Increasing concerns about energy shortages and serious environmental problems in the modern society requires reasonable utilization sustainable energy sources, for which efficient large-scale electric energy storage technology is urgently demanded [1-6]. Owing to the low cost and abundant sodium resources in earth, sodium-ion batteries (SIBs) are considered as a promising candidate to substitute lithium-ion batteries (LIBs) for large-scale grid energy storage [7, 8]. However, the lack of high performance electrode materials, limits the practical application and commercialization of SIBs. In the past decades, various cathode materials have been intensively investigated [9-12]. Among them, layered sodium transition metal oxides Na_xMO₂ (M is a transition metal, Co, Mn, Fe, Ni, etc.), have been considered as one of the most potential cathode materials, which can be classified into P2 and O3 phases according to the Na occupation sites and repeated unit cell [13]. Especially, the manganese-based P2 layered structure (Na_0.7MnO₂) have attracted considerable attention due to their low cost, high theoretical capacity and environment friendliness of Mn [14-17]. Nevertheless, such structures usually undergo a structural degradation and serious voltage decay upon Na⁺ (de)intercalation, thus cannot fully satisfy the demands of advanced applications. Thus, voltage fade is the pivotal problem that urgently needs to be overcome.

As is generally accepted, the irreversible phase transition, Jahn–Teller lattice distortion of Mn(III) and Na⁺/vacancy ordering occurred in the electrochemical process are mainly responsible for the SIB fadings [18-20]. Various approaches including lattice doping and surface/interface modifications, have been intensively investigated to alleviate the structural change upon cycling. Cationic doping with heteroatoms, such as Li⁺, Mg²⁺, Al³⁺, Ti⁴⁺, has been proved as an effective method to promote the performance of cathode materials [21-25]. For instance, the incorporation of more Cu ions can suppress the valance change of Fe ions and voltage decay in the Fe-based layered oxides [26]. Li as an inert element, is also found can act as the structural support point to stabilize the crystal structure during Na⁺ intercalation/deintercalation [27]. Meanwhile, considering the cost of production in practice, trace amount doping with a heteroatom concentration generally two to three orders lower is placed on the agenda. Trace Ti-doping is recently reported to effectively modify the microstructure of LiCoO₂ and stabilize the surface oxygen at high voltages, resulting in a promoted cycling stability at 4.6 V [28].

Despite high relevance in structure and electrochemistry, strong oxygen redox accompanied by the oxygen gas release from the Mn-based material was revealed, but none in the high-valent ion doped materials. Inspired by the above work, we report a novel low-concentration Nb doping of Na_0.7Ni_0.3Co_0.1Mn_0.6O₂ as high performance SIB cathode material. Considering the relationship between transition-metal ion migration and voltage decay, the size of doped ions and the bond dissociation between metal and oxygen play an important role in suppressing voltage decay. Herein, 5d metal niobium has been incorporated into manganese-based layered oxides [29-32]. Particularly, the ionic radius of Nb⁵⁺ (0.64 Å) is close to that of Mn³⁺ (0.645 Å) and Co³⁺ (0.61 Å), ensuring the successful doping of Nb element in the layered structure. Besides, the Nb-O bonds generally demonstrate higher metal-oxygen bond energy [33-36], which is expected to enhance structural stability, and alleviate the structural degradation upon cycling. The improved electrochemical performance is resulted from the enhanced oxygen stability induced by local structural variation around doping Nb. Consequently, the as prepared Na_0.7[Ni_0.3Co_0.1Mn_0.6]_0.98Nb_0.02O₂ demonstrates a high capacity retention of 87.9% after 200 cycle at 0.5 C, showing a 2.4 fold improvement as compared to that of the control sample Na_0.7Ni_0.3Co_0.1Mn_0.6O₂ (37%). It is worth noting that, even in trace amounts, Nb-doping suppresses significantly the voltage decay of Na_0.7Ni_0.3Co_0.1Mn_0.6O₂. Na_0.7[Ni_0.3Co_0.1Mn_0.6]_0.98Nb_0.02O₂ shows much lower discharge midpoint voltage decay (0.132 V) than that of pristine one (0.319 V) after 200 cycles. Remarkably, the as prepared material also demonstrates good performance at low temperature (−20 ℃).

The Nb-doped materials with Na_0.7[Ni_0.3Co_0.1Mn_0.6]_1-xNb_xO₂ were synthesized using two steps: co-precipitation and the solid phase sintering method. Nickel, cobalt and manganese acetate was dissolved in 30 mL deionized water according to the stoichiometric ratio, which corresponds to a concentration of 2 mol/L for the metal ions. Then, 90 mL Na₂CO₃ solution (2 mol/L) was added into the acetate solution. The above solution was stirred for 12 h, then filtered, washed and dried at 80 ℃ to obtain the carbonate precursor [Ni_0.3Co_0.1Mn_0.6]CO₃. Next, based on 0.01 mol of carbonate precursor, add a certain proportion of Na₂CO₃ and Nb₂O₅, and mix it thoroughly. The powder was calcined at 500 ℃ for 10 h, then sintered again at 900 ℃ for 15 h in air with heating rate of 5 ℃/min and cooled down to room temperature.

The diffraction data were collected by the powder X-ray diffraction (PXRD, Rigaku P/max 2200 VPC) equipped with Cu-Kα (λ₁=0.15406 nm, λ₂=0.154439 nm) radiation source within the range from 5° to 80° at a scan rate of 2°/min. The SEM (Zeiss SUPRA 55) and TEM (Hitachi HT-7700, 120 kV) were employed to observe the morphology of the cathode materials. X-ray photoelectron spectroscopy (XPS) performed on a Thermo Fisher Multi-element (K-alpha) high-transmission spectrometer input lens, with energy ranging from 200 eV to 3 keV.

The Nb-doped Na_0.7[Ni_0.3Co_0.1Mn_0.6]_1-xNb_xO₂ (x=0, 0.02, denoted as NCM, NCMN) samples were successfully synthesized through a classical solid-state reaction. The inductively couple plasma (ICP) results (Table S1 in Supporting information) of NCMN confirm the ratio of 0.723:0.290:0.099:0.568:0.013 for Na: Ni: Co: Mn: Nb, which is very close to the expected stoichiometry. The XRD results of the obtained samples show all the diffraction peaks are well indexed to a typical P2-type structure with the P63/mmc space group (Fig. S1a in Supporting information). As shown in Fig. 1a and Table S2 (Supporting information), the Rietveld refinement of the NCMN gives the lattice parameters a=b=2.8771(6) Å, c=11.1332(5) Å and V=79.8143 ų with a reliability index (R_wp) of 4.53%. The molecular model structure diagram for the P2-NCMN is shown in inset of Fig. 1a, which is composed of alternating MnO₆ slabs and Na layers. Besides, niobium ions occupy the transition metal sites in metal oxide octahedral. The refined XRD patterns of NCM sample is also displayed in Fig. S1b (Supporting information) and corresponding crystallographic parameters are listed in Table S3. After Nb-replacement, the (002) peak shifts to lower degree (Fig. 1b), which reveals that Nb has been successfully doped into Na_0.7Ni_0.3Co_0.1Mn_0.6O₂ without affecting the hexagonal structure. The crystallographic lattice parameters after Rietveld refinement are summarized in Table S4 (Supporting information). The d-spacing of Na⁺ layers is decreased by 0.181 Å, whereas the slab thickness of TMO₂ is increased 0.135 Å after Nb substitution. Thereby c parameter is generally decreased from 11.225 Å to 11.133 Å. The possible reason is the comprehensive effect of larger size and high valence state of Nb, which leads to the shrinkage of Na layer in c axis and expands TMO₆ octahedron [37].

The transmission electron microscopy (TEM) and high-resolution TEM (HRTEM) images further confirm that the as-prepared sample are highly crystalline (Figs. 1c and d). The morphologies of NCM is shown in Fig. S2 (Supporting information). It is clear that there is no obvious change in morphology before and after doping. From the HRTEM image in Fig. 1d, the interplanar distance between neighboring lattice fringes is 0.565 nm, corresponding to the (002) planes of the layered structure. Furthermore, the selected area electron diffraction (SAED) patterns (Fig. 1e) are indexed to the P2-type crystalline structure viewed from the [001] direction, in accordance with the results of HRTEM images. The scanning electron microscopy (SEM) image shows that the morphology of the NCMN, which is an irregular layered structure in general with size ranging from 2 μm to 5 μm. The results of EDS mapping on the crystallites reveal that Na, Ni, Co, Mn, Nb and O elements are homogenously distributed in the materials (Fig. 1f).

X-ray photoelectron spectroscopy (XPS) was employed to investigate the oxidation states of elements in NCMN electrode (Fig. 2). Dominant peaks located at 854.90 and 872.15 eV are attributed to the peak of the Ni 2p line, indicating that the valence of nickel ion is +2 [38]. Two peaks of Co 2p_3/2 and Co 2p_1/2 respectively correspond to 780.29 and 795.46 eV, demonstrating the presence of Co³⁺. In addition, Mn 2p_3/2 and Mn 2p_1/2 peaks are located at 642.47 and 653.95 eV, respectively, which clearly reveals that the Mn³⁺ and Mn⁴⁺ coexist in the material [39]. The XPS peaks at 209.53 and 206.64 eV provide direct evidence that valence state of Nb is +5, in accord with previous report [40].

In order to study the effect of Nb doping on the materials, electrochemical performance of both samples in Na half cells were tested in the voltage range of 2–4.25 V. As displayed in Fig. 3a, the first three cyclic voltammetry (CV) curves of the NCMN show three pairs of reversible cathodic/anodic peaks, which are located at 2.54/2.21 V, 3.37/3.21 V and 3.76/3.49 V. The peaks around 2.5 V are relative to the redox reaction of Mn⁴⁺/Mn³⁺, while these multiple peaks above 3.0 V could be attributed to the redox reactions of Ni and Co and the simultaneously occurred Na⁺/vacancy ordering [41-44]. The redox reactions of Ni and Co are the major contributors of the capacity in the charge/discharge cycles. Compared to the CV curves of the NCM (Fig. S3a in Supporting information), the sample NCMN exhibits smaller polarization and better reversibility, which may be attributed to the influence of Nb dopant. Figs. S3b and c (Supporting information) compare the typical charge/discharge voltage profiles for both cathode materials from 2 V to 4.25 V at a rate of 0.1 C (1 C=180 mA/g). Obviously, the charge/discharge curves of NCM and NCMN exhibit a long voltage platform over 3.5 V, then followed by a sloping region above 3.75 V accompanied with by the sequential oxidation of Ni and Co, which is consistent well with the CV results. The initial reversible discharge capacity of NCMN is estimated to be ≈ 84.2 mAh/g, the discharge curves of the first three cycles are almost identical, indicating that the material has good reversibility. Compared with NCM, the smoother charge/discharge curves are observed for the NCMN. This certifies that the substitution of Nb indeed greatly suppresses the Na⁺/vacancy ordering [45]. As shown in Fig. 3b, the electrodes were cycled at various current densities ranging from 0.1 C to 5 C. The reversible capacities of the NCMN are approximately 94, 87.1, 80.5, 76.9, 73.1 and 55.5 mAh/g, whereas the pristine delivers a discharge capacity of 94.4, 85.6, 77.1, 69.4, 50.6 and 0.1 mAh/g at 0.1 C, 0.2 C, 0.5 C, 1 C, 2 C and 5 C, respectively. Further compared with the rate performance of both electrodes (Fig. S4 in Supporting information), the galvanostatic charge/discharge curves show the slight increase of voltage polarization for NCMN at various rates. The excellent kinetic capability of Na⁺ extraction/insertion is also proved by small voltage polarization in the first charge/discharge galvanostatic intermittent titration technique (GITT) curves (Fig. 4a). The cycling performances for both the Na_0.7[Ni_0.3Co_0.1Mn_0.6]_1-xNb_xO₂ electrodes at 0.5 C are compared in Figs. 3c and d. The initial discharge specific capacities of NCM and NCMN are 84.3 and 78.5 mAh/g, corresponding to the capacity retention of 37% and 87.9% after 200 cycles, respectively. Compared to the pristine NCM, although the initial capacity of the Nb-doped material slightly decreases, the presence of Nb leads to higher capacity retention.Fig. 3e shows the discharge midpoint voltage (MPV) of the pristine and Nb-doped electrode. It can be apparently seen that the discharge midpoint voltage decay of NCMN is 0.132 V after 200 cycles, whereas that of NCM is 0.319 V. Obviously, trace amount of Nb doping can enhance the cycling stability and mitigate the voltage decay. In addition to the superior rate capabilities and cycling stability of the as-prepared NCMN, noting that it displays outstanding low-temperature performance. The electrochemical performance at low temperature (−20 ℃) is also investigated and shown in Figs. 3f-g. A reversible capacity of 63.6 mAh/g is obtained at −20 ℃ at 0.5 C, which is 81% of that at room temperature. Significantly, at −20 ℃, all the charge-discharge curves give two evident voltage plateaus from 1^st to 300^th cycles indicating the very limited voltage decay upon cycling. The initial discharge specific capacities of NCMN are 78.5 and 63.6 mAh/g at 25 ℃and −20 ℃ respectively, corresponding to the capacity retention of 79.3% and 88.5% after 300 cycles (Fig. 3g). The higher capacity retention at −20 ℃ should be associated with the more sluggish Na-transport kinetics at low temperature. Besides, Fig. S5 shows the discharge capacity of NCMN remains 68.2 mAh/g at 2 C for 100 cycles with a high capacity retention of 94.8%, even achieving a capacity retention of 68.4% after 500 cycles.

To further explain the structural stability and internal mechanism, the morphology and electrochemical behavior of NCMN were examined before and after cycling, and the results are displayed in Fig. S6 (Supporting information). In terms of morphology, the main layered structure was still remained even after 200 cycles at 0.5 C as shown in Figs. S6a and b, suggesting the good structure stability for NCMN. Figs. S6c and d show the major oxidation/reduction peaks located at about 3.5 V is still sharp after 200 cycles, which further confirms that NCMN has high structure reversibility upon Na⁺ (de) intercalation. This is probably a result from reducing migration of Ni²⁺ and dissolution of Mn³⁺ in the transition metal layers, and to some extent, restraining the phase transition after Nb substitution [26, 28, 45, 46]. These results indicate that substitution of Nb can effectively stabilize the structure of the crystal and suppress voltage fading during the charging/discharging process, further improving the electrochemical properties of the materials.

The galvanostatic intermittent titration technique (GITT) and electrochemical impedance spectroscopy (EIS) are employed to further examine the effects of Nb-doping on kinetics behavior of Na⁺ ion. The test method was described in supporting information. Fig. 4b compares the diffusion coefficient of Na⁺ (D_Na) calculated from the GITT curves. The values of D_Na in the NCMN electrode vary from 7.29×10⁻¹² cm²/s to 2.24×10^-10 cm²/s, while NCM shows lower diffusion coefficients (8.97×10^-13 cm²/s to 6.21×10^-11 cm²/s). Although the trend of diffusion coefficients is similar, variation of D_Na versus voltage plots in NCMN is slightly smaller than that of NCM. The slower diffusion rate corresponds to the charge plateaus, and sodium ions diffuse faster in the sloping region, which results from a phase transition or order-disorder transition during sodium intercalation/deintercalation in layered oxide cathode for SIBs [47]. Figs. 4c and d are electrochemical impedance spectroscopy (EIS) results of both samples at open circuit voltage (OCV), which provide further evidence for the better rate capability of the Nb-doped material. The D_Na of NCMN determined from EIS test is 1.16×10⁻¹² cm²/s larger than that of NCM (5.15×10^-13 cm²/s), which is in good agreement with the GITT results. Both plots consist of two semicircles at the high frequency range, and a straight line at the low frequency range attributed to Warburg impedance of Na⁺ diffusion in the bulk material. The first semicircle reflects sodium ion migration through the interface between the surface of the particles and the electrolyte, the second semicircle could be regarded as the charge-transfer resistance. Obviously, the first semicircle of NCMN is smaller, which demonstrates that the Nb-doped samples show improved mobility of Na⁺ in bulk electrode materials than that of the pristine counterpart. In addition, total impedance of two parts decreases from 631.5 Ω (NCM) to 480.3 Ω (NCMN) after doping. As discussed above, it is plausible that the Nb-doping can increase the electronic conductivity and accelerate the migration rate of the sodium ions, thus enhancing the rate capability of cathode material [37]. The strong Nb-O ionic bond inhibits the variation of the TM-O bond or the distortion of the TMO₆ octahedron, maintaining the integrity of the surface structure [47]. The incorporation of Nb into the transition metal layer and the Mn⁴⁺ oxidation state can improve the structural stability as well due to the presence of tough Nb-O bonds relative to Co/Mn/Ni-O bonds. These improvements are responsible for the promoted cycling and rate performances of NCMN.

In summary, we have successfully synthesized a novel trace Nb-doped cathode materials and explored the influence of Nb substitution on the electrochemical performance of P2-Na_0.7Ni_0.3Co_0.1Mn_0.6O₂ composites. The NCMN electrodes deliver an initial discharge capacity of 78.5 mAh/g at 0.5 C (tested at 2~4.25 V), with retention of 87.9% and negligible voltage decay of 0.132 V after 200 cycles. Besides, it also keeps initial capacity of 63.6 mAh/g and retention of 88.5% at 0.5 C under −20 ℃ after 300 cycles. EIS and GITT tests demonstrate that the pronounced rate performance is attributed to high sodium diffusivity during the intercalation/deintercalation. The results indicate that Nb-doping has a positive effect on the improvement cycling stability and the sodium ion diffusion coefficient. In particular, it can dramatically suppress the discharge voltage decay during the cycling process. Nb-doping is a possible strategy to maintain structural stability for suppressing the detrimental voltage fade to improve energy density for practical applications.

The authors declare no competing interests.

We thank the financial supports from the National Natural Science Foundation of China (No. 51774251), Hebei Natural Science Foundation for Distinguished Young Scholars (No. B2017203313), Hundred Excellent Innovative Talents Support Program in Hebei Province (No. SLRC2017057), Talent Engineering Training Funds of Hebei Province (No. A201802001), and the Opening Project of the State Key Laboratory of Advanced Chemical Power Sources (No. SKL-ACPS-C-11).

Supplementary material related to this article can be found, in the online version, at doi: