Regensburg 2019 – wissenschaftliches Programm
MM 8.7: Vortrag
Montag, 1. April 2019, 18:15–18:30, H43
Investigation of phase stability in high-entropy alloys with the use of machine-learning interatomic potentials — •Tatiana Kostiuchenko1, Alexander Shapeev1, Fritz Körmann2, 3, and Jörg Neugebauer2 — 1Skolkovo Institute of Science and Technology, Moscow, Russia — 2Max-Planck-Institut für Eisenforschung GmbH, Düsseldorf, Germany — 3Delft University of Technology,The Netherlands
High-entropy alloys (HEA) is a class of materials which consist of at least four different chemical elements and have a specific structure. These alloys have high ductility and yield strengthening, they are widely used as construction and heat-resistant materials. Experimental methods of HEAs investigation are time-consuming, and by this reason, computational methods of HEAs investigation are of the high interest. In this work, a new data-driven approach for investigation of solid solution stability in HEA is proposed. It is based on Canonical Monte Carlo algorithm with the use of machine-learning potential, namely low-rank potential (LRP) [Shapeev A., 2017]. The approach was examined by comparing it with the existing works. The key difference of the LRP from the other ``on-lattice'' models is its ability to take into account local lattice distortions, which is critical for the materials behavior. The parameters of the LRP were fitted on quantum-mechanical data, the LRPs prediction accuracy was 1 meV/atom. Thus, the temperature of the order/disorder phase transition was accurately calculated for the equiatomic NbMoTaW system. The low-energy structures and the mechanisms of chemical ordering were also investigated.