Reassessing structural models of graphitic carbon nitride for reliable computational predictions

Accurate structural modelling is critical for reliable prediction of materials properties in computational design workflows. graphitic carbon nitride, a widely studied metal-free photocatalyst for hydrogen evolution, is typically represented using idealized planar layer models; however, the validity of these assumptions remains largely unexamined. In this work, we employ first-principles density functional theory (DFT) to systematically reassess the structural landscape of heptazine-based g-C?N? across monolayer and bulk forms, with the aim of establishing physically consistent models for predictive simulations.

We demonstrate that commonly adopted planar configurations correspond to metastable states and can lead to qualitatively incorrect predictions of electronic properties. Instead, a buckled heptazine framework is consistently identified as the energetic ground state across all systems considered.

This finding is independently supported by reverse Monte Carlo analysis of experimental data, providing direct validation of the computational model. Extending beyond monolayers, we explore a range of stacking registries and uncover previously unreported low-energy corrugated bulk structures (P1 symmetry) that exhibit significant variation in electronic band gaps, several of which closely reproduce experimentally observed values (~2.7 eV).

To further probe structure–property relationships, substitutional doping with P and Ni at identical lattice sites is evaluated across planar, buckled, and corrugated hosts, revealing distinct electronic responses that depend sensitively on the underlying structural motif. These findings demonstrate that structural degrees of freedom, especially buckling and stacking registry, are decisive parameters in computational screening, and their explicit inclusion is essential for achieving physically reliable and predictive models.

To be updated shortly..