Redox Engineering of Fe‐Rich Disordered Rock‐Salt Li‐Ion Cathode Materials

The pursuit of high‐performance and cost‐effective Li‐ion batteries emphasizes the need for cathode materials composed of abundant elements, such as Fe. Disordered rock‐salt (DRX) cathode materials, known for their high compositional flexibility, offer a unique opportunity in this regard. However, Fe‐rich DRX (Fe‐DRX) cathodes, potentially the most cost‐effective among all DRXs, have seen limited research interest due to their comparatively restrained performance. This limitation stems from the inaccessibility of the Fe3+/Fe4+ redox in the DRX structure, prompting the need for redox engineering to enable Fe‐DRXs with readily utilizable redox mechanisms. In this work, utilizing both experiments and theoretical study, reversible Fe2+/Fe3+ redox in an Fe2+‐based DRX cathode is demonstrated. This design minimizes the reliance on O redox, resulting in a high capacity (≈290 mAh g−1) and energy density (≈700 Wh kg−1), as opposed to an Fe3+‐based DRX operating on the limited Fe3+/Fe4+ redox and extensive O redox upon cycling. Overall, the study introduces a novel approach to redox engineering to develop low‐cost, high‐performing Fe‐rich cathode materials.


Introduction
The global shift toward electric vehicles and large-scale energy storage demands cost-effective alternatives to commercial Ni/Co-based layered cathodes (e.g., LiNi 0.6 Mn 0.2 Co 0.2 O 2 ) DOI: 10.1002/aenm.202400402 in Li-ion batteries (LIBs). [1][4] These alternatives hold promise for enhancing the affordability of LIBs, leveraging iron's status as the most affordable and abundant transition metal (TM). [5]While the demand for iron-based cathode chemistries (namely LiFePO 4 ) has increased in recent years and is forecast to reach almost an equal global share as Ni-based chemistries by 2035, [6] a current challenge is the lower energy density of Fe-rich cathodes compared to Ni/Co-based layered cathodes. This limitation hinders their application in LIBs, particularly in scenarios where energy density (Wh/kg; Wh/l) is crucial.
The recent discovery of the disordered rock-salt (DRX) cathode's ability to accommodate a variety of TMs and anions presents a significant opportunity for identifying LIB cathodes with high energy density (≈900 Wh kg −1 ) made using abundant TMs, including Fe. [7][8][9][10][11][12][13] Compared to LiFePO 4 , Ferich DRX also presents an opportunity for supply chain resilience by reducing reliance on rock phosphate reserves, most of which are concentrated in a single jurisdiction. [14][17][18] For instance, Li 1.21 Fe 3+ 0.37 Ti 0.42 O 2 delivers only ≈138 mAh g −1 (≈374 Wh kg −1 ) at a very slow rate of ≈2 mA g −1 , with its voltage profile containing a significant voltage hysteresis. [19]Fe-DRXs have been synthesized as Fe 3+ -containing compounds (Fe 3+ -DRX), and their limited performance is primarily attributed to the prohibitively high Fe 3+ /Fe 4+ redox potential, resulting in a substantial overlap of the Fe 3d band with the O 2p band and causing overcompensation of O redox capacities (Figure 1a). [20]23]  One approach to the design of DRX involves intentionally minimizing reliance on O redox and maximizing TM redox.In the case of Fe 3+ -DRXs, reducing the Fe 3+ /Fe 4+ redox potential proves beneficial in enhancing the Fe-redox capacity before resorting to O-redox.This concept was recently demonstrated in Li 2 Fe 3+ 0.5 Mn 3+ 0.5 O 2 F, where the observed Fe 3+ /Fe 4+ redox potential decreased when combined with Mn 3+ /Mn 4+ . [24]owever, further studies are required to determine the feasibility of applying a similar strategy to Fe 3+ -rich DRXs.Unlike Li 2 Fe 3+ 0.5 Mn 3+ 0.5 O 2 F, where significant Mn-redox capacities (known to be reversible) and a diluted amount of Fe can mask the inherent instability of the Fe 3+ /Fe 4+ redox process in DRXs, Fe-rich DRXs primarily rely on Fe-redox, making it essential to explore alternative stabilization methods.
Given the challenges associated with Fe 3+ /Fe 4+ redox in Fe 3+ -DRXs, there is a potential advantage in designing an Fe 2+based DRX (Fe 2+ -DRX) with Fe 2+ /Fe 3+ redox occurring before Fe 3+ /Fe 4+ or O redox, achieved through partial F − substitution for O 2− .This strategy could prevent overcompensation of O redox upon cycling by introducing ample Fe 2+ /Fe 3+ capacity before the onset of Fe 3+ /Fe 4+ or O redox processes, similar to how the addition of Mn 2+ /Mn 3+ redox before Mn 3+ /Mn 4+ and O redox could increase the Mn-redox capacity in Mn-based DRXs (Mn-DRX) to mitigate the O-redox-related side reactions. [16,25] In this study, we investigate the accessibility of Fe 2+ /Fe 3+ in Fe 2+ -DRX and evaluate its performance and redox mechanisms compared to Fe 3+ -DRX.Our examination includes Li  DRX) for the first time.Notably, LFNOF achieves ≈290 mAh g −1 and ≈700 Wh kg −1 , marking one of the highest capacities and energy densities among any Fe-DRX or other Fe-rich cathode materials reported to date.X-ray Photoelectron Spectroscopy (XPS), Electron Paramagnetic Resonance (EPR) spectroscopy, X-ray diffraction (XRD), and Density-Function Theory (DFT) calculations confirm that Fe 2+ /Fe 3+ redox is readily accessible, which delays and reduces the participation of O redox in LFNOF, leading to more stable cycling performance than LFNO.Overall, our study highlights the importance of redox engineering in developing Fe-rich Li-ion cathode materials with high energy density.

Synthesis and Characterization of Fe-Rich DRX Compounds
We synthesized LFNO (Fe 3+ /Fe 4+ redox available with O redox, theoretical Fe 3+ /Fe 4+ capacity = 174 mAh g −1 ) and LFNOF (Fe 2+ /Fe 3+ redox additionally available, theoretical Fe 2+ /Fe 3+ /Fe 4+ capacity = 341 mAh g −1 ) using mechanochemical synthesis.Detailed information can be found in the Methods section.Nb exists in the Nb 5+ state in these compounds, minimally participating in the redox process.Figure 1c,d shows the XRD patterns of the as-synthesized LFNO and LFNOF powders.All peaks can be indexed to a targeted disordered rock-salt structure (space group: Fm3̅ m). [30]The broad diffraction peaks indicate the nanoparticle character of the as-synthesized materials, consistent with the SEM images. [31]Rietveld XRD refinement reveals that LFNO has a lattice parameter of 4.1807 Å while that of LFNOF is 4.2031 Å. Detailed information about the Rietveld refinement of both powder materials can be found in Tables S1 and S2 (Supporting Information).The larger lattice parameter of LFNOF is likely due to LFNOF containing Fe as Fe 2+ , a bigger cation than Fe 3+ in LFNO.XPS on the as-synthesized LFNO and LFNOF powders reveals that Fe in LFNO predominately exists as Fe 3+ or Fe 4+ , whereas that of LFNOF compounds are largely Fe 2+ or Fe 3+ (Figure 1e,f).Thus, overall, the Fe oxidation state is lower for LFNOF than LNFO and is close to Fe 2+ for LFNOF and Fe 3+ for LFNO.Finally, the energy dispersive X-ray spectroscopy (EDS) mapping with the scanning transmission electron microscopy (STEM) performed on the assynthesized LFNO and LFNOF powders suggests a uniform distribution of the elements (Figure S1, Supporting Information).

Figure 2a
,b shows the voltage profiles of LFNO and LFNOF when cycled between 1.3 and 4.8 V at 40 mA g −1 at room temperature.LFNOF has a lower open-circuit voltage of ≈2.2 V than LFNO (≈2.5 V) before cycling, indicating a more reduced chemical state (≈Fe 2+ ) in LFNOF than LFNO (≈Fe 3+ ).LFNO delivers the 1 st discharge capacity and specific energy of 243 mAh g −1 and 599 Wh kg −1 , whereas LFNOF achieves 292 mAh g −1 and 704 Wh kg −1 , respectively.[4] For both materials, the first charging includes a low voltage plateau at ≈2.7 V and a high voltage plateau at ≈4.2 V.The ≈4.2 V plateau is predominant for LFNO, whereas LFNOF exhibits a notable ≈130 mAh g −1 charge capacity at ≈2.7 V before the ≈4.3 V charging.Previous work by Dahn et al. demonstrated that the ≈4.2 V charging plateau in Li 1.2 Fe 0.4 Ti 0.4 O 2 (Fe 3+ -based DRX) primarily originates from O oxidation with partial Fe 3+ /Fe 4+ oxidation. [19]Thus, the prolonged ≈4.2 V charging plateau in LFNO is most likely due to extensive O oxidation with limited Fe 3+ /Fe 4+ oxidation.In contrast, more readily accessible Fe 2+ /Fe 3+ oxidation (prior to Fe 3+ /Fe 4+ and O oxidation) appears to contribute to the notable ≈2.7 V charging capacity observed for LFNOF.We note that the first charging capacity of LFNOF at ≈2.7 V (≈130 mAh g −1 ) is smaller than its theoretical Fe 3+ /Fe 4+ capacity of ≈170 mAh g −1 .This difference may stem from partial Li extraction from LFNOF (resulting in slight oxidation) before cycling, possibly due to factors such as air exposure, despite our efforts to minimize air exposure during synthesis and processing.This pre-Li extraction could also explain why the discharging capacity of LFNOF upon first discharge exceeds its charging capacity, as it allows the discharging process to fill in the pre-formed Li vacancies.
Comparing the 2nd cycle voltage profile of LFNO and LFNOF, a significantly larger voltage hysteresis/polarization is observed in LFNO (Figure 2c).The O-redox process in Li-excess layered or DRX materials is known to contribute to voltage hysteresis due to unequal redox mechanisms during charging and discharging, involving the formation of O dimers (e.g., peroxide, superoxide, trapped O 2 ) at the top of charging.These dimers get reduced and dissociated only after deep discharging to a low voltage. [32,33]In this context, the substantial voltage hysteresis in LFNO suggests a higher involvement of O redox in the material compared to LFNOF.
The dQ/dV plots of LFNO and LFNOF reveal a more pronounced change for LFNO than LFNOF (Figure 2d,e).Notably, the high-voltage charging peaks >4 V (associated with the ≈4.2 V plateau) rapidly disappear after 10 and 20 cycles for LFNO, and LFNO's 1 st discharging dQ/dV peak at ≈2.7 V is lost after cycling.This shift of dQ/dV charging and discharging peaks to lower voltages for LFNO indicates voltage fading, a phenomenon attributed to irreversible O loss from various Li-excess cathode materials. [34,35]Conversely, the changes in the dQ/dV plot are less pronounced for LFNOF, aligning with the slower discharge voltage fading for LFNOF compared to LFNO (Figure 2f): the average discharge voltage decreases by 254 mV for LFNO but only by 57 mV for LFNOF after 10 cycles.
Interestingly, we find that the rate at which the 1st discharge specific energy decreases with a lowered upper cut-off voltage (from 4.8 to 4.4 V at 40 mA g −1 ) is much slower for LFNO than for LFNOF (Figure 3f).The 1st discharge specific energy of LFNO decreases from 599 to 539 Wh kg −1 (≈10% reduction), whereas the decrease is from 704 to 481 Wh kg −1 (≈32% reduction) for LFNOF, with the lowered upper cut-off voltage from 4.8 to 4.4 V.4] LFNO's specific discharge energy being less sensitive to the upper cut-off voltage is primarily due to its average discharge voltage barely changing with the upper cut-off voltage (Figure 3f).In fact, LFNO's average discharge voltage "increases" from 2.47 V to 2.53 V with a "decreasing" charge cut-off voltage from 4.8 V to 4.4 V, contrasting the typical behavior of a cathode material showing a higher discharge voltage with a higher cut-off voltage.This result is likely due to oxygen loss from LFNO upon charging to a high voltage, making the LFNO a more reduced compound (rather than a more oxidized compound) after high-voltage charging, leading to a decrease in the operating voltage rather than an increase.This observation is consistent with its fast voltage fading observed during 1.3-4.8V cycling.In this regard, LFNO shows a clear ≈3.7 V discharge plateau when the charge cut-off is limited to 4.4 V, which was barely seen after a 4.8 V charge, leading to more symmetric charging and discharging voltage profiles.On the other hand, LFNOF shows the typical behavior of operating voltage lowered with decreasing upper charge cut-off voltage.The ≈540 Wh kg −1 achieved by LFNO between 1.3 and 4.4 V and ≈700 Wh kg −1 achieved by LFNOF between 1.3 and 4.8 V suggests an opportunity to design 600-700 Wh kg −1 -class Fe-based DRX cathodes via redox process engineering.
LFNO and LFNOF demonstrate comparable rate capabilities, with LFNOF exhibiting a slight advantage over LFNO (Figure 3g,h).When charged at 20 mA g −1 up to 4.8 V and discharged at different rates of 10, 20, 40, 100, 200, 400, and 1000 mA g −1 to 1.3 V, LFNO's discharge capacity decreases from ≈272 mAh g −1 (at 10 mA g −1 ) to ≈131 mAh g −1 (at 1000 mA g −1 ).Similarly, for LFNOF, the capacity changes from ≈314 mAh g −1 (at 10 mA g −1 ) to ≈152 mAh g −1 (at 1000 mA g −1 ) for LFNOF.The higher capacity achieved by LFNOF is in line with the slightly higher chemical Li diffusivity in LFNOF compared to LFNO during the 1st discharging, as determined from Galvanostatic Intermittent Titration Technique (GITT) tests (Figures 3i; S2 and S3, Supporting Information).However, the estimated Li diffusivity of 10 −16 −10 −15 cm 2 s −1 in LFNO and LFNOF is still quite low.Therefore, the reasonably high capacity of the materials, even at a high rate of 1000 mA g −1 , is likely attributable to the pulverized nanoparticle morphology of the two compounds, minimizing the criticality of long-range Li diffusion for achieving high capacity. [36]

Redox Mechanisms of LFNO and LFNOF
To understand the LFNO's and LFNOF's redox mechanism, we examined the Fe and O oxidation states in the materials at different states-of-charge (SoC) using ex situ XPS (Figure 4a,b).Nb is barely redox active in the compounds (Figure S5, Supporting Information).Additionally, the fraction (%) of Fe and O species at each SoC was estimated through XPS fitting (Figure 4c,d).The LFNO and LFNOF cathode films were cycled at 40 mA g −1 , with an increment of 80 mAh g −1 until the top-of-charge (ToC) at 4.8 V, and then to the end-of-discharge (EoD) at 1.3 V.The XPS measurement was conducted using an air-tight XPS holder to prevent air exposure.Due to the near-random ionic distribution in the DRX structure, various local environments exist for Fe and O ions, leading to a broad distribution of electron energies (and thus Binding Energy, BE, in XPS) for ions with the same nominal charge (e.g., Fe 3+ ).Therefore, the labels on each peak in the XPS Fe 2p spectra should be interpreted with caution.It is important to note that XPS scans over a surface depth of ≈5 nm, which is often considered inadequate for probing bulk redox mechanisms of cathode materials.However, the nanoparticle morphology of LFNO and LFNOF, with a characteristic size of d≈100 nm, implies that the ≈5 nm surface (and consequently the XPS signal) represents a significant portion (≈30%) of the particles' total volume and, by extension, the redox mechanism.
For LFNO, a subtle increase in the average Fe oxidation states is observed from ≈Fe 3+ toward ≈Fe 4+ after charging to 80 mAh g −1 (C80).This is evidenced by the growth of a peak at ≈716 eV (assigned as Fe 4+ ), accompanied by a reduction in the peak at ≈712 eV (assigned as Fe 3+ ).However, a significant portion of Fe remains in the Fe 3+ state (or possibly Fe 2+ ).Further charging to 160 mAh g −1 (C160) and reaching the ToC results in a decrease of the average Fe oxidation state back towards ≈Fe 3+ .This is supported by the increased prominence of the ≈712 peak over the ≈716 eV peak, likely attributed to the reductive coupling of Fe with oxygen during charging. [20]The O 1s XPS spectra reveal that O 2− ions undergo oxidation as early as 80 mAh g −1 charge (C80), and the population of oxidized O species increases upon further charging (Figure 4c; Figure S6, Supporting Information).These findings suggest limited Fe 3+ /Fe 4+ oxidation in LFNO during the 1st charging and O oxidation, which concurrently occurs with Fe oxidation during the early charge but becomes the predominant oxidation process at the later stage of charging for LFNO, possibly in conjunction with Fe─O reductive coupling. [20]Upon discharging (D80, D160, EoD), the Fe 2p 3/2 spectrum shifts to lower BEs, signifying Fe reduction.Simultaneously, the population of oxidized O species decreases, that is, O reduction (Figure 4a,c).Notably, the intensity of the Fe 2+ peak (at ≈710 eV) at the EoD is greater than before cycle (BC), likely due to O loss upon the 1st charging, allowing Fe to be reduced more toward Fe 2+ in discharging compared to BC., that is, LFNO becomes a more reduced compound after the first cycle, as reflected in the Fe 2p 3/2 spectrum appearing at lower BEs after the 1st discharge than BC.
Compared to LFNO, we observe much clearer participation of Fe redox in LFNOF from the Fe 2p 3/2 XPS spectra, mainly involving the Fe 2+ /Fe 3+ redox couple (Figure 4b,d).Upon charging to 80 and 160 mAh g −1 (C80, C160), the peak at ≈710 eV (assigned as Fe 2+ ) gradually decreases in intensity, replaced by growing peaks at ≈712 eV (assigned as Fe 3+ ) and ≈716 eV (assigned as Fe 4+ ).At the ToC, the Fe 3+ peak shows the strongest intensity and dominates the Fe 2p 3/2 XPS spectra instead of the Fe 4+ peak, indicating that Fe 3+ /Fe 4+ oxidation is also limited in LFNOF.The O1s XPS spectra show concurrent O oxidation with Fe upon charging, but the detection of oxidized O species is less compared to the case of LFNO (Figures 4d; S6, Supporting Information).These results suggest that Fe 2+ /Fe 3+ oxidation occurs readily in LFNOF to reduce the O oxidation upon charging, explaining the long ≈2.7 V 1 st charging plateau for LFNOF that is barely seen for LFNO.Upon discharging, the Fe 2+ peak recovers its intensity over the Fe 3+ or Fe 4+ peaks, and oxidized O species are detected less in the XPS spectra, indicating Fe and O reduction.
To further confirm the Fe redox activity in the LFNO and LFNOF, we performed EPR spectroscopy, characterizing the Fe oxidation state at BC, at the ToC to 4.8 V, and at the EoD during the 1 st cycle (Figure 4e,f). For LFNO, the Fe 3+ EPR signal at ≈3330 G is prominently observed in both BC, ToC, and EoD samples, underscoring the predominant presence of Fe 3+ in LFNO throughout the 1st cycle (Figure 4e).In the case of LFNOF, the EPR signal is silent for the BC sample (Fe 2+ is EPR silent).Subsequently, the Fe 3+ EPR signal prominently appears at the ToC, and the EPR signal becomes silent again at the EoD.This finding aligns with our XPS observation that Fe 2+ /Fe 3+ oxidation is readily accessible in LFNOF, whereas Fe 3+ /Fe 4+ oxidation is limited for both LFNO and LFNOF.
Concerning the structural changes during cycling, we observe a characteristic "lattice breathing" behavior in DRX cathodes from LFNOF, where the volume (lattice parameter) contracts (decreases) upon charging and expands (increases) upon discharging, reflecting ionic radius changes.In contrast, LFNO exhibits a non-intuitive volume expansion (lattice parameter decrease) during the 1st charge.Figure 4g,h depicts the (002) XRD peak of LFNO and LFNOF during the 1st cycle, respectively.For LFNOF, the (002) peak gradually shifts to a higher angle with charging, returning to a position similar to the pre-cycling state after the 1st discharge (Figure 4h).The XRD refinement indicates that this peak shift corresponds to LFNOF's lattice parameter changing from ≈4.215 Å (BC) to ≈4.174 Å (ToC) and then to ≈4.220 Å (EoD) (Figure 4j), showcasing a typical structural evolution observed in other DRXs. [36,39,40]Meanwhile, for LFNO, the (002) peak shifts to a lower angle after the 1st charge, corresponding to the lattice parameter increase from ≈4.191 Å to ≈4.197 Å (Figure 4g,i).Subsequently, the (002) peak further shifts to a lower angle (≈4.222Å) after the 1st discharge (EoD).The increase in LFNO's lattice parameter (volume) during the 1st charge differs from the typical behavior of DRXs, where the volume usually decreases upon charging due to ions (TM or O) getting smaller upon oxidation.We currently speculate that LFNO's volume expansion upon 1st charging is related to dominant O oxidation in the structure, leading to excessive O dimer formation.This process may cut some Fe─O bonds, ultimately increasing the average TM-O distance in LFNO.Overall, the volume changes observed in LFNO and LFNOF during the first cycle (from minimum to maximum volume) are ≈2.7% and ≈3.2%, respectively (Figure 4i,j).These figures are smaller than the ≈6.8% volume change exhibited by LiFePO 4 , [2] suggesting potential advantages for developing a mechanically stable electrode.We note that the volume changes of LFNO aggravate with cycling from 2.7% (1 st cycle) to 4.0% (2 nd cycle), 5.8% (5 th cycle), and 6.8% (10 th cycle), whereas that of LFNOF is limited to ≈4% after 10 cycles, indicating improved structural stability for LFNOF (Figure S7, Supporting Information).
DFT calculations on LFNO and LFNOF corroborate our experimental findings on redox mechanisms.The calculations were carried out on disordered rock-salt supercells (Li representing LFNOF) with the most stable cation arrangement generated using a genetic algorithm based on a monoclinic LiMO 2 structure. [41]The fully relaxed DFT structures were subsequently subjected to AIMD simulations at 1000 K, followed by DFT relaxation for simulated ion migrations to achieve more energetically stable structures.The calculated structures of different compositions were then selected for post-analysis, including the calculated voltage profile and occupancy of atomic species based on the convex hull (Figure S8, Supporting Information).The occupancy of atomic species of all calculated compositions (i.e., including compositions excluded from the convex hull) can be found in Figure S9 (Supporting Information).
Our DFT voltage profiles reproduce the two-phase-reactionlike near-flat 1st charging voltage observed in GITT for LFNO, as well as the more sloped voltage profile with distinct ≈2.Based on the magnetic moments and Bader charge analysis on the structures on the DFT convex hull, we investigated the Fe and O oxidation states in LFNO and LFNOF (Figure 5c,d).Detailed information about the oxidation-state assignment is provided in the Methods section and Figures S10 and S11 (Supporting Information).For LFNO, concurrent Fe 3+ /Fe 4+ and O 2− /O n− oxidation is observed, with only ≈30% and ≈40% of Fe atoms in the structure found as Fe 4+ at highly delithiated states of x = 1.111 and 1.222 in Li 1.222-x Fe 0.556 Nb 0.222 O 2 , respectively (Figure 5c).This indicates limited Fe 3+ /Fe 4+ oxidation, supporting our experimental findings.For LFNOF (Figure 5d), our DFT results suggest complete oxidation of Fe 2+ to Fe 3+ after removing 0.666 Li (x = 0.666 in

Outlook on the Fe-Rich DRX Cathodes
The cost-effectiveness and abundance of Fe compared to other TMs (e.g., Ni, Co, Mn, V, Mo) make Fe-rich cathode materials highly desirable for LIBs. [1]Our investigation suggests the possibility ofdesigning Fe-rich Li-ion cathodes with high energy density (≈700 Wh kg −1 ) by tuning the Fe redox processes within the DRX structure.In essence, the readily accessible Fe 2+ /Fe 3+ redox in LFNOF (an Fe 2+ -DRX) reduces dependence on the challenging Fe 3+ /Fe 4+ and O redox processes during cycling compared to LFNO, an Fe 3+ -DRX like Li 1.2 Fe 3+ 0.4 Ti 0.4 O 2 .While it is necessary to narrow the voltage window, the notable performance of LFNOF, achieving ≈704 Wh kg −1 and ≈290 mAh g −1 (1.3-4.8V, 40 mA g −1 ), stands out.To the best of the authors' knowledge, this represents the highest energy density and capacity among reported Fe-DRX cathodes.Also, it ranks among the highest energy densities achieved by any Fe-redox-based cathode, surpassing benchmarks such as LiFePO 4 (≈560 Wh kg −1 ) and LiFeSO 4 F (≈500 Wh kg −1 ).Interestingly, when considering more practical voltage windows, LFNO (Fe 3+ -DRX) demonstrates superior energy density compared to LFNOF.Specifically, LFNO achieves ≈590 Wh kg −1 in the range of 1.3-4.6V and ≈540 Wh kg −1 in 1.3-4.4V.This observation suggests potential benefits in combining Fe 3+ /Fe 4+ and O-redox with Fe 2+ /Fe 3+ for enhanced performance.While Fe 2+ /Fe 3+ redox provides better reversibility, the less reversible Fe 3+ /Fe 4+ and O-redox, with a higher operating voltage than Fe 2+ /Fe 3+ , offers an option for increasing the energy density of Fe-DRX within a more practical voltage window.Furthermore, while not reaching the near-zerostrain observed in V-based DRXs facilitated by reversible V migration and the use of t 2g V electrons, [42][43] the volume changes in LFNOF (3.2% in the 1 st cycle and ≈4% in the later cycles) are still small, especially when compared to other Fe-based cathode materials like LiFePO 4 (≈6.8%).These modest volume changes are advantageous for constructing high-loading and high-capacity electrodes, contributing to improved mechanical stability during cycling.
Meanwhile, the observation that both LFNO (Fe 3+ -DRX) and LFNOF (Fe 2+ -DRX) exhibit a high capacity (>250 mAh g −1 ) at a low average discharge voltage of ≈2.5 V suggests the need for strategies to enhance their operating potential.If achieving an average discharge voltage of ≈3.0 V in Fe-DRXs is feasible, their energy density could be comparable to widely studied Mn-DRXs (e.g., Li-Mn-Ti/Nb-O, 800-1000 Wh kg −1 ). [1]This potential equivalence allows DRX cathodes to reduce costs by utilizing cheaper Fe compared to Mn while maintaining the energy density, paving the way for the development of ultimate Fe-rich cathode materials for LIBs.A potential suggestion involves incorporating P or B cations into the Fe-DRX structure to promote P─O or B─O bonds.By leveraging the inductive effect of P or B through P(B)─O─Fe orbital hybridization, the Fe─O bond can become more ionic, leading to a lowering of the Fe 3d band energy.These dopants may thereby elevate the operating potential of Fe-DRXs, similar to their impact on the LiFePO 4 cathode. [2]Furthermore, the P(B)─O bond has the potential to stabilize O n− anions formed during O oxidation, thereby enhancing the reversibility of O redox in Fe-DRXs.Encouragingly, successful P/B doping has been demonstrated in Mn-DRXs, [44][45] providing optimism for the application of a similar strategy to enhance the operating voltage and cycling stability of Fe-DRXs.
Finally, we find the need for novel electrolytes for improved cycling stability of the Fe-DRXs.While reducing the Fe-oxidation state toward Fe 2+ can introduce more Fe-redox capacity and decrease the reliance on the problematic Fe 3+ /Fe 4+ and O-redox capacity, TM cations with lower oxidation states tend to dissolve into the electrolyte more than those with higher oxidation states. [46]n this context, highly concentrated electrolytes would be worth investigating, as they have previously demonstrated reduced dissolution of Mn-DRX oxyfluorides (e.g., Li 2 MnO 1.5 F 1.5 ), which share similar chemical characteristics (i.e., oxyfluorides containing low-valent TM) with LFNOF. [25]

Conclusion
In summary, this study introduced Fe 2+ /Fe 3+ redox in a Fe-rich DRX cathode, Li 1.2 Fe 2+ 0.6 Nb 0.2 O 1.4 F 0.6 (LFNOF), through fluorination and compared its performance with Li 1.2 Fe 3+ 0.6 Nb 0.2 O 2 (LFNO), which relies on limited Fe 3+ /Fe 4+ redox and appreciable, largely irreversible O redox for high capacity.The combined experimental and computational investigation revealed that the easily accessible Fe 2+ /Fe 3+ redox in LFNOF enables improved performance with reduced dependence on O redox.This results in mitigated voltage/capacity fading and voltage hysteresis, leading to higher capacity (≈290 mAh g −1 ) and energy density (≈700 Wh kg −1 ) in LFNOF compared to LFNO (≈600 Wh kg −1 ) when cycled between 1.3 and 4.8 V at 40 mA g −1 .These achievements represent some of the highest reported values for Fe-DRXs and other Fe-rich cathodes.Meanwhile, reducing the charge cutoff voltage to 4.4 V resulted in LFNO surpassing LFNOF in energy density (≈540 Wh kg −1 compared to ≈460 Wh kg −1 ).This observation suggests a potential advantage in combining Fe 3+ /Fe 4+ with Fe 2+ /Fe 3+ in Fe-DRXs to achieve high energy density in a narrower voltage window.In conclusion, our study offers new strategies for designing low-cost, high-performing Fe-rich DRX cathodes with reduced reliance on O redox.

Experimental Section
Material Synthesis: To synthesize Li  O (Alfa Aesar, 99.5%), Fe (Sigma-Aldrich, 99%), Fe 2 O 3 (Sigma-Aldrich, 96%), LiF (Sigma-Aldrich, 99.99%), and Nb 2 O 5 (Alfa Aesar, 99.5%) were used as precursors.A total amount of 2 g precursors was mixed with a Fritsch Planetary Micro Mill (PULVERISETTE 7) at a rate of 500 rpm for 40 h in stainless steel vials assembled in an Argon-filled glovebox.The grinding media were fifteen 10 mm diameter stainless balls and ten 5 mm diameter balls.After the mixing, the synthesized powders were collected in an Argon-filled glovebox.
Electrochemical Tests: For the preparation of a cathode film, the synthesized powders (140 mg) and Super C65 carbon black (40 mg) were mixed in the PULVERISETTE 7 for 1 h at 300 rpm, with 3 mm diameter (20 g) grinding media in stainless-steel vials assembled in an Argon-filled glovebox.Upon collection in an Argon-filled glovebox, the as-mixed powders (90 mg) and polytetrafluoroethylene (PTFE, 10 mg) were manually mixed with a mortar and pestle.The mixture was then rolled into a thin film inside an Argon-filled glovebox.The weight ratio between the active material, carbon black, and PTFE is 70:20:10 in the cathode film.Coin cells (CR2032) were assembled with the cathode film, the Li-counter electrode, a polypropylene separator (Celgard 2400), and 1 м solution of LiPF 6 in a mixture of ethyl carbonate/dimethyl carbonate (EC/DMC, 1:1 v/v) electrolyte in an Argon-filled glove box.The galvanostatic charge/discharge and rate-capability tests were performed using a Landt CT3002A battery testing system at room temperature.The specific capacity was calculated based on the amount of active materials (LFNO, LFNOF) in the cathode film.
Material Characterization: The X-ray diffraction (XRD) patterns were collected on a Malvern PANalytical Empyrean X-ray diffractometer (Cu source) in the 2 range of 10-90°.To perform XRD on the as-cycled electrodes, coin cells were disassembled in an Argon-filled glovebox and washed with DMC for 30 s.Then, the cathode film was placed on an airtight polycarbonate-domed sample holder with a zero-background plate.The Rietveld refinement on the collected XRD patterns was performed using the PANalytical X'pert HighScore Plus software.Scanning electron microscopy (SEM) was performed with the Hitachi SU-8000 SEM.Scanning transmission electron microscopy (STEM) imaging and energy dispersive X-ray spectroscopy (EDS) mapping were acquired using a Talos F200X G2 STEM (Thermo Scientific) equipped with a high-angle annular dark-field (HAADF) detector.The HAADF detector collection angles ranged from 58 to 200 mrad.X-ray Photoelectron Spectroscopy (XPS) of the as-synthesized powders and as-cycled cathode films was performed on a Thermo-Scientific K-Alpha with Al K-alpha radiation as the X-ray source for excitation.The cycled cathode films were collected from the disassembled coin cells, washed with DMC for 30 s, and dried under vacuum in the Argon-filled glovebox overnight.The dried, washed, cycled cathode films were then transferred into the XPS spectrometer with a vacuum transfer module (Thermo Scientific) to prevent any air exposure.The post-mortem XPS measurements on the as-cycled electrodes were conducted after 30 s Argon sputtering with 0.5 keV ion energy.The binding energy scale was charge-corrected using the C 1s peak at 284.8 eV from the hydrocarbon contamination.The peak positions and areas were optimized using 70% Gaussian and 30% Lorentzian line shapes using Avantage (Thermo Scientific) software.XPS quantification was performed based on Scofield's relative sensitivity factors.
They were employed with a plane-wave energy cutoff of 520 eV, k-point mesh density of at least 1000 per number of atoms, and converged to an electronic and a force convergence criteria of 10 −6 eV and 0.05 eV Å −1 , respectively.Perdew-Burke-Ernzerhof (PBE) functional was used for spin-polarized generalized gradient approximation calculations with a Hubbard U correction (GGA+U) for only Fe (5.3 eV). [49] The oxidation states of the Fe and O ions were determined based on their Bader charges processed by the Henkelman group's code and their magnetic moments. [52]Detailed explanation for the assignment of the oxidation states of Fe and O species based on the Bader charge and magnetic moment analyses can be found in Figures S10  and S11 (Supporting Information).
To obtain the disordered rock-salt structures with the most stable cation and anion arrangements based on the electrostatic force, the home-made genetic algorithm (GA) method was used to generate the 3 × 3 × 2 supercells based on the monoclinic LiMO 2 with similar cation composition at every layer along the c axis. [53]A population size of 16 was used for GA and a total of 1000 generations was generated.Among them, thirty configurations with the lowest electrostatic energies were fully relaxed using GGA+U to obtain the ground-state DRX structures with different levels of lithiation.
AIMD simulations were performed with GGA+U and the NVT canonical ensemble using a Nosé-Hoover thermostat with a period of 80 fs, a planewave energy cutoff of 400 eV, and a minimal gamma-centered 1 × 1 × 1 k-point grid.Temperatures were initialized at 100 K and scaled to 1000 K for 300-time steps, followed by a hold at 1000 K for 10000-time steps, totalling 20 ps.Among them, twenty configurations were sampled after 1000-time steps (equilibrium phase) at every 450-time step and fully relaxed with GGA+U.

Figure 2 .
Figure 2. Electrochemical characterization of the LFNO and LFNOF cathodes.Voltage profiles of the initial three cycles of a) LFNO and b) LFNOF when cycled between 1.3 and 4.8 V at 40 mA g −1 .The insets show the capacity retention.c) Charging and discharging profiles of LFNO and LFNOF in their second cycle.d,e) Differential capacity (dQ/dV) of (d) LFNO and (e) LFNOF in the 1 st , 10 th , and 20 th cycle.f) The corresponding average discharge voltage of LFNO and LFNOF upon cycling.

Figure 3 .
Figure 3.The performance of the LFNO and LFNOF at various upper cut-off voltages and rates of cycling.a-d) The initial three-cycle voltage profiles of (a,b) LFNO and (c,d) LFNOF when cycled at 40 mA g −1 between 1.3 and 4.6 and 1.3-4.4V, respectively.The insets show capacity retention.e) The average discharge voltage evolution of LFNO and LFNOF was obtained from Figure 3a-d.f) The first discharge specific energy and average discharge voltage of LFNO and LFNOF when cycled at 40 mA g −1 and different voltage windows (1.3-4.8/4.7/4.6/4.5/4.4V). g,h) The discharge voltage profiles of LFNO and LFNOF when charged at 20 mA g −1 and discharged at different rates of 10, 20, 40, 100, 200, 400, and 1000 mA g −1 between 1.3 and 4.8 V. i) The estimated Li diffusivity from the GITT measurements as in Figure S2 (Supporting Information).Refer to Figure S3 (Supporting Information) for more information about the calculation.

Figure 4 .
Figure 4. Redox mechanisms of LFNO and LFNOF cathodes.a,b) The Fe 2p 3/2 X-ray photoelectron spectroscopy (XPS) spectra after 30 s Ar sputtering of (a) LFNO and (b) LFNOF at progressive 1 st cycling stages at 40 mA g −1 : before cycle (BC), charged to 80 and 160 mAh g −1 (C80, C160), top-of-charge (ToC at 4.8 V), discharge to 80 and 160 mAh g −1 (D80, D160), and at the end-of-discharge (EoD at 1.3 V).The complete Fe 2p spectra can be found in Figure S4 (Supporting Information).c,d) The evolution of Fe 2+ , Fe 3+ , Fe 4+ , O 2− , and O n-species in (c) LFNO and d) LFNOF during each of the cycling stages as estimated via XPS fitting profile areas.e,f) Electron paramagnetic resonance (EPR) spectra of the evolution of Fe 3+ species in (e) LFNO and (f) LFNOF in various initial cycling stages at 40 mA g −1 :ToC and EoD as well as non-cycled (pristine).g,h) The evolution of the ex situ XRD (002) peak of (g) LFNO and (h) LFNOF during cycling (40 mA g −1 , 1.3-4.8V) and i,j) their corresponding lattice parameters calculated from the XRD refinement.

Figure 5 .
Figure 5. Computational study of the redox mechanisms in LFNO and LFNOF.a,b) The DFT voltage profiles and GITT 1st charge profiles of (a) LFNO and (b) LFNOF.c,d) The percentage of Fe and O species with different oxidation states with respect to the Li content in (c) LFNO and (d) LFNOF.e,f) The DFT calculated structures of (e) LFNO and (f) LFNOF with O dimers encircled by black dashed lines and their O─O bond lengths.The red, white, light green, dark green, and brown spheres represent O, F, Li, Nb, and Fe ions, respectively.
7 and ≈4.3 V plateaus in LFNOF (Figure 5a,b), providing a reliable voltage prediction.For LFNO, a highly stable DRX structure containing two O dimers in the supercell forms at x = 1.111 in Li 1.222-x Fe 0.556 Nb 0.222 O 2 , promoting a two-phase-reaction across a wide range of Li contents (Figure S8, Supporting Information).In LFNOF, structures with intermediate Li contents (x = 0.444, 0.666, and 1.111 in Li 1.222-x Fe 0.556 Nb 0.222 O 1.444 F 0.556 ) also exist on the convex hull, contributing to a more stepped voltage profile (Figure S8, Supporting Information).
Li 1.222-x Fe 0.556 Nb 0.222 O 2 ), during which O oxidation appears to be minimal.Subsequently, O 2− /O n− oxidation occurs simultaneously with slight Fe 3+ /Fe 4+ oxidation upon further Li extraction, as inferred from the population change of species in the structure (Figures 5d; S12, Supporting Information).Overall, our DFT results affirm the readily accessible Fe 2+ /Fe 3+ oxidation in LFNOF, which reduces O oxidation in the material compared to LFNO, where the limited Fe 3+ /Fe 4+ oxidation requires earlier and more significant O oxidation during charge.In this context, we also note a lower presence of O dimers in LFNOF (one O dimer per supercell with 36 O atoms) compared to LFNO (two O dimers per supercell) after complete Li extraction (Figure 5e,f).