Design of photovoltaic materials assisted by machine learning and the mechanical tunability under micro-strain

Abstract

In order to address the limited mechanical properties of silicon-based materials, this study designed 12 B-site mixed-valence perovskites with s⁰ + s² electronic configurations. Five machine learning models were used to predict the bandgap values of candidate materials, and Cs₂AgSbCl₆ was selected as the optimal light absorbing material. By using first principles calculations under stress and strain, it has been determined that micro-strains can achieve the goals of reducing material strength, enhancing flexible characteristics, directionally adjusting the anisotropy of stress concentration areas, improving thermodynamic properties, and enhancing sound insulation ability without significantly affecting photoelectric properties. According to device simulations, tensile strain can effectively increase the theoretical efficiency of solar cells. This work elucidates the mechanism of mechanical property changes under stress and strain, offering insights into new materials for solar energy conversion and accelerating the development of high-performance photovoltaic devices.