A machine learning approach to identifying perovskite materials in A2BB′X6 compounds

Abstract

Double perovskite materials are attracting increasing attention due to their outstanding photoelectric properties and broad application prospects in many fields such as energy conversion, optoelectronics, and catalysis. Among the many double perovskite structures, the A₂BB′X₆ type is favored for its excellent physicochemical properties. However, not all A₂BB′X₆-type compounds can successfully form perovskite structures. This study successfully developed an efficient machine learning model that can accurately predict the formation of perovskite structures in A₂BB′X₆-type double perovskite materials. Based solely on the atomic statistical characteristics of the material, this model was able to successfully distinguish the compounds that could form perovskite structures in A₂BB′X₆-type compounds with an accuracy rate of 92.1%. In addition, the atomic statistics at the B site and B′ site have a significant impact on the formation of perovskite structures. This study provides new insights into the design and synthesis of perovskite materials while reducing the reliance on expensive and time-consuming experimental methods.