Oxygen Redox Activity in Cathodes: A Common Phenomenon Calling for Density-Based Descriptors

Abstract

First-principle computations are crucial for the understanding of the oxygen redox mechanism in lithium-excess transition metal oxide materials. An important tool for the assignment of the redox-active species is the projected density of states (PDOS). A topological analysis of the charge density, on the other hand, suggests substantial oxygen redox activity in many transition metal oxide compounds beyond the ones commonly associated with it. This can be linked to the shortcomings of the spherical approximation for ions in crystalline compounds used to compute the PDOS, which fails to describe the charge density topology in transition metal oxides and leads to an erroneous description of the nature of charge transfer compared to a charge density-based approach. The density-based approach, due to the nonspherical nature of its domains, is more apt to properly describe oxygen redox contributions. This raises the question how meaningful the commonly employed descriptors of oxygen redox activity are.

Density-Based Descriptors of Redox Reactions Involving Transition Metal Compounds as a Reality-Anchored Framework: A Perspective

“…In accordance with the charge self-regulation principle, the Bader charge analysis of these three Li-excess compounds reported in the DFT study in Ref. [24] shows significant changes in the near-oxygen charge density upon reduction for all compounds, while the TM charges are less affected by the electron transfer from Li (which in all cases is almost fully ionized). While quantitatively there is a slightly stronger charge gain on O upon lithiation of the oxygen redox-active LiMnO 3 , the computed charge state changes are overall remarkably similar to LiCrO 3 and LiVO 3 .…”

“…This effect was described as charge self-regulation of transition metal ions by Raebiger et al Consequently, the integrated charge density around TM centers remains nearly constant over a wide range of oxidation states, as was corroborated in Ref. [26] and several other computational studies since [22][23][24]27]. In Figure 2c, the charge stability is shown on the example of a Co dopant in Cu 2 O, where the charge density increase due to TM gap states, and the charge loss from valence band states becoming less TM-like, offset each other, keeping the integrated local charge nearly constant, while the dopant's FOS is changing with increasing total charge of the system.…”

“…The integrated PDOS can be expressed as a partial charge and their comparison in Ref. [24] does not show a significant charge gain on O, while there is a more pronounced change for the TM partial charge in LiCrO 3 and LiVO 3 upon reduction. It should be noted that for plane-wave basis sets, which are ubiquitously used for DFT computations in the solid state, an appropriate radius for the projection in Equation (3) needs to be set.…”

“…For computations utilizing the projector-augmented wave (PAW) method [109], the PAW augmentation sphere radii are often used for the PDOS projection, leading to a severe underestimation of the charge present in the cell, especially around O, as shown in Ref. [24], where oxygen partial charges were found positive with the default projection volumes. Increasing the radii to match the Bader basin volumes of the different species improves the total charge within the simulation cell but does not reproduce the same charges and charge state changes found from the Bader analysis.…”

“…The Fermi level is indicated by a dashed line and spin-up and spin-down DOS plotted as negative and positive values, respectively. Reprinted with permission from Ref [24]…”