Conjugated-engineering covalent organic framework for synergistic artificial photosynthesis of hydrogen peroxide and high-value chemicals

INTRODUCTION

Since the discovery and initial reporting of H₂O₂ in 1818, it has garnered extensive attention worldwide^[1-3]. Among the various strategies developed, photocatalytic H₂O₂ production is considered the most promising^[4-15]. In this process, O₂ captures a photogenerated electron from the conduction band (CB) to form a superoxide radical anion (O₂^•-), which then captures a proton to form hydroperoxyl radicals (•OOH). These intermediates subsequently undergo further proton- and electron-coupling steps to yield H₂O₂. Simultaneously, other electron donors, such as benzyl amine, can donate electrons to photogenerated holes in the valence band (VB). This process is followed by a series of steps that lead to the formation of imines [Figure 1A]^[16-27]. While the artificial synthesis of H₂O₂ and high-value chemicals provides new insights and methodologies for enhanced solar energy utilization and value-added product synthesis, achieving their synergistic artificial photosynthesis remains challenging. Prolonged reaction times lead to H₂O₂ decomposition, whereas short reaction times yield low amounts of high-value chemicals. Constructing a photocatalyst that minimizes H₂O₂ decomposition is key to effectively controlling the yield of high-value chemicals and the H₂O₂ generation rate. In 2024, Liu et al. constructed donor-acceptor covalent organic frameworks (COFs) for highly efficient H₂O₂ photosynthesis coupled with oxidative organic transformations^[28]. However, the range of high-value chemicals produced was limited to benzyl amines and the process necessitated the use of a Xenon (Xe) lamp. Developing a novel synergistic photosynthesis of H₂O₂ and high-value chemicals remains crucial for the development of artificial photosynthesis.

Figure 1. Conjugated-engineering COF for synergistic artificial photosynthesis of H₂O₂ and high-value chemicals. (A) Mechanism of synergistic artificial photosynthesis and challenge; (B) Structure and advantages of TAAB-Py-COF. COF: Covalent organic framework; LEDs: light-emitting diodes.

Metal organic frameworks (MOFs) or COFs are porous crystalline materials built from organic units, which are prized in photocatalysis for their customizable porosity, extended π-conjugation, tunable bandgaps, high surface areas, and stability^[29-36]. Recent studies indicated that 2D COFs could serve as heterogeneous photocatalysts, driving light-induced chemical reactions^[37-40]. Among them, TAAB-Py-COFs, synthesized by condensing 1,2,4,5-tetrakis(4-aminoaryl)-benzene (TAAB) and Py-CHO, have shown great photocatalytic ability in various organic reactions^[41-46]. One of the key advantages of TAAB-Py-COFs is their broad donor-acceptor (D-A) interface, which has electron-withdrawing and electron-donating traits, making it easy to abstract and donate electrons to substrates. Additionally, aryl groups with different conjugated systems can be introduced at the 1,2,4,5 positions of central benzene in TAAB to tailor the photocatalytic properties for various applications. Furthermore, Py-COFs have been successfully employed as photocatalysts in H₂O₂ generation, suggesting their potential as efficient catalysts in synergistic artificial photocatalysis processes^[47-50]. We envisaged that reasonably designed TAAB-Py-COFs with well-optimized conjugation and pore structures are of great significance to the efficiency and sustainability of the synergistic photocatalytic systems.

Based on the aforementioned considerations, we engineered naphthyl group instead of phenyl group at the 1,2,4,5 positions of central benzene in TAAB, yielding TANB-Py-COF. Naphthyl substituents enhance π-π stacking and electronic delocalization, often improving crystallinity, porosity, and optoelectronic properties compared to phenyl groups^[51,52]. This material exhibited the expected photocatalytic performance in a heterogeneous visible-light catalytic system, which facilitated synergistic artificial photosynthesis of H₂O₂ and a series of high-value chemicals [Figure 1B]. Using TANB-Py-COF as the catalyst in CH₃CN under an air atmosphere, an apparent quantum yield (AQY) of 7.87% was achieved. Moreover, the synergistic artificial photosynthesis of imines or sulfoxides, via amine couplings or the photooxidations of thioethers, was achieved with yields of up to 99%. This work provided a precedent synergistic artificial photosynthesis system of hydrogen peroxide and high-value chemicals via TAAB-Py-COF.

RESULTS AND DISCUSSION

Initially, we synthesized the TANB-Py-COF by the condensation between 1,2,4,5-tetrakis(6-aminonaphthyl)-benzene (TANB) and 4,4’,4’’,4’’’-(pyrene-1,3,6,8-tetrayl)tetrabenzaldehyde (Py-CHO) [Figure 2A and B]. Compared to known TAPB-Py-COF, TANB-Py-COF exhibited stronger electron-withdrawing properties, suggesting a greater potential to abstract electrons during the single electron transfer (SET) process from benzyl amine to benzyl amine radical cation. Additionally, the more extensive conjugated area of TANB-Py-COF enhanced its ability to be excited by blue light-emitting diodes (LEDs). The powder X-ray diffraction (PXRD) pattern of TANB-Py-COF exhibited two prominent diffraction peaks, with the most intense at 5.32° and another at 10.68°. These peaks were assigned to the (010) and (103) facets, respectively [Figure 2C]. The simulated PXRD patterns of the eclipsed AA stacking conducted by Materials Studio Software were in good agreement with the experimental PXRD patterns with a P1 space group. The Pawley refinement cell parameters were a = 25.9 Å, b = 5.0 Å, c = 27.4 Å, α = 90°, β = 90° and γ = 90°, with good agreement factors R_wp = 5.23% and R_p = 4.11% for TANB-Py-COF [Figure 2C and Supplementary Table 1]. Furthermore, almost identical PXRD patterns to the fresh sample were obtained after treatment of TANB-Py-COF in 3M NaOH, triethylamine, 3M HCl, trifluoroacetic acid and boiling water [Figure 2D], suggesting the high structural stability of TANB-Py-COF. Fourier transform infrared (FT-IR) spectra of TANB-Py-COF were analyzed by comparison with the monomer TANB and Py-CHO. The disappearance of the N-H vibration at 3,371 cm^-1 and the C=O vibration at 1,695 cm^-1 together with appearance of C=N vibration at 1,697 cm^-1 confirmed the high condensation degree of TANB-Py-COF [Figure 2E]. Chemical shifts at 144.8 and 145.2 ppm assigned to C=N bond in the solid-state ¹³C cross-polarization/total sideband suppression (¹³C-CP/TOSS) spectrum also indicated the successful formation of TANB-Py-COF with the imine linkage [Figure 2F] and other chemical shifts were assigned to the C=C bonds of TANB-Py-COF. The TANB-Py-COF featured a microporous structure with a Brunauer-Emmett-Teller (BET) surface area of 459.9 m²/g, pore volume of 0.33 cm³/g and pore size distribution with a diameter of 1.13 nm [Figure 2G and H]. TANB-Py-COF demonstrated the capability to remain stable up to a temperature of 400 °C in nitrogen atmosphere, as indicated by thermogravimetric analysis (TGA), thereby highlighting its excellent thermal stability [Figure 2I]. The abrupt mass loss above 400 °C corresponds to cleavage of the organic linkers and the subsequent release of small molecules. The scanning electron microscopy (SEM) images revealed a rod-like morphology of TANB-Py-COF [Supplementary Figure 1A]. The transmission electron microscopy (TEM) images displayed distinct, ordered fringes, which not only confirmed the excellent crystallinity of TANB-Py-COF but also revealed a quadrilateral pore structure with a periodicity of approximately 1.1 nm, matching with the BET analysis results [Supplementary Figure 1B].

Figure 2. Structure and characterization of TANB-Py-COF. (A) Chemical structure of TANB-Py-COF and TAPB-Py-COF; (B) Graphic view of one layer of TANB-Py-COF simulated by Material Studio. (azure, pink, C; blue, N; white, H); (C) PXRD patterns of TANB-Py-COF; (D) PXRD patterns of TANB-Py-COF before and after treatment in triethylamine, 3M NaOH, HOAc, 3M HCl and boiling water for 3 days; (E) Fourier transform infrared (FT-IR) spectroscopy spectra of TANB-Py-COF, TANB and Py; (F)¹³C-CP/TOSS NMR spectrum of TANB-Py-COF; (G) Nitrogen adsorption and desorption isotherm at 77 K of TANB-Py-COF; (H) Pore size distribution of TANB-Py-COF; (I) Thermogravimetric analysis of TANB-Py-COF. COF: Covalent organic framework; PXRD: powder X-ray diffraction; ¹³C-CP/TOSS: ¹³C cross-polarization/total sideband suppression; NMR: nuclear magnetic resonance.

After successfully synthesizing TANB-Py-COF, we investigated its potential in artificial photosynthesis for the production of hydrogen peroxide and N-benzyl imines. Using benzyl amine 1a as a model substrate, the reaction was conducted under a 40 W blue LED light source with a wavelength of 410 nm in an air atmosphere at room temperature [Supplementary Figure 2]. TANB-Py-COF catalyzed the reaction to yield N-benzyl imines 2a with an impressive 99% yield in CH₃CN [Table 1, entry 1], while also generating hydrogen peroxide at a rate of 18.32 mmol g^-1 h^-1. In contrast, TAPB-Py-COF, synthesized via condensation between 1,2,4,5-tetrakis(4-aminophenyl)-benzene (TAPB) and Py-CHO, exhibited a lower hydrogen peroxide generation rate of 16.24 mmol g^-1 h^-1 [Table 1, entry 2]. One hour was insufficient for complete conversion [Table 1, entry 3], and extending the reaction time to two hours resulted in a disappointing hydrogen peroxide generation rate due to H₂O₂ decomposition [Table 1, entry 4]. The LED wavelength selection process indicated that the catalytic reactions were most effective in 410 nm [Table 1, entry 1 vs. entries 5-7]. As anticipated, no product was formed in the absence of light or under a nitrogen atmosphere [Table 1, entries 8 and 9]. An oxygen atmosphere resulted in a lower rate of hydrogen peroxide generation at 22.18 mmol g^-1 h^-1 [Table 1, entry 10]. A catalyst loading of 1 mol% was found to be disappointing for photocatalysis [Table 1, entry 11]. Therefore, the optimized conditions involved using TANB-Py-COF as the catalyst in CH₃CN under an air atmosphere [Table 1, entry 1], achieving an AQY of 7.87%.

Table 1

Optimizations of the reaction

Entries	Catalysts	Wavelength/nm	Time/h	Conv./%	Generation rate of H2O2/mmol g-1 h-1
1	TANB-Py-COF	410	1.5	99	18.32
2	TAPB-Py-COF	410	1.5	99	16.24
3	TANB-Py-COF	410	1	78	16.83
4	TANB-Py-COF	410	2	99	11.33
5	TANB-Py-COF	390	1.5	64	5.65
6	TANB-Py-COF	420	1.5	57	5.45
7	TANB-Py-COF	450	1.5	74	7.56
8a	TANB-Py-COF	410	1.5	0	0
9b	TANB-Py-COF	410	1.5	0	0
10c	TANB-Py-COF	410	1.5	99	22.18
11d	TANB-Py-COF	410	1.5	86	13.22

All reactions were conducted on a 0.1 mmol scale with 1.0 eq. benzyl amine 1a and 2 mol% COF in 5.0 mL CH₃CN under blue LED irradiation at room temperature in air. Conversion was determined by GC-MS. ^aIsolated yield. ^a performed in the dark. ^b performed under N₂ atmosphere. ^c performed under O₂ atmosphere. ^d performed under 1 mol% COF. COF: Covalent organic framework; LED: light-emitting diode; GC-MS: gas chromatography-mass spectrometry.

Both TANB-Py-COF and TAPB-Py-COF were evaluated for their optical band gaps and electronic structures using ultraviolet/visible (UV/Vis) diffuse reflectance spectroscopy (DRS) and Mott-Schottky measurements. As shown in Figure 3A, both COFs demonstrated similar optical absorption with an edge around 550 nm. The optical band gaps were calculated to be approximately 2.29 eV for TANB-Py-COF and 2.31 eV for TAPB-Py-COF from the Tauc plots [Figure 3B]. Mott-Schottky measurements further revealed that both COFs are n-type semiconductors, with flat band potentials (E_fb) of -0.60 V and -0.65 V vs. Ag/AgCl for TANB-Py-COF and TAPB-Py-COF, respectively [Figure 3C]. Based on well-established principle that CB bottom nearly equals the E_fb in n-type semiconductors^[53,54], the CB potentials were estimated to be around -0.40 V and -0.45 V vs. Normal Hydrogen Electrode (NHE) for TANB-Py-COF and TAPB-Py-COF, respectively. These potentials are more negative than the theoretical potential for O₂ reduction to O₂^•-. The VB positions were deduced to be 1.89V vs. NHE for TANB-Py-COF and 1.86 V vs. NHE for TAPB-Py-COF, combining the band gap data from UV-DRS [Figure 3D]. These results confirmed that both COFs possessed suitable energy potentials to activate benzyl amine radical cations and O₂^•-. To further elucidate the photoelectrochemical properties of TANB-Py-COF and TAPB-Py-COF, transient photocurrent responses and electrochemical impedance spectroscopy (EIS) were conducted. As shown in Figure 3E, TANB-Py-COF exhibited a significantly higher photocurrent density compared to TAPB-Py-COF, indicating faster photoresponse and more efficient photoinduced charge transfer. The Nyquist impedance diagram of TANB-Py-COF displayed a smaller semicircle than that of TAPB-Py-COF [Figure 3F], indicating lower internal charge transfer resistance and more efficient charge transportation in TANB-Py-COF.

Figure 3. The photoelectrochemical measurements of TANB-Py-COF (green line) and TAPB-Py-COF (blue line). (A) UV/Vis diffuse reflectance spectroscopy; (B) Tauc plots; (C) Mott-Schottky plots; (D) Band structure diagram as determined from Mott-Schottky plots and UV-DRS; (E) Transient photocurrent measurements; (F) EIS Nyquist plots. COF: Covalent organic framework; UV/Vis: ultraviolet/visible; DRS: diffuse reflectance spectroscopy; EIS: electrochemical impedance spectroscopy.

Under the optimized reaction conditions, we investigated the substrate scope using various benzyl amines to assess the generality of this artificial photosynthesis process [Figure 4]. The reaction proved to be versatile, accommodating a wide range of benzyl amines with different substituents, including methyl, halogens, and alkoxy groups (2a-d), and consistently delivering the corresponding imines with 99% yields while maintaining good to excellent H₂O₂ generation rates. Both electron-donating and electron-withdrawing substituents on the benzene ring were well tolerated. Additionally, 2-thienyl substituted amine was effectively accommodated, yielding imine 2e. Furthermore, tetrahydroisoquinoline served as an effective substrate, producing product 2f in high yield with a H₂O₂ generation rate of 34.45 mmol g^-1 h^-1. Notably, alkyl amine also emerged as a suitable substrate for this reaction protocol, yielding the corresponding product 2g in excellent yield and with good H₂O₂ generation rates. The yields of 2f and 2g were detected by effective carbon number (ECN) concept^[55]. The response factors (R_f) of standard compounds were determined with Gas Chromatography-Mass Spectrometry (GC-MS) using

Figure 4. All reactions were carried out on a 0.1 mmol scale with 1.0 eq. amines 1a-g and 2 mol% TANB-Py-COF in 5.0 mL CH₃CN under blue LED irradiation at room temperature in air. Isolated yield. The orange segments in the figure denote the generation rates of H₂O₂. ^aYield detected by effective carbon number (ECN) concept. COF: Covalent organic framework; LED: light-emitting diode.

where C_c is the carbon moles of the model compound, C_is is the carbon moles of the internal standard (biphenyl), A_c is the area of the model compound, A_is is the area of the internal standard, R_f is the response factor of the compound, ECN_is is the effective carbon number of the internal standard, and ECN_c is the effective carbon number of the compound.

Beyond carbon-nitrogen bond construction, the versatility of this artificial photosynthetic platform extends to the selective oxygenation of thioethers [Figure 5]. Under the same photoredox conditions, a broad spectrum of thioethers 3a-f underwent controllable oxidation to deliver the corresponding sulfoxides or sulfones 4a-f in a single step. Alkyl thioethers proved particularly compatible and the conversion of simple dialkyl sulfides to sulfones 4e-f proceeded smoothly, furnishing isolated yields of 62%-85 % while simultaneously sustaining steady H₂O₂ evolution rates of 120-150 μmol g^-1 h^-1. These results underscore the catalysts’ capacity to orchestrate both photoredox transformations and selective oxygen-atom transfer, thereby expanding the synthetic utility of COF-based photocatalysis beyond amination to encompass sulfur functionalization.

Figure 5. All reactions were carried out on a 0.1 mmol scale with 1.0 eq. thioethers 3a-f and 2 mol% TANB-Py-COF in 5.0 mL CH₃CN under irradiation of blue LEDs in air atmosphere at room temperature. Isolated yield. The orange segments in the figure denote the generation rates of H₂O₂. ^aYield detected by effective carbon number (ECN) concept. COF: Covalent organic framework; LEDs: light-emitting diodes.

To further demonstrate the practicality of this reaction, we conducted a recycle experiment with TANB-Py-COF after the reaction was completed. In the artificial photosynthesis of H₂O₂ and benzyl amine, TANB-Py-COF was successfully retrieved and reused across six subsequent cycles, with negligible diminution in its catalytic performance [Figure 6A]. In addition, we carried out the PXRD and FT-IR measurements of TANB-Py-COF after multiple cycles. PXRD analyses of the recycled TANB-Py-COF revealed only a minor loss of crystallinity [Figure 6B]. FT-IR measurements indicated the stability of the C=N bonds, suggesting that the key functional groups remained intact [Figure 6C]. This result highlighted the robustness and reusability of TANB-Py-COF in this system. To elucidate the underlying mechanism of this reaction, a series of mechanistic studies were carried out. To delineate the roles of photogenerated electrons and holes, various scavengers were introduced into the system^[56,57]. As shown in Figure 6D, the yield dropped sharply to 0% when tetramethylpiperidine N-oxyl (TEMPO) and butylated hydroxytoluene (BHT, superoxide scavenger), (NH₄)₂C₂O₄ (hole scavenger), and K₂S₂O₈ (electron scavenger) were added, indicating the critical involvement of both charge carriers. These results suggested that the photogenerated electrons and holes played essential roles in the catalytic process, facilitating the formation of reactive species necessary for the reaction to proceed efficiently. Electron paramagnetic resonance (EPR) spectroscopy, employing 5,5-dimethyl-1-pyrroline N-oxide (DMPO) as a spin trap, was utilized to probe the active radical intermediates generated during the reaction [Figure 6E]. The EPR analysis demonstrated that spin adduct signals were detectable exclusively under light irradiation. Under an argon atmosphere, a substantial amount of Ph(•CH₂)NH₂ was identified, whereas both O₂^•- and Ph(•CH₂)NH₂ were observed under an oxygen atmosphere. These EPR findings robustly corroborate the high efficiency of TANB-Py-COF in generating active radical intermediates, thereby indirectly affirming the pivotal roles of O₂^•- and Ph(•CH₂)NH₂ in the associated redox reactions.

Figure 6. Practicality of the reactions and mechanistic studies. (A) Recycle experiments of TANB-Py-COF; (B) PXRD patterns of TANB-Py-COF before and after six consecutive photocatalytic cycles; (C) FT-IR patterns of TANB-Py-COF before and after six consecutive photocatalytic cycles; (D) Artificial photosynthesis with different radical scavengers; (E) EPR analysis. COF: Covalent organic framework; PXRD: powder X-ray diffraction; FT-IR: fourier transform infrared; BHT: butylated hydroxytoluene; TEMPO: tetramethylpiperidine N-oxyl.

The local polarization and charge separation behavior of the TANB-Py-COF fragment were studied using density functional theory (DFT) calculations at the B3LYP/6-31G* level of theory using Gaussian 16. The highest occupied molecular orbital (HOMO, -5.33 eV) is primarily localized on the Py core, while the lowest unoccupied molecular orbital (LUMO, -2.16 eV) is delocalized along the imine bond linker into the naphthyl substituent of the TANB moiety [Figure 7A], resulting in a HOMO-LUMO gap of 3.17 eV. Electrostatic potential (ESP) mapping highlights the intrinsic polarization of the fragment; the Py unit shows pronounced negative potential, whereas the imine linker and naphthyl-substituted TANB moiety show alternating electron-rich and electron-deficient regions [Figure 7B]. The optimized fragment possesses a dipole moment of 1.73 D, consistent with a polarized conjugated network. The naphthyl group extends conjugation, enhancing LUMO delocalization while preserving the polarized network, ensuring the TANB-Py-COF suitability for efficient optoelectronic function. These electronic features assist exciton separation and charge transport, similar to the performance seen in COFs with different imine bond orientations^[58]. To determine charge separation efficiency, the S_r index was calculated using time-dependent DFT (TD-DFT), yielding a value of 0.691, revealing significant spatial separation between hole (blue) and electron (green) densities [Figure 7C] and consistent with efficient intrafragment charge transport^[59]. Further analysis reveals that the imine linker and extended conjugation serve as the primary electron transport pathway, whereas the Py moiety functions as the electron-donating site.

Figure 7. DFT calculations (A) HOMO and LUMO orbitals and energy levels of the TANB-Py-COF fragment; (B) ESP map with dipole moment of the TANB-Py-COF fragment; (C) Electron-hole distribution of the TANB-Py-COF fragment. DFT: Density functional theory; HOMO: highest occupied molecular orbital; LUMO: lowest unoccupied molecular orbital; ESP: electrostatic potential; COF: covalent organic framework.

CONCLUSIONS

In this work, we successfully synthesized a conjugated-engineered COF, designated as TANB-Py-COF, and explored its potential as an efficient catalyst for the photosynthesis of H₂O₂ under blue-LED irradiation. This innovative system demonstrated an impressive AQY of 7.87% at 420 nm, along with a robust H₂O₂ production rate of 18.32 mmol g^-1 h^-1. Moreover, the TANB-Py-COF enabled the synergistic photosynthesis of imines through both amine couplings and photooxidations of thioethers, achieving yields as high as 99%. These results highlight the versatility and efficiency of the TANB-Py-COF in facilitating multiple chemical transformations under visible-light irradiation. Future research will focus on further optimizing the COF structure and exploring its applications in other photocatalytic reactions.