Allylic C(sp3)-H alkylation via synergistic organo- and photoredox catalyzed radical addition to imines.

A new catalytic method for the direct alkylation of allylic C(sp3)-H bonds from unactivated alkenes via synergistic organo- and photoredox catalysis is described. The transformation achieves an efficient, redox-neutral synthesis of homoallylamines with broad functional group tolerance, under very mild reaction conditions. Mechanistic investigations indicate that the reaction proceeds through the N-centered radical intermediate which is generated by the allylic radical addition to the imine. This journal is © The Royal Society of Chemistry.

Introduction

The selective C–H bond functionalization has expanded rapidly over the past decade owing to its tremendous potential to streamline the synthesis of natural products, pharmaceuticals, agrochemicals and materials.[1,2] Allylic C–H activations are of particular interest due to the synthetic versatility of the resulting alkenes and opportunities for further functionalization. Despite the recent progress in the direct oxidation,[3] amination,[4] carboxylation,[5] alkylation[6] and arylation[7] of allylic C–H bonds, direct C(sp3)–C(sp3) bond formations remain a major challenge for synthetic organic chemistry.

Homoallylic amines are versatile building blocks and precursors for the synthesis of bioactive molecules and functional materials.[8] Thus, routes for their efficient preparation have received considerable interest (Scheme 1A and B).[8] Although advances in conventional organic synthesis have been achieved, several issues need to be addressed. These include the frequent requirement of cryogenic conditions, multi-step syntheses of the organometallic reagents and the generation of stoichiometric amounts of by-products.[9] Recent developments in visible light photocatalyzed reactions have enabled new synthetic pathways towards the functionalization of imines.[10,11] Notably, the polarity-reversed allylation of imines with prefunctionalized alkenes has been reported.[10] This reaction, however, requires stoichiometric amount of Hantzsch ester as electron/proton donor and activator. We recently questioned whether the direct coupling between an aminoalkyl moiety and an allylic C(sp3)–H bond of simple olefins could be feasible. This strategy is generically useful considering it can overcome the challenges of functional group compatibility associated with preparation of allylation precursors and the use of excess reductants. Therefore, it could offer a wealth of opportunities for elaboration of both complex molecules and feedstock olefins.

Scheme 1

(A) The reaction development of direct allylic C–H activation of olefins. (B) Classical methods towards homoallylic amines synthesis. (C) Photoredox activation of allylic C(sp3)–H bonds for the homoallylic amines synthesis, initial proposed reaction pathways.

In line with our recent reports on reductive Umpolung of imines[10] as well as recent development on synergistic organo- and photoredox catalysis,[12] we envisioned that upon visible light irradiation, the resulting oxidizing excited state of a photocatalyst PC(* could be quenched by a thiol catalyst RSH through proton coupled electron transfer (PCET)[13] in the presence of a basic additive[14] to give the corresponding reduced PC( complex and a transient thiyl radical RS·.[15] The generated electrophilic thiyl radical would then react with the alkene substrate (allylic C–H bond dissociation energy BDE is about 340 kJ mol−1) via hydrogen atom transfer to form an allylic radical along with the regeneration of thiol catalyst (S–H BDE is typically 365 kJ mol−1).[16]

Next, the oxidation of PC( species with imine ArCH = NR′ via single electron transfer would result in the formation of the aminoalkyl radical anion[17] and the following radical–radical cross coupling[18] with the allylic radical[19] would result in the desired C–C bond formation (pathway 1). Alternatively, the allylic radical could add to the imine to form a N-centred radical, which is subsequently reduced by PC( to give the desired product (pathway 2). Herein, we report the successful development of a new protocol for the direct alkylation of allylic C(sp3)–H bonds via synergistic organo- and photoredox catalysis (Scheme 1C) and provide insight into the reaction mechanism of this efficient addition reaction.

Results and discussions

Based on our mechanistic hypothesis, we first evaluated the reaction conditions with imine 1a and cyclohexene 2a. To our delight, we found that the reaction of 1a with 2a (5 equiv) in the presence of [Ir(ppy)2(dtbbpy)]PF63a (ppy = 2,2′-bipyridine, dbbpy = 4,4′-di-tert-butyl-2,2′-bipyridine, vs. SCE in MeCN)[20] (1 mol%), organocatalyst 4 (5 mol%) and Li2CO3 (20 mol%) in DMF under irradiation with 24 W blue LEDs (λmax = 405 nm) afforded the desired homoallylic amine product 5a in 81% yield (Table 1, entry 1). Remarkably, we did not observe any reductive homocoupling product under these reaction conditions.[10] Control experiments confirmed that the reaction did not proceed in the absence of the photocatalyst and light (entries 2 and 3). Moreover, only traces of the desired product were obtained in the absence of thiol 4, indicating the necessity of the hydrogen atom transfer (HAT) catalyst (entry 4). The use of other Ir-based photocatalysts gave 5a in lower yields (entries 5–7). In sharp contrast, only 5% of the desired coupling product was obtained when less reducing Ru(bpy)3(PF6)2 (EII/I1/2 = −1.35 V vs. SCE)[21] was employed as photocatalyst (entry 8). The organic photocatalyst mesityl acridinium which is known as a weak reductant in its reduced form was next examined,[22] but no product was observed (entry 9). Interestingly, a moderate yield was obtained in the absence of a basic additive (entry 10). In addition, we found that DMF was a suitable solvent while MeCN was inferior (entry 11). Notably, 34% of 5a was obtained when 2a was used as limiting reagent (entry 12). The use of other thiols failed to improve the yield (see ESI†).

Optimization of the reaction conditionsa

Entrya	Deviation from standard conditions	Yieldb (%)
1	None	81
2	No photocatalyst	0
3	No light	0
4	No thiol 4	Trace
5	Photocatalyst 3b instead of 3a	38
6	Photocatalyst 3c instead of 3a	58
7	Photocatalyst 3d instead of 3a	64
8	Photocatalyst 3e instead of 3a	5
9	Photocatalyst 3f instead of 3a	0
10	No Li2CO3	55
11	MeCN as solvent	66
12	2a as limiting reagent	34

Reaction conditions: 1a (0.2 mmol), cyclohexene 2a (1.0 mmol), Li2CO3 (20 mol%), iPr3SiSH 4 (5 mol%), 3a–f (1 mol%), rt, solvent (2 mL), 24 W LEDs (450 nm), 24 h.

Yield of product after isolation.

Under the optimized reaction conditions, we next screened various imines 1 in reaction with cyclohexene 2a. As shown in Table 2, a wide range of N-aryl imines with various substituents on the aryl ring underwent the aminoalkylation smoothly, affording the desired homoallylic amines 5b–k in good to excellent yields (61–92%) regardless of their electronic nature as well as substituent position. Notably, imines containing heterocyclic groups such as thiophene and pyridine were compatible with the reaction conditions and gave the corresponding products 5l and 5m in 46% and 63% yield, respectively. Moreover, the reaction proceeded in excellent yields for imines bearing the readily removable N-p-methoxyphenyl group (5n–q, 74–83%). In addition to N-aryl imines, imine 1r derived from an alkyl amine was also tolerated, albeit in moderate yield (5r, 58%).

Scope of various imine substratesa

Reaction conditions: imine 1 (0.3 mmol), cyclohexene 2a (1.5 mmol), Li2CO3 (20 mol%), iPr3SiSH 4 (5 mol%), 3a (1 mol%), DMF (1.5 mL), 24 W LEDs (450 nm), rt, 24 h; yields after isolation; dr between 1 : 1 to 1.2 : 1.

Encouraged by these results, we next evaluated more challenging ketimine substrates for the construction of all-carbon quaternary centers. Not surprisingly, aromatic ketimines were converted to the corresponding homoallylamines in good yields (5s–v, 51–87%). Next, we evaluated the scope of allylic radicals that can be generated with this protocol. Under the optimized conditions, a variety of cyclic and acyclic alkenes were successfully aminoalkylated using imine 1a, furnishing the corresponding products in good yields (Table 3). In addition to cyclohexene 2a described above, several other cyclic alkenes 2b–d (five, seven and eight membered derivatives) were tested and the corresponding homoallylic amines 6a–d were prepared in good to excellent yields under the reaction conditions (6a–d, 62–90%). An excellent regioselectivity was observed with 1-(tert-butyl)cyclohex-1-ene (2d), indicating that steric effects play a predominant role in determining the regioselectivity. For asymmetrically substituted alkene 2f which has two different C–H bonds at the allylic sites, isomers of aminoalkylation were observed. The selective formation of 6f and 6f′ suggests that the proton abstraction of allylic methine C–H bonds is favoured over primary C–H bonds due to the formation of more stabilized carbon centred radical. It is notable that unprotected allylic alcohol 2g was also aminoalkylated providing the corresponding product 6g in 51% yield.

Scope of alkene substratesab

Reaction conditions: imine 1a (0.2 mmol), alkene 2 (1.0 mmol), Li2CO3 (20 mol%), iPr3SiSH 4 (5 mol%), 3a (1 mol%), DMF (1.0 mL), 24 W LEDs (450 nm), rt, 24 h; yields after purification.

dr between 1 : 1 to 1.2 : 1.

iPr3SiSH 4 (10 mol%).

Owing to the wide applications of natural product modification in the fields of medicine, health science, pharmacy and biology, we subsequently focused on the aminoalkylation of several naturally occurring substrates as well as their derivatives (Fig. 1). Both regioisomers of pinenes could be converted to the homoallylic amines 6h and 6i with excellent regioselectivities at the methylene C–H bonds. This can be explained by the electrophilic thiol radical selectively abstracts the allylic C–H bond to form the more stabilized radical of pinenes. Meanwhile, the steric effect also plays a major part in determining the regioselectivity, the C–C bond formation selectively takes place in the sterically less-hindered allylic position. Steroid derivatives such as cholesterol and pregnenolone could also be functionalized under the photoredox conditions and the aminoalkylation took place selectively at the allylic position, affording the corresponding diastereomers 6j–l in good yields. In comparison with literature precedent,[7] the low to moderate diastereoselectivities indicates that a possible different reaction pathway was involved for the current transformation.

Fig. 1

Late-stage diversification of complex molecules. Reaction conditions: imine 1 (0.2 mmol), alkene 2 (1.0 mmol), Li2CO3 (20 mol%), iPr3SiSH 4 (5 mol%), 3a (1 mol%), DMF (1.0 mL), 24 W LEDs (450 nm), rt, 24 h, for dr see ESI.†2 (0.6 mmol), iPr3SiSH 4 (10 mol%), DMF (2 mL). Isomers ratios: 6h (1.2 : 1 : 1 : 1), 6i (3 : 3 : 1 : 1), 6j (1.6 : 1.6 : 1 : 1), 6k (1.2 : 1 : 1 : 1), 6l (1.9 : 1.9 : 1).

In order to gain some insight into the reaction mechanism, we first performed radical inhibition experiments. The reaction was greatly suppressed in the presence of various radical inhibitors, indicating the involvement of a radical process. It is noteworthy that radical initiators failed to give any product even at elevated temperature (see ESI†). Next, a series of experiments were conducted to better understand the nature of photoinduced single electron transfer step. The photocatalyst [Ir(ppy)2(dtbbpy)]PF63a on photoexcitation can result in a long-lived triplet excited state *3a (τ0 = 454.71 ± 0.30 ns in DMF, Fig. S1†), which can result in SET in the presence of suitable quencher. Therefore, Stern–Volmer quenching experiments were performed in order to confirm the quenching of 3a. Initially, steady-state fluorescence quenching of 3a using different concentrations of silanethiol 4 was carried out in the presence of Li2CO3, where a predominant quenching was observed with a linear correlation (Fig. 2a). Further, a linear correlation similar to steady-state experiments was also observed by time-resolved emission spectroscopy studies where excited-state life-time of *3a is quenched by different concentrations of 4 in the presence of Li2CO3 (Fig. 2a). Linear correlation by both the time-resolved and steady-state Stern–Volmer quenching experiments reveal the dynamic quenching of triplet excited-state of *3a by 4 in the presence of base. Next, a similar quenching experiments was carried out in the absence of base which also displayed considerable quenching of *3a (Fig. 2a). In-line with the observed result, a moderate yield of 55% of homoallylic amine 5a was obtained in the absence of a basic additive (Table 1, entry 10). Simultaneous steady-state Stern–Volmer quenching studies using 1a proved that there is no quenching of excited-state of *3a even in the absence and presence of base (Fig. 2b). In order to understand the electron transfer event better, the electron transfer rate constant kET was determined by time-resolved emission spectroscopy measurements. An electron transfer rate constant, kET = (2.22 ± 0.06) × 109 L mol−1 s−1 was obtained by plotting the difference between the observed rate constant (kobs) and rate constant without quencher (k0) versus different concentrations of 4 in presence of base which gave a linear fit (Fig. 2c and d).

Fig. 2

(a) Steady-state and time-resolved Stern–Volmer quenching of 3a with 4 and the mixture of 4 + base (Li2CO3); (b) Steady-state Stern–Volmer quenching experiment with 1a and the mixture of 1a + base (Li2CO3); (c) phosphorescence lifetimes of excited-state photocatalyst *3a (1 × 10−5 M) at different concentrations of quencher mixture 4 + base (B = Li2CO3). (d) Stern–Volmer analysis yielded a rate constant, kET, of 2.22 × 109 L mol−1 s−1 by the PCET between *3a and 4 in the presence of base; (e & f) proposed mechanistic pathways for aminoalkylation of allylic C(sp3)–H bonds.

Initially, we assumed that the alkylation process may proceed through the radical–radical coupling pathway (Fig. 2e, pathway-1), where the allylic radical couples with the persistent radical anion intermediate that is generated by the SET reduction of imine. However, the lack of informative byproducts formation including reduced amine and vicinal diamine derived from the N-centered radical anion prompted us to think otherwise. And if such an intermediate is involved in the reaction process, there should be a yield dependence on the electronics of the imine. Based on our substrate scope both electron rich (1b) and electron deficient (1g) imines gave the expected product in good yield which indicates the possibility of an alternate pathway. Furthermore, the reduction potential of imines 1a and 1c (E1/2 = −1.91 and −2.01 V vs. SCE respectively)[10] is very low compared to the redox potential of photocatalyst 3a (IrIII/IrII = −1.51 V) which does not favour the formation of radical anion intermediate. Nevertheless, both the imines 1a and 1c gave the expected product in good yield. All these observations strongly indicate that the reaction proceeds through the N-centered radical intermediate which is generated by the allylic radical addition to the imine. This N-centered radical subsequently gets reduced by IrII to give the expected product (Fig. 2f, pathway-2).[23]

In conclusion, we developed a new method for the efficient synthesis of homoallylic amines through the direct alkylation of allylic C(sp3)–H bonds of simple alkenes with imines by using the combination of photoredox and organocatalysis. The reaction tolerates a wide range of functional groups and proceeds under mild reaction conditions. The reactivity and utility of this protocol can be further emphasized in the late stage allylic C(sp3)–H bond functionalization of a range of complex natural products and derivatives. The strategy developed herein could be of relevance for organic synthesis as well as pharmaceutical development and may inspire the discovery other direct allylic C(sp3)–H functionalizations.

Conflicts of interest

There are no conflicts to declare.