[1P01-52] Poster Presentation

[1P48] Theoretical analysis of thermal conductivity of GaN containing defects using machine learning potential

*Rintaro Tobita¹, Shimizu Koji¹, Satoshi Watanabe¹ (1. Department of Materials Engineering, Graduate School of Engineering, The University of Tokyo)

Deepening our understanding on the effects of defects on thermal conduction behavior of Gallium Nitride (GaN) is important to improve performance of GaN devices. As a first step toward this goal, in this study, a high-dimensional neural network potential, a kind of machine learning potentials, was constructed for GaN with N vacancies. To improve the prediction accuracy compared with our previous work (Phys. Rev. B 106 (2022) 054108), additional structural data for training dataset were generated by ab initio molecular dynamics calculations. The prediction accuracy of the constructed potential was examined on various physical quantities.