Phase transformation in BN/diamond heterojunctions using machine-learned force fields

Cubic boron nitride (cBN) is a promising ultrawide bandgap material for high power electronics due to its superior mechanical strength, thermal conductivity, and huge breakdown field (Eb > 5 MV/cm). However, its use is limited by poor crystallinity and phase purity due to challenging growth conditions. cBN is often grown on diamond due to its low lattice mismatch (1.4%) and similar properties, but competing phases, especially hexagonal BN (hBN), are commonly observed at the interface. These phases introduce long-range effects that can hamper device performance and are difficult to model. Thus there is still a lack of detailed understanding of the phase stability of BN in this system. To study the pressure- and temperature-dependent phase-stability of the BN/diamond system, we developed machine-learned force fields (MLFFs) which are trained on ab-initio molecular dynamics simulations. We utilize these MLFFs to study the pressure-, temperature-, and defect-mediated phase transitions of BN in the heterostructure. Using MLFFs, we find that the presence of both BN defects and high pressures can aid the hBN to cBN phase transition. Our findings represent a pivotal step toward identifying the ideal growth conditions for cBN, paving the way for its controlled and scalable synthesis.