Kinetic and Thermodynamic Control of C(sp²)–H Activation Enable Site-Selective Borylation

Abstract

Catalysts that distinguish between electronically distinct C–H bonds without relying on steric effects or directing groups are challenging to design. Here cobalt precatalysts supported by N-alkyl-imidazole substituted pyridine dicarbene (ACNC) pincer ligands are described that enable undirected, meta-selective borylation of fluoroaromatics encompassing electron-rich arenes, pyridines, and tri- and difluoromethoxylated arenes, addressing one of the major limitations of first-row transition metal C–H functionalization catalysts. Mechanistic studies established a kinetic preference for C–H bond activation at the meta-position despite cobalt-aryl complexes resulting from ortho C–H activation being thermodynamically preferred. Switchable site selectivity in C–H borylation as a function of the boron reagent employed was thereby preliminarily demonstrated using a single precatalyst.

C–H functionalization has enhanced chemical synthesis by providing a direct method for the activation of otherwise inert bonds, providing a straightforward route to molecules with applications as pharmaceuticals, agrochemicals, and aromatics relevant to materials science. The development of site-selective reactions remains a grand challenge in metal-catalyzed C–H functionalization (1). A recent analysis of benzenoid-containing small molecule active pharmaceutical ingredients (APIs) approved through 2019 revealed that the most common substitution patterns (1,2-; 1,4-; and 1,2,4-) are accessible by electrophilic aromatic substitution (S_EAr), while the least common (1,3-; 1,3,5-; and 1,2,3,5-) are absent due to the lack of methods for their synthesis (2). Thus, remote C–H functionalization methods for the selective formation of one constitutional isomer are transformative. Site-selective C(sp²)–H borylation represents one of the most impactful C–H functionalization methods in organic synthesis as the C–B bond installed is readily and selectively transformed to a variety of other substituents using robust and well-established chemistry (3–7).

Progress in C(sp²)–H borylation has centered on state-of-the-art iridium precatalysts such as [Ir(COD)OMe]₂ (COD = 1,5-cyclooctadiene) in combination with bipyridine ligands that operate principally with sterically-driven site selectivity(8–11). To overcome the limitations of steric control, developments in regioselective iridium-catalyzed C–H borylation have relied on directing groups and intricate ligand engineering to exploit non-covalent interactions of specific functional groups with a recognition element on the ligand or an additive that orients the distal C–H bonds in proximity to the metal (12–16) (Figure 1A). While advances in ligand design have addressed the inherent sterically-driven site-selectivity, remote C(sp²)–H borylation of medicinally relevant molecules such as fluoroarenes remain a significant challenge due to the weak coordinating ability of fluoro substituents and their similar size to hydrogen (17). As a result, a general method for remote C–H borylation of fluoroarenes has not been reported. Given the prevalence of fluorine in pharmaceuticals and the paucity of methods for non-directed, site-selective remote borylation, methods to synthesize or directly functionalize fluoroarenes are desirable (Figure 1A). Inspired by reports of cobalt catalysts supported by electron-rich pincer ligands offering complementary regioselectivity compared to iridium, the development of a cobalt-catalyzed remote C–H borylation was pursued (18–21).

Figure 1. Background and catalyst development.

A) Transition-metal catalyzed regioselective arene borylation. B) This work: Electronically controlled remote C(sp²)–H borylation.

The overall performance of first-row transition metal C–H borylation catalysts is limited by low activity which in turn limits scope and overall synthetic utility (20, 21). To overcome the limitations of previously reported cobalt catalysts, rigid but sterically-accessible pincer ligands were targeted that are designed to limit destabilizing steric interactions during C–H activation and are supported by strong σ-donors resulting in a more electron-rich metal center (Figure 1C). For example, computational studies on C–H activation with ortho-to-fluorine selective pyridine bisphosphine cobalt complex (4-Me-(^iPrPNP)Co(H)₂BPin) (1-(H)₂BPin) established a late transition state for C–H oxidative addition arising from distortion of the flexible pincer and steric interactions between the alkyl substituents on the phosphine (22, 23). While the rigid terpyridine cobalt alkyl complex 5,5′′-Me₂-(ArTpyCo)(CH₂SiMe₃) (2-CH₂SiMe₃) enabled meta-to-fluorine selectivity by rate- and selectivity-determining C–H activation, the electron-deficient nature of the ligand inherently limits its performance as only a subset of electronic deficient fluorinated arenes provided modest yields and catalyst deactivation was observed at elevated temperatures (21). As a result, N-alkyl-imidazole substituted pyridine dicarbene cobalt complexes were pursued as N-heterocyclic carbenes (NHCs) are strong electron donors, offering the possibility of a sterically accessible catalyst that is also thermally robust (24). Here, we describe the synthesis of sterically-attenuated, NHC-based, pincer-supported cobalt alkyl and aryl complexes and their activity for C(sp²)–H activation resulting in the borylation of non-fluorinated arenes, heteroarenes, and, ultimately, meta-selective C(sp²)–H borylation of fluoroarenes and tri- and difluoromethoxylated arenes (Figure 1B). The increased activity enabled switchable site selectivity in catalytic C–H functionalization as cobalt-aryl isomerization was competitive with C–B bond formation.

Synthesis of well-defined precatalyst.

Our studies commenced with the synthesis of 3,5-Me₂-(^iPrACNC)Co(Br)₂ (3-Br₂). Cobalt alkyl complexes with this class of pincer bearing less sterically-encumbering N-alkyl rather than sterically-encumbering 2,6-diaryl substituents have been elusive owing to the propensity for competitive formation of bis(chelate) metal complexes (25, 26). To circumvent this issue, an optimized procedure was developed whereby 3,5-Me₂-(^iPr ACNC)(HBr)₂ (S1) was added to a cold (−95 °C) hexanes solution containing excess Co(HMDS)₂ (See supplementary materials for experimental details). This procedure, conducted on gram-scale, furnished a red solid in 93% yield identified as 3-Br₂ that was characterized by x-ray diffraction (Figure 2A). Addition of 2.1 equivalents of MeLi to a suspension of 3-Br₂ in diethyl ether followed by filtration and subsequent analysis by ¹H NMR spectroscopy in benzene-d₆ established formation of a diamagnetic, C_2v symmetric cobalt compound along with CH₃D formation. Single crystals obtained from this procedure established the identity of the product as the cobalt-aryl complex, 3-C₆D₅ arising from C–D activation by a putative cobalt methyl complex (Figure 2A). Repeating this procedure in THF enabled isolation of 3-Me and dissolution of the product in benzene-d₆ rapidly generated the cobalt-phenyl complex, establishing facile C–H activation reactivity. By contrast, the analogous cobalt methyl complex bearing aryl substituents (4-Me) on the NHCs was unreactive toward activation of benzene-d₆ even upon heating to 80 °C for 24 hours. To rationalize the differences in reactivity between 3-Me and 4-Me, the solid-state structures were analyzed and used to generate topographical steric maps using SambVca 2.1 (27) (Figure 2B). Comparison of the two plots shows significant shielding above and below the XZ plane defined by the CNC pincer plane by the wing-tips of 4-Me. Notably, 4-Me has a significantly higher buried volume (%V_Bur) compared to 3-Me. These results illustrate that multiple catalyst design principles must be satisfied. For example, despite being supported by strong donors, the steric disposition of the wing-tip substituents on 4-Me diminishes reactivity toward C–H activation.

Figure 2. Precatalyst synthesis and characterization.

A) Activation of benzene by 3-Me. B) Steric maps of [(^iPrCNC)Co] and [3,5-Me₂-(^iPrACNC)Co] from the solid-state structures of the corresponding cobalt-methyl complexes generated in SambVca2.1. See supplementary materials for experimental details.

Reaction development.

Given the rapid activation of benzene, the C(sp²)–H borylation activity of 3-Me was evaluated. Standard conditions for catalytic experiments employed 1 equivalent of arene and 1 equivalent of B₂Pin₂ with 5 mol% of 3-Me at room temperature in THF for 24 hours. A series of 3-substituted (6a–6e) fluoroarenes was evaluated and the desired meta-fluoroarylboronates were obtained in high yield (>75%) and regioselectivity (>85:15 m:o). The substrates evaluated contain directing groups commonly employed in metal-catalyzed directed functionalization and led to lower yields when previously reported 2-CH₂SiMe₃ was employed (e.g., 6b-6d) (28). In previous studies, a substrate-controlled selectivity was observed with fluorinated-aryl boronate esters (29). For example, when 6f and 6g were functionalized with precatalyst 1-(H)₂BPin, the selectivity was inverted from 94:4 meta-to-ortho to 20:80, respectively. Although mechanistically informative, changes in selectivity with minor modifications to the substrate limit predictability and ultimately hamper synthetic utility. Improved yields and catalyst-controlled high meta-selectivity were observed with 3-Me, highlighting the superior performance of the NHC-based pincer cobalt precatalyst. A series of 2,6-difluoroaryls containing functional groups, including boronates, ethers, and esters (6h–6m), was examined, and in all cases, high yields (>80%) and selectivities (>87:13 m:o) were obtained. The para-to-boron and meta-to-fluorine selectivity was observed independent of the substrate, highlighting the insensitivity of 3-Me to substrate modifications and providing a more general and synthetically useful C–H borylation route to mixed diborylarenes. To probe the borylation selectivity of arenes with sterically-accessible ortho- (to fluorine), meta-, and para-C–H bonds, a series of 2-substituted fluoroarenes was examined. The meta-arylboronate esters were the major products for all substrates studied (6o–6s).

Pyridines are an important class of substrate as the second most encountered heterocycle in FDA-approved pharmaceuticals, and they present electronically distinct C–H bonds (e.g., based on BDFEs and pKa) (30). Borylation of 6-fluoro-2-picoline (6t) with 3-Me furnished the (hetero)arylboronate product in 91% yield with 99:1 regioselectivity favoring the 4-position. The previously reported cobalt precatalyst 2-CH₂SiMe₃ suffered from low yields while iridium/bipyridine catalysts provided poor selectivity (31–33). Borylation of a series of substituted pyridines (6u–6aa) with 3-Me furnished the desired 4-functionalized products in good yields and selectivity greater than 87:13. The borylation of pyridines typically requires a directing group or inclusion of sterically demanding aluminum-based Lewis acids to achieve synthetically useful selectivities (34).

Next, the overall performance of the designed pyridine dicarbene cobalt precatalyst was evaluated and compared to previously reported precatalysts. Standard conditions for catalytic experiments employed one equivalent of both arene and B₂Pin₂ in the presence of 5 mol% of cobalt precatalyst in THF at ambient or elevated temperatures. Among the precatalysts evaluated, 3-Me was the only one effective for the catalytic borylation of 1,3-dimethoxybenzene (6ab), reaching 50% conversion at 80 °C. With relatively electron-rich substrates (6ab-6ae), solvent or excess quantities are often required for borylation activity with prior cobalt catalyst. Trifluoromethoxybenzene (6af) and a related derivative (6ag) were borylated at 23 °C using 3-Me to afford the meta-functionalized product with regioselectivity ratios of 91:9 and >99:1 (m:o), respectively. Both 1-(H)₂Bpin and 2-CH₂SiMe₃ failed to catalyze any detectable formation of the desired products even upon extended heating at 80 °C. High meta-selectivity (84:16) was also observed with difluoromethoxy groups (6ah). Trifluoromethoxy (OCF₃) and difluoromethoxy (OCF₂H), typically introduced by radical addition, are highly desirable structural motifs in drug development processes due to their ability to fine-tune the physicochemical properties of drug candidates (35). These results highlight the importance of developing more active first-row metal catalysts as electronically-controlled selectivity can be compatible with functional groups not amenable to directing group strategies.

The cobalt precatalyst with a phosphine-based pincer, 1-(H)₂Bpin promotes para-to-boronate ester selectivity; however, the scope is limited to electron-deficient fluorinated arylboronate esters (29). To generalize the electronically-controlled selectivity, phenylboronic acid pinacol ester (6ai) and phenylboronic acid neopentylglycol ester (6aj) were examined and failed to give any of the desired product with any previously reported catalyst. Borylation with 3-Me provided the desired products in 58% and 56% yield with 77:23 and 82:18 respective regioselectivity favoring functionalization at the para-to-boronate site. Among the precatalysts evaluated, 3-Me was the only one effective for the catalytic borylation of methyl benzoate (6ak) providing the desired product in 64% yield with a regioselectivity of 77:23 favoring functionalization at the para-to-ester site. To evaluate the electronic influence of multiple substituents on an arene, substrates 6al, 6am, and 6an were evaluated. One regioisomer was obtained when 6al was evaluated highlighting the reinforced para-to-ester and meta-to-fluorine selectivity. Borylation of 6am led to one regioisomer while 6an led to preferential functionalization at the para-to-boronate site overriding the meta-to-fluorine selectivity. 1,4 disubstituted fluoroarenes were also viable substrates. For example, borylation of 4-fluorotoluene and 4-fluorobenzotrifluoride led to preferential functionalization ortho-to-fluorine over sterically inaccessible C2 functionalization (see figure S133, S134). The presence of Csp²–X bonds in aromatics presents a chemoselectivity challenge as previously reported cobalt catalyzed Csp²–H borylation methods are incompatible with aryl halide substituents. For example, chlorobenzene (6ao) was an incompatible substrate with previously reported cobalt catalyst 1-(H)₂Bpin and 2-CH₂SiMe but was borylated in 61% yield and 86:14 regioselectivity favoring meta-to-chlorine product with 3-Me highlighting the unique kinetic preference for C–H over C–Cl borylation. To demonstrate the utility of the cobalt-catalyzed method for gram-scale synthesis, etoxazole (5ap), a chitin synthesis inhibitor, underwent borylation in 97% yield to afford 6ap. In situ activation of 3-Br₂ with KBHEt₃ using standard Schleck technique could also be applied, resulting in the borylation of etoxazole (5ap) in 52% yield (Figure S9). Recent advancements in cross-coupling with earth-abundant metals have been developed between alkyl and aryl halides with neopentylglycol(hetero)aryl boronic esters (36). To benefit from the unique site-selectivity, a method was developed to synthesize neopentylglycolaryl boronic esters with 3-Me and applied to the derivatization of etoxazole. Under the optimized conditions employing bis(neopentyl glycolato)diboron, the desired product 7 was obtained in 91% isolated yield and characterized by x-ray diffraction. Performing the reaction with 2-CH₂SiMe did not lead to the desired product, highlighting the limitations of previously reported cobalt catalysts for remote C–H borylation.

Mechanistic studies.

To gain insight into the origin of selectivity for C–H borylation, the synthesis of cobalt(I)-fluoroaryl complexes relevant to catalysis were pursued. Treatment of a benzene-d₆ solution of 3-Me with excess fluoroarene (8) at room temperature resulted in the formation of the corresponding cobalt(I)-aryl complex 3-Ar^F arising from C–H activation at the ortho-to-fluorine site (Figure 4A). No other signals corresponding to isomers of 3-Ar^F were detected by ¹⁹F NMR spectroscopy. Notably, the preferred C–H bond activated during catalysis is the meta-to-fluorine site leading to the meta-to-fluoroarylboronate product. To probe for intramolecular or HBPin induced isomerization following the generation of the cobalt-aryl complex, the reaction of 3-Ar^F with HBPin was examined. Treatment of 3-Ar^F with excess HBPin resulted in the exclusive formation of the ortho-to-fluorine borylated product (Figure 4A): no other isomers of the arylboronate were detected by ¹⁹F NMR spectroscopy ruling out isomerization prior to C–B bond formation. To rationalize the discrepancy between the stoichiometric experiments leading to the ortho-Co(I) aryl complex and the major product observed in catalysis arising from C–B bond formation at the meta-site, the stoichiometric reaction of 3-Ph with a series of fluoroarenes was monitored by NMR spectroscopy. Treatment of 3-Me with excess 1,3-difluorobenzene at room temperature furnished 3-(3,5-C₆H₃F₂) (wherein the fluorine substituents are both meta to the cobalt–aryl bond ) as the major product after 30 minutes. After 24 hours, 3-(3,5-C₆H₃F₂) was converted to 3-(2,6-C₆H₃F₂) wherein both fluorine substituents occupy ortho-positions. With fluorobenzene, similarly distinct kinetic and thermodynamic preferences was observed as the meta-aryl isomer formed at early reaction times, then underwent isomerization to the ortho-variant as the reaction progressed (Figure S6). No detectable quantities of the para-isomer were observed by ¹⁹F NMR spectroscopy throughout the course of the experiment (Figure 4B). To probe the mechanism of cobalt–aryl isomerization, a kinetic ratio of 3-(3,5-C₆H₃F₂) and 3-(2,6-C₆H₃F₂) in 0.4 mL of THF-d₈ was transferred to a J-Young NMR tube, and the mixture was directly analyzed by ¹⁹F NMR spectroscopy (Fig 4C). After 24 h, the ratio of 3-(3,5-C₆H₃F₂) to 3-(2,6-C₆H₃F₂) did not change. However, upon addition of excess 1,3 difluorobenzene, complete conversion to 3-(2,6-C₆H₃F₂) was observed. These results demonstrate excess fluoroarene is required for cobalt-aryl isomerization to occur. These observations support kinetically preferred C–H activation at the position meta-to-fluorine and thermodynamically favored formation of the ortho-isomer (37, 38). The free energies of the fluoroaryl isomers were calculated with DFT using the ωB97XD functional and the def2-TZVP basis set and corroborate the experimental observations that 3-(2,6-C₆H₃F₂) is lowest in energy consistent as a result of a thermodynamic ortho-to-fluorine effect (Fig 4B). Experimental support for these trends with first-row metals has not previously been reported due to the lack of reactivity of the metal-aryl complexes which are typically synthesized by addition of an organometallic reagent to the metal dihalide precursors.

Figure 4. Mechanistic considerations and kinetic vs. thermodynamic selectivity.

A) Reaction of 3-Me with 8 and probing isomerization of 3-Ar^F upon treatment with HBPin. B) Thermodynamic and kinetic preferences of C–H activation. C) Switchable C–H borylation. See supplementary materials for experimental details.* 4 equiv. of 5a.^†3 equiv. of 5c.

The observed aryl isomerization suggested that switchable site selectivity in catalytic C–H borylation should be possible. We postulated that if cobalt-aryl isomerization were faster than C–B bond formation, the selectivity would be inverted to favor the ortho-fluoroarylboronate product (Figure 4C). Stirring a THF solution of 8 with HBPin with 5 mol% of 3-Me at 50 °C produced boronate ester 9a in 94% yield and a 24:76 ratio favoring the ortho-isomer (ortho-8a). The switch in selectivity is attributed to a slower rate of C–B bond formation when using HBPin.

In order to assess the rate of C–B bond formation as a function of boron reagent, the catalytic borylation of 5a with B₂Pin₂ and HBPin was monitored by ¹⁹F NMR spectroscopy as a function of time (Figure S24–S25). With HBPin, 19% conversion was observed after 19 h. In contrast, with B₂Pin₂, 80% conversion was observed in 10 minutes. These results support a faster rate of C–B bond formation with B₂Pin₂. Repeating the experiment with DBPin resulted in no significant deuterium incorporation in 9a (Figure S27). If DBPin were responsible for cobalt-aryl isomerization, hydrogen isotope exchange of the free arene should be observed and the ortho-functionalized aryl boronate would contain significant deuterium incorporation at the meta-to-fluorine site. An inverse relationship was also observed between the selectivity for the ortho-fluoroaryl boronate ester and the equivalents of HBPin, which does not support an HBPin induced cobalt–aryl isomerization mechanism (Table S3). A series of additional mechanistic experiments (Figure S26) suggest the isomerization to thermodynamically preferred cobalt-aryl takes place by intermolecular C–H activation of a second arene by meta-to-fluoroaryl cobalt intermediate. The preferred site of borylation was also switched when 5a and 5c were employed as substrate (Figure 4D). The different selectivities were obtained from a single catalyst by simply changing between two commercially available boron reagents. These observations contrast traditional approaches to switchable selectivity where different ligands or transition metal complexes are required (39). Switchable C–H borylation was unique to 3-Me as the selectivity did not alter significantly when 1-(H)₂Bpin and 2-CH₂SiMe₃ were employed as precatalyst (Figure S23). Based on insights from the stoichiometric experiments, the selectivity to favor formation of the meta-fluoroarylboronate was improved by increasing the equivalents of B₂Pin₂. For example, stirring a THF solution of 8 and 5c with two equivalents of B₂Pin₂ with 5 mol% of 3-Me for 24 hours improved the regioselectivity from 3:1 to 4.5:1 and 3:1 to 5.6:1 in favor of meta-9a and meta-6c, respectively. Further studies are on-going to expand the switchable C–H borylation scope and lower the equivalents of arene when using HBPin as the boron reagent (See Figure S22).

In conclusion, introduction of a sterically-attenuated, electron-donating pincer has enabled the development of a cobalt precatalyst that exhibits high activity for C–H activation and C(sp²)–H borylation. A combination of stoichiometric experiments and site-selectivity from catalytic examples demonstrate a kinetic preference for functionalization of sites meta to fluorine that in the presence of arene substrate isomerize over time through C–H activation to ortho-aryl complexes. This feature enables switchable selectivity in catalytic C(sp²)–H borylation where tuning the relative rates of C–H activation and C–B bond formation modulates the preference for ortho- versus meta-functionalized products. These findings provide key catalyst design principles for site selective C–H functionalization exploiting kinetic and thermodynamic preferences of M–C versus C–H bond formation.

Figure 3.

C(sp²)–H borylation of (hetero)arenes. Yields and selectivities were determined by ¹⁹F NMR spectroscopy using a 4-fluorotoluene internal standard. *Isolated yield. ^†3 equiv. of B₂Pin_2. ^‡0.3 mol% of 3-Me. ^§50 °C. ^¶ 80 °C.^# 3 equiv. of arene

Data and materials availability:

All other data are available in the main text or the supplementary materials. X-ray data are available in the Cambridge Crystallographic Data Center under CCDC numbers 2246986, 2246987, 2246988, and 2246989.