Reassessing structural models of graphitic carbon nitride for reliable photocatalytic predictions

Abstract

Among emerging photocatalysts, graphitic carbon nitride (g-C3N4) has attracted considerable attention as a metal-free, visible-light-active semiconductor for sustainable hydrogen production and environmental remediations. Here we demonstrate that many of the obvious models for the layer structure and preferred stacking arrangements of g-C3N4 lead to incorrect predictions of photocatalytic properties. We present a systematic structural exploration of monolayer and bulk geometries of heptazine-based g-C3N4 using first principles density functional theory (DFT) to assess their stability and elucidate how these configurations influence their electronic properties. Across all systems investigated, we show that using a buckled heptazine structure—rather than the conventional planar geometry—represents the true energetic ground-state structure. A buckled structure is also found by reverse Monte Carlo analysis of experimental data, also reported here. Further analysis of buckling and different stacking registries in g-C3N4 layered structures reveals newly identified low-energy corrugated stackings (with P1 symmetry) that give rise to a broad range of electronic band gaps, several of which closely match experimentally reported values of ∼2.7 eV. Moreover, introducing non-metal P and metal Ni dopants at identical lattice sites across 2D-planar/buckled and 3D-corrugated hosts shows distinct electronic responses. Our findings demonstrate that accurate structural modelling, including buckling and stacking registry, is crucial for reliably capturing the stability and tunable electronic properties of g-C3N4.