Comparison of meta-GGAs, DFT+U corrections, and hybrid functionals for polaronic point defects in layered MnO2 , NiO2

Defects in a material can significantly tune properties and enhance utility. Hybrid functionals like HSE06 are often used to describe solids with such defects. However, geometry optimization (including accounting for effects such as Jahn-Teller distortion) using hybrid functionals is challenging for the large supercells needed for defect study. The proposed ${\mathrm{r}}^{2}\mathrm{SCAN}+\mathrm{rVV}10+$U$+{U}_{d}$ method, which is computationally much cheaper and faster than hybrid functionals, can successfully describe defects in materials with the proper choice of U (for the $d$ orbitals of the host atom) and ${U}_{d}$ (for those of the defect atom), as shown here for small polaron defects in layered transition-metal oxides. For a range of U and ${U}_{d}$ around literature values (from solid-state reaction energies) for a given transition-metal ion and its oxidation state, we find that this approach predicts localized polaronic states in band gaps, as hybrid functionals do. The layered materials birnessite $({\mathrm{K}}_{\mathit{n}}{\mathrm{MnO}}_{2},n=0.03)$ and ${\mathrm{K}}_{\mathit{n}}{\mathrm{NiO}}_{2},n=0.03$, with one K atom intercalated between layers in a supercell, are found to have one localized occupied ${e}_{g}$ polaronic state on the transition metal ion reduced by the insertion of the K atom, when the geometry is calculated as above using published U values. The expected Jahn-Teller distortion is not observed when $U={U}_{d}=0$. Layered cobalt oxide with additional potassium ions intercalated $({\mathrm{K}}_{\mathit{n}}{\mathrm{CoO}}_{2},n=1.03)$ is different due to a dramatic difference in electronic configuration of the defected Co(II) ion: A single extra K atom in the supercell leads to four localized electrons in the band gap, using standard U values, and even for $U={U}_{d}=0$.