Performance optimization of CsSnI3-based perovskite solar cells using SCAPS-1D and machine learning analysis

Tin-based perovskite solar cells (PSCs) can be employed in place of lead-based PSC, delivering equivalent efficiency and stability while eliminating the use of lead. The work comprises simulation of CsSnI3-based PSC by using the SCAPS-1D software. The device performance with several electron transport layers such as ZnO, TiO2, WO3, CdS, IGZO and MoS2, CuSbS2, Cu2O, V2O5, CuInS2 QD as hole transport layers is studied. The several factors such as thickness of the absorber layer, operating temperature, work function and defect density (N-t) are varied. The optimized thickness of the absorber layer is 400 nm. The optimized architecture, FTO/CdS/CSnI3/MoS2/Pd has achieved a maximum efficiency of 23.45% at 300 K and N-t of 1 x 10(13) cm(-3). The ML model predicts the effect of distinct parameters on device performance and PCE using SHAP plot, XGB and RF algorithm. XGB model predicts performance with accuracy of 99.93%. The ML model is helpful for the fabrication of efficient CSnI3 PSCs.