Machine learning-driven analysis of activation energy for metal halide perovskites

Metal halide perovskite single crystals (MHPSCs) are highly promising materials for optoelectronic applications, but their stability is hindered by ion migration, thereby impacting their performance. A key factor to understand this issue is calculating the activation energy. Electrochemical Impedance Spectroscopy (EIS) is a powerful technique for separating ionic and electronic processes, yet traditional analysis is labour-intensive, involving extensive measurements, circuit fitting, and manual data interpretation. In this study, we introduce a machine learning (ML)-driven approach to fully automate EIS analysis. EIS data, collected for MAPbI₃ and MAPbBr₃ across temperatures from 263 K to 343 K, enabled the creation of a large database. The developed ML model predicts EIS spectra at unknown temperatures, fits the appropriate electrical circuit, and automatically extracts passive component values to calculate the activation energy via an Arrhenius plot. This automated workflow streamlines the calculation process, offering fast and reliable activation energy predictions even when temperature data are incomplete or missing. Our approach enhances the efficiency of EIS analysis, providing valuable insights into the stability and performance of MHP SCs.