Recent Advances in Electrode Materials with Anion Redox Chemistry for Sodium-Ion Batteries

The development of sodium-ion batteries (SIBs), which are promising alternatives to lithium-ion batteries (LIBs), offers new opportunities to address the depletion of Li and Co resources; however, their implementation is hindered by their relatively low capacities and moderate operation voltages and resulting low energy densities. To overcome these limitations, considerable attention has been focused on anionic redox reactions, which proceed at high voltages with extra capacity. This manuscript covers the origin and recent development of anionic redox electrode materials for SIBs, including state-of-the-art P2and O3type layered oxides. We sequentially analyze the anion activity–structure–performance relationship in electrode materials. Finally, we discuss remaining challenges and suggest new strategies for future research in anion-redox cathode materials for SIBs.


Introduction
Lithium-ion batteries (LIBs) are one of the most efficient energy storage devices to power not only portable electronics but also electric vehicles owing to their high energy density and good cycle life. The electrification of vehicles has confirmed the feasibility of LIBs as medium-or large-scale energy devices; hence, the application of LIBs is being expanded toward grid-scale applications to store electricity generated from renewable applications or power plants. This rising demand directly affects the fluctuation of prices of raw materials such as lithium, cobalt, and nickel resources. Economical and ethical concerns for mining raw materials encourage us to search for alternatives to LIBs. Recently, sodium-ion batteries (SIBs) have reemerged as alternatives to LIBs [1,2], with emphasis on the merit of the raw material costs; namely, lithium resources are unevenly distributed on the Earth's crust, whereas sodium is present everywhere. Expensive cobalt and nickel elements are the main redox centers for LIBs, whereas inexpensive manganese-based moieties can provide stable electrochemical activity for SIBs. Because of the difference in the standard electrode potential between Li (−3.04 V versus standard hydrogen electrode (SHE)) and Na (−2.7 V versus SHE), there is evident loss of energy density for SIBs compared to that for LIBs when assuming that both systems deliver the same specific capacity [3]. This fact encourages the rational design of high-capacity cathode materials for SIBs to achieve capacities that are comparable to those for LIBs.
The large sodium ion (1.02 Å) can be stabilized in both prismatic (P) and octahedral (O) environments in layered structures, in which the former has a larger sodium layer than the latter owing to the size of the prismatic environment. The oxygen stacking sequence is combined with the environment of sodium ions in the sodium layer, for example, P2, P3, and O3, as proposed by Delmas et al. [4]. O3 layer compounds are stabilized when the sodium content ranges between 0.9 and 1 in Na x TMO 2 (TM: metal), where the TM has an average oxidation state of~3+; however, sodium-deficient Na x TMO 2 (0:7 ≤ x ≤ 0:4) compounds are crystallized into P2 and P3 structures that show average oxidation states of Me over 3.3+. This affects the first charge capacity; namely, the first discharge capacity is always higher than the charge capacity for P2 and P3 compounds because of the sodium deficiency in the sodium layers. Thus, additional presodiation or the use of additives is needed to increase the first charge capacity and achieve a coulombic efficiency (CE) close to 1 [5,6]. The large size of sodium ions also induces successive structural changes during the extraction and insertion of sodium ions. The resulting interlayer distance of the O3 layer structure is smaller than those of P2 and P3, such that it is common to see more phase transitions in O3 layered compounds. Therefore, it is generally accepted that P2-and P3type layered cathode materials typically deliver higher capacity, reaching approximately 200 mAh g −1 , with better capacity retention than O3 layered compounds, which usually have capacities of approximately 120 mAh g −1 based on the redox reaction of transition metal elements.
Indeed, the capacity contributed by transition metal elements is limited in those layered compounds. The anionic redox process enables delivery of additional capacity, such that sodium ions can be additionally de/intercalated from/into the structure; namely, the combination of cationic and anionic redox reactions provides more capacity. This type of chemistry has been demonstrated in Li-rich manganese oxide systems (Li 2 MnO 3 [7][8][9], Li 1.2 TM 0.8 O 2 [10][11][12], and their derivatives [13][14][15][16]) that have provided capacities to their theoretical limit. These anionic redox can contribute to additional capacity, thereby increasing the specific energy density of the battery. This phenomenon has been observed in overstoichiometric lithium compounds, where lithium partially replaces transition metals (TMs) with typical feature of Li[Li x TM 1-x ]O 2 (TM: Ni, Co, Fe, Cu, etc.) [12][13][14] or Li 1 +x TM 1-x O 2 (TM: Ru and Ir) [15,16]. These materials provide a higher capacity than the theoretical value obtained from a redox pair TM. The former compounds based on 3d metal (Mn) are of interest due to their high capacity in excess of 250 mAh (g-oxide) -1 . However, they still suffer from voltage drop and irreversible capacity loss associated with migration of cations in the crystal lattice and the release of oxygen during the charging process. The latter compounds based on 4d (Ru) and 5d (Ir) metals have attracted considerable attention from a fundamental and theoretical point of view. It was found that the transition from 3d (Mn) to 4d (Ru) or 5d (Ir) metals can increase the covalence of TM-O and stabilize oxygen-redox reactions. This higher covalence increases the structural rigidity and reduces the stress associated with the removal of lithium ions from the structure. This concept is also applicable to SIBs, which has triggered the intensive investigation of cathode materials from their synthesis to the elucidation of the mechanism of the anionic redox ( Figure 1(a)). During deintercalation of charge carriers, the Fermi level lowers to the top of O 2p bands, spontaneously generating excess electrons accompanying deintercalation of charge carriers to prevent structural collapse, and vice versa during intercalation (Figure 1(b)). That is, electron transfer is solely contributed by the oxygen activity, of which the resulting TM-O distance becomes shortened as a result of the oxidation of oxygen. Therefore, additional capacity is delivered along with the contribution by the redox reaction of transition metal elements. Electrochemical oxidation of anions is more facile than the reaction during reduction, and this approach can balance the abnormal CE during the first cycle, potentially obviating the need for presodiation or additives to compensate for the low first charge capacity for P2 and P3 layered compounds.
The selection of elements in transition metal layers is of great importance in improving the reversibility and controlling the operation voltage of the anionic redox reaction, namely, Na x [A y TM 1-y ]O 2 (A: Li [17][18][19][20][21][22][23][24][25][26][27][28][29], Na [30][31][32][33][34][35][36][37][38][39][40][41][42][43][44][45], Mg [46][47][48][49][50][51][52][53][54][55][56], Zn [57][58][59][60], Ni [61][62][63][64][65][66][67][68], Cu [67][68][69][70], Fe [71,72] or vacancy [54,55,[73][74][75][76][77][78][79][80][81]; TM: Mn [17-29, 46-70, 73-85], Ru [30-42, 85, 86], or Ir [43][44][45]87]). There are widely accepted requirements for oxygen-redox reaction, namely, a local Na-O-A coordination medium for the redox activation of oxygen in layered Na x [ATM]O 2 for P2, P3, and O3-type layered compounds. The presence of the Na-O-A configuration triggers anionic reactions that depend on structures through the irreversible release of O 2 , reversible redox process, and hysteresis process. The migration of these Li and Na in the transition metal layers to the Na layers causes the formation of a lone pair of electrons in the O 2p orbital, so that the high density of state energy for the oxygen allows the oxidation of oxygen although the reaction is kinetically slow. Herein, we review the current status of research and remaining challenges for the anionic reaction and associated mechanisms for different structures and elements of cathode materials for SIBs. 3 Energy Material Advances capacity was also sufficiently high (approximately 200 mAh g −1 ) and was retainable with cycling. The authors proposed that partial oxygen loss from the lattice and inplane rearrangement by partial lithium extraction from the TM layers activates the inactive tetravalent Mn to deliver high capacity. The authors also claimed that the Li in the TM layers is responsible for the increased operation voltage compared with that of Na 2/3 MnO 2 . Later work by de la Llave and coworkers [21] explored the origins of the good electrode performance of P2 Na 0.6 [Li 0.2 Mn 0.8 ]O 2 , which delivered a capacity of~190 mAh g −1 . Their thermodynamic investigation demonstrated an energy state of oxygen located close to the Fermi level, which enabled oxidation of oxygen as sodium was extracted from the layered structure ( Figure 2(b)). Yabuuchi et al. [17] proposed that lithium from the TM layers migrates toward the sodium layer when the octahedral environment is present at a highly desodiated state such as the O2 or OP4 phase. In this state, the density of states for O 2p was located at a higher energy state than that for Mn 4+ 3d [25]; however, the net charge of Mn did not vary during the extraction of sodium ions in the structure (Figure 2(c)). In addition, oxygen was not released from the oxide lattice as the oxidation of oxygen progressed (Figure 2(d)). The continuous reduction of the a-axis parameter was also indicative of the gradual oxidation from O 2− to O n− (n < 2) during charge, compensating for the charge imbalance occurring in Mn 4+ . Li et al. [24] observed the migration of lithium using NMR in Na 0.72 [Li 0.24 [20]. However, the aforementioned P2 Na x [Li y Mn 1-y ]O 2 compounds exhibited hysteresis between charge and discharge, as evident in Figure 2(a).

Sodium-Deficient Layered Structures
House and coworkers [27] found that the migration of lithium causes in-plane migration of manganese in the structure on charge, such that the lithium in the Na layers moves back to different sites of the TM layers. This disordered arrangement of Li and Mn in the structure is thought to be one of the reasons for the hysteresis in the layered P2 Na x [-Li y Mn 1-y ]O 2 compounds with honeycomb structure (Figure 2(a)). They also compared the hysteresis using superstructured P2 Na 0.6 [Li 0.2 Mn 0.8 ]O 2 (Figure 2(e)). It is worth mentioning that a flat voltage plateau, induced by the O 2-/nredox pair at~4.1 V, was observed during charge and discharge for the superstructured P2 Na 0.6 [Li 0.2 Mn 0.8 ]O 2 . In this case, Mn migration is suppressed although lithium moves to the Na layers (Figure 2(f)); the improved in-plane ordering is responsible for the reversible flat plateau with high operation voltage on charge and discharge. Notwithstanding, the behavior became less evident as repetitive lithium migration and recovery progressed because of the in-plane disorder in the superstructure.
This series of cathode materials utilize less than 36% oxygen in the lattice, implying that most cases have a high average oxidation state of Mn of higher than 3.5+. Because Jahn-Teller distortion is predominant in Mn 3+ O 6 octahedra, it is mainly observed in the deeply discharged state, which usually accompanies the formation of the P'2 structure. However, the aforementioned P2 Na x [Li y Mn 1-y ]O 2 compounds first reduce the oxidized oxygen (O n− ) to O 2− in the high-voltage region, after which Mn initiates its reduction from Mn 4+ to Mn 3+ until the end of discharge. Through this process, the average oxidation state of Mn is usually higher than 3.5; therefore, the Jahn-Teller distortion appears less dominant at the end of sodiation, Na x [Li y TM 1-y ]O 2 (0:9 ≤ x ≤ 1, y ≤ 1/3), without notable formation of the P2 structure, which is affected by the presence of Jahn-Teller distortion.
Cao et al. [29] stabilized the oxygen-redox chemistry in P2 Na 0.66 [Li 0.22 Ru 0.78 ]O 2 . Lowering of the transition metal d energy is possible using 4d or 5d elements such as Ru and Ir; hence, the oxygen redox can be active in the nonbonding state of O 2p. Ru 4+ was first oxidized to Ru 5+ , and further desodiation led to oxidation of oxygen, resulting in additional capacity but no evolution of O 2 gas from the crystal structure. The series of reactions resulted in a discharge capacity of 160 mAh g −1 . As more covalent character is seen for the elements on the right side of the periodic table, the migration of Li to the Na layers was suppressed even though the oxygen redox was dominant in the Z-phase region, like the O2 or OP4 phase that provides octahedral coordinates, for the highly desodiated state. The suppression of lithium migration and lack of O 2 release enabled reversible electrochemical reaction for 500 cycles.  [46] proposed the highly reversible electrochemical activity of P2 Na 2/3 [Mg 0.28 Mn 0.72 ]O 2 , which does not include vacancies in the TM layers and for which Mn has a valence state of 3.85 + . It was proposed that 0.15e − could be used to induce the change of Mn to Mn 4+ during charge, which may result in~40 mAh g −1 of capacity assuming a Mn 3+/4+ redox. Surprisingly, the compound delivered a charge capacity of 150 mAh with a plateau over 4.1 V undergoing a phase transition from P2 to O2, whereas the recovery on discharge was 210 mAh g −1 even though there was a gradual decay in the capacity with cycling (Figure 3(a)). They reasoned that oxygen-related activity or the partial loss of oxygen was associated with the activity of the compound, although a small portion derived from the oxidation of Mn 3+ to Mn 4+ (~15%) was available for the first charge capacity. Clement et al. [47] observed structural stability in Na x [Mg y Mn 1-y ]O 2 (y = 0 -0:1), which led to the suppression of the Jahn-Teller distortion by Mn 3+ and potential Mn 3+ /Mn 4+ ordering in the structure. Later, Maitra et al. [48] confirmed the availability of the oxygen-redox chemistry in P2 Na 2/3 [Mg 0.28 Mn 0.72 ]O 2 , which does not require excess alkali metal such as lithium in the TM layers of the compound. According to their combined studies of O K-edge X-ray absorption (XAS) and resonant inelastic X-ray scattering (RIXS), the voltage plateau above 4.1 V, with the phase transition from P2 to O2 or OP4, can be attributed to the reaction of the electron-  (Figure 3(c)). Hence, the authors proposed that lithium migration results in oxygen underbonding in the lattice, which triggers oxygen loss together with lithium loss from the surface of Na 0.78 [Li 0.25 (Figure 3(d)). The emission energy emerging at 523.7 eV repeatedly appeared at different charge ends, such as the 1st, 10th, 50th, and 100th cycles, whereas the signal was not visible at the end of discharge. This finding indicates that the obtained capacity can be attributed to both cationic and anionic redox processes during cycling.

Energy Material Advances
The presence of vacancies □ in TM layers also helps accelerate the oxygen-redox reaction by forming Na-O-Mg and Na-O-□ Mg configurations [54,55]. Interestingly, the Mg-deficient P2 Na 0. 63  In addition, it is interesting to see the additional voltage plateau over 4 V induced by the oxidation of oxygen in the lattice, even though the P2 phase was dominant in the plateau region. Their operando XRD study confirmed that the variation in the a-axis observed for Na 0.63 [Mg 0.143 Mn 0.820 □ Mg0.036 ]O 2 occurred after the reaction associated with the oxidation of Mn 3+ to Mn 4+ . The same effect was observed in Na 2/3 [Mg 1/9 Mn 7/9 □ Mg1/9 ]O 2 , with not only the absence of the phase transition but a higher capacity than that of the vacancy-free compound [55]. DFT calculation indicated that the oxygen close to the vacancies (Na-O-□ Mg ) provides more charge than the oxygen coordinated with Mg (Na-O-Mg) on charge, resulting in more charge compensation with the presence of vacancies that induce the lower-level voltage plateau for the oxygen reaction. Therefore, these properties were responsible for the increased capacity with improved reversibility in these vacancy-containing compounds [51].
Efforts have been made to improve the sluggish oxygenredox reaction in terms of the operation voltage, capacity retention, and rate capability by introducing transition metals (Ni [51] and Co [51]). Tapia-Ruiz et al. [49] demonstrated that increased operation voltage was achieved with partial substitution of Mg 2+ by Ni 2+ , Na 2/3 [Mg 1/3- , with the added Ni resulting in the suppression of the P2 to O2 phase transition. Kim et al. [51] utilized the nature of the overlapping of density of states between cobalt 3d and oxygen 2p observed in LiCoO 2 , facilitating electron transfer. The effect of Co in Na 0.6 [Mg 0.2-Co 0.2 Mn 0.6 ]O 2 resulted in not only a high discharge capacity, 214 mAh g −1 , including the capacity provided by the oxygen redox, but also good capacity retention for 1000 cycles at a rate of 5C (1.82 A g −1 ). Their pDOS demonstrated the effect of Co, which sufficiently lowered the bandgap energy to 0.61 from 2.65 eV, and the high DOS energy of oxygen relative to that of Mn and Co (Figure 3(e)). However, further investigation of these compounds is suggested to minimize the polarization during the oxygen redox process observed in the OP4-phase region. likely to induce covalence in the structure. Zn migration or oxygen loss was not observed during de/sodiation, which may result in an increase in the Na-O-Zn bonding, resulting in an unpaired electron that can trigger O 2−/1− redox, which agrees with the results proposed by Bai et al. [58] and Zheng et al. [59]. Although a strong covalence was expected, the phase transition from P2 to OP4 was inevitable in the highvoltage region. Importantly, the average oxidation Mn was higher than 3.5 + , which can effectively minimize the cooperative Jahn-Teller effect during cycling. These intrinsic properties resulted in good capacity retention upon cycling (Figure 4(b)); however, further elaboration is required to not only improve the rate capacity but also raise the operation voltage. Konarov et al. [60] circumvented these demerits of P2 Na x [Zn y Mn 1-y ]O 2 by replacing half of the Zn 2+ with Ni 2+ . The observed average discharge voltage was approximately 3.5 V, and their Na 2/3 [(Ni 0.5 Zn 0.5 ) 0.3 Mn 0.7 ]O 2 was able to deliver a capacity of over 70 mAh g −1 even at a rate of 10C ( Figure 4(c)). Such improvement was attributed to the presence of the Ni element, which provided improved electrical conductivity and activity of the Ni 2+/4+ redox reaction. The desodiation induced the oxidation of Ni 2+ toward Ni 4+ , and the oxidation of oxygen was also confirmed by XAS when Ni was oxidized to 4 + , which corresponds to the plateau seen over 4.1 V. Later, Cheng et al. [62] and Dai et al. [67] employed mRIX to confirm the oxygen activity in Na 2/3 [Ni 1/3 Mn 2/3 ]O 2 . Similarly, the two-electron reaction was dominant for Ni, whereas the upper voltage plateau over 4.1 V was mainly governed by the oxygen-redox reaction. The signature seen at 523 eV in emission energy was not visible after discharge. Zuo et al. [66] proposed a possible process for the oxygen redox behavior using pDOS data. In the highly desodiated state, the energy of O 2p in the e g * (Ni-O) becomes higher than that of Ni3d, which triggers the oxygen-redox reaction. Zhang et al. [65] observed O 2 release from the structure at the highly charge state, which produces a dense Ni 2 Mn 2 O 7 layer on the outer surface of Na 2/3- This, in turn, plays a role in impeding Na + diffusion, causing irreversible capacity in the first cycle. They circumvented the O 2 release by introducing a small amount of Fe 3+ , Na 2/3 [Fe 2/9 Ni 2/9 Mn 5/9 ]O 2 (Figure 4(d)), such that the first irreversible capacity was dramatically reduced to approximately 4%. According to their thermodynamic calculation, such improvement was related to the redistribution of electrons in the Fe-O-O configuration, for which the energy of oxygen in the pDOS was above the Fermi energy level after  [64]. Note that a prerequisite for oxygen redox is the unhybridized O 2p orbital that is     Figure 5(a)) [88,89]. Even though the capacity was limited to approximately 100 mAh g −1 in the operation range of 3-4.5 V, it is worth noting that the obtained capacity was activated by the pure O 2−/1− redox pair. The associated two-phase reaction was responsible for the flat charge and discharge curves; however, the desodiated new phase was refined as another P3 phase with a smaller interlayer distance. Therefore, the migration of Li and Mn from TM layers to the Na sites was difficult due to the larger trigonal prismatic site than the octahedral one, as revealed using neutron diffraction. This immobile feature, in turn, resulted in suppression of O 2 release from the lattice. Wu et al. [90] visualized the activity of oxygen for P3 Na 0.6 [Li 0.2 Mn 0.8 ]O 2 , which was available for cycling ( Figure 5(b)). They also pointed out that the observed nonlattice reaction on charge affected the irreversible capacity. Therefore, the P3 Na 0.6 [ (Figure 3(a) [95]. Investigation of the oxygen-redox reaction of P3 has not progressed as much as that for P2 compounds. The voltage plateaus attributed to the oxygen-redox reaction were generally more sloppy and shorter than those observed in the P2 structure. In addition, the cyclability was inferior to that in P2 compounds. These findings may be related to the low crystallinity of the P3 structure compared with that of the P2 compound. In addition, the irreversible O 2 evolution, which may induce the formation of a byproduct like Na 2 CO 3 , would result in the growth of CEI layers on the surface of cathodes. The suggested concomitant deterioration of electrodes can be circumvented by surface engineering of the active materials to make the oxygen-redox reaction more sustainable throughout cycling. superstucture) is displayed in Figure 6(a) [74]. Na 2 Mn 3 O 7 , with a valence of Mn 4+ , crystallizes in the triclinic P 1 structure, with half of the sodium ions occupying the distorted prism sites (P) and the other half occupying distorted octahedral sites (O). Na 2 Mn 3 O 7 possesses a notable performance with the smallest voltage hysteresis (~50 mV) among the known P2/P3/O3-type cathode materials with oxygen redox (Figure 6(b)) [74]. Furthermore, multiple reports have shown that the delivered capacity is~210-220 mAh g −1 in the voltage range of 1.5-4.5 V, which is one of the highest attainable capacities for sodium cathodes with simultaneous cationic and anionic redox processes [75][76][77]. Despite the promising low-voltage hysteresis and high capacity, Na 2 Mn 3 O 7 still suffers from degradation during cycling. Furthermore, the mechanism of the anion redox reaction with such a small hysteresis is still under debate [78,79]. Na 2 Mn 3 O 7 was initially studied by Adamczyk and Pralog in the low-voltage range, and it showed a specific capacity of 160 mAh g −1 at~2 V, based only on the Mn 3+ /Mn 4+ redox reaction [73]. Later, several groups showed that the desodiation of Na 2 Mn 3 O 7 at high voltage was governed only by the oxygen-redox reaction [75][76][77]. Inherent manganese vacancies in the TM layer lead to nonbonding 2p orbitals of oxygen □ Mn , giving extra oxygen-redox capacity of~70-80 mAh g −1 at high voltage without making the Mn-O bond labile (Figures 6(c) and 6(d)) [74]. In situ synchrotron XRD revealed negligible structural changes in the high-voltage 9 Energy Material Advances range, which differs from the behavior of common P2-O2 or P2-OP4 phase transitions in P2-type materials [75]. The high structural stability of the material at high voltage (100% after 45 cycles) was explained by the presence of intrinsic vacancies, which lead to a MnO 6 octahedron with six different Mn-O bond lengths that is more robust to desodiation. Thus, the [□ 1/7 Mn 6/7 ]O 2 slab can self-regulate its deformation and improve the structural stability of the material. However, at low voltage, which is attributed to the manganese Mn 3+ /-Mn 4+ redox reaction, a distorted lattice appeared, leading to the formation of a new phase [77]. This distorted structure was induced by the strong Jahn-Teller effect, associated with the presence of Mn 3+ upon additional sodium insertion in the low-voltage region.

P 1-Type
In light of the common high-voltage hysteresis of oxygenredox reactions, the origin of the impressive low-voltage hysteresis in Na 2 Mn 3 O 7 was studied by Song et al. using ex situ/in situ paramagnetic resonance and XRD [80]. The authors claimed that the well-maintained oxygen stacking sequence together with the absence of irreversible gliding of the oxygen layers and cation migrations resulted in the highly reversible oxygen redox with a negligible voltage hysteresis between charge and discharge. The authors showed that Na-ion extraction from the octahedral site is an essentially zero-strain process, proceeding through a single-phase reaction. The extraction of sodium ions from the prismatic site occurred through a two-phase reaction (P 1 -R3) with the shrinkage/expansion process of the vacant MnO 6 octahedron and a larger volume change during charge/discharge (Figure 6(e)). However, the overall stacking sequence of oxygen ions was barely changed during both steps of Na extraction/insertion, which differs from the behavior in P2/P3 cathodes, where extraction/insertion of Na lead to the formation of O-P phases with simultaneous decrease of the interlayer distance.
In a recent work, Tsuchimoto et al. confirmed a unique behavior of O in Na 2 Mn 3 O 7 using DFT calculations and magnetic and spectroscopic measurements (Figure 6(f)) [78]. The existence of thermodynamically favorable O −• over the peroxide-like O 2 2− dimers was predicted by computations and showed that hole stabilization occurred through a (σ + π) multiorbital Mn-O bond. Similar predictions were reported in the work of Kitchaev et al., where a π-bonded Mn-d and O-p orbital network formed a collective delocalized redox center [81]. Therefore, the authors concluded that such a π-network rather than any local bonding environment was responsible for the two-step voltage profile with a low-

Stoichiometric Sodium Transition-Metal Layer Structures Na x TMO 2 (x =~1)
Sodium stoichiometric layered compounds, Na x TMO 2 (x =~1) known as O3 type, typically deliver lower discharge capacities than P2/P3-type sodium cathodes. The diffusion of sodium ions in O3-type compounds occurs from one octahedral site to another through face-shared interstitial tetrahedral sites and is characterized by relatively slow diffusion compared with that in P2/P3 structures. However, sufficient Na content reserved in the alkali layer guarantees a high first charge capacity and high coulombic efficiency, which is an advantage for commercial applications. In this part of the article, we will discuss the O3-type cathodes with anionic redox, in which sodium ions are located only in the alkali layer.  (Figures 7(a) and 7(b)). Furthermore, their theoretical calculations were confirmed by the experimental work of Wang et al. [85]. Interestingly, the as-synthesized Na[Li 1/3 Mn 2/3 ]O 2 material was not sensitive to moisture, and even after soaking in distilled water, it maintained its structure and crystallinity [85]. The delivered first charge capacity was~250 mAh g −1 (0.9 Na + extraction), attributed to the oxygen O 2− /O 1− redox reaction accompanied by partial Li migration from the TM layer to the tetrahedral Na sites and O 2 release. The resulting discharge capacity was 190 mAh g −1 and was maintained during 40 cycles. In situ XRD revealed O3(I)-O3(II)-O3(III) structural transformations with a rapid decrease of the c lattice parameter during the O3(II)-O3(III) phase transition at the first charge. Because of the O 2 release and irreversible Li migration, the pristine state of the structure could not be fully recovered after discharge. In addition, partial Li + migration to the interlayer space (alkali-) sites resulted in the formation of vacan-cies in the TM layer, which caused the in-plane rearrangement of Mn and loss of the honeycomb ordering in the TM layer. In addition to the in situ XRD analysis, Wang [87]. The Li atoms preferred to stay in the center of the Ir honeycombs in the TM layer, whereas Na atoms occupied all the octahedral sites in the interlayer space. Cycling the material in a Na cell resulted in the extraction of 1.5 Na + /Li + through a complicated multiphase process and subsequent segregation of Lirich Li x IrO 2 and Na-rich Na x IrO 2 phases. Because of the absence of calculations and further experimental investigation of the reaction mechanism together with the Na/Li exchange during cycling, the authors were not able to arrive at a clear conclusion on the perspectives of this material. Therefore, substituting Li + with less mobile and smaller Mg 2+ or Zn 2+ could be a promising future direction in Irbased stoichiometric compounds with oxygen redox.

Cr-Based Compounds with Sulfur Redox.
Although cathode materials with oxygen-redox reaction have received more attention to date, some studies have also considered cathode materials with sulfur redox. For example, sulfur redox has been recently investigated in stoichiometric Na sulfides, such as O3-NaCrS 2 and O3-NaCr 2/3 Ti 1/3 S 2 [97,98]. Originally, the sulfur-redox chemistry in chalcogenides can be traced back to the work of Whittingham on TiS 2 in 1976 in Li cells [99], and the reaction mechanism is explained 11 Energy Material Advances

12
Energy Material Advances in detail by Rouxel in 1996 [100]. However, only recently, attempts have been made to utilize sulfur redox in Cr-based Na cathodes. In the work of Shadike et al., NaCrS 2 underwent Na + extraction/insertion through a solid-solution reaction with the occurrence of Cr/Na vacancy antisite (cation migration to the Na layer) with high-voltage hysteresis and poor reversibility [97]. A similar phenomenon of Cr migration to the Li sites was observed earlier in LiCrO 2 oxide [101]. In LiCrO 2 , the oxidation of Cr from 3+ to 4+ on charge resulted in disproportionation and further migration of Cr to tetrahedral sites, forming a rocksalt structure. The authors claimed that in NaCrS 2 , the migration of Cr occurred without its oxidation, which triggered sulfur redox. The prerequisite for sulfur redox was an unhybridized S 3p orbital that was provided by Cr migration to the Na layer and changed the configuration symmetry around S, resulting in the formation of the Na-S-□ configuration.
In an effort to increase the delivered capacity and decrease the voltage hysteresis, which is associated with TM disorder, the authors spearheaded attempts to stabilize Cr in the TM layer through doping of Ti 3+ into the NaCrS 2 structure [98]. The capacity of NaCr 2/3 Ti 1/3 S 2 was boosted to 186 mAh g −1 due to the synergetic effect of the anion S 2 − /S 1− and cation Ti 3+ /Ti 4+ redox reactions without Cr participation (Figure 7(c)). The in situ XRD results showed that the material underwent a sequence of O3-P3-O1' phase transitions with the shortening of the a / b lattice parameter and a large reduction of the c lattice parameter in the region of the P3-O1' transition (Figure 7(d)). Doping of Ti 3+ was unsuccessful in preventing Cr migration from the TM layer to Na vacancies in the P3-O1' region. However, the authors claimed that migration of Cr was a highly reversible process and that the main reason for the capacity loss was not migration of Cr but the loss of sulfur (Figure 7(e)). Using experimental and theoretical methods, including K-edge XAS, STEM, XPS, and DFT+U, it was possible to trace the different stages of the compensation mechanism such as the formation of electron holes, anionic dimers, and disulfide-like species as well as the precipitation of sulfur. According to those results, various anionic redox chemistries were proposed as follows: Electron holes of S were formed at the top of 3p band, which was accompanied with P3 phase evolution process (Equation (1)). Further, the formation of (S 2 ) ndimers was occurred, and it was triggered by noncoordinated S 3p states (Equation (2)). The weakening of Cr-S electrostatic repulsion caused Cr migration from TM layer to Na layer, resulting in P3-O1' phase transformation. The direct observation of Cr migration was obtained using STEM. The anion defects and formation of (S 2 ) 2-(Equation (3)), followed with irreversible oxidation or disproportionation of disulfides to sulfur (Equation (4)), occurred on the surface of the particles with the subsequent sulfur dissolution in electrolyte. The dissolution of sulfur was proved by measuring XPS spectra of the glass fiber separator before and after 10 and 50 cycles.
In conclusion, despite the TM cation migration and lower operation voltage in sulfides than in oxides, the high reversibility of the sulfur redox pushes further research works on chalcogenide materials for SIBs. For mitigation of cation migration, different doping types can be implemented to improve the cycling stability; furthermore, Cr-based chalcogenides can be expanded to different TM chalcogenides.
3.5. Other Compounds. Besides the above-discussed compounds with anion-redox participation, some more Nastoichiometric cathodes, such as α-NaFeO 2 and NaVO 3 , have been reported. α-NaFeO 2 (R 3 m) is a promising low-cost layered material with a typical O3 structure. However, it suffers from irreversible capacity with poor cycling performance. The underlying reaction mechanism and such irreversible behavior are still under debate. α -NaFeO 2 material was first reported in 1985 by Kikkawa et al., [102] followed by a research of Yabuuchi et al. [103], Zhao et al. [104], and Lee et al. [105] groups. α-NaFeO 2 delivered 80-100 mAh g −1 of reversible capacity with a flat voltage plateau at 3.3 V vs. Na (Figure 8(a)). The reversible capacity in the range of Na 1- x FeO 2 , x < 0:5, was ascribed to Fe 3+ /Fe 4+ redox reaction and was confirmed by 57 Fe Mössbauer spectrometry [104]. The electrode performance beyond x > 0:5 was deteriorated and exhibited irreversible structural behavior, which the authors suggest is due to Fe ion migration to neighboring tetrahedral sodium sites and subsequently blocking the diffusion sodium pathways [103]. Lee et al. first showed the nonequilibrium phase transformation during charge/discharge process from hexagonal (R 3 m) to monoclinic (C2/m) phase, accompanied with the evidence of the chemical instability of Fe 4+ species and electrolyte decomposition in the battery cell environment [105].
Recently, Li et al. [106] observed not only the irreversible Fe migration from TM layer to Na layer at the atomic scale by aberration-corrected scanning transmission electron microscopic (STEM), but also oxidation of oxygen during desodiation of α-NaFeO 2 using XAS spectra. Moreover, DFT calculations showed that near Fermi level, the Fe 3d and O 2p states are highly overlapping, which means that both Fe and O contribute to charge compensation. In contrast to the Li et al.'s work, Susanto et al. [107] revealed that only oxygen redox responsible for charge compensation from the beginning of charging of α-NaFeO 2 . When more than 0.5 Na was extracted, O 2 gas was released together with Fe migration and the formation of Fe 3 O 4 (cubic spinel Fd 3 m) locally on the surface of the particles (Figure 8(b)). Therefore, reversible charge compensation mechanism through solely oxygen redox was limited up to 0.5 of Na extraction. To address this critical issue, partial substitution of Fe with other TM has been reported previously. However, the participation of oxygen redox has not been shown in sufficient detail, which emphasizes the need for further systematic studies of such materials.

Energy Material Advances
In addition, oxygen-redox participation has also been investigated in stoichiometric layered NaVO 3 material [108,109].The material crystallizes into a layered structure with monoclinic C2/c space group, where layers of NaO 6 octahedra and VO 4 tetrahedra alternate (Figure 8(c)). On the basis of in situ synchrotron XRD results, the structure of the material showed insignificant changes during Na + (de-)intercalation with a and b lattice parameter variation only by 0.13 and 0.19%, respectively [109]. A more detailed study on electrochemical process revealed that NaVO 3 undergoes an oxygen-redox reaction during initial charge and cationic V 4+ /V 5+ and anionic O 2-/O 1redox during subsequent discharge (Figure 8(d)). Oxygen charge compensation mechanism was proved by DFT calculation showing the domination of O 2p states for partially and fully desodiated structure. The delivered reversible capacity was reached 245 mAh g -1 , leading to one of the most highest energy densities (566 Wh kg −1 ) for SIB materials with anion-redox participation.

Sodium-Rich Transition Metal Layer
Structures Na x TMO 2 (x > 1) Na-rich materials, Na 1 [Na x TM 1-x ]O 2 , which are analogs of Li-rich materials, are promising materials for highperformance SIBs owing to their high sodium content, offering an opportunity to increase the energy density of SIBs. In sodium-rich materials, sodium ions are located in both alkali and TM layers (octahedral sites), in contrast to full sodiumstoichiometry and deficient materials, in which sodium ions are located only in the alkali layer. The extra Na content may provide extra capacity, delivered by cumulative cationic and anionic redox, potentially exceeding 1 Na + per formula unit.

Mn-Based Compounds.
In contrast to their lithium-ion analogs, where lithium-rich Li 2 TMO 3 (Na[Na 1/3 TM 2/3 ]O 2 )type oxides (TM: Mn, Mo, Ru, and Ir) have been synthesized and studied for 3d TM (Mn) and 4d/5d TM (Mo, Ru, and Ir), sodium-rich oxides Na 2 TMO 3 have been successfully obtained only for 4d (Ru) and 5d TM (Ir). The reason for that is likely the larger ionic radius mismatch between Na + (1.02 Ǻ) and Mn 4+ (0.53 Ǻ). Therefore, to date, there is no a clear evidence of the crystal structure or electrochemical performance of Na 2 MnO 3 . The material has been discussed only in computer simulations. According to the theoretical work of Gao et al., it is possible to extract 1.75 Na + per formula unit through partial O 2− /O 1− redox reaction leaving much of the local structure intact [110].

Ru-Based
Compounds. The oxygen-redox chemistry was demonstrated in sodium-rich Ru-based cathodes with good structural stability [30][31][32][33][34][35][36][37][38][39][40][41][42]. The first research paper on such materials was published in 2013 by the group of Tamaru et al. on Na 2 RuO 3 (R 3 m), in which Ru was stabilized as Ru 4+ [30]. The material delivered a specific capacity of 140 mAh g −1 , which exceeded 7%, the theoretical capacity from only Ru 4+ /Ru 5+ redox. Sloppy charge/discharge curves were observed with an average potential of 2.8 V vs. Na/Na + . It was indicated that the reaction mechanism proceeded through a solid-solution reaction (Na 2-x RuO 3 0 < x < 0:5), followed by a two-phase reaction (0:5 < x < 0:6). Later in 2015, Rozier's group designed Sn-doped layered Na 2 Ru 1 −y Sn y O 3 materials [31]. The voltage profiles of the Sndoped materials were similar to those of Li-rich materials with two distinct voltage plateaus on charge (2.8 and 3.8 V) and an S-curved shape on discharge. Using XPS analysis, the authors proved that the lower voltage plateau was associated with cationic Ru 4+ /Ru 5+ redox and that the higher one was associated with anionic O 2− /O 2 n− redox. Similar to De Boisse et al.'s work, a solid-solution-two-phase-solid solution mechanism was observed during Na deintercalation.
In a more recent study, De Boisse et al. synthesized honeycomb ordered O-Na 2 RuO 3 (C2/m) and the disordered analog D-Na 2 RuO 3 (R 3 m) [32] and investigated the importance of structural order/disorder of alkali and TM ions in the TM layer (Figure 9(a)). The authors claimed that the order in the TM layer is a prerequisite for the activation of oxygen redox, induced by frontier orbital O(2p)-Ru(t g ) reorganization with the short O-O distances in distorted RuO 6 . Because of the inplane honeycomb ordering, the O-Na 2 RuO 3 electrode demonstrated enhanced capacity of 180 mAh g −1 compared with Capacity (mAh g -1 ) 200 250 14 Energy Material Advances that of 135 mAh g −1 for D-Na 2 RuO 3 (Figure 9(b)). The authors showed that 30% extra capacity was achieved by the spontaneously ordered intermediate ilmenite O1-Na 1 RuO 3 (R 3) phase, which accommodates the cooperative distortion of the RuO 6 octahedra. In contrast, in the case of disordered material, the intermediate P3-Na 1 RuO 3 phase exhibited strain frustration, which did not allow activation of the oxygen-redox reaction. Further study of ordered O-Na 2 RuO 3 showed O3-O1-O1' structural phase transformations during charge with an intermediate O1-Na 1 RuO 3 phase and O1'-Na 1/2 RuO 3 (P 3 1 m) phase stabilized in the fully charged state (Figures 9(c) and 9(d)) [33]. The in situ XRD and DFT calculation results showed that the existence of ordered Na vacancies played an essential role in increasing the O 2p electronic population near the Fermi level, which not only stabilized the phase transformations during cycling but also facilitated reversible oxygen-redox reactions (Figure 9(e)) [34]. Moreover, in a later work of Liu et al., the Mn 4+ -substitution strategy was adopted in Na 2 Ru 1-x Mn x O 3 (x = 0 -0:3) material [38]. Mn 4+ doping resulted in an increase of the voltage of the material due to the increase in the M-O band ionicity and charge on O. In addition, Mn 4+ doping suppressed the O3-P3 phase transition and prevented the formation of the spinel phase in the highly desodiated state and enhanced the robustness against water attack. A surprising super long cycling stability with a capacity retention of 70% was achieved at 5C after 1000 cycles. Na 3 RuO 4 (Na[Na 1/2 Ru 1/2 ]O 2 ) (C2/m) is a further expansion of the Na 2 RuO 3 cathode material toward higher sodium content (O/TM ratio) Ru-based materials with oxygen-redox activity [40][41][42]. The crystal structure of Na 3 RuO 4 is described as a layered structure with a Ru 5+ framework forming isolated tetramers of the edge-sharing RuO 6 octahedra in the Na 1/2 Ru 1/2 O 2 layer (Figure 10(a)) [41]. The reaction mechanism in Na 3 RuO 4 is currently under debate by several groups [40][41][42]. The first work of Qiao et al. on the chemical extraction of Na from Na 3 RuO 4 indicated that capacity is delivered by merely O redox reaction through the formation of peroxo-based O-O (de)bonding, which was confirmed by in situ Raman analysis (Figure 10(b)) [40]. On the basis of the XPS and XANES analyses, the authors claimed that Na extraction/insertion proceeded along the inert redox character of Ru 5+ in the octahedral position. However, Otoyama et al. later revisited this compound and showed that the charge compensation mechanism includes the participation of both Ru and O redox reactions [41]. First, the oxidation of Ru 5+ to Ru 6+ occurs, leading to the formation of Na 2 RuO 4 (P 2 1 /c), followed by oxygen reduction at the end of charge through a solid-solution process with the formation of amorphous Na x RuO 4 (x = 0 -1). The presented O-redox reaction proceeded together with the dissolution of the active material, which subsequently led to the irreversible reaction and poor recovery of the Na 3 RuO 4 structure with its initial crystallinity [41]. In the work of Hu et al. [42], similar results were shown, clarifying that both Ru 5+ /Ru 6+ and O 2− /O 1− are active, and it was shown that the oxygen-redox activity decreased with a retention of 36% after 30 cycles, which was the main reason for the large capacity fading and limited reversibility.
To conclude, Na 3 RuO 4 showed electrostatically more unstable behavior upon deep desodiation than Na 2 RuO 3 , which could be explained by the higher Na vacancy content

Ir-Based
Compounds. The O3-type Na 2 IrO 3 (Na[Na 1/3 Ir 2/3 ]O 2 ) (C2/m) compound is another example of a Na-rich composition that displays oxygen-redox activity [43]. In contrast to 3d (Mn) and 4d (Ru) TM compounds, which present drastic evolution of the voltage profile through the first cycle, the 5d (Ir) Na 2 IrO 3 material did not show an evolution from a two-plateau to S-shape voltage profile (Figure 11(a)). The enhanced structural rigidity of the compound allowed cycling of the cell reversibly upon high sodium extraction/insertion (1.5 Na + per formula unit, capacity~130 mAh g −1 ) with neither oxygen release or cationic migrations to interlayer (alkali-) sites. During extraction of sodium ions, the O3-type layered structure first converted to O1' (P 1) and further to O1 (C 2m) through gliding of the oxygen close-packed planes (Figure 11(b)). The increased cationic repulsion with charge density loss on the O atom leads to a high covalent TM-O bond, which reduces the possibility of O 2 release and effectively stabilizes the structure. Using DFT calculations, the authors indicated that both the Ir and O bands lay just below the Fermi level, which involve both oxygen O 2− /O 1− and Ir 4+ /Ir 5+ contributions in charge compensation (Figures 11(c) and 11(d)). Interestingly, that oxygen-redox activity was observed in the early stage of the desodiation process, which the author linked to the slightly distorted IrO 6 octahedra in the pristine Na 2 IrO 3 material (Figures 11(c) and 11(d)).
To  [45]. β-Na 1.7 IrO 3 was obtained by the electrochemical exchange of Li into Na in β-Li 2 IrO 3 material. In the structure of β-Na 1.7 IrO 3 , the Na and Ir layers were shifted relative to each other; instead of a layered structure, the material acquires a Fddd hyperhoneycomb structure with a rigid 3D network (Figures 11(e) and 11(f)). It turned out that such a hyperhoneycomb structure was much more stable, with reversible uptake of nearly 1.3 Na + and good capacity retention of 82% after 100 cycles in a full-cell configuration against hard carbon. Despite the fact that the β-Na 1.7 IrO 3 material underwent multiple structural phase transitions with different lattice parameters and a large volume change~26% during sodiation/desodiation, it maintained the same IrO3 framework (Figure 11(g)). The charge-compensation mechanism revealed that both the anionic (O 2 ) n− redox and cationic activity of Ir 5+ /Ir 4+ were responsible for the delivered capacity.
In summary, 5d Ir-based compounds exhibit more reversible charge/discharge curves than 4d TM and especially 3d TM compounds, suggesting higher structural integrity due to more diffused orbitals against the irreversible reactions of oxidized oxygen. However, because Ir is one of the rarest metals in the Earth's crust and because of its high cost, the use of such cathodes is of little practical importance.

Biphasic Layered Structures
To achieve better electrode performance, an attractive approach of the combination of P/O phases in one material was recently proposed. The synergetic effect of the   [112]. Combining solid-state NMR, EPR, XPS, and HRTEM analyses, the authors demonstrated that Ti 4+ substitution could not only effectively suppress lithium migration from the TM layer to the Na layer but also enhanced the structural stability by alleviating the formation of irreversible surface cracks on particles during cycling.
In conclusion, successful utilization of O/P composites as a cathode material with anion redox to increase the energy density and improve the first coulombic efficiency appears to be a promising strategy, further expanding the variety of potential anion redox materials for SIBs.

Cation-Disordered Rocksalt Oxides
Over the last five years, Li-rich cation-disordered rocksalttype structure oxides with oxygen redox compensation mechanism have emerged as potential high energy density cathodes for LIBs with high reversible capacities 300 mAh g -1 and extremely high energy densities 1000 W h kg -1 [113]. Oxygen-redox activity in such systems associated with a highly ionic character of d 0 charge compensator metal ions, such as Ti 4+ /Zr 4+ /Nb 5+ /Mo 6+ .
The concept of Li-rich rocksalt structures was recently applied to Na-rich materials [114,115]. Sato

17
Energy Material Advances process. The reversible capacity was~200 mAh g -1 for both materials, despite the different reaction mechanism.
For Na 1.3 Nb 0.3 Mn 0.4 O 2 , the detailed investigation of electrochemical process revealed that Mn 3+ /Mn 4+ redox prevailed at low voltage, while O oxidation accompanied with O 2 loss dominated at high voltage on first charge. As a result, accumulation of carbonate species and O 2 loss were detected at fully charged state in Na 1.3 Nb 0.3 Mn 0.4 O 2 electrode. Therefore, only reversible two-electron Mn 2+ /Mn 4+ redox rather than reversible oxygen redox were observed on the subsequent cycles. By substituting Nb 5+ with Ti 4+ , the reversibility of oxygen redox was significantly improved in Na 1.14 Mn 0.57-Ti 0.29 O 2 . Reversible cationic Mn 3+ /Mn 4+ without reduction to Mn 2+ and anionic O 2− /O n− redox reactions were responsible for delivered capacity in Na 1.14 Mn 0.57 Ti 0.29 O 2 . Moreover, not only the reversibility of anionic redox has been highly improved compared with the former Na 1.3 Nb 0.3 Mn 0.4 O 2 material, but also cycling stability of Na 1.14 Mn 0.57 Ti 0.29 O 2 was greatly increased. Therefore, further research in Narich cation-disordered rocksalt cathodes from abundant elements, such as manganese and titanium, with reversible anionic redox is required for the development of high energy density rechargeable SIBs.

Conclusions
The use of cathode materials with both anion-and cationredox reactions represents a promising approach toward important gains in the energy density of SIBs. The estimation of the gravimetric energy density for P2-type materials with cationic and anionic redox is~500-600 Wh kg −1 and that for O3-type materials is~400 Wh kg −1 , which is very promising ( Figure 12). To date, considerable progress has been achieved for Na-rich (O3-type) and Na-deficient (P2-type) materials. Most of the emphasis on Na-rich materials has been on rare and expensive 4d TM (Ru) and 5d TM (Ir) elements; however, Na-deficient materials rely on the use of low-cost 3d TM (Mn). Therefore, from a practical viewpoint, P2-type materials have considerable merits such as low cost, high performance, and relatively stable oxygen-redox reactions, making them promising for SIBs with improved energy density. However, recent work of Wang et al. [85] showed that a new Na-stoichiometric compound O3-Na[Li 1/3 Mn 2/3 ]O 2 not only offers an effective possibility of utilizing anionic redox but also does not have the drawback of low coulombic efficiency. Moreover, from this viewpoint, mixed P2/O3-type oxides represent an interesting strategy, leading to high-performance materials with enhanced air stability and relatively high sodium content.
From a practical viewpoint, the main drawbacks of materials with anion redox are their severe microstructural and electrochemical instabilities. Therefore, it is critical to understand the origin of these issues and to find ways to mitigate them. Among the microstructural issues, the most significant involve oxygen evolution, cation migration, and surface reconstruction. The principal electrochemical problems in anion redox reactions in SIBs are the sluggish reaction kinetics, high-voltage hysteresis (compared to cationic redox), and voltage fade during cycling, which in turn adversely affect the electrochemical performance. The slow reaction kinetics is mostly caused by drastic rearrangements of bonding configurations; however, high-voltage hysteresis and voltage fade have largely been attributed to the structural cation disordering in plane and out of plane in the material during cycling. Therefore, we would like to provide some important solutions for optimization of the properties of cathode materials with cationic-anionic redox for SIBS.
(1) One of the critical concerns for anion-redox cathodes is the nature of the oxidized species, which needs to be better understood and characterized, especially because it defines the chemical reactivity and solubility of the material. Therefore, the use of a combination of experimental and theoretical techniques, such as O-K mRIXS, in situ XRD/ND, XAS, and Raman methods together with DFT calculations, is critical because these observations and predictions provide a critical benchmark for the detection and determination of the nature of the oxidized species in cathode materials (2) The concept of an ordered compositional approach has been successfully realized in P2-Na 0.6 [Li 0.2 Mn 0.8 ]O 2 , O3-Na 2 RuO 3 , and Na 2 Mn 3 O 7 materials and has resulted in beneficial properties with good structural stability, high reversibility of anion redox, and lowvoltage hysteresis. However, a nonhysteresis profile has been demonstrated only for Na 2 Mn 3 O 7. This unique behavior makes it a key compound for understanding the chemical, structural, and electronic properties for engineering materials with truly reversible nonhysteretic anion redox. Therefore, further theoretical predictions and additional approaches should be expanded to determine the reasons for the nonhysteretic behavior of Na 2 Mn 3 O 7 and to identify more compounds with high electrochemically reversible and kinetically easy anion redox processes 200 Na-full stoichiometry and Na-rich oxides Na-deficient oxides