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

Fig. S1.jpg shows the effect of system length on the room-temperature interfacial thermal conductance of the representative 0h-N-s grain boundary. hBN.extxyz and hBN.cfg contain the AIMD datasets generated in this work, provided in extended XYZ (EXTXYZ) and CFG formats, respectively. In the CFG files, atom types 5 and 7 correspond to boron (B) and nitrogen (N), respectively. hBN.mtp is the trained MTP developed in this work for modeling the mechanical response and thermal transport of grain boundaries in monolayer h-BN. In this potential file, atom types 0 and 1 correspond to boron (B) and nitrogen (N), respectively. Light_hBN.mtp is a faster MTP version for accelerated simulations. AIMD.tar.gz contains sample AIMD input files in the native VASP format used to generate the training data. GB_structures.tar.gz contains the 12 constructed grain-boundary models in LAMMPS data format. The structures should be replicated along the width (Y) direction before simulation.