Machine-Learning-Assisted Investigation on Benign Ion Migration in Metal Halide Perovskites.

Defect-assisted ion migration is one of the important issues that results in instability and non-radiative losses in hybrid organic-inorganic metal halide perovskite solar cells. In this work, based on the deep potential (DP) model, a long-time-scale molecular dynamics (MD) simulation has been employed to capture the interstitial-assisted iodine migration process. The results indicate that, when interstitial iodine (I) begins to migrate, the serious structural distortion becomes mild, weakening the electron-vibration interaction.