 Chapter 3 Methods and Materials .................................................................................... 67 3.1 Tribological Study ............................................................................................ 67 3.1.1 Substrate Materials ................................................................................... 67 3.1.2 NiTi Interlayer and Hard Coating Deposition .......................................... 68 3.1.3 Special Treatments of DLCzNiTi Interfaces ............................................ 70 3.1.4 Characterization of NiTi Interlayer Properties ......................................... 70 3.1.5 Tribological Testing Methods .................................................................. 74 3.2 Training and Characterization of The Two-way Shape Memory Effect .......... 75 3.3 Finite Element Modeling of Spherical Indentation on Superelastic NiTi ........ 78 3.3.1 Formulation of Three-dimensional Constitutive Model ........................... 78 3.3.2 Formulation of Slip Plasticity ................................................................... 81 3.3.3 Finite Element Program Implementation ................................................. 83 3.3.4 Validation and Discussion ........................................................................ 85 Chapter 4 Results and Discussion .................................................................................... 90 4.1 Tribological System with NiTi Interlayers ....................................................... 90 4.1.1 Residual and Phase Transformational Stress in NiTi Interlayers ............. 91 4.1.2 Wear Resistance of CrN Coatings on 6061 Al Substrates with NiTi Interlayers ............................................................................................................. 94 4.1.3 Contact Stress Induced Phase Transformation in NiTi Interlayer ............ 99 4.1.4 Wear Resistance of DLC Coatings on Soft Substrates with NiTi Interlayers ........................................................................................................... 102 4.1.5 Wear Resistance of CrN Coatings on Hard Substrates with NiTi Interlayers ........................................................................................................... 104 4.1.6 Interfacial Adhesion between CrN and NiTi .......................................... 105 vi 4.1.7 Interfacial Adhesion between DLC:NiTi ............................................... 107 4.1.8 Finite Element Modeling of Indentation on CrN:NiTi:A1 and CrN:Al.. 110 4.2 Indentation-induced Two-way Shape Memory Effect ................................... 112 4.2.1 Indentation-induced TWSME fi'om Thermomechanical Training ......... 113 4.2.2 One-cycle Deformation Training and Reversible Surface Topography. 117 4.2.3 Layered Planarization ............................................................................. 121 4.2.4 Small Hysteresis and “Self Training” of Reversible Surface Protrusion 123 4.2.5 Mechanism of Indentation-induced TWSME ........................................ 127 Chapter 5 Conclusions and Future Work ....................................................................... 131 5.1 Conclusions: Hard Coating Systems with NiTi Interlayer ............................. 131 5.2 Conclusions: Indentation-Induced TWSME .................................................. 133 5.3 Suggestions to Future Work ........................................................................... 134 Appendices ....................................................................................................................... 136 References ........................................................................................................................ 229 vii List of Tables Table 1. Comparison of crystallographic data of NiTi martensite a, Otsuka et al [20]; b, Hehmann and Sandrock [235]; c, Michal and Sinclair [21]; d, Kudoh et al [22]. .. 136 Table 2. Lattice structure and parameters of precipitates in NiTi alloys ................. 136 Table 3. Deposition parameters for NiTi interlayers on 6061 T6 substrates and M2 steel substrates, before CrN deposition ................................................................... 137 Table 4. Deposition parameters for NiTiCu interlayers on 6061 T6 substrates, before CrN deposition ........................................................................................................ 137 Table 5. Deposition parameters for NiTi interlayers on 6061 T6 substrates, before DLC deposition ....................................................................................................... 138 Table 6. Deposition parameters for NiTi interlayers on M2 tool steel substrates, before CrN deposition ........................................................................................................ 138 Table 7. Phase transformation temperatures of thin film and bulk NiTi used in TWSME study ........................................................................................................ 139 Table 8. Material parameters of NiTi and reference materials used in finite element modeling. (The “Ref” stands for reference material) ........................................... 139 viii List of Figures Figure 1. Phase diagram of a NiTi alloy by Massalski[241], to which the phase equilibrium between the B2 and Ti3Ni4 phases are added [242]. ........................... 140 Figure 2. Transmission electron micrographs of Ti3Ni4 precipitates in 51.3at%NiTi thin films which were solution-treated at 973 K for 3.6 ks followed by age-treatment at 773 K for (a) 3.6 ks; (b) 36 ks and (c) 360 ks, at 673 K for (d) 3.6 ks; (e) 36 ks and (f) 360 ks, and at 573 K for (g) 3.6 ks; (h) 36 ks and (i) 360 ks (after reference [51]). ................................................................................................................................. 140 Figure 3. Schematic illustration of parent and martensite with invariant plane as a habit plane. .............................................................................................................. 141 Figure 4. Stereographic projection of 24 habit plane variants from WLR theory, for <01—l > Type II twinning. ..................................................................................... 142 Figure 5. Ms temperature as a function of Ni content for binary NiTi alloys. Different data symbols represent data fi'om different authors. The solid line is given by thermodynamic calculations (after reference [34]). ................................................ 143 Figure 6. Schematic stress-temperature diagram for isochronal annealing of sputtered NiTi films on (100) silicon. Dark curves P1, P2 and P3 show the general intrinsic stress—deposition temperature characteristics for different pressures of the working gas. Trajectories a, b, c and d are for amorphous films deposited at different pressures and at temperatures below Ted, the minimum in-situ crystallization temperature. Isochronal crystallization occurs at Tc giving a transient tensile stress spike at x. Curve d additionally shows the effect of an isothermal hold inserted in the isochronal sequence. Trajectories e, f and g are for crystalline films deposited above T64. Cooling from any point on a trajectory generally proceeds along a path with slope 1.5 MPa/K, to points such as d‘, a‘ and g‘, allowing a wide range of stress levels to be obtained at T a. (after reference [58]) ................................................... 144 Figure 7. Stress-strain curves as a function of temperature of 50.6 at.% NiTi alloy, which was solution-treated at 1273 K for l h. The critical stress for inducing martensites and that for reverse transformation are indicated by an arrow (—-)) and by a double arrow (—-)—->) in (i), respectively. (after reference [68]) .................. 145 Figure 8. Critical stress as a function of temperature for inducing martensites (open circles) and for reverse transformation (solid circles), which were plotted fi'om the ix data in Figure 7. (after reference [68]) ................................................................. 146 Figure 9. Schematic representation for the appearance of the shape memory effect and superelasticity in temperature—stress space. (after reference [4]) .................. 147 Figure 10. Critical stress for inducing martensites and for reverse transformation of a 50.3at.%NiTi single crystal, which was solution-treated at 1273 K for 1 h followed by aging at 673 K for 1 h. (after reference [243].) ............................................... 147 Figure 11. Stress—strain curves as a function of temperature of a Ti—50.5Ni single crystal, which was aged at 673 K for l h after solution-treatment. The curves are related with R-phase transformation or deformation of R-phase. T’m=Rs. (after reference [72]) ......................................................................................................... 148 Figure 12. (a) Critical stresses for R-phase deformation and superelasticity, and (b) the recoverable strain of R—phase superelasticity, which is a function of temperature. (after reference [72]) ............................................................................................... 149 Figure 13. Elongation vs. temperature curves under 40 MPa load for Ti—Ni—Cu alloys with various compositions indicated. (after reference [73]). ............................... 150 Figure 14. Stress—strain curves of 50.6 at.%NiTi alloy, which was deformed at 243 K (>Aj). The alloy was solution-treated at 1273 K for 1 h. Dashed lines represent the recovered strain upon heating to 373 K. The symbol (x) represents the fracture point. (b) Plots of various strains in (a) as a function of total strain. (afier reference [68]) ................................................................................................................................. 1 5 1 Figure 15. Deformation stage of martensitic NiTi alloy sample (after reference [78] ). ................................................................................................................................. 152 Figure 16. TEM images of in-situ tensile loading of superelastic NiTi. (a) martensite plates 1, 2, 3, 4 formed at 118 um displacement. (b) High density of looped dislocations (marked by ABCD) formed at 150 um displacement. (c) Magnified photo of the area between martensite plates 1, 2 and 3, in (b). (d) Diffraction pattern corresponding to the area indicated by single-headed arrow in (c). (After reference [79].) The martensite plate 2 shows some detwinning at the lower part; but the plate 3 does not seems to detwin, even though dislocations have been generated. ......... 153 Figure 17. (above) Tensile cyclic stress—strain response of single crystal NiTi oriented in the [112] direction. The test temperature is above the A, transformation X temperature. (below) Compressive cyclic stress-strain response of single crystal NiTi oriented in the [112] direction. The test temperature is above the A f transformation temperature. (after reference [76]) ........................................................................ 154 Figure 18. Observation of Luders deformation by photography for Ti-50.5Ni alloy. Black regions represent the parent phase, and lighter colored regions martensite. The strain increases with increasing numbers flour (1) to (13). (afier reference [83]).155 Figure 19. Effect of thermomechanical treatment on the stress—strain curves as a function of deformation temperature for 49.8at.%NiTi alloy. Solution-treatment at 1273 K for 1 h followed by annealing at 673 K for 1 h. (after reference [244]) 155 Figure 20. Effect of thermomechanical treatment on the stress—strain curves as a function of deformation temperature for 49.8at.%NiTi alloy. Annealed at 673 K for 1 h immediately after cold-working. (after reference [244]) .................................. 156 Figure 21. TEM images of 49.8at.% NiTi alloy, which were annealed for l h at various temperatures indicated after cold-working; (a) as cold-worked, (b) at 67 3 K, (c) at 773 K and (d) at 873 K. (after reference [245]) .......................................... 157 Figure 22. Stress—strain curves as a function of testing temperature of 50.6at.%NiTi alloy, which was aged at 773 K for 1 h after solution-treatment at 1273 K for l h. ................................................................................................................................. 1 5 8 Figure 23. Effect of aging temperature on the critical stresses for inducing martensites and for reverse transformation in a 50.6at.% NiTi alloy, which was aged at various temperatures for 1 h after solution treatment. (after reference [244]). 159 Figure 24. TEM images and corresponding diffraction patterns of 50.6at.% NiTi alloy specimens, which was aged at various temperatures after solution-treatment at 1273 K for 1 h; (a and a’) at 673 K, (b and b’) at 773 K. (after reference [245]) 160 Figure 25. Stress—strain curves as a function of deformation temperature of 50.6at.%NiTi alloy, which was aged at 673 K immediately after cold-working. (after reference [245]) ....................................................................................................... 161 Figure 26. The procedure of processing TWSME TiNi SMAs spring (afier reference [87]) ......................................................................................................................... 162 Figure 27. The two-way shape memory effect obtained after a single deformation by xi 6.2% of equiatornic NiTi alloy (after reference [102]). .......................................... 162 Figure 28. Optical microscopy verification of the predicted microstructure memory for an aged Auled 49 martensite and its disappearance after aging in the parent phase. (a) Domain pattern of the aged martensite (aged at room temperature for six months). The sample is heated up to (b) parent phase (60 °C), and immediately cooled down to (c) martensite. Then the sample is heated up again to (d) parent phase (60 °C) and aged for 30 minutes, followed by cooling down to (e) martensite. (after reference [104]) ....................................................................................................................... 163 Figure 29. TWSME due to internal tensile residual stress: elongation during cooling and shrinkage during heating. ................................................................................. 164 Figure 30. External Stress and Internal Compressive Stress Field. The tensile external stress increases the internal tensile stress, elevating phase transformation temperatures but compressive external stress decreases the internal tensile stress. 164 Figure 31. Strain-temperature curves during (a) cooling and (b) heating through the transformation region of a trained specimen at different external stresses. The external stresses are indicated in the rectangulars (in MPa). The pure elastic deformation which is caused by the loading in the austenitic condition has been subtracted to obtain a common curve in the austenitic condition. .......................... 165 Figure 32. DSC curves of bulk NiTi samples used in TWSME study .................... 166 Figure 33. Stress distributions at the surface and along the axis of symmetry caused by (left) uniform pressure and (right) Hertz pressure acting on a contacting radius a (after K.L. Johnson, Contact Mechanics, Cambridge University Press, 1985, pp.94, [108]). ................................................................................................................................. 167 Figure 34. Two-dimensional photo-elastic fringe patterns (contours of shear stress). (a) point load; (b) uniform pressure; (c)rigid flat punch; ((1) contact of cylinders. (after reference [108], pp. 103). ............................................................................... 168 Figure 35. The “pile-up” (above) where contact radius is larger than linear case and, “sink-in” (below) where contact radius is smaller. ................................................. 169 Figure 36. Representative strain vs. hardness in spherical indentation, and its correlation with true stress-true strain in uniaxial tension. Metal A is steel and metal B is copper. (after reference [I I I, I 12]) .................................................................... 170 xii Figure 37. Finite element modeling of indentation of an elastic-plastic half space by a rigid sphere: development of the plastic zone. Broken line: contour of .12. Py is the normal load needed to first initiate plastic flow. (after reference [114]) .............. 171 Figure 38. Recovery of spherical indents in NiTi alloy by the shape memory effect. (after reference [231] ) ............................................................................................ 172 Figure 39. Load-displacement curves of (a) Berkovich and (b) spherical indentations in SE NiTi. (after reference [1 1 7] ) ......................................................................... 172 Figure 40. TEM images from the indent-affected zone in 50.9at%NiTi single crystals. Images (a) to (c) were taken adjacent to the same indent at different magnifications. (afier reference [I 18]). ............................................................................................ 173 Figure 41. Images of wear marks on NiTi after 200 cycles of scanning scratch under different normal loads and temperatures. A f is 16 °C. (afier reference [144] ) ....... 174 Figure 42. Calculated stresses due to an isolated unit normal line load applied at the origin as a function of substrate modulus E2. Variation of bearing stress 03, and shear stress try at the film-substrate interface are shown for a 1000 nm film with a reference elastic modulus E1=l . (after reference [198]) ....................................................... 175 Figure 43. Calculated tensile stress a, at the film-substrate interface for a 1000 mm film with a reference elastic modulus E1=1 for a unit normal line load at the origin. Substrate modulus E2 is varied in the upper figure. The effect of varying the film thickness on the flexure stress is shown in the lower figure for a substrate of modulus E2=0.2. (after reference [198]) ............................................................................ 176 Figure 44. Wear test results of samples with] um CrN coating and with: (from left to right) no interlayer, with 4 um SE NiTi interlayer (S1), 4 um Cr interlayer and 4pm SMA NiTi interlayer (82). (after reference [3] ) .................................................. 177 Figure 45. SEM Images of scratch track end with (a) lum CrN coating, no interlayer, (b) 5 um CrN coating no interlayer, (0) 1pm CrN coating 4pm SE NiTi interlayer, (d) lum CrN coating and 4pm SMA NiTi interlayer and (e) 1pm CrN coating and 4m Cr interlayer. (after reference [3]) ............................................................................. 178 Figure 46. Vogit (parallel) model for stress induced martensite in austenite phase. (after reference [204] ) ............................................................................................ 179 xiii Figure 47. Series model for stress induced martensite in austenite phase. (afier reference [204] ) ...................................................................................................... 179 Figure 48. X ray diffraction patterns of NiTi thin films (sample #1) at various temperatures ............................................................................................................ 180 Figure 49. X ray diffraction patterns of NiTi thin films (sample #2) at various temperatures sample ................................................................................................ 180 Figure 50. Temperature scanning X-ray diffraction of NiTi thin film on 6061T6 substrates, from which phase transformation temperatures were estimated and listed in TABLE. CrN coating was later deposited onto the NiTi thin films. ................ 181 Figure 51. Temperature scanning X-ray diffraction of NiTiCu thin film on 6061T6 substrates. CrN coatings were later deposited on the NiTi thin films .................. 181 Figure 52. Temperature scanning X-ray diffraction of NiTi thin film on 6061 T6 aluminum substrates. The NiTi thin films were interlayers of DLC coatings ...... 182 Figure 53. The normalized load-depth curves from Berkovich indentation of the NiTi thin films at different temperatures. ........................................................................ 182 Figure 54. Diagram of thermomechanical training method to create two-way indent depth change. .......................................................................................................... 183 Figure 55. Stress-strain curves of simple loading conditions from the 3D constitutive model for superelastic effect with plasticity. For the biaxial loading curve, the strain rate is set the same in two loading directions. The dashed curve represents a high uniaxial load which causes plastic hardening and permanent plastic strain. .......... 183 Figure 56. Finite element mesh of indentation model. The refined region is the contact region. Total node number is around 15000. ........................................... 184 Figure 57. Absolute (a) and normalized (b) load—displacement curves of SE NiTi and reference materials from finite element modeling. Indentation-induced superelasticity enables large or 100% indent depth recovery. ................................ 185 Figure 58. Recovery ratios vs. representative strains at different indentation depths, and the size of indentation-induced phase transformation region at different xiv indentation depths, measured in contact radius. ..................................................... 186 Figure 59. Comparison of experimental and calculated load-displacement curves. The experimental results are adapted from Ni et 01.6 Modeling SE NiTi has a phase transformation hysteresis A (A= (A f- A,)/2-(Ms - Mf)/2 ) 15K and 30K for the two computed curves. .................................................................................................... 186 Figure 60. Contours of the austenite phase volume fiaction at indentation depth of 0.018um (a), and complete indent recovery after unloading (b), where few residual martensites are present after unloading ................................................................... 187 Figure 61. Contours of the austenite phase volume fraction (a) and plastic strains (b) at indentation depth of 0.10am. (c) Contours of the austenite volume fiaction after unloading. For the partially recovered indent (recovery ratio 71%), there were much more residual martensites than in FIG. 5(b). .......................................................... 188 Figure 62. Residual stress evolution of NiTi thin film on Al substrates with temperature, for three heating-cooling cycles ......................................................... 189 Figure 63. First cycle residual stress evolution of NiTi thin film on Al substrates with temperature. ............................................................................................................ 189 Figure 64. The stress/temperature rate of NiTi thin film on Al substrates during martensite phase transformation. ............................................................................ 190 Figure 65. The stress/temperature rate of NiTi thin film on Al substrates from CTE mismatch between thin film and substrates, without the influence of phase transformation, ........................................................................................................ 190 Figure 66. Curvature change of 2‘SDLCaniTigégAl sample between room temperature around Mfand elevated temperature above A f. The curvature change mainly resulted from residual stress change in NiTi interlayer. ............................. 191 Figure 67. Temperature scanning wear deformation of CrN:NiTi:Al and CrN:Al samples, under 0.49 and 0.98 N load, using a 200 um tip radius spherical indenter. Superelastic NiTi improves the wear resistance by limiting plastic deformation to the soft aluminum substrates ......................................................................................... 192 Figure 68. SEM images of temperature wear tracks of 3"’CerA1 samples (top row) and 3“S'CrN:'5NiTi353:Al sample (bottom row), under 0.98N load. The wear data was XV presented in Figure 67 ............................................................................................. 193 Figure 69. SEM images of temperature scanning wear tracks of 3‘9CerAl samples (first row) and 3'9CrN: lsNiTi353,:Al sample (second row), under 0.49N. The wear data was presented in Figure 67. .................................................................................... 193 Figure 70. SEM images of temperature scanning wear tracks of l'7CrN:9NiTiCu343:Al samples. The wear data was presented in Figure 71 . ............................................ 194 Figure 71. Temperature scanning wear deformation of "7CermiTiCu343zAl and l'7CrN2Al samples, under 0.49 and 0.98 N load. ..................................................... 194 Figure 72. The effect of interlayer and CrN coating thickness on wear resistance at room temperature. The wear test was done using 6.35mm diameter WC under 5N load. Thicker coating and interlayer gave better wear resistance. ........................ 195 Figure 73. Wear loss vs. temperature using a different loading condition (6.3 5mm ball, 0.98N load). Wear resistance was not improved at temperature above .................. 196 Figure 74. Elastic contact stress in elastic halfspace from 3.175mm radius and 200nm radius sphere under different loads. ........................................................................ 196 Figure 75. Temperature scanning wear tests of 2'SDLCzNiTimzAl samples, using 200nm tip radius diamond indenter under 0.49N load. Thick NiTi interlayer and SE NiTi interlayer at temperature above A f showed better wear. ................................. 197 Figure 76. SEM images of temperature scanning wear tracks of 2'5DLC :NiTi363zAl and 2‘SDLC:Al samples. The wear data was presented in Figure 88. .......................... 198 Figure 77. The wear resistance of CrN:NiTi:M2 samples at room temperature as a function of CrN coating thickness. ......................................................................... 199 Figure 78. The wear resistance of CrN:NiTi:M2 samples at room temperature as a function of interlayer thickness ............................................................................... 199 Figure 79. SEM images of temperature scanning scratch tracks on 3'S’CergNiTi353:Al. Images of the whole scratch tracks (1 mm in length )are placed at the upper right comer. ..................................................................................................................... 200 xvi Figure 80. SEM images of temperature scanning scratches tracks on 3'S’CerAl. Images of the whole scratch tracks (1 mm in length )are placed at the upper right corner. ..................................................................................................................... 200 Figure 81. Acoustic emission of scratch tests on 3'9Cer9NiTi353:Al. It showed strong temperature dependence that matched phase transformation temperatures of NiTi interlayers ................................................................................................................ 201 Figure 82. Acoustic emission of scratch tests on 3'9CrN:Al. There is no significant temperature dependence .......................................................................................... 201 Figure 83. EDS spectrum of the delarrrinated area in scratch tracks on 3'9Cer9NiTi353zAl, showing that delamination occurred at the interface between CrN and NiTi interlayer. ................................................................................................. 202 Figure 84. EDS spectrum of the delaminated area in scratch tracks on 3‘9CerAl, showing the delamination occurred at the interface between CrN and the Cr interlayer. The Cr interlayer ( type II twinning (a lattice invariant shear), phenomenological crystallographic calculation gives 24 h.p.v. which are shown in a stereographic projection in Figure 4. The elastic strains associated with martensitic transformations are reduced by the invariant plain strain condition across the austenite/martensite interface. However, this condition cannot eliminate the transformation strains completely, because out-of-plane shear components are not eliminated under this condition, but if two habit plane variants with opposite shearing directions are formed side by side, the shear strains are eliminated macroscopically. The hierarchy in reducing elastic strain energy upon transformation by the combination of multiple variants is called self accommodation of martensite. The mechanism of self-accommodation in most of the B-phase alloys such as Cu—Al—Ni, Ag—Cd were successfully analyzed using phenomenological crystallographic theory [33]. According to the mechanism, four or two clusters of habit plane variants (h.p.v.) around 10 [011] poles self-accommodate, and the averaged shape strain matrix becomes close to the identity matrix, which represents the strain free state. 2.1.4 Phase Transformation Temperatures There are four phase transformation temperatures: Ms, Mf( martensite start and finish temperatures ) and A s, A f( austenite start and finish temperatures). Ms and Marc the temperatures that the martensitic phase starts to form (Ms) and completes the phase transformation (Mf), respectively. The A s and A f temperatures, the austenite start and finish temperatures, are defined in the same way. If MS and Marc close to A S and A f, then the shape memory alloy has a narrow phase transformation hysteresis and vice versa. One way to define hysteresis, A, is: A: AS+Af _MS+Mf 2 2 ' Transformation temperatures determine the temperature range in which shape memory effect (SME) or superelastic effect (SE) can be observed, which is important for practical applications. Experimentally it is known that the martensite transformation temperatures are strongly dependent on composition. For example, one percent change in Ni content above 50 at.% can change transformation temperature by more than 100 K. Figure 5 shows the phase transformation temperature dependency on the Ni content in NiTi alloy [34]. Therefore, the most effective way of obtaining desired transformation 11 temperature is by tuning composition. Ternary alloying elements are added to the binary NiTi system to alter transformation temperatures. Most alloying elements lower transformation temperatures, but Pd [35], Pt [36], Zr [3 7, 38] and Hf [39-41] increase transformation temperatures. Pd and Hf are used in making shape memory alloys with high phase transformation temperatures. The addition of Cu changes the lattice structure of martensite fi'om B19’ to B19 when the Cu content exceeds 8-10 at.%. Compared to the B2-B19’ phase transformation, the B2-B19 transformation has a narrower temperature hysteresis. The size and distribution of precipitates in the ternary NiTiX system has strong influence on mechanical properties and shape memory characteristics [40, 42]. Aging processes can also influence phase transformation temperatures, especially the Ms temperature, in binary NiTi alloy. The solubility of Ni in the B2 matrix is shown in the phase diagram in Figure 1. Quenching from solution temperature around 1173K can prevent excess Ni fiom precipitating. It enables the study of transformation temperatures vs. Ni content. Ni4Ti3 and other precipitates are formed in the aging treatment, which lower the Ni content in matrix and elevate Ms temperatures [43-45]. Since excess Ni can dissolve into the matrix while Ti cannot, the transformation temperatures of bulk Ti-rich NiTi are more stable [46]. 2.2 Sputter-Deposited NiTi Thin Films Development of NiTi thin films was driven by the prospect for microactuators in micro-electrical-mechanical system (MEMS) [9]. Magnetron sputter deposition, among other physical vapor deposition (PVD) methods, like evaporation and laser ablation [47], is the most effective method of making NiTi thin films with strong SE and SME [48-50]. Sputter—deposited NiTi thin fihns have grain size ranging from 0.5m to several microns, almost ten times smaller than that of bulk NiTi [51, 52]. The fine grain sizes give good mechanical properties such as higher yield stress, which is another reason for using NiTi thin films for actuation. 2.2.1 Fabrication of NiTi Thin Films Magnetron sputter deposition is a complex process. During the magnetron sputter deposition process, inert gas ions are accelerated by an electric field and bombard the cathode surface. Atoms of the cathode are dislodged by this energetic process and are deposited on the substrate. Many factors influence the resulting thin films’ composition, mechanical properties and SME characteristics. The most important parameters are target chemical composition, deposition power and voltage, Ar pressure, and deposition temperature. They are discussed briefly here: (1) At steady state, deposition flux matches the target composition even though 13 different elements have different sputter yield, because the target surface composition is altered by the differential sputter rate. Differential sputtering only affects the growing film because it too suffers sputter ejection and doesn’t reach a steady state. As a result, the sputter-deposited NiTi thin films usually have different chemical composition compared to the target composition. (2) Deposition power determines the deposition rate of thin films. Higher deposition power gives more substrate heating and affects trapping times. (3) Higher Ar pressure may produce thin films with a columnar structures while lower Ar pressure gives a featureless structure [51]. Because the kinetic energy of sputtered atoms may be reduced by collision with Ar ions and the absorbed Ar atoms may impede the diffusion of Ni and Ti atoms by reducing the kinetic energy driving surface diffusion [51]. (4) The deposition temperature determines whether NiTi thin films are amorphous or crystalline. The thin fihns are heated by the kinetic energy of deposited atoms, normally around 100K above from room temperature. Extra substrate heating is needed to elevate the temperature of thin films above around 650K so that Ni and Ti atoms have enough kinetic energy to crystallize [5 0, 53]. 2.2.2 Effect of Stress on Transformation: the Clausius-Clapeyron Equation The Clausius-Clapeyron equation is used widely to describe the pressure dependence l4 of the equilibrium temperature in the reversible first order phase transformation between gas-liquid, gas-solid, and solid-solid phases. Let the Gibbs fiee energy of phase A and B be G .4 and G B . For first order phase transformation, the Gibbs free energy is continuous but its derivative is not. do, = d0, , (eq. 2.1) 610 = Vdp — SdT , (eq. 2.2) For each phase, V is the molar volume, p is the hydrostatic (tri-axial) pressure, T is the temperature and S is the entropy. Use (eq. 2.2) in (eq. 2.1) to give: VAdp - SAdT = Vde — SBdT, (eq. 2.3). If the phase transformation direction is from phase A to phase B, then AS: 2S3 —SA , . 2.4 AV,” = V, —V, (6‘1 ) where AH j is the enthalpy change from phase A to phase B Rearrange (eq. 2.3) to give dp _ 45,? dTEQ - AVAB , (eq. 2.5) (eq. 2.5) is the basic form of the Clausius-Clapeyron equation. The dTEQ is the change in the equilibrium phase transformation temperature due to the pressure. Under the conditions of constant pressure and constant temperature, and no external work except fiom volume change from pressure, an entropy change can be expressed in terms of an enthalpy change: 15 Thus (eq .25) can be expressed as dp _ AH j dTEQ TEQ . AVAB , (eq. 2.7). For shape memory alloys, uniaxial tension and compression are the more common loading conditions than hydrostatic pressure. The uniaxial loading also changes phase transformation temperatures. This phenomenon can be explained by a modification of the Clausius-Clapeyron equation. If the pressure is replaced by uniaxial stress: (10 = Edd — SdT , (eq. 2.8). (eq. 2.5) becomes do _ASf_ AH: dTEQ A8? TEQ-AaA ’ (eq' 2‘9) The shift of phase transformation temperatures due to external stress can be estimated for NiTi SMA. Consider uniaxially loading of a lmx lmx 1m NiTi sample cooled from austenite phase to martensite phase. Typical material properties are: transformation enthalpy: AH f = 20J / g , equilibrium temperature: T=333K, density: 6.45 g/cm3, Ni atomic number: 58.7, Ti atomic number: 47.8, transformation strain: A3]? = 0.05 16 B B 6 3 3 d_a'= AS; : AHAB ___ 20J/gx6.45x10 g/m xlm =7.74x106 ”(57.74 MPa/K dT AaA ToAaA 333Kx0.05 Experimental value shows % = (7 i 1)MPa/ K [4, 54-56]. It is close to the theoretical prediction because enthalpy change and transformation strain may be different for different materials. [5 7] 2.3 Superelastic Effect of NiTi Shape Memory Alloy 2.3.1 Residual Stress in NiTi Thin Films The residual stress in NiTi thin films comes fi'om two origins: the intrinsic stress from the deposition process, and the thermal stress from mismatch of the coefficient of thermal expansion (CTE) between the substrate and the thin film [58-62]. The residual stress in a thin film can be measured by the wafer-curvature method using the well-known Stoney equation [63, 64], assuming the curvature change is elastic. The intrinsic residual stress is the result of the non-equilibrium characteristics of the sputter deposition process, which involves rapid thermal energy dissipation and simultaneous energetic particle bombardment during film growth. It is natural that the intrinsic residual stress is affected by deposition parameters such as Ar pressure, deposition temperature, and whether the thin fihns are crystalline or amorphous. The intrinsic residual stress can be relaxed by post annealing. Grummon er al. proposed a general stress model (Figure 6) describing the stress relaxation processes of NiTi thin films [58]. Stress relaxation path of isochronal and isothermal annealing of NiTi thin films are given for different deposition conditions. The intrinsic stress in a NiTi thin film is completely relaxed at crystallization temperature around 753K. Cooling from that temperature, the coefficient of thermal expansion (CTE) mismatch between silicon substrates and NiTi gives rise to thermal stress. For instance, the CTE mismatch between NiTi and Silicon substrates gives -1.0 MPa/K thermal stress rate, which is derived as: am, =11.4x10_6,as,. =3x10‘6 EMT, = 83GPa, v = 0.33 , E 3 dflz—L-(as —-af) 2 83x10 (-8.4x10‘6) z —l.011lfl’a/K dT l—v 1—0.33 where aNm , a s: are CTE of NiTi thin films and silicon substrates, E NiTi and v are elastic modulus and Poisson’s ratio of NiTi thin fihns. The thermal stress can be relaxed by martensite transformation partly or completely, and restored by the reverse phase transformation. The transformational stress relaxation rate is governed by the thermodynamics of martensite phase transformation. It can be expressed in the form of Clausius-Clapeyron equation, which was discussed in 2.2.2. This transformational stress relaxation and restoration cause a quite significant but reversible wafer curvature /displacement change within a relatively small temperature hysteresis of about 80K. This is the basis of some NiTi thin fihn micro-actuators for MEMS applications [60, 65, 66]. 18 2.3.2 Temperature Dependence of Superelasticity Superelasticity originates from the stress-induced phase transformations upon loading, and the subsequent reverse transformations upon unloading. The external stress changes the free energy and induces phase transformations according to the Clausius-Clapeyron equation discussed in section 2.2.2: For instance, tensile loading an austenitic NiTi specimen at temperature 50K above MS requires a = 7.74MPa/ K x 50K = 387MPa tensile stress to induce the martensitic phase transformation at that temperature. Superelasticity is therefore closely dependant on testing temperatures. The highest temperature at which martensite can no longer be stress induced is called Md. Above Md NiTi alloy deforms like ordinary metals by slip-plasticity. Below the A 3 temperature, martensite can be stress induced but does not completely revert to austenite. Thus, superelasticity appears in a temperature range from near A fand up to M4. The greatest ability to recover occurs just above A f [4]. Figure 7 shows the stress-strain curves for uniaxial loading of a NiTi specimen at different temperatures. The critical stress to initiate the plateau region where stress-induced martensites begin to form, increases with increasing temperature, which is seen in Figure 8, where open circles represent the critical stresses upon loading, and closed ones are critical stresses upon unloading. The dependence of the shape memory effect and superelasticity on test temperatures can be further presented in the schematic diagram of Figure 9. The line with a positive slope represents the critical stress to induce martensite, which is similar to the one with 19 open circles in Figure 8. The lines with negative slopes represent the critical stress for slip-plasticity. If the critical stress to introduce plasticity is very low (line B), complete superelasticity can not be expected, because slip is an irreversible process and has a detrimental effect on the thermoelastic phase transformations. If the critical stress for slip is high enough (line (A)), superelasticity can appear at temperatures above A fwithin the shaded region. Reverse transformation occurs upon unloading, since the martensite is unstable at temperatures above A fin the absence of stress. Similarly, at temperatures below A s the deformed martensites are stable, and the strain can recover only by reverse transformation on subsequent re-heating. In the temperature regime between A s and A f the martensites are partly unstable. The Md temperature is the maximum temperature that martensite phase can be stress induced. Above the M4 temperature slip plasticity is the only inelastic deformation mechanism in austenitic NiTi. 2.3.3 Superelasticity and Phase Transformations Superelasticity stems from the stress-induced phase transformations. Although the parent phase is invariably the B2 austenite, there are three low-temperature phases that can emerge after phase transformations, depending on chemical composition and thermal-mechanical history. The superelasticity associated with each low temperature phase is discussed briefly here. 20 BZ-BI 9’ Transformations Stress induced B2-B19’ phase transformation is the most important to the superelasticity of NiTi. The mechanism of superelasticity is the stress-induced martensite phase transformation upon loading and reverse phase transformation upon unloading. There is direct experimental evidence in Cu-Al-Ni single crystals to support this mechanism [6 7]. In uniaxial loading of a Cu-Al-Ni single crystal, a particular variant of martensite was stress-induced with the habit plane that was most favorable under the applied external stress. The parent single crystal changed into a single crystal of that martensite variant at the end of the stress plateau. Upon unloading the reverse transformation occurred with the same habit plane and the strain recovered completely upon unloading [6 7]. The recoverable strain also agreed with crystallographic predictions. The verification of the mechanism of superelasticity in NiTi has some complications. Solution-treated single crystal NiTi has low critical stress for plastic slip, which is the case of line B discussed in the section 2.3.2 and Figure 9. When such a specimen is subjected to external stress at temperature above A f, plastic flow started before the onset of stress-induced phase transformations and thus excluded superelasticity. However aged single crystal Ni-rich NiTi had higher critical stress for plastic slip due to precipitation strengthening. Superelasticity was achieved for aged NiTi single crystals, which are shown in Figure 10. But the aged single crystals did not give full recovery strain because untransformed regions were left due to the presence of precipitates. As a result, the recoverable strain was smaller than theoretical value [68]. 21 How superelasticity from B2-B1 9’ martensite phase transformation is affected by thermomechanical treatments is discussed in 2.3.5. BZ—R Transformation The B2-R transformation is an independent martensite phase transformation between B2-R-B19’ transformations. However if the martensite phase start temperature MS is higher than R phase start temperature R3, the R-Bl9’ phase transformation is suppressed. Superelasticity is also attainable with R-phase transformation, although the recoverable strain is about 0.5%. Ling and Kaplov studied the shape memory effect fi'om R—phase transformation in bending tests coupled with X-ray diffraction measurements [69, 70]. More systematic work on superelasticity and shape memory effects were done by Miyazaki et a1. by tensile tests on both 50 at.% Ti- 47 Ni—3 Fe polycrystals [71] and aged 50.5at.%NiTi single crystals [72]. The stress-strain curves of aged single crystal NiTi specimens at different temperatures are shown in Figure 11. R-phase transformation temperatures are different from that of B2-B19’ transformations. It is interesting to note that the recoverable strain increases with decreasing temperature. The reason is that the rhombohedra angle of R-phases changes with temperature. Figure 12 shows the critical stress for stress-induced R-phase transformation increased quite rapidly with increasing temperature, because the recoverable strains is so small that the slope of da/ dT in the Clausius-Clapeyron equation (discussed in 2.2.2) becomes very large. The available shape memory or 22 recoverable strain of R-phase transformation is small (less than 1% in polycrystals). But the small transformation strain may contribute to the small temperature hysteresis of R-phase transformation ( 1 -2K), which is a very useful property for actuator applications [4]- BZ-BI 9 Transformations B2-B19 (cubic to orthorhombic) transformation exists in NiTiCu SMAs, when the Cu content exceeds more than 8%-10%. The theoretical recovery strain associated with B2—B1 9 transformation was calculated for 40 at.% Ni-10Cu-50Ti alloy as a function of orientation by Nam et al. [73]. The maximum recoverable strain of 5.3% is obtained at <110>p orientation, which is much smaller than the maximum 10.5% strain in B2—B19’ transformation. But there is no supporting data from single crystal NiTiCu experiments. Polycrystal NiTiCu SMAs have smaller temperature hysteresis in superelasticity, compared to binary NiTi (Figure 13). The temperature hysteresis decreases with increasing Cu content, the minimum being 4K for 20at% Cu. It was suggested that volume change in B2-B19 phase transformation decreases with increasing Cu content up to 12.5 at.% Cu [73]. Further addition of Cu would increase the volume change. The small volume change may be one factor for the small temperature hysteresis but that alone cannot explain all the experimental data. Another mechanism proposed is the parent phase strengthening by solution of Cu atoms [74]. The frictional stress for twin boundary movement is suppressed if the introduction of dislocations is avoided. 23 2.3.4 Mechanical and Microstructural Behaviors in Superelasticity The loading and unloading of a superelastic NiTi sample involves complicated changes in microstructure and multiple stages. In stress-strain curves, the loading stage has a plateau region, after stress has reached a critical level. The plateau region involves stress-induced martensite phase transformation that is triggered by a resolved shear stress over the most favorably oriented martensite habit plane. An internally twinned structure is formed within a single martensite plate. Such a martensite plate can be called a correspondence variant pair (CVP) because two related martensite single crystal variants develop. This is the first stage of superelastic loading that is well-understood. The second stage is characterized by the strain hardening with slightly increasing superelastic strain, which was noticed by Miyazaki in multistage loading of superelastic NiTi alloy ( Figure 14) [68]. It is suggested the detwinning process may explain this behavior [75 -77]. The detwinning is that one martensite variant of the CVP grows at the expense of the other and finally saturates the whole martensite plate. The detwinning process has been observed in TEM of loaded martensitic NiTi alloy (Figure 15 and [78]), which occurred after the alignment of martensite twins (CVPs) with external stress was complete. Finally the martensite plate consists of only one variant instead of twins. Dislocations are generated in the late stage or after the detwinning process has finished (Figure 15). From then on slip-plasticity is the main deformation mechanism [75, 78], which is the third stage. In-situ TEM observation of stress-induced martensite 24 transformation and plastic deformation was carried out by J iang et al. [79]. However they did not intend to study the detwinning in NiTi alloy, although the TEM micrographs showed some evidence of detwinning (Figure 16). Tension-compression asymmetry was found present in single crystal and textured polycrystalline NiTi SMA [76, 77, 80-82]. The reason is that the CVP with the largest resolved shear stress component is induced preferentially than the rest 23 CVPs, which can be described by the Schmid law [76]. The tension-compression asymmetry arises from the different Schmid factors between tensile and compressive directions. Figure 17 shows the tension-compression asymmetry of [211] orientated single crystal NiTi alloy [76] In polycrystal NiTi samples without texture, the stress induced martensite transformation plateau is found to be inhomogeneous, like Luders deformation, that involves nucleation at some point within the specimen and the subsequent propagation, according to direct observation with an infrared imaging radiometer by Kyriakides and Shaw (Figure 18, [83]). 2.3.5 Effects of Thermo-Mechanical Treatments on Superelasticity Thermomechanical treatments, mainly by slip-plastic deformation and aging treatments, are important to the superelasticity and shape memory effects. Thermoelastic phase transformations are reversible but thermomechanical treatments 25 introduce irreversible changes to the microstructure. Their main purpose is to increase the critical stress for plastic slip by strengthening the parent phase. As discussed in 2.3.2, low critical stress for slip in solutionized single crystal NiTi makes the realization of superelasticity impossible. Higher critical stress for slip means that superelasticity can be obtained over a wider temperature range and with larger recoverable strain. The full potential of NiTi SMA can only be brought out after suitable thermomechanical treatments. Cold-working followed by annealing enhances superelasticity. Cold work introduces dislocations into the specimen; post annealing is a recovery process during which dislocations relax and rearrange. The critical stress for plastic slip is increased after this process. As a result, the plateau stress of superelastic loading can also be increased by loading at higher temperature. The temperature and stress range for superelasticity is thus increased. Figure 19 shows a series of stress-strain curves from solution-treated 49.8at.% NiTi wires. There is only partial superelasticity at temperature above Af. After uniaxial cold drawing and post annealing at 673K for 1h, good superelasticity was achieved (Figure 20). Figure 21 shows the TEM images of specimens at different stages of thermomechanical treatments. NiTi alloy was in a recovered but not recrystallized state after annealing at 673K, and led to the best superelasticity. Increasing the annealing temperature to 773K resulted in recrystallization, which again lead to lowered critical stress of slip and decreased superelasticity. 26 Aging treatment enhances superelasticity by precipitation strengthening in the parent phase. Figure 22 shows the stress-strain curves of 50.6at.% NiTi wire that were solutionized and aged at 673K for 1h. Superelasticity was achievable only in a narrow temperature range. Figure 23 shows that after solution treatment and aging at 773K for 1h, much better superelasticity was obtained. TEM study (Figure 24 ) showed that precipitates were small but densely distributed in the matrix after aging at 673K; precipitates were larger and less densely distributed after aging at 773K. Small and densely distributed precipitates were more effective to strengthen the parent phase, which results in improved superelasticity. Combined effects of cold-working and aging enhance superelasticity. Combined use of work-hardening and age-hardening is more effective to increase the critical stress for slip, which is presented in Figure 25. Solutionized 50.6at.%NiTi wires were cold drawn and aged at 673K for 1h. Compared to Figure 22, complete superelasticity was available at higher stress level. Besides, the combined treatment was useful for increasing cyclic stability of superelasticity [84]. 2.4 The Two-way Shape memory Effect 2.4.1. Introduction The two-way shape memory effects (TWSME) is the reversible shape change during martensite and reverse phase transformations, without the stimulus of external stress. In 27 conventional one-way shape memory effect, the martensite transformation does not result in any macroscopic shape change because the martensite variants are self-accommodated to minimize elastic and transformation strain. The two-way shape memory effect is not an intrinsic effect but can only be obtained by certain thermomechanical treatments, termed as “training”. It was first noted in 1965 by Wang et al [85]. In 1974 Nagasawa et a1. did the first detailed research on this “reversible shape memory effect” in NiTi and CuZn SMA[86]. After the training procedures, microstructural and mechanical anisotropies are introduced into the material. The martensite variant formation follows these structural anisotropies during phase transformations to minimize internal energy. Macroscopic strain would appear because martensite variants form that have preferential alignment. Reverse austenite transformation cancels the shape change. More details about the TWSME mechanism will be discussed in 2.4.2. 2.4.2. Training Methods and Mechanisms of TWSME Various training methods have been developed to realize TWSME: (l) constrained aging, (2) thermomechanical cycling, (3) plastic deformation by slip. There are variations in configurations of each category. The mechanism associated with each of these training methods is discussed below. 28 Constrained Aging Training Figure 26 illustrates the process of an constrained aging training of NiTi springs [8 7]. The spring was aging at 773K under constraint, which was an over stretched position. After aging, the spring was deformed in martensite phase to a crushed position. The reverse transformation expanded the spring while the martensite transformation shortened the spring, completing the two-way cycle. However the authors did not propose any possible precipitates that might be induce the TWSME. Isalgue reported that the aging process under external stress created the oriented growth of y precipitates in Cu-Zn-Al SMA [88]. Zhang further found that oriented strain field around these 7 precipitates were distributed along one direction in austenite, regardless of the grain orientation [89]. But the 7 precipitates are embedded in the austenite phase coherently. While in the martensite phase, 7 precipitates did not result in any strain fields. The reversible strain fields around precipitates that were produced by constrained aging may explain the TWSME. Fukuda et al. aged a single crystal 51 .at% NiTi under compressive stress along [111] [90]. Ni4Ti3 precipitates were formed coherently to the matrix along [111]. They explained the resulting TWSME from an energy approach: the volume fraction of precipitates was about 9%, based on TEM study of the size, shape and distribution of the precipitates. The matrix was expanded elastically about 0.2% on average along the [111] direction. The elastic energy accumulated in the matrix was estimated to be about 0.3 MJ/m3; it was about 1.5% of the latent heat of the B2 --> R transformation. The 29 thermally induced R-phase was composed of a single variant and the spontaneous strain along the [111] direction was about 0.8% of expansion. The maximum work from the spontaneous shape change was estimated to be about 0.17 MJ/m3. Chang et al. reported constrained aging at room temperature [91], where NiTi wires were first solutionized and aged at elevated temperature and then bended in round shape in room temperature for 24 h. This, however, is more likely to be the deformation training method. Precipitates formed in constrain aging training can be observed using TEM [89, 90]. It has been shown that the oriented strain fields around precipitates in NiTi can cause martensite variants to align in preferential directions. The stress fields around the precipitates can be recovered again after the reverse austenite phase transformation [89]. Molecular dynamics simulation indicated that oriented tension stress fields in the middle line of the dislocation band are favorable sites for the martensite nucleation and growth of preferred variants [89]. Thermomechanical Training Thermomechanical training methods have different variations, depending on the researcher’s interpretation of mechanisms of TWSME. The most common setup was cycling between temperatures above A f and below M; when the specimen was subjected to external tension, compression or bending stress [92-95]. Pons et al. used the following training cycles: cooling down below the Mftemperature under a constant tensile uniaxial stress of 35 MPa, and then unload and heat up to A fwith zero stress [96]. This low stress was used to train a Cu-based shape 30 memory alloy. Prader et al. kept the tensile strain constant while thermal cycling between Mr and M4 temperatures. There is no consensus about the optimal training method. Explanations of the development of the TWSME often involve either the presence after training of a low proportion of locally stabilized martensites in the parent phase, or the presence after training of an internal stress field in the material. Stalmans [92] et al. carried out a well-cited research about thermomechanical cycling on Cu-Zn-Al SMA and discussed the mechanism of TWSME. They disproved these two previous explanations. Following their study, the most important effect of the dislocation arrays which are introduced during the training is not to develop internal stresses but rather to give rise to a microstructural anisotropy inducing a thermodynamic anisotropy. The thermodynamic anisotropy means that after training, the crystallographically equivalent martensite variants are no longer thermodynamically equivalent: the trained martensite variants are thermodynamically preferred to the non-trained variants. The authors argued that this thermodynamic anisotropy must have come from certain changes in microstructures as the result of training. The microstructural basis of the thermodynamic anisotropy was termed microstructural anisotropy. However, the exact nature of the microstructural anisotropy was not explained clearly. The authors speculated that dislocations arrays generated during the training should contribute to it. Manach et al. carried out a similar thermomechanical training procedure but in the pure shear loading conditions [97]. They supported Stalmans conclusion that internal stress field had only a weak influence 31 on the development of TWSME, and dislocation arrays introduced by training created not only internal stresses but also a thermodynamic anisotropy. They further suggested the anisotropy is due to the friction energy which is after training no longer equal for all martensite variants. One cycle slip-plasticity deformation training The slip-plastic deformation training was actually the training method used in the original discovery [85] and initial study of TWSME [86]. The specimens were deformed in the martensite phase to introduce a certain level of dislocations [98-103]. Slip-plasticity training has the advantage of requiring fewer training cycles. Figure 27 shows the TWSME obtained from the uniaxial tensile deformation training. As discussed previously, the mechanism of TWSME from slip-plasticity training is believed to be that the microstructural anisotropies (oriented precipitates, dislocations, residual martensites) and their associated mechanical anisotropies would select martensite variants in specific orientations that are energetically favorable. Although dislocations can be observed in transmission electron microscopy (TEM), they look like densely tangled black lines. Although possible, it is difficult to get information such as strains and strain gradient from direct TEM observation of dislocations. The localized stress fields can not be observed in TEM directly. because these obstacles, the mechanism of TWSME has not been clarified using TEM. The one-to-one correspondence between TWSME and microstructure “memory 32 effect” was found in Au-Cd alloy, where the specimen exhibited same variant configuration upon heating and cooling (Figure 28) under optical microscope [104]. Similar results are more difficult to get from NiTi alloy due to the fine size of martensite variants. 2.4.3 Phase Transformation Temperatures of TWSME The phase transformation temperatures are affected by the presence of TWSME. Thermomechanical training of NiTi wires lowered the A S and A f temperatures significantly but had a less influence on MS and Mftemperatures [105]. Stalmans et al. showed that external stress opposing the direction of TWSME shifted all transformation temperatures downwards [106]. If a NiTi bar was trained by slip-plasticity method using uniaxial tensile stress, the resulting TWSME would be: shrinkage during heating and elongation during the cooling. Assume there is a tensile residual stress field in the NiTi bar that is responsible for inducing the TWSME (Figure 29). Applying external compressive stress to NiTi bar decreases the internal tensile stress, if the level of the external compressive stress is not higher than that of the internal tensile stress. A schematic diagram is shown Figure 30. The actual stress inside the specimen is decreased by the external stress. According to the Clausius-Clapeyron equation, lowered stress also lowers the phase transformation temperatures. 33 However, the external compressive stress is in the opposite direction of the TWSME. During cooling the external compressive stress impedes the shrinkage of NiTi bar. If large enough, the compressive stress can suppress the TWSME completely. In experiments, both lowering phase transformation temperatures and suppressing of TWSME were observed when applying external stress in the opposite direction of TWSME (Figure 31) [106]. In Figure 31, strain change vs. temperature is presented for the cooling (3) and heating (b) direction, for various external stress levels. The TWSME with zero external stress is also shown. Note in Figure 31 the compressive stress is in the opposite direction of TWSME, which suppresses completely the TWSME at -52.lMPa. The phase transformation temperatures are elevated by the tensile stress and lowered by the compressive stress in the opposite direction of TWSME. Stalmans et al. also proposed a thermodynamic explanation for this interesting phenomenon [106]. The Clausius-Clapeyron equation derived in 2.2.2 can be expressed as 5:. AH? AT TAef’ (eq. 1.10). The enthalpy change from austenite phase to martensite phase is positive (heat flows into NiTi): AH f > O , which is why heat flow jumps in the cooling DSC curves when phase transformation occurs (Figure 32). The left hand side of the C-C equation 3%:- M A TAs‘;w ' must has the same sign as the right hand side If a normal SMA specimen was loaded by tensile stress and cooled from austenite to martensite phase, the phase 34 M transformation strain was also tensile. Since AH j" > 0, A8;M > 0 and T Ag“ > 0 , A the phase transformation temperatures were shifted upwards (AT > 0) compared to free standing specimen, because the tensile stress A0 >0. When heating from martensite to austenite phase, the AH f; < 0 . The recovery of phase transformation strain is in the opposite direction of tensile strain: A8; < 0. As a A AH result AH :4 < 0, A3; < 0 —) T A; > 0 . There was no change to the tensile stress, M so the phase transformation temperatures were still increased (AT > 0). In TWSME, however, the situation is different depending on the direction of external stress. Consider a specimen that was trained to possess TWSME giving elongation during austenite-martensite transformation on cooling (As): > 0) and shrinkage during martensite-austenite transformation on heating ( A8}; < O ). Tensile stress still elevates phase transformation temperatures as described in the previous case. But compressive stress lowered the phase transformation temperatures becomes: for cooling from austenite M AH to martensitephase, AHf>0,A8f >0 and TAefM >0. Since Aa'<0 A (compressive), so AT < 0 that the left hand side of (eq. 1.10) was positive as the right A M TAef, hand side; is still positive when heating fi'om martensite to austenite phase, because AH 3', <0, A8; <0. Since Aa < 0 (compressive), AT < 0. The key point is that the external stress was in opposite direction against strain change in TWSME as compared to the case in SME, which caused the negative temperature shift. Figure 31 35 shows the strain-temperature curves in Stalmans’ study that exhibited the lowering phase transformation temperatures. 2.5 Contact Mechanics and Tribological Study The subject of contact mechanics may be said to have started in 1882 with the publication by Heinrich Hertz of his classic paper 0n the contact of elastic solids [107]. His interest in the problem was aroused by the question whether the force of holding glass lenses could have a significant influence on the pattern of optical interference fiinges. However, developments in the theory did not appear in the literature until the early 20th century, stimulated by engineering developments on the railways, in marine reduction gears, and in the rolling contact bearing industry. Now contact mechanics are used widely in science and engineering disciplines like geology, astronomy, defense, materials and mechanical engineering, where impact and contact is investigated. Most importantly contact mechanics serves as the theoretical basis to tribological testing methods like indentation, wear and scratch tests. 2.5.1 Elastic Contact Problems Analytical solutions are given for many elastic contact problems. The elastic contact problems of spherical and sharp indenters into a half space are especially 36 important because they are directly related to the measurement of hardness. The mean contact pressure Pm is simply normal load P on the indenter over contact area 7:02 : P pm: 2 7m In the elastic spherical indentation, the contact radius a can be calculated: 3PR 1/3 a: (a) , where R is the radius of indenter, and E . is the reduced modulus, a function of the modulus of the indenter and the substrate: 1 1 l - + E * E E indenter substrate On the contact surface the contact pressure is not uniformly distributed. The maximum contact pressure p0 is expressed as 3P 6PE*2 ”3 _ 27m2 _ 71'3R2 P0 The vertical displacement of the surface within the contact radius, measured with respect to the original specimen free surface, is given by: 2 — Usubstrate ”p0 (2‘12 _ l u, (r) = -— 4a r2), rSa, substrate where Umbsmm, is the Poisson’s ratio of substrate [108]. The stress distribution is depicted in Figure 33 [108]. The maximum shear stress is located at a distance of half contact radius below surface. Figure 34 presents the photo-elastic fringes pattern of shear stress fi'om indenters of various geometries. They all show concentric 37 hemispherical contours. For uniform pressure and cylindrical contact, the highest stress is located at a distance from the contact surface. This distance is much shorter in the case of point loading. The flat punch shows stress concentration at the contact edges. 2.5.2 Elastic-Plastic Contact Problems and Instrumented Indentation In metals, slip-plasticity develops when the elastic contact stress exceeds the material’s yield stress. As the indenter is pressed deeper into the substrate, a plastic zone appears below the indenter tip, surrounded by an elastic region. This is called the “contained mode”. The plastic zone grows with indentation depth and finally breaks out to the fi'ee surface. From then on the displaced material is free to escape by plastic flow to the sides of the indenter. This is the uncontained mode of indentation. It begins when mean contact pressure pm reaches about triple the yield stress: P p :——:CO’.Id. m 7:02 ”8 When p," is between 1 and 3, the plastic zone is contained by elastic materials [108]. The mean pressure is also defined as the hardness of the material. Indentation is a traditional method to measure the hardness of materials. Instrumented indentation is the advanced form of indentation, where both displacement and load of the indenter is monitored continuously. It has been developed during the past decade driven by the need to measure modulus and hardness of thin films efficiently. 38 Nix et al. first proposed that the contact radius could be calculated from load-displacement curves, assuming that the initial part of the unloading curve is linear [109]. However when the “sink-in” or “pile-up” situation is present (Figure 35), the Doemer and Nix method is not accurate enough because the contact radius is larger in the “pile-up” situation and smaller when the “sink-in” occurs. This method was later improved by Oliver and Pharr [110], whose method is now used to calculate the contact radius. Analytical solutions for most elastic-plastic contact problems are not readily available, especially with hardening present. But most metals are elastic-plastic solids with hardening behavior. Both finite element analysis and semi-analytical models are commonly used in analyzing elastic-plastic contact problems. A semi analytical: spherical cavity model was proposed by Johnson to explain the elastic-plastic indentation process [108]. In this spherical cavity model, the contact surface of the indenter is encased in a hemispherical “core” with a hydrostatic component fi. Outside the core it is assumed that the material will respond the same way as in an elasfic half space containing a spherical cavity under hydrostatic pressure F. The elastic-plastic boundary is somewhere outside the spherical cavity. The hydrostatic pressure fl below the indenter can be expressed as: W+4(1-20) -2 Y —3{l+ln[ 60w) 1}. *VIEI 39 where Y is the yield stress of indented material. The .3 is the angle of indenter. For the Berkovich indenter, 8:19. 3°. The representative strain proposed by Tabor is also used to characterize spherical indentation into elastic-plastic solids [I 1 1, 112]. The representative strain is a depth- sensitive parameter that is defined as: a 8, = 0.2 — , R where a is the contact radius and R the indenter radius. When the representative strain is equal to the true strain, the hardness H is about 2.8 times of the true stress (Figure 36, [1 I 1, 1 12]): E- = 2.8, when 8, = 8 Y Tabor arrived at this conclusion after comparing the experimental data of spherical indentation and uniaxial deformation of copper and steel. The indentation and uniaxial deformation data set collapsed into a single line (Figure 36) when applying the above two equations. The correlation between spherical indentation experiments and uniaxial ’ tensile data, although an empirical law, indicates that using the concept of representative strain, spherical indentation experiments may possibly be used to measure the true stress-true strain relationship. Finite element analysis is an effective tool for investigating elastic-plastic contact problems [113]. Figure 37 shows the finite element modeling of the expansion of plastic zone under spherical indentation [114]. Sinclair et al. carried out a detailed finite 40 element analysis of the spherical indentation on elastic-plastic solids [115]. They confirmed Tabor's simple experimental correlations for work-hardening materials using the representative strain, and found generally good agreement between theory and other physical evidence. However they also found there was high strain gradient near the contact edge that was not predicted before. 2.5.3 Indentation-induced Shape Memory Effect Shape memory effect under contact loading conditions has received very little attention previously. This is partly due to the complexity of the problem and the absence of suitable characterization methods. Ni et al. reported the shape memory and superelastic effects in NiTi alloy during spherical and sharp indentation [116, 11 7] using a well-controlled instrumented indentation method. A recovery ratio r] was defined to quantify the indent recovery: 77 = hrnax - hresidual hmax ’ where hmx is the maximum indentation depth, and hm Mm, is the indent depth after recovery. Spherical indents in NiTi alloy recovered almost completely (7] =100%) due to the shape memory or superelastic effect, if the indentation representative strain was less than 5% [116, 117]. Spherical indents in martensitic NiTi recovered after being heated above Af (Figure 38) while in austenitic NiTi the indents recovered during 41 unloading (Figure 39 (a)). The recovery ratio was only about 45% during sharp indentation (Berkovich), because slip-plasticity impeded the martensite-austenite transformation and affected SME adversely. Deeper spherical indentation also resulted in decreased recovery ratio. Gall et al. did Vickers indentation on NiTi single crystals and observed the existence of dislocations in the vicinity of indents in TEM Gigure 40). The same study also suggested that the precipitates size and crystal orientation also affect the recovery ratio [118, 119]. Indentation on single crystal CuAlNi SMA also showed indentation-induced SME [120, 121]. The above research shows robust shape memory and superelastic effects under contact loading conditions in NiTi and Cu—based SMA. This new phenomenon of shape memory alloys could be used to engineer novel applications in surface engineering, information storage and optics. It also prompts new questions about the mechanism of the shape memory effect. 2.5.4 Wear and Scratch Damage Wear is the removal of material from a solid surface as a result of mechanical action. Wear can be divided into four categories: adhesive wear, abrasive wear, corrosive wear and surface fiacture wear [122, 123]. In addition, there are several compound types of wear: erosion, fretting, and cavitation erosion [122, 123]. 42 Adhesive wear arises as a result of a process by which isolated spots on two sliding surfaces adhere together momentarily. When shear occurs, the shear takes place at some place other than the original interface. As a result, an adhering particle is formed, which may come loose at a later stage. Adhesive wear is considered the purest and most important form of wear because it is the only one which is always present. Unlike other types of wear, adhesive wear can not be completely eliminated. A quantitative expression of adhesive wear was derived by Archard [124-126] : .4“? 3H’ V where V is the wear volume formed after sliding through a distance x; P is the normal load and H the hardness of the tested material. The non-dimensional constant It represents the probability that a junction between two surfaces would lead to the formation of a wear particle. It would be ideal but actually difficult to relate the wear coefficient k to basic material properties (modulus and hardness) and surface properties (surface roughness and lubrication conditions). Abrasive wear is produced by a hard protuberance plowing a groove in a softer surface so that the material in the groove is worn away. A similar expression to Archard’s equation is also obtained: _ tanepx 3H ’ where tan 0 is the tangent of the average effective roughness angle for the hard surface, 43 or for hard abrasive grains at an interface [127]. Values of fa—h—é under various circumstances have been tabulated [127]. This equation can explain some wear situations, for example, the reduction in wear rate when the abrasive hardness and the tested material hardness are nearly equal; the reduction in wear rate when smaller abrasive particles are used [127]. Corrosive wear is a form of wear where corrosion of a surface first occurs, then sliding occurs and removes the corrosion products, allowing further corrosion. Quantitative information on this form of wear is not readily available. Because the wear rate is governed by so many factors: the rate of corrosion, the rate of removal of corrosion products, the geometry of the sliding surfaces, and whether the whole of the corrosion products are removed all at once or in part. Surface fracture wear is common in rolling elements like ball and roller bearings, wheels and rails, and gears. In this case, as the stresses come and go during continued rolling service, cracks grow in or below the surface of the element, and eventually a spalled particle is produced. This phenomenon is closely related to fatigue failure. Wear tests are used to measure wear resistance. ASTM standards exist regarding various test materials and experimental configurations [128]. The pin-on-disk or ball-on-disk wear test is used most widely [129]. More details will be discussed in Chapter 3. The scratch test is a semi-quantitative method to evaluate a coating or thin films strength and interfacial adhesion. Small-radius spherical indenters are used most widely 44 as the scratchers. The radius is between 100 and 200 um. Different scratch methods, like progressively increasing load, constant load, and repeated scratch are available on various testing platforms. Coating failure occurs when the scratch load is increased to a certain value termed the critical load. The failure can happen at the interface between coating and substrate, or within the coating itself, or a combination of both. The critical load is the most important numerical indication of interfacial adhesion from a scratch test. Although scratch tests are used widely and are easily understood intuitively, and there are ongoing efforts to establish a standard [13 0]. It remains difficult to interpret scratch test results quantitatively and consistently. Description of the sudden failure of the coating is beyond the established realm of continuum elasticity and plasticity. Acoustic emission (AE) signals result either fiom the sudden release of elastic energy, or from surface interactions such as friction and adhesion. Sudden energy release can occur during unstable crack growth, high-speed phase transformations, or plastic instabilities. A transducer to measure the acoustic emission is attached to the scratch slider and its output can be correlated to events which occur along the scratch track. At constant load, as the scratch is created a baseline AE response is established due to plastic deformation and friction/adhesion. Individual AE responses fiom fiacture are superimposed upon this signal. The magnitude of the acoustic emission signal depends on the size of the crack produced since the energy in the AE signal scales with the energy released in the process [131]. For coatings thicker than a micron, the area of 45 adhesive failure is large compared to the area of through-thickness cracks. Therefore, a large jump in the acoustic signal is thus a reasonable indication of adhesive failure. Besides establishing the critical load, microscopic observation of the failed region also provides important information on the failure mechanism. Bull et al. proposed several failure modes from the appearance of the failed hard coating, such as through- thickness cracking, coating detachment, chipping, buckling, wedge spallation [132, 133]. The geometry and distribution of coating cracks also depends on the mechanical properties of coating and substrate. 2.5.5 Tribological Study on Bulk NiTi Shape Memory Alloy A number of researchers have investigated the erosion wear behavior of NiTi alloys in different wear modes and compared NiTi alloys to conventional engineering materials like steels, Ni-based and Co-based tribo-alloys [134-I44]. It is found that NiTi alloy has a better erosion wear resistance than conventional wear-resistant materials, especially during cavitation erosion processes [14], I 43]. In addition, NiTi alloy is resistant to corrosive wear [145]. The wear resistance of conventional tribo-materials strongly depends on their mechanical properties such as hardness, modulus, hardening [146, 147]. Each of the mechanical properties contributes in greater or lesser degree to the wear resistance in different wear modes. It has been suggested that the high wear resistance of NiTi alloy 46 can be attributable to superelasticity [142]. Richman et al. demonstrated the superior cavitation resistance of NiTi alloy compared to a number of standard and wear-resistant materials including cast martensitic stainless steels and Co-based alloys (Stellite) [141]. Shida and Sugimoto observed remarkable erosion resistance of NiTi alloys during a water-jet erosion test [142]. Liang et al. observed that NiTi alloy showed high resistance to wear under impact during a sand blasting test. They compared NiTi alloy to a wear resistant alloy Ni-hard—4 (CrggNisngpFe) and showed that NiTi has a lower erosion rate than the Ni-hard-4 alloy. They noticed that there was a strong correlation between the wear resistance and the superelastic strain. Bulk superelastic NiTi also shows excellent adhesive and abrasive wear resistance. Li et al. carried out pin-on-disk and block-on-ring wear tests on superelastic NiTi and found the adhesive wear resistance was superior to steels and other alloys [148-150]. But few researchers compared the wear behavior of superelastic materials at different temperatures. Qian et al. studied the microscopic wear resistance of SE ( superelastic ) NiTi using a Berkovich indenter under uN load at various temperatures, which is an abrasive wear condition [144]. The wear resistance of NiTi is significantly better in the temperature range around A f. This temperature dependence of wear resistance was shown in Figure 41. It is clear that at temperatures 50K above A f=289K the wear performance deteriorated. 47 2.5.6 Hard Coatings and Interlayers Hard coatings are widely used to reduce wear in tribo-contacts. Material loss by abrasive wear in tribo-contacts is minimized by using a coating with hardness greater than that of the counterface. Material transfer to the counterface by adhesive interactions and the accompanying material loss may also be reduced. Lubricating the contact, even if it is only boundary lubrication, serves the same purpose. Chromium nitride (CrN) has a FCC NaCl lattice structure, and exhibits high hardness (~20 GPa), low electrical resistivity (640 ufl-cm) and high melting point (1553K) [151-158]. The outstanding properties of CrN have generated practical applications in the protective coating, tribological, and superconductive disciplines. Reactive magnetron sputtering combined with a negative substrate bias is the most common method of making chromium nitride fihns. Argon gas is normally used as sputter gas and N2 as the reactive gas. The application of nitride hard coatings on cutting and forming tools has been successfully established over the past decade. Hard coatings like CrN normally contain a high degree of intrinsic compressive stress owing to lattice distortion and CTE mismatch between coating and substrate. Residual compressive stress in the GPa range is not unusual. The residual stress increases with coating thickness, and thick CrN coatings tend to suffer from premature interfacial failure. It is difficult to produce single-layer nitride coating thicker than 6-7urn without encountering adhesion problems, which limits the expansion of CrN applicability on into other areas. 48 Diamond-like carbon (DLC) is a metastable form of amorphous carbon containing a significant fiaction of sp3 bonds. It has a high hardness (12-17 GPa), chemical inertrress, optical transparency, and it is a wide band gap semiconductor [159]. DLC films have widespread applications as protective coatings in areas including optical windows, magnetic storage disks, car parts, biomedical coatings, and micro-electro-mechanical devices. The most common industrial process for deposition of DLC is direct-current or radio-frequency sputtering of a graphite electrode in an Ar plasma. Other methods like ion beam assisted sputter deposition, pulsed laser deposition and cathodic arc deposition are also used. DLC has a hardness around 12-17 GPa, slightly lower than that of CrN. The high compressive stress in DLC also limits the maximum thickness of the coating. DLC coatings are notable for their low fiiction coefficient compared with other hard coatings [160-163]. The friction coefficient of DLC on steel can be below 0.1 without lubrication, close to the fiiction coefficient of steel on steel with lubrication [164, 165]. Interlayers have long been used to boost interfacial adhesion, moderate intrinsic stress and increase performance. For example, the problem of high residual stress in electroplated nickel was overcome by introducing a Cu interlayer [166-168]. NiP interlayers were used to increase corrosion resistance and tribological performance of TiN and TiAlN coatings because the porosity of these PVD coatings was sealed off by the NiP interlayer [169-1 72]. Interestingly, the TiAlN/NiP structure on steel drill bits did not meet expectations because NiP can not withstand high impact forces during service, which shows the importance of the mechanical properties of interlayers [169]. 49 There is more discussion about interlayers between hard coatings and soft substrates in section 2.6. 2.6 Tribological Study of Hard Coatings on Soft Substrates 2.6.1 Experimental Research PVD hard coatings on soft lightweight alloy substrates are intended for potential application in machine components in automobile, aerospace and other industries. Here, the tribological system is closed, so that wear debris including coating fiagrnents circulates within the system and can cause severe consequences. The control of wear resistance and coating properties are more stringent for coated machine components. It represents a wider application area and more complicated problem than traditional hard coating:hard substrate system for tool materials. The economic and engineering significance of hard coating:soft substrate tribological systems is being recognized, leading to a rapidly expanding research literature on this topic since 2000 [1 73-196]. A rich selection of coatings have been applied to aluminum substrates by various methods: nitride coatings by PVD [181, 185, 191, 194, 196] and are ion plating [1 74, I 76, 183-185, I 90], ferrous coating by thermal spray [182], ceramic coating by micro-arc oxidation [185], plasma spray [192] and Mo-WC coating by laser surface cladding [189]. The majority of research efforts are devoted to coating processes and tribological characterization. 50 Experimental studies offer valuable information and solid support for the hard coating: soft substrate tribological systems. Zhou et al. used are ion plating to deposit TiN and CrN onto 6061 and 7075 aluminum alloy substrates [183, 184]. They compared the tribological performances of hard coatings on these two alloy substrates and suggested the substrates’ mechanical properties had an influence on the tribological properties. They pointed out that the plastic deformation of substrate during wear tests can cause hard coating fragmentation. Miki et al. depositied TiN on aluminum alloy substrates by are ion plating (AIP) method and an ion beam mixing (IBM) method [190]. They focused on the measurement of residual stress in hard coatings using X-ray diffraction. Different residual stress was obtained by altering substrate bias voltage during the deposition process with residual stresses of 2.3 GPa developed at lower N2 pressures. However they found that wear resistance of nitride coatings was not sensitive to the compressive stress. Kuruppu compared tribological performance of CrN and Cr2N coatings on A2 tool steel, 52100 stainless steel, and 2024 aluminum substrates, with and without Cr interlayers [194]. The coated alurrrinurn samples showed slightly lower scratch resistance than hard substrates. Multilayered coatings on aluminum substrates showed better scratch resistance than a single coating structure. The substrate deformation was apparent in some wear tests, but the hard coatings did not wear or delarninate significantly. Murakami and Zhou et al. also investigated a similar problem. But they found single layer CrN on 2024 aluminum substrates has higher scratch resistance than multilayer structures but suffered dramatic coating delamination during 51 wear tests [1 74, 183, 184]. Xia et al. studied diamond-like carbon coated 2024 aluminum alloy samples with four different intermediate layers [180, 186]. Results showed that a thicker interlayer improves tribological performance. Specifically, stable fiiction coefficients for these four samples were nearly same and about 0.1, but the number of sliding cycles giving a stable fiiction coefficient increased with increasing interlayer thickness. The wear rate of the samples also decreased with increasing interlayer thickness. The same group also investigated gradient layers of DLC/T iN/T i on nitrogen-implanted 2024 aluminum substrates. The results indicate that the gradient layer shows a gradual change in hardness, compact surface appearance and good tribological properties. Xu et al. used a laser cladding method to coat 2024 aluminum alloys substrates with composite Ti-Al-Fe-B coating [1 73]. The coated samples exhibited improved wear resistance compared to bare aluminum samples. Cain et al. investigated TiN coated 6061 aluminum alloy samples for electrical contact resistance improvement [195]. These were intended for electromagnetic shields with better corrosion and wear protection, and lower contact resistance at mated contact joints, which is important for contacts which are periodically operated in aggressive environments, such as the doors and access panels on mobile communications facilities used by the military in the field. It was concluded that TiN:Al contacts have significantly lower electrical contact resistance than comparable chromate conversion coatings or bare aluminum alloy contacts, when cycled through several thousand contact closures. 52 Some general points can be summarized from these studies on hard coating:soft substrates systems. (1) Hard coatings on aluminum substrates generally exhibit satisfactory interfacial adhesion and scratch resistance, although in some cases poor interfacial adhesion was regarded as the reason for coating delamination during wear tests. (2) Hard coatings provide wear and scratch resistance to soft substrates as expected. Thick coatings offer better tribological properties than thin coatings. (3) Substrate deformation occurred during wear and scratch tests, which may contribute to coating failure. (4) An interlayer can provide better tribological properties, but it depends closely on the process, interlayer properties and thickness. 2.6.2 Theoretical Research Soft and compliant substrates pose a fundamentally different contact situation than hard and stiff substrates. A hard and stiff substrate does require a hard coating for abrasion resistance. The purpose of the coating is to guard against diffusive material transport under sliding contact conditions. The soft and compliant substrates have a lower load bearing capacity, which can partly be compensated in design by limiting contact stresses. However it is necessary to compare the differences in contact stress between coated hard substrates and soft substrates, to understand the benefits and 53 limitations of soft substrates. So far theoretical research on this problem lags in comparison with numerous experimental studies. Souza et al. used frrrite element analysis to model spherical indentation on TiN and SiC coated 6061 aluminum substrates [1 79, 188]. They focused on circular crack patterns in hard coatings on soft substrates generated during spherical indentation, and compared modeling with experiments [188]. The number and location of circular cracks was found to be related to several factors: overall contact geometry, the friction coefficient between coating and indenter, the modulus mismatch between coating and substrate, the level of residual stress and coating fiacture toughness. Another study attempted to relate pile-up from spherical indentation in TiNzAl with plastic strain [1 79], but without much success. The relation between pile-up and plastic strain is an important topic in instrumented indentation. These could have been more informative studies if the authors had shifted their attention to comparison of the contact stress situations of hard coating on soft substrates and hard substrates. Rarnalingam et al. used the displacement formulation method to compare the elastic contact stresses at the interface between coatings and substrates [197, 198]. It is found that stiff (high modulus) substrates and compliant (low modulus) substrates contribute to a quite different stress condition. Different loading conditions are also considered: normal point load (Figure 42), shear point load (Figure 43), normal strip load and shear strip load. Point loading represents the concentrated contact loading where stress and strain is localized, while strip load is the distributed contact loading over a length scale an 54 order of magnitude larger than the coating thickness. Results are summarized in the following table, which are compared to the case with stiff (high modulus) substrates. O Displacements are distributed in compliant substrates over larger lateral dimensions than in stiff substrates. It is good for compliant substrates because the strain gradient is smaller. 0 The maximum normal bearing stress 0' y at Normal line load coating:substrate interface is lower in compliant substrates; (Figure 42 and the maximum interfacial shear stress Txy is also lower in Figure 43) compliant substrates. It is good for compliant substrates because it is beneficial for coating adhesion. O The interfacial tensile stress 0', is higher in compliant substrates, but can be lowered by increasing coating thickness. It is good for compliant substrates because increasing coating thickness could have a beneficial effect. O Displacements are distributed in compliant substrates over Shear Line Load larger lateral dimensions than in stiff substrates. It is good for compliant substrates. 55 O Interfacial shear stress 7,9. is lower in compliant substrates; interfacial tensile stress 0'Jr is higher in compliant substrates than in stiff substrates. It is good for compliant substrates. 0 Stiff substrates have a higher positive 0' y , which tends to lift the coating fiom substrates, while 0'), is moderated in compliant substrates. It is good for compliant substrates because it is beneficial for coating adhesion. O 0' 3 becomes compressive instead of tensile, regardless of Normal Strip Load substrates modulus. 0', is more compressive in compliant (contact lengthle times coating substrates than in stiff substrates. It is not good for thickness) compliant substrates. 0 Higher tensile stress 0' x at coating:substrate interface in compliant substrates. It is not good for compliant substrates Shear Strip Load 0 Lower lifting stress 0', in compliant substrates. It is good for coating adhesion on compliant substrates. Bilayer composite structures, which involve an interlayer between a stiff coating and the substrate, have also been considered [197]. It was found that an interlayer reduced the maximmn tensile stress 0'I up to 75%. Decreases in interfacial tensile stress are 56 accompanied by increases in shear and bearing stresses at the interface, but the rate of increase is low. For shear traction, similar but smaller gains in reducing tensile stresses are obtained with stiff interlayers. The lift-off (positive) 0' y is enhanced by introduction of an interlayer. The authors suggested even a thin interlayer can be helpful in lowering the maximum film tensile stress in heavily loaded conditions. Tensile and shear stresses are affected more strongly than the bearing or lifting stresses. The Ramalingarn study showed that there were both advantages and disadvantages of compliant substrates as support structures compared with stiff substrates. But since only elastic contact was considered, the plastic deformation on both soft and compliant substrates encountered during experimental research still posed a major problem. Introduction of an interlayer has the positive effect of reducing interfacial stress. Experimental research has also pointed to this direction. 2.6.3 Preliminary Study of CrN:NiTi:Al Tribological System Bulk SE NiTi has shown excellent wear resistance [134-144]. The wear and scratch track in martensitic NiTi also showed recovery after austenite phase transformation [2]. This corresponds to the indentation-induced shape memory and superelastic effect discussed in 2.5.3. However the high cost of NiTi alloy production and processing would inhibit its tribological application. What is an economic and versatile way to utilize the tribological advantages of NiTi shape memory alloy? The NiTi interlayer 57 approach may well be the right answer. Ni et al. studied the tribological performances of CrN:NiTizAl tribological systems, where CrN is nitride hard coating, NiTi is the interlayer, and 6061 aluminum alloy is the substrate [1-3]. It was found that a SE NiTi interlayer greatly improves the wear resistance compared to SMA NiTi or Cr interlayers (Figure 44). In scratch tests, the CrN coating shows no delamination with SE NiTi interlayer, while severe damage in CrN coating without NiTi interlayer was obvious Figure 45. These studies provided no systematic information on how the evolution of martensite-austenite phase transformations affects tribological performance. This knowledge can only be obtained from wear and scratch tests carried out over the complete transformation temperature range. Another key problem is how coating:interlayer thickness ratio affects tribological performance. The residual stress of CrN coating and NiTi interlayer interact with each other, and with contact stresses during tribological loading. Residual stresses also tend to increase with coating and interlayer thickness. Detailed knowledge of this problem can be obtained from tribological tests over a matrix of coating:interlayer thickness ratios, and over a temperature range that covers the phase transformation temperatures. 58 2.7 Constitutive Modeling of Shape Memory Effect 2.7.1 One dimensional Constitutive Model of SME Computational modeling of the shape memory effect has been developed with increasing applications of shape memory alloys. The macroscopic mechanical behavior of shape memory alloys is usually modeled following either a phenomenological or a micromechanical approach. One-dimensional phenomenological models are suitable for engineering practice, because they make use of measurable quantities as parameters and are often computationally efficient to implement. Uniaxial phenomenological models have been proposed by many researchers: Tanaka and Nagaki [199], Liang and Rogers [200], Ivshin and Pence [201-203], and Brinson and Huang [204], Boyd and Lagoudas [205, 206]. The main factor distinguishing these one—dimensional models is the particular kinetics governing phase transformations. Because the transformation strain is large compared to the elastic strain, the phenomenological transformational kinetics play a significant role. The micromecharrics approach tries in various ways to follow very closely the crystallography, using thermodynamics to describe the phase transformation [207-211]. These models often consider martensite variants as transforming inclusions and use micromecharrics to calculate the interaction energy due to the phase transformation. Stresses and strains are obtained as volume averages in which many possible martensite variants are included. The micromechanical models are much more complicated than phenomenological models and are computationally demanding. But based more closely 59 on physical metallurgy, these models may yield a more accurate constitutive law. One-dimensional phenomenological models start with how to model the mixture of martensite and austenite phases. Both parallel (Figure 46) and series (Figure 47) approaches can be adopted [204]. In the parallel model, the relationship between stress and strain is expressed as: 0':(1—§)O'A +‘f'0M GM = M(8-§6)$0A =EA(5_§5) 0:(§°EM +(1—6).EAX8_;6) In the series model, it is: 8:581” +(l-§)£A £M=5+i,€A=£- EM A _ («e—:6) _5_ ti (EM+ EA) where 8M , 0' M and 8A , a", are the strain and stress contribution from the martensite phase and austenite phases, respectively. The 0' and e are the total stress and strain. The 2,‘ is the martensite volume fiaction, and 6 is the phase transformation strain. E M and EA are the modulus of martensite and austenite phases. As pointed out by Brinson and Huang [204], the key difference between various constitutive models of shape memory materials is the kinetic equations governing the phase transformations. The kinetic equation specifies how the martensite volume fraction changes with stress, temperature, and loading history. For instance, in the Boyd-Lagoudas model [205, 206], the kinetic equation takes the form of exponential 60 frmction: ézl-exp[aM(Ms —T)+bM0'] M ln('01) bM __q: (M, —M,)’ CM ' where a The three-dimensional model developed in the present work is based on the one-dimensional Ivshin-Pence model [201, 202], whose kinetic equation is more complex. 2.7.2 One-dimensional Ivshin-Pence Model The Ivshin-Pence model is a one—dimensional phenomenological model of shape memory effects. The transformation kinetics were carefully constructed to capture various aspects of the shape memory effect. The shape memory alloy is treated as homogenous mixture of martensite and austenite phases. The total strain a is expressed as: e=(1—a)eM +a8A, where 8M and 8 A are the strain contributions from martensite phase and austenite phase. The variable a is the austenite volume fraction, used in the Ivshin-Pence model instead of the martensite volume fiaction f. The variable 8,, is purely elastic but 8,, has both an elastic component (El) and transformational component (6): M 6] What distinguishes the Ivshin-Pence model from other 1D constitutive models is its transformation kinetics that govern the hysteresis exhibited by the austenite phase fraction a . The rate form for the austenite volume fiaction is: d_a_ a(tK) _(_6_fl_ 6T +_ 613 do do: —.<_ 0 d—t amax(,6(T(tK),o(tK))) 6?: 6T 6t an dt dt 1‘1: 1—a(tK) amm ____(_6_,B _6__T+_6_fl_ do da>0 dt l-amin(fl(T(tK ),o(tK))) ddfl 6T 6t 6o dt dt where ,6 characterizes neutrality curves that is a function of stress and temperature but do not involve hysteresis of phase transformations. The austenite volume fiaction a is an explicit function of ,B. The variables amax and “min are envelope functions describing the outermost hysteresis behavior for the martensite phase transformation and the reverse austenite phase transformation, respectively. The variable t,{ is the previous time when the transformation directions switch. ,8(T,o) is a function of temperature and stress which characterize the driving force of phase transformations: ,6(T,o)= T+ 1 {(E“ -E,.)q2 ~50}. (53 —s;’,) 215MB, where S 3 and S j), a re the specific entropies of the austenite and martensite phases at reference temperature, E M and E A are the Young’s modulus of the respective martensite and austenite phase, and 6 is the martensite phase transformation strain. 62 The envelope functions can take various forms. The following form is used by the original authors and also in this study: a 3,, =O.5+0.5tanh(k,,6+k2) a:mm =0.5+O.5tanh(k3,6+k4)' The austenite volume fraction (1 can be integrated using amax and am)": d__a_ a(tK) amam _(6__,6_ 6T +_6__,6 do da —.<_ dt am(,6(T(tK),0(tK))) dd: 6T 6t +60 dt ’dt da_ 1—a(tK)___(6_fl_amin6T +_ 6,6 do)d _> 0 dt l—amin(,B(T(tK),o(tK))) dds ar 6t 6o dt dt The k parameters are determined by the phase transformation temperatures: amax(fl(M 30)) = amin(fl(As’O)) = W: am(,6(MS,0)) = amin(,8(Af,0)) =1—t/l ’ The parameter W is an arbitrary small number because w = 0 is a numerical singularity. In Ivshin-Pence model, {/1 = 0.02 is used [201, 202]. 2.7.3 Three Dimensional Constitutive Models of SME A number of three-dimensional constitutive models have been proposed based on existing one-dirnensional models. For phenomenological models, the expansion fi'om one to three dimensions is based on generalized plasticity theory, assuming the martensite-austenite phase transformation also holds for the assumptions of classical plasticity theory. These assumptions are: 1) yield stress is insensitive to hydrostatic stress; 2) inelastic deformation due to the shape memory effect, like plastic deformation, 63 is volume conserving; 3) the plastic strain increment in each direction is proportional to the stress deviator component in that direction. Current three-dimensional constitutive models of SME do not have enough experimental support from multidirectional loading [212-21 7]. Only one-dimensional models are well calibrated with uniaxial experimental data that is more readily available. The unique behavior of indentation-induced SME may thus be a suitable proving ground for three dimensional models. Modeling of indentation-induced SME also offers better understanding of complex SME under contact loading conditions. Lubliner and Auricchio et al. proposed a framework of three dimensional constitutive model of shape memory effect based on generalized plasticity [21 7]. Generalized plasticity is an intemal-variable model of rate-independent inelasticity that includes conventional or classical plasticity as a special case. It was developed in order to account for the behavior of elastic-plastic solids in which, following initial plastic loading and elastic unloading, the reloading is not necessarily elastic up to the state at which unloading began; such solids include graphite, some stainless steels and some minerals. The model is based on some fundamental axioms and on results fi'om elementary set theory and topology, which makes it also applicable to other mechanical behavior like shape memory effect. In the case of shape memory effect, the transformation strain rate tensor 8;- was determined by: . . J— e;=,i—3—S) 2J7,” where Si] was the deviator stress, J; the second invariant of stress, and A is the effective transformation (inelastic) strain, whose rate form was determined by the kinetics of martensite-austenite phase transformations. 2.8 Summary This chapter has briefly reviewed recent work in various research areas involved in this dissertation. The literature on shape memory alloys, shape memory and superelastic effect, contact mechanics, and constitutive modeling of shape memory materials is very extensive. It is difficult to cover these topics completely in this chapter. More attention is given to the superelastic effect, the Clausius-Clapeyron equation, and the Ivshin-Pence constitutive model due to their importance in this work. Although there is also abundant literature on two-way shape memory effect, most of them are focused on training methods instead of reaching deep into the mechanism of the two-way shape memory effects. The few that focused on the mechanism and were well recognized were reviewed. There presently exists no literature about two-way shape memory effects from training methods involving indentation. For hard coatings on soft and compliant substrates, more and more research papers have been published in recent years, reflecting a growing interest in these tribological 65 systems. The majority are devoted to coating deposition and tribological tests. Very few concern mechanics and modeling. Both experimental and theoretical work on this tribological system has been reviewed. 66 Chapter 3 Methods and Materials 3.1 Tribological Study 3.1.1 Substrate Materials Both M2 tool steel and 6061 T6 aluminum alloy. were used as substrate materials. The M2 tool steel was supplied by Crucible Service Center (Syracuse, NY.) The 6061 T6 aluminum was supplied by United Aluminum Corp. (North Haven, C.T.) Two mm thick M2 steel plates were cut into 25 mmx 25 mm square pieces by electrical discharge machining. The steel substrates were polished to 50 nm surface roughness with 0.25 pm grit size diamond paste. The surface roughness was measured by optical surface profilometer. The samples were cleaned ultrasonically in acetone and then in methanol for 10 minutes before being introduced into the deposition chamber. 3mm thick 6061 T6 aluminum alloy plates were cut into 30mm><30mm square pieces and then mechanically polished to 50-70nm surface roughness using 0.05 pm grit size alumina suspension. The polished aluminum substrates were then cleaned with acetone and methanol. ‘ The T6 temper in the as-supplied material was later changed by processing. 67 3.1.2 NiTi Interlayer and Hard Coating Deposition NiTi thin films were deposited by DC magnetron sputter deposition. The sputter cathode compositions were Ni 46.5 Ti 53,5, Ni 43 Ti 52, and Ni 33,5 Ti 535 Cu 8 (atomic percentage) for various experiments. The thin films compositions was Ni 433, Ti 51.7, Ni 51,3 Ti 43,2 and Ti 51.7 Ni 4;; Cu 6,] respectively, measured using a wet chemical analysis (the materials were dissolved with nitric and hydrofluoric acid and measurements were made by inductively coupled plasma optical emission microscopy by Wangyang Ni at GM R&D Center [218]). The detailed deposition parameters are listed in Table 3 through Table 6. For samples intended for later diamond-like carbon (DLC) deposition, a Cr layer of 0.3 pm thick was deposited onto NiTi interlayers in the same chamber, without breaking vacuum. The purpose was to insulate the NiTi interlayer fi'om contamination. The thicknesses of deposited films were measured by the spherical abrasion method (CALOTEST, CSM Instruments SA)T, and Wyko NT surface profilometer at film edges produced by masking. The surface roughness of the deposited thin films was measured to be 35i10 nm with a Wyko NT surface profilometer. The NiTi:Al samples were then transferred to a TEER UDP 550 unbalanced magnetron sputter system for deposition of the hard coatings. Surfaces were sputter- cleaned for 600 seconds before deposition of CrN, and 3.6 ks before deposition of DLC. I A rotating steel sphere wears through the coating, which gives a rim and then into the substrate which gives a crate. Coating thickness is determined from the diameters of the rim and the crate, measured optically. More details may be found at http://www.csm-instruments.com 68 400 V was applied to the samples during sputter cleaning. CrN coatings were deposited on both M2 steel and 6061 T6 aluminum substrates, with and without NiTi interlayers. DLC coatings were only deposited on the aluminum substrates with and without NiTi interlayers. A Cr layer of 0.1 pm thickness was deposited before the deposition of CrN and DLC to facilitate bonding, which was indispensable for CrN and DLC coatings deposition. CrN coatings were deposited by sputtering two chromium targets of 99.99% purity in a nitrogen- containing argon environment. Argon of 99.999% purity was used as the sputtering gas. The base pressure of the system was 8><10‘4 Pa and the working partial Ar pressure was 0.13 Pa. The partial pressure of nitrogen was 0.13 Pa. The deposition power was 600W with 4 A at 150 V. The DLC coating was deposited by sputtering from one chromium and two graphite targets. The base pressure was 1.33 X 10'4 Pa and working Ar pressure was 0.13 Pa. The deposition power was 525 W with 3.5 A at 150 V. NiTi interlayer and CrN layer thicknesses were measured at masked thin film edges by a surface profilometer. CrN and DLC coatings were also deposited onto reference aluminum and steel substrates without NiTi interlayers. Reference samples went through one same thermal history so that the hardness of the substrates was the same as the samples with a NiTi interlayer. The hardness of the as-received 6061 T6 aluminum alloy was 94 HRE (62 HRB), which dropped to 21 HRE after deposition of NiTi interlayers and CrN coatings. 69 3.1.3 Special Treatments of DLC:NiTi Interfaces Special treatments were given to create a strong DLC:NiTi interface, because the CrN:NiTi interface was found to be weaker than desired during scratch tests. A 0.3 pm Cr layer was deposited onto NiTi interlayers after the deposition and annealing process, without breaking vacuum. The Cr layers shields the NiTi interlayers from any potential contamination. After the samples were set into the TEER deposition system, the Cr layers were subjected to 3.6 ks sputter cleaning. It had been calibrated that the 3.6 ks sputter cleaning would remove about 0.1 um Cr layer. As a result, about 0.2 urn thick Cr layers still remain on the NiTi interlayers. Before the DLC deposition, another 0.1 um thick Cr layer was deposited onto the sputter cleaned Cr surfaces. Thus,.the DLC coatings and NiTi interlayers connected at the Cr-Cr interface, while the CrN coating and NiTi interlayers had a Cr-NiTi interface. 3.1.4 Characterization of NiTi Interlayer Properties The phase transformation temperatures of NiTi interlayers on aluminum and steel substrates were estimated using temperature scanning X-ray diffraction (XRD). For room temperature martensitic NiTi interlayers, the measurement was carried out on a Scintag 2000 diffractometer (Cu Ka radiation) using a thermoelectric module for heating. The temperature of the NiTi interlayer was measured by a K-type thermocouple that was adhered to the interlayer surface. An XRD scan was taken after each designated 70 temperature was stabilized. Room temperature austenitic NiTi interlayers needed a special cooling stage to reach sub-ambient temperature. The thermoelectric cooler was not capable of reaching steady state below 283K. Therefore additional measurements were carried out on a Rigalcu AF C6S (Cu Ko. radiation) diffractometer with a low temperature sample stage which was cooled by liquid nitrogen. Since the thin films were attached to the substrates during the temperature controlled XRD measurements, the measured transformation temperatures included the effects of residual stresses in the thin films. Results of temperature scanning XRD shown in Figure 48 shows that room temperature martensitic NiTi on steel substrates has A f350i5K and Ms (the martensite start temperature) 3 10:t5 K. Figure 49 shows that room temperature austenitic NiTi has Af310Kzt5K and Ms 270i5K. The results of NiTizAl samples are shown in Figure 50, Figure 51 and Figure 52. The phase transformation stress of NiTi thin fihns on aluminum substrates was measured by the wafer curvature scanning method, which is based on the Stoney equation [58, 60, 219]. The surface reflectivity of DLC and CrN hard coatings are not good enough for laser scanning stress measurement, but good enough for optical profilometer. The optical profilometer can also measure curvature over longer length, which is more than 20mm. Longer length provides more accurate curvature measurement. However the laser scanning stress measurement is capable of providing real time stress measurement vs. changing temperature, from which residual stress evolution can be 71 obtained. The optical profilometer method can only take “snap-shot” of the curvature at certain stable temperature. This first-order approximation is valid for to < ts / 100 . The Stoney equation calculates the stress oc in the thin film: 2 oc = E"‘ K 6t ’ C where EC is the biaxial modulus of the substrate, tc and ts are the thicknesses of thin film and substrate, respectively. K is the curvature of thin fihn:substrate bilayer, which is measured by the laser scanning method. The aluminum substrates that supported NiTi thin films were plastically deformed during thin film deposition and annealing by the intrinsic residual stress in the NiTi thin fihn (section 4.1.1). As a result the total curvature measured was not caused entirely by the residual stress in the NiTi thin film. Part of the curvature came from plastic deformation of the aluminum substrates. However the Stoney equation is still valid for measuring the relative residual stress change in the NiTi thin film. Let Kplasfic be the curvature from plastic deformation; K1 and K2 are total curvatures measured at two different temperatures; a; and of are the residual stresses in the NiTi thin films at two different temperatures: 2 1 _ Ests _ ac — 61c (Kl Kplastic) 2 2 _ Ests 0'6 _ 6! (K2 —Kplastic) C . . 2 2 l The change in resrdual stress A0] = O'c - O'c between two temperatures can be 72 calculated by Stoney equation: 2 1 Ests 6t C 2_ 2 _ A01 —O-C— C— (Kz‘Kl). A resistance-heating furnace was used to heat the samples, which were placed on the surface of the heater type thermocouple was inserted in a tight-fitting hole on the side of the furnace. Because the thermocouple did not contact the samples, actual temperature could have been lower than the measured temperatures. The CrN hardness and elastic modulus were 20i2GPa and 200i20GPa, measured from nine indentation tests at different locations on the coating; the DLC hardness and modulus were 10:1:1GPa and 160i20GPa; both were measured by nanoindentation on Nano Indenter XP (MTS Systems Corp.). Temperature scanning indentations were also done on NiTi interlayers on steel substrates to verify indentational superelasticity at elevated temperatures. The indentation experiments were carried out on a Micro Materials (Wrexham, U.K.) Nano Test 600 system with a Berkovich diamond indenter. The samples were immersed in liquid nitrogen for 10 minutes before experiments to obtain a fillly martensite phase in the interlayers. The indenter and substrate holder had resistance-heating ability. The temperature at the indenter and sample surface was measured by two thermocouples and controlled by independent temperature controllers. After the designated temperature was reached the experimental set-up was left overnight to reach thermal equilibrium and minimize thermal drift. The highest temperature used was 473K, far below the typical 73 aging temperature of 773 K for NiTi. As a result, the microstructure and mechanical properties of thin fihns are not expected to change by the heating step. The maximum load was set at 25 N so that the indentation depth would not exceed 10% of the film thickness in order to minimize the substrate influence on the indentation response of the films. The indentation speed was 5 nm/s, which was small enough to minimize any dynamic effects. Load-depth curves are shown in Figure 53. 3.1.5 Tribological Testing Methods Scratch and wear tests were used to evaluate tribological performance. The scratch tests were performed on a Micro-Scratch Tester (CSM Instruments SA), which can also record acoustic emissions from the scratch. A 107 um tip radius conical diamond indenter was used as the scratcher. (Conical indenters actually have a small-radius spherical tip.) If the indent or scratch dimension was small compared to the radius of the spherical tip, the conical indenter functions as a spherical indenter. A thermoelectric cooler heated the samples from below and the sample’s surface temperature was measured by attaching a thermocouple. The wear tests were performed on an ISC-200 Tribometer (Implant Sciences Corp.) equipped with heating sample holder. The wear counterpart was a 200 um tip radius diamond spherical indenter under 0.49 N and 0.98 N loads, and a 3.175mm radius WC 74 ball under 0.98N, 1N and 5N loads. The geometry of the wear tracks and scratch tracks were characterized by Wyko NT surface profilometer and an Hitachi SC-2500C SEM. The measurement of wear rate was made according to ASTM standard G 99 pin-on-disk wear test [220]: Volume of the Wear Track Total moving distance of the Ball Wear Rate = Volume of the Wear Track=(Cross-sectional area) X (Circumference) Total moving distance of the Ball = (Number of Total circles) >< (Circumference) 3.2 Training and Characterization of The Two-way Shape Memory Effect Both bulk NiTi and thin film NiTi SMAs were used in the two-way shape memory study. The bulk NiTi metal was purchased from Special Metals Corp. (New Hartford, New York). The nominal composition was N15031 Ti49,69. The phase transformation temperatures were measured by differential scanning calorimetry using TA Instruments modulated 2920 DSC. The DSC scanning curves are presented in Figure 32. The phase transformation temperatures were: A s =350K, A f=404K; Ms =344K, and Mf =287K. The as-received material was electrical-discharge machined to small pieces with dimensions of 2.45cmX2.45cm>< 1cm. Surface roughness was reduced to 0.05pm in three steps of mechanical polishing using 3, 1, and 0.5 um grit size diamond paste. The NiTi thin films were sputter-deposited on to a (100) silicon wafer at 673K in 45 75 minutes, followed by annealing at 773K for 3.6ks. The film thickness was measured to be 4.5 pm. The chemical composition was Ni 43.3 Ti 5L7 by wet chemical analysis. The phase transformation temperatures, A s =343K, A f=390K; MS =329K, and Mf=282K, were estimated by the temperature scanning wafer curvature method A simple “ball-and-clamp” method (Figure 54) was used to induce cyclic thermomechanical training to the NiTi sample. The NiTi sample was first cooled in liquid nitrogen for 5 min to ensure fully martensitic structure. A 3.175 mm diameter tungsten carbide ball was then clamped into the sample using a steel c-clamp with a fixed number of rotations of the clamping screw. The indentation depth was 170nm measured by an optical surface profilometer. The representative strain was about 9%. The whole fixture was then placed in a resistance-heating oven for 2 minutes to reach 423i10K. Afterwards, it was quenched into ice water for another 2 minutes, which concluded one training cycle. A total of 30 training cycles were performed. Since the number of rotations of the c-clamp screw is kept the same during the training procedure, the thermomechanical training is considered under a constant strain condition. The profile of indents was measured using Wyko NT optical surface profilometer. A thermoelectric cooler (Marlow Industries Inc., Dallas, Texas) was placed below the sample for heating and cooling. A thermocouple was taped onto the side of the sample to measure one temperature. The profiles of the indents were taken after the sample was heated to 400i5K and measured again after the sample was cooled to 300i3K. During the process the position of the sample was kept the same. 76 Another training method, using a single-cycle deformation approach, was carried out by the following procedures. A 3 X3 matrix of spherical indents was created on the bulk sample surface using a 1.59 mm diameter steel ball indenter under a 100kg load. Also a scratch was made with a 107m tip radius spherical indenter under 15N load, and nanoindents were created on the NiTi thin films by a 10pm tip radius spherical diamond indenter using a Nano XP indenter (MTS). Before indentation, the samples were cooled in liquid nitrogen for 5min to ensure a fully martensitic structure. The samples were then heated to 423K on a hot plate for 10min to let the spherical indents recover, i.e., transform from the martensite to austenite. They were cooled again in liquid nitrogen for 5 min. The profiles of the indents and scratches were measured using a Wyko NT optical surface profilometer (V eeco Instruments, Inc.). A thermoelectric device (Marlow Industries, Inc.) was placed below the sample for heating and cooling. A thermocouple was taped to the back side of the sample to measure temperature. The profiles of the indents were taken after the sample was heated to 400i2K and taken again after it was cooled to 300i2K. The residual indents and scratches on the bulk NiTi were removed by mechanical polishing: first using 240, 400 and 600 grit size silicon carbide grinding paper, and then with 6 um and 1 um grit diamond paste. The indents on the surface of NiTi thin film were polished using 0.05pm alumina suspensions. The profiles of surface protrusions were then measured using the Wyko surface profilometer together with the thermoelectric device for heating and cooling. 77 In the layered planarization experiment, the depth of successively polished layers was measured by comparing of the thicknesses of the polished NiTi sample and a reference sample, using optical surface profiler. 3.3 Finite Element Modeling of Spherical Indentation on Superelastic NiTi Spherical indentation-induced superelasticity in NiTi was simulated in FEA and compared to the results of experiments, including load-displacement curves and recovery ratios. A three dimensional constitutive model is developed based on the one-dimensional Ivshin-Pence constitutive model [202, 203] and generalized plasticity theory [21 7], to describe the shape memory effect under contact loading conditions. Details can be found in section 2.7. The von-Mises slip-plasticity is also included in this model to capture the mechanical response of NiTi SMA in large deformation. Implementation of the model in a commercial finite element program (ABAQUS) is done by writing a user material subroutine. 3.3.1 Formulation of Three-dimensional Constitutive Model 78 First the 3D constitutive model is presented in terms of total strain rate Eij . This total strain rate is decomposed into elastic part 85 and a transformation strain (inelastic) T part8ij I . .T .5 56=EU+3U. (1) Both the martensite phase (with volume fiaction f) and austenite phase (with volume fraction a) contribute to the elastic response, with 5 + a = 1. The elastic strain rate is: 85: = Ell-7:1 old Eli-'21 =(1-a)Ei},n-a + “Elly-n. (2) where E 1:41, ij k I , 2,1,)“ are the respective stiffness tensor for martensite and austenite phase, respectively. For superelastic behavior, the transformational strain is dominant. The simplification that E Maj/Cl = E Asijkl = Ey-k] can be made so that: E _ -l 81-]. — Elmo/r1, (3) 1 The E5” is represented in terms of Young’s modulus E and Poisson’s ratio 0. If loading a tensile specimen of superelastic NiTi to 6% total strain using 400MPa tensile stress, it is assumed that 80% of the structure is stress induced martensitic phase. Without the simplification, the elastic strain is: 79 400Wa 400Wa 35:20% +80%——=1.1%, 80GPa 3OGPa With the simplification that E Mj-jk, = E Mk, = Egg, , the elastic strain is 5’3 = W :- 0. 5% . SOGPa There is 1.1%-O.5% = 0.6% difference in the elastic strain. This difference is 10% of the 6% total strain. So the simplification causes about 10% difference in the model’s behavior. Lubliner and Auricchio proposed a 3D frame work of the shape memory effect based on the generalized plasticity theory [21 7]. The transformation strain rate is determined by: .T . J3 1 (2 172 J), (4) where Si]- is the deviator stress, J 2 the second invariant of stress, and A the effective transformation (inelastic) strain, whose rate form is determined by the austenite volume fiaction rate: i=—56. (3 Here a is again the austenite volume fraction, whereas 6 is a material constant that gives the maximum transformation strain that can be induced by the shape memory effect. This maximum transformation strain 6 follows fi-orn crystallographic considerations and is typically about 5%. The austenite volume fraction a = a(x,t) is a material dependant function of current von Mises stress and the overall loading history. 80 The a value is calculated using the Ivshin and Pence constitutive model. Combining the elastic part and transformation part, the complete strain rate was expressed as: . .fi _1. ei'=—6a —S~ +E.. O'kl J (2J2 l1) ,1” (6) This constitutive relation between stress and strain is strictly in accordance with classical von Mises plasticity theory. The assumptions and limitations of the classical plasticity theory, like insensitivity to hydrostatic pressure and tension-compression asymmetry, were also implied in the model. Experimental results showed some shape memory materials had tension-compression asymmetry [77], and the phase transformation behavior shifted with elevating hydrostatic pressure [221, 222]. Inclusion of these factors into the current model will be the topic of future studies. 3.3.2 Formulation of Slip Plasticity Slip plasticity dominates the deformation of SE NiTi if stress-induced martensite phase transformation is exhausted as the dominant deformation mechanism. Indentation usually involved strains larger than 7 %. Plasticity was thus necessary to be included in the constitutive model for indentation modeling. A simple but effective algorithm is used in this constitutive model to shift the deformation mechanism from superelasticity to von Mises plasticity where ever the austenite volume fiaction has decreased to 5 % (95 % transformed from austenite to martensite). At this threshold superelasticity is replaced 81 by von Mises plasticity as the mechanism of plastic deformation. The strain rate is still decomposed in the same way as equation (1) with the elastic part expressed as equation (3) and inelastic part as (4). Equation (5) is replaced by an alternative expression so as to model von Mises plasticity with isotropic hardening. The complete form of equation (5) is expresses as: r 1 J5 . — S..o.--, a_<_0.05, von Mises lastici hZJJ—Z ,1 J ( P W) 9 (7) l _ 6a, (a > 0.05, superelasticity) pe ll where h is the plastic hardening slope. After the superelastic constitutive model is switched to the slip-plasticity model, a is taken as fixed. The plastic strain continues to change according to equation (7) as long as the plastic loading is in effect while .7 2 > 0. The von Mises stress and plastic strain is increasing as plastic deformation proceeded. Unloading occurs whenJ 2 < 0 . Initial unloading from the plastic deformation stage is elastic until the von Mises stress returns to the value associated with the previous transition from superelasticity to slip plasticity. Further unloading is governed by the constitutive model of superelasticity where the austenite volume fraction increases (a > 0 ). The constitutive model is implemented in ABAQUS finite element program by customizing the user material subroutine. It is first tested under simple loading conditions of uniaxial tension, biaxial compression and shear. The stress-strain curves 82 from this test are presented in Figure 55. The constitutive model can simulate the superelastic behavior in multi-direction loading directions. Plastic deformation in uniaxial loading results in the residual strain, which came from residual martensites. The materials properties used are representative of typical SE NiTi, but not for specific samples, which were summarized in Table 8. 3.3.3 Finite Element Program Implementation The finite element implementation is accomplished using the User Material Interface provided by ABAQUS. The strain increment is passed to the user material subroutine and the corresponding stress increment calculated from the constitutive relations using the return mapping algorithm. The state variables, including austenite volume fiaction and effective plastic strain, are also updated. The strain increment can be expressed as: J5 day. = —6da(—S,-j) + Egfildou, 2w: Multiplying both sides by elastic modulus gives the trial stress increment as: 73 l 2 The austenite volume fiaction increment is expressed as: do da- 6a eff do,j 6oefi. do,j (9) 83 where 0' eff is the von Mises stress (effective stress). 6a / 50817 is determined by the one dimensional Ivshin-Pence model. The rate form of austenite volume fiaction in Ivshin-Pence model is expressed as: d3: a(tK) do:max (6)6 6T + 6,3 610.5) da < 0 dt am(,B(T(tK ),o,fl(t,, ))) dfl 6T 6t 6o dt ’ dt ’ _c_l_a_ l—a(tK) mm(6,8 6T +66 610.17 fl) do >0 dt l-am (,6(T(tK),0',fl(tK))) (its 6T 6t 6o dt ’ dt ‘ In the modeling of superelasticity at a constant temperature, the above equation becomes: 6a = a(tK) docmax 6,8 da 50 6a ={ l—a(tK) }damm an da >0 50.1; l-am (fl(T(tr),a(tr))) dB 60.1; dt _ For isothermal phase transformation, the switching constant t), are the times of stress reversal. In the 3D generalization considered here the switching constants become times of von Mises stress reversal (equivalent to J 2 reversal). Insert equation (9) into (8), the elastic trial stress is expressed as: t _ 5680' :17” CE After Cijkl is calculated, the real stress increment dog. can be obtained by correction of the elastic trial stress increment d0;- The indentation model was comprised of 15000 axisymmetric 4-node bilinear elements. Part of the mesh around contact region is shown in Figure 56. The indenter 84 was modeled as rigid body sphere using the corresponding feature in ABAQUS to increase computation efficiency. The fiiction coefficient between the indenter and the substrate is assumed to be zero. A reference elastic-plastic material was simulated under the same indentation conditions for comparison. Its elastic modulus and Poisson’s ratio were the same as that of SE NiTi; its von Mises plasticity duplicated the loading stage of SE NiTi. The reference material properties are summarized in Table 8. 3.3.4 Validation and Discussion The absolute and normalized load-displacement curves of indentations are shown in Figure 57 for three different indentation depths. The recovery ratio r] is defined as a measure of the indent depth recovery: h max — h residual ’7 — h , max where hm is the maximum indentation depth and hmidual the residual indent depth after unloading. The recovery ratio 1] is smaller at deeper indentation depths for both SE NiTi and reference material. Experimental results has shown complete spherical indents recovery (:7 = 100%) on SE NiTi [11 7], which were reproduce computationally here at shallow indentation depth, by the modeled indentation-induced superelasticity. The reference elastic-plastic materials’ recovery ratios are much smaller than that of the SE NiTi at each indentation depths. 85 Tabor introduced the representative strain for spherical indentations: a 8, = 0.2 — , R where a is the contact radius and R the indenter radius [112], which is also reviewed in 2.5.2. Ni et al. found that the spherical indentation-induced SME has 100% recovery when a, s 5% [116]. This is comparable to results of uniaxial loading of SE NiTi: when uniaxial strain was smaller than 5%-7%, the recovery of deformation strain upon unloading can be complete. The recovery ratio vs. representative strain from FEA modeling is shown in Figure 58. However, the complete recovery of r] =100% is obtained only at the lower representative strain of 1.5%. The cause of this discrepancy may be due to the constitutive modeling. Another likely cause is the stress-strain concentration at the contact edge, discussed below. Figure 59 compares two computational load-displacement curves with experimental results from Ni et a1 [11 7]. The load-displacement curves are from two modeled NiTi materials with same properties but different phase transformation hysteresis A, where A = (A f +A s)/2-(MS+Mfi/2. The modeled SE NiTi with A=30K is used only in Figure 59, to show the effect of phase transformation hysteresis on load-displacement curves. It is obvious that larger A also increases the hysteresis of indentation load-displacement curves. The two computational curves match the experimental results quite closely. The A=15K curve is more close to the experimental results in the loading stage while A=30K curve is more close in the unloading stage. In contrast, the experimental SE 86 NiTi’s A is more than 70 K [11 7]. Better material calibration is needed to obtain a closer match between experiments and the finite element model. The distribution and size of the indentation-induced phase transformation region is difficult to determine from experimental measurements, but is easily available fi'om finite element modeling. Figure 60 depicts the indentation-induced phase transformation regions at peak load and at the end of unloading. The indentation depth is shallow so that the recovery ratio is 100 %. The contours of the transformed area consist of concentric hemi-spherical rings with maximum value at 0.6 a (contact radius) and size of 1.8 a. This corresponds to the contours of von Mises stress by the one dimensional constitutive relation. The shallow indentation does not exhaust the stress induced martensite phase transformation, where maximum austenite volume fiaction is less than 60 %. Reverse transformation upon unloading lead to a perfect indent recovery and there are very few residual martensites (Figure 58 (b)). In comparison, the reference material has a recovery ratio of 56 %. Figure 61 (a) shows that a deeper indentation resulted in a larger transformed area of 2.2a. The maximum transformed area is still located at 0.60 but inclined towards the contact edge, because there is stress-strain concentration at contact edge at deeper indentation depth. Although this concentration is not predicted by contact mechanics, the situation is also present in the modeling using reference materials and in results from other researchers [1 15]. The plastic strain contour in Figure 61(b) shows the largest plastic strain initiated at the contact edge. The plastic deformation impedes the reverse 87 phase transformation during the unloading stage and the indent recovery is not complete ()7 = 71%). Contours of residual martensite expressed as austenite volume fraction is shown in Figure 61(c), which has an intriguing irregular pattern.I More residual martensite segregates around the contact edge and near surface area and fade into austenite phase around 1.5a. The size of the indentation-induced transformed region is also plotted in Figure 58. The transformed region is approximately two times the contact radius, which increases slightly with indentation depth. In summary, spherical indentation on SE NiTi is investigated using finite element method, which incorporates a three dimensional constitutive model of the shape memory effect with plasticity: l) The computational load-displacements curves from spherical indentation modeling capture indentation-induced superelasticity, which has exceptionally large indent recovery ratios compared to elastic-plastic materials. 2) Martensite phase transformation is driven by contact stresses, whose hemispherical contours have an outer limit of approximately two times the contact radius for indents with perfect recovery, and increase in size at deeper indentation depths. 3) For shallow indentation depths, the stress-induced martensite phase transformation is the deformation mechanism and no plasticity is involved. The indents recover perfectly after unloading, leaving very few residual martensites. For deep ‘ The contours did not change with different mesh densities. Thus the pattern is not likely to come from numerical errors in the FEA. 88 4) indentations, plasticity develops first around contact edges. Plasticity impedes the reverse austenite to martensite phase transformation and results in partial indent recovery, where much more residual martensites are left after unloading. The computational load-displacement curves match experimental results closely. There is disparity between the phase transformation hysteresis of computational and experimental SE NiTi. The perfect indent recovery predicted by the finite element model occurs at the lower representative strains than experimental values. Closer agreement between experiments and finite element models would need more accurate input parameters from better material calibration. 89 Chapter 4 Results and Discussion 4.1 Tribological System with NiTi Interlayers The tribological performance of hard-coatings, such as chromium nitride and diamond like carbon, depends intimately on mechanical properties of the support structure beneath coating, which was review in section 2.6. In this work, the hypothesis of using superelastic NiTi interlayers to moderate contact stress and strain, and shield the soft substrates from plastic deformation, was tested by experiments. To investigate how the mechanical properties of NiTi interlayers affect the tribological performance of the whole coating system, it is best that the tribological performances can be compared with martensite and austenite (superelastic) NiTi interlayers, within one sample. The temperature scanning methods complemented previous wear test results at room temperature by showing that in one single sample the tribological performance could change dramatically with the phase transformations of NiTi interlayers. The various CrN:NiTi:Al specimen forms prepared in this study are designated by the following notation: ACrN: B NiTiAFzAl , where the superscript A is coating thickness (um) and B is interlayer thickness (um). The subscript “AF” is the austenite finish temperature A fof NiTi interlayer. For instance, a sample with a 3.9 um thick CrN coating and a 15 um NiTi interlayer with A f 353 K on an aluminum substrate would be 3'9CrN:liNiTi353:Al. A sample with a 2.5 pm thick DLC coating and a 21 um NiTi interlayer with A f368 K on an aluminum substrate would be Z'SDLC:21NiTi363:Al. 4.1.1 Residual and Phase Transformational Stress in NiTi Interlayers The measurement of phase transformational stress is to prove that NiTi interlayers on the aluminum substrates, even covered with hard coatings, can change the mechanical behavior of the whole composite structure significantly, due to martensitic phase transformations. It has been shown in temperature scanning X-ray diffraction (Figure 50 to Figure 52) that the NiTi thin fihns on aluminum substrates undergo thermoelastic martensite phase transformation. Figure 62 to Figure 66 show that NiTi interlayers, attached to the aluminum substrates and further covered by the hard coatings, could have the stress-strain response similar to those from flee-standing thin fihns. Figure 62 shows the stress evolution in NiTi thin films on aluminum substrates continuously over three heating-cooling cycles, while Figure 63 shows the details of the first heating-cooling cycle. Figure 64 shows the slope of the residual stress change vs. 91 temperature due to CTE mismatch. Figure 65 shows the residual stress change due to phase transformation. The residual stress in NiTi interlayers comes fi'om two origins: the intrinsic stress from PVD process, and the extrinsic stress from coefficient of thermal expansion (CTE) mismatch, which was discussed in section 2.3. 1. The aluminum substrates’ CTE is larger than that of NiTi thin films. Thermal stress change is tensile during heating and compressive during cooling. The CTE mismatch between NiTi thin films and silicon substrates gives -1 .0 MPa/K thermal stress rate (derivation can be found in section 2.3.1). The residual stress can be relaxed partly or completely by martensite transformation, and restored by the reverse phase transformation. The transformational stress relaxation rate is governed by the thermodynamics of martensite phase transformation, which is described by the Clausius-Clapeyron equation in section 2.2.2. The transformational stress relaxation and restoration cause a significant but reversible wafer curvature change/displacement within a relatively small temperature hysteresis of about 80K. The most significant problem encountered in the stress measurement is the plastic deformation of the 6061 T6 aluminum substrates occurs during deposition or heat treatment process where temperatures reached around 800K. This temperature is more than 2/3 of the melting point of aluminum (933K). The intrinsic compressive stress in NiTi thin fihns could plastically warp the aluminum substrate at this temperature to give a large plastic curvature. Were this curvature elastic, it would require 2-3GPa residual stress in the NiTi thin films, which is not possible for the NiTi thin films. It is difficult 92 to distinguish the plastic curvature and elastic curvature in the total curvature. The residual stress shown fi'om Figure 62 to Figure 65 is compressive (negative). The actual temperature could be lower than the measured temperatures because the thermocouple was inserted in a tight-fitting hole on the side of the furnace, not contacting the samples. There are some fine details about the phase transformations revealed in the stress evolution. First there are nonlinear stress relaxations before the phase transformations in both heating and cooling directions. The stress relaxation flattens to the theoretical 1.01MPa/K curvature to a smaller or even negative values before actual phase transformation starts. The R phase transformation might also be present in the second cycle (red curves) in the cooling direction (Figure 62). There was a stress relaxation about 20MPa before the martensitic phase transformation that can be caused by R-phase transformation. But the complicated precursor behaviors are not the focal point of the residual stress measurements here. Relaxation of the residual stress is also obvious in Figure 62 between three consecutive heating-cooling curves, because the stress loops do not overlap each other. The stress relaxation might also come from the creep of the aluminum substrate. But generally the phase transformation behavior of NiTi thin films on aluminum substrate is quite strong and stable. Figure 66 shows the curvature of 2'SDLC:'2NiTi3(,3:Al sample at different 93 temperatures, measured by optical profilometer scanning, from which the phase transformation stress in the NiTi interlayer can be measured. M; and A; temperatures were reached because further cooling and heating about 10 K did not change the curvature significantly. To support tribological experimental data presented from 4.1.2 to 4.1.7, it is necessary to consider the residual stress in the composite structure of hard coating:interlayerzsubstrate. Because of the residual stress in DLC coating, the stress in the NiTi interlayer, and their coupling, plus the plastic curvature in aluminum substrate, it is very difficult to obtain full details of the real residual stress. One important piece of information that can be obtained in Figure 66 is the residual stress change due to the phase transformation of the NiTi interlayer, with the covering DLC coating. The phase transformation relaxed 450MPa compressive stress in the NiTi interlayer when cooling from austenitic phase to martensitic phase. This is presented as proof that the NiTi interlayer has robust phase transformation that can affect the mechanical response of the tribological system. 4.1.2 Wear Resistance of CrN Coatings on 6061 Al Substrates with NiTi Interlayers The wear resistance was measured in terms of wear loss at different temperatures for samples with and without NiTi interlayers. The definition of wear loss is introduced in 94 2.5.4. Figure 67 shows the temperature scanning wear test results on 3'9CerlsNiTi353:Al and 3""CerAl samples using the 200m tip radius spherical indenter under 0.49N and 0.98N loads. 3 '9CrN215N iTi353:Al had two orders of magnitude improvement in wear resistance at temperatures above A ;, compared to 3""CerAl. For 3“S’CerlsNiTi353zAl, the superelastic (austenitic) NiTi interlayer gave two orders of magnitude improvement in wear resistance as compared to the martensitic interlayer. The improvement in wear resistance decreased as the temperature was decreased towards room temperature. At room temperature with a martensitic interlayer, the wear resistance of 3“S'CerisNiTi353zAl was slightly better (40% less for 0.49N load ) than that 3'9CrN2Al. The wear resistance of samples without a NiTi interlayer changed very little over the temperature range. The CerAl samples’ wear results had no temperature dependence while CrN:NiTizAl samples had a strong temperature dependence. Since the only difference between CrN:NiTi:Al and CerAl samples is the NiTi interlayer, the temperature dependence can only be explained by the temperature dependant mechanical properties of the NiTi interlayer. The improvement in wear resistance was thus apparently due to the superelasticity of the NiTi interlayer at temperatures above A ;. The superelasticity of the NiTi interlayer accommodated the contact strain, which was recovered after the wear counterpart was moved. Much less plastic deformation was left with each passing of the wear counterpart. The cyclic wear test amplified this effect. The wear loss of 3'S’CrN:'5NiTi:Al (Figure 67) with SE interlayer is 40 times smaller than the same sample 95 with a martensite interlayer. SEM images of the wear tracks where are shown in Figure 68 (0.98N load) and Figure 69 (0.49N load). The wear tracks from 0.49N shown in Figure 69 were similar to that in Figure 68, only the extent of deformation was much less. The better wear resistance is obvious in the less deformed wear tracks that were made above A ;. 0.98N load gave especially deep wear tracks in 3'9CerAl in Figure 68, with part of the CrN coatings delaminated. The wear track in 3'9CerAl also showed large spherical craters lying across the wear tracks. It is natural that improvement in wear behavior with superelastic interlayers under 0.49N load is more significant than that under 0.98N. Under 0.49N load, the improvement in wear loss with SE NiTi interlayers compared to 3‘9CerAl samples was higher than the case under 0.98N load. This is because lower load produced less deformation, and the percentage of the deformation that could be recovered by the superelastic response of the NiTi interlayer was larger, which means the superelasticity of NiTi interlayer was more prominent. During the wear test, the motion of spherical indenter pushed the soft substrate material forward and piled the material ahead of the indenter by plastic deformation. But with the increasing piling-up, resistance to the motion of the indenter also increased, which eventually forced the indenter to “j ump” across the piled material to a flat area. However wear tests on samples with superelastic NiTi interlayers did not show this behavior because the superelastic interlayer significantly reduced the plastic deformation 96 of the aluminum substrate. Wear tracks with SE NiTi interlayers in Figure 68 had a smooth and shallow geometry, showing strong evidence of superior wear resistance. Wear tracks produced at lower temperature with martensitic NiTi interlayer had a resemblance to the wear tracks in CrN:Al, also showing slight delamination of CrN and substrate deformation. It is quite remarkable that the wear resistance changed so dramatically in the same sample when temperature was changed less than 100K. The temperature scanning wear tests were repeated with a NiTiCu interlayer. The ternary NiTiCu interlayer has a narrower hysteresis than NiTi interlayer, which can be seen in the X-ray difiraction data (Figure 51). Tribological behavior can be potentially by influenced by the hysteresis of superelastic interlayers. Figure 70 shows the SEM images of the wear tracks at different temperatures. The wear track produced with the presence of superelastic NiTiCu interlayer was narrower and shallower than wear track produced with a martensitic interlayer. Figure 71 shows the wear data on l‘llCergNiTiCu3432Al samples. At room temperature, martensitic NiTiCu interlayer gave around 2 times improvement in wear resistance compared to the sample without interlayer, while 3'QCerlsNiTi:Al had only about 2 times improvement to compared to 3'9CrN:Al at room temperature. This may be because the l'7CrN:9NiTiCu343:Al had larger ratio of interlayer to hard coating thickness (5.3) than 3'9Cer'5NiTi353zAl (3.8). At temperature above A ;, the wear resistance of l‘7CrN:9NiTiCu343:Al is about 10 97 times higher than that the sample without interlayer. But the improvement in wear resistance with a superelastic interlayer was only 5 times compared to a martensitic interlayer. This number was smaller than the 80 times improvement in 3‘S’CerlsNiTi353zAl samples when the NiTi interlayer transformed from martensite to austenite. This can only be explained by the difference in the mechanical properties between NiTi and NiTiCu interlayers. The superelasticity of the NiTiCu interlayer may be worse than that of the NiTi interlayer, which means the NiTiCu interlayer had smaller recoverable strain. The smaller recoverable strain limited the improvement in wear resistance. It would be best if there are tensile test results of both NiTiCu and NiTi thin films so that the differences in mechanical properties could be compared. The SE NiTi interlayer thickness affects the wear resistance of CrN coating significantly. Figure 72 shows at room temperature how CrN and NiTi interlayer thickness affect wear resistance. The wear test was done using 6.35mm diameter WC under 5N load. For aluminum substrates, thicker CrN and NiTi interlayers improved the wear resistance. It is quite easy to understand that thicker CrN coating has better wear resistance, because thicker CrN coating bears more contact stress due to its high hardness, providing the increased residual stress associated with the thicker coating does not compromise interfacial adhesion. CrN coating’s intrinsic compressive stress scales with thickness [223-225]. Thick CrN coatings on steel substrates usually suffer performance degradation for the large intrinsic stress weakens the interface between CrN coating and substrate. This situation 98 is proved in the tribological tests of CrN:NiTi:M2, which is discussed in the next paragmph. The martensitic NiTi interlayer also increases the wear resistance, compared to samples without interlayers. There is 40 times improvement between l'7CrN:'5NiTi353:Al and l'7CerAl at room temperature. Where did this improvement come from? In scratch test results discussed in 4.1.6, it is obvious that the interfacial adhesion between CrN and NiTi at room temperature is worse than adhesion between CrN and Al. The improvement in wear resistance can not come fiom the adhesion of CrN coatings. It is more likely that the improvement in wear resistance comes from the better mechanical properties of martensitic NiTi interlayers compared to the soft aluminum substrates. The fatigue resistance of martensitic NiTi interlayer may be superior to those of the aluminum substrates. The martensitic NiTi interlayer may be able to accommodate large strain by aligning martensite variants and twins without introducing significant defects. The yield stress of NiTi interlayer (~200 MPa) [51, 52] is close to that of the fully annealed aluminum substrate (HRE 62, ~180MPa). 4.1.3 Contact Stress Induced Phase Transformation in NiTi Interlayer The experimental results discussed in the previous sections clearly show that contact stress induced phase transformation in NiTi interlayer has a strong influence on tribological performance. Here two topics will be discussed about the contact stress 99 induced phase transformation in NiTi interlayer that will provide some interesting information. The first is about contact loading conditions that trigger the phase transformation. The second is about finite element modeling of spherical indentation on CrN:NiTizAl. Similar to uniaxial loading of SE NiTi, whether there is superelastic effect in NiTi interlayer is also dependant on the contact load conditions. If the contact stress is too small, the NiTi interlayer behaves like normal elastic material, and there is no difference in wear resistance between martensite and austenitic NiTi interlayer. Figure 73 shows the temperature scanning wear tests on l'7CrN:15NiTi353:Al and ”CerAl samples using a 3.175mm radius WC ball under 0.98N. There is no improvement of wear resistance when the NiTi interlayer is superelastic at temperature above A ;. The reason is that under same 0.98N load, 3.175mm radius WC ball gave much less contact stress than the 200m radius diamond indenter used in section 4.1.2. The contact stress was not large enough to initiate the superelastic effect in NiTi interlayer. In other words, the contact strain can be accommodated by the elastic deformation of CrN coating and NiTi interlayer and superelasticity is irrelevant. In Figure 73, the martensitic NiTi interlayer improves the wear resistance of the CrN coating by nearly 100 times. This improvement is higher than the 40 times improvement in l'7Cerl5NiTi353zAl and l'7CrN: Al discussed in section 4.1.2 with Figure 72 because the load used here was decreased to 0.98N instead of 5N, but the wear counterpart remained a 6.35mm WC ball. It has been discussed that thicker martensitic 100 NiTi interlayers resulted in better wear resistance. So under lower wear load, the thicknesses of NiTi interlayers were virtually increased because the contact stress contours are geometrically similar (lower load gave smaller stress contours). Figure 74 plots the elastic contact stress in elastic half space under 3.175mm radius sphere and 200m radius sphere as a function of load using the equation introduced in 2.5.1: 3P [6PE *2 )“3 p0 _ 27m2 _ n3 R2 1 1 1 ’ _*. = + E E indenter E substrate where P is normal load, a is contact radius, R is the indenter radius, E. is composite modulus. The 3.175mm radius sphere under 0.49N has 558MPa elastic contact stress. This number is further attenuated by the CrN coating, and its intrinsic compressive residual stress, which in the end is not strong enough to trigger the superelastic response of NiTi interlayer. The elastic deformation of the CrN:NiTizAl composite structure would be enough to support the contact stress around 550 MPa. But on the other hand, the 200m radius sphere gave a very high elastic contact stress (>2000MPa), which was impossible to support elastically. It has to be mediated by the deformation in the coating:interlayerzsubstrate system. The superelastic NiTi interlayer can accommodate the deformation superelastically while a martensitic NiTi layer and the aluminum substrate cannot, which is why the wear resistance improvement is present only under higher contact stress/strain loading conditions. 101 4.1.4 Wear Resistance of DLC Coatings on Soft Substrates with NiTi Interlayers Temperature scanning wear tests were performed on the DLC:NiTizAl and DLCzAl samples. The DLC (diamond-like-carbon) coating has different mechanical properties and residual stress compared to the CrN coatings. How these differences change the tribological behaviors with the presence of NiTi interlayers would provide valuable information on this complex problem, along with more convincing evidence that SE NiTi interlayer could enhance tribological performance. First the DLC:NiTizAl samples also showed improvement in wear resistance with the superelastic NiTi interlayers at elevated temperature. Figure 75 presents the wear resistance of2'5DLCz9NiTi363:Al, 2-5I)Lc:‘5NiTi363:A1, 2‘5DLC:2‘NiTi363:Al and 2'SDLCzAl samples at different temperatures. The NiTi interlayers have 9pm, 15m, 21 um thicknesses. The SEM images of the wear tracks are compared in Figure 76. The 21 um SE NiTi interlayer improved the wear resistance 340 times compared to 2'SDLC:AI samples at temperature above A ;. The 15pm SE NiTi interlayer improved the wear resistance about 60 times compared to 2'SDLCzAl. This improvement is similar to the 3'9CerlsNiTi353zAl with 15 um SE NiTi interlayer, which had a 80 times improvement compared to 3'9CerAl. 2'SDLCz9NiTi363:Al sample only improved wear resistance about 4 times compared to 2"(’DLC:A1 sample. The thicker NiTi interlayers greatly increased wear resistance. The Z'SDLC29NITI3682A1 sample with a thin 9pm SE NiTi interlayer did not improve the wear resistance significantly, while 21 um SE NiTi interlayer produced a much better 102 improvement. Thicker NiTi interlayer (21 um) even improved the wear resistance of DLC coating at room temperature in martensite phase quite significantly (10 times), compared to the DLC:Al sample. The improvement in wear resistance as a function of interlayer thickness is not linear, but logarithmic. The reason of why thicker interlayers are better lies in the fact that contact stress and strain decays with depth. The basics of contact mechanics are reviewed in section 2.5.1 and 2.5.2. The contact stress and strain fields are hemispherical contours, the intensities of which decay with depth. For a given coating thickness, a thicker interlayer means larger portion of the contact strain is supported by the coating and interlayer, while the substrate is subjected to smaller stress and strain. Smaller contact stress and strain in aluminum substrates gave smaller plastic deformation. It is true that thicker NiTi interlayer costs more in deposition time and target material but the increase in cost is linear, which justifies the benefits of thicker NiTi interlayer. The DLC coating is known to have higher compressive stress than CrN coating but lower modulus and hardness. But the low friction coefficient of DLC coating makes it favorable in many applications. However no significant change in fiiction coefficient was observed with superelastic NiTi interlayer and martensitic NiTi interlayer in this work. 103 4.1.5 Wear Resistance of CrN Coatings on Hard Substrates with NiTi Interlayers The wear results from CrN:NiTi:M2 system are quite different fiom that fi'om CrN:NiTizAl system, which vividly illustrates the importance of the mechanical properties of supporting substrates. Figure 77 shows that the addition of a NiTi interlayer affects adversely on the wear resistance of CrN hard coatings. The increase in wear loss was biggest for 1.7 pm thin CrN coating and thick 12 um NiTi interlayer, showing that large interlayer/coating thickness ratio is not suitable for hard substrates. On the other hand, small interlayer/coating thickness ratio for 6 pm NiTi interlayer and 3.9 um CrN coating only caused moderate increase in wear loss compared to 3.9 um CrN coating without an interlayer. The best result is seen in a thick coating/interlayer combination: 3.9 um CrN coating and 12 um NiTi interlayer, which is slightly better than the case without the NiTi interlayer. Figure 78 presents the same data in Figure 77 but as a function of NiTi interlayer thickness. Although a thicker NiTi interlayer improves the wear resistance of samples with 2.8 um and 3.9 pm CrN coatings, the wear resistance is still not improved compared with cases without the NiTi interlayers. For samples without interlayers, the wear resistance was highest for samples with the intermediate 2.8um CrN coating. The wear results on CrN:NiTi:M2 systems indicate that the NiTi interlayer may not be suitable for improving the wear resistance of hard coating/hard substrate combinations. The modulus and yield stress of NiTi interlayers are substantially lower than the CrN and 104 M2 tool steel. As a result, strong interfacial acflresion must be realized for a relatively weak interlayer between hard coating and hard substrate to function as an adhesive layer. However, scratch test results in section 4.1.6 shows the evidence of a weak interface between CrN and martensitic NiTi. Although the substrate is aluminum in that case, the CrN and NiTi interface is still the same here on steel substrates. The weak interface may very well be the reason for the worse wear resistance of CrN:NiTi:M2 compared to CerM2. The interfacial adhesion problem can be solved, as shown in the DLC:NiTizAl system discussed in section 4.1.7. It may yet be possible to improve the wear resistance of CrN:NiTi:M2 by promoting the interfacial adhesion between CrN and NiTi in a future study. 4.1.6 Interfacial Adhesion between CrN and NiTi Figure 79 shows temperature scanning scratch tracks on a 3"”Cer9NiTi353zAl sample. There is a significant temperature dependence of the scratch resistance. At temperature above A ;there was minimum coating delamination and failure but there is very significant coating delamination and failure at temperature around MS in 3'9Cer9NiTi353zAl. Figure 80 shows the scratch tracks of 3'9CerAl samples that were made at the same temperature. There is no change in scratch resistance with changing temperature. This comparison proves that the temperature dependence of the scratch 105 resistance of 3'S’CergNiTi353zAl sample only comes fi‘om the NiTi interlayer, which has temperature-dependant mechanical properties. It is direct evidence that the stress-induced superelasticity of NiTi interlayer can improve scratch resistance and lead to better tribological performance. The scratch resistance can be quantified by a “critical load”, which is the normal scratch load where interfacial failure occurs. The critical load is determined after scrutiny of various experimental data: the appearance of scratch tracks, changes in friction coefficient, and acoustic emission signals. In scratch tests on 3'9Cer9NiTi353zAl samples, the acoustic emission signals, shown in Figure 81 and Figure 82, are good indication of the critical coating failure because of the sudden jump in acoustic emission when dramatic coating delamination happened. The critical coating failure load determined from acoustic emission can also be verified by inspecting SEM images of the scratch tracks (Figure 83 and Figure 84). The critical load determined fi'om acoustic emission and SEM images agreed each other well. It is shown in Figure 85 that 3'9CergNiTi353zAl has improved scratch resistance above A ;. The critical load jumped from 1.8N at room temperatures around MS to 4.2N at temperatures above A ;. For 3'9CerAl, the scratch resistance remains relatively stable across the temperature range. However it is noticeable in Figure 79 that 3'9CrN:9NiTi353:Al sample suffered more severe coating delamination than 3'9CerAl sample at temperature below A ;. The EDS obtained from delarninated region showed that the failure occurred at the CrN:NiTi interface (Figure 83) and CrN:Al interface (Figure 84), which suggested that the 106 CrN:NiTi interface was weaker than CerAl interface. Considering CrN and NiTi deposition took place at two different vacuum systems, the exposure of NiTi surface to air before deposition of CrN may contribute to the poorer interfacial adhesion. However the problem of insufficient adhesion is improved in subsequent DLC:NiTizAl experiments. 4.1.7 Interfacial Adhesion between DLC:NiTi The interfacial adhesion between DLC and NiTi was improved by two methods: interface chemistry and a sputter cleaning process. After deposition of NiTi and heat treatment, a lum pure Cr layer was deposited onto the NiTi surface without breaking vacuum. The NiTizAl samples were then transferred to the TEER deposition system for DLC deposition. Aggressive sputter cleaning (3600 ks) was employed to remove about 0.5 um of the Cr. Another thin Cr layer (0.1um) was then deposited onto the sputtered Cr surface before DLC deposition. The potentially weak interface between DLC and NiTi was replaced by cleaned and thin Cr surface. The reference samples without NiTi interlayer also went through this preparation procedures. Scratch tests were used again to measure the interfacial adhesion of Z'SDLCIISN1T13631AI and 2'SDLCzAl samples. The phase transformation temperatures of the NiTi interlayer were determined by temperature scanning X-ray diffraction (Figure 52). Figure 86 shows the scratch track on the DLCzAl sample produced by 107nm tip 107 radius spherical indenter and 7N linear progressive load, at room temperature. Figure 87 shows the scratch tracks on “DLC: ”NiTiggngl produced under the same conditions. There is very severe DLC delamination and failure in the 2'SDLC:Al sample (Figure 86), while no significant coating fracture occurs in 2‘SDLC:”NiTi3(58:Al (Figure 87) except at the edges of the scratch track. It is evident the interfacial adhesion between DLC and NiTi is much better than that between DLC and Al, even at room temperature with a martensitic NiTi interlayer. The delamination and dramatic failure of DLC coatings on aluminum substrates without NiTi interlayer at 7N scratch load made it unnecessary to conduct the scratch test at higher load. Figure 88 shows the scratch track on 2'SDLC:‘5NiTi363:Al produced by the very high 30N scratch load. This load is more than 4 times higher than the case in Figure 86 and Figure 87. Under the high scratch load, the DLC coating failure is inevitable, which is manifest as fine cracks inside the scratch track. However the strong interface adhesion kept the fractured DLC coating attached to the NiTi interlayer. Otherwise the fractured coating would have the fragmented appearance in Figure 86 and Figure 79. The delamination interface was determined by EDS spectrum shown in Figure 89. How does interfacial adhesion affect the appearance of the scratch tracks? Consider the scratch of 3'9CrN29NiTi353:Al in Figure 79. The destructive nature of scratch test inevitably leads to some damage to the coating, CrN or DLC. A strong interface keeps the coating adhered to the interlayer or substrate, so that the fractures, which accommodate the contact strains, are uniformly spaced over the scratch track. This is 108 the case in the scratch tracks at temperature above A; in Figure 79, also in 2'5DLC2'5NiTi3éngl (Figure 87 and Figure 88). If the interface is weak, the coating would be fiactured and then delaminated, damaged by the moving scratch head, resulting in the highly damaged look in scratch tracks at temperatures around Ms as seen in 3‘S'CergNiTi353zAl (Figure 79), and also in the 2‘SDLC:A1 sample (Figure 86). Whether there are dislodged coating fragments are good indication of whether there is good interfacial adhesion. The temperature scanning scratch test in the 2""DLC:'5NiTi;;(53:Al sample did not show significant temperature dependence, which is different than the temperature scanning test on 3‘9Cer9NiTi353:Al sample. The reason is that the interface between DLC and NiTi is intrinsically so strong that the influence of SE NiTi interlayer on the interfacial adhesion is greatly reduced. Due to the good adhesion, scratch at temperature around Ms did not cause delamination of DLC coating; the enhanced adhesion with SE NiTi interlayer at temperatures above A; still gives similar scratch tracks because fiacture in the DLC is something SE NiTi interlayer can not prevent. So under the improved adhesion between DLC and NiTi, there is no significant temperature dependence in the scratch tracks in 25DLC:"NiTi363:Al sample. 109 4.1.8 Finite Element Modeling of Indentation on CrN:NiTi:Al and CrN:Al Single-cycle spherical indentation on 3'S’CerlsNiTizAl and 3'9CrN:Al samples is simulated in FEA. Although the indentation loading is simpler than the cyclic sliding loading in the actual ball-on-disk wear tests, spherical indentation captures the main features of the contact loading conditions. Figure 90 shows the load—displacement curve from the indentation simulation. The maximum indentation depth in 39(3er lsNiTizAl and 3'9CrN:Al was 2.1um and 2.5um, respectively, because the SE NiTi interlayer had higher yield stress than the fully annealed aluminum substrate. After unloading, the 3'9CrN:'5NiTi:Al had only 1 pm residual depth while the 3'9CerAl had 1.9pm residual depth. The shallower residual depth in 3“S’Cerfi’NiTizAl probably results fiom a combination of strain-recovery in the SE NiTi interlayer and a decrease in the amount of plastic deformation occurring in the aluminum substrate. The result is generally consistent with the experimental result showing that the 3"’lCerlsNiTizAl provided better wear resistance. Figure 91 and Figure 92 show the strain on and eyy calculated by FEA of 3"’CerlsNiTizAl and 3'9CrN:Al, in the loaded and unloaded state. The contours for the loaded state of 3'9CerlsNiTizAl showed that the highest strain was at the NiTizAl interface, but without much strain mismatch across the interface; the strains in CrN:NiTi:Al were smaller compared to strains in 3'9CerAl. After unloading, the strain contours in CerA1 were similar to that at the loaded state. 110 Figure 93 shows the shear strain (exy) contours in 3'9CerlsNiTi:Al and 3‘9CrN:Al. A high shear strain of about 6.5% appears at the position under the edge of contact, which indicates that the material was being pushed outside of the contact zone. The hard CrN coating was not affected by this shearing flow due to its high hardness. As a result, the high shear strain gradient was located beneath the hard coating interface. In 3'S’CerAl, the majority of the shear strain was plastic deformation; in 3'9CrN:'5NiTi:Al, the NiTi interlayer accommodated most of the shearing strain superelastically. There were also a shear strain gradient across NiTizAl interface, which was caused by different material properties across the interface. Overall, the shearing strains in aluminum substrate were less in 3'9CrN:‘5NiTi:Al both in the loaded and unloaded state, as compared to ”CerAl. Figure 94 depicts the martensite volume fraction from stress induced phase transformation in loaded and unloaded states. There was a highly transformed area around the contact edge due to the stress concentration there. The lower center near the interface between NiTi and Al also showed more stress-induced martensite, which accommodated the strains from the vertical displacement of the spherical indenter. Figure 95, Figure 96 and Figure 97 show the stress components in 3'9CerlsNiTi:Al and ”CNN. The highest stress always was located within the CrN coating, although in the terms of strain, the CrN coating always had the smallest amount of deformation. This shows that the hard coating on softer substrate can shield the substrate from high contact stress, but not from deformation. On the other hand, the differences in stress components in 3'9CrN:15NiTi:Al and 3""CerAl are not as significant as that in strains. 111 This means that the addition of NiTi interlayer can shield the softer substrate from large plastic deformation. With the hard coating supporting contact stress and the SE NiTi accommodating contact strain, the finite element analysis proves that the CrN:NiTi is a useful combination for the protection of softer substrates. 4.2 Indentation-induced Two-way Shape Memory Effect The Two-Way Shape Memory Effect (TWSME) has been achieved in NiTi and other shape memory alloys under uniaxial and bending loading, which has been reviewed in section 2.4. In this work TWSME under contact loading conditions is demonstrated for the first time by two different training methods. Textured surfaces can have different tribological properties, for instance, lower fiiction coefficient, compared to smooth surfaces [226-230]. Surfaces with reversible texture that is controlled by temperature can then have variable friction coefficient and other controllable tribological and optical properties. Such surfaces can be realized using indentation-induced TWSME. The experimental demonstrations of indentation-induced TWSME are presented in sections 4.2.1 and 4.2.2. Before the discussion of the mechanisms of indentation-induced TWSME in 4.2.5, two experiments are introduced and discussed in 4.2.3 and 4.2.4, which reveal some interesting aspects of this complicated problem. 112 4.2.1 Indentation-induced TWSME from Thermomechanical Training Thermomechanical training of shape memory alloy is accomplished by cycling between temperatures above A; and below M; when a specimen is subjected to external stress or constraint, which has been reviewed in section 2.4.2. In this study a complex contact loading condition was used instead of simple axial loading conditions. However the experimental setup was simple. The details are described in section 3.2. Briefly, a 3.175mm diameter tungsten carbide sphere was pressed into a martensitic NiTi alloy sample to a certain depth and then the sample was cycled between temperatures below M S and above A ;. After the treatment, spherical indents exhibit two-way depth recovery with a shallow indent depth observed at high temperature and deep indent depth at low temperature. The reversible depth change is about 45% of the total indent depth and is stable over many cycles. Figure 98 (a) shows the 3D indent profiles taken at temperatures above A; and below Ms, respectively. Figure 98 (b) shows the cross section profiles of the heated and cooled indent. DA was the depth after heating the sample to 400:1:5K. Since this temperature was near A ;, the NiTi was in the austenite phase. DM was the depth after cooling the sample to ambient temperature of 30013K. Since the ambient temperature was close to M;, the sample was in the martensitic structure. After the first heating step, the indent depth decreased from the original indent depth of about 170 pm to 80 um (Figure 99), an approximately 53% indent depth recovery. This is reminiscent of the one-way indentation depth recovery reported by Ni et al., who 113 showed that the shape memory effect exists under complex loading conditions produced by spherical and sharp indentations [116, 231]. After cooling to the ambient temperature of 300K, the indent depth increased from D; of about 80 um to DM of about 140 um, an approximately 75% increase in the depth of indent. The subsequent heating-cooling cycles produced nearly constant indent depth ratio, (DM-DA)/DM, of about 45% (Figure 100). Figure 99 and Figure 100 show that the two-way indent depth change was quite stable over the five heating-cooling cycles tested. Both DA and DM slightly increased over the thermal cycles, with a small decrease in the two-way depth change DA-DM over the five thermal cycles. It is so far unclear the fatigue behavior of the two-way depth change (DA-DH) over more thermal cycles. The depth changed most significantly between the first and second thermal cycles. DA increased 2.5um; DM increased 6.3 um and DA-DM increased 3.7um. But their values stabilized over the next 4 cycles with DA increasing 0.75um/cycle, DM 1.4 uni/cycle and DA-DM 0.65um/cycle. Figure 101 shows the ratio of (DM-DA)/DM, which compares the depth change to the depth of the indent at martensite phase. The smallest ratio of depth change, around 0.5, is at the center of the indent. Although the ratio increases to 1 further away from the indent, both DM and DA become small and dominant by the surface roughness, causing the fluctuation in the ratio. Ideally the (DMDJ/DM should approach zero for surfaces far away from indents, because the surface does not move between austenite phase and martensite phase. But (DH-DA) is not zero because: 1) small surface roughness change 114 in NiTi between martensitic and austenitic phase, 2) measurement error between two measurements. The “ball-and-clamp” training used in this study was used to induce the low temperature shape into the sample by constraining the high temperature recovery. Under this “constant-strain” training procedure, the stress was not constant during the heating and cooling cycles. When heating temperatures above A ;, the stress under the sphere increased as the recovery was constrained. When cooling below M;the stress was decreased. Since the displacement of the clamped ball was constant, the only effect that thermal cycling had created was the alternation between high and low contact stress. The resulting “sink-in” of the area around the indent indicated that additional dislocations were generated during the training, which most likely occurred during the heating stage when contact stress increased. Undoubtedly plastic deformation was introduced in the martensite phase during the initial ball clamping or indentation process. The initial contact radius was about 0.8 mm. The representative strain 8}; : a, = 0.2-5’- : 0.2x 0'8 R 1.5875 =0.10 The 10% representative strain was high enough to introduce slip-plasticity in the NiTi sample. These dislocations would have a spherically symmetric distribution due to the symmetry of the spherical indentation. When first heated above A ;, the intensified contact stress due to constrained recovery must have generated additional dislocations 115 into the austenite phase. The process should be equivalent to the process of an unconstrained recovery followed by indentation to the same depth in the austenite phase. These dislocations first introduced in the austenitic phase are different in mechanics compared to the previous set of dislocations from indentation in martensitic NiTi. It is also very possible that these dislocations would move when the sample was cooled back to martensitic phase, because internal stress and strain field changed. When the sample was heated back to austenitic phase, new dislocations might be introduced because the previous dislocations in the austenitic phase have moved due to phase transformations. The training cycles might stabilize these two sets of dislocations so that no more additional dislocations are introduced during heat-cooling cycles. However it is also possible that the stabilization is not achieved during training so that new dislocations are introduced during each training cycle, which might contribute to the “sinking-in” effect described below. It is noticeable in Figure 98 that the area around the indent had a “sinking-in” effect in the martensite phase but was leveled in the austenite phase. The radius of the sinking-in area was about 1.8 mm, which was about 2.6 times of 700 um indent radius. This sinking-in area and its reversible change over heating-cooling cycles indicate that the training procedure has affected not only the microstructure beneath the sphere but also a large portion of material around the indent. Since the surrounding area is not directly under the indenter, either propagation of dislocations or residual stress field causes the “sink-in” effect. Additional research is necessary to determine the 116 mechanisms of indentation-induced TWSME. 4.2.2 One—cycle Deformation Training and Reversible Surface Topography The One-cycle deformation training method is an efficient and effective alternative training method to the thermomechanical training because it does not involve cyclic heating and cooling. Section 2.4.2 reviews previous research on the one-cycle deformation training method. The experimental procedures were described in section 3.2 of Chapter 3. First, spherical indents or scratches were made on the surface of a NiTi alloy in its martensite phase. Second, the indented or scratched surface was polished mechanically to restore a flat surface in a step referred to as “planarization”. Two-way reversible circular and line protrusions were thus produced by altering the temperature to drive martensite-to-austenite phase transformations. Figure 102 schematically illustrates the preparation method and the proposed mechanisms responsible for the two-way reversible surface protrusions [57]. Spherical indentation was performed on a martensitic NiTi sample so that after being heated above A ;, the indent recovered partially. The experimental details are described in 3.2 of Chapter 3. Figure 103 showed that residual indent had two-way reversible depth change with heating and cooling: indents were shallower in the austenitic condition and deeper in the martensitic condition. The average reversible depth change was 13.6:1:0.3 um over five thermal cycles tested, which is about 24% of the total indent depth. But the reversible depth change is not constant: it increased after first two thermal cycles. 117 This “self—training” effect is discussed in section 4.2.4. Reversible surface topography can be realized fi'om the indentation-induced TWSME. Figure 104 shows a 3X3 matrix of circular protrusions. Figure 105 shows the sectional profile of the reversible surface protrusion and its height over first five thermal cycles. Figure 106 shows the actual shape of the circular surface protrusion, with same length scale in the horizontal and vertical directions. Although in Figure 106 it seems the circular surface protrusion is not very significant because the aspect (height/ width) ratio is low, it is a strongly visible phenomenon. Surface morphology can be characterized by several different kinds of surface fluctuations: form error, waviness, and roughness (Figure 107). Surface roughness is generally referred to the shortest-wavelength irregularities of a surface, and waviness is the intermediate-wavelength deviation, and form is the longest-wavelength deviation. Reversible surface protrusions can be potentially created to change the roughness, waviness or form error, depending on the loading conditions of the indentations. The surface protrusion shown in Figure 106 creates a surface form error, which is a long-wavelength deviation from a flat surface. Figure 108 shows that imposing surface protrusions close to each other and Figure 111 shows that nanoscale protrusions can be created. These indicate that close-packed micro- and nanoscale protrusions can be made on the NiTi surface to change surface waviness and roughness. Figure 108 shows a reversible GM logo on the surface of a 2.54 cm diameter NiTi SMA sample. The images were taken from a optical microsc0pe, but also visible to 118 naked eye. The letters were made by joining circular surface protrusions closely adjacent to each other. The indenter and indentation load are the same as those in making circular protrusions shown in Figure 104, except that the indents are close to each other with around 0.8 mm spacing. Comparing Figure 106 with Figure 108, it seems the reversible surface protrusions are more significant optically than in physical dimensions. The reason is that the reversible surface protrusions are big change in terms of surface form. The polished flat surface of NiTi SMA had a roughness about 60 nm. The reversible surface protrusions have a height above 10 um, which is very big compared to the surface roughness. It is also shown in Figure 108 that making overlapping indents does not significantly diminish the reversible surface protrusions. Although the stress and strain field of each indent should be affected by the adjacent indents, it seems there is no detrimental effect. Currently it is unclear whether this would change fatigue behavior. Figure 109 shows a line protrusion after an indented and planarized bulk NiTi sample was heated to the austenite phase. These protruding structures disappeared when the sample was cooled down to the martensite phase. The height of the circular and line protrusions was 13.6:t0.1um and 0.83:0.05um over the first five heating-cooling cycles, respectively. The geometry and size of reversible surface topography (RST) can be readily controlled by indenter shape and applied load. Furthermore, intricate patterns of RST 119 can be efficiently laid out by arranging positions of indentations or length and direction of scratches accordingly. The reversible surface topography was also realized in sputter-deposited NiTi thin film on silicon substrates. This means the reversible surface topography could be readily integrated into the micro-electro-mechanical systems (MEMS). Figure 110 shows the same effect achieved in a 4.5 pm thick sputtered NiTi thin film deposited on a (100)-silicon substrate. The height of the circular protrusion was 33.2:h7 .2 run over the first five temperature cycles. However the protrusion deteriorated over the first five thermal cycles more significantly than RST on bulk’NiTi samples (Figure 112), which is discussed in section 4.2.3. Figure 111 shows the reversible line protrusions on NiTi thin films, which were planarized from a scratch that was produced by manually sliding a razor blade on the NiTi thin film. The line protrusions had three parallel lines. This may had come fi'om some imperfections in the razor blade. There was also a non-reversible “x” shaped line protrusions nearby, which were from scratches in mechanical polishing. Indeed, not every surface deformation in NiTi SMA could give reversible surface protrusions. Spherical indentation and sliding were shown to be effective loading conditions. But it is possible that more “blunt” indenters, like flat end punch, or more “sharp” indenters, like Berkovich indenter, could not successfully induced TWSME, which is for future research to clarify. The RST in NiTi thin fihn on silicon substrate shown in Figure 110, deteriorated over 120 the first five thermal cycles (Figure 112). It is quite different from the RST in bulk NiTi. This can be explained by the effect of the NiTi thin film thickness. The indent diameter is about 7 pm. Therefore 7 pm NiTi thin fihn was needed to fully contain the deformation zone that supported the TWSME (see section 4.2.3 for detail). But the NiTi thin fihn thickness used in the experiment had a thickness of only 4.5 pm, less than the contact diameter size that needs to support reversible surface protrusion. The compromised deformation region underneath indent caused the deterioration of reversible surface protrusion in Figure 112. 4.2.3 Layered Planarization Insight into the mechanism underlying cyclic TWSME fiom indentation of NiTi-based shape memory alloys can be obtained by experiments in which successively deeper surface-removal steps are performed, until the reversible displacement effect is found to vanish. Results of such an experiment are shown in Figure 113. An indent was again made in martensitic NiTi, with a 1.59 mm diameter ball, and the specimen polished just sufficiently to remove any trace of the indent, while maintaining the temperature below Mf. After heating past A fthe height of the thermally-reversible surface protrusion was determined, and subsequently measured again as additional surface layers were successively polished away in each case, while the specimen was martensitic. The protrusion amplitude was found to monotonically decrease fi'om initial 121 maximum of just over 11 pm until it vanished after ~05 mm (extrapolated) of the surface had been removed. Protrusion formation thus apparently involves transformational behavior in a subsurface zone nearly 40 times the protrusion height, indicating that the effect can be produced with an average of ~2.3% transformational strain in the subsurface volume, well within the known capability of TWSME in NiTi alloys [4]. Furthermore, when the contact mechanics of indentation are considered, the depth at which the cyclic protrusion amplitude vanished is consistent with existing theories on the mechanisms of TWSME in NiTi alloys. Figure 114 adapted from Sarnuels and Mulhearn [232] shows theoretical predictions and experimental results that give the distribution of plastic strain beneath a spherical indenter in Cu-30%Zn brass. Contours are shown for plastic strains of 7%, 1% and 0%, the latter contour corresponding to the elastic-plastic boundary. Of particular significance is the location of the 7% contour, occurring at a depth approximately equal to the radius of the contact perimeter of the indent. In the present experiment, the initial indent contact radius was 450 +/- 25 pm. Therefore, the surface removal experiment, in which the effect vanished after removal of ~0.47 mm (roughly equal to the initial indent contact radius), indicates that cyclic protrusion formation depends on TWSME occurring in material that experienced more than 6 to 7% plastic strain (in the martensite phase) during indentation. This corresponds closely to the maximum strain (6-8%) that can be tolerated by thermal martensite before the onset of plastic flow via dislocation production. It is therefore concluded that the initial indentation event created, within a near-surface zone whose 122 depth corresponded to the indenter radius, sufficient numbers of dislocations in the martensite that could later cause a shape strain to occur during the austenite-to-martensite transformation. At greater depth below the indent, where smaller plastic strains occurred, little dislocation production is expected since strains in this region can be accommodated entirely by martensite variant-boundary motion (deformation twinning). This deeper material effectively self-accommodated the transformational shears on cooling, and was therefore unable to contribute significantly to TWSME in the present context. 4.2.4 Small Hysteresis and “Self Training” of Reversible Surface Protrusion The relaxation and recovery of surface protrusion has a much smaller temperature hysteresis compared to the phase transformation hysteresis of annealed NiTi alloy sample. Figure 115 showed the surface protrusion height vs. temperature, for planarization of a spherical indent that was produced under 60kg load using 1.587mm diameter ball. The protrusion started to rise before the nominal A S temperature. Before the temperature reached (A s +A ;)/2, surface protrusion height has already reached 90% of maximum height, with a rate 0.26um/K. During the cooling stage, the surface protrusion decreased slowly before MS but started a rapid descent well above MS. Again 90% of the protrusion height was diminished before (MS+M;)/2 with a rate 0.31 rim/K. 123 As can be seen in Figure 115, the martensite-austenite phase transformation temperatures shifted to lower temperatures in the heating direction; the austenite-martensite transformation temperatures shifted to higher temperatures in the cooling direction. The hysteresis of the reversible surface protrusion was 30:1:3K compared to the phase transformation hysteresis of 60i5K, from DSC measurement of free standing sample (Figure 32). How could, in TWSME, the martensite—austenite transformation temperatures shifted downwards to lower temperatures in the heating direction? Stalmans et al. noticed the lowering of phase transformation temperatures in TWSME when the sample was subjected to external stress that was “opposing” the direction of TWSME, and provided a plausible explanation from the Clausius-Clapeyron equation [106] ( reviewed in detail in section 2.4.3) . The main reason is that the directions of the transformation strain in TWSME are set by the training process; the external stress posed afterwards can either be “along” or “opposing” the direction of TWSME. The former elevates phase transformation temperatures and the later suppresses phase transformation temperatures. For instance, if a specimen was trained to possess such TWSME: shrinkage occurs during austenite-martensite transformation in the cooling direction (Ate;w < 0) and elongation occurs during martensite-austenite transformation in the heating direction (A813 > 0 ). The tensile stress is in the opposite direction of TWSME because it suppresses the strain change in the cooling direction. 124 The Clausius-Clapeyron equation can be used to explain this by simply considering the sign of enthalpy, strain and stress. The C-C equation is: Ao _ AHfi’ AT Maj" Tensile stress lowers the phase transformation temperatures as follows: on in cooling M from austenite to martensite phase, A1124 > 0, A87 < 0 and :54” < 0. The left hand 8A side of the Clausius-Clapeyron equation :—; must has the same sign as the right hand M A side, Since Ao > 0 (tensile), AT must be negative to give :—; < 0 so that M. 8A the left hand side of C-C equation was negative. Here the tensile stress lowers the phase transformation temperature in the cooling direction. A In the heating from martensrtlc to austemtlc phase, TA ”f, 18 stlll negatlve because 8M AH A”; <0, A8,”; >0. Both enthalpy and strain change flip sign. Note that in one-way SME, the strain change does not change sign. Since Ao > 0 (tensile), AT is still negative. Therefore, small hysteresis can be obtained in indentation-induced TWSME only if the residual stress meets the following requirements: It must be in the opposite direction of TWSME during heating so that phase transformation temperatures are lowered; It must not be in the opposite direction of TWSME during cooling so that phase transformation temperatures are elevated. 125 If the same TWSME specimen in the previous paragraph were subjected to compressive stress during cooling (A324 < 0 ), the austenite-martensite phase transformation temperatures would be shifted upwards in the cooling direction; if the stress was changed to tensile during heating (As A: > 0 ), the martensite-austenite transformation temperatures were shifted downwards in the heating direction. Thus small hysteresis can be achieved by moving the transformation temperatures towards each other. In the above situation, the direction of the stress can actually assist the deformation: tensile stress with positive strain change and compressive with negative strain change. This could stabilize the TWSME. A similar situation may be what actually happened in the small hysteresis behavior of reversible surface topography, which is shown in a schematic in Figure 116. It is known that residual stress exists in elastic-plastic indentation [108]. The distribution and extent of the residual stress in indentation-induced TWSME is very complicated, considering the coupling of martensite phase transformation and planarization process. However the residual stress may very possibly act in the direction that elevates the phase transformation temperatures during heating and suppresses the transformation temperatures during cooling, creating the small hysteresis. Other experimental evidence comes from the “self training” effect of the indentation-induced TWSME. As can be seen in Figure 103 (reversible depth change from one-cycle deformation training) and Figure 105 (reversible circular surface 126 protrusions), initial thermal cycles after training process did not lead to immediate deterioration of the TWSME, but instead acted to strengthen the TWSME slightly. It was in sharp contrast to the fatigue behavior of uniaxial TWSME, where the two-way effect deteriorated significantly during the first several thermal cycles [101]. It has been pointed out in the previous paragraphs that for TWSME to have small hysteresis, the direction of external stress must be in directions to assist the deformation. The residual stress in indentation-induced TWSME may very possibly act this way, strengthening the TWSME, curbing the deterioration, and narrowing the temperature hysteresis. 4.2.5 Mechanism of Indentation-induced TWSME The successive layered planarization experiment revealed that a zone underneath indent with the size of the indent diameter is mainly responsible for the reversible surface topography. Theoretical calculation based on contact mechanics and experimental measurements by Samuel and Mulhearn [232] also pointed out this zone is a highly deformed region with at least 7% plastic strain. In one-cycle deformation training of TWSME, it is believed that the dislocation structure and the associated stress field create rrricrostructural, thermodynamic and stress anisotropy within the martensite phase, as discussed in section 2.4.2. The thermodynamic anisotropy means that after training, the crystallographically equivalent martensite variants are no longer thermodynamically equivalent: the trained martensite 127 variants are thermodynamically preferred to the non-trained variants. The thermodynamic anisotropy must have come fi'om certain changes in microstructure as the result of training. The microstructural basis of the thermodynamic anisotropy was termed microstructural anisotropy. As a result, martensite variants in certain orientations are thermodynamically favored over others when cooled from the austenite phase. The preferred nucleation and growth of these directional martensite variants enable the SMA to “remember” its low temperature shape. As is shown section 2.4.2, slip-plasticity is effective to introduce TWSME under uniaxial loading. In indentation-induced TWSME, dislocations should play a similar important role to induce the reversible effect. Comparing to simple loading conditions such as tension and compression, indentation and scratching are effective to introduce dislocations locally and symmetrically. Several authors have suggested that indentations are accommodated by circular loops of “geometrically necessary dislocations” under spherical indenters [233, 234] (Figure 117). It is assumed that the indentation is accommodated by circular loops of geometrically necessary dislocations with Burgers vectors normal to the plane of the surface. As the indenter is forced into the surface of a single crystal, geometrically necessary dislocations are required to account for the permanent shape change at the surface. These geometrically necessary dislocations lead to spherically symmetric microstructures and residual stresses beneath the indents, which may induce TWSME. On the other hand, the small hysteresis of the reversible surface topography and 128 “self-training” effect indicated the residual stress played a significant role in altering phase transformation temperatures and stabilizing the two-way effect. However, the distribution and extent of the residual stress in indentation-induced TWSME are very complicated, especially considering the coupling with the martensite phase transformation, which undoubtedly changes the residual stress significantly. The residual stress and strain fields in an as-indented martensitic NiTi should be close to those in an elastic-plastic material. But how the stress and strain fields change after the first recovery is unclear. How those change in the following thermal cycles would be a useful finite element analysis to shed some light on indentational TWSME. Indentational shape memory effect has not been modeled in this work due to the difficulty determining appropriate boundary conditions. Therefore a finite element analysis with general approximation was made here, which is an indentation in an austenitic NiTi. The reason is that the stress-strain field in austenitic NiTi, which includes the effects of stress-induced phase transformation, may be close to those after a thermally driven phase transformation in an austenitic NiTi. Figure 118 presents a finite element analysis of the ow residual stress field (normal to the surface) in austenitic NiTi after spherical indentation. There exits a highly compressive stress “core” extending fiom one quarter of the indent diameter to one indent diameter beneath the residual indent. The compressive stress also agrees with Figure 116, where there should be a compressive stress at temperature above A fto pull the surface protrusion down. When cooling down from the austenite phase to the 129 martensite phase, this compressive stress in the vertical direction may be able to help pull material back from surface, which would create a deeper indent. How this residual stress evolves in phase transformations by heating and cooling would help provide more information. 130 5.1 Chapter 5 Conclusions and Future Work Conclusions: Hard Coating Systems with NiTi Interlayer Based on temperature-controlled scratch and dry sliding wear experiments on CrN and DLC coated 6061 T6 aluminum, tested both with and without NiTi interlayers, the following conclusions may be drawn: 1) 2) 3) Magnetron sputter deposition is capable of producing excellent adhesion of NiTi on aluminum substrates, without elaborate surface preparation or other adhesion-promoters. The NiTi thin films on aluminum substrates display robust martensitic transformational behavior, as shown in temperature-scanning X-ray diffraction and wafer-curvature measurements. The wafer-curvature measured on DLC:NiTizAl also shows strong phase transformational stress relaxation and recovery in the NiTi interlayer. Superelastic NiTi interlayers on very soft aluminum substantially improves adhesion of nitride hard coatings, as indicated by an increase in the critical load for delamination as measured in temperature-controlled scratch tests. The marked increase in performance at test temperatures near A fstrongly suggests that stress-induced A->M transformation is responsible for the observed 131 4) 5) 6) 7) improvement. The adhesion of hard coatings to martensitic NiTi interlayers can be improved by deposition of a protective Cr layer and subsequent aggressive sputter cleaning. This procedure could also be replaced by depositing NiTi interlayers and hard coatings within the same vacuum chamber. NiTi interlayers dramatically improve the wear behavior of CrN and DLC films on soft aluminum and can increase wear resistance by a factor of 100-3 00. The best performance is noted for test temperatures near or above A f. Martensitic NiTi interlayers can improve the wear resistance by a factor of 40—100. Thicker NiTi interlayers are more effective. Performance monotonically improves as the interlayer thickness is increases to 21 um. Thicker CrN coating also increases wear resistance monotonically for thicknesses up to 4 pm. It is unclear now whether there is an upper limit to hard coating and interlayer thickness on aluminum substrates. The NiTi interlayers between CrN and hard M2 steel substrates do not improve wear resistance significantly, compared to the samples without NiTi interlayers. It is suspected that the weak CrN:NiTi interface is responsible for the worse wear resistance of CrN:NiTi:M2. However the interfacial adhesion can be improved, which might lead to better wear performance of CrN:NiTi:M2 in future studies. Placing an interlayer that is softer and more compliant than the substrate can lead to larger strains in the hard coating, and larger interfacial 132 ‘8) mismatches. Finite element analysis proves that the use of superelastic NiTi interlayers can shield the soft aluminum substrate fi'om contact strains, while the CrN hard coating can support contact stress. The combination of CrN hard coating and superelastic NiTi interlayer successfillly protects the aluminum substrates. 5.2 Conclusions: Indentation-Induced TWSME 1) 2) The two-way shape memory (TWSME) effect can be realized at the micro- and nano-scale using spherical indentation. Both thermomechanical cycling and slip-plastic deformation were proven to be effective training methods to induce the two-way shape memory effect. Thermomechanical cycling results in larger reversible indent depth change and a “sink-in” effect. The slip-plastic deformation is an efficient one-cycle training method and there is no “sink-in” effect. TWSME does not decay significantly over initial thermal cycles. Reversible surface protrusions (RSP) are produced fi'om planarizing spherical indents with TWSME. The height of the RSP is the same as the two-way indent depth change, which means the planarization process does not affect the TWSME. Layered planarization experiment shows that the RSP arises from transformational activity in a zone with the size of the contact radius beneath the indent, which is about the boundary of 7% plastic strain. 133 3) The residual stress underneath the indents may possibly strengthen the TWSME. The RSP has a smaller temperature hysteresis than that of the bulk NiTi SMA. Second is the “self-training” effect, which means the RSP does not decay but strengthens slightly during initial thermal cycles. From the above two phenomena and thermodynamic analysis, it is concluded that the residual stress from contact loading may be able to promote the TWSME and RSP. 5.3 Suggestions to Future Work 1) 2) 3) Mechanisms of indentation-induced shape memory, superelasticity and two-way shape memory effect. New experimental tools like synchrotron X-ray diffraction and in-situ TEM indentation can be used to analyze the microstructures beneath indents in NiTi alloys, along with traditional experimental techniques like cross-sectioning samples with indents. Tribological applications of indentation-induced two-way shape memory effect. Surfaces with reversible surface protrusions could have different fiiction coefficients compared to flat surface. The difference in friction coefficients is also dependent on testing conditions like loading configurations and lubrications. Development of finite element modeling of shape memory effect is the next step after current modeling of superelasticity. FEA of shape memory effect can provide stress-strain distributions beneath indents in NiTi before and after phase 134 4) 5) transformations, which can be used to explain indentation-induced shape memory and two-way shape memory effect. Tribological tests on CrN:NiTizAl and DLC:NiTi:Al samples with re-heat treated aluminum substrates. The temper of 6061 T6 aluminum substrates were destroyed during NiTi interlayer deposition, which led to low hardness (HRE 62). It is possible to re-heat treat the aluminum alloy to restore the hardness to HRB 60, after the deposition of NiTi interlayers and hard coatings. Finite element modeling using cohesive elements and user material subroutine for superelasticity. The cohesive elements can be used to model interfacial delamination between coatings, interlayers and substrates. It is possible to show in FEA that superelastic NiTi interlayers discourage interfacial delamination under contact loading, which supports experimental results. Parametric studies can also be carried out to find the optimal mechanical properties of hard coating, NiTi interlayer and substrate to maximize tribological performance. 135 Appendices Table 1. Comparison of crystallographic data of NiTi martensite a, Otsuka et a1 [20]; b, Hehmann and Sandrock [235]; c, Michal and Sinclair [21]; d, Kudoh et a1 [22]. 088 (a) HS (b) MS (c) KTSO (d) a (run) 0.2889 0.2883 0.2885 0.2898 b (nm) 0.4120 0.41 17 0.4120 0.4108 c (nm 0.4622 0.4623 0.4622 0.4646 [1 (°) 96.8 96.8 96.8 97.78 v (nm’) 0.05463 0.05449 0.5455 0.05479 Atoms per unit cell 4 4 4 4 SPace 8mm ”’0 P21/m P21/m P21/m Table 2. Lattice structure and parameters of precipitates in NiTi alloys Name Ti3N14 Ti2Ni3 TiNi; TiN i2 Space group R3 I4/mmm D024 Fd3m (rhombohedral) (tetragonal) (hexagonal) (cubic) Lattice a = 0.67nm a = 0.3095nm a = 0,501,222), a = 1.132nm parameter a = 113.8° c =1.3585nm c = 0.8307nm c/ a = 1.6284 Reference [236, 23 7] [238] [239] [240] 136 Table 3. steel substrates, before CrN deposition Deposition parameters for NiTi interlayers on 6061 T6 substrates and M2 Target Deposition Substrate- Deposition Ar Pressure Composition Power Target Distance Temperature (Pa) (at.%) (W) (mm) (K) 53.5 Ti 46.5Ni 200 0.35 63.5 483 Film Deposition Time Base Pressure NiTi Film Heat Composition (hour) (Pa) Thickness (um) treatment (at.%) 800K for 51. 7 Ti 48.3 Ni 0.45,].5,2.5 5.2><10'5 4.5, 9, 15 7200 ks Table 4. Deposition parameters for NiTiCu interlayers on 6061 T6 substrates, before CrN deposition Target Deposition Substrate- Deposition Ar Pressure Composition Power Target Distance Temperature (Pa) (at'/o) (W) (mm) (K) 53.5 Ti 8 Cu 200 0.35 63.5 743 38.5 Ni Film Deposition Time Base Pressure NiTi Film Heat Composition (hour) (Pa) Thickness (pm) treatment (at.%) 51.7 Ti 6.] Cu 793K for 1.5 5.3 x 10'5 9 42.2 Ni 7200 ks 137 Table 5. Deposition parameters for NiTi interlayers on 6061 T6 substrates, before DLC deposition Target Deposition Ar Pressure Substrate- Deposition Composition Power (Pa) Target Distance Temperature (at.%) (W) (Hill!) (K) 53.5 Ti 46.5 200 0.53 63.5 523 Ni Film Deposition Time Base Pressure NiTi Film Heat Composition (hour) (Pa) Thickness (um) treatment (at.%) 773K for 51.7 Ti 48.3 Ni 25,35 2.0x10“ 9,15,21 9000 ks Table 6. Deposition parameters for NiTi interlayers on M2 tool steel substrates, before CrN deposition Target Deposition Substrate- Deposition Ar Pressure Composition Power Target Distance Temperature (P8) (at.%) (W) (mm) (K) 53.5 Ti 46.5 Ni 200 0.33 63.5 723 51.2 Ti 48.8 Ni Film Deposition Time Base Pressure NiTi Film Heat Composition (hour) (Pa) Thickness (pm) treatment (at.%) 51.7 Ti 48.3 Ni 793K for 1,2 4.3x 10'5 6,12 48.2 Ti 51.8 Ni 7200ks 138 Table 7. Phase transformation temperatures of thin film and bulk NiTi used in TWSME study . , Martensite Austenite start Austenite finish Martensite start finish (As) (A1) (Ms) (M1) Bulk NiTi 350K 404K 344K 287K Thin Film 343K 390K 329K 282K NiTi Table 8. Material parameters of NiTi and reference materials used in finite element modeling. (The “Ref” stands for reference material) Austenite Austenite Martensite start , start Martensite finish finish Phase transformation (111;) 318 K 338 K 323 K 303 K k1 *2 k3 *4 k3 parameters 0.1946 -7.784 0.1946 -10.622 . . Elastic Plastic . . Yield Transformation Testing Modulus hardening Porsson’s stress Strain Temperature , , K slope It ratio (NlTl and Ref) (Ni’l‘i and Ref) (Ref) 5% 90K 80 GPa 10 GPa 0.3 533.5 MPa I39 2000 8’ . Q 00 _ 1455 3 15 1380 e u or no a. 1304 E 1118 Ni 1- TiNi looo - . Ni3'fi2 “‘8“ . Nuria‘l' }Metastable 500 . . . . . I I . . . . . T 20 “ZN' 40 , 6o "'3“ 80 Ni Atom % Nl Figure 1. Phase diagram of a NiTi alloy by Massalski[241], to which the phase equilibrium between the B2 and Ti3Ni4 phases are added [242]. T" .. i ‘ (C) Figure 2. Transmission electron micrographs of Tl3Nl4 precipitates in 51.3at%NiTi thin films which were solution-treated at 973 K for 3.6 ks followed by age-treatment at 773 K for (a) 3.6 ks; (b) 36 ks and (c) 360 ks, at 673 K for (d) 3.6 ks; (e) 36 ks and (f) 360 ks, and at 573 K for (g) 3.6 ks; (h) 36 ks and (i) 360 ks (after reference [51]). 140 Parent Phase Martensite I ’ e— martenslte matrix e—twln Invariant plane (habit plane) Figure 3. Schematic illustration of parent and martensite with invariant plane as a habit plane. 141 (010)82 (100182 Figure 4. Stereographic projection of 24 habit plane variants from WLR theory, for <01—1> Type II twinning. 142 m _1 L l l 1 1 l l 350 , r A 310-4 r- 5 g 250~ - (B 5.; aoo~ - Q E. 1' 150 d o Kormllov E” + Hanlon 100 - v Melton '- x Miyazaki 50d 0 Nishida _ In Wasilewski a Smith 0 r r 1 W r 1 r r 0.48 0.49 0.50 0.51 0.52 Mole-Fraction NI Figure 5. Ms temperature as a function of Ni content for binary NiTi alloys. Different data symbols represent data from different authors. The solid line is given by thermodynamic calculations (after reference [34]). 143 1.2 GPa 600 1\ q 1 1.5 W . ~ ~ 1 ., I m w 85 . E ., g o ‘L — ““““““ 1.— -200 400 -600 ’2 x 00mm We: soo P1>Pz>P3 e 0 473 573 873 Temperature, K Figure 6. Schematic stress-temperature diagram for isochronal annealing of sputtered NiTi films on (100) silicon. Dark curves P1, P2 and P3 show the general intrinsic stress—deposition temperature characteristics for different pressures of the working gas. Trajectories a, b, c and d are for amorphous films deposited at different pressures and at temperatures below T a), the minimum in-situ crystallization temperature. Isochronal crystallization occurs at Tc giving a transient tensile stress spike at x. Curve d additionally shows the effect of an isothermal hold inserted in the isochronal sequence. Trajectories e, f and g are for crystalline films deposited above T 04. Cooling from any point on a trajectory generally proceeds along a path with slope 1.5 MPa/K, to points such as d‘, a‘ and g‘, allowing a wide range of stress levels to be obtained at T ,1. (after reference [58]) 144 153.0 K 164.1 K O '82.?“ 285K l93.3l( ~203.0K ' 25LOK ? 24I.OK . w - g 223.7K —o r a w o l = Q a, A A - # A A __- 3 O O 0 O '5 2765K 5 273.2K I" 263.4K (9) 283.7 K M9190 K Af=221 K o 2 4. 2 4o é‘To 2 4‘ Strain (“/o) Figure 7. Stress—strain curves as a function of temperature of 50.6 at.% NiTi alloy, which was solution-treated at 1273 K for 1 h. The critical stress for inducing martensites and that for reverse transformation are indicated by an arrow (—)) and by a double arrow (—>->) in 0), respectively. (after reference [68]) 145 Tensrle Stress (MP0) 5 8 zoo 04k ° F 0‘0 O/ . so / \o O \ / ’I / O ’1 /. o_‘_€} L j 1; v A L n :50 “s 200 A, 250 300 Temperature (K) Figure 8. Critical stress as a function of temperature for inducing martensites (open circles) and for reverse transformation (solid circles), which were plotted from the data in Figure 7. (after reference [68]) 146 I drape memory effect " " ' ".o‘ 0“.“ ~“.:.'o" ':. .3... .. a . 0.: :.o. o O...- ... . a .. .- . :- +§;.I~... '0‘- ., ;".t. ...v.._ $ , . ‘. "P .- .x. . h: 9 . 0 , ., 37:1... IQ crtlcal stress for sllp (B) -.. . lit-.1 o .:.‘._.. 3.. .. _ '. V ~5‘~~‘- ',.v .' tf";'; ' \i ‘5‘. ' ‘-" .. :'- ' ' '.° ' 3f Mst AsAf Temperature —> Figure 9. Schematic representation for the appearance of the shape memory effect and superelasticity in temperature—stress space. (after reference [4]) 500' Ti-50.3%Ni single crystal 1273 K IQ- 673 I0 400 - Ms=244 K ,. Af =2e4 K ', ’ Tensile Stress (MPa) 8 or 9 8 100 0 R 4 J Vlfi A j 150 200 250 300 350 Deformation Temperature (K) Figure 10. Critical stress for inducing martensites and for reverse transformation of a 50.3at.%NiTi single crystal, which was solution-treated at 1273 K for 1 h followed by aging at 673 K for l h. (afier reference [243].) 147 80 203 K 220K 243 K 40 J + M O ‘- 0....ng 1 4— 1’ l- 262 K 233 K 289 K 295 K 80 40 0 , Stress (MPa) 200 b . Ms=233 K 160 .298 K 302 K 307 K Af= 285 K Tm=315 K 120 80 40 o 040 0.4 00.403 Strain (%) 0 Figure 11. Stress—strain curves as a function of temperature of a Ti—50.5Ni single crystal, which was aged at 673 K for l h after solution-treatment. The curves are related with R—phase transformation or deformation of R-phase. T’m=Rs. (after reference [72]) 148 120- (a) :\ IOO V Tensale Stress (MPa) 1) o 8 _ "" Dfl=szb:-—r>— 20 ' ( w—A—a— 0 4 . . - r . . m )- “-.o~“~ 08» (b) ox‘k‘No . \0 A ‘\ 2‘3 04) ‘o .9 x, .5.’ ‘ ‘9 U) Q 0 . . . . . L . a m 220 240 260 280 300 320 Temperature (K) Figure 12. (a) Critical stresses for R-phase deformation and superelasticity, and (b) the recoverable strain of R-phase superelasticity, which is a function of temperature. (after reference [72]) 149 4— a = 40 MPG 5 at?‘ Cu 12.5 otx Cu 2.. 0r- L - - A 7.5 ates Cu 3 4“ N " 15 am Cu .5 r T: a; 2- N -+ «12 K ‘" o _ t 4— '0 ‘3’" C“ - 20 am Cu Mf ’5 2" § _. .11 K " O3 ' ) -"‘ “‘K 0C 1 Ms , _. Ms ,1 230 270 310 350 250 290 330 370 Temperature (K) Figure 13. Elongation vs. temperature curves under 40 MPa load for Ti—Ni—Cu alloys with various compositions indicated. 150 (after reference [73]). <—Stoge I->(<—Stoge H~—->l<—~—— Stage ll—ll—--> IOOO ® a 5! E a 800 U I 2% ___°;. Deformed at243 k E Ms=190 K .2 Af=221 K E 4" \~————-/' '\._‘I —fi]:¥ 0 l 20 50 (a) Strain (%) .Stago 14+— Stage II —-t-—— StageIII ---+—- [5 A (or 8s 93 (e) 8, ‘2 0L (Al SP '8 t— p m o 5- ‘ O O . . A O A A . . A : o s to Is 20 (b) Total Strain (“16) Figure 14. Stress—strain curves of 50.6 at.%NiTi alloy, which was deformed at 243 K (>Af). The alloy was solution-treated at 1273 K for l h. Dashed lines represent the recovered strain upon heating to 373 K. The symbol (x) represents the fracture point. (b) Plots of various strains in (a) as a function of total strain. (after reference [68]) 151 6(1) Deformation of a polycrystalline martensitic NiTi SMA under tension ‘JP' ‘v‘ {D "L‘..~' . ".sjw he» i .. ,. {a . I... \ . . cu CL :5 400» vi to 2 :75 .E’ 200- a 0 .E U) ch o . 0 3 o 9 12 15 Engineering Strain, % (8) Twin Bot-fiery um My mmmummrwmdnmmm don comma (0W ‘ " "“ “"‘, (ammm (b) mmmuw‘ ' “fomddomfllon Figure 15. Deformation stage of martensitic NiTi alloy sample (after reference [73])- 152 Figure 16. TEM images of in-situ tensile loading of superelastic NiTi. (a) martensite plates 1, 2, 3, 4 formed at 118 um displacement. (b) High density of looped dislocations (marked by ABCD) formed at 150 um displacement. (c) Magnified photo of the area between martensite plates 1, 2 and 3, in (b). (d) Diffraction pattern corresponding to the area indicated by single-headed arrow in (c). (After reference [79].) The martensite plate 2 shows some detwinning at the lower part; but the plate 3 does not seems to detwin, even though dislocations have been generated. 153 I . 1 l I Tensron 12m ._ [211] Orientation __ Single Crystal Ti-50.8at%Ni . 1000 Aged 1.5 hours a 823 K 3% strain A '— .— n de/dt=1x10'3 E 800 l— _ O ‘ 600 — - I; cycle I _ 2 10 16 .00 _ 7 _ - 200 _ o l l o 0 0 Strain, s (°/o) 1400 _ ompressron 1200 _ [21110rientatlon Single Crystal Ti-50.8at%Ni Aged 1.5 hours a 823 K 1000 — 4 ‘3 dc / dt = 6x10 E 800 _ c cycle 1 2 10 XX) g: 600 - — m 400 — 200 l .- l o l.‘ 0 0 0 Strain, e (%) Figure 17. (above) Tensile cyclic stress—strain response of single crystal NiTi oriented in the [1 12] direction. The test temperature is above the A f transformation temperature. (below) Compressive cyclic stress—strain response of single crystal NiTi oriented in the [1 12] direction. The test temperature is above the A f transformation temperature. (after reference [76]) 154 a. * ,.,. ,- ‘ ”w “’W Infill“... Figure 18. Observation of Luders deformation by photography for Ti-50.5Ni alloy. Black regions represent the parent phase, and lighter colored regions martensite. The strain increases with increasing numbers from (1) to (13). (after reference [83]). . ..... H421 Tensile street (MPa) 3 . x g ‘ C.--_‘_‘_I‘ ‘0 l.-_—: H 332 It 1149.8 at‘llr Ni 1273K 10-673 IQ M9294 K Af=318 K .0 2 4 5 I 101214 Skin 0‘) Figure 19. Effect of thermomechanical treatment on the stress—strain curves as a function of deformation temperature for 49.8at.%NiTi alloy. Solution-treatment at 1273 K for 1 h followed by annealing at 673 K for 1 h. (after reference [244]) 155 v Tenslle Stress (MPa) «- 3 ‘C) Ti=4SJ 1')“- Hi 673 K l0 Ms=251K M=313K 8.246.246. 00246024 Strain (%) Figure 20. Effect of thermomechanical treatment on the stress—strain curves as a fimction of deformation temperature for 49.8at.%NiTi alloy. Annealed at 673 K for l h immediately after cold-working. (afier reference [244]) 156 (a) as-rolled O I. 3 J‘, m_ Figure 21. TEM images of 49.8at.% NiTi alloy, which were annealed for 1 h at various temperatures indicated after cold-working; (a) as cold-worked, (b) at 673 K, (c) at 773 K and (d) at 873 K. (after reference [245]) 157 (0)209K (b12llK (c) 221K (6)229K (8)24OK Omani h-.-’ \aoncvi \----’ 6.--.l 6----d O 001296K “is" O. a [Z (I) 8 L. \..J O 6 g soot (ml 324K I!) S tkl304K tl)3l4l< !- , Ti-50.60t%Ni 4oor I273K to -* 773K (hr 10 Ms=ZSOK 200 Af-292K o 02‘4L680244681l0‘ S tr 0 in {%) Figure 22. Stress—strain curves as a function of testing temperature of 50.6at.%NiTi alloy, which was aged at 773 K for 1 h after solution-treatment at 1273 K for l h. (after reference [245]) 158 800~ 0" 0,, Heat treatment A A l273K 10* 773K [0 O O s - 673K 10 o I = ~573K IO Ti-50.6 at°/o Ni 0') O O Tensile Stress (MPa) .3 3 am I- O G :30 200 ' Temperature (K) Figure 23. Effect of aging temperature on the critical stresses for inducing martensites and for reverse transformation in a 50.6at.% NiTi alloy, which was aged at various temperatures for 1 h afier solution treatment. (after reference [244]) 159 6‘; (a) 573K m, ,5, .. "lf$1:\ 1}“: n. kilns" . 5'?! V “1‘5“ png\ I”? _ . .2.; 3} it”; a. ‘ét’: - . v . .- . A;.. §\“‘.5}§‘€:': o.‘ .u‘ R: “I? — . (b) 773K" “‘8 "‘ \ \K x“; \Q Q"\ ‘- \\ ‘ ‘&\‘ r\\ s y‘ \ \V\\\ \ $15!. is ' \‘ \\ \ ‘80 K. ; fk‘V‘h‘ \ ‘ \ \\ \\ ‘\ ~‘I “(it s: \ ‘3. Figure 24. TEM images and corresponding diffraction patterns of 50. 6at. % NiTi alloy specimens, which was aged at various temperatures afier solution-treatment at 1273 K for l h; (a and a’) at 673 K, (b and b’) at 773 K. (after reference [245]) Tenslle Stress (Ina) Strain fl) Figure 25. Stress—strain curves as a function of deformation temperature of 50.6at.%NiTi alloy, which was aged at 673 K immediately after cold-working. (afier reference [245]) 161 Anealed at fill) C for 1h, air cooled WW Dimaeter - 4.5 mm, Pitch .0.1 mm 1 Deformed and constrained atthb shape and annealed at different temperatures [\/\/\/\/\/\/| 1 Diameter - 4.5 mm, Pitch - 12 mm a. Deformed at martensite state Diameter -4.5mm, Pitch-0 mm Heated up to temperature higher than M NWWVl l Cooled to temperature lower than If Figure 26. The procedure of processing TWSME TiNi SMAs spring (afier reference [87]) 1 o 6.2 'llr defamation l 346 K i a? ‘ A .5 ‘2 ' g s s; a -3 - -4 k tramway J J L memory . -5 _ — -6 b L 1 L .1 L L 1 _l L L t __ 260 280 300 320 340 360 380 Temperature (K) Figure 27. The two-way shape memory effect obtained after a single deformation by 6.2% of equiatomic NiTi alloy (afier reference [102]). I62 a T: 22 C b T=60 C<1 min . c T: 22 C d T=60 0.30 min Figure 28. Optical microscopy verification of the predicted nricrostructure memory for an aged AU51Cd 49 martensite and its disappearance after aging in the parent phase. (a) Domain pattern of the aged martensite (aged at room temperature for six months). The sample is heated up to (b) parent phase (60 °C), and immediately cooled down to (c) martensite. Then the sample is heated up again to (d) parent phase (60 °C) and aged for 30 minutes, followed by cooling down to (e) martensite. (after reference [104]). _- 63 Austenite with tensile residual stress field l i Cool l l Heat Martensite Figure 29. TWSME due to internal tensile residual stress: elongation during cooling and shrinkage during heating. fill. i i M NH (ill Austenite ’1‘ T l l (if) Cool Heat Cool Heat Martensite Martensite Figure 30. External Stress and Internal Compressive Stress Field. The tensile external stress increases the internal tensile stress, elevating phase transformation temperatures but compressive external stress decreases the internal tensile stress. 164 .5538 03:89:“ 05 E 3.30 88an a £390 8 38838 soon was 5338 03563 05 E 9552 ofi .3 333 mm 5033 song—Homo“. owns—o 0.3a 25. A32 :0 mam—swasoou 05 E 33083 03 mammoth 35038 RE. $88.5. 353x”. Beanbag we eogoomm 358“ a mo common cougomaab 05 .3385 wagon 3v can mam—coo A3 wet—6 noise o§eH&EoH-qum ._m onE 3: 22360... 3: 839.030... . mm . mmm . awn . mum . mmm . mmm . men man i . ~31 r l. . . --.----... new .8 . ................ w ........ ll. .. - Lfin .2. . . .ll itiitttitiitt 4 in“)... iiiiiii e ix... 174%.? (ii. . 41./v1 “ a _. . rmd / t l E 2 n Eu m . . n . m . T3 w 33 a .0 m. t (E. u E e .06 m. 6.0 m. . m. u r a Em. .2 M _E :3 ./. 0" ll‘ ll PHD . E . lllill E E . i la .2 VB 3 l. E . pg. n E _ . In I film. .2 QB .2 H l E T. E a 3. .3 165 °‘2 50.51 °C °" * 13.96°C g 1 62.15°C _g 0.0 . 25.41 J/g '§ 103.09%: I 25.20J/g 77.05°C -°-1 * 31.56°C 112.64°C -0.2 . , T r . , . . . . j . . , . . -50 0 50 100 150 200 Temperature (°C) Figure 32. DSC curves of bulk NiTi samples used in TWSME study 1.5 oz/p oalp ' arlp 0.5 -— -- - -’ - —- -:I’ --“~‘ F a “—4 ”"’ o,/p ‘ ~' . l’ or/pm Uniform pressure p Hertz pressure — I .0 —0.5 o —0.5 —1.0 ~15 Figure 33. Stress distributions at the surface and along the axis of symmetry caused by (lefi) uniform pressure and (right) Hertz pressure acting on a contacting radius a (afier K.L. Johnson, Contact Mechanics, Cambridge University Press, 1985, pp.94, [108]). I67 (c) (d) Figure 34. Two-dimensional photo—elastic fi'inge patterns (contours of shear stress). (a) point load; (b) uniform pressure; (c)rigid flat punch; (d) contact of cylinders. (after reference [108], pp. 103). I68 Figure 35. The “pile-up” (above) where contact radius is larger than linear case and, “sink-in” (below) where contact radius is smaller. I69 200 b 281' l- r 5‘ E . E g 100 - E P D- e , 2-8? ' B i. , / equivalent defer-attain % i '9 1.5 y I T I 01 0-2 0-3 0.4 0.5 0-6 0-7 0.3 0-9 dID Figure 36. Representative strain vs. hardness in spherical indentation, and its correlation with true stress-true strain in uniaxial tension. Metal A is steel and metal B is copper. (after reference [1 1 I, 1 12]) 170 Figure 37. Finite element modeling of indentation of an elastic-plastic half space by a rigid sphere: development of the plastic zone. Broken line: contour of J2. Py is the normal load needed to first initiate plastic flow. (after reference [114]) I7] Before Heating Afier Heating 9“'-‘ r' _‘ A 518 N I‘_‘ " if. _O‘ I. ,2 t." .‘ 4.1M Figure 38. Recovery of spherical indents in NiTi alloy by the shape memory effect. (afier reference [231] ) 12 1 . E 0.8 ‘ I: 0.6 \ Superelastic Ll. 0.4 ‘ "m 0.2 o o 02 o 4 0.6 0.3 1 1 2 h/h max (a) 1.2 1 . x 0.3 4 E Superelastic & 06 m LL 0.‘ 02 . Cu 0 4 , . - 0 0.2 0.4 0.8 0.8 1 1.2 Mb max (b) Figure 39. Load-displacement curves of (a) Berkovich and (b) spherical indentations in SE NiTi. (afier reference [11 7] ) 172 Figure 40. TEM images from the indent-affected zone in 50.9at%NiTi single crystals. Images (a) to (c) were taken adjacent to the same indent at difiermt magnifications. (after reference [1 18]). I73 Figure 41. Images of wear marks on NiTi afier 200 cycles of scanning scratch under different normal loads and temperatures. A fiS 16 °C. (afier reference [144] ) I74 -0.80 -0.40 Stress c’y at the fllm-matrlxlnterface 4.00 4.00 -0.25 r -0.13 0.00 0.13 Stress Txy at the film-matrix Interface 0.25 ---i...i...i-.- 4.00 -2.00 0.00 2.00 4.00 Distance from normal line load applied at x=0.0 Figure 42. Calculated stresses due to an isolated unit normal line load applied at the origin as a function of substrate modulus E2. Variation of bearing stress 0,, and shear stress 1,, at the film-substrate interface are shown for a 1000 nm film with a reference elastic modulus E1=1. (afier reference [198]) I75 P 8 Stress “x atthe fllm-matrlxlnterface I - 8 'vvv'vviv . —o— £2=29 + Ef=50 1"” . . l . L g . . l . . 4. -2.00 0.00 2.00 4,00 Distance from normal line load applied at x=0.0 0.20 Substrate Modulus E2=0.2 Stress “x at the film-matrix interface 050 J r . i - . . i - . . i , - . 4.00 -2.00 0.00 2.00 4.00 Distance from normal line load applied at x=0.0 Figure 43. Calculated tensile stress a, at the film-substrate interface for a 1000 nm film with a reference elastic modulus E1=l for a unit normal line load at the origin. Substrate modulus E2 is varied in the upper figure. The effect of varying the film thickness on the flexure stress is shown in the lower figure for a substrate of modulus E2=0.2. (afier reference [198]) 176 ‘|.OE+07 - l CrN1-Al ’ A 1.0E+06 i z_ I g (I .0E+05 CTN-32'“ 5 \ l "E 1 .0E+O4 5 c - . 8 1 .0E+03 m 31 A' 2 g 1 .OE+02 3 'l .OE+O‘I ’l .OE+00 ,, " fl * Figure 44. Wear test results of samples withl um CrN coating and with: (from lefi to right) no interlayer, with 4 um SE NiTi interlayer (S1), 4 um Cr interlayer and 4pm SMA NiTi interlayer (S2). (afier reference [3] ) l77 (c) with -—‘ Figure 45. SEM Images of scratch track end with (3) 1pm CrN coating, no interlayer, (b) Sum CrN coating no interlayer, (c) 1pm CrN coating 4pm SE NiTi interlayer, (d) 1pm CrN coating and 4pm SMA NiTi interlayer and (e) 1pm CrN coating and 4pm Cr interlayer. (after reference [3]) I78 transformed M f gig/141 ////// i Figure 46. Vogit (parallel) model for stress induced martensite in austenite phase. (after reference [204] ) I I /r / tmnsfonnedM . j////////// ' VKKKKKKKKKKK Ill IIIIIIIIII/i \\\\ \\\\\\\\\\\ III III Ill/l1 \\\\\\ s\\\\\ Ill/ll III/Ii \\\\\\ s\\\\\ I;////////////. \\\\\\\\\\~.\\\\ Il/./_/_/./_/./£/_/,/!/’/_/.. ‘5 \ \ \ \ \ N. \ \ A Figure 47. Series model for stress induced martensite in austenite phase. (after reference [204] ) I79 A 0) v U) .fi 300K :—l o o U 35 OK 0 ii C‘. '3 32 BK r0 (0 :1: 3 0 OK 38 40 42 44 46 2theta Figure 48. X ray diffi'action patterns of NiTi thin films (sample #1) at various temperatures (b) 433aFe 2 1 8 K 0 L A 6‘ 112R 300R .5 2 63K :—i o 0 U 3 0 3K or .5 286K 4.) A 8 2 2 3K :13 38 40 42 44 46 2theta Figure 49. X ray diffraction patterns of NiTi thin films (sample #2) at various temperatures sample 180 NlTizAl A, W 0) g 300R (.8) 328K 1 10A 349K .8 AI ‘8 020M 1-1 1M 298K I f 3’8'4'0'4'2'4'4'4'6'4'8 Ztheta Figure 50. Temperature scanning X-ray diffraction of NiTi thin film on 6061T6 substrates, from which phase transformation temperatures were estimated and listed in TABLE. CrN coating was later deposited onto the NiTi thin fihns. NiTiCuzAl 318K 323K W 338K WWW Al 110A Al Cooling 348K 331K MM 002M 012M 298K ' ' ' 4'0 4'0 42 44 45 2 theta Temperature scanning X-ray diffraction of NiTiCu thin fihn on 6061T6 CrN coatings were later deposited on the NiTi thin films Heating Figure 51. substrates. 181 15 um NiTizAl ii 300K 338K A a.) W 110A 378K I l A ll 111M 020 M ‘ T l 38 40 42 44 46 48 2 9 Temperature scanning X—ray diffraction of NiTi thin film on 6061 T6 The NiTi thin films were interlayers of DLC coatings Cooling Heating Figure 52. aluminum substrates. 0.9 - — 300K — 353K —— 433K ——473K x 0.6 ‘ (U E E D. 0.3 ~ 0.0 r r r l l 0.0 0.2 0.4 0.6 0.8 1.0 h/hmax Figure 53. The normalized load-depth curves fi'om Berkovich indentation of the NiTi thin films at different temperatures. 182 3.175mm WC Ball NiTi alloy Figure 54. Diagram of thermomechanical training method to create two-way indent depth change. Plastic 6004 x4 . , Uniaxial l BlaXIal ' 480 - i 360- ........... 240 4 von Mises Stress (MPa) 120~ 0 1' . . . T 0.000 0.015 0.030 0.045 0.060 Strain Figure 55. Stress-strain curves of simple loading conditions from the 3D constitutive model for superelastic effect with plasticity. For the biaxial loading curve, the strain rate is set the same in two loading directions. The dashed curve represents a high uniaxial load which causes plastic hardening and permanent plastic strain. 183 Figure 56. Finite element mesh of indentation model. The refined region is the contact region. Total node number is around 15000. (a) 6000 1 — SE Nrri ----- Reference 2 3 E o 3000 q _I 15004 0.00 0.03 0.06 0.09 Depth (um) — SE NITI ----- Reference max) 04* 0.2 - Normalized Load (F/F ‘l I I A I I I 0 0.2 0.4 0.6 0.8 1.0 Normalized Depth (dldm) Figure 57. Absolute (a) and normalized (b) load-displacement curves of SE NiTi and reference materials from finite element modeling. Indentation-induced superelasticity enables large or 100% indent depth recovery. 185 f; 804 ‘ a. | \. “9" 60 § 1 Reference 2‘ - 4 m g 40« 2 O \I —-——-> g 8 A ‘ DC ........ e ._ 2 0: 20- A , , ‘6 Transformed reglon suze g (in contact radius) 5 0 . . . 0 o 0 1 2 3 4 Representative Strain (%) Figure 58. Recovery ratios vs. representative strains at different indentation depths, and the size of indentation-induced phase transformation region at different indentation depths, measured in contact radius. A 1-00‘ FE Modeling u-é - Experiment 0; 0.75] 'U (U 3 .0 0.50- A=30K (D .5 E E 0.25 O z 0.00 0.25 0.50 0.75 1.00 Normalized Depth (d/dmax) Figure 59. Comparison of experimental and calculated load—displacement curves. The experimental results are adapted from Ni et alf‘ Modeling SE NiTi has a phase transformation hysteresis A (A= (A f- As)/2-(M, - Mf)/2 ) 15K and 30K for the two computed curves. 186 z/a Figure 60. Contours of the austenite phase volume fraction at indentation depth of 0.018um (a), and complete indent recovery after unloading (b), where few residual martensites are present after unloading. 187 Figure 61. Contours of the austenite phase volume fraction (a) and plastic strains (b) at indentation depth of 0.10pm. (c) Contours of the austenite volume fraction afier unloading. For the partially recovered indent (recovery ratio 71%), there were much more residual martensites than in FIG. 5(b). I88 Compressive (Negative) —- Cycle 1 —— Cycle 2 —— Cycle 3 100 200 300 Temperature (C) 01 Figure 62. Residual stress evolution of NiTi thin fihn on Al substrates with temperature, for three heating-cooling cycles. 0 100 200 300 Temperature (C) Figure 63. First cycle residual stress evolution of NiTi thin film on A] substrates with temperature. 189 l .“p 50 MP3 "I .I "" 'w l 1 I 70 80 90 Temperature (C) Figure 64. The stress/temperature rate of NiTi thin film on Al substrates during martensite phase transformation. / " 11 l‘ 30 MPa ,“i' 1"” a “|lfl l 1’ do/dT=1.6 MPa/K . I" l 11 210 260 200 Temperature (C) Figure 65. The stress/temperature rate of NiTi thin film on Al substrates from CTE mismatch between thin film and substrates, without the influence of phase transformation, 190 Z'SDLC:'2NiTim:Al 12- n E 9‘ Austenite 375K 1 I .g 64 I 31 Martensite,303K, R: 15.94m 0 I l I l I O 5 10 15 20 Width (mm) Figure 66. Curvature change of Z'SDLC112N1T13632A1 sample between room temperature around M; and elevated temperature above A f. The curvature change mainly resulted from residual stress change in NiTi interlayer. 191 100 3"CrN:"NiTi353:Al _/ Deformation (umz) —I— CrN:NlTizAl, 0.98N 0-1 —o— CrN:NlTizAl, 0.49N o -- l:l-- CerAl, 0.98N ~ O-- CerAI, 0.49N 0.01 - I I ' ' ' 300 320 340 360 330 Temperature (K) Figure 67. Temperature scanning weer deformation of CrN:NiTi:A1 and CrN:Al samples, under 0.49 and 0.98 N load, using a 200 um tip radius spherical indenter. Superelastic NiTi improves the wear resistance by limiting plastic deformation to the sofi aluminum substrates. 192 298K 333K 353K 378K Figure 68. SEM images of temperature wear tracks of 3‘9CerAl samples (top row) and 3'S’CerlsNiTimzAl sample (bottom row), under 0.98N load. The wear data was presented in Figure 67 298K 378K Figure 69. SEM images of temperature scanning wear tracks of 3“’CerA1 samples (first row) and 3‘9CrN:'5NiTi353:Al sample (second row), under 0.49N. The wear data was presented in Figure 67. 193 ._ - _‘. _»~:-"‘ _. . , 720011111. 298 K 338 K 358 K Figure 70. SEM images of temperature scanning wear tracks of l"'CrN:$’NiTiCu;i43:Al samples. The wear data was presented in Figure 71. 1o_o, 200 um tip radius conical indenter, 0.49N Load [j ........................... D .............. El """"" NA ‘ 7cmuu E 3 .5 ' ‘5 1.0 q \ E I l 23 ‘-’cm:°NITICu,,,:AI\. O a a) 3 0.1 . 'Y'T‘I‘s “iii“ . 300 320 340 360 Temperature (K) Figure 71. Temperature scanning wear deformation of l‘7Cer9NiTiCU343zAl and 1'7CrN2Al samples, under 0.49 and 0.98 N load. 194 8.0x10“), ”cum: 3\ j\ 0’1 1 ‘-’cm:’Niri,,3:AI H 1 “cl-nun .U‘ .‘l 0 U1 X X _L _x O O or 4 Wear Loss (nm2) 2.5x105- “CrN:NiTimml 0.0 l l I i l 0 4 8 12 16 NiTi Interlayer Thickness (um) Figure 72. The effect of interlayer and CrN coating thickness on wear resistance at room temperature. The wear test was done using 6.3Smm diameter WC under 5N load. Thicker coating and interlayer gave better wear resistance. 195 6.35mm WC Ball, 0.98N Load 1051 'g T' ' .9 g ”CrN:AI E I 5 . 3 10“: o : _l . . a . g 3 l o/.* . "/ 1° 1 ‘-7cm:“Nirim:Al 300 330 360 330 Test Temperature(K) Figure 73. Wear loss vs. temperature using a different loading condition (6.35mm ball, 0.98N load). Wear resistance was not improved at temperature above 200 pm radius 4000 ~ 0.98N, 3498MPa 3000 - 0.49N, 2776MPa 2000 - 3.175 mm radius 0.98N, 554MPa 0.0 0.6 112 1.8 Load (N) 1000~ Elastic Contact Stress (MPa) Figure 74. Elastic contact stress in elastic halfspace from 3.175mm radius and 200um radius sphere under different loads. 196 1 0 1 1 2-5I)Lc:Al ”E 1 _ "Killi; 1:; 1:" """"""" ' :1. 5 O \\\\\ c i I Q 4 3531-09:'FT‘i;c.-_.:.-5.l E . .15 E 0 1 _ \ :\'DLC. NiTi333. Al O o “5 \\o D \e g 0.01 -. g ' 2-50L0:2‘NITI3,,;AI 1E-3 . . . . ' . 300 320 340 360 A f 380 Temperature (K) Figure 75. Temperature scanning wear tests of 2‘SDLC:NiTi363:Al samples, using 200um tip radius diamond indenter under 0.49N load. Thick NiTi interlayer and SE NiTi interlayer at temperature above A I showed better wear. 197 2-5DLc:Al 2'5DLC:AI 298K, 50 turns 378K, 50 turns 2"DLC:“NiTi3s:Al 2"DLC:‘5NiTim:AI 298K, 100 turns 378K, 100 turns 200 um 2"DLC:2‘NiTim:AI 2"DLC:21NiTim:AI 298K, 200 turns 378K, 2000 turns Figure 76. SEM images of temperature scanning wear tracks of 2‘SDLCzNiTi353zAl and 2'SDLCzAl samples. The wear data was presented in Figure 88. 198 1.2x106- o 9.0x1051 CrN:6NiTi31o:MZ NA ‘ E. "" g 6.0x105 ~ 3 o a CrN:12NiTi31o:M2 A é’ 3.0x105 - D CerMZ 0 0.0 . ”'1' """""" D"; . . 0 1 2 3 4 5 CrN Thickness (um) Figure 77. The wear resistance of CrN:NiTi:M2 samples at room temperature as a function of CrN coating thickness. 1 .2x106 - 1.7 . ~D--- CrN-M2 ”CrN:NiTiszZ WA 2'BCrN:hfl2 9.0x1054 --~o--~"‘"CrN:M2 "3» 2.8 . - - , CTN.NIT|31O.M2 6.0x105‘ A .0 5 3'gCerNiTisz2 3.0x10 - $.91. . 0.0 ' I I I I I I o 2 4 6 8101T2 NiTi Interlayer Thickness (um) Figure 78. The wear resistance of CrN:NiTi:M2 samples at room temperature as a function of interlayer thickness. Wear Loss (nmz) 353K Figure 79. SEM images of temperature scanning scratch tracks on 3'9CrN:9NiTi353:Al. Images of the whole scratch tracks (1 mm in length )are placed at the upper right corner. , - 50 run 306 K 331K 353K 375 K 398 K Figure 80. SEM images of temperature scanning scratches tracks on 3'9CerAl. Images of the whole scratch tracks (1 mm in length )are placed at the upper right corner. 200 298K 333K JV 1 1111' 91% L 1” 403K ”1 1' i 3 4 :7 Normal Load (N) Figure 81. Acoustic emission of scratch tests on 3‘S’Cer9NiTi3s3zAl. It showed strong temperature dependence that matched phase transformation temperatures of NiTi interlayers ex 101 w W 1' i 3 4 6 Normal Load (N) Acoustic Emissions Figure 82. Acoustic emission of scratch tests on 3'9CerAl. There is no significant temperature dependence. 201 Figure 83. EDS spectrum of the delarninated area in scratch tracks on 3'9C1'N39N1T13531A1, showing that delamination occurred at the interface between CrN and NiTi interlayer. Friends-13 cps Canoe: cerium ._ Figure 84. EDS spectrum of the delarninated area in scratch tracks on 3‘9CerAl, showing the delamination occurred at the interface between CrN and the Cr interlayer. The Cr interlayer ( 0.1 -12 , r f u 0 10 20 30 40 Horizontal (pm) Figure 112. Sectional profiles of the nano circular surface protrusions and peak height over five thermal cycles. 222 _a N 4\ ’’’’ [ ho p1. _ _. #15111:sz 1111 .. --- - —‘ Protrusion amplitude, hp ,microns O f I ' I T l O 200 400 Polished layer depth, dpl, microns V I Figure 113. Decay of protrusion amplitude with surface-removal. When successive surface layers of thickness dpl are removed afier initial planarization to a depth ho, the amplitude of protrusion appearance and disappearance, hp, diminishes and approaches zero when material has been removed to a depth roughly equal to the radius of the initial indent contact. 223 —l l 1/U *k‘9 9% ' (elastic-plastic O boulndary) \ l ‘K __ db/D = 0.51 I I l l Depth, y/D N / 2 l O Diameter,x/D Figure 114. Plastic strain distribution under an indent. Theoretical and experimental results , adapted from Samuel and Mulhearn [232] showing the expected magnitude of plastic strain under an indent in a ductile metal. 224 ATp = 62 AM‘ = “6 AMs = +34 _3 l Cooling JEN O / : In .t’ C D .d h < _d 4.: Q. a) I CD I c ' _ .9 : ATn — 28 I H I I (U I | H I C I a) i “o l K E ~ l O /O/O/ U \“1 Heating m G _ \ I I I < I AAS = -24 AAf = .4 0 50 100 l 50 Temperature, °C Figure 115. Height of reversible circular protrusions vs. temperature. The phase transformation temperature of NiTi sample is also marked. The hysteresis of the reversible protrusion is much narrower. 225 Austenite llll Martensite 7 llll Residual stress Figure 116. Schematic of Residual Stress in Reversible Surface Protrusion. The residual stress is in the opposite direction of TWSME during heating from martensite to austenite phase, and not in the opposite direction during cooling, narrowing hysteresis. The residual stress also stabilizes the TWSME. 226 Geometrically Necessary Dislocations Figure 1 17. Geometrically necessary dislocations. It is assumed that the indentation is accommodated by circular loops of geometrically necessary dislocations with Burgers vectors normal to the plane of the surface. As the indenter is forced into the surface of a single crystal, geometrically necessary dislocations are required to account for the permanent shape change at the surface. 227 Figure 118. Residual stress in y direction (ayy) afier spherical indentation in austenitic NiTi alloy, by finite element analysis. direction. 228 The negative sign is compressive References 1. W. Y. Ni et al., "Effects of the ratio of hardness to Young's modulus on the friction and wear behavior of bilayer coatings", Applied Physics Letters 85, 18, 4028 (2004). 2. W. Ni, Y. T. Cheng, D. S. Grummon, "Shape recovery and stress—induced martensite in TiNi following indentation and wear loading", Journal De Physique Iv 112, 1147 (2003). 3. W. Y. Ni et al., "Novel layered tribological coatings using a superelastic NiTi interlayer", Wear 259, 842 (2005). 4. K. Otsuka, C. M. Wayman, "Shape memory materials" (Cambridge University Press, New York, 1998) 5. A. Olander, "The crystal structure of Aqu", Zeitschrifi Fur Kristallographie 83, 1/2, 145 (1932). 6. A. B. Greninger, V. G. Mooradian, "Strain transformation in metastable beta copper-zinc and beta copper-tin alloys", Transactions Of The American Institute Of Mining And Metallurgical Engineers 128, 337 (193 8). 7. G. V. Kurdyumov, L. G. Khandros, "O Termouprugom Ravnovesii Pri Martensitnykh Prevrashcheniyakh", Doklaafy Akademii Nauk Sssr 66, 2, 211 (1949). 8. L. C. Chang, T. A. Read, "Plastic Deformation And Diffusionless Phase Changes In Metals - The Gold-Cadmium Beta-Phase", Transactions Of The American Institute Of Mining And Metallurgical Engineers 191, 1, 47 (1951). 9. Y. Q. Fu, H. J. Du, W. M. Huang, S. Zhang, M. Hu, "TiNi-based thin fihns in MEMS applications: a review", Sensors and Actuators a-Physical 1 12, 2-3, 395 (2004) 10. M. I. Frecker, "Recent advances in optimization of smart structures and actuators", 229 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. Journal of Intelligent Material Systems and Structures 14, 4-5, 207 (2003). C. Mavroidis, "Development of advanced actuators using shape memory alloys and electrorheological fluids", Research in Nondestructive Evaluation 14, l, l (2002) Y. Luo, M. Salvia, "SMA actuators in medical applications", Annales De Chimie-Science Des Materiaux 29, 6, 33 (2004). N. B. Morgan, "Medical shape memory alloy applications - the market and its products", Materials Science and Engineering a-Structural Materials Properties Microstructure and Processing 378, 1-2, 16 (2004). L. G. Machado, M. A. Savi, "Medical applications of shape memory alloys", Brazilian Journal of Medical and Biological Research 36, 6, 683 (2003). A. R. Pelton, S. M. Russell, J. DiCello, "The physical metallurgy of Nitinol for medical applications", Jam-Journal of the Minerals Metals & Materials Society 55, 5, 33 (2003). S. A. Shabalovskaya, "Surface, corrosion and biocompatibility aspects of Nitinol as an implant material", Bio-Medical Materials and Engineering 12, 1, 69 (2002). F. J. Gil, J. A. Planell, "Shape memory alloys for medical applications", Proceedings of the Institution of Mechanical Engineers Part H-Journal of Engineering in Medicine 212, H6, 473 (1998). S. A. Shabalovskaya, "On the nature of the biocompatibility and on medical applications of NiTi shape memory and superelastic alloys", Bio-Medical Materials and Engineering 6, 4, 267 (1996). B. P. Philip TV, Trans AIME 209, 1269 (1958). K. Otsuka, T. Sawarnura, K. Shimizu, "Crystal Structure and Internal Defects of Equiatomic Tini Martensite", Physica Status Solidi a-Applied Research 5, 2, 457 (1971) 230 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. G. M. Michal, R. Sinclair, "The Structure of Tini Martensite", Acta Crystallographica Section B-Structural Science 37, Oct, 1803 (1981). Y. Kudoh, M. Tokonami, S. Miyazaki, K. Otsuka, "Crystal-Structure of the Martensite in Ti-49.2 at-Percent-Ni Alloy Analyzed by the Single-Crystal X-Ray-Diffraction Method", Acta Metallurgica 33, 11, 2049 (1985). V. N. Khachin, Y. I. Paskal, V. E. Gunter, A. A. Monasevich, V. P. Sivokha, "Structure Transformation, Physical-Properties and Memory Effects in Titanium Nickelide and Its Alloys", F izika Metallov I Metallovedenie 46, 3, 511 (1978). M. B. Salamon, M. E. Meichle, C. M. Wayman, "Premartensitic Phases of Ti50Ni47Fe3", Physical Review B 31, 11, 7306 (1985). T. Tadaki, C. M. Wayman, "Electron-Microscopy Studies of Martensitic Transformations in Ti50Ni50-Xcux Alloys.1. Compositional Dependence of 1/3 Reflections from the Matrix Phase", Metallography 15, 3, 233 (1982). T. Tadaki, C. M. Wayman, "Electron-Microscopy Studies of Martensitic Transformations in Ti50Ni50-Xcux Alloys.2. Morphology and Crystal-Structure of Martensites", Metallography 15, 3, 247 (1982). T. H. Nam, T. Saburi, Y. Kawarnura, K. Shimizu, "Shape Memory Characteristics Associated with the BZ-Reversible-Bl 9 and B19-Reversible-Bl 9' Transformations in a Ti-40Ni-10Cu (at-Percent) Alloy", Materials Transactions Jim 31, 4, 262 (1990). M. Nishida, C. M. Wayman, T. Honma, "Precipitation Processes in near-Equiatomic Tini Shape Memory Alloys", Metallurgical Transactions a-Physical Metallurgy and Materials Science 17, 9, 1505 (1986). K. Otsuka, K. Shimizu, "Memory Effect and Thermoelastic Martensite Transformation in Cu-Al-Ni Alloy", Scripta Metallurgica 4, 6, 469 (1970). J. W. Christian, "The theory of transformations in metals and alloys " (Pergarnon Press, Oxford, 1965) 231 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. D. S. Lieberman, M. S. Wechsler, T. A. Read, "Cubic To Orthorhombic Diffusionless Phase Change - Experimental And Theoretical Studies Of Aucd", Journal of Applied Physics 26, 4, 473 (1955). M. S. Wechsler, D. S. Lieberman, T. A. Read, "On The Theory Of The Formation Of Martensite", Transactions Of The American Institute Of Mining And Metallurgical Engineers 197, 11, 1503 (1953). T. Saburi, C. M. Wayman, "Crystallographic Similarities In Shape Memory Martensites", Acta Metallurgica 27, 6, 979 (1979). W. J. Tang, "Thermodynamic study of the low-temperature phase 819' and the martensitic transformation in near-equiatomic Ti-Ni shape memory alloys", Metallurgical and Materials Transactions a-Physical Metallurgy and Materials Science 28, 3, 537 (1997). H. S. Wu, Q. C. Tian, "The superelasticity of TiPdNi high temperature shape memory alloy", Intermetallics ll, 8, 773 (2003). T. Inarnura et al., "Martensitic transformation behavior and shape memory properties of Ti-Ni-Pt melt spun ribbon", Journal of the Japan Institute of Metals 69, 8, 628 (2005). S. F. Hsieh, S. K. Wu, "A study on ternary Ti-rich TiNin shape memory alloys“, Materials Characterization 41, 4, 151 (1998). P. Olier, J. C. Brachet, J. L. Bechade, C. F oucher, G. Guenin, "Investigation of transformation temperatures, microstructure and shape memory properties of NiTi, NiTin and NiTin alloys", Journal De Physique Iv 5, C8, 741 (1995). X. L. Meng, Y. Wu, W. Cai, L. C. Zhao, "Two-way shape memory effect and its stability in Ti-Ni-Hf high temperature shape memory alloy", Transactions of Nonferrous Metals Society of China 15, 2, 340 (2005). L. W. Chang, D. S. Grummon, "Phase transformations in sputtered thin films of Ti-x(Ni,Cu)(1-x).2. Displacive transformations", Philosophical Magazine 232 fl 41. 42. 43. 44. 45. 46. 47. 48. 49. a-Physics of Condensed Matter Structure Defects and Mechanical Properties 76, 1, 191 (1997). X. D. Han et al. , "The martensite structure and aging precipitates of a TiNin high temperature shape memory alloy", Journal De Physique Iv 5, C8, 753 (1995) L. W. Chang, D. S. Grummon, "Structure evolution in sputtered thin films of Ti-x(Ni,Cu)(1-x).l. Diffusive transformations", Philosophical Magazine a-Physics of Condensed Matter Structure Deflects and Mechanical Properties 7 6, 1, 163 (1997). J. S. Zhang, W. Cai, X. B. Ren, K. Otsuka, M. Asai, "The nature of reversible change in M-s temperatures of Ti-Ni alloys with alternating aging", Materials Transactions Jim 40, 12, 1367 (1999). T. F ukuda, T. Saburi, T. Kakeshita, M. Kitayama, "Shape memory behavior of Ti-40.5Ni-10Cu alloy affected by C11(b)-type precipitates", Materials Transactions Jim 38, 2, 107 (1997). T. F ukuda, M. Kitayama, T. Kakeshita, T. Saburi, "Martensitic transformation behavior of a shape memory Ti-40.5Ni-10Cu alloy affected by the C11(b)-type precipitates", Materials Transactions Jim 37, 10, 1540 (1996). H. C. Lin, S. K. Wu, J. C. Lin, "The Martensitic-Transformation in Ti-Rich Tini Shape-Memory Alloys", Materials Chemistry and Physics 37, 2, 184 (1994). H. D. Gu et al. , "Growth of TiNin shape memory alloy thin films by laser ablation of composite targets", Applied Surface Science 129, 579 (1998). A. Ishida, A. Takei, S. Miyazaki, "Shape-Memory Thin-Film of Ti-Ni Formed by Sputtering", Thin Solid Films 228, 1-2, 210 (1993). J. A. Walker, K. J. Gabriel, M. Mehregany, "Thin-Film Processing of Tini Shape Memory Alloy", Sensors and Actuators a-Physical 21, 1-3, 243 (1990). 233 50. 51. 52. 53. 54. 55. 56. 57. 58. D. S. Grummon, L. Hou, Z. Zhao, T. J. Pence, "Progress on sputter-deposited thermotractive titanium-nickel films", Journal De Physique Iv 5, C8, 665 (1995). S. Miyazaki, A. Ishida, "Martensitic transformation and shape memory behavior in sputter-deposited TiNi-base thin films", Materials Science and Engineering a-Structural Materials Properties Microstructure and Processing 275, 106 (1999) L. Hou, D. S. Grummon, "Transformational Superelasticity in Sputtered Titanium-Nickel Thin-Films", Scripta Metallurgica Et Materialia 33, 6, 989 (1995). Y. H. Li et al., "Grain size and its distribution in NiTi thin films sputter-deposited on heated substrates", Chinese Physics 13, 8, 1315 (2004). P. Wollants, J. R. Roos, L. Delaey, "Thermally-Induced and Stress-Induced Thermoelastic Martensitic Transformations in the Reference Frame of Equilibrium Thermodynamics", Progress in Materials Science 37, 3, 227 (1993). P. Wollants, M. Debonte, L. Delaey, J. R. Roos, "Thermodynamic Analysis of the Work Performance of a Martensitic-Transformation under Stressed Conditions.2. Numerical-Analysis", Zeitschrifi Fur Metallkunde 70, 5, 298 (1979). P. Wollants, M. Debonte, L. Delaey, J. R. Roos, "Thermodynamic Analysis of the Work Performance of a Martensitic-Transformation under Stressed Conditions. 1 . Theoretical Considerations", Zeitschrifi Fur Metallkunde 70, 3, 146 (1979). T. M. Adams, S. R. Kirkpatrick, Z. Wang, A. Siahmakoun, "NiTi shape memory alloy thin films deposited by co-evaporation", Materials Letters 59, 10, 1161 (2005). D. S. Grummon, J. P. Zhang, "Stress in sputtered films of near-equiatomic TiNiX on (100) Si: Intrinsic and extrinsic stresses and their modification by thermally activated mechanisms", Physica Status Solidi a-Applied Research 186, l, 17 (2001) 234 59. 60. 61. 62. 63. 65. 66. 67. 68. D. S. Grummon, "Fabrication, microstructure and stress effects in sputtered TiNi thin films", Shape Memory Materials 327-3, 295 (2000). D. S. Grummon, J. P. Zhang, T. J. Pence, "Relaxation and recovery of extrinsic stress in sputtered titanium-nickel thin films on (100)-Si", Materials Science and Engineering a-Structural Materials Properties Microstructure and Processing 275, 722 (1999). R. X. Wang, Y. Zohar, M. Wong, "Residual stress-loaded titanium-nickel shape-memory alloy thin-film micro-actuators", Journal of Micromechanics and Microengineering 12, 3, 323 (2002). Y. Q. Fu, H. J. Du, "Relaxation and recovery of stress during martensite transformation for sputtered shape memory TiNi fihn", Surface & Coatings Technology 153, l, 100 (2002). R. W. Hoffman, " The Use of Thin Films in Physical Investigations" (Academic Press, New York, 1966) M. Ohring, "Materials Science of Thin Films" (Academic Press, New York, 1992) A. P. Lee et al., "A practical microgripper by fine alignment, eutectic bonding and SMA actuation", Sensors and Actuators a—Physical 54, 1-3, 755 (1996). S. A. Mathews, M. Wuttig, Q. M. Su, "The effect of substrate constraint on the martensitic transformation of Ni-Ti thin films", Metallurgical and Materials Transactions a-Physical Metallurgy and Materials Science 27, 9, 2859 (1996). K. Otsuka, C. M. Wayman, K. Nakai, H. Sakamoto, K. Shimizu, "Superelasticity Effects And Stress-Induced Martensitic Transformations In Cu-Al-Ni Alloys", Acta Metallurgica 24, 3, 207 (1976). S. Miyazaki, K. Otsuka, Y. Suzuki, "Transformation Pseudo-Elasticity And Deformation-Behavior In A Ti-50.6at-Percent Ni-Alloy", Scripta Metallurgica 15, 3, 287 (1981). 235 69. 70. 71. 72. 73. 74. 75. 76. 77. H. C. Ling, R. Kaplow, "Stress-Induced Shape Changes And Shape Memory In The R And Martensite Transformations In Equiatomic Niti", Metallurgical Transactions a-Physical Metallurgy and Materials Science 12, 12, 2101 (1981). H. C. Ling, R. Kaplow, "Phase-Transitions And Shape Memory In Niti", Metallurgical Transactions a-Physical Metallurgy and Materials Science 1 1, 1, 77 (1980). S. Miyazaki, K. Otsuka, "Mechanical-Behavior Associated With The Premartensitic Rhombohedral-Phase Transition In A Ti50ni47fe3 Alloy", Philosophical Magazine a-Physics of Condensed Matter Structure Defects and Mechanical Properties 50, 3, 393 (1984). S. Miyazaki, S. Kimura, K. Otsuka, "Shape-Memory Effect And Pseudoelasticity Associated With The R-Phase Transition In Ti-50.5 At—Percent-Ni Single-Crystals", Philosophical Magazine a-Physics of Condensed Matter Structure Defects and Mechanical Properties 57, 3, 467 (1988). T. H. Nam, T. Saburi, K. Shimizu, "Cu-Content Dependence Of Shape Memory Characteristics In Ti-Ni-Cu Alloys", Materials Transactions Jim 31, 11, 959 (1990). T. Hashinaga, S. Miyazaki, T. Ueki, H. Horikawa, "Transformation and deformation behavior in sputter-deposited Ti-Ni-Cu thin films", Journal De Physique Iv 5, C8, 689 (1995). H. Sehitoglu et al. , "Detvvinning in NiTi alloys", Metallurgical and Materials Transactions a-Physical Metallurgy and Materials Science 34, 1, 5 (2003). K. Gall et al. , "On the mechanical behavior of single crystal NiTi shape memory alloys and related polycrystalline phenomenon", Materials Science and Engineering a-Structural Materials Properties Microstructure and Processing 317, 1-2, 85 (2001). K. Gall, H. Sehitoglu, Y. I. Chumlyakov, I. V. Kireeva, "Tension-compression asymmetry of the stress-strain response in aged single crystal and polycrystalline NiTi", Acta Materialia 47, 4, 1203 (1999). 236 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. Y. Liu, Z. Xie, J. Van Humbeeck, L. Delaey, "Some results on the detwinning process in NiTi shape memory alloys", Scripta Materialia 41, 12, 1273 (1999). X. P. J iang et al. , "In situ observation of stress-induced martensitic transformation and plastic deformation in TiNi alloy", Materials Science and Engineering a-Structural Materials Properties Microstructure and Processing 238, 2, 303 (1997) H. Sehitoglu et al., "Compressive response of NiTi single crystals", Acta Materialia 48, 13, 3311 (2000). K. Gall, H. Sehitoglu, "The role of texture in tension-compression asymmetry in polycrystalline NiTi", International Journal of Plasticity 15, 1, 69 (1999). L. Orgeas, D. Favier, "Stress-induced martensitic transformation of a NiTi alloy in isothermal shear, tension and compression", Acta Materialia 46, 15, 5579 (1 998). J. A. Shaw, S. Kyriakides, "Initiation and propagation of localized deformation in elasto-plastic strips under uniaxial tension", International Journal of Plasticity 13, 10, 837 (1997). S. Miyazaki, T. Irnai, Y. Igo, K. Otsuka, "Effect Of Cyclic Deformation On The Pseudoelasticity Characteristics Of Ti-Ni Alloys", Metallurgical Transactions a-Physical Metallurgy and Materials Science 17, l, 1 15 (1986). F. B. Wang, W. J. Buehler, S. J. Pickart, "Crystal Structure and a Unique Martensitic Transition of Tini", Journal of Applied Physics 36, 10, 3232 (1965). A. Nagasawa, K. Enarni, Y. Ishino, Y. Abe, S. Nenno, "Reversible Shape Memory Effect", Scripta Metallurgica 8, 9, 1055 (1974). Z. G. Wang et al. , "Two-way shape memory effect of TiNi alloy coil extension springs", Materials Science and Engineering a-Structural Materials Properties Microstructure and Processing 360, 1-2, 126 (2003). 237 88. 89. 90. 91. 92. 93. 94. 95. 96. A. Isalgue, F. C. Lovey, M. Sade, "Oriented Growth Of Gamma Precipitates And Twsme In Cu-Zn-Al", Scripta Metallurgica Et Materialia 28, 10, 1183 (1993). X. M. Zhang, J. Fernandez, J. M. Guilemany, "Origin of two way shape memory effect in Cu-based shape memory alloys", Journal De Physique Iv 112, 487 (2003). T. F ukuda, M. Takahata, T. Kakeshita, T. Saburi, "Two-way shape memory properties of a Ti-51Ni single crystal including Ti3Ni4 precipitates of a single variant", Materials Transactions Jim 42, 2, 323 (2001 ). C. Y. Chang, D. Vokoun, C. T. Hu, "Two-way shape memory effect of NiTi alloy induced by constraint aging treatment at room temperature", Metallurgical And Materials Transactions A—Physical Metallurgy And Materials Science 32, 7, 1629 (2001) R. Stalmans, J. Van Humbeeck, L. Delaey, "Thermomechanical Cycling, 2 Way Memory and Concomitant Effects in Cu-Zn-Al Alloys", Acta Metallurgica Et Materialia 40, 3, 501 (1992). W. Siegert, M. Mertrnann, K. Neuking, G. Eggeler, "Influence of thermomechanical treatment on a NiTiNb shape memory alloy for coupling devices", Materialwissenschafi Und Werkstofiiechnik 31, 9, 822 (2000). H. W. Kim, "A study of the two-way shape memory effect in Cu-Zn-Al alloys by the thermomechanical cycling method", Journal of Materials Processing Technology 146, 3, 326 (2004). J. M. Guilemany, J. Fernandez, "Mechanism of 2—Way Shape-Memory Effect Obtained by Stabilized Stress-Induced Martensite", Journal De Physique Iv 5, C2, 355 (1995). J. Pons, M. Masse, R. Portier, "Thermomechanical cycling and two-way memory effect induced in Cu-Zn-Al", Materials Science and Engineering A 273-275, 610 (1999) 238 97. 98. 99. 100. 101. 102. 103. 104. 105. 106. P. Y. Manach, D. F avier, "Origin Of The 2-Way Memory Effect In Niti Shape-Memory Alloys", Scripta Metallurgica Et Materialia 28, 11, 1417 (1993). X. L. Meng, Y. F. Zheng, W. Cai, L. C. Zhao, "Two-way shape memory effect of a TiNin high temperature shape memory alloy", Journal of Alloys and Compounds 372, 1-2, 180 (2004). T. Omori, Y. Sutou, J. J. Wang, R. Kainuma, K. Ishida, "Effect of microstructure on two-way shape memory effect in Cu-Al—Mn alloys", Journal De Physique Iv 112, 507 (2003). R. Lahoz, L. Gracia-Villa, J. AlPuertolas, "Training of the two-way shape memory effect by bending in NiTi alloys", Journal of Engineering Materials and Technology—Transactions of the Asme 124, 4, 397 (2002). H. Schemgell, A. C. Kneissl, "Generation, development and degradation of the intrinsic two-way shape memory effect in different alloy systems", Acta Materialia 50, 2, 327 (2002). Y. Liu, Y. Liu, J. Van Humbeeck, "Two-way shape memory effect developed by martensite deformation in NiTi", Acta Materialia 47, 1, 199 (1998). D. A. Miller, D. C. Lagoudas, "Influence of cold work and heat treatment on the shape memory effect and plastic strain development of NiTi", Materials Science and Engineering a-Structural Materials Properties Microstructure and Processing 308, 1-2, 161 (2001). X. B. Ren, K. Otsuka, "Universal symmetry property of point defects in crystals", Physical Review Letters 85, 5, 1016 (2000). P. Prader, A. C. Kneissl, "Deformation behaviour and two-way shape memory effect of NiTi alloys", Zeitschrifi Fur Metallkunde 88, 5, 410 (1997). R. Stalmans, J. Van Humbeeck, L. Delaey, "The 2 Way Memory Effect in Copper-Based Shape Memory Alloys - Thermodynamics and Mechanisms", Acta Metallurgica Et Materialia 40, 11, 2921 (1992). 239 107. H. Hertz, "0n the contact of rigid elastic solids and on hardness". J. a. Schott, Ed., 108. 109. Miscellaneous Papers by H. Hertz (Macmillan, London, 1896) K. L. Johnson, "Contact mechanics" (Cambridge University Press, New York, NY, 1985)pp. 174-176. M. F. Doemer, W. D. Nix, "A method for interpreting the data from depth-sensing indentation instruments", Journal of Materials Research 1, 4, 601 (1986). 110. 111. 112. 113. 114. 115. 116. W. C. Oliver, G. M. Pharr, "An Improved Technique For Determining Hardness And Elastic-Modulus Using Load And Displacement Sensing Indentation Experiments", Journal of Materials Research 7, 6, 1564 (1992). D. Tabor, "Indentation hardness: Fifiy years on - A personal view", Philosophical Magazine a-Physics of Condensed Matter Structure Defects and Mechanical Properties 74, 5, 1207 (1996). D. Tabor, "A Simple Theory Of Static And Dynamic Hardness", Proceedings Of The Royal Society Of London Series A-Mathematical And Physical Sciences 192, 1029, 247 (1948). J. Mackerle, "Finite element modelling and simulation of indentation testing: a bibliography (1990-2002)", Engineering Computations 21, 1, 23 (2004). C. Hardy, G. V. Tordion, C. N. Baronet, "Indentation Of An Elastic-Perfectly-Plastic Half-Space By A Hard-Sphere", Journal Of Basic Engineering 94, 1, 251 (1972). G. B. Sinclair, P. S. Follansbee, K. L. Johnson, "Quasi-Static Normal Indentation Of An Elasto-Plastic Half-Space By A Rigid Sphere.2. Results", International Journal of Solids and Structures 21, 8, 865 (1985). W. Y. Ni, Y. T. Cheng, D. S. Grummon, "Microscopic shape memory and superelastic effects under complex loading conditions", Surface & Coatings Technology 177, 512 (2004). 240 117. W. Y. Ni, Y. T. Cheng, D. S. Grummon, "Microscopic superelastic behavior of a nickel-titanitun alloy under complex loading conditions", Applied Physics Letters 82, 17, 2811 (2003). 118. K. Gall, K. Juntunen, H. J. Maier, H. Sehitoglu, Y. I. Chumlyakov, "Instrumented micro-indentation of NiTi shape-memory alloys", Acta Materialia 49, 16, 3205 (2001). 119. K. Gall et al., "Micro and macro deformation of single crystal NiTi", Journal of Engineering Materials and Technology-Transactions of the Asme 124, 2, 238 (2002) 120. C. Liu, Q. P. Sun, Y. P. Zhao, T. X. Yu, "Depth dependence of nanohardness in a CuAlNi single crystal shape memory alloy", International Journal of Nonlinear Sciences and Numerical Simulation 3, 3-4, 535 (2002). 121. C. Liu, Y. P. Zhao, T. X. Yu, "Measurement of microscopic deformation in a CuAlNi single crystal alloy by nanoindentation with a heating stage", Materials & Design 26, 5, 465 (2005). 122. E. Rabinowicz, "The Least Wear", Wear 100, 1-3, 533 (1984). 123. J. T. Burwell, C. D. Strang, "On The Empirical Law Of Adhesive Wear", Journal of Applied Physics 23, 1, 18 (1952). 124. J. F. Archard, "Contact And Rubbing Of Flat Surfaces", Journal of Applied Physics 24, 8, 981 (1953). 125. R. Liu, D. Y. Li, "Modification of Archard's equation by taking account of elastic/pseudoelastic properties of materials", Wear 250, 956 (2001). 126. A. Leyland, A. Matthews, "On the significance of the H/E ratio in wear control: a nanocomposite coating approach to optimised tribological behaviour", Wear 246, 1-2, 1 (2000). 127. E. Rabinowicz, "Friction and Wear of Materials " (Wiley-Interscience, New York, 241 128. 129. 130. 131. 132. 133. 134. 135. 136. 137. 138. ed. 2 nd, 1995) "Compilation of AST M Standard Definitions" (ASTM, Philadelphia, PA, 1994) "Standard Test Method for Wear Testing with a Pin-on-Disk Apparatus ", Compilation of ASTM Standard Definitions (ASTM, Philadelphia, PA, 1994) R. Jacobs et al. , "A certified reference material for the scratch test", Surface & Coatings Technology 174, 1008 (2003). N. I. Tymiak, A. Daugela, T. J. Wyrobek, O. L. Warren, "Highly localized acoustic emission monitoring of nanoscale indentation contacts", Journal of Materials Research 18, 4, 784 (2003). S. J. Bull, E. G. Berasetegui, "An overview of the potential of quantitative coating adhesion measurement by scratch", Tribology International 39, 2, 99 (2006). S. J. Bull, "Failure mode maps in the thin film scratch adhesion test", Tribology International 30, 7, 491 (1997). H. C. Lin, H. M. Liao, J. L. He, K. C. Chen, K. M. Lin, "Wear characteristics of TiNi shape memory alloys", Metallurgical and Materials Transactions a-Physical Metallurgy and Materials Science 28, 9, 1871 (1997). H. C. Lin, H. M. Liao, J. L. He, K. M. Lin, K. C. Chen, "Wear characteristics of ion-nitrided Ti50Ni50 shape memory alloys", Surface & Coatings Technology 92, 3, 178 (1997). Y. N. Liang, S. Z. Li, Y. B. Jin, W. Jin, S. Li, "Wear behavior of a TiNi alloy", Wear 198, 1-2, 236 (1996). D. Y. Li, "Wear behaviour of TiN i shape memory alloys", Scripta Materialia 34, 2, 195 (1996). J. Singh, A. T. Alpas, "Dry Sliding Wear Mechanisms In A Ti50ni47fe3 242 139. 140. 141. 142. 143. 144. 145. 146. 147. 148. Interrnetallic Alloy", Wear 181, 302 (1995). C. A. Zirnmerly, O. T. Inal, R. H. Richman, "Explosive Welding Of A Near-Equiatomic Nickel-Titanium Alloy To Low-Carbon Steel", Materials Science and Engineering a-Structural Materials Properties Microstructure and Processing 188, 1-2, 251 (1994). P. Clayton, "Tribological Behavior Of A Titanium Nickel-Alloy", Wear 162, 202 (1993). R. H. Richman, A. S. Rao, D. E. Hodgson, "Cavitation Erosion Of 2 Niti Alloys", Wear 157, 2, 401 (1992). Y. Shida, Y. Sugimoto, "Water Jet Erosion Behavior Of Ti-Ni Binary-Alloys", Wear 146, 2, 219 (1991). F. T. Cheng, P. Shi, H. C. Man, "Cavitation erosion resistance of heat-treated NiTi", Materials Science and Engineering a-Structural Materials Properties Microstructure and Processing 339, 1-2, 312 (2003). L. M. Qian, X. D. Xiao, Q. P. Sun, T. X. Yu, "Anomalous relationship between hardness and wear properties of a superelastic nickel-titanium alloy", Applied Physics Letters 84, 7, 1076 (2004). J. Koskinen, E. Hainri, A. Mahiout, V. K. Lindroos, S. P. Hannula, "Superelastic NiTi coatings with good corrosive wear resistance", Journal De Physique Iv 112, 1 137 (2003). I. M. Hutchings, " T ribology Friction and Wear of Engineering Materials " (Edward Arnold, London, 1992) W. A. Glaeser, "Materials for Tribology" (Elsevier Science Publishers, Amsterdam, The Netherlands, 1992) R. Liu, D. Y. Li, "Experimental studies on tribological properties of pseudoelastic TiN i alloy with comparison to stainless steel 304", Metallurgical and Materials 243 149. 150. 151. 152. 153. 154. 155. 156. 157. 158. Transactions a-Physical Metallurgy and Materials Science 31, 11, 2773 (2000). D. Y. Li, "Exploration of TiN i shape memory alloy for potential application in a new area: tribological engineering", Smart Materials & Structures 9, 5, 717 (2000). H. Z. Ye, R. Liu, D. Y. Li, R. Eadie, "Development of a new wear-resistant material: TiC/TiNi composite", Scripta Materialia 41, 10, 1039 (1999). S. PalDey, S. C. Deevi, "Single layer and multilayer wear resistant coatings of (Ti,Al)N: a review", Materials Science and Engineering a-Structural Materials Properties Microstructure and Processing 342, 1-2, 58 (2003). M. O. Vasil'yev, V. Y. Panarin, A. A. Tkachuk, "Titanium-nitride coatings deposited by the method of a vacuum electric arc (review ", Metallofizika I Noveishie Tekhnologii 22, 11, 58 (2000). L. Hultman et al., "Review of the thermal and mechanical stability of TiN-based thin films", Zeitschrifi Fur Metallkunde 90, 10, 803 (1999). Z. B. Zhao, Z. U. Rek, S. M. Yalisove, J. C. Bilello, "Phase formation and structure of magnetron sputtered chromium nitride films: in-situ and ex-situ studies", Surface & Coatings Technology 185, 2-3, 329 (2004). X. M. He et al. , "Structure, hardness, and tribological properties of reactive magnetron sputtered chromium nitride films", Journal of Vacuum Science & Technology A 18, 1, 30 (2000). M. Nordin, M. Larsson, S. Hogrnark, "Mechanical and tribological properties of multilayered PVD TiN/CrN, TiN/MON, TiN/NbN and TiN/TaN coatings on cemented carbide", Surface & Coatings Technology 106, 2-3, 234 (1998). M. Nordin, M. Larsson, S. Hogrnark, "Mechanical and tribological properties of multilayered PVD TiN/CrN", Wear 232, 2, 221 (1999). P. Ameodo Larochette, M. Ahlers, "Grain-size dependence of the two-way shape 244 159. 160. 161. 162. 163. 164. 165. 166. 167. memory effect obtained by stabilisation in Cu—Zn-Al crystals", Materials Science and EngineeringA 361, 1-2, 249 (2003). J. Robertson, "Diamond-like amorphous carbon", Materials Science & Engineering R-Reports 37, 4—6, 129 (2002). M. Wang, K. Schmidt, K. Reichelt, H. Dimigen, H. Hubsch, "The Properties of Titanium-Containing Amorphous Hydrogenated Carbon-Films", Surface & Coatings Technology 47, 1-3, 691 (1991). F. M. Kimock, B. J. Knapp, "Commercial Applications of Ion-Beam Deposited Diamond-Like Carbon (ch) Coatings", Surface & Coatings Technology 56, 3, 273 (1993). A. Grill, V. Patel, "Tribological Properties of Diamond-Like Carbon and Related Materials", Diamond and Related Materials 2, 5-7, 597 (1993). J. Koskinen et al., "Microstructural changes in DLC films due to tribological contact", Surface & Coatings Technology 109, 1-3, 385 (1998). A. K. Gangopadhyay, P. A. Willerrnet, M. A. Tamor, W. C. Vassell, "Amorphous hydrogenated carbon films for tribological applications. 1 . Development of moisture insensitive films having reduced compressive stress", T ribology International 30, 1, 9 (1997). A. K. Gangopadhyay, P. A. Willerrnet, W. C. Vassell, M. A. Tamor, "Amorphous hydrogenated carbon films for tribological applications.2. Films deposited on aluminium alloys and steel", Tribology International 30, 1, 19 (1997). A. Haseeb, J. P. Celis, J. R. Roos, "An Electrochemical Deposition Process For The Synthesis Of Laminated Nanocomposites", Materials And Manufacturing Processes 10, 4, 707 (1995). J. P. Celis, K. Vanacker, K. Callewaert, P. Vanhoutte, "Residual-Stress Measurements In Electrolytic Copper-Nickel Compositionally Modulated Multilayers", Journal Of The Electrochemical Society 142, 1, 70 (1995). 245 168. 169. 170. 171. 172. 173. 174. 175. 176. A. Haseeb, J. P. Celis, J. R. Roos, "Dual-Bath Electrodeposition Of Cu/Ni Compositionally Modulated Multilayers", Journal W The Electrochemical Society 141, 1, 230 (1994). K. L. Lin, W. H. Chao, C. D. Wu, "The performance and degradation behaviours of the TiAlN/interlayer coatings on drills", Surface & Coatings Technology 89, 3, 279 (1997). A. Leyland et al. , "Tin And Cm Pvd Coatings On Electroless Nickel-Coated Steel Substrates", Surface & Coatings Technology 60, 1-3, 474 (1993). J. L. He, M. H. Hon, "Sliding Wear-Resistance Of Tool Steel Coated With Electroless Ni-P And Cathodic Arc Plasma Tin", Surface & Coatings Technology 53,1,87(1992) J. L. He, M. H. Hon, "Corrosion Protection Of Tin-Coated Low-Carbon Steel With Electroless Ni-P As An Interlayer", Surface & Coatings Technology 53, 1, 93 (1992). J. Xu, W. J. Liu, Y. D. Kan, M. L. Zhong, "Microstructure and wear properties of laser cladding Ti-Al-Fe-B coatings on AA2024 aluminum alloy", Materials & Design 27, 5, 405 (2006). R. Murakami, Y. H. Kim, K. Kimura, D. Yonekura, D. H. Shin, "Evaluation of adhesive behaviors of chromium nitride coating films produced by are ion plating meth ", Jsme International Journal Series A-Solid Mechanics And Material Engineering 49, l, 123 (2006). T. H. Le, Y. H. Chae, S. S. Kim, "Sliding wear behavior of plasma sprayed zirconia coating on cast aluminum against silicon carbide ceramic", Journal W Materials Science & Technology 21, 5, 666 (2005). T. B. Wei, F. Y. Yan, J. Tian, "Characterization and wear- and corrosion-resistance of microarc oxidation ceramic coatings on aluminum alloy", Journal W Alloys And Compounds 389, 1-2, 169 (2005). 246 177. 178. 179. 180. 181. 182. 183. 184. 185. J. X. Liao et al. , "The tribological properties of a gradient layer prepared by plasma-based ion implantation on 2024 aluminum alloy", Surface & Coatings Technology 183, 2-3, 157 (2004). I. A. Podchemyaeva, A. D. Panasyuk, V. V. Shchepetov, D. V. Yurechko, V. U. Gromenko, "Surface modification of AL9 alloy by electric-spark alloying with materials of the AlN-Ti(Zr)B-2-Ti(Zr)Si-2 system", Powder Metallurgy And Metal Ceramics 43, 3-4, 156 (2004). E. A. R Perez, R. M. Souza, "Numerical and experimental analyses on the contact stresses developed during single and successive indentations of coated systems", Surface & Coatings Technology 188-89, 572 (2004). L. F. Xia, Z. H. Yan, J. X. Liao, "Effects of intermediate layers on the tribological behavior of DLC coated 2024 aluminum alloy", Wear 257, 5-6, 599 (2004). H. A. Samir, H. C. Qian, B. C. Xia, S. M. Wu, "Comments on process of duplex coatings on aluminum alloys", Journal W Central South University W Technology 11, 3, 239 (2004). S. Uozato, K. Nakata, M. Ushio, "Corrosion and wear behaviors of ferrous powder thermal spray coatings on aluminum alloy", Surface & Coatings Technology 169, 691 (2003). F. Zhou, C. M. Suh, S. S. Kim, R. Murakami, "Sliding-wear behavior of TiN- and CrN-coated 2024 aluminum alloy against an A1203 ball", Tribology Letters 13, 3, 173 (2002). F. Zhou, C. M. Suh, S. S. Kim, R. Murakami, "Tribological behavior of CrN coating on aluminum alloys deposited by are ion plating", Journal W Materials Research 17, 12, 3133 (2002). J. Tian, Z. Z. Luo, S. K. Qi, X. J. Sun, "Structure and antiwear behavior of micro-arc oxidized coatings on aluminum alloy", Surface & Coatings Technology 154, 1, 1 (2002). 247 186. 187. 188. 189. 190. 191. 192. 193. 194. 195. J. X. Liao, L. F. Xia, Y. Sun, "Study of the structure of AlN/DLC layer of aluminum alloy implanted by PBII", Acta Metallurgica Sinica 37, 9, 922 (2001). K. Nakata, M. Ushio, "Wear resistance of plasma sprayed Al-Si binary alloy coatings on A6063 Al alloy substrate", Surface & Coatings Technology 142, 277 (2001) R M. Souza, A. Sinatora, G. G. W. Mustoe, J. J. Moore, "Numerical and experimental study of the circular cracks observed at the contact edges of the indentations of coated systems with soft substrates", Wear 251, 1337 (2001). P. H. Chong, H. C. Man, T. M. Yue, "Microstructure and wear properties of laser surface-cladded Mo-WC MMC on AA6061 aluminum alloy", Surface & Coatings Technology 145, 1-3, 51 (2001). Y. Miki, T. Taniguchi, T. Hanabusa, K. Kusaka, T. Matsue, "X-ray stress measurement and mechanical properties of TiN films coated on aluminum and aluminum alloy substrates by are ion plating and ion beam mixing", Materials Science Research International 6, 4, 243 (2000). S. K. Kim, T. H. Kim, J. Wohle, K. T. Rie, "TiCN coatings on aluminum alloy formed by MO-PACVD", Surface & Coatings Technology 131, 1-3, 121 (2000). Y. H. Chae, S. S. Kim, "Sliding wear behavior of ceramic, plasma sprayed on casting aluminum alloy against SiC ball", T ribology Letters 8, l, 35 (2000). G. Y. Liang, T. T. Wong, J. Y. Su, C. H. Woo, "Amorphous structure in a laser clad Ni-Cr-Al coating on Al-Si alloy", Transactions W Nonferrous Metals Society W China 10, 2, 220 (2000). M. L. Kuruppu, G. Negrea, I. P. Ivanov, S. L. Rohde, "Monolithic and multilayer Cr/CrN, Cr/Cr2N, and Cr2N/CrN coatings on hard and sofi substrates", Journal W Vacuum Science & Technology A-Vacuum Surfaces And Films 16, 3, 1949 (1998). B. L. Cain, W. J. Croisant, "Electrical contact resistance of titanium nitride 248 196. 197. 198. 199. 200. 201. 202. 203. 204. coatings for electromagnetic shielding applications", Surface & Coatings Technology 82, 1-2, 83 (1996). J. Senf, E. Broszeit, "Wear and corrosion protection using Cr and CrN (PVD coating on Aland Mg)", Materialwissenschafl Und Werkstofiiechnik 30, 5, 262 (1 999). L. S. Zheng, S. Ramalingam, "Multi-layer and composite structures for advanced coatings", Surface & Coatings Technology 81, 1, 52 (1996). S. Ramalingarn, L. S. Zheng, "Film-Substrate Interface Stresses And Their Role In The Tribological Performance Of Surface-Coatings", T ribology International 28, 3, 145 (1995). K. Tanaka, S. Nagaki, "A Thermomechanical Description of Materials with Internal Variables in the Process of Phase-Transitions", Ingenieur Archiv 51, 5, 287 (1982). C. Liang, C. A. Rogers, "One-dimensional thermomechanical constitutive relations for shape memory materials (Reprinted from Journal of Intelligent Material Systems and Structures, vol 1, pg 207-234, 1990)", Journal of Intelligent Material Systems and Structures 8, 4, 285 (1997). Y. Ivshin, T. J. Pence, "A Constitutive Model for Hysteretic Phase-Transition Behavior", International Journal of Engineering Science 32, 4, 681 (1994). Y. Ivshin, T. J. Pence, "A Thermomechanical Model for a One Variant Shape-Memory Material", Journal of Intelligent Material Systems and Structures 5, 4, 455 (1994). Y. W. Chi, T. J. Pence, H. Y. Tsai, "Axisymmetric plane stress states of an annulus subject to displacive shear transformation", Journal of Applied Mechanics-Transactions of the Asme 72, 1, 44 (2005). L. C. Brinson, M. S. Huang, "Simplifications and comparisons of shape memory alloy constitutive models", Journal of Intelligent Material Systems and Structures 249 205. 206. 207. 208. 209. 210. 211. 212. 7,1,108(1996) J. G. Boyd, D. C. Lagoudas, "A thermodynamical constitute model for shape memory materials. 1. The monolithic shape memory alloy", International Journal of Plasticity 12, 6, 805 (1996). J. G. Boyd, D. C. Lagoudas, "A thermodynamical constitutive model for shape memory materials.2. The SMA composite material", International Journal of Plasticity 12, 7, 843 (1996). Q. P. Sun, K. C. Hwang, "Micromechanics Modeling for the Constitutive Behavior of Polycrystalline Shape Memory Alloys]. Derivation of General Relations", Journal of the Mechanics and Physics of Solids 41, 1, 1 (1993). Q. P. Sun, K. C. Hwang, "Micromechanics Modeling for the Constitutive Behavior of Polycrystalline Shape Memory Alloys.2. Study of the Individual Phenomena", Journal of the Mechanics and Physics of Solids 41, l, 19 (1993). E. Patoor, M. O. Bensalah, A. Eberhardt, M. Berveiller, "Thermomechanical Behavior Determining of Shape-Memory Alloys Using a Therrnodynamical Potential Optimizing", Revue De Metallurgie-Cahiers D Informations Techniques 90, 12, 1587 (1993). M. Huang, L. C. Brinson, "Multivariant model for single crystal shape memory alloy behavior", Journal of the Mechanics and Physics of Solids 46, 8, 1379 (1998). B. C. Goo, C. Lexcellent, "Micromechanics-based modeling of two-way memory effect of a single crystalline shape-memory alloy", Acta Materialia 45, 2, 727 (1997) F. Auricchio, L. Petrini, "A three-dimensional model describing stress-temperature induced solid phase transformations: solution algorithm and boundary value problems", International Journal for Numerical Methods in Engineering 61, 6, 807 (2004). 250 213. 214. 215. 216. 217. 218. 219. 220. 221. 222. 223. F. Auricchio, U. Stefanelli, "Numerical analysis of a three-dimensional super-elastic constitutive model", International Journal for Numerical Methods in Engineering 61, 1, 142 (2004). D. C. Lagoudas, P. B. Entchev, "Modeling of transformation-induced plasticity and its effect on the behavior of porous shape memory alloys. Part I: constitutive model for fully dense SMAs", Mechanics of Materials 36, 9, 865 (2004). Y. H. Wang, D. N. Fang, "A three-dimensional constitutive model for shape memory alloys", International Journal of Nonlinear Sciences and Numerical Simulation 4, l, 81 (2003). M. Brocca, L. C. Brinson, Z. Bazant, "Three-dimensional constitutive model for shape memory alloys based on microplane model", Journal of the Mechanics and Physics of Solids 50, 5, 1051 (2002). J. Lubliner, F. Auricchio, "Generalized plasticity and shape-memory alloys", International Journal of Solids and Structures 33, 7, 991 (1996). W. Y. Ni, Ph.D Dissertation, Michigan State University (2003). Z. B. Zhao, Master Thesis, Michigan State University (2000). A. C. o. Terminology, "Compilation of AST M Standard Definitions" (ASTM, Philadelphia, PA, 1994) T. Kakeshita et al. , "Effect Of Hydrostatic Pressures On Thermoelastic Martensitic Transformations In Aged Ti-Ni And Ausaged Fe-Ni-Co-Ti Shape Memory Alloys", Materials Transactions Jim 33, 1, 1 (1992). Y. J. Zhang, Y. T. Cheng, D. S. Grummon, "Indentation stress dependence of the temperature range of microscopic superelastic behavior of nickel-titanium thin films", Journal of Applied Physics 98, 3, (2005). M. Gelfi, G. M. La Vecchia, N. Lecis, S. Troglio, "Relationship between through-thickness residual stress of CrN-PVD coatings and fatigue nucleation 251 224. 225. 226. 227. 228. 229. 230. 231. 232. 233. sites", Surface & Coatings Technology 192, 2-3, 263 (2005). W. J. Chou, G. P. Yu, J. H. Huang, "Mechanical properties of TiN thin film coatings on 304 stainless steel substrates", Surface & Coatings Technology 149, 1, 7 (2002). T. W. Kim, S. W. Park, in Fracture And Strength W Solids, Pts I And 2. (2000), vol. 183-1, pp. 1279-1284. U. Pettersson, S. Jacobson, "Influence of surface texture on boundary lubricated sliding contacts", Tribology International 36, l 1, 857 (2003). G. Ryk, Y. Kligerman, I. Etsion, "Experimental investigation of laser surface texturing for reciprocating automotive components", Tribology Transactions 45, 4, 444 (2002). I. Etsion, G. Halperin, "A laser surface textured hydrostatic mechanical seal", Tribology Transactions 45, 3, 430 (2002). A. Ronen, I. Etsion, Y. Kligerman, "Friction-reducing surface-texturing in reciprocating automotive components", Tribology Transactions 44, 3, 359 (2001). P. Barungart, D. J. Krajnovich, T. A. Nguyen, A. C. Tam, "A New Laser Texturing Technique For High-Performance Magnetic Disk Drives", Ieee Transactions 0n Magnetics 31, 6, 2946 (1995). W. Y. Ni, Y. T. Cheng, D. S. Grummon, "Recovery of microindents in a nickel-titanium shape-memory alloy: A "self-healing" effect", Applied Physics Letters 80, 18, 3310 (2002). L. E. Sarnuels, T. O. Mulhearn, "An Experimental Investigation of the Deformed Zone Associated with Indentation Hardness Impressions", Journal of the Mechanics and Physics of Solids 5, 2, 125 (1957). J. G. Swadener, E. P. George, G. M. Pharr, "The correlation of the indentation size effect measured with indenters of various shapes", Journal of the Mechanics 252 234. 235. 236. 237. 238. 239. 240. 241. 242. 243. and Physics of Solids 50, 4, 681 (2002). W. D. Nix, H. J. Gao, "Indentation size effects in crystalline materials: A law for strain gradient plasticity", Journal of the Mechanics and Physics of Solids 46, 3, 411 (1998). R. F. Hehemann, G. D. Sandrock, "Relations between Premartensitic Instability and Martensite Structure in Tini", Scripta Metallurgica 5, 9, 801 (1971). M. Nishida, C. M. Wayman, R. Kainuma, T. Honma, "Further Electron-Microscopy Studies of the Til 1Nil4 Phase in an Aged Ti-SZat-Percent-Ni Shape Memory Alloy", Scripta Metallurgica 20, 6, 899 (1986) T. Saburi, S. Nenno, T. Fukuda, "Crystal-Structure and Morphology of the Metastable X-Phase in Shape Memory Ti-Ni Alloys", Journal of the Less-Common Metals 125, 157 (1986). T. Hara, T. Ohba, K. Otsuka, M. Nishida, "Phase transformation and crystal structures of TiZNi3 precipitates in Ti-Ni alloys", Materials Transactions Jim 38, 4,277(1997) A. Taylor, R. W. Floyd, "Precision Measurements of Lattice Parameters of Non-Cubic Crystals", Acta Crystallographica 3, 4, 285 (1950). M. H. Mueller, H. W. Knott, "Crystal Structures of Ti2Cu, Ti2Ni, Ti4NiZO, and Ti4Cu20", Transactions of the Metallurgical Society of A ime 227, 3, 674 (1963). O. H. Massalski TB, et al., "Binary Alloy Phase Diagrams", Binary Alloy Phase Diagrams, ASM Internaltional 3, (1990). K. Otsuka, T. Kakeshita, "Science and technology of shape-memory alloys. New developments", Mrs Bulletin 27, 2, 91 (2002). S. Miyazaki et al., "Shape Memory Effect And Pseudoelasticity In A Ti-Ni Single-Crystal", Scripta Metallurgica 17, 9, 1057 (1983). 253 244. 245. 246. S. Miyazaki, Y. Ohmi, K. Otsuka, Y. Suzuki, "Characteristics Of Deformation And Transformation Pseudoelasticity In Ti-Ni Alloys", Journal De Physique 43, NC-4, 255 (1982). K. Otsuka, X. Ren, "Physical metallurgy of Ti-Ni-based shape memory alloys", Progress in Materials Science 50, 5, 511 (2005). "Surface Texture, Surface Roughness, Waviness and Lay:ASME B4 6. I -2002 " (ASME, 2003) 254