C(sp²)–H Borylation of Heterocycles by Well-Defined Bis(silylene)pyridine Cobalt(III) Precatalysts: Pincer Modification, C(sp²)–H Activation and Catalytically Relevant Intermediates

Abstract

Well-defined bis(silylene)pyridine cobalt(III) precatalysts for C(sp²)–H borylation have been synthesized and applied to the investigation of the mechanism of the catalytic borylation of furans and pyridines. Specifically, [(^ArSiNSi)CoH₃]·NaHBEt₃ (^ArSiNSi = 2,6-[EtNSi(N^tBu)₂CAr]₂C₅H₃N, Ar = C₆H₅ (1-H₃·NaHBEt₃), 4-MeC₆H₄ (2-H₃·NaHBEt₃)) and trans-[(^ArSiNSi)Co(H)₂BPin] (Ar = C₆H₅ (1-(H)₂BPin), 4-MeC₆H₄ (2-(H)₂BPin), Pin = pinacolato) were prepared and employed as single component precatalysts for the C(sp²)–H borylation of 2-methylfuran, benzofuran and 2,6-lutidine. The cobalt(III) precursors, 2-H₃·NaHBEt₃ and 2-(H)₂BPin also promoted C(sp²)–H activation of benzofuran, yielding [(^ArSiNSi)CoH(Bf)₂] (Ar = 4-MeC₆H₄, 2-H(Bf)₂, Bf = 2-benzofuranyl). Monitoring the catalytic borylation of 2-methylfuran and 2,6-lutidine by ¹H NMR spectroscopy established the trans-dihydride cobalt(III) boryl as the catalyst resting state at low substrate conversion. At higher conversion two distinct pincer modification pathways were identified, depending on the substrate and the boron source.

Graphical Abstract

Introduction.

The metal-catalyzed borylation of C–H bonds is one of the most widely applied methods of C–H functionalization, due in part to the versatility and synthetic utility of the resulting aryl- or heteroaryl boronic ester products.¹ State-of-the-art methods rely on in situ-generated iridium catalysts derived from an iridium(I) precursor and a bidentate supporting ligand, typically a (bis)phosphine or substituted 2,2’-bipyridines.² The mechanism of iridium-catalyzed C–H borylation has been studied and it is well-accepted that an Ir(III)-Ir(V) cycle is operative where an iridium(III) trisboryl is responsible for turnover limiting C–H activation.³ Highly active iridium catalysts have been discovered with predictable site-selectivity that is typically sterically driven, where sterically accessible C–H bonds undergo borylation with statistical selectivity. Approaches involving hydrogen-bonding interactions,⁴ directing group strategies⁵ and specialized ligand designs⁶ have been developed to overcome sterically-driven site-selectivity.

Recent interest has focused on the use of catalysts with more Earth-abundant first-row transition metals to promote C–H borylation and examples with Fe,⁷ Co⁸ and Ni⁹ are now known. In addition to possible economic benefits and reduced environmental impact, these catalysts offer the opportunity for selectivity distinct or orthogonal to state-of-the-art iridium.¹⁰ With cobalt, the bis(diisopropylphosphine)pyridine pincer, [^iPrPNP], imparts a strong ligand field and enables the two-electron oxidative addition of H₂, C–H and B–H bonds to cobalt(I)–alkyl precursors.^11,12 In catalytic C–H borylation (Scheme 1a), the [(^iPrPNP)Co] family of precatalysts offers unique ortho-to-fluorine site selectivity in the borylation of fluorinated arenes, arising from reversible, thermodynamically-controlled oxidative addition where formation of the strongest M–C_aryl bond governs borylation selectivity.¹³ In all cases examined, a Co(I)–Co(III) cycle is favored where a cobalt(I)–boryl intermediate is responsible for cleavage of the C(sp²)–H bond.^14,15,16

Scheme 1.

Comparison of cobalt pincer precatalysts for the C(sp²)–H borylation of heterocycles and motivations for the present study.

Our laboratory has explored a variety of tridentate pincer ligands that support active cobalt catalysts for C–H borylation.^6,8b Most exhibited inferior activity to the [(^iPrPNP)Co] family. In addition, rapid formation of catalytically inactive bis(chelate) metal complexes was also deleterious. Driess, Cui and coworkers reported a notable exception with the efficient C–H borylation of arenes and heteroarenes employing [(^PhSiNSi)CoBr₂] (^PhSiNSi = (2,6-[EtNSi(N^tBu)₂CPh]₂C₅H₃N)) activated in situ with NaHBEt₃ in the presence of excess cyclohexene (Scheme 1b).¹⁸ The introduction of silylene donors generates one of the most electron-donating pincers known, and delineation of the consequences of such an electronic effect in catalysis has not been established.^19,20,21

Despite their competence in catalytic C–H borylation, little is known about the identity of the active species, the composition of the cobalt complex responsible for C–H activation, the role of the cyclohexene additive and what redox cycle is operative during catalysis. Here we describe the synthesis and characterization of well-defined cobalt(III) complexes supported by [SiNSi]-type pincers (Scheme 1c) and establish key elements of the catalytic cycle including the redox states accessed during catalysis, demonstration of catalyst modification during turnover and insights into the role of the cyclohexene additive. These studies uncover the unique consequences of introduction of such an electron rich pincer on elementary steps relevant to C–H borylation.

RESULTS AND DISCUSSION

Synthesis of Bis(silylene)pyridine Cobalt(III) Complexes.

To gain insight into the mechanism and the redox cycle operative in C–H borylation promoted by [SiNSi]-supported cobalt compounds, well-defined coordination precatalysts were targeted. Furthermore, single component precatalysts eliminate the need for air-sensitive and pyrophoric activators that because of their extreme reactivity, limit synthetic applications. To date, the cobalt(II) dibromide, [(^PhSiNSi)CoBr₂]¹⁸ is the only reported cobalt complex of this pincer also inspiring additional investigations in the coordination chemistry and accessible oxidation states of these complexes.

Inspired by synthetic route used to prepare [(^iPrPNP)Co(I)] complexes,^8a two equivalents of NaHBEt₃ (1.0 M solution in toluene) were added to a frozen toluene solution containing [(^ArSiNSi)CoCl₂] (Ar = C₆H₅, 4-MeC₆H₄) at −95 °C (see Supporting Information). Filtration and subsequent recrystallization resulted in the isolation of pale yellow solids identified as [(^ArSiNSi)CoH₃]·NaHBEt₃ in 40% (Ar = C₆H_5, 1-H₃·NaHBEt₃) and 43% (Ar = 4-MeC₆H₄, 2-H₃·NaHBEt₃) yield respectively (Scheme 2). The p-tolyl variant of the [SiNSi] ligand was employed to improve the crystallinity of the products.

Scheme 2.

Synthesis of [(^ArSiNSi)CoH₃]·NaHBEt₃ (Ar = C₆H₅, 1-H₃·NaHBEt₃; Ar = 4-MeC₆H₄, 2-H₃·NaHBEt₃).

The ¹H and ¹³C NMR spectra of 1-H₃·NaHBEt₃ in benzene-d₆ exhibited signals consistent with a C_2v symmetric compound. The ethyl groups of the NaHBEt₃ ligand were observed as broad signals at 0.86 ppm (CH₂) and 1.50 ppm (CH₃) that were maintained in the spectrum even after repeated washing with Et₂O, supporting an interaction between the cobalt complex and the NaHBEt₃ that is preserved in benzene solution. A singlet was observed at −17.19 ppm consistent with rapid exchange of the three hydride ligands on the NMR timescale at 23 °C. Upon cooling a toluene-d₈ solution to −75 °C, two broad resonances were observed at −13.32 (2 H) and −24.33 ppm (1 H), corresponding to the apical and equatorial hydrides respectively (Figure S6). Increasing the temperature to −35 °C resulted in coalescence of the signals at −17.19 ppm. A similar dynamic was reported for [(^iPrPNP)CoH₃] lacking the coordinated NaHBEt₃.¹³ A minimum T₁ value of 113 ms was measured at −5°C for 1-H₃·NaHBEt₃, supporting its formulation as a classical cobalt trihydride, although the impact of the coordinated NaHBEt₃ on this value is unknown.^13,22

A pathway to rationalize the formation of 1-H₃·NaHBEt₃ and 2-H₃·NaHBEt₃ involves initial reduction of the cobalt(II) complex [(^ArSiNSi)CoCl₂] by NaHBEt₃ to the cobalt(I) derivative, [(^ArSiNSi)CoCl] with concomitant formation of H₂. Generation of the cobalt(I) hydride, [(^ArSiNSi)CoH], followed by oxidative addition of the H₂ present in the reaction and interaction with a molecule of NaHBEt₃ yields the isolated cobalt(III) product.

Slow evaporation of a diethyl ether solution of 2-H₃·NaHBEt₃ at ambient temperature produced yellow crystals suitable for single-crystal X-ray diffraction (Figure 1). A distorted octahedral geometry is observed with the pincer and the three hydride ligands defining the coordination sphere around the cobalt. The sodium of the NaHBEt₃ bridges two of the hydride ligands (H(50) and H(51)) and the Co(1)–Si(1) and Co(1)–Si(2) bond distances of 2.1226(5) and 2.1101(5) Å respectively, are consistent with the presence of two silylene donors.²¹ The dihedral angles between the planes of the p-tolyl substituents (C(11)–C(12)–C(13) and C(19)–C(20)–C(21)) and the plane defined by the pincer ligand (Co(1)–N(4)–C(7)) are 11.9(2)° and 5.2(2)° respectively, supporting a nearly coplanar orientation of the p-tolyl substituents with the SiNSi ligand.

Figure 1.

Solid-state structure of 2-H₃·NaHBEt₃ at 30% probability ellipsoids. The hydrogen atoms bonded to cobalt (H(50), H(51) and H(52)) have been located and freely refined. All hydrogen atoms except those bonded to cobalt, a B-CH₂ group and two of the tert-butyl groups in the pincer have been omitted for clarity.

Cobalt(I) complexes of [(^iPrPNP)Co] are useful precatalysts for C–H borylation and are routinely prepared by treatment of [(^iPrPNP)CoCl₂] with two equivalents of an alkyl lithium reagent, RLi (R = Me, CH₂SiMe₃).^8a,15 Attempts to synthesize analogous cobalt(I) complexes starting from [(^ArSiNSi)CoCl₂] and MeLi, PhLi or CH₂SiMe₃Li were unsuccessful, producing decomposition (with MeLi or PhLi) or unidentified mixtures of products (with LiCH₂SiMe₃). Likewise, treatment of (^RPNP)CoCl₂ complexes with one equivalent of NaHBEt₃ has been used to synthesize the corresponding (^RPNP)CoCl derivatives.²³ Applying this methodology to [(^ArSiNSi)CoCl₂] resulted in lower yield and purity of the hydride adducts, 1-H₃·NaHBEt₃ and 2-H₃·NaHBEt₃.

The synthesis of a cobalt(III) dihydride boryl was also pursued as these complexes have been identified as catalyst resting states and precatalysts for C–H borylation.^{8a,10c,13,14,15} Addition of excess (~ 6 equiv) of HBPin to 1-H₃·NaHBEt₃ or 2-H₃·NaHBEt₃ in THF and heating to 80 °C for 10 minutes followed by washing with Et₂O resulted in isolation of yellow solids identified as trans-[(^ArSiNSi)Co(H)₂BPin] (Ar = Ph, 1-(H)₂BPin; Ar = 4-MeC₆H₄, 2-(H)₂BPin) in 65% and 68% yield, respectively (Scheme 3). The ¹H NMR spectrum of 1-(H)₂BPin in benzene-d₆ exhibited the number of signals consistent with a C_2v symmetric molecule. A diagnostic singlet was observed at −10.93 ppm assigned to the two equivalent trans hydrides. No signals were observed between 0.50 to 1.50 ppm attributable to the ethyl groups of NaHBEt₃, demonstrating elimination of this salt from the coordination sphere of the cobalt. The 4-Me-^iPrPNP (4-methyl-2,6-ⁱPr₂P-methylpyridine) analog of 1-(H)₂BPin and 2-(H)₂BPin, [(4-Me-^iPrPNP)Co(H)₂BPin], has been previously synthesized by treatment of the corresponding cobalt(II) dichloride complex with two equivalents of NaHBEt₃ in the presence of 4 equivalents of HBPin.¹⁴ Attempts to apply this method to the synthesis of 1-(H)₂BPin produced the desired cobalt(III) product along with other unidentified products as judged by ¹H NMR spectroscopy.

Scheme 3.

Synthesis of trans-[(^ArSiNSi)Co(H)₂BPin] (Ar = C₆H₅, 1-(H)₂BPin; Ar = 4-MeC₆H₄, 2-(H)₂BPin).

Slow evaporation of a diethyl ether solution of 2-(H)₂BPin at room temperature furnished yellow crystals suitable for single-crystal X-ray diffraction (Figure 2). As with 2-H₃·NaHBEt₃, a distorted octahedral geometry about the cobalt is observed in 2-(H)₂BPin, with the two hydride ligands in a mutually trans disposition and the BPin ligand trans to the pyridyl ring. The Co(1)–B(1) (1.912(2) Å) and Co(1)–H(22) (1.44(3) Å) bond distances are similar to those in the ^iPrPNP analog, trans-[(^iPrPNP)Co(H)₂BPin] (Co–B is 1.926(4) Å and Co–H is 1.49(4) Å).^6a In contrast, the Co(1)–Si(1) (2.1128(4) Å) bond distances are shorter than the Co–P distance in [(^iPrPNP)Co(H)₂BPin] (2.1296(6) Å), and the Co(1)–N(1) bond (2.0378(16) Å) is slightly longer than that in [(^iPrPNP)Co(H)₂BPin] (2.016(3) Å). Both metrics are in agreement with the silylenes in [SiNSi] being stronger electron-donors than the phosphines in [^iPrPNP]. The strong electron-donating ability of the silylene donors is likely the origin of the elongation of the Co-N(pyridine) bond in 2-(H)₂BPin, as previously reported for the [(^PhSiNSi)FeCl₂] complex, where the pyridine ring does not coordinate.^21a

Figure 2.

Solid-state structure of 2-(H)₂BPin at 30% probability ellipsoids. Hydrogen atoms except those bonded to cobalt have been omitted for clarity.

The conversion of 1-H₃·NaHBEt₃ to 1-(H)₂BPin was monitored by ¹H NMR spectroscopy in the presence of excess HBPin at 25 °C, 50 °C and 80 °C with the goal of detecting intermediates. The transformation of 1-H₃·NaHBEt₃ to 1-(H)₂BPin occurred in 7 hours, 40 minutes and 5 minutes respectively and neither the cobalt(I) hydride, [(^PhSiNSi)CoH], nor the cis-[(^PhSiNSi)Co(H)₂BPin] complex were detected (Figure S32).

C(sp²)–H Borylation of Heterocycles with Well-Defined Bis(silylene)pyridine Cobalt(III) Complexes as Precatalysts.

The synthesis of well-defined cobalt(III) dihydride boryl complexes prompted exploration of the performance of these compounds in catalytic C–H borylation. Three representative substrates were selected: 2-methylfuran, benzofuran and 2,6-lutidine (Scheme 4). The borylations of 2-methylfuran and benzofuran proceeded to >98% conversion in 24 hours in THF at 80 °C with one equivalent of HBPin in the presence of 5 mol% of either 1-(H)₂BPin, 2-(H)₂BPin or 1-H₃·NaHBEt₃. The borylation of 2,6-lutidine with 1 equiv of HBPin at 100 °C with 5 mol% of 2-(H)₂BPin over 24 hours produced only small (<5% conversion) amounts of product. In contrast, when B₂Pin₂ was used as the borylating agent under the same conditions, 46% conversion of the starting material to borylated product was observed. As with [(^iPrPNP)Co] catalysts, B₂Pin₂ was the preferred boron reagent for the C–H functionalization of 2,6-lutidine.¹⁴

Scheme 4.

C(sp²)–H borylation of 2-methylfuran, benzofuran and 2,6-lutidine employing 1-H₃·NaHBEt₃, 1-(H)₂BPin or 2-(H)₂BPin as precatalysts. The reported percentages are substrate conversions determined by GC with mesitylene as an internal standard.

The bis(silylene)pyridine cobalt(III) precatalysts are proposed to operate by a redox Co(I)-Co(III) cycle, where reductive elimination of H₂ and loss of NaHBEt₃ from 1-H₃·NaHBEt₃ and of H₂ or HBPin in 1-(H)₂BPin and 2-(H)₂BPin generates a Co(I) intermediate responsible for activation of the C(sp²)–H bond. Because only Co(III) complexes have been detected or isolated, a redox-neutral Co(III) cycle where bond activation occurs by σ-bond metathesis is also plausible although seegminly unlikely because of the high electron count and a hexacoordinate cobalt. Pathways involving reversible modification of the pincer assists in oxidative addition, reductive elimination or other elementary steps are also possible and supported by the various unique modifications processes discovered in this study.

Determination of the Catalyst Resting State: Furan Borylation.

With well-defined cobalt(III) precatalysts in hand, the mechanism of 2-methylfuran borylation was studied. Monitoring the catalytic reaction by ¹H NMR spectroscopy with either 1 equivalent of HBPin and 5 mol% of 1-H₃·NaHBEt₃ or 0.5 equivalents of B₂Pin₂ and 15 mol% of 1-H₃·NaHBEt₃ at 80 °C in THF-d₈ resulted in observation of 1-(H)₂BPin as the catalyst-resting state in both experiments (Scheme 5).

Scheme 5.

(a) Catalyst resting-state in the borylation of 2-methylfuran using 1-H₃·NaHBEt₃ as a precatalyst (b) aromatic region of the ¹H NMR spectrum in THF-d₈ after 1 hour and 40 minutes of reaction (35% substrate conversion) and after 10 hours of reaction (54% substrate conversion).

When the reaction was carried out with 0.5 equivalents of B₂Pin₂ in the presence of 15 mol% of 1-H₃·NaHBEt₃ at 80°C, complete conversion was reached in 40 minutes whereas with 1 equivalent of HBPin and 5 mol% of 1-H₃·NaHBEt₃, it took 10 hours to reach 54% conversion. In both cases, the initial formation of 1-(H)₂BPin (δ Co–H = −11.46 ppm, THF-d₈) was accompanied by formation of two other cobalt species (δ Co–H = −12.23 (major) and −12.26 (minor) ppm, THF-d₈) whose concentration increased over time at expense of 1-(H)₂BPin. Formation of these compounds was slower when the borylation was carried out with 0.5 equivalents of B₂Pin₂ than with 1 equivalent of HBPin (see Figures S46–S49). When B₂Pin₂ was employed as the borylating agent, the ratio of the 1-(H)₂BPin to the new cobalt hydrides was 1:0.04 at the completion of the reaction, in contrast, when HBPin was used the ratio was 1:1.3 at 54% conversion.

Addition of 4 atm of H₂ to THF solutions of 1-H₃·NaHBEt₃, 2-H₃·NaHBEt₃ or 1-(H)₂BPin and heating at 80 °C for 16 hours produced the same cobalt hydrides present in the catalytic reaction. From the hydrogenation of 1-H₃·NaHBEt₃, a brown solid was isolated in a 27% yield and was identified as a 4:1 ratio of a mixture of stereoisomers of [(^Ar,HSiNSi)CoH₂(H₂)] (3-H₂(H₂), ^Ph,HSiNSi = 2-[EtNSi(N^tBu)₂CHPh]-6-[EtNSi(N^tBu)₂CPh]C₅H₃N). Based on analogy, the tolyl variant of the compound was assigned accordingly as 4-H₂(H₂) = ^ptol,HSiNSi = 2-[EtNSi(N^tBu)₂CH(p-tol)]-6-[EtNSi(N^tBu)₂C(p-tol)]C₅H₃N) (Scheme 6a). The products arise from hydride migration from the cobalt to one of the imine groups on the silylene donor along with trapping by σ-H₂ coordination.²⁴ As a consequence of this migration, the [^Ar,HSiNSi] ligand becomes anionic, as the Si-donor bearing the new C–H group is best described as a silyl donor. As a result of hydride migration, both the incipient carbon (C7) and the cobalt are stereogenic. The oxidation state of the cobalt is maintained at +3 and both 3-H₂(H₂) and 4-H₂(H₂) are diamagnetic with two hydride ligands and a σ-H₂ ligand. A related cobalt(III) complex has been reported by Heinekey and coworkers.²⁵

Scheme 6.

(a) Independent synthesis of the catalyst resting-state II (at high substrate conversion) in the borylation of 2-methylfuran with HBPin from hydrogenation of 1-H₃·NaHBEt₃, 2-H₃·NaHBEt₃ or 1-(H)₂BPin. (b) Borylation of benzofuran with 3-H₂(H₂) as the precatalyst. The reported percentage is the substrate conversion determined by GC using mesitylene as an internal standard.

Integration of the hydride signals in the ¹H NMR spectrum of 3-H₂(H₂) (δ Co–H= −11.57 (major); −11.73 (minor), benzene-d₆) established a 4:1 diastereomeric ratio. Because the minor isomer is formed in such small amounts and many of the ligand signals overlap with the major isomer, only the signals for the Co–H and the ^tBu substituents are resolved. The major isomer exhibits the number of ¹H NMR resonances consistent with a C_s symmetric product. Accordingly, a singlet at 5.86 ppm in the ¹H NMR spectrum and a signal at 76.8 ppm in the ¹³C NMR spectrum were assigned to the newly formed C(sp³)–H group. A minimum value of T₁ of 94 ms was measured at 0 °C for 4-H₂(H₂), in the range diagnostic for σ-H₂ ligands.²⁸ At −85 °C, the exchange between the hydrides and the dihydrogen ligand in 4-H₂(H₂) was slowed and the ¹H NMR spectrum in toluene-d₈ exhibited two broad signals in the hydride region at - 10.05 (3 H) and −22.65 (1 H) ppm that were assigned as the mutually trans hydride and σ-H₂ ligands and to the hydride trans to the pyridyl ring, respectively (see Figure S51).

The observation of 3-H₂(H₂) from hydrogenation of 1-(H)₂BPin accounts for its formation during catalytic C–H borylation. The borylation of 2-methylfuran with HBPin generates H₂ as the stoichiometric byproduct; capture of the H₂ by the 1-(H)₂BPin resting state produces the cobalt product with the modified pincer. This also accounts for the higher conversion of 1-(H)₂BPin to 3-H₂(H₂) when HBPin is used in place of B₂Pin₂. With the latter boron source, HBPin is the stoichiometric byproduct and the concentration of H₂ is low and hence little modified product was observed.

The catalytic performance of the modified pincer complex, 3-H₂(H₂) was evaluated in the borylation of benzofuran. With 1 equivalent of HBPin and 5 mol% of 3-H₂(H₂), 83% conversion to product was observed at 80 °C over the course of 24 hours (Scheme 6b). The conversion was lower than when 1-H₃·NaHBEt₃ or 1-(H)₂BPin were used, suggesting that 3-H₂(H₂) is less ative. The catalytic borylation of 2-methylfuran with HBPin and 5 mol % of 3-H₂(H₂) was monitored by ¹H NMR spectroscopy in THF-d₈ to gain insight into the catalyst resting state. A Co–H resonance was observed at −11.97 ppm throughout catalytic turnover corresponding to an unidentified cobalt compound, along with those for the 3-H₂(H₂) mixture (Figure S54). Detectable quantities of 1-(H)₂BPin were not observed during in these experiments supporting the irreversibility of the intramolecular hydride migration process and suggesting a different cycle operative than that with the precatalysts with the intact ligand.

Intramolecular hydride or alkyl migration to imine ligands in transition metal complexes has been reported,²⁶ and, more importantly, has been recognized as a deactivation pathway for pyridine(diimine) iron-catalyzed olefin polymerization.²⁷ It is particularly remarkable that in the case of the borylation of 2-methylfuran by bis(silylene)pyridine cobalt complexes, this process does not constitute a deactivation pathway, and 3-H₂(H₂) is a competent precatalyst for C(sp²)–H borylation. The mechanism of operation, active redox cycle and nature of the C–H activating species are unknown.

C–H Activation of Benzofuran by Bis(silylene) pyridine Cobalt(III) Complexes.

Because cobalt(III) dihydride boryl and cobalt(III) trihydrides complexed with NaHBEt₃ containing the bis(silylene)pyridine pincer are efficient precatalysts for the borylation of heterocycles, additional insights into the C–H activation step were sought. Addition of 6 equivalents of benzofuran to a THF solution of 2-H₃·NaHBEt₃ or 2-(H)₂BPin followed by heating (50 °C for 2-H₃·NaHBEt₃; 80 °C for 2-(H)₂BPin) for 1 h and 30 min, yielded a single new cobalt compound that was isolated as an orange powder in moderate (46% and 40% respectively) yield (Scheme 7). The product was identified as cis-[(^ptolSiNSi)CoH(Bf)₂] (2-H(Bf)₂, Bf = 2-benzofuranyl) based on a combination of 1D and 2D NMR characterization. The ¹H and ¹³C NMR spectra of 2-H(Bf)₂ in benzene-d₆ exhibited the number of signals consistent with a C_s symmetric complex. A singlet at −11.41 (1H) ppm in the ¹H NMR spectrum was assigned to the hydride ligand while the C(4)-H groups of each benzofuranyl ligand occurred at 7.51 and 7.48 ppm respectively. Both signals did not show any crosspeak in the ¹H, ¹H-COSY spectrum supporting activation of the C(2)–H bond of benzofuran. A ¹H,¹³C-HMBC spectrum allowed for the assignment of the ¹³C NMR signals at 203.7 and 194.6 ppm to the Co-bonded C(5) atoms.

Scheme 7.

Net C–H activation of benzofuran by [(SiNSi)Co(III)].

Complex 2-H(Bf)₂ exhibited a distorted octahedral geometry with the two benzofuranyl ligands in a mutually cis arrangement and the hydride ligand trans to one of the benzofuranyl ligands (Figure 3). The benzofuranyl ligand trans to the hydride is tilted 12.2(3)° from the plane defined by Co(1)–N(4)–C(42), whereas the benzofuranyl ligand trans to the pyridyl ring is tilted 45.4(4)° from the plane that contains Co(1)–N(4)–C(1). The Co(1)–C(42) bond distance is longer than the Co(1)–C(50) as a result of an stronger trans influence of the hydride in comparison to the pyridyl ring. Co(1)–Si(1) and Co(1)–Si(2) bond distances (2.1552(10), 2.1371(10) Å respectively) are slightly longer than those found for 2-(H)₂BPin (see above), whereas the Co(1)–N(4) bond distance (2.031(3) Å) is similar for both, supporting both as cobalt(III) complexes.

Figure 3.

Solid-state structure of 2-H(Bf)₂ at 30% probability ellipsoids. The hydrogen atom bonded to cobalt (H(57)) has been located. All other hydrogen atoms and two of the tert-butyl groups in the pincer have been omitted for clarity.

To gain additional insight into the formation of 2-H(Bf)₂, the reaction of 2-H₃·NaHBEt₃ or 2-(H)₂BPin with 6 equivalents of benzofuran was monitored by ¹H NMR spectroscopy at 50 °C in THF-d₈. Three cobalt complexes were observed prior to formation of 2-H(Bf)₂. The diastereomers of the modified pincer complex, 4-H₂(H₂) were identified as one of the cobalt complexes present and increased over the course of the C–H activation reaction. The other two compounds were identified as cis-[(^ptolSiNSi)CoH₂(Bf)] (cis-2-H₂(Bf)) and trans-[(^ptolSiNSi)CoH₂(Bf)] (trans-2-H₂(Bf)) based on 1D and 2D NMR spectroscopic experiments. Based on these observations, a pathway accounting for formation of cis-2-H₂(Bf) and trans-2-H₂(Bf) from 2-H₃·NaHBEt₃ is presented in Scheme 8.

Scheme 8.

Proposed pathway for the formation of 2-H(Bf)₂.

Loss of NaHBEt₃ and reductive elimination of H₂ generates a putative and unobserved cobalt(I)–hydride, [(^ptolSiNSi)CoH], that promotes rapid C–H oxidative addition of benzofuran accounting for cis-2-H₂(Bf) and trans-2-H₂(Bf) depending on the trajectory of the substrate. Reductive elimination from cis-2-H₂(Bf) generates a cobalt(I)–benzofuranyl that undergoes oxidative addition of benzofuran to form 2-H(Bf)₂. At initial stages of the reaction (from 5 min to 1 h 45 min), when 2-H₃·NaHBEt₃ has not been fully consumed, the cis:trans ratio of the 2-H₂(Bf) complexes was 1:0.9 and remained constant. When all the starting material was consumed (1 h 45 min), the cis:trans ratio gradually changed to 1:2.1, supporting that formation of 2-H(Bf)₂ takes place preferentially from the cis isomer. The fact that the cis isomer is present in the reaction until all the trans complex is consumed to form 2-H(Bf)₂ supports that trans-2-H₂(Bf) isomerizes to cis-2-H₂(Bf) which ultimately yields the final product. During the reaction, the amount of 4-H₂(H₂) gradually increased (cis-2-H(Bf)₂:4-H₂(H₂) 1:0.4 at 45 min, 1:1.4 at 2h 30 min and 1:5.4 at 3 h 5 min) reaching a 2-H(Bf)₂:4-H₂(H₂) ratio of 1:0.2 at the completion of the reaction. The generation of H₂ during this process accounts for formation of 4-H₂(H₂). Because pincer modification is irreversible, this compound accumulates over the course of the C–H activation reaction.

The same three cobalt products were obtained when the reaction was conducted with one rather than six equivalents of benzofuran. Notably, 2-H₃·NaHBEt₃ remained in solution after all the benzofuran was consumed. Observation of the same product independent of whether 2-H₃·NaHBEt₃ or 2-H₂BPin was used as the starting material demonstrates that the coordinated NaHBEt₃ plays little role in the C–H activation of benzofuran. It is important to note that no spectroscopic evidence has been obtained for any [(SiNSi)Co(I)] intermediate, in stark contrast to the PNP cases where such complexes are common precatalysts, highlighting the impact of the increased electron donation of the silylene-based pincer.

To probe the presence of a cobalt(I) hydride in the C–H activation of benzofuran, attempts were made to synthesize a cobalt(I) complex capable of C–H activation. Addition of neohexene to a benzene-d₆ solution of 1-H₃·NaHBEt₃ resulted in decomposition with no evidence for a cobalt(I) alkyl. Addition of π-acidic carbon monoxide was explored in an attempt to stabilize the lower oxidation state. Addition of 1 atm of CO to a benzene-d₆ solution of 1-H₃·NaHBEt₃ or 2-(H)₂BPin yielded yellow solids in 75% and 71% yields, respectively. The products were identified as the cobalt(I) dicarbonyl derivatives, [(^ArSiNSiH)Co(CO)₂] (Ar = C₆H₅, 5-(CO)₂; Ar = 4-MeC₆H₄, 6-(CO)₂) where the pincer has been modified to an anionic silylene-pyridine-silyl ligand formed from cobalt-to-silicon hydride migration (Scheme 9).

Scheme 9.

Reaction of 1-H₃·NaHBEt₃ or 2-(H)₂BPin with CO to yield 5-(CO)₂ and 6-(CO)₂ respectively.

The ¹H NMR spectrum of 5-(CO)₂ in benzene-d₆ exhibited resonances consistent with an asymmetric SiNSi ligand, e. g., four singlets for the four tBu groups (at 1.54, 1.29, 1.27 and 1.10 ppm, 9 H each) supporting the loss of the horizontal and vertical mirror planes in the molecule. This symmetry contrasts the C_s symmetry observed for 3-H₂(H₂) or 4-H₂(H₂), further supporting hydride migration to the amidinate C=N bond in the latter compounds. A singlet at 6.49 ppm was assigned to the silicon-bonded hydrogen atom and was assigned based on a ¹H,¹³C-HSQC experiment, which did not exhibit a crosspeak for that signal, establishing that the hydrogen atom is not bonded to a C atom. The ¹³C NMR spectrum exhibited two signals at 206.7 and 205.6 ppm that were assigned to the two inequivalent carbonyl ligands. Moreover, the C(25) in the amidinate (see Scheme 10) is upfield shifted (163.0 ppm) in agreement with a localized double bond and contrasts with the chemical shift of C(10) (172.6 ppm) engaged in delocalized N-C=N bonding. The IR spectrum of 5-(CO)₂ in KBr showed two bands at 1871 and 1860 cm⁻¹ in the carbonyl region, consistent with the presence of two CO ligands and a band at 1925 cm⁻¹ that is assigned as the Si–H stretch.²⁸

Scheme 10.

(a) Catalyst resting-states for the borylation of 2,6-lutidine. (b) aromatic region of the ¹H NMR spectrum after 5 minutes of reaction (5% substrate conversion), showing signals for 1-(H)₂BPin, and after 4 hours of reaction (45% substrate conversion), showing signals of a mixture of 1-(H)₂BPin and 7-(H)₂BPin.

Single crystals of 5-(CO)₂ suitable for X-ray diffraction were obtained from a pentane solution stored at −35 °C for three days (Figure 4). An idealized trigonal bipyramidal geometry was observed with a Co(1)–Si(1) bond distance of 2.1734(4) Å contracted from the value of 2.2392(4) Å of Co(1)–Si(2), consistent with Si(1) being a silylene donor and Si(2) being a silyl. In addition, the C(25)–N(6) bond distance of 1.4046(17) Å is consistent with a C–N single bond while the C(25)–N(7) bond length of 1.2769(18) Å is in agreement with a C=N bond. In contrast, the C(10)–N(1) and C(10)–N(2) bond distances on the unmodified amidinate donor are 1.3385(17) and 1.3425(17) Å, supporting a delocalized double bond.

Figure 4.

Solid-state structure of 5-(CO)2 at 30% probability ellipsoids. The hydrogen atom bonded to silicon has been located and freely refined. All other hydrogen atoms have been omitted for clarity.

Formation of 5-(CO)₂ from carbonylation of 1-H₃·NaHBEt₃ likely proceeds by initial H₂ reductive elimination and elimination of NaHBEt₃ to yield an undetected cobalt(I) hydride complex, [(^PhSiNSi)CoH(CO)] (Scheme 9). This intermediate then undergoes cobalt-to-silicon hydride migration promoted by the coordination of a second molecule of CO to yield the final product. During the course of this migration, one of the Si–N bonds breaks and the silylene donor converts to an anionic silyl ligand. A similar pathway is envisioned for the formation of 6-(CO)₂ from 2-(H)₂BPin, however, in this case 2-(H)₂BPin should undergo an initial step of HBPin reductive elimination (instead of H₂) to yield the final product.

Determination of the Catalyst Resting State for the Borylation of 2,6-lutidine with Well-Defined Bis(silylene)pyridine Cobalt (III) Precatalysts.

Monitoring the borylation of 2,6-lutidine with one equivalent of B₂Pin₂ in THF-d₈ at 100 °C in the presence of 15 mol% of 1-H₃·NaHBEt₃ or 1-(H)₂BPin resulted in observation of two cobalt complexes by ¹H NMR spectroscopy (Scheme 10). At low substrate conversion (5%), 1-(H)₂BPin remained as the predominant cobalt species while at higher conversion (45%), a new cobalt compound identified as trans-[(^BPinSiNSi)Co(H)₂BPin] (^BPinSiNSi = 2,6-[EtNSi(NtBu)₂C(3-BPinC₆H₄)]₂C₅H₃N) (7-(H)₂BPin) was observed. The latter arises from borylation of the meta positions of both phenyl rings in the [^PhSiNSi] pincer during catalysis. A trace amount of 3-H₂(H₂) was also detected at high (74%) substrate conversion, presumably formed from reaction of 1-(H)₂BPin with the H₂ formed from the borylation of lutidine with the HBPin byproduct.

The independent synthesis of 7-(H)₂BPin was pursued from treatment of 1-H₃·NaHBEt₃ with excess of B₂Pin₂ in THF in the absence of substrate (Scheme 11). At 25 °C, the major species was 1-(H)₂BPin while raising the temperature to 100 °C afforded 7-(H)₂BPin as the major product. Attempts to access pure 7-(H)₂BPin have been unsuccessful and the borylation of the pincer did not reach completion even when 20 equivalents of B₂Pin₂ and longer reaction times of 16 h at 100 °C were used. Nevertheless, diagnostic features of 7-(H)₂BPin have been identified that signal its formation. The Co–H appears at −11.42 ppm, close to the value of −11.44 ppm observed for 1-(H)₂BPin in THF-d₈. A diagnostic singlet at 7.92 ppm also appeared at the expense of the Ph signals of 1-(H)₂BPin. The fact that this signal was a singlet enabled the assignment as the ortho-H (H_o) of the Ph ring, supporting selective borylation at the meta position.

Scheme 11.

Attempted independent synthesis of 7-(H)₂BPin by treatment of 1-H₃·NaHBEt₃ with B₂Pin₂.

To further explore the role of C–H activation under catalytically relevant conditions as well as to gain insight into reductive elimination from cobalt(III), 1-(H)₂BPin was treated with excess DBPin in THF at 100 °C. A mixture of isotopologues with the general formula [(^Ph-dSiNSi)CoD_xH_2-x(BPin)] (x = 0, 1, 2) was observed as judged by ²H and ¹H NMR spectroscopies. Deuterium incorporation was observed in the Co–H bonds as well as on the Ph rings of the [^PhSiNSi] ligand (15 % d-incorporation), demonstrating that oxidative addition of the C(sp²)–H bonds of the Ph groups in the SiNSi pincer is operative. In contrast, addition of excess DBPin to 2-(H)₂BPin under the same conditions produced a mixture of isotopologues with the general formula [(^ptolSiNSi)CoD_xH_2-x(BPin)] (x = 0, 1, 2). Notably, deuterium was not observed into the p-tolyl substituents of the pincer ligand, a result of the steric hindrance imposed by the methyl groups in the para position that inhibits oxidative addition.

Competing borylation of the 4-position of the pyridyl ring in the pincer was also observed in the catalytic borylation of 2,6-lutidine with B₂Pin₂ with [(^iPrPNP)CoCH₂SiMe₃].¹⁴ Notably, the borylation of the pincer was faster than the substrate and deactivated the cobalt toward turnover limiting C–H oxidative addition. To determine the effect of pincer borylation on the catalytic performance of (^PhSiNSi)-based cobalt catalysts, the borylation of 2,6-lutidine with 2-H₃·NaHBEt₃ or 2-(H)₂BPin was carried out as the para-methyl groups on the aryl rings prevent C–H borylation. Monitoring the catalytic reaction by ¹H NMR spectroscopy in THF-d₈ at 100 °C established 2-(H)₂BPin as the exclusive cobalt resting state confirming that borylation of the phenyl rings did not take place (Scheme 12). This result is also consistent with the lack of observable deuterium incorporation in stoichiometric experiments with DBPin.

Scheme 12.

(a) Catalyst resting-state for the borylation of 2,6-lutidine employing 2-H₃·NaHBEt₃ as precatalyst. (b) aromatic region of the ¹H NMR spectrum after 5 minutes of reaction (5% substrate conversion) and after 11 hours and 30 minutes of reaction (62% substrate conversion) showing signals of 2-(H)₂BPin.

Comparing the rate of disappearance of 2,6-lutidine in the borylation with B₂Pin₂ and either 1-H₃·NaHBEt₃ or 2-H₃·NaHBEt₃ established that the reaction was approximately 1.4 times faster with the former (Figure S103). This result demonstrates that formation of 7-(H)₂BPin for pincer borylation has little discernable effect on the rate of the catalytic reaction. However, integration of the hydride signal of 2-H₂BPin against the mesitylene internal standard revealed that deactivation occurred during turnover (see Figure S102) but the origin of this deactivation process has not been identified.

Having identified pincer borylation with well-defined cobalt precatalysts and obtaining the appropriate spectroscopic signatures raised the question as to whether this process occurs with in-situ generated catalysts. Monitoring the catalytic borylation of 2,6-lutidine with B₂Pin₂ using 5 mol% of (^PhSiNSi)CoCl₂ activated with 2 equiv of NaHBEt₃ by ¹H NMR spectroscopy revealed formation of 1-(H)₂BPin and 7-(H)₂BPin, supporting access to the same catalytic cycle as with the well-defined 1-H₃·NaHBEt₃ precatalyst. The same catalyst resting-states were observed when the borylation was initiated from 1-H₃·NaHBEt₃ in the presence of a stoichiometric amount of cyclohexene under the same conditions again supporting a similar catalytic cycle. However, in this case, the reaction was found to be 2.4 times faster than in the absence of cyclohexene (Figure S106). In addition, a signal at 1.44 ppm (THF-d₈) was observed corresponding to cyclohexane that increased as the catalytic reaction progressed. These observations support the role of cyclohexene as a H₂ acceptor that likely facilitates generation of the cobalt(I) complex responsible for C–H activation.²⁹ When the borylation of 2,6-lutidine with B₂Pin₂ was carried out with 2-(H)₂BPin and in the presence of one equivalent of cyclohexene, >98% substrate conversion was observed also consistent with reductive elimination from cobalt(III).

To further support the role of cyclohexene as a hydrogen acceptor, excess cyclohexene (6 equivalents) was added to a solution of 1-H₃·NaHBEt₃ in benzene-d₆ and the reaction was monitored by ¹H NMR spectroscopy at 25°C with mesitylene as an internal standard. Formation of cyclohexane was observed and the concentration of 1-H₃·NaHBEt₃ decreased over time consistent with hydrogen transfer from cobalt(III) to form a putative cobalt(I) complex that is unstable preventing observation.

CONCLUSIONS

In summary, well-defined cobalt(III) precatalysts with an electron-donating bis(silylene)pyridine pincer have been synthesized and structurally characterized. These complexes proved effective single-component precatalysts for the borylation of furans and 2,6-lutidine. These compounds were also identified from in-situ catalyst activation procedures, supporting common active species using both procedures. While the strong electron donating properties of the pincer precluded synthesis or observation of cobalt(I) derivatives, a Co(I)-Co(III) redox cycle is plausible for catalysis although steps where the pincer undergoes reversible modification are also possible. Support for this conclusion derives from a modest rate acceleration upon addition of cyclohexene, which serves as a hydrogen acceptor presumably facilitating generation of cobalt(I) intermediates.

Pincer modification processes were also identified that were competitive with catalytic turnover and were observed at higher conversions of substrate. When HBPin was used as the borylating agent for the functionalization of furans, the stoichiometric H₂ byproduct proved deleterious, prompting a hydride migration from the cobalt to an imine position of the pincer. The resulting cobalt complex proved less active cobalt σ-H₂ complex. For the borylation 2,6-lutidine with B₂Pin₂, borylation of the phenyl rings on the pincer was observed with little effect on catalytic activity. The results of these studies highlight the consequences of introducing a more electron rich pincer. These insights should prove useful for the design of next generation catalysts with improved performance.