A-SITE TUNING OF HALIDE PEROVSKITE SOLAR CELLS VIA MACHINE LEARNING

Abstract

In the realm of material science for solar cell applications, the prediction of critical properties such as tolerance factor, bandgap, refractive index, and dielectric constant in hybrid organic-inorganic perovskite materials holds paramount importance. While bandgap prediction traditionally relies on costly techniques such as Density Functional Theory (DFT), this study aims to substitute this method with a more computationally affordable approach using machine learning models. A comprehensive machine learning framework has been proposed to predict these crucial properties. This framework incorporates machine learning techniques, including polynomial regression, gradient boosting, and random forest. The model performance is evaluated based on the R-squared (r2) score to determine the most accurate and reliable model for prediction. The tolerance factor, a pivotal parameter for assessing a material's suitability for solar cell applications, is among the factors investigated. The machine learning model's ability to predict these parameters offers a promising avenue for cost-effective and efficient analysis, potentially replacing expensive and time-consuming DFT methods in predicting the bandgap. The findings showcase the potential of machine learning techniques to efficiently predict fundamental properties essential for evaluating hybrid organic-inorganic perovskite materials for solar cell applications. This research provides a foundation for further exploration and development in material science and photovoltaics.