Versatile Photocatalytic Systems for H2 Generation in Water Based on an Efficient DuBois-Type Nickel Catalyst

The generation of renewable H2 through an efficient photochemical route requires photoinduced electron transfer (ET) from a light harvester to an efficient electrocatalyst in water. Here, we report on a molecular H2 evolution catalyst (NiP) with a DuBois-type [Ni(P2R′N2R″)2]2+ core (P2R′N2R″ = bis(1,5-R′-diphospha-3,7-R″-diazacyclooctane), which contains an outer coordination sphere with phosphonic acid groups. The latter functionality allows for good solubility in water and immobilization on metal oxide semiconductors. Electrochemical studies confirm that NiP is a highly active electrocatalyst in aqueous electrolyte solution (overpotential of approximately 200 mV at pH 4.5 with a Faradaic yield of 85 ± 4%). Photocatalytic experiments and investigations on the ET kinetics were carried out in combination with a phosphonated Ru(II) tris(bipyridine) dye (RuP) in homogeneous and heterogeneous environments. Time-resolved luminescence and transient absorption spectroscopy studies confirmed that directed ET from RuP to NiP occurs efficiently in all systems on the nano- to microsecond time scale, through three distinct routes: reductive quenching of RuP in solution or on the surface of ZrO2 (“on particle” system) or oxidative quenching of RuP when the compounds were immobilized on TiO2 (“through particle” system). Our studies show that NiP can be used in a purely aqueous solution and on a semiconductor surface with a high degree of versatility. A high TOF of 460 ± 60 h–1 with a TON of 723 ± 171 for photocatalytic H2 generation with a molecular Ni catalyst in water and a photon-to-H2 quantum yield of approximately 10% were achieved for the homogeneous system.


* S Supporting Information
ABSTRACT: The generation of renewable H 2 through an efficient photochemical route requires photoinduced electron transfer (ET) from a light harvester to an efficient electrocatalyst in water. Here, we report on a molecular H 2 evolution catalyst (NiP) with a DuBois-type [Ni(P 2 R ′N 2 R ″) 2 ] 2+ core (P 2 R ′N 2 R ″ = bis(1,5-R′diphospha-3,7-R″-diazacyclooctane), which contains an outer coordination sphere with phosphonic acid groups. The latter functionality allows for good solubility in water and immobilization on metal oxide semiconductors. Electrochemical studies confirm that NiP is a highly active electrocatalyst in aqueous electrolyte solution (overpotential of approximately 200 mV at pH 4.5 with a Faradaic yield of 85 ± 4%). Photocatalytic experiments and investigations on the ET kinetics were carried out in combination with a phosphonated Ru(II) tris(bipyridine) dye (RuP) in homogeneous and heterogeneous environments. Time-resolved luminescence and transient absorption spectroscopy studies confirmed that directed ET from RuP to NiP occurs efficiently in all systems on the nano-to microsecond time scale, through three distinct routes: reductive quenching of RuP in solution or on the surface of ZrO 2 ("on particle" system) or oxidative quenching of RuP when the compounds were immobilized on TiO 2 ("through particle" system). Our studies show that NiP can be used in a purely aqueous solution and on a semiconductor surface with a high degree of versatility. A high TOF of 460 ± 60 h −1 with a TON of 723 ± 171 for photocatalytic H 2 generation with a molecular Ni catalyst in water and a photon-to-H 2 quantum yield of approximately 10% were achieved for the homogeneous system.

■ INTRODUCTION
The sunlight-driven generation of the energy carrier H 2 from water using earth-abundant materials is considered one of the key processes to generate more sustainable fuels in a postfossil age. 1 Synthetic first-row transition-metal complexes containing Co, 2 Fe, 3 and Ni 4 are under intense development as scalable alternatives to the benchmark H 2 evolution catalysts platinum 5 and hydrogenases. 6 Complexes with a catalytic bis(1,5-R′diphospha-3,7-R″-diazacyclooctane)nickel(II) core, [Ni-(P 2 R ′N 2 R ″) 2 ] 2+ , developed by DuBois and co-workers, emerged in the past decade as probably the most active synthetic 3d transition-metal electrocatalysts for proton reduction. 4b,c,7 The catalytic cycle has been studied experimentally 8 and computationally, 9 and the catalyst mimics important features of the hydrogenase active site. 4c,7c, 10 Photocatalytic H 2 generation requires the efficient coupling of an efficient proton reduction electrocatalyst with a lightharvesting component. In particular, the development and investigation of molecular H 2 evolution catalysts made of abundant elements that show stability and activity in water are of major interest. 11 Many reports on homogeneous photocatalytic systems for water reduction using Fe 3a,e,12 or Co 2b,13 complexes are available, but there are relatively few using Nibased catalysts. 4f−i, 14 Heterogenization of well-defined molecular catalysts on semiconductor surfaces is an emerging approach for photocatalytic H 2 generation. Systems that have been shown to have notable H 2 production efficiencies include cobaloximes immobilized on TiO 2 nanoparticles 13f−i or CdSe/ ZnS quantum dots, 13j and [FeFe]-hydrogenase mimics on CdTe 12a or ZnS 15 quantum dots. Efforts have also been made to integrate molecular catalysts on photoelectrodes, such as an [FeFe] complex on InP, 12c a cobaloxime catalyst on p-type GaP, 16 and Fe(dithiolato)(diphosphine) 17 or cobaloxime catalysts 18 on dye-sensitized NiO films. However, the frequent use of organic solvents, poor light-to-H 2 conversion efficiencies, and photoinstabilities are drawbacks in these systems.
Despite the promising properties of DuBois-type catalysts, they were previously only applied in homogeneous photocatalytic schemes in the presence of organic solvents or a biological matrix. 4g In this work, we report on a novel DuBois-type catalyst (NiP; Figure 1), which not only contains a highly electroactive 2+ core, but also is surrounded by an outer sphere containing four dangling phosphonic acid moieties. This outer-sphere feature provides the catalyst with good solubility in aqueous solutions and anchors for immobilization onto metal oxide surfaces. 21 Electrochemical studies of NiP in an aqueous electrolyte solution demonstrated sustained electrocatalytic activity in the absence of organic solvents. Subsequently, we studied NiP in aqueous photochemical systems with a phosphonated bipyridine-based Ru(II) photosensitizer (RuP; Figure 1) in both a homogeneous system and heterogeneous photocatalytic environments where RuP and NiP were immobilized on different metal oxide semiconductors.
The kinetics of charge separation and recombination were studied by time-correlated single photon counting (TC-SPC) and transient absorbance spectroscopy (TAS). This spectroscopic study reveals that distinct reaction mechanisms are possible in these remarkably versatile photochemical systems. In particular, we consider whether this coimmobilization strategy on metal oxide surfaces reduces the rapid electron/ hole recombination losses typically observed when such photosensitizers and catalysts are covalently attached via molecular linker groups. 22 The visible-light-driven H 2 generation of the different systems was also studied in bulk experiments and revealed that NiP can operate with a high photocatalytic turnover frequency for a Ni-based molecular proton reduction catalyst in purely aqueous systems.

■ EXPERIMENTAL SECTION
Materials and Methods. All synthetic procedures involving air-or moisture-sensitive materials were carried out under N 2 by using either a glovebox or Schlenk techniques. Chemicals for the synthetic part of this work were purchased from commercial suppliers and used without further purification. Solvents were dried using standard purification procedures under an N 2 atmosphere. Chemicals for analytical measurements were purchased in the highest available purity. TiO 2 nanoparticles (Aeroxide TiO 2 P25 particles; anatase/rutile (8/2) mixture, average particle size 21 nm) were a gift from Evonik Industries, and ZrO 2 nanoparticles (99.9%, 20−30 nm) were obtained from Skyspring Nanomaterials Inc. Nanostructured anatase TiO 2 and ZrO 2 films were prepared by the Doctor Blading technique from colloidal pastes as reported previously. 13f, 23 The films were annealed at 450°C for 30 min prior to use, and the anatase phase for TiO 2 was confirmed by X-ray diffraction studies after heating. The resulting film thicknesses, determined by profilometry (Tencor Instruments), were 4 μm. [Ni(P Ph 2 N PhCH 2 PO(OEt) 2 2 ) 2 ](BF 4 ) 2 (NiP Et ) 7a ( Figure S1, Supporting Information) and RuP 24 were prepared according to published procedures.
Physical Measurements. 1 H and 31 P NMR spectra were recorded on a Bruker 400 MHz spectrometer. 1 H NMR spectra were referenced to the solvent residual peaks as an internal reference, 25 and 31 P NMR spectra were referenced to an external standard (85% H 3 PO 4 in D 2 O). UV−vis spectra were recorded on a Varian Cary 50 UV−vis spectrophotometer using quartz glass cuvettes. High-resolution electrospray ionization mass spectra (HR-ESI-MS) were recorded on a Quattro LC spectrometer, and the theoretical and experimental isotope distributions were compared. Elemental analysis was carried out by the microanalytical laboratory in the Department of Chemistry of the University of Cambridge.
Synthesis of [Ni(P Ph 2 N PhCH 2 P(O)(OH) 2 2 ) 2 ]Br 2 ·HBr (NiP). Bromotrimethylsilane (0.25 mL, 1.86 mmol) was added dropwise to a deep red solution of NiP Et (250 mg, 0.143 mmol) in degassed dichloromethane (15 mL) under a N 2 atmosphere at room temperature. The color changed to deep purple, and the solution was stirred for 2 days at room temperature. The solvent was removed, followed by addition of deoxygenated MeOH (HPLC grade, 15 mL) to the dark solid residue and stirring under N 2 for 1 day at room temperature. The volume of the solvent was reduced to approximately one-third, and the product was precipitated with Et 2 O (∼25 mL). After the mixture was stirred for 30 min, the deep purple precipitate was filtered off under N 2 , washed with Et 2 O, and dried in vacuo. The solid was redissolved in MeOH, the solution was stirred for several minutes, and the solvent was evaporated under high vacuum to give a purple solid. Electrochemistry. Electrochemical measurements were performed on an IviumStat or Ivium CompactStat potentiostat under an argon atmosphere using a three-electrode configuration. A glassy-carbon disk (3 mm diameter) working electrode, a platinum-wire counter electrode, and a Ag/AgCl/KCl(sat) reference electrode were used for cyclic voltammetry (CV). All potentials were converted to the normal hydrogen electrode (NHE) by addition of +0.197 V. 26 Controlled-potential electrolysis (CPE) was carried out in a threenecked flask with a three-electrode setup using glassy-carbon rods as working and counter electrodes (surface area in solution ∼2 cm 2 ) and a Ag/AgCl/KCl(sat) reference electrode. A solution of NiP (0.18 mM) in ascorbic acid (AA, 0.1 M, pH 4.5) was purged with N 2 containing CH 4 (2%) as the internal gas chromatography (GC) standard, and a potential of −0.5 V vs NHE was applied during CPE for 2 h. The Faradaic yield was calculated from the amount of H 2 accumulated in the headspace, as measured by GC. CPE of catalystfree electrolyte at the same potential showed no production of H 2 .
Preparation of RuP-Sensitized TiO 2 and ZrO 2 Films Loaded with and without NiP for Spectroscopic Studies. To prepare TiO 2 and ZrO 2 films sensitized with RuP, an aqueous solution of RuP (10 μL of 4 μM) was spread onto the films (geometrical surface area 1.5 cm 2 ), and the solvent was dried in air for 30 min. For TiO 2 and ZrO 2 films cofunctionalized with RuP and NiP, first a solution of NiP (10 μL of 8 μM) in MeOH was spread onto the surface of the film, followed by drying for 30 min in air. Then, an aqueous RuP solution (10 μL of a 4 μM) was spread onto the surface of the film and was dried for an additional 30 min.
Spectroscopic Characterization. Spectroscopic measurements on homogeneous solutions and functionalized TiO 2 and ZrO 2 films were carried out in water or in aqueous AA solution (0.1 M) carefully adjusted to pH 4.5 with NaOH (0.1 M), unless otherwise stated. A fresh AA solution was prepared prior to any measurements. The aqueous solutions were purged with N 2 for 15 min prior to the measurements. The UV−vis and fluorescence spectra of the solutions and films were recorded using a quartz cuvette (1 cm path length) on a Perkin−Elmer Lambda 35 UV−vis spectrophotometer and a Horiba Jobin Yvon Fluorolog luminescence spectrophotometer, respectively. TC-SPC measurements were performed by using a Horiba Jobin Yvon TBX Fluorocube system. As the excitation source, a pulsed laser with 467 nm nominal wavelength at a repetition rate of 100 kHz was used. The photoluminescence intensity of RuP at λ em 650 nm was measured as a function of time after the excitation pulse, for a fixed data collection period to ensure matched densities of absorbed photons between samples (600 s). The instrument response was measured at the full width at half-maximum and showed typically a 200−250 ps value. All TC-SPC experiments undertaken on TiO 2 and ZrO 2 films were measured under aqueous conditions or a 0.1 M AA solution at pH 4.5 in air.
The microsecond to second transient absorption decays were measured using a Nd:YAG laser (Big Sky Laser Technologies Ultra CFR Nd:YAG laser system, 6 ns pulse width). The second and third harmonics of the laser (corresponding to 532 and 355 nm, respectively) were used to excite RuP. The laser intensity was adjusted using neutral density filters as appropriate, with experiments typically employing 350 μJ cm −2 , and the frequency of the laser pulse was fixed to 1 Hz. A liquid light guide with a diameter of 0.5 cm was used to transmit the laser pulse to the sample. The probe light source was a 100 W Bentham IL1 tungsten lamp, and the probing wavelength was selected by using a monochromator (OBB-2001, Photon Technology International) placed prior to the sample. Several highpass, low-pass and neutral-density filters (Comar Optics) were used to decrease the light arriving to the detector. Transient absorption data were collected with a Si photodiode (Hamamatsu S3071). The information was passed through an amplifier box (Costronics) and recorded using a Tektronics TDS 2012c oscilloscope (microsecond to millisecond time scale) and a National Instruments (NI USB-6211) DAQ card (millisecond to second time scale). The decays observed were the average of 500 laser pulses. The data were processed using home-built software based on Labview.
Photocatalytic H 2 Evolution Experiments. All photocatalytic experiments were carried out using a Solar Light Simulator (Newport Oriel, 100 mW cm −2 ) equipped with an air mass 1.5 global filter (AM 1.5G). UV irradiation was filtered using a 420 nm cutoff filter (UQG optics), and IR irradiation was filtered by a water filter (path length 10 cm). The photoreactor was held at a constant temperature of 25°C in all experiments. Samples were generally prepared in air protected from light by an Al foil. The reaction vessel was sealed with a rubber septum, and air was replaced by N 2 containing 2% CH 4 (internal GC standard). The irradiated cross section of the solution in the vials was approximately 3.3 cm 2 . H 2 evolution was monitored by GC measurements with an Agilent 7890A Series GC equipped with a 5 Å molecular sieve column. The GC oven temperature was kept constant at 45°C, N 2 was used as a carrier gas at an approximate flow rate of 3 mL min −1 , and a thermal conductivity detector (TCD) was used. Samples (15−20 μL) for headspace gas analysis were taken from the reaction vessel headspace in 30 min intervals for 2 h in the screening experiments. For long-term measurements longer intervals were chosen. The response factor of the thermal conductivity detector for H 2 compared to CH 4 was 1.91.
Preparation of Homogeneous System RuP-NiP. In a typical experiment, a freshly prepared AA solution was titrated to the desired pH using NaOH Immobilization of RuP and NiP on TiO 2 and ZrO 2 : RuP-TiO 2 -NiP and RuP-ZrO 2 -NiP. AA (0.1 M) was prepared as described above. ZrO 2 or TiO 2 nanoparticles (2.5 mg) were placed in a vial containing an AA solution and dispersed by sonication for 5 min. To load the particles, first NiP was added and after stirring for several min RuP was added, resulting in a final volume of 2.25 mL, leaving 5.59 mL of gas headspace in the vial.
Quantification of Catalyst Loading on the Nanoparticles by UV−Vis Spectrophotometry. The adsorption of NiP on TiO 2 and ZrO 2 nanoparticles was quantified by recording the difference UV−vis spectrum of a solution of NiP in AA (0.1 M, pH 4.5, 2.25 mL) before and after exposure to TiO 2 or ZrO 2 nanoparticles (2.5 mg). The solution was stirred with the nanoparticles for 30 min, followed by centrifugation (7000 rpm, 5 min) between the measurements. Adsorption of RuP on nanoparticles after coadsorption with NiP was estimated by adding RuP (0.05 μmol) to a suspension of NiPsensitized TiO 2 or ZrO 2 particles (0.02 or 0.1 μmol NiP loading on 2.5 mg nanoparticles). The suspensions were purged with N 2 during catalyst loading. Quantitative adsorption of 0.05 μmol RuP and 0.02 μmol NiP would result in an estimated surface coverage of approximately 400 RuP and 150 NiP molecules per nanoparticle and approximately 45% coverage of the TiO 2 surface (see the Supporting Information).
Determination of Photon to H 2 Quantum Efficiency. The external quantum (photon to H 2 conversion) efficiency (EQE) was determined by an LED light source (Modulight, Ivium) using blue light (λ 460 nm, 5 mW cm −2 ). The light intensity was measured with a Newport thermopile detector (818P-020-12) coupled with an optical power meter (1916-R). Samples of RuP (0.3 μmol) and NiP (0.1 μmol) in solution were used. Aliquots of headspace gas were subjected to GC analysis during irradiation. The efficiency was determined from the amount of H 2 produced after 2 h of irradiation using the equation Treatment of Data. All analytical measurements were repeated at least three times. The obtained data were treated as follows: for a sample of n observations x i , the unweighted mean value x 0 and the standard deviation σ were calculated using the equations A minimum σ of 10% was assumed in all experiments. The light sources and gas chromatographs were calibrated regularly to ensure reproducibility throughout all experiments.

■ RESULTS AND DISCUSSION
Synthesis and Characterization of NiP. The catalyst NiP was synthesized through hydrolysis of the octaethyl phosphonate ester analogue NiP Et 7a by dealkylation of the ethyl ester groups with trimethylsilyl bromide in dichloromethane 27 ( Figure S1, Supporting Information). NiP was obtained in 81% yield and characterized by 1 H and 31 P NMR spectroscopy, high-resolution mass spectrometry, and elemental analysis for C, H, N, P, and Br. NiP is soluble in water (∼0.3 mg mL −1 ) and dissolves well in an aqueous ascorbic acid solution (0.1 M, pH 4.5) with a solubility of more than 5 mg mL −1 . NiP therefore is an example of a DuBois-type Ni catalyst that is soluble in aqueous solutions and can be employed homogeneously in organic solvent free aqueous catalytic systems.
Electrocatalytic Activity of NiP in Aqueous Solution. Before studying NiP in a purely aqueous electrolyte solution, we compared NiP with the previously studied complex NiP Et Journal of the American Chemical Society  (Figure 2A and Figures S2 and S3 (Supporting Information)). 4g Addition of triflic acid induced a catalytic current attributed to proton reduction below w s pH 28 ∼4.9 following the reduction of Ni II to Ni I with an onset potential of −0.3 V vs NHE ( Figure  2A). The small cathodic shift in potential of the reduction wave upon acidification is presumably due to protonation of the dangling phosphonate moieties. Thus, the [Ni(P 2 R ′N 2 R ″) 2 ] 2+ core and electrocatalytic proton reduction activity are intact in NiP and the electrochemical response is indeed comparable to that of NiP Et in the presence of an organic acid ( Figures S2−  S4, Supporting Information).
Electrocatalytic proton reduction by NiP was also observed in the absence of organic solvent in an aqueous solution buffered with ascorbic acid (AA) or Na citrate (0.1 M, pH 4.5). A single, irreversible reduction wave (a) was observed at E p = −0.35 V vs. NHE, followed by the onset of a catalytic wave (b) at approximately −0.48 V vs NHE ( Figure 2B and Figure S5 (Supporting Information)). Therefore, the catalyst operates with a small overpotential requirement of approximately 0.2 V in comparison to the thermodynamic potential for proton reduction of −0.27 V vs NHE at pH 4.5. The reduction potentials of intermediates of RuP, which are likely to be formed during the quenching process after photoexcitation (see below) are also indicated in Figure 2B. The position of these potentials and the conduction band edge of TiO 2 suggest that enough driving force for photocatalytic proton reduction with NiP would be available when using RuP and RuP-sensitized TiO 2 as a light harvester, as described in detail in the kinetic and photocatalytic sections.
Controlled-potential electrolysis with NiP (0.18 mM) in an aqueous ascorbic acid solution (0.1 M, pH 4.5) on a glassycarbon-rod working electrode (surface area ∼2 cm 2 ) at −0.5 V vs NHE for 2 h confirmed the generation of H 2 gas with a Faradaic yield of 85 ± 4% (H 2 in the headspace quantified by GC). Rinsing the electrode with H 2 O after CPE and immersing it in a fresh electrolyte solution (in the absence of NiP catalyst) did not result in the formation of H 2 when continuing with CPE at the same potential. Thus, H 2 production originates from a dissolved catalyst and not from electrodeposited decomposition products on the electrode. Thus, the watersoluble NiP displays electroactivity for the reduction of aqueous protons with a high Faradaic yield in the absence of organic solvents, a prerequisite for the use of NiP in photocatalytic schemes in water.
Kinetics and Mechanisms of ET in Photocatalytic Schemes. The suitability of NiP in photocatalytic H 2 generation was studied with RuP, 24,29 AA (0.1 M) as a sacrificial electron donor, and buffer (pK a 4.17) in aqueous solution. 30 Photoexcitation of RuP to RuP* (λ max 455 nm, MLCT) can result in either oxidative or reductive quenching of the photoexcited state (Figures 3 and 4). 31 First, the kinetics and mechanisms of ET between RuP and NiP in homogeneous aqueous solution (RuP-NiP system) and anchored onto metal   Figure 3, and the mechanistic and kinetic details are summarized in Figure 4 and described below. The RuP-NiP and RuP-ZrO 2 -NiP systems generate H 2 through a reductive quenching mechanism of RuP in solution or "on the particle", respectively. RuP-TiO 2 -NiP undergoes oxidative quenching and electrons are transferred "through the particle".
TiO 2 was observed to cause oxidative quenching of the immobilized RuP*, resulting in the oxidized intermediate RuP + (E(RuP + /RuP*) = −0.95 V vs NHE) 32 by electron injection into the conduction band of TiO 2 (E CB = −0.55 V vs NHE at pH 4.5). 33 These kinetics were measured by TC-SPC to take place in approximately 180 ps, with an injection efficiency of >95% ( Figure 5). 13f The transient absorption spectrum of a photoexcited RuP-TiO 2 film shows a maximum transient absorption peak at 700 nm, corresponding to the absorption spectrum of the oxidized RuP + ( Figure S6, Supporting Information). In the absence of AA, the TiO 2 conduction band electrons recombine with RuP + within approximately 1 ms. Upon reduction of RuP + by AA (t 50% ≈ 50 μs), corresponding to the regeneration of RuP, the resulting photoinjected TiO 2 electrons exhibit a lifetime of 0.5 s. Following the codeposition of NiP (molecular ratio RuP/ NiP of 1/2), the decay of these electrons is accelerated to approximately 1 ms, assigned to ET to NiP ( Figure S7, Supporting Information).
The oxidative ET mechanism from RuP to NiP through the semiconductor ("through particle" system, Figure 3C), is possible with TiO 2 but not with ZrO 2 , due to the energetic mismatch between RuP* and the conduction band of ZrO 2 (E CB = −1.26 V vs NHE at pH 4.5). 34 ZrO 2 is therefore unable to accept electrons from RuP* and can only be used as a matrix to immobilize the compounds in close proximity on the particle ("on particle" system, Figure 3B).
The reductive quenching through intermolecular ET from AA (E = 1.17 V vs NHE) 35 to RuP* (E(RuP*/RuP − ) = 1.08 V vs NHE) 31 was measured by TC-SPC to take place in approximately 250 ns with an estimated efficiency of 70% in homogeneous RuP-NiP and heterogenized "on particle" RuP-  37 This reduced state of the dye, RuP − , is a strong reducing agent and has a large driving force for the reduction of NiP (E(RuP/RuP − ) = −1.09 V vs NHE, Figure 2). 31 The addition of increasing amounts of NiP (from 0 to 16 μM) to a solution containing RuP (4 μM) and AA (0.1 M, pH 4.5) results in the linear decrease of the lifetime of RuP − from 700 to 29 μs, following first-order kinetics with respect to NiP concentration for the intermolecular ET between RuP − and NiP (Figure 6). The second-order rate constant of this ET is k ET = 1.4 × 10 9 M −1 s −1 , indicating that the ET kinetics are diffusion limited ( Figure 6B and Table S1 (Supporting Information)). In addition, the presence of NiP resulted in the appearance of a long-lived (t 50% ≈ 0.1 s) bleaching signal assigned to the reduction of Ni II species ( Figure S9    are ongoing to monitor the nickel catalytic species involved in the H + reduction reaction. The reductive quenching of RuP* also occurs through intermolecular ET from AA to RuP* in RuP-ZrO 2 -NiP. The photoluminescence intensity and lifetime of RuP* on nanostructured ZrO 2 films decrease upon the addition of an aqueous AA solution (0.1 M, pH 4.5), and TC-SPC measurements reveal that reductive quenching of RuP* to form RuP − occurs in approximately 250 ns with an estimated ET efficiency of 70% ( Figure 5B). The cofunctionalization of ZrO 2 with NiP (8 μM) and RuP (4 μM) in water does not affect the luminescence of RuP, indicating that ET between the two molecules requires the formation of RuP − in the presence of a sacrificial electron donor. A transient absorption peak at 500 nm corresponding to the signal of RuP − was also observed in the RuP-ZrO 2 system in the presence of AA (0.1 M at pH 4.5), with the decay of RuP − accelerating following the codeposition of NiP ( Figure   S10 (Supporting Information)). It can be concluded that the same reductive ET mechanism and similar kinetics (dependent upon component loadings/concentrations) take place both in homogeneous media and on anchoring of the molecules onto the surface of a material that does not allow for dye electron injection. Although the reductive ET mechanism is the same for both homogeneous and the "on-particle" ZrO 2 systems, charge separation is achieved through two different processes: while in homogeneous media charge separation occurs through the diffusion of molecules into the solution, when the dye and catalyst are immobilized in RuP-ZrO 2 -NiP charge separation can take place through the intermolecular ET between neighboring molecules. 38 Standard Conditions for Photocatalytic H 2 Evolution. The spectroscopic studies revealed that all three photocatalytic systems drive the efficient photoreduction of NiP. We subsequently studied photocatalytic generation with NiP in bulk experiments. All three systems were indeed efficient in producing H 2 during irradiation with visible light. Consequently, the systems were studied and optimized by varying the pH value of the AA solution and concentrations of RuP and NiP (Table 1 and Tables S2−S4 (Supporting Information)). The photocatalytic activity and longevity of H 2 production was dependent on the conditions and type of system employed. The photocatalytic performance of NiP generally varied in all three environments, and the rate of photogenerated H 2 was constant over at least 1 h in all experiments ( Figure 7A). The turnover frequencies based on NiP (TOF NiP ) were calculated from the amount of H 2 accumulated in a photoreactor after 1 h irradiation (Tables S3 and S4).
Photo-H 2 Generation with Homogeneous RuP-NiP. In a homogeneous RuP-NiP system, a TOF NiP value of 460 ± 60 h −1 was observed when RuP (0.05 μmol) was used with a low amount of NiP (0.02 μmol) in aqueous AA (0.1 M, 2.25 mL). This RuP-NiP system was photoactive for 2 h with a final TON NiP value of 723 ± 171 (15 ± 3 μmol of H 2 ). An increasing amount of NiP (0.1 μmol) resulted in a decreased initial TOF NiP concomitant with an increased system lifetime and a comparable overall TON NiP value of 651 ± 30 (65 ± 3 μmol of H 2 ) after 30 h irradiation (Table 1 and Figure 7B). This observation might be explained by the less likely double reduction of a single molecular catalyst at high NiP concentrations. Addition of RuP or AA to a photodegraded system did not result in reactivation, suggesting that decomposition of NiP occurred after approximately 700 TONs. Addition of NiP to a deactivated homogeneous system did result in reactivation, but full photoactivity was not restored, indicating that photodegradation affected not only NiP but also other system components such as RuP. Indeed, photobleaching of RuP became evident after 1 h irradiation by recording an electronic absorption spectrum of the homogeneous solution after irradiation in a gastight quartz cuvette under standard conditions ( Figure S11 (Supporting Information)).
Several sets of control experiments were carried out, which showed that no or only negligible amounts of H 2 were produced in the absence of RuP, NiP, AA or light. The presence of additional buffers such as citrate and acetate did not impede the photocatalytic activity of the original system. Replacement of NiP by different Ni salts such as NiCl 2 and NiBr 2 in combination with 4 equiv of a water-soluble phosphine ([2-(dicyclohexylphosphino)ethyl]trimethylammonium chloride) resulted only in negligible amounts of H 2 . Homogeneous  (Table S1 (Supporting Information)). See the caption to Table S1 for details of the analysis. systems using commercially available dyes such as [Ru(bipy) 3 ]Cl 2 (TOF NiP = 220 ± 20 h −1 ) or the organic, noble-metal-free dye Eosin Y (used as the disodium salt; TOF NiP = 189 ± 31 h −1 ) also resulted in efficient photocatalytic H 2 production in an aqueous AA solution (0.1 M, pH 4.5) ( Table 1 and Table S6 (Supporting Information)).
Photon-to-H 2 Conversion Efficiency. The external quantum efficiency (EQE) of the homogeneous photocatalytic system with RuP (0.3 μmol) and NiP (0.1 μmol) in AA (0.1 M, pH 4.5) was measured using an LED light source (λ 460 nm, 5 mW cm −2 ). An EQE value of 9.7 ± 1.2% was determined after 2 h irradiation. We note that this external yield assumes that all photons emitted by the light source were absorbed by RuP and therefore represents a lower estimate of the quantum efficiency of the system. The photon-to-H 2 quantum yield observed for our homogeneous RuP-NiP system is remarkable in comparison to other photocatalytic systems with molecular 3d transition-metal catalysts operating homogeneously or immobilized in aqueous solution. Recently reported quantum yields for homogeneous photocatalytic systems ranged from 0.23 to 0.6% for a Co-pentapyridine catalyst with [Ru(bpy) 3 ] 2+ as photosensitizer 39 in an aqueous system at neutral pH to 4.6% for a cobaloxime-based H 2 generation system with an Alporphyrin dye in a water/acetone solvent system. 13k In systems with a molecular catalyst immobilized on a solid-state material, EQEs of approximately 1−1.5% for a cobaloxime immobilized on RuP-sensitized TiO 2 13f and a Ni(TEOA) 3 2+ (TEOA = triethanolamine) complex on graphitic carbon nitride, 40 respectively, were reported.
EQEs for self-assembled photocatalytic systems using Ni salts and 2-mercaptoethanol (24.5%) 4f or dihydrolipoic acid (36% QE) 14 ligands are still benchmarks; however, a defined catalytically active species has not been reported for these systems.
Benchmark for Synthetic Photochemical Systems. Various homogeneous photocatalytic schemes have been reported for the reduction of aqueous protons with molecular catalysts. Examples of efficient photo-H 2 generating systems include a molecular dye combined with [Ni(P 2 R ′N 2 R ″) 2 ] 2+type 4g and Fe-based molecular catalysts, 3c,15 but in these systems water/organic solvent mixtures were employed. Photocatalytic schemes for H 2 generation which operate in pure aqueous systems are highly desirable. 11 Very recently, the water-insoluble [Ni(P 2 Ph N 2 Ph ) 2 ](BF 4 ) 2 was incorporated into the photosynthetic protein Photosystem I (PSI) for photocatalytic H 2 production in aqueous solution, 19b but only a very limited amount of Ni catalyst can be loaded onto dilute enzyme systems and the photostability of the hybrid assembly remained an issue (more than half of the photoactivity ceased after 30 min of irradiation).
Nickel-thiolate 4i and cobalt-dithiolene 13l catalysts have been reported to achieve several thousand turnovers in organic solvent/water mixtures. Cobaloximes have been reported to evolve H 2 in a homogeneous system 13g and on RuP-sensitized TiO 2 in water, 13f but the TOF Co value never exceeded 20 h −1 and the photon-to-H 2 efficiency was only 1%. The higher photocatalytic activity and EQE of NiP in comparison to those of cobaloxime catalysts may be related to the similar reduction  potentials required for both reduction steps in NiP. We have previously shown that the photocatalytic activity of a cobaloxime complex is significantly limited by its relatively slow second reduction kinetics, attributed at least in part to the relatively unfavorable energetics for this second reduction. 41 The use of a Co-pentapyridine catalyst 39 and a [FeFe]hydrogenase mimic 12a in photocatalytic schemes in an aqueous solution resulted in TOFs of approximately 20 and 50 h −1 , respectively. Thus, our RuP-NiP system compares favorably with previously reported systems in organic solvent free aqueous solution.
Photo-H 2 Generation with Heterogenized Catalysts. The immobilization of a molecular catalyst on a photoelectrode is a prerequisite for fuel generation in a photoelectrochemical device, and the attachment on a semiconductor material is particularly desirable. Therefore, we extended our studies using NiP for H 2 generation in heterogeneous photocatalytic assemblies with nanoparticle suspensions as a first step towards an electrode assembly.
We first determined the maximum loading capacity of RuP 6c and NiP on the metal oxide particles by spectrophotometry. Approximately 0.05 μmol of RuP or NiP can be immobilized per milligram of TiO 2 or ZrO 2 when adding an excess of phosphonated catalyst (see the Supporting Information). 6c,13f The metal oxide nanoparticles were loaded by the following procedure: NiP was added to a suspension of TiO 2 or ZrO 2 in aqueous AA solution (2.5 mg in 2.25 mL), and then RuP was added. NiP was loaded first due to the optimized geometry of adsorption of phosphonated bipyridine ligands of RuP. 29b The photoactivity of the suspensions was studied under irradiation with visible light (AM1.5G, 100 mW cm −2 , λ >420 nm). Upon investigating the heterogeneous photocatalytic systems, notably different trends in performance were observed in comparison to the homogeneous system at lower NiP loadings (Figure 8).
In a RuP-ZrO 2 -NiP system, a low amount of NiP (0.02 μmol) with RuP (0.05 μmol) on 2.5 mg of ZrO 2 resulted in a TOF NiP value of 27 ± 3 h −1 . The results obtained in TAS measurements suggest that direct interaction of the quenched dye (RuP − ) with NiP is required to drive the reaction. ET between RuP and NiP "on the particle" as observed by the spectroscopic studies above becomes difficult under such dilute conditions due to the spatial separation of the compounds on the ZrO 2 surface. When the amount of NiP added to ZrO 2 is increased to 0.1 μmol, a significant enhancement in TOF NiP to 92 ± 26 h −1 was observed ( Figure 8 and Table S4 (Supporting Information)).
We note that, at low concentrations of compounds carrying phosphonic acid groups, attachment on ZrO 2 or TiO 2 is not quantitative, presumably due to competitive binding of AA to metal oxides. 42 Spectrophotometric studies indicate that NiP is almost quantitatively (>80%) adsorbed on ZrO 2 , whereas only approximately 10% of RuP adsorbs on NiP-modified ZrO 2 (Table S7 (Supporting Information)). When the ZrO 2 nanoparticles were loaded with NiP (0.02 μmol) and RuP (0.05 μmol), separated by centrifugation, and redispersed in fresh AA (0.1 M, pH 4.5), the amount of H 2 produced was 0.16 μmol after 1 h irradiation (in comparison to 0.54 μmol of H 2 before centrifugation). Thus, the photodriven H 2 production in bulk experiments can best be described as a mixture of ET between surface-immobilized catalysts through an "on particle" mechanism 38 and from solubilized RuP − to surface-bound NiP.
A RuP-TiO 2 -NiP system displays a TOF NiP value of 51 ± 7 h −1 at a low NiP loading of 0.02 μmol on 2.5 mg of TiO 2 ( Table 1). The rate of H 2 production reached a maximum with a NiP loading of 0.1 μmol, whereupon a TOF NiP value of 72 ± 5 h −1 and an overall TON NiP value of 278 ± 19 h −1 (after 30 h) was obtained (Table S5 (Supporting Information)).
Under these conditions, at least 80% of NiP and more than 20% of RuP are attached on TiO 2 , as measured by spectrophotometry (Figures S12 and S13 (Supporting Information)). Once NiP or RuP is bound to the TiO 2 surface, it cannot easily be removed from the solid-state  material. Redispersion of loaded particles in a fresh AA solution did not result in the detection of significant amounts of RuP or NiP in solution. Loading the TiO 2 nanoparticles with NiP (0.02 μmol) and RuP (0.05 μmol), centrifugation, and resuspension in fresh AA (0.1 M, pH 4.5) resulted in a TOF NiP value of 24 h −1 during irradiation. Thus, 50% of the photocatalytic activity remained, thereby establishing the importance of the role of the conduction band and the "through particle" mechanism observed in the spectroscopic study. This experiment and the significantly higher rate of photocatalytic H 2 production for RuP-TiO 2 -NiP in comparison to RuP-ZrO 2 -NiP at very low loading of the particles with NiP (Table 1) support a preferential "through particle" ET mechanism for TiO 2 : a mechanism which does not require the direct electronic communication of RuP and NiP as needed on ZrO 2 .
These results show that photocatalytic H 2 production can be achieved with a DuBois-type catalyst attached on a solid-state semiconductor. However, to this point limitations by the loading capacity of materials used and by competitive binding of electron donor or electrolyte are still being faced when investigating such hybrid materials. Work is in progress to overcome these limitations by investigating the binding modes of the molecular components and photocatalytic activity of RuP-NiP systems on thin films and electrode materials.

■ CONCLUSIONS
In summary, we describe a novel [Ni(P R ′ 2 N R ″ 2 ) 2 ] 2+ -type H 2 evolution catalyst (NiP), which is soluble in water and can be immobilized on metal oxide surfaces. NiP is electrocatalytically active in organic solvent free aqueous solution and evolves H 2 with an onset potential of only −0.48 V vs NHE under mild conditions (pH 4.5). Photocatalytic and spectroscopic studies were performed with NiP in three different systems in a purely aqueous solution containing AA. A homogeneous RuP-NiP system operates through reductive quenching of RuP* in solution. The heterogeneous RuP-ZrO 2 -NiP system shows the same ET mechanism, and ZrO 2 acts merely as a matrix to retain the attached molecules closely together, hence facilitating ET "on the particle". In RuP-TiO 2 -NiP, ET occurs via a "through particle" mechanism, where RuP* is oxidatively quenched upon injection of an electron into the conduction band of TiO 2 , which can subsequently be harvested by NiP.
A high TOF value of 460 ± 60 h −1 for light-driven H 2 evolution with a molecular 3d transition metal catalyst in pure aqueous solution was obtained, with TONs of approximately 700 for NiP. Advanced spectroscopic methods (TC-SPC and TAS) confirmed that directed ET from RuP to NiP occurs efficiently in all systems on the nano-to microsecond time scale. Losses due to charge recombination are minimized, as ET occurs efficiently within the lifetimes of the excited species. The highly efficient ET from dye to proton reduction catalyst is also reflected in the high photon to H 2 quantum yield of the homogeneous system of almost 10% in the presence of the sacrificial electron donor AA. Work is in progress to assemble a photoelectrode with NiP for use in a photoelectrochemical water splitting cell.

* S Supporting Information
Text and tables giving experimental data from kinetic and photocatalytic experiments and additional figures detailing cyclic voltammetry, kinetic measurements, and photocatalytic experiments. This material is available free of charge via the Internet at http://pubs.acs.org.

Notes
The authors declare no competing financial interest.

■ ACKNOWLEDGMENTS
The work at Cambridge was supported by the Christian Doppler Research Association (Austrian Federal Ministry of Economy, Family and Youth and National Foundation for Research, Technology and Development), the OMV Group, and the EPSRC (EP/H00338X/2). Imperial College London was supported by the ERC (project Intersolar), the Spanish Ministry of Education, and the European Commission Marie Curie CIG. This work was also supported by the "COST Action CM1202". The authors thank Dr. Xiaoe Li for assistance in TiO 2 and ZrO 2 film preparation and Ms. Bettina Basel for some help with the experimental part of the work.