Persistent semiquinone radicals enable efficient near-infrared-driven H₂O₂ photosynthesis

Abstract

Photosynthesis of H₂O₂ under sunlight is a sustainable method; however, most developed photocatalysts utilize limited near-infrared light, which accounts for 52% of the solar spectrum. In typical near-infrared photocatalysts, excited electrons fall into low-energy sub-gap states, reducing the driving force for H₂O₂ generation. Here, a polydopamine-loaded porphyrin supramolecular photocatalyst efficiently utilizes near-infrared light for H₂O₂ production from H₂O and O₂, achieving an apparent quantum yield of 2.8% at 1020 nm. This substantial near-infrared utilization significantly boosts activity under full-spectrum irradiation, with an H₂O₂ generation rate of 3.37 mM/h and solar-to-chemical conversion efficiency of 2.2%. Persistent semiquinone radicals in polydopamine are demonstrated to enable ultrafast sub-gap electron transfer (ca. 79 fs) from porphyrin to polydopamine and facilitate near-infrared-driven •OOH radical generation, thereby accelerating H₂O₂ production. This study sheds light on the potential of near-infrared-responsive photocatalysts and offers insights into optimizing their performance for sustainable H₂O₂ synthesis.

A porphyrin-polydopamine photocatalyst enables efficient near-infrared light use for hydrogen peroxide production from water and oxygen, achieving high quantum efficiency and solar conversion through ultrafast electron transfer mediated by persistent radicals.

Introduction

Hydrogen peroxide (H₂O₂) has emerged as a prominent green oxidizing agent with widespread applications in medicine, bleaching, and environmental remediation^1,2. The current global market for H₂O₂ is valued at ca. 6 billion dollars, with projections indicating growth to ca. 10 billion dollars by 2030³. Traditionally, H₂O₂ is synthesized industrially via the anthraquinone process, which relies on high-pressure conditions, costly palladium catalysts, and explosive hydrogen gas⁴. In recent years, photocatalytic methods have gained attention as a sustainable alternative for producing H₂O₂ using water (H₂O), oxygen gas (O₂), and sunlight in the presence of suitable photocatalysts. To maximize the utilization of solar energy, the photocatalysts are expected to efficiently harness near-Infrared (NIR, λ > 800 nm) light, which constitutes approximately 52% of the solar spectrum. However, most developed NIR-responsive photocatalysts suffer from significant limitations^5–8, including low H₂O₂ production rate (e.g., ~ 100 μM/h), a narrow NIR-responsive range (λ < 900 nm), and minimal contribution from NIR light (< 10 %) to overall activity under full-spectrum illumination.

Tetrakis(4-carboxyphenyl) porphyrin (TCPP) supramolecule exhibits remarkable NIR photocatalytic activity for H₂O₂ production through a novel dual-pathway mechanism⁹. Unlike previously reported photocatalysts, which rely solely on oxygen reduction (ORR) for H₂O₂ generation, the TCPP supramolecule demonstrates both ORR and water oxidation reduction (WOR) to produce H₂O₂. Specifically, ORR occurs on the pyrrole ring, while WOR is driven by photo-thermal transitions involving the carboxylic groups and peroxy acid intermediates (Fig. 1a). Notably, a key feature of the TCPP supramolecular photocatalyst is its light absorption capability extending into the NIR region (up to 1100 nm). Under monochromatic NIR light (940 ± 10 nm), SA-TCPP achieves a H₂O₂ production rate of 0.15 mM h⁻¹, with an apparent quantum yield (AQY) of up to 1.1%. Despite these advancements, the contribution of NIR-driven activity to the overall photocatalytic performance remains limited. In most NIR-responsive photocatalysts, electrons excited by NIR light primarily transition to sub-gap energy levels (trap states)^10,11, which possess lower energy than the conduction band, reducing the potential driving force for effective H₂O₂ production (Fig. 1b). Consequently, this inherent energy limitation may hinder the efficient utilization of NIR light in TCPP-based systems. Radicals generated within the photocatalyst framework have been reported to smoothly mediate ORR to H₂O₂ by facilitating the utilization of electrons in the conduction band¹². However, the generation of such reactive radicals typically requires UV or visible excitation. Moreover, under NIR irradiation, even when •O₂^- or HOO• intermediates during H₂O₂ production are generated by such internally formed radicals, they exhibit a strong tendency toward back reactions originating from rapid electron-hole recombination.

Fig. 1. Utilization of NIR-excited sub-gap electrons for •OOH radical generation.

a Dual-pathway mechanism of H₂O₂ generation on the porphyrin photocatalyst (The dashed circles highlight the N-H and COOH groups on SA-TCPP, which serve as the active sites for the ORR and the WOR, respectively). b Schematic energy-level diagram illustrating •OOH radical formation and subsequent H₂O₂ generation driven by NIR-excited trap electrons. c Calculated free energy for the generation of •OOH radical and H₂O₂ on PDA. Inset is the structure of the prepared PDA (O: Red ball, N: Blue ball, C: Gray ball, H: White ball). d Experimental and simulated EPR spectra of PDA under NIR irradiation (λ > 800 nm) with N₂ bubbling and DMPO as a radical capture reagent. e In situ FTIR spectra recorded with and without NIR irradiation (Xe lamp with an 800 nm cutoff filter). *OOH represents the •OOH radicals adsorbed on the PDA surface. f UV-Vis-NIR DRS for SA-TCPP. Inset is the schematic relationship between the bandgap and Urbach sub-bandgap. g Potential diagram illustrating the utilization of sub-gap electrons for •OOH radical generation in the PDA/SA-TCPP system (The NIR photon energy (< 1.55 eV) was estimated from an excitation wavelength of 800 nm).

Besides the inherent energy limitation, the high energy barrier for the ORR to H₂O₂ further constrains the efficient utilization of sub-gap NIR-excited electrons. Thermodynamically, effective ORR requires that the conduction band edge potentials be more negative than the reduction potential of O₂ to H₂O₂ (0.68 V vs. RHE)¹³. During the actual ORR process, O₂ is initially reduced to free or adsorbed •OOH intermediates, followed by the subsequent reduction of these intermediates to H₂O₂. While the further reduction of •OOH intermediates to H₂O₂ occurs at potentials around 1 V, their initial formation requires a more negative potential (approximately 0.30 V vs. RHE, based on the average of reported data)^14–16. Accordingly, the ability to form •OOH intermediates using NIR-excited electrons occupying sub-gap states plays a crucial role in the efficient utilization of NIR light for H₂O₂ production.

This study addresses the challenge of utilizing NIR light for H₂O₂ photosynthesis by integrating polydopamine (PDA), a well-established full-spectrum absorption material^17,18, onto self-assembled TCPP porphyrin supermolecules (SA-TCPP). Although PDA has been employed to promote the photocatalytic H₂O₂ production¹⁹, its use to enhance NIR-driven H₂O₂ generation remains rare, despite its broad solar-spectrum absorption. Moreover, while radical-mediated fast electron transfer has been reported¹², the utilization of the persistent semiquinone radicals intrinsically present in PDA to mediate the transfer of low-energy, NIR-excited sub-bandgap electrons remains poorly understood.

Here, we investigate H₂O₂ generation from only H₂O and O₂ using PDA/SA-TCPP composite photocatalysts, demonstrating that NIR-driven ORR contributes nearly 30% to the overall activity. By enhancing NIR utilization, the overall H₂O₂ generation rate reaches 3.37 mM h⁻¹, with a solar-to-chemical conversion efficiency (SCC) of 2.2%. We elucidate the mechanism by which persistent semiquinone radicals in PDA facilitate rapid transfer of NIR-excited sub-gap electrons in SA-TCPP, promoting the formation of •OOH radicals and H₂O₂. This work establishes persistent semiquinone radicals as effective electron shuttles for driving NIR-excited sub-gap ORR-to-H₂O₂ conversion, providing guidance for the design of efficient organic photocatalysts for scalable solar H₂O₂ production under sunlight.

Results

Theoretical feasibility of semiquinone-mediated sub-gap ORR of SA-TCPP under NIR light

PDA is synthesized via a self-polymerization method using dopamine monomers. The structural composition of the prepared PDA, shown in Fig. 1c, includes dihydroxyindole (hydroquinone-type, QH₂) and indoledione (quinone-type, Q) units with varying degrees of (un)saturation. This structural characterization is confirmed through the Fourier transform infrared (FTIR), Raman, and X-ray photoelectron spectroscopy (XPS) (Supplementary Figs. 1–3) measurements, consistent with findings reported in the literature^19,20. Scanning electron microscopy (SEM) images in Supplementary Fig. 4 reveal that the prepared PDA exhibits a nanoparticle morphology, with particle sizes ranging from 50 to 250 nm. Electron Paramagnetic Resonance (EPR) is further employed to elucidate the PDA structure. A distinct signal at g = 2.0035 with a line width of 5.5 G is observed for the PDA powder. (Supplementary Fig. 5) This is attributed to the presence of substantial persistent semiquinone radicals (SQ•), consistent with previous reports^21,22.

Ultraviolet-Visible-Near Infrared diffuse reflection spectrum (UV-Vis-NIR DRS) in Supplementary Fig. 6 shows the absorption band edge of the prepared PDA extends up to 1400 nm, attributed to the stacking of indole conjugated structural units. This absorption profile together with the band structure analyses (Supplementary Fig. 6) demonstrates a large overlap with the solar spectrum, encompassing not only the visible light region but also the NIR region, indicating the strong potential of PDA for NIR-light responsiveness.

To further explore the theoretical ORR pathway for H₂O₂ production on PDA and verify the NIR utilization, Gibbs free energy calculations are performed (Fig. 1c). The formation energy of SQ• radicals from QH₂ and Q is only 0.38 eV, indicating that the generation of SQ• radicals is highly favorable and supports their persistent presence within the PDA structure. Comparable formation energies (0.2–0.5 eV) have also been reported for other SQ• systems²³. Both the reaction barrier for SQ• radical generation from QH₂/Q and the subsequent O₂ addition to SQ• to form adsorbed Q*OOH are lower than the energy provided by NIR photons(energy < 1.55 eV), indicating that these steps are energetically accessible under NIR excitation. However, the formation of H₂O₂ requires the desorption of the *OOH intermediate followed by its reaction with a neighboring SQ• or QH₂ unit. In contrast, the subsequent desorption of the *OOH intermediate and its conversion to H₂O₂ via reaction with neighboring SQ• or QH₂ units involves substantially higher activation barriers (TS3-1,2) that exceed the energy accessible under NIR excitation, rendering these steps kinetically unfavorable. EPR measurements are conducted for PDA to examine its NIR-light driven •OOH generation from ORR. The experiment is conducted in the dark and under NIR light irradiation within an anoxic environment, using 5,5-dimethyl-1-pyrroline N-oxide (DMPO) as the •OOH radical trapper (Fig. 1d). The anoxic environment is specifically used to slow down the O₂ reduction on PDA, allowing the accumulation of intermediates. In the dark, PDA exhibits negligible signals of •OOH radicals, indicating that O₂ is not spontaneously reduced on PDA. In contrast, a distinct semiquinone radical (SQ•) is observed, consistent with the EPR signals for powders (Supplementary Fig. 5). With NIR irradiation, adduct EPR signals are observed. The signals consist of the SQ• radical adduct, with an intensity 2.6 times as high as that in the dark, along with newly emerged •OOH radicals, as confirmed by simulation results (details in Supplementary Fig. 7). The result suggests •OOH radicals are generated from 1 electron transfer from SQ• radicals to O₂. Previous studies have demonstrated that SQ• radicals exhibit high reaction rate constants with O₂ (10⁶ ~ 10⁸)^24,25, further supporting our observations.

In situ attenuated total reflectance infrared absorption (ATR-FTIR) spectroscopy is employed to investigate the NIR-light response of the prepared PDA. The experimental setup is illustrated in Supplementary Fig. 8, incorporating a silicon (Si) prism as the IR-transparent element. The absorption changes, determined by subtracting the baseline signal of the pristine PDA surface, provide insights into the surface reactions of PDA with H₂O and O₂ under NIR irradiation. Under NIR light irradiation (λ > 800 nm), distinct vibrational peaks at 1112 cm⁻¹ and 842 cm⁻¹ are observed. They are attributed to stretching vibrations of the adsorbed •OOH radicals and O-O in the radicals (Supplementary Table S1). Additional in-situ FTIR observations with ¹⁸O₂ isotopic labeling further confirm that the observed peak changes belong to •OOH radicals generated onto the PDA surface (Supplementary Fig. 9). Control experiments with different light wavelength cutoff filters verify that this PDA-induced generation of •OOH radicals enables substantial NIR utilization with the light wavelength extending up to 1020 nm (Supplementary Fig. 10).

With increasing irradiation time from 0 to 30 min (Fig. 1e), the intensity of the two peaks at 1112 cm⁻¹ and 842 cm⁻¹ increases. However, after the light is turned off, the intensity of these peaks gradually diminishes to nearly negligible levels, indicating that NIR light-induced •OOH generation is highly reversible. Furthermore, the prepared PDA exhibits negligible H₂O₂ generation under NIR irradiation (Supplementary Fig. 11), suggesting that the generated •OOH radicals cannot be further reduced, reinforcing the high reversibility of •OOH generation under NIR light. Under NIR light irradiation, the QH₂ and Q moieties may initially interact to increase the amount of persistent SQ• radicals, which subsequently react with O₂ to form •OOH radicals and regenerate Q moieties (as supported by FTIR and ¹³C solid-state nuclear magnetic resonance (NMR) results in Supplementary Figs. 12, 13).

While prior studies reported the H₂O₂ generation on PDA alone (0.5-1.0 mM h⁻¹ g⁻¹) via •OOH intermediates under sunlight irradiation²⁶, the current work highlights the facile generation of •OOH radicals on PDA under NIR irradiation. Although PDA cannot directly utilize NIR light to produce H₂O₂, the efficient generation of •OOH radicals on it under NIR irradiation remains crucial for enabling NIR-driven H₂O₂ production. In particular, in the case of NIR-driven H₂O₂ generation on SA-TCPP, •OOH radical formation with the excited sub-gap electrons represents a key rate-limiting step (Fig. 1b). Therefore, the facile generation of •OOH radicals presents a promising strategy for enhancing the efficiency of NIR-driven sub-gap H₂O₂ production on SA-TCPP.

The band structure of SA-TCPP is investigated using UV-Vis-NIR DRS to assess the potential for charge transfer from its sub-bandgap states to PDA. As shown in Fig. 1f, SA-TCPP displays a strong absorption band around 730 nm and a weak Urbach tail extending from 730 to 1400 nm. These features correspond to a bandgap energy (E_g) of 1.77 eV and an Urbach sub-bandgap energy (E_Urbach) of 1.21 eV (Supplementary Fig. 14). The Urbach energy is typically attributed to trap states within the SA-TCPP structure, which may arise from structural distortions, impurities, and lattice imperfections associated with its amorphous nature (as supported by the XRD results in Supplementary Fig. 15). The relationship between E_g and E_Urbach is illustrated in the inset of Fig. 1f. These trap states are generally distributed both above the valence band maximum (VBM) and below the conduction band minimum (CBM). Ultraviolet photoelectron spectroscopy (UPS) measurements determine the VBM and the trap state level above the VBM to be 1.79 eV and 1.51 eV, respectively. By subtracting E_g and E_Urbach from these values, the CBM and the trap state level below the CBM are calculated to be 0.02 eV and 0.30 eV, respectively (Supplementary Fig. 14).

It is noteworthy that in SA-TCPP, the electrons excited by NIR light predominantly accumulate at the trap state level located below the conduction band minimum (CBM), with a potential of 0.30 V vs. RHE. This value is only 0.02 V lower than the reduction potential of O₂ to •OOH radicals (0.32 V), indicating a minimal driving force and, consequently, limited utilization of NIR light for •OOH radical generation. In contrast, PDA exhibits a more favorable energetic alignment: the photon energy of NIR light (< 1.55 eV) is substantially higher than the energy barrier (0.51 eV) required for •OOH radical generation (Fig. 1c), enabling efficient 1-electron transfer from persistent semiquinone (SQ•) radicals to O₂. Furthermore, the quinone moieties formed in PDA during this process possess a redox potential (Q/SQ•) of 0.52 V vs. RHE (Supplementary Figs. 16, 17), which is 0.22 V positive relative to the trap state level in SA-TCPP (0.30 V vs. RHE). This favorable potential difference suggests that the transfer of sub-gap electrons from SA-TCPP to PDA is thermodynamically viable. Therefore, coupling SA-TCPP with PDA may enhance NIR light utilization through efficient sub-gap electron transfer, as illustrated in Fig. 1g.

Boosting H₂O₂ Generation under NIR Light by loading PDA on SA-TCPP Photocatalysts

To prepare PDA and SA-TCPP composite photocatalysts (PDA/SA-TCPP), an appropriate amount of the prepared PDA nanoparticles is introduced during the self-assembly process used for fabricating TCPP nanosheets.

Figure 2a, b shows the SEM and atomic force microscope (AFM) images of the prepared PDA/SA-TCPP photocatalysts. In the composite structure, small PDA nanoparticles ranging in size from 50 nm to 250 nm with a thickness of ca. 20–80 nm (Supplementary Fig. 18) are predominantly dispersed onto SA-TCPP nanosheets (Supplementary Fig. 19). The particle size of PDA remains similar to its original dimensions before loading onto the SA-TCPP nanosheets, indicating that the material experiences minimal chemical change during the self-assembly process of TCPP.

Fig. 2. Structural characterizations of the prepared PDA/SA-TCPP composite.

a SEM (The dashed circles highlight the PDA particles) and (b) AFM for the prepared PDA/SA-TCPP photocatalysts (The numbers labeled in the figure represent the thickness of the samples). c FTIR for PDA, SA-TCPP and PDA/SA-TCPP. d UV-Vis-NIR DRS for PDA, SA-TCPP and PDA/SA-TCPP.

Figure 2c presents the FTIR spectrum of PDA/SA-TCPP, alongside the spectra of SA-TCPP and PDA recorded as controls to investigate the formation of PDA loading onto the SA-TCPP nanosheet. Owing to the presence of similar functional groups in both compounds, most peaks in the spectra of PDA and SA-TCPP overlap. Nevertheless, three distinct peaks at 1693 cm⁻¹, 1399 cm⁻¹, and 1176 cm⁻¹ are exclusively attributed to SA-TCPP with minimal interference from PDA. Specifically, the peaks at 1693 cm⁻¹ and 1176 cm⁻¹ correspond to the stretching vibrations of C = O and C-O in the carboxyl groups²⁷, respectively, while the peak at 1399 cm⁻¹ is assigned to C = C stretching vibrations in the pyrrole ring of the porphyrin structure²⁸. Upon PDA loading, the peaks associated with the carboxyl groups shift by 17 cm⁻¹ and 9 cm⁻¹ to lower wavenumbers, whereas the peaks associated with the porphyrin core remain unchanged. This shift indicates the formation of hydrogen bonds between the SA-TCPP and PDA²⁹.

The formation of hydrogen bonds between PDA and SA-TCPP is further supported by the UV-Vis-NIR DRS spectra. As shown in Fig. 2d, pure PDA has a strong and broad absorption spectrum up to 1500 nm, and the Soret band (corresponding to the S₀ → S₂ transition) of SA-TCPP appears at 357 nm and redshifts to 385 nm after PDA loading. This redshift is likely attributed to the formation of edge-to-edge hydrogen bonds between PDA and SA-TCPP, which enhance the transition dipole moment, facilitate exciton splitting, and lower the energy of the S₂ state^30,31. In addition, isotope-labeled FTIR, solid-state ¹³C NMR, and XPS measurements further confirm the presence of hydrogen-bond interactions between PDA and SA-TCPP (Supplementary Figs. 20, 21). The hydrogen-bonding is expected to contribute to the stabilization of PDA on the SA-TCPP nanosheet during the catalytic reaction

Next, the H₂O₂ generation activity of the PDA/SA-TCPP photocatalyst is evaluated. The optimal concentration of the PDA/SA-TCPP photocatalyst in the reaction is 4.0 g/L (Supplementary Fig. 22). Figure 3a shows the time-dependent H₂O₂ generation on PDA/SA-TCPP under Xe lamp irradiation with λ > 420 nm and λ > 800 nm cutoff filters, respectively. Under the visible + NIR light (λ > 420 nm) irradiation, PDA/SA-TCPP exhibits a high H₂O₂ generation rate of 3.37 ± 0.32 mM/h in the first hour, reaching a cumulative concentration of 11.24 mM after 4 h. Notably, PDA/SA-TCPP achieves a high H₂O₂ production rate, placing it among the top-performing photocatalytic systems reported to date (Supplementary Table S2).

Fig. 3. H₂O₂ generation activity of the PDA/SA-TCPP photocatalyst.

a Time-dependent H₂O₂ generation on PDA/SA-TCPP with cutoff filters of λ > 420 nm (Visible + NIR light) and λ > 800 nm (NIR light). b AQY of PDA/SA-TCPP and SA-TCPP for H₂O₂ generation under light irradiation using various bandpass filters. c Comparison of H₂O₂ generation rates for PDA/SA-TCPP and SA-TCPP under different Xe lamp spectral regions. d Long-term H₂O₂ generation with PDA/SA-TCPP, SA-TCPP, and PDA photocatalysts (The numbers indicate the accumulated H₂O₂ concentration after ₂4 h). Reaction conditions: Cat., 200 mg; Sln., 50 mL H₂O with O₂ bubbling; Light source, Xe lamp equipped with different cut-off filters (λ > 420 nm, 200 mW/cm²; 420 nm < λ < 780 nm, 120 mW/cm²; λ > 800 nm, 80 mW/cm²); Reaction temperature, 353.15 K. Error bars represent the standard deviation of the mean, n = 3.

Under the NIR light (λ > 800 nm) irradiation, PDA/SA-TCPP still achieves a substantial H₂O₂ generation rate of 0.99 ± 0.30 mM/h in the first hour, with 3.30 ± 0.26 mM accumulated after 4 h. The H₂O₂ generation continues to increase with reaction time, which is different from the observations for PDA alone where negligible H₂O₂ generation is detected (Supplementary Fig. 12). In addition, loading PDA onto pre-assembled SA-TCPP nanosheets yielded a photocatalytic H₂O₂ generation activity comparable to that of the in-situ self-assembled PDA/SA-TCPP system (Supplementary Fig. 23), indicating that the enhanced activity does not originate from the incorporation of PDA during the self-assembly process. By contrast, a physical mixture of PDA and SA-TCPP is prepared, which exhibited less than half the activity of PDA/SA-TCPP (Supplementary Fig. 24). These results clearly indicate that the integration of SA-TCPP with PDA enables efficient NIR photon utilization for H₂O₂ production in this photocatalytic system.

Figure 3b depicts the AQY of PDA/SA-TCPP measured at different wavelengths (400–1020 nm). In the visible light region (400–800 nm), the average AQY of PDA/SA-TCPP is 22.3 ± 0.2% at 420 nm (Supplementary Fig. 25), which is 1.9 times higher than that of SA-TCPP and competitive among most reported values for H₂O₂ generation from H₂O and O₂ (Supplementary Table S2). In the NIR region, the AQY values of PDA/SA-TCPP at 850 nm, 940 nm, and 1020 nm are 7.1 ± 0.1%, 4.8 ± 0.1%, and 2.8 ± 0.1%, respectively, (Supplementary Figs. 26, 27), approximately 3.0–4.0 times as high as those of SA-TCPP (1.6% at 850 nm, 0.87% at 940 nm, and N/A at 1020 nm), which are competitive among most reported NIR light-responsive photocatalysts (Supplementary Table S3). The AQY results confirm that the enhanced photocatalytic activity is attributed to the improved NIR photon utilization enabled by PDA loading.

Figure 3c compares the H₂O₂ generation rates of PDA/SA-TCPP and SA-TCPP under Xe lamp irradiation with different cutoff filters (λ > 420 nm, 420 nm < λ < 800 nm, and λ > 800 nm) to better understand the contribution of NIR light utilization to overall H₂O₂ generation. Under visible and NIR irradiation (λ > 420 nm), the H₂O₂ generation rate of the PDA/SA-TCPP (3.37 ± 0.12 mM/h) is ca. 2.5 times higher than that of the SA-TCPP alone (1.35 ± 0.08 mM/h). Under NIR-only irradiation (λ > 800 nm), PDA/SA-TCPP achieves a rate of 0.99 ± 0.06 mM/h, nearly 7.1 times higher than SA-TCPP (0.14 ± 0.03 mM/h), and contributes about 29% to the total activity under λ > 420 nm irradiation, indicating a substantial enhancement in NIR photon conversion efficiency. By contrast, the enhancement under visible light alone (420 nm < λ < 800 nm) is relatively modest at 1.9 times.

Figure 3d compares the long-term H₂O₂ accumulation between PDA/SA-TCPP, SA-TCPP, and PDA photocatalysts. Leveraging its superior NIR utilization, which also helps minimize H₂O₂ decomposition, PDA/SA-TCPP accumulates 52.2 mM H₂O₂ (0.16 wt%) after 24 h irradiation, which is 2.6 times higher than SA-TCPP alone and 130 times relative to pure PDA. It is worth noting that higher H₂O₂ generation activity on PDA through chemical (thermal-driven) or photochemical pathways has been reported²⁶. However, such reactions predominantly occur under neutral or alkaline conditions. Under the acidic conditions used in this study, the H₂O₂ production via chemical and photochemical pathways are determined to be negligible compared with the H₂O₂ production observed for PDA/SA-TCPP (Supplementary Figs. 28−31). According to the above, the integration of PDA with SA-TCPP significantly enhances overall photocatalytic H₂O₂ production under broad-spectrum light, primarily due to improved NIR light utilization.

Rapid sub-gap electron transfer enabling efficient NIR Utilization

The NIR photocatalytic activity enhancement mechanism for PDA/SA-TCPP is studied next. In the PDA/SA-TCPP system, the peroxy acid intermediate—similar to that in SA-TCPP—is still observed under irradiation by comparative ¹³C NMR, ¹H NMR, TOF-MS, and FTIR analyses of PDA/SA-TCPP and SA-TCPP after room temperature reaction. (Supplementary Fig. 32) The similarity of the spectra excludes hole transfer from SA-TCPP to PDA, which indicates that holes remain on SA-TCPP. Further scavenger experiments (Supplementary Fig. 33) suggest the contribution of holes to the NIR-driven H₂O₂ production is nearly half similar as we reported in the SA-TCPP. KPFM measurements (Supplementary Fig. 34) of PDA/SA-TCPP before and after light irradiation further indicate that the photogenerated holes remain predominantly on SA-TCPP rather than transferring to PDA. These accumulated holes may oxidize the carboxyl groups on SA-TCPP to form peroxy acid intermediates (-COOH + 2 h⁺ + H₂O → -CO₃H + 2H⁺). Subsequently, the generated -CO₃H undergoes thermal hydrolysis—the primary role of heat throughout the photothermal catalytic process, as supported by the data in Supplementary Figs. 30, 31, thereby regenerating -COOH and yielding H₂O₂ (-CO₃H + H₃O → -COOH + H₂O₂)⁹. Moreover, PDA loading onto Zn-TCPP, with a similar structure to SA-TCPP, enhances H₂O₂ generation by eightfold (Supplementary Fig. 35). In-situ XPS shows the electron transfer from SA-TCPP to PDA on light. (Supplementary Fig. 36) The results show that PDA loading promotes electron transfer to accelerate the ORR to H₂O₂ under visible light, which is consistent with previous studies^32,33. However, while this electron transfer explains the activity boost, it does not account for the much greater enhancement under NIR light (7.4 times) compared to visible light (1.9 times).

We then conduct control experiments to elucidate the effect of PDA on the NIR activity enhancement. We employ capacitance measurements, specifically utilizing Mott-Schottky (M-S) curves (Supplementary Fig. S37), both in the dark and under NIR irradiation, to investigate the transfer of NIR light-excited electrons from SA-TCPP to PDA. Generally speaking, in the M-S curves, the slope at low bias reflects the charge density at defect levels, while the one at high bias reflects the charge density in the band³⁴. Supplementary Fig. 37a shows that in SA-TCPP, the slope of the curve at a low bias (−0.1-0.2 V) remains unchanged before and under irradiation, suggesting most NIR-excited electrons at defect levels recombine with photogenerated holes. In contrast, Supplementary Fig. 37b shows that in PDA/SA-TCPP, the slope decreases significantly, accounting for the defect electron density estimated to be 8.1 × 10¹⁶cm^-3 and 1.3 × 10¹⁷cm⁻³ before and under NIR irradiation (Supplementary Fig. 37c). Note that for PDA-alone, little contribution to this variation is observed after NIR irradiation (Supplementary Fig. 37d). The increase of defect electron density for PDA/SA-TCPP after NIR light irradiation indicates more NIR light-excited electrons transfer from trap levels to PDA rather than bulk recombination.

To gain further insight into the charge transfer, femtosecond transient absorption spectra (fs-TAS) of PDA/SA-TCPP and SA-TCPP are recorded. (Fig. 4a, b) The pump wavelength is set at 660 nm (S₀ → S₁ transition) to more accurately probe electron relaxation and transfer at the trap levels^35,36. For SA-TCPP, a ground-state bleaching (GSB) signal appears at 779 nm at 20 fs. At longer delay times ( > 170 fs), SA-TCPP exhibits an excitonic-state absorption (ESA) signal at 792 nm, and by 550 fs, two additional ESA signals emerge at 824 nm and 858 nm through an intersystem crossing (ISC) process^37,38. The signals at 779 nm and 792 nm are attributed to the ¹T states of SA-TCPP coupling with the electron spin. Notably, the peak intensities of the 824 nm and 858 nm signals remain essentially unchanged with further increasing the decay time, suggesting that photogenerated electrons localized at these trap states hardly undergo transfer and are prone to recombination with valence-band holes.

Fig. 4. Mechanism of NIR utilization enhancement by PDA loading.

fs-TAS signals of (a) SA-TCPP and (b) PDA/SA-TCPP recorded at different timescales (pump wavelength: 660 nm). c TAS decay and fitted curves of the 824 nm signal for SA-TCPP and PDA/SA-TCPP. d Schematic of exciton decay dynamics in SA-TCPP with and without PDA loading, along with the calculated LUMO + 2, LUMO + 1, and LUMO orbitals of PDA (isosurface value: 0.04 a.u.) and the PDA structural model (Schematic structural formula and corresponding ball-and-stick model) used for the calculation. TCPP^•+-TCPP^•- and TCPP^•+-PDA^•- represent polaron separation (O: Red ball, N: Blue ball, C: Gray ball, H: White ball).

In contrast, PDA/SA-TCPP displays distinct decay behavior compared to SA-TCPP. The GSB signal shifts to 792 nm. Although trap-state signals also appear at 560 fs, as observed in SA-TCPP, their intensity decreases with increasing delay time. At 640 fs, the peak at 824 nm diminishes, accompanied by the emergence of a new peak at 899 nm. As the delay time further increases to 690 fs, 730 fs, and 790 fs, the trap-state signals continue to decrease, while the 899 nm peak shifts progressively to 944 nm and 1026 nm. This rapid decay likely reflects efficient interfacial electron transfer from SA-TCPP to PDA³⁹. Meanwhile, the overall signal intensity is an order of magnitude stronger than that of SA-TCPP, highlighting the efficient utilization of trap-state electrons enabled by PDA loading.

Transient absorption kinetics at 824 nm are then compared for SA-TCPP and PDA/SA-TCPP to analyze exciton decay dynamics at trap states with and without PDA loading (Fig. 4c). Lifetimes and amplitudes at the fitted wavelengths are summarized in Supplementary Table S4. A bi-exponential model is employed to fit the decay curves, where τ₁ corresponds to polaron formation from excitons and τ₂ to polaron separation, with an additional constant term accounting for recombination processes beyond the measurement window⁴⁰. For SA-TCPP, τ₁ and τ₂ are determined as 380 fs and 225 ps, respectively. In contrast, PDA/SA-TCPP exhibits much faster dynamics, with τ₁ of 79 fs and τ₂ of 7.14 ps. These results demonstrate that PDA loading drives excitons at the trap states of SA-TCPP to form interfacial polarons instead of bulk polarons. Compared with bulk polaron formation without PDA, interfacial polaron formation is 5-fold faster, and their subsequent dissociation into free charges is 30-fold more efficient. The TRPL results further confirm that the acceleration of charge separation after PDA loading (Supplementary Fig. 38).

A schematic summary of the exciton processes in SA-TCPP and PDA/SA-TCPP together with the calculated absolute energy levels are shown in Fig. 4d and Supplementary Fig. 39. The presence of PDA clearly enhances interfacial charge transfer by forming interfacial polarons. Notably, the transferred electrons in PDA/SA-TCPP exhibit a longer recombination lifetime (> 8 ns) compared with those in SA-TCPP (> 2 ns), as indicated by the offset constants of the TAS signals at 1026 nm and 824 nm, respectively (Supplementary Fig. 40). In PDA/SA-TCPP, the transferred electrons on PDA eventually occupy the LUMO states, where they can be further captured. Since the TAS spectra display sequential ESA features at 899 nm, 944 nm, and finally 1026 nm after PDA loading, the 1026 nm peak is attributed to electrons localized in the PDA LUMO states, while the 899 nm signal likely corresponds to the LUMO + 2 orbitals. These LUMO + 2 states, originating mainly from persistent SQ• radicals (the orange orbital mapping shown in Fig. 4d), facilitate electron transfer from trap states of SA-TCPP probably by providing additional electron-pair energy. The acceleration of charge transfer mediated by the SQ•-related LUMO + 2 states was further supported by subsequent experimental evidence. The involvement of SQ• in ultrafast sub-gap electron transfer is further supported by TAS measurements performed after suppressing SQ• (Supplementary Fig. 41). Upon SQ• suppression, the interfacial trap-electron transfer process is prolonged, ranging from 570 fs to 820 fs, whereas with SQ•, it ranges from 590 fs to 690 fs. The TAS spectra further demonstrate that the trap-electron transfer proceeds via hydrogen-bonding interactions (Supplementary Fig. 42). Moreover, analyses of the interfacial electron density difference, Bader charge, and density of states are consistent with these observations (Supplementary Fig. 43).

To further verify that the trap electrons in SA-TCPP can be more effectively utilized for •OOH radical generation upon PDA loading and elucidate the role of SQ• radicals on H₂O₂ production, EPR measurements are carried out in air and anoxic conditions. As shown in Supplementary Fig. 44, PDA/SA-TCPP exhibits a pronounced •OOH radical signal under NIR irradiation, verified by radical-quenching experiments and isotope-labeling FTIR results (supplementary Figs. 45–47), markedly stronger compared with pristine SA-TCPP. The EPR fitting results together with quantitative H₂O₂ analysis further demonstrate the effective utilization of trap electrons for •OOH radical and H₂O₂ production is mediated by SQ• (supplementary Fig. 48). In addition, transient photocurrent responses and electrochemical impedance spectroscopy (EIS) measurements (Supplementary Fig. 49 and Table S7) also reveal that PDA loading enhances the transfer of NIR-excited trap electrons from SA-TCPP to PDA. The transferred trap electrons subsequently facilitate the further reduction of •OOH radicals adsorbed on Q moieties of PDA to form C-OOH intermediates (TS-1 in Fig. 1c) and then spontaneously hydrolyze to form H₂O₂, as confirmed by in situ FTIR and in situ Raman analyses (Supplementary Fig. 50). Because this pathway avoids the desorption of •OOH radicals from Q moieties (Fig. 1c), the electron transfer for their further reduction will be more favorable than the direct reduction of •OOH radicals in PDA alone. Also note that although the formation of •OOH and H₂O₂ may consume SQ• by converting it to Q, subsequent electron transfer from SA-TCPP efficiently regenerates SQ•, as illustrated by the quantitative EPR analysis before and after the extended reaction in Supplementary Fig. 51. According to the above discussions, the complete H₂O₂ production process mediate by SQ• radicals is summarized in Supplementary Fig. 52.

The enhancement of NIR utilization by combining PDA with persistent SQ• radicals is also verified in additional systems. Four NIR-adsorption photocatalysts^41,42, including Bi₁₂CoO₂₀, Black-TiO₂, (Tetracarboxyphthalocyaninato) cobalt (II), and Zn-TCPP, are selected as model systems. Upon PDA loading, each material exhibits an approximately 2 ~ 8-fold increase in photocatalytic activity compared with its pristine counterpart, demonstrating the general applicability of the PDA modification strategy. (Supplementary Fig. 53).

H₂O₂ generation on PDA/SA-TCPP photocatalysts under simulated and real sunlight

Encouraged by the high NIR-driven H₂O₂ production originating from rapid charge transfer from the trap states of SA-TCPP to PDA, we further evaluated the performance of PDA/SA-TCPP under both simulated and real sunlight. Figure 5a shows the H₂O₂ generation activity of PDA/SA-TCPP when the reaction suspension is irradiated and heated solely under simulated sunlight (AM 1.5 G). The H₂O₂ generation rate reached 1.07 ± 0.11 mM h⁻¹, corresponding to an SCC efficiency of 2.2%. This value is nearly twice that of pristine SA-TCPP, and it is competitive among most reported photocatalysts (Supplementary Table S2). Notably, besides the rapid charge transfer, the high SCC may also be owing to the superior NIR photothermal conversion of PDA compared to SA-TCPP alone. In Supplementary Figs. 54 and 55, the solution temperature of PDA/SA-TCPP rapidly increases to 369.45 K within only 40 min of irradiation, nearly 30 K higher than that of SA-TCPP. This intrinsic heating effect also alleviates the need for external thermal input that is otherwise required in standalone SA-TCPP systems to sustain high H₂O₂ generation activity.

Fig. 5. H₂O₂ generation on PDA/SA-TCPP under simulated and natural sunlight.

a H₂O₂ generation and the corresponding SCC efficiency of the PDA/SA-TCPP photocatalyst under AM 1.5 G illumination (The SCC was calculated based on the average H₂O₂ concentration). b Cycling performance of the PDA/SA-TCPP photocatalyst under AM 1.5 G illumination. Reaction conditions: Cat., 200 mg, Sln.,50 mL H₂O with O₂ bubbling, Light source, Xe lamp equipped with AM1.5 G filter (100 mW/cm²). c Photograph and H₂O₂ accumulation on PDA/SA-TCPP cast on a Petri dish under natural sunlight, with corresponding light intensity. Measurement period: Sept. 12-14, 2024. Diameter: 10.8 cm. Location: Wuxi, China (120.27° E, 31.48° N). Ambient temperature: 303.15 K. d Photograph of the degradation setup using in situ produced H₂O₂ on PDA/SA-TCPP. MB dye (5 ppm) is used as a representative pollutant. The outlet pipe transferring the generated H₂O₂ to the MB solution is covered with a semi-permeable membrane to prevent PDA/SA-TCPP powder from entering the MB solution. The inset shows the MB solution after degradation. Error bars represent the standard deviation of the mean, n = 3.

The stability of PDA/SA-TCPP is then examined. In ten consecutive cycles (40 h in total), an average of 3.76 mM H₂O₂ is produced per cycle without any observable activity decay. (Fig. 5b) FTIR spectra and SEM images after the cycling reaction confirm negligible morphological and structural change in PDA/SA-TCPP before and after the stability tests (Supplementary Fig. 56). Long-term operation under AM 1.5 G illumination for three extended cycles (72 h, Supplementary Fig. 57) also demonstrated stable performance, with each cycle yielding 14.67 mM H₂O₂. Motivated by this durability and efficiency, we further evaluated H₂O₂ generation using a lab-built reaction–evaporation device (Supplementary Fig. 58). Remarkably, the device produced H₂O₂ at a concentration of 1.6 wt% within 10 h, reaching a commercially relevant level.

Next, we perform outdoor tests under natural sunlight for H₂O₂ generation in the air to further confirm its robustness. For three consecutive days using the same photocatalyst, the PDA/SA-TCPP powder achieves a SCC efficiency of 1.81% in average, slightly lower than that obtained under laboratory conditions due to solar intensity and solution temperature (Supplementary Fig. 59 and Table 8). Moreover, the photocatalyst film is cast onto a Petri dish with a diameter of 10.8 cm. In particular, the PDA/SA-TCPP powder is ultrasonically dispersed in ethanol, and the resulting suspension is poured into the Petri dish and dried at room temperature to obtain a film. The film exhibits good adhesion to the Petri dish substrate and doesn’t peel off during the outdoor tests. After 1 h reaction under ambient air with simultaneous heating and irradiation by the real sunlight, the H₂O₂ generation amount still reaches above 1 mM, and the corresponding SCC efficiency is ca. 0.12% for this large area film (Fig. 5c). The obtained activity remains relatively high among reported supported systems (Supplementary Table 9). The accumulation amount reaches ca. 4 mM, and the SCC efficiency remains essentially unchanged during an additional 3 h of operation, with the solution temperature stabilizing at ca. 333.15 K. The SCC efficiency also surpasses the average solar-to-biomass conversion efficiency of natural photosynthesis in plants (0.1-1.0%), highlighting the potential of PDA/SA-TCPP for practical solar-driven H₂O₂ production^43–45.

Given the excellent photocatalytic H₂O₂ production activity of PDA/SA-TCPP, an integrated system for in-situ H₂O₂ generation and pollutant degradation is designed, as illustrated in Fig. 5d. Representative pollutants, including methylene blue (MB) and ciprofloxacin (CIP), are employed to evaluate the removal efficiency of the system. In the device, the concentration of in-situ produced H₂O₂ reaches a stable level of 2.1 ± 0.2 mM during the reaction at a flow speed of 2 mL/min. Within 12 min, nearly 95% of both pollutants are removed, corresponding to a degradation rate of 0.23 min⁻¹ and 0.21 min⁻¹, respectively. (Supplementary Fig. 60) These results demonstrate the effectiveness of the H₂O₂-producing system for pollutant removal and water purification.

Discussion

In conclusion, our study demonstrates that loading PDA nanoparticles onto SA-TCPP supramolecular photocatalysts dramatically enhances the activity for photosynthesis of H₂O₂ from only H₂O and O₂. The integration of PDA improves the conversion of NIR photons to H₂O₂, achieving the AQY values of 7.1%, 4.8% and 2.8% at 850 nm, 940 nm, and 1020 nm, respectively. The advantage contributes to a remarkable H₂O₂ generation rate of 3.37 mM/h and an SCC of 2.2% under full-spectrum irradiation. Persistent SQ• radicals in PDA are demonstrated to enable the rapid transfer (79 fs) of sub-gap electrons from SA-TCPP to PDA and facilitate the generation of •OOH radicals under NIR light irradiation. This work not only offers a simple approach to designing effective H₂O₂ generating photocatalysts that harness NIR energy but also holds potential for application across a broader range of materials.

Methods

Materials

Ammonia solution (NH₃·H₂O, 28–30.0%, Sinopharm Chemical Reagent Co., Ltd.); ethanol, anhydrous (EtOH, ≥ 99.7%, Sinopharm Chemical Reagent Co., Ltd.); hydrochloric acid (HCl, 36–38.0%, Sinopharm Chemical Reagent Co., Ltd.); sodium hydroxide (NaOH, ≥ 96.0%, Sinopharm Chemical Reagent Co., Ltd.); isopropanol (IPA, ≥ 99.7%, Sinopharm Chemical Reagent Co., Ltd.); sodium sulfate (Na₂SO₄, 99%, Sinopharm Chemical Reagent Co., Ltd.); 2,2,6,6-tetramethylpiperidine 1-oxyl (TEMPO, 98%, Innochem); nitroblue tetrazolium (NBT, 98%, Innochem); formic acid (HCOOH, 88%, Sinopharm Chemical Reagent Co., Ltd.). All commercial chemicals were used without further purification. Ultrapure water (18.20 MΩ) was used in all experiments.

Preparation of polydopamine (PDA)

PDA is prepared by a self-polymerization method from dopamine hydrochloride (98%, Beijing InnoChem Science & Technology Corp., Ltd., China)⁴⁶. In a typical process, 2.0 mL ammonia aqueous solution (NH₃·H₂O, 28-30%) is added to a 130 mL ethanol solution (30 vol%). The solution is stirred at room temperature for 30 min, and then, 10.0 mL dopamine hydrochloride solution (50.0 mg/mL) is added to the solution. The mixture solution becomes pale brown at first. During stirring, the mixture solution converts into dark brown and then gradually turns into black, indicating the formation of PDA particles. After 48 h of reaction, the PDA is collected by centrifugation (5440 × g) and washed with water three times. The collected solids are dried in a vacuum box at 333.15 K for 12 h to obtain the final product.

Preparation of SA-TCPP supermolecule photocatalysts⁹

Typically, 500.0 mg of meso-Tetra(4-carboxyphenyl) porphyrin (TCPP, 97%, Beijing InnoChem Science & Technology Co, Ltd, China) is fully dissolved with 10.0 mL of 0.5 M KOH aqueous solution. 0.1 M HCl is gradually introduced using a constant-pressure funnel until the pH of the solution reaches 4.0. Subsequently, the suspension is stirred for 72 h to obtain the self-assembled TCPP (SA-TCPP). The resulting precipitate is isolated by centrifugation (5440 × g) and repeatedly washed with ultrapure water until pH ≈ 7.0. The collected SA-TCPP is finally dried under vacuum at 313.15 K prior to further use. Note that to be consistent with the preparation of PDA/SA-TCPP composites, the preparation of SA-TCPP photocatalysts is carried out at room temperature, rather than at 353.15 K in an oil bath as previously reported by us. Though the preparation temperature change leads to different crystallization of SA-TCPP, the H₂O₂ generation activity of SA-TCPP is similar, probably due to the same short-range ordering (J-aggregation) as verified by the PL spectra (Supplementary Fig. 61).

Preparation of PDA/SA-TCPP composites

Typically, PDA/SA-TCPP composites are prepared by a self-assembly precipitation method at room temperature. The prepared 220.0 mg of PDA and 200.0 mg of TCPP are added into 10.0 mL of 0.5 M NaOH solution, with then 100.0 mL of water is added to the mixture, and the mixture is stirred 2 h to form a transparent dark purple solution.

0.1 M HCl aqueous solution is dropped into the solution by a constant pressure dropping funnel until the pH of the solution decreases to 4.0. Then the black precipitants are formed in the solution, and the final products (PDA/SA-TCPP composites) are centrifuged (5440 × g) and washed with water three times until the pH is 7.0. The obtained PDA/SA-TCPP is dried in a vacuum box at 353.15 K overnight. It is worth noting that the term “PDA/SA-TCPP” mentioned in the article specifically refers to the photocatalyst with an optimal loading amount ratio (m_PDA: m_SA-TCPP = 1.1: 1). In addition to the optimal PDA loading amount, other mass ratios between PDA and SA-TCPP photocatalysts are prepared, and including the mass ratios of PDA: TCPP are 0.5, 1, 1.2, and 1.5 (Supplementary Figs. 62–64). The sample is denoted as x-PDA/SA-TCPP. Note that the PDA/SA-TCPP are prepared from commercially available chemicals, easily manipulated methods, and common laboratory equipment, indicating good potential for scale-up preparation.

Characterization

The lattice structure is recorded by powder X-ray diffraction (XRD) on a D8 diffractometer equipped with Cu Kα radiation (Bruker AXS, Germany). The sample morphology is characterized using a transmission electron microscope (TEM JEM-2100, JEOL, Japan) operating at an acceleration voltage of 200 kV, and a field-emission scanning electron microscope (FESEM S-4800, Hitachi, Japan). Optical absorption properties in the ultraviolet-visible-near-infrared region are analyzed with a UV-3600 spectrophotometer (Shimadzu, Japan). The bonding information is analyzed by Fourier transform infrared (FTIR) spectroscopy with a Nicolet iS5 instrument (Thermo, USA) employing the KBr pellet method. The band structure is investigated by ultraviolet photoelectron spectroscopy (UPS) using an ESCALAB Xi⁺ system (Thermo, USA). The related radicals are characterized by electron paramagnetic resonance (EPR) at X-band frequency (9.8 GHz) on a Bruker EMXplus-6/1 spectrometer (Germany).

Photocatalytic H₂O₂ production tests

The photocatalytic performance of PDA/SA-TCPP for H₂O₂ production is evaluated under continuous O₂ bubbling (100 mL min^-1). The light irradiation is achieved by using a 300 W Xe lamp equipped with a 420 nm cut-off filter (Beijing Perfectlight Technology Co., Ltd, China). In a typical experiment, 0.2000 g photocatalyst is ultrasonically dispersed in 50 mL of water (pH ≈ 4, adjusted by adding 2-3 drops of 0.1 M HCl) in a 100 mL glass bottle. A rubber stopper is capped onto the glass bottle. The rubber stopper is pinned with a PTFE pipe and a needle, which serve as the gas inlet and outlet, respectively⁴⁷. The H₂O₂ activity is evaluated at 353.15 K unless otherwise specified, the temperature shown to cause negligible decomposition of the generated H₂O₂ (Supplementary Figs. 65–67). Note that despite the decrease in O₂ solubility at 353.15 K, the elevated temperature simultaneously enhances the diffusivity of the dissolved O₂ and the volume-mass transfer rate of O₂ is estimated as ca. 33.8-91.3 mM/h, far exceeding the O₂ consumption (1.6 mM/h) based on the H₂O₂ production rate of PDA/SA-TCPP. (Supplementary Note 1)

Hydrogen peroxide concentrations were determined using the well-established potassium titanium(IV) oxalate spectrophotometric method^48,49. In acidic medium, H₂O₂ reacts with Ti(IV) oxalate to form a yellow Ti-peroxo complex, which exhibits a characteristic absorption at 400 nm. After the desired reaction time, 1.0 mL of solution was filtered through a 0.45 μm syringe filter and mixed with 1.0 mL of 0.02 M Ti(IV) oxalate solution and 3.0 mL of water. The resulting yellow complex was quantified by measuring the absorbance at 400 nm using a UV-3600 spectrophotometer (Shimadzu, Japan). The calibration curve is linear over 0.5–2.0 × 10^-3M (absorbance 0.5–2.0, 1 cm cells), with a precision of ~ 1% (relative mean deviation 1.1%)⁵⁰. Although this method can be affected by fluoride ions or colored species, in our system, this interference was eliminated by using ultrapure water (18.20 MΩ). The method shows comparable accuracy to the conventional iodometric method (Supplementary Fig. 65).

Measurements for apparent quantum yield (Φ_AQY) and solar-to-chemical conversion efficiency (η_SCC)

The photocatalytic reaction is carried out with a 200.0 mg photocatalyst in a glass bottle containing 50.0 mL of water by a 300 W Xe lamp equipped with a quartz optical fiber as the light source. The bottle is kept at 353.15 K in an oil bath. During the reaction, the Xe lamp is set over the reaction solution with a constant distance of 4.0 cm, O₂ is bubbled into the bottle. The reaction is performed for 0.5 h with magnetic stirring. The monochromatic light for AQY measurements is obtained using a series of band-pass filters (400 ± 10, 420 ± 10, 490 ± 10, 550 ± 10, 600 ± 10, 650 ± 10, 700 ± 10, 850 ± 10, 940 ± 10, and 1020 ± 10 nm). The photon number entered into the reaction bottle is determined with a UV spectral illuminance meter (OHSP-350 UV, Hangzhou, China)⁹. The apparent quantum yields for H₂O₂ generation are determined using the equation:

$${\Phi }_{\mathrm{AQY}}\left(\%\right)=\frac{\left({\mathrm{H}}_2{\mathrm{O}}_2\mathrm{formed}\left(\mathrm{mol}\right)\times {\mathrm{N}}_{\mathrm{A}}\times \mathrm{h}\times \mathrm{C}\right)}{\left(I\times A\times \lambda \times t\right)}\times 100$$ 1

where N_A is the Avogadro constant (6.02 × 10²³mol⁻¹); h is the Planck constant (6.626 × 10⁻³⁴ J·s); C is the light speed (3.0 × 10⁸m/s); I is the light intensity (W/m²); A is the irradiation area (0.25π × 10⁻⁴ m²); λ is the wavelength of incident light (m), and t is the reaction time (1800 s).

The SCC efficiency for the H₂O₂ production is determined under AM 1.5 G simulated sunlight irradiation (Xe lump with AM 1.5 G cut-off filter, Perfect Light, Beijing, China)⁹. The light intensity is measured by a PL-MW 2000 photoradiometer (Perfect Light, Beijing, China). The η_SCC is determined using the following equation:

$${\eta }_{\mathrm{scc}\left(\%\right)}=\frac{\mathrm{\Delta }\mathrm{G}\times {\mathrm{H}}_2{\mathrm{O}}_2\mathrm{formed}\left(\mathrm{mol}\right)}{t\times S\times {\mathrm{I}}_{\mathrm{AM}}}\times 100$$ 2

where ΔG is the free energy for H₂O₂ production (117 kJ/mol); t is the irradiation time, S is the irradiation area (0.25π × 10⁻⁴ m²), and I_AM is the light intensity (1000 W/m²)

Calculation for the photothermal conversion efficiency

The photothermal conversion efficiencies of the SA-TCPP and PDA/SA-TCPP are determined according to the literature⁵¹. Detailed calculations are given below:

$$\eta =\frac{hS\times \left(T_{max}-T_{surr}\right)-Q_{dis}}{I\times \left(1-{10}^{-A_{940}}\right)}$$ 3

$$hS=-\frac{m_D\times C_D\times \ln \left(\frac{T-T_{surr}}{T_{\max }-T_{surr}}\right)}{t}$$ 4

Where h is the heat transfer coefficient, S is the surface area of the container, and the value of hS is obtained from Eq. (3). T_max is the maximum steady temperature of the solution (311.45 K for SA-TCPP and 317.15 K for PDA/SA-TCPP), T_Surr is the environmental temperature (288.35 K), I is light power (0.49 W), A₉₄₀ is the absorbance of the photocatalyst at 940 nm (For SA-TCPP and PDA/SA-TCPP, the absorbance values are 0.17 and 0.60 respectively), Q_Dis is the heat dissipated from the light absorbed by the solvent and container (0.21 W), m_D is the mass of the solvent (0.1 g), C_D is the specific heat capacity of the solvent (4.2 J/g·K), and t is the irradiation time (900 s).

Electrochemical and PEC measurements

All electrochemical and PEC measurements (photocurrent, Mott-Schottky, and electrochemical impedance) are performed in 0.5 M sodium sulfate solution (pH = 6.8 ± 0.2) through the traditional three-electrode system using the CHI 660E electrochemical workstation (Shanghai Chenhua Technology Co., LTD). PDA/SA-TCPP/fluorine-doped tin oxide (FTO) glass (1.5 × 1.0 cm²) is used as a working electrode; a platinum net is employed as a counter electrode (1.5 × 1.5 cm², Shanghai Chenhua Technology Co., LTD), and a standard Ag/AgCl electrode (E_Ag/AgCl  =  0.210 V vs. RHE, Shanghai Chenhua Technology Co., LTD) is used as a reference electrode. A 300 W Xe lamp is used as the light source. The resistance of the electrochemical cell was measured using the built-in iR test of the electrochemical workstation prior to electrochemical measurements. The obtained cell resistance was 22.7 ± 0.2 Ω, averaged from three independent measurements.

The PDA/SA-TCPP/FTO is used as the working electrode. The electrode is obtained by finely grinding the PDA/SA-TCPP photocatalyst (10 mg) and then ultrasonicating it in 1.0 mL of ethanol for 1 h to form a uniform suspension. Next, 20.0 µl of the well-dispersed suspension is uniformly cast onto an FTO glass by dropping. The PDA/SA-TCPP/FTO is then dried at room temperature for 2 h before use.

Mott-Schottky plots are measured at alternating current (AC) frequencies of 2000 Hz. The electron density of the defect levels is calculated as follows:

$$\frac{1}{C_d^2}=\frac{2}{A^{2}q{\epsilon N}_A}\left(V_{bi}-V_{DC}\right)$$ 5

where A is the junction area (1.0 cm²), q is the elementary charge (1.6×10^-19C), and ε = ε₀ε_r is the semiconductor permittivity (ε₀ the permittivity of vacuum (8.8 × 10⁻¹⁰F/cm), ε_r the relative dielectric constant of the semiconductor (1.13), N_A is the electron density at defect level, V_bi is the built-in potential and V_DC is the applied DC bias. The plot of 1/C² vs V_DC (Mott-Schottky plot) yields a straight line in the range of small forward DC bias (− 0.1 V− 0.2 V), N_A, and V_bi is extracted by fitting the straight line with Eq. (5).

The rotating disk electrode test

The rotating disk electrode (RDE) test is performed on a CHI-750E electrochemical workstation, operated via computer control and a conventional three-electrode setup. The glassy carbon disk coated with the sample serves as the working electrode, while an Ag/AgCl electrode and a platinum mesh electrode are employed as the reference and counter electrodes, respectively⁹. The electrolyte consists of 100 mL of 0.5 M Na₂SO₄ solution containing 80 mg of dopamine hydrochloride (DA·HCl), which is the monomer precursor of polydopamine. Linear sweep voltammetry scans are conducted at a scan rate of 10 mV s⁻¹ under argon saturation across various rotation speeds. After each scan, argon was bubbled for 30 minutes to remove residual oxygen (O₂). The average electron transfer number (n) involved in the redox process from SQ· to Q is determined from the slopes of the Koutecky−Levich plots, using the following equations:

$$j^{-1}=j_k^{-1}+B^{-1}{\mathrm{\omega }}^{-1/2}$$ 6

$$B=0.2nF{\mathrm{\nu }}^{-1/6}C{D}^{2/3}$$ 7

where j is the measured current, j_k is the kinetic current density, B is the proportionality coefficient, and ω is the electrode rotating speed (rpm), respectively. F is the Faraday constant (96485 C mol⁻¹), ν is the kinetic viscosity of water (0.01 cm²·s⁻¹), C is the concentration of DA⋅HCl in solution (6.52 × 10⁻⁶ mol/cm³), and D is the diffusion coefficient (6.5 × 10⁻⁶  cm²/ s) of DA⋅HCl⁵².

The calculations of volume-mass transfer rate of O₂

The diffusion coefficients of O₂ in water at 298.15 K and 353.15 K are calculated according to the Wilke-Chang correlation⁵³.

$$D_{AB}=\frac{7.4\times {10}^{-8}{\left(\varphi M_B\right)}^{1/2}T}{\eta V_A^{0.6}}$$ 8

where D_AB is the diffusion coefficient (cm²/s), φ is the solvent association factor (2.26 for water), M_B is the molecular weight of the solvent (18.0 g/mol), T is the absolute temperature (K), η is the dynamic viscosity (298.15 K: η = 0.890 cP, 353.15 K: η = 0.355 cP), and V_A is the molar volume of O₂ at its boiling point (25.6 cm³/mol)⁵⁴.

The volume-mass transfer rate of O₂ at 353.15 K is calculated according to the diffusion coefficients. In our experimental setup, O₂ is continuously bubbled throughout the photocatalytic process. In such a condition, the instantaneous dissolved O₂ concentration is not solely governed by temperature-dependent solubility. Instead, the supply of O₂ to our photocatalyst is determined by the volume-mass transfer rate of O₂, which in the liquid phase can be expressed by the classical gas-liquid mass transfer model as ref. ⁵⁵:

$$R_{O_2}=\frac{k_{L}a\left(C^{*}-C\right)}{M_{O_2}}=\frac{Sh{D}_{O_2}}{L}\times \frac{a\left(C^{*}-C\right)}{M_{O_2}}$$ 9

where R_O2 represents the molar supply rate of O₂ per unit volume (mM/h), K_La is the volume-mass transfer coefficient, D_O2 is the liquid-phase diffusion coefficients of O₂ (6.72 × 10⁻⁵cm²/s at 353.15 K), Sh is the Sherwood number (10)⁵⁶, L is a characteristic length (distance from the outlet of the gas-bubbling pipe to the surface of suspended photocatalysts, set in this work as the average of the liquid depth, 2 cm), C^* is the solubility of O₂ in water (4.0 mg/L at 1 atm and 353.15 K), a is gas-liquid interface area ratio (300–800 cm⁻¹)⁵⁷, C is the instantaneous dissolved O₂ concentration in water (2.0 mg/L, which is the residual O₂ concentration after per minute H₂O₂ production), and M_O2 is the molar mass of oxygen (32 g/mol).

Determination of light intensity

Photocatalytic reactions were carried out using a 300 W xenon lamp equipped with a 420 nm cut-off filter. The average light intensity incident on the reaction area was determined to be 200 mW cm⁻². Considering the cross-sectional area of our reactor, the intensity was measured at five representative positions within a circular area of 3 cm radius, including the center (spot A) and four positions at the edge (spots B-E) (Supplementary Fig. 68).

The average light intensity I was calculated using the following weighted equation:

$$I=\frac{\pi r_1^2\times I_{\text{A}}}{\pi r_3^2}+\frac{1}{2}\times \frac{\pi \left(r_2^2-r_1^2\right)\times \left(I_{\text{B}}+I_{\text{D}}\right)}{\pi r_3^2}+\frac{1}{2}\times \frac{\pi \left(r_3^2-r_1^2\right)\times \left(I_{\text{C}}+I_{\text{E}}\right)}{\pi r_3^2}$$

The measured light intensities were as follows: spot A, 900 mW cm^-2; spot B, 220 mW cm⁻²; spot C, 57 mW cm⁻²; spot D, 200 mW cm⁻²; and spot E, 51 mW cm⁻².r₁, r₂, and r₃ correspond to distances of 1 cm, 2 cm, and 3 cm, respectively.

DFT calculation

The DFT calculations are performed on Material Studio with the Dmol3 package⁵⁸. The geometry optimization is performed first to relax the reactants and products, respectively, and then, the transition state calculations are carried out to obtain the energy barriers for SQ⋅ with O₂ to produce H₂O₂. For the transition state calculations, the linear synchronous transit/quadratic synchronous transit (LST/QST) method is used. The exchange and correlation functions are treated by the Becke three-parameter Lee−Yang−Parr (B3LYP) level of theory, taking into account the bulk solvent effects of water. The tolerances for geometry optimization and TS calculations are set with the total energy of 5 × 10⁻⁵Ha/atom, the maximum force of 0.002 Ha/Å, the maximum ionic displacement of 5 × 10⁻³Å, and the maximum root-mean-square (RMS) forces of 0.01 Ha/Å. To calculate the LUMO orbitals of PDA, a structural model consisting of nine moieties containing Q, SQ•, and QH₂ units with varying saturation degrees of the indole is constructed. The exchange and correlation functions are also treated by the B3LYP. Geometry optimization is performed with convergence criteria of 1 × 10⁻⁴ Ha/atom for total energy, 0.02 Ha/Å for maximum force, and 0.05 Å for maximum ionic displacement. Note that, considering the inherent challenges in accurately modeling the exact material under realistic catalytic conditions, the DFT calculations in this work are intended primarily to provide supportive insight for the experimental results.

Author contributions

S.D. conducted the experiments and performance evaluation. Y.N.Z. performed DFT calculations and analysis of the TAS data. J.X. and Y.Z. performed data analysis. J.Z., Y.D., and Y.L. participated in the paper discussions and interpretation of the results. Y.F.Z. supervised the project. C.P. wrote the manuscript with critical revision from all authors.

Peer review

Peer review information

Nature Communications thanks Minsu Gu, Qingrui Zhang and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

The data that support the findings of this study are available within the main text and its Supplementary Information. Atomic coordinates of the optimized computational models of PDA/SA-TCPP are available in Supplementary Data 1. Source data are provided in this paper.

Competing interests

The authors declare no competing financial interest.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-70130-x.