A machine-learning-based investigation on the mechanical/failure response and thermal conductivity of semiconducting BC2N monolayers

B Mortazavi and IS Novikov and AV Shapeev, CARBON, 188, 431-441 (2022).

DOI: 10.1016/j.carbon.2021.12.039

Graphene-like lattices consisting of neighboring elements of boron, carbon and nitrogen are currently among the most attractive two- dimensional (2D) nanomaterials. Most recently, a novel graphene-like lattice with a BC2N stoichiometry has been grown over nickel catalyst via molecular precursor. Inspired by this experimental advance and exciting physics of h-BxCyNz lattices, herein extensive theoretical calculations are carried out to investigate physical properties of three different h-BC2N lattices. Density functional theory (DFT) results confirm direct-gap semiconducting electronic nature of the BC2N monolayers. In this work, state-of-the-art models based on the machine- learning interatomic potentials (MLIPs) are employed to elaborately explore the mechanical/failure and heat transport properties of various BC2N monolayers under ambient conditions. Outstanding accuracy of the developed MLIP-based classical models are confirmed by comparing the estimations with those by DFT. MLIP-based models are also found to outperform empirical interatomic potentials. It is shown that while the mechanical/failure responses are close for different BC2N lattices, the change of an atomic configuration can result in around four-fold differences in the lattice thermal conductivity. The obtained results confirm the robustness of MLIP-based models and moreover provide an extensive vision concerning the critical physical properties of the BC2N nanosheets and highlight their outstanding heat conduction, mechanical, and electronic characteristics. (C) 2021 Elsevier Ltd. All rights reserved.

Return to Publications page