Superelasticity of Geometrically Graded NiTi Shape Memory Alloys

Abstract

A stress plateau with a strain of 5–8% normally occurs during the stress-induced martensite transformation (SIM) of NiTi shape memory alloys. Within the stress plateau, the correlation between the stress and strain is lost, which limits their application in certain fields which require accurate control of inelastic deformation. In order to address this limitation, a series of step-like NiTi samples with graded cross-sectional area were designed and fabricated. Multiple stress plateaus were achieved by varying the sample width and adjusting the number of steps; for instance, two and three stress plateaus were obtained in the samples with two and three steps, respectively. Also, linear force–strain response was obtained by changing gradually the width of the samples. The functional stability of the geometrically graded samples was significantly improved by incomplete recrystallization annealing (600 °C) followed by low-temperature (250 °C) aging treatment. The incompletely recrystallized specimens contained many dislocations and grain and sub-grain boundaries, which promoted the uniform precipitation of Ni₄Ti₃ nanoparticles during aging treatment. The homogeneously and densely dispersed Ni₄Ti₃ nanoparticles were able to strengthen the matrix considerably and prevent plastic activities during stress-induced martensite transformation. As a result, the functional stability of the geometrically graded NiTi samples was much improved. After aging at 250 °C for 120 h, all the samples showed a small residual strain of <1.0% after 20 loading–unloading cycles.

1. Introduction

NiTi shape memory alloys (SMAs) show unique superelasticity (SE) and shape memory effect (SME), which originate from the reversible thermoelastic martensitic transformation between a B2-structured austenite (A) phase and a B19′-structured martensite (M) phase [1,2,3,4]. These unique properties have attracted widespread interest, including for aerospace [5,6,7], biomedicine [8,9,10], and micro-electromechanical systems [11,12]. During the stress-induced martensite transformation (SIM), a stress plateau occurs over a large strain of 5–7% [13,14]. Within the stress-plateau, the correlation between the stress and strain is lost, which limits their application in certain fields that require accurate control of inelastic deformation, such as guide wires, micro-actuators, and orthodontics [15,16,17].

To address this limitation, many studies have shown that microstructurally [18,19], compositionally [20,21], or geometrically [22,23] graded NiTi alloys could be designed to achieve flexible control of the mechanical performance. Microstructurally graded NiTi alloys have been achieved through gradient aging [24] or annealing [25,26,27], which widens the transformation stress window (i.e., the stress difference between the start and the end of SIM). For example, a linear stress–strain response for SIM was achieved by introducing a gradient distribution of Ni₄Ti₃ precipitates through gradient aging in a Ni-rich NiTi wire [24]. A gradient superelastic behavior expands the stress window for the forward transformation to ~130 MPa. Laser surface annealing [18,28] has been reported as a promising method to tune locally the microstructure (e.g., the size of recrystallized grains) of NiTi plates to make functionally graded samples, in which transformation temperature and functional properties are tuned by varying the laser energy and scan rate. By alternately varying two sets of process parameters [29] or using a repetitive scanning strategy [30] to regulate the Ni evaporation, NiTi alloys with a compositional gradient were prepared using laser powder bed fusion techniques. Similarly, local heat treatment of Ni-rich NiTi wire by laser processing resulted in more pronounced vaporization of Ni than Ti, and thus a multi-stage SIM plateau could be achieved [31]. By welding NiTi wires with different compositions into a single unit along their length, a sample with graded transformation temperatures and graded critical stress for SIM was achieved [32].

Geometrically graded NiTi alloys are featured with a gradient change in the shape perpendicular [33] or parallel to the loading axis [34]. Compared with the abovementioned microstructurally or compositionally graded alloys, the fabrication of geometrically graded NiTi alloys is quite simple, since it is more feasible to change the shape of the samples. It has been reported that the stress window for the forward transformation could be expanded to ~220 MPa in a NiTi sample with trapezoidal shape [34]. Different stress gradients have been obtained by designing different geometries, such as linear strips, concave strips, and convex strips [35]. Geometrically graded NiTi structures with parallel design configuration were also designed [36]. During uniaxial tension, two transformation plateaus were observed in geometrically graded NiTi alloys [33,37,38], i.e., a flat stress plateau and an inclined one with a positive slope. These changes in plateau are attributed to the progressive propagation of the SIM perpendicular to the loading axis. However, inhomogeneous geometries lead to uneven deformations with locally concentrated stress, which are prone to induce plastic deformation. As a result, the geometrically graded NiTi alloys are more susceptible to functional degradation than the homogeneous specimens.

Precipitation strengthening has been widely utilized to improve the performance of metallic materials [39,40,41,42]. Ni₄Ti₃ nanoprecipitates can be introduced by aging at a temperature range from 200 °C to 550 °C in Ni-rich NiTi alloys [43,44,45]. However, a high-temperature solution treatment is typically required prior to the aging treatment [3], which results in the coarsening of the grains. It has been reported frequently that the distribution of Ni₄Ti₃ nanoprecipitates is not homogeneous in coarse grains [41,43,46]. Kolobova et al. [47] studied the precipitation behavior from a kinetic point of view. It was found that the precipitates are much smaller at the grain or sub-grain boundaries than in the grain interior. The heterogeneous distribution of Ni₄Ti₃ nanoprecipitates may undermine the effect of precipitation hardening [48,49]. Previous studies have shown that controlling the grain size can alter the distribution of the Ni₄Ti₃ particles. Low-temperature aging treatment of the fine-grained Ni-rich NiTi alloys (e.g., 1.7 μm [42,50]) has yielded a microstructure with homogeneously distributed Ni₄Ti₃ nanoprecipitates, and thus improved functional stability [42]. For coarse-grained NiTi alloys, the uniform precipitation of Ni₄Ti₃ nanoprecipitates can be achieved by introducing dislocation networks prior to aging treatment. Several approaches have been reported to introduce the dislocations, including repetitive temperature-induced phase transformation (i.e., thermal cycles between liquid nitrogen and boiling water [51]), stress-induced phase transformation, and ultrasonic shot peening [52]. As a result, the functional stability of the coarse-grained NiTi alloys has been improved.

In this work, a series of geometrically graded samples with different stepwise shapes was designed and fabricated. The effect of sample shape on the stress–strain response was investigated. The influence of incomplete recrystallization (600 °C) and subsequent low-temperature (250 °C) aging treatment on the functional stability of geometrically graded NiTi alloys was studied. This work will provide insight into the development of high-performance functionally graded NiTi alloys.

2. Materials and Methods

A commercial cold-rolled NiTi plate (Youyan Medical Co., Ltd., Beijing, China) with a nominal composition of Ti-50.8 at.% Ni and thickness of 0.5 mm was used. The accumulated cold deformation was around 10%. Geometrically graded samples were cut from the as-received NiTi plates along the rolling direction using wire electrical discharge machining. Figure 1 shows the detailed dimensions of the samples. A homogeneous specimen with a gauge width of 3 mm and length of 20 mm was used for comparison (Figure 1a). Figure 1b shows a two-step sample with step widths of 3 mm and 4 mm and a length of 10 mm for each step. Figure 1c shows a three-step sample with step widths of 3 mm, 3.4 mm, and 4 mm and a length of 10 mm for each stage. Figure 1d shows a sample with a continuous width transition from 3 mm to 4 mm over a length of 20 mm, named the N-step sample.

All the NiTi specimens were annealed at 600 °C for 30 min, followed by water quenching at room temperature. The characteristic transformation temperatures of the NiTi alloys, measured by differential scanning calorimetry (DSC), were −61.8 °C (M_f, martensite transformation finish temperature), −38.8 °C (M_s, martensite transformation start temperature), −22.7 °C (A_s, austenite transformation start temperature), and −6.5 °C (A_f, austenite transformation finish temperature). Low-temperature aging treatment was conducted at 250 °C for different durations from 1 h to 120 h. The microstructure of the annealed and aged samples was characterized using electron backscatter diffraction (EBSD) and transmission electron microscopy (TEM). EBSD testing was carried out using the NordlysMax3 system (Oxford Instruments, Abingdon, UK) on a JSM-7800 scanning electron microscope (JEOL, Tokyo, Japan). Samples for EBSD observation were electropolished in a solution of 21 vol% perchloric acid and 79 vol% acetic acid at room temperature. TEM characterization was performed using a Talos F200 microscope (ThermoFisher, Waltham, MA, USA), operated at 200 kV. The phase transformation behavior was studied using differential scanning calorimetry (DSC) in a DSC 3500 Sirius calorimeter (Netzsch, Selb, Germany), with a cooling/heating rate of 10 °C min⁻¹. The superelasticity was tested on a tensile test machine with a strain rate of 1.67 × 10⁻⁴ s⁻¹ at room temperature.

3. Results and Discussion

3.1. Microstructure

Figure 2 shows the EBSD crystallographic orientation map and inverse pole figures of the as-annealed Ti-50.8 at.% Ni plate. Figure 2a indicates that the deformed microstructure is largely annihilated. But several deformation bands can also be observed in some areas, indicating that recovery and recrystallization occurs when annealing at 600 °C, but the short annealing duration (30 min) was insufficient for full recrystallization. The average grain size is around 27 μm, which was manually determined using the linear intercept method described by Sutou et al. [53]. Figure 2b shows the inverse pole figures with reference to the rolling direction (RD), transverse direction (TD), and normal direction (ND) of the NiTi plate. The following relationship could be identified: RD//<110>_B2, ND//<111>_B2, indicating the presence of (111)//<110> texture in the as-annealed sample.

Figure 3a–f show the TEM results of the as-annealed Ti-50.8 at.% Ni plate. Figure 3a shows that some deformed (i.e., elongated) grains still remain in the as-annealed sample. A large number of dislocations are also observed inside the elongated grains. This result is consistent with the EBSD results; the short annealing duration (30 min) was insufficient for full recrystallization. Selected area electron diffraction (SAED) pattern (Figure 3b) indicates that the as-annealed sample is in B2 austenite phase at room temperature. Some ultrafine grains with size of <500 nm, as a result of recrystallization, are also observed in Figure 3c. Figure 3d shows that a high density of dislocation also exists at the grain boundaries and in the grain interior of the equiaxial grains. Some particles with size around 1 μm are detected in Figure 3e. The corresponding SAED pattern (Figure 3f) indicates that the particles are Ti₂Ni. This indicates that the low annealing temperature (600 °C) was insufficient to dissolve the Ti-rich particles.

Figure 3g–l show the TEM images of the as-annealed sample after aging at 250 °C for 120 h. Some elongated grains are observed in Figure 3g. Figure 3h–j show the high-resolution TEM (HRTEM) image of the NiTi matrix, and the corresponding fast Fourier transforming (FFT) pattern and schematic presentation of the FFT pattern, respectively. Figure 3i,j clearly show the 1/7{321}_B2 spots (marked by green arrow) for the Ni₄Ti₃ nanoprecipitates along the [111]_B2 zone axis. This indicates that Ni₄Ti₃ particles are formed in the matrix after long-term aging treatment, consistent with previous studies [42,50]. The nucleation of Ni₄Ti₃ precipitates is energetically favorable in the places of B2 austenite lattice distortion, dislocations, grain boundaries, and sub-grain boundaries [3,54]. The diffraction spots at 1/3<110>_B2 (marked by yellow arrow) are also identified, indicating that the aged samples are in the R-phase at room temperature [55]. Some Ti₂Ni particles are also observed in the NiTi alloy (Figure 3k,l). According to EDS point analysis in Figure 3k, the Ti content of the particle is about twice the Ni content. The HRTEM image and the corresponding FFT pattern (Figure 3f) indicate the Ti₂Ni structure. The Ti₂Ni particles may have formed during the casting process.

3.2. Transformation Behavior

Figure 4a–g show the transformation behavior of the NiTi samples after aging at 250 °C for different durations from 0 h to 120 h. The martensite (M_p) and R-phase (R_p) transformation peak temperatures are summarized in Figure 4h. The transformation heat of martensite and R-phase transformation are summarized in Figure 4i. The as-annealed sample shows a one-stage A→M transformation during cooling, and M→A transformation during heating. The latent heat of the martensitic transformation is 19.3 J/g (Figure 4i), which is less than that reported in the literature (21–29 J/g [56]). This is because the martensitic transformation is suppressed due to the incomplete recrystallization, where numerous dislocations and fine grains are present within the samples (Figure 3). Upon aging at 250 °C for short times (i.e., 1 h and 4 h), a two-stage A→R→M transformation is observed during cooling. Further increasing the aging time to 12 h-120 h, the martensite transformation peak disappears, and only A→R transformation during cooling is detected.

According to Figure 4h, as the aging time increases from 1 h to 4 h, the M_p temperature decreases, while the R_p temperature increases. The reason behind this phenomenon is two-fold. First, a strain field is formed around the coherent Ni₄Ti₃ nanoprecipitates, which inhibits the martensitic transformation. The strength of strain field increases with the growth of Ni₄Ti₃ nanoprecipitates, leading to greater inhibition of the martensitic transformation. This leads to a reduction in martensitic transformation temperature (e.g., M_p) and the latent heat of the martensitic transformation (Figure 4i). Secondly, the strain field generated by the coherent Ni₄Ti₃ nanoprecipitates has a negligible effect on the R-phase transformation due to its small transformation strain associated with the A→R transformation (around 1%) [3]. Therefore, the R-phase transformation temperature is mainly related to the Ni content in the matrix. The formation of Ni₄Ti₃ nanoparticles causes the depletion of Ni in the matrix, leading to an increase in the R-phase transformation temperature. The latent heat of the R-phase transformation also increases (Figure 4i). After aging for 12 h, only the A→R peak temperature is detected during cooling due to the further suppression of the martensitic transformation. The R→M transformation takes place continuously over a wide temperature range, resulting in a non-detectable transformation peak temperature. However, an evident M→R peak temperature is observed during heating.

It has been frequently reported that multi-stage martensitic transformation with two-stage R-phase occurs in the coarse-grained (e.g., average grain size > 20 μm) Ni-rich NiTi alloys after low-temperature aging treatment [48]. The multi-stage transformation is mainly due to the inhomogeneous distribution of Ni₄Ti₃ precipitates between the vicinity of grain boundaries and the grain interior. However, in this work, the multi-stage transformation is absent, indicating a homogeneous distribution of the Ni₄Ti₃ nanoprecipitates. This is mainly due to the presence of dislocations and ultra-fine grains, which helps the precipitation of the Ni₄Ti₃ particles in the grain interior. As a result, the homogeneous aging microstructure could be achieved.

3.3. Superelasticity

Figure 5 shows the tensile loading–unloading curves of the homogeneous and geometrically graded NiTi samples (with a tensile strain of 10%). Figure 5a shows the deformation behavior of the homogenous sample. During loading, elastic deformation of the austenite phase (stage I) occurs first, followed by the stress-induced martensite (SIM) transformation (stage II), which produces a stress (or force) plateau. Once the SIM is complete, the sample undergoes the elastic deformation of the martensite (stage III). Upon unloading, elastic strain is recovered and the stress-induced martensite reverse transformation occurs, producing a plateau with lower force level. A small residual strain is generated, caused by the plastic activities during A↔M transformation.

Figure 5b shows the mechanical behavior of the two-step sample (10 mm for each step), where sections a and b represent the regions with widths of 3 mm and 4 mm, respectively. Upon loading, SIM starts at the narrow region giving rise to the first force plateau (stage II). After the completion of the SIM in stage II, section a undergoes the elastic deformation of stress-induced martensite (stage III), in which the force increases with the increase in strain. When the force reaches a critical value, the SIM in the wide region is triggered, resulting in a second force plateau (section IV). Therefore, two force plateaus are obtained by changing the width of the samples.

Figure 5c shows the mechanical behavior of the three-step NiTi sample (widths of 3, 3.4, and 4 mm). Similar to the two-step specimen, SIM occurs sequentially in three different regions as the applied force increases, resulting in three plateaus, namely, sections II, IV, and VI. However, a large residual strain is observed upon unloading, since the force required to induce the SIM in section VI (i.e., widths of 4 mm) exceeds the critical value for plastic deformation at its narrowest part (i.e., widths of 3 mm). The deformation behavior of the N-step sample is given in Figure 5d. As the loading force increases, SIM occurs successively in the trapezoidal region (stage II). The force required for the transformation gradually increases from the narrowest side (3 mm) to the widest side (4 mm), resulting in a linear force–strain relationship over a large strain (from 1% to 8%). Upon unloading, most of the strains are recovered.

Figure 6 schematically shows the sequence that triggers the SIM during loading in the homogeneous and geometrically graded NiTi samples. Figure 6a shows that the homogeneous specimen gradually undergoes SIM from austenite to martensite upon loading. The two-step specimen produces a two-step phase transformation (Figure 6b): (i) SIM occurs first in the narrow region at low loading force, and (ii) SIM occurs in the wide region at high loading force. For the N-step specimen, the cross section does not change abruptly but transitions progressively. The SIM occurs successively from the narrow side to the wide side (Figure 6c), resulting in a linear stress–strain response. This is consistent with previous studies [34,37,38].

As discussed above, different sample geometries result in different stress–strain responses, which offers the possibility to control the nucleation location and propagation of the SIM. This improves the controllability and flexibility of the geometrically graded NiTi samples as an actuator.

Figure 7 shows the superelastic stability of the homogeneous and geometrically graded NiTi samples after annealing. All the specimens were subjected to cyclic loading–unloading for 20 cycles with a strain of 8%. All the samples show a fast degradation of the functional stability. This is because the cyclic A↔M transformation is accompanied by plastic slip, resulting in the accumulation of residual strain. Figure 7a shows that the residual strain of homogeneous specimens is 3.4% after 20 cycles. The residual strain of the two-step sample is 4.1% (Figure 7b). The three-step sample has a residual strain of 4.2% after 20 cycles (Figure 7c). As compared with the uniform specimen, the larger residual strain of the geometrically graded specimen is attributed to the locally concentrated deformation at the narrower region that induces more plastic deformation. However, the room temperature superelasticity of the N-step sample exhibits a residual strain of 3.5% (Figure 7d). This is mainly due to the gradual transition of the cross section that homogenizes the overall deformation and avoids the locally concentrated plastic activities. In addition, the stress–strain response (i.e., the stepwise shape of the loading–unloading curve) of geometrically graded NiTi samples disappears due to the plastic activities accompanying the cyclic A↔M transformation. For instance, the multiple force plateaus of the two-step and three-step samples disappear after 2 cycles. The results in Figure 7 indicate that geometrically graded samples are more susceptible to functional degradation than homogeneous specimens. Enhancement of the strength of the matrix is highly required to suppress plastic deformation, thus improving their functional stability.

Figure 8 shows the superelastic curves of the homogeneous and geometrically graded NiTi samples after aging at 250 °C for different durations (1, 4, 12, 24, 48, and 120 h). For homogeneous samples, aging for 1 h has a negligible effect on the functional stability, and the residual strain remains 2.8% after 20 loading–unloading cycles (Figure 8a). With increasing aging time, the residual strain gradually decreases. For example, the accumulated residual strain decreases to 1.0% as the aging time increases to 120 h (Figure 8f). The improvement in superelasticity is attributed to the formation of homogeneously and densely dispersed Ni₄Ti₃ nanoprecipitates that hinders the plastic deformation during cyclic A↔M transformation.

Similarly, the functional stability of the geometrically graded specimen is also much improved with the increase in aging time (Figure 8g–x). Although the as-annealed geometrically graded specimens are more prone to introduce plastic deformation (Figure 7), the functional stability could be improved to a level comparable to that of the homogeneous specimens after long-term aging treatment (e.g., 120 h). It is suggested that the accumulated residual strains of all the geometrically graded samples were less than 1% after 20 loading–unloading cycles after aging treatment at 250 °C for 120 h (Figure 8l,r,x).

Figure 8 also shows that the aging treatment helps to maintain the stepwise shape of the cyclic loading–unloading curves of the geometrically graded NiTi samples. For the specimens with short-term aging treatment (e.g., 1 h and 4 h), plastic deformation is generated and accumulated rapidly during cyclic A↔M transformation due to less precipitation of Ni₄Ti₃ nanoprecipitates. This results in the disappearance of the multiple force plateaus in the geometrically graded specimens after a few loading–unloading cycles. For example, the three-step specimen aged for 4 h exhibits three force plateaus in the first 5 cycles, while the plateau disappears in the subsequent 6 to 20 cycles (Figure 8n). With increasing aging time, the uniform dispersion of Ni₄Ti₃ nanoprecipitates suppresses the plastic activities, thereby increasing the number of cycles that can maintain the multiple force plateaus. After aging treatment for 120 h, the stepwise shape of the loading–unloading curves of the two-step and three-step geometrically graded samples is still observed after 20 loading–unloading cycles (Figure 8l,r). The above results indicate that long-time aging treatment not only improves functional stability but also helps to maintain the stepwise stress–strain response of geometrically graded samples.

Figure 9 summarizes the total residual strain of the different geometrically graded samples. For the as-annealed and the short-term aged (i.e., 1 h and 4 h) specimens, a large residual strain is generated in the first few cycles due to the rapid accumulation of plastic deformation. In subsequent cycles, the residual strain increment is gradually reduced. However, the overall residual strain increment of long-term-aging-treated (i.e., 120 h) specimens is small, which is attributed to the inhibition of plastic activities by the uniform precipitation of Ni₄Ti₃ nanoparticles.

4. Conclusions

In this study, a series of geometrically graded NiTi samples were designed and fabricated. Multiple force plateaus were achieved by varying the geometries of each step. The superelastic stability of geometrically graded samples was improved through a combination of incomplete recrystallization treatment and long-term low-temperature aging treatment. The main conclusions can be drawn as follows:

(1)

Altering the width and the number of steps of the geometrically graded samples can adjust their stress–strain response, resulting in multiple force plateaus over a large strain of 8%.

(2)

Geometrically graded samples are more susceptible to functional degradation than homogeneous specimens due to the locally concentrated deformation at the narrower region that induces more plastic deformation.

(3)

Incomplete recrystallization treatment at 600 °C for 30 min leads to a high density of dislocations and grain and sub-grain boundaries in the specimen, which are energetically favorable for the nucleation of Ni₄Ti₃ precipitates during the subsequent low-temperature aging process. This significantly improves the functional stability of the geometrically graded samples. All the geometrically graded samples aged at 250 °C for 120 h show good superelasticity, with a total residual strain of <1.0% after 20 loading–unloading cycles.

(4)

The stepwise shape of the stress–strain curve (i.e., stress–strain response) of the geometric gradient samples can be maintained after long-term low-temperature aging treatment (i.e., 120 h) due to the suppression of plastic activities.