Optimization of reaction conditions. Our optimization began by reacting 2-vinylnaphthalene 1a with oxime ester 3a and TMSCN in a ratio of 1:3:3 in DMA under the dual photoredox and copper catalysis (Table 1). After some experimentation63, we found that the target three-component reaction indeed occurred to give a moderate yield of desired product 6aa with 88% ee, when using photocatalyst fac-Ir(ppy)3 (1 mol%) and a combination of Cu(CH3CN)4PF6 (0.5 mol%) and Box-type ligand L1 (0.6 mol%) under irradiation of purple LEDs (Table 1, entry 1). However, several competing processes were also involved. For example, in addition to 6aa, a significant amount of side products sp-1, sp-2, and sp-3 were also detected, which might result from acyl radical 3a-II-mediated two-component cross-coupling with 1a and TMSCN, or its own dimerization. A brief screening of typical solvents such as DMF, CH3CN, and THF showed that DMA was still the best of choice in terms of reaction efficiency (Table 1, entry 1 vs 2–5). Notably, when the catalyst loading of fac-Ir(ppy)3 was decreased to 0.8 mol%, a cleaner reaction was observed, and a 64% yield of 6aa was obtained without effect on the enantioselectivity (Table 1, entry 6). An extensive survey of other commonly used copper salts and chiral ligands established that the combination of Cu(CH3CN)4PF6 and Box-type ligand L1 were superior to others (Table 1, entries 6–8). Further optimization studies with respect to the loading of copper salt and concentration confirmed that a combination of 1.5 mol% of Cu(CH3CN)4PF6 and 2.25 mol% of ligand L1 with 0.8 mol% of fac-Ir(ppy)3 at a concentration of 0.04 M gave the best results, with 6aa being isolated with 74% yield and 90% ee (Table 1, entry 11). It was postulated that rational tuning of catalyst loading can help regulate the concentration of the reactive radical species, thus suppressing the competing side reaction pathways. As expected, a series of control experimental results established that each component (light, photocatalyst, and copper salt) is critical to this asymmetric alkene vicinal dicarbofunctionalization reaction63.
Substrate scope. With optimized reaction conditions established, we first investigated the substrate scope of alkenes by reacting with 3a and TMSCN (Fig. 2). Notably, most of these alkenes are inexpensive and commercially available feedstock chemicals. As shown in Fig. 2A, aside from 1a, simple neutral styrene 1b and a range of styrene derivatives 1c-j with electron-donating (e.g., Me, tBu, Ph) or electron-withdrawing (e.g., F, Cl, Br, OAc, Bpin) functional groups at the para-position of the aromatic ring are well tolerated, furnishing the corresponding products 6ba-ja in 64–79% yields with 86–90% ee. Notably, halide substituents, F, Cl, and Br, as well as Bpin moiety remained intact after the reaction, thereby facilitating further modifications at their positions (e.g., products 6fa-ha and 6ja). A 1.0 mmol scale reaction of 1b also proceeded smoothly to give comparable results (6ba, 74% yield, 90% ee), demonstrating the scalability of this process. Moreover, the reactions with alkenes 1 k-o bearing common substituents such as methoxy, methyl, fluoro, and bromo at the meta- or ortho-positions also worked well; and the expected products 6 ka-oa were isolated in 61–81% yields with 89–90% ee. 2-Vinylnaphthalene 1p having a methoxy group and heterocycle-containing alkenes 1q-s all proved to be suitable coupling partners, leading to 6pa-sa with good yields and 86–93% ee. Remarkably, this protocol can also be successfully extended to biologically relevant molecule and pharmaceutical-derived styrene analogues (Fig. 2B). For instance, estrone, febuxostat-, and simple amino acid-derived alkenes 1t-v reacted well to give the desired acylcyanation products 6ta, 6ua, and 6va with good stereoselectivity, respectively. As a result, our protocol should be of potential use for late-stage structural modification of drug and complex compounds. Unfortunately, the current catalytic system is not applicable to simple unactivated or electron-deficient alkenes.
Then, we continued to evaluate the generality of this asymmetric three-component reaction by using a representative set of oxime esters, which can be easily prepared in two steps from the relevant ketone precursors. As shown in Fig. 3A, a range of aryl ketone-derived oxime esters 3b-g with electronically diverse functional groups (e.g., Me, OMe, tBu, F, Cl, or Br) at the para-position of the phenyl ring reacted well with 1b and TMSCN. And the expected alkene acylcyanation products 6bb-bg were obtained with good yields (70–86%) and excellent enantiomeric excess (83–92% ee). As shown in the cases of 3 h-k, the change of the substitution pattern and steric hindrance of their phenyl ring has deleterious effect on the reaction efficiency or enantioselectivity, with the corresponding products 6bh-bk being obtained with 67–85% yields and 89–92% ee. Single crystals of product 6bj were obtained, and the absolute stereochemistry was determined to be S by X-ray crystallographic analysis 64, and all other coupling products were tentatively assigned by analogy with 6bj. Remarkably, oxime esters 3 m-p derived from aliphatic ketone with various lengths of alkyl chains also reacted well with 2-vinylnaphthalene 1a and TMSCN (Fig. 3B). The relative products 6am-ap were isolated with 61–70% yields and 86–90% ee.
Encouraged by these results, we further attempted to extend the current dual photoredox and copper catalysis strategy to the asymmetric three-component vicinal DCF reaction of cycloketone-derived oxime esters, alkenes, and TMSCN (Fig. 4). Minor modification of the reaction conditions identified that a combination of organic photocatalyst Ph-PTZ (1.25 mol%) and Cu(CH3CN)4PF6 (0.5 mol%)/ligand L1 (0.6 mol%) enabled the desired reaction to proceed smoothly under irradiation of 2 × 3 W purple LEDs at room temperature63. This process also exhibited broad substrate scope and good functional tolerance with respect to both alkenes and oxime esters. As shown in Fig. 4A, a wide variety of commercially available styrenes containing neutral (2a), alkyl (2b-c), electron-withdrawing (2e-h), or aryl (2i-j) groups at the para-position of the phenyl ring could react well with oxime ester 4a. The corresponding dinitrile products 7aa-ja were obtained in 51–75% yields with 83–96% ee. The absolute stereochemistry of 7ja was also confirmed to be S-configuration by X-ray diffraction64. Again, as shown in the reactions of alkenes 2 k-n, variation of the substitution pattern and steric hindrance of the aromatic ring could be well tolerated, leading to formation of products 7 ka-na with 51–78% yields and 82–97% ee. Moreover, the reactions of 2-vinylnaphthalene 2o and substrates containing heterocycle-fused ring (2p) or heteroaryl groups (e.g., 2q, 2r) all proceeded well to give products 7oa-ra with moderate to good yields and excellent enantioselectivity (84–90% ee). Notably, styrenes (e.g., 2 s and 2t) derived from dihydroartemisinin and gibberellic acid could also participate in the reaction with good stereoselectivity, suggesting that the method can potentially be used in the late-stage modification of pharmaceutically relevant compounds.
Finally, we turned our attention to study the substrate scope of cyclobutanone oxime esters by reacting with styrene 2i and TMSCN (Fig. 4B). Both oxetan-3-one and 1-Cbz-3-azetidinone derived oxime esters 4b and 4c reacted well to afford 7ib and 7ic in moderate to good yields with excellent enantioselectivity (87–89% ee). Mono-substituted oxime ester 4d could participate in the reaction smoothly to deliver product 7id as a mixture of diastereomers with good yield and high enantioselectivity. Note that sterically demanding oxime esters 4e-g also proved to be compatible with the reaction, giving the expected products 7ie-ig in good yields with 85–92% ee.
Synthetic applications. To showcase the potential synthetic utility of this asymmetric method in the construction of valuable skeletons, we performed diverse further transformations with the chiral β-cyano ketones and alkyldinitriles (Fig. 5)65–66. For example, the cyano group of 6ba and 7ia could be easily converted to amide group by Pd-catalyzed hydrolysis using stoichiometric acetaldoxime in refluxing aqueous EtOH, giving the corresponding products 8 and 9 with good yields, respectively (Fig. 5a). Moreover, treatment of 7ia with NiCl2/NaBH4 and Boc2O in MeOH allowed efficient sequential reduction and protection of both cyano groups, with aliphatic chiral amine 10 being obtained with 66% yield and 90% ee (Fig. 5b). The synthesis of chiral ester 11 can also be achieved by treatment of 7ia with alcoholysis. Notably, no notable loss of optical purity was detected in these manipulations.
Mechanistic studies. To gain some insight into the mechanism, we carried out several control experiments by using the substrates 1b, 3a, and TMSCN (Fig. 6). The target three-component reaction was completely inhibited, when stoichiometric radical scavenger 2,2,6,6-tetramethyl-1-piperidinyloxy (TEMPO) was introduced (Fig. 6a). Instead, the relevant TEMPO-adduct 12 was obtained in 76% yield, suggesting the possible involvement of acyl radical 3a-I in this process. Moreover, the reaction of radical clock substrate 13 having a cyclopropyl moiety also proceeded smoothly to give ring-opening product 14 in 64% yield with good stereoselectivity (Fig. 6b). These results indicated the intermediacy of radical species 13-A and 13-B, as well as the radical property of the reaction. Similar control experimental results were also observed in the case of cycloketone oxime ester-based asymmetric three-component reaction63.
On the basis of these mechanistic studies and related literature19–21, 44-47, we proposed a dual photoredox and copper-catalyzed mechanism for the present asymmetric three-component reaction as depicted in Fig. 7. The reaction starts with SET reduction of redox-active oxime esters 3a and 4a by the excited state photocatalyst to give iminyl radicals 3a-I and 4a-I, with release of carboxylic anion (RCO2-). Then, 3a-I and 4a-I undergoes C-C bond β-cleavage to form acyl and cyanoalkyl radicals 3a-II and 4a-II. Further facile trap of these carbon radicals by styrene derivatives 1 and 2 forms relatively more stable benzylic radicals III. On the other hand, the initially formed carboxylic anion (RCO2-) could also facilitate the ligand exchange between L1/copper(I) complex and TMSCN to form L1Cu(I)CN species. Such L1Cu(I)CN complex can further be oxidized by the oxidizing photocatalyst (PC•+) via a SET process, and undergoes another ligand exchange with TMSCN to form L1Cu(II)(CN)2 complex, regenerating ground-state photocatalyst to close the photoredox catalysis cycle. Finally, L1Cu(II)(CN)2 traps the prochiral benzylic radical III to form a chiral high-valent Cu(III) complex IV, which undergoes reductive elimination to afford the coupled product 6 or 7, with regeneration of L1Cu(I)CN species to complete the copper catalysis cycle. It should be noted that an alternative process involving direct cyano transfer from the L1Cu(II)(CN)2 complex to the benzylic radical III through an outer-sphere pathway is also possible. Notably, the whole process is redox-neutral and does not need any external stoichiometric oxidants or reductants.