Metal-ion batteries, mainly lithium-ion batteries (LIBs) and sodium-ion batteries (SIBs), have attracted a considerable attention over the past few decades because of their applications in portable electronic devices, electric vehicles, and grid storage solutions, which have a direct impact on our quotidian life^[1,2,3,4,5]. The practical applications call for metal-ion batteries of excellent performance, i.e., higher energy density, lower cost, longer lifespan, and greater safety. To achieve good performance, the development of new electrode materials or optimization of available electrode materials that can work at high cell voltage and high specific capacity are the major important task in the field^{[6,7,8,9,10,11]}.

To develop the good electrode materials, it is important to have a fundamental understanding structure-activity relationship, that is the working mechanism and the controlling factors of the materials. Therefore, different kinds of characterization techniques (ex-situ, in-situ, operando) has been developed to monitor the composition and structure evolution of the electrode materials^[12].

Among various characterization techniques, magnetic resonance techniques, including nuclear magnetic resonance (NMR, including magnetic resonance imaging (MRI)) and electron paramagnetic resonance (EPR, including electron paramagnetic resonance imaging (EPRI)) are unique tools to probe structural information and phase information at atomic-level and to measure ionic motion at various time-scales.

Solid-state NMR spectroscopy (ssNMR) can provide a wealth of information on the materials in a non-disruptive and non-invasive manner, and is suitable for characterizing common light elements in electrode materials such as ^6/7Li, ²³Na, ³¹P, ¹⁹F, ¹H, ¹³C, etc., especially ¹⁷O, which provides a powerful tool for studying atom environmental changes and valence processes^[13]. For different types of observable atomic nuclei, suitable advance ssNMR methods (e.g., pjMATPASS: projected magic-angle-turning phase-adjusted-sideband-separation, Hahn-echo, HMQC: heteron-uclear multiple quantum coherence) can be deliberately selected to analyze the evolution of the short-range structure and electronic structure of battery materials. EPR, which is adept in studying metal ions (e.g., V, Co, Ni, Mn, Fe) containing unpaired ele-ctrons on the atomic orbital with high sensitivity, can help to characterize the electronic state and coordination state, and then elucidate the redox evolution of the metal center^[14]. ssNMR and EPR are two complementary techniques.

MRI and EPRI can provide chemical environment information at different spatial locations^[15,16]. MRI could be used to study the metal anodes and different kinds of electrolytes targeting at the specific nuclei (^6/7Li and ²³Na). However, the 2D (two-dimension) or 3D (three-dimension) MRI often has low spatial resolution (~ mm) due to the use of low-power gradients. To improve the spatial resolution, MRI is often performed in the 1D mode, using high-power Z-axis gradient. However, in the 1D mode, NMRI could detect the growth of the Li- dendrite in side view, but could not catch the Li growth in front view. To improve the spatial resolution, EPRI could be employed to monitor the metal anodes with the relatively high resolution of ca. 50 ~ 100 μm per pixel.

In this review, we focus on the topic of the electrode materials studied by NMR and EPR in our group. For solid polymer electrolytes, readers could refer to another review of our group on this topic. In the following part, the application of NMR will be discussed first, followed by the application of EPR, and finally, the combined use of NMR and EPR will be discussed.

To the best of our knowledge, previous NMR studies of LiCoO₂ have focused on ⁶Li, ⁷Li and ⁵⁹Co species, no ¹⁷O NMR studies have reported to probe the oxygen atoms in LiCoO₂. Our group has successfully produced the ¹⁷O-labeled LiCoO₂ and obtained the ¹⁷O NMR spectra to investigate the structural evolution of Li_xCoO₂ (0 ≤ x ≤ 1) during delithiation processes (Figure 1)^[17]. Here x denotes the mole fraction of Li⁺ in Li_xCoO₂.

From 0.85 ≤ x ≤ 1.0, only one peak at -636 ppm with the roughly the same intensity could be observed. This peak belongs to O3I insulator phase with low conductivity, which is a kind of O3H (O3 with hexagonal structure) phase. From 0.75 ≤ x ≤0.5, the peak at -636 ppm decreased quickly, suggesting that a large amount of paramagnetic Co⁴⁺ makes the ¹⁷O signal unobservable. This phase belongs to O3II conductor phase with high conductivity, which is also a kind of O3H phase. Due to the same chemical shift, O3I and O3II should have the very similar crystal structure. However due to the extraction of Li-ions, the c axis might be different, leading to different XRD diffractions for O3I and O3II phases.

At x = 0.5, a new set of peaks at 900 ppm began to appear. The results by fitting this set of peaks showed two peaks with similar C_Q (quadrupolar coupling constant). These two peaks could be assigned to the monoclinic Li_0.5CoO₂ phase, indicating the occurrence of a hexagonal-monoclinic phase transition (O3H → O3M). Here, it should be pointed out that Li_0.5CoO₂ is proposed for this intermediate phase at x = 0.5, while the previous ⁷Li NMR suggested an intermediate phase of Li_0.75CoO₂. The phase at 0.15 ≤ x ≤ 0.45 has been assigned to O3III phase, however, this phase should have the very similar structure to O3M phase at x = 0.5, as suggested by NMR results. This set of two peaks disappeared at x = 0.18, and then a new peak at 1250 ppm was observed. The peak at 1250 ppm can be assigned to the H1-3 phase.

Accordingly, the ⁵⁹Co spectra have also been acquired. The ⁵⁹Co NMR spectra had only one peak that only existed when x ≥ 0.5. It is suggested here that a large amount of Co⁴⁺ generated after x < 0.5, and then the electrons would hop between two Co⁴⁺ ions with t_2g⁵e_2g⁰ configuration through t_2g-t_2g overlap, resulting in electron delocalization, which makes the ⁵⁹Co NMR signal unobservable.

The C_Q value (~ 7 MHz) for the monoclinic phase obtained the ¹⁷O NMR results is quite large, indicating that there is very strong covalent bonding between Co and O atoms. Therefore, when charging at high-voltage, the Co-O bond is difficult to be distorted due to the directionality of the strong covalent bond, and the hexagonal-to-monoclinic phase transition takes place instead of local structural distortion. Thus, the charge of LiCoO₂ higher than 4.2 V with x < 0.5 will lead to the structural instability.

In recent years, in-situ NMR technology for the real-time analysis of the battery has been well developed. Most in-situ NMR studies were performed with a plastic cell in the static mode without magic angle spinning, and static NMR spectra show very broad line-shapes, which is difficult to be analyzed. Recently, a novel jelly-roll cell design was proposed for in-situ NMR and MAS (magic angle spinning) NMR spectra could be acquired for high resolution of the peaks^[18]. However, when the NMR data acquisition started, the wires had to be removed. During the detachment time of the wires, the system state would relax back to the equilibrium state. Subsequently, the wires were reattached for the following charging or discharging process.

Shimoda et al. conducted in-situ NMR studies on the charge and discharge processes for Li||LiCoO₂ with plastic cell^[19]. Our group also employed in-situ ⁷Li NMR to study uncoated and Ba-Ti-coated (barium and titanium coated) LiCoO₂ samples with plastic cell^[20]. The pristine LiCoO₂ is in the O3I phase with insulator character. In-situ NMR results suggested that the uncoated LiCoO₂ sample could not return to the pristine state after being charged to a high voltage of 4.5 V and then discharged to 3.0 V. The final state of uncoated LiCoO₂ after charge/discharge process had O3I-O3II coexisted, that is, the conductor phase and the insulator phase coexisted. This result is identical to the result of Shimoda^[19]. The Ba-Ti-coated LiCoO₂ samples could return to the pristine O3I phase through O3II-O3I phase transition upon the same charge/discharge process, because this sample had higher electrical conductivity due to the formation of TiO₂ layer and BaTiO₃ particles on the surface after the Ba-Ti coating. Therefore, it could be anticipated that the Ba-Ti coating can greatly improve the electrical conductivity and make the conductor-insulator phase transition more reversible.

NaCrO₂ is a candidate cathode material for sodium-ion batteries due to its safety, low cost, and high-power density. To increase the energy density, it is important to extract more Na ions at higher voltage, however, this often leads to fast capacity decay and irreversible structural transformation to CrO on the surface.

In-situ EPR was carried out to monitor the evolution of NaCrO₂ cycled between 2.2 ~ 3.6 V and 2.2 ~ 4.5 V^[21]. The electrochemical process was reversible for the cutoff voltages of 3.6 V (Figure 2(B)), whereas the discharge capacities decreased fast as the cutoff voltage rises to 4.5 V (Figure 2(D)). As shown in Figure 2(A), the in-situ EPR spectra for the first two cycles with an upper cutoff voltage of 3.6 V clearly reveals the good reversibility during the charge/discharge processes. It should be pointed out here that the EPR intensity at g = 1.962 decreased and vanished within 3 h (ca. 0.12 Na removed) when it starts charging (Figure 2(B)). The loss of the EPR signal, is possibly originated from the electronic delocalization within the [CrO₂] layers due to the overlap between the adjacent Cr t_2g orbitals. An increase of the EPR intensity after ca. 18 h was observed when fully discharged, implying the relocalization of the electrons and the Cr reduction to +3. The in-situ EPR spectra cycled between 2.2 and 4.5 V are shown in Figure 2(C). It could be observed that the signal at g = 1.962 was not recovered in the discharge process after charging to 4.5 V. An unexpected new signal was observed with g = 1.977 at the beginning of the voltage plateau with respect to the irreversible phase transition (Figure 2(D)).

EPR imaging (EPRI) was carried out on the in-situ cell in order to determine its spatial distribution of this unknown signal. It could be observed in Figure 3 that this signal was distributed in an area of 8 × 4 mm and a thickness of 1.5 mm, consistent with the size of the separator used in the cell. Therefore, this unknown species associated with the irreversible phase transition should be located in the electrolyte, which is then attributed to the Cr⁵⁺ ions dissolved with the surface reconstruction. This is the first report on the presence of soluble Cr⁵⁺ species from Cr-based cathode material. A disproportionation-dissolution mechanism is suggested here that the Cr disproportionation on the surface could take place as Cr³⁺ + Cr⁴⁺ ↔ Cr²⁺ + Cr⁵⁺ and then dissolution of Cr⁵⁺ into the electrolyte would lead to a densified rock-salt CrO as the reaction proceeds to the right side.

Ex-situ X-band EPR was employed to study lithiation/delithiation process of monocline Li₃V₂(PO₄)₃ calcinated from vanadium metal-organic framework precursor MIL-101(V)^[22]. All the EPR spectra were collected at the temperature of 2 K, as shown in Figure 4. In the perpendicular mode, V⁴⁺(t_2g¹e_g⁰, S = 1/2) is often observable while V³⁺(t_2g²e_g⁰, S = 1) is “EPR-silent” with no observable signal because its zero-field splitting in S = 1 ground state is usually larger than the available microwave quantum. V⁵⁺(t_2g⁰e_g⁰, S = 0) has no EPR signal with 3d⁰ configuration.

The pristine sample and the sample at 3.63 V showed a narrow signal at g ~ 2.00 corresponding to free electrons from the super-P conducting additives. When the sample was charged to 3.72 V or discharged to 3.56 V, a new peak could be observed at ~ 200 G, which arised from a new spin state S = 3/2 due to the ferromagnetic coupling between V³⁺ and V⁴⁺. Therefore, this peak at ~ 200 G suggests the existence of V⁴⁺ at this voltage. This ferromagnetic coupling also confirmed the existence of V³⁺/V⁴⁺ ordering for Li₂V₂(PO₄)₃ at the charging voltage of 3.72 V. Although V³⁺/V⁴⁺ ordering had been suggested by NMR result at ca. 3.72 V^[23], the EPR result is a stronger proof of V³⁺/V⁴⁺ ordering.

After charging to 4.12 V, the peak at ~ 200 G disappeared and a new peak at ~ 3426 G (g_eff ≈ 1.96) could be observed, suggesting that most of V³⁺ ions have been transformed to V⁴⁺ ions, which cancelled the ferromagnetic coupling and hence the peak at ~ 200 G vanished. Further charging to 4.60 V led to the reduction of V⁴⁺ signal, because V⁴⁺ was oxidized to V⁵⁺ which has no EPR signal. The EPR spectra of discharge process show a typical reversible feature with respect to the charge process, demonstrating the good reversibility of this Li₃V₂(PO₄)₃ sample during the charge/discharge processes. Using EPR technique, the valence states of V-ion were explicitly revealed; especially, the ferromagnetic coupling could be revealed.

Coordination polymers (CPs) or metal-organic fra-meworks (MOFs) constructed by metal ion nodes and bridging ligands, are emerging as a novel class of energy storage materials for rechargeable batteries. The electrochemical process of cobalt nitrilotriacetic acid CP nanorods (r-CoHNta) was analyzed by continuous-wave (CW) X-band EPR spectra recorded at 2 K, as shown in Figure 5^[24].

The intense sharp signal (g ≈ 2.00 at 340 mT) can be attributed to the free electron from super-P conductive additives. The EPR spectrum of the pristine r-CoHNta was dominated by a broad derivative lineshape with peak intensity around g ~ 4.3 at 130 mT, which is characteristic of rhombic high-spin (S = 3/2) Co²⁺ (t_2g⁵e_g²), disclosing that the Co-ions in pristine r-CoHNta are weak 6-coordination. The most interesting result is that Li-ion insertion into r-CoHNta (the curve of 0.01d in Figure 5(A)) led to a Dysonian-lineshape EPR absorption around ~ 312 mT (g ~ 2.07). It is well established that the Dysonian lineshape of EPR signal is the feature of delocalized conducting electron. The delocalized conducting electron might come from delocalized high-spin Co²⁺ or from Co-metal clusters. Therefore, this result suggests that the high spin Co²⁺ releases some electron spins from the localized electrons after Li⁺ intercalation. Several MOF anodes were also investigated by EPR techniques^[25,26,27].

Our group has studied the series samples of Na₃V₂(PO₄)₂F_3-2yO_2y(y = 1.0, 0.8, and 0.6), produced by a microwave-assisted solvothermal procedure^{[28, 29]}. The three samples were called as NVOPF, NV_3.8OPF and NV_3.6OPF, among which the average valence states of V-ion are +4, +3.8 and +3.6, respectively. The NVOPF had only one ²³Na peak at 91.5 ppm (Figure 6(A)), suggesting that the synthesis method could eliminate, to a great extent, the defects, which was observed previously by Park et al^[30]. For NV_3.8OPF, two major peaks at 89.4 and 128.7 ppm were observed, which comes from Na⁺ close to V⁴⁺ and V³⁺ ions, respectively, considering that paramagnetic hyperfine interaction of V³⁺ to Na is stronger compared to V⁴⁺ ions. The ²³Na MAS NMR spectrum of NV_3.6OPF was broader, implying the stronger paramagnetic environment due to more V³⁺ ions generated. The ³¹P NMR spectra also demonstrated the broadening effect of the V³⁺ ions (Figure 6(B)).

Continuous-wave (CW) X-band EPR spectra were recorded under 2 K to further interpret the valence state of V-ions. The NVOPF exhibited a featureless first-differential spectrum centered at g-value of 1.93 with ~ 100 G peak-to-trough (Figure 6(C)). This g-value was close to that of free electron (g_e = 2.0023), suggesting the existence of V⁴⁺ with electron configuration t¹_2ge⁰_g (S = 1/2). However, the EPR signal intensities of NV_3.8OPF and NV_3.6OPF were much weaker than that of NVOPF (namely, NVOPF > NV_3.8OPF > NV_3.6OPF), indicating a lower content of V⁴⁺ centers in these systems. Here the signal observed only came from V⁴⁺, since V³⁺ (t²_2ge⁰_g, S = 1) is “EPR-silent” as it is a non-Kramer's ion with integral J in LS coupling. In addition, the EPR peak of NV_3.8OPF was much broader than that of NVOPF (Figure 6(D)), suggesting the presence of V³⁺ centers that significantly broaden the peak of V⁴⁺ due to strong electron dipole-dipole couplings between V³⁺ and V⁴⁺.

In-situ²³Na NMR spectra of Na/NV_3.8POF cell cycled at a current rate of 0.05C during the first charge and discharge processes were first acquired. Due to the low resolution the in-situ static NMR spectra, ²³Na MAS spectra were then acquired^[29]. For the pristine NV_3.8OPF (Na₃V₂(PO₄)₂F_1.4O_1.6) sample, only Na⁺ signals close to V³⁺ and V⁴⁺ were observed at ~ 130 and ~ 90 ppm. After the extraction of two Na ions from Na₃V₂(PO₄)₂F_1.4O_1.6, Na₁V₂(PO₄)₂F_1.4O_1.6 with average covalence state of 4.8 was produced, having two Na peaks neighboring to V⁴⁺ and V⁵⁺ at ~ 90 and ~ 6.5 ppm, respectively. To confirm V⁵⁺ ion being existed in this system, the ⁵¹V MAS ssNMR spectrum of Na₁V₂(PO₄)₂F_1.4O_1.6 was acquired, having an obvious signal located in the region of [-550, -720] ppm.

The ²³Na MAS ssNMR spectra for the charge process of NV_3.8POF (Figure 7(A)) could be grouped into two different stages (I and II): In Stage I, the main peak at ~ 90 ppm showed a continuous shift to lower frequencies, and this symmetric lineshape became asymmetric with apparent shoulder peaks at the right side. In Stage II, this main peak showed an shift back to higher frequency. Such a phenomenon was not observed in the NV₄POF system. The NMR result of NV_3.8POF in the early stage II is counterintuitive, because it was believed that upon Na extraction, V³⁺ should be converted to V⁴⁺ and V⁴⁺ should be converted to V⁵⁺, leading to an overall shift to lower frequency. This observation could be explained by EPR measurements. Parallel-mode EPR revealed two dispersion peaks at g = 4.01 and g = 1.93 (Figure 7(B))^[29]. The signal at g = 4.01 indicates the transition between |+1> and |-1> doublet of an S = 1 spin state, which can be ascribed to V³⁺ with t²_2ge⁰_g configuration. It should be mentioned that S = 1 configuration could not be observed with perpendicular-mode EPR. The increase of the V³⁺ content in Na₂V₂(PO₄)₂F_1.4O_1.6 (Na0.48c, V^4.3+ covalence state) was obtained as compared with that in bare Na₃V₂(PO₄)₂F_1.4O_1.6, which inferred that during the Na-ion extraction, the charge disproportionation of two V⁴⁺ to a mixture of V³⁺ and V⁵⁺ possibly occurred in the stage II, which leads to abnormal phenomenon in the ²³Na NMR spectra in Stage II. In the charged NV₄POF sample, that is, Na₂V₂(PO₄)₂F₁O₂, no charge disproportionation was observed and no generated V³⁺ as suggested by the NMR spectra.

Another skin system boron-doped Na₃V₂(PO₄)₃ had also been studied by NMR and EPR with similar techniques^[31]. EPR results under 1.8 K confirmed the generation of V²⁺ with rhombohedral distortion upon the fourth Na⁺ intercalation process of Na₃V₂(PO₄)₃ at low voltage of ~ 1.6 V, which is further confirmed by ²³Na ssNMR experiments. The results from NMR and EPR suggested that the boron substitution can widen the range of solid-solution reaction, facilitate the structural transformation toward V²⁺-containing phase, and mitigate the short scale heterogeneity of P and Na nuclei^[31].

Two cathode materials, Na_0.72Li_0.24Mn_0.76O₂ (NLMO) and Na_0.72Li_0.24Mn_0.66Ti_0.1O₂ (NLMTO-0.1) were studied by NMR and EPR^[32]. Cyclic charge/discharge studies show that the stability of Ti-doped NLMTO-0.1 far exceeded NLMO. The ⁷Li pj-MATPASS ssNMR spectra (Figure 8) were used to probe the structural evolution. For pristine NLMO (Figure 8(C)), the typical resonance observed at ~ 746 ppm can be attributed to Li ions in the Na layer, whereas the resonances centered at ~ 1490 and ~ 1850 ppm can be assigned to Li sites in a honeycomb-like arrangement within the TMO₂ layer. In particular, after Ti-doping, most of the peaks for NLMO-0.1 had been largely broadened (Figure 8(A)), due to more distorted coordination environments. Figure 8(B) shows that when Ti-doped NLMTO-0.1 was charged to 4.5 V and then discharged to 1.5 V, Li would return to the MnO₂ layer (the yellow region in Figure 8). However, when the NLMO sample went through the same process, Li would prefer to stay in the Na layer (the blue region in Figure 8). It should be pointed out here that the Li staying in the Na layer will lead to a large number of vacancies in the MnO₂ layer. The existence of vacancies makes Mn tend to move and cause its structural instability.

The EPR experiments (Figure 9) showed that in these two systems, the O₂^n- signal with an effective g-value of 2.007 would appear after charging to 4.2 V, but the signal evolution of NLMTO-0.1 became more reversible. Especially, the EPR signal of NLMTO moved along the x-axis upon charging and discharging, while that of NLMTO-0.1 stayed at the similar position, further showing the better reversibility of oxygen redox reaction in the NLMTO-0.1 system. The combined use of NMR and EPR shows that Ti doping can make Li return to the MnO₂ layer, and then make the oxygen reaction more reversible, thereby increasing the stability of the cathode materials.

Similar techniques were also applied to Sn/Zr-substituted P2-Na_0.66Li_0.22Mn_0.78O₂ system^[33]. The ²³Na ssNMR spectra shows that when charged to 4.5 V, the ²³Na signal would be changed from a sharp peak at ~ 1708 ppm to a broad peak ranging from 1560 to -300 ppm, signifying a considerable disordering around the residual Na ions, which further confirms the generation of stacking faults at high voltage of 4.5 V that disrupt the long-range order. The ⁷Li ssNMR after long 50 cycles at low voltage of 1.5 V suggested the gradual loss of the ⁷Li in the MnO₂ layer. EPR also showed the existence of O₂^n- signal. NMR and EPR results validate that (i) Li migrates from TM layer to AM layer upon charge, and do not fully return to the original sites in TM layer upon discharge; (ii) the extraction/insertion of Na leads to the local structural transition which is not fully reversible; (iii) Li/Mn migrations keep going, leading to irreversible Li/Mn losses and Na-site local disorder upon prolonged cycling.

We also used EPR and NMR to study the changes in the local atomic environment and electronic structure of the cation-disordered cathode material Li_1.2Ti_0.4Mn_0.4O₂ during the first cycle of charge and discharge^[34]. It was found by ⁷Li ssNMR that Li ions were preferentially extracted from the Li-rich environment, which forms 0-TM percolation channel. However, the irreversible loss of Li in the diamagnetic environment after one cycle was observed in ⁷Li NMR, indicating that the 0-TM channel was partially destroyed. The ¹⁷O NMR reveals that an anomaly occurred in both ⁷Li and ¹⁷O shifts at the end of charging to 4.8 V, since a decrease of the ¹⁷O peak shift and a steep rise in the ⁷Li weighted average shift were observed. Such an anomaly implies the oxygen redox reactions.

The EPR spectrum (Figure 10) of pristine Li_1.2Ti_0.4-Mn_0.4O₂ had no signal, however, upon charging, a prominent broad line resonating at ca. 90 mT (g ~ 7.5) was observed, which arises from the ferromagnetic resonance. This result suggests the presence of short-range ferromagnetism due to the local clusters of Mn ions. At the beginning of oxygen oxidation (C230, ~ 4.3 V), a new weak EPR signal appearing at g ~ 2.0 is assigned to (O₂)^n- species. However, this (O₂)^n- signal disappeared for C310 (~ 4.5 V) as oxidation proceeded. When charging to a high voltage of 4.8 V (C358), a signal analogous to that of the (O₂)^n- species appeared with stronger intensity, which should be ascribed to the localized electron holes on the lone-pair states of O whose other 2p orbitals are bonded to Mn. Furthermore, the results of Fast Fourier transforms (FFTs) of the TEM images suggested that short-range order (SRO) occurs when charging begins and the TM cation migration takes place during O oxidation to relax the distorted lattices, resulting in a local disorder-to-order transition, which deteriorates the diffusion kinetics of Li ions.

The combined use of NMR and EPR confirmed the migration and the aggregation of Mn ions, together with the change of 0-TM environment and oxygen environment during charge and discharge, which leads to an undesirable voltage hysteresis and low coulombic efficiency.

To conclude, ssNMR/EPR is a powerful tool to investigate electrode materials. NMR could probe the environment evolution of different kinds of nuclei with high-resolution, such as ⁷Li, ²³Na, ¹⁹F, ¹⁷O and especially ⁵⁹Co, providing direct evidence for structural evolution together with the subtle information of phase transition. EPR could probe different kinds of metal ions (e.g., V, Co, Ni, Mn, Fe) containing unpaired electrons and several metal electrodes (e.g., Li, Na) with high sensitivity. It could be employed to characterize the electronic state and the redox state of metal ions. Table 1 summarizes different NMR/EPR techniques and their applications. The explicit physical background could be found in the ref.^[35]. It should be pointed out here that as the electrode material system becomes more and more complex, the advance NMR/EPR techniques, together with DFT (Density Functional Theory) calculations are urgently needed. Our group has used VASP and Quantum Espresso (QE) software together to calculate ¹⁷O NMR and EPR parameters (e.g., g-value), laying a foundation for the explanation of NMR and EPR signals^[36]. Readers could read DFT-related literatures for details^{[37,38,39,40,41]}. The results from the combination of NMR and EPR have important implications for the interpretation of the working mechanism and structure-activity relationship.

Above all, these investigations might pave the way towards better understanding of the structural evolution mechanism and the design of the stable high-capacity electrode materials in the future. More application using NMR and EPR techniques to electrode materials is expecting.

Authors are grateful for the funding supported by National Natural Science Foundation of China (No. 21872055, No. 21874045), Shanghai Science and Technology Innovation Action Plan (No. 19142202900) and Fundamental Research Funds for Central Universities and Open Foundation of ECNU (No. 42125102).