Molecular dynamics simulations of austenite-martensite interface migration in NiTi alloy

The mechanism of austenite-martensite interface migration is a key component in understanding the phase transformations in shape memory alloys. It is also intimately tied to their observed hysteresis. Molecular dynamics simulations offer a unique capability to study phase transformations in detail, however, their associated timescales prevent the observation of interface formation via nucleation and growth near the transformation temperature. To address this challenge, we present a simulation methodology in which steady-state austenite-martensite interfaces are allowed to form close to equilibrium. The resulting structures contain well-defined interfaces which can be perturbed from equilibrium to study their migration. In NiTi specifically, the austenite-martensite interfaces are semicoherent, made up of terrace planes separated by structural disconnections. The disconnections advance via kink pairs and provide an atomic-scale mechanism for interface migration. The methodology and results presented here provide a foundation toward further leveraging molecular dynamics simulations to better understand how the atomic-scale structure of austenite-martensite interfaces impacts macroscopic properties such as hysteresis in shape memory alloys.

Figure 1

(a) Bulk free energy of B2 and B19' phases calculated via nonequilibrium TI. Each data point is averaged over five independent simulations, with error bars representing the standard deviation. (b) Cooling and heating of a bulk NiTi simulation cell (16 000 atoms) demonstrating the thermal hysteresis of the martensitic phase transformation.

Figure 2

(a) Bulk free-energy difference of B2 and B19' phases, as calculated via nonequilibrium TI, plotted for different concentrations of Ni. Each data point is averaged over five independent simulations, with error bars representing the standard deviation. (b) Transition temperatures determined from free-energy calculations compared with experimental results from various NiTi compositions. Experimental data courtesy of Ref. [4].

Figure 3

Cooling and heating of spherical NiTi nanoparticles, demonstrating the dependence of thermal hysteresis on surfaces and hence particle size. The bulk data from Fig. 1 is reproduced for comparison. The bulk transformation temperature ($T_0$), which falls outside the hysteresis loops, is indicated by a dotted line.

Figure 4

Schematic depicting the MD simulation setup for creation of austenite-martensite interfaces as well as its perturbation from equilibrium to drive interface migration. Snapshots show a cross-sectional view of the center of the spherical NiTi nanoparticle. Atoms are colored according to CNA: B2—blue, other—gray.

Figure 5

(a) Growth of the B19' plate during NVT MD simulations. (b) Austenite-martensite interface velocities determined from plate thickness values in (a). The data point at 1 ns for 410 K has been omitted because the nanoparticle completely transformed prior to this.

Figure 6

(a) Snapshot of the NiTi nanoparticle after NVE equilibration with a B19' nucleus shaped as a spherical cap. (b) Snapshot of the same nanoparticle after the B19' nucleus has grown—alternatively this can be viewed as the starting state for a B2 nucleus. Atoms are colored according to CNA: B2—blue, other—gray.

Figure 7

(a) Snapshot of the growing B19' plate at 415 K with atomic positions averaged over 1000 time steps. The interfaces are made up of disconnection arrays. (b), (c) Magnified images of interfaces 1 (b) and 2 (c). (d)–(f) Successive snapshots show the migration of interface 1 via the advancement of disconnections. The black line of the initial interface structure (d) is superimposed on all three images to emphasize the motion of the disconnections via the addition of new B19' atoms. Atoms are colored according to CNA: B2—blue, other—gray.

Figure 8

Top-down view of the austenite-martensite interface showing the array of disconnections as well as a number of kinks which facilitate their motion. B2 atoms have been removed for clarity, and the remaining atoms are colored according to their position along the habit plane normal, enabling the interfacial steps to be visualized. The bright and dark rings on the perimeter correspond to surface atoms and can be disregarded.

Figure 9

Schematic of the formation of a disconnection in NiTi when two crystals of B2 and B19' with opposite step directions are joined together. Each crystal has an associated step vector ($\mathbf{l}$) and step height ($h$). Summation of the two step vectors leads to the Burgers vector of the disconnection (${\mathbf{b}}_{\mathbf{D}}$).