Introduction

Lithium-ion batteries (LIBs) have attracted significant interest as the power sources of numerous portable consumer electronics due to their high safety, high energy density (both volumetric and gravimetric), long lifetime, wide operating temperature range, and environmental friendliness [1, 2]. However, LIBs cannot meet the ever-increasing marketable demands due to the booming cost and limited availability of lithium reserves [3,4,5,6]. It is necessary and significant to develop alternative with high-performance to replace the LIBs. Recently, sodium-ion batteries (SIBs) has been a research hotspot in the fields of energy storage and conversion. Sodium has similar physical and chemical properties with lithium [7, 8], and the abundance is 430 times of lithium resources. Nevertheless, the larger radius of Na+ is considered to be a restriction because it makes the de-intercalation of Na+ more difficult and affects the structural stability of host materials. Therefore, the crucial issue in the research of SIBs is to develop the adaptive cathode materials facilitating the migration of Na+. For the past few years, various cathode materials have been developed and modified, such as NaVO3 [9], NaFePO4 [10,11,12], NaVPO4F [13], MnO [14], and Na3V2(PO4)3 [15, 16]. Among the abovementioned electrodes, Na3V2(PO4)3 has generated considerable research interests due to its highly open three-dimensional NASICON structure and the advantages of large capacity and easy preparation [17,18,19].

Na3V2(PO4)3, with high thermal stability, NASICON structure, and high-power density, is deemed to be one of the most prospective candidates for SIBs. It was firstly proposed by Delmas in 1978. The triclinic Na3V2(PO4)3 is constructed by PO4 tetrahedron and VO6 octahedron, which has a highly open three-dimensional NASICON structure [20]. The unique crystal structure can be beneficial to the transition of Na+. However, it suffers from the low intrinsic electronic and ion conductivity [21,22,23]. Generally, carbon coating is one of the most effective methods to improve the surface electrical conductivity of Na3V2(PO4)3. But it has little effects on the mobility of Na+ [24, 25]. So, it is significant to search for a solution to realize the rapid and repeated transport of Na+, improving the electrochemical kinetics and cyclability of SIBs [26]. Cation doping is a successful approach to enhance the bulk phase characteristics. A slight amount of cation ions partially substituting of V in Na3V2(PO4)3, such as Mg [24], Al [25], Ti [26], and Mn [27], have been widely used as alternatively effective methods. In addition, doping K+ to partially substituting of Na can enhance the stability of the structure and channel [28]. Our previous investigation indicates that the optimum content of the doped K+ in Na3-xKxV2(PO4)3 is 0.04 (Fig.S1). Therefore, carbon coating combining with cation co-doping may be an effective way to improve electrochemical performance of Na3V2(PO4)3.

In this work, Na2.96K0.04V2-xZr(3/4)x(PO4)3/C(x = 0.01, 0.04, 0.07, 1) were synthesized by a solid-phase method to enhance the bulk electronic conductivity and ion conductivity. The electrochemical properties of Na3V2(PO4)3 were remarkably improved by the design of carbon coating and K+/Zr4+ co-doping. Conducting carbon coating constructs a conductive network to increase the surface electronic conductivity of particles. Replacing sodium site with potassium can improve the structural stability. In addition, the Zr4+ substitution can facilitate the migration of Na+ by broadening the diffusion pathway of Na+. Compared with the undoped composite, the co-doped samples exhibit superior cyclic and rate performance owing to the enhanced electrical and ionic conductivity. Notably, Na2.96K0.04V1.93Zr0.0525(PO4)3/C composite exhibits the best electrochemical performance: it delivers a high discharge capacity of 107.3 mAh g−1 at 0.1 C; it remains 92.3 mAh g−1 after 400 cycles at 2 C, corresponding to the capacity retention of 92.02%; it can deliver a high discharge capacity of 100.0 mAh g−1 even at 10 C rate.

Experiment

Materials preparation

Na2.96K0.04V2-xZr (3/4)x(PO4)3/C(x = 0.01, 0.04, 0.07, 0.1) were synthesized by a solid-phase method. ZrO2, K2CO3, Na2CO3, NH4H2PO4, V2O5, and C6H12O6 · H2O were used as raw materials. Carbon-coated design was prepared with C6H12O6 · H2O as carbon source. The raw materials were mixed uniformly with stoichiometric ratio of 9x:0.24:17.76:36:(12-6x):(4-2x). Then, it was ball-milled for 8 h in a high energy ball-milling machine. Afterwards, the yellow fluid was put in the drying box and dried at 80 °C for 12 h. It was ground for 40 min and then calcined at 450 °C for 4 h under argon gas. The active composites, Na2.96K0.04V2-xZr (3/4)x(PO4)3/C, were obtained after final heated at 700 °C for 6 h under argon gas. The synthesized samples Na2.96K0.04V2-xZr(3/4)x(PO4)3/C (x = 0, 0.01, 0.04, 0.07, 0.1) were abbreviated as NVP/C, KZ0.01, KZ0.04, KZ0.07, and KZ0.1, respectively.

Characterization

The crystal structure of the materials were determined by X-ray diffraction (XRD, Bruker D8 Advance) with radiation between 10°–80° and a scanning speed of 2° min−1. Scanning electron microscopy (SEM, HITACHI SU8010) and transmission electron microscopy (TEM, JEM-2100F) were used to observe the particle size and morphology. X-ray photoelectron spectroscopy (XPS, ESCALAB 250X) was performed to analyze the binding energies of vanadium, potassium, and zirconium elements. Thermogravimetric analysis (TGA, Mettler-Toledo) was employed to analyze the weight percentages of carbon. The small variation of each doping concentration is determined by inductively coupled plasma-mass spectrometer (ICP-MS) using NexIONTM 350D, PerkinElmer (PE), USA.

Electrochemical measurement

The electrochemical properties of the samples were tested on LAND battery test system by button cell (LIR2016). The cathode electrode was prepared by adding 80 wt% active material, 10 wt% acetylene black conductive agent, and 10 wt% polyvinylidene chloride (PVDF) binder to NMP solvent. The mass loading of the active material in each coin cell is about 1.2 mg cm−2. The battery was assembled in a glove box filled with pure argon gas. Both cyclic voltammetry and electrochemical impedance spectroscopy tests were conducted by an electrochemical workstation (AUIVIUM). Cyclic voltammetry was measured at a scan rate of 0.1 m V s−1 from 2.4 to 4.1 V. Electrochemical impedance spectroscopy was carried out with an AC amplitude of 10 mV from 0.01 Hz to 100 kHz.

Results and discussion

The XRD patterns of all samples are revealed in Fig. 1a. The vast majority of diffraction peaks can be attributed to the rhombohedral structure with R-3c space group. A few impurity peaks at around 18° are detected, which represent the existence of NaVP2O7 and NaVO2. These by-products are favorable to improving the conductivity of Na+ [29, 30]. Moreover, refined results are carried out for all samples to acquire more accurate information of crystal structure. As shown in Fig. 1b and Fig.S2, the XRD patterns of KZ0.07 sample were clearly assigned to the framework of a rhombohedral NASICON structure with the R3c space group (ICDD#01-078-7289). The Rietveld-refined XRD patterns of KZ0.07 sample are shown in Fig.S3. The refinements are carried out using GSAS program suite, and a space group of R-3c is chosen as the refinement model. The K and Zr elements have substituted at Na1 and V site correspondingly. As can be seen in Table.S1, the occupancy of K and Zr in KZ0.07 sample is 0.0112 and 0.0312, respectively. The sharp peaks with high intensity in Fig. 1b reveal the great crystallinity of KZ0.07 composite. This also demonstrates that co-doping with small amounts of K+ and Zr4+ does not change the crystal structure of Na3V2(PO4)3. Meanwhile, no obvious diffraction peaks of the coated carbon layer are discovered in all samples, which can be attributed to the low content and amorphous state [23]. The refined lattice parameters of KZ0.01, KZ0.04, KZ0.07, KZ0.1, and pure NVP are listed in Table 1 for comparison. The ionic radius of K+ and Zr4+ is apparently larger than those of Na+ and V3+, suggesting that co-doping can contribute to a bit expanded unit cell volume. The enlarged unit cell volume can broaden the migration pathway of Na+. Moreover, the ionic radius of Zr4+ is very close to V3+, indicating that Zr4+ tends to be doped in the V site in NVP. On the other hand, K+ tends to be doped in the Na site in NVP. Compared with pure NVP, the co-doped composites have larger value of c. Especially, enlarged c parameter by co-doping facilitates the movement of Na+ during charge and discharge process [24].

Fig. 1
figure 1

a The XRD patterns of Na2.96K0.04V2-xZr (3/4)x(PO4)3/C (x = 0, 0.01, 0.04, 0.07, 0.1). b Rietveld-refined XRD patterns of KZ0.07

Full size image
Table 1 The calculated lattice parameters of NVP, KZ0.01, KZ0.04, KZ0.07, and KZ0.1
Full size table

To determine the carbon content of NVP/C and KZ0.07 samples, the TGA curves are collected and shown in Fig. 2. The carbon contents of NVP/C and KZ0.07 are 3.92 wt% and 3.77 wt%, respectively. The carbon content of these two samples is about 4 wt%, which can be attributed to the excess carbon source, facilitating the electronic conductivity efficiently. Moreover, the small variation of each doping concentration (0.01, 0.04, 0.07, 0.1) is determined by inductively coupled plasma-atomic spectrometer (ICP-AES). The results are listed in Table.S2. The values of ICP measurements are slightly lower than that of theoretical contents.

Fig. 2
figure 2

TGA curves of NVP/C and KZ0.07 samples

Full size image

The SEM image of pure NVP and co-doped Na2.96K0.04V2-xZr(3/4)x(PO4)3/C (x = 0.01, 0.04, 0.07, 0.1) samples are shown in Fig.S4 and Fig. 3. The different doping amounts of Zr4+ result in different morphologies. Na2.96K0.04V2-xZr(3/4)x(PO4)3/C (x = 0.01, 0.04, 0.1) samples display the serious agglomeration with aggregated sizes of up to 9 μm. This phenomenon can be attributed to the high-temperature calcination in the preparation process. Combined with the XRD results (Fig. 1) and refined lattice constants (Table 1), the SEM results indicate that K+/Zr4+ doped into the crystal structure of NVP, which did not destroy the structure stability and had little effects on the particle morphology. In comparison, the particles of KZ0.07 composite are smaller with more even size distribution. The primary size of the KZ0.07 composite is about 2 μm. The reduced particle size can shorten the diffusion path of Na+ and increase the specific surface area to improve electrochemical performance. The SEM image of KZ0.07 sample at 100 K magnification is shown in Fig. 3e.

Fig. 3
figure 3

The SEM images of (a) KZ0.01, (b) KZ0.04, (c) KZ0.07, (d) KZ0.1, and (e) KZ0.07

Full size image

HRTEM was used to further elucidate the components in KZ0.07 sample. As displayed in Fig. 4, the surface of particle is covered by a thin carbon layer, which is beneficial to improving the electronic conductivity effectively. This carbonaceous phase is built as a rough layer of 10 nm surrounding the particles. The carbon layer would induce the formation of a mixed conducting network, which is beneficial for both electron and ion transport. In addition, the interplanar spacing of 0.62 nm coincides with the (012) planes in Na3V2(PO4)3 system. The lattice fringes correspond to the well-crystallized structure.

Fig. 4
figure 4

The HRTEM images of KZ0.07 sample

Full size image

XPS measurements are conducted to detect elementary binding energy and oxidation state. As shown in Fig. 5a, the C 1s peak is corrected to 284.6 eV for the further study of other elements. The peak at 284.6 eV corresponded to the C 1s can furtherly prove the presence of carbon coating in the sample. The two peaks at 517 eV and 524.15 eV can be attributed to the excitations of V3+ 2p3/2 and V3+ 2p1/2 in KZ0.07 composite, respectively, which are similar to those in reference for the oxidation state of vanadium in NVP without any peak shift [25]. The low-intensity K2p peaks may result from the low doping amount of K+. As depicted in Fig. 5c, two peaks at binding energy of 295.9 eV and 292.9 eV are corresponding to K+2p1/2 and K+2p3/2. The peaks representing Zr4+3d5/2 and Zr4+3d3/2 locate at 183.2 eV and 185.65 eV correspondingly. The abovementioned results indicate that both K+ and Zr4+ have successfully doped into Na site and V site in KZ0.07 sample. Co-doping of K+ and Zr4+ in samples and the maintenance of V3+ suggest that vacancies can be produced in V site. Appropriate vacancies are in favor of rapid transport for Na+ in electroactive-doped particles. The specific fitting results are listed in Table 2.

Fig. 5
figure 5

The XPS spectra of KZ0.07 sample

Full size image
Table 2 The XPS fitting results of the sample
Full size table

The cyclic voltammetry curves of all samples are displayed in Fig. 6. The scanning speed was 0.1 mVs−1 and the scan voltage range was 2.3–4.1 V. The CV curves of all samples are composed of one oxidation peak and two reduction peaks, which corresponds to the extraction and insertion of Na+. The curves reveal similar, indicating that Na+ and Zr4+ co-doping does not change the electrochemical reaction process of NVP. It is also clearly reflected the intercalation and extraction processes of Na+. For comparison, the co-doped samples have redox peaks with higher intensity and larger areas, suggesting larger electrochemical capacity. Especially, the extra occurrence of a small reversible redox couple at 3.7 V is observed in the co-doped samples, corresponding to the partial redox reaction of V4+/V5+. This consequence is discovered in the related doped NVP composites, such as Na3V1.98(PO4)3/C [29] and Na3V1.98(PO4)2.9F0.3/C [30].

Fig. 6
figure 6

Cyclic voltammograms of Na2.96K0.04V2-xZr (3/4)x(PO4)3/C (x = 0.01, 0.04, 0.07, 0.1) and pure NVP at 0.1 mV s−1

Full size image

Figure 7 a and b show the galvanostatic charge/discharge curves of NVP/C and Na2.96K0.04V2-xZr (3/4)x(PO4)3/C(x = 0.01, 0.04, 0.07, 0.1) in the first cycle at 0.1 C and 2 C, respectively. As shown in Fig. 7a, all the samples exhibit one discharge plateau at 3.4 V, which can be assigned to the intercalation/deintercalation of Na+ at M2 site. Notably, the K/Zr co-doped samples reveal a new voltage platform at around 3.7 V, corresponding to the V4+/V5+ reaction resulting from the reversible extraction of the third Na+ at M1site [30,31,32,33]. The initial discharge capacities at 0.1C rate of NVP/C, KZ0.01, KZ0.04, KZ0.07, and KZ0.1 are 94.5, 91, 100.3, 107.3, and 97.1 mAh g−1, respectively. When it turns to be 2 C rate, the values are 85.6, 91.2, 96, 107.1, and 93.5 mAh g−1. Moreover, the KZ0.07 composite exhibits the exceedingly slight polarization voltage, indicating the favorable reversibility of Na+ migration. It indicates that the co-doping of K+ and Zr4+ can enhance the charge and discharge characteristics significantly. All the doped samples reveal the typical voltage plateaus of NVP with lower polarization voltage than that of the NVP/C sample, which can be attributed to the enlarged diffusion pathway caused by K+ and Zr4+ co-doping, inducing the fast insertion and extraction of sodium-ion.

Fig. 7
figure 7

a First cycle of charge/discharge curves of all samples at 0.1 C. b First cycle of charge/discharge curves of all samples at 2 C. c Long-cycle performance of all samples at 2 C for 400 cycles. d 400th cycle charge/discharge curves of all samples at 2 C. e Rate performance of all samples. Charge/discharge curves at different densities for NVP/C (f) and KZ0.07-NVP/C (g) samples. h Discharge specific energy curves for all samples at 2C

Full size image

Figure 7c exhibits the cycling performances of all samples at 2 C in the voltage range of 2.4–4.1 V. KZ0.07 composite possesses the best cyclability among the abovementioned materials. The highest initial discharge capacity is 100.3 mAh g−1, and it remains 92.3 mAh g−1 after 400 cycles, corresponding to the capacity retention of 92.02%. By comparison, NVP/C sample only has the 77.5 mAh g−1 after a cycle and the capacity retention is about 76.5%. The coulombic efficiency of KZ0.07 maintains in 99.93% during the 400 cycles at 2 C rate, suggesting the good reversibility. Moreover, the 400th charge/discharge curves of all samples are displayed in Fig. 7d. Obviously, the value of polarization voltage for NVP/C has greatly increased, leading to the poor electrochemical property. The tiny plateau at 3.7 V representing the reaction of V4+/V5+ has gradually disappeared for KZ0.01, KZ0.04, and KZ0.1. However, it still contributes some capacity for the KZ0.07 composite. The improved cycle performance of KZ0.07 mainly relates with the large and stable channels for the extraction/insertion of Na+, which can be ascribed to the co-doping of K+ and Zr4+. This result can correspond to the XRD results. A small amount of cationic co-doping can enlarge the ionic transfer channel of NVP to improve the electrochemical performance. In a certain range, the electrochemical performance of NVP can be improved with the increase of doping amount because of the larger channel and more suitable amounts of vacancies. However, when the excess cations take up the position of the sites, the cell shrinkage occurs and the transport channel compresses, which impedes the rapid diffusion of Na+ in the bulk of electroactive material and is unfavorable for the enhancement of structural stability.

In order to investigate the influence of K+ and Zr4+ co-doping on the rate capability, NVP/C and Na2.96K0.04V2-xZr (3/4)x(PO4)3/C (x = 0.01, 0.04, 0.07, 0.1) samples were charged and discharged at diverse density from 0.1 to 10 C then back to 1 C. Figure 7e shows that discharge capacities of undoped NVP/C electrode are much lower than that of KZ0.07 at the same measured rates. When NVP/C discharges at 0.1, 0.2, 0.5, 1, 5, 10, and 1 C, its discharge capacities are 69.5, 64.1, 59.3, 56.8, 53.1, 49.0, and 65.5 mAh g−1. Its reversible capacity drops rapidly with the increase of current rates. The capacity retention ratio at 10 C is 70.50%. While KZ0.07 composite presents a significantly enhanced rate capability and exhibits discharge capacities as high as 108.2, 107.2, 106.0, 104.6, 103.4, 100.0, and 104.2 mAh g−1 at 0.1, 0.2, 0.5, 1, 5, 10, and 1 C, respectively. The capacity retention ratio at 10 C corresponds to 92.42% of that at 0.1 C. The outstanding electrochemical performance of KZ0.07 at different current rates was attributed to the introduction of K+ and Zr4+ which is beneficial to the fast Na+ diffusion, especially at high current rates. For further investigation of the rate capability, the charge/discharge curves at different current densities for NVP/C and KZ0.07 are revealed in Fig. 7f and g. The high overlap ratio of curves in Fig. 7g for KZ0.07 demonstrates the excellent reversible capability even at high current densities. By contrast, the front opening of curve in Fig. 7f for NVP/C gets larger with the increase of current density, which suggests the poorer electrochemical property. Due to the appearance of higher platform at 3.7 V for the co-doped samples, the specific energy must be improved accordingly. As shown in Fig. 7h, the discharge specific energy of NVP/C is 241.3 Wh kg−1, much lower than that of the co-doped composites. KZ0.07 exhibits the highest value of 349.5 Wh kg−1 at 2 C rate. The elevated energy density is ascribed to the enhanced capacity and potential.

To further understand the effects of co-doping of K+ and Zr4+ on the electrochemical behaviors, electrochemical impedance spectroscopy (EIS) tests were performed on Na2.96K0.04V2-xZr (3/4)x(PO4)3/C (x = 0.01, 0.04, 0.07, 0.1) composites and pure NVP/C. Figure 8 shows that all composites possess typical Nyquist characteristics. The linear relationship of Zre and ω−1/2 based on the low frequency region is shown in Fig. 8b. The Na+ diffusion coefficients (DNa+) are calculated according to the equation:

$$ {Z}_{\mathrm{re}}={R}_{\mathrm{s}}+{R}_{\mathrm{ct}}+{\upsigma \upomega}^{-0.5} $$
(1)
$$ D={R}^2{T}^2/2{A}^2{n}^4{F}^4{{\mathrm{C}}_{\mathrm{Na}}}^2{\upsigma}^2 $$
(2)

where σ is the Warburg factor, which is associated with Zre in Eq. (1). Additionally, ω, R, T, A, n, F, and CNa are frequency, gas constant, absolute temperature, surface area of the cathode, number of electrons per molecule during oxidization, Faraday constant, and the concentration of sodium ions (3.47 × 10−3 mol cm−3), respectively. The relationship of Zre and ω−1/2 can be fitted by a straight line whose slope is σ in Eq. (2). The calculated DNa+ is listed in Table 3. The DNa+ of KZ0.07 composite is approximately one order of magnitude higher than that of NVP/C. The increased DNa+ may be owing to the slightly enlarged unit cell volume, which expands the Na+ migration pathway and facilitates the insertion/extraction of Na+. Moreover, the charge transfer resistance of Na2.96K0.04V2-xZr (3/4)x(PO4)3/C(x = 0.01, 0.04, 0.07, 0.1) is significantly lower than that of the pure NVP/C sample. It means that the doped K+ and Zr4+ can significantly improve the apparent and intrinsic electrical conductivities of NVP. In particular, the lowest Rct value of KZ0.07 suggests that it has the best electrochemical properties among all the samples, which is in accordance with the results in Fig. 7.

Fig. 8
figure 8

a Nyquist plots and fitting curves of Na2.96K0.04V2-xZr (3/4)x(PO4)3/C (x = 0.01, 0.04, 0.07, 0.1) and pure NVP at the initial state. b The corresponding profiles of the relationship between Zre and ω−1/2

Full size image
Table 3 The calculated sodium diffusion coefficients of all composites
Full size table

In addition, we compared the electrochemical performance of KZ0.07 with modified NVP materials by other teams in recent years [34,35,36,37,38,39]. As shown in Fig. 9, it is obvious that the KZ0.07 prepared by this work has the highest specific discharge capacity after a long cycle at a high rate, indicating that it has an excellent electrochemical property.

Fig. 9
figure 9

Comparison of specific capacity of modified samples

Full size image

Conclusions

In summary, the modified Na2.96K0.04V2-xZr (3/4)x(PO4)3/C(x = 0.01, 0.04, 0.07, 0.1) composites have been prepared by a solid-phase method in this study. The investigation of influence resulted from doped K+ and Zr4+ on the Na3V2(PO4)3 is explored in detail. The physicochemical and electrochemical performances for the obtained powders are analyzed by XRD, XPS, SEM, CV, and EIS. The results indicate that the co-doping does not destroy the crystal structure of Na3V2(PO4)3. The co-doped strategy can not only expand and stabilize the migration pathway of Na+ but also improve the apparent and intrinsic electrical conductivities. Accordingly, the co-doped materials exhibit better electrochemical performance than that of the undoped one. Especially, KZ0.07 composite shows the best performance among all the co-doped systems. Hence, the K+ and Zr4+ co-doping and combining with carbon coating have a practical application prospect to synthesize the high-rate and long-life Na3V2(PO4)3 for SIBs.