Molecular Orbital Principles of Oxygen-Redox Battery Electrodes

Oct 10, 2017 - Correlations between Transition-Metal Chemistry, Local Structure, and Global Structure in Li2Ru0.5Mn0.5O3 Investigated in a Wide Voltag...
1 downloads 20 Views 4MB Size
Download PDF
Review www.acsami.org

Molecular Orbital Principles of Oxygen-Redox Battery Electrodes Masashi Okubo and Atsuo Yamada* Department of Chemical System Engineering, School of Engineering, The University of Tokyo, Hongo 7-3-1, Bunkyo-ku, Tokyo 113-8656, Japan ABSTRACT: Lithium-ion batteries are key energy-storage devices for a sustainable society. The most widely used positive electrode materials are LiMO2 (M: transition metal), in which a redox reaction of M occurs in association with Li+ (de)intercalation. Recent developments of Li-excess transition-metal oxides, which deliver a large capacity of more than 200 mAh/g using an extra redox reaction of oxygen, introduce new possibilities for designing higher energy density lithium-ion batteries. For better engineering using this fascinating new chemistry, it is necessary to achieve a full understanding of the reaction mechanism by gaining knowledge on the chemical state of oxygen. In this review, a summary of the recent advances in oxygen-redox battery electrodes is provided, followed by a systematic demonstration of the overall electronic structures based on molecular orbitals with a focus on the local coordination environment around oxygen. We show that a π-type molecular orbital plays an important role in stabilizing the oxidized oxygen that emerges upon the charging process. Molecular orbital principles are convenient for an atomic-level understanding of how reversible oxygen-redox reactions occur in bulk, providing a solid foundation toward improved oxygen-redox positive electrode materials for high energy-density batteries. KEYWORDS: battery, cathode, transition-metal oxides, oxygen-redox reaction, molecular-orbital method, orphaned oxygen orbital

1. INTRODUCTION Electrochemical energy storage in the form of batteries is crucial to foster a sustainable and clean-energy society. For example, electrified transportation such as electric vehicles to reduce reliance on fossil fuels requires the development of batteries with a gravimetric energy density of approximately 350−500 Wh/kg, enabling a vehicle range of 500 km per charge. However, the currently used lithium-ion batteries have a gravimetric energy density of ca. 200−300 Wh/kg.1,2 To realize lithium-ion batteries with higher energy density, it is necessary to (i) increase the specific capacity of negative electrode materials, (ii) increase the reaction voltage of positive electrode materials, and (iii) increase the specific capacity of positive electrode materials. Based on the first strategy, alloying reactions (Si, Sn, P, etc.), conversion reactions (transition-metal oxides), and Li metal plating/stripping reactions have been studied intensively.3−5 The second strategy has been adopted primarily for polyanionic compounds in which their reaction voltage can be increased using the inductive effect of anions with large electronegativity such as SO42− or F−.6,7 In this review, we focus on the third strategy specifically applied to lithium/sodium-excess transition-metal oxides, namely oxygenredox electrodes, which have recently become a central target for battery scientists with a large number of key publications.8 The conventional and most widely used positive electrode materials are α-NaFeO2-type layered transition-metal oxides with a general chemical formula of LiMO2 (M: transition metal); these transition-metal oxides undergo an electrochemical intercalation reaction: LiM 3+O2 ↔ xLi+ + xe− + Li(1−x)M 3+(1−x)M 4+xO2.9 However, despite having a large theoretical one-electron capacity (ca. 270−290 mAh/g for x = © XXXX American Chemical Society

1), the available reversible capacity of LiMO2 is typically less than 200 mAh/g (x ∼ 0.7, Figure 1) because of irreversible structural damages induced in the Li-poor regions, thus limiting the practical capacity by the amount of reversible Li

Figure 1. Performance of selected oxygen-redox positive electrode materials Li1.2Ni0.13Co0.13Mn0.54O2, Li1.3Nb0.3Mn0.4O2, and β-Li2IrO3. For comparison, performance of conventional positive electrode materials based on transition-metal redox reactions (LiCoO2, LiNi1/3Co1/3Mn1/3O2, LiMn2O4, and LiFePO4) is also shown. Received: July 7, 2017 Accepted: October 3, 2017

A

DOI: 10.1021/acsami.7b09835 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

Review

ACS Applied Materials & Interfaces

orbitals) focusing on M.32 However, the state of the oxygen orbital is of primary interest for oxygen-redox battery electrodes. Before considering the oxygen-redox reaction, it is necessary to redefine the molecular orbitals in the band structure because the local coordination environment around oxygen in Li-excess transition-metal oxides has a point-group symmetry that is different from that in LiMO2. In this review, after a summary of recent advances in oxygenredox electrode materials, a systematic derivation of oxygenredox chemistry from a molecular-orbital viewpoint is presented to understand the electronic states of oxygen during the oxygen-redox reactions, clarifying how the oxidized oxygen is stabilized to deliver a reversible oxygen-redox capacity.

intercalation rather than by the whole one-electron M redox capacity.10,11 One effective strategy to overcome this limitation is to use Li-excess transition-metal oxides, where partial replacement of M with Li in LiMO2 to form Li[LiyM1−y]O2 can realize the electrochemical reaction Li[LiyM1−y]O2 ↔ (x + y)Li+ + (x + y)e− + Li(1−x)[Li0M1−y]O2 with a capacity greater than that of LiMO2. This Li-excess strategy has been exploited using various transition-metal oxides such as Li1.11Ni0.33Mn0.56O2,12 Li1.2Ni0.18Co0.1Mn0.52O2,13 Li1.3Nb0.3Mn0.4O2,14 and β-Li2IrO3,15 which indeed deliver large reversible capacities exceeding 200 mAh/g (Figure 1). However, a simple charge compensation scheme cannot explain the large capacities because the initial valence state of M in Li-excess transition-metal oxides is higher than that of LiMO2 as a result of aliovalent substitution (M 3+ → Li+1/3M 4+ 4+ is the highest valence state of the 2/3). Assuming that M fourth period M (3d transition metal) in an octahedral oxygen environment, the redox capacity of M in Li[LiyM1−y]O2 should decrease with increasing y. This prediction clearly contradicts the large capacities previously reported for many Li-excess transition-metal oxides, making the simple M redox scheme seem questionable. The issue of the extra capacities in Li-excess transition-metal oxides has been under debate since 2001 when the research group led by Dahn showed a large capacity for solid solution Li2MnO3−LiMO2.12,16 Several groups postulated possible reaction mechanisms focusing on the electrode surface such as Li2 O extraction, 17 O 2 gas evolution, 18 Li + /H + exchange,19 carbonates formation/decomposition,20 and/or CO2 evolution.21 The possible surface reactions and resulting structural changes in Li-excess transition-metal oxides were previously reviewed in detail.22,23 However, after the research group led by Croguennec experimentally suggested in 2013 the participation of an oxygen-redox reaction as a charge compensation mechanism,24,25 much effort has been devoted to understanding how lattice oxygen undergoes a reversible redox reaction. Although the possible oxygen contribution to the capacity of oxide electrode materials has been discussed for decades,26,27 oxygen-redox chemistry in Li-excess transitionmetal oxides is yet to be fully understood due to the lack of the full set of experimental evidence and theoretical insights into oxygen states. An important current interest in oxygen-redox battery electrodes is the change in the nature of chemical bonds between M and O during oxygen-redox reactions. For example, some groups reviewed the oxygen redox reactions of various battery electrodes using a rigid band model, which intuitively considers only a bandwidth and a band center energy level of each atomic orbital.23,28,29 However, the rigid band model cannot estimate the interactions between M and O. Chemists know that a molecular orbital method is an efficient approach to describe the interaction between the atoms and the phenomena of closed-shell molecules.30 Although this Hartree−Fock level method is not precise for open-shell (magnetic) molecules due to configuration interactions such as ligand-to-metal charge transfer, a qualitative interpretation is provided based on topology, orbital overlap, and orbital energy level. For example, the band structure of LiMO2 is usually labeled based on the Oh point-group molecular-orbital notation of a MO6 octahedron (a1g, t1u, eg, and t2g),31 which greatly aids the understanding of the redox reaction of M. Indeed, the research group led by Ceder attempted to rationalize the oxygen-redox reactions of various Li-excess transition-metal oxides based on a conventional band picture (Oh molecular

2. OVERVIEW OF OXYGEN-REDOX ELECTRODE MATERIALS It has been reported that a wide range of Li-excess transitionmetal oxides can deliver an overcapacity exceeding the limit of the M redox reaction. We provide a brief overview of these high-capacity electrode materials and clarify the points that the research community shares and those that they do not share. 2.1. Li2MnO3. Li2MnO3 crystallizes into a layered structure, where Li and [Li1/3Mn2/3] layers stack alternatively, giving a formula of Li[Li1/3Mn2/3]O2 in conventional layered LiMO2 notation.33,34 All cations occupy octahedral sites, and the [Li1/3Mn2/3] layers have a honeycomb-type cation arrangement, where oxygen is coordinated by two Mn and four Li to form an OMn2Li4 octahedron (Figure 2a).

Figure 2. Crystal structures of selected oxygen-redox electrode materials. (a) Li-excess layered transition-metal oxide Li2MnO3, (b) Li-excess random rock salt Li1.3Nb0.3Mn0.4O2, (c) Li-excess tridimensional transition-metal oxide β-Li2IrO3, and (d) Na-excess layered transition-metal oxide Na2RuO3. The coordination environment of oxygen is also shown for each compound. All compounds have oxygen coordinated by two transition metals and four alkali-metal ions with a Li−O−Li (or Na−O−Na) coordination geometry.

From an electronic-structure viewpoint, Li-ion deintercalation from Li2MnO3 is difficult because the frontier energy level of the occupied t2g orbitals of Mn4+ is too low to undergo oxidation. However, in 1999, Kalyani et al. reported that Li2MnO3 undergoes Li-ion deintercalation upon charging.35 The research group led by Bruce re-examined this phenomenon and confirmed 1.4 Li-ion deintercalation per the formula unit without the oxidation of Mn4+.19,36 The group proposed two charge-compensation mechanisms: (1) oxygen loss; Li2MnO3 → (x/2)Li2O + xe− + Li(2−x)Mn4+O(3−x/2), and (2) solvent oxidation/proton exchange; Li2MnO3 + solvent → xLi+ + xe− + HxLi(2−x)Mn4+O3 + decomposed solvent. Although oxidation of B

DOI: 10.1021/acsami.7b09835 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

Review

ACS Applied Materials & Interfaces O2− was presumed to trigger the above reactions on the electrode surface, a reversible oxygen-redox reaction in bulk was not postulated in the early 2000s. 2.2. Li2MnO3−LiMO2 (M = Cr, Mn, Co, Ni). Almost concurrently with the studies on Li2MnO3, Yamanaka et al. reported the synthesis of solid solution Li2MnO3−LiMO2 (M = Co) in 1997.37,38 A series of solid solution compounds, Li[Lix/3Mn2x/3Co1−x]O2, was successfully synthesized for 0 ≤ x ≤ 1 because both Li2MnO3 and LiCoO2 are isostructural with α-NaFeO2. The structure of the Li2MnO3−LiMO2 solidsolution compounds is essentially the same as that of Li2MnO3, while the Li and Mn sites in the honeycomb [Li1/3Mn2/3] layers are partially replaced with M to form [LiMnM] layers. Oxygen is coordinated either by two transition metals and four Li to form OM2Li4 octahedra or three transition metals and three Li to form OM3Li3 octahedra. In 2001, Dahn et al. found a large capacity of solid solution Li2MnO3−LiMO2 (M = Ni0.5Mn0.5),12,16,17 and this finding immediately stimulated extensive studies on various other systems (e.g., M = Ni1/3Co1/3Mn1/3).39,40 The unique electrochemical properties of Li2MnO3−LiMO2 are (i) a large voltage plateau around 4.5 V vs Li/Li+ upon the first charge, (ii) sloping discharge/charge curves without a voltage plateau after the first charge, (iii) a large irreversible capacity at the first cycle, and (iv) a large reversible capacity after the first charge (Figure 3a, 0.4Li 2 MnO 3 −0.4Li(Ni 1/3 Co 1/3 Mn 1/3 )O 2 or Li1.2Ni0.13Co0.13Mn0.54O2).41 The potential plateau observed at the first charge is not observed in subsequent cycles, suggesting that a structural change occurs only during the first charge process. By the early 2010s, many groups were focused on surface reactions such as oxygen loss or Li+ /H + exchange.17−21,42 The participation of bulk oxygen in providing the extra capacity of Li2MnO3−LiMO2 was reported by the research group led by Croguennec in 2013.24,25 Bruce et al. then quantitatively examined the reversible oxygen-redox reaction using operando mass spectroscopy for 18 O-labeled Li1.2Ni0.13Co0.13Mn0.54O2 and Li1.2Ni0.2Mn0.6O2.43,44 The group concluded that, although oxygen loss accounts for the initial irreversible capacity, an oxygen-redox reaction is the key reaction to achieve the extra capacity. 2.3. Li2MO3 (M = Ru, Ir). Almost simultaneously with the first postulation of the bulk oxygen-redox chemistry in Li2MnO3−LiMO2, the research group led by Tarascon proposed reversible oxygen redox reactions in a series of Li2MO3 (M = Ru, Ir).45−49 The group found that solid solution Li2RuO3−Li2MO3 (M = Mn4+, Sn4+) shows a large potential plateau at 4.3 V vs Li/Li+ upon the first charge with an extra capacity exceeding that of the Ru5+/Ru4+ redox reaction.45,46 The existence of an initial irreversible capacity in the first cycle and the disappearance of the charging potential plateau in the subsequent cycles are features similar to those found in the Li2MnO3−LiMO2 system. Systematic studies on the reversible oxygen-redox reactions in Li2RuO3−Li2MO3 (M = Mn4+, Sn4+) and Li2IrO3 using X-ray photoelectron spectroscopy (XPS),45,46 transmission electron microscopy (TEM),47 electron paramagnetic resonance (EPR),48 and ab initio calculations49,50 indicated that a peroxo-like O−O bond is formed to facilitate the oxygen-redox reactions and to stabilize oxidized oxygen. Indeed, a short O−O distance of ca. 2.5 Å was observed in Li0.5IrO3 after the oxygen oxidation. 2.4. Random Rock Salts. In 2014, the research group led by Ceder found that transition-metal oxides with a random

Figure 3. First through third charge−discharge curves for (a) Li-excess layered transition-metal oxide Li1.2Ni0.13Co0.13Mn0.54O2, (b) Li-excess random rock salt Li1.2Ti0.4Mn0.4O2 (50 °C),52 (c) Li-excess tridimensional transition-metal oxide β-Li2IrO3, and (d) Na-excess layered transition-metal oxide Na2RuO3. The first charge−discharge curves are plotted in red for clarity. (c) Reprinted with permission from ref 15. Copyright 2017 Springer Nature.

rock-salt structure can deliver a large capacity when excess Li opens a Li-ion percolating network.51 This percolation strategy was fused with oxygen-redox chemistry by Komaba et al.:14 a Li-excess random rock salt, Li1.3Nb0.3Mn0.4O2, delivers a large capacity due to an oxygen-redox reaction as well as percolative Li-ion migration. Furthermore, Yabuuchi et al. found another random rock salt Li1.2Ti0.4Mn0.4O2 with a large capacity of 300 mAh/g.52 Prominently, Li1.2Ti0.4Mn0.4O2 exhibits a stable potential plateau upon cycling, suggesting that it possesses a structural integrity against the oxygen-redox reaction (Figure 3b). Other random rock salts such as LiNi0.5Ti0.5O2− Li1.6Mo0.4O2 and Na1.3Nb0.3Mn0.4O2 were reported as oxygenredox electrode materials.53,54 From a structural perspective, cations in a random rock salt occupy octahedral sites, and the coordination environment of oxygen varies from OM6 to OLi6 due to the disordered C

DOI: 10.1021/acsami.7b09835 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

Review

ACS Applied Materials & Interfaces arrangement of the cations (Figure 2b). The probability that the oxygen in Li1+xM1−xO2 has a Li−O−Li axis exceeds 73% for x = 0.2; thus, Li-excess random rock salts contain a large amount of oxygen with a Li−O−Li axis. 2.5. β-Li2MO3 (M = Ir). β-Li2IrO3 is the first example of a well-defined three-dimensional framework that provides extra capacity from an oxygen-redox reaction.15 In β-Li2IrO3, Li and Ir occupy octahedral sites in an ABCABC close-packed oxideion array, and Li2Ir layers stack in a zigzag alignment to form a three-dimensional framework (Figure 2c).55,56 Oxygen is coordinated by two Ir and four Li to form an OIr2Li4 octahedron. Charge−discharge curves show reversible (de)intercalation of 2.0 Li+ per the formula unit with 3 voltage plateaus at 3.4, 4.4, and 4.6 V vs Li/Li+ (Figure 3c), where a reversible capacity for β-Li2IrO3 exceeds the limit of the Ir5+/ Ir4+ redox reaction. Tarascon et al. proposed that an oxygenredox reaction occurs simultaneously with the Ir5+/Ir4+ redox reaction, even at the low voltage plateau (Li2IrO3 ↔ Li1IrO3). 2.6. Na2MO3 (M = Ru, Ir). Oxygen-redox chemistry was also confirmed in positive electrode materials for sodium-ion batteries with a pronounced stability of the layered structure. Na2MO3 or Na[Na1/3M2/3]O2 (M: Ru, Ir) has a layered structure, where the Na and [Na1/3M2/3] layers stack alternatively (Figure 2d).57,58 Na and M occupy octahedral sites, while oxygen is coordinated by two M and four Na to from OM2Na4 octahedra. Tarascon et al. reported the electrochemical properties of Na2RuO3 in which charge− discharge curves show two potential plateaus at 2.5 and 3.7 V vs Na/Na+ with a capacity of 180 mAh/g (Figure 3d), exceeding the limit of the Ru5+/Ru4+ redox capacity.59 Yamada et al. clarified that, using two polymorphs of Na2RuO3 with and without a honeycomb-type arrangement of Ru and Na, an oxygen-redox reaction is triggered by cooperative distortion of the honeycomb lattice.60 The potential plateau at the high voltage was observed in many cycles, suggesting that Na2RuO3 possesses structural integrity against the oxygen-redox reaction. A similar oxygen-redox reaction was also reported for Na2IrO3.61

Figure 4. Molecular orbital energy diagrams of (a) an OA3M3 octahedron in layered transition-metal oxides AMO2 and (b) an OA4M2 octahedron in A-excess layered transition-metal oxides A2MO3. The electronic structure of M is assumed as t2g6eg0 for simplicity. The number of each molecular orbital is per oxygen. Symmetry labels for AMO2 correspond to the conventional Oh point group of MO6 coordination, while those for A2MO3 correspond to the C2v point group of OM2A4 coordination.

3. MOLECULAR ORBITALS IN TRANSITION-METAL OXIDES As described in Figure 2, oxygen-redox electrode materials commonly have oxygen coordinated by two transition metals and four alkali-metal ions. The importance of this specific coordination environment was described by the research groups led by Bruce and Ceder in 2016.32,43,44 In this section, we briefly summarize orbital hybridization in conventional transition-metal oxides and Li-excess transition-metal oxides. On the basis of the point symmetry of the oxygen coordination environment, we relabel the band structure of Li-excess transition-metal oxides. 3.1. AMO2. Consider oxygen coordinated by three M of the nd group (n = 3, 4, and 5) and three alkali-metal ions A such as in AMO2 (Figure 4a). The atomic orbitals of A can be neglected for molecular-orbital formation because their frontier 2s/3s energy level is much higher than that of O 2p. The nd, (n +1)s, and (n+1)p of M and 2p of O have an energy-level sequence of E(2p) < E(nd) < E((n+1)s) < E((n+1)p), which enhances O 2p−M nd hybridization. However, because the radial distributions of (n+1)s and (n+1)p are much more diffusive than that of nd, the overlap integrals between O 2p and M (n+1)s/(n+1)p are larger than that between O 2p and M nd, and the most bonding a1g/t1u and antibonding a1g*/t1u*

molecular orbitals are formed from O 2p−M (n+1)s/(n+1)p (Figure 4a). For the molecular orbitals from M nd, the nd(eg) orbitals are directed toward oxygen to form strong σ bonds (eg and eg*).31 Because the number of M orbitals exceeds the number of O 2p orbitals in AMO2, all O 2p orbitals form σ-type molecular orbitals with M nd(eg)/(n+1)s/(n+1)p, and M nd(t2g) remains a nonbonding orbital. Due to the low energy level of O 2p relative to that of most M orbitals, the bonding molecular orbitals have a predominant O 2p character, while the antibonding molecular orbitals are governed by M orbitals (Figure 5a). As the Fermi level of most AMO2 lies at antibonding eg* or nonbonding t2g with a predominant M nd orbital character, the capacity essentially arises from the M redox reaction (Figure 6a), even though a generated hole on M delocalizes to O 2p to a certain extent through orbital hybridization.62,63 All O 2p electrons have a strong bonding character forming a band at a much lower energy (Figure 4a and Figure 5a), thereby hindering any extra capacity from the additional oxygen oxidation 3.2. A2MO3. When oxygen is coordinated by two M and four A such as in A2MO3 (Figure 4b), the point symmetry of the OM2A4 octahedron is C2v. On the basis of this point symmetry, D

DOI: 10.1021/acsami.7b09835 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

Review

ACS Applied Materials & Interfaces

while two O 2p orbitals along an A−O−A axis and two M nd(t2g) form π-type less-bonding (b1) and less-antibonding (b1*) molecular orbitals (Figures 4b, 5b, and 5c).64 For high-valent M (e.g., Mn4+, Co4+, Ni4+, Ru5+, or Ir5+) that are included in typical oxygen-redox electrodes, the energy level of the occupied M nd(t2g) is lower than that of O 2p (Figure 5b). Therefore, the occupied b1* molecular orbital primarily consists of O 2p, which is responsible for an additional oxygenredox reaction (Figure 6b). Importantly, before oxygen oxidation occurs, this π-type interaction between the occupied O 2p and the occupied M nd(t2g) is weak (almost nonbonding), and the O 2p orbital near the Fermi level appears to be “orphaned”.32 However, after oxidation of this antibonding b1* molecular orbital occurs, the corresponding M−O bond should become more bonding with larger b1/b1* splitting, thus stabilizing the oxidized oxygen (Figure 6b). On the other hand, M of the d0 configuration (e.g., Ti4+, or Nb5+) has no occupied b1* molecular orbitals for the oxygen-redox reaction; therefore, a poor oxygen-redox activity is predicted (Figure 6c). Comparing M of the 3d, 4d, and 5d groups, M of the 5d group has the largest b1/b1* splitting because of having the most diffused orbitals, which significantly stabilizes the oxidized oxygen. For example, Na2RuO3 and β-Li2IrO3 exhibit reversible charge/discharge curves for many cycles, suggesting structural integrity against the oxygen-redox reaction.15,60,65 Note that, as shown in Figure 6d, the b1* molecular orbitals with M of the 4d/5d groups have substantial M character because of the large orbital overlap, and the corresponding additional oxygen-redox reaction should be accompanied by a certain degree of the M redox reaction.

Figure 5. Schematic band structures for (a) layered transition-metal oxides AMO2 and (b) A-excess layered transition-metal oxides A2MO3. The electronic structure of M is assumed as t2g6eg0 for simplicity. The red color represents the O 2p orbital contribution to each state. Symmetry labels for AMO2 correspond to the conventional Oh point group of MO6 coordination, while those for A2MO3 correspond to the C2v point group of OM2A4 coordination. (c) Selected molecular orbitals for the OM2A4 cluster. The O 2p orbital along an A−O−A axis forms π-type b1 and b1* molecular orbitals with M nd(t2g) orbitals.

4. THEORETICAL APPROACHES As most oxygen-redox electrode materials undergo irreversible reactions (e.g., oxygen loss, cation densification, and electrolyte decomposition) near the surface,17−21 experimental techniques suffer from these surface contaminations, rendering the true mechanism of the oxygen-redox reaction controversial. Therefore, theoretical calculations are important to understand how the oxygen-redox reaction occurs. Theoretical calculations of the oxygen-redox electrode materials were systematically conducted by the research group led by Ceder in 2016.32 It is well-known that density functional theory (DFT) calculations based on local density approximation (LDA) or generalized gradient approximation (GGA) cannot accurately describe electronic structures of transition-metal oxides due to self-interaction errors, especially with localized electrons.66,67 Therefore, they calibrated transition-metal and oxygen orbital levels by optimizing a screening parameter in Heyd−Scuseria−Ernzerhof (HSE) hybrid functional.68,69 Using the hybrid functional, they found that Li-excess transition-metal oxides have an orphaned O 2p orbital along the Li− O−Li axis near the Fermi level and that its labile O 2p electrons (Figure 6b) participate in an oxygen-redox reaction. After oxygen oxidation occurs, a hole orbital diffuses along the Li−O−Li axis (Figure 7a). The group also reported that less directional d0 or d10 metals facilitate the rotation of the orphaned O 2p orbitals for peroxolike σ-type O−O bond formation (Figure 7b), as experimentally observed in Li2Ru0.5Sn0.5O3.45 The energy level of the M orbitals relative to that of the orphaned O 2p determines whether the M redox reaction or oxygen-redox reaction occurs. As the redox-active O 2p is essentially nonbonding, Ceder et al. concluded that hybridization of M nd is not an effective mechanism to control the oxygen-redox reactions. However, the proposed nonbonding O 2p scheme ignores the π-type b1/b1* molecular-orbital formation between the M and O orbitals, as described in Figures 4 and 5. In this context, the research group led by Doublet emphasized the importance of orbital hybridization in M−O bonds using DFT

Figure 6. Schematic energy diagram of (a) eg/t2g molecular orbitals of an MO6 octahedron with M of 3d(t2g6egn) and b1/b1* molecular orbitals of an OM2Li4 octahedron with M of (b) 3d(t2gn), (c) nd(t2g0), and (d) 4d/5d(t2gn) before and after oxygen oxidation. The energy level of each orbital is qualitative and altered by configuration interactions and broadening under periodic potential.

the molecular orbitals can be relabeled as follows: four O 2p orbitals form σ-type bonding (a1 and b2) and antibonding (a1* and b2*) molecular orbitals with M nd(eg), (n+1)s, and (n+1)p, E

DOI: 10.1021/acsami.7b09835 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

Review

ACS Applied Materials & Interfaces

Figure 8. Schematic diagram of core-level spectroscopies. (a) Oxygen partial density of states (pDOS) for A2MO3, (b) XAS spectroscopy, (c) XES spectroscopy, (d) elastic RXES, and (e) RIXS. corresponds to excitation from occupied 2p states to unoccupied 2p states, giving resonant inelastic X-ray scattering (RIXS, Figure 8e). Therefore, RIXS directly probes occupied O 2p orbitals. The probing depth of XAS (fluorescence yield) and XES is ∼100 nm, which allows monitoring the oxygen orbitals in bulk. The O K-edge XAS experiments were first conducted for the Liexcess layered transition-metal oxide Li1.16Ni0.15Co0.19Mn0.5O2 by the research group led by Ogumi in 2012 (Figure 9a),74 before the first postulation of an oxygen-redox reaction. Therefore, the authors did not mention the oxygen-redox reaction. Nevertheless, when we

Figure 7. Visualized oxygen-hole orbitals in (a) lithium-excess layered transition-metal oxide Li0.33Ni0.25Mn0.58O2 (Li: green, Ni: black, Mn: purple, O: red) and (b) Li0.5Ru0.5Sn0.5O3 (Ru: brawn, Sn: gray).32 (c) Energy diagram of Mott−Hubbard, charge-transfer, and intermediate regimes. The dotted circles highlight the redox-active electrons. U and Δ represent the on-site Coulombic repulsion and the charge-transfer parameter, respectively.50 (a and b) Reprinted with permission from ref 32. Copyright 2016 Springer Nature. (c) Reprinted with permission from ref 50. Copyright 2017 Royal Society of Chemistry. calculations on the level of GGA+U.49,50 For example, considering that M in the 4d/5d groups have much more stable σ-type molecular orbitals than do M of the 3d group, an increased M−O covalency suppresses oxygen loss upon oxygen oxidation, which explains the small irreversible capacity for the 4d or 5d transition-metal oxides such as β-Li2IrO3 or Na2RuO3. Furthermore, transition-metal oxides in the charge-transfer regime (U ≫ Δ, U: on-site Coulomb repulsion, Δ: charge transfer energy) were claimed to have an inaccessibly stable M nd electron relative to the O 2p electron, and the unstable oxidized oxygen leads to oxygen gas release (Figure 7c). Indeed, oxygen gas release is frequently observed for oxygen-redox electrode materials composed of 3d transition metals (e.g., Mn4+, Co4+, Ni4+),41,42 with their large U set in the charge transfer regime.

5. EXPERIMENTAL APPROACHES From a molecular-orbital viewpoint, the occupied b1* molecular orbital with a predominant O 2p character (Figure 5c) plays a crucial role in the reversible oxygen-redox reaction. However, it is difficult to directly investigate the oxygen states, especially in the form of composite electrodes. For example, XPS has a typical probing depth of a few nanometers, and information on the oxygen-redox reaction in bulk is, in many cases, veiled by surface contaminations. Although many reports have relied on diffraction techniques (X-ray diffraction, neutron diffraction, and electron diffraction),70−72 they provide only indirect and limited information on the target orbitals. Experimental techniques capable of direct oxygen probing include soft X-ray absorption and emission spectroscopies (XAS and XES, Figures 8a−e).73 A soft X-ray in the energy range of 525−550 eV excites O 1s core electrons to unoccupied O 2p (K-edge), allowing the direct probing of unoccupied states (XAS, Figure 8b). After the excitation occurs, a portion of the excited electrons directly relaxes to the 1s core hole (without a secondary excitation) to give an elastic resonant X-ray emission spectrum (elastic RXES, Figure 8d). The other 1s core holes get filled with electrons relaxed from occupied 2p states, where energy loss from the absorption to the emission

Figure 9. Oxygen-redox reaction monitored by the core-level spectroscopies. XAS for (a) Li-excess layered transition-metal oxide Li 1.16−xNi 0.15Co 0.19 Mn0.5 O 2 ,73 (b) Li-excess random rock salt Li1.2−xTi0.4Mn0.4O2,52 and (c) Na-excess layered transition-metal oxide Na2−xRuO3 (the samples i, ii, iii, and iv are x = 0, 0.5, 1.0, and 1.4, respectively).60 (d) RIXS for lithium-excess layered transitionmetal oxide Li1.2−xNi0.13Co0.13Mn0.54O2 (the samples i, ii, iii, and iv are x = 0, 0.4, 1.0, and 1.1, respectively).43 (a) Reprinted with permission from ref 74. Copyright 2013 Elsevier. (d) Reprinted with permission from ref 43. Copyright 2016 Springer Nature. F

DOI: 10.1021/acsami.7b09835 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

Review

ACS Applied Materials & Interfaces

oxidation. This new peak may correspond to the occupied bonding b1 molecular orbital, which was predicted by the molecular-orbital theory. On the basis of the reported XAS and RIXS spectra with a molecular orbital expression of the related orbitals, we schematically describe a partial density of states of oxygen in oxygen-redox battery electrodes, as shown in Figure 10. “A strong restructuring of the electronic distribution around oxygen sites”35 upon oxygen oxidation is fully explained by the formation of the bonding/antibonding b1/b1* molecular orbitals through a π-type O 2p−Mn 3d(t2g) interaction (Figures 5c and 6b).

examine the reported spectra upon charge in a low voltage region, a new shoulder emerges in the lowest energy region (527.5 eV), suggesting oxidation of the 3a1*/3b2*/b1* molecular orbitals with predominant Ni/Co 3d character (i.e., Ni4+/Ni2+ and Co4+/Co3+, Figure 5b). A further charge in the oxygen-redox region generates a new shoulder in the high-energy region (530.5 eV), which could be attributed to the unoccupied b1* molecular orbital primarily with an O 2p character (Figure 6b). In general, by oxidation of an element, the energy level of core 1s is lowered more significantly than that of the frontier orbitals (Figure 10), thus shifting the XAS peaks to a higher-

7. SUMMARY AND OUTLOOK A molecular orbital approach predicted that the b1/b1* molecular orbitals play a key role in oxygen-redox reactions. Labile electrons in antibonding b1* with a predominant O 2p character are responsible for the extra oxygen-redox capacity in Li/Na-excess transition-metal oxides, while enhanced π-type orbital hybridization between O 2p and M t2g upon oxygen oxidation yields a large b1/b1* energy splitting, thus stabilizing the oxidized oxygen. Despite the tremendous experimental and theoretical efforts devoted to oxygen-redox chemistry as described in this review, additional points still need to be clarified for the practical application of oxygen-redox electrodes in batteries. For example, there is no direct and quantitative evidence of the π-type interaction between O 2p and M t2g. As this interaction governs the oxygen-redox capacity, experimental and theoretical examinations are necessary. Full knowledge of the O 2p−M t2g interaction would reveal which M is the best transition metal for stable and high-capacity oxygen-redox electrode materials. Potential experimental techniques to quantify this interaction are XAS and XES. Theoretical calculations are also expected to illuminate the key molecular orbital once oxygen-redox reactions occur. Another important issue is the role played by M with d0/d10 configurations in oxygen-redox reactions. From a molecularorbital viewpoint, M with d0/d10 configurations should not stabilize oxidized oxygen (Figure 6c), although many electrode materials with d0/d10M such as Li2RuO3−Li2MO3 (M = Sn4+) or Li1.2Ti0.4Mn0.4O2 undergo stable oxygen-redox reactions.37,44 Therefore, it is important to clarify the contribution of these highly ionic M−O bonds to stabilize the extra oxygen-redox reaction. XAS and XES can provide quantitative information on the covalency/ionicity of the M−O bonds. In conclusion, recent advances in oxygen-redox battery electrodes have highlighted the importance of the chemical states of oxygen. Because oxygen-redox chemistry is an important foundation to develop better oxygen-redox electrode materials, the redefined electronic structure focusing on oxygen in this review allows generalization of the reaction mechanisms of various oxygen-redox battery electrodes, offering a rationale for materials design for higher energy-density batteries.

Figure 10. Schematic illustration of oxygen pDOS before/after the oxygen oxidation of battery electrodes. energy side. Therefore, it is reasonable to assign the emergence of the high-energy XAS peak to oxygen oxidization. The assignment to the unoccupied b1* orbital is also supported by the similar O K-edge XAS results reported for Li1.2Ni0.13Co0.13Mn0.54O2 and Li1.2Ni0.2Mn0.6O2.35,36 The O K-edge XAS experiments were also conducted for the Liexcess random rock salt Li1.2Ti0.4Mn0.4O2 (Figure 9b).52 Similar to the above results for the layered compounds, a new XAS peak emerges at 530 eV upon oxygen oxidation, suggesting again the lowering of the O 1s energy with oxygen oxidation. Because Ti4+ cannot form a redoxactive molecular orbital with a predominant O 2p character (Figure 6c), the oxidized oxygen in Li1.2Ti0.4Mn0.4O2 may be stabilized by the π-type b1* molecular orbital formation in an OMn2Li4 octahedron. The O K-edge XAS spectrum for Na2RuO3 after oxygen oxidation exhibits a new broad XAS peak in the high energy region of 533 eV (Figure 9c),60 possibly because the diffused Ru 4d(t2g) orbital results in a large orbital hybridization through an enhanced π-type interaction. Alternatively, as proposed by Tarascon et al., a short O−O distance induces a σ-type interaction to further stabilize the oxidized oxygen.45−49 The O K-edge RIXS spectra for Li-excess layered transition-metal oxides Li1.2Ni0.13Co0.13Mn0.54O2 and Li1.2Ni0.2Mn0.6O2 were reported by the research group led by Bruce (Figure 9d).43,44 On the basis of an enhanced elastic RXES peak after oxygen oxidation (the dotted region in Figure 9d), the group concluded that a localized oxygen hole exists, which contradicts the molecular-orbital prediction of π-type b1* molecular-orbital formation of the oxidized oxygen. Here, it is important to note that the elastic scattering becomes intense under a resonant condition.75 In the experiments conducted by the Bruce group (excitation by 531.8 eV photons), a new XAS peak emerged at 531.8 eV after oxygen oxidation, whereby the resonant condition was applicable, which intensifies the elastic scattering. Therefore, the enhanced elastic scattering cannot provide evidence for the existence of a localized oxygen hole. Focusing on the reported RIXS spectra rather than the elastic RXES peak (Figure 9d), a new RIXS peak appears at 523 eV after oxygen

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Masashi Okubo: 0000-0002-7741-5234 Atsuo Yamada: 0000-0002-7880-5701 Notes

The authors declare no competing financial interest. G

DOI: 10.1021/acsami.7b09835 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

Review

ACS Applied Materials & Interfaces

■

(17) Lu, Z.; Dahn, J. R. Understanding the Anomalous Capacity of Li/Li[NixLi(1/3−2x/3)Mn(2/3‑x/3)]O2 Cells Using In Situ X-ray Diffraction and Electrochemical Studies. J. Electrochem. Soc. 2002, 149, A815−A822. (18) Armstrong, A. R.; Holzapfel, M.; Novak, P.; Johnson, C. S.; Kang, S. H.; Thackeray, M. M.; Bruce, P. G. Demonstrating Oxygen Loss and Associated Structural Reorganization in the Lithium Battery Cathode Li[Ni0.2Li0.2Mn0.6]O2. J. Am. Chem. Soc. 2006, 128, 8694− 8698. (19) Robertson, A. D.; Bruce, P. G. The Origin of Electrochemical Activity in Li2MnO3. Chem. Commun. 2002, 2790−2791. (20) Yabuuchi, N.; Yoshii, K.; Myung, S. T.; Nakai, I.; Komaba, S. Detailed Studies of a High-Capacity Electrode Material for Rechargeable Batteries, Li2MnO3-LiCo1/3Ni1/3Mn1/3O2. J. Am. Chem. Soc. 2011, 133, 4404−4419. (21) Hong, J. H.; Lim, H. D.; Lee, M.; Kim, S. W.; Kim, H.; Oh, S. T.; Chung, G. C.; Kang, K. Critical Role of Oxygen Evolved from Layered Li-Excess Metal Oxides in Lithium Rechargeable Batteries. Chem. Mater. 2012, 24, 2692−2697. (22) Hy, S.; Liu, H.; Zhang, M.; Qian, D.; Hwang, B. J.; Meng, Y. S. Performance and design considerations for lithium excess layered oxide positive electrode materials for lithium ion batteries. Energy Environ. Sci. 2016, 9, 1931−1954. (23) Hong, J.; Gwon, H.; Jung, S. K.; Ku, K.; Kang, K. ReviewLithium-Excess Layered Cathodes for Lithium Rechargeable Batteries. J. Electrochem. Soc. 2015, 162, A2447−A2467. (24) Koga, H.; Croguennec, L.; Menetrier, M.; Mannessiez, P.; Weill, F.; Delmas, C. Different Oxygen Redox Participation for Bulk and Surface: A Possible Global Explanation for the Cycling Mechanism of Li1.20Mn0.54Co0.13Ni0.13O2. J. Power Sources 2013, 236, 250−258. (25) Koga, H.; Croguennec, L.; Menetrier, M.; Douhil, K.; Belin, S.; Bourgeois, L.; Suard, E.; Weill, F.; Delmas, C. Reversible Oxygen Participation to the Redox Processes Revealed for Li1.20Mn0.54Co0.13Ni0.13O2. J. Electrochem. Soc. 2013, 160, A786−A792. (26) Aydinol, M. K.; Kohan, A. F.; Ceder, G.; Cho, K.; Joannopoulos, J. Ab initio study of lithium intercalation in metal oxides and metal dichalcogenides. Phys. Rev. B: Condens. Matter Mater. Phys. 1997, 56, 1354−1365. (27) Ohzuku, T.; Nagayama, M.; Tsuji, K.; Ariyoshi, K. High-capacity lithium insertion materials of lithium nickel manganese oxides for advanced lithium-ion batteries: toward rechargeable capacity more than 300 mAh g−1. J. Mater. Chem. 2011, 21, 10179−10188. (28) Rozier, P.; Tarascon, J. M. Review-Li-Rich Layered Oxide Cathodes for Next-Generation Li-Ion Batteries: Chances and Challenges. J. Electrochem. Soc. 2015, 162, A2490−A2499. (29) Yabuuchi, N. Solid-State Redox Reaction of Oxide Ions for Rechargeable Batteries. Chem. Lett. 2017, 46, 412−422. (30) Szabo, A.; Ostlund, N. S. Modern Quantum Chemistry: Introduction to Advanced Electronic Structure Theory; Dover Publications: New York, 1996. (31) Cox, P. A. Transition Metal Oxides: An Introduction to Their Electronic Structure and Properties; Oxford Press: Oxford, 1992. (32) Seo, D. H.; Lee, J.; Urban, A.; Malik, R.; Kang, S. Y.; Ceder, G. The Structural and Chemical Origin of the Oxygen Redox Activity in Layered and Cation-Disordered Li-Excess Cathode Materials. Nat. Chem. 2016, 8, 692−697. (33) Jansen, V. M.; Hoppe, R. Zur Kenntnis der NaCl-Strkturfamilie: Neue Untersuchungen an Li2MnO3. Z. Anorg. Allg. Chem. 1973, 397, 279−289. (34) Strobel, P.; Lambert-Andron, B. Crystallographic and Magnetic Structure of Li2MnO3. J. Solid State Chem. 1988, 75, 90−98. (35) Kalyani, P.; Chitra, S.; Mohan, T.; Gopukumar, S. Lithium Metal Rechargeable Cells Using Li2MnO3 as the Positive Electrode. J. Power Sources 1999, 80, 103−106. (36) Robertson, A. D.; Bruce, P. G. Mechanism of Electrochemical Activity in Li2MnO3. Chem. Mater. 2003, 15, 1984−1992. (37) Numata, K.; Sakaki, C.; Yamanaka, S. Synthesis of Solid Solutions in a System of LiCoO2-Li2MnO3 for Cathode Materials of Secondary Lithium Batteries. Chem. Lett. 1997, 26, 725−726.

ACKNOWLEDGMENTS We acknowledge the financial support received from the Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan; the Grant-in-Aid for Specially Promoted Research no. 15H05701. M.O. was financially supported by the Iketani Science and Technology Foundation. We are grateful to B. Mortemard de Boisse, E. Watanabe, and Y. Harada at the University of Tokyo, D. Asakura at the National Institute of Advanced Industrial Science and Technology, and M. Nakayama at the Nagoya Institute of Technology for their fruitful discussions.

■

REFERENCES

(1) Van Noorden, R. A Better Battery. Nature 2014, 507, 26−28. (2) Tarascon, J. M.; Larcher, D. Towards Greener and More Sustainable Batteries for Electrical Energy Storage. Nat. Chem. 2014, 7, 19−29. (3) Chevrier, V. L.; Obrovac, M. N. Alloy Negative Electrodes for LiIon Batteries. Chem. Rev. 2014, 114, 11444−11502. (4) Poizot, P.; Laruelle, S.; Grugeon, S.; Dupont, L.; Tarascon, J. M. Nano-Sized Transition-Metaloxides as Nagative-Electrode Materials for Lithium-Ion Batteries. Nature 2000, 407, 496−499. (5) Lin, D. C.; Liu, Y. Y.; Cui, Y. Reviving the Lithium Metal Anode for High-Energy Batteries. Nat. Nanotechnol. 2017, 12, 194−206. (6) Barpanda, P.; Ati, M.; Melot, B. C.; Rousse, G.; Chotard, J. N.; Doublet, M. L.; Sougrati, M. T.; Corr, S. A.; Jumas, J. C.; Tarascon, J. M. A 3.90 V Iron-Based Flourosulphate Material for Lithium-Ion Batteries Crystallizing in the Triplite Structure. Nat. Mater. 2011, 10, 772−779. (7) Barpanda, P.; Oyama, G.; Nishimura, S.; Chung, S. C.; Yamada, A. A 3.8 V Earth-Abundant Sodium Battery Electrode. Nat. Commun. 2014, 5, 4358. (8) Grimaud, A.; Hong, W. T.; Shao-Horn, Y.; Tarascon, J. M. Anionic Redox Processes for Electrochemical Devices. Nat. Mater. 2016, 15, 121−126. (9) Mizushima, K.; Jones, P. C.; Wiseman, P. J.; Goodenough, J. B. LixCoO2 (0