English

• Hydrogen peroxide (H₂O₂) has emerged as an eco-friendly chemical with broad applications spanning water purification, healthcare, and renewable energy systems [1]. Its appeal as an energy carrier stems from a high energy density (3.0 MJ/L), surpassing even compressed hydrogen (2.8 MJ/L at 35 MPa), while offering safer storage and transport [2]. Despite these benefits, conventional H₂O₂ production relies on the energy-intensive anthraquinone oxidation process, which demands costly noble metal catalysts, generates hazardous byproducts, and requires complex separation steps [3,4]. To address these limitations, sustainable alternatives such as electrocatalytic synthesis via the 2e⁻ oxygen reduction reaction (ORR) have gained traction, enabling on-demand H₂O₂ generation under ambient conditions [5,6]. However, practical challenges persist, such as low oxygen solubility in electrolytes, rapid H₂O₂ decomposition on metal-based catalysts, and the competing 4e^- ORR pathway for H₂O formation. These factors collectively limit industrial viability, often restricting operational currents to < 100 mA/cm² and minimal H₂O₂ yields < 0.1 wt% [6,7]. While metal-free carbon catalysts offer a promising alternative, their adoption is hindered by unresolved questions about active site identification and design principles [8,9]. Thus, developing energy-efficient, cost-effective methods for H₂O₂ synthesis remains a pressing priority.

  Artificial photosynthesis presents a compelling route, enabling sustainable H₂O₂ production from water and oxygen under sunlight [10–12]. Current approaches focus on the 2e^- ORR, which proceeds via single-step or two-step mechanisms [13,14]. However, natural water sources contain vastly more water molecules than dissolved oxygen (< 10 mg/L), suggesting that enhancing water oxidation reactions (WOR) could substantially boost photocatalytic H₂O₂ yield [15]. According to previous reports, three primary WOR pathways exist: The 1e^- pathway generates hydroxyl radicals (^•OH) that combine to form H₂O₂; the direct 2e^- pathway produce H₂O₂; and the 4e^- pathway yield oxygen for subsequent reduction to H₂O₂ [16–18]. The 1e^- pathway is particularly attractive, as it not only enables H₂O₂ formation but also supports ^•OH-driven applications like pollutant degradation. Yet, conventional inorganic semiconductors like TiO₂ exhibit poor efficiencies for H₂O₂ generation via WOR, with external quantum efficiencies below 1% for ^•OH-mediated routes [19,20]. These limitations arise from material-reaction mismatches: The strong oxidative potentials of metal oxides (e.g., ~3.0 V vs. NHE) and nitrides (~2.0 V) thermodynamically favor 4e^- or 2e^- pathways [21]. Balancing these thermodynamic factors is therefore critical to optimizing the 1e^- WOR pathway for efficient H₂O₂ production.

  Recently, covalent organic frameworks (COFs) have gained increasing attention for controllable photocatalytic synthesis and decomposition of H₂O₂ due to their light-harvesting capabilities, customizable electronic properties, and abundant catalytic sites [22,23]. However, a fundamental challenge persists: Photogenerated electrons (e^–) and holes (h⁺) in COFs recombine rapidly due to strong electrostatic attraction, leading to energy loss and lower catalytic efficiency [9,24]. To address this, researchers have focused on reducing the energy barrier for charge separation, known as exciton binding energy (E_b), by integrating donor-acceptor (D-A) units into COF backbones [25,26]. The electron push-pull interaction between D and A units promotes exciton dissociation and suppresses recombination, while enabling precise tuning of redox properties in facilitating the WOR and ORR pathway [27,28]. Despite these advances, structural imperfections in COFs, such as torsional strain in linear linkages, introduce large angles between D and A units, disrupting molecular planarity and impeding electron delocalization [29]. Thus, achieving high-performance COF photocatalysts requires not only functional frameworks but also precise alignment of energy levels between D-A components.

  In this study, we present a straightforward and adaptable D-π-A structural extension strategy aimed to reduce E_b value in COFs materials, thereby enhancing the photocatalytic performance for H₂O₂ generation. By examining the impact of various D-π-A units on the coplanarity of the COFs, 2,5-dimethoxy-1,4-benzenedicarboxaldehyde serves as the D unit, while the A units are modulated through interconnection with 4,4′,4′'-(1,3,5-triazine-2,4,6-triyl)trianiline (TAPT–OCH₃), 1,3,5-tri(4-aminophenyl)benzene (TAPB–OCH₃), or tris(4-aminophenyl)amine (TAPA–OCH₃) as the A units, respectively. Among these, TAPT–OCH₃ exhibited significantly higher ORR and WOR activity, achieving a rate of 924 μmol g^-1 h^-1 from oxygen and water without sacrificial agents, with a notable apparent quantum yield (AQY) of 2.21% at 420 nm and a superior solar-to-chemical conversion (SCC) efficiency of 0.23%. For practical application, TAPT–OCH₃ was immobilized onto a continuous-flow reactor to consecutively produce H₂O₂, achieving in-situ disinfection of bacteria and degradation of organic contaminants in various real water samples. The strategic design of D-π-A structures presents a promising avenue for developing efficient low-E_b COF photocatalysts in future research.

  To systematically examine how D and A units govern molecular geometry and electronic properties, a series of COFs with tailored D-A architectures were synthesized using varying monomers (Texts S1-S5 in Supporting information). As depicted in Figs. 1a-c, 2,5-dimethoxy-1,4-dialdehyde served as the D unit, leveraging its electron-donating methoxy groups to enhance π-electron delocalization. Complementary A units, TAPT featured an electron-deficient triazine core; TAPB and TAPA delocalized electron-deficient aromatic systems, were selected to induce controlled molecular torsion, optimizing electronic interactions for improved charge mobility and photocatalytic activity. Electrostatic potential (ESP) mapping (Figs. 1d-f) demonstrated that geometric isomerization amplifies dipole moments, enabling precise electronic redistribution. Dihedral angle analysis revealed structural variations: TAPT-OCH₃ exhibited near-planarity (70.2°), while TAPA-OCH₃ and TAPB-OCH₃ showed angles of 108.2° and 102.5°, respectively (Fig. S1 in Supporting information). The triazine π-bridge in TAPT-OCH₃ enhanced coplanarity, whereas the amine chains in TAPA-OCH₃ introduced rotational flexibility. The spatial separation in TAPT-OCH₃ calculated from the highest occupied and lowest unoccupied molecular orbitals (HOMO and LUMO) confirms the formation of a charge-transfer excited state (Fig. S2 in Supporting information). This observation was further supported by localized orbital locator (LOL-π) analysis, which highlighted extensive π-electron delocalization in TAPA-OCH₃, consistent with strong D-A coplanarity and efficient charge separation (Figs. 1g-i).

  Figure 1

  

  Figure 1.  (a-c) Schematic diagrams. (d-f) Electrostatic potential maps models. (g-i) LOL-π isosurfaces of the obtained COFs.

  DownLoad: Full-Size Img PowerPoint

  The crystallinity and structural parameters of the synthesized COFs were analyzed using powder X-ray diffraction (PXRD) and theoretical simulations [30]. The PXRD patterns of these samples exhibited sharp and similar peaks, indicating the high crystallinity and isostructural nature (Figs. 2a-c). Sharp, well-defined PXRD peaks confirmed the high crystallinity and isostructural nature of the three COFs. Distinct primary diffraction peaks at 2.84°, 2.89°, and 2.95° were observed for TAPT-OCH₃, TAPB-OCH₃, and TAPA-OCH₃, respectively, corresponding to the (100) crystallographic plane [31]. Additional peaks for the (110), (200), and (210) planes appeared at higher angles: 4.83°, 7.47°, and 9.51° for TAPT-OCH₃; 5.63°, 7.43°, and 9.74° for TAPB-OCH₃, while the (200) and (210) planes appeared at higher angles of 6.52° and 10.87° for TAPA-OCH₃ [32]. All samples displayed a broad peak near 25°, attributed to π-π stacking interactions between COF layers along the (001) plane [33]. Furthermore, the Pawley refinement of the PXRD data revealed detailed unit cell parameters. For TAPT–OCH₃, the refined lattice matched a honeycomb structure (R_p = 1.74%, R_wp = 1.35%) with unit cell dimensions of a = 37.72 Å, b = 36.29 Å, c = 3.97 Å, and angles α = β = 90°, γ = 120°, consistent with the idealized AA-stacking configuration [30]. Similar refinements for TAPB–OCH₃ and TAPA-OCH₃ (Table S1 in Supporting information) confirmed analogous crystallographic parameters and atomic coordinates, supporting their syn-AA-stacked architectures. Based on the PXRD patterns and simulated stacking models, the hexagonal pore diameters of TAPT-OCH₃, TAPB-OCH₃, and TAPA-OCH₃ were calculated as 34.65, 34.35, and 30.34 Å, respectively (inset). The interlayer π-π stacking distances, derived from the (001) plane d-spacings, were 3.97 Å (TAPT–OCH₃), 3.75 Å (TAPB–OCH₃), and 4.13 Å (TAPA–OCH₃). These findings demonstrate that the three COFs possess well-defined crystalline structures and excellent crystallinity.

  Figure 2

  

  Figure 2.  (a-c) PXRD patterns with the top and side view of corresponding simulated packing structures. (d-f) SEM images. (g) N₂ sorption isotherms with calculated pore size distributions. (h) The contact angle for the synthesized COFs. (i) TEM image of TAPT–OCH₃.

  DownLoad: Full-Size Img PowerPoint

  The surface morphology of the COFs was further examined using scanning electron microscopy (SEM). TAPT–OCH₃ presented a grass-like morphology consisting of well-bedded nanowires with diameters of 200 nm (Figs. 2d-f), while TAPB–OCH₃ and TAPA–OCH₃ displayed blocky structures composed of aggregated microspheres. The porosity and surface area of the as-synthesized COFs were further analyzed using N₂ adsorption-desorption isotherms [34]. As shown in Fig. 2g, Brunauer-Emmett-Teller (BET) surface area calculations revealed values of 1889.1, 1460.7, 249.1 m²/g of TAPT–OCH₃, TAPB–OCH₃, and TAPA–OCH₃, respectively. Corresponding pore size distributions were 3.01, 2.63, and 1.19 nm, confirming the mesoporous nature of these frameworks for oxygen adsorption and water diffusion. The denser molecular stacking in the latter two frameworks resulted in smoother surfaces and reduced specific surface areas, consistent with the BET surface area measurements. The contact angle measurements for the synthesized COFs were conducted to evaluate their surface hydrophilicity (Fig. 2h). In contrast to TAPB–OCH₃ and TAPA–OCH₃ (> 95°), TAPT–OCH₃ showed a notably smaller contact angle (35.4°), suggesting a slight improvement in water affinity upon incorporating the more polar triazine core into the TAPT–OCH₃ framework. As shown in (Fig. 2i), TAPT–OCH₃ displayed a clear honeycomb-like porous structure of ca. 0.31 nm, in accordance with the in-plane pore channels in the proposed AA-stacked structure. The highly ordered spatial arrangements with well-defined chemical structures in TAPT–OCH₃ are critical for understanding the structure-activity property relationship of porous organic semiconductors.

  The structural integrity of the synthesized COFs was verified through Fourier-transform infrared (FT-IR) spectra and ¹³C solid-state nuclear magnetic resonance (¹³C NMR). As shown in Fig. 3a, the FT-IR spectra exhibited a characteristic imine bond (C=N) vibration peak at 1579 cm^-1, alongside triazine ring-associated signals at 1512, 1360, and 812 cm^-1, confirms the formation of imine linkages via Schiff-base condensation [35]. Additionally, the absence of the C=O stretching band (1035 cm^-1) from the TAP–OCH₃ monomer and the diminished intensity of N—H stretching bands (3360 cm^-1) from TAPT monomers (Fig. S3 in Supporting information) confirm the complete polymerization of TAPT–OCH₃. Complementing these findings, the ¹³C NMR spectra (Fig. 3b) displayed a resonance at 152 ppm, consistent with C=N bond formation, and a peak at 168 ppm corresponding to triazine ring carbons [36]. Elemental composition and bonding configurations were further corroborated by X-ray photoelectron spectroscopy (XPS). The high-resolution N 1s spectrum of TAPT–OCH₃ (Fig. 3c) revealed two peaks at 397.91 and 400.17 eV, assigned to C=N-C and C=N bonds, respectively, Meanwhile, peaks at around 397.75 eV can be assigned to the sp² hybridized -C=N- in triazine ring [37]. Notably, the C 1s binding energy of TAPT–OCH₃ shifted higher (Fig. S4 in Supporting information), while the N 1s energy shifted lower compared to TAPB–OCH₃ and TAPA–OCH₃. These trends align with the anticipated electronic modulation induced by the D-A configuration of C=N bonds, which enhances charge transfer efficiency [38]. The synthesis was highly reproducible, yielding the target COF in an average isolated yield of 85% (±3%) over five separate batches, with uniformity confirmed by identical FT-IR spectra (Fig. S5 in Supporting information). Furthermore, the framework demonstrated excellent stability, retaining its structural integrity after 48 h of exposure to various conditions, as confirmed by post-exposure FT-IR analysis (Fig. S6 in Supporting information). This robustness is a critical attribute for its application in photocatalytic systems.

  Figure 3

  

  Figure 3.  (a) FT-IR spectra, (b) ¹³C NMR spectra, (c) XPS spectra of N 1s, (d) UV–vis DRS spectra, (e) calculated band alignments, (f) the calculated TDOS, (g) temperature-dependent PL spectral, (h) steady-state PL spectra, (i) time-resolved PL spectra of synthesized COFs.

  DownLoad: Full-Size Img PowerPoint

  The photocatalytic performance of the synthesized COFs depends critically on their light absorption capabilities and electronic band structure [39]. As shown in Fig. 3d, all COFs displayed strong visible-light absorption, with optical bandgaps of 2.23 eV for TAPA–OCH₃, 2.33 eV for TAPB–OCH₃, and 2.37 eV for TAPT–OCH₃. Critically, reported data indicate that differences in visible-light absorption alone cannot account for the significant variations in H₂O₂ production efficiency observed among these materials [40]. Mott-Schottky measurements (Fig. 3e and Fig. S7 in Supporting information) revealed that the conduction band potentials of all COFs were sufficiently negative to drive oxygen reduction to H₂O₂, while their valence band potentials were sufficiently positive to enable water oxidation, thermodynamically enabling photocatalytic H₂O₂ generation. Considering partial density of states (PDOS), further analysis revealed that TAPT–OCH₃ exhibits smaller bandgaps near the Fermi level compared to TAPA–OCH₃ and TAPB–OCH₃ (Fig. 3f), suggesting improved electron mobility and conductivity. To evaluate charge transfer dynamics, electrochemical impedance spectroscopy and photocurrent response measurements were performed. TAPT–OCH₃ demonstrated the smallest semicircle radius in Nyquist plots (Fig. S8 in Supporting information) and the highest photocurrent density (Fig. S9 in Supporting information), reflecting its superior charge separation efficiency and reduced interfacial resistance.

  Temperature-dependent photoluminescence (TD-PL) studies provided insights into exciton behavior: the full-width at half-maximum of PL spectra broadened (Fig. 3g) as temperatures decreased from 300 K to 240 K. Using the Arrhenius equation, the E_b value of TAPT–OCH₃ was calculated as 45.2 meV, significantly lower than those of TAPB–OCH₃ (64.2 meV) and TAPA–OCH₃ (82.2 meV). This lower E_b suggests that TAPT–OCH₃ facilitates easier exciton dissociation into free charges, likely due to its rotation-inhibited π-conjugated structure. Steady-state PL spectra (Fig. 3h) showed weaker emission intensity for TAPT–OCH₃, implying reduced electron-hole recombination, a feature attributed to its D-π-A configuration. Time-resolved PL decay analysis (Fig. 3i) further revealed prolonged charge carrier lifetimes in TAPT–OCH₃, allowing greater participation of photogenerated charges in catalytic reactions. Collectively, these findings highlight optimized photophysical and electrochemical properties of TAPT–OCH₃, which minimize energy loss and electron recombination, ultimately improving photocatalytic efficiency.

  The photocatalytic performance for H₂O₂ generation was evaluated in pure water without sacrificial agent. As shown in Fig. 4a, TAPT–OCH₃ demonstrated a steady increase in H₂O₂ accumulation upon visible-light irradiation (λ > 420 nm), reaching a production rate of 924.1 μmol g^-1 h^-1. This performance surpassed that of TAPB–OCH₃ by 2.6-fold (355.5 μmol g^-1 h^-1) and TAPA–OCH₃ by 14.1-fold (65.24 μmol g^-1 h^-1) under identical conditions, underscoring the superior efficiency imparted by the triazine linkage in TAPT–OCH₃. Remarkably, even TAPT–OCH₃ outperformed many previously reported COF-based photocatalysts under comparable conditions, achieving this high efficiency without the use of any sacrificial agents, which highlights its superior intrinsic activity for photocatalytic H₂O₂ production (Fig. 4b and Table S2 in Supporting information). Consistent with its strong visible-light absorption, TAPT–OCH₃ achieved an apparent quantum yield (AQY) of 2.21% at 420 nm, with diminishing yields at longer wavelengths (Fig. 4c). Moreover, its solar-to-chemical efficiency of 0.23% also exceeded many existing COF systems, positioning it as a leading photocatalyst for H₂O₂ production. Critically, TAPT–OCH₃ retained full photocatalytic activity and structural stability over seven consecutive cycles (Fig. 4d), with minimal H₂O₂ decomposition observed during irradiation (Fig. S10 in Supporting information), highlighting its potential for long-term use. Post-reaction PXRD (Figs. S11a–c in Supporting information) and FT-IR (Figs. S11d–f in Supporting information) analyses confirm the robust structural integrity of TAPT–OCH₃, whose high activity and long-term stability make it a leading candidate for sustainable photocatalytic H₂O₂ production.

  Figure 4

  

  Figure 4.  (a) Time course of photocatalytic H₂O₂ production. (b) Performance comparison with reported photocatalysts. (c) Wavelength-dependent AQY values and solid-state UV-visible spectrum. (d) Cycling stability tests. (e) Bacterial inactivation. (f) Pollutants degradation. (g) Photoactivity in the presence of different scavengers. (h) CV test. (i) RRDE analysis for electron transfer number and H₂O₂ selectivity.

  DownLoad: Full-Size Img PowerPoint

  To enable practical application, TAPT–OCH₃ powders were integrated into a continuous-flow reactor, maintaining a stable H₂O₂ generation rate of 716 μmol g⁻¹ h⁻¹ over 20 h of visible-light exposure (Figs. S12a and b in Supporting information) [41,42]. The generated H₂O₂ demonstrated efficacy in real-world applications: The filtrate from TAPT–OCH₃ effectively neutralized antibiotic-resistant bacteria in dark conditions (Fig. S13 in Supporting information), while the material itself enabled in-situ bacterial inactivation (Fig. 4e) and degradation of pollutants such as norfloxacin under visible light (Fig. 4f). To exclude the potential role of reactive chlorine species (RCS), a control test in PBS buffer showed nearly identical disinfection kinetics to those in NaCl solution (Figs. S14a–c in Supporting information), confirming that bactericidal activity originates primarily from material-derived ROS rather than RCS. Quenching assays with various scavengers (Fig. S15 in Supporting information) identified ^•OH as the primary ROS responsible for bacterial inactivation, based on significantly reduced efficacy upon adding TBA (^•OH scavenger) [43]. The similar inhibition by methanol (scavenges ^•OH and ^•Cl) and lack of enhancement with high NaCl concentration jointly confirm that ^•Cl plays a negligible role. Notably, H₂O₂ production remained robust across a broad pH range (3–11; Fig. S16 in Supporting information) and in diverse water matrices, including tap, river, and seawater, with rates exceeding 880 μmol g^-1 h^-1 (Fig. S17 in Supporting information). Catalytic tests in water with various common ions (K⁺, Mg²⁺, Ca²⁺, Zn²⁺, Cl^-, HCO₃^-, CO₃^2-, SO₃^2-, HPO₄^2-, NO₃^-) show a maintained high H₂O₂ production rate (> 850 μmol g^-1 h^-1), demonstrating remarkable tolerance to common ions (Figs. S18a and b in Supporting information). Additionally, the material maintained excellent photocatalytic H₂O₂ production performance after 12–48 h of ultraviolet light exposure and with the addition of calcium carbonate (5–20 mmol/L) as simulated dirt (Fig. S19 in Supporting information), demonstrating high durability. Collectively, these results highlight TAPT–OCH₃ as a stable, efficient, and adaptable materials for advanced water purification and environmental remediation technologies.

  To clarify the reaction pathways underlying H₂O₂ photosynthesis, control experiments were performed under varying gas atmospheres and scavenger conditions. As illustrated in Fig. 4g, introducing an electron scavenger (AgNO₃) under N₂-saturation caused a slight reduction in H₂O₂ yield, suggesting that both the 2e^– ORR and 1e^– WOR pathways drive H₂O₂ photosynthesis in TAPT–OCH₃ system. Meanwhile, the absence of detectable O₂ in water oxidation experiments further ruled out the involvement of 4e^– WOR in the H₂O₂ photosynthesis process by TAPT–OCH₃ (Fig. S20 in Supporting information). Further insights into the reaction mechanism were obtained using p-benzoquinone (p-BQ) as a scavenger for superoxide radicals (O₂^•–) [44]. The significant decline in H₂O₂ production upon adding p-BQ highlighted O₂^•– as a key intermediate. This finding was corroborated by electron spin resonance (ESR) spectroscopy, which detected DMPO—O₂^•– signals under visible light (Fig. S21a in Supporting information), and nitroblue tetrazolium (NBT) assays, which quantitatively confirmed O₂^•– presence (Fig. S21b in Supporting information). Cyclic voltammetry (CV) in an O₂-saturated environment revealed a prominent oxygen reduction peak at −0.23 V vs. RHE for TAPT–OCH₃ (Fig. 4h), indicating strong oxygen reduction activity with low energy demand. Rotating ring-disk electrode (RRDE) analysis further supported this, showing an average electron transfer number of 2.34 and a H₂O₂ selectivity of 82% ± 2% (Fig. 4i). Together, these results confirm that H₂O₂ generation in TAPT–OCH₃ system proceeds via a two-step indirect ORR mechanism, where O₂ is first reduced to O₂^•–, followed by its subsequent conversion to H₂O₂.

  To elucidate the mechanism of H₂O₂ generation via 1e^– WOR, ^•OH trapping experiments were conducted using t-butyl alcohol (TBA) under N₂ atmosphere. As shown in Fig. 4g, the addition of TBA resulted in a slight reduction in H₂O₂ yield, suggesting that ^•OH contributes to the photocatalytic H₂O₂ production in the TAPT–OCH₃ system. This observation aligns with the DMPO-^•OH signals detected in EPR spectra (Fig. S22 in Supporting information). Further isotopic experiments employing H₂¹⁸O and ¹⁸O₂ confirmed that H₂O₂ originates directly from the ORR and WOR by TAPT–OCH₃ (Fig. 5a). Electrochemical analyses provided additional insights into the reaction pathway. Under N₂ atmosphere, the ring electrode potential was fixed at −0.3 V (vs. Ag/AgCl) to monitor oxygen evolution. The absence of detectable ring current for TAPT–OCH₃ (Fig. 5b) rules out a 4e^- WOR pathway. Conversely, when the ring electrode was set to the H₂O₂ oxidation potential (0.6 V vs. Ag/AgCl), a distinct oxidation current emerged (Fig. 5c), confirming H₂O₂ generation via the 1e^- WOR process. These results corroborate the ^•OH quenching experiments, solidifying the dominance of the 1e^- WOR pathway. Accordingly, the proposed mechanism involves exciton dissociation in TAPT–OCH₃ under light irradiation, generating photogenerated e^--h⁺ pairs. The photogenerated e^- reduce oxygen to form O₂^•–, and subsequently converted to H₂O₂, while the h⁺ oxidize water to adsorbed ^•OH, which then combine to yield H₂O₂. These results provide solid support that TAPT–OCH₃ undergo the 2e⁻ ORR process on H₂O₂ production, whereas a 1e⁻ WOR demonstrates remarkably enhanced overall photocatalytic activity.

  Figure 5

  

  Figure 5.  (a) H₂¹⁸O and ¹⁸O₂ isotope labeling experiment. RRDE voltammograms set at (b) −0.3 V vs. Ag/AgCl to detect O₂ and set at (c) 0.6 V vs. Ag/AgCl to detect H₂O₂. Calculated energy profile for (d) reduction of oxygen into H₂O₂ and (e) oxidation of water into H₂O₂. (f) In-situ DRIFT spectra of TAPT–OCH₃ during H₂O₂ generation under light irradiation. (g) Life cycle assessment label: 1-Global warming, Human health; 2-Global warming, Terrestrial ecosystems; 3-Global warming, Freshwater ecosystems; 4-Stratospheric ozone depletion; 5-Ionizing radiation; 6-Ozone formation, Human health; 7-Fine particulate matter formation; 8-Ozone formation, Terrestrial ecosystems; 9-Terrestrial acidification; 10-Freshwater eutrophication; 11-Marine eutrophication; 12-Terrestrial ecotoxicity; 13-Freshwater ecotoxicity; 14-Marine ecotoxicity; 15-Human carcinogenic toxicity; 16-Human non-carcinogenic toxicity; 17-Land use; 18-Mineral resource scarcity; 19-Fossil resource scarcity; 20-Water consumption, Human health; 21-Water consumption, Terrestrial ecosystem; 22-Water consumption, Aquatic ecosystems.

  DownLoad: Full-Size Img PowerPoint

  To elucidate the photocatalytic mechanism underlying H₂O₂ generation, first-principles calculations were performed. For the 2e⁻ ORR process, TAPT–OCH₃ exhibited a smaller energy barrier during the initial *O₂ formation step and a substantially reduced Gibbs free energy difference (ΔG) for the rate-determining *OOH intermediate than TAPA–OCH₃ and TAPB–OCH₃ (Fig. 5d and Fig. S23 in Supporting information). These results highlight superior oxygen affinity and catalytic efficiency of TAPT–OCH₃ in facilitating ORR. Regarding the WOR step, the ΔG of *OH adsorption on TAPT–OCH₃ was found to be optimal for H₂O₂ production (Fig. 5e and Fig. S24 in Supporting information). Excessively strong *OH adsorption could trigger over-oxidation to O₂, while weaker adsorption would release free ^•OH, both detrimental to selective H₂O₂ production [39,41]. Notably, TAPT–OCH₃ demonstrated a low energy barrier (0.52 eV) for the critical *OOH → H₂O₂ step but higher barriers for further oxidation to *O (2.06 eV) or *OH (1.24 eV), favoring selective 2e⁻ oxidation over competing 4e⁻ pathways. Complementary in-situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) identified key intermediates and active sites (Fig. 5f). Under light irradiation, the vibration peak at 976 cm^-1 in TAPT–OCH₃ were assigned to adsorbed peroxy species (-O-O) respectively [45]. Additional peaks of C—O (1410 cm⁻¹) and C=C-O (1213 cm⁻¹) confirmed O₂ reduction at C atoms of TAPT–OCH₃. During water oxidation, intensified peaks at 1150 cm^-1 (N—O) and 1367 cm⁻¹ (O—H), along with the emergence of TzH⁺ 1518 cm^-1, indicating that the water oxidation surrounding the N atoms in triazine rings of TAPT–OCH₃ [39]. Accordingly, its coplanar π-conjugated structure of TAPT–OCH₃ enhances exciton dissociation, yielding abundant charge carriers. Photoexcited electrons drive sequential O₂ reduction via *O₂^- intermediates, while holes oxidize water through *OH species, sustaining the photocatalytic cycle for H₂O₂ synthesis.

  Life cycle assessment (LCA) is critical for evaluating the environmental sustainability of the obtained COFs and informing strategies to align their development with global sustainability goals [46]. Using experimental data, the LCA checklists for the three COF synthesis were established in Tables S3-S5 (Supporting information). A range of descriptive indicators pertinent to wastewater treatment processes were employed in Fig. 5g, such as global warming, stratospheric ozone depletion and human carcinogenic toxicity, which helps us comprehensively evaluate the stability applications [47]. Further analysis using the endpoint ReCiPe methodology (Text S6) quantified impacts on human health, ecosystems, and resource depletion [48]. Toxicity effects (indicators 12–16) cover human carcinogenic, non-carcinogenic, and ecological toxicity, particularly relevant in systems involving chemicals, industrial processes, or e-waste. Ecosystem quality is assessed via eutrophication potential (indicators 10–11), crucial for systems with agricultural activities or wastewater discharge. Water resource impacts (indicators 20–22) apply to water-sensitive or high-consumption systems, evaluating effects on human health and aquatic ecosystems. Resource consumption is captured by mineral and fossil resource scarcity (indicators 18–19), essential for systems dependent on such materials. The results (Fig. S25 in Supporting information) indicated that TAPT–OCH₃ COF exhibit exceptional chemical resilience and reusability, retaining the initial pollutant removal efficiency even after multiple cycles, consistent with the durability long-term material consumption and waste generation, aligning with circular economy principles derived in Fig. 5g. Notably, the TAPT–OCH₃ COF exhibited notably higher environmental impacts, primarily attributed to the lower E_b for exciton dissociation in its rotation-inhibited π-conjugated structure. This trade-off positions it as a promising candidate for practical wastewater treatment despite synthesis challenges.

  In summary, we present a straightforward yet impactful strategy to enhance photocatalytic H₂O₂ production by optimizing the coplanar structure of the TAPT–OCH₃ COF. Through detailed analysis of the methoxy-triazine-modified TAPT–OCH₃, we demonstrated that this design promotes exciton dissociation and improves charge-carrier separation and migration. The system also achieved a solar-to-chemical conversion efficiency of 0.23% and an apparent quantum yield of 2.21% at 420 nm using only water and oxygen under simulated sunlight, surpassing current benchmarks for metal-free photocatalysts. The enhanced activity stems from the extended D-A units within the two-dimensional framework, which simultaneously optimizes the 2e^- ORR and 1e^- WOR pathways. This dual improvement ensures atom utilization efficiency, enabling highly efficient overall H₂O₂ photosynthesis. Our findings pave the way for developing sustainable, high-performance metal-free materials for green H₂O₂ production and other energy conversion applications.

  Declaration of competing interest

  The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

  CRediT authorship contribution statement

  Jun Gao: Writing – original draft, Investigation, Formal analysis, Data curation, Conceptualization. Lichao Wang: Formal analysis, Data curation, Conceptualization. Shunwei Huang: Supervision, Software, Data curation. Hao Du: Software, Formal analysis, Data curation. Huayue Zhu: Supervision, Software. Derek Hao: Validation, Supervision, Software. Yanling Wu: Methodology, Investigation, Formal analysis. Qi Wang: Writing – review & editing, Funding acquisition, Formal analysis, Conceptualization. Limin Jin: Writing – review & editing, Writing – original draft, Supervision, Data curation, Conceptualization.

  Acknowledgments

  This work was supported by "Pioneer" and "Leading Goose" R&D Program of Zhejiang (No. 2025C02240), National Natural Science Foundation of China (No. 22276168).

  Supplementary material

  Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2025.112283.

  
  
• 1. [1]

     C. Wang, X. Xie, F. Qiu, et al., Chin. Chem. Lett. 37 (2026) 111604. doi: 10.1016/j.cclet.2025.111604

  2. [2]

     F.N. Habarugira, D. Yao, W. Miao, et al., Chin. Chem. Lett. 35 (2024) 109886. doi: 10.1016/j.cclet.2024.109886

  3. [3]

     L. Li, J. Bai, J. Li, B. Zhou, Chem. Eng. J. 506 (2025) 159828. doi: 10.1016/j.cej.2025.159828

  4. [4]

     J. Li, N. Xu, Y. Zhang, et al., Chin. Chem. Lett. 36 (2025) 110470. doi: 10.1016/j.cclet.2024.110470

  5. [5]

     Q. Yang, Y. Zhang, P. Xiao, et al., Environ. Sci. Technol. 58 (2024) 19058–19069. doi: 10.1021/acs.est.4c06638

  6. [6]

     Y. Liu, J. Li, B. Zhou, et al., Environ. Chem. Lett. 7 (2009) 363–368. doi: 10.1007/s10311-008-0180-z

  7. [7]

     L. Jin, J. Xu, H. Zhu, et al., Chin. Chem. Lett. (2025), doi: 10.1016/j.cclet.2025.111865.

  8. [8]

     C. Zhang, W. Shen, K. Guo, et al., J. Am. Chem. Soc. 145 (2023) 11589–11598. doi: 10.1021/jacs.3c00689

  9. [9]

     T. Huang, W. Zhang, S. Yang, et al., Smart Mat. 5 (2024) e1309.

  10. [10]

      J.Y. Yue, L.P. Song, Y.F. Fan, et al., Angew. Chem. Int. Ed. 62 (2023) e202309624. doi: 10.1002/anie.202309624

  11. [11]

      L. Jin, M. Deng, J. Gao, et al., Chem. Eng. J. 515 (2025) 163722. doi: 10.1016/j.cej.2025.163722

  12. [12]

      Q. Li, M. Deng, J. Gao, et al., Sep. Purif. Technol. 362 (2025) 131722. doi: 10.1016/j.seppur.2025.131722

  13. [13]

      L. Wang, H. Du, X. Wang, et al., Environ. Res. 272 (2025) 121166. doi: 10.1016/j.envres.2025.121166

  14. [14]

      L.M. Jin, M. Sun, J.P. Yang, et al., Appl. Catal. B: Environ. 339 (2023) 123150. doi: 10.1016/j.apcatb.2023.123150

  15. [15]

      R. Wang, J. Liu, Q. Zhang, et al., Water Res. 252 (2024) 121209. doi: 10.1016/j.watres.2024.121209

  16. [16]

      J. Zhang, Z. Zhou, Z. Feng, et al., Environ. Sci. Technol. 56 (2022) 1331–1340. doi: 10.1021/acs.est.1c06368

  17. [17]

      L. Jin, Y. Wang, Q. Li, et al., Appl. Catal. B: Environ. 373 (2025) 125328. doi: 10.1016/j.apcatb.2025.125328

  18. [18]

      X. Lu, Y. Chang, S. Wang, et al., Chin. Chem. Lett. 36 (2025) 110277. doi: 10.1016/j.cclet.2024.110277

  19. [19]

      S.K. Loeb, P.J.J. Alvarez, J.A. Brame, et al., Environ. Sci. Technol. 53 (2019) 2937–2947. doi: 10.1021/acs.est.8b05041

  20. [20]

      M.Y. Sun, L. Zhang, Y. Li, et al., Chin. Chem. Lett. 36 (2025) 110035. doi: 10.1016/j.cclet.2024.110035

  21. [21]

      Z. Teng, H. Yang, Q. Zhang, et al., Nat. Chem. 16 (2024) 1250–1260. doi: 10.1038/s41557-024-01553-6

  22. [22]

      X.X. Wang, C.R. Zhang, R.X. Bi, et al., Adv. Funct. Mater. 35 (2025) 2421623. doi: 10.1002/adfm.202421623

  23. [23]

      L. Wang, D. Ma, A. Liu, et al., Angew. Chem. Int. Ed. 64 (2025) e202518097. doi: 10.1002/anie.202518097

  24. [24]

      H. Li, R. Chen, L. Sun, et al., Adv. Mater. 36 (2024) 2408778. doi: 10.1002/adma.202408778

  25. [25]

      S. Li, S. Xu, E. Lin, et al., Nat. Chem. 17 (2025) 226–232. doi: 10.1038/s41557-024-01690-y

  26. [26]

      A. Dey, F.A. Rahimi, S. Barman, et al., J. Mater. Chem. A 11 (2023) 13615–13622. doi: 10.1039/d3ta00901g

  27. [27]

      X. Wang, H. Li, S. Zhou, et al., Adv. Funct. Mater. 35 (2025) 2424035. doi: 10.1002/adfm.202424035

  28. [28]

      R.J. Wei, X. Luo, G.H. Ning, D. Li, Acc. Chem. Res. 58 (2025) 746–761. doi: 10.1021/acs.accounts.4c00774

  29. [29]

      H. Yan, M. Shen, Y. Shen, et al., Proc. Natl. Acad. Sci. U. S. A. 119 (2022) e2202913119. doi: 10.1073/pnas.2202913119

  30. [30]

      H. Chen, D. Li, M. Lin, et al., Adv. Mater. 37 (2025) 2500063. doi: 10.1002/adma.202500063

  31. [31]

      T. Ma, L. Wei, L. Liang, et al., Nat. Commun. 11 (2020) 6128. doi: 10.1038/s41467-020-19858-8

  32. [32]

      J. He, H. Liu, F. Zhang, et al., Adv. Funct. Mater. 32 (2022) 2110030. doi: 10.1002/adfm.202110030

  33. [33]

      X. Zhang, S. Cheng, C. Chen, et al., Nat. Commun. 15 (2024) 2649. doi: 10.3390/ma17112649

  34. [34]

      Z. Chen, B. Yang, J. Li, et al., Chin. Chem. Lett. 36 (2025) 110320. doi: 10.1016/j.cclet.2024.110320

  35. [35]

      L. Zhang, C. Wang, Q. Jiang, et al., J. Am. Chem. Soc. 146 (2024) 29943–29954. doi: 10.1021/jacs.4c12339

  36. [36]

      W. Zeng, Y. Dong, X. Ye, et al., Chin. Chem. Lett. 35 (2024) 109252. doi: 10.1016/j.cclet.2023.109252

  37. [37]

      Y. Yang, H. Niu, L. Xu, et al., Appl. Catal. B: Environ. 269 (2020) 118799. doi: 10.1016/j.apcatb.2020.118799

  38. [38]

      X. Chen, X. Zhang, X. Xiao, et al., Angew. Chem. Int. Ed. 62 (2023) e202216010. doi: 10.1002/anie.202216010

  39. [39]

      F. Liu, P. Zhou, Y. Hou, et al., Nat. Commun. 14 (2023) 4344. doi: 10.1038/s41467-023-40007-4

  40. [40]

      C. Qin, X. Wu, L. Tang, et al., Nat. Commun. 14 (2023) 5238. doi: 10.1038/s41467-023-40991-7

  41. [41]

      M. Xie, W. Zheng, M. Sun, et al., Appl. Catal. B 345 (2024) 123708. doi: 10.1016/j.apcatb.2024.123708

  42. [42]

      B. Shen, H. Du, A. Liu, et al., J. Clean Prod. 523 (2025) 146458. doi: 10.1016/j.jclepro.2025.146458

  43. [43]

      A. Liu, X. Ma, B. Shen, et al., Chem. Eng. J. 504 (2025) 158929. doi: 10.1016/j.cej.2024.158929

  44. [44]

      S. Chen, J. Gao, P. Wang, et al., Sep. Purif. Technol. 378 (2025) 134672. doi: 10.1016/j.seppur.2025.134672

  45. [45]

      C. Yang, L. Hou, J. Yang, et al., Chem. Eng. J. 510 (2025) 161631. doi: 10.1016/j.cej.2025.161631

  46. [46]

      G. Ren, J. Zhang, T. Rao, et al., Appl. Catal. B: Environ. 373 (2025) 125349. doi: 10.1016/j.apcatb.2025.125349

  47. [47]

      Y. Ren, J. Wang, L. Yang, et al., Environ. Sci. Technol. 59 (2025) 11414–11425. doi: 10.1021/acs.est.5c04310

  48. [48]

      E. Chatzisymeon, S. Foteinis, D. Mantzavinos, T. Tsoutsos, J. Clean Prod. 54 (2013) 229–234. doi: 10.1016/j.jclepro.2013.05.013

• 

  Figure 1  (a-c) Schematic diagrams. (d-f) Electrostatic potential maps models. (g-i) LOL-π isosurfaces of the obtained COFs.

  下载: 全尺寸图片 幻灯片
  

  Figure 2  (a-c) PXRD patterns with the top and side view of corresponding simulated packing structures. (d-f) SEM images. (g) N₂ sorption isotherms with calculated pore size distributions. (h) The contact angle for the synthesized COFs. (i) TEM image of TAPT–OCH₃.

  下载: 全尺寸图片 幻灯片
  

  Figure 3  (a) FT-IR spectra, (b) ¹³C NMR spectra, (c) XPS spectra of N 1s, (d) UV–vis DRS spectra, (e) calculated band alignments, (f) the calculated TDOS, (g) temperature-dependent PL spectral, (h) steady-state PL spectra, (i) time-resolved PL spectra of synthesized COFs.

  下载: 全尺寸图片 幻灯片
  

  Figure 4  (a) Time course of photocatalytic H₂O₂ production. (b) Performance comparison with reported photocatalysts. (c) Wavelength-dependent AQY values and solid-state UV-visible spectrum. (d) Cycling stability tests. (e) Bacterial inactivation. (f) Pollutants degradation. (g) Photoactivity in the presence of different scavengers. (h) CV test. (i) RRDE analysis for electron transfer number and H₂O₂ selectivity.

  下载: 全尺寸图片 幻灯片
  

  Figure 5  (a) H₂¹⁸O and ¹⁸O₂ isotope labeling experiment. RRDE voltammograms set at (b) −0.3 V vs. Ag/AgCl to detect O₂ and set at (c) 0.6 V vs. Ag/AgCl to detect H₂O₂. Calculated energy profile for (d) reduction of oxygen into H₂O₂ and (e) oxidation of water into H₂O₂. (f) In-situ DRIFT spectra of TAPT–OCH₃ during H₂O₂ generation under light irradiation. (g) Life cycle assessment label: 1-Global warming, Human health; 2-Global warming, Terrestrial ecosystems; 3-Global warming, Freshwater ecosystems; 4-Stratospheric ozone depletion; 5-Ionizing radiation; 6-Ozone formation, Human health; 7-Fine particulate matter formation; 8-Ozone formation, Terrestrial ecosystems; 9-Terrestrial acidification; 10-Freshwater eutrophication; 11-Marine eutrophication; 12-Terrestrial ecotoxicity; 13-Freshwater ecotoxicity; 14-Marine ecotoxicity; 15-Human carcinogenic toxicity; 16-Human non-carcinogenic toxicity; 17-Land use; 18-Mineral resource scarcity; 19-Fossil resource scarcity; 20-Water consumption, Human health; 21-Water consumption, Terrestrial ecosystem; 22-Water consumption, Aquatic ecosystems.

  下载: 全尺寸图片 幻灯片
•