A machine learning-assisted exploration of the structural stability, electronic, optical, heat conduction and mechanical properties of C₃N₄ graphitic carbon nitride monolayers

Among all members of the extensive family of two-dimensional (2D) materials, graphitic carbon nitrides of the C₃N₄ stand out as one of the most successful nanomembranes, owing to their exceptional electro-optical and chemical properties, accompanied by remarkable achievements in their large scale synthesis. Nevertheless, the complex corrugated nature of their most stable structures poses a challenge in the accurate evaluation of their electronic, optical, mechanical and thermal properties from a theoretical point of view. To address the aforementioned challenge, we herein employed a combination of the machine learning interatomic potentials (MLIPs) and density functional theory (DFT) calculations. The MLIPs facilitated the detection of dynamically stable configurations and subsequently allowed us to accurately evaluate the lattice thermal conductivity and mechanical properties of four different C₃N₄ lattices, one of which is theoretically predicted in this work. Using the HSE06-based DFT calculations, the electronic band structure, optical properties and photocatalytic potential of the considered nanomembranes were also investigated. Thanks to the robustness of the proposed combined MLIP-DFT approach, this study not only presents the first comprehensive understanding of the stability, thermal conductivity, mechanical strength, electronic and optical properties of the C₃N₄ nanosheets, but also highlights the crucial influence of the structural corrugations on the resulting theoretical predictions for the nanoporous 2D material properties.