Electronic correlations in the ferroelectric metallic state ofLiOsO3

Abstract

LiOsO3 has been recently identified as the first unambiguous "ferroelectric metal", experimentally realizing a prediction from 1965 by Anderson and Blount. In this work, we investigate the metallic state in LiOsO3 by means of infrared spectroscopy supplemented by Density Functional Theory and Dynamical Mean Field Theory calculations. Our measurements and theoretical calculations clearly show that LiOsO3 is a very bad metal with a small quasiparticle weight, close to a Mott-Hubbard localization transition. The agreement between experiments and theory allows us to ascribe all the relevant features in the optical conductivity to strong electron-electron correlations within the t2g manifold of the osmium atoms. Introduction -Ferroelectric materials display a spontaneous polarization due to an inversion symmetry breaking induced by the macroscopic ordering of local dipole moments. Multiferroic materials are defined by coexistent and coupled magnetic and ferroelectric order. It is a natural expectation that ferroelectric (and consequently multiferroic) ordering can only happen in insulators in order to avoid the metallic charge carriers to screen out the ferroelectric polarization. The existence of "ferroelectric metals" challenging this expectation has been hypothesized in 1965 by Anderson and Blount [1], who have shown that ferroelectricity can occur in a metal as long as the electrons at the Fermi level are decoupled from the ferroelectric distortions. Under these circumstances, the ferroelectric ordering can take place through a second-order transition. In 2013 the first unambiguous realization of this proposal has been reported in LiOsO 3 , where a second-order transition leads to a ferroelectric ionic structure below 140 K while the material remains conducting [2]. This behavior is accompanied by a large residual resistivity which exceeds by two orders of magnitude that of prototypical metals as gold, and by a CurieWeiss like behavior of the magnetic susceptibility in the ordered phase which suggests the presence of almost localized magnetic moments, characteristic precursors of Mott localization [2].

Coexistence of polar distortion and metallicity in PbTi1−xNbxO3

“…Such materials should be very hard to obtain because the free carriers screen the long-range coulomb interaction, and therefore weaken the ferroelectric polar distortion. Nevertheless, not long ago, an unambiguous experimental evidence of ferroelectric-like structure transition in metallic material LiOsO3 has reattracted much attention on the metals with non-centrosymmetric structures, which was observed in low temperature below 150 K. [2][3][4][5][6][7][8] Yet, they remain challenging to discover. 8,9 So far no room temperature candidate for the coexistence of polar distortion and metallicity has been obtained.…”

Possible Origin of the Absence of Magnetic Order in LiOsO3: Spin–Orbit Coupling Controlled Ground State

“…In ref. , ( U = 2.3 eV, J H = 0.345 eV) are adopted in the LDA + DMFT calculation which can lead to a large local moment (≈2.5 μ B /Os) . Then a crucial question is what are the proper values of U (and J H ) for LiOsO 3 , which are decisive for the ground state in the first‐principles calculations.…”

Static and Fluctuating Magnetic Moments in the Ferroelectric Metal LiOsO3

“…[7] had formula Li 0.98 OsO 3 , it has been argued that the Li-vacancy level may need to be closer to a Li 0.94 OsO 3 composition in order to allow the ferroelectric metallic phase to exist [14]. It has also been suggested that strong electron-electron correlations bring LiOsO 3 close to a Mott-Hubbard transition, but these are not strong enough to produce insulating behaviour [12,13].…”

“…In high-defect structures, it is possible that magnetism may be destroyed entirely. In the case that the ferroelectric transition were to be caused by band hybridization, however, the AFM state is disfavoured as it would result in an insulating ground state instead of a metallic one [13].…”

“…Around 140 K, a centrosymmetric (R3c) to noncentrosymmetric (R3c) phase transition occurs in LiOsO 3 (which is equivalent to the ferroelectric transition in LiNbO 3 and LiTaO 3 [8,9]), involving a shift in the mean position of the A-site Li [10]. There have been several competing explanations for the mechanism behind this transition, including that it may be fundamentally order-disorder-like [7,11] or alternatively that it is caused by the hybridization of electronic orbitals [12,13].…”