Janus Halogenated\nSilicene for Photocatalytic and\nPhotovoltaic Applications: A Machine Learning Combined with First-Principles\nStudy

In this paper, we integrate machine learning (ML) with\ntwo-dimensional\nmaterial screening to develop novel halogenated silicene-based materials\nwith considerable potential for photocatalytic and photovoltaic applications.\nWe proposed a data-driven framework that incorporates first-principles\nelectronic structure calculations and 286 self-contained databases\ntrained by supervised learning to accurately predict the structural\nstability and electronic properties of halogenated silicene-based\ncompounds. A total of 110 potential candidates for photocatalytic\nwater splitting were identified, and five materials with high predicted\nstability exhibited ideal band edges and demonstrated high carrier\nmobility rates. Among the 321 heterojunctions composed of halogenated\nsilicene compounds with potential type-II band alignments, we randomly\nscreened a Si<sub>8</sub>HFCl<sub>5</sub>I/Si<sub>8</sub>FClBrI<sub>5</sub> heterojunction and validated that it can achieve a high photoelectric\nconversion efficiency of 21.04%.