Development of Cobalt Catalysts for the meta-Selective C(sp²)–H Borylation of Fluorinated Arenes

Abstract

Cobalt precatalysts for the meta-selective borylation of fluorinated arenes are described. Initial screening and stoichiometric reactivity studies culminated in the preparation of a cobalt alkyl precatalyst supported by the sterically protected terpyridine (5,5″-Me₂^ArTpy = 4′-(4-N,N′-dimethylaminophenyl)-5,5″-dimethyl-2,2′:6′,2″-terpyridine). Under the optimized conditions, borylation with this precatalyst afforded up to 16 turnovers and near-exclusive meta regioselectivity with a range of substituted fluoroarenes in cyclopentyl methyl ether solvent at room temperature. Deuterium kinetic isotope effects of 2.9(2) at 23 °C support a turnover-limiting and selectivity-determining C(sp²)–H activation step, and stoichiometric C–H activation experiments provided insights into the identity of the C–H activating intermediate in catalysis. Analysis of the relevant Co–C and C–H bond thermodynamics support that the thermodynamics of C–H activation favor ortho-to-fluorine selectivity, providing additional, indirect support for kinetic control of C–H activation as the origin of meta selectivity.

Graphical Abstract

INTRODUCTION

Methods for the functionalization of carbon–hydrogen (C–H) bonds are an active area of research with demonstrated progress ranging from sustainable chemistry to complex molecule synthesis.¹ Of the established methods, C–H borylation is among the most impactful due to the synthetic versatility of the readily handled arylboronate products.² Notably, C–H borylation is unique among C–H functionalization reactions in that borylation methods generally do not require directing groups nor superstoichiometric quantities of substrate to achieve synthetically useful reactivities.³ Originally independently pioneered by the groups of Smith⁴ and Hartwig⁵ for catalytic C(sp²)–H borylation, significant progress in iridium catalyst design over the past two decades has resulted in improved performance, including competent catalysts for more challenging C(sp³)–H borylation.⁶ Experimental^5c,7 and computational⁸ investigations of bipyridine iridium-catalyzed C(sp²)–H borylation support a pathway involving turnover limiting C–H oxidative addition to an Ir(III) triboryl to furnish iridium(V) intermediates en route to organoboronate products. Site selectivity with the most widely used catalysts is sterically controlled, occurring at positions that are not ortho to a methyl group or larger substituent.⁹

While highly active catalysts have been discovered and used on large scale, sterically-driven site selectivity imposes a limitation for the functionalization¹⁰ of substrates with multiple sterically accessible C–H bonds. When these sites are inequivalent, this method is predisposed to produce mixtures of organoboronate products, often requiring separations and generating waste. Fluorinated arenes are a valuable class of substrates due to their ubiquity in the pharmaceutical industry resulting from the impact of fluorine substitution on the lipophilicity and metabolic properties of the molecule.¹¹ While thoughtfully engineered directing groups,¹² templating strategies,¹³ and secondary substrate-catalyst interactions¹⁴ have enabled methods that are selective for a given position relative to specific functional groups, these methods are not readily extended to site selective C(sp²)–H borylation of aryl fluorides due to the small atomic radius and poor coordinating ability of fluorine. The realization of regioselective methods for the functionalization of fluorinated arenes requires advancement of the underdeveloped concept of electronic control of site selectivity in C–H activation and functionalization, which would ultimately enable methods and synthetic strategies complementary to established iridium-catalyzed methods.

Cobalt-catalyzed borylation methods have been discovered that offer regioselective chemistry distinct from iridium catalysts.¹⁵ Use of bis(phosphino)pyridine (PNP) pincers produced catalysts with high ortho to fluorine selectivity including examples that are functionalized indiscriminately by iridium catalysts to yield statistical mixtures of products.^15c,16 Mechanistic studies¹⁷ established that the ortho to fluorine selectivity arises from rapid C(sp²)–H oxidative addition to the cobalt(I) pincer that generates cobalt(I)-aryl intermediates arising from a thermodynamically-controlled “ortho-to-fluorine” effect.¹⁸ Subsequent studies with the same pincer cobalt catalyst demonstrated that subtle changes to the substrate result in dramatic changes to the regioselectivity due to alterations of fundamental steps in the catalytic cycle.¹⁹ For example, 2,6-difluorophenylboronate pinacol ester undergoes turnover-limiting C(sp²)–H oxidative addition and kinetic control of this step results in a preference for borylation of the position para to the boronate ester (and meta to both fluorines) (Scheme 1a). Replacement of one of the fluoride substituents with a more electron-withdrawing trifluoromethyl²⁰ substituent sufficiently accelerates C(sp²)–H oxidative addition relative to subsequent steps of the catalytic cycle to the extent that it is no longer turnover-limiting, restoring ortho-to-fluorine regioselectivity.¹⁹

Scheme 1.

(a) Previous Demonstration of Regioselectivity Inversion in Cobalt-Catalyzed Fluoroarene Borylation with Varied Substrate Electronics and (b) Catalyst-Controlled Regioselectivity Inversion to Afford meta Selectivity in Fluoroarene Borylation (this work).

The observation of substrate-controlled regioselectivity highlights an impactful subtlety of [(^iPrPNP)Co] catalysts. Namely, the activation barriers associated with the slowest steps (e.g., C–H oxidative addition or C–B reductive elimination) are close in energy such that orthogonal regioselectivities may be observed with similar substrates due to a change in the turnover limiting step.²¹ This realization inspired extension of this concept to catalyst-controlled regioselectivity whereby a cobalt catalyst bearing a less electron donating pincer would enable turnover-limiting and selectivity-determining C(sp²)–H activation; if operative such a pathway may result in meta-to-fluorine regioselectivity using the same substrates where ortho selectivity was observed with [(^iPrPNP)Co]. Notably, 1,3-disubstituted and 1,3,5-trisubstituted benzenoid rings are among the least represented arene substitution patterns in drug candidates, a trend that has been attributed to the lack of meta-selective synthetic methods.²² Therefore, general catalyst design principles that translate on to meta selective C–H functionalization methods are attractive for their potential to access underrepresented substructures in drug discovery.

Here we describe the rational application of relatively electron poor terpyridine-based cobalt complexes to promote meta selective C(sp²)–H borylation of fluorinated arenes (Scheme 1b). Precatalyst evaluation and stoichiometric reactivity studies culminated in the introduction of a highly regioselective (>20:1 in most cases) catalyst which was readily prepared from inexpensive, commercially available starting materials. Deuterium kinetic isotope effect (KIE) experiments supported turnover limiting C–H activation, and a single-turnover experiment established that C–H activation is responsible for the observed selectivity rather than subsequent steps in the catalytic cycle.

RESULTS AND DISCUSSION

Initial Precatalyst Selection and Optimization of Conditions.

Our studies commenced with catalytic borylation of fluoroarene 1a with the (terpyridine)Co(II)-bis(carboxylate) complex 3-(OAc)₂ as the precatalyst (Table 1). This compound was selected due to its previously demonstrated activity for arene borylation,²³ as well as for the reduced electron-donating ability compared to the analogous phosphine-based pincer.¹⁷ In addition, P–C bond cleavage has been identified as a deactivation pathway with [(PNP)Co]^15b and chelates lacking these linkages were of interest rather than variants of [(PNP)Co].

Table 1.

(a) Evaluation of precatalysts and conditions (b) Cobalt complexes studied.

Entry	[Co]	Solvent	Temp. (°C)	%conv.	m:o
1	3-(OAc) 2	CPME	80	23%	82:18
2	3-CH 2 SiMe 3	CPME	23	14%	92:8
3	4-CH 2 SiMe 3	CPME	23	42%	94:6
4	4-CH 2 SiMe 3	Et2O	23	49%	95:5
5	4-CH 2 SiMe 3	CyH	23	43%	94:6
6	4-CH 2 SiMe 3	THF	23	36%	93:7

Using conditions modified slightly from those previously reported for (Tpy)Co-catalyzed arene borylation,²³ stirring a cyclopentyl methyl ether (CPME) solution of 1a and B₂Pin₂ with 5 mol% of 3-(OAc)₂ at 80 °C produced 23% conversion to boronate ester favoring meta-2a in an 82:18 ratio. Inspired by the relatively high meta selectivity, attention was devoted to catalyst improvements to further increase selectivity and activity. Because the mode of activation of 3-(OAc)₂ is not well understood and known to produce deleterious boron-containing side products,²³ emphasis was placed on use of a well-defined terpyridine-based cobalt(I) precursor. The alkyl derivative, 3-CH₂SiMe₃ was evaluated for its catalytic borylation performance. At 23 °C, borylation of 1a using B₂Pin₂ with 5 mol% of 3-CH₂SiMe₃ produced 14% conversion to arylboronate products 2a with an improved 92:8 selectivity favoring the meta isomer. The conversion improved to 42% when 4-CH₂SiMe₃, bearing the commercially available 4,4′,4″-tri-tert-butyl-2,2′:6′,2″-terpyridine (^tBu₃Tpy) ligand, was used as the precatalyst in CPME. Analogous reactions in diethyl ether, cyclohexane, and THF exhibited similar activity (36–49% conversion) and regioselectivities (from 93:7 to 95:5 favoring meta-2a). The high selectivity and low yields of the reaction motivated investigations into mechanism and speciation of the cobalt complex to better understand the reaction and to improve turnover numbers. Combustion analysis of samples of 4-CH₂SiMe₃ provided evidence for the presence of small amounts of paramagnetic impurities that were not detected by ¹H and ¹³C NMR spectroscopy.

Synthesis and Stoichiometric Reactivity of Catalytically Relevant Complexes.

The independent synthesis and reactivity of metal-aryl and -boryl complexes in [(bipy)Ir]-^5c and [(^iPrPNP)Co]-catalyzed¹⁷ C(sp²)–H borylation, were foundational for providing mechanistic insights and for cobalt, insights into the site selectivity of the reaction. Analogous [(tpy)Co] examples were targeted with the goal of improving selectivity and turnover number of the catalytic borylation reactions. Derivatives of ^tBu₃Tpy were used because of their crystallinity and superior catalytic performance compared to 3-(OAc)₂ and 3-CH₂SiMe₃.

Addition of two equivalents of 4-fluorophenyl magnesium bromide to a suspension of 4-Cl₂ in Et₂O at −116 °C resulted in an initial color change from gray to green, followed by a more gradual color change to purple with formation of a darkly colored precipitate. The isolated solid contained a major component that was identified by ¹H and ¹⁹F NMR spectroscopies as the desired terpyridine cobalt(I)-aryl complex, 4-(p-C₆H₄F). The minor component of the mixture was identified as 4,4’-difluorobiphenyl by comparison of spectroscopic data to an authentic sample. Isolation of pure 4-(p-C₆H₄F) was accomplished by washing with cold (−35 °C) pentane. Single crystals were obtained by slow evaporation of a diethyl ether solution at −35 °C over two days and the solid-state structure was determined by X-ray diffraction (Scheme 2b). An idealized square planar geometry was observed with a N2–Co1–C28 angle of 179.27(15)° with the sum of cobalt–ligand bond angles equal to 360.00(14)°. Analysis of the bond distances of the pincer support a terpyridine ligand reduced by one electron and bound to a cobalt(II) center, consistent with the electronic structure previously proposed for terpyridine cobalt-alkyl complexes.^24,25

Scheme 2.

(a) Preparation and (b) solid-state structure of 4-(p-C₆H₄F) at 30% probability ellipsoids with hydrogens atoms omitted for clarity.

To assess whether isomerization events contribute to the meta selectivity observed in fluoroarene borylation, the reaction of 4-(p-C₆H₄F) with HBPin was investigated. Addition of two equivalents of HBPin to a benzene-d₆ solution of 4-(p-C₆H₄F) resulted in complete conversion to the para-borylated product as judged by ¹⁹F NMR spectroscopy, supporting the possibility that cobalt(I)-aryl complexes are cataytically competent (Scheme 3). No detectable quantities of the ortho or meta isomers of the boronate were observed spectroscopically, indicating that C–B bond formation occurred without isomerization. This stoichiometric reaction likely occurs by initial B–H oxidative addition to generate a Co(III) intermediate which then undergoes C–B reductive elimination, although a concerted sigma bond metathesis cannot be ruled out. This experiment supports that the regioselectivity in [(Tpy)Co]-mediated fluoroarene borylation originates with the initial formation of a Co–C bond and corresponds to the initial C–H activation event, ruling out that C–B bond formation plays a role in determining meta-to-fluorine regioselectivity in catalytic borylation.

Scheme 3.

Reaction of 4-(p-C₆H₄F) with HBPin.

Inspired by previous studies where iridium-^5c or cobalt-boryl^15b compounds are known to be the C–H activating species in catalytic borylation, the synthesis of a terpyridine cobalt–boryl complex was targeted. Addition of one equivalent of B₂Pin₂ to 4-CH₂SiMe₃ in THF generated an intractable mixture of products as judged by ¹H NMR spectroscopy (Figure S3). Attempts to isolate any component of the crude mixture for complete characterization were unsuccessful. Slow evaporation of a concentrated pentane solution of the residue produced a small number of crystals suitable for single crystal X-ray diffraction studies that established C–H activation of the terpyridine pincer in a trimetallic cobalt complex (5) (Figure 1). One metal hydride (H95) was located on the difference map that bridges Co2 and Co3, and it is possible that other metal hydrides are present despite the absence of significant electron density corresponding to their locations. Given the generation of 5 from stoichiometric experiments, the activation of the chelate suggested that competing terpyridine C–H activation interferes with productive catalysis. The extent of this side reaction is challenging to assess due to the low yield associated with 5. Attempts to isolate or observe a cobalt-boryl complex have been unsuccessful. These results contrast those obtained with (PNP)Co, where cobalt(I) boryl intermediates were observed spectroscopically following either stoichiometric addition of B₂Pin₂ to a cobalt(I) alkyl or in situ during catalytic borylation of 2,6-lutidine.^15b

Figure 1.

Representation of the solid-state structure of the [(^tBu₃Tpy)Co]₃ (5) at 30% probability ellipsoids with tert-butyl substituents and hydrogen atoms except for H95 omitted for clarity.

Synthesis and Catalytic Borylation Activity of a Sterically Protected Terpyridine Cobalt Alkyl.

To prevent competing terpyridine C–H activation, a Co(I)-alkyl precatalyst (6-CH₂SiMe₃) bearing the terpyridine ligand 4′-(4-N,N′-dimethylaminophenyl)-5,5″-dimethyl-2,2′:6′,2″-terpyridine (5,5″-Me₂^ArTpy) was prepared (Figure 2a). Using a procedure analogous to that used for ^ArTpy,²³ 5,5″-Me₂^ArTpy was synthesized on multigram scale in two steps from inexpensive (less than 1 USD/gram), commercially available starting materials (see the Supporting Information for full details of ligand and precatalyst synthesis). Metalation of 5,5″-Me₂^ArTpy with CoCl₂ and subsequent alkylation with LiCH₂SiMe₃ generated 6-CH₂SiMe₃ in 51% yield. The solid-state structures of both 4-CH₂SiMe₃ and 6-CH₂SiMe₃ were determined by single crystal X-ray diffraction and illustrate the steric protection of the flanking pyridyl substituents conferred by the methyl groups in 6-CH₂SiMe₃. (Figure 2b).

Figure 2.

(a) Motivation for the synthesis of the sterically protected terpyridine cobalt alkyl 6-CH₂SiMe₃ for use in borylation catalysis and (b) representations of the solid-state structures of 4-CH₂SiMe₃ and 6-CH₂SiMe₃ at 30% probability ellipsoids with hydrogen atoms omitted for clarity.

The performance of 6-CH₂SiMe₃ as a precatalyst for C(sp²)–H borylation was then evaluated using selected solvents and temperatures (Table 2). Catalytic borylation of 1a with B₂Pin₂ in the presence of 5 mol% 6-CH₂SiMe₃ at 23 °C produced high (99:1) meta selectivities in 30% and 40% conversion in Et₂O and THF, respectively (entries 1–2). Reaction in hexanes resulted in 44% conversion with diminished (88:12) selectivity (entry 3). Conversion increased to 75% when borylation was attempted in cyclohexane solvent, and 91:9 selectivity favoring the meta product was observed (entry 4). Borylation in CPME was found to be optimal, as the reaction reached 81% conversion with 98:2 (meta:ortho) regioselectivity (entry 5). Increasing the temperature to 50 °C (entry 6) reduced the conversion, suggesting that higher temperatures increase the rate of decomposition over productive turnover. An experiment analogous to entry 5 where HBPin was used in place of B₂Pin₂ produced 10% conversion with a 96:4 selectivity favoring meta-2a. This result indicates that the use of HBPin rather than the diboron reagent does not erode meta regioselectivity but rather diminishes overall catalytic borylation activity, consistent with previous results in cobalt-catalyzed fluoroarene borylation with (PNP)Co catalysts.¹⁷

Table 2.

Evaluation of conditions for the borylation of 1a using 5 mol% 6-CH₂SiMe₃ as precatalyst.

Entry	Solvent	Temp. (°C)	%conv.	m:o
1	Et2O	23	30%	99:1
2	THF	23	40%	99:1
3	hexanes	23	44%	88:12
4	CyH	23	75%	91:9
5	CPME	23	81%	98:2
6	CPME	50	56%	98:2

With improved conditions for meta-selective C(sp²)–H borylation of 1a in hand, the regioselectivity of the reaction with various fluorinated substrates with differing substitution patterns and electronic properties was examined (Scheme 4). A selection of 3-substituted fluoroarenes was borylated using one equivalent B₂Pin₂ at 23 °C in CPME and furnished meta-fluoroarylboronates 2a-d, with regioselectivities greater than or equal to 97:3. Reactions of electron-deficient arenes including those bearing sulfonamide (1a) and trifluoromethyl (1b) substituents proceeded with higher turnover numbers, and arenes bearing electron-donating substituents produced lower activity. For example, reaction of 3-fluorotoluene proceeded with less than 5% conversion to arylboronate products (see the Supporting Information for a complete list of substrates which did not undergo borylation efficiently). The high selectivity of the reaction extended to pyridines, as 2-fluoropicoline (2d) was selectively functionalized at the 4-position.

Scheme 4.

Borylation of selected fluoroarenes using 6-CH₂SiMe₃ as precatalyst.^a,b,c

^aBlue circles designate the position at which the minor product is borylated.^bYields and selectivities were determined by integration of the ¹⁹F NMR spectrum of the product mixture using 4-fluorotoluene standard.^cValues in parentheses correspond to isolated yields following column chromatography.

Substrates with two meta-disposed fluorine substituents underwent borylation selectively at the position meta to both fluorines to generate products 2e-h with excellent regioselectivities. Most notably, reaction of meta-difluorobenzene (1f) generated the 1,3,5-trisubstituted product 2f with exclusive selectivity. In contrast, borylation of this substrate using (^iPrPNP)Co(H)₂BPin as catalyst exclusively occurs at the site between the fluorine substituents.¹⁹ Arenes with chemically inequivalent C–H sites ortho, meta, and para to a fluorine substituent were also examined and provided the meta-borylated products 2i-k with high regioselectivities. With this class of substrates, the most significant minor product was the product arising from para to fluorine borylation. These observations established that borylation activity with 6-CH₂SiMe₃ trends in the order meta (to fluorine) > para > ortho. Additional substrate exploration demonstrated that borylations using 6-CH₂SiMe₃ suffered from many of the same functional group incompatibilities as those catalyzed by (^iPrPNP)Co(H)₂BPin (see the Supporting Information for complete details), as nitriles, chlorides, bromides, and iodides were not tolerated.¹⁵

Determination of Deuterium Kinetic Isotope Effects.

To elucidate the origins of the observed meta regioselectivity observed in fluoroarene borylation, the turnover limiting step of the catalytic reaction was explored. Monitoring the catalytic borylation of 1a with 5 mol% 6-CH₂SiMe₃ in 0.15 M CPME solution by ¹⁹F NMR spectroscopy over time revealed significant catalyst decomposition or product inhibition during the reaction, as the conversion to arylboronate product meta-2a gradually increased to approximately 50% during the first 60 minutes of the time course before remaining constant at this value (Figure S1). Attempts to remove the stoichiometric HBPin byproduct by addition of LiOMe as reported previously²³ did not improve the yield, consistent with catalyst deactivation as the source of the decline in the productivity of the reaction over time. The instability of the cobalt catalyst over a sufficient timeframe and number of half-lives (vide infra) prohibited reliable determination of the overall rate law.

Competition deuterium kinetic isotope effect experiments (KIEs) were conducted to gain insight into the (ir)reversibility of the C–H activation step in the catalytic cycle. Equimolar amounts of 1a and 1a-d₂ were borylated in the presence of 5 mol% 6-CH₂SiMe₃ and two equivalents of B₂Pin₂ at 23 °C in CPME (Scheme 5a). Analysis of the resulting reaction mixture by ¹⁹F NMR spectroscopy revealed that the consumption of 1a exceeded that of 1a-d₂ by approximately threefold, affording a k_H/k_D ratio of 2.9(2) (see the Supporting Information for full experimental details and KIE calculation). The observation of a normal, primary KIE is consistent with irreversible cleavage of the C(sp²)–H, though a competition experiment cannot be used to determine whether this is the slow step of the catalytic reaction.²⁶ To assess whether this irreversible C–H activation is turnover-limiting, an additional experiment was conducted whereby the KIE was determined in separate vessels (Scheme 5b). The borylations of 1a and 1a-d₂ were performed in separate vials with 5 mol% of 6-CH₂SiMe₃ and one equivalent of B₂Pin₂. Quenching both reactions after one hour and analysis by gas chromatography established a conversion ratio (conv_H/conv_D = 2.6(3), 23 °C) favoring reaction of the natural abundance compound, supporting that C–H bond cleavage occurs during the turnover-limiting step of the borylation of 1a. These results stand in contrast to the reported value of 1.1(1) (50 °C) obtained for k_H/k_D in parallel KIE experiments using 1a and 1a-d₂ with 5 mol% (^iPrPNP)Co(H)₂BPin as precatalyst, a reaction which produces a 95:5 ratio of products favoring borylation ortho to the fluorine substituent.¹⁷

Scheme 5.

(a) Competition deuterium kinetic isotope experiment and (b) Parallel conversion experiments for the borylation of 1a and 1a-d₂.

The results of KIE experiments support a turnover-limiting C(sp²)–H activation step which is necessarily selectivity-determining. This raises the question of whether the kinetic preference for the meta position is predominantly steric or electronic in origin. Catalytic borylation of most substrates examined with 6-CH₂SiMe₃ generated the meta products with approximately 99:1 regioselectivities, and it appears unlikely that such high selectivities arise from the slightly larger size of a fluorine substituent (van der Waals radius of 1.47 Å; A value of 0.15) compared to hydrogen (van der Waals radius of 1.20 Å) ,^11b especially considering the lack of steric bulk proximal to the cobalt in the terpyridine complexes. Additionally, the borylation of substrates with sterically accessible C–H sites para to fluorine (1i-1k, see Scheme 4) also produced high meta selectivity, with no more than 12% of the products borylated at the para position despite its nearly identical steric accessibility to that of the meta position. While a rigorous analysis of the steric and electronic contributions to meta selectivity in fluoroarene C–H activation is ongoing, the current data support an electronically-driven selectivity.

Generation of a Cobalt-Fluoroaryl Complex by C–H Activation.

To gain additional insight into fundamental steps of the catalytic reaction, preparation of a terpyridine cobalt(I)–aryl complex by C–H activation was pursued. Difluorinated substrate 1f was selected for these studies due to its anticipated reactivity toward C–H activation and ease of product assignment by ¹⁹F NMR spectroscopy; a cobalt(I)–aryl complex arising from C–H activation of the position meta to both fluorines would generate two ¹⁹F resonances in a 3:2 integration ratio, while the other regioisomeric product of C–H activation would give three ¹⁹F NMR signals in a 3:1:1 ratio, corresponding to the trifluoromethyl group and the fluorine substituents ortho and para to the cobalt center. Dissolution of 6-CH₂SiMe₃ in neat 1f produced no reaction by ¹⁹F NMR spectroscopy after 24 h at 23 or 80 °C (Scheme 6a). These initial experiments suggested that the terpyridine cobalt(I)-alkyl complex is not a suitable starting point for C–H activation of activated fluoroarenes.

Scheme 6.

Stoichiometric C–H activation experiments with 1f and 6-CH₂SiMe₃ under (a) 1 atm dinitrogen and (b) 0.1 atm dihydrogen, with proposed mechanism illustrated.

Repeating the procedure in the presence of 0.1 atm dihydrogen generated one new fluorine-containing cobalt product with ¹⁹F NMR signals located at −56.7, −83.5, and −115.6 ppm in a 3:1:1 ratio. The chemical shifts and integration are most consistent with formation of the fluoroaryl cobalt isomer^17,19 (6-Ar^F) with fluorine substituents ortho- and para- to cobalt. Analysis of the reaction mixture by ¹H NMR spectroscopy established approximately 33% conversion to product (Scheme 6b). These results support the competence of terpyridine cobalt complexes in stoichiometric C–H activation while also offering insight into the nature of the C–H activating intermediates in catalysis. The generation of 6-Ar^F from 6-CH₂SiMe₃ and excess 1f supports that C–H activation occurs from a putative terpyridine cobalt hydride. However, the regioselectivity of C–H activation from this procedure is opposite to that observed from the catalytic reaction, suggesting that the putative cobalt hydride is unlikely to be the intermediate responsible for C–H activation in catalysis. Based on precedent with [(^iPrPNP)Co] catalysts,^15b a terpyridine cobalt boryl is also plausible as the C–H activating species; however, attempts to observe and characterize either a cobalt boryl or cobalt hydride have been unsuccessful. That 6-Ar^F was generated selectively under H₂ may result from reversible C–H oxidative addition to generate a cis-Co(III) aryl dihydride, followed by reductive elimination of H₂. In this scenario, oxidative addition of the arene is expected to favor the ortho-fluoroaryl complex under thermodynamic control,^17,18 a possibility which will be discussed in the following section. Addition of 2 equivalents of HBPin to 6-Ar^F resulted in selective generation of the ortho-borylated product, demonstrating that no isomerization occurred prior to C–B bond formation similarly to the analogous experiment with 4-(p-C₆H₄F) (vide supra).

Analysis of Co–C and C–H Bond Thermodynamics.

Previous studies¹⁷ on [(^iPrPNP)Co]-catalysts for fluoroarene borylation established that the ortho selectivity observed in these reactions arises from fast and reversible C–H activation, resulting in ortho selectivity due to the well-established phenomenon whereby ortho fluorine substituents stabilize transition metal aryl intermediates.¹⁸ To discern whether the [(Tpy)Co] catalysts operate by a different mechanism or simply represent an exception to the typical thermodynamic effects of ortho fluorine substitution, a bond dissociation free energy (BDFE) correlation study was performed using BDFEs calculated with DFT using the ωB97XD functional²⁷ and the def2-TZVP basis set. Geometry optimizations were initially performed on potential Co(III) intermediates bearing hydride, boryl, and aryl ligands, but in some cases these optimized to Co(I) compound and free HBPin or arylboronate esters. To avoid variability in BDFE computations arising from inconsistencies in the identities of the optimized complexes, Co(I)–aryls analogous to 4-(p-C₆H₄F) and 6-Ar^F were computed for thermodynamic analyses.

According to convention, the relative increase in Co–C_aryl BDFE with increasing ortho fluorine substitution for Co(I)–aryl complexes supported by a truncated 5,5″-Me₂Tpy ligand was plotted on the y-axis, while the C–H BDFEs for the corresponding arenes were plotted on the x-axis. A plot with a correlation (R²=0.893) and a slope (termed R^MC/CH) of 2.47 was obtained, consistent with previous computational studies on the magnitude of the ortho fluorine effect for first-row transition metal complexes (Figure 3; see Table S7 for tabulated Co–C and C–H BDFEs.).^18b For comparison, our previous BDFE correlation studies of the [(^iPrPNP)Co] catalyst system produced a value of 2.87 for R^MC/CH.¹⁷ A slope of R^MC/CH is significantly greater than unity indicates that Co–C bonding is stabilized by ortho fluorine substitution to a greater extent than the corresponding C–H bonds in free arenes; in other words, ortho-fluoroaryl complexes of [TpyCo] are expected to be stabilized relative to other regioisomers consistent with other previously examined transition metal complexes. To compare the stabilities of [(Tpy)Co]-fluoroaryl isomers more directly, the free energies of the ortho, meta, and para-fluorinated cobalt complexes of the truncated 5,5″-Me₂Tpy ligand were calculated (Figure 4). Consistent with the thermodynamic ortho fluorine effect, the calculated energy of the ortho isomer was the lowest. The meta and para isomers were found to be higher in energy by 3.4 and 5.2 kcal mol⁻¹, respectively. These results indicate that the meta regioselectivity observed in fluoroarene borylation is not attributable to a thermodynamic preference for formation of meta-fluoroaryl intermediates. Combined with the observation of KIEs greater than unity for the borylation of 1a and 1a-d₂ and the selective reaction of 4-(p-C₆H₄F) with HBPin, the BDFE correlation study supports that meta selectivity in catalytic borylation arises from the kinetic preference of an irreversible C–H activation event. This contrasts with selectivity determination by [(PNP)Co] catalysts, which are proposed to undergo reversible C–H oxidative addition and favor borylation ortho to fluorine substituents under thermodynamic control.

Figure 3.

Correlation between the Co–C bond dissociation free energies (BDFEs) of (5,5″-Me₂Tpy)Co–Ar^F and the C–H BDFEs of the corresponding arenes calculated using DFT (ωB97XD/def2-TZVP). The zero points for the x and y axes correspond to the values for the C-H BDFE in benzene and the Co–C BDFE in (5,5″-Me₂Tpy)Co–Ph, respectively. The equation of the trendline is y = 2.47x + 0.03 (R² = 0.893).

Figure 4.

Comparison of the calculated (DFT, ωB97XD/def2-TZVP) free energies of Co(I)-fluoroaryl isomers bearing a 5,5”-dimethylated terpyridine ligand.

CONCLUDING REMARKS

In summary, cobalt precatalysts for the meta-selective borylation of fluorinated arenes have been discovered. A sterically protected terpyridine pincer was prepared from inexpensive, commercially available starting materials, and complexes derived from this ligand exhibited greater activity and regioselectivity than those from other terpyridine ligands examined. Deuterium kinetic isotope effect experiments and stoichiometric reactivity studies supported an irreversible and selectivity-determining C–H activation step in catalytic borylation, and analysis of computed Co–C and C–H bond thermodynamics supported that activation at the position meta to fluorine is disfavored thermodynamically compared to at the ortho position. In the absence of boron reagents, the cobalt(I) alkyl precatalyst was unreactive towards a fluoroarene at room temperature, and a putative cobalt(I) hydride activated the substrate with high selectivity for the site ortho to fluorine due to thermodynamic control of C–H activation in this case. These results, in combination with stoichiometric experiments ruling out isomerization following Co–C bond formation, suggest that a cobalt(I) boryl is the most likely C–H activating intermediate in catalytic borylation. All available data support that a slow, kinetically-controlled C–H activation step results in meta selectivity in catalysis. This proof-of-concept study for catalyst-controlled regioselectivity in C–H functionalization by electronic means illustrates the potential of first-row metal catalysts to access structures with underrepresented substitution patterns.