Two-dimensional nonlinear optical materials predicted by network visualization

Two-dimensional (2D) materials with nonlinear optical (NLO) effects have emerged as promising candidates for nanoscale laser devices. However, only a few monolayers have been experimentally explored. Herein, starting from 258 compounds that have been predicted to be readily exfoliable, we built networks based on the optical properties of the compounds with machine learning and graph theory to illustrate the importance and connection of their elements. The results show that iodine, bromine, oxygen and chlorine play very important roles in these materials; metal chalcogenides also play a large role; and hydrogen, which is usually negligible in bulk crystals, may represent a breakthrough in 2D systems. The first-principles calculations are consistent with previous publications both theoretically and experimentally. This method can also be applied to other functional material portfolios.