Diffusion Monte Carlo Study on Relative Stabilities of Boron Nitride Polymorphs

TL;DR

This study uses advanced diffusion Monte Carlo calculations to determine the most stable polymorph of boron nitride, resolving previous debates and confirming hexagonal BN as the thermodynamically favored form.

Contribution

The paper applies fixed-node diffusion Monte Carlo to accurately assess the relative stabilities of BN polymorphs, providing reliable reference energies and resolving conflicting theoretical predictions.

Findings

hBN is the most stable polymorph at 0 K and 300 K

Results agree with experimental data and high-level methods

Demonstrates FNDMC's effectiveness for van der Waals materials

Abstract

Although Boron nitride (BN) is a well-known compound widely used for engineering and scientific purposes, the phase stability of its polymorphs, one of its most fundamental properties, is still under debate. The ab initio determination of the ground state of the BN polymorphs, such as hexagonal and zinc-blende, is difficult because of the elusive Van der Waals interaction, which plays a decisive role in some of the polymorphs, making quantitative prediction highly challenging. Hence, despite multiple theoretical studies, there has been no consensus on the ground state yet, primarily due to contradicting reports. In this study, we apply a state-of-the-art ab initio framework - fixed-node diffusion Monte Carlo (FNDMC), to four well known BN polymorphs, namely hexagonal, rhombohedral, wurtzite, and zinc-blende BNs. Our FNDMC calculations show that hBN is thermodynamically the most stable among the four polymorphs at 0 K as well as at 300K. This result agrees with the experimental data of Corrigan~{\it et al.} and Fukunaga. The conclusions are consistent with those obtained using other high-level methods, such as coupled cluster. We demonstrate that the FNDMC is a powerful method to address polymorphs that exhibit bonds of various forms. It also provides valuable information, like reliable reference energies, when reliable experimental data are missing or difficult to access. Our findings should promote the application of FNDMC for other van der Waals materials.