Impact of the U parameter on the predicted defect chemistry of materials: The example of PuO 2 ± x

Density functional theory (DFT) provides a powerful tool for describing the electronic properties of materials, however, self-interaction errors in semilocal functionals complicate the accurate modeling of correlated materials. The DFT + U method is a popular and computationally cost-effective solution for mitigating self-interaction; though, this reduces the aspect of DFT calculations, as the outcome now becomes dependent on the user's choice of U parameters. Atomistic modeling of plutonium dioxide ( PuO 2 ) is employed to provide insight into its evolution in storage or into its properties as mixed-oxide fuel. There is no single U parameter that can reproduce all the experimental properties of PuO 2 accurately and, as such, it is important that the U parameter is selected with careful consideration. In this work, we use noncollinear DFT + U simulations to thoroughly examine the defect chemistry of PuO 2 using U = 4 and 7 eV , in order to understand the implication the choice of U parameter can have on the predicted defect chemistry. We find that both U parameters predict the same intrinsic defect chemistry, with the main discrepancy being on the preferred charge state of the oxygen vacancies: + 2 for U = 4 eV and neutral for U = 7 eV . Additionally, we show that the choice of U can impact the defect formation energies and preferred charge states of a dopant. When uranium is placed onto a plutonium lattice site, it tends to favor the neutral and + 1 charge state (indicative of U 4 + and U 5 + ) with U = 4 eV , whereas with U = 7 eV the − 1 and neutral charge states are more stable.