Scripta Materialia 50 (2004) 819823 www.actamat-journals.

com

In situ TEM observation of two-step martensitic transformation in aged NiTi shape memory alloy
L. Tan a, W.C. Crone
b

b,*

a Materials Science Program, University of Wisconsin Madison, 1509 University Avenue, Madison, WI 53706, USA Department of Engineering Physics, University of Wisconsin Madison, 1500 Engineering Drive, Madison, WI 53706-1687, USA

Received 10 August 2003; received in revised form 10 December 2003; accepted 15 December 2003

Abstract Martensitic transformation was investigated in an aged NiTi alloy with DSC and a temperature controllable TEM specimen stage to observe the inuence of Ti11 Ni14 precipitates and R-phase on martensitic transformation in situ. The R-phase, conventional martensitic twins, and a new morphology of interwoven austenite/martensitic structure were observed. 2003 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
Keywords: Martensitic transformation; Ageing; TEM; DSC

1. Introduction Shape memory eect, rst discovered in binary alloys of CuZn alloys and CuSn alloys in 1938 [1], was not widely applied until the discovery in NiTi alloys in 1963 [2]. In addition to the shape memory eect giving the material the ability to return to a predetermined shape when heated, NiTi shape memory alloys (SMAs) are capable of displaying pseudoelasticity giving the material the ability to transform between phases upon loading and unloading and recover to its original zero strain shape after signicant deformation. With the properties such as repeatability, wear resistance, corrosion resistance and biocompatibility, NiTi is the most commercially successful shape memory alloy [35]. The shape memory eect of NiTi is related to a martensitic transformation between the high-temperature austenite phase and the low-temperature martensite phase. Unlike fully annealed near-equiatomic NiTi alloys which transform from austenite to martensite directly, post-aged near-equiatomic NiTi alloys normally transform in two steps showing two peaks on the dif-

Corresponding author. Tel.: +1-608-262-8384; fax: +1-608-2637451. E-mail address: [email protected] (W.C. Crone).

ferential scanning calorimetry (DSC) curve in the cooling direction. The rst DSC peak correlates with the transformation from the austenite in a cubic structure (B2) to the R-phase in a rhombohedral structure; the second DSC peak correlates with the transformation from the R-phase to the martensite in a monoclinic structure (B190 ) [6]. Dlouhy et al. [7] proposed a new explanation based on their in-situ TEM observation that the second DSC peak in the cooling direction may include two transformations: R-phase to B190 in the grainboundary regions with precipitates, and B2 directly to B190 in the precipitate-free grain interiors. Multi-step (three or more steps) transformation could occur in appropriately aged near-equiatomic NiTi alloys. The explanations for the conditions of multi-step transformation are still debated [710], but it is clear that Ti11 Ni14 precipitates enhance the R-phase formation which is favorable for the shape memory eect [11]. It is known that Ti11 Ni14 precipitate, which helps martensitic transformation, appears in the aged NiTi shape memory alloys with nickel compositions greater than 50.5 at.% [12]. The the R-phase nucleates at Ti11 Ni14 precipitates because they have the same rhombohedral structure with a 0:738 nm and c 0:532 nm [13] for the R-phase and a 1:124 nm and c 0:5077 nm [12] for Ti11 Ni14 . Their same crystal structure and close lattice parameters makes formation of R-phase at

1359-6462/$ - see front matter 2003 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.scriptamat.2003.12.019

820

L. Tan, W.C. Crone / Scripta Materialia 50 (2004) 819823

the boundary of Ti11 Ni14 easier than the formation of the monoclinic martensite phase. In order to ensure the presence of the R-phase, an ageing temperature below 600 C was chosen in this study. In addition to the dierential scanning calorimetry (DSC) used to analyze the behavior of phase transformation, transmission electron microscopy (TEM) was employed to characterize the microstructural evolution during martensitic transformation in situ. The phase relationships among Ti11 Ni14 precipitate, R-phase and martensite, and a new martensite/austenite morphology are reported below.

3. Results and discussion The DSC results of the sample aged at 550 C for 20 min is shown in Fig. 1. Two distinct peaks in the cooling direction and two superimposed peaks in the heating direction are observed, indicating a two-step phase transformation. The two peaks in the cooling direction corresponding to the R-phase and the martensite phase (M) were determined to have transformation temperatures Rs )9 C, Rf )36 C, Ms )30 C, Mf )76 C. The subscript s denotes the temperature at which the phase transformation starts, and the f denotes the temperature at which the phase transformation nishes. The peak in the heating direction corresponds to the superposition of the R0 -phase [15] and the austenite (A) with R0s )23 C, R0f )1 C, As )16 C and Af )3 C determined by deconvoluting the curve. Thus the aged samples with Af )3 C is in the austenite phase at room temperature. The area under each power peak or valley separating the solid-state phases, austenite (A), martensite (M), rhombohedral-phase (R and R0 ), represents the heat of transformation. As can be seen from Fig. 1, the heat of transformation for A M is approximately equal to that of M A. Using TEM at room temperature, the material was shown to be in the austenite phase with selected area diraction (SAD) as is shown in the inset of Fig. 2. The two-step transformation (austenite to R-phase then martensite) observed by DSC was veried by TEM upon cooling with liquid nitrogen. At low ux of liquid nitrogen in the TEM specimen stage, R-phase formed rst generating a stress eld which reduced the phase transformation barrier of martensite. As shown in Fig. 2, the parallel structure of the R-phase grows from the nucleus sites at the surface boundaries of the Ti11 Ni14 precipitates, which is in accordance with the ndings of Kainuma et al. [17]. (No preferential distribution of the

2. Sample preparation and characterization The material used in this study was a commercial Ti 50.7at.%Ni alloy procured from Nitinol Devices & Components (NDC), Inc., in the form of as-rolled polycrystalline sheet (0.81 mm in thickness) and annealed condition. The sheet was cut into 5 mm by 10 mm at samples with electro-discharge machining (EDM). An ageing treatment was performed on the samples at 550 C for 20 min followed by quenching in water at room temperature. To eliminate the eect of the thick oxide layer created during thermal treatment, the samples were chemically etched with an 1HF + 4HNO3 5H2 O solution to remove the surface oxides, and mechanically polished using SiC paper in successive grades from 240 to 1200 grit prior to DSC analysis. The phase transformation characteristics of the aged samples were determined with a Perkin-Elmer DSC 7 instrument under sub-ambient conditions and analyzed according to the ASTM standardsF2004 [14] and F2005 [15]. During the DSC test, a sample weight of approximately 10 mg was analyzed with an empty aluminum pan as the reference. A temperature range from 60 to )100 C was scanned at a rate of 10 C/min during cooling and heating. JEOL 200CX-II TEM was used at 200 kV to characterize the microstructural evolution of the aged samples. Bright-eld images and selected area electron diraction patterns were obtained from cross-section specimens. Sections of 0.5 mm in thickness were cut from the cross-section of an aged sample and sandwiched between two silicon wafers with epoxy. The specimens were polished until the silicon appeared yellow, indicating that the specimen thickness is thinner than 5 lm [16]. Final thinning was done by argon ion milling after the specimen had been mounted on a copper grid with epoxy. Since the aged sample is in austenitic phase at room temperature, the martensitic transformation was investigated with a special TEM specimen stage cooled by liquid nitrogen. The ux of the liquid nitrogen was controlled to obtain desirable temperatures for detecting the stable phases occurring at dierent temperatures.

0.2 0.1

M R
Cooling

Heat Flow (W/g)

0.0 Heating -0.1 -0.2 -0.3

-0.4 -80

HMA = -18.6 J/g HAM = 18.4 J/g

R'+A
0 20 40

Temperature (C)

Fig. 1. DSC curves of Ti50.7at.%Ni alloy aged at 550 C for 20 min indicating a two-step phase transformation occurred during cooling and heating the NiTi alloy.

L. Tan, W.C. Crone / Scripta Materialia 50 (2004) 819823

821

Fig. 2. Bright-eld image of the aged Ni-rich NiTi SMA demonstrating the correlation of Ti11 Ni14 precipitates, R-phase (R), and interwoven austenite/martensite (M/A) phases. The inset is the select area diraction pattern taken at the region labelled M/A prior to cooling, which indicates the parent phase was austenite with [1 1 1]B2 zone.

Ti11 Ni14 precipitates was observed in this material in our previous work [18].) The Ti11 Ni14 phase can introduce internal tensile stress originating from the dierent density between the Ti11 Ni14 and matrix (B2) (qTi11 Ni14 =qmatrix 1:06) and the mist strain [19]. The stress state created by the presence of these precipitates decreases the barrier to R-phase transformation and the martensitic transformation that follows. The restriction of the Ti11 Ni14 precipitates on the growth of the R-phase is also indicated in Fig. 2, which conrms the suggestion that the high density of Ti11 Ni14 precipitates in the matrix of the aged Ni-rich NiTi SMAs are related to the smaller R-phase domain [20]. A martensite phase domain forming in a region of R-phase is shown in the upper left corner of Fig. 2. Martensite does not appear around the Ti11 Ni14 precipitates due to the larger nucleation barrier to martensite compared to that with R-phase [10]. The parallel structure of the R-phase is magnied in Fig. 3, and its corresponding diraction pattern with [1  0]B2 zone is shown in the inset. The 1 fringe contrast of the R-phase is also called the APB (antiphase boundary)-like contrast. This contrast was proposed to be due to a two dimensional crystal defect (or displacement boundary) that arises when a diusionless transformation originates in dierent regions of the crystal on dierent sublattices [21]. The diraction pattern of R-phase is consistent with the work of Wu and Wayman [20] who provided the detailed diraction information correlating with reciprocal lattice of Rphase. At high ux of liquid nitrogen in the TEM specimen stage, an interwoven austenite/martensite structure

Fig. 3. Bright-eld image of R-phase in the shape of parallel lines with 30 nm width. The inset is the diraction pattern of the R-phase in [1  0]B2 zone. 1

originates at the boundary of the R-phase and parent phase and grows into both the parent phase and the Rphase regions. The R-phase fraction diminished with the growth of the martensite, which may support the argument of Dlouhy et al. [7] that the R-phase transforms to B190 in addition to the direct transformation from B2 to B190 . An interwoven austenite/martensite structure, shown in Fig. 4, with white cells denoting the austenite phase and black and gray cells denoting the martensite variants was stabilized at this lower temperature. The corresponding diraction pattern of Fig. 4 with zone axis [1 1 1]B2 is shown in the inset of Fig. 4. The large bright diraction spots forming an equilateral triangle

822

L. Tan, W.C. Crone / Scripta Materialia 50 (2004) 819823

Fig. 4. Bright-eld image of martensitic bands in 50 nm width. White cells correspond to the austenite phase and black and gray to the martensite variants. The diraction pattern with the zone axis [1 1 1]B2 corresponds to the bright-eld image, where the large bright diraction spots in equilateral triangle correspond to austenite phase, and the small diraction spots beside the diraction spots of austenite phase correspond to the martensite phase.

Fig. 5. (a) Growth of conventional martensitic twins are shown during cooling. (b) Stabilized martensitic twins are shown after cooling with liquid nitrogen. The region shown in (b) is at the lower left corner of (a) and close to the edge of the sample.

4. Conclusions A two-step martensitic transformation occurring in an aged NiTi shape memory alloy was analyzed with DSC and TEM. A specimen stage cooled with liquid nitrogen was employed during TEM analysis to characterize the correlation of Ti11 Ni14 precipitates, R-phase, and martensite. It was revealed that the R-phase nucleates at Ti11 Ni14 precipitates and its growth is constrained by these precipitates. In addition to the growth of the conventional martensitic twins, a new morphology of interwoven austenite/martensitic twin structure was observed in this material, which forms preferentially at the defect-free and precipitate-free regions.

result from the austenite phase; and the streaks beside the austenite diraction spots result thin lamellae of the martensite variants. The dim diraction spots in a hexagonal pattern surrounding the large austenite diraction spots correspond to the R-phase in the up-right corner of the bright-eld image (Fig. 4). This is the rst time that this type of interwoven austenite/martensite structure has been reported to be observed in NiTi. In contrast to the situation in bulk shape memory alloys where it is not possible to form an interface between austenite and a single variant of martensite, in thin lms the situation can be quite different [22]. A similar interwoven austenite/martensite structure to that observed experimentally was predicted analytically for CuAlZn SMAs by Sreekala and Ananthakrishna [23] with a two-dimensional model. According to the in situ TEM observations shown here, this interwoven structure forms preferentially in defectfree and precipitate-free regions which are surrounded by the R-phase in the thin planar TEM sample. The formation of the interwoven austenite/martensite structure is thought to be a result of the strains surrounding the R-phase, which favor the growth of the slabs along the certain directions [23]. In addition to the interesting interwoven austenite/ martensite structure, conventional martensitic twins were also observed in the material during and after cooling with liquid nitrogen. As shown in Fig. 5(a), poorly dened martensitic twin boundaries start to grow when cooling at low ux of liquid nitrogen. After cooling at high ux of liquid nitrogen, straight and welldened martensitic twin boundaries are stabilized as shown in Fig. 5(b).

Acknowledgements Support for this research was provided by Air Force Oce of Scientic Research (award #F49620-01-0146).

References
[1] Greninger AB, Mooradian VG. Trans AIME 1938;128:337. [2] Buehler WJ, Gilfrich JW, Wiley RC. J Appl Phys 1963;34:1475. [3] Duerig T, Pelton A, St ckel D. Mater Sci Eng A 1999;273 o 275:149. [4] Humbeeck JV. Mater Sci Eng A 1999;273275:134. [5] Otsuka K, Kakeshita T. MRS Bull 2002;2:91. [6] Saburi T. In: Otsuka K, Wayman CM, editors. Shape memory materials. Cambridge University Press; 1998. p. 49. [7] Dlouhy A, Khalil-Alla J, Eggeler G. Philos Mag 2003;83:339. [8] Morawiec H, Stroz D, Goryczka T, Chrobak D. Scr Mater 1996;35:485. [9] Bataillard L, Bidaux JE, Gotthardt R. Philos Mag A 1998;78:327. [10] Khalil-Alla J, Ren X, Eggeler G. Acta Mater 2002;50:793. [11] Fuentes JMG, G mpel P, Strittmatter J. Adv Eng Mater u 2002;4:437.

L. Tan, W.C. Crone / Scripta Materialia 50 (2004) 819823 [12] Saburi T, Nenno S, Fukuda T. J Less-Common Metals 1986;125:157. [13] Goo E, Sinclair R. Acta Metall 1985;33:1717. [14] ASTM StandardsF2004: Standard Test Method for Transformation of NickelTitanium Alloys by Thermal Analysis. Annual book of ASTM standards, vol. 13.01, 2002. p. 1480. [15] ASTM StandardsF2005: Standard Terminology for Nickel Titanium Shape Memory Alloys. Annual book of ASTM standards, vol. 13.01, 2002. p. 1482. [16] Mccarey JP, Hulse J. Micron 1998;29:139.

823

[17] Kainuma R, Matsumoto M, Honma T. Proc ICOMAT-86, 1986. p. 717. [18] Tan L, Crone WC. Acta Mater 2002;50:4449. [19] Li DY, Wu XF, Ko T. Philos Mag A 1991;63:603. [20] Wu SK, Wayman CM. Acta Metall 1989;37:2805. [21] Goo E, Sinclair R. Scr Metall 1985;19:1257. [22] Bhattacharya K. Microstructure of martensite: Why it forms and how it gives rise to the shape-memory eect. Oxford University Press; 2003. p. 15977. [23] Sreekala S, Ananthakrishna G. Phys Rev Lett 2003;90:135501.