Dual‐strategy modification on P2‐Na0.67Ni0.33Mn0.67O2 realizes stable high‐voltage cathode and high energy density full cell for sodium‐ion batteries

P2‐type Na0.67Ni0.33Mn0.67O2 is considered as a potential cathode material for sodium‐ion batteries due to the merits of high voltage, low cost, and air stability. However, the unsatisfied cycling stability and rate performance caused by the destructive phase transition and side reactions hinder its practical application. Herein, we present a feasible dual strategy of Mg2+ doping integrated with ZrO2 surface modification for P2‐Na0.67Ni0.33Mn0.67O2, which can well address the issues of phase transition and side reactions benefitting from the enhanced structural and interfacial stabilities. Specifically, it exhibits a decent cycling stability with a capacity retention of 81.5% at 1 C and promising rate performance with a discharge capacity of 76.6 mA h g−1 at 5 C. The in situ X‐ray diffraction measurement confirms that the damaged P2–O2 phase transition is suppressed with better reversibility in high‐voltage region, whereas the side reactions are inhibited due to the protective ZrO2 surface modification. Commendably, the full cell achieves an outstanding operating voltage of 3.57 V and a fabulous energy density of 238.91 W h kg−1 at 36.73 W kg−1, demonstrating great practicability. This work is expected to provide a new insight for designing stable high‐voltage cathode materials and high energy density full cells for sodium ion batteries.

portable electronic equipment and electric vehicles due to the merits of high energy/power density and appreciable lifespan. [8][9][10] However, recent concerns about the cost and abundance of lithium resources compel researchers to develop suitable alternatives of LIBs. 11 Among the numerous candidates, sodium-ion batteries (SIBs) have been placed great expectations by the virtue of the high abundance of sodium resources in the crust as well as the identical working principle to LIBs, which have been considered the most prospective candidates for EES application. [12][13][14] However, the development of SIBs has been seriously hindered due to the lack of suitable cathode materials. Therefore, it is of great significance to develop satisfactory cathode materials with decent capacity, high operating voltage, and excellent cycling stability. [15][16][17] Up to now, various compounds, such as layered transition metal oxides (LTMOs), polyanions and Prussian blue analogues, have been widely investigated as promising cathodes for SIBs. Among these cathodes, sodiumbased layered oxides (Na x TMO 2 , 0 < x ≤ 1, TM = Mn, Ni, Fe, Co, Cr, etc.) have attracted tremendous attention because of their high theoretical capacity and plain synthesis process. 18 According to the arrangement of transition metal layers and the coordination environment of sodium ions, LTMOs are usually divided into two types: O3 type and P2 type. 19 In general, P2-type materials exhibit better cycling and rate performance because of their open frameworks as well as better moisture resistance. 20 Specially, P2-type Na 0.67 Ni 0.33 Mn 0.67 O 2 has been extensively studied as cathode materials for SIBs due to the features of high theoretical specific capacity (173 mA h g −1 ) and high average operating voltage (∼3.6 V) from the redox couple Ni 2+ -Ni 4+ with two electron transfers. 21,22 However, when charged above 4.2 V, it suffers from an irreversible P2-O2 phase transition with large volume change (∼23%), leading to dramatic capacity fading and voltage decay. 23,24 Moreover, the electrolyte consumption and transition metal dissolution caused by the side reactions between cathode and electrolyte also aggravate the capacity and voltage decay. [24][25][26] To solve these problems, many works, such as ion doping, surface coating, nano engineering, have been reported. 25,[27][28][29] For instance, Chen's 30 group developed an Mg 2+ doped Na 0.66 Mn 0.95 Mg 0.05 O 2 cathode, in which a prominently improved electrochemical performance of 82% retention after 50 cycles under ultrahigh voltage of 4.5 V can be successfully achieved. Guo's 31 group developed a Ti 4+ -doped Na 2/3 Ni 1/3 Mn 1/3 Ti 1/3 O 2 , and they found the Na + /vacancy ordering can be inhibited through Ti 4+ doping. Wang's 32 group developed an Li + doped Na 0.85 Li 0.12 Ni 0. 22 Mn 0. 66 O 2 cathode and realized the complete solid-solution reaction within the voltage range of 2.0-4.3 V. Sun's 33 group sys-tematically studied the effect of Al 2 O 3 , ZrO 2 , and TiO 2 coating on P2-type Na 0.66 (Mn 0.54 Co 0.13 Ni 0.13 )O 2 cathode via atomic layer deposition method, in which they found that Al 2 O 3 coating can significantly improve the cyclic stability, whereas ZrO 2 is more effectively in improving the rate performance, providing valuable guidance for follow-up works. However, it is still unsatisfactory to meet the practical application in terms of capacity and voltage stability with just single modification method. Based on the existing studies, ion doping has been proved effective to suppress the damaged phase transition, whereas surface modification can play a positive role in inhibiting side reactions. Therefore, designing a dual strategy of ion doping integrated with surface modification is believed can combine the advantages of both aspects simultaneously, thus achieving better electrochemical performance for both half and full cells. However, up to now, only a very small number of works are reported, [34][35][36] which means it is of importance to carry out studies with dual-strategy modification.
Among the various doping ions and coatings, Mg 2+ doping shows better cycling stability than other ions as well as the lighter atomic mass that is believed to be helpful to improve the theoretical specific capacity. ZrO 2 possesses excellent fracture toughness and electrical conductivity of the metal oxide coating layers, which can buffer inherent stress and strain as well as suppress the side reactions between cathode and electrolyte. 33 Herein, we report a stable high-voltage cathode Na 0.67 Ni 0.28 Mg 0.05 Mn 0.67 O 2 with ZrO 2 surface modification (NNMMO-Zr). By the virtue of the synergistic effect of Mg 2+ doping and ZrO 2 surface modification, the destructive phase transition and side reactions between cathode and electrolyte are well suppressed. Such a dual-strategy modification significantly improves the electrochemical performance of Na 0.67 Ni 0.33 Mn 0.67 O 2 . Specifically, the NNMMO-Zr delivers a decent capacity retention rate of 84.7% after 50 cycles at 0.1 C and 81.5% after 150 cycles at 1 C (1 C = 173 mA g −1 ) as well as a prominent rate performance of 76.6 mA h g −1 at 5 C. In situ X-ray diffraction (XRD) measurement shows that the damaged P2-O2 phase transition is suppressed and transforms into a P2-OP4 phase transition with good reversibility when charged to high voltage, whereas the side reactions between cathode and electrolyte are also inhibited due to the protective ZrO 2 surface modification at the same time. What is more, beyond the significantly improved electrochemical performance in half-cell, the full cell can output an outstanding operating voltage of 3.57 V and a fabulous energy density of 238.91 W h kg −1 at the power density of 36.73 W kg −1 based on the NNMMO-Zr cathode and commercial hard carbon (HC) anode and can still maintain extraordinary cell voltage of 3.11 V and energy density of 106.73 W kg −1 even at the high power F I G U R E 1 Structural and morphological characterizations of NNMMO-Zr: (A) schematic illustration of synthesis process; (B) Rietveld refinement X-ray diffraction (XRD) pattern; (C) schematic diagram of P2-type crystal structure; (D) field-emission scanning electron microscopy (FESEM) image; (E-G) HRTEM image and corresponding lattice fringes; (H) elemental mapping results density of 1601.04 W kg −1 , which is superior to most LTMOs cathode materials.

RESULTS AND DISCUSSION
The Na 0.67 Ni 0.28 Mg 0.05 Mn 0.67 O 2 with ZrO 2 surface modification sample (denoted as NNMMO-Zr) is synthesized by a citric acid sol-gel method followed by a simple wet chemical method, as illustrated in Figure 1A. First, the Na 0.67 Ni 0. 28  Powder XRD and Rietveld refinement are employed to investigate the crystallographic structure of the synthesized materials, as shown in Figures 1B and S1 and S2. All peaks of NNMMO-Zr can be well indexed to the hexagonal P2-type layered structure with the space group of P6 3 /mmc (JCPDS-54-0894) with slight peaks of NiO impurity (Fm3m, JCPDS-89-7131) attributed to the limited solubility of Ni 2+ in P2 phase. 37 The control samples of NNMO and NNMMO also exhibit the identical XRD patterns, indicating that the Mg 2+ doping and a small amount of ZrO 2 surface modification will not affect the crystal structure ( Figure S1). It is worth to mention that the characteristic peaks of ZrO 2 cannot be observed in XRD patterns, which may be attributed to the low content of ZrO 2 . Moreover, Table S1 shows the lattice parameters of three samples obtained from Rietveld refinement XRD. It can be observed that the lattice parameters of NNMMO-Zr and NNMMO are very close but increase in both a and c compared with that of NNMO, which may be ascribed to the larger ionic radius of Mg 2+ (0.72 Å) compared with Ni 2+ (0.69 Å). 27 The detailed structure schematic diagrams of P2-type layered oxide ( Figure 1C) show that the transition metal layers (TMO 2 ) are formed by the transition metal-oxygen (MO 6 ) octahedrons shared edges with each other and the Na-ions reside in the prismatic sites to form the Na layers. Field-emission scanning microscopy F I G U R E 2 Chemical states analysis of NNMMO-Zr by X-ray photoelectron spectroscopy (XPS) measurement: (A) survey spectrum; high-resolution spectra of Ni 2p (B), Mn 2p (C), Mg 1s (D), and Zr 3d (E) (FESEM) image of NNMMO-Zr ( Figure 1D) shows that a typical plate-like morphology can be observed, and some isolated particles can be detected compared with control samples of NNMO and NNMMO ( Figure S3), which can be further confirmed in TEM image ( Figure 1E), implying the successful surface modification with ZrO 2 . Moreover, the distance of the lattice fringes of the spheral-like substance ( Figure 1F) is measured to be about 2.84 Å, which can be well indexed to the (1 1 1) planes of ZrO 2 , whereas the lattice fringes of the bulk material ( Figure 1G) exhibit an interlayer distance of 2.34 Å corresponding to the (0 1 2) planes of NNMO, which provide direct evidence for the successful introduction of ZrO 2 at surface. The elemental mapping results ( Figure 1H) also suggest the notable signal of Zr element besides the Na, Ni, Mg, Mn and O elements.
The X-ray photoelectron spectroscopy (XPS) is carried out to verify the valence states of the synthesized materials. The survey spectrum of NNMMO-Zr ( Figure 2A) shows that all the elements of Na, Ni, Mg, Mn, Zr, and O exist. The peak at 1070.8 eV in Na 1s spectrum ( Figure  S4) can be attributed to the typical binding energy of Na + in layered structure. 38 Ni 2p high-resolution spectrum is shown in Figure 2B, in which the characteristic peaks at the binding energy of 872.2 and 854.7 eV can be attributed to the Ni 2p 1/2 and Ni 2p 3/2 , and their corresponding satel-lite peaks are located at 878.7 and 861.1 eV, respectively, indicating the divalent chemical state of nickel. 39,40 The Mn 2p spectrum is shown in Figure 2C, in which the peaks at 653.8 and 642.3 eV belonging to the Mn 2p 1/2 and Mn2p 3/2 , indicating that the valence state of manganese is tetravalent. 41,42 Figures 2D and S5A,B show the Mg 1s high-resolution spectrum of NNMMO-Zr, NNMMO, and NNMO; it is distinguished that there is no signal in the NNMO sample, whereas it is obvious in the NNMMO and NNMMO-Zr samples, indicating the successful doping of Mg 2+ . The high-resolution spectrum of Zr 3d in NNMMO-Zr sample ( Figure 2E) shows two obvious peaks at 184.4 and 182.0 eV corresponding to the Zr 3d 3/2 and Zr 3d 5/2 of ZrO 2 , 43 respectively, whereas it cannot be detected in NNMMO and NNMO samples ( Figure S5C,D), indicating the existence of ZrO 2 in NNMMO-Zr. Moreover, the high-resolution spectra of Ni 2p and Mn 2p in the control samples ( Figure S6) indicate that their valence states are also divalent and tetravalent, respectively, proving that Mg 2+ doping and ZrO 2 surface modification will not change the chemical state of Ni and Mn ions.
The comparing investigations on electrochemical performances of NNMMO-Zr, NNMMO, and NNMO are carried out to illustrate the synergistic effects of Mg 2+ doping integrated with ZrO 2 surface modification. The F I G U R E 3 Electrochemical performances in half-cell configuration: (A) galvanostatic charge/discharge curves of NNMMO-Zr, NNMMO, and NNMO samples at 0.1 C; (B) galvonostatic charge/discharge curves of NNMMO-Zr from 1st to 50th cycles at 0.1 C; (C) comparing cycling performance at 0.1 C; (D) rate performance with different current densities ranging from 0.1 to 5 C; and (E) comparing cycling performance at a relatively high rate of 1 C for NNMMO-Zr, NNMMO, and NNMO samples, respectively charge/discharge curves of the three samples at 0.1 C (1 C = 173 mA g −1 ) are shown in Figure 3A. It can be clearly observed that the plateau in high-voltage region above 4.2 V in NNMMO-Zr and NNMMO is significantly shortened compared with that of NNMO, which may be ascribed to the suppressed P2-O2 phase transition. 23,32 The specific capacity of NNMO, NNMMO, and NNMMO-Zr are 143.3, 124.5, and 121.9 mA h g −1 , respectively, where the decreasing trend after Mg 2+ doping could be due to the replacement of electrochemical active Ni 2+ by the inactive Mg 2+ , 27,44 and further deduction of capacity upon ZrO 2 surface modification could be derived from its electrochemical inactivity. The first 50 galvano-static charge/discharge curves at 0.1 C for NNMMO-Zr ( Figure 3B) and control samples of NNMMO and NNMO ( Figure S7) reflect the better structural stability of NNMMO-Zr. Furthermore, the cycling performance tests are performed in half-cells at 0.1 C to illustrate the positive effects of the dual-strategy modification, as shown in Figure 3C. Specifically, the NNMMO-Zr sample delivers the best capacity retention rate of 84.7% after 50 cycles, whereas those for NNMMO and NNMO samples are 74.7% and 48.6%, respectively. This phenomenon may be attributed to the enhanced structural stability after Mg 2+ doping and suppressed side reactions due to the protective ZrO 2 surface modification, 27,45 indicating the feasibility of dual-strategy modification for stabilizing the P2-type NNMO. Besides, the corresponding average voltage during cycling ( Figure S8) shows that the NNMMO-Zr sample exhibits a high operating voltage of 3.67 V at 0.1 C in half-cell configuration. To further investigate the suitable content of ZrO 2 surface modification, the cycling performances of the samples with different ZrO 2 surface modification contents at 0.1 C are also provided in Figure  S9, in which the samples with 1 wt% ZrO 2 content exhibit best performance. Moreover, the NNMMO-Zr can deliver a gorgeous rate performance ( Figures 3D and S10) with a high capacity retention rate of 62.9% from 0.1 to 5 C, which can be well recovered as high as 98% of the original capacity when the current density is reduced to 0.1 C. The comparing cycling performance at a relatively high current density of 1 C ( Figure 3E) intuitively demonstrates the unique merit of dual-strategy manipulation, where an outstanding capacity retention rate of 81.5% after 150 cycles can be achieved in NNMMO-Zr. Moreover, the comparing electrochemical performance of NNMO, NNMMO, and NNMMO-Zr samples is also studied at a higher cutoff voltage of 4.5 V, which is provided in Figure S11, in which NNMMO-Zr also exhibits the enhanced performance, displaying a good specific capacity of 126.9 mA h g −1 at 0.1 C as well as a decent cycling stability of 66.8% after 150 cycles at 1 C.
In situ XRD measurement is carried out to investigate the structural evolution of the NNMMO-Zr sample, as shown in Figure 4. The peaks marked with asterisk are attributed to the Al foil, Be/BeO window, and in situ cell parts, which will not change during charge/discharge positions. Parts A and B of Figure 4 reveal the evolution process of XRD patterns for NNMMO-Zr during the first charge/discharge process. During the charge process, the (0 0 2) and (0 0 4) peaks gradually shift to lower angle, indicating the expansion of the lattice parameter c during the Na + extraction, which is mainly caused by the increasing electrostatic repulsive force between TMO 2 layers when Na + content is reduced. 46 Moreover, a new peak at about 16.5 • appears when charged to high-voltage region, which can be ascribed to the emergence of OP4 phase. 44,47,48 In the subsequent discharge process, the peaks gradually recover to the P2 phase as pristine state, implying the good reversibility of P2-OP4 phase transition in the NNMMO-Zr electrode. 27,44 Contrastively, a new peak at about 20 • appears in the NNMO electrode when charged in the highvoltage region ( Figure S12A), indicating the P2-O2 phase transition occurs. The interlayer distance of O2 phase is smaller than that of P2 phase, leading to a large volume contraction of about 23%, which is responsible for the dramatic capacity decay. 24,49 By contrast, the OP4 phase is an intergrowth structure of O-type and P-type layers, which is closer to P2 structure compared with O2 phase; thus, the lattice mismatch and volume change is much smaller for P2-OP4 phase transition than that of P2-O2 phase transition, resulting in an enhanced cycling stability. 44,50,51 In addition, to determine whether Mg 2+ doping or ZrO 2 surface modification inhibits the destructive P2-O2 phase transition, the ex situ XRD measurement is carried out on NNMO-Zr (NNMO with 1 wt% ZrO 2 surface modification) and NNMMO samples, as presented in Figure S12B,C. It can be observed that the new peak at around 20 • attributed to O2 phase can be also detected in NNMO-Zr sample when charged into high-voltage region, which is same as that of NNMO sample. However, this peak does not appear in an NNMMO sample while the (0 0 2) peak of P2 phase can still maintain even under the high voltage of 4.3 V, indicating the enhanced structural stability can be achieved after Mg 2+ doping. These results demonstrate that Mg 2+ doping plays the dominant role in inhibiting the P2-O2 phase transition, which is consistent with other previous works. 27,44,52 The structural evolution of the NNMMO-Zr cathode during charge/discharge process can be schematically demonstrated as Figure 4C, the damaged P2-O2 phase transition is effectively constrained, and a P2-OP4 phase transition with better reversibility occurs.
In order to further investigate the possible origins of the enhanced electrochemical performance for NNMMO-Zr, cyclic voltammetry (CV) and galvanostatic intermittent titration technique (GITT) measurements are carried out to characterize the kinetic process. The comparing CV curves from first to fourth cycles of NNMO, NNMMO, and NNMMO-Zr samples in the voltage range of 2.5-4.35 V are shown in Figure 5A-C, respectively.
It can be seen that the CV curves of NNMMO and NNMMO-Zr samples are boarder and smother than that of NNMO, implying that Mg 2+ doping can suppress the Na + /vacancy ordering, which is unfavorable to Na + diffusion. 31 53 Moreover, the peaks concerning phase transition in the NNMMO-Zr sample overlaps better than that of NNMMO and NNMO samples, indicating the structural stability is improved after dual-strategy modification, which is consistent with the electrochemical tests results. Besides, Multi-sweep CVs are performed to study the Na + diffusion rate in all three samples ( Figure S13). The calculated average diffusion rate of Na + is 6.95 × 10 −11 cm 2 s −1 in the NNMMO-Zr sample, which is much larger than that of NNMO, indicating an enhanced kinetic process due to the dual-strategy modification. Furthermore, it is indicated that the apparent diffusion coefficient of Na + can be calculated by the following simplified equation if the functional relationship between E and τ 1/2 is linear 54,55 :  Figure S14), the calculated diffusion coefficient of Na + ( Figure 5H-I) clearly suggests that the NNMMO-Zr and NNMMO samples exhibit similar Na + diffusion coefficient, but much larger than that of NNMO sample, indicating that after Mg 2+ doping, the kinetic process is enhanced, which is consistent with other works. 23,44 To characterize the optimized interface between cathode and electrolyte introduced by the ZrO 2 surface modification, the SEM and the electrochemical impendence spectra (EIS) before and after cycling are employed. The morphology ( Figure 6A-C) of NNMO, NNMMO, and NNMMO-Zr samples after 150 cycles at 1 C suggest that a large amount of cracks can be observed in the NNMO sample, whereas in the NNMMO sample, the number of cracks decreases significantly, and cracks almost cannot be observed in the NNMMO-Zr sample. The occurrence of cracks is mainly relative to two reasons: the volume change concerning the phase transition during charge and discharge process and the side reactions between cathode and electrolyte during cycling. 11,25,56 In detail, the volume change caused by phase transition will generate stress within the particle, and when the stress is larger than atomic bonding strength, the cracks generate. On the other hand, the side reactions between cathode and electrolyte that dissolves the transition metal ions will promote the growth of cracks and make them deeper. After Mg 2+ doping, the P2-O2 phase transition with large volume change transforms into the P2-OP4 phase transition with much less volume change, as mentioned above; thus the number of cracks is greatly reduced. In addition, after ZrO 2 surface modification, the number of cracks further reduced, indicating that the suppressed side reactions between cathode and electrolyte are thanks to the protection of ZrO 2 surface modification. The EIS evolution results of NNMO, NNMMO, and NNMMO-Zr at different cycles are shown in Figure 6D-F, respectively. In addition, an equivalent circuit applied to describe the interface is shown in Figure S15, in which R e represents the solution resistance, R SEI represents the Na + migration resistances through the SEI films formed on the cathode surface (corresponding to a semicircle in the high-frequency region in Nyquist plot), and R ct represents the charge transfer resistance (corresponding to a semicircle in the medium frequency region in Nyquist plot). 57,58 It can be clearly observed that the NNMMO-Zr exhibits the smallest resistance of the three samples, indicating best charge transfer process in NNMMO-Zr, which is consistent with the above kinetic analysis. It is worth to mention that the resistances in all samples decreased after five cycles, according to the previous report 23,59 ; this phenomenon can be attributed to an activation process of the electrode due to the stabilization of the SEI film and the gradual formation of channels for sodium-ion transfer during charge/discharge process. After 50 cycles, the resistance of NNMO and NNMMO increases fast. However, the resistance in the NNMMO-Zr sample exhibits much less increase rate; especially in R SEI , it can be witnessed that the R SEI almost unchanged after 50 cycles in NNMMO-Zr sample, whereas it increases obviously in NNMMO sample, implying that the ZrO 2 surface modification can significantly suppress the side reactions between cathode and electrolyte and thus inhibiting the growth of SEI. Detailed fitted resistance values of all three samples are shown in Figure S16. XPS is employed to further investigate the surface compositions of the electrodes based on NNMMO-Zr, NNMMO ( Figure 6G-I), and NNMO ( Figure S17) after cycling at 0.1 C. In the C 1s spectra, four peaks of C-F (291.08 eV), C=O (288.51 eV), C-O (286.13 eV), and C-C (284.8 eV) can be divided. 35,56 It can be observed that the intensity of C-O and C=O peaks corresponding to the decomposition of carbonate solvent in the NNMMO sample is higher than that of in the NNMMO-Zr sample. Besides, in O 1s spectra, three peaks of C-O (533.41 eV), C=O (531.55 eV), and lattice oxygen (529.9 eV) can be divided, 35,60 and it can be clearly observed that the peak intensity of lattice oxygen is higher in the NNMMO-Zr sample, whereas the intensity of C-O and C=O peaks is lower than that of the NNMMO sample, suggesting severer side reactions occur in NNMMO. Moreover, in F 1s spectra, two peaks of C-F (688.09 eV) associated F I G U R E 6 SEM images of NNMO (A), NNMMO (B), and NNMMO-Zr (C) after 150 cycles at 1 C; electrochemical impendence spectra (EIS) evolution at 1st, 5th, and 50th cycles collected at 0.1 C of NNMO (D), NNMMO (E), and NNMMO-Zr (F); X-ray photoelectron spectroscopy (XPS) spectra of C 1s (G), O 1s (H), and F 1s (I) for NNMMO and NNMMO-Zr electrode after cycling with polyvinylidene fluoride (PVDF) and Na-F (685.38 eV) result from the decomposition of fluoroethylene carbonate (FEC) can be observed. 35 The intensity of Na-F peak in NNMMO is higher than NNMMO-Zr, indicating the FEC decomposition is suppressed in NNMMO-Zr. As for the NNMO electrode, it shows similar peaks as that of NNMMO, as shown in Figure S17. Detail information of peaks content in three samples are provided in Table S2. From the above analysis, we can conclude that ZrO 2 surface modification can play a positive role in inhibiting the side reactions and optimize the interface between cathode and electrolyte, which is consistent with other studies relative to oxide surface modification. 43,61,62 Inspired by the high operating voltage and decent cycling stability in half-cell, the full-cell device based on the NNMMO-Zr cathode and commercial HC anode (denoted as NNMMO-Zr//HC) is configured to demonstrate the potential in practical application, as schematically shown in Figure 7A. According to the charge balance principle and their electrochemical properties ( Figure 7B,C), the mass ratio of cathode to anode is 1.4-1.6, and the voltage range is 2.4-4.25 V for full-cell test. Figure 7D,E shows the rate performance and the galvanostatic charge/discharge curves of the NNMMO-Zr//HC full cell, in which a high specific capacity of 112.7 mA h g −1 at 0.1 C can be obtained, and it can still deliver a specific capacity of 58.6 mA h g −1 at 5 C (based on the cathode mass). It is worth to mention that the specific capacity is 76.6 mA h g −1 at 5 C in half-cell, which is higher than that of in full cell; the specific capacity difference between full cell and half-cell is owing to the poor rate performance of HC, as shown in Figure S18. Thus, it is of great F I G U R E 7 Electrochemical performance evaluation of NNMMO-Zr//HC full cell: (A) schematic diagram of NNMMO-Zr//HC full cell; cyclic voltammetry (CV) (B) and galvanostatic charge/discharge curves (C) of NNMMO-Zr cathode and hard carbon anode; (D) rate performance of full-cell device at the current density range from 0.1 to 5 C; (E) corresponding galvanostatic charge/discharge curves at different rates; (F) corresponding Ragone plot; (G) cycling performance of full-cell device at the current density of 0.1 C; (H) cycling performance of the full-cell device first cycled 10 cycles at 0.1 C and then cycled 100 cycles at 1 C importance to develop HC anode with good rate performance. The calculated Ragone plot is shown in Figure 7F, in which a fabulous energy density of 238.91 W h kg −1 at the power density of 36.73 W kg −1 can be observed based on the total mass of active cathode and anode materials. Moreover, the full cell can still deliver a decent energy density of 106.73 W h kg −1 even at a high power density of 1601.04 W kg −1 . Besides, Figure S19 shows the operating voltage at corresponding power density; a prominent operating voltage of 3.57 V can be observed at the power density of 36.73 W kg −1 and can still maintain extraordinary cell voltage of 3.11 V even at the high power density of 1601.04 W kg −1 , which is higher than most sodium-ion cathode materials and shows large potential for practical applications. Besides, the comparing average operating voltage and energy density with other full-cell devices are provided in Table S3. Figure S20 shows that a circuit board with "USTC" logo composed of 38 green light-emitting diodes can be lightened up, making a small demonstration for practical application. Besides, the cycling performance of NNMMO-Zr//HC full cell is also evaluated, as shown in Figure 7G; a decent capacity retention of 71.1% can be achieved after 50 cycles at 0.1 C. In addition, a relatively large rate of 1 C is also used to further estimate its cycling performance ( Figure 7H); it can deliver a capacity retention of 76.8% after 100 cycles. The cycling performance of full cell is inferior to that of half-cell, which may be attributed to the larger polarization of full-cell suffering from the sluggish kinetic behavior of sodium ions in HC, which has been widely reported by previous works. 63,64 The comparing charge/discharge curves between full cell and half-cell are provided in Figure S21; it can be intuitively observed that the full cell exhibits larger polarization during cycling compared with half-cell. In addition, the assembly process may also influence the performance of full cell, 65 in laboratory environmental, many process conditions cannot control as well as that of enterprise, which may also cause the inferior cycling performance of full cell. Despite these, from the above electrochemical performance assessment of full cell, it can be concluded that the NNMMO-Zr//HC full cell with high operating voltage and high energy density exhibits charming potential in practical application.

CONCLUSION
In summary, a dual-strategy of Mg 2+ doping integrated with ZrO 2 surface modification is adopted on P2-Na 0.67 Ni 0.33 MnO 2 through a sol-gel followed by a wet-chemical method. Thanks to the enhanced structure and interface stability brought by the dual-strategy modification, the electrochemical performance is effectively improved. Specifically, the NNMMO-Zr cathode can deliver decent cycling stability with a capacity retention rate of 81.5% at 1 C as well as a promising rate performance with a discharge capacity of 76.6 mA h g −1 at 5 C in half-cell configuration. In addition, exhaustive characterization methods have been employed to investigate the mechanism of improved performance. The dual-strategy modification material exhibits the following advantages: First, the damaged P2-O2 phase transition that occurs in NNMO is suppressed and transforms into a P2-OP4 phase transition with better reversibility when charge in high voltage, which is verified by in situ XRD. Second, profiting by the dual-strategy modification, the kinetic process is also enhanced, which is beneficial to Na + insertion/extraction. Third, the side reactions are also alleviated and the interphase between cathode and electrolyte gets optimized due to the ZrO 2 surface modification.

Characterization of materials
The crystalline structure of the acquired materials is characterized by a powder X-ray diffractor (XRD, Rigaku, Ultima IV). Rietveld refinement is performed by FullProf program. 66 The morphology of the materials is characterized by FESEM (HITACHI, SU8200). The field-emission transmission electron microscopy (FETEM, JOEL, Talos F200X) is employed to characterize the microstructure. The valence states of elements are characterized by XPS (ESCALAB 250). As for the in situ XRD measurement, the in situ XRD cell is purchased from Beijing Scistar Technology Co. Ltd. A beryllium window at the top of the in situ cell is used for X-ray penetration, and Al foil is used as a current collector. During in situ measurement, each scan is recorded with an increment of 0.02 • in the range of 10 • -40 • every 10 min.

Electrochemical measurement
To fabricate the cathode electrode, the active material, carbon black (Super-P), and PVDF is well mixed and dispersed in N-methylpyrrolidone (NMP) solution in a mass ratio of 7:2:1. After that, the slurry is cast on the Al foil and then dried at 70 • C in the air for 12 h and 100 • C in the vacuum oven for another 12 h to totally evaporate the NMP. The mass loading is 1.4-1.6 mg cm −2 for each cathode electrode. The electrochemical performance is evaluated in the CR2016 coin cell, and a piece of metallic Na disc is used as both counter and reference electrode. NaClO 4 of 1 M in propylene carbonate with 5 vol% of FEC is employed as the electrolyte, whereas the glass fiber (Whatman, GF/F) is used as a separator. The galvanostatic charge/discharge tests are performed on a Neware battery test system (CT4008) in the voltage range of 2.5-4.35 V at room temperature. The collected CV curves and EIS are conducted in a CHI660E electrochemical station.

Preparation of the full cell
The anode electrode is prepared by well mixing the commercial HC (95 wt%) and CMC (5 wt%) with appropriate deionized water, and then the slurry is pasted on the copper foil. The drying procedure is same as that of positive electrode. The prepared anode is presodiated via a precycle procedure in half-cell. The cathode and anode are assembled in a CR2016 coin cell with the weight ratio of about 1.4-1.6. The electrolyte and separator are same as that used in the half-cell, and the full cell is investigated in the voltage range of 2.4-4.25 V.

C O N F L I C T O F I N T E R E S T
The authors declare no conflict of interest.