Machine learning prediction of 2D perovskite photovoltaics and interaction with energetic ion implantation

Abstract

Atomic-level prediction combined with machine learning (ML) and density functional theory (DFT) is carried out to accelerate the fast discovery of potential photovoltaics from the 2D perovskites. Based on the ML prediction, stability test, optical absorption, and the theoretical power conversion efficiency (PCE) evaluation, two promising photovoltaics, i.e., Sr2VON3 and Ba2VON3, are discovered with PCE as high as 30.35% and 26.03%, respectively. Cu, Ag, C, N, H, and He ion implantation are adopted to improve the photovoltaic performance of the high-efficiency and best stable perovskite Sr2VON3. The time-dependent DFT electronic stopping calculations for energetic ion implanted Sr2VON3 indicate that the excited electrons from the valence band contribute to the electron–phonon coupling, the evolution and formation of the defects, and the photovoltaic performance. This work opens the way to the high-accuracy fast discovery of the high-efficiency and environmentally stable 2D perovskites solar cells and the further engineering improvement in photovoltaic performance by ion implantation.