|About this Abstract
||2022 TMS Annual Meeting & Exhibition
||Advances in Multi-Principal Elements Alloys X
||High-throughput simulation of finite-temperature elasticity and phase stability of TiZrHfTax alloys with near-DFT accuracy via machine-learning interatomic potentials
||Konstantin Gubaev, Yuji Ikeda, Blazej Grabowski, Fritz Körmann
|On-Site Speaker (Planned)
Accurate simulation of finite-temperature properties of a range of metallic alloys is often infeasible by means of DFT calculations.
In the present work we tackle the challenge of investigating the relation between elastic properties and dynamical instability of the bcc-phase towards the 𝜔-phase in TiZrHfTax alloys in a wide temperature range with varying Ta concentration.
For bcc-𝜔 phases distinction we use recently proposed scheme based on analysis of atomic displacements along transformation directions during finite-temperature MD.
We achieve a great performance boost while not sacrificing much accuracy (compared to DFT) by using both accurate interpolation of ab-initio data and the active learning principle.
The research involves using such software as VASP (for ab-initio energies/forces), LAMMPS (for MD simulations) and MLIP (for interpolation of DFT data).
The framework proposed in the present work is essentially generic and can be used for other high-entropy alloys as well.
||Computational Materials Science & Engineering, High-Entropy Alloys, Modeling and Simulation