Effect of Pincer Methylation on the Selectivity and Activity in (PNP)Cobalt-Catalyzed C(sp²)–H Borylation

Abstract

Cobalt complexes supported by a tetramethylated PNP pincer ligand (Me₄^iPrPNP = 2,6-(ⁱPr₂PCMe₂)₂(C₅H₃N)) have been synthesized and structurally characterized. Examples include cobalt(I)–choride, –methyl, –aryl and –benzofuranyl derivatives. The performance of these compounds was evaluated in the catalytic borylation of fluorinated arenes using B₂Pin₂ as the boron source. While P–C bond cleavage, a known deactivation pathway in [(PNP)Co]-catalyzed borylation was suppressed, the overall activity and selectivity of the borylation of fluoroarenes was reduced as compared to the previously reported [(PNP)Co] catalyst lacking isopropylene spacers. Stoichiometric reactions support an increased barrier for oxidative addition to cobalt(I), a result of the increased steric profile and decreased conformational flexibility of the pincer resulting from methylation distal to the active site. With a more activated substrate such as benzofuran, catalytic borylation with cobalt(I) precatalysts and HBPin was observed. Monitoring the progress of the reaction by NMR spectroscopy revealed the presence of cobalt(III) intermediates during the course of the borylation, supporting a cobalt(I)-(III) redox cycle.

Graphical Abstract

INTRODUCTION

Metal-catalyzed methods for selective C(sp²)–H functionalization are attractive strategies for the synthesis of ubiquitous organic molecules or late-stage active pharmaceutical ingredients.^1–7 Among different functionalization reactions, the borylation of C(sp²)–H bonds is particularly powerful due to the synthetic versatility of the organoboron products.^4,8–10 Following the pioneering work of Smith, Maleczka, and Hartwig where borylation of arenes using [(η⁵-C₅Me₅)Ir] precatalysts was demonstrated,^11–13 iridium catalysts for direct borylation of arenes and heteroarenes with increased rates and high yields have been extensively studied including iridium(I) complexes with phosphine¹⁴ or bipyridine- and phenanthroline-derived ligands.^15–20 The combination of [Ir(OMe)(COD)]₂ (COD = 1,5-cyclooctadiene) and bipyridine or phenanthroline ligands is now one of the most widely used catalysts for C(sp²)–H borylation. These catalysts exhibit high activity and often predictable site-selectivity; where sterically accessible arene and heteroarene C(sp²)–H bonds are functionalized preferentially.²¹ In substrates that have multiple sterically accessible bonds, statistical mixtures of functionalized products are obtained. For example, 1:1 ortho:meta selectivity was observed for (dtbpy)Ir-catalyzed C–H borylation of 1-fluoro-3-(trifluoromethyl)benzene (dtbpy = 4,4’-di-tert-butyl-2,2’-bipyridine, Figure 1A).²²

Figure 1.

State-of-the-art iridium and cobalt (pre)-catalysts for C(sp²)–H borylation.

Our group has reported cobalt catalysts for both C(sp²)–H^23–28 and C(sp³)–H borylation.²⁹ Among them, cobalt catalysts supported by [^iPrPNP] pincers (^iPrPNP = 2,6-(ⁱPr₂PCH₂)₂(C₅H₃N)) are most effective for C(sp²)–H borylation of arenes and heteroarenes, often with distinct site selectivity.^23–25 For the borylation of 1-fluoro-3-(trifluoromethyl)benzene with B₂Pin₂ (Pin = pinacolato), [(^iPrPNP)Co] exhibits high selectivity towards ortho-to-F sites (95:5 ortho:meta, Figure 1B).²² To gain insight into the mechanism of action and the origin of the electronically driven site selectivity, experimental and computational studies were carried out and notable features are summarized in Figure 2. One potential catalytic intermediate, (^iPrPNP)CoH undergoes P–C bond cleavage to generate (^iPrPNP)Co(H)(PHⁱPr₂) in a catalyst deactivation pathway (Figure 2A).²⁵ Mechanistic studies established that the oxidative addition of C(sp²)–H bonds of certain electron-poor, fluorinated arenes are fast and reversible, and under thermodynamic control. The differences in Co–C bond dissociation free energies between sites is greater than the corresponding C(sp²)–H bond strengths, giving rise to the ortho fluorine effect and ultimately the high ortho-to-fluorine selectivity for C–H borylation (Figure 2B).²⁶

Figure 2.

Notable mechanistic features of [(^iPrPNP)Co]-catalyzed C(sp²)–H borylation.

DFT calculations established that conformational preferences of the methylene spacers between the pyridine and the phosphine donors influence fundamental transformations relevant to catalysis (Figure 2C).³⁰ The pincer ligand interconverts between folded (cisoid) and unfolded (transoid) geometries by rotating the phosphine arms. The flexibility of the pincer ligand allows for interaction with the arene substrate by spreading apart the isopropyl groups on phosphines (in cisoid conformations) and stabilizes octahedral cobalt(III) intermediates by minimizing steric hindrance (in transoid conformations). While phosphine arm dissociation,³¹ σ-bond metathesis,^32–34 and ligand de- and re-aromatization^35,36 were considered as alternative mechanisms for catalytic turnover, computational studies ruled out these scenarios.

Based on the computational insights and the identification of P–C bond cleavage as a catalyst deactivation pathway, substitution of the methylene spacers on the PNP was targeted to explore effects on catalyst stability, activity and selectivity. The tetramethylated variant, Me₄^iPrPNP (2,6-(ⁱPr₂PCMe₂)₂(C₅H₃N)), where all of the benzylic hydrogens are replaced with methyl groups, was targeted (Figure 1C). This ligand was previously synthesized and reported by Khusnutdinova and coworkers and applied to the isolation and characterization of 4-coordinate Ni(I) and Ni(II) compounds.^37,38 Pincer methylation was introduced to suppress metal-ligand cooperation a well as to provide additional steric protection of the metal with the goal of influencing both the geometry and stability of the resulting nickel complexes.

The increased steric profile and rigidity imparted by benzylic methylation was hypothesized to raise the barrier of oxidative addition which in turn could render selectivity under kinetic rather than thermodynamic control. Here we describe the synthesis of a series of cobalt compounds supported by Me₄^iPrPNP ligand and their application to catalytic C(sp²)–H borylation of fluoroarenes and benzofuran. While catalyst deactivation by P–C bond cleavage was suppressed, so was overall activity, providing insights on the impact of sterics and flexibility of ligands on catalytic reactivity.

RESULTS AND DISCUSSION

Our studies commenced with the synthesis of [(Me₄^iPrPNP)Co] complexes relevant to C(sp²)–H borylation (Scheme 1). The synthesis of the free ligand was accomplished according to the reported procedure.^37,39 Analytically pure 1-Cl₂ was obtained by straightforward metalation with CoCl₂ and the subsequent washing with pentane as a dark olive powder in 89% yield. A solid-state magnetic moment of 4.1(1) μ_B (23 °C, magnetic susceptibility balance) was measured for 1-Cl₂, consistent with an S = 3/2 ground state. Treatment of 1-Cl₂ with one equivalent of NaEt₃BH followed by recrystallization from a 1:9 mixture of toluene and pentane afforded (Me₄^iPrPNP)CoCl, 1-Cl as a dark brown solid in 79% isolated yield. Similar to (^iPrPNP)CoCl which has an S = 1 ground state, a solid-state magnetic moment at 23 °C produced a value of 2.9(1) μ_B, consistent with the spin-only value for two unpaired electrons and an S = 1 ground state.

Scheme 1. Synthesis of (Me4iPrPNP)Cobalt Complexes.^a.

^aSee Supporting Information for reaction conditions.

Methylation of 1-Cl was accomplished by addition of a stoichiometric quantity MeLi and subsequent filtration and recrystallization from pentane furnished 1-Me as a brown solid in 78% yield. The ¹H and ³¹P{¹H} NMR spectra in benzene-d₆ exhibit resonances consistent with a diamagnetic compound with idealized C_2v symmetry. A diagnostic Co–CH₃ resonance was located as a triplet (³J_P–H = 9.6 Hz) centered at –1.41 ppm while a sharp ³¹P resonance was observed at 71.6 ppm. Direct conversion from 1-Cl₂ to 1-Me was also achieved by addition of two equivalents of MeLi followed by filtration.

The solid-state structures of 1-Cl₂, 1-Cl, and 1-Me were determined by single-crystal X-ray diffraction and representations of the molecular structures are presented in Figure 3. The geometry of 1-Cl₂ is best described as distorted trigonal bipyramidal with the two phosphines occupying axial positions and the nitrogen of the pyridine, the two chlorines defining the equatorial plane (Figure 3A). Dihedral angles between N–Co–P(1)–C(1) and N–Co–P(2)–C(7) are 31.36(7)° and –30.92(6)° respectively, confirming that two spacers of the PNP pincer ligand are folded in different directions. The (Me₄^iPrPNP)CoCl compound, 1-Cl, has a pseudo-tetrahedral structure, similar to the geometry observed with (^iPrPNP)CoCl,⁴⁰ although the cobalt and chlorine of 1-Cl appear to be disordered (Figure 3B). In 1-Cl as well as (^iPrPNP)CoCl, the methylene spacers of the pincer are in a cis configuration.

Figure 3.

Solid-state structures of 1-Cl₂, 1-Cl, 1-Me, and (^iPrPNP)CoMe at 30% probability ellipsoids. Hydrogen atoms except for those at the åbenzylic positions omitted for clarity.

For comparison, (^iPrPNP)CoMe was prepared following published procedures^40–42 and recrystallization from a concentrated pentane solution at –35 °C furnished dark-brown crystals suitable for X-ray diffraction. A representation of the molecular structure, depicted in Figure 3D, established a planar geometry about the cobalt and confirmed that the pincer ligand remained intact.⁴³ Both (^iPrPNP)CoMe, and 1-Me are planar, consistfent with a diamagnetic ¹H NMR spectrum (Figure 3C,D). Despite the similar planar geometries, a significant distortion is found at the both spacers of PNP pincer ligand of 1-Me, whereas in (^iPrPNP)CoMe only one of two methylene spacers deviates from the idealized P–N–P pincer plane (dihedral angles of Co–P(2)–C(9)–C(1) and Co–P(1)–C(6)–C(5) are 31.62(11)° and 29.51(11)°, respectively and the corresponding values in (^iPrPNP)CoMe are –8.25(10)° and –25.56(9)°). This observation established the steric impedance imposed by four methyl groups of the methylene spacer of the pincer and enabled quantification of the steric profiles of modified- and non-modified PNP ligands, which will be discussed below. The solid-state structure of 1-Me clearly established trans-arrangement of two methylene spacers. The ligand conformations varied depending on the environment around the cobalt center. In planar geometry (1-Me) or with higher coordination number (1-Cl₂), the modified pincer existed as a transoid conformer, whereas a cisoid conformation is observed with a 4-coordinate distorted tetrahedral complex (1-Cl), implying that reactions at the metal center would be accompanied by pincer arms flipping.

To investigate the impact of pincer modifications on the overall steric properties of the cobalt alkyl complexes, the solid-state structures of (^iPrPNP)CoMe, 1-Me, and (^tBuPNP)CoMe (^tBuPNP = 2,6-(^tBu₂PCH₂)₂(C₅H₃N)) were analyzed and the steric maps of those ligands were generated using SambVca 2.1 (Figure 4).⁴⁴ The plot illustrates that the upper part of the metal coordination sphere is mostly shielded by pincer substituents but the openness of the southern hemisphere varies depending on the specific pincer. Lower percent buried volume (%V_bur) was observed with ^iPrPNP ligand, corresponding to the high level of flexibility of the pincer ligand, which in turn reflected to the highest catalytic activity for C(sp²)–H borylation. Introduction of tert-buryl substituents on the phosphines, by contrast, effectively shielded the cobalt and reduced open space in all quadrants. Such difference rendered the ^tBuPNP ligand more rigid, as a significant crashes between tert-butyl substituents would be accompanied by phosphine arms flipping. Accordingly, (^tBuPNP)CoMe was inactive for borylation of fluoroarenes even at 80 °C.^45,46 Steric hindrance from isopropylene spacers of Me₄^iPrPNP ligand forces the isopropyl groups on the phosphines to locate away from the PNP pincer ligand, thus resulting in a less open coordination sphere, akin to [(^tBuPNP)Co].

Figure 4.

Steric maps of (^iPrPNP)Co, (Me₄^iPrPNP)Co, and (^tBuPNP)Co generated from the solid-state structures for the corresponding methyl complexes generated with SambVca2.1. The Co atom defines the center of the xyz coordinate system. The P–N–P plane defines the xz-plane, and the N–Co line defines the z-axis. Colors indicate occupied space toward (-z, blue) or away from (+z, red) the PNP pincer ligand. Total percent buried volume (%V_bur) and local percent buried volume listed for each quadrant of a sphere of radius, r = 3.5 Å. Bondi radii scaled by 1.17 Å and mesh spacing for numerical integration set to 0.1 Å.

With a route to 1-Me in hand, stoichiometric reactions with H₂, B₂Pin₂, and HBPin were carried out to generate potential precatalysts or intermediates for C(sp²)–H borylation and to explore fundamental oxidative addition chemistry (Scheme 2). Exposure of a benzene-d₆ solution of 1-Me to 4 atm of H₂ generated the cobalt(III) trihydride, 1-H₃ within 15 minutes, demonstrating the propensity to participate in two-electron oxidative addition of H₂. Stirring a diethyl ether solution of 1-Me under 4 atm of H₂ in a thick-walled reaction vessel followed by filtration and recrystallization from Et₂O at –35 °C resulted in the isolation of 1-H₃ as a dark-orange crystalline solid in 86% yield (Scheme 2A, top). The benzene-d₆ NMR spectrum recorded at ambient temperature exhibits a broad singlet centered at −13.69 ppm, assigned to the cobalt hydrides that are undergoing exchange on the NMR timescale. The solid-state structure of 1-H₃ was determined by X-ray diffraction (Figure 5) and the metrical parameters are indistinguishable from those reported for (^iPrPNP)CoH. Unlike (^iPrPNP)CoH which undergoes instantaneous formation of (^iPrPNP)Co(H)(PHⁱPr₂), 1-H₃ proved stable in the absence of H₂ with no evidence for P–C bond cleavage (Figure S1–S3).

Scheme 2.

Reactivity of (Me₄^iPrPNP)CoMe.

Figure 5.

Representation of molecular structure of 1-H₃ at 30% probability ellipsoids. Hydrogen atoms except those attached to cobalt omitted for clarity. Each of the cobalt hydrides was located and freely refined.

In contrast to H₂, the oxidative addition of B–B bonds proved more challenging. Addition of one equivalent of B₂Pin₂ to a THF-d₈ solution of 1-Me produced no reaction after 24 hours at 23 °C as judged by ¹H and ³¹P NMR spectroscopies (Scheme 2A, middle). However, treatment of 1-Me with two equivalents of HBPin produced 1-(H)₂BPin along with a trace but detectable amounts of 1-H₃. For 1-(H)₂BPin, a broad ³¹P resonance was observed at 127.1 ppm, supporting the formation of a cobalt(III) product. Two resonances were observed in the hydride region of ¹H NMR spectrum; one broad signal at –8.29 ppm assigned to 1-(H)₂BPin and the other at –13.71 ppm for 1-H₃. Attempts were made to prepare 1-(H)₂BPin directly from 1-Cl₂ from addition of two equivalents of NaHBEt₃ to a suspension of 1-Cl₂ in the presence of 4 equivalents of HBPin. Unlike with (^iPrPNP)CoCl₂ where the cobalt(III)–dihydride–boryl product is formed,²⁴ decomposition was observed.

Although the previously reported method to synthesize 1-BPin by addition of B₂Pin₂ to of 1-Me was unsuccessful, 1-BPin was generated by heating a mixture of the cobalt-methyl derivative in the presence of one equivalent of B₂Pin₂ and 1-fluoro-3-(trifluoromethyl)benzene (2a) at 80 °C for one hour (Scheme 2B). The role of fluoroarene in synthesis of 1-BPin is yet unclear, but its presence is essential as heating the reaction mixture of 1-Me and B₂Pin₂ in THF-d₈ produced no discernable change in ¹H and ³¹P NMR spectra in the absence of arene (Scheme 2A, middle, see Figure S12–S13). The reactivity of 1-Me towards 2a in the absence of B₂Pin₂ was also examined, but both 1-Me and 2a remained intact for 24 hours at 80 °C, as evidenced by ¹H, ¹⁹F, and ³¹P NMR spectroscopy (see Figure S14–S16).

The catalytic performance of the isolated cobalt complexes was examined for the C–H borylation of 1-fluoro-3-(trifluoromethyl)benzene using B₂Pin₂ as the boron source (Scheme 3A). No reaction was observed at ambient temperature, in contrast to (^iPrPNP)Co(CH₂SiMe₃) where full conversion to borylated product was observed. Increasing the temperature of the reaction to 50 °C resulted in borylated product as assayed by GC but the yield was low (< 5% conversion). Performing the catalytic reaction at 80 °C resulted in complete conversion to product and produced a 51:49 mixture of ortho:meta substituted boronate ester products. Based on this observation, 1-H₃, 1-BPin, and 1-(H)₂BPin were used as precatalysts at 80 °C and proved competent for C(sp²)–H borylation of 2a and yielded 1:1 mixtures of ortho- and meta-to-F borylated products.

Scheme 3. C(sp²)–H Borylation of Fluoroarenes with [(Me4iPrPNP)Co] Precatalysts.^a.

^aReaction conducted with 0.55 mmol of fluoroarene (2a–d), 0.55 mmol of B₂Pin₂, 5 mol% of precatalyst in 1 mL of THF at the specified temperature for 24 h. Conversions and ortho:meta regioselectivities determined by GC using cyclooctane as an internal standard or by ¹⁹F NMR spectroscopy using 4-fluorotoluene as an internal standard.

The scope of the catalytic borylation reaction was explored with other fluorinated arenes with varied electronic properties (Scheme 3B). Each catalytic experiment utilized 5 mol% of 1-Me and B₂Pin₂ as the boron source. This specific cobalt precursor was selected because of its straightforward synthesis and purification. With the relatively electron poor sulfonamide arene 2b, similar conversion and regioselectivity to 2a were observed (>95% conversion and 49:51 o:m selectivity). However, arenes 2c and 2d produced significantly lower conversions (24% and 12%, respectively) and with slight deviation from 1:1 ortho:meta product mixtures (63:35 and 43:57, respectively). The regioselectivity observed with 2d regioselectivity is similar to the [Ir(COD)OMe]₂/dtbpy catalyst which exhibits about 40:60 ortho:meta product ratio.²²

The previously proposed mechanism for (^iPrPNP)Co-catalyzed C–H borylation is presented in Scheme 4A. The cobalt(I)–boryl complex, generated by reductive elimination of H₂ from (^iPrPNP)Co(H)₂BPin precatalyst, undergoes fast and reversible oxidative addition of fluorinated arenes.²⁶ Aryl and boryl groups are trans to each other in the resulting cobalt(III)–hydride–boryl–aryl complex. Instead of C–B bond formation, reductive elimination of HBPin from the Co(III) intermediate occurs and generates a cobalt(I)–aryl intermediate, which then reacts with HBPin to afford cobalt(III) complex with aryl and boryl groups cis to each other. Reductive elimination of the borylated product from the cobalt complex forms the cobalt(I) hydride which can regenerate the C–H activating cobalt(I)–boryl complex upon addition of B₂Pin₂ and loss of HBPin.

Scheme 4. Reactivity of [(Me4iPrPNP)Co] Complexes.

^aSee Supporting Information for reaction conditions. ^b1-Ar^F contained 30% of 4,4’-difluorobiphenyl.

To determine if an analogous pathway is operative with the [(Me₄^iPrPNP)Co] catalyst, stoichiometric reactions were performed (Scheme 4B). The fluorinated arene, 2a was added to a THF-d₈ solution of 1-BPin to evaluate its potential as an intermediate for C–H activation. No reaction was observed upon mixing 1-BPin with 2a at ambient temperature; decomposition to an intractable mixture within an hour upon heating at 80 °C. Because attempts to prepare a cobalt(I)–aryl complex from 1-BPin were unsuccessful, an alternate route involving transmetalation of a Grignard reagent to 1-Cl was explored. Relevant to catalysis, the addition of 3-fluoro-5-(trifluoromethyl)phenylmagnesium bromide to 1-Cl was conducted but the targeted cobalt(I)–aryl complex was not obtained as no reaction was observed. Instead, a less sterically hindered and ideally more reactive Grignard reagent, 4-fluorophenylmagnesium bromide was used, where the 4-fluoro substituent was used as an ¹⁹F NMR probe. Addition of one equivalent of the Grignard reagent to a thawing ether solution of 1-Cl produced 1-Ar^F along with inseparable biaryl byproducts arising from homocoupling, as evidenced by two distinct resonances at −129.22 ppm (1-Ar^F) and −115.71 ppm (4,4’-difluorobiphenyl) in ¹⁹F NMR spectrum.

With a cobalt(I)–aryl complex in hand, its reactivity toward boron reagents was studied. To a benzene-d₆ solution of 1-Ar^F, one equivalent of B₂Pin₂ was added and the progress of the reaction was monitored by ¹H, ¹⁹F, and ³¹P NMR spectroscopies. No reaction was observed over extended periods at 23 °C or 80 °C. The reactivity of 1-Ar^F with HBPin was also studied. Addition of four equivalents of HBPin to 1-Ar^F produced no reaction at 23 °C but heating the reaction mixture to 80 °C resulted in decomposition of cobalt complex as judged by the appearance of multiple resonances in the ³¹P NMR spectrum. The only observable compounds in the ¹⁹F NMR spectrum were 4,4’-difluorobiphenyl (from the synthesis of 1-Ar^F) and fluorobenzene with no evidence of the formation of p-fluorophenylboronic acid pinacol ester.

These results contrast previous studies with 4-substituted PNP-supported cobalt catalysts.²⁶ The cobalt(I)–boryl complex with the 4-pyrrolidinyl ^iPrPNP pincer undergoes rapid C(sp²)–H oxidative addition upon the addition of 2a, forming the corresponding cobalt(I)–aryl complex. In addition, the cobalt(I) aryl with 4-methyl ^iPrPNP ligand reacted instantaneously with B₂Pin₂ or HBPin and yielded the aryl boronate ester. The disparity between the reactivity of (^iPrPNP)cobalt complexes and those of (Me₄^iPrPNP)cobalt complexes demonstrates that methylation of the methylene spacers raises the barrier for oxidative addition with cobalt(I). The inability to observe any stoichiometric reactions involving [(Me₄^iPrPNP)Co(I)] complexes raised the possibility of heterogeneity of the catalytic species in C–H borylation of fluoroarenes. A filtration test was carried out to explore this possibility and no change in activity or selectivity were observed following filtration after 2 hours of reaction time. While consistent with a molecular catalyst, this experiment does not definitively exclude the possibility of nanoparticles or a heterogeneous catalyst.^47,48

The poor catalytic borylation activity of 1-BPin prompted assessment of the C–H activating ability of the (Me₄^iPrPNP)cobalt(I) hydride, 1-H. As complete conversion of 1-Me to 1-H₃ was observed upon addition of 4 atm H₂, in situ generation of 1-H from 1-H₃ may access a more reactive cobalt(I) compound for C(sp²)–H oxidative addition. A THF solution of 1-Me was exposed to 1 atm D₂ in the presence of an excess of 2a and vigorously stirred at 80 °C for 24 hours. Deuterium incorporation into the ortho- and meta-to-F sites of the arene was assayed by ¹H NMR spectroscopy. An identical reaction was also set up with (^iPrPNP)CoMe instead of 1-Me for comparison. As shown in Scheme 5, (^iPrPNP)CoMe predominantly exchanged deuterium into the ortho-to-F C–H sites (84% D) over those meta to the fluorine (37% D). However, with 1-Me, no deuterium incorporation into 2a was observed. One possibility is that under these conditions, 1-H₃ fails to generate 1-H by H₂ reductive elimination, consistent with the stability of 1-H₃ under N₂ or vacuum. The other possibility is that 1-H is formed but does not promote C(sp²)–H oxidative addition. In either case, substitution at the methylene spacers of the pincer results in an increased barrier for conversion between Co(I) and Co(III) complexes.

Scheme 5. Hydrogen Isotope Exchange (HIE) Experiments with Fluorinated Arenes and D₂ Gas.^a.

^aReaction conditions: 1-fluoro-3-(trifluoromethyl)benzene (0.50 mmol), 1-Me (0.025 mmol, 5 mol%), THF (1 mL) under 1 atm D₂, 23 °C, 24 h. Percent deuterium incorporations determined by ¹H NMR spectroscopy.

Efforts were made to identify the resting state during catalytic C(sp²)–H borylation. A THF-d₈ solution containing 1-Me, 2a, and B₂Pin₂ was prepared in a J-Young tube, and the reaction mixture was heated to 80 °C and monitored by ¹H, ¹⁹F, and ³¹P NMR spectroscopies as a function of time. During the course of the experiment, formation of the boronate ester products was confirmed by ¹⁹F NMR spectroscopy and a 1:1 ratio of the ortho- and meta-to-F products was maintained throughout (Figure S47). Identification of the cobalt compounds present during catalysis was complicated by the broadening of the ¹H NMR signals, likely from formation of small amounts of particulate cobalt (Figure S45). The ³¹P NMR spectra recorded over the same time intervals had several peaks emerge and disappear over the course of 24 hours, none of which correlated to a known cobalt compound (Figure S46). Monitoring of the reaction progress by GC confirmed that there was an induction period required for conversion of precatalyst 1-Me to an active catalyst.

Because the cobalt complexes supported by Me₄^iPrPNP offered few observables during the catalytic borylation of 2a, attention was devoted to the borylation of benzofuran as the oxidative addition was expected to be faster and may produce cleaner organometallic chemistry. As depicted in Scheme 6, in the presence of 5 mol% of 1-Me and one equivalent of HBPin, benzofuran undergoes borylation at 23 °C in THF.

Scheme 6. (Me4iPrPNP)Co-catalyzed C–H Borylation of Benzofuran.^a.

^aReaction conditions: benzofuran (0.55mmol), HBPin (0.55 mmol), 1-Me (0.0275 mmol, 5 mol%), THF (1 mL), 23 °C, 24 h. Conversion determined by GC using cyclooctane as an internal standard.

Monitoring the reaction between 1-BPin and 3 equivalents of benzofuran in THF-d₈ by NMR spectroscopy revealed complete conversion of cobalt boryl to the corresponding benzofuranyl compound, 1-Bf in 15 minutes (Scheme 7A). After 6 hours, the cobalt was present at 1-Bf and (2-benzofuranyl)BPin was detected. A plausible pathway for this transformation is presented in Scheme 7B, where reaction of 1-Bf and HBPin generates the borylated benzofuran and the cobalt hydride, 1-H, which promotes rapid C–H activation of benzofuran and regenerates 1-Bf. This supports that both 1-H, and 1-BPin are capable of C(sp²)–H activation.

Scheme 7. Observation of 1-Bf During the Cobalt-Catalyzed Borylation of Benzofuran and Proposed Reaction Pathway.

^aReaction conditions: 1-BPin (0.012 mmol), benzofuran (0.036mmol, 3 equiv.), THF-d₈ (0.8 mL), 23 °C, 6 h. ^bConversion and yield calculated with respect to 1-BPin. >95% conversion to 1-Bf. >95% yield of (2-benzofuranyl)BPin. 90% benzofuran left in the solution. Bf = 2-benzofuranyl

The progress of the cobalt-catalyzed C(sp²)–H borylation of benzofuran with HBPin was monitored by ¹H and ³¹P NMR spectroscopies to determine the catalyst resting state (Figure 6). After 30 minutes of reaction time, corresponding to <5% conversion of (2-benzofuranyl)BPin, 1-Bf was observed as the principal cobalt compound with a trace amount of 1-(H)₂BPin and 1-H₃. As the reaction proceeded, 1-H₃ accumulated accompanied by a decrease in the concentration of 1-Bf. Notably, catalyst deactivation by P–C bond cleavage of the pincer ligand, often observed with (^iPrPNP)Co system to generate catalytically inactive (^iPrPNP)Co(H)(PHⁱPr₂), was absent, suggesting that introduction of the methyl groups in the methylene spacers successfully suppressed this deleterious reaction. Having identified 1-H₃ and 1-Bf as the major cobalt compounds formed during catalytic turnover, the reactivity of these compounds was investigated (Scheme 8). Addition of excess benzofuran to a THF-d₈ solution of 1-H₃ resulted in 80% conversion to 1-Bf as judged by ¹H and ³¹P NMR spectroscopies. The reaction was driven to complete conversion by repeated freeze-pump-thaw cycles. Rapid deuterium incorporation was observed at the C2 position of benzofuran upon exposure of a THF-d₈ solution of 1-Bf to 1 atm D₂, supporting fast and reversible C(sp²)–H activation of benzofuran. By contrast, addition of HBPin to a solution of 1-Bf resulted in gradual formation of borylated product over the course of 24 h. This observation supports a relatively high barrier for the reaction with HBPin likely due to the increased steric profile of the pincer ligand.

Figure 6.

³¹P NMR spectra of the reaction mixture of benzofuran (0.10 mmol), HBPin (0.11 mmol), and 1-Me (0.005 mmol, 5 mol%) in THF-d₈ (0.8 mL) as a function of time.

Scheme 8. Stoichiometric Reactions.

^aConversions determined by ³¹P NMR spectroscopy. ^bPercent deuterium incorporation determined by ¹H NMR spectroscopy with respect to 1-H₃. The cobalt(III) deuteride, 1-D₃ was observed as the sole organometallic product. ^cYield determined by ¹H NMR spectroscopy with respect to 1-H₃. A mixture of cobalt complexes was observed including: 1-Bf (73%), 1-(H)₂BPin (15%), and 1-H₃ (12%).

Based on the experimental data, a plausible mechanism for the cobalt-catalyzed C–H borylation of benzofuran is presented in Scheme 9. The cobalt alkyl, 1-Me enters the catalytic cycle forming 1-(H)₂BPin, which undergoes isomerization and reductive elimination of H₂ to generate 1-BPin. While 1-BPin is unreactive toward fluoroarenes at ambient temperature, this compound proved more reactive with benzofuran as 1-Bf is formed rapidly. In the presence of H₂, the 1-Bf compound is in equilibrium with 1-H and benzofuran, supported by a deuterium labelling experiment. By contrast, the reaction of 1-Bf with HBPin to liberate the borylated product and 1-H proceeds slowly. The cobalt hydride undergoes rapid oxidative addition with either H₂ or HBPin to generate 1-H₃ or 1-(H)₂BPin respectively, accounting for the observation of these compounds along with 1-Bf during the catalysis.

Scheme 9.

Proposed Mechanism for the Catalytic Borylation of Benzofuran.

CONCLUSIONS

Methylation of the methylene positions in a PNP-pincer ligand was explored in the context of cobalt-catalyzed C(sp²)–H borylation. A family of cobalt(I) compounds including chloride, methyl, hydride, benzofuranyl and aryl derivatives were synthesized and structurally characterized. The catalytic performance of this class of cobalt precatalysts in C(sp²)–H borylation of fluoroarenes was assayed. While methylation of the pincer suppressed P–C bond cleavage, a known catalyst deactivation pathway with [(^iPrPNP)Co] catalysts, diminished activity and selectivity in borylation was observed as a statistical mixture of ortho- and meta-to-F boronate ester products was obtained. The reduced activity is likely due to the slightly increased steric profile and the concomitant descreased flexibility of the modified pincer ligand raises the barrier for C(sp²)–H oxidative addition. Determination of the buried volume of three pincer cobalt alkyl complexes supports this assertion. Borylation of benzofuran, a more activated substrate, was achieved and mechanistic investigations support a pathway involving a cobalt(I)-(III) redox cycle with observable cobalt(III) intermediates.