First-principles study of iron oxide polytypes: comparison of GGA+ U and hybrid functional method

Iron oxides are materials of wide interest that exhibit diverse electric, magnetic, optical, and catalytic properties; therefore, many studies to gain complete understanding of their polytypic phase boundary have been pursued. However, first-principles investigations of iron oxides using conventional density functional theory (DFT) calculations often yield a gross error due to the strong electron correlation that is poorly described within (semi) local approximations. This limitation often can be overcome using either the Hubbard correction (DFT+U) or a hybrid functional DFT method. Here, we investigate the diverse polytypic phases of iron monoxide (FeO) by comparing DFT+U and the hybrid-functional method (particularly B3PW91). We found that both methods show reasonable agreement in predicting the properties of the experimentally observed phases (B1, B8, iB8, and B2). However, the DFT+U method overestimates the equilibrium volume of B1 phase and predicts the experimentally undiscovered B4 phases to be nearly as stable as the naturally abundant B1 phase. In addition, B3PW91 predicts a local Jahn-Teller distortion pattern of the B1 phase that is more similar than that predicted by DFT+U to the result of a reported low-temperature neutron diffraction experiment. Using B3PW91, which is considered more convincing, we further discuss that there is no clear phase boundary between the monoclinic and rhombohedral B1 phases under compression but that the compression gradually reduces the local anisotropy to yield a rhombohedral-like phase, which agrees with previous experimental diffraction results. We expect that our comprehensive study demonstrates the virtue of using hybrid-functional DFT methods, particularly in exploring various known and unknown polytypic phases of transition-metal oxides.