Machine Learning-Driven Optimization of CuAl _{1– <i>x</i>} Fe _<i>x</i> S ₂ -based Solar Cells with Oxygen-Implanted FeS ₂ Thin Films as the Hole Transport Layer

Incorporating a hole transport layer (HTL) in a solar cell provides a promising approach to enhance charge extraction and suppress recombination; however, identifying a compatible HTL with a given absorber material remains challenging. This study investigates oxygen-implanted FeS2 thin films as HTLs in CuAl1–xFexS2-based solar cells using a machine learning-assisted framework. A total of 187,200 device simulations were carried out in the SILVACO ATLAS simulator by systematically varying absorber and HTL parameters. Random Forest regression coupled with SHAP analysis identified the absorber bandgap as the most influential factor, with an optimal value of 1.63 eV, corresponding to 41% Fe-substitution in CuAlS2. Among the evaluated device structures, the device incorporating oxygen-implanted FeS2 with a dose of 9 × 1014 ions cm–2 as HTLs achieved the highest power conversion efficiency of 14.77%, along with an open-circuit voltage of 0.92 V, due to reduced interfacial recombination losses.