Zwitterion $π$–conjugated network polymer based on guanidinium and $β$-ketoenol as a heterogeneous organo-catalyst for chemical fixation of CO₂ into cyclic carbonates

Carbon dioxide (CO₂) is one of the dominant anthropogenic greenhouse gases, which is believed to be contributing to global warming and climate change; hence, CO₂ chemistry (capture and conversion) has appealed worldwide as the most important and top priority for the scientific community.^1–7 Recently, the CO₂ concentration increased to 415 ppm in the atmosphere and continues increasing.

On the other hand, CO₂, an abundant, nontoxic, cheap, and nonflammable carbon source, which is formed from living organisms and in industrial processes, is a significant resource and can be easily handled at an industrial level. Chemical fixation of CO₂ is a growing interest in the development of the ecofriendly processes, as CO₂ can be used as a cheap and safe building block in organic synthesis.⁸ Among them, one of the most promising methodologies is the synthesis of five-membered cyclic carbonates via the cycloaddition reaction of CO₂ to epoxides.^9–14 These cyclic carbonates are important as precursors of raw materials in the production of plastics, pharmaceutical chemical intermediates, aprotic polar solvents, and biomedical applications.^15–17 Recently, numerous catalyst systems, such as ammonium and phosphonium salts, ionic liquids, cyclic amidines and guanidines, porous organic polymers (POPs), and metal-organic frameworks (MOFs), have been developed for this conversion.^18–33 Generally, homogeneous catalysts show effective catalytic activity with well-defined mechanisms, but main drawbacks are the need for separations and the purification of products. Heterogeneous catalysts are more beneficial in terms of scalability, rapid catalyst recovery, and reusability.^34–37 But the leading heterogeneous catalysts are found to be restricted due to poor stability and low activity in the harsh reaction conditions. Some of the promising recent reports include the work by Byun³⁸ which shows an imidazolium ionic liquid catalyst that facilitated the selective formation of cyclic carbonates under ambient reaction conditions. Recently, we have investigated a new type of pyridyl salicylimine polymeric catalyst that exhibits good conversion without any cocatalysts.³⁹ Zhang and He⁴⁰ showed the importance of guanidine based organocatalyst in the field of CO₂ fixation. Therefore, it is still a challenging field to explore heterogeneous catalysts with excellent stability and activity under mild conditions for the nonredox chemical fixation of CO₂.

Guanidine-based functional motifs in the form of neutral (guanidine), cationic (guanidinium), and anionic (guanidinate) units have shown wide use in chemistry, such as in crystal engineering and asymmetric catalysis. These guanidine groups exist not only in enzymes and proteins but also in many natural drugs products. The guanidinium salts promote the reactions by using ion-pair interactions and hydrogen bonding to accelerate reaction rates. The strong basicity owing to resonance stabilization of the corresponding conjugate and ease for structural modification offer guanidine derivatives significant value in organic synthesis, especially in the field of CO₂ conversion.

In terms of framework building, dynamic covalent chemistry assists self-assembly of reactive building blocks into structurally complex yet robust materials, such as covalent organic frameworks (COFs), covalent organic polymers (COPs), and porous organic polymers (POPs).^41–48 In literature, boronic acid self-condensation, aldehyde/amine condensation (Schiff base reaction), polycondensation with aromatic diols, and nitroso dimerization mainly show this dynamic mechanism. Among them, the π–conjugated COPs can be designed by Schiff base reaction formation of imine bonds that exhibit similar dynamic covalent chemistry.

Herein, a unique zwitterionic active site on guanidinium–β-keto-enol based π–conjugated network polymers (COP-213) has been designed and synthesized which show good activity and selectivity in cyclic carbonate formation without the participation of cocatalysts or solvents and under mild condition. The structure of the materials has been analyzed in detail and proved by FTIR-attenuated total reflection (ATR), PXRD, thermogravimetric analysis (TGA), XPS, surface area Brunauer–Emmett–Teller (BET), and CO₂ uptake. To the best of our knowledge, guanidine and β-keto-enol-based network organic polymer-catalyzed transformation of CO₂ is not yet reported.

The design for a cycloaddition catalyst necessitates a quaternary ammonium center and a polar/heterocyclic backbone that is stable enough for the reactive conditions. One way to achieve this goal is to include multiple modes of connectivity and nucleophilic units. Hydrazone formation through a β-keto-enol and hydrazine reaction would yield a promising architecture. Our catalyst (COP-213), therefore, consists of these multifunctional moieties, the β-keto-enol and zwitterionic guanidium units shown in Fig. 1. To prepare β-keto-enol, we revised a synthetic strategy that was reported earlier^49,50 but applying to the triphenyl benzene core. This also marks the first time the tris β-keto-enol derivative of the triphenyl benzene was synthesized.

In order to incorporate guanidinium units within the polymers, we have prepared triaminoguanidinium halide (TGCl) from well-known procedures⁵¹ and reacted with C3 symmetric β-ketoenol at 120 °C in the presence of dioxane:H₂O mixture (2:0.6 ml) (Fig. 2). The resulting yellow powder was filtered and washed with dioxane and water several times. After that, materials were purified by Soxhlet extraction [tetrahydrofuran and dichloromethane solvent over 24 h] and dried under vacuum at 100 °C for 12 h. Computational calculations based on Density Functional Theory (DFT) conferred with a 2 × 2 unit network with a pore diameter of 8 Å (Fig. S9).

The polymer catalyst (COP-213) was examined first by FTIR-ATR spectra. In the FT-IR spectrum of COP-213, the peaks of —C—N (bending), C=O (stretch), C=C (stretch), =C—H (stretch), and N—H (stretch) were shown at 1280, 1669, 1,555, 3031, and 3404 cm⁻¹, respectively, which are the common features of beta-keto enol and guanidium polymers, while characteristic primary amine stretching frequency (1679 cm⁻¹) of the precursors was absent [Fig. 3(a)]. The morphology analysis was then studied by scanning electron microscopy image to show the nature of the polymers. The powder X-ray diffraction pattern showed largely amorphous feature of COP-213 [Fig. 3(b)]. Permanent porosity of COP-213 was verified using N₂ adsorption/desorption isotherms of the activated samples at 77 K. The Brunauer–Emmett–Teller (BET) surface area of COP-213 was found to be 1.8 m²/g. The low BET value indicates a strong hydrogen bonding of polymer networks. However, it shows good CO₂ uptake performance with the values of 1.12 mmol/g (25.27 cm³/g) at 298 K [Fig. 3(c)]. A moderate BET surface area of 126 m²/g is calculated based on CO₂ as a probe molecule. The pore size distribution from CO₂ adsorption data revealed that pores predominantly remain under 9 Å (Fig. S9). The thermal stability of COP-213 was studied under nitrogen atmosphere and showed that it is stable up to 350 °C [Fig. 3(d)]. To analyze the active guanidium sites in catalyst, XPS was carried out. According to the data, N1s spectra can be convoluted into three different peaks of 399.09, 400.41, 401.33 eV, which could be assigned to —C=N, —NH—, and —C=NH+, respectively [Fig. 3(f)]. Cl 2p spectra revealed a major peak at 200.09 eV, which confirms free chloride ions in the frameworks [Fig. 3(e)].

To optimize the catalytic activity for chemical fixation of CO₂, epichlorohydrin was selected as a model reaction according to literature reports. A typical experiment for catalytic cycloaddition of CO₂ to the epoxide was as follows: epichlorohydrin (5 mmol) and catalysts (30 mg) under solvent-free conditions were charged into a reactor that connects with a CO₂ cylinder under atmospheric pressure [Fig. 4(a)]. The reactor vessel was placed into an oil bath at 100 °C and kept under stirring. After the given time, the reaction mixture was cooled to room temperature. Conversion and selectivity were measured by a 300 MHz ¹H NMR analysis. The catalyst COP-213 shows excellent catalytic activity (>99% conversion) after 48 h.

The remarkable catalytic activity of polymer COP-213 for the cycloaddition of CO₂ to epoxides allowed us to obtain structural facts on the grafting of guanidinium moieties. Both cation and anion parts of the guanidinium unit are expected to have a significant effect on catalytic activities. Selection of the halide ion is also crucial because counter-anion needs to show a nucleophilicity to activate the epoxide ring and then have a good leaving ability. Among them, chloride ions show the highest activity through a S_N2 mechanism. It is proposed that the epoxide is activated along with the electron deficient guanidinium unit of the polymer. On the other hand, the bulkiness of the guanidinium ion influences the electrostatic interaction between the cations and anions, which limits the counter-anion to be more nucleophilic.

Various factors including time, loading of the catalyst, and substrate variations were investigated to find the optimal reaction condition and evaluate the catalytic performance. The dependence of product yield and selectivity on reaction time was studied under identical conditions shown in Table I. The yield of cyclic carbonate increased rapidly in the initial time and stayed steady after 48 h. The effect of varying catalyst loading was investigated for different catalyst amount at 100 °C and atmospheric CO₂ pressure for 24 h. Minimum catalyst loading shows negligible yield. However, increased loading realized a significant enhancement in activity, and a steady increment was observed at 30 mg catalyst loading [Fig. 4(b)]. We also investigated whether synthetic components themselves were as active as the network polymer itself. The catalytic performance of guanidium, TGCl, beta-keto-enol, and the physical mixture of TGCl & enol showed negligible activities in comparison with COP-213 (Fig. S7).

To monitor industrial viability, the substrate scope was also examined. Propylene oxide, an industrially very important substrate, shows good conversion and selectivity. The epibromohydrin also yielded excellent results. The hard substrates styrene oxide and 2-(4-Fluorophenyl)oxirane led to a low conversion but high selectivity in cyclic carbonates. Less reactive and hard substrates as 1,2-Epoxybutane, 1,3-Butadiene Monoepoxide, and 1,2-Epoxyhexane conversion of cyclic carbonate showed low conversions. We are still studying these substrates to see how we can improve further. The turnover numbers and frequencies were found to be moderate for cycloaddition reaction (TON: 2–20) (Table I).

We studied the reusability of COP-213 using model substrate under optimized reaction conditions and found that the catalyst can be reused at least 3 times with minimum loss in its activity [Fig. 4(c)]. In addition, we carried out hot filtration test, where the catalyst is filtered during the catalytic reaction while the reaction kept going. In a standard cycloaddition conditions, we filtered the catalyst after 6 h (29% conversion) and let the reaction continue for another 6 h. The analysis of the samples from 6 h to 12 h revealed that there were no catalysts left in the solution that provided further conversion (Fig. S8).

Based on the literature examples,^52–58 the reported catalytic performances, and the review of the catalyst structures, a plausible synergistic catalytic mechanism for this cycloaddition of CO₂ to epoxides in the presence of designed catalysts was carefully proposed, as shown in Fig. 5. Interestingly, the guanidinium cation carries the positive charges on the central N atom (Fig. 2) and forms an anion-binding site. The Cl⁻ anion in the catalyst is stabilized by the positive charges distributed over the guanidinium cation and is poised to attack epoxide. Nucleophilic attack by the Cl- anion gives intermediate A via five membered transition state which causes ring opening and generates the intermediate B. Once CO₂ is weakly bound to give complex C, the C–O bond forms via transition state, in which a negatively charged O atom of CO₂ is stabilized by the guanidinium cation. The cation interacts with the two negatively charged O atoms, through electrostatic stabilization of the guanidinium cation. This CO₂ insertion reaction is very exothermic (ΔH° ≈ −20 kcal/mol). Finally, the cyclic carbonate D is produced by an S_N2 type intramolecular elimination, the guanidinium cation interacts tightly with the two negatively charged O atoms of the carbonate rather than with the leaving Cl atom, at the same time regenerating the catalyst for another cycle.

In conclusion, a novel π–conjugated zwitterionic network polymer, based on β-keto-enol and guanidinium units, has been synthesized and characterized successfully. The synthesized organocatalyst exhibits high activity and selectivity for cycloaddition reaction of CO₂ to epoxide without a cocatalyst under a solvent-free condition in atmospheric pressure. The conversion of CO₂ to various cyclic carbonates takes place effectively using this catalyst because of the “electrophile–nucleophile” synergistic effect for epoxide ring opening. It is expected that guanidine-based polymers could significantly impact the development of heterogeneous catalysis through rational design and play an important role in CO₂ capture and utilization to value-added chemicals in industries.

See supplementary material for more details on materials and methods, characterization of organocatalysts, and products.

This work was supported by the Saudi Aramco-KAIST CO₂ Management Center and the National Research Foundation of Korea (NRF) grant funded by the Korea Government (MSIP) (Grant Nos. NRF-2017M3A7B4042140 and NRF-2017M3A7B4042235).