High-performance lead-free Ba 3 SbI 3 -based perovskite solar cells: device simulation and machine learning-based optimization

Abstract In this study, a fully inorganic, lead-free Perovskite solar cell (PSC) architecture employing Ba 3 SbI 3 has been proposed using both simulation and a machine learning approach. A detailed multi-variable charge transport layers are analyzed, and it reveals that the TiO 2 -CuSbS 2 pair offers a balanced optimal choice as the electron and hole transport layers (E/HTLs) respectively. A thorough numerical assessment has been performed to investigate the device parameters, including the thickness of different layers, doping concentration and defect levels to obtain the optimum device performance. From our proposed lead-free Al/FTO/TiO 2 /Ba 3 SbI 3 /CuSbS 2 /Au solar cell model, a power conversion efficiency of 32.91% was obtained with an open-circuit voltage of 1.18 V, a short-circuit current density of 32.16 mA cm −2 and a fill factor of 86.38%. In addition, a random forest regression (RFR) model has been incorporated to predict photovoltaic performance and to estimate the relative impact of critical device parameters, which significantly influence the device performance. Our proposed RFR model achieved impressive predictive accuracy ( R 2 > 0.92) across all output parameters. Correlation matrix analysis supports these findings and offers a data-driven pathway for further optimization. This combined simulation and machine learning approach certainly exhibits a valuable insight for advancing efficient, lead-free PSCs for promoting environmentally friendly and sustainable energy solutions in the near future.