Reaction Development and Optimization

We began our investigation by examining the cross-nucleophile coupling between alkylboronic pinacol ester 1 a and trimethylsilyl cyanide (TMSCN). We examined various copper sources, chiral bis(oxazoline) (BOX) ligands, chemical oxidants, and solvents (see Supporting Information, Table S1 for full optimization details). We identified a combination of CuOAc and ligand L1 as the optimal catalyst mixture, which provided cyanation product 2 a with the highest yield and enantioselectivity (entries 1–4, Table 1).

Table 1. Optimization of the Enantioconvergent Deborylative Cyanation of Alkylboronic Pinacol Ester 1a.^a

• ^aReaction conditions unless otherwise noted: 2-(1-([1,1′-biphenyl]-4-yl)ethyl)-4,4,5,5-tetramethyl-1,3,2-dioxaborolane (1 a, 0.10 mmol), trimethylsilyl cyanide (TMSCN, 0.15 mmol), aniline (0.10 mmol), specified catalyst mixture, 1-fluoro-3,3-dimethyl-1,3-dihydro-1 l3-benzo[d][1,2]iodoxole (3, 0.25 mmol), and benzene (0.2 M); reaction yields were determined by ¹H NMR spectroscopy of the crude reaction mixture using 1,1,2,2-tetrachloroethane as an internal standard (see the Supporting Information for details). The enantiomeric ratio (er) of product 2 a was determined by chiral high-performance liquid chromatography (HPLC) analysis. ^bSolid-state molecular structure of P1 with thermal ellipsoids at 50 % probability level. Color Scheme: Cu, gold; N, blue; O, red; C, gray; H, white.

Among the oxidants tested, commercially available fluorinated hypervalent iodine(III) reagent 3 afforded the highest yield compared to other commonly employed oxidants in radical relay reactions (entries 5–6 and Table S1).¹⁶ We anticipate that the Lewis basic additive is vital in modulating the single-electron reactivity of alkylboron reagents.^{14b, 17} A range of substituted anilines and neutral N-lone-pair donors such as DMAP and pyridine were tested as the Lewis basic additive.^14a The highest yield and enantioselectivity were obtained with a stoichiometric amount of aniline (entries 11–15). To simplify the reaction protocol, a precatalyst (L1)CuOAc (P1) was prepared and utilized in subsequent experiments. The use of P1 afforded 2 a in similar yield and enantioselectivity to that obtained using a mixture of CuOAc and L1 (entry 16).

Substrate Scope of the Enantioconvergent Deborylative Cyanation Protocol

With the optimized reaction conditions identified for the formation of 2 a, we assessed the functional group tolerance and scope of compatible alkylboronic pinacol esters (Table 2). The reaction proceeds effectively with substrates bearing both electron-donating and –withdrawing functional groups. A range of heterocycles commonly found in pharmaceuticals were also well-tolerated, including pyridine (2 g, 2 t), quinoline (2 l), thiazole (2 h, 2 j), benzoxazole (2 c, 2 x), dibenzothiophene (2 b), pyrrole (2 q), indazole (2 w), benzothiophene (2 i), benzothiazole (2 k, 2 u), dibenzofuran (2 e), and oxetane (2 r).¹⁸ The absolute configuration of the major enantiomer was determined by obtaining the solid-state structures of products 2 b and 2 u (Figure S6).

Table 2. Substrate Scope of the Enantioconvergent Deborylative Cyanation Reaction.^a

• ^aAll yields were obtained using 0.50 mmol of alkylboronic pinacol ester substrates unless otherwise noted. Enantiomeric ratio (er) was determined by chiral HPLC analysis. ^bReaction carried out on a 5.00 mmol scale of substrate 1 a. ^cSolid-state molecular structure of 2 b with thermal ellipsoids at 50 % probability level. Color scheme: S, yellow; N, blue; C, gray; H, white.

Benzylic boronic pinacol esters featuring ortho- (2 n, 2 o, 2 v), meta- (2 f, 2 o, 2 v), and para- (2 a, 2 j, 2 m) substituents are all suitable coupling partners. Cyanation of substrates containing electron-deficient aza-heterocycles (2 g, 2 t, 2 w) proceeds effectively with no detectable side products resulting from radical addition. The protocol is compatible with weak allylic (2 m) and benzylic C−H bonds (2 o), offering complementary reactivity to C−H functionalization strategies.^{15a, 19} To further highlight the synthetic utility of the enantioconvergent cyanation protocol, the synthesis of 2 a was carried out on a 5.0 mmol scale, resulting in comparable yield and enantioselectivity to the 0.5 mmol scale reactions.

Having implemented the new strategy to employ alkylboronic esters in enantioconvergent cyanation, we next sought to delineate the reaction mechanism and lay the groundwork for our future expansion of the oxidative cross-coupling platform. Specifically, through a combined experimental and computational approach, we are interested in (1) probing our hypothesized aniline-assisted C−B bond homolysis for alkyl radical generation, (2) examining the electronic structure features of the reactive organometallic intermediate that enables the cross-coupling reactivity, and (3) elucidating potential pathway²⁰ for transition metal-mediated radical functionalization.²¹

Mechanistic Studies of Aniline-Assisted Alkyl Radical Generation

Cyanation of radical clock substrate 4 led to the formation of both direct cyanation product 5 a as well as ring-opening product 5 b (Figure 2A), suggesting the intermediacy of an alkyl radical. Subjecting enantiomerically enriched (R)-1 a to the cyanation conditions with either (+)-L1 or (−)-L1 as the ligand resulted in the formation of product 2 a with similar yields and opposite configurations (Figure 2B), indicating the loss of stereochemical information following alkyl radical formation.

Electrochemical studies were carried out to examine the mechanism of aniline-assisted C−B bond homolysis. 1,2-Difluorobenzene was used as the solvent for its ability to effect the cyanation, albeit with a lower yield (Table 1, entry 8).²² To evaluate the thermodynamic driving force for the single-electron activation of 1 a, we measured the cyclic voltammogram of alkoxide 6 a and observed an irreversible oxidation event at 1.52 V (vs. SCE, Figure 2C, red trace). No oxidation event of 1 a was observed within the solvent potential window (green trace), consistent with the high oxidation potential of alkylboronic pinacol esters^13a and the necessity of a Lewis basic additive (Table 1, entry 15). Upon the addition of a stoichiometric amount of aniline (E_p=+0.40 V vs SCE),²³ an irreversible oxidation event was observed with a peak potential of +1.37 V (vs SCE, blue trace).^14b The new oxidation event could be attributed to (1) the formation of an aniline-1 a adduct¹⁴ or (2) the association of 1 a with in situ generated aminyl radical (Figure 3, vide infra)²⁴ and the subsequent cleavage of C−B bond.

The feasibility of these two pathways for alkyl radical generation was assessed through computational studies using 1-phenylethyl boronic pinacol ester (7). We first considered the formation of an aniline-7 adduct (8). Computational analysis indicated that the binding of aniline to 7 is endergonic (ΔG=12.1 kcal/mol) with an activation barrier of 14.0 kcal/mol (Scheme S5). However, single electron transfer between 8 and alkoxy radical 6 b proceeding C−B bond homolysis is highly thermodynamically unfavorable (ΔG=52.6 kcal/mol, Scheme S6). Alternatively, the formation of an aminyl radical 9 through hydrogen atom transfer (HAT) to 6 b was computed to be exergonic (ΔG=−8.5 kcal/mol) with a small kinetic barrier (ΔG^≠=6.8 kcal/mol, Figure 3). Association of the resulting aminyl radical 9 with 7 leads to adduct 10 (ΔG=+3.5 kcal/mol) from which homolytic substitution (S_H2) can occur easily. We found a flat potential energy surface (PES) in which this occurs in a formal stepwise fashion, albeit with a shallow intermediate that is tetravalent at the boron center (10). Subsequent facile C−B bond scission (ΔG^≠=0.5 kcal/mol) from 10 produces alkyl radical 11 in a highly exergonic fashion (ΔG=−29.6 kcal/mol). The computed low barriers of this pathway are also consistent with the reported alkyl radical generation from reactions between alkylboronic pinacol esters and aminyl radicals.^13c

Spectroscopic Characterization and Electronic Structure Analysis of Reactive Organometallic Intermediates

Upon formation of the prochiral alkyl radical, we hypothesized that a ligated cupric dicyanide complex (L1)Cu(CN)₂ (14) enables the enantioselective cyano group transfer.^{15a, 25} We propose that the highly covalent nature of Cu(II)–CN bond distributes the open-shell character across both the metal center and cyano ligand, enabling its single-electron cyano group transfer reactivity. Intermediate 14, however, has yet to be spectroscopically characterized. In order to access 14, complex (L1)CuOAc (P1) and (L1)Cu(OAc)₂ (15, Figure S8) were prepared and structurally characterized. Exposing P1 to stoichiometric TMSCN in benzene resulted in a white suspension over two minutes (Figure S3). We propose that the white precipitate is (L1)Cu(CN) (16) in the form of an insoluble coordination polymer.²⁶ Infrared spectroscopy measurement of the solid product revealed a C≡N stretch of 2102 cm⁻¹, consistent with previously characterized four-coordinate Cu(CN) coordination polymers.²⁷ The sequential addition of 3 and TMSCN to 16 or treatment of 15 with two equivalents of TMSCN instantaneously resulted in a dark purple solution, potentially due to the formation of 14. A similar observation of the formation of a purple species was reported in the synthesis of complex (bpy)Cu(CN)₂.²⁸ The paramagnetic species 14 is highly unstable even at −78 °C (in toluene or 2-methyltetrahydrofuran), with the purple solution effervescing and changing into a white suspension within seconds.²⁹ IR measurement of the resulting white precipitate confirmed its identity as 16. This observation is consistent with the redox buffering ability of the cyanide nucleophile in facilitating radical relay reactions.³⁰ The thermal instability of 14 limited our attempts at its isolation and structural characterization. Instead, 14 was characterized using EPR spectroscopy under cryogenic conditions, and its electronic structure was probed using DFT calculations. The X-band CW-EPR spectrum of 14 collected at 77 K in toluene revealed an S=¹/₂ spin state with the unpaired electron occupying a pseudo-axial environment (Figure S4, g₁=2.306, g₂=2.083). A hyperfine constant of A_1,Cu=379.3 Hz was derived from the g₁=2.306 signal. EPR spectroscopic analysis of a catalytic reaction mixture after 30 min revealed a similar signal. Computational DFT studies corroborated these experimental data, indicating that the formation of 14 from either 15 in the presence of two equivalents of TMSCN or the sequential addition of 3 and TMSCN to 16 is thermodynamically favorable (ΔG=−12.8, −20.4 kcal/mol, respectively. Schemes S13 and S14).

In an open-shell system, ligand-based spin density contributes to a range of radical-type reactivities, such as the H-atom abstraction and radical rebound in C−H hydroxylation catalyzed by cytochrome P450.³¹ We hypothesized that the facile radical cyano group transfer reactivity of 7 similarly results from a significant degree of ligand-based radical density. From a molecular orbital perspective, the unpaired electron in 14 resides in a singly occupied molecular orbital (SOMO) with contributions from both the metal 3d and carbon 2p orbitals. To gain insight into the electronic structure and frontier molecular orbitals of the Cu(II) biscyano complex 14, particularly the distribution of the unpaired electron across the Cu−CN σ-bond, natural bonding orbital (NBO)³²-based natural population and intrinsic bonding orbital (IBO)³³ analyses were performed. Natural population analysis of 14 indicated that the unpaired electron in the SOMO is delocalized across the central Cu atom (0.513) and the carbons of the two cyano ligands (0.365, see Scheme S15). Effective oxidation state (EOS)³⁴ calculations for 14, performed at the B3LYP(D3-BJ)/def2-TZVPPD level of theory, closely corroborated these data showing a partial spin delocalization on Cu (0.569) and the cyano carbons (0.310) (Figure S20). NBO-based orbital composition analysis of the highest occupied alpha SOMO of 14 features a significant contribution from cyanide carbon atoms (44.0 %) compared to the copper center (14.9 %) (Figure 4), a result from the highly covalent nature of the Cu−C bond.

Theoretical Calculation of the Cu-Catalyzed Cyanation Pathway

Computational studies were next carried out to better understand the proposed mechanism of this transformation and to elucidate the origins of enantioconvergence. Dispersion-corrected DFT calculations were performed at the B3LYP-D3(BJ) level of theory with 6-311+G(d,p)//LANL2DZ (F, I)//LANL2TZ (Cu) basis sets using an implicit SMD model of benzene.³⁵ We modeled the cyanation of 1-phenylethyl boronic pinacol ester (7) with ligand (+)-L1. DFT calculations, in conjunction with experimental studies, indicated that the formation of catalytically active 14 was expected to proceed exergonically from complex 16 (ΔG=−25.9 kcal/mol, Figure 5). A stepwise pathway was computed. The oxidation of 16 by fluorine atom transfer from hypervalent iodine reagent 3 furnishes complex 17 and oxygen-based radical 6 b (ΔG=−10.8 kcal/mol).³⁶ Subsequent ligand exchange between 17 and TMSCN forms 14 in an exergonic fashion (ΔG=−15.1 kcal/mol) through a four-membered ligand exchange transition state TS4.

DFT calculations indicated that the formation of a formal Cu(III) intermediate can occur in one of two binding modes, with 11 assuming either a pseudoequatorial or -apical position with respect to the BOX ligand (Figure 5). Conformational analysis of both binding modes revealed that placing the benzylic group in the less-congested equatorial position and the electronegative cyano groups in the apical positions as in 18 and 19 is more favorable (Scheme S3). In contrast, addition of 11 along the apical direction of 14 results in more thermodynamically disfavored intermediates (ΔΔG_pro-R=6.9 kcal/mol, ΔΔG_pro-S=9.1 kcal/mol, Scheme S4). This is likely the result of (1) a more sterically congested coordination environment and (2) the strong trans effect between the cyano ligands and alkyl groups. Transition states corresponding to the formation of 18 and 19 cannot be located despite attempts at various levels of theory and broken symmetry calculations.³⁷ The relaxed potential energy surface scan by modulating the Cu−C distance showed no inflection point, consistent with previous studies (see Supporting Information for details).^37b

The subsequent C−C bond formation was calculated to occur via reductive elimination transition structures (TSs) TS5 (S) and TS5 (R) with retention of configuration at the benzylic carbon. Assuming reversible formation of the diastereomeric formal Cu(III) complexes 18 and 19, the reductive elimination is enantiodetermining. Application of the Curtin Hammett principle results in a ΔΔG^≠ of 1.3 kcal/mol in favor of the major R-enantiomer, consistent with our experimental observation.^{15a, 37b} We ascribe the preferential formation of the R-enantiomer to additional stabilizing dispersion interactions in the reductive elimination TSs (Figure S23). Direct benzylic radical attack on the cyano groups in 14 following an outer-sphere pathway was also probed computationally. The result reveals a diminished ΔΔG^≠ of 0.6 kcal/mol favoring the R-enantiomer and suggests that the inner-sphere pathway is kinetically more favorable (see Supporting Information for details).

Our combined experimental and computational mechanistic studies led to the following proposed catalytic cycle (Figure 6). Single electron oxidation of Cu(I) catalyst 16 generates concurrently alkoxy radical 6 b and Cu(II) fluoride complex 17. Hydrogen atom transfer between 6 b and aniline produces aminyl radical 9, while transmetalation between 17 and TMSCN furnishes reactive Cu(II)(CN)₂ intermediate 14. The association between aminyl radical 9 and substrate 7 followed by C−B bond homolysis leads to the prochiral alkyl radical 11, which subsequently reacts with Cu(II)(CN)₂ intermediate 14 to generate enantiomerically enriched product 2 s.