Machine‐Learning Assisted Optimized Photovoltaic Performance of Sb 2 S 3 /Sb 2 Se 3 Heterojunction Using DFT and SCAPS‐1D

Antimony chalcogenides have earned significant attention as potential absorber materials owing to their suitable bandgaps, high absorption coefficients, and Earth abundance. This work presents a multiscale computational framework combining density functional theory (DFT), SCAPS‐1D simulations, and machine learning (ML) to investigate an Sb 2 S 3 /Sb 2 Se 3 double‐absorber heterojunction. DFT is employed to analyze the structural, electronic, and optical properties of Sb 2 S 3 (Sb 2 Se 3 ) yielding bandgap values of 1.73 eV (1.28 eV) alongwith good optical absorption across the vis–UV region. DFT‐derived material parameters are used as input in SCAPS‐1D to simulate FTO/ZnSe(ETL)/Sb 2 S 3 /Sb 2 Se 3 /CuSbS 2 (HTL)/Au device, while the impacts of absorber layer thickness, defect density, radiative recombination coefficient, surface recombination velocity, and temperature are systematically investigated. To further predict the power conversion efficiency (PCE), the datasets used for ML models are generated in SCAPS‐1D by varying absorber thickness, bulk defect density, shallow acceptor density, and one output parameter, PCE. The ML models, such as SVR, RF, XGB, and stacked RF‐SVR, are employed to find a suitable model for PCE prediction. Among them, XGB exhibited superior predictive performance with R 2 value of 98.96% (97.55%) for Sb 2 S 3 (Sb 2 Se 3 ). Eventually, the optimized device delivers a PCE value of 20.29%, V oc of 1.14 V, J sc of 20.69 mA/cm 2 and FF of 85.87%.