A Quantum Monte Carlo Study of the Structural, Energetic, and Magnetic Properties of Two-Dimensional H and T Phase VSe₂

Abstract

Previous works have controversially claimed near-roomtemperature ferromagnetism in two-dimensional (2D) VSe 2 , with conflicting results throughout the literature. These discrepancies in magnetic properties between both phases (T and H) of 2D VSe 2 are most likely due to the structural parameters being coupled to the magnetic properties. Specifically, both phases have a close lattice match and similar total energies, which makes it difficult to determine which phase is being observed experimentally. In this study, we used a combination of density functional theory, highly accurate diffusion Monte Carlo (DMC), and a surrogate Hessian line-search optimization technique to resolve the previously reported discrepancy in structural parameters and relative phase stability. With DMC accuracy, we determined the free-standing geometry of both phases and constructed a phase diagram. Our findings demonstrate the successes of the DMC method coupled with the surrogate Hessian structural optimization technique when applied to a 2D magnetic system.

Predictions of delafossite-hosted honeycomb and kagome phases

“…We follow the former approach. In diffusion Monte Carlo calculations of correlated oxides, a U value between 2 and 6 eV is often found 38–51 with a value near 3 eV occurring most often, and so we selected U = 3.0 eV in all calculations. In SCF and structural relaxation calculations a k -point grid of 13 × 13 × 13 per ABO 2 primitive cell (or finer) was used and relaxed structures were obtained with 10 −6 eV and 0.01 eV Å −1 tolerances on the total energy and maximum atomic force, respectively.…”

Accelerated Data-Driven Discovery and Screening of Two-Dimensional Magnets Using Graph Neural Networks

“…Nonetheless, many magnetic 2D materials exhibit strongly correlated effects and are better suited for more advanced methods that account for correlation in the outer electronic shells, such as DFT + U , − dynamical mean field theory, and quantum Monte Carlo . These methods have proven to be more accurate in describing materials manifesting strong correlation effects. − …”