Mechanistic Investigations of Cobalt-Catalyzed, Aminoquinoline-Directed C(sp²)–H Bond Functionalization

Abstract

Monoanionic, bidentate-auxiliary-directed, cobalt-catalyzed C–H bond functionalization has become a very useful tool in organic synthesis. A comprehensive investigation into isolated organometallic intermediates and their reactivity within the catalytic cycle is lacking. We report here mechanistic studies of cobalt-catalyzed, aminoquinoline-directed C(sp²)–H bond functionalization. A number of organometallic Co(III) intermediates have been isolated and structurally characterized, including, for the first time in the aminoquinoline system, complexes arising from migratory insertion into cobalt–carbon bonds. The catalytic and stoichiometric reactions of cobalt(III) aryls with alkenes, alkynes, carbon monoxide, cyclic secondary amines, and aminoquinoline benzamides have been explored. The oxidation state of cobalt intermediates in the product-forming step depends on the nature of the coupling component. Specifically, annulation with alkynes and carbonylation with CO likely proceed via a Co(I)/Co(III) catalytic cycle. Carbon–hydrogen bond functionalization with alkenes and amines, as well as benzamide homocoupling, likely proceed via a (formally) Co(IV) species and involve oxidatively induced reductive elimination.

Graphical Abstract

INTRODUCTION

Since its introduction in 2005, the aminoquinoline directing group has emerged as one of the most utilized removable auxiliaries in carbon–hydrogen bond functionalization.^1,2 The initial studies were directed toward its use in palladium catalysis.^2,3 More recently, C–H functionalization by cheaper and more abundant first-row transition metals such as copper, iron, cobalt, nickel, manganese, and chromium has been explored.^2,4,5,6 The success of the aminoquinoline directing group can be explained by its propensity to facilitate the C–H activation step and its ability to stabilize high-valent metal intermediates via a tridentate, dianionic coordination mode.^2,3,4

Cobalt is perhaps the most widely used first-row transition metal employed for aminoquinoline-directed C–H functionalization. In 2014, we reported the first protocol for cobalt-catalyzed, aminoquinoline-directed coupling of arene C–H bonds with alkynes (Scheme 1a).^5a Subsequently, we disclosed sp² C–H bond coupling with alkenes (Scheme 1b),^5b carbon monoxide (Scheme 1c),^5c and dimerization of aminoquinoline benzamides (Scheme 1d).^5d We also reported the use of arylsulfonic^5e and arylphosphinic acid^5f aminoquinoline amides in sp² C–H functionalization. Since these initial reports, aminoquinoline-directed, cobalt-catalyzed C–H functionalization has been extensively explored by a number of research groups.⁶ Furthermore, other bidentate, monoanionic directing groups such as picolinamide, pyridine N-oxide, 2-hydrazinyl-pyridine, and 2-(hydroxymethyl)pyridine have been employed to accomplish arene C–H functionalization.^5a,7 In most of these cases, the combination of simple cobalt(II) salts such as Co(OAc)₂ or Co(acac)₂ with an Mn salt and a base affords the best results.

Scheme 1.

First Reports of Cobalt-Catalyzed, Aminoquinoline-Directed C(sp²)–H Bond Functionalization

Aminoquinoline-directed, metal-catalyzed C–H functionalization has been the subject of extensive mechanistic studies. Notably, copper-catalyzed reactions have been explored by Stahl and Sanford, while nickel catalysis was investigated by the Sanford, Hoover, and Schafer groups.⁸ Cobalt-catalyzed C–H functionalization has also attracted significant attention, with many research groups conducting mechanistic investigations and characterizing potential organocobalt intermediates.^9,10 Despite relatively extensive studies, little is known about the reactivities of putative organocobalt intermediates. Furthermore, different coupling partners may result in distinct catalytic cycles involving Co(III) or Co(IV) intermediates. Oftentimes, conflicting mechanisms have been proposed. It appears that intermediates arising from migratory insertion into cobalt–carbon bonds have not been characterized in aminoquinoline or related auxiliary-directed C–H functionalizations. Isolation Functionalization of such intermediates allows the direct study of the product-forming step in the catalytic cycle.

We report here on mechanistic studies of cobalt-catalyzed, aminoquinoline-directed C(sp²)–H bond functionalization. Several distinct organometallic Co(III) intermediates have been isolated and structurally characterized, including, for the first time in an aminoquinoline system, complexes arising from migratory insertion into cobalt–carbon bonds. Catalytic as well as stoichiometric reactions with alkenes, alkynes, carbon monoxide, cyclic secondary amines, and dimerization of aminoquinoline benzamides have been explored. The oxidation state of cobalt intermediates in the product-forming step depends on the nature of the coupling component.

RESULTS AND DISCUSSION

General Considerations.

It is believed that, in most cases, cobalt-catalyzed, monoanionic, bidentate auxiliary-directed C–H functionalization reactions proceed via a Co(I)/Co(III) catalytic cycle.^{9c–l,n,p–r,t,u,w} The individual steps involve substrate binding to the catalyst (1 to 2, Scheme 2), C–H bond activation (2 to 3), migratory insertion (4 to 5), and reductive elimination to liberate Co(I) species (5 to 6). The low-valent Co(I) is then reoxidized back to the state of Co(III). The Co(I)/(III) catalytic cycle has been proposed based on evidence gathered from kinetic studies, calculations, and isolated organometallic Co(III) complexes.

Scheme 2.

Mechanistic Considerations

In the published mechanistic studies, nearly all Co(III) intermediates that have been isolated and characterized are formed in the C–H activation step.^{9g,k,l,o,p,r,t,v,w,10e} No examples of migratory insertion products have been isolated and characterized, and consequently, their reactivity toward reductive elimination has not been explored. In a few cases, in situ HRMS analysis suggests the presence of migratory insertion products.^9v,x,10f In addition, recently, several research groups have proposed that a Co(II)/Co(IV) catalytic cycle is operative, involving an oxidatively induced reductive elimination pathway for carbon–nitrogen and carbon–oxygen bond formation.^{9o,v,10a,b,k,l} Due to the high reactivity of Co(IV) species, authors collected evidence from experimental, computational, and cyclic voltammetry studies. Reports on the involvement of Co(IV) intermediates are scarce, and the intermediacy of high-valent Co species in aminoquinoline-directed carbon–carbon bond-forming reactions has not been conclusively established.

Analysis of Potential Reaction Pathways.

Possible mechanistic pathways are depicted in Scheme 2. Reaction of aminoquinoline benzamide 1 with a Co(II) salt generates Co(II) amide 2. Next, the oxidation and C–H bond activation step produces a Co(III) aryl intermediate 3. In the coupling with alkyne, alkene, or CO, coordination of the reagent, followed by migratory insertion would lead to 4 and then 5 via pathway I. Reductive elimination from a Co(III) (or Co(IV)) species 5 liberates product 6.

Pathway II describes the aminoquinoline benzamide dimerization. Complex 7 is formed after ligand exchange places two aminoquinoline benzamides on a Co(III) center. Next, a second C–H activation and reductive elimination yield dimer 8. Alternatively, the pathway involving the second C–H activation and oxidation of the intermediate to a Co(IV) species, followed by reductive elimination could deliver the product, as suggested by Ackermann for a related pyridine oxide system, and by Budnikova and Sanford for aminoquinoline derivatives.^9o,z,10b

Pathway III could be operative for amination and other C–X bond-forming reactions. Accordingly, Co(III) aryl 3 would react with an amine to afford amine complex 9. After ligand exchange and deprotonation, complex 10 is formed. After reductive elimination, which could proceed either from Co(III) complex 10 or, more likely, via an oxidatively induced reductive elimination pathway involving a high-valent Co species, the amination product 11 is liberated. The C–H functionalization reactions listed in Scheme 2 will be analyzed with respect to each of the isolated Co(III) complexes, as discussed below.

Synthesis and Reactivity of the Co(III) Amide 2a.

We started our studies by preparing cobalt(III) amide 2a, which is a potential catalytic intermediate before the C–H activation step. Related cobalt(III) amide complexes with acetylacetonate ligands were obtained by Maiti and Deb.^9c,t,y The Co(III) amide 2a was synthesized in 18% yield by reacting aminoquinoline p-nitrobenzamide 1a with a stoichiometric amount of cobalt(II) nitrate in the presence of sodium acetate and dipivaloylmethane ligand in trifluoroethanol under air (Scheme 3). Complex 2a was purified by column chromatography and exists as a moderately stable brownish-green solid that is soluble in most common organic solvents, including hexanes.

Scheme 3.

Synthesis of 2a

Crystals of 2a suitable for X-ray analysis were grown from pentane. The ORTEP diagram of complex 2a is shown in Figure 1. An aminoquinoline amide and two dipivaloylmethane ligands are coordinated to cobalt in a slightly distorted octahedral fashion.

Figure 1.

ORTEP view of the molecular structure of Co(III) complex 2a. Thermal ellipsoids are drawn to encompass 50% probability. Hydrogens are omitted for clarity. Selected bond distances (Å) and angles (deg): Co(1)–N(1) 1.9125(9), Co(1)–N(11) 1.9678(9), Co(1)–O(28) 1.8780(7), N(1)–Co(1)–N(11) 82.79(4), N(11)–Co(1)–O(28) 96.42(3).

Attempts to convert Co(III) amide 2a to C–H activated species failed despite multiple attempts under different conditions (Scheme 4a).¹¹ The reactivity of complex 2a with alkynes was studied under stoichiometric and catalytic conditions, which are similar to the ones reported earlier.^5a Reaction of 2a with 2-butyne in the presence of Mn(OAc)₂ afforded coupling product 12a in 98% yield (Scheme 4b). If the C–H bond functionalization reaction was performed in the absence of the Mn(OAc)₂ additive,¹² product 12a was isolated in 25% yield (Scheme 4c). The reaction under air-free conditions with a catalytic or stoichiometric Mn(OAc)₂ additive gave 12a in 54% and 60% yields, respectively (Scheme 4d,e). A possible explanation for the Mn(OAc)₂ effect is the acceleration of ligand exchange in 2a, which results in the substitution of one dipivaloyl methane ligand by acetate.¹² The C–H activation via a CMD or BIES mechanism by metal-bound acetate would give cobalt aryl, which reacts further to afford 12a.

Scheme 4.

Reactivity of Co(III) Amide Complex 2a

Reaction of benzamide 1a with diphenylacetylene using a catalytic amount of Co(III) complex 2a gave 12b in 40% NMR yield (Scheme 4f). These results indicate that the identity of the ligands has a major impact on the reactivity of Co(III) species. In our original methodology for C–H functionalization with alkynes, the Co(OAc)₂ catalyst was used.^5a It is likely that a Co(III) amide–acetate complex undergoes faster ligand exchange compared to 2a, which may explain higher yields for Co(OAc)₂ catalysis.

Kinetics of Benzamide C–H Bond Functionalization with Alkenes.

To gain insight into the kinetics of the C–H bond activation and functionalization, we measured intramolecular, intermolecular, and independent initial rate KIE for C–H bond functionalization with alkenes. The results are summarized in Scheme 5.¹¹ First, we determined that the C–H activation step is irreversible and did not observe the incorporation of deuterium in substrate 1 and in product 12c when 1 was reacted with styrene under previously reported conditions using trifluoroethanol-d₃ solvent (Scheme 5a).^5b Intramolecular competition experiments with ortho-deuterated benzamide 1-d₁ displayed a significant isotope effect of 3.0 regardless of the conversion (Scheme 5b). Next, we determined the intermolecular KIE by reacting equimolar amounts of benzamide 1 and deuterated analogue 1-d₅ with styrene (Scheme 5c). The reactions were run in duplicate and were stopped after 10% and 30% conversions. Analysis by ¹H NMR showed identical reactivity of 1-d₅ and 1, giving KIE = 1. Similarly, 1 and 1-d₅ were reacted with styrene in parallel reactions, and the initial reaction rates were compared after reaching 22% and 25% conversions, respectively (Scheme 5d). No KIE was observed. Intermolecular, intramolecular, and independent initial rate KIE measurements indicate that the C–H bond activation step is not the RDS, and that it occurs after the RDS of the reaction.

Scheme 5.

Kinetic Experiments for the Reaction with Alkenes

Next, we examined the dependence of the reaction order on alkene concentration (Scheme 5e). Benzamide 1 was reacted with 1, 2, 5, and 10 equiv of styrene in parallel reactions under the same conditions. Aliquots were taken and analyzed by ¹H NMR at certain time intervals to obtain the initial reaction rates. We did not observe any initial rate dependency on styrene concentration, suggesting that alkene is not involved in the RDS of the reaction. Similarly, parallel experiments were run with 1, 2, 5, and 10 equiv of benzamide 1, and reaction rates were measured. The reaction rate is not dependent on the concentration of benzamide (5e). These experiments suggest that C–H activation and migratory insertion steps cannot be rate-determining, that the RDS likely occurs before the C–H activation step, and is plausibly ligand exchange introducing a carboxylate base.¹³

Synthesis and Characterization of Cobalt(III) Aryls 3.

During the catalytic reactions, TLC analysis showed the formation of several cobalt-containing intermediates. Cobalt-(III) complexes were isolated by flash column chromatography, and their structures were determined by NMR and X-ray crystallographic analysis. Two distinct Co(III) aryl types were formed (Scheme 6). Reaction of benzamides 1a and 1b with a stoichiometric amount of Co(dpm)₂ or Co(acac)₂ in the presence of NaOPiv and Mn(OAc)₃·2H₂O gave Co(III) aryl complexes 3a–3d. In these complexes, one aminoquinoline amide and one diketone moiety are coordinated to the cobalt center. The aminoquinoline amide acts as a tridentate, dianionic CNN pincer ligand, and the sixth coordination site is occupied by a weakly bound solvent molecule. Note that the weakly bound ligand is situated at the cis-position relative to the cobalt–carbon bond. Crystals of 3c suitable for X-ray analysis were grown from methanol. The ORTEP diagram of complex 3c is shown in Figure 2. The second type of cobalt(III) complex 7a was synthesized from benzamide 1c and a stoichiometric amount of Co(acac)₂ in the presence of a pivalate base. In 7a, two aminoquinoline benzamides are coordinated to the cobalt center. Several analogous complexes have been reported and characterized by other research groups.^9k,l,o,p,t,v In 7a, one of the amides is coordinated in a κ-2 fashion, while the other is bound to cobalt in a κ-3 fashion. Only one of the amides contains a Co–C bond. Dark red crystals suitable for X-ray analysis were obtained by crystallization from a CD₃OD solvent. The ORTEP diagram of complex 7a is shown in Figure 3. Complexes of type 7a are the most commonly proposed intermediates in aminoquinoline-directed, cobalt-catalyzed C–H functionalization reactions.

Scheme 6.

Synthesis of Cobalt Aryls 3a–3d and 7a

Figure 2.

ORTEP view of the molecular structure of Co(III) complex 3c. Thermal ellipsoids are drawn to encompass 50% probability. Hydrogens are omitted for clarity. Selected bond distances (Å) and angles (deg): Co(1)–N(2) 2.022(2), Co(1)–N(3) 1.900(2), Co(1)–O(6) 1.957(2), Co(1)–C(10) 1.907(3) N(2)–Co(1)–N(3) 83.48(8), N(3)–Co(1)–O(4) 88.09(7).

Figure 3.

ORTEP view of the molecular structure of Co(III) complex 7a. Thermal ellipsoids are drawn to encompass 50% probability. Hydrogens are omitted for clarity. Selected bond distances (Å) and angles (deg): Co(1)–N(30) 1.898(2), Co(1)–N(10) 1.971(2), Co(1)–O(41) 1.955(2), N(30)–Co(1)–N(19) 93.28(9), N(30)–Co(1)–O(41) 89.20(8), N(10)–Co(1)–N(39) 96.75 (8).

Cobalt(III) aryl complexes 3b and 3d react with an excess of aminoquinoline benzamide 1b in trifluoroethanol at 80 °C to form Co(III) aryl complex 7b (Scheme 7). In the case of dipivaloylmethane-ligated 3b, 7b was formed in 26% yield in 20 min. Acetylacetone-ligated 3d under identical reaction conditions gave Co(III) complex 7b in 50% NMR yield, indicating that ligand exchange occurs faster for the acetylacetonate ligand. No product formation was observed under similar reaction conditions at room temperature even after 16 h. Addition of Mn(III) had no effect on the rate of ligand exchange.

Scheme 7.

Formation of Cobalt(III) Aryl Complex 7b

Reactivity Studies of Co(III) Aryl Complexes 3.

Reactivity of Co(III) complex 3a (L = MeOH) with alkynes, CO, and alkenes was investigated under stoichiometric and catalytic conditions (Scheme 8). The reaction of Co(III) complex 3a with diphenylacetylene at 30 °C yielded annulation product 12b nearly quantitatively in less than 5 min (Scheme 8a). Importantly, the presence of an oxidant was not required. This was confirmed by the reaction under air-free conditions, which gave nearly identical results.¹¹ Complex 3a reacted smoothly with CO at room temperature to give carbonylation product 13 in a near-quantitative yield by ¹H NMR (Scheme 8b). These experiments suggest that complex 3a is a competent intermediate for cobalt-catalyzed C–H bond functionalization with alkynes and CO, and reductive elimination from Co(III) is facile without the requirement of further oxidation at the metal center (Scheme 2). Complex 3a reacted with styrene at room temperature to yield annulation product 12c in only 39% yield (Scheme 8c). No product was formed if oxygen was excluded from the reaction mixture (Scheme 8c)

Scheme 8.

Reactivity of Co(III) Aryl Complex 3a

The addition of Mn(III) acetate increased the product yield to a quantitative yield in less than 5 min at room temperature (Scheme 8d). To understand the role of Mn(III) acetate, additional experiments were carried out. Manganese acetate can either facilitate ligand exchange or act as an oxidant to access Co(IV) species, which can undergo reductive elimination. Reaction of 3a with styrene under air-free conditions using the Mn(II) acetate additive gave only traces of 12c (Scheme 8e). Addition of the Cp₂FePF₆ oxidant gave product 12c in 81% yield in just 3 min (Scheme 8f). This result indicates that high-valent, formal Co(IV) species are likely involved in C–H bond functionalization with alkenes. Under strictly air-free conditions, the Mn(III) acetate additive gave the product in 70% yield, showing that it can act as a competent oxidant in the absence of oxygen (Scheme 8g).

Similarly, the reactivity of Co(III) aryl complex 7a was studied (Scheme 9). The reaction of complex 7a with diphenylacetylene in trifluoroethanol at room temperature yielded annulation product 12d in 66% yield by ¹H NMR analysis and was accompanied by liberation of one equivalent of 1c (Scheme 9a).

Scheme 9.

Reactivity of Co(III) Aryl Complex 7a

Reaction with CO gave the carbonylation product 13a in 70% yield at room temperature (Scheme 9b). Surprisingly, the Co(III) complex 7a reaction with styrene under oxidative conditions did not yield annulation product 6. Instead, amidation product 11a was formed in moderate yield (Scheme 9c).^9o Dimerization from complex 7a proceeded smoothly and yielded 8a in quantitative yield (Scheme 9e). Only 3% yield of 8a was observed when the reaction was performed under strictly inert conditions, suggesting the intermediacy of a high-valent cobalt species (Scheme 9f).^9z

Next, catalytic reactions with both Co(III) aryl complex types were explored (Scheme 10). The C–H bond functionalization with diphenylacetylene was performed under reported reaction conditions^5a by replacing the Co-(OAc)₂ hydrate catalyst with Co(III) aryl complex 3d (Scheme 10a). Annulation product 12e was formed in a quantitative yield. Reaction under standard conditions^5c with CO gas using 3d as the catalyst yielded carbonylation product 13a in 70% yield (Scheme 10b). Reaction with styrene^5b gave product 12f in 60% yield (Scheme 10c). Diphenylacetylene reaction with 1c catalyzed by Co(III) aryl complex 7a gave 12d in 50% yield (Scheme 10d). However, reaction with CO gas gave a carbonylation product in only 13% yield (Scheme 10e). The addition of a catalytic amount of dipivaloylmethane ligand increased the yield of 13a to 35% (Scheme 10f). Complex 7a was also tested for benzamide 1c reaction with styrene, but 12g was not formed. Instead, the amination product 11a was produced (Scheme 9c). These results indicate that Co(III) aryl complexes 3 and 7 are active in catalysis and might be viable intermediates in some C–H bond functionalization reactions. However, experimental observations clearly indicate that the ligand plays an important role. The acetylacetonate ligand undergoes faster ligand exchange compared to dipivaloylmethane or benzamide (Schemes 7–9). Even though Co(III) aryl diamide complex 7a exhibits moderate reactivity in reactions with alkynes and carbon monoxide, it is likely not the active Co(III) intermediate in these C–H functionalization reactions.

Scheme 10.

Catalytic Reactions with Co(III) Aryl Complexes 3b, 3d, and 7a

Our observations show that ligand exchange from acetylacetonate to benzamide in Co(III) complex 3c is relatively slow, whereas coordination/migratory insertion of a reagent occurs much faster (Schemes 7–9). In the absence of other C–H functionalization reagents, such as in benzamide dimerization, and under more forcing conditions, Co(III) aryl diamide complex 7 could be a viable intermediate. However, the presence of bis(aminoquinoline amide)Co(III) complexes in catalytic reactions cannot be completely excluded, as the rate of ligand exchange under such conditions might be increased due to the higher substrate-to-cobalt ratio.

Synthesis and Reactivity Studies of Co(III) Migratory Insertion Products 5a and 5b.

We observed that both Co(III) aryl complexes 3a and 7a under oxidant-free conditions failed to give a coupling product with styrene. We hypothesized that in cobalt-catalyzed benzamide reactions with alkenes, high-valent, formally Co(IV) species might be involved via the oxidatively induced reductive elimination pathway. Based on these experiments, we assumed that a Co(III) migratory insertion complex could be sufficiently stable for isolation and characterization. We were pleased to find that the complex 3a reaction with ethyl acrylate or ethylene in ethanol at 30 °C gave migratory insertion complexes 5a and 5b, which were isolated in good yields (Scheme 11).

Scheme 11.

Synthesis of Migratory Insertion Complexes 10a and 10b

Both complexes 5a and 5b were purified by column chromatography and fully characterized. Suitable complex 5a crystals for XRD analysis were grown from EtOH at −5 °C. The ORTEP diagram of complex 5a is shown in Figure 4.

Figure 4.

ORTEP view of the molecular structure of Co(III) complex 5a. Thermal ellipsoids are drawn to encompass 50% probability. Hydrogens are omitted for clarity. Selected bond distances (Å) and angles (deg): Co(2)–C(68) 1.994(3), Co(2)–N(55) 1.963(2), Co(2)–O(82) 1.887(2) (1), Co(2)–O(74) 2.142(2), N(55)–Co(2)–O(82) 90.75(9), N(45)–Co(2)–N(55) 83.7(1), N(55)–Co(2)–C(68) 96.5(1), C(67)–C(68)–Co(2) 111.9(2).

Ethylene insertion complex 5b was recrystallized from methanol/pentane, affording X-ray quality crystals. The ORTEP diagram of complex 5b is shown in Figure 5. In complex 5b, a 7-membered cobaltacycle is formed by ethylene insertion into a cobalt–aryl bond. The sixth coordination site is occupied by the MeOH molecule in the trans-position relative to the alkyl moiety. Alternatively, if the crystals for XRD analysis were obtained by slow evaporation of a saturated C₂H₄/methylene chloride solution, the sixth coordination site was occupied by the carbonyl group of the neighboring molecule within the unit cell, forming a six-membered supramolecular cluster.¹¹

Figure 5.

ORTEP view of the molecular structure of Co(III) complex 5b. Thermal ellipsoids are drawn to encompass 50% probability. Hydrogens are omitted for clarity. Selected bond distances (Å) and angles (deg): Co(1)–C(23) 1.951(2), Co(1)–N(2) 1.941(1), Co(1)–O(1) 1.883(1), Co(1)–O(3) 2.119(1), N(1)–Co(1)–O(1) 87.88(5), N(1)–Co(1)–N(2) 83.33(6), N(2)–Co(1)–C(23) 95.17(7), C(24)–C(23)–Co(1) 115.18, C(15)–C(24)–C(23) 113.33(14).

Next, the reactivity of Co(III) migratory insertion complex 5b was examined (Scheme 12). As expected, Co(III) migratory insertion product 5b under external oxidant-free conditions did not undergo reductive elimination even at elevated temperatures (Scheme 12a,d). Additionally, no reductive elimination from Co(III) species occurred in the presence of a base (Scheme 12b) or a Lewis acid additive (Scheme 12c). Only when a stoichiometric amount of external oxidant was added, we observed product 12h formation (Scheme 12e), indicating the requirement for oxidation to a formally Co(IV) species for reductive elimination.

Scheme 12.

Reactivity Studies of Migratory Insertion Intermediate 5b

We investigated the redox potential of the Co(III) complex 5b using cyclic voltammetry. The CV of 5b, performed at room temperature and recorded from 0 to 1.0 V vs Ag/Ag⁺ at different scan rates, showed a quasi-reversible redox event at the potential of E^p,ox = 0.54 V vs Ag/Ag⁺ (at a 100 mV/s scan rate), which was assigned to the Co(III/IV) redox couple (Figure 6). The obtained results are in good agreement with literature data, demonstrating that the formation of high-valent Co species is plausible under our reaction conditions.^{9o,v,10a,b,k,l}

Figure 6.

CVs of complex 5b. Conditions: glassy carbon as the working electrode, Ag/AgNO₃ (silver wire in 0.1 M NBu₄ClO₄/CH₃CN solution; c(AgNO₃) = 0.01 M; E⁰ = −87 mV vs Fc/Fc⁺ couple) as the reference electrode, and a platinum wire as the counter electrode, nBu₄PF₆ (0.2 M), 5b (0.002 M) in CF₃CH₂OH, recorded from 0 to 1.0 V vs Ag/Ag⁺ at different scan rates.

Intrigued by the stability of Co(III) complex 5b, we wondered whether the second migratory insertion step is feasible in the presence of an alkyne. We conducted a ¹H NMR experiment, where an excess of 4-methoxyphenylacetylene was added to the solution of complex 5b in 1,1,2,2-tetrachloroethane-d₂. At 80 °C, we observed the formation of annulation product 12i arising from benzamide C–H bond functionalization with 4-methoxyphenylacetylene and liberation of ethylene (Scheme 13a). This result demonstrates the reversibility of the alkene migratory insertion step. Similarly, β-carbon elimination was observed when PPh₃ was introduced, yielding complex 3e and liberating ethylene (Scheme 13b).

Scheme 13.

β-Carbon Elimination in Migratory Insertion Complex 5b

The facile β-carbon elimination reaction was unexpected.^9m,14 Most of the literature examples for β-carbon elimination in late transition metal systems involve metal alkoxides, strained rings, or norbornene derivatives.¹⁵ Well-behaved, reversible β-carbon elimination examples are particularly scarce. Only a few carbon–carbon bond cleavage examples have been described in aminoquinoline amides, and most of them involve palladium complexes or strained rings.^9m,16 Thus, we further explored this unusual reaction. The reaction between 5b and 10 equiv of PPh₃ gave complex 3e. Kinetic analysis by ¹H NMR (Scheme 14a) showed the formation of 3e with ΔG^‡ = 24.9 kcal/mol.¹⁷ The PPh₃-bound 3e is in equilibrium with complex 5c, as the ¹H NMR experiment showed that 3e readily inserts ethylene to form 5c (Scheme 14b). Kinetic analysis by ¹H NMR shows that the transformation of 3e to 5c proceeds with ΔG^‡ = 23.3 kcal/mol. Complex 3e reacts with 10 equiv of 4-methoxypyridine, yielding pyridine-bound complex 3f with ΔG^‡ = 23.3 kcal/mol, determined by low-temperature ¹H NMR analysis (Scheme 14c).

Scheme 14.

Kinetic Analysis

Plausible mechanisms for β-carbon elimination are depicted in Scheme 15. The weakly bound ethanol ligand in 5b dissociates, opening a free coordination site and forming a 16e complex 5d, which can either trap the added PPh₃ ligand or β-carbon eliminate to afford an ethylene-bound complex 3g. After dissociation of ethylene, triphenylphosphine coordinates to cobalt via 3h to generate complex 3e. The measured barriers are likely the values for dissociating the phosphine ligands (5c to 5d and 3e to 3h). The barrier for β-carbon elimination is likely lower than those values.

Scheme 15.

Proposed Mechanistic Pathway for β-Carbon Elimination

Cobalt-Catalyzed Amination Studies.

At an early stage of the development of the cobalt-catalyzed C–H bond functionalization methodology, we discovered an arene C–H bond amination method (please see the Supporting Information for reaction conditions and substrate scope, Scheme S3, Tables S1, and S2). However, this transformation was not investigated further, as the reaction under cobalt catalysis is limited to cyclic six-membered ring amines, while our earlier copper-catalyzed methodology allows the use of nearly any amine and employs oxygen from air as a terminal oxidant.^4h Both types of Co(III) aryl intermediates, 3a and 7a, were investigated with respect to carbon–nitrogen bond formation with morpholine under stoichiometric and catalytic reaction conditions developed in our group (Scheme 16).¹⁸ First, we observed that morpholine easily replaces the weakly bound MeOH ligand in the Co(III) aryl complex 3a, forming 9a. Alternatively, Co(III) morpholine complex 9a was isolated in 78% yield when benzamide 1a was reacted with a stoichiometric amount of Co(III) nitrate under oxidative conditions in the presence of NaOAc base, dipivaloylmethane ligand, and morpholine.

Scheme 16.

Synthesis of Co(III) Complex 9a

Complex 9a is stable at room temperature and, after purification by column chromatography, was isolated as a dark orange solid. Crystals of complex 9a suitable for XRD analysis were grown by layering pentane over the solution of the complex in CH₂Cl₂ at room temperature. The ORTEP diagram of Co(III)–morpholine complex 9a is shown in Figure 7.

Figure 7.

ORTEP view of the molecular structure of Co(III) complex 9a. Thermal ellipsoids are drawn to encompass 50% probability. Hydrogens are omitted for clarity. Selected bond distances (Å) and angles (deg): Co(1)–N(1) 2.016(2), Co(1)–N(2) 1.906(1), Co(1)–O(5) 1.910(1), Co(1)–C(16) 1.904(2), Co(1)–N(3) 2.012(1), N(1)–Co(1)–N(3) 89.58, N(2)–Co(1)–C(16) 84.41(6), and C(16)–Co(1)–N(1) 166.46(6).

Under various conditions tested, no C–N bond formation occurred using Co(III) aryl complex 7a and morpholine (Scheme 17a). Next, Co(III) complex 9a was heated at 80 °C in the presence of the NaOAc base. However, after 16 h, neither 11a nor 11b was formed (analysis by ¹H NMR, Scheme 17b). Similar to the cobalt-catalyzed benzamide reaction with alkenes, C–N bond formation giving 11b occurred only when an external oxidant was added (Scheme 17c), indicating that a Co(IV) intermediate might be required for the reductive elimination step. Interestingly, only 11b was obtained, suggesting that the formation of 11c is slower than the conversion of 11c to 11b. Reaction of benzamide 1a with morpholine under optimal reaction conditions using 30 mol % of Co(III) complex 9a as the catalyst yielded 13% of 11c along with 47% of diamination product 11b (Scheme 17d). These results demonstrate that Co(III) complex 9a is a competent intermediate in cobalt-catalyzed C–H bond amination with morpholine.

Scheme 17.

Reactivity of Co(III) Complexes 7a and 9a

Cyclic voltammetry studies of the Co(III) complex 9a confirmed the plausibility of a Co(IV)/Co(II) catalytic cycle. CV of complex 9a was performed at room temperature and recorded from 0 to 1.2 V versus Ag/Ag⁺ at different scan rates. In the voltammogram, we observed a quasi-reversible redox event at the potential of E^p,ox = 0.71 V vs Ag/Ag⁺ (at a 100 mV/s scan rate), which was assigned to the Co(III/IV) redox couple (Figure 8).

Figure 8.

CV of complex 9a. Conditions: glassy carbon as the working electrode, Ag/AgNO₃ (silver wire in 0.1 M NBu₄ClO₄/CH₃CN solution; c(AgNO₃) = 0.01 M; E⁰ = −87 mV vs Fc/Fc⁺ couple) as the reference electrode, and a platinum wire as the counter electrode, nBu₄PF₆ (0.2 M), 12 (0.002 M) in MeOH, recorded from 0 to 1.2 V vs Ag/Ag⁺ at different scan rates.

Cobalt-Mediated Benzamide Dimerization Studies.

The cobalt-mediated aminoquinoline benzamide 1 homocoupling mechanism has been studied previously. In 2020, MacBeth and Musaev performed detailed mechanistic studies and proposed a Co(I)/(III) catalytic cycle involving either a charge neutral or an anionic pathway for benzamide 1 dimerization.^9l Recently, two groups have suggested the involvement of a high-valent Co(IV) intermediate.^9o,z We showed that Co(III) aryl complex 7a is a competent intermediate in the benzamide dehydrogenative dimerization reaction (Scheme 9). We were interested in exploring the reactivity of Co(III) aryl complexes 3a and 3c for the benzamide homocoupling reaction. The reaction of 3a with benzamide 1a in the presence of an oxidant gave only traces of dimerization product 8b. Instead of product 8b, we observed the formation of two Co(III) complexes 14a and 14b, which did not react further (Scheme 18a,b). Reaction of 3c with benzamide 1a in the presence of Na₂CO₃ base and the Mn(OAc)₂/O₂ oxidant system yielded 78% of the dimerization product (Scheme 18c). Using the Mn(OAc)₃·2H₂O oxidant, dimerization product 8b was obtained in 99% ¹H NMR yield (Scheme 18d). These results suggest that aminoquinoline can displace the acac ligand under the reaction conditions, whereas ligand exchange in the case of dipivaloyl methane is less productive.

Scheme 18.

Dimerization of Benzamide 1a Using Co(III) Complexes 3a and 3c

We found that 14c and 14d could be synthesized in acceptable yields starting from benzamide 1a using a stoichiometric amount of Co(acac)₂ salt in the presence of NaOPiv base (Scheme 19). The structures of 14c and 14d were determined by using XRD analysis. Related complexes were reported in 2016, although no further studies were performed with these intermediates.¹⁹ The ORTEP diagram of complex sym-14c is shown in Figure 9, and that of the Co(III) complex unsym-14d is shown in Figure 10.

Scheme 19.

Synthesis of Co(III) Complexes 14c and 14d

Figure 9.

ORTEP view of the molecular structure of sym-14c. Thermal ellipsoids are drawn to encompass 50% probability. Hydrogens are omitted for clarity. Selected bond distances (Å) and angles (deg): Co(1)–N(1) 2.016(2), Co(1)–N(2) 1.906(1), Co(1)–N(9) 1.938(2), Co(1)–N(10) 1.917(2), Co(1)–O(8) 1.895(2) N(9)–Co(1)–N(11) 101.17(9), N(9)–Co(1)–N(10) 98.66(9).

Figure 10.

ORTEP view of the molecular structure of unsym-14d. Thermal ellipsoids are drawn to encompass 50% probability. Hydrogens are omitted for clarity. Selected bond distances (Å) and angles (deg): Co(1)–N(11) 1.946(1), Co(1)–N(2) 1.914(2), Co(1)–N(34) 1.965(2), Co(1)–O(50) 1.900(2), N(11)–Co(1)–N(2) 83.88(6), N(11)–Co(1)–O(50) 85.45(6).

Complexes 14c and 14d are stereoisomers that contain a biaryl diamide dimerization product complexed to the Co(III) center via aminoquinoline and amide nitrogen atoms, with acetylacetonate as a bidentate ligand. In Co(III) complex sym-14c, both aminoquinolines are NMR-equivalent, while in unsym-14d, all carbon and hydrogen atoms are nonequivalent.

The formation of the relatively stable Co(III) complexes under the reaction conditions explains the necessity for 50 mol % Co(II) loading for the dimerization reaction. When reacted with a base in EtOH, different reactivities of complex 14 isomers were observed (Scheme 20). The unsym-14d liberated dimerization product 8b with a significantly lower reaction rate.

Scheme 20.

Reactivity Studies of sym-14c and unsym-14d

CONCLUSIONS

In summary, we have synthesized and fully characterized a number of organometallic key reaction intermediates in cobalt-catalyzed 8-aminoquinoline-directed C(sp²)–H bond functionalization. The reactivity of isolated Co(III) intermediates in stoichiometric and catalytic reactions was explored. Kinetic studies of the aminoquinoline benzamide coupling with alkenes show that the likely rate-determining step involves ligand exchange on cobalt. Two different Co(III) aryl intermediate types were isolated. The first possesses two aminoquinoline ligands around the cobalt center, one of which contains a cobalt–aryl bond. The second type contains one aminoquinoline-bound cobalt aryl and a ligated 1,3-diketone moiety. Interconversion between these two types of complexes is facile at higher temperatures but slow under milder conditions. All isolated intermediates are low-spin, diamagnetic, octahedral Co(III) complexes.

Two types of catalytic cycles are operative for aminoquinoline-directed cobalt-catalyzed C–H functionalizations. Annulation with alkynes and carbonylation with CO likely proceed via a Co(I)/Co(III) catalytic cycle. However, C–H bond functionalization with alkenes and amines, as well as dehydrogenative dimerization reactions, likely proceed via a (formally) Co(IV) species and involve oxidatively induced reductive elimination. This is broadly consistent with the classical work of Bergman and Kochi, showing that reductive eliminations of C(sp²) fragments are more facile compared to those of C(sp³) moieties, and that higher oxidation states of metals typically facilitate reductive elimination.²⁰

We have isolated and fully characterized the first complexes arising from migratory insertion into cobalt–aryl bonds in the aminoquinoline system. Specifically, complexes arising from insertion reactions with ethyl acrylate and ethylene were prepared. Interestingly, the reactivity of the ethylene insertion product shows that alkene migratory insertion is a reversible and relatively low-energy process, demonstrating facile β-carbon elimination in organometallic cobalt systems.

The ligand environment around the Co(III) center plays a crucial role in Co(III) aryl complex reactivity. If cobalt is ligated with a bulky dipivaloylmethane ligand, aminoquinoline amide dimerization is suppressed, allowing more difficult reactions, such as C–H amination, to occur.