Cross metathesis with acrylates: N‐heterocyclic carbene (NHC)‐ versus cyclic alkyl amino carbene (CAAC)‐based ruthenium catalysts, an unanticipated influence of the carbene type on efficiency and selectivity of the reaction

Olefin metathesis has been widely explored as a handle for chemical diversification, a feature critical across chemical sectors. Cross metathesis (CM) with acrylic acid derivatives is an example of important but, due to the low catalyst's efficiency, industrially non‐utilized transformation. Here we report on systematic evaluation of ruthenium‐based catalysts bearing N‐heterocyclic carbene (NHC) or cyclic alkyl amino carbene (CAAC) ligands in cross metathesis with methyl acrylate. Dramatic influence of the carbene type on the reaction's efficiency and selectivity has been found. Density functional theory (DFT) calculations suggest that the kinetic selectivity is the main factor differentiating NHC‐ and CAAC‐based ruthenium complexes. Productive turnover number (TON) of 49 900 at 10 ppm loading of nitro‐substituted Hoveyda‐Grubbs complex (nitro‐Grela catalyst) was obtained in the studied reaction, representing the highest efficiency reported to date for this transformation. High efficiency and selectivity of nitro‐Grela catalyst was then utilized in cross metathesis of trans‐anethole with 2‐ethylhexyl acrylate to efficiently produce octyl methoxycinnamate (86 % yield), an antioxidant used in sunscreen formulations.

Olefin metathesis has been widely explored as a handle for chemical diversification, a feature critical across chemical sectors. Cross metathesis (CM) with acrylic acid derivatives is an example of important but, due to the low catalyst's efficiency, industrially non-utilized transformation. Here we report on systematic evaluation of ruthenium-based catalysts bearing Nheterocyclic carbene (NHC) or cyclic alkyl amino carbene (CAAC) ligands in cross metathesis with methyl acrylate. Dramatic influence of the carbene type on the reaction's efficiency and selectivity has been found. Density functional theory (DFT) calculations suggest that the kinetic selectivity is the main factor differentiating NHC-and CAAC-based ruthenium complexes. Productive turnover number (TON) of 49 900 at 10 ppm loading of nitro-substituted Hoveyda-Grubbs complex (nitro-Grela catalyst) was obtained in the studied reaction, representing the highest efficiency reported to date for this transformation. High efficiency and selectivity of nitro-Grela catalyst was then utilized in cross metathesis of trans-anethole with 2-ethylhexyl acrylate to efficiently produce octyl methoxycinnamate (86 % yield), an antioxidant used in sunscreen formulations.
The environmental impact and limited resources of petrochemicals drive the shift towards renewable, efficient and waste averse industrial processes. Utilization of plant oils, a natural alternative for fossil-derived chemicals remains a focus of both, the chemical industry and academia. [1] The broad availability of these raw materials, their inherent chemical functionalities along with competitive costs make them desirable in a variety of commercial applications. [2] The enormous potential of olefin metathesis in chemical industries inspired efforts to develop efficient processes implementing this technology. [3] The requirements for economically viable processes are particularly stringent in production of commodity and specialty chemicals. As estimated by The Dow Chemical Company, turnover numbers of at least 50 000 and 35 000, respectively, are required for economical production of these materials. [4] Despite the intense research over past decades, the efficiency of metathesis catalysts remains unsatisfactory for many transformations, prohibiting its wider application.
Cross metathesis (CM) offers an efficient pathway to introduce diverse functionalities thus allowing a straightforward access to higher value, multifunctional compounds. The cross metathesis with acrylates has been widely studied in the context of depolymerization, [5] transformation of oils and their derivatives [6] as well as other natural products. [7] These type III olefins (according to Grubbs' classification) [8] are known for their low metathesis reactivity therefore cross metathesis is regarded challenging and requires concerted optimization efforts that include selection of a matching metathesis catalyst. Phosphinecontaining catalysts e. g. Grubbs 2nd generation catalyst (1, Figure 1) are considered incompatible with electron deficient metathesis partners, since phosphine ligands released into the reaction mixture upon metathesis initiation lead to undesired side reactions. [9] As demonstrated by e. g. Fogg et al. [9c] and Lipshutz et al. [9d] efficient scavenging of phosphines during CM with acrylates resulted in dramatic improvement of performance of catalyst 1. However, the reaction has not been optimized to the level acceptable by the industry as maximal TONs of less than 1 000 were achieved.
On the other hand, phosphine-free, Hoveyda type catalysts allowed for relatively efficient CM with acrylates. [10] Miao et al. investigated synthesis of polyamide precursors from renewable 10-undecenenitrile and methyl acrylate via olefin CM. [6d] The best productivity was obtained using continuous injection of second generation Hoveyda-Grubbs catalyst 2 (50 ppm) into the reaction mixture at 100°C. Under these conditions 63 % yield of the fatty nitrile was obtained which corresponds to TON of 12 600. Abbas and Slugovc screened a series of ruthenium catalysts in cross metathesis of 1,9-decadiene with methyl acrylate and reached full conversions with 100 ppm per double bond by using catalyst 3. [11] In this case, however, the good efficiency came at the price of long reaction times (4 h).
In our recent work on cross metathesis of ethyl 10undecenoate with acrylonitrile we found that substitution of the NHC ligand in the ruthenium catalyst by a properly selected cyclic alkyl amino carbene (CAAC) ligand (e. g. catalysts 6 and 7, Figure 1) led to superior catalytic productivity. [12] For CAACbased complex 6 TON of 28 500 was observed whereas catalyst 5 bearing N-heterocyclic carbene (NHC) ligand reached TON of only 12 000. Experimental results and DFT calculations linked improved performance of 6 with higher stability of the catalyst's active forms and the related resistance to decomposition induced by acrylonitrile. Importantly, kinetic selectivity towards cross metathesis product was similar for 5 and 6, whereas very stable and typically efficient complex 7 performed poorly due to its low kinetic selectivity. [12] The unexpected results of our study related to cross metathesis with acrylonitrile prompted us to investigate cross metathesis with acrylates. Two classic catalysts from NHC-based group (4 and 5, Figure 1) were compared to representative ruthenium catalysts from CAAC-based series (6 and 7, Figure 1). The goal of this exploration was to find the most efficient and selective catalyst and optimal reaction conditions for cross metathesis with methyl acrylate, as well as to understand factors which determine the catalyst's performance. Ultimately, the most favorable catalyst for this industrially important reaction should assure economically relevant efficiency expressed in high TONs.
Bulk chemical production requires careful consideration of economics in industrial processes development. In the case of commodity and specialty chemicals, well-defined rutheniumbased metathesis catalysts come at relatively high cost per kg compared to the product's price per kg, thus the catalyst loading often becomes one of the key contributors to the overall costs of the process. The most relevant parameter to estimate catalyst performance is the productive turnover number (TON). Reaction yields, although important, can be less critical, as large-scale processing often allows recycling of unreacted substrates. Therefore, in this study, the reaction conditions were adjusted to avoid full conversions in order to register fine performance differences in the selected catalytic systems.
According to the results reported in the literature methyl acrylate (9) shows no poisoning effect on Ru-based catalysts (this applies to Hoveyda-Grubbs type catalysts) [10] while its excess improves the reaction selectivity. Therefore, we ran initial experiments with 5 equivalents of this reagent using toluene as a solvent. While high conversions in all cases were observed, selectivity differences were striking. Contrary to the results observed in CM with acrylonitrile, CAAC-based catalyst 6 proved to have dramatically low selectivity (24 %) towards desired CM product 10 (Table 1, entry 3). On the other hand, complex 7 showed very high, 89 % selectivity, albeit to self-metathesis product 11, which makes it useless in CM with 9 (selectivity toward desired CM product 10 was 11 %, Table 1, entry 4). However, it is also worth noting that this intriguing and unique type of selectivity could potentially be used in some specific synthesis cases where metathesis with an acryloyl group would be undesired.
Both NHC-based catalysts were much more selective than CAAC catalysts (Table 1, entries 1 and 2). Unexpectedly, the nitro substituted Hoveyda-Grubbs complex 4, commonly known as nitro-Grela catalyst, [1] provided noticeably higher productive turnover number (TON 10 34 110) (Table1, entry 1) than its ChemCatChem Full Papers doi.org/10.1002/cctc.202001268 bulkier derivative 5 (TON 10 26 784) (Table1, entry 2). Therefore, we decided to optimize the reaction conditions for catalyst 4 with the aim to maximize TON 10 ( Table 2). Following the findings of Abbas and Slugovc [11] we setup the experiments without typical organic solvent (toluene). Instead, even higher excess of methyl acrylate (8 equivalents) was used to serve both as a reagent and solvent. As could be expected selectivity increased from 77 % to 88 %, but at the same time conversion dropped from 89 % to 74 %, leading to TON 10 lower than that observed in the initial test ( Table 2, entry 2). Very high improvement in efficiency, especially at 10 ppm catalyst loading, was gained upon temperature increase to 80 or 90°C ( Table 2, entries 4 and 5). Observed at 90°C TON 10 of 49 967 at 10 ppm catalyst loading (59 % yield) ( Table 2, entry 5) and TON 10 39 480 at 20 ppm catalyst loading (89 % yield) ( Table 2, entry 6) substantially outperformed results previously reported for CM with acrylates. Sharp decrease in TON 10 was noted at 100°C, which was most probably due to the fast catalyst initiation and short life time of the active species at this temperature ( Table 2, entries 7 and 8). Rather surprisingly, similar optimization protocol did not result in improved TON 10 with bulky complex 5 (see ESI, Table S1).
Further studies were performed in order to understand reasons for poor TON 10 obtained in CM between 8 and 9 with CAAC-bearing complexes 6 and 7. First, the efficiency of examined complexes in ring closing metathesis (RCM) of diene 12 was determined (Table 3). We assumed that catalysts initiate via association of the terminal, more electron-rich double bond to the activated catalysts which, after release of the product, leads to the formation of methyl ester substituted methylidenes   1  70  10  22  84  19  18 649  2  70  20  74  88  65  32 594  3  80  10  47  84  40  39 680  4  80  20  84  89  74  37 048  5  90  10  59  85  50  49 967  6  90  20  89  89  79  39 480  7  100  10  23  79  18  18 025  8  100  20  30  80  24  12 105 [a] Results at 10 and 20 ppm of the catalyst obtained in the same experiment in neat 8; samples measured at 30 and 90 min reaction times respectively.  (Table 3). No significant differences in TON were observed for complexes 4-7. This result together with high total TON's achieved by catalysts 6 and 7 in CM between olefin 8 and acrylate 9 allowed us to conclude that the poor selectivity of CAAC based catalysts in CM with 9 is not related to the poisoning effect of acrylate. Next, we performed theoretical studies to gain mechanistic understanding of the reaction of interest. Due to the similar selectivity represented by both NHC-based catalysts and both CAAC-based catalysts we performed calculations for the three representative structures, namely 4-6. We used DFT calculations on the crucial intermediates in the catalytic cycle to assess their relative Gibbs free energies and use them to obtain thermodynamic ratios of the final products. The schematic representation of the most important intermediates in the initiation part of the catalytic pathways leading to the homo-metathesis products 11 and dimethylfumarate is presented in Scheme 1, while in the pathway leading to the cross-metathesis product 10 in Scheme 2. It is worth noting that using this approach we can estimate the selectivities of catalysts based on Gibbs free energies of different intermediates, e.g. taking into account only those with associated olefins (int1, int3 …) or only ruthenacy-clobutanes (int1_mcb, int3_mcb …). After performing all calculations we decided to base our estimates on intermediates with associated olefins, as their relative Gibbs free energies are higher than those of ruthenacyclobutanes for virtually all investigated cases. At the end of this section we also shortly considered selectivities obtained from ruthenacyclobutane intermediates for the sake of completion.
The initiation of the entire catalytic cycle commences with the association of either ethyl undecenoate 8 (int1) or acrylate 9 (int6) to the activated precatalyst. All three examined complexes react preferentially with 8 to provide intermediate int2 in paths 1 and 3. In these pathways the association of 8 to int2 is favored over the association of 9 by approximately 0.5-1.1 kcal mol À 1 for all three catalysts and thus mostly the side product 11 is formed in this step, accompanied by the generation of ruthenium methylidenes (int4) which proved to be the critical catalytic species. Based on the calculated Gibbs free energies of the intermediates (Table 4) we can estimate the 10/11 ratios in the first phase of the catalytic cycle (initiation) only to be 33 : 67 for complex 6, 26 : 74 for complex 5 and 33 : 67 for complex 4, which are not in agreement with the Scheme 1. A schematic representation of the most important intermediates in the catalytic cycle initiation phase and the first half of the catalytic cycle (from the precatalyst to the methylidene intermediate with the formation of the product) of the homo-metathesis of either ethyl 10-undecenoate (8) (PATH 1) or acrylate 9 (PATH 2). Numbers represent the relative Gibbs free energies of the intermediates with respect to the corresponding precatalyst (in kcal/mol). experimental data presented in Table 1 (see the ESI for the details of these calculations).
The association of 8 to int4 leading to the formation of int11 and, in the next step, the unselective int2 is favored over association of 9 to form int12 by ca. 2 kcal mol À 1 for 6 (see Scheme 3 and Table 4). For 5 and 4 the situation is the opposite, with association of 9 to form int12 being favored by ca. 1-2 kcal mol À 1 over the association of 8 to form int11. The energy differences are the main reason for selectivity contrast observed for both catalysts (Scheme 3). For 6, int11 is favored over int12 by 2.0 kcal mol À 1 resulting in a 19 : 1 selectivity promoting the association of the ethyl 10-undecenoate. Related to the relatively low kinetic selectivity of int2, the final computationally-estimated ratio of 10/11 is approximately 21 : 79, close to the experimental selectivity of 24 % (24 : 76). On the other hand, the good selectivities of catalysts 4 and 5 stem mostly from the lower Gibbs free energy of int12 which leads to int7. The selectivity in the discussed CM reaction is calculated based on ethyl 10-undecenoate (8). From this perspective int7 offers a full selectivity towards main product even though formation of int9 and generation of dimethyl fumarate is almost equally likely as formation of int8. [13] Selective formation of int6 for catalyst 5 and 4 translates to 91 : 9 and 77 : 23 selectivities towards the main product, respectively. The theoretical value for 4 is in perfect agreement with the experimental one, while for 5 it is noticeably higher that that observed experimentally (64 : 36).
We hypothesized that the gap between the experimental and computational results for 5 can be explained by the reuptake of 2-isopropoxy-4-nitro styrene (the so-called boomerang effect) by int4 to form int10. [14] Such event would reduce selectivity of complex 5, and possibly also 4, since the nonselective intermediate int2 would be formed. On the other Scheme 2. A schematic representation of the most important intermediates in the catalytic cycle initiation phase and the first half of the catalytic cycle (from the precatalyst to the methylidene intermediate with the formation of the product) of the cross-metathesis of 10-undecenoate (8) with acrylate 9. Numbers represent the relative Gibbs free energies of the intermediates with respect to the corresponding precatalyst (in kcal/mol).

ChemCatChem
Full Papers doi.org /10.1002/cctc.202001268 reaction and there may be certainly cases where it plays a vital role in the final ratio of obtained products. A more detailed explanation of the relationship between the structural parameters and differences in relative Gibbs free energies for 4-6 is provided in the ESI.
As mentioned above analogous calculations may be performed for ruthenacyclobutane intermediates to obtain selectivities for the studied catalysts. Based on the results presented in Schemes 1 and 2 we obtained the following values for 10 : 11 ratios: 6 : 94 for 4, 1 : 99 for 5 and 89 : 11 for 6 (see ESI). These values are not in agreement with the experimental data suggesting, that the energetics of the ruthenacyclobutane intermediates does not play a crucial role in the final selectivities of the investigated catalysts.
It must be noted that olefin metathesis is a reversible process and the thermodynamic stability of the final products should also be taken into account. In contrast to our previous study with acrylonitrile our current calculations showed that product 10 of CM with methyl acrylate (8) is favored over dimeric side product 11 by only 0.21 kcal mol À 1 . Considering the expected accuracy of our approach of approximately 1 kcal mol À 1 and lack of poisoning effect of methyl acrylate we can suggest that the kinetic selectivity is the only factor that determines the selectivity observed experimentally. [15] In the final experiment, we attempted to extrapolate the superiority of the nitro-Grela complex (4) in cross metathesis with acrylates to the synthesis of antioxidant octyl methoxycinnamate (17), an active ingredient in sunscreen formulations with the trade name Octinoxate (Table 5). Anethol 15 could not be purified by our standard protocol probably due to the poor thermal stability. Moreover, aryl substituted internal C=C in 15 is much less reactive than terminal C=C in 8 making the catalyst more exposed to impurities present in the starting material. This was reflected in a relatively poor TON of 4 686 at 65 % yield (Table 5, entry 2). Fogg et al. studies showed advantages of applying a phenol-functionalized polymer when phosphine ligands were present in the precatalysts. [9c] The authors demonstrated that the beneficial effect of phenol groups arises from protonation of the enolate formed from the acrylate. Phosphine free nitro-Grela catalysts (4) bypasses this liability. On the other hand, chemists from Sasol demonstrated positive effect of phenols on efficiency of Grubbs 1st generation catalyst and negative effect on Grubbs 2nd generation catalysts. [16] In our hands, productive TON in CM between 15 and 16 was almost doubled in the presence of 1 000 ppm of butylated hydroxytoluene (BHT) ( Table 5, entry 4). Further increase of BHT amount to 1 equivalent allowed full conversion of 15 and 86 % yield with only 100 ppm of catalyst 4 (this result is comparable to the previously reported yields obtained with 1 000-5 000 ppm of complex 1) ( Table 5, entry 6).
In summary, DFT results suggest that the kinetic selectivity plays crucial role in CM of ethyl 10-undecenoate with methyl acrylate due to the lack of thermodynamic driving force towards CM product. Classic NHC-bearing ruthenium benzylidene catalysts are much more selective, and therefore effective in this transformation than the emerging CAAC-based catalysts. The best selectivity and overall reaction productivity were obtained with nitro-Grela catalyst (4) at 90°C with turnover number of 49 967 for 50 % yield (10 ppm of the catalyst) and 39 480 TON for 79 % GC yield (20 ppm of the catalyst). Rather surprisingly, complex 4 outperformed its close, bulkier analogue 5 in terms of productive TON.
The nitro-Grela catalyst (4) was used to efficiently obtain octyl methoxycinnamate (17), an active ingredient in sunscreen formulations. The synthesis was accomplished with 86 % GC yield using only 100 ppm of 4 in the reaction between transanethole (15) with 2-ethylhexyl acrylate (16). Strong positive influence of a free radical scavenger (BHT) on TON in this transformation was revealed.