Machine learning interatomic potential study of grain boundaries thermal conductance and tensile strength in hexagonal boron nitride

Abstract

Chemical vapor deposition enables the scalable synthesis of large-area two-dimensional (2D) materials, but the resulting grain boundaries (GBs) can strongly influence their thermal and mechanical performance. In this work, we employ machine learning interatomic potentials (MLIPs) to systematically investigate thermal transport and tensile response across representative 5|7-based grain boundaries in monolayer hexagonal boron nitride (h-BN). The considered GBs span different defect densities and include both symmetric and asymmetric configurations containing B–B and N–N homonuclear bonds. The developed potential is first validated against first-principles calculations, showing excellent agreement for the phonon dispersion and tensile strength of pristine h-BN under thermal ground-state conditions. Non-equilibrium molecular dynamics simulations predict a room-temperature thermal conductivity of approximately 600 W/m·K for pristine h-BN, consistent with the range reported in previous experimental and theoretical studies. The grain boundary calculations reveal a clear increase in thermal conductance with decreasing defect density, indicating reduced phonon scattering as the spacing between neighboring defect pairs increases. In parallel, tensile simulations exhibit predominantly brittle failure, consistent with the strong covalent bonding characteristic of 2D h-BN. Interestingly, the mechanical response follows an opposite trend: grain boundaries with higher defect densities can exhibit distinctly higher tensile strengths. This counterintuitive strengthening is attributed to the more uniform arrangement of closely spaced defect pairs, which promotes a more homogeneous stress distribution and suppresses stress concentrations. As the defect density decreases and the separation between neighboring defect pairs increases, the tensile strength progressively approaches a saturation-like regime. These results demonstrate that grain boundary defect density can exert contrasting effects on thermal transport and mechanical strength, highlighting the critical role of GB topology in governing the thermal and mechanical properties of polycrystalline h-BN. The findings provide atomistic insights for investigating the thermal and mechanical properties of polycrystalline and defective 2D materials.