Catalytic Enantioselective α-Alkylation of Carbonyl Compounds by Unactivated Alkyl Electrophiles

Abstract

Carbonyl groups that bear an α stereocenter are commonly found in bioactive compounds, and intense effort has therefore been dedicated to the pursuit of stereoselective methods for constructing this motif. While the chiral auxiliary-enabled coupling of enolates with alkyl electrophiles represented groundbreaking progress in addressing this challenge, the next advance in the evolution of this enolate-alkylation approach would be to use a chiral catalyst to control stereochemistry. Herein we describe the achievement of this objective, demonstrating that a nickel catalyst can accomplish enantioselective intermolecular alkylations of racemic Reformatsky reagents with unactivated electrophiles; the resulting α-alkylated carbonyl compounds can be converted in one additional step into a diverse array of ubiquitous families of chiral molecules. Applying a broad spectrum of mechanistic tools, we have gained insight into key intermediates (including the alkylnickel(II) resting state) and elementary steps of the catalytic cycle.

Graphical Abstract

INTRODUCTION

Because carbonyl groups that bear an α stereocenter are a common motif in biologically active molecules,¹ the development of efficient methods for generating this subunit has been a long-standing objective in organic synthesis (Figure 1A).^2,3,4 One straightforward approach to accessing such structures is through the α-alkylation of enolates, a process that accounted for 11% of all carbon–carbon bond-forming reactions run in bulk in a Pfizer research facility between 1985 and 2002.⁵ The development of methods by Evans, Myers, and others that achieve the stereoselective α-alkylation of carbonyl derivatives through the use of a covalently bound chiral auxiliary represents one of the landmark advances in the field of asymmetric synthesis (Figure 1B);^6,7,8 such methods have been widely applied throughout academia and industry,⁹ including syntheses on a metric-ton scale.¹⁰

Figure 1.

Catalytic enantioselective intermolecular α-alkylation of carbonyl compounds with unactivated alkyl electrophiles. (A) A long-standing challenge in organic synthesis: Stereoselective α-alkylation of carbonyl compounds with unactivated alkyl electrophiles. (B) Landmark advance: Diastereoselective α-alkylation through the use of a stoichiometric chiral auxiliary. (C) This study: Enantioselective α-alkylation through the use of a chiral catalyst. Ar = 3,5-di-t-butylphenyl; NaDA = sodium diisopropylamide; R = carbon substituent; X = leaving group (e.g., halide); Y = H, carbon, nitrogen, or oxygen.

Controlling the stereochemistry of the α carbon with a chiral catalyst, rather than a stoichiometric chiral auxiliary, is a logical next step in the development of this strategic carbon–carbon bond-forming process. However, to our knowledge, asymmetric catalysis of the fundamental transformation illustrated in Figure 1A has not yet been described, although some progress has been reported for certain specialized substrates.^{11,12,13,14–15}**consolidate/global**) For example, in the case of generic/unactivated electrophiles (versus a specialized/activated electrophile, such as an allyl electrophile¹⁵), only phase-transfer catalysis has provided good enantioselectivity, but the carbonyl coupling partner must include an activating, generally an α-imino, substituent.^11,14 In the case of specialized electrophiles, some progress has been described with phase-transfer reagents, transition metals, Brønsted acids, and amines.^11–15

Catalyzing the α-alkylation of alkali-metal (e.g., lithium) enolates is challenging, due to their high nucleophilicity and Brønsted basicity, which can lead to undesired processes such as direct, uncatalyzed α-alkylation (to furnish racemic product) and acid–base reactions (including loss of HX from the electrophile and enolization/racemization of the product). Due to their attenuated nucleophilicity and Brønsted basicity, readily available Reformatsky reagents (broadly defined as α-zincated carbonyl compounds) are an attractive alternative to alkali-metal enolates for use in catalyzed reactions.¹⁶

Although metal-catalyzed couplings of Reformatsky reagents with alkyl electrophiles have not been described,¹⁷ progress has been reported in the development of metal-catalyzed couplings of other alkyl nucleophiles with alkyl electrophiles, including asymmetric processes.^18,19 To achieve the goal of catalytic enantioselective α-alkylation of carbonyl compounds with unactivated alkyl electrophiles, we sought to develop a method that would solve the two key challenges: catalysis of a new carbon–carbon bond-forming process (the coupling of a Reformatsky reagent with an unactivated alkyl electrophile) and effective control of stereochemistry.

Herein we describe the realization of our objective, establishing that a chiral catalyst based on nickel, an earth-abundant metal, can achieve this strategic carbon–carbon bond-forming process with good yield and ee (Figure 1C). The α-alkylated products of the coupling can be transformed in one step, without racemization, into a broad spectrum of useful classes of chiral molecules. Our mechanistic studies provide significant insight into the reaction pathway; of particular note is our structural characterization of the predominant resting state of nickel during catalysis, specifically, an alkylnickel(II) complex in which nickel is bound to the stereogenic α-carbon of the carbonyl group, the stereochemistry of which corresponds to that observed in the major enantiomer of the coupling product.

RESULTS AND DISCUSSION

Synthesis.

In our initial studies, we employed as the nucleophile a racemic Reformatsky reagent generated through the reaction of an α-bromoamide with zinc metal, and we determined that a chiral nickel catalyst can provide good yield and enantioselectivity in an α-alkylation with an unactivated alkyl iodide (Figure 2A, Method A) (for an overview of the impact of changes in various reaction parameters on the yield and the enantioselectivity, see the Supporting Information). We subsequently established that the corresponding Reformatsky reagent generated via deprotonation of the amide,²⁰ followed by in situ nickel-catalyzed asymmetric α-alkylation, affords similar yield and ee in a one-pot process (Figure 2A, Method B). No detectable racemization of the potentially labile α stereocenter of the product is observed with either method.

Figure 2.

Nickel-catalyzed asymmetric α-alkylation of carbonyl compounds, including transformations of the enantioenriched product. All data represent the average of two experiments, and the percent yield represents purified product. (A) Scope (0.6-mmol scale, unless otherwise noted). ^a Reaction performed with 10 mol% NiBr₂·glyme and 13 mol% L*. ^b Reaction performed with 3.0 mol% NiBr₂·glyme and 3.8 mol% L*. (B) Transformation to other useful families of enantioenriched compounds (all proceed with essentially no racemization (≤1%)).

A variety of Reformatsky reagents and unactivated alkyl electrophiles serve as suitable coupling partners in these catalytic asymmetric intermolecular α-alkylations, generally leading to carbon–carbon bond formation with similar yield and enantioselectivity for the two methods (Figure 2A). With respect to the nucleophile, the α-alkyl substituent may vary in size from methyl to isopropyl, and various functional groups may be present (products 1–9; for additional information on functional-group compatibility, see the Supporting Information). Although azetidines have been employed as effective stoichiometric chiral auxiliaries in diastereoselective α-alkylations of enolates,²¹ the stereochemistry of the nickel catalyst, not that of the azetidine, predominantly controls the stereochemistry of the product in the case of an azetidine that bears a stereocenter (products 8 and 9). The high yield and the high enantioselectivity for couplings to form products such as 2, taken together, establish that the chiral nickel catalyst is achieving a stereoconvergent reaction that utilizes both enantiomers of the racemic nucleophile, not merely carrying out a simple kinetic resolution.

With respect to the electrophile, an array of unactivated alkyl iodides are suitable coupling partners in these nickel-catalyzed enantioconvergent α-alkylations. Primary alkyl electrophiles that vary in steric demand, including a β-branched substrate, engage in carbon–carbon bond formation with good enantioselectivity (products 10–12 in Figure 2A). A variety of functional groups can be present in the electrophile, including a silyl ether, aryl ether, primary alkyl fluoride, primary alkyl chloride, trifluoromethyl group, ester, carbamate, alkylboronate ester, or acetal (products 13–23). On a gram scale in the presence of 3.0 mol% nickel, catalytic enantioselective α-alkylation proceeds with similar yield and ee as for a reaction conducted on a 0.6-mmol scale (product 10).

The enantioenriched N-acylazetidines that are generated in these nickel-catalyzed asymmetric α-alkylations are particularly attractive due to their ready transformation into an array of useful families of compounds (Figure 2B). In addition to the previously described conversion of N-acylazetidines to ketones,²² the enantioenriched N-acylazetidine can be converted into other carbonyl compounds (an aldehyde and a carboxylic acid), and it can be reduced to an alcohol or an amine. All of these transformations proceed with essentially no racemization (≤1%) of the potentially labile α stereocenter.

Mechanism.

Only two types of racemic alkyl nucleophiles, both somewhat specialized in comparison with Reformatsky reagents, have previously been shown to serve as useful partners in enantioconvergent couplings with unactivated alkyl electrophiles;^23,24 there have been no detailed experiment-based mechanistic studies of such coupling processes. This deficiency provided a strong impetus for us to investigate the mechanism of our new method for the catalytic asymmetric α-alkylation of carbonyl compounds, with a particular interest in the structure and reactivity of the nucleophile (the Reformatsky reagent), as well as any alkylnickel intermediate derived from the nucleophile.

Our mechanistic studies began with an investigation of the Reformatsky reagent. A number of crystal structures of zincated carbonyl derivatives have been reported, including compounds that are oxygen-bound only (e.g., zinc enolates of ketones^25,26) as well as carbon-bound (~tetrahedral carbon; amide derivatives^26,27). Although we originally hypothesized that our nucleophile, derived from an amide, likely involved a carbon-bound zinc, we also considered the possibility that the ring strain of the azetidine might impede electron donation by the nitrogen lone pair to the carbonyl group,²² leading to an oxygen-bound zinc enolate, as observed for ketone derivatives. To resolve this question, we pursued the structural characterization of a representative nucleophile.

X-ray crystallographic analysis of Zn-2 revealed a dimeric carbonyl-bridged C-metalated structure in the solid state (Figure 3A), consistent with prior studies of Reformatsky reagents derived from amides.^26,27 The dimeric Reformatsky reagents are heterochiral, and the C2–C3–C4, C2–C3–Zn1, and C4–C3–Zn1 bond angles of the stereogenic carbon range from 106–115°.

Figure 3.

Mechanistic studies. (A) Crystallographic characterization of racemic organozinc nucleophile Zn-2 (thermal ellipsoids at 50% probability; hydrogen atoms have been omitted for clarity). (B) UV-vis spectroscopy: Ni_C is the resting state of the catalyst during a coupling reaction in progress (THF, r.t.). (C) Synthesis of nickel(II) complexes, including the resting state during catalysis (Ni_C) (thermal ellipsoids at 50% probability; most hydrogen atoms have been omitted for clarity).

Recognizing that the structure of the nucleophile in solution is more relevant to catalysis than its structure in the solid state, we investigated the Reformatsky reagent via NMR spectroscopy, and we determined that the α-CH group of Zn-2 appears at δ 1.79 in the ¹H NMR spectrum and at δ 37.6 in the ¹³C NMR spectrum (THF-d₈, r.t.), which are consistent with a C-metalated structure and inconsistent with an O-bound zinc enolate. Furthermore, Zn-2 exhibits an absorption in its infrared spectrum at 1597 cm⁻¹, which we assign to the C=O stretching of a carbonyl group.

Understanding the dynamics of the Reformatsky reagent, specifically, the rate of racemization of the α stereocenter, could help to frame our consideration of the origin of stereoselectivity in our nickel-catalyzed enantioconvergent alkylations; to the best of our knowledge, the barrier for this interconversion has not previously been measured for Reformatsky reagents. We have determined that the diastereotopic methylene protons in the β position of Reformatsky reagent Zn-2 can be distinguished in THF-d₈ at room temperature, indicating that the α stereocenter is configurationally stable on the NMR timescale under these conditions, corresponding to a barrier for interconversion of at least 14 kcal/mol.

With useful information in hand about the Reformatsky reagent, we sought to identify the nickel-containing species that are present during catalysis. When monitoring through electron paramagnetic resonance (EPR) spectroscopy the coupling that provides product 2 (Figure 2A), we observe no signal, consistent with the absence of detectable amounts of nickel(I) and nickel(III) species during the reaction. On the other hand, the UV–vis spectrum of this coupling reveals an absorbance with λ_max = 449 nm (red trace on the left side of Figure 3B). On the basis of our previous mechanistic studies of other nickel-catalyzed enantioconvergent couplings,^28,29 we speculated that a nickel(II) complex might be responsible for this absorbance.

Treatment of NiBr₂·glyme with bisoxazoline ligand L* in THF at room temperature provides a 94% yield of NiBr₂L* (Ni_A), which exhibits a UV–vis spectrum that is distinct from a catalyzed coupling in progress (red versus black traces on the left side of Figure 3B). Because bisoxazoline L* bears a potentially labile methylene proton between the two oxazolines, we sought to determine whether the ligand may be deprotonated under the reaction conditions. Treatment of Ni_A with LiHMDS, a strong, non-nucleophilic Brønsted base, generates a 98% yield of Ni_B (Figure 3C; isolated as a LiBr/diglyme adduct, Ni_B^LiBr; for the crystallographic characterization of Ni_B^LiBr, see the Supporting Information), which also displays a UV–vis spectrum different from a catalyzed coupling (red versus blue traces on the left side of Figure 3B) but essentially identical to the spectrum produced upon treatment of Ni_A with one equivalent of Reformatsky reagent Zn-2 (blue versus green traces on the left side of Figure 3B), indicating that Ni_B may be formed via deprotonation of Ni_A by Zn-2 at the outset of catalysis (Figure 3C).³⁰

Treatment of Ni_B with an additional equivalent of Reformatsky reagent Zn-2 leads to a UV–vis spectrum that is very similar to that of a coupling in progress (red versus black traces on the right side of Figure 3B). Hypothesizing that an organonickel(II) complex might be responsible for these UV–vis spectra, we independently synthesized and characterized two such complexes through the reaction of Ni_B^LiBr with the lithium enolates of two amides (Ni_C^Et and Ni_C^OMe; Figure 3C). The UV–vis spectra of both of these alkylnickel(II) complexes are very similar to the spectrum of a catalyzed coupling (red, blue, and green traces on the right side of Figure 3B), consistent with such a complex being the resting state of the catalyst during a reaction.

Next, we determined the structure of Ni_C^OMe via X-ray crystallography (Figure 3C). The metalated amide is bound to an approximately square-planar nickel through the α carbon ((S) stereochemistry with (S,S)-L*) and the carbonyl group (C36–O3: 1.291(2) Å and C36–C37: 1.451(2) Å, which are similar to the corresponding bonds in Zn-2: C4–O1: 1.275(2) Å and C3–C4: 1.459(2) Å). The configuration of the coupling product generated by (S,S)-L* corresponds to alkylation of the carbon–nickel bond of Ni_C^OMe with retention of stereochemistry.

Analysis of Ni_C^Et via ¹H NMR spectroscopy reveals the presence of both epimers at the α carbon in THF-d₈ at room temperature (4:1 ratio). NOESY experiments establish that the major diastereomer has (S) stereochemistry at the α carbon, which is presumably favored because the α-H, rather than the α-Et, is proximal to the bulky aryl substituent of the chiral ligand (see C37 in the crystal structure of Ni_C^OMe in Figure 3c). We have determined that the diastereomers interconvert with ΔG^‡(S→R) = 19.5 kcal·mol⁻¹ and ΔG^‡(R→S) = 18.5 kcal·mol⁻¹ (Supporting Information).

Building on our previous mechanistic studies,^28,29 we provide in Figure 4A a possible pathway for these nickel-catalyzed enantioconvergent α-alkylations. At the outset, NiBr₂·glyme, L*, and the Reformatsky reagent react to generate Ni_C, which is the resting state of nickel during a catalyzed coupling (Figure 3B and 3C) and is formed within five minutes of mixing the reaction components (Supporting Information). A nickel(I) metalloradical, Ni_E (vide infra), abstracts a halogen atom from the electrophile (R–X) to afford an alkyl radical (R•) and Ni_B (which reacts with the Reformatsky reagent to produce Ni_C). Alkyl radical R• couples with the resting state of nickel, Ni_C, to furnish a dialkylnickel(III) complex, Ni_D, which reductively eliminates to provide the desired enantioenriched α-alkylation product and to regenerate nickel(I) complex Ni_E.

Figure 4.

Mechanistic studies (cont.). (A) Outline of a possible mechanism. For the sake of simplicity, all steps are drawn as irreversible, and the full coordination sphere around nickel is not always provided. (B) Ni_C as a catalyst. (C) Beneficial role of Ni_E^cod. (D) Support for the intermediacy of an organic radical, R•.

Consistent with the proposed mechanism (Figure 4A), Ni_C^Et and Ni_B^LiBr serve as suitable catalysts for enantioconvergent α-alkylation, affording product 2 (Figure 2A) with a similar rate and enantioselectivity as NiBr₂·glyme/L* under the standard reaction conditions (Supporting Information). To gain insight into the reactivity profile of postulated intermediate Ni_C, we carried out a crossover experiment wherein we employed Ni_C^Et, which bears an amide with an α-ethyl substituent, as a catalyst in the cross-coupling of an α-methyl-substituted Reformatsky reagent (Figure 4B). Our observation that most of the α-ethyl-substituted product is generated at the outset of the reaction suggests that exchange of the organic groups between Ni_C^Et and the Reformatsky reagent is not occurring extremely rapidly relative to the rate of carbon–carbon bond formation from Ni_C^Et.

We have also investigated stoichiometric couplings of Ni_C^Et with an alkyl electrophile. A modest yield and good enantioselectivity (27% yield, 91% ee) are observed upon treating Ni_C^Et with a primary alkyl iodide (Figure 4C). As outlined in Figure 4A, we hypothesize that a nickel(I) metalloradical, Ni_E, serves as a chain-carrying radical in the catalytic cycle; under our standard coupling conditions, such species may be generated, for example, by a comproportionation reaction or by bond homolysis from a nickel(II) complex. To provide support for the beneficial impact of a nickel(I) complex on carbon–carbon bond formation, we independently synthesized nickel(I)/L* derivative Ni_E^cod (for the crystal structure, see the Supporting Information), and we determined that the presence of a small amount of this complex (0.1 equiv) does indeed enhance the yield of the coupling reaction (50% yield, 91% ee; Figure 4C).

As our proposed catalytic cycle suggests that the R group of the alkyl electrophile (R–X) binds to nickel via a radical pathway (Figure 4A), we sought evidence that R• is formed under our coupling conditions. Using 6-iodo-1-heptene as a mechanistic probe, we observe coupling products wherein the electrophile has cyclized to generate a 3.4:1 mixture of cis and trans isomers of cyclopentanes, the same ratio of isomers that has been reported for the cyclization of radical I (Figure 4D).³¹ Furthermore, since I cyclizes with a rate constant of ~1 × 10⁵ s⁻¹ at 25 °C,³¹ much slower than the rate of diffusion (generally ~10⁸–10⁹ s⁻¹ at 25 °C),³² our observation of cyclized product is consistent with out-of-cage coupling of R•, as required by the mechanism illustrated in Figure 4A. Taken together, our various studies are fully congruent with the catalytic enantioselective radical-based pathway outlined in Figure 4A, which complements classic approaches to stereoselective α-alkylation that have relied on polar reactions and stoichiometric chiral auxiliaries.^6–8

CONCLUSION

The catalytic enantioselective α-alkylation of carbonyl compounds with unactivated alkyl electrophiles is a classic, long-standing challenge in asymmetric synthesis. A recent review of transition-metal-catalyzed reactions of enolates concluded that a “methodology for building stereogenic centers with non-functionalized sp³-hybridized electrophiles…is…missing at this point”.¹² In this study, we address this deficiency, demonstrating that a chiral nickel catalyst can achieve enantioconvergent couplings of racemic Reformatsky reagents with unactivated alkyl electrophiles; the method displays good functional-group tolerance, and the products of the coupling can be transformed without racemization into a wide range of other families of useful enantioenriched compounds. Exploiting an array of mechanistic tools, we have gained insight into key intermediates and elementary steps of this enantioconvergent α-alkylation, including crystallographic characterization of an alkylnickel(II) complex that contains a nickel-bound stereogenic carbon and serves as the resting state of the catalyst under the reaction conditions. This work demonstrates the ability of nickel catalysis to address a long-sought objective in asymmetric synthesis, via a novel mechanism. Additional efforts to exploit earth-abundant metals in enantioselective catalysis are underway.