Generation and Reactions of ε-Carbonyl Cations via Group 13 Catalysis

The generation of ε-carbonyl cations and their reactions with nucleophiles is accomplished readily without transition metal cation stabilization, using the ε-bromide dienoate or dienone starting materials and GaCl3 or InCl3 catalysis. Arene nucleophiles are somewhat more straightforward than allyltrimethylsilane, but allyltrimethylsilane and propiophenone trimethysilyl enol ether each react successfully with InCl3 catalysis. The viability of these cations is supported by DFT calculations.


Introduction
The reaction of electrophilic allyl and propargyl compounds with nucleophiles is a commonly used technique in organic chemistry. However, this chemistry becomes challenging when the system involves an electron-withdrawing group, such as a carbonyl. As a result, the generation and reaction of cations at the site γto a carbonyl or carbonyl equivalent (1 and 2) has seen only limited work, although it constitutes a fundamental form of umpolung chemistry (Figure 1). A modest but growing number of methods have been developed to obtain synthetic equivalents of these species. Propargyldicobalt [1] and allyliron [2] cations bearing electron withdrawing groups have been successfully generated and reacted with nucleophiles at the γ-site and are highly electrophilic. Activated cyclopropanes may serve as γ-carbonyl cation equivalents in the presence of Lewis acids, and they have close to the same level of electrophilicity [3]. Allylpalladiums and -iridiums bearing EWG's are significantly less electrophilic but act catalytically and react well with stronger nucleophiles [4][5][6][7][8]. Methods giving an equivalent overall transformation, involving cationic species but not γ-carbonyl cations themselves, are known [9]. Nevertheless, methodology involving direct generation of γ-carbonyl cations without additional stabilization has remained elusive.

Introduction
The reaction of electrophilic allyl and propargyl compounds with nucleophiles is a commonly used technique in organic chemistry. However, this chemistry becomes challenging when the system involves an electron-withdrawing group, such as a carbonyl. As a result, the generation and reaction of cations at the site γ-to a carbonyl or carbonyl equivalent (1 and 2) has seen only limited work, although it constitutes a fundamental form of umpolung chemistry (Figure 1). A modest but growing number of methods have been developed to obtain synthetic equivalents of these species. Propargyldicobalt [1] and allyliron [2] cations bearing electron withdrawing groups have been successfully generated and reacted with nucleophiles at the γ-site and are highly electrophilic. Activated cyclopropanes may serve as γ-carbonyl cation equivalents in the presence of Lewis acids, and they have close to the same level of electrophilicity [3]. Allylpalladiums and -iridiums bearing EWG's are significantly less electrophilic but act catalytically and react well with stronger nucleophiles [4][5][6][7][8]. Methods giving an equivalent overall transformation, involving cationic species but not γ-carbonyl cations themselves, are known [9]. Nevertheless, methodology involving direct generation of γ-carbonyl cations without additional stabilization has remained elusive. Research on vinylogous versions of γ-carbonyl cations, specifically on equivalents of ε-carbonyl cation equivalents (3), is still more scattered (Figure 2). The Green group has reported vinylogous Nicholas reactions involving compounds 4-5 to functionalize the site ε-to the carbonyl or carbonyl equivalent [10]. Activated vinylcyclopropanes (6) can, in principle, serve as ε-carbonyl cation equivalents, but Lewis acid mediated openings of these systems often favor reaction at the γ-site [3,[11][12][13]. Transition metal mediated Research on vinylogous versions of γ-carbonyl cations, specifically on equivalents of ε-carbonyl cation equivalents (3), is still more scattered (Figure 2). The Green group has reported vinylogous Nicholas reactions involving compounds 4-5 to functionalize the site εto the carbonyl or carbonyl equivalent [10]. Activated vinylcyclopropanes (6) can, in principle, serve as ε-carbonyl cation equivalents, but Lewis acid mediated openings of these systems often favor reaction at the γ-site [3,[11][12][13]. Transition metal mediated couplings are, in general, ε-selective, but again are only modestly electrophilic [14][15][16][17][18][19][20][21]. As a consequence, there the is an absence of work on ε-carbonyl cations or their equivalents that features both catalysis and high electrophilicity. Furthermore, the existence of a number of natural products containing ε-arylated carbonyls indicates significant synthetic utility to any methods capable of accessing such cations [22][23][24]. Unlike the γ-carbonyl cations themselves, the further conjugation possible to ε-carbonyl potentially ameliorates the effect of the electron-withdrawing group. As a result, we considered it worth investigating whether the ε-carbonyl cations themselves (3) could be generated, and whether this would be amenable to Lewis acid catalysis.
Molecules 2022, 27, x FOR PEER REVIEW 2 of 13 couplings are, in general, ε-selective, but again are only modestly electrophilic [14][15][16][17][18][19][20][21]. As a consequence, there the is an absence of work on ε-carbonyl cations or their equivalents that features both catalysis and high electrophilicity. Furthermore, the existence of a number of natural products containing ε-arylated carbonyls indicates significant synthetic utility to any methods capable of accessing such cations [22][23][24]. Unlike the γ-carbonyl cations themselves, the further conjugation possible to ε-carbonyl potentially ameliorates the effect of the electron-withdrawing group. As a result, we considered it worth investigating whether the ε-carbonyl cations themselves (3) could be generated, and whether this would be amenable to Lewis acid catalysis. Figure 2. Existing ε-carbonyl cation equivalent precursors.
Molecules 2022, 27, x FOR PEER REVIEW 2 of 13 couplings are, in general, ε-selective, but again are only modestly electrophilic [14][15][16][17][18][19][20][21]. As a consequence, there the is an absence of work on ε-carbonyl cations or their equivalents that features both catalysis and high electrophilicity. Furthermore, the existence of a number of natural products containing ε-arylated carbonyls indicates significant synthetic utility to any methods capable of accessing such cations [22][23][24]. Unlike the γ-carbonyl cations themselves, the further conjugation possible to ε-carbonyl potentially ameliorates the effect of the electron-withdrawing group. As a result, we considered it worth investigating whether the ε-carbonyl cations themselves (3) could be generated, and whether this would be amenable to Lewis acid catalysis. Figure 2. Existing ε-carbonyl cation equivalent precursors.
Scheme 1. Allyl-and dienyl bromide ionization reactions. The results of the calculations were promising. The ionization energy of 8a to dienyl cation 8a+ was unsurprisingly the most favored, the process being 16.8 kcal/mol lower in energy than allyl cation generation. Somewhat to our surprise, the ionizations of the ε-carbonyl cation precursors 8b and 8c also were found to be favored substantially (by 10.9 kcal and 12.7 kcal, respectively), relative to the process with allyl bromide. Finally, the analogous ionization of ethyl 4-bromocrotonate was found to be 6.9 kcal/mol higher in energy than that of allyl bromide, consistent with the difficulty in discrete generation of γ-carbonyl cations. As a result of these findings, we chose to test these observations with an experiment. Given the notably mild conditions reported in the group 13 catalyzed electrophilic reactions of allyl bromides [25][26][27][28], we chose to pursue the analogous approach for ε-carbonyl cations.
The results of the calculations were promising. The ionization energy of 8a to dienyl cation 8a+ was unsurprisingly the most favored, the process being 16.8 kcal/mol lower in energy than allyl cation generation. Somewhat to our surprise, the ionizations of the εcarbonyl cation precursors 8b and 8c also were found to be favored substantially (by 10.9 kcal and 12.7 kcal, respectively), relative to the process with allyl bromide. Finally, the analogous ionization of ethyl 4-bromocrotonate was found to be 6.9 kcal/mol higher in energy than that of allyl bromide, consistent with the difficulty in discrete generation of γ-carbonyl cations. As a result of these findings, we chose to test these observations with an experiment. Given the notably mild conditions reported in the group 13 catalyzed electrophilic reactions of allyl bromides [25][26][27][28], we chose to pursue the analogous approach for ε-carbonyl cations.

Scheme 2. Preparation of phenyl ketone 8c.
In addition, a third substrate chosen for the study was 10, employing an aryl spacer rather than one of the alkene spacers between the ester and bromide. Compound 10 was prepared by the radical bromination of cinnamate ester derivative 11 (10, 77%) (Scheme 3), itself being prepared by the Wittig reaction of o-tolualdehyde [32].

Scheme 3. Preparation of benzylic bromide 10.
Experimental work began with ethyl 6-bromohexadienoate (ethyl 6-bromosorbate, 8b). Test reactions were undertaken with mesitylene (5 equiv) as the nucleophile, and catalytic amounts (10 mol%) of Lewis acids CuCl, SnCl4, InCl3, GaCl3, and BiI3, in CH2Cl2 with 4 Å molecular sieves ( Table 2, Scheme 4). CuCl and BiI3 afforded no product and minimal amounts of product, respectively. Conversely, GaCl3, InCl3, and SnCl4 gave more significant amounts of conversion to 12a over 24 h, although small amounts of starting material remained. Repetition of the reactions at reflux afforded complete starting material consumption, but also gave some polar decomposition byproduct. Ultimately, GaCl3 at room temperature proved to be the most successful Lewis acid, giving 12a in a 68% yield. Reducing the amount of GaCl3 to 5 mol% decreased the yield noticeably (47%), while an increase to 15 mol% made a negligible difference (67% yield). Omission of the 4 Å molecular sieves also gave a decrease in the yield of 12a (51%, 58% brsm).

Scheme 2. Preparation of phenyl ketone 8c.
In addition, a third substrate chosen for the study was 10, employing an aryl spacer rather than one of the alkene spacers between the ester and bromide. Compound 10 was prepared by the radical bromination of cinnamate ester derivative 11 (10, 77%) (Scheme 3), itself being prepared by the Wittig reaction of o-tolualdehyde [32].
The results of the calculations were promising. The ionization energy of 8a to dienyl cation 8a+ was unsurprisingly the most favored, the process being 16.8 kcal/mol lower in energy than allyl cation generation. Somewhat to our surprise, the ionizations of the εcarbonyl cation precursors 8b and 8c also were found to be favored substantially (by 10.9 kcal and 12.7 kcal, respectively), relative to the process with allyl bromide. Finally, the analogous ionization of ethyl 4-bromocrotonate was found to be 6.9 kcal/mol higher in energy than that of allyl bromide, consistent with the difficulty in discrete generation of γ-carbonyl cations. As a result of these findings, we chose to test these observations with an experiment. Given the notably mild conditions reported in the group 13 catalyzed electrophilic reactions of allyl bromides [25][26][27][28], we chose to pursue the analogous approach for ε-carbonyl cations.

Scheme 2. Preparation of phenyl ketone 8c.
In addition, a third substrate chosen for the study was 10, employing an aryl spacer rather than one of the alkene spacers between the ester and bromide. Compound 10 was prepared by the radical bromination of cinnamate ester derivative 11 (10, 77%) (Scheme 3), itself being prepared by the Wittig reaction of o-tolualdehyde [32]. Experimental work began with ethyl 6-bromohexadienoate (ethyl 6-bromosorbate, 8b). Test reactions were undertaken with mesitylene (5 equiv) as the nucleophile, and catalytic amounts (10 mol%) of Lewis acids CuCl, SnCl4, InCl3, GaCl3, and BiI3, in CH2Cl2 with 4 Å molecular sieves ( Table 2, Scheme 4). CuCl and BiI3 afforded no product and minimal amounts of product, respectively. Conversely, GaCl3, InCl3, and SnCl4 gave more significant amounts of conversion to 12a over 24 h, although small amounts of starting material remained. Repetition of the reactions at reflux afforded complete starting material consumption, but also gave some polar decomposition byproduct. Ultimately, GaCl3 at room temperature proved to be the most successful Lewis acid, giving 12a in a 68% yield. Reducing the amount of GaCl3 to 5 mol% decreased the yield noticeably (47%), while an increase to 15 mol% made a negligible difference (67% yield). Omission of the 4 Å molecular sieves also gave a decrease in the yield of 12a (51%, 58% brsm). Experimental work began with ethyl 6-bromohexadienoate (ethyl 6-bromosorbate, 8b). Test reactions were undertaken with mesitylene (5 equiv) as the nucleophile, and catalytic amounts (10 mol%) of Lewis acids CuCl, SnCl 4 , InCl 3 , GaCl 3 , and BiI 3 , in CH 2 Cl 2 with 4 Å molecular sieves (Table 2, Scheme 4). CuCl and BiI 3 afforded no product and minimal amounts of product, respectively. Conversely, GaCl 3 , InCl 3 , and SnCl 4 gave more significant amounts of conversion to 12a over 24 h, although small amounts of starting material remained. Repetition of the reactions at reflux afforded complete starting material consumption, but also gave some polar decomposition byproduct. Ultimately, GaCl 3 at room temperature proved to be the most successful Lewis acid, giving 12a in a 68% yield. Reducing the amount of GaCl 3 to 5 mol% decreased the yield noticeably (47%), while an increase to 15 mol% made a negligible difference (67% yield). Omission of the 4 Å molecular sieves also gave a decrease in the yield of 12a (51%, 58% brsm).

Discussion
An analysis of the results suggests several issues worth discussing. First of all, despite Scheme 6. Reactions of ε-bromo aryl alkenoate 10. The reaction with allyltrimethylsilane was again more difficult than for arene nucleophiles with GaCl 3 catalysis. In this case, while 10 mol% GaCl 3 showed no significant conversion, 50 mol% GaCl 3 gave a 46% yield of 13g. InCl 3 again proved to be a superior catalyst with allyltrimethylsilane; 10 mol% of InCl 3 afforded a 29% yield of 13g, while raising the catalyst amount to 20 mol% InCl 3 gave 13g in 64% (78% brsm). Finally, a switch to higher temperature reaction conditions (1,2-dichloroethane, reflux) demonstrated that propiophenone trimethylsilyl enol ether was also amenable to reaction with 10 (13h, 82% yield) with the use of InCl 3 as the catalyst.

Discussion
An analysis of the results suggests several issues worth discussing. First of all, despite the unmanageable superficial appearance of ε-carbonyl cations, they are quite viable. Transition metal stabilization of the cationic dienyl (or enynyl) unit is not mandatory. The use of dienyl bromides and Ga(III) or In(III) catalysts is capable of generating ε-carbonyl cations that react with nucleophiles in moderate yields with 8b-c, and in good yields with 10. The reactions require somewhat more vigorous conditions than with allyl bromide itself, and we attribute this to the presence of the Lewis basic carbonyl functions in the substrates, and in some cases, the reacting nucleophiles. Arene nucleophiles react with greater facility than allylsilanes using GaCl 3 , although conditions can normally be found using InCl 3 that give synthetically useful yields of 12f and 13g. InCl 3 also allows the successful reaction of an enol silane (13h). The successful incorporation of benzene as a nucleophile (13f) indicates that the current protocol can allow incorporation of less reactive nucleophiles than the Nicholas reaction-based ε-carbonyl cation equivalents [10] and far less reactive nucleophiles than the analogous transition metal catalyzed equivalents [14][15][16][17][18][19][20][21]. The question of competitive conjugate addition does not appear problematic with the arene, allylsilane, or enol silane nucleophiles. For example, the crude reaction product of 8b and allyltrimethylsilane showed no evidence of conjugate addition byproducts. Conversely, trial reactions with triethylsilane, a substantially stronger nucleophile than arenes or allyltrimethylsilane [33], appeared to give mixtures whose 1 H NMR spectra included multiple aliphatic resonances, suggesting the conjugate addition may be a major reaction pathway there.

Materials and Methods
The starting materials and reagents involved in the reactions were purchased from commercial sources, unless otherwise noted. GaCl 3 and InCl 3 were stored under an inert atmosphere prior to use. Purification of synthesized products was conducted by either column chromatography (using SilaFlash ® P60, 230-400 mesh, SiliCycle, Quebec City, QC, Canada), preparative TLC (SiliaPlate, 1000 µm thickness, SiliCycle, Quebec City, QC, Canada) or radial chromatography (Silica gel, 2000 µm thickness, EM Science, Gibbstown, NJ, USA). Analytical thin layer chromatography (TLC) was performed using Silicycle aluminum-backed sheets (SiliCycle, Quebec City, QC, Canada). Dichloromethane and tetrahydrofuran solvents (Sigma-Aldrich Canada, Milton, ON, Canada) were obtained from a solvent purification system. All of the reactions were performed under an atmosphere of nitrogen unless otherwise stated. Prior to reaction, all glassware was dried in an oven at 110 • C for a minimum of one hour and subsequently cooled in a desiccator. Reactions conducted at greater than 25 • C were conducted in a heated oil bath.
All of the NMR spectral analyses were conducted on 300 MHz and 500 MHz spectrometers (Bruker Canada, Milton, ON, Canada) at room temperature in solutions of CDCl 3 (CIL, Andover, MA, USA). The residual CHCl 3 peak was set to 7.27 ppm and 77.0 ppm for the 1 H NMR and 13 C NMR spectra, respectively. 1 H NMR spectral data are listed with units of ppm for peak position (δ) and Hz for coupling constant (J). The following symbols were used for peak appearance: s, singlet; d, doublet; t, triplet; dd, doublet of doublets; dt, doublet of triplets; q, quartet; m, multiplet. The 1 H and 13 C NMR spectra are available in the Supplementary Materials. The IR analysis was conducted on an ATR infrared (FTIR) spectrometer (Bruker Canada, Milton, ON, Canada). For IR spectra listed in the characterization of compounds and the absorption peaks with the greatest functional group relevance are reported in wavenumbers (cm −1 ). High resolution mass spectrometry results were obtained by direct insertion probe on a Waters Xevo G2-XS Time-of-Flight Mass Spectrometer (Waters, Toronto, ON, Canada) in ASAP(+) mode at the University of Windsor Mass Spectrometry lab. The computational calculations were conducted with Gaussview 5.0.9 and B3LYP/6-311++G(d,p) to optimize the structures studied, both with and without solvation in dichloromethane. Final coordinates are available in the Supplementary Materials.

Methyl 3-[2-(Bromomethyl)phenyl]acrylate (10)
Bromination was conducted with methods derived from those described by Snead [34]. To a suspension of GaCl 3 (0.009 g, 0.05 mmol, 10 mol%) and 4Å molecular sieves (ca. 0.4 g), CH 2 Cl 2 (6 mL) was added to mesitylene (0.37 mL, 2.67 mmol, 5 equiv.) and 8b (0.1161 g, 0.5299 mmol) at room temperature. The reaction was stirred under N 2 and monitored by TLC for 26 h. Following removal of volatiles under reduced pressure and flash chromatography (10:1 PE:Et 2 O), 12a (0.0902 g, 68%) was isolated as a yellow oil. This compound was also made by methods outlined below in General Procedure 1, where the reaction was brought to reflux for 22 h after the reagents were added. This afforded the product 12a in a 63% yield. IR