Diamond -> beta-tin phase transition in Si within diffusion quantum Monte Carlo

TL;DR

This study uses diffusion quantum Monte Carlo methods to analyze the diamond to beta-tin phase transition in silicon, highlighting the impact of wave function choice, pseudopotentials, and vibrational effects on the transition pressure.

Contribution

It demonstrates the importance of wave function form and pseudopotential selection in accurately modeling phase transitions in silicon using DMC.

Findings

Backflow wave functions lower DMC energies.

Choice of pseudopotential significantly affects transition pressure.

Empirical pseudopotential yields transition pressure close to experimental value.

Abstract

We have studied the diamond -> beta-tin phase transition in Si using diffusion quantum Monte Carlo (DMC) methods. Slater-Jastrow-backflow trial wave functions give lower DMC energies than Slater-Jastrow ones, and backflow slightly favors the beta-tin phase with respect to the diamond phase. We have investigated the changes in the equation of state that result from the use of different pseudopotentials, the inclusion of either zero-point motion or finite-temperature vibrations, and the application of corrections for finite-size effects. Our tests indicate that the choice of pseudopotential can significantly affect the equation of state. Using a Dirac-Fock pseudopotential leads to an overestimation of the transition pressure but an empirical pseudopotential designed for use in correlated calculations gives a transition pressure in quite good agreement with experiment.