Accurate and Efficient Prediction of Highly Disordered Bi₂O₃with Optimum Structure Pool: Combined Approach of the Special Quasirandom Structure Method and Structure Sampling

Materials exhibiting outstanding ion transport properties have been investigated for extensive applications, and there has been an increasing demand for the rational design of ion-conductive materials. δ-Bi₂O₃, which shows the highest reported oxygen ionic conductivity, possesses a notoriously defective and disordered structure, and this leads to make the initial guess of the atomic coordinates difficult. Therefore, there is a lack of fundamental understanding of the thermodynamic stability, doping effects, and surface reactions of δ-Bi₂O₃. Herein, we suggest an accurate and efficient way to describe the disordered nature of the δ-Bi₂O₃oxygen sublattice, involving structure modeling and structure sampling processes. Using a special quasirandom structure method, we developed the disordered structure pool of δ-Bi₂O₃and verified our modeling process by accurately predicting its material properties. Subsequently, we introduced a structure sampling process to efficiently determine the size of the Bi₂O₃structure pool based on the convergence of the disorder variable while maintaining the prediction accuracy.