[10a-N302-1~11] 18.1 Materials Informatics

11:15 AM - 11:30 AM JST(2:15 AM - 2:30 AM UTC)

[10a-N302-9] Gradient-Based Inverse Composition Design for Dielectric Material Discovery

〇Daisuke Hirai¹, Keitaro Fujii¹, Akihiro Fujii², Anh Khoa Augustin Lu^3,2, Satoshi Watanabe² (1.Murata Manufacturing, 2.Univ. of Tokyo, 3.NIMS)

Keywords:

inverse problem,dielectrics

To enable efficient discovery of high-dielectric-constant materials, we developed a materials design approach that combines a neural-network-gradient-based inverse composition design method with a crystal structure generation model. As a result, candidate materials with previously unreported compositions and crystal structures were identified. Furthermore, density functional theory calculations confirmed their high dielectric constants, demonstrating the effectiveness of the proposed inverse design method.