Evaluation of reaction conditions. On the basis of our previously reported Cu-catalyzed anti-Markovnikov hydrosilylation of alkynes,61 the investigation was initiated with using 4-ethynyl-1,1'-biphenyl 1a and (2,6-dimethoxyphenyl)(methyl)silane 2a (Table 1). The reaction was conducted at 50 oC for 15 h in THF with Cu(CH3CN)4PF6 as catalyst and (S)-BINAP as ligand. The desired product 3aa was obtained in 56% yield with 92:8 er (entry 1). Inspired by this preliminary result, we then tested several Cu catalysts such as Cu(CH3CN)4BF4, Cu(CH3CN)4OTf, CuI and Cu(OAc)2 to promote this reaction. Unfortunately, the reaction catalyzed by other copper catalysts did not generate the desired product. Then, other commercially available bisphosphine ligands were screened (entries 6-12). The reaction proceeded smoothly with (S)-Tol-BINAP (L2) as ligand and afforded the desired product with 94:6 er. But a bulkier ligand (S)-Xyl-BINAP (L3) was examined and provided insufficient yield and enantioselectivity. A higher enantioselectivity (95:5 to 99:1 er) could be observed in the presence of (R)-DTBM-SEGPHOS (L5), (R)-tBu-MeOBIPHEP (L6) or (R,R)-Ph-BPE (L7) as ligand, but the bulky and electron-rich bisphosphine ligands cannot maintain high yields (4-24%) of 3aa. Furthermore, various solvents were investigated for this hydrosilylation reaction catalyzed by Cu(CH3CN)4PF6. Replacement of THF with 2-MeTHF could increase the yield (72%), whereas a slightly decreased enantiomeric ratio (91:9) was obtained (entry 13). When this reaction was conducted in PhOMe, toluene, or m-xylene, the desired product was obtained in low yield but with good er values (entries 14-16). Thus, the mixed solvents were tried, and the mixture of THF and m-xylene [1:4 (v/v)] furnished the product 3aa in 52% yield with 95:5 er (entry 17). When THF was replaced by 2-MeTHF as solvent in mixed system, the desired product was formed in 60% yield with 95:5 er (entry 18). Additionally, when a complex Cu(CH3CN)PF6·L2 as precatalyst instead of Cu(CH3CN)4PF6 and L2 was used to ensure efficient generation of the catalytic species and a precise ratio of ligand and copper, the reaction gave 3aa in 66% yield without loss of enantioselectivity. Moreover, decreasing the catalyst loading had no obvious change on the yield or enantioselectivity. The absolute configuration of (S)-3aa was determined by X-ray diffraction analysis (CCDC: 2183555).
Scope of the reaction. After establishing the optimized reaction conditions, we turned to evaluate the substrate scope and limitation of this Cu-catalyzed asymmetric hydrosilylation of alkyne (Scheme 2). A broad range of aryl substituted alkynes (1a-z, 1a′-j′) were applicable under the reaction condition with moderate to good yields and enantioselectivities. Simple phenylacetylene was converted to styrenylhydrosilane (3ba) in 66% yield with 94:6 er. Different alkyl substituents installed on phenyl ring did not affect the reaction efficiency and enantioselectivities, affording the chiral organosilane products in 56-77% yield with 93:7-95:5 er (3ca–3fa). Other electron-donating groups such as –OMe (1g), –SMe (1h), –OBn (1i) and –N(Ph)2 (1j) on phenyl ring were all well tolerated to provide the corresponding products in 54-79% yields with 93:7-94:6 er. In addition, the substrates with electron-withdrawing substituents were also investigated. It was found that the substrate with 4-fluoro, 2-chloro, 3-chloro or 4-bromo halide group at phenyl ring still worked well with dihydrosilane (2a) and furnished the desired products (3ka-3na). When the 4-CO2Me-substituted phenylacetylene was applied in this reaction, the product (3oa) was delivered in 50% yield with a slightly diminished enantioselectivity. Next, other aryl rings
such as thienyl, naphthyl, and 9H-fluorenyl substituted terminal alkynes were tested to react with dihydrosilane 2a, all took place smoothly to afford the corresponding products (3pa, 3qa, and 3ra) in good results, respectively. Interestingly, excellent chemoselectivity was observed when the internal alkynyl and terminal alkenyl group substituted phenylacetylenes were subjected to this transformation, the products 3sa and 3ta were formed with good enantioselectivities. Moreover, reaction of the conjugated enynes 1u and 1v with 2a under the optimized reaction conditions afforded the linear conjugated dienylsilane products (3ua, 3va) in 55% and 71% yield with 93:7 and 92:8 er, respectively. Other aliphatic groups substituted alkynes displaying various functional substituents, such as carbonyl (1w), ester (1x and 1y), chloride (1z), and ether (1a′, 1b′) were all well tolerated, providing the desired products (3wa-3b′a) with good to excellent yields and enantioselectivities. It was noted that pure aliphatic chain substituted alkynes (1c′ and 1d′) were also suitable substrates and converted to the desired products in 68% and 65% yields with the same 92:8 er.
To demonstrate the synthetic potential and diversity of this methodology, we applied the reaction to late-stage functionalization of natural products, bioactive molecules, pharmaceutical and material building blocks, such as β-estradiol (3e′a), (-)-borneol (3f′a), geraniol (3g′a), estrone (3h′a), D-menthol (3i′a), and cholestanol (3j′a). An array of chiral Si-stereogenic alkenylhydrosilanes were obtained in moderate to good yields with excellent enantioselectivities, regardless of existing diverse functional groups or complex molecular structures.
Furthermore, to investigate the possible roles of the 2,6-dimethoxy functional group in substrates 2 in controlling the enantioselectivity and/or yield, a series of dihydrosilanes 2b–2o were synthesized and subjected to this reaction. Simple methylphenylsilane (2b) and n-pentylphenylsilane (2c) generally led to products with eroded enantioselectivities but maintained moderate yields. More bulky substituents on aryl ring of the silane could slightly improve the enantioselectivities (2d–2j). Notably, mesityl(methyl)silane (2h) led to an increased enantioselectivity of product 3ah (91:9 er), albeit with a decreased yield. A series of silanes with electron-donating (2k–2n) and electron-withdrawing (2o) groups were explored to examine the electronic effect. The results revealed that dimethoxy functional group in dihydrosilane 2a not only has steric effect to control the enantioselectivities and yields, but also has an electron-donating effect favor this transformation.
Synthetic transformations of compound 3a′a. Finally, the reaction was performed on a gram scale, which gave alkenylhydrosilane 3a′a in 73% yield with 93:7 er. To further demonstrated the utility of Si-stereogenic vinylhydrosilanes, several transformations of 3a′a were performed. Oxidation of the Si-stereogenic monohydrosilane 3a′a with m-CPBA provided the corresponding chiral silanol 4 in 75% yield with a decreased enantioselectivity.62 Then, hydrosilylation of 3a′a could smoothly convert it to 5 in 68% yield without loss of enantiopurity in the presence of [Rh(cod)Cl]2.39 In addition, hydrogenation of 5 successfully delivered silane 6 in a good yield with 93:7 er via a hydroboration/protodeboronation strategy.63
Computational studies of enantioselectivity controlled. To elucidate the origin of the enantiocontrol, we resorted to DFT calculation of the only step involving generation of the stereocenter, i.e., the metathesis between the incipient styrenyl Cu(I) and the silane. We systematically sampled the conformational space by including all permutations of the major flexible elements in the interaction of alkenyl Cu(I) with silane, namely, (i) the relative position of silane and styrenyl moiety around the tetrahedral Cu(I) center (denoted as 1 or 2 respectively), (ii) the s-cis or s-trans orientation of the styrenyl moiety relative to the C‒P bond projecting toward the silane (denoted as a or b respectively), (iii) orientation of the silane phenyl opposite to or toward the styrenyl Cu(I) (denoted as c or d respectively), and (iv) engagement of the pro-(R) or pro-(S) silane hydride (denoted as S or R respectively) (Figure S1). Such consideration leads to altogether 16 transition structures whose geometries were located and their Gibbs free energies determined using the double hybrid functional PWPB95,64 with Grimme’s DFT-D3(BJ) empirical dispersion65 and the SMD implicit solvation model66 (Figure 1a). The calculations suggest that the conformers having the styrenyl moiety at one side (Set 1) generally favor the observed (S)-configuration, while the structures in the other set (Set 2) all favor the minor (R)-antipode. Importantly, the former set of conformers includes several members that are appreciably lower in energy and are thus dominating in population. Quantitatively, an 85:15 er is theoretically predicted (Table S5), agreeing reasonably well with the experimental observation. For a comparison, calculations with three typical hybrid functionals M0667, PBE068 and ω-B97XD69 with DFT-D3 empirical dispersion all agree with the five low-lying conformers within 2 kcal/mol range accountable for the enantiocontrol, while they differ in the order of the relative energy of a couple of them (Figure S3). According to the PWPB95-D3(BJ) functional, 1ad_S (62%), 1ac_S (17%) and 1bd_R (7%) are the most dominant. Inspection of the geometries of the five lowest-lying transition structures (Figure 1B) shows no obvious correlation between the key bond length parameters with the relative stability of the structures, so distortion interaction analyses70 were performed to probe the origin of the energy difference (Figure 1B). First, comparison of the most dominant TS, 1ad_S and its “epimer” 1ad_R suggest that the major factor contributing to the energy difference is the appreciably larger distortion energy of the silane fragment in the latter, likely rising from shorter C‒Si bond (2.19Å vs 2.27Å) and slightly longer Si‒H bond (1.61Å vs 1.60Å). Interaction region indicator (IRI) analysis71 (processed by Multiwfn 3.8 dev.72) revealed the appreciable differences in non-covalent interaction regions: the former features a larger and stronger van der Waals (vdW) attraction region between the Si‒Me moiety and one of the P-Ar of the ligand as compared to the latter (Scheme 1C, red circled part in i vs ii), which likely helps increase the interaction energy, while reduce the strain of the fragments. Besides this dominant pair, 1ac_S, having the phenyl moiety of the silane being stacked with the P-Ar of the other side of the styrenyl moiety, also makes non-negligible contribution (17%) to the formation of the major enantiomer. In contrast, its “epimer”, 1ac_R, is 4.33 kcal/mol higher in energy and makes almost no contribution, due also to the much larger distortion energy of the silane moiety (Si-H bond length: 1.79 Å vs 1.73 Å in 1ac_S) being not sufficiently counterbalanced by the interaction energy. The opposite is observed for the least-contributing pair (1bd_S vs 1bd_R). Here 1bd_R is slightly lower in energy (0.26 kcal/mol) despite of its larger distortion energies from both fragments, as this effect is offset by the larger interaction energy, resulting in a later yet lower-energy transition state. For the latter two pairs, differences in the interaction region diagrams are non-obvious (Figure S4).