Data-Driven Strategies for Analysis of Perovskite Solar Cell Performance: Insights from Scaps and ML

Traditional perovskite solar cells are poisonous and unstable, making commercialization difficult. High-performance lead-free inorganic perovskites may overcome these challenges. This study used the Solar Cell Capacitance Simulator (SCAPS-1D) to analyze the SnO₂/Cs₂TiBr₆/Cu₂O structured solar cells’ performance. The impact of the thickness of the absorbing layer, defect density and temperature was investigated. The analysis reveals that when the thickness of the absorbing perovskite material (Cs₂TiBr₆) increases, the short-circuit current density (Jsc), open circuit voltage (Voc), fill factor (FF), and efficiency improve due to the generation of more electron-hole pairs. In contrast, the efficiency of the simulated solar cell device decreases as the ambient operation temperature increases as a result of the decrease in Voc. The optimized structure of the designed solar cell was fabricated using spin coating method. The absorbing layer Cs₂TiBr₆ was characterized by x-ray diffraction (XRD) which confirms the crystallinity of the layer. Subsequently, we developed a machine learning (ML) model to predict the performance parameters of the solar cells. ML & data science techniques are used with the linear regression model to predict power conversion efficiency (PCE), which demonstrates a high level of accuracy, as indicated by a significant R-squared (R2) score.