Infinite-layer LaNiO2: Ni1+ is not Cu2+

Abstract

The Ni ion in $\mathrm{La}\mathrm{Ni}{\mathrm{O}}_2$ has the same formal ionic configuration $3d^9$ as does Cu in isostructural $\mathrm{Ca}\mathrm{Cu}{\mathrm{O}}_2$, but it is reported to be nonmagnetic and probably metallic whereas $\mathrm{Ca}\mathrm{Cu}{\mathrm{O}}_2$ is a magnetic insulator. From ab initio calculations we trace its individualistic behavior to (1) reduced $3d–2p$ mixing due to an increase of the separation of site energies $({\epsilon }_d–{\epsilon }_p)$ of at least $2\mathrm{eV}$, and (2) important $\mathrm{Ni}3d(3z^2-r^2)$ mixing with $\mathrm{La}5d(3z^2–r^2)$ states that leads to Fermi surface pockets of $\mathrm{La}5d$ character that hole dope the $\mathrm{Ni}3d$ band. Correlation effects do not appear to be large in $\mathrm{La}\mathrm{Ni}{\mathrm{O}}_2$. However, ad hoc increase of the intra-atomic repulsion on the Ni site (using the $\mathrm{LDA}+\mathrm{U}$ method) is found to lead to a correlated state: (i) the transition metal $d(x^2-y^2)$ and $d(3z^2-r^2)$ states undergo consecutive Mott transitions; (ii) their moments are antialigned leading (ideally) to a “singlet” ion in which there are two polarized orbitals; and (iii) mixing of the upper Hubbard $3d(3z^2–r^2)$ band with the $\mathrm{La}5d\left(xy\right)$ states leaves considerable transition metal $3d$ character in a band pinned to the Fermi level. The magnetic configuration is more indicative of a ${\mathrm{Ni}}^{2+}$ ion in this limit, although the actual charge changes little with $U$.