Dirhodium(II)/Phosphine Catalyst with Chiral Environment at Bridging Site and Its Application in Enantioselective Atropisomer Synthesis

A dirhodium(II)/phosphine catalyst with a chiral environment at the bridging site was developed for the asymmetric arylation of phenanthrene-9,10-diones with arylboronic acids. In contrast to the classic chiral bridging carboxylic acid (or derivatives) ligand strategy of bimetallic dirhodium(II) catalysis, in this reaction, tuning both axial and bridging ligands realized the first Rh2(OAc)4/phosphine-catalyzed highly enantioselective carbonyl addition reaction. The kinetic analysis reveals that dirhodium(II) and arylboronic acid follow the first-order kinetics, while phenanthrene-9,10-dione is zeroth-order. These data supported the proposed catalytic cycle, where the key intermediate in the rate-determining step involved the dirhodium(II) complex and arylboronic acid. Finally, axially chiral biaryls were prepared based on a newly developed oxidative ring-opening reaction of α-hydroxyl ketones with a base and molecular oxygen, which featured a central-to-axial chirality transfer radical β-scission step.


■ INTRODUCTION
Dirhodium(II) complexes, such as Rh 2 (OAc) 4 or its carboxylic analogues, display unique reactivities, which rely on the unique bimetallic structure, Rh(II)−Rh(II), and each has octahedral molecular geometry. The chiral bridging ligands, which connect two rhodium atoms in the coordination complex, well tune the properties of the complex to realize stereocontrol. However, these reactions mainly locate on the carbenoid or nitrenoid involved transformations ( Figure 1a). 1−7 Besides the bridging ligands, it was also found that fine-tuning of axial ligand(s) of dirhodium(II) complexes can explore new reactivities. In 2001, Furstner and Krause first proved that IMes, the axial ligand of the dirhodium(II) complex, enabled the dirhodium(II) complexes to show new catalytic activity: the arylations of aldehydes with boronic acids. 8 In the next two decades, the strategy of adjusting the axial ligands was proven to be a valuable way to tune the reactivities of dirhodium(II) complexes, which boosted the applications of these complexes in organic synthesis. 9−16 Despite the great successes in asymmetric C−H functionalization and cyclopropanation of carbenoids by using bimetallic dirhodium(II) catalysts with chiral bridging ligands, the asymmetric synthesis by employing achiral bridging ligands and a chiral axial ligand is very challenging. In Bolm, Ma, and Li's pioneering works, there was no example of excellent stereoinduction catalyzed by dirhodium(II) complex with chiral axial ligand as the only chiral source. 17−20 Over the last 20 years, the enantiomeric excesses of all the reported arylation reactions of carbonyls with a combination of achiral dirhodium(II) and chiral axial ligand only ranged from ∼20−52%. These observations were possibly due to the chiral environment being far from the reactive center: the chiral ligand is at one axial position of dirhodium(II) complexes, and the reactive center is at the opposite axial site (Figure 1b). In contrast to the well-established Rh(I)-catalyzed addition of arylmetal reagents to aldehydes, there are limited examples of Rh-catalyzed highly enantioselective carbonyl addition reactions, which directly give optically active alcohols, including sterically bulky tertiary alcohols. 21−28 These reactions still suffer from the narrow substrate scope, particularly the lack of methods for high stereoselective addition to ketones. Herein we report the development of a dirhodium(II)/Josiphos catalyst, which bears an axial phosphine ligand to tune the reactivity and bridging ligand to create a chiral environment (Figure 1c). The new catalyst has been applied in asymmetric arylation of diketones. Moreover, these obtained optically active α-hydroxyl ketones could be efficiently converted to axially chiral biaryls by treatment with a base and molecular oxygen.
■ RESULTS AND DISCUSSION Reaction Conditions Optimization. 4,5-Dimethylphenanthrene-9,10-dione 1a and phenylboronic acid 2a have been selected as model substrates, and the initial trials focused on Rh 2 (OAc) 4 as the catalyst with various chiral phosphine ligands (Table 1). Notably, without the assistance of phosphine ligand the reaction is very sluggish, and only a trace amount of product was observed (entry 1). A survey of diverse chiral phosphines indicated that these ligands not only induce stereochemistry but also dramatically improve the reactivity, which is beneficial for catalytic asymmetric reactions by minimizing the background reaction. The Josiphos (L1− L6) resulted promising stereoselectivity, where L6 gave the best outcomes regarding both the yield and ee value (entries 2−7). Furthermore, the reaction is fairly robust and can be scaled up to 1.0 mmol without deteriorating either the yield or stereoselectivity (entry 8). With K 3 PO 4 , or even in the absence of a base, the reaction still worked; however, the yields of 3a decreased and the same enantiomeric excesses were obtained (entries 9 and 10). For comparison, several rhodium(I) catalysts were tested, all trials gave reduced yield or stereoselectivity, showing the superiority of Rh 2 (OAc) 4 in this asymmetric arylation reaction (entries 11−13).
Reaction Scope and Selectivity. With the validity of our concept established, the dirhodium(II)/Josiphos-catalyzed asymmetric arylation was applied to different substituted phenanthrene-9,10-diones and boronic acids (Scheme 1). The substituent effect at the para and meta positions of phenylboronic acids was investigated. It can be alkyl, halogen atom, methoxy, trifluoromethoxy, as well as vinyl group (3b−3q). The reactions with phenylboronic acids bearing multisubstituents also proceed uneventfully (3r−3v). The stereoselectivity slightly decreased when the arylboronic acids contain electron-donating group, i.e., methoxy, methylthio, and [1,3]-dioxol (3g, 3o, 3p, and 3t). Both the reactivity and selectivity are sensitive to ortho substituted phenylboronic acids: compound 3w was obtained in 78% yield with 75% ee. Introduction of dimethyl groups to 3,6-positions of phenanthrene-9,10-diones slightly decreased the stereoinduction (3x), while the reactions of 2,4,5,7-tetramethyl or 4,5-dichloro phenanthrene-9,10-diones worked as efficiently as the model compound (3y and 3z). The binaphthyl analogue resulted in a decreased yield, while the enantiomeric excess still reached 94% [(S)-3aa (CCDC 2095229)]. The arylation of nonsymmetric phenanthrene-9,10-diones by replacing one methyl group with either an isopropyl or chloro group gave poor regioselectivity (3bb and 3cc). For heteroaromatic boronic acids, the thiophene and carbazole derivatives can be obtained in moderate to excellent yields, with the ee values still maintaining at high levels (3dd and 3ff). However, 3furanylboronic acid resulted in a dramatic decrease of reactivity, and the enantioselectivity also drops significantly (3ee). 3-Pyridinylboronic acid is not a compatible substrate,  which is possibly due to the coordinative ability of the pyridine unit (3gg). Atropisomerism is one of the key elements of chirality, and biaryl atropisomers are widespread in natural products and pharmaceuticals. 29−37 Herein, we anticipate preparing axially chiral biaryls via ring-opening reaction 38−46 from these optically active α-hydroxyl ketones 3. However, the challenge is to develop an efficient method that can break the C−C bond at mild conditions, and at the same time the central chirality of tertiary alcohol can be efficiently transferred to the axial chirality. 47−56 It is well-established that α-hydroperoxy ketones readily undergo C−C bond breaking reaction via the degradation of hydroperoxide to the oxygen-centered radical, followed by β-scission to form acyl radial. After investigation of a coupling of metal oxidants, we were surprised to find the oxidative ring-opening reaction proceeded efficiently to give carboxylic acid under 1 atm of O 2 upon the treatment of excess NaH or KOtBu at 0°C. In order to make the purification and enantiomeric excess analysis easy, the carboxylic acids were esterified to the methyl esters. Notably, this reaction condition is fairly mild and it proceeded with 97% to full transfer of chirality. The para and meta-substituents at the phenyl ring had a marginal effect on this ring-opening reaction (4a−4o), except the substrate bearing methylthio group (4p) (Scheme 2). The absolute configuration of 4b was determined by X-ray crystallographic analysis (CCDC 2095230). Although the reaction might proceed through a radical pathway (see below for a detailed discussion), the aryl-halogen bond as well as the vinyl group were well tolerated (4d−4f, 4l−4n, and 4q). This reaction could be applied to substrates bearing big aryls (4r− 4v); however, the o-methylphenyl derivative proceeded in a quite low efficacy (4w). Introducing γ,γ′or β,β′-dimethyl did not alter the virtue of this base-promoted chirality transfer reaction (4x and 4y), while dichlorobiphenyl or binaphthyl atropisomers were obtained in decreased yields (4z and 4aa). Baeyer−Villiger oxidation of 4a with mCPBA and triflic acid gave inseparable regioisomeric esters 5a and 5a′. Subsequently, hydrolysis, followed by treatment with TMSCHN 2 gave bicarboxylic ester 6a in 37% overall yield, along with 39% yield of methyl ether 6a′ in a decreased ee value.
Mechanistic Studies. After establishing the two-step sequence for axially chiral biaryl synthesis, we became interested in demonstrating the possible mechanism of this reaction. The coordination of phosphine to Rh 2 (OAc) 4 was reversible, which was monitored by 1 H and 31 P NMR upon gradually increasing the loading of L6. Notably, with 1:1 ratio of Rh 2 (OAc) 4 and L6, the reaction seemed to form a mixture of Rh 2 (OAc) 4 (L6) and Rh 2 (OAc) 4 (L6) 2 , along with a small amount of insoluble Rh 2 (OAc) 4 (Figure 2). A pair of doublet of doublets signals in 31 P NMR were observed by the analysis of the metal residue after the reaction. Furthermore, replacing phenyboronic acid with 4-fluorophenyboronic acid would change the chemical shift of these two doublet of doublets peaks ( Figure S11), which were assigned to the dirhodium(II) complex M4 (vide infra). The UV−vis spectra of Rh 2 (OAc) 4 (L6) 2 has a shoulder absorption at ∼500 nm, which is ascribed to the Rh−Rh π* to Rh−Rh σ* HOMO− LUMO transition. 57 The metal residue after the reaction also displayed an absorption at ∼500 nm. These observations confirm that the dirhodium(II) complex did not dissociate to the monomer (Figure 3a). The oxidation state of the rhodium catalyst was still +2 based on the X-ray photoelectron spectroscopy (XPS) analysis of the reaction mixture ( Figure  3b). It was found that the ee value of 3a was perfectly linear with the optical purity of L6, which possible indicated the catalytically active Rh complex binding with only one molecule of L6 (Figure 3c). Kinetic analyses were performed to learn the orders in the dirhodium(II) catalyst and each reagent. All the reactions displayed a short induction time (see the Supporting Information for details), and Rh 2 (OAc) 4 (L6) 2 and phenylboronic acid 2a displayed first-order kinetics versus the reaction rate (Figure 3d,e). It was interesting to observe that 1a showed a good zeroth-order kinetics, indicating 1a did not involve in the rate-determining step (Figure 3f). These kinetic behaviors are different from the observations in Gois's catalytic system, 9 where a concerted three-component interaction model of dirhodium(II)−(phenylboronic acid)−(aldehyde) was proposed.
On the basis of the above studies and previous reports by other groups, 58 we tentatively proposed a pathway for the Rh 2 (OAc) 4 (L6) 2 catalyzed arylation reaction ( Figure 4A). The coordination of L6 with dirhodium(II) species is a kinetic process, where complex M1 releases one molecule of L6 to form M2. The second arm of the phosphine ligand in M2 coordinated to the Rh atom to push one of the bridging acetates to fully bind to the other Rh atom (M3), which was the possible process causing a short period of induction time in kinetic studies ( Figure S9). This coordination mode of acetate with dirhodium(II) complexes has been observed by Gois and Li. 9,59 Luckily, an independent reaction of Rh 2 (OAc) 4 (1.0 equiv) and L6 (1.0 equiv) in the absence of aryl boronic acid and diketone in toluene at 80°C gave C−H metallic rhodium complex M6 (Figure 4B), which is collateral evidence of the above coordination mode. The formation of dirhodium-aryl species, i.e., M4, via transmetalation of M3 with arylboronic acid had been previously studied by Doyle and co-workers. 12 Transmetalation is considered as the rate-determining step, which is consistent with the kinetic behaviors observed in Figure 3d−f. Subsequently, stereoselective carbonyl addition delivered the oxyrhodium M5. Finally, releasing the alkoxyl  group with the assistance of the acetate anion would regenerate either M2 or M3 rhodium catalyst to complete the catalytic cycle.
It was observed that the colorless solution of 3a in THF quickly changed to deep blue upon the addition of tBuOK over several seconds. The deep blue color would last for around 30 s, and the color then turned to light yellow, which indicates the end point of the ring-opening reaction (Figure 5a). To elucidate the possible rationale of the base-promoted oxidative ring-opening reaction, a couple of control experiments were conducted (Figure 5b). A standard Schlenk technique with N 2 atmosphere, tBuOK, or NaH (4.0 equiv) still promoted the ring-opening reaction to deliver 7a in decent yield at 60°C, along with 7−15% of unknown compound (based on the mass isolated). tBuONa resulted in a slightly reduced yield. Even in a glovebox, compound 7a was still formed, albeit in a relatively lower yield. These results indicated that a trace amount of O 2 still could promote this reaction upon heating. The oxidative ring-opening reaction is not affected by light: excellent yields can be obtained either under visible light irridiation or under dark conditions (Figure 5c). The oxidative ring-opening of 3aa followed by methylation gave 6aa in 54% yield, along with 8aa (CCDC 2095231) in 36% yield (Figure 5d). The addition of TEMPO dramatically suppressed the reaction, and the corresponding TEMPO adduct was confirmed by HRMS (see the Supporting Information). Moreover, a prominent EPR signal was observed by a spin trapping experiment with DMPO, whose hyperfine splitting (hfs) constants (a N = 1.393 mT; a Hβ = 2.087 mT) well match the classic acyl-DMPO radicals (Figure 5e).
After the analysis of the configuration and conformation of crystal structures of 3aa and 4b, a stereochemical model for the transfer of chirality was proposed (Figure 5f) . It is calculated that (S a ,S)-3a was thermodynamically 7.9 kcal/mol less stable than (R a ,S)-3a, 60 whose anion species 9a is readily oxidized by O 2 to deliver 10a. β-Scission of oxygen-centered radical 10a gave a more stable acyl radical 11a, 61−64 which reacted with oxygen, followed by peroxide radical degradation, to produce the axially chiral carboxylic acid. Under nitrogen atmosphere, reduction of (R a ,S)-3a via a single electron transfer (SET) process afforded 13. 65−67 The second SET reduction, followed by elimination of hydroxide anion, gave the side product phenol.

■ CONCLUSIONS
We have developed a dirhodium(II)-catalyzed asymmetric arylation of 4,5-disubstituted phenanthrene-9,10-diones, which  demonstrated the first chiral phosphine ligand controlled dirhodium(II)-catalyzed highly enantioselective carbonyl addition reaction. Kinetic analysis revealed that the ratedetermining step involved dirhodium catalyst and arylboronic acid, while the phenanthrene-9,10-dione is zeroth-order for this reaction. Moreover, axially chiral biaryls could be efficiently prepared through a base-promoted oxidative ringopening reaction of optically active α-hydroxyl ketones. The process conceived a radical type β-scission reaction, which had a high degree of central-to-axial chirality transfer. A notable feature of this reaction is that the radical was initiated under very mild conditions without adding extra metal oxidant. ■ ASSOCIATED CONTENT
Experimental procedures, characterization of products, and spectroscopic data (PDF) Crystallographic data for 3aa (CIF) Crystallographic data for 4b (CIF) Crystallographic data for 8aa (CIF) Crystallographic data for M6 (CIF) Crystallographic data for S26 (CIF) ■ AUTHOR INFORMATION