Theoretical insights on alleviating lattice-oxygen evolution by sulfur substitution in Li1.2Ni0.6Mn0.2O2 cathode material

Here, we demonstrate that the lattice oxygen release on the high-capacity cathode, Li1.2Ni0.6Mn0.2O2 (LNMO) surface can be successfully suppressed through S-anion-substitution using density functional theory (DFT) calculations and ab initio molecular dynamics (AIMD) simulations. The oxygen evolution mechanisms on pristine and sulfur (S)-substituted LNMO (003) surfaces in the presence of an electrolyte mixture are compared. Over-oxidation of O2− anions during delithiation in the pristine surface results in oxygen evolution and subsequent structural deformation. Whereas, in the S-substituted LNMO, S2− anions primarily participate in charge compensation and further inhibit oxygen evolution and O vacancy formation at high degrees of delithiation. Furthermore, the S-substitution effectively prevents the formation of Ni3+ ions and Jahn-Teller distortion, retaining the layered structure during delithiation. Our findings provide insight into improving the structural stability of the LNMO (003) surface, paving the way for developing Li-rich LNMO cathode materials for next-generation LIBs.


INTRODUCTION
In the past few decades, high-energy-density lithium-ion batteries (LIBs) have been extensively investigated and widely used as a major power source owing to the emerging market for electronic devices and electric vehicles [1][2][3][4][5] . However, increasing the energy density of commercial LIBs will be necessary to meet the growing demand for large-scale energy storage devices. Therefore, the design and development of innovative cathode materials with high specific capacities and redox potential becomes of utmost importance 6,7 . Recently, Li-rich layered transition-metal (TM) oxides, including Li 1.2 Ni 0.2 Mn 0.8 O 2 (LNMO), have attracted considerable attention due to their large theoretical specific capacity (up to 300 mA h g −1 ) 2, [8][9][10] . Furthermore, oxygen vacancies in LNMO cathode materials are considered beneficial defects since they can regulate the electrochemical properties and electrical conductivity of cathode materials [11][12][13][14] . Tang et al. have shown that the formation of oxygen vacancies in LNMO cathode materials can have a significant impact on LIBs performance, such as charge transfer resistance, specific capacity, and the rate of charging and discharging 11 . However, an excess of oxygen vacancies in LNMO materials resulted in the structural transformation layered to spinel-like or rock-salt phases, which leads to low efficiency and fast-fading capacity as well as fading voltage in LIBs 1,5,[15][16][17][18][19][20] . In addition, lattice-oxygen evolution is strongly associated with oxygen vacancy formation in these LNMO materials during delithiation 11,21 . The over-oxidation of some O 2− anions during delithiation is widely assumed to be the primary cause of the formation of oxygen vacancies and lattice-oxygen evolution in cathode materials [22][23][24] .
Understanding the triggering factors for the oxygen evolution mechanism and circumventing capacity fading and structural degradation limitations in LNMO cathode materials are therefore essential for developing better cathode materials in LIBs. Tarascon et al. have shown that under-coordinated surface oxygen prefers over-oxidation during charging, resulting in the formation of O 2 molecules and peroxide-like (O 2 ) n− species 25 . They also demonstrated that a particular environment of metal-oxygen hybridization in LNMO cathode materials has a significant impact on oxygen evolution. Ceder et al. identified that the O 2− p-orbital along with the Li-O-Li direction generated a specific non-bonding band with a higher energy level, increasing the O redox activity and facilitating oxygen evolution of LNMO cathode materials 26 . To overcome this limitation, element doping or inert-layer coating are commonly used to suppress oxygen evolution or alleviate structural deformation during cycling. Zhang et al. reported that the inherent stability of the structure can be ameliorated, and lattice-oxygen release can be suppressed by fabricating a spinel Li 4 Mn 5 O 12 coating on a Li-rich layered cathode material 27 . In addition, surface doping is a promising modification method that can improve the structural stability of LNMO cathode materials and suppress irreversible oxygen evolution on the surface [28][29][30] . Various heteroatom dopants can affect the structural stability and electronic properties of cathode materials in different ways. For instance, our previous study demonstrated that doping the Ni site of the Li-rich LNMO material with Cd ions can effectively inhibit the structural change and suppress the Jahn-Teller distortion during the delithiation process 31 . Similarly, Nayak et al. demonstrated that the addition of Al 3+ in the LNMO cathode material has a surface stabilization effect, improving structural stability by alleviating oxygen evolution 32 . In addition to cationic substitutions, Jiang et al. have shown that sulfurdoped LiMn 2 O 4 exhibits improved cycling stability and a higher capacity during cycling 33 . Additionally, sulfur doping into MnO 2 nanosheets is also being investigated as a high-performance cathode for Zn-ion batteries due to the addition of Zn storage sites that exhibit pseudocapacitive properties 34 . Along with the effects of sulfur doping on these cathode materials, Zhang et al. also demonstrated that S-substitution in the O site of LNMO could stabilize and improve the cycling stability and rate performance (a capacity of 117 mA h g −1 is maintained) 35 .
However, a fundamental understanding of the formation of O vacancy, lattice-oxygen-evolution mechanism, and structural deformation during delithiation in S-substituted LNMO cathode materials have not been thoroughly examined, particularly the role of S-substitution at the O site. We recently reported the oxygen oxidation mechanisms in the LNMO (003) surface by considering different local oxygen-coordination environments and investigating their oxidation process during various degrees of delithiation 36 . We observed that the oxygen-dimer formation could be attributed to the over-oxidation of the O 2− anions by forming an irreversible superoxide O 2 − at a high degree of delithiation. However, the lattice-oxygen evolution and structural stability of the LNMO (003) surface in the presence of an electrolyte mixture were not considered in our previous study. Thus, in this work, we used DFT and AIMD simulations to investigate the oxygen evolution mechanism and the oxidation state of TM ions and O 2− anions in both pristine and S-substituted LNMO surfaces with electrolytes, with a focus on the role of sulfur substitution in lattice-oxygen formation and structural degradation.

RESULTS AND DISCUSSION
Oxygen evolution mechanism and electronic characteristics during delithiation To further investigate the oxygen evolution mechanism in LNMO (003) surfaces during delithiation, we have compared each delithiated structure and the oxidation state of O 2− anions and TM metals using AIMD simulations. During the delithiation process, we considered different delithiation states from x = 0 to 0.8 in the pristine Li 1.2 − x Ni 0.2 Mn 0.6 O 2 (003) surfaces. A step-bystep delithiation process was performed on the Li 1.2 − x Ni 0.2 Mn 0.6 O 2 (003) surface by removing four lithium atoms at a time. The simulated pristine model before delithiation is illustrated in Fig. 1a, and the optimized geometries of electrolyte components are shown in Fig. 1b. After the removal of Li ions from the LNMO (003) surface, the system was equilibrated for 5 ps, during which it demonstrated a structural change and reached a new equilibrium. In addition, the energy was determined by calculating the lowest total energy over the final half of the trajectory. The intercalation potential (V) vs. Li Þare the calculated total energies of the system before and after Li-ion  . The calculated Bader charge shows the formation of oxygen via a sudden increase in the atomic charge (larger than −0.4|e|), which can serve as a baseline for oxygen evolution. Furthermore, we observed the second oxygen formation at the delithiation level x = 0.67, and the value of the atomic charge also exceeds the baseline of oxygen evolution from −0.9 to −0.3|e| (green line in Fig. 3a).
To further understand the detailed mechanism of oxygen evolution in the LNMO (003) surface, we calculated the change in atomic charge (ΔQ) for the O anions and TM ions (Ni and Mn) in the LNMO (003) surface at various delithiation levels, and the results are plotted in Fig. 3b. A positive value of ΔQ indicates that the atom loses electrons and increases its oxidation state. The ΔQ value of Mn atoms shows a slight increase and then remains almost constant during the entire delithiation process, indicating that Mn atoms are not involved in the charge compensation process. The ΔQ values of Ni atoms undergo a relatively significant increase during the delithiation process compared to those of Mn atoms, and it demonstrates that the Ni atoms are primarily involved in the oxidation process, which is consistent with our previous and other studies 22,23,31 . The value of ΔQ for O anions significantly increases, implying that the oxidation activity of O anions in LNMO (003) surface is more than those of Ni 2+ and Mn 4+ ions. The charge compensation process in the LNMO (003) surface during the whole delithiation process is dominated by the O 2− / O 2 n− couple, as observed in Fig. 3a, b.

Effects of S-substitution on oxygen evolution
Oxygen release in Li-rich LNMO cathode materials usually causes undesired structural degradation, resulting in the rapid fading of capacity/voltage in LIBs. Therefore, proposing a strategic method for suppressing oxygen release was a challenge in this study.
Considering the similar electronic configurations of S and O, we employed S −2 substitution in the LNMO (003) surface by replacing the O −2 anions. The simulated S-substituted LNMO before delithiation is illustrated in Fig. 1a. As a further investigation into the effects of S-substitution on oxygen evolution during delithiation, we calculated the intercalation potential of the S-substituted LNMO cathode material as a function of the delithiation state, as shown in Fig. 1c. The calculated voltage value for the S-substituted LNMO (003) surface is~3.5 V at the initial delithiation level, which is lower than the value for the pristine system (4.0 V). We also illustrated the optimized configurations of electrolytes on the S-substituted LNMO (003) surface from low to high degrees of delithiation in Fig. 4, which demonstrates that sulfite-like species (SO 2 ) n− are generated on the S-substituted LNMO (003) surface at delithiation levels of x = 0.4 and 0.8. At delithiation level x = 0.53, the formation of (SO 3 ) m− species on the S-substituted LNMO surface is observed, but there is no oxygen evolution on the S-substituted surface. The formation of (SO 2 ) n− /(SO 3 ) m− indicates a shift in the redox center from the more electronegative O atoms to the less electronegative S atoms after the sulfur substitution, which plays an important role in charge compensation during delithiation. Although S-substitution lowers the electrode potential, our structural analysis of the S-substituted LNMO (003) surface at different delithiation levels shows that the substitution of S anion could effectively suppress oxygen evolution.
To compare the oxidation state of O atoms in the pristine and S-substituted LNMO (003) surfaces, we calculated the Bader charge corresponding to the O atoms in the S-substituted system at various delithiation levels and are illustrated in Fig. 3c. The average atomic charge of the O atoms in the S-substituted LNMO (003) surface gradually increased during delithiation, as we noticed in the pristine LNMO surface (Fig. 3a). However, all the O atomic charge in the S-substituted system is below the baseline of −0.4|e|, suggesting that there is no over-oxidation of lattice oxygen anions at the S-substituted LNMO surface. This result indicates that S-substitution effectively suppresses the redox activity of O atoms and inhibits the over-oxidation of O atoms; therefore, no O 2 molecule formation occurred even at a higher degree of delithiation. Furthermore, we analyzed the change in the atomic charge (ΔQ) of anions and TMs on the S-substituted LNMO surface to confirm the suppression of oxygen  Table 1, where positive values represent electron loss. Notably, S anions exhibit significant electron loss upon delithiation, confirming that the redox center is transferred from O atoms to S atoms. On the other hand, the ΔQ value for the O anions in the S-substituted system is changed to 0.064|e|/atom during delithiation, which is lower than 0.46|e|/atom in the pristine system (Fig. 3d). The large ΔQ for S anions is due to the formation of sulfite-like (SO 2 ) n− /(SO 3 ) m− species on the S-substituted LNMO (003) surface, as shown in Fig. 4. Furthermore, the huge oxidation of S anions can be considered as an electron buffer; consequently, S anions can participate in charge compensation to effectively prevent lattice-oxygen evolution. This result supports that the presence of S anions can significantly inhibit oxygen oxidation in LNMO cathode materials.
We further investigated the redox activity of TM atoms in the S-substituted LNMO surface by calculating the change in atomic charge (ΔQ) and electronic properties during the delithiation process, and the results are given in Table 1. The ΔQ values of Ni atoms are changed significantly after delithiation compared to those of Mn atoms, indicating that the Ni atoms in the S-substituted LNMO surface are primarily involved in the oxidation process, as like pristine surface. To further explore the changes in the electronic properties of Ni upon extraction of Li atoms in the pristine and S-substituted LNMO surfaces, we projected the density of states (PDOS) of the 3d-orbital of Ni and Mn atoms in the first and second TM layers at delithiation levels of x = 0 and 0.4, as shown in Fig. 5. In the pristine system, the Ni d states are shifted from below the Fermi level to above the Fermi level at the delithiation level x = 0.4 compared to the initial state (x = 0), indicating that Ni atoms lose electrons from the Ni d-orbital (Fig. 5a, b). Moreover, we integrated the areas under the PDOS peaks above the Fermi level at the delithiation levels of x = 0 and x = 0.4, which are 16.1 and 17.0, respectively. The increased oxidation of the Ni atoms at x = 0.4 is supported by the increase in the area above the Fermi level during delithiation.
The PDOS of the S-substituted LNMO surface (Fig. 5c) shows that the Ni valance band is significantly up-shifted and close to the Fermi level at x = 0 compared to the pristine system (Fig. 5a), suggesting that Ni oxidation is more favorable in the S-substituted system than in the pristine system. As shown in Fig. 5d, the 3d states of Ni atoms in the S-substituted system are shifted more from the valence band to the conduction band after delithiation (x = 0.4) than in the pristine system (Fig. 5b). Furthermore, the integral areas above the Fermi level for the 3d states of Ni atoms in the S-substituted LNMO (003) surface at delithiation levels of x = 0 and x = 0.4 are 16.2 and 17.4, respectively, demonstrating that S-substitution can improve Ni atoms' ability to lose electrons compared to the pristine system. In general, increasing the oxidation of Ni atoms can improve capacity retention significantly 2,29 . Apart from the capacity retention, the oxidation change of Ni atoms during delithiation would also influence the stability

Effects of S-substitution on phase transition
The irreversible phase transition from the layered structure to a new phase caused by oxygen vacancies and TM migration during cycling affects the structural stability of LNMO cathodes, resulting in voltage decay and capacity fading 15 . To understand this, we explored the detailed configurations of each TM ion during the delithiation of both pristine and S-substituted LNMO (003) surfaces. The number of octahedral and tetrahedral sites for Mn and Ni atoms in the first layer of the pristine and S-substituted LNMO (003) surfaces are listed in Table 3. Figure 7 illustrates the structural deformation before and after delithiation in the pristine and S-substituted Li 1.2 − x Ni 0.2 Mn 0.6 O 2 (003) surfaces. We observed that structural deformation occurred at the delithiation level of x = 0.4 in the pristine LNMO surface, owing to the increased amounts of Mn and Ni atoms from the octahedral to tetrahedral configuration. For the pristine system, the MnO 6 and NiO 6 octahedra are changed to MnO 4 and NiO 4 tetrahedra, with surface oxygen formation from the Mn-O environment at the delithiation level of x = 0.4. Furthermore, structural deformation in the S-substituted surface can be observed when the delithiation state reaches x = 0.67, and the extent of this structural change is less than that of the pristine surface at a higher degree of delithiation, as shown in Table 2. This result suggests that the structural change of S-substituted LNMO (003) is insignificant at low degrees of delithiation state. In addition, the S-substituted LNMO (003) maintains a layered structure at the delithiation level of x = 0.4 (Fig. 7), suggesting that S-substitution can effectively prevent structural deformation during the charging process. Thus, we can conclude that S-substitution in the Li 1.2 − x Ni 0.2 Mn 0.6 O 2 cathode material plays a significant role in the electron buffer to participate in charge compensation, which further circumvents lattice-oxygen formation and release and effectively reduces structural deformation during the delithiation process.
In conclusion, we systematically investigated and compared the O vacancy formation and the mechanism of lattice-oxygen evolution in the pristine and S-substituted LNMO (003) surfaces in the presence of electrolyte mixture using DFT and AIMD simulation methods. The structural and electronic properties of the pristine LNMO (003) surface are analyzed during delithiation, and some overoxidized O anions are observed at a high level of delithiation, and these O anions undergo oxygen evolution, causing O vacancy formation. We found that the O vacancy formation on the pristine LNMO (003) surface is thermodynamically favorable and forms spontaneously at a high degree of delithiation. The favorable formation of O vacancy implies that oxygen evolution and release  could easily occur on the pristine surface. However, the oxygen formation and O vacancy on the LNMO surface can be significantly suppressed by S-substitution at the O site because of the formation of (SO 2 ) n− /(SO 3 ) m− species associated with the electron buffer behavior; these species primarily participate in charge compensation and further preserve the reversibility of the O 2− /O 2 n− redox reaction at a higher level of delithiation. Additionally, the formation of (SO 2 ) n− /(SO 3 ) m− species also creates appropriate O vacancies on the S-substituted surface. It is noticed that appropriate O vacancies can reduce charge transfer resistance and increase the capacity and the rate of charging and discharging of the LNMO cathode material 11 . Moreover, (SO 2 ) n− /(SO 3 ) m− species inhibit the excess formation of O vacancies resulting from the oxygen evolution, thereby preventing severe structural deformation at a high level of delithiation. Besides, S-substitution can effectively prevent the formation of Ni 3+ ions, inhibiting the Jahn-Teller distortion during delithiation. According to the delithiated structures, we observed that the structural deformation of S-substituted LNMO (003) during delithiation is negligible, and the layered structure is maintained during cycling. Our results contribute to the understanding of the oxygen evolution mechanism in LNMO cathode materials, with a particular emphasis on a strategy for suppressing lattice-oxygen formation for developing high-performance cathode materials with high reversibility and high structural stability for LIBs.

METHODS
The DFT and AIMD calculations were performed using the projector augmented wave (PAW) method implemented in the Vienna Ab initio Simulation Package (VASP) [40][41][42] . The generalizedgradient approximation (GGA) and the long-range dispersion correction incorporated in the optB88-vdW functional was adopted to describe the van der Waals (vdW) interactions 43 . The Hubbard U corrections of 5.96 and 5.10 eV were considered to include the self-interaction energy of the correlated d-electrons of Ni and Mn atoms in the system, respectively 44,45 . Similar to our previous study 36 , we considered bulk LNMO unit cell with the rhombohedral symmetry (R3m space group). All the geometrical parameters are relaxed until the total energy is converged to 10 −4 eV and the Hellmann−Feynman forces are smaller than 0.01 eV/Å. The optimized structure of bulk LNMO unit cell and simulated XRD pattern are shown in Supplementary Fig. 1. In comparison to previous experimental XRD patterns, both our simulated and experimental results demonstrate the (003) orientation as the preferred one 31,36,46 . Furthermore, we considered different terminations, such as O-terminated, TM-terminated, and Li-terminated surfaces. Supplementary Fig. 2 illustrates the optimized structures and total energy of each termination surface. Accordingly, the Literminated LNMO (003) surface shows the lowest total energy and was constructed for the following calculations.
To study the doping effects on oxygen evolution and structural stability during delithiation, we considered sulfide anion (S −2 ) doping into the first layer of the O site in the LNMO (003) surface. A suitable site for sulfide anion substitution was determined by calculating the formation energies (E f ) of S −2 anions substituted at different O sites of the LNMO (003) surface, which is defined as Eq. (2)  To study the oxygen evolution in the LNMO (003) surface, we constructed a cathode/electrolyte interface to study the influences of electrolytes on the structural stability of Li 1.2 − x Ni 0.2 Mn 0.6 O 2 cathode materials in different delithiation states using AIMD simulations. Organic electrolytes often consist of mixtures of multiple solvents and different lithium salts. In this study, we considered Li salt (LiPF 6 ) and the solvent mixture (ethylene carbonate (EC) and diethyl carbonate (DEC)) and optimized these molecules at the B3LYP/6-311 + G (d, p) level of theory using the Gaussian 09 package [48][49][50][51] . The p(4 × 1) supercell of the LNMO (003) surface is generated to performed AIMD simulations for 1 M LiPF 6 in an EC/DEC mixture (3:7 v/v%). The numbers of EC and DEC molecules in the simulation cell were chosen as 4 and 6, corresponding to densities of 1.32 and 0.975 g cm −3 for EC and DEC, respectively. The Brillouin zone was sampled using a 1 × 1 × 1 Gamma k-point mesh, and the cut-off energy of the plane-wave basis expansion was set to 500 eV. The convergence criteria for electronic self-consistent iteration were set to 10 −4 . Besides, we considered one layer of He atoms and fixed their positions to prevent the electrolyte interactions between the neighboring slabs due to the periodic boundary conditions in the AIMD simulations. To simulate the delithiation process, a step-by-step delithiation process was performed on the Li 1.2 − x Ni 0.2 Mn 0.6 O 2 (003) surface by removing four lithium atoms at a time. After removing four lithium atoms, the AIMD simulations were equilibrated at 400 K in a canonical ensemble (NVT) for a total time of 5 ps with a time interval of 1 fs. Then, we optimized the LNMO structures with electrolytes according to the lowest total energy over the final half of the trajectory for further electronic analyses such as Bader charge and projected densities of states (PDOS) 52,53 . Figure 8 shows an overview of the AIMD simulation procedure carried out in this study.

DATA AVAILABILITY
The authors declare that the main data supporting the findings of this study are contained within the paper and its associated Supplementary Information. All other relevant data are available from the corresponding author upon reasonable request.

CODE AVAILABILITY
All density functional theory calculations in this work were performed using the VASP code (version 5.4.4), which is a licensed software package.