Unlocking nucleophilic reactivity in Co-MHAT catalysis: Stereospecific backside-backside displacement by anion-activated organocobalt

Abstract

The recent renaissance of olefin hydrofunctionalization based on cobalt-catalyzed hydrogen atom transfer (Co-MHAT) has primarily focused on two pathways: the homolytic cleavage of the alkyl–Co^III intermediate and its electrophilic reaction upon oxidation. In contrast, a third pathway, which leverages the two-electron nucleophilic potential of alkyl–Co^III, has remained unexplored because of the inherently low polarization of the C–Co bond. Here, we report an axial-coordination activation strategy to unlock this reactivity space in Co-MHAT catalysis. Through stoichiometric organometallic studies using well-defined alkyl–Co^III complexes, we demonstrate that an alkoxide ligand can render alkyl–Co^III sufficiently nucleophilic to displace a suitable leaving group, via a stereospecific backside-backside approach. This pathway is further applied in catalysis, enabling a diastereoselective redox neutral Co-MHAT cyclopropanation reaction.

Anion-activation unlocks reactivity and stereochemistry in MHAT.

INTRODUCTION

Metal hydride hydrogen atom transfer (MHAT) has emerged as a powerful tool for olefin hydrofunctionalization in modern organic synthesis (1). The cobalt salen-type frameworks (salen, salicylaldehyde-ethylenediamine–derived planar tetradentate ligands) are among the most widely used catalysts, partly owing to their ability to form well-defined and versatile alkyl–Co^III intermediates by MHAT (Co-MHAT; Fig. 1A) (2–5). Early investigations in this field primarily leveraged the homolytic cleavage of the labile alkyl–Co^III bond (type A) (6–18). Notably, alkyl group transfer to Ni or Cr enables reaction with an electrophile, as demonstrated by Shenvi (19–21). Over the past decade, an alternative oxidative functionalization pathway has been identified involving [alkyl–Co^IV]⁺ intermediates as carbocation surrogates (type B) (22–33). In contrast to these well-established reactivities, the two-electron nucleophilic potential of alkyl–Co^III, i.e., directly displacing an alkyl halide, remains unexplored in Co-MHAT catalysis (type C).

Fig. 1. Exploring reactivity space in CoSalen-catalyzed MHAT reactions.

(A) Three possible functionalization pathways of the key organocobalt intermediate; (B) typical catalytic cyclopropanation using homoallylic/allylic electrophiles; (C) hypothesized redox neutral Co-MHAT cyclopropanation via intramolecular type C trapping, which could be enabled by axial coordination of a basic anion; and (D) this work, a demonstration of type C reactivity in Co-MHAT reactions, from stoichiometric organometallic studies to catalysis.

In search for this reactivity, we specifically targeted cyclopropanation reactions (34–39). Cyclopropanes are prevalent in bioactive molecules and a privileged bioisostere motif in medicinal chemistry (40). A simplest approach for cyclopropanation involves cyclizations. For instance, anionic radical-polar crossover (41–46) and metal-catalyzed alkylation chemistry (37, 47, 48) allow for mild and catalytic cyclopropanation from homoallylic and allylic electrophiles, respectively (Fig. 1B). However, anion-stabilizing substitutions, such as aryl, carbonyl, and trifluoromethyl groups on the olefin moiety, are usually necessary.

In this context, we envisioned that a redox-neutral type C Co-MHAT hydroalkylation could offer an alternative, enabling cyclopropanation of unactivated alkenes (Fig. 1C). A major challenge in this proposal lies in the minimal polarization of the C–Co^III bond, which reflects the intermediate’s innate reluctance to undergo polar reactions. For instance, alkylcobaloximes resist acid cleavage unless facilitated by a β-hydroxy group (49–51). To address this, we took a strategy based on the trans influence (52, 53). We hypothesized that axial binding of a basic anion could enhance the nucleophilicity of the alkyl in the opposite position, thereby promoting the desired transformation. On the other hand, a potential complication could arise from base-mediated deprotonation of the putative cobalt hydride, which might disrupt the MHAT process (54, 55).

Here, we report the successful implementation of this design, first in stoichiometric experiments with well-defined cobalt complexes and then catalytic reactions (Fig. 1D). A Co-catalyzed redox-neutral MHAT hydroalkylation of unactivated homoallylic electrophiles is developed. Experimental evidence is found for the anion-enabled type C ring closure, more specifically, a stereospecific backside-backside coupling between the alkyl–Co^III and the alkyl electrophile. Good diastereoselectivity by 1,2-induction favoring the trans-1,2–disubstituted cyclopropanes is observed [diastereomeric ratio (d.r.) up to 17.2:1], whereas the stereoselectivity decreases for 1,3-induction. This work establishes a viable strategy to effect type C trapping in Co-MHAT catalysis, opening reactivity space with broad potential for synthetic method development.

RESULTS

Stoichiometric studies

We began our investigation by density function theory (DFT) analysis of organocobalt complex 1 bearing different axial ligands L (Fig. 2A). While natural population analysis (56) suggests that the Co^III–C bond in the parent complex is nearly nonpolar (−0.037 at Me), a buildup of negative charge becomes evident upon introduction of an increasingly basic ligand (−0.20 at Me for [1–OMe]⁻). This effect is further corroborated by the energy decomposition analysis with the combination of natural orbital for chemical valence (57, 58) analysis of complex [1–OMe]⁻, which clearly depicts the electron flow from the methoxide to the methyl via the Co d(z²) orbital, supporting our hypothesis of enhanced alkyl nucleophilicity.

Fig. 2. Stoichiometric studies.

(A) DFT calculation reveals increasing nucleophilicity at the Me in complex 1 with a more basic anionic axial ligand L. Major natural orbital for chemical valence (NOCV) pair in the energy decomposition analysis–NOCV analysis of [1-OMe]⁻ describes the interaction between 1 and OMe⁻. Charge flows from red to blue. The contour value is |Δρ| = 0.003 atomic unit (a.u.) DFT calculations computed at the B3LYP-D3(BJ)/6-311+G(d,p)//B3LYP-D3(BJ)/6-31+G(d), polarizable continuum model (PCM)(EtOH) level. (B) Synthesis of well-defined 2° alkyl–Co^III intermediates; (C) cyclic voltammetry in THF/MeOH with 0.10 M TBAPF₆ at a scan rate of 0.1 V/s. To a solution of 2a (black line) was added 1.0 equiv Cs₂CO₃ (blue dots) and subsequently neutralized by adding 2.0 equiv acetic acid (red dash); (D) left: x-ray structure of anti-2b shown with 50% thermal ellipsoids; right: 2b undergoes stereospecific displacement in the presence of methoxide; DBU, 1,8-diazabicyclo(5.4.0)undec-7-ene.

Encouraged by the computational results, we set out to experimentally validate this trend using well-defined organocobalt complexes. Treatment of [Co-1]OTs with allylbenzene and phenylsilane in MeOH at room temperature afforded complex 2a in 51% yield (Fig. 2B). Cyclic voltammetry of 2a in tetrahydrofuran (THF)/MeOH revealed a reversible redox event at E_1/2 ~ −0.25 V versus Fc⁺/Fc (Fig. 2C). Notably, addition of carbonate-induced irreversibility along with a cathodic shift, while subsequent neutralization essentially restored the original trace. Control experiments in anhydrous THF showed negligible potential changes, suggesting that the formation of a more electron-rich cobalt species requires both a carbonate base and a protic solvent (fig. S4) (59).

We then followed the same protocol and prepared complex 2b containing a γ-leaving group (Fig. 2B). 2b was isolated as a mixture of diastereomers (d.r. = 9.8:1), with the major isomer (anti-2b) unambiguously assigned by x-ray crystallography (Fig. 2D). 2b was found stable over more than a week at ambient temperature in MeOH. In contrast, addition of sodium carbonate powder triggered rapid cyclization within a few minutes, producing cyclopropane 4a (d.r. = 9.3:1) with an apparent stereochemical inversion at the nucleophilic carbon bound to Co^III. In THF, the same reaction occurred in the presence of sodium methoxide but not carbonate, implying methoxide’s pivotal role in activating the system. Further base screening in THF showed that only strong anionic bases yielded 4a, with methoxide being the most efficient. Collectively, our stoichiometric studies using well-defined organocobalt complexes suggest the possibility that a new Co^III–alkyl species could arise from methoxide coordination, presumably [2b–OMe]⁻, where the alkyl becomes a good nucleophile so that displacement of a suitable leaving group takes place in a stereospecific manner.

Catalytic reaction development

On the basis of the stoichiometric experiments, we proceeded to investigate the catalytic cyclopropanation reaction (Fig. 3). In a model reaction, homoallylic tosylate 3a was treated with a catalytic amount of [Co-1]OTs in EtOH at room temperature, in the presence of Ph₂SiH₂ and Na₂CO₃ (entry 1). The product 4a was obtained smoothly with a d.r. (12.2:1) that aligns with the results from the stoichiometric reactions. Control experiments confirmed the essential roles of the cobalt complex, silane, and base, respectively (entry 2). Performing the reaction open to air resulted in a substantially decreased yield due to competing Mukaiyama hydration (entry 3). Alternative Co complexes afforded lower yields, which could be attributable to the more dissociable C–Co^III bonds (entries 4 and 5) (60). A protic solvent was found necessary (entry 6).

Fig. 3. Catalytic reaction optimization.

0.10 mmol scale. Yield and d.r. were determined by analyzing the crude reaction mixture using gas chromatography-flame ionization detection. The major diastereomer (trans-) is shown. h, hours. r.t., room temperature.

A balanced solubility and basicity of the base was found critical (entries 7 to 9). For instance, using a less carbonate-solubilizing solvent (^tBuOH, meanwhile affording a bulkier anion) or a weaker base (NaOAc) led in reduced yields, presumably due to insufficient axial-coordination activation. On the other hand, more soluble combinations involving either MeOH or Cs₂CO₃, although proved effective in stoichiometric reactions, completely suppressed the catalytic reaction. Similarly, an organic base, 1,8-diazabicyclo(5.4.0)undec-7-ene, was unsuitable. In these cases, little conversion was observed and no base-mediated elimination was detected. Instead, a substantial amount of the neutral complex Co-1 was spectroscopically detected (fig. S12). These results could be attributable to the deprotonation of the weakly acidic Co–H species to give Co(I) at high base concentrations (54, 55), which reduces Co(III) and leads to accumulation of inactive Co(II) species. Last, we introduced MgO and molecular sieves (MSs) to scavenge the potential acid byproduct, which substantially improved the yield (entry 10).

Reaction scope

With the optimized protocol in hand, we sought to investigate the scope and limitations of this cyclopropanation reaction (Fig. 4). We initially focused on 1,2-disubstituted cyclopropane synthesis. A variety of functional groups were found well tolerated (3a-h), for example. aryl halides and thioethers. The scope of the electrophiles was then studied. 1°/2° alkyl sulfonates and 2° benzyl chlorides were successfully used (3i-j, l-n), whereas 1° alkyl acetates and chlorides failed to cyclize (3 k/k′; other failed examples are shown in fig. S1). This leaving group preference well aligns with a polar S_N2-like mechanism rather than S_H2 (18). Cyclopropanation product can also be obtained from a homopropargylic tosylate, likely resulting from in situ terminal alkyne hydrogenation under the standard conditions (3m). As Co-MHAT is sterically sensitive, terminal alkenes can react selectively in the presence of an internal alkene (3o) (61). Regarding the stereoselectivity, 1,2-induction typically leads to high d.r. favoring the trans-diastereomers (3a-j), while 1,3-induction was found rather inefficient (3 l-o). In addition, monosubstituted cyclopropanes can be synthesized from internal olefins, albeit in low yields (3p-q). We note that a major competing reaction was the olefin hydrogenation, presumably via the protonation of the activated organocobalt with axial-anion coordination, which is difficult to eliminate at this stage as a protic solvent was found necessary (14).

Fig. 4. Evaluation of substrate scope.

Standard conditions: 3 (0.40 mmol), [Co-1]OTs (0.040 mmol), Na₂CO₃ (1.20 mmol), Ph₂SiH₂ (0.80 mmol), MgO (0.40 mmol), and 4-Å MS (100 mg) in 2.0 ml of EtOH at room temperature for 24 hours. Yields refer to isolated compounds. d.r. were determined by ¹H NMR analysis of the crude reaction mixture. n.d., not detected; n.a., not applicable.

Next, we turned to the synthesis of 1,2,3-trisubstituted cyclopropanes, whose stereochemistry can offer valuable mechanistic insight. Subjecting both syn- and anti-3r diastereomers to the standard conditions consistently afforded essentially a single diastereomer in each case. Analysis of nuclear Overhauser effect spectroscopy and ¹H nuclear magnetic resonance (NMR) coupling constants reveals stereoinversion at the electrophilic carbon (C–OTs), reminiscent of a classic S_N2-type process. Smaller allylic substituents in the substrate led to attenuated 1,2-induction and thus lower diastereoselectivity regarding the terminal methyl group (3s-u). Nonetheless, complete stereoinversion at the electrophilic site was consistently observed.

Mechanistic hypothesis

On the basis of the findings above, a plausible catalytic cycle is depicted (Fig. 5A). MHAT of a Co^III–H species to the homoallylic tosylate 3 followed by cage collapse generates an alkyl–Co^III complex 2, a process that is supported by both stoichiometric synthesis and deuterium-labeling study (Fig. 5B). Subsequently, this intermediate could be activated by axial coordination of an ethoxide anion to produce [2-OEt]⁻, where the Co^III–alkyl bond becomes highly nucleophilic. As a result, [2–OEt]⁻ undergoes type C cyclization to yield the cyclopropane product 4, releasing a Co^III–OEt complex, which reacts with the silane to regenerate the Co^III–H species. We reason that the judicious combination of EtOH and weakly soluble Na₂CO₃ produces a low yet steady concentration of ethoxide. This controlled release ensures that the ethoxide has higher chance to react with the more abundant species, specifically complex 2, rather than the transient Co^III–H species to disrupt MHAT.

Fig. 5. Working hypothesis and computational studies of the key cyclization step.

(A) A plausible catalytic cycle; (B) deuterium-labeling experiment and test of enantioselectivity using a chiral CoSalen catalyst; (C) DFT-calculated free energy profile of the cyclization step. [Co] = Co-1. (D) Intrinsic bond orbital (IBO) analysis of TS-4 showing the electron flow via the favored backside-backside orbital overlap. (E) Computed barriers for the formation of four- and five-membered rings. (F) Computed barriers for the cyclization of diastereomeric organocobalt complexes. At the B3LYP-D3(BJ)/6-311+G(d,p)//B3LYP-D3(BJ)/6-31+G(d), PCM(EtOH) level.

The stereospecificity of the key intramolecular displacement step is particular noteworthy. As mentioned above, experimental evidence suggests that this step proceeds with simultaneous stereoinversion at both the nucleophilic (Co^III–C) and the electrophilic (C–OTs) carbons. To shed light on the underlying mechanism, a model reaction was computationally evaluated (Fig. 5C). Axial ethoxide coordination of the alkyl–Co^III complex substantially reduces its activation barrier toward an S_N2-type cyclization (Int-1 via TS-1/2 versus Int-2 via TS-3/4). The calculations demonstrated a pronounced preference (10.8 kcal/mol difference) for backside approaching of the electrophile relative to the Co^III–C bond (TS-4) over frontal attack (TS-3), providing quantitative support for the experimentally observed stereospecificity. As a visual illustration, the effective overlap between the backsides of the C–O σ* and Co^III–C σ orbitals in TS-4 can be clearly captured by intrinsic bond orbital analysis (Fig. 5D) (62). Although stereospecific backside displacement at Co^III–C σ bonds is uncommon in C–C coupling reactions, it finds analogy in S_E2-type transmetallation processes between alkyl–Co^III complexes and Hg^II, Ti^III, or Pd^II (63–65). The anionic organocobalt is both sterically hindered and charge-delocalized, which, as concluded by Friedman (66, 67), would disfavor the formation of larger ring systems than cyclopropanes during intramolecular displacement. DFT calculations on reaction barriers support this trend (Fig. 5E). This likely accounts for our failed attempts of synthesizing four- and five-membered rings (fig. S1).

Given that the cyclization step is stereospecific, the product’s stereochemistry can either be controlled by MHAT or a Curtin-Hammett scenario during the final cyclization, depending on whether an effective equilibrium is established epimerizing the Co^III–C stereogenic center. We computationally evaluated the cyclization from epimers Int-5 and Int-6, respectively (Fig. 5F). The difference in transition state energies is consistent with the observed diastereoselectivity (ΔΔG^‡ = 1.7 kcal favoring the trans-isomer). Shigehisa suggested a radical chain process (S_H2 at cobalt) accelerating the epimerization (68). However, we note that experimentally, the establishment of a Curtin-Hammett scenario may vary across different reactions. This corroborates with the observed low d.r. in the presence of dioxygen, which could inhibit such a radical chain (Fig. 3, entry 3). Last, we were able to obtain a low but detectable enantioselectivity (15% ee) by using a chiral cobalt complex in a preliminary attempt (Fig. 5B).

DISCUSSION

In conclusion, we disclose type C (nucleophilic) reactivity of the organocobalt intermediate in redox-neutral Co-MHAT catalysis. Key to this transformation is the anion activation strategy, which unlocks the nucleophilicity of the alkyl–Co^III complex without interfering with the MHAT process. As a demonstration, a diastereoselective cyclopropanation of homoallylic electrophiles is developed. Stoichiometric organometallic studies, catalytic reactions, and computation collectively establish that the organocobalt-mediated nucleophilic displacement proceeds via a stereospecific backside-backside trajectory. This study expands the fundamental reactivity landscape of CoSalen catalysis. We anticipate the development of exciting new transformations harnessing this nucleophilic alkyl–Co^III intermediate, either in MHAT or other catalytic scenarios.

MATERIALS AND METHODS

Materials

Reagents were purchased at the highest commercial quality and used without further purification unless otherwise stated. 1-phenyl-2-propene (CAS # 300-57-2) and diphenylsilane (CAS # 775-12-2) were purchased from Innochem. Sodium carbonate (CAS # 497-19-8, powder) was purchased from Sinopharm Chemical Reagent Company. Cesium carbonate (CAS # 534-17-8), phenylsilane (CAS # 694-53-1), tert-butyl hydroperoxide (CAS # 75-91-2), and 4-Å MS (CAS # 70955-01-0) were purchased from Energy Chemical. Magnesium oxide (CAS # 1309-48-4) was purchased from Bidepharm. Sodium acetate (CAS # 127-09-3) was purchased from J&K Chemical. Triethylsilane (CAS # 617-86-7) was purchased from Heowns. MS (4 Å) was activated at 180°C under vacuum for 12 hours before use.

General procedure for the synthesis of 4 from 3

An oven-dried 25-ml resealable Schlenk tube equipped with a Teflon-coated magnetic stir bar was charged with [Co-1]OTs (30.8 mg, 0.040 mmol, 0.10 equiv.), Na₂CO₃ (127 mg, 1.2 mmol, 3.0 equiv.), MgO (16 mg, 0.40 mmol, 1.0 equiv.), and 4-Å MS (100 mg). The reaction vessel was then briefly evacuated and backfilled with nitrogen (this sequence was repeated a total of three times). 3 (0.40 mmol, 1.0 equiv.), anhydrous ethanol (2.0 ml), and diphenylsilane (147 mg, 0.80 mmol, 2.0 equiv.) were added to the reaction vessel via syringe sequentially. The reaction mixture was stirred at room temperature for 24 hours. The mixture was filtered through a short pad of silica gel with CH₂Cl₂ as an eluent. The solvents were removed in vacuo, and the residue was purified by silica gel flash column chromatography to afford the corresponding product 4. d.r. of 4 was determined by ¹H NMR analysis of the crude reaction mixture.

Supplementary Materials

This PDF file includes:

Supplementary Text

Figs. S1 to S14

Tables S1 to S7

NMR Spectra

Cartesian Coordinates of DFT Optimized Structures

References