SCAPS Empowered Machine Learning Modelling of Perovskite Solar Cells: Predictive Design of Active Layer and Hole Transport Materials

Recently, organic-inorganic perovskites have manifested great capacity to enhance the performance of photovoltaic systems, owing to their impressive optical and electronic properties. In this simulation survey, we employed the Solar Cell Capacitance Simulator (SCAPS-1D) to numerically analyze the effect of different hole transport layers (HTLs) (Spiro, CIS, and CsSnI3) and perovskite active layers (ALs) (FAPbI(3), MAPbI(3), and CsPbI3) on the solar cells' performance with an assumed configuration of FTO/SnO2/AL/HTL/Au. The influence of layer thickness, doping density, and defect density was studied. Then, we trained a machine learning (ML) model to perform predictions on the performance metrics of the solar cells. According to the SCAPS results, CsSnI3 (as HTL) with a thickness of 220 nm, a defect density of 5 x 10(17) cm(-3), and a doping density of 5 x 10(19) cm(-3) yielded the highest power conversion efficiency (PCE) of 23.90%. In addition, a 530 nm-FAPbI(3) AL with a bandgap energy of 1.51 eV and a defect density of 10(14) cm(-3) was more favorable than MAPbI(3) (1.55 eV) and CsPbI3 (1.73 eV) to attain a PCE of >24%. ML predicted the performance matrices of the investigated solar cells with similar to 75% accuracy. Therefore, the FTO/SnO2/FAPbI(3)/CsSnI3/Au structure would be suitable for experimental studies to fabricate high-performance photovoltaic devices.

Automated summary

Recently, organic-inorganic perovskites have shown great potential in enhancing the performance of photovoltaic systems due to their exceptional optical and electronic properties. In this study, the influence of different hole transport layers (HTLs) and perovskite active layers (ALs) on the performance of solar cells was numerically analyzed using the Solar Cell Capacitance Simulator (SCAPS-1D). The results showed that CsSnI3 as the HTL and FAPbI(3) as the AL achieved the highest power conversion efficiency (PCE) of 23.90%. Machine learning (ML) was used to predict the performance metrics of solar cells with approximately 75% accuracy.