Synergistic Screening of High-Performance TMDs/2D-LHPs Heterostructures for Solar Cells via Deep Learning and DFT

Abstract

Two-dimensional layered hybrid perovskites (2D-LHPs) and transition metal dichalcogenides (TMDs) heterostructures exhibit exceptional optoelectronic properties, making them highly promising for photovoltaic and optoelectronic applications. However, due to the vast number of possible heterojunction combinations, efficiently screening high-performance materials remains a challenge. In this study, a deep learning model is employed to systematically predict the band alignment types of TMDs/2D-LHPs heterostructures, identifying 3,510 potential Type-II heterostructures. Further screening criteria are applied to select 99 heterostructures for high-throughput density functional theory (DFT) calculations, evaluating their photovoltaic conversion efficiency (PCE). The results reveal that 10 heterostructures achieve a PCE exceeding 20%, with the highest reaching 22.43%. Notably, some of these heterostructures exhibit low effective masses and high carrier mobilities (~10⁴ cm²/V·s). Additionally, optical absorption coefficient calculations indicate that all 10 heterostructures possess strong light absorption capabilities (~10⁵ cm⁻¹), highlighting their significant potential for solar energy applications. Furthermore, deep learning methods are utilized to predict the PCE of the remaining 3,411 TMDs/2D-LHPs heterostructures based on computational data, providing valuable guidance for both experimental and theoretical research.