Description of Paramagnetism, gap formation and atomic
displacements in 3d oxide perovskites in polymorphous DFT
Energy Institute, University of Colorado, Boulder, Colorado 80309, USA
ABO3 perovskite oxides with a 3d transition metal element on the B site (“Mott Insulators”) have held the solid-state physics community in constant fascination ever since Mott and Peirles have characterized them as failures of band theory. These compounds are usually insulators in their high temperature spin-disordered paramagnetic phase, and insulators (or rarely metals) in their low temperature spin-ordered phases. Textbooks descriptions based on the celebrated Mott-Hubbard Hamiltonian usually ascribe this behavior to symmetry-conserving dynamical electronic correlations codified by the on-site repulsion energy U. Structural distortions such as bond disproportionation in RNiO3 or Jahn-Teller motions in LaMnO3 appears as secondary effects, not related to gapping.
Band theory was assumed for long to fail to describe Mott insulators, in particular in the PM phase. However, this failure occurred under rather restrictive assumptions, such as using in band theory a primitive unit cell containing but a single, ABO3 formula unit (a monomorphous description with a single local motif , so structural symmetry breaking could not be accommodated), along with a non-spin polarized description. Under such restrictions, a system with an odd number of electrons, unable to form magnetic states, must have a Fermi level that intersects a band, in contradiction with experiment.
However, such naïve (N) Density Functional Theory (N-DFT) does not represent what band theory can do. Indeed, recent work, to be discussed here, had shown that when the PM phase is described by a supercell, where different transition metal sites can ‘see’ different local environments (a polymorphous description) with local moments that add up to zero globally, but not necessarily locally, the system could lower its total energy by coupling to atomic displacements and magnetic moments. This can open gaps when such coupling exist ( most ABO3), but leaves the system metallic when such coupling is negligible (e.g SrVO3). In conjunction with exchange-correlation functionals that distinguish occupied from unoccupied states, such a polymorphous DFT applied to all ABO3 3d perovskites as well as 3d binary oxides NiO, MnO, FeO, CoO has shown that the observed trends in LT vs HT magnetism and gaping can be systematically accounted for by considering symmetry breaking modes that simultaneously lower the systems total energy. This is true whether one uses “U” as in in DFT+U, or even without U, using a functional that better cancels the self- interaction error (SCAN). This illustrates how the basic trends in Mott insulator ground state properties can be accounted for using single-determinant DFT, and exemplifies Mott Insulators without Mott-Hubbard U.