Cationic Polymers Bearing Quaternary Ammonium Groups-Catalyzed CO₂ Fixation with Epoxides

It has been well documented that the nature of the halide counter anion can play a crucial role in such ammonium salt catalyzed CO₂-incorporation reactions [48, 49] and we thus tested Cl and I-containing polymers A–F for this study. Our investigations started by using simple styrene oxide 1a (R = Ph, Scheme 1) as the epoxide component and carrying out the CO₂ fixation under solvent free conditions with an atmospheric pressure of CO₂ (by using a simple balloon) (Table 1).

As the developed ionene synthesis provides the chloride salts directly we tested these compounds as catalysts in the beginning (entries 1–15). The first results obtained when using the 1,4-diazabicyclo[2.2.2]octane (DABCO)-based ionenes A were rather disappointing as no dioxolane 2a was formed at 60 °C (entries 1 and 2) and only small amounts were formed after 4 h at 120 °C (entries 3–5). The same rather low reactivity was then observed by using the B-chloride series with a C2 ammonium group linker, (entries 6–8). By using the other polymers C–F under these conditions however, some very interesting observations were made (entries 9–15). With increasing linker length the catalytic performance of the ionene chlorides improved significantly, especially for the 1,4-regioisomers like compound C^1,4 (entry 11) and compound D^1,4 (entry 13). In both cases catalyst quantities of 0.5–1 mol% allowed us to reach conversions of more than 50% after 4 h reaction time (120 °C), and almost complete conversion when doubling the reaction time in the presence of 0.5 mol% D^1,4 (entry 14). In addition, also the cyclohexyl-based polymer E and the pyridinium-based F showed some promising initial catalytic potential (entries 15, 16).

To investigate the influence of the counter anion we next tested the corresponding iodide-based polymers (entries 17–24). These salts were prepared by carrying out a chloride to iodide exchange of the initially used salts (see the supporting information for details). Very interestingly, this lead to a significant improvement in case of the A and B series (entries 17–20), which performed rather slow with chlorides only. On the other hand, those polymers that performed well for the chloride salts already (i.e. C^1,4 and D^1,4) performed less satisfactory when using the iodides instead (entries 21,22). According to mechanistic studies by others [48, 49], the nucleophilic counter anion plays a crucial role in the initial epoxide opening. The results obtained in our experiments may thus be rationalized by different halide-binding affinities of the different polymers. While for the shorter ionenes it seems reasonable (based on the observed trends in reactivity) that chlorides are more closely bound to the polymer backbone and thus less readily available for the nucleophilic attack of the epoxide (compared to the bigger and more easily polarizable iodide), it seems that iodides are on the other hand more tightly attached to the polymers in those cases where longer linkers between the ammonium groups are present (thus explaining the different trend in reactivity in these cases). Finally, we tested the use of chloride-based ionenes and adding additional NaI to the reaction mixture (entries 25–30). Interestingly, in these experiments again the A and B series performed clearly superior compared to the sole use of the Cl-ionenes (compare entries 25 vs. 4, 26 vs. 5, 27 vs. 3, and 28 vs. 8). It should be noted that NaI alone allows for approximately 10% conversion under these conditions in the absence of any catalyst, thus a synergistic effect of the use of NaI together with the cationic polymers was clearly proven herein. The results are again not that much better for the C and D series (entries 29,30) which is in accordance to the observations made when using the isolated salts alone (the slightly higher yields may be rationalized by some minor catalytic contribution of the excess NaI). Finally, it was also shown that addition of NaBr has more or less no positive effect (entry 31).

Altogether this screening of different cationic polymers revealed a rather complex structure-catalytic activity relationship and with respect to ease of operation we thus decided to carry out further studies (i.e. investigation of the application scope) with the chloride-containing ionene D^1,4 under the conditions shown in entry 14. It was also clearly shown that the catalyst can easily be recycled by a simple filtration from the crude reaction mixture and reused for at least four further reactions without any significant loss in activity. As shown in Table 2, the application scope of this reaction is relatively broad and a variety of differently substituted aliphatic and aromatic epoxides performed very well under the catalysis of ionene D^1,4. For comparison, the reactions were all run at 120 °C with 0.5 mol% of the catalyst for either 1, 4, or 10 h. The only real limitation was observed when using nitro-styrene oxide (entry 14), which gave almost no cyclic carbonate but instead decomposed totally under the reaction conditions.

Mechanistically, the CO₂ fixation with epoxides in the presence of nucleophilic catalysts has been discussed and investigated in much detail [23, 40, 48‐50]. It is commonly believed that the epoxide is activated towards nucleophilic attack by the Lewis- or Bronsted-acidic catalyst and that the nucleophilic source opens the epoxide by addition to the less substituted carbon (for two recent mechanistic investigations see Ref. [40, 49]). As a simple test to investigate whether epoxide opening really occurs preferably via nucleophilic addition to the less-substituted carbon or if maybe a more S_N1-type mechanism proceeding via attack to the benzylic position is likely we tested the CO₂-fixation with the deuterium-labelled styrene oxides 1a–D2 and 1a–D1 (Scheme 2). It was clearly shown that the di-deuterated 1a–D2 reacts measurably slower than the parent non-deuterated 1a and the epoxide deuterated in the benzylic position (1a–D1). This kinetic isotope effect is therefore a clear hint that nucleophilic attack indeed occurs at the less-substituted carbon, as proposed by others for different catalyst systems [40, 49].