Fabrication process analysis on Sb2(SxSe1-x)3-based material properties and solar cell performance via machine learning

Abstract

The optimization of the fabrication process of Sb₂(S_xSe_1-x)₃ thin-films and studying the interplay between the parameters of their growth conditions to improve the performance of solar cells using traditional experimental methods requires more time and resources. In this work, we explore the application of machine learning (ML) techniques using the experimental data from peer-reviewed reports to optimize the fabrication process of Sb₂(S_xSe_1-x)₃ thin films, targeting to enhance the device performance. The optimized ML models demonstrate high accuracy in predicting the power conversion efficiency with a root mean square error of 1% and a Pearson coefficient of 0.9. Furthermore, the Shapley additive explanations method is employed to rank the fabrication parameters that have an impact on the solar cell performance. Finally, the results obtained are validated through their consistency with theory and experimental verification.