Method for locating low-energy solutions withinDFT+U

Abstract

The widely employed DFT+ U formalism is known to give rise to many self-consistent yet energetically distinct solutions in correlated systems, which can be highly problematic for reliably predicting the thermodynamic and physical properties of such materials. Here we study this phenomenon in the bulk materials UO 2 , CoO, and NiO, and in a CeO 2 surface. We show that the following factors affect which self-consistent solution a DFT+ U calculation reaches: ͑i͒ the magnitude of U; ͑ii͒ initial correlated orbital occupations; ͑iii͒ lattice geometry; ͑iv͒ whether lattice symmetry is enforced on the charge density; and ͑v͒ even electronic mixing parameters. These various solutions may differ in total energy by hundreds of meV per atom, so identifying or approximating the ground state is critical in the DFT+ U scheme. We propose an efficient U-ramping method for locating low-energy solutions, which we validate in a range of test cases. We also suggest that this method may be applicable to hybrid functional calculations.

Stratified construction of neural network based interatomic models for multicomponent materials

“…A way to relax the architecture-defined constraint for both fitting schemes is to link the inputs to higher-level neurons as well. In fact, generation of reliable DFT data is a greater challenge for strongly correlated transition metal oxides due to the known difficulty of defining the U parameter for the full composition range and obtaining converged DFT+U energies [89,90].…”

Computational study of the energetics of charge and cation mixing in U1−xCexO2Hanken1, Stanek2, Grønbech‐Jensen3 et al. 2011Phys. Rev. B370251View full textAdd to dashboardCite

“…An important issue that has been recently discussed in the literature [35][36][37] concerns the propensity for DFT+U calculations to converge to multiple self-consistent solutions corresponding to different orbital occupations. To ensure that such calculations converge to solutions that are the lowest-energy electronic states (or near them), several different methods have been proposed [35][36][37] .…”

“…To ensure that such calculations converge to solutions that are the lowest-energy electronic states (or near them), several different methods have been proposed [35][36][37] . In the current work we employ the approach described by Meredig et al 37 , which involves a slow localization of the f electrons in a series of DFT+U calculations with incrementally increasing values of U eff . Specifically, one begins by performing a GGA (with U eff =0) calculation until ionic positions and charge density are converged.…”

Transport properties of C and O in UN fuels

“…The idea behind the DFT+U method is to treat the strong on-site Coulomb interaction of localized electrons, which is not correctly describe by LDA or GGA, with an additional energy term (Hubbard) 17 . However, it is known that this method may introduce metastable states 18 with no physical meaning. This issue was extensively investigated for the UN system by Claisse et al 19 , where the occupation matrix control (OMC) scheme was used to avoid metastable states in this material 20 .…”