|Home | About | Journals | Submit | Contact Us | Français|
Inspired by advances in our understanding of biological processes, new reactions employing organic molecules as catalysts have grown significantly over the last two decades.1 In the broadest of terms, the most successful of these catalysts can be classified either as Brønsted acids,2,3 hydrogen-bond donors,4–7 or Lewis bases.8–11 Each of these catalytic manifolds activates substrates in biological settings and provides an inspiring blueprint to create smaller, synthetic versions of these impressive biocatalysts. Lewis base catalysis is presently an exciting area of research and encompasses a wide variety of strategies to initiate both established and new chemical processes.12,13 An elegant and key biological transformation utilizes the cofactor thiamine, a coenzyme of vitamin B1, to transform α-keto acids into acetyl CoA, a major building block for polyketide synthesis. In this process, a normally electron-deficient molecule (e.g., pyruvic acid) is converted into an intermediate that possesses electron density on the carbon atom that was initially part of the carbonyl system. These carbonyl or acyl anions are unusual since they have “umpolung” (reversed polarity) when compared to the initial keto acids. In 1954, Mizuhara and Handler proposed that the active catalytic species of thiamine-dependent enzymatic reactions is a highly unusual divalent carbon-containing species,14 later on referred to as an N-heterocyclic carbene (NHC). An alternative description of the active thiamine cofactor employs the term zwitterion, which can be viewed as a resonance form of the carbene description. This unique cofactor accomplishes fascinating Lewis base catalyzed transformations by utilizing the lone pair of electrons at C-2.
In the early 1960s, Wanzlick and co-workers realized that the stability of carbenes could be dramatically enhanced by the presence of amino substituents, and they attempted to prepare a carbene center at C-2 of the imidazole ring.15,16 However, only the dimeric electron-rich olefin was isolated. Later on, Wanzlick's group demonstrated that potassium tert-butoxide can deprotonate imidazolium salts to afford imidazol-2-ylidenes, which can be trapped with phenyl isothiocyanate and mercury salts.17–19 However, Wanzlick's group never reported the isolation of the free carbene. Following these results, Arduengo et al. isolated in 1991 a stable crystalline N-heterocyclic carbene by the deprotonation of 1,3-di(1-adamantyl)imidazolium chloride with sodium or potassium hydride in the presence of a catalytic amount of either potassium tert-butoxide or dimethyl sulfoxide.20 The structure was unequivocally established by single-crystal X-ray analysis, and the carbene was found to be thermally stable, which stimulated extensive research in this field.
The nature of the stabilization is ascribed to the steric and electronic effects of the substituents: The two adamantyl substituents hinder reactions of the carbene center with external reagents as well as prevent dimerization. In 1992, Arduengo et al. expanded this carbene class by successfully isolating the carbene from 1,3,4,5-tetramethylimidazolium chloride by treating the latter with sodium hydride and catalytic amounts of potassium tert-butoxide in tetrahydrofuran.21 The successful isolation of carbenes with less bulky substituents demonstrates that electronic factors may have greater impact on the stability of the carbene than steric ones. Such electronic factors operate in both the π and σ frameworks, resulting in a “push-pull” synergistic effect to stabilize the carbene. π donation into the carbene from the out-of-the-plane π orbital of the heteroatoms adjacent to C-2 stabilizes the typical electrophilic reactivity of carbenes. The electronegative heteroatoms adjacent to C-2 provide additional stability through the framework of σ bonds, resulting in a moderation of the nucleophilic reactivity of the carbene (Figure 1).22 The combination of these two effects serves to increase the singlet–triplet gap and stabilize the singlet-state carbene over the more reactive triplet-state one.
The electronic properties of NHCs are a key determinant of the unique reactivity of these catalysts. Lewis bases are normally considered as single electron-pair donors. However, the singlet carbenes of NHCs are distinct Lewis bases that have both σ basicity and π acidity characteristics. These attributes allow for the generation of a second nucleophile in the flask. Nucleophilic addition of the carbene to an aldehyde results in the formation of a new nucleophile. The “doubly” nucleophilic aspect is unique to the carbenes. The combination of these characteristics allows NHCs to react as powerful nucleophiles, which has driven the development of a distinct class of catalytic processes during the last decade. This review highlights our work in this young and promising field.
The earliest application of these unique Lewis bases was the development of the benzoin reaction.23 This umpolung process
utilizes a thiamine-related NHC as a nucleophilic catalyst. In 1943, Ugai and co-workers reported that thiazolium salts can catalyze the self-condensation of benzaldehyde to produce benzoin.23 This process is clearly related to the earliest reports of laboratory organocatalysis from Wöhler and Liebig in 1832 detailing the cyanide-catalyzed benzoin reaction.24 Based on Ugai's report, Breslow proposed the mechanism in which the active catalytic species is a nucleophilic carbene derived from a thiazolium salt to generate the carbanion known as the Breslow intermediate.1b In 1966, Sheehan and Hunneman reported the first investigations into an asymmetric variant of the benzoin condensation employing a chiral thiazolium salt as precatalyst.25a Most recently, Enders and Kallfass accomplished the first high-yield and highly enantioselective intermolecular benzoin condensation (Scheme 1).25b This seminal work by Enders on carbene catalysis using triazolium salts focused the interest of the community on these unique structures and moved interest away from thiazolium catalysts.
In the 1970s, Stetter demonstrated that catalysis with thiazolium species can be employed to accomplish the addition of acyl anions to 1,4-conjugate acceptors.26 This transformation is a useful carbon–carbon-bond-forming strategy that has attracted the attention of researchers interested in producing 1,4-dicarbonyl species. Like the benzoin reaction, the addition of an NHC to an aldehyde generates the acyl anion equivalent, and this initial step is typically facile due to the highly electrophilic nature of the aldehyde. However, this electrophilicity is also detrimental to processes other than dimerization in that multiple side-products are formed, because the aldehyde is at least as reactive as the secondary electrophile needed for a benzoin or Stetter reaction. A possible approach to circumvent this problem is to utilize acylsilanes.27–29 Disclosures by Heathcock's30 and then Degl'Innocenti's31 groups have shown that, upon exposure of acylsilanes to charged nucleophilic species (e.g., fluoride, cyanide), the carbonyl carbon can participate in alkylation reactions or conjugate additions. Acylsilanes have become a useful alternative to aldehydes in the generation of acyl anions, because the sterically congested nature of the silyl group precludes problematic dimerization reactions (e.g., benzoin reaction).32,33
Inspired by this early acylsilane work of Heathcock and Degl'Innocenti, we began a research program in 2002 directed toward the investigation of catalytic carbonyl addition reactions. Well-established routes to acyl anion equivalents from the combination of aldehydes and NHCs have been heavily investigated,33,34 and we hypothesized that a complementary approach utilizing acylsilanes would enhance known reactions (e.g., Stetter reaction) and provide a platform for the discovery of new reverse polarity. In this process, nucleophilic addition of an NHC to an acylsilane would facilitate a Brook 1,2 rearrangement35 with concomitant formation of the acyl anion equivalent or Breslow intermediate. However, at the onset of our investigations, it was unknown if a nucleophile larger than fluoride or cyanide would add to the sterically congested carbonyl carbon of the acylsilane, let alone facilitate the requisite 1,2 migration of the silyl group from the carbon to the oxygen.
We first explored the NHC-catalyzed 1,4 addition of acylsilanes to chalcone. The use of stoichiometric amounts of thiazolium salt 1 and DBU led to the formation of the desired 1,4-diketone in 71% yield from benzoyltrimethylsilane and chalcone (eq 1).36–38 While this result was reassuring, rendering this reaction catalytic in azolium salt was not straightforward. When the amount of 1 was reduced to 30 mol %, the isolated yield became only 43%. Interestingly, no product was observed when one equivalent of thiazolium 2 was employed. This lack of catalytic activity led us to believe that the alcohol moiety in 1 played a pivotal role in the reaction. Indeed, upon addition of four equivalents of 2-propanol to an acylsilane reaction containing 30 mol% 1, the isolated yield improved to 77%. A similar yield was obtained with thiazolium 3 and four equivalents of 2-propanol, further supporting our contention. Attempts to incorporate other azolium salt derivatives such as imidazolium, benzimidazolium, and triazolium salts proved to be unsuccessful and highlighted the importance of the catalyst structure for this sila-Stetter process. This divergent reactivity among different azolium salts provided a strong impetus to explore varying catalyst structures in several different reaction pathways.
As illustrated, this catalytic acyl anion addition is compatible with a wide range of α,β-unsaturated ketones (eq 2).36 Both electron-withdrawing and electron-donating substituents are accommodated on either aryl ring of the chalcone core and give rise to good yields. Other classes of α,β-unsaturated carbonyl electrophiles emphasized the utility of this sila-Stetter reaction. Acylsilanes reacted with diethyl fumarate and dimethyl maleate to furnish the corresponding conjugate addition products in good yields. Unsubstituted α,β-unsaturated compounds such as methyl vinyl ketone and ethyl acrylate were also competent coupling partners in this process. The compatibility of a wide range of highly reactive unsaturated carbonyl components is an impressive feature of this reaction. These compounds are notorious for being susceptible to polymerization reactions and are exposed to multiple nucleophilic species in solution. In spite of this precarious situation, the acyl anion addition products are isolated in good yields.
The reaction was then examined with respect to the acylsilane component (eq 3).36 Aromatic acylsilanes with methyl or chloro substitution are competent reaction partners, producing the desired product in 70% and 82% yield, respectively. Interestingly, para-chloro substitution renders the acylsilane most reactive, most likely due to the increased stabilization of the anion generated in the reaction. Acylsilanes containing enolizable protons are successful substrates for this reaction as well. Several 1,4-dicarbonyl compounds can be synthesized with varying substitution patterns under extremely mild conditions. The combination of an NHC and acylsilane bypasses the need for toxic cyanide catalysis and provides a highly practical and safe method for the construction of 1,4-dicarbonyl products, which can be further telescoped to provide useful furans and pyrroles (Scheme 2).37,38
We succeeded in enhancing the utility of the Stetter reaction with acylsilanes in our first carbene-catalyzed reaction. These initial studies were a pivotal step for our catalysis program and demonstrated the ability to access Breslow-type intermediates without aldehydes. Our early successful 1,4 additions allowed us to explore 1,2 additions as a means to fully realize the potential of the NHC-catalyzed generation of acyl anions from acylsilanes. Since we could access competent acyl anions without reactive carbonyl groups present, there was a strong chance that new electrophile classes could be employed to engage these useful nucleophilic intermediates formed in situ. An appropriate manifold for the investigation of 1,2 additions would be the reaction of acylsilanes with activated imines. In addition to expanding the breadth of possible reaction platforms, a successful application would produce valuable α-amino ketones directly in a particular oxidation state.39–41
The choice of imine protecting group proved pivotal to the success of the reaction. Attempts to incorporate N-Bz, N-sulfinyl, and N-sulfonyl imines were fruitless, whereas N-phosphinoyl imines provided the right balance of activation to be successful substrates as reported by Weinreb and Orr.42 This stark contrast in reactivity demonstrates a crucial consideration of any reaction utilizing carbenes as Lewis base catalysts, namely selective reaction of the NHC with one of two electrophiles present. During these carbene processes, there is the primary electrophile (e.g., the aldehyde) and a secondary electrophile, in this case the imine. Importantly, an irreversible reaction with the imine would preclude the desired productive carbene addition to the acylsilane. This intrinsic reaction characteristic makes the development of new carbene-catalyzed processes a significant challenge. In contrast, when a Lewis acid fails to promote a reaction, a stronger Lewis acid can be employed to further activate the electrophile. However, the failure of an NHC to catalyze a reaction cannot
be solved by simply increasing the nucleophilicity of the catalyst or the electrophilicity of the reaction partner, as there are too many facets of the reaction that have to be considered, such as primary and secondary electrophiles as well as key proton-transfer events.
The optimal reaction conditions turned out to be similar to those of the 1,4-addition reaction. A wide variety of substitution was accommodated on the acylsilane (eq 4).43 Both alkyl and aryl acylsilanes provided the desired α-amino ketones in good yields. Several phosphinoyl imines with various substitution patterns, including both electron-withdrawing and electron-donating groups, were suitable substrates for this transformation. In addition, aromatic heterocycles such as thiophene provided the desired products in high yields. Unfortunately, N-phosphinoyl imines derived from aliphatic aldehydes do not furnish the desired products. This limitation is attributed to the ability of the imine to readily undergo conversion into the more stable enamide, rendering it unsusceptible to nucleophilic addition.
The strategy of employing acylsilanes with NHCs as acyl anion precursors has allowed for the successful addition of acyl anion equivalents to conjugate acceptors and activated imines. This catalytic process generates 1,4-diketones and α-amino
ketones in good-to-excellent yields. This initial discovery by our group propelled us to investigate new NHC-catalyzed reactions and polarity reversal (umpolung) strategies.
The combination of NHCs and aldehydes has led to useful new chemistry in our laboratory beyond the area of umpolung catalysis. While a plethora of the existing carbene catalysis is focused on polarity reversal chemistry, alternative modes of reactivity are possible, and have indeed been explored. One different avenue is oxidation of the initial tetrahedral intermediate formed from the addition of an NHC to an aldehyde. Collapse of this intermediate would generate an acyl azolium species with concomitant formation of a hydride equivalent.
In 2006, we disclosed the application of this route in the context of an NHC-catalyzed hydroacylation (Scheme 3).44 In this Tishchenko-like process, an aromatic aldehyde–NHC adduct generates a hydride equivalent in the presence of an organic oxidant, an α-keto ester. The initial collapse of the tetrahedral intermediate to produce an activated ester is unprecedented, and adds an interesting facet to the potential avenues of NHC-catalyzed reactions. Once the ketone undergoes reduction, the resulting alkoxide regenerates the catalyst through addition to the acyl azolium intermediate. In aprotic solvents, such as CH2Cl2, the hydroacylation products are isolated in good yields when triazolium precatalyst 5 is used. Additionally, when the reaction is conducted in MeOH, the α-hydroxy ester can be isolated as the sole product due to catalyst regeneration by the solvent. One limitation of this methodology is the requirement that the aldehyde starting material possess a nonenolizable α-carbon atom.
In order to further explore this reaction pathway, a crossover experiment was performed with an α-keto ester and 0.5 equiv each of deuterated benzaldehyde and p-tolualdehyde.44 In this reaction, all four potential products were observed, supporting our contention that reduction and acylation are separate steps in the reaction pathway. Additionally, when benzoin is exposed to the reaction conditions, the hydroacylation product is isolated suggesting that the benzoin reaction is reversible and may be operating under the reaction conditions. In this new reaction, the carbene catalyst is responsible for two distinct processes: oxidation and acylation. The novelty of this reaction is further illustrated through the combination of two components (aldehyde and ketone) participating in a disproportionation reaction.
Following this initial report, we focused on the use of more conventional oxidants to facilitate the formation of acyl azoliums. An interesting, but underutilized, approach to unsaturated esters has been the Corey–Gilman oxidation. In this process, an allylic alcohol is oxidized in the presence of 10 to 20 equivalents of cyanide and MnO2, first to the corresponding aldehyde and then to the ester. This streamlined process to convert alcohols into esters would be incredibly useful if the reaction avoided the use of superstoichiometric amounts of cyanide. Our previous success of replacing cyanide with NHCs in the context of the Stetter reaction encouraged us to pursue a similar strategy with regard to this oxidation. Importantly, MnO2 was chosen in order to allow the presence of the inactivated nucleophilic alcohol required for catalyst regeneration.45
We chose to evaluate this process to demonstrate the use of carbenes as oxidation co-catalysts and to develop a practical oxidation procedure. When butanol is used as solvent, several allylic and benzylic alcohols are successfully oxidized to the corresponding unsaturated butyl esters with 10 mol% precatalyst 5 (eq 5).45 In many cases, the nucleophilic alcohol may be too costly to use as solvent. In this circumstance, slight modifications to the reaction conditions allow the alcohol to be used as a reagent as opposed to solvent. In toluene, just 5 equivalents of the nucleophilic alcohol are required to facilitate ester formation.
During these initial carbene-catalyzed oxidation investigations, we recognized that the addition of an NHC to any aldehyde (activated or inactivated) should generate a transient benzylic-type alcohol! The catalyst itself is aromatic and should induce mild oxidations of the resulting intermediate. Indeed, the addition of azolium salt 5 to a variety of saturated aldehydes in the presence of MnO2 and a nucleophilic alcohol enables oxidation of the aldehydes to their respective esters (Scheme 4).46 It is noteworthy that aldehydes with electron-rich aromatic rings are accommodated under these reaction conditions, whereas typical Pinnick oxidation conditions result in significant chlorination of the aromatic ring. For example, when 3-(2,4,6-trimethoxyphenyl)propanal is oxidized using Pinnick conditions a substantial amount of monochlorination of the aromatic trimethoxyphenyl ring occurs, whereas under NHC-catalyzed conditions only the desired ester is produced.47,48
Interestingly, the use of a chiral NHC in this reaction offers the opportunity to desymmetrize meso diols through the in situ formation of chiral activated ester equivalents. In a proof of concept experiment, in the presence of triazolium salt 7 with the combination of K2CO3 and a proton sponge as a base, a meso diol is acylated with modest enantioselectivity (eq 6).45 Problems with base-catalyzed intramolecular acyl-transfer reactions presumably inhibit higher selectivities for this process, but these initial results pave the way for carbenecatalyzed stereoselective acylation reactions using simple aldehydes.
From 2003 onward, our substantial interest in accessing Breslow intermediates with acylsilanes and understanding the mechanistic aspects of this process stimulated our thinking about new potential applications of related structures. An
intriguing possibility was the “extension” of the nucleophilic character at C-1 of the aldehyde to the distal site C-3 through a C–C multiple bond. In this approach, the addition of carbenes to aldehydes containing an additional unsaturation unit could relocate the electron density in the Breslow intermediate to the B carbon of the aldehyde (Scheme 5).49 This transient nucleophile is a “vinylogous” carbonyl anion or a homoenolate.
Our group began investigating this electronic reorganization with the goal of intercepting this possible homoenolate intermediate with a suitable electrophile for β functionalization and a suitable nucleophile for subsequent acylation.49,52 We initially chose to explore this reaction with a simple electrophile: a proton. The proposed pathway for this process begins with the initial 1,2 addition of the NHC to the α,β-unsaturated aldehyde (Scheme 6).49,52 Following proton migration, the electron density that would be typically located at the carbonyl carbon is extended to the B position with formation of the extended Breslow intermediate (I). Addition of a proton generates enol II, which tautomerizes to activated acylating agent III. In the presence of a nucleophile, the NHC catalyst is regenerated and the catalytic cycle is restarted. The obvious choice for a proton source is an alcohol that also serves as a nucleophile in the last step to promote catalyst turnover.
Initial probing of this reaction was moderately successful. Exposure of cinnamaldehyde to 30 mol% 6, DBU, and phenol in toluene led to a 55% isolated yield of the desired saturated ester. A serendipitous discovery was made when the reaction was run in CHCl3, which had been passed through basic Al2O3 but not distilled. With phenol being used as the proton source, a large amount of the saturated ethyl ester was isolated. In hindsight, it became clear that the source of the ethanol was the chloroform, which uses ethanol as a stabilizer. With the knowledge that two alcohols can be used simultaneously (one as a proton source and one as the nucleophile), we quickly discovered that high yields could be obtained for this process when phenol is employed as a proton source and a second alcohol is used as a nucleophile.
Under these new reaction conditions, which employ 5 mol% of azolium salt 6, 2 equiv of phenol, and 4 equiv of the nucleophilic alcohol, a variety of saturated esters can be synthesized (Scheme 7).49,52 Both secondary and primary alcohols are accommodated under the reaction conditions. Optically active alcohols retain their integrity in this process to generate enantioenriched esters. Not surprisingly, tert-butyl alcohol was unreactive. Additionally, the reaction tolerates a variety of α,β-unsaturated aldehydes with both alkyl and aryl substitution. Aldehydes with additional substitution at the A position, as well as substrates with β,β-diaryl substitution, afford the corresponding products in good yields. Our group has also explored asymmetric variants of this process.
This initial report demonstrated that homoenolate activity could be generated and utilized in a productive fashion. The formation of simple saturated esters provided a platform for the investigation of homoenolates and introduced our group to a new and exciting area of chemistry. Following the success of this reaction, we actively pursued new applications with these atypical nucleophiles. The most significant challenge encountered with these reactions is the vinylogous benzoin reaction. Homoenolate addition to an equivalent of unsaturated aldehyde starting material must be avoided in these reactions if a secondary electrophile is incorporated into a new reaction sequence. The most reasonable approach to this problem would be to increase the electrophilicity of this secondary electrophile. This adjustment, however, can make the addition of the NHC to the secondary electrophile more likely, thus inhibiting the reaction. These complications highlight the impressive nature of all homoenolate reactions reported to date and demonstrate the balance of electronic and steric effects that are required for future reaction development.
A carbene-catalyzed homoenolate addition that forms new carbon–carbon bonds would clearly be a valuable reaction and enhance the applicability of this process. An appropriate strategy would be to utilize an electrophile, which, upon homoenolate addition, generates a transient nucleophile to aid in catalyst regeneration (Scheme 8). In this vein, ylides contain the appropriate functionality, and the ability of these dipolar species to undergo cycloadditions is well-precedented.53–56
3-Oxopyrazolidin-1-ium-2-ides are stable compounds that can be prepared in gram quantities, and can thus be practical coupling partners57–60 for the investigation of this reaction.61 To our gratification, a variety of azolium salts catalyzed the reaction between cinnamaldehyde and the diphenyl-substituted azomethine imine in the presence of DBU with excellent diastereoselectivity albeit in low yields. The optimal reaction conditions were obtained using 20 mol% of azolium salt 9 and DBU in CH2Cl2 while heating the reaction at 40 °C (eq 7).62 A survey of α,β-unsaturated aldehydes revealed that a variety of electron-rich aromatic rings are accommodated in the reaction. Unfortunately, electron-deficient unsaturated aldehydes are not compatible. β-Alkyl substitution and extended dienylic substitution are also tolerated. Investigation of the reaction with respect to the azomethine imine component demonstrated that several substitution patterns are allowed. However, azomethine imines derived from aldehydes with enolizable protons do not afford any tetrahydropyridazinone product. Successful azomethine imine substrates in this reaction typically possess phenyl substitution in the 5 position, primarily due to the insolubility of the unsubstituted analogues.
We investigated this reaction platform further with the incorporation of a second 1,3-dipole: nitrones. In addition to being well-known partners for cycloadditions,63–66 successful homoenolate addition to these ylides would potentially produce γ-amino acid67–69 derivatives as well as γ-lactams. While we were pleased to discover that the homoenolate addition does occur, the initial 6-membered-ring heterocyclic product was unstable toward chromatography. Addition of NaOMe to the reaction mixture upon consumption of starting material helped bypass this problem and afforded γ-hydroxy amino ester products (Scheme 9).70 The products can be manipulated further with Pd(OH)2/C and H2 to provide γ-amino esters. An impressive aspect of this reaction is the synthesis of optically active products with chiral azolium 10. The high degree of asymmetric induction is surprising considering the distance between the reactive carbon atom and the stereogenic centers of the homoenolate intermediate.
We then wondered whether this methodology could be extended to heteroatom electrophiles such as in a homoenolate-based amination reaction, which would constitute a reverse-polarity approach to creating B-amino carbonyl compounds. These carbonyl compounds are commonly synthesized through conjugate additions of nitrogen nucleophiles.71–74 When we employed diazenes (RN=NR’) as elecrophilic amination partners, significant complications were encountered due to the reactive nature of the diazene functional group. A notable side product in these reactions was 1-benzoyl-2-phenylhydrazine [PhC(O)NHNHPh]. Based on the previously discussed hydroacylation and oxidation reactions, we posit that the initial tetrahedral aldehyde–carbene adduct can behave as a hydride source and perform conjugate reductions on the diazene. The damage to the outcome of the reaction is heightened by not only sacrificing the diazene, but an equivalent of the aldehyde as well. Fortunately, employing precatalyst 11 and lowering the temperature of the reaction circumvented these problems (eq 8).75
Our success with generating homoenolate reactivity using NHCs led us to believe that enolate generation and utilization were possible. The current mechanistic understanding of the homoenolate process includes the possible generation of a short-lived enol (Scheme 10, structure II).76,77 The tautomerization of this enol followed by nucleophilic attack on the transient acyl azolium intermediate drives catalyst regeneration. This observation provides an opportunity to tap into the powerful ability of the carbene to functionalize the α, β, and carbonyl carbons of an A B-unsaturated aldehyde in a single flask! Our
goal was to intercept this nucleophile (II) with a competent electrophile and thus expand the number of NHC-catalyzed reactions.
Toward this end, we synthesized substrates that would not only maximize the potential success of the reaction but also provide interesting structural motifs (eq 9).77 This three-atom functionalization proceeded as envisaged in Scheme 10. While the β-protonation step is not well-understood, it has
been observed that weaker bases, such as (i-Pr)2EtN, and their conjugate acids, are more accommodating in this process. An intramolecular Michael addition follows the β-protonation step and results in the construction of a five-membered ring. Under these conditions, catalyst regeneration is afforded by the O-acylation of the newly formed (second) enol. However, the addition of methanol is required to avoid hydrolysis of the initial labile lactone products and to facilitate purification. Importantly, when aminoindanol-derived precatalyst 7 is used in combination with (i-Pr)2EtN, excellent diastereoand enantioselectivities are achieved for a wide range of substrates.
The success achieved with this highly diastereo- and enantioselective intramolecular NHC-catalyzed Michael addition led our group to investigate an intramolecular aldol reaction.78,79 Readily prepared symmetrical 1,3-diketones undergo intramolecular aldol reactions to afford optically active cyclopentene rings. In this reaction, the enol generated from the addition of chiral, optically active NHC 10 to the aldehyde performs a desymmetrization of the 1,3-diketone. Acylation of the resulting alkoxide is coupled with a decarboxylation step to afford the cyclopentene adducts with excellent enantiocontrol (eq 10).78,79 Importantly, degassing of the solvent leads to a dramatic increase in yield. In some cases, unsaturated acids are observed, and they are thought to originate from the oxidation of the homoenolate intermediate.
The high selectivity achieved with this system is believed to arise from a 6-membered hydrogen-bonded feature in the Breslow-type intermediate. The enol proton behaves as a bridge between the enol oxygen and the ketone oxygen, which predisposes the complex to undergo the aldol reaction and minimizes the nonbonding interactions between the catalyst and the keto group not undergoing attack. The regeneration of the catalyst is also a result of the hydrogen bonding in the adduct since an anti disposition of the alkoxide and acyl azolium groups in the adduct would inhibit subsequent intramolecular acylation.
In order to demonstrate the intrinsic value of this desymmetrization process, we adapted this methodology to the synthesis of the bakkenolide family of natural products.80,81 The bakkanes are comprised of a cis-fused 6,5-cyclic system with two quaternary stereogenic centers, one of which contains an angular methyl group. This key structural element provided an excellent opportunity to apply our methodology and provide a modern demonstration of the power of carbene catalysis in total synthesis.82–84 The crucial NHC-catalyzed bond-forming
The formation of reactive enols through carbene catalysis is an exciting area. The use of α,β-unsaturated aldehydes in this process requires extensive atom and electronic reorganization. In addition to our continued studies along these lines, we are also investigating new approaches using carbenes to generate enols or enolates. Specifically, we hypothesized that carbon–carbon-bond-forming reactions were possible with acetate-type enols derived from the addition of NHCs to α-aryloxyacetaldehydes. The aryloxy (ArO) group would not only facilitate enol formation through an elimination event, but would also assist in catalyst regeneration by adding to the acyl azolium. While this new concept to generate acetate enolates has been successful (vide infra), the initial approach requiring a functional group to behave as both a good leaving group and a competent nucleophile was challenging. A good leaving group may initiate the formation of the enol faster, but would not be effective at catalyst regeneration.85 Thus, the optimal ArO group must strike a balance between good-leaving-group ability and sufficient nucleophilicity.
In order to explore the potential of this process, we chose to incorporate enones with tethered aldehydes. The strategy of tethering the conjugate acceptor to the potential nucleophile not only increases the chance of a productive bond formation, but also forms privileged 3,4-dihydrocoumarin structures. Indeed, exposure of these enones to 10 mol% 11 in MeCN affords 3,4-dihydrocoumarins with varying substitution patterns (eq 11).86a
The fragmentation required for this reaction to occur was a strong impetus for investigating the reaction pathway. First, the possibility of the alkoxide undergoing acylation and then performing a Michael addition was a plausible route. To discount this option, the potential intermediate was synthesized and exposed to the reaction conditions. No reaction was observed, thus refuting the presence of the acylated phenoxide in the catalytic cycle. A crossover experiment was also performed to determine if fragmentation was occurring. When two A-aryloxy aldehydes were exposed to the reaction conditions in a single flask, a randomized mixture of 3,4-dihydrocoumarin products was obtained, supporting the contention that the starting material fractures during the reaction.86
These interesting enolate precursors were further applied in Mannich-type reactions. In this case, we sought to synthesize an enolate precursor that would allow for facile elimination and subsequent enol formation, but yet retain enough nucleophilicity to assist in catalyst regeneration (Scheme 12).86b After much optimization, we discovered that a 4-nitrophenoxide anion was the most suitable leaving group. In contrast to the previous studies of acyl anion additions to imines with N-phosphoryl protecting groups, N-tosylimines proved to be the most compatible. Additionally, as opposed to the previously described NHC-catalyzed reactions, which use a consortium of amine bases to deprotonate the azolium precatalyst, the most successful base in this process is sodium 4-nitrophenoxide. One drawback of this process is that the initial aryl β-amino ester products are unstable toward column chromatography. However, addition of benzylamine upon consumption of the starting material circumvents this problem and affords a wide variety of β-amino amides (eq 12).86b Interestingly, these acetate-type enols add to imines with
excellent levels of stereoselectivity despite the problems typically associated with 1,2 additions of acetate enolates.
The highly selective Mannich-type reaction of α-aryloxy aldehydes provides an opportunity to synthesize products that are valuable to the chemical and biological communities. In addition to α-unbranched β-amino amides, synthesis of the corresponding β-amino acids is also possible with exposure to NaOH (Scheme 13).86b Formation of 1,3-amino alcohols is accomplished with LiBH4, and the synthesis of more stable esters is demonstrated by the addition of sodium methoxide. Lastly, β-peptide formation is possible by in situ interception with benzyl-protected (S)-alanine.86b These two reaction manifolds, the Michael and Mannich reactions, demonstrate the viability of this type of “rebound” catalysis, and will surely open doors to new reactions.
As discussed above, the combination of NHCs with α,β-unsaturated aldehydes can result in two divergent pathways: (i) oxidation of the aldehyde or (ii) internal redox reactions through addition of the homoenolate to an electrophile followed by oxidation of the aldehyde carbon. In many cases, both pathways operate under the same reaction conditions.49,87 Even though NHC-promoted reactions have garnered significant attention in recent years, the discovery of the conditions required for new reactions has remained empirical. Methods to predict and control which pathway is likely to be operating are imperative to the development of new reactions.
One course of action following the addition of the NHC to the aldehyde is the formation of the homoenolate. As previously stated, this option requires a formal 1,2-proton shift to succeed the formation of the tetrahedral intermediate. The second possible pathway includes a collapse of the tetrahedral intermediate, generating an acyl azolium intermediate and a formal reducing equivalent. Our goal, in collaboration with Cramer's group at the University of Minnesota, was to apply computational models toward this complex problem in order to better understand the potential reaction pathways and factors that control the reaction preference for one pathway or the other.88
Using crotonaldehyde as a model system for α,β-unsaturated aldehydes, enthalpies of the proposed intermediates were obtained using Density Functional Theory (DFT) calculations with solvation models providing a correction for the solution environment.89
The results of these calculations are in agreement with the experimental findings: the homoenolate pathway is more dependent on the choice of catalyst, while the choice of solvent is less influential. In the corresponding experimental studies, a 10:1 mixture of solvent to methanol was employed to assist in catalyst turnover. Four different, but relatively simple, azolium salts were surveyed as precatalysts over a narrow range of solvents. While the GC yields of these reactions were low, the ratio of oxidation product to homoenolate product was the important statistic. The results indicated that polar protic solvents such as methanol favor the oxidation pathway; but as solvent polarity decreases, the homoenolate pathway becomes more favored.88 While catalysts 5 and 9 showed that catalyst structure could be used to favor the oxidation pathway, the choice of solvent also played a key role in the distribution of products in the case of 1,3-dimethylimidazolium chloride and 1,3-dimethylbenzimidazolium iodide.
These initial results suggest that a desired pathway can be favored with a specific choice of catalyst and solvent, a choice that is informed by a rational theoretical exploration of the reaction parameters. This first computational exploration of these Lewis base reactions will hopefully lead to a more systematic approach toward the development of NHC-catalyzed reactions.
Our laboratory has been inspired by nature and by the pioneering work of Ugai and Breslow to develop whole new families of acyl anion, homoenolate, enolate, and redox processes. The structural diversity of these intriguing azolium catalysts allows them to effect several transformations and leads to the conclusion that their potential has not been realized. A high degree of stereocontrol is possible through the use of chiral, optically active N-heterocyclic carbenes. The integration of experimental data with computational analysis has provided the first study suggesting that the further development and/or optimization of carbene-catalyzed reactions need not be based solely on empirical approaches. With continued mechanistic investigation, additional insight into these powerful reactions will drive further development of the carbene catalysis field. In the future, new discoveries that allow for lower catalyst loadings may enable the incorporation of these catalysts into synthetic plans for the construction of complex molecules. Ultimately, the continued exploration of carbene catalysis will undoubtedly produce new reactions and strategies, and we look forward to participating in these discoveries.
We thank Northwestern University, the National Institute of General Medical Sciences (R01GM73072), 3M, Abbott Laboratories, Amgen, AstraZeneca, Boehringer Ingelheim, GlaxoSmithKline, and Novartis for their generous support of this work. A.C. thanks The Dow Chemical Company for financial support. E.M.P. is a recipient of an ACS Division of Organic Chemistry Graduate Fellowship Sponsored by Organic Reactions.
Eric Phillips was born in 1983 in Grand Rapids, MI. He received his B.S. degree in chemistry from Western Michigan University. In 2005, he joined the laboratory of Professor Karl A. Scheidt at Northwestern University, where he is currently a fifth-year graduate student. The majority of his graduate work has focused on the development of reactions catalyzed by N-heterocyclic carbenes, and has received an ACS Division of Organic Chemistry Graduate Fellowship sponsored by Organic Reactions. Upon completion of his Ph.D. requirements, he will join the laboratory of Professor Jon A. Ellman at the University of California, Berkeley, as a postdoctoral fellow.
Audrey Chan was born in São Paulo, Brazil, and immigrated to Brooklyn, New York, at the age of six. She received her B.S. degree in chemistry in 2002 from Cornell University. After working at Merck Research Laboratories as a medicinal chemist, she joined the group of Professor Karl A. Scheidt at Northwestern University in 2003. As a Dow Chemical Company Predoctoral Fellow, she studied N-heterocyclic carbene catalyzed homoenolate and hydroacylation reactions. Upon completion of her Ph.D. requirements in 2008, she joined Cubist Pharmaceuticals in Lexington, Massachusetts, where her research focuses on drug development for anti-infectious diseases.
Karl Scheidt became interested early in science because of his father, W. Robert Scheidt, a prominant inorganic chemistry professor at the University of Notre Dame. He received his Bachelor of Science degree from Notre Dame in 1994 while working in the laboratory of Professor Marvin J. Miller. Under the direction of Professor William R. Roush, Karl earned his Ph.D. degree from Indiana University, and was a National Institutes of Health Postdoctoral Fellow with Professor David Evans at Harvard University. Since joining Northwestern University in 2002, Karl's research has focused on the development of new catalytic reactions and the total synthesis of molecules with important biological and structural attributes. He currently holds the Irving M. Klotz Research Chair in Chemistry and is the Alumnae of Northwestern Teaching Professor. He is a fellow of the Alfred P. Sloan Foundation, an American Cancer Society Research Scholar, and the recipient of a National Science Foundation CAREER Award. His recent honors include: The GlaxoSmithKline Scholar Award (2008), AstraZeneca Excellence in Chemistry Award (2007), Novartis Chemistry Lecture Award (2007), Amgen Young Investigator Award (2006), Boehringer Ingelheim New Investigator Award in Organic Chemistry (2005), Northwestern University Distinguished Teaching Award (2005), 3M Nontenured Faculty Award (2005), Abbott Laboratories New Faculty Award (2005), and the Amgen New Faculty Award (2004).