Enhancing photocatalytic performance of covalent organic frameworks via ionic polarization

Abstract

Covalent organic frameworks have emerged as a thriving family in the realm of photocatalysis recently, yet with concerns about their high exciton dissociation energy and sluggish charge transfer. Herein, a strategy to enhance the built-in electric field of series β-keto-enamine-based covalent organic frameworks by ionic polarization method is proposed. The ionic polarization is achieved through a distinctive post-synthetic quaternization reaction which can endow the covalent organic frameworks with separated charge centers comprising cationic skeleton and iodide counter-anions. The stronger built-in electric field generated between their cationic framework and iodide anions promotes charge transfer and exciton dissociation efficiency. Moreover, the introduced iodide anions not only serve as reaction centers with lowered H* formation energy barrier, but also act as electron extractant suppressing the recombination of electron-hole pairs. Therefore, the photocatalytic performance of the covalent organic frameworks shows notable improvement, among which the CH₃I-TpPa-1 can deliver an high H₂ production rate up to 9.21 mmol g⁻¹ h⁻¹ without any co-catalysts, representing a 42-fold increase compared to TpPa-1, being comparable to or possibly exceeding the current covalent organic framework photocatalysts with the addition of Pt co-catalysts.

Covalent organic frameworks (COFs) show great promise in photocatalysis but are limited by slow charge transfer. Here, the authors report a strategy to enhance the built-in electric field of COFs via ionic polarization, resulting in a hydrogen evolution rate of 9.21 mmol g^-¹ h^-¹ without Pt co-catalysts.

Introduction

Converting solar energy to “solar fuels” has been recognized as one of the most promising solutions for addressing the global energy and environmental crisis. Photocatalytic water splitting stands out among these solutions, since it harnesses renewable resources to yields green hydrogen product which possesses the advantages of green, sustainable and high energy density^1–3. Despite decades of investigation, the performance of most photocatalysts still falls short in real-world applications. Developing highly efficient and stable photocatalysts to promote water splitting remains at the forefront of our agenda.

Covalent organic frameworks (COFs), a kind of porous crystalline materials, are assembled by the condensation of organic building units through covalent bond^4–6. Compared with traditional semiconductors, COFs display incomparable advantages in photocatalysis, including large specific surface area, excellent optical properties, high thermal/chemical stability, and functional designability^7–9. However, it’s important to note that there are also some drawbacks potentially impeding their photocatalytic efficiency, such as high exciton dissociation energy and sluggish charge transfer. To overcome these issues, some strategies, such as heterojunction construction^10–13, defect engineering^14,15, metal doping^16–19, and donor-acceptor system construction^20–23, have been proposed. Nevertheless, these methods remain unsatisfactory due to inherent structural limitations, which contribute to the requirement for expensive Pt co-catalysts. Actually, the reliance on expensive co-catalysts extends beyond COFs and presents a universal challenge in the field of photocatalysis, stemming from the scarcity of active sites for hydrogen evolution through H* formation^24,25.

Recently, polarization strategy has been proven to be an alternative method to enhance the charge separation/migration properties of the photocatalysts^26–30. From a perspective of mechanism, polarization can enhance the dipole moment of the photocatalysts, and cause a change in the distribution of positive and negative charges on/in the surface/bulk of the materials, thus establishing a built-in electric field (BIEF) to provide a strong and persistent driving force for promoting the electron-hole pairs separation. Moreover, the enhancement of the dipole moment can also lower the exciton binding energy of the metal-free organic polymers^31–34. However, the polarization strategy has rarely been employed for COFs materials due to the difficulty in polarizing. Yet, we notice that endowing COFs materials with an ionic skeleton can not only improve electronic conductivity and charge transfer but also theoretically alter their positive and negative charge distribution and implant large numbers of polar sites into the frameworks, thereby achieving ionic polarization^35–37. Besides, ionic polarization may also open up new possibilities for new active sites for H* formation, potentially reducing the necessity for Pt co-catalysts. However, due to steric hindrance effects and the electrostatic repulsion of ionic building units, it is difficult to directly synthesize COFs with ionized frameworks using traditional bottom-up synthetic methods^38–40.

Among various different kinds of COFs, β-keto-enamine-based COFs have several advantages in photocatalysis. At first, they exhibit high structural stability due to the irreversible keto-enamine linkages, which contribute to their resistance against harsh chemical environments^41,42. Besides, β-keto-enamine-based COFs also have strong visible-light absorption capabilities and efficient charge separation, making them highly effective in visible-light photocatalysis^43,44 In addition, following the transformation of C = N linkage into the C-N bond in keto-enamine-based COFs, post-synthetic quaternization reaction can take place at the N sites of C-N bond, resulting in the formation of ammonium groups, thus endowing COFs with cationic frameworks to achieve ionic polarization.

Herein, we explore the possibility of using post modification approach to endow β-keto-enamine-based COFs with ionic structure. A series of β-keto-enamine-based COFs are successfully endowed with cationic frameworks by a well-designed post-synthetic quaternization reaction. As a result of the enhanced BIEF induced by ionic polarization, the migration/separation efficiency of photogenerated electron-hole pairs can be significantly improved. It can also enhance the dipole moment of COFs, thus lowering the exciton dissociation energy. Moreover, it is found that the incorporated counteracting iodide anions can serve as electron-withdrawing group (active sites) to promote electron transfer and lower H* formation energy barrier. With these dual benefits, ionized COFs materials demonstrate higher photocatalytic activities toward water splitting. The H₂ evolution rate of CH₃I-TpPa-1 reaches as high as 9.21 mmol g⁻¹ h⁻¹ without the use of Pt co-catalysts, surpassing that of most COFs photocatalysts assisted by Pt co-catalysts.

Results and discussion

In this work, a classical β-keto-enamine-based COFs named TpPa-1⁴⁵, formed via polymerization of 1,3,5-triformylphloroglucinol (Tp) and p-phenylenediamine (Pa), is first selected as a representative model for examining ionic polarization effects. Subsequently, cationic COFs containing iodide counterions (CH₃I-TpPa-1) are obtained through a quaternization modification. To verify the universality of the ion polarization approach, two additional β-keto-enamine-based COFs with different diamines linkers, such as 4,4’-diaminobiphenyl (BD) and 4,4’-(ethene-1,2-diyl) dianiline (EDDA), TpBD⁴⁶ and TpEDDA⁴⁷ have also been prepared, and they are endowed with cationic structure (CH₃I-TpBD and CH₃I-TpEDDA) by the same method (Fig. 1a). The high crystallinity of these samples is verified by powder X-ray diffraction (PXRD) tests. As shown in Fig. 1b, the diffraction peaks observed at 4.7° and 27° are indexed to the (100) and (001) reflection planes of TpPa-1, respectively. However, the (100) diffraction peaks of TpBD and TpEDDA shift toward smaller angles as the length of their diamine linker increases (Supplementary Figs. S2, S4, and S7). This phenomenon is related to the increased pore size, which can be measured according to the simulated COFs structure associated with ligand extension. Based on the comparison of peak location and relative intensities, the experimental PXRD profiles of all the samples are in good agreement with the simulated patterns, and their diffraction peaks consistently adhere to the AA stacking mode (Supplementary Figs. S1, S3, and S6). All of the above results demonstrate the successful preparation of the pristine COFs samples. After quaternization modification, the crystallinity of the CH₃I-TpPa-1 remains well maintained, as evidenced by the sharp (100) diffraction peaks at 4.7°, which is in accordance with those observed in the pristine TpPa-1 (Fig. 1b). Similar results can also be observed in other two modified COFs (CH₃I-TpBD and CH₃I-TpEDDA), illustrating that the post-modification process does not destroy or alter the original structure of the pristine COFs (Supplementary Figs. S5 and S8). To assess the permanent porosity of the samples, nitrogen sorption isotherm at 77 K is conducted. As expected, all the samples display microporous adsorption behavior, and the incorporation of iodide anion decreases the Brunauer-Emmett-Teller (BET) surface area from 715 m² g⁻¹ of the TpPa-1 to 407 m² g⁻¹ for CH₃I-TpPa-1 (Fig. 1c). The same trends are observed in the other two COFs (Supplementary Figs. S10 and S12). The pore size distribution results also demonstrate that the post-synthesis treatments have a negligible effect on the pore structure of the COFs (Supplementary Figs. S9, S11, and S13). Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) images show that the TpPa-1 nanorod crystals stack together to exhibit a flower-like morphology (Fig. 1d and Supplementary Fig. S14a). After treatment with CH₃I, the CH₃I-TpPa-1 maintains its original morphology and exhibits a flower-like appearance (Fig. 1e and Supplementary Fig. S14b). TpBD displays a coral-like morphology, while TpEDDA shows a stacked structure composed of small rod-like components. Similarly, both TpBD and TpEDDA maintain their original morphology after quaternization modification (Supplementary Figs. S15, S17). Furthermore, the energy dispersive spectrometer (EDS) mapping images of CH₃I-TpPa-1, CH₃I-TpBD, and CH₃I-TpEDDA clearly reveal that apart from the C, N, O elements, I element is also uniformly distributed within the framework, providing additional evidence for the successful quaternization modification (Fig. 1f and Supplementary Figs. S16, S18).

Fig. 1. Synthesis, structure, and morphological characterization of samples.

a Schematic illustration for post-quaternization modification of series β-keto-enamine-based COFs to prepare ionic COFs. b PXRD patterns of TpPa-1, CH₃I-TpPa-1 and simulated TpPa-1. c N₂ adsorption–desorption isotherms of TpPa-1, CH₃I-TpPa-1. d TEM images of TpPa-1. e, f TEM image of CH₃I-TpPa-1 and the corresponding elemental distributions of C, N, O, and I.

To further identify the successful quaternization modification of TpPa-1, a combination of X-ray photoelectron spectroscopy (XPS, Fig. 2a–d) and solid-state ¹³C cross-polarization magic angle spinning nuclear magnetic resonance (¹³C CP/MAS NMR, Fig. 2e) tests are performed. The XPS survey spectra demonstrate that C, N, O, and I elements existed in the modified samples with no other impurity peak observed. In the high-resolution C 1 s XPS spectrum of TpPa-1, three peaks located at 284.4, 285.5, and 288.5 eV can be observed, corresponding to C = C, C-N, and C = O bonds, respectively (Fig. 2b). The high-resolution N 1 s spectrum reveals a single peak located at 399.5 eV, matching well with the C-N-H bond, revealing the structural transition from Enol to Keto and the successful construction of TpPa-1 (Fig. 2c). As for the modified sample, CH₃I-TpPa-1 displays additional peeks at 286.8 and 402.5 eV in C 1 s and N 1 s spectra, respectively, ascribed to the formation of C-N⁺ bonds (Fig. 2a–c)⁴⁸. Moreover, the I 3 d spectrum displays two peaks at 629.4 and 617.9 eV, corresponding to I 3d_3/2 and I 3d_5/2, respectively, indicating that I element is present in COFs as a monovalent anion (Fig. 2d). Similar results can also be obtained in other two pairs of COFs (Supplementary Figs. S19–S26). Interestingly, in comparison with TpPa-1, a significant peak appears at 57 ppm in the ¹³C CP/MAS NMR spectrum of CH₃I-TpPa-1, representing the presence of N⁺-CH₃ in the modified framework (Fig. 2e)⁴⁹. Moreover, the NMR spectrum of the product obtained by treating the monomer synthesized from Tp and aniline with CH₃I further confirms the feasibility of the methylation reaction (Supplementary Fig. S27)¹⁶. Combining the above results, the successful functionalization of COFs can be fully validated. In addition, the binding energy between iodine anions and the cationic COF framework is calculated as high as − 3.23 eV (see atomic coordinates of optimized computational models in Supplementary Data 1), demonstrating that the structure of CH₃I-TpPa-1 is not only rational but also highly stable (Fig. 2f).

Fig. 2. The bonding structure of materials.

a The survey XPS spectra of TpPa-1 and CH₃I-TpPa-1. b C 1 s High-resolution XPS spectra of TpPa-1 and CH₃I-TpPa-1. c N 1 s High-resolution XPS spectra of TpPa-1 and CH₃I-TpPa-1. d I 3 d High-resolution XPS spectra of CH₃I-TpPa-1. e ¹³C CP/MAS NMR spectra of TpPa-1 and CH₃I-TpPa-1. f The simulated structure of a fragment of CH₃I-TpPa-1.

The photocatalytic tests are performed to validate our hypothesis that ion polarization can make great contributions to promoting the photocatalytic process, with initial experiments focusing on photocatalytic H₂ production under different sacrificial reagents. When methanol or triethanolamine are used as the sacrificial reagent, only trace amounts of H₂ are detected. However, when employing L-ascorbic acid sodium salt or L-ascorbic acid as the sacrificial reagent, the H₂ production rate is significantly increased, with L-ascorbic acid demonstrating the highest efficiency on H₂ production (Fig. 3a). Following these findings, subsequent photocatalytic tests are conducted under visible light irradiation with L-ascorbic acid as a sacrificial reagent. It is worth mentioning that the entire photocatalytic system is operated without any other metal or Pt co-catalysts. Due to the well-recognized issues such as high exciton dissociation energy and sluggish charge transfer, the original TpPa-1 does not exhibit high photocatalytic activity, only with an H₂ evolution rate of 0.22 mmol g⁻¹ h⁻¹. However, after quaternization modification, the photocatalytic performance of CH₃I-TpPa-1 is significantly enhanced, resulting in an H₂ evolution rate as high as 9.21 mmol g⁻¹ h⁻¹. This represents an approximately 42-fold increase compared to the original TpPa-1, even surpassing most of those reported COFs photocatalysts with Pt as co-catalysts (Fig. 3b and Supplementary Table S2)^50–52. The experiment with varying amounts of catalyst is also conducted, confirming that the catalyst exhibits high photocatalytic activity at different dosages (Supplementary Fig. S28). At the same time, photocatalytic hydrogen production experiments are conducted under seawater conditions, and the comparably photocatalytic activity suggests the potential for significant future applications in hydrogen production from seawater (Supplementary Fig. S29). In order to confirm the universality of ion polarization, we also investigated the photocatalytic activities of the other two pairs of COFs (TpBD vs CH₃I-TpBD and TpEDDA vs CH₃I-TpEDDA). Similar activity trends can be observed that after the quaternization modification, the activity of β-keto-enamine-based COFs series can be significantly enhanced (Fig. 3c, d and Supplementary Figs. S30, S32). In addition to catalytic activity, recyclability is also a crucial factor for evaluating the practical value of photocatalysts. As shown in Fig. 3e and Supplementary Figs. S31, S33, after undergoing 4 times (20 h) cycling tests, the cationic COFs can still maintain their high photocatalytic performance without any observable decline. The crystallinity and morphology of the three COFs can also be maintained after the reaction (Supplementary Figs. S34–S39). Moreover, inductively coupled plasma spectrometry (ICP) is used to quantitatively analyze the leakage of I^- after the cycling tests. The results show only a minimal increase in iodine content in the solution, proving that there is almost no leakage of iodine during the reaction (Supplementary Table S1). The above findings have all substantiated the recyclability of the material. In addition, the apparent quantum efficiency (AQY) of CH₃I-TpPa−1, CH₃I-TpBD, and CH₃I-TpEDDA at a wavelength of 500 nm is calculated to be 1.99, 0.85 and 0.7 %, respectively (Fig. 3f and Supplementary Figs. S40, 41). Furthermore, the other β-keto-enamine-based COFs, such as TpBpy⁵³ and TpAzo⁵⁴, with higher nitrogen content, are also synthesized and subjected to modification and photocatalytic tests under the same conditions (Supplementary Figs. S42, S43). The results are equally gratifying, as their activities are greatly enhanced, further validating the universality of this method (Fig. 3c, d).

Fig. 3. Evaluation of H₂ production performance.

a Photocatalytic hydrogen evolution performance of CH₃I-TpPa-1 under different sacrificial agents. b–d Photocatalytic hydrogen evolution activity of the synthesized products. e Recycling performance of CH₃I-TpPa-1. f Wavelength-dependent apparent quantum efficiency (AQY) of CH₃I-TpPa-1.

To explore the factors contributing to the improvement in the photocatalytic performance of β-keto-enamine-based COFs after CH₃I treatment, the optical absorption characteristics of samples are first studied using UV-vis diffuse reflectance spectroscopy (UV-vis DRS). These three COFs exhibit significant absorption in the visible light region, indicating great potential for visible light photocatalysis. The absorption edge of the samples is only slightly changed after quaternization modification with CH₃I, suggesting a minor variation in their band gaps. According to the Kubelka-Munk (KM) equation and Tauc plots, the band gap energies of TpPa-1 and CH₃I-TpPa−1 are determined to be 2.10 and 2.09 eV, respectively. For TpBD and CH₃I-TpBD, the band gaps are found to be 2.24 and 2.19 eV, and it is calculated to be 2.13 and 2.17 eV for TpEDDA and CH₃I-TpEDDA, respectively (Fig. 4a–c and Supplementary Figs. S44–S49). Subsequently, Mott-Schottky measurements are performed to further elucidate their energy band structures. All samples exhibit n-type semiconductor behavior, as evidenced by the positive slopes observed in their curves. According to the value of intersection at the x-axis, the flat-band potentials (E_fb vs. Ag/AgCl) of TpPa-1, TpBD, and TpEDDA are determined to be − 0.73, − 0.57, and − 0.84 eV, respectively (insert images of Fig. 4b, c and Supplementary Figs. S50, S52). According to the equation E(NHE) = E(Ag/AgCl) + 0.2 eV, the normal hydrogen electrode (NHE) potentials of TpPa-1, TpBD, and TpEDDA are estimated to be − 0.53, − 0.37 and − 0.64 V, respectively. It is widely accepted that for n-type semiconductors, the lowest unoccupied molecular orbital (LUMO) is typically around 0.2 eV above the E_fb⁵⁵. Consequently, the LUMO energies of TpPa-1, TpBD, and TpEDDA are estimated to be − 0.73, − 0.57, and − 0.84 eV, respectively. In comparison with pristine COFs, intriguing patterns in experimentation can be observed that after CH₃I treatment, all the modified samples exhibit a more negative LUMO energy to be − 0.92, − 0.63, and − 0.90 eV, indicating an enhancement in their reduction potential, which would be more favorable for H⁺ reduction (Fig. 4c, d and Supplementary Figs. S51 and S53). This phenomenon may be attributed to the electron-withdrawing nature of the iodide anion in their framework, which can stabilize the LUMO and lower its energy. Moreover, in conjugation with the band gaps results, their highest occupied molecular orbitals (HOMO) can also be easily obtained (Fig. 4d). It is well acknowledged that a built-in electric field refers to a generalized electric field existing within a material. It is primarily formed due to charge redistribution caused by spontaneous electron flow at material interfaces or by charge movement within the material under the influence of a weak internal electric field. The built-in electric field (BIEF) extends from regions with low charge density to regions with high charge density within the material. In this study, the enhancement of the BIEF is mainly attributed to the iodide-induced ionic polarization effect. This effect strengthens the non-uniform charge distribution within the COF framework, thereby significantly enhancing the BIEF. To illustrate the effectiveness of quaternization modification in polarizing the COFs and establishing a stronger BIEF in their framework, the charge density distribution of the TpPa-1 and CH₃I-TpPa-1 are first studied through density functional theory (DFT) method^47,56–59. According to the results of electrostatic potential distribution (ESP), CH₃I-TpPa-1 (1.35 Debye) exhibits a larger dipole moment compared to TpPa-1 (0.46 Debye), along with a more significant potential difference (Fig. 4e and Supplementary Fig. S54, S55). The electrostatic potential of the iodide ion unit is significantly higher than that of the Tp unit, highlighting an uneven electron distribution. This imbalance arises from the induced polarization effect, which is caused by electron transfer between the iodide ion unit and the Tp unit. Consequently, the BIEF is established between these two units, oriented from the Tp unit toward the iodide ion unit³⁷. Moreover, the BIEF of both materials is quantitatively measured through the kelvin probe force microscopy (KPFM) tests^60,61. As shown in Fig. 4f, g, CH₃I-TpPa-1 has a significantly higher surface potential (ΔE = 281 mV) than that of TpPa-1 (ΔE = 51 mV). Utilizing the method reported by Kanata et al.⁶², it is confirmed that CH₃I-TpPa-1 exhibits a 6.4-fold enhancement in the intensity of its BIEF compared with TpPa-1 (Fig. 4h), as calculated by combining the surface charge density and the surface voltage results. All these results consistently indicate that ionic polarization has effectively enhanced the BIEF of the COFs.

Fig. 4. Electronic characterizations of COFs samples.

a UV–vis diffuse reflectance spectra of TpPa-1 and CH₃I-TpPa-1. b, c Tauc plot and Mott–Schottky plots of TpPa-1 and CH₃I-TpPa-1. d the band structure of series β-keto-enamine-based COFs and modified samples. e The calculated electrostatic potential maps and the corresponding diagrams for TpPa-1 and CH₃I-TpPa-1. f, g Surface potentials of TpPa-1 and CH₃I-TpPa-1 tested by the Kelvin probe force microscopy (KPFM). h A comparison of BIEF intensity for TpPa-1 and CH₃I-TpPa-1.

To identify the important role of ionic polarization in promoting the electron-hole pairs migration/separation efficiency, several photoelectrochemical experiments are additionally conducted. The photocurrent response of the CH₃I-TpPa-1 is much stronger than that of the TpPa-1, indicating a more charge transfer/separation efficiency (Fig. 5a). The Nyquist curve of CH₃I-TpPa-1 exhibits a significantly smaller semicircle diameter compared to TpPa-1, demonstrating a significant improvement in electron transfer efficiency (Fig. 5b). Similar results can also be found in other two pairs of COFs (Supplementary Figs. S56–S59), suggesting the universality of the ionic polarization method. Surface photogenerated voltage (SPV) spectra can also be employed to identify the charge transfer property. As shown in Fig. 5c, the signal of CH₃I-TpPa-1 reaches 80 μV, far surpassing that of TpPa-1, which is only 8.9 μV. This significant disparity implies that CH₃I-TpPa-1 with a stronger SPV signal has a higher charge separation efficiency. All these findings strongly suggest that the enhanced BIEF associated with ionic polarization can make great contributions to promoting charge migration and separation. The linear sweep voltammetry (LSV) curves clearly show that the electrocatalytic H₂ evolution potential of CH₃I-TpPa-1 is dramatically lower than that of TpPa-1, implying an enhanced reduction capability for facilitating the H₂ evolution reaction after ionic polarization (Fig. 5d). The photoluminescence (PL) and time-resolved photoluminescence (TRPL) spectra are then used to investigate the separation and dynamic behavior of photogenerated charges. Under excitation at 362 nm, the fluorescence intensity of the pristine TpPa-1 is higher than that of the CH₃I-TpPa-1, which can be attributed to the role of BIEF in enhancing carrier separation (Fig. 5e). Similar trends are observed in other two pairs COFs (Supplementary Figs. S60–S63). Based on the TRPL results, the average lifetime of CH₃I-TpPa-1 is 3.47 ns, while that of TpPa-1 is 1.21 ns, indicating an extended lifetime for photo-induced charges in CH₃I-TpPa-1 (Fig. 5f). Excitons dissociation efficiency is another decisive factor for COFs photocatalysts^63,64. Thus, temperature-dependent PL measurements are also conducted at different temperatures ranging from 110 to 265 K. The PL intensity of both COFs decreases as the temperature increases, primarily due to the thermally activated nonradiative recombination process. The exciton binding energies E_b of TpPa-1 and CH₃I-TpPa-1 are estimated to be 75.3 and 63.9 meV, respectively (Fig. 5g, h). This suggests that the excitons generated in CH₃I-TpPa-1 are more likely to undergo dissociation^65,66, highlighting the beneficial effect of ionic polarization in lowering the exciton dissociation energy for COFs. Moreover, contact angle tests are also performed to detect possible changes in the hydrophilicity of materials, and the results indicate that the three modified COFs have superior hydrophilicity, which will improve the H⁺ transfer efficiency and promote the H⁺ reduction process (Supplementary Figs. S64–S66). The formation of H* is typically recognized as the rate-determining step for H⁺ reduction⁶⁷, while Pt is well recognized as an active site for this process due to favorable adsorption and desorption. Here, we verify our experimental results by computation. The results show that the Gibbs free energy (ΔG_H*) for the iodine anion as an active site in CH₃I-TpPa-1 is only 0.12 eV, much lower than the oxygen site in TpPa-1, which is 0.40 eV, implying that the iodine anion is more likely to serve as the active site for hydrogen production in the modified samples can more effectively balance the hydrogen ions adsorption and desorption processes (Supplementary Figs. S67 and S68), playing a similar function of Pt. In contrast, the higher ΔG_H* for TpPa-1 suggests limited release of hydrogen ions, thereby inhibiting photocatalytic activity (Fig. 5i).

Fig. 5. Photoelectrochemical characterization and Gibbs free energy calculations.

a The transient photocurrent response of TpPa-1 and CH₃I-TpPa-1. b EIS Nyquist images of TpPa-1 and CH₃I-TpPa-1. c SPV spectra of TpPa-1 and CH₃I-TpPa-1. d LSV plots of TpPa-1 and CH₃I-TpPa-1. e, f The photoluminescence spectra, and time-resolved fluorescence emission decay spectra of TpPa-1 and CH₃I-TpPa-1. g, h Temperature-dependent PL spectra with excitation wavelength at 362 nm and extracted exciton binding energies of TpPa-1 and CH₃I-TpPa-1. i Calculated free energy diagrams of hydrogen evolution for TpPa-1 and CH₃I-TpPa-1.

After validating through the aforementioned experimental and theoretical calculations tests, the boosted photocatalytic mechanism can be demonstrated. Through a unique post-synthetic quaternization reaction, ionic polarization can be achieved in a series of β-keto-enamine-based COFs, thus endowing them with separate charge centers comprising cationic skeleton and iodide counter-anions. Due to the different electrostatic potential values between the iodide anions unit and the cationic framework unit, a local electric field can be formed between the iodide anions unit and the cationic framework unit, with the direction from the cationic framework unit to the iodide anions unit, thus establishing robust BIEF within their framework. Benefitting from the enhanced BIEF, the charge separation and the exciton dissociation efficiency of the modified COFs can be significantly enhanced, which can be proved by a series of optical and spectroscopy tests. In addition, in pristine β-keto-enamine-based COFs, the oxygen atom can act as the active site to promote the H₂ evolution. After being modified through quaternization reaction, due to the electron-withdrawing effect, the introduced iodide anion not only serves as an additional active site to reduce the energy barrier of H* formation but also acts as an electron extractant to promote the electron-hole pairs separation/migration, confirmed by series of photoelectric and theoretical calculation tests. Therefore, under the synergistic effect, the photocatalytic activity of series modified β-keto-enamine-based COFs materials for hydrogen production is significantly enhanced.

In conclusion, we have successfully constructed a series of cationic covalent organic frameworks (COFs) incorporating iodine anions via a post-synthetic method. The ionic framework of these COFs induces ionic polarization, resulting in a stronger built-in electric field, as validated through KPFM tests and DFT calculations. Moreover, the presence of iodine ions can provide more electron-rich reduction centers, thereby improving the efficiency of electron-hole pair separation and promoting photocatalytic reactions. It is worth mentioning that the addition of Pt co-catalysts is typically required in most reported COFs photocatalysts. However, with the help of ionic polarization, the photocatalytic H₂ evolution performance of the modified samples is significantly improved and can even display commendable H₂ evolution performance without the addition of Pt co-catalysts. This is particularly evident in CH₃I-TpPa-1, where the H₂ production rate reaches as high as 9.21 mmol g⁻¹ h⁻¹. Overall, this work paves the way for addressing the issues of high exciton dissociation energy and sluggish charge transfer via rational molecule-level structure design of ion-polarized COFs. Moreover, the great potential of ion-polarized COFs as highly efficient photocatalysts and is significant for clean energy generation.

Methods

Preparation of COFs

TpPa-1 was synthesized based on previously reported methods with slight adjustments. Initially, 0.1 mmol (purity >  8%) of 1,3,5-triformylphloroglucinol (Tp) and 0.15 mmol (purity > 99.3%) of p-phenylenediamine (Pa-1) ligands were effectively dispersed in 3 mL of N, N-dimethylformamide (DMF, purity > 99.5%) utilizing ultrasound. Next, 0.5 mL of 3 M acetic acid was introduced into the tube. After achieving uniform ultrasonication of the solution, the tube was transferred to liquid nitrogen (77 K) for rapid freezing and subjected to three rounds of vacuum treatment. Then, the tube was hermetically sealed and placed inside an oven, where it was subjected to heating at 120 °C for a period of 3 days. The resulting product was obtained through centrifugation, followed by multiple washes with anhydrous tetrahydrofuran (THF, purity > 99.5%) and anhydrous acetone (purity > 99.8%). Finally, the product was dried under vacuum at 60 °C for a duration of 24 h to get TpPa-1. Using a similar approach, the replacement of p-phenylenediamine (Pa-1) with 0.15 mmol (purity > 99.5%) of 4,4’-Diaminobiphenyl (BD) or 0.15 mmol (purity > 99.7%) of 4,4’-(ethyne-1,2-diyl)dianiline (EDDA) was carried out to synthesize TPBD or TPEDDA.

TpBpy was prepared using a solvothermal synthesis method. In this process, 0.3 mmol of Tp and 0.45 mmol of (2,2’-bipyridine)-5,5’-diamine (Bpy) (purity > 98.8%) were dissolved in a solvent mixture containing 4.5 mL of N,N-dimethylacetamide (DMAc) (purity > 99.0%) and 1.5 mL of ortho-dichlorobenzene (ODCB). In addition, 0.6 mL of 6 M acetic acid was added. The resulting solution was ultrasonicated for 15 min to ensure complete dispersion and then subjected to three cycles of freeze-pump-thaw to remove dissolved gases. After degassing, the sealed reaction mixture was heated at 120 °C for 72 h. Once the reaction was complete, the product was collected by filtration and washed thoroughly with deionized water and DMAc to remove unreacted precursors and oligomeric byproducts. The final purification step involved rinsing the material with acetone, followed by vacuum drying at 60 °C for 24 h⁶⁸.

TpAzo was synthesized via a solvothermal method. Initially, 0.3 mmol of Tp, along with the corresponding diamine 4,4′-azodianiline (Azo) (purity > 99.2%), 3 mL of DMAc, and 3 mL of ODCB, were combined and sonicated for 10 min to form a uniform dispersion. The mixture was then quickly frozen in a liquid nitrogen bath (77 K) and degassed by performing three cycles of freeze-pump-thaw. After sealing the tube, it was heated at 120 °C for 3 days. The final sample was washed with acetone and then dried under a vacuum at 60 °C for 24 h.

Preparation of Iodine ion modified COFs

20 mg of the above product was added to 3 mL of DMF, followed by the addition of 1 mL of CH₃I. The mixture was stirred in the dark at room temperature for 24 hours. The resulting product was then washed twice with acetone, collected by centrifugation, and finally dried under vacuum at 60 ^oC for 4 h. The product was named CH₃I-TpPa-1. The same method was applied to obtain CH₃I-TpBD and CH₃I-TpEDDA. The product was named CH₃I-TpPa-1. The same method was applied to obtain CH₃I-TpBD, CH₃I-TpEDDA, CH₃I-TpBpy, and CH₃I-TpAzo.

Photocatalytic H₂ production rests

Photocatalytic H₂ production experiments were conducted using a full glass automatic online trace gas analysis system (Labsolar-6A, Beijing Perfectlight Technology Co., Ltd.). The light source was a 300 W Xenon lamp with a 420 nm filter (PLS-SXE300/300UV, light intensity: 100 mW·cm⁻², Beijing Perfectlight Technology Co., Ltd.). Hydrogen detection and quantitative analysis were performed using a GC7900 gas chromatography system (Shanghai Tianmei Scientific Instrument Co.) with nitrogen as the carrier gas at 30-minute intervals. The catalyst was placed in a custom quartz glass reactor, and the reaction temperature was maintained at 5 °C using a water condenser (Supplementary Fig. S69). 10 mg of catalyst was added to 100 mL of deionized water containing 100 mg of L-Ascorbic acid as a hole scavenger. The suspension was then stirred in a 200 mL custom quartz reactor. The suspension was then transferred to the custom quartz reactor for the photocatalytic reaction. Prior to each photocatalytic reaction, the system was vacuum-treated several times to remove dissolved air. The entire reaction system was irradiated with Xe light using a UV cut-off filter (λ ≥ 420 nm). The vertical distance between the Xe lamp and the reactor was approximately 3 cm. In addition, the apparent quantum efficiency (AQE) was measured using a xenon lamp equipped with bandpass filters. Photocatalytic experiments were performed with 10 mg of catalysts, using filters with wavelengths of 475, 500, 520, 550, and 600 nm, respectively.

Photoelectrochemical characterization

The electrochemical measurements were conducted using an electrochemical workstation, model CHI 760. The experiment employed a three-electrode system, where a Pt sheet served as the counter electrode and an Ag/AgCl electrode as the reference electrode (The reference electrode calibration method: First, clean the reference electrode using deionized water to remove any contaminants. Then, adjust the potential of the reference electrode via an external circuit until it reaches a stable state. Finally, allow the potential to stabilize and check again). 0.2 M Na₂SO₄ solution (pH = 7) was used as the electrolyte and prepared fresh before each use (Supplementary Fig. S70). A xenon lamp with a filter (λ ≥ 420 nm) provided the visible light source. To prepare the working electrode, 5 mg of the catalyst sample and 20 μL of Nafion solution were dissolved in 1 mL of anhydrous ethanol, sonicated for 30 min to ensure thorough mixing, and then applied onto FTO glass (1 cm × 1 cm). (The above electrochemical measurement results have not been corrected for iR compensation.)

Author contributions

M.T. and S.X. proposed the project and designed experiments. Z.J. and Hu.H. synthesized and characterized the samples and performed experiments and data analysis. L.H. and Hu.H. contributed to the analyses of KPFM results. S.X. wrote the paper. S.X., M.T. and L.X. polished the writing of the paper. Z.J. and L.X. contributed equally to this work. All authors discussed the results and commented on the paper.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

All data are included in the main text or the supplementary information. Source data are provided in this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-024-53834-w.