The integration of hexagonal boron nitride (hBN) into next-generation 2D heterostructures requires a precise mapping of how structural defects modulate its exceptional thermal and mechanical properties. In this work, we utilize a machine-learned Atomic Cluster Expansion (ACE) potential to systematically investigate the structure–property relationships of monolayer hBN resulting from eight distinct defect types. Our simulations reveal a clear hierarchy of the impact of defects on thermal conductivity, where square-octagon ($4|8$) triplets and vacancies act as “super-scatterers” of phonons because of their massive local strain fields and disruption of planar continuity. Mechanically, we discover a strain-induced self-healing mechanism unique to Stone-Wales (SW) defect pairs; specifically, under uniaxial strain along the armchair direction, the work of tensile deformation drives the reorientation of all the homonuclear bonds back into the pristine hexagonal framework. This restoration, combined with the unfolding of initial lattice corrugations (“decrumpling”), results in an anomalous increase in fracture strain compared to the pristine limit. In contrast, all other defects lack this restorative capacity and trigger a “weakest-link” catastrophic failure. We also discover a new “hole” defect (H defect) that emerges when antisite nitrogen defects are strained along the armchair direction. These defects provide additional stability, increasing the fracture stress by $\approx 10\%$ compared with that under straining along the zigzag direction. These H defects remain stable after the strain is released. By bridging the gap between local atomic topology, lattice corrugation, and global failure limits, this study provides a framework to facilitate the functional design of defect-engineered hBN in high-performance flexible electronics and thermal management systems.

Close

PrevNext

View figure in article

• Defects
• Mechanical & acoustical properties
• Strain
• Thermal properties
• Boron nitride
• Monolayer films
• Machine learning
• Molecular dynamics