Introduction

Cathode materials are a key component of lithium-ion batteries (LIBs) to determine electrochemical properties such as energy densities and rate performance [1,2,3]. Li-rich layered oxides (LLOs) have attracted much attention due to their high theoretical capacity of 250 mAh/g, which arises from the cumulative redox reactions of TM and oxygen ions [4,5,6,7,8]. It is widely accepted that the lattice oxygen with a linear Li–O–Li configuration provides extra specific capacity [9,10,11,12,13]. However, the irreversibility of oxygen redox makes LLOs undergo lattice oxygen release, structural degradation, and poor cycle performance [14, 15]. Consequent problems such as low initial Coulombic efficiency (ICE) and severe capacity/voltage fading seriously hinder commercial applications of LLOs [16,17,18,19].

LLOs structures with good reversibility of oxygen redox are being sought. It is widely accepted that LLOs consist of twin LiTMO2 (TM = Mn, Co, Ni, etc.) and Li2MnO3-like domains [20, 21]. The key to controlling oxygen redox chemistry is to modulate the chemical environment of lattice oxygen with redox activity, that is, to modulate Li2MnO3-like domains. According to Pauling’s rule [22], in the Mn-based LLOs, the coordination configuration OLi4Mn2 with Mn4+ is stable, making aggregated Li2MnO3-like domains popular [23, 24]. Recent researches have demonstrated that the large-sized and aggregated Li2MnO3-like domain would cause irreversible TM migration and oxygen release [25,26,27]. Cho’s group found that the weak Mn–O covalency promoted excessive oxidation of oxygen ions and chemical irreversibility [25, 28]. Liu discovered that oxygen redox reversibility and cation migration could be suppressed by constructing Li2ZrO3 slabs into Li2MnO3-like domains [29]. Yu’s study found that the 3:7 component with the least Li2MnO3-like component among xLi2MnO3·yLiMn0.42Ni0.42Co0.16O2 (x:y = 3:7, 5:5, 7:3) had the best electrochemical performance and the higher reversibility of oxygen redox [12]. Bruce found that the oxidized oxygen ions in the Na0.75[Li0.25Mn0.75]O2 and Li1.2Ni0.13Co0.13Mn0.54O2 with a honeycomb superstructure cannot be reduced and a large voltage hysteresis occurred, while the oxidization of lattice oxygen ions for systems with a band superstructure had good reversibility [30, 31]. Sun found that the aggregated Li2MnO3-like domains can lead to irreversible TM migration and oxygen release, while the dispersed Li2MnO3-like domains have inhibitory effects on TM migration and oxygen release [13]. Therefore, to control the size of Li2MnO3-like domains would be an effective strategy to improve the electrochemical performance of LLOs.

In our recent work, we demonstrated a highly stable layered Li1.2Ni0.2Mn0.6O2 LLOs with dispersed small-sized Li2MnO3-like domains by tuning the oxygen partial pressure during high-temperature calcination [32]. Our work revealed that ONiMn4+Mn3+Li3 coordination configurations stabilized the Li-rich Li-Ni-Mn–O structure with small-sized Li2MnO3-like domains [32]. The small-sized Li2MnO3-like domains suppressed the O-release and enhanced the reversibility of oxygen redox and cycling performance by a special oxidation mechanism [32].

In LLO materials, high Ni content can contribute to the high capacity. Though LLO materials with the high Ni content have been a hot research topic now, the mixed arrangement of Li/Ni is still a problem to influence the cycle performance [33,34,35,36]. Layered Li-rich Li-Co–Ni-Mn–O LLOs are often used as one of the most promising cathode materials for the next generation battery because of their high capacity [37, 38]. However, the cost of Co element is expensive. Therefore, to develop Co-free cathode materials for LIBs is being pursued. Fe resources are rich in the earth. Its electronic structure is similar to other TM elements such as Co and Ni in manganese-based cathode materials and the contribution of Fe to the capacity is similar to that of Co in the terms of redox pairs [39, 40]. Previous studies have shown that Fe-substitution can suppress the oxygen release and improve the electrochemical properties [39,40,41,42,43,44,45]. Of course, these performances were also affected by compositions of the electrolyte and electrode–electrolyte interphases and so on [46, 47].

Therefore, the work selects a cobalt-free Li-rich layered material, Li29Fe3Ni3Mn13O48(LFNMO), as the research object. Two models with dispersed Li2MnO3-like domains (D-LFNMO) and aggregated Li2MnO3-like domains (A-LFNMO) are built to study the Fe function and the effect of Li2MnO3-like domain size on the oxidation process, structures, and structural stability. This is different with our previous work in which the influence of the Li2MnO3-like domain size on electrochemical performance was only studied [32].Oxidation process of cations and anions and the oxidation mechanism are explored using density of state (DOS) and electric density. Additionally, the TM ions (particularly Mn4+) migration in LLOs with different Li2MnO3-like domain distribution also needs to be explored, since it is directly related to irreversible structural transformations and lattice oxygen release [48,49,50,51]. Consequently, structural stability is completed by calculating oxygen release energies and migration energies of Mn4+ ions from the TM layer to the Li vacancies at the TM layer and the Li layer. The study would provide a solid theoretical basis for the design of Li-rich cathode materials for cobalt-free lithium-ion batteries with high specific capacity and structure stability.

Calculation method and model

All calculations were performed using the VASP (Vienna Ab-initio simulation package) program. The spin-polarized generalized-gradient approximation (GGA) method with the Perdew-Burke-Ernzerhof (PBE) exchange correlation function [52] was used to study the Li29Fe3Ni3Mn13O48 system (C2/m space group, α = γ = 90°, β ≠ 90°, a ≠ b ≠ c) containing 96 atomic positions (Li1.208Fe0.125Ni0.125Mn0.542O2 formula). The core electrons were represented by the projector-augmented wave (PAW) potentials [53], and the valence electrons Li 2s1, Fe 3d74s1, Ni 3d84s2, Mn 3d64s1, and O 2s22p4 were expanded at a plane wave with the cutoff energy of 500 eV. Considering that the system contains the TM element with high angular quanta, we used the GGA + U method proposed by Dudarev et al. [54]. The included parameters U and J are not set separately, and only the difference value U − J is meaningful. Therefore, we simplified the parameters to U, and U values of Fe, Ni, and Mn were 4.9 eV, 6.0 eV, and 4.9 eV, respectively, similar to our previous work [55]. The energy and force convergence criteria were set to 10−4 eV and 0.02 eV/Å, respectively. Ion and unit cell relaxations and the electronic structures were performed on the 96-atom-position model using a 2 × 1 × 4 and 4 × 2 × 8 Monkhorst–Pack (MP) k-points, respectively. Various possible ferromagnetic (FM) and antiferromagnetic (AFM) structures are considered. Calculations show that the most stable magnetic structure of Li29Fe3Ni3Mn13O48 is AFM, see Fig. 1(a) and (b).

Fig. 1
figure 1

A-LFNMO (a) and D-LFNMO (b) models of Li29Fe3Ni3Mn13O48. The ΔHLi of A-LFNMO (c) and D-LFNMO (d) as x in Li29-xFe3Ni3Mn13O48. The stable de-lithiation structures of A-LFNMO (e) and D-LFNMO (f) as x in Li29-xFe3Ni3Mn13O48

Full size image

According to Pauling’s rule [22], in the Mn-based LLOs, the coordination configuration OLi4Mn2 with Mn4+ is stable, making aggregated Li2MnO3-like domains popular [23, 24]. Recently, we have successfully synthesized layered Li-Ni-Mn–O LLOs with dispersed small-sized and aggregated Li2MnO3-like domains by tuning the oxygen partial pressure during high-temperature calcination, respectively, named A-LNMO and D-LNMO [32]. The average grain size of Li2MnO3-like domains is 23.57 nm for D-LNMO, which is almost half that of A-LNMO (56.25 nm) [32]. D-LNMO possesses highly reversible oxygen redox and exceptional structural stability, exhibiting superior cycling stability of high capacity [32]. DFT calculation [32] selects Li-rich aggregated and dispersed models with 5 Å × 9 Å × 5 Å to simulate the LLOs with dispersed small-sized and aggregated Li2MnO3-like domains and well understands the experimental result in the terms of atomic and electronic structures. Therefore, this work employs a similar model to explore the effect of Li2MnO3-like domain size and Fe on the structural stability and oxidization. Herein, a cobalt-free Li-rich layered Li29Fe3Ni3Mn13O48 with dispersed Li2MnO3-like domains (D-LFNMO) and aggregated Li2MnO3-like domains (A-LFNMO) is designed, see Fig. 1. In A-LFNMO and D-LFNMO, all possible structures with various atomic occupations are considered, and structures with the lowest energies are shown in Fig. 1(a) and (b), respectively. Ni likely occupies the rich-Li site in layered Li2MnO3 structures and Fe preferably occupies the Mn site in layered Li2MnO3 structures. The calculation finds that the energy difference between A-LFNMO and D-LFNMO structure with the lowest energies is 0.245 meV/fu (slight smaller than electron energy of 26 meV at the room temperature), namely, they can both exist in the synthetic compounds. In addition, we have found that the interfacial O ions between twin Li2MnO3-like and LiTMO2 domains were easily triggered to participate into charge compensation [32].

In order to evaluate the structural stability of the Li29-xFe3Ni3Mn13O48 (0 ≤ x ≤ 24) system during charging, the ΔHLi was calculated and the corresponding formula was followed:

$$\Delta {\it{H}}_{\text{Li}}={\it{E}}\left({\text{Li}}_{29 - {\it{x}}}{\text{Fe}}_{3}{{\text{Ni}}}_{3}{{\text{Mn}}}_{13}{{\text{O}}}_{48}\right)-\frac{\it{x}}{{29}} \, {\it{E}}\left({\text{Fe}}_{3}{{\text{Ni}}}_{3}{{\text{Mn}}}_{13}{{\text{O}}}_{48}\right)-(1-\frac{\it{x}}{{29}}){\it{E}}\left({\text{Li}}_{29}{{\text{Fe}}}_{3}{{\text{Ni}}}_{3}{{\text{Mn}}}_{13}{{\text{O}}}_{48}\right)$$
(1)

where E(Li29Fe3Ni3Mn13O48) and E(Fe3Ni3Mn13O48) were total energies of the Li29Fe3Ni3Mn13O48 systems with all and no Li+ ions, respectively. E(Li29-xFe3Ni3Mn13O48) was the total energy of the Li29-xFe3Ni3Mn13O48 system.

Instability of lattice oxygen in LLOs from aggregated Li2MnO3-like domains causes the irreversibility of oxygen redox and consequent structural damage and severe capacity/voltage fading. Herein, the enthalpy changes ΔHO of the Li29-xFe3Ni3Mn13O48 (0 ≤ x ≤ 24) expressing the O–release was calculated and the corresponding formula was as followed:

$$\Delta {\it{H}}_{\text{O}}={\it{E}}\left({\text{Li}}_{29 - {\it{x}}}{\text{Fe}}_{3}{{\text{Ni}}}_{3}{{\text{Mn}}}_{13}{{\text{O}}}_{47}\right)+1/2{\it{E}}({\text{O}}_{2})-{\it{E}}\left({\text{Li}}_{29 - {\it{x}}}{\text{Fe}}_{3}{{\text{Ni}}}_{3}{{\text{Mn}}}_{13}{{\text{O}}}_{48}\right)$$
(2)

where E(Li29-xFe3Ni3Mn13O48) and E(Li29-xFe3Ni3Mn13O47) were the total energies of Li29-xFe3Ni3Mn13O48 systems without and with one O vacancy, respectively. E(O2) was the energy of the O2 molecule. In addition, the average redox potential Vave of Li29-xFe3Ni3Mn13O48 as de-lithiation amount x was calculated, and Eq. (3) was as follows:

$${\it{V}}_{\text{ave}}\left(x1\le x\le x2\right)= -\frac{{\it{E}}\left({\text{Li}}_{29 - {\it{x}}{1}}{\text{Fe}}_{3}{{\text{Ni}}}_{3}{{\text{Mn}}}_{13}{{\text{O}}}_{48}\right)-{\it{E}}\left({\text{Li}}_{29 - {\it{x}}{2}}{\text{Fe}}_{3}{{\text{Ni}}}_{3}{{\text{Mn}}}_{13}{{\text{O}}}_{48}\right)-({\it{x}}2-{\it{x}}1){\it{E}}(\text{Li})}{({\it{x}}2 - {\it{x}}1){\it{e}}}$$
(3)

where E(Li) was the energy of one Li atom and E(Li29-x1Fe3Ni3Mn13O48) and E(Li29-x2Fe3Ni3Mn13O48) were the total energies of the Li29-xFe3Ni3Mn13O48 system as de-lithiation amount x is x1 and x2, respectively.

Bruce found that the oxidized O2− ions on charging retained in the Na0.75[Li0.25Mn0.75]O2 and Li1.2Ni0.13Co0.13Mn0.54O2 bulk through the in-plane migration of Mn4+ ions to form a local TM-O2 structure [30, 31]. In addition, the out-plane migration of Mn4+ ions to the Li vacancy at the Li layer may happen and transform the layered structure to the spinel structure [56]. These structural transformations are irreversible, leading to the capacity/voltage fading [30, 31]. Therefore, in-plane and out-plane migrations of Mn4+ ions to Li vacancies are investigated by a climbing image nudged elastic band (CI–NEB) algorithm.

Results and discussion

Structure and energy evolutions of A-LFNMO and D-LFNMO during de-lithiation

Total energies of various possible de-lithiation structures for A-LFNMO and D-LFNMO are calculated as a criterion, see Fig. 1(c) and (d). Several most stable de-lithiation structures as examples are shown in Fig. 1(e) and (f). Calculations show that the Li-rich removal at the TM layer is often accompanied by the formation of LiO4 tetrahedrons such as the structures with two tetrahedrons at x = 8, 16, and 24. The structural evolutions during the de-lithiation often are related with structural instability and the irreversibility of oxygen redox. Therefore, herein, the unit cell parameters, the O–O layer spacing (d) between the TMO2 layers, the volume change before and after de-lithiation (ΔV), and Ni–O, Fe–O, and O–O bond lengths as the de-lithiation amount x in Li29-xFe3Ni3Mn13O48 are given, see Table 1 and Fig. 2.

Table 1 The cell parameters, O–O layer spacing (d), volume change (ΔV), and average de-lithiation potential (Vave) of A-LFNMO and D-LFNMO (values in brackets are calculations of D-LFNMO)
Full size table
Fig. 2
figure 2

Variations of the Ni–O, Fe–O, and O–O bond lengths in A-LFNMO (a) and D-LFNMO (b). Oxygen release enthalpy ΔHO (c) and average de-lithiation potential Vave (d) of A-LFNMO and D-LFNMO

Full size image

The evolution rules of the unit cell parameters and the O–O layer spacing d of A-LFNMO are similar to D-LFNMO. With the increase of the de-lithiation amount x, the lattice constant a remains invariable, the lattice constant b slowly decreases, the lattice constant c gradually increases, the angle β first decreases and then increases, and the O–O layer spacing d shows a whole increase trend due to the coulomb repulsion of oxygen ions between two TMO2 layers, as shown in Table 1. These results are similar to our previous research finding on other Li-rich Li-Fe–Ni-Mn–O composition [55]. Regardless of D-LFNMO and A-LFNMO, the ΔV values are small and the largest value is only 0.88%. The ΔV values of D-LFNMO are larger than those of A-LFNMO due to larger d between the TMO2 layers, see Table 1. The d changes are related to oxidization characteristic of lattice oxygen ions. More interfacial lattice O2− ions in D-LFNMO participate into the oxidization (more interfacial O2− ions in D-LFNMO [13], see the part on Oxidation Process and Mechanism); thus, the oxidization degree of every oxidized O2− ions in D-LFNMO is slighter than that in A-LFNMO. Consequently, the Coulomb repulsion in D-LFNMO is larger than that in A-LFNMO, leading to larger d between the TMO2 layers.

All the Ni–O, Fe–O and O–O bond lengths of every de-lithiation structures for A-LFNMO and D-LFNMO are listed. Their changes are related to the oxidation of Ni2+, Fe3+, and O2−. For instance, when one or two among three Ni are oxidized, there are various Ni–O bond lengths, including shortened bonds due to Ni2+ ions toward Ni3+ or Ni4+ ions and elongated bonds to maintain the structures, well reflected in Fig. 2. The distribution range of Ni–O bond lengths first becomes wider and last converges to the shorter bonds till all Ni2+ ions are oxidized to Ni4+ at x = 6 for A-LFNMO and at x = 8 for D-LFNMO. Fe–O bond lengths become short due to charge compensation of Fe3+ ions after finishing oxidization of Ni2+ ions, and the distribution range is wide. Last, shortened O–O bond lengths appear due to the oxidization of oxygen ions, and shortened O–O bonds are concentrated on A-LFNMO, which is explained by their different oxidation characteristic of oxygen ions. This would maybe bring out their different stability of lattice oxygen.

To estimate the stability of lattice oxygen, O-release enthalpy ΔHO with the de-lithiation amount x in Li29-xFe3Ni3Mn13O48 is calculated, shown in Fig. 2(c). The ΔHO in D-LFNMO is higher than that in A-LFNMO when the oxidization of lattice oxygen starts at x > 9 in Li29-xFe3Ni3Mn13O48. The higher ΔHO means that the release of the lattice oxygen ions is suppressed, which would maybe improve the redox reversibility of the lattice oxygen ions in D-LFNMO, due to the different oxidation characteristic of oxygen with A-LFNMO. More interfacial O2− ions in D-LFNMO participate in the oxidation (the detailed sees the next part and Sun’s work [13]), that is, the dispersed Li2MnO3-like domains would provide the better reversibility of the oxygen redox. In addition, the oxidation process is divided into two parts, one is the oxidation of Ni2+ and Fe3+, corresponding to redox pairs Ni2+/Ni4+ and Fe3+/Fe4+(x = 0–12 for A-LFNMO and x = 0–10 for D-LFNMO), and another is the oxidation of lattice oxygen (x ≥ 8 for A-LFNMO and x ≥ 10 for D-LFNMO), shown in Fig. 2(d) and Table 1.

Oxidation process and mechanism of A-LFNMO and D-LFNMO during de-lithiation

Densities of states (DOS) and partial DOS (PDOS), which directly reflect the electronic structure changes during de-lithiation, are calculated to study the evolutions of electronic structures of D-LFNMO and A-LFNMO as the de-lithiation, see Fig. 3. For Ni, at x = 0, the spin-up and spin-down states of t2g and the spin-up state of eg* occupy electrons, and the spin-down state of eg* is empty, indicating that the chemical valence of Ni is + 2, that is, Ni2+ (t2g)6(eg*)2. When de-lithiation starts, the spin-up dz2 for Ni loses electrons, that is, Ni2+ ions are oxidized to Ni3+. As the de-lithiation goes forward, more spin-up dz2 loses electrons until the eg* state of Ni2+ ions become empty, namely all Ni2+ ions are oxidized to Ni4+ ions. DOS for Fe and O does not pass the Fermi level, that is, they are not oxidized. For Fe, at x = 0, all the spin-up states of t2g and eg* are occupied, and all the spin-down states are empty, indicating that Fe is + 3 valence with the (t2g)3(eg*)2 configuration. During the de-lithiation, more dx2-y2 of Fe becomes empty, indicating that Fe3+ ions are oxidized to + 4. Total state of lattice oxygen ions in D–LFNMO to exceed the Fermi level is later than that in A–LFNMO, and D–LFNMO happens at x = 10 and A–LFNMO happens at x = 8. This is confirmed by different initial oxidization potential Vave of O in A–LFNMO and D–LFNMO, see Fig. 2(d) and Table 1.

Fig. 3
figure 3

DOS evolutions of Ni, Fe, and total O for A-LFNMO (a) and D-LFNMO (b) and the compared DOS for A-LFNMO and D-LFNMO (c) of O1 in the linear Fe–O-Li configuration, O2 in the linear Li–O-Li configuration, their sum and total O as the de-lithiation amount x in Li29-xFe3Ni3Mn13O48. Fermi level is set to 0 eV

Full size image

In all, A–LFNMO and D–LFNMO have a similar charge compensation process during the de-lithiation process: Ni2+ is first oxidized to Ni4+, then Fe3+ is oxidized to Fe4+, last lattice oxygen participates into charge compensation, see Fig. 3(a) and (b). These results are similar to our previous research finding on other Li-rich Li-Fe–Ni-Mn–O composition [55]. As x > 9, oxygen begins to participate in charge compensation due to their states pass the Fermi level, see Fig. 3(a) and (b). At the energy range of 0–1.0 eV corresponding to the oxidization of lattice oxygen, we find that the lattice oxygen ions participated into charge compensation are O1 in the linear Fe–O-Li configuration, O2 in a linear Li–O-Li configuration and other O, samples at x = 10,14, and 18 as examples are presented, see Fig. 3(c). The DOS for other de-lithiation amount has a similar rule. O in OTMFeLi4 (TM = Mn, Ni), not only in linear Fe–O-Li configuration but also in a linear Li–O-Li configuration, is named O2, because the 2p orbital of these oxidized O2− ions is along a linear Li–O-Li direction, the unpaired O2p electron orbital direction with higher energy level, see Fig. 3(c). The amount of O1 in the linear Fe–O-Li configuration for A–LFNMO and D–LFNMO is the same, so they have the same contribution to the charge compensation in A–LFNMO and D–LFNMO if they have the same coordination environment. In fact, more OTM2FeLi3(TM = Mn, Ni, Fe) coordination configurations in A–LFNMO (Fig. 1(a)) make more O1 participate into charge compensation, see Fig. 3(c). Based on Sun’s work that dispersed and aggregated Li2MnO3-like domains interfacial O between Li2MnO3-like and LiTMO2 domains participated into the charge compensation [13], it is reasonably believed that the other lattice oxygen ions oxidized are interfacial O between Li2MnO3-like and LiTMO2 domains. It is seen from Fig. 3(c) that D–LFNMO has more interfacial O between Li2MnO3-like and LiTMO2 domains and less O1 in the linear Fe–O-Li configuration. Therefore, it can be concluded that the dispersion of Li2MnO3-like domains is more important to trigger more lattice oxygen ions to participate into charge compensation and the doping of Fe has a certain function.

Migration of Mn4+ ions of A-LFNMO and D-LFNMO during de-lithiation

Excessive oxidation of lattice oxygen ions makes LLOs unstable and brings out the structural transition and sequent capacity/voltage fading during cycling [48,49,50,51]. It has been found that the structural transformations such as Mn4+ ion migration to in-plane (forming the local TM–O2 structure) [30, 31] or out-plane Li–vacancy (forming the spinel structure) are irreversible, affecting the cycle performances of LLOs. Therefore, the migration structures and energies of Mn4+ ions are explored, see Fig. 4. The lattice oxygen ions at x = 16 in A-LFNMO and x = 18 in D-LFNMO start to become unstable, so the migration energy, intermediate state, and transition state structures of Mn4+ ions around the de-lithiation amount x = 16 and x = 18 are considered.

Fig. 4
figure 4

The out-plane (a, b) and in-plane (c, d) migration path diagram of Mn4+ ions and their migration structures and energies of A-LFNMO (e) and D-LFNMO (f) ((a), (c), (e), at x = 18 in A-LFNMO; (b), (d), (f), at x = 20 in D-LFNMO). The out-plane migration paths: S1, S2, S3, S4, and S5; the in-plane migration paths: D1, D2, and D3. In product structures of the in-plane migration, the dotted lines are possible formed O–O dimer by one isolated O0 and its adjacent O1, O2, O3, or O4

Full size image

As examples with x = 18 in A-LFNMO and x = 20 in D-LFNMO are used, it is found that the more stable migration product structures correspond to lower migration energies, shown in Fig. 4(e) and (f). Therefore, the migration paths with the most stable product structures are further investigated to simplify the calculations. Considering the calculation time, the sample at the deep de-lithiation amount x = 20 as an in-plane (DA-LFNMO and DA-LFNMO) and out-plane (SA-LFNMO and SA-LFNMO) migration example for A-LFNMO and D-LFNMO to completely studied. Figure 5 shows that the in-plane migration energies of D-LFNMO are all higher than those of A-LFNMO and migration energies of Mn4+ ions reduce with the de-lithiation amount x. For example, when x = 15, the in-plane migration energy of Mn4+ ions for D-LFNMO is 2.00 eV, larger than that for A-LFNMO (1.23 eV), and the in-plane migration energies of Mn4+ ions for A-LFNMO at x = 15, 18, and 20 are, respectively, 1.23 eV, 0.75 eV, and 0.51 eV. The small value of 0.51 eV indicates that the migration product with the local TM-O2 structure would be easily formed. The out-plane migration energies of Mn4+ ions for A-LFNMO at x = 18 and 20 are, respectively, 1.68 eV and 1.57 eV, respectively larger than those of in-plane migration energies of 0.75 eV and 0.51 eV, showing that the out-plane migration is difficult. In all, the in-plane migration of Mn4+ ions is inclined to happen in A-LFNMO and regardless of A–LFNMO and D–LFNMO; the out-plane migration of Mn4+ ions is difficult because of large migration energies. These results indicate that the dispersed Li2MnO3-like domains suppress the in-plane migration of Mn4+ ions to form the local TM-O2 structure, stabilizing the structures and consequently the improved the reversibility of the oxygen redox and cycle performances of LLOs.

Fig. 5
figure 5

Comparison of various migration energies of A-LFNMO and D-LFNMO. DA-LFNMO and DD-LFNMO or SA-LFNMO and SD-LFNMO are in-plane or out-plane migration of A-LFNMO and D-LFNMO, respectively

Full size image

Conclusion

In conclusion, de-lithiation structures, structural stability, and oxidization mechanism of lattice oxygen ions for Co-free LLOs with dispersed Li2MnO3-like domains (D–LFNMO) and aggregated Li2MnO3-like domains (A–LFNMO) are studied. Besides O2− ions in the linear Li–O-Li configuration, the Fe-doping and especially dispersed Li2MnO3-like domains trigger more lattice oxygen ions, avoiding the peroxidation of lattice oxygen. During deep de-lithiation states, the oxygen release enthalpies of D–LFNMO are higher than those of A-LFNMO. At x = 20, the in-plane migration energy of Mn4+ ions with 0.51 eV for A–LFNMO shows that the irreversible migration product with the local TM-O2 structure would be easily formed, vs. 1.72 eV, the large in-plane migration energy of Mn4+ ions for D–LFNMO. In addition, the large out-plane migration energies of Mn4+ ions (the largest migration energies at x = 20 are 1.41 eV for A–LFNMO and 1.93 eV for D–LFNMO) show that the out-plane migrations of Mn4+ ions would be difficult. In all, the in-plane migration of Mn4+ ions is inclined to happen in A-LFNMO and the out-plane migration of Mn4+ ions is difficult, regardless of A–LFNMO and D–LFNMO. D-LFNMO with the dispersion of Li2MnO3-like domains would stabilize the structures and improve cycle performance by suppressing the migration of Mn4+ ions and the oxygen release. This work would provide new insights and feasible strategies for developing high-performance Li-rich cathodes for advanced LIBs.