Reaction development. α,β-Unsaturated carbonyl compounds are one class of common Michael acceptors in organic synthesis. Initially, we chose cyclohexane (1a) as the model substrate, α,β-unsaturated carbonyl compound bearing an N-acylpyrazole moiety (2a) as the reaction partner, 5,7,12,14-pentacenetetrone (PC1) as the HAT photocatalyst,38,52 and the bisoxazoline nickel complex generated in situ ([L1-Ni]) as the chiral catalyst (Table 1). The reaction of 1a and 2a failed to produce an addition product under visible light conditions (entry 1). We next tested several SO2 surrogates in the photochemical system, and found that DABCO∙(SO2)2 was the appropriate reaction component (entries 2,3). Under irradiation with a 24 W blue LEDs lamp (λmax = 455 nm) at 20°C under argon, the reaction of 1a, 2a and DABCO∙(SO2)2 in dichloroethane delivered the chiral sulfone product (3a) in 58% conversion and with only 3% ee (entry 3). α,β-Unsaturated carbonyl substrates bearing a different auxiliary group (Z) were examined, and it was found that N-acylpyrazole (2a) was a suitable substrate (entries 4–7). Other photocatalysts (PC2–PC4) failed to catalyze this transformation (entries 8–10). In order to improve the reaction efficiency and enantioselectivity, a range of chiral ligands (L2–L11) were screened (entries 11–20). Chiral bis(oxazoline) ligands (L2–L4) and a tridentate ligand (L5), which have been successfully used for asymmetric induction in some radical-mediated transformations,53−56 were found to be ineffective in this photocatalytic reaction (entries 11–14), but replacement of L1 with an indane-derived BOX ligand (L6) led to a full conversion and significantly increased enantiomeric excess (82% ee) (entry 15). Based on this observation, we modified this type of ligands and synthesized several sterically more bulky analogs for creation of more precise chiral environment (L7–L9). One of these, L7 was identified as the best ligand with regard to the yield and enantioselectivity (86% ee) (entry 16). Finally, the reaction was further improved by reducing the reaction temperature and increasing the loading of chiral catalyst, which provided 3a with a full conversion and 95% ee (entry 23). The reaction with only 1.0 equiv. of cyclohexane also proceeded smoothly at a lower but reasonable reaction rate (52% conversion within 96 h) and the similar enantioselectivity of 93% ee (entry 24). Such conditions would be useful with more expensive C(sp3)−H precursors such as natural products or drug molecules.
Reaction scope investigation. With the optimal reaction conditions in hand, we investigated the substrate scope of this photocatalytic asymmetric three-component sulfonylation reaction (Fig. 2). It was revealed that unsubstituted cycloalkanes such as cyclohexane (1a), cyclopentane (1b), cycloheptane (1c), cyclooctane (1d) and cyclododecane (1e) were excellent substrates, delivering the chiral sulfone products (3a–3e) in 62–74% yield and with 91–95% ee. A trisubstituted cycloalkane, 1,3,5-trimethylcyclohexane (1f) exhibited high regioselectivity towards the tertiary C(sp3)−H bonds (> 50:1 rr) and provided lower enantioselectivity (78% ee). The observation of high regioselectivity was consistent with an inherent difference of bond dissociation energies (BDEs) of secondary and tertiary C(sp3)−H bonds as well as stability of the corresponding carbon radicals. The crystal structure of product 3b revealed the absolute configuration (R) of the major enantiomer, and those of the other products were accordingly assigned as R by analogy. Adamantane has a significantly higher 3° C(sp3)−H bond dissociation energy of 99 kcal∙mol-1 caused by its rigid cage structure, which exceeds the bond dissociation energy of its 2° C(sp3)−H bonds (96 kcal∙mol-1) and those of most other hydrocarbons.57 The unusually strong tertiary C(sp3)−H bonds of adamantine-type compounds present a remarkable challenge to their selective tertiary C−H functionalization,58 but such transformations proceeded very well in our photochemical system. Adamantane (1g), 1-methyladamantane (1h), 1-ethyladamantane (1i) and 1,3-dimethyladamantane (1j) were all selectively functionalized at the tertiary C(sp3)−H position, affording the products (3g–3j) as single regioisomers with 84–90% ee. These results indicate that the catalyst-control also plays a crucial role in the selectivity in addition to the inherent difference of bond dissociation energies and carbon radical stabilities. The reaction exhibited high functional group compatibility, which would be desirable for its synthetic application to structurally diverse compounds. For example, adamantine derivatives bearing a fluoro- (1k), chloro- (1l), bromo- (1m), cyano- (1n), ester- (1o), keto- (1p), or hydroxyl- (1q) substituent all provided a reasonable yield (59–71%), perfect regioselectivity (> 50:1), and satisfactory enantioselectivity (82–91%). Moreover, the photochemical reactions of tertiary alkanes also proceeded smoothly under the standard conditions and delivered the desired chiral sulfones (3r–3u) as exclusive regioisomers with 75–82% ee.
To further investigate the generality of this method, other C(sp3)−H precursors were examined. The reaction with toluene and its derivatives were more rapid than those of cycloalkanes and alkanes. The primary (1v–1z), secondary (1za–1ze) and tertiary benzylic hydrocarbons (1zf–1zh) all worked well and delivered the chiral products (3v–3zh) in 58–75% yield and with 62–94% ee. Sterically more demanding substrates such as diphenylmethane (1zc) and cyclohexylbenzene (1zg) tended to give lower enantioselectivity. The reactions of toluene derivatives containing completing reactive sites displayed some degree of regioselectivity. For example, p-isopropyl toluene (1zh) provided a very moderate regioselectivity of 1.5:1, while 1-(p-tolyl)adamantine bearing benzylic C(sp3)−H bonds, secondary and tertiary C(sp3)−H bonds in the admantyl moiety (1zi) afforded a single regioisomer with the benzylic functionalization. It was found that heteroaromatic α-C(sp3)−H precursors such as 3-methylthiophene (1zj), 2-methylthiophene (1zk), 3-methylbenzothiophene (1zl) and 3-methylbenzofuran (1zm) were compatible with the reaction and gave products with 76–88% ee. Interestingly, ethers such as tetrahydrofuran (1zn) and methyl tert-butyl ether (1zo), which have strong α-C(sp3)−H bonds (BDEs = ~ 92 kcal∙mol-1),59 were also identified as excellent substrates by the yields (55–73%) and enantioselectivities (80–92% ee).
We next evaluated the scope of α,β-unsaturated N-acylpyrazoles. As summarized in Fig. 3, β-substituents including a linear alkyl group (products 3zp–3zr), a branched alkyl group (products 3zs, 3zt), a cylcoalkyl group (product 3zu), and an aryl substituent (products 3zv–3zy) were all compatible with regards to yields (53–71%) and enantioselectivity (64–93% ee). Typically, β-alkyl substituted substrates gave better enantioselectivity in comparison to those containing a β-aryl group, perhaps due to the trend of Z/E isomerization of the latter under photochemical conditions.65 For example, in the reaction of 1a+DABCO∙(SO2)2+2l®3zv, 28% of Z-isomer of 2l was isolated (see more details in Supplementary Information). Such byproducts were not observed in the conversions of β-alkyl α,β-unsaturated N-acylpyrazoles.
Mechanistic studies. Several control experiments were conducted to gain insight into the reaction mechanism (Fig. 4). For example, addition of a radical quencher (2,2,6,6-tetramethylpiperidine-1-oxyl, TEMPO, 3 equiv.) to the photochemical reaction 1v+DABCO∙(SO2)2+2a®3v was found to completely inhibit the transformation to 3v, instead affording a TEMPO-carbon radical cross-coupling product (4) detected by HRMS analysis. The three-component sulfonylation reaction of 4-(cyclopropylmethyl)-1,1’-biphenyl (1zp) under the standard conditions provided a ring-opened product (5) in 63% yield and with 95% ee, which further confirmed the reaction pathway via benzylic radicals. Employing cyclopropyl-substituted N-acylpyrazole (2p) as a substrate for the radical clock experiment provided product 6 in 65% yield and did not give any ring-opened product, thus excluding mechanisms involving the β-carbon radicals of N-acylpyrazole complexes.60 The reaction of 2a with sodium cyclopentanesulfinate was examined under the standard conditions, and failed to produce any desired product 3b (Fig. 4b). This result indicated that the reaction might not proceed through sulfonyl anion intermediates. Moreover, removal of the nickel catalyst in the reaction of 1v+DABCO∙(SO2)2+2a®3v led to the failure of product formation, while replacing the nickel catalyst by other Lewis acids such as a copper, zinc, iron or cobalt complex of the same chiral ligand (L7) still afforded product 3v in a moderate yield (17−31%). The cobalt complex with similar octahedral configuration even gave good enantioselectivity of 80% (Fig. 4c). These results suggested that the nickel complex most likely only provided Lewis acid activation for α,β-unsaturated N-acylpyrazoles. Finally, luminescence quenching experiments quenching experiments revealed that the C(sp3)-H precursors such as toluene (1v) was capable of quenching the excited state of PC1 and initiating the radical process (Fig. 4d).
Mechanistic proposal. On the basis of the initial experiments and mechanistic studies, we propose a plausible reaction mechanism (Fig. 5a). The chiral nickel catalyst ([L*-Ni]) undergoes fast ligand exchange with the α,β-unsaturated N-acylpyrazole (2) to generate an intermediate complex (A). On the other hand, the organophotocatalyst (PC) is excited to its triplet state (B), which performs hydrogen atom abstraction from the C(sp3)–H precursor (1) to give the semiquinone-type radical intermediate (C) and the transient carbon radical (D).52 The radical (D) is rapidly trapped by the sulfur dioxide released in situ to produce the stabilized sulfonyl radical (E).41 Owing to the electronic and steric effects, E reacts with the metal-coordinated Michael acceptor (A) through an outer-sphere rather than an inner-sphere pathway, affording the radical complex (F).61−66 Such an outer-sphere attack might be critical to avoid side reactions such as self-coupling or elimination and to achieve a high level of asymmetric induction in the photochemical reaction.67−80 Subsequent single electron transfer and proton transfer among intermediates C, F and small amount of water in the solution lead to formation of the neutral complex (G) and the organophotocatalyst (PC). Ultimately, ligand exchange between intermediate G and the substrate (2) gives the chiral sulfone product (3) and regenerates the coordinated α,β-unsaturated N-acylpyrazole (A).
A crystal structure of nickel complex [L6-Ni] exhibits an octahedral geometry, in which the six coordination sites are occupied by one chiral ligand and four water molecules (Fig. 5b, left). Accordingly, an intermediate [L7-Ni-2a] is simulated by Gaussian 09, and a proposed transition state is modeled by CYLview 1.0 (Fig. 5b, middle).81 The sulfonyl radical (E) interacts with the C=C double bond of the coordinated α,β-unsaturated N-acylpyrazole from Re-face with less steric hindrance, that is consistent with an observed R-configuration in product 3b (Fig. 5b, right). The modeled transition state structure also illustrated that the extended phenyl substituents on the chiral ligand (L7) were critical for a high level of asymmetric induction.
Synthetic utility. A mmol-scale reaction was performed to demonstrate the synthetic utility of the reaction. A mixture of toluene (1v, 533 μL, 5.0 mmol), 2a (164 mg, 1.0 mmol) and DABCO∙(SO2)2 (180 mg, 0.75 mmol) was irradiated with a blue LEDs lamp under the standard conditions (Fig. 6a), leading to the production of 227 mg of 3v (71% yield, 91% ee). The yield and enantioselectivity were basically the same as in the small-scale reaction. Next, we investigated further transformations of the reaction product. For example, an alcohol derivative (7) was obtained in 93% yield and with 91% ee by treatment of the chiral sulfone (3v, 91% ee) with NaBH4 in a mixed solvent of THF and H2O at 0 °C to room temperature. 3v could also be converted into the corresponding ester (8) or amide (9) in good yield and with retention of enantiomeric excess by substitution of the pyrazole moiety with an ethoxy or an amino group, respectively (Fig. 6b). Finally, the photochemical reaction was applied to the late-stage modification of bioactive molecules (Fig. 6c). Under the standard conditions, the reaction of a lopid derivative afforded its chiral sulfone derivative (10) as a single regioisomer in 64% yield and with 86% ee. Such precise recognition of two very similar benzylic C(sp3)–H bonds further confirmed the powerful catalyst-control of selectivity in the photochemical reaction. Using the similar protocol, celestolide and a synthetic intermediate of differin could be converted to the corresponding sulfone products 11 and 12 with excellent regioselectivity and high enantioselecitivity, respectively.