Thermal cycling degradation in Cs₃Sb₂Br₉-based solar cell: A simulation-to-ML framework

This study combines density functional theory (DFT), SCAPS-1D device simulation, and machine learning (ML). It investigates thermal degradation and performance stability of lead-free Cs₃Sb₂Br₉ perovskite solar cells (PSCs) on repeated heating and cooling cycles. The device architecture is FTO/TiO₂/Cs₃Sb₂Br₉/CFTS/Au. The device is optimized for efficient charge transport and improved interfacial stability. DFT calculations were used to determine the structural and optical properties. The results showed a suitable direct bandgap and strong visible-light absorption. These properties make Cs₃Sb₂Br₉ a promising absorber material. These calculated parameters were inserted into SCAPS-1D. SCAPS-1D was used to analyse charge transport, recombination mechanisms, and band alignment across the layers. An XGBoost regression model was trained using combined experimental and simulated datasets. The model was used to predict degradation trends under different temperature and humidity conditions. It achieved a high R² value of 0.992. The optimized device showed a short - circuit current density (Jsc) = 30.17 mA/cm², an open-circuit voltage (Voc) of 1.18 V, a fill factor (FF) of 63.75% and power conversion efficiency (PCE) of 21.66%. These results show strong photovoltaic performance and good thermal resilience. The novelty of this work lies in the integration of density functional theory (DFT), SCAPS-1D device simulation, and machine learning to investigate thermal cycling-induced degradation in Cs₃Sb₂Br₉ PSCs under realistic operating conditions. This multiscale approach links material-level properties with device performance and degradation behaviour. The analysis further identifies temperature, relative humidity, and operational cycling as key parameters governing degradation, with a nonlinear dependence of device performance on these factors.