.,. I..,':"_,Q‘,!l w M a ‘ ‘r #311" 1’5. , 5;? 615;. 23%,? ‘1’". ANNIE-11.1 ‘ :1: 5011:. 2,- J ‘ram‘J‘: -v~ ‘ r. . ‘ .sztJ‘i’ 5- 41.}.5' . .r'. " 4:2: 2.... :32. ~ 0. if"? $.51 #1, *rd‘"’*;’?:% 'afigs“ ”$2? i'SE-fi‘g'm" w 1 v ir'LCL h ,z.:: ¢ .J‘ M *1“??? .3532? ' M =. 96.15.. r' ""12" sv’gfézwfiyrflr' '1' "7m“. "” . ’ ,._{..’ g“... ”3&1, m4... ‘ 253:“ m -- ~..1..;:-""1: “NZ! “ m. - 4.5“qu “my 3:”? ‘1»? gu {Arm 4C..M-x-'n‘-:' '.,:> .. -,x,;...r.a ‘1' (“’4‘ "“W‘“' v: wk: 'é'vP-kymiifi. h ‘z #4- " , . . V "xgr'Zgr‘" . ‘71 Si... u I" if“; "M! ‘ w ..- "ab...” 2;: w. “M Mfifisi M . 4 in. , :ifiéfim :ic::,:;'n flgrlz'fll: a "" 1“"..33'31'!’ ' " IE"? M" .,u. 3'1" . “‘2: .73, '7“; . 5,. g "g? a; ‘ «u 9'. . “.3: 3‘ {fig-2’2 L'wgg‘mx ) Wm H I" J 31,3 .3” 'iwgc " ”Fig.1. 2* ”I? . g; «€37 fig: I“? . “.58”? 31% yr)?“ Wham-.19 314"!“ * A: - 5.135% It"? " it}. .545 .3312}. £356 ‘3‘; fféy W”; . I 13%.... ~ ‘3 53’“? was. .5 16...»? ‘ ’W .7 12:.5. Nruv‘jég'w 'h ”1.. e 5?: l' O! V\V V V I I ‘willlilllllllllmlllll ‘ 3 1293 00785 15 This is to certify that the thesis entitled Ni-Ti-Cu Thin Films by a DC Sputtering System—a presented by Chimin Hu Simpson has been accepted towards fulfillment of the requirements for MASTER degree in Metallurgy, Mechanics, and Materials Science " L/ /\1 Major pro essor Jun 2 1 Date 8’ ’ 989 0.7639 MS U is an Affirmative Action/Equal Opportunity Institution ' LIBRARY Michigan State University ' —_ fl vo-Nq Q PLACE Ill RETURN BOX to remove this chockout from your record. TO AVOID FINES return on or before duo duo. DATE DUE DATE DUE DATE DUE l I ‘1 I . A —— .— pr ’ . MSU Is An Aflirmdlvo Action/Equal Opportunity lmtiMion ommt NI-TI-CU THIN FILMS BY A DC SPUTTERING SYSTEM By Chimin Hu Simpson A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Metallurgy, Mechanics, and Materials Science 1989 ’4ai7q #1 u- ABSTRACT NI-TI-CU THIN FILMS BY A DC SPUTTERING SYSTEM By Chimin Hu Simpson Ti-Ni-Cu thin films produced by a DC sputtering system were studied in the present work. TiSO.OSN144.99CUS.96 was the sputtering source used. Thin films with 2500A, 5000A, and 13800A thicknesses were sputtered on pure copper and salt (KCl) substrates. The compositions of the films, determined by x-ray microanalysis, were slightly different from the sputtering source material. X-ray diffractometry and transmission electron microscopy were used to investigate structures. The as-sputtered films were found to be amorphous and had zone 1 and zone T morphologies. After annealing the films at 933K for 30 minutes, the monoclinic structure of martensite and rhombohedral structure associated with premartensitic phase were observed. From selected area diffraction patterns, 1/3 reflections representing the rhombohedral structure and 1/2 reflections signifying suboxide precipitates were found. The thermoelastic effect in the Ti-Ni-Cu films was confirmed by in-situ heating and cooling in the TEM. To my parents and Scott He He He ACKNOWLEDGMENTS I wish to thank Dr. D.S. Grumman, my advisor, for his patient guidance, creative ideas and scientific expertise. I would also like to thank Dr. W. Pratt, Professor of Physics, for his generosity in providing liberal access to the sputtering system. Dr. C. Fierz who operated the sputtering system, Dr. K. Klomparens and Mr. V. Shull who taught me the techniques involved in EDS, were also appreciated. iv TABLE OF CONTENTS TITLE LIST OF TABLES LIST OF FIGURES 1. INTRODUCTION 2. REVIEW OF THE LITERATURE 2.1 Fabrication of Ni-Ti Thin Films 2.2 Sputtering Deposition 2.3 Physical Properties of Thin Films 2.4 Phase Transformations and Crystal Structures in Ni-Ti Alloys 2.5 Microstructures of Ni-Ti Alloys 2.6 Studies of Ni-Ti-Cu Alloys 3. EXPERIMENTAL PROCEDURE 3.1 Sputtering Experiments 3.1.1 Sputtering Source Preparation 3.1.2 Substrate Materials and Preparation 3.1.3 Sputtering System 3.1.4 Sputtering Procedure 3.2 Composition Analysis 3.2.1 Preparation of the Thin Films for EDS 3.2.2 Preparation of the Standard Material for EDS 3.2.3 Set-up Page viii. ix \4 8‘ $‘ $‘ F‘ 10 11 14 14 14 15 15 19 20 22 22 23 (Continued) 3.3 X-Ray Diffractometry of the As-sputtered Films 3.4 Transmission Electron Microscopy of the As-sputtered Films 3.4.1 Preparation of the Thin Films on Cu Substrates 3.4.2 Preparation of the As-sputtered Films on Salt Substrates 3.5 Annealing Process 3.6 X-Ray Diffractometry of the Annealed Films 3.7 Transmission Electron Microscopy of the Annealed Films 4. EXPERIMENTAL RESULTS AND ANALYSIS 4.1 Composition Analysis and Results 4.2 Structure Identification Results 4.3 Observation of In-situ Crystallization 4.4 Microstructure Determination Results 4.4.1 Microstructure of the As-sputtered Films 4.4.2 Microstructure of the Annealed Films 4.5 Thermoelastic Effect Verification 5.DISCUSSION 5.1 Compositions of the Films 5.2 Structures of As-sputtered Films 5.3 Morphology of As-sputtered Films 5.4 Rhombohedral Structure 5.5 Morphology of Annealed Films 5.6 Interstitial Ordering of Oxygen 5.7 Transformation Temperature vi 25 25 25 26 26 27 27 28 28 35 39 45 45 54 64 73 73 75 75 76 76 77 78 6. CONCLUSIONS 7. REFERENCES (Continued) vii 80 81 Table 10 11 12 13 14 LIST OF TABLES Sputtering Yields for Several Pure Metals. Composition of the Sputtering Source Material. Transformation Temperatures of the Sputtering Source Material. Set-up Conditions for Sputtering System. EDS Results for 5000A Thin Films EDS Results for 13800A Thin Films. EDS Results for the Standard Material for 5000A Thin Films. EDS Results for the Standard Material for 13800A Thin Films. K Values for the Standard Material. Compositions of the As-sputtered Films Indexed Peaks of X-Ray Diffraction Patterns for the Sputtering Source Material. Indexed Peaks of X-Ray Diffraction Patterns for the Annealed 5000A Thin Film. Indexed Peaks of X-Ray Diffraction Patterns for the Annealed 13800A Thin Film. D-Values Obtained from Figure 15 (d) viii Page 14 14 21 30 31 32 33 34 35 43 44 44 48 Figure 10 ll 12 l3 14 15 LIST OF FIGURES Structure zone model [Thornton, 1974]. M as a function of Ti Contents, x binary NiTi aIloy,0 Ni-Ti-Swt% Cu alloy [Mercier, 1979]. Lattice parameter a of the martensite phase, plotted as a function of Cu content [Bricknell, 1979]. Lattice parameters of the martensite phase(fl-97.5°), plotted as a function of Cu content [Bricknell, 1979]. Disc sample holder. A schematic view of the sputtering system. Standard preparation, for composition analysis, by an ion mill system. A typical EDS spectrum. X-ray diffraction pattern for an as-sputtered 5000A thin film. ‘ X-ray diffraction pattern for an as-sputtered 138OOA thin film. X~ray diffraction pattern for the background (salt + glass). X-ray diffraction pattern for an annealed 5000A thin film. X-ray diffraction pattern for an annealed 138OOA thin film. X-ray diffraction pattern for the sputtering source material. A sequence of electron diffraction patterns taken from an as-sputtered 5000A thin film on Cu substrate, (a) at room temperature before annealing, (b) at 880K after annealing for 10 minutes, (c) at 880K after annealing for 15 minutes and (d) at room temperature after annealing for 30 minutes. ix Page 12 12 16 17 24 29 36 37 38 40 41 42 46 16 17 18 19 2O 21 22 (Continued) Transmission electron micrograph (a) and corresponding electron diffraction pattern (b) taken from an as-sputtered 13800A thin film at room temperature; (a) bright field image, showing Zone T structure and (b) electron diffraction pattern, showing amorphous structure. Transmission electron micrograph (a) and corresponding electron diffraction pattern (b) taken from the same sample as Figure 16 at room temperature; (a) bright field image, showing Zone 1 structur and (b) electron diffraction pattern, showing amorphous S truc ture . Transmission electron micrograph (a) and corresponding electron diffraction pattern (b) taken from the same sample as Figure 16 at room temperature; (a) bright field image, showing Zone 1 and Zone T structures found together and (b) electron diffraction pattern, showing amorphous structure. Transmission electron micrograph (a) and corresponding electron diffraction pattern (b) taken from an as-sputtered 5000A thin film at room temperature; (a) bright field image, showing Zone 1 and Zone T structures and (b) electron diffraction pattern, showing amorphous structure. Electron diffraction patterns, taken at room temperature from (a) an as-sputtered 5000A thin film on Cu substrate and (b) an as-sputtered 2500A thin film on Cu substrate. showing amorphous structure. Transmission electron micrographs, taken at room temperature from (a) an annealed 5000A thin film and (b) an annealed 13800A thin film, showing the polycrystalline structure after annealing the films at 930K for 30 minutes. Electron diffraction patterns, taken at room temperature from (a) an annealed 5000A thin film, [111] zone and (b) an annealed 13800A thin film, [1—12] zone, showing monoclinic structure associated with martensite. 49 50 51 52 53 55 56 23 24 25 26 27 28 29 3O 31 (Continued) Electron diffraction pattern [001] zone, taken at room temperature from the sputtering source material, showing martensite structure. 57 Electron diffraction pattern (a) and transmission electron micrograph (b), taken from an annealed 13800A thin film at room temperature; (a) diffraction pattern [100]B zone, showing 1/2 , 1/4 and 1/3 reflections and (b) dark field image, formed by one of 1/3 reflections encircled in (a), showing antiphase boundaries. 58 Electron diffraction pattern (a) and transmission electron micrographs (b) and (c), taken at room temperature from the same sample as Figure 24; (a) diffraction pattern, [111] 2 zone, showing 1/2 and 1/3 reflections, (b) dark field image, formed by one of 1/3 reflections encircled (A area) in (a), showing antiphase boundaries and (c) dark field image fromed by one of 1/2 reflections encircled (8 area) in (a), showing oxide precipitates which not observed in (b). 59 Electron diffraction pattern [100]B zone, taken at room temperature from an annealed 5000A thin film, showing 1/2 and 1/3 reflections. 61 The ESCA spectrum of the sputtering source material, showing Ti, Ni, Cu, and 0 peaks. 63 The ESCA spectrum of an as-sputtered 138OOA thin film, showing Ti, Ni, Cu, and 0 peaks. The 0 peaks not so obvious as those in Figure 27. 65 Electron diffraction patterns, taken from an annealed 13800A thin film at room temperature; (a) [112]82 zone and (b) [110]B2 zone, showing 1/2 and 1/3 reflections. 66 Electron diffraction patterns, taken from the sputtering source material at room temperature; (a) [100] zone, (b) [110]82 zone, (c) [112]B and (d) [III] zone, showing 1/2 reflections gut 1/3 . B2 . reflections only seen in (d). 67 Transmission electron microscophs, taken from an annealed 5000A thin film at different temperatures by controlling the electron beam; (a) bright field image, taken as beam converged and (b) bright field image, same place as (a), taken as beam spread, showing the martensite plate growing in the high temperature structure in (a) during the cooling process. 70 xi 32 (Continued) A sequence of diffraction patterns, taken from an annealed 13800A thin film; (a) at room temperature, (b) as the electron beam converged and (c) as electron beam spread, showing 1/3 reflections in (a) disappeared in (b) when temperature raised, and appeared in (c) as temperature reduced. xii 71 1. INTRODUCTION The present study has been undertaken to investigate the feasibility of producing superelastic thin films in the Ni-Ti system by DC sputtering. These films are of interest as potential surface microalloys for the purpose of improving fatigue crack initiation life in metal alloys by inhibiting the penetration of persistent slip bands to the free surface [Meyers, 1988]. In fatigue tests, two characteristics of fatigue crack initiation are: (1) the initiation phase of fatigue can play a substantial role in the total fatigue life; and (2) surface processes can strongly influence the nucleation of fatigue cracks. Furthermore, fatigue crack initiation at low strain amplitudes involves two main processes: the localization of plastic strain forming persistent slip bands in the interior of the material, and the progressive development of stress concentrations on the surface where the persistent slip bands emerge. The latter leads rapidly to the initiation of fatigue cracks. It is thus desirable to attempt to modify the surface of the material in order to restrict persistent slip bands penetrating to the free surface and therefore prolong fatigue life [Grummon, 1988]. However, so far no surface microalloy has been found which significantly inhibits persistent slip band penetration. In the present work, a new class of surface modification alloys were explored which may have potential for improving fatigue properties in metals due to the unique superelastic stress-strain behavior. Superelastic NiTi bulk alloys have superior properties in 1 ductility, strength, fatigue, and resistance to corrosion, and it has been shown that some of these properties, particularly fatigue estimate, result specifically from the superelastic behavior inherent in these materials. Anelastic strain larger than 5% in NiTi alloys can more or less be recovered upon unloading. NiTi alloys have been noted to endure more that 107 cycles of the fatigue under constant plastic strain of 1% and applied stresses up to four times the yield strength [Buehler, 1968]. This beneficial fatigue property has not been found in other superelastic alloys, such as Cu-Zn-Al, Cu—Zn-Sn and Cu-Al-Ni. These facts strongly suggest that NiTi surface alloys would be able to accomodate high plastic strain amplitudes without suffering excessive damage and thus delay persistent slip bands from forming on the surface. Furthermore, Nil- xTiCux alloys have been shown to retain superelastic properties for values of x as high as 0.7 [Melton, 1979]. This observation presents the possibility of applying NiTi surface alloys to Cu based alloys, , without serious complications from Cu contamination of superelastic surface alloys. In the present work, Ti-Ni-Cu thin films produced from a T150 5N1 target by a DC triode sputtering were 44.99C“4.96 investigated. Energy dispersive x-ray spectrometry (EDS) was used to determine the compositions of the films. The compositions of the thin films and of the sputtering source material will be discussed, and data will be presented from x-ray diffractometry and transmission electron microscopy (TEM) which characterize the structures of these films. Crystallization of the films was observed in the TEM by using its heating stage. To measure oxygen content in the sputtering source and thin films, ESCA experiments were performed. Since thermoelastic effect is accompanied with the superelastic effect in Ni-Ti binary and ternary alloys, the thin films were observed in the TEM during in situ heating and cooling experiments to see if thermoelastic effect occurred. 2. REVIEW OF THE LITERATURE 2.1 Fabrication of Ni-Ti Thin Films Recently, several techniques have been applied for forming Ni-Ti thin films. Rai, et. a1. [Rai, 1987] fabricated 850 nm thick Ni Ti amorphous films by sputtering deposition from composite 52 48 NiSOTi50 targets. Multiple layer vacuum evaporated composite layers of 140 nm thickness were also prepared which were mixed and amorphized by heavy ion irradiation. Magnetron sputter deposition has been used to make films up to 10 micrometers thick [Kim, 1986] with a nominal composition of Ni56T144. Liu, et. a1. [1983] also produced ion beam mixed surface alloys by sequential vacuum evaporation of 100 nm thick Ni and Ti layers. These films possessed amorphous structure after irradiation and transformed to the equilibrium crystalline phase at 730 to 830K. 2.2 Sputtering Deposition Deposition of films by sputtering was first studied in 1882 by Grove. The principles of sputtering deposition are as follows: Typically, the target (sputtering source) is connected to a negative voltage supply, and the substrate holder grounded or biased (usually negative potential) as an anode. An inert gas is introduced to provide a sputtering medium. When the gas glow discharge is started, the positive ions strike the target (cathode) and the atoms on the surface of the target are removed by momentum transfer. These neutral target atoms then condense into thin films. Most metal elements have high sputtering yields [Vossen, 1978]. Examples in 4 Table 1, indicate that it does not take a great amount of time in order to form the desired thicknesses. Compared with thermal evaporation deposition, sputtering deposition is more likely to maintain a constant composition in films sputtered from alloy targets [Tarng, 1971]. Theoretically, if a target is kept cold enough so the diffusion of atoms in the target can be prevented, then after the steady state is established the sputtered films should have the same compositions as the target. The steady state is a very important factor in the sputtering process. It means that the composition of the sputtered flux leaving the alloy target equals that of the bulk alloy. Every element has its own sputtering yield. In a compound alloy target, when the sputtering process starts, the elements with higher sputtering yields will be sputtered faster than the elements with lower sputtering yields. Therefore, in the early stage of the sputtering process, the sputter flux has a different composition than the target: the higher the sputtering yield, the richer the sputter flux. However, the composition on the surface of the target changes due to the different sputtering yield. The elements with lower sputtering yields become enriched on the surface of the target since they are sputtered away from the surface more slowly than the elements with higher sputtering yields. Because the surface composition is altered, the sputtering flux is changed too; the richness of the elements in the sputter flux will be reduced since their surface concentrations are decreased. So, after a period of sputtering, the sputter flux will be the same as the composition of the bulk target material, although the composition of the target surface is altered. If the shutter of the substrate holder is Table l Sputtering Yields for Several Pure Metals at SOOeV [Vossen, et a1. 1978] \ Gas I He Ne Ar Kr Metal \ | I Al | 0.16 0.73 1.05 0.96 I I Fe | 0 15 0.88 1 10 l 07 | I Ti | 0 07 0 43 0 51 0 48 | I Ni | 0.16 1.10 1.45 1.30 I I Cu | 0 24 1.80 2 35 2 35 I Pd | 0 13 1.15 2 O8 2 22 | | Ag | 0 2 1 77 3.12 3 27 I Au | 0 07 1.08 2 40 3 06 | opened, the composition of the thin film should then be identical to the target. Practically, however, compositions of films are dependent on several factors, such as substrate bias, angular ejection distribution, sticking coefficient and resputtering effect. Although a few workers have studied the relationship between these factors [Cuomo, 1975; Winters, 1969; Haff, 1976; Sigmund, 1980], composition control remains very complicated. A number of suggestions for managing the compositions of the films were summarized [Vossen, 1978] as follows: (1) The target material must be homogeneous, (2) The target must be presputtered to establish equilibrium, (3) The target needs to be maintained cold to prevent bulk diffusion of the target material. Still, most variables such as the presputtering time, the bias potential and setup for the sputtering process, need to be obtained. empirically for each type of material. 2.3 Physical Properties of Thin Films Physical properties such as morphologies of the films and grain size, depend on the preparation conditions of deposition, namely, two factors: substrate temperature and argon gas flow. The microstructures of thick films related to the parameter T/Tm ( T is the substrate temperature and Tm is the melting temperature for the target material) were first described by Movchan and Demchishin [Movchan, 1969]. They developed a structure model with the parameter T/Tm (the homologous temperature). Thornton (1974) added an additional parameter, sputtering gas pressure, and proposed a diagram (Fig.1) named the "structure zone model." Zone 1 structure is an open structure, resulting from adatom insufficient diffusion or rough substrates. Zone 2 structure possesses columnar grains which increase in width with T/Tm. The Zone T structure, consisting of a dense array of poorly defined fibrous grains, is a transition structure located between Zone 1 and Zone 2; however, it is difficult to draw a line separating Zone T and Zone 1. Zone 3 structures, with equiaxed grains, grow on the substrates when T/Tm is higher than 0.8. Although this model was developed primarily for crystalline thick films, some amorphous semiconductor thin films were also observed to possess these zone structures [Ross, 1981, 1983; Messier, 1982]. 2.4 Phase Transformations and Crystal Structures in Ni-Ti Alloys Phase transformations in Ni-Ti or Ni-Ti-X alloys (X= Au, Fe, Al, Cu) have been studied by using the TEM, x-ray diffractometry, and neutron diffraction. The sequence of the transformations upon cooling for most Ni-Ti alloys can be described as follows: Parent phase (82) ---> incommensurate phase (distorted cubic) ---> commensurate phase (rhombohedral) ---> martensite phase (monoclinic) [Hwang, 1984; Miyazaki, 1986]. For Ni-Ti binary alloys, the parent phase is known to possess a 8 structure with a unit cell of 2 a a 3.015A [Philip, 1961]. The rhombohedral phase (R-phase), 0 was considered the premartensitic phase, with suggested parameters of a-9.03A, and amin-89.3° by Chandra et a1. (1968). The lattice parameters of the monoclinic structure associated with the martensite phase were summarized as follows: a=(2.889i 0.005)A, ZoneT'o' Iii/MW” D TIT m Figure 1 Structure zone model [Thornton. 1974] 100 \ \ ~50 46 45 wt°/oTi 44 Figure 2 M alloy, 5 as a function of Ti Contents, x binary NiTi Ni-Ti-Swt% Cu alloy [Mercier, 1979]. 10 b-(4.120i0.012)A, c-(4.622i0.016)A, and fi-(96.8i0.32)° [Hehemann, 1971; Otsuka, 1971; Michal, 1981; Buhrer, 1982]. These parameters vary slightly for Ni-Ti-X alloys due to the difference in size between Ni atoms and X atoms (Fe, Cu, Au, Al). The lattice parameters of Ni-Ti-Cu will be described in 2.6. 2.5 Microstructures of Ni-Ti Alloys Images and selected area diffraction patterns (SADPs) of Ni-Ti or Ni-Ti-X alloys reveal the microstructures of these alloys, and have been published in a number of papers. The relationship of the martensite plates and prior austenite grains were observed in the present work [Sinclair, 1978]. The martensite phase was found to consist of many differently oriented variants and was internally twinned. SADPs taken from Ti-49.75 at% Ni alloys [Otsuka, 1971] substantiate that the 82 structure is the parent phase and that the martensite phase has a monoclinic structure. The existence of the premartensitic phase, named the R-phase, with a rhombohedral structure was verified by extra 1/3 reflections in SADPs, and antiphase domains in the dark field images from 1/3 reflections were also observed [Hwang, 1983]. From the plot of electrical resistivity vs. temperature, Hwang et a1. noted that electrical resistivity started to increase when temperature decreased to T , and then resistivity reduced when temperature decreased further to MS. The structure in this region was found to be rhombohedral. Moreover. between T1) and Ms 1/3 reflections were observed between the temperatures Ms and Tp, which verifies that the 1/3 reflections result from the presence of the rhombohedral structure. In situ 11 observation [Otsuka, 1971] showed that the 1/3 reflections appeared upon cooling, and disappeared while heating, confirming the thermoelastic effect (82 a R-phase) of Ni-Ti alloys. Besides 1/3 reflections, 1/2 reflections were also observed in TEM. Recently, the 1/2 reflections were suggested to result from the interstitial ordering of hydrogen and/or oxygen [Wu, 19883]. Hydrogen may be from the jet-polishing solutions used during the thinning process for TEM samples, and oxygen may result from the raw Ti material or from the arc melting process which produces Ni-Ti bulk materials. Interstitial hydrogen atoms are not stable in Ni-Ti alloys and diffuse out at room temperature. However, interstitial oxygen atoms occupying octahedral sites remain stable in Ni-Ti alloys and form suboxides. The stoichiometric formula for the suboxide was determined [Wu, 1988b] to be Nil6Ti1604 in binary Ni-Ti alloys. 2.6 Studies of Ni-Ti-Cu Alloys One important characteristic of NiTi xCuX alloys is that the MS 1- temperature is near or above room temperature when x>3 [Tadaki, 1982]. The Ms temperatures of Ni-Ti-Cu alloys do not depend on the titanium content as much as in binary Ni-Ti alloys [Mercier, 1979]. The crystal structures of Ni-TiLCu alloys were investigated by Bricknell et a1. [1979]. From their studies, we know that when copper atoms substitute for nickel atoms, the lattice parameters a of the austenite are expanded (Fig. 3) since the size 0 of copper atoms is larger than that of nickel atoms. The structure parameters for martensite exhibit a similar tendency (Fig.4). The phase transformations in Ni-Ti-Cu alloys undergo the same thermal 12 o 1 A 3.1 ‘ I"‘--- ‘--.-‘-‘----. Am!" so J} 2.9 1 ‘ ‘ ‘ ‘# 5 1O 15 20 5 at%Cu Figure 3 Lattice parameter a of the martensite phase, plotted as a function of Cu content ?Bricknell, 1979]. o A E 4-61 '”"-O--o------- --3------ C 4.4 1 4.2 . , ’_"_‘______.....,..---_.. b 1" 4.04 w’ 3'0 ‘0‘.“__,._-..----------'I---..., a 2.8 ‘ 5 10 15 20 25 81% CU Figure 4 Lattice parameters of the martensite phase( fi=97.5°), plotted as a function of Cu content [Bricknell, 1979]. 13 sequence as binary alloys when x<19; however at x-l9, the rhombohedral structure did not occur [Tadaki, 1982]. In situ TEM observation of Ti-hO.5Ni-lOCu (at%), using the technique of converging and spreading the electron beam to effect a change of temperature [Saburi, 1986], showed that the martensite phase nucleated at certain stress concentrations, such as the matrix-oxide interface, and boundary dislocations, proving that thermoelastic behavior can take place in TEM thin foils. 3. EXPERIMENTAL PROCEDURE 3.1 Sputtering Experiments 3.1 l Sputtering Source Preparation: The composition of the sputtering source material for the sputtering experiment is listed in Table 2. Table 2 Composition of the Sputtering Source Ti Ni Cu Weight percentage 44.78% 49.34% 5.88% Atomic percentage 50.05% 44.99% 4.96% The material, donated by U.S. Nitinol Co., was specified to have the transformation temperatures given in Table 3: Table 3 Transformation Temperatures of the Sputtering Source Temperature 46°C 63°C 40°C 24°C The sputtering source material was machined to 57 mm diameter x 4 mm thickness. Before the sputtering source was placed in the sputtering 14 15 machine, the sputtering source was cleaned by methanol and deionized water . 3.1.2 Substrate Materials and Preparation Two materials were selected to be the substrate materials: pure copper (99.90%) sheet purchased from the EM Science Company for TEM study, and salt material (KCL) needed for experimentally determining compositions of the thin films by Energy Dispersive Spectrometry. Due to the size (11 mm x 13 mm) of substrate holders of the sputtering system, salt material was made the same size as the substrate holders. However, for the transmission electron microscope experiment, the copper substrate material was mechanically punched into 3 mm diameter discs. These 3 mm diameter discs were placed in special disc holders designed for mounting them into the substrate holders of the sputtering system. These disc holders were made of aluminum and are described further in 3.1.4. 3.1.3 Sputtering System A schematic view of the DC planar triode sputtering system is displayed in Figure 6. The substrate plate and disk plate are grounded as anodes and the sputtering source installed below the substrate plate is the cathode. The sputtering sources and all accessories are mounted on ultra-high-vacuum conflat flanges. There was no provision for adjusting the temperature of the substrate plate or the disc plate. The temperatures of these plates were estimated to be approximately 310K. There are four identical sputtering sources in the sputtering 16 Figure 5 Disc Sample Holder 17a Figure 6 A schematic view of the sputtering system. P‘F‘ HOOQVO‘U‘DUJNH stepping motor. . rotary feedthrough. rotary feedthrough. . meissner trap. ionization gauge. . convectron gauge. . wobble stick. . ultra high purity argon tank. . cold trap gas purifier. . hydrox gas purifier. . gas flow controllers. 12. l3. 14. 15. 16. 17. 18. 19. 20. 21. 22. rotary feedthrough. residual gas analyser. cryopump. gate valve. mechanical pump. bellows sealed valve. FTM plate. Substrate plate. disk plate. chimney assembly. sputtering sources. 17 E |_.I E U E] E I= 7 Figure 6 A Schematic Wow of the SputteringSystem: 18 system. Each can be removed without disturbing the others. The chimney assembly, positioned above the sputtering source, is fixed to a rotating table. Therefore, the sputtering sources can be covered with the closed chimneys by rotating the table. The substrate materials can be placed in either the substrate plate or in the disk plate. The substrate plate with two 5 cm diameter holes and the disk plate with ten 5 cm diameter holes are both made of aluminum. A round plate (5 cm diameter), accommodating two substrate holders with the size 11 mm x 13 mm, is fixed into each hole. The substrate materials can be placed in the substrate holders. The thicknesses of the sputtered films can be measured by two quartz crystal film thickness monitors (FTMs) located above the substrate plate. The cryopump (TI cryo-Torr8) provides high pumping speeds (1500 l/s air, 4000 l/s water) with no oil vapor contamination. Typically this sputtering system can be pumped to 10-8 torr with the cryopump and the mechanical pump working in tandem. After gently heating to about 320K overnight, a vacuum better than 5x10.9 torr can be obtained. The sputtering gas comes from a cylinder of ultra-high purity argon gas which is further purified by first passing through a cold trap maintained about 100K in order to freeze out water vapor. Furthermore, a gas purifier (Matheson Hydrox purifier, 8301) connected with the cold trap gas purifier can extract impurities such as oxygen and nitrogen by reacting with a hot Ti-based alloy. The gas flow controller limits the fluctuation of the sputtering gas flow to within a i0.5% error range. The voltages and currents of the sputtering sources are manually kept constant to maintain steady 19 sputtering rates. The sputtering source voltage and current affect the deposition rates; a high negative voltage, resulting in high current through the sputtering source, gives a high deposition rate for producing the films. 3.1.4 Sputtering Procedure The Ti-Ni-Cu alloy sputtering source was placed in one of the four sputtering source holders. The remaining holders were covered by the closed chimneys. The disk plate was used in the present work instead of the substrate plate, because the former plate can hold more substrate materials. The salt substrate materials were placed in the regular 11 mm x 13 mm substrate holders. However, for fixing the 3 mm diameter copper discs, a few disc sample holders made of aluminum were designed (Figure 5). Each disc sample holder can retain five 3 mm diameter copper discs. The disc sample holders were stationed within the 11 mm x 13 mm substrate holders so that the copper discs could be fixed in the disk plate of the sputtering system. Neither disk plate nor substrate plate can have a bias potential applied. After pumping the sputtering system overnight, the pressure of the chamber approached less than 1x10.8 torr. Argon gas was then bled into the chamber and the pressure of argon gas was maintained at an approximate level of 0.0025 torr. The sputtering source gun was turned on next; however, the substrate materials were now covered by the masks mounted within the substrate holders. After the sputtering source material gun was presputtered for 10 minutes, the 20 masks of one salt substrate and one copper substrate were removed under computer control. After 5000A thickness films were formed on the substrates, the masks of the deposited substrates were closed. Next, the mask for one salt substrate was opened and a 13800A thin film was formed on the salt substrate. After the mask of this salt substrate was closed, the mask of one copper substrate was removed, and the deposition started forming 2500A thin films on the copper discs. Table 4 shows the conditions for the sputtering process of the different thickness Ni-Ti-Cu thin films on two substrate materials. 3.2 Composition Analysis Energy Dispersive X-ray Spectroscopy (EDS) was used for analyzing the composition of the thin films. This technique can determine the presence of various elements by analyzing the different energies of their characteristic x-rays, which are stimulated by incident electrons. Although the probe of incident electrons is very small, the area where x-rays are produced is much wider than the electron probe and reach deeply into the sample. Typically, this affected area has a tear drop shape. The size of the tear drop depends on several variables, such as material, energy of the incident electrons, and tilt angle. Furthermore, x-rays counts are influenced by three main factors: atomic number factor (Z), absorption factor (A), and fluorescence factor (F). These factors are correlated with the tear drop shape: the bigger the volume of the tear drop produced, the greater the ZAF correction required. If standards with known compositions have the same thicknesses as the 21 Table 4 Set-up Conditions for Sputtering System 5000A 5000A 13800A 2500A on Cu on KCL on KCL on Cu Ar pressure (torr) .00252 .00252 .0025 .00251 Deposition rate(nm/sec) 0.608 0.608 0.692 0.686 Voltage of sputtering source(V) -521 4775 -521.4775 -543.0684 -538.9658 Current of sputtering source(A) 0.90498 0.90498 0.94492 0.94539 Deposition time 13min 28sec 13min 285ec 32min 38sec 5min 57sec Finished Voltage of -518.6445 -518.6445 -543.0459 -54l.6543 sputtering source (V) Finished Current of 0.89125 0.89125 0.94165 0.94283 sputtering source (A) 22 unknown samples, the tear drops produced in standards and unknown samples would be similar, which means ZAF factors influence both samples in the same manner. Compositions of unknown samples then can be computed accurately by comparing them with the standards. 3.2.1 Preparation of the Thin Films for EDS The films (5000A and 13800A) on the salt substrates were chosen for the EDS experiments instead of the films on copper substrates in order to reduce the additional X-ray counts coming from the copper substrates. The salt substrates were removed from the distilled water and then acetone was used to clean the films. After the films were dry, with carbon paint the films were glued flat on the graphite sample holder designed for EDS equipment. 3.2.2 Preparation of the Standard Material for EDS In order to analyze the compositions of the thin films accurately, standards with the same thicknesses as the films are necessary for reference purposes. However, uniform standards with 5000A and 13800A are not very easily prepared. Therefore, the following technique was applied in this study: a rectangular piece of the sputtering source material was cut by a diamond saw. A 3 mm diameter disc (about 1mm thickness) was core drilled from the rectangular piece. This disc was carefully mechanically polished to a 0.08 mm thickness. Ion milling was used to thin this disc. There are two ion guns in the ion mill machine. The ion gun located above the sample plate thins the sample from the top, while the ion gun installed below the sample plate thins the sample from the bottom. 23 The angles between the ion beams and the surface of the sample can be adjusted by tilting the ion guns. The ion guns can be tilted between 0 and 90 degrees. In the present work, only the top gun was operated and the angle between the ion beam and the surface of the sample was set at 17 degrees. As shown in Figure 7, the locations of regions with thicknesses of 5000A and 13800A can be calculated from simple geometry. The distance (18 from the hole to the location with 5000A thickness is 1.63 pm, and the distance db from the hole to the location with 13800A thickness is 4.51 pm. Although the distances are very minute, after magnifying the details 50 x 103 times in the electron microscope, these regions can be readily found. 3.2.3 Set-up A x-ray detector (LINK System) was used in a JEOL lOOCX scanning transmission electron microscope with a LaB6 filament. SEM mode was selected since the thin films were too thick (13800A and 5000A) for electrons to penetrate them. 40 RV was used in the present work and the current was kept constant at 5 pA for all samples. The samples were tilted 30 degrees to face the x-ray detector, which was stationed above the sample plate. The live time was fixed at 200 seconds. Three energy windows for Ni, Ti and Cu were set up individually by computer program (ANlOOOO). The computer program also integrated the area of three windows and displayed the gross integral counts and net counts for each element set by the energy windows. The dead time was shown simultaneously on the computer screen . 24 da= ha x cot 17 ° = 50001 x cot 17° = 16354A 0 db = hb x cot 17 = 13800A x cot 17° = 451381 Figure 7 Standard preparation, for composition analysis, by an ion mill system. 25 3.3 X-ray Diffractometry of the As-sputtered Films The x-ray source of the x-ray diffractometer (XBD-S, General Electric Co.) provided Cu Ka radiation at a wavelength of 1.542A. In this study, the detector rotated from 10 to 130 degrees. The applied voltage was 30 KV. The entire experiment was carried out at room temperature. The thin films (5000A and 13800A) on the salt substrate materials were taped on a glass strip, which was inserted in the sample holder of the x-ray diffractometer. The salt substrates were not removed in this experiment. In addition to the thin films, the salt material (KCI) was taped to the glass strip to serve as a reference. 3.4 Transmission Electron Microscopy of the As-sputtered Films The Hitachi H-800 transmission electron microscope with a tungsten filament was utilized and the applied voltage was 200 KV. The TEM was operated at room temperature (295K) for the as-sputtered films. 3.4.1 Preparation of the Thin Films on Cu Substrates Because the thin films (2500A and 5000A) on copper substrates are thin enough for electron penetration, a South Bay Tech., Model 550 jet-thinner was used for thinning the copper substrates. The jet-polishing solution for pure copper was: phosphoric acid (90%) and deionized water (10%), (used at room temperature). The applied voltage and current were set 5V and 6.8 mA. The copper substrate materials were almost polished away in order to eliminate the 26 diffusion of copper from the substrates to the thin film during the annealing process. The copper substrates, like the copper grids, had big holes (about 2 mm diameter) in the centers after the polishing process. 3.4.2 Preparation of the As-sputtered Films on Salt Substrates The salt substrates under the films were dissolved in the distilled water. The thin films (5000A and 13800A) sandwiched by double copper grids (100 mesh / 200 mesh) were ion milled individually so that electron beams could penetrate the thin areas. 3.5 Annealing Process For the in-situ annealing, after polishing the Cu substrates, the films on the Cu substrates were annealed in the transmission electron microscope by using the heating stage. The annealing temperature was 880K and annealing time was approximately 30 minutes. The annealing temperature was raised by applying current to the sample holder. The current was controlled externally. The temperature can be determined by the current and a plot of current vs. corresponding temperature, supplied by Hitachi Co. The vacuum in the TEM was held better than 10.7 torr. During the annealing process, in situ crystallization was observed in the TEM. The thin films removed from the salt substrates were annealed in a vacuum furnace at 930K for 30 minutes and then the films were air quenched in the quartz tube. The films were trapped with two thin pieces of alumina which should reduce the reaction between 27 oxygen and Ni-Ti-Cu thin films. The vacuum was maintained about 10-5 during the annealing process. 3.6 X-ray Diffractometry of the Annealed Films The set-up was the same one described in 3.3, but the thin films were taped directly onto the glass strip without the salt substrates. A specimen of the sputtering source material, annealed at the same conditions as the films (in a vacuum furnace, 930K, 30 min.), was also analyzed for comparison. 3.7 Transmission Electron Microscopy of the Annealed Films The preparation of the thin films for the TEM was identical to 3.4.2. A 3 mm diameter disc of the sputtering source material, for use as reference, was also prepared by the technique of ion thinning. For observation of thermoelastic behavior in the annealed thin films, the technique of converging and spreading the electron beams was used to heat and cool the films. This was preferred to use of the heating stage because copper's diffusion from the grid to the films, and also the thermal shift could be eliminated, as well as vibration of the films reduced, during the heating process. 4. EXPERIMENTAL RESULTS AND ANALYSIS 4.1 Composition Analysis and Results Figure 8 exhibits a typical EDS spectrum for the thin films and the standard materials. The peaks of Ni, Ti, and Cu are readily distinguished. Occasionally, the peak of Pb was noticed, but it was from the carbon paint. Tables 5 through 8 show the EDS results for a 5000A thin film, a 13800A thin film and the corresponding standards. The tables include the net counts and qualitative analysis results, which were calculated by a computer program (ANlOOO, LINK System). The principles of analyzing the compositions for the films in the present study are explained as follows. The relationship between compositions, and net counts of x-ray is: - * CTi/ CNi KTiNi ITi/ INi - * CCu/ cNi KCuNi ICu/ INi' Cj is the weight percentage for j element, and Ij is the intensity (net counts) for j element. K is a material dependent constant. The weight percentage for each element of the standard is known (Table 2). Tables 6 and 7 give the net x-ray counts for Ni, Ti and Cu. Therefore, the K values can be computed by using the formula mentioned above. The results are given below, in Table 9. 28 29 HEHI: SPBSPCSLBZ ulNDON START END WIDTH GROSS NE‘I EFF. XAGE LABEL kev ch CHANS INTEGRAI INIEGRAI FACTOR TOTAL Tiha 4.30 a 72 22 8612 7259 1.00. «4.01 Nina 7.22 7 72 26 9381 8380 1.00 $0.80 CuKa 7.80 8 3O 26 3144 856 1.00 5.19 ' 1 >c.-F:Fi‘.- e - - . - a c I 'fi' -' n - O - - Li'u'e- .s'UU: F‘reSEU .-:.U-U°; Remaining: 0;. - .- _ r. - Real: did: -4 Dead 1 30 ..C - \ -0” Pa! 1'1E111 ' 31°23 133L192 If. ‘s,l’ Figure 8 A Typical EDS Spectrum. 30 Table 5 EDS Results for 5000A Thin Films Ti cnts. % Ni cnts. Cu cnts. %. Dead time 1 12681 48.72 12691 48.75 659 .53 7 % 2 13077 47.67 13520 49.29 834 .04 7 % 3 13139 47.70 13416 48.71 988 .59 7 % 4 15521 48.43 15763 49.18 767 .39 7 % 5 15425 47.18 15691 48.01 1571 .81 7 % 6 15413. 48.08 15665 48.86 980 .06 7 % 7 13569 47.92 13803 48.74 946 .34 6 % 8 13433 47.27 13866 48.80 1117 .93 6 % 9 13345 46.69 14013 49.03 1224 .28 6 % AVG. 13956 47.74 14270 48.81 1009 .45 EDS Results for 13800A Thin Films 31 Table 6 Ti cnts. % Ni cnts. Cu cnts. % Dead time 1 122030 49.21 115648 46.63 10319 .16 22 % 2 119881 49.03 113565 46.45 11065 .53 21 % 3 118603 48.72 114617 47.08 10226 .20 21 % 4 145123 48.24 141564 47.05 14169 .71 23 % 5 141012 48.16 138303 47.24 13464 .60 23 % 6 139809 48.70 135302 47.13 11950 .16 23 % 7 _130752 47.81 130336 47.66 12383 .53 21 % 8 131856 48.23 128011 46.82 13545 .95 22 % 9 130935 48.19 128266 47.21 12498 .60 21 % AVG. 131111 48.46 127290 47.04 12180 .50 EDS Results for the Standard Material for 5000A Thin Films 32 Table 7 Ti cnts. % Ni cnts. % Cu cnts. % time 1 12199 50.03 10974 45.01 1210 4.96 % 2 12254 51.06 10458 43.57 1289 5.37 % 3 12335 50.42 10770 44.03 1358 5.55 % 4 15363 50.95 13130 43.54 1663 5.51 % 5 15379 50.90 13333 44.13 1501 4.97 % 6 16327 49.89 14462 44.19 1939 5.92 % AVG. 13976 50.53 12188 44.07 1493 5.40 33 Table 8 EDS Results for the Standard Material for the 13800A Thin Films Ti cnts. % Ni cnts. % Cu cnts. % Dead time 1 139043 51.35 120187 44.39 11544 4.26 21 % 2 139475 51.28 120531 44.31 11996 4.41 21 % 3 139647 51.43 120915 44.53 10986 4.05 21 % 4 141080 51.54 121510 44.39 11119 4.06 21 % 5 142039 51.60 122438 44.48 10768 3.91 21 % 6 142313 51.65 122184 44.34 11052 4.01 21 % AVG. 140599 51.48 121294 44.41 11244 4.11 34 Table 9 K Values for the Standard Material 5000A standard KTiNi-O'7915 KCuNi- 0.97737 13800A standard KTiNi- 0.7830 KCuNi- 1.28667 The K values for the standards may then be applied to the as- sputtered thin film data. The net counts of the films for the three elements are listed in Tables 4 and 5. The ratio of the weight percentages for Ni, Ti and Cu can be thus calculated from the following formulae: - K * CTi/ CNi TiNi ITi/ INi - 71’ CCu/ CNi KCuNi ICu/ INi and CTi + CNi + CCu-l. The compositions of the as-sputtered films are displayed in Table 10: 35 Table 10 Compositions of the As-sputtered Films Thickness Ti wt% Ni wt% Cu wt% 5000A 42.00 54.26 3.74 13800A 41.80 51.82 6.38 Thickness Ti at% Ni at% Cu at% 5000A 47.14 49.70 3.16 13800A 47.02 47.57 5.41 4.2 Structure Identification Results Figures 9 and 10 display the scattering curves of the x-ray diffractometer for the as-sputtered thin films (5000A and 13800A). In these two figures, several peaks actually are from the background (salt+g1ass). Comparing Figures 9 and 10 with Figure 11, which is the x-ray diffraction pattern for the background, the peaks located at 20 -29.5°, 32.7°, and 67.2° in Figures 10 and 26=67.2° in Figure 9 are from the background. The as-sputtered films, therefore, possess an amorphous structure. Interestingly, the 5000A film demonstrates the second order peak but the 13800A film does not. The fact that the films crystallized during the annealing 36 RELATIVE INTENSITY . u—n III—ll-ll-ll— «I..- .--.;‘f 3= m ——n:: .' Ann-- “II-ll Inn- sewn-“mu. “I-" Figure 9 X-ra Diffraction Pattern for an As-sputtered 50 A Thin Film. 20 (DEGREE) 37 RELATIVE INTENSITY ._......._... :_.___::£; "mun—l. _—n_-_uunn_—— ‘ un- 'I' ======$“_=:L'==T_S:=:g=f_'£::===—rf==i Figure 10 X-ray Diffraction Pattern for an As-sputtered 13800A Thin Film. 20 (DEGREE) 38 RELATIVE INTENSITY Figure 11 X-ra Diffraction Pattern for the Ba round (Salt + Glass). 20(0EGREE) 39 process is confirmed by Figure 12 and 13. Since the salt substrates had been removed from the annealed films, all the peaks of the x-ray diffraction patterns belong to the thin films. However, the increasing background at low angles is from the glass strips, to which the films were taped. The x-ray diffraction pattern from the sputtering source material, as a comparison, is exhibited in Figure 14. From this plot, the lattice parameters for martensite structure and residual B2 structure can be determined [Edington, 1975] by: 1 1 H2 K2*SIN26 L 2*H*L*COSB d2(H K L) SINZB a b c a*c For the sputtering source material Ti the 50.05N144.99C“5.96’ martensite structure is monoclinic with the parameters: a-(2.902i0.01)A, b-(4.245t0.01)A, c-(4.523i0.01)A and 8-(97.5:0.5)°. The lattice parameter of B structure is 3.020A. The indexed peaks 2 for the sputtering source material are given in Table 11. Comparing Figures 12 and 13 to Figure 14, the films were found to have crystal structures similar to those of the sputtering source material; however, the intensity of each peak is different. The indexed peaks for the various films are listed in Tables 12 and 13. 4.3 Observation of In-situ Crystallization The 5000A thin film on a 3 mm diameter copper disc (as described in 3.4.1) was annealed in the TEM by using the heating stage. Figure 15 (a)-(d) shows the results. The diffraction pattern in Figure 15 (a) taken at room temperature before annealing, has 40 RELATIVE INTENSITY I." 14.-III... lush-nop- _:. Janna-m...“ _II n- '34.?- Em. Figure 12 X-ray Diffraction Pattern for an Annealed 5000A Thin Film. 20(DEGREE) 41 RELATIVE INTENSITY CD (D V N —=="_."2::..’r ===‘-‘-— ==§EE§E=EE g3 Figure 13 X-ray Diffraction Pattern for an Annealed 13800A Thin Film. 2 @(DEGREE) 42 RELATIVE INTENSITY un-unI—u-I- "AIIIIW maul-mucus.- 1.. lm I nun-—-. _. _—— ——.——_-—I—I_u——-b—‘x__n_ —_ u-I-mI-—l——e~ 1'“ —-= _-l-_——- —.-— Figure 14 X-ray Diffraction Pattern for ' Source Material. the Sputtenng 20(DEGREE) 43 Table 11 Indexed Peaks of X-ray Diffraction Patterns for the Sputtering Source Material 29 9 sine obsv. cal. (H L) 40. 20.15 .34448 .23816 .2419 (0 2)m 41. 20.65 .35266 .18625 .2041 (1 i)m 42. 21.1 .35600 .14168 .1222 (0 0)In .1354 (1 0) a 45. 22 6 .38430 .00627 .0149 (1 1) 59. 29.9 .49849 .54668 .5412 (0 2) 61. 30.6 .50904 .51461 .5100 (0 2)a 77. 38.7 .62524 .23312 .2329 (1 2) 44 Table 12 Indexed Peaks of X-ray Diffraction Patterns for the Annealed 5000A Thin Film 29 9 sine dobsv. dcal. ' (H K L) 29.3 14.65 .25291 3.0485 3.0824 (0 1 1)m 31.65 15.825 .27270 2.8273 2.8768 (1 0 0)m 41.4 20.7 .35347 2.18120 2.2041 (1 1 i)m 42.6 21.3 .36325 2.12250 2.1222 (0 2 0)m 2.1354 (1 1 0)a 44.9 22.45 .38188 2.01880 2.0149 (1 1 l)m 60.0 30.0 .50000 1.54200 1.5412 (0 2 2)m Table 13 Indexed Peaks of X-ray Diffraction Patterns for the Annealed 13800A Thin Film 29 9 sine dobsv. dcal. (H K L) 41. 20.6 .35184 2.19133 .2041 (1 i)m 42. 21.25 .36244 2.12726 .1222 (0 0) .1354 (1 0)m a 44. 22.45 .38188 2 01880 .0149 (1 1) 59. 29.8 .49697 1.55139 .5412 (0 2) 61. 30.8 .51204 1.50573 .5100 (0 2) 77. 38.8 .62660 1.23044 .23291 (1 2) 45 only one ring which verifies the as-sputtered thin film posseses amorphous structure. After annealing the thin film at 880K for 10 and 15 minutes, the diffraction patterns changed; there were more than one ring in the diffraction patterns (Figure 15 (b) and (c)) which display the film was crystallizing. Figure 15 (d) was taken at room temperature after the film was annealed at 880K for 30 minutes. The diffraction pattern has six obvious rings which presents a crystalline structure. The reciprocal d-values obtained from Figure 15 (d) were listed in Table 14. The camera length is approximate 1.6 m. The sequence of diffraction patterns shows that the films had crystallized during the annealing process. 4.4 Microstructure Determination Results 4.4.1 Microstructure of the As-sputtered Films The selected area electron diffraction pattern (SADP) in Figure 16 (b) taken from the 13800A as-sputtered film on the salt substrate shows this film had an amorphous structure. SADPs from the as-sputtered 5000A film on the salt substrate (Figure 19 (b)) and as-sputtered films (5000A and 2500A) on the copper substrates (Figure 20 (a),(b)) also show the same result. Regarding morphology, the as-sputtered films were found to possess the Zone 1 and Zone T structure (described in 2.2). Figure 16 (a) shows the smooth Zone T structure and Figure 17 (a) displays Zone 1 structure; however, Zone 1 and Zone T were also found together in both 13800 A (Figure 18) and 5000A films (Figure 19 (a)). Furthermore, the Zone 1 and Zone T structures each possess an amorphous structure [Figures 16 (b) and 17 (b)). Figure 15 A sequence of electron diffraction patterns taken from an as-sputtered 5000A thin film on Cu substrate, (a) at room temperature before annealing, (b) at 880K after annealing for 10 minutes, (c) at 880K after annealing for 15 minutes and (d) at room temperature after annealing for 30 minutes. 47 (d) Figure 15 (Continued) 48 Table 14 D-Values Obtained from Figure 15 (d) Order r (cm) d(A) (h k 1) deal. 1 1.15 3.48799 (0 1 1)m 3.08245 2 1.65 2.43102 (1 1 0)m 2.38135 3 1.75 2.29211 (1 0 1)In 2.28927 4 2.65 1.51366 (0 0 2)a 1.51461 5 3.25 1.23421 (1 1 2)a 1.23312 6 3.75 1.06965 (2 0 2) 1.06773 49 (b) Figure 16 Transmission electron micrograph (a) and corresponding electron diffraction pattern (b) taken from an as-sputtered 13800A thin film at room temperature; (a) bright field image, showing Zone T structure and (b) electron diffraction pattern, showing amorphous structure. 50 Figure 17 Transmission electron micrograph (a) and corresponding electron diffraction pattern (b) taken from the same sample as Figure 16 at room temperature; (a) bright field image, showing Zone 1 structure and (b) electron diffraction pattern, showing amorphous structure. 51 Figure 18 Transmission electron micrograph (a) and corresponding electron diffraction pattern (b) taken from the same sample as Figure 16 at room temperature; (a) bright field image, showing Zone 1 and Zone T structures found together and (b) electron diffraction pattern, showing amorphous structure. 52 Figure 19 Transmission electron micrograph (a) and corresponding electron diffraction pattern (b) taken from an as-sputtered 5000A thin film at room temperature; (a) bright field image, showing Zone 1 and Zone T structures and (b) electron diffraction pattern, showing amorphous structure. Figure 20 Electron diffraction patterns, taken at room temperature from (a) an as-sputtered 5000A thin film on Cu substrate and (b9 an as-sputtered 2500A thin film on Cu substrate. showing amorphous structure. 54 4.4.2 Microstructure of the Annealed Films Figure 21 (a) and (b) taken from 5000A and 13800A thin films annealed in the vacuum furnace show that the residual austenite grain sizes are quite small. The 13800A films have equiaxed residual austenite grains and the grain size is about 1 pm; however, the residual austenite grains in 5000A films are not all symmetrical. Some of the grains possess the shape of rectangle-like bars. Martensite plates grew in the residual austenite grains along longitudinal directions. In addition, the martensite plates could be observed more clearly in 5000A films than in 13800A films. On the other hand, in Figure 21 (a), the domains appear in the form of very thin plates with {1 l O}B habit planes in the matrix of B 2 structure. These domains are said to oscillate between the high and 2 low temperature phases [Aboelfotoh, 1978]. The SADPs taken from 5000A and 13800A [Figure 22(a), (b)] show the structure to be monoclinic associated with martensite, in agreement with previous x-ray diffractometry studies. The SADP in Figure 23 taken from the sputtering source material for comparison also displays the same result. In addition to the martensite structure, some residual 82 structure was observed, which was anticipated from the x-ray diffraction studies. SADPs shown in Figs. 24 (a), 25 (a) and 26, taken from a 13800A film, display the reflections from 82 structure. However, extra "1/2", "1/4" and "1/3" reflections were apparent in the SADPs. "1/3" reflections have been associated with 55 Figure 21 Transmission electron micrographs, taken at room temperature from (a) an annealed 5000A thin film and (b) an annealed 13800A thin film, showing the polycrystalline structure after annealing the films at 930K for 30 minutes. 56 (b) Figure 22 Electron diffraction patterns, taken at room temperature from (a) an annealed 5000A thin film, [lll]m zone, and (b) an annealed 13800A thin film, [1-12] zone, showing monoclinic structure associated with marten51te. 57 Figure 23 Electron diffraction pattern [001]m zone, taken at room temperature from the sputtering source material, showing martensite structure. 58 v1 100 nm (b) Figure 24 Electron diffraction pattern (a) and transmission electron micrograph (b), taken from an annealed 13800A thin film at room temperature; (a) diffraction pattern [100] zone, showing 1/2 , 1/4 and 1/3 reflections, and (b) dark field image, formed by one of 1/3 reflections encircled in (a), showing antiphase boundaries. 59 (a) Figure 25 Electron diffraction pattern (a) and transmission electron micrographs (b) and (c), taken at room temperature from the same sample as Figure 24; (a) diffraction pattern, [111] zone, showing 1/2 and 1/3 reflections, (b) dark field image, formed by one of 1/3 reflections encircled (A area) in (a), showing antiphase boundaries, and (c) dark field image fromed by one of 1/2 reflections encircled (B area) in (a), showing oxide precipitates which not observed in (b). 60 Figure 25 (Continued) 61 Figure 26 Electron diffraction pattern [100]B zone, taken at room temperature from an annealed 5000A thin film. showing 1/2 and 1/3 reflections. 62 the "premartensitic" phase and with rhombohedral structure. This phenomenon substantiates that Ni-Ti-Cu thin films produced in the present work transformed the 82 structure to rhombohedral structure and then to monoclinic structure in the same way as bulk alloys do. The anti-phase boundaries were also observed in the dark field images [Figs. 24 (b) and 25 (b)], but were not very obvious. The extra "1/2" and "1/4" reflections were recently suggested to result from interstitial ordering of hydrogen or oxygen, and not from premartensitic structure (Wu, 1988a). Wu et a1. explained that hydrogen atoms may result from the jet-polishing process, and the oxygen atoms come from Ti raw materials as well as the arc-melting process. In the present study, these extra 1/2 and 1/4 reflections were observed in the annealed films and sputtering source materials. Neither the thin films nor the sputtering source material were jet- polished, so the extra reflections should not have been caused by the ordering of hydrogen, but are most likely from the ordering oxygen due to the reasons mentioned above. The dark field image from 1/2 reflections [Figure 25 (c)] shows precipitates in the matrix. These bright precipitates were not displayed in the dark field image taken from 1/3 reflections [Figure 25 (b)]. It is speculated that they are suboxides. Figure 26 exhibits a SADP of [l O 0] zone taken from an annealed 5000A film. This SADP contains 1/2 and 1/3 reflections; however, 1/3 reflections are faint and 1/4 reflections do not appear. The ESCA spectrum (Figure 27) demonstrates the oxygen peaks for sputtering material. According to quantitative analysis of ESCA, the sputtering source material includes 15at% oxygen. For the as- 63 >0 .5529 95!: 9.9 9.89 9.8~ 9.999 9.9? 9.999 9.999 9.99N . 9.199 . 9 ”999 . 9 9999 V. p u u u u n u q _ . a . u . I .. amaz Aavso : -. are -- -- amaa : -. A9 .>9zuzu 92H92H9 a... as: 9.9% 9.89 9.8.. 9.89 9.89 35 9.89 9.89 9.8: x . u u u u i u u n u i i l 4 u i n 3. u a .- an; m: “83 :9 -- .. ~ H a... a: H m -. as .. -. 2:: a: a; .. v .- -. m .. -- 9 -- 1. -- .1 N .- a; as .. a .. .. a u x n x x u 1 u 4 i n i u i i u x i a a x 999 9: >0 9m9.9~uu>9mmzu 9999 mxo x 99~.~ .999.—m upu9uu9 .99p999 99999 “as: 5:34 Ex 3:; Huang c_c 99.9uuzmp 999 909 9v "94929 99xv~xv rm>u=9 999m ‘. 3/(3)N Figure 28 The ESCA spectrum of an as-sputtered 13800A thin The 0 peaks not so film, showing Ti, Ni, Cu, and 0 peaks. obvious as ' Figure 27. In those 66 (b) Figure 29 Electron diffraction patterns, taken from an annealed 13800A thin film at room temperature; (a) [112] zone and (b) [110]B2 zone, showing 1/2 and 1/3 reflections. 67 (b) Figure 30 Electron diffraction patterns, taken from the sputtering source material at room temperature; (a) [100]B2 zone, (b) [110] zone (c) [112]B and (d) [111] zone, showing 1/2 reflections but 1/3 reflections only seen in (d). 68 Figure 30 (Continued) 69 transformed to the high temperature structure when the beam was converged, as shown in Figure 31 (a). The electron beam then was spread, decreasing the temperature of the local volumn. The martensite plates were observed to grow back, as displayed in Figure 31 (b). This figure demonstrates that the martensite plates grew in the matrix (high temperature structure) as the temperature decreased. Furthermore, SADPs taken at different temperatures show that the thermoelastic effect occurred in the films. Figure 37 verifies that the differences of SADPs correspond to various temperatures: in Figure 32(a), 1/3 and 1/2 reflections were observed. When the temperature was elevated by converging the electron beam, the 1/3 and 1/2 reflections in Figure 37(a) disappeared, as shown in Figure 32 (b). Only the reflections from 82 structure were viewed, which shows that the rhombohedral structure transformed to 82 structure when the temperature was raised. When the temperature was diminished by spreading the electron beam, the extra 1/3 and 1/2 reflections were visible again, as displayed in Figure 32 (c) which verifies that 82 structure transformed back to rhombohedral structure when the temperature was reduced. These variations in both SADPs and bright field images substantiate that phase transformation did take place in the thin films. 70 (b) Figure 31 Transmission electron microscophs, taken from an annealed 5000A thin film at different temperatures by controlling the electron beam; (a) bright field image,taken as beam converged and (b) bright field image, same place as (a), taken as beam spread, showing the martensite plate growing in the high temperature structure in (a) during the cooling process. 71 (b) Figure 32 A sequence of diffraction patterns, taken from an annealed 13800A thin film; (a) at room temperature, (b) as the electron beam converged, and (c) as electron beam spread, showing 1/3 reflections in (a) disappeared in (b) when temperature raised, and appeared in (c) as temperature reduced. 72 (C) Figure 32 (Continued) 5. DISCUSSION 5.1 Compositions of the Films From the results of the composition analysis, we acquired ambiguous results. For the 5000A thin films, titanium was reduced from 50.05 at% to 47.17 at%. Copper reduced from 4.96 at% to 3.16 at%, but nickel increased from 44.99 at% to 49.70 at%. However, for the 13800A thin films, titanium also decreased from 50.05 at% to 47.02 at%, but copper increased from 5 at% to 5.41 at%. Nickel increased from 44.99 at% to 47.57 at%. Even if the steady state is established in the sputtering system, the films still would not have the same composition as the sputtering source due to a number of factors decribed in 2.1. Bias of the substrate is probably the most significant factor. When the bias is zero or positive in the films, elements with higher sputtering rates are enriched, compared with elements with lower sputtering rates [Tarng, 1971]. Tarng et a1. explained that if a negative bias voltage was more than the discharge enclosure, more material would be removed from the substrate than deposited, due to the resputtering effect, and the elements with higher sputtering rate would be less rich on the films. If the substrate has the same bias as the discharge enclosure, the film would have the same composition as the sputtering source. Therefore, if the substrate was biased with a positive potential or grounded (which means the potential of the substrate is positive compared with the negative discharge enclosure), elements with higher sputtering rates would be richer on 73 74 the films. In this study, the substrates were placed in the grounded disk plate. Copper has the highest sputtering rate, followed by nickel and titanium, which has the lowest sputtering rate (Table l). Supposedly, in the films, a higher amount of copper should be retained than in the sputtering source. The opposite relationship should be found concerning titanium; the films should contain a lower amount of titanium than the sputtering source. This corresponds with the composition result for the 13800A thin film that the amounts of copper and nickel are higher in this film than in the sputtering source. But why copper is diminished in the 5000A film is still unresolved. In fact, although substrates were grounded, the surfaces of salt substrates had different potentials because salt is an insulating material. This would lead to nonuniform bombardment and result in a compositional shift. Therefore, usually, a negative bias was applied to the substrates for obtaining films with desired compositions, but the bias value needs to be obtained empirically. Moreover, 5000A thin films were produced first among these sputtered films. The sputtering source might not be the steady state. The factor of presputtering time for establishing the steady state in this case also needs to be acquired empirically. The applied sputtering source voltage and pressure of argon gas for 5000A are slightly different compared with 13800A films. These factors may influence the compositions of the films. Anyway, in order to understand the relationship between compositions and variables, more study is needed. 75 5.2 Structures of As-sputtered Films According to the x-ray diffraction patterns and SADPs, the as- sputtered films on both Cu and KCl substrates possess an amorphous structure. This substantiates that the surface diffusion is very slow at substrate temperatures achieved in the sputtering system. For most pure metals, the surface mobility is large enough to form crystalline states, even at the temperature of liquid helium. However, alloys and compounds often need more thermal energy in order to increase the mobilities of atoms for growing crystalline structures. For tenary alloys, it is reasonable to suppose that three different kinds of atoms require more energy than one or two would for forming an ordered phase. The result that the Ni-Ti-Cu thin films growing on the substrates at about 310K possess an amorphous structure is not surprising. 5.3 Morphology of As-sputtered Films The Zone T morphology was observed by TEM for the as-sputtered thin films. This result corresponds with the structure zone model [Thornton, 1974]. The ratio of the substrate temperature and melting temperature (T/Tm) is approximately 0.2 and the argon pressure was set at 3.3 pPa so the Zone T structure was expected (Figure 1). Still, the Zone 1 structure was found, and an array of the Zone 1 structure extruded from the Zone T background, which was also reported by Spalvins et a1. (1977). The Zone 1 structure may result from rough substrates, or from being ion bombarded in the sputtering system (Thornton, 1978). Furthermore, the boundary between Zone 1 76 and Zone T in the zone model is not easily identified, so finding Zone 1 together with Zone T is understandable. 5.4 Rhombohedral Structure Coincidently, nickel is rich in the as-sputtered films. According to Miyazaki et al.(1986), the rhombohedral structure associated with the premartensitic phase should be more easily observed due to the increased nickel content. SADPs taken from the annealed thin films and sputtering source material verify this point: 1/3 reflections in the [l 0 0] zone, [1 1 0] zone, and [1 1 2] zone are more visible in the films than in the sputtering source. However, results of the x-ray diffractometer for the thin films still do not display peaks manifested either from 1/3 reflections or 1/2 reflections. These peaks may be buried in the background noise. 5.5 Morphology of Annealed Films The TEM photographs demonstrate that the average residual austenite grain size of 13800A thin films is approximately 1 pm. This value is similar to the thickness of the films, which means that the grains equiaxed in three dimensions. However, the 5000A films do not contain equiaxed grains, the lack of which might be caused by abnormal grain growth, or low angle grain boundary adjustment. This phenomenon can not be explained in the present work, and needs more study. Moreover, the idea that thinner films with less atoms piling up need less stress in order to form the martensite can explain why martensite plates were more visible in the annealed 5000A films than in the annealed 13800A films. 77 5.6 Interstitial Ordering of Oxygen The extra 1/2 and 1/4 reflections possibly result from interstitial ordering of impurities. In this work, hydrogen may not be the dominant impurity because ion milling was used instead of jet-polishing. Oxygen would be the most likely interstitial impurity in the thin films, which may result from the sputtering source, or may occur as a contaminant during the annealing process. The dark field images of 1/2 reflections taken from the sputtering source material show minute precipitates with sizes from about 100A to 300A. These precipitates were speculated to be suboxide in Ni-Ti-Cu alloys, a conclusion supported by previous work [Wu, 1988a, 1988b]. Furthermore, although the sputtering source material has 15 at% oxygen and the transformation temperature might be affected, the martensite transformation still can occur. For the annealed thin films, the transformation temperatures might have been decreased since oxygen atoms, occupying the octhedal sites in the 32 structure, would be a barrier to phase transformation. The high temperature structure now involves one more element, oxygen atoms, in addition to Ni, Ti and Cu atoms. Therefore, more stress is needed for distorting this high temperature structure (with oxygen) into the martensite structure than the stress needed for distorting the high temperature phase without oxygen atoms. The MS temperature may be reduced for this reason. In order for maintaining the desired transformation temperatures, the problem of oxidation must be resolved. Obviously, improving the furnace vacuum can diminish oxidation. Raising the substrate temperature of the sputtering 78 system, in order to increase the atomic mobility on the surface of thin films, may be considered so that the crystalline structure can grow directly on the substrates. The high temperature annealing procedure could then be skipped, and oxidation of the films avoided. However, the temperature of the substrate must not be so high as to cause the compositions of the films to change due to diffusion of substrate materials during the sputtering process. Another suggestion is to replace the copper substrates with the elements or alloys, such as iron, titanium or nickel, materials with low diffusion rates. The benefits are, first, that the compositions would not be varied during the heat treatment in the vacuum furnace and second, the higher temperature of the substrates in the sputtering system would not affect the compositions of the films. 5.7 Transformation Temperature Figures 31 and 32 exhibit the changes in the images and SADPs by using electron beams to alter the temperature at the local regions. This can verify that transformation temperatures of the thin films are close to room temperature, since we observe rhombohedral phase and martensite phase both in the annealed films at room temperature. MS temperature should be higher than the room temperature and Mf lower than room temperature. These transformation temperatures of the thin films are close to those of the sputtering source material. Although the compositions of the as-sputtered films are different than the sputtering source, the transformation temperatures would not vary too much due to the characteristic of Ni-Ti-Cu alloys [Mercier, 1979]. Because the amount of nickel is greater in 79 the films than in the sputtering source material, it is no surprise that Ms and M of the thin films would be slightly reduced. This f assumption can be verified by differential scanning calorimetry. (1) (2) (3) (4) (5) (6) (7) (8) 6. CONCLUSIONS The compositions of Ni-Ti-Cu thin films produced by a DC sputtering system are slightly different from the sputtering source material: Ti The composition of 50.05Nl44.99cu4.96' the 5000A thin film was Ti and that 47.14N149.70C”3.16’ of the 13800A thin film was Ti 47.02Ni47.47C“5.41° The as-sputtered thin films have an amorphous structure under the conditions set for the sputtering system. Zone 1 and zone T film morphologies were identified in the films. Crystallization of the amorphous film was observed at 880K. After annealing, martensite with monoclinic structure was found in the films at room temperature. The rhombohedral phase was also found in the annealed films at room temperature. From (5) and (6), we speculate that Ms > room temperature > Mf. The thermoelastic effect was confirmed to take place in N1— Ti-Cu thin films by the SADPs and images from the TEM. Rhombohedral structure was observed to transform to B 2 structure at high temperature, and B2 structure transformed back to rhombohedral structure at low temperature; martensite plates disappeared when temperature was elevated, and grew back when temperature was reduced. 80 7. REFERENCES M.O. Aboelfotoh, H.A. Aboelfotoh, and J. Washburn, "Observations of Pretransformation Lattice Instability in Near-Equiatomic NiTi Alloy", Journal of Applied Physics, 49 (1978), 5230. R.H. Bricknell, K.N. Melton and O. Mercier, "The Structure of NiTiCu Shape Memory Alloys", Metallurgical Transactions A, 10A (1979), 693. W.J. Buehler and F.E. Wang, " A Summary of Recent Research on the Nitinol Alloys and Their Potential Application in Ocean Engineering", Ocean Engineering, 1 (1968), 105. W. Buhrer, R. Gotthardt, A. Kulik, and O. Mercier, "Neutron Diffraction Measurements in Equiatomic NiTi Alloys", Journal de Physique, Colloque C4 (1982), 219. K. Chandra and G.R. Purdy, "Observations of Thin Crystals of TiNi in Premartensitic States", Journal of Appled Physics, 39(1968), 2176. I J.J. Cuomo and R.J. Gambino, "Influence of Sputtering Parameters on the Composition of Multicomponent Films", Journal of Vacuum Science and Technology, 12 (1975), 79. J.W. Edington, "Electron Diffraction in the Electron Microscope", in Practical Electron Microscopy, MacMillian Press, New York, Volumn 2 (1975), 80 D.S. Grummon and J.W. Jones, "Fatigue Damage Accumulation in Nickel Modified by Ion Beam Surface Microalloying", Metallurgica Transaction A, 19 (1988), 2775. P.K. Haff and Z.E. Switkowski, "On the Sputtering of Binary Compounds", Applied Physics Letters, 29 (1976), 549. R.F. Hehemann and G.D. Sandrock, "Relations Between the Premartensitic Instability and the Martensite Strucutre in TiNi", Acta Metallugica, 5 (1971), 801. C. M. Hwang, M. Meichle, M. B. SalamonN .and C. M. Wayman, "Transformation Behaviour of a Ti Fe Alloy. I. Premartensitic Phenomena and Incommensurate PhaseQN iPhilosophical Magazine A, 7(1983),9 ‘/C.M. Hwang and C.M. Wayman, "Diffuse Electron Scattering from an Incommensurate Phase in a Ti Ni Al Alloy", Acta Metallugica, 32 (1984), 183.58’7 37.5 3'8 J.J. Kim, P. Moine, and D.A. Stevenson, "Crystallization Behavior of Amorphous Ni-Ti Alloys Prepared by Sputter Deposition", Scripta Metallurgica, 20 (1986), 243. 81 82 B. Liu, W.L. Johnson and M.A. Nicolet, "Amorphous Film Formation by Ion Beam Mixing in Binary Metal Systems", Nuclear Instruments and Methods, 209/210 (1983), 229. K.N. Melton and O. Mercier, "Fatigue of NiTi And CuZnAl Shape Memory Alloys", in Strength of Metals and Alloys, P. Haasen et al., eds., Vol.2 (1979), 911. V/fo. Mercier and K. Melton, "The Substitution of Cu for Ni in NiTi Shape Memory Alloys", Metallurgical Transactions A, 10A (1979), 387. V/ R. Messier and R.C. Ross, "Evolution of Microstructure in Amorphous Hydrogenated Silicon", Journal of Applied Physics, 53 (1982), 6220. C. Meyers and D.S. Grummon, "A Finite Element Analysis of the Interaction of Persistent Slip Bands with Hard Surface Films During Fatigue", TMS Fall Meeting, Chicago, Sept. 1988)]. C.M. Michal and R. Sinclair, "The Structure of TiNi Martensite", Acta Crystallographica, B37 (1981), 1803. \/'S. Miyazaki and K. Otsuka, "Deformation and Transition Behavior Associated with the R-Phase in Ti-Ni Alloys", Metallurgical Trasactions A, 17A (1986), 53. B.A. Movchan and A.V. Demchishin, "Study of the Structure and Properties of Thick Vacuum Condensates of Nickel, Titanium, Tungsten, Aluminium Oxide and Zirconium Dioxide", Physics of Metals and Metallography, 28 (1969), 83. K. Otsuka, T. Sawawura, and K. Shimizu, "Crystal Structure and Internal Defectss of Equiatomic TiNi Martensite", Physica Status Solidi (a), 5 (1971), 457. G.R. Purdy and J.C. Parr, Transactions, " A Study of the Titanium- Nickel System Between TiZNi and TiNi", The Metallurgy Society AIME, 221 (1961), 636. A.K. Rai and R.S. Bhattacharya, "Sputter-Deposited and Ion-Mixed Ni- Ti and Ni-Cr Thin Films", Materials Science and Engineering, 85 (1987), 139. ‘v' R.C. Ross and R. Messier, "Microstructure and Properties of Rf- Sputtered Amorphous Hydrogenated Silicon Films", Journal of Applied Physics, 52 (1981), 5329. (/ R.C. Ross and R. Messier, "Reactive Sputtering of Amorphous Silicon in Ne, Ar, and Kr", Journal of Applied Physics, 54 (1983), 5744. T. Saburi and S. Nenno, "In Situ Observations of the Nucleation 83 and Growth of the Thermoelastic Martensite in a Ti-Ni-Cu Alloy", Proceedings of The International Conference on Martensitic Transformations (1986), 671. )// P. Sigmund, "Sputtering of Single and Multiple Component Metals", Journal of Vacuum Science and Technology, 17 (1980), 396. R. Sinclair and H.A. Mohamed, "Lattice Imaging Study of a Martensite-Austenite Interface", Acta Metallurgica, 26 (1978), 623. T. Spalvins and W.A. Brainard, "Nodular Growth in Thick-Sputtered Metallic Coatings", Journal of Vacuum Science and Technology, 11 (1974), 1186 - T. Tadaki and C. M. Wayman, "Electron NMicroscopy Studies of Martensitic Transformations in Ti Cu Alloys. Part I. Compositional Dependence of One- Tao rd 5Reflections from the Matrix Phase", Metallography, 15 (1982), 233. . . arn an . . e ner, A o putterin tu 1es wit in l/ M L T g d C K W h " 11 y S ' g S d' ' h ' situ Auger Electron Spectropy", Journal of Applied Physics, 42 (1971), 2449. l" J.A. Thornton, "Influence of Apparatus Geometry and Deposition Conditions on the Structure and Topography of Thick Sputtered Coatings", Journal of Vacuum Science and Technology, 11 (1974), 666. L» J.A. Thornton and A.S. Penfold, "Cylindrical Magnetron Sputtering", in Thin Film Processes", J.L. Vossen and W. Kern, eds., Academic Press (New York), 1978, 75 p~ J.L. Vossen and J.J. Cuomo, "Glow Discharge Sputter Deposition", in Thin Film Process, J.L. Vossen and W. Kern, eds., Academic Press, New York, 1978, 12. H.F. Winters, D.L. Raimondi and D.E. Horne, "Proposed Model for the Composition of Sputtered Multicomponent Thin Films", Journal of Applied Physics, 40 (1969), 2996. S.K. Wu and C.M. Wayman, "Interstitial ordering of Hydrogen and Oxygen in TiNi Alloys", Acta Metallurgica, 36 (1988a), 1005. S.K. Wu, A.G. Khachaturyan, and C.M. Wayman, "Superstructure of Intersitial Ordering of Oxygen in TiNi Alloys", Acta Metallurgica, 36 (1988b), 2065.