Layered-rocksalt intergrown structure

Li_1.2Ni_0.4Ru_0.4O₂ was prepared by a solid-state reaction and the crystal structure at pristine state is carefully examined by a joint synchrotron X-ray diffraction (sXRD) and neutron diffraction (ND). All the reflections in Fig. 1a can be well indexed, based on layered R$$ \overline{3}$$m structure. In particular, no super-lattice peaks in the 2θ region of 5°–9° (λ = 0.4127 Å), originating from the Li/TM ordering in the TM slabs, are noticed in sXRD of Li_1.2Ni_0.4Ru_0.4O₂. Rietveld refinement of sXRD for pristine Li_1.2Ni_0.4Ru_0.4O₂ based on R$$ \overline{3}$$m space group leads to a good fit in the peak position. However, a discrepancy in the peak intensity is revealed, especially for the intensity ratio of reflection (003)/(104), which is an important indicator of the degree of cation ordering in layered R$$ \overline{3}$$m phase (Supplementary Fig. 2). Low intensity ratio of reflection (003)/(104) in layered R$$ \overline{3}$$m could be due to Li/Ni intermixing because of the similar ionic radius of Li⁺ (0.76 Å) and Ni²⁺ (0.69 Å)^42–45. Simulation of XRD patterns based on R$$ \overline{3}$$m clearly shows the decreased intensity of (003) reflection with respect to (104) reflection when the level of Li/Ni mixing increases (Supplementary Fig. 3). Additional simulated XRD and ND patterns for R$$ \overline{3}$$m and Fm$$ \overline{3}$$m are presented in Supplementary Fig. S4. Therefore, further joint refinement of sXRD and ND is performed based on single R$$ \overline{3}$$m phase model with Li/Ni intermixing (Supplementary Fig. 5). Single R$$ \overline{3}$$m phase model that allows Li/Ni intermixing results in a better fit in peak intensity with a final R-factor of 13.8%, revealing ~8.0% Li/Ni intermixing. Meanwhile, close comparison of the calculated and observed XRD patterns reveals the deviation of the reflections around 9.9°, 11.4°, and 16.1° (λ = 0.4127 Å), which are in accordance with the characteristic reflections of the rocksalt phase. Furthermore, a joint refinement based on layered-rocksalt biphasic model leads to an even lower R-factor of 9.4% and the optimal refinement indicates the final product is composed of 70 mol% layered and 30 mol% rocksalt phase (Fig. 1a, Supplementary Fig. 6, and Supplementary Table 1).

Fig. 1

Li_1.2Ni_0.4Ru_0.4O₂ of layered-rocksalt intergrown structure.

a sXRD and ND patterns with Rietveld fits for Li_1.2Ni_0.4Ru_0.4O₂; refinement is performed based on R$$ \overline{3}$$m and Fm$$ \overline{3}$$m biphasic model, indicating 70 mol% R$$ \overline{3}$$m (a = 2.94194(7) Å, c = 14.40729(2) Å, and V = 107.990(1) ų) and 30 mol% Fm$$ \overline{3}$$m (a = 4.15874(6) Å and V = 71.926(1) ų); critical reflections are indexed as (hkl)_L and (hkl)_R for R$$ \overline{3}$$m and Fm$$ \overline{3}$$m, respectively. b Representative HAADF-STEM image and FFT of the selected areas highlighted by orange and red box, respectively, showing the layered-rocksalt intergrown structure along the [110] zone axis. Scale bar is 2 nm. c Schematic of layered-rocksalt intergrown structure along the [110] zone axis, pink highlight showing the structurally compatible region with different TM arrangements in the Li and TM slabs.

To further verify the structure of Li_1.2Ni_0.4Ru_0.4O₂, high-angle annular dark field-scanning transmission electron microscopy (HAADF-STEM) was employed to directly visualize the atomic distribution of TMs. A number of particles are examined and a representative HAADF-STEM image is shown in Fig. 1b, with additional images from different particles presented in Supplementary Fig. 7 supporting this global feature. HAADF-STEM clearly reveals the typical layered arrangement of TMs in one domain (left) and rocksalt pattern in the other domain (right), also confirmed by fast Fourier transformation (Fig. 1b). More importantly, the layered-rocksalt components are not randomly separated in crystal grains; instead, they exhibit an intergrown structure that firmly anchors the rocksalt domain in the main layered phase. Careful examination of the boundary between layered and rocksalt domains reveals a structurally compatible region, where a gradual transition is clearly distinguished by different TM distribution in the Li and TM slabs, as opposed to a grain boundary. Electron energy loss spectroscopy (EELS) mapping of the pristine Li_1.2Ni_0.4Ru_0.4O₂ (Supplementary Fig. 8) reveals uniform elemental distribution in both rocksalt and layered regions. Therefore, our combined sXRD, ND, and STEM analysis consistently and unambiguously reveals the new layered-rocksalt intergrown structure for pristine Li_1.2Ni_0.4Ru_0.4O₂ (Fig. 1c).

Electrochemical characterization

Electrochemical activity of Li_1.2Ni_0.4Ru_0.4O₂ with a layered-rocksalt intergrown structure (Fig. 2) is investigated directly on the as-produced material with a particle size of ~500 nm (Supplementary Fig. 9) without further modification. It is initially subjected to galvanostatic charge and discharge testing at various charge cutoff voltages, ranging from 3.9 to 4.8 V. We reveal continuous Li⁺ extraction and increased Li⁺ uptake upon increasing charge cutoff voltage (Fig. 2a). Li_1.2Ni_0.4Ru_0.4O₂ displays the best electrochemical reversibility between 4.6 and 2.5 V, featured by ~1.1 Li⁺ extraction and 0.95 Li⁺ re-insertion (244 mAh g⁻¹ and 904 Wh kg⁻¹) during charge and discharge, respectively. Given the total of 1.2 Li⁺ inventory in the material, such layered-rocksalt intergrown oxide enables high % of Li⁺ extraction/insertion, which is comparable to that in Li-rich layered oxide and disordered rocksalt. No additional reversible capacity is gained beyond 4.6 V charge cutoff. Further expanding the voltage window to 4.8–1.5 V leads to a discharge capacity of 333 mAh g⁻¹ (Supplementary Fig. 10). The differential capacity curves (Fig. 2b) are characterized by a sharp anodic peak around 3.8 V upon charge with a common cathodic peak around 3.75 V upon discharge, perhaps relating to Ni redox. Ni²⁺/Ru⁵⁺ or Ni³⁺/Ru⁴⁺ combination is possible in Li_1.2Ni_0.4Ru_0.4O₂. Nickel can be electrochemically active through Ni²⁺/Ni⁴⁺ (2e⁻) or Ni³⁺/Ni⁴⁺ (1e⁻), but only Ru⁴⁺/Ru⁵⁺ redox is possible for Ruthenium. In either case, TM redox can only account for 0.8 Li⁺ (206 mAh g⁻¹). Interestingly, an additional cathodic peak around 4 V starts to evolve when the charge cutoff voltage reaches to 4.6 V, indicating the possible contribution of oxygen redox in the high-voltage region. Meanwhile, Li_1.2Ni_0.4Ru_0.4O₂ demonstrates better capacity retention at cutoff voltages < 4.3 V (Fig. 2c). The rate capability is also evaluated directly on Li_1.2Ni_0.4Ru_0.4O₂ at the rates ranging from C/50 to 1C between 4.6 and 2.5 V. The material delivers a discharge capacity of ~200 and 165 mAh g⁻¹ at C/2 and 1C, respectively (Fig. 2d). With increasing current density, the charge and discharge profiles mostly retain, characterized by a pair of anodic/cathodic peaks around 3.75 V (Fig. 2e), whereas the cathodic peak around 4 V remains at low rates and becomes less pronounced at ≥C/10, in accordance with slightly high polarization at >4 V discharge observed by Galvanostatic intermittent titration technique (Supplementary Fig. 10). In general, the layered-rocksalt intergrown Li_1.2Ni_0.4Ru_0.4O₂ displays a high capacity and good rate capability; more importantly, it largely mitigates the notorious hysteresis of typical Li-rich layered oxides during the initial cycles.

Fig. 2

Electrochemical characterization of Li_1.2Ni_0.4Ru_0.4O₂.

a The first cycle voltage profiles, b dx/dV plots, and c voltage profiles during the first five cycles at different charge cutoff voltages. d Voltage profiles and e dQ/dV plots at different rates. Cells are cycled at 5 mA g⁻¹ in a–c and between 4.6 and 2.5 V in d, e.

Isotropic and nearly zero-strain structural evolution

To investigate the evolution of layered-rocksalt intergrown structure upon electrochemical cycling, in situ sXRD patterns are collected on a pouch cell composed of Li_1.2Ni_0.4Ru_0.4O₂//Li between 4.8 and 2.5 V at C/10 (Fig. 3a). Here, in situ sXRD analysis mainly focuses on the general structural change upon delithiation/lithiation, because the reflections of layered R$$ \overline{3}$$m and Fm$$ \overline{3}$$m rocksalt partially overlap with those of in situ pouch cell. Clearly, there is no new phase formation upon electrochemical cycling. Strikingly, the lattice parameters a and c of layered R$$ \overline{3}$$m component exhibit an isotropic change (Supplementary Fig. 11), as evidenced by all reflections shifting to a slightly higher diffraction angle upon charging and shifting back upon discharging. Such isotropic change in crystal lattice upon electrochemical cycling is further verified by ex situ sXRD collected on the cycled electrodes at different states of charge (Supplementary Fig. 12). More importantly, the layered-rocksalt intergrown structure still remains even after 100 cycles (Supplementary Fig. 13). Indeed, conventional layered oxides of R$$ \overline{3}$$m structure experiences an anisotropic change, which is characterized by a gradual increase in c lattice parameter accompanied by a slight decrease in a lattice upon delithiation, due to the change in ionic radius of TM and repulsion between the TM slabs at different states of charge^46,47. In sharp contrast, although with 70 mol% layered R$$ \overline{3}$$m phase, Li_1.2Ni_0.4Ru_0.4O₂ displays an isotropic structural change, resembling that of Li-excess disordered rocksalt. Therefore, 30 mol% intergrown rocksalt can effectively manipulate the slabs of the layered matrix so that an isotropic lattice change becomes dominant upon delithiation/lithiation.

Fig. 3

Nearly zero-strain isotropic structural evolution of Li_1.2Ni_0.4Ru_0.4O₂ upon delithiation/lithiation.

a In situ sXRD of Li_1.2Ni_0.4Ru_0.4O₂, the black pattern at the bottom is the background from the in situ cell; cell was cycled between 4.8 and 2.5 V at C/10. b sXRD of Li_xNi_0.4Ru_0.4O₂ (x = 1.2, 0.5, 0.2, 0) prepared by chemical delithiation method. c Joint refinement of sXRD and ND patterns of Li_0.2Ni_0.4Ru_0.4O₂. Critical reflections are indexed as (hkl)_L and (hkl)_R for R$$ \overline{3}$$m and Fm$$ \overline{3}$$m, respectively.

Furthermore, a series of Li_xNi_0.4Ru_0.4O₂ samples at varied states of delithiation have been prepared via a chemical delithiation method for the detailed analysis of the structural evolution of each component. sXRD patterns of chemically delithiated Li_xNi_0.4Ru_0.4O₂ (Fig. 3b) are consistent with those obtained from the electrochemical cells. All the characteristic reflections retain upon delithiation. Close comparison reveals a very small shift towards the higher 2θ angle. For Li_0.5Ni_0.4Ru_0.4O₂, the most pronounced change is the decrease in the intensity of the reflections at 10.9, 12.7, and 17.9° (λ = 0.4577 Å), which are characteristic of rocksalt component, suggesting the delithiation perhaps starts from Fm$$ \overline{3}$$m rocksalt. The refinement of the ND reflection of chemically delithiated Li_0.5Ni_0.4Ru_0.4O₂ sample also indicates the preference of Li⁺ extraction from rocksalt rather than layered phase during initial Li⁺ extraction, as evidenced from the slightly higher lithium content in the R$$ \overline{3}$$m phase than in the Fm$$ \overline{3}$$m phase (Supplementary Fig. 14). A specific constrain is created in the refinement process (Supplementary Note 1). These results imply the two-dimensional Li⁺ diffusion channels of layered R$$ \overline{3}$$m can help facilitate the extraction of Li⁺ from the intergrown Fm$$ \overline{3}$$m rocksalt.

Chemically delithiated Li_0.2Ni_0.4Ru_0.4O₂ sample is further investigated, as it is close in composition to the electrode electrochemically charged to 4.6 V, with almost 1 Li⁺ extracted from Li_1.2Ni_0.4Ru_0.4O₂. Joint sXRD and ND refinements based on biphasic model generate a final R-factor of 8.0% (Fig. 3c and Supplementary Table 2). It is worth noting that both lattice parameter a and c of the layered R$$ \overline{3}$$m component show an exceptionally low change of ~1%, which can be referred to as “nearly zero-strain” electrode. Moreover, the fraction of layered R$$ \overline{3}$$m and Fm$$ \overline{3}$$m rocksalt is consistent with that of the pristine state, suggesting the delithiation process does not alter the overall phase composition. Therefore, the layered-rocksalt intergrown phase displays an excellent structural robustness with the minimal change in lattice parameters upon delithiation/lithiation.

Cationic TM redox mechanism

In parallel with the structural evolution study, the oxidation states of Ni and Ru at pristine and different states of charge are probed using hard X-ray absorption spectroscopy (XAS), to determine the charge compensation mechanism of TMs. Samples that are of interest are selected for a detailed characterization based on the dQ/dV plot (Fig. 4a). From hard XAS (Fig. 4b–e), the energy of Ru and Ni at half maxima is consistent with charged Li₂RuO₃ (red dot line in Fig. 4b, c) and LiNi_1/3Mn_1/3Co_1/3O₂ (black dot line in Fig. 4d, e), implying Ru⁵⁺/Ni²⁺ combination in pristine Li_1.2Ni_0.4Ru_0.4O₂. When the electrode is charged to 3.9 V, Ru K-edge energy remains consistent with Ru⁵⁺ reference, because Ru⁵⁺ cannot be further oxidized, whereas the Ni K-edge shifts from Ni²⁺ reference to a higher energy, a clear indication of Ni oxidation. Interestingly, both Ru and Ni K-edge show an abnormal shift to a lower energy upon further charging to 4.3 V and beyond, up to 4.8 V, indicating the “reduction” of Ru and Ni upon charging in the high-voltage region. Ni reduction at a highly charged state is further verified by a two-dimensional transmission X-ray microscopy (TXM) measurement (Supplementary Fig. 15). These results further confirm that other oxidation reaction beyond TM such as O accounts for the Li⁺ extraction in the high-voltage region. Upon discharge, K-edge energy remains same until 2.0 V for Ru and 3.9 V for Ni, implying no cationic TM redox by 3.9 V. Therefore, cathodic peak at 4.0 V (Fig. 2b, e) can be unambiguously attributed to anionic O reduction. Further discharging to 1.5 V leads to Ru reduction to 4+, because Ru K-edge energy at 1.5 V matches that of pristine Li₂RuO₃ (black dot line in Fig. 4b, c). After 3.9 V discharge, Ni K-edge shows a significant shift to a lower energy, close to pristine 2+, which is fully recovered at 2 V, and shows no further change upon discharging to 1.5 V; therefore, Ni redox largely accounts for the anodic/cathodic peaks around 3.75 V (Fig. 2b, e). Such a trend in TM oxidation state change upon charging/discharging can be easily visualized in Fig. 4f, also confirmed by extended X-ray absorption fine structure (Supplementary Fig. 16). Overall, Ni and Ru are present as 2+ and 5+ at pristine state; Ni redox mainly accounts for cationic TM redox, whereas Ru remains inactive. We also infer that O participates in the electrochemistry in the high-voltage region, accounting for the second redox around 4 V, which will be further discussed below.

Fig. 4

Cationic redox mechanism of Li_1.2Ni_0.4Ru_0.4O₂.

a Voltage profiles and dQ/dV plot of Li_1.2Ni_0.4Ru_0.4O₂, showing samples (open circles) for ex situ XAS analysis. b, c XANES of Ru K-edge. d, e XANES of Ni K-edge. f Ni and Ru K-edge energy measured at half maxima at different states of charge.

Anionic oxygen-redox mechanism

Both electrochemistry (Fig. 2) and TM XAS (Fig. 4) indicate that anionic oxygen participates in the electrochemistry of Li_1.2Ni_0.4Ru_0.4O₂ in the high-voltage region. We therefore perform high-efficiency mapping of resonant inelastic X-ray scattering (mRIXS) at O K-edge, which has been established as a reliable probe of lattice oxygen redox^48,49. In general, the mRIXS images (Fig. 5) are dominated by three broad features around 525 eV emission energy (horizontal axis), which are typical O²⁻ features for oxides with excitation energies (vertical axis) of 528–533 eV and above 535 eV, corresponding to the TM-d and -s/p states hybridized to O-2p states, respectively⁵⁰. It is noteworthy that the excitation energy here is the same as that in typical O-K soft XAS spectra, but mRIXS is capable of differentiating the intrinsic oxidized oxygen signals from the dominating TM characters along the new dimension of emission energy, revealing a fingerprinting feature of lattice oxygen-redox state at 523.7 eV emission energy (red arrows in Fig. 5)^48,49. This particular mRIXS feature corresponds to the electron excitation into unoccupied O-2p states, thus fingerprinting the lattice oxidized oxygen, because O²⁻ has no unoccupied 2p states⁵⁰. This oxidized oxygen feature emerges when the Li_1.2Ni_0.4Ru_0.4O₂ electrode is charged to 4.1 V. Given about 0.8 Li⁺ is extracted from Li_1.2Ni_0.4Ru_0.4O₂ at 4.1 V charge, oxygen oxidation takes place with almost full oxidation of Ni²⁺ to Ni⁴⁺, consistent with our TM XAS results (Fig. 4). The intensity of the lattice O redox feature increases upon further charging, while the hybridization features along 525 eV emission energy also get enhanced due to the increasing covalency of the overall system upon charging. At 4.6 and 4.8 V charged states, the two groups of growing features get overlapped, but the oxidized oxygen feature remains clear via a direct comparison of the individual RIXS spectra cut from the mRIXS image along the 531 eV excitation energy (Supplementary Fig. 17). In addition, such oxidized lattice oxygen feature completely disappears at 2.5 V discharged state, indicating a reversible oxygen-redox reaction.

Fig. 5

Anionic redox mechanism of Li_1.2Ni_0.4Ru_0.4O₂.

O K-edge mRIXS results at different electrochemical states. Red arrows indicate the fingerprinting feature of oxidized oxygen at the excitation and emission energy of 531 and 523.7 eV, respectively. The feature emerges at 4.1 V during charge and disappears in the following discharge, clearly revealing a reversible lattice oxygen-redox reaction.

It is noteworthy that typical oxygen-redox-active Li-rich compounds always display a finite amount of broadening of the mRIXS features after discharge compared to pristine state, because of their severe structural changes during the initial cycle^17,50. In contrast, Li_1.2Ni_0.4Ru_0.4O₂ electrode at the discharged state here recovers completely to its pristine state. Again, this is highly consistent with the robustness of such layered-rocksalt intergrown structure upon cycling. The reversible lattice oxygen redox during the charge and discharge of Li_1.2Ni_0.4Ru_0.4O₂ is further supported by the gas evolution measured by operando differential electrochemical mass spectrometry (DEMS) (Supplementary Fig. 18), showing minimal oxygen and CO₂ gas release during the first cycle. A burst of CO₂ evolution at 4.3 V charge mostly originates from the carbonate residual from the synthesis⁵¹. Therefore, we clearly revealed that the lattice oxygen redox is mostly reversible in Li_1.2Ni_0.4Ru_0.4O₂ with negligible irreversible O loss, which is of critical importance not only for practical utilization of combined cationic and anionic redox reactions but also for fundamental understanding to differentiate these two oxygen activities, i.e., lattice oxygen redox and oxygen loss²⁸.

Synthesis

Li-rich metal oxides, Li_1.2Ni_0.4Ru_0.4O₂, along with other Li-Ni-Ru-O derivatives were prepared using Li₂CO₃, Ni(OH)₂, and RuO₂ as precursors. The precursors at designated stoichiometric amounts were first mixed on a Spex 8000 mill for 3 h, then fired at 450 °C for 3 h and 950 °C for 15 h in air, unless noted otherwise (Supplementary Fig. 25). The synthesis of this material via a solid-state reaction is very reproducible. Chemically delithiated samples were prepared by reacting Li_1.2Ni_0.4Ru_0.4O₂ with stoichiometric amounts of 0.1 M nitronium tetrafluoroborate (NO₂BF₄) in acetonitrile inside an Ar-filled glovebox (H₂O < 0.1 p.p.m.) overnight. Onefold excess NO₂BF₄ was used to prepare fully delithiated sample. The final products were obtained by filtering and thoroughly washing the resulting mixtures by acetonitrile until the residual solution was clear, then drying under vacuum overnight. The compositions of final Li_xNi_0.4Ru_0.4O₂ (1.2 ≤ x ≤ 0) and other Li-Ni-Ru-O derivatives were determined by inductively coupled plasma mass spectrometry analysis (Supplementary Table 7). Li-Ni-Fe-Mn-O derivatives were prepared by using Li₂CO₃, Ni(OH)₂, FeC₂O₄, and MnCO₃ as precursors. The precursors of designated stoichiometry were first mixed on a Spex 8000 mill for 3 h, followed by a calcination process at 450 °C for 3 h and 900 °C for 16 h in air.

Characterization

sXRD was taken at the Advanced Photon Source at Argonne National Laboratory on beamline 11-BM. The beamline uses a sagittal focused X-ray beam with a high precision diffractometer circle and perfect Si (111) crystal analyzer detection for high sensitivity and resolution. In situ and ex situ sXRD was performed on beamline 11-3 equipped with a bent flat, side-scattering Si (311) monochromator at a fixed energy of 12,700 eV (λ = 0.976 Å) at Stanford Synchrotron Radiation Lightsource (SSRL). Powder XRD patterns were collected on a Bruker D2-Phaser with Cu Kα radiation (λ = 1.54178 Å). XRD patterns were analyzed by the conventional Rietveld method using the general structure analysis system package with the graphical user interface (EXPGUI). Scanning electron microscopy was performed on a JEOL JSM-7000F. High-resolution STEM images and electron diffraction patterns were obtained on JEM-2100F (JEOL) and HD2700C-dedicated STEM (Hitachi) with probe corrector at an accelerating voltage of 200 kV. HAADF-STEM images were filtered using Digital Micrograph software. Hard XAS measurements were performed on beamline 2-2 at SSRL in a transmission mode using a (220) monochromator. Higher harmonics in the X-ray beam was reduced by detuning the Si (220) monochromator. Energy was calibrated using the first inflection points in the spectra of Ni and Ru foil reference. XANES data were analyzed by SIXPACK software with the Photoelectron Energy Origin E₀ determined by the first inflection point of the absorption edge jump. To visualize the valence state change of Ni at different states of charge, TXM was performed at beamline 6-2C at SSRL by loading the samples into quartz capillary tubes and sealing them inside a glovebox to avoid air exposure. A stack of transmission images (nominal spatial resolution of ~30 nm) recorded as the energy of the incoming X-rays was scanned across the Ni absorption K-edge. In the near-edge region, the energy step was set to be 1 eV for sufficient energy resolution. Over the pre-edge and post-edge regions, the energy was scanned at a larger step size of 10 eV, to cover a wide energy window for normalization of the X-ray absorption spectra. TXM data were analyzed using an in-house software package known as TXM-Wizard. RIXS maps were collected in the newly commissioned ultrahigh efficiency iRIXS endstation at Beamline 8.0.1 at the Advanced Light Sources. Sample surface was mounted 45° to the incident beam and the outgoing photon direction along the RIXS spectrograph is 90°. RIXS resolving power, other technical details, and data processing could be found in our previous report. All the cycled electrodes were immediately collected from the cells at designated voltages to minimize the side reactions between cycled electrodes and electrolyte, then vigorously washed by dimethyl carbonate solvent to remove the soluble surface species. All the dried electrodes were transferred into the experimental vacuum chamber through a specially designed sample transfer kit in an Ar-filled glovebox with no exposure to air. Operando DEMS measurements were taken on a customized Swagelok type cell connected to a high-pressure gas chromatography valve. The DEMS cells were initially rested at an open-circuit voltage for 6 h and the charge/discharge was done under potentiostatic control using a Bio-Logic SP-300 potentiostat.

Electrochemistry

Electrodes were prepared from slurries containing 80 wt% of active material, 10 wt% of polyvinylidene fluoride binder, and 10 wt% acetylene carbon black (Denka, 50% compressed) in N-methylpyrrolidone solvent. The slurries were casted on carbon-coated aluminum current collectors (Exopack Advanced Coatings) using a doctor blade and then dried under vacuum at 120 °C overnight. Typical loading of the active materials was ~2.5 mg cm⁻². 2032-type coin cells (Hohsen Corp.) containing Li metal, a Celgard 2400 separator, and 1 M LiPF₆ electrolyte solutions in 1 : 2 w/w ethylene carbonate–diethyl carbonate (Daikin America) were assembled inside an Ar-filled glovebox (H₂O < 0.1 p.p.m.). Galvanostatic charge and discharge were performed on a Maccor 4200 cycler at designated rates and voltages. A 1C current density was defined as 250 mA g⁻¹.