Introduction

Being one of the most basic functional groups with remarkable reactivity, ketones play a fundamental role in synthetic organic chemistry. In addition, they are also important and ubiquitous motifs in natural products, pharmaceuticals and functional materials1,2,3,4. Classic approaches to ketones include the oxidation of alcohols and acylation reaction with activated acyl electrophiles (Fig. 1a), such as acyl chlorides, amides, anhydrides and thioesters, via Friedel-Crafts acylation5, substitution with organometallic reagents6,7, and transition metal-catalyzed coupling with organostannanes, boronic esters, and halides8,9,10. In modern organic synthesis, direct acylation with aldehydes is an alternative and attractive approach to ketones, featuring high atom- and step-economy11. Most acylation reactions with aldehydes rely on transition metal (TM) catalysis (Fig. 1b). The TM-catalyzed coupling of aldehydes with organoboronic reagents12,13,14, hypervalent iodine reagents15,16 and aryl/alkyl halides17,18,19 has been well established. In addition, the TM-catalyzed aryl-H acylation with aldehydes were also reported20,21,22,23. Recently, the pioneering acylation of benzylic C(sp3)-H bonds was developed via photoredox Ni catalysis24. The direct acylation of unactivated alkyl C(sp3)-H bonds with aldehydes is one of the most appealing approaches for ketone synthesis but far less developed.

Fig. 1: Ketone synthesis.
figure 1

a Classical ketone synthesis from alcohols and acyl electrophiles. b Ketone synthesis via TM-catalyzed coupling with aldehydes. c Ketone synthesis via NHC-catalyzed acylation of activated alkylation reagents. d This work: ketone synthesis via photoredox NHC/Pd-catalyzed acylation of alkyl C(sp33)-H. X = halide. TM = transition metal. NHC = N-heterocyclic carbene. NHPI = N-hydroxyphthalimide. hv = photoirradiation. RN = reaction.

Full size image

In the past few decades, N-heterocyclic carbenes (NHCs) have emerged as powerful organocatalysts to construct structurally diverse molecules25,26,27,28,29,30,31. NHC-catalyzed reactions of aldehydes with carbonyl compounds (benzoin reaction), imines (aza-benzoin reaction), and activated alkenes (Stetter reaction) have been well established for the synthesis of functionalized ketones via the umpolung of aldehydes32,33. In addition, NHC catalysis via radicals has opened up new avenues for the C–C bonds formation34,35,36,37, including the synthesis of ketones from aldehydes (Fig. 1c). In 2019, Ohmiya and co-workers pioneered the NHC-catalyzed synthesis of ketones via the decarboxylative coupling of N-hydroxyphthalimide (NHPI) esters with aldehydes38. Lately, other single electron acceptors, such as Katritzky salts39, oxime esters40 or ethers41, Togni reagents42,43, and disulfides44, were successfully employed for the synthesis of ketones with a similar strategy. In addition, the synthesis of ketone from acyl electrophiles under cooperative photoredox/NHC catalysis has also been reported45,46,47,48,49,50.

The efficient and selective functionalization of C(sp3)-H bond has been a long-standing challenge in organic chemistry due to the intrinsic chemical inertness and similarity in electronic and steric properties of C(sp3)-H bonds51,52. The NHC-catalyzed acylation of C(sp3)-H bonds has broad potential application in ketone synthesis with reductive in the number of the synthetic steps, but is not well developed. The primary progress has been achieved in the acylation of activated benzylic and α-amino C(sp3)-H bonds by Rovis53, Studer54, Wang55, and Ohmiya groups56. In contrast, the NHC-catalyzed acylation of ubiquitous unactivated alkyl C(sp3)-H bonds remains a considerable challenge57. Very recently, Li and co-workers reported NHC-catalyzed remote C(sp3)-H acylation of benzylic and unactivated alkyl C(sp3)-H bonds (only two special examples) of amides through N-centered radical-mediated 1,5-hydrogen atom transfer (HAT) mechanism58.

In this work, we report the photoredox cooperative NHC/Pd-catalyzed synthesis of ketones via two and three-component reactions with alkyl C(sp3)-H functionalization (Fig. 1d). This process involves palladium-triggered primary carbon-centered radical-mediated 1,n-HAT to activate remote C(sp3)-H bonds59,60,61 and the following coupling of the resultant radicals62. This reaction features mild reaction conditions and broad functional group tolerance. The potential application of this methodology is further illustrated by the late-stage acylation of the natural products.

Results

Condition optimization of two-component reaction

The two-component model reaction of iodomethylsilyl ether of aliphatic alcohol 1a and picolinaldehyde 2a was investigated under photoredox NHC/Pd catalysis (Table 1). We were encouraged that the reaction gave the desired acylation product 3a after 1,6-HAT in 24% yield with similar amount of undesired direct coupling product 4a and other byproducts59 via dehydrogenation, when carried out in the presence of 20 mol% of thioazolium preNHC N1, 10 mol% of Pd(OAc)2, 20 mol% of bidentate diphenyl-Xantphos L1 with 2.0 equivalent Cs2CO3 in trifluorotoluene under blue LED irradiation (entry 1). Screening of preNHCs found that the preNHC N2 with free hydroxyl group performed better to give ketone 3a in 37% yield without 4a (entry 2). When the preNHC N3 with hydroxyl protected was used instead of N2, the yield of 3a decreased and with the formation of 4a (entry 3). Screening of ligands revealed that the reaction using dicyclohexyl-Xantphos L2 instead of diphenyl L1 gave the desired product 3a in some decreased yield but without 4a (entry 4). Thus, the monodentate tricyclohexylphosphine was then employed, resulting in dramatical improvement of the yield of 3a without 4a (entry 5). Other palladium source, such as Pd(PPh3)4, Pd(TFA)2, and PdCl2 is inferior to Pd(OAc)2 (entries 6–8). Decreasing the loading of preNHC, Pd, or ligand resulted in decreased yields (entries 9–11). The yield of 3a was decreased to 29% under white LEDs irradiation (entry 12). Control experiments revealed all of preNHC, Pd and photoirradiation are crucial for the reaction (entries 13–15).

Table 1 Condition optimizationa
Full size table

Substrate scope of two-component reaction

With the optimized reaction conditions in hand, the substrate scope of aldehyde was then explored (Figs. 2 and 3a–ah). Picolinaldehyde with 3-chloro resulted in decreased yield (3b), possible due to difficulty of the formation of Breslow intermediate stemming from the hindrance of the chloro substituent. The reaction of picolinaldehyde with different substituents, such as Me, OMe and Cl, at 4 or 5-position, all reacted smoothly to give ketones (3c–3g) in moderate to good yields. 6-Methyl substituent on picolinaldehyde resulted in low yield (3h), possibly owing to the increasing steric hindrance. Alkynyl substituent was well tolerated, giving ketone 3i in 90% yield. The picolinaldehydes with a range of additional substituted aryl group (Ar’ = 4-MeOC6H4, 4-CNC6H4, 3,5-(CF3)2C6H3), heteroaryl group (Ar’ = 2-thienyl) and fused aryl group (Ar’ = 1-naphthyl, 9-anthranyl, 9-phenanthryl), all worked well, providing the corresponding products 3j-3q in moderate to good yields. Isoquinoline-3-carbaldehyde performed much better than quinoline-2-carbaldehyde (3r vs. 3s), possibly due to the increasing steric hindrance. Pyrazine-2-carbaldehyde afforded the desired product (3t) in 47% yield. The reaction of thiazole-2-carbaldehydes furnished the corresponding products 3u-3v in decreased yields.

Fig. 2: Two-component reaction.
figure 2

Conditions: 1 (0.3 mmol), 2 (1.5 equiv.), Pd(OAc)2 (10 mol%), PCy3 (20 mol%), N2 (20 mol%), Cs2CO3 (2.0 equiv.), 3.0 mL PhCF3, 36 W Blue LEDs, r.t., under N2, 16 h; awith 5-(trifluoromethyl)pyridin-2-ol (20 mol%); bwith 4-methylpyridin-2-ol (20 mol%); cdr = 1:1; dAfter treating with TBAF (2.0 equiv.), 2 h.

Full size image
Fig. 3: Three-component reaction.
figure 3

Conditions: 1 (0.3 mmol), 5 (2.0 equiv.), 2 (1.5 equiv.), Pd(OAc)2 (10 mol%), PCy3 (20 mol%), N2 (20 mol%), Cs2CO3 (2.0 equiv.), 3.0 mL PhCF3, 36 W Blue LEDs, r.t., under N2, 16 h; then TBAF (2.0 equiv.); aWithout TBAF; bdr = 5:1.

Full size image

Unexpectedly, benzaldehyde did not work under the same conditions as picolinaldehyde. Considering the possible coordination of palladium with picolinaldehyde, 2-hydroxypyridine derivatives were added as additive for the reaction of benzaldehyde (see Supplementary Table 3). We were pleased to find that the reaction of benzaldehyde with aliphatic silyl ether 1a afforded multi-functionalized tetrahydrofuran 3w’ in 61% yield after desilylation, when 20 mol% of 2-hydroxy-5-trifluoromethylpyridine was added as the additive. The corresponding reaction with γ-phenyl silyl ether proceeded smoothly to produce 3x in 72% yield with acylation of benzylic C(sp3)-H bond. The scope with respect to benzaldehyde was evaluated. Benzaldehydes bearing electron-donating (X = 4-Me) or electron-withdrawing groups (X = 4-F, 4-Cl, 4-Br, 4-CF3, 4-CO2Me, 4-CN) on the para-position all worked well to give the corresponding ketones 3y-3ae in moderate to good yields. The benzaldehydes with meta- or ortho-substituent (3-Cl, 2-CN) were also tolerated affording the products 3af and 3ag in 48% and 88% yield, respectively. In addition, cinnamyl aldehyde was also compatible to produce the α,β-unsaturated ketone 3ah albeit with decreased yield. However, simple aliphatic aldehydes did not work for the reaction under the current reaction conditions.

Silyl ether derived from varied alcohol was then examined (Figs. 2 and 3ai–az). Primary alcohols, such as 3-methylbutan-1-ol and 2-cyclohexylethan-1-ol, furnished the desired ketones 3ai and 3aj in good yields. It was noteworthy that the reaction of non-symmetric secondary alcohol bearing two tertiary C(sp3)-H gave the ketone 3ak with exclusive γ-regioselectivity via 1,6-HAT. The terminal chloro in secondary alcohol was tolerated, affording the ketone 3al in 67% yield. Interestingly, hexan-2-ol and cyclooctanol derived ether without tertiary C(sp3)-H could also participate in the reaction with exclusive γ-regioselectivity at secondary C(sp3)-H albeit in low yields (3am & 3an). Beyond the favored 1,6-HAT, the reaction of alkyl ether with β- and δ-tertiary C(sp3)-H gave the corresponding ketones 3ao and 3ap in acceptable yields via 1,5- and 1,7-HAT, respectively. 4-Arylbutan-2-ols bearing different substituents (X = 4-H, 4-OMe, 4-NMe2, 4-Br, 2-Br) all worked well to give the corresponding ketones 3aq-3au in moderate to good yields. The reaction of iodomethylsilyl ether of 1-phenylpropan-2-ol afforded the corresponding product 3av via 1,5-HAT in 94% yield. The silyl ether with γ-benzyloxyl worked for the reaction, giving the corresponding product 3aw with C-H acylation at the alpha-position of the benzyloxy in 76% yield. In addition, ester was tolerated to deliver the corresponding product 3ax in 63% yield. Importantly, cholesterol and dehydroepiandrosterone worked well for the reaction, yielding the corresponding ketones 3ay and 3az in good yields, which showed potential application of this methodology in late-stage functionalization.

Substrate scope of three-component reaction

The carboacylation of alkenes provide easy access to ketones. Recently, a variety of groups have developed TM-catalyzed intramolecular63,64 or intermolecular carboacylation65,66,67,68 with acyl electrophiles. The NHC-catalyzed carboacylation of alkenes using aldehydes was also reported57,62. However, these reactions require active alkylation reagents as radical precursor. Following the two-component photoredox NHC/Pd-catalyzed C(sp3)-H acylation reaction, we then tested the corresponding three-component reaction with the addition of styrenes (Fig. 3). We were happy to find that the three-component reaction went smoothly under same conditions as the two-component reaction, and the corresponding ε-hydroxylketones 6 were obtained in good to high yields via alkylacylation of styrenes and following desilylation with tetrabutylammonium fluoride (TBAF).

The reaction of picolinaldehydes with different substituents at varied position (4-Cl, 4-OMe, 5-Me, 5-OMe, 5-Cl and 6-OMe) of the pyridine ring, went smoothly to give ε-hydroxylketones (6a–6g) in good to high yields. 4-Alkynylpicolinaldehyde was also tolerated, giving the ketone (6h) with multifunctional groups in 79% yield. Picolinaldehydes with varied additional 5-aryl groups (Ar’= C6H5, 4-OMeC6H4, 3,5-(CF3)2C6H3) or fused aryl groups (Ar’ = 1-naphthyl, 9-anthranyl, 9-phenanthryl) all worked well to afford the corresponding ε-hydroxylketones (6i-6n) in good to high yields. Several other heteroaryl aldehydes, such as quinoline-2-carbaldehyde, isoquinoline-3-carbaldehyde and thiazole-4-carbaldehyde, performed as well, providing the ε-hydroxylketones (6o-6q) in good yields.

A variety of styrenes were then tested for the reaction. It is worth noting that all styrenes with both electron-donating (X = OMe, tBu) and electron-withdrawing (X = F, Cl, Br, CN, CF3) groups at the para-position worked well, leading to ε-hydroxyl 6r-6x in good yields. Substituents at the meta-, ortho-position were tolerated to afford the products 6y-6z in high yields. The reaction of 2-vinylnaphthalene gave ketone 6aa in 69% yield. The reaction with 2-vinylpyridine or 4-vinylpyridine produced ketones 6ab and 6ac in good yields. Notably, the reaction of 1,1-diphenyl-buta-1,3-diene with conjugated C=C bonds showed exclusive regioselectivity (6ad), giving only the 1,2-alkylacylation of the terminal C=C bond in 98% yield. ω-Chlorine of the alkyl silyl ether was tolerated for the reaction, giving the ketone 6ae in 58% yield. In addition, the reaction of menthol derived silyl ether gave the corresponding product 6af with exclusive regioselectivity at the tertiary carbon of the isopropyl group even in the presence of two other tertiary C(sp3)-H. However, the reaction with internal or aliphatic olefins gave no desired ketone products under current conditions.

To verify the practicality of this protocol, gram-scale experiments and further chemical transformations of the products were carried out (Fig. 4). The reaction of alkyl silyl ether 7a from alcohol 7 (1.07 g, 6 mmol) with picolinaldehyde via γ-C(sp3)-H acylation under the standard conditions gave the silyl ether of chloroketone 3z, which was desilylated to give hemiketal 8 (1.01 g, 59% overall yield from 7) of the corresponding γ-hydroxyl ketone. Reduction of hemiketal 8 with LiAlH4 gave the corresponding 1,4-diol 9 in 56% yield. The reaction of ω-chlorohemiketal 8 with TMSCN afforded the corresponding silylated cyanohemiketal 10 in 55% yield. In addition, 1.12 g of ε-hydroxyl ketone 6a was also obtained from 4-methylpentan-2-ol via the process of silylation, three-component coupling with styrene and picolinaldehyde, and desilylation (60% overall yield from alcohol). Dess-Martin oxidation of ε-hydroxyl ketone 6a delivered diketone 11 in 79% yield.

Fig. 4: Gram-scale synthesis and further chemical transformations.
figure 4

a Si reagent (1.5 equiv.), imidazole (2.0 equiv.), THF, r.t. b Picolinaldehyde (1.5 equiv.), Pd(OAc)2 (10 mol%), PCy3 (20 mol%), N2 (20 mol%), Cs2CO3 (2.0 equiv.), PhCF3, 36 W Blue LEDs, r.t. c TBAF (2.0 equiv.), THF, r.t. d LiAlH4 (2.0 equiv.), Et2O, 0 °C. e TMSCN (3.0 equiv.), KF (3.0 equiv.), DMF, 80 °C. f (1) Si reagent (1.5 equiv.), imidazole (2.0 equiv.), THF, r.t.; (2) Styrene (2.0 equiv.), picolinaldehyde (1.5 equiv.), Pd(OAc)2 (10 mol%), PCy3 (20 mol%), N2 (20 mol%), Cs2CO3 (2.0 equiv.), PhCF3, 36 W Blue LEDs, r.t.; (3) TBAF (2.0 equiv.), THF, r.t. g Dess-Martin periodinane (2.0 equiv.), CH2Cl2, r.t.

Full size image

Discussion

Mechanistic studies

A series of control experiments were carried out, in order to investigate the mechanism of this reaction. No acylation product 3a was detected when PhSeSePh was added as a radical scavenger, while the adduct 12 from the alkyl radical and PhSeSePh was detected by HRMS and isolated in 43% yield (Fig. 5a). These results indicated that the existence of primary alkyl radical via dehalogenation of the iodomethylsilyl ether. Crossover experiment with (iodomethyl)trimethylsilane as radical precursor and silyl ether 1a’ without the iodomethyl substituent as proton source of possible intermolecular HAT was carried out, which resulted in no formation of ketone 14 via intermolecular HAT but ketone 13 in 73% yield without the participation of silyl ether 1a’ (fully recovered) (Fig. 5b). This result rules out the intermolecular HAT pathway and suggests the formation of tertiary/secondary alkyl radical from primary alkyl radical via the intramolecular 1,n-HAT for the titled reaction. In addition, the reaction could not be fully inhibited by the addition of D2O or MeOH to trap the possible carbon cation or anion (Fig. 5c), which support the radical coupling pathway instead of radical/polar crossover pathway for the reaction.

Fig. 5: Mechanistic experiments.
figure 5

a Radical trapping experiment with diphenyl diselenide. b Crossover experiment with the addition of (iodomethyl)trimethylsilane. c Cation and anion trapping experiment with the addition of deuterium oxide or methanol. N.D. not detected.

Full size image

The light on/off experiments revealed that blue light was essential for this reaction and the chain reaction process was excluded (see Supplementary Fig. 1). The UV−visible absorption spectra of the starting materials and catalysts were measured (see Supplementary Fig. 2). It was found that no apparent absorption for the alkyl silyl ether 1a, styrene 5a, picolinaldehyde 2a, carbene N2, phosphine ligand PCy3 observed. There is weak absorption for Pd(OAc)2, which could be dramatically enhanced by the addition of phosphine ligand and/or NHC. These results suggest that Pd species worked as photocatalysts.

Based on the mechanistic experiments and previous works59,62, the plausible catalytic cycle of this reaction was shown in Fig. 6. The Pd(0)L complex undergoes excitation by blue light to form the active [Pd(0)L]* catalyst. The single electron transfer (SET) between the [Pd(0)L]* catalyst and iodomethylsilyl ether 1 generates Pd(I) complex and methyl radical I with the leaving of iodide. The following intramolecular 1,n-HAT (n = 5, 6, 7) of methyl radical I produces secondary/tertiary radical species II. Meanwhile, the Breslow intermediate anion III, generated from aldehyde 2 under NHC catalysis, is single electron oxidized by Pd(I) complex to afford persistent ketyl radical IV and regenerates Pd(0)L catalyst. The coupling of persistent ketyl radical IV and transient alkyl radical II gives the adduct V of the two-component reaction, which is fragmented to afford the C(sp3)-H acylation product 3 and release NHC catalyst (two-component reaction).

Fig. 6: Proposed mechanism.
figure 6

The plausible mechanism for cooperative NHC/Pd-catalyzed functionalization of remote C(sp3)-H bond of alcohols.

Full size image

When styrene 5 is added, the addition of alkyl radical II to styrene 5 affords more stable benzylic radical VI, which is then coupled with ketyl radical IV to furnish the three-component adduct VII. The fragmentation of adduct VII gives the alkylacylation product 6’ and regenerates NHC catalyst (three-component reaction).

In summary, this work demonstrates the generation of ketyl radical and alkyl radical under the photoredox cooperative NHC/Pd catalysis, and provides two and three-component reactions for the synthesis of ketones via unactivated alkyl C(sp3)-H functionalization with aldehydes as the acylation reagents. The two-component reaction of iodomethylsilyl alkyl ether with aldehydes gave a variety β-, γ- and δ-silyloxylketones via 1,n-HAT (n = 5, 6, 7) of silylmethyl radicals to generate secondary or tertiary alkyl radicals and following coupling with ketyl radical from aldehydes under photoredox NHC catalysis. The three-component reaction with the addition of styrenes gave the corresponding ε-hydroxylketones via the generation of benzylic radical by the addition of alkyl radical to styrenes and following coupling with ketyl radical. Further investigation on photoredox cooperative NHC/TM catalysis is underway in our laboratory.

Methods

General procedure for two-component reaction

A 5 mL vial equipped with a stir bar was charged with N2 (0.06 mmol), Pd(OAc)2 (0.03 mmol), PCy3 (0.06 mmol), additive (0.06 mmol for benzaldehydes and enals) and 2.0 mL of PhCF3. After stirring for 30 min in glove box, to the solution was added Cs2CO3 (0.6 mmol), aldehydes 2 (0.45 mmol), alkyl silyl ethers 1 (0.3 mmol), and 1.0 mL of PhCF3. The reaction mixture was removed from the glove box and stirred under 36 W blue LED lights at room temperature for 16 h. The solution was concentrated under reduced pressure, and purified by column chromatography on silica gel to afford the desired ketones 3.

General procedure for three-component reaction

A 5 mL vial equipped with a stir bar was charged with N2 (0.06 mmol), Pd(OAc)2 (0.03 mmol), PCy3 (0.06 mmol) and 2.0 mL of PhCF3. After stirring for 30 min in glove box, to the solution was added Cs2CO3 (0.6 mmol), aldehydes 2 (0.45 mmol), styrenes 5 (0.6 mmol), alkyl silyl ethers 1 (0.3 mmol), and 1.0 mL of PhCF3. The reaction mixture was removed from the glove box and stirred under 36 W blue LED lights at room temperature for 16 h. Then, TBAF (0.6 mmol, 1.0 M in THF) was added. After being stirred for 2 h, the solution was concentrated under reduced pressure, and purified by column chromatography on silica gel to afford the desired ketones 6.