Transaction B: Mechanical EngineeringVol. 16, No. 3, pp. 248{252c Sharif University of Technology, June 2009

Characterization of Sputtered NiTiShape Memory Alloy Thin Films

S. Sanjabi1;�, M. Naderi1, H. Zare Bidaki1 and S.K. Sadrnezhaad2

Abstract. During recent years, many investigations have been carried out to determine how to selectdi�erent materials for the making of Micro Electro Mechanical Systems (MEMS) and bio-MEMS. TheNiTi shape memory alloy thin �lm has been much regarded as a promising candidate for MEMS due toits unique shape memory e�ect and high energy output. In this research, NiTi thin �lm was fabricatedusing a sputtering technique from separate elemental Ni and Ti targets. The characterizations of thedeposited �lms were investigated using di�erent analysis techniques, such as Field Emission SEM, DSC,XRD, electrical resistivity measurement and nanoindentation.

Keywords: NiTi thin �lm; Sputtering; Characterization; Shape memory e�ect.

INTRODUCTION

Shape memory is the unique property of some al-loys, which returns to its original shape with heating.Shape-Memory Alloys (SMAs) possess an array ofdesirable properties: High power to weight (or forceto volume) ratio (thus having the ability to recoverlarge transformation stresses and strains upon heatingand cooling), peudoelasticity (or superelasticity); ahigh damping capacity and good chemical resistanceand biocompatibility [1,2]. More recently, thin �lmSMA has been recognized as a promising and highperformance material in the �eld of Micro-Electro-Mechanical Systems (MEMS) since it can be patternedwith standard lithography techniques and fabricated ina batch process [3].

Thin �lm SMA has only a small amount ofthermal mass to heat or cool, thus the cycle (re-sponse) time can be reduced substantially and thespeed of operation may be increased signi�cantly. Theapplication of SMA �lms in MEMS also facilitatesthe simpli�cation of mechanisms with a exibility ofdesign and the creation of clean, friction free and non-

1. Department of Materials Science and Engineering, TarbiatModares University, Tehran, P.O. Box: 14115-143, Iran.

2. Department of Materials Science and Engineering, SharifUniversity of Technology, Tehran, P.O. Box 11155-9466,Iran.

*. Corresponding author. E-mail: sanjabi@modares.ac.ir

Received 12 May 2007; received in revised form 26 February2008; accepted 27 May 2008

vibration movements. MEMS-based micropumps andmicrovalves are attractive for many applications, suchas implantable drug delivery, chemical analysis andanalytical instruments etc. [4]. There are di�erentdesigns for TiNi �lm based micropumps or microvalvesand most of them use TiNi. Membrane graspingand manipulating small or micro-objects with highaccuracy is required for a wide range of importantapplications, such as the assembly in microsystems,endoscopes for microsurgery and drug injection micro-manipulators for cells.

NiTi alloys are sensitive to the composition ratioof Ti and Ni: only 1% variation of composition cancause a 100�C shift in transformation temperatures [5].From a practical point of view, only sputter depositionhas succeeded so far and a perfect shape memory e�ectsimilar to bulk materials can be obtained [6].

In sputtering, an alloy target is generally used andthe compositions of the sputtered �lms are always Ni-rich in comparison with the target (the �lms are Tipoor with respect to the target (by 2-4at%)), becausethe sputtering yield for Ni is higher than Ti [7]. Thesimplest and most common solution to this problemis to place small pieces of pure Ti onto the target toachieve the correct �lm stochiometry [8]. Compositionadjustment using Ti (or Ni) plates requires the controlof numerous parameters, such as number, geometry,size and position, and any subsequent change of thedeposition parameters will require re-adjustment. Thistechnique leads to high impurity levels, due to the largesurface contribution.

Sputtered NiTi Shape Memory Alloy Thin Films 249

The control of the �lm composition from separatetargets controlled by the power ratio is more exibleand can overcome these problems. The control ofcomposition uniformity is important, requiring appro-priate deposition geometry (e.g. inclining the axes ofthe targets [9], optimizing the substrate-target distanceand the size and shape of targets). In this study, shapememory NiTi �lms were fabricated by a simultaneoussputter deposition from separate elemental Ni and Titargets (the mixing of elements before arrival at thesubstrate) and annealing at a relatively low temper-ature. The deposited �lms were characterized withrespect to structure, transformation temperature andmechanical property.

EXPERIMENT DETAILS

Films were deposited by ultra high vacuum DC mag-netron sputtering onto an unheated Si (1 0 0) substrateof dimension 10 mm � 5 mm. The deposition systemallowed two magnetrons (target size: 35 mm � 55 mm),plus associated heater leads, instrumentation feed-throughs, viewing port and rotary shaft, to be placedon one standard 200 mm O.D. ange, inserted into a150 mm I.D. nitrogen-cooled can. A constant ow of Ar(99.999%) was controlled with a leak valve during �lmdeposition and di�erent sputtering gas pressures wereachieved by throttling the gate valve. The thicknessof the �lm was measured by surface pro�lometry(Talysurf 6, Taylor-Hobson) using the step height ona masked silicon substrate: the deposition rate wascalculated to be about 1 Am/h and the thickness ofthe �lms was around 2 �m.

as-deposited �lms were annealed at 500�C for 1 hin a vacuum furnace (base pressure < 10�5 Pa) withheating and cooling rates of approximately 50�C/min.The composition of as-deposited �lms was determinedby energy dispersive X-ray spectroscopy, using a JEOLJSM-5800LV operating at 15 keV. Films were examinedby observing fracture cross-sections in a Field EmissionScanning Electron Microscope (FESEM) using a JEOLJSM-6340F. A Siemens D500, with a Cu-K�(� =1:54056 A�) X-ray source at 40 kV and 40 mA wasused for XRD as a function of temperature, and DSC(Q1000, TA instruments with the minimum requiredmass = 0.5 mg) was used to indicate the transfor-mation temperatures at heating and cooling rates of10�C/min. A standard four-point electrical resistivitymeasurement also characterized the transformationtemperatures. Nanoindentation was performed using aNanotest 600 (Micro-Materials Ltd., UK). A diamondindenter (tip radius of 10 micron) was used for sphericalindentation, loading and unloading at 0.1 mN/s, givingindentation depths ranging from 20 to 500 nm. Bothload and displacement were recorded during the entireloading and unloading cycle.

RESULTS AND DISCUSSION

Without substrate rotation, suitable powers were ap-plied to each target to give similar deposition rates(of Ni and Ti) at the center of the substrate support.Figure 1 shows the resulting composition variationacross the substrate support (the power ratio of Ti toNi was around 3). The composition uniformity wasa�ected by the deposition geometry. For example,a small deviation of the substrate holder from thehorizontal plane led to around 1% variation across the100 mm substrate holder diameter (0.01%/mm).

Important parameters which a�ect the qualityof the �lms are target power (and hence the deposi-tion rate), Ar gas pressure, substrate-target distance,substrate temperature, purity of material targets andthe deposition environment. The Ar pressure andthe target-substrate distance a�ect the energies of thedepositing species, and thus �lm density, structuralintegrity and stress. At high Ar pressures, �lms oftenshow low density containing structural defects, whileat low pressures the structure is denser with fewerdefects [10]. Films spanning a range of compositions,prepared at high Ar pressures (> 1 Pa), had a brittlestructure and cracks were observed on the �lm surface;Figure 2a shows the extensive delamination at theNiTi/Si interface and cracks caused by in-plane, biaxialtensile stresses. Films covering a similar range ofcomposition, prepared at low Ar pressures, exhibited asmooth, featureless surface structure; Figure 2b showsthe cross-section of a �lm deposited at 0.6 Pa.

XRD of as-deposited �lms showed only a broadpeak around 2� = 42� and no crystalline peaks,suggesting an amorphous structure (see Figure 3).After annealing, in spite of similar composition, �lmsdeposited at two di�erent gas pressures showed verydi�erent �lm structures. The XRD spectrum of a51.2% Ti and �lm deposited at 1.2 Pa (see Figure 3a),

Figure 1. The composition variation across the substrateholder.

250 S. Sanjabi, M. Naderi, H. Zare Bidaki and S.K. Sadrnezhaad

Figure 2. Scanning electron micrographs of �lmsdeposited at di�erent Ar gas pressures: (a) plan view of a�lm deposited at 1.2 Pa, showing severe surfacedelamination, and (b) FESEM cross-section of a �lmdeposited at 0.6 Pa.

Figure 3. Room temperature XRD pro�les of �lmsdeposited at di�erent Argon pressures (annealed at500�C): (a) 1.2 Pa with composition of Ti51:2Ni48:8 and(b) 0.6 Pa with composition of Ti50:6Ni49:4. XRD of anas-deposited �lm is also shown.

showed only an austenite phase (peaks at 42.7�and 62�,corresponding to 110 and 200 austenite re ections) atroom temperature.

The shape memory e�ect of this �lm was char-acterized by electrical resistivity, as a function oftemperature (see Figure 4a), con�rming that thetransformation temperature of this �lm is below roomtemperature. The loop of the shape memory duringcooling and heating is not clear and one peak can beobserved around 40�C. Following the deposition of asimilar �lm composition at lower Ar pressure (0.6 Pa),the annealed �lm showed martensite peaks at roomtemperature. The existence of the martensite peaksat room temperature indicates that the shape memorye�ect and phase transformation occur above roomtemperature, as con�rmed by the electrical resistivitymeasurement shown in Figure 4b.

These results illustrate that in spite of controllingthe �lm composition, a poor shape memory e�ect canbe obtained.

Thin �lms with di�erent compositions were fab-ricated at the Ar gas pressure of 0.6 Pa and then

Figure 4. Electrical resistivity measurement as a functionof temperature for �lms deposited at di�erent Ar gaspressures and annealed at 500�C: (a) 1.2 Pa withcomposition of Ti51:2Ni48:8 and (b) 0.6 Pa withcomposition of Ti50:6Ni49:4.

Sputtered NiTi Shape Memory Alloy Thin Films 251

Figure 5. XRD trace of near-equiatomic (Ti50:6Ni49:4)(a) room temperature, and (b) 100�C (see text forindexing of peaks).

annealed at 500�C for 1 h. Figure 5 shows theXRD traces of annealed near equiatomic �lms at roomtemperature and 100�C. For an equiatomic NiTi �lm atroom temperature, di�raction patterns showed peaksat 2� = (38.2�, 39.1�, 41.3�, 44.05�, 45.1�, 60.2

�)

indexed respectively as 110, 002, 111, 020, 012 and 022planes of the B190 (monoclinic) martensite structure(Figure 5a). During heating to 100�C, the martensitepeaks gradually disappeared and were replaced by asharp peak at 2� = 42.7�and a minor peak at 2�= 62�(Figure 5b) which corresponds to austenite 110and 200.

Annealed NiTi thin �lms undergo a martensiticphase transformation when cooled below their transfor-mation temperature. The transformation temperaturesof near equiatomic and Ti-rich �lms are above roomtemperature, while that of the Ni-rich �lm is lower.In Figure 6, which shows the DSC results of the nearequiatomic NiTi �lm, a one-stage transformation isobserved during heating, corresponding to B190 to

Figure 6. DSC trace of near-equiatomic (Ti50:6Ni49:4).

B2 transformation. The transformation enthalpy of22.3 J/g con�rms that this peak is related to a directtransformation from martensite to austenite phases;the transformation starts from 71�C and �nishes at86�C. During cooling, a two-stage transformation isobserved, corresponding to a transformation betweenB2, R-phase and B19' phases. The austenite phasestarts to transform to R-phase at around 60�C. TheR-phase starts to change to the martensite phase at atemperature around 50�C and �nishes at 30�C. Thisevaluation indicates that the martensite structure isdominant at room temperature and austenite at highertemperature (around 100�C) as indicated by XRD(Figure 5).

Figure 7 shows the indenter load versus theindentation depth as a function of the �lm compo-sition (all indentation tests were conducted at roomtemperature). For the Ni-rich �lm, high elastic re-covery was seen in the unloading curve, indicating thepredominance of elastic or psuedo-elastic deformation.For near-equiatomic (Figure 7) there is much lessevidence of such elastic recovery in the unloadingcurve, indicating predominantly plastic deformation.Interpreting the indentation response is a complex taskdue to dependence on the test temperature, strainlevel, applied stress and type of indenter. Recentreports suggest that a spherical indenter is optimumfor understanding the shape memory e�ect and pseudo-elasticity. Our other measurements show that bothTi-rich and near-equiatomic �lms have the martensitestructure at room temperature, while the Ni-rich �lmis austenitic. The mechanism of deformation underindentation has been discussed for both martensite andaustenite bulk material structures; for austenite struc-tures, both martensite formation and detwinning can

Figure 7. Load vs. displacement curve fornear-equiatomic (Ti50:6Ni49:4).

252 S. Sanjabi, M. Naderi, H. Zare Bidaki and S.K. Sadrnezhaad

occur, while in the martensitic phase only detwinningtakes place. Martensite formation in the austeniticphase is metastable, and thus reverses on unloading,leading to recoverable strain, while detwinning resultsin permanent deformation unless a subsequent heatingstep is included.

CONCLUSIONS

The fabrication of shape memory TiNi thin �lmsusing simultaneous sputter deposition from separateelemental targets was demonstrated. The in uenceof the sputtering parameters on the quality of the�lm and shape memory e�ect has been illustrated; inspite of composition control, at higher Ar gas pressurethe transformation temperature can occur below roomtemperature and the shape memory hysteresis loopduring heating and cooling is not clear. Followingdeposition at low gas pressure, phase transformationsshowed a qualitatively similar behavior as a function of�lm composition to Ni-Ti bulk materials. Simultaneoussputter deposition from two targets o�ers exibility incontrol of the required composition and it is simpleto add another target for the fabrication of ternaryshape memory thin �lms. Successful implementation ofTiNi micro-actuators requires a good understanding ofthe relationship between the processing, microstructureand properties of TiNi �lms.

REFERENCES

1. Otsuka, K. and Ren, X. \Recent developments in theresearch of shape-memory alloys", Intermetallics, 7,pp. 511-528 (1999).

2. Humbeeck, J.V. \Non-medical applications of shape-

memory alloys", Mater. Sci. Eng. A273-A275, pp. 134-148 (1999).

3. Krulevitch, P., Lee, A.P., Ramsey, P.B., Trevino, J.C.,Hamilton, J. and Northrup, M.A.M.A. \Thin �lmshape-memory alloy micro-actuators", J. MEMS, 5,pp. 270-282 (1996).

4. Reynaerts, D., Peirs, J. and Van Brussel, H. \An im-plantable drug-delivery system based on shape mem-ory alloy micro-actuation", Sensors & Actuators A, 61,pp. 455-462 (1997).

5. Ikuta, K., Fujishiro, H., Hayashi, M. and Matsuura,T. \Laser ablation of Ni-Ti shape memory alloy thin�lm", Proceedings of the First International Confer-ence on Shape Memory and Superelastic Technologies,Asilomar Conference Center, Paci�c Grove, California,USA, p. 13 (1994).

6. Miyazaki, S. and Ishida, A. \Martensitic transforma-tion and shape memory behavior in sputter-depositedTiNi-base thin �lms", Mater. Sci. Eng., A Struct.Mater.: Prop. Microstruct. Process., 106, pp. 273-275(1999).

7. Grummon, D.S., Hou, L., Zhao, Z. and Pence,T.J. \Progress on sputter-deposited thermotractivetitanium-nickel �lms", J. de. Phys. IV, 5, p. 665(1995).

8. Yamamoto, T. and Miyazaki, S. \Fabrication of TiNi-based shape memory alloy thin �lms by simultaneousmulti-target sputtering method", J. de Phys. IV, 112,p. 869 (2003).

9. Ishida, A., Takei, A. and Miyazaki, S. \Shape memorythin �lm of Ti-Ni formed by sputtering", Thin SolidFilms, 228, p. 210 (1993).

10. Thornton, J.A. \The microstructure of sputter-deposited coatings", J. Vac. Sci. Technol., A, Vac.Surf. Films, 4(6), p. 3059 (1986).