Study of the lithium diffusion properties and high rate performance of TiNb6O17 as an anode in lithium secondary battery

TiNb6O17 and TiNb2O7 were synthesized using a solid-state method. The techniques were used to assess the electrochemical performance and lithium diffusion kinetics of TiNb6O17 related to the unit cell volume with TiNb2O7. The charge-discharge curves and cyclic voltammetry revealed TiNb6O17 to have a similar redox potential to TiNb2O7 as well as a high discharge capacity. The rate performance of TiNb6O17 was measured using a rate capability test. SSCV and EIS showed that TiNb6O17 had higher lithium diffusion coefficients during the charging. From GITT, the lithium diffusion coefficients at the phase transition region showed the largest increase from TiNb2O7 to TiNb6O17.

Therefore, this study examined the accurate lithium diffusion kinetics and electrochemical performance of TiNb 6 O 17 compared to TiNb 2 O 7 which has the smallest unit cell volume among the TNO materials and can clearly be compared with TiNb 6 O 17 . The materials were synthesized using a solid-state method. For electrochemical analysis, the charge-discharge curves and rate capability tests were conducted to determine their electrochemical performance. To examine the lithium diffusion kinetics, SSCV, electrochemical impedance spectroscopy (EIS), and a galvanostatic intermittent titration technique (GITT) were used. As a result, TiNb 6 O 17 showed higher discharge capacity (284mAh/g vs. 264mAh/g) and better rate performance than TiNb 2 O 7 (82mAh/g vs. 20mAh/g at 30 C). In addition, TiNb 6 O 17 showed higher lithium diffusion coefficients than TiNb 2 O 7 (mean value 10 -12 S 2 /m vs. 10 -13 S 2 /m).

Experimental
Synthesis of the active materials and characterization. TiNb 2 O 7 and TiNb 6 O 17 were synthesized by a solid-state reaction method using TiO 2 (99.9%, Rare Metallic) and Nb 2 O 5 (99.99%, Sigma-Aldrich) powders as the starting materials. TiO 2 and Nb 2 O 5 were mixed by ball milling at a stoichiometric molar ratio for 4 h at 300 rpm. The mixed powder was pressed into pellets and calcined in air 1300 °C for 12 h (5 °C/min). The morphology and Ti and Nb content in the two TNO materials were observed by field-emission scanning electron microscopy (FE-SEM, Jeol JSM6500F) and energy dispersion spectroscopy (EDS) attached to FE-SEM. The crystalline structures of the materials were analyzed by X-ray powder diffraction (XRD, Rigaku, Ultima4) was conducted using Ka1 radiation at 45KV/40 mA in the range, 10-100° (2θ). Fourier-transform infrared spectroscopy (FT-IR, Shimadzu IR AFFInity-1S) and X-ray photoelectron spectroscopy (XPS, ThermoFisher K-alpha) were used to examine the chemical bonding and oxidation state of the TNO materials, respectively.
Coin cell assembly and electrochemical analysis. The composition of the TNO anodes was a mixture of active material (TiNb 2 O 7 or TiNb 6 O 17 , 70 wt. %), conducting agent (Super-P, 20 wt. %), and polyvinylidene fluoride binder (PVdF 5130, 10 wt. %). The materials were mixed by ball-milling in 1-methyl-2-pyrrolidinone (NMP) until a viscous slurry formed and cast on Cu foil. The electrochemical properties were tested in CR2032-type coin cells. The cells were assembled with a TNO electrode as the working electrode and lithium metal as the counter electrode separated by a membrane with polypropylene in an Ar-filled glove box. The electrolyte was 1 M LiPF 6 dissolved in a mixture of ethylene carbonate (EC) and dimethyl carbonate (DMC) with a volume ratio 1:2. Cyclic voltammetry (CV) was conducted using a battery cycler (Won A tech, WBCS3000) at a scan rate of 0.1mVs −1 and ranging from 0.05-0.3 mVs −1 from 3.0 to 1.0 V (versus Li/Li + ). Galvanostatic charge-discharge tests were performed using the battery cycles at 0.1 C (38.7mAg −1 of TiNb 2 O 7 and 39.7mAg −1 of TiNb 6 O 17 ) from 3.0 to 1.0 V. The rate capabilities were conducted over the voltage range of 3.0-1.0 V with a current density range 1.0 C to 30 C at room temperature. EIS was carried out by applying an AC signal of 5 mV amplitude over the frequency range from 100KHz to 10mHz using an electrochemical analyzer (NeoSience, SP-300). GITT was tested at a current density of 0.1 C over the voltage range of 3.0-1.0 V using the electrochemical analyzer. The procedure of GITT consisted of galvanostatic charge pulses for each duration time (15 min), followed by a relaxation time (30 min).

Results and Discussion
Characterization. Figure 2 A comparison of the particle size and morphology was not accurate due to irregular particle formation by solid state synthesis. On the other hand, the morphologies of the two materials were similar in principle. The mean particle size of the two samples was approximately 1-3 μm. Figure 2(c)~(f) and (g)~(j) present SEM images of (c) TiNb 2 O 7 and (g) TiNb 6 O 17 (magnification ×5,000) and EDS mapping images of (d) oxygen, (e) titanium, and    12 . The BET specific surface area and volume of the TNO materials were studied by nitrogen adsorption techniques; Fig. 4(b) shows the corresponding isotherm. The specific surface area of TiNb 2 O 7 and TiNb 6 O 17 is 2.66 m 2 /g and 2.36 m2/g; the mean pore volume of the materials is 0.11 cm 3 /g and 0.10 cm 3 /g respectively. As the measurement was conducted by using standard multi point BET, the specific surface area of two materials is almost same. The results are corresponded to the SEM images showing similar particle size of two materials. Therefore, the surface area of the electrodes made by two TNO materials is also same and have not an effect on the electrochemical analysis such as lithium diffusion analysis.
XPS was used to analyze the chemical oxidation state of Ti and Nb in the samples, as shown in Fig. 5. . The noise of the Ti spectra was attributed to the smaller content than Nb. In particular, the spectra of Ti in TiNb 6 O 17 showed more noise than that of TiNb 6 O 17 . This may be because TiNb 6 O 17 is composed of a lower Ti content than TiNb 2 O 7 . These results match the results of EDS analysis and the mapping images. Figure 5 These regions mean the redox reaction of Ti 4+ ↔ Ti 5+ and Nb 3+ ↔ Nb 4+ , respectively. Region 2 is a two-phase O 17 showed three current peaks at the oxidation and reduction state, respectively. Each peak is expressed in the curves (C p and A p mean the cathodic peaks and anodic peaks). Although the reduction peaks were C p1 (Ti 4+ → Ti 3+ ), C p2 (Nb 5+ → Nb 4+ ), and C p3 (Nb 4+ → Nb 3+ ), A p1 , A p2 , and A p3 mean the oxidation reaction of Nb 3+ → Nb 4+ , Nb 4+ → Nb 5+ , and Ti 3+ → Ti 4+10, 15 . These potential regions of the current peaks were matched with the plateau regions in charge and discharge curves. These results show that the reaction mechanisms of the two TNO materials are the same. In addition, the reaction of Nb 4+ ↔ Nb 5+ , which is corresponded to two-phase regions in the charge and discharge curves, showed the highest current peak area and is regarded as the main reaction. Compared to the CV curves of TiNb 2 O 7 and TiNb 6 O 17 , TiNb 6 O 17 exhibits higher reactivity and reversibility from the peak area at all cycles. In addition, the decrease in the peak intensity during the cycle, particularly A p2 and C p1 , suggests that the reversibility of TiNb 6 O 17 is better than TiNb 2 O 7 . This is in agreement with the results of the charge and discharge tests.
To understand the electrochemical performance of the lithium diffusion properties of TiNb 2 O 7 and TiNb 6 O 17 , the rate capabilities were performed at various C-rates from 1 C to 30 C (discharge rate was fixed at 1 C). Figure 7 presents the rate performance of the two TNO materials. A comparison with the average capacities for the 5 th cycle at each C-rate revealed TiNb 6 O 17 to have charge capacities of 252, 230, 206, 187, 107, and 80 mAhg −1 at 1 C, 2 C, 5 C, 10 C, 20 C, and 30 C, respectively. These values are larger than that of TiNb 2 O 7 (234, 210, 174, 152, 52, and 19 mAhg −1 ). In particular, the difference in the charge capacities at a high rate (20 C and 30 C) was distinct. When calculating the ratio of the average charge capacity, 30 C/1 C, the ratio was 8.12% for TiNb 2 O 7 and 31.7% for TiNb 6 O 17 , which suggests that TiNb 6 O 17 has better rate properties than TiNb 2 O 7 13 . In addition, a comparison of the cycling retention at 5 C to 30 C revealed TiNb 6 O 17 to have better cycling properties, whereas TiNb 2 O 7 exhibited a rapid decrease in capacity. This means the better electrochemical reversibility of the TiNb 6 O 17 . These studies including the results of the charge and discharge tests and CV indicated that lithium ion transport of TiNb 6 O 17 is faster than the rate of TiNb 2 O 7 due to the larger theoretical capacity and better lithium diffusion kinetics by larger lithium site. Figure 8 presents the CV data of (a) TiNb 2 O 7 and (b) TiNb 6 O 17 at various scan rates in the range, 0.05-0.3 mVs −1 . CV at various scan rates is usually used to study the oxidation and reduction properties in electrochemical reactions and obtain the apparent chemical diffusion coefficient of Li-ions [16][17][18][19][20] . With increasing scan rate, the anodic peaks move to a low potential and the cathodic peaks move to a high potential due to the increasing  Fig. (a) and nitrogen adsorption-desorption isotherm in Fig. (b) of two TNO materials.
polarization. In addition, the peak intensities of anodic and cathodic reaction increase with increasing scan rate. The peak current density (I p ) revealed a linear relationship with the square root of the scan rate (v −0.5 ), which is expected for a diffusion-controlled process in Fig. 8 17 20-22 . Each color means the linearity of three anodic and cathodic peaks (Black: A p1 and C p1 , Pink: A p2 and C p2 , and Purple: A p3 and C p3 ). The relationship and chemical diffusion coefficient can be determined from the Randles-Sevcik equation (Eq. 1) 16,17,23 : where n is the charge transfer number; F is Faraday's constant; C Li + is the Li-ion concentration in TiNb 2 O 7 and TiNb 6 O 17 ; S is the surface area per weight of active materials; R is the gas constant; and T is the absolute temperature (K). + D Li is the Li-ion diffusion coefficient, and v is the scan rate. In this study, + D Li around three anodic and three cathodic peaks in Fig. 6(c) and (d) was calculated using the above equation.  (12 and 15 times). In addition, the anodic and cathodic reaction of the TNO anodes means the de-lithiation and lithiation process during oxidation and reduction, respectively. Therefore, the lithium diffusion properties of TiNb 6 O 17 were better than those of TiNb 2 O 7 . The reason is that TiNb 6 O 17 has a larger unit cell volume and more open Li-ion sites than TiNb 2 O 7 . The advanced crystal structure of TiNb 6 O 17 leads to a larger size and number of Li-ion transport paths in the crystal structure, facilitating Li-ion transport during the de-lithiation and lithiation processes 10,12,18 .  Figure 9 presents the Nyquist plots of TiNb 2 O 7 and TiNb 6 O 17 by EIS. EIS has been used to examine electrode materials because it can reveal the relationship between the crystal lattice with the electrochemical properties [24][25][26][27][28][29] . This technique provides kinetic information that can be related to a specific state-of-charge or discharge (SOC,  SOD), because the measurement is run by applying a low amplitude signal around an equilibrium state [26][27][28][29] . Figure 9(a) shows the Nyquist plot of TiNb 2 O 7 and TiNb 6 O 17 at the open circuit voltage (OCV) and an equivalent circuit (insert image). Each Nyquist plot was composed of a high-frequency semicircle and Warburg tail region followed by a steep sloping line in the low-frequency region 27 . The R 1 and C dl are the ohmic resistance between the electrolyte and surface of the electrode and double layer capacitance. The high-frequency semicircle means the charge-transfer resistance (R ct ) relevant to the interfacial Li-ion transfer. The Z w is the Warburg impedance, which is related to Li-ion diffusion to the structure of the active materials and corresponds to the tail at a low frequency. Compared to R ct , the TiNb 6 O 17 anode shows a smaller R ct (58Ω) than that of the TiNb 2 O 7 cell (85 Ω). This means that the TiNb 6 O 17 anode has a faster Li insertion process in the surface area than TiNb 2 O 7 . Figure 9(b) presents a plot of the real part resistance with the inverse square root of the angular speed in the low-frequency range of TiNb 2 O 7 and TiNb 6 O 17 anodes at the OCV state. The Warburg factor (σ) is determined from the slope, and is substituted using equation (Eq. 2 and 3):   where Z' is the real part resistance; ω is the angular frequency; R is the gas constant; T is the absolute temperature; A is the surface area of the electrode; F is the Faraday constant; and C is the molar concentration of Li ion in an active material. Equations (2) and (3) were used to calculate the Warburg factor and lithium diffusion coefficient, respectively.   Nb 4+ → Nb 5+ (1.68 V), and Ti 3+ → Ti 4+ (1.98 V) in Fig. 6 (c),(d). Before the EIS experiments, the discharge and charge were processed during 1 cycle and the discharge was then conducted to the cut off potential of 1.0 V. Figure 9(c) TiNb 2 O 7 and (d) TiNb 6 O 17 present Nyquist plots of the two anodes from EIS (Inert images: plot of the real part resistance with the inverse square root of angular speed in the low-frequency range at three oxidation potential). The calculated + D Li value is listed in Table 2 with a value at the OCV state. Compared to + D Li of two TNO anodes from EIS, TiNb 6 O 17 showed higher + D Li values of 2.94 × 10 −13 cm 2 s −1 , 1.12 × 10 −11 cm 2 s −1 , and 1.85 × 10 −12 cm 2 s −1 at 1.36 V, 1.68 V, and 1.98 V, respectively, than those of TiNb 2 O 7 (6.64 × 10 −14 cm 2 s −1 , 1.12 × 10 −11 cm 2 s −1 , and 4.57 × 10 −11 cm 2 s −1 ). The  (20, 12, and 38 times at A p1 -1.36 V, A p2 -1.68 V, and A p3 -1.98 V, respectively, from SSCV) but exhibited similar tendency showing higher + D Li at all redox potentials than TiNb 2 O 7 . In particular, the gap of + D Li at 1.68 V (A p2 peak at CV) meaning that the two phase regions coincide well with the results of SSCV (14 and 12 fold, respectively.) Therefore, TiNb 6 O 17 has better lithium diffusion properties than TiNb 2 O 7 due to its structure inducing a larger open lithium site and a number of Li-ion transport paths during the charge processes.
GITT was conducted to determine the Li + chemical diffusion coefficient and analyze the phase transition of the two TNO materials. The techniques developed by Weppner and Huggins assumed one-dimensional diffusion in a solid solution electrode and a uniform current distribution throughout the electrode and estimated the electrochemically active area from the structure of the active material particles not for the diffusion reaction between the electrode surface and electrolyte [31][32][33][34][35][36][37] . At the transitional GITT, a small constant current was applied to an electrode during a short time and the electrode was left to stand after reaching the OCV state [31][32][33][34][35] . In this study, GITT was performed on the TNO materials to determine the + D Li at a single phase and two phase region as a function of the voltage for the cut off range of the charge/discharge cycle, 1.0-3.0 V. Figure 10 shows the GITT curves of (a) TiNb 2 O 7 and (b) TiNb 6 O 17 during the second cycle. The cells were first discharged at a constant current (0.1 C) for a duration time of 15 min and a rest time of 30 min. The curves showed a similar shape and exhibited three plateau regions meaning the solid-solution regions (Ti 4+ ↔ Ti 5+ and Nb 3+ ↔ Nb 4+ ) and two-phase reaction (Nb 4+ ↔ Nb 5+ ) with the charge-discharge curves. These regions also showed the cyclic voltammetry peaks of C p1 (Ti 4+ → Ti 3+ ), C p2 (Nb 5+ → Nb 4+ ), and C p3 (Nb 4+ → Nb 3+ ); A p1 , A p2 , and A p3 mean the oxidation reaction of Nb 3+ → Nb 4+ , Nb 4+ → Nb 5+ , and Ti 3+ → Ti 4+ in Fig. 6 where V M is the molar volume of the active material; M B is molecular weight of the materials; m B is the mass of the active materials in an electrode; L is the lithium diffusion distance (thickness of the electrode); A is the electrode area; and τ is the duration time. When the change in cell voltage with duration time exhibited a linear relationship on plotting against τ 1/2 , equation (4) can be changed to the following simple equation 32 This equation assumes that the molar volume (V M ) is stable with the change in Li content in an active material. In this study, the Li + diffusion coefficient of the two TNO materials could be calculated, as shown in Fig. 10. (c)-(e). Figure 10.(e) shows the linear relationship between the single steps in Fig. 10. (c),(d). The Li + diffusion coefficients of the two TNO materials from the GITT results are presented as a function of SOC (%) vs. Log ( + D ) Li during the charge state in Fig. 10. (f). The coefficients were calculated at all steps during the GITT measurements except for the 1 st step and the end two steps of the end due to the large voltage variations. The two cells showed three minimum Li + diffusion coefficient points in Fig. 10.(f) and the voltages representing the points are shown. These minimum diffusion coefficients suggest a phase transition for strong attractive interactions between the intercalation species and the host matrix or some order-disorder transition during cycling 24,32 17 and also corresponds to the EIS results. These trends suggest that the oxidation reaction is a two phase transition region of TNO materials with the charge-discharge curves and cyclic voltammetry results (the most reaction region). In the event of SSCV, the measurements showed a different tendency with EIS and GITT. The coefficients of the (Nb 4+ → Nb 5+ ) reaction (A p2 ) showed the largest values and the diffusion coefficients of the (Nb 3+ → Nb 4+ ) reaction showed the largest increase from TiNb 2 O 7 to TiNb 6 O 17. This may be due to the inaccuracy of the SSCV measurements in this study. Compared to A p2 , both A p1 and A p3 showed a small peak current and a broad shape. Therefore, the diffusion coefficients of the two peaks may be not precise values.

Conclusions
Galvanostatic charge-discharge, cyclic voltammetry, and rate capability test were conducted to analyze the electrochemical performance and properties of TiNb 6 O 17 and TiNb 2 O 7 . From the results, two TNO materials showed three similar plateau regions and three redox peaks corresponding to two Nb redox and one Ti redox reaction. TiNb 6 O 17 showed higher capacities of 284mAh/g than that of TiNb 2 O 7 264mAh/g. In the rate capability test, TiNb 6 O 17 exhibited improved rate capacity of 80mAh/g at 30 C than 19mAh/g for TiNb 2 O 7 . SSCV, EIS, and GITT measurement were taken to investigate the performance and lithium diffusion properties related to the unit cell volume of the two TNO materials. The anodic and cathodic Li + diffusion coefficients from SSCV were in the range of 10 −14 to 10 −15 cm 2 s −1 for TiNb 2 O 7 and 10 −13 to 10 −14 cm 2 s −1 for TiNb 6 O 17 . The anodic diffusion coefficients of TiNb 6 O 17 were 5 times (Nb 3+ → Nb 4+ ), 15 times (Nb 4+ → Nb 5+ ), and 14 times (Ti 3+ → Ti 4+ ). From the EIS measurement, the coefficients were in the range of 10 −12 to 10 −14 cm 2 s −1 of TiNb 2 O 7 and 10 −11 to 10 −13 cm 2 s −1 of TiNb 6 O 17 at the OCV state and three oxidation potential region of the two TNO materials during the charging process. The three minimum diffusion coefficients points were determined from the GITT measurement. The diffusion coefficients of the two phase transition region (Nb 4+ → Nb 5+ ) were improved 10 fold compared to that of TiNb 2 O 7 . CV, EIS, and GITT indicated that TiNb 6 O 17 has better lithium diffusion kinetics and electrochemical performance than TiNb 2 O 7 because of its large unit cell volume and more open Li + insertion site.