Alkene 1,2‐Difunctionalization by Radical Alkenyl Migration

Abstract

Transition‐metal‐free radical α‐perfluoroalkylation with the accompanying vicinal β‐alkenylation of unactivated alkenes is presented. These radical cascades proceed by means of 1,4‐ or 1,5‐alkenyl migration by electron catalysis on readily accessed allylic alcohols. The reactions comprise a regioselective perfluoroalkyl radical addition with subsequent alkenyl migration and concomitant deprotonation to generate a ketyl radical anion that sustains the chain as a single‐electron‐transfer reducing reagent.

Vicinal alkene difunctionalization is a very powerful strategy to increase complexity in organic compounds by a modular approach.1 In the past decades, radical 1,2‐difunctionalization of alkenes has attracted great interest and significant results have been achieved.2, 3 Radical alkene perfluoroalkylation with concomitant β‐functionalization has received great attention,4, 5 owing to the fact that a perfluoroalkyl group (R_f) improves the solubility, bioavailability, lipophilicity, and metabolic stability of an organic compound.6, 7 Therefore, the development of methods for alkene perfluoroalkylation is of great importance.

Vicinal difunctionalization through radical perfluoroalkylation and subsequent intramolecular formyl,8a aryl,8b cyano,8c and heteroaryl4e, 8d migration has been documented previously (Scheme 1). Moreover, Zhu and co‐workers and our group reported cascade reactions involving a radical 1,4‐alkynyl‐group migration.9 Although there are reports on radical 1,2‐ or 1,3‐vinyl migrations,10 the corresponding 1,4‐ or 1,5‐vinyl migrations are not yet established in synthesis.11 Considering the mechanisms of these reactions, 1,2‐ and 1,3‐vinyl migrations proceed through radical 3‐exo and 4‐exo cyclizations that lead to highly strained three‐ and four‐membered intermediates, which readily undergo ring opening by radical β‐cleavage. Accordingly, 1,4‐ and 1,5‐vinyl migrations have to proceed through 5‐exo or 6‐exo cyclizations to give thermodynamically more stable five‐ or six‐membered cyclized radicals, which lack any strong driving force for ring opening. Indeed, 5‐exo and 6‐exo radical cyclizations are highly valuable in synthesis and belong to the most intensively studied radical processes to date. The β‐cleavage of a cyclized radical resulting from a 5‐exo process is only observed if trapping is very slow and if the ring‐opened radical shows high thermodynamic stability.12 Considering these prerequisites, we designed radical cascades comprising 5‐exo or 6‐exo cyclizations with subsequent ring opening leading to highly stabilized ketyl radical anions. These cascades should proceed in the absence of any efficient radical‐trapping reagent, allowing the challenging β‐C−C bond cleavage to occur. To suppress endo‐type cyclization an R⁴‐substituent should be installed (Scheme 1).

Scheme 1.

Intramolecular‐radical‐group migration.

Herein, we introduce a simple and efficient method for α‐perfluoroalkylation with concomitant β‐alkenylation of unactivated alkenes involving a radical 1,4‐ or 1,5‐alkenyl migration.

(E)‐3‐Methyl‐1‐phenylhepta‐1,6‐dien‐3‐ol 1 a was chosen as a model substrate. The alcohol 1 a was first reacted with lithium hexamethyldisilazide (LiHMDS, 1.2 equiv) in 1.25 mL of 1,2‐dimethoxyethane (DME) at room temperature for 0.5 h. After deprotonation, 1,4‐diazabicyclo[2.2.2]octane (DABCO, 1.5 equiv)13 and perfluorobutyl iodide 2 a (1.8 equiv) were added sequentially and the mixture was stirred under visible‐light irradiation [using a Philips Master HPI‐T Plus (400 W) bulb] at 50 °C for 18 hours. To our delight, the 1,4‐alkenyl migration product 3 a was obtained in 34 % yield with complete E selectivity (Table 1, entry 1). The yield increased to 40 % when LiOH was used as a base (Table 1, entry 2); however, with NaOH or KOH the yield decreased (Table 1, entries 3 and 4). In the presence of Na₂CO₃ or K₂CO₃ 3 a was formed in 35 or 38 % yield, respectively (Table 1, entries 5 and 6), and KOtBu was found not to be an efficient base to mediate this cascade (Table 1, entry 7). The highest yield in this series (41 %) was obtained by using K₃PO₄ (Table 1, entry 8). A solvent screen revealed that dichloroethane (DCE) provided an improved result (52 %) (Table 1, entries 9–13). Replacing DABCO by other amines, such as N,N,N′,N′‐tetramethylethane‐1,2‐diamine (TMEDA), 1,8‐diazabicyclo[5.4.0]undec‐7‐ene (DBU), and N,N,N′,N′‐tetramethyl‐1,3‐propanediamine (TMPDA), afforded lower yields (Table 1, entries 14–16).14 Varying the amount of base and DABCO showed that the highest yield (67 %) was obtained by using 1 a (0.1 mmol), 2 a (1.8 equiv), K₃PO₄ (2.0 equiv), and DABCO (1.2 equiv) in 1.25 mL of DCE with stirring under visible light at 50 °C for 24 h (Table 1, entry 17). We also examined other phosphate salts, such as Na₃PO₄, Li₃PO₄, K₂HPO₄, and KH₂PO₄, but lower yields were obtained (Table 1, entries 18–21). Notably, the cascade reaction did not proceed without visible‐light irradiation (Table 1, entry 22) and very low yields were achieved in the absence of K₃PO₄ or DABCO (Table 1, entries 23 and 24).

Table 1.

Reaction optimization.^[a]

Entry	Base	Solvent	Amine	Yield of 3 a [%][b]
1	LiHMDS	DME	DABCO	34
2	LiOH	DME	DABCO	40
3	NaOH	DME	DABCO	25
4	KOH	DME	DABCO	27
5	Na2CO3	DME	DABCO	35
6	K2CO3	DME	DABCO	38
7	KOtBu	DME	DABCO	16
8	K3PO4	DME	DABCO	41
9	K3PO4	1,4‐dioxane	DABCO	34
10	K3PO4	DMA	DABCO	44
11	K3PO4	DMF	DABCO	–
12	K3PO4	DCM	DABCO	44
13	K3PO4	DCE	DABCO	52
14	K3PO4	DCE	TMEDA	21
15	K3PO4	DCE	DBU	10
16	K3PO4	DCE	TMPDA	–
17[c]	K3PO4	DCE	DABCO	67 (63)[d]
18	Na3PO4	DCE	DABCO	38
19	Li3PO4	DCE	DABCO	29
20	K2HPO4	DCE	DABCO	52
21	KH2PO4	DCE	DABCO	29
22[c,e]	K3PO4	DCE	DABCO	–
23[c,f]	K3PO4	DCE	–	3
24[c,g]	–	DCE	DABCO	10

[a] The reaction was conducted with 1 (0.1 mmol), 2 a (1.8 equiv), base (1.2 equiv), and amine (1.5 equiv) in 1.25 mL of solvent under visible‐light irradiation [using a Philips Master HPI‐T Plus (400 W) bulb] at 50 °C for 18 h. [b] Determined by ¹H NMR analysis by using 1‐fluoro‐4‐methylbenzene as the internal standard. [c] K₃PO₄ (2.0 equiv) and DABCO (1.2 equiv) were used for 24 h. [d] Yield of isolated product in parenthesis. [e] The reaction was conducted without visible‐light irradiation. [f] The reaction was conducted without DABCO. [g] The reaction was conducted without base. DMA=N,N‐dimethylacetamide; DMF=N,N‐dimethylformamide; DCM=dichloromethane.

With the optimized reaction conditions in hand, we investigated the scope of the reaction by keeping perfluorobutyl iodide 2 a as the C‐radical precursor and systematically varied the migrating styrenyl group (Table 2). Electronic effects at the para position in the aryl moiety are not pronounced and the corresponding products 3 b–3 f were formed in moderate‐to‐good yields (Table 2, entries 1–5). The meta‐ and ortho‐methyl substituted congeners 1 g and 1 h provided the targeted 3 g and 3 h in 61 and 52 % yield, respectively (Table 2, entries 6 and 7). Alcohols bearing di‐ and trisubstituted styryl groups, such as 1 i and 1 j, afforded the corresponding ketones 3 i and 3 j in 51 and 53 % yield (Table 2, entries 8 and 9), indicating that steric effects at the aryl moiety in the migrating styrenyl group do not play a major role. Notably, the 1‐naphthyl and 2‐pyridyl groups are both tolerated as substituents (see 3 k, 3 l, Table 2, entries 10 and 11). The silylated allylic alcohol 1 m also worked well and 3 m was isolated in 50 % yield (Table 2, entry 12). Cyclohexyl‐substituted allylic alcohol 1 n was also suitable for this migration reaction; however, only a moderate 23 % yield of 3 n was obtained (Table 2, entry 13). Next, we studied the radical styrenyl migration on various alcohols of type 1 by varying the R¹ and R² substituents (Scheme 2). The butyl‐ and isopropyl‐substituted allylic alcohols 1 o and 1 p worked well and ketones 3 o and 3 p were isolated in 62 and 67 % yield, respectively. A higher yield was obtained with the tertiary benzylic alcohol 1 q to give 3 q (74 %). Notably, styrenyl migration also works for tertiary alkyl radicals, as documented by the successful preparation of 3 r bearing an all‐carbon quaternary center (84 %). Importantly, the reaction is not limited to the 1,4‐alkenyl migration: The allylic alcohol 1 s reacted in 50 % yield to provide the ketone 3 s, resulting from a 1,5‐alkenyl migration. As expected, initial perfluoroalkyl radical addition on the trisubstituted alkene 1 t occurred at the less hindered site and the ketone 3 t, derived from a 1,5‐alkenyl migration, was isolated in 49 % yield. Phenyl‐substituted alkene 1 u could also undergo this migration reaction and 3 u was obtained in moderate 30 % yield as a 2.6:1 diastereoisomeric mixture.

Table 2.

Variation of the alkenyl substituent.^[a]

Entry	R	Product	Yield [%][b]
1	1 b, 4‐MeC6H4	3 b	70
2	1 c, 4‐tBuC6H4	3 c	69
3	1 d, 4‐MeOC6H4	3 d	56
4	1 e, 4‐FC6H4	3 e	50
5	1 f, 4‐ClC6H4	3 f	72
6	1 g, 3‐MeC6H4	3 g	61
7	1 h, 2‐MeC6H4	3 h	52
8	1 i, 2,4,6‐Me3C6H2	3 i	51
9	1 j, 3,5‐(MeO)2C6H3	3 j	53
10	1 k, 1‐naphthyl	3 k	45
11	1 l, 2‐pyridyl	3 l	43
12	1 m, iPr3Si	3 m	50
13[c]	1 n, Cy	3 n	23

[a] The reaction was conducted with 1 (0.1 mmol), 2 a (1.8 equiv), K₃PO₄ (2.0 equiv), and DABCO (1.2 equiv) in 1.25 mL of DCE under visible‐light irradiation [using a Philips Master HPI‐T Plus (400 W) bulb] at 50 °C for 24 h. [b] Yield of isolated product. [c] The reaction was conducted at 0.2 mmol scale.

Scheme 2.

Variation of the radical acceptor and the perfluoroalkyl iodides. [a] 3.6 equivalents of CF₃I were used.

Other perfluoroalkyl groups, including the important trifluoromethyl moiety, could also be introduced by this method as documented by the preparation of ketones 3 v–3 y, which were isolated in good‐to‐excellent yields. The reaction of 1 r with ICF₂CF₂Cl provided 3 z (84 %) and products derived from chloride fragmentation were not identified in this transformation. Other alkyl‐radical precursors, such as ethyl 2‐iodoacetate, 2‐iodoacetonitrile, 2‐iodo‐2‐methylpropanenitrile, 2‐bromo‐1‐phenylethan‐1‐one, and 1‐iodoadamantane were not suitable for this migration reaction.

Based on the above results, a plausible mechanism is suggested in Scheme 3. Initiation occurs by visible‐light irradiation of the halogen‐bond (XB) complex14 formed between the perfluoroalkyl iodide and DABCO to give the corresponding perfluoroalkyl radical. This radical adds at the terminal position of the alkene in alcohol 1 to give the adduct radical A. The internal double bond is well shielded by the neighboring quaternary carbon center and, therefore, the internal double bond remains unreacted at this stage. Radical 5‐exo or 6‐exo cyclization leads to the cyclized radical B. Unlike our initial design in which the cascade reactions were planned to be conducted on deprotonated allylic alcohols, reaction optimization revealed that these transformations work most efficiently with K₃PO₄ in combination with DABCO. Both of these bases are too weak to deprotonate a tertiary alcohol. We assume that the phosphate anion undergoes hydrogen bonding with the tertiary alcohol and this leads to an activation of the β‐C−C bond towards homolytic cleavage. This proposal is reminiscent of the β‐C−H bond weakening in phosphate‐complexed alcohols suggested by MacMillan and co‐workers.15 Radical β‐C−C bond cleavage will then afford the ketyl radical anion C. Owing to the increase of the acidity of the hydroxy group in α‐hydroxy‐α‐alkyl carbon radicals compared with that of the parent alcohols, we believe that deprotonation by K₃PO₄ occurs during β‐C−C cleavage. As previously shown, such ketyl radical anions are very good single‐electron transfer (SET) reducing reagents.16 Hence, SET reduction of the perfluoroalkyl iodide will give the corresponding ketone 3, along with the perfluoroalkyl radical sustaining the chain. The overall cascade is part of an electron‐catalyzed process.17

Scheme 3.

Proposed mechanism.

In summary, we have developed a novel and efficient method for the radical perfluoroalkylation of unactivated alkenes with accompanying β‐alkenylation. The radical cascade proceeds by a 1,4‐ or 1,5‐alkenyl migration, a reaction that is currently not established in synthetic methodology. The chain reaction belongs to an electron‐catalyzed process and does not require any transition‐metal‐based redox catalyst.

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.

Supplementary

X. Tang, A. Studer, Angew. Chem. Int. Ed. 2018, 57, 814.