Covalent organic frameworks (COFs) are a series of crystalline porous materials composed of light elements (e.g., C, H, N, B, O) [1-4]. COFs have been extensively studied in recent years because of their potential applications in various areas, such as gas storage/separation [5-9], sensing [10, 11], and energy conversion [12-19]. Moreover, flexible regulation of the pores and easy introduction of functional active sites onto the skeleton of COFs make them promising platforms for many catalysis applications. Thus, many efforts have been focused on either post-synthetic modification or a bottom-up strategy that was shown to be successful in the introduction of catalytic active sites [20-27]. In contrast to post-synthetic modification, the bottom-up strategy generally requires a tedious solvothermal condition, particularly if a bulky catalytic site is attached to building units, it will be difficult to obtain crystalline COFs. Post-synthetic modification is relatively easy and simple, but this strategy generally endows the resulting COFs with only a certain function. To optimize a COF for use in a wide variety of different applications, it is important to be able to tailor its functionality in a straightforward fashion.

Nanosized pores in ionic architectures exhibit unique properties in terms of adsorption and separation of gas molecules [28, 29]. This charged nature also enables ionic materials to permanently incorporate, through ion exchange, other extra-framework counter ions, thus making it possible to obtain special properties, such as those useful in chemical sensing, photonics, and catalysis, to realize many unconventional applications [30]. Ionic COFs (I-COFs) have recently been reported, and these ionic frameworks exhibit exceptional proton/ion conduction, electrolyte performance, and removal of pollutants [31-34]. However, little research has been done on the utilization of I-COFs for catalysis by far. It is known that many positively charged species, including metal ions and coordinated metal complexes, exhibited excellent catalytic performance, and their high catalytic activity remained when they were loaded within the ionic framework as counter ions. Therefore, we expect that I-COFs can be used as versatile catalysts that maintain the high catalytic activity of either metal ions or coordinated metal complexes. Moreover, the high porosity, designable topology, and easy modification of I-COFs make them ideal materials for catalysis applications.

Herein, we report the synthesis of a negatively charged I-COF through a post-synthetic modification strategy. A chemically stable imine-based COF, DhaTab, was first constructed by the condensation reaction of 2, 5-dihydroxyterephthalaldehyde (Dha) and 1, 3, 5-tris(4-aminophenyl)benzene (Tab) [35]. The ring-opening reaction of 1, 3-propane sultone with phenolic hydroxyl on the skeleton of DhaTab affords a sulfoacid-based COF material. In the presence of a dilute NaOH solution, two different counter ions, metallic Mn²⁺ and a coordination complex of manganese(Ⅱ) bipyridine complexes ([Mn(bpy)₂]²⁺), were successfully incorporated via a simple ion exchange process (Scheme 1). The resulting I-COFs, as heterogeneous catalysts, exhibited excellent catalytic activities in epoxidation reactions. The proposed I-COF strategy provides a promising platform for the development of versatile materials for catalysis.

All starting materials and solvents, unless otherwise specified, were obtained from commercial sources and used without further purification. Dha was synthesized according to a previously published procedure [36]. All reactions were performed under ambient laboratory conditions, and no precautions were taken to exclude oxygen or atmospheric moisture, unless otherwise specified.

A pyrex tube was charged with Tab (0.16 mmol, 56 mg) and Dha (0.24 mmol, 32 mg), 1.0 mL of ortho-dichlorobenzene (o-DCB), 1.0 mL of n-butanol, and 0.2 mL of 6 mol L^‒1 acetic acid. This mixture was sonicated for 5 min and then flash-frozen at ‒196 ℃ (liquid N₂ bath) and degassed by three freeze-pump-thaw cycles. The tube was sealed off and then heated at 120 ℃ for 3 d. A yellowish precipitate appeared and was collected by filtration. The solid was then thoroughly washed with dimethylacetamide (DMAc), water, and then ethanol. The collected powder was solvent-exchanged with ethanol three times and then dried at 100 ℃ under vacuum overnight to give a yellowish powder in ca. 80% (38 mg) isolated yield. Elemental analysis (%) calcd. for C₆₀H₄₈N₆O₆: C (75.92), H (5.06), N (8.86); found C (73.36), H (4.51), N (7.69).

To 10.0 mL of a toluene suspension of DhaTab (10 mg) was added 5.0 mL of 1, 3-propane sultone. The reaction was refluxed for 6 h, and the solid was filtered and then thoroughly washed with DMAc, then water, and then ethanol. The solid was dried at 100 ℃ under vacuum overnight to give a deep red powder in 81% yield.

To 10.0 mL of a methanol suspension of [SO₃H]-DhaTab COF (10 mg) was added 10.0 mL of aqueous NaOH 1 mol L^‒1. The suspension was stirred for 2 h, and the solid was filtered and washed with methanol three times. Next, the resulting [SO₃Na]-DhaTab COF was added to 10.0 mL of a methanol solution containing 2 mg of Mn(OAc)₂. The mixture was stirred at 80 ℃ for 12 h, and the solid was filtered and thoroughly washed with DMAc, then water, and then ethanol. The solid was dried at 100 ℃ under vacuum overnight to give a deep red powder in 92% yield.

To 10.0 mL of a toluene suspension of [SO₃Mn]-DhaTab (10 mg) was added 2 mg of bipyridine. The reaction was refluxed at 120 ℃ for 12 h. The solid was thoroughly washed with toluene, then DMAc, then water, and then ethanol. The solid was dried at 100 ℃ under vacuum overnight to give a deep red powder in 87% yield.

To take an example, [SO₃Mn]-DhaTab was selected as a catalyst for the epoxidation reaction. For the [SO₃Mn]-DhaTab- catalyzed epoxidation of stilbene to trans-stilbene oxide, stilbene (1.0 mmol), indole-3-butyric acid (IBA, 3.0 mmol), and [SO₃Mn]-DhaTab (15 mg, containing about 0.005 mmol of Mn) in CH₃CN (1.0 mL) was stirred at room temperature for 12 h in air. The product was detected by ¹H nuclear magnetic resonance (NMR) using CH₂Cl₂ as an inter-standard.

Elemental analysis was performed using an organic elemental analyzer (vario MACRO cube, Elementar, Germany). Inductively coupled plasma optical emission spectroscopy (ICP-OES) was conducted using an ICP-OES 7300DV apparatus (PerkinElmer). The sample was first calcined at 1000 ℃ in air for 12 h to burn out organic moieties. The residue was dissolved in aqua regia and then diluted by water for ICP-OES testing. Fourier transform infrared (FTIR) measurements were 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 Cu K_α radiation (λ = 1.5432 Å ). Thermogravimetric analysis (TGA, STA449F3, NETZSCH, Germany) was performed from room temperature to above 750 ℃ at a heating rate of 10 ℃ min⁻¹ and a N₂ flow rate of 20.0 mL min⁻¹. A nitrogen physisorption experiment was conducted at ‒196 ℃ on a QUADRASORB SI gas sorption system (Quantachrome Instruments), which was degassed at 120 ℃ under vacuum before testing. Brunauer-Emmett-Teller (BET) analysis was used to determine the specific surface areas (m² g⁻¹) using desorption branches over 0.003-0.051 (P/P₀). The micropore volumes (V_p, cm³ g⁻¹) were determined using the Dubinin-Radushkevich model of nitrogen isotherms across the region of 0.01 < P/P₀ < 0.23. In all the isotherm plots, filled circles indicate adsorption data points, and open circles represent desorption data points. The pore size distribution of all the COFs was calculated from the adsorption isotherms by the nonlocal density functional theory (NLDFT) method using the Ar-zeolite/silica cylindrical pores at ‒186 ℃ kernel (applicable pore diameters, 3.5-1000 Å ) as implemented in the AUTOSORB iQwin data reduction software (version 3.01). X-ray photoelectron spectroscopy (XPS) was performed using an ESCALAB 250Xi spectrometer equipped with Al K_α radiation (1486.6 eV, 200 W) on sample powder-pressed pellets. ¹H and ¹³C NMR spectra were recorded by a Bruker Advance Ⅲ 400 MHz NMR spectrometer (Bruker BioSpin Corporation, F llanden, Switzerland).

The formation of DhaTab was confirmed by FTIR spectroscopy, where a characteristic peak of imine appeared at 1612 cm⁻¹ (Fig. 1). The crystallinity of DhaTab was determined from PXRD data (Fig. 2). Diffraction peaks appeared at 2θ = 2.8°, 4.9°, 5.6°, 7.4°, 9.8°, and 26.3°, which correspond to the (100), (110), (200), (120), (220), and (001) facets, respectively. The use of the lattice modeling and Pawley refinement processes produced an eclipsed AA stacking model that could reproduce the PXRD results in terms of the peak position and intensity. In contrast, an alternatively staggered AB model did not match the observed data (not shown here). The unit cell was created with a P3 space group of a = b = 36.2 Å , c = 3.4 Å , and α = β = 90°, γ = 120°. After the alkyl sulfonic acid group was linked, the modified COF, [SO₃H]-DhaTab, exhibited an XRD pattern similar to that of DhaTab, indicating that it possesses a similar crystal structure. After ion exchange, the pristine crystal structure of DhaTab remained (Fig. 3). Successful introduction of the alkyl sulfonic acid group on the skeleton was confirmed by FTIR spectroscopy, where the characteristic stretching peaks of -SO₃H appeared at 2600, 1348, 1158, and 1035 cm⁻¹ (Fig. 1). TGA revealed that both DhaTab and [SO₃H]-DhaTab exhibited high thermal stability with a decomposition temperature above 300 ℃ (Fig. 4). The excellent chemical stability of DhaTab has been confirmed by Banerjee's group [35]. The good thermostability and chemical stability revealed that the COFs could be used as catalysts or catalyst carriers.

The pore properties of DhaTab and [SO₃Mn]-DhaTab were determined by nitrogen sorption isotherms measured at ‒196 ℃. DhaTab exhibited a typical type-Ⅳ isotherm characteristic of mesoporous materials, whereas [SO₃Mn]-DhaTab exhibited a type-Ⅰ isotherm (Fig. 5(a)), which is characteristic of microporous materials. The BET surface areas were calculated to be 1312 and 477 m² g⁻¹, respectively. The pores were estimated to be 3.4 and 1.6 nm in diameter, respectively, using the NLDFT (Fig. 5(b)). The decrease in the BET surface area and pore size again indicates successful modification of DhaTab. The Mn²⁺ content of [SO₃Mn]-DhaTab was determined to be 1.9 wt% by ICP-AES, which is lower than the theoretical content (8.3 wt%) owing to incomplete conversion in the etherification reaction of 1, 3-propane sultone with the phenolic hydroxyl of DhaTab. We intended to control the grafting degree of the alkyl sulfonic acid group by controlling the reaction time because full grafting will lead to destruction of the crystal structure. XPS was performed to determine the valence state of the exchanged manganese (Fig. 6). In comparison with those of manganese acetate [Mn(OAc)₂], both the Mn 2p_1/2 and Mn 2p_3/2 signals of [SO₃Mn]-DhaTab shifted to slightly higher energy values owing to the stronger bonding between Mn²⁺ and -SO₃⁻. No valence change was observed for Mn²⁺ when it was introduced into the frameworks.

The I-COF with Mn²⁺ exchanged into the frameworks was explored for use as the heterogeneous catalyst for epoxidation of different olefins to epoxides. The epoxidation of stilbene was first used as a model reaction. A high conversion of 99% with a yield of 99% was observed (Table 1, entry 1), showing that [SO₃Mn]-DhaTab is highly active in this reaction. To further judge the highly catalytic activity of [SO₃Mn]-DhaTab, the epoxidation of styrene was also investigated. The conversion of styrene was 99%, and the yield was 74% (Table 1, entry 4), which is higher than that for the reaction without the catalyst and for the experiment in which Mn(OAc)₂ and [SO₃H]-DhaTab were used as catalysts (Table 1, entries 5 and 6).

Evidently, the conversion in this heterogeneous catalytic system is even higher than that in the homogeneous counterpart. The heterogeneity of [SO₃Mn]-DhaTab was confirmed by hot filtration of the catalysts after 30 min of reaction, which resulted in negligible additional yield of the product up to 18 h after filtration (Fig. 7). This indicated that [SO₃Mn]-DhaTab was a heterogeneous catalyst, and no catalytically active species were released into solution. The recyclability test of [SO₃Mn]-DhaTab clearly shows that it can be easily isolated from the reaction suspension by filtration and can be reused without significant loss of activity in the third run (Table 1, entries 2 and 3). The crystalline structure of the scaffold of the COF was found to be well preserved after three cycles (Fig. 8(a)), and the FTIR spectrum (Fig. 8(b)) of [SO₃Mn]-DhaTab after the third run showed almost the same patterns as the pristine one, indicating the high stability of this catalyst in the catalytic environment. We also explored a range of substrates for this oxidation reaction. The conversions of cyclohexene, cyclooctene, and 1-hexene were found to be 84%, 82%, and 80%, with corresponding yields of 83%, 80%, and 62%, respectively (entries 7-9). These results further indicate the high reactive activity of [SO₃Mn]-DhaTab in the epoxidation of olefins.

Ion exchange in [SO₃H]-DhaTab can in principle be carried out with a range of cations. Inspired by several significant studies where a known "ship-in-the bottle" approach was used to immobilize transition metal complexes ([Mn(bpy)₂]²⁺), leading to improved catalytic stability and recyclability [37-40], we used a similar procedure for immobilizing [Mn(bpy)₂]²⁺ into the pores of [SO₃Mn]-DhaTab through direct complexation of Mn²⁺ in [SO₃Mn]-DhaTab and bipyridyl [40]. The epoxidation of trans-stilbene with [SO₃Mn(bpy)₂]-DhaTab as the catalyst was investigated. It was found that within only 1 h, complete conversion was accomplished, with a selectivity of 99% for the formation of 2, 3-diphenylethylene oxide (Table 1, entry 10). The high reactive activity of [Mn(bpy)₂]²⁺ was well preserved in the negatively charged I-COF.

In summary, we developed a post-synthetic approach to realize ionization on the channel walls of a chemically stable DhaTab COF. Either metallic Mn²⁺ or the transition metal complex [Mn(bpy)₂]²⁺ was loaded within the pores of the COF through a simple ion exchange process. The resulting [SO₃Mn]-DhaTab and [SO₃Mn(bpy)₂]-DhaTab exhibited excellent catalytic activity in the epoxidation of olefins to epoxides. We consider that with this type of ionic architecture, a variety of other functional cations could be exchanged into the frameworks, thus making the COF a versatile platform for different applications, such as ion exchangers and ion conductors. Moreover, we predict that the pore volume and pore size of this COF can be tailored by ion exchange so that these parameters can be adjusted for adsorption and separation of gas molecules.