Unravelling the pH-dependence of a molecular photocatalytic system for hydrogen production† †Electronic supplementary information (ESI) available. See DOI: 10.1039/c5sc01349f Click here for additional data file.

The electron-donating ability of the sacrificial agent and the protonation of the catalyst determine the optimum pH for hydrogen production.


Introduction
The photochemical production of H 2 from water is a rapidly expanding research eld that aims to store solar energy in a chemical fuel. 1 From the viewpoint of sustainability and economic viability, this proton reduction reaction should be carried out in aqueous conditions and use stable and Earth abundant materials. 2 Current investigations for solar H 2 synthesis include molecular dyes and electrocatalysts based on nickel, iron and cobalt, either in solution or immobilised onto the surface of a semiconductor. [3][4][5][6][7][8][9][10][11][12] These photocatalytic systems typically require the use of sacricial chemical reductants to provide the electrons to regenerate the oxidised dye following proton reduction.
The efficiency of H 2 evolving photo-and electrocatalytic systems is typically strongly pH dependent. [13][14][15][16] Understanding the origins of this pH dependence is critical to guide further system development and optimisation. In particular, it is essential to determine whether such pH dependencies derive from the availability of protons to the molecular catalyst, from the function of the molecular light-harvesting unit or from the sacricial electron donor.
We have recently reported a homogeneous photocatalytic system based on a molecular ruthenium photosensitiser (RuP) and a nickel catalyst (NiP) capable of producing H 2 in pure water with a quantum efficiency near 10% in the presence of ascorbic acid (AA) as a sacricial electron donor (Fig. 1). 17 In this system, the electron transfer from the photoreduced dye (RuP À ) to NiP takes place following reductive quenching of the photoexcited dye in the presence of the sacricial agent, AA (Scheme 1). Under visible light irradiation, optimum performance of this photocatalytic system was observed at pH 4.5. In contrast, when used as an electrocatalyst, the proton reduction efficiency of the NiP catalyst was observed to increase towards more acidic pH. 17 This pH dependence is typical of this type of nickel-based molecular electrocatalysts, and has been attributed to the presence of pendant amines with low pK a , which are thought to act as a proton relay between the solvent and the metal centre. 13,[18][19][20] Studies reporting the dependence of H 2 evolution on the acidity of the aqueous media for molecular photocatalytic systems have typically focused on the overall system efficiency as a function of pH. [13][14][15]21 Reaction mechanisms, where studied, have been addressed through theoretical calculations and experimental techniques such as nuclear magnetic resonance spectroscopy, electrochemistry and steady state spectroscopy; 7,22-24 and to a lesser extent, time-resolved absorption spectroscopy. 15,[24][25][26][27][28][29][30][31][32][33][34] Herein, we report on the inuence of the solution acidity on the formation of the photo-reduced RuP À species, the electron transfer kinetics between the optically active RuP À and NiP, as well as the pH dependence of H 2 evolution observed in electrochemical and bulk photocatalytic experiments. We have employed transient absorption spectroscopy, combined with electrochemical experiments, to determine the working principles of this photocatalytic system. The correlation of these results allowed us to determine the pHdependent rate-limiting steps in the photocatalytic system and give a rational explanation for the observed optimal activity at pH 4.5, as well as to provide a timescale for the electron transfer (ET) reactions between the sacricial electron donor, the dye and the catalyst. Experimental details are described in the ESI. †

Results and discussion
At pH 4.5, photoexcitation of RuP in the presence of AA leads to the efficient formation of RuP À within t 50% $ 250 ns through a reductive quenching mechanism, with a quantum yield estimated from transient emission studies of approximately 70%. 17 The reduced photosensitiser RuP À shows a transient absorption peak at l ¼ 500 nm with a lifetime (t 50% , calculations detailed in Fig. S1 †) of 500-700 ms (Fig. 2). 35 The yield of RuP À produced at different pH values can be determined from the initial amplitude (at $10 ms) of this RuP À transient absorption signal at l ¼ 500 nm. It is apparent (Fig. 2, inset) that this assay of the yield of RuP À increases with increasing pH, reaching a maximum at pH ¼ 5. This behaviour can be explained by the different reactivity of two protonation states of ascorbic acid present in the pH range studied herein. At low pH, ascorbic acid exists primarily in its undissociated form H 2 A (pK a ¼ 4.17), whereas the monoprotic ascorbate anion (HA À ) predominates at higher pH values (pK a ¼ 11.57). The ascorbate anion is a stronger reducing agent than its protonated form, and thus the reductive quenching of the excited dye, RuP*, is favoured at pH >4, where HA À is the dominating species. [36][37][38] Aer the formation of RuP À , electrons should be transferred from the reduced dye to the catalyst. In the presence of NiP, the positive transient absorption signal corresponding to RuP À absorption at l ¼ 500 nm is rapidly quenched (within 50-100 ms on the range of pH values studied herein), leading to the appearance of a negative signal at longer timescales (500 ms to 1 s; Fig. 3 and S2 †). This negative signal is assigned to electron transfer from RuP À to NiP, resulting in bleaching of ground state absorption of NiP. 17 This bleach is not observed in the absence of either RuP or NiP (see for example Fig. 2 and S3 †), suggesting that it is due to intermolecular electron transfer (ET) between RuP À and NiP (rather than the direct photoexcitation of NiP). The fast electron transfer kinetics between RuP À and NiP at all studied pH values suggests that this process is not limiting the catalytic activity of NiP (Fig. S2 †). However, the long-lived transient absorption bleach signal corresponding to reduced NiP indicates that the subsequent protonation step is more likely to be the rate limiting reaction. We can also estimate the yield of NiP reduced by RuP À from the amplitude of the bleach (Fig. 3 inset). Thus, a greater negative signal indicates the reduction of more NiP due to ET from RuP À . It is apparent  that the yield of reduced NiP increases as the pH is increased, reaching a maximum at pH ¼ 5. Fig. 4 compares the pH dependence of the 500 nm transient absorption bleach signal assigned to the yield of reduced NiP (blue circles) and the TOF NiP (H 2 ) per catalyst molecule of the system (red squares) determined from bulk photocatalysis experiments reported previously (see ESI †). 17 Also shown in Fig. 4 is the ratio of reduced NiP per RuP À (black triangles, calculations detailed in the ESI †). It is apparent that whilst both the TOF and yield of reduced NiP are strongly pH dependent, the ratio of reduced NiP/RuP À is independent of pH. Thus, our results suggest that the yield of reduction of NiP by RuP À is pH independent. In contrast, from pH 2 to 4.5, both the NiP reduction yield and the TOF NiP increase. As the efficiency of electron transfer from RuP À to NiP is pH independent, the increase in the yield of reduced NiP with higher pH can be assigned directly to the increased efficiency of RuP À formation due to the pH dependence of the electron donating function of the ascorbic acid as discussed above. It is also striking from Fig. 4 that at pH >4.5, the TOF NiP rapidly decreases despite the yield of reduced NiP remaining high. Such a sharp maximum in the pH dependence of TOF catalyst has also been observed in many other photocatalytic systems. 14,15,17,25,39,40 As the yield of reduced catalyst is approximately constant between pH 4.5 and 6, the drop on hydrogen generation towards neutral pH is strongly indicative of a decreasing activity in proton reduction catalysed by NiP. The exact catalytic mechanism for proton reduction using nickel bis(diphosphine) catalysts is still not fully elucidated, with little evidence of the catalytic intermediates in aqueous media. 41,42 Although protonation of the reduced Ni species may in principle occur at the pendant amines of the ligand or directly at the Ni metal centre, DFT calculations support protonation of the amines. 41 This agrees with the dependence of the electrocatalytic activity on acid concentration of bis(diphosphine) nickel electrocatalysts, which has been explained by the presence of pendant amines in the second coordination sphere. These amines with a relatively low pK a have been suggested to act as proton relays between the solvent and the metal centre. 13,18,19,43 Although these studies were mainly performed in pure organic solvents or aqueous-organic solvent mixtures in the presence of strong acids, the electrocatalytic proton reduction activity of NiP was observed to increase towards more acidic pH. 17 In this article, we detail the dependence of the catalytic activity of NiP on pH in pure water. In order to further investigate the drop in the H 2 production yield of the photocatalytic system towards neutral pH, the protonation state of NiP at different pH values was studied. The titration of NiP with NaOH (0.1 M) shows two equivalence points, at pH $5 and pH $9 (Fig. 5). In agreement with previous reports, these processes are assigned to the deprotonation of the pendant amines and the second deprotonation of the phosphonic acid groups, respectively. 20,44 The assignment of the deprotonation of the amines is further conrmed by the presence of only one equivalence point at pH $5 for the titration of an analogous bis(diphosphine) nickel complex where the phopsphonic acid substituents are protected with ethyl ester groups (NiP Et ) (Fig. S4 †). A pK a $3 is calculated from the Henderson-Hasselbach equation for the pendant amines in the ligand with an equivalence point at pH $5 (see ESI † for details), meaning that at pH >5, the amines are largely deprotonated. Since these amines are considered to play an important role as proton relays between the solvent and the nickel metal centre, 18,19 it is likely that, at less acidic media, the catalytic efficiency is limited by a poor degree of protonation of the pendant amines of the catalyst, which inhibits the ability of NiP to reduce protons to H 2 . It is worth noting that the photosensitiser employed in our studies contains phosphonic acid substituents. This dye was chosen for consistency and to allow for direct comparison with our previous studies. 17 The pK a  values of RuP have been reported to be $1 and $12, suggesting that the buffer capacity of RuP within the pH range employed in this study is limited. 45,46 Our results match well with the strong pH-dependencies reported with other proton reduction photocatalytic systems that employ either AA, triethanolamine (TEOA) or ethylenediaminetetraacetic acid (EDTA) as sacricial electron donors. 14,15 In acidic media, the sacricial electron donor molecules become protonated, resulting in a poor electrondonating ability due to the anodic shi of the reduction potential. 4,17 Hence our studies show that the optimum pH of active homogeneous proton reduction systems is a compromise between electron donating ability of the sacricial agent and the optimum working environment for the catalyst.

Conclusions
In summary, we have used transient absorption spectroscopy, combined with titration studies, electrochemistry and bulk photocatalytic experiments, to study the pH-dependence of the electron transfer reactions of a ruthenium-based photosensitiser and a nickel bis(diphosphine) catalyst for the production of H 2 under visible light irradiation. Our results suggest that the yield and kinetics of the electron transfer from the sensitiser to the catalyst are independent of the pH. However, at pH <4.5, the catalysis is limited by the number of RuP À molecules available to reduce the catalyst due to the poor reducing character of undissociated AA. In contrast, at less acidic pH, low TOF NiP (H 2 ) are observed despite the large concentration of RuP À molecules available to reduce NiP. Titration studies of NiP with NaOH show that at pH >5, the amines are largely deprotonated and electrochemical studies conrm the lower activity at such pH values. 17 Since these amines have been reported to play an important role as proton relays between the solvent and the nickel metal centre, it is likely that the catalytic efficiency is limited by the lack of protonated amines in the nickel catalyst. In the wider context, our studies suggest that the pH of photocatalytic systems using a sacricial agent has to be adjusted to match the pH at which the dye is effectively reduced by the sacricial electron donor and the pH at which the catalyst can be efficiently protonated. We have also demonstrated how transient absorption spectroscopy, bulk photocatalytic and titration studies and electrochemical experiments can be combined for a rational analysis of limiting factors in a homogeneous photocatalytic system.