Enhancing Oxygenic Photosynthesis by Cross-Linked Perylenebisimide “Quantasomes”

As the natural-born photoelectrolyzer for oxygen delivery, photosystem II (PSII) is hardly replicated with man-made constructs. However, building on the “quantasome” hypothesis (Science1964, 144, 1009−101117811607), PSII mimicry can be pared down to essentials by shaping a photocatalytic ensemble (from the Greek term ”soma” = body) where visible-light quanta trigger water oxidation. PSII-inspired quantasomes (QS) readily self-assemble into hierarchical photosynthetic nanostacks, made of bis-cationic perylenebisimides (PBI2+) as chromophores and deca-anionic tetraruthenate polyoxometalates (Ru4POM) as water oxidation catalysts (Nat. Chem.2019, 11, 146−15330510216). A combined supramolecular and click-chemistry strategy is used herein to interlock the PBI-QS with tetraethylene glycol (TEG) cross-linkers, yielding QS-TEGlock with increased water solvation, controlled growth, and up to a 340% enhancement of the oxygenic photocurrent compared to the first generation QS, as probed on 3D-inverse opal indium tin oxide electrodes at 8.5 sun irradiance (λ > 450 nm, 1.28 V vs RHE applied bias, TOFmax = 0.096 ± 0.005 s–1, FEO2 > 95%). Action spectra, catalyst mass-activity, light-management, photoelectrochemical impedance spectroscopy (PEIS) together with Raman mapping of TEG-templated hydration shells point to a key role of the cross-linked PBI/Ru4POM nanoarrays, where the interplay of hydrophilic/hydrophobic domains is reminiscent of PSII-rich natural thylakoids.


Equipment and Methods
NMR spectra were recorded on Bruker 400 Advance III HD equipped with a BBI-z grad probe head 5 mm and Bruker 300 equipped with a BBI-ATM-z grad probe head 5 mm. The chemical shifts (δ) for 1 H and 13 C are given in ppm relative to residual signals of the solvents (CHCl3 @ 7.26 ppm 1 H NMR, 77.16 ppm 13 C NMR). Coupling constants are given in Hz. The following abbreviations are used to indicate the multiplicity: s, singlet; d, doublet; t, triplet; q, quartet; m, multiplet; bs, broad signal. Diffusion ordered spectroscopy (DOSY) NMR experiments were carried out with the sample at a concentration of 2 mM in D2O to evaluate the aggregate size in solution. The value obtained for the translational diffusion coefficient is subsequently converted into the hydrodynamic radius applying the Stokes-Einstein equation D = kBT/(6πηR) (see Table S2).
Steady-state absorption spectroscopy studies have been performed at room temperature on a Varian Cary 5000 UV-Vis-NIR double beam spectrophotometer or on a Varian Cary 100 UV-Vis beam spectrophotometer. 10 mm path length Hellma Analytics 100 QS quartz cuvettes have been used.
Steady-state fluorescence spectra have been recorded on a Varian Cary Eclipse Fluorescence spectrophotometer or on a Horiba FluoroMax3 spectrophotometer; 10 mm path length Hellma Analytics 100F QS quartz cuvettes have been used.
Fluorescence decay dynamics studies have been performed using 405 nm laser pulses on a FLS1000 by Edinburgh Instruments equipped with a PMT-980 detector. 10 mm path length Hellma Analytics 100F QS quartz cuvettes have been used.
Fluoresce Quantum Yield Φf measurements were performed with Rhodamine 6G in ethanol (literature quantum yield 0.91 at 490 nm) as the standard. 4 The fluorescence quantum yields were calculated, exciting the samples at 480 nm, according to equation: F is the integral photon flux, f is the absorption factor, n is the refractive index of the solvent and Φf is the quantum yield. The index x denotes the sample, and the index st denotes the standard. The Φf reported in the work are averages of 5 measurements at different concentrations.
( ) = ( / ) + 0,205 + 0,0592 • . A LOT-QuantumDesign solar simulator, equipped with an AM 1.5 G filter, was used as the illumination source. The power of the light-source for light management experiments was measured using a THOR LABS PM100D power meter coupled with a CCD THOR LABS S370C and then to minimize the IO-ITO electrodes contribution, a 450 nm cut-off filter was used. Front illumination was adopted for all the experiments and unless otherwise stated, the majority of the measurements were recorded at 850 mW/cm 2 in order to better appreciate the photocurrent delivered by the photoanodes. Current-voltage curves were all recorded at a scan rate of 10 mV/s. Current-voltage and current-time curves under chopped illumination (10 s) were acquired by manually chopping the excitation source. All the photoelectrochemical data represent the average of three different samples. All the experiments were performed before on blank samples and then on co-deposited photoanodes. The electrodes were cleaned electrochemically scanning three times in dark CV (range 0.62 -1.62 V vs RHE) until reaching a superimposable current response. This procedure enabled the elimination both of impurities coming from electrodes synthesis and of halides (i.e bromide) present in PBIs that could affect the measurements by behaving as redox mediators. 7 In particular, regarding IO-ITO|QS-TEGLock and IO-ITO|QS-TEGunlock, EDX experiments after the electrochemical cleaning procedure confirmed the absence of Br signal (Scheme S1). Electrochemical analysis of voltammetric scans after cleaning procedures (spanning from 0.62 V to 1.62 V vs RHE) is reported in section 3.21 for IO-ITO|QS, or QS-TEGlock and QS-TEGunlock electrodes.
A) EDX experiments on IO-ITO|QS-TEGlock 12 nmol cm -2 after electrochemical cleaning procedure. where Φ − and Φ ℎ are the flux of electrons and incident photons (mol/s m 2 ), respectively, Jλ is the steady state photocurrent density (µA/cm 2 ) generated at λ (nm), F is the Faraday constant, Iλ is the irradiance (W m -2 ), NA is Avogadro's number and Eλ = hc/λ × 10 -9 is the monochromatic photon energy (eV). The APCE value was then calculated from the IPCE data based on the absorption features of the photoanode, calculated as the light harvesting efficiency (LHE) at each wavelength, according to equation:

B) EDX experiments on IO-ITO|QS-
where T(λ) is the transmittance and A(λ) is the apparent absorbance calculated from reflectance at λ.

Determination of Faradaic Efficiency
In order to measure the O2 evolved from the photoelectrochemical water oxidation by IO-ITO|QS or QS-TEGlock and QS-TEGunlock, the Generator-Collector method, originally introduced by the Mallouk group and subsequently adapted by Meyer, Finke and our group, was employed. 1,[8][9][10] Briefly, the IO-ITO|QS or QS-TEGlock and QS-TEGunlock photoanode was illuminated under an applied bias, thus acting as O2 generator. An FTO electrode (previously cleaned via 10 min sonication in KOH/iPrOH, 10 min sonication in iPrOH and annealed at 500 °C for 30 min) was sandwiched to the photoanode (with both conducting sides facing inward), in order to be used as the collector, i.e. the electrode at which the reduction of the evolved O2 takes place. The sandwiched device was held together by 3 layers of unstretched parafilm (ca. 200 µm spacing), sealed together by employing a cell made with polyether ether ketone engineered by our group (see Scheme S2). The parafilm was designedly cut to form a U-shaped chamber to minimize O2 loss, while still allowing the electrolyte access by capillary forces. Both the photoanode generator and the FTO collector were contacted using Cu tape (also covered by the parafilm layers), and respectively connected to the two working electrodes of a bipotentiostat (PGSTAT302N), while an Ag/AgCl and a Pt wire were used as the reference and the counter electrode respectively. The sandwich was then immersed in degassed NaHCO3 0.1 M solution (pH 7) and illuminated at 850 mW/cm 2 (using a calibrated LOT-QuantumDesign solar simulator, equipped with an AM 1.5 G filter plus a 450 nm cut off filter to minimize IO-ITO contribution). The generator electrode was held at 1.12 V and 1.52 V vs RHE bias, while the collector at -0.28 V vs RHE, which was identified to be the optimal O2 reduction potential. In a typical experiment, currents at both the working electrodes were recorded for 100 s in the dark, then 100 s of illumination, then in the dark for additional 300 s, to allow the diffusion of oxidation products across the solution. The faradaic efficiency for O2 production, ηO2, can be calculated by: where Qcoll is the integrated current measured at the Collector electrode, Qgen is the integrated photocurrent measured at the Generator electrode, and ηcoll is the collector efficiency (which must be quantified under the specific set-up and experimental conditions used). In particular, to the registered current values, we have subtracted the corresponding stable current values measured in the initial dark step. This value was usually negligible in the case of photoanodes, while on the FTO collector the reduction of some residual O₂ was registered. The corrected current values were then integrated (to yield Qgen and Qcoll) using the "subtract baseline" tool in Origin, in order to account also for the current registered in the final dark step of the experiments. As regards ηcoll, it has been estimated to be 78%, from generator-collector experiments registered on FTO-FTO sandwiches prepared in the same way of the analogous photoanode-FTO ones. Firstly, it was registered the dark current associated to water oxidation on FTO in NaHCO3 0.1 M solution (pH 7), then once the potential at which the faradaic current begins was identified, one of the two FTO (which acts as the generator) was held at such potential (up to 1.70 V vs RHE), while the other FTO electrode (acting as the collector) was set at -0.28 V vs RHE (the optimal oxygen reduction potential). Current at both working electrodes was recorded, subtracting at the generator the current at open circuit potential, i.e. before applying the oxygenic overpotential. The corrected current values were then integrated on origin allowing the calculation of the collection efficiency from the ratio between the areas calculated. In order to measure O2 evolved from the electrochemical water oxidation by IO-ITO|QS or QS-TEGlock and QS-TEGunlock the same method was employed, registering dark current in NaHCO3 0.1 M solution (pH 7) in the potential window used to screen photoanodes. IO-ITO|QS or QS-TEGlock and QS-TEGunlock photoanodes were held at applied bias where faradaic dark current start to be appreciable, thus acting as O2 generators. In particular one potential was identified: the one at which falls the first dark oxygenic manifold (i.e. 1.40 V vs RHE for IO-ITO|QS and 1.45 V vs RHE for IO-ITO|QS and QS-TEGlock and QS-TEGunlock). While the FTO collector was held at -0.28 V vs RHE. Current at both the working electrodes was recorded, subtracting at the generator the current delivered at non faradaic potential (i.e. 1.00 V vs RHE). The corrected current values were then integrated on origin using the same procedure named before, allowing the calculation of the faradaic efficiency from the ratio between the areas calculated.
Picture of the cell made with polyether ether ketone and sealed with Teflon screws along with a Generator-Collector sandwich.

Preparation of IO-ITO
FTO cleaning was performed by sonicating the slides in Alconox® solution, in Milli-Q water and then in 2propanol, every time for 10 minutes. After every step, FTO was rinsed with Milli-Q water. Subsequently, electrodes were sintered at 450 °C for 30 min. IO-ITO was prepared following a literature procedure. 11 A mixed dispersion of ITO nanoparticles (< 50 nm diameter) and polystyrene beads (750 nm diameter, 2.5% solids (w/v)) was prepared as follows: ITO nanoparticles (35 mg) were dispersed by sonication in 300 uL of a MeOH/water (6:1 v/v) mixture (300 μL taken from a mother solution of 10 mL) for 3 h. The dispersion of polystyrene beads (1 mL) was centrifuged for 1 hour at 4000 rpm, the supernatant removed, and the polystyrene pellet redispersed in MeOH (1 mL). The polystyrene dispersion was centrifuged again with same parameters. The supernatant was removed, and the dispersion of ITO nanoparticles added to the polystyrene pellet. This mixture was thoroughly vortexed and sonicated for 5 min in ice cold water (< 5 °C) to give the polystyrene−ITO dispersion. 30 minutes before deposition, electrodes were placed in a chamber with controlled atmosphere of the mother solution (MeOH/water, 6:1 v/v). On the electrodes, a magic tape mask to delimit the geometrical area of the deposition to 0.25 cm 2 was placed. An amount of 4.2 μL of the described polystyrene−ITO dispersion were deposited corresponding to a 10 ± 2 μm thick IO-ITO structure. The electrodes were then heated with 1°C min −1 to 500 °C and annealed at this temperature for 20 min. The IO-ITO electrodes were then cleaned by placing them in a mixture of 30% H2O2/ H2O / 30% NH4OH (1:5:1 v/v) at 70 °C for 15 min, rinsed with water, and heated for 1 h at 180 °C to give a contamination-free hydrophilic ITO surface.

Determination of IO-ITO ECSA (Electrochemical Surface Area) by Double-Layer Capacitance Measurements
Experiments were carried out using an Autolab PGSTAT302N potentiostat in a three-electrode setup using as counter electrode Au and as reference electrode Ag/AgCl (NaCl 3M). Potentials are then converted to RHE using the correlation ( ) = ( / ) + 0,205 + 0,0592 • . CV scans for IO-ITO were recorded at scan rates in the range of 2 − 50 mV/s, spanning ca. ± 30 mV of the OCP, a range where no faradic processes occur. 12 The current values were divided by the geometric area of the electrodes. From the CV traces, the capacitive current was then calculated as (Ja − Jc)/2, where Ja and Jc are, respectively, the anodic and cathodic current densities at OCP. The resulting values (in A/cm 2 ) were plotted against the scan rate of the CV experiments (in V/s) and the data fitted with a linear equation. The slope of the linear regression gives the capacitance of the electrode (in F/cm 2 ). Assuming the FTO to be featureless (roughness factor, RF = 1 by definition), the RFs of IO-ITO electrodes can be calculated by dividing the corresponding capacitance values by the capacitance of the FTO foil used as the reference. For each anode, the OCP value was directly read on the potentiostat display after connecting all the three electrodes. The reading was stable.

Quantasome deposition on IO-ITO electrodes
IO-ITO|QS and QS-TEGlock electrodes were prepared by co-deposition of the quantasome building blocks with 5:1 PBI/Ru4POM stoichiometry yielding a nominal quantasome loading of 2.4, 4.8, 7.2, 9.6, 12 nmol cm -2 (Scheme S3). The optimized deposition protocols is as follows: aliquots of a Ru4POM solution in water are sequentially codeposited with aliquots of the PBI-derivatives solution in water, as specified in the following Table, in order to reach the desired quantasome loading amount. 10:1 and 5:2 PBI/Ru4POM stoichiometries were also codeposited by varying the concentration of the mother solutions and the deposited aliquots. Scheme S3. Concentration and aliquots deposited of the mother solutions used in co-deposition.

Determination of surface-active sites of Ru4POM
The active sites of Ru4POM deposited on IO-ITO|QS or QS-TEGlock and QS-TEGunlock were calculated from the oxidation redox active peak at ca. 0.80 V vs RHE associated to a monoelectronic process as reported in Pourbaix diagram from literature. 13,14 Figure S30 shows an example of the calculation of the oxidation peak area of Ru4POM extrapolated from cyclic voltammetry recorded at 10 mV s -1 in sodium hydrogen carbonate NaHCO3 0.1M (adjusted at pH 7). 15 Calculated area and charge are reported in Table S17. The number of electrons is extracted from charge. From the number of electrons calculated above, moles of Ru4POM active sites can be obtained dividing for Avogadro constant. 16

PBI-YNYL:
A suspension of 200 mg of PBI-1 (0.37 mmol, MW = 532) in 1.6 ml of propargyl bromide 80% solution in toluene was stirred at 50°C overnight. The following day the product was precipitated adding 25 ml of THF to the mixture, which was consecutively filtered on a gooch washing abundantly with THF. The dry solid was then redissolved in milliQ water and filtered on a gooch to remove unreacted material. The filtrated was recovered and water was removed by rotavapor yielding 44 % of PBI-YNYL.

Synthesis of TEG-OTs
Adapted from previous reported procedure 16 TEG-OTs: tetraethyleneglycol (0.35 g, 1.8 mmol) was added dropwise to a solution of p-toluenesulfonyl chloride (0.70 g, 3.7 mol) in anhydrous pyridine (5 mL) stirring in an ice bath at 0 °C, over 3 h. The resulting mixture was added to 10 mL ice water and extracted with CH2Cl2 (3×10 mL). The organic layer was then washed with 20 mL of HCl 6 M, followed by 20 mL of saturated NaCl solution and dried over anhydrous MgSO4. The solvent was removed under reduced pressure affording a yellow oil in a 30% yield (0.27 g) without further purification.

Synthesis of TEG-N3
Adapted from previous reported procedure 16 TEG-N3: NaN3 (130 mg, 2 mmol) and TEG-OTs (200 mg, 0.5 mmol, MW = 502) were dissolved in anhydrous DMF. The solution was degassed with N2 and stirred at 50°C overnight. After cooling to room temperature, 100 ml of ice water were added to the reaction mixture and subsequently extracted with DCM. The organic fraction was washed with saturated NaCl and dried over magnesium sulphate to yield a 100 mg of a yellow solid in 80 % yield.

Synthesis of PBIn-TEGlock
Adapted from previous reported procedure 16 PBIn-TEGlock: 20 mg PBI-YNYL (0.026 mmol, MW = 770) and 20 mg of TEG-N3 (0.08 mmol, MW = 244) were dissolved in 10 ml of MilliQ water and degassed with N2 for 40 min. In a second flask 8 mg of CuSO4·5H2O and 4.5 mg of sodium ascorbate were dissolved in 5 ml degassed MilliQ water and transferred to the former flask. The solution was stirred at room temperature overnight under N2. The following day the aqueous solution was extracted three time with DCM to remove the unreacted azidoglycole. The solution was concentrated under evaporation and passed through a Sephadex G50 column conditioned with water. The fractions were preliminary evaluated by UV-Vis in DMF to identify the successful locked aggregation system. Once the aqueous fraction was concentrated, acetonitrile was added and the solid was centrifugated three times before yielding 18 mg of a dark red solid. The molecular weight of a PBIn-TEGlock monomer is estimated to be 1002 resulting in 70 % yield. 1

Synthesis of 2-(2-(2-methoxyethoxy)ethoxy)ethyl p-toluenesulfonate
Adapted from previous reported procedure 17 2-(2-(2-methoxyethoxy)ethoxy)ethyl p-toluenesulfonate: 2 mL of polyethylenglycol monomethylether (12.67 mmol) was added dropwise to a solution of 3.14 g of p-toluenesulfonyl chloride (16.5 mmol) and 3.6 mL of triethylamine (25 mmol) in 130 mL of MeCN at 0 °C. The mixture was stirred for 3 h at room temperature. The suspension was filtered and the solvent was removed under reduced pressure. The residue was dissolved in DCM and washed with 100 mL of 10% hydrochloric acid, followed by 100 mL of saturated NaCl solution and dried over anhydrous magnesium sulfate. The solvent was removed under reduced pressure affording a yellow oil. The product was separated by chromatographic column (n-hexane/ethyl acetate 3:1) to furnish a yellow oil in a 50% yield.

Synthesis of PBI-TEGunlock
Adapted from previous reported procedure 16,17 PBI-TEGunlock: 25 mg PBI-YNYL (0.032 mmol, MW = 770) and 155 mg of TEG-OMe (0.08 mmol, MW = 189) were dissolved in 20 ml of Milli-Q water and degassed with N2 for 30 min. In a second flask 10 mg of CuSO4•5H2O and 8 mg of sodium ascorbate were dissolved in 10 ml degassed MilliQ water and transferred to the former flask. The solution was stirred at room temperature overnight under N2. The following day the aqueous solution was extracted three times with diethyl ether to remove the unreacted azidoglycole. The solution was concentrated under evaporation and passed through a Sephadex G50 column conditioned with water. Once the aqueous fractions were concentrated, acetone was added and the solid was centrifugated three times before yielding 22.8 mg (62%) of a dark red solid.  Figure S1. FTIR of the precursor PBI-YNYL and of the locked PBIn-TEGlock in the region 3600 cm -1 -1000 cm -1 . Figure S2. Zoom in the 1800 to 1000 cm -1 region of normalized FTIR of the precursor PBI-YNYL and the locked PBIn-TEGlock. Figure S3. 1 Figure S4. Superimposed absorption spectra of PBI-YNYL and PBIn-TEGlock in DMF showing the multi-PBI aggregation of the locked superstructure. Figure S5. 1 Figure S9. A) UV-Vis spectrophotometric titration of PBIn-TEGlock (12.5 µM, blue trace) in milliQ water with Ru4POM (up to 2.5 µM, 0.2 eq., red trace). B) plot of the absorbance at 500 nm versus the equivalents of Ru4POM. C) Emission spectrophotometric titration of PBIn-TEGlock (12.5 µM, blue trace) in milliQ water with Ru4POM (up to 3.12 µM, 0.25 eq., red trace). Excitation wavelength 500 nm. D) plot of the emission intensity at 550 nm versus the equivalents of Ru4POM. E) ζ-Potential titration of PBIn-TEGlock (12.5 µM) upon addition of Ru4POM (up to 15 µM, 0.6 eq.) in milliQ water.  Figure S11. SEM images at different magnification factors of IO-ITO samples prepared. A)-C) cross-section images. D)-F) top view images. Measured thickness is 10 ± 2 m. Pores dimension is 600 ± 100 nm.  Table S3.                             1.92 ± 0.06 a Onset potential, b peak potential and c peak current of the first oxygenic manifold in 0.1 M NaHCO3, pH 7, scan rate 10 mV/s; d Dark current related to water oxidation at 1.62 V vs RHE in 0.1 M NaHCO3, pH 7, scan rate 10 mV/s. e Turnover frequency of the catalyst calculated from registered current at 1.62 V vs RHE considering the active sites measured in section 3.20. Every parameter is obtained from the mean value of at least 3 different samples.       Figure S36. Prolonged photoelectrochemical response of IO -ITO|QS-TEGlock (red trace) and of IO -ITO|QS-TEGunlock (green trace) (12 nmol cm -2 loading) registered by chronoamperometry (CA) at 1.12 V vs RHE in 0.1 M NaHCO3, pH 7, with solar simulator equipped with AM 1.5 G filter, 850 mW cm -2 , λ > 450 nm.  38 % 36 % a Applied sun power. b Loss of reflectance intensity at 500 nm converted in Kubelka Munk units after recording experiments at different light intensities (F(R) = intensity after 300 s chronoamperometries at 1.12 V and 1.52 V vs RHE; F(R)0 = intensity of the photoanodes freshly deposited). Samples prepared via co-deposition of 12 nmol cm -2 of QS or QS-TEGlock on IO-ITO electrodes. Every measure is repeated at least 3 times and the value here reported is the mean value.