Alternative treatment of relativistic effects in linear augmented plane wave (LAPW) method: application to Ac, Th, ThO2 and UO2

TL;DR

This paper proposes new methods to incorporate relativistic effects in the LAPW computational approach, improving accuracy for actinides and oxides, and reveals their impact on material properties and electronic structures.

Contribution

It introduces alternative radial functions and correction schemes for relativistic effects in LAPW, enhancing the modeling of actinide and oxide materials.

Findings

Relativistic treatment affects lattice constants and elastic moduli significantly.

UO2 is characterized as a semimetal with a small forbidden gap.

New radial functions improve the description of filled 6p bands.

Abstract

We examine the influence of the relativistic effects within the linear augmented plane wave method (LAPW) for solids and propose a few alternative ways to accurately take them into account: (1) we introduce new radial dependencies for LAPW (Bloch-type) basis functions, based on two actual radial solutions of the Dirac equation for j=l-1/2 and j=l+1/2 states. The proposed radial 6p functions receive more weight from the Dirac p-1/2 solution and, due to this, can on average correctly describe completely filled $6p$ bands even without the additional p-1/2 local atomic function, as is done in the LAPW+p-1/2 method; (2) the canonical LAPW matrix elements for the spherically symmetric component of the potential, assuming non-relativistic radial wave functions, should be corrected; (3) we argue that for a realistic spin-orbit (SO) energy splitting of the semicore 6p-states the spin-orbit interaction constant zeta(p) should be calculated with the 6p-3/2 radial component, because the value of zeta(p) obtained with the canonical mixing of the 6p-1/2 and 6p-3/2 components overestimates the SO splitting. Different ways of taking into account relativistic effects can change the equilibrium lattice constant up to 0.15 A and the elastic modulus up to 26 GPa. We find that in the full treatment of the spin-orbit coupling UO2 has a small gap of forbidden states (0.2-0.4 eV) at the Fermi level, which persists for all k-vectors and, therefore, UO2 should be classified as a semimetal. We also discuss the peculiarities of the electron band structure of actinium, which result in an overestimation of its lattice constant.