Finite-temperature interplay of structural stability, chemical complexity, and elastic properties of bcc multicomponent alloys from ab initio trained machine-learning potentials

High-throughput finite-temperature phase stability and materials property predictions directly from first-principles are extremely resource demanding. Here, the authors combine density functional theory calculations with accurate machine-learning potentials and an active learning scheme to obtain fast, yet highly accurate interatomic potentials. The approach is used to analyze the bcc-$\ensuremath{\omega}$ phase stability and elastic properties of a series of bcc TiZrHfTa${}_{x}$ alloys. The phase stability is evaluated by analyzing the projections of atomic displacements from molecular dynamics trajectories revealing fingerprints of bcc-$\ensuremath{\omega}$ martensitic transformations.