Stereoselective Copper-Catalyzed Olefination of Imines

Abstract

Alkenes are an important class of organic molecules found among synthetic intermediates and bioactive compounds. They are commonly synthesized through stoichiometric Wittig-type olefination of carbonyls and imines, using ylides or their equivalents. Despite the importance of Wittig-type olefination reactions, their catalytic variants remain underdeveloped. We explored the use of transition metal catalysis to form ylide equivalents from readily available starting materials. Our investigation led to a new copper-catalyzed olefination of imines with alkenyl boronate esters as coupling partners. We identified a heterobimetallic complex, obtained by hydrocupration of the alkenyl boronate esters, as the key catalytic intermediate that serves as an ylide equivalent. The high E-selectivity observed in the reaction is due to the stereoselective addition of this intermediate to an imine, followed by stereospecific anti-elimination.

Graphical Abstract

A new look at catalytic Wittig-type olefinations. We explored the use of transition metal catalysis to form ylide equivalents from readily available starting materials leading to the development of a highly E-selective copper-catalyzed olefination of imines with alkenyl boronate esters as coupling partners.

Introduction

Wittig-type olefination is a powerful synthetic tool that enables the construction of a new carbon-carbon σ-bond with precise control over the geometry of the newly formed π-bond.^[1] These features are shared by two state-of-the-art catalytic methods for alkene synthesis: olefin cross-metathesis^[2] and the hydroalkylation of alkynes.^[3] What makes Wittig-type olefination reactions particularly valuable is that they operate on carbonyls and imines, which are ubiquitous in organic chemistry. Examples of such reactions include Wittig,^{[1,a, 1,c, 4]} HWE,^[5] Peterson,^[6] and Julia^[7] olefinations, which all follow the general form shown in Scheme 1a.

Scheme 1.

Stoichiometric and metal-catalyzed Wittig-type reactions.

The key intermediate in Wittig-type olefination reactions are carbanions stabilized by an oxo/aza-philic moiety. While exact mechanisms of individual reactions vary, the nucleophilicity of these intermediates generally drives the formation of a new σ-bond in a reaction with a π-electrophile. The oxo/aza-philic group then facilitates the elimination to form the new π-bond.^{[1,d, 4,a]} For example, with phosphonium ylides the phosphonium group stabilizes the adjacent negative charge and facilitates alkene formation through elimination of the phosphine oxide.

In contrast to the state-of-the-art methods for alkene synthesis, Wittig-type reactions have been conceived as stoichiometric processes. Over time, significant effort has been devoted to the development of catalytic Wittig-type reactions, with a major focus on recovering the anion stabilizing group G (see 2).

For example, P(III)/P(V) oxidation state cycling has been leveraged in turning over phosphine oxide and accomplishing phosphine-catalyzed Wittig/HWE reactions.^[8] However, this approach does not address the key issue related to ylide synthesis: to access ylides, the anion precursors (1) are prepared from more readily available starting materials and then treated with a full equivalent of a base. As a result, the catalytic Wittig reactions centered on phosphine recycling work with a very narrow range of specialized substrates that facilitate the ylide formation. Efforts to transform other Wittig-type reactions into similar catalytic processes have not been as fruitful.^[9] Therefore, stoichiometric forms of Wittig-type reactions remain by far the most common.

Transition metal catalysis has so far played a relatively minor role in the development of catalytic Wittig-type olefination reactions.^[10] In the most common approach to metal-catalyzed Wittig reactions, phosphonium ylides have been accessed through a catalytic reaction of phosphines with diazo alkanes^[11] or alkyl halides^[12] (Scheme 1b). To be efficient, both reactions require the presence of an additional electron-withdrawing group (EWG) in the ylide precursor.^[13] Despite these limitations, these examples have suggested a different approach to developing catalytic variants of Wittig-type reactions, in which metal catalysis enables efficient and convenient access to ylides and ylide-equivalents from readily available starting materials.

Exploring the idea of transition metal catalyzed formation of ylide equivalents, we focused on copper boryl heterobimetallic complex 4. This complex can be accessed directly from alkynes or alkenyl boronate esters through copper hydride catalysis and has been identified as a key intermediate in copper-catalyzed reactions developed by our group and by others (Scheme 1c).^{[3,a, 14]} We recognized that the two reactive sites of this complex make it a functional equivalent of an ylide. As Meek, Cho and Yun groups have demonstrated, the copper boryl complex can react with π-electrophiles to form a new σ-bond.^{[14,c, 15]} Further, known bora-Wittig reactions rely on the ability of a boryl group to facilitate elimination of oxygen/nitrogen-based functional groups and promote the formation of a new π-bond.^[16]

These unique properties of heterobimetallic complex 4 allowed us to envision the catalytic olefination of π-electrophiles shown in Scheme 1d. In the proposed transformation, the heterobimetallic intermediate is obtained by the reaction of alkenyl boronate ester with copper hydride.

The addition of the intermediate to a carbonyl or an imine is followed by elimination to furnish the alkene product. Additionally, transmetalation of the addition product with a hydride donor allows catalyst turnover and the overall transformation to proceed without the need for a stoichiometric amount of a base.

Results and Discussion

Reaction Development

We initiated the development of the proposed olefination reaction by examining a variety of π-electrophiles in a reaction with alkenyl boronic esters, performed in the presence of a copper catalyst and a silane (Scheme 2). Aldehydes produced the desired alkene product in low yield, in part due to the reduction of the carbonyl outcompeting the desired reaction. We explored in situ formation of the aldehyde through the copper hydride mediated reduction of pyridoate esters. Keeping the aldehyde concentration low improved the yield, but the E:Z ratio of the alkene product remained low. In an effort to improve the stereoselectivity of the reaction, we explored imine electrophiles and found that selectivity was heavily influenced by the substituent on the imine nitrogen.^[17] Gratifyingly, under the same conditions simple aniline-derived imines provided 7 with E:Z ratio greater than 30:1.

Scheme 2.

π-Electrophiles. Reactions were performed on 0.05 mmol scale. Yields determined by GC using trimethoxybenzene as an internal standard. pin = pinacolato, dppbz = 1,2-Bis(diphenylphosphanyl)benzene, DMMS = dimethoxymethylsilane, KOPyr = potassium pyridoate, THF = tetrahydrofuran. R = Ph(CH₂)₃ and Ar = p-MeOC₆H₄.

Using the preliminary results described in Scheme 2 as a starting point, we developed the copper-catalyzed Wittig-type olefination of aryl imines, shown in Table 1. The best results were obtained with 2 mol% of a copper/dppbz catalyst, TMDSO as the hydride source, and a substoichiometric amount of potassium pyridoate (conjugate acid pKa = 17.0).^[18] While other bidentate phosphine ligands, such as Xantphos (entry 3), gave the desired alkene with good selectivity and yield, simple monodentate phosphine (entry 4) and NHC-supported (entry 2) copper complexes were not productive in the reaction.

Table 1.

Reaction development.

^[a]Reactions were performed on 0.05 mmol scale and yields were determined by GC using 1,3,5-trimethoxybenzene as an internal standard. TMDSO = 1,1,3,3-tetramethyldisiloxane, HMTSO = 1,1,3,3,5,5-hexamethyltrisiloxane, DME = dimethoxyethane. R = p-CH₃OC₆H₄O(CH₂)₃.

The identity of the substoichiometric base was very important for the success of the reaction. Replacing the potassium counter ion with sodium (entry 5) or switching to potassium tert-butoxide (entry 6) led to drastically lower selectivity and yield, with full consumption of the starting materials. Similarly, TMDSO and closely related HMTSO (entry 7) were the only silanes that gave good results, where other silanes, such as DMMS (entry 8), predominantly led to hydrosilylation of the alkenyl boronate ester. Most ethereal solvents gave good to excellent yields and selectivities (entry 9 and table S9) while aromatic hydrocarbons, such as toluene (entry 10), gave lower yields. Additionally, we found that a mixture of THF and 2-MeTHF gave the most consistent results throughout our exploration of the substrate scope. The reaction temperature also had a strong effect on the reaction outcome. At lower temperatures we observed diminished yield but increased selectivity, while at higher temperatures (entries 11 and 12) the yield was unchanged, and selectivity was lower. Finally, in the absence of the copper precatalyst, ligand, silane or base, we did not observe any of the desired alkene product (entry 13).

While alkenyl boronate esters are straightforward to synthesize and are generally stable, we wanted to explore if alkynes could be used as starting materials for the reaction in a one-pot procedure. As shown in equation 1, an alkyne (11) can be transformed to an alkenyl boronate ester, which is then used in the olefination reaction without isolation under standard reaction conditions (see Table 1).^[19] The yield and diastereoselectivity of the desired styrene are comparable to those obtained using an alkenyl boronate ester as the starting material.

(eq. 1)

Substrate Scope

After establishing reaction conditions that provided excellent yield and E-selectivity of the model substrate, we explored the scope of our olefination reaction (Scheme 3). We found that a variety of functional groups could be tolerated on the alkenyl boronate ester, including terminal alkenes (12), both aryl nucleophiles and electrophiles (13 and 14), a protected allylic alcohol (15) and a Boc-protected amine (17), as well as nitrogen-containing heterocycles (16 and 21). The reaction was also amenable to simple vinyl boronate esters (20) and β-styrenyl boronate esters (19), although with slightly diminished yield, likely due to the altered electronic properties of the starting π-bond.

Scheme 3.

Substrate scope. Yields of isolated products are reported. E:Z ratios were determined by GC. Reactions performed on 0.3 mmol scale with CuOAc (2.0 mol%), dppbz (3.0 mol%), TMDSO (2.0 equiv), KOPyr (0.2 equiv), and imine (1.5 equiv) at 45 °C for 24 h in a 1:1 mixture of THF and 2-MeTHF [0.1 M] with respect to alkenyl Bpin (1.0 equiv), unless otherwise stated. [a] Reaction was performed with CuOAc (2.0 mol%), (p-CF₃)dppbz (3.0 mol%), TMDSO (2.0 equiv), KOPyr (0.4 equiv), and imine (2.0 equiv) at 45 °C for 24 h in a 1:1 mixture of THF and 2-MeTHF [0.1 M] with respect to alkenyl Bpin (1.0 equiv). [b] Reactions performed on 0.5 mmol scale with CuOAc (2.0 mol%), dppbz (3.0 mol%), HMTSO (1.5 equiv), KOPyr (0.2 equiv), and imine (2.0 equiv) at 60 °C for 48 h in 2-MeTHF [0.2 M] with respect to alkenyl Bpin (1.0 equiv). [c] Reaction performed on 0.3 mmol scale. [d] SciOPP (1,2-Bis[bis[3,5-di(t-butyl)phenyl]phosphino]benzene) used instead of dppbz. R = p-MeOC₆H₄O(CH₂)₃. Ar = p-MeOC₆H₄.

We next turned our attention to the aryl imine scope and found that our reaction tolerates a variety of ortho, meta and para substituents (26, 22, 23, and 28). In addition, several nitrogen-containing heterocycles are tolerated (24 and 27). Further, conjugated imines provide 1,3-dienes in good yield (29). We found that under our standard reaction conditions, electron deficient aryl imines were readily reduced and did not produce the desired styrene products. However, using imines derived from more electron-rich anilines and a less electron-donating ligand, we were able to realize the transformation of m-cyano and p-trifluoromethyl aryl imines into the corresponding styrenes (25 and 31). We believe that the successful formation of these two products can be attributed to the decreased reduction of the starting imines and slower copper-catalyzed isomerization of the styrene π-bond under the modified reaction conditions.

Further, we found that our method could be applied to the synthesis of alkenes with quaternary carbons at the allylic position. These highly sterically encumbered alkenes are not only a challenge to synthesize using catalytic methods but are also formed in low yields using stoichiometric Wittig-type reactions.^[20] By changing the hydride source and increasing the temperature to 60 °C we were able to produce alkenes showcasing a variety of substitution patterns and excellent selectivity (>200:1 selectivity). A simple tert-butyl group (32), as well as various cyclic and fused ring systems were tolerated (35 and 36). A cyclopropyl moiety (37), a simple methyl ester (39) and a tetrahydrofuran (34) were all tolerated under the reaction conditions. Curiously, when 74 and 71 were submitted to the reaction conditions, we did not observe the expected α-phenyl ether and α-methoxy ester alkenes, but instead we isolated the products arising from rearrangement prior to elimination (see SI for more details).

We also identified some general limitations of the reaction (see Scheme 3, bottom). More highly substituted alkenyl boronate esters featuring trisubstituted alkenes did not provide significant amount of the desired products, and instead led to the reduction of an imine. The presence of good leaving groups, such as tosylates, also prevented the formation of the desired alkene. Alkyl imines containing α-hydrogen atoms gave less than 10% yield of the desired aliphatic alkene product under a variety of reaction conditions. Similarly, imines derived from ketones or alkynyl aldehydes gave a complex mixture of products under the standard reaction conditions, as did ethyl N-(phenyl)formimidate.

Reaction Mechanism

To better understand the mechanism of this reaction we performed several mechanistic experiments. Our first goal was to confirm the proposed role of a copper containing heterobimetallic complex as the intermediate in the reaction. Although phosphine-ligated 1,1-boryl-copper heterobimetallic complexes have been proposed as catalytic intermediates before, they have only been characterized via ³¹P-NMR analysis.^[14,e] To gain a stronger and more specific confirmation for the involvement of the phosphine ligated heterobimetallic intermediate, we used a ¹³C-enriched alkenyl boronate ester (41) as a substrate (Scheme 4). Submitting 41 to a solution of dppbzCuH gave rise to a new broad signal in the ¹³C-NMR spectrum at δ = 17.8 ppm. The addition of imine 9 to this solution produced styrene 43, along with disappearance of the broad signals corresponding to both the starting alkenyl boronate ester and the proposed intermediate (42).

Scheme 4.

Heterobimetallic intermediate. Reactions were performed on 0.05 mmol scale. Ar = p-MeOC₆H₄.

While we were unable to isolate the dppbz-ligated intermediate due to facile decomposition, a related known heterobimetallic species bearing an NHC ligand (44) was isolated and showed a broad signal in ¹³C-NMR spectrum at δ = 16.4 ppm corresponding to the α-carbon, further validating our assignment of the phosphine ligated α-boryl alkyl copper species.^{[14,a, 14,d, 14,e, 14,g]}

Next, we monitored the reaction of alkenyl Bpin 6 and aryl imine 9 by ¹H-NMR spectroscopy (Scheme 5a). Within 30 minutes we observed consumption of starting material accompanied by the formation of two diastereoisomers of 1,2-boryl amine addition product 45. As the reaction progressed, one isomer appeared to gradually give rise to the product styrene and was fully consumed after 24 hours, while the other isomer accumulated to a constant concentration. Through careful isolation, conversion to the corresponding oxazolidinone (46), and NOESY and J-coupling analysis (Scheme 5b), we determined that the accumulated adduct was the threo diastereoisomer. The threo isomer of 45 appears not to eliminate under the reaction conditions, suggesting that the product is formed solely via the erythro isomer. In the reaction with alkyl imines, we observed the formation of a single 1,2-boryl amine diastereoisomer that was fully consumed after 48 hours (Scheme 5c). Through similar analysis using ¹H-NMR spectroscopy we identified this isomer as the erythro-diastereoisomer.

Scheme 5.

Reaction intermediates and stereochemistry of olefination. Reactions performed on 0.05 mmol scale. Yields determined by ¹H-NMR spectroscopy with 1,3,5-trimethoxybenzene as an internal standard and E:Z ratios determined via GC analysis of crude reaction mixtures. [a] See Table 1. R = Ph(CH₂)₃ and Ar = p-MeOC₆H₄.

Together, these findings indicate that the erythro diastereoisomer of the addition intermediates gives rise to both the E-styrenes and E-aliphatic alkenes. Therefore, in both cases, the alkene formation involves anti-elimination,^[16,a] with no indication that the syn-elimination from the erythro isomer occurs. Furthermore, the threo isomer, when formed, does not undergo elimination under the reaction conditions (Scheme 5d). We can speculate about explanations for the observations we made. First, it is likely that the increased bulk of the tert-butyl group prevents the formation of the syn-diastereoisomer in the initial addition to the alkyl imines. Second, the erythro isomer of the addition product, in both reactions of alkyl and aryl imines, encounters the least steric hinderance undergoing anti-elimination, while the aryl threo isomer must overcome increased steric hinderance in either syn- or anti-elimination.

Interestingly, the understanding of the stereochemistry of the addition and the elimination processes provided unexpectedly little insight into the origin of the Z-alkene products and therefore the E:Z selectivity. The first clue about the origins of the Z-isomer in the reactions of aryl imines came from the observation that the E:Z ratio diminishes during the reaction, suggesting isomerization of the styrene products.

Previously, copper hydride has been implicated in the isomerization of styrenes via hydrocupration and subsequent β-hydride elimination.^[21] As expected, when we submitted an E-styrene 7 to the standard reaction conditions, we observed an erosion in the E:Z ratio from >100:1 to 43:1 (Scheme 5e). We also found that copper hydride mediated isomerization was not as facile in the presence of starting materials.

Overall, we conclude that E-styrene is formed with high selectivity in the elimination reaction and is then slowly isomerized in the process mediated by the copper catalyst. This conclusion is consistent with the observed dependence of E:Z selectivity on the electronics of the product styrene. We also noticed that for high selectivity, styrenes bearing electron-withdrawing groups required a more electron deficient copper catalyst, presumably to slow the reversible hydrometallation and isomerization.

In analogous experiments with aliphatic alkene 32 we found no evidence for its isomerization under the reaction conditions (see SI). This observation suggests that aliphatic alkene products do not undergo hydrocupration under the reaction conditions and is consistent with high E-selectivity observed in reactions of aliphatic imines.

We found that the starting alkenyl boronate esters also undergo isomerization under the reaction conditions (Scheme 5f). When Z-enriched alkenyl boronate ester 6 was allowed to react under the standard conditions, we observed the erosion of the diastereomeric ratio from (1:172) to (1:6) before full consumption and formation of the product alkene. This result shows that the initial hydrocupration event is likely reversible.

On the basis of these experiments, we propose that the reaction proceeds according to the mechanism outline in Scheme 6. The first step involves the reversible hydrocupration of an alkenyl boronate ester to form the key hetero-bimetallic intermediate.^{[14,a, 22]} Subsequently, this intermediate reacts with the imine electrophile, followed by regeneration of copper hydride via an equivalent of silane.^[23] The resulting 1,2-boryl N-silylamine eliminates, possibly aided by the Lewis-basic potassium pyridoate, to furnish the E-alkene product, which when R¹ is aryl can isomerize to give a mixture of E/Z styrenes.

Scheme 6.

Proposed mechanism.

Conclusion

In summary, we have developed a new method for the catalytic olefination of imines leveraging a catalytically generated copper-boryl heterobimetallic intermediate as an ylide equivalent. The new reaction is compatible with aryl and sterically hindered alkyl imines and can be performed in the presence of an array of functional groups. Mechanistic studies corroborate the proposed catalytic cycle, confirming the existence and function of the proposed heterobimetallic intermediate, and provide insights into the origins of selectivity of the newly formed π-bond.