Basicity-controlled synthesis of Li₄Ti₅O₁₂ nanocrystals by a solvothermal method

Abstract

Spinel-type lithium titanate (Li₄Ti₅O₁₂) nanocrystals were synthesized at a low temperature with control of the basicity by a solvothermal method without any calcination processes. The Li–Ti–O precursor was produced under relatively high pH conditions (pH > 11) in the initial stage. The spinel phase was then formed through delithiation and dehydration from the precursor by lowering the pH of the system (pH < 10). The direct synthesis of the Li₄Ti₅O₁₂ nanocrystals was achieved through self-control of the basicity using a two-phase system of water–ethanol and toluene–oleic acid. The Li₄Ti₅O₁₂ crystals were formed from the precursor by decreasing the basicity. The average size of the nanocrystals prepared by the one-step method was ca. 13 nm, and the specific surface area was 158 m² g⁻¹. On the basis of the electrochemical measurement, the Li₄Ti₅O₁₂ nanocrystals produced in a low-temperature solution system are applicable for an anode material of a lithium-ion rechargeable battery.

---

1. Introduction

In recent years, a wide variety of cathode and anode materials have been studied to improve the capacity and durability of rechargeable lithium-ion batteries. Spinel lithium titanate (Li₄Ti₅O₁₂) is a stable anode material having a reaction potential of 1.55 V (vs. Li/Li⁺) and a theoretical capacity of 175 mA h g⁻¹ for lithium-ion batteries. This oxide is well known as zero-strain insertion material having high durability against the insertion–desertion of charge–discharge cycles.^1–4 The size reduction of the crystalline particles is important for further improvement of the capacity and rate performance through expansion of the interface area for the electrochemical reaction and shortening of the distance for the lithium-ion diffusion.⁵ Thus, the nanosized particles, hierarchical structures and mesoporous frameworks of Li₄Ti₅O₁₂ were reported to enhance the performance.^6–11 However, calcination processes above 400 °C are required to produce the spinel structure, and their nanostructures would be degraded at high temperatures.^8–16 In the current work, we directly synthesized the nanocrystals of Li₄Ti₅O₁₂ in a solution system by controlling the basicity without any subsequent calcination processes.

Hydrothermal and solvothermal methods are useful for the synthesis of nanocrystals and hierarchical structures of various metal oxides.^17–23 However, they generally have been hindered by problems such as controlling their chemical composition, size distribution and crystallinity. Particularly, exploration of the optimum conditions for the synthesis of complex oxides is required. In previous works, alkali-earth titanium oxides having a stable perovskite structure (BaTiO₃, SrTiO₃, and CaTiO₃) and layered alkali titanium oxides (Na₂Ti₆O₁₃, KTi₈O_16.5, and Cs_xTi_2−x/4□_x/4O₄) were prepared through hydrothermal routes by adjusting conditions including temperature and pH.^24–29 Although Li₂TiO₃ was obtained by a hydrothermal process, the preparation of crystalline spinel Li₄Ti₅O₁₂ has not been achieved without subsequent calcination.^30,31 While the diffraction pattern assignable to Li₄Ti₅O₁₂ was reported for a product obtained through a solvothermal process by Fattakhova et al., the structural analysis was insufficient and the preparation condition was ambiguous.³² In recent papers, formation of the Li₄Ti₅O₁₂ phase was achieved through a nonaqueous or supercritical synthesis without an annealing process.^33,34 However, these methods required restricted conditions such as an inert atmosphere or high temperature and high pressure. Moreover, the Li₄Ti₅O₁₂ prepared by the nonaqueous synthesis included TiO₂ as an impurity, or Li₂TiO₃ appeared after annealing. Therefore, development of the aqueous-solution system for preparation of spinel-type Li₄Ti₅O₁₂ is still important to produce its nanometric structure with a simple technique for various applications.

In present study, single-phase Li₄Ti₅O₁₂ nanocrystals were directly obtained by a two-step solvothermal method (method I) or a one-step two-phase method (method II) as shown in Scheme 1. The basicity control after formation of the Li–Ti–O precursor was found to be essential for synthesis of the spinel crystal. The resultant Li₄Ti₅O₁₂ crystals, ca. 13 nm in diameter, exhibited a good capacity as an anode material for lithium-ion batteries even without heat treatment. The cycle stability of the nanocrystals was greatly enhanced through improvement of the crystallinity by calcination. The basicity-controlled solvothermal technique would be useful for preparation of a wide variety of complex oxide materials.

Scheme 1 Solvothermal methods for basicity-controlled syntheses of the Li₄Ti₅O₁₂ nanocrystals.

2. Experimental section

The pH-controlled two-step solvothermal method (method I in Scheme 1) was examined as follows. All the reagents in the present work were used without further purification. A 1.0 mol dm⁻³ LiOH solution was prepared by dissolving LiOH·H₂O (Kanto Chemical) in 20 cm³ of a mixture of purified water and ethanol (1/1 volume). Bis(ammonium lactate) titanium dihydroxide (Sigma-Aldrich) was mixed into the solution (molar ratio of Li/Ti: 10), and a white precipitate formed immediately. The first solvothermal process was carried out in a Teflon-lined stainless steel autoclave at 200 °C for 12 h. After cooling to room temperature, the pH of the mixture was decreased to a value between 9 and 10 by the addition of 0.014 mol of acetic acid. The second solvothermal process was then carried out at 200 °C for 12 h. The obtained precipitates were washed by ethanol and purified water and dried at 60 °C in air.

We directly prepared Li₄Ti₅O₁₂ crystals using a basicity-controlled two-phase solvothermal method (method II in Scheme 1). The same mixture of purified water and ethanol containing LiOH·H₂O and bis(ammonium lactate) titanium dihydroxide as for the two-step method was used for the one-step two-phase method. Toluene (20 cm³) containing 32 mmol of oleic acid (Wako Pure Chemical) and tert-butylamine (Wako Pure Chemical) was carefully poured to the water–ethanol mixture. The solvothermal treatment of the two-phase system was carried out at 200 °C for 12 or 24 h. The obtained precipitates were washed by ethanol and purified water and dried at 60 °C in air.

The crystal phases were identified by X-ray diffraction (XRD: Rigaku MiniFlex II and Bruker D8 Advance) with CuKα radiation. Crystalline domain sizes D_(hkl) are calculated by the Scherrer equation. The values of full width at half maximum (FWHM) were calibrated with a highly crystalline standard sample. The morphologies of particles were observed by a transmission electron microscope (TEM: FEI TECNAI-F20) operated at 200 kV. The specific surface area was measured by the N₂-adsorption/desorption measurement (Shimazu Trister 3000). The electrochemical performance was measured with a beaker-type three-electrode cell. The working electrode was a slurry composed of 80 wt% Li₄Ti₅O₁₂ or other Li–Ti–O products, 10 wt% acetylene black carbon and 10 wt% polyvinylidene difluoride binder dispersed in N-methyl pyrrolidinone. The slurry was dropped on a Ni mesh and dried at 80 °C in vacuum. A lithium metal was used for the counter and reference electrodes on a Ni mesh. The electrolyte was 1 mol dm⁻³ of LiClO₄ in ethylene carbonate and diethyl carbonate (1/1 volume). The cell was assembled in an argon-filled glove box. Galvanostatic discharge–charge measurements were performed in a potential range between 1.0 and 2.7 V vs. Li/Li⁺ at room temperature.⁸

3. Results and discussion

3.1. Basicity-controlled two-step solvothermal method (method I)

The precipitate was obtained through the solvothermal reaction with lithium and titanium ions at a pH value over 11 in the first step. According to the XRD pattern (Fig. 1a), the product of the first step was the Li–Ti–O precursor, which was reported in previous works for hydrothermal synthesis of lithium titanates.^{8,9,13,14,34–36} The Li–Ti–O precursor was deduced to be nominal cubic α-Li₂TiO₃ due to the following reasons. The diffraction pattern of the intermediate phase was very similar to the XRD profile assigned to α-Li₂TiO₃ by Laumann et al.³⁶ Basically, two major diffraction peaks of (200) and (220) appear for α-Li₂TiO₃ in the range of 2θ = 10–80°. The spacings for d₍₂₀₀₎ of α-Li₂TiO₃, d₍₋₁₃₃₎ of monoclinic β-Li₂TiO₃, and d₍₄₀₀₎ of cubic Li₄Ti₅O₁₂ are 0.2075 nm (ref. 36), 0.2075 nm (ICDD: 033-0831), and 0.2091 nm (ICDD: 049-0207), respectively. As shown in Fig. 1, the diffraction peak at 2θ = 43.7° from the intermediate phase was corresponding to 0.2075 nm. This result supports that the Li–Ti–O precursor was not Li₄Ti₅O₁₂. Several diffraction peaks (002), (200), (006) and (062) assigned to β-Li₂TiO₃ were not observed for the Li–Ti–O precursor. Thus, the intermediate phase was deduced to be nominal α-Li₂TiO₃, namely some lithium defects in the α-Li₂TiO₃.

Fig. 1 XRD patterns of precipitates obtained by method I. The precursor obtained in the first stage (a), the final product in the second step (b), and the final product after calcination at 500 °C for 3 h (c). The top of this figure shows a diffraction pattern of Li₄Ti₅O₁₂ (ICDD: 049-0207).

The diffraction peaks of the final product after the second solvothermal step at a relatively low pH value (9–10) agreed with those of spinel Li₄Ti₅O₁₂, although the diffraction bands corresponding to the (111) and (311) planes were broadened (Fig. 1b). The diffraction peak position (2θ = 43.3°) of the final product, which corresponded to the spinel Li₄Ti₅O₁₂, shifted from that of the Li–Ti–O precursor (2θ = 43.7°). Moreover, the diffraction peaks of (020), (110), (−111), and (021) for β-Li₂TiO₃ were not observed in the range of 2θ = 20–22°. Thus, we succeeded in the production of spinel Li₄Ti₅O₁₂ from the Li–Ti–O precursor by the solvothermal method without subsequent heat treatment.

Table 1 shows various sizes of the obtained Li₄Ti₅O₁₂ samples. The table includes the FWHM of the diffraction peaks, the crystalline domain sizes, D₍₁₁₁₎ and D₍₄₀₀₎, the particle sizes D_TEM estimated by TEM, and the particle sizes D_BET calculated from specific surface area, D_BET = 6/(S_BETρ). ρ is the density of the Li₄Ti₅O₁₂. The weak (111) signal of Li₄Ti₅O₁₂ is deduced to be a lattice distortion in the [111] direction. In the second solvothermal step, Li₄Ti₅O₁₂ was formed from α-Li₂TiO₃ which was produced in the first step. The (400) plain of Li₄Ti₅O₁₂ is inherited from the (200) plain of cubic α-Li₂TiO₃ because of the agreement of their d-spacings. On the other hand, the (111) plane of Li₄Ti₅O₁₂ would be gradually formed because no similar planes exist in the precursor phase. Thus, we observed a strong signal due to the highly ordered (400) plane and a relatively weak signal due to the disordered (111) plane.

Table 1 FWHM of the diffraction peaks, the crystalline domain sizes D_(hkl), the particle sizes D_TEM observed by TEM, the particle sizes D_BET calculated from specific surface area, and the specific surface areas S_BET for obtained Li₄Ti₅O₁₂ samples. D_BET = 6/(S_BETρ). ρ is the density of the Li₄Ti₅O₁₂

	FWHM(111)/degree	FWHM(400)/degree	D(111)/nm	D(400)/nm	DTEM/nm	DBET/nm	SBET/m^2 g^−1
a These values are affected by the lattice distortion.
Method I	2.97	0.71	2.7^a	13.0	11.7 (4.9)	10.8	159.1
Method II	2.07	0.72	3.9^a	12.7	13.0 (4.1)	10.9	157.9
Method I calcined	0.83	0.63	10.2	14.9	11.7 (3.3)	12.7	134.8
Method II calcined	0.67	0.67	13.0	13.8	13.9 (4.9)	17.1	100.5

---

Fig. 2 shows TEM images of the products. Round-shaped particles with a diameter of ca. 13 nm were observed for the Li–Ti–O precursor (Fig. 2a). Almost the same sized particles of Li₄Ti₅O₁₂ crystal were obtained by the second step. The lattice fringes corresponding to the (111) plane of Li₄Ti₅O₁₂ support that the crystalline nanoparticles were generated by the second step (Fig. 2b). The crystalline domain sizes calculated by the Scherrer equation are D₍₁₁₁₎ = 2.7 nm (as-prepared sample), D₍₄₀₀₎ = 13.0 nm (as-prepared sample), D₍₁₁₁₎ = 10.2 nm (calcined sample), and D₍₄₀₀₎ = 14.9 nm (calcined sample). The value of D₍₄₀₀₎ = 13.0 nm for the as-prepared sample was almost corresponded to the particles size D_TEM = 11.7 nm observed by TEM. On the other hand, the value of D₍₁₁₁₎ was smaller than D_TEM. The difference suggests the presence of a lattice distortion in the [111] direction. Because D₍₁₁₁₎ was clearly increased by the calcination, the crystallinity of the calcined sample is better than that of the as-prepared sample.

Fig. 2 TEM images of the precipitates obtained by method I. The precursor obtained in the first stage (a), the final product in the second step (b), and the final product after calcination at 500 °C for 3 h (c). The insets in (b) and (c) show the lattice fringes of the (111) plane of Li₄Ti₅O₁₂ (d₍₁₁₁₎ = 0.48 nm).

Here we discuss the formation mechanism of the spinel phase under the solvothermal conditions (Formulas (1)–(3)). Amorphous titanium dioxide was precipitated through a hydrolysis of bis(ammonium lactate) titanium dihydroxide under a basic condition before the first solvothermal step. The Li–Ti–O precursor, nominal α-Li₂TiO₃, then formed under a basic condition over pH 11 in the first solvothermal step. In the previous study, the transformation of the Li–Ti–O precursor to Li₄Ti₅O₁₂ requires heat treatment at temperatures above 400 °C. On the other hand, α-Li₂TiO₃ is delithiated even at room temperature in pure water or an acid solution.^14,36 Thus, (Li_2−xH_x)TiO₃ could be formed through delithiation of the precursor by lowering the pH to ∼10 by the addition of acetic acid in the second step. Finally, the spinel Li₄Ti₅O₁₂ is produced by dehydration of (Li_2−xH_x)TiO₃. (x = 1.2 is confirmed for formation of the spinel Li₄Ti₅O₁₂). When the pH value was decreased to below pH 9 by the addition of a large amount of acetic acid, we observed the formation of anatase TiO₂ through H₂TiO₃ due to excess delithiation. This fact supports the formation mechanism of the spinel phase in the solution system proposed here. When hydrochloric acid or nitric acid was used as a pH controller instead of acetic acid, spinel Li₄Ti₅O₁₂ was also produced. These results support the role of the acid in the reaction.

2LiOH + TiO₂ → Li₂TiO₃ + H₂O (>pH 11) (1)

Li₂TiO₃ + xH⁺ → (Li_2−xH_x)TiO₃ + xLi⁺ (delithiation) (2)

(Li_2−xH_x)TiO₃ → Li_2−xTiO_3−x/2 + x/2H₂O (dehydration) (3)

3.2. Basicity-controlled one-step two-phase method (method II)

Based on the reaction process of the two-step solvothermal method mentioned above, a delithiation process induced by decreasing the basicity is essential for the formation of the spinel Li₄Ti₅O₁₂ phase in the solution system. Next we developed the one-pot synthesis of the spinel Li₄Ti₅O₁₂ nanocrystal with self-control of the basicity using the two-phase system of water–ethanol and toluene–oleic acid. As shown in Fig. 3, the pure spinel phase was directly obtained after the solvothermal reaction at 200 °C for 24 h, whereas the Li–Ti–O precursor was observed for 12 h. The Li–Ti–O precursor was produced under a relatively high pH condition due to the dissolution of lithium hydroxide in the initial stage. The Li₄Ti₅O₁₂ crystals were then formed from the precursor by decreasing the basicity in the progressive stage. This means that a gradual transfer of a certain amount of the organic acid from the toluene phase lowers the basicity of the water phase by disassociation. tert-Butylamine was needed for the two-phase solvothermal process because a slight amount of anatase TiO₂ as an impurity formed without using tert-butylamine. The presence of tert-butylamine may support decreasing the pH stably or stabilizing titanium hydroxide ions as dissolution species. Additionally, the formation of the Li₄Ti₅O₁₂ nanocrystal required a solvothermal temperature equal to or higher than 180 °C (Fig. S2†). The particle size and the specific surface area of the products were ca. 13 nm (Fig. 4) and 158 m² g⁻¹, respectively. The crystallinity was improved by calcination at 500 °C with a slight increase in the particle size.

Fig. 3 XRD patterns of the precursor obtained by method II in the initial stage (12 h) (a), the Li₄Ti₅O₁₂ crystals in the progressive stage (24 h) (b), and after calcination at 500 °C for 3 h (c).

Fig. 4 TEM images of the Li₄Ti₅O₁₂ crystals prepared by method II (a) and after calcination at 500 °C for 3 h (b).

3.3. Electrochemical analyses of Li₄Ti₅O₁₂ nanocrystals

The electrochemical performance was measured for the Li–Ti–O precursor and Li₄Ti₅O₁₂ nanocrystals obtained through methods I and II. The samples calcined at 500 °C were also characterized under the same conditions. Fig. 5 and 6 show charge–discharge curves and the cycle and rate performances, respectively, at several current rates in the voltage between 1.0 and 2.7 V. We observed unclear voltage plateaus and low capacity values for the Li–Ti–O precursor (a) and the calcined one (a′). On the other hand, the Li₄Ti₅O₁₂ nanocrystals obtained by method I (b) showed a reaction voltage plateau around 1.55 V (vs. Li/Li⁺), and the first discharge and reversible capacity were 113 and 102 mA h g⁻¹, respectively, at 0.20 C (1.0 C = 175 mA g⁻¹). The first reversible capacity of the nanocrystals was increased to 128 mA h g⁻¹ by the calcination (b′). The capacity of the Li₄Ti₅O₁₂ crystals obtained by method II (c) was 143 mA h g⁻¹, which was even higher than that of the calcined one (c′) (137 mA h g⁻¹) and was comparable to the results of well-crystallined Li₄Ti₅O₁₂ reported in a previous article.¹⁵ These results indicate that the Li₄Ti₅O₁₂ nanocrystals obtained directly by method II showed a good capacity even without heat treatment. The capacity value of as-prepared Li₄Ti₅O₁₂ nanocrystals (b and c) was decreased by increasing the rate current. On the other hand, the performance of the Li₄Ti₅O₁₂ nanocrystals was improved by the subsequent calcination (b′ and c′). A high crystallinity would be required for a fast electrode reaction at a high rate. Because the difference in the crystallinity of the calcined samples is smaller than that of the as-prepared samples, the difference in reversible capacity of the calcined samples obtained method I and II was decreased. According to the difference of FWHM₍₁₁₁₎ (2.97° (method I) and 2.07° (method II)), the crystallinity of the as-prepared sample of method II was higher than that of method I. We think that the performance was not affected by tert-butylamine because the organic molecules were easily removed in the washing process. In the one-step process, the Li₄Ti₅O₁₂ nanocrystals are directly formed through the transformation of the fresh precursor phase in the solution system. Thus, the higher crystalline products are easily obtained by this simple technique. At a low current rate (0.20–1.0 C), the performance of as-prepared Li₄Ti₅O₁₂ nanocrystals by method II (c) was higher than that of the calcined one. The higher performance of the Li₄Ti₅O₁₂ nanocrystals may be attributed to the coexistence of the high crystallinity and the high specific surface area due to loose aggregation without the calcination process. This suggests the advantage of direct synthesis of Li₄Ti₅O₁₂ nanocrystals by the solvothermal method. Tailing was observed in the first discharge profiles (0.20 C). According to ref. 34, the irreversible capacity was ascribed to the surface reaction of nanosized Li₄Ti₅O₁₂ samples. Therefore, the origin of tailing in the first discharge curves for our samples would be attributed to the presence of a high surface area originating form nanosized Li₄Ti₅O₁₂ particles. In the previous works, the electrochemical performance of as-prepared sample obtained by the solvothermal method was quite lower than that of calcined samples due to their lower crystallinity. Highly crystalline products can be prepared even through the low-temperature process by adjusting the preparation conditions. The direct synthesis of the products exhibiting a high performance would enable a low-temperature technique which is needed such for low energy consumption process, fabrication of nanoscale textures, and surface modification with organic molecules.

Fig. 5 Charge–discharge curves at various current rates (1.0 C = 175 mA g⁻¹) of the Li–Ti–O precursor (a), the Li₄Ti₅O₁₂ nanocrystals obtained by methods I and II (b and c, respectively), and the calcined samples at 500 °C for 3 h (a′, b′, and c′).

Fig. 6 Rate and cycle performances of the Li–Ti–O precursor (a), the Li₄Ti₅O₁₂ nanocrystals obtained by methods I and II (b and c, respectively), and the calcined samples at 500 °C for 3 h (a′, b′ and c′). The numbers above the plots indicate current rates per C (1.0 C = 175 mA g⁻¹).

4. Conclusions

Spinel Li₄Ti₅O₁₂ nanocrystals were produced through the pH-controlled two-step solvothermal process. We found that the formation of Li₄Ti₅O₁₂ crystals requires the nucleation of the Li–Ti–O precursor under a high pH condition in the initial stage and its transformation with a decrease of pH in the progressive stage of the solvothermal reaction. Then we succeeded in the direct synthesis of spinel Li₄Ti₅O₁₂ nanocrystals by the one-step two-phase solvothermal process using a toluene–oleic acid system as a novel basicity-control method. The Li₄Ti₅O₁₂ nanocrystals of ca. 13 nm in diameter prepared by the one-step two-phase solvothermal process showed a high specific capacity for application as an anode material.

Acknowledgements

This work was partially supported by the Advanced Low Carbon Technology Research and Development Program of Strategic Basic Research Programs from Japan Science and Technology Agency.

Notes and references

1. T. Ohzuku, A. Ueda and N. Yamamoto, J. Electrochem. Soc., 1995, 142, 1431–1435.
2. K. M. Colbow, J. R. Dahn and R. R. Haering, J. Power Sources, 1989, 26, 397–402.
3. K. Zaghib, M. Simoneau, M. Armand and M. Gauthier, J. Power Sources, 1999, 81–82, 300–305.
4. F. Ronci, P. Reale, B. Scrosati, S. Panero, V. R. Albertini, P. Perfetti, M. di Michiel and J. M. Merino, J. Phys. Chem. B, 2002, 106, 3082–3086.
5. H.-G. Jung, S.-T. Myung, C. S. Yoon, S.-B. Son, K. H. Oh, K. Amine, B. Scrosati and Y.-K. Sun, Energy Environ. Sci., 2011, 4, 1345–1351.
6. L. Kavan and M. Grätzel, Electrochem. Solid-State Lett., 2002, 5, A39–A42.
7. A. S. Prakash, P. Manikandan, K. Ramesha, M. Sathiya, J.-M. Tarascon and A. K. Shukla, Chem. Mater., 2010, 22, 2857–2863.
8. C. Jiang, E. Hosono, M. Ichihara, I. Honma and H. Zhou, J. Electrochem. Soc., 2008, 155, A553–A556.
9. L. Shen, C. Yuan, H. Luo, X. Zhang, K. Xu and Y. Xia, J. Mater. Chem., 2010, 20, 6998–7004.
10. Y. Tang, L. Yang, Z. Qui and J. Huang, J. Mater. Chem., 2009, 19, 5980–5984.
11. Y.-S. Lin and J.-G. Duh, J. Power Sources, 2011, 196, 10698–10703.
12. D. Fattakhova and P. Krtil, J. Electrochem. Soc., 2002, 149, A1224–A1229.
13. A. Zhang, Z. M. Zheng, F. Y. Cheng, Z. L. Tao and J. Chen, Sci. China: Chem., 2011, 54, 936–940.
14. J. Lu, C. Nan, Q. Peng and Y. Li, J. Power Sources, 2012, 202, 246–252.
15. H. Yan, Z. Zhu, D. Zhang, W. Li and Qilu, J. Power Sources, 2012, 219, 45–51.
16. Y. F. Tang, L. Yang, Z. Qiu and J. S. Huang, Electrochem. Commun., 2008, 10, 1513–1516.
17. H. Cheng, J. Ma, Z. Zhao and L. Qi, Chem. Mater., 1995, 7, 663–671.
18. Q. Wu, X. Chen, P. Zhang, Y. Han, X. Chen, Y. Yan and S. Li, Cryst. Growth Des., 2008, 8, 3010–3018.
19. V. Subramanian, H. Zhu, R. Vajtai, P. M. Ajayan and B. Wei, J. Phys. Chem. B, 2005, 109, 20207–20214.
20. H.-X. Mai, L.-D. Sun, Y.-W. Zhang, R. Si, W. Feng, H.-P. Zhang, H.-C. Liu and C.-H. Yan, J. Phys. Chem. B, 2005, 109, 24380–24385.
21. C.-T. Dinh, T.-D. Nguyen, F. Kleitz and T.-O. Do, ACS Nano, 2009, 3, 3737–3743.
22. F. Lu, W. Cai and Y. Zhang, Adv. Funct. Mater., 2008, 18, 1047–1056.
23. G. Tian, Y. Chen, W. Zhou, K. Pan, Y. Dong, C. Tian and H. Fu, J. Mater. Chem., 2011, 21, 887–892.
24. A. Testino, M. T. Buscaglia, V. Buscaglia, M. Viviani, C. Bottino and P. Nanni, Chem. Mater., 2004, 16, 1536–1543.
25. M.-H. Um and H. Kumazawa, J. Mater. Sci., 2000, 35, 1295–1300.
26. W. Dong, G. Zhao, B. Song, G. Xu, J. Zhou and G. Han, CrystEngComm, 2012, 14, 6990–6997.
27. D.-S. Seo, H. Kim and J.-K. Lee, J. Cryst. Growth, 2005, 275, e2371–e2376.
28. C.-Y. Xu, Q. Zhang, H. Zhang, L. Zhen, J. Tang and L.-C. Qin, J. Am. Chem. Soc., 2005, 127, 11584–11585.
29. N. Masaki, S. Uchida and T. Sato, J. Mater. Chem., 2002, 12, 305–308.
30. M. Tomiha, N. Masaki, S. Uchida and T. Sato, J. Mater. Sci., 2002, 37, 2341–2344.
31. M. Sugita, M. Tsuji and M. Abe, Bull. Chem. Soc. Jpn., 1990, 63, 1978–1984.
32. D. Fattakhova, V. Petrykin, J. Brus, T. Kostlánová, J. Dědeček and P. Krtil, Solid State Ionics, 2005, 176, 1877–1885.
33. S.-H. Yu, A. Pucci, T. Herntrich, M.-G. Willinger, S.-H. Baek, Y.-E. Sung and N. Pinna, J. Mater. Chem., 2011, 21, 806–810.
34. A. Laumann, M. Bremholm, P. Hald, M. Holzapfel, K. T. Fehr and B. B. Iversen, J. Electrochem. Soc., 2012, 159, A166–A171.
35. A. Laumann, K. M. Ø. Jensen, C. Tyrsted, M. Bremholm, K. T. Fehr, M. Holzapfel and B. B. Iversen, Eur. J. Inorg. Chem., 2011, 2221–2226.
36. A. Laumann, K. T. Fehr, M. Wachsmann, M. Holzapfel and B. B. Iversen, Solid State Ionics, 2010, 181, 1525–1529.

---

Footnote

† Electronic supplementary information (ESI) available: Fig. S1 and S2. See DOI: 10.1039/c4ra06646d

---