Core excitations of the solid oxygen ε phase: periodic hybrid density functional theory studies with localized atomic basis

Abstract

The e and the superconducting ζ oxygen phases at high pressures consist of (O2)4 cuboid unit cells. Previous lattice optimizations and vibrational calculations with periodic hybrid density functional theory using atomic basis sets addressed the evolution of the lattice parameters, Raman and infrared spectra as well as the e–ζ phase transition pressure; in all cases, the results found with hybrid exchange–correlation (XC) functionals are in excellent agreement with experiments. We address here the evolution of the core 1s–σ u * and 1s–1π g * transitions for the low-pressure regime of the e phase with GGA and hybrid-type XC functionals. Again, hybrid XC functionals yield a better agreement with K-edge results for the pressure evolution in the 10–40 GPa range for which experimental data are available. However, a slower dependence of the 1π g * excitation with pressure is found with respect to experimental data, irrespective of whether a GGA or hybrid-type XC functional is used. We provide arguments to explain the poorer theoretical description for the 1π g * excitation as arising from the rather strong multireference character of the unit cell singlet wavefunction in the low-pressure regime. Some general conclusions are drawn concerning the accuracy of hybrid versus GGA-type exchange correlation functionals for this complex system.