Covalent Organic Framework Photocatalysts: Decoding Linkage Chemistry in Hydrogen Peroxide Synthesis From Air and Water

ABSTRACT

Covalent organic frameworks (COFs) provide a polymer platform for exploring covalent linkages to design ordered skeletal and porous architectures. However, the role of linkages in controlling structural and functional evolutions remains to be well explored. In this study, we reported hexaphenyltriphenylene COF photocatalysts constructed with ketazine or azine linkages that differ by a single pinpoint methyl substituent, enabling a controlled interrogation of linkage chemistry. Unexpectedly, the ketazine linkage enhances water uptake, accelerates transport, and directs water confinement within trigonal pores. Simultaneously, it modulates the π‐electronic structure through hyperconjugation and inductive/resonance effects, extends light absorption, lowers exciton binding energy, prolongs charge‐separated lifetimes, and promotes balanced charge transport. These synergistic structural and electronic evolutions translate into exceptional photocatalysis for hydrogen peroxide production from air and water under ambient conditions. Ketazine‐HPTP‐COF achieves a production rate of 8.17 mmol g⁻¹ h⁻¹ with an apparent quantum yield of 15.1% at 420 nm, outperforming azine‐linked, amorphous, and other photocatalysts. The system operates under sunlight, enabling scalable production, and maintains activity across tap water, rainwater, and seawater. Mechanistic studies reveal dense yet spatially resolved photocatalytic sites, where linkage sites mediate oxygen reduction and knot units drive water oxidation, promoting photosynthesis through efficient charge and mass transport.

Condensing hexaphenylbenzene knots with hydrazine creates ketazine‐linked trigonal frameworks featuring dense photocatalytic sites and mass‐transporting 1D channels. The ketazine‐linked frameworks direct water confinement, promote exciton splitting, and accelerate charge transport, producing H₂O₂ efficiently from air and diverse water resources under ambient conditions and outperforming azine‐linked, amorphous, and other photocatalysts.

1. Introduction

Covalent organic frameworks (COFs) are an emerging class of crystalline porous materials with predesignable backbones and well‐defined porous architectures [1, 2, 3, 4, 5, 6, 7], offering a versatile platform for designing photocatalysts to drive solar‐to‐chemical energy conversion [8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23]. Through topology design, π building units can be spatially organized into ordered π arrays, to enable efficient light harvesting, charge transport, and catalytic site integration, while precise control over pore size, shape, and environment governs mass transport, reactant enrichment, and product release [1, 24, 25, 26, 27, 28, 29]. COFs are constructed by linking knots and linkers via covalent bonds; these linkages dedicate both the local structure and the global environment of both skeleton and pore [27, 28, 29, 30]. Despite their importance, how the linkage chemistry governs structural and functional evolutions remains poorly understood.

Hydrogen peroxide (H₂O₂) is an environmentally benign oxidant and energy carrier with broad applications [31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41]. Its industrial production, however, relies predominantly on the anthraquinone process, which is energy‐intensive and associated with safety and environmental concerns [42, 43, 44, 45]. Artificial photosynthesis provides a sustainable alternative by utilizing light energy to drive H₂O₂ production from water and oxygen under mild conditions [46, 47, 48]. Achieving this transformation requires orchestrating multielectron oxygen reduction and water oxidation reactions, with precise coupling of charge separation, carrier transport, and mass supply [31, 32, 33, 34, 49, 50, 51, 52]. Addressing these challenges necessitates identifying the key structures in COFs that govern photosynthesis.

In this study, we elucidate the role of linkage chemistry in COF photocatalysis through the designed synthesis of hexaphenyltriphenylene COFs featuring ketazine or azine linkages that differ by a single methyl substituent. This pinpoint structural variation enables direct interrogation of linkage effects on both framework structure and function. Combined experimental and computational studies reveal that ketazine linkages profoundly modulate the evolution of porous, skeletal, and photocatalytic structures and functions. For effects on porous structure and function, the ketazine‐linked COFs enhance uptake of water and dissolved oxygen, accelerate water transport, and direct water confinement within trigonal nanogrooves. For effects on π electronic structure and function, the ketazine‐linked COFs promote exciton splitting, prolong charge separation, facilitate carrier transport, and enhance photocurrent generation. These effects work synergistically in photocatalysis to achieve exceptional activity in H₂O₂ synthesis from air and water under ambient conditions. The ketazine‐linked COFs display a production rate of 8.17 mmol g⁻¹ h⁻¹ and achieve an apparent quantum yield of 15.1% at 420 nm, greatly surpassing azine‐linked, amorphous, and other photocatalysts. The system maintains high activity under natural sunlight across tap water, rainwater, and natural seawater. These findings open a way to ketazine linkage for structural and functional design and establish linkage chemistry as a decisive strategy for developing high‐performance photocatalysts.

2. Results and Discussion

2.1. Design and Synthesis

We selected trigonal COFs as structural motifs because they provide the highest π density and the smallest pore dimensions among reported COF topologies. The dense π stacks offer strong light harvesting, high densities of photocatalytic sites, and short charge transfer distance. Concurrently, the 1D trigonal supermicropores generate pronounced capillary effects, facilitating mass transport of reactants to catalytic sites and product release (Figure 1). These features are expected to synergistically improve photocatalytic performance.

FIGURE 1.

Synthesis (a) Schematic of Ketazine‐HPTP‐COF. (b, c) Reconstructed structures of (b) one layer and (c) ten layers of Ketazine‐HPTP‐COF. (d) Schematic of Azine‐HPTP‐COF. (e, f) Reconstructed structures of (e) one layer and (f) ten layers of Azine‐HPTP‐COF.

The ketazine‐linked Ketazine‐HPTP‐COF (Figure 1a–c) and azine‐linked Azine‐HPTP‐COF (Figure 1d–f) were synthesized via solvothermal condensation of hydrazine with 1,1′,1″,1‴,1⁗,1‴″‐(triphenylene‐2,3,6,7,10,11‐hexaylhexakis(benzene‐4,1‐diyl)) hexakis(ethan‐1‐one) and 2,3,6,7,10,11‐hexakis(4‐formylphenyl) triphenylene in 71% and 83% yields, respectively. Structurally, the two COFs differ only by the methyl substituent at the linkage site (Figure 1a,d), thereby providing a well‐controlled framework platform to disclose linkage effects. Unexpectedly, this pinpoint chemical variation at the C═N linkage site triggers pronounced effects on porous architecture, π‐electronic structure, and photocatalytic activity.

2.2. Chemical and Crystal Structures

Both COFs are crystalline, as evident by powder X‐ray diffraction (PXRD) signals (Figure 2). Ketazine‐HPTP‐COF exhibited a main peak at 4.86° indexed to the (100) facet, along with peaks at 8.58°, 9.82°, 14.78°, and 22.61°, corresponding to the (110), (200), (300), and (001) facets, respectively (Figure 2a, black cross). In comparison, Azine‐HPTP‐COF displayed diffractions at 4.86°, 8.45°, 9.72°, 14.58°, and 22.82°, assigned to the (100), (110), (200), (300), and (001) facets, respectively (Figure 2c, black cross). Structural modeling indicates that both COFs adopt AA‐stacking mode (Figure 2a, blue curve; Figure 2c, orange curve), with matching peak intensities and positions to the experimental PXRD results. Pawly refinements further reproduce the PXRD profiles (Figure 2a,c, red curves) with minimal differences (grey curves) from the experimental results, as revealed by R _w and R _p values of 2.73% and 1.69% for Ketazine‐HPTP‐COF, and 3.63% and 2.73% for Azine‐HPTP‐COF, respectively. The refined unit cell parameters are a = b = 2.14 nm, c = 3.75 Å, α = β = 90.00°, and γ = 58.09° for Ketazine‐HPTP‐COF (Figure 2b), and a = 2.12 nm, b = 2.15 nm, c = 3.92 Å, α = β = 90.00°, and γ = 60.01° for Azine‐HPTP‐COF (Figure 2d).

FIGURE 2.

Chemical structures. (a) PXRD patterns of Ketazine‐HPTP‐COF of experimental results (black cross), Pawley refined (red curve), their difference (grey curve), and simulated AA stacking mode (blue curve). (b) Unit cell of Ketazine‐HPTP‐COF. (c) PXRD patterns of Azine‐HPTP‐COF of experimental results (black cross), Pawley refined (red curve), their difference (grey curve), and simulated AA stacking mode (orange curve). (d) Unit cell of Azine‐HPTP‐COF. (e, f) FTIR spectra of (e) Ketazine‐HPTP‐COF and (f) Azine‐HPTP‐COF. (g, h) SS ¹³C CP/MAS NMR spectra of (g) Ketazine‐HPTP‐COF and (h) Azine‐HPTP‐COF.

Fourier transform infrared spectroscopy (FTIR) revealed that Ketazine‐HPTP‐COF exhibited characteristic C═N and C─H (CH₃) vibrational bands at 1637 cm⁻¹ and 1363 cm⁻¹, respectively (Figure 2e), whereas Azine‐HPTP‐COF displayed a C═N stretching band at 1623 cm⁻¹ (Figure 2f). Solid‐state ¹³C cross polarization magic angle spinning nuclear magnetic resonance spectroscopy (SS ¹³C CP/MAS NMR, Figure 2g,h) confirmed the framework structures, with Ketazine‐HPTP‐COF (Figure 2g) displaying a distinct carbon signal of methyl group at 8.6 ppm. Scanning electron microscopy reveals that Ketazine‐HPTP‐COF is particle (Figure S1), while Azine‐HPTP‐COF is stick in morphology (Figure S2).

Ketazine‐HPTP‐COF is robust in organic solvents, acid (HCl, 4 M), and base (NaOH, 4 M), retaining crystallinity (Figure 3a) and showing less weight loss (Figure 3b). In comparison, Azine‐HPTP‐COF retained PXRD intensity (Figure 3c) but showed greater weight loss (Figure 3d). The interlayer stacking energy is 187.8 and 169.3 kJ mol⁻¹ for Ketazine‐HPTP‐COF and Azine‐HPTP‐COF, respectively, indicating stronger π–π interactions in Ketazine‐HPTP‐COF (Table S1).

FIGURE 3.

Stability and effects of the methyl group. (a–d) PXRD patterns and residual ratio of (a, b) Ketazine‐HPTP‐COF and (c, d) Azine‐HPTP‐COF after dispersion in different solvents. (e) Electronic effects of the methyl group on the linkage.

Ketazine‐HPTP‐COF has a thermal stability over 400°C under N₂ (Figure S3), which is much higher than that (∼ 300°C) of Azine‐HPTP‐COF (Figure S4). This enhanced robustness of Ketazine‐HPTP‐COF is attributed to the methyl substituent in ketazine linkages. It induces hyperconjugation, inductive, and resonance stabilization effects to reinforce the framework by hyperconjugation and inductive effects and to soften the interlayer charge repulsion that enhances stacking structures (Figure 3e). In addition, the methyl groups may contribute through steric effects by optimizing local packing geometry and suppressing unfavorable and less efficient stacking modes, which is consistent with the stronger stacking interactions observed for Ketazine‐HPTP‐COF (Table S1). In the absence of such effect, Azine‐HPTP‐COF exhibits weaker structural stability.

2.3. Effects on the Evolution of Pore Structures and Functions

Nitrogen sorption isotherm measurements at 77 K confirmed that both Ketazine‐HPTP‐COF (Figure 4a) and Azine‐HPTP‐COF (Figure 4c) are microporous materials. Ketazine‐HPTP‐COF and Azine‐HPTP‐COF exhibited a Brunauer–Emmett–Teller (BET) surface area of 576 and 530 m² g⁻¹, respectively. Ketazine‐HPTP‐COF displayed pore sizes of 0.92 and 1.40 nm (Figure 4b), while Azine‐HPTP‐COF slightly increased the pore sizes to 1.02 and 1.41 nm, consistent with the absence of methyl groups on the pore walls (Figure 4d). Both Ketazine‐HPTP‐COF and Azine‐HPTP‐COF displayed a pore volume of 0.20 cm³ g⁻¹ (Table S2). Interestingly, in Ketazine‐HPTP‐COF, the methyl groups protrude into the trigonal nanochannels to cause partial physical blockage, which restricts the accessibility of small pore and reduces its contribution to the total pore volume. In contrast, for Azine‐HPTP‐COF, the absence of methyl substituents leaves the small triangular channels unobstructed. Consequently, these small pores remain fully accessible and contribute significantly to the overall pore volume.

FIGURE 4.

Porosity and water sorption. (a, c) Nitrogen sorption isotherm curves (filled circle: adsorption; open circle: desorption) of (a) Ketazine‐HPTP‐COF and (c) Azine‐HPTP‐COF. (b, d) Pore size (blue and orange circles) and its distribution (black circles) profiles of (b) Ketazine‐HPTP‐COF and (d) Azine‐HPTP‐COF. (e, f) Water sorption curves of (e) Ketazine‐HPTP‐COF and (f) Azine‐HPTP‐COF at different temperatures (each from top to bottom: 283.15, 288.15, and 298.15K). (g, h) Dynamic water vapor sorption (g) and transfer rate–time variation curves (h) of Ketazine‐HPTP‐COF (blue curve) and Azine‐HPTP‐COF (orange curve).

Water vapor sorption measurements (Figure 4e,f and Table S3) revealed distinct pore environments arising from linkage variation. Ketazine‐HPTP‐COF (Figure 4e) exhibited broad induction regions (P/P ₀ = 0–0.474, 0–0.476, and 0–0.499) and extended capillary condensation regimes (P/P ₀ = 0.474–0.848, 0.476–0.848, and 0.499–0.849) across 283.15–298.15 K. In contrast, Azine‐HPTP‐COF (Figure 4f) showed narrower induction region (P/P ₀ = 0–0.275, 0–0.274, and 0–0.299), and earlier capillary condensation (P/P ₀ = 0.275–0.401, 0.274–0.426, and 0.299–0.424). This shift is attributed to the methyl groups lining the pore walls in Ketazine‐HPTP‐COF, which protrude into the trigonal channels and increase the pressure required for vapor penetration along the 1D pores.

Despite this, surprisingly, Ketazine‐HPTP‐COF achieved a high water‐uptake capacity of 0.19 g g⁻¹ at P/P ₀ = 0.9 and 298.15 K (Table S3), with a heat of adsorption (Q _st) of 45.8 kJ mol⁻¹ and a pore occupancy exceeding 95%. In contrast, Azine‐HPTP‐COF exhibited a low uptake capacity of 0.11 g g⁻¹, with a low Q _st of 44.6 kJ mol⁻¹ and a low pore occupancy of 54% (Figures S5–S7). Although the methyl groups increase the vapor pressure required for initial pore penetration, once water enters the channels, the methyl‐functionalized trigonal nanogrooves provide stronger confinement and cooperative water–water interactions. As a result, Ketazine‐HPTP‐COF exhibits a much higher uptake capacity (0.19 g g⁻¹), higher pore occupancy (> 95%), and faster transport kinetics than Azine‐HPTP‐COF. These results highlight the critical role of methyl‐functionalized linkages in tuning pore accessibility and adsorption capacity.

Dynamic vapor sorption (DVS) measurements further elucidate water uptake kinetics at specific relative humidity (R.H.) and temperature. As shown in Figure 4g, Ketazine‐HPTP‐COF (blue curve) and Azine‐HPTP‐COF (orange curve) exhibited a water vapor sorption capacity of 103.0 and 95.1 mg H₂O g⁻¹ COF at 80% R.H. and 298.15 K, respectively. Kinetic fitting using biexponential models yields excellent agreement (R ²> 0.99, Figures S8 and S9; Table S4). Notably, Ketazine‐HPTP‐COF (Figure 4h, blue curve) displayed an initial water uptake rate of 17.25 mg H₂O g⁻¹ COF min⁻¹, which is much higher than that (12.75 mg H₂O g⁻¹ COF min⁻¹) of Azine‐HPTP‐COF (orange curve). Analysis of water flow rate versus adsorption extent further confirms faster transport in Ketazine‐HPTP‐COF (Figure S10), indicating promoted mass transfer within the methyl‐functionalized channels at the early stage.

Electrostatic potential mapping revealed that both Ketazine‐HPTP‐COF (Figure 5a) and Azine‐HPTP‐COF (Figure 5b) develop polarized pore surfaces that interact with water molecules. However, adsorption simulations show distinct spatial distributions. In Ketazine‐HPTP‐COF, water molecules preferentially localize within trigonal nanogrooves, providing confined environments that coincide with water oxidation sites (Figure 5c). In contrast, Azine‐HPTP‐COF exhibits a more uniform distribution of water molecules across the pores (Figure 5d). This spatial confinement effect in Ketazine‐HPTP‐COF establishes a direct structure–function relationship between linkage chemistry, pore architecture, and catalytic site utilization.

FIGURE 5.

Pore structures and environments. (a, b) Electrostatic potential maps of (a) Ketazine‐HPTP‐COF and (b) Azine‐HPTP‐ COF. (c, d) Water adsorption density fields of (c) Ketazine‐HPTP‐COF and (d) Azine‐HPTP‐COF (blue spots: water).

2.4. Effects on π Electronic Structures and Functions

Ketazine‐HPTP‐COF and Azine‐HPTP‐COF absorb visible photons with different electronic absorption band edges (Figure 6a). Noticeably, Ketazine‐HPTP‐COF (blue curve) enables light absorption beyond 550 nm, suggesting a better light‐harvesting capability compared to Azine‐HPTP‐COF (orange curve). This redshift is related to a more compact interlayer stacking, which strengthens π–π interactions in Ketazine‐HPTP‐COF (Figure 2b,d). The extended absorption tail is likely associated with enhanced interlayer electronic coupling and low‐energy electronic transitions arising from the ketazine linkage, which contributes to visible‐light harvesting but has a limited influence on the optical bandgap determined from the absorption onset. Therefore, Tauc analysis revealed comparable bandgaps of 2.77 eV and 2.79 eV for Ketazine‐HPTP‐COF (Figure 6b, blue curve) and Azine‐HPTP‐COF (orange curve), respectively. Mott–Schottky and valence band measurements place the conduction band (CB) (vs. NHE, pH = 6.8) at –1.27 V and –1.22 V and the valence band (VB) at 1.50 V and 1.57 V for Ketazine‐HPTP‐COF and Azine‐HPTP‐COF, respectively (Figure 6c; Figures S11 and S12). Both satisfy the thermodynamic requirements for oxygen reduction and water oxidation; however, Ketazine‐HPTP‐COF provides a larger driving force for oxygen reduction (more negative CB) and a reduced barrier for water oxidation (lower VB).

FIGURE 6.

π Electronic structure and function. (a–c) U V–Vis absorption spectra (a), tauc plots (b), and band alignment (c) of Ketazine‐HPTP‐COF (blue curves) and Azine‐HPTP‐COF (orange curves). (d) PDOS of Ketazine‐HPTP‐COF. (e, f) Photocurrent density (e) and impedance spectra(f) of Ketazine‐HPTP‐COF (blue curves) and Azine‐HPTP‐COF (orange curves).

Density functional theory calculations reveal delocalized frontier orbitals for both Ketazine‐HPTP‐COF (Figure S13) and Azine‐HPTP‐COF (Figure S14). The projected density of states (PDOS) shows that Ketazine‐HPTP‐COF possesses a higher density of p‐orbital states near the valence band edge (Figure 6d, blue curve, Figure S15), suggesting an increased probability of photoexcitation. Consistently, Ketazine‐HPTP‐COF exhibits stronger photocurrent responses (Figure 6e, blue curve) and a smaller semicircle in Nyquist plots (Figure 6f, blue curve), indicating enhanced photoconductivity and reduced charge‐transfer resistance relative to Azine‐HPTP‐COF (Figure 6e,f, orange curves).

Temperature‐dependent photoluminescence measurements yield an exciton binding energy (E _b) of 62 meV for Ketazine‐HPTP‐COF (Figures 7a,b and S16), significantly lower than that (80 meV) of Azine‐HPTP‐COF (Figure S17). This difference between Ketazine‐HPTP‐COF and Azine‐HPTP‐COF likely arise from the unique electronic effects of the methyl groups and the strength of interlayer interaction, which polarizes the framework to facilitate exciton dissociation. As a result, Ketazine‐HPTP‐COF facilitates exciton splitting to generate charges.

FIGURE 7.

Exciton dynamics and photosynthesis. (a–e) Temperature‐dependent fluorescence (a), its corresponding kinetics trace (b), transient absorption (c) and its kinetics (d), and time‐correlated single photon counting measurements (e) of Ketazine‐HPTP‐COF (blue curves) and Azine‐HPTP‐COF (orange curves). (f) Photocatalytic H₂O₂ generation from water and air with different amounts of Ketazine‐HPTP‐COF. (g) H₂O₂ production rates of Ketazine‐HPTP‐COF (blue bars) and Azine‐HPTP‐COF (orange bars). h) Apparent quantum yields (blue circles) and electronic absorption spectrum (black curve) of Ketazine‐HPTP‐COF. (i) Photocatalytic H₂O₂ production from water and air under sunlight and open‐air conditions for 1 h at seven different days.

Femtosecond transient absorption spectroscopy (fs‐TA) reveals a broad excited‐state absorption (ESA) band at 600–700 nm, characteristic of π–π* transitions, partially overlapping with a ground‐state bleach (GSB) near 500 nm (Figures 7c and S18) [53, 54, 55]. This ESA feature is consistent with the steady‐state absorption spectra. Notably, the GSB signal showed negligible decay within the experimental time window (2 ns), suggesting the existence of long‐lived excited species. To quantify the exciton behavior, decay traces were extracted at 700 nm (Figures 7d and Figure S19). The kinetic profile of Ketazine‐HPTP‐COF was well‐fitted with a biexponential function, yielding a fast component of τ₁ = 1.2 ps and a slow component of τ₂ = 340 ps (Table S5). The former is attributed to initial vibrational relaxation or exciton cooling, while the latter represents the recombination of the main exciton population. Compared to Azine‐HPTP‐COF (τ₂ = 272.55 ps), the prolonged τ₂ in Ketazine‐HPTP‐COF indicates a stabilized excited state.

Time‐correlated single photon counting (TCSPC) confirms prolonged exciton lifetimes. Ketazine‐HPTP‐COF exhibits a long‐lived decay component of 4.72 ns, after initial decay with τ = 0.7 ns (Figure 7e, blue curve, Figure S20), which is comparable with that of Azine‐HPTP‐COF (1.80 ns, Figure 7e, orange curve,). This behavior suggests effective exciton transfer to trap states, suppressing nonradiative relaxation and extending the time window for energy harvesting toward charge transport and redox reactions.

Flash‐photolysis time‐resolved microwave conductivity (FP‐TRMC) measurements were performed under air, SF₆, and NEt₃ atmospheres to identify the dominant charge carriers and their transport in Ketazine‐HPTP‐COF (Figure S21) and Azine‐HPTP‐COF (Figure S22) [56, 57]. As shown in Table S6, Azine‐HPTP‐COF exhibited a photoconductivity (ΦΣμ) of 2.0 × 10⁻⁹ m² V⁻¹ s⁻¹ under ambient air, which decreased to 1.0 × 10⁻⁹ and 1.3 × 10⁻⁹ m² V⁻¹ s⁻¹ under SF₆ and NEt₃, respectively. The resulting electron and hole conducting ratio was estimated to be 1.3, indicating a slightly dominant electron transport and supporting the n‐type character of Azine‐HPTP‐COF. In contrast, Ketazine‐HPTP‐COF displayed a higher ΦΣμ value of 4.6 × 10⁻⁹ m² V⁻¹ s⁻¹ under ambient air. This improvement can be ascribed to the presence of the methyl groups. They reinforce the framework through hyperconjugation/inductive effects, mitigate interlayer charge repulsion, and thereby promote more efficient charge separation and transport. The Ketazine‐HPTP‐COF shows significantly higher stability of photogenerated electron‐hole pairs as represented by the transient electric conductivity after ∼10 µs. Under SF₆ and NEt₃ atmospheres, the ΦΣμ values decreased to 1.8 × 10⁻⁹ and 1.5 × 10⁻⁹ m² V⁻¹ s⁻¹, respectively (Table S6). Photoconductivity analysis indicates a balanced electron‐hole transport (ratio = 0.8) in Ketazine‐HPTP‐COF. In this study, the balanced charge transport indicates a comparable transport efficiency of photogenerated electrons and holes and hence concurrently promotes the two half‐reactions of oxygen reduction and water oxidation, as evidenced by the electron‐to‐hole transport ratio approaching unity. The methyl substituents in ketazine linkages enhance local π‐electron density, mitigate interlayer electrostatic repulsion, and promote balanced charge transport.

The above experimental and computational studies revealed the profound impacts of the linkages on the π electronic structures and functions of COFs through multiple mechanisms: (1) enhancing light‐harvesting capability via hyperconjugation/inductive effects; (2) optimizing band energetics to favor both oxygen reduction and water oxidation; (3) facilitating exciton dissociation; and (4) prolonging charge‐separated states. In addition, the confinement of water molecules within trigonal nanogrooves enriches reactants around the water oxidation sites and facilitates their interactions with photogenerated holes. These synergistic effects establish ketazine‐linked COFs as an advanced platform for efficient solar‐to‐chemical energy conversion.

2.5. Synergistic Impacts on Photocatalysis

Ketazine‐HPTP‐COF and Azine‐HPTP‐COF promote photosynthesis of H₂O₂ from water and air upon irradiation with visible light (λ ≥ 420 nm) under ambient conditions (Figure S23, see detailed experimental procedures in Section 1.22). The photocatalyst loading was systematically optimized (Figures 7f and S24). As a typical protocol, Ketazine‐HPTP‐COF (1.0 mg in 15‐mL aerated deionized water in a 50‐mL Schlenk tube) achieves a H₂O₂ production rate of 8.17 mmol g⁻¹ h⁻¹ at an optimal COF concentration of 0.07 g L⁻¹ (Figure 7g), whereas Azine‐HPTP‐COF reaches 5.00 mmol g⁻¹ h⁻¹ at its optimal condition with a higher optimal loading of 0.17 g L⁻¹ (2.5 mg COF in 15‐mL H₂O, Figure 7g). The production rate decreased with increased concentration of photocatalyst, due to the increased light scattering and the inefficient utilization of COFs in the photocatalytic system. Ketazine‐HPTP‐COF exhibited an apparent quantum yield of 15.1%, 7.5%, 3.1%, and 0.5% at 420, 450, 500, and 550 nm, respectively (Figure 7h).

Control experiments further underscore the importance of structural order. The amorphous Ketazine‐HPTP polymer and HPTP monomer exhibited much lower activity, with H₂O₂ production rate of 3.10 and 0.13 mmol g⁻¹ h⁻¹, respectively (Figure S25). These results highlight the critical role of crystalline framework organization in enabling efficient photocatalysis.

Ketazine‐HPTP‐COF also demonstrates good operational stability, maintaining its activity over multiple cycles with negligible performance decay (Figure S26), retaining its crystalline and chemical structures (Figures S27 and S28), confirming its structural robustness under reaction conditions.

Notably, the production rate increased to 11.78 mmol g⁻¹ h⁻¹ upon irradiated with 300‐W Xe lamp (λ > 300 nm) (Figure S29), indicating efficient utilization of a broader solar spectrum. Remarkably, under natural sunlight and open‐air conditions, Ketazine‐HPTP‐COF (50 mg in 1–L water) sustains photocatalytic activity over 7 days (Figure 7i), delivering an average production rate of 3.06 mmol g⁻¹ h⁻¹ and cumulating over 7 L of H₂O₂ solution using only water, air, and sunlight (Figures S30 and S31).

The system demonstrates strong tolerance to complex aqueous environments. Ketazine‐HPTP‐COF produces H₂O₂ from tap water, rainwater, and natural seawater at rates of 5.15, 5.85, and 3.00 mmol g⁻¹ h⁻¹, respectively (Figure S32), highlighting its applicability across diverse and nonideal water sources. Compared with previously reported COF photocatalysts, Ketazine‐HPTP‐COF achieves competitive high performance despite its structurally simple design (Table S7), Unlike many high‐performing COF photocatalysts that rely on donor–acceptor architectures, heteroatom‐rich building units, or complex functional‐group engineering, the present system employs only a pinpoint methyl substitution at the linkage site. This minimal structural modification simultaneously optimizes pore environment, water management, exciton dynamics, and charge transport, leading to substantial enhancement in photocatalytic performance. These results highlight linkage engineering as a simple yet powerful strategy for developing efficient COF photocatalysts.

2.6. Mechanistic Studies

In a typical H₂O₂ photosynthetic reaction, two half‐reactions of oxygen reduction and water oxidation are involved in the photochemical processes. At first, we excluded the possibility of metal interferences originated from the coupling reactions for monomer synthesis by inductively coupled plasma mass spectrometry (Table S8).

By revealing the photocatalytic sites and reactive intermediates, we conducted a series of control experiments and theoretical calculations (Figure 8).

FIGURE 8.

Mechanistic studies. (a) EPR spectra of Ketazine‐HPTP‐COF using DMPO as scavenger. (b) Gibbs free energy diagram for oxygen reduction reaction over the linkage of Ketazine‐HPTP‐COF (blue bar) and Azine‐HPTP‐COF (orange bar). (c) Gibbs free energy diagram for water oxidation at the phenyl site of Ketazine‐HPTP‐COF. (d–g) DRIFTS of Ketazine‐HPTP‐COF collected at different irradiation times during the photosynthesis under O2 (d), enlarged region of 1315–1680 cm⁻¹ (e), 1050–1110 cm⁻¹ (f) and 850–910 cm⁻¹ (g). (h) Active sites for oxygen reduction reaction and cycle. (i) Active sites for water oxidation reaction and cycle.

For oxygen reduction half‐reaction, we performed the following systematic investigations to explore the catalytic sites and reactive species. (1) There was no H₂O₂ evolution in the dark, proving that the photocatalytic reactions are driven by light (Figure S33). (2) No H₂O₂ was detected upon replacing air with argon, while its production rate boosted to 9.41 mmol g⁻¹ h⁻¹ using O₂‐saturated water (Figure S33). (3) Electron paramagnetic resonance (EPR) spectroscopy was conducted by using 5,5‐dimethyl‐1‐pyrroline N‐oxide (DMPO) as a spin trapping reagent, which reacts with superoxide anion radical (^•O₂ ^–) to form DMPO–^•O₂ ^– adduct to exhibit characteristic EPR signals. As shown in Figure 8a, the clear peaks suggested the involvement of ^•O₂ ^– in the reaction system and greatly increased their intensity with prolonged reaction time, thereby confirming a two‐step oxygen reduction mechanism (Figure 6c, dashed pink line, –0.35 V). (4) Computational studies screened a series of possible O₂ adsorption sites at the linkage, phenyl, and triphenylene core (Figure S34), where Ketazine‐HPTP‐COF exhibited higher O₂ affinity at each site over Azine‐HPTP‐COF, with the most favorable O₂ adsorption site at the ketazine linkage (Figure S35). (5) Charge distribution calculations revealed that both Ketazine‐HPTP‐COF and Azine‐HPTP‐COF possess the most positively charged carbon atoms at their linkages (Figure S36), which enable to accept electrons from the knot to serve as catalytic site for the oxygen reduction half‐reaction. (6) Based on Gibbs free energy calculations (Figure 8b), Ketazine‐HPTP‐COF presented smaller overpotential for this process, compared to Azine‐HPTP‐COF, demonstrating an easier reduction pathway proceeded with Ketazine‐HPTP‐COF.

Next, we investigated the catalytic intermediates by using in situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) (Figure 8d–g). Interestingly, the peaks at 1637 cm⁻¹ and 1363 cm⁻¹, corresponding to the ketazine linkage, increased gradually over the time course of reaction, which can be attributed to the increased polarity of chemical bond due to the adsorption of reactive intermediates (Figure 8e). This also indicates the catalytic site on the linkage and hence is in good agreement with the computational results. Noticeably, the characteristic peak at 1083 cm⁻¹ was assigned to the vibration of ^•O₂ ⁻ intermediate (Figure 8f). We observed the prominent growth of signals from H₂O₂, including O–H vibration at 2841 cm⁻¹, H─O─O bending vibration at 1380–1405 cm⁻¹, and O─O stretching vibration at 871 cm⁻¹ (Figure 8d–g). These peaks increased their intensity with the irradiation time, indicating a progressed accumulation of H₂O₂ in the system under light irradiation.

For water oxidation half‐reaction, we revealed the pathway and catalytic site by conducting the following control experiments and computational studies. (1) H₂O₂ production rate was unchanged upon addition of tert‐butanol as hydroxyl radical (^•OH) scavenger to the reaction system, excluding the involvement of ^•OH in the process (Figure S33). This result is also in line with the CB levels of the two COFs, which unfit the energy requirement for ^•OH‐involved water oxidation pathway (Figure 6c). (2) H₂O₂ production was inhibited by adding NaIO₃ as electron scavenger to the reaction system, suggesting that the process of one‐step direct water oxidation reaction is not involved in the reaction (Figure S33). (3) Charge distribution calculations revealed that both COFs possessed the most negatively charged carbon atoms at the phenyl spacers, which accumulate holes to proceed water oxidation half‐reaction (Figure S36). (4) Gibbs free energy calculations revealed that water oxidation reaction possibly occurs on the phenyl site via single‐site (*OOH) process rather than dual‐site (*O*OH) pathway as less rate‐determining process involves (Figure 8c).

With the above systematic investigations, the photochemical events for H₂O₂ evolution can be elaborated as follow: Ketazine‐HPTP‐COF harvests visible photons to generate electrons and holes via photoinduced electron transfer, which can be spatially separated on the linkage and knot units, respectively. These liker and knot sites function as catalytic centers for two‐electron oxygen reduction and four‐electron water oxidation. Parallelly, the supermicropores transport water and dissolved oxygen instantly to these catalytic sites, while the π arrays transport electrons and holes in a good balance. The linkage sites promote oxygen reduction (O₂ + 2H⁺ + 2e⁻ → H₂O₂) (Figure 8h), while the phenyl sites facilitate water oxidation (2H₂O → O₂ + 4H⁺ + 4e^–) (Figure 8i). This integrated system enables a closed photocatalytic cycle (2H₂O + O₂ → 2H₂O₂) under ambient conditions without sacrificial agents or cocatalysts, highlighting the synergistic coupling of electronic structure and mass transport governed by the linkage.

3. Conclusion

This work establishes linkage chemistry as a decisive element for governing multiscale structure–function relationships in COFs. By constructing hexaphenyltriphenylene COFs with ketazine and azine linkages that differ by a pinpoint methyl substituent, we disclose and quantify linkage effects on pore environments, π‐electronic structures, and photocatalytic activity.

The ketazine linkage introduces methyl arrays that reshape the pore landscape, leading to higher water uptake, accelerated transport, and spatial confinement of water within trigonal nanogrooves. Concurrently, it modulates the electronic structure through hyperconjugation and inductive/resonance effects, resulting in enhanced light absorption, reduced exciton binding energy, prolonged charge‐separated lifetimes, and improved carrier transport with balanced electron–hole conduction. These coupled effects translate into markedly enhanced photocatalytic performance for H₂O₂ production from air and water, with high rates, quantum efficiency, scalability under natural sunlight, and robust operation across diverse water sources.

Mechanistic investigations reveal that the ketazine‐linked frameworks enable spatially resolved catalysis, where linkage sites drive two‐electron oxygen reduction and knot phenyl units mediate four‐electron water oxidation, operating synergistically without sacrificial agents or cocatalysts. The integration of ordered π‐arrays for charge transport and supermicroporous channels for mass delivery synchronizes photophysical and catalytic processes, lowering kinetic barriers and maximizing efficiency. These findings demonstrate that even minimal chemical modification at the linkage level can reprogram both electronic and porous architectures, providing a unified strategy to couple exciton dynamics, charge transport, and mass transfer. This study not only advances the fundamental understanding of COF photocatalysis but also offers a generalizable design principle for developing scalable, sustainable systems for solar‐to‐chemical energy conversion.

Author Contributions

D. Jiang conceived the idea and led the project. Y.Z. Chen, X.Y. Mu conducted the experiments and measurements. S.L. Ji, K‐A. Kuo, Y.Y. Guo, Y. Tsutsui, and S. Seki performed the photoconductivity, transient absorption and fluorescence lifetime measurements. T. Wang and N. Yan conducted the DRIFTS measurements. X. Chen conducted temperature‐dependent photoluminescence measurements. Y.Z. Chen, X.Y. Mu, and D. Jiang wrote the paper. All authors have read and commented on the paper.

Conflicts of Interest

The author declares no conflicts of interest.

Supporting information

Supporting File 1: anie73563‐sup‐0001‐SuppMat.pdf.

Supporting File 2: anie73563‐sup‐0002‐cif.zip.

Data Availability Statement

The data that supports the findings of this study are available in the supplementary material of this article.