Comparison of amorphous , pseudohexagonal and orthorhombic Nb 2 O 5 for high-rate lithium ion insertion †

Amorphous, pseudohexagonal and orthorhombic Nb2O5 nanoparticles were synthesized using a facile and green sol–gel process followed by thermal treatment at 450 °C, 600 °C and 800 °C for 3 h in air, respectively. The resulting materials have been subjected to a detailed experimental study and comparison of their structural, electrical and electrochemical properties. The experiments have demonstrated that the pseudohexagonal Nb2O5 (TT-Nb2O5) exhibited higher storage capacity, largely due to its high specific surface area and small crystallites, and better cycling performance than both amorphous Nb2O5 (a-Nb2O5) and orthorhombic Nb2O5 (T-Nb2O5); such experimental findings were found to be associated with and thus ascribed to the lower charge transfer resistance and higher lithium ion diffusion coefficient of TTNb2O5 than those of a-Nb2O5 and T-Nb2O5. This research contributes to a better fundamental understanding of the relationship between the crystal structure and the crystallinity and electrochemical properties, particularly Li-ion storage properties, and leads to a possible new advancement in the research field of lithium ion batteries and pseudocapacitors.


Introduction
Electrochemical energy storage devices have attracted considerable attention in powering future advanced communication equipment, portable electronics, electric vehicles and smart grids. [1][2][3] For electric vehicles (EVs), the current commercial lithium-ion batteries (LIBs) commonly using graphite as the anode cannot meet the stringent requirements of high power density, because of poor rate performance and especially safety concerns. 4 Thus, developing advanced electrode materials, which possess better safety characteristics and high rates, becomes imperative. There are several promising candidates including silicon and tin-based compounds being actively studied as anodic materials for the next generation of batteries as they offer high lithium-ion storage capacities; [5][6][7][8] however, they suffer from huge volume changes accompanied by lithium ion insertion and extraction. 9 Lithium titanate (Li 4 Ti 5 O 12 ) is another most noticeable anodic material under intensive study as it offers a high charge/discharge voltage of 1.55 V (vs. Li/Li + ) and very stable cyclic stability since there is a very small volume change during lithium ion insertion and extraction; however, its lithium ion storage capacity is 170 mA h g −1 theoretically. 10 Niobium pentoxide (Nb 2 O 5 ) is another anodic material which attracts a lot of attention lately, due to its unique fast ion diffusion property during lithium ion intercalation and extraction, and most research are based on pseudocapacitors. 11 It has been reported that the high rate is attributed to the fact that lithium ion insertion and extraction is rate-limited by the capacitive effect, not by lithium ion diffusion through the electrode. 12 Compared to commercial graphite anodes, Nb 2 O 5 exhibits a similar insertion/extraction voltage to Li 4 Ti 5 O 12 , which is higher than the 1 V voltage (vs. Li/Li + ) of SEI formation and suppresses lithium dendrite deposition on the surface of the anode. 13 In addition, Nb 2 O 5 has a good theoretical lithium-ion storage capacity of 200 mA h g −1 , 14 which leads to a higher energy density and makes it a good candidate for niche applications.
Due to its attractive physical properties, Nb 2 O 5 has been extensively studied and applied in various technical applications such as electrochemical pseudocapacitors, 11,12 gas sensors, 15 catalysts, 16 electrochromics 17 and solar cells. 17,18 The electrochemical properties of Nb 2 O 5 were first investigated by Bard et al., who demonstrated that Nb 2 O 5 exhibits lithiumion intercalation properties, which led to further research of Nb 2 O 5 as an anode for lithium ion batteries. 19,20 In recent years, various nanostructures of Nb 2 O 5 as electrodes for lithium ion batteries have been studied, such as nanotubes, nanowires, nanofibers and spherical nanoparticles. 14,[21][22][23][24] Some research studies revealed that Nb 2 O 5 -C composites and Ti-or Ta-substituted Nb 2 O 5 possessed improved electrochemical properties. [25][26][27] The Li intercalation properties of other Nb-based oxides are summarized in a recent review paper. 28 11 These research studies have clearly indicated that Nb 2 O 5 polymorphs are very promising lithium ion storage materials. What's more, fast transport kinetics is another important advantage along with high power density for lithium ion batteries besides large capacity. Thus, more studies based on a detailed comparison of the structural, electrical and electrochemical properties of a-Nb 2 O 5 , TT-Nb 2 O 5 and T-Nb 2 O 5 phase materials applied in lithium-ion batteries becomes necessary.
This paper reports the synthesis of nanostructured Nb 2 O 5 via a facile and green sol-gel process with forced hydrolysis and condensation and investigates the electrochemical properties of a-Nb 2 O 5 , TT-Nb 2 O 5 and T-Nb 2 O 5 nanoparticles, aimed at achieving a better fundamental understanding of the relationship between the electrochemical properties, in particular Li-ion storage properties, and the crystallinity and nanostructures. Specifically, we introduced NH 4 OH and H 2 O 2 into the precursor solution to catalyze and promote rapid hydrolysis and condensation reactions. The resulting Nb 2 O 5 nanoparticles were amorphous in nature and the subsequent annealing at elevated temperatures converted the amorphous niobium pentoxide to the pseudohexagonal phase and orthorhombic crystals. Upon evaluation as anode materials for lithium-ion batteries, the three Nb 2 O 5 phases all showed very stable cycling performance, and the TT-Nb 2 O 5 nanoparticles showed higher lithium ion storage capacity than both amorphous Nb 2 O 5 and T-Nb 2 O 5 . Compared with the results for T-Nb 2 O 5 and M-Nb 2 O 5 reported in the literature, the pseudohexagonal crystal has the advantage of low processing temperature which can save energy in materials processing.

Experimental method
Materials synthesis a-Nb 2 O 5 , TT-Nb 2 O 5 and T-Nb 2 O 5 nanoparticles were synthesized via a facile sol-gel process followed by annealing in air at 450°C, 600°C and 800°C, respectively. In a typical reaction, 1 g of NbCl 5 was dissolved in 2 ml of ethanol, and a clear yellow NbCl 5 EtOH solution was formed. A white precipitate was subsequently obtained from this solution by dropwise addition of 5% dilute ammonia, and the pH value was tuned at 7. 31 The precipitate was isolated from the solution and washed with deionized water four times by centrifugation at 6000 rpm. Then, 4 ml of 30 wt% aqueous hydrogen peroxide was added to the precipitate. The mixture was stirred for 5 minutes in an ice bath, and then a transparent yellow sol was obtained. This sol was dried in an oven at 70°C for 24 hours and then a clear yellow xerogel was obtained. The xerogel was ground into white particles and then annealed under ambient atmosphere at 450°C (a-Nb 2 O 5 ), 600°C (TT-Nb 2 O 5 ) and 800°C (T-Nb 2 O 5 ) for 3 hours.

Materials characterization
The crystalline structure of the particles obtained was characterized by X-ray diffraction (XRD, D8 Bruker X-ray diffractometer with Cu-Kα radiation (λ = 1.5418 Å)) within the range of 10-70°(2θ). X-ray photoelectron spectroscopy (XPS) analysis was performed using a VG Multilab 2000 with Al Kα as the X-ray source. The accelerating voltage and current were 40 kV and 40 mA, respectively. Scanning electron microscopy (SEM, JEOL, JSM-7000F) and transmission electron microscopy (TEM, FEI, Tecnai G2 F20) were used to examine the morphology and microstructure of the as-prepared samples. Thermogravimetric analysis (TGA) was performed on a TG instrument (NETZSCH STA 409C). Nitrogen adsorption-desorption isotherms were measured using a Quantachrome NOVA 4200e system. Samples were degassed at 250°C under vacuum overnight prior to measurements. The specific surface areas and micropore volumes were determined by multipoint Brunauer-Emmett-Teller (BET), t-method and Barrett-Joyner-Halenda (BJH) desorption analyses, respectively.

Electrochemical property measurement
The electrochemical performance of the Nb 2 O 5 polymorphs was tested in coin-type cells, which were assembled using the as-prepared samples in an argon-filled glove-box (Innovative Technology, IL-2GB). The as-prepared active materials were mixed and ground with poly(vinylidene fluoride) (PVDF, Sigma-Aldrich) as a binder and Super P conductive carbon (TIMCAL Graphite & Carbon) as the conductive assistant materials at a respective weight ratio of 70 : 20 : 10 in an N-methyl-2-pyrrolidone (NMP, Alfa Aesar) solution. The asprepared mixture was uniformly spread and pressed onto an aluminum foil and dried in a vacuum oven at 80°C overnight for use as the working electrode. The mass of the active material is approximately 0.5 mg. A pure lithium foil was used as the counter and reference electrodes. A polypropylene membrane film (Celgard 2400) was used as the separator. 1 M LiPF 6 in ethylene carbonate (EC)/dimethyl carbonate (DMC) was used as the electrolyte. Galvanostatic charge/discharge measurements were performed in the voltage range of 1.2 to 3.0 V (vs. Li/Li + ) under different current densities. Cyclic voltammetry (CV) was conducted on an electrochemical analyzer (CH Instruments, model 605C) in the voltage range of 1.2 to 3.0 V (vs. Li/Li + ) at a scan rate of 0.2 mV s −1 . Electrochemical impedance spectroscopy (EIS) studies were performed on a Solartron 1287A in conjunction with a Solartron 1260FRA/impedance analyzer with an amplitude of 5.0 mV in the frequency range from 100 kHz to 0.1 Hz. The half-cells were tested at various current rates based on the weight of the active material.

Results and discussion
When NbCl 5 was dissolved in absolute ethanol, the following reaction would occur, leading to the formation of niobium chloride ethoxide: 32 Continuous stirring would allow the partial removal of HCl through evaporation, promoting reaction (1) further, and also the absorption of H 2 O from air, resulting in partial hydrolysis: When NH 4 OH was subsequently added, it reacted with HCl, increased the pH value, and promoted the hydrolysis and condensation reactions of niobium alkoxide as a catalyst, and the white precipitate of NbĲOH) 5 was subsequently obtained: NH 4 Cl is a white precipitate in ethanol. 32 However, in the present work, no NH 4 Cl precipitate was observed, probably due to the low percentage of ethanol.
Collectively, the purpose of adding NH 4 OH to the niobium oxide sol is to promote both the hydrolysis and condensation reactions, and it consequently resulted in a rapid nucleation process with a high density of nuclei of narrow size distribution. The addition of H 2 O 2 dissolved the precipitate and turned the colour from ivory to yellow; upon drying in an oven at 70°C, the release of the oxygen gas by-product benefits the nucleation process as well. The release and removal of oxygen gas bubbles from the sol helped homogenize the concentration and temperature of the sol so as to ensure a uniform size of the nanoparticles, and also prevented the possible formation of agglomerated nanoparticles. The SEM images shown in Fig. 4b and d have demonstrated the difference between the as-synthesized Nb 2 O 5 nanoparticles synthesized with and without the addition of H 2 O 2 through a solgel process under otherwise the same conditions. The Nb 2 O 5 nanoparticles synthesized with the addition of H 2 O 2 are smaller and are dispersed better, i.e., less aggregated.
Through a thermal treatment process, different Nb 2 O 5 polymorphs were formed using the above-mentioned method. The crystalline structure and crystallinity of the as-prepared materials annealed at different temperatures were examined by XRD, and the results are displayed in Fig. 1. The intensity and shape of the diffraction peaks vary greatly as the annealing temperature changes. The as-prepared Nb 2 O 5 xerogel was amorphous. At the annealing temperature of 450°C, the diffraction peaks are mild, and the pattern has a shape close to that of an amorphous crystal. This might be attributed to the fact that the particles could not change into crystals very well at this annealing temperature. When the annealing temperature is 600°C, pseudohexagonal crystals of Nb 2 O 5 (TT-Nb 2 O 5 ) are formed. All the diffraction peaks are clear and in good agreement with those of TT-Nb 2 O 5 (space group: P6/mmm, JCPDS: 28-317), though the intensity of the peaks is weak, the peaks are broad and the signal-to-noise ratio is low, which are all indicative of small crystal size and/or poor crystallinity. The average crystallite size of TT-Nb 2 O 5 was calculated to be approximately 22.5 nm using the Scherrer equation. Between 600°C and 800°C, there is a crystal transition. When the annealing temperature is raised to 800°C, the diffraction peaks are consistent with the peaks of orthorhombic Nb 2 O 5 (T-Nb 2 O 5 , space group: Pbam, JCPDS: 30-0873). The average crystallite size was estimated to be ∼113.3 nm using the Scherrer equation. The crystallites of the T-phase are significantly larger than those of TT-Nb 2 O 5 , as the annealing temperature is much higher. The high intensity and narrow width of the T-Nb 2 O 5 peaks imply that it crystallized very well at the annealing temperature of 800°C.
In order to study the crystalline water loss and the crystal transition during the annealing process, TGA was carried out in flowing air from 30°C to 830°C with a heating rate of 10°C min −1 and the TGA curve of the pure Nb 2 O 5 xerogel is shown in Fig. 2. From the TGA curve, we can see that the weight loss started with a mild curve from room temperature to around 90°C which is supposed to be the physically adsorbed water molecules. A large mass loss of about 20% from 90 to 300°C mainly involves fragments of organic molecules (C x H y ) and HCl, and it is attributed to the build-up of the inorganic amorphous network. 33 The lower mass loss of 2.2% at around 580°C accompanying the crystallization process is due to a transformation from an uncharacterized niobium pentoxide hydrate, also known as niobic acid or columbic acid (Nb 2 O 5 ·nH 2 O), gel to a Nb 2 O 5 crystal (pseudohexagonal). 31 From 600°C to 800°C, the weight loss is less than 0.5%, which means that although the morphology variation is significant and the crystal structure changes from pseudohexagonal to orthorhombic, the weight loss is inconspicuous. This result agrees well with the works published in the literature. 17,24 In these works, no appreciable weight loss or crystallization peaks (from their differential thermal analysis results) were observed for the crystal structure change from pseudohexagonal to orthorhombic.
XPS analyses were carried out in the region of 0-800 eV to elucidate the chemical composition. Fig. 3 shows the typical XPS spectra taken from the niobium and oxygen regions of TT-Nb 2 O 5 . The binding energy scales of the spectra were calibrated by assigning the most intense C 1s peak a bind-ing energy of 284.8 eV. Fig. 3a reveals that the sample contains the Nb, O and C elements. In Fig. 3b, the Nb 3d level binding energies of TT-Nb 2 O 5 were 207.3 and 210.0 eV, which are attributed to the doublet Nb 3d 5/2 and Nb 3d 3/2 core levels of Nb 5+ , respectively. These binding energies of Nb 5+ ions are in close agreement with the values reported. 34,35 Fig. 3c shows the binding energies of the O 1s peaks. Among these peaks, the sharp one located at 529.6 eV and the low one located at 531.9 eV are assigned to O 2− in Nb-oxides and surface oxygen, respectively. 24 Another O 1s peak located at 530.7 eV could be the peak of C-O from some possible residual groups.
SEM and TEM were carried out to investigate the microstructure and morphology of the as-synthesized samples. As shown in Fig. 4a, a-Nb 2 O 5 annealed at 450°C has smaller particle size, better dispersion and higher porosity. From Fig.  4b, it can be clearly seen that the particles are formed homogeneously after being annealed at 600°C for 3 h in air, and the TT-Nb 2 O 5 particles have a narrow size distribution at around 20 nm. When the powders were annealed at 800°C (Fig. 4c), larger-sized T-Nb 2 O 5 particles of around 100 nm size were obtained. This means that the rise in temperature not only modified the crystalline phase but also increased the size. Fig. 4d presents the SEM image of the Nb 2 O 5 particles without addition of hydrogen peroxide and annealed at 600°C for 3 h in air. In this image, connections between particles can be found. By comparing Fig. 4b and d, it can be clearly seen that the particles in Fig. 4b have much better uniform dispersion, which also proved that the addition of H 2 O 2 can improve the dispersity of Nb 2 O 5 nanoparticles. The high-resolution TEM images provide more detailed structural information as seen in Fig. 5. Compared to a-Nb 2 O 5 shown in Fig. 5a, the lattice fringes of the TT-Nb 2 O 5 and T-Nb 2 O 5 particles can be clearly observed in Fig. 5b and c, suggesting the presence of a well-defined crystal structure. The periodic lattice fringe spacings in Fig. 5b were found to be 0.32 and 0.25 nm corresponding to the interplanar spacings of the (100) and (101)    The presence of pores and the porous distribution of the Nb 2 O 5 polymorphs were confirmed by the corresponding nitrogen sorption isotherms and Barrett-Joyner-Halenda (BJH) desorption analyses which are shown in Fig. 6. The results in Fig. 6a indicated that the pore size of a-Nb 2 O 5 centrally distributed at 3.5 nm which was the only mesopore distribution can be found, and the Brunauer-Emmett-Teller (BET)-derived surface area was determined to be 129.6 m 2 g −1 . Within the TT-Nb 2 O 5 sample, the majority of the pores are around 15 nm (Fig. 6b) and exhibit a very narrow pore size distribution. The BET surface area of TT-Nb 2 O 5 was 57.8 m 2 g −1 , which is a reasonable value providing excellent channels and cavities for complete and homogeneous Li + diffusion and intercalation throughout the material. 36 For the T-Nb 2 O 5 sample (Fig. 6c), no obvious pore size distribution was observed in our study, and the peaks of pore size distributions at 3.5 and 5 nm are very weak. The surface area of T-Nb 2 O 5 turns out to be only 43.0 m 2 g −1 , which is far lower than those of a-Nb 2 O 5 and TT-Nb 2 O 5 .
In order to investigate the storage properties of the Nb 2 O 5 polymorphs, CV, galvanostatic charge/discharge measurements and electrochemical impedance analysis have been performed. The CV curves of the first, second and fifth runs for the a-  (Fig. 7c), the current response of the cathodic sweep increases greatly below 2.1 V (vs. Li/Li + ), and two obvious reduction reactions located at 1.72 and 1.54 V correspond to the valence variation during the intercalation process. On the reverse sweep, an ambiguous broad peak at 1.82 V reveals that the continuous variation from Nb 4+ back to Nb 5+ takes place during the redox process. 30 The cathodic and anodic peaks of TT-Nb 2 O 5 (Fig. 7b) are broad and located at 1.5 V and 1.85 V, respectively. In contrast, the CV peaks of the a-Nb 2 O 5 material (Fig. 7a) are very obscure except for the first cycle which exhibits a small reduction peak at 2.0 V and a wide shift of the CV curve to the other cycles. Fig. 8 displays the charge-discharge profiles of the Nb 2 O 5 polymorphs for the first three cycles at a current rate of 10 mA g −1 in the voltage range of 1.2-3.0 V (vs. Li/Li + ). As shown in Fig. 8a, a-Nb 2 O 5 exhibited an initial discharge capacity of 291 mA h g −1 , but it dropped to 150 mA h g −1 , nearly half of the first discharge capacity, after the second cycle. Similarly, TT-Nb 2 O 5 and T-Nb 2 O 5 showed a decrease in capacity following the first cycle; however, the proportion of reversible capacity in the second cycle is much larger than that for a-Nb 2 O 5 , at around 80% for both TT-Nb 2 O 5 and T-Nb 2 O 5 . a-Nb 2 O 5 exhibited the highest initial discharge capacity and the largest decay in the second cycle. This agrees well with the CV curves in Fig. 7a, which possess a wide shift between the first and second reduction curves, suggesting a large   capacity decrease. In contrast, the CV curves for crystalline Nb 2 O 5 (Fig. 7b and c) are more reversible between the first two cycles. In addition, from Fig. 7a, a mild reduction peak located at 2.0 V revealed that an irreversible chemical reaction with Li could exist. 37 On the other hand, from the TGA analysis in Fig. 2, a-Nb 2 O 5 should involve the loss of 2.2% located at 580°C in the form of hydroxyl groups (O-H), which could bring about an undesirable irreversible electrochemical reaction with Li in the initial cycle. 38,39 The existence of O-H in amorphous materials has been proven in previous reports, and it is also not detected in a fully crystallized material. 40 From the figures of TT-Nb 2 O 5 and T-Nb 2 O 5 , curves of a similar shape can be found which have no plateaus, implying that the charge and discharge processes of the TT-Nb 2 O 5 and T-Nb 2 O 5 samples are very fast, which are in good agreement with the characteristics of pseudocapacitance. 12,41 Among the three samples, TT-Nb 2 O 5 shows the highest reversible charge/ discharge capacities of about 190 mA h g −1 which is much higher than the capacity of 110 mA h g −1 reported by Kodama et al. 30 Such a high capacity might be the result of a high spe-cific surface area and the nanosized crystallites. For the larger-sized T-Nb 2 O 5 , the reversible charge/discharge capacities in the present study are similar to the results reported in the literature. 27 From here, we see that the difference is probably because of the fact that the nanoscale TT-Nb 2 O 5 particles obtained by this sol-gel method can provide a larger surface area for Li + de/intercalation and thereby improve the capacity. 3 More capacities under different charge rates and stabilities are shown in Fig. 9.
The rate performance of the a-Nb 2 O 5 , TT-Nb 2 O 5 and T-Nb 2 O 5 samples at various charge-discharge rates is compared in Fig. 9. The charge-discharge curves show that all these three Nb 2 O 5 polymorphs have very regular performance. No obvious fading can be found after 120 cycles. Among these samples, the TT-Nb 2 O 5 electrode shows the highest capacity at each current rate. The discharge capacities measured in the voltage window from 1.2 V to 3 V are 191, 164, 146, 130, 122 and 114 mA h g −1 at the current densities of 50, 100, 200, 300, 400 and 500 mA g −1 , respectively. More importantly, when the current rate was reversed back to 100 mA g −1 after testing at the abovementioned different rates, the discharge capacity went back up to 172 mA h g −1 which is even higher than the initial discharge capacity at the same rate (164 mA h g −1 ). This implies that the decline in capacity at high rates is limited. After 120 charge-discharge cycles, the TT-Nb 2 O 5 electrode demonstrated a specific discharge capacity of 170 mA h g −1 at a current density of 100 mA g −1 , which is close to the capacity of the hightemperature phase, the pristine monoclinic phase, in a   report. 27 Possessing relatively stable charge-discharge performance, the T-Nb 2 O 5 electrode showed much lower capacities than the TT-Nb 2 O 5 electrode at each current rate mentioned above, and its specific discharge capacity was stable at 110 mA h g −1 during 120 charge-discharge cycles. Apparent fading was observed during the first 20 cycles in the rate capability of the a-Nb 2 O 5 electrode. What's more, the a-Nb 2 O 5 electrode showed even lower capacities at different current rates; especially, when the current rate was 500 mA g −1 , the capacity fell to as low as 50 mA h g −1 . The intercalation reaction is sensitive to doping, preparation temperature and particle size which was also documented in TiO 2 as well as other systems. 42 It should be noted that research on other transition-metal oxides such as Fe 2 O 3 and MnO x showed that the amorphous phase possessed much enhanced performance compared to its crystalline counterpart. 37,43 The reason for these results has been explained to be their unique structures which are beneficial for lithium ion storage. Upon comparing the CV curves between the amorphous and crystalline phases, only small changes could be found such as slight shifts of peak location, peak intensity or/and peak width. However, the main CV curves between the amorphous and crystalline samples are very similar, which implies that, for these two materials, there are no significant mechanism differences in the lithiation process caused by changing the crystalline material into an amorphous one. In contrast, our CV results reveal that there are either some mechanism changes in the lithiation process or some capacity deficiency because the CV curve of a-Nb 2 O 5 significantly changed with a lack of redox reaction peaks compared with that of crystalline Nb 2 O 5 . Those different lithiation mechanisms may explain the result that crystalline Nb 2 O 5 showed better lithium ion storage properties than a-Nb 2 O 5 in the present work.
For the sake of comparison of the electrochemical properties of M-Nb 2 O 5 , the Nb 2 O 5 precursor was annealed at 1100°C in air for 3 h and the resulting monoclinic phase was obtained as confirmed by the XRD patterns shown in Fig. S1a (ESI †). The electrochemical properties (shown in Fig. S1, ESI †) turned out to be in good agreement with those reported in the literature. 24 More specifically, the discharge-charge profiles showed a clear discharge plateau at a potential around 1.65 V, which is the same as that reported in the literature. 24 However, the CV curves were slightly different in appearance, which was likely due to the high scan rate used in the current study as compared with that in the literature (0.2 mV s −1 vs. 0.058 mV s −1 ). The resulting M-Nb 2 O 5 demonstrated a high reversible capacity of 185 mA h g −1 at a low rate of 10 mA g −1 . The capacity was found to be 120 mA h g −1 at a high rate of 100 mA g −1 . The M-Nb 2 O 5 in the current investigation possessed a smaller capacity as compared to TT-Nb 2 O 5 .
To investigate the electrochemical kinetics of the Li + intercalation/deintercalation process within the Nb 2 O 5 polymorphous electrode, EIS measurement was performed over a frequency range of 100 kHz to 0.1 Hz with the results shown in Fig. 10. The impedance data were recorded for fresh cells af-ter 5 discharge-charge cycles at a current rate of 0.5 C. The voltage was 3.0 V under the charging status. As shown in the figure, the Nyquist plots of the a-Nb 2 O 5 , TT-Nb 2 O 5 and T-Nb 2 O 5 electrodes display a single semicircle in the highfrequency region and a straight line in the low-frequency range. The EIS was simulated through Z-view software using the equivalent circuit shown in the inset. 35,44,45 It consists of the solution resistance (R s ), charge transfer (R ct ) resistance, a constant phase element (CPE 1 ) and diffusional component Warburg impedance (Z w ). The semicircle at the highfrequency region is associated with the charge transfer resistance (R ct ). 46,47 The fitted R ct values are 332.9 Ω, 261.5 Ω and 400.9 Ω for the a-Nb 2 O 5 , TT-Nb 2 O 5 and T-Nb 2 O 5 electrodes, respectively.
With the lowest R ct value, TT-Nb 2 O 5 demonstrates better electrical conductivity, which can be a reason for its higher capacity and cycling performance. The straight line in the low-frequency range stands for the Warburg impedance. 48 The diffusion coefficient of lithium ions, D Li , as one of the main kinetic factors in the rate-determining step of the intercalation process, can be calculated using several techniques, such as CV, EIS and the galvanostatic intermittent titration technique (GITT). These testing methods are based on their respective formula derived by using Fick's law and the Nernst equation. The differences mostly come from adopting different boundary conditions, initial conditions and numerical analysis methods. The CV technique can provide quantitative information on the electrode process, and the reversibility, stability and phase transformations during the intercalation and deintercalation reactions. The GITT is established to be a reliable technique for investigating the D Li with highly resolved data for intercalation compounds of varying lithium content or potential. EIS is also a powerful technique for calculating the D Li due to the fact that the low-frequency Warburg contribution of the impedance response is directly related to the lithium-ion diffusion process in an electrode material. 49 Slight differences in the diffusion coefficient may exist when different testing techniques were applied; however, D Li obtained from the same technique can still be used to analyze and compare different systems and different electrode materials. The D Li here was calculated from the plots in the low-frequency region. The equation for the calculation of D Li values by EIS can be expressed as follows: [50][51][52] where T is the absolute temperature, R is the gas constant, n is the number of electrons per molecule during oxidization, A is the surface area, F is Faraday's constant, C Li is the concentration of lithium ions, ω is the angular frequency, and σ is the Warburg factor, which has a relationship with Z′. The Z′-ω −0.5 plots are presented in Fig. 10b. Based on the fitting linear equation in Fig. 10b, the Li-ion diffusion coefficients of a-Nb 2 O 5 , TT-Nb 2 O 5 and T-Nb 2 O 5 were calculated and the results are 1.39 × 10 −21 cm 2 s −1 , 3.86 × 10 −20 cm 2 s −1 and 2.56 × 10 −20 cm 2 s −1 , respectively. The lower surface film and charge transfer resistance, higher Li-ion diffusion coefficient and electrical conductivity of the TT-Nb 2 O 5 electrode lead to its superior rate capabilities and other electrochemical properties. Table 1 presents a summary of BET and EIS data of the three Nb 2 O 5 polymorphs. In general, a larger surface area provides more reaction sites and shorter diffusion paths for ionic transport and electronic conduction, leading to a higher power density. 53 In this work, the crystallinity and crystal structure exhibit an obvious influence on the electrical conductivity as well as the ionic diffusivity. The surface area of a-Nb 2 O 5 is twice as large as that of TT-Nb 2 O 5 ; however, its resistance is lower than that of TT-Nb 2 O 5 . This means that during the process of crystal phase transition between the amorphous and TT phases, better crystallinity significantly reduced the value of resistance. In general, better crystallinity means more ordered arrangement of ions and fewer defects, which collectively offer more ordered passage for charge and mass transport, leading to fast diffusion and high conductivity (low resistance). During the transformation from the amorphous to TT phase, the crystals tend to be perfect. When the crystal phase transformed from the TT phase to the T phase, although the crystallinity improved to some extent, the particle size became larger, too. A larger particle size brings about long diffusion paths for ionic transport which can be a reason for high resistance.
Conclusions a-Nb 2 O 5 , TT-Nb 2 O 5 and T-Nb 2 O 5 nanoparticles were synthesized using a facile and environmentally benign sol-gel process followed by thermal treatment. The size of the TT-Nb 2 O 5 particles is around 20 nm, which is much smaller than that of T-Nb 2 O 5 which is around 100 nm. The Nb 2 O 5 nanoparticles synthesized with the addition of H 2 O 2 are smaller and are dispersed better. Upon evaluation as anode materials for lithium-ion batteries, the a-Nb 2 O 5 , TT-Nb 2 O 5 and T-Nb 2 O 5 electrodes demonstrated specific discharge capacities of 149 mA h g −1 , 191 mA h g −1 and 152 mA h g −1 , respectively, at a current density of 50 mA g −1 . The large capacity of TT-Nb 2 O 5 is most likely attributable to its large specific surface area. All these three Nb 2 O 5 polymorphs have very stable cyclic stability within 120 cycles at a current density of 100 mA g −1 . In addition to its large capacity, TT-Nb 2 O 5 possesses lower charge transfer resistance, a higher lithium ion diffusion coefficient and better electrical conductivity than both a-Nb 2 O 5 and T-Nb 2 O 5 , which might be ascribed to its favourable crystal structure.