As an important and promising chemical fixation pathway of carbon dioxide (CO₂), its cycloaddition with epoxides to form cyclic carbonates has attracted considerable attention in recent years owing to its excellent merits, including high atom economy, green reaction, and the versatapplication in the syntheses of cyclic carbonates [1–6]. Various kinds of homogeneous catalysts such as metal complexes [7–9], metal salts [10], N-heterocyclic carbenes [11], and ionic liquids (ILs) [12–14] and heterogeneous catalysts such as metal oxides [15], metal-organic frameworks (MOFs) [16–18], porous organic polymers [19–21], and N-doped carbons [22] have been developed for this transformation. Among these catalysts, environmentally friendly ILs are found to be a good candidate for both CO₂ capture and fixation, and various ILs have been produced [23–27]. Although these IL-based homogeneous catalytic systems are efficient, their complex recycling processes, high viscosity, and relatively low utilization efficiency severely impede their industrial applications.

To overcome these existing problems, ILs are generally immobilized on different solid materials (e.g., silica [28–30], polymer [31–34], and MOFs [35, 36]), forming heterogeneous catalysts. However, a low catalytic efficiency is often encountered for the heterogeneous systems because of the loss of active sites after immobilization. A Lewis acid is required to form Lewis acid/ionic liquid catalytic system, because the synergetic catalytic effect between the Lewis acid and nucleophile could significantly promote the reaction [12]. Therefore, from a mechanistic point of view, the conventional immobilization methods, which result in the spatial separation of active sites within the rigid framework, will certainly block the dual activation mode. However, it is still highly desirable to develop IL-based heterogeneous catalysts that can promote the cooperative effects of the different active sites, for the CO₂ cycloaddition reactions.

Recently, it was found that the cooperative catalytic effects could be greatly enhanced through proper arrangement of the different active species in close proximity within a confined nanospace [35–43]. Especially, linear macromolecular ionic polymers with flexible and soluble properties were confined instead of IL catalysts. By doing so, not only the "free" state, similar to their homogeneous counterparts, could be maintained during reaction, but also an extraordinary recyclability could be obtained due to the existing twining effect of the long chain linear polymers with large molecular weight [37, 44]. Furthermore, while utilizing Lewis acid-doped porous support as the host material, the cooperative effect between the heterogeneous Lewis acid sites and poly-ILs can be enhanced, thus leading to a superior catalytic system. To date, different porous structures, including covalent organic frameworks [37] and metal organic frameworks [35, 36], have been investigated to accommodate linear ionic polymer catalysts and promote the cooperative effects. In addition to these porous materials, there are also other promising candidates that can be utilized as a host matrix. Ordered mesoporous materials, possessing well-defined pore structures, controlled pore size (2–50 nm), and high surface area, for example, have attracted great attention in the last two decades [45–48]. Catalytically active guest species can either be loaded into its mesoporous network or be dispersed on the pore surface through grafting or a co-condensation process. Thus, a synergetic catalytic effect may work between the two different kinds of functional groups introduced.

In this study, we report the confinement of a linear imidazolium-based ionic polymer catalyst in the mesopores of ordered mesoporous silica nanospheres (MSN) that are pre-anchored with zinc species through in situ polymerization. Such a unique, active, and co-decorated solid species could efficiently catalyze the cycloaddition reaction of epoxides to form cyclic carbonates. It could also exhibit a much higher activity than its corresponding individual catalytic components. The superior catalytic activity is mainly attributed to the enhanced cooperative effect between the zinc species that acted as a Lewis acid and the ionic liquid that served as a nucleophile in the confined mesoporous structure.

All the chemicals were used directly as received. Sodium hydroxide (NaOH), cetyltrimethylammonium bromide (CTAB), tetraethyl orthosilicate (TEOS), propylene oxide (PO), and ZnBr₂ were purchased from Shanghai Chemical Reagent Company of the Chinese Medicine Group. 2-(4-Chlorosulphonylphenyl)ethyl trichlorosilane (CSPE) in methylene chloride (47–53 wt%) was obtained from J&K Scientific LTD. 2, 2'-Azobisisobutyronitrile (AIBN) was purchased from Aladdin (China). Epichlorohydrin (99%), 1, 2-epoxybutane, styrene oxide, cyclohexene oxide, allyl glycidyl ether, and glycidyl phenyl ether were purchased from Beijing Innochem. 1-Vinyl-3-butylimidazolium bromide was prepared according to the method given in the literature [49].

A certain amount of NaOH solution (0.35 mL, 2 mol/L) was added to 45 mL water, in which 0.10 g CTAB was dissolved. After stirring at 70 ℃ for 15 min, TEOS (0.45 g) was added to the mixture as a silane precursor and prehydrolyzed for about 5 min. Then, CSPE in methylene chloride (0.32 g) was added dropwise to the system and stirred for another 2 h at 70 ℃. The powder product was isolated by filtration and washed with water and ethanol. To remove the surfactant, the synthesized material was dispersed in 60 mL ethanol, in which 40 mg ammonium nitrate was dissolved. After being treated at 60 ℃ for 0.5 h, the solid material was isolated and washed with ethanol. Such a process was repeated thrice to remove the surfactant completely. As for the synthesis of the MSN without the zinc species, a similar procedure as above was performed, except that only TEOS, and no CSPE, was utilized as the silane precursor.

The above solid powder was dispersed into 5.0 mL ZnBr₂ (1.0 g) solution at room temperature to allow for an overnight ion exchange process. After exchanging for three cycles, the final product was washed with water and ethanol sufficiently and dried under vacuum. The product was denoted as ArSO₃ZnBr-MSN.

A quantitative amount of 1-vinyl-3-butylimidazolium bromide (232 mg) as the monomer and AIBN (5% relative to vinyl group) as the initiator was dissolved in a mixture of ethanol and water (1.0 mL, 20:80 v/v). Then, the above solution was impregnated into 0.50 g ArSO₃ZnBr-MSN. The yielded powder was pumped to remove the air and heated at 80 ℃ under N₂ for 12 h for polymerization. It was washed with ethanol to remove the unreacted monomers and then air-dried to obtain the final product, denoted as ILP@ArSO₃ZnBr-MSN, signifying that the ionic liquid polymers (ILPs) are confined within the zinc species-anchored mesoporous silica nanospheres. To synthesize samples with different ILP contents, the dosages of 1-vinyl-3-butylimidazolium bromide monomer were changed to 150 and 350 mg, yielding L-ILP@ArSO₃ZnBr-MSN and H-ILP@ArSO₃ZnBr-MSN, respectively.

Transmission electron microscopy (TEM) was performed on a JEOL-JEM-2000EX microscope. Samples for observation were prepared by dispersing the solid powder in ethanol by sonication and then allowing a drop of the suspension to evaporate on a copper gird covered with a holey carbon film. X-ray diffraction (XRD) patterns were obtained using a desktop powder X-ray diffractometer at 40 kV and 30 mA with Cu K_α radiation. Thermo-gravimetric analysis (TGA) was performed under an air atmosphere at a heating rate of 10 ℃/min on a TA SDT Q600 analyzer. Fourier transform infrared spectra (FT-IR) were obtained on a Nicolet iS5 spectrophotometer using a KBr pellet. Nuclear magnetic resonance (NMR) spectra were recorded on a Bruker AC 400 spectrometer using CDCl₃ as the solvent. Solid-state NMR spectra were recorded on an Infinityplus 300 MHz spectrometer. The spectra of the ¹³C CP-MAS NMR experiments were acquired at 75.4 MHz resonant frequency, 10.0 kHz spin rate, 5.0 s pulse delay, and 1.0 ms contact time, using hexamethyl benzene as a reference compound. Gas chromatography-mass spectrometry (GC-MS) analysis was performed on an Agilent 7890-5977A MSD using an HP-5 column. Nitrogen-sorption analysis was carried out on a Micromeritics ASAP 2020 analyzer. Before measurement, the samples were degassed at 120 ℃ under vacuum for 6 h. The specific surface area was calculated from the adsorption branch in the relative pressure range 0.05−0.15, using the Brunauer-Emmett-Teller (BET) method, while the total pore volume was estimated from the amount adsorbed at a P/P₀ value of 0.99. Pore size distributions were determined from the adsorption branches using the Barrett-Joyner-Halenda (BJH) method.

Typically, solid catalyst (ILP@ArSO₃ZnBr-MSN, 100 mg) and PO (780 mg, 13.4 mmol) were charged into a 15 mL stainless steel reactor equipped with a magnetic stirrer. After purging with CO₂ several times, the pressure was adjusted to 2 MPa, and the reactor was transferred into an oil bath to carry out the reaction at 110 ℃ under stirring for a certain time. Then, the reactor was cooled quickly by putting the autoclave into an ice-water bath. Subsequently, the CO₂ pressure was released slowly, and the solid catalyst was separated by centrifugation. The supernatant was analyzed on an Agilent 7890 GC instrument equipped with a PEG column (30 m × 0.25 mm × 0.25 mm). For the recycling test, the solid catalyst obtained by centrifugation was washed with ethyl acetate and dried under vacuum. Then, the catalyst was used for the subsequent cycloaddition of CO₂ with PO. The same procedure was conducted four times to examine the recyclability.

All the steps for the synthesis of the ordered mesoporous silica decorated with dual-active-species are illustrated in Fig. 1. As shown, mesoporous silica anchored with sulfonate groups was first prepared via a prehydrolysis-induced co-condensation strategy using CTAB as the surfactant under basic conditions. During this process, 2-(4-chlorosulphonylphenyl)ethyl trichlorosilane was chosen as the co-condensed organosilane, because its chlorosulphonylphenyl group could be easily transferred in situ to the sulfonate group under the conditions of aqueous base synthesis owing to its high activity. A prehydrolysis method is utilized here because such a process is demonstrated to be helpful for introducing abundant organic groups while simultaneously maintaining the morphology of the silica materials [50]. Then, after a simple ion exchange process with overmuch ZnBr₂, zinc species were anchored on the surface of mesoporous silica materials, serving as Lewis acid sites. Subsequently, the obtained host material was impregnated with 1-vinyl-3-butylimidazolium bromide monomers for the thermally induced radical polymerization, to afford ILPs within the nanopores.

To verify the chemical compositions and structures of the yielded hybrid catalysts, various characterizations were performed. Fig. 2A shows that compared with the pure silica sample, MSN, the FT-IR spectrum of ArSO₃ZnBr-MSN not only exhibits the characteristic bands for silica around 3430, 1642, and 1086 cm^–1 for ν(O–H), δ(O–H), and ω(Si–O), respectively, but also shows typical peaks ascribable to the vibrations of organic groups, such as the C-H stretching vibrations at 2800–2950 cm^–1, C=C vibration of phenyl at 1450–1540 cm^–1, and the breathing vibration of the phenyl ring at 650–730 cm^–1, confirming the incorporation of the –Ar groups within the mesoporous silica. The confined polymerization was confirmed from the additional peaks observed around 1160 and 1040 cm^–1 arising from the imidazolium ring in ILP@ArSO₃ZnBr-MSN. Stronger peaks at 2800–2950 cm^–1 for the C–H stretching vibrations were also observed, which were mainly attributable to the higher content of the organic groups. These observations demonstrated the successful incorporation of the proposed ILP into ArSO₃ZnBr-MSN. To provide an additional proof, solid-state ¹³C MAS NMR analysis was performed. As shown in Fig. 2B, the signals arising from the aromatic carbons (in the range 120–150 ppm) and −CH₂CH₂− groups (around 30 ppm) are observed in ArSO₃ZnBr-MSN, demonstrating the successful anchoring of the –Ar groups. For the ILP sample, the chemical shifts attributable to the carbon atoms of both the imidazolate ring (ranging from 120 to 138 ppm) and CH₂ groups linked to the imidazole ring (ranging from 12 to 53 ppm) were observed. Integrating with similar signals of ILP@ArSO₃ZnBr-MSN, it is obvious that the ionic polymers are successfully incorporated. The ILP content was measured by TGA. In Fig. 2C, the weight loss below 200 ℃ is mainly ascribed to the physically adsorbed water, while the weight loss between 200 and 800 ℃ corresponds to the degradation of the organic groups. For ArSO₃ZnBr-MSN and ILP@ArSO₃ZnBr-MSN, the weight losses due to the organic groups are calculated to be 25.8 wt% and 52.9 wt%, respectively. The loading amount of ILP can be estimated to be about 27.1 wt%. Notably, ILP@ArSO₃ZnBr-MSN affords two consecutive steps of weight loss at 200 and 420 ℃, suggesting the uniform accommodation of ILP in the mesopores. According to the inductively coupled plasma-mass spectrometry (ICP-MS) analysis, the amount of zinc is measured to be 0.86 mmol/g, which is slightly lower than that of the ArSO₃– groups (1.15 mmol/g).

The mesostructures of the obtained materials were characterized by low angle powder XRD. It is clear from Fig. 2D that both MSN and ArSO₃ZnBr-MSN exhibit an intense diffraction peak assignable to the (100) reflection, along with two weak peaks corresponding to the (110) and (200) faces, suggesting that ArSO₃ZnBr-MSN has the same 2-D hexagonal mesostructure as MSN. ILP@ArSO₃ZnBr-MSN exhibits a similar XRD pattern, suggesting the successful incorporation of the ordered mesoporous structure. However, after the inclusion of the ILP within the mesopores, the intensities of the corresponding diffraction peaks were weakened, indicating a disturbance in the mesostructure. The textural properties of these samples were detected by nitrogen sorption. Fig. 2E shows that MSN and ArSO₃ZnBr-MSN exhibit typical type IV isotherms with a sharp capillary condensation step at relative pressures, P/P₀, of 0.2 to 0.4, which confirm their well-defined mesoporous structure. The BJH pore size distribution curves further show a uniform mesopore size centered at 2.4–2.6 nm (Fig. 2F and Table 1). Such mesopores are large enough for the following accommodation of ILP. As shown in Table 1, the BET surface area and total pore volume of ArSO₃ZnBr-MSN are measured to be as high as 702 m²/g and 0.69 m³/g, respectively, which are slightly lower than those of MSN (984 m²/g and 0.97 m³/g, respectively). After the incorporation of ILP, the BET surface area and pore volume of ILP@ArSO₃ZnBr-MSN sharply decreased to 59 m²/g and 0.32 m³/g, respectively. As the amount of incorporated ILP increases, the BET surface area decreases from 141 to 19 m²/g due to the dense occupancy of the ILPs in the nanopores.

The structures and morphologies of the yielded materials were further characterized by TEM. ArSO₃ZnBr-MSN consists of monodisperse nanoscale particles with spherical morphologies, with particle size in the range 100–200 nm, which is similar to that of MSN (Fig. 3a and 3b). Ordered mesopores with diameter 2–3 nm are clearly seen from the images, suggesting the mesostructured character, which has been confirmed by the XRD and nitrogen sorption analysis. No bulk ILPs were formed on the external surface of ILP@ArSO₃ZnBr-MSN nanospheres, and their mesopores were indistinct from those of ArSO₃ZnBr-MSN (Fig. 3c and 3d), demonstrating that the polymerization occurs entirely within the mesopores. To reveal the distribution of ILPs in ArSO₃ZnBr-MSN, HAADF-STEM was performed. As shown, uniform elemental distributions of Si, O, Br, N, S, and Zn within the silica nanospheres are clearly observed, suggesting that the Lewis acid salts and ILPs are dispersed uniformly within the mesoporous silica material.

Considering the flexibility of linear ILP in ILP@ArSO₃ZnBr-MSN and its potential cooperative ability with the anchored Lewis acid active sites (zinc species) on the silica walls and the porous structures, we set out to evaluate the catalytic performance in the cycloaddition reaction of epoxides. Encouragingly, ILP@ArSO₃ZnBr-MSN exhibits an excellent catalytic activity, and almost complete conversion (> 99%, Fig. 4a) could be obtained in the cycloaddition of PO with 2 MPa CO₂ at 110 ℃, in the absence of any co-catalyst. This is comparable with the homogeneous catalytic system of IL and ZnBr₂. Under identical conditions, the samples with individual active sites, ArSO₃ZnBr-MSN and ILP@MSN (the sample prepared through IL polymerization in pure MSN), give only 4.5% and 80% conversions, respectively. Due to the negative textural parameters of ILP@ArSO₃ZnBr-MSN (BET surface area: 59 m²/g, pore volume: 0.32 m³/g), the significantly enhanced catalytic activity observed might mainly be attributed to the strengthened cooperative effects between two different catalytic components, namely Br^- anions in ionic liquid (nucleophile) and the zinc species (Lewis acid). This is further supported by a control experiment using a simple physical mixture of ArSO₃ZnBr-MSN and ILP@MSN as a solid catalyst, in which only 83% conversion could be obtained under the same reaction conditions. The active species are spatially isolated in the last scenario, thereby leading to the obstructive contact of the two active sites. To further substantiate the cooperative effect, ArSO₃ZnBr-MSN, together with the soluble ILP component, were used to catalyze the reaction. An enhanced activity was observed, and 96% conversion was achieved. However, it is still slightly lower than that in ILP@ArSO₃ZnBr-MSN. This is easy to understand. In the dissolved state, the free mobility of the linear ILP species enables the nuclephile (Br^- ions) and Lewis acid sites to work synergistically, thus facilitating the reactions. However, a portion of ILP is distributed in the solution and will not participate in the cooperative catalytic reaction, thus resulting in a relative lower utilization efficiency. However, in ILP@ArSO₃ZnBr-MSN, almost all the ILP components are dispersed in the vicinity of the zinc species within the nano-confined spaces; therefore, the chances of cooperative processes between the two different types of catalytic components are increased significantly, leading to higher activity. These observations also highlight the superiority of our in situ polymerization method, through which the ILP are confined homogeneously in the mesopores. Nevertheless, the obtained catalytic performances are not so impressive compared with some of those reported in the literature (listed in Table S1). This might be ascribed to the compact structure and low surface area of the solid catalyst.

For further investigating the catalytic performance of ILP@ArSO₃ZnBr-MSN, the reaction kinetics were recorded. Fig. 4b shows that the cycloaddition of PO proceeds smoothly as the reaction time increases, demonstrating the excellent catalytic activity. The influence of the ionic polymer content on the catalytic performance of the ILP@ArSO₃ZnBr-MSN catalysts was also investigated (Table 2). As evident, on increasing the polymer content, the activity of the composite catalyst increased gradually. For H-ILP@ArSO₃ZnBr-MSN with a high ILP loading, complete conversion was obtained within 3 h, while a longer time of 5 h was needed for L-ILP@ArSO₃ZnBr-MSN. The influences of reaction conditions, such as the amounts of catalyst, temperature, and CO₂ pressure, were also investigated. Table 2 shows that a longer time of up to 5 h was needed to complete the reaction when the amount of solid catalyst was decreased to 50 mg. However, the reaction was already over within 1.5 h as the amount of solid catalyst was increased to 200 mg. On raising the reaction temperature from 45 to 130 ℃, the reaction rate increased significantly. For example, the PO substrate could be completely converted within 2 h at 130 ℃, while 94.6% and 97% conversions were achieved within 24 and 7 h at 45 and 90 ℃, respectively. Notably, when the CO₂ pressure was changed to 1 MPa, 96% conversion could still be achieved within 7 h, further demonstrating the good efficiency of the solid catalyst.

Encouraged by the above results, we then investigated the recycling stability under the same reaction conditions, which is an important feature of a heterogeneous catalyst [51]. After the reaction, the solid catalyst was separated by simple centrifugation and used directly for the next cycle. Remarkably, ILP@ArSO₃ZnBr-MSN could be readily recycled for at least four cycles, highlighting its heterogeneous nature (Fig. 5a). This is interesting because the absence of covalent bonds between the mesoporous silica and ILP might result in the leaching of ILP from the solid catalyst during the reaction. The remarkable stability might originate from the existing twining effect of the linear polymers within the nanoporous scaffold. This is further verified by the low amount of leached ILP after treating the solid catalyst under harsh conditions (Fig. S1). However, the conversion decreased to 85% in the fourth cycle. For exploring the causes of the decreased activity, the zinc species were detected using ICP, and it was found that 5.8% zinc was present in the reaction solution. To further investigate the cause of deactivation, nitrogen sorption and TG measurements of the recycled catalyst were conducted (Fig. 5b-d). After the recycling tests, the solid catalyst exhibited slightly larger BET surface area and comparable pore volume as compared with the fresh catalyst (Table 1, 135 vs 59 m²/g and 0.269 vs 0.32 m³/g, respectively). This might resulted from the partial escape of ILP from the nanopores. Although TGA characterization demonstrated that only a small quantity of ILP was leached from the solid catalyst (47.5% vs 52.9%), the partial escape of ILP reduced the chances of contact between the two cooperative species. Therefore, the decreased activity might be related to the diminished cooperative effects.

Based on the outstanding performance of ILP@ArSO₃ZnBr-MSN in the CO₂ cycloaddition with PO, the reactions with other epoxides were studied (the corresponding products were determined by GC-MS and NMR, see Supporting Information). As listed in Table 2, excellent conversions were obtained in almost all the reactions under the test conditions, including those of terminal epoxides with both electron-withdrawing and electron-donating substituents. An exceptionally low conversion (40%) was achieved for cyclohexene oxide even at longer reaction times, which might result from the much larger steric hindrance of cyclohexene oxide compared with that of other terminal epoxides. These results indicate that the composite catalyst can be broadly used in such reactions.

On the basis of some previous reports [12, 37, 39, 52] and combining the dual-active species-anchored structure of the solid ILP@ArSO₃ZnBr-MSN catalyst and the observed catalytic performance, a possible intramolecular synergistic mechanism involving the Lewis acidic zinc species and nucleophilic Br^– is proposed for the CO₂ cycloaddition of epoxides to form cyclic carbonates. As illustrated in Fig. 6, the oxygen atom of the diffused epoxide first binds to the coordinatively unsaturated zinc sites (Lewis acid) that is dispersed on the silica wall, forming Zn–O coordination bond. This leads to the activation of the epoxy ring. Simultaneously, the nucleophilic Br⁻ in ILP attacks the less-hindered carbon atom of the activated epoxide at an appropriate angle. This process promotes the opening of the epoxy ring, affording an intermediate Zn-coordinated bromo alkoxide. Subsequently, CO₂ inserts into the open epoxy ring to form a zinc carbonate intermediate. Finally, the reaction reaches completion through an intramolecular cyclization of the acyclic ester, resulting in the formation of the desired cyclic carbonate and regeneration of the solid catalyst.

In summary, a dual-active species-based cooperative solid catalyst was constructed successfully by confining a linear imidazolium-based ionic polymer in the mesopores of the ordered mesoporous silica nanospheres that were pre-anchored with zinc species. Due to the strengthened synergistic activation of the active metal center and halogen anion, such a catalyst exhibited excellent activity in the cycloaddition reaction of the epoxides to form cyclic carbonates. Moreover, the catalyst could be easily separated and reused several times by a simple method of centrifugation. Because of the diversity of both the mesoporous host and the ionic polymer guest, this work might provide a novel avenue for the development of heterogeneous catalysts with integrated superior properties for other reactions.