Synthesis of cyclic carbonates from diols and CO 2 catalyzed by carbenes †

The synthesis of cyclic carbonates from epoxides and CO 2 is a well-established reaction, whereas the synthesis of cyclic carbonates from diols and CO 2 is considerably more challenging, and few eﬃcient catalysts areavailable. Here, we describeheterocyclic carbenecatalysts, including one derived from a cheap and eﬃcient thiazolium salt, for this latter reaction. The reaction proceeds at atmospheric pressure in the presence of an alkyl halide and Cs 2 CO 3 . Reaction mechanisms for the transformations involved are also proposed.

Utilization of carbon dioxide (CO 2 ) in the production of fine chemicals and synthetic fuels would contribute towards a more sustainable chemical industry. However, CO 2 is a challenging molecule to activate as it is thermodynamically stable and kinetically inert in many transformations. Accordingly, only a few energy-efficient processes which employ CO 2 have been commercialized. 1 From a thermodynamic perspective, oxygenated cyclic carbonates are particularly suitable synthetic targets from CO 2 . These compounds have been exploited as electrolytes for lithium ion batteries, 2 building blocks for polymeric materials, 3,4 solvents 5,6 and intermediates in the synthesis of compounds such as dimethyl carbonate (DMC) 7 and ethylene glycol. 8 Industrial production of cyclic carbonates involves either the transesterification of diols with phosgene in an energy-intensive process 9 or the cycloaddition of CO 2 to epoxides. [10][11][12] Despite the latter route exhibiting 100% atom economy and industrial scalability, the synthesis of epoxides combined with their high reactivity and volatility are problematic. Recently, more stable, biodegradable 1,2-diols have been proposed as promising alternatives for the synthesis of cyclic carbonates with CO 2 . 13 Their reaction with CO 2 is, however, neither kinetically nor thermodynamically-favored due to the formation of water as the sole by-product. 14 Attempts have been made to by-pass this problem by the implementation of a suitable catalyst system and a dehydrating agent. Both heterogeneous and homogeneous catalysts have been proposed for this reaction. For example, a heterogeneous cascade catalysis comprising CeO 2 and 2-cyanopyridine is arguably the most efficient system. 15 However, this process requires harsh reaction conditions (150 1C and 50 bars of CO 2 ), an expensive reagent (2-cyanopyridine) and the activity is highly sensitive to the size of ceria particles. A number of homogeneous metal-free catalysts run under milder conditions and, interestingly, all are based on the 1,8-diazabicyclo [5.4.0]undec-7-ene (DBU) aided insertion of CO 2 . Different reagents are used to facilitate the subsequent alkylation step to afford cyclic carbonates in good yield under only 10 bars of CO 2 . 16 The reaction may even proceed at an atmospheric pressure of CO 2 if DBU and the alkyl halide are used in large excess. 17 The same mild conditions are employed in a system in which tosyl chloride and triethylamine are used to afford cyclic carbonates with 6-membered rings in good yields. 18 Ultimately, only a few efficient processes exist and finding an increasingly sustainable process for this reaction remains important.
Recently, N-heterocyclic carbenes (NHCs) have gained interest as catalysts for reactions which employ CO 2 as a substrate. [19][20][21][22][23] This stems from their ability to act as nucleophiles which activate CO 2 via the formation of imidazolium carboxylates. 24,25 Interestingly, these intermediates have been previously reported to catalyze the synthesis of cyclic carbonates from diols employing DMC as the carbonyl source rather than CO 2 . 26 Herein, we show the utility of carbene catalysts for the synthesis of cyclic carbonates from diols and CO 2 and, based on key experiments, propose plausible mechanisms for this transformation.
Initially, reaction conditions were optimized using 1-phenyl-1,2-ethanediol (1a) as the substrate, see Table 1. Several imidazolium and thiazolium carbene catalysts (1c-4c) were evaluated. NHCs 1c and 3c 19 and the thiazolium carbene catalysts 1b and 1d 27 have been previously shown to catalyze the N-methylation of amines using CO 2 as the carbon source. The efficiency of a variety of bases and alkyl halides was also studied as they are essential for the reaction to proceed (see below). 16,18 The ability of cesium carbonate (Cs 2 CO 3 ) to activate CO 2 and other small molecules 28-32 encouraged us to employ it as a base in the reaction. Dibromomethane (CH 2 Br 2 ) was also used due its efficiency in forming an effective leaving group. 16 The activity of 1c-4c was investigated in the presence of 2 eq. of CH 2 Br 2 and 2 eq. of Cs 2 CO 3 . The highest yields of styrene carbonate (1b) were obtained with catalysts 1c and 2c (Table 1, entries 1 and 2). In contrast, 3c and 4c resulted in lower product yields (Table 1, entries 3 and 4). The effect of quantities of CH 2 Br 2 and Cs 2 CO 3 on the reaction was studied. Increasing CH 2 Br 2 to 5 eq. resulted in 61% yield of styrene carbonate 1b (Table 1, entry 6). Interestingly, a larger excess of the base (3 eq. instead of 2 eq.) led to a slight decrease in the yield of 1b (Table 1, entry 7). It should be noted that the reaction proceeds in low yield using Cs 2 CO 3 as the base in the absence of CO 2 (Table S2, entry 2, ESI †). However, 13 C labeled CO 2 was used to confirm that the main source of the carbonyl group incorporated in the cyclic carbonate product originates from CO 2 (see NMR spectra comparing the 13 C NMR spectra of non-labeled and 13 C labeled products in the ESI †).
The enhanced activity of catalyst 2c might be due to a greater stability to moisture; note that 1b was not observed in a control experiment in which water was introduced into the system (Table S2, entry 6, ESI †). In the initial catalytic runs the active carbene catalyst was generated prior to reaction by the deprotonation of the corresponding salt with NaH. Subsequently, we found that the in situ generation of the carbene catalyst yielded 1b in 71% in presence of 3 eq. of Cs 2 CO 3 ( Table 1, entry 8). Interestingly, in a previous study using the ionic liquid 1-butyl-3-methylimidazolium hexafluorophosphate ([bmim][PF 6 ]) as solvent, the increased carbonate yield was attributed to the increased solubility of CO 2 . 16 Presumably, an active carbene was also generated by the deprotonation of the imidazolium salt by DBU -the ability of DBU and Cs 2 CO 3 to deprotonate [bmim][BF 4 ] to form a NHC has been reported. 33 DBU was evaluated under our conditions, but yielded 1b in a significantly lower yield ( Table 1, entry 11). 16 Na 2 CO 3 and K 2 CO 3 were evaluated in place of Cs 2 CO 3 , but afford the product in 0 and 5% yield, respectively, presumably due to the lower solubility of these carbonates in DMF (Table 1, entries 9 and 10). No product was observed with Et 3 N (Table 1, entry 12). We speculate that Et 3 N, which is often employed in the Stetter reaction, may undergo a Menshutkin reaction with CH 2 Br 2 thereby inhibiting the reaction. Notably, Cs 2 CO 3 was found to be the optimal base in this reaction owing to its ability to generate the active carbene catalyst as well as to act as a minor carbonyl donor and a dehydrating agent.
Dimethylformamide (DMF) was selected as a reaction solvent as it can activate CO 2 . 34 As expected, other polar aprotic solvents (dimethyl sulfoxide (DMSO) or dimethylacetamide (DMA)) could also be used (Table 1, entries 21 and 22), whereas no reaction was observed in toluene (Table 1, entry 23).
The optimum reaction temperature is 90 1C, with lower temperatures leading to a decrease in product yield (Table 1, entries 13 and 14) and with more elevated temperatures, e.g. 110 1C, leading to deactivation of the catalytic system (Table 1, entry 15). The alkyl halide also affects the reaction, in particular, 2 eq. of bromobutane (C 4 H 9 Br) results in a higher yield than 5 eq. of CH 2 Br 2 ( Based on the optimized conditions, which afford 1b in up to 81% yield, the scope of the reaction was explored using catalyst 2c ( Table 2). The substrates varied from 1,2-diols to 1,3-diols (2a-4a) bearing functional groups with varying steric influence. The diols were subjected to the optimized conditions of 2 eq. bromobutane, 3.2 eq. Cs 2 CO 3 at 90 1C and 1 atm CO 2 pressure.
Scheme 1 presents the principal mechanism for the carbenecatalyzed reaction. As mentioned above, both the base and alkyl halide are essential in the reaction, as confirmed in control experiments in which no carbonate was formed in their absence (Table S1, entries 4 and 5, ESI †). Scheme 2 represents the mechanism for the minor non-catalytic formation of cyclic carbonate in the absence of CO 2 (Table S1, entry 2, ESI †). C 4 H 9 Br is included in the second mechanism due to detection of dibutyl carbonate and n-butanol in the reaction mixture using GC-MS, see ESI. † However, a similar mechanism is likely to take place in presence of other alkyl halides. Moreover, both of these mechanisms appear to occur concurrently to form the cyclic carbonate. This hypothesis is based on our finding that while 25% of 1b was obtained in the absence of CO 2 (Table S1, entry 2, ESI †), addition of CO 2 increased the yield of 1b to 81% (Table 1, entry 17).
In the mechanism in Scheme 1, step 2 involves the generation of an alkoxide I and the parallel attack of the carbene-CO 2 adduct on CH 2 Br 2 after activation of CO 2 by the carbene in step 1. Nucleophilic attack of the alkoxide I on intermediate II in step 3 results in the elimination of the leaving group and formation of intermediate III. In step 4, the secondary hydroxyl group of the diol is deprotonated, leading to the generation of intermediate IV and, in the final step (step 5), the intramolecular addition of the alcohol occurs in intermediate IV, which affords the cyclic carbonate and regenerates the catalyst. Notably, bromomethanol is eliminated as a leaving group, however bromomethanol is unstable, and hence it is believed to decompose to a mixture of hydrogen bromide (HBr) and formaldehyde (CH 2 O). 37 Note, the formation of these side-products was not detected by spectroscopic or chromatographic studies, possibly due to neutralization of HBr by Cs 2 CO 3 and the volatility of CH 2 O.
The secondary (non-catalytic) reaction in Scheme 2 proceeds by the attack of Cs 2 CO 3 on C 4 H 9 Br in step 1, leading to the formation of intermediate II In summary, the work presented here offers an approach for the synthesis of cyclic carbonates from diols and CO 2 . The proposed system benefits from the use of environmentallyfriendly metal-free carbene catalysts. Using this methodology cyclic carbonates were obtained under mild conditions (90 1C and atmospheric pressure of CO 2 ) in good yield and comparable or better to those obtained with other catalysts that operate under more forcing conditions. Based on labelling studies and other experiments two-mechanisms are proposed, one non-catalytic and one catalytic that account for the overall reaction. Scheme 1 Tentative mechanism for the carbene-catalyzed reaction of diols and CO 2 to form cyclic carbonates. The substituents of the catalyst are omitted for clarity.
Scheme 2 Proposed mechanism for the non-catalytic reaction of diols and CO 2 to form cyclic carbonates.