Introduction

Chiral carbon centers are present in a wide variety of functional molecules, including natural products, pharmaceuticals, and agrochemicals1. Extensive study of asymmetric hydrogenation reactions over the last several decades has led to the establishment of a number of remarkable methods involving homogeneous transition-metal catalysis, providing reliable access to chiral compounds from abundant unsaturated substrates2,3,4,5,6,7,8,9,10,11,12,13,14,15. To achieve high enantioselectivity, researchers have designed chiral catalysts that can distinguish between the two prochiral faces of substrates on the basis of differences between the substituents on the double bonds16,17,18,19,20,21,22,23,24,25,26,27,28,29. However, general methods for discriminating between the re and si faces of double bonds bearing two different aryl groups, which would provide compounds with chiral gem-diaryl motifs30,31,32,33,34,35,36,37,38,39,40,41, are still under development owing to the small difference in electronic properties between the two aromatic rings and to their similar planar structures.

Depending on the positions of the substituents on the aromatic rings of 1,1-diaryl unsaturated compounds such as 1,1-diarylethylenes and benzophenones, they can be divided into two categories. High enantioselectivities have been achieved for category a, compounds in which one of the aromatic rings bears an ortho substituent (Fig. 1a)42,43,44,45,46,47,48,49,50,51,52. For example, Wang et al. reported rhodium/Duanphos-catalyzed asymmetric hydrogenation of 1,1-diarylethylenes bearing a ortho-hydroxy directing group44. In addition, Song et al. accomplished elegant syntheses of enantiomerically enriched 1,1-diarylethanes by means of iridium-catalyzed hydrogenation reactions assisted by an ortho carboxylic acid group45. Sterically bulky substituents without a chelating atom can also facilitate catalyst differentiation of the prochiral faces of double bonds. For example, Mazuela et al. accomplished asymmetric hydrogenation of 1,1-diarylethylenes by using an iridium phosphite–oxazoline catalyst42, and Chen et al. used a chiral oxazoline iminopyridine–cobalt catalyst for enantioselective hydrogenation of 1,1-diarylethylenes48. Ohkuma et al.49 and Touge et al.52 found that ruthenium-catalyzed asymmetric hydrogenation of benzophenones also requires an ortho substituent for high enantioselectivity. In addition, owing to the less difference of steric hinderance, chiral induction in hydrogenation of unsaturated compounds bearing two ortho-substituted aromatic rings is challenging; a highly enantioselective catalytic method has yet to be reported.

Fig. 1: Background of metal-catalyzed asymmetric hydrogenation of 1,1-diarylethylenes and benzophenones.
figure 1

a Metal-catalyzed asymmetric hydrogenation of 1,1-diarylethylenes and benzophenones with ortho-substituent. b Metal-catalyzed asymmetric hydrogenation of 1,1-diarylethylenes and benzophenones without ortho-substituent.

In stark contrast, asymmetric hydrogenation of 1,1-diaryl-substituted double bonds without the assistance of a proximal substituent is fundamentally difficult, and enantioselectivities are generally low (Fig. 1b)53,54,55,56. Exceptionally, double bonds bearing highly electronically biased aromatic rings have been successfully hydrogenated with good enantioselectivities. For example, Bess et al. achieved asymmetric hydrogenation of 1,1-diarylalkenes bearing a 3,5-dimethoxylphenyl ring by using an iridium/phosphoramidite catalyst53, and Touge et al. observed high enantioselectivities in ruthenium/1,2-diphenyl-1,2-ethylenediamine-catalyzed transfer hydrogenation of benzophenones with electron-poor aromatic rings, such as a 3,5-dinitrophenyl ring52.

To provide a reliable method for asymmetric hydrogenation of 1,1-diaryl-substituted double bonds, we wondered a transition-metal unit would facilitate the differentiation of the two prochiral faces by ligating with one of the aromatic rings. Metal η6-benzene complexes, such as (benzene)Cr(CO)3, have different electronic properties than free benzene and have three-dimensional structures57,58,59,60,61,62,63,64,65,66. In 1995, Corey and co-workers conducted an elegant study of the electronic effects of remote substituents on aromatic rings during Corey-Bakshi-Shibata (CBS) reduction of benzophenones and found that Cr(CO)3-ligated p’-chlorobenzophenone synthesized from (benzene)Cr(CO)3 can be reduced to the corresponding alcohol with 94% ee (Fig. 2a)67,68,69,70. Given that the stability of a metal η6-arene complex varies with the ligated aromatic ring (for an example involving chromium, see Fig. 2b)71,72,73, we envisioned that combining selective arene coordination with a metal unit and transition-metal-catalyzed asymmetric hydrogenation would provide a reliable solution to the challenge posed by category b. Two questions needed to be answered: Could a metal unit distinguish and coordinate selectively to only one aromatic ring of a conjugated compound, and would the metal complexes be compatible with homogenous hydrogenation conditions74,75?

Fig. 2: Asymmetric reduction of chromium η6-arene complexes.
figure 2

a Synthesis and Corey-Bakshi-Shibata (CBS) reduction of Cr(CO)3 benzophenone complex. b Relative stability of Cr(CO)3 arene complex. c This work: synthesis of compounds bearing a chiral gem-diaryl motif from either 1,1-diarylethylenes or benzophenones by a sequence involving selective arene exchange and stereoselective asymmetric hydrogenation.

Herein, we report a relay strategy for synthesis of compounds bearing a chiral gem-diaryl motif from either 1,1-diarylethylenes or benzophenones by a sequence involving selective arene exchange with (naphthalene)Cr(CO)3 (2) and subsequent ruthenium-catalyzed asymmetric hydrogenation (Fig. 2c).

Results

Reaction optimization

To explore the feasibility of our strategy, we began by investigating arene exchange reactions between different metal species and 4-chlorobenzophenone (1a) as a model substrate (see Supplementary Fig. 1). To our delight, upon reaction of 1a with (naphthalene)Cr(CO)3 (2)76,77,78,79,80 in dioxane at 80 °C, arene exchange took place selectively on the unsubstituted benzene ring (1a-Cr: 1a-Cr’ > 10:1).

The arene exchange conditions showed great generality, allowing for ligation of an array of benzophenones with generally high regioselectivities (Fig. 3). Specifically, the selectivity for the two aromatic rings was in line with the trend in the relative stabilities of the Cr(CO)3–arene complexes shown in Fig. 2b. In addition, we found that the selectivity varied somewhat with temperature. For example, after a mixture of Cr-ligated 4-methoxylbenzophenone (1b) was heated at 140 °C for 3 h, the 1b-Cr/1b-Cr’ ratio increased to 4.4:1 from the 1.5:1 ratio that was observed under our kinetically controlled conditions, and a complex bearing one Cr(CO)3 units on each aromatic ring was detected (see Supplementary Fig. 3). We attributed this result to arene exchange between complexes under thermodynamic conditions, and metal migration along the conjugated system may also have taken place81,82,83.

Fig. 3: Selective formation of Cr(CO)3-benzophenone complex.
figure 3

Arene exchange reactions between (naphthalene)Cr(CO)3 (2) and an array of benzophenones. Reaction conditions, unless otherwise stated: benzophenones (0.10 mmol), 2 (0.15 mmol), dioxane (1.0 mL), N2, 80 °C, 24 h. a2 (0.11 mmol). Ratios were determined by 1H NMR spectroscopy.

Next, we investigated the asymmetric hydrogenation of Cr(CO)3-ligated 1a as a model substrate by using various chiral Ru catalysts (see Supplementary Fig. 4a). To our delight, the Ru(II)-NHC-diamine catalyst developed by Glorius group84 afforded the corresponding reduced complex with high enantioselectivity (93% ee) upon reaction with H2 gas (5 atm) in the presence of NaOtBu. The free alcohol could be readily released from the chromium with no obvious loss of enantiomeric purity by irradiation of the crude reaction mixture at 440 nm under air85. In contrast, when 1a was directly subjected to the same hydrogenation conditions, the enantioselectivity was much lower (43% ee) (see Supplementary Fig. 4b). Beside Ru-1, we also found that the spiro diphosphines supported ruthenium catalyst developed by Zhou group86 also provided alcohol 4a with high enantioselectivity (91% ee) (see Supplementary Fig. 4a).

Substrate scope

Having optimized the conditions for the selective arene exchange and the asymmetric hydrogenation, we evaluated the scope of the sequence by applying it to an array of benzophenones 1 (Fig. 4). Monosubstituted benzophenones were suitable substrates (4a–4j), providing the desired alcohols in moderate to good yields with high enantioselectivities, regardless of whether the substituent was in the para or meta position. In addition, substrates bearing bisfunctionalized aromatic rings (4k–4u) and polyfunctionalized aromatic rings (4v, 4w), were suitable and gave high enantioselectivities (up to 99% ee). Notably, an array of functional groups, such as carboxylate (4h), amide (4i), piperonyl (4k), and siloxy (4q), were well tolerated under the reaction conditions.

Fig. 4: Substrate scope of Ru-catalyzed hydrogenation reactions.
figure 4

Synthesis of compounds bearing a chiral gem-diaryl motif from either 1,1-diarylethylenes or benzophenones by a sequence involving selective arene exchange and stereoselective asymmetric hydrogenation. Reaction conditions, unless otherwise stated: (1) 1 (0.2 mmol), 2 (0.3 mmol), dioxane (2 mL), 80 °C, 24 h. (2) Ru-1 (3 mol%), NaOtBu (15 mol%), toluene (1.5 mL), H2 (5 atm), 0 °C, 36 h; then irradiation with 440 nm LEDs for 2 h in air at room temperature. a(1) 2 (0.22 mmol). b(2) Room temperature. c(2) Hexane (1.5 mL), H2 (50 atm). Isolated yields were reported. The absolute configurations of 4a, 4e, 4h, and 4m were S, as determined by comparison of their optical rotations with reported data.

Next, we evaluated the utility of the sequence for asymmetric hydrogenation of several 1,1-diarylethylenes (Fig. 4). By using hexane as the solvent for hydrogenation instead of toluene, we could obtain 1,1-diarylethanes in good yields with moderate to high enantioselectivities (up to 99% ee); both mono- (5a–5c) and disubstituted (5d) 1,1-diarylethylenes were suitable substrates.

Complexation of an unsymmetrically 1,2- or 1,3-disubstituted arene ring to a metal via η6-coordination results in a molecule lacking symmetry elements60. To determine whether planar chirality affected the stereochemical control of the asymmetric hydrogenation, we subjected a 1:1 mixture of both enantiomers of Cr(CO)3 complexes of 2-methyl-4’-chlorobenzophenone to the ruthenium-catalyzed hydrogenation conditions (Fig. 5a). This reaction yielded two pair of diastereomers with > 20:1 dr and 2.9:1 dr respectively. After removing the Cr(CO)3 unit, we obtained the free alcohol (4x) in overall 98% yield and 74% ee, implying that chiral catalyst Ru-1 dominated the differentiation of the two prochiral faces during the hydrogenation. Interestingly, shortening the reaction time to 15 minutes resulted in kinetic resolution of 1x-Cr87, providing alcohol 4x in 92% ee.

Fig. 5: Asymmetric hydrogenation of benzophenones 1x and 1b.
figure 5

a Synthesis of 4x by a sequence involving selective arene exchange and stereoselective asymmetric transfer hydrogenation. b Synthesis of 4b by a sequence involving selective arene exchange and stereoselective asymmetric transfer hydrogenation.

Our strategy was also applicable to substrates that formed chromium complexes less selectively (Fig. 5b). For example, reaction of 4-methoxybenzophenone (1b) with (naphthalene)Cr(CO)3 (2) gave a mixture of two isomers, 1b-Cr and 1b-Cr’, which could be separated. Each of the isomers was then converted to the same enantioenriched alcohol (4b) by reduction catalyzed by the catalyst Ru-1 or its enantiomer Ru-1’, respectively.

In addition, we tested the versatility of our relay strategy by replacing H2 gas with other easy-handle hydrogenation sources, and we found that asymmetric hydrogenation of Cr(CO)3-ligated 4-chlorobenzophenone catalyzed by a 1,2-diphenyl-1,2-ethylenediamine-derived ruthenium complex Ru-288,89,90 with sodium formate provided alcohol 4a with high enantioselectivity. Then we evaluated the substrate scope of the combination of arene exchange and asymmetric hydrogenation (Fig. 6). To our delight, an array of diaryl ketones were suitable for this alternative method, providing enantioenriched 1,1-diarylmethanols with HCO2Na as the hydrogen source instead of H2 gas. Notably, substrates with a vinyl group (4ac) were tolerated under the reaction conditions.

Fig. 6: Substrate scope of Ru-catalyzed transfer hydrogenation reactions.
figure 6

Synthesis of compounds bearing a chiral gem-diaryl motif from benzophenones by a sequence involving selective arene exchange and stereoselective asymmetric transfer hydrogenation. Reaction conditions, unless otherwise noted: (1) 1 (0.2 mmol), 2 (0.3 mmol), dioxane (2.0 mL), 80 °C, 24 h. (2) Ru-2 (7 mol%), NaCO2H (2 mmol), DMF (1.0 mL), H2O (0.1 mL), 35 °C, 36 h; then irradiation for 2 h with 440 nm LEDs. Isolated yields were reported. The absolute configurations of 4a, 4e, 4h and 4m were S, as determined by comparing their optical rotations with reported data.

Mechanistic studies

Next, we turned our attention to the origin of the high enantioselectivity achieved in the hydrogenation of Cr(CO)3-ligated substrates. The steric bulk of the Cr(CO)3 unit can be expected to have hindered rotation of the aromatic rings relative to the extent of rotation possible in the unbound precursor. Indeed, density functional theory calculations involving a relaxed scan of the dihedral angle formed by the carbonyl group and the aromatic ring connected to the carbonyl group were consistent with this expectation. The maximum barrier to rotation for each ring of 4-chlorobenzophenone (1a) was around 3 kcal/mol, and there was no obvious energetical difference between the para-Cl-substituted and unsubstituted rings (Fig. 7a). In contrast, the barrier to rotation of Cr(CO)3-ligated phenyl group of 1a-Cr was remarkably higher than that of the unbound phenyl ring, which may require a higher energy cost to adapt the benzophenone to the catalytic pocket (Fig. 7b).

Fig. 7: Energy diagram of a relaxed scan of the dihedral angle between the carbonyl group and the aromatic ring in 1a and 1a-Cr.
figure 7

The relative Gibbs free energies are given in kcal/mol. a Energy diagram of a relaxed scan of the dihedral angle between the carbonyl group and the aromatic ring in 1a. b Energy diagram of a relaxed scan of the dihedral angle between the carbonyl group and the aromatic ring in 1a-Cr.

To gain more insight into the mechanism of the stereoselectivity of the ruthenium-catalyzed hydrogenation, we performed additional density functional theory calculations. On the basis of previous studies52,84, we began by investigating the reaction of noncoordinated 4-chlorobenzophenone (Fig. 8, left part), and we found that the transition states for the different stereoselectivities were quite similar in structure; there was only a tiny energy difference (0.3 kcal/mol) between them, which is consistent with the observed absolute stereochemical outcome and the low enantioselectivity (43% ee). Moreover, similar weak interactions (the green area, see IGMH analysis) between the aromatic substituents of the catalyst and the 4-chlorobenzophenone ring in transition states TS-1 and TS-2 lead to quite close in energy.

Fig. 8: Computational studies of transition states for asymmetric hydrogenation of 1a and 1a-Cr.
figure 8

The relative Gibbs free energies are given in kcal/mol. Left part, computational studies of transition states for asymmetric hydrogenation of 1a. Right part, computational studies of transition states for asymmetric hydrogenation of 1a-Cr.

In contrast, calculations on the asymmetric hydrogenation of Cr(CO)3-ligated 4-chlorobenzophenone revealed relatively lower energy barriers to the formation of both enantiomeric products, which can probably be attributed to the electron-withdrawing effect of the Cr(CO)3 unit (for atomic charge population and conceptual density functional analysis see Supplementary Tables 2 and 3). Furthermore, transition state TS-4, formed by reduction from the re face, was favored by 2.1 kcal/mol over TS-3, formed by reduction from the si face (Fig. 8, right part). Specifically, in TS-3, the chromium-ligated benzene ring was located in a rigid pocket formed by a phenyl group of the diamine ligand and a naphthyl group of the carbene ligand, leading not only to greater steric repulsion (for activation strain model, see Supplementary Table 2), but also to a weaker π–π stacking interaction between the phenyl ring of the diamine ligand and the chromium-ligated benzene ring (the corresponding green area in TS-3 was smaller than that in TS-4, see IGMH analysis) because rotation of the latter was hindered, as indicated by the clearly smaller | O1-C1C2-C3 | angle (27.7°) relative to other transition states TS-1 (39.2°), TS-2 (39.1°), and TS-4 (41.9°).

In conclusion, we have developed a relay strategy for accessing compounds with a chiral carbon center bearing two aromatic substituents by means of a combination of selective arene exchange with (naphthalene)Cr(CO)3 and subsequent enantioselective hydrogenation mediated by a chiral ruthenium catalyst. Density functional theory calculations revealed that the chromium markedly affected the energy of the transition state for addition of the ruthenium hydride species to the carbonyl group. Work aimed at extending this strategy to related transformations is currently underway in our laboratory.

Methods

General procedure for synthesis of Cr complexes

In a glovebox, a solution of Cr(CO)3(naphthalene) 2 (0.3 mmol, 1.5 equiv.), 1 or 3 (0.2 mmol, 1.0 equiv.) in 1,4-dioxane (2 mL) in a 4 ml glass vial was stirred at room temperature for 2 min, then the sealed reaction vial was taken out of the glovebox and the reaction mixture was stirred at 80 °C in the dark for 24 h. After cooling to room temperature, the solution was evaporated under reduced pressure. The residue was purified by column chromatography eluting with Ethyl acetate/Petroleum ether, affording complex 1-Cr or 3-Cr.

General procedure for asymmetric hydrogenation

(Figure 4) To a 5 ml glass tube, Ru-1 (3 mol%), NaOtBu (15 mol%), 1-Cr or 3-Cr (1.0 equiv.), and toluene or hexane (0.10 M) were added in a glovebox. The glass tube was taken out of the glovebox and placed in a pre-cooled (0 °C) stainless steel autoclave, then the autoclave and pressurized and depressurized with hydrogen gas five times before the indicated pressure (5 atm or 50 atm) was set. The reaction mixture was stirred at 0 °C for 36 h. After the autoclave was carefully depressurized, the tube was irradiated under 440 nm LEDs (20 W) for 2 hours at room temperature in air atmosphere. Then the product was purified by flash column chromatography on silica gel. The enantiomeric excess was determined by SFC or HPLC analysis using chiral column. (Fig. 6) To a 4 ml glass vial, Ru-2 (7 mol%), HCO2Na (10 equiv.), 1-Cr (1.0 equiv.), and DMF/H2O (10:1, 0.18 M) were added. The reaction mixture was stirred at 35 °C for 36 h. Then the vial was irradiated under 440 nm LEDs (20 W) for 2 hours at room temperature in air atmosphere. Then the product was purified by flash column chromatography on silica gel. The enantiomeric excess was determined by SFC or HPLC analysis using chiral column.