| Abstrakt: |
Perovskite materials exhibit excellent optical and electronic properties, but they are highly sensitive to water, heat, and light, limiting their commercial applications. Although the stability of various perovskites has been previously studied, establishing general patterns remains challenging due to a limited number of samples. In this work, machine learning interpretability methods were employed to investigate the impact of halogen elements on the stability of organic–inorganic hybrid perovskites. Molecular dynamics simulations, modeling, and interpretability analyses were conducted on over a thousand organic– inorganic hybrid perovskites, and the conclusions were experimentally validated. The results reveal that chlorine (Cl) has the strongest ability to reduce the adsorption energy of perovskites, followed by bromine (Br), while iodine (I) exhibits the lowest degree of influence. The findings of this study generalize the relationship between the perovskite structure and properties, contributing to the screening and synthesis of perovskite materials with enhanced stability. [ABSTRACT FROM AUTHOR] |
