Recently, considerable research has been focused on chemical transformations to useful chemicals using CO₂ as a feedstock. This is because the accumulation of CO₂ in the atmosphere is accelerating at a frightening pace and is considered the leading cause of the greenhouse effect over the past few decades [1-4]. Among these transformations, the synthesis of cyclic carbonates from epoxides and CO₂ has attracted increasing attention due to their high efficiency in industrial processes [5-10]. For example, cyclic carbonates can be widely used as aprotic high-boiling polar solvents, electrolytes for lithium-ion batteries, precursors of polymeric materials, and fine chemical intermediates.

In the last decade, various catalysts have been developed for this transformation, including organocatalysts (e.g., Schiff bases [11] and ionic liquids [12-17]), metal complexes (e.g., salen complexes [18] and metalloporphyrins [19]). Among these catalysts, ionic liquids (ILs), such as ammonium [12, 13], phosphonium [14], imidazolium [15, 16], and pyridinium salts [17], have shown high catalytic activity for CO₂-epoxide cycloaddition. Regrettably, these homogeneous catalysts are not easily separated from the organic constituents of the reaction mixture, limiting their large-scale applications. In order to overcome this issue, the development of IL-based heterogeneous catalysts is considered an effective alternative solution. Accordingly, ILs have been successfully grafted to various solid supports, including silica materials [20-22], polymers [23-25], and metal-organic frameworks [26, 27].

Covalent organic frameworks (COFs) are a class of crystalline porous organic polymers composed of light-weight elements and constructed via strong covalent bonds [28-30]. Due to their unique physical and chemical properties, COFs have been widely used for gas adsorption-separation, and in the photoelectric field, catalytic field, as well as other fields [31, 32]. Especially in the catalysis field, the flexible regulation of pore parameters, including pore size, shape, pore volume, and size distribution, as well as the easy introduction of catalytically active sites to the framework of COFs render these materials promising platforms for the immobilization of catalysts in organic synthesis. In particular, one-dimensional (1D) open channels found in 2D-COFs would not only remarkably enhance the diffusion of substances, but can also be used to immobilize catalysts through the post-modification of the building blocks of COFs. As such, some reactions have been conducted using COFs as catalysts, such as the Michael addition reaction [33], Diels-Alder reaction [34], oxygen evolution reaction [35] and Heck-epoxidation tandem reaction [36]. Recently, the concept of incorporating ILs into COFs has been realized by our group and others [37-40]. COF materials can thus be used as an ideal catalyst or catalyst carrier in a heterogeneous reaction process.

To further explore the application of the COFs in heterogeneous catalysis, in this study, we have immobilized an IL, 1-alkyl-3-methylimidazolium bromide, on the microporous channel walls of H₂P-DHPh COF through a two-step modification step (Scheme 1). The H₂P-DHPh COF material was first prepared and selected in terms of its special pore structures, high surface area, and abundant surface hydroxyl groups. The as-synthesized catalyst (AMIMBr@H₂P-DHPh) was characterized and evaluated for the coupling of CO₂ and epoxides. Moreover, the recyclability of our catalyst system was examined for a selected coupling reaction between CO₂ and epichlorohydrin.

All the reagents were used as received without further purification. Pyrrole, 4-nitrobenzaldehyde, 1, 4-dimethoxybenzene, deuterium chloride, 1, 2-dichlorobenzene (o-DCB), and 1-butanol (n-BuOH) were purchased from TCI Chemicals. Hexamethylenetetramine, potassium carbonate, acetic acid, acetic anhydride, pyridine, 1, 4-dioxane, formaldehyde, hydrochloric acid, sulfuric acid, chloroform, acetone, tetrahydrofuran (THF), dichloromethane, ethyl acetate, hydrobromic acid, magnesium sulfate, methanol, ammonia solution, 1, 4-dibromobutane, propionic acid, Tin(Ⅱ) chloride, and 1-methylimidazole were purchased from Sinopharm Chemical Reagent Co.

Synthesis of building blocks. 5, 10, 15, 20-Tetrakis(p- tetraphenylamino)porphyrin (H₂P) and 2, 5- dihydroxyterephthalaldehyde (DHTA) were prepared according to reported procedures [41, 42].

Synthesis of H₂P-DHPh COF.H₂P (27.0 mg, 0.04 mmol), and DHTA (13.5 mg, 0.08 mmol) were placed in a glass ampule vessel (10 mL), followed by the addition of a solution of o-DCB/n-BuOH/3 M acetic acid (5/5/1 by volume; 2.2 mL) [42]. The mixture was sonicated for 10 min, then flash-frozen in liquid nitrogen, and degassed by three freeze-pump-thaw cycles. The tube was flame-sealed, then heated at 120 ℃ for 3 d. The resulting precipitate was filtered out, washed sequentially with THF (five times) and acetone (three times) to give a purple colored powder, which was dried at 120 ℃ under vacuum overnight to give the desired product in 70% yield.

Synthesis of AMIMBr@H₂P-DHPh COF.H₂P-DHPh COF (40 mg) and potassium carbonate (118.4 mg) were weighed into a 50-mL glass vial, to which dehydrated THF (10 mL) was added. The mixture was heated at reflux. To the hot mixture, 1, 4-dibromobutane (1 mL) was added, and the mixture was maintained at reflux, and stirred overnight in nitrogen atmosphere. After cooling the mixture to room temperature, the precipitate of the product was collected by filtration, washed with water and acetone. The powder was dried overnight under vacuum at 80 ℃ to give the corresponding product of BuBr@H₂P-DHPh COF, quantitatively.

The obtained BuBr@H₂P-DHPh COF was stirred in 1-methylimidazole (5 mL) at 80 ℃ under nitrogen protection for 4 h. After cooling down to room temperature, the solvent was removed. The residue was washed thoroughly with water and acetone to get the targeted product AMIMBr@H₂P-DHPh COF.

Hydrolysis of AMIMBr@H₂P-DHPh COF.The AMIMBr@H₂P- DHPh COF (2 mg) sample was hydrolyzed by a DMSO-d₆/DCl (4/1 by vol., 1 mL) solution [43]. The mixture was sonicated for over 15 min to obtain a homogenous solution. The content of 1-alkyl-3-methylimidazole was calculated from the proton integrates of its ¹H NMR spectrum.

All the reactions were carried out in a 60-mL stainless steel reactor with a glass-lined tank. The mixture of epoxides (7.5 mmol) and AIMIBr@H₂P-DHPh COF (10 mg) was added to the reactor without solvent. The reactor was pressurized with CO₂ (1.0 MPa), and the temperature was raised to 120 ℃. The reaction was conducted for 24 h. After the reaction, the reactor was cooled down in ice-cold water. Unreacted CO₂ was vented out, and the catalyst was separated by centrifugation. The products were analyzed by gas chromatography (Agilent HP 6890 A) using a capillary column (HP-5, 30 m × 0.25 mm) and a flame ionization detector. The product yield was determined by using toluene as an internal standard.

Fourier-transform infrared (FTIR) spectroscopy was carried out on a Bruker spectrophotometer (Model TENSOR27) with powder-pressed KBr pellets. Powder X-ray diffraction (PXRD) analysis was carried out on a Rigaku RINT D/Max 2500 powder diffraction system, using a Cu Kα radiation (λ = 1.5432 Å). Thermogravimetric analysis (TGA, STA449F3, NETZSCH, Germany) was performed by heating from room temperature to above 800 ℃ at a rate of 10 ℃ min^-1 and a N₂ flow rate of 20 mL min^-1. The nitrogen physisorption experiment was conducted at 77 K on a QUADRASORB SI gas sorption system (Quantachrome Instruments), which was degassed at 120 ℃ in vacuum before testing. The specific surface areas were calculated by the Brunauer-Emmett-Teller (BET) method. The pore size distribution was evaluated by the nonlocal density functional theory (NLDFT) method. Gas chromatography (GC, Agilent 7890A) equipped with a capillary column (HP-5, 30 m × 0.25 mm) using a flame ionization detector. ¹H and ¹³C nuclear magnetic resonance (NMR) spectra were recorded on a Bruker Advance Ⅲ 400-MHz NMR spectrometer (Bruker BioSpin Corporation, F llanden, Switzerland).

Herein we designed and prepared a two-dimensional cationic covalent organic framework by a post-synthesis strategy. A porous imine-linked porphyrin COF, H₂P-DHPh COF, was utilized as the precursor, which was initially synthesized by a two-component condensation system. AMIMBr@H₂P-DHPh COF was synthesized through three main steps: (1) preparation of the H₂P-DHPh COF according to a known procedure described by Jiang et al. [42]; (2) post-modification of H₂P-DHPh COF with excess 1, 4-dibromobutane to obtain an alkyl halide-grafted BuBr@H₂P-DHPh COF through an etherification reaction that used potassium carbonate as a neutralizing reagent; (3) ionization of the BuBr@H₂P-DHPh COF to AMIMBr@H₂P-DHPh COF using 1-methylimidazole through a quaternization reaction. The general pathway is represented in Scheme 1.

The crystalline structures and unit cell parameters of H₂P-DHPh COF were first determined by PXRD analysis in combination with structural simulations. Diffraction peaks of H₂P-DHPh COF were observed at 2θ = 3.5°, 7.1°, and 22.9°, which were attributed to the (100), (200), and (001) facets, respectively (Fig. 1(a), black curve). The use of lattice modeling and Pawley refinement processes led to an eclipsed AA stacking model that could reproduce the PXRD results with regards to the peak position and intensity (Fig. 1(a), green curve). The Pawley refinement (Fig. 1(a), red curve) using a unit cell of α = γ = 90°, β = 89.98°, a = 24.53 Å, b = 3.80 Å, and c = 24.52 Å confirmed the peak assignment, as evidenced by their negligible difference (Fig. 1(a), blue curve). The pseudo-Voigt function was used for whole profile fitting, and the Berar-Baldinozzi method was used for asymmetry correction during the refinement processes. The final Rwp and Rp values were 4.63% and 3.62%, respectively, for the H₂P-DHPh COF. Clearly, the crystalline structure of H₂P-DHPh COF was the same as was reported by Jiang et al. [42].

After grafting 1, 4-dibromobutane and 1-methylimidazole successively onto the channel walls of H₂P-DHPh COF, all the reflection peak positions for the modified samples BuBr@H₂P-DHPh COF and AMIMBr@H₂P-DHPh COF were similar to those for the as-synthesized H₂P-DHPh COF, indicating that they possessed similar crystal structures (Fig. 1(b)). A similar situation was also observed in our previous study [39]. When BuBr@H₂P-DHPh COF was further ionized through the quaternization of 1-methylimidazole, the reflection peaks were significantly decreased, suggesting the occupation of the pores by the grafted IL. We do not exclude the possibility of undesirable damage to the crystal structure of H₂P-DHPh COF during the post-modification and ionization processes [44].

The immobilization of 1, 4-dibromobutane and further 1-methylimidazole on H₂P-DHPh COF has been characterized by FT-IR spectroscopy, N₂ adsorption-desorption, and TGA techniques. Fig. 2 shows the FT-IR spectra of H₂P-DHPh COF, BuBr@H₂P-DHPh COF, and AMIMBr@H₂P-DHPh COF. The appearance of an intensive band around 1620 cm⁻¹ in H₂P-DHPh COF confirmed the formation of imine bonds. After post-modification and ionization, the C=N bands were retained, suggesting that the covalent connectivity of the frameworks was maintained. A weak peak also appeared at 660 cm⁻¹ in the FT-IR spectrum of BuBr@H₂P-DHPh COF, corresponding to C-Br stretching vibrations of 1, 4-dibromobutane, indicating the success of the etherification reaction. Furthermore, a peak at 1556 cm⁻¹ was also observed, which can be assigned to the C=C band of the 1-methylimidazole group, revealing that the 1-methylimidazole groups were successfully grafted onto the channel walls of the BuBr@H₂P-DHPh COF.

The porous properties of these COFs were determined from the nitrogen adsorption-desorption isotherms recorded at 77 K. As shown in Fig. 3(a), these COF materials exhibited the typical type-Ⅳ isotherms with an H4 hysteresis loop. The BET surface areas were measured to be 1123, 856, and 513 m² g⁻¹ for H₂P-DHPh COF, BuBr@H₂P-DHPh COF, and AMIMBr@H₂P-DHPh COF, while their corresponding total pore volumes were 0.91, 0.80, and 0.62 cm³ g⁻¹. The pore size distributions of the different samples were evaluated by DFT. The pore diameter gradually decreased from 2.0 to 1.3 nm when H₂P-DHPh COF was modified through 1, 4-dibromobutane and 1-methylimidazole, successively; this result is consistent with that of the PXRD patterns. Despite the occupation of the pores by the grafted IL, AMIMBr@H₂P-DHPh COF maintained a porosity of up to 513 m² g⁻¹. The porosity data of the COFs are summarized in Table 1. It needs to be clarified that the intensity of the diffraction peaks of AMIMBr@H₂P-DHPh COF was remarkably decreased when 1-methylimidazole was introduced into the channels of BuBr@H₂P-DHPh COF. The reason is that the COFs are comprised of light elements, like C, H, O, and N. All these light elements have weak scattering powers, leading to a low-resolution XRD pattern. Therefore, we consider that the ordered framework structure remained intact to a certain extent, even though the H₂P-DHPh COF was successively modified twice. The thermal behavior of these COFs was studied by thermogravimetric analysis (TGA). TGA showed good thermal stability for these as-prepared materials, which survived up to 300 ℃ without decomposing. The high porosity and good thermal stability of the COFs mean that these COFs can be used as catalysts or catalyst carriers.

The content of the 1-methylimidazole groups grafted onto the channel walls of BuBr@H₂P-DHPh COF was analyzed by ¹H NMR spectroscopy. The AMIMBr@H₂P-DHPh COF sample was digested with DCl (6 M) in DMSO-d₆. We found that about 16 mol% 1-methylimidazole was grafted onto the channel walls of the COF, i.e. the content of 1-methylimidazole on the AMIMBr@H₂P-DHPh COF was calculated to be 4.9 wt%. It is worth noting that an excessive ionization of the pores will lead to the exfoliation of the nanosheets of 2D-COFs [45]; therefore, the reaction time was limited to 4 h.

The CO₂ adsorption capability of the COFs was first investigated in this work. The CO₂ adsorption isotherms of H₂P-DHPh COF and AMIMBr@H₂P-DHPh COF were measured at pressures of up to 1 bar at 0 and 25 ℃, respectively. AMIMBr@H₂P-DHPh COF exhibited a relatively higher adsorption capacity than that of H₂P-DHPh COF at 0 ℃, although the specific surface area of AMIMBr@H₂P-DHPh COF was remarkably reduced after modification because the decreased pore size would enhance the interaction between CO₂ guest molecules and the pore surface of the COF [46]. The introduction of highly charged groups on the surface of porous materials can afford high affinities for CO₂ due to the high quadrupole moment of CO₂ [46]. H₂P-DHPh COF and AMIMBr@H₂P-DHPh COF exhibited CO₂ uptake capacities of 31.9, and 31.4 cm³ g^–1 (approximately 6.3 and 6.2 wt%) at 0 ℃ and 1 bar, which decreased to 23.9 and 22.0 cm³ g^–1 (approximately 4.3 and 4.0 wt%), respectively, when the temperature was increased to 25 ℃, 1 bar (Fig. 5(a)). The CO₂ adsorption isotherms of the COFs were calculated to be 35.8 kJ mol^-1 for AMIMBr@H₂P-DHPh COF and 25.5 kJ mol^‒1 for H₂P-DHPh COF (Fig. 5(b)), confirming that there are stronger interactions between CO₂ and the pore surface of AMIMBr@H₂P-DHPh COF than the pore surface of H₂P-DHPh COF. The high CO₂ adsorption capability of the COF makes it possible for AMIMBr@H₂P-DHPh COF to exhibit excellent catalytic performance for the cycloaddition of CO₂ to epoxides.

To test the catalytic performance of the obtained COF catalyst, the cycloaddition of CO₂ to epoxides was chosen as the probe reaction in this study. The reaction conditions were carried out at 120 ℃, 1.0 MPa CO₂, and a reaction time of 24 h. The experimental results are shown in Table 2. In the absence of the catalyst, a conversion of 38% of the epichlorohydrin to the cyclic carbonate can be obtained under the above reaction conditions (Table 2, entry 1). A similar result (a conversion of 40%) was obtained when the same experiment was repeated with H₂P-DHPh COF as a heterogeneous catalyst (Table 2, entry 2). However, a conversion of 91% was observed when functionalized AMIMBr@H₂P-DHPh COF was used as the catalyst (Table 2, entry 3), indicating the outstanding catalytic performance of AMIMBr@H₂P-DHPh COF.

The catalytic activity of AMIMBr@H₂P-DHPh COF in the cycloaddition of CO₂ to different epoxides under identical conditions has been examined. The epoxides used for this purpose include propylene oxide, 1, 2-epoxyhexane, 1, 2-epoxyoctane, butyl glycidyl ether, and styrene oxide (Table 2, entries 4-8). The catalytic ability of AMIMBr@H₂P-DHPh COF was more effective for the smaller epoxides (such as epichlorohydrin and propylene oxide with 91% and 95% conversions, respectively) than the epoxides with longer alkyl chains, such as 1, 2-epoxyoctane (16% conversion) and butyl glycidyl ether (21% conversion). These results could be ascribed to the limited space in the channels of AMIMBr@H₂P-DHPh COF, which gave an obvious size-selectivity towards epoxides.

The reusability of the as-synthesized AMIMBr@H₂P-DHPh COF and the reproducibility of its catalytic performance has been investigated based on the experimental results of constant cyclic tests. In each cycle, AMIMBr@H₂P-DHPh COF was removed by centrifugation and then rinsed with epichlorohydrin. After drying, the catalyst was reused for the next run. The yields of cyclic carbonate in the first five consecutive runs (Fig. 6) indicated that the catalytic activity of AMIMBr@H₂P-DHPh COF could be retained for up to five cycloaddition series. After five runs, the BET surface area of AMIMBr@H₂P-DHPh COF was measured to be 386 m² g⁻¹, suggesting that the high porosity nature of the COF catalyst remained. This result reveals that the 1-alkyl-3-methylimidazolium IL-modified COF can thus be considered a renewable and stable catalyst. The heterogeneous nature of the COF catalyst was verified via a leaching test. In a typical experiment, the supernatant after one cycle of the catalytic reaction was isolated and mixed with fresh reactants for another cycle of the cycloaddition reaction without the heterogeneous catalyst. Expectedly, no target product was obtained under such conditions.

The data obtained from this study were compared to the activities of several other reported catalysts for the same reaction (Table 3). According to the data listed in Table 3, we can regard AMIMBr@H₂P-DHPh COF as a prominent heterogeneous catalyst for the cycloaddition reaction because this synthesis can be carried out under relatively mild conditions compared to other catalysts, without a co-catalyst like a quaternary ammonium salt.

In summary, the ionic AMIMBr@H₂P-DHPh COF was successfully synthesized by grafting IL onto the channel walls of H₂P-DHPh COF and used as a catalyst for the synthesis of cyclic carbonates from epoxides and CO₂ under mild reaction conditions. The results show that this ionized porous COF material showed good catalytic activity even in a solvent-free and co-catalyst-free environment. The H₂P-DHPh COF with the immobilized IL significantly improved the catalytic activity as compared to the pure H₂P-DHPh COF. In addition, AMIMBr@H₂P-DHPh COF showed apparent size-selectivity towards epoxides due to the limited space of the channels after a two-step modification. Furthermore, the AMIMBr@H₂P-DHPh COF has been successfully recycled five times after easy separation, without deteriorating its activity and selectivity under equivalent reaction conditions.