[18p-B3-1~12] CS.7 Code-sharing Session of 6.1 & 13.3 & 13.5

3:45 PM - 4:00 PM JST(6:45 AM - 7:00 AM UTC)

[18p-B3-11] Investigation of modeling HfO2 crystals using machine learning potentials

〇(D)Yuki Itoya¹, Masaharu Kobayashi^1,2 (1.Tokyo Univ. Inst., 2.Tokyo Univ. d.lab)

Keywords:

Ferroelectric material,Simulation

HfO₂-based ferroelectrics are expected to be put to practical use in polycrystalline form. Molecular dynamics simulations for large systems such as polycrystals are computationally expensive. Therefore, a method has been proposed to lighten the computation by replacing the computationally heavy potential calculation part with an energy potential surface created by machine learning. In this study, we performed modeling using Ephemeral Data Derived Potentials proposed by Pickard et al. and evaluated the relationship between learning parameters and modeling accuracy.