The rapid development of hybrid electric vehicles and largescaled grid systems force us to explore improved energy storage devices with high energy density and stable cycling performance [1-5]. Among them, lithium-ion batteries (LIBs) are widely used in portable devices owing to their high specific energy density and high power output [6-10]. Currently graphite is the commonly used anode in LIBs, but it suffers from poor rate capability and low lithiation potential (~0.1 V); the low lithiation potential may arouse the safety concern regarding the formation of lithium dendrites when operated at high current density [11-15]. For searching safety anode materials, lithium titanium oxide (Li₄Ti₅O₁₂) is a promising candidate, which affords a flat discharge plateau of ~1.55 V and excellent cycling performance, however, the low electronic conductivity of Li₄Ti₅O₁₂ limits its practical application [16-18]. So, there is a motivation to finding anode materials with high conductivity and safe operating voltage windows like Li₄Ti₅O₁₂ [12, 16, 19, 20].

Recently, intercalation-type materials have been considered as a promising electrode materials for LIBs due to their high energy and superior cycle stability [21-25]. Among them, the niobium pentoxide (Nb₂O₅) has attracted great attention due to its relatively inexpensive and environmentally benign. Although, Nb₂O₅ owns a relative low theoretical capacity of 200 mAh/g, its excellent rate capability could result in a high power density. Moreover, Nb₂O₅ with a voltage plateau of ~1.7 V is safe to avoid the formation of lithium dendrites. In addition, the Nb⁵⁺ with an ionic radius of 64 pm enables a large (001) interplanar lattice spacing of 3.9 Å, which could offer fast lithium ion diffusion, resulting in good rate capability [11, 26-28]. Recently, Dunn's group discovered an extrusive pseudocapacitive Li⁺ intercalation mechanism occurred in orthorhombic Nb₂O₅ as an anode material for LIBs, which played a critical role in the rapid diffusion and storage of lithium ions [24, 29, 30]. Density functional theory (DFT) analysis was conducted to explain the pseudocapacitative behavior of Nb₂O₅ by Ganesh et al., indicating that the high rate intercalation pseudocapacitative behavior was ascribed to the local charge-transfer at all adsorption sites (leading to high energy), open channels to reduce the diffusion barrier for lithium-ions (leading to high rates and power density), and the lack of first-order phase transformation upon lithiation/delithiation (leading to high reversible capacity and good cycling stability) [31]. Therefore, Nb₂O₅ holds great application potential for high-performance LIBs.

The Nb₂O₅ exists mainly in five different crystal structures: amorphous Nb₂O₅ (a-Nb₂O₅), pseudohexagonal Nb₂O₅ (TT-Nb₂O₅), orthorhombic Nb₂O₅ (T-Nb₂O₅), tetragonal Nb₂O₅ (B-Nb₂O₅) and monoclinc Nb₂O₅ (M-Nb₂O₅) [32, 33]. The phase stability is affected by environmental temperature and pressure, and the existence of polymorphs depends on the modes of heating treatment, precursors, and preparation methods [34]. Nb₂O₅ with different crystal structures showed different intercalation/deintercalation characteristics [34, 35]. For example, Le Viet et al. found that the MNb₂O₅ delivered highest capacity (~242 (±3) mAh/g) compared with that of T-Nb₂O₅ and TT-Nb₂O₅, and the capacity of T-Nb₂O₅ was superior than that of TT-Nb₂O₅ [36]. While, Cao and his workers indicated that the TT-Nb₂O₅ delivered an obviously higher specific discharge capacity of 191 mAh/g than that of a-Nb₂O₅ (149 mAh/g) and T-Nb₂O₅ (152 mAh/g) [17]. Griffith et al. also showed different results based on different Nb₂O₅ phase structures: the TT-Nb₂O₅ (< 150 mAh/g at 1 A/g) and T-Nb₂O₅ (150 mAh/g at 1 A/g) showed better high-rate performance, while the MNb₂O₅ delivered a higher specificcapacity (235 mAh/g at 20 mA/g) but with inferior rate (< 50 mAh/g at 1 A/g) capability [34]. These results all showed that the electrochemical performances of Nb₂O₅ are highly related on the phase structures. However, there is little research on controlling the morphology, size and phase of Nb₂O₅ through a simple method and exploring their effects on lithium storage characteristics.

In this work, we have successfully synthesized TT-Nb₂O₅ and TNb₂O₅ by solvothermal method, followed by a thermal conversion process. The crystal structures of the synthesized Nb₂O₅ were related to the annealing temperatures. The TT-Nb₂O₅ material can be obtained at a relatively low temperature (500 ℃ or 600 ℃), and the T-Nb₂O₅ nanoparticles were obtained at high temperature (700 ℃). The synthesized T-Nb₂O₅-700 nanoparticles demonstrated the best electrochemical performance. For example, it showed the higher capacity of 175 mAh/g than that of TT-Nb₂O₅-500 nanosheets (127 mAh/g) and TT-Nb₂O₅-600 nanoparticles (39 mAh/g) after 100 cycles at a current density of 50 mA/g and delivered good rate performance with 219 mAh/g at 50 mA/g and a capacity of 140 mAh/g at 1.0 A/g. The excellent rate capability and cycling stability of T-Nb₂O₅-700 could be attributed to its dominant contribution of pseudocapacitive effect in the high charge/discharge current densities.

Pseudohexagonal and orthorhombic Nb₂O₅ samples were synthesized via a simple hydrothermal method followed by a thermal treatment, which is similar to previous report [37].1 mmol niobium chloride was dissolved in 20 mL ethanediamine and 15 mL distilled water with continuous stirring at room temperature for 50 min. This precursor solution was slowly poured into the Telfonlined autoclave, which kept at 180 ℃ for 24 h in oven. After cooling to room temperature naturally, the Nb₂O₅-precursor was centrifuged and rinsed with DI water and ethanol thoroughly for 3 times and dried at 65 ℃ for 12 h. To prepare different phase structures of Nb₂O₅, the dried Nb₂O₅-precursor was annealed at 500 ℃, 600 ℃, and 700 ℃ for 2 h in a tube furnace in air, respectively.

The crystal structures were confirmed by X-ray power diffraction (XRD) (Rigaku D/MAX-2500) at 50 kV and 250 mA with Cu Kα radiation. The bonding natures and chemical states were analyzed by X-ray photoelectron spectroscopy (XPS) (Thermo ESCALAB 250 spectrometer). Morphologies of the prepared Nb₂O₅ samples were characterized by scanning electron microscopy (SEM) (HitachiS-4800) with an accelerating voltage of 15 kV and transmission electron micoscopy (TEM) (JEOL JEM-3010) at 200 kV. Surface area and pore diameter distribution were performed by Brunauer-Emmett-Teller (BET) measurements using an AutosorbiQ (Quantachrome Instruments).

The Nb₂O₅ electrodes were prepared as follows: active material (70 wt%), acetylene carbon black (20 wt%) and carboxyl methyl cellulose (CMC) (10 wt%) were mixed together in water and ethanol to form a slurry, then the slurry was uniformly coated onto a piece of Cu foil and subsequently dried at 60 ℃ for 24 h under vacuum. The mass loading of active material Nb₂O₅ was controlled at about 1.0-1.5 mg/cm².

The cells were tested in a 2025 type coin-cells. The obtained Nb₂O₅ was employed as cathode, the lithium metal was used as anode with a porous polypropylene separator. For electrolyte, 1 mol/L LiPF₆ in amixture of ethylene carbonate (EC) and dimethyl carbonate (DMC) (V/V = 1:1) was used. All cells were assembled in a glove box with the H₂O and O₂ level below 0.1 ppm. The electrochemical performance measurements were carried out at different current densities by using Neware Battery Testing system. Cyclic voltammetry (CV) and the impedance tests were carried out by using a CHI 660e electrochemical workstation. The cells were operated in the voltage window of 1-2.6 V.

The synthesis process of Nb₂O₅ mainly involved two steps, as shown in Fig. 1. First, the Nb₂O₅-precursor was prepared via a solvothermal method, then pseudohexagonal Nb₂O₅ and orthorhombic Nb₂O₅ were obtained by annealing the Nb₂O₅-precursor at 500 ℃, 600 ℃, 700 ℃ for 2 h in air, respectively, which were labeled as TT-Nb₂O₅-500, TT-Nb₂O₅-600 and T-Nb₂O₅-700. The TTNb₂O₅-500 sample displayed a nanosheets structure, while the TTNb₂O₅-600 and the T-Nb₂O₅-700 samples were shown the nanoparticles morphology. The corresponding crystal structures are also shown in Fig. 1. Structurally, the unit cell of TT-Nb₂O₅ contains an essential defect of an oxygen atom and each Nb atom is at the center of four, five or six oxygen atoms on the ab-plane and an Nb-O-Nb-O chain structure is along the c-axis. T-Nb₂O₅ is comprised of orthorhombic unit cell; six or seven oxygen atoms surround a Nb atom forming distorted octahedra or pentagonal bipyramids [38]. TT-Nb₂O₅ is described as a disordered modification of T-Nb₂O₅ [34]. Some of the oxygen atoms of T-Nb₂O₅ are replaced by impurities such as Cl^- (from residual precursors) or vacancy in the TT-Nb₂O₅ [39]. This unique crystal structure endows T-Nb₂O₅ high lithium storage during the charge/discharge processes [30, 39].

Fig. 2 shows the XRD patterns of the synthesized Nb₂O₅ samples. All the peaks of the TT-Nb₂O₅-500 and TT-Nb₂O₅-600 samples are well accorded with the diffraction peaks of pesudohexagonal Nb₂O₅, (JCPDS-07-0061, a = 3.607 Å, b = 3.607 Å, c = 3.925 Å; space group: P6/mmm), indicating their pesudohexagonal phase structure [40]. While, the T-Nb₂O₅-700 samples belongs to an orthorhombic phase (JCPDS-30-0873, a = 6.175 Å, b = 29.175 Å, c = 3.930 Å; space group: Pbam (55)) [17, 41-43]. Compared with the XRD patterns of the Nb₂O₅-precursor (Fig. S1 in Supporting information), the annealed samples all exhibit better crystallinity. Moreover, the intensity of these diffraction peaks obviously increases with the increase of annealing temperature, which implies an enhanced crystallinity.

The morphology and structure of all prepared Nb₂O₅ samples were investigated by scanning electron microscopy (SEM) and transmission electron microscopy (TEM). As shown in SEM and TEM images (Figs. 3a and b), the TT-Nb₂O₅-500 sample is composed of nanosheets with an average thickness of ~ 17 nm, which is well preserved the original morphology of the Nb₂O₅-precursor, as shown in Fig. S2 (Supporting information). The TTNb₂O₅-600 sample (Figs. 3d and e) and T-Nb₂O₅-700 sample (Figs. 3g and h) both show particle-like structure. Moreover, it is also noticed that with the increase of annealing temperature, the particle sizes of prepared TT-Nb₂O₅-600 and T-Nb₂O₅-700 became bigger. The structural details of all Nb₂O₅ samples were confirmed by high-resolution TEM. The crystalline lattice distance of 0.310 nm in TT-Nb₂O₅ (Figs. 3c and f) are corresponding to the (001) planes, which is consistent with the XRD results. The T-Nb₂O₅-700 shows a parallel lattice fringes with a d spacing of 0.390 nm (Fig. 3i), indicating an orthorhombic phase.

To further characterize differences among these three samples, nitrogen adsorption/desorption analysis was employed, as shown in Fig. 4. The specific surface area of the prepared TT-Nb₂O₅-500, TT-Nb₂O₅-600 and T-Nb₂O₅-700 samples are 92.4, 29.4 and 37.7 m²/g, respectively. Moreover, the pore-size distribution curves obtained through the BJH method indicates that the smaller pore sizes distribution of the T-Nb₂O₅-700 (9.08 nm) than that of TTNb₂O₅-500 (16.98 nm) and TT-Nb₂O₅-600 (17.14 nm). The T-Nb₂O₅-700 nanoparticles show higher surface area and smaller pore size distribution than that of TT-Nb₂O₅-600 nanoparticles, which could be associated with the thermal treatment process in the different temperatures.

The charge/discharge profiles of TT-Nb₂O₅-500, TT-Nb₂O₅-600 and T-Nb₂O₅-700 (Figs. 5a-c) were measured between 1.00 V and 2.60 V at a current density of 50 mA/g. In the first cycle, these three electrodes displayed the similar discharge/charge behaviors with the initial discharge capacities of 353 mAh/g, 260 mAh/g and 270 mAh/g for TT-Nb₂O₅-500, TT-Nb₂O₅-600 and T-Nb₂O₅-700, respectively. From the second cycle, the obtained three samples all show high reversible specific capacities: 249 (TT-Nb₂O₅-500), 146 (TT-Nb₂O₅-500) and 216 (T-Nb₂O₅-700) mAh/g, respectively. The charge/discharge curves are almost overlapped during the subsequent cycles, implying an excellent cycling stability [19]. The capacity drop in the second cycle can be ascribed to the following reasons: the Faradaic contributions of side reaction with the electrolyte and/or the reversible formation of polymer/gel-like film through a pseudo-capacitance-type behavior [44, 45]; the irreversible reactions including the formation of SEI film and further lithium consumption via interfacial reactions due to the charge separation at the metal/Li₂O phase boundary [46-49]. No obvious platform can be observed from the charge-discharge curves, which may be explained by that the electrochemical reactions of Nb₂O₅ with lithium through multiple steps for its decomposition and formation [20]. The electrochemical impedance spectroscopy (EIS) analysis of the Nb₂O₅ electrodes were carried out, and the Nyquist plots are shown in Fig. 5d. It clearly demonstrated that the semicircular diameter of T-Nb₂O₅-700 in the high frequency region is smaller than that of TT-Nb₂O₅-500 and TT-Nb₂O₅-600 samples, indicating the high temperature annealing could improve the conductivity of materials and thus favor the transport of lithium ions.

Fig. 5e exhibits a decreasing trend in specific capacities for all Nb₂O₅ electrodes with the increasing of current density. When the current density gradually increase from 50 mA/g to 1000 mA/g, the specific capacities of the TT-Nb₂O₅-500, TT-Nb₂O₅-600 and TNb₂O₅-700 electrodes drops from 185, 135 and 219 mAh/g to 138, 53 and 140 mAh/g, respectively. When the current density was reversed back to 50 mA/g, the capacities of the TT-Nb₂O₅-500, TTNb₂O₅-600 and T-Nb₂O₅-700 electrodes can be restored to 171, 126 and 199 mAh/g, respectively. The T-Nb₂O₅-700 sample shows the best rate capability among them, which can be ascribed to its unique crystal and mesoporus structure, which could reduce the energy barrier with high charge transfer rate and allow fast lithium ion transportation [50].

Cycling performance was conducted to characterize the stability of the lithium storage behavior of the obtained Nb₂O₅ samples. Fig. 5f shows the cycling performances and coulombic efficiency curves of the TT-Nb₂O₅-500, TT-Nb₂O₅-600 and TNb₂O₅-700 electrodes within a voltage window of 1.0-2.6 V at a current density of 50 mA/g. The initial discharge capacity of TT- Nb₂O₅-500 electrode is 353 mAh/g, which is higher than the initial discharge capacity of TT-Nb₂O₅-600 (261 mAh/g) and T-Nb₂O₅-700 (270 mAh/g) electrodes. The retained capacity of TT-Nb₂O₅-500, TT-Nb₂O₅-600 and T-Nb₂O₅-700 electrodes are 127, 39 and 175 mAh/g after 100 cycles, respectively. Kodama et al. proved the continuous variation of valence state from Nb⁵⁺ to Nb⁴⁺ during the discharge process, as xLi + xe^- + Nb₂O₅ → Li_xNb₂O₅ (x = 0-2) [16]. Thus, the crystal structure changes of Nb₂O₅ could deteriorate the contact between active material and conducting additives, leading to the capacity fading [51].

To further evaluate the electrochemical kinetics of Nb₂O₅ electrodes, CV measurements were conducted at a series of scan rates from 0.1 mV/s to 5 mV/s, as shown in Figs. 6a-c. TT-Nb₂O₅-500 and T-Nb₂O₅-700 show similar CV profiles with broad peaks during both anodic and cathodic processes. Interestingly, when the scan rate increased from 0.1 mV/s to 5 mV/s, peak separations could be almost ignored, which indicates the low polarization at high charge/discharge rates. Therefore, TT-Nb₂O₅-500 and TNb₂O₅-700 present significantly enhanced lithium ion kinetics.

It was reported that the T-Nb₂O₅ displayed significant pseudocapacitive behavior [52-54]. The CV curves with bump shape are consistent with the typical pseudocapacitive Li⁺ intercalation mechanism [55]. To further understand the pseudocapacitive effect of the TT-Nb₂O₅-500 and T-Nb₂O₅-700 electrodes, electrochemical kinetics based on the CV profiles were studied. The capacitive and diffusion-controlled features can be analyzed through the relationship between current (i) and sweep rates (v), as shown by Eq. (1) [30, 56]:

(1)

(2)

(3)

where a and b are the discretionary constants. When the value of b is close to 0.5, the diffusion-controlled process dominates the whole capacity; while if the value of b is close to 1, the capacitive-controlled process arising from the surface faradaic redox reaction dominates the whole capacities [50]. If diffusion-controlled process and capacitive process both work simultaneously in the charge/discharge process, the combined current (i_c) can be expressed as Eqs. (2) and (3) [57].

According to the above two equations, k₁v and k₂v^1/2 presents the capacitive process and the diffusion-controlled process, respectively. The ratio contribution of the capacitive process and diffusion-controlled process can be determined by the ratio of k₁ and k₂. The relationship between the i_c/v^1/2 and v^1/2 could be plotted during the lithiation and delithiation process from the Eq. (3) (Figs. 6e and h), where the slope can be calculated by linear fitting. Fig. 6d shows the CV profile of the TT-Nb₂O₅-500 electrode at a scan rate of 5 mV/s with capacitive contribution (red region), and the capacitive contribution is 78.91%, which is much lower than that (90.89%) of the T-Nb₂O₅-700 electrode (Fig. 6g). This result reveals that more the lithium storage sites in the T-Nb₂O₅- 700 are easily contact with electrolyte to improve the capacitive capacity contribution due to its unique phase and relatively high surface area. The capacitive contributions of the TT-Nb₂O₅-500 and T-Nb₂O₅-700 electrodes were also calculated at different scan rates (Figs. 6f and i), which show an obvious trend. With the increasing of the scan rate, the contribution of capacitive process increase. At a scan rate of 5.0 mV/s, charge storage from capacitive process accounts for 92.4% in the T-Nb₂O₅-700 electrode, which is higher than that in the TT-Nb₂O₅-500 electrode (78.9%), indicating that the capacitive process plays a dominant role in lithium storage in the all Nb₂O₅ electrodes at high scan rates [50, 57]. It was interesting to discover in our work that even at a low scan rate of 0.1 mV/s, the capacitive process still accounts for about 67.8% in the whole charge storage, which is comparatively high because of the high capacitive behavior of T-Nb₂O₅ nanoparticles. Based on the above electrochemical analysis, T-Nb₂O₅-700 nanoparticles with numerous mesopores should contribute additional capacitive capacity in contrast with TT-Nb₂O₅-500 nanosheets, which provides more reaction sites for Li storage and faciles Li transportation through the whole electrode duo to its unique crystal structure among different phases [50]. For TT-Nb₂O₅-600 samples, kinetic analysis was also conducted, given in Fig. S3 (Supporting information).

In summary, we demonstrated the realization of the morphology, size and phase controlled Nb₂O₅ by simply adjusting the annealing process. Electrochemical results indicate that the cycling and rate performance of these materials are dependent on their structures. The T-Nb₂O₅-700 shows the highest capacity of 175 mAh/g after 100 cycles at a current density of 50 mA/g. This outstanding performance mainly contributed to the pseudocapacitive effect during cycling due to the unique crystal phase. Moreover, the TT-Nb₂O₅-500 nanosheets exhibits better cycling and rate capacity than TT-Nb₂O₅-600 nanoparticles owing to its high surface area related to the morphology and size.

Acknowledgments

This work was supported by the National Natural Science Foundation of China (No. 51302079), and the Natural Science Foundation of Hunan Province (No. 2017JJ1008).

Appendix A. Supplementary data

Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.cclet.2018.11.018.