Coordination‐Driven Direct C─H Metalation of N‐Heterocycles With a Superbasic Co(II) Amide Co(TMP)₂

ABSTRACT

Metalation of N‐heterocycles such as pyridines or diazines with conventional alkali metal bases is particularly challenging due to the limited activation of their C─H bonds coupled with the thermal fragility of the resulting organometallic intermediates. Building upon previous work in this area, we demonstrate that the superbasic cobalt(II) amide Co(TMP)₂ (TMP = 2,2,6,6‐tetramethylpiperidide) enables high‐yielding and highly regioselective C─H metalation of a broad range of N‐heterocycles via direct, redox‐neutral Co(II)–H exchange. This reactivity extends beyond single‐site activation, with controlled di‐cobaltation of non‐activated substrates such as pyrimidine and pyrazine. Via a combination of NMR reaction monitoring, isolation and characterisation of key organometallic intermediates, and density functional theory calculations, we establish that regioselectivity is dictated by coordination effects at positions α to nitrogen, overriding conventional trends based solely on C─H acidity. Exploiting cobalt's redox flexibility, oxidative treatment of the metalated intermediates with iodine enables selective C─N bond formation, giving sterically demanding TMP‐substituted heterocycles. Collectively, these findings establish Co(TMP)₂ as a mechanistically distinct and versatile reagent for the selective functionalisation of N‐heterocycles.

By enabling regioselectivities that are inaccessible with conventional bases, this study introduces a Co(II) amide platform for the deprotonative C─H metalation of sensitive N‐heterocycles. Subsequent interception of the resulting Co(II) intermediates with external oxidants affords a family of sterically congested, TMP‐substituted heterocycles via C─N bond formation. (TMP = 2,2,6,6‐tetramethylpiperidide).

1. Introduction

Typified by pyridine and pyrimidine, azines and diazines constitute an important family of N‐heterocyclic motifs within the structures of natural products, pharmaceuticals, agrochemicals, and functional materials [1, 2, 3, 4, 5, 6, 7, 8]. Despite their extensive utility, their metalation with conventional s‐block reagents such as alkyllithiums or lithium amides remains challenging. The low‐lying lowest unoccupied molecular orbitals (LUMOs) of these electron‐deficient heterocycles render them prone to nucleophilic addition and other side reactions upon exposure to strongly nucleophilic alkyl or aryl lithium bases [9]. As a result, extremely low temperatures and large excesses of base are often required, and even under such conditions undesired pathways occur frequently [9, 10, 11, 12, 13, 14]. For example, lithiation of pyridine with the Lochmann–Schlosser superbase (nBuLi/KOtBu) gives a mixture of regioisomers (C2:C3:C4 = 6:1:3) even at −105°C (Figure 1a) [13]. Improved, though still cryogenic, the selectivity is achieved with the unimetallic superbase nBuLi.LiDMAE (LiDMAE = Me₂N(CH₂)₂OLi), which affords exclusive α‐lithiation at −78°C when used in excess [15]. More recently, Thomas and coworkers reported that sodiation of pyridine with nBuNa at −78°C furnishes a 1:3.5 ratio of the C3 and C4 isomers (Figure 1b) [16]. Notably, the metalated intermediates generated in these transformations are highly fragile, even under cryogenic conditions, so their detailed structures have remained concealed [9].

FIGURE 1.

Selected examples of pyridine and pyrimidine metalation.

To address these limitations, several heterobimetallic bases that combine Li with a less electropositive metal such as Mg, Zn, or Ga have been developed [17, 18, 19, 20, 21, 22, 23]. These reagents often generate less labile organometallic intermediates, which allows milder reaction conditions and improved regioselective control. An exemplar is Knochel's C2‐zincation of pyrimidine using TMPZnCl.LiCl (TMP = 2,2,6,6‐tetramethylpiperidide), which proceeds with high selectivity (Figure 1c) [18]. Our groups have similarly shown that combining NaTMP with the Fe(II) amide [Fe{N(SiMe₃)₂}₂] affords a potent sodium ferrate capable of promoting direct ferration of pyridine via a redox‐neutral Fe(II)–H exchange. However, the selectivity of this method is limited, and quenching studies using D₂O gave a 2.2:1 mixture of the C3 and C4 metalated isomers (Figure 1d) [24].

Motivated by the search for more basic, earth‐abundant transition metal amides that can function as regioselective bases for aromatic metalation, we recently explored the amides M(TMP)₂ (M = Fe and Co). These reagents promote direct ferration or cobaltation of multi‐fluorinated arenes such as pentafluorobenzene and 1,2,4,5‐tetrafluorobenzene [25]. Their reactivity exhibits a distinctive superbasic character that resembles the behaviour of alkali‐metal amides, but with the added benefit that they form thermally stable organometallic intermediates.

Encouraged by these findings, we next investigated whether this approach could be extended to the regioselective functionalisation of non‐substituted pyridine and diazines. These substrates are considerably less activated than fluoroarenes in terms of C─H acidity (pK _a values) and lack pre‐installed directing groups. Here, we systematically evaluate the reactivity of these earth‐abundant transition metal amides toward a broad range of pyridines and diazines and uncover their unique ability to promote direct Co─H exchange with unprecedented control over regioselectivity (Figure 1e). Through NMR spectroscopic studies, isolation and characterisation of key organometallic intermediates, supplemented by density functional theory (DFT) calculations, we delineate the possible mechanism of these transformations and highlight the pivotal role of coordination effects in guiding the Co(II)─H exchange process.

2. Results and Discussion

2.1. Metalation of Pyridine With M(TMP)₂

We began our investigation using pyridine (py) as the model substrate for metalation. Relative to polyfluoroarenes, pyridine is poorly activated toward C─H metalation, in line with the higher pK _a values of its aromatic protons (C2 = 43.6; C3  = 41.0; C4 = 40.3 in DMSO versus 29.0 for C₆H₅F) [26]. Consistent with this, ¹H NMR monitoring of equimolar reactions of M(TMP)₂ (1^M ) (M = Fe, Co) and pyridine in C₆D₆ at room temperature showed paramagnetically shifted pyridine resonances (31.54 to −16.74 ppm for Fe and 24.18 to −10.07 ppm for Co), indicating quantitative formation of pyridine coordination adducts [(py)Fe(TMP)₂] (2^Fe ) and [(py)Co(TMP)₂] (2a). The absence of TMP(H) under these conditions supported the conclusion that no C─H metalation takes place at ambient temperature. Rerunning the reactions in pentane, 2^Fe and 2a could be isolated as crystalline solids in respective 93% and 73% yields (Figure 2). X‐ray crystallography confirmed that both complexes are monomeric in the solid state. Although severe twinning prevented a detailed discussion of bond parameters for 2^Fe , the presence of an Fe─N_py bond could still be unambiguously established (Figure S52).

FIGURE 2.

Reaction of 1^M (M = Fe, Co) with pyridine. Isolated crystalline yields. Molecular structures of 2a (left) and 3a (right). H atoms are omitted for clarity and displacement ellipsoids are displayed at 50% probability level. Selected bond lengths (Å) and angles (°) in 2a: Co1N1 2.095(4), Co1–N2 1.913(7), Co1–N3 1.890(7); N2–Co1–N3 147.2(2); in 3a: Co1–N1 2.026(2), Co2–C1 2.033(3), Co1–Co2 2.8830(5); Co1–N4–Co2 88.90(8).

Structurally, 2a features a trigonal planar Co(II) centre made up of two TMP ligands and one pyridine. The sum of the N─Co─N angles is 360(4)° ranging from 105.5(4) to 147.2(2)°. The Co─N_py distance of 2.095(4) Å lies at the upper limit of reported values for monopyridine adducts of Co(II) bis(amides) [2.055(5) Å in [(py)Co{N(SiMe₃)₂}₂] and 2.091(3) Å in [(py)Co{N(SiMe₃)(Dipp)}₂] (Dipp = 2,6‐diisopropylphenyl)], which is consistent with the steric demand of the TMP groups [27, 28]. In C₆D₆ solution at 300 K, 2^Fe and 2a exhibit magnetic moments of 5.18 and 4.06 μ _B (Evans method), respectively, values that agree well with other trigonal planar high‐spin Fe(II) and Co(II) complexes [25, 29, 30, 31].

Heating the reaction mixtures to 70°C for four hours led to divergent outcomes for Fe and Co. For 1^Co , the solution colour changed from yellow‐green to dark orange, and ¹H NMR monitoring showed conversion of 2a into a new paramagnetic species accompanied by the appearance of TMP(H), suggesting successful C─H metalation (Figure S36). Slow evaporation of the solvent at room temperature afforded dark‐coloured crystals suitable for X‐ray crystallography, which were identified as the C2‐metalated pyridine complex [(2‐pyridinyl)Co₂(TMP)₃] (3a) (Figure 2).

In 3a, the two Co(II) centres are bridged by one TMP ligand and one C2‐pyridinyl anion. The pyridinyl fragment binds via carbon to Co2 [2.033(3) Å] and via nitrogen to Co1 [2.026(2) Å], thereby forming a five‐membered {CoNCoNC} ring. Each cobalt centre also binds to a terminal TMP ligand to complete a distorted trigonal environment. From a structural viewpoint, 3a can be regarded as the metalation product [(2‐pyridinyl)Co(TMP)] coordinated to unreacted 1^Co or, equivalently, as a cationic dinuclear assembly that stabilises the anionic pyridinyl fragment.

Evidence that the dinuclear structure persists in solution was obtained by ¹H NMR spectroscopy (see Supporting Information for details). The spectrum of 3a displays sharp aromatic singlet resonances at 86.97, 66.56, 54.81, and 54.39 ppm, together with three distinct sets of signals for the TMP ligands. The well‐resolved features, despite the paramagnetic nature of the complex, indicate that the solid‐state structure is retained in solution. Consistent with this picture, the magnetic moment of 3a measured at 300 K is 3.75 μ _B, which is substantially lower than the spin‐only value of 5.48 μ _B expected for two uncoupled high‐spin Co(II) centres (S  = 3/2) [29, 30, 31]. Although this value is higher than those typically observed for strongly antiferromagnetically coupled Co(II) systems (1.95 μ _B at room temperature) [25], it nevertheless indicates the presence of partial magnetic coupling between the two cobalt centres. This interpretation is supported by their close spatial proximity in the solid‐state structure [Co1···Co2 = 2.8830(5) Å] (Figure 2).

Remarkably, 3a is a thermally robust pyridine‐metalated cobalt complex, in marked contrast to the highly fragile metalated pyridine intermediates typically witnessed with s‐block bases. Repeating the metalation of 1^Co with 0.5 equivalents of pyridine gave 3a cleanly and quantitatively according to ¹H NMR monitoring, and the complex could be isolated in 79% yield (Figure 2, see Supporting Information for details). Notably, this reactivity is unique to cobalt, as 1^Fe showed no evidence of pyridine metalation under identical conditions but instead decomposed over sixteen hours at 70°C (Figure S39).

These observations establish that 1^Co functions as a base capable of direct Co(II)–H exchange with pyridine, a substrate that is significantly less activated in terms of pK _a than fluorobenzenes. This reaction proceeds selectively at the least acidic C─H position, in contrast with the behaviour of s‐block reagents such as the Lochman–Schlosser superbase or nBuNa [13, 16], since these typically metalate more acidic sites (Figure 1). Although C2‐lithiation of pyridine can be accomplished using nBuLi.LiDMAE, that protocol requires strict temperature control and immediate trapping at −78°C to prevent decomposition of the organolithium intermediate [15].

While 3a forms through a direct Co(II)–H exchange, it is helpful to place this result within the broader landscape of C2‐functionalisation of pyridines. Transition metal‐mediated C─H activation at C2 has been reported in both catalytic and stoichiometric regimes, taking place typically with a change in the oxidation state of the metal [32, 33, 34, 35, 36]. For instance, Chirik and co‐workers reported a Co(I) complex supported by a tridentate PNP pincer ligand that promotes catalytic C2‐borylation via a Co(I)/Co(III) redox manifold [34]. In addition, several stoichiometric C2‐selective activations of pyridines have been reported which rely on bimetallic cooperation [37, 38, 39, 40]. This includes an iron–aluminylene system reported by Crimmin and co‐workers, which engages in reductive deprotonation of pyridine operating as a frustrated Lewis pair [38].

Because 3a retains three TMP ligands, we next examined whether any of them remained sufficiently basic to metalate a second equivalent of pyridine. However, neither prolonged heating at 70°C for 16 h nor the addition of LiCl or nBu₄NCl [25], promoted further reactivity, and 3a was recovered unchanged in all cases. These experiments suggested that cobaltation terminates after formation of the dinuclear complex since its steric surrounds prevent access to the cobalt centres.

To rationalise the unusual reactivity and the regioselectivity of 1^Co toward pyridine, as well as the stability of 3a, we turned to DFT calculations at the ωB97XD level (see Supporting Information for details). As shown in Figure 3a, pyridine initially coordinates to 1^Co to give the stable adduct 2a (−10.1 kcal mol⁻¹), and this association facilitates the subsequent Co(II)–H exchange via transition state TS1 with a barrier of +23.4 kcal mol⁻¹. Alternative pathways involving metalation at the meta or para positions were also examined and were significantly higher in energy (Figure S76). The intermediate that follows, I1, displays a Co─N_py bond rather than the Co─C_py connectivity characteristic of the metalated product. This feature highlights the strong cobalt–nitrogen interaction that stabilises transition state TS1 and enables selective deprotonation at the least acidic C─H site. This mechanistic picture is reminiscent of our earlier work on pentafluorobenzene, where substrate pre‐coordination was identified as an important contributor to metalation and the Co–substrate interactions were primarily non‐covalent [25]. In the present system, however, non‐covalent interaction (NCI) analysis of TS1 shows no sign of non‐covalent Co···N_py contacts, which is consistent with the formation of a covalent C─N bond (Figure 3b). Quantum theory of atoms in molecules (QTAIM) analysis further reinforces this interpretation, revealing a bond critical point (BCP) with relatively high electron density [ρ(r) = 0.315 a.u.], a negative Laplacian [∇²ρ(r) = −8.811 a.u.], and negative electronic energy density [H(r) = −3.244 a.u.], all characteristic of covalent bonding (Figure 3c). Complementary electron density difference maps show depletion at N_py and accumulation around Co upon interaction, along with charge buildup at the formed carbanion.

FIGURE 3.

a) Gibbs energy profile for the C─H metalation of pyridine by 1^Co at 298 K and 1 atm. Optimised structures of the transition states are shown as insets, with relevant bond lengths (Å); some H atoms are omitted for clarity. b) Non‐covalent interaction (NCI) analysis (isovalue = 0.50 a.u.) of TS1, highlighting attractive interactions between the pyridine and Co(TMP)₂ fragments. c) Quantum theory of atoms in molecules (QTAIM) parameters and 3D electron density difference map (isovalue = 0.80 a.u.) between the pyridine and Co(TMP)₂ fragments in TS1. Electron density accumulation (red) between Co···N_py indicates a covalent bond, with electron density flowing from the N_py atom to the Co center. d) Natural bond orbital (NBO) analysis of TS1, depicting the main orbital interactions between pyridine and Co(TMP)₂ fragments as isosurfaces (isovalue = 0.05 a.u.). Key interaction energies for the Co···N_py bond, calculated using second‐order perturbation theory (ΔE_SOPT), are reported in kcal mol⁻¹. Further details on the software used for the analyses in panels b)–d) can be found in the Supporting Information.

Natural bond orbital (NBO) analysis provides additional insight into this interaction (Figure 3d). Two principal donor‐acceptor interactions correspond to donation from the N_py lone pair into distinct Co valence orbitals and together stabilise the system by −26.0 kcal mol⁻¹. A third interaction involves donation from the nitrogen lone pair into Co─N_TMP antibonding orbitals (−8.4 kcal mol⁻¹), suggesting that weakening of both Co─N_TMP bonds precedes TS1 and helps facilitate deprotonation. These combined observations rationalise how TMP ligands can activate N‐containing arenes despite their comparatively higher α‐C─H pK _a values and explain the regioselectivity observed experimentally in the formation of 3a.

To probe this mechanistic interpretation experimentally, we evaluated the reactivity of 1^Co‐IMes (IMes = 1,3‐bis(2,4,6‐trimethylphenyl)imidazol‐2‐ylidene), which contains a non‐labile NHC ligand coordinated to the Co(II) centre. In this case, no pyridine metalation was observed, even under forcing conditions (70°C, 16 h; see Supporting Information for details), a result consistent with the inability of pyridine to displace IMes and form adduct 2a.

From I1, the reaction proceeds through the low‐barrier transition state TS2 (+0.8 kcal mol⁻¹) to give the C‐bound isomer I2, which then undergoes co‐complexation with unreacted 2^Co to form the dinuclear product 3a together with TMP(H) (Figure 3a). This step is highly exergonic (–12.3 kcal mol⁻¹) and drives the overall reaction forward. We also evaluated a possible alternative pathway in which I2 dimerises with elimination of two equivalents of TMP(H) to generate a putative [{(2‐pyridinyl)Co(TMP)}₂] species analogous to those proposed in fluoroarene metalation. However, this pathway is endergonic by +15.7 kcal mol⁻¹ and therefore was not considered further (Figure S77).

Taken together, these findings show that coordination of the Co(II) centre to the pyridine nitrogen has a dual role: it controls regioselectivity at TS1 and stabilises the metalated intermediate I2 through co‐complexation with 2^Co and elimination of TMP(H), thereby preventing the reversal of the cobaltation.

Finally, to understand why Fe(TMP)₂ does not metalate pyridine, we performed analogous calculations for the iron system. The sequence of intermediates (TS1^Fe , I1^Fe , TS2^Fe and I2^Fe ) resembles that calculated for cobalt (Figures S78 and S79), yet the subsequent co‐complexation step differs markedly. Formation of a putative [(2‐pyridinyl)Fe₂(TMP)₃] complex (3a^Fe , Figure S79) does not provide sufficient stabilisation to render the process thermodynamically favourable (+2.1 kcal mol⁻¹). This distinction accounts for the lack of pyridine metalation observed experimentally with Fe(TMP)₂ and highlights the critical influence of the greater Lewis acidity of Co(II) relative to Fe(II) in enabling this low‐polarity metalation.

2.2. Extending Co(II)–H Exchanges to Other N‐Heterocyclic Substrates

We next explored the scope of this cobaltation method, focusing on N‐heterocycles (Figure 4). Reactions were carried out in pentane, using 1^Co and the relevant substrate in a 2:1 ratio, furnishing the cobaltation products 3b‐m, which were subsequently isolated and characterised spectroscopically.

FIGURE 4.

Substrate scope of cobaltation with 1^Co . Reaction conditions: 1^Co (0.2 mmol), substrate (0.1 mmol), pentane (2 mL), rt. Isolated crystalline yields are shown; conversions by ¹H NMR spectroscopy are given in parentheses. [a] Benzene as solvent, 70°C. [b] 1:1 mixture of C2‐ and C6‐metalated products determined by ¹H NMR. Inset: Molecular structures of selected complexes 3b, 3c, 3j, 3l, and 3m. H atoms omitted for clarity and displacement ellipsoids displayed at 50% probability level.

Their solid‐state structures were established by X‐ray crystallography (see Supporting Information for details). Isolated crystalline yields are shown in Figure 4, with ¹H NMR conversions given in parentheses. In all cases, dinuclear complexes with a similar structure to that discussed for 3a were obtained, and cobaltation occurred at a C─H site located in close proximity to a nitrogen centre. Except for 3c (vide supra), complexes 3b‐m display the same dinuclear tris‐TMP five‐membered ring motif described for 3a and exhibit similar structural parameters in most cases (see Supporting Information for details). The Co···Co distances in 3b, 3d, 3f‐k, and 3m range from 2.7349(2) to 2.9282(3) Å, and their solution magnetic moments lie between 3.43 and 3.98 μ _B in C₆D₆ at 300 K. These values are consistent with two antiferromagnetically coupled high‐spin Co(II) centres (S = 3/2). In contrast, longer Co···Co distances are observed in 3c [3.1177(4) Å] and 3l [3.0908(7) Å]; this is reflected in their higher solution magnetic moments [both 4.68 μ _B in C₆D₆ at 300 K by Evans method], which point to a reduced degree of antiferromagnetic coupling in these two complexes [29, 30, 31].

The substrate 2‐phenylpyridine was chosen because its metalation has been widely explored with main‐group bases. In particular, using the lithium alkyl/lithium alkoxide combination nBuLi.LiDMAE favours regioselective metalation at the C6‐position on the pyridine ring [41], whereas the Turbo Grignard base TMPMgCl.LiCl reported by Knochel and co‐workers promotes C2‐magnesiation on the phenyl ring, although in the latter case an excess of base (2.0 equivalents) and prolonged heating (55°C, 24 h) are required [42]. Under our conditions, treatment of 2‐phenylpyridine with 1^Co leads to quantitative Co─H exchange at the C6‐position of the pyridinyl ring after 16 h at room temperature (or after 3 h at 70°C), affording complex 3b in 74% crystalline yield. X‐ray crystallography confirmed that 3b adopts the same structural motif as 3a, in which the pyridine nitrogen acts as a σ‐donor to a Co(TMP)₂ unit (Figure 4). This suggests that formation of the bimetallic motif is better stabilised when metalation occurs on the pyridinyl ring rather than on the phenyl ring. Although this regioselectivity resembles that observed in organolithium chemistry (vide supra), it contrasts with the behaviour reported for many transition‐metal‐catalysed C─H functionalisation protocols of 2‐phenylpyridine [43, 44, 45, 46, 47, 48, 49, 50, 51], which typically activate the C2‐position of the phenyl ring and gain additional stabilisation through coordination with the pyridine nitrogen [52, 53, 54].

We then turned to methyl‐substituted pyridines. For 2‐picoline, cobaltation proceeds cleanly at the benzylic position at room temperature within one hour to give complex 3c in 68% crystalline yield, and ¹H NMR monitoring in C₆D₆ shows this transformation is quantitative. The molecular structure of 3c is related to those of 3a and 3b; however, metalation at the methyl position generates a six‐membered {CoNCoNCC} ring, which results in a larger Co···Co distance [3.1177(4) Å vs 2.8830(5) Å in 3a]. In addition, the Co─C bond in 3c [2.0898(10) Å] is slightly longer than the Co─C bonds in 3a and 3b [2.033(3) and 2.0321(13) Å], consistent with its sp³ character. The electron density of the monoanionic picolyl fragment is delocalised across the azaallyl moiety, as evidenced by the increased double bond character of C1–C2 [1.4426(15) Å] compared to the corresponding bond in free 2‐picoline [1.507 Å] [55]. The observed regioselectivity, which targets the most acidic site of the molecule (pK _a = 34.5 in THF) [56], agrees with trends reported for main‐group metal bases [57, 58, 59, 60], while still allowing pyridine nitrogen coordination and formation of the dinuclear assembly via the Co1─N1 bond (Figure 4). Interestingly, for 4‐picoline, full conversion into the C2‐metalated product 3d was achieved upon heating at 70°C for 5 h, and 3d could be isolated as a crystalline solid in 56% yield. For 3‐picoline, which contains two non‐equivalent α‐C─H positions relative to the nitrogen, cobaltation produces 3e in 77% yield as a 1:1 statistical mixture of the C2‐ and C6‐metalated products (see Supporting Information). For these substrates, preferential α‐metalation of the pyridine rings contrasts with the more commonly observed benzylic metalation when using alkali‐metal alkyls or amides [57].

We next investigated substituted fluoro‐ and methoxypyridines, which contain classical directing ortho‐metalation (DoM) groups. Remarkably, reaction of 1^Co with 2‐, 4‐, and 3‐methoxy pyridine leads to quantitative formation of complexes 3f, 3g, and 3h, respectively, each arising from regioselective α‐deprotonation of the pyridine ring. The products were isolated in 48%, 67%, and 79% yields, respectively, with reactions for 3f and 3g requiring heating at 70°C. For 3‐methoxypyridine, which has two non‐equivalent α positions, cobaltation takes place exclusively at the site that is also ortho to the methoxy group.

Further insight into this behaviour was obtained from reaction monitoring. ¹H NMR studies of 1^Co with one equivalent of 2‐methoxypyridine revealed initial formation of the coordination adduct [(2‐methoxy‐py)Co(TMP)₂] (2f), followed by its conversion into the metalated 3f upon heating. This sequence underscores the importance of preliminary Co─N coordination in these metalation processes (see Supporting Information). Overall, these results indicate that for 2‐methoxypyridine the α‐directing effect of the pyridine nitrogen overrides the classical ortho‐directing effect of the methoxy group. This outcome contrasts sharply with zincation of similar substrates using bimetallic bases [MZn(TMP)R₂] (M = Li, K; R = tBu, Et), where selective ortho‐zincation at the OMe group is observed [61, 62]. In a related case, Mongin and coworkers examined the metalation of 2‐methoxypyridine with an in situ generated LiFe(TMP)₃ mixture, which after quenching with iodine afforded 3‐iodo‐2‐methoxypyridine and 2,2′‐dimethoxy‐3,3′‐bipyridine, outcomes that are again consistent with favoured ortho metalation [63].

On moving to fluoropyridine substrates, the same α‐to‐nitrogen regioselectivity was observed, affording complexes 3i, 3j, and 3k in 66%, 75%, and 71% isolated yields, respectively. These results are particularly striking because fluorine exerts a strong inductive effect and typically acidifies ortho C─H positions in DoM chemistry. For example, in 2‐fluoropyridine the pK _a value of the C─H bond ortho to the C─F unit is 34.0, whereas the C─H bond α to the pyridine nitrogen has a pK _a of 40.7 [64]. Accordingly, s‐block metal bases such as the Lochmann/Schlosser superbase, alkali‐metal magnesiates and zincates metalate this substrate at C3 [64, 65], in clear contrast to the unique C6‐regioselectivity observed with 1^Co . This divergence is especially notable in 3j, where the C─H position doubly activated by two fluorine substituents remains untouched.

Adding significance, the cobaltation strategy was found to extend beyond simple pyridines. For caffeine, 1^Co promotes C8‐metalation to afford 3l in 73% crystalline yield. Reaction of 1^Co with N‐benzylideneaniline leads to selective α‐metalation at the imine CH═NPh unit, giving 3m in 78% crystalline yield. This regioselectivity differs from that observed for LiTMP, which metalates ArCH═NCy aldimines (Cy = cyclohexyl) at the ortho position of the aryl group [66]. It is also worth noting that benzylideneaniline is known to undergo C─H activation at the same ortho position with various transition metal complexes, giving cyclometalated products [67, 68, 69].

Collectively, these results demonstrate the excellent regioselectivity of 1^Co toward α‐cobaltation of a broad range of pyridines and related N‐containing substrates under mild conditions. In line with the DFT studies on pyridine cobaltation, coordination of the substrate nitrogen to Co(II) appears to fix the regioselectivity of these reactions and, in many cases, overrides the intrinsic preferences implied by the pK _a values of the available C─H sites (see Supporting Information). This interpretation is further supported by the lack of reactivity of 1^Co toward 2,6‐difluoropyridine. In this substrate, the C─H bonds ortho to the C─F units are relatively acidic (pK _a = 31.9 in THF) [64], yet ¹H NMR monitoring shows that 2,6‐difluoropyridine does not coordinate to 1^Co and is inert toward metalation. Similarly, no reaction occurs with 1‐methylindole, which lacks a coordinating nitrogen atom, or dimethylaniline and anisole, even though the ortho C─H bonds in anisole are considerably more activated that those in pyridine.

Interestingly, when 1^Co was treated with methylpicolinate, the complex reacted preferentially as a nucleophile rather than a base. In this case, nucleophilic substitution furnished pyridine‐2‐yl(2,2,6,6‐tetramethylpiperidin‐1‐yl)methanone together with the cobalt complex [(MeO)Co₂(TMP)₃] (4) (Figure 5). ¹H NMR reaction monitoring showed that formation of these species is quantitative. In a similar way, reaction of 1^Co with benzonitrile affords the ketimide complex [{(Ph)(TMP)C═N}Co₂(TMP)₃] (5) in 88% crystalline yield as the product of nucleophilic addition across the C─N triple bond. These findings demonstrate that, despite the exceptional basicity of 1^Co , the TMP ligands bound to Co(II) retain significant nucleophilic character. In this context, we recently reported the ability of 1^Co to undergo double insertion of CO₂ into its two Co─N_TMP bonds to form a bis(carbamate) species [25]. This nucleophilic behaviour contrasts with that of LiTMP, which acts as a base in metalating benzonitrile at −78°C to give a mixture of mono‐ and di‐ortho‐substituted products [70].

FIGURE 5.

Reaction of 1^Co with methyl picolinate and benzonitrile. Isolated crystalline yields are shown. Molecular structures of 4 (left) and 5 (right). H atoms omitted for clarity and displacement ellipsoids displayed at 50% probability level. Selected bond lengths (Å) and angles (°) in 5: Co1─Co2 2.6630(15); N5–C16 1.2787(10).

Isolated as crystalline products, compounds 4 and 5 were characterised by X‐ray crystallography (Figure 5 and Supporting Information). Both complexes display a dinuclear {Co₂(TMP)₃} structural core similar to those in 3a‐m, where the two Co(II) centres connect by a methoxy O bridge in 4 or by the ketimido N═C(Ph)(TMP) N bridge in 5.

2.3. Direct Cobaltation of Diazines and Diazanaphthalenes

We next evaluated the ability of 1^Co to promote the metalation of diazines. Although these molecules are important building blocks in pharmaceutical chemistry, their selective deprotonation with s‐block metal bases is particularly challenging [14, 20, 71, 72], especially for non‐substituted substrates that lack a directing group. Typically, strict temperature control is required and, even under cryogenic conditions, unwanted side reactions can occur because of the extreme fragility of the metalated intermediates. These issues can be partially mitigated by using cooperative bimetallic reagents that combine Li with a less electropositive metal such as Zn [18, 20, 61, 72, 73], Ga [19], or Cd [9]. Reaction of pyrazine (C₄H₄N₂, pyz) with one equivalent of 1^Co in pentane at room temperature causes an immediate colour change from dark red to dark orange. Slow evaporation of the solvent at −30°C afforded dark orange crystals of [(2,5‐C₄H₂N₂‐pyz)Co₄(TMP)₆] (6a) in 48% yield.

X‐ray crystallographic analysis revealed that 6a contains a two‐fold deprotonated pyrazine ring, substituted at the less sterically hindered 2,5‐positions by Co(II) atoms (Figure 6). The complex adopts a centrosymmetric tetranuclear motif in which each deprotonated carbon binds to one cobalt center [Co2–C1, 2.0145(14) Å], and the remaining two cobalt atoms coordinate to the nitrogen atoms of the heterocycle [Co1–N1, 2.0083(13) Å]. The two types of cobalt centres, Co1 and Co2, are linked by a bridging TMP ligand, and each completes a distorted trigonal planar coordination sphere with a terminal TMP ligand. This coordination arrangement resembles that found in 3a‐m but here generates a distinctive fused three‐ring (5,6,5) framework. The tricyclic backbone in 6a is nearly planar, with Co1 and the bridging N3 lying slightly out of the mean plane by 0.171 and 0.435 Å, respectively, which helps relieve steric congestion from the adjacent TMP ligands. Compound 6a was also characterised by ¹H NMR spectroscopy in C₆D₆, displaying a distinct resonance at 126.26 ppm for the remaining aromatic C─H protons of the dimetallated pyrazine ring, along with paramagnetically shifted signals for the TMP ligands spanning from 32.77 to −25.45 ppm. As far as we can ascertain, 6a represents the first structurally characterised example of di‐cobaltation of an N‐heterocycle.

FIGURE 6.

Substrate scope of (di)‐cobaltation with 1^Co . Reaction conditions: 1^Co (0.4 or 0.2 mmol), substrate (0.1 mmol), pentane (2 mL), rt. Isolated crystalline yields are shown; conversions by ¹H NMR are given in parentheses. [a] Benzene as solvent, 70°C. Inset: Molecular structures of complexes 6a and 7e. H atoms omitted for clarity except those on the aromatic ring; displacement ellipsoids shown at 50% probability level.

To improve the efficiency of the process, pyrazine was treated with four equivalents of 1^Co at 70°C for four hours, which allowed isolation of 6a in 85% yield (Figure 6). Monitoring this reaction by ¹H NMR spectroscopy showed that, at room temperature and using only two equivalents of 1^Co , an additional paramagnetic species was present in solution together with 6a. This species was assigned to a C2‐monometalated intermediate (see Supporting Information for details).

The regioselectivity of this transformation is particularly notable. Although pyrazine can be 2,5‐dimetallated using superbasic bimetallic systems that combine LiTMP with zinc, gallium, or cadmium reagents, direct use of LiTMP alone is far less selective. At −78°C, a four‐fold excess of LiTMP gives mainly C2‐substituted pyrazine together with only small amounts (2%–16%) of C2,C5‐disubstituted products after electrophilic quench [14]. In contrast, cobaltation with 1^Co provides 6a cleanly, indicating that 1^Co is not only more selective but also more efficient than classical LiTMP‐based protocols for this sensitive heterocycle.

To try to rationalise this unusual reactivity, we performed additional DFT calculations for pyrazine. Coordination of pyrazine to 1^Co gave intermediate I1 in an exergonic process (−7.3 kcal mol⁻¹), identifying I1 as the most stable species in solution (Figure 7). From this intermediate, C─H cobaltation proceeds via a pathway analogous to that described for pyridine (Figure 3). Co(II)–H exchange through TS1 (∆G^‡ = +19.1 kcal mol⁻¹) forms I2, in which the deprotonated pyrazine remains bound through nitrogen. I2 then undergoes a change in coordination via TS2 (∆G^‡ = +0.6 kcal mol⁻¹), affording the mono‐C2‐cobaltated intermediate I3, formulated as [(TMPH)Co(C₄H₃N₂‐pyz)(TMP)]. This rearrangement is highly exergonic (−15.8 kcal mol⁻¹) and is followed by coordination of one equivalent of I1 with displacement of TMP(H), which produces the di‐cobalt intermediate [(C₄H₃N₂‐pyz)Co₂(TMP)₃], I4 (−11.4 kcal mol⁻¹) and pyrazine. Although I4 could not be isolated, its formation was observed spectroscopically (vide supra).

FIGURE 7.

Gibbs energy profile computed for the first and second C─H metalations of pyrazine by 1^Co at 298 K and 1 atm. Optimized structures of the transition states are shown as insets, with relevant bond lengths (Å); some H atoms are omitted for clarity.

Intermediate I4 can then bind an additional equivalent of I1 through its remaining nitrogen to give I5 (−15.5 kcal mol⁻¹). This coordination facilitates the second Co(II)–H exchange at C5. The second metalation proceeds via TS3 and TS4 to afford I7, which ultimately yields 6a upon co‐complexation with another equivalent of I1 and release of TMP(H) and pyrazin. Overall, formation of 6a is exergonic by −23.8 kcal mol⁻¹ with the rate‐limiting step being the second metalation (∆G ^‡ = +21.5 kcal mol⁻¹). Importantly, binding of 1^Co to I4 to generate I5 is thermodynamically favourable by −4.1 kcal mol⁻¹, making I5 the lowest‐energy intermediate on the reaction profile. This, together with the favourable energetics toward 6a, explains why 6a forms even when only two equivalents of 1^Co are used experimentally. The fact that the second metalation is rate‐limiting also accounts for the observation of a C2‐monometalated intermediate in solution alongside 6a.

Having established that 1^Co can promote both mono‐ and di‐cobaltation of pyrazine, we next examined whether this reactivity could be extended to other diazines. The synthetic metalation approach also proves effective for pyrimidine (C₄H₄N₂, pym), a substrate that cannot be metalated by LiTMP even under strictly controlled low‐temperature conditions [14]. Reacting pyrimidine with four equivalents of 1^Co at 70°C for 16 h furnished [(4,6‐C₄H₂N₂‐pym)Co₄(TMP)₆] (6b) in 78% crystalline yield. X‐ray diffraction analysis showed that 6b is isostructural with 6a and features the same fused three‐ring (5,6,5) motif in which four cobalt atoms coordinate to the doubly deprotonated [C₄H₂N₂]^2– dianion (Figure 6). Its paramagnetic ¹H NMR spectrum in C₆D₆ displays two well‐resolved resonances at 132.68 and 126.21 ppm for the remaining C─H hydrogens of the dimetallated ring (see Supporting Information). To our knowledge, 6b represents the first example of selective dimetallation of pyrimidine.

Notably, decreasing the loading of 1^Co to two equivalents and performing the reaction at room temperature led instead to the mono‐C4‐cobaltated product 7b, which was isolated in 48% crystalline yield (>95% conversion by ¹H NMR; see Supporting Information). Pyrimidine is another particularly challenging substrate in metalation chemistry. Knochel and co‐workers showed that TMPZnCl.LiCl promotes selective C2‐zincation through coordination‐controlled effects, whereas combinations of LiTMP with equimolar Ga(CH₂SiMe₃)₃ or ZnCl₂ instead furnish C4‐metalated products [18, 19, 20]. The behaviour of 1^Co therefore aligns with this latter trend while enabling direct cobaltation under comparatively mild conditions.

We then assessed the reactivity of 1^Co toward pyridazine (C₄H₄N₂, pdz). Reaction at room temperature selectively produced [(3‐C₄H₃N₂‐pdz)Co₂(TMP)₃] (7c), which was isolated in 49% crystalline yield and forms quantitatively according to ¹H NMR monitoring (Figure 6). Neither altering the stoichiometry nor raising the temperature resulted in di‐cobaltation. This outcome can be attributed to the steric congestion imposed by the arching {Co₂(TMP)₃} fragment in 7c, which appears to hinder binding of an additional equivalent of 1^Co to the remaining nitrogen atom.

Finally, we explored whether cobaltation would extend to more complex N‐heterocycles. Treatment of quinoxaline and quinazoline with 1^Co at room temperature for two hours gave complexes 7d and 7e in respective 76% and 52% crystalline yields, respectively, with metalation occurring selectively at C2 in quinoxaline and C4 in quinazoline (Figure 6). The structures of 7b‐e were confirmed by X‐ray crystallographic analysis (see Supporting Information).

2.4. Applications in Oxidative C─N Coupling Reactions

To explore how the cobaltated intermediates might be used in downstream transformations, we probed their reactivity under oxidative conditions. Reacting 3a with I₂ gave 2‐(2,2,6,6‐tetramethylpiperidin‐1‐yl)pyridine (8a) in 50% yield, arising from selective coupling of a TMP group with the C2‐pyridinyl fragment (Figure 8). Comparable results were obtained when N‐iodosuccinimide or 5‐(trifluoromethyl)dibenzothiophenium trifluoromethanesulfonate were used as oxidants, each providing 8a in similar yields (see Supporting Information).

FIGURE 8.

C─N coupling of selected (di)cobaltated species mediated by iodine. Yields were determined by ¹H NMR using C₆Me₆ (10 mol%) as an internal standard; isolated yields are given in parentheses.

Encouraged by this initial result, we evaluated whether related cobaltated complexes would undergo analogous oxidative transformations. Indeed, complexes 3b, 6a, and 7b delivered the corresponding TMP‐substituted products 8b, 8c, and 8d in respective 59%, 65%, and 70% yields (Figure 8). Notably, 8c, in which two TMP groups participate in C─N bond formation, could be isolated in crystalline form and structurally characterised by X‐ray diffraction.

These findings demonstrate that the Co(II) system not only activates heterocycles through an initial C─H metalation step but also enables subsequent oxidative C─N bond formation. The resulting amines 8b‐d are new, and their sterically encumbered TMP substituents make them difficult to access by conventional C─N cross‐coupling methods such as Buchwald–Hartwig reactions. In this regard, the cobaltation–oxidation sequence provides a complementary strategy for assembling highly hindered C─N bonds from simple N‐heterocyclic precursors.

3. Conclusions

This study has revealed the extraordinary power of the homoleptic cobalt(II) bis‐amide Co(TMP)₂ to promote the regioselective metalation of a broad range of N‐heterocycles via direct Co(II)–H exchange, including pyridine and highly sensitive diazines. These substrates are notoriously difficult to metalate using conventional lithium‐ or magnesium‐based reagents. While cobalt complexes have previously been reported in C─H activation chemistry of heteroarenes, such processes typically proceed through changes in the oxidation state of the metal. In contrast, Co(TMP)₂ operates in redox‐neutral processes displaying a reactivity profile that closely parallels that of main‐group metal amides, while simultaneously affording access to thermally robust organocobalt intermediates.

By combining detailed experimental and computational studies, we elucidate the mechanistic origins of this reactivity, and particularly the decisive role played by coordination effects. Initial binding of the N‐heterocycle to the cobalt centre not only enables Co–H exchange but also dictates the regioselectivity of such metalations, consistently favouring C─H sites α to nitrogen. Subsequent stabilisation of the metalated intermediate through co‐complexation with a second equivalent of Co(TMP)₂ renders the cobaltation thermodynamically favourable and effectively suppresses reversibility. This coordination‐driven activation leads to regioselectivities that override trends predicted solely by C─H acidity (pK _a values). Notably, even weakly acidic substrates such as pyrazine and pyrimidine undergo controlled two‐fold di‐cobaltation when stoichiometry and temperature are judiciously adjusted.

Finally, the redox flexibility of cobalt can be exploited to transform these organometallic intermediates into valuable organic products. Exposure of the metalated species to external oxidants such as iodine triggers selective C─N bond formation, enabling coupling of the heteroaryl fragment with a sterically demanding TMP group.

Overall, this work firmly establishes Co(TMP)₂ as a versatile and mechanistically distinct reagent for the selective functionalisation of N‐heterocycles. In bridging concepts from main‐group deprotonative metalation and transition metal coordination chemistry, this non‐precious transition metal reagent opens up new avenues for the controlled activation and derivatisation of challenging heteroaromatic substrates.

Author Contributions

Na Jin: investigation, writing – original draft, formal analysis, data curation. Manting Mu: investigation, writing – original draft, formal analysis. Max García‐Melchor: conceptualisation, writing – review and editing, funding acquisition, project administration, supervision. Eva Hevia: conceptualisation, writing – review and editing, funding acquisition, project administration, supervision.

Funding

HORIZON‐Europe, Research Ireland, and Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung.

Conflicts of Interest

The authors declare no conflict of interest.

Supporting information

Supporting File: anie73155‐sup‐0001‐SuppMat.pdf.

Data Availability Statement

The data that support the findings of this study are openly available in BORIS at https://boris‐portal.unibe.ch/handle/20.500.12422/230779. All the DFT data underlying this work, including the cartesian coordinates and energies of all the modelled structures, is openly accessible via the following ioChem‐BD online dataset: doi.org/10.19061/iochem‐bd‐6‐618.