Photochemical water splitting mediated by a C1 shuttle†

N. P. Alderman ^a, J. M. Sommers ^a, C. J. Viasus ^a, C. H. T. Wang ^a, V. Peneau ^a, S. Gambarotta *^a, B. Vidjayacoumar ^b and K. A. Al-Bahily ^b
aDepartment of Chemistry, University of Ottawa, Ottawa, Ontario K1N 6N5, Canada. E-mail: sgambaro@uottawa.ca
^bAdvanced Catalysis SABIC CDR Centre at KAUST, Thuwal, 23955, Saudi Arabia

First published on 31st October 2016

---

---

1. Introduction

Large-scale water splitting is currently sought as an alternative to carbon based energy production. For this reason, research efforts to identify routes that afford large amounts of inexpensive hydrogen are currently flourishing around the world in both academic and corporate sectors.^1–5

In sight of the future extraction of hydrogen from water on a large scale, the issue of oxygen/hydrogen separation is a major concern^6,7 because it is expected to substantially increase production costs.^8,9 To solve this problem a system similar to an electrolytic device, producing hydrogen and oxygen in two separate compartments, would be an attractive solution. However, this is challenging for a purely photochemical device where the photosemiconductor simultaneously produces the two gases either from the same surface^10–12 or from two surfaces of a composite film.¹³ Stimulating ideas on this line of research have been elaborated by Milstein with the consecutive thermal and photochemical water splitting¹⁴ and by Cronin with the two step use of an electrochemical cell combined with a W⁴⁺/W⁶⁺ polyoxoanion¹⁵ or a quinone derivative.¹⁶ The concept is to split the catalytic cycle into two elemental redox events of hydrogen and oxygen release and to carry them out in two separate stages.

Inspired by the naturally occurring photosystem (PS) (II),¹⁷ where the oxygen release is coupled to the transport of electrons and protons by a quinoid system,¹⁸ we have now discovered a new possibility to carry out water splitting using simple shuttle chemistry. We have developed a catalytic redox system shuttling between its reduced and oxidized forms, in the process of releasing hydrogen and oxygen in two separate stages with the input of water, light and two catalysts. In other words, we have used one molecule capable of working toward the storage of hydrogen in its reduced form, releasing H₂ on demand, and affording its oxidized form in a selective fashion. In turn, the oxidized shuttle can be re-hydrogenated to the original reduced form via water oxidation and oxygen release. Although for this study we have focused on the formaldehyde/formate anion as a redox-hydrogen shuttling couple (Scheme 1), the same concept could also apply to the amplest variety of inorganic and organic shuttles. In any event, two separate catalysts and an energy input are certainly necessary for this purpose. The two catalysts should be compatible (not reacting with each other) and the two redox events could ideally be performed in two separate compartments. Herein, we report our findings.

2. Results and discussion

Several molecules, either organic (e.g. quinone/hydroquinone) or inorganic (e.g. hydrogen storing devices, polyoxo anions, etc.), may, in principle, be considered as shuttles. For this study, we have focused on the lightest C1 organic shuttle: the redox couple formaldehyde (methanediol in aqueous solutions)/formic acid (formate anion). Two very good catalytic systems for para-formaldehyde (p-FA) dehydrogenation were recently independently reported by Prechtl and Fukuzumi based on Ru and Ir, respectively, capable of efficiently converting hydrated formaldehyde into H₂ and CO₂ (at 95 °C and room temperature, respectively).^19,20 The formic acid/CO₂ couple^21–23 was not considered as a shuttle in this study as it poses a formidable challenge due to the thermodynamic stability of CO₂. Reducing CO₂ with water as a hydrogenating agent is insofar possible only via the so-called “reverse combustion” process.^24–30 This fascinating process, regrettably lacks selectivity unless it is combined with photo-electrochemical techniques.^31–33 Therefore, we have identified a light-switchable catalyst for formaldehyde dehydrogenation selectively affording formate anions and one for the photochemical reduction of Na(formate) to formaldehyde using water as a hydrogenating agent. Interestingly, one of these air-stable and chemically robust systems works in river and seawater, and only requires visible light.

2.1 Hydrogen production via dehydrogenation of formaldehyde

When a three-component solution, consisting of a catalytic amount of commercial sodium ferrocyanide and equimolar amounts of NaOH and p-FA, is exposed to visible light in a pyrex vessel (either 300 W Xe lamp or direct sunlight), a vigorous evolution of H₂ is observed. The reaction is accompanied by the formation of sodium formate according to eqn (1) as follows:

CH2O(l) + NaOH(aq) → H2(g) + HCOONa(aq) (ΔGf = −91 kJ mol−1).	(1)

The GC analysis of the H₂ effluent showed no trace of contamination by CO₂ or CO (ESI Fig. 1 and 2†).

The disappearance of formaldehyde was monitored by titration, whereas the formation of sodium formate by colorimetry (Fig. 1). The presence of a base is a necessity and its role is most likely that of ensuring a rapid transformation of p-FA into methanediol and its partial deprotonation.³⁴

While the productivity of hydrogen and formate followed a similar trend, the slight overproduction of formate, with respect to hydrogen, is attributable to the Cannizzaro's slow and yet unavoidable disproportionation of formaldehyde in a basic medium (eqn (2)).³⁵

2CH2O(l) + NaOH(aq) → HCOONa(aq) + CH3OH(l).	(2)

The H₂ evolution only occurs when using Fe^II(CN)₆⁴⁻ catalysts while the Fe(III) corresponding salt is completely inert. The reaction is rapid and efficient and both p-FA and NaOH are consumed during the hydrogen evolution cycles. During the catalyst activation period, the pH has an initial decrease with respect to H₂ production versus the blank. After the initial decrease, the pH quickly tracks with the blank rate, which is likely showing the background Cannizzaro reaction. The only deviation in pH change at the beginning of the reaction leads us to believe that NaOH is also required to activate the catalyst (ESI Fig. 3†).

The efficiency of the catalytic reaction is a function of the initial concentration of both p-FA and NaOH. The total production of hydrogen as depending on the initial amount of p-FA at a constant amount of NaOH was measured (ESI Fig. 4†). At low initial concentrations of p-FA, the conversion to hydrogen can be as high as 100%, but decreases as the concentration of p-FA is increased. This can be attributed to the longer time required to dehydrogenate larger levels of formaldehyde, resulting in a larger proportion of formaldehyde undergoing the Cannizzaro reaction. The maximum total productivity can be reached when the amount of p-FA is approximately equimolar with NaOH (p-FA/NaOH = 1.2). Higher ratios show a detrimental effect on hydrogen production by showing a sharp decline.

It is interesting to note that during experiments carried out with low p-FA loading and a consequent large excess of NaOH, a complete conversion of methanediol to formate occurs, where 100% yields can be obtained (Table 1).

---

Although the reaction is energetically permissible, constant illumination is a necessity, with the type of illumination not playing a significant role (ESI Fig. 5†). When irradiation with visible light is interrupted, the hydrogen evolution stops and it can be either restarted or arrested by intermittently turning the light on and off (Fig. 2). In other words, the catalytic system is light-switchable.

The hydrogen rate of formation and productivity (Fig. 3) were monitored by regularly adding p-FA and NaOH in both 1:5 and 1:1 molar ratio. Initially, 0.25 moles of NaOH were added to both solutions to ensure that the pH was suitable for the deprotonation of methanediol.³⁴ High hydrogen evolution rates were observed with the 1:5 p-FA:NaOH solution, which quickly tapered off before the next sample was added. However, with each addition of p-FA and NaOH, the maximum rate decreased rapidly to zero after the 10^th addition. This drop is attributed to the decomposition of catalyst to Fe₂O₃ in highly basic conditions. Addition of 1:1 p-FA:NaOH instead results in steady productivity being reached, with no decline of catalytic activity for 350 minutes. With each addition, a reduction in the initial spike of hydrogen production was observed, which stabilized into a nearly continuous release of 10 mL of hydrogen per minute.

The catalyst, in addition to being oxygen insensitive and capable of fully operating in air and under ambient conditions, is also remarkably resilient to impurities. The effect of water purity on catalytic activity was investigated in water obtained from three sources (ESI Fig. 6†). Regardless of whether the reaction occurred in distilled water, 3.5% NaCl solutions (to match the average salinity of the ocean), or water taken directly from city canal water, there was no significant difference in the activity of the catalyst.

Catalyst resiliency was probed by acidifying a standard reaction mixture to pH = 1.0 using concentrated sulfuric acid. Under an acidic pH, the reaction was halted; when the pH was increased back to 13.0 with NaOH and illuminated with light, the reaction restarted.

Reactions uninterruptedly carried out for 16 h with periodical additions of NaOH/p-FA every 30 min showed that the catalytic system slowly decomposes after having produced a substantial amount of 4.8 L of pure hydrogen (14.5% based on p-FA). At this stage a significant amount of iron oxide (ESI Fig. 7†) becomes visible and can be isolated and characterized.

There are two main mechanistic questions arising from this catalytic cycle. First, during the release of hydrogen, the reaction can be switched on and off simply by the presence or absence of light. A lag time of a few minutes was observed for hydrogen production to completely cease upon the removal of light, whilst the opposite was observed upon illumination. As has previously been shown, the hydrogen release from p-FA is a thermodynamically permissible process.³⁶ Secondly, the reaction exclusively produces sodium formate with no sign of further oxidation to CO₂. The catalyst is a saturated 18-electron complex with Fe in an octahedral, strong ligand-field. However, it is known to embark upon reversible dissociation reactions upon visible light irradiation.^37,38 Thus, the role of the irradiation is to induce dissociation of one cyanide anion and to form a transient penta-coordinated 16-electron species isolobal with an organic carbocation (eqn (3)). Such an electrophilic species is expected to readily react with nucleophiles (eqn (4)).

[Fe(CN)6]4− ↔ [Fe(CN)5]3− + CN−	(3)

[Fe(CN)5]3− + OH− → [Fe(CN)5OH]4−	(4)

The necessity for the presence of excess NaOH to ensure vigorous hydrogen evolution in fact suggests the formation of a Fe–OH intermediate as a realistic possibility. We suggest that such a species is responsible for the attack of formaldehyde, in its hydrated form, followed by reductive elimination of hydrogen and consequent formation of the formate anion. Alternatively, the partly deprotonated form of methanediol may also directly coordinate to the 16-electron [Fe(CN)₅]³⁻ intermediate to form the same species. Unfortunately, our attempts to isolate the intermediate resulted in the crystallization of only Na₄Fe(CN)₆·10H₂O, whereas attempts to analyse the intermediates by ES-MS-TOF resulted in the observation of complex metal-ion clusters with no adducts with bound intermediates visible. While there is no doubt that light is necessary for the initial dissociation of one cyanide anion,^37,38 DFT-B3LYP calculations (Fig. 4) indicated that the coordination of partially-deprotonated methanediol is the most energy demanding step. Interestingly, the exergonic dissociation of a second cyanide anion favours the release of hydrogen from methanediol. At this stage, the coordinated methandiol anion seems to be incapable of adopting the chelating mode. This tentatively suggests that the presence of an empty coordination site around the metal is necessary to trigger H₂ release. The re-coordination of cyanide results in a closed-shell structure with a terminally bonded formate.

2.2 Oxygen evolution via formate photo-reduction with water

The reduction of sodium formate using water (eqn (5)) is a thermodynamically uphill process.

	(5)

Since the input of energy under a form of light is a must, a number of semiconductor metal oxides, which have shown promising catalytic activity for CO₂ reduction such as BiVO₄,²⁷ TiO₂²⁴ and SrTiO₃³⁰ were prepared using literature procedures and tested for the purpose. The most successful catalyst was Bi₂WO₆³⁹ which showed no compatibility issues with Na₄Fe(CN)₆ and was thus adopted for the subsequent testing.

A mixture of Bi₂WO₆ and sodium formate was illuminated with UV/Visible light by a mercury lamp for 24 hours (Fig. 5). This choice was made due to the large bandgap of Bi₂WO₆ (470 nm, ESI Fig. 8†).

The reaction afforded both formaldehyde and oxygen. Presence or absence of additional NaOH did not affect the reaction outcome. After 5 hours, both the oxygen and formaldehyde levels had reached a maximum, with around 6 μmoles of product being obtained. The slight overproduction of oxygen is, again, attributed to the Cannizzaro disproportionation, affording methanol and formate anion ready for another cycle of re-hydrogenation and oxygen production.

Reactivation of the Bi₂WO₆ catalyst was attempted by heating at 300 °C (ESI Fig. 9†). Although the catalyst was indeed reactivated, the production decreased upon each reactivation cycle due to the leaching out of tungsten from the structure during the reaction (ESI Fig. 10†). As with the previous catalytic system, even Bi₂WO₆ proved to be resilient to impurities and the reduction cycles seemed to be unaffected by impurities or salinity of the water medium.

Isotopic labeling experiments using H₂O¹⁸ were performed to confirm that the oxygen evolution was from the oxidation of water. In this case, ¹⁸O₂ was detected by EI MS, which was not present in the standard reaction (ESI Fig. 11†). This confirms that the oxygen produced during the reduction of sodium formate is derived from water and not from sodium formate.

2.3 Combination of oxidation and reduction catalysts

We have finally probed the initial idea of using shuttle chemistry for water splitting and have been able to combine the two hydrogen and oxygen evolution systems into an overall H₂/O₂ production cycle, which is shown in Scheme 1, where catalyst 1 is Na₄Fe(CN)₆·10H₂O and catalyst 2 is Bi₂WO₆. Since formate reduction is the slower part of the cycle, we started the overall water splitting experiment using the two catalysts and sodium formate as a starting point. In this cycle, sodium formate was reduced to formaldehyde under illumination using water while releasing oxygen. The formaldehyde (in basic conditions) was oxidized back to formate by sodium ferrocyanide, releasing hydrogen.

Although the two stages of the overall cycle can be easily carried out in two separate vessels for the separate recovery of H₂ and O₂, we were concerned about the possible incompatibility of the two catalysts during repeated cycles of H₂/O₂ production. We therefore decided to attempt experiments under the most unfavorable conditions, which was by one-pot experiments. This certainly implies that, in this particular case, H₂ and O₂ would be released as a mixture.

Upon irradiation of a solution of sodium formate, sodium hydroxide, sodium ferrocyanide and a stirred suspension of bismuth tungstate, both hydrogen and oxygen were generated (Fig. 6). In all cases, no CO₂ was observed.

The reaction was repeated three times, and in all three cases both hydrogen and oxygen were produced. For a pure water splitting system, a 2:1 hydrogen:oxygen ratio has to be expected. In our case, this is not what was observed due to the parasitic Cannizzaro reaction, which produces both methanol and formate from formaldehyde in a basic medium.

To confirm that the hydrogen was from the dehydrogenation of the photogenerated formaldehyde, the reaction was repeated on removing one of the two catalysts (ESI Fig. 12†). When one of the two catalysts was missing from the reaction mixture, no hydrogen was detected, confirming that hydrogen was indeed generated from the dehydrogenation of the formaldehyde produced from the reduction of formate. When the two catalysts were present and no formate was added, no gas evolution was detected.

3 Conclusions

With this study we have identified catalysts for selective dehydrogenation of formaldehyde to sodium formate and for the re-hydrogenation of sodium formate to formaldehyde using water to overall produce highly pure hydrogen and oxygen and no trace of CO₂. The two catalysts are compatible and can work in tandem to shuttle the C1 molecule between the reduced and oxidized forms, in the process completing a whole cycle of water splitting. This study proves the concept of shuttle chemistry for hydrogen production from water.

Although the current system is quite appealing since its successful implementation may solve the issue of gas mixture separation, the formaldehyde/formate system is clearly not the ideal shuttle for commercial production of hydrogen. The unavoidable Cannizzaro reaction requires a continuous re-injection of C1 units (under the form of either formaldehyde or formate) and eventually elimination (or recycling) of the methanol accumulating during the cycles. Further research will be required to identify more efficient redox shuttles along with their appropriate catalysts.

Acknowledgements

Saudi Basic Industry Corporation (SABIC) is gratefully acknowledged for financial support through a research grant. The authors wish to thank Dr Sharon Curtis and Dr Alexander Mommers from the John L. Homes Mass Spectrometry Facility for their help with the isotopic experiments. Parts of this study have been filed as patent applications US 15/150,680, US 62/300,175 and PCT/IB2016052676.

Notes and references

1. Y. Qiu, W. Liu, W. Chen, W. Chen, G. Zhou, P.-C. Hsu, R. Zhang, Z. Liang, S. Fan, Y. Zhang and Y. Cui, Sci. Adv., 2016, 2, 1–8 Search PubMed .
2. K. Maeda, K. Teramura, D. Lu, T. Takata, N. Saito, Y. Inoue and K. Domen, Nature, 2006, 440, 295–295 CrossRef CAS PubMed .
3. A. Fujishima and K. Honda, Nature, 1972, 238, 37–38 CrossRef CAS PubMed .
4. S. Y. Reece, J. A. Hamel, K. Sung, T. D. Jarvi, A. J. Esswein, J. J. H. Pijpers and D. G. Nocera, Science, 2011, 334, 645–648 CrossRef CAS PubMed .
5. D. J. Martin, P. J. T. Reardon, S. J. A. Moniz and J. Tang, J. Am. Chem. Soc., 2014, 44, 2–5 Search PubMed .
6. J. W. Phair and S. P. S. Badwal, Sci. Technol. Adv. Mater., 2006, 7, 792–805 CrossRef CAS .
7. V. Schröder, B. Emonts, H. Janßen and H. P. Schulze, Chem. Eng. Technol., 2004, 27, 847–851 CrossRef .
8. G. G. Haselden, Gas Sep. Purif., 1989, 3, 209–215 CrossRef CAS .
9. P. Rao and M. Muller, 2007 ACEEE Summer Study Energy Effic. Ind., 2007, pp. 124–135 .
10. L. Liao, Q. Zhang, Z. Su, Z. Zhao, Y. Wang, Y. Li, X. Lu, D. Wei, G. Feng, Q. Yu, X. Cai, J. Zhao, Z. Ren, H. Fang, F. Robles-Hernandez, S. Baldelli and J. Bao, Nat. Nanotechnol., 2014, 9, 69–73 CrossRef CAS PubMed .
11. H. Kato, K. Asakura and A. Kudo, J. Am. Chem. Soc., 2003, 125, 3082–3089 CrossRef CAS PubMed .
12. R. Li, H. Han, F. Zhang, D. Wang and C. Li, Energy Environ. Sci., 2014, 7, 1369 CAS .
13. Q. Wang, Y. Li, T. Hisatomi, M. Nakabayashi, N. Shibata, J. Kubota and K. Domen, J. Catal., 2015, 328, 308–315 CrossRef CAS .
14. S. W. Kohl, L. Weiner, L. Schwartsburd, L. Konstantinovski, L. J. W. Shimon, Y. Ben-David, M. A. Iron and D. Milstein, Science, 2009, 324, 74–77 CrossRef CAS PubMed .
15. B. Rausch, M. D. Symes, G. Chisholm and L. Cronin, Science, 2014, 345, 1326–1330 CrossRef CAS PubMed .
16. B. Rausch, M. D. Symes and L. Cronin, J. Am. Chem. Soc., 2013, 135, 13656–13659 CrossRef CAS PubMed .
17. J. Barber, Curr. Opin. Struct. Biol., 2002, 12, 523–530 CrossRef CAS PubMed .
18. O. Yehezkeli, R. Tel-Vered, J. Wasserman, A. Trifonov, D. Michaeli, R. Nechushtai and I. Willner, Nat. Commun., 2012, 3, 742 CrossRef PubMed .
19. T. Suenobu, Y. Isaka, S. Shibata and S. Fukuzumi, Chem. Commun., 2015, 51, 1670–1672 RSC .
20. L. E. Heim, N. E. Schlörer, J.-H. Choi and M. H. G. Prechtl, Nat. Commun., 2014, 5, 3621 Search PubMed .
21. S. Kakuta and T. Abe, ACS Appl. Mater. Interfaces, 2009, 1, 2707–2710 CAS .
22. Y. Maenaka, T. Suenobu and S. Fukuzumi, Energy Environ. Sci., 2012, 5, 7360 CAS .
23. A. Boddien, D. Mellmann, F. Gartner, R. Jackstell, H. Junge, P. J. Dyson, G. Laurenczy, R. Ludwig and M. Beller, Science, 2011, 333, 1733–1736 CrossRef CAS PubMed .
24. S. N. Habisreutinger, L. Schmidt-Mende and J. K. Stolarczyk, Angew. Chem., Int. Ed., 2013, 52, 7372–7408 CrossRef CAS PubMed .
25. P. Usubharatana, D. McMartin, A. Veawab, P. Tontiwachwuthikul, F. Engineering, V. Uni and S. S. Canada, Ind. Eng. Chem. Res., 2006, 45, 2558–2568 CrossRef CAS .
26. Y. Liu, B. Huang, Y. Dai, X. Zhang, X. Qin, M. Jiang and M. H. Whangbo, Catal. Commun., 2009, 11, 210–213 CrossRef CAS .
27. J. Mao, T. Peng, X. Zhang, K. Li and L. Zan, Catal. Commun., 2012, 28, 38–41 CrossRef CAS .
28. G. Gao, Y. Jiao, E. R. Waclawik and A. Du, J. Am. Chem. Soc., 2016, 138, 6292–6297 CrossRef CAS PubMed .
29. W. Kim, G. Yuan, B. A. McClure and H. Frei, J. Am. Chem. Soc., 2014, 136, 11034–11042 CrossRef CAS PubMed .
30. T. Arai, S. Sato, T. Kajino and T. Morikawa, Energy Environ. Sci., 2013, 6, 1274–1282 CAS .
31. E. E. Barton, D. M. Rampulla and A. B. Bocarsly, J. Am. Chem. Soc., 2008, 130, 6342–6344 CrossRef CAS PubMed .
32. G. Seshadri, C. Lin and A. B. Bocarsly, J. Electroanal. Chem., 1994, 372, 145–150 CrossRef CAS .
33. C. Graves, S. D. Ebbesen, M. Mogensen and K. S. Lackner, Renewable Sustainable Energy Rev., 2011, 15, 1–23 CrossRef CAS .
34. E. Norkus, R. Pauliukaite, A. Vaskelis, E. Butkus, Z. Jusys and M. Kreneviciene, J. Chem. Res., 1998, 4, 320–321 RSC .
35. S. Cannizzaro, Justus Liebigs Ann. Chem., 1853, 88, 129–130 CrossRef CAS .
36. S. Kapoor, F. A. Barnabas, M. C. Sauer, D. Meisel and C. D. Jonah, J. Phys. Chem., 1995, 99, 6857–6863 CrossRef CAS .
37. S. Jimori, Z. Anorg. Allg. Chem., 1927, 167, 145–172 CrossRef CAS .
38. I. M. Kolthoff and E. A. Pearson, Ind. Eng. Chem., Anal. Ed., 1931, 3, 381–382 CrossRef CAS .
39. F. Amano, K. Nogami, R. Abe and B. Ohtani, J. Phys. Chem. C, 2008, 112, 9320–9326 CAS .

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6dt03658a

This journal is © The Royal Society of Chemistry 2017