Deconstructive Radical–Radical Coupling for Programmable Remote Acylation

Abstract

Dicarbonyl compounds are structural motifs that have been extensively utilized in synthetic chemistry, and their downstream transformations have proven valuable in synthesizing numerous heterocycles. Conventional methods for accessing such compounds include the Stetter reaction, the Michael reaction, and Friedel–Crafts acylation. However, a flexible and enabling platform for obtaining all types of 1,n‐dicarbonyls remains undeveloped. Reported herein is a unified approach to access a variety of 1,n‐dicarbonyls through a radical‐promoted deconstructive process utilizing dual photocatalysis/carbene catalysis. The utility of this strategy is demonstrated with a broad scope with robust functional group tolerance along with application in the preparation of γ‐amino esters, three‐component manifolds, and the first enantioselective deconstructive synthesis of 1,5‐dicarbonyls using a chiral NHC catalyst.

A unified approach to access a variety of 1,n‐dicarbonyls through a radical‐promoted deconstructive process utilizing dual photocatalysis/carbene catalysis is reported. The utility of this strategy is demonstrated with a broad scope with robust functional group tolerance along with application in the preparation of γ‐amino esters, three‐component manifolds, and the first enantioselective deconstructive synthesis of 1,5‐dicarbonyls using a chiral NHC catalyst.

Selective remote functionalization to forge new C─C bonds remains a fundamental challenge in organic synthesis.^{[ 1 ]} In recent years, transition‐metal catalyzed C–H functionalization has emerged as a viable strategy for distal C–H activation to install a diverse C─C bonds including alkylation,^{[ 2 ]} arylation,^{[ 3 ]} and olefination.^{[ 4 ]} However, these established methods typically require a pre‐installed Lewis‐basic directing group to selectively activate one specific C─H bond, thereby limiting functionalization to only a singular remote site (Figure 1a).^{[ 5 ]} Alternatively, radical‐based methods often utilize hydrogen atom transfer (HAT) to activate distal C─H bonds,^{[ 6 ]} either by employing an external HAT reagent^{[ 7} , ^{8 ]} or through internal 1,5‐HAT processes.^{[ 9 ]} While these methods have proven useful, they are inherently restricted by site selectivity issues, i.e., targeting one C─H bond over numerous others.^{[ 10 ]} As a complementary approach, C─C bond activation can potentially be highly selective and maintain modularity to achieve a diverse general platform for distal C─C functionalization.^{[ 11 ]}

Figure 1.

a) Standard fixed directing group functionalization. b) Potential programmable approach from cyclic ketones. c) Reaction design with alcohols and acid derivatives d) this work.

Radical‐promoted deconstructive processes have provided the opportunity to unmask dormant functional groups, such as hydroxyls, which can be rapidly transformed into two tethered functional groups.^{[ 12 ]} In these processes, the distance between these two groups is determined by the size of the ring (Figure 1b). However, the direct utilization of hydroxyls by breaking the O─H bond is challenging due to the high bond‐dissociation free energy (BDFE), of ∼105 kcal mol⁻¹.^{[ 13 ]} Recent efforts have been focused on exploring mild conditions to activate cyclic alcohols.^{[ 14} , ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ , ²¹ , ²² , ²³ , ^{24 ]} Seminal work by the Knowles group has pioneered the use of proton‐coupled electron transfer (PCET) approach leveraging a combination of a highly oxidizing photocatalyst and a critical phosphate base, enabling access to a series of linear ketones.^{[ 25} , ²⁶ , ²⁷ , ^{28 ]} Additionally, the Zuo group has successfully generated useful alkyl radical intermediates through a ligand‐to‐metal charge transfer (LMCT) process, which can further react with di‐tert‐butyl azodicarboxylate.^{[ 29} , ³⁰ , ^{31 ]} Most recently, the Lin group identified a related reaction system that utilizes frustrated radical pairs (FRPs) between oxoammonium ions and tertiary alcohols. This system undergoes single electron transfer (SET) followed by radical coupling, leading to the formation of a series of aminoxyl products.^{[ 32 ]} While these methods mentioned above have mainly focused on the formation of C─X bonds (where X can be H, F, Cl, Br, N, or O), the study of C─C bond construction has been limited to nickel‐mediated cross‐coupling with aryl halides^{[ 33} , ^{34 ]} or Giese addition to Michael acceptors.^{[ 35} , ^{36 ]} A general process for cross‐coupling between two carbon‐centered radicals has not yet been realized to the best of our knowledge (Figure 1c).

Alcohols and carboxylic acids are abundant materials found in both natural and commercial sources.^{[ 37} , ^{38 ]} Traditional coupling reactions typically lead to the formation of esters via C─O bond formation.^{[ 39 ]} Recently, the MacMillan group reported a novel strategy utilizing metallaphotoredox radical coupling to construct C(sp ³)─C(sp ³) bonds.^{[ 40 ]} We anticipate that utilizing alcohols and carboxylic acid derivatives in radical coupling processes should advance the development of novel disconnection strategies. Radical N‐heterocyclic carbene (NHC) organocatalysis has seen rapid growth over the past five years,^{[ 41} , ⁴² , ⁴³ , ⁴⁴ , ⁴⁵ , ⁴⁶ , ⁴⁷ , ⁴⁸ , ⁴⁹ , ⁵⁰ , ⁵¹ , ⁵² , ^{53 ]} with one of the key discoveries being the generation of persistent ketyl radicals, which can participate in radical cross‐coupling reactions under transition metal‐free conditions.^{[ 54} , ⁵⁵ , ⁵⁶ , ⁵⁷ , ⁵⁸ , ⁵⁹ , ⁶⁰ , ⁶¹ , ⁶² , ⁶³ , ⁶⁴ , ⁶⁵ , ⁶⁶ , ⁶⁷ , ⁶⁸ , ⁶⁹ , ⁷⁰ , ⁷¹ , ^{72 ]} As part of our ongoing interest in NHC radical organocatalysis, we envisaged the use of cyclic alcohols and acyl azoliums under photoredox conditions could separately generate alkyl radicals and ketyl radicals concomitantly. These nascent radicals could then subsequently undergo cross‐coupling based on the persistent radical effect^{[ 73 ]} and eject the carbene to form C(sp ³)─C(sp ²) bonds. (Figure 1c) Herein, we report the radical cross‐coupling reaction between tertiary alcohols and carboxylic acid derivatives to produce various types of 1,n‐dicarbonyls (Figure 1d).^{[ 74} , ⁷⁵ , ⁷⁶ , ⁷⁷ , ⁷⁸ , ^{79 ]} The overall process is programmable in the sense that the size of the ring undergoing fragmentation directs the site of eventual acylation enabling a series of 1,n‐dicarbonyl products.

Initial experiments commenced with an examination of the coupling reaction between diethyl pyrocarbonate (1a, 2 equiv.) with cyclobutanol (2a). Following a series of condition screenings, we identified optimal conditions using an azolium (Az‐1, 20 mol%), 5CzBN (PC‐1, 2.5 mol%), and potassium carbonate (40 mol%). With these conditions, the desired 1,5‐keto ester (3a) was isolated in 80% yield (87% NMR yield) after 4 h of 427 nm blue LEDs irradiation in 0.1 M acetonitrile (Table 1, entry 1). A screen of various bases revealed that the reaction is highly sensitive to the basic environment. For example, a strong base such as potassium tert‐butoxide significantly lowered the yield to 15% (entry 2), and the formation of linear ketone (see Supporting Information for details) as a side product was observed. Sodium ethoxide and sodium acetate both provided slight decreases in yield (60% and 81%, entries 3 and 4), while cesium carbonate afforded a similar yield to potassium carbonate at 87% (entry 5). Two different types of azoliums (Az‐2 and 3) were screened, both failed to form any of the desired product (entries 6 and 7, see Supporting Information for other azoliums screening). This result indicates that the reaction is sensitive to both the substitution pattern and the counterion of the azolium catalyst.^{[ 55 ]} Tuning the oxidative potential of the photocatalyst by employing more oxidizing photocatalysts, such as 4CzIPN (PC‐2) or an iridium photocatalyst (PC‐3), resulted in lower yields (30% and 19%, entries 8 and 9). This reaction also demonstrated sensitivity to the solvent environment, as evidenced by the results with THF (71%, entry 10) and dichloromethane (0%, entry 11), while toluene yielded a similar result with acetonitrile (82%, entry 12). Additionally, decreasing the loading of the base from 40 to 20 mol% led to a lower yield (69%, entry 13). A higher concentration also negatively impacted the formation of product (3a), providing a yield of 58% (entry 14), no product formation in absence of azolium, photocatalyst, or light (entries 15–17).

Table 1.

Optimization of reaction conditions.

^a) Reaction conditions unless otherwise indicated: 1a (0.2 mmol), 2a (0.1 mmol), Az (0.02 mmol), base (0.4 mmol), PC (2.5 µmol), and solvent (0.10 M; THF = tetrahydrofuran) irradiated for 4 h.

^b) 1H NMR yields using 1,3,5‐trimethoxybenzene as an internal standard. An isolated yield is given in parentheses.

We next explored the reaction scope by surveying various cycloalkanols (Table 2). Different ring size of the cycloalkanols all successfully provided the desired 1,n‐keto esters in excellent yields. Notably, the use of cyclopropanol (2d), which can serve as a primary radical source, was tolerated under these conditions. The use of a‐tertiary cycloalkanols generated the desired 1,n‐keto esters bearing a quaternary left, which can be challenging to access through traditional two‐electron processes. Various functional groups were successfully incorporated, yielding the desired products in good yields (68%‐76%, 3e‐3i). Different substituents on the aryl ring of the cycloalkanols were examined, and both electron‐rich and electron‐poor substituents were well tolerated. Additionally, a more hindered mesityl substitution produced product 3m in 74% yield. Replacing the para‐methoxy group with 2,4‐dimethoxy (3o), 2‐naphthyl (3p), or phenanthrene (3q) proved effective under the standard conditions. Finally, modifying a camphor derivative (2t) furnished the desired cyclopentane (3t) bearing two quaternary lefts, achieving a modest yield of 40%. Subsequently, we examined different carbonate precursors. While increasing the steric hindrance of the ester negatively impacted the yield, e.g., the methyl ester (3u) was formed with an 80% yield with the isopropyl ester (3v) yielded 77%. In contrast, the tert‐butyl ester (3w) provided a decreased yield of 63%. Additionally, other types of esters, including modifications of L‐menthol derivative(1ab), were well tolerated under these conditions (3x‐3ab, 70%–80% yields).

Table 2.

Substrate scope for ester synthesis.

^a) Reaction conditions, unless otherwise indicated: 1 pyrocarbonate (0.2 mmol), 2 (0.1 mmol), Az‐1 (0.02 mmol), K₂CO₃ (0.4 mmol), PC‐1 (2.5 µmol), and CH₃CN (0.10 M) irradiated for 4–16 h, see Supporting Information for more detailed conditions.

^b) 1 Pyrocarbonate (0.5 mmol) and the reaction time was 48 h.

^c) 1 Chloroformate (0.2 mmol) and Cs₂CO₃ (0.12 mmol) were used as the coupling partner and base.

Having demonstrated the success of our strategy for the formation of 1,n‐keto esters, we next examined the radical cross‐ coupling reaction between acyl imidazole (4) and alcohol (5). By slightly modifying the standard reaction conditions, we obtained the desired 1,4‐diketone (6a) with a yield of 67% (Table 3). These conditions were further tested with a secondary radical source (6b, 68%) and various ring sizes (6c‐6e, 77%–81%). The use of heterocycles such as oxetane and azetidine also resulted in the corresponding 1,5‐diketones (6f) and (6g) in good yields. Finally, different acyl imidazoles were evaluated, yielding the desired 1,5‐diketones in satisfactory results (6h‐6j, 61–78%). The radical‐relay reaction of alkenes offers a novel pathway for rapidly increasing the complexity of the product. The success of this type of reaction relies on the ability of a transient radical to be captured by an alkene, such as styrene, to form a more stabilized benzylic radical, which can then undergo radical coupling with the persistent radical. Based on this strategy, 2.0 equiv. of styrene was added to our standard conditions from Table 1, and the desired coupling product (8) was obtained in 45% yield without further optimization (Scheme 1a).^{[ 80 ]} We envisaged the use of cyclic amines in this strategy to promote a similar radical ring opening, producing C‐lefted radicals and tethered iminium ions. Pioneering studies by Zheng have explored this process in a [3 + 2] cycloaddition with olefins.^{[ 81} , ⁸² , ⁸³ , ^{84 ]} More recently, Huang demonstrated the feasibility of photoredox and cobalt co‐catalyzed ring‐opening dehydrogenative functionalization of cyclicamines.^{[ 85 ]} As of now, radical coupling reactions via the deconstructive process of amines have not been realized. Motivated by this, we investigated our reaction system using cyclopropyl amine (9) and successfully observed the generation of product (II) through ¹H‐NMR and ESI‐HRMS analysis (see Supporting Information for detailed results). Following this, sodium cyanoborohydride and p‐toluenesulfonic acid were added in a one‐pot reaction, resulting in the desired γ‐amino esters with good yields (Scheme 1b, 10a and 10b, 64% and 52%). To demonstrate the utility of the 1,n‐keto esters 3, we successfully transformed them into the corresponding amino ketones (11a and 11b, 65% and 60%) through a sequential hydrolysis/acyl azide formation/Curtius rearrangement process (Scheme 1c). Additionally, the first enantioselective deconstructive radical coupling has been achieved under transition metal free conditions, providing the desired product 3ac in 65% yield and 91:9 er (Scheme 1d).

Table 3.

Substrate scope for ketone synthesis.

^a) Reaction conditions, 4 (0.2 mmol), 5 (0.1 mmol), Az‐1 (0.02 mmol), K₂CO₃ (1.2 mmol), PC‐1 (2.5 µmol), and CH₃CN (0.10 M) irradiated for 4–16 h, see supporting Information for more detailed conditions.

Scheme 1.

a) Radical relay study. b) γ‐amino ester synthesis. c) Remote amine installation. d) Enantioselective 1,5‐keto ester synthesis.

To investigate the mechanism of this reaction, we conducted two experiments. A TEMPO trapping experiment was performed under standard reaction conditions and showed no formation of the desired 1,5‐keto ester (3a) with the addition of 3.0equivalents of TEMPO. Additionally, TEMPO successfully formed adducts with the cyclohexyl radical and the azolium radical intermediate, both of which were observed via ESI‐HRMS (TEMPO adducts 12 and 13, Scheme 2a). Next, we conducted Stern‐Volmer quenching experiments to support the photoredox cycle (Scheme 2b). Pyrocarbonate (1a) showed no quenching at concentrations of up to 0.1 M, while cyclohexanol (2c) and equimolar mixtures of (2c) with potassium carbonate provided lower rate of photocatalyst quenching. In contrast, the acyl azolium intermediate (IV) displayed a quenching rate significantly higher than that of any other reaction component, implying an oxidative quenching mechanism. However, we cannot completely rule out the possibility of an operative reductive quenching pathway due to the quenching demonstrated by 2c with base.

Scheme 2.

a) Radical trapping study. b) Stern–Volmer fluorescence experiment.

Based on the previous observations, we propose the reaction mechanism depicted in Figure 2. The cycle begins with the generation of the acyl azolium IV through the addition of the free carbene to the carbonate or acyl imidazole. This intermediate has a reduction potential (acyl azolium IV, e.g., R = OEt, E _1/2  = −1.16 V vs. SCE^{[ 86 ]} or R = Ph, E _1/2  = −1.29 V vs. SCE^{[ 87 ]}) that can be reduced by the excited photocatalyst (e.g., 5CzBN, E _1/2 = −1.42 V vs. SCE).^{[ 88 ]} Following the single electron transfer (SET) process, it generates the azolium radical intermediate V and oxidized photocatalyst. The oxidation potential of the cyclic alcohol (e.g., 2c, E _1/2 = +1.48 V vs. SCE)^{[ 23 ]} exceeds that of the oxidized photocatalyst (e.g., 5CzBN, E _1/2 = +1.41 V vs. SCE).^{[ 88 ]} The deprotonated alcohol VI has a lower oxidative potential (e.g., 2c, E _1/2 = +1.40 V vs. SCE), which is sufficient for oxidation by the photocatalyst. This provides the O‐lefted radical which readily undergoes β‐scission to provide radical intermediate VII. Subsequently, radical–radical coupling leads to the formation of the tertiary azolium intermediate VIII. Under basic conditions, this intermediate expels the carbene catalyst, ultimately yielding the desired 1,n‐dicarbonyl 3.

Figure 2.

Proposed reaction pathway.

We have successfully established a dual NHC‐ and photocatalyzed apparent remote acylation reaction through the deconstructive ring‐opening of cyclic alcohols. This reaction demonstrates remarkable versatility, accommodating a wide range of alcohols to access different forms of 1,n‐dicarbonyls with high efficiency. Further exploration has expanded its application to two distinct reaction modes: a three‐component reaction facilitating remote carboxylation and the first radical coupling reaction with cyclic amines via a ring‐opening process. These advancements underscore this methodology's potential applicability across a wide range of programmable acylation events controlled by the ring size of the starting material alcohol. This deconstructive radical process provides a new strategy for functionalization that compliments current C–H functionalization strategies. When integrated with the SET process in conjunction with NHC‐mediated intermediates, this approach enables the formation of C─C bonds to install carbonyl groups at remote positions. Ongoing studies in our group are currently focused on exploring novel reactivity patterns using this deconstructive radical coupling strategy.

Conflict of Interests

The authors declare no conflict of interest.

Supporting information

Supporting Information

Cao J., Schull C. R., Scheidt K. A., Angew. Chem. Int. Ed. 2025, 64, e202507542. 10.1002/anie.202507542

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.