Machine-learning-guided efficiency engineering of In2AgSbCl6 double perovskite photovoltaics

PSCs, or perovskite solar cells, are a next-generation photovoltaic technology that is gaining popularity due to their unique properties. The optimum designs of the In 2 AgSbCl 6 double perovskite-based solar cells are investigated using a combination of density functional theory (DFT) and SCAPS-1D. In this research, we executed a DFT analysis to evaluate and assess the structural, electronic, and optical properties of the In 2 AgSbCl 6 absorber. The electronic band gap (E g ) of the In 2 AgSbCl 6 absorber was 1.321 eV and 2.582 eV, respectively, using the GGA and Tb-mBJ approaches of DFT. This study aimed to examine the In 2 AgSbCl 6 -based perovskite absorber material and model the outcomes using the SCAPS-1D simulation program. Different configurations of In 2 AgSbCl 6 - based solar cells were designed using an ITO substrate, ZnS, PCBM, ZnSe, and C 60 as Electron Transport Layer (ETL) components, In 2 AgSbCl 6 as the absorber, MoS 2 as the Hole Transport Layer (HTL), and Pt as the back contact. The thickness of the absorber and the ETL have a big impact on the device’s performance. Additionally, the comprehensive analysis was conducted to evaluate the impact of several factors, including the absorber defect density, shunt and series resistance, temperature, Mott-Schottky behavior, capacitance, recombination, generation rates, J-V (current–voltage density) characteristics, and quantum efficiency (QE). In 2 AgSbCl 6 -based solar cell structures are investigated four best optimized combinations, and power conversion efficiencies (PCE) of devices optimized with 27.16%, 27.13%, 27.08%, and 26.31% of ETLs and HTLs combination. ITO/ZnS/In 2 AgSbCl 6 /MoS 2 /Pt displays the best performance, with a power conversion efficiency of 27.16%, a current density (J SC ) of 34.82 mA/cm 2 , an open circuit voltage (V OC ) of 0.926 V, and a fill factor (FF) of 84.20%. Subsequently, a machine learning (ML) framework was developed to estimate the key performance parameters of the fabricated solar cells. High-fidelity SCAPS-1D data were employed to train and validate an ensemble ML framework for predicting essential photovoltaic metrics such as PCE, V OC , J SC, and FF. Compared with previous research, this is a thorough and important advancement that provides valuable insights for the development of highly efficient and reasonably priced PSCs, greatly expanding solar technology.