Atomic Scale Modeling of Microstructural Features and Defects in Shape Memory Alloys

Modeling of shape memory alloys (SMAs) is inherently a multiscale endeavor, ranging from continuum models of macroscopic behavior, down to atomic scale calculation of properties. Modeling at the atomic scale provides the unique advantage of being from “first principles,” i.e., no underlying assumptions about the material of interest must be made; properties are calculated purely on the basis of atomic interactions. Methods which employ quantum mechanical techniques, e.g., density functional theory (DFT), to describe atomic interactions have proven successful in predicting important SMA bulk properties such as crystal structures, elastic constants, and transition temperatures. The critical weakness of such techniques, however, is their great computational expense, which limits calculations to only a few unit cells at a time. This precludes investigation of important microstructural features and defects, such as dislocations, precipitates, grain boundaries, and interphase boundaries. Extending calculations to these length scales, while maintaining atomistic resolution, requires employing cheaper semi-empirical or machine-learned descriptions of the underlying atomic interactions. Here, we present some examples of how such techniques can be used to provide novel insights into SMA behavior. Specifically, we investigate the kinetics of austenite-martensite interface motion in NiTi and NiTiHf alloys. The kinetics are influenced in different ways by the presence of both precipitates and point defects. We discuss how these interactions can have consequences on important SMA properties such as hysteresis. With this example in mind, we discuss how such large-scale atomistic calculations can be used as a predictive engineering tool for SMA design.