Fabrication of Li4Ti5O12-TiO2 Nanosheets with Structural Defects as High-Rate and Long-Life Anodes for Lithium-Ion Batteries

Development of high-power lithium-ion batteries with high safety and durability has become a key challenge for practical applications of large-scale energy storage devices. Accordingly, we report here on a promising strategy to synthesize a high-rate and long-life Li4Ti5O12-TiO2 anode material. The novel material exhibits remarkable rate capability and long-term cycle stability. The specific capacities at 20 and 30 C (1 C = 175 mA g−1) reach 170.3 and 168.2 mA h g−1, respectively. Moreover, a capacity of up to 161.3 mA h g−1 is retained after 1000 cycles at 20 C, and the capacity retention ratio reaches up to 94.2%. The extraordinary rate performance of the Li4Ti5O12-TiO2 composite is attributed to the existence of oxygen vacancies and grain boundaries, significantly enhancing electrical conductivity and lithium insertion/extraction kinetics. Meanwhile, the pseudocapacitive effect is induced owing to the presence of abundant interfaces in the composite, which is beneficial to enhancing specific capacity and rate capability. Additionally, the ultrahigh capacity at low rates, greater than the theoretical value of spinel Li4Ti5O12, may be correlated to the lithium vacancies in 8a sites, increasing the extra docking sites of lithium ions.

On the other hand, grain boundary can provide fast lithium ion insertion/extraction as the diffusion of lithium ion along the grain boundaries can be orders of magnitude faster than bulk diffusion in the grains, as illustrated in these literatures 18,19 . Therefore, it is very much expected that if the materials are rich in grain boundaries, they will possess excellent lithium transport and storage properties. To increase grain boundary density with large interfacial areas, it would be wise to choose binary or multinary compounds rather than single compound. Rahman et al. 20 synthesized carbon coated Li 4 Ti 5 O 12 -TiO 2 composites with high grain boundary density, exhibiting a high capacity of 166 mA h g −1 at 0.5 C (1 C = 175 mA g −1 ), good cycling stability, and excellent rate capability. Shen et al. 21 developed a versatile hydrothermal route to fabricate porous LTO/rutile-TiO 2 nanosheet arrays, delivering a high initial discharge capacity of 184.6 mA h g −1 at 200 mA g −1 and possessing excellent electrochemical stability with only 8.3% loss of specific capacity at 1 A g −1 after 1000 cycles.
Here, we report on a new route to synthesize a LTO-TiO 2 composite with structural defects. Our purpose is to combine the advantages of both oxygen vacancy and grain boundary to improve the electrochemical performance of LTO. The strategy in this study shows multiple advantages: (1) high concentration of oxygen vacancy enhances the electrical conductivity of electrode materials, (2) grain boundaries, which usually act as channels to facilitate lithium ion diffusion towards the internal of bulk materials, enhance lithium insertion/extraction kinetics, (3) extra lithium ions can be stored in the structural defects including grain boundaries and lithium vacancies, significantly improving the specific capacity, (4) the well-defined TiO 2 scattering on the surface of LTO suppresses the stacking of LTO nanosheets, therefore shortening the path length for lithium ion transportation. The introduction of structural defects obviously enhances lithium storage properties of the LTO-TiO 2 composite in terms of ultrahigh specific capacity (182.1 mA h g −1 at 1 C, 1 C = 175 mA g −1 ), remarkable rate capability (170.3 mAh g −1 at 20 C, 168.2 mAh g −1 at 30 C) and long cycle life (161.3 mAh g −1 remaining after 1000 cycles at 20 C). In addition, the present strategy may provide a new idea to develop high-performance electrode materials for LIBs.

Results
The novel LTO-TiO 2 composite (denoted as LTO-RT-AT) was synthesized using a facile hydrothermal process with subsequent calcination. For a comparative purpose, pure LTO and LTO-rutile TiO 2 composite (denoted as LTO-RT) were also prepared by changing the molar ratios of Li/Ti in the precursor solution. The detailed experimental processes are presented in the methods section.
The X-ray diffraction (XRD) patterns of the precursors of pure LTO, LTO-RT and LTO-RT-AT are shown in Fig. 1a. It is clear that all characteristic diffraction peaks of these precursors are consistent with those of orthorhombic ( Figure 1b shows the XRD patterns of the three samples after calcinated at 600 °C for 6 h in air. The phase combination of the products show strong dependence on the molar ratio of Li/Ti. With a ratio of 4.5:5, the pure LTO was obtained in the final product, all diffraction peaks of the product correspond to the (111), (311), (400), (331), (333), (440), (531), (533), (622) and (444) planes of well-crystallized spinel LTO (JCPDS No. 49-0207). With a decrease in the molar ratio of Li/Ti to 4.2:5 (LTO-RT-AT), two weak peaks at 25.26° and 27.39° arise in Fig. 1c, which can be indexed as the typical peaks of anatase-TiO 2 and rutile-TiO 2 22,23 , respectively, indicating a ternary-phase composite consisting of dominant LTO and a low amount of rutile-TiO 2 and anatase-TiO 2 was obtained. When the ratio further drops to 4:5 (LTO-RT), the peak of anatase-TiO 2 completely disappears and the peaks of rutile-TiO 2 increase further, indicating that a dual-phase composite including LTO and rutile-TiO 2 was obtained. On the basis of the Scherrer equation, the average grain sizes of pure LTO, LTO-RT and LTO-RT-AT are evaluated to be 32.6, 31.1 and 26.5 nm, respectively. This indicates that the LTO-RT-AT exhibits the smallest grain size.
The XRD patterns of the calcinated samples with Li/Ti molar ratio of 4.2:5 at 400 °C (LTO-AT-400) and 500 °C (LTO-AT-RT-500) are shown in Fig. S1a (Supporting Information). It is found that the LTO-AT-400 exhibits the almost pure LTO phase with a tiny broad peak at ~25.3°. For LTO-AT-RT-500, not only the peak at 25.3° becomes shaper, but also additions peaks at 37.3° and 48.1° appear. All these three peaks can be assigned to the three strongest peaks of anatase-TiO 2 22 . Meanwhile, the peaks of rutile-TiO 2 can be observed at 27.5 and 54.3°2 4 . These observations confirm the existence of the TiO 2 phases (anatase and rutile), and a phase transition from anatase to rutile could start when the calcination temperature exceeds 400 °C. Thermogravimetric (TG) and differential scanning calorimetry (DSC) results of the precursor of LTO-RT-AT are shown in Fig. S1b. There are two major weight-loss regions. The first weight loss occurred between 130 and 250 °C due to desorption of crystal water from    25 . In the DSC curve, a broad endothermic peak at ~447 °C is found and this could be related to the conversion of anatase-TiO 2 to rutile-TiO 2 as mentioned above.
Additional information on the structure and oxidation states of pure LTO, LTO-RT and LTO-RT-AT was obtained from X-ray photoelectron spectroscopy (XPS) analysis. Figure 2a shows the high-resolution XPS spectra of Ti 2p for pure LTO. Two peaks at 464.05 and 458.23 eV correspond to the characteristic Ti 2p 1/2 and Ti 2p 3/2 peaks of Ti 4+ 26, 27 , respectively. In addition, a tiny peak at 463.30 eV can be attributed to Ti 3+ 26 . Note that almost all of the peak areas are assigned to Ti 4+ and only a small peak area belongs to Ti 3+ with a content of 3.02%. In comparison to pure LTO, the Ti 2p peaks of LTO-RT (Fig. 2b) and LTO-RT-AT (Fig. 2c) show a slight shift to lower binding energy, and the contents of Ti 3+ increase to 6.31% for LTO-RT and 4.15% for LTO-RT-AT. These differences are probably ascribed to the increased oxygen vacancies in these LTO-TiO 2 composites than those in pure LTO 28 . The incremental oxygen vacancies can decrease electron cloud density surrounding the titanium cores, therefore the electronic screening effect in the nanoparticles would be strengthened, causing the slightly decreased bonding energy values of the Ti 2p 1/2 and Ti 2p 3/2 signals 29 .
In order to further verify the above mentioned statement, the O 1 s spectra of the three samples are also shown in Fig According to previous reports 32, 33 , the formation energy of oxygen vacancy significantly decreases in and around the grain boundary regions compared to the bulk and also the bulk-like structures. That is to say, grain boundary is beneficial for the formation of oxygen vacancy. Note that the grain size of LTO-RT-AT (26.5 nm), significantly smaller than that of pure LTO (32.6 nm) and LTO-RT (31.1 nm), resulting in a higher grain boundary density in the material. Therefore, we believe that the formation of more oxygen vacancies in LTO-RT-AT is closely related to its abundant grain boundaries. It has been reported that grain boundary can serve as reservoir for accumulating oxygen vacancies 34 , at the same time, oxygen vacancies can also stabilize the grain boundary structure 35 .
Such incremental oxygen vacancies are helpful for lithium ion insertion and electron transfer, which is beneficial to improving rate capability of the LTO-RT-AT. However, it is inexplicable that the LTO-RT-AT with the highest content of oxygen vacancies has a relatively lower content of Ti 3+ in comparison with LTO-RT. Based on equilibrium of chemical valence, the content of oxygen vacancies is usually related to the content of Ti 3+ in titanium-based oxides 24 . Therefore, the likely reason is the existence of lithium vacancies in the LTO-RT-AT. The oxygen vacancies and lithium vacancies charge compensate each other.
The morphology of pure LTO, LTO-RT and LTO-RT-AT were examined using a field emission scanning electron microscope (FESEM). As shown in Fig. 3, the three samples exhibit similar two-dimensional plate-like morphology, but the significantly different size, thickness and uniformity. The length and width of the pure LTO plates can reach over 1000 and 500 nm, respectively (Fig. 3a). By measuring the thickness at high magnification SEM (the insert of Fig. 3a), the thickness of the plates is 30~50 nm. In contrast, the LTO-RT (Fig. 3b) and LTO-RT-AT (Fig. 3c) plates exhibit better uniformity and smaller size. The thicknesses are 25~35 nm for LTO-RT (the insert of Fig. 3b) and 15~25 nm for LTO-RT-AT (the insert of Fig. 3c).
The transmission electron microscope (TEM) images show that these plates compose of smaller nanosheets, as shown in Fig. 3d-f. For pure LTO (Fig. 3d), the nanosheets also show the largest size (~500 nm in length and ~300 nm in width), while both LTO-RT and LTO-RT-AT nanosheets exhibit smaller size with length of ~200 nm and width of ~100 nm (Fig. 3e for LTO-RT; Fig. 3f for LTO-RT-AT). The above results clearly indicate that the TiO 2 has important effects on the size of LTO nanosheets and contributes to suppressing their stacking, which may influence their electrochemical performance owing to the different path length for lithiumion transport 36,37 .
More significantly, TEM analysis shows there are some smaller sizes polygons scattered on the surface of these LTO nanosheets in the LTO-RT-AT in  38 . It is found that the peak potentials and shapes of LTO-RT-AT are slightly changed, and the potential differences between the anodic and cathodic potential peaks decrease with cycles, indicating that the polarization becomes weak gradually in LTO-RT-AT 39,40 . In contrast, the peak potentials remain almost unchanged for pure LTO and LTO-RT. In addition, we do not find the characteristic redox peaks of TiO 2 from CV results of the LTO-RT and LTO-RT-AT, suggesting that the electrochemical activity of TiO 2 may be weak in this study, or it may be due to the low content of TiO 2 in these LTO-TiO 2 composites.
To demonstrate the potential application of LTO-RT-AT as LIBs anode materials, the galvanostatic measurements of test cells were performed at a current density of 1 C (1 C = 175 mA g −1 ) in the voltage range of 1.0 to 2.5 V (vs. Li/Li + ). Figure 6a shows the steady-state charge/discharge curves of pure LTO, LTO-RT and LTO-RT-AT. The LTO-RT-AT exhibits the highest specific capacity and the smallest voltage difference between the charge and discharge (the insert of Fig. 6a), suggesting weaker polarization 41 . The finding agrees with the CV results well. The small polarization generally suggests a low internal resistance for electron and lithium ion transport in active materials in line with the excellent insertion/extraction behavior during the high rate cycle. The Nyquist plots and corresponding equivalent circuit based on electrochemical impedance spectroscopy (EIS) measurements are shown in Fig. S3 (Supporting Information). The weights and thicknesses of these tested electrodes are similar. EIS datas were collected after 3 cycles of each cell. It can be seen that the charge transfer resistance (R ct ) of LTO-RT-AT is 23.69 Ω that is smaller than that of pure LTO (35.88 Ω) and LTO-RT (27.14 Ω).
In addition, a long sloped region (marked with arrow in Fig. 6a) between 1.55 and 1.0 V for the LTO-RT-AT is related to the pseudocapacitive lithium storage behavior 19 . To investigate the pseudocapacitive effect of the LTO-RT-AT electrode further, CV at different scan rates were performed and shown in Fig. S4 (Supporting Information). It is well known that the relation between peak current and scan rate reveals the different electrochemical reaction characteristics, including solid phase diffusion-controlled process or surface-confined charge-transfer process 19,29 . For LTO-RT-AT (Fig. S4c), the lower potential peaks at 1.45, 1.40, 1.36, 1.29 and 1.19 V in the cathodic process appear with increasing the scan rate (marked with arrows) and proportionate with the square roots of the scan rates, suggesting an obviously pseudocapacitive effect in the LTO-RT-AT electrode. Whereas such peaks cannot be observed in pure LTO (Fig. S4a) and LTO-RT (Fig. S4b), indicating only the diffusion-controlled process in the two samples.
The above results indicate that the electrode reaction of LTO-RT-AT is controlled simultaneously by the diffusion-controlled and faradaic pseudocapacitive processes. It is well established that the pseudocapacitive effect can be induced due to the presence of abundant interfaces in the LTO-TiO 2 composites, which is conducive to enhancing the specific capacity and rate capability of electrode materials. Similar results were also reported in these previous studies 29,[42][43][44] . By contrast, the pure LTO and LTO-RT have no obviously pseudocapacitive effect, suggesting their lower grain boundary density than the LTO-RT-AT.   To investigate the effect of rulite-TiO 2 on the capacity of the LTO-TiO 2 composites, pure rutile-TiO 2 was synthesized without adding LiOH·H 2 O in the precursor solution. Figure S5a shows its XRD pattern (Supporting Information). The electrochemical result indicates that its specific capacity is very low (less than 20 mA h g −1 ) at a current density of 175 mA g −1 , as shown in Fig. S5b. It is reasonable to believe that the rutile-TiO 2 in LTO-RT and LTO-RT-AT can accommodate only small amounts of lithium ions and difficultly contribute to high capacity. This could be attributed to the highly anisotropic diffusion of lithium ions in rutile crystal structure hampering its performance 45 . Moreover, Liu et al. 46 also found that the contribution of rulite-TiO 2 to the overall capacity of the LTO-TiO 2 composites is negligible at high current density.
Therefore, the ultrahigh specific capacity of LTO-RT-AT, even slightly higher than the theoretical value of LTO (175 mA h g −1 ), could be attributed to the following factors. Firstly, anatase-TiO 2 has a higher specific capacity than LTO. Secondly, lithium ions can be stored in the structural defects of LTO-RT-AT. In consideration of the extremely low content of anatase-TiO 2 (~0.456%) calculated according to the ratio of XRD peak area (anatase-TiO 2 /LTO-RT-AT), we believe the contribution from anatase-TiO 2 to high capacity of LTO-RT-AT is very limited. The stepwise scanned XRD pattern and the areas of all the diffraction peaks of LTO-RT-AT are shown in Fig. S6 (Supporting Information). Therefore, the main reason for the ultrahigh specific capacity of LTO-RT-AT can be attributed to the lithium storage in the structural defects.
Benefiting from the high grain boundary density in LTO-RT-AT, a large number of lithium ions can be stored in these interfacial areas, leading to an obviously pseudocapacitive effect (Fig. S4c), increasing the specific capacity of the material. This is called as 'interface lithium storage' mechanism 47,48 . Moreover, the lithium vacancies in the 8a sites of spinel LTO could allow more external Li ions insertion to 16c sites, as schematically illustrated in Fig. 7. Note that before lithiation, not all the tetrahedral 8a sites are occupied by Li, therefore the initial composition can be depicted as Li 4−x Ti 5 O 12 . Upon lithiation, Li ions migrate from the tetrahedral 8a sites to the octahedral 16c sites, and external 3 + x Li ions insert to 16c sites to attain the final Li 7 Ti 5 O 12 composition. It has been reported that holes can be trapped by lithium vacancies in lithium tetraborate (Li 2 B 4 O 7 ) crystals 49 . A similar mechanism may also exist in the LTO-RT-AT. The lithium vacancies could also trap Li ions druing lithiation process. In light of this, the extra capacity beyond the theoretical value for LTO-RT-AT can be partly attributed to additional x Li ions. Figure 6b also compares the rate capability of three samples measured at various current densities of 1 to 30 C. The capacities of the three samples decline slowly and show excellent rate capabilities from 1 to 10 C. When the current densities are above 20 C, nevertheless, the capacities of pure LTO and LTO-RT decrease significantly, whereas the capacity of LTO-RT-AT decreases much more slowly at the same rate. At the highest rate of 30 C, the LTO-RT-AT still delivers a capacity of 168.2 mA h g −1 , which is much higher than that of pure LTO (134.6 mA h g −1 ) and LTO-RT (123.2 mA h g −1 ). A capacity of 181.6 mA h g −1 is retained for the LTO-RT-AT after 160 cycles when the current density recovers to 1 C. The remarkable rate capability of LTO-RT-AT can be attributed to its unique structural features. A large of structural defects significantly improved electrical conductivity and lithium ion diffusion kinetics properties of LTO-RT-AT electrode during the electrochemical reaction.
To explore the potential applications of LTO-RT-AT in electric vehicles (EV) and hybrid electric vehicles (HEV) which require both high-rate and long-life electrode materials, the fast charge and discharge performance at a high rate of 20 C for long-term cycles were measured and shown in Fig. 6d. It is observed that the LTO-RT-AT not only delivers an ultrahigh initial capacity of 171.3 mA h g −1 , but also maintains long-term cycle stability with Scientific RepoRts | 7: 2960 | DOI:10.1038/s41598-017-03149-2 a capacity of 94.2% retention (161.3 mA h g −1 ) after 1000 cycles. The corresponding coulombic efficiency is held close to 100%, which is very important to the implementation and commercialization. In contrast, the capacities of pure LTO and LTO-RT decrease dramatically at the same condition. The lower capacities of 94.9 mA h g −1 for pure LTO and 98.3 mA h g −1 for LTO-RT are obtained after 1000 cycles, and the capacity retention rates of the two samples only remain at 62% and 69.9%, respectively.

Discussion
In light of the above results, the ultrahigh reversible capacity, remarkable rate capability and long-term cycling stability of LTO-RT-AT can be mainly attributed to its abundant structural defects, including oxygen vacancies and grain boundaries. On the one hand, the oxygen vacancies are beneficial to improving the electrical conductivity of the electrode (increased carrier density) that facilitate the transport of charge carriers. On the other hand, the abundant grain boundaries not only provide more channels for lithium ion transport, enhancing lithium insertion/extraction kinetics, but also cause an obviously pseudocapacitive effect, significantly enhancing specific capacity and rate capability. Therefore, the unique structure of LTO-RT-AT is regarded as the crucial point, resulting in the superior electrochemical performance. This makes the reported LTO-RT-AT a promising new anode material for high-rate and long-life lithium-ion batteries and even for the potential applications in large scale energy storage.

Synthesis of LTO-AT-RT composite.
In a typical synthesis, 20 mM of tetrabutyl titanate was added into 40 mL of anhydrous ethanol under mild stirring for 30 min. Meanwhile, 16.8 mM of LiOH·H 2 O was added into 40 mL of ultrapure water with stirring to allow for complete dissolution. The molar ratio of Li/Ti is 4.2:5. Then, the lithium hydroxide solution was dropwise added into the tetrabutyl titanate ethanol solution with vigorously stirring. The mixture solution was strongly stirred for 2 h to uniform the components. Subsequently, the solution was transferred to a 100 mL Teflon-lined autoclave, then sealed and heated at 180 °C for 24 h. The precipitates were centrifuged and washed thoroughly with ultrapure water and ethanol and then dried at 60 °C for 12 h under vacuum. The resulting product was calcined at 600 °C for 6 h in air to enhance crystallinity. Pure LTO and LTO-RT composite were also prepared using the same method, but the molar ratios of Li/Ti were 4.5:5 and 4:5, respectively.

Materials Characterizations.
The as-obtained samples were initially characterized by X-ray diffraction (XRD) using an X-ray diffraction analyzer with Cu-Kα radiation (D8-Discover, Bruker). Thermogravimetric (TG) and differential scanning calorimetry (DSC) were examined from the STA449 F3 thermal analyzer with a heating rate of 10 °C min −1 over the range 50~600 °C in an air atmosphere. X-ray photoelectron spectroscopy (XPS) measurements were performed on the product using a VG MultiLab 2000 system with a monochromatic Al Ka X-ray source (Thermo VG Scientific). The morphology of the samples were examined using a field emission scanning electron microscope (FESEM, Sirion, FEI). The detailed informations of the LTO-TiO 2 composite were investigated using a high-resolution transmission electron microscope (HRTEM, Tecnai F20, FEI).
Electrochemical Measurements. The electrochemical performance of the novel material was evaluated using CR2032-type coin cell. The working electrodes were fabricated by mixing the active materials with super P and polyvinylidene difluoride (PVDF) in a weight ratio of 75:15:10. Then, an appropriate amount of N-methylpyrrolidone (NMP) solvent was slowly added to the mixture to produce a slurry. The uniform slurry was coated onto a copper foil and dried at 120 °C for 24 h under vacuum. These electrodes were punched into the form of 13 mm diameter disks and assembled into half cells in an Ar-filled glove box. The mass loading of the active materials in each electrode was calculated to be ~2 mg cm −2 by weighing the mass discrepancy of the electrode and Cu foil. Lithium foils were used as the counter electrodes and the Celgard 2400 films were used as the separators. The commercial electrolyte was used which contained 1 M solution of LiPF 6 in ethylene carbonate (EC) and dimethyl carbonate (DMC) with in a volume ratio of 1:1. The galvanostatic charge/discharge measurements were conducted using a LAND battery test system in the voltage range of 1.0 to 2.5 V (vs. Li/Li + ) at different current densities. Cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS) measurements were performed using a CHI660e electrochemical station.