Understanding the epsilon and zeta High-Pressure Solid Phases of Oxygen. Systematic Periodic Density Functional Theory Studies Using Localized Atomic Basis

摘要

The experimentally characterized epsilon and zeta phases of solid oxygen are studied by periodic Hartree-Fock (HF) and Density Functional Theory calculations at pressures from 10 to 160 GPa using different types of exchange-correlation functionals with Gaussian atomic basis sets. Full geometry optimizations of the monoclinic C-2/m (O-2)(4) unit cell were done to study the evolution of the structural and electronic properties with pressure. Vibrational calculations were performed at each pressure. While periodic HF does not predict the epsilon-zeta phase transition in the considered range, Local Density approximation and Generalized Gradient approximation methods predict too low transition pressures. The performance of hybrid functional methods is dependent on the amount of non-local HF exchange. PBE0, M06, B3PW91, and B3LYP approaches correctly predict the structural and electronic changes associated with the phase transition. GGA and hybrid functionals predict a pressure range where both phases coexist, but only the latter type of methods yield results in agreement with experiment. Using the optimized (O-2)(4) unit cell at each pressure we show, through CASSCF(8,8) calculations, that the greater accuracy of the optimized geometrical parameters with increasing pressure is due to a decreasing multireference character of the unit cell wave function. The mechanism of the transition from the non-conducting to the conducting zeta phase is explained through the Electron Pair Localization Function, which clearly reveals chemical bonding between O-2 molecules in the ab crystal planes belonging to different unit cells due to much shorter intercell O-2-O-2 distances.