Thermodynamic stability descriptor of A2BX6-type perovskite materials

The toxicity issue and unsatisfactory thermodynamic stability of Pb-based perovskites pose the most critical challenge for practical commercialization. Among the remaining alternatives, A 2 BX 6 -type vacancy ordered double perovskites have attracted significant attention. With the success achieved by A 2 BX 6 -type materials, the design of stable and non-toxic A 2 BX 6 -type materials has consistently remained a prominent concern. The incorporation of stability descriptors has played a crucial role in the design and exploration of conventional single and double perovskites. However, due to the significantly more relaxed arrangement exhibited by the octahedral structure of A 2 BX 6 compared to that of ABX 3 and A 2 B 1 B 3 X 6 , traditional descriptors have proven ineffective in establishing a structure-stability relationship for A 2 BX 6 . An explicit descriptor for the stability of A 2 BX 6 -type perovskite materials in this study has been developed through the utilization of symbolic regression algorithm. The accuracy of ∼90 %, which is significantly higher than traditional stability descriptors, and also comparable to that of black-box machine learning models. The utilization of this efficient descriptor facilitates the design stable vacancy ordered double perovskite materials. • Thermodynamic stability descriptor of A 2 BX 6 compounds has been constructed. • Symbolic regression algorithm has been applied to explore descriptors. • The descriptor performs significantly better than traditional descriptors for A 2 BX 6 perovskites.