Selective Cross‐Ketonization of Carboxylic Acids Enabled by Metallaphotoredox Catalysis

Abstract

Carboxylic acids are attractive building blocks for synthetic chemistry because they are chemically stable, abundant, and commercially available with substantial structural diversity. The process of combining two carboxylic acids to furnish a ketone is termed ketonization. This is a potentially valuable transformation that has been underutilized in organic synthesis due to the harsh reaction conditions generally required and the lack of selectivity obtained when coupling two distinct carboxylic acids. We report herein a metallaphotoredox strategy that selectively generates unsymmetrical ketones via cross‐ketonization of two structurally dissimilar carboxylic acids. Cross‐selectivity is achieved by exploiting divergent reactivity of differentially substituted acids towards critical one‐ and two‐electron processes in the proposed coupling mechanism. This method is broadly applicable to a variety of functionalized carboxylic acids. It can also be applied to acids of similar steric profile by exploiting differences in their relative rates of decarboxylation.

Unsymmetrical ketones can be prepared through selective coupling of two structurally distinct carboxylic acids without the need for preactivation of either partner. This photoredox cross‐ketonization strategy exploits differences in the rates of key one‐ and two‐electron activation steps using the two reactants.

The conversion of carboxylic acid derivatives into ketones is an important chemical transformation, due both to the prevalence of ketones in bioactive drugs and natural products and to their central role as versatile reactants in synthetic chemistry. ^[1] Classical methods to effect this transformation rely upon the reaction of carboxylic acid derivatives (e.g., Weinreb amides, anhydrides, or acid chlorides) with organometallic nucleophiles such as organolithium, organomagnesium, or organocopper reagents (Figure 1A). ^[2] Modern transition metal‐catalyzed strategies are more attractive, employing milder reactants such as organoboron, organozinc, or organosilicon nucleophiles. ^[3] However, these organometallic reagents ultimately derive from alkyl halides, and Ni‐catalyzed cross‐electrophile coupling strategies have increasingly become valued as attractive alternatives that preclude the need for a prefunctionalization step. ^[4] Although alkyl halides represent a valuable synthetic starting point, alkyl carboxylic acids constitute an even more abundant set of commercially available building blocks (Figure 1B). ^[5] We wondered if diverse ketone products could also be accessed from the coupling of two different carboxylic acids, where one acid serves as an acyl electrophile, and the other serves as a nucleophile via decarboxylation. The coupling of two carboxylic acids to afford a ketone is termed ketonization ^[6] and is a powerful transformation that has been largely overlooked in synthetic organic chemistry.

Figure 1.

A) Summary of methods for ketone synthesis. B) Relative commercial abundance of alkyl‐functionalized feedstock chemicals.

Ketonization was first discovered over 150 years ago by Friedel, who demonstrated that the dry distillation of calcium acetate affords acetone. ^[7] Since then, ketonization has primarily been studied as a strategy for biomass conversion, enabling simple alkyl carboxylic acids to be upgraded into high molecular weight ketones. ^[8] Current research has mainly focused on the identification of new metal and zeolite catalysts that enhance reactivity and cross‐selectivity. Efficient heterocoupling has generally required the use of a large excess of one trivial coupling partner, such as acetic acid, ^[9] or the use of an aryl carboxylic acid that resists decarboxylation. ^[10] However, complex, unsymmetrical dialkyl ketones are inaccessible using these strategies. The difficulty of achieving cross‐selectivity and the generally harsh reaction conditions required (>300 °C) have greatly limited the application of direct ketonization methods in organic synthesis.

Several strategies for ketone synthesis from carboxylic acid derivatives have emerged in recent years. ^[11] Generally, these reactions have proceeded through selective generation of an acyl oxidative addition complex from one partner in conjunction with radical decarboxylation of the other; radical combination and subsequent reductive elimination can then selectively furnish an unsymmetric ketone. However, to distinguish between the two different precursors, these protocols have relied on strategies that prefunctionalize the carboxylic acid feedstocks. In 2019, Weix and co‐workers reported the nickel‐catalyzed synthesis of ketones using a thioester, poised for oxidative addition, together with an N‐hydroxyphthalimide ester that could serve as a precursor for a carbon‐centered radical. ^[12] Conceptually similar approaches using alternate activated acid derivatives have subsequently been reported by Opatz, Qi, and Yuan. ^[13] There have also been limited examples of formal cross‐ketonization requiring prefunctionalization of only one carboxylic acid. MacMillan and Chi have reported that acid chlorides and acyl imidazoles could be used in conjunction with acids to form unsymmetrical ketones using a metallaphotoredox strategy. ^[14] Additionally, Yang and Xia reported that N‐hydroxyphthalimide esters could engage with in situ activated carboxylic acids in an electrochemical setting. ^[15] Cross‐selective ketonization using two native carboxylic acid inputs, however, has not yet been realized.

We hypothesized that this goal could be achieved using a metallaphotoredox strategy ^[16] that capitalizes on contrasting reactivity between differentially substituted carboxylic acids (Scheme 1). We surmised that the less sterically demanding of the pair should undergo preferential in situ activation to form mixed anhydride 2, which would activate it towards subsequent oxidative addition to a catalytic nickel center. ^[17] A more substituted arylpropionic acid, on the other hand, would favor photoinduced radical decarboxylation to form stabilized benzylic radical 1. ^[18] Radical addition to the oxidative addition complex would generate nickel intermediate 3, which is poised to undergo reductive elimination to produce unsymmetric ketone 4. We also considered an alternative mechanism in which the radical combination with nickel would precede the acyl oxidative addition event or formation of a benzylic anion via radical decarboxylation and subsequent reduction; however, the selectivity control elements would remain identical in either case. ^[19]

Scheme 1.

Reaction design for the current work and optimization of the cross‐selective ketonization. ^aYields were determined by ¹H NMR analysis of the crude reaction mixture using dibromomethane as an internal standard. ^bNo dark prestir phase. ^cAdded 2.0 equiv H₂O. ^dOmitted 4CzIPN. ^eOmitted nickel catalyst. ^fOmitted Boc₂O. ^gReaction run wrapped in aluminum foil.

Initial studies probing the feasibility of this design focused on the coupling of ketoprofen (5) with primary carboxylic acid 6. The desired cross product is formed when using Boc₂O as an in situ activating reagent (Scheme 1B, Entry 1). When using DMDC (dimethyl dicarbonate) or DEDC (diethyl dicarbonate) (Entries 2, 3), we observed reduced yields, and the methyl and ethyl esters were formed as significant byproducts. ^[20] Lower yields were also obtained with PivCl or Piv₂O (Entries 4, 5). Furthermore, although we typically prestirred the reaction mixture for 30 min in the dark to maximize the formation of mixed anhydride, the yield and selectivity were still acceptable without the prestir phase (Entry 6). The primary carboxylic acid was employed in excess to favor formation of the mixed anhydride intermediate, however, good yields and selectivity were observed with lower equivalents (see Supporting Information for details). The reaction proved to be water sensitive, as the addition of 2 equiv of water resulted in a decrease in yield and a significant increase in the protodecarboxylation product 10 (Entry 7). ^[21] Control experiments indicate that productive coupling requires the photocatalyst, nickel catalyst, Boc₂O and light (Entries 8–11), consistent with the proposed metallaphotoredox strategy.

The cross‐selective ketonization protocol is applicable to a wide variety of substituted carboxylic acids capable of undergoing decarboxylation (Figure 2). Substituted phenylacetic acid derivatives yielded products (11–19) in moderate to good yields; although, electron rich arenes (18) showed decreased yields. Heterocyclic products (20, 21) and carbocyclic ketones (22) also showed diminished yields. A variety of drug‐based α‐methyl phenylacetic acids could be incorporated in good yields, including those bearing heterocycles (23–27, 7). Cyclohexanecarboxylic acid, which generates a comparably less stable radical intermediate, afforded product 28 in 22 % yield, and an adamantyl group (29) could be incorporated in similar yields. In addition to substituted α‐aryl acids, α‐heteroatom‐containing carboxylic acids also underwent ketonization. Acids bearing α‐phenoxy motifs (30, 31) reacted in good yields, and an α‐methoxy group was tolerated (32). Acids containing saturated or benzofused heterocycles afforded products (33–35) in moderate yields, and a highly substituted α‐phenoxy acid was successful in generating product 36. Similarly, using bezafibrate, product 37 was isolated in 24 % yield. Beyond α‐oxy acids, α‐amino acids were also tolerated in the cross‐coupling; alanine, methionine, glutamate, and proline deliver the corresponding ketones in moderate yields (38–41). The highly substituted dimethylamino acid analogue (42) was also furnished in 20 % yield. Notably, tertiary radical intermediates generated the desired ketonization products in reduced yields, likely attributable to increased steric hindrance. ^[13b] Products derived from phenylacetic acid, diphenylacetic acid, and 2‐phenylisobutyric acid, however, could not be obtained in useful yields.

Figure 2.

Reactions performed on 0.30 mmol scale following Table 1, Entry 5. ^aUsing 2.7 equiv linear acid, 2.3 equiv Boc₂O, 4.0 equiv K₂CO₃. ^bUsing bpy instead of dtbbpy. ^c1.2 mmol scale, 5 mol % Ni, 7.5 mol % dtbbpy, 2 mol % 4CzIPN, 1.7 equiv linear acid, 3 equiv K₂CO₃, 1.3 equiv Boc₂O, 0.1 M DME.

A variety of acyl donors were investigated in the cross‐coupling as well. Simple alkyl derivatives (43–50) were well tolerated; however, hydrocinnamic acid showed a depressed yield (47). Protected indoles (51) and alkyl chlorides (52) could be included in the cross‐coupling as well. Succinic acid derivative 53 was obtained in reduced yields. However, when the ethyl ester was changed to a sterically bulky tert‐butyl ester (54) or when alkyl tether length was increased (55), the yield increased. This might be rationalized by coordination of the ester stabilizing the oxidative addition complex and slowing turnover of the Ni catalyst. Other acids bearing potentially coordinating groups furnished products in slightly reduced yields (56, 57). A similar trend was observed between product 57 and 58, wherein the longer tether containing a protected amine showed increased yield. Azetidine‐based carboxylic acids were successfully coupled (59), and amino acid sidechain functionalization was also showcased with product 60 in 43 % yield. Product 61, bearing a bulky adamantanol substituent, was tolerated without protection of the tertiary alcohol. Products derived from lineoic acid (62), cis‐pinonic acid (63), and lithocholic acid (64) were all obtained in moderate yields. Finally, alkenyl acids were also tolerated, as demonstrated by crotonic acid (65), and β,β‐dimethylacrylic acid (66). Enone product 67, containing a piperidine unit, was also generated, albeit in lower yields. Notably, cinnamic acid and benzoic acid derivatives yielded no product in this transformation.

The strategy for achieving selectivity in this method for cross‐ketonization is premised on the hypothesis that one acid partner is predisposed towards acylation, leading to the corresponding acyl‐Ni oxidative insertion complex, and that the other will undergo more efficient radical decarboxylation, affording the corresponding organoradical intermediate. Several control experiments are consistent with this reaction design. First, when primary carboxylic acid 6 is treated with Boc₂O and K₂CO₃, good conversion to the expected mixed anhydride is observed within 2 h (Scheme 2A). In the analogous experiment using secondary carboxylic acid 5, there is no observable formation of the mixed anhydride at the same time point, with good recovery of the carboxylic acid. Second, when the reaction is conducted in the absence of Boc₂O, precluding the formation of any mixed anhydrides, we observe good yields of 10 arising from protodecarboxylation of the secondary carboxylic acid (Scheme 2B). Only traces of products assignable to protodecarboxylation of the primary acid (70) could be identified by ¹H NMR analysis of the unpurified reaction mixture.

Scheme 2.

Effect of the degree of substitution on in situ activation and radical decarboxylation. ^aYields were determined by ¹H NMR analysis of the crude reaction mixture using dibromomethane as an internal standard.

These results are consistent with the general observation that yields of cross‐ketonization are highest using carboxylic acids that would provide stabilized radicals upon decarboxylation. Lower yields are observed when decarboxylation would afford relatively unstabilized secondary or tertiary radicals (e.g., 28 and 29). As a direct assessment of the importance of intermediate radical stability, we conducted a competition experiment subjecting a 1 : 1 mixture of 2‐phenoxypropionic acid and cyclobutanecarboxylic acid to cross‐ketonization conditions with a model primary aliphatic carboxylic acid (Scheme 3A). As expected, this experiment favors formation of product 28 in 20 % yield, with only a trace of 71. Interestingly, we also observed a substantial yield of 72, arising from coupling of the two secondary carboxylic acids. This would suggest that aliphatic secondary carboxylic acids might be suitable oxidative addition partners that would enable selective cross‐ketonization between two distinct secondary carboxylic acids. To test this hypothesis, we examined the coupling of 2‐phenoxypropionic acid with cyclohexanecarboxylic acid and obtained product 73 in 50 % isolated yield with high selectivity for the crossed product (Scheme 3B). This strategy was shown to be applicable to a variety of secondary‐secondary cross‐ketonizations in synthetically useful yields (72–76), demonstrating that this novel metallaphotoredox method offers a route to a wide range of structurally diverse, unsymmetrical ketone products.

Scheme 3.

Cross‐coupling of two secondary carboxylic acids. ^aYields determined by ¹H NMR analysis of the crude reaction mixture using phenanthrene as an internal standard. ^bRatio was determined by ¹H NMR analysis of the crude reaction mixture.

In conclusion, we have developed a metallaphotoredox strategy for cross‐selective ketonization that does not require prefunctionalization of either carboxylic acid precursor. Subtle differences in substitution underpin selectivity for the cross product, wherein less sterically bulky acids undergo in situ acylation to enable a two‐electron oxidative addition with a nickel catalyst. Conversely, a more sterically bulky acid can selectively undergo a radical decarboxylation to generate a more stable radical intermediate. The metal catalyst mediates the C−C bond forming event to generate an unsymmetric ketone. A variety of carboxylic acids were utilized, including a number of acid‐containing drug molecules. The protocol was further extended to the cross‐coupling of two secondary acids, driven by understanding the relative rates of decarboxylation.

Conflict of interest

The authors declare no conflict of interest.

Supporting information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

A. Whyte, T. P. Yoon, Angew. Chem. Int. Ed. 2022, 61, e202213739; Angew. Chem. 2022, 134, e202213739.

Data Availability Statement

The data that support the findings of this study are available in the Supporting Information of this article.