Exploration of stable atomic configurations in graphene-like BCN systems by Bayesian optimization

• Main Authors: Hara, Taichi, Kusaba, Akira, Kangawa, Yoshihiro, Kuboyama, Tetsuji, Bowler, David, Kawka, Karol, Kempisty, Pawel
• Format: Journal Article
• Language: English
• Published: 07.11.2024
• Subjects: Physics - Materials Science
• DOI: 10.48550/arxiv.2411.04758

h-BCN is an intriguing material system where the bandgap varies considerably depending on the atomic configuration, even at a fixed composition. Exploring stable atomic configurations in this system is crucial for discussing the energetic formability and controllability of desirable configurations. In this study, this challenge is tackled by combining first-principles calculations with Bayesian optimization. An encoding method that represents the configurations as vectors, while incorporating information about the local atomic environments, is proposed for the search. Although the optimization did not function with the conventional one-hot encoding that had been effective in other material systems, the proposed encoding proved efficient in the search. As a result, two interesting semiconductor configurations were discovered. These configurations exhibit qualitatively similar patterns to conventional models, and one of the two is more stable than the conventional ones with the same periodicity. Furthermore, the optimization behavior is discussed through principal component analysis, confirming that the ordered BN network and the C configuration features are well embedded in the search space. The proposed encoding method, which is easy to implement, is expected to expand the applicability of atomic configuration search using Bayesian optimization to a broader range of material systems.