Novel 3D flower-like CoNi₂S₄/carbon nanotube composites as high-performance electrode materials for supercapacitors

Abstract

Novel 3D flower-like CoNi₂S₄/carbon nanotube composites with a porous structure were designed through a facile precursor transformation approach as electrode materials for supercapacitors. The resulting samples are characterized by X-ray diffraction, Raman spectra, energy dispersive spectroscopy, field emission scanning electron microscopy, transmission electron microscopy, and nitrogen adsorption–desorption. Their electrochemical performance was investigated by means of cyclic voltammetry, galvanostatic charge–discharge, impedance spectra and cycle life. By selecting carbon nanotubes as the conductive support for the growth of CoNi₂S₄, the as-obtained CoNi₂S₄/carbon nanotube composites displayed an ultrahigh specific capacitance of 2094 F g⁻¹ at 1 A g⁻¹ and a good rate capability (72% capacity retention at 10 A g⁻¹). These results above suggest the great potential of the unique flower-like CoNi₂S₄/carbon nanotube composites in the development of high-performance electrode materials for supercapacitors.

---

1. Introduction

Nowadays, supercapacitors have attracted considerable attention as promising energy storage devices applied in portable electronics and electric or hybrid electric vehicles because of their advantages of fast charge–discharge capability, long lifespan and high energy and power density.^1,2 Generally, the performance of supercapacitors mainly depends on the properties of their electrode materials. The three types of electrode materials used in the supercapacitors are carbon materials, conductive polymers and metal oxides. Carbon based materials have advantages in the aspects of abundant resources, high stability and conductivity, but they have a relatively low specific capacitance from their electronic double layer capacitance. Conductive polymers have higher specific capacitance than carbons, but with poor cycling stability.³ Transition metal oxides usually possess rich oxidation states that are in favor of a high specific capacitance^4,5 and the stability is between that of carbons and conductive polymers; however, their conductivity needs to be further improved. Compared with one-component metal oxides NiO and Co₃O₄, for example, two-component metal oxides, such as spinel NiCo₂O₄, exhibit a higher electronic conductivity and more redox reactions could be carried out to produce faradic capacitance. However, NiCo₂O₄ as an electrode material still suffers the disadvantages of slow redox kinetics, poor electrical conductivity and low mechanical stability. In order to solve these problems, highly conductive carbon materials such as active carbon, carbon nanotubes (CNTs) and graphene are combined with metal oxides to form high-performance composite electrode materials, such as NiO/graphene,⁶ MnO₂/CNT,⁷ V₂O₅/CNT,⁸ V₂O₅/graphene,⁹ NiCo₂O₄/CNT,^10,11 NiCo₂O₄/graphene¹² and RuO₂/graphene.¹³

More recently, NiCo₂S₄ has been reported to exhibit higher conductivity than NiCo₂O₄ owing to the lower optical band gap energy than that of NiCo₂O₄.¹⁴ Similar with NiCo₂O₄, CoNi₂S₄ is also a binary metallic species and possesses richer valence states than the corresponding one-component metallic sulfides nickel sulfide (NiS) and cobalt sulfide (Co₉S₈).¹⁵ Chen¹⁴ and his co-workers synthesized an urchin-like NiCo₂S₄ electrode material by a precursor transformation method and the as-obtained product exhibits a high specific capacitance of 1025 F g⁻¹ at 1 A g⁻¹. More recently, Yu et al. synthesized uniform Ni_xCo_3−xS₄ hollow nanoprisms¹⁶ with thioacetamide (TAA) as the sulfur source. Zhang et al.¹⁷ synthesized urchin-, tube-, flower-, and cubic-like NiCo₂S₄ microstructures through controlling the composition of the solvents and found that the solvent polarity and solubility have a significant effect on the process of nucleation and crystal growth. In order to increase the efficient utilization of electrodes, electroactive materials supported by conductive substrates, such as Ni foam,¹⁸ graphene^19,20 and carbon cloth,^21,22 were also reported. Their electrochemical properties were greatly improved due to the high conductivity, large surface area and three-dimensional network of these substrates. Besides them, carbon nanotubes are also highly conductive and have a large surface area, in favor of electron transportation and material loading. However, there are not very many reports on the study of the combination of CNTs and nickel cobalt sulfides, so more research needs to be done in this area, especially in the area of constructing a novel structure using them.

In this work, a new composite electrode material CoNi₂S₄/CNT was designed by a two-step approach, including the deposition of the Co–Ni precursor on CNTs and the transformation of the Co–Ni precursor via anion exchange. Before the reaction, CNTs were pretreated with acid to increase the amount of oxygen-containing functional groups on the CNT surface in order to improve the deposition of the Co–Ni precursor on the CNT substrate. In the synthesis process, urea, a weak base, was used as a precipitant to control the crystal growth of the nickel–cobalt precursor in the first step and Na₂S was taken as the sulfur source for the anion exchange reaction in the second step. The as-obtained CoNi₂S₄/CNT composites were characterized by XRD, Raman spectra, EDS, FESEM, BET and electrochemical tests. The relation between their unique structural properties and capacitive performance was investigated.

2. Experimental

2.1 Preparation of CoNi₂S₄/CNT composite

Nickel sulfate hexahydrate (NiSO₄·6H₂O), cobalt sulfate heptahydrate (CoSO₄·7H₂O), and urea (CO(NH₂)₂) were purchased from Sigma-Aldrich. All reagents are of analytical grade. MWCNTs (95% purity, average pore size: <8 nm, length: 10–30 μm) were purchased from Timesnano, China. The MWCNTs were pretreated by refluxing in 30 wt% nitrate acid for 12 h at 100 °C. Then the acid-treated CNTs were collected by vacuum filtration and dried at 80 °C for 12 h.

A facile hydrothermal method was used for the synthesis of the Ni–Co sulfide/CNTs composite. Firstly, 4 mmol NiSO₄·6H₂O, 2 mmol CoSO₄·7H₂O and 24 mmol CO(NH₂)₂ were added into 40 mL DI water and stirred for 30 minutes to form a homogeneous solution. Meanwhile, 40 mg of acid-treated CNTs and 15 mg of polyvinyl pyrrolidone (PVP) were put in 40 mL of ethanol and underwent ultrasonic treatment for 30 min. Then the two solutions were mixed and underwent ultrasonic treatment for 30 minutes. Secondly, the mixed solution was transferred into a Teflon-lined stainless steel autoclave and maintained at 120 °C for 6 h. After cooling down to room temperature, the precursor was separated by centrifugation, washed with DI water and ethanol several times, and dried at 80 °C for 12 h.

0.15 g of precursor and 0.54 g of Na₂S·9H₂O were added into 80 mL of DI water with vigorous stirring for 30 minutes. Then the mixture was transferred into a 100 mL Teflon-lined stainless steel autoclave and maintained at 160 °C for 12 h. After cooling down, the product was collected and washed with DI water and ethanol several times, and finally dried at 80 °C.

To study the influence of the CNT content on the composite, different CNT amounts, 0, 30, 40, 50 and 60 mg, were added in the synthesis. The obtained composite samples are denoted as M-0, M-30, M-40, M-50 and M-60, respectively.

2.2 Materials characterization

A CoNi₂S₄/CNTs composite was finally obtained through the above precursor transformation approach. The crystal structure of the composite was studied by X-ray diffraction (XRD, D/max250VB3 + /PC, Rigaku) at a scan rate of 1° s⁻¹ from 10° to 80° (with Cu Ka1 radiation). The Raman spectra of the samples were studied by Laser Micro-Raman spectrometry (Raman, Invia). The morphology was observed by field emission scanning electron microscopy (FESEM, Hitachi S-4800 and JEOL JSM-6060). The elemental compositions were analyzed by employing energy dispersive spectroscopy (EDS, 200 kV). The specific surface area and pore size distribution of the prepared samples were estimated according to N₂ adsorption/desorption isotherms at 77 K (Micromeritics, Tristar 3000 gas adsorption analyzer).

2.3 Electrochemical measurements

The electrochemical properties of the CoNi₂S₄/CNT working electrode were investigated on an electrochemical workstation (CHI660C, Chenhua, Shanghai) by using a standard three-electrode system. The working electrodes were fabricated by mixing the CoNi₂S₄/CNT composite, acetylene black and polytetrafluoroethylene (PTFE) in a weight ratio of 80 : 10 : 10 and then adding about 5 mL of ethanol into the mixture and dispersing in an ultrasonic bath to form a homogeneous paste. Finally, when the ethanol was evaporated, the prepared mixtures were pressed onto a nickel grid under a pressure of 20 MPa for 60 s. 6 mol L⁻¹ KOH aqueous solution was used as the electrolyte; a nickel grid and a saturated calomel electrode (SCE) served as the counter and reference electrodes, respectively. The specific capacitance was strictly and carefully calculated from the discharge curves based on the following equation:where C_p is the specific capacitance (F g⁻¹), I is the discharging current (A g⁻¹), Δt is the discharge time (s), and ΔV is the potential window during the discharge process (V).

3. Results and discussion

3.1 Structure characterization and morphology analysis

Fig. 1a shows the XRD patterns of the precursor. The XRD pattern of the precursor shows nearly the same pattern as that of Ni₂(OH)₂CO₃·4H₂O, except that all the diffraction peaks slightly shift to the high diffraction direction, in accordance with the fact that Co²⁺ or Ni²⁺ always form carbonate hydroxide in a hydrothermal environment in the presence of urea. And this result further suggested that the partial substitution of Ni ions by Co ions does not change the crystal structure except several lattice parameters, owing to the similar atomic radii of Ni and Co.^14,17 After anion exchange reaction with Na₂S, CO₃²⁻ and OH⁻ in the precursor, nickel–cobalt carbonate and hydroxide were gradually replaced by S²⁻ and the precursor was chemically converted into its NiCo₂S₄/CNT counterpart.

Fig. 1 XRD patterns of Co–Ni precursor (a) and the final product CoNi₂S₄/CNT (b).

Fig. 1b shows the XRD patterns of the as-prepared CoNi₂S₄/CNT sample. The characteristic peaks at 16.3°, 26.7°, 31.5°, 38.2°, 47.4°, 50.3° and 55.0°, respectively, correspond to (111), (220), (311), (400), (511) and (440) planes of the cubic CoNi₂S₄ phase (JCPDS Card No. 24-0334). Besides, the two weak peaks in the range between 20° and 25° are believed to be caused by the acid-treated CNTs, which always have a typical peak at 26°.^23,24

Fig. 2 presents the Raman spectra of the pure acid-treated carbon CNTs and the CoNi₂S₄/CNT composites in the range from 800 to 4000 cm⁻¹. Three typical graphic peaks can be seen in both the Raman spectra of the pure acid-treated carbon CNTs and the CoNi₂S₄/CNT composites: the peak at 1348 cm⁻¹, the D band, is attributed to the breathing modes of the rings or κ-point photons of A_1g symmetry, the symbol of the defects on the CNT surface;²⁵ the peak at 1585 cm⁻¹, the G band, is attributed to the splitting E_2g stretching mode of graphite; the peak at 2086 cm⁻¹, the 2D band, results from a harmonic of the D band.²⁶ These two Raman spectroscopies give a good demonstration about the existence of CNTs in the composites.

Fig. 2 Raman spectra of the acid-treated CNTs (a) and the CoNi₂S₄/CNTs composites (b).

The EDS spectrum proves the existence of Co, Ni, S, C and O elements in the CoNi₂S₄/CNT sample as shown in Fig. 3. The elemental molar ratio of Co and Ni is 1 : 2.03, which is very close to the ratio of Ni and Co atoms in the formula of CoNi₂S₄. Noticeably, the content of C is 37.87%, including both the small amount of C in the substrate and other C in the CNTs. The O element is supposed to come from oxygen functional groups on the surface of the CNTs or residual oxygen atoms, because of incomplete sulfur treatment.

Fig. 3 EDS spectrum of the CoNi₂S₄/CNT sample.

The morphological features of the CoNi₂S₄/CNT samples were studied by FESEM, as shown in Fig. 4. Fig. 4(a and b) displays the FESEM image of the precursor, which appears to be a 3D flower-like structure with a diameter of approximately 2 μm and is composed of lots of interconnected nanosheets of about 10 nm in thickness. Besides, it can be found that the MWCNTs (short and tiny rods) are partly enwrapped in the nanosheet and partly lie between nanosheets, as shown in the red rectangle in the high-magnification image of Fig. 4b, in favor of forming the basic 3D flower texture of the flower structure. And the shape was well preserved after sulfidation as can be seen in Fig. 5a.

Fig. 4 FESEM images of the Co–Ni precursor.

Fig. 5 FESEM images of CoNi₂S₄/CNT.

Moreover, it can be observed that the surface of the nanosheets became rather rough and porous, as shown in the high-magnification images in Fig. 5b. Here, it is worthwhile to point out that in the process of crystal growth, PVP, a non-ionic surfactant, could induce faces toward different directions to grow with different speed and lead to the final flower-like structure.

Besides, the TEM characterization of CoNi₂S₄ (a) and CoNi₂S₄/CNT (b) was also performed, as can be seen in Fig. 6. The edges of their nanosheets are presented and compared. There seem to be many slim rods in the nanosheet of CoNi₂S₄/CNT, which is a symbol of the existence of CNTs in the sample. But the lattice spacing was not given due to its weak crystallinity, which is in accordance with the XRD characterization results.

Fig. 6 TEM images of CoNi₂S₄ (a) and CoNi₂S₄/CNT (b).

3.2 Electrochemical properties

Cyclic voltammetry (CV), galvanostatic charge–discharge and electrochemical impedance spectroscopy (EIS) were performed in order to investigate the electrochemical properties of the as-obtained CoNi₂S₄/CNTs composites as shown in Fig. 5.

Fig. 7a shows that the oxidation and reduction peaks can be clearly observed in the potential range of −0.3–0.6 V, suggesting good pseudocapacitive characteristics. With an increase in the scan rate from 5 mV s⁻¹ to 30 mV s⁻¹, the anodic peak shifts to a higher potential and the cathodic peak shifts to a lower potential. And more peaks appear at around 0.2 V, which may result from the abundant Faradic redox processes of Ni²⁺/Ni³⁺ and Co²⁺/Co³⁺/Co⁴⁺ transitions associated with the anion OH⁻. The redox reactions in the alkaline electrolyte are based on the following mechanism:²⁷

NiS + OH⁻ → NiSOH + e⁻ (2)

CoS + OH⁻ → CoSOH + e⁻ (3)

CoSOH + OH⁻ → CoSO + H₂O + e⁻ (4)

Fig. 7 Electrochemical test results of M-40 electrode. (a) CV curves; (b) galvanostatic charge–discharge tests; (c) specific capacitance as a function of discharge current density; (d) cycling stability of 1000-cycles of charge–discharge at 10 A g⁻¹ (the inset shows the last 10 cycles).

The distortion of the CV curves becomes more and more serious as the scan rate increases from 5 mV s⁻¹ to 30 mV s⁻¹. It is because the ionic diffusion rate in the case of high scan rate was not fast enough to satisfy electronic neutralization during the redox reactions.

Fig. 7b shows the galvanostatic charge–discharge curves of the CoNi₂S₄/CNTs electrodes within the potential window between 0 and 0.35 V at the discharge current density of 1, 2, 5, 7 and 10 A g⁻¹, respectively. The shape of the charge–discharge curves exhibits a typical pseudocapacitive behavior. Encouragingly, for M-40, the specific capacitance values are as high as 2094, 2022, 1855, 1694 and 1510 F g⁻¹ at current densities of 1, 2, 5, 7 and 10 A g⁻¹, respectively, as shown in Fig. 7c, which are much higher than those of other NiCo₂S₄ and CoNi₂S₄ materials reported in the literature as listed in Table 1. The higher capacitance of the CoNi₂S₄/CNTs electrode probably stemmed from its novel 3D flower-like structure with many pores formed between interconnected nanosheets, which has a high efficiency of the space and provided enough active sites for electrochemical reactions with electrolytes.

Table 1 Specific capacitance and capacitive retention of CoNi₂S₄ materials from the literature and prepared in this work

Sample	Specific capacitance (F g^−1/1 A g^−1)	Rate retention (%)	Current range (A g^−1)	BET (m^2 g^−1)	Ref.
Urchin-like NiCo2S4	1025	77.3	1–20	—	14
CoNi2S4/graphene	2009	49.8	1–20	48.4	28
Nanostructured NiCo2S4	1050	66.2	1–50	20.3	18
NiCo2S4 hollow	895	65.4	1–20	30.0	16
Nanoprisms
NiCo2S4 nanotubes	933	58.9	1–5	32.0	29
NiCo2S4 nanosheets on graphene	1300 (3 A g^−1)	52.3	3–20	79.1	20
Porous NiCo2S4 hexagonal nanoplates	966	95.5	1–10	17.4	30
CoNi2S4 nanoparticles	1169	60.1	1–5	21.6	18
Mesoporous NiCo2S4 nanoparticles	972 (2 A g^−1)	86.8	2–20	42.8	31
Ball-in-ball hollow NiCo2S4 spheres	1036	68.1	1–20	—	32
Hollow tubular NiCo2S4	1046 (2 A g^−1)	68.0	2–20	108.4	33
3D flower-like CoNi2S4	2094	72.0	1–10	31.5	Our work

---

Cycle stability is also an important factor determining the electrochemical performance of the electrode materials. In the present work, cycle stability experiments were carried out at a charge–discharge current density of 10 A g⁻¹ in the potential range of 0–0.35 V for 1000 repetitive cycles, as shown in Fig. 7(d). The specific capacitance loss after 1000 cycles was 16%, suggesting that the capacity retention was as high as 84%. The last 10 cycles are presented as an inset. It was found that the shapes of the charge–discharge curves were nearly unchanged, indicating the good reversibility and cyclic life of CoNi₂S₄/CNT electrodes.

3.3 The influence of the CNTs added to the composite on the electrochemical properties

Nitrogen adsorption–desorption and electrochemical tests were performed in order to investigate the influence of different CNT amounts added to the composite on the properties of the final samples.

Fig. 8(a) shows the typical CV curves of M-0, M-30, M-40, M-50, and M-60 at a scanning rate of 5 mV s⁻¹ in the potential range from −0.3 V to 0.6 V. Evidently, M-40 has the largest CV area, meaning that the M-40 electrode has the highest specific capacitance. The specific capacitance of the composite electrodes increases with increasing CNT addition from 0 mg to 40 mg, but decreases from 40 mg to 60 mg. Thus the addition of 40 mg of CNTs to the composite is the optimal value. The galvanostatic charge–discharge at 1 A g⁻¹ in a potential window of 0–0.35 V is shown in Fig. 8(b). The sample M-40 exhibits the longest discharge time, suggesting that it has the highest specific capacitance. Reasonably, it is because of the synergistic effect of the pseudocapacitance of CoNi₂S₄ and the double-layer capacitance of the CNTs.³⁴ But when the content of CNTs exceeds 40 mg, the double-layer capacitance of the CNTs has a greater effect on the electrodes compared with the pseudocapacitance of CoNi₂S₄, so that the capacitance is decreased instead.

Fig. 8 (a) CVs; (b) charge–discharge curves; (c) impedance spectra; (d) specific change with different current density of the samples M-0, M-30, M-40, M-50, and M-60.

The test result of electrochemical impedance spectroscopy (EIS) is shown in Fig. 8(c); all EIS curves have a similar shape consisting of a semicircle on the left and an inclined line on the right. Among them, the inclined line of M-40 in the low-frequency region is almost parallel to the vertical axis, which means that the Warburg resistance is rather small and exhibits a fast ion-diffusion behavior in the electrolyte. The change in capacitance with the increasing current density, also called rate capacity, is another important factor other than the specific capacitance and cycle life. As shown in Fig. 8(d), when the current density changes from 1 A g⁻¹ to 10 A g⁻¹, the capacitance retention of M-0, M-30, M-40, M-50, and M-60 are 56%, 57%, 72%, 73%, and 83%, respectively. Interestingly, the rate capacity was improved with the increasing amount of CNTs, which may due to the increase in the conductivity that facilitates access for electron transfer and benefits the fast charging and discharging.¹⁴

BET characterization was carried out to investigate the specific surface area and pore-size distribution of the as-prepared samples. The adsorption/desorption isotherms are of the combination of Type I and Type V isotherms as shown in Fig. 9(a), and with a typical H3 hysteresis loop, suggesting the presence of mesopores for CoNi₂S₄/CNTs composite samples M-0 and M-40. Their pore size distributions are shown in Fig. 9(b). It is clear that most pores are in the range of 4–15 nm. The mesoporous structure possibly came from the slot pores³⁵ between the interconnected nanosheets in the flower structure.

Fig. 9 (a) Nitrogen adsorption–desorption isotherms and (b) the corresponding pore size distributions of M-0 and M-40.

The specific surface area, pore volume and average pore size details are listed in Table 2.

Table 2 Specific surface area, pore volume and average pore size of CoNi₂S₄/CNTs composite samples

Samples	S BET (m^2 g^−1)	V tot (cm^3 g^−1)	PS (nm)
S BET, BET specific surface area; Vtot, total pore volume; PS, average pore size.
M-0	16.2	0.081	19.9
M-40	31.5	0.108	13.7

---

It is noteworthy to point out that the specific area of M-40 was much larger and the average pore size became smaller than that of M-0. The improved specific surface area can be attributed to its hierarchical structure with CNTs embedded in and between the sheets consisting of the flower structure, which is beneficial to the mass and charge transport during the electrode reaction. Besides, the average pore size of M-40 shifted to a smaller size than that of M-0, leading to higher specific surface area and more active sites for electrochemical reactions. Thus both the improved specific area and reasonable mesopore size distribution are responsible for the enhanced electrochemical properties.

4. Conclusion

A 3D flower-like CoNi₂S₄/carbon nanotube composite was synthesized by a facile Co–Ni precursor transformation approach via anion exchange with Na₂S. It exhibits ultrahigh specific capacitance of 2094 F g⁻¹ at 1 A g⁻¹ and a high capacity retention rate of 72% when the discharge current density is increased by ten times from 1 A g⁻¹ to 10 A g⁻¹. Additionally, a desirable cycling stability (83.4% of the initial capacity retention after 1000 cycles at 10 A g⁻¹) was achieved. Such outstanding capacitive behaviors of the CoNi₂S₄/carbon nanotube composite are mainly ascribed to the rich redox reactions that are contributed by both nickel and cobalt ions, the high conductivity of carbon nanotubes, the novel 3D flower-like structure and the optimal mesopore size. The novel CoNi₂S₄/carbon nanotube composite could be a promising electrode material for high-performance supercapacitors.

Acknowledgements

This work is supported by the Tongji University Research Foundation (1380219039).

References

1. D. P. Dubal, P. Gomez-Romero, B. R. Sankapal and R. Holze, Nano Energy, 2015, 11, 377.
2. L. Yu, H. Wu, T. Wu and C. Yuan, RSC Adv., 2013, 3, 23709.
3. H. Chen, J. Jiang, L. Zhang, D. Xia, Y. Zhao, D. Guo, T. Qi and H. Wan, J. Power Sources, 2014, 254, 249.
4. F. Cao, G. X. Pan, X. H. Xia, P. S. Tang and H. F. Chen, J. Power Sources, 2014, 264, 161.
5. Y. Wang, A. Pan, Q. Zhu, Z. Nie, Y. Zhang, Y. Tang, S. Liang and G. Cao, J. Power Sources, 2014, 272, 107.
6. X. Feng, J. Zhou, L. Wang, Y. Li, Z. Huang, S. Chen, Y. Ma, L. Wang and X. Yan, New J. Chem., 2015, 39, 4026.
7. L. Chen, Y. Zhou, H. Dai, T. Yu, J. Liu and Z. Zou, Nano Energy, 2015, 11, 697.
8. I. Shakir, Z. Ali, J. Bae, J. Park and D. J. Kang, Nanoscale, 2014, 6, 4125.
9. F. Cai, Y. Kang, H. Chen, M. Chen and Q. Li, J. Phys. Chem. A, 2014, 2, 11509.
10. I. Shakir, Electrochim. Acta, 2014, 132, 490.
11. X. Wang, X. Han, M. Lim, N. Singh, C. L. Gan, M. Jan and P. S. Lee, J. Phys. Chem. C, 2012, 116, 12448.
12. E. Mitchell, A. Jimenez, R. Gupta, B. K. Gupta, K. Ramasamy, M. Shahabuddin and S. R. Mishra, New J. Chem., 2015, 39, 2181.
13. Y. Yang, Y. Liang, Y. Zhang, Z. Zhang, Z. Li and Z. Hu, New J. Chem., 2015, 39, 4035.
14. H. Chen, J. Jiang, L. Zhang, H. Wan, T. Qi and D. Xia, Nanoscale, 2013, 5, 8879.
15. J. Xiao, L. Wan, S. Yang, F. Xiao and S. Wang, Nano Lett., 2014, 14, 831.
16. L. Yu, L. Zhang, H. B. Wu and X. W. Lou, Angew. Chem., 2014, 53, 3711.
17. Y. Zhang, M. Ma, J. Yang, C. Sun, H. Su, W. Huang and X. Dong, Nanoscale, 2014, 6, 9824.
18. W. Du, Z. Zhu, Y. Wang, J. Liu, W. Yang, X. Qian and H. Pang, RSC Adv., 2014, 4, 6998.
19. S. Peng, L. Li, C. Li, H. Tan, R. Cai, H. Yu, S. Mhaisalkar, M. Srinivasan, S. Ramakrishna and Q. Yan, Chem. Commun., 2013, 49, 10178.
20. J. Tang, J. Shen, N. Li and M. Ye, Ceram. Int., 2015, 41, 6203.
21. M. Sun, J. Tie, G. Cheng, T. Lin, S. Peng, F. Deng, F. Ye and L. Yu, J. Mater. Chem. A, 2015, 3, 1730.
22. S. Liu, Y. Liu, W. Song, J. Song, C. Wang, G. Shao and X. Qin, J. Solid State Electrochem., 2015, 19, 1321.
23. H. Cheng, A. D. Su, S. Li, S. T. Nguyen, L. Lu, C. Y. H. Lim and H. M. Duong, Chem. Phys. Lett., 2014, 601, 168.
24. Z. Tang, C. Tang and H. Gong, Adv. Funct. Mater., 2012, 22, 1272.
25. X. Wang, X. Han, M. Lim, N. Singh, C. L. Gan, M. Jan and P. S. Lee, J. Phys. Chem. C, 2012, 116, 12448.
26. M. S. Dresselhaus, G. Dresselhaus, R. Saito and A. Jorio, Phys. Rep., 2005, 409, 47.
27. T. Zhu, Z. Wang, S. Ding, J. S. Chen and X. W. Lou, RSC Adv., 2011, 1, 397.
28. W. Du, Z. Wang, Z. Zhu, S. Hu, X. Zhu, Y. Shi, H. Pang and X. Qian, J. Mater. Chem. A, 2014, 2, 9613.
29. H. Wan, J. Jiang, J. Yu, K. Xu, L. Miao, L. Zhang, H. Chen and Y. Ruan, CrystEngComm, 2013, 15, 7649.
30. J. Yang, W. Guo, D. Li, Q. Qin, J. Zhang, C. Wei, H. Fan, L. Wu and W. Zheng, Electrochim. Acta, 2014, 144, 16.
31. Y. Zhu, Z. Wu, M. Jing, X. Yang, W. Song and X. Ji, J. Power Sources, 2015, 273, 584.
32. L. Shen, L. Yu, H. B. Wu, X. Y. Yu, X. Zhang and X. W. Lou, Nat. Commun., 2015, 6, 6694.
33. Y. M. Chen, Z. Li and X. W. Lou, Angew. Chem., Int. Ed., 2015, 54, 10521.
34. P. Liu, Z. Hu, Y. Liu, M. Yao and Q. Zhang, J. Alloys Compd., 2015, 622, 805.
35. D. Du, Z. Hu, Y. Liu, Y. Deng and J. Liu, J. Alloys Compd., 2014, 589, 82.

---