Polymeric Benzothiadiazole, Benzooxadiazole, and Benzoselenadiazole Photocathodes for Photocatalytic Oxygen Reduction to Hydrogen Peroxide

Visible‐light‐driven semiconductor photoelectrodes are promising new devices for on‐demand photocathodic generation of hydrogen peroxide. Herein, the fabrication of organic polymeric photocathodes employing poly(4,7‐di(thiophen‐2‐yl)benzo[c][1,2,5]thiadiazole) (pThBTD), and comparatively seven other related derivatives is reported. The monomer dithienobenzodithiazole can be directly polymerized on an electrode via two methods: electropolymerization or iodine‐vapor‐assisted polymerization. Both give polymers with wide visible light absorption and suitable stability for photoelectrodes. These methods yield different active layer morphologies, with electropolymerization yielding photocathodes with better performance. Critical issues affecting oxygen reduction photocurrents are evaluated, namely thickness, wettability, and pH. Photocathodic oxygen reduction currents, as well as photovoltages, are among the highest reported for an organic photoelectrocatalyst, and pThBTD films can stably produce H2O2 with high faradaic yield over at least 8 h. This study shows that single‐component organic semiconductor devices can be highly competitive versus more complex heterostructures and that such low‐bandgap organic polymers can afford remarkable stability.


Introduction
Catalytic oxygen reduction reactions (ORR) to hydrogen peroxide have recently gained attention as an attractive concept both for sustainability and industrial/agricultural applications, [1][2][3] as well as biology and physiology where local generation of peroxide is a Figure 1. Operation principles of a polymeric semiconductor photocathode for oxygen reduction reactions to peroxide. Light absorption by the polymer results in the formation of excitons. Oxygen accepts the excited electron and leaves a hole which is transported through the polymer and extracted at the underlying electrode. Light absorption, selective one-or two-electron reduction of oxygen, and good p-type conductivity are all essential for the successful operation of a polymeric photocathode to produce hydrogen peroxide. To maximize peroxide accumulation, a membrane or salt bridge (symbolized by the dotted blue line) is used to separate the cathode and anode chambers to prevent peroxide reoxidation at the counter electrode.
photovoltaic effect and a photocurrent of negative polarity. Photocathodic operation involves light absorption by an organic semiconductor and subsequent transfer of photoexcited electrons to oxygen, acting as an electron acceptor. Positive charges remain in the semiconductor layers, and are transported to the underlying conductor layer (photocathodic current). Suitable photocathodic materials must therefore have p-type electronic transport in order to operate. This photocathodic process is schematized in Figure 1.
The first example of photocathodes based on organic materials for peroxide generation was published in 2016, featuring hydrogen-bonded small molecules as active materials. [8] These materials demonstrated high faradaic efficiencies (90%+) for peroxide generation, however, they had spectral absorption limited to the green and blue parts of the spectrum. More efficient photocathodes should absorb further into the red region of the spectrum. This is critical for applications involving solar energy harvesting, or biological applications which necessarily rely on light in the red region for safe penetration through the biological sample. The archetypical p-type polymeric semiconductor poly(3-hexylthiophene) (P3HT) has been demonstrated by several research groups [4,9,10] to be a suitable candidate for photocathodic hydrogen peroxide generation, however, it has limited absorption (primarily green light) and poor stability, suffering from photo-oxidative degradation. [10,11] A number of reports from the group of Winther-Jensen have explored the optimization of polythiophene-based photocathodes, as well as various copolymers which afford better performance and stability than P3HT. [12] Aside from improved photocathodic performance, these studies have introduced the notion of fabricated photocathodes via direct polymerization from monomers at the electrode surface, either by electrochemical polymerization or chemical polymerization using iodide vapors. [13] This route to photocathodes increases the range of possible polymeric materials beyond those obtainable by solution processing. This detail is important, since solution-processable semiconducting polymers are usually soluble in organic solvents, necessitating the use of hydrophobic side chains. This hydrophobicity is problematic for aqueous photocathode operation, as the photochemical reduction of oxygen requires protons from water. Therefore, direct polymerization of monomers can lead to semiconducting films that do not dissolve in water, but are also not restrictively hydrophobic as to impede photocathodic activity. Aside from favorable wettability, direct polymerization at the electrode can offer possibilities of morphology tuning. Perhaps most interestingly, from a synthetic chemistry point of view, this route enables testing of a wide range of monomeric species which otherwise can be challenging to obtain and process in polymeric form. Currently, the highest photocurrents and photovoltages for organic photocathodes are achieved by donor-acceptor heterojunction structures, with optimized charge generation, charge transport, and catalytic layers. [14,15] These photocathodes achieve ORR current densities approaching the maximum possible ones for diffusion-limited ORR, and faradaic yields for peroxide production in the range of 80%-90%. However, these structures are complex and fragile, and suffer from morphological instability. Simpler photocathode architectures, offering stability yet preserving the same high photocurrent and faradaic yield performance, are a worthwhile goal.
The aim of our work is to attempt to achieve the best possible photocathodic oxygen-to-peroxide performance, using simple donor-acceptor units polymerized directly at the electrode. To do this, we target the following building blocks: benzothiadiazole, benzooxadiazole, and benzoselenadiazole as acceptor units, and various thiophene-based donor components (Figure 2). These acceptor units are relatively well-studied in the context of polymeric materials tailored for bulk heterojunction solar cells. In state-ofthe-art bulk heterojunction organic solar cells, a key principle is the use of donor-acceptor polymers which account for excellent broad-band absorption of light and good, usually ambipolar, charge transport. Nevertheless, such materials are still excitonic in nature, meaning that light absorption leads to a strongly bound exciton. For photovoltaic applications, this necessitates the mixing of at least two polymers, to form a heterojunction to polarize the exciton into free charge carriers. However, if the goal is ORR, oxygen can act as the electron-acceptor to polarize the exciton (as shown in Figure 1), therefore in principle, a polymer with broad light absorption and good charge transport could achieve the same photocathodic efficiencies as an optimized multilayer heterojunction structure. As this work will show, through judicious selection of material and optimization of deposition parameters, it is possible to achieve single-material photocathodes which yield the same performance as a complex multilayer junction.

UV−vis Spectroscopy Characterization
The UV-vis absorption spectra of studied compounds dissolved in dichloromethane or toluene were recorded using a Hitachi UV-2300II spectrometer. The emission spectra of solutions were recorded using Edinburgh FS5 equipped with an enhanced range photomultiplier detector (PMT-EXT). All measurements were carried out at room temperature, according to published procedures. [24] Suprasil quartz cuvettes (10.00 mm) were used. 1.5 nm slits were used for absorption and emission spectra. To eliminate any background emission, spectrum of pure solvent was subtracted from the samples' spectra. The photoluminescence quantum yields (Φ) were determined in diluted solutions (A ≈ 0.1 for longest wavelength band) by comparison with known standards -fluorescein in 0.1 M NaOH (Φ = 0.92). [25] The concentration of compounds was adjusted to reach similar absorbance to absorbance of reference solution at the excitation wavelength ( = 465 nm).. Emission spectra of thin films were measured using front-face geometry. Quantum yields of thin films were measured using an integrating sphere (Edinburgh Instruments) according to a known procedure. [26] Thin films were obtained via drop casting on quartz substrates. Toluene solutions were deposited on quartz plates, let to dry, and vacuumdried. Samples were excited at 450 nm. Fluorescence lifetime measurements were acquired using Time Correlated Single Photon Counting system equipped with a picosecond pulsed 340 nm EPLED source. Absorption spectra of the polymer thin films deposited on an ITO electrode were recorded on a Varian Cary 5000 spectrometer.

Electrochemical Characterization
Cyclic voltammetry (CV) and differential pulse voltammetry experiments were performed in a one-compartment three-electrode electrochemical cell with a platinum disk (surface area = 2 mm 2 ) or ITO/FTO working electrode, platinum counter electrode, and an Ag/0.1 M AgNO 3 /CH 3 CN reference electrode, the potential of which was verified at the end of each set of experiments using the ferrocene couple (Fc/Fc + ). The monomer concentration was 1 × 10 −3 M in an electrolyte consisting of 0.1 M Bu 4 NPF 6 in anhydrous CH 2 Cl 2 . For all electrochemical experiments, a potentiostat (Autolab, EcoChemie) was used.

Polymer Film Preparation Methods
Thin films of all investigated polymers were deposited via electropolymerization either on a platinum disk, an ITO (indium-tin oxide)-coated glass slide, or on an FTO (fluorine-doped tin oxide) coated glass. FTO/ITO samples were used for photoelectrochemical investigations and determination of H 2 O 2 concentration, as well as measurements of the polymers in organic electrolyte: 0.1 M Bu 4 NPF 6 in CH 3 CN. Additionally, in the case of Th-BTD, the iodine-vapour-assisted polymerization (IVP) technique was used as an alternative to electropolymerization. IVP was performed as follows: A Th-BTD solution in chloroform (0.25 ml) was spin-coated onto an ITO substrate (previously ultrasonically cleaned in a series of solvents: water, isopropyl alcohol, and chloroform) at 1500 rpm for 3 s followed by 600 rpm for 25 s. The Th-BTD coated sample was placed in a glass chamber with iodine and polymerized under rt conditions over 24 h. Then the sample was removed from the chamber and carefully washed with CH 3 CN and dried in air.
In the selected cases, ITO substrates were additionally modified through silanization. To achieve this, cleaned ITO substrates were placed for 15 min in a vessel with a mixture of 25% ammonia, 30% hydrogen peroxide, and water (1:3:5 v/v/v) at 60°C. Subsequently, after rinsing with distilled water, they were transferred to a chamber with triethoxy(octyl)silane (OTS) and heated to 80°C for 60 min. In the following step, the layers were cleaned of residual silane by sonification in acetone.

Photoelectrochemical Measurements
Photoelectrochemical measurements of the photocathodes were performed in a three-electrode system, with a Pt counter electrode, and an Ag/AgCl wire in 3 M KCl as a reference electrode, using an Ivium technologies Vertex.One potentiostat. All electrodes were placed in a Redox.me MM PEC double-sided cell, containing 15 ml of electrolyte in all cases. To measure the faradaic efficiency, the system was equipped with a salt bridge that was submerged in an external chamber, along with a Pt counter electrode. The active area of the ITO/FTO sample mounted in the cell was limited to 1.0 cm 2 . The cyclic voltammetry experiments were performed with a 50 mV s −1 scan rate within +0.4 V to −0.4 V range, at pH 1, pH 7, and pH 12 in the dark and under 100 mW cm −2 (1 sun) irradiation. As a light source, a halogen lamp 93638 EKE OSRAM with water filter was used, after its calibration with light meter to a desired light intensity. All samples were illuminated from the substrate side, i.e., through the ITO/FTO. Chronoamperometry was conducted at constant potential of 0 V versus Ag/AgCl under pulsed (20 s or 60 s) irradiation, the same bias was used in the long experiment of H 2 O 2 photogeneration. The assembled cell was purged with argon for 10-20 minutes for inert conditions or constantly with O 2 for oxygen reduction conditions.

Characterization of Photocathode Surface Morphology
SEM images were taken with SEM Zeiss Sigma-500 at an acceleration voltage of 3-6 kV, using the in-lens secondary electron detector. The samples were washed with DI water and dried with a stream of N 2 before imaging. The uncovered areas of samples were contacted to the SEM stage with copper tape and placed in the microscope chamber.

Determination of Hydrogen Peroxide Concentration and Faradaic Efficiency
Hydrogen peroxide produced during the experiment was quantified using spectrophotometric analysis. Herein the oxidation of 3,3′,5,5′-tetramethylbenzidine (TMB) in the presence of horseradish peroxidase and citric acid-phosphate buffer solutions was monitored, and absorbance values were measured at a wavelength of 653 nm using a Varian Cary 5000 spectrometer. Depending on hydrogen peroxide concentration in the sample, different volumes of aliquots (up to 10 μL) were taken and added to corresponding volumes of HRP/TMB/buffer solution, resulting in a total volume of 1000 μL. Based on calibration curve calculations, the resulting absorbance values were converted to concentration values (R2 > 0.998 in every case). The Faradaic efficiencies of the photoelectrolysis processes were determined based on the concentration of hydrogen peroxide, electrolyte volume, and theoretical molar amount of hydrogen peroxide generated.

Electrochemical and Spectroscopic Studies
To prepare polymers with an alternating donor-acceptor-donor architecture, nine suitable monomers were first synthesized (as shown in Figure 2 and Schemes S1-S2, Supporting Information). Optical and electrochemical characterization of all monomers were collected in the Supporting Information (as shown in Figures S1-S4 and Tables S1-S2, Supporting Information). The studied compounds can be classified into two groups based on the central acceptor and donor units used. The first group consists of molecules where acceptor units (benzothiadiazole, benzooxadiazole, benzoselenadiazole, and naphthobisthiadiazole) were symmetrically substituted with thiophene. The second group investigates the electro-donating effect of different moieties such as thiophene, thienothiophene, bithiophene, and ethylenedioxythiophene on the same benzothiadiazole core.
For eight of the monomers, uniform polymer films were deposited on the working electrode through oxidative electropolymerization under potentiodynamic conditions. Electropolymerizations were carried out by cyclic voltammetry (CV) with 10 cycles in 0.1 M Bu 4 NPF 6 in CH 2 Cl 2 within the range between 0.75 V and 1.00 V (vs Fc/Fc + ). Representative electropolymerization scans are shown in Figure S2 of the Supporting Information. Only one of the nine studied compounds, Th-BTDCN, did not electropolymerize. This can be attributed to the high oxidation potential value of this compound (E ox > 1.25 V vs Fc/Fc + ), which exceeds the stability window of the electrolyte employed in the experiment. Consequently, the complete current response of the anodic process could not be recorded, nor was it possible to deposit a polymer layer of Th-BTDCN using electrochemical techniques. Nevertheless, a two-step reduction process leading to the formation of the radical anion and dianion, respectively, can be observed in this case ( Figure S1, Supporting Information). Notably, the peak registered in the positive potential range of the CV of Th-BTDCN originates from the solvent used for the measurement. Figure 3 compares cyclic voltammograms of electropolymerization products. After 10 cycles, polymer films on a platinum electrode surface were washed with CH 2 Cl 2 and taken into a monomer-free electrolytic solution containing 0.1 M Bu 4 NPF 6 in CH 3 CN. All electrochemically deposited films exhibited electroactivity for both negative and positive potentials. However, their redox potentials differed depending on the electronaccepting properties of their central unit and electron-donating properties of peripherally substituted moieties. Cyclic voltammograms of polymers showed quasi-reversible reduction and irreversible oxidation. Reduction of polymers with benzothiadiazole core (pTh-BTD, pBTh-BTD, and pEDOT-BTD) started at lower potentials than in polymers with other central units studied in this research, which manifests the weakest electron-accepting properties of BTD. In the case of pTh-BBTD, where two benzothiadiazole rings are conjugated, it could be expected that the reduction of the core should proceed in a two-step manner. However, both the monomer and polymer CVs ( Figure S1f and S3, Supporting Information) clearly showed only one current signal originating from the formation of the anion radical. Electrochemically determined ionization potentials (IP) of benzothiadiazole derivatives were found to strongly depend on the nature of the donor unit, whereas the difference range of electron affinity (EA) values is smaller (see Table 1). Replacing the thiophene moiety with thienothiophene (pTT-BTD), bithiophene (pBTh-BTD), and ethylenedioxythiophene (pEDOT-BTD) resulted in significant differences in IP values, with the highest value observed for pTT-BTD. This indicates that pTT-BTD is more difficult to oxidize compared to its analogs, especially pEDOT-BTD, which readily oxidizes in the air but is hard to reduce. Moreover, introducing fluorine substituents into the BTD core (pTh-BTDF) led to an increase in the IP to a value of 5.74 eV compared to pTh-BTD (5.44 eV). The presence of electronegative fluorine atoms in the polymer structure, which are known to exert an inductive effect, causes changes in the electron density within the thiadiazole ring and thiophene units. As a consequence, the formation of the anion radical is facilitated, while the ionization potential increases.
Simultaneously, the current magnitude of the polymer with the EDOT moiety is comparatively higher than that of thiophene, thienothiophene, and bithiophene. This is caused by lower oxidation potential and higher capacitance of the material. [27] For spectroscopic investigations, thin polymer layers were electrodeposited on an ITO substrate. UV-vis spectra with appropriate photographs are shown in Figure 4 and the absorption maxima are collected in Table 1. A broad spectral absorption is evident ranging from ≈450 nm to greater than 800 nm for all polymers. For instance, the benzoselenadiazole band is strongly bathochromically shifted to 623 nm as compared to the case of pTh-BTD (549 nm). While all the polymers appear from UV-vis to be in their semiconductive form, pEDOT-BTD stands out from the rest with a strong polaronic absorption Adv. Mater. Interfaces 2023, 10, 2300270 Table 1. Electronic and optical data of the studied electropolymerization products.  signaling that the polymer is doped. Since the polymer contains the highly electron-rich EDOT unit, it is not surprising that ambient oxygen can lead to p-doping in the ground state. The determined optical band gaps (Eg opt ) of studied polymers are lower than the electrochemical ones (Eg el ), which was a common occurrence in the similar low-and high-molecular-mass electroactive compounds. [28] The optical bandgaps values, together with electrochemical ones and IP and EA values calculated on the basis of these electrochemical data, are listed in Table 1.

pTh-BTD
The experiments described in Section 3.2 on testing photoelectrochemical ORR and hydrogen peroxide production revealed pTh-BTD to be the best-performing polymer in the series. As a result, in the following section of this work, pTh-BTD was subjected to extensive photoelectrochemical research and optimization. Herein pTh-BDT thin films were obtained not only by electropolymerization but also by a chemical polymerization technique: iodine-vapor assisted polymerization (IVP). For this purpose, we used the modified procedure reported previously by Oka et al. [12,13] Both polymerization methods give a product with wide optical absorption and good electrochemical stability (Figure 5a,b). The SEM images (Figure 5c) of electrodeposited pTh-BTD reveal its highly porous structure, whereas pTh-BTD IVP layer forms smooth surfaces with visible nano-walls.

Photoelectrochemical Studies
Photocathodes were fabricated by polymerization of the given monomer directly onto ITO and then testing of the photocathodic performance was done in aqueous, oxygenated electrolyte. Photocathode characterization was performed either by using cyclic voltammetry, in dark versus illuminated conditions (1 sun = 100 mW cm −2 intensity), or chronoamperometrically at a fixed potential. As it was established in seminal studies of organic semiconductor photocathodes, [8,30] a simple point of reference is to record photocathodic current traces using pulses of light illumination using a fixed potential of 0 V versus Ag/AgCl. The magnitude of photocapacitive and photofaradaic currents can thus be clearly observed. [31] A brief comparison of such chronoamperometric scans of polymeric photocathodes based on the material series is plotted in Figure 6. Most materials gave moderate performance with photocurrent densities in the range of 10-50 μA cm −2 . As a reference point, the archetypical p-type material poly(3-hexylthiophene), generated between 5 and 10 μA cm −2 under the same measurement conditions, [10] while more optimized P3HT formulations are reported to achieve ≈20 μA cm −2 . [32] This comparison with P3HT is therefore a good indication that this series of donor-acceptor polymers is promising. One material, pEDOT-BTD, yielded no photocurrent, which can easily be explained by the fact of the high-lying HOMO level (see Figure 7), which indicates that the polymer will be p-doped under the ambient experimental conditions, explaining the quenching of photocurrent. As can also be rationalized from its UV-vis spectrum, pEDOT-BTD is in its conducting and not semiconductive form, therefore photocurrent is not generated. The material from the series that emerges as the most promising and reproducible in terms of photocathodic oxygen reduction is pTh-BTD and for this reason, the remaining detailed experiments on photocathodic oxygen reduction to peroxide are performed exclusively on this   material. pTh-BTD yielded photocurrents already greater than 100 μA cm −2 in this comparative assay.
A more detailed investigation was therefore done for pTh-BTD. First, we focused on layer thickness, comparison of chemical versus electropolymerization, wettability of the layer to water, and finally maximizing photocurrent by stirring to increase oxygen diffusion. Comparisons of photocurrent measurements for different thicknesses of pTh-BTD on an ITO substrate are plotted in Figure 8. For electropolymerization, thickness was dictated by the number of cyclic voltammetry cycles to form the film. For IVP, thickness was controlled by the concentration of the monomer which was solution-processed onto the substrate prior to oxidative polymerization. For electropolymerized layers, photocurrent magnitude appeared to be essentially independent of thickness, while for IVP films, overall photocurrents were lower, and more monomer/high thickness resulted in increased photocurrents (Figure 8a,b). Thicker films absorb more light, and therefore could in principle generate more photocathodic current. However, if the photoreduction reaction with oxygen and water can only occur on the interface where semiconductor/oxygen/water meet, that is the surface, increasing thickness may not lead to better performance. This is due to the limited diffusion of photogenerated excitons, on one hand, and the question of how far water and oxygen penetrate into the bulk of the film. From the SEM images of pTh-BTD layers (Figure 5c), it is apparent that electropolymerized films have a rough and porous nano/microstructure, while IVP films appear smoother and with much less porous structure. In principle, such a porous structure could be expected to provide higher photocurrent, however, provided that the surface is wettable to electrolyte. It has been shown for P3HT, that porous structures are in fact less efficient in generating photocurrent despite higher surface area, however after a hydrophilization treatment, the porous films become much more efficient than nonporous ones. [10] We followed this logic of wettability by immersing the polymer films in a 1% surfactant solution for 10 s and rinsing it with water. Our studies employed three common surfactants: sodium dodecyl sulfate (SDS), hexadecyltrimethy-lammonium bromide, and hexadecylpyridinium chloride (CPC). Both anionic and cationic surfactants help, but anionic appear to yield higher photocurrent ( Figure S5, Supporting Information). The surfactant treatment applied, the photocurrents from electropolymerized films increased by more than a factor of two and had a clear thickness dependence of thicker films giving more photocurrent. On the other hand, surfactant treatment had no effect on IVP films, indicating that wettability in this case is not the photocurrent-limiting factor. The electropolymerized films emerge from this comparison as having higher ORR photocurrents. Using a series of optimized electropolymerized films, we then sought to see how high photocurrents they could produce when solution mixing was introduced, in order to increase the possible diffusion-limited ORR current. It should be noted that from earlier studies it was established that devices with current densities <100 μA cm −2 do not benefit from mixing, as these values are not diffusion-limited current but photogenerationlimited current. There the mixing of electrolyte is done only on the best-performing samples, shown in Figure 8c. In the final stage of our optimization process, we conducted tests also on hydrophobicized ITO, FTO, and Au-coated PET foil. As shown in Figure S5, Supporting Information, the best photocurrent densities ranging from 150-200 μA cm −2 and over 300 μA cm −2 were recorded for polymer films electrodeposited on ITO and FTO, respectively. These values represent the highest ORR photocurrents achieved for a polymeric photocathode to-date, based on the best of our knowledge.
The photocurrent transient recordings shown in Figures 6  and 9 are a good way to quantitatively compare samples, however for the application of ORR for peroxide generation, it is important to establish the magnitude of ORR photocurrents over a larger potential range, to exclude other processes. We recorded cyclic voltammograms under illumination and in the dark, in argon-or oxygen-saturated conditions (Figure 9). These measurements, performed in a wide pH range (1, 7, and 11 pH) reveal that there is a large light/dark current ratio, and that argon-saturated measurements give almost no photocurrent effect. Therefore, it can be concluded that the dominant photocurrent process is in fact ORR, and that this process is more efficient with decreasing pH, corresponding to the mechanism of Reaction 1 and/or 5 ( Figure 1).
Having established photocurrent generation and its correspondence to ORR, we next performed longer-term photoelectrolysis experiments to quantify the hydrogen peroxide yield (Figure 10). The pTh-BTD/FTO photocathodes begin with photocurrents ≈300-400 μA cm −2 and faradaic yields for peroxide generation in excess of 90% (Figure 10a). The faradaic yield remain above 90% during the period of 8 hours, as concentrations of peroxide reach values above 3 mM. Compared to other measurements on organic photocathodes, this behavior and decrease in performance are similar, though such performance over a period as long as 8 h is longer than most published measurements. In the case of P3HT, over the course of long photoelectrolysis experiments, the semiconductor layer degrades heavily, with obvious bleaching which proceeds until the entire layer is destroyed. In the case of pTh-BTD, it appears that the long electrolysis experiment does not lead to irreversible degradation/oxidation of the material itself. This we probed by dissolving the layer in chlorobenzene and measuring the optical absorption of pristine samples and those after an 8 h run. This method of solution-absorption has been established before as a reliable way to probe layer photodegradation. [11,14] Measuring in solution is done in order to discount possible absorption changes in solidstate that arise from morphology change. From these measurements (Figure 10d), it is apparent that the photocathode material does undergo some degradation and/or doping process. It appears this is not reversible. We imaged the morphology of layers before and after photoelectrolysis using scanning electron microscopy, as well as testing the water contact angle. The morphology is obviously changed after the electrolysis: the sample is less rough and porous, and the contact angle drops from a relatively Figure 10. a) Chronoamperometry of the photocathode at pH 1 and 0 V versus Ag/AgCl over 8 hours of irradiation (100 mW cm −2 ). The illumination was shut off periodically to remove aliquots of electrolyte to measure peroxide content. These result in dark-current "spikes" in the current trace. b) The time dependence of the faradaic efficiency (red line) and measured solution concentration of hydrogen peroxide (black line). c) Comparison of the UV-vis spectra for the electrodeposited pThBTD photocathodes before (red line) and after (blue line) 8 h of photoelectrolysis in 1500 μL of chlorobenzene (averaged over 5 samples) Shaded areas represent standard deviation of averaged absorbance. d) SEM images and water-content angle measurements of the active layer before and after 8 hours experiment run.
hydrophobic 134 degrees to a very hydrophilic 17 degrees. This signals that there are significant morphological and/or chemical changes which may be resulting in much less efficient photocathode performance with respect to ORR. This shows that the catalytic activity of organic semiconductor photocathodes may be very dependent on morphology and overall material stability.

Conclusions
We have proposed macromolecular organic semiconductors as catalysts for the generation of hydrogen peroxide through photocathodic reduction of oxygen. The starting point of the research was synthesis of nine monomeric compounds of donor-acceptordonor block-type structure. The acceptor units, i.e., benzothiadiazole (BTD), benzooxadiazole (BOD), benzoselenadiazole (BSeD), and naphthobisthiadiazole (BBTD) were symmetrically substituted with electron-donating moieties: thiophene (Th), thienoth-iophene (TT), bithiophene (BTh), and ethylenedioxythiophene (EDOT) (Figure 1). Our study follows the logic that electropolymerization directly at the substrate can lead to rapid prototyping of organic building blocks with respect to their photocathode performance. Electrochemical and spectroscopic studies determined structure-property relationships in terms of electron and optical parameters. Thin films of polymers obtained by the electrochemical method exhibited electroactivity for both negative and positive potentials. In all cases, the reduction and oxidation of the polymers were quasi-reversible, signaling the potential of ambipolar charge transport. Absorption spectra displayed particularly broad absorption bands in visible and ultraviolet range. Electroactive polymers with broad absorption in the visible spectrum are potential candidates for photocathodic oxygen reduction to hydrogen peroxide. We benchmarked the photocathodic responses of the different polymers, finding in most cases photocurrent values exceeding those reported for the well-known photocathodic polymer P3HT. The material which emerged as the most promising in terms of photocurrent was pTh-BTD, which was thus selected for further study and optimization. Photocathodic performance scaled with pH, with the photocurrent trend being pH 1 > pH 7 > pH 11. This implicates a proton-coupled pathway to peroxide generation. This can be either the thermodynamicallyfavored 2H + /2e − reduction of oxygen, or the less favorable but potentially kinetically facile single-electron reduction to superoxide, followed by further reduction and protonation of superoxide. Competing pathways to peroxide accumulation are 1) the 2electron hydrogen evolution reaction, which may occur though is much less favored than ORR thermodynamically; and 2) further 2-electron reduction of peroxide to water (Reaction 3). 4-electron ORR is not considered a competitive process, as the concerted transfer of four electrons is highly demanding and unlikely to occur without a transition metal mediator. The photocurrent magnitude could be increased by up to 50% by treating the polymer film with a surfactant (sodium dodecyl sulfonate). This indicates that wettability of the polymer film to water is an important limiting factor, as protons are necessary to the generation of peroxide. It is assumed that this has a minor effect on oxygen transport into the film, as due to poor solubility of molecular oxygen in water it is reasonable to assume that oxygen can accumulate in the polymer layer to a higher concentration than in the surrounding water. Once photocurrent densities are over ≈100 μA cm −2 , oxygen diffusion from the electrolyte clearly becomes a limitation, as introducing mechanical mixing doubles the photocurrent magnitude. Photocathodic operation over the course of 8 h shows that peroxide concentrations up to ≈3 mM are achievable. Over this time, faradaic yields for peroxide production drop remain >90%. It is helpful to interpret these results in the context of other organic molecule and polymeric photocathodes. Compared to other reports on photocathodes based on a single organic material, the ORR photocurrent values obtained with pTh-BTD are the highest, to the best of our knowledge. The faradaic yields with respect to peroxide generation are in-line with the best-reported values for organic photocathodes, which have been reported to be >90% with similar concentrations of peroxide. Evaporated small molecule photocathodes, [8] as well as recent polymeric ones, [13] give total peroxide concentrations in the range of several mM. These previously-reported polymeric polythiophene-based systems show the best performance at alkaline conditions, while our donor-acceptor polymers work better in acidic conditions. This would indicate that different organic polymers may have quite different mechanisms of oxygen reduction and peroxide generation. Indeed, reports exist of optimum-alkaline, [13] optimum acidic, [14] and optimum acidic or alkaline, with a minimum at neutral conditions. [8] This underscores the need for better understanding of mechanistic pathways to reach higher efficiencies in the future. A final issue that our work shows, in line with previous studies on different organic photocathode systems, is stability. Stability remains a limiting factor, as efficiency drops over the course of several hours of continuous operation, and it appears irreversible. This degradation effect appears to be surface-activity related, and/or morphological, but also with a component of chemical degradation of the semiconducting polymer itself. With deeper understanding of this process, progress in this field can be made. The approach of electropolymerization, however, has some advantages and potential applications. While at present, the notion of using such materials for peroxide generation in the context of industrial applications and sustainability seems precluded due to limited efficiency and stability, the moderate micromolarmillimolar concentrations which can be generated hold applications in biology, [5,6] where local on-demand peroxide delivery is a desirable property. The ability to electropolymerize photoactive peroxide-generating layers should enable unique applications in this space, which can be achieved at low cost and with minimal equipment.

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