Synthesis of Polycarbonates from CO2 Promoted by Immobilized Ionic Liquid Functionalized di‐Mg Complex Catalyst

The capture and conversion of anthropogenic CO2 is a paramount challenge for our global ecosystem. An optimal way of to cope with the emitted CO2 is to efficiently convert it to value‐added materials. Whereas nature sequesters CO2 by making sugar‐based polymers, utilizing CO2 to make highly demanded synthetic polymers such as polycarbonates is of great value. The present work reports the synthesis of a new supported ionic liquid (IL) functionalized organic ligand that is able to accept metal sites. After incorporating di‐nuclear magnesium, it was utilized as a single‐component solid catalyst for the copolymerization of CO2 and cyclohexene oxide. The obtained solid catalyst was found to be active under mild CO2 pressures of (1–15 atm) giving a turnover number of up to 283 and turnover frequency up to 11.8 h−1. To the authors knowledge, these rates are the highest obtained using a heterogeneous catalyst, maintaining 96–99 % polycarbonates selectivity and 97–99 % carbonate repeat units. In addition, the obtained polymers showed high molecular weights (16.7 to 11.7 kg/mol) with 1.05 to 1.6 dispersity (Đ). The catalyst was recycled 4 times, under regular laboratory conditions and without any intermediate reactivation steps, which provided ∼3 g of polycarbonate for ∼0.03 g catalyst (100 : 1) at 80 °C in neat cyclohexene oxide and 15 atm CO2.


Introduction
The synthesis of green materials through the utilization of anthropogenic carbon dioxide (CO 2 ) can have a pivotal effect on environmental mitigation of CO 2 emissions. [1][2][3] Even more so, if the CO 2 is effectively converted to green polymers, it can also reduce the environmental negative impact of non-degradable plastics. [4] In this context, the copolymerization of CO 2 with epoxides to make degradable aliphatic polycarbonates (PC) is a highly attractive catalytic pathway. [5,6] Since the discovery of the first diethylzinc catalyst by Inoue and co-workers in 1969, [7,8] tremendous efforts have been devoted to discover improved catalysts for this chemical transformation. [9][10][11] Coates and coworkers demonstrated that the use of β-diketiminate (BDI) zinc complexes in dimeric state was more active than the monomeric species in the homogenously catalyzing the alternating copolymerization of CO 2 with epoxides. [12,13] This effect was attributed to the close proximity of two zinc sites, which enabled the dual activation of the epoxide and CO 2 simultaneously to give higher activity. More recently, Williams and coworkers developed a series of robust molecular homogeneous catalysts composed of homo-and heterodinuclear coordinated by macrocyclic ancillary ligands [14][15][16][17][18] These class of catalysts were shown to have good activity and selectivity for the alternating copolymerization of CO 2 with epoxides under atmospheric CO 2 pressure. [19] Interestingly, their di-Mg bis(acetate) catalyst promoted the reaction in non-pure streams of CO 2 containing amine and thiol impurities. [20] Moreover, the same catalyst was found to maintain its activity and selectivity in the presence of up to 400 equivalents of water and to produce low molecular weight (600-9000 g/mol) polycarbonate polyols. [20] From a practical perspective, the heterogenization of this class of well-defined and active molecular catalysts is of great interest. However, in many cases the constraint applied by the solid surface on the configurational degrees of freedom of the molecular catalyst renders it less active and, in some cases, even affecting its product selectivity. [21,22] Whereas, heterogeneous catalysts for olefin polymerization have been extensively documented, [23][24][25][26] heterogeneous catalysts for CO 2 /epoxide polymerization to polycarbonates (PC) received far less attention. Moreover, examples of a single-component catalyst (i. e. not requiring a cocatalyst or a promoter) are scarce. [21] To the best of our knowledge, only three types of solid catalysts have been reported to promote the alternating copolymerization of CO 2 with epoxides: (i) zinc dicarboxylates, [27][28][29][30][31][32][33] (ii) zinc-cobalt(III) double metal cyanide nanoparticles (ZnÀ Co(III) DMC) [34][35][36][37][38][39][40][41][42][43][44] and (iii) silica-immobilized zinc β-diiminate. [21] The reactions using propylene oxide promoted by zinc dicarboxylate catalysts had a relatively moderate activity (TOF: 3 to 6 h À 1 ) at 40 to 70 bar CO 2 and yielded high molecular weight (M n ) PC with a broad dispersity (Đ: 2 to 8). [27][28][29][30][31][32][33] The ZnÀ Co(III) DMC catalysts were tested under 40 to 100 bar of CO 2 with terminal epoxides, yielded PC with 70-95 % carbonate content and a broad dispersity (2.2 to 8). [34][35][36][37][38][39][40][41][42][43][44] The zinc β-diiminate supported on silica showed moderate activity at significantly lower pressure of 7 bar CO 2 , providing PC at a TOF of 5 to 21 h À 1 with a carbonate content up to 78 % and a low Đ (1.29 to 1.89). [21] Notably, all these catalysts suffer from deactivation due to sensitivity toward air or moisture, making the use of these catalysts challenging and impractical. This highlights the importance and need for the development of new solid catalysts that can be recycled under standard ambient conditions and can promote the CO 2 /epoxide copolymerization to highly alternating polycarbonates with high M n and narrow MWD's. Inspired by the metalloenzymes multimetallic cooperativity [45][46][47][48][49][50][51] in nature and robustness of the reported homogeneous di-Mg bis(acetate) catalyst, [20] herein, we designed and synthesized a new silica grafted di-Mg complex based on a reduced Robson ligand functionalized with a vinylimidazolium-based IL side chain (IL-(Cl) 2 [LMg 2 ](OAc) 2 À SÀ SiO 2 ), Figure 1, Scheme S1 and S2 in Supporting Information. The presence of the imidazolium-based IL side chains, secondary amines and the oxo-groups of dinuclear sites are aimed at trapping the CO 2 more effectively and hence lower the required working pressure.This is approach is reasoned by the known high CO 2 uptake capacity of imidazolium based ionic liquids as well as the vast commercial use of amine solvents for CO 2 capture from flue gas. [52,53] We envisioned that the molecular architecture as in Figure 1 will promote the trapping of CO 2 by the imidazolium group facilitating the transfer to the two Mg centers, thereby promoting the cooperative activation and reaction with the epoxide. Our results show that the obtained di-Mg(II) catalyst is highly active as a single-component solid catalyst for CO 2 and cyclohexene oxide (CHO) copolymerization and produced > 99 % polymer selectivity and 99 % polycarbonate linkages under 15 bar CO 2 . Moreover, we were able to reuse the catalyst despite exposure to regular laboratory ambient conditions containing moisture and oxygen.

Results and Discussion
The new IL-functionalized Robson ligand, IL(Cl) 2 [H 4 L](ClO 4 ) 2 was synthesized from commercial reagents using a three-step chromatography-free synthetic route, to give an overall yield of 85 % (Scheme S1, see the Supporting Information for full synthesis details). The products in each step were confirmed using 1 H and 13 C NMR along with APCI-MS and elemental analysis (Figures S1-S9, Supporting Information). The product of the first step, 5-(chloromethyl)-2-hydroxyisophthalaldehyde, was also confirmed by single crystal x-ray analysis ( Figure S10 and Table S1, Supporting Information). We were unsuccessful in producing the reduced homogeneous analog of the IL-functionalized ligand, as the reduction process caused the ligand to polymerize. Hence, the divinyl-macrocyclic ligand (IL-(Cl) 2 [H 4 L](ClO 4 ) 2 ) was immobilized covalently onto a thiol terminated silica (SiO 2 -SH) via radical chain mechanism using azobisisobutyronitrile (AIBN) as a radical initiator (Scheme S2, Supporting Information). [54,55] The SiO 2 À SH was easily prepared by reacting commercially available fumed silica with 3-mercaptopropyl silane under inert conditions (Scheme S2, Supporting Information). [54,55] The grafting of the (IL(Cl) 2 [H 4 L](ClO 4 ) 2 ) onto SiO 2 -SH was confirmed by the appearance of two new bands at 1655 and~1530 cm À 1 (Figure 2d) in the diffuse reflectance infrared spectroscopy (DRIFTS) which correspond to C=N (imino group) and C=C (aromatic ring) vibrational stretching of the unreduced homogeneous complex respectively ( Figure 2c). [56][57][58][59] The reduced form of the Robson type macrocycle, IL(Cl) 2 [H 2 L]À -SÀ SiO 2 , was readily obtained by reduction of the tetraimino groups in the grafted complex (IL(Cl) 2 [H 4 L](ClO 4 ) 2 À SÀ SiO 2 ) material using NaBH 4 (Scheme S2, Supporting Information). The successful reduction was evidenced by the disappearance of the bands at 1655 and 1530 cm À 1 in FT-IR spectra and appearance of new bands at 1613 cm À 1 and 1476 cm À 1 which correspond to NÀ H bending vibrations [60,61] of tetraamino diphenol macrocycle in IL(Cl) 2    details in the Supporting Information file). The FT-IR spectrum of the latter showed a slight enhancement of the band at 1609 cm À 1 , which corresponds to the C=O vibration of the acetate groups ( Figure 2f). [62] The Mg loading on the grafted material analyzed using ICP-OES was found to be 0.2 mmol/ g silica . This is consistent with the mass loss of the complex detected by thermal gravimetric analysis coupled to a mass spectrometer (TGA-MS). The data showing that the main mass loss in the range of 200-700°C is accompanied by an increase in CO 2 and water release, confirming the combustion of the organic part of the immobilized complex ( Figures S11-13 (Figures S11-13, Supporting Information). Notably, the onset for oxidative decomposition of the reduced form of the grafted complex, IL(Cl) 2 [H 2 L]À SÀ SiO 2 , was found to be~230°C, which indicates the thermal stability of the immobilized ligand. [63,11,[27][28][29][30][31][32][33] The present di-Mg solid catalyst was tested for the copolymerization of CO 2 with CHO. Reaction testing were carried out at 1 or 15 atm of CO 2 and 80°C ( Table 1). The catalytic activity at 15 atm CO 2 pressure was found to be 2 times greater than what was reported for solid zinc-dicarboxylate catalysts at 40 to 70 atm, with a TOF of 11.8 h À 1 for the former. [11,[27][28][29][30][31][32][33] Analyzing the reaction product using 1 H NMR showed a broad signal at δ 4.60 ppm, corresponding to the aliphatic methyne protons of poly(cyclohexene carbonate)'s (PCHC). The signals related to undesired cyclic carbonate and polyethers at δ 4.0 ppm and 3.3 ppm were found to be relatively weak, confirming the high selectivity to polycarbonates (Figures 3 and S14, S15, Supporting Information). Integrating the 1 H NMR signals gave a PCHC selectivity > 99 % with 99 % carbonate linkages. This high activity and selectivity was previously shown by William's and coworkers for the homogeneous analog catalyst to proceed via a cooperative relationship between the two adjacent metal centers, which is facilitated by the coordinative flexibility of the ligand. [63] According to the proposed mechanism, [63] the epoxide ring-opening and CO 2 insertion takes place sequentially on both metal centres due to the proximity of the two-metal centers, which could facilitate a bidentate carboxylate chain binding mode, lowering the energy barrier to CO 2 insertion. Interestingly, in the current work, it seems that the presence of the solid surface did not diminish the catalyst ability to work cooperatively. Moreover, even at 1 atm CO 2 pressure our di-Mg solid catalyst promoted the copolymerization reaction to yield a high molecular weight polymer (10.8 kg/mol) with a carbonate repeat units (98 %) and a TOF of 1.9 h À 1 ( Table 1, entry 1), which to the best of our knowledge is observed for the first time using any heterogeneous catalyst. The IR absorption band of C=O stretching of the polycarbonate at ν =~1735 cm À 1 was measured to have high intensity, supporting the presence of high number of polycarbonate linkages. Whereas the intensity related to the cyclic carbonate carbonyl band at ν =~1803 cm À 1 was relatively weak (Figures 4 and S16 in Supporting Information).
Analysis by gel permeation chromatography (GPC) showed that all the produced polymers had a M n in the range of 16.7 to 11.7 kg/mol, (Table 1) (Figures 5 and S17-S20, Supporting Information). These values are similar in range to the results obtained by Williams et al., using the homogeneous di-Mg complex. [14] The relatively low M n is commonly observed in CO 2 / epoxide copolymerization and is attributed to the effect of residual water in the reaction media. [64] This phenomenon is evidenced by the observed dihydroxy and acetate end-capped polycarbonates (HOÀ PCHCÀ OH and HOÀ PCHCÀ OAc) as measured by matrix assisted laser desorption ionization-time of flight (MALDI-ToF-MS) spectrum ( Figure 6). To remove trace water intensive thermal/chemical drying procedures are required, which will damage the integrity of the complex or its connectivity to the surface of SiO 2 . Moreover, it was our initial intent to evaluate our catalysts performance without the use of extreme measures. The effect of trace water presence is not evidenced at low monomer conversions because the polymers are short and the chain transfer occurrence is limited. For low conversions the GPC trace shows a unimodal peak and a narrow Đ, Figure S17 and Figure S20 in Supporting Information. With the propagation of the reaction the effect of water becomes more prominent resulting in the formation of a bimodal peak in the GPC spectrum, Figure 5 and Figures S18-S19, Supporting Information. A distinct shoulder is observed on the main GPC peak associated to the presence of both HOÀ PCHCÀ OH and HOÀ PCHCÀ OAc polycarbonates, giving a slightly broader Đ, Table 1, entry 2, 3 and 4. [64] The di-Mg solid catalyst was successfully reused up to four times with no intermediate reactivation or special handling steps to give polycarbonate polyols (Table 1, entry 2-5). Importantly, ICP-OES analysis showed that between runs less than 0.01 ppm (3 � 1 %) Mg leached. Regardless, we did record a loss in activity between the 1 st cycle (TON of 283) to the 2 nd cycle (TON of 182) and to the 4 th cycle (TON of 155) (Table 1, entry 2, 3, and 5). Consistent with the reported results of the homogeneous analog, [14] in this work as well, the presence of water did not affect the reaction product selectivity, as measured by 1 H NMR and FTIR analysis ( Figure S21 and S22, Supporting Information). In this reaction the water acts as a chain transfer agent and hence may affect the obtained M n and Đ but, should not reduce the activity or alter the selectivity. [14] Since, in this work we observed the loss in activity in each cycle with no change to reaction selectivity, suggests that a change in the Mg bound initiating group (e. g. acetate group) after each cycle. [14] An alternative, and perhaps a more plausible explanation for the loss in activity and stable in selectivity could be attributed to the blockage of active sites by adsorbed oligomeric species. Notably, the 4 consecutive reaction cycles yielded~3 g of polycarbonate per 0.2 g of solid catalyst (0 .03 g complex) without the need for intermediate reactivation or special catalyst handling.  Table 1).  Table 1) showing carbonate content.  Table 1).

Conclusions
In the current work we showed the synthesis and covalent immobilization of a robust di-Mg bis(acetate)-IL complex to the surface of non-porous silica. Using this catalyst we demonstrated, for the first time, the synthesis of highly alternating poly(cyclohexene carbonate)'s through a coordination-insertion copolymerization of CO 2 with cyclohexene oxide over a singlecomponent solid catalyst. The catalyst was recycled 4 times under regular laboratory conditions, without the need for reactivation and only minor total leaching of < 3 mol% was observed. The di-Mg solid catalyst was shown to significantly out-performs previously reported heterogeneous catalysts, giving a TON up to 283 and TOF of up to 11.8 h À 1 under a mild CO 2 pressure (15 atm). The copolymerization was achieved with high polycarbonate linkages (97-99 %).

Synthesis of chloromethyl dialdehyde product (P-1)
A mixture of di-formyl (8.0 mmol, 1.2 g), paraformaldehyde (64.0 mmol, 1.92 g) and 21 ml conc. HCl were stirred in a flask for 5 min, and 1.5 mL of POCl 3 was added slowly. The resulting mixture was refluxed at 45°C for 5 days. After the reaction, the product was extracted by ether and then washed by saturated NaHCO 3 solution and brine. The organic layer was dried over anhydrous Na 2 SO 4 and evaporated to obtain P-1 as a pale yellow crystalline solid (2.0 g, 92 % yield). 1

Synthesis of 1-Vinylimidazole-tethered dialdehyde product (P-2)
A 100 ml round-bottom flask quipped with a magnetic stirrer was charged with CH 3 CN in which choro methylated di-formyl (3.92 mmol, 0.778 g) was clearly dissolved. To this CH 3 CN solution of vinyl imidazole (3.92 mmol, 0.369 g) was slowly added and the mixture was stirred at RT for 12 h. The obtained pale-yellow precipitate was separated by centrifugation and washed with acetonitrile (10 ml × 2) followed by a diethyl ether (10 ml × 2). After washing the precipitate was dried in vacuum.

Synthesis of IL(Cl) 2 [H 4 L](ClO 4 ) 2 À SÀ SiO 2
The preparation of immobilized di-Mg catalyst is illustrated in Scheme S2, in Supporitng Information. In a typical synthesis the silica support (CAS no: 112945-52-5) was first dehydrated at 200°C at 0.1 Torr for 24 h. The dried silica material was modified with 3mercaptopropyltrimethoxysilane as described by previous literature. [54,55] To

Synthesis of IL(Cl) 2 [H 2 L]À SÀ SiO 2
The solid IL(Cl) 2 [H 4 L](ClO 4 ) 2 À SÀ SiO 2 (1 g) material was suspended and stirred in 70 ml methanol. The suspension was cooled to 0°C, and NaBH 4 (320 mg) was added, slowly. As NaBH 4 was reacted, the orange suspension turned to pale yellow over a period of 0.5 h. The suspension was allowed to stir at room temperature for 3 h, after which water (70 ml) was added slowly, and the resulting mixture was left over night. The product was centrifuged and washed with methanol (40 ml × 3) followed by diethyl ether (40 ml × 2). After washing the obtained pale-yellow precipitate was azeotrope with dry toluene (25 ml

Material characterizations
All materials and chemicals required for synthesis were purchased from Sigma-Aldrich and used as received unless otherwise noted. The monomer cyclohexene oxide (CHO) for polymerization was dried over CaH 2 overnight, vacuum-distilled, and stored in the glovebox for further use. CP grade carbon dioxide was used for polymerization studies. Fourier transform infrared spectra (FTIR) were recorded on a Bruker (Tensor 27) Fourier Transform InfraRed spectrometer equipped with an attenuated total reflection (ATR) stage. ICP-OES analyses were done on a Thermo Scientific ICP Spectrometer ICAP 6300 DUO. TGA-MS measurements were performed on a SETARAM Labsys-Evo coupled to a Hiden QGA-pro using synthetic air as the carrier gas (20 % O 2 in Ar, 30 ml/min, 5°C/ min).

General experiment for polymerization reaction at 1 atm CO 2
The reactions were carried out in air tight 25 ml Schlenk flasks, using standard Schlenk techniques. In a typical reaction, di-Mg solid catalyst (0.2 g, 0.0305 mmol) was homogeneously dispersed in dry cyclohexene oxide (CHO, 0,015 mol), which is placed in Schlenk flask equipped with a magnetic stirrer. The cyclohexene oxide was degassed, before being left for stirring by refluxing with a CO 2 balloon at 80°C for the desired period. After which, the crude reaction mixture was then taken up in CH 2 Cl 2 and centrifuged to remove the solid catalyst and evaporated and dried in vacuo overnight. The product was analyzed by 1 H NMR spectroscopy without further purification as the vacuum was sufficient to remove unreacted cyclohexene oxide.
in a 25 mL stainless-steel Parr reaction vessel (which was dried in an oven at 140 o C overnight) under argon. After being sealed, the reactor was carefully evacuated and refilled with CO 2 and brought up to 15 atm. The reaction was carried out at the specified temperature and for the desired period of time. After the reaction, the reactor was cooled in an ice-water bath and slowly depressurized. Work-up was carried out in the same manner as above.

Recyclability procedure
The reactions were carried out in 25 ml stainless-steel Parr reaction vessel using the same quantities of di-Mg solid catalyst (0.2 g, 0.0305 mmol), cyclohexene oxide (CHO, 0.061 mol) and the same reaction conditions as mentioned above. After the reaction, the crude reaction mixture was then taken up in CH 2 Cl 2 and centrifuged to remove the solid catalyst. The solid catalyst was further washed with CH 2 Cl 2 5 times (20 ml × 5) under regular laboratory ambient conditions containing moisture and oxygen to remove any physically adsorbed polycarbonates traces from solid surface. After washings, the centrifuged solid catalyst was dried in vacuo for 6 h. The obtained solid catalyst was again reused for three consecutive cycles by applying the same reaction conditions and following the same washing procedure. In every cycle the obtained polycarbonates showed without appreciable loss of selectivity (97-99 %) confirmed from 1 H NMR analysis and negligible Mg leaching (< 0.01 ppm) calculated from ICP-OES analysis.