Solar H2 evolution in water with modified diketopyrrolopyrrole dyes immobilised on molecular Co and Ni catalyst–TiO2 hybrids† †Electronic supplementary information (ESI) available: Experimental details, synthetic procedures, additional tables and figures. See DOI: 10.1039/c6sc05219c Click here for additional data file.

A series of diketopyrrolopyrrole (DPP) dyes with a terminal phosphonic acid group for attachment to metal oxide surfaces were synthesised and the effect of side chain modification on their properties investigated.


Preparation of mesoporous ITO electrodes
ITO-coated glass slides (1 x 2 cm 2 ) were cleaned by immersing them in a solution of distilled water, ammonia (25%) and hydrogen peroxide (30% w/v) in the ratio of (5:1:1 v/v) and heating at 70 °C for 30 min. The ITO-coated glass slides were sonicated in distilled water and dried at room temperature. An ITO suspension (20 wt% of ITO nanopowder in 5 M acetic acid in ethanol) was doctor-bladed onto the cleaned ITO glass slides with a circular area (0.28 cm 2 ) using Scotch® tape (3M) as spacers to prepare the mesostructured coating. The resulting ITO slides with the mesoporous ITO film (thickness approximately 3 μm) 6 were left to dry in air before removing the tapes and then annealed using a Carbolite furnace under atmospheric conditions using the following temperature program: heating from 25 °C to 350 °C (5 °C min -1 ), holding at 350 °C for 20 min before slowly cooling down to room temperature in the furnace chamber. The mesoporous ITO electrodes were cleaned using an ozone cleaner (BioForce Nanoscience) for 15 min before soaking them into 1 mM solution of each dye in THF (except for RuP, 1 mM in H2O) for 16 h. Subsequently, the electrodes were washed with ethanol and were kept in the dark before use.

Preparation of dye-sensitised TiO 2 films
Glass slides (2 x 2 cm 2 ) were cleaned by successively immersing the electrodes for 10 min in a solution of distilled soapy water, distilled water and then ethanol while sonicating. TiO 2 paste was slot-coated onto glass with a rectangular area (1 x 2 cm) using Scotch ® tape (3M) as spacers. After removing the tape, the slides were sintered using a Carbolite furnace under atmospheric conditions using a previously published heating ramp. 7 The thickness of the resulting TiO 2 layers was estimated around 6 m by scanning electron microscopy. The dyes were immobilised by soaking the electrodes into a DPP solution in THF (0.25 mM) or a RuP solution in H2O (0.25 mM) for 16 h.
Subsequently, the electrodes were washed with THF or H 2 O and kept in the dark before use.

Electrochemical measurements
A three-electrode electrochemical setup was used to determine the redox potential for the DPP and RuP dyes. Tetrabutylammonium tetrafluoroborate (TBABF 4 , 0.1 M, ≥99.0%, Sigma, recrystallised from water and dried overnight at 80 °C under vacuum) in acetonitrile was used as an electrolyte solution and was purged with N 2 for 15 min before each experiment. The cell employed a dye-sensitised mesoporous ITO working, a platinum mesh counter and a silver wire coated with AgCl as a pseudo-reference S4 electrode. The reference potential was calibrated by adding ferrocene (Fc) as internal standard at the end of the experiment. The applied potentials were subsequently referenced versus NHE by addition of 0.63 V (Fc + /Fc = +0.63 V vs NHE). 8 All cyclic voltammetry experiments were performed at room temperature at a scan rate of 50 mV s −1 using an IviumStat potentiostat. Quantification of DPP1 and DPP4 loading on TiO 2 nanoparticles 2.5 mg of TiO 2 nanoparticles were dispersed in a TEOA or AA SED solution via sonication for 10 min, followed by addition of DPP1 or DPP4 (0.05 or 0.25 μmol; 1 mM in THF). The mixture (total volume of 3 mL) was stirred for 10 min, centrifuged (8000 rpm, 10 min) and the UV-vis spectra of the supernatant recorded after passing it through a syringe filter (0.2 μm membrane). The amount of dye attached to TiO 2 was determined by comparing the UV-vis spectrum of the dye before and after addition of TiO 2 .

Photocatalytic experiments
TiO 2 (2.5 mg) nanoparticles were dispersed in a SED solution (2.95 mL-V cat ) in the photoreactor via sonication for 10 min and a solution containing the H 2 evolution catalyst was subsequently added (0.01-0.2 μmol, 10-200 μL (= V cat ) of a 1 mM CoP or NiP solution in H 2 O or methanol, respectively). After stirring the resulting suspension for 10 min in air, RuP (0.05 μmol, 50 μL of a 1 mM solution in water H 2 O) or DPP (0.05 μmol, 50 μL of a 1 mM solution in THF) was added. In the case of TiO 2 -Pt, pre-platinised TiO 2 (2.5 mg) was dispersed in a SED solution (2.95 mL) in the photoreactor via sonication for 10 min. RuP The photoreactor was sealed, kept in the dark after addition of the dye and purged with N 2 containing 2% methane (as internal standard for gas chromatography analysis) for 10 min. The total volume of the photocatalyst suspension was 3 mL leaving a gas headspace S5 volume of 4.84 mL. A LOT solar light simulator (1000 W Xenon lamp) irradiated an area of approximately 3.3 cm 2 of the stirred dye-sensitised photocatalyst suspension. The light source was equipped with an AM 1.5G filter, a water filter to remove IR irradiation and a 420 nm cut-off filter to avoid direct UV band gap photo-excitation of TiO 2 . The light simulator was calibrated to 1 sun irradiation intensity (100 mW cm -2 ). Samples were kept at 25 °C with a temperature controlled water bath and stirred during the course of the reaction.
Gas chromatography was used to analyse the headspace of the photoreactor in regular time intervals. The gas chromatograph (GC, Agilent 7890A Series) was equipped with a 5 Å molecular sieve column (held at 45 °C) and a thermal conductivity detector.
Nitrogen was used as carrier gas at a flow rate of approximately 3 mL min -1 . The GC was calibrated in regular intervals to determine the response factor of hydrogen to the internal standard methane. All experiments were performed at least in triplicate (unless otherwise noted) and the mean values and standard deviations (error) are reported (see below for statistical analysis). A minimum of 10% error was assumed for all experiments.

Determination of external quantum efficiency (EQE)
The EQE was determined for RuP│TiO 2 │NiP and DPP2│TiO 2 │NiP at the following wavelengths:  = 400, 450, 475, 500, 550 and 575 nm. A LED light source (Ivium Modulight) was used and the light intensity (3.05 or 3.13 mW cm -2 , see Table S7) measured with a radiometric detector coupled to an optical power meter (ILT 1400 radio and photometer).
The irradiated area was 0.283 cm 2 . Samples were prepared as described above using TiO 2 (2.5 mg), NiP (0.025 µmol) and RuP or DPP2 (0.05 µmol) in a total volume of 3 mL (AA, 0.1 S6 M). An airtight quartz cuvette was used as photoreactor, filled with the photocatalytic suspension and sealed with a rubber septum. Samples were purged with N 2 (including 2% methane as internal standard) for 10 min prior to the measurements and samples (40 μL) of the remaining headspace (0.89 mL) of the cuvette were analysed via GC after 2 h of irradiation. Experiments were run at least in duplicate and the EQE was determined using the following equation: Si photodiode (Hamamatsu S3071). Data < 1 ms was recorded by an oscilloscope after passing through amplifying electronics (Costronics) while data > 1 ms was simultaneously recorded on a National Instrument DAQ card. A few hundred laser pulses were averaged to obtain the kinetic traces. Samples were purged with inert gas (N 2 or argon) before acquisitions. Data was acquired and processed using home-built software written in the Labview environment.

Calculation of regeneration efficiencies
In order to estimate the overall regeneration efficiencies of each dye, we obtained the mean lifetime from stretched exponential fits (Fig. S11) according to:

S7
In order to compare the decays with and without AA and take into account the regeneration that took place on timescales shorter than our experimental setup could measure, we fixed the amplitude A in the stretched exponential expression to that obtained from the fitting in H 2 O when fitting the decay in AA. In this fashion, we normalise the decays to the same amount of photogenerated dye cation, as expected under the assumption that the injection yield has not been influenced by the addition AA. Simplifying to first order competitive kinetics, the quantum yield of generation is then calculated as: We calculate regeneration efficiencies of 52% (DPP1), 94% (DPP2), 88% (DPP5), 100% (RuP).