Remote Allylation of Unactivated C(sp³)–H Bonds Triggered by Photogenerated Amidyl Radicals

Abstract

The allylation reaction is a highly versatile transformation in chemical synthesis. While many elegant direct C(sp²)–H allylation reactions have been developed, the direct allylation of unactivated C(sp³)–H bonds is underdeveloped. By applying photoredox catalysis and a [1,5]-HAT process, herein we report a direct allylation of unactivated C(sp³)‒H bonds. This photocatalyzed transformation is tolerant of several functional groups in the amide and allylic chloride substrates. Various allyl-substituted amide products were obtained with good yields and high δ-selectivity.

Graphical Abstract

The allylation reaction is one of the most fundamental and powerful transformations to construct new C–C bonds in organic synthesis.^1–3 The allylic moiety in the product enables further transformations due to a versatile double bond that can be manipulated synthetically. Traditional allylations include nucleophilic allylation of carbonyl compounds and imines,^1a allylic substitution of carbanions,² transition-metal-catalyzed decarboxylative allylations^1b and allylic cross-couplings of aromatic halides.³ Among these methods, pre-installed functional groups are required in the starting materials.

To pursue higher reaction efficiency and atom economy, the direct functionalization of C–H bonds represents one of the most intriguing and advanced technologies in synthetic chemistry.⁴ Many successful direct allylation reactions of aromatic and vinylic C(sp²)–H bonds have been disclosed,^1d which allows for the rapid generation of molecular complexity (Scheme 1A). However, examples of C(sp³)–H allylation are rare. Current achievements in this field are limited to activated C(sp³)–H bonds,⁵ such as allylic and dibenzylic C(sp³)–H bonds, or C(sp³)–H bonds adjacent to carbonyl groups and heteroatoms (Scheme 1B). The direct allylation of unactivated C(sp³)–H bonds is still underdeveloped.

Scheme 1. Allylation Reactions of C–H Bonds.

Recently, radical mediated hydrogen-atom transfer (HAT) has provided an efficient and mild path to cleave unactivated C(sp³)–H bonds that allows for subsequent functionalizations.⁶ Breakthroughs mediated by Nitrogen-⁷ and Oxygen-centered⁸ radicals have been achieved by photocatalysis. Flechsig and Wang recently reported an allylation of sp³ C–H bonds with allyl sulfones activated by electron-withdrawing groups via photocatalytic fragmentation of pre-functionalized aryloxy amides (Scheme 1C).⁹ Herein, we report a direct allylation of unactivated C(sp³)–H bonds in unfunctionalized amides with simple allylic chlorides (Scheme 1D). By applying a 1,5-HAT process triggered by photoredox-generated amidyls, various allyl-substituted amides are obtained with good yields, high δ-selectivity, and high functional group tolerance.

As a model reaction for the direct allylation of unactivated C(sp³)–H bonds, trifluoroacetamide 1a and allylic chloride 2a were irradiated in the presence of a [Ir(dF(CF₃)ppy)₂(dtbpy)]PF₆ photocatalyst, Ni(COD)₂, bisoxazoline L1 and K₃PO₄ under 34W blue LED for 48 hours to afford the C–H allylation product 3aa in 72% isolated yield and high δ-selectivity (Table 1, entry 1).¹⁰ The corresponding allylic bromide and iodide generated considerable amounts of the undesired N–H allylation product 3aa’ (entries 2 and 3), which is most likely formed via nucleophilic substitution between the amide and the more electrophilic allylic halides. We did not observe any C–H allylation product in the absence of blue LED excitation (entry 4), K₃PO₄ (entry 5), or [Ir(dF(CF₃)ppy)₂(dtbpy)]PF₆ (entry 6). The use of other photoredox catalysts, such as [Ir(ppy)₃] (entry 7) and [Ru(bpy)]²⁺ (entry 8),¹¹ suppressed the formation of the desired product. Although the nickel catalyst was not necessary for product formation (entry 9, 48% yield), the yield of 3aa was improved in the presence of nickel catalyst (entry 1, 76%). Interestingly, the effect of the nickel catalyst on the efficiency of the reaction was sensitive to the structure of the nickel complex (entries 10–13). Other bidentate ligands L2–L4 resulted in lower yields of 3aa (entries 11–13). Additionally, the absence of L1 resulted in a significant loss in yield (entry 14). The combination of Ni(COD)₂ and L1 was therefore deemed optimal for product formation.

Table 1.

Reaction Conditions for δ-Allylation^a

Entry	Variation From Standard Conditions	Yieldb (3aa, %)	Yieldb (3aa’, %)
1a	none	76 (72c)	0
2	instead of 2a	28	52
3	instead of 2a	0	85
4d	No blue LED	0	8
5d	No K3PO4	0	0
6d	No lr catalyst	0	12
7d	[Ir(ppy)3] instead of lr catalyst	0	10
8d	[Ru(bpy)3]2+ instead of lr catalyst	0	8
9	No Ni(COD)2 and no L1	48	0
10	NiBr2 · (CH2OMe)2 instead of Ni(COD)2	36	0
11	L2 instead of L1	38	0
12	L3 instead of L1	58	0
13	L4 instead of L1	66	0
14	no L1	34	0

^aReaction conditions: [Ir(dF(CF₃)ppy)₂(dtbpy)]PF₆ (2 mol%), Ni(COD)₂ (10 mol%), L1 (11 mol%), 1a (0.1 mmol), 2a (0.15 mmol), K₃PO₄ (0.2 mmol), MeCN (0.5 mL), 34 W blue LED, 48 h. Conversion was >95% by ¹H NMR.

^b1H NMR yields with 1,3,5-tribromobenzene as internal standard.

^cIsolated yield.

^dConversion was 0–15% by ¹H NMR.

We evaluated the scope of the δ-selective C–H allylation with various amides 1a‒l and allylic chloride 2a (Table 2A). Amides bearing a methine carbon on the δ-position were examined (1a-1f). The reaction exhibited good tolerance to different hydrocarbon substituents, such as methyl (3aa, 3ba, 3da), ethyl (3ca), n-butyl (3ca), neopentyl (3ba), phenyl (3da), cyclopentyl (3ea), and cyclohexyl (3fa). Amides with methylene δ-carbons were also tested (3ga–3ja). Although only moderate yields were obtained (38–58%), the δ-selectivity was retained. To further demonstrate the synthetic potential of the reaction, we employed amide substrates derived from natural products, such as 1k from (L)-menthol and 1l from (S)-leucine. In both instances, the desired δ-allylation products 3ka and 3la were obtained in satisfying yields.¹²

Table 2.

Substrate Scope^a

^aReaction conditions: [Ir(dF(CF₃)ppy)₂(dtbpy)]PF₆ (2 mol%), Ni(COD)₂ (10 mol%), L1 (11 mol%), 1a (0.1 mmol), 2a (0.15 mmol), K₃PO₄ (0.2 mmol), MeCN (0.5 mL), 34 W blue LED, 48 h.

Remote sp³ C‒H allylations with various allylic chlorides 2b-2n were also investigated (Table 2b). Substrates bearing aliphatic substitution at the internal C2-position of the allylic system exhibited good tolerance to functional groups, such as an aromatic ring (3ab), trimethylsilane (3ac), TBS-protected alcohol (3ad), and alkene (3ae). The steric bulk of the C2-substituent did not influence the yield or δ-selectivity of product formation (3af-3ah). Even a sterically demanding adamantyl group afforded the desired product 3ah in 64% yield. Various aromatic C2-substituents were also examined (3ai-3am). The desired δ-selective C–H allylation products were obtained in synthetically useful yields (62–76% yield). A substrate with an electron-withdrawing group was also tested. With an ester group on the C2-position, allylic chloride 2n afforded the desired δ-allylation product 3an in 38% yield, with significant amounts of the N-allylation byproduct (47% yield).¹³

Although we do not fully understand the role of the Ni (0) complex in improving the yield of the photocatalyzed allylation of unactivated C(sp³)-H bonds, we conducted a series of deuterium-labeled experiments to gain some insight (Scheme 2). When deuterated substrate 4 ((3-chloroprop-1-en-2-yl-3,3-d₂) benzene) was used under optimal conditions (Scheme 2A), the ratio of products 6:7 was the same in the presence or absence of the Ni (0) complex, which suggests that the Ni (0) complex did not significantly alter the reaction pathway. Based on these observations, we can rule out the formation of a Ni(II) (Ƞ³) allyl-π-complex from allylic chloride, which would have formed equal amounts of 6 and 7.

Scheme 2. Deuterium Labeled Experiments.

We also propose the Ni (0) complex improves the yield of the photocatalyzed allylation by stabilizing the allylic chloride substrate. The high ratio of 6:7 (88:12) is consistent with a C–C bond forming mechanism via radical addition to the double bond of the allylic chloride. The formation of small amounts of allylation product 7 can be explained by the photoisomerization of allylic chloride 4 to 5. When deuterated allylic chloride 4 was mixed with 2 mol% Ir(III) photocatalyst under blue LED light (Scheme 2B, entry 1), considerable isomerization from 4 to 5 was observed. However, this isomerization was depressed to 11% in the presence of Ni (0) complex (entry 2). To understand this observation, we examined the byproducts generated in the presence and absence of Ni (0) complex. When Ni (0) complex was omitted from the optimized reaction conditions (Scheme 2A), we observed considerable amounts of the allylic chloride dimerization byproduct. We believe this dimer is formed through the radical pair shown in Scheme 2B, which is prevalent in the absence of the Ni (0) complex. Although the presence of the Ni (0) complex suppresses photoisomerization, it does not completely shut down photoisomerization (Scheme 2B, entry 2). We surmise that the small amount of allylation product 7 formed in the optimized reaction with the Ni (0) complex (Scheme 2A) may be the result of small amounts of photoisomerization even in the presence of the Ni (0) complex. Therefore, the Ni (0)/L1 complex stabilizes the allylic chloride substrate to isomerization and decomposition during exposure to blue LED irradiation and Ir photoredox catalyst.

On the basis of these investigations, we propose the mechanism depicted in Scheme 3. The activated photocatalyst Ir(III)* oxidizes deprotonated amide 1a by single electron transfer to generate amidyl 8, which undergoes a 1,5-hydrogen atom transfer (1,5-HAT) to cleave a sp³ C‒H bond at the δ-position. The generated carbon radical 9 is trapped by allylic chloride 2a to provide 10, which yields δ-allylation product 3aa and chlorine radical through β-scission.¹⁴ The chlorine radical then oxidizes the Ir(II) photocatalyst to regenerate Ir(III) via another single electron transfer process to complete the catalytic cycle. During this process, the Ni(0) complex presumably stabilizes the allylic chloride substrate to isomerization and decomposition, which results in improved yield of the desired allylation product 3aa.

Scheme 3. Proposed Mechanism.

In conclusion, by using photoredox catalysis and a [1,5]-HAT process, we have developed a direct allylation of unactivated C(sp³)‒H bonds. This photocatalyzed transformation has a broad substrate scope of substituted amides and allylic chlorides. Various allyl-substituted amide products were obtained with good yields and high δ-selectivity.