Uniformly deposited Pt nanoparticles onto crosslinked ionic liquids wrapped carbon nanotubes for methanol electrooxidation

Abstract

A type of skin-core structured hybrid with a multi-walled carbon nanotube (MWCNT) center was synthesized by in situ free-radical polymerization of vinyl-benzyl ionic liquid and divinylbenzene on the outer surface of MWCNTs. Pt nanoparticles were uniformly deposited on the hybrid using a layer of crosslinked ionic liquid non covalently coated on MWCNTs. The physical characteristics were identified by X-ray diffraction and transmission electron microscopy. The Pt nanoparticles with a small diameter and narrow size distribution were uniformly deposited on the surface of the skin-core structured MWCNTs with an average diameter of 3.3 nm. Cyclic voltammetry results showed that the as-prepared Pt-deposited MWCNT hybrids had a high electrochemical surface area (67.5 m² g⁻¹) and large current density (425.0 mA mg⁻¹) for methanol electrooxidation.

---

1. Introduction

With increasing energy demand and shortage of fossil fuel resources, fuel cells are expected to become useful as high efficiency and environmentally friendly energy conversion devices.^1,2 Direct methanol fuel cells (DMFCs), which constitute one type of proton-exchange membrane fuel cells (PEMFCs), are promising candidates for the storage and transportation of methanol and are safe at low operating temperature.^3–5 Methanol electrooxidation is a simple reaction wherein the C–C bond is not broken; thus, the anodic reaction occurs at a high rate with high energy efficiency in DMFCs.¹ At present, Pt and its alloy are still the most practical methanol electrooxidation reaction catalysts in DMFCs.⁶ Many studies have been conducted to disperse the catalyst nanoparticles on the support, expecting an increase in catalytic activity and stability.^7–14

Carbon nanotubes (CNTs) possess unique electrical, structural, and mechanical properties as ideal building blocks in hybrid materials, which have drawn increasing interest in several applications, particularly in electrochemical sensors.¹⁵ However, the application of CNTs has been limited over the years because of their inert nature (i.e., CNTs have no modification tendency to agglomerate) and the weak interaction between CNTs and hybrids.¹⁶ General approaches are commonly used to functionalize CNTs through chemical oxidation in a mixture of HNO₃/H₂SO₄ solution, thereby introducing defects on the sidewalls and tube tips. However, the defects show a negative impact on their electronic conductivity as well as corrosion resistance.^17–19 Therefore, non-covalent functionalization of CNTs by surfactants, polymers, or organic materials was investigated to introduce binding sites for catalysts and minimal damage to the structure of CNTs such as poly(diallydimethylammonium chloride),²⁰ propyleneglycol alginate sodium sulfate,²¹ poly(ethyleneimine),²² poly(allyamine hydrochloride),²³ and poly(benzimidazole).²⁴

Since Fukushima et al. found that ionic liquids (ILs) can effectively disperse CNTs, extensive research has been conducted on electrocatalysis because of the wide potential application and great conductivity of ILs.²⁵ Niu et al. and Guo et al. used CNTs covalently modified with ILs to load Au and Pt nanoparticles, but the CNTs must undergo acid-oxidation pretreatment.^26,27 Chen et al. used raw CNTs covalently modified with polymerized IL (PIL) to support Pt and Pt–Ru nanoparticles, obtaining a superb catalyst for electrochemical performance toward methanol electrooxidation.^28,29 However, the functionalization of CNTs by covalent grafting is more difficult than non-covalent methods and the efficiency is less as well.

In this study, a new facile method was developed to wrap CNTs by polymerizing IL with the cross-linking of divinylbenzene (DVB) by in situ free-radical polymerization. The special conjugated structure of DVB molecule is strongly attracted on the surface of CNTs by π–π interaction. The process of free-radical polymerization would occur between DVB and IL because the multi-functional groups of DVB reduce the damage of CNTs by a ring-opening reaction with monomer. The CNTs can be coated with a cross-linked PIL layer because of the introduction of DVB as a cross-linking agent;³⁰ thus, a type of skin-core structured CNT hybrid was synthesized. As the polymerization reaction occurred between ILs and DVB rather than ILs and CNTs, the integrity and electronic structure of CNTs remained unaffected. With the assistance of the layer of cross-linked IL coated on CNTs non-covalently, Pt nanoparticles were deposited on the CNTs and electrocatalysis of Pt/PIL-CNTs for methanol electrooxidation were compared with those of Pt/CNTs.

2. Experimental section

2.1 Chemicals

All chemical reagents used in this experiment were of analytical grade and used without further purification. The multiwall CNTs (MWCNTs, >95% purity, outer diameter 40–60 nm, 5–15 μm) were purchased from Shenzhen Nanotechnologies Port Co. Ltd. 1-Vinyl-3-ethyl imidazolium bromide ([VEIM]Br) was purchased from Shanghai Cheng Jie Chemical Co. Ltd. DVB was purchased from J&K.

2.2 Synthesis of PIL-CNT

The modified CNTs are made by the following procedure (Scheme 1). In detail, 200 mg of MWCNTs, different amounts of [VEIM]Br, and 50 mg of 2,2′-azobisisobutyronitrile in 100 mL of methanol were sonicated for 30 min and then transferred to a 250 mL round-bottomed flask equipped with a condenser and magnetic stirrer. Moreover, a suitable amount of DVB, which is about a half molar of [VEIM]Br, was added into the solution. Then, the suspension was kept at 353 K in an oil bath for 12 h under vigorous stirring and N₂ protection. The reaction mixture was filtered using a nylon membrane with 0.22 μm pore size and washed with double-distilled water and ethanol several times to thoroughly remove physically absorbed copolymer and unreacted monomer from the surface of the CNTs. The as-functionalized CNTs were dried in a vacuum oven at 333 K for 12 h and collected.

Scheme 1 Schematic of synthetic procedures of Pt nanoparticles on PIL-CNTs. AIBN: 2,2′-azobisiso butyronitrile, EG: ethylene glycol.

2.3 Synthesis Pt/PIL-CNTs

Scheme 1 shows the preparation of Pt/PIL-CNTs composites. In a typical process, 20 mg PIL-CNTs and 200 μL H₂PtCl₆ (51.5 mM) were added to the solvent of 20 mL ethylene glycol under ultrasonication for 30 min. Then, the reaction mixtures were kept at 120 °C in an oil bath for 4 h with continuous magnetic stirring. Thereafter, the mixture was filtered through a nylon membrane and washed with double-distilled water and ethanol several times. The as-prepared Pt/PIL-CNT catalysts were dried at 80 °C in a vacuum oven for 12 h. For comparison, Pt nanoparticle-deposited raw CNTs (Pt/CNTs) were prepared in accordance with the previously described procedure.

2.4 Characterization

Determination of the amount of PIL copolymer and the Pt loading of the catalyst in air was conducted by thermogravimetric analysis (TGA, Netzsch, Model TG209F3) in a nitrogen atmosphere. The copolymer coating and distributions of Pt nanoparticles on PIL-CNTs or CNTs were characterized using transmission electron microscopy (TEM) (Tecnai, Model G²20). X-ray photoelectron spectroscopy (XPS) was recorded on an ESCALAB 250 spectrometer (Thermo Electron Corp.). X-ray powder diffraction (XRD) characterization was done on a Shimadzu XRD-6000 diffractometer with a Cu Kα source (1.54056 Å) in the range of 10°–90°.

The electrochemical measurements of catalysts for electrocatalytic oxidation of methanol were conducted at room temperature in a standard three-electrode cell by an Autolab PGSTAT128N electrochemical workstation. A Pt foil and a saturated calomel electrode (SCE) were used as the counter electrode and reference electrode, respectively. The detailed process of the working electrode is described as follows. Approximately 5 mg of catalyst was mixed with 50 μL Nafion (5 wt%) dispersed in 1 mL ethanol solution by ultrasonication for 30 min to form a uniform ink, and then 5 μL of this catalyst ink was dropped on a glassy carbon electrode (3 mm, diameter) dried at room temperature. The cyclic voltammogram (CV) curves of the catalysts were measured in N₂-saturated 0.5 M H₂SO₄ solution at a scan rate of 50 mV s⁻¹ and the methanol electrooxidation was measured in 0.5 M H₂SO₄ + 1 M CH₃OH solution at a scan rate of 50 mV s⁻¹ between −0.2 and 1.0 V vs. SCE. The electrolytic solution was bubbled with a steady stream of N₂ for 10 min to remove O₂.

3. Results and discussion

3.1 Characterization of Pt/PIL-CNT catalysts

The different weight percentages of PIL copolymer on CNTs are measured with TGA, and the curves at a heating rate of 10 °C min⁻¹ in a N₂ atmosphere are shown in Fig. 1. Weight losses of approximately 6%, 18%, and 50% are observed at a temperature range of 250–800 K, which correspond to different layer thicknesses attributed to the different amounts of PIL-copolymers on the CNT surface. The results show successful cladding of PIL on CNTs. In this study, 18% PIL-CNTs are selected as the support for the deposition of Pt nanoparticles because of their suitable thickness and uniformity of coating layer. The Pt loading on CNTs and PIL-CNTs is also obtained through TGA analysis at a heating rate of 10 °C min⁻¹ in an O₂ atmosphere. In accordance with the TGA curves in Fig. 2, the Pt loading on PIL-CNTs and CNTs is 24.1% and 17.3%, respectively.

Fig. 1 TGA curves of PIL-CNTs with varied thickness of coating layer.

Fig. 2 TGA curves of Pt/CNT and Pt/PIL-CNT catalysts recorded in air at 10 K min⁻¹.

The Raman spectra of the raw CNTs and PIL-NTs are shown in Fig. 3. The peak at 1327 cm⁻¹ is assigned to the A_1g breathing mode of disordered graphite structure (D band). In addition, the peak at 1560 cm⁻¹ is assigned to the E_2g stretching mode of graphite (G band). The G band reflects the structure of the sp² hybridized carbon atom. The D bands are due to the defect sites in the hexagonal framework of graphite materials; thus, the change of D peak usually represents the modification of carbon nanomaterials.³¹ The extent of the defects in graphite materials can be quantified by the intensity ratio of the D to G bands (i.e., ID/IG). As shown in Fig. 3, the ID/IG ratios are 0.90 and 0.82 for the raw CNTs and PIL-CNTs, respectively. The ID/IG ratios of both the raw CNTs and PIL-CNTs are similar, and the value of PIL-CNTs decreases slightly because the CNTs are covered by the sp² hybridized carbon atom from the DVB parts. The results indicate that the PIL copolymer modification process leads to no structural damage of CNTs and retains great electric conductivity.

Fig. 3 Raman spectra of raw CNTs and PIL-CNTs.

The coating of the PIL copolymer on the CNTs surface is directly observed and compared to that of raw CNTs by TEM. The micrographs in Fig. 4 show that the PIL layers with different thicknesses cover the CNT surface. Fig. 5 shows the distribution of Pt nanoparticles on the outer space of raw and coated CNTs. As shown in Fig. 5a and b, some of the Pt nanoparticles agglomerate and in some places of CNT, the surfaces are bare. In Fig. 5d and e, the Pt dispersion on the outer space of PIL-CNTs is homogeneous and the Pt nanoparticles are densely coated on PIL-CNTs. The average particle size of Pt supported on CNTs is approximately 4.4 nm with a size distribution range from 3 nm to 7 nm (Fig. 5c). In contrast, the average particle sizes of Pt supported on PIL-CNTs is 3.3 nm with a size distribution range from 2 nm to 5 nm (Fig. 5f). The results show that the Pt nanoparticles deposited on the PIL-CNTs have a narrower size distribution and smaller average particle size than those on the raw CNTs. This condition proves that the PIL copolymer layer uniformly covered the CNTs, thereby offering abundant anchor points on the surface of CNTs. The PIL copolymer modified CNTs have a large amount of positive charge on the surface and attract PtCl₆²⁻ negative ions, which have a positive influence on the nucleation, growth, and binding. This result further shows that PIL copolymer modified CNTs have a smaller particle size, better dispersion, and narrower size distribution than the raw CNTs.

Fig. 4 TEM image of (a) 6%, (b) 18%, and (c) 50% PIL-CNTs.

Fig. 5 TEM (a) and HRTEM (b) images of Pt/CNTs; size distribution histogram of the Pt/CNT nanoparticles (c); and TEM (d) and HRTEM (e) images of Pt/PIL-CNTs; size distribution histogram of Pt/PIL-CNT nanoparticles (f).

The XRD patterns of Pt/PIL-CNTs and Pt/CNTs are shown in Fig. 6, in which the broad peak at 2θ = 26.2° is assigned as the (002) plane of the graphite-like structure of the CNTs. Peaks at 2θ = 39.87°, 46.33°, 67.6°, and 80.97° are assigned to the (111), (200), (220), and (311) planes of Pt, respectively (JCPDS no. 04-0802), displaying a typical diffraction pattern of the face-centered cubic lattice for Pt. Based on the XRD spectra shown in Fig. 6, the diffraction peak of Pt (220) is used to estimate the Pt particle size by the Scherrer equation for no interference of carbon diffraction peak. Calculating the Pt particle size accurately is impossible because the peak of Pt (220) is extremely weak. However, the peaks of Pt (111) and Pt (220) of the Pt/PIL-CNTs are broader and weaker than those of the Pt/CNTs, indicating a smaller size than Pt/CNTs, which is in accordance with the results of TEM analysis.

Fig. 6 XRD patterns of (a) Pt/PIL-CNTs and (b) Pt/CNTs.

The XPS spectra of Pt/PIL-CNTs and Pt/CNTs are presented in Fig. 7. As shown in Fig. 7a, the highest peak at 285 eV is due to C 1s and the peak due to O 1s is observed at 533 eV. In the survey spectra in Fig. 7b, the Pt/PIL-CNT sample exhibits doublet N 1s peaks at 400.1 eV and 402.1 eV, whereas no detectable N peak is observed in the Pt/CNTs. The binding energy at 402.1 eV indicates the presence of the protonated ammonium ions and 400.1 eV is the characteristic peak for the amine-like nitrogen atoms.³² The Pt (4f) spectrum of Pt/PIL-CNTs shows doublet peaks in Fig. 7c, where the intense peaks at 71.8 eV (Pt 4f_7/2) and 74.9 eV (Pt 4f_5/2) are characteristic peaks of metallic Pt, and the lower intensity peaks at 71.2 eV and 75.8 eV are assigned to Pt²⁺and Pt⁴⁺, respectively. These spectra show that the CNTs are covered by the PIL copolymer, in which the positive electricity layer attracts PtCl₆²⁻ by electrostatic adsorption effect and then is reduced to metallic Pt distributed on the CNT surface.

Fig. 7 XPS survey scan of (a) Pt/CNTs and Pt/PIL-CNTs, (b) N 1s spectrum of Pt/PIL-CNTs, and (c) Pt 4f spectrum of Pt/PIL-CNTs.

3.2 Electrochemical investigation of Pt/PIL-CNT catalysts

The electrochemical behavior of Pt/PIL-CNT catalysts is measured by cyclic voltammetry and chronoamperometry (Fig. 8 and 9). Fig. 8a shows the CVs of Pt/PIL-CNT catalysts in N₂-saturated 0.5 M H₂SO₄ solution at a scan rate of 50 mV s⁻¹. The peaks at −0.2 to 0.1 V are associated with the adsorption/desorption processes of hydrogen on the Pt nanoparticle surfaces, which can be used to calculate the ECSA of the Pt-based catalysts.³³ In accordance with the hydrogen adsorption/desorption charges, the ECSA of Pt/PIL-CNTs and Pt/CNTs is calculated by using the following formula (eqn (1)):where Q_H is the average value of the amount of charge exchanged during the hydrogen adsorption/desorption processes and 210 μC cm⁻² represents the charge required to oxidize a monolayer of H₂ on bright Pt. The ECSA value of the Pt/PIL-CNT catalysts is calculated to be 67.5 m² g⁻¹ of Pt, which is higher than that of Pt/CNTs (43.4 m² g⁻¹).The ECSA of Pt/PIL-CNT catalysts show a larger value than that of Pt/CNTs because of the small particle size and good dispersion of Pt nanoparticles on Pt/PIL-CNTs. For the same Pt loading on CNT surfaces covered by the PIL copolymer, our ECSA shows a similar value compared with that of dispersing Pt nanoparticles under microwave as reported by Chen et al.²⁸ using the traditional chemical deposition.

Fig. 8 (a) CVs of the Pt/CNTs and Pt/PIL-CNT catalyst electrodes measured in 0.5 M H₂SO₄ electrolyte at a scan rate of 50 mV s⁻¹; (b) mass activity of the Pt/CNTs and Pt/PIL-CNT catalysts in 0.5 M H₂SO₄ + 1.0 M CH₃OH at a scan rate of 50 mV s⁻¹; (c) specific activity of the Pt/CNTs and Pt/PIL-CNT catalysts in 0.5 M H₂SO₄ + 1.0 M CH₃OH at a scan rate of 50 mV s⁻¹.

Fig. 9 Chronoamperometry curves of Pt/PIL-CNTs and Pt/CNTs measured in 0.5 M H₂SO₄ + 1.0 M CH₃OH at a potential of 0.65 V.

Furthermore, the catalytic activity of the prepared Pt/PIL-CNTs toward methanol electrooxidation is investigated in 0.5 M H₂SO₄ + 1.0 M CH₃OH solution by CV at room temperature. Two significant parameters, the onset potential and the current, should be compared for the relative activity of the catalyst, and the good catalysts should produce oxidation current at a relatively low overpotential.³⁴ CV is conducted in the potential range from −0.2 to 1.2 V at a scan rate of 50 mV s⁻¹. As shown in Fig. 8b and c, the onset potential on Pt/PIL-CNTs is 0.08 V, which is approximately 95 mV more negative than that of the Pt/CNTs (0.17–0.18 V). From the specific activity (mA per mg of Pt or mA per cm²), the forward peak current of Pt/PIL-CNT catalysts is 425.0 mA mg⁻¹ (68.74 mA cm⁻²), which is 2.16 (3.02) times higher than that of the Pt/CNT catalysts (196.6 mA mg⁻¹, 22.72 mA cm⁻²). The specific activity (mA per cm²) of these two catalysts indicates the intrinsic activity of the electrocatalysts. All the results imply that the Pt/PIL-CNT catalysts possess better catalytic activity than Pt/CNTs, which can be attributed to the increased dispersion and decreased size of the Pt nanoparticles on the CNT support; these findings are in agreement with the ECSA results. In addition, the Pt/PIL-CNT catalysts have better catalytic activity than the Pt/CNTs because of the conservation of the structural integrity of CNTs by coating PIL copolymer.

In order to further evaluate the durability of the electrocatalysts, chronoamperometry was performed in N₂-saturated 0.5 M H₂SO₄ + 1.0 M CH₃OH solution at 0.6 V with a scan rate of 50 mV s⁻¹. Fig. 9 shows the typical chronoamperometry curves of the Pt/PIL-CNTs and Pt/CNTs catalysts and the two electrooxidation current curves decrease rapidly at the beginning because of the poisoning of the intermediate species, such as CO_ads, CHO_ads generated from methanol electrooxidation process.^35,36 The reduction of the curves was at a stable stage after 600 s and the current of methanol electrooxidation on the Pt/PIL-CNT catalysts is 168 mA mg⁻¹, which is much higher than that of Pt/CNT catalysts (83.2 mA mg⁻¹). The results indicate that the Pt/PIL-CNT catalysts have a much better activity and durability than Pt/CNT catalysts for methanol.

4. Conclusions

A facile method to synthesize Pt catalysts on PIL-copolymer-coated CNTs was devised successfully. Compared with the previous CNT functionalization with a linear PIL polymer, the CNTs were wrapped with a crosslinked PIL network, which showed minimal damage to the integrity and electronic structure of the CNTs. Physical characterization revealed that the non-covalently functionalized CNTs by PIL copolymer provided uniform anchor points for the metal nanoparticles, and the Pt nanoparticles with high density and small size were easily deposited on the CNT surface. Electrochemical characterization demonstrated that compared with the Pt/CNTs, the as-prepared Pt/PIL-CNTs have larger ECSA and better electrochemical performance toward methanol electrooxidation. The results showed that the PIL-CNTs are good candidates for catalyst support.

Acknowledgements

The authors gratefully acknowledge the financial support from the National Natural Science Foundation of China (NSFC) (No. 51573010).

References

1. C. Bussy, H. Ali-Boucetta and K. Kostarelos, Acc. Chem. Res., 2013, 46, 692–701.
2. B. H. Steele and A. Heinzel, Nature, 2001, 414, 345–352.
3. V. Castranova, P. A. Schulte and R. D. Zumwalde, Acc. Chem. Res., 2013, 46, 642–649.
4. Z. Bai, L. Yang, J. Zhang, L. Li, C. Hu, J. Lv and Y. Guo, J. Power Sources, 2010, 195, 2653–2658.
5. J. Li, Y. Liang, Q. Liao, X. Zhu and X. Tian, Electrochim. Acta, 2009, 54, 1277–1285.
6. M. Cao, D. Wu and R. Cao, ChemCatChem, 2014, 6, 26–45.
7. C. Hu, Y. Guo, Y. Cao, L. Yang, Z. Bai, K. Wang, P. Xu and J. Zhou, Mater. Sci. Eng., B, 2011, 176, 1467–1473.
8. T. Fujigaya, M. Okamoto and N. Nakashima, Carbon, 2009, 47, 3227–3232.
9. S. Chen, Z. Wei, L. Guo, W. Ding, L. Dong, P. Shen, X. Qi and L. Li, Chem. Commun., 2011, 47, 10984–10986.
10. L. Tao, S. Dou, Z. Ma and S. Wang, Electrochim. Acta, 2015, 157, 46–53.
11. X. Yang, X. Liu, X. Meng, X. Wang, G. Li, C. Shu, L. Jiang and C. Wang, J. Power Sources, 2013, 240, 536–543.
12. Y. Kuang, Y. Cui, Y. Zhang, Y. Yu, X. Zhang and J. Chen, Chem.–Eur. J., 2012, 18, 1522–1527.
13. S. Wang, X. Wang and S. P. Jiang, Langmuir, 2008, 24, 10505–10512.
14. B. Wu, Y. Kuang, X. Zhang and J. Chen, Nano Today, 2011, 6, 75–90.
15. X. M. Sun, T. Chen, Z. B. Yang and H. S. Peng, Acc. Chem. Res., 2013, 46, 539–549.
16. Y. Liu, Y. L. Zhao, B. Y. Sun and C. Y. Chen, Acc. Chem. Res., 2013, 46, 702–713.
17. E. Yoo, T. Okada, T. Kizuka and J. Nakamura, J. Power Sources, 2008, 180, 221–226.
18. L. Zhao, Z.-B. Wang, X.-L. Sui and G.-P. Yin, J. Power Sources, 2014, 245, 637–643.
19. M. Ghasemi, M. Ismail, S. K. Kamarudin, K. Saeedfar, W. R. W. Daud, S. H. A. Hassan, L. Y. Heng, J. Alam and S.-E. Oh, Appl. Energy, 2013, 102, 1050–1056.
20. W. Hong, Y. Liu, J. Wang and E. Wang, J. Power Sources, 2013, 241, 751–755.
21. W. Yang, X. Wang, F. Yang, C. Yang and X. Yang, Adv. Mater., 2008, 20, 2579–2587.
22. Y. Cheng and S. P. Jiang, Electrochim. Acta, 2013, 99, 124–132.
23. M. Grzelczak, M. A. Correa-Duarte, V. Salgueiriño-Maceira, B. Rodríguez-González, J. Rivas and L. M. Liz-Marzán, Angew. Chem., Int. Ed., 2007, 46, 7026–7030.
24. T. Fujigaya and N. Nakashima, Adv. Mater., 2013, 25, 1666–1681.
25. T. Fukushima, A. Kosaka, Y. Yamamoto, T. Aimiya, S. Notazawa, T. Takigawa, T. Inabe and T. Aida, Small, 2006, 2, 554–560.
26. Z. Wang, Q. Zhang, D. Kuehner, X. Xu, A. Ivaska and L. Niu, Carbon, 2008, 46, 1687–1692.
27. S. Guo, S. Dong and E. Wang, Adv. Mater., 2010, 22, 1269–1272.
28. B. Wu, D. Hu, Y. Kuang, B. Liu, X. Zhang and J. Chen, Angew. Chem., Int. Ed., 2009, 48, 4751–4754.
29. B. Wu, Y. Kuang, Y. Zhang, X. Zhang and J. Chen, J. Mater. Chem., 2012, 22, 13085.
30. Y. Ren, Z. Zhou, G. Yin, G.-X. Chen and Q. Li, Mater. Lett., 2016, 166, 133–136.
31. K. Androulaki, K. Chrissopoulou, D. Prevosto, M. Labardi and S. H. Anastasiadis, ACS Appl. Mater. Interfaces, 2015, 7, 12387–12398.
32. X. Feng, W. Gao, S. Zhou, H. Shi, H. Huang and W. Song, Anal. Chim. Acta, 2013, 805, 36–44.
33. M. S. Saha, R. Li and X. Sun, J. Power Sources, 2008, 177, 314–322.
34. Z. Cui, C. Liu, J. Liao and W. Xing, Electrochim. Acta, 2008, 53, 7807–7811.
35. P. Ferrin and M. Mavrikakis, J. Am. Chem. Soc., 2009, 131, 14381–14389.
36. T. H. M. Housmans, A. H. Wonders and M. T. M. Koper, J. Phys. Chem. B, 2006, 110, 10021–10031.

---