Fabrication of a novel tin gas diffusion electrode for electrochemical reduction of carbon dioxide to formic acid

Abstract

Current gas diffusion electrodes (GDEs) for electrochemical reduction of carbon dioxide to formic acid (ERCF) suffer from poor catalyst utilization or high cost. In this work, we developed a novel Sn GDE (SGDE) consisting of a PTFE-bound gas diffusion layer produced using the rolling-press method and a Nafion-bound catalyst layer synthesised by spraying. The influence of Sn loading, Nafion fraction and electrolytic potential was studied. The results showed that increasing the Sn loading and Nafion fraction in an appropriate range both led to an improvement in ERCF via the increase in the total area of the gas–liquid–solid three phase interface. The shift of the electrolytic potential in the negative direction accelerated ERCF, but shifting too far caused the occurrence of side reactions. The highest Faraday efficiency (72.99 ± 1.91%) and current density (13.45 ± 0.72 mA cm⁻²) for ERCF were obtained at −1.8 V vs. Ag/AgCl with the SGDE when the Sn loading and Nafion fraction were 5 mg cm⁻² and 50 wt%, respectively. This Faraday efficiency is one of the highest values for Sn GDEs under ambient pressure. Moreover, the fabrication cost of the SGDE is low (ca. 80 % m⁻²). This work indicates the great potential of the SGDE for ERCF in large scale applications.

---

1. Introduction

Formic acid is an important feedstock in chemical industry and could also be used as a fuel in fuel cells.¹ It is also a promising hydrogen storage material.² The electrochemical reduction of CO₂ to formic acid (ERCF) is of growing interest, since it offers an approach for the reduction of CO₂, which is widely regarded as one of the most significant contributions to climate change.^3,4 Moreover, ERCF consumes less energy than traditional chemical reduction processes and can proceed at a moderate temperature and atmospheric pressure.^4,5 Many researchers have announced that ERCF as a large scale application is economically feasible.^6–13 The economic feasibility of ERCF intensely depends on the cathode, including its electrochemical performance (e.g., Faraday efficiency, current density) and cost.^8,12,13 Most of the current cathodes for ERCF are metal electrodes and CO₂ is thus provided by sparging in bulk. As the solubility of CO₂ in water under ambient conditions is relatively low (ca. 0.033 M), the current density of ERCF is inevitably limited by the mass transfer of CO₂ from the bulk to the cathode surface. For the sake of an increase in the current density of ERCF, gas diffusion electrodes (GDEs) have been introduced. The GDE is a porous composite electrode, usually consisting of a gas diffusion layer (GDL), a current collector (CC) and a catalyst layer (CL).¹⁴ The GDL has a double function of gas delivery and waterproofing. The CC is used to reduce the electrode ohmic losses. The CL possesses a gas–liquid–solid three phase interface (TPI), which is the reaction site for ERCF. It has been reported that the current density of ERCF can be increased by one to two orders of magnitude using GDEs.^14,15 Two types of GDEs have been widely used for ERCF in current studies. In the first one, the GDL is a mixture of carbon black and polytetrafluoroethylene (PTFE) binder and the CL is composed of catalyst, carbon black and PTFE binder. The CC is usually made of metal mesh/foam.^14,15 This type is noted as PTFE-GDE here. Significant advantages, including even porosity, ideal air permeability and good mechanical strength as well as the low fabrication cost, have been found in the GDL (including CC) of the PTFE-GDE.¹⁶ By using a tin-impregnated PTFE-GDE, Mahmood et al. obtained a Faraday efficiency of 49% for ERCF.¹⁴ But this Faraday efficiency is insufficient to support the industrial application,¹³ more probably owing to the inadequacy of the catalyst utilization caused by PTFE which cannot conduct protons.¹⁷ In the second one, the CL consists of catalyst and Nafion binder and the GDL is made of carbon paper/cloth.^8,9 This type is denoted as Nafion-GDE here. By using Nafion-GDE with a Sn catalyst in the CL, Prakash et al. obtained a Faraday efficiency of 70% for ERCF.¹⁸ This improvement in the Faraday efficiency probably results from better catalyst utilization in the Nafion-GDE since Nafion is a proton conductor. However, the cost of the Nafion-GDE is too high (ca. 1050 % m⁻²), impeding its large scale application severely. It should be due to the expensive GDL (1000 % m⁻²,¹⁹ ∼33 times higher than the GDL in the PTFE-GDE), which accounts for 95% of the overall investment in the Nafion-GDE.

In this work, a GDE with Sn catalyst (SGDE) was developed for further improvement in ERCF by integrating the superiority of the current GDEs. Nafion was chosen as the binder in the CL to allow utmost utilization of the catalyst. Low-cost conductive carbon black and PTFE binder were chosen as the primary materials of the GDL. Copper mesh was used as the CC to increase the conductivity and mechanical strength of the electrode. The morphology of the SGDE was examined using scanning electron microscopy (SEM). The influences of the crucial fabrication parameters including Sn loading and Nafion fraction on ERCF were investigated using cyclic voltammetry (CV), electrochemical impedance spectroscopy (EIS) and constant potential electrolysis. The effect of the electrolytic potential on ERCF was also studied.

2. Experimental

2.1 SGDE fabrication

The SGDE consisted of a roll-pressed GDL (including a copper mesh CC) with PTFE binder and a sprayed CL with Sn catalyst and Nafion binder. The detailed fabrication procedure of the roll-pressed GDL is presented in the ESI (Fig. S1†). The CL of the SGDE was prepared by spraying the as-prepared Sn catalyst ink onto the roll-pressed GDL. The Sn catalyst ink was made by mixing Sn catalyst particles (prepared using a chemical reduction method²⁰), Nafion ionomer (DuPont, USA), isopropanol (Tianjin Concord Technology Co. Ltd, China) and deionized water, followed by ultrasonicating the mixture for 1 hour. In order to study the influence of Sn loading on ERCF, four samples with different Sn loading (3, 5, 7, 9 mg cm⁻²) were prepared with the same Nafion fraction of 50 wt%. To study the influence of the Nafion fraction on ERCF, four samples with different Nafion fractions (20, 33, 50, 60 wt%) were prepared with the same Sn loading of 5 mg cm⁻².

2.2 Characterization and analysis

The crystal structure of the as-prepared Sn particles was characterized using X-ray diffraction (XRD, Rigaku D/MAX-2500, Cu Kα, Japan) at a scan rate of 4° min⁻¹ in a 2θ range from 20° to 90°. The morphology of the as-prepared Sn particles and the SGDE were observed using SEM (S-3500N, Hitachi Limited, Japan).

CV and EIS were carried out using an electrochemical workstation (CHI600D, Shanghai Chenhua Instruments Co., China) at ∼298 K in a three-compartment electrochemical cell, which is shown in Fig. S2.† The working electrode chamber and counter electrode chamber were separated by a proton exchange membrane (PEM, Nafion 117, Dupont, USA). More precisely, a gas chamber was designed. The SGDE (7 cm²) and Pt sheet (1 cm², Tianjin Aidahengsheng Technology Co. Ltd., China) were chosen as working electrode and counter electrode, respectively. The reference electrode was an Ag/AgCl electrode (+0.197 V vs. NHE), which was extended to the surface of the working electrode using a Luggin capillary. All of the potential values were in reference to the Ag/AgCl electrode unless otherwise noted. The electrolyte was a 0.5 M KHCO₃ solution. CO₂ (99.99%, Tianjin Sizhi gas Co. Ltd., China) was continuously fed into the gas chamber during the measurement. CV was conducted between 0 V and −2 V at a scan rate of 0.1 V s⁻¹. EIS was performed over a frequency range of 100 kHz to 0.1 Hz with an AC signal amplitude of 0.005 V. The applied potential was −1.3 V. The EIS data were analyzed using the Zsimpwin software (ver. 3.10).

Constant potential electrolysis for ERCF was done at ∼298 K in the same three-compartment electrochemical cell, which is shown in Fig. S2.† The electrolyte was circulated using a peristaltic pump (BT-yz1515, Tianjin Sabo Instruments Co., China) at a flow rate of 25 mL min⁻¹. The flow rate of CO₂ was 30 mL min⁻¹. An electrolysis time of 0.5 h was applied for each batch. The formic acid concentration in the catholyte after each batch was determined using high performance liquid chromatography (HPLC, Beijing Puxitongyong Instruments Co., China) equipped with a C18 reversed phase column (250 mm × 4.6 mm × 5 μm) using UV detection at 210 nm. The Faraday efficiency (FE) of formic acid production was calculated as previously described.⁵ The current density of formic acid production (j_HCOOH) was expressed as the current used for forming formic acid divided by the geometric area of the SGDE.

3. Results

3.1 SGDE characterization

The crystal structure of the as-prepared Sn particles is shown in Fig. S3.† The XRD pattern clearly showed a Sn phase (JCPDS Card no. 65-7657), which indicates that the Sn particles were prepared successfully. The average particle size of the as-prepared Sn particles was calculated to be ∼0.1 μm using the Scherrer equation.²¹ According to SEM, numerous small Sn particles agglomerated together to form a bigger one with average sizes of 0.3–2 μm (Fig. S4†). The morphology of the SGDE was also observed using SEM. The cross-sectional image of the SGDE showed that the GDL and CL were tightly combined with each other and the thickness of the CL was ∼20 μm (Fig. 1a). An abundant pore structure was also observed for the GDL of the SGDE (Fig. 1b), which was formed during the annealing process of the GDL.²² This pore structure would benefit CO₂ transport and prevent electrolyte leakage.²² In the CL (Fig. 1c), a number of pores existed in the agglomerated Sn particles, where the gas–liquid–solid TPI formed.

Fig. 1 SEM images of cross-section (a), gas diffusion layer (b) and catalyst layer (c) of the SGDE. Sn loading: 5 mg cm⁻², Nafion fraction: 50 wt%.

3.2 Influence of fabrication parameters

3.2.1 Influence of Sn loading. The CV curves obtained for the four SGDEs with different Sn contents (3, 5, 7, 9 mg cm⁻²) are given in Fig. 2a. A rapid increase in the reduction current on the cathodic end of the voltammogram can be observed in each case. This possibly resulted from ERCF and the hydrogen evolution reaction (HER).^11,18 The reduction current increased with an increase in Sn loading from 3 mg cm⁻² to 7 mg cm⁻², but it decreased slightly when the Sn loading further increased to 9 mg cm⁻². This indicates that increasing the Sn loading within a certain range (<7 mg cm⁻² here) can expedite the reduction reactions (involving ERCF and HER).

Fig. 2 CV curves (a), Nyquist plots (b) and constant potential electrolysis results (c) obtained for the SGDEs with different Sn loading in 0.5 M KHCO₃ solution. Nafion fraction: 50 wt%. Error bars based on three duplicate measurements for each experiment.

The electrochemical performance of the four SGDEs was further surveyed using EIS measurements (Fig. 2b). All impedance plots consisted of two semicircles and could be modeled by the same equivalent circuit (Fig. S5†). The solution resistances (R_s) and ohm resistances (R₁) for the four SGDEs were similar (Table S1†). Whereas obvious differences were found in the charge transfer resistances (R₂) among the electrodes, which decreased as follows: 3 mg cm⁻² (7.588 Ω) > 5 mg cm⁻² (7.078 Ω) > 9 mg cm⁻² (6.742 Ω) > 7 mg cm⁻² (6.453 Ω). This suggests that in a certain range (<7 mg cm⁻² here), increasing the Sn loading can accelerate the reduction reactions (involving ERCF and HER), which is in agreement with the aforementioned CV results.

The results of the ERCF experiments operated at −1.8 V are shown in Fig. 2c. The maximum FE was obtained when the Sn loading was 5 mg cm⁻² (72.99 ± 1.91%), which was 14% and 53% higher than those when the Sn loading was 3 mg cm⁻² (64.09 ± 0.84%) and 9 mg cm⁻² (47.61 ± 2.44%), respectively. The variation tendency of j_HCOOH with the Sn loading was the same as FE. The maximum j_HCOOH was achieved when the Sn loading was 5 mg cm⁻² (13.45 ± 0.72 mA cm⁻²), which was 45% and 13% higher than those when the Sn loading was 3 mg cm⁻² (9.26 ± 0.58 mA cm⁻²) and 9 mg cm⁻² (11.88 ± 1.24 mA cm⁻²), respectively. Hence, an optimal Sn loading of 5 mg cm⁻² in the SGDE for ERCF was achieved under our conditions.

3.2.2 Influence of Nafion fraction. As shown in Fig. 3a, the reduction current resulting from ERCF and HER increased with an increase in the Nafion fraction from 20 wt% to 60 wt%. This indicates that increasing the Nafion fraction within a certain range (<60 wt% here) can expedite the reduction reactions (involving ERCF and HER).

Fig. 3 CV curves (a), Nyquist plots (b) and constant potential electrolysis results (c) obtained for the SGDEs with different Nafion fractions in 0.5 M KHCO₃ solution. Sn loading: 5 mg cm⁻². Error bars based on three duplicate measurements for each experiment.

Nyquist plots of the four SGDEs with different Nafion fractions are compared in Fig. 3b. All impedance plots consisted of two semicircles and could be modeled by the same equivalent circuit (Fig. S5†). R_s and R₁ for the four SGDEs with different Nafion fractions were almost similar (Table S2†). Significant differences were observed for R₂, which decreased as follows: 8.522 Ω (20 wt%) > 7.654 Ω (33 wt%) > 7.078 Ω (50 wt%) > 4.649 Ω (60 wt%). This suggests that increasing the Nafion fraction in a certain range (<60 wt% here) can accelerate the reduction reactions (involving ERCF and HER), which is in line with the aforementioned CV results.

ERCF tests were further carried out with the four SGDEs of different Nafion fractions at −1.8 V. As shown in Fig. 3c, the FE increased with an increase in Nafion fraction from 20 wt% to 50 wt%, but it decreased when the Nafion fraction further increased to 60 wt%. The maximum FE was obtained when the Nafion fraction was 50 wt% (72.99 ± 1.91%), which was 22% and 18% higher than those when the Nafion fractions were 20 wt% (59.59 ± 1.32%) and 60 wt% (61.99 ± 2.66%), respectively. Similar to the FE, the maximum j_HCOOH was also achieved when the Nafion fraction was 50 wt% (13.45 ± 0.72 mA cm⁻²), which was 43% and 14% higher than those when the Nafion fractions were 20 wt% (9.38 ± 0.55 mA cm⁻²) and 60 wt% (11.81 ± 0.45 mA cm⁻²), respectively. Hence, an optimal Nafion fraction of 50 wt% in the SGDE for ERCF was achieved under our conditions.

3.3 Influence of electrolytic potential

In order to determine the influence of the electrolytic potential on ERCF with the optimal SGDE, a series of experiments at different electrolytic potentials (from −1.3 V to −2.2 V) were carried out (Fig. 4). The FE increased when the electrolytic potential shifted from −1.3 V to −1.8 V, but it decreased when the electrolytic potential further shifted to −2.2 V. The maximum FE (72.99 ± 2.91%) was achieved at an electrolytic potential of −1.8 V, which was three times higher than that at −1.3 V (25.07 ± 0.55%). Similar to the FE, the maximum j_HCOOH (13.45 ± 0.72 mA cm⁻²) was also obtained at the electrolytic potential of −1.8 V, which was 224 times higher than that at −1.3 V (0.22 ± 0.02 mA cm⁻²). This is probably because of the fact that when the electrolytic potential was more positive than −1.8 V, ERCF became more slack due to the low standard potential of CO₂/CO₂⁻ in aqueous media (−2.33 V vs. SCE²³). While when the electrolytic potential was more negative than −1.8 V, side reactions (such as HER) were more likely to take place, which also led to a decrease in ERCF.¹¹ Therefore, the optimal electrolytic potential for ERCF using the optimal SGDE is −1.8 V, which is in accordance with the results reported by Prakash et al. with a Sn Nafion-GDE under similar conditions.¹⁸

Fig. 4 Variations of current density and FE for formic acid formation at different electrolytic potentials with the optimal SGDE. Error bars based on three duplicate measurements for each experiment.

4. Discussion

By using our SGDE, ERCF occurs at the TPI made up of CO₂, electrolyte and Sn catalyst in the CL. Obviously, the Sn loading and Nafion fraction are the two crucial parameters for the establishment of the TPI. It has been shown that Sn particles agglomerated together, creating a number of pores between these agglomerates for CO₂ diffusion (see Fig. 1c). While Nafion connected those agglomerates to form a continuous frame network for the electrolyte adsorption.²⁴ Thus, abundant TPIs were established in the CL of the SGDE, which are essential to an efficient ERCF. The morphologies of the SGDEs with different Sn loading and Nafion fractions were observed using SEM and are shown in the ESI (Fig. S6 and S7†). For the SGDE with different Sn loading, it can be seen that when the Sn loading increased from 3 to 5 mg cm⁻² (Fig. S6a and S6b†), the number of pores between the agglomerated Sn particles increased. This allowed an enhancement in the total area of the TPI. When the Sn loading further increased from 7 to 9 mg cm⁻² (Fig. S6c and S6d†), the number of pores decreased. So the total area of the TPI decreased. For the SGDEs with different Nafion fractions, it can be seen that the Sn particles were packaged by Nafion. When the Nafion fraction increased from 20 to 50 wt% (Fig. S7a–c†), more and more Sn particles were packaged by Nafion. This led to an increase in proton conduction near the Sn catalyst’s surface, which favored ERCF. When the Nafion fraction was higher than 50 wt% (Fig. S7d†), CO₂ permeability would be impaired, resulting from blocking of the pores between the Sn agglomerates. This led to a decrease in the total area of the TPI. According to the aforementioned results of CV, EIS, constant potential electrolysis and SEM, it can be inferred that the increase in Sn loading (<5 mg cm⁻² in this work) leads to an increase in the total area of the TPI, resulting in an improvement in ERCF.²⁵ However, excessive Sn loading (>5 mg cm⁻² in this work) will inhibit the diffusion of CO₂, leading to a decrease in the total area of the TPI and further in ERCF.²⁵ The increase in Nafion fraction (<50 wt% in this work) leads to an increase in proton conduction near the Sn catalyst’s surface, resulting in an enhancement in ERCF. Whereas an excessive Nafion fraction (>50 wt% in this work) will impair CO₂ permeability by blocking the pores between the Sn agglomerates, resulting in a decrease in the total area of the TPI and further in ERCF.²⁶ In addition, an excessive Nafion fraction will impair electron conduction to the Sn catalyst’s surface since Nafion is non-conducting, also resulting in a decrease in ERCF. Hence, there should be an optimal Sn loading and Nafion fraction in the SGDE for ERCF and the values are 5 mg cm⁻² and 50 wt% in our experiments.

ERCF is an intricate electrochemical process whose performance is highly dependent on the electrolytic potential.⁸ It has been shown that increasing the electrolytic potential in the negative direction accelerated ERCF but shifting it too far caused the occurrence of side reactions (such as HER). Thus, the optimal electrolytic potential for ERCF should exist and the value is −1.8 V here.

By using the optimal SGDE for ERCF, the kinetic performance was evaluated using the Tafel plot (Fig. 5). The exchange current density j₀ for the optimal SGDE (3.49 × 10⁻⁴ A cm⁻²) was somewhat higher than that for the previously reported Nafion-GDE with Sn catalyst (1.60 × 10⁻⁴ A cm⁻²).¹⁸ j₀ has a close relationship with the intrinsic catalytic activities of the electrode materials.²⁷ Therefore the intrinsic catalytic activity of the optimal SGDE is somewhat higher than that of the Nafion-GDE with Sn catalyst for ERCF. This is supported by the results of the ERCF experiments with the optimal SGDE in 0.5 M KHCO₃ solution at −1.8 V, which shows that the highest FE and j_HCOOH are 72.99 ± 1.91% and 13.45 ± 0.72 mA cm⁻², respectively. This FE is somewhat higher than that for the Nafion-GDE with Sn catalyst in 0.5 M NaHCO₃ solution in a fuel cell-like device (70%),¹⁸ while j_HCOOH is somewhat lower than that for the latter (18.9 mA cm⁻²)¹⁸ but still in the same order of magnitude. This mainly results from the differences in the ERCF experiment conditions, such as reactor structure, electrolyte, electrolytic time, etc.⁸ The FE of 72.99 ± 1.91% obtained in this work is one of the highest values found in the literature for previously reported GDEs with Sn catalyst under similar conditions.^{9,13,14,18,25,28,29} Although the electrochemical performance of the optimal SGDE for ERCF is equal to that of the Nafion-GDE with Sn catalyst, the fabrication cost of the optimal SGDE (ca. 80 % m⁻²) is 92% lower than that of the latter (ca. 1050 % m⁻²).

Fig. 5 Tafel plot for the optimal SGDE obtained from the corresponding voltammograms.

ERCF tests were also conducted with a Sn GDE that was similar to our SGDE except that the CL was PTFE-bound (Sn loading: 5 mg cm⁻²; PTFE fraction: 50 wt%) at an electrolytic potential of −1.8 V. The FE (63.89 ± 1.11%) and j_HCOOH (6.62 ± 0.21 mA cm⁻²) obtained from this Sn GDE were 12% and 51% lower than those obtained for the optimal SGDE, respectively. This indicates that the Sn GDE with the Nafion-bound CL shows a more excellent electrochemical performance in ERCF than that with the PTFE-bound CL.

The long term performance of the optimal SGDE in ERCF was also studied in this work at an electrolytic potential of −1.8 V for 12 h of operation (Fig. S8†). It can be seen that FE slightly decreased with the electrolysis time and could maintain ∼70% for at least 3 h. The FE at the electrolysis time of 12 h (63.42 ± 3.10%) was only 13% lower than that at 0.5 h (72.99 ± 1.91%). The degradation in performance in ERCF has been reported previously^6,9,11,13,30 and it can be attributed to several reasons, for example,^6,9,11,30 (1) accumulated formic acid crossed the membrane to the anode side to be oxidized to CO₂, (2) formic acid reduction at the cathode side, etc. To alleviate the aforementioned effect on ERCF, intermittent electrolysis with an electrolysis time of 3 h in each run and four consecutive runs was carried out (Fig. S8†). It was found that the FE was almost unchanged. This indicates that the optimal SGDE exhibits high long-term stability in ERCF, and intermittent electrolysis is more reasonable for ERCF.

In short, the optimal SGDE shows a more excellent electrochemical performance than the Sn PTFE-GDE and costs less than the Sn Nafion-GDE. In addition, the optimal SGDE also exhibits high stability in ERCF. Thus, it exhibits more potential for ERCF in large scale applications.

5. Conclusions

A novel gas diffusion electrode with Sn catalyst (SGDE) consisting of a PTFE-bound gas diffusion layer (GDL), synthesised using the rolling-press method, and a Nafion-bound catalyst layer (CL), produced by spraying, was developed for electrochemical reduction of CO₂ to formic acid (ERCF). The optimal Sn loading and Nafion fraction are 5 mg cm⁻² and 50 wt%, respectively, for the establishment of a gas–liquid–solid three phase interface for ERCF. An optimal electrolytic potential of −1.8 V was achieved. The highest Faraday efficiency (72.99 ± 1.91%) and current density (13.45 ± 0.72 mA cm⁻²) for ERCF were reached using the optimal SGDE at the optimal electrolytic potential of −1.8 V in 0.5 M KHCO₃ solution. This Faraday efficiency is one of the highest values found in the literature for previously reported Sn GDEs under ambient pressure. The electrocatalytic activity of the SGDE for ERCF is stable. The mechanical strength of the SGDE is good and the fabrication cost of the SGDE is low (ca. 80 % m⁻²). The SGDE has great potential in promoting ERCF applications on a large scale.

Acknowledgements

The authors gratefully acknowledge financial support by the Major National Science & Technology Projects of China on Water Pollution Control and Treatment (2012ZX07501002-001) and the Research Project of Tianjin City for Application Foundation and Advanced Technology (BE026071).

References

1. X. Yu and P. G. Pickup, J. Power Sources, 2008, 182, 124–132.
2. M. Grasemann and G. Laurenczy, Energy Environ. Sci., 2012, 5, 8171–8181.
3. J. H. Jeon, P. M. Mareeswaran, C. H. Choi and S. I. Woo, RSC Adv., 2014, 4, 3016.
4. Y. Koo, R. Malik, N. Alvarez, L. White, V. N. Shanov, M. Schulz, B. Collins, J. Sankar and Y. Yun, RSC Adv., 2014, 4, 16362.
5. H.-Z. Zhao, Y. Zhang, Y.-Y. Chang and Z.-S. Li, J. Power Sources, 2012, 217, 59–64.
6. H. Li and C. Oloman, J. Appl. Electrochem., 2007, 37, 1107–1117.
7. M. Alvarez-Guerra, S. Quintanilla and A. Irabien, Chem. Eng. J., 2012, 207–208, 278–284.
8. X. Lu, D. Y. C. Leung, H. Wang, M. K. H. Leung and J. Xuan, ChemElectroChem, 2014, 1, 836–849.
9. R. L. Machunda, H. Ju and J. Lee, Curr. Appl. Phys., 2011, 11, 986–988.
10. S. R. Narayanan, B. Haines, J. Soler and T. I. Valdez, J. Electrochem. Soc., 2011, 158, A167.
11. W. Lv, R. Zhang, P. Gao and L. Lei, J. Power Sources, 2014, 253, 276–281.
12. A. S. Agarwal, Y. Zhai, D. Hill and N. Sridhar, ChemSusChem, 2011, 4, 1301–1310.
13. C. Oloman and H. Li, ChemSusChem, 2008, 1, 385–391.
14. M. Mahmood, D. Masheder and C. Harty, J. Appl. Electrochem., 1987, 17, 1159–1170.
15. G. Lu, H. Wang, Z. Bian and X. Liu, Sci. World J., 2013, 1–8.
16. L. Cindrella, A. M. Kannan, J. F. Lin, K. Saminathan, Y. Ho, C. W. Lin and J. Wertz, J. Power Sources, 2009, 194, 146–160.
17. M. S. Wilson and S. Gottesfeld, J. Appl. Electrochem., 1992, 22, 1–7.
18. G. K. S. Prakash, F. A. Viva and G. A. Olah, J. Power Sources, 2013, 223, 68–73.
19. J. Wei, P. Liang and X. Huang, Bioresour. Technol., 2011, 102, 9335–9344.
20. X. Zhao, Y. Xuan, Y. Liu and C. Zhang, Tribol. Lubr. Technol., 2007, 32, 108–110.
21. S. Zhang, P. Kang and T. J. Meyer, J. Am. Chem. Soc., 2014, 136, 1734–1737.
22. H. Dong, H. Yu, H. Yu, N. Gao and X. Wang, J. Power Sources, 2013, 232, 132–138.
23. V. Lates, A. Falch, A. Jordaan, R. Peach and R. J. Kriek, Electrochim. Acta, 2014, 128, 75–84.
24. A. Suzuki, U. Sen, T. Hattori, R. Miura, R. Nagumo, H. Tsuboi, N. Hatakeyama, A. Endou, H. Takaba, M. C. Williams and A. Miyamoto, Int. J. Hydrogen Energy, 2011, 36, 2221–2229.
25. J. Wu, P. P. Sharma, B. H. Harris and X.-D. Zhou, J. Power Sources, 2014, 258, 189–194.
26. M. Eguchi, K. Baba, T. Onuma, K. Yoshida, K. Iwasawa, Y. Kobayashi, K. Uno, K. Komatsu, M. Kobori, M. Nishitani-Gamo and T. Ando, Polymers, 2012, 4, 1645–1656.
27. S. Zhang, P. Kang, S. Ubnoske, M. K. Brennaman, N. Song, R. L. House, J. T. Glass and T. J. Meyer, J. Am. Chem. Soc., 2014, 136, 7845–7848.
28. J. Qiao, Y. Liu, F. Hong and J. Zhang, Chem. Soc. Rev., 2014, 43, 631–675.
29. D. T. Whipple, E. C. Finke and P. J. Kenis, Electrochem. Solid-State Lett., 2010, 13, B109–B111.
30. J. Wu, F. G. Risalvato, F. S. Ke, P. J. Pellechia and X. D. Zhou, J. Electrochem. Soc., 2012, 159, F353–F359.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c4ra10775f

---