Vapor-Solid Reaction Growth of Rutile TiO2 Nanorods and Nanowires for Li-Ion-Battery Electrodes.

A new synthetic method to grow O-deficient rutile TiO2(s) nanorods (NRs) and nanowires (NWs) by a vapor-solid reaction growth method is developed. TiCl4(g) was employed to react with commercially supplied CaTiO3(s) (size 2-4 μm) at 973 K under atmospheric pressure to generate TiO2(s) NRs (diameters 80-120 nm, lengths 1-4 μm). The reaction employing TiCl4(g) and CaO(s) at 973 K also generated CaTiO3(s) (size 4-13 μm) as the intermediate which reacted further with TiCl4(g) to produce NWs (diameters 80-120 nm, lengths 4-15 μm). This is the first report of 1D rutile TiO2(s) nanostructure with such a high aspect ratio. Both of the NRs and the NWs, with compositions TiO1.81 and TiO1.65, respectively, were single crystals grown in the [001] direction. Their morphology was affected by the reaction temperature, the concentration of TiCl4(g), and the particle size of CaTiO3(s). The NRs and the NWs were investigated as anode materials for Li+-ion batteries. At constant current rates 1, 2, and 5 C (1 C = 170 mA g-1) for 100 cycles, the cycling (1.0-3.0 V) performance data of the NRs were 146, 123, and 104 mA h g-1, respectively. On the other hand, the cycling performance data of the NWs were 120, 80, and 52 mA h g-1, respectively. This is attributed to the high Li+ ion diffusion rate (D Li+ ) of the NRs (29.52 × 10-15 cm2 s-1), which exceeds that of the NWs (8.61 × 10-15 cm2 s-1). Although the [001] growth direction of the NR crystals would provide the fastest channels for the diffusion of Li+ ions and enhance the battery capacity, the extremely long channels in the NWs could hamper the diffusion of the Li+ ions. The O-deficiency in the structure would increase the conductivity of the electrode material and improve the stable cycling stability of the batteries also. The long-term cycling test at 2 C for the battery constructed from the NRs retained 121 mA h g-1 after 200 cycles and 99.2 mA h g-1 after 800 cycles. The device has an excellent long-term cycling stability with a loss of only 0.04% per cycle.

Introduction

Metal oxides are very important solid materials for energy storage applications because of their good chemical stability, nontoxicity, and low cost.[1,2] For example, anatase TiO2(s) can be employed as an anode material for lithium-ion batteries (LIBs) because of the following major advantages.[3,4] First, it demonstrates excellent Li+ ion insertion/extraction reversibility. Also, its volume expansion during repeated cycling shows a negligible value of 3–4%, so that the mechanical degradation of the electrode material can be avoided.[5,6] Furthermore, LIBs can be operated at a relatively high voltage (>1 V vs Li+/Li), which reduces the formation of the solid-electrolyte interphase on the anodes.[7,8] In addition to anatase TiO2(s), other major phases, such as rutile and brookite, are intensively investigated.[9] Among them, the rutile phase is generally regarded as the most thermodynamically stable one.[10] However, bulk rutile TiO2(s) is a poor electron conductor (10–13 S cm–1). Because Li+ ion intercalation into the crystal lattice is anisotropic, that is, 10–6 cm2 s–1 along the c-axis but 10–15 cm2 s–1 along the ab-plane, the phenomenon limits the storage capacity of the devices.[11−13] To increase the performance and to improve the cycle stability, nano-sized electrode material could be incorporated to reduce the Li+ ion diffusion path (equilibrium time: τeq = L2/2DLi, L: diffusion length, and DLi: diffusion coefficient) and to provide more space to alleviate the structural strain.[14] Another effective method is to increase the Ti3+ and the O vacancy concentrations in the TiO2(s) structure to improve the conductivity.[15,16]

Here, we report employing low-cost TiCl4(g) to react with commercially supplied perovskite CaTiO3(s) via a simple one-step vapor–solid reaction growth (VSRG) method to prepare single crystalline rutile TiO2(s) nanorods (NRs). In addition, we also discover that the reaction between TiCl4(g) and CaO(s) generated CaTiO3(s). The perovskite intermediate reacts further with TiCl4(g) to produce unusually long (>15 μm) high aspect ratio single crystalline rutile TiO2(s) nanowires (NWs). Even though the process is an improvement of a previously reported VSRG route, to our knowledge, the observation is the first report of rutile TiO2 in this morphology.[17] The as-prepared TiO2(s) NRs and NWs structures are O-deficient and show excellent electrochemical behavior as the anode material in LIBs.

Result and Discussion

Preparation and Characterization of Rutile TiO2(s) NRs

Through a new VSRG route, TiCl4(g) was reacted with commercially supplied CaTiO3(s) (size 2–4 μm) for 20 h in a hot-wall reactor at 973 K to generate a blue solid product (Figure S1a in the Supporting Information). The size of the reactant CaTiO3(s) influences the product morphology greatly and will be discussed more below. As shown by the X-ray diffraction (XRD) pattern in Figure S1b, the product was identified to be rutile TiO2(s) embedded in a matrix of CaCl2(s) (Joint Committee of Powder Diffraction Standard, JCPDS no. 21-1276 and 74-0992).[18,19] After CaCl2(s) was removed by washing the solid product with deionized (DI) water, pure rutile TiO2(s) (P973) was isolated. P973 and other samples investigated are summarized in Table S1 in the Supporting Information. The scanning electron microscopic (SEM) images of P973 (Figure a,b) display the presence of numerous NRs (widths 80–120 nm and lengths 1–4 μm). The electron-dispersive X-ray spectrum (EDX) of P973, shown in Figure c, indicates that the sample contains both Ti and O elements. The XRD pattern (Figure d) does not show signals from the residual CaTiO3(s) reactant (JCPDS no. 22-153). This confirms that P973 is pure rutile TiO2(s). A transmission electron microscopic (TEM) image in Figure e shows an individual NR with a length 1.2 μm and a width 80 nm. From the high-resolution TEM (HRTEM) image presented in Figure f, the spacings are estimated to be 0.29 and 0.20 nm and assigned to the (0 0 1) and (2 1 0) planes of rutile TiO2(s), respectively.[20] The selected-area electron diffraction (SAED) image of P973 displays a dot pattern in Figure g.[21] This reveals that the sample is a single crystal with the zone axis indexed to be [−1 2 0] of rutile TiO2(s).

Figure 1

(a) Low- and (b) high-magnification SEM images, (c) EDX, (d) XRD (*: signals from the sample holder), (e) low- and (f) high-magnification TEM images, and (g) SAED of P973.

To further investigate the chemical states of the surface elements of P973, X-ray photoelectron spectra (XPS) of the sample were taken. As shown in the survey spectrum in Figure S2, the sample displays signals from Ti and O atoms majorly. In addition, signals from traces of C, Cl, and Ca elements are observed. In Figure a, the high resolution spectra of P973 are presented. Figure a shows two major peaks at 458.1 and 463.8 eV from 2P3/2 and 2P1/2 electrons of Ti4+ atoms on the surface, respectively.[22] Two shoulder peaks at 457.1 and 463.3 eV are attributed to 2P3/2 and 2P1/2 electrons of Ti3+ atoms, respectively.[23] A semi-quantitative analysis of the peak intensities suggests a Ti3+/Ti4+ ratio of 0.10. The presence of Ti3+ sites suggests that the NRs may be deficient in O atoms. Figure b shows the XPS of the O 1s electrons at 529.1 and 531.4 eV.[23] They are assigned to the O atoms of TiO2(s) and those of the surface OH groups.[24]Figure c displays two signals of Ca 2P1/2 and 2P3/2 electrons at 354.5 and 347.2 eV, respectively.[25]Figure d indicates weak Cl 2P1/2 and 2P3/2 peaks at 200.2 and 198.1 eV, respectively.[26] They are assigned to a trace amount of the unremoved CaCl2 byproduct in the sample.

Figure 2

High-resolution XPS spectra of (a) Ti 2p, (b) O 1s, (c) Ca 2p, and (d) Cl 2p electrons of P973.

After successfully obtaining the one-dimensional (1D) TiO2(s) NRs, we explored the factors that influenced the crystal growth. These factors include reaction time, precursor concentration, reaction temperature, and CaTiO3(s) size. The data of the samples grown from reactions carried out at 973 K for 5–20 h are shown in Figures S3 and S4 in the Supporting Information. The XRD reflections from rutile TiO2(s) were observed after 5 h (Figure S3). As the reaction time prolongs, the CaTiO3(s) signals decreased while the rutile signals increased. After 20 h, CaTiO3(s) was completely converted into rutile TiO2(s). The SEM study in Figure S4 shows the growth of NRs and the diminishing of the original particles. After 20 h, particles no longer existed while numerous clusters of NRs were observed. The NR growth was affected by the TiCl4(g) concentration in the vapor phase also, as suggested by the XRD and the SEM analyses shown in Figures S5 and S6, respectively. By adjusting the TiCl4(l) bubbling rates (1–4 bubbles s–1), the XRD analyses in Figure S5 confirm that CaTiO3(s) was fully converted to pure rutile TiO2(s) after the reaction was carried out at 973 K for 20 h. As demonstrated in the SEM images in Figure S6, the sizes and the uniformity of the as-produced NRs increased as the TiCl4(l) bubbling rates were increased. As shown in Figure S6c, the NRs with widths 320–400 nm and lengths 2–4 μm were isolated from the reaction employing the highest concentration of TiCl4(g). This product is P973, which is mentioned above. The reaction temperature influenced the product morphology greatly. As the reaction temperature was increased from 923 to 1023 K, the XRD patterns in Figure S7 confirm the formation of the rutile phase TiO2(s) in all experiments. The corresponding SEM images in Figure S8 display their morphology. Isolated from the reaction carried out at 923 K, P923 was obtained. As presented in Figure S8a,b, it contained irregularly shaped particles (100–300 nm). P973 was the product obtained from the reaction performed at 973 K. As discussed above, the NR-shaped particles display widths of 80–120 nm and lengths of 1–4 μm in Figures a,b and S8c,d. For the product collected from the reaction performed at 1023 K (P1023), the particle width increased further to 600–800 nm (Figure S8e,f). The last important factor which influences the product morphology is the particle size of the reactant CaTiO3(s). This will be discussed further below.

Preparation and Characterization of Rutile TiO2(s) NWs

As reported previously, we discovered that in the reaction between CaO(s) and TiCl4(g) at 973 K for 36 h, a dark blue product composed of rutile TiO2(s) and CaCl(s) was obtained (Figure S9).[17] After the removal of the byproduct CaCl2(s), the sample was named C973 and identified to be bundles of rutile TiO2(s) NWs with lengths exceeding 10 μm (Figure S10). The detailed images of C973 are exhibited in Figure . The SEM image in Figure a indicates that C973 contained numerous NWs with widths 80–140 nm and lengths 4–15 μm. The TEM image (Figure b) of a NW isolated from C973 shows a length of 11 μm and a width of 80 nm. From the HRTEM image shown in Figure c, the spaces between two sets of parallel fringes of the crystalline structure are measured to be 0.29 and 0.32 nm. They correspond to the spacings of the (0 0 1) and (1 1 0) planes of rutile TiO2(s), respectively. The SAED image shows a dot pattern in Figure d.[20] It suggests that the single crystal sample can be indexed to the [−1 1 0] zone axis of rutile TiO2(s). The XPS survey spectrum of C973 is shown in Figure S11. In Figure S12a, two major peaks at 457.4 and 463.1 eV from Ti4+ 2P3/2 and 2P1/2 electrons are observed, respectively. Also, after curve fittings, two shoulder peaks at 457.1 and 462.5 eV are attributed to Ti3+ 2P3/2 and 2P1/2 electrons, respectively. The molar ratio of Ti3+/Ti4+ ions in C973, determined by semi-quantitative analyses of the peak intensities, is 0.33. Figure S12b shows the XPS signals of O 1s electrons at 529.4 and 531.8 eV.[23] They are assigned to the O atoms and OH groups on the surface of TiO2(s).[24] The signals of Ca 2P and Cl 2P electrons from the trace amount of the CaCl2(s) byproduct are presented in Figure S12c,d.[25,26]

Figure 3

(a) SEM, (b) low- and (c) high-magnification TEM images, and (d) ED of C973.

Further investigation of the reaction carried out for only 20 h provided a dark blue product also. The XRD pattern in Figure S13 shows that the solid is a mixture of several products. In addition to TiO2(s) and CaCl2(s), perovskite CaTiO3(s) was also formed at this stage, probably as an intermediate. As discussed above, it may react further with TiCl4(g) to grow the TiO2(s) NWs. As shown by the SEM and EDX data in Figure S14, some as-synthesized CaTiO3(s) crystals with sizes 4–13 μm, much larger than the ones found in the commercial powders (size 2–4 μm, Figure S4), are found in the sample. On the other hand, the XRD pattern and the SEM image in Figure S15 reveal that in the sample, there also coexisted numerous rutile TiO2(s) NWs about 10 μm in length with partially reacted large CaTiO3(s) crystals. In addition, as shown in Figure S16, CaTiO3(s) crystals with numerous steps on the faces are found in the sample. After the reaction time extends to 36 h, the product contained rutile TiO2(s) NWs embedded in a matrix of CaCl2(s) only, as demonstrated by the data discussed previously in Figure S9. No CaTiO3(s) was observed anymore.

In summary, from the reactions between TiCl4(g) and CaO(s), we discover that large CaTiO3(s) (ca. 4–13 μm) crystals were generated as the intermediate and reacted further to produce long NWs, such as the ones observed in C973. In contrast, from the small size commercial CaTiO3(s) (size 2–4 μm), the NRs in P973 were relatively short, about 4 μm in length. From these observations, we conclude that the particle sizes of the perovskite CaTiO3(s) influence the lengths of the rutile 1D structure greatly. A reaction pathway will be proposed and discussed further below.

The effect of the reaction temperature (923–1023 K) on the NW growths, as shown by the XRD and the SEM images, is presented in Figures S17 and S18, respectively. Based on the XRD patterns displayed in Figure S17, all products, C923 (grown at 923 K), C973, and C1023 (grown at 1023 K), are determined to be rutile-phase TiO2(s). The SEM images of C923 (Figure S18a,b) display uneven NRs with lengths 100–300 nm and widths 80–140 nm. As discussed above, C973 shows numerous long NWs with widths 80–140 nm and lengths 10–15 μm in Figure and in Figure S18c,d. For C1023 observed in Figure S18e,f, long rods are found. With widths 300–400 nm and lengths exceeding 10 μm, they appear to be aggregates of NWs.

O-Deficiency in TiO2(s) NRs and NWs

The XPS data and the color of the as-prepared TiO2(s) products suggest that they are O-deficient.[27] From the results of the thermogravimetric analyses (TGA) presented in Figure a, P973 and C973 heated in a stream of O2(g) turned white and showed 3.8 and 7.5 wt % increases, respectively. Consequently, the original P973 and C973 compositions are determined to be TiO1.81 and TiO1.65, respectively. As suggested by the TGA studies, bulk samples of P973 and C973 were heated at 773 K in air to offer the corresponding oxidized samples PA973 and CA973. As shown in the Raman spectra in Figure b, the vibrations of P973, C973, PA973, and CA973 are assigned to the second-order scattering Eg (232 cm–1), Eg (447 cm–1), and A1g (612 cm–1) modes of rutile TiO2(s).[28] For P973 and C973 with O-deficiencies, the Eg signals are found at 437 and 439 cm–1 while the A1g signals are observed at 632 and 629 cm–1, respectively.[29] For PA973 and CA973, the Eg and the A1g modes are observed at 446 and 636 cm–1, respectively. They agree well with the characteristic rutile TiO2(s) peaks. Clearly, the blue shifts of the Eg and A1g vibrations of P973 and C973 from the reported rutile values are proposed to be originated from the O-deficiencies. The electron paramagnetic resonance (EPR) results of the samples are shown in Figure c. For P973 and C973, the observed g factors 1.96 and 1.94 are attributed to the Ti3+ in the O-deficient sites in the bulk and in the subsurface layers of rutile crystals, respectively.[30] C973, with more O-deficiencies, shows stronger EPR signals than P973 does. As displayed in Figure c, CA973 and PA973 are nearly EPR-silent after being annealed in air to remove the O vacancies. Figure S19 shows the UV–vis diffuse reflectance spectra and the plots of the spectra transformed according to the Kubelka–Munk function.[31,32] In Figure S19a, P973 and C973 display absorptions at 438 and 450 nm, respectively. Because of their O-deficiencies, the absorptions show significant red-shifts from that of rutile TiO2(s), reported at 410 nm. On the other hand, the oxidized products PA973 and CA973 demonstrate the literature value at 410 nm. From Figure S19b, the optical band gaps of the samples are derived from the plots of (F(R∞)hν)2 versus hν. After the analyses, the band gap energies are calculated to be 2.88, 2.79, 2.96, and 2.95 eV for P973, C973, PA973, and CA973, respectively. The observations agree with the literature report, indicating that TiO2(s) samples with more O-deficiencies show lower band gaps and darker blue colors.[32]

Figure 4

(a) TGA of C973 and P973 (in O2(g), 300–1100 K, 4 K/min). (b) Raman and (c) EPR spectra of TiO2(s) samples P973, C973, PA973, and CA973.

Proposed Reaction Pathway

Although 1D crystalline rutile TiO2 nanostructures were prepared previously via solution routes, including hydrothermal and electrochemical methods followed by high-temperature treatments, our experiments demonstrate the first successful solvent-free growth of the material with extremely high aspect ratio.[9,10,20,24] Thermodynamically, reacting CaTiO3(s) with TiCl4(g) favors the formation of TiO2(s) as shown by its exothermic nature in eq (33)

In addition, CaTiO3(s) and TiO2(s) are also produced exothermically from CaO(s) and TiCl4(g), as expressed in eqs and 3

By summarizing the experimental observations discussed above, a schematic illustration of the VSRG process of 1D TiO2(s) is depicted in Scheme . During the reaction, phase separation of the solid products grown at the interface between the vapor and the solid reactants shapes the product morphology.[34−37] In this study, TiCl4(g) reacts with CaTiO(s), both commercially supplied and as-synthesized, to produce a mixture of TiO2(s) and CaCl2(l) (Figures S1 and S9) on the surface of the solid reactant. As presented in Scheme , we propose that because of the high melting point of TiO2(s) (1903 K) and its low solubility in the as-produced CaCl2(l) (mp 1045 K), the oxide NRs and NWs elongate along the CaTiO3(s) faces. This is supported by the steps observed on the sides of the reacted CaTiO3(s) displayed in Figures S16.[38] As the reaction prolongs, nanosized TiO2 clusters grow into the high aspect ratio crystals while each individual one is embedded in and isolated by the as-formed molten CaCl2(l) media at 973 K. After the removal of CaCl2(s), the 1D TiO2(s) products are isolated. From the commercial CaTiO3(s), short NRs were obtained because the crystal sizes in the perovskite reactant were small (Figure S4a). In contrast, long NWs were isolated from the large synthesized perovskite crystals (Figure S14). The proposed reaction pathway agrees with the results observed.

Scheme 1

Proposed 1D TiO2(s) Growth Process

Performance as Li-Ion Battery Electrodes

The 1D solids were examined by cyclic voltammetric (CV) measurements using CR2032 type half-cell devices constructed from P973 and C973 and a piece of Li(s) foil as the electrodes. The CV results presented in Figure S20 show five cycles of CV profiles (0.5 mV s–1, 1.0–3.0 V vs Li/Li+). For both samples, obvious cathodic peaks located at 1.4 V are observed in the first sweep. It is assigned to the conversion of TiO2(s) into LiTiO2(s) according to the equation TiO2(s) + xLi+ + xe– → LiTiO2(s).[39] In the anodic scan, a peak at 1.95 V is observed. This represents the extraction of Li+ ions from LiTiO2(s).[40,41] The shape of the later CV sweeps does not change significantly. Galvanostatic charge/discharge voltage profiles of P973 and C973 at 1 C (170 mA g–1, 1.0–3.0 V vs Li/Li+. The C rate is defined as the current to discharge the nominal capacity in 1 h) are displayed in Figure S21. The irreversible capacity loss between 1.4 and 1.0 V during the first cycle is related to the irreversible change of the rutile structure during deep Li+-ion insertions.[42] As suggested by the literature, a short plateau appeared at approximately 1.45 V in the discharging curve, which is attributed to the result of the insertion of Li+ ions into the octahedral positions in the lattice.[43] The desertion of Li+ ions is observed at ca. 1.95 V as a plateau in the charging curve. This conclusion coincides with the result of CV measurement discussed above. Figure shows the specific discharge capacities of several half-cell devices constructed from the TiO2(s) samples at a current rate 1 C (1.0–3.0 V vs Li/Li+). The corresponding initial charge/discharge data of the samples are displayed in Figure S22 in the Supporting Information. Among them, P973 demonstrates the best performance. It exhibits the first-cycle discharge capacity 166.4 mA h g–1 with an initial Coulombic efficiency (capacity discharged/capacity charged x 100%) of 84.3%. As shown in Figure , other samples do not perform as well as P973 does. The corresponding cycling performance data are summarized in Table S2. After 100 cycles, P973 provides a capacity of 143.3 mA h g–1, which is the best performance among all samples investigated. The capacities of P973 and C973 at different charging rates (1, 2, and 5 C, 1.0–3.0 V) are shown in Figure S23. Again, P973 exhibits better performance at all charging rates than C973 does. The results of electrochemical impedance spectroscopic (EIS) analyses of P973 and C973 are summarized in Figure S24 in the Supporting Information. The corresponding Nyquist plots are drawn in Figure S24a by using the equivalent circuit shown in S24b. From the impedance spectra, the internal resistance Rs, the charge transfer resistance Rct, and the Warburg impedance Zw are estimated and listed in Table S3. By using the data, σ, the diffusion coefficient, and DLi, the diffusion rate of Li+ ions in the electrode material, are calculated and provided in Table S3.[44] The results indicate that the Rs and Rct values of P973, 9.39 and 134.21 Ω are higher than those of C973, 5.75 and 86.20 Ω, respectively. However, the capacity of P973, 143.3 mA h g–1, exceeds that of C973, 102.3 mA h g–1, greatly. Thus, the low Rs and Rct values of C973 do not enhance its battery performance. On the other hand, the DLi value of P973, 29.52 × 10–15 cm2 s–1, is nearly four times that of C973, 8.61 × 10–15 cm2 s–1. We attribute that increasing DLi in the solid electrode material enhances the battery capacity. Clearly, DLi is the major factor affecting the battery performance. This agrees with the literature observations.[41,45−47] It has been reported that the rate of Li+ ion intercalation into rutile TiO2(s) along c-axis is 10–6 cm2 s–1 while through ab-plane is 10–15 cm2 s–1.[12] Therefore, the capacity performance of the battery depends highly on the amount of the channels along c-axis, which is the [001] direction in the lattice. Previously, the TEM images of the NR in P973 (Figure e) and the NW in C973 (Figure b) display that the crystals have long lengths in the [001] direction. Consequently, we propose that Li+ ions could insert and desert through the channels in that direction more easily. Because the NRs are much shorter than the NWs, the diffusion of Li+ ions in the NRs should be more effective than in the NWs, as demonstrated by the corresponding DLi values.[43] To demonstrate that shorter crystals would improve the battery performance, we shorten the length of the NWs in C973 by ultrasonicating the sample in ethanol for 120 min. The shortened NWs, with lengths 1–4 μm, is named m-C973 and shown in Figure S25. After being assembled into a half battery, the m-C973 device shows improved performance as demonstrated in Figure S26. For example, at 2 C, the capacity of the m-C973 device increases to 108.4 mA h g–1. Moreover, DLi of m-C973 derived from the EIS analysis (Figure S27) increases to 14.12 × 10–15 cm2 s–1. The value is about twice of the result of C973.

Figure 5

Cycling performance of P923, P973, P1023, C923, C973, C923, PA973, and CA973 electrodes at 1 C (170 mA g–1, 1.0–3.0 V vs Li/Li+. The C rate is defined as the current to discharge the nominal capacity in 1 h).

Both P973 and C973 were further tested for variable current rate charging and discharging studies. In Figure , the results of the discharge/charge at 0.5 C followed by increasing those to 1, 2, 5, 10 C, and returning to 1 C sequentially for 10 cycles each are shown. P973 demonstrates better performance with average reversible capacities 152.7, 139.3, 124.1, 104.4, and 75.7 mA h g–1 at 0.5, 1, 2, 4, and 10 C, respectively. After the current rate was reduced back to 1 C, a respectable capacity 133.8 mA h g–1 was observed. The experiment indicates a stable reversibility of P973. The long-term cycling performance test result of P973 at 2 C is presented in Figure . At the 10th cycle, the capacity was recorded to be 136.8 mA h g–1. After 200 cycles, the capacity was retained at 121.3 mA h g–1. After 800 cycles, the capacity was measured to be 99.2 mA h g–1, which is 72.5% of the value recorded at the 10th cycle. The result indicates that the device possesses an excellent long-term cycling stability with a capacity loss of only 0.04% per cycle. The Coulombic efficiency for each cycle is excellent as well, ranging from 99.3 to 99.8%.

Figure 6

Rate performance of P973 and C973 at various current rates from 0.5 to 8 C (1 C = 170 mA g–1).

Figure 7

Cycling performance of P973 at a current density of 2 C and the corresponding Coulombic efficiency of P973 (1 C = 170 mA g–1).

Conclusions

In this study, we successfully developed a new synthetic method to grow O-deficient rutile TiO2(s) NRs and NWs by a VSRG method. The process employed TiCl4(g) to react with commercially supplied CaTiO3(s) (size 2–4 μm) at 973 K under atmospheric pressure to generate TiO2(s) NRs (diameters 80–120 nm, lengths 1–4 μm). In the reaction which employed TiCl4(g) to react with CaO(s) at 973 K also generated CaTiO3(s) (size 4–13 μm) as the intermediate. This reacted further with TiCl4(g) to produce NWs (diameters 80–120 nm, lengths 4–15 μm). This is the first report of the 1D rutile TiO2(s) nanostructure with such a high aspect ratio. Both of the NRs and the NWs, with compositions TiO1.81 and TiO1.65, respectively, were single crystals grown in the [001] direction. Their morphology was affected by the reaction temperature, the concentration of TiCl4(g), and especially the particle size of CaTiO3(s). As indicated by the steps observed on the sides of the reacted CaTiO3(s) crystal surface, we suggest that during the reaction, the growths of TiO2(s) and CaCl2(l) were assisted by phase separation along the vapor–solid reaction interface. Consequently, when large size CaTiO3(s) crystals were reacted, long 1D TiO2(s) NWs were grown from the large crystal faces.

The NRs and the NWs were investigated as anode materials for LIBs. At constant current rates 1, 2, and 5 C for 100 cycles, the cycling (1.0–3.0 V) performance data of the NRs were 146, 123, and 104 mA h g–1, respectively. On the other hand, the cycling performance data of the NWs were 120, 80, 52 mA h g–1, respectively. This is because the Li+ ion diffusion rate (DLi) of the NRs (29.52 × 10–15 cm2 s–1) is higher than that of the NWs (8.61 × 10–15 cm2 s–1). Although the [001] growth direction of the NR crystals would provide the fastest channels for the diffusion of Li+ ions and enhance the battery capacity, the extremely long channels in the NWs could hamper the diffusion of the Li+ ions. Another important factor which affects the performance is the O-deficient structure, which increases the conductivity and improves the cycling stability of the batteries.[48] As a result, the long-term cycling test at 2 C for the battery constructed from the NRs retained 121 mA h g–1 after 200 cycles and 99.2 mA h g–1 after 800 cycles. The device has an excellent long-term cycling stability with a loss of only 0.04% per cycle. While the performance of the device is comparable to those of other rutile-phase TiO2(s) examples reported in literature (Table S4), it is interesting to note that the crystal size of the NRs far exceeds the literature cases.[42,43,49−52] This indicates that while the surface area of the NRs is comparatively small, it provides an efficient electrolyte/electrode interface. We expect the simple VSRG method can be further improved so that better TiO2(s) products may be prepared more efficiently in the future.

Experimental Section

Preparation of Rutile TiO2(s) NRs

Before the reaction, a hot-wall reactor system composed of a quartz tube (length: 85 cm, diameter: 27 mm) placed inside a Lindberg tubular furnace was heated at 1273 K and 10–5 Pa overnight. After the apparatus was loaded with CaTiO3(s) (0.3 g) at room temperature, it was heated to 923–1023 K. Then, TiCl4(l) (Fluka 98%, 1.84 mL), evaporated with the assistance of a stream of Ar(g) (flow rate 4–16 sccm) at 1–4 bubbles s–1 under 1 Pa for 20 h, reacted with the solid. After the reaction completed, the product was washed with DI water several times to remove the soluble portion and dried at 353 K overnight to offer blue solid products (ca. 0.2 g) P923, P973, and P1023, formed at 923, 973, and 1023 K, respectively. A summary of the experimental conditions and the obtained products are listed in Table S1 in the Supporting Information. P973 was further heated at 773 K in air for 5 h to remove the O-deficiency. After the treatment, the white product PA973 was collected.

Preparation of Rutile TiO2(s) NWs

An analogous reaction route, reported previously by our group, was employed.[17] In the process, CaO(s) was reacted with TiCl4(g) using the same reaction setup under similar conditions. The dark blue products, C923, C973, and C1024, obtained from the reactions carried out at 823, 973, and 1024 K, respectively, were collected (Table S1). C973 was further heated at 773 K in air for 5 h to remove the O-deficiency. This offered the white product CA973.

Characterizations

All samples were verified by using a powder XRD (Bruker AXS D8 ADVANCE) with Cu Kα1 radiation (40 kV, 20 mA). The SEM images and the EDX spectra of the samples were investigated by using JEOL JSM-7401F operated at 15 keV. The TEM, SAED, HRTEM, and EDX data were recorded from a JEOL JEM-200 FTM at 200 kV. The XPS results were obtained by a ULVAC-PHI (PHI 5000 Versaprobe). The UV–vis absorption spectra were recorded on a Hitachi 3010 UV–vis spectrophotometer. A Renishaw Raman spectrometer using a 632.8 nm laser was used to characterize the vibrational information of the samples. The EPR spectra were recorded using a JES-FA200 spectrometer. The TGA studies were performed with a NETZSCH STA 409PC under an O2(g) atmosphere.

Electrochemical Tests

All electrochemical tests were examined by using CR2032 coin-type cells assembled in a dry room. The working electrodes were manufacture by the following steps. A N-methyl pyrrolidone (Timcal) slurry was mixed with carbon black (Super-P) (Timcal) and polyvinylidene fluoride in a weight ratio of 8:1:1. The slurry was then uniformly coated on a Cu foil (Furukawa, thickness 14 μm) and dried at 383 K in vacuum overnight. The coated Cu foil was cut into disk electrodes (diameter 13 mm). The electrolyte was composed of LiPF6(s) in a solution of ethylenecarbonate (1.0 M) and dimethyl carbonate (1.0 M) in a volume ratio of 1:2. The active material was weighed (ca. 12 mg) and loaded for each battery. A piece of Li(s) foil (thickness 0.3 mm, diameter 14 mm, ca. 6 mg) was used as the counter electrode and the reference electrode, whereas a piece of membrane (Celgard 2400, pore size: 20 μm) was employed as the separator. A BAT-750B (UbiQ technology) cell test instrument was used to investigate the galvanostatic discharge/charge behaviors of the cells (1.0–3.0 V vs Li/Li+).

The CV experiments were measured with a CHI 6082C electrochemical analyzer (1.0–3.0 V, 0.5 mV/s). The EIS tests (100 kHz to 0.1 Hz, 5 mV) were carried out using a CHI 6082C electrochemical analyzer.