Superelastic Properties of Aged FeNiCoAlTaB Cold-Rolled Shape Memory Alloys

: In the present study, microstructure and cyclic tensile tests were used to measure the su-perelastic responses of Fe 40.95 Ni 28 Co 17 Al 11.5 Ta 2.5 B 0.05 (at.%) shape memory alloys after 97% cold rolling. Cold-rolled samples underwent annealing heat treatment (1250 °C/1 h) followed by quenching in water or aging heat treatment (700 °C/6 h and 700 °C/12 h) followed by quenching in water. The microstructure results showed that the average grain size increased from 210 µm to 1570 µm as annealing times increased from 0.5 h to 1 h. X-ray diﬀraction (XRD) spectra for FeNiCoAlTaB (NCATB) showed that in cold-rolled alloys after solution, the strong peak was in the face-centered cubic ( γ, FCC) <111> structure. In aged samples, a new peak ( γ’, FCC) emerged, the intensity of which increased as aging times rose from 6 to 12 h. Transmission electron microscope (TEM) images showed that the average precipitate size was around 10 nm in 700 °C/6 h specimens and 18 nm in 700 °C/12 h specimens. The precipitate was enriched in Ni, Al, and Ta elements and exhibited an L1 2 crystal structure. Tensile samples aged at 700 °C for 6 and 12 h exhibited recoverable strains of 1% and 2.6%, respectively, at room temperature. Digital image correlation (DIC) results for the sample aged at 700 °C for 12 h showed that two martensite variants were activated during the superelastic test. Such variants can form corresponding variant pairs (CVPs), which promote tensile deformation. The tensile sample exhibited a gradual cyclic degradation, and a large irrecoverable strain was observed after the test. This irrecoverable strain was the result of residual martensite, which was pinned by dislocations.


Introduction
Shape memory alloys (SMAs) are characterized by (1) superelasticity and (2) the shape-memory effect.SMAs are now used in many industrial applications, including automobile manufacturing, the aerospace industry, and biomedical engineering [1].Recently, SMAs have attracted much attention for applications in aerospace and space fields.In aerospace applications, SMAs can be usefully applied in the design of morphing wings, tailoring of the inlet geometry of propulsion systems, optimization of variable geometry chevron, and reduction in power consumption.Space applications include the vibrator, actuators, satellites, and solar sails [2].Ferrous superelastic alloys are valued for their (1) low material cost, (2) excellent cold workability, and (3) high ductility and strength [3].Therefore, Fe-based SMAs are now regarded as promising alternatives to NiTi SMAs in aerospace and space applications.Recent studies using different methods of thermome-chanical processing have shown that FeNiCoAlXB (X = Ta, Ti, or Nb) SMAs possess recoverable strains of superelasticity, which are higher than 4% at room temperature [4][5][6][7][8][9][10][11][12].In the systems of such alloys, the crystal structures of austenite (γ) and martensite are facecentered cubic (FCC) and body-centered tetragonal (BCT), respectively.The structure of the precipitate is the Ni3Al-type γ′ (L12) phase.Good superelastic properties in an FeNiCoAlXB alloy system are derived from the following: large grain size; strong texture in <100> orientation; and appropriate size of precipitate [3].
Zhang et al. [13,14] studied the effects of aging heat treatment and proportions of cold deformation on grain morphology and superelasticity in an FeNiCoAlTaB alloy.When aging time or temperature was increased, the Ta element was first generated at the triple-junction boundaries and along the grain boundaries, preferring to accumulate at boundaries with high angles.After Ta was sufficiently accumulated at the boundaries, the β phase (B2) preferred to form a Ta-rich region.Ta and β phases not only increased transformation temperatures but also reduced the ductility of samples.An aged sample (600 °C-72 h) with a high proportion of low-angle boundaries was found to exhibit a recoverable strain of 2.5% under tension testing.
The impacts of solution temperature (1150 °C, 1200 °C, 1250 °C or 1300 °C) on microstructure and superelasticity in an FeNiCoAlTaB alloy were studied by Zhao et al. [15].Microstructure results revealed the presence of intermetallics in samples that were treated at 1150 °C and 1200 °C.When sample solutions were treated at 1250 °C or 1300 °C, a complete austenitic parent was found in each sample after quenching.However, when the solution temperature was 1300 °C, cracks were generated at the sample surface.A strong rolling texture of the brass type was formed using 98.5% cold-rolling at 1250 °C for 1 h.The cyclic tensile stress-strain result showed that the tensile sample aged at 600 °C/72 h exhibited 1.6% recoverable strain and 885 MPa tensile strength.
Fu et al. [16] used directionally solidified methods (with drawing velocity equal to 0.5 mm/min) to determine the grain size and orientation of FeNiCoAlTaB SMAs.Cyclic tensile tests were carried out on samples before aging and after aging for 1, 3, 6, 9, 12, and 24 h.Superelastic behavior was observed in the samples aged for 6, 9, and 12 h.The recoverable strains in the 6, 9, and 12 h tensile samples were found to be 1.7%, 1%, and 1.5%, respectively.Microstructure test results showed that, in the sample aged for 1 h, there was no martensitic phase transformation.The martensitic start temperature (Ms) and austenite finish temperature (Af) both increased as aging heat treatment times increased from 6 h to 12 h.Precipitate diameters of samples aged for 1 h, 6 h, and 12 h were approximately 2 nm, 13 nm, and 20 nm, respectively.
The superelastic properties of FeNiCoAlTaB polycrystalline wires were first investigated by Choi et al. [17].Wires which were solutionized at 1300 °C for 1 h were found to exhibit a bamboo-like grain structure.Superelastic wires aged for 72 h at 600 °C exhibited a maximum 7% recoverable strain, 200-250 MPa stress hysteresis, and 1.2 GPa strength.Atom probe tomography (APT) revealed that precipitate volume fraction was 30% and size ranged between 3 and 10 nm.In addition, the superelastic behavior of aged FeNiCoAlTaB polycrystalline wire was characterized by high cyclic stability.The wire exhibited good superelasticity until the 37th cycle, before failure occurred on the 38th cycle.Finally, constant-strain testing at different temperatures revealed that the stress-temperature slope was 5.4 MPa/°C in a temperature range from −40 to 30 °C, and the addition of boron to the wires effectively suppressed the β-NiAl formation.
Zhang et al. [18] investigated recrystallization textures, grain morphologies, and mechanical properties in FeNiCoAlTaB and FeNiCoAlCrB alloys.They found that the addition of Cr to an FeNiCoAl-based system decreased the stacking fault energy.The stacking fault energy also influenced grain orientation in samples that underwent cold rolling and annealed heat treatment.When the stacking fault energy was low, it was difficult for dislocations to contain stacking faults.As a result, a strong recrystallization texture in <100> orientation could be achieved by adding the Cr element to the alloy system.In FeNiCoAlCrB alloys, which underwent 70% cold rolling and annealed heat treatment at 1300 °C for 1 h, the grain size was found to be around 420 µm, and the proportion of lowangle boundaries was 39%.
Czerny et al. [19] compared the microstructures and superelastic responses of FeNiCoAlTa and FeNiCoAlTaB single crystals to investigate the effect of the addition of boron to these alloy systems.For FeNiCoAlTa, compressive results revealed that, when annealing times were longer than 10 h, the plastic deformation of austenite produced a permanent strain, resulting in a large irrecoverable strain.However, in NCATB single crystals which were aged for longer than 10 h, two martensite variants were observed during compressive testing.Transmission electron microscope (TEM) images revealed the presence of FeNiCoAlTaB single crystals after aging at 700 °C for variable times (0.5, 1, 5, 10, and 24 h).In addition, precipitate sizes in FeNiCoAlTaB were found to be 3, 3.5, 4.8, 6, and 6.5 nm after aging for 0.5, 1, 5, 10, and 24 h, respectively.However, under the same aging conditions, precipitate sizes in FeNiCoAlTa were found to increase more quickly than in FeNiCoAlTaB.Finally, the addition of boron was found to inhibit the formation of the brittle β phase and reduce the size of precipitates.
Wójcik et al. [20] conducted superelastic cycling experiments at −216 °C to determine compressive recoverable strain in FeNiCoAlTa and FeNiCoAlTaB single-crystal samples.Each of the two materials was subjected to annealing at 1300 °C /1 h, followed by aging at 700 °C/0.5 h.The FeNiCoAlTa single crystal exhibited a recoverable strain of 15% under compression at −216 °C.In addition, compressive stress-strain results revealed stress-drop behaviors during the cyclic test.In the case of the FeNiCoAlTaB single crystal, the recoverable strain was found to be around 14.3% at −216 °C.The addition of boron increased stress values for each onset of stress.For the FeNiCoAlTa single crystal, under the same aging condition (700 °C/0.5 h), the volume fraction of the precipitate was 2%, and the size of the precipitate was around 3 nm.The experimental results also showed that the addition of boron to single crystals inhibited the diffusion of precipitates during aging heat treatment.Finally, TEM results showed that, in FeNiCoAlTa single crystal samples, the volume fraction and size of precipitate was around 4% and 5 nm, respectively.
The gradual cyclic degradation and large irrecoverable strain have been reported by Zhang et al. [14,21] and Fu et al. [16] for the specimens aged at 600 °C and 700 °C during the superelastic test.In their study, they did not investigate how microstructures affect the gradual cyclic degradation and large irrecoverable strain of superelasticity.As a result, the present work aims to answer these questions by focusing on the superelastic properties of Fe40.95Ni28Co17Al11.5Ta2.5B0.05 (at.%) cold-rolled alloys so that their microstructural and superelastic properties could be determined.The digital image correlation (DIC) was used to observe each cycle of loading to investigate the martensite morphology during the superelastic test.The TEM was used to observe the size of the precipitate in two different aging conditions.

Materials and Methods
The Fe40.95Ni28Co17Al11.5Ta2.5B0.05 polycrystalline in at.% (NCATB) SMA was prepared by induction melting and cast into an alloy bar.Wire electrical discharge machining (EDM) was then used to cut the bar into several blocks, with each block having a length, width, and thickness of 100 mm, 25 mm, and 25 mm, respectively.The block was cold rolled (CR) with a reduction ratio of 97% (CR97) to a thickness of 0.75 mm (cold-rolling reduction is defined as ε = (t0 − t1)/(t0) × 100%, where t0 and t1 were the thicknesses of the sheets before and after cold rolling, respectively).Tensile samples were cut from the 97% cold-rolled sheets, with gauge dimensions of 0.75 mm × 3 mm × 8 mm.Each tensile sample was then sealed in a quartz tube and annealed at 1250 °C for 1 h, followed by water quenching.The annealed samples were aged at 700 °C for 6 and 12 h (PXL700 °C-6 h and PXL700 °C-12 h).
A tension experiment was used to characterize superelastic responses at room temperature.Tensile samples were successively loaded with strain, with increments of ap-proximately 0.5% strain, during each cycle.Strain levels were increased during each loading/unloading cycle until sample failure occurred.During tensile tests, the non-contact method of digital image correlation (DIC) was used to measure deformation in tensile samples.The commercial VIC-2D version 7 software was used to analyze strain distribution in tensile samples during each cycle.
The hardness of CR samples under different heat-treatment conditions was measured using Vickers hardness-testing machines.X-ray diffraction (XRD) measurements were used to determine the phases of the CR97 samples.CR97 samples were subjected to annealing heat treatment at 1250 °C for 1 h, PXL700 °C-6 h and PXL700 °C-12 h.
In order to investigate the grain morphology, such as grain size and texture in NCATB, the CR sample was annealed at 1250 °C for 0.5 h and 1 h, respectively.The recrystallization texture and grain growth of samples were characterized by electron backscatter diffraction (EBSD).Inverse pole figures (IPFs) were used to show grain orientation distribution for three different sample directions: rolling direction (RD), normal direction (ND), and transverse direction (TD).An electropolishing solution was used consisting of 90% C2H5OH and 10% HClO4.The nomenclature of this study is shown in Table 1.

Hardness in NCATB
Table 2 presents hardness results for CR97 NCATB samples under various thermomechanical processing conditions.The average value for the CR97 sample was 423 HV.After annealing, this value was 225 HV.For samples subjected to aging, hardness values were 460 HV and 486 HV for PXL700 °C-6 h and PXL700 °C-12 h specimens, respectively.Hardness values increased with increasing aging times because of precipitation hardening.

Crystal Structure in NCATB
Figure 1a presents room-temperature X-ray diffraction spectra recorded for alloy samples under different thermomechanical processing conditions.It can be seen that austenite (γ) phases were observed only in the CR97 sample and after annealing at 1250 °C for 1 h.The corresponding diffraction planes were (111), (200), and (220).Figure 1b shows XRD measurement ranges with a diffraction angle (2 theta) between 42 and 46 degrees.In the figure, (111)γ with strong intensity can be observed for the annealed sample without aging.After aging at 700 °C for 6 and 12 h, a new peak with (111)γ' and a higher diffraction angle appeared between 44 and 45 degrees.This new peak resulted from the γ′ (L12) precipitates.As aging times increased from 6 to 12 h, the intensity of the new peak was found to increase when the (111)γ peak decreased.A similar observation was made in a previous study by Zhou [21] in which an NCATB alloy aged at 600 °C for 96 h exhibited a strong peak intensity of (111)γ'.Based on the TEM results for samples aged for 6 h, the average precipitate size was found to be around 10 nm.For the sample aged at 700 °C for 6 h, the diffraction pattern results identified the austenite phase as having an FCC structure and the precipitate an L12 structure in the (110) zone axis.Figure 2c,d show bright-field TEM images of PXL700 °C-12 h samples.A high density of precipitates was distributed in these samples because of high magnetization, which made it difficult to obtain high-resolution bright-field TEM images.Based on the measurement of the Figure 2d bright-field TEM image, we found that when the aging time was prolonged to 12 h, the size of the precipitate increased dramatically to an average of 18 nm.Previously, Fu et al. [16] found that after aging for 6 h, the size of precipitates was around 13 nm, but when aging time was prolonged to 12 h, precipitates attained a size of around 16 nm.The coherency of the precipitate with martensite is dependent on the size of the precipitate.The authors of [4] reported that small-sized precipitates were able to maintain coherency not only with austenite but also with martensite during martensitic transformation [4].In other studies of the FeMnAlNi alloy involving average precipitate sizes as large as 10 nm, it has been found that the precipitate maintains coherency with martensite, a phenomenon attributed to the elastic distortion of the precipitate [22].Figure 2e

Development of Texture in NCATB
Figure 3a,b show EBSD IPFs for NCATB (CR97) after annealing at 1250 °C for 0.5 h and 1 h, respectively.It can be seen that the grain orientations of specimens in the rolling direction were along the <100> orientation.For CR97 samples subjected to annealing for 0.5 h, the maximum value of texture intensity was 13.98 mud.When the annealing time was increased to 1 h, the maximum texture intensity slightly decreased to 10.94 mud.In FeNiCoAlNbB SMAs, the recrystallization texture was along the <110> orientation, which is in line with the results reported by Omori [5].

Grain Morphology in NCATB
In the present study, it was important to increase the volume fraction of low-angle boundaries (LABs) to avoid the formation of β-NiAl phases along grain boundaries.Figure 4a,b show that the proportions of low-angle boundaries (<15 degrees) in annealing for 0.5 h and 1 h samples were 15.4% and 39.5%, respectively.For the FeNiCoAlXB (X = Ta, Ti, and Nb) alloy systems, the proportion of low-angle boundaries was 11% for FeNiCoAlNbB (CR98.5)and 36% for FeNiCoAlTiB (CR98.5).These values were lower than the corresponding figure of almost 60% previously reported by Tanaka [4].
In Fe-based SMAs, superelastic properties are influenced by average grain size.Consequently, in the present study, data relating to average grain size were calculated based on the EBSD results.Figure 4c,d visually summarize average grain sizes in NCATB alloys under different annealing conditions.It can be seen that after annealing for 0.5 h, the average size of grain was 210 µm.When the annealing time increased to 1 h, the average grain size increased dramatically to 1570 µm.The proportions of grains with a size of less than 500 µm were around 89% and 75% under annealing for 0.5 and 1 h, respectively.In order to understand the grain growth mechanism, a method for calculating average grain size was adopted following the work of Toth et al. [23].Figure 4e illustrates the relationship between grain size (d) and annealing time (T) when T = 30 and 60 min.Figure 4f shows relative grain size development with annealing time.A classical kinetic approach for describing the phenomenon of gradual coarsening was previously suggested by Burke and Turnbull [24][25][26], who used the following equations: In the above equations, n values of 2 and 1/2 are used for normal grain growth and abnormal grain growth, respectively.It can be seen in Figure 4e that the data point is close to the n = 2 line when the annealing time is 30 min, with grain growth close to the abnormal type.However, as the annealing time approaches 60 min, the grain growth changes to the normal type.Figure 4f shows that the relative grain size (210 µm/750 µm) was below 1 for the first 30 min, but the relative grain size (1570 µm/750 µm) increased to 2.1 when the annealing time reached 60 min.Zhang et al. [14] proposed that during superelastic testing, relative grain size should have a value above 1 to reduce grain constraint and promote superelastic behaviors.In the present study, the relative grain size was above 1 when the annealing time was 1 h.We may say, then, that tensile samples should be subjected to a one-hour annealing condition to reduce grain constraint and promote superelastic properties.

Superelastic Properties of NCATB
The superelastic properties of NCATB include critical stress, as well as levels of recoverable and irrecoverable strain.In Figure 5a, εrec indicates the recoverable strain, εirrec the irrecoverable strain, and σc the critical stress for stress-induced martensite transformation.Figure 5b,c show superelastic responses for PXL700 °C-6 h and PXL700 °C-12 h samples, respectively, at room temperature.The cross indicates the fracture point.Recoverable and irrecoverable strains expressed as functions of applied strain levels are summarized in Figure 5d,e for PXL700 °C-6 h and PXL700 °C-12 h samples, respectively.It can be seen that, for PXL700 °C-6 h, the maximum recoverable strain was 1%, and the maximum irrecoverable strain was 0.5%.The sample experienced failure during the 1.5% strain test, and the fracture stress level was 795 MPa. Figure 5e shows that the maximum recoverable strain was 2.6% when the level of applied strain level was 4.8%.The sample failed during the next cyclic test, when the strain level was 5.5% and the fracture strength was 675 MPa.Moreover, the critical stresses for PXL700 °C-6 h and PXL700 °C-12 h samples were found to be 765 MPa and 560 MPa, respectively.These results indicate that the transformation temperature increased its value as aging time increased from 6 to 12 h so that less stress was required to induce martensitic transformation.These results are also in agreement with previously reported superconducting quantum interference device (SQUID) results, which indicated austenite finish temperature (Af) values of -73 °C and -50 °C for 700 °C-6 h and 700 °C-12 h aged samples, respectively [27].The relationship between transformation temperatures and critical stress is summarized in Figure 5f.Recorded values for critical stress, fracture stress, maximum recoverable strain, and ductility for both PXL700 °C-6 h and PXL700 °C-12 h samples are summarized in Table 3.The martensitic transformation of the NCATB involves a change from the martensite phase (BCT) to the austenite phase (FCC).According to the Bain distortion theory, two martensite variants are able to accommodate tensile deformation, but only one martensite variant is able to accommodate compressive deformation [28].In order to understand the mechanism of tensile deformation, a digital image correlation (DIC) was performed for each cyclic test of PXL700 °C-6 h and PXL700 °C-12 h samples.In the DIC measurement results, the numbers 1, 2, and 3 represent loading, loading to target strain levels, and postunloading, respectively.Figure 6a-c show DIC results for the PXL700 °C-6 h sample under each cyclic test.It can be seen that the sample only exhibited a full recovery in the first and second cycles of tension.In the third cycle of tension, the sample was fractured, and a single martensite variant was observed in the DIC result.These results indicate that the critical stress was higher than the fracture stress during the third cycle.As a result, the sample experienced failure.In the first, second, and third cycles, the sample exhibited a full recovery, and residual martensite was not observed after testing.In the fourth cycle, DIC results revealed that two variants of martensite were formed in the tensile sample during the superelastic test.As a result, two martensite variants could contribute equally to tensile deformation.Two martensite variants can easily form corresponding variant pairs (CVPs).The formation of a CVP enables internal twinning to accommodate tensile deformation and provides a more coherent interface with the austenite.The tensile result showed a small and smooth increase in stress with increased strain during the test.In the seventh cycle, as illustrated in Figure 7g, the DIC result revealed a large volume fraction of residual martensite in the sample after the test.This was because the martensite interaction became intense, so that martensite was retained in the sample.As a result, the sample exhibited gradual cyclic degradation due to the retained martensite.Such residual martensite has been found to be due to the generation of dislocations [22].Dislocations are generated at the austenitemartensite interface; these pin the martensitic phase and prevent it from reversing back to austenite.As a result, a large volume fraction of the martensitic phase is retained in the tensile sample, and this contributes to irrecoverable strain.
The precipitate size for PXL700 °C-12 h was found to be around 18 nm.Previous studies like that of Tanaka et al. [4] indicated that precipitates with a size of about 3 nm have a positive effect on the recoverable strain of superelasticity.If the precipitate size is large, it may lose coherency and affect the nucleation of martensitic phases.The loss of coherency between precipitates and martensite accompanies the generation of misfit dislocations at the precipitate/matrix interfaces [8,29].As a result, the large size of the precipitate leads to more barriers to the transformation, and the interaction between martensite variants is accelerated.If the precipitates cannot transform, they cannot contribute to recoverable strain.In the present study, the maximum recoverable strains in the PXL700 °C-6 h and PXL700 °C-12 h samples were found to be 1% and 2.6%, respectively.In addition, the size of the precipitate in the PXL700 °C-12 h sample was larger than that previously reported [3].The large size of the precipitate inhibited not only the forward but also the reversed martensitic transformation during the superelastic test.As a result, the tensile result revealed gradual cyclic degradation.

Conclusions
In the present study, the superelastic properties of cold-rolled FeNiCoAlTaB polycrystalline alloys were investigated for two aging conditions (PXL700 °C-6 h and PXL700 °C-12 h).In conclusion, the following results may be stated: 1.Under an annealing condition of 0.5 h, for cold-rolled NCATB SMAs, the average size of grain was 210 µm.When the annealing time increased to 1 h, the average grain size increased to 1570 µm.Moreover, the fraction of low-angle boundaries increased from 15.4% to 39.5% when the annealing time was extended from 0.5 h to 1 h.Normal grain growth was exhibited when the annealing time was 0.5 h.When the annealing time was increased from 0.5 h to 1 h, the mechanism of grain growth changed to abnormal growth.2. TEM results indicated that the sizes of the precipitates were between 10 nm for the PXL700 °C-6 h sample and around 18 nm for the PXL700 °C-12 h sample.The precipitates had an L12 crystal structure, and the austenite had an FCC structure.The precipitates were also found to be enriched with Ni, Ta, and Al. 3. A superelastic strain of 3% and a maximum tensile elongation of 6% were recorded for the NCATB polycrystalline alloy with a high proportion of low-angle grain boundaries and large grain sizes in a tensile test at room temperature.PXL700 °C-6 h exhibited higher critical stress values than PXL700 °C-12 h due to its lower transformation temperatures.The DIC results revealed that two martensite variants could activate during the superelastic test, with the recoverable strain decreasing due to retained martensite.Dislocations pinned the martensite phase to prevent it from reversing back to austenite.As a result, the large volume fraction of martensite was retained in the tensile sample, and this contributed to irrecoverable strain.

Figure 2a ,
Figure 2a,b show high-resolution TEM images and diffraction patterns of precipitate in the NCATB for PXL700 °C-6 h samples.Yellow circles show the positions of precipitates.Based on the TEM results for samples aged for 6 h, the average precipitate size was found to be around 10 nm.For the sample aged at 700 °C for 6 h, the diffraction pattern results identified the austenite phase as having an FCC structure and the precipitate an L12 structure in the (110) zone axis.Figure2c,dshow bright-field TEM images of PXL700 °C-12 h samples.A high density of precipitates was distributed in these samples because of high magnetization, which made it difficult to obtain high-resolution bright-field TEM images.Based on the measurement of the Figure2dbright-field TEM image, we found that when the aging time was prolonged to 12 h, the size of the precipitate increased dramatically to an average of 18 nm.Previously, Fu et al.[16] found that after aging for 6 h, the size of precipitates was around 13 nm, but when aging time was prolonged to 12 h, precipitates attained a size of around 16 nm.The coherency of the precipitate with martensite is dependent on the size of the precipitate.The authors of[4] reported that small-sized precipitates were able to maintain coherency not only with austenite but also with martensite during martensitic transformation[4].In other studies of the FeMnAlNi alloy involving average precipitate sizes as large as 10 nm, it has been found that the precipitate maintains coherency with martensite, a phenomenon attributed to the elastic distortion of the precipitate[22].Figure2epresents high-angle annular dark field scanning transmission electron microscopy (HAADF-STEM) and elemental distribution energy dispersive spectroscopy (EDS) maps.Based on the results illustrated in the figure, we determined that the precipitates were mainly composed of Ni, Al, and Ta.

Figure 2 .
Figure 2a,b show high-resolution TEM images and diffraction patterns of precipitate in the NCATB for PXL700 °C-6 h samples.Yellow circles show the positions of precipitates.Based on the TEM results for samples aged for 6 h, the average precipitate size was found to be around 10 nm.For the sample aged at 700 °C for 6 h, the diffraction pattern results identified the austenite phase as having an FCC structure and the precipitate an L12 structure in the (110) zone axis.Figure2c,dshow bright-field TEM images of PXL700 °C-12 h samples.A high density of precipitates was distributed in these samples because of high magnetization, which made it difficult to obtain high-resolution bright-field TEM images.Based on the measurement of the Figure2dbright-field TEM image, we found that when the aging time was prolonged to 12 h, the size of the precipitate increased dramatically to an average of 18 nm.Previously, Fu et al.[16] found that after aging for 6 h, the size of precipitates was around 13 nm, but when aging time was prolonged to 12 h, precipitates attained a size of around 16 nm.The coherency of the precipitate with martensite is dependent on the size of the precipitate.The authors of[4] reported that small-sized precipitates were able to maintain coherency not only with austenite but also with martensite during martensitic transformation[4].In other studies of the FeMnAlNi alloy involving average precipitate sizes as large as 10 nm, it has been found that the precipitate maintains coherency with martensite, a phenomenon attributed to the elastic distortion of the precipitate[22].Figure2epresents high-angle annular dark field scanning transmission electron microscopy (HAADF-STEM) and elemental distribution energy dispersive spectroscopy (EDS) maps.Based on the results illustrated in the figure, we determined that the precipitates were mainly composed of Ni, Al, and Ta.

Figure 6 .
Figure 6.DIC results for PXL700 °C-6 h sample under each cyclic test.(a-c) First to third cycles, respectively.The numbers 1, 2, and 3 represent loading, loading to target strain levels, and postunloading, respectively.

Figure
Figure 7a-h present DIC results for the PXL700 °C-12 h sample under each cyclic test.In the first, second, and third cycles, the sample exhibited a full recovery, and residual martensite was not observed after testing.In the fourth cycle, DIC results revealed that two variants of martensite were formed in the tensile sample during the superelastic test.As a result, two martensite variants could contribute equally to tensile deformation.Two martensite variants can easily form corresponding variant pairs (CVPs).The formation of a CVP enables internal twinning to accommodate tensile deformation and provides a more coherent interface with the austenite.The tensile result showed a small and smooth increase in stress with increased strain during the test.In the seventh cycle, as illustrated in Figure7g, the DIC result revealed a large volume fraction of residual martensite in the sample after the test.This was because the martensite interaction became intense, so that martensite was retained in the sample.As a result, the sample exhibited gradual cyclic degradation due to the retained martensite.Such residual martensite has been found to be due to the generation of dislocations[22].Dislocations are generated at the austenitemartensite interface; these pin the martensitic phase and prevent it from reversing back to austenite.As a result, a large volume fraction of the martensitic phase is retained in the tensile sample, and this contributes to irrecoverable strain.The precipitate size for PXL700 °C-12 h was found to be around 18 nm.Previous studies like that of Tanaka et al.[4] indicated that precipitates with a size of about 3 nm have a positive effect on the recoverable strain of superelasticity.If the precipitate size is large, it may lose coherency and affect the nucleation of martensitic phases.The loss of

Figure 7 .
Figure 7. DIC results for PXL700 °C-12 h sample under each cyclic test.(a-h) First to eighth cycles, respectively.The numbers 1, 2, and 3 represent loading, loading to target strain levels, and postunloading, respectively.

Table 1 .
Nomenclature of this study.