A Triad Photoanode for Visible Light-Driven Water Oxidation via Immobilization of Molecular Polyoxometalate on Polymeric Carbon Nitride

is shown to act as an effective and non-sacrificial electrostatic linker for immobilization of the anionic CoPOM onto the negatively charged surface of CN x . The optimized deposition of CoPOM using the PEI linker translates directly into improved efficiency of the transfer of photogenerated holes to water molecules and to enhanced oxygen evolution. This work thus provides important design rules for effective immobilization of POM-based catalysts into soft-matter photoelectrocatalytic architectures for light-driven water polymer that has been previously reported to enable improved deposition of CoPOM onto various metal oxides, [47,48,60] can also act as a highly effective electrostatic linker for immobilization of the anionic CoPOM onto the negatively charged surface of carbon nitride. Mecha-nistic studies revealed that the optimized deposition of CoPOM using the PEI linker translates directly into improved efficiency of the transfer of photogenerated oxidizing equivalents (holes) to water molecules and thus to enhanced oxygen evolution. On the other hand, the charge separation efficiency in triad photoanodes was largely unaffected by the CoPOM loading, and remained rather low (below 10% at moderate bias potentials), suggesting that primary recombination is a key performance bottleneck in triad photoanodes. Importantly, we also show that the PEI linker is effectively stabilized in the presence of the CoPOM catalyst that efficiently extracts the holes from PEI, preventing thus the oxidative degradation that takes place in the absence of CoPOM. This work thus highlights the importance of careful design of multi-component photoelectrocatalytic systems, and provides a simple protocol for effective immobilization of POM-based catalysts into soft matter-based photoelectrocatalytic architectures for light-driven water oxidation.

absorption by the light absorber. This problem could be only partially overcome by using, for example, ultrasmall (1-2 nm) CoO(OH) x nanoparticles that exhibit a larger bandgap and correspondingly better transparency in the visible range due to quantum size effects. [26] This led us to hypothesize that, in contrast to conventional bulk metal oxide-based water oxidation catalysts, a molecular-scale catalyst might be favorable for preventing the undesired light absorption by the catalyst and enabling also more controllable cocatalyst deposition. With this motivation in mind, we turned our attention to water oxidation catalysts based on well-defined molecular polyoxometalates (POMs), such as [Co 4 (H 2 O) 2 (PW 9 O 34 ) 2 ] 10 = CoPOM, a tetra-cobalt-doped polyoxometalate, that have attracted much attention with regards to catalytic applications, [42][43][44][45][46] and have been previously utilized as cocatalysts on various semiconducting metal oxides (e.g., TiO 2 , Fe 2 O 3 ) to fabricate photoanodes for light-driven water splitting. [47,48] However, to the best of our knowledge, no studies on CN x -based photoanodes comprising molecular POMs cocatalysts for photoelectrocatalytic water-splitting have been reported so far.
Herein, we report for the first time a triadic design of a photoanode consisting of porous TiO 2 as an electron collector, CN x as a sensitizer for visible light, and CoPOM as a molecular water oxidation catalyst. The triad photoanode enables visible (λ > 420 nm) light-driven oxidation of water to dioxygen at moderate bias potentials (>0.2 V vs RHE). Notably, we show that effective immobilization of CoPOM into the porous structure of the photoanode plays a crucial role in photoanode performance and can be significantly improved using polyethyleneimine (PEI), a cationic polymer that can act as a non-sacrificial, electrostatic linker between the surface of CN x and the CoPOM that are both charged negatively (Scheme 1). The thus achieved optimized immobilization of CoPOM is demonstrated to result in a more efficient transfer of photogenerated holes to water molecules and enhanced oxygen evolution.

Synthesis of CoPOM
The CoPOM complex was synthesized according to literature. [42] Briefly, Na 2 WO 4 ⋅2H 2 O (50.89 g), Na 2 HPO 4 ⋅7H 2 O (4.60 g) and Co(NO 3 ) 2 ⋅6H 2 O (9.98 g) were dissolved in 50 mL deionized water in a 200 mL round-bottom flask. The pH was adjusted to 7 by HCl under magnetic stirring. The solution was then stirred and refluxed at 100 °C for 2 h and cooled down to room temperature. CoPOM was finally obtained by recrystallization and washed by deionized water. The purity of CoPOM is confirmed by Attenuated total reflection Fourier transform infrared (ATR-IR) spectroscopy.

Preparation of FTO/TiO 2 Substrates
TiO 2 layers on FTO were prepared using an established doctorblading protocol. [26] Briefly, 0.25 g TiO 2 powder (Hombikat UV-100, pure anatase) was added to 1.25 mL anhydrous ethanol. The mixture was treated in ultrasonic bath for 10 min to produce a well-dispersed suspension. The FTO glass substrates with a size of 1.5 × 2.5 cm were first cleaned by acetone for removing residual organic contaminants by ultrasonication for 20 min. The cleaned FTO glass was then etched in 0.1 m NaOH and rinsed with deionized water. Two FTO glass pieces were placed between microscope glasses and fixed using a 3M scotch tape as frame and spacer, leaving an exposed area of ≈1.5 × 1.5 cm. Then, 200 µL TiO 2 suspension was dropped on the microscope glass and gently swept by a glass stick onto the FTO glass pieces. After drying at 70 °C for 20 min, the TiO 2 films were pressed at 10 4 N to improve the mechanical stability. All samples were calcined at 450 °C for 30 min in air before any tests or further treatments. The FTO/TiO 2 substrates were abbreviated as TiO 2 in this report.

Deposition of CN x
CN x was deposited by chemical vapor deposition of urea pyrolysis products according to the previous report. [26] Two pieces of TiO 2 electrodes were placed in a Schlenk tube connected to a round-bottom flask containing 1 g of urea. Before the CN x deposition was started, the muffle oven (Carbolite, Germany) was preheated to 425 °C. Then, the reactor was directly placed into the muffle oven and heated at 425 °C for 30 min. Finally, the reactor was cooled down to room temperature in air. The resulting electrodes are denoted as CN x -TiO 2 .

Immobilization of CoPOM on CN x -TiO 2
First, CoPOM (5 × 10 −3 m) and PEI (6 × 10 −3 m, based on monomer) solutions in an aqueous solution of sodium phosphate Scheme 1. The fabrication of the CN x /TiO 2 hybrid photoanode with an anionic molecular CoPOM cocatalyst immobilized via electrostatic attraction on the negatively charged CN x surface using a positively charged cationic PEI linking agent.
(80 × 10 −3 m, pH 5) and NaCl (137 × 10 −3 m) were prepared. Then, CN x -TiO 2 electrode was first dipped into the PEI solution for 5 min, then rinsed with distilled water and dried in air, followed by dipping in the CoPOM solution for another 5 min. The dipping processes in PEI and CoPOM were alternately repeated for five times to acquire desired amount of CoPOM and the sample was named as CoPOM-PEI-CN x -TiO 2 . The reference sample without PEI was prepared by dipping CN x -TiO 2 only into the CoPOM solution for five times and named as CoPOM-CN x -TiO 2

Material Characterization
The electronic absorption spectra were measured using a UV-vis spectrophotometer (UV-2600, Shimadzu, Japan) equipped with the integrating sphere and the absorptance (Abs.) was calculated by the equation: The baselines were recorded using an FTO glass and a BaSO 4 plate as references for transmittance and reflectance, respectively. Scanning electron microscopy (SEM) and focused Ga ion beam (FIB) cross sections were performed using a NVision 40 Ar (Carl Zeiss Microscopy, Germany) SEM/FIB crossbeam machine. energy-dispersive X-ray spectroscopy (EDX) investigations were done with the attached Octane Elite (AMETEK EDAX, USA) EDX system. Photoluminescence (PL) spectra were recorded on an RF-6000 spectrofluorophotometer (Shimadzu, Japan) using excitation wavelength of 360 nm with a 400 nm cut-off filter placed in front of the emission detector. ATR-IR spectroscopy was performed by the FT-IR spectrometer (Alpha II, Bruker, Germany). X-ray photo electron spectroscopy (XPS) measurements were performed with monochromatized Al Kα radiation using a PHI Quantera SXM system (ULVAC-PHI, Japan). The binding energies were calibrated based on C 1s peak of adventitious carbon (284.8 eV).

Photoelectrocatalytic Measurements
The photoelectrochemical measurements were conducted using an SP-300 BioLogic potentiostat and a typical 3-electrode system consisting of a Pt wire counter electrode, a Ag/AgCl (3.5 m KCl, 0.207 V vs SHE) reference electrode and tested photoanodes as working electrodes with geometric irradiation area of 0.5 cm 2 . Photoanodes were irradiated by visible light (λ > 420 nm) using a 150 W Xe lamp (L.O.T.-Oriel) with light power density of ≈150 mW cm -2 , equipped with a KG-3 (LOT-Quantum Design) heat-absorbing filter and a 420 nm longpass optical filter. All electrodes were illuminated from backside (through FTO glass). All photoelectrochemical measurements at hybrid photoanodes were carried out at least in triplicate (at three different electrodes), and representative average data is shown.
The charge separation efficiency (η sep ) and the hole transfer efficiency (η tr ) were evaluated using the approach reported by Dotan et al., [49] and utilizing sodium sulfite as a readily oxidizable reducing agent. [50] The hole transfer efficiency (η tr ) was determined using the equation: and Na SO 2 3 J are the photocurrents measured in the absence and presence of additional hole scavenger (Na 2 SO 3 ), respectively. The charge separation efficiency (η sep ) was estimated by the equation: where J max is the maximal photocurrent obtained by integrating the Abs spectrum (Figure 1b) over the AM1.5G solar spectrum (ASTM G-173; [51] 1.5 sun intensity) from 420 to 600 nm with Abs at 600 nm as a baseline. The oxygen evolution was recorded by the FireSting optical fiber oxygen meter (PyroScience, GmbH) in a home-made air-tight two-compartment cell with the oxygen collection efficiency as ≈75%, which was estimated by a direct electrolysis using a Pt working electrode. The volume of the photoanode compartment was 5 mL. The oxygen concentrations were not corrected for the losses in the gaseous headspace. The electrolyte was purged with argon before the electrodes were illuminated under applied potential of 1.12 V versus RHE. The incident monochromatic photon-to-current conversion efficiency (IPCE) was recorded using a photoelectric spectrometer Figure 1. ATR-FTIR spectra of all CN x -containing electrodes and CoPOM powder a); UV-vis electronic absorption spectra of the photoanodes; Abs., absorptance b). The nonzero baseline can be ascribed to the differences in internal reflection and scattering at the FTO/TiO 2 interface in the transmittance and reflectance measurement modes. [57] (Instytut Fotonowy Sp. z o.o.) equipped with a tunable monochromatic light source provided with a 150 W Xenon lamp and a grating monochromator with a bandwidth of ≈10 nm. The value of photocurrent density was the difference between current density under irradiation and in the dark in steady-state conditions with a wavelength sampling interval of 10 nm. The IPCE value for each wavelength was calculated according to equation: where i ph is the photocurrent density, h is Planck's constant, c is the velocity of light, P is the light power density, λ is the irradiation wavelength, and q is the elementary charge. The electrolyte for all photoelectrochemical measurements was 0.1 m sodium borate electrolyte with pH value of 8.0. Na 2 SO 3 (0.1 m) was dissolved in the electrolyte when photocurrents were measured in the presence of sacrificial electron donor. All potentials are recalculated and reported versus RHE.

Results and Discussion
The immobilization of the negatively charged [Co 4 (H 2 O) 2 (PW 9 O 34 ) 2 ] 10 (CoPOM) water oxidation catalyst onto polymeric carbon nitride is challenging since the surface of carbon nitride is known to be negatively charged due to large amount of unprotonated surface Brønsted base moieties that can be protonated only by highly concentrated strong acids. [52,53] In order to effectively immobilize the anionic CoPOM cocatalyst onto the internal surface of our porous CN x -TiO 2 photoelectrodes, we have therefore utilized the layer-bylayer technique demonstrated by Jeon et al. for immobilization of CoPOM onto various metal oxides. [47] The CN x -TiO 2 electrodes carrying a negative surface net charge were sequentially immersed into a solution of the cationic PEI and a solution of the anionic CoPOM for the desired number of times to fabricate the CoPOM-PEI-CN x -TiO 2 photoanode. It is known that PEI can be protonated in a wide pH range (pH 3-10) when dissolved in aqueous solutions. [54] The positively charged cationic PEI thus plays a role of an electrostatic linker between CN x and CoPOM that are both charged negatively. For comparison, the CN x -TiO 2 electrodes were also only dipped into the CoPOM solution resulting in the reference, linker-free electrodes CoPOM-CN x -TiO 2 . To evaluate and compare the CoPOM loading of the two CoPOM-containing CN x -TiO 2 electrodes, energy dispersive X-ray (EDX) spectra were recorded. The CoPOM-PEI-CN x -TiO 2 electrode shows significantly higher concentration of elements contained in CoPOM, with 0.35 at% Co, 0.13 at% P, and 0.97 at% W, compared to 0.05 at% Co, 0.07 at% P, and 0.03 at% W in case of the PEI-free CoPOM-CN x -TiO 2 reference electrode ( Figure S1, Supporting Information). The higher CoPOM loading of CoPOM-PEI-CN x -TiO 2 was further corroborated by XPS, showing the increase of surface concentration of Co by the factor of 3.4 in photoelectrodes comprising the PEI linker ( Figure S2, Supporting Information). In addition, the EDX mapping analysis (Figures S3-S6, Supporting Information) shows the presence of CoPOM within the whole volume of the porous CN x -TiO 2 film and confirms that the CoPOM loading and the homogeneity of its distribution throughout the whole thickness of the CN x -TiO 2 porous film is enhanced in the presence of the PEI linker. Finally, only in the case of CoPOM-PEI-CN x -TiO 2 , the characteristic IR fingerprint of CoPOM [42] is detectable (Figure 1a), which confirms improved CoPOM immobilization compared to electrodes without the cationic PEI linker. Hence, we conclude that the more effective immobilization of CoPOM in the CoPOM-PEI-CN x -TiO 2 photoanodes is due to beneficial effect of the electrostatic attraction between the positively charged PEI linker and the negatively charged CN x and CoPOM components. Figure 1b depicts the UV-vis electronic absorption spectra of TiO 2 , CN x -TiO 2 , CoPOM-CN x -TiO 2 , and CoPOM-PEI-CN x -TiO 2 .
All CN x -containing electrodes exhibit a significant red shift compared to the optical absorption of pristine anatase TiO 2 (3.2 eV, ≈390 nm) and CN x (2.9 eV, ≈428 nm), which we ascribe to effective sensitization of TiO 2 by CN x , including formation of a charge-transfer complex between CN x and TiO 2 , making thus possible also the direct optical electron transfer from the HOMO of CN x to the conduction band of TiO 2 . [55] The optical absorption edge of the hybrid photoanodes (≈2.6 eV, ≈477 nm) determined from the Tauc plots ( Figure S7, Supporting Information) is larger than the value typically obtained in our previous studies (≈2.3-2.5 eV), [26,[34][35][36][37][38][39][40] which can be explained by the inherent limitations of the Tauc formalism as applied for bandgap determination of hybrid materials, [56] and to the fact that in previous studies we determined the bandgap using the Kubelka-Munk function calculated from diffuse reflectance spectra of corresponding powders, while here we use Abs data obtained from measurements on complete photoanodes. Notably, the change in electronic absorption properties upon the deposition of the CoPOM catalyst is negligible, which indicates that the parasitic light absorption by the CoPOM catalyst is very low. This clearly highlights the advantage of using the molecular CoPOM as compared to, for example, cobalt oxide catalysts that typically show significant light absorption in the visible range due to their fully developed band structure and correspondingly low bandgap, blocking thus partially the visible light absorption by the light absorber. [35] The photoelectrocatalytic properties of our photoanodes were investigated under visible light (λ > 420 nm) using an appropriate cutoff-filter, in order to effectively shut off the intrinsic UV light absorption of TiO 2 . The porous TiO 2 layer thus serves solely as an electron-collecting scaffold that transports electrons injected under visible light irradiation from CN X into TiO 2 to the underlying FTO glass support, whereas the oxidizing equivalents (i.e., photoholes) photogenerated in CN x should be ideally channeled to the water-oxidizing CoPOM catalyst to drive dioxygen evolution from water. First, potential-dependent photocurrents (Figure 2a) were measured for all electrodes utilizing illumination by chopped visible light (λ > 420 nm, 150 mW cm -2 ) in borate electrolyte (pH 8). As expected, no photocurrents could be detected at the pristine TiO 2 substrate since anatase TiO 2 cannot be excited by visible light. The photocurrents recorded for the CN x -TiO 2 electrode without any CoPOM catalyst are attributed to the photocorrosion processes presumably at the CN x /TiO 2 interface as no oxygen evolution was detected at this electrode under identical experimental conditions (see Figure 4). In contrast, the photocurrent values significantly increased in the presence of both CN x sensitizer and CoPOM cocatalyst. Importantly, the polymeric PEI linker-containing CoPOM-PEI-CN x -TiO 2 photoanode showed the highest photocurrents within the whole potential range and also the highest monochromatic quantum efficiencies (IPCE) measured at a constant bias potential of 1.12 V versus RHE (Figure 2b). Importantly, the CoPOM-PEI-CN x -TiO 2 photoanode exhibited also the most negative photocurrent onset potential of 0.2 V versus RHE, which clearly indicates an improved rectifying behavior due to more effective extraction of photogenerated holes from CN x . Figure 2c shows photocurrent transients under the same visible light irradiation conditions at a constant potential of 0.78 V versus RHE, and indicates a relatively good short-term stability of the photocurrent response. Interestingly, for the CoPOM-free electrode (black line), the current spikes after switching on the light and negative current overshoots appearing after the light is switched off become significantly more pronounced. Such spikes and overshoots are a typical fingerprint of intense surface recombination processes, [58] indicating that, in the absence of water oxidation catalyst, the photoholes in CN x do not undergo the desired interfacial transfer, but instead accumulate in the CN x layer and subsequently either recombine or induce photocorrosion. In contrast, the current spikes are less pronounced and the overshoots are nearly absent in both CoPOM-containing electrodes, which again indicates that the CoPOM cocatalyst can extract holes generated in the CN x layer and trigger the desired water oxidation reaction.
In order to shed more light on the factors governing the photoresponse of CoPOM-containing photoanodes, we performed an analysis of charge separation (η sep ) and hole transfer (η tr ) efficiencies according to established protocols. [49,50,59] This analysis is based on the assumption that the measured photocurrent density in water oxidation can be calculated by multiplying the maximum possible photocurrent (obtained from the absorbed photon flux) by η sep and η tr , whereby the hole transfer efficiency η tr in the presence of a readily oxidizable reducing agent (here Na 2 SO 3 ) is taken as 100%; for details see the experimental section. The calculated η tr and η sep values for the both CoPOMcontaining electrodes are depicted in Figure 2d, calculated from data in Figure 2a and Figure S8 (Supporting Information). As expected, both efficiencies show clear dependence on the applied potential as stronger positive applied bias is beneficial for both charge separation and hole transfer. Notably, the charge separation efficiency η sep is very similar for both PEI linker-free CoPOM-CN x -TiO 2 and PEI-containing CoPOM-PEI-CN x -TiO 2 , which is also in line with only minor differences in PL spectra of the corresponding photoanodes that are possibly related to slightly more efficient quenching of emissive states in CN x ( Figure S9, Supporting Information). In stark contrast, the CoPOM-PEI-CN x -TiO 2 exhibits significantly higher charge transfer efficiencies than CoPOM-CN x -TiO 2 in the whole potential range. In other words, these data suggest that the superior photoelectrocatalytic behavior of the PEI linker-containing CoPOM-PEI-CN x -TiO 2 photoanode arises from a more efficient transfer of photogenerated holes from CN x to water, which can be attributed to the higher CoPOM loading due to the beneficial effect of the cationic PEI polymer linker.
However, in addition to its beneficial effect on the CoPOM immobilization, a question arises whether the cationic PEI polymer could potentially also act as a sacrificial electron donor that can be simply more easily oxidized than water by holes from CN x . In order to directly address this issue, we measured the photocurrents from the CN x -TiO 2 electrode modified with PEI polymer only (dipped in PEI solution five times), and the measurements at PEI-CN x -TiO 2 photoanodes were repeated subsequently in four cycles. The deposition of PEI enhanced photocurrents, but a gradual decrease of photocurrents was observed, ending up at the same values as those for the CN x -TiO 2 photoanode (Figure 3a). Hence, in the absence of the CoPOM catalyst, the PEI does effectively extract the holes from CN x , but is thereby oxidatively degraded. As a next step, the same protocol was also applied to the CoPOM-PEI-CN x -TiO 2 photoanode. Contrary to the gradual decline of photocurrents observed for PEI-CN x -TiO 2 , the photocurrents at CoPOM-PEI-CN x -TiO 2 remain completely stable over all four cycles (Figure 3b). This is also in line with the chronoamperometric photocurrent measurements of the three electrodes (Figure 3c), which show that PEI-CN x -TiO 2 exhibits a fast decline in photocurrent due to deg-radation of PEI in the absence of CoPOM, whereby the photocurrent at CoPOM-PEI-CN x -TiO 2 is practically stable. Therefore, we conclude that in the presence of CoPOM, the holes are efficiently transferred from PEI to the CoPOM catalyst where they drive water oxidation, and the cationic PEI linker is thereby effectively stabilized.
In order to unambiguously prove the dioxygen evolution at CoPOM modified photoanodes, we performed photoelectrocatalytic OER measurements (Figure 4a) in a borate solution (pH 8) under prolonged (1 h) visible light irradiation (λ > 420 nm, 150 mW cm -2 ). Both CoPOM-containing hybrid photoanodes, CoPOM-CN x -TiO 2 and CoPOM-PEI-CN x -TiO 2 , clearly exhibit OER activity under visible light illumination. This confirms our assumption that charge transfer from CN x to CoPOM is feasible and that the presence of the CoPOM cocatalyst is necessary to trigger the OER. Importantly, at the PEI-containing electrode the oxygen evolution rate was doubled compared to the counterpart photoanode without PEI. Importantly, no oxygen evolution was observed at the CoPOM-free CN x -TiO 2 photoanode despite substantial photocurrents that can be ascribed to photocorrosion. [38] This result is also in line with our previous studies that confirmed that the presence of an effective OER catalyst is absolutely necessary to observe oxygen as a product of water oxidation at CN x -TiO 2 hybrid photoanodes. [26,[34][35][36][37][38][39][40][41] On the other hand,  ; measured under polychromatic visible light irradiation (λ > 420 nm, 150 mW cm -2 ) at 1.12 V versus RHE in a borate electrolyte (0.1 m, pH 8.0). The oxygen evolution was measured at least at three different electrodes (representative curves are shown), and the error is taken as 2σ (σ = standard deviation; 95% confidence interval). CN x -free pristine TiO 2 photoanodes modified with CoPOM exhibited neither photocurrents nor oxygen evolution since pristine TiO 2 does not absorb in the visible range. The apparent (based on dissolved O 2 and uncorrected for losses in the headspace) Faradaic efficiencies (FE) of oxygen evolution for CoPOM-PEI-CN x -TiO 2 (15% ± 4%) and for CoPOM-CN x -TiO 2 (12% ± 4%) are rather low, which suggests that even in the best photoanodes the overall utilization of holes generated in CN x for water oxidation is still far from optimum, and a substantial portion of holes does not induce the OER but instead contributes to the photocorrosion of CN x (Figure 4b). The improved photocurrent onset potential, higher oxygen production rate, and FE at the PEI-containing CoPOM-PEI-CN x -TiO 2 photoanode clearly confirm the beneficial effect of the cationic PEI polymer that serves as an effective linker by establishing the electrostatic attraction between the CN x sensitizer and CoPOM catalyst that are both negatively charged. This results in a more efficient hole extraction from CN x and more effective utilization of photogenerated holes for water oxidation due to the more effective immobilization (i.e., higher loading) of the CoPOM water oxidation catalyst.

Conclusion
For the first time, a triad photoanode comprising a molecular cobalt polyoxometalate (CoPOM) embedded in the porous structure of hybrid photoanodes consisting of polymeric carbon nitride deposited onto an electron collecting porous TiO 2 layer is reported. The photoanodes exhibit complete water oxidation to dioxygen under visible (λ > 420 nm) light irradiation, with photocurrents down to relatively low bias potentials of 0.2 V versus RHE. Importantly, it is demonstrated that PEI, a positively charged cationic polymer that has been previously reported to enable improved deposition of CoPOM onto various metal oxides, [47,48,60] can also act as a highly effective electrostatic linker for immobilization of the anionic CoPOM onto the negatively charged surface of carbon nitride. Mechanistic studies revealed that the optimized deposition of CoPOM using the PEI linker translates directly into improved efficiency of the transfer of photogenerated oxidizing equivalents (holes) to water molecules and thus to enhanced oxygen evolution. On the other hand, the charge separation efficiency in triad photoanodes was largely unaffected by the CoPOM loading, and remained rather low (below 10% at moderate bias potentials), suggesting that primary recombination is a key performance bottleneck in triad photoanodes. Importantly, we also show that the PEI linker is effectively stabilized in the presence of the CoPOM catalyst that efficiently extracts the holes from PEI, preventing thus the oxidative degradation that takes place in the absence of CoPOM. This work thus highlights the importance of careful design of multi-component photoelectrocatalytic systems, and provides a simple protocol for effective immobilization of POM-based catalysts into soft matter-based photoelectrocatalytic architectures for light-driven water oxidation.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.