Reaction optimization
To develop a site- and enantioselective carbene insertion reaction into B − H bond of o-carboranes, reaction conditions were extensively explored (see Supplementary Information Table S1 for details). Our investigation began with the reaction of 1,2-(DMPS)2-o-C2B10H10 (1a, DMPS = dimethylphenylsilyl) with methyl 2-diazo-2-phenylacetate (2a) in dichloromethane (DCM) at 40 oC using various dirhodium tetracarboxylate catalysts. Upon detailed examination of these reactions, we found that two regioisomers [3aa-B(9) and 4aa-B(8)] were obtained as an inseparable mixture, while no other regioisomers were observed. The regioisomeric ratio (r.r.) of 3aa and 4aa and the enantiomeric excess (e.e.) of 3aa were determined through 1H NMR and chiral HPLC analysis, respectively. Among the achiral catalysts, Rh2(OAc)4 gave the mixture of regioisomers in 87% yield with 2.6:1 r.r. as racemate (entry 1). On the other hand, Rh2(oct)4 provided a relatively low yield (71%) but high site-selectivity (14:1) (entry 2). Furthermore, we investigated various chiral catalysts to achieve the enantioselective reaction (entries 3–11)31. As a result of examining chiral Rh(II) catalysts, it was revealed that Rh2(S-TCPTTL)4 showed quantitative yield, high site-selectivity (29:1), and enantioselectivity (25% e.e.) (entry 10). Rh2(R-BTPCP)4, the most sterically encumbering catalyst, gave trace amount of conversion of 1a, probably due to the steric effect of carborane cluster (entry 11)32. The enantioselectivity of the present reaction was affected by the solvents such as dichloroethane (DCE), cyclohexane, benzene, and trifluorotoluene (PhCF3). Especially, benzene and PhCF3 enhanced the enantioselectivity to 44% e.e. and 42% e.e., respectively, and then PhCF3 was chosen as an optimum solvent (entry 15)33. When the reaction temperature was lowered from 40°C to 0°C, the enantiomeric excess increased from 42–51% (entry 17). We were pleased to observe that quantitative yield was obtained even with 1.0 mol % catalyst loading (entry 19). When 2a (1.5 equiv) was used, the yield was reduced to 79% (entry 21).
Substrate scope
Based on these results, the substrate scope of o-carboranes was next investigated (Fig. 2). When unsubstituted, silyl- or benzyl-disubstituted o-carboranes (1a-1d) were treated with methyl 2-diazo-2-phenylacetate (2a), the yields of the desired products (3aa-3da) were all quantitative, but the selectivity was affected by the substituents of o-carborane. We found that 1,2-(TMS)2-o-C2B10H10 (1b; TMS = trimethylsilyl) gave rise to 3ba in high site- and enantioselectivity (22:1 r.r. and 55% e.e.). After close examination of diazo substrate scope in the reaction with 1b, it is disclosed that substituents on aryl and ester group play an important role in enantioselectivity (3bb-3bo). As a result, 2,2,2-trichloroethyl 2-(4-bromophenyl)-2-diazoacetate (2d) underwent the B − H insertion reaction to produce the desired product 3bd in 96% yield with high site- and enantioselectivity (25:1 r.r. and 99% e.e.). This result indicates that trichloroethyl (TCE) group is very effective to the B − H insertion34. When there was no substituent on the aryl group, the desired 3be was obtained in 98% yield with 25:1 r.r. and 89% e.e., suggesting that para-substituents on aryl group are essential for excellent enantioselectivity. Then, we evaluated various electron-withdrawing groups on para-position of the aryl ring. TCE aryl diazoacetates possessing chloro, iodo, trifluoromethyl, ketone, and ester groups provided the corresponding B − H insertion products (3bf-3bj) in high yields ranging from 89–99% with excellent site- and enantioselectivity (up to > 50:1 r.r. and 99% e.e.). p-Nitro-substituted diazoacetate (2k) was reacted with 1b at 40°C, resulting in the formation of the desired product (3bk) in 69% yield with > 50:1 r.r. and 93% e.e.. In addition, a variety of electron-donating groups such as methyl, tert-butyl, phenyl, and methoxy group on para-position were tolerable, affording the desired carboranes (3bl-3bo) in high yields with site- and enantioselectivities. Diazo compounds possessing bromo, methyl, and methoxy groups on meta-position (2p-2r) and fluoro, bromo, and methyl groups on ortho-position (2s-2u) gave the corresponding products (3bp-3bu) in good to excellent site-selectivities, enantioselectivities (up to 97% e.e.), and yields (up to 99%). TCE aryl diazoacetates that possess 3,4-dichlorophenyl, 3,5-dimethylphenyl, and 2-naphthyl groups were also compatible with the present reaction conditions (3bv-3bx). TCE heteroaryl diazoacetates including thiophene and pyridine (2y and 2z) successfully applied to the present reaction. The enantiomeric excesses of 3bk, 3bq, 3bt, and 3by were determined after desilylation because of the difficulty in separation of enantiomers.
Encouraged by these results, a wide range of carboranes were investigated in the reaction with 2,2,2-trichloroethyl 2-(4-bromophenyl)-2-diazoacetate (2d) to verify if the excellent site- and enantiooselectivity would be maintained. When o-C2B10H12, 1,2-dibenzyl- and 1,2-dimethyl-o-C2B10H10 (1c-1e) were treated with 2d, the desired products (3cd-3ed) were obtained in high yield with 99% e.e.. However, these substrates exhibited inferior site-selectivity (4.9:1 ~ 6.3:1 r.r.), suggesting that silyl groups on the cage carbon of the carborane play a critical role. To demonstrate the versatility of these Rh-catalyzed cage B − H insertion reactions, we examined whether the substrates possessing substituent on the cage boron could be employed. It is noteworthy that both 1f and 1g were smoothly converted to the desired products (3fd and 3gd) in 89% and 97% yields, respectively, without any regioisomers. Although 3,6-diphenyl-o-C2B10H10 (1f) showed 35% enantioselectivity, 1-methyl-7,11-diphenyl-o-C2B10H9 (1g) exhibited excellent enantioselectivity (99% e.e.). The structures of (R)-3bd and (R)-3gd were confirmed by X-ray crystallography (see Supplementary Information). Crystal structure of (R)-3gd was obtained after transformation of ester to carboxylic acid because of difficulty in crystal formation.
Next, we applied the present method to m- and p-carboranes (Fig. 2). Gratifyingly, the carbenes on phthalimido Rh catalyst smoothly underwent B − H insertion reactions with m- and p-carboranes. When m-C2B10H12 (1h) was treated with 2d under optimum reaction conditions, the corresponding product 3hd was obtained in 76% yield with excellent enantioselectivity (99% e.e.). 1,7-(TMS)2-m-C2B10H10 (1i) was transformed to the desired product (3id) in 92% yield with 95% e.e. with 3.0 equivalents of 2d. p-Carborane (1j) having equivalent ten B − H bonds can react with two or more carbenes to give multialkylated products. To suppress repetitive B − H insertion reaction, steric influence of the rhodium catalyst was enhanced, and it was revealed that Rh2(S-TPPTTL)4 is suitable for mono-selective B − H insertion reactions of p-carboranes, affording 3jb-3jd in good yield with high enantioselectivity (up to 97% e.e.). The structures of (R)-3hd and (R)-3jd were confirmed by X-ray crystallography (see Supplementary Information).
To prove the practicability of the present catalytic procedure, the B − H insertion reaction of o-carboranes was examined on a large scale using 1.01 g (3.50 mmol) of 1b. After completion of Rh-catalyzed B − H insertion reaction, the one-pot desilylation reaction was successfully carried out, leading to the desilylated products 5 and 6 in high yields (85% and 91%, each) with excellent enantioselectivity (99% e.e.).
CD spectra, reactivity comparison, and synthetic applications
Circular dichroism (CD) spectra of (R)-3bd and (S)-3bd obtained with Rh2(S-TCPTTL)4 and Rh2(R-TCPTTL)4 catalyst exhibited unambiguously mirror images to each other, indicating a pair of enantiomers (Fig. 3a). Furthermore, the absolute configuration of (R)-3bd and (S)-3bd was confirmed by X-ray crystallography.
To examine the reactivity of carboranes with rhodium carbenoids, competition experiments were conducted with various carbenophiles using 1.0 equivalent of 2d under the optimum reaction conditions (Fig. 3b). First, we initiated competition experiment with 1,4-cyclohexadiene (1,4-CHD) that rapidly undergoes allylic C − H insertion reactions with rhodium carbenoids35. As a result, 7 was obtained in 70% yield without the formation of 3bd, suggesting that reactivity of 1,4-CHD is strong compared to 1b. Next, competition reaction of 1b with dioxolane furnished 3bd (21%) and 8 (47%). This result implies that reactivity of dioxolane has slightly better than that of 1b. Finally, since 3bd was only produced from competition experiment of 1b and tetrahydrofuran (THF), relative reactivity order of these carbenophiles could be listed as follows 1,4-CHD > > dioxolane > TMS-carborane (1b) > > THF.
To explore the application of these reactions, further functionalization of 5 and 6 was attempted. When 5 was treated with DIBAL-H, the corresponding alcohol 10 was obtained in 75% yield without erosion of the stereochemical fidelity. Trichloroethyl ester was successfully transformed to carboxylic acid 11 in 89% yield using zinc and acetic acid also with no erosion of enantiomeric excess. Enantiomeric excess of 6 was slightly deteriorated under coupling reaction conditions. As a result of Sonogashira and Buchwald-Hartwig cross-coupling reactions with 6, desired internal alkyne 12 and diaryl amine 13 were produced in 86% (89% e.e.) and 68% yields (86% e.e.), respectively.
Mechanistic studies
A deuterium labeling experiment revealed that B − H insertion reaction occurs through concerted mechanism because the deuterium atom substituted at B(9)-position of o-carborane 1b-[D]n was transferred to the α-carbon adjacent to cage boron of the product 3bd-[Dn] without a change in the H/D ratio. When 1b-[D]n or 1b was treated with H2O or D2O under the optimum reaction conditions, deuterium scrambling was not observed at all (Fig. 4a).
In addition to the experimental data, computations were also conducted to understand the site- and enantioselectivity of this dirhodium-catalyzed carbenoid B–H insertion reaction into icosahedral cage o-carborane using density functional theory (DFT). The applicability of DFT for studying dirhodium-catalyzed reactions and C–H bond insertions have been explored by others36–46. It is noteworthy that despite significant efforts by multiple research groups, these pioneering efforts reveal the enormous challenges and complexities involved with computing transition structures of large and conformationally flexible systems. Most computational studies of dirhodium carbenoid insertion processes have been rationalizations from ground state structures. To date, there are only two computed transition state studies involving the full dirhodium-catalyzed carbenoid insertion for C–H bonds, and none for B–H bond insertions using carboranes. Houk, Davies, and co-workers reported an enantioselective functionalization of a non-activated primary C–H bond using an alkyl substrate, but this study involved a relatively conformationally rigid catalyst and a judicious choice of QM/MM methodology to deal with the cost of computing such large structures40. Tantillo and co-workers reported ab initio molecular dynamics simulations to rationalize the origins of selectivity in a C–H functionalization involving an intramolecular 1,4-shift46. Herein, we are pleased to report the first fully quantum mechanically computed transition structures of a chiral dirhodium-catalyzed carbenoid B–H insertion reaction of carboranes involving the complete experimentally used ligands and substrates with no structural simplifications. Our DFT results reproduce experimentally observed site- and enantioselectivity. In addition, we reveal a tool to visualize and highlight the structurally subtle, but energetically critical, steric close contacts in a topographical view to elucidate the specific functional groups and moieties. All computations and structures presented in this paper were performed at the PBE-D3BJ level of theory in conjunction with the LANL2DZ(Rh, Br, Cl) & 6-31G* (for all other atoms) basis sets as implemented in Gaussian 16. CPCM(C6H6) solvation corrections were also used at 0°C. Single point energy refinements were performed at the PBE-D3BJ level of theory with the Ahlrich def2-TZVP basis set (see Supplementary Information for details).
The proposed catalytic cycle for the synthesis of product (R)-3bd begins with the decomposition of the diazo compound 2d by the dirhodium catalyst Rh2(S-TCPTTL)4 to afford the dirhodium carbenoid intermediate I (∆G = 0.0 kcal/mol) with the release of molecular nitrogen gas (Fig. 4b). The highly reactive dirhodium carbenoid I undergoes B − H insertion with the incoming o-carborane 1b, forming the major three-member transition state (TS) (II-TS(R)−B(9), ∆G‡ = 6.92 kcal/mol), which gives the site selective at B(9)-position and enantioselective preference (R)-enantiomer at the exohedral carbon-stereocenter, which is a carbon-sterocenter adjacent to cage boron of the carborane. This major II-TS(R)−B(9) leads to the following ground state product complex III (∆G = − 37.5 kcal/mol) wherein the desired product is embedded in the dirhodium catalyst pocket. A second diazo compound 2d releases the major product (R)-3bd (∆G = − 51.1 kcal/mol), as well as molecular nitrogen gas, resulting in regeneration of the dirhodium carbenoid I for the next catalytic cycle. The complete reaction coordinate diagram for this proposed mechanism and the energies are shown in the Supplementary Information (Fig. S8).
Previous reports by Houk and Davies hypothesized that the helical arrangement of the phthalimide ligands of the chiral dirhodium catalyst observed in the ground state as important in determining the selectivities of the C − H insertion process in their studies. The conformational complexity and substantial molecular size of this chiral dirhodium-catalyzed carbenoid B–H insertion of carboranes that were challenging to DFT compute also posed significant difficulties to discover and explain where the origins of selectivity arose within the large transition structure complexes. To address these challenges, we employed a Topographical Proximity Surfaces (TPS) visualization to analyze the steric repulsions that exist in the TSs (Fig. 4c). First, taking the DFT optimized II-TS, the electron isodensity surface of the dirhodium carbenoid complex I was rendered. Then the surface was color-coded to reflect the close steric contact between it and TMS groups of the o-carborane 1b. In this manner, the intensity of the color represents the severity of steric interactions. Hence, this TPS approach can reveal close steric contacts in large, complex transition structures and aid in the rationalization of reaction selectivities.
In the favored major (R)-B(9)-insertion TS (II-TS(R)−B(9), ∆G‡ = 6.92 kcal/mol), the dirhodium carbenoid insertion occurs at B(9) position of o-carborane 1b to give the (R)-configuration product at the exohedral carbon-stereocenter adjacent to cage boron of the carborane. The unfavored epimeric dirhodium carbenoid insertion results in the minor (S)-B(9)-insertion TS (II-TS(S)−B(9), ∆G‡ = 8.96 kcal/mol, i.e. stereoselectivity of 2.04 kcal/mol), and the unfavored regioisomeric insertion results in the minor (R)-B(8)-insertion TS (II-TS(R)−B(8), ∆G‡ = 9.00 kcal/mol, i.e. site-selectivity of 2.08 kcal/mol). These DFT results agree with the experimental site- and enantioselectivity of 2.00 kcal/mol and 2.87 kcal/mol, respectively. The TPS visualization of the major (R)-B(9)-insertion TS (II-TS(R)−B(9)) reveals a comparatively diminished steric repulsion between TMS groups of the o-carborane 1b and the dirhodium carbenoid complex I (Fig. 4c). This is a result of the o-carborane angle and positioning of the TMS groups into the phthalimide ligand cavity (movie S1). In contrast, the TPSs of the epimeric (S)-B(9)-insertion TS (II-TS(S)−B(9)) and the regioisomeric (R)-B(8)-insertion TS (II-TS(R)−B(8)) both show greater steric repulsion of the TMS groups against the phthalimide ligands of the dirhodium carbenoid complex I. In the former, in order to achieve the epimeric insertion of the minor (S)-configuration product, it necessitates the angle and positioning of the o-carborane such that the TMS groups clash into the wall of the phthalimide cavity (II-TS(S)−B(9), movie S2). Similarly, in the regioisomeric (R)-B(8)-insertion TS (II-TS(R)−B(8)), the rotation of the o-carborane to achieve insertion at the B(8) position not only results in steric interactions between the TMS groups with the phthalimide ligands, but also with the aryl substituent of the carbene substrate itself (movie S3). These results visually reveal the extent and severity of steric interactions that govern the preference for the favored B–H insertion process by this large and conformationally flexible dirhodium catalyst, Rh2(S-TCPTTL)4.