In situ growth of metallic 1T-WS₂ nanoislands on single-walled carbon nanotube films for improved electrochemical performance

Abstract

Layered tungsten disulfide (WS₂) is a potential electrode material for electric double layer capacitance (EDLC) and hydrogen evolution reaction (HER). However, the electrochemical performance of WS₂ has been hindered by the semiconducting nature and poor active sites. Herein, we have demonstrated a bottom-up hydrothermal approach to fabricate metallic 1T-WS₂ nanoislands in situ grown on flexible single-walled carbon nanotube nonwovens (1T-WS₂@SWCNT). The robust hybrids with a tight interface possess nanoscopic few-layered WS₂ pieces with an abundance of exposed sites, along with a unique woven-architecture originating from the high conductive carbon nanotube network. The in situ-growing enhanced interface between metallic WS₂ nanoislands and SWCNTs provides a relatively strong electrical coupling integrity, which facilitates charge transfer during electrochemical reactions. The merits of rich active sites, excellent conductivity and well bonding-interactions are significantly beneficial to improve the electrochemical performance. Particularly, in contrast to the pure material, the as-obtained hybrids are found to exhibit higher EDLC capacity (226 mF cm⁻²), almost 646-fold higher than pure 1T-WS₂, smaller Tafel slope (57 mV per decade) and lower HER overpotential (∼25 mV) than any WS₂-based materials reported so far.

---

1. Introduction

Tungsten disulfide (WS₂), as one of the most studied and layered transition metal dichalcogenides (LTMDs), generally consists of hexagonal S–W–S three atomic layers that are combined with covalent bonds in plane.^1,2 Similar to the graphene layers in graphite, the individual S–W–S layers are weakly bound by van der Waals interactions out of plane. Recently, with wide applications of excellent graphene materials, WS₂ and other TMDs are also drawing tremendous attention owing to their excellent electrochemical performance and potential applications in terms of energy storage and generation.^3–26 In recent years, the layered WS₂ is extensively explored as the electrode materials in hydrogen evolution reaction (HER),^11,12,23,25 Li-ion battery,^27,28 electric double layer capacitance (EDLC),^8,34,36,37 and so on. However, WS₂ also shows some disadvantages that urgently need to be resolved, such as the lack of good conductivity, unsatisfactory shape or size, and insufficient edges. To conquer these challenges, some high-conducting substrates have been used to mix with WS₂ for improving its electrochemical ability, especially carbon-based materials including carbon nanotubes,^29,30 graphene^{10,24,26,30,31} and carbon cloth.³² In fact, it is still a big challenge to realize good interface between WS₂ and substrates due to crude contact and low yield. On the other hand, the electrochemical activity for WS₂ nanomaterials is largely depending on their exposed active edge sites.²⁶ In this case, various exfoliation methods have been employed for preparing mono- or few-layers WS₂ with rich active sites.^{11,24,25,33–35} More importantly, WS₂ layer has two distinct symmetries depending on the arrangement of S atoms, i.e. semiconducting 2H phase and metallic 1T phase. It is naturally to believe that metallic 1T-WS₂ will pose more exposed active sites and much better conductivity than the semiconducting 2H-WS₂, which will result in better electrochemical performance. Recently, many efforts have been drawn to metallic 1T-WS₂.^33–35 Despite of some significant achievements have obtained in the preparation of 1T-WS₂, most of these products are less stable or require dangerous and complex synthesis process. Therefore, it is still highly desirable to produce stable 1T-WS₂ tightly adhered onto high conducting substrates for applications in electrochemical and other related fields.

Herein, we have developed a bottom-up solvothermal route to facile assemble numerous metallic 1T-WS₂ nanoislands with size less than 10 nm onto single-walled carbon nanotube films (1T-WS₂@SWCNT). Interestingly, the 1T-WS₂ pieces can gain abundant exposed active catalytic sites and few-layered property when the SWCNT was used as a substrate for the in situ formation and growth processes. These advantages from the intense electrical coupling between 1T-WS₂ and SWCNT, to a large extent, contribute to the greatly enhancing electrochemical performances in contrast with pure materials.

2. Experimental

2.1 Synthesis of SWCNT

SWCNT non-woven material was prepared using a modified floating CVD technique reported by our previous work.³⁸ In the process, a special designed quartz tube with diameter of 50 mm together with an inner tube with diameter of 10 mm was placed into furnace (MTK Co. Ltd) and extra heating wire was employed to heat the catalyst (1 : 16 mole ratio S : ferrocene). To better understand, we showed the configuration in Scheme S1.†

SWCNT film has been synthesized successfully with the following heating process: main heating zone was heated to 1000 °C with a ramping rate 30 °C min⁻¹ and a suitable argon airflow 200 sccm introduced as protective atmosphere. Then, the temperature was increased to 1100 °C with rate 10 °C min⁻¹, and meanwhile, furnace 1 was heated to 90 °C to heat catalysts. When the temperature was reached to 1100 °C, the argon flow was increased to 1000 sccm and mixed with extra methane with rate 3 sccm. Two hours later, the heating was stopped and cooled to room temperature naturally. The as prepared SWCNT film with successive textile structure was obtained.

To remove amorphous carbon and remaining iron, the as-grown SWCNT films were purified by oxidation in air (∼400 °C) for 24 h and then immersed into concentrated hydrochloric acid for one week to remove metal oxides, along with DI water cleaning for several times.

2.2 Synthesis of metallic 1T-WS₂

In a typical experiment, 1 mmol of tungsten chloride (WCl₆) and 10.1 mmol of thioacetamide (CH₃CSNH₂, TAA) were dissolved in 30 mL of DMF and stirred 1 h at room temperature. Then the solution was transferred to a 40 mL Teflon-lined stainless steel autoclave, heated up to 220 °C, and kept for 24 h. After cooling naturally, the product was washed with absolute alcohol several times, and dried in vacuum oven at 60 °C for 24 h.

2.3 Synthesis of 1T-WS₂@SWCNT

1T-WS₂@SWCNT hybrid films were synthesized by the same solvothermal reaction conditions as that for 1T-WS₂ sheets. After stirring, a SWCNT film about 1 cm × 0.5 cm was added to the mixture of tungsten chloride and thioacetamide, and kept still for 10 minutes. Then the solution was transferred to a 40 mL Teflon-lined stainless steel autoclave, heated up to 220 °C, and kept for 24 h. After cooling naturally, the black film was washed with absolute alcohol several times under the sonication at power of 40%, and dried in vacuum oven at 60 °C for 24 h.

2.4 Electrochemical measurements

Electrochemical measurements were carried out using a 3-electrode cell and a 0.5 M sulfuric acid (H₂SO₄) electrolyte solution. Platinum net (Sigma Aldrich) and saturated calomel (Ag/AgCl) electrode (Pine Research Instrument) have been used as counter electrode and reference respectively. The reference electrode was calibrated with respect to reversible hydrogen electrode (RHE) using platinum sheets as working and counter electrode. In 0.5 M H₂SO₄: E_RHE = E_Ag/AgCl + 0.2046 V. Normally, the 1T-WS₂@SWCNT hybrid film was cut into a small piece with a glassy carbon electrode (GCE, 3 mm in diameter), and then was scattered on a glassy carbon electrodes of 3 mm in diameter (loading 400 ± 5 μg cm⁻²) and protected by 5 μL Nafion solution. The cyclic voltammetry (CV) was measured with scan rates of 20, 40, 60, 80, 100, 120, 140, 160, and 180 mV s⁻¹, respectively, in the potential ranging from 0.10 V to 0.30 V (vs. RHE) to estimate the electrochemical double-layer capacitances. And then the linear sweep voltammetry (LSV) experiments were performed from 0.05 to −0.30 V (vs. RHE) with 2 mV s⁻¹ scan rate using a CHI660E electrochemical workstation and electrodes were cycled at least 500 cycles prior to any measurements. The Nyquist plots were measured at frequencies ranging from 100 KHz to 0.1 Hz at the overpotential −150 mV (vs. RHE) with an amplitude voltage of 5 mV. Cyclic voltammetry (CV) experiments were performed for 600 cycles and then 3000 cycles with a scan rate of 50 mV s⁻¹ in the potential ranging from −0.25 V to 0.05 V (vs. RHE) to investigate the cycling durability of 1T-WS₂@SWCNT film.

2.5 Materials characterizations

Our samples were characterized by X-ray powder diffraction (XRD) by a Philips X'Pert Pro Super diffractometer equipped with Cu Kα radiation (λ = 1.54178 Å). Field emission scanning electron microscopy (FE-SEM) images were taken via a JEOL JSM-6700F SEM. The morphologies were obtained from JEM-2100F field emission electron microscopy (TEM/HRTEM) with an acceleration voltage of 200 kV and high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) on a JEOL JEM-ARF200F TEM/STEM (200 keV) with a spherical aberration corrector. Zeta potential was tested by Nano Zetasizer 90 from Malvern Instruments. Raman spectra were detected by a Renishaw RM3000 Micro-Raman system with a 532 nm Ar laser. W L₃-edge XAFS spectra were collected at the Shanghai Synchrotron Radiation Facility (BL14W1, SSRF) and the Beijing Synchrotron Radiation Facility (1W1B, BSRF).

3. Results and discussion

The growth of 1T-WS₂@SWCNT hybrid is proposed and the scheme is illustrated in Fig. 1a. In order to obtain homogeneously dispersed precursor solutions, the WCl₆ as W source and TAA as S source and reductant have been selected. The process of growth is occurred in a sealed autoclave system via a solvothermal method, similar to our previous work.³⁹ Accordingly, ammonium ions from the hydrolysis of TAA can act as the electron donor to distort the W–W band, resulting in the stable metallic WS₂, which will be proved by the following characterizations. More detailed mechanism can also be found in our previous report.³⁹

Fig. 1 (a) Schematic synthesis process of 1T-WS₂@SWCNT hybrid via a solvothermal method. (b) TEM image of the hybrid's hierarchical nanostructure and (c) HRTEM image of 1T-WS₂ small pieces with standing edges on SWCNT. (d) A typical HADDF-STEM image of 1T-WS₂ on the SWCNT shows bonds of W atoms reveal an obvious zigzag chain superlattice. (e) The Raman spectra of the SWCNT (black curve) and 1T-WS₂@SWCNT hybrid (red curve). The inset shows the zoom-in Raman spectrum at 80–550 cm⁻¹ range.

In order to directly observe the microstructures of our as-obtained materials, the transmission electron microscopy (TEM) was employed. Fig. 1b shows that numerous small islands with size less than 10 nm homogeneously dispersed and adhered on SWCNT bundles. The HRTEM image in Fig. 1c reveals that the islands exhibit loosely stacked layers with space distance range of 9.0–9.4 Å. The expanded spacing distance is also consistent with the XRD result (see Fig. S3†), indicating the hybrid of few-layered WS₂ pieces with SWCNT film. It is worth noting that this hybrid exhibits the uniform dispersion of WS₂ nanoislands in contrast with bare WS₂ nanosheets, which were largely aggregated to form flower-like clusters (see Fig. S1a and b†).

High-angle annular dark field (HAADF) imaging mode in an aberration-corrected scanning transmission electron microscope (STEM) was used to observe the microstructure of the as-grown hybrid. It can be clearly seen from Fig. 1d that the WS₂ islands are very small with an abundance of exposed edges, active surface, and numerous vacancies. Notably, the nearest distance of W–W for WS₂ islands is about 2.7 Å with unique zigzag-chain superlattice, strongly indicating the metallic 1T phase in the WS₂ islands grown on SWCNT. In combination with STEM intensity profile and X-ray absorption fine structure analysis (see Fig. S4, S5, and Table S1†), it can be seen that the hybrid is composed of metallic 1T-WS₂ nanoislands and flexible SWCNT. More importantly, 1T-WS₂ nanoislands with an abundance of exposed sites and metallic character can be uniformly dispersed and adhered onto the woven-like SWCNT films, forming unique hybrid nanoarchitecture that is very different with pure aggregated WS₂ sheets.

In order to further identify the structure and composition of the hybrid, the Raman spectroscopy was carried out using a 532 nm excitation laser in ambient. In contrast to the bare SWCNT, the spectrum of 1T-WS₂@SWCNT hybrid in Fig. 1e shows several characteristic, including the carbon-like G peaks at around 1581 cm⁻¹ and D peak at around 1340 cm⁻¹ which originate from SWCNT.³⁸ Interestingly, a clear red shift (∼8 cm⁻¹) of G peak was observed. Taking the small-curvature of interface between SWCNT and 1T-WS₂ nanoislands into account, the electromagnetic field enhancement effect can be ignored, and the large shift could be considered from the simultaneously intensive electronic and vibrational coupling at their interfaces.⁴⁰ The insert of Fig. 1e reveal that richer peaks were observed for the hybrid as compared to bare SWCNT, which can be assigned to RBM peaks and metallic 1T-WS₂ peaks.³⁹ Notably, there are no obvious scattering peaks around 350 cm⁻¹ and 420 cm⁻¹ related to E¹_2g (in-plane) and A_1g (out-of-plane) of 2H-WS₂, further implying the WS₂ nanoislands among the hybrid is mostly metallic 1T phase, not semiconducting 2H phase. Furthermore, we tested the Raman spectra of fresh and aging 1T-WS₂@SWCNT hybrid (see Fig. S2†). These two samples showed the similar Raman signals, indicating the phase stability and unchanged components after ten-month placement in ambient.

The chemical components and surface states of 1T-WS₂@SWCNT hybrid were proved by using X-ray photoelectron spectroscopy (XPS). As shown in Fig. 2a, the C1s can be deconvoluted into two peaks. The main peak at 284.5 eV is attributed to sp² hybridized graphite-like carbon and the peak around 285.9 eV is assigned to the carbon–oxygen, which is similar to the values for acid-treated SWCNTs reported previously. The high-resolution spectra for W4f, shown in Fig. 2b, display several peaks. There are two main peaks occurred at 31.9 eV and 34.2 eV (red curves), respectively (Fig. 2b), which is consistent to the Li-intercalated WS₂.³³ This to some extent indicates the presence of metallic 1T-WS₂ at a large content. However, another two peaks at about 33.2 eV and 35.4 eV (green curves) also were observed, showing the simultaneously existence of partial semiconducting phase. Moreover, similar results of S2p can also be observed (Fig. 2c). Furthermore, energy dispersive X-ray spectrum (EDS) was utilized to demonstrate the elemental distribution, which showed the co-existence of W, S, and C (Fig. 2d). The Cu signal here was from the copper substrate.

Fig. 2 (a–c) High-resolution XPS C1s, W4f, and S2p spectra, respectively. (d) TEM-EDS of the 1T-WS₂@SWCNT hybrid with general elements ratio.

To investigate the practicability in fundamentally and technologically important applications, we carried out the capacitance measurements on the high-quality 1T-WS₂@SWNT hybrids. The EDLC values were estimated by method similar to previous reports.³⁶ Fig. 3a shows the current response window used for the CV (0.1–0.3 V vs. RHE) at different scan rates (20–180 mV s⁻¹), indicating the double-layers capacitance behavior. It is observing that the CV curves of 1T-WS₂@SWCNT have a slightly distorted rectangular shape, which further implies that the classic EDLC behavior dominated this process. The corresponding Δj/2_{0.2 V vs. RHE}-scan rate curve was shown in Fig. 3b, and the calculated C_dl was about 226 mF cm⁻², significantly larger than previously reported WS₂-related materials (see Table S2†). The good performance can be contributed to the large surface area from our hierarchical structure in electrolyte, agreeing with the above TEM observations. For further comparison, the EDLC value of pure 1T-WS₂ was measured. A value of ∼0.35 mF cm⁻² was obtained, extremely smaller than that of the hybrid (see Fig. S6†).

Fig. 3 (a) Voltammograms of the 1T-WS₂@SWCNT hybrid at various scan rates (20–180 mV s⁻¹). (b) The calculated C_dl from (a). (c) The extended Stern electric double-layer model. (d) The zeta potential distribution of 1T-WS₂ in water dispersion.

To further explain the high EDLC value, we tested the zeta potential of our sample (Fig. 3d). The result showed very low average potential (−38.1 mV), indicating 1T-WS₂ is highly negatively charged, which would simultaneously contributes to the diffusion of protons to electrodes together with external power. This leads to formation of effective and narrower Stern layers through its electrostatic force with the negatives, consequently the higher EDLC obtained (see Fig. 3c).

Considering high EDLC, which to some extent demonstrated the abundant exposure of electrochemical active sites, we also measured electrocatalytic HER performance of 1T-WS₂@SWCNT hybrid. The hybrid were deposited on a glassy carbon electrode and tested by using the standard three-electrode electrochemical allocation in 0.5 M H₂SO₄ electrolyte. The polarization curves (not iR corrected), displaying the normalized current density versus potential for the 1T-WS₂, SWCNT and 1T-WS₂@SWCNT hybrid along with Pt/C and bulk WS₂ powder samples, as reference purposes, were shown in Fig. 4a. It can be seen that the pure 1T-WS₂ and SWCNT electrodes have very lower onset of HER voltages ranging from 150 mV to 170 mV, above which the current density increases rapidly, and the bulk WS₂ has much higher overpotential as reported before.⁴¹ Notably, the hybrids exhibit excellent HER behavior, the overpotential is reduced to 25–30 mV in the case of 1T-WS₂@SWCNT hybrid, which is better than any WS₂-based catalysts ever reported thus far (see Table S3†). These results are possibly attributed to the electrical coupling between 1T-WS₂ with abundant exposed active sites and the SWCNT substrate possessing high conductivity due to its special 1D structure, which favors the transfer of electrons and enhance the surface activity. Besides, the 1T-WS₂@SWCNT hybrids also shows notably hydrogen evolution (∼10 mA cm⁻²) at the very lower overpotential ∼108 mV vs. RHE, which is much lower than previous reported values for WS₂ or MoS₂ electrodes.

Fig. 4 (a) Polarization curves and (b) corresponding Tafel plots recorded on glassy carbon electrodes with a catalyst loading of 400 μg cm⁻², and (c) alternating current impedance spectra of SWCNT (red curve), 1T-WS₂ (black curve), and 1T-WS₂/SWCNT hybrid (blue curve). (d) Polarization curves revealing that a little bit degradation of HER activity is observed after 3000 CV cycles. (e) The scheme of possible HER mechanism for 1T-WS₂/SWCNT catalyst.

To further illustrate the enhanced HER property, we also showed the corresponding Tafel plots for the hybrid and Pt/C recorded on glassy carbon electrodes, as shown in Fig. 4b. The lower Tafel slope of the hybrid (∼57 mV per decade) further suggest excellent HER activity, which is possibly duo to a large number of exposed active sites of few-layered 1T-WS₂. Exchange current density was also gained by the extrapolation of Tafel curve (Fig. 4b). The 1T-WS₂@SWCNT hybrid displays a large exchange current density of 225 μA cm⁻², indicating the lower hinder from kinetics and better reversibility of electrode reaction.

In addition, we used the electrochemical workstation to test the EIS (at −0.15 V vs. RHE) to investigate the electrical capacity under the catalytic HER conditions. EIS Nyquist plots (Fig. 4c) for the 1T-WS₂@SWCNT hybrid revealed greatly low charge-transfer resistances (R_ct ∼ 4 Ω). Similarly SWCNT samples (Fig. 4c) also showed low charge transfer resistance (R_ct ∼ 3 Ω), but only 1T-WS₂ has a large one. The small R_ct obtained from hybrid shows the importance of in situ synthesis on conductive substrates, which promotes the valid transport of electrons and decreased the rate of recombination. The EIS results suggested that 1T-WS₂@SWCNT hybrid was dramatically active catalyst that exhibit facile process towards HER. Electrochemical stability, as one of the most important properties for practical applications, is essential for any electrode materials, so we investigated the durability of 1T-WS₂@SWCNT hybrid under the condition as electrocatalytic operation at the range −0.25 to +0.05 V vs. RHE. In Fig. 4d, it is indicated that overpotential just has a negligible increasing trend for our sample, and the current density is always obtaining the ideal value eventually after 3000 cycles, which indicates the stable nature of hybrid used in this work. Fig. 4e shows the mechanism of electrochemical HER. The electrons from electrode can efficiently transfer to catalyst by SWCNT, and the H⁺ absorbed onto surface would be reduced and then released in the form of H₂ bubbles.

To prove the stability of our sample and possible Pt deposition, we have tested the XPS of 1T-WS₂@SWCNT hybrids after HER test for over 1000 cycles in the range of −0.30 V to 0.1 V vs. RHE. As shown in Fig. 5a, the C1s was deconvoluted into two peaks. The two main peaks are consistent to the values in Fig. 2a. Also, the high-resolution spectra for W4f and S2p were shown in Fig. 5b and c. This two also stay the same with fresh 1T-WS₂@SWCNT hybrids. These resluts above to some extent indicate the stability of 1T-WS₂@SWCNT hybrids during long time testing. What's more, Pt4f XPS was detected and displayed in Fig. 5d. There is no obvious signals observed, revealing the absence of Pt, and the reasonable HER result above.

Fig. 5 (a–d) High-resolution XPS spectra of C1s, W4f, S2p, and Pt4f, respectively for 1T-WS₂@SWCNT hybrids after HER test for over 1000 cycles in the range of −0.30 V to 0.1 V vs. RHE.

To view these results more intuitively, herein we collate some consequences and list them into following two charts. The value of EDLC is almost due to the contact area and the distance between electrode materials and charges around with solvent molecules. As comparison, the 1T-WS₂@SWCNT hybrids show higher EDLC value than any other MS₂ (M = W or Mo)-based materials reported before (Fig. 6a). It is noteworthy that owing to the hierarchical hybrid (high contact area) and highly negative electrical property (short distance), the capacitance from Stern layer will be larger. Though the capacitance of diffusive layer is almost unchanged, the total capacitance from these two capacitors in series will be greatly improved.

Fig. 6 (a) The different EDLC values in several recent literatures are listed for comparison. (b) The HER performance for various MS₂ (M = W or Mo)-based materials.

We also compare HER performance of the hybrid with previous reports (Fig. 6b). The overpotential as low as ∼25 mV and small potential of ∼108 mV for 10 mA cm⁻² were obtained, which are the highest HER behavior of MS₂ (M = W or Mo)-based catalyst reported so far, which are mostly due to the tight contact between metallic 1T-WS₂ nanoislands and SWCNT and high active area from our hierarchical structure to promote electrons transfer to surface of catalyst.

4. Conclusions

In summary, we presented a facile, effective and large-scale solvothermal method for in situ synthesis of 1T-WS₂@SWCNT hybrids. Our characterizations revealed that numerous nanoscopic few-layered WS₂ pieces with an abundance of exposed sites and metallic character can be uniformly adhered onto the highly conducting woven-like SWNT films, forming unique hybrid nanoarchitecture in strong contrast to pure aggregated WS₂ sheets that were prepared in solution without SWCNT. Benefit from such unique structure character, the as-obtained hybrids exhibited superior electrocatalytic activity in the hydrogen evolution reaction with very positive overpotentials of 25–30 mV and a small Tafel slope of ∼57 mV per decade, which shows competitive compared with most of previous reported WS₂ materials. Furthermore, this innovative hybrid opens up a new window for 2D material in the fields of specific EDLC, HER, and other related applications.

Acknowledgements

This work was supported by the National Basic Research Program of China (2014CB848900), the National Natural Science Foundation of China (U1232131, U1532112, 11375198, 11574280), CUSF (WK2310000053), the Unsers with Potential project (2015HSC-UP020) and Key Lab of Advanced Energy Materials Chemistry (Ministry of Education) Naikai University. L. S. thanks the recruitment program of global experts, the CAS Hundred Talent Program. We also thank the Shanghai synchrotron Radiation Facility (14W1), the Beijing Synchrotron Radiation Facility (1W1B and soft-x-ray endstation) and the Hefei Synchrotron Radiation Facility (MCD and Photoemission endstation) for helps in SR-characterizations.

Notes and references

1. D. J. L. Dattatray, S. R. Chandra, C. Disha and R. Satyajit, Can. Chem. Trans., 2015, 3, 118–157.
2. S. F. Matthew and S. Jin, Energy Environ. Sci., 2014, 7, 3519–3542.
3. X. Fang, C. Hua, C. Wu, X. Wang, L. Shen, Q. Kong, J. Wang, Y. Hu, Z. Wang and L. Chen, Chem.–Eur. J., 2013, 19, 5694–5700.
4. X. Zou and Y. Zhang, Chem. Soc. Rev., 2015, 44, 5148–5180.
5. Z. Wu, B. Fang, B. Arman, S. Aokui, P. W. David and D. Wang, Appl. Catal., B, 2012, 125, 59–66.
6. K. Jakob, Z. Chen, N. R. Benjamin and F. J. Thomas, Nat. Mater., 2012, 11, 963–969.
7. J. Xie, H. Zhang, S. Li, R. Wang, X. Sun, M. Zhou, J. Zhou, X. Lou and Y. Xie, Adv. Mater., 2013, 25, 5807–5813.
8. A. L. Mark, S. D. Andrew, F. Meng, F. Audrey, L. Li and S. Jin, J. Am. Chem. Soc., 2013, 135, 10274–10277.
9. T. Chen, Y. Chang, C. Hsu, K. Wei, C. Chiang and L. Li, Int. J. Hydrogen Energy, 2013, 38, 12302–12309.
10. Y. Yan, X. Ge, Z. Liu, J. Wang, J. Lee and X. Wang, Nanoscale, 2013, 5, 7768–7771.
11. L. Cheng, W. Huang, Q. Gong, C. Liu, Z. Liu, Y. Li and H. Dai, Angew. Chem., Int. Ed., 2014, 53, 7860–7863.
12. M. Rachel, H. Trevor and A. W. Colin, Chem. Mater., 2014, 26, 3986–3992.
13. N. Zhang, S. Gan, T. Wu, W. Ma, D. Han and L. Niu, ACS Appl. Mater. Interfaces, 2015, 7, 12193–12202.
14. Y. Yang, H. Fei, G. Ruan, Y. Li and M. James, Adv. Funct. Mater., 2015, 25, 6199–6204.
15. F. J. Thomas, P. J. Kristina, B. Jacob, H. N. Jane, H. Sebastian and C. Ib, Science, 2007, 317, 100–102.
16. L. G. Alan, A. Vincent, J. Bruno, D. T. Phong, G. Nicolas, M. Romain, F. Aziz, P. Serge and F. Marc, Science, 2009, 326, 1384–1387.
17. I. K. Hemamala, J. Z. Christopher and R. L. Jeffrey, Nature, 2010, 464, 1329–1333.
18. J. Xie, J. Zhang, S. Li, F. Grote, X. Zhang, H. Zhang, R. Wang, Y. Lei, B. Pan and Y. Xie, J. Am. Chem. Soc., 2013, 135, 17881–17888.
19. D. Kong, H. Wang, J. C. Judy, P. Mauro, J. K. Kristie, J. Yao and Y. Cui, Nano Lett., 2013, 13, 1341–1347.
20. H. Zhu, M. Du, M. Zhang, M. Zou, T. Yang, S. Wang, J. Yao and B. Guo, Chem. Commun., 2014, 50, 15435–15438.
21. M. Gao, J. Liang, Y. Zheng, Y. Xu, J. Jiang, Q. Gao, J. Li and S. Yu, Nat. Commun., 2014, 6, 5982–5988.
22. W. K. Ho, J. C. Yu, J. Lin, J. Yu and P. Li, Langmuir, 2004, 20, 5865–5869.
23. L. C. Charina, J. Feng, Y. Li, W. Justin, Z. Alla, T. Reshef and H. Dai, Nano Res., 2013, 6, 921–928.
24. Y. Jieun, V. Damien, J. A. Seong, D. Kang, Y. K. Ah, C. Manish and S. S. Hyeon, Angew. Chem., Int. Ed., 2013, 52, 1–5.
25. Y. Zhang, J. Shi, G. Han, M. Li, Q. Ji, D. Ma, Y. Zhang, C. Li, X. Lang, Y. Zhang and Z. Liu, Nano Res., 2015, 8, 2881–2890.
26. A. S. Tofik, F. Wang, Z. Cheng, X. Zhan, Z. Wang, K. Liu, S. Muhammad, L. Sun and J. He, Nanoscale, 2015, 7, 14760–14765.
27. R. Bhandavat, L. David and G. Singh, J. Phys. Chem. Lett., 2012, 3, 1523–1530.
28. Y. Du, X. Zhu, L. Si, Y. Li, X. Zhou and J. Bao, J. Phys. Chem. C, 2015, 119, 15874–15881.
29. Y. Liu, W. Wang, H. Huang, L. Gu, Y. Wang and X. Peng, Chem. Commun., 2014, 50, 4485–4488.
30. R. Chen, T. Zhao, W. Wu, F. Wu, L. Li, J. Qian, R. Xu, H. M. A. Hassan, A. S. Al-Bogami, D. A. El-Hady, J. Lu and A. Khalil, Nano Lett., 2014, 14, 5899–5904.
31. R. Satyajit and S. R. Chandra, ACS Appl. Mater. Interfaces, 2013, 5, 11427–11433.
32. Y. Yan, Y. Xia, N. Li, Z. Xu, F. Adrian and X. Wang, J. Mater. Chem. A, 2015, 3, 131–135.
33. V. Damien, Y. Hisato, J. Li, S. Rafael, C. B. A. Diego, F. Takeshi, M. Chen, A. Tewodros, B. S. Vivek, E. Goki and C. Manish, Nat. Mater., 2013, 12, 850–855.
34. A. L. Mark, S. D. Andrew, R. E. Caroline, F. Meng, F. Audrey, J. H. Robert and S. Jin, Energy Environ. Sci., 2014, 7, 2608–2613.
35. M. Benoit, H. Veronika, L. Kristine and A. O. Geoffrey, J. Am. Chem. Soc., 2014, 136, 14121–14127.
36. F. Wang, J. Li, F. Wang, A. S. Tofik, Z. Cheng, Z. Wang, K. Xu, X. Zhan, Q. Wang, Y. Huang, C. Jiang and J. He, Adv. Funct. Mater., 2015, 25, 6077–6083.
37. D. Li, U. Maiti, J. Lim, D. Choi, W. Lee, Y. Oh, G. Lee and S. Kim, Nano Lett., 2014, 14, 1228–1233.
38. L. Song, L. Ci, L. Li, Z. Zhou, X. Yan, D. Liu, H. Yuan, Y. Gao, J. Wang, L. Liu, X. Zhao, Z. Zhang, X. Dou, W. Zhou, G. Wang, C. Wang and S. Xie, Adv. Mater., 2004, 16, 1529–1534.
39. Q. Liu, X. Li, Z. Xiao, Y. Zhou, H. Chen, K. Adnan, T. Xiang, J. Xu, W. Chu, X. Wu, J. Yang, C. Wang, Y. Xiong, C. Jin, M. A. Pulickel and L. Song, Adv. Mater., 2015, 27, 4837–4844.
40. E. Barros and M. Dresselhaus, Phys. Rev. B: Condens. Matter Mater. Phys., 2014, 90, 035443.
41. H. Shi, G. Li, H. Sun, T. An, H. Zhao and P. Wong, Appl. Catal., B, 2014, 158–159, 301–307.

---

Footnotes

† Electronic supplementary information (ESI) available: SEM, TEM, Raman, XRD, CV and EXAFS data. See DOI: 10.1039/c6ra19680b

‡ The authors contributed equally to this work.

---