Zhiqiang Zhang,
Liancai Xu* and
Wenkai Feng
Department of Material and Chemical Engineering, Zhengzhou University of Light Industry, China. E-mail: miss_xulc@126.com
First published on 15th January 2015
Azaphosphatranes were reportedly efficient and metal-free catalysts for CO2 fixation to epoxides, however, the mechanism remains unclear. DFT investigations reveal that intermolecular proton transfer is essential for the reaction while CO2 insertion into the P–N bond of the catalyst will result in catalytic deactivation.
The synthesis of PC in industrial scale is usually carried out using Lewis acid or base catalysts, which require high temperatures and pressures. The conditions have limited the process in terms of energy and economics. As the utilization of PC is substantially increased due to that the PCs are widely used as electrolyte components in lithium batteries, polar aprotic solvents, and intermediates in the production of pharmaceuticals and fine chemicals,11,12 new commercially viable catalysts and processes which can be operated under atmospheric pressure and close to room temperature are required to minimize the energy costs of the PCs production. Many efforts have been made to study the catalytic systems for the production of PC. Most of the catalysts are metal complexes.13–16 Besides metal compounds, bromine,17 KI,18,19 N-heterocyclic compounds,20–22 and ionic liquids,23–27 were also reported to be effective in production of PCs. However, in most of these cases, additives or cocatalysts as well as organic solvents are usually required.
Recently, Chatelet and his coworkers reported that azaphosphatranes, also known as Verkade's superbases,28–35 can serve as single-component, metal-free organocatalysts for the production of PCs from CO2 and styrene oxide at atmospheric pressure and the temperature of 80–100 °C.36 They also proposed a mechanism in which CO2 was activated via insertion into P–N bond of catalyst. However, their mechanism seems contradict with their kinetic observations that the catalysts with bulky P–N bond protecting group exhibit high catalytic activity. Two questions naturally arise: whether the CO2 insertion is essential for the catalytic reaction? Which property of the catalyst is related to the catalytic activity?
In this contribution, we theoretically investigated the reported reaction36 with simplified model (see Chart 1) at 100 °C, 1 atm, and in toluene solution. The structures of catalyst and its derivatives were shown in Chart 2. All geometries of minima and transition states were optimized at B3LYP/6-31G(d) theoretical level for gas phase molecules. For all molecules, the electronic energy (Eelectron) was improved by a single-point calculation at B3LYP/6-311++G(2d,p) level. For an intermediate or transition states, the Gibbs energy in gas phase (Ggas) was calculated as eqn (1):
Ggas = Eelectron + Edispersion + ZPE + Gcorrect,gas | (1) |
Gsol = Ggas + Gcorrect,sol | (2) |
For the catalyst C1a, the Gibbs energy profiles for this catalytic reaction were depicted in Fig. 1. The reaction starts with the epoxide ring-opening step in which the secondary carbon of 1 is attacked by chloride, whilst the phosphonium of C1a supplies a proton to the ring-opened 1 to stabilize the intermediate 3. This elementary step is exothermic in gas phase (ΔrGgas = −9.08 kcal mol−1) while endoenergic in solution by ΔrGsol = 0.94 kcal mol−1. The energy barriers for this step are 35.27 and 26.90 kcal mol−1 in gas phase and in solution, respectively. In intermediate 3, the O–H bond of 1 points towards to the P atom of C1a resulting in a short O–H⋯P (the distance between H and P is 2.43 Å).
The secondly elementary step is CO2 addition which starts with the approaching of CO2 to ring-opened 1. When the C atom of CO2 gets close to the O atom of ring-opened 1, the transition states TS3-4 formed. In this stage, the C⋯O distance is 1.75 Å indicating no covalent bond is formed, whilst the O–H bond is elongated to 1.37 Å and the H⋯P distance is decreased to 1.59 Å. This addition reaction has a large barrier (Δ‡Ggas = 30.97, Δ‡Gsol = 28.32 kcal mol−1). However, the reverse reaction has an extremely small barrier (Δ‡Gsol = 9.79 kcal mol−1). In adduct 4, the C–O bond has been formed between 1 and CO2, whilst the proton returns to the P atom of C1a. The formation of 4 is an endothermic process whether in gas phase or in solution.
The last elementary step is the conversion from 4 to 2 in which the carbonate group attacks the secondary carbon of the ring-opened epoxide followed by the chloride leaving and lactone ring closure. After the separation of cyclic carbonate and azaphosphatrane, the catalyst is regenerated and the catalytic cycle is completed. This step requires a relatively low barrier (Δ‡Ggas = 14.85, Δ‡Gsol = 9.48 kcal mol−1), and it is an exothermic reaction. Among the three steps, the CO2 addition is the rate determining step. The intermolecular proton transfer is completed in the first two steps of the catalytic reaction.
In order to test the role of the catalyst, the objective reaction was investigated using the same method but with the absent catalyst, see page S4–S5, (ESI†). Without the catalyst, actually a proton donor, the epoxide ring-opening step cannot occur due to the highly reactive oxide anion in the ring opened intermediate. After CO2 addition to the ring-opened epoxide, the following ring-closure step for the formation of cyclic carbonate requires releasing the proton. Thus, the intermolecular proton transfer is required by the reaction. The protonated phosphonium moiety of the catalyst is the active site which acts as a proton transfer station in the catalytic cycles.
The P–N bond is reportedly sensitive to CO2, which has been approached to CO2 capture.41 The catalyst C1a is able to be converted to a tricyclic phosphorylcarbamate structure (C2a) via insertion of CO2 into the P–N bond. The CO2 insertion into the catalyst C1a was subsequently investigated to evaluate its influence on the catalytic reaction. Fig. S2a and S3a (ESI†) present the Gibbs energy profiles of the CO2 insertion into acidic and basic form of C1a, respectively. According to the DFT calculations, the basic form of C1a is more sensitive to CO2 insertion than its acidic form. The computed activation barrier (Δ‡Ggas = 37.26, Δ‡Gsol = 32.78 kcal mol−1) for the CO2 insertion into P–N bond is relatively higher than the barrier for the rate determining step of the objective reaction catalyzed by C1a, it is also within a realistic range for a reaction occurring at 100 °C. Thus, the CO2 insertion into the basic form of C1a is a side reaction.
The formation of C2a through CO2 insertion into C1a is endothermic by about 35 kcal mol−1 (as shown in Fig. 2) which indicated the process is not thermodynamically allowed. However, we still investigated the catalytic activity of C2a to reveal the structure–activity relationship of azaphosphatranes. The Gibbs energy profiles for this reaction were depicted in Fig. 2. The catalyst C2a obviously lowers the barrier of the first step compared with C1a, nevertheless, it significantly heightens the barriers for the remaining steps. Thus, the catalytic activity of C2a is lower than C1a. The intermolecular proton transfer in the catalytic reaction with C2a is also different from C1a. In the epoxide ring-opening step, the proton transfers from C2a to epoxide, thus the intermediate 5 becomes stable. In this stage, the newly formed O–H bond of 1 points towards to the catalyst's carbonyl O atom instead of P atom. In the CO2 addition step, the proton shifts to the carboxyl of the CO2 adduct rather than the catalyst during the production of 6. The proton returns to the catalyst until the last step. These differences in the intermolecular proton transfer cycle imply that CO2 insertion into the P–N bond reduces the alkalinity of the azaphosphatrane. Furthermore, the catalytic activity may be related with the alkalinity of the catalyst.
To reveal the relationship between the catalytic activity and the alkalinity of the catalysts, a series azaphosphatranes were theoretically investigated. The proton affinity, see Table S1 (ESI†), was firstly calculated since it can reflect the alkalinity of the base. Among the three analogs (C1a–C1c), C1b is the most protophilic azaphosphatrane and followed by C1c and C1a. The order of the calculated proton affinities is in good agreement with the reported catalytic activity (C1b > C1c > C1a).36
On the other hand, the ionization state of an azaphosphatrane, which can greatly change the catalytic reaction pathway, is primarily determined by the alkalinity. Thus, the ionization states of a series azaphosphatranes (see Table 1) were theoretically investigated. Under the reaction condition, the superbase C1a almost entirely exists in its acidic form indicated by an ionization degree of 99.8%. However, it completely changes to the basic form after CO2 insertion indicated by an extremely low ionization degree of <0.001% for C2a. The deprotonation of the active site will result in the alteration of catalytic reaction pathway, because the catalyst can no longer serve as a proton transfer station which is required by the catalytic process. We carefully explored the possible reaction paths with the basic form of C2a as catalyst. However, no realistic one was found. These results suggest that C2a, the CO2 insertion product of C1a, is hard to be yielded under the reaction conditions. Unfortunately, C2a almost entirely exists in its basic form which is inactive, even it has been yielded. Thus, the CO2 insertion into C1a will result in deactivation of the catalyst.
Given the observations from the present DFT calculations and reported precedents,42–44 we proposed the reaction mechanism which is illustrated in Scheme 1. The epoxide ring breaks by the nucleophilic attack of the chloride at the secondary carbon of 1. The ring-opened epoxide is stabilized by capture of proton from the azaphosphatrane. The consequent attack of the ring-opened 1 by CO2 at the resultant alcohol moiety leads to the formation of CO2 adduct, whilst the catalyst is regenerated by recapture of proton from CO2 adduct. Subsequent ring-closure forms the product 2. There is an equilibrium existing between the acidic and basic form of azaphosphatranes. Although the basic form is not dominated compared with the acidic form, the insertion of CO2 to the basic form will result in a displacement of the equilibrium to the left which cause the consumption of the catalyst. The present mechanism can explain the thermodynamic observations reported by Chatelet and his coworkers that C1a deactivated in a few hours but C1b and C1c did not under the same condition.36 It is probably due to that the bulky substituents in C1b and C1c act as protecting groups avoiding CO2 insertion into the P–N bond of the catalyst. This was also confirmed by the calculations results (Fig. S2b and c and S3b (ESI†)). Chatelet et al. also proposed a reaction mechanism in which the CO2 is activated via insertion into P–N bond of azaphosphatrane and subsequently attacked by ring-opened epoxide compound to yield the final product through a ring-closure step.36 Their mechanism seems contradict with their kinetic observations, while those observations are in good agreement with the present mechanism proposed based on our DFT calculations. According to the present mechanism, the chloride of C1a acts as a nucleophile to attack and break the epoxide ring, while the protonated azaphosphatrane, the cation part of C1a, acts as a proton transfer center to facilitate the following cyclic carbonate production. We think that an appropriate nucleophile, such as chloride or bromide but not confined to halogens, and an adequate base which is protonated and insensitive to carbon dioxide can serve as a catalyst for cyclic carbonate synthesis. This needs to be confirmed by further experimental and theoretical studies. However, it supplies us with an idea of the design of catalysts for cyclic carbonate synthesis.
Natural Science Foundation of China (U1204209, 21101142) and Projects in Henan Province department of Education (12A150027) are gratefully acknowledged.
Footnote |
† Electronic supplementary information (ESI) available: Computational details, electronic energies, corrections to Gibbs free energy and Cartesian coordinates. See DOI: 10.1039/c4ra11081a |
This journal is © The Royal Society of Chemistry 2015 |