Electrochemical behavior of interconnected

0 downloads 0 Views 3MB Size Report
Mar 30, 2017 - high energy and power density at low cost [4]. For instance, with .... 10% fluoroethylene carbonate (FEC) as the electrolyte. Pure Li foil.
Electrochimica Acta 236 (2017) 451–459

Contents lists available at ScienceDirect

Electrochimica Acta journal homepage: www.elsevier.com/locate/electacta

Electrochemical behavior of interconnected Ti2Nb10O29 nanoparticles for high-power Li-ion battery anodes De Pham-Conga , Jinwoo Kimb , Van Tan Trana , Su Jae Kima , Se-Young Jeonga , Jun-Hee Choic, Chae Ryong Choa,* a

Department of Nanoenergy Engineering, College of Nanoscience and Nanotechnology, Pusan National University, Busan 609735, South Korea Department of Materials Science and Engineering and Frederick Seitz Materials Research Laboratory, University of Illinois at Urbana-Champaign, IL 61801, USA c Device & System Research Center, Samsung Advanced Institute of Technology, Samsung Electronics, Suwon 443803, South Korea b

A R T I C L E I N F O

Article history: Received 11 December 2016 Received in revised form 23 March 2017 Accepted 27 March 2017 Available online 30 March 2017 Keywords: Ti2Nb10O29 nanofibers single crystal kinetics diffusion coefficient

A B S T R A C T

We synthesized polycrystalline Ti2Nb10O29 nanofibers (NFs) via a simple post-annealing process of aselectrospun polymeric NFs as an anode material for Li-ion batteries (LIBs). During the first discharge/ charge process, the Ti2Nb10O29 NFs annealed at 900  C exhibited insertion/extraction capacities of up to 344 and 304 mAh g1, corresponding to 19.1 and 16.7 mol Li+ per formula unit, respectively. This material exhibited excellent rate capability (93 mAh g1 at 15 A g1) and a higher average diffusion coefficient (DLi = 1.5  1012 cm2 s1) than Ti2Nb10O29 powder (6.9  1013 cm2 s1). This performance can be attributed to the unique nanostructure of firmly interconnected, highly crystalline Ti2Nb10O29 nanograins, which facilitates the Li+ and electron transport. The kinetics obtained from current-voltage curves indicate a mixture of diffusion-limited and capacitive processes. The suggested electro-spinning/post annealing approach can effectively provide a simple route towards high-quality Ti2Nb10O29 NF-based anodes for high-performance LIBs. © 2017 Elsevier Ltd. All rights reserved.

1. Introduction High-performance Li-ion batteries (LIBs) have been used increasingly in numerous applications, such as portable electronics, hybrid vehicles, and stationary storage systems, in tandem with renewable sources of electrical energy since their introduction to the market by Sony in the early 1990s [1–3]. LIBs have been extensively investigated for high-power applications that require high energy and power density at low cost [4]. For instance, with electric vehicle (EV) applications, high energy density is required for the batteries to fit in the available limited space, as well as for the long-term operation of hybrid or pure EVs. However, current LIBs that use a graphite anode suffer from low kinetic properties at high power due to graphite's low Li diffusion coefficient and safety concerns. As a result, the anode cannot meet the strict requirements required of these applications [5–9]. Indeed, electrode potential below 1 V vs. Li/Li+ during discharge typically leads to the formation of a solid electrolyte interphase (SEI) layer on the

* Corresponding author. Tel.:+ +82 55 350 5297. E-mail address: [email protected] (C.R. Cho). http://dx.doi.org/10.1016/j.electacta.2017.03.203 0013-4686/© 2017 Elsevier Ltd. All rights reserved.

graphite surface during the first charge cycle because electrolyte decomposition leads to irreversible capacity loss and the production of secondary-phase active materials [10,11]. Titanium-based oxides, such as zero-strain Li4Ti5O12 (LTO) with a spinel structure, and TiO2 with anatase or rutile structures are considered to be the alternative to replace graphite because of their high structural and chemical stability and suitable voltage range of 1.0–2.0 V vs. Li/Li+ [12–14]. However, their low theoretical capacities (175 mAh g1 and 335 mAh g1 for LTO and TiO2, respectively) have limited their potential for use in high-power applications. The development of safe materials with high theoretical capacities and ultrafast charging/discharging rates is urgently needed for portable electronic devices and EV applications. Ti-Nb-O compounds with high redox potentials, general chemical formulae, and high theoretical capacities, such as TiNb2O7 (388 mAh g1), Ti2Nb10O29 (396 mAh g1), and TiNb6O17 (397 mAh g1) have recently emerged as new and promising active materials for replacing graphite in LIBs [15–20]. Most Ti-Nb-O compounds have theoretical capacities that are two times higher than that of LTO, and even exceed that of graphite (372 mAh g1). Among these materials, TiNb2O7 and Ti2Nb10O29 have a monoclinic shear ReO3 structure with space group C2/m, consisting of MO6

452

D. Pham-Cong et al. / Electrochimica Acta 236 (2017) 451–459

(M = Ti, Nb) octahedra sharing edges and corners. They show practical reversible capacities of 250–300 mAh g1 in the voltage range of 1.0  3.0 V vs. Li/Li+. The multi-redox couples Nb5+/Nb4+, Nb4+/Nb3+, and Ti4+/Ti3+ can avoid possible Li plating from the Li electrode in half-cell batteries and enable extremely high rates and safe operation because the average voltage in the Ti-Nb-O compound is 1.66 V, which is close to that of LTO (1.55 V). The higher redox potentials of Ti-Nb-O compounds endow them with greater overcharge protection than graphite. However, the development of rechargeable LIBs with Ti-Nb-O anodes has been limited because of these materials’ low electronic and ionic conductivities, and small Li-diffusion coefficients [16]. Introducing nanostructure to obtain high diffusion coefficients and long diffusion lengths may be the key to improve Li+-insertion kinetics [21,22]. Compared with TiNb2O7, few reports of the electrochemical properties of Ti2Nb10O29 exist. Solid-state reaction methods based on commercial TiO2 and Nb2O5 powders have been used as simple methods to synthesize Ti2Nb10O29 anodes in some works, but these methods are limited in producing sub-micrometer size particles because of high-temperature crystallization (1000–1200  C) and long-term calcination (20–40 h) [23–26]. Thus, new and simple processing methods are needed to synthesize Ti2Nb10O29 anodes with excellent performance. One relevant approach is to tailor the morphology and particle size of the Ti2Nb10O29 material. Among the various possible nanostructures, one-dimensional (1D) nanostructures, such as nanotubes, nanowires, and nanorods, have attracted considerable interest because of their rapid electron transport along the 1D direction and short ionic diffusion distance along the radial direction. Numerous methods have been introduced for generating 1D nanostructures, such as soft chemistry and the use of template bases. Of these, electrospinning is a simple scalable method for nanofibers (NFs), hollow NFs, and porous NFs [27,28]. To the best of our knowledge, there has been limited research focusing on the use of Ti-Nb-O compounds as anode materials, and no research relating to Ti2Nb10O29 NFs has been reported. In this study, we report on the synthesis of Ti2Nb10O29 NFs as a high-performance anode material for LIBs. The structural and morphological properties of Ti2Nb10O29 NFs were investigated via a comparison with Ti2Nb10O29 powder. The electrochemical properties, including cyclic voltammetry (CV), charge/discharge behavior, cycling and C-rate tests, and Li+-diffusion behavior, of both Ti2Nb10O29 NFs and powder were studied in detail. 2. Experimental section 2.1. Synthesis of Ti2Nb10O29 NFs The Ti2Nb10O29 NFs were synthesized by a simple electrospinning technique. The precursors, 2.79 g of Nb(V) ethoxide (Nb (OC2H5)5), 0.6 g of Ti(IV) butoxide (Ti(OC4H9)4), and 1.6 g of PVP (Sigma-Aldrich, Mw = 1,300,000) were dissolved in 20 mL of ethanol, and then 4 mL of acetic acid was added. After stirring for 5 h, the prepared transparent solution was transferred into a 12-mL syringe with a 25-gauge blunt needle tip. A voltage of 17 kV was applied between the tip and aluminum foil at a distance of approximately 20 cm. The as-electrospun polymeric NFs were collected by controlling the sol-gel solution with a syringe pump at a flow rate of 0.5 mL h1. The dried polymeric NFs were crystallized to obtain 1D Ti2Nb10O29 NFs by annealing at various temperatures (700, 800, and 900 C) for 7 h with a ramping rate of 2 C min1 chosen based on the thermal decomposition and combustion of the polymer.

2.2. Synthesis of Ti2Nb10O29 powders For comparison with Ti2Nb10O29 NFs, Ti2Nb10O29 powder was synthesized as follows: first, 0.4 g of anatase TiO2 (P25, 2030 nm, Sigma-Aldrich) and 3.325 g of Nb2O5 (Sigma-Aldrich), corresponding to a molar ratio of TiO2:Nb2O5 = 2:5, were mixed in 10 mL of ethanol (Sigma-Aldrich). Then, the dried sample was annealed at 900  C for 20 h with a ramping rate of 1  C min1. 2.3. Materials characterization The sample morphologies were characterized by SEM (Hitachi, S4700) at an operating voltage of 15 kV. The crystal structures of the materials were determined by XRD (PANalytical) in the u-2u scan mode, with 2u ranging from 10 to 90 , at scan rate of 1 min1 using Cu-Ka1 radiation. The samples’ crystalline structures and elemental mapping were collected using TEM equipped with EDX (Jeol, JEM2011). Raman spectra were obtained using a Raman spectrometer (Dongwoo Optron, MonoRa500i) with 532-nm excitation from an Ar laser. The chemical bonding states of the samples were analyzed using XPS (Thermo Fisher Scientific, ESCALAB250) with monochromic Al-Ka (1,486.6 eV) radiation. The binding energies of the spectra were calibrated to carbon (C1s, 284.6 eV). Specific surface area of samples was measured using BET system (Micromeritics, ASAP2010). The samples were pre-treated at vacuum to remove organic residual and isothermal gas adsorption depending on relative pressure was observed at 77 K using N2 gas. 2.4. Electrochemical measurements The electrochemical properties of the Ti2Nb10O29 samples were evaluated using CR2032 coin cells with Li metal as the counter electrode. The working electrode was prepared by spreading a slurry composed of active materials, carbon black, and carboxymethyl cellulose (CMC) (weight ratio of 70:15:15, respectively) using a scalpel on Cu foil and then drying it in a vacuum oven at 100  C for 12 h. The mass of active materials in each working electrode disc (diameter of 14 mm) was calculated to be approximately 0.91.1 mg cm1. The dried sample was punched using a pressure of 4 MPa and inserted into an Ar-filled glove box to fabricate coin cells with 1 M LiPF6 in a mixture of ethylene carbonate and dimethyl carbonate (EC/DMC [1/1 vol%]) containing 10% fluoroethylene carbonate (FEC) as the electrolyte. Pure Li foil was used as the counter electrode, and Celgard 2400 was used as the separator. The electrochemical properties were determined using a multichannel potentiostat/galvanostat (Wonatech, WMPG 1000) over a voltage range from 1.0 to 3.0 V vs Li/Li+ at room temperature and current densities ranging from 0.2 to 15 A g1 (1C = 397 mA g1). CV measurements were performed at various scan rates from 0.1 to 10 mV s1 in the voltage range of 1.03.0 V. The coin cells were disassembled to investigate the surface morphologies and crystal structures of the electrodes after cycling, and the working electrodes were washed and dried in vacuum at 50  C for 12 h. The AC impedance spectra were obtained by applying a sine wave with an amplitude of 5.0 mV over the frequency range from 100 kHz to 0.01 Hz. One of the electrochemical method to determine the Li diffusion coefficient in electrode active material is galvanostatic intermittent titration (GITT) technique [29,30]. The GITT is a useful procedure to retrieve both thermodynamics and kinetics parameters. The GITT curves were collected by applying a constant small current flux of 50 mA g1 for 30 min with a rest time interval of 4 h in the voltage range of 1.03.0 V vs. Li/Li+ at room temperature using the Wonatech battery test system.

D. Pham-Cong et al. / Electrochimica Acta 236 (2017) 451–459

We present a novel strategy to create high-performance anodes and successfully demonstrate a simple electrospinning method to synthesize Ti2Nb10O29 NFs, as shown in Fig. 1. The typical synthetic process mainly included two steps: (i) a solution of polyvinylpyrrolidone (PVP), tetrabutyl orthotitanate, and Nb ethoxide (Nb/ Ti = 1/5 mole ratio) was fluently electrospun to form NFs. The polymer and solvent components were removed by annealing, and then, Ti2Nb10O29 NFs with diameters of 100  300 nm and lengths of tens micrometers were formed, as described in the experimental section. (ii) Ti2Nb10O29 powder was generated by the solid-state reaction of commercial TiO2 and Nb2O5 powders, which were heat treated at 900  C for 20 h in air. 3. Results and discussion Scanning electron microscopy (SEM) images of Ti2Nb10O29 powder revealed that the average particle size was approximately 0.9 mm as illustrated in Fig. S1 (Supporting information, Fig. S1). In contrast, the shape of the as-electrospun NFs was uniform and smooth, and their diameters were 200–400 nm (Supporting information, Fig. S2). After annealing in air at various temperatures (700  C, 800  C, and 900  C) for 7 h, the products were denoted as Ti2Nb10O29-700 NFs, Ti2Nb10O29-800 NFs, and Ti2Nb10O29-900 NFs, respectively (Fig. S2c  h). All the Ti2Nb10O29 NFs formed had dense structures. The morphologies of the Ti2Nb10O29 NFs clearly varied with the annealing temperature. The smooth surface of aselectrospun Ti2Nb10O29 NFs became rough on annealing at 700  C. A unique NF structure consisting of strongly interconnected faceted-grains was formed at 800  C, and an enlarged, faceted grain structure with very smooth surface was formed at 900  C. The annealed NF has a length of about several micrometers with grain size of 50  200 nm. X-ray diffraction (XRD) was performed to evaluate the crystallinity and identify the various phases of the NFs for different annealing temperatures (Supporting information, Fig. S3a). The diffraction patterns annealed at 700  C revealed the existence of anatase-TiO2 and Nb2O5 peaks, which disappeared above 800  C. This means that full phase transformation into Ti2Nb10O29 can be verified above 800  C. The resemblance of diffraction peaks between the powder and the NF suggest high crystallinity of the

453

calcined Ti2Nb10O29 NFs. All of the samples’ diffraction peaks are highly consistent with those of JCPDS card no. 72-0159, corresponding to a monoclinic crystallographic system. The lattice parameters calculated from the diffraction pattern are a = 2.0530 nm, b = 0.3814 nm, c = 1.5565 nm, and b = 113.51, all of which are in good agreement with the reported values [31]. In the C2/m space group of the layered monoclinic structure, the disordered Ti4+ and Nb5+ occupy octahedral sites sharing edges and corners and form an open tunnel-like interstitial space for Li+ insertion. To further understand the crystal structure, the Raman spectra of both Ti2Nb10O29 powder and Ti2Nb10O29-900 NFs were performed (Fig. S3b). The Raman modes measured in the range of 50  1200 cm1 were generally assigned to the edge-sharing and corner-sharing NbO6 octahedra. The peaks located at 1017 and 906 cm1 are attributed to the Nb  O stretching modes, and the Ti  O stretching modes in the edge-sharing and corner-sharing TiO6 octahedra were observed at 648 cm1 and 543  118 cm1, respectively. No secondary phases of Nb2O5 or TiO2 were found, which is in good agreement with the XRD result, suggesting that the reaction was complete under these condition. X-ray photoelectron spectroscopy (XPS) analysis was performed to investigate the chemical bonding states of Ti, O, and Nb in the active anode nanostructures (Supporting information, Fig. S4). All the samples displayed similar Ti 2p, Nb 3d, and O 1 s high-resolution spectra, respectively (Fig. S4b  d). The binding energy (530.1 eV) confirms that the O2 states bound to Nb or Ti within the lattice structure. The Ti 2p3/2 binding energy was 458.5 eV, suggesting the existence and occupation of Ti4+ in an octahedral arrangement [32,33]. Two binding energy peaks were located at 206.8 eV and 209.5 eV and are attributed to Nb 3d5/2 and Nb 3d3/2, respectively, indicating the presence of Nb5+ [34,35]. The microstructures of the Ti2Nb10O29-900 NFs and Ti2Nb10O29 powder were investigated using high-resolution transmission electron microscopy (HRTEM). The NF consists of strongly interconnected faceted grains, where each Ti2Nb10O29 grain is highly single crystalline with sharp surface (Fig. 2a and c). Fig. 2d clearly shows that the intervals between Fourier-transformed lattice fringes in the Ti2Nb10O29-900 NF had d = 0.41 nm and 0.48 nm, corresponding to the (20-5) and (040) planes, respectively. The selected-area electron diffraction (SAED) pattern along the

Fig. 1. Schematic of the synthesis of Ti2Nb10O29 NFs and powders via electrospinning and solid-state reaction methods, respectively.

454

D. Pham-Cong et al. / Electrochimica Acta 236 (2017) 451–459

Fig. 2. HRTEM images and lattice images of (a–d) Ti2Nb10O29 NFs and (e–h) Ti2Nb10O29 powder, respectively. The electron diffraction patterns are displayed in Fig. 2b and f. The insets of Fig. 2d and h show the SAED patterns corresponding to the boxes in Fig. 2c and g. (d, h) the lattice fringes are Fourier-transformed from the SAED patterns.

[010] zone-axis confirms the single crystallinity of the faceted grain (Fig. 2d, inset). An inspection of Fig. 2e  h reveals that Ti2Nb10O29 powder is composed of loosely connected, single crystalline microparticles (Fig. S1d). It should be noted that, due to the particle or grain size difference, SAEDs for nanofibers and powders under large electron beam size yield similar patterns showing polycrystalline as shown in the XRD results. The corresponding energy-dispersive X-ray (EDX) maps of Nb, Ti, and O for Ti2Nb10O29 powder and Ti2Nb10O29-900 NF samples are shown in Fig. S5 (Supporting information, Fig. S5) using a color scale. These three elements are dispersed uniformly throughout the Ti2Nb10O29-900 NFs and Ti2Nb10O29 powder. The redox kinetics of the Ti2Nb10O29 powder and Ti2Nb10O29900 NF anodes were investigated by galvanostatic chargedischarge measurements in the cut-off voltage range of 1.03.0 V vs. Li/Li+ at a scan rate of 0.1 mV s1 (Fig. 3a and b). Three oxidation peaks are observed for both samples. The oxidation peaks during anodic sweeping appeared at 1.19 and 1.91 V and are attributed to the Nb4+/Nb3+ and Ti4+/Ti3+ redox couples, respectively. One pair of sharp anodic and cathodic peaks can be observed at 1.69 and 1.63 V and may be attributable to the valence variation of Nb5+/Nb4+ based on reported results for Nb2O5 [36]. The cathodic peaks at 1.59 V for Ti2Nb10O29 powder and 1.57 V for Ti2Nb10O29-900 NFs in the first cycle and at 1.63 V in the subsequent cycles correspond to the voltage platforms in the discharge process, during which Li+ intercalates into Ti2Nb10O29. These high working potentials can prevent the formation of thick SEI layers and Li dendrites on Ti2Nb10O29 surfaces. However, during the subsequent cycles, the cathodic peaks are positively shifted to a higher voltage relative to the first cycle in both samples because of the polarization effect of the electrolyte during the first cycle [37]. In contrast, the anodic and cathodic peak intensity of Ti2Nb10O29900 NFs is higher than that of Ti2Nb10O29 powder, which can be attributed to its lower activation energy for Li+ insertion, higher electronic conductivity, and smaller particle size [38]. Cyclic voltammetry (CV) was performed to elucidate the Li+storage mechanism of the various Ti2Nb10O29 samples at different scan rates from 0.1 to 10 mV s1 (Supporting information, Fig. S6). As the scan rate increased, the cathodic peaks shifted to higher potentials, whereas the anodic peaks shifted to lower potentials. In addition, the peak intensity exhibited a linear relationship with the square root of the scan rate (v1/2), which corresponds to a typical feature of diffusion-controlled processes, as shown by the RandlesSevcik equation [39]:

Ip = 2.69  105  AC  DLi1/2  n3/2  v1/2

(1)

where Ip is the peak current, A is the electrode surface area, C is the concentration of Li+ in the electrode, n is the number of electrons involved in the redox process, and v is the scan rate. The Ti2Nb10O29-900 NFs possesses the largest slopes in all samples. In addition, the DLi values of Ti2Nb10O29-900 NF electrode calculated from Eq. (1) showed 24.6 times higher during the Li+-insertion process than the Ti2Nb10O29 powder, Ti2Nb10O29-700 NFs, and Ti2Nb10O29-800 NFs (Supporting information, Table S1). We note that the lowest DLi value of the Ti2Nb10O29-700 NFs may be attributed to the incomplete phase transformations of TiO2 and Nb2O5. We hypothesize that smaller particle size and high crystallinity of Ti2Nb10O29-900 NFs might increase DLi. In addition, several processes could contribute to total stored charge, and they can be determined by analyzing the CV at various scan rate. Assuming that the current obeys a power-law relationship with the scan rate, this leads to: I = a  vb

(2)

where a and b are adjustable values. In particular, the b-value of 0.5 represents a total diffusion-limited process, whereas I indicates a capacitive process [40]. By fitting log I vs log v, calculated b-values of Ti2Nb10O29-900 NF for the cathodic and anodic peaks were 0.7 and 0.8, indicating that the kinetics are the mixture of diffusionlimited and capacitive process [41]. The galvanostatic discharge/ charge profiles measured at a low current density of 50 mA g1 (0.13C) over several cycles using Ti2Nb10O29 powder/Li and Ti2Nb10O29-900 NFs/Li cells are presented in Fig. 3c and d. Consistent with the CV results, each discharge/charge curve can be divided into three regions with different slopes: region 1 (opencircuit voltage) from 3.0 to 1.9 V, region 2 from 1.9 to 1.6 V, and region 3 from 1.6 to 1.0 V. The charge and discharge plateaus at 1.63 V and 1.69 V vs. Li/Li+ in region 2 are in good agreement with the two intense peaks of the two-phase reaction shown in the CV curves. The Ti2Nb10O29 powder electrode exhibited first charge (delithiation) and discharge (lithiation) capacities of 278 and 309 mAh g1, which correspond to 15.1 mol and 17.1 mol Li+ extraction/insertion (Coulombic efficiency of 89.9%), respectively. In contrast, the first discharge capacity of the Ti2Nb10O29-900 NF electrode is 344 mAh g1 (19.1 mol Li+), and its charge capacity is 304 mAh g1 (16.7 mol Li+), corresponding to a Coulombic efficiency of 88.4%. The larger specific capacities of the Ti2Nb10O29-900 NF are attributable to the larger contact area

D. Pham-Cong et al. / Electrochimica Acta 236 (2017) 451–459

455

Fig. 3. (a, b) CV curves for various cycles at scan rate of 0.1 mV s1, (c, d) charge-discharge curves under a current density of 50 mA g1 (0.13C), and (e, f) capacity performance as a function of cycle number at a current density of 50 mA g1 for the Ti2Nb10O29 powder and Ti2Nb10O29 NFs.

between the electrode and electrolyte and the shortened Li+ insertion/extraction pathways as compared to Ti2Nb10O29 powder. The overall Li+-insertion/extraction reaction mechanism can be described using the following equation: Ti2Nb10O29 + xLi+ + xe $ LixTi2Nb10O29

(3)

The cycling tests of Ti2Nb10O29 powder and Ti2Nb10O29-900 NFs were performed at a low current rate of 50 mA g1, as shown in Fig. 3e and f, respectively. The Ti2Nb10O29-900 NFs exhibited high structural stability with a reversible capacity as high as 289 mAh g1, even after 35 cycles at 50 mA g1. In contrast, the Ti2Nb10O29 powder showed a lower capacity and higher fading behavior from 275 mAh g1 in the second cycle to 228 mAh g1 in the 35th cycle. The Ti2Nb10O29-900 NFs exhibited only an 8% decrease. Note that the capacity values of both Ti2Nb10O29-900 NFs and Ti2Nb10O29 powder at low current density (50 mA g1) are still much higher than that of LTO. The high Coulombic efficiency ( > 88%) at the first cycle indicates a small initial irreversibility, suggesting that

Ti2Nb10O29-900 NFs and Ti2Nb10O29 powder electrodes can also be operated above 1.0 V without SEI layer formation. The stable Ti2Nb10O29-900 NFs should afford excellent cycle life under severe charge/discharge cycling conditions as shown in Fig. 4. To investigate the cycle life, the charge rate performances of the Ti2Nb10O29-900 NFs and Ti2Nb10O29 powder electrodes were determined with a constant discharge rate of 0.5 A g1 and with gradually increasing charge rates (Fig. 4a). Little capacity loss between 0.5 and 15 A g1 was observed in both samples. The specific capacity of the Ti2Nb10O29-900 NFs (234 mAh g1) is higher than that of the Ti2Nb10O29 powder (203 mAh g1), even at a high charging current density of 15 A g1 (95 s for full charge). Indeed, the charging voltage clearly plateaued even at high current density and after long cycling (Supporting information, Fig. S7). Furthermore, it gradually increases because of electrode polarization as the charging rate increases [42,43]. The Ti2Nb10O29-900 NFs also displays better rate capabilities than the other samples (Fig. 4b, and supporting information in Fig. S8ac). The specific

456

D. Pham-Cong et al. / Electrochimica Acta 236 (2017) 451–459

Fig. 4. Charge-discharge performance of Ti2Nb10O29 powder and Ti2Nb10O29-900 NFs in the voltage range of 1.03.0 V. The specific capacities measured (a) at various charge current densities with discharge fixed at 0.5 A g1, (b) at various charge-discharge current densities in the range of 0.2–15 A g1, and (c) at a current density of 5 A g1 for longterm cycling.

capacities of Ti2Nb10O29-900 NFs are as high as 290, 271, 254, 194, and 138 mAh g1 at current densities of 0.2, 0.5, 1, 5, and 10 A g1, respectively. Even at a high current density of 15 A g1, the Ti2Nb10O29-900 NFs were still able to deliver a charge capacity of 93 mAh g1, exceeding those of the Ti2Nb10O29-800 NFs (29 mAh g1), Ti2Nb10O29-700 NFs (4 mAh g1), and Ti2Nb10O29 powder (11 mAh g1) under this test conditions. From the cyclic performances for the Ti2Nb10O29-700 NFs and Ti2Nb10O29-800 NFs (500 cycles), these samples exhibited strong fading because of their lower crystallinities (Fig. S8d). The samples’ specific capacities were measured according to the cycle number at a low current density of 0.5 A g1 in the voltage range of 1.03.0 V vs. Li/Li+ (Fig. S8e). The specific capacity of the Ti2Nb10O29-900 NFs was 210 mAh g1 in the first cycle, and after 350 cycles, these samples’ values were 1.53.8 times higher than those of the other samples. Fig. 4c shows the cyclic performances and corresponding

Coulombic efficiencies over 2000 cycles at a current density of 5 A g1 (12.6C). The excellent reversible specific capacity of the Ti2Nb10O29-900 NFs reached 212 mAh g1 at a current density of 5 A g1 and was maintained at 109 mAh g1 after 2000 cycles, corresponding to a capacity retention of 49.6%. In contrast, the discharge capacity of the Ti2Nb10O29 powder electrode dropped rapidly to 27 mAh g1 after 200 cycles. The Coulombic efficiency of the Ti2Nb10O29-900 NFs quickly increased to near 100% within 50 cycles and remain the high value afterwards. Moreover, the Ti2Nb10O29-900 NFs also showed excellent cycle performance at a charge current density of 10 A g1 (142 s for full charge) and a discharge current density of 1 A g1 (Supporting information, Fig. S9a). The reversible capacity of Ti2Nb10O29-900 NFs remained at 202 mAh g1, corresponding to 80% retention of the initial capacity after 150 cycles. According to Arrhenius law, temperature significantly influences the performance, safety, and cycle lifetime

D. Pham-Cong et al. / Electrochimica Acta 236 (2017) 451–459

of LIBs [44–46]. The cycling performances of all samples were measured at different temperatures in the range of 2050  C at a current density of 2 A g1 (Fig. S9b). The discharge capacity of the Ti2Nb10O29-900 NFs was more stable and higher than those of other samples in all temperatures. To investigate the structural changes in Ti2Nb10O29 electrodes during cycling, the cells were carefully decapped before and after cycling at a current density of 5 A g1. Fig. S10 (Supporting information, Fig. S10) presents the SEM images and XRD patterns of Ti2Nb10O29-900 NFs and Ti2Nb10O29 powder after 2000 and 600 cycles, respectively. The polycrystalline microsized Ti2Nb10O29 powder became smaller after 600 cycles because of the stress generated by Li+-insertion/ extraction and the enlarged defective layer [47–49]. However, no distinct cracks were observed on the Ti2Nb10O29-900 NF electrode even after 2000 cycles because the stress was effectively released through the radial direction of NFs during the lithiation/ delithiation process. A thin layer on the surfaces of Ti2Nb10O29 powder is believed to be a SEI layer formed by solid-liquid reactions at the interface, causing the fading effects observed for these samples. The XRD patterns of both samples showed no distinct additional peaks after the cycling test. To confirm the Li+ diffusion coefficient (DLi), the GITT was performed at room temperature under a small current flux of 50 mA g1 for 30 min with a rest time interval of 4 h in a voltage range of 1.0–3.0 V (Fig. 5a and b). GITT can provide more precise DLi than electrochemical impedance spectroscopy (EIS) [50–52]. DLi can be calculated using the following equation, which is based on Fick’s second law of diffusion: DLi = [4(mB  Vm)2  (DEs)2]/[pt(MB  A)2  (DEt)2]

(4)

457

where MB and mB are the relative formula mass and active mass in the electrode, Vm is the molar volume of the material, A is the contact area between the electrolyte and the electrode, DEt is the total change in cell voltage during the application of the current flux, and DEs is the steady-state voltage throughout the galvanostatic titration, respectively. The overall DLi was calculated as a function of the voltage during the charge-discharge process from GITT data (Fig. 5c and d). The discharge DLi values of Ti2Nb10O29 powder and Ti2Nb10O29-900 NFs ranged from 0.5  1013 to 1 1012 cm2 s1 and from 1.5  1013 to 1.2  1012 cm2 s1, respectively, whereas the charge DLi values of Ti2Nb10O29 powder and Ti2Nb10O29-900 NFs ranged from 2.8  1013 to 0.8  1012 cm2 s1 and from 2.8  1013 to 1.1 1012 cm2 s1, respectively. Therefore, these materials can exhibit large DLi values over a wide range of x, resulting in larger discharge capacities. The maximum Li+ intercalation amount in Ti2Nb10O29-900 NFs (x in LixTi2Nb10O29) increased to 5, which is much larger than that of the Ti2Nb10O29 powder. In addition, higher DLi of Ti2Nb10O29-900 NFs, significantly improves the intrinsic kinetic properties and enables superior rate capabilities. The increased DLi and larger x of the Ti2Nb10O29-900 NFs during charge and discharge may be attributed to its well-aligned crystallinity, which is free of the lattice distortion evident in the Ti2Nb10O29 powder (Fig. 2c and f). EIS measurements were performed in the fully lithiated state to further understand the superior rate performance of the Ti2Nb10O29-900 NFs. Fig. S11 (Supporting information, Fig. S11) shows the Nyquist plots of the samples before and after chargedischarge cycling (2000 cycles for Ti2Nb10O29-900 NFs, 600 cycles for Ti2Nb10O29 powder and Ti2Nb10O29-800 NFs, and 550 cycles for Ti2Nb10O29-700 NFs). In the fresh cells, the Ti2Nb10O29 electrodes

Fig. 5. GITT curves for the Ti2Nb10O29 powder and Ti2Nb10O29-900 NF electrodes as functions of (a) time and (b) stoichiometric x values between 1.0 and 3.0 V. The DLi values were calculated from the GITT (c) discharge and (d) charge data of the samples as a function of the stoichiometric x value.

458

D. Pham-Cong et al. / Electrochimica Acta 236 (2017) 451–459

show one semicircle in the high-frequency range, possibly reflecting the charge-transfer process at the conductive agent interface related to the dominant reduction reaction at the uppervoltage plateau. The inclined line corresponds to Li+-diffusion in the low-frequency range. Notably, the interfacial resistance decreases after long-term continuous cycling, possibly because of the pulverization of the Ti2Nb10O29 surface and the subsequent increase of the surface area. Among all the samples, Ti2Nb10O29900 NFs showed the lowest resistance because of its relatively high crystallinity and Li+-diffusion coefficient. Fig. 6 presents the schematic descriptions of designed structures for the Ti2Nb10O29-900 NFs and Ti2Nb10O29 powder, which suggest the enhanced charge transfer mechanism associated with the continuous 1D framework. Compared to the Ti2Nb10O29 powder, the Ti2Nb10O29 NFs have large surface areas and high length/diameter ratios, which increase the electrolyte/ electrode contact area and provide short pathways for Li+ and electron transport (Fig. 6a). The specific surface areas of Ti2Nb10O29 powder and Ti2Nb10O29 NFs are 1.66  0.02 m2/g and 6.95  0.07 m2/g, respectively. The surface area of nanofibers was about 4 times larger than powder sample. The high value (1.5  1012 cm2 s1) of the Li+ diffusion coefficient in Ti2Nb10O29900 NFs significantly facilitates the intrinsic kinetic properties and enables superior rate capabilities. In addition, the Ti2Nb10O29-900 NFs consists of high-quality crystalline particles connected to particles, which provides a shorter diffusion length for Li+-

insertion/extraction and, thereby, improves the charge transfer and electron conduction along the length direction [53,54]. Finally, these interconnected channels can migrate the lattice stress–strain generated from the volume expansion and contraction associated with lithium insertion and extraction, resulting in the enhanced cycling stability and Coulombic efficiency of the electrodes. Meanwhile, the Ti2Nb10O29 powder may induce cracks and form SEI layer due to the stress generated within the particles and at defective sites on the Ti2Nb10O29 surface during high Li+-insertion/ extraction rate and long cycling test, respectively, causing the fading capacity of Ti2Nb10O29 powder faster than Ti2Nb10O29-900 NFs (Fig. 6b). The above attractive properties clearly confirm that Ti2Nb10O29 NFs is one of the promising candidates for practical applications in lithium ion batteries. 4. Conclusions In this paper, we have successfully fabricated Ti2Nb10O29 NFs composed of linked single crystallites through a simple and facile electrospinning method without any chemical treatment. Compared to the Ti2Nb10O29 powder synthesized by the solid-state reaction, the Ti2Nb10O29-900 NFs show high Li+ diffusion coefficients, an extremely high rate performance and long cycle life as demonstrated by over 2000 cycles. The excellent electrochemical properties may be related to the favorable kinetics supported by the 1D nanostructure, which has a high Li+-diffusion

Fig. 6. Schematic of the Li+-insertion/extraction and electron-transfer mechanisms of the (a) Ti2Nb10O29-900 NF and (b) Ti2Nb10O29 powder electrodes under high C-rate and long cycling number condition. Conversion(M.E.)

D. Pham-Cong et al. / Electrochimica Acta 236 (2017) 451–459

coefficient because of the connection of highly crystalline particles. The electrospinning method represents an effective, simple, and versatile technique for producing Ti2Nb10O29 NF electrode materials with high gravimetric energy densities and promising applications in next generation LIBs. Acknowledgments

[20] [21] [22] [23] [24] [25] [26]

D. Pham-Cong and J. Kim contributed equally to this work. This work was supported by a research program (grant NRF2015R1D1A3A01018611) through the Ministry of Education of Korean Government.

[27] [28] [29] [30]

Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version,at http://dx.doi.org/10.1016/j. electacta.2017.03.203. References

[31] [32] [33] [34] [35] [36] [37]

[1] Q. Zhang, E. Uchaker, S.L. Candelaria, G. Cao, Chem. Soc. Rev. 42 (2013) 3127– 3171. [2] C. Chae, H. Noh, J.K. Lee, B. Scrosati, Y. Sun, Adv. Funct. Mater. 24 (2014) 3036– 3042. [3] Y. Tang, Y. Zhang, J. Deng, J. Wei, H.L. Tam, B.K. Chandran, Z. Dong, Z. Chen, X. Chen, Adv. Mater. 26 (2014) 6111–6118. [4] M. Armand, J.M. Tarascon, Nature 451 (2008) 652–657. [5] L. Wang, N. Lin, J. Zhou, Y. Zhu, Y. Qian, Chem. Commun. 51 (2015) 2345–2348. [6] M. He, K. Kravchyk, M. Walter, M.V. Kovalenko, Nano Lett. 14 (2014) 1255– 1262. [7] X. Wang, R.A. Susantyoko, Y. Fan, L. Sun, Q. Xiao, Q. Zhang, Small 10 (2014) 2826–2829. [8] J.S. Cho, Y.J. Hong, Y.C. Kang, ACS Nano 9 (2015) 4026–4035. [9] C. He, S. Wu, N. Zhao, C. Shi, E. Liu, J. Li, ACS Nano 7 (2013) 4459–4469. [10] C. Guan, X. Wang, Q. Zhang, Z. Fan, H. Zhang, H.J. Fan, Nano Lett. 14 (2014) 4852–4858. [11] S.S. Zhang, J. Power Sources 161 (2006) 1385–1391. [12] D. Pham-Cong, J.H. Kim, S.Y. Jeong, J.H. Choi, J.W. Kim, C.R. Cho, Electrochem. Commun. 60 (2015) 204–207. [13] J. Kim, J.Y. Kim, D. Pham-Cong, S.Y. Jeong, J. Chang, J.H. Choi, P.V. Braun, C.R. Cho, Electrochim. Acta 193 (2016) 35–44. [14] L. Yu, H.B. Wu, X.W. Lou, Adv. Mater. 25 (2013) 2296–2300. [15] H. Park, H.B. Wu, T. Song, X.W. Lou, U. Paik, Adv. Energy Mater. 5 (2015) 1401945. [16] X. Lu, Z.L. Jian, Z. Fang, L. Gu, Y.S. Hu, W. Chen, Z.X. Wang, L.Q. Chen, Energy Environ. Sci. 4 (2011) 2638–2644. [17] T. Takashima, T. Tojo, R. Inada, Y. Sakurai, J. Power Sources 276 (2015) 113–119. [18] G. Liu, B. Jin, R. Zhang, K. Bao, H. Xie, J. Guo, M. Wei, Q. Jiang, Int. J. Hydrogen Energy 41 (2016) 14807–14812. [19] C. Lin, S. Yu, H. Zhao, S. Wu, G. Wang, L. Yu, Y. Li, Z.Z. Zhu, J. Li, S. Lin, Sci. Rep. 5 (2015) 17836.

[38] [39] [40] [41] [42] [43] [44] [45] [46] [47] [48] [49] [50] [51] [52] [53] [54]

459

C. Lin, G. Wang, S. Lin, J. Li, L. Lu, Chem. Commun. 51 (2015) 8970–8973. J. Maier, Nat. Mater. 4 (2005) 805–815. Y. Tang, Y. Zhang, W. Li, B. Ma, X. Chen, Chem. Soc. Rev. 44 (2015) 5926–5940. X. Wu, J. Miao, W. Han, Y.S. Hu, D. Chen, J.S. Lee, J. Kim, L. Chen, Electrochem. Commun. 25 (2012) 39–42. Q. Cheng, J. Liang, Y. Zhu, L. Si, C. Guo, Y. Qian, J. Mater. Chem. A 2 (2014) 17258– 17262. W.L. Wang, B. Oh, J. Park, H. Ki, J. Jang, G. Lee, H. Gu, M. Ham, J. Power Sources 300 (2015) 272–278. L. Perfler, V. Kahlenberg, C. Wikete, D. Schmidmair, M. Tribus, R. Kaindl, Inorg. Chem. 54 (2015) 6836–6848. M.E. Im, D. Pham-Cong, J.Y. Kim, H.S. Choi, J.H. Kim, J.P. Kim, J. Kim, S.Y. Jeong, C. R. Cho, J. Power Sources 284 (2015) 392–399. H. Xu, X. Hu, Y. Sun, W. Luo, C. Chen, Y. Liu, Y. Huang, Nano Energy 10 (2014) 163–171. W. Weppner, R.A. Huggins, J. Electrochem. Soc. 124 (1977) 1569–1578. D.W. Dees, S. Kawauchi, D.P. Abraham, J. Prakash, J. Power Sources 189 (2009) 263–268. A.D. Wadsley, Acta. Cryst. 14 (1961) 660–664. C.C. Li, Y.P. Xheng, T.H. Wang, J. Mater. Chem. 22 (2012) 13216–13222. M.V. Reddy, N. Sharma, S. Adams, R.P. Rao, V.K. Peterson, B.V.R. Chowdari, RSC Adv. 5 (2015) 29535–29544. H. Liu, N. Gao, M. Liao, X. Fang, Sci. Rep. 5 (2015) 7716. S. Guo, X. Zhang, Z. Zhou, G. Gao, L. Liu, J. Mater. Chem. A 2 (2014) 9236–9243. A.L. Viet, M.V. Reddy, R. Jose, B.V.R. Chowdari, S. Ramakrishna, J. Phys. Chem. C. 114 (2010) 664–671. A. Nyman, T.G. Zavalis, R. Elger, M. Behm, G. Lindbergh, J. Electrochem. Soc. 157 (2010) A1236–A1246. T. Takashima, T. Tojo, R. Inada, Y. Sakurai, J. Power Sources 276 (2015) 113–119. S.R. Das, S.B. Majumder, R.S. Katiyar, J. Power Sources 139 (2005) 261–268. V. Augustyn, J. Come, M.A. Lowe, J.W. Kim, P.-L. Taberna, S.H. Tolbert, H.D. Abruña, P. Simon, B. Dunn, Nat. Mater. 12 (2013) 518–522. H. Lindström, S. Södergren, A. Solbrand, H. Rensmo, J. Hjelm, A. Hagfeldt, S.-E. Lindquist, Phys. Chem. B 101 (1997) 7717–7722. X.P. Li, J. Mao, New J. Chem. 39 (2015) 4430–4436. P.L. Taberna, S. Mitra, P. Poizot, P. Simon, J.M. Tarascon, Nat. Mater. 5 (2006) 567–573. T.M. Bandhauer, S. Garimella, T.F. Fuller, J. Electrochem. Soc. 158 (2011) R1– R25. A. Senyshyn, M.J. Muhlbauer, O. Dolotko, H. Ehrenberg, J. Power Sources 282 (2015) 235–240. F. Leng, C.M. Tan, M. Pecht, Sci. Rep. 5 (2015) 12967. B. Song, T. Sui, S. Ying, L. Li, L. Lu, A.M. Korsunsky, J. Mater. Chem. A 3 (2015) 18171–18179. P. Barai, P.P. Mukherjee, J. Electrochem. Soc. 160 (2013) A955–A967. M. Ebner, D.W. Chung, R.E. Garcia, V. Wood, Adv. Energy Mater. 4 (2014) 1301278. M. Park, X. Zhang, M. Chung, G.B. Less, A.M. Sastry, J. Power Sources 195 (2010) 7904–7929. A. Van der Ven, J. Bhattacharya, A.A. Belak, Acc. Chem. Res. 46 (2013) 1216– 1225. D.D. Lecce, J. Hassoun, J. Phys. Chem. C 119 (2015) 20855–20863. K. Kang, H.S. Lee, D.W. Han, G.S. Kim, D. Lee, G. Lee, Y.M. Kang, M.H. Jo, Appl. Phys. Lett. 96 (2010) 053110. H.G. Wang, S. Yuan, D.L. Ma, X.B. Zhang, J.M. Yan, Energy Environ. Sci. 8 (2015) 1660–1681.