Rational wiring of photosystem II to hierarchical indium tin oxide electrodes using redox polymers †

Photosystem II (PSII) is a multi-subunit enzyme responsible for solar-driven water oxidation to release O 2 and highly reducing electrons during photosynthesis. The study of PSII in protein film photoelectrochemistry sheds light into its biological function and provides a blueprint for artificial water-splitting systems. However, the integration of macromolecules, such as PSII, into hybrid bio-electrodes is often plagued by poor electrical wiring between the protein guest and the material host. Here, we report a new benchmark PSII–electrode system that combines the efficient wiring afforded by redox-active polymers with the high loading provided by hierarchically-structured inverse opal indium tin oxide (IO-ITO) electrodes. Compared to flat electrodes, the hierarchical IO-ITO electrodes enabled up to an approximately 50-fold increase in the immobilisation of an Os complex-modified and a phenothiazine-modified polymer. When the Os complex-modified polymer is co-adsorbed with PSII on the hierarchical electrodes, photocurrent densities of up to B 410 m A cm (cid:2) 2 at 0.5 V vs. SHE were observed in the absence of diffusional mediators, demonstrating a substantially improved wiring of PSII to the IO-ITO electrode with the redox polymer. The high photocurrent density allowed for the quantification of O 2 evolution, and a Faradaic efficiency of 85 (cid:3) 9% was measured. As such, we have demonstrated a high performing and fully integrated host–guest system with excellent electronic wiring and loading capacity. This assembly strategy may form the basis of all-integrated electrode designs for a wide range of biological and synthetic catalysts.


Introduction
The immobilisation of photosynthetic proteins onto electrodes is of importance to a range of current and future innovations, including biosensors, [1][2][3] biophotovoltaic systems [4][5][6][7] and photoelectrochemical (PEC) platforms. 8,9Photosystem II (PSII) is a photosynthetic enzyme with the ability to photocatalyse water oxidation, a bottleneck reaction in artificial photosynthesis, at theoretical rates of up to 250 mol O 2 (mol PSII monomer) À1 s À1 . 10,11][16][17] Several strategies for the integration of PSII into electrodes are currently employed, each with unique advantages.Before these approaches are discussed, some knowledge of the mechanism behind PSII water oxidation is required.Briefly, light is absorbed by pigments within PSII, and funnelled into the reaction centre complex where charge formation and separation at the P680 primary electron donor site occurs.The photogenerated electrons are then transferred via pheophytin and plastoquinone A (Q A ) to the terminal electron acceptor plastoquinone B (Q B ), which is located on the stromal side of the enzyme.Holes generated at the P680 are transferred in the lumenal direction, via a redox-active tyrosine side chain (Tyr Z ) to the oxygen-evolving complex (OEC), where water is oxidised to liberate H + and O 2 . 18,19If the PSII is adsorbed in the correct orientation on an electrode, direct electron transfer from the Q A /Q B to the electrode can take place. 9,20owever, a Q B mimic, such as 2,6-dichloro-1,4-benzoquinone (DCBQ), is typically required as a diffusional mediator between the insufficiently wired PSII and the electrode to substantially enhance photocurrent generation. 19 traditional approach for the immobilisation of photosynthetic reaction centres is to align the proteins on chemically-modified electrodes functionalised with linkers such as quinonoid, 21 N-hydroxy-succinimidyl ester, 22 nickel nitrilotriacetic acid, 23,24 cytochrome c 25,26 and carboxylic acid/amino groups. 27However, the magnitude of the photocurrent is limited by the attachment of a single monolayer of photosynthetic reactions centres that can be assembled on the electrode.
A strategy to enhance the loading of electrically wired PSII onto electrodes is to entrap PSII in a redox-active polymer matrix on an electrode surface. 28,29In this approach, PSII of any orientation can in principle be efficiently wired to the electrode by the redox-active moieties that are homogeneously distributed in the matrix, which can mediate charge transfer via an electron hopping mechanism. 30The benchmark system using this approach consists of a flat gold electrode on which PSII is embedded in an Os complex-modified polymer (E 1/2 = 0.395-0.505V versus the standard hydrogen electrode; vs. SHE). 31Photocurrents of up to 45 mA cm À2 at an applied potential (E app ) of 0.5 V vs. SHE were reported for this biophotoanode.Despite its advantages, the performance of this system was limited by the intrinsic properties of the polymeric matrix.Independently of the total loading at the electrode surface, the amount of electroactive enzyme is defined by the rate of charge transfer via electron hopping, which limits the maximum (photo-)electrocatalytic response that can be detected. 32On modified flat electrodes where enzymes are entrapped in redox polymers, the current generation typically arises from catalysts present within a thin layer (a few micrometer thick) at the electrode/hydrogel interface; the remaining catalysts in the outer layers of the film are electro-inactive and do not contribute to current generation. 33n emerging and effective enzyme immobilisation strategy involves the adoption of highly structured electrode morphologies [34][35][36] to increase the available surface area for enzyme adsorption. 9,27,37In a recent benchmark system, PSII was adsorbed on a hierarchically-structured indium tin oxide (ITO) electrode that incorporated macroporosity (for enhanced enzyme and substrate penetration) and mesoporosity (to enhance the effective surface area and enzyme anchoring) with high thickness. 9As a result, a 16 000-fold increase in PSII loading was observed compared to conventional flat electrodes. 6,31However, insufficient wiring at the PSII-electrode interface was still apparent, with non-mediated photocurrents of 20 mA cm À2 being observed in contrast to 1 mA cm À2 in the presence of a freely diffusing mediator.A further limitation of the electrode is poor PSII photostability, with the electrode exhibiting a photocurrent half-life time of only a few minutes.
Here, we report a high performing PSII-electrode system that combines the best aspects of two advanced enzyme immobilisation strategies: the use of a redox polymer matrix to enable efficient PSII wiring, and the use of high surface area hierarchicallystructured ITO electrodes to enable high loading of both the polymer and the PSII (Fig. 1a).The highly structured electrode scaffold increases the polymer-electrode interface and reduces the average charge transfer distance between the PSII and the electrode surface via the polymeric matrix.This enables the wiring of a large population of PSII to the electrode, which translates to high effective loading.We first compared the performance of two promising redox polymers differing in chemical and redox properties as electron conducting matrices for PSII in inverse opal mesoporous ITO (IO-ITO) electrodes (Fig. 1b).We then focused on the optimisation of the lead ITOpolymer-PSII system to ultimately deliver high photocurrents in the absence of diffusional mediators, at an extended operating lifetime.

Physical characterisation
The surface morphology of the electrodes was analysed by scanning electron microscopy (SEM; Philips SFEG XL30).A 5804 Eppendorf centrifuge and Carbolite furnace (ELF 11/14B/301) were used for electrode preparation.UV-vis absorption spectra were recorded on a Varian Cary 50 or Agilent Cary 60 UV-vis spectrophotometer, using cuvettes with an optical path length of 1 cm.Nuclear magnetic resonance (NMR) experiments were conducted with a Bruker 200 DPX spectrometer with a proton resonance frequency of 200.13 MHz (the residual solvent peak was used as internal standard).All dynamic light scattering (DLS) measurements were carried out with a Malvern Zetasizer Nano ZS (laser wavelength: 633 nm, measurement angle: 1731 backscatter).The buffers were filtered through 450 nm membrane filters (polypropylene membranes bearing a borosilicate prefilter, Alltech) before dissolution of the polymers for DLS measurements; cuvettes were rinsed with buffer solution prior to measurements.For the filtration of polymer suspensions, non-pyrogenic 200 nm polyethersulfone membranes (Filtropur S, Sarstedt) were used.

Preparation of IO-ITO|PSII electrodes
The IO-ITO electrodes were fabricated according to a previously reported co-assembly procedure. 9A standard IO-ITO electrode macropore diameter of 750 nm, film thickness of 20 mm and geometrical surface area of 0.25 cm 2 were used, unless stated otherwise.An amount of 4.2 mL of the described polystyrene-ITO dispersion on a 0.25 cm 2 geometrical surface area corresponds to a 10 mm thick IO-ITO structure.To obtain higher film thicknesses, the polystyrene-ITO mixture (4.2 mL) was deposited several times with a 4 h drying period in between.All current densities (mA cm À2 ) are reported with respect to the geometrical surface area.The IO-ITO|PSII modified electrodes were prepared as follows: a PSII stock solution (1 mL, 2.6 mg Chl a mL À1 ) was drop-cast onto the IO-ITO electrode and incubated in the dark for 15 min at room temperature.It was determined that 1 mL of PSII stock solution provided an excess of PSII for 20 mm thick IO-ITO and ensured full enzyme coverage on the electrode surface.Prior to electrochemical studies, the IO-ITO|PSII electrode was rinsed (3 Â 500 mL) with the electrolyte solution to remove all unbound enzyme from the electrode surface.

Preparation of IO-ITO|polymer-PSII electrodes
A PSII to polymer ratio of 1 : 1 (v/v) was defined based on 1 mL PSII stock solution (2.6 mg Chl a mL À1 ) and 1 mL polymer solution (10 mg mL À1 ).The PSII to polymer ratio was optimised on a 20 mm thick IO-ITO electrode by keeping the PSII solution volume (1 mL) and concentration (2.6 mg Chl a mL À1 ) constant, and varying the polymer solution concentration at constant volume (1 mL).The optimal PSII to polymer ratio was found to be 1 mL of PSII solution (2.6 mg Chl a mL À1 ) to 1 mL of the polymer solution (10 mg mL À1 ) per 20 mm thickness of IO-ITO.
The IO-ITO|polymer-PSII electrodes were prepared as follows: the PSII stock solution (1 mL) was mixed with a redox polymer solution (1 mL) and the polymer-PSII mixture was drop-cast (2 mL) onto the IO-ITO electrode (20 mm thick) and incubated in the dark for 15 min at room temperature.Prior to electrochemical studies, the IO-ITO|polymer-PSII electrode was rinsed (3 Â 500 mL) with the electrolyte solution.

Determination of PSII and polymer loading on IO-ITO
The amount of PSII on the IO-ITO surface was quantified by scratching off the IO-ITO from the glass substrate and washing with MeOH (500 mL) to extract Chl a from the electrode surface into a centrifuge vial.The vial was centrifuged (10 000 rpm, 1 min), and the UV-vis spectrum of the supernatant was recorded (Fig. S6a, ESI †).The band with an absorption maximum of l max = 665 nm assigned to Chl a (extinction coefficient e = 79.95(Chl a mg) À1 mL cm À1 ) 40 was used to calculate the amount of PSII monomers assuming 35 Chl a molecules per PSII monomer. 41he Os complex loading in the P Os polymer was determined by UV-vis spectroscopy in DMSO, using the freely diffusing Os complex analogue, cis- For the mediated photocurrent measurements, a DCBQ solution in DMSO (40 mL, 100 mM) was added to give a final concentration of 1 mM in the electrolyte solution.The following correction factor was used to convert the reduction potential to SHE: ). IO-ITO|polymer-PSII electrodes were typically exposed to cycles of 30 s dark and 30 s light irradiation in the PF-PEC measurements.The photocurrent response was defined as the baseline-corrected photocurrent peak after the third light exposure, accounting for charging effects and to avoid overestimation. 19The action spectra were recorded using a Xenon lamp Solar Light Simulator (300 W) coupled to a monochromator (MSH300; both from LOT Quantum design).The light intensity was measured as a function of wavelength with a photodiode detector (SEL033/F/QNDS1/W) and power meter (ILT1400).For the O 2 evolution measurements, an Ivium Modulight LED module (l = 660 nm; E e = 10 mW cm À2 ) and a gastight two-compartment glass cell with the IO-ITO|polymer-PSII working electrode separated from the counter electrode by a glass frit were employed in an anaerobic (O 2 level o1 ppm) MBraun glovebox.The error analysis was based on the standard deviations resulting from at least three experiments.

Product analysis
Quantification of O 2 was performed with a calibrated fluorescence O 2 sensor (Neofox; Ocean Optics FOSPHOR probe) inside an MBraun glovebox to avoid leakage of atmospheric O 2 .The probe was placed inside the cell headspace, protected from direct irradiation and the background signal was subtracted from all measurements using the OriginPro 9.0 software.The reported O 2 values were corrected for dissolved O 2 using Henry's Law.
Equations ( 1)-( 5): where d -diffusion layer thickness of the electrons (m), D e -apparent electron diffusion coefficient (m 2 s À1 ), R -ideal gas constant (8.314J K À1 mol À1 ), T -temperature (K), n -number of electrons transferred, F -Faraday constant (96485.332C mol À1 ) and v -scan rate (V s À1 ). 42 where G -surface coverage of the electrochemically-active redox centres (mol m À2 ), Q -total charge passed (C) and A -geometric electrode area (m 2 ). 42 where G PSII -surface coverage of immobilised PSII (mol m À2 ), A Dl -background (l = 750 nm) corrected UV-vis absorption of Chl a, V MeOH -MeOH volume (mL) and M Chl a -molecular mass of Chl a (893.5 g mol À1 ). 40 where EQE -external quantum efficiency (defined as the number of incident photons converted to electrons at a selected irradiation wavelength), I e -electron flux at the external circuit (mol m À2 s À1 ),

Synthesis and characterisation of IO-ITO and polymers
This study uses a hierarchical IO-ITO electrode, which has previously demonstrated a high loading capacity for the large enzymes, PSII and hydrogenase (Fig. S1a, ESI †). 9 The macropores with diameter of 750 nm and channels of 150 nm are also suitable for the penetration of macromolecular polymers; the mesopores with a diameter of approximately 50 nm provide a high effective surface area of B115 Â 10 6 m 2 m À3 for polymer/ enzyme adsorption. 9The tunability of the film thickness (up to 80 mm, Fig. S1d, ESI †) provides extra flexibility in the optimisation of guest loading.The PSII used was isolated from the thermophilic cyanobacterium Thermosynechococcus elongatus given that cyanobacterial PSII is relatively well characterised, 41,43,44 and it exhibits high activity and relative robustness. 38,45he polymers chosen for this study include the Os complexmodified polymer P Os (Fig. 2a), which has demonstrated excellent integration of PSII on flat electrodes; 31 and the purely organic P Phen (Fig. 2b), which has a better matched redox potential with the Q A /Q B cofactors and has also demonstrated favourable wiring of PSII to flat electrodes. 6Both polymers are compatible with PSII and are stable under the acidic pH conditions for photocurrent measurements. 6,39The chemical structure, purity and size of the polymers were confirmed by 1 H NMR (Fig. S2, ESI †), UV-vis spectroscopy (Fig. S3, ESI †) and DLS (Fig. S4, ESI †), respectively.The 1 H NMR spectra of the polymer backbones correspond to the expected structure (Fig. S2, ESI †).Based on the integral ratio between methyl groups of terminal isopropyl units and the intra-chain imidazole unit, as well as the two signals assigned to the polymer chain, a molecular weight of B26 AE 3 kDa was estimated for the P Os backbone.For the backbone of the P Phen polymer, analysis of the molecular weight via NMR spectroscopy was not possible due to overlapping signals in the spectrum of the backbone.
The total number of Os complexes in P Os was quantified using UV-vis spectroscopy in DMSO (0.74 AE 0.04 mmol g À1 polymer, Fig. S3, ESI †), which is consistent with ICP-AES measurements (0.67 AE 0.05 mmol Os g À1 polymer).Cis-[Os II Cl(1-(n-butyl)imidazole)(bipy) 2 ](PF 6 ) (Fig. S3c, ESI †), which can be regarded as the freely diffusing analogue to the Os complex moiety in the P Os , was used as a reference for characterisation by UV-vis spectroscopy.The spectrum of the freely diffusing complex and the polymer exhibit the same spectral features (Fig. S3a, ESI †).Thus, for the calculation of the total number of Os complexes within the polymer, we assume that both species exhibit the same extinction coefficients.From the UV-vis studies, the ratio of noncomplexed imidazole units to Os complex moieties was calculated to be E7 : 1, which corresponds to a molecular weight of B44 AE 5 kDa for P Os .The same analysis was performed with the freely diffusing toluidine blue (Fig. S3c, ESI †) and P Phen .The spectral shapes of both species are again similar (Fig. S3d, ESI †), but the extinction coefficient of the toluidine blue moiety is increased upon covalent attachment to the polymer backbone (the primary amine in the toluidine blue monomer is converted to a secondary or even to a tertiary amine upon reaction with the epoxide functionality of the polymer backbone of P Phen ).Thus, the estimation of the exact number of toluidine blue species was not possible (calculated values exceed the theoretical values).
The hydrodynamic particle diameter of P Os and P Phen was determined using DLS (Fig. S4, ESI †) to be 16 AE 1 nm and B500 nm (broad distribution), respectively, which indicate the agglomeration of smaller polymer chains.Since both polymer solutions were filtered through a membrane with 200 nm pore size, it was concluded that the P Phen polymer chains form weak agglomerates that can be easily disassembled.The estimated sizes and agglomeration properties of P Os and P Phen are expected to allow them to enter into the IO-ITO structure either by diffusional transport or by convection due to the capillary forces induced by pore filling and H 2 O evaporation.

Integration of PSII and polymer into IO-ITO electrodes
The polymers (1 mL, 10 mg mL À1 ) were drop-cast onto the IO-ITO and allowed to adsorb for 15 min at room temperature.The redox properties of the adsorbed polymers on the IO-ITO electrode (IO-ITO|polymer) were characterised using cyclic voltammetry (CV; Fig. 2c and Fig. S5, ESI †).The redox waves of P Os and P Phen were attributed to the Os 3+/2+ (1e À transfer process) and Phen + /PhenH (2e À /H + transfer) redox couples, respectively. 6The positive reduction potential of the P Os polymer (E 1/2 = 0.44 V vs. SHE) is expected to provide a large driving force for electron transfer from the Q A and Q B (E 1/2 = À0.14V and À0.06 V vs. SHE, respectively) 19 to the redox centres of the polymer (Fig. 1b).However, electron transfer between Q A /Q B and the Os complexes results also in a substantial potential loss (40.4 V). 5 The P Phen hydrogel exhibits a less positive reduction potential (E 1/2 = 0.04 V vs. SHE), which matches the Q B more closely (Fig. 1b).The reversibility of the electron transfer process for the surface-adsorbed redox polymers is evident in the almost symmetrical shape in the CV scans of P Os and P Phen , which show minimal peak separation between the oxidation and reduction potentials (DE p = 0.02 AE 0.01 and 0.01 AE 0.005 V for P Os and P Phen , respectively).Furthermore, an anodic to cathodic peak current ratio close to unity ( Ip a Ip c ¼ 0:97 and 0.83 for P Os and P Phen , respectively) can be observed, and the current density is linearly proportional to the scan rate up to 100 mV s À1 (Fig. 2c inset and Fig. S5, ESI †). 42The observed slight increase in the DE p at scan rates 410 mV s À1 (Fig. S5, ESI †) was attributed to the rate limiting charge transfer between the polymer and the electrode surface. 46In particular, P Phen showed small shoulder waves at high scan rates that could arise from the slow 2e À /H + transfer rate at the iminium cation site.The voltammetric features of P Os , even at the relatively fast scan rates used here, are characteristic for surfaceconfined species.The corresponding diffusion layer thicknesses of the electron (d, calculated from eqn (1)) give an estimate of the film thickness that is accessible to the electrochemical process assuming planar semi-infinite diffusion.Based on the previously reported apparent electron diffusion coefficient of the electron for P Os (D e of 4.00 AE 0.47 Â 10 À9 cm 2 s À1 ), 39 the d value corresponding to the scan rate of 100 mV s À1 320 nm.Hence, the diffusional range of the electrons within P Os is in the range of the IO macropore radius (375 nm; Fig. S1a, ESI †) even at fast scan rates.As such, the IO structure should increase the total polymer loading that can participate in electron transfer in a given geometric surface area by taking advantage of the thick 3-D architecture.
No photocurrent originating from the IO-ITO|polymer electrodes during irradiation (l = 685 nm, E e = 10 mW cm À2 ) was observed (Fig. 2c).The surface coverage (G) of the electrochemically-active redox centres connected to the electrode surface was calculated for each polymer using eqn (2); the total charge was calculated by integrating the area under the CV curve minus the background.A substantial enhancement in polymer loading (Fig. S5, ESI †) was observed for IO-ITO compared to flat electrodes.The polymer loading increased approximately linearly with the electrode thickness (Fig. 2d).Loadings of 1.7 AE 0.2 nmol cm À2 and 1.7 AE 0.5 nmol cm À2 were observed for flat ITO electrodes with the adsorbed polymers, P Os and P Phen , respectively, which is comparable to previously reported values (1.8 AE 0.1 nmol cm À2 ) on flat glassy carbon electrodes. 39 S1, ESI †).The number of electrochemically-active Os complexes on 20 mm thick IO-ITO was found to be B85 AE 10% of the total number of immobilised Os atoms, quantified by ICP-AES and UV-vis spectroscopy.The IO-ITO|P Os electrode exhibited excellent stability, showing no significant desorption or decomposition after 100 CV cycles at 10 mV s À1 scan rate (Fig. 2c).The IO-ITO|P Phen electrode exhibited lower stability This journal is © The Royal Society of Chemistry 2016 (63% and 38% G Phen remaining after the second and third CV cycle at 10 mV s À1 scan rate, respectively, Fig. S5e, ESI †).The imidazole functionality in the P Os is also likely to have a strong affinity for the ITO surface and act as an anchoring group, analogous to histidine-tagged enzymes. 47The toluidine blue centres of the P Phen (heterocyclic N and S atoms pK a o 7) are most likely deprotonated and the polymer backbone groups (amine functions pK a 4 7) are protonated at pH 6.5.The hydrogel nature of the polymers allows the diffusion of small molecules throughout the network, although the lack of anchoring groups in P Phen prevents stable loading.
Following the assembly and characterisation of the IO-ITO|P Os and IO-ITO|P Phen electrode systems, PSII was introduced into the electrode system.PSII (1 mL, 2.6 mg Chl a mL À1 ) and the redox polymer (1 mL, 10 mg mL À1 ) were mixed together and immediately drop-cast on the IO-ITO electrode (20 mm thick) as a uniform blend, then allowed to adsorb in the dark for 15 min at room temperature.The amount of PSII entrapped in the polymer matrix inside the electrode (G PSII ) was quantified based on the absorption amplitude of Chl a (l max = 665 nm, eqn (3)), extracted from PSII using MeOH (Fig. S6a, ESI †).UV-vis spectra of polymer solutions (0.02 mg mL À1 ) in the electrolyte solution and MeOH (Fig. S6b, ESI †) showed a negligible absorption at the irradiation wavelength used in PF-PEC (l = 685 nm).Exceptionally high PSII loadings were observed for IO-ITO|P Os -PSII (144 AE 21 pmol cm À2 ), IO-ITO|P Phen -PSII (149 AE 7 pmol cm À2 ) and IO-ITO|PSII (162 AE 17 pmol cm À2 ) (Fig. S6c, ESI †).The slightly higher PSII loading in the PSII-only system could be explained by more space being available (that could be filled by the enzymes) in the absence of polymers.The SEM images of the IO-ITO electrodes taken before and after P Os -PSII and P Phen -PSII deposition (Fig. S1,  ESI †) indicate no evident channel or pore blockages.
The effective assembly of PSII with the polymers can be attributed to favourable non-covalent interactions between the protein shell and the polymers.The hydrophilic nature of the polymers is bestowed primarily by the cationic Os complex/ phenothiazine dye, with some contributions by the multiple polar functional side groups (P Os : imidazole and amine groups; P Phen : polyethylene glycol side chains and OH-functions).At pH 6.5, P Os is expected to behave as a cationic polyelectrolyte since the primary amine (pK a 10) and imidazole groups (pK a 7) are protonated.This also contributes to the close to optimal polymer solvation and swelling, supported by high D e value previously observed. 39-PEC with IO-ITO|polymer-PSII electrodes PF-PEC measurements were performed at 25 1C using an IO-ITO|polymer-PSII working, a Pt wire counter and a Ag/AgCl (3 M KCl) reference electrode.The electrolyte solution was adjusted to pH 6.5 and contained CaCl 2 (20 mM), MgCl 2 (15 mM), KCl (50 mM) and MES (40 mM).The action spectra of the IO-ITO|PSII, IO-ITO|P Os -PSII and IO-ITO|P Phen -PSII photoelectrodes were recorded to determine appropriate wavelengths of light for photocurrent generation (Fig. 3 and Fig. S7, ESI †).In a typical experiment, the wavelength was decreased in steps of 20 nm starting from 760 nm at an applied potential of 0.5 V vs. SHE and the photoresponse was measured at each wavelength.The maximum photocurrent was observed at 680 nm, which matches the UV-vis absorption spectrum of PSII and supports the integrity of PSII in its native state during the immobilisation on the IO-ITO electrode.21,48 The action spectra of the control samples IO-ITO, IO-ITO|P Os and IO-ITO|P Phen corresponded to the UV-vis absorption spectra of the respective polymers (Fig. S6b, ESI †) and confirmed no significant contribution to the photocurrent generation from the polymers (Fig. S7, ESI †).
Stepped chronoamperometry with chopped red-light irradiation (l = 685 nm, E e = 10 mW cm À2 ) was performed to characterise the onset potential (E onset ) of photocurrents in each IO-ITO|polymer-PSII system (Fig. S8, ESI †).In a typical experiment, the applied potential was gradually increased in steps of 0.1 V in the anodic direction.A summary of the photoresponse as a function of the E app is shown in Fig. 4. The IO-ITO|P Phen -PSII system showed an E onset value of B0.1 V vs. SHE, which is slightly more positive than expected, possibly due to other minor interference charge transfer pathways.However, the E onset of IO-ITO|P Phen -PSII is still clearly more negative than that of the IO-ITO|P Os -PSII electrode (E onset = B0.3V vs. SHE; Fig. 4 inset), which is consistent with the lower E 1/2 of P Phen (E 1/2 = 0.04 V vs. SHE) compared to the P Os (E 1/2 = 0.44 V vs. SHE).The photocurrents for both the IO-ITO|P Os -PSII and IO-ITO|P Phen -PSII electrodes reach a plateau at B0.5 V vs. SHE.No photoactivity and negligible dark current were observed for the IO-ITO|P Os and IO-ITO|P Phen electrodes (Fig. S8d, ESI †).Upon prolonged irradiation at more positive potentials (E app 4 0.6-0.7 V vs. SHE), a drop in photocurrent was observed.This drop in photocurrent is irreversible, as shown by the low photoresponse given by a backward scan in the negative direction (at 0.5 V vs. SHE, Fig. S9a,  ESI †).CV scans of the IO-ITO|P Os -PSII electrode (Fig. S9b, ESI †) confirmed the stability and homogeneity of the integrated PSII-polymer film on the electrode surface in the dark.However, CV scans performed with red light irradiation (Fig. S9c, ESI †) show a significant decrease in photocurrents after 3 potential sweep cycles over the range 0.1-0.8V vs. SHE, which is indicative of P Os -PSII film photodegradation (PSII-limiting system). 49o investigate the quality of the wiring between the PSII and the ITO electrode in the IO-ITO|polymer-PSII systems, chronoamperometry at an applied potential of 0.5 V vs. SHE was performed in the presence and absence of the diffusional mediator, DCBQ, with chopped light irradiation (Fig. 5).Typical photocurrent densities for optimised 20 mm thick IO-ITO|PSII, IO-ITO|P Os -PSII, and IO-ITO|P Phen -PSII electrodes in the absence of a diffusional mediator (Fig. 5a) were approximately 15, 230 and 45 mA cm À2 , respectively, which compares favourably with PF-PEC of previously reported PSII-electrodes. 9,19Bare IO-ITO and IO-ITO|polymer electrodes exhibited photocurrent densities below 100 nA cm À2 .
The relatively large photoresponse observed for the IO-ITO|P os -PSII system is indicative of efficient electronic communication between PSII and the electrode.An external quantum efficiency (EQE) of 4.4% (derived using eqn (4)) was obtained for the IO-ITO|P Os -PSII system, which is 15-fold higher than for IO-ITO|PSII (EQE = 0.3%) and the highest reported so far for a diffusional mediator-free PSII-electrode. 9,19The photoresponse in the IO-ITO|P Phen -PSII system (EQE = 0.8%) is improved compared to IO-ITO|PSII, however the enhancement is not as great as the IO-ITO|P Os -PSII system, which indicates that the P Phen is less efficient at wiring PSII to the electrode, possibly because of its significantly lower driving force for electron transfer.
The addition of DCBQ (0.36 V vs. SHE) 9 to the IO-ITO|P Os -PSII system gave rise to a further 1.5-fold photocurrent density increase (375 mA cm À2 , EQE = 7.7%, Fig. 5b).Similarly, the addition of DCBQ to the IO-ITO|P Phen -PSII system gave rise to a 6-fold photocurrent density increase (236 mA cm À2 , EQE = 4.6%).The addition of DCBQ to the IO-ITO|PSII system gave rise to an 18-fold increase in photoresponse (265 mA cm À2 , EQE = 5.1%).This observation demonstrates that a significantly higher proportion of PSII was electrically connected to the electrode in the IO-ITO|P Os -PSII system compared to IO-ITO|P Phen -PSII, and that the IO-ITO|polymer-PSII electrodes were better connected than the IO-ITO|PSII system.Addition of bifunctional crosslinkers (PEGDGE for P Os 31 and 2,2 0 -(ethylenedioxy)diethanethiol for P Phen 6 ), to the IO-ITO|polymer-PSII systems resulted in no further photocurrent increase.This may be attributed to the stabilisation of the PSII-polymer matrix inside the 3-D-interconnected porous electrode framework. 9hese results indicate favourable interactions between the P Os and PSII, most likely between the side groups of the polymer (positively-charged Os 3+ complex, primary amine and imidazolium units) and the polar residues of PSII, 41,50 in particular the negatively charged region at the stromal side of PSII and near the Q A site. 37,51 In addition, a high number of electrochemicallyactive Os centres is estimated to be in close proximity to each PSII unit (based on the Os centre to PSII ratios (G Os /G PSII B 175)  This journal is © The Royal Society of Chemistry 2016 co-adsorbed on the electrodes), which explains the favourable photoelectrochemical response of the system discussed earlier.The P Phen can also interact with PSII via its hydrophilic side chains and residual epoxide groups to give rise to possible cross-linking. 41,50However, the P Phen is expected to have weaker interactions with the ITO electrode surface (Fig. S5e, ESI †), and is more likely to undergo polymer aggregation, as indicated by DLS, to result in significantly lower polymer entrapment and retention of PSII.The estimated number of toluidine blue units per PSII unit is 108, which is significantly lower than in the IO-ITO|P Os -PSII system.

Comparison of P Os and P Phen
In the preceding experiments, PF-PEC was used to systematically compare the performance of two benchmark polymers for PSII entrapment when they are integrated into high surface area electrodes.The P Os exhibited the most stable integration in 20 mm thick IO-ITO electrodes.When embedded with PSII, the IO-ITO|P Os -PSII electrodes delivered high photocurrent densities that were at least 5-fold higher than systems connected by P Phen (Fig. 5a).Despite the fact that P Phen is free of noble metals and has a better matched E 1/2 to the Q A and Q B (giving rise to earlier onset potentials for water oxidation), it exhibits lower adsorption stability on 20 mm thick IO-ITO electrodes.The IO-ITO|P Phen -PSII systems showed lower overall photoresponses compared to IO-ITO|P Os -PSII, which can also be rationalised by their more negative redox potential values (providing less driving force) and slower (H + diffusion-dependent) electron hopping process (2e À /H + vs. 1e À transfer, respectively).Overall, IO-ITO|P Os -PSII electrodes demonstrated higher performance and more efficient wiring between the PSII and the ITO electrode.

IO-ITO|P Os -PSII performance
To determine the enhancement of the photoresponse with film thickness in IO-ITO|polymer-PSII, IO-ITO electrodes with varying thickness (from 20 to 80 mm) were prepared and studied by PF-PEC.The focus was placed on the optimisation of the top performing IO-ITO|P Os -PSII systems.
The maximum loading of PSII and P Os on IO-ITO electrodes of different thicknesses are shown in Fig. 6a.P Os and PSII loadings increase linearly as the thickness rises from 0 to 80 mm.In comparison, an adsorbent saturation point was reached for IO-ITO|P Os -PSII electrodes beyond 40 mm.This was attributed to the accumulation of moderately viscous P Os -PSII aggregates over deposition time, which limits the penetration depth of the P Os due to the formation of channel blockages.No significant losses due to desorption upon long-term (60 min) immersion in the electrolyte solution with constant light irradiation were observed.
The dependence of photocurrent density on the IO-ITO|P Os -PSII electrode thickness is shown in Fig. 6b.A saturation photocurrent density of 381 AE 31 mA cm À2 (EQE = 6.9 AE 0.9%) for 40 mm thick electrodes was observed, which correlates with the maximum PSII loading reached at this thickness.Upon DCBQ addition, a further 1.35-fold photocurrent density increase was detected (513 AE 29 mA cm À2 , EQE = 9.3 AE 1.2%).The IO-ITO|PSII  The theoretical TOF PSII of water oxidation was estimated (assuming 100% Faradaic efficiency) according to eqn (5) for the IO-ITO|P Os -PSII electrodes of different thicknesses as shown in Fig. 6c.The maximum TOF PSII of 4.0 AE 0.4 s À1 was achieved using 20 mm thick IO-ITO|P Os -PSII electrodes, which could be increased to 6.7 AE 0.7 s À1 by the addition of DCBQ.This is a 1.7-fold increase compared to the IO-ITO|PSII system in the presence of DCBQ, and indicates that the mediated IO-ITO|P Os -PSII system is overall more efficiently wired than the mediated IO-ITO|PSII system due to the presence of the P Os matrix.
The long-term photostability of the IO-ITO|P Os -PSII system was evaluated at a relatively mild E app = 0.5 V vs. SHE and the results are presented in Fig. 6d.To determine the photocurrent half-life time (t 1/2 ), the photocurrent generated by IO-ITO|P Os -PSII electrode under continuous light irradiation for 60 min was recorded starting at the third photoresponse peak (Fig. S10, ESI †).Across the entire thickness range, the IO-ITO|P Os -PSII systems exhibited a 2-fold t 1/2 increase (maximum of 4.3 AE 0.4 min) compared to the IO-ITO|PSII systems (2.2 AE 0.2 min) in the absence of DCBQ.In the presence of DCBQ, further enhancement of the t 1/2 can be seen to reach B10 min in 80 mm thick IO-ITO|P Os -PSII electrodes.After 60 min of constant light irradiation, B7% and 11% of the initial photocurrent was detected from the IO-ITO|P Os -PSII electrode, without and with DCBQ addition, respectively.In contrast, less than 2% of the initial photocurrent was detected from the IO-ITO|PSII electrode.This can in part be attributed to the physical stabilisation of the PSII by the polymer matrix and the IO-ITO electrode architecture.The increased t 1/2 in the IO-ITO|P Os -PSII system can also be partly attributed to reduced accumulation of pigments in the excited state due to more efficient electron transfer between PSII and the Os centres in P Os . 49The higher efficiency in charge transfer would result in dampened formation of reactive oxygen species and deterioration of the D1 subunit in PSII. 52inally, the photocurrent generated by the IO-ITO|P Os -PSII electrode is high enough to enable the quantification of O 2 evolution (Fig. 7).Controlled potential electrolysis at E app = 0.5 V vs. SHE was carried out in a two-compartment cell in the glovebox employing an optimised 40 mm thick IO-ITO|P Os -PSII electrode upon light irradiation for 60 min (l = 660 nm, E e = 10 mW cm À2 ).The passage of 0.12 AE 0.03 C cm À2 charge was measured and the evolution of 0.24 AE 0.03 mmol O 2 cm À2 was detected by a fluorescence O 2 sensor, which corresponded to 85 AE 9% Faradaic efficiency.A turnover number TON PSII of 946 AE 96 mol O 2 (mol PSII) À1 , and an initial PSII-based TOF PSII of 3.6 AE 0.3 mol O 2 (mol PSII) À1 s À1 was calculated based on quantified O 2 and PSII using eqn ( 6) and ( 5), respectively.Previously, the generation of 0.23 AE 0.01 C cm À2 charge and the evolution of 0.45 AE 0.01 mmol O 2 cm À2 (75 AE 4% Faradaic efficiency), corresponding to TON PSII of 4200 AE 200 mol O 2 (mol PSII) À1 and TOF PSII of 12.9 AE 0.4 mol O 2 (mol PSII) À1 s À1 were reported for the IO-ITO|PSII system in the presence of DCBQ. 9 The absence of diffusion-limited mediators enables an all-integrated electrode design and eliminates problems such as those associated with concentration-dependent electron transfer.It also overcomes the issue of diffusional mass transport that may interfere with processes at the counter electrode and limit the performance of PSII-based PEC assemblies.Lastly, this electrode prototype allows all catalytic/electroactive material to be confined inside the porous electrode architecture, minimising the presence of high concentration catalytic/electroactive material in the bulk solution.

Conclusions
The present study has introduced a new benchmark PSII-based electrode, which was developed as a result of a rational design process that incorporated the best aspects of two leading enzyme immobilisation strategies.We integrated the stabilisation and efficient electronic wiring of enzymes within redox polymer matrices with the high loading capacity of hierarchicallystructured electrodes.This enabled the demonstration of high photocurrent densities, TOFs and levels of evolved O 2 that could be obtained for a PSII-driven PF-PEC system without the requirement for diffusional additives in the bulk solution.The photocurrents arising from PSII reported here also compare favourably with those reported for other wired photosynthetic proteins such as bacterial reaction centres 53 or photosystem I. 4,39,54 The development of this IO-ITO|polymer-PSII system provides the basic concepts needed for the future design of enzymedriven semi-artificial photosynthetic systems, including PEC tandem systems that incorporate other reaction centre or pigment-based proteins.We anticipate that this approach will also serve as an inspiration in the design of synthetic PEC watersplitting architectures.In the future, we expect that improvements in polymer design will lead to favourable changes to the electrode stability, electron hopping efficiency and formal redox potentials to better match the energy levels of the protein terminal electron acceptors.Lastly, hierarchical IO electrodes have demonstrated the potential to be highly versatile as a host system and may be used in various applications outside of PF-PEC, including batteries, fuel cells and solar cells.

Fig. 1
Fig. 1 (a) Schematic representation of PSII wired via a redox polymer network to a hierarchically-structured IO-ITO electrode (species size not drawn to scale), indicating the electron transfer from photoexcited PSII to the electrode via the redox-active centres.The SEM image of IO-ITO is also shown.(b) Energy level diagram showing electron transfer pathways betweenPSII, the redox polymer (P Phen or P Os , at pH 6.5) and the IO-ITO electrode (E app refers to the applied electrochemical potential, which determines the Fermi level at the ITO electrode).Abbreviations: P680 -primary electron donor site, Phe -pheophytin, Q A /Q B -electron acceptor plastoquinones, Tyr Z -tyrosine, Mn 4 Ca -oxygen-evolving complex.
[Os II Cl(1-(n-butyl)-imidazole)(bipy) 2 ](PF 6 ) for calibration, and confirmed by inductively-coupled plasma atomic emission spectroscopy (ICP-AES), obtained by washing off the P Os from the IO-ITO electrode with aq.conc.HNO 3 solution and measuring the concentration of the Os 2+ metal ions relative to Osmium ICP standard (1 mg Os mL À1 in 20% HCl, Ricca Chemical).Protein film photoelectrochemistry (PF-PEC) measurementsAll electrochemical experiments (with the exceptions of O 2 quantification and action spectra measurements) were performed with an Ivium Compactstat potentiostat with a purposebuilt monochromatic red-light LED lamp (l = 685 nm), collimated by two plano-convex lenses (THORLABS N-BK7 A Coated, + = 7.5 cm, f = 5.0 cm).A light intensity flux (irradiance) (E e ) of 10 mW cm À2 was used, unless stated otherwise.Chronoamperometry and cyclic voltammetry (CV) measurements were carried out in a water-jacketed glass one-compartment cell at 25 1C with a three-electrode setup using an IO-ITO working, a Pt wire counter and a Ag/AgCl (3 M KCl) reference electrode.Measurements of the IO-ITO|polymer-PSII system were carried out in 4 mL aqueous pH 6.5 electrolyte solution containing CaCl 2 (20 mM), MgCl 2 (15 mM), KCl (50 mM) and MES (40 mM).

Fig. 2
Fig. 2 (a) Chemical structures of the P Os and (b) P Phen polymers.(c) CV scans of P Os adsorbed on 20 mm thick IO-ITO (adsorbed 25 AE 4 nmol Os cm À2 ) showing excellent stability on the electrode surface.The first scan and the 100 th scan (dark, 10 mV s À1 ) are shown by the red solid trace and the dark green dotted trace, respectively.The second scan was the only scan measured during irradiation (l = 685 nm; E e = 10 mW cm À2 ), and is shown by the blue dashed line.A CV scan with P Os -modified flat ITO (dark, 10 mV s À1 ) is shown for comparison (adsorbed 1.7 AE 0.2 nmol Os cm À2 ).The inset shows a linear dependence of the peak current density J p with the scan rate n, confirming electron transfer of a surface-confined redox species.(d) Linear dependence of the redox-active centres loading for both polymers P Os and P Phen (up to 75 AE 3 and 96 AE 16 nmol cm À2 , respectively) with the electrode thickness confirms effective infiltration of the polymers into the electrode matrix.The error bars correspond to the standard deviation (N = 4).All experiments were carried out in a MES electrolyte solution (pH = 6.5, T = 25 1C).

Fig. 3
Fig. 3 Action spectra (solid traces) showing the photocurrent density (left Y axis) vs. irradiation wavelength of the IO-ITO|PSII (black), IO-ITO|P Os -PSII (red) and IO-ITO|P Phen -PSII (blue) photoelectrodes (20 mm thickness) recorded with monochromatic light measured in 20 nm steps (E e = 3.25 to 6.26 mW cm À2 ) at E app = 0.5 V vs. SHE (pH = 6.5, T = 25 1C) in MES electrolyte solution (see Fig. S7 for raw data and more detailed information, ESI †).The error bars correspond to the standard deviation (N = 3).The UV-vis absorption spectrum of the PSII (1 mL, 2.6 mg Chl a mL À1 ) in MES electrolyte solution (0.5 mL) (dashed green line, right Y axis) matches the photocurrent response of PSII on the electrodes.

Fig. 4
Fig. 4 Photocurrent density as a function of the applied potential (E app ) for the IO-ITO|polymer-PSII photoanodes determined by stepped potential chronoamperometry (pH = 6.5, T = 25 1C) (see Fig. S8 for raw data, ESI †).The inset shows a magnified region of the plot close to the onset potentials of the polymers.The photoresponse for PSII-free IO-ITO|polymer electrodes are shown for comparison.The error bars correspond to the standard deviation (N = 4).

Fig. 6
Fig. 6 Characterisation of the IO-ITO|P Os -PSII photoanode as a function of the electrode thickness: (a) PSII loading quantified by the amplitude of absorption at l = 665 nm, and Os 3/2+ redox centres loading determined by CV (Fig. S5b, ESI †); (b) photocurrent densities, (c) corresponding TOF PSII values and (d) photocurrent half-life times (t 1/2 ) measured upon light illumination (l = 685 nm; E e = 10 mW cm À2 ) at a fixed potential of 0.5 V vs. SHE without any additional diffusional mediator and upon addition of 1 mM of freely diffusing DCBQ mediator (pH = 6.5, T = 25 1C).The error bars correspond to the standard deviation (N = 4).

Fig. 7
Fig. 7 Quantification of O 2 evolution and determination of Faradaic yield (85 AE 9%) for the IO-ITO|P Os -PSII electrode (40 mm thickness) during continuous light illumination (l = 660 nm; E e = 10 mW cm À2 ) between 30 and 90 min with continuous stirring at E app = 0.5 V vs. SHE (pH = 6.5, T = 25 1C, red line).The chronoamperogram is shown in the inset.A control experiment in the absence of PSII is also shown (black curve).
144J s), c -speed of light (3.00Â 10 8 m s À1 ), J -photocurrent density (A m À2 ), e -electron charge (1.602 Â 10 À19 C), N A -Avogadro constant (6.022Â 10 23 mol À1 ), l -irradiation wavelength (m) and E e -light intensity flux (irradiance) (W m À2 ).14 9his journal is © The Royal Society of Chemistry 2016 (6.626 Â 10 2 (mol PSII monomer) À1 ) and t -time (s).9 almost ideal linear increase in photocurrent densities with the ITO film thickness, which is also consistent with the trend of PSII loading in IO-ITO|PSII electrodes.Maximum photocurrent values of 33 AE 5 and 577 AE 21 mA cm À2 from 80 mm thick electrodes were observed in the absence and presence of DCBQ, respectively.The comparable maximum photocurrent densities reached by the IO-ITO|P Os -PSII electrode in the absence of DCBQ and the IO-ITO|PSII electrode in the presence of DCBQ indicate efficient wiring of the PSII to the ITO surface by the P Os matrix.