Stereogenic and Conformational Diversity in Multicomponent Ir-Catalyzed Borylation of Arenes: Unveiling the Origin of Enantioselectivity by Conformational Sampling

Abstract

To date, multicomponent transition-metal-catalyzed asymmetric reactions have led to pivotal breakthroughs in modern catalysis. However, investigations on the mechanism of these reactions are inherent challenges due to the coeffect of the complicated stereostructures and multiple noncovalent interactions. Here, we proposed a stepwise computational workflow combining density functional theory (DFT) calculations and molecular dynamics (MD) simulations to investigate a multicomponent Ir-catalyzed asymmetric borylation of arenes induced by chiral cations. Based on a definite rate-determining step investigated by DFT calculations, we employed MD simulations with constrained harmonic potential for coordinate bonds to completely sample the conformations of enantioselectivity-determining transition-state ensembles. Interestingly, results reproduce well the experimental enantioselectivity (ee value) and clearly reveal that both quantity and energy of conformations are responsible for the enantioselectivity. Systematical analysis on the conformations suggests that the advantages of tert-butyl group substitution at the meta position of the outer chiral cation for enantioselective C–H borylation do not arise from steric repulsion of the tert-butyl group destabilizing the unfavorable R-forming transition states but other two factors: (i) the limited conformational freedom due to the bulky tert-butyl group and (ii) the dispersion interactions between the tert-butyl group and the methyl group of the Bpin ligand stabilize the S-forming transition state more than R-one. This work underscores the fundamental significance and importance of conformational space accessibility and multiple noncovalent interactions for elucidating the origin of asymmetry induction.

Introduction

Asymmetric C–H bond activation of arenes has emerged as a powerful tool in organic synthesis, allowing for the selective functionalization of C–H bonds to access chiral molecules. , In recent years, this field has witnessed significant advancements and garnered increasing attention due to its potential to achieve the synthesis of complicated organic compounds with high levels of stereocontrol. − Among those outstanding works, the design of chiral ligands for transition metal catalysts has been a focal point in the development of asymmetric C–H functionalization. In general, a chiral ligand is essential for inducing enantioselectivity in transition-metal-catalyzed reactions by favoring the formation of one enantiomer over another. Numerous research groups, including those of Hartwig, , Noyori, Nishiyama, Shi, , Xu, , Jacobsen, Zhang, and Yu have devoted considerable efforts to employing chiral ligands (e.g., PyOX, BINAP, PyBOX, Taddol, CBL, Salen, and Sadphos) to enhance the enantioselectivity of these transition-metal-catalyzed reactions, providing new opportunities toward diversity-oriented asymmetric synthesis.

In addition to transition-metal catalysts with chiral ligands, various alternative chiral sources, such as chiral solvents, cations, anions, − aluminum Lewis-acidic cooperative catalysts, and chiral-at-metal complexes, , have been demonstrated to enable enantioselective reactions, as shown in Scheme . Similar to the function of chiral ligands, when these additional chiral auxiliaries are introduced to the catalytic system, the interactions among active species, substrates, additives, and even solvents in a stereoselective manner allow the complexes to exhibit the stereochemistry of a chiral transition-metal catalyst, guiding the formation of chiral products. , For instance, Meggers and co-workers designed an array of concise chiral metal catalysts, which are solely composed of optically inactive ligands and chiral metal centers (e.g., Ir, Ru, and Fe), achieving the asymmetric induction in the course of creating a new stereogenic center. ,,− Ishihara and co-workers have reported a series of asymmetric organocatalytic systems without chiral ligands , and developed an ion pair consisting of a radical cation and a chiral counteranion to induce an excellent level of enantioselectivity. Li and Wang et al. recently proposed that metal-centered chirality can serve as a source of enantioselectivity in binaphthols-metal asymmetric catalytic systems, providing valuable guidance for the design and optimization of related chiral catalysts.

1. Asymmetric Catalysis with Various Chiral Sources.

In the catalytic systems utilizing chiral cations, the unique design of these carefully tailored ions allows for dynamic stereostructures and multiple noncovalent interactions. For instance, Phipps et al. developed an anionic bipyridine ligand with a remote sulfonate group, which enabled iridium-catalyzed C–H borylation at the arene meta position via noncovalent interactions (hydrogen bonding) between the sulfonate group and the substrate. − Recently, they employed a series of chiral cations derived from dihydroquinine (DHQ) with varying N-benzyl substitution instead of the achiral tetrabutylammonium counterion to achieve an enantioselective remote C–H borylation of a prochiral symmetrical benzhydrylamide substrate 2a, as shown in Scheme a. This is also the first asymmetric transition-metal-catalyzed approach for the combination of a chiral cation with an anionic achiral ligand reported to date. Several interesting and important experimental findings have been reported about this enantioselective borylation, as summarized below: (i) in the ion-paired ligands, a variety of chiral cations were investigated by placing substituted aromatic rings at the 3- and 5-positions of the quaternizing N-benzyl group, with L·1g (L and 1g denote the anionic achiral ligand and the chiral cation, respectively) exhibiting the best enantioselectivity for generating meta-borylated S-product (S)-3a′ (73% ee), but the reaction with L·1k was not enantioselective (0% ee), as shown in Scheme b; (ii) the variant of 1g gave a decreased ee value, as shown in Scheme a, in which 1g-OMe (the quinine hydroxyl group is methylated) and epi-1g (the stereochemistry of the quinine hydroxyl group is inverted) gave 72% ee and 11% ee, respectively. Importantly, the ligand L·1j possessing a diastereomeric chiral cation derived from quinidine gave R-product (R)-3a′ instead of S-product with −90% ee; and (iii) the N-methylated variant of 2a gave only 8% ee (Scheme b) and the product was racemic when ion-paired ligand L·1g was replaced with neutral 5,5′-dimethylbipyridine (Scheme c) together with the optimal chiral cation as its bromide salt (Br·1g), indicating the necessary hydrogen bonding interaction of substrate with ligand and the requirement for ligand and chiral cation to be associated to achieve asymmetric induction. Considering these findings, we aim to tackle three critical questions: (a) why do the remote quaternizing N-benzyl groups of chiral cations have a significant effect on enantioselectivity? (b) What is the underlying mechanism behind the dependence of enantioselectivity on the stereostructure of chiral cation and the interactions among the multicomponents, such as chiral cation, achiral ligand, and substrate? (c) What is the precise way that the chiral cation induces enantioselectivity in these reactions?

2. (a) Ir-Catalyzed Enantioselective Desymmetrizing C–H Borylation of Benzhydrylamides; (b) Enantioselectivity Is Dependent on the Chiral Cation (Solvent: THF) .

^a COD, 1,5-cyclooctadiene; rt, room temperature; t-Bu, tert-butyl; B₂pin₂, bis­(pinacolato)­diboron; ee, enantiomeric excess.

3. (a) Expanded Cations and Experimental Results about Enantioselectivity (Solvent: CPME, Temperature: −10 °C, 2 Equivalent B₂pin₂); (b) Control Experiments to Probe Ligand–Substrate Interactions; (c) Control Experiments to Probe Ligand–Cation Interactions .

Generally, the density functional theory (DFT) calculations are able to evaluate the differences between chemo-/regio-/enantioselectivity-determining transition states (TSs) by providing detailed potential energy surfaces and electronic properties. However, diverse conformations could be presented in the bulky chiral cation, ligand, and substrate of this multicomponent reaction, which is very difficult for a DFT calculation study because such a reaction contains a considerable system size (263 atoms), a great number of rotatable bonds, and multiple dynamic noncovalent interactions. The accessible conformational spaces are difficult to be fully traversed in a DFT calculation, leading to a limited set of enantioselective models and a risk of missing low-energy conformations. Moreover, because the absolute configuration of the cation is typically intricately connected to the variable Ir-centered chirality (Λ and Δ) and borylated product chirality (S and R), a plenty of diastereomeric intermediates further increases the computational complexity. It is worth noting that Ermanis et al. have reported a computational study on the mechanism, noncovalent interactions, and the possible origin of selectivity for the desymmetrizing borylation of amide substrate. Their work underscores the importance of the conformational space of diastereomers in asymmetric catalytic reactions. Their results suggest that the trifluoroacetamide group of the substrate in the R-forming transition state is clashing with the large chiral cation, thereby causing it to rotate into a less favorable and higher-energy conformation, providing a rationalization for the observed S-product selectivity. Besides conformational diversity, the variable chirality at the Ir center and the absolute chirality of cations could play an important role in controlling enantioselectivity, which has been discovered recently. Therefore, in addition to conformational sampling using molecular dynamics (MD) simulations as reported in previous works, ,− a thorough consideration of all diastereomeric transition-state ensembles determined by the chiralities of Ir center, cations, and borylated products is necessary.

Herein, we described a detailed computational study on revealing the origin of a complicated Ir-catalyzed enantioselective borylation of arenes with multicomponents. We proposed a systematic and stepwise workflow in this study to investigate the reaction mechanism and disclose important factors controlling the selectivity of transition-metal-catalyzed reactions with multicomponents. Particularly, we employed a MD simulation method with a constrained harmonic potential for coordinate bonds to completely sample the conformations of enantioselectivity-determining transition-state ensembles. Further analysis clearly shows how extensive conformational space sampling can lead to a better understanding of stereogenic and conformational diversity in the Ir-catalyzed enantioselective borylation of arenes.

Computational Details and Models

All DFT calculations were conducted with the Gaussian16 program unless otherwise noted. The geometry optimizations were performed using B3PW91 functional , augmented with D3 dispersion correction by Grimme and co-workers and 6-31G­(d) basis set, with SDD basis set on iridium (BS-I). In our previous work, this combination reproduced well experimental structures of the related iridium­(III) boryl complexes. ,, Frequency analysis was employed for all obtained geometries at the optimization level of theory to characterize the structures as minima (no imaginary frequency) or TS (only one imaginary frequency). To accurately evaluate potential energy, single-point calculations were performed using the ωB97XD functional and a larger basis set system (BS-II) based on the geometries optimized at the B3PW91-D3/BS-I level. In BS-II, two f polarization functions were added to Ir and 6-311+G­(d,p) basis sets were used for all other atoms. − The quality of this computational method was checked by comparing energy changes between the ωB97XD and DLPNO-CCSD­(T) calculations , of a full system using the ORCA program , (5.0.4); details are presented in Figures S1, S2, and Table S1. Our previous work also suggested that this computational method is reliable for several model reactions of iridium complexes. ,, The solvation effect (THF) was evaluated with the polarizable continuum model − in the single-point calculations. Since Gaussian does not support CPME parameters, we employed THF solvent parameters in place of CPME because the majority of solvent used for control experiments of chiral cations was THF. The constrained MD simulations using the xtb program (v.6.5.1) were performed to sample the conformational space of TS, and the chemical bonds that are about to form or break were constrained to specific values by harmonic potential. ,, The steric map calculations were conducted by the SambVca (2.1) web application to quantitatively evaluate the steric hindrance. The independent gradient model based on Hirshfeld partition of molecular density (IGMH) analysis for evaluating the noncovalent interactions and energy decomposition analysis (EDA) was performed using Multiwfn wave function analysis code and the sobEDA method. TS and the important noncovalent interactions were visualized using VMD (v.1.9.4), and structures were visualized with CYLview20.

In this work, the Gibbs free energy changes were used for discussion. Thermal correction and entropy contribution to the Gibbs free energy were evaluated at the experimental temperature (298.15 K) and 1 atm, where the translational entropy in solution was corrected by the method of Whitesides et al. Enantiomeric excess (ee) was calculated as a difference between the Boltzmann factors of the energy barriers corresponding to various conformational isomers of transition-state structures for the formation of S and R enantiomers.

Results and Discussion

The overall flowchart of this study on the mechanism, stereogenic diversity, and conformational space sampling is outlined in Scheme . We will first discuss the plausible catalytic cycle and Gibbs free energy profiles to identify the rate-determining step. Then, we will discuss the potential diastereomeric active species, important intermediates, and TS derived from the Ir-centered chirality in the desymmetrizing borylation of the substrate. In addition, a deconstructive, stepwise conformational sampling approach is described to obtain the ensemble of diastereomeric TS emerging from these conformational possibilities; detailed procedures and the calculation results will be discussed later. Finally, we will make a careful analysis of the key noncovalent interactions involved and provide a fuller description of the enantioselectivity.

4. Overall Workflow of the Study Combining DFT Calculations and MD Simulations, including the Catalytic Mechanisms Investigation, Stereogenic Diversity Analysis, and Conformational Space Sampling of TS.

Catalytic Mechanism of the Ir-Catalyzed C­(sp²)–H Borylation

For the most common bipyridine and related ligands, − the catalytic cycle of the Ir-catalyzed C–H borylation is well established. In addition to experimental mechanism studies, many computational studies using DFT methods have been performed to explore the mechanisms and selectivities, including our previous works. , As shown in Figure , several important processes are presented in the catalytic cycle of Ir-catalyzed C–H borylation, including the formation of active species, C–H oxidative addition, isomerization, B–C reductive elimination, and catalyst regeneration. Here, we calculated the Gibbs free energy profile of this catalytic cycle to identify the rate-determining step in the presence of various interactions between the sulfonated bipyridine ligand, the symmetrical benzhydrylamide substrate, and the outer chiral cation, although the C–H oxidative addition is demonstrated to be the rate- and selectivity-determining step in most aromatic borylation reactions. −

1.

Plausible catalytic cycle for the Ir-catalyzed C­(sp²)–H borylation of symmetrical benzhydrylamide, HBpin, and pinacolborane.

We first considered the relatively simple quinine-derived chiral cation 1k (dihydroquinine with N-benzyl) in the experimental catalyst system and briefly discuss these elementary steps because the basic mechanism is well-known. As shown in Figure , starting with INT3, consisting of a substrate coordination to the Ir atom of (N,N)­Ir­(Bpin)₃, the oxidative addition of the C–H bond occurs via transition state TS2 to afford seven-coordinated complex (N,N)­Ir­(Bpin)₃(H)­(phenyl) INT4 with a Gibbs free energy of activation (ΔG°^‡, 15.6 kcal mol^–1) and a Gibbs free energy of reaction (ΔG°, 4.1 kcal mol^–1) relative to INT3. Then, intermediate INT4 undergoes isomerization with position changes of three Ir–B bonds via transition state TS3 (ΔG°^‡, 10.9 kcal mol^–1 relative to INT3) to afford a good geometry for the B–C reductive elimination. An alternative possible isomerization pathway (hydride migration) was also considered, as shown in Figure S3, but with a higher energy barrier (21.9 kcal mol^–1). Subsequently, the B–C reductive elimination occurs via a transition state TS4 to form borylated product (S)-3a′ and iridium­(III) species INT6 with a moderate ΔG°^‡ value of 9.5 kcal mol^–1 and ΔG° value of −2.4 kcal mol^–1 (relative to INT3). The remaining steps are regeneration of the catalyst by the oxidative addition of B₂Pin₂ and reductive elimination of HBpin. After (S)-3a′ dissociates from the Ir atom, INT6 undergoes oxidative addition of B₂Pin₂ to the Ir­(III) atom followed by reductive elimination of HBpin to generate the iridium­(III) tris-boryl complex INT9. The ΔG°^‡ values are 14.5 and 3.7 kcal mol^–1 for the oxidative addition and reductive elimination, respectively. Finally, INT9 easily reacts with one more B₂pin₂ to afford INT1 with a ΔG°^‡ value of 9.3 kcal mol^–1. All of these steps have lower energy barriers than the irreversible C–H oxidative addition. In this context, C–H oxidative addition is clearly identified as the rate-determining step, with an energy barrier of 15.6 kcal mol^–1.

2.

Gibbs free energy profile (in kcal mol^–1) for the Ir-catalyzed C­(sp²)–H borylation of symmetrical benzhydrylamide and the corresponding schematic representation of geometry changes. Above the profile is the representation of the TS, while below it is the representation of the intermediates.

Conformational Sampling for Evaluation of Enantioselectivity

Having shown that the C–H oxidative addition is the rate-determining step (enantioselectivity-determining step), we next performed a comprehensive conformational sampling of the transition-state structures to obtain the corresponding ensemble, probing the origin of enantioselectivity. Due to the presence of variable Ir-centered chirality and absolute configuration of chiral cations, as illustrated in Scheme , the formation of S and R borylated enantiomers is dependent on four potential diastereomeric TS of the C–H oxidative additions (these TSs are named Λ–S, Λ–R, Δ–S, and Δ–R, respectively). Among them, Λ–S had already been shown in Figure (TS2), and other TS of diastereomeric Δ–R, Λ–R, and Δ–S were also constructed by DFT calculations. The structures of these four TS serve as the initial input for conformational sampling, which undergoes several steps, including MD simulation, restraint minimization and redundancy elimination, and full DFT optimization of the low-energy conformations (the workflow is shown in the third part of Scheme ), leading to four ensembles of TS. Notably, an additional redundancy elimination is still necessary after the DFT optimization to ensure that independent conformations are sampled. The number of conformations for all target TS at each level is shown in Table S2. In the previous work, conformational searches for the TS were conducted using the OPLS3 force field and a 50/50 mixture of Monte Carlo/Low-Mode Following Algorithm. This method, well-established for thorough exploration of conformational space, can be computationally demanding for extensive systems. In contrast, we employed the xtb program based on the semiempirical GFN0-xTB and GFN2-xTB methods for conformational sampling (Scheme ), which facilitates a more streamlined workflow compared to that of traditional force field-based approaches. In addition, since routine force fields for MD simulation are not suitable for the simulation of the motion of coordinate bonds, we added a constrained harmonic potential in MD simulations for these bonds. A careful evaluation of the constrained parameters is shown in Figure S4. We briefly discuss why constrained harmonic potential is necessary in the MD simulations for conformational sampling. As shown in Figure S4a, when the constrained harmonic potential was not employed in the MD simulation, the C–H bond that is supposed to be at the transition state (1.61 Å), was completely cleaved at 5 ps. Although some coordinate bonds (Ir–N bonds and partial Ir–B bonds) in the initial structure survived, one of the Bpin ligands apparently dissociated from the Ir atom at 5 ps and remained uncoordinated to Ir in subsequent simulation. If these bonds were constrained by harmonic potential, as shown in Figure S4b,c, the transition-state regions and coordinate bonds were well maintained, and these bond lengths remained essentially unchanged with the force constant set to 2.0.

With the constrained harmonic potential for the key coordinate bonds, MD simulations (250 ps) of the transition state of C–H oxidative addition were performed, generating about 5000 frames. During the simulation, the conformations of the outer flexible groups of chiral cation, ligand, and substrate, except for the constrained moieties, were constantly changing, resulting in a continuously fluctuating root-mean-square deviation (RMSD), as shown in Figure S5a. For highly flexible transition-metal complexes, the recurring conformational variation is the equilibrium state of the system. Then, multiple steps of restraint minimization and redundancy elimination were performed to sort the remaining conformations by energy and filtered for low-energy conformations (energy difference: 0–3 kcal mol^–1). Wherein, the threshold for redundancy elimination was also examined carefully, as shown in Figure S5b, where the remaining conformations have been essentially balanced at the criteria of 0.50.

Systematic conformational sampling was carried out for these four diastereomers to obtain the corresponding transition-state ensembles, as shown in Figure . For each ensemble, the number of squares represents the quantity of conformations, and the color (from blue to red) and percentage indicate the fraction of the Boltzmann distribution for each conformation. After final redundancy elimination following the optimization by DFT calculation, a total of 24 conformations were identified, spanning a range of 8.4 kcal mol^–1. As shown in Figure S6, we elected to graphically represent the TS ensembles for each generalized structure as overlays of the corresponding TS with the lowest-energy transition state per diastereomeric category represented in bold. In Figure , the Boltzmann contributions corresponding to the lowest-energy transition-state conformations for the formation of S and R borylated products are 23.0% and 42.6%, respectively, and it appears that the selectivity of the R product is higher than that of the S product. However, the Boltzmann contributions of all conformations are non-negligible for each transition-state ensemble. The calculations of the ee values should consider both the quantity and the energy of each conformation rather than focusing only on the energy of one conformation. Therefore, the ee values are calculated using the following formulas

$$k^S=A\sum _i\exp (-\mathrm{\Delta }{G_{S_i}^o}^{‡}/RT)$$ 1

$$k^R=A\sum _i\exp (-\mathrm{\Delta }{G_{R_i}^o}^{‡}/RT)$$ 2

$$\mathrm{e}\mathrm{e}\%=\frac{k^R-k^S}{k^R+k^S}\times 100$$ 3

where k ^S/R is the macroscopic rate constants for the formation of each product and $\mathrm{\Delta }{G}_{S_i/R_i}^{{◦}^{‡}}$ are the activation free energies for the formation of S or R products, respectively. The calculations predicted an ee value of 0.2% at 298.15 K, in excellent agreement with experimental observations (0% ee). These results clearly suggest that both the quantity and energy of complete conformations are responsible for the enantioselectivity.

3.

Boltzmann distribution of each conformation in the transition-state ensembles for the formation of S and R borylated enantiomers. For each ensemble, the number of squares represents the quantity of conformations in the corresponding ensemble, and the color (from blue to red) and percentage indicate the fraction of the Boltzmann distribution for each conformation; the color scale is given. Energy range for all conformations: 0–8.4 kcal mol^–1. The predicted enantioselectivity: 0.2% ee.

In stark contrast to chiral cation 1k, quinine-derived chiral cation 1g with tert-butyl-substituted meta positions of the outer arenes was shown to have a strong influence on the quantity and energy distribution of the TS. The final ensembles are presented in Figure , with a total of only 13 conformations within 4.2 kcal mol^–1. Of the four diastereomeric transition-state categories, the Λ–S ensemble contained the lowest-energy transition state and accounted for the highest Boltzmann contribution (39.9%) to enantioselectivity. In a similar manner, the second- and third-lowest Boltzmann contributions (27.1% and 15.4%) were also found to favor the S borylated product. However, the highest Boltzmann contribution for the formation of the R borylated product was only 6.6%, and the quantity of low-energy TS in the corresponding ensembles (Λ–R and Δ–R) was insufficient. The final ensembles are also shown in Figure S7 as overlays of TSs from each category (lowest transition state in bold). Enantioselectivity was calculated as mentioned above, predicting an ee value of 66.8% in favor of the S borylated product, in good agreement with the experimental findings (73% ee). We recalled the experimental results that the variants of the 1g chiral cation had lower ee values (CPME as a solvent, Scheme a), with 1g-OMe and epi-1g giving 72% ee and 11% ee, respectively, whereas 1j exhibited a similar enantioselectivity but generated opposite stereochemistry of the borylated product (−90% ee, favoring R enantiomer). We compared the energy difference between the lowest-energy TS for the formation of the S and R products with these cations and in this way estimated the ee values. As shown in Table S3, the results are consistent with the experimental trends (Scheme a), this confirmed the reliability of the conformational model.

4.

Boltzmann distribution of each conformation in the transition-state ensembles for the formation of S and R borylated enantiomers. For each ensemble, the number of squares represents the quantity of conformations in the corresponding ensemble, and the color (from blue to red) and percentage indicate the fraction of the Boltzmann distribution for each conformation; the color scale is given. Energy range for all conformations: 0–4.2 kcal mol^–1. The predicted enantioselectivity: 66.8% ee.

Origin of Enantioselectivity Induced by Chiral Cations

In the experiment (Scheme ), chiral cation 1g, a variant of chiral cation 1k with tert-butyl-substituted meta positions of the outer arenes of the teraryl system, gave a dramatic increased enantioselectivity for S borylated product (form 0% ee to 73% ee). Our proposed conformational sampling protocol for evaluating enantioselectivity based on the total Boltzmann distribution of all transition-state ensembles reproduced this result well. To fully understand the intrinsic reasons why the substituents of chiral cation significantly improve the selectivity of S borylated product, the energetically favorable features of the S-forming transition state should be revealed. In terms of the lowest-energy transition state, the R-forming TS was more stable than the S-forming TS in the conformational ensembles of the L·1k ion-paired ligand (Boltzmann contribution of 23.0% in Λ–S (1k) and 42.6% in Δ–R (1k) ensembles, respectively), whereas an inverted stability was obtained for the L·1g ion-paired ligand (Boltzmann contribution of 39.9% in Λ–S (1g) and 5.8% in Δ–R (1g) ensembles, respectively). It is worth noting that although the Λ–R (1g) ensemble gave the highest Boltzmann contribution for the L·1g ion-paired ligand (6.6%, Figure ), the Δ–R (1g) ensemble had more transition state structures whose sum of Boltzmann contributions exceeded that of the Λ–R (1g) ensemble. Altogether, considering both the quantity and energy distribution of TS, the most favorable TS for the formation of the R borylated product should be attributed to the Δ–R (1g) ensemble. Thus, an analysis of these four TS for the different chiral cation cases should give the origin of the improved selectivity of the S borylated product. To understand the asymmetric induction, distortion/interaction (D/I) , and EDA were conducted to analyze the distortion energies and different types of cation–ligand–substrate interactions in the above four diastereomeric TS. As shown in Figure S8, the full system of the transition state of C–H oxidative addition was separated into three fragments, substrate (frag-a), tris-boryl Ir catalyst with an anionic bipyridine ligand bearing a remote sulfonate group (frag-b), and chiral cation (frag-c) moieties. The computed activation energies (E _a) were dissected into two terms

$$E_{\mathrm{a}}=\mathrm{\Delta }{E}_{\mathrm{d}\mathrm{i}\mathrm{s}\mathrm{t}}+\mathrm{\Delta }{E}_{\mathrm{int}}$$ 4

where ΔE _dist is the total energy required to distort the fragments from the equilibrium structures into their transition-state geometries. ΔE _int is the interaction energy between the fragments. The interaction energy can be further dissected using the EDA method

$${\mathrm{\Delta }E}_{\mathrm{int}}={\mathrm{\Delta }E}_{\mathrm{e}\mathrm{ls}}+{\mathrm{\Delta }E}_{\mathrm{x}\mathrm{r}\mathrm{e}\mathrm{p}}+{\mathrm{\Delta }E}_{\mathrm{o}\mathrm{r}\mathrm{b}}+{\mathrm{\Delta }E}_{\mathrm{DFTc}}+\mathrm{\Delta }{E}_{\mathrm{dc}}$$ 5

where ΔE _els, ΔE _xrep, ΔE _orb, ΔE _DFTc, and ΔE _dc are the energies of electrostatic interactions, exchange-repulsion term (sum of exchange energy and Pauli repulsion), orbital interactions (i.e., polarization and charge transfer), DFT correlation, and dispersion interactions, respectively.

The values for the individual energy terms were computed for all four diastereomeric TS of C–H oxidative addition (L·1k (Λ–S), L·1k (Δ–R), L·1g (Λ–S), and L·1g (Δ–R), Figures and ). To identify factors that control the enantioselectivity by chiral cations, we analyzed the difference in each energy term (ΔΔE) between pairs of TS leading to the major S and minor R products (Figure ). Distortion/interaction analysis , suggested that the distortion energy is not responsible for the overall ΔE _a values to be negative in the case of L·1g ion-paired ligand, and the interactions are the dominant factors that affect the enantioselectivity. The interactions between cation, ligand, and substrate moieties in the TS of L·1g (Λ–S) are significantly more favorable than those in the TS of L·1g (Δ–R) (ΔΔE _int = −6.8 kcal mol^–1), while the interactions of transition-state L·1k (Λ–S) are comparable to those of transition-state L·1k (Δ–R) (ΔΔE _int = −1.2 kcal mol^–1). Then, we performed EDA calculations to understand the factors that play key roles in the interaction. Because the ωB97XD functional is not supported by sobEDA analysis, the popular B3LYP-D3 for general situations is used in EDA calculations. The results (relative energies) of these two functionals in distortion/interaction analysis are essentially the same, as shown in Figure S9, thus using the B3LYP-D3 for D/I and EDA analysis can give consistent conclusions with the ωB97XD functional. EDA analysis (Figure ) indicated that dispersion is the major interaction due to the most negative value of the relative ΔΔE in the dispersion interaction term (ΔΔE _dc(L·1g) – ΔΔE _dc(L·1k) = −5.4 kcal mol^–1), resulting in a more stable S-forming TS (L·1g (Λ–S)) than the disfavored R-one (L·1g (Δ–R)).

5.

Distortion/interaction analysis of four diastereomeric TS of C–H oxidative addition. All energies are in kcal mol^–1.

6.

EDA to reveal factors that determine the enantioselectivity of the Ir-catalyzed enantioselective borylation of arenes. All energies are in kcal mol^–1.

As mentioned above, the dispersion involved in the reaction system of L·1g favor S-forming TS. Therefore, we turned our attention to exploring possible noncovalent interactions between the chiral cation, anionic bipyridine ligand, and the substrate by IGMH analysis. As shown in Figure , the cation–substrate interaction (π–π), cation–ligand interaction (H-bond), and cation–dipole interaction between the electronegative substituent (CF₃) of the substrate and the quaternary nitrogen of the chiral cation were identified in two diastereomeric TS of Λ–S and Δ–R with L·1k chiral cation, reflecting the precise nature of cation–substrate and cation–ligand interactions probed by control experiments. These interactions were also found in the TS of Λ–S and Δ–R with the L·1g chiral cation, which are barely changed. Strikingly, the differentiation by attractive dispersion effects was illustrated in the form of the large interface in the TS of Λ–S with L·1g and the small interface in the TS of Δ–R with L·1g. The extended teraryl group with tert-butyl-substituted meta positions allows dispersion interaction between the alkyl groups on the outer chiral cation and methyl groups on the Bpin ligand of the inner tris-boryl Ir complex. It is worth noting that the strategy of controlling enantioselectivity by exploiting the critical attractive dispersion interaction between alkyl groups in flexible transition-metal catalytic systems has been reported by recent experimental and theoretical studies. − Additionally, we reoptimized the previously reported transition-state structures using the computational method in this work to investigate the effect of dispersion correction on these structures. As shown in Figure , introducing dispersion correction during geometry optimization causes the tert-butyl group of the cation to be closer to the anionic Ir-catalyst moiety (8.98 Å → 5.94 Å). IGMH analysis reveals that more noncovalent interactions appear in the transition-state structures involving dispersion corrections, and these structures are more stable by 2.26 kcal mol^–1 for the S-forming transition state and 3.35 kcal mol^–1 for the R-forming transition state, respectively (Figure S11). These results indicate that dispersion correction during the geometry optimization stage is crucial for such complicated catalytic systems. In this context, the advantages of tert-butyl group substitution at the meta position of the outer chiral cation for enantioselective C–H borylation are associated with two factors: (i) the limited conformational freedom due to the bulky tert-butyl group, and (ii) the dispersion interactions between tert-butyl group and methyl group of the Bpin ligand in the lowest-energy Λ–S stabilize the S-forming transition state more in favor of the formation of S borylated product. To further understand the reversal of enantioselectivity in the case of 1j, the conformational search on the TS of the C–H activation with the 1j cation was performed. As shown in Table S4, the ee value was calculated to be −55.2%, which is close to the calculated value of −56.6% based on the most-stable transition state structure of 1g (Table S3). Importantly, noncovalent interaction analysis revealed that additional dispersive interactions are present in the R-forming transition state for 1j (Figure S10), which enhances the selectivity for the R-product and thus results in the observed reversal of enantioselectivity.

7.

IGMH analysis and visualization of noncovalent interactions of the lowest-energy TSs of C–H oxidative addition for the formation of S and R products. The value of isosurface is 0.007 eV/ų.

8.

Optimization of the previously reported conformations of S-forming TS, including dispersion corrections.

To further explore the potential influence of steric hindrance except for dispersion, the buried volume and the steric map calculations were performed to quantitatively evaluate the differences of steric hindrance between these four TS. As shown in Figure , the difference of the Ir-centered buried volume between the S- and R-forming TS with L·1k and L·1g ion-paired ligands (ΔV _S–R ) is only 0.4%, which is reasonable because the Ir centers are enantiomeric and independent of the different substituents in the outer cation. Further examining the S-centered buried volume of the outer-sphere ligands and cations, the difference (ΔV _S–R = 0.5%) is likewise negligible. It can be concluded that the 1g chiral cation did not improve the selectivity of the S product by changing the difference in the steric hindrance of the S- and R-forming TS. On the other hand, the contribution of all conformations to enantioselectivity was also evaluated by statistical distortion/interaction analysis. As shown in Table S5 and Figure S12, the interaction energy between fragments after considering all transition-state conformations is indeed the primary factor influencing the energy difference between the S- and R-forming TS, which agrees with the above discussion. Although the ΔΔE _xrep and ΔΔE _dc values for the second- and third-most stable TS conformations with relatively minor Boltzmann contributions are similar (Figure S13), the difference in dispersion interactions between the most stable R- and S-forming TS conformations with the largest Boltzmann contributions can be identified as the primary factor in the increased interaction energy observed in the S-forming transition state (Figure ). Overall, all these calculations show that cation–ligand–substrate dispersion interactions constitute the dominant factor that controls the enantioselectivity of the desymmetrizing borylation of arenes.

9.

Steric maps and buried volume calculations based on the lowest-energy TSs of C–H oxidative addition for the formation of S and R products using a 3.5 Å sphere around Ir atom and S atom, respectively. Two ion-paired ligands of L·1k and L·1g were analyzed.

Conclusions

In this work, we investigated the Ir-catalyzed enantioselective C–H borylation of benzhydrylamides by the combination of DFT calculations and MD simulations with a constrained harmonic potential for coordinate bonds. Mechanism studies showed that the C–H oxidative addition is the rate-determining and enantioselectivity-determining step in the presence of chiral counter-cations. Due to the variable Ir-centered chirality and absolute configuration of chiral cations, four potential diastereomeric configurations of TS (Λ–S, Λ–R, Δ–S, and Δ–R) of the C–H oxidative additions for the formation of S and R borylated enantiomers were proposed to be responsible for the enantioselectivity.

Additionally, we applied a conformational sampling method tailored for these diastereomeric configurations to fully traverse the entire space of accessible conformations. The Boltzmann contribution of each conformation in the final conformational ensembles was counted, and the predicted ee values (0.2% for chiral cation 1k and 66.8% for chiral cation 1g) are in good agreement with the contrasting enantioselectivities exhibited by the two chiral cations in the experiment (0% for 1k and 73.0% for 1g). These results were rationalized by distortion/interaction analysis, EDA, and buried volume calculations, which revealed a dispersion interaction between the bulky tert-butyl-substituted teraryl group on the outer chiral cation and methyl groups on the Bpin ligand of the anionic iridium complex responsible for enantioselectivity. Importantly, the key noncovalent interactions involved in such a challenging catalytic system (π–π, hydrogen bonding, and cation–dipole interactions between the chiral cations, anionic ligands, and substrates) were identified, leading to a greater recognition of the importance of noncovalent interactions in enantioselective reactions. These findings provide a systematic computational workflow and a clear perspective to reveal the impact of stereogenic and conformational diversity on the asymmetric induction in enantioselective reactions.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacsau.5c00830.

• Benchmark calculations to select DFT functionals for energy evaluation, two pathways for the isomerization of hydride intermediate, snapshots of MD simulations, RMSD evolves with simulation frames, TS ensembles as overlays of the TS, calculated ee values of other chiral cations, and distortion/interaction and EDA analysis (PDF)

• Cartesian coordinates for optimized geometries (XYZ)

∥.

Jian-Sen Wang and Xiao-Xia You contributed equally to this work.

The authors declare no competing financial interest.