Introduction
Shape memory alloys (SMAs) are a unique class of materials due to their unique properties of the shape memory effect and pseudoelastic behaviour [
1]. These properties originate from their thermoelastic martensitic transformations. The most common SMAs are the NiTi-based alloys, which exhibit a B2 austenite to monoclinic B19′ martensite transformation. The B2 → B19′ transformation is associated with a crystalline lattice distortion, which may manifest as global strains that are recoverable upon the reversal of the phase transformation. This self-enabled shape strain, associated with either the forward B2 → B19′ transformation or the reverse B19′ → B2 transformation, is the underlying mechanism of these two unique properties that have rendered them suitable for many applications, e.g., as actuators and sensors [
2,
3].
The performance of these alloys in such applications is obviously dependent on the magnitude of the transformation-enabled recoverable strain, as well as the forces they generate in doing so and the critical temperature at which the transformation occurs. For polycrystalline NiTi commercial alloys, the recoverable strain is typically 5–7% [
1]. The magnitude of the recoverable strain is directly dictated by the crystallographic lattice distortion of the transformation, of which the monoclinic angle of the B19′ martensite is the most important single parameter [
4].
The general knowledge of the B19′ phase is that it has a monoclinic angle of 97.8°, as has been experimentally determined [
5]. However, some studies have shown that the monoclinic angle of the B19′ phase in NiTi-based alloys can be influenced by various factors. Prokoshkin et al. [
6] determined the lattice constants of the B19′ martensite of binary NiTi alloys from X-ray diffraction (XRD) analysis. They found that the monoclinic angle decreased from 97.9° to 96.4° when increasing the Ni content from 50.0 at. % to 50.7 at. %. Similarly, Zarinejad et al. [
7] studied the crystal structures of the B19′ martensite of NiTiHf SMAs by means of XRD analysis and found that the monoclinic angle of pseudo-equiatomic Ni(Ti
50−xHf
x) alloys increased with the increase of Hf content, from 98.0° in Ni
50Ti
45Hf
5 to 100.1° in Ni
50Ti
30Hf
20. It was also reported that increasing the Cu content in pseudo-equiatomic Ti(Ni
50−xCu
x) system decreased the monoclinic angle of B19′ from 96.2° at 5 at. % Cu to 95.2° at 10 at. % Cu [
8]. Ahadi et al. [
9] measured the monoclinic angle of binary NiTi by means of in-situ neutron diffraction and found that the angle increased by ~ 1° when the temperature was decreased from 300 to 50 K.
In addition to these experimental observations, the monoclinic angle has also been found to change by various factors in theoretical analyses, including density functional theory (DFT) calculations [
10], molecular dynamics (MD) simulations [
11], and crystallographic analysis [
4]. Bakhtiari et al. [
10] investigated the change in monoclinic angle and lattice constants of B19″ martensite phase in Ti
50Ni
50−xCu
x alloys by means of DFT calculation and showed that the monoclinic angle decreases with increasing Cu content, which is consistent with the experimental observations reported by Nam et al. [
8,
12]. Mansouri Tehrani et al. [
11] studied the effects of vacancy and anti-site defects on equiatomic NiTi B19′ structure by means of MD simulation and showed that these defects can reduce the monoclinic angle to 92.5°. A similar finding was also reported by Ko [
13], who examined the martensite structure by using harmonic approximation within the DFT calculation.
Echoing the findings of the variations of the monoclinic angle of the B19′ phase, the shape memory strain of NiTi-based SMAs has also been reported to vary under various conditions in the literature. For example, pseudoelastic deformation cycling [
14‐
16] and two-way shape memory [
17,
18] thermal cycling have been found to reduce the transformation strain. Increasing the amount of cold work decreases the transformation strain [
19,
20]. Increasing the temperature of recovery anneal (prior to recrystallization) after cold working increases the transformation strain [
20,
21]. Increasing the level of stress of deformation (e.g., by increasing the testing temperature) increases the pseudoelastic strain [
22,
23]. Furthermore, Ni-rich NiTi alloys appear to have smaller transformation strains than near-equiatomic NiTi alloys [
6,
24]. It is also common knowledge that third element alloying may change the magnitude of the strain output of the B2 → B19′ transformation. For example, substitution of 5 at. % Cu for Ni in NiTi reduces the B2 → B19′ transformation strain to 3.3% [
12], and the maximum recoverable strain of (Ni
47Ti
44)
100−xNb
x decreases with increasing Nb content [
25]. Saghaian et al. [
26] observed that the recoverable strain decreased with increasing Ni content in Ni-rich NiTi-20Hf SMAs.
Whereas there may be many reasonable factors, both intrinsic to the transformation crystallography and extrinsic as embodied within the microstructure, that can change the macroscopic transformation strain, there is no reason to exclude the monoclinic angle of the B19′ phase as a contributing factor. To investigate this, the effects of structural and compositional point defects on the monoclinic angle of the B19′ phase in NiTi-based alloys were investigated in this study by means of DFT calculation. Four types of point defects were studied, including vacancies, Ni–Ti anti-site swapping, third element substitution, and Ni or Ti enrichment in the B2 stoichiometry.
DFT Calculations
The effects of the four types of point defects on the structure of the B19′ martensite in NiTi was studied by means of DFT calculation. It is known that the monoclinic B19′ phase is observed experimentally but does not exist as a minimum energy state in DFT calculation for the Ni–Ti system, and that the monoclinic B19″ phase is often used as a proxy for the B19′ phase [
10]. Given that both B19′ and B19″ are product phases of martensitic transformations of the B2 phase, they have identical atomic position and unit cell configuration except the monoclinic angle. This gives the validity for using B19″ in DFT calculation as a proxy for the experimentally observed B19′. Figure
1a shows a comparison of the unit cells selected for the DFT calculation of the B2, B19′, and B19″ phases. The line drawing presents the B2 structure. The blue dashed lines define the unit cell selected. The three sphere model structures show the selected B2, B19′, and B19″ unit cells. For the calculations, supercells of three different sizes of 2 × 2 × 2 B19″ unit cells with 32 atoms, 3 × 3 × 3 B19″ unit cells with 108 atoms, and 4 × 4 × 4 B19″ unit cells with 256 atoms were used. Figure
1b shows the 2 × 2 × 2 B19″ supercell.
The DFT calculation was performed using the Vienna ab initio Simulations Package (VASP) [
27]. The PBE (Perdew–Burke–Ernzerh) exchange–correlation functionals [
28] and the projector-augmented wave method (PAW) [
29] with a cut-off energy of 500 eV were employed to perform the simulations. The
k-point mesh density was at least 50 per Å
−1, the convergence criterion of the total energy was 10
–6 eV, and the maximum force on each atom was 5 × 10
–3 eV Å
−1. The simulation temperature was 0 K. Each structure was fully relaxed in all degrees of freedom. After relaxation, if the other two angles (α and γ) of the unit cell are within the range of 90° ± 1°, the structure is considered monoclinic and vice versa.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.