Machine learning and meta-heuristic algorithm assisted optimization of perovskite/Sb₂Se₃ tandem solar cells

Abstract Tandem solar cells are anticipated to possess higher efficiencies than single-junction counterparts originating from the broader absorption spectra. In recent years, machine learning methods have demonstrated to be efficient tools for the design and optimization of photoelectric devices. Herein, we proposed a tandem solar cell based on absorber layers of MAPbI 3 and Sb 2 Se 3 and employed three different machine learning models to predict its photovoltaic response. Additionally, a co-simulation model of metaheuristic algorithm (GA, PSO, SA) and machine learning model (LR, RR, MLP, DT, RF) is constructed to optimize the bandgap of the absorption materials, as well as the thicknesses of the ETL, HTL, and absorber layers. Through the proposed inverse design, the optimized power conversion efficiency (PCE) of the concerned cell is demonstrated to reach ∼39.86%. The current work is believed to be insightful for the design and boosting the efficiency of tandem solar cells.