Reply to “Comment on ‘Conduction states in oxide perovskites: Three manifestations ofTi3+Jahn-Teller polarons in barium titanate’”

Abstract

A concise overview on the available theoretical and experimental data concerning the structure of oxygen vacancies in the oxide perovskites is given. In this context the statements made in the Comment are discussed.There is no doubt that neutral oxygen leaves an oxygen perovskite, OP, ͑ABO 3 ͒ crystal when annealed in a reducing atmosphere. For each oxygen atom lost two electrons remain in the crystal.What is the structure of the resulting defects? We start by characterizing the present knowledge concerning isolated oxygen vacancies as based on quantum chemical studies. Then we compare these different approaches and discuss the apparent discrepancies. In the light of these findings we assess the available relevant experimental data and point out directions for further research.Recently theoretical evidence cumulates (Refs. 1-4 and references therein) that the isolated oxygen vacancy ͑V O ͒ in the OPs has an electronic groundstate characterized by the full, unbroken, a 1g type, symmetry of the defect. This implies that there is equal electron density at both B-cation neighbors of V O , and finite density at this site is not forbidden. The results support earlier findings by Tsukada et al. 5 They are valid no matter whether V O contains one electron, as for BaTiO 3 1 and KNbO 3 , 2 or two electrons, investigated for KNbO 3 , 2 PbTiO 3 , 3 and SrTiO 3 . 4 These recent theoretical results, independently obtained by four groups employing rather flexible ab initio methods (for details see , are at variance with earlier ones using the Green function procedure (Refs. 6 and 7 and references therein). These predicted a groundstate symmetry having a nodal line along the axis of V O and xy-type symmetry at both B-neighbors, e.g., Ti 3+ ͑3d xy ͒. Based on phenomenological or semiempirical 7 arguments Prosandeyev et al. additionally claimed that the electron density is asymmetrically localized at only one of the two cation neighbors of V O . In a rather detailed way, Donnerberg et al. 1,8 modeled this situation, using their flexible method, and found that this corresponds to a higher energy than the a 1g state, ruling out the xy-type state as the theoretical groundstate of isolated V O . This argument was cited in the commented paper. 9,10 Not only the ground state symmetries but also the predicted level energies differ in both types of approaches: The a 1g states are deep, at least 0.4 eV below the conduction band edge, for the doubly filled V O in SrTiO 3 , 4 and they are found to be still lower for the other compounds and, of course, for the singly filled V O . 1,2 The xy-type states, on the other hand, are predicted to be shallow, about 0.3 eV below the conduction band edge, for V O containing one electron, and higher for the doubly filled one. 6,7 Electron paramagnetic resonance (EPR) can yield the most detailed structural information on defects. The oxygen vacancy in the OPs has been definitely identified in this way only when neighboring a paramagnetic transition metal ion, e.g., ͑Ni 3+ ͑3d 7 ͒ − V O ͒ 11 or other case...

Probing ferroelectricity in highly conducting materials through their elastic response: Persistence of ferroelectricity in metallic BaTiO3−δ

“…Each V O in undoped BaTiO 3 donates up to two electrons, which are generally localized on Ti, so that a completely ionized V O induces the Ti 3+ state in two Ti 4+ ions. 53,54 The external electronic shell of Ti +3 is 3d 1 , or an electron is doped in the empty 3d shell. The five degenerate d orbitals are already split into three T 2g (d xy , d yz and d xz ) and two E g (d 3z 2 −r 2 and d x 2 −y 2 ) orbitals by the octahedral environment of Ti, which can be further split by octahedral distortions, giving rise to the Jahn-Teller (JT) effect.…”

“…The five degenerate d orbitals are already split into three T 2g (d xy , d yz and d xz ) and two E g (d 3z 2 −r 2 and d x 2 −y 2 ) orbitals by the octahedral environment of Ti, which can be further split by octahedral distortions, giving rise to the Jahn-Teller (JT) effect. 55 The Ti 3+ ions are usually observed in acceptor doped titanates, 53,54 where, at low concentrations, they form polarons or bipolarons, without particular effects on the structure. It is also suggested that the JT distortions of Ti 3+ in BaTiO 3 may stabilize the hexagonal phase (h-BT) over the cubic one under strongly reducing conditions, but this occurs at the very high sintering temperatures ∼ 1400 • C, which can be slightly lowered by doping strongly JT active ions in the Ti site.…”

First-principles calculations of the atomic and electronic structure ofFcenters in the bulk and on the (001) surface ofSrTiO3

“…7,[20][21][22][23] For perovskite oxides such as SrTiO 3 , the electronic structure of the F-type defects is a subject of a long debate. 24 Several theoretical studies have suggested major and identical contributions from the 3d orbitals of the two Ti ions on both sides of O vacancy to the electron density of the F center, 13,19,26 although others indicate a partial or total localization of the electrons in the vacancy. 27 The position of the F center level in the optical band gap is also somewhat controversial 19,25 although indirect experimental data based on conductivity measurements suggest that the neutral F center is a quite shallow defect.…”

Identification of vacancy defects in a thin film perovskite oxide

“…However, convincing evidence is lacking that the isolated native vacancy defects have accessible EPR active states in perovskite oxides [9]. Direct imaging of O vacancies has been demonstrated using aberration corrected transmission electron microscopy (TEM) [10], or annular-dark-field TEM, supported by electron energy-loss spectroscopy (EELS) [11].…”