Experimental and theoretical study for CO2 activation and chemical fixation with epoxides

The synthesis of five-membered cyclic carbonates via catalytic cycloaddition reaction of CO2 with epoxides is considered to be an effective technology for alleviation of the energy crisis and global warming. Various commercial organic bases and ionic salts were used as catalysts, while the relationship of catalytic activity and compound structure has been seldom explored. Herein, a facilely obtained binary catalytic system based on triethylamine/NBu4Br was developed for CO2 activation and chemical fixation. The highly efficient catalytic system showed outstanding conversion and above 99% selectivity under metal-free mild reaction conditions (100 °C, 1 atm) in one hour. The detailed process of CO2 activation and chemical fixation was investigated at the molecular level by a series of experiments and theoretical calculation, which provided a mode for the design and synthesis of a highly efficient catalytic system for conversion of CO2 under mild conditions.


Introduction
Carbon dioxide (CO 2 ) excessive emission is a main component of greenhouse gases in the atmosphere. 1-3 Its conversion to value-added products has offered one promising way to alleviate the energy crisis and global warming. [4][5][6][7] As an economical and abundant C 1 source, CO 2 is renewable, inexpensive and nontoxic, [8][9][10][11] but its chemical inertness is still the bottleneck for its applicability as a raw material in industry. The catalytic synthesis of cyclic carbonate from epoxides and CO 2 is considered to be one of the most promising pathways to CO 2 utilization. [12][13][14] Cyclic carbonates have wide applications as green polar aprotic solvents, fuel additives, and chemical intermediates. [15][16][17][18] The chemical xations of CO 2 into cyclic carbonates are relatively simple from a chemical reactivity point of view, while in fact, the high temperature, high pressure, high catalyst loading, or a combination of these are required to make the reaction effective, which is not economically suitable and poses safety concerns as well. [19][20][21][22] To optimize the reaction conditions, a great variety of catalysts for the synthesis of cyclic carbonates have been developed so far, including alkali-metal halides, metal complexes, metal oxides, cellulose, MOFs, zeolites, ionic liquids, carbon nitride and so on. [23][24][25][26][27][28][29][30][31][32] Though most catalysts were used to activate epoxide by a metal ion or hydrogen bond center, the effect of CO 2 activity on the reaction efficiency has rarely been studied deeply. CO 2 is the highest oxidation state of carbon, and it is thermodynamically stable and kinetically inert, which will consequently hinder the development of efficient catalysts that achieve CO 2 activation and subsequently its functionalization. 23,33,34 Thus the activation of CO 2 is pivotal for its effective conversion. The introduce/use of Lewis basic species and transition metal system have been highly considered, while the detailed CO 2 activation process was unclear, 35-38 so developing a catalytic system that provides molecular level insight for CO 2 activation process is still highly desired. Taking the aforementioned concerns into account, we propose to explore a new catalytic system that can activate and convert CO 2 with epoxides under mild conditions, and serve as an ideal mold for providing a detailed mechanistic understanding of CO 2 activation and xation process.
Herein, a simple and efficient binary catalytic system based on organobase/NBu 4 Br was developed for CO 2 cycloaddition reaction with epoxides under metal-free mild conditions. It was found that the synergistic effect between two components in this new catalytic system promote the cycloaddition reaction occur under atmospheric pressure during a short time period of 1 hour. Moreover, the relationship of catalytic activity and catalyst structure was investigated at molecular level by a series of experiments and theoretical calculation, which could not only offer in-depth understanding of the reaction mechanism but also provide a theoretical basis for the effect of triethylamine (NEt 3 ) in activating CO 2 and promoting the reaction process.

Results and discussion
To understand the effects of the counter anions on the catalytic activity, the catalytic cycloaddition reaction of CO 2 with epoxides were initially investigated in the presence of 0.5 mL NEt 3 . As shown in Table 1, NBu 4 Br afforded a full conversion of 2-(chloromethyl)oxirane to 4-(chloromethyl)-1,3-dioxolan-2-one under 1.0 atm. CO 2 at 100 C for 1.0 h (entry 1), and the catalytic conversion showed unconspicuous change when the NBu 4 Br was replaced by NBu 4 Cl, NBu 4 I, respectively (entries 2-3). Notably, the yield of 2-(chloromethyl)oxirane to 4-(chloromethyl)-1,3-dioxolan-2-one decreased signicantly when NBu 4 PF 6 was used under the same condition, which is probably due to the weakest nucleophilicity of PF 6 À in selected Cl À , Br À , I À , and PF 6 À , suggesting the counter anions play a dominant role in cycloaddition reaction of CO 2 with epoxides. 39 When the common organic base NEt 3 was used alone under the same condition, 36% yield of 2-(chloromethyl)oxirane to 4-(chloromethyl)-1,3-dioxolan-2-one was obtained (entry 6), indicating that NEt 3 can activate CO 2 in this reaction. 40 Based on the above results, the high efficiency of the binary system is probably attributed to a synergistic effect between NBu 4 Br and NEt 3 during the catalytic conversion of CO 2 with epoxides to cyclic carbonates. Additionally, the inuence of the basicity on the conversion of 2-(chloromethyl)oxirane was investigated in the presence of NBu 4 Br with various organic bases at 100 C for 1 h ( Fig. 1a and  b). As shown in Fig. 1b, the yield of 4-(chloromethyl)-1,3dioxolan-2-one greatly related to the basicity of organic bases, and NEt 3 gave rise to the highest conversion of 2-(chloromethyl) oxirane to corresponding product due to its strongest alkalinity and minimum steric hindrance. The activities of organic bases decreased sharply from 99% to 92 and 63% when the pKa decrease from 18.8 to 10.2, respectively. As reported previously, the base was weaker, the DG of this reaction was lower, 41 so the basicity order might be DIPEA > NEt 3 > TBA > TMEDA > MIm > Py > DMBA. However, the catalytic activity order of the organic bases is not in strict accordance with the established pKa, that was NEt 3 > TBA > Py > TMEDA > MIm > DIPEA > DMBA, which indicates that the basicity of the organic bases is one important factor for promoting catalytic activity, but the steric-hindrance also play a role in this catalytic reaction. 23 The dependence of the cycloaddition reaction of CO 2 and 2-(chloromethyl)oxirane on temperature is shown in Fig. 2a. The results indicated that the activity of this catalytic system is highly dependent on the reaction temperature. In the lower temperature region (25 to 50 C), the yield of 4-(chloromethyl)- a Reaction conditions: 2-(chloromethyl)oxirane (12.8 mmol), NBu 4 X (X ¼ Br À , Cl À , I À , PF 6 À ) (0.06 mmol), NEt 3 (0.5 mL), CO 2 (1 atm), 100 C. b Yield was determined by GC and 1 H NMR. The possibility of byproduct was 3-chloro-1,2-propanediol.  1,3-dioxolan-2-one increases slowly with increasing temperature. A further increase in temperature from 50 to 100 C has signicant effects on the 2-(chloromethyl)oxirane conversion, and gave the target product in 99% GC yield at 100 C for 1 h.
The kinetic curve for catalytic conversion of CO 2 into cyclic carbonates was also investigated in the presence of NEt 3 / NBu 4 Br at 100 C. As shown in Fig. 2b, the yield of 4-(chloromethyl)-1,3-dioxolan-2-one increased rapidly in the rst 40 min and then went up slowly. The complete consumption of 2-(chloromethyl)oxirane and synchronous formation of the desired product was achieved in 1 h. It should mentioned that the selectivity remains above 99% in the entire catalytic process. It could be obviously seen that the reaction pressure showed a great effect on the cycloaddition reaction (Fig. 2c). With the increase of CO 2 from 0.25 to 1 atm, the 4-(chloromethyl)-1,3dioxolan-2-one yield increases from 20 to 99%. A further increase in the CO 2 pressure from 1 to 1.5 atm results in a same level in 2-(chloromethyl)oxirane conversion. A similar effect of CO 2 pressure on catalytic activity was observed in other related catalytic systems. 43,44 The inuence of NEt 3 to NBu 4 Br ratio on the yield of 4-(chloromethyl)-1,3-dioxolan-2-one was also investigated at 100 C for 1 h with xed NBu 4 Br (0.06 mmol). As shown in Fig. 2d, when the ratio of NEt 3 to NBu 4 Br increases from 1.6 to 18.2, the 4-(chloromethyl)-1,3-dioxolan-2-one yield increased rapidly. Then the 2-(chloromethyl)oxirane conversion stayed almost constant when ratio of NEt 3 to NBu 4 Br increased further.
To further understand the underlying principles of CO 2 activation, the computation by the DFT (M06-2X) calculations was studied. Preliminary calculations indicated no involvement of NBu 4 + cation (Fig. S1 in ESI †), which was consequently neglected from the elaborate calculations reported herein. 47 With bromide as catalyst (Fig. 3), the ring opening through the attack of a nucleophile on epoxide and CO 2 addition takes place simultaneously in a TS1, which is considered to be the ratedetermining step with a free energy of 22.3 kcal mol À1 . Subsequently, this reaction tended to undergo the carbonate ringclosure step ( a TS2) has the energy of 10.8 kcal mol À1 compared to the reactants. This is consistent with the inefficient reaction under the mild experimental conditions. The efficiency of the reaction is promoted signicantly when NEt 3 is introduced as the catalyst in addition to the system. To gain further insight into the synergetic catalytic role played by NEt 3 and NBu 4 Br, more detailed calculations were carried out. When the reaction was calculated in NEt 3 solution, there different mechanisms were revealed (Fig. 4). The bromidecatalyzed process (black energy prole shown in Fig. 4) in NEt 3 solution shows the same transition states and intermediates as in gas-phase. But the energy barrier of rate-determining step is 31.1 kcal mol À1 , which is 8.8 kcal mol À1 larger this procedure without NEt 3 solvent. Also the carbonate ring-closure step ( a TS2) in NEt 3 solution has an energy of 18.2 kcal mol À1 , which is 7.4 kcal mol À1 larger than bromide-catalyzed process in gas. For comparison, we also calculated the reaction with NEt 3 as catalyst in the absence of bromide (red prole). The NEt 3 -catalyzed epoxide ring-opening procedure has a high energy barrier ( b TS1, 42.4 kcal mol À1 , red prole in Fig. 4), which is 11.3 kcal mol À1 larger than bromide-catalyzed transition state in NEt 3 solution. Aerwards, the CO 2 addition step to form b TS2 has an energy of 46.0 kcal mol À1 relative to the energy of the reactants. Finally, this mechanism of ring-closure process may undergo a b TS3 transition states, which has an energy of 44.7 kcal mol À1 above the reactants. However, the energy barrier of 42.4 kcal mol À1 is quite high and consistent with a sluggish reaction and unable to take place under mild experimental conditions, 48 indicating that NEt 3 does not merely act as a solvent but also inuences the course of the reaction. Thus, the mechanism when NEt 3 participation tends to a bromine and NEt 3 jointly catalyzed process (blue prole shown in Fig. 4), a new intermediate C 1 appears, resulting from the CO 2 addition upon interaction with NEt 3 and bromine. This rate-determining step presented no transition states and has an energy of 17.5 kcal mol À1 the above reactants, 4.8 kcal mol À1 energy favorable than pure bromine-catalyzed process in gas-phase, 13.8 kcal mol À1 decrease than bromine-catalyzed procedure in NEt 3 solvent, which allows the reaction to be performed under mild conditions. A potential energy surfaces scan calculation based on C 1 intermediate validates transition states nonexistent (Fig. S3 in ESI †). It is conceivable that the ringopening of epoxide and CO 2 addition might undergo synchronously with two catalysts. Aerwards, ring-closure takes place through c TS1. NEt 3 then dissociates from the reacting system, which reverts to A 1 and evolves to product through a TS2 with a free energy of 27.7 kcal mol À1 above the reactants. The calculated rate constant is 1.850 Â 10 À3 s À1 at 373 K. Then the corresponding half-life is ten minute, which accord with the experimental data. The intermediate of A 1 with C 1 were selected to compare the structure changes in reaction, the C-O bond of CO 2 change slightly as shown in Fig. S4, † while the C-O bond between C atom of CO 2 moiety and O atom of epoxide moiety decreases from 1.46Å to 1.43Å, then the interaction between CO 2 and epoxide moieties enhanced with NEt 3 as solution and catalyst. Moreover, the Mulliken charge of carbon atom of CO 2 moiety decreases from 0.3 to 0.07 when compared C 1 with A 1. Therefore, the bond length and charge analysis suggest that the lone pair electron of nitrogen in NEt 3 stabilized the formation of C 1 and c TS1. Overall, a plausible mechanism for triethylamine-promoted catalytic conversion of CO 2 into cyclic carbonates has been proposed based on the aforementioned results (Fig. 5). First, the epoxide ring opens through nucleophilic attack on the less sterically hindered b-carbon atom by Br À to produce an alkoxide ion, and simultaneously CO 2 is activated by NEt 3 via electrostatic interaction to form the carbamate salt. Then a new intermediate C 1 is produced resulting from the nucleophilic attack on carbamate salt by alkoxide ion. Finally, cyclic carbonate is produced by following intramolecular ring-closure reaction of C 1 .

Conclusions
An efficient binary catalytic system containing triethylamine and NBu 4 Br was screened for catalytic conversion of CO 2 and epoxides into cyclic carbonates under metal-free mild conditions. Especially, NEt 3 could activate CO 2 via electrostatic interaction and remarkably reduce the reaction energy to promote the reaction in the catalytic system. This work not only presents a simple and useful route for CO 2 chemical xation into high-value chemicals, but provides a detailed mechanistic understanding of CO 2 activation and xation process.

Experimental
General Chemicals including NEt 3 , NBu 4 Br, NBu 4 Cl, NBu 4 I, NBu 4 PF 6 , epoxides and CO 2 are commercially available and used directly without further purication. Gas chromatography (GC) was performed on a Shimadzu GC-2014 equipped with a capillary column (RTX-5, 30 m Â 0.25 mm) using a ame ionization detector.

Typical catalytic reaction
The cycloaddition reaction was carried out by magnetic stirring, trimethylamine (0.5 ml), NBu 4 Br (0.06 mmol) and 2-(chloromethyl)oxirane (12.8 mmol) were added into a reactor at room temperature. Then, the reactor was sealed and purged with CO 2 to remove air. CO 2 was introduced into the reactor and the pressure was adjusted to 1 atm at room temperature. The reactor was placed into pre-heated oil bath and temperature was maintained at 100 C. Aer the reaction was completed, the reactor was cooled to 0 C in ice-water bath, and then the excess of CO 2 was carefully vented. The mixture was diluted with ethyl acetate. The conversion of epoxide and yield of cyclic carbonate were determined by gas chromatography (Shimadzu GC-2014, a ame ionization detector) and 1 H NMR.

Computational details
The M06-2X functional 49 was employed in this article to perform all the calculations. Our structure optimizations were as follows. In NBu 4 Br involved reaction, only gas-phase calculations were performed. For carbonate formation with bromine as catalyst, the structures were rstly optimized base on the level of 6-31+G(d,p) in gas-phase, then the solvent structure optimizations were carried out based on gas-phase results. As for the rest of other structure optimizations, the calculations were performed in solvent. Vibrational frequency analyses at the same basis sets were used on all optimized structures in order to characterize stationary points as local minima or transition states. Furthermore, the intrinsic reaction coordinate (IRC) calculations at the same level have been applied to validate that transition states connect appropriate reactants and products. The Gibbs free energy were further calculated by single-point energy calculations using M02-2X/6-311+G(d,p) method on previously 6-31+G(d,p) structures and thermal corrections at 298.15 K and 1 atm. No conformational sampling calculations were performed in this work. The continuum SMD model 50 was applied. The Gaussian 09 package 51 was used for all of our calculations in NEt 3 solvent.

Conflicts of interest
There are no conicts to declare.