One-dimensional nanofiber architecture of an anatase TiO₂–carbon composite with improved sodium storage performance

Abstract

Titanium-based oxides have been considered promising anode materials for rechargeable sodium (Na)-ion batteries (NIBs); however, the slow reaction kinetics of TiO₂ due to its low electrical conductivity remains a critical problem to be addressed for its practical application in NIBs. Here, we report one-dimensional (1D) TiO₂–C composite nanofibers for use as an anode for NIBs, which were fabricated via an electrospinning technique combined with optimized heat-treatment. The TiO₂–C composite nanofibers are composed of tiny anatase TiO₂ nanocrystals surrounded by and/or embedded in the conducting carbon matrix. When tested in an electrochemical Na cell, the TiO₂–C nanofiber electrode retained a high charge capacity of 140 mA h g⁻¹, even after 100 cycles at a current density of 60 mA g⁻¹, and it exhibited excellent rate-capability (105 mA h g⁻¹ at 3 A g⁻¹), proving its feasibility as an NIB anode. The superior Na storage performance of the nanofiber anode is mainly attributed to the unique composite architecture, that is, a large number of active sites per volume due to the high aspect ratios and short lengths for Na transport through 1D nanofibers and TiO₂ nanocrystals as well as highly conducting electron paths.

---

Introduction

Recently, ambient temperature sodium (Na)-ion batteries (NIBs) have attracted great attention as alternative energy storage systems for today's lithium-ion batteries (LIBs) for large-scale stationary applications due to the low cost and wide availability of Na resources.^1–3 Although the operation mechanism of NIBs is closely analogous to that of LIBs, there are several fundamental differences in behavior between lithiation and sodiation reactions involved in battery operations. In particular, the larger radius of Na⁺ ions in comparison to Li⁺ ions significantly limits the electrochemical performance of electrode materials used for NIBs.^1–3 The discovery and optimization of promising electrode materials with high capacity, rate-capability, cyclability, and safety are therefore crucial to the commercial success of NIB technology.

To date, a wide range of materials, including carbons and metallic and non-metallic compounds, have been explored as anodes for NIBs.^4,5 For carbonaceous materials, much research focus has been on disordered carbons with low degrees of crystallinity because graphite is known to accommodate only a very small amount of Na. Disordered carbons, such as hard carbon and amorphous carbon, can deliver reversible capacities of 200–350 mA h g⁻¹, depending on their structural properties; however, the sodiation process of carbon at low potentials (close to Na/Na⁺ redox potential) can cause metallic Na plating at high currents, leading to a safety concern.^6–9 Metallic and non-metallic elements based on Sn, Sb, As, and/or P have proven to be electrochemically active for the sodiation reaction to form Na alloys and to show much higher specific capacities compared to carbon.^10–13 As in the case of LIBs, however, Na alloy compounds suffer from very large volume changes upon sodiation and desodiation, resulting in considerable loss of the structural integrity of the electrodes during repeated charge–discharge cycles.

Another class of anode materials that is known to possess Na storage capability is titanium-based compounds.^14–17 In particular, TiO₂ polymorphs offer many advantages, i.e., natural abundance, non-toxicity, and low cost. Recent studies have shown that anatase TiO₂ could deliver specific capacities in the range of ca. 100–200 mA h g⁻¹, depending on its particle size and morphology as well as the surface chemistry.^18–26 Kim et al. provided experimental evidence that the sodiation/desodiation processes are accompanied by the Ti⁴⁺/³⁺ redox reaction, suggesting that Na could be reversibly intercalated into and deintercalated from anatase TiO₂ without causing substantial structural/morphological changes.¹⁹ In addition, a safety issue related to Na plating can be avoided with TiO₂ due to relatively higher potentials for Na intercalation (ca. 0.2–2.0 V vs. Na/Na⁺) in TiO₂ than in carbon.

The sluggish reaction kinetics of TiO₂ due to its low electrical conductivity represents a critical problem to be addressed for practical application in NIBs. Several attempts have been made to overcome this problem, including the preparation of nano-sized particles with reduced transport lengths, surface modification with conductive carbon, and doping of metallic elements into the structure.^18–26 According to the recent experimental studies mentioned above, an effective engineering approach for high-performance NIB anodes would be to fabricate a composite architecture with controlled shape and morphology in which tiny TiO₂ nanocrystals are surrounded by conductive carbonaceous materials.

Keeping this design principle in mind, herein, we report one-dimensional (1D) composite nanofibers comprising anatase TiO₂ nanocrystals and a conductive carbon network as an anode for NIBs. These 1D nanofiber architectures with high aspect ratios provide additional benefits, such as large surface areas per volume and short transport lengths in the radial direction, which make them attractive for use as battery electrodes. Although TiO₂-based composite nanofibers have been extensively investigated for LIB applications, few studies on their electrochemical behaviors in NIBs have been reported yet. In this work, the TiO₂–C composite nanofibers were synthesized by an electrospinning technique combined with optimized heat-treatment, and then, their Na storage performance was characterized in terms of capacity, cycling stability, and rate-capability. As will be shown later, the results demonstrate that the composite nanofiber anode exhibits promising electrochemical properties as an NIB anode owing to the fibers' unique architectures.

Experimental

Materials preparation

As schematically illustrated in Fig. 1, the composite nanofibers (NFs) of anatase TiO₂ and carbon (TiO₂–C) were fabricated by electrospinning combined with a post-heat-treatment process. In a typical synthesis, 0.514 g of poly-vinylpyrrolidone (PVP, M_w = 1 300 000, Aldrich) and 1.5 g of titanium(IV) butoxide (TiBuO₄ (C₁₆H₃₆O₄Ti), 97%, Aldrich) were dissolved in 10 mL of ethanol and 3 mL of acetic acid. Then, the mixture was continuously stirred at room temperature for 48 h to form a homogeneous solution. The resulting solution was electrospun through a needle under a DC voltage of 17 kV and collected on a drum as a fibrous mat. After this, TiO₂–C NFs were obtained by two-step heat-treatment of the as-electrospun PVP–TiBuO₄ fibers in air at 280 °C for 5 h and in Ar at 500 °C for 2 h. Anatase TiO₂ nanocrystals and amorphous carbon are believed to be produced simultaneously via hydrolysis of Ti(OC₄H₉)₄ (Ti(OC₄H₉)₄ + 2H₂O → TiO₂ + 4C₄H₉OH) and carbonization of PVP, respectively, during the heat-treatment process at 500 °C.²⁶ The electrical properties of the PVP-derived carbon could have been improved by increasing the carbonization temperature beyond 500 °C. During heat-treatment at high temperatures (>550 °C), however, anatase TiO₂ undergoes the phase transformation to the rutile phase with a lower specific capacity. In this work, therefore, TiO₂–C NFs were fabricated at 500 °C to prevent the formation of rutile TiO₂. For comparison, carbon-free, pure TiO₂ NFs were prepared through two-step heat-treatment of the PVP–TiBuO₄ fibers in air at 280 °C for 5 h and in air at 450 °C for 6 h.

Fig. 1 Schematic illustration of the fabrication process of the TiO₂–C composite nanofibers.

Materials characterization

Phase and crystal structure analysis was performed with an automated HPC-2500 X-ray diffractometer (Gogaku) using Cu K_α radiation (λ = 1.5405 Å). Thermogravimetric analysis (TGA) was carried out using a TGA/SDTA851e-METTLER instrument with a heating rate of 2 °C min⁻¹ in air. Raman spectra were recorded using a DXR Raman spectrometer (Thermo Fisher Scientific) with 532 nm laser excitation. The morphology, microstructure and composition of the synthesized samples were examined by scanning electron microscopy (SEM, Hitachi S4700) and transmission electron microscopy (TEM, TECNAI G2 F30S-Twin) in conjunction with energy dispersive X-ray spectroscopy (EDS). Brunauer–Emmett–Teller (BET) surface areas were determined from N₂ sorption isotherms by using a BEL-SORP mini system. X-ray photoelectron spectroscopy (XPS) was conducted using a Thermo MultiLab 2000 system with a monochromatic Al K_α X-ray source.

Electrochemical experiments

To make the electrodes, a slurry was prepared by mixing the active material (70 wt%), carbon black (15 wt%, super-C), and PVdF-co-HFP binder (15 wt%, Kynar2801) in N-methylpyrrolidone (NMP). The same amount of carbon black was incorporated into the carbon-free and carbon-containing TiO₂ NF electrodes in an attempt to make a comparative evaluation of the electrochemical properties of the two synthesized materials. The slurry was then coated on a Cu foil, followed by drying under vacuum at 60 °C for 12 h. The electrodes were finally pressed with a twin roller. The total weight of the electrode (active material, carbon black, and binder) was ca. 1.5 mg cm⁻². Electrochemical experiments were performed using a coin-type cell (CR2032) with a sodium metal counter electrode. The separator was a glass fiber sheet, and the electrolyte was 1 M NaClO₄ dissolved in a mixed solvent of ethylene carbonate (EC) and propylene carbonate (PC) (1 : 1 in volume). All of the cells were assembled in a glove box filled with purified Ar gas. The galvanostatic charge–discharge experiments were performed with Toscat 3100 series at various current densities of 0.03–3 A g⁻¹ in a voltage range of 0.02–2.0 V vs. Na/Na⁺. Note that the specific capacity was calculated based on the electrode's weight excluding the weights of binder and carbon black. Electrochemical impedance spectroscopy was carried out at an open circuit potential using a Zahner IM6. The impedance spectra were obtained with an ac signal of 5 mV amplitude over a frequency range of 10⁻² to 10⁵ Hz.

Results and discussion

Fig. 2(a–c) presents SEM images of the PVP–TiBuO₄, TiO₂, and TiO₂–C NFs, respectively. All of the NF samples show typical 1D fibrous morphologies of <200 nm diameters with bead-free, smooth surfaces. The SEM micrographs also confirm that the post-heat-treatment process caused no fragmentation of the fibers, resulting in long 1D TiO₂ and TiO₂–C NFs with high aspect ratios. The TGA analysis (Fig. S1†) indicated that the TiO₂–C NFs contained ca. 25 wt% carbon. The structural information of the TiO₂ and TiO₂–C NF samples was acquired by X-ray diffraction (XRD) analysis (Fig. 2(d)). The XRD data for both of the NF samples show similar profiles with diffraction peaks for the tetragonal anatase TiO₂ phase (space group I4₁/amd, JCPDS no. 21-1272). No secondary or impurity phases were detected. The diffraction peaks for TiO₂ in TiO₂–C NFs were found to have lower intensities and larger widths compared to those for the TiO₂ NFs, which may be attributed to the presence of the PVP-derived carbon in TiO₂–C NFs. To examine the structural characteristics of the PVP-derived carbon in TiO₂–C NFs, the Raman spectra were measured as shown in Fig. 2(e). For comparison, the data were also obtained from TiO₂ NFs. In Fig. 2(e), the characteristic bands observed at 140–650 cm⁻¹ correspond to the vibrational modes of anatase TiO₂ phase: 145 cm⁻¹ and 637 cm⁻¹ for E_g, 396 cm⁻¹ for B_1g, and 514 cm⁻¹ for A_1g.^21,27 It was reported that PVP undergoes complete decomposition below around 250 °C.²⁸ The Raman spectrum of TiO₂–C NFs shows additional broad bands at ca. 1350 cm⁻¹ and ca. 1580 cm⁻¹ corresponding to D (disordered) and G (ordered) bands of carbon, respectively, without any residual organic components.²⁹ The value of I_D/I_G (relative intensity of D to G band) for TiO₂–C NF was measured to be ca. 0.94, which indicates the formation of disordered (amorphous) carbon at 500 °C.

Fig. 2 SEM micrographs of (a) PVP–TiBuO₄, (b) TiO₂ and (c) TiO₂–C NFs, (d) XRD patterns, and (e) Raman spectra of TiO₂ and TiO₂–C NFs.

The structural details of the synthesized NFs in nano- and micro-scales were investigated using various tools as shown in Fig. 3. The key experimental findings include the following. First, high-resolution TEM analysis (Fig. 3(a)) revealed that TiO₂ NFs consist of tiny nanocrystals with sizes of ca. 3–8 nm and meso-scale pores as seen in the figure. The lattice fringes with d-spacing values of 0.352 and 0.190 nm are assigned to the (101) and (200) planes of anatase TiO₂, respectively. Furthermore, the selected area electron diffraction (SAED) pattern (the inset in Fig. 3(a)) shows the characteristic diffraction rings corresponding to the (101), (004), and (200) planes of the tetragonal anatase TiO₂ phase, which is in accordance with the XRD result (Fig. 2(d)). Second, the TEM micrograph of Fig. 3(b) shows that TiO₂–C NFs are made of smaller TiO₂ crystals (ca. 2–5 nm) (compared with TiO₂ NFs) and an amorphous carbon network. The EDS mapping analysis (Fig. 3(c)) confirmed the uniform distribution of Ti, O, and C throughout TiO₂–C NFs. Third, the analysis of the nitrogen adsorption–desorption isotherms (Fig. 3(d)) indicated that TiO₂ NFs have a mesoporous structure as evidenced by a type-IV isotherm, whereas TiO₂–C NFs show a greatly reduced porosity compared to TiO₂ NFs.^18,30 The BET surface areas of TiO₂ and TiO₂–C NFs were estimated to be 55 m² g⁻¹ and 12 m² g⁻¹, respectively. The pore size distribution curves (Fig. S2†) reproduced from Fig. 3(d) clearly show that TiO₂–C NFs have a significantly reduced amount of meso-scale pores as compared with TiO₂ NFs. This indicates that the reduced porosity of TiO₂–C NFs is mainly due to the formation of a carbon network in mesopores among TiO₂ nanocrystals. In summary, the TEM, EDS, and BET analyses confirmed that a composite architecture with 1D fibrous morphologies was successfully fabricated, in which tiny anatase TiO₂ nanocrystals are surrounded by and/or embedded in a conductive carbon network. This unique architecture is expected to play a beneficial role in achieving high electrochemical performance of TiO₂–C by providing (i) very short lengths for Na transport through 1D NFs and TiO₂ nanocrystals and (ii) conducting electron paths around TiO₂.

Fig. 3 TEM micrographs of (a) TiO₂ and (b) TiO₂–C NFs. The inset in (a) displays the selected area diffraction patterns for the TiO₂ NF sample. (c) EDS mapping of Ti, O, and C elements in TiO₂–C NFs. (d) N₂ adsorption–desorption isotherms for TiO₂ and TiO₂–C NFs.

To evaluate the electrochemical properties of TiO₂–C NFs, we constructed and tested a half-cell using a metallic Na and NaClO₄–EC:PC as the counter electrode and electrolyte, respectively. Fig. 4(a and b) presents the galvanostatic discharge (sodiation) and charge (desodiation) profiles of TiO₂ and TiO₂–C NFs, respectively, measured for the first five cycles. The experiments were conducted with cut-off voltages of 0.02 and 2.0 V vs. Na/Na⁺ at a current density of 60 mA g⁻¹. The initial discharge and charge capacities (see inset in Fig. 4(a)) of the TiO₂ NF anode were determined to be 504 and 176 mA h g⁻¹, respectively, resulting in a coulombic efficiency as low as 35% (Fig. 4(a)). It should be noted from Fig. 4(b) that the TiO₂–C NF anode showed a similar initial coulombic efficiency (ca. 34%), but it delivered higher capacities (619 mA h g⁻¹ for discharge and 212 mA h g⁻¹ for charge) in comparison with the TiO₂ NF anode (the possible contribution of carbon to the total capacity of TiO₂–C NFs is discussed in Fig. S3†). Some broad peaks were observed in the voltage range of 1.2–0.2 V vs. Na/Na⁺ on the plot of differential capacity (dQ/dV) vs. voltage (Fig. 4(c)) for the initial discharge step; however, those peaks disappeared during subsequent charge–discharge cycles. The low coulombic efficiency and irreversible dQ/dV peaks during the first cycle indicate that considerable parasitic reactions, such as the formation of solid electrolyte interphase (SEI) and electrolyte decomposition, take place during the first sodiation process, which is typical of TiO₂ anodes in Na-based cells.^18–20,22

Fig. 4 Discharge–charge profiles of the (a) TiO₂ and (b) TiO₂–C NF anodes measured at the current density of 60 mA g⁻¹ during the initial 5 cycles. (c) Differential capacity (dQ/dV) vs. voltage plots reproduced from the 1st, 3rd and 5th discharge–charge profiles of the TiO₂–C NF anode. (d) Ti 2p XPS spectra for the pristine, discharged, and recharged TiO₂–C NF anodes.

The XPS spectra of the Ti 2p region for TiO₂–C NFs were measured to follow the evolution of the Ti oxidation states upon sodiation and desodiation, as shown in Fig. 4(d). The observed spectrum of the pristine electrode shows two major peaks at binding energies (BEs) of 458.8 eV and 464.4 eV for Ti 2p_3/2 and Ti 2p_1/2, respectively, indicating that Ti is predominantly present in a tetravalent Ti⁴⁺ form.^18,21 The peak shift in the lower BE direction observed in the XPS spectrum of the discharged electrode indicates that sodiation induces the partial oxidation of Ti⁴⁺ to Ti³⁺. Upon charge, the XPS peaks shifted back to the original BEs, which implies that Ti³⁺ is re-oxidized to Ti⁴⁺ during desodiation. The XPS results presented here further support the view that the electrochemical reaction mechanism is based on reversible Na intercalation/deintercalation in the TiO₂–C NF anode.²²

The cycling performance results of the Na cells with the TiO₂ and TiO₂–C NF anodes are shown in Fig. 5(a). The capacity decreased rapidly within 15 cycles and subsequently reached an almost constant value. After 100 cycles, the TiO₂–C NF anode delivered a charge capacity as high as 140 mA h g⁻¹, which is much higher than that of the TiO₂ NF anode (54 mA h g⁻¹). When normalized with respect to the value of the charge capacity at the 15th cycle, the capacity retentions for the TiO₂ and TiO₂–C NF anodes at the 100th cycle were approximately 83% and 56%, respectively. We speculate that the enhanced cyclability of the TiO₂–C NF anode is largely due to the improved structural integrity of the electrode as well as the presence of highly electronic conducting paths. This is further supported by the ac-impedance results in Fig. 5(b). The interfacial polarization resistance of the TiO₂ NF anode, determined from the magnitude of a high-frequency arc, increased from ca. 115 Ω to ca. 468 Ω after 30 cycles, which is more than a four-fold increase. On the other hand, the interfacial resistance for the TiO₂–C NF anode increased only by ca. 120 Ω, indicating that the carbon network formed among TiO₂ nanocrystals is probably responsible for the improved stability observed for TiO₂–C NFs. This is supported by the results of a previous work on LIBs, which demonstrated that the carbon network in a TiO₂ fiber–graphene nanocomposite provided highly electronically conducting pathways and also worked as a protective layer to keep the active material integrated by effectively accommodating the strain during lithium intercalation/deintercalation.²⁹ Although the TiO₂–C NF electrode exhibited stable cycling performance after the initial capacity decay, comprehensive studies should be conducted to find optimum combinations of electrolyte salts and solvents that can work with the TiO₂–C NF anode for NIBs.

Fig. 5 (a) Plots of capacity vs. cycle number for the TiO₂ and TiO₂–C NF anodes. (b) Ac-impedance spectra of the TiO₂ and TiO₂–C NF anodes after the 1st and 30th discharge steps. (c) Discharge–charge profiles of the TiO₂–C NF anode measured at various current densities of 0.03–3 A g⁻¹. (d) Comparison of the rate-capability between the TiO₂–C NF and carbon-coated TiO₂ NF anodes.

The rate-capability of the TiO₂–C NF anode was evaluated as shown in Fig. 5(c). The TiO₂–C NF anode exhibited good charge capacity retention at various current densities of 0.03–3.0 A g⁻¹. At a high current density of 3.0 A g⁻¹, in particular, the charge capacity was determined to be as high as 105 mA h g⁻¹. It is evident that the conductive carbon network, TiO₂ nanocrystals and 1D fibrous structures play a synergistic role in achieving high rate-capability. In an attempt to demonstrate the beneficial role of a “composite” architecture having the carbon network and TiO₂ nanocrystals, we fabricated another TiO₂-based NF anode in which carbon was mostly “coated” on the outer surfaces of TiO₂ NFs. Carbon coating was achieved through the impregnation of TiO₂ NFs in a sucrose-dissolved solution followed by the thermal decomposition and carbonization of sucrose on the NF surfaces in air.²² As shown in Fig. S4,† a thin layer of carbon with several nanometers in thickness was produced on the fiber surfaces rather than the formation of a continuous network throughout the fibers. The galvanostatic discharge–charge measurements at various rates demonstrate that the rate-capability of the carbon-coated TiO₂ NF anode is inferior to that of the TiO₂–C NF anode (Fig. 5(d)). This strongly suggests that the conductive carbon network formed throughout the whole NF structure has indeed a key role in improving electrochemical performance.

Conclusions

In summary, we have developed an anatase TiO₂–C composite nanofiber for use as an NIB anode, which is produced by electrospinning and post-heat-treatment. The TiO₂–C composite nanofiber anode exhibits improved sodium storage capacity and cycle performance in comparison with a carbon-free TiO₂ nanofiber anode. The enhanced performance of the composite nanofibers is mainly due to the unique 1-D composite architecture consisting of tiny TiO₂ nanocrystals surrounded by a conductive carbon network, which provides short lengths for Na transport and highly conducting electron paths. In addition to demonstrating the feasibility of the anatase TiO₂–C composite nanofibers as an anode for NIBs, this work clearly demonstrated that the controlled nanostructural design of electrodes could be an effective approach to improving Na storage performance.

Acknowledgements

This work was supported by the R&D Convergence Program (National Research Council of Science & Technology, Project No. CAP-14-2-KITECH).

References

1. V. Palomares, P. Serras, I. Villaluenga, K. B. Hueso, J. Carretero-González and T. Rojo, Energy Environ. Sci., 2012, 5, 5884.
2. S.-W. Kim, D.-H. Seo, X. Ma, G. Ceder and K. Kang, Adv. Energy Mater., 2012, 2, 710.
3. M. D. Slater, D. Kim, E. Lee and C. S. Johnson, Adv. Funct. Mater., 2013, 23, 947.
4. F. Klein, B. Jache, A. Bhide and P. Adelhelm, Phys. Chem. Chem. Phys., 2013, 15, 15876.
5. M. Dahbi, N. Yabuuchi, K. Kubota, K. Tokiwa and S. Komaba, Phys. Chem. Chem. Phys., 2014, 16, 15007.
6. D. A. Stevens and J. R. Dahn, J. Electrochem. Soc., 2000, 147, 1271.
7. S. Komaba, W. Murata, T. Ishikawa, N. Yabuuchi, T. Ozeki, T. Nakayama, A. Ogata, K. Gotoh and K. Fujiwara, Adv. Funct. Mater., 2011, 21, 3859.
8. E. M. Lotfabad, J. Ding, K. Cui, A. Kohandehghan, W. P. Kalisvaart, M. Hazelton and D. Mitlin, ACS Nano, 2014, 8, 7115.
9. Y. Bai, Z. Wang, C. Wu, R. Xu, F. Wu, Y. Liu, H. Li, J. Lu and K. Amine, ACS Appl. Mater. Interfaces, 2015, 7, 5598.
10. J. W. Wang, X. H. Liu, S. X. Mao and J. Y. Huang, Nano Lett., 2012, 12, 5897.
11. A. Darwiche, C. Marino, M. T. Sougrati, B. Fraisse, L. Stievano and L. Monconduit, J. Am. Chem. Soc., 2012, 134, 20805.
12. J. Qian, X. Wu, Y. Cao, X. Ai and H. Yang, Angew. Chem., Int. Ed., 2013, 52, 4633.
13. W. Li, S.-L. Chou, J.-Z. Wang, J. H. Kim, H.-K. Liu and S.-X. Dou, Adv. Mater., 2014, 26, 4037.
14. P. Senguttuvan, G. Rousse, V. Seznec, J.-M. Tarascon and M. R. Palacín, Chem. Mater., 2011, 23, 4109.
15. A. Rudola, K. Saravanan, S. Devaraj, H. Gong and P. Balaya, Chem. Commun., 2013, 49, 7451.
16. Y. S. Wang, X. Q. Yu, S. Y. Xu, J. M. Bai, R. J. Xiao, Y. S. Hu, H. Li, X. Q. Yang, L. Q. Chen and X. J. Huang, Nat. Commun., 2013, 4, 2365.
17. J. P. Huang, D. D. Yuan, H. Z. Zhang, Y. L. Cao, G. R. Li, H. X. Yang and X. P. Gao, RSC Adv., 2013, 3, 12593.
18. Y. Xu, E. M. Lotfabad, H. Wang, B. Farbod, Z. Xu, A. Kohandehghan and D. Mitlin, Chem. Commun., 2013, 49, 8973.
19. K.-T. Kim, G. Ali, K. Y. Chung, C. S. Yoon, H. Yashiro, Y.-K. Sun, J. Lu, K. Amine and S.-T. Myung, Nano Lett., 2014, 14, 416.
20. L. Wu, D. Buchholz, D. Bresser, L. G. Chagas and S. Passerini, J. Power Sources, 2014, 251, 379.
21. L. Wu, D. Bresser, D. Buchholz, G. Giffin, C. R. Castro, A. Ochel and S. Passerini, Adv. Energy Mater., 2015, 5, 1401142.
22. S.-M. Oh, J.-Y. Hwang, C. S. Yoon, J. Lu, K. Amine, I. Belharouak and Y.-K. Sun, ACS Appl. Mater. Interfaces, 2014, 6, 11295.
23. J. Lee, Y.-M. Chen, Y. Zhu and B. D. Vogt, ACS Appl. Mater. Interfaces, 2014, 6, 21011.
24. D. Yan, C. Yu, Y. Bai, W. Zhang, T. Chen, B. Hu, Z. Sun and L. Pan, Chem. Commun., 2015, 51, 8261.
25. D. Bresser, B. Oschmann, M. N. Tahir, F. Mueller, I. Lieberwirth, W. Tremel, R. Zentel and S. Passerini, J. Electrochem. Soc., 2015, 162, A3013.
26. Y. Ge, J. Zhu, Y. Lu, C. Chen, Y. Qiu and X. Zhang, Electrochim. Acta, 2015, 176, 989.
27. N. D. Petkovich, S. G. Rudisill, B. E. Wilson, A. Mukherjee and A. Stein, Inorg. Chem., 2014, 53, 1100.
28. S. M. Hwang, S. Y. Kim, J.-G. Kim, K. J. Kim, J.-W. Lee, M.-S. Park, Y.-J. Kim, M. Shahabuddin, Y. Yamauchi and J. H. Kim, Nanoscale, 2015, 7, 8351.
29. G. Gu, J. Cheng, X. Li, W. Ni, Q. Guan, G. Qu and B. Wang, J. Mater. Chem. A, 2015, 3, 6642.
30. J. Wang, Y. Zhou, Y. Hu, R. O'Hayre and Z. Shao, J. Phys. Chem. C, 2011, 115, 2529.

---

Footnote

† Electronic supplementary information (ESI) available: Thermogravimetric analysis data, pore size distribution, discharge–charge profile, and transmission electron micrograph. See DOI: 10.1039/c5ra14655k

---