Keto-enol tautomerism as dynamic electron/hole traps promote charge carrier separation for hydrogen peroxide photosynthesis

Abstract

Covalent organic frameworks (COFs) are promising photocatalysts for H₂O₂ photosynthesis, but charge carrier separation remains a critical challenge. Donor-acceptor COFs enhance charge separation, but the slow kinetics of water oxidation and oxygen reduction reactions lead to carrier accumulation, thereby decreasing efficiency. Here, we report T-C type COFs (T = trap units, C = catalytic units), demonstrating that units with keto-enol tautomerism can serve as dynamic electron/hole traps (T) to mitigate Coulomb forces. This design effectively facilitates swift charge transfer and extends carrier lifetimes, thereby enhancing reactions at the C units. Imine COFs derived from 2,4,6-trihydroxybenzaldehyde (Tp) outperform those based on 1,3,5-benzenetricarboxaldehyde due to tautomerization. The optimal Tp COF (TpBpy) achieves an H₂O₂ generation rate of 37.9 μmol h⁻¹ (or 8350 μmol h⁻¹ g⁻¹) under simulated light, and a solar-to-chemical conversion efficiency of 0.038% in a flow reactor under natural sunlight. This work provides molecular design strategies and standard criteria for efficient H₂O₂ photocatalysts.

This study shows that keto-enol tautomerism in 2,4,6-trihydroxybenzaldehyde (Tp) imine COFs acts as variable electron/hole traps, boosting exciton dissociation and charge transfer. Tp COF (TpBpy) achieves an photocatalytic H₂O₂ generation rate of 8350 μmol h^-1 g^-1 under simulated light, and a solar-to-chemical conversion efficiency of 0.038% in a flow reactor under natural sunlight.

Introduction

Covalent organic frameworks (COFs) represent an innovative category of photocatalytic materials that have gained significant attention for the photocatalytic production of hydrogen peroxide (H₂O₂)^{1, 2–3}. The advantages of COFs can be summarized as follows: (1) The highly organized porous architecture of COFs significantly increases the number of exposed active sites while reducing the carrier migration distance to the sub-nanometer range, thus enhancing the efficiency of photocatalytic processes; (2) The spectral absorption range of COFs can be broadened to encompass the visible, and infrared regions through targeted molecular design, thereby creating an satisfactory platform for harnessing visible light energy, and infrared thermal energy; (3) Efficient design of specific reaction active sites can be achieved by considering the unique characteristics of various photocatalytic reactions^{4, 5, 6, 7, 8, 9, 10, 11–12}. However, the current photocatalytic performance of COFs in H₂O₂ photosynthesis is far from meeting industrial requirements. The primary reason is that, although the sub-nanometer migration distance of charge carriers in COFs provides convenience for catalytic reactions, this same scale of migration distance is incapable of sufficiently weakening the Coulomb forces between photogenerated electrons, and holes. Consequently, this limitation hampers the separation of charge carriers¹³. In other words, the charge-separating ability of COFs is limited, which poses a considerable obstacle to the sustained separation of photogenerated carriers, thereby challenging their practical application in H₂O₂ photosynthesis¹⁴.

Over the past decade, researchers have been extensively focusing on the problem of photogenerated charge carrier separation in COFs. The most promising approach is to take advantage of the effective molecular design features of COFs to construct a series of donor acceptor (D-A) structured COFs^{15, 16–17}. By regulating the charge distribution difference between donor, and acceptor units, a push-pull effect mechanism is induced, which facilitates the migration of photogenerated electrons, and holes to the acceptor, and donor units, respectively¹⁸. However, although the D-A structure holds promise for improving the separation of photogenerated charge carriers, a considerable disparity persists between the swift movement of these carriers, and the relatively sluggish kinetics of the water oxidation, and oxygen reduction reactions occurring at the D and A units. The reaction rates are generally several orders of magnitude lower than the migration rates of the charge carriers. This imbalance results in the accumulation of photogenerated carriers, which leads to significant Coulombic interactions among them. Such interactions hinder the separation efficiency of the photoinduced electron-hole pairs in D–A structured COFs, ultimately restricting the photoactivity of COFs for H₂O₂ photosynthesis. This challenge is currently a widespread issue encountered by nearly all existing photocatalysts, and photocatalytic processes¹⁹.

At present, the most effective approach to addressing the conflict between swift carrier migration and sluggish oxidation-reduction reactions is the addition of hole-sacrificial reagents to the photocatalytic H₂O₂ production system. These sacrificial reagents rapidly consume the photogenerated holes present on the catalyst surface, thereby diminishing the Coulombic forces among photogenerated carriers. This process results in an increase in the number of photogenerated electrons that are available for O₂ reduction, ultimately enhancing H₂O₂ photosynthesis^20,21. However, this method also possesses several significant drawbacks. Firstly, this strategy is ineffective in simultaneously enhancing both photocatalytic H₂O oxidation and O₂ reduction reactions, as the photogenerated holes are consumed by the sacrificial reagents rather than being employed for the water oxidation process^{22, 23, 24, 25, 26–27}. Secondly, the introduction of sacrificial reagents complicates the catalytic system, and has an adverse effect on the purity of the H₂O₂ solution. Lastly, the incorporation of these sacrificial reagents increases economic costs. Moreover, the majority of these sacrificial reagents are not environmentally friendly, giving rise to considerable environmental problems^{27, 28, 29, 30, 31, 32, 33, 34–35}. Therefore, to address the discrepancy between rapid carrier migration and slow oxidation-reduction reactions, it remains essential to concentrate catalyst design. Here, we draw inspiration from the sacrificial reagent scheme to create units that function as dual traps for electrons and holes within the catalyst. These dual-trap units are engineered with two main functions. Capturing photogenerated electrons allows holes to participate in photocatalytic oxidation reactions, and alternately, capturing photogenerated holes allows electrons to participate in photocatalytic reduction reactions. This methodology effectively reduces the Coulombic forces between photogenerated electrons, and holes, enabling ultrafast charge transfer, and longer-lived free charge carriers. This ultimately efficient photocatalytic H₂O oxidation, and O₂ reduction reactions, promoting H₂O₂ photosynthesis.

Based on the above considerations, we have designed a series of Tp (2,4,6-trihydroxybenzaldehyde-1,3,5-tricarbaldehyde) imine COFs with keto-enol tautomerism, which exhibit significantly higher photocatalytic performance compared to the BT (1,3,5-benzenetricarboxaldehyde) series COFs that lack of keto-enol tautomerism. Mechanistic studies indicate that Tp-series imine COFs display enhanced photoinduced dynamic behaviour, including increased photocurrent density, decreased charge transfer resistance, and enhanced surface photopotential. Taking the optimal TpBpy photocatalyst as a case study, in situ infrared spectroscopy, femtosecond time-resolved spectroscopy, and theoretical calculations have demonstrated that the Tp unit in TpBpy alternates between electron, and hole traps during the keto-enol tautomerism process. When Tp is in the ketone form, it acts as an electron trap, facilitating the migration of electrons towards Tp and holes towards Bpy, resulting in the oxidation of adsorbed water on Bpy to generate H₂O₂. When Tp is in the enol forms, it acts as a hole trap, facilitating the migration of holes towards Tp and electrons towards Bpy-H⁺, resulting in the reduction of adsorbed oxygen on Bpy-H⁺ to generate H₂O₂. Consequently, TpBpy exhibits a rapid rate of H₂O₂ photosynthesis, reaching values of 37.9 μmol h⁻¹ or 8350 μmol h⁻¹ g⁻¹, which establishes it as the most effective catalyst for H₂O₂ photosynthesis known to date. Our experimental results support the distinctive function of the electron/hole trap interconversion unit in diminishing the Coulombic interactions between photogenerated electrons and holes. This directly accelerates charge transfer and enhances the photocatalytic redox process. This charge carrier separation mechanism directly correlates the notable photocatalytic performance of the Tp-series imine COFs with the presence of keto-enol tautomerism.

Results and discussion

To affirm our hypothesis, we analyzed and summarized over forty-nine published papers on Tp imine COFs (Table S1). We found that the photocatalytic performance of all Tp imine COFs was much higher than that of BT imine COFs (Fig. 1a). To further demonstrate the universality of this phenomenon, we designed and synthesized eight Tp imine COFs and eight corresponding BT imine COFs, with their unit structures shown in Fig. S1. The synthesis procedures, material characterization findings, and photoelectrochemical tests of these COFs are provided in the supporting information (Figs. S2, S4 for BTPa and TpPa; Figs. S5–S7 for BTDpt and TpDpt; Figs. S8–S10 for BTTAPT and TpTAPT; Figs. S11-S13 for BTTBPT and TpTBPT; Figs. S14–S16 for BT-2F and Tp-2F; Figs. S17–S19 for BT-CH₃ and Tp-CH₃; Figs. S20–S22 for BT-COOH and Tp-COOH; Figs. S23–S28 for BTBpy and TpBpy). Then, we assessed the photocatalytic activity of these COFs for H₂O₂ photosynthesis. As depicted in Fig. 1b and Table S2, the Tp series of imine COFs exhibited significantly enhanced performance in comparison to the BT series of imine COFs, with improvements ranging from several to dozens of times, regardless of the incorporation of monomers with varying lengths, diverse functional group modifications, and differing numbers of linkages (C2 and C3). Among them, TpBpy exhibited the superior photocatalytic activity (Fig. 1b), achieving H₂O₂ photosynthesis rate of 37.9 μmol h⁻¹ (Fig. 1c and Fig. S27). Notably, TpBpy achieved an apparent quantum yield (AQY) of 8.1% (Fig. S28a). This rate is greater than those previously recorded for COFs across a range of bonding types. This includes the covalent triazine series³⁶, sp² carbon-conjugated series³⁷, polyimide series³⁸, hydrazone-linked series³⁹, imine series^{40, 41, 42, 43, 44, 45, 46, 47, 48–49}, and other bonding categories⁵⁰, in addition to polymers such as resins⁵¹ and g-C₃N₄⁵², as illustrated in (Fig. 1d and Table S3). Additionally, TpBpy demonstrated satisfactory cycling performance (Fig. S28b). As shown in Fig. S28b, after five cycles, the catalytic performance of the catalyst remained virtually unchanged. Moreover, in the FTIR spectrum of the TpBpy catalyst after H₂O₂ photosynthesis, apart from the characteristic peaks of the enol bonds appeared in FTIR (Fig. S28d), there were negligible changes. Meanwhile, the XRD patterns (Fig. S28c) displayed negligible changes before, and after the reaction. This observation clearly indicates the stability of the catalyst. Simultaneously, TpBpy shows significantly greater activity than BTBpy in photocatalytic hydrogen generation, benzoamine conversion, and CO₂ reduction (Fig. S29). Clearly, the photocatalytic outcomes discussed above align with the various findings mentioned in Fig. 1a. This further confirms our initial conjecture that studying imine COF synthesized from Tp is meaningful, and also provides additional evidence that the potential keto-enol tautomerism in Tp imine COFs has a significant enhancing effect on H₂O₂ photosynthesis.

Fig. 1 Photocatalytic performance comparison of Tp- and BT-Based COFs for H₂O₂ synthesis. [Images not available. See PDF.]

a A diagram depicting the structural and performance disparity between Tp- and BT-based COFs. b Photocatalytic H₂O₂ production of eight-group COFs samples in H₂O₂ synthesis (Conditions: 298 K, Xenon lamp, λ > 420 nm, light intensity of 100 mW cm^-2, 10 mg catalyst, 10 mL water, and air flowing). c H₂O₂ generation activity of TpBpy and BTBpy in pure water. The error bars in c are the standard deviations calculated from measurements of three independent samples across parallel experiments. d H₂O₂ production rate comparison of various COFs, highlighting the superior performance of TpBpy (Conditions: 298 K, Xenon lamp, λ > 420 nm, light intensity of 100 mW cm⁻², 1 mg catalyst, 10 mL water, and air flowing) relative to other reported COFs.

To illustrate the significance of keto-enol tautomerism in Tp imine COFs, we focused on the imine/amine bonds, and enol/keto groups inherent in the Tp unit of TpBpy. Initially, we compared the H₂O₂ photosynthesis activity of TpBpy with that of BTBpy, which does not contain hydroxyl groups, and with OMe-Bpy, where methoxy groups replace the enol/keto groups within Tp units. The material characterization, and photoelectrochemical performance of OMe-Bpy are presented in Figs. S30–S32. The results reveal that the H₂O₂ production rates for both BTBpy and OMe-Bpy are considerably lower than that of TpBpy (Table S4), highlighting the critical role of hydroxyl groups in the keto-enol tautomerism process. Furthermore, regarding the imine/amine bonds, we compared the activity of TpBpy with S-TpBpy (Fig. S33–S35) and CN-SP₂ (Fig. S36–S38), both of which lack bonds capable of converting imine to amine. Their markedly low photocatalytic activity further revealed that imine/amine bonds are important functional groups for the keto-enol tautomerism process (Fig. 2a and Table S4). The Fourier transform infrared (FT-IR) spectra (see Fig. S2c for BTTa and TpPa; Fig. S5c for BTDpt and TpDpt; Fig. S8c for BTTAPT and TpTAPT; Fig .S11c for BTTBPT and TpTBPT; Fig. S14c for BT-2F and Tp-2F; Fig. S17c for BT-CH₃ and Tp-CH₃; Fig. S20c for BT-COOH and Tp-COOH; Fig. S23c for BTBpy and TpBpy) of the Tp imine COFs showed the presence of keto and amine bonds, while enol and imine bonds were not detected. These FTIR findings indicate that the Tp-derived imine COFs, represented in E₃K₀ state (as illustrated in Fig. 2a) have undergone a thermodynamic transformation to β-ketoenamine-linked COFs, referred to E₀K₃ state for TpBpy^{16,53, 54–55}. However, it remains to be investigated whether the fully ketonic TpBpy can transition to enol-containing forms under varying external conditions such as light exposure, as well as environments with oxygen, and water. To explore this, the chemical structure of TpBpy subjected to H₂O₂ photosynthesis was examined using FT-IR (Fig. 2b,c), solid-state ¹³C nuclear magnetic resonance (NMR, Fig. 2d), and X-ray photoelectron spectroscopy (XPS, Fig. 2e, f). After H₂O₂ photosynthesis, new infrared vibration peaks were detected, corresponding to C-OH at 1417 cm^-1, along with PyH⁺ and C=NH⁺ peaks in the range of 1500 to 1590 cm^-1. Simultaneously, solid-state 13C NMR also exhibited changes, particularly a new chemical shift appearing at 163.9 ppm. A comparison of the data before, and after photocatalysis indicates that the signal at 163.9 ppm was detectable in the sample prior to the reaction but remained very weak due to the extremely low concentration of enol (phenol C-OH). After the photocatalytic reaction, the increased concentration of phenol C-OH resulted in a notable enhancement of this signal, as indicated by the upward arrow in Fig. 2d. Furthermore, a decrease in phenyl carbonyl (C=O: 180–185 ppm) was also observed, as illustrated by the downward arrow in Fig. 2d^16,54,55. Additionally, XPS peaks also indicating new C=N (399.27 eV) and C-O bonds (532.01 eV) emerged. These findings provide evidence for the occurrence of amine-to-imine tautomerism, in line with other research⁵⁶. Photoisomerization results in the conversion of keto-amine structures to enol-imine forms (E₁K₂ and E₂K₁ states for TpBpy, as depicted in Fig. 2a). Thus, the improved photocatalytic performance of Tp imine COFs can be plausibly ascribed to the keto-enol tautomerism phenomenon. Nevertheless, further research is needed to comprehensively elucidate their action mechanisms in photocatalysis.

Fig. 2 Structural and compositional analysis of TpBpy before and after H₂O₂ photosynthesis. [Images not available. See PDF.]

a Schematic of enol/keto and imine/amine group roles (E = enol, K = keto). b, c FT-IR spectral changes of TpBpy during H₂O₂ synthesis. d NMR spectra of TpBpy before and after H₂O₂ photosynthesis (spin sidebands marked with *). e, f N1s and O1s XPS spectra of TpBpy before and after H₂O₂ photosynthesis.

The ability of the samples to dissociate excitons was assessed by investigating the temperature-dependent photoluminescence (PL) spectra to determine the exciton binding energy (E_b). The integrated PL intensity of both TpBpy and BTBpy decreased consistently as the temperature rose from 80 to 280 K (Fig. 3), primarily due to thermally activated non-radiative recombination processes^{57, 58–59}. By using the Arrhenius equation I(T) = I₀/(1 + Aexp(‒E_b/k_BT))⁵⁶ to fit the experimental data, the E_b values of TpBpy and BTBpy were obtained (Fig. 3c). The E_b value for TpBpy was found to be 35.7 meV, significantly lower than that of BTBpy (44.2 meV), suggesting that the incorporation of Tp moieties can accelerate exciton dissociation, and thus promote the generation of long-lived photogenerated charge carriers. Subsequently, the surface potentials of TpBpy and BTBpy were measured using a Kelvin probe force microscopy (KPFM) under both dark (Fig. S39) and under light irradiation (Fig. 3d,e) conditions. When exposed to light, excited electrons are generated within the bulk material, and a portion of these charges can be separated and transported to the material surface. This results in the buildup of charges on the surface, leading to a change in surface potential. As depicted in Fig. 3f, g TpBpy exhibits a significant increase in average surface potential of 33.31 mV under light conditions, while BTBpy shows a minor increase of 5.09 mV compared to their respective dark states. The substantial variation in surface potential for TpBpy suggests that it possesses higher efficiency in charge separation and transport, allowing for a greater number of electrons and holes with longer lifetimes to drive the photoredox reactions. It is important to highlight that we have prepared amorphous TpBpy (AP-TpBpy). While the carrier separation efficiency in amorphous materials is typically inferior to that of crystalline substances due to the absence of periodic structure, AP-TpBpy demonstrates a superior carrier separation effect compared to BTBpy. Additionally, as illustrated in Fig. S40, the surface potential of AP-TpBpy exhibited a change of 23.30 mV before, and after exposure to light (Fig. S40). As expected, the transient photocurrent response of TpBpy shows a greater enhancement compared to that of BTBpy (Fig. 3h). Additionally, the Nyquist plot of TpBpy under irradiation exhibits a smaller semicircle (the radius corresponds to the catalyst/electrolyte interface resistance, R_ct.) than that of BTBpy (Fig. 3i). Consistent with the E_b and surface potential variations, both the transient photocurrent, and Nyquist results indicate that Tp moieties are more effective in separating, and transferring charges than BT moieties. It is worth noting that the transient photocurrent responses and Nyquist plots of other synthesized COFs were also measured, and these findings consistently suggest higher efficiencies in separating and transferring photogenerated charge carriers in the Tp imine COFs with keto-enol tautomerism property.

Fig. 3 Photophysical and photoelectrochemical characterizations of BTBpy and TpBpy COFs. [Images not available. See PDF.]

a, b Temperature-dependent PL spectra with excitation wavelength at 455 nm and c extracted exciton binding energies of BTBpy and TpBpy. AFM images in surface potential mode for d BTBpy and e TpBpy under light illumination. f, g Two surface potential profiles along the lines in (d, BTBpy) and (e, TpBpy) under dark and light, respectively. h Photocurrent response and i EIS spectra of BTBpy and TpBpy.

To investigate the dynamic behavior of excited state charge in TpBpy and BTBpy systems, femtosecond transient absorption spectroscopy (fs-TA) was utilized to observe the processes post photoexcitation of the photocatalyst. The samples were excited with 400 nm pump pulses, and the fs-TA spectrum was recorded using a probe pulse. As shown in Fig. 4a, b, in the wavelength range of 450–515 nm, the negative signal corresponds to ground-state bleaching (GSB) since the BTBpy can absorb these wavelengths (as shown in the UV-vis absorption spectra in Fig. S23d). Subsequently, after 0.75 ps, the negative signal transitions to a positive signal attributed to excited-state absorption (ESA). Additionally, a stimulated emission (SE) feature at 570 nm, consistent with its steady-state PL emission spectra (Fig. S41), emerges. The rise, stabilization, and decay processes of the signals at 475 nm and 570 nm, as depicted in Fig. 4c and Table S5, occur almost simultaneously. The kinetic fitting of the signal transition at 475 nm and 570 nm reveals a rapid rise time constant of 1.66 ps, corresponding to the transfer of excited electrons, and holes via vibrational relaxation to the conduction band (CB) bottom and valence band (VB) top, as illustrated in Fig. 4g. Immediately following photoexcitation, another ESA at 720 nm was observed. The positive ESA signal likely originates from hole transfer from electron-acceptor Bpy moieties to electron-donor BT moieties, as the 2,2′-bipyridine unit possesses stronger electron-accepting ability than the benzene ring unit in BTBpy⁶⁰. The kinetic fitting of the signal transition at 720 nm shows a rapid kinetic with the time constant of 2.37 ps. Notably, the fitting of these kinetics reveal two transition time constants of 1.66–2.37 ps and 247–255 ps, with the fast constant associated with rapid charge transfer ( ${\tau }_{\mathrm{T}}$ ) involving annihilation or exciton (and polaron) dissociation aided by transfer and trapping^27,61,62. Once the exciton or polaron is dissociated, the separated state is long-lived. Consequently, the slow-transition constant is associated with carrier recombination ( ${\tau }_{\mathrm{R}}$ ), as illustrated in Fig. 4I^63,64.

Fig. 4 Transient absorption spectroscopy and charge transfer kinetics of BTBpy and TpBpy COFs. [Images not available. See PDF.]

TA spectra of a BTBpy and d TpBpy excited at 400 nm. Time slices of the TA spectra of b BTBpy and e TpBpy in water. TA kinetics of c BTBpy probed at 475 nm, 570 nm, and 720 nm, and f TpBpy probed at 510 nm and 650 nm. Schematic of electron transitions for g BTBpy and h TpBpy. i Charge transfer kinetic constants derived from TA kinetics.

Alongside the UV–vis absorption spectra of TpBpy presented in Fig. S23d, and considering that the absorption peaks of the Tp and Bpy monomers (illustrated in Fig. S42) are situated in the ultraviolet range, it can be concluded that the negative signal observed at 510 nm corresponds to the GSB of TpBpy. In comparison to BTBpy, the GSB signal of TpBpy shows a notable enhancement (as seen in Fig. 4d, e), indicating that the electron in the ground state of TpBpy is more easily excited to the excited state. Besides, there are noticeable negative signals even after 1000 ps across the entire measuring wavelength range, suggesting the generation of abundant long-lived free charge carriers. Interestingly, TpBpy demonstrates a broadened SE ranging from 540 nm to 750 nm (consistent with its steady-state PL emission spectra in Fig. S41). The emission at 650 nm increases at a slower rate compared to the direct excited state at 510 nm. As depicted in Fig. 4f and Table S6, the rise at 510 nm (corresponding to GSB) is limited by the instrument response function (IRF) and takes ~80 fs, while the rise time at 650 nm (corresponding to SE) takes 0.90 ps. The enhancement of the SE signal reveals that within around 0.90 ps, electrons and holes are continuously captured, resulting in an accumulation of their quantities. This accumulation is accompanied by an increase in the intensity of electron-hole recombination, which is manifested as the enhancement of the SE signal. As schematically depicted in Fig. 4h, keto-enol tautomerism generates E₁K₂ or E₂K₁ states at local positions. These states can be regarded as impurity states (in the figure, their corresponding energy levels are marked in red, which differ from the energy-band structure corresponding to the steady-state E₀K₃ structure, marked in cyan). Due to keto-enol tautomerism, Tp moieties can serve as both electron-trapping and hole-trapping sites, namely double traps. This finding is consistent with our subsequent theoretical calculations. The rapid charge transport is achieved precisely through this electron, and hole trapping effect of Tp moieties. Moreover, because of the diverse local environments of TpBpy (in some areas, the Tp units are in the E₀K₃ state, while in other regions, they are in the E₁K₂ or E₂K₁ states), the SE signal shows inhomogeneous broadening. After the signal enhancement reaches its peak, the signal at 650 nm begins to decay. To analyze this decay process, we perform kinetic fitting on the signal transition at 650 nm (Fig. 4f). The results of this kinetic fitting reveal a decay time constant of 3.14 ps, along with an additional long-lived component (Table S6). The slower decay kinetic (3.14 ps) is probably a result of the trapping process operating in reverse (being released), as shown in Fig. 4h. The kinetic fitting of the signal transition at 510 nm reveals two decay time constants of 1.68 ps and 1.08 ns. It is evident that the lifetime ( ${\tau }_{\mathrm{R}}$ ) of photogenerated free charge carriers in TpBpy (1.08 ns) has been significantly extended in comparison to BTBpy (0.25 ns)⁶⁵. Additionally, TpBpy exhibits a smaller ${\tau }_{\mathrm{T}}$ (0.90–1.68 ps), signifying quicker exciton dissociation (in line with a lower exciton binding energy, E_b) and charge transfer processes. The enhanced charge transport, and prolonged carrier lifetime of TpBpy can be attributed to keto-enol tautomerism.

Furthermore, AP-TpBpy as well exhibits a GSB near 500 nm and SE signal in the range of 600–730 nm. The fs-TA spectra of AP-TpBpy (Fig. S43a, b) closely resemble those of TpBpy (Fig. 4d, e). However, the overall fs-TA signal for the amorphous phase is weaker compared to the crystalline phase, as evidenced by the maximum ΔA value of −2.4 mOD for AP-TpBpy, in contrast to −6.8 mOD for TpBpy. This observation aligns with the conclusions drawn from KPFM. AP-TpBpy demonstrates a swift rate of H₂O₂ photosynthesis (Fig. S44), achieving levels of 9.9 μmol h⁻¹. Although this is lower than that of crystalline TpBpy (37.9 μmol h⁻¹), it still surpasses the performance of BTBpy and other samples listed in Table S2. This suggests that factors such as crystallinity, specific surface area, pore structure, and hydrophilicity are not the primary reasons for the performance differences observed between the Tp and TB series.

In situ Fourier transform infrared (in-situ FTIR) spectrometry is a valuable tool for divulging the photocatalytic mechanisms. The in-situ FTIR spectra of TpBpy and BTBpy during H₂O₂ photosynthesis under a continuous steam-saturated O₂ flow, and water vapor are depicted in Fig. 5a. Vibrations corresponding to C=C (1342 cm⁻¹), C-H (1378 cm⁻¹), and benzene ring (1441 cm⁻¹) for TpBpy are clearly observed in both dark, and light conditions. Notably, the signal intensities of vibrations related to C-OH (1172, 1402 cm⁻¹), O-H (from C-OH, at 3252 cm⁻¹), C=N (1647 cm−¹)^40,66, PyH⁺ (from 1500 to 1580 cm⁻¹), C=NH⁺ (1615 cm⁻¹), and N-H⁺ (2727 cm⁻¹) increase with prolonged illumination, while the signal intensities of vibrations of C=O (1677 cm⁻¹)⁴² and C-N (1280 cm⁻¹) decrease. This indicates a gradual transformation from the keto-amine structure of TpBpy stabilized in the dark to the enol-imine structure. A similar phenomenon is also observed in AP-TpBpy (refer to the in-situ FTIR spectra of AP-TpBpy in Fig. S45). Additionally, new infrared vibration signals at 982 cm⁻¹ and 1198 cm⁻¹ (Fig. 5a) are attributed to O-O bonding, and *OOH intermediate species, respectively^41,48,67. The vibrational intensities of these signals increase gradually with the duration of irradiation. The variations in the aforementioned vibration peaks indicate keto-amine/enol-imine photoisomerization concurrently facilitates the adsorption of oxygen, and the formation of hydrogen peroxide intermediates.

Fig. 5 In situ FT-IR and charge density analysis of COF catalysts during H₂O₂ photosynthesis. [Images not available. See PDF.]

a In situ FT-IR spectra in water vapor with O₂ flow for TpBpy and b BTBpy, c Charge density difference of TpBpy during H₂O₂ photosynthesis.

In the case of BTBpy (Fig. 5b), vibrations corresponding to C=C (1338 cm⁻¹), C-H (1384 cm⁻¹), and benzene ring (1432 cm⁻¹) can be observed in both dark, and light conditions. Unlike TpBpy, the intensity of these peaks does not change significantly with prolonged illumination. And there are no obvious infrared vibrations of C­OH, C-N, and PyH⁺, C=NH⁺, and N-H⁺, suggesting that there is no keto-amine/enol-imine photoisomerization. Therefore, even though BTBpy can adsorb oxygen (947 cm⁻¹ for O-O vibration) and water molecules (3103 cm⁻¹ for O-H vibration), no obvious infrared vibrations of *OOH were observed. In-situ FTIR data provide insight into the significantly low efficiency of the photocatalytic synthesis of H₂O₂ by BTBpy, which can be attributed to the lack of keto-amine/enol-imine photoisomerization. At end, the in-situ IR spectra of TpBpy were compared under different conditions to investigate the factors affecting the occurrence, and degree of keto-to-enol tautomerism. Condition 1 involved TpBpy in water vapor with O₂ flowing (Fig. 5a). Condition 2 involved TpBpy in water vapor with Ar flowing. Condition 3 involved TpBpy with Ar flowing. It was observed that keto-to-enol tautomerism occurs, but to a lesser extent when there is no O₂ flowing, as revealed by the in-situ FTIR spectra illustrated in Fig. S46a. Additionally, there is a slight enhancement in protonation. However, in the absence of O₂ and water, even after 30 min of photoexcitation, there were minimal keto-to-enol tautomerism or protonation signals (Fig. S46b). These findings suggest that light is a crucial prerequisite for keto-to-enol tautomerism, and the presence of O₂ and water vapor as reactants significantly enhances the degree of keto-to-enol tautomerism⁶⁸.

Considering the structure tautomerism, intermediate species, and the protonation of N, a keto-enol tautomerism-mediated redox process leading to the generation of H₂O₂ can be summarized as depicted in Fig. 5g. To further investigate the impact of keto-enol tautomerism on charge separation, theoretical calculations of electronic configurations for the initial and intermediate structures presented in Fig. 5g were performed. The thermodynamically stable β-ketoenamine (E₀K₃) structure serves as the initial structure, followed by the adsorption of water molecules on the Bpy active sites. The adsorption energy barriers for water molecules at different positions (Fig. S47) disclose that the optimal adsorption site for TpBpy (E₀K₃) is situated between the two nitrogen atoms of the Bpy unit, as illustrated in Fig. S47. Subsequently, the photoisomerization of β-ketoenamine (keto-to-enol tautomerism) results in the formation of E₁K₂ and E₂K₁, utilizing photogenerated holes to produce hydrogen peroxide and protonation of pyridine. Notably, before the photocatalytic water oxidation reaction (WOR), the charge on the Tp unit was −0.751 |e| and −0.570 |e|, whereas after WOR, the formation of E₁K₂ and E₂K₁ through keto-to-enol tautomerism resulted in Tp unit charges of −1.324 |e| and −1.380 |e|, respectively. Correspondingly, prior to WOR, the Bpy unit side charge in the E₀K₃ structure was +0.577 |e| and +0.711 |e|, while after WOR, the positive charge on the Bpy unit side increased to +0.723 |e| and +0.692 |e|. The observed changes in the charge distribution indicate that the enol-free Tp units (E₀K₃) function as electron traps, guiding the flow of electrons towards the Tp side and accumulating them, thereby promoting the flow of holes towards the Bpy active sites, where the adsorbed water was photocatalytically oxidized to generate H₂O₂. Similarly, comparing the charge density difference before and after the photocatalytic oxygen reduction reaction (ORR), the charge amounts of the Tp unit in E₁K₂ and E₂K₁ before ORR are −1.278 |e| and −1.158 |e|, respectively. After the ORR, the enol-to-keto tautomerism forms E₀K₃, with the charge amount on the Tp unit side being −0.751 |e|. Correspondingly, in E₁K₂ and E₂K₁ before ORR, the positive charge amounts on the Bpy unit side are +1.193 |e| and +0.802 |e|, respectively, while in E₀K₃ after ORR, the positive charge amount on the Bpy unit side decreases to +0.577 |e|. It can be observed that the enol-containing Tp units (E₁K₂ and E₂K₁) formed by keto-to-enol tautomerism act as hole traps, guiding and trapping the holes on the Tp side, thereby promoting the flow of electrons towards the Bpy active sites for photocatalytic reduction reactions, where the adsorbed oxygen on Bpy-H⁺ is reduced to H₂O₂. It is worth noting that the raw data of the corresponding model in Fig. 5g is provided in supplementary data 1. The ORR (involving O₂) and the WOR (involving H₂O) are synergistically coupled through keto-enol tautomerism to form a well-established catalytic cycle, thereby significantly accelerating hydrogen peroxide (H₂O₂) photosynthesis. As shown in Fig. S48, H₂O₂ accumulates over time even in the absence of O₂, demonstrating that the WOR pathway alone can produce H₂O₂. However, upon introducing O₂, the H₂O₂ production rate increases dramatically from 3.08 μM min⁻¹ to 33.71 μM min⁻¹ (a nearly 11-fold enhancement). The underlying reason lies in the inability to form a complete reaction cycle (as illustrated in Fig. 5c) in an O₂-free system. This observation is echoed by Fig. S46b, where keto-enol tautomerism occurs but to a much lesser extent under anaerobic conditions. Notably, during oxidation and reduction processes, the charge flow direction in donor-acceptor (D-A) type COFs is consistent and unidirectional, with electrons moving towards the acceptor and holes transferring to the donor. In contrast, in Tp-derived imine COFs, like TpBpy, when in the keto form (E₀K₃), electrons flow towards Tp causing oxidation on Bpy side; while in the enol-containing forms (E₁K₂ and E₂K₁), holes move towards Tp resulting in reduction on Bpy side.

The primary goal of H₂O₂ photosynthesis is its industrial application. Therefore, a separation-free and continuous-flow reaction process is a fundamental requirement (see Fig. 6a). Furthermore, utilizing natural sunlight, which encompasses a broad spectral range, consumes no energy, and demonstrates satisfactory environmental compatibility, is another essential factor (refer to Fig. 6b). To begin with, a flow reactor was utilized to evaluate the performance of H₂O₂ photosynthesis. This reactor was designed to continuously produce products without the necessity of subsequent separation procedures. In this setup, TpBpy served as the photocatalytic filling material (Fig.S49), while a peristaltic pump regulated the flow rate of ultrapure water throughout the system. Initially, a Xenon lamp was employed as the simulated sunlight source in a laboratory environment (Fig. 6c). The average H₂O₂ concentration recorded over a 20-hour period was 172 μM, and the concentration of H₂O₂ produced by TpBpy remained relatively stable (Fig. 6d). The photocatalytic generation rate of H₂O₂ in our custom-built flow reactor reached 1429 mM h⁻¹ m⁻² for TpBpy, considerably higher than the recently reported sunlight-driven synthesis rate in a flow reaction system, such as TAPT–FTPB COFs (376 mM h⁻¹ m⁻²)³. Furthermore, TpBpy demonstrated a notably improved solar-to-chemical conversion (SCC) efficiency of 0.040%, exceeding the performance of TAPT–FTPB COFs at 0.010%³. Subsequently, to achieve the ultimate goal of H₂O₂ photosynthesis, natural sunlight was used to as the light source in the open-air environment (Fig. 6e). The H₂O₂ concentration produced by photosynthesis and the corresponding solar-to-chemical conversion (SCC) efficiency were found to be positively and negatively correlated with natural sunlight intensity, respectively. The average photocatalytic generation rate of H₂O₂ reached 1030 mM h⁻¹ m⁻², and an SCC efficiency reached 0.038% (Fig. 6f). This consistent stability over the course of the 3-day outdoor experiment and 20-hour indoor experiment for H₂O₂ production demonstrates the remarkable cyclic stability of TpBpy.

Fig. 6 Photocatalytic H₂O₂ synthesis in reactor systems under simulated and natural sunlight. [Images not available. See PDF.]

a Development history of photocatalytic synthesis of H₂O₂ for industrial applications in reactors, b Comparison between simulated sunlight and natural sunlight, c Laboratory-scale flow-reactor system for H₂O₂ photocatalysis under simulated sunlight, d Flow-method H₂O₂ production and solar-to-chemical (SCC) efficiency under simulated sunlight, e Open-air flow-reactor system for H₂O₂ photocatalysis under natural sunlight, f Flow-method H₂O₂ production and SCC efficiency under natural sunlight irradiation.

To date, there are few reports on the use of COFs in flow reactions for the outdoor production of H₂O₂. Consequently, to compare the H₂O₂ photocatalytic efficiency of the separation-free system under natural sunlight, we investigated all the reported immobilization systems (Table S7). As shown in Fig. S50, by immobilizing the TpBpy catalyst onto a glass slide (0.3 m × 0.4 m) as a coating film and subsequently submerging the catalyst-coated glass slide in stagnant ultrapure water (18.25 MΩ cm), the immobilization reactor utilizing TpBpy achieved a H₂O₂ photosynthesis SCC of 0.029%. This exceeds all the recently reported performance of immobilization systems, such as COF-2CN, (0.0075%)⁶⁹, PI-BD-TPB (0.024%)^70,71, and COF-N32 (0.019%)⁴⁰. Additionally, the SCC of dispersion systems under natural sunlight was also compared. TpBpy showed superior SCC (Table S7) among all reported photocatalytic materials under similar conditions (Figs. S51,S52). Therefore, whether in the flow phase, immobilization phase, or dispersion phase, TpBpy exhibited superior SCC performance compared to other photocatalysts under natural sunlight.

It is evident that the SCC of all photocatalysts for H₂O₂ production under natural sunlight remains far lower than the 0.10% associated with natural photosynthesis. Therefore, improving photocatalytic performance remains the foremost challenge for the industrial application of photocatalytic H₂O₂ synthesis. Although recent reports indicate that SCC values have exceeded 1.00% under laboratory conditions, these high SCC values often result from continuous optimization of various experimental parameters. Some studies even fail to provide complete experimental details and data, leading to a misleading assessment of the maturity of photocatalytic H₂O₂ synthesis technology. When utilizing fixed-bed reactors that mimic the structure of plant leaves and employing natural sunlight as the light source, the SCC for photocatalytic H₂O₂ synthesis holds greater scientific significance. Therefore, we advocate for the use of SCC values obtained under natural sunlight in immobilization or flowing phase reactors as a comparative metric for evaluating the performance of photocatalytic H₂O₂ synthesis.

Methods

Materials

2,2′-bipyridine-5,5′-diamine (Bpy, >97%), 2,4,6-trihydroxybenzaldehyde-1,3,5-tricarbaldehyde (Tp, >97%), 4,4′-diamino-2,2′-difluo-biphenyl (2F, >98%), 2,2′-dimethyl-[1,1′-biphenyl]-4,4′-diamine (CH₃, >97%), 2,4,6-tris(4-aminophenyl)-1,3,5-triazine (TAPT, >98%), 4′,4′′′,4′′′′′-(1,3,5-triazine-2,4,6-triyl)tris(([1,1′-biphenyl]-4-amine)) (TBPT, >97%) and 5,5′-bis(cyanomethyl)-2,2′-bipyridine (CN, >97%) were purchased from Shanghai Tensus Biotech Co., Ltd. 1,3,5-benzenetricarboxaldehyde (BT, >99.76%), 4,4′-diamino-p-terphenyl (Dpt, >98.86%) and 2,5-diaminobenzoic acid (COOH, >98%) were purchased from Jilin Yanshen Technology Co., Ltd. Solvents including dichloromethane (DCM, AR), o-dichlorobenzene (o-DCB, 99%), acetic acid (AcOH, ≥99.8%), N, N-dimethylacetamide (DMAC, 99.8%, extra dry, with molecular sieves, water ≤50 ppm), dioxane (spectral pure, ≥99.5%) and p-phenylenediamine (AR, 97%) were purchased from Aladdin. Tetrahydrofuran (AR, 99.0%) and 1,3,5-trimethylbenze (AR, 97%) were purchased from Macklin. 38% HCl was procured from Chengdu Kolon Chemical Co. All reagents were used as received without further purification. N, N-Diethyl-p-phenylenediamine (DPD, AR, 98.0%) and horseradish peroxidase (POD, >200 U/mg) were both procured from Aladdin.

Synthesis of covalent organic frameworks (COFs)

Synthesis of TpBpy

Tp (0.3 mmol, 63 mg) and Bpy (0.45 mmol, 83 mg) were mixed with 4.5 mL DMAC, 1.5 mL o-DCB, ensuring uniform dispersion through ultrasonication. Subsequently, 0.6 mL of 6 M AcOH aqueous solution was swiftly introduced and sonicated again over 30 min. After degassing with liquid nitrogen via three freeze-thaw cycles, the reaction was conducted at 120 °C for 3 days. The solid was then washed away with a large amount of DMAC, acetone, n-BuOH, tetrahydrofuran and, dichloromethane, and dried under vacuum. The resulting TpBpy COF was collected.

Synthesis of BTBpy

BT (0.5 mmol, 81 mg) and Bpy (0.75 mmol, 139.5 mg) were mixed in 8 mL n-BuOH, ensuring uniform dispersion through ultrasonication. Subsequently, 0.6 mL of AcOH aqueous solution was swiftly introduced and sonicated again over 30 min. After degassing with liquid nitrogen via three freeze-thaw cycles, the reaction was conducted at 120 °C for 3 days. The solid was then washed away with a large amount of DMAC, acetone, n-BuOH, tetrahydrofuran and, dichloromethane, and dried under vacuum. The resulting BTBpy COF was collected.

Synthesis of BTPa

BT (0.3 mmol, 48.6 mg) and Pa (0.45 mmol, 48.6 mg) were mixed in 6 mL n-BuOH, ensuring uniform dispersion through ultrasonication. Subsequently, 0.4 mL of 6 M AcOH aqueous solution was swiftly introduced and sonicated for an additional 30 min. After degassing with liquid nitrogen through three freeze-thaw cycles, the reaction was conducted at 120 °C for 3 days. The solid was then washed thoroughly with a large amount of DMAC, acetone, n-BuOH, tetrahydrofuran, and dichloromethane, and dried under vacuum. The resulting BTPa COF in yellow color was collected.

Synthesis of TpPa

Tp (0.3 mmol, 63 mg) and Pa (0.45 mmol, 48.6 mg) were mixed with 4.5 mL of DMAC and 1.5 mL of o-DCB, ensuring uniform dispersion through ultrasonication. Subsequently, 0.4 mL of 6 M AcOH aqueous solution was swiftly introduced and sonicated for an additional 30 min. After degassing with liquid nitrogen via three freeze-thaw cycles, the reaction was conducted at 120 °C for 3 days. The solid was then washed thoroughly with a large amount of DMAC, acetone, n-BuOH, tetrahydrofuran, and dichloromethane, and dried under vacuum. The resulting TpPa COF in yellow color was collected.

Synthesis of BT-2F

BT (0.5 mmol, 81 mg) and 2 F (0.75 mmol, 165 mg) were mixed in 8 mL n-BuOH, ensuring uniform dispersion through ultrasonication. Subsequently, 0.6 mL of 6 M AcOH aqueous solution was swiftly introduced and sonicated again over 30 min. After degassing with liquid nitrogen via three freeze-thaw cycles, the reaction was conducted at 120 °C for 3 days. The solid was then washed away with a large amount of DMAC, acetone, n-BuOH, tetrahydrofuran, and dichloromethane, and dried under vacuum.The resulting Tp-2F COF in yellow color was collected.

Synthesis of Tp-2F

Tp (0.3 mmol, 63 mg) and 2 F (0.45 mmol, 99 mg) were mixed with 4.5 mL DMAC and 1.5 mL o-DCB, ensuring uniform dispersion through ultrasonication. Subsequently, 0.6 mL of 6 M AcOH aqueous solution was swiftly introduced and sonicated again over 30 min. After degassing with liquid nitrogen via three freeze-thaw cycles, the reaction was conducted at 120 °C for 3 days. The solid was then washed away with a large amount of DMAC, acetone, n-BuOH, tetrahydrofuran, and dichloromethane, and dried under vacuum. The resulting Tp-2F COF was collected.

Synthesis of BT-CH₃

BT (0.6 mmol, 97 mg) and CH₃ (0.9 mmol, 191 mg) were mixed in 8 mL n-BuOH, ensuring uniform dispersion through ultrasonication. Subsequently, 0.6 mL of 6 M AcOH aqueous solution was swiftly introduced and sonicated again over 30 min. After degassing with liquid nitrogen via three freeze-thaw cycles, the reaction was conducted at 120 °C for 3 days. The solid was then washed away with a large amount of DMAC, acetone, n-BuOH, tetrahydrofuran, and dichloromethane, and dried under vacuum. The resulting BT-CH₃ COF was collected.

Synthesis of Tp-CH₃

Tp (0.6 mmol, 126 mg) and CH₃ (0.9 mmol, 191 mg) were mixed in 8 mL n-BuOH, ensuring uniform dispersion through ultrasonication. Subsequently, 0.6 mL of 6 M AcOH aqueous solution was swiftly introduced and sonicated again over 30 min. After degassing with liquid nitrogen via three freeze-thaw cycles, the reaction was conducted at 120 °C for 3 days. The solid was then washed away with a large amount of DMAC, acetone, n-BuOH, tetrahydrofuran, and dichloromethane, and dried under vacuum. The resulting Tp-CH₃ COF was collected.

Synthesis of BTDpt

BT (0.1 mmol, 16.2 mg) and Dpt (0.15 mmol, 39 mg) were mixed in 8 mL n-BuOH, ensuring uniform dispersion through ultrasonication. Subsequently, 0.6 mL of 6 M AcOH aqueous solution was swiftly introduced and sonicated again over 30 min. After degassing with liquid nitrogen via three freeze-thaw cycles, the reaction was conducted at 120 °C for 3 days. The solid was then washed away with a large amount of DMAC, acetone, n-BuOH, tetrahydrofuran, and dichloromethane, and dried under vacuum. The resulting BTDpt COF was collected.

Synthesis of TpDpt

Tp (0.1 mmol, 21 mg) and Dpt (0.15 mmol, 39 mg), were mixed in 6 mL n-BuOH, ensuring uniform dispersion through ultrasonication. Subsequently, 0.6 mL of 6 M AcOH aqueous solution was swiftly introduced and sonicated again over 30 min. After degassing with liquid nitrogen via three freeze-thaw cycles, the reaction was conducted at 120 °C for 3 days. The solid was then washed away with a large amount of DMAC, acetone, n-BuOH, tetrahydrofuran and, dichloromethane, and dried under vacuum. The resulting TpDpt COF was collected.

Synthesis of BT-COOH

BT (0.3 mmol, 48.6 mg) and COOH (0.45 mmol, 68 mg) were mixed with 4 mL 1,4-dioxane and 4 mL m-xylene, ensuring uniform dispersion through ultrasonication. Subsequently, 0.6 mL of 6 M AcOH aqueous solution was swiftly introduced and sonicated again over 30 min. After degassing with liquid nitrogen via three freeze-thaw cycles, the reaction was conducted at 120 °C for 3 days. The solid was then washed away with a large amount of DMAC, acetone, n-BuOH, tetrahydrofuran and dichloromethane, and dried under vacuum. The resulting BT-COOH COF was collected.

Synthesis of Tp-COOH

Tp (0.3 mmol, 63 mg) and COOH (0.45 mmol, 68 mg) were mixed with 4 mL 1,4-dioxane and 4 mL m-xylene, ensuring uniform dispersion through ultrasonication. Subsequently, 0.6 mL of 6 M AcOH aqueous solution was swiftly introduced and sonicated again over 30 min. After degassing with liquid nitrogen via three freeze-thaw cycles, the reaction was conducted at 120 °C for 3 days. The solid was then washed away with a large amount of DMAC, acetone, n-BuOH, tetrahydrofuran and dichloromethane, and dried under vacuum. The resulting Tp-COOH COF was collected.

Synthesis of BTTAPT

BT (0.1 mmol, 16.2 mg) and TAPT (0.1 mmol, 35.4 mg) were mixed with 1.5 mL 1,4-dioxane and 1.5 mL tetrahydrofuran, ensuring uniform dispersion through ultrasonication. Subsequently, 0.3 mL of 6 M AcOH aqueous solution was swiftly introduced and sonicated again over 30 min. After degassing with liquid nitrogen via three freeze-thaw cycles, the reaction was conducted at 120 °C for 3 days. The solid was then washed away with a large amount of DMAC, acetone, n-BuOH, tetrahydrofuran and, dichloromethane, and dried under vacuum. The resulting BTTAPT COF was collected.

Synthesis of TpTAPT

Tp (0.15 mmol, 31.5 mg) and TAPT (0.15 mmol, 53.2 mg) were mixed in 3 mL 1,4-dioxane, ensuring uniform dispersion through ultrasonication. Subsequently, 0.3 mL of 6 M AcOH aqueous solution was swiftly introduced and sonicated again over 30 min. After degassing with liquid nitrogen via three freeze-thaw cycles, the reaction was conducted at 120 °C for 3 days. The solid was then washed away with a large amount of DMAC, acetone, n-BuOH, tetrahydrofuran and dichloromethane, and dried under vacuum. The resulting TpTAPT COF was collected.

Synthesis of BTTBPT

BT (0.1 mmol, 16.2 mg) and TBPT (0.1 mmol, 58.3 mg) were mixed with 1.5 mL 1,4-dioxane and 1.5 mL tetrahydrofuran, ensuring uniform dispersion through ultrasonication. Subsequently, 0.3 mL of 6 M AcOH aqueous solution was swiftly introduced and sonicated again over 30 min. After degassing with liquid nitrogen via three freeze-thaw cycles, the reaction was conducted at 120 °C for 3 days. The solid was then washed away with a large amount of DMAC, acetone, n-BuOH, tetrahydrofuran and dichloromethane, and dried under vacuum. The resulting BTTBPT COF was collected.

Synthesis of TpTBPT

Tp (0.1 mmol, 21.0 mg) and TBPT (0.1 mmol, 58.3 mg) were mixed with 1.5 mL 1,4-dioxane and 1.5 mL tetrahydrofuran, ensuring uniform dispersion through ultrasonication. Subsequently, 0.3 mL of 6 M AcOH aqueous solution was swiftly introduced and sonicated again over 30 min. After degassing with liquid nitrogen via three freeze-thaw cycles, the reaction was conducted at 120 °C for 3 days. The solid was then washed away with a large amount of DMAC, acetone, n-BuOH, tetrahydrofuran and dichloromethane, and dried under vacuum. The resulting TpTBPT COF was collected.

Synthesis of OMe-Bpy

OMe (0.3 mmol, 75 mg) and Bpy (0.45 mmol, 83 mg) were mixed with 4.5 mL DMAC and 1.5 mL o-DCB, ensuring uniform dispersion through ultrasonication. Subsequently, 0.6 mL of 6 M AcOH aqueous solution was swiftly introduced and sonicated again over 30 min. After degassing with liquid nitrogen via three freeze-thaw cycles, the reaction was conducted at 120 °C for 3 days. The solid was then washed away with a large amount of DMAC, acetone, n-BuOH, tetrahydrofuran and dichloromethane, and dried under vacuum. The resulting OMe-Bpy COF was collected.

Synthesis of CN-SP₂

Tp (0.3 mmol, 63 mg) and CN (0.45 mmol, 105.4 mg) were mixed with 4.5 mL DMAC and 1.5 mL o-DCB, ensuring uniform dispersion through ultrasonication. Subsequently, 0.6 mL of 6 M AcOH aqueous solution was swiftly introduced and sonicated again over 30 min. After degassing with liquid nitrogen via three freeze-thaw cycles, the reaction was conducted at 120 °C for 3 days. The solid was then washed away with a large amount of DMAC, acetone, n-BuOH, tetrahydrofuran and dichloromethane, and dried under vacuum. The resulting CN-SP₂ was collected.

Synthesis of S-COF-TpBpy

The synthesized COF-TpBpy was placed in octathiocyclohexane (catalyst to octathione ratio of 1:10) and calcined in a muffle furnace for 16 h to obtain black solid powder. Following that, the vulcanized powder was then extracted using a tetrahydrofuran Soxhlet extractor for 12 h and ground to obtain brown-red S-COF-TpBpy.

Synthesis of AP-TpBpy

A total of 0.3 mmol (63 mg) of Tp and 0.45 mmol (83 mg) of Bpy were placed into a 10 mL glass ampoule. Subsequently, dioxane solvent was introduced, and the mixture was sonicated to achieve uniform dispersion. Following this, 0.6 mL of a 6 M AcOH aqueous solution was rapidly added, and sonication was performed once more. After sealing the ampoule, the reaction was conducted at 120 °C for a duration of 3 days. The resulting solid was then thoroughly washed with a substantial volume of DMAC, acetone, and dichloromethane, and subsequently dried under vacuum.

Photocatalytic H₂O₂ generation and other reactions

Simulated light (dispersion)

The catalyst with different masses (1, 2, 5, 10, or 20 mg) and 10 mL of water were placed in a sealed apparatus consisting of quartz tubes and sealing elements. The suspension was well dispersed by sonication for 15 min. Prior to the photocatalytic experiment, the suspension was stirred, and dry air was bubbled into the suspension for 30 min to reach adsorption–desorption equilibrium. A 300 W Xe lamp (PLS-SXE300D, Beijing Perfectlight) was used as the light source, and the concentration of H₂O₂ was measured with a UV spectrophotometer. For analysis, 1 mL of liquid was sampled with a 0.22 μm filter to remove the photocatalyst. The samples were mixed and reacted with pre-prepared phosphate buffer solution, POD solution and DPD solution, and the concentration of H₂O₂ was determined by UV–vis spectrophotometer.

Natural sunlight (dispersion)

To facilitate comparisons with dispersion methods reported in other studies, we also assessed the SCC values by using TpBpy dispersion solution. The experiment utilized a 50 mL beaker (diameter = 4 cm), 45 mg of catalyst, and 30 mL of deionized water. To ensure precise area measurements, we opted to cover the sides of the beaker with tin foil, thereby minimizing the impact of external experimental variables (Fig. S51). Additionally, to prevent the settling of the catalyst, a magnetic stirrer was incorporated into the setup. The SCC test was performed on September 18, 2024, in the Student Building of Three Gorges University, from 14:53 to 16:53, during which the light intensity was recorded at 52.7 mW cm⁻², based on the average of six measurements taken over the two-hour period.

Natural sunlight (immobilization reactor)

In order to compare with the flow reaction, an immobilization reactor was utilized for the photocatalytic generation of H₂O₂ under natural sunlight. The immobilization reactor, measuring 50 × 30 × 5 cm, featured TpBpy applied via drop-coating onto a glass slide with an area of 1148 cm², with a mass loading of ~0.8 mg cm⁻², using 5 L of pure water (illustrated in Fig. S49,S50). The H₂O₂ generation was conducted on September 23, 2024 from 09:30 to 11:30, with Longitude of 111.311539 and latitude of 30.714216. The immobilization reactor utilizing TpBpy achieved a H₂O₂ photosynthesis rate of 207.12 μM h⁻¹ and an SCC of 0.0288 % under an average light intensity of 50.83 mW cm⁻².

Simulated light (flow reactor)

To ensure the consistent and stable production of H₂O₂ using a flow reactor, we developed a reaction apparatus in which the brown-red TpBpy powder is filled into the channels, featuring a surface area of 17 cm², as depicted in Fig. S48. A 300 W xenon lamp (100 mW cm⁻²) was employed to provide continuous illumination for a duration of 20 h, during which we conducted real-time monitoring of both light intensity and H₂O₂ concentration at every moment. This experiment further demonstrated that the flow reactor is capable of producing H₂O₂ in a stable and continuous manner.

Natural sunlight (flow reactor)

Photocatalytic generation of H₂O₂ using natural sunlight and TpBpy was performed on a partially sunny day, specifically from 10:00 to 17:00 h, on August 8, 9, and 13, 2024. This experiment took place on the fifth floor of the Mechanical building at Three Gorges University, located at a longitude of 111.311539 and a latitude of 30.714216. The light intensity and the H₂O₂ concentration of the mobile phase flow reactor at each time of the three days were sampled and measured. The flow rate of the peristaltic pump is controlled at 2 mL min⁻¹, and the irradiated area of the flow reactor is 17 cm².

Photocatalytic H₂ evolution

Photocatalytic hydrogen evolution was carried out in a closed photocatalytic glass system with a reactor volume of 150 mL and a circulating condensate system. The specific experimental procedure is as follows: 20 mg of photocatalyst powder (BTBpy or TpBpy) and 0.1 M ascorbic acid as the sacrificial agent were mixed into a solution consisting of 0.3% Pt colloidal solution (co-catalyst) and 40 mL of deionized water. The reactor was then placed in an ultrasonic water bath until all the catalyst powder was uniformly dispersed. After sonication, the reactor was connected to the closed photocatalytic glass system. The circulating condensate was activated at 5.5 °C ( ± 0.1) and the vacuumization process was initiated. Vacuumization was stopped once the system pressure remained stable and unchanged for a duration of 30 min. A suitable cut-off filter (to allow the light wavelength ≥420 nm to pass through) was attached to a 300 W Xe lamp. After light illumination, the gas produced was analyzed by a gas chromatography equipped with a thermal conductivity detector (TCD) with argon (Ar) as the carrier gas.

Photocatalytic CO₂ reduction

First, a stock solution was prepared by dispersing 10 mg of finely powdered catalyst (BTBpy or TpBpy) in 40 mL of deionized water, followed by sonication for 10 min. Next, highly pure CO₂ gas (99.99%) was purged into the reaction mixture for 30 min. The setup was then sealed, leaving 24.5 mL of void space. After sonication, the reactor was connected to a closed photocatalytic glass system. The circulating condensate was activated at 5.5 °C ( ± 0.1), and vacuumization was performed until the system pressure remained stable and unchanged for 30 min. A suitable cut-off filter (to allow the light wavelength ≥420 nm to pass through) was attached to a 300 W Xe lamp. After light illumination, the gas produced was analyzed by a gas chromatography equipped with a thermal conductivity detector (TCD) with argon (Ar) as the carrier gas. The main gases targeted for detection were CO and CH₄.

Photocatalytic aniline oxidation conversion

5 mg of catalysts (BTBpy or TpBpy) was added to 2 mL of acetonitrile solvent. Then, 0.5 mmol of benzyl alcohol was added. The reaction was conducted under a blue LED irradiation for 12 h. Subsequently, the plate was exposed to observe the reaction progress and the conversion rate of benzyl alcohol to benzaldehyde was determined.

Material characterization

X-ray powder diffraction (XRD) analysis was conducted using a Rigaku Ultima IV diffractometer equipped with an X-ray generator with a power of 3 KW (Cu-Kα radiation). The microstructures and morphologies of the catalysts were examined through transmission electron microscopy (TEM, FEI JEOL-2100F). The compositions and element states of the catalysts were determined by X-ray photoelectron spectroscopy (XPS, Escalab 250Xi X-ray, Thermo Scientific). UV–vis absorption spectra were acquired using a UH4150 (HITACHI) to analyze diffuse reflectance spectroscopy over the 300–1200 nm range, while time-resolved PL spectra were obtained with an FLS980 multifunctional steady-state and transient fluorescence spectrometer. The surface potential was measured in the surface potential mode using an AFM (Bruker Dimension Icon). For the fs-TA spectroscopy measurement, fs-TA was determined through the combined use of a Helios Pump-Probe System (Ultrafast Systems LLC) and a Femtosecond Laser System (Coherent). The BEL SORP-MAX volume adsorption analyzer and analyzer (Novatouch, Quantachrome) were used to carry out N₂ (77 K) adsorption-desorption measurement and BET analysis. The solid phase ¹³C nuclear magnetic resonance (NMR) spectra were obtained on an JEOL JNM-ECZL 600G solid state NMR spectrometer. Contact angles were acquired using a fully automatic contact angle tester (Theta Lite Biolin, Finland). Fourier transform infrared (FTIR) spectra of the samples were acquired using a Nicolet iS-50 instrument. In situ Fourier transform infrared spectroscopy (in-situ FTIR) was also performed on a Nicolet iS-50 instrument. The sample was placed into an in-situ FTIR cell, and O₂, H₂O or Ar vapors were introduced into the cell through the CaF₂ window using a fiber source (FX300, Beijing Perfect Light Technology Co., Ltd., Beijing, China). Before measurements, the samples were degassed at 423 K for 4 h. The baseline was obtained before the sample reached O₂ or Ar adsorption equilibrium within 1 h. The electrochemical impedance and photocurrent response of the catalysts were evaluated using an electrochemical workstation (CHI660D, CHI Instruments, Shanghai, China).

Detection of H₂O₂ concentration

Different concentrations of H₂O₂ solutions were prepared using a 30% H₂O₂ standard solution, and the absorbance was measured with a UV–Vis spectrophotometer to obtain the concentration–absorbance standard curve. In this experiment, 11.935 g of KH₂PO₄ and 2.876 g of K₂HPO₄·3H₂O were dissolved in 200 mL of pure water to prepare a phosphate buffer solution, which was used to adjust the pH of the samples. Then, the DPD and POD stock solutions were prepared (0.1 g of DPD was dissolved in 10 mL of 0.05 M H₂SO₄ solution, and 10 mg of POD was dissolved in 10 mL of pure water). During the experiment, 2.5 mL of the sample solution was added to a quartz tube, and 0.4 mL of phosphate buffer solution, 50 μL of POD solution, and 50 μL of DPD solution were added in sequence and mixed well. The absorbance at 552 nm was measured with a UV-2600 (Shanghai Tianmei Scientific Instruments Co., Ltd.).

Photoeletrochemical measurements and Calculations of SCCs and AQYs

To prepare working electrodes, 10 mg fully ground sample and 1 mL ethanol were mixed under sonication for 30 min to completely disperse the sample. The resulting slurry was dropped onto a piece of fluorinated tin (FTO) glass substrate, covering an area of 1 cm². Electrochemical impedance spectroscopy (EIS) measurements, photocurrent responses and photoelectrochemical water oxidation reactions were performed using a CHI660E electrochemical workstation (CHI Instruments, Shanghai, China). FTO glass coated with the prepared samples, platinum wires and Ag/AgCl electrodes were used as working, counter and reference electrodes, respectively. A 300 W xenon lamp (λ > 400 nm) served as the light source and Na₂SO₄ (0.5 M, pH = 6.5) aqueous solution was used as the electrolyte throughout the measurements. The working electrode was illuminated by a 300 W Xe lamp (λ > 400 nm) (PLS-SXE300D, Beijing Perfectlight). The potential was calculated using the formula: E(vs. RHE) = E(vs. Ag/AgCl) + 0.197 V + 0.0591*pH. The equation for solar conversion efficiencies (SCCs): $$\mathrm{SCC}\mathrm{efficiency}\left(\%\right)=\frac{\left[△\mathrm{for}{\mathrm{H}}_2{\mathrm{O}}_2\mathrm{generation}\left(\mathrm{J}{\mathrm{mol}}^{-1}\right)\right]\times \left[{\mathrm{H}}_2{\mathrm{O}}_2\mathrm{formed}\left(\mathrm{mol}\right)\right]\times \left[{\mathrm{H}}_2{\mathrm{O}}_2\mathrm{formed}\left(\mathrm{mol}\right)\right]}{\left[\mathrm{Total}\mathrm{input}\mathrm{power}\left(\mathrm{W}\right)\right]\left[\mathrm{Reaction}\mathrm{time}\left(\mathrm{s}\right)\right]}\times 100\%$$

Where the free energy for H₂O₂ formation is 117 kJ mol⁻¹. For measuring apparent quantum yields (AQYs), the photocatalytic reaction was carried out in pure deionized water (400 ml) with photocatalyst (600 mg) in a foil reflective light-concentrating reactor. After ultrasonication and Air bubbling, the bottle was irradiated by an Xe lamp (light intensity 100 mW cm⁻²). The optical power was determined by a PL-MW 2000 photoradiometer (Beijing Perfect Light Technology Co., Ltd., Beijing, China). The incident light was monochromated by band-pass glass filters. $$\mathrm{AQY}=\frac{{2N}_{H_2{O}_2}}{N_P}=\frac{2\times \mathrm{the}\mathrm{number}\mathrm{of}\mathrm{evolved}{\mathrm{H}}_2{\mathrm{O}}_2\mathrm{molecules}}{\mathrm{the}\mathrm{number}\mathrm{of}\mathrm{incident}\mathrm{photons}}\times 100\%$$

Theoretical calculations

All the density functional theory (DFT) calculations were performed using the Vienna Ab initio Simulation Package. The projector augmented wave (PAW) method and plane-wave basis functions with a cutoff energy of 520 eV were utilized to represent the core and valence electrons. These calculations were based on the Perdew-Burke-Ernzerhof (PBE) approximation and the generalized gradient approximation (GGA) method. To consider the long-range van der Waals (vdWs) interactions, Grimme’s DFT-D3 correction scheme was used. The convergence level set such that the forces on each atom were less than 0.05 eV Å⁻¹. A vacuum layer of 15 Å was introduced along the z-axis to eliminate interactions between periodic images in the TpBpy model. The K-point mesh in the Brillouin zone was sampled by 1 × 1 × 1. The Bader charge analysis was using the Bader code by Henkelman group.

Acknowledgements

This work is supported by the National Natural Science Foundation of China (Grant No. 22476109 to L.Y., 22136003 and 22476110 to Y.H.), the Hubei Provincial Natural Science Foundation of China (2022CFA065 to L.Y.), and the 111 Project (D20015 to N.H.). X.Y.K. acknowledges the support from the Lee Kuan Yew Postdoctoral Fellowship with start-up grant (024042-00001). T.M. acknowledged the Australian Research Council (ARC) through Future Fellowship (FT210100298), Discovery Project (DP220100603), Linkage Project (LP210200504, LP220100088, LP230200897) and Industrial Transformation Research Hub (IH240100009) schemes, the Australian Government through the Cooperative Research Centres Projects (CRCPXIII000077), the Australian Renewable Energy Agency (ARENA) as part of ARENA’s Transformative Research Accelerating Commercialization Program (TM021), and European Commission’s Australia-Spain Network for Innovation and Research Excellence (AuSpire).

Author contributions

F.M. and T.G. performed and analyzed the majority of experiments; N.H., L.Y., T.M., X.S. and X.-Y.K. wrote the manuscript draft; G.L. and A.J. assisted in the Femtosecond transient absorption spectroscopy experiment; C.H., Y.W., Y.Z., H.W., L.W., B.J., Y.H., H.H. and H.L. participated in the data analysis and results discussion; L.Y., N.H. and T.M. surpervised the studies and revised the manuscript.

Peer review

Peer review information

Nature Communications thanks Yusuke Yamauchi 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 authors declare all data supporting the findings of this study are available within the article and supplementary Information file. Source data are provided with this paper (https://doi.org/10.6084/m9.figshare.29422913).

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41467-025-62286-9.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

1. Côté, AP et al. Porous, crystalline, covalent organic frameworks. Science; 2005; 310, pp. 1166-1170.2005Sci..310.1166C [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/16293756]

2. Banerjee, T; Podjaski, F; Kröger, J; Biswal, BP; Lotsch, BV. Polymer photocatalysts for solar-to-chemical energy conversion. Nat. Rev. Mater.; 2021; 6, pp. 168-190.2021NatRM..6.168B

3. Liu, Y. et al. Enhanced hydrogen peroxide photosynthesis in covalent organic frameworks through induced asymmetric electron distribution. Nat. Synth. https://doi.org/10.1038/s44160-024-00644-z (2024)

4. Xie, Z et al. Well-defined diatomic catalysis for photosynthesis of C₂H₄ from CO₂. Nat. Commun.; 2024; 15, 2024NatCo.15.2422X [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/38499562][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10948895]2422.

5. Xia, Y et al. A triple tandem reaction for the upcycling of products from poorly selective CO₂ photoreduction systems. Nat. Synth.; 2024; 3, pp. 406-418.2024NatSy..3.406X

6. Zhang, Q et al. Designing covalent organic frameworks with Co-O₄ atomic sites for efficient CO₂ photoreduction. Nat. Commun.; 2023; 14, 2023NatCo.14.1147Z [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/36854683][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9975230]1147.

7. Liu, L et al. Linear conjugated polymers for solar-driven hydrogen peroxide production: the importance of catalyst stability. J. Am. Chem. Soc.; 2021; 143, pp. 19287-19293. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/34757722][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8630703]

8. Teng, Z et al. Atomically dispersed antimony on carbon nitride for the artificial photosynthesis of hydrogen peroxide. Nat. Catal.; 2021; 4, pp. 374-384.

9. Krishnaraj, C et al. Strongly reducing (diarylamino) benzene-based covalent organic framework for metal-free visible light photocatalytic H₂O₂ generation. J. Am. Chem. Soc.; 2020; 142, pp. 20107-20116. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/33185433][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7705891]

10. Xu, X et al. Pore partition in two-dimensional covalent organic frameworks. Nat. Commun.; 2023; 14, 2023NatCo.14.3360X [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/37291160][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10250421]3360.

11. Lv, M et al. Precisely engineering asymmetric atomic CoN₄ by electron donating and extracting for oxygen reduction reaction. Angew. Chem. Int. Ed.; 2024; 63, e202315802.

12. O’Leary, LER; Fallas, JA; Bakota, EL; Kang, MK; Hartgerink, JD. Multi-hierarchical self-assembly of a collagen mimetic peptide from triple helix to nanofibre and hydrogel. Nat. Chem.; 2011; 3, pp. 821-828. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/21941256]

13. Vitak, E et al. Phenazine-based covalent organic framework cathode materials with high energy and power densities. J. Am. Chem. Soc.; 2020; 142, pp. 16-20.

14. Haase, F; Lotsch, BV. Solving the COF trilemma: towards crystalline, stable and functional covalent organic frameworks. Chem. Soc. Rev.; 2020; 49, pp. 8469-8500. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/33155009]

15. Chen, Y et al. Hierarchical assembly of donor–acceptor covalent organic frameworks for photosynthesis of hydrogen peroxide from water and air. Nat. Synth.; 2024; 3, pp. 998-1010.

16. Yang, Y et al. Engineering β-ketoamine covalent organic frameworks for photocatalytic overall water splitting. Nat. Commun.; 2023; 14, 2023NatCo.14.593Y [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/36737616][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9898260]593.

17. Li, Z et al. Three-component donor-π-acceptor covalent-organic frameworks for boosting photocatalytic hhydrogen evolution. J. Am. Chem. Soc.; 2023; 145, pp. 8364-8374.

18. Shu, C et al. Mixed-linker strategy for the construction of sulfone-containing D-A-A covalent organic frameworks for efficient photocatalytic hydrogen peroxide production. Angew. Chem. Int. Ed.; 2024; 63, e202403926.

19. Li, J et al. Covalent organic frameworks: reversible 3D coalesce via interlocked skeleton-pore actions and impacts on πelectronic structures. J. Am. Chem. Soc.; 2023; 145, pp. 26383-26392. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/37983008]

20. Wang, H et al. Integrating suitable linkage of covalent organic frameworks into covalently bridged inorganic/organic hybrids toward efficient photocatalysis. J. Am. Chem. Soc.; 2020; 142, pp. 4862-4871. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/32073853]

21. Liao, J et al. Flexible units induced three-dimensional covalent organic frameworks with a heteromotif molecular junction for photocatalytic H₂O₂ production. ACS Catal; 2024; 14, pp. 3778-3787.

22. Wang, J et al. Photocatalytic Z/E isomerization unlocking the stereodivergent construction of axially chiral alkene frameworks. Nat. Commun.; 2024; 15, 2024NatCo.15.3254W [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/38627395][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11021481]3254.

23. Zhang, X et al. Enhancing photocatalytic H₂O₂ production with Au co-catalysts through electronic structure modification. Nat. Commun.; 2024; 15, 2024NatCo.15.3212Z [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/38615063][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11016070]3212.

24. Tan, R et al. Liberating C-H bond activation: achieving 56% quantum efficiency in photocatalytic cyclohexane dehydrogenation. J. Am. Chem. Soc.; 2024; 146, pp. 14149-14156. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/38717984]

25. Jia, G et al. Charge redistribution of a spatially differentiated ferroelectric Bi₄Ti₃O₁₂ single crystal for photocatalytic overall water splitting. Nat. Commun.; 2024; 15, 2024NatCo.15.4746J [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/38834546][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11150255]4746.

26. Lin, X et al. Solar overall water-splitting by a spin-hybrid all-organic semiconductor. Nat. Commun.; 2024; 15, 2024NatCo.15.5047L [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/38871750][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11176302]5047.

27. Abe, R; Shinmei, K; Koumura, N; Hara, K; Ohtani, B. Visiblelight-induced water splitting based on two-step photoexcitation between dye-sensitized layered niobate and tungsten oxide photocatalysts in the presence of a triiodide/iodide shuttle redox mediator. J. Am. Chem. Soc.; 2013; 135, pp. 16872-16884. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/24128384]

28. Zhai, G et al. Highly efficient, selective, and stable photocatalytic methane coupling to ethane enabled by lattice oxygen looping. Sci. Adv.; 2024; 10, eado4390. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/38941471][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11637002]

29. Wang, Y et al. Coupling photocatalytic CO₂ reduction and CH₃OH oxidation for selective dimethoxymethane production. Nat. Commun.; 2024; 15, [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/39025876][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11258228]6047.

30. Huang, Y et al. Achieving a solar-to-chemical efficiency of 3.6% in ambient conditions by inhibiting interlayer charges transport. Nat. Commun.; 2024; 15, [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/38926358][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11208529]5406.

31. Zhao, K et al. Peri-Fused polyaromatic molecular contacts for perovskite solar cells. Nature; 2024; 632, pp. 301-306.2024Natur.627.301Z [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/39048825]

32. Wang, Y et al. General negative pressure annealing approach for creating ultra-high-loading single-atom catalyst libraries. Nat. Commun.; 2024; 15, [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/38971885][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11227521]5675.

33. Li, H et al. Layered Pd oxide on PdSn nanowires for boosting direct H₂O₂ synthesis. Nat. Commun.; 2022; 13, 2022NatCo.13.6072L [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/36241626][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9568611]6072.

34. Gu, D et al. Material-structure-performance integrated laser-metal additive manufacturing. Science; 2021; 372, 932.

35. He, T et al. Integrated interfacial design of covalent organic framework photocatalysts to promote hydrogen evolution from water. Nat. Commun.; 2023; 14, 2023NatCo.14.329H [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/36658157][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9852592]329.

36. Sun, R et al. Unprecedented photocatalytic hydrogen peroxide production via covalent triazine frameworks constructed from fused building blocks. Angew. Chem. Int. Ed.; 2024; 63, e202416350.

37. Xu, H et al. Integrating multipolar structures and carboxyl groups in sp²-carbon conjugated covalent organic frameworks for overall photocatalytic hydrogen peroxide production. Angew. Chem. Int. Ed.; 2024; 63, e202408802.

38. Chen, D et al. N-rich 2D heptazine covalent organic frameworks as efficient metal-free photocatalysts. ACS Catal.; 2022; 12, 616623.

39. Pan, G et al. The hydration-initiated pathway of water oxidation over photoexcited covalent organic frameworks. ACS Catal.; 2022; 12, pp. 14911-14917.

40. Liu, F et al. Covalent organic frameworks for direct photo synthesis of hydrogen peroxide from water, air and sunlight. Nat. Commun.; 2023; 14, 2023NatCo.14.4344L [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/37468482][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10356944]4344.

41. Xu, T et al. Constructing photocatalytic covalent organic frameworks with aliphatic linkers. J. Am. Chem. Soc.; 2024; 146, pp. 20107-20115. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/38842422]

42. Feng, S et al. Rational design of covalent organic frameworks with redox-active catechol moieties for high-performance overall photosynthesis of hydrogen peroxide. ACS Catal.; 2024; 14, pp. 7736-7745.

43. Qin, C et al. Dual donor-acceptor covalent organic frameworks for hydrogen peroxide photosynthesis. Nat. Commun.; 2023; 14, 2023NatCo.14.5238Q [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/37640726][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10462664]5238.

44. Zhu, Q; Shi, L; Li, Z; Li, G; Xu, X. Protonation of an imine-linked covalent organic framework for efficient H₂O₂ photosynthesis under visible light up to 700 nm. Angew. Chem. Int. Ed.; 2024; 63, e202408041.

45. Zhang, X et al. Keto-anthraquinone covalent organic framework for H₂O₂ photosynthesis with oxygen and alkaline water. Nat. Commun.; 2024; 15, 2024NatCo.15.2649Z [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/38531862][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11258313]2649.

46. Luo, Y et al. Sulfone-modified covalent organic frameworks enabling efficient photocatalytic hydrogen peroxide generation via one-step two electron O₂ reduction. Angew. Chem. Int. Ed.; 2023; 62, e202305355.

47. Yue, J et al. Regulating the topology of covalent organic frameworks for boosting overall H₂O₂ photogeneration. Angew. Chem. Int. Ed.; 2024; 63, e202405763.

48. Liao, Q et al. Regulating relative nitrogen locations of diazine functionalized covalent organic frameworks for overall H₂O₂ photosynthesis. Angew. Chem. Int. Ed.; 2023; 62, e202310556.

49. Yue, J et al. Regulating the H₂O₂ photosynthetic activity of covalent organic frameworks through linkage orientation. ACS Catal.; 2024; 14, pp. 4728-4737.

50. Das, P et al. Integrating bifunctionality and chemical stability in covalent organic frameworks via one-pot multicomponent reactions for solar-driven H₂O₂ Production. J. Am. Chem. Soc.; 2023; 145, pp. 2975-2984. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/36695541]

51. Shiraishi, Y et al. Resorcinol-formaldehyde resins as metal-free semiconductor photocatalysts for solar-tohydrogen peroxide energy conversion. Nat. Mater.; 2019; 18, pp. 985-993.2019NatMa.18.985S [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/31263224]

52. Zeng, X et al. Simultaneously tuning charge separation and oxygen reduction pathway on graphitic carbon nitride by polyethylenimine for boosted photocatalytic hydrogen peroxide production. ACS Catal.; 2020; 10, pp. 3697-3706.

53. Mi, Z et al. Covalent organic frameworks enabling site isolation of viologen-derived electron-transfer mediators for stable photocatalytic hydrogen evolution. Angew. Chem. Int. Ed.; 2021; 60, pp. 9642-9649.

54. Wu, S et al. The keto-switched photocatalysis of reconstructed covalent organic frameworks for efficient hydrogen evolution. Angew. Chem. Int. Ed.; 2023; 62, e202309026.

55. Chu, F et al. Regulating keto-enol tautomerism of β‑ketoenamine covalent-organic frameworks for photocatalytic oxidative coupling of amines. ACS Catal.; 2023; 13, pp. 13167-13180.

56. Lin, Z; Liu, S; Weng, W; Wang, C; Guo, J. Photostimulated covalent linkage transformation isomerizing covalent organic frameworks for improved photocatalytic performances. Small; 2023; 20, 2307138.

57. Hu, Y et al. Molecular recognition with resolution below 0.2 angstroms through thermoregulatory oscillations in covalent organic frameworks. Science; 2024; 384, pp. 1441-1447. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/38935724]

58. Zhen, C et al. Liquid metal-embraced photoactive films for artificial photosynthesis. Nat. Commun.; 2024; 15, 2024NatCo.15.1672Z [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/38395923][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10891066]1672.

59. Mase, K; Yoneda, M; Yamada, Y; Fukuzumi, S. Seawater usable for production and consumption of hydrogen peroxide as a solar fuel. Nat. Commun.; 2016; 7, 2016NatCo..711470M [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/27142725][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4857479]11470.

60. Pan, Q et al. Ultrafast charge transfer dynamics in 2D covalent organic frameworks/Re-complex hybrid photocatalyst. Nat. Commun.; 2022; 13, 2022NatCo.13.845P [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/35149679][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8837612]845.

61. Qiu, J et al. COF/In₂S₃ S-scheme photocatalyst with enhanced light absorption and H₂O₂-production activity and fs-TA investigation. Adv. Mater.; 2024; 36, 2400288.

62. Yang, J et al. Protonated imine-linked covalent organic frameworks for photocatalytic hydrogen evolution. Angew. Chem. Int. Ed.; 2021; 60, pp. 19797-19803.

63. Tan, KT et al. Water cluster in hydrophobic crystalline porous covalent organic frameworks. Nat. Commun.; 2021; 12, 2021NatCo.12.6747T [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/34799574][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8604923]6747.

64. Wang, Y et al. Molecular engineering of methylated sulfone-based covalent organic frameworks for back-reaction inhibited photocatalytic overall water splitting. Angew. Chem. Int. Ed.; 2024; 63, e202408697.

65. Wang, C et al. Cage escape governs photoredox reaction rates and quantum yields. Nat. Chem.; 2024; 16, pp. 1151-1159. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/38499849][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11230909]

66. Bunck, DN; Dichtel, WR. Bulk synthesis of exfoliated two-dimensional polymers using hydrazone-linked covalent organic frameworks. J. Am. Chem. Soc.; 2013; 135, pp. 14952-14955. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/24053107]

67. Gottschling, K; Savasci, G; Vignolo-González, H; Schmidt, S; Lotsch, BV. Rational design of covalent cobaloxime-COF hybrids for enhanced photocatalytic hydrogen evolution. J. Am. Chem. Soc.; 2020; 142, pp. 12146-12156. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/32564604][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7366382]

68. Liu, X et al. Tris(triazolo)triazine-based covalent organic frameworks for efficiently photocatalytic hydrogen peroxide production. Angew. Chem. Int. Ed.; 2024; 63, e202411546.

69. Hou, Y et al. Efficient photosynthesis of hydrogen peroxide by cyano-containing covalent organic frameworks from water, air and sunlight. Angew. Chem. Int. Ed.; 2024; 63, e202318562.

70. Chi, W et al. A photocatalytic redox cycle over a polyimide catalyst drives efficient solar-to-H₂O₂ conversion. Nat. Commun.; 2024; 15, [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/38909037][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11535368]5316.

71. Liu, R et al. Linkage-engineered donor-acceptor covalent organic frameworks for optimal photosynthesis of hydrogen peroxide from water and air. Nat. Catal.; 2024; 7, pp. 195-206.