The Ionic Organic Cage: An Effective and Recyclable Testbed for Catalytic CO₂ Transformation

Abstract

Porous organic cages (POC) are a class of relatively new molecular porous materials, whose concept was raised in 2009 by Cooper’s group and has rarely been directly used in the area of organic catalysis. In this contribution, a novel ionic quasi-porous organic cage (denoted as Iq-POC), a quaternary phosphonium salt, was easily synthesized through dynamic covalent chemistry and a subsequent nucleophilic addition reaction. Iq-POC was applied as an effective nucleophilic catalyst for the cycloaddition reaction of CO₂ and epoxides. Owing to the combined effect of the relatively large molecular weight (compared with PPh₃⁺I⁻) and the strong polarity of Iq-POC, the molecular catalyst Iq-POC displayed favorable heterogeneous nature (i.e., insolubility) in this catalytic system. Therefore, the Iq-POC catalyst could be easily separated and recycled by simple centrifugation method, and the catalyst could be reused five times without obvious loss of activity. The molecular weight augmentation route in this study (from PPh₃⁺I⁻ to Iq-POC) provided us a “cage strategy” of designing separable and recyclable molecular catalysts.

1. Introduction

In 2009, Cooper and co-workers discovered a new type of 3D discrete molecular cages with a central cavity and molecule accessible windows, and then named them porous organic cages (POC) [1]. Research in this field has been on the rise in the last decade [2,3,4,5,6,7,8,9,10,11], with imine condensation [12,13,14,15,16,17,18] and boronic esters formation [19,20,21,22,23,24,25,26] being the two main strategies for constructing POC materials. When thinking at a higher level about the assembly, scientists found that POC materials are assembled by two organic synthons through dynamic covalent chemistry (DCC) [5], which means POC materials are constructed by reversible covalent bonds only, and this is the main characteristic that distinguishes them from other cage materials (such as metal-based coordination cages [27]). Due to their porous nature and “solution processable” feature, POC materials have found widespread application in the areas of gas storage and separation materials [28,29,30,31,32,33], host-guest chemistry [34,35,36,37], sensors [21,22,38,39,40], packing materials for chromatography [41,42,43,44] and synthetic templates for metal nanoparticles (MNPs) [45,46,47,48,49,50,51,52,53,54].

Despite the research of POC materials growing substantially in the last decade (POC-stabilized MNPs in catalysis are relatively common [47,48,49,50,51,52,53,54,55,56,57,58,59]), the direct using of pure POC molecules in catalysis has rarely been reported. To the best of our knowledge, only a few studies (less than five) have been reported. For instance, Smith and co-workers developed an iron porphyrin-decorated POC molecule as a highly efficient catalyst for electrochemical CO₂ reduction [60]. Jiang’s group elaborated a tubular organic porphyrin cage and elucidated its heterogeneous photocatalytic superiority in the heterogeneous photo-oxidation of primary amines [61]. Very recently, Kim’s group reported a gigantic porphyrin cage (P₁₂L_24, built with 12 square-shaped porphyrins (P) and 24 bent linkers (L)), and it was applied as an efficient catalyst for the photooxidation of 1,5-dihydroxynaphthalene in the presence of oxygen and visible light [62]. Patra et al. reported an organic cage for catalytic CO₂ conversion [63]. In this contribution, we introduce a quaternary phosphonium salt-functionalized cage as an effective and recyclable catalyst for catalytic CO₂ transformation.

Today in the 21st century, the greenhouse gas CO₂, which is mostly produced by the burning of petroleum, coal and natural gas, is a big problem due to climate change, ocean acidification and the global sea-level rise [64]. On the other hand, CO₂ could be an economical, renewable, nontoxic and abundant source of carbon, and therefore an attractive C1 synthetic building block for some useful fine chemicals. Additionally, to a certain extent, the conversion of CO₂ into value-added fine chemicals helps the carbon cycle of the earth, which is a fascinating goal in the area of green chemistry and catalysis [65,66]. Notably, the transformation of CO₂ into cyclic carbonates has attracted much attention because of the extensive applications of cyclic carbonates, such as the electrolytes of lithium-ion batteries, polar aprotic solvents and important intermediates for fine chemicals [67,68,69,70]. Recent research suggested that cyclic carbonates can be straightforwardly converted to methanol (CH₃OH) and diols through a catalytic hydrogenation process, and this process might be an alternative pathway to produce CH₃OH from CO₂ (Scheme 1a) [71,72].

To synthesize the cyclic carbonates from epoxides and CO₂, a catalyst is essential during the ring-opening of epoxides. Various molecular catalysts have been developed to realize the ring-opening process of epoxides. However, in the end, in almost all cases, a nucleophile as the catalyst is essential for the reaction, and a cocatalyst (such as a Lewis acid and another oxophilic compound) is not essential, although the cocatalyst also plays an important role (Scheme 1b) [73]. Among the various nucleophiles reported, ionic liquids [74,75,76], quaternary ammonium salts [77,78,79] and quaternary phosphonium salts [80] are the three most recognized catalysts because of the powerful nucleophilicity of their counter anions. However, almost all the homogeneous nucleophile catalysts are hard to recover and reuse; therefore, from the viewpoint of green catalysis, it is necessary to develop a recyclable nucleophile catalyst [81]. In this contribution, a novel quasi-ionic porous organic cage (Iq-POC), a quaternary phosphonium salt, was synthesized and utilized as a recyclable nucleophile catalyst (Figure 1a). We came up with the name quasi-porous organic cages (q-POC) because their pore windows are too small for external molecules to enter their inner cavities (the crystal data and nitrogen adsorption experiment can prove that). From one point of view, our nucleophile cage catalyst Iq-POC could be viewed as an amplified version of methyltriphenylphosphonium iodide (5). As a matter of experience, small organic molecules with molecular weights (Mw) < 500 are generally soluble in common solvents; however, strongly polar compounds with large molecular weights (Mw) > 1000 are usually poorly soluble in common solvents. Therefore, the molecular amplifying process from methyltriphenylphosphonium iodide to Iq-POC can be understood as the heterogenized (from homogeneous to heterogeneous) process of a homogeneous catalyst. The synthetic route of the nucleophile catalyst Iq-POC is illustrated in Figure 1a. Two equivalents of triphenylphosphine-based tri-aldehyde (1) and three equivalents of (R,R)-1,2-cyclohexanediamine were assembled together ([2+3] assembly) to afford the PPh₃-based cage compound (3) through dynamic covalent synthesis. The crystal structure of cage 3 is illustrated in Figure 1b, which provides us a clear idea of its structure by using different viewing angles. Subsequently, cage compound 3 was reacted with iodomethane—a nucleophilic reaction—to form the target molecule Iq-POC (Figure 1a; for detailed synthetic procedures and characterizations, see Supplementary Materials). The cage compound Iq-POC could be recognized as an amplified version of methyltriphenylphosphonium iodide (5), and Iq-POC exhibited very poor solubility in epoxides and cyclic carbonates; therefore, it could be easily separated and reused in the catalytic synthesis of cyclic carbonates through the solvent-free cycloaddition reaction of CO₂ and epoxides.

2. Results and Discussion

The synthesis of Iq-POC was mainly divided into two steps, i.e., the synthesis of cage 3 through dynamic covalent chemistry (DCC) and the subsequent nucleophilic addition reaction. DCC is essentially a kind of reversible chemistry which has a self-correcting function [5]. A subsequent nucleophilic addition reaction is frequently used in organic syntheses, and the yield (91%) in this case was high. The cage Iq-POC was poorly soluble in almost all solvents, so it could not be characterized by liquid nuclear magnetic resonance (NMR). However, we obtained the high-resolution mass spectra (HRMS) spectrum; its molecular weight can be measured at very dilute concentrations. The peak at m/z = 478.2386 corresponds to [M − 2I]²⁺ (Figure 2, theoretical value: m/z = 478.7412, z = 2). To further confirm its structure, the solid-state ³¹P NMR spectrum was obtained for our ionic cage, and the result showed just one signal at 22.0 ppm (see Supplementary Materials), which is the characteristic peak of a “quaternary phosphonium salt.” The ³¹P NMR peak of the cage 3 was at −7.8 ppm (very similar to PPh₃), which is quite different from that of a quaternary phosphonium salt. Moreover, the powder X-ray diffraction (PXRD) indicated the ionic cage was relatively amorphous (see Supplementary Materials), and the N₂ adsorption isotherm revealed the cage was nonporous (very low adsorbed volume; see Supplementary Materials), which could be ascribed to the narrow windows and dense packing structure. With the prepared nucleophilic cage Iq-POC in hand, we then test the catalytic activities of Iq-POC/ZnX₂ catalysts toward cycloaddition of CO₂ and propylene oxide. For the sake of simplicity and cost, ZnX₂ combinations were chosen as the Lewis acid cocatalysts. The catalyst combination of Iq-POC/ZnBr₂ displayed better activity (turnover frequency (TOF) = 1550 h⁻¹,

Table 1. Catalytic activities of different catalytic systems in the cycloaddition reaction of CO₂ and propylene oxide ¹.

Entry	Catalyst	Gas Chromatography (GC) Yield (%)	TOF (h−1)
1	Iq-POC/ZnBr2	31	1550
2	Iq-POC/ZnCl2	21	1050
3	Iq-POC/ZnI2	28	1400
4	ZnBr2	trace	—
5	Iq-POC	2	100
6	CH3P+Ph3I−/ZnBr2	38	1900

¹ Reaction conditions: propylene oxide (2.9 g, 50 mmol), an initial pressure of 3 MPa CO₂, substrate/catalyst = 5000, Iq-POC (12.1 mg, 0.01 mmol), ZnX₂ (0.1 mmol), CH₃P⁺Ph₃I⁻ (8.1 mg, 0.02 mmol), 1 h, 120 °C. The selectivities of all the results were >99%. All the results were averaged over two runs.

, entry 1) than Iq-POC/ZnCl₂ (TOF = 1050 h⁻¹, Table 1, entry 2) and Iq-POC/ZnI₂ (TOF = 1400 h⁻¹, Table 1, entry 3). The order of the activity of different Lewis acidic cocatalysts was found to be ZnBr₂ > ZnI₂ > ZnCl₂, which is constant with a previous report [73]. To check if the imine bonds of the cage play a role in the catalytic cycle, methyltriphenylphosphonium iodide (CH₃P⁺Ph₃I⁻)/ZnBr₂ was used as the catalyst; the iodide ion was taken as the active center; the CH₃P⁺Ph₃I⁻/ZnBr₂ system (Table 1, entry 6) exhibited slightly higher activity than the Iq-POC/ZnBr₂ system. This might be explained by the better solubility of CH₃P⁺Ph₃I⁻ compared with the amplified cage analogue; the result also told us the imine bonds have little effect on the catalytic cycle. When the Lewis acid cocatalyst ZnBr₂ was used alone, only a trace amount of propylene carbonate was afforded (Table 1, entry 4), which indicates that the nucleophilic cage Iq-POC is essential for the catalytic system. When the nucleophilic cage Iq-POC was used alone without the cocatalyst, only 2% yield of propylene carbonate was obtained (Table 1, entry 5), which implies the Lewis acidic cocatalyst also plays an important role in this case. With the optimized catalytic system of Iq-POC/ZnBr₂ in hand, we then investigated the effect of reaction time on the yield of propylene carbonate. As illustrated in Figure 3a, the conversion of propene epoxide proceeded relatively quickly within the first 1 h (31% yield of propylene carbonate was afforded). After 7 h, a yield of 87% was obtained, and then the yield reached a plateau; a high yield (>95%) could be obtained when the reaction time was longer than 9 h.

Then, the recyclability test of the nucleophilic cage Iq-POC was performed in the model reaction of cycloadditon of CO₂ and propene oxide. Owing to the amplified cage structure, the nucleophilic cage Iq-POC is insoluble in propylene oxide and propylene carbonate; therefore, after the reaction, Iq-POC could be easily separated from the reaction system through filtration or centrifugation. To test the recyclability more accurately, a conversion of approximately 30% was chosen to evaluate whether there was a decline in activity, and the results show that recycling the Iq-POC cage over four runs did not lead to a significant loss of propylene carbonate yields (Figure 3b). Analysis of the reaction solution after each cycle by ICP (inductively coupled plasma) showed P and I element leaching did not reach the detection limit; however, we found ZnBr₂ salt is soluble in the reaction system; hence, the cocatalyst of ZnBr₂ salt has to be reloaded for the next run, while the Iq-POC was reused. Moreover, as shown in Figure 4, the versatility of our catalyst was further investigated. Under similar reaction conditions—though the reaction was prolonged for 10 h—satisfactory yields of corresponding cyclic carbonates were obtained, except for cyclohexene oxide (34% yield, 12 h), which was probably due to the steric hindrance of both the substrate and Iq-POC.

3. Materials and Experiments

3.1. General Remarks

Triphenylphosphine-based trialdehyde was synthesized according to [82,83]. All the other reagents were commercial grade and used without further purification. (R,R)-1,2-cyclohexanediamine (Alfa Aesar Co., Ltd., Tianjin, China), 4-bromobenzaldehyde diethyl acetal (Energy Chemical Co., Ltd., Shanghai, China), phosphorus trichloride (Energy Chemical Co., Ltd.) and all the epoxides were purchased from J&K Scientific. Co., Ltd. (Shanghai, China) and Alfa Aesar Co., Ltd. The carbon dioxide used is commercially available with 99.99% purity.

Yields refer to isolated yields of compounds estimated to be ≥95% pure as determined by ¹H NMR (25 °C). The liquid NMR spectra were recorded on a Bruker AVANCE III NMR spectrometer (Bruker, Karlsruhe, Germany), 400 MHz for ¹H spectra and 100 MHz for ¹³C spectra. Chemical shifts are reported as δ-values in ppm relative to the deuterated solvent peak: CDCl₃ (δ_H: 7.26; δ_C: 77.16). The high-resolution mass spectra (HRMS) data were collected on an Agilent Q-TOF6540 spectrometer (Agilent, Santa Clara, CA, USA). The single crystal X-ray diffraction data (SXRD) were collected on an Agilent GeminUltra diffractometer (Agilent, Santa Clara, CA, USA) with Mo Kα radiation, and liquid nitrogen purging was necessary to prevent crystal cracking. Inductively coupled plasma spectroscopy (ICP) was performed on a PerkinElmer apparatus Optima 7300 DV (PerkinElmer, Fremont, CA, USA). The N₂ adsorption experiment was performed on the Quantachrome Autosorb-1 instrument (Quantachrome, Boynton Beach, Florida, USA) at 77 K. GC analyses were performed on an Agilent 7890A (Agilent, Santa Clara, CA, USA) equipped with a capillary column (HP-5, 30 m length, 0.32 mm diameter), using a flame ionization detector.

3.2. Synthesis of the Cage 3

Ethyl acetate (150 mL) was added to triphenylphosphine-based trialdehyde (1, 429 mg, 1.24 mmol) in a beaker at room temperature. The suspension was stirred with a glass rod and became clear. Next, a solution of (R,R)-1,2-cyclohexanediamine (2, 212 mg, 1.86 mmol) in ethyl acetate (30 mL) was added dropwise, and a turbid solution was observed during the addition process. The resulting mixture was left covered for 72 h without stirring. Some precipitation was formed after around 1–2 h. After reaction, the resulting mixture was filtrated, and the filtrate was concentrated to around 5 mL by using a rotary evaporator. Then 100 mL methanol was added to the precipitate the product as a white powder. The cage product (81% yield, 465 mg) was collected by filtration and dried under reduced pressure. ¹H NMR (CDCl₃, 400 MHz): δ (ppm) = 7.67 (s, 6H), 7.30 (d, 12H, J = 4.0 Hz), 6.99 (t, 12H, J = 8.0 Hz), 3.19 (t, 6H, J = 8.0 Hz), 2.13 (d, 6H, J = 12.0 Hz), 1.92 (d, 12H, J = 8.0 Hz), 1.51 (t, 6H, J = 12 Hz). ¹³C NMR (CDCl₃, 100 MHz): δ (ppm) = 163.1, 139.1, 139.0, 136.6, 133.2, 133.0, 127.7, 127.6, 73.2, 32.0, 24.4. ³¹P NMR (CDCl₃, 400 MHz): δ (ppm) = −7.8. Electrospray ionization high resolution mass spectrometry (ESI-HRMS): m/z calculation for [C₆₀N₆P₂H₆₀]: 926.4355; found 927.4415 [M + H]⁺.

3.3. Synthesis of Iq-POC

Cage 3 (1.0 mmol, 0.93 g) was dissolved in 35 mL THF (tetrahydrofuran); then CH₃I (1.4 g, 10 mmol) was introduced to the cage solution. The mixed solution was refluxed for 48 h. After reaction, the cage Iq-POC was isolated by filtration and washed with THF (10 mL × 3). The crude product was dried under vacuum at 60 °C for 5 h to provide the final pure cage Iq-POC (1.1 g, 91% yield). Note: The cage Iq-POC was poorly soluble in almost all solvents, so it could not be characterized by liquid NMR. However, we obtained the HRMS spectrum; its molecular weight can be measured at very dilute concentrations. Electrospray ionization high resolution mass spectrometry (ESI-HRMS): m/z calculation for [C₆₂N₆P₂H₆₆]²⁺: m/z = 478.7412, z = 2; found 478.2386. Solid-state ³¹P NMR: δ (ppm) = 22.

3.4. General Procedure of Catalytic Reactions

Reactions were performed in a 25 mL autoclave with vigorous stirring. For a typical catalytic run, propylene oxide (1.16 g, 20 mmol), Iq-POC catalyst (4.8 mg, 0.004 mmol, substrate/catalyst = 5000) and ZnBr₂ (9 mg, 0.04 mmol) were added into the autoclave without any solvent. After sealing and purging with CO₂ 3 times, the pressure was adjusted to 3 MPa as an initial pressure. Then the autoclave was put into a preheated oil bath and the solution was stirred at 120 °C for several hours. After the reaction was completed, the autoclave was cooled to ambient temperature, and the excess CO₂ was carefully vented. The catalyst was separated by filtration and the aqueous solutions containing the products were analyzed using gas chromatograph. n-Butanol was used as an internal standard and the GC was also calibrated using known amounts of substrate and products. The recycled catalyst could be reused for the next run.

4. Conclusions

In conclusion, in this work we have developed a novel ionic quasi-porous organic cage (Iq-POC) through dynamic covalent chemistry and a nucleophilic addition reaction. This quaternary phosphonium salt-type cage was applied as a nucleophilic catalyst for the cycloaddition reaction of CO₂ and epoxides. Satisfactory activity and substrate scope were achieved by using our catalytic system. Owing to the amplified cage structure, the nucleophilic cage Iq-POC is insoluble in propylene oxide and propylene carbonate; therefore, after reaction, Iq-POC could be easily separated from the reaction system through filtration. This “heterogenized cage catalyst” may be a new solution in the area of homogeneous catalyst recycling. However, it is a pity that the co-catalyst cannot be recycled at present; in pursuit of a complete catalyst cycle, we need to try other catalytic systems in the future. Other interesting cage catalysts are under development in our laboratory.