Empowering higher energy sodium‐ion battery cathode by oxygen chemistry

Sodium (Na) ion batteries (SIBs) promise low‐cost energy storage systems but are still restricted by insufficient energy density. Introducing oxygen (O) redox into the design of the Na‐storage cathode is presently considered an effective avenue to generate extra capacity in solving the energy density bottleneck. The succeeding issues are how to overcome the irreversible electrochemical behavior accompanied by O release. Meanwhile, the O redox chemistry and subsequent structural evolution remain ambiguous so far. Here, we deliberate on the O redox mechanism in Na‐storage transition metal oxides. Challenges associated with the reaction irreversibility and structural collapse are summarized by virtue of the advanced characterization techniques. Beyond that, strategies that potentially enhance the electrochemical properties of O redox and future research perspectives on exploring useable O redox cathode materials are outlined.


| INTRODUCTION
The emergence of intermittent renewable energy and grid peaking has created a substantial driving force for the development of large-scale electrochemical energy storage technology. Such application scenarios require rechargeable batteries that possess a cheap supply of materials and are easy to mass produce. [1] Although lithium (Li) successfully boosts the huge market of Li-ion batteries (LIBs), its cost risk remains unresolved due to uneven elemental distribution and geopolitics. Alternatively, sodium (Na) is an extremely abundant and accessible element capable of acting as the charge carrier shuttling between the electrodes. Na ion batteries (SIBs) therefore have a clear commercial imperative to address the cost demand. [2] Regrettably, Na has a larger atomic mass compared to Li, which inevitably lowers the theoretical specific capacity of electroactive materials for storing Na. This is particularly true for the cathode materials that play in battery capacity and cost. [3] Therefore, the research on high-specific-capacity cathode for Na storage is beyond doubt.
Generally, the redox center of cathode materials solely stems from transition metal (TM) ions involved with the valance change upon the Na + insertion/ extraction. As such, theoretical specific capacity is limited by the number of electrons that can be exchanged per unit mass of TM ions. [4] Recently, the strategies to increase the capacity of Na-storage cathodes have been gradually given with emphasis on the introduction of oxygen redox into the layered TM oxides (TMOs). [1] The main reason is that apart from the active centers from TM cations, [5] the oxygen (O) redox can provide excess electrons to achieve extra capacity during the cycling procedure. [6] In this context, understanding O-redoxbased electrochemistry for emerging high-energy-density SIB cathodes has begun to gain significant attention. In this perspective, we first deepen the understanding of the origin of O anion redox. Subsequently, the role of O redox in layered TMO cathode materials for SIBs and the associated challenges is analyzed. Potential solutions targeting structural collapse and O loss at highly desodiated states are proposed based on the information provided by advanced characterization techniques. Finally, prospects of the future research direction of O redox in high-energy-density SIBs are offered, shedding light on the development of layered TMO cathode materials.

| ORIGIN OF O REDOX
The O-redox electrochemistry has been discovered several years ago and is thought to occur in the TMO where the 2p-band of the O overlaps with the TM d-band. [3] The O redox is triggered due to the presence of free electrons in O that overlap the TM orbital. As such, upon charging, O 2− in the crystalline structure releases charge to transform into [O 2− /O 2 ] n− species. However, such O redox is unusable due to the irreversible TM-O coordinating environment deterioration and leads to the eventual release of gaseous O 2 . Given the fact that the coordination environments and TMs species largely affect the complex charge transfer system of O redox, the relevant triggering mechanism should be described.
It is generally understood that the relative energy of the TM versus O states determines which species (TM vs. O) are oxidized first upon charging. As the energy levels of those states depend on the species in a compound and their chemical bonding, the local coordinating environment in a crystal structure thus plays a crucial role in the redox process and the O redox involved. For both Li-ion and Na-ion layered TM oxides, the often-discussed prerequisite for O redox is the presence of diverse local O configurations [7] (AM-O-AM or AM-O-vacancy, AM: alkali metal), which results in a high-energy density of states in O 2p orbitals further that generate labile electrons of O redox activities ( Figure 1A-B). Besides, the promotion of reversible O redox in TMO is recognized as being triggered by the degree of metal-ligand (cation-anion) covalent bonding. In pure TM redox cathode materials, compounds consisting of TM and O usually result in a clear separation between the O 2p-band and the TM d-band. Therefore, relatively reversible O redox can be achieved by using TMs species with higher d-bond to overlap the O 2p orbitals. Correspondingly, the species generated by O oxidation are closely related to the amplitude of the TM-O charge transfer (Δ) with respect to the Coulombic interactions in the d shell (U). [9] This is coupled with the splitting of the O 2p into σ, σ* π, and π* with their energy change denoted as for σand π-bond states, respectively ( Figure 1E). The (O-O) n− peroxide species must be charge-balanced with the TM via its reduction, namely the reductive coupling mechanism. This reduction can be predicted by the Δ CT energy (difference in energy between the lowest TM d band and the O 2p). Following the horizontal blue line of Figure 1E, (O-O) n− species are stable as long as Δ CT > ∆ O−O σ (before ①). And the system undergoes reversible O redox because the electronic states involved in oxidation and reduction are similar. If the metallic band lies between the π* and σ* bands (between ① and ②), the (O 2 ) 2− peroxides can form and the system produces voltage hysteresis. When Δ CT is lower than ∆ O−O σ , the system evolves O 2 irreversibly concomitantly cationic reduction. [8] This also implies that the reduction of the product depends on the capacity used in the redox process.
In addition to the above-mentioned charge transfer systems with electronic structure and covalent degree, another method is to describe the O redox parameter as a hole in the O sublattice (h°). [8] Whether the system is disordered-coordinating or chemically substituted, the underlying effect on the O redox is the resultant distribution of the different numbers of non-bonding O 2p. The energy of nonbonding O 2p is primarily depending on the electrostatic field exerted by the surrounding cationic charges. Therefore, the lower the cationic charge around one O, the higher in energy nonbinding O 2p state in the electronic structure. The number of O holes generated on the O sublattice thus depends on the relative energy of the O sublattice and the total capacity available for O redox. From this aspect, the accompanying larger electrostatic field change leads to easier generation of O-holes during desodiation. This also explains why O redox can be realized in various Na-layered transition materials (especially P2-type Na-TMO cathode materials) compared to Li-rich materials. For homogeneous O networks, the reversible O redox means that the number of h°just works with the electron of TM ions to ensure the deintercalation of all AM ions. However, the anionic process should be associated with a larger polarization than the cationic process, due to the structural reorganization of the O network. Therefore, the O network at high oxidation evolve disproportionately into O 2− , (O 2 ) 2− , and O 2 .

| O REDOX IN THE SIB CATHODE
Recent attention is increasingly paid to the introduction of O redox into the SIB cathode due to their inherent natures. First, both Na-deficient and Na-rich cathodes are able to provide additional capacity through O redox. Second, the activated voltage for O redox in SIBs is generally lower than that required for LIBs. Furthermore, the larger ionic radius Na + can suppress TM migration. These factors make O redox more promising in SIB cathode.
Over the past few years, many theories have emerged referring to what exactly renders a specific TMOs O redox-active and reversible. Given the reason that the overlapping of TM d and O 2p orbitals is a prerequisite for the occurrence of O redox, different electron orbitals TMs are worth to be discussed. One typical example is Na x MnO 2 cathode material [10] , which has high operating voltage and reversible capacity owing to excellent redox characteristics. The potential slope region was derived from the Mn 3+ to Mn 4+ redox couple, while [O 2− /O 2 ] n− participated in the redox process during the plateau in the high-voltage range (Figure 2A). [2] X-ray photoelectron spectroscopy (XPS) also confirmed that  [6] (E) Qualitative electronic structure of transition metal oxides with reversible, hysteresis, and irreversible anionic redox. The vertical axis denotes the relative energy levels and the horizontal axis denotes the decrease in bond distance between O-O or increasing anionic capacity from left to right. Reproduced with permission: Copyright 2019. Springer Nature. [8] ligand formed in 4d/5d electrons of TM ions. The Na 2 RuO 3 , a typical 4d-TM-based layered cathode material, also has excellent cationic and anionic redox characteristics. When initially reported, the Na 2 RuO 3 material had a high capacity of 140 mAh g −1 at an average potential of 2.8 V versus Na/Na + . [13] Subsequently, when activates both Ru 4+ /Ru 5+ and [O 2− /O 2 ] n− redox in Na 2 RuO 3 , higher capacity was achieved because the large delocalization of the 4d orbital enables the strong covalent Ru-O bond. [14] The strong covalent is also capable of preventing the Ru migration from the TM layer to the interlayer position. As for 5d-TM-based layered cathode materials, typically Ir-based, that feature a stronger degree of covalency and higher orbital energy level with O redox, a more stable O coordinated environment to store and release the charge is achievable. Meanwhile, ordered O-O distance was observed in charged/discharged states by scanning transmission electron microscopy (STEM): The change of O-O distance with O-redox has been first affirmed in the study of Na 2 IrO 3 material [11] (Figure 2E). The high redox reversibility of Ir 4+ /Ir 5+ and [O 2− /O 2 ] n− is attributed to the stable intermediate phase in Na 2 IrO 3 . Besides, it is worth emphasizing that the reversibility and stability of the cation/anion redox in layered cathode materials can be further improved by introducing a small amount of AM ions (Li + , Mg 2+ , etc.) to occupy TM sites. [12] One typical example is Na 2/3 Mg 1/3 Mn 2/3 O 2 . The high reversible activities of O redox at different SOC were probed by advanced mapping of resonant inelastic X-ray scattering (RIXS) technology ( Figure 2F-G). These AM ions limit the driving force of TM migration and tune the O 2p orbital solitary electron density, eventually improving O redox activities, rather than directly initiating O redox.
Further, in recent years, structural optimization, especially the design of orderly structures, has provided a promising path for improving the reversibility of O redox. Compared with a disordered structure [15] , an ordered honeycomb has an increased capacity of 30%. Similarly, the uniformly distributed vacancy structure Na 4/7 [□ 1/7 Mn 6/7 ]O 2 (□: Mn vacancy) exhibits better O redox performance at 4.1 V, which is attributed to optimization of the nonbonded 2p orbital of O adjacent to Mn vacancy. [16] Such a uniform TM vacancy structure is of critical concern in designing highly reversible O redox cathode. Similarly, the structural ordering of other TMO components also exhibits a highly O redox reversibility. [17] For example, smaller voltage hysteresis and less O evolution can be achieved by forming a ribbon-ordered rather than a honeycomb-ordered Li/Mn layer in Na 0.6 [Li 0.2 Mn 0.8 ]O 2 cathode [1] .

| POTENTIAL SOLUTION
The O redox holds great promise to increase the specific capacity of Na-TMO cathodes for the higher energy density of SIBs. This scenario requires a highly charged environment to activate the O redox. However, such a high-voltage operating condition undoubtedly deteriorates structural stability. [18] For example, the voltage hysteresis between the charge-discharge curves caused by cation migration and disordered gliding of the O layer begins to appear and evolve after the first cycle. As accessing high states of desodiation, it is difficult to restore the distorted octahedral sites, leading to increasingly irreversible voltage polarization. Therefore, to suppress voltage hysteresis, structural modification in Na-TMO cathodes is necessary. An effective approach is to enhance the TM-O covalent bond strength by using 3d, 4d, and 5d TM ions (Cu, Ru, Ir, etc.). This method is capable of reducing the disorder gliding of the O layer and simultaneously achieving a lower potential O redox. [19] Designing nonbonding 2p orbitals of O neighboring TM vacancies □ to suppress TM migration and disordered gliding of the O layer is considered another feasible strategy. [20] Recent attention has been increasingly paid to orderly structural optimization that can play an instrumental role in alleviating the voltage hysteresis of O redox. Locally ordered TM ions and AM ions offer an ultralow-strain environment for lattices which ensures a highly reversible voltage profile.
Another  [21] Correspondingly, the benefits of suppressing O 2 release by replacing O with less electronegative sulfur or selenium moieties to increase the M(3d)-L(np) orbital overlap have also been demonstrated. [22] Further, the stabilization of TM-O bonds via doping metal ions with fully empty or fully filled d-orbital. For example, the TM ions sites have positive effects in stabilizing the process of O-redox which helps to reduce O 2 release. [23] Beyond that, the synergistic combination of the above potential solutions may further facilitate the better utilization of stable and reversible O-redox in high-energy-density Na storage cathodes.

| SUMMARY AND PERSPECTIVES
In summary, Na-layered TMO cathode materials, either Na-rich or Na-deficiency types, hold a great chance of triggering the usable O redox that offers an additional capacity for SIBs to increase their energy densities. Yet puzzles remain which in turn direct the future development of O redox. First thing first, an in-depth understanding of the O redox reaction pathways and the end products is required. Fortunately, some typical phenomena of the O redox have been identified through advanced characterization technologies [24,25] , including the evolutionary behavior of O redox (RIXS), the chemical valence state of O (X-ray absorption spectroscopy), and the paramagnetic interactions of O with their environment (nuclear magnetic resonance). Several possible final products including O holes, bulk O 2 molecules, and [O-O] 2− dimers have been proposed. However, a widely accepted reaction pathway and detailed O-species interaction will certainly call for more effort. Also, how to obtain more reversible O redox by impelling studies on d-orbital elements and superstructures is expected to be a major direction for future research.