rsta.royalsocietypublishing.org
Effect of crystallographic compatibility and grain size on the functional fatigue of sputtered TiNiCuCo thin films
Research
C. Chluba1 , W. Ge2 , T. Dankwort1 , C. Bechtold3 , R. Lima
Cite this article: Chluba C, Ge W, Dankwort T, Bechtold C, Lima de Miranda R, Kienle L, Wuttig M, Quandt E. 2016 Effect of crystallographic compatibility and grain size on the functional fatigue of sputtered TiNiCuCo thin films. Phil. Trans. R. Soc. A 374: 20150311. http://dx.doi.org/10.1098/rsta.2015.0311
de Miranda3 , L. Kienle1 , M. Wuttig2 and E. Quandt1 1 Institute for Materials Science, Faculty of Engineering, University of
Kiel, Kiel, Germany 2 Department of Materials Science and Engineering, University of Maryland, College Park, MD, USA 3 Acquandas GmbH, Kiel, Germany CC, 0000-0001-9337-9846; EQ, 0000-0002-2314-5179
Accepted: 19 April 2016 One contribution of 16 to a discussion meeting issue ‘Taking the temperature of phase transitions in cool materials’. Subject Areas: materials science Keywords: NiTi, shape memory alloys, fatigue, thin films, superelasticity Author for correspondence: E. Quandt e-mail: [email protected]
The positive influence of crystallographic compatibility on the thermal transformation stability has been already investigated extensively in the literature. However, its influence on the stability of the shape memory effect or superelasticity used in actual applications is still unresolved. In this investigation sputtered films of a highly compatible TiNiCuCo composition with a transformation matrix middle eigenvalue of 1 ± 0.01 are exposed to thermal as well as to superelastic cycling. In agreement with previous results the thermal transformation of this alloy is with a temperature shift of less than 0.1 K for 40 cycles very stable; on the other hand, superelastic degradation behaviour was found to depend strongly on heat treatment parameters. To reveal the transformation dissimilarities between the differently heat-treated samples, the microstructure has been analysed by transmission electron microscopy, in situ stress polarization microscopy and synchrotron analysis. It is found that good crystallographic stability is not a sufficient criterion to avoid defect generation which guarantees high superelastic stability. For the investigated alloy, a small grain size was identified as the determining factor which increases the yield strength of the composition and decreases the functional degradation during superelastic cycling. This article is part of the themed issue ‘Taking the temperature of phase transitions in cool materials’.
2016 The Author(s) Published by the Royal Society. All rights reserved.
1. Introduction
.........................................................
rsta.royalsocietypublishing.org Phil. Trans. R. Soc. A 374: 20150311
The effects observed in shape memory alloys (SMAs) are based on the structural transformation of the high-temperature austenite phase to the low-temperature martensite phase [1]. This first order transformation is characterized by a change in crystallographic symmetry, in the case of TiNi-based compositions, from cubic austenite to monoclinic or orthorhombic martensite. The structural change is the basis of all related effects. The transformation can be induced by raising the temperature beyond austenite start and finish temperatures As and Af for the formation of austenite, and, by lowering the temperature below martensite start and finish temperature Ms and Mf for the reverse transformation. If an additional stress is applied, a pseudoelastic deformation can be observed, caused by the alignment of the long axis of martensite variants parallel to the axis of external tensile stress. Upon heating above the austenite finish temperature the original shape can be recovered. This shape memory effect can be used for actuator applications and has the potential to replace conventional linear actuators as a temperature variation is sufficient for actuation [2]. The phase boundary of martensite and austenite phases with respect to critical transformation stress and transformation temperatures is described by the Clausius–Clapeyron equation. This equation also describes the reversible stress-induced transformation at temperatures > Af from austenite to stress-induced martensite, known as superor pseudoelasticity. The superelastic transformation is characterized by large reversible strains up to 10% which makes it attractive for highly flexible structural materials and is already extensively used for medical applications [2]. Additionally, the structural change during the superelastic transformation along with the interrelated transformation enthalpy is promising for the development of novel cooling applications based on the elastocaloric effect [3]. Up to now, the limiting factor for a broad commercial device implementation of superelasticity and shape memory effect is the long-term stability. Structural fatigue describes the failure of a material due to crack nucleation and growth similar to the mechanisms known from conventional metals and strongly depends on surface quality, dislocation mobility and defects that can act as nucleation sites. Crack formation and propagation have been studied extensively by in situ synchrotron [4] and in situ scanning electron microscopy [5]. It has been found that in the austenite phase, stress-induced martensite can form at a crack tip. This transformation allows for stress relaxation at the stress tip. Conversely, the martensite phase has a lower resistance of crack extension compared with austenite [5]. Owing to these contradictory effects the understanding of the stress-induced martensite on the fracture mechanics is complicated and not entirely understood [6]. Functional fatigue of SMAs describes the change of thermally or stress-induced transformations and their related effects as a function of cycling number. Different cases have to be distinguished. The first case is the thermally induced transformation without an external stress field. If no external stress is applied a decrease of Ms is observed over the course of repeated transformations. This is attributed to the build-up and rearrangement of dislocations which can impede the martensitic transformation interface [7]. It has been observed that the crystallographic compatibility of the austenite and martensite phases, first described by the middle eigenvalue [8] and later refined through additional cofactor conditions [9], improves the thermal transformation stability significantly. This is attributed to the smaller transformation interface misfits which results in smaller local stresses at the interface and, consequently, in a reduction of dislocation formation [9,10]. Additionally, by approaching the ideal compatibility conditions a reduction of hysteresis for several SMA compositions has been predicted and was observed: e.g. a reduced hysteresis in TiNiCu [8], zero hysteresis in TiNiCuPd [11] and high reversibility of TiNiPd [10] and ZnAuCu [9]. In the second case, an additional global stress field is applied. As most SMA applications are actuators [2], or are based on superelasticity an additional stress or strain field is required. In this case, defects are not only created by local invariant lattice strains but by the global stress field. Also the martensite variants have to accommodate to the global field besides the local constraints.
2
Dogbone-shaped structured TiNiCuCo samples have been prepared by multilayer sputter deposition of 500 layers using a 4 TiNiCu target and an 8 Co target. Further information on the structuring process can be found in [24,27]. The as deposited amorphous samples with a thickness of 18 µm have been homogenized and crystallized in a sequential annealing step using a rapid thermal annealing device. Two different annealing temperatures have been chosen, 700◦ C for 15 min and 600◦ C for 15 min. No further heat treatment has been applied. Tensile tests have been performed using a Zwick Roell Z0.5 tensile testing device with an attached climate chamber. Functional fatigue tests of the superelastic transformation have been performed at slow strain rates of 1% min−1 for 200 cycles at room temperature (RT). In order
.........................................................
2. Experimental methods
3
rsta.royalsocietypublishing.org Phil. Trans. R. Soc. A 374: 20150311
In general, higher local stresses can be expected. Thus, for thermally induced transformation with an additionally applied strain or stress field different fatigue behaviour is observed. In contrast with the stress-free case an increase of Ms is reported. This is explained by an assisted martensitic transformation by dislocations and defect arrangement [7]. Superelastic fatigue is characterized by the change of the superelastic stress–strain behaviour from a flat to a steeper transformation curve progression, with a lower critical stress for transformation and the accumulation of remnant strain [12]. This degradation is commonly explained by slip deformation along with the formation of local stress fields that assist the stress-induced martensite transformation and enable the build-up of remnant martensite. Transformation bands, plastic deformation and the accumulation of remnant martensite during superelastic cycling have been investigated by numerous researchers using light microscopy [13,14]. One underlying deformation process is identified as the movement of dislocations that is investigated by in situ strain transmission electron microscopy (TEM) [15]. If high functional stability is required dislocation movement has to be hindered effectively without impeding the transformation [16]. Apart from some basic research on single crystals common NiTi-based alloys are in a polycrystalline state. The extent of dislocation mobility and slip resistance strongly depends on several microstructural features like work hardening [17], grain size [18], texture [19] or precipitates [20–24]. It was found that coherent, small Ni4 Ti3 precipitates in NiTi influence the transformation [20], suppress dislocation movement and improve the functional stability [21]. A similar stabilization effect has been found for precipitates in the TiNiZr system [22] and in the TiNiCu system [23,24]. Grain-size reduction increases the martensitic transformation temperature [18] and can even completely impede the martensitic transformation [25]. On the other hand, grain refinement is an effective measure to increase the yield strength of a material and to avoid plastic deformation [26]. The underlying microstructural mechanisms for functional fatigue have been studied extensively and are mostly understood. Nevertheless, it remains ambiguous to what extent microstructural features like grain size, precipitates and crystallographic compatibility obstruct slip and the formation of local stresses, remnant martensite and permanent strain. This is the case as the variation of a single microstructural parameter is an experimental challenge. Generally, by changing processing parameters several microstructural features are influenced simultaneously. In this work, we report on two nearly precipitate-free TiNiCuCo thin films of the same composition but with different grain size adjusted by the heat treatment. Crystallographic compatibility, investigated by in situ stress synchrotron analysis, is similar for both compositions and they fulfil the ideal compatibility condition almost perfectly. Thermally induced transformation stability and superelastic fatigue behaviour, measured by differential scanning calorimetry (DSC) and tensile tests, are being compared. The differences are attributed to the microstructural changes observed by polarization microscopy, TEM and synchrotron analysis. This investigation should reveal if good cofactor conditions are sufficient to establish a stable stress-induced transformation behaviour besides the already-known thermally induced transformation stability and how the grain size acts on the transformation mechanisms.
(a) Macroscopic transformation behaviour (i) Tensile cycling Ti50.8 Ni34.2 Cu12.5 Co2.5 films, annealed at 700◦ C, show a severe functional fatigue behaviour characterized by accumulation of permanent strain until stable conditions of 0.8% remnant strain after approximately 50 cycles are reached (figure 1). The critical stress of transformation changes from a distinct value of approximately 200 MPa in the first cycle to a wider range of lower critical stress values which are difficult to quantify accurately. By contrast, the sample annealed at 600◦ C shows a stable transformation for 200 cycles. At the start of the superelastic transformation, a stress peak can be observed that is almost reversible for 200 cycles and approximately 50 MPa higher than the actual superelastic transformation plateau. This stress peak for the nucleation of a martensite band is also observed for bulk materials [31]. The stress hysteresis of the first cycles is 86 MPa and 53 MPa wide for the samples annealed at 700◦ C and 600◦ C, respectively.
.........................................................
3. Results
4
rsta.royalsocietypublishing.org Phil. Trans. R. Soc. A 374: 20150311
to investigate the yield strength of the austenite phase, tests have been performed at high temperatures of 100–130◦ C, to prevent the stress-induced martensite transformation. DSC measurements have been conducted at heating rates of 10◦ C min−1 for cycles 1, 10, 20, 30 and 40 in the temperature range of −40◦ C to 40◦ C using a Perkin Elmer 1 DSC, calibrated with an indium standard. The cycles in between have been performed at a higher heating rate of 40◦ C min−1 using the same temperature range. Synchrotron measurements have been executed at APS BM11 (wavelength 0.11168 Å) measured in transmission mode using an area detector. The samples are mounted in a self-built micro tensile holder consisting of an actuator (Pi M226-26S) and a load cell which enables the acquisition of stress–strain curves during the X-ray diffraction (XRD) measurement. The area diffraction patterns are integrated by using azimuthal integration (fit2d) along the stress direction and perpendicular to the stress direction [28]. Superelastic tensile tests up to 51 cycles have been performed at RT. For the 1st, 51st and 200th cycle, XRD patterns are acquired stepwise along the complete superelastic stress–strain curve. The same sample holder has been applied to measure the microscopic transformation behaviour by using polarization microscopy. This measurement exploits the varying surface roughness of austenite and martensite phases and is a widely used technique to image the transformation interface and investigate changes of the transformation path after multiple transformation cycles [13]. The used polarization microscope Zeiss Axiomat is equipped with a video camera and 100 cycles of full superelastic transformation at RT have been recorded. Polishing of the sample is not necessary because sputter deposition yields surfaces with a low roughness. TEM analysis has been performed on two microscopes. Firstly, a FEI Tecnai F30 G2 (field emission gun, 300 kV) and second a Jeol JEM-2100 (LaB6 cathode, 200 kV). TEM samples have been prepared by focused ion beam (FEI Helios NanoLab) by cutting parallel to the mechanical load direction. Investigations of both heat-treated samples have been performed in the pristine state and after 51 of 200 cycles (the cycled samples are the same as used for the synchrotron measurements). First, bright field imaging was used to estimate the grain size of the samples. To investigate the accumulation of defects with cycling, weak beam dark field (WBDF) analysis was used which is a long established technique based on the change of Bragg condition within the stressed surrounding of a defect [29]. This method has been reported to be applicable for NiTibased alloys to image dislocations in the austenite phase [31]. In this work, the [100] direction of the austenite phase close to the g(2 g) or g(3 g) conditions is analysed. No thermal control was applied as Af is below RT.
(a)
(b)
350
5
700°C 15 min
600°C 15 min
stress (MPa)
200 cycle 1 cycle 10 cycle 50 cycle 100 cycle 200
150 100 50 0 0
0.01
0.02 strain
0.03
0
0.01
0.02 strain
0.03
Figure 1. Two hundred full superelastic cycles of Ti50.8 Ni34.2 Cu12.5 Co2.5 films annealed at (a) 700◦ C for 15 min and (b) 600◦ C for 15 min measured at RT.
(ii) Differential scanning calometry cycling For both heat treatments, the forward and reverse thermally induced transformation shows high reversibility for 40 cycles. The tangent method is used to determine the austenite and martensite finish temperatures (figure 2a). After 40 cycles, the measured Ms temperature is shifted by 0.02 K and 0.01 K for 700◦ C and 600◦ C, respectively. These values are smaller than the DSC measurement error of 0.1◦ C. Thus, apart from the statement that the temperature shifts are smaller than 0.1 K for 40 cycles, a reliable quantification is not possible. The peak shape of the 700◦ C sample with Ms − Mf = 8.4 K is broader compared with Ms − Mf = 4.4 K for the 600◦ C sample. The hysteresis is calculated according to dT = (Af + As − Mf − Ms )/2 which results in 9.8 K and 14.2 K for the samples annealed at 700◦ C and 600◦ C, respectively.
(iii) Tensile yield stress To measure the yield strength, cyclic tensile tests have been performed by sequentially increasing maximum stress using steps of 50 MPa (figure 3). In this work, the yield strength is defined as the critical stress with 0.2% irreversible strain remaining after unloading. The sample annealed at 700◦ C measured at 100◦ C has a yield strength of 390 MPa. The sample annealed at 600◦ C is measured at 100◦ C and 130◦ C to distinguish between plastic deformation and stress-induced martensite. As the deformation stress plateau is changing with temperature, we assume a combination of plastic deformation and martensitic stress-induced transformation. Because the experimental set-up is limited to 130◦ C, it can only be concluded that the austenite yield strength of the sample annealed at 600◦ C is at least 880 MPa which is at least 490 MPa higher compared with the 700◦ C heat-treated sample.
(b) Microstructural analysis XRD investigations of the undeformed samples at TTest < Af reveal the B2 structure as the dominant phase for both heat treatments. Weak additional peaks at 3.15◦ and 4.71◦ occur in both patterns which fit either the Ti2 Ni or the Ti2 Cu precipitate structure (figure 4).
(i) Remnant martensite The XRD azimuthal integrated patterns, parallel and perpendicular to the stress direction, are compared to investigate the evolution and texturing of remnant martensite during mechanical cycling. Integration area and specimen alignment with stress direction can be found in figure 4b.
.........................................................
250
rsta.royalsocietypublishing.org Phil. Trans. R. Soc. A 374: 20150311
300
700°C 15 min (a)
600°C 15 min (b)
martensite–austenite
–15
–10
–5
0
5
austenite–martensite
(c)
–10
heating
–5
0
–20
10
Ms –9.8°C
Mf –14.2°C
cooling –25
5
austenite–martensite
(d)
Ms –11.3°C
Mf –19.7°C
Af 3.7°C
10 K min–1
0.2 W g–1
heating
As –0.7°C
cooling –15 –10 temperature (°C)
–5 –20
–15
–10 –5 temperature (°C)
0
Figure 2. DSC measurements for 40 cycles of Ti50.8 Ni34.2 Cu12.5 Co2.5 annealed at 700◦ C (a,c) and 600◦ C (b,d). Martensite– austenite and austenite–martensite peaks are shown separately with the corresponding transformation start and finish temperatures evaluated using the tangent method which is exemplarily shown in (a). 700°C 15 min; TTest 100°C 600°C 15 min; TTest 100°C 600°C 15 min; TTest 130°C
1400 1200 stress (MPa)
1000 800 600 400 200 0 0
0.02
0.04 strain
0.06
0.08
Figure 3. Tensile tests of the samples heat treated at 700◦ C and 600◦ C to determine the yield strength of the austenite phase. Tests are performed at temperatures of 100◦ C and 130◦ C to avoid stress-induced martensitic transformation. The measurement is performed without applied stress in the undeformed state and after 51 full superelastic transformations. For the sample annealed at 600◦ C, no significant change of the XRD pattern after 51 cycles could be observed (figure 4c). For the sample annealed at 700◦ C
.........................................................
Af 0.4°C As –8.7°C
rsta.royalsocietypublishing.org Phil. Trans. R. Soc. A 374: 20150311
cycle 1 cycle 20 cycle 10 cycle 30 cycle 40
6
martensite–austenite
3 4 2q (°)
1
5
2
(211)B2
(200)B2
(111)B2
(211)B2
integrated area ^ s
(110)B2
(113)B19 (220)
(200) (200)B2
(111)B2
(110)B2
(100)B2
(020)
B19
B19
B19
(103)B19
(022)B19
(002)B19 /(021)B19
(002)B19
(001)B19
intensity (arb. units)
stress direction s
7
600°C
3 4 2q (°)
5
Figure 4. Influence of superelastic cycling on the formation of stabilized martensite in Ti50.8 Ni34.2 Cu12.5 Co2.5 annealed at 700◦ C (a) and 600◦ C (c). XRD patterns are measured without external stress using an area detector (b). Areas parallel (a,c, top) and perpendicular (a,c, middle) to the stress direction are integrated. Undeformed state and patterns after 51 stress-induced transformation cycles are compared and an accumulation of textured stabilized martensite (indexed with B19) for the sample annealed at 700◦ C is demonstrated. XRD patterns of Ti50.8 Ni34.2 Cu12.5 Co2.5 annealed at 700◦ C and 600◦ C depict B2 as the dominant phase. Small additional peaks appear at 3.15◦ and 4.71◦ .
(a)
(b) [100]B2
(100)B2 (100)B2
B2 B19 [001]B19 c
2
XRD measurement directions
[110]B2
a a
b [010]B19
]B 00 [1
[110]B2
a [100]B2
(110)B2 (110)B2 tension s
[100]B19 a l1 (100)B2/(100)B19 l2 (110)B2/(010)B19
Figure 5. (a) Distortion of the B2 unit cell with [100] and [110] orientation with respect to the mechanical loading direction. The change of the plane distances parallel and perpendicular ⊥ to the stress direction are denoted. (b) Unit cell transformation from B2 to B19 with the transformation matrix eigenvalues λ1 and λ2 which describe misfit of the lattice planes at the transformation interface during the transition.
(figure 4a) additional B19 peaks occur after 51 cycles with different intensities for the pattern along the strain direction compared with the pattern perpendicular to the strain direction. Along the strain direction (001), (002), (012), (021), (022), (103) and (113) B19 peaks appear, whereas perpendicular to the strain direction (020), (200) and (220) B19 peaks occur. This is a clear indication for the accumulation of remnant martensite with a [001] texture along the stress direction. The investigation of the B2 peaks also shows a significant increase in broadening which can be interpreted as the accumulation of stress or grain refinement.
(c) Crystallographic compatibility The aim of this investigation is to measure the lattice constants of the austenite and martensite phases under mechanical load during the superelastic transformation. This enables the calculation of the transformation matrix eigenvalues which describe the crystallographic
.........................................................
2
integrated area˙˙ s
(c)
rsta.royalsocietypublishing.org Phil. Trans. R. Soc. A 374: 20150311
1
cycle 51,˙˙ s cycle 51, ^ s cycle 1
(100)B2
(b)
700°C
intensity (arb. units)
(a)
uncycled cycled
(100)B2
3.04 Å
(b)
3.04
3.03 3.02
ΩΩ
3.01
Å
^ 100
200
300
400
^
139 2 GPa 0
100
200
300
400
(d)
2.14
2.14
2.13
2.11
ΩΩ
3.02 3.00
0
Å 2.12
3.03
(100)B2 m = 0.4 55.3 0.7 GPa
3.01
3.00
(c)
600°C 15min 3.05
70.2 0.5 GPa
2.13 (110)B2 transformation
Å
2.11
(010)B19
2.10 0
100
4.49 4.48 4.47
300
0
400
(001)B19
600°C
700°C
2.89 2.88 0
100
200 stress
300
400
200
300
400
2.13 2.12 2.11 600°C 700°C
1.000 l 2 0.995 0.990 0.985 0.980
(100)B19
100
(110)B2 transformation (010)B19
fitted (110)B2, (010)B19 and l2
(f) Å
Å
170.8 2 GPa m = 0.41
2.10 200
martensite a and c lattice constants
(e)
2.12
0
100
200 stress
300
400
Figure 6. (a,b) Variation of (100)B2 lattice plane distances with applied stress for the samples annealed at 700◦ C and 600◦ C. Measurements parallel, , and perpendicular, ⊥, to the stress direction are shown, also extensional modulus and the contraction perpendicular to the direction of applied stress μ are denoted for the sample annealed at 600◦ C. (c,d) Variation of the (100)B2 planes and the transition of the (110)B2 planes to the (010)B19 planes during the stress-induced transformation, measured perpendicular to the stress direction. (e) Variation of the stress-induced martensite planes (001)B19 (parallel to stress direction) and (100)B19 (perpendicular to stress direction) with applied stress. (f ) The change of (110)B2 and (010)B19 during the first cycle, shown in (c) and (e), is fitted linearly (top) to approximate the λ2 value for different transformation stresses (below).
compatibility during the transformation. Details of the compatibility theory can be found in [32]. Before the transformation starts the austenite unit cell is elastically deformed. Indices and orientations used in this evaluation can be found in figure 5a,b. The XRD area patterns of both heat-treated samples are measured during sequential mechanical loading and unloading of the first cycle and the 51st/200th cycle for the samples annealed at 700◦ C/600◦ C, respectively (figure 6). Azimuthal integrated patterns along the stress direction and perpendicular to the stress direction have been evaluated, similar to figure 4. The peaks are fitted numerically using a Gaussian fit function to determine the plane distances for the specific applied stress. Fitting errors, larger than the symbol size in figure 6, are denoted by error bars. Similar to the ex situ investigation in figure 4, peaks along the stress direction differ from those perpendicular to the stress direction. The austenite (100)B2 and (110)B2 peaks can be evaluated for both directions and reveal a distortion along the stress direction. In figure 6a–d, the distortion of the lattice planes for 600◦ C and 700◦ C are shown where denotes lattice planes under tensile stress along the stress direction and ⊥ describes the perpendicular direction under compressive stress (figure 5a). For the samples annealed at 600◦ C, this enables the calculation of
.........................................................
700°C 15min 3.05
rsta.royalsocietypublishing.org Phil. Trans. R. Soc. A 374: 20150311
(a)
8
The λ2 value during the first cycle is approximately 0.99 for 600◦ C and 700◦ C, respectively. While for 600◦ C the overall transformation behaviour of (100)B2 to (010)B19 does not change with increasing cycle number, for 700◦ C the (010)B19 plane distance changes gradually and accommodates much better to (110)B2 after 51 cycles. Synchrotron tests were performed at room temperature only, which implies a given transformation stress plateau. To extend the determination of the eigenvalues to other temperatures (other transformation stresses), the change of (110)B2 and (010)B19 with stress for the first cycle has been fitted and extrapolated numerically (figure 6f ). The change of λ2 with stress is very similar for both heat treatments; thus, the crystallographic compatibility at the interface is comparable for both samples. Nevertheless, differences between both samples can be observed. For the sample annealed at 700◦ C, no significant compression of (100)B2 perpendicular to the strain direction can be measured (figure 6a). Also a high remnant tensile strain of (100)B2 is detectable already after the first cycle. Additionally, the martensite lattice constants a (100)B19 and c (001)B19 are different for both heat treatments (figure 6e). (100)B19 , which can only be observed perpendicular to the stress direction, is another measure for crystallographic compatibility because it describes the second transformation direction of the habit plane from (100)B2 to (100)B19 . During the transformation, the crystallographic plane changes for the 700◦ C sample from 3.012 Å ((100)B2 ) to 2.886 Å ((100)B19 ) at 250 MPa and for 600◦ C from 3.005 Å ((100)B2 ) to 2.893 Å ((100)B19 ) at 330 MPa. As a result, the lattice misfit (100)B2 /(100)B19 of 3.8% for the sample annealed at 600◦ C is smaller compared with 4.3% for the sample annealed at 700◦ C. The (001)B19 planes can only be observed along the stress direction. In theory, this value is proportional to the transformation strain of the superelastic stress–strain curve. Here the sample annealed at 700◦ C shows a higher value despite the smaller superelastic transformation strain, even during the first superelastic cycle.
(d) Transmission electron microscopy investigations Microstructural properties like grain size, precipitate composition and formation of dislocations after superelastic cycling have been investigated by TEM analysis. Bright field and high angle annular dark field scanning transmission electron microscopy (HAADF–STEM) images show a large difference in grain size for both samples ranging from approximately 300 nm and greater than 5 µm for the samples annealed at 600◦ C and 700◦ C, respectively (figure 7a,b). WBDF technique is performed to investigate the lattice distortions caused by local defects [29]. The exact condition for the WBDF images could not be determined due to the low intensity of the Kikuchi lines; however, for all cases recorded the conditions are close to g(2 g) or g(3 g). For the 600◦ C samples before and after mechanical cycling (51 cycles) no formation of dislocations could be
.........................................................
bB19 d(010)B19 λ2 = √ = . d(110)B2 2aB2
9
rsta.royalsocietypublishing.org Phil. Trans. R. Soc. A 374: 20150311
the extensional modulus and the contraction perpendicular to the direction of applied stress of specific planes (figure 6b,d). The values of the extensional modulus under tension are determined to be 55.3 GPa for the [100] direction and 77.2 GPa for the [110] direction and are thus lower when compared with the values found for binary NiTi single crystals which are 92 GPa and 113 GPa, respectively [33]. It has to be pointed out that the sample annealed at 700◦ C is excluded from the calculation because the error of the single d-values and the nonlinear scattering with increasing stress is much higher. This is caused by the large grain size of approximately 5 µm compared with the XRD spot size of 500 µm which reduces the statistical significance. After a few cycles the dotted polycrystalline rings recorded by the area detector become smoother and the error of the evaluation decreases. During superelastic transformation the (110)B2 plane transforms to the (010)B19 plane which is only observed perpendicular to the stress direction (figure 6c,d). The compatibility at the habit plane is described by the middle eigenvalue λ2 which can be calculated for the orthorhombic martensite according to [8] (figure 5b):
(a)
10
(e)
(c)
(b)
0.5 mm (d)
(f)
Ni L
200 nm Co K
Cu K
Ti K
g 1 mm
0.5 mm
100 nm
Figure 7. (a) Bright field overview image of the sample annealed at 600◦ C and (b) HAADF–STEM overview image of the sample annealed at 700◦ C. For samples annealed at 600◦ C and 700◦ C, the grain size is approximately 300 nm and greater than 5 µm, respectively. For the latter case, a so-called curtaining effect is observed due to sample preparation by FIB cutting. (c,d) WBDF image of the 700◦ C heated sample in the undeformed state and after 51 superelastic cycles, respectively. After superelastic cycling the formation of dislocations is observed. (e) Mass-sensitive HAADF–STEM micrograph of the sample annealed at 700◦ C (before superelastic cycling) and corresponding EDX elemental maps (f ) reveal the formation of Ti-rich precipitates.
observed. However, owing to the small grain size, the use of WBDF is limited. WBDF images of the undeformed sample annealed at 700◦ C show no formation of dislocations (figure 7c). However, bright spots could be observed and likely originate from additional excitation due to Ti-rich precipitates which were observed by elemental sensitive STEM–HAADF and STEM– EDX measurements (figure 7e,f ). In comparison, the same material (annealing temperature 700◦ C) shows the formation of dislocations (bright lines) after 51 superelastic transformation cycles which clearly indicates the occurrence of irreversible plastic deformation (figure 7d).
(e) Microscopic transformation behaviour To investigate the macroscopic transformation behaviour, polarization microscope images have been obtained during the superelastic transformation to observe the evolution of the transformation interfaces with increasing cycle number (figure 8). Polarization microscopy is a widely used technique based on the difference in surface roughness between the martensite and austenite phase. The sample annealed at 600◦ C shows a single band transformation which does not change significantly for 200 cycles. Also the surface of austenite and martensite has a similar and low roughness. Before the first cycle and after 200 cycles atomic force measurements have been applied and no significant increase in surface roughness of the austenite phase is observed. The sample annealed 700◦ C shows a much higher initial surface roughness of the stress-induced martensite phase. The initially smooth austenite surface becomes rougher with superelastic cycling even after the release of external force. Additionally, the transformation path changes with increasing cycle number from a single interface to a transformation that takes place simultaneously and continuously within the whole sample.
.........................................................
0.5 mm
rsta.royalsocietypublishing.org Phil. Trans. R. Soc. A 374: 20150311
g
700°C 15 min cycle 1
cycle 5
600°C 15 min cycle 100
11
cycle 200
2 1 500 mm 3
Figure 8. Polarization microscope investigation to image martensite and austenite phase during superelastic loading of the sample annealed at 700◦ C for cycles 1, 5 and 100 and for the sample annealed at 600◦ C for cycle 200. Different loading conditions during the superelastic transformation are shown (left): (1) unloaded austenite state, (2) superelastic transformation and (3) loaded stress-induced martensite state. (Online version in colour.)
4. Discussion Both heat-treated samples exhibit a stable thermally induced transformation and a small hysteresis (figure 2). This observation combined with the high crystallographic compatibility measured by stress-induced XRD (figure 6) are in agreement with former investigations of different low hysteresis SMA systems like TiNiX (X = Cu, Pd, Pt, Au) [34] or ZnAuCu [9]. The transformation temperature stability is comparable with approximately 0.1 K reported for TiNiPd [10], ZnAuCu [9] and TiNiCuPd [35], which are alloys exhibiting ideal austenite/martensite compatibility of λ2 = 1. Nevertheless, annealing at 600◦ C compared with 700◦ C results in a significantly different functional degradation behaviour of the superelastic transformation. The sample annealed at 600◦ C reveals a stable stress–strain behaviour (figure 1) and a stable transformation path characterized by the reversible movement of a single transformation band (figure 8). The stress peak observed at the start of the superelastic plateau is assumed to be the energy barrier for stress-induced martensite transformation. Compared with the nucleation stress, the actual movement of the martensite transformation interface during the transformation is 50 MPa lower. For the back transformation, no similar nucleation peak is observed probably because remaining austenite provides a movable transformation interface. For the sample annealed at 700◦ C no similar nucleation behaviour is observed but there is a severe degradation of the superelastic transformation with increasing cycle number. Polarization microscope investigations show that the transformation path changes from distinct transformation interfaces in the first cycle to blurred transformation interfaces with higher cycle number. The same relation of superelastic degradation and the change of the transformation path has been investigated by several groups using IR thermography [36] and polarization microscopy [13]. The polarization microscope investigation also reveals a much rougher stressinduced martensite surface of the 700◦ C sample compared with the 600◦ C sample, already during the first cycle. By contrast, for the thermally induced martensite transformation no increase of surface roughness is observed. Apparently, the martensite variant structure of the thermally induced martensite is different from the stress-induced case. Microstructurally, the superelastic degradation of the 700◦ C sample is characterized by the accumulation of dislocations (figure 7) and the build-up of textured stabilized martensite. The high dislocation movement is caused by local stresses during the transformation because the yield stress (figure 3) is not exceeded by the macroscopic plateau stress. XRD analysis reveals a small
.........................................................
3 2
rsta.royalsocietypublishing.org Phil. Trans. R. Soc. A 374: 20150311
1
Two TiNiCuCo samples with similar crystallographic compatibility and precipitate content but different grain sizes of greater than 5 µm and approximately 300 nm are prepared by sputter deposition and sequential annealing. A strong asymmetry between thermally and stress-induced transformation stability has been found. While both grain sizes result in high thermally induced transformation stability, for the stress-induced transformation this is only the case for the fine grained sample. From these results, we conclude that good crystallographic compatibility of λ2 = 1 ± 0.01 is not a sufficient criterion for good superelastic functional fatigue behaviour. Apparently, additional parameters are mandatory to avoid the accumulation of dislocations and the formation of remnant martensite. For the investigated TiNiCuCo compositions, grain refinement results in a higher yield strength and a better resistance to dislocation nucleation and movement. Data accessibility. All data used for this publication are available from the corresponding author. Authors’ contributions. C.C. sample preparation, tensile tests, DSC measurements, TEM analysis, construction of in situ test sample holder, synchrotron measurements and manuscript preparation. W.G. coordination of synchrotron experiments and data evaluation. T.D. TEM analysis and evaluation. C.B. polarization microscopy measurements. R.L.d.M. sample preparation. L.K., M.W. and E.Q. conception, coordination and revision of the investigations. All authors gave final approval for publication. Competing interests. The authors declare that they have no financial or non-financial competing interests. Funding. The work of C.C. and E.Q. was funded by the German research foundation (DFG) within the priority program 1599 ‘Ferroic cooling’. M.W. is supported by grant no. NFS 1206397 and W.G. by grant no. DOE DESC0005448. Acknowledgements. The authors thank Christiane Zamponi for the FIB preparation of the TEM samples, Lars Bumke for the AFM measurements and Prof. McCord for the polarization microscope measurements. The authors T.D. and C.C. thank Dietrich Häußler for his comments on weak beam imaging. Use of the Advanced Photon Source—Argonne National Laboratory was supported by the US Department of Energy (DOE) Office of Science.
References 1. Otsuka K, Ren X. 2005 Physical metallurgy of Ti-Ni-based shape memory alloys. Prog. Mater. Sci. 50, 511–678. (doi:10.1016/j.pmatsci.2004.10.001) 2. Mohd Jani J, Leary M, Subic A, Gibson MA. 2014 A review of shape memory alloy research, applications and opportunities. Mater. Des. 56, 1078–1113. (doi:10.1016/j.matdes.2013.11.084) 3. Fähler S et al. 2012 Caloric effects in ferroic materials: new concepts for cooling. Adv. Eng. Mater. 14, 10–19. (doi:10.1002/adem.201100178) 4. Daymond MR, Young ML, Almer JD, Dunand DC. 2007 Strain and texture evolution during mechanical loading of a crack tip in martensitic shape-memory NiTi. Acta Mater. 55, 3929–3942. (doi:10.1016/j.actamat.2007.03.013)
.........................................................
5. Conclusion
12
rsta.royalsocietypublishing.org Phil. Trans. R. Soc. A 374: 20150311
amount of precipitates contained in both samples, which are identified as a Ti-rich phase using STEM–EDX analysis (figure 7). A more detailed structure determination was not possible due to their small size. As the precipitates are found in both samples, precipitation hardening can be excluded as the reason for the huge difference in functional fatigue. The TEM analysis also reveals a large grain size difference of both samples. Because other microstructural parameters like precipitates and crystallographic compatibility are similar for both samples, the grain size seems to be the major reason for the difference in functional stability. Fine grains can act in two ways to stabilize the transformation. First, by limiting the dislocation movement which is a convenient measure to increase the yield strength of conventional metals [37] as well as for TiNi [16]. One the other hand, it was also found that a high yield strength is not necessarily a warrant for stable functional behaviour [38]. Second, the fine-grained structure might act as a constraint for the martensite variants. This results in finer martensite twins which reduce the surface roughness of the stress-induced martensite and thereby avoid local stress peaks which are the reason for plastic deformation and stabilized remnant martensite.
13 .........................................................
rsta.royalsocietypublishing.org Phil. Trans. R. Soc. A 374: 20150311
5. Gollerthan S, Young ML, Baruj A, Frenzel J, Schmahl WW, Eggeler G. 2009 Fracture mechanics and microstructure in NiTi shape memory alloys. Acta Mater. 57, 1015–1025. (doi:10.1016/j.actamat.2008.10.055) 6. Eiselstein L, Sire RA, James BA. 2005 Review of fatigue and fracture behavior in NiTi. In ASM Symp. on Materials and Processes for Medical Devices Conf., Boston, MA, USA, 13–16 November 2005. 7. Morgan NB, Friend CM. 2001 A review of shape memory stability in NiTi alloys. J. Phys. IV Fr. 11, 325–332. (doi:10.1051/jp4:2001855) 8. Cui J et al. 2006 Combinatorial search of thermoelastic shape-memory alloys with extremely small hysteresis width. Nat. Mater. 5, 286–290. (doi:10.1038/nmat1593) 9. Song Y, Chen X, Dabade V, Shield TW, James RD. 2013 Enhanced reversibility and unusual microstructure of a phase-transforming material. Nature 502, 85–88. (doi:10.1038/ nature12532) 10. Chen X, Srivastava V, Dabade V, James RD. 2013 Study of the cofactor conditions: conditions of supercompatibility between phases. J. Mech. Phys. Solids 61, 2566–2587. (doi:10.1016/j.jmps.2013.08.004) 11. Zarnetta R et al. 2010 Identification of quaternary shape memory alloys with near-zero thermal hysteresis and unprecedented functional stability. Adv. Funct. Mater. 20, 1917–1923. (doi:10.1002/adfm.200902336) 12. Miyazaki S, Hashinaga T, Ishida A. 1996 Martensitic transformations in sputter-deposited Ti-Ni-Cu shape memory alloy thin films. Thin Solid Films 281–282, 364–367. (doi:10.1016/ 0040-6090(96)08627-0) 13. Brinson LC, Schmidt I, Lammering R. 2004 Stress-induced transformation behavior of a polycrystalline NiTi shape memory alloy: micro and macromechanical investigations via in situ optical microscopy. J. Mech. Phys. Solids 52, 1549–1571. (doi:10.1016/j.jmps.2004.01.001) 14. Iadicola M, Shaw J. 2002 The effect of uniaxial cyclic deformation on the evolution of phase transformation fronts in pseudoelastic NiTi wire. J. Intell. Mater. 13, 143–155. (doi:10.1106/ 104538902028220) 15. Jiang X, Hida M, Takemoto Y, Sakakibara A, Yasuda H, Mori H. 1997 In situ observation of stress-induced martensitic transformation and plastic deformation in TiNi alloy. Mater. Sci. Eng. A 238, 303–308. (doi:10.1016/S0921-5093(97)00422-X) 16. Hornbogen E, Heckmann A. 2003 Improved fatigue resistance of pseudo-elastic NiTi alloys by thermo-mechanical treatment. Materwiss. Werksttech. 34, 464–468. (doi:10.1002/ mawe.200390094) 17. Miller DA, Lagoudas DC. 2000 Thermomechanical characterization of NiTiCu and NiTi SMA actuators: influence of plastic strains. Smart Mater. Struct. 9, 640–652. (doi:10.1088/ 0964-1726/9/5/308) 18. Gil FJ, Manero JM, Planell JA. 1995 Effect of grain size on the martensitic transformation in NiTi alloy. J. Mater. Sci. 30, 2526–2530. (doi:10.1007/BF00362129) 19. Gall K, Sehitoglu H, Anderson R, Karaman I, Chumlyakov YI, Kireeva IV. 2001 On the mechanical behavior of single crystal NiTi shape memory alloys and related polycrystalline phenomenon. Mater. Sci. Eng. A 317, 85–92. (doi:10.1016/S0921-5093(01)01183-2) 20. Khalil-Allafi J, Dlouhy A, Eggeler G. 2002 Ni4 Ti3 -precipitation during aging of NiTi shape memory alloys and its influence on martensitic phase transformations. Acta Mater. 50, 4255–4274. (doi:10.1016/S1359-6454(02)00257-4) 21. Gall K, Maier HJ. 2002 Cyclic deformation mechanisms in precipitated NiTi shape memory alloys. Acta Mater. 50, 4643–4657. (doi:10.1016/S1359-6454(02)00315-4) 22. Kim HY, Mizutani M, Miyazaki S. 2009 Crystallization process and shape memory properties of Ti-Ni-Zr thin films. Acta Mater. 57, 1920–1930. (doi:10.1016/j.actamat.2008.12.036) 23. Ishida A, Sato M, Gao Z. 2014 Effects of Ti content on microstructure and shape memory behavior of Tix Ni(84.5−x) Cu15.5 (x = 44.6–55.4) thin films. Acta Mater. 69, 292–300. (doi:10.1016/j.actamat.2014.02.006) 24. Chluba C, Ge W, Lima de Miranda R, Strobel J, Kienle L, Quandt E, Wuttig M. 2015 Ultralow-fatigue shape memory alloy films. Science 348, 1004–1007. (doi:10.1126/science. 1261164) 25. Waitz T, Antretter T, Fischer FD, Simha NK, Karnthaler HP. 2007 Size effects on the martensitic phase transformation of NiTi nanograins. J. Mech. Phys. Solids 55, 419–444. (doi:10.1016/j.jmps.2006.06.006)
14 .........................................................
rsta.royalsocietypublishing.org Phil. Trans. R. Soc. A 374: 20150311
26. Sergueeva A, Song C, Valiev R, Mukherjee A. 2003 Structure and properties of amorphous and nanocrystalline NiTi prepared by severe plastic deformation and annealing. Mater. Sci. Eng. A 339, 159–165. (doi:10.1016/S0921-5093(02)00122-3) 27. Lima de Miranda R, Zamponi C, Quandt E. 2012 Micropatterned freestanding superelastic TiNi films. Adv. Eng. Mater. 15, 66–69. (doi:10.1002/adem.201200197) 28. Hammersley A. 1997 FIT2D: an introduction and overview. Intern. Rep. 29. Cockayne DJH, Ray ILF, Whelan MJ. 1969 Investigations of dislocation strain fields using weak beams. Philos. Mag. A 20, 1265–1270. (doi:10.1080/14786436908228210) 30. Simon T, Kröger A, Somsen C, Dlouhy A, Eggeler G. 2010 On the multiplication of dislocations during martensitic transformations in NiTi shape memory alloys. Acta Mater. 58, 1850–1860. (doi:10.1016/j.actamat.2009.11.028) 31. Sun Q, Li Z, Tse K. 2001 On superelastic deformation of NiTi shape memory alloy micro-tubes and wires—band nucleation and propagation. In Smart structures and structronic systems (eds U Gabbert, HS Tzou). Solid Mechanics and Its Applications, vol. 89, pp. 113–120. Dordrecht, The Netherlands: Springer. (doi:10.1007/978-94-010-0724-5_15) 32. James RD, Zhang Z. 2005 A way to search for multiferroic materials with ‘unlikely’ combinations of physical properties. In Magnetism and structure in functional materials (eds A Planes, L Mañosa, A Saxena). Springer Series in Materials Science, vol. 79, pp. 159–175. Berlin, Germany: Springer. (doi:10.1007/3-540-31631-0_9) 33. Rajagopalan S, Little AL, Bourke MAM, Vaidyanathan R. 2005 Elastic modulus of shapememory NiTi from in situ neutron diffraction during macroscopic loading, instrumented indentation, and extensometry. Appl. Phys. Lett. 86, 081901. (doi:10.1063/1.1863437) 34. Zhang Z, James RD, Müller S. 2009 Energy barriers and hysteresis in martensitic phase transformations. Acta Mater. 57, 4332–4352. (doi:10.1016/j.actamat.2009.05.034) 35. Zarnetta R, Ehmann M, Savan A, Ludwig A. 2010 Identification of optimized Ti-Ni-Cu shape memory alloy compositions for high-frequency thin film microactuator applications. Smart Mater. Struct. 19, 065032. (doi:10.1088/0964-1726/19/6/065032) 36. Ossmer H, Chluba C, Gueltig M, Quandt E, Kohl M. 2015 Local evolution of the elastocaloric effect in TiNi-based films. Shape Mem. Superelast. 1, 142–152. (doi:10.1007/s40830015-0014-3nn) 37. Lasalmonie A, Strudel J. 1986 Influence of grain size on the mechanical behaviour of some high strength materials. J. Mater. Sci. 21, 1837–1852. (doi:10.1007/BF00547918) 38. Gall K, Tyber J, Wilkesanders G, Robertson SW, Ritchie RO, Maier HJ. 2008 Effect of microstructure on the fatigue of hot-rolled and cold-drawn NiTi shape memory alloys. Mater. Sci. Eng. A 486, 389–403. (doi:10.1016/j.msea.2007.11.033)
Effect of crystallographic compatibility and grain size on the functional fatigue of sputtered TiNiCuCo thin films
Research
C. Chluba1 , W. Ge2 , T. Dankwort1 , C. Bechtold3 , R. Lima
Cite this article: Chluba C, Ge W, Dankwort T, Bechtold C, Lima de Miranda R, Kienle L, Wuttig M, Quandt E. 2016 Effect of crystallographic compatibility and grain size on the functional fatigue of sputtered TiNiCuCo thin films. Phil. Trans. R. Soc. A 374: 20150311. http://dx.doi.org/10.1098/rsta.2015.0311
de Miranda3 , L. Kienle1 , M. Wuttig2 and E. Quandt1 1 Institute for Materials Science, Faculty of Engineering, University of
Kiel, Kiel, Germany 2 Department of Materials Science and Engineering, University of Maryland, College Park, MD, USA 3 Acquandas GmbH, Kiel, Germany CC, 0000-0001-9337-9846; EQ, 0000-0002-2314-5179
Accepted: 19 April 2016 One contribution of 16 to a discussion meeting issue ‘Taking the temperature of phase transitions in cool materials’. Subject Areas: materials science Keywords: NiTi, shape memory alloys, fatigue, thin films, superelasticity Author for correspondence: E. Quandt e-mail: [email protected]
The positive influence of crystallographic compatibility on the thermal transformation stability has been already investigated extensively in the literature. However, its influence on the stability of the shape memory effect or superelasticity used in actual applications is still unresolved. In this investigation sputtered films of a highly compatible TiNiCuCo composition with a transformation matrix middle eigenvalue of 1 ± 0.01 are exposed to thermal as well as to superelastic cycling. In agreement with previous results the thermal transformation of this alloy is with a temperature shift of less than 0.1 K for 40 cycles very stable; on the other hand, superelastic degradation behaviour was found to depend strongly on heat treatment parameters. To reveal the transformation dissimilarities between the differently heat-treated samples, the microstructure has been analysed by transmission electron microscopy, in situ stress polarization microscopy and synchrotron analysis. It is found that good crystallographic stability is not a sufficient criterion to avoid defect generation which guarantees high superelastic stability. For the investigated alloy, a small grain size was identified as the determining factor which increases the yield strength of the composition and decreases the functional degradation during superelastic cycling. This article is part of the themed issue ‘Taking the temperature of phase transitions in cool materials’.
2016 The Author(s) Published by the Royal Society. All rights reserved.
1. Introduction
.........................................................
rsta.royalsocietypublishing.org Phil. Trans. R. Soc. A 374: 20150311
The effects observed in shape memory alloys (SMAs) are based on the structural transformation of the high-temperature austenite phase to the low-temperature martensite phase [1]. This first order transformation is characterized by a change in crystallographic symmetry, in the case of TiNi-based compositions, from cubic austenite to monoclinic or orthorhombic martensite. The structural change is the basis of all related effects. The transformation can be induced by raising the temperature beyond austenite start and finish temperatures As and Af for the formation of austenite, and, by lowering the temperature below martensite start and finish temperature Ms and Mf for the reverse transformation. If an additional stress is applied, a pseudoelastic deformation can be observed, caused by the alignment of the long axis of martensite variants parallel to the axis of external tensile stress. Upon heating above the austenite finish temperature the original shape can be recovered. This shape memory effect can be used for actuator applications and has the potential to replace conventional linear actuators as a temperature variation is sufficient for actuation [2]. The phase boundary of martensite and austenite phases with respect to critical transformation stress and transformation temperatures is described by the Clausius–Clapeyron equation. This equation also describes the reversible stress-induced transformation at temperatures > Af from austenite to stress-induced martensite, known as superor pseudoelasticity. The superelastic transformation is characterized by large reversible strains up to 10% which makes it attractive for highly flexible structural materials and is already extensively used for medical applications [2]. Additionally, the structural change during the superelastic transformation along with the interrelated transformation enthalpy is promising for the development of novel cooling applications based on the elastocaloric effect [3]. Up to now, the limiting factor for a broad commercial device implementation of superelasticity and shape memory effect is the long-term stability. Structural fatigue describes the failure of a material due to crack nucleation and growth similar to the mechanisms known from conventional metals and strongly depends on surface quality, dislocation mobility and defects that can act as nucleation sites. Crack formation and propagation have been studied extensively by in situ synchrotron [4] and in situ scanning electron microscopy [5]. It has been found that in the austenite phase, stress-induced martensite can form at a crack tip. This transformation allows for stress relaxation at the stress tip. Conversely, the martensite phase has a lower resistance of crack extension compared with austenite [5]. Owing to these contradictory effects the understanding of the stress-induced martensite on the fracture mechanics is complicated and not entirely understood [6]. Functional fatigue of SMAs describes the change of thermally or stress-induced transformations and their related effects as a function of cycling number. Different cases have to be distinguished. The first case is the thermally induced transformation without an external stress field. If no external stress is applied a decrease of Ms is observed over the course of repeated transformations. This is attributed to the build-up and rearrangement of dislocations which can impede the martensitic transformation interface [7]. It has been observed that the crystallographic compatibility of the austenite and martensite phases, first described by the middle eigenvalue [8] and later refined through additional cofactor conditions [9], improves the thermal transformation stability significantly. This is attributed to the smaller transformation interface misfits which results in smaller local stresses at the interface and, consequently, in a reduction of dislocation formation [9,10]. Additionally, by approaching the ideal compatibility conditions a reduction of hysteresis for several SMA compositions has been predicted and was observed: e.g. a reduced hysteresis in TiNiCu [8], zero hysteresis in TiNiCuPd [11] and high reversibility of TiNiPd [10] and ZnAuCu [9]. In the second case, an additional global stress field is applied. As most SMA applications are actuators [2], or are based on superelasticity an additional stress or strain field is required. In this case, defects are not only created by local invariant lattice strains but by the global stress field. Also the martensite variants have to accommodate to the global field besides the local constraints.
2
Dogbone-shaped structured TiNiCuCo samples have been prepared by multilayer sputter deposition of 500 layers using a 4 TiNiCu target and an 8 Co target. Further information on the structuring process can be found in [24,27]. The as deposited amorphous samples with a thickness of 18 µm have been homogenized and crystallized in a sequential annealing step using a rapid thermal annealing device. Two different annealing temperatures have been chosen, 700◦ C for 15 min and 600◦ C for 15 min. No further heat treatment has been applied. Tensile tests have been performed using a Zwick Roell Z0.5 tensile testing device with an attached climate chamber. Functional fatigue tests of the superelastic transformation have been performed at slow strain rates of 1% min−1 for 200 cycles at room temperature (RT). In order
.........................................................
2. Experimental methods
3
rsta.royalsocietypublishing.org Phil. Trans. R. Soc. A 374: 20150311
In general, higher local stresses can be expected. Thus, for thermally induced transformation with an additionally applied strain or stress field different fatigue behaviour is observed. In contrast with the stress-free case an increase of Ms is reported. This is explained by an assisted martensitic transformation by dislocations and defect arrangement [7]. Superelastic fatigue is characterized by the change of the superelastic stress–strain behaviour from a flat to a steeper transformation curve progression, with a lower critical stress for transformation and the accumulation of remnant strain [12]. This degradation is commonly explained by slip deformation along with the formation of local stress fields that assist the stress-induced martensite transformation and enable the build-up of remnant martensite. Transformation bands, plastic deformation and the accumulation of remnant martensite during superelastic cycling have been investigated by numerous researchers using light microscopy [13,14]. One underlying deformation process is identified as the movement of dislocations that is investigated by in situ strain transmission electron microscopy (TEM) [15]. If high functional stability is required dislocation movement has to be hindered effectively without impeding the transformation [16]. Apart from some basic research on single crystals common NiTi-based alloys are in a polycrystalline state. The extent of dislocation mobility and slip resistance strongly depends on several microstructural features like work hardening [17], grain size [18], texture [19] or precipitates [20–24]. It was found that coherent, small Ni4 Ti3 precipitates in NiTi influence the transformation [20], suppress dislocation movement and improve the functional stability [21]. A similar stabilization effect has been found for precipitates in the TiNiZr system [22] and in the TiNiCu system [23,24]. Grain-size reduction increases the martensitic transformation temperature [18] and can even completely impede the martensitic transformation [25]. On the other hand, grain refinement is an effective measure to increase the yield strength of a material and to avoid plastic deformation [26]. The underlying microstructural mechanisms for functional fatigue have been studied extensively and are mostly understood. Nevertheless, it remains ambiguous to what extent microstructural features like grain size, precipitates and crystallographic compatibility obstruct slip and the formation of local stresses, remnant martensite and permanent strain. This is the case as the variation of a single microstructural parameter is an experimental challenge. Generally, by changing processing parameters several microstructural features are influenced simultaneously. In this work, we report on two nearly precipitate-free TiNiCuCo thin films of the same composition but with different grain size adjusted by the heat treatment. Crystallographic compatibility, investigated by in situ stress synchrotron analysis, is similar for both compositions and they fulfil the ideal compatibility condition almost perfectly. Thermally induced transformation stability and superelastic fatigue behaviour, measured by differential scanning calorimetry (DSC) and tensile tests, are being compared. The differences are attributed to the microstructural changes observed by polarization microscopy, TEM and synchrotron analysis. This investigation should reveal if good cofactor conditions are sufficient to establish a stable stress-induced transformation behaviour besides the already-known thermally induced transformation stability and how the grain size acts on the transformation mechanisms.
(a) Macroscopic transformation behaviour (i) Tensile cycling Ti50.8 Ni34.2 Cu12.5 Co2.5 films, annealed at 700◦ C, show a severe functional fatigue behaviour characterized by accumulation of permanent strain until stable conditions of 0.8% remnant strain after approximately 50 cycles are reached (figure 1). The critical stress of transformation changes from a distinct value of approximately 200 MPa in the first cycle to a wider range of lower critical stress values which are difficult to quantify accurately. By contrast, the sample annealed at 600◦ C shows a stable transformation for 200 cycles. At the start of the superelastic transformation, a stress peak can be observed that is almost reversible for 200 cycles and approximately 50 MPa higher than the actual superelastic transformation plateau. This stress peak for the nucleation of a martensite band is also observed for bulk materials [31]. The stress hysteresis of the first cycles is 86 MPa and 53 MPa wide for the samples annealed at 700◦ C and 600◦ C, respectively.
.........................................................
3. Results
4
rsta.royalsocietypublishing.org Phil. Trans. R. Soc. A 374: 20150311
to investigate the yield strength of the austenite phase, tests have been performed at high temperatures of 100–130◦ C, to prevent the stress-induced martensite transformation. DSC measurements have been conducted at heating rates of 10◦ C min−1 for cycles 1, 10, 20, 30 and 40 in the temperature range of −40◦ C to 40◦ C using a Perkin Elmer 1 DSC, calibrated with an indium standard. The cycles in between have been performed at a higher heating rate of 40◦ C min−1 using the same temperature range. Synchrotron measurements have been executed at APS BM11 (wavelength 0.11168 Å) measured in transmission mode using an area detector. The samples are mounted in a self-built micro tensile holder consisting of an actuator (Pi M226-26S) and a load cell which enables the acquisition of stress–strain curves during the X-ray diffraction (XRD) measurement. The area diffraction patterns are integrated by using azimuthal integration (fit2d) along the stress direction and perpendicular to the stress direction [28]. Superelastic tensile tests up to 51 cycles have been performed at RT. For the 1st, 51st and 200th cycle, XRD patterns are acquired stepwise along the complete superelastic stress–strain curve. The same sample holder has been applied to measure the microscopic transformation behaviour by using polarization microscopy. This measurement exploits the varying surface roughness of austenite and martensite phases and is a widely used technique to image the transformation interface and investigate changes of the transformation path after multiple transformation cycles [13]. The used polarization microscope Zeiss Axiomat is equipped with a video camera and 100 cycles of full superelastic transformation at RT have been recorded. Polishing of the sample is not necessary because sputter deposition yields surfaces with a low roughness. TEM analysis has been performed on two microscopes. Firstly, a FEI Tecnai F30 G2 (field emission gun, 300 kV) and second a Jeol JEM-2100 (LaB6 cathode, 200 kV). TEM samples have been prepared by focused ion beam (FEI Helios NanoLab) by cutting parallel to the mechanical load direction. Investigations of both heat-treated samples have been performed in the pristine state and after 51 of 200 cycles (the cycled samples are the same as used for the synchrotron measurements). First, bright field imaging was used to estimate the grain size of the samples. To investigate the accumulation of defects with cycling, weak beam dark field (WBDF) analysis was used which is a long established technique based on the change of Bragg condition within the stressed surrounding of a defect [29]. This method has been reported to be applicable for NiTibased alloys to image dislocations in the austenite phase [31]. In this work, the [100] direction of the austenite phase close to the g(2 g) or g(3 g) conditions is analysed. No thermal control was applied as Af is below RT.
(a)
(b)
350
5
700°C 15 min
600°C 15 min
stress (MPa)
200 cycle 1 cycle 10 cycle 50 cycle 100 cycle 200
150 100 50 0 0
0.01
0.02 strain
0.03
0
0.01
0.02 strain
0.03
Figure 1. Two hundred full superelastic cycles of Ti50.8 Ni34.2 Cu12.5 Co2.5 films annealed at (a) 700◦ C for 15 min and (b) 600◦ C for 15 min measured at RT.
(ii) Differential scanning calometry cycling For both heat treatments, the forward and reverse thermally induced transformation shows high reversibility for 40 cycles. The tangent method is used to determine the austenite and martensite finish temperatures (figure 2a). After 40 cycles, the measured Ms temperature is shifted by 0.02 K and 0.01 K for 700◦ C and 600◦ C, respectively. These values are smaller than the DSC measurement error of 0.1◦ C. Thus, apart from the statement that the temperature shifts are smaller than 0.1 K for 40 cycles, a reliable quantification is not possible. The peak shape of the 700◦ C sample with Ms − Mf = 8.4 K is broader compared with Ms − Mf = 4.4 K for the 600◦ C sample. The hysteresis is calculated according to dT = (Af + As − Mf − Ms )/2 which results in 9.8 K and 14.2 K for the samples annealed at 700◦ C and 600◦ C, respectively.
(iii) Tensile yield stress To measure the yield strength, cyclic tensile tests have been performed by sequentially increasing maximum stress using steps of 50 MPa (figure 3). In this work, the yield strength is defined as the critical stress with 0.2% irreversible strain remaining after unloading. The sample annealed at 700◦ C measured at 100◦ C has a yield strength of 390 MPa. The sample annealed at 600◦ C is measured at 100◦ C and 130◦ C to distinguish between plastic deformation and stress-induced martensite. As the deformation stress plateau is changing with temperature, we assume a combination of plastic deformation and martensitic stress-induced transformation. Because the experimental set-up is limited to 130◦ C, it can only be concluded that the austenite yield strength of the sample annealed at 600◦ C is at least 880 MPa which is at least 490 MPa higher compared with the 700◦ C heat-treated sample.
(b) Microstructural analysis XRD investigations of the undeformed samples at TTest < Af reveal the B2 structure as the dominant phase for both heat treatments. Weak additional peaks at 3.15◦ and 4.71◦ occur in both patterns which fit either the Ti2 Ni or the Ti2 Cu precipitate structure (figure 4).
(i) Remnant martensite The XRD azimuthal integrated patterns, parallel and perpendicular to the stress direction, are compared to investigate the evolution and texturing of remnant martensite during mechanical cycling. Integration area and specimen alignment with stress direction can be found in figure 4b.
.........................................................
250
rsta.royalsocietypublishing.org Phil. Trans. R. Soc. A 374: 20150311
300
700°C 15 min (a)
600°C 15 min (b)
martensite–austenite
–15
–10
–5
0
5
austenite–martensite
(c)
–10
heating
–5
0
–20
10
Ms –9.8°C
Mf –14.2°C
cooling –25
5
austenite–martensite
(d)
Ms –11.3°C
Mf –19.7°C
Af 3.7°C
10 K min–1
0.2 W g–1
heating
As –0.7°C
cooling –15 –10 temperature (°C)
–5 –20
–15
–10 –5 temperature (°C)
0
Figure 2. DSC measurements for 40 cycles of Ti50.8 Ni34.2 Cu12.5 Co2.5 annealed at 700◦ C (a,c) and 600◦ C (b,d). Martensite– austenite and austenite–martensite peaks are shown separately with the corresponding transformation start and finish temperatures evaluated using the tangent method which is exemplarily shown in (a). 700°C 15 min; TTest 100°C 600°C 15 min; TTest 100°C 600°C 15 min; TTest 130°C
1400 1200 stress (MPa)
1000 800 600 400 200 0 0
0.02
0.04 strain
0.06
0.08
Figure 3. Tensile tests of the samples heat treated at 700◦ C and 600◦ C to determine the yield strength of the austenite phase. Tests are performed at temperatures of 100◦ C and 130◦ C to avoid stress-induced martensitic transformation. The measurement is performed without applied stress in the undeformed state and after 51 full superelastic transformations. For the sample annealed at 600◦ C, no significant change of the XRD pattern after 51 cycles could be observed (figure 4c). For the sample annealed at 700◦ C
.........................................................
Af 0.4°C As –8.7°C
rsta.royalsocietypublishing.org Phil. Trans. R. Soc. A 374: 20150311
cycle 1 cycle 20 cycle 10 cycle 30 cycle 40
6
martensite–austenite
3 4 2q (°)
1
5
2
(211)B2
(200)B2
(111)B2
(211)B2
integrated area ^ s
(110)B2
(113)B19 (220)
(200) (200)B2
(111)B2
(110)B2
(100)B2
(020)
B19
B19
B19
(103)B19
(022)B19
(002)B19 /(021)B19
(002)B19
(001)B19
intensity (arb. units)
stress direction s
7
600°C
3 4 2q (°)
5
Figure 4. Influence of superelastic cycling on the formation of stabilized martensite in Ti50.8 Ni34.2 Cu12.5 Co2.5 annealed at 700◦ C (a) and 600◦ C (c). XRD patterns are measured without external stress using an area detector (b). Areas parallel (a,c, top) and perpendicular (a,c, middle) to the stress direction are integrated. Undeformed state and patterns after 51 stress-induced transformation cycles are compared and an accumulation of textured stabilized martensite (indexed with B19) for the sample annealed at 700◦ C is demonstrated. XRD patterns of Ti50.8 Ni34.2 Cu12.5 Co2.5 annealed at 700◦ C and 600◦ C depict B2 as the dominant phase. Small additional peaks appear at 3.15◦ and 4.71◦ .
(a)
(b) [100]B2
(100)B2 (100)B2
B2 B19 [001]B19 c
2
XRD measurement directions
[110]B2
a a
b [010]B19
]B 00 [1
[110]B2
a [100]B2
(110)B2 (110)B2 tension s
[100]B19 a l1 (100)B2/(100)B19 l2 (110)B2/(010)B19
Figure 5. (a) Distortion of the B2 unit cell with [100] and [110] orientation with respect to the mechanical loading direction. The change of the plane distances parallel and perpendicular ⊥ to the stress direction are denoted. (b) Unit cell transformation from B2 to B19 with the transformation matrix eigenvalues λ1 and λ2 which describe misfit of the lattice planes at the transformation interface during the transition.
(figure 4a) additional B19 peaks occur after 51 cycles with different intensities for the pattern along the strain direction compared with the pattern perpendicular to the strain direction. Along the strain direction (001), (002), (012), (021), (022), (103) and (113) B19 peaks appear, whereas perpendicular to the strain direction (020), (200) and (220) B19 peaks occur. This is a clear indication for the accumulation of remnant martensite with a [001] texture along the stress direction. The investigation of the B2 peaks also shows a significant increase in broadening which can be interpreted as the accumulation of stress or grain refinement.
(c) Crystallographic compatibility The aim of this investigation is to measure the lattice constants of the austenite and martensite phases under mechanical load during the superelastic transformation. This enables the calculation of the transformation matrix eigenvalues which describe the crystallographic
.........................................................
2
integrated area˙˙ s
(c)
rsta.royalsocietypublishing.org Phil. Trans. R. Soc. A 374: 20150311
1
cycle 51,˙˙ s cycle 51, ^ s cycle 1
(100)B2
(b)
700°C
intensity (arb. units)
(a)
uncycled cycled
(100)B2
3.04 Å
(b)
3.04
3.03 3.02
ΩΩ
3.01
Å
^ 100
200
300
400
^
139 2 GPa 0
100
200
300
400
(d)
2.14
2.14
2.13
2.11
ΩΩ
3.02 3.00
0
Å 2.12
3.03
(100)B2 m = 0.4 55.3 0.7 GPa
3.01
3.00
(c)
600°C 15min 3.05
70.2 0.5 GPa
2.13 (110)B2 transformation
Å
2.11
(010)B19
2.10 0
100
4.49 4.48 4.47
300
0
400
(001)B19
600°C
700°C
2.89 2.88 0
100
200 stress
300
400
200
300
400
2.13 2.12 2.11 600°C 700°C
1.000 l 2 0.995 0.990 0.985 0.980
(100)B19
100
(110)B2 transformation (010)B19
fitted (110)B2, (010)B19 and l2
(f) Å
Å
170.8 2 GPa m = 0.41
2.10 200
martensite a and c lattice constants
(e)
2.12
0
100
200 stress
300
400
Figure 6. (a,b) Variation of (100)B2 lattice plane distances with applied stress for the samples annealed at 700◦ C and 600◦ C. Measurements parallel, , and perpendicular, ⊥, to the stress direction are shown, also extensional modulus and the contraction perpendicular to the direction of applied stress μ are denoted for the sample annealed at 600◦ C. (c,d) Variation of the (100)B2 planes and the transition of the (110)B2 planes to the (010)B19 planes during the stress-induced transformation, measured perpendicular to the stress direction. (e) Variation of the stress-induced martensite planes (001)B19 (parallel to stress direction) and (100)B19 (perpendicular to stress direction) with applied stress. (f ) The change of (110)B2 and (010)B19 during the first cycle, shown in (c) and (e), is fitted linearly (top) to approximate the λ2 value for different transformation stresses (below).
compatibility during the transformation. Details of the compatibility theory can be found in [32]. Before the transformation starts the austenite unit cell is elastically deformed. Indices and orientations used in this evaluation can be found in figure 5a,b. The XRD area patterns of both heat-treated samples are measured during sequential mechanical loading and unloading of the first cycle and the 51st/200th cycle for the samples annealed at 700◦ C/600◦ C, respectively (figure 6). Azimuthal integrated patterns along the stress direction and perpendicular to the stress direction have been evaluated, similar to figure 4. The peaks are fitted numerically using a Gaussian fit function to determine the plane distances for the specific applied stress. Fitting errors, larger than the symbol size in figure 6, are denoted by error bars. Similar to the ex situ investigation in figure 4, peaks along the stress direction differ from those perpendicular to the stress direction. The austenite (100)B2 and (110)B2 peaks can be evaluated for both directions and reveal a distortion along the stress direction. In figure 6a–d, the distortion of the lattice planes for 600◦ C and 700◦ C are shown where denotes lattice planes under tensile stress along the stress direction and ⊥ describes the perpendicular direction under compressive stress (figure 5a). For the samples annealed at 600◦ C, this enables the calculation of
.........................................................
700°C 15min 3.05
rsta.royalsocietypublishing.org Phil. Trans. R. Soc. A 374: 20150311
(a)
8
The λ2 value during the first cycle is approximately 0.99 for 600◦ C and 700◦ C, respectively. While for 600◦ C the overall transformation behaviour of (100)B2 to (010)B19 does not change with increasing cycle number, for 700◦ C the (010)B19 plane distance changes gradually and accommodates much better to (110)B2 after 51 cycles. Synchrotron tests were performed at room temperature only, which implies a given transformation stress plateau. To extend the determination of the eigenvalues to other temperatures (other transformation stresses), the change of (110)B2 and (010)B19 with stress for the first cycle has been fitted and extrapolated numerically (figure 6f ). The change of λ2 with stress is very similar for both heat treatments; thus, the crystallographic compatibility at the interface is comparable for both samples. Nevertheless, differences between both samples can be observed. For the sample annealed at 700◦ C, no significant compression of (100)B2 perpendicular to the strain direction can be measured (figure 6a). Also a high remnant tensile strain of (100)B2 is detectable already after the first cycle. Additionally, the martensite lattice constants a (100)B19 and c (001)B19 are different for both heat treatments (figure 6e). (100)B19 , which can only be observed perpendicular to the stress direction, is another measure for crystallographic compatibility because it describes the second transformation direction of the habit plane from (100)B2 to (100)B19 . During the transformation, the crystallographic plane changes for the 700◦ C sample from 3.012 Å ((100)B2 ) to 2.886 Å ((100)B19 ) at 250 MPa and for 600◦ C from 3.005 Å ((100)B2 ) to 2.893 Å ((100)B19 ) at 330 MPa. As a result, the lattice misfit (100)B2 /(100)B19 of 3.8% for the sample annealed at 600◦ C is smaller compared with 4.3% for the sample annealed at 700◦ C. The (001)B19 planes can only be observed along the stress direction. In theory, this value is proportional to the transformation strain of the superelastic stress–strain curve. Here the sample annealed at 700◦ C shows a higher value despite the smaller superelastic transformation strain, even during the first superelastic cycle.
(d) Transmission electron microscopy investigations Microstructural properties like grain size, precipitate composition and formation of dislocations after superelastic cycling have been investigated by TEM analysis. Bright field and high angle annular dark field scanning transmission electron microscopy (HAADF–STEM) images show a large difference in grain size for both samples ranging from approximately 300 nm and greater than 5 µm for the samples annealed at 600◦ C and 700◦ C, respectively (figure 7a,b). WBDF technique is performed to investigate the lattice distortions caused by local defects [29]. The exact condition for the WBDF images could not be determined due to the low intensity of the Kikuchi lines; however, for all cases recorded the conditions are close to g(2 g) or g(3 g). For the 600◦ C samples before and after mechanical cycling (51 cycles) no formation of dislocations could be
.........................................................
bB19 d(010)B19 λ2 = √ = . d(110)B2 2aB2
9
rsta.royalsocietypublishing.org Phil. Trans. R. Soc. A 374: 20150311
the extensional modulus and the contraction perpendicular to the direction of applied stress of specific planes (figure 6b,d). The values of the extensional modulus under tension are determined to be 55.3 GPa for the [100] direction and 77.2 GPa for the [110] direction and are thus lower when compared with the values found for binary NiTi single crystals which are 92 GPa and 113 GPa, respectively [33]. It has to be pointed out that the sample annealed at 700◦ C is excluded from the calculation because the error of the single d-values and the nonlinear scattering with increasing stress is much higher. This is caused by the large grain size of approximately 5 µm compared with the XRD spot size of 500 µm which reduces the statistical significance. After a few cycles the dotted polycrystalline rings recorded by the area detector become smoother and the error of the evaluation decreases. During superelastic transformation the (110)B2 plane transforms to the (010)B19 plane which is only observed perpendicular to the stress direction (figure 6c,d). The compatibility at the habit plane is described by the middle eigenvalue λ2 which can be calculated for the orthorhombic martensite according to [8] (figure 5b):
(a)
10
(e)
(c)
(b)
0.5 mm (d)
(f)
Ni L
200 nm Co K
Cu K
Ti K
g 1 mm
0.5 mm
100 nm
Figure 7. (a) Bright field overview image of the sample annealed at 600◦ C and (b) HAADF–STEM overview image of the sample annealed at 700◦ C. For samples annealed at 600◦ C and 700◦ C, the grain size is approximately 300 nm and greater than 5 µm, respectively. For the latter case, a so-called curtaining effect is observed due to sample preparation by FIB cutting. (c,d) WBDF image of the 700◦ C heated sample in the undeformed state and after 51 superelastic cycles, respectively. After superelastic cycling the formation of dislocations is observed. (e) Mass-sensitive HAADF–STEM micrograph of the sample annealed at 700◦ C (before superelastic cycling) and corresponding EDX elemental maps (f ) reveal the formation of Ti-rich precipitates.
observed. However, owing to the small grain size, the use of WBDF is limited. WBDF images of the undeformed sample annealed at 700◦ C show no formation of dislocations (figure 7c). However, bright spots could be observed and likely originate from additional excitation due to Ti-rich precipitates which were observed by elemental sensitive STEM–HAADF and STEM– EDX measurements (figure 7e,f ). In comparison, the same material (annealing temperature 700◦ C) shows the formation of dislocations (bright lines) after 51 superelastic transformation cycles which clearly indicates the occurrence of irreversible plastic deformation (figure 7d).
(e) Microscopic transformation behaviour To investigate the macroscopic transformation behaviour, polarization microscope images have been obtained during the superelastic transformation to observe the evolution of the transformation interfaces with increasing cycle number (figure 8). Polarization microscopy is a widely used technique based on the difference in surface roughness between the martensite and austenite phase. The sample annealed at 600◦ C shows a single band transformation which does not change significantly for 200 cycles. Also the surface of austenite and martensite has a similar and low roughness. Before the first cycle and after 200 cycles atomic force measurements have been applied and no significant increase in surface roughness of the austenite phase is observed. The sample annealed 700◦ C shows a much higher initial surface roughness of the stress-induced martensite phase. The initially smooth austenite surface becomes rougher with superelastic cycling even after the release of external force. Additionally, the transformation path changes with increasing cycle number from a single interface to a transformation that takes place simultaneously and continuously within the whole sample.
.........................................................
0.5 mm
rsta.royalsocietypublishing.org Phil. Trans. R. Soc. A 374: 20150311
g
700°C 15 min cycle 1
cycle 5
600°C 15 min cycle 100
11
cycle 200
2 1 500 mm 3
Figure 8. Polarization microscope investigation to image martensite and austenite phase during superelastic loading of the sample annealed at 700◦ C for cycles 1, 5 and 100 and for the sample annealed at 600◦ C for cycle 200. Different loading conditions during the superelastic transformation are shown (left): (1) unloaded austenite state, (2) superelastic transformation and (3) loaded stress-induced martensite state. (Online version in colour.)
4. Discussion Both heat-treated samples exhibit a stable thermally induced transformation and a small hysteresis (figure 2). This observation combined with the high crystallographic compatibility measured by stress-induced XRD (figure 6) are in agreement with former investigations of different low hysteresis SMA systems like TiNiX (X = Cu, Pd, Pt, Au) [34] or ZnAuCu [9]. The transformation temperature stability is comparable with approximately 0.1 K reported for TiNiPd [10], ZnAuCu [9] and TiNiCuPd [35], which are alloys exhibiting ideal austenite/martensite compatibility of λ2 = 1. Nevertheless, annealing at 600◦ C compared with 700◦ C results in a significantly different functional degradation behaviour of the superelastic transformation. The sample annealed at 600◦ C reveals a stable stress–strain behaviour (figure 1) and a stable transformation path characterized by the reversible movement of a single transformation band (figure 8). The stress peak observed at the start of the superelastic plateau is assumed to be the energy barrier for stress-induced martensite transformation. Compared with the nucleation stress, the actual movement of the martensite transformation interface during the transformation is 50 MPa lower. For the back transformation, no similar nucleation peak is observed probably because remaining austenite provides a movable transformation interface. For the sample annealed at 700◦ C no similar nucleation behaviour is observed but there is a severe degradation of the superelastic transformation with increasing cycle number. Polarization microscope investigations show that the transformation path changes from distinct transformation interfaces in the first cycle to blurred transformation interfaces with higher cycle number. The same relation of superelastic degradation and the change of the transformation path has been investigated by several groups using IR thermography [36] and polarization microscopy [13]. The polarization microscope investigation also reveals a much rougher stressinduced martensite surface of the 700◦ C sample compared with the 600◦ C sample, already during the first cycle. By contrast, for the thermally induced martensite transformation no increase of surface roughness is observed. Apparently, the martensite variant structure of the thermally induced martensite is different from the stress-induced case. Microstructurally, the superelastic degradation of the 700◦ C sample is characterized by the accumulation of dislocations (figure 7) and the build-up of textured stabilized martensite. The high dislocation movement is caused by local stresses during the transformation because the yield stress (figure 3) is not exceeded by the macroscopic plateau stress. XRD analysis reveals a small
.........................................................
3 2
rsta.royalsocietypublishing.org Phil. Trans. R. Soc. A 374: 20150311
1
Two TiNiCuCo samples with similar crystallographic compatibility and precipitate content but different grain sizes of greater than 5 µm and approximately 300 nm are prepared by sputter deposition and sequential annealing. A strong asymmetry between thermally and stress-induced transformation stability has been found. While both grain sizes result in high thermally induced transformation stability, for the stress-induced transformation this is only the case for the fine grained sample. From these results, we conclude that good crystallographic compatibility of λ2 = 1 ± 0.01 is not a sufficient criterion for good superelastic functional fatigue behaviour. Apparently, additional parameters are mandatory to avoid the accumulation of dislocations and the formation of remnant martensite. For the investigated TiNiCuCo compositions, grain refinement results in a higher yield strength and a better resistance to dislocation nucleation and movement. Data accessibility. All data used for this publication are available from the corresponding author. Authors’ contributions. C.C. sample preparation, tensile tests, DSC measurements, TEM analysis, construction of in situ test sample holder, synchrotron measurements and manuscript preparation. W.G. coordination of synchrotron experiments and data evaluation. T.D. TEM analysis and evaluation. C.B. polarization microscopy measurements. R.L.d.M. sample preparation. L.K., M.W. and E.Q. conception, coordination and revision of the investigations. All authors gave final approval for publication. Competing interests. The authors declare that they have no financial or non-financial competing interests. Funding. The work of C.C. and E.Q. was funded by the German research foundation (DFG) within the priority program 1599 ‘Ferroic cooling’. M.W. is supported by grant no. NFS 1206397 and W.G. by grant no. DOE DESC0005448. Acknowledgements. The authors thank Christiane Zamponi for the FIB preparation of the TEM samples, Lars Bumke for the AFM measurements and Prof. McCord for the polarization microscope measurements. The authors T.D. and C.C. thank Dietrich Häußler for his comments on weak beam imaging. Use of the Advanced Photon Source—Argonne National Laboratory was supported by the US Department of Energy (DOE) Office of Science.
References 1. Otsuka K, Ren X. 2005 Physical metallurgy of Ti-Ni-based shape memory alloys. Prog. Mater. Sci. 50, 511–678. (doi:10.1016/j.pmatsci.2004.10.001) 2. Mohd Jani J, Leary M, Subic A, Gibson MA. 2014 A review of shape memory alloy research, applications and opportunities. Mater. Des. 56, 1078–1113. (doi:10.1016/j.matdes.2013.11.084) 3. Fähler S et al. 2012 Caloric effects in ferroic materials: new concepts for cooling. Adv. Eng. Mater. 14, 10–19. (doi:10.1002/adem.201100178) 4. Daymond MR, Young ML, Almer JD, Dunand DC. 2007 Strain and texture evolution during mechanical loading of a crack tip in martensitic shape-memory NiTi. Acta Mater. 55, 3929–3942. (doi:10.1016/j.actamat.2007.03.013)
.........................................................
5. Conclusion
12
rsta.royalsocietypublishing.org Phil. Trans. R. Soc. A 374: 20150311
amount of precipitates contained in both samples, which are identified as a Ti-rich phase using STEM–EDX analysis (figure 7). A more detailed structure determination was not possible due to their small size. As the precipitates are found in both samples, precipitation hardening can be excluded as the reason for the huge difference in functional fatigue. The TEM analysis also reveals a large grain size difference of both samples. Because other microstructural parameters like precipitates and crystallographic compatibility are similar for both samples, the grain size seems to be the major reason for the difference in functional stability. Fine grains can act in two ways to stabilize the transformation. First, by limiting the dislocation movement which is a convenient measure to increase the yield strength of conventional metals [37] as well as for TiNi [16]. One the other hand, it was also found that a high yield strength is not necessarily a warrant for stable functional behaviour [38]. Second, the fine-grained structure might act as a constraint for the martensite variants. This results in finer martensite twins which reduce the surface roughness of the stress-induced martensite and thereby avoid local stress peaks which are the reason for plastic deformation and stabilized remnant martensite.
13 .........................................................
rsta.royalsocietypublishing.org Phil. Trans. R. Soc. A 374: 20150311
5. Gollerthan S, Young ML, Baruj A, Frenzel J, Schmahl WW, Eggeler G. 2009 Fracture mechanics and microstructure in NiTi shape memory alloys. Acta Mater. 57, 1015–1025. (doi:10.1016/j.actamat.2008.10.055) 6. Eiselstein L, Sire RA, James BA. 2005 Review of fatigue and fracture behavior in NiTi. In ASM Symp. on Materials and Processes for Medical Devices Conf., Boston, MA, USA, 13–16 November 2005. 7. Morgan NB, Friend CM. 2001 A review of shape memory stability in NiTi alloys. J. Phys. IV Fr. 11, 325–332. (doi:10.1051/jp4:2001855) 8. Cui J et al. 2006 Combinatorial search of thermoelastic shape-memory alloys with extremely small hysteresis width. Nat. Mater. 5, 286–290. (doi:10.1038/nmat1593) 9. Song Y, Chen X, Dabade V, Shield TW, James RD. 2013 Enhanced reversibility and unusual microstructure of a phase-transforming material. Nature 502, 85–88. (doi:10.1038/ nature12532) 10. Chen X, Srivastava V, Dabade V, James RD. 2013 Study of the cofactor conditions: conditions of supercompatibility between phases. J. Mech. Phys. Solids 61, 2566–2587. (doi:10.1016/j.jmps.2013.08.004) 11. Zarnetta R et al. 2010 Identification of quaternary shape memory alloys with near-zero thermal hysteresis and unprecedented functional stability. Adv. Funct. Mater. 20, 1917–1923. (doi:10.1002/adfm.200902336) 12. Miyazaki S, Hashinaga T, Ishida A. 1996 Martensitic transformations in sputter-deposited Ti-Ni-Cu shape memory alloy thin films. Thin Solid Films 281–282, 364–367. (doi:10.1016/ 0040-6090(96)08627-0) 13. Brinson LC, Schmidt I, Lammering R. 2004 Stress-induced transformation behavior of a polycrystalline NiTi shape memory alloy: micro and macromechanical investigations via in situ optical microscopy. J. Mech. Phys. Solids 52, 1549–1571. (doi:10.1016/j.jmps.2004.01.001) 14. Iadicola M, Shaw J. 2002 The effect of uniaxial cyclic deformation on the evolution of phase transformation fronts in pseudoelastic NiTi wire. J. Intell. Mater. 13, 143–155. (doi:10.1106/ 104538902028220) 15. Jiang X, Hida M, Takemoto Y, Sakakibara A, Yasuda H, Mori H. 1997 In situ observation of stress-induced martensitic transformation and plastic deformation in TiNi alloy. Mater. Sci. Eng. A 238, 303–308. (doi:10.1016/S0921-5093(97)00422-X) 16. Hornbogen E, Heckmann A. 2003 Improved fatigue resistance of pseudo-elastic NiTi alloys by thermo-mechanical treatment. Materwiss. Werksttech. 34, 464–468. (doi:10.1002/ mawe.200390094) 17. Miller DA, Lagoudas DC. 2000 Thermomechanical characterization of NiTiCu and NiTi SMA actuators: influence of plastic strains. Smart Mater. Struct. 9, 640–652. (doi:10.1088/ 0964-1726/9/5/308) 18. Gil FJ, Manero JM, Planell JA. 1995 Effect of grain size on the martensitic transformation in NiTi alloy. J. Mater. Sci. 30, 2526–2530. (doi:10.1007/BF00362129) 19. Gall K, Sehitoglu H, Anderson R, Karaman I, Chumlyakov YI, Kireeva IV. 2001 On the mechanical behavior of single crystal NiTi shape memory alloys and related polycrystalline phenomenon. Mater. Sci. Eng. A 317, 85–92. (doi:10.1016/S0921-5093(01)01183-2) 20. Khalil-Allafi J, Dlouhy A, Eggeler G. 2002 Ni4 Ti3 -precipitation during aging of NiTi shape memory alloys and its influence on martensitic phase transformations. Acta Mater. 50, 4255–4274. (doi:10.1016/S1359-6454(02)00257-4) 21. Gall K, Maier HJ. 2002 Cyclic deformation mechanisms in precipitated NiTi shape memory alloys. Acta Mater. 50, 4643–4657. (doi:10.1016/S1359-6454(02)00315-4) 22. Kim HY, Mizutani M, Miyazaki S. 2009 Crystallization process and shape memory properties of Ti-Ni-Zr thin films. Acta Mater. 57, 1920–1930. (doi:10.1016/j.actamat.2008.12.036) 23. Ishida A, Sato M, Gao Z. 2014 Effects of Ti content on microstructure and shape memory behavior of Tix Ni(84.5−x) Cu15.5 (x = 44.6–55.4) thin films. Acta Mater. 69, 292–300. (doi:10.1016/j.actamat.2014.02.006) 24. Chluba C, Ge W, Lima de Miranda R, Strobel J, Kienle L, Quandt E, Wuttig M. 2015 Ultralow-fatigue shape memory alloy films. Science 348, 1004–1007. (doi:10.1126/science. 1261164) 25. Waitz T, Antretter T, Fischer FD, Simha NK, Karnthaler HP. 2007 Size effects on the martensitic phase transformation of NiTi nanograins. J. Mech. Phys. Solids 55, 419–444. (doi:10.1016/j.jmps.2006.06.006)
14 .........................................................
rsta.royalsocietypublishing.org Phil. Trans. R. Soc. A 374: 20150311
26. Sergueeva A, Song C, Valiev R, Mukherjee A. 2003 Structure and properties of amorphous and nanocrystalline NiTi prepared by severe plastic deformation and annealing. Mater. Sci. Eng. A 339, 159–165. (doi:10.1016/S0921-5093(02)00122-3) 27. Lima de Miranda R, Zamponi C, Quandt E. 2012 Micropatterned freestanding superelastic TiNi films. Adv. Eng. Mater. 15, 66–69. (doi:10.1002/adem.201200197) 28. Hammersley A. 1997 FIT2D: an introduction and overview. Intern. Rep. 29. Cockayne DJH, Ray ILF, Whelan MJ. 1969 Investigations of dislocation strain fields using weak beams. Philos. Mag. A 20, 1265–1270. (doi:10.1080/14786436908228210) 30. Simon T, Kröger A, Somsen C, Dlouhy A, Eggeler G. 2010 On the multiplication of dislocations during martensitic transformations in NiTi shape memory alloys. Acta Mater. 58, 1850–1860. (doi:10.1016/j.actamat.2009.11.028) 31. Sun Q, Li Z, Tse K. 2001 On superelastic deformation of NiTi shape memory alloy micro-tubes and wires—band nucleation and propagation. In Smart structures and structronic systems (eds U Gabbert, HS Tzou). Solid Mechanics and Its Applications, vol. 89, pp. 113–120. Dordrecht, The Netherlands: Springer. (doi:10.1007/978-94-010-0724-5_15) 32. James RD, Zhang Z. 2005 A way to search for multiferroic materials with ‘unlikely’ combinations of physical properties. In Magnetism and structure in functional materials (eds A Planes, L Mañosa, A Saxena). Springer Series in Materials Science, vol. 79, pp. 159–175. Berlin, Germany: Springer. (doi:10.1007/3-540-31631-0_9) 33. Rajagopalan S, Little AL, Bourke MAM, Vaidyanathan R. 2005 Elastic modulus of shapememory NiTi from in situ neutron diffraction during macroscopic loading, instrumented indentation, and extensometry. Appl. Phys. Lett. 86, 081901. (doi:10.1063/1.1863437) 34. Zhang Z, James RD, Müller S. 2009 Energy barriers and hysteresis in martensitic phase transformations. Acta Mater. 57, 4332–4352. (doi:10.1016/j.actamat.2009.05.034) 35. Zarnetta R, Ehmann M, Savan A, Ludwig A. 2010 Identification of optimized Ti-Ni-Cu shape memory alloy compositions for high-frequency thin film microactuator applications. Smart Mater. Struct. 19, 065032. (doi:10.1088/0964-1726/19/6/065032) 36. Ossmer H, Chluba C, Gueltig M, Quandt E, Kohl M. 2015 Local evolution of the elastocaloric effect in TiNi-based films. Shape Mem. Superelast. 1, 142–152. (doi:10.1007/s40830015-0014-3nn) 37. Lasalmonie A, Strudel J. 1986 Influence of grain size on the mechanical behaviour of some high strength materials. J. Mater. Sci. 21, 1837–1852. (doi:10.1007/BF00547918) 38. Gall K, Tyber J, Wilkesanders G, Robertson SW, Ritchie RO, Maier HJ. 2008 Effect of microstructure on the fatigue of hot-rolled and cold-drawn NiTi shape memory alloys. Mater. Sci. Eng. A 486, 389–403. (doi:10.1016/j.msea.2007.11.033)
