The metal-catalyzed C-H borylation reaction has received significant attention over the past two decades based on the e ciency and utility of converting unreactive C-H bonds into versatile and reactive C-B bonds. [1][2][3][4] Early catalytic reactions provided sterically controlled regioselectivity: [5][6][7] In the case of alkanes, primarily ruthenium and rhodium catalysts have been used to mediate the borylation of alkane C-H bonds, which is selective for primary C-H bonds. 8 For arenes, iridium has been employed most consistently, providing selectivity for C-H bonds without neighboring substituents. 1,3-Disubstituted and 1,2-symmetrically disubstituted arenes provide high regioselectivity, along with heteroarenes when there is a particularly reactive C-H bond. 6,9 Since 2008, a significant focus of C-H borylation methods has been on the use of a directing group (DG) to provide complementary regioselectivity to the nondirected systems. 4,[10][11][12][13][14][15] The predominant approach to achieve directed C-H borylation has relied on modifying the reported active catalyst for nondirected C-H borylation, an iridium(III) trisboryl complex (I, Figure 1) with a rigid bidentate ligand (bipyridine-or phenanthroline-based). [16][17][18] The rigid struc-ture of the active catalyst limited early e orts to extend the method to directed C-H borylation based on the single open coordination site on the catalyst, which does not allow a traditional Lewis basic directing group to coordinate to the catalyst with an open coordination site remaining for C-H activation. Hartwig provided the first general example that overcame this limitation, using a silyl tether to direct ortho C-H borylation. 15,19 In this case, a silane, after a facile oxidative addition (Si-H bond)/reductive elimination (B-H bond) sequence, replaces one of the boryl ligands, maintaining the necessary reactivity in C-H borylation (II, Figure 1).
Silyl-directed C-H borylation initiated a wave of approaches to achieve directed C-H borylation, 4,10,12 including those that are directed to meta [20][21][22][23][24][25][26][27][28] or para C-H bonds. [29][30][31][32][33][34] While the diversity of approaches cannot be detailed in this report, 4 some key approaches to ortho C-H borylation deserve mentioning. Sawamura developed solid-supported monodentate phosphine ligands which are believed to force the active catalyst to maintain two open coordination sites (III, Figure 1), providing a variety of directing groups that can mediate ortho C-H borylation. [35][36][37] Smith and Maleczka 38 and Li 39 developed rigid bidentate pro-ligands with a cleavable bond (e.g., Si-H or Si-B), which replaces one of the X-type boryl ligands on iridium (IV), opening an additional coordination site for Lewis basic directing groups. Chattopadhyay recently reported alternative pro-ligands that undergo cyclometalation with iridium to form C-/N-bound XL-type ligands while keeping two open coordination sites for directed C-H borylation. 40,41 Our group reported a parallel approach that uses a cationic iridium complex to open an additional coordination site for phosphine-directed C-H borylation (V). [42][43][44][45][46][47] Smith and Maleczka have also developed outer-sphere hydrogen bonding with the N-H of aniline carbamates (VI) 48,49 which has led to the use of a number of noncovalent interactions to achieve directed C-H borylation. 12,50,51 Lassaletta [52][53][54] and our group 55,56 have shown that hemilabile ligands can be used to temporarily open a coordination site for hydrazone-and amine-directed C-H borylation (VII).
In spite of the flurry of directed C-H borylation methods over the past decade, there has not been a comprehensive mechanistic study completed on any of the systems described above. Thus far, mechanistic work has been primarily theoretical, providing limited insights on a subset of 40,50,[57][58][59][60][61][62][63] We became interested in providing a greater understanding of this growing area of directed C-H borylation by conducting a full investigation into the mechanism of amine-directed C-H borylation using hemilabile ligands. 55,56 There have been two studies of directed C-H borylation that provide a context for our mechanistic investigation. The first one is a reaction that was studied computationally based on an ester-directed C-H borylation reported by Miyaura, which utilizes an electron-deficient monodentate phosphine, tris [3,5-bis(trifluoromethyl)phenyl]phosphine with an [Ir(μ-OMe)(COD)] 2 precatalyst, to allow an ester to coordinate to iridium and maintain an open coordination site for C-H activation. 64 Jover and Maseras completed a detailed computational study that was consistent with the reactivity observed by Miyaura. 60 The second study was published by Sunoj and Chattopadhyay, 65 in which the reported transformation involved a ligand-free system for pyridine-directed C-H borylation of arenes. In this latter case, some experimental mechanistic studies were coupled with computational work to provide a more detailed picture of the mechanism.
Seminal studies on the mechanism of nondirected, iridiumcatalyzed arene C-H borylation have been done computationally by Sakaki 66 and experimentally by Hartwig. 16 A simplified account of the catalytic cycle is provided in Scheme 1a, which highlights the rate-determining C-H activation of the arene by the iridium(III) trisboryl active catalyst. [16][17][18] In addition to the kinetic and computational evidence for this Scheme 2. Context of C-H Borylation with No or Inverse Isotope E ect and Kinetic Dependence on Concentration of the Boron Sources turnover-limiting step, the kinetic isotope e ect (KIE) was determined by examining the catalytic C-H borylation of benzene and benzene-d 6 with a preformed iridium(III) trisboryl complex, running the two reactions in separate vessels to provide a large k H /k D of 5.0 (Scheme 2a). Stoichiometric reactions with the same catalyst gave a similar k H /k D (4.6). Two additional KIE studies were done by Hartwig on di erent types of substrates. The rate of C-H borylation of pyridine was compared to pyridine-d 5 , providing a k H /k D of 2.9 for borylation in the 3-position and 2.4 in the 4position (Scheme 2b). 67 The benzylic C-H borylation of toluene was also compared to toluene-d 8 , resulting in a k H /k D of 2.2 (Scheme 2c). 68 In the recent work by Sunoj and Chattopadhyay on a ligand-free pyridine-directed C-H borylation, the intermolecular KIE depended on the extent of conversion and the precatalyst utilized, but nevertheless fell in the range of 0.97-1.18 (Scheme 2d). 65 It is important to note that, in all reported kinetic studies, the reaction rate is independent of the concentration of the boron source. 16,67,68 Directed C-H borylation necessitates alternative mechanistic descriptions that account for the regiochemistry of the transformation. In the case of amine-directed C-H borylation, our initial proposed mechanism involved partial dissociation of a hemilabile diamine ligand. 55,56 Both experimental and computational experiments support this model and provide a detailed picture of the catalytic cycle that does not involve turnover-limiting C-H activation. Instead, isomerization of the C-H activated complex is the turnover-limiting step (see Scheme 1b for a summary of the proposed catalytic cycle). We herein report a detailed mechanistic study of this system that involves both experimental and computational insights into the amine-directed C-H borylation reaction. Turnover-limiting reassociation of pinacolborane (HBpin) has the highest free energy of activation in the regeneration of the active catalyst. HBpin, which is formed during the course of the reaction, also serves as an autocatalyst involving the slowest step. This study is highlighted by an unexpected inverse isotope e ect on the activated C-H bond and no isotope e ect in an intramolecular KIE study (Scheme 2e). Furthermore, di erent from previous kinetic studies, we provide an example in which a first-order dependence is observed for B 2 Pin 2 , which indicates that the directing group can alter the mechanism of C-H borylation.
All procedures involving air-or moisture-sensitive reagents were performed in oven-or flame-dried glassware and under purified nitrogen, either in an inert atmosphere glovebox or by standard Schlenk techniques. In all procedures, unless otherwise noted, the concentration was performed by rotary evaporation or by subjecting the material to high vacuum using a Schlenk line. Thin layer chromatography (TLC) analysis was performed on Whatman 60 Å silica layer fluorescence ultraviolet (UV) plates. Flash column chromatography was carried out on hand-packed columns of silica gel, 40-63 μm, 60 Å. NMR spectra were collected on a UNITY Inova spectrometer at 500 or 400 MHz for 1 H NMR, 125 or 100 MHz for 13 C NMR, 128 MHz for 11 B NMR. The 11 B NMR spectra were typically processed by adding 15 points of backward linear prediction to remove the glass peaks from the broadband probe and 5 Hz of apodization was applied. 1 H NMR spectra are referenced to CDCl 3 at 7.26 ppm, C 6 D 6 at 7.16 ppm, tol-d 8 at 2.08 ppm, or to an internal tetramethylsilane (TMS) standard at 0.00 ppm. The 1 H NMR spectral data are reported as follows: chemical shift, parts per million; multiplicity (s = singlet, br s = broad singlet, d = doublet, t = triplet, q = quartet, qn = quintet, hex = hextet, sep = septet, oct = octet, m = multiplet), coupling constants (Hz), and integration. 13 C NMR spectra are referenced to CDCl 3 at 77.0 ppm or C 6 D 6 at 128.0 ppm. Carbon atoms attached to boron are typically not observed and accounted for in the analysis. 11 B NMR spectra were referenced to an external BF 3 • Et 2 O sample in CDCl 3 or C 6 D 6 (0.0 ppm). High-resolution mass spectrometry was obtained at the University of California, Irvine Mass Spectrometry Facility. All solvents were dried and degassed by standard procedures unless employed for extraction or purification. Dioxane and toluene were obtained from a solvent purification system, degassed by freeze-pumpthaw cycles (×3), and stored in the glovebox. Tetrahydrofuran was obtained from a solvent purification system, dried over sodium metal and benzophenone, distilled, followed by freeze-pump-thaw cycles (×3), and stored in the glovebox. Benzene-d 6 was distilled from CaH 2 , degassed by freezepump-thaw cycles (×3), and stored in a glovebox. Chloroform-d was distilled from MgSO 4 , degassed by freeze-pumpthaw cycles (×3), and stored in a glovebox. Bis(pinacolato)diboron, pinacolborane, di-μ-methoxobis(1,5-cyclooctadiene)diiridium(I), 2-benzylaminopyridine, phenyltrimethylsilane (PhTMS), tert-butyllithium solution (1.9 M in pentane), dimethylamine solution (40% in water), and all required benzyl bromides were purchased and employed without purification. Substrate 3 and aryl boronate ester 4 were synthesized as previously reported and matched spectral data. 55 General Procedure for Kinetics Studies. All stock solutions used for kinetics were made with toluene-d 8 in the glovebox. Stock solutions of the di-μ-methoxobis(1,5cyclooctadiene)diiridium(I) precatalyst, borylation substrate (with PhTMS as an internal standard), B 2 pin 2 , and 2benzylaminopyridine were made using 1, 2, or 5 mL volumetric flasks. The solution of the precatalyst was made fresh before any kinetics experiments, as old solutions showed drastic and unpredictable changes in reaction rate. The stock solutions were made to the concentrations shown in Table S1 unless the total volume of the reaction mixture exceeded 0.600 mL. In those cases, the ligand solution was made more concentrated. The PhTMS internal standard was always included in the substrate solution to minimize addition errors.
Kinetics trials were done in triplicate or quadruplicate using the corresponding number of NMR tubes equipped with Teflon screwcaps (J-Young tubes). Microliter syringes were used to add stock solutions of each reagent to the J-Young tubes. Stock solutions were always added in the order: substrate 3 (with PhTMS internal standard), iridium precatalyst, 2-benzylaminopyridine, B 2 pin 2 , HBpin (if added), and toluene-d 8 to make up a total volume of 0.600 mL. Once mixed, the samples were sealed with Teflon screwcaps and removed from the glovebox. The samples were then placed into a 70 °C oil bath, which was maintained by a thermocouple stirring plate. A sample holder was made from a block of Teflon with 4 holes drilled to accommodate the NMR tubes. This sample holder allowed for operationally simple and reproducible placement of the samples in the oil bath. The samples were heated in the oil bath for a monitored amount of time. At the end of each heating interval, the samples were removed from the oil bath and promptly placed in an ice bath to halt the progression of the reaction while NMR spectra were acquired.
Computational Methodologies. All reported geometries [minima and transition states (TSs)] were fully optimized utilizing the Gaussian 09 D.01 69 implementation of the PBE0 70 density functional with the D3 version of the Grimme 71 dispersion correction with Becke-Johnson damping with basis set BS1 (where BS1 is the electron core potential and modified LANL2DZ basis set [72][73][74] for iridium and the 6-31G(d′) basis sets [75][76][77] for the remaining atoms). TSs were located by a combination of direct TS and quadratic synchronous transit QST3 techniques followed by confirmation of minimums, which along with atomic displacements along the imaginary modes ensure located TSs connect to corresponding minima.
Exploring the e ects of solvent, in select cases, single-point solvation computations were also performed on gas-phase geometries utilizing the solvation model based on density (SMD) 78 implicit solvation model via the self-consistent reaction field (SCRF) with parameters consistent with toluene modeled as the solvent.
Hopmann has previously shown that the pure PBE functional 79,80 combined with D2 dispersion correction 81 and the integral equation formalism variant of the polarizable continuum solvation model (IEFPCM, commonly referred to as PCM) [82][83][84] produces the most accurate results for iridium organometallic reaction chemistry. 85 Hopmann also found that when increasing the basis set size from 6-311G(d,p) to 6-311+G(2d,2p) for nonmetallic atoms and LanL2DZ(f) to LanL2TZ(f) for iridium does not improve the results, in fact, it even worsens them! Therefore, single-point energy computa-tions with this recommended functional/basis set combination were performed to study the activation free energy of the catalytic process. Due to the size of the system, we use PBE0-D3/BS1 to optimize gas-phase geometries of TDI and TDTS:PCM-PBEPBE-D2/BS2//PBE0-D3/BS1 in Scheme 15. BS2 is the LANL2DZ valence basis set with one fpolarization function 86 for iridium and the LANL2DZ e ective core potential and the 6-311G(d,p) basis sets for the remaining atoms. As a matter of convenience, in selected cases in the text, the Pople notation has been truncated; e.g., SMD//PBE0-D3/ BS1 explicitly means SMD(Toluene)-PBE0-D3/BS1//PBE0-D3/BS1.
A summary of the experimental and computational mechanistic study is shown in the catalytic cycle depicted in Scheme 3. The directed borylation by the iridium catalyst begins at agostic complex A where the substrate coordinates to the metal center. Utilizing the p orbital of the boron center in Bpin, the complex undergoes a σ-bond metathesis step to activate the C-H bond and produce complex B. An isomerization process via dissociation of HBpin (B to C) followed by reassociation of HBpin (C to D) a ords the appropriate proximity between the B pin and C aryl ligands for C-B bond formation. The association of HBpin with C was found to be the TDTS (TOF determining transition state from the energetic span model). 87 Then, another σ-bond metathesis occurs to form the C-B pin bond to convert isomer D to the aryl borate product E. Next, a ligand exchange process with bis-Scheme 3. Proposed Mechanism for the Iridium-Catalyzed C-H Borylation of Benzylic Amines (pinacolato)diboron (B 2 Pin 2 ) makes complex F. Finally, the tris-Bpin iridium species is regenerated by another σ-bond metathesis to yield complex G, a geometry with σ-bound HBpin, which was found to be TDI c (TOF determining intermediate catalyst from the energetic span model). From complex G, either another equivalent of substrate can replace HBpin and start a new cycle or B 2 Pin 2 can coordinate to the catalyst and produce stable complex H, which stays outside of the catalytic cycle and inhibits the reaction. Our experimental and computational evidence supporting the proposed mechanism will be discussed.
Attempted Isolation of Active Catalyst. Attempts to isolate an active catalyst for amine-directed C-H borylation were previously reported by our group as unsuccessful. 56 Addition of HBpin or B 2 pin 2 and 2-benzylaminopyridine (the optimized ligand) to [Ir(μ-OMe)(COD)] 2 (complex 1) in various orders of addition, as done by Ishiyama, Miyaura, and Hartwig, 16,88 and Smith and Maleczka 89 to synthesize iridium trisboryl complexes for nondirected systems, 16,88,89 provided complex mixtures impeding the synthesis of an active catalyst species (see the Supporting Information for spectra). Addition of 2-benzylaminopyridine to [Ir(μ-Cl)(COD)] 2 provided quantitative yield of square planar iridium(I) complex 2 (Scheme 4), 90 but attempts to form a boron-containing complex by the addition of HBpin or B 2 pin 2 were unsuccessful. Complex 2 did, however, provide catalytic activity comparable to the separate addition of [Ir(μ-Cl)(COD)] 2 and 2benzylaminopyridine in the presence of a boron source. 56 Going forward, [Ir(μ-OMe)(COD)] 2 was used as the precatalyst in all kinetic studies as it provides increased reaction e ciency over [Ir(μ-Cl)(COD)] 2 as the precatalyst. Our computational study also shows that [Ir(μ-Cl)(COD)] 2 leads to a more favorable isolated complex than the more common methoxy-substituted precatalyst [Ir(μ-OMe)-(COD)] 2 , see Scheme S9 in the SI. The reaction order with respect to B 2 pin 2 was interrogated by examining the kinetic profile at 0.2500, 0.3750, 0.5000, and 0.6250 M of [B 2 pin 2 ]. As shown in Figure 2, the rates increased at a first-order dependence on the B 2 pin 2 concentration, demonstrating that the rate law is first order in B 2 pin 2 through this concentration regime.
Identical concentration data points were obtained with varied concentrations of substrate 3 (0.2500 to 0.6250 M) and a 0.2500 M [B 2 pin 2 ] to examine the rate dependence on the substrate (Figure 3). In this case, the reaction is first order at low concentrations of substrate but results in saturation at higher concentrations of 3.
The dependence of the rate on the catalyst loading was then examined. The 2:1 ratio of 2-benzylaminopyridine/1 was maintained with each kinetic run, and the concentration of 1 was varied from 0.0025 M (2 mol % Ir) to 0.0125 M (10 mol % Ir). The reaction is first order in precatalyst concentration, shown in Figure 4 used as a catalytic additive, or as the sole boron source. 2,4,42,91 The reaction progress was therefore examined with varying combinations of B 2 pin 2 and HBpin (Figure 5). Each reaction was heated to 90 °C followed by 1 H NMR spectroscopy using the method described for the kinetics experiments. The standard kinetics run (1 equiv of B 2 pin 2 ) was followed for 250 min, resulting in a 70% NMR yield of 4. Under these conditions, the 11 B NMR spectrum of the reaction mixture confirmed the presence of HBpin at 28.1 ppm. Over the course of the reaction, the formation of hydrogen gas was also observed in the 1 H NMR spectrum at 4.47 ppm (see the Supporting Information for 1 H and 11 B NMR spectra of the reaction over the course of the reaction). Furthermore, the presence of two metal hydride peaks was observed at -11.8 ppm (-11.79 and -11.85 ppm). We believe that these peaks represent σ-complexes of HBpin to iridium (such as B, D, and G in Scheme 3) and are consistent with an analogous σcomplex of HBpin to iridium reported by Heinekey with a chemical shift of -13.15 ppm. 92 Addition of a catalytic amount of HBpin (0.1 equiv) to the reaction mixture provided a significant rate acceleration, reaching a similar amount of product formation in 80 min (66% NMR yield) and reaching 86% NMR yield after 250 min. The addition of a full equivalent of HBpin to the reaction mixture further accelerated the reaction, reaching 67% NMR yield in 60 min, and proceeding to ≥95% NMR yield in 140 min. The necessity of B 2 pin 2 was then probed by adding HBpin as the sole boron source. With both 1 and 2 equiv of HBpin, the reaction was ine ective. Both reactions plateaued, resulting in 10% (1 equiv HBpin) and 22% NMR yield (2 equiv HBpin) after 200 min at 90 °C. Notably, when B 2 Pin 2 is used as the sole boron source, there is an induction period that extends through 1200 s (see expansion in Figure 5). The induction period could result from the conversion of the dinuclear methoxide precatalyst to the active mononuclear iridium boryl species. 16,93 ■ COMPUTATIONAL STUDIES Computed Energy Profile. The proposed mechanism for the amine-directed C-H borylation was derived from the results of density functional theory (DFT) computations. The system was modeled with benzylamine 5 as substrate, which produces the corresponding boronate ester 6 (Scheme 6).
Five key stages were identified in the reaction mechanism for the formation of the aryl boronate ester product (see Scheme 7): (1) C-H bond cleavage (complex 7 to complex 8); (2) rate-determining isomerization of the iridium complex through dissociation and reassociation of HBpin (complex 8 to complex 10); (3) C-B bond formation (complex 10 to complex 11); (4) product dissociation (complex 11 to complex 14); and (5) catalyst regeneration (complex 14 to complex 15). The rate-determining association of HBpin with complex 9 (resulting from the dissociation of HBpin from complex 8) is distinct from other experimental and computational work that typically involves rate-determining C-H activation. In this case, however, the σ-complex that results from C-H activation (i.e., complex 8) does not have a lowenergy C-B bond-forming step that is energetically accessible. The lowest-energy pathway requires dissociation of the HBpin (complex 8 to complex 9), followed by reassociation in a location trans to the newly formed iridium-carbon bond (complex 9 to complex 10). This isomerization allows for a low-energy transition state in forming the C-B bond. 94 C-H Bond Cleavage. In previous computational studies of C-H borylation reactions, the key step is often the activation of the C-H bond. For example, Sakaki and co-workers reported a computational study on the iridium-catalyzed borylation of benzene with B 2 pin 2 in which the C-H bond is broken in a rate-determining oxidative addition step. 66 Yao and co-workers also found that C-H oxidative addition step is the most energetically demanding process with a bulky ligand, leading to meta regioselectivity. 59 Both of these examples lack a directing group, which has the potential to alter the energy profile through the precoordination of the substrate to the iridium complex. The recent mechanistic work by Sunoj and Chattopadhyay, however, included computations for pyridinedirected C-H borylation. In their system, C-H activation was also shown to be the rate-determining step. 65 Examination of complexes resulting from coordination of the benzylic amine to the iridium(III) trisboryl complex expected in this system resulted in low-energy complex 7 with an agostic interaction to the ortho C-H bond. The subsequent C-H activation step is no longer the highest point in the energy profile (Scheme 7), with a relative free energy of activation (Δ ‡ G°) of approximately 16.7 kcal/mol (for TS-7-8) Scheme 6. Catalytic Reaction Used in the Computational Studies Scheme 7. Proposed Pathway for Amine-Directed C-H Borylation a a Relative Gibbs free energies in kcal mol -1 . compared to the highest free energy of activation of 20.6 kcal/ mol (vide infra). A competing N-H bond activation was also considered but was found to have a significantly higher free energy of activation via TS-7-17, (Δ ‡ G°= 25.8 kcal/mol; see S3.2 in the SI).
The key C-H bond activation step is known to undergo some form of two limiting mechanisms: (1) standard oxidative addition of the C-H bond, resulting in an iridium(V) complex; 66,68 (2) a σ-bond metathesis step that involves an interaction of the boron p-orbital, generating an iridium(III) complex with a B-H σ-complex. 68 In this case, the computed transition state proceeds through a σ-bond metathesis step, resulting in the direct formation of complex 8 (Scheme 8). The di erence between metathesis and oxidative addition in this situation is a fundamental conceptual di erence. If Ir(III) is oxidized to Ir(V) in the C-H bond cleavage step then the hydrogen bound to the iridium center would be a formal hydride ligand. However, in this case, the hydrogen is transferred through a σ-bond metathesis from the agostic C-H bond transferred directly to a σ-bound HBpin. Although most mechanistic studies for C-H borylation conclude an Ir(V) intermediate through standard oxidative addition, a closer look at the calculations is not decisively in favor of this mechanism for C-H activation. For example, Sakaki and coworker calculations show that pentaboryl Ir(V) complex has one set of Bpin ligands that are closer in distance to each other than to the other Bpin group. Between those two boron atoms, d(B-B) is 2.1 Å and is significantly shorter than the other pair of cis Bpin ligands, d(B-B) from 2.5 to 2.8 Å. 66 Isomerization. The typical mechanism for C-B bond formation in nondirected C-H borylation is reductive elimination from an iridium(V) complex. 16 The distinct feature in C-B bond formation in this system seems to result from the geometrical constraints imposed by the directing group, and therefore an isomerization process is required to accommodate the interaction between C aryl and the boron of Bpin. The directing group makes the architecture more rigid and prohibits the formation of the C-B bond with the axial Bpin group due to a lack of proper orbital overlap without straining the five-membered ring metallacycle (Scheme 9). 95 Further attempts to force this C-B bond formation from complexes 8 and 10 in a constrained potential energy surface also show these interactions are not accessible (see S3.3 in the SI).
Complex 10 was identified as the required isomer to achieve C-B bond formation with an accessible free energy of activation. To access isomer 10 (HBpin in the distal equatorial position) from complex 8 (initially formed upon C-H activation), several pathways were considered. Hartwig previously proposed internal isomerization by σ-bond metathesis transferring the hydride to an adjacent Bpin substituent. 16 In this case, however, the location of the hydride only allows formation of complex 18 with HBpin occupying the axial position, which has a higher free energy of activation to reductive elimination (vide infra; see Scheme 12). Hall and co-workers also reported the isomerization process by an HBpin dissociation pathway for a rigid pincer cobalt complex. 96 Applying this pathway to this system, HBpin completely dissociates from complex 8 and then recombines with a di erent orientation in order to form isomer 10. The free energy for releasing HBpin is exergonic by 6.5 kcal/mol. However, this isomerization is quite a bit more complicated and involves a series of hydride relays in the proposed mechanism that lower the apparent free energy of activation and speeds up the reaction (see S3.4 in the SI). 97 The Scheme 8. Computed Energies for C-H Activation from 7 a a Relative Gibbs free energies in kcal mol -1 .
introduction of HBpin to complex 9 produces dipole-dipole complex 9-HBpin. The transition state TS-9-10 is the highest point in the energy profile, Δ ‡ G°= 20.6 kcal/mol, and yields complex 10, in which the aryl carbon and the boron are cis, with the orbitals aligned for C-B bond formation. Overall, the isomerization from complex 8 to complex 10 has a net Gibbs free energy, ΔG°= -13.7 kcal/mol, which illustrates a thermodynamically feasible process.
Kinetic Isotope E ect Studies. An intermolecular KIE was examined with 5-deutero-2-fluoro-N,N-dimethylbenzylamine (3-d) under the same reaction conditions used for the parent kinetic analysis (Scheme 10), providing a similar rate constant [(6.95 ± 0.21) × 10 -6 M/s] to the protio sample [(6.33 ± 0.23) × 10 -6 M/s]. The resulting k H /k D of 0.91 ± 0.04 is consistent with either a small inverse isotope e ect or no isotope e ect. A small inverse KIE was also observed by Sunoj and Chattopadhyay in their system. 65 The isotope e ect in their study was attributed to the dominance of an equilibrium isotope e ect (EIE) in the preequibrilium over KIE of TDTS (TOF determining transition state), which is the C-H bond activation. There are other examples in the literature showing that EIE is the main contribution to the observed inverse KIE. 98,99 However, omitting the primary KIE contribution in the slowest microscopic steps is questionable, especially those having isotopic labeled atoms involved significantly in motion in the TDTS such as C-H bond activation. In this study, we provide another interpretation of a small KIE without the contribution of EIE and based on examining the di erent mechanism in which HBpin reassociation is the TDTS. For the simulation of the KIE, TDI (TOF Determining Intermediate, complex 15) and TDTS (TS-9-10) were used. Based on these stationary points, the intermolecular KIE simulation gives a fortuitously close k H / k D value of 0.911 when compared to 0.91 from the experiment. This value could be understood as an inverse secondary KIE. Within the TS-9-10 transition state, the formation of an agostic B-H bond occurs, thus enabling a back-bonding interaction from Iridium to the σ*(B-H). The interaction contribution from p and d orbitals results in analogous out-ofplane motion, which is similar to the well-known secondary isotope e ect in sp 2 carbon centers.
An intramolecular KIE study was examined next, which typically will provide an isotope e ect even if the C-H or C-D bond is not broken in the rate-determining step. 100 Deuterated substrate 5-d was subjected to the reaction conditions (70 °C, 2 mol % [Ir] 2 , 4 mol % ligand in toluene-d 8 ) and monitored throughout the reaction to 70% conversion of 5-d to 6 and 6-d (Scheme 11, moderate conversion was used to avoid bis-borylation products that are observed at high reaction conversion). The reaction was performed in triplicate and resulted in an intramolecular KIE of ∼1.0 at all time points examined (29, 45, and 70% NMR yield of 6 and 6-d). See the Supporting Information for details.
The fact that a di erence between the intramolecular and intermolecular KIE is observed can be explained by the turnover-limiting step occurring after the C-H bond cleavage.
This observation is consistent with the current proposed mechanism in which HBpin dissociates and reassociates (complex 8 to complex 10). Similar to the equilibrium between H-and D-substrates described in the work by Yi, 101 the recombination mechanism enables reactions to take place at hydrogen of 5-d while still exhibiting a deuterium e ect, and vice versa. As seen in Scheme 11, either HBpin or DBpin can be generated during the reaction with compound 5-d. In the reassociation step (TS-9-10, which is assigned to be the TDTS), either HBpin or DBpin can react with intermediate, complex 9, or complex 9-d. Therefore, four di erent possibilities occur and form two di erent products, 6 and 6d in the intramolecular KIE experiment. It should be noted that the kinetic di erence between substrate and deuterated substrate only comes from the HBpin or DBpin reassociation step. There is no di erentiation made between complexes 9 and 9-d because the motion in the imaginary mode of TS-9-10 involves only the hydrogen or deuterium atom in HBpin or DBpin. As a result, the overall kinetic picture to produce product 6 will be the sum of path 1 and path 2 with k = k H + k D and it is exactly the same value to create product 6-d through the sum of path 3 and path 4, k = k H + k D . Even though there is a di erence in the zero-point energies between the two isotopic compounds, this cross reaction during the reassociation of HBpin or DBpin will cancel out the KIE in the intramolecular experiment. These results further a rm the dissociation and reassociation mechanistic proposal and demonstrate the unique mechanistic feature imposed by the directing group.
C-B Bond Formation. C-B bond formation requires one of the -Bpin groups to be adjacent to the Ir-C aryl bond. The two isomers that meet this criterion with an equatorial-Bpin bond (complex 10 and complex 18), were found to undergo reductive elimination with significantly di erent free energies of activation. Reductive elimination from complex 10 (TS-10-11) has a free energy of activation of 13.9 kcal/mol, whereas reductive elimination from complex 18 has a free energy of activation of 24. The products (complexes 11 and 19) of the C-B bond formation transition states are also quite dissimilar. Complex 11 has an unusual stabilization of the bound product, with the metal-bound hydride strongly interacting with the boron of the product, essentially forming a borate with the B-H σ-bound to the iridium. From complex 11, a subsequent transition state is required to break this boron-hydrogen interaction to form complex 12, which has a nitrogen-bound product and a terminal hydride trans to the nitrogen of the pyridyl ligand (a hydride trans to the pyridyl nitrogen is lower in energy than other isomers here and throughout the computations). This spatial arrangement is consistent with the work of Fujita 102 and likely governed by the relative order of trans influence for the ligands involved: Bpin ≥ hydride ≥ pyridine ≥ tertiary amine. 103 Complex 19 has a nitrogen-bound product and a σbound H-Bpin.
Product Dissociation. Upon formation of complex 11 by reductive elimination, the desired C-B bond is installed, but the product remains bound to the iridium metal through both the nitrogen directing group and the newly installed Bpin group. Simple dissociation of the amine product would produce the unsaturated species (20, Scheme 13), which has a high free energy, ΔG°= 25.1 kcal/mol (see S3.5 in the SI) and, by necessity, an even higher free energy of activation. If this pathway were the only one for product formation, this TS would be the highest free energy point and would be the ratedetermining step. Upon further examination, however, interactions with the solvent, substrate, or ligand were found to provide more stable complexes with lower free energies of activation so that this higher-energy species can be avoided. The dissociation pathway with an additional interaction from B 2 Pin 2 , solvent (toluene), or substrate (Me 2 NBn) forms intermediates 13 (ΔG°= 4.5 kcal/mol, relative to 7), 21 (ΔG°= 12.8 kcal/mol), or 23 (ΔG°= 16.2 kcal/mol), respectively. Coordination of B 2 Pin 2 provides the lowestenergy path for product dissociation.
Catalyst Regeneration. The B 2 pin 2 -ligated complex 14 is poised for catalyst regeneration, which requires the net replacement of a hydride with a Bpin ligand. A low free energy activation σ-bond metathesis step was found to provide a low-energy pathway to complex 15 (Δ ‡ G°= 0.3 kcal/mol), which has a σ-bonding interaction with H-Bpin and is the lowest-energy complex in the catalytic cycle (Scheme 14).
There is an equilibrium where Me 2 NBn replaces HBpin in complex 15 and starts the new cycle with regenerated complex 7 (7′), or dissociates HBpin and forms the active catalyst (ΔG°= 3.4 kcal/mol) before coordinating with Me 2 NBn. The energy di erence between complexes 7 and 7′ is the thermodynamic energy for the overall reaction after each cycle, ΔG r = -13.5 kcal/mol (Scheme 7). The most stable complex 16, produced by ligand exchange to replace σ-bound HBpin with σ-bound B 2 Pin 2 (ΔG°= -17.4 kcal/mol), stays outside the catalytic cycle, can inhibit the catalyst regeneration process, and is assigned to turnover-limiting intermediate (TDI).
Rate Law and E ect of HBpin. According to our proposed mechanism, the steady-state approximation provides the following expression for the reaction rate:
2 as detailed in Section 2.5 in the SI. This expression shows first-order kinetics with respect to the concentrations of B 2 Pin 2 and substrate (with saturation at higher concentrations) and half-order kinetics with respect to the concentration of the precatalyst. The saturation kinetics with respect to substrate 3 at higher concentrations of 3 can be explained by the necessary coordination of substrate 3 to the Iridium center leading to the formation of complex 7, which is at low concentration (see Scheme 7). As the concentration of 3 continues to increase, however, there is no longer su cient iridium complex remaining to bind to substrate 3 and increase the concentration of complex 7.
In terms of the kinetic order of B 2 Pin 2 , although 1 equiv of B 2 Pin 2 leads to the catalytic o -cycle species TDI, the catalytic process requires 2 equiv of B 2 Pin 2 to regenerate the same TDI complex in the following turnover number (one for the reaction and one bound to the iridium center). In total, B 2 pin 2 has a net order of 1 and is consistent with our kinetic measurements. This result is di erent from previously reported kinetic studies for nondirected borylation, in which the rate law was independent of the concentration of the boron source in all cases. 16,67,68 The observed half-order kinetics for the precatalyst suggests the existence of a reversible process between [Ir(μ-OMe)(COD)] 2 and trisboryliridium species, as previously reported by Hartwig and colleagues in their initial kinetic study. 16 Most importantly, the rate law includes the concentration of HBpin, which is a side product that explains the autocatalytic behavior. The HBpin concentration dependence of the rate law is consistent with the observation of an induction period (about 10 min) when no HBpin is added. When either HBpin is added or a high enough concentration of HBpin is generated, the reaction rate is accelerated. To further support the role of HBpin as an autocatalyst and not merely involved in catalyst generation, a time course experiment was completed (Figure 6) that started with B 2 pin 2 as the boron source (using the same conditions as those represented in Figure 5). After 50 min and a similar concentration of newly formed ArBpin 4, 0.1 equiv of HBpin was added to the reaction. The rate of the reaction was immediately accelerated above the rate of the reaction with B 2 pin 2 only. Because HBpin was added after the initiation period, the increased rate of the reaction is consistent with HBpin playing a role in the reaction mechanism outside of the initial generation of active catalyst.
Besides the autocatalytic e ect of HBpin, other hypotheses were also considered: HBpin can be consumed as an independent boron source in the separated catalytic cycle or provides for a better energetic pathway in generating the active species. Because the addition of HBpin alone and B 2 pin 2 alone are slower than the addition of both reagents and the rate acceleration with both reagents is much higher than the addition of HBpin alone (Figure 5), it is unlikely that the increased rate results from product formation through a second catalytic cycle. In terms of the possibility of precatalyst activation, from our computational analysis, HBpin does not provide a more favorable pathway for generating the active species compared to the B 2 Pin 2 pathway (see S3.6 in the SI for more details).
Free Energy of Activation. The analogous computation was performed with monofluoro compound 3 to correlate the calculated free energy of activation to the experimental Δ°G ‡ (Scheme 15). The kinetic experiments with substrate 3 further demonstrate the validity of this hypothesis and show that the experimentally determined free energy of activation (22.4 kcal/ mol) is very close to the value computed at the PBE0-D3/BS1 level theory (22.3 kcal/mol).
To explore the computational e ects of toluene as a solvent on the Δ°G ‡ , single-point SMD solvent model computations were performed by using gas-phase geometries. The results summarized in Table S2 show a good correlation (R 2 = 0.966) between the models with and without solvation (Figure S4). The same conclusions can be drawn about the mechanistic pathways. In terms of the DFT methodology, Hopmann's work shows that pure PBE functional combined with the D2 dispersion correction and the integral equation formalism variant of the polarizable continuum solvation model (IEFPCM) produces the most accurate results for iridium chemistry. 85 Thus, we report here a comparison with singlepoint computations using Hopmann's methodology (albeit on slightly di erent optimized geometries) using solvent param-eters for toluene. In Scheme 15, it is interesting that the closest value was obtained by PBE0-D3/BS1 which is a gas phase and utilizes a double-ζ basis set. However, the computational results in general are in agreement with the experimental kinetic value, regardless of which methodology is used.
Experimental and computational studies were utilized to examine the mechanism of amine-directed C-H borylation. The turnover-limiting step of the catalytic cycle is the isomerization of the iridium(III) C-H-activated complex rather than the C-H activation step itself. This unusual outcome is supported by the lack of both inter-and intramolecular KIE, and by computational analysis that shows a low free energy of activation for C-H activation by σ-bond metathesis. The computationally determined turnoverlimiting step involves the isomerization of the complex that results from C-H activation. That isomerization involves the dissociation and reassociation of HBpin to align the complex in a way that allows C-B bond formation. Prior to isomerization, an inaccessibly high free energy of activation is observed for C-B bond formation. The resulting computational energetic profile provides an overall free energy of activation that matches the experimental results very well.
There is a clear distinction between the mechanism of directed and nondirected C-H borylation reactions that are unveiled through this comprehensive mechanistic study. Coordination of the directing group results in changes in the rate-determining step of the catalytic cycle by constraining the geometry of the multiple transition states. Ultimately, this change in the energy profile of the reaction resulted in two key experimental observations that are strongly corroborated by the results from the computations. First, the inverse KIE for an intermolecular labeling study and no KIE for an intramolecular competition experiment. Second is the role of HBpin as an autocatalyst when B 2 pin 2 is the sole boron source and to accelerate the reaction as an additive. The latter result has potential implications in a wide range of C-H borylation reactions as it could render ine ective transformations as e cient upon addition of borane. The details of the mechanism provided in this work are expected to provide critical new insights into directed C-H borylation reactions, a steadily expanding area of C-H functionalization reactions.
The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acscatal.3c03316.
https://doi.org/10.1021/acscatal.3c03316 ACS Catal. 2023, 13, 12877-12893 Downloaded via UNIV OF CINCINNATI on September 21, 2023 at 18:03:54 (UTC).See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
https://doi.org/10.1021/acscatal.3c03316 ACS Catal. 2023, 13, 12877-12893