(2021) Gubaev, Konstantin; Ikeda, Yuji; Tasnádi, Ferenc; Neugebauer, Jörg; Shapeev, Alexander V.; Grabowski, Blazej; Körmann, Fritz
An active learning approach to train machine-learning interatomic potentials (moment tensor potentials) for multicomponent alloys to ab initio data is presented. Employing this approach, the disordered body-centered cubic (bcc) TiZrHfTax system with varying Ta concentration is investigated via molecular dynamics simulations. Our results show a strong interplay between elastic properties and the structural ω phase stability, strongly affecting the mechanical properties. Based on these insights we systematically screen composition space for regimes where elastic constants show little or no temperature dependence (elinvar effect).
Local lattice distortions in a series of body-centered cubic alloys, including refractory high-entropy alloys, are investigated by means of atomic volumes, atomic charges, and atomic stresses defined by the Bader charge analysis based on first-principles calculations. Analyzing the extensive data sets, we find large distributions of these atomic properties for each element in each alloy, indicating a large impact of the varying local chemical environments. We show that these local-environment effects can be well understood and captured already by the first and the second nearest neighbor shells. Based on this insight, we employ linear regression models up to the second nearest neighbor shell to accurately predict these atomic properties. Finally, we find that the elementwise-averaged values of the atomic properties correlate linearly with the averaged valence-electron concentration of the considered alloys.
