Over the past three decades, Li-ion batteries have dominated the rechargeable battery technology for powering portable electronic devices, such as cell phones, laptop computers and iPads that have transformed global communications^[1]. More recently, with the growing demand for secure and sustainable energy supply of modern society, new applications of Li-ion batteries quickly emerge especially for powering electric vehicles and assembling grid energy storage systems for renewable energies^[2−4]. However, to meet these large-scale applications, it is important to improve the energy density of Li-ion batteries to over 500 Wh/kg, which is still beyond the current Li-ion technology. This is because, essentially, all of today’s Li-ion batteries use intercalation reactions at both electrodes, which are in most cases restricted to the one-electron redox process per transition metal (TM) ion^[5]. This redox mechanism imposes a fundamental limitation on the capacity of both electrodes, especially the cathodes, of Li-ion batteries. For instance, the two representative cathodes used today are LiCoO₂^[6] and LiFePO₄^[7], both of which show limited capacities of below 200 mAh/g.

Notably, the discovery of Li-rich TM oxides, Li_1+xTM_1−xO₂, as high-capacity cathodes at the beginning of this century may provide a promising way to address this challenge^[8−11]. It has been demonstrated that materials, such as Li_1+xNi_yCo_zMn_1−x-y-zO₂, can deliver a capacity of over 250 mAh/g, much higher than that of layered TM oxides^[8]. The excess capacity of Li_1+xTM_1−xO₂ is now recognized to come from the anionic redox activity involving O-dominated states, which is beyond the traditional cationic redox chemistry^[12−14]. However, the real-world application of these high-capacity cathodes is still plagued by critical issues, such as sluggish kinetics, voltage hysteresis, as well as long-term capacity and voltage fade^[15−19]. These degradations of Li_1+xTM_1−xO₂ have been frequently ascribed to their structural changes upon delithiation, associated with the formation of various defects, such as TM migration and lattice oxygen loss^{[20, 21]}. Nevertheless, the origin of these detrimental structural changes and their precise relationship to the redox chemistry of Li_1+xTM_1−xO₂ remains unknown, leaving the question of how to control and improve the performances of these promising cathodes open. Notably, different from their 3d counterparts, the recent experiments have demonstrated that 4d and 5d Li_1+xTM_1−xO₂, such as Li₂RuO₃ and Li₂IrO₃, can possess much better cycling reversibility with negligible capacity fade and O₂ release^[22−25]. These observations imply that the atomic configuration of TM ions may have significant influences on the structural stability of Li_1+xTM_1−xO₂ and thus provide a promising way to improve their cycling performances by adjusting TM components, whereas the underlying mechanism behind this phenomenon is still not fully understood.

To address these challenges, here we performed first-principles calculations to systematically study the electronic structures and defect properties of Li₂MnO₃ and Li₂RuO₃, which are the representative 3d and 4d Li_1+xTM_1−xO₂, respectively. Our calculations revealed that the structural and cycling stability of these two cathodes depend closely on their electronic structures, especially the energy of their highest occupied electronic states (HOS), from which the compensating electrons would be removed upon delithiation. For Li₂MnO₃, we found that its HOS are mainly determined by O 2p orbitals, thus having a relatively low energy. As a consequence, injection of holes into these HOS upon delithiation would spontaneously trigger the formation of various defects, such as Mn−Li antisites (Mn_Li), Mn interstitials (Mn_i), Li interstitials (Li_i), and O vacancies (V_O), in Li₂MnO₃, which are detrimental to its structural stability. The underlying mechanism is that, due to the low-lying HOS, the charge transfer from defect levels to the injected holes upon delithiation could gain an energy that is sufficiently large to compensate the energy cost for the defect formation in Li₂MnO₃. While for Li₂RuO₃, on the other hand, we found that its HOS have a predominant Ru 4d component that promotes these states to a higher energy than that of Li₂MnO₃. Consequently, the charge transfer from defect levels to hole states could gain only a small energy that is not sufficient to trigger the defect formation in Li₂RuO₃. Thus, by carefully comparing these two prototype cathodes, our calculations uncovered an electronic origin for the structural degradation in 3d Li_1+xTM_1−xO₂ and meanwhile explained why the cycling reversibility can be significantly improved in their 2p counterparts.

The first-principles calculations were carried out by using the projector augment wave (PAW) method^[26] within the framework of density functional theory (DFT) as implemented in VASP^[27], with the exchange-correlation energy treated by the strongly constrained and appropriately normed (SCAN) functional^[28]. The wave functions were expanded by using the plane waves up to a kinetic energy cutoff of 520 eV and the Brillouin-zone integration was approximated with a 12 × 12 × 12 k-mesh. During structural optimizations, the lattice vectors and atomic coordinates were all fully relaxed until the force on each atom is less than 0.01 eV/Å. The defect calculations were performed by using the DASP code^[29] with a supercell of 432 atoms and a gamma-only k point.

The crystal structure of Li₂MnO₃ consists of cubic closed-packed O ions with alternative sheets of octahedral sites between them occupied by Li and [Li_1/3Mn_2/3], as show in Fig. 1(a). If Li and Mn ions in [Li_1/3Mn_2/3] layers are fully disordered, then the structure resembles that of layered LiTMO₂ with the R3¯m symmetry. However, the experimental measurements have identified that [Li_1/3Mn_2/3] layers actually possess a honeycomb-like cation ordering, which leads to the structure of Li₂MnO₃ with a lowered symmetry of C2/m^[30]. Like Li₂MnO₃, Li₂RuO₃ also crystalizes into a layered structure, where Li and [Li_1/3Ru_2/3] layers stack alternatively, giving a formula of Li[Li_1/3Ru_2/3]O₂ in the conventional layered LiTMO₂ notation, as shown in Fig. 1(b). However, compared to Li₂MnO₃, a minor structural difference still exits for Li₂RuO₃ as its unit cell has the C2/c symmetry with a slightly distorted O framework^[31]. The calculated lattice constants of the unit cells for both Li₂MnO₃ and Li₂RuO₃ are given in Table 1, which show good agreement with the experimental measurements^{[30, 31]}.

Fig. 2(a) plots the spin-polarized band structure and projected density of states (PDOS) for Li₂MnO₃. It is clear to see that our calculations with the SCAN functional indicated that Li₂MnO₃ is a semiconductor with a band gap of 2.1 eV, which is in consistent with the previous DFT + U calculations^[32]. Moreover, from the PDOS, we found a profound hybridization between the Mn 3d and O 2p orbitals within a wide energy range from −6 to 5 eV, indicating strong covalent bonds formed between the Mn and O ions. The conduction band minimum (CBM) and valence band maximum (VBM) of Li₂MnO₃ were identified to have the predominant Mn 3d and O 2p characters, respectively, which can be further understood by the molecular-orbital diagram, as shown in Fig. 2(b). Here, since each Mn ion is octahedrally coordinated with eight O ions, its 3d orbitals would split into the triply degenerate t2g and doubly degenerate eg states, which can further couple with O 2p orbitals to form the bonding and antibonding molecular orbitals (MOs). For Li₂MnO₃, the Mn⁴⁺ ions have the 3d3 atomic configuration and the high-spin magnetic ground state, as shown in Table 1. Thus, its CBM and VBM correspond to the Mn-dominated and O-dominated MOs, respectively. Based on this band structure analysis, it is natural to expect that, as Li ions are extracted from Li₂MnO₃ upon delithiation, the compensating electrons would be removed from the O-dominated MOs, thereby suggesting an anionic redox process, in accordance with the experimental observations^[9]. However, it should be noted that injection of holes into O 2p orbitals may not be stable, since it usually leads to the formation of compensating defects^{[33, 34]}.

To check the structural stability of Li₂MnO₃ upon delithiation, its defect properties were calculated, following the standard defect calculation method^{[29, 34]}. Fig. 3(a) shows a chemical potential region for Li₂MnO₃ with the boundaries set by the competing phases in the Li-Mn-O phase diagram. This chemical potential region represents the thermodynamically equilibrium conditions under which Li₂MnO₃ could be stabilized. Here, to represent the chemical environment in Li-ion batteries, we focused on the Li-rich condition, as labeled in Fig. 3(a). Fig. 3(b) illustrates the calculated formation energies for a series of defects in Li₂MnO₃, including vacancies, antisites, and interstitials, as a function of the electron Fermi energy (EF). At the Li-rich condition, we found that the EF is pinned at 0.27 eV below the CBM by a pair of donor and acceptor defects (VLi− and Lii+), and at this EFpin, VO0 has the lowest formation energy of 0.56 eV among all the considered defects. These results indicated that, at the Li-rich condition, the defect formation in Li₂MnO₃ should be difficult, due to the relatively large energies that are needed to create defects. Nevertheless, it is important to note that, as Li ions are extracted upon delithiation, the EF would be continuously pushed down to the VBM. In this case, our calculations showed that various defects, such as MnLi3+, Mni4+, VO2+, Lii+, LiO3+, and MnO4+, could form spontaneously in Li₂MnO₃, as they have negative formation energies. This result is in general consistent with the experimental observations that the cation migration and lattice oxygen loss usually occur in Li₂MnO₃ upon delithiation^{[20, 21]}.

To get more insight into the defect formation in Li₂MnO₃ upon delithiation, we proposed a defect charge transfer mechanism, as schematically illustrated in Fig. 3(c). As discussed above, before delithiation, Li₂MnO₃ is a semiconductor with its HOS (i.e., the VBM states) fully occupied. In this case, creating a donor defect in Li₂MnO₃ would yield a defect level with occupied electrons, which usually leads to a large formation energy since it needs to break the strong Mn−O bonds, as shown in the left panel of Fig. 3(c). As a consequence, before delithiation, defects cannot form easily in Li₂MnO₃. While, on the other hand, when Li ions are extracted from Li₂MnO₃ upon delithiation, holes are injected into its HOS. Thus, the electrons originally occupied the defect levels can now drop into these hole states and gain an energy (ΔE), as illustrated in the right panel of Fig. 3(c). If ΔE is large enough to compensate the energy cost for breaking the Mn−O bonds, defects could form spontaneously. Given that the HOS of Li₂MnO₃ are mainly determined by O 2p orbitals and thus have a low energy, ΔE could be large, which should be mainly responsible for the observed defect formation in Li₂MnO₃ upon delithiation^{[20, 21]}.

Compared to Li₂MnO₃, it has been widely observed that Li₂RuO₃ could have a greatly improved cycling reversibility^{[22, 23, 25]}. To understand this phenomenon, we next discussed the electronic structure and defect properties of Li₂RuO₃. Fig. 4(a) plots the spin-polarized band structure and PDOS for Li₂RuO₃. Different from its 3d counterpart, our calculations indicated that Li₂RuO₃ is a metal with its EF crossing only the spin-down channel. This happens because the large crystal field splitting of Ru 4d orbitals results in a low-spin magnetic ground state (see Table 1), as illustrated by the molecular-orbital diagram in Fig. 4(b). Accordingly, the HOS of Li₂RuO₃ should correspond to the t2g*↓ states that have a predominant Ru 4d character. Thus, compared to Li₂MnO₃, we can expect that the HOS of Li₂RuO₃ should have a much higher energy, which would prevent it from the defect formation, according to the defect charge transfer mechanism. Actually, by aligning the band structures of these two cathodes, we found that the HOS of Li₂RuO₃ have an energy of 1.68 eV higher than that of Li₂MnO₃. Moreover, we also calculated the formation energies of several important defects in Li₂RuO₃ at the Li-rich condition, including Ru_Li, Ru_i, Li_i, and V_O. The results indicated that all these considered defects have relatively large formation energies, with the smallest value of 3.33 eV found for V_O. Notably, different from Li₂MnO₃, there are nearly infinite numbers of electrons and holes at the HOS of metallic Li₂RuO₃ and thus the electrons originally occupying the defect level may jump into the HOS. For instance, the formation energy of VO2+ was calculated to be 2.35 eV, which is smaller than that of VO0, due to the charge transfer from the defect level to the HOS. It should be noted that the delithiation would not significantly change the energy of HOS and thus the defect properties of Li₂RuO₃. These calculations provide a good explanation on the improved cycling reversibility of Li₂RuO₃.

In this work, based on the first-principles calculations with the SCAN functional, we systematically studied the electronic structures and defect properties of two prototype Li-rich TM oxides, Li₂MnO₃ and Li₂RuO₃, in order to get more understanding on the structural degradation of these high-capacity cathodes upon delithiation. Our results indicated that the structural and cycling stability of Li₂MnO₃ and Li₂RuO₃ depend closely on their electronic structures, especially the energy of their HOS, from which the compensating electrons would be extracted upon delithiation. We found that Li₂MnO₃ is a semiconductor with a band gap of 2.1 eV and thus its HOS correspond to the VBM states that are mainly determined by O 2p orbitals. The defect calculations showed that, before delithiation, Li₂MnO₃ is stable against the defect formation, due to the relatively large energies that are needed to create defects. Whereas, upon delithiation, we found that various defects could form spontaneously in Li₂MnO₃, as their formation energies become negative. The underlying mechanism of this phenomenon was ascribed to the charge transfer from defect levels into the injected holes at the HOS of Li₂MnO₃ upon delithiation, which could gain an energy that is sufficiently large to compensate the energy cost for the defect formation. While, on the other hand, different from Li₂MnO₃, Li₂RuO₃ was found to be a metal with its HOS predominately determined by Ru 4d orbitals and thus having a relatively high energy. As a consequence, the charge transfer from defect levels to hole states at its HOS could gain only a small energy that is not sufficient to trigger the defect formation, which explains why a good cycling reversibility could be realized in Li₂RuO₃.