Linear Conjugated Polymers for Solar-Driven Hydrogen Peroxide Production: The Importance of Catalyst Stability

1Hydrogen peroxide (H2O2) is used on a huge scale in applications, such as paper and textile bleaching, chemical synthesis, wastewater treatment, and (on a smaller scale) fuel cells. H2O2 is mainly prepared via the anthraquinone process, which uses large amounts of energy and creates a lot of chemical waste. The clean production of H2O2 via the photocatalytic reaction of H2O and O2 under solar illumination has therefore sparked much recent interest.

2To achieve efficient H2O2 production, photocatalysts must absorb sunlight, produce separated charges, and drive redox reactions. A variety of organic materials have been investigated, including graphitic carbon nitride (g-C3N4),− supramolecular coordination complexes, covalent organic frameworks (COFs), triazine-based frameworks (CTFs), and polymer resins.− Recently, promising H2O2 production efficiencies were reported for some organic photocatalysts.,− However, to date, few organic photocatalysts have shown good performance for photocatalytic H2O2 synthesis without sacrificial agents, and efficiencies are far from satisfying industrial requirements.

3Here we used high-throughput experiments to identify a linear conjugated polymer (LCP), poly(3-4-ethynylphenyl)ethynyl)pyridine (DE7), with good activity for photocatalytic H2O2 production from H2O and O2 under visible light. The apparent quantum yield (AQY) of DE7 was 8.7% at 420 nm over short irradiation times, but over longer periods (>50 h), both DE7 and a resorcinol-formaldehyde catalyst− were found to become deactivated.

4An initial library of 60 candidate materials was designed (Figure 1a–f and Figures S1–4) that included g-C3N4, TiO2, and known photocatalytic conjugated polymers (P1, P7, and P10) as benchmarks. This functionally diverse library of branched and linear conjugated polymers was then screened for H2O2 production in pure water (no added sacrificial donors) in air using a modified high-throughput screening platform. DE7 showed the highest H2O2 production in this library over an irradiation period of 1.5 h using simulated solar light (Figure 1a). Other derivatives of DE7 were also synthesized (DE7-D-n, Figures S5–S7), but DE7 still showed the highest activity. We found no simple correlation between the H2O2 production rate and any single physical property such as band gap, fluorescence lifetime, particle size, or water contact angle (Figure S8), much as for sacrificial hydrogen evolution.,

5DE7 shows poor activity for photocatalytic hydrogen (H2) evolution, and more broadly, sacrificial H2 production and H2O2 generation are somewhat contraindicated (Figure S9), although several catalysts (e.g., DE3, TE2, TE3, TE9, TE10, TE12) are poor catalysts for both reactions. DE7 was synthesized by Pd-catalyzed Sonogashira coupling of 1,4-diethynylbenzene and 2,5-dibromopyridine (Figures S10–S15). To investigate the effect of residual metals, on catalytic performance, DE7 was prepared with different amounts of Pd(PPh3)Cl2 and CuI. Excessive quantities of Pd and Cu in the polymer decreased the catalytic performance, although only at very high metal loadings (Figure S16), perhaps due to Fenton-like reactions or effects on charge transfer pathways. Other Sonogashira-coupled polymers contained variable quantities of Cu (Table S1; mostly <0.5 wt %). Polymers in the DE7-n series had similar Cu contents to DE7 but showed significantly lower catalytic performance. As such, we do not believe that variable Cu contents in these polymers account for their variable catalytic performances. Microwave heating was also used to synthesize DE7-M (that is, DE7 synthesized under microwave conditions), shortening the synthesis time from 2 days to 2 h along with enhanced photoelectric properties (Figures S17 and S18). The optical gap of DE7-M was calculated from the UV–visible spectrum to be 2.34 eV (Figure 2a). Powder X-ray diffraction (PXRD) patterns (Figure 2b) showed that DE7-M was semicrystalline. The polymer mainly consisted of plicated sheets (Figure 2c,d). The contact angle of DE7-M against water was 67°, indicating moderate wetting behavior (Figure S19).

6The AQY was measured at different wavelengths to evaluate the photocatalytic H2O2 production performance. The AQY was determined to be 8.7% at 420 nm (Figure 3a), which is higher than most reported organic photocatalysts for H2O2 production using pure water, such as PEI/C3N4 (2.21% at 420 nm) and RF523 (∼8.0% at 420 nm), but lower than Sb-SAPC15 (17.6% at 420 nm) and OCN-500 (10.2% at 420 nm) (Table S2). The AQY profile followed the absorption spectrum, supporting a photoinduced H2O2 generation process (Figure 3a). The solar-to-chemical conversion efficiency was 0.28% in the first hour, falling to 0.23% after 5 h (Figure S20). As shown in Figure 3b, no H2O2 was detected under a nitrogen atmosphere. The amount of H2O2 produced under pure O2 (99%) was 1.6 times greater than in air, again indicating that O2 is essential for H2O2 production. This was further confirmed by isotopic labeling experiments using 18O2 (Figure 3c): the percentage of 18O2 detected by mass spectroscopy increased from 0% to 59% over 24 h. Electron paramagnetic resonance (EPR) measurements detected radicals using 5,5-dimethyl-1-pyrroline N-oxide (DMPO) as the trapping agent (Figure 3d). The typical signals of •O2– were found after illumination, indicating the one-electron reduction of O2.

7The H2O2 production was quenched when the •O2– scavenger p-benzoquinone (p-BQ) was added. In the presence of Sc(NO3)3 (a strong Lewis acid), the H2O2 production performance of DE7-M was significantly enhanced (Figure 3e), which we attribute to the inhibition of back electron transfer from •O2– after binding of Sc3+ to •O2–. On the basis of these results, we suggest that the photoinduced H2O2 production of DE7-M involves the stepwise reduction of O2.

8To gain deeper insight into the mechanism, we explored the half-reaction of DE7-M for H2O2 production by the addition of sacrificial electron-donor agents (Figure 3e and Figure S21). The increase in rate in the presence of isopropanol (IPA) suggests that H2O2 production might not be occurring via water oxidation by photoinduced holes, since these would be consumed by such electron donors. No H2O2 was detected when alkaline sacrificial reagents were used (Figure S22), probably because of decomposition of H2O2. AgNO3 also functioned as an electron-acceptor sacrificial agent in the half-reaction for O2 production (Figure S23), indicating that DE7-M can oxidize H2O to O2. The conduction band (CB) and valence band (VB) levels for DE7-M were estimated to be −0.93 V (vs NHE) and 1.41 V (vs NHE), which suggests that reduction and oxidation of oxygen and water are thermodynamically possible (Figure 3f and Figure S24). On the basis of these various results, we propose a mechanism for photocatalytic H2O2 production by DE7-M (Figure 3f).

9The experiments discussed so far relate to short reaction times (<5 h). To be practically useful, long-term photostability of catalysts is essential. We therefore tested the photostability of DE7-M. First, the DE7-M material was used in five sequential 2 h photocatalytic H2O2 production runs (Figure 4a). The activity was found to decline over the last four runs, and there was dramatic catalyst mass loss, suggesting decomposition. In a continuous, 55.5 h experiment (Figure 4b), the H2O2 production rate leveled off after about 30 h. H2O2 consumption (decomposition) started to occur after 47.5 h, suggesting that the reaction had ceased (Figure S25). Similar photocatalytic H2O2 production profiles were observed for composites of procyanidins-methoxybenzaldehyde (PM) dipolymers with carbon dots. Bubbling fresh O2 into the reactor did not recover the production rate (Figure 4b). For comparison, we also tested the published resorcinol formaldehyde catalyst RF523, which we synthesized using the reported conditions (Figure S26). RF523 was found to be significantly less active than DE7-M at short reaction times under these conditions (Figure 4b), but it, too, became deactivated over time (Figure 4b) and underwent chemical changes (Figure S26c); again, the H2O2 production rate tended to zero after about 50 h of irradiation. Previously, carbon nitrides have shown good stability,, but their catalytic activity is generally low for photocatalytic H2O2 production from pure water, which may in part account for this.

10The morphology of DE7-M was maintained over the first 24 h, but after 55.5 h of irradiation, the particle size was notably reduced (Figure S27). Also, the absorption spectrum of DE7-M became blue-shifted after 55.5 h of irradiation (Figure 4c). PXRD patterns (Figure S28) and the elemental composition (Table S3) also changed. FTIR spectra showed a growing peak at 1672 cm–1, which was assigned to carbonyl functionalities, suggesting oxidation of DE7-M (Figure S29). Further analysis showed that the alkyne bonds in DE7-M are oxidized, ultimately to produce 1,4-terphthalic acid and 2,5-pyridinedicarboxylic acid (Figure S30a).

11There are at least two possible pathways for the degradation of DE7-M: the polymer could be attacked by •O2– produced via photoinduced one-electron reduction of O2, or it could be self-oxidized by photoinduced holes. To study this, a series of control experiments was carried out (Figure 4d). There was no observable change in the FTIR for DE7-M under dark conditions in the presence of H2O/air or H2O2/air. DE7-M was not inherently unstable to H2O2 at these concentrations (∼3 mM). The polymer also remained intact after illumination under a nitrogen atmosphere. In the presence of IPA (a hole scavenger), both the rate of H2O2 production and the polymer decomposition rate were increased significantly (Figure S31), indicating that water oxidation is the slow step during photoinduced H2O2 production. This also suggested that photoinduced holes in the polymer are not the main cause of the decomposition since these holes might be expected to be scavenged by IPA. In support of this, DE7-M neither decomposed nor produced H2O2 in the presence of AgNO3, an electron scavenger (Figure 4d). In the presence of p-BQ, an •O2– scavenger, the decomposition of DE7-M was significantly mitigated (Figure 4d). Taken together, these data suggest that the main pathway for DE7-M decomposition involves •O2– (Figure S30b).

12We next measured the AQY in the presence of IPA, which reflects the intrinsic efficiency of the photoinduced exciton generation, separation, and migration; this was found to be 20.4% at 420 nm, highlighting the potential for DE7-M or similar materials as photocatalysts, providing that decomposition pathways can be suppressed. One strategy might be to load cocatalysts onto the surface of DE7-M, as exemplified in photocatalytic overall water splitting,, to promote O2 activation to H2O2, and to accelerate the H2O oxidation reaction,, respectively. Cocatalysts might also suppress reactions between the polymer and photoinduced electrons and holes,, providing that they do not also catalyze H2O2 decomposition., Another strategy would be to develop materials that are intrinsically less susceptible to •O2– attack or to adjust the band energy positions to produce H2O2 via a one-step, two-electron reduction pathway, avoiding the production of •O2– intermediates.,

13In summary, a conjugated polymer, poly(3–4-ethynylphenyl)ethynyl)pyridine, DE7, is an efficient organic photocatalyst for H2O2 production from H2O and O2 under visible light illumination without added sacrificial agents. DE7 outperforms many other organic catalysts over short reaction times. However, DE7 also decomposes over quite short irradiation time scales, rendering practical applications impossible at this stage. Similar deactivation was observed for a resorcinol formaldehyde resin, RF523. This suggests that more attention must be paid to catalyst stability, as well as increasing photocatalytic rates. Given that longer-term measurements are relatively straightforward, we would suggest that kinetic experiments of at least 24 h duration, and preferably longer (Figure 4b), should be mandatory for new catalysts.