Article pubs.acs.org/accounts
Explaining Anomalies in Enamine Catalysis: “Downstream Species” as a New Paradigm for Stereocontrol Jordi Burés,†,‡ Alan Armstrong,† and Donna G. Blackmond*,†,‡ †
Department of Chemistry, The Scripps Research Institute, La Jolla, California 92037 United States Department of Chemistry, Imperial College London, London SW72AZ, U.K.
‡
CONSPECTUS: Enantioselective organocatalysis by diarylprolinol ethers is remarkably selective and efficient for a wide range of transformations involving a number of different activation modes, including HOMO activation via enamines. A simple steric model based on facial discrimination in the attack of an enamine on an electrophile has been invoked to rationalize high enantioselectivity. In a number of reactions, however, experimental observations have persistently left us with mechanistic puzzles that fail to fit neatly into this simple picture. Further studies involving both kinetic profiling of reaction networks and NMR spectroscopic characterization of the structures of intermediate species helped us to address these puzzles. This work led to the proposal of a new paradigm for stereocontrol in asymmetric aminocatalysis, demonstrating that the ultimate stereochemical outcome may not, in fact, be determined solely in the stereogenic bond-forming step between enamine and electrophile. The identification of stable species occurring downstream from the addition of an electrophile to an enamine, and the discovery of kinetic features that are diagnostic of the presence of such species, allows development of a new mechanistic framework that reveals a hierarchical selection. Both kinetic and thermodynamic processes associated with downstream intermediates can exert an influence on the ultimate enantioselectivity. Interestingly, the role of these species may be either to enhance or to erode selectivity established at the enamine−electrophile step. The reversibility of steps preceding and subsequent to the stereogenic bond-forming step is an important factor, as are reaction parameters that may stabilize or destabilize intermediates, including the nature of the electrophile counterion and the solvent. These concepts hold implications for the future design and optimization of asymmetric catalytic processes, because such design does not necessarily feature the same parameters at the second hierarchical level as it does at the first. Examples presented to highlight these issues include the conjugate addition of aldehydes to nitroalkenes and the α-chlorination and α-selenylation of aldehdyes using diarylprolinol ether catalysts commonly assumed to follow a steric model for enantioselectivity. While the new paradigm for stereocontrol involving downstream intermediates is developed here for enamine catalysis, the same concepts may hold for other organocatalytic modes of activation.
■
INTRODUCTION
been, there remain puzzling experimental observations from a number of different reactions that are not easily explained by the model shown in Schemes 1 and 2. Indeed, our detailed mechanistic studies6−8 of a number of reactions using these catalysts revealed general experimental features that suggest a more complex mechanism may operate for enamine catalysis by diarylprolinol ethers. In some cases, we found that enantioselectivity is ultimately determined not at the step of electrophile−enamine interaction but instead further along the catalytic cycle after the initial stereogenic bond-forming step (in steps not elaborated in Scheme 1), leading us to highlight the importance of what we term “downstream intermediate species”. The concept holds important implications for the design and optimization of reactions based on diarylprolinol
The development of a wide range of highly selective transformations using pyrrolidine-based organocatalysts has led to their elevation as “privileged” structures in asymmetric catalysis.1 The diarylprolinol ethers introduced independently by the groups of Jørgensen2 and Hayashi3 enjoy a particularly broad usage because of their ability to employ several different activation modes.4 Scheme 1 shows the conventionally accepted mechanism for enamine activation. Diarylprolinol ethers contain a bulky catalyst side chain designed to offer excellent facial differentiation, illustrated in Scheme 2 for enamine intermediate I. Scheme 2 shows that enantiocontrol in the steric model I is in the opposite sense as found for proline enamine II directed by Bronsted acid cocatalysis. The simple steric model I rationalizes the experimental results in a remarkably wide variety of reactions catalyzed by diarylprolinol ethers.5 However, as successful as the model has © XXXX American Chemical Society
Received: August 25, 2015
A
DOI: 10.1021/acs.accounts.5b00394 Acc. Chem. Res. XXXX, XXX, XXX−XXX
Article
Accounts of Chemical Research Scheme 1. Enamine Mechanism
Scheme 3. Diarylprolinol Ether Catalyzed Reactions
Scheme 2. Transition State Models
chlorination (Scheme 4, Table 1),11 and MacMillan reported high enantioselectivities using imidazolidinone catalysts such as
ethers, and it should be considered that such mechanistic nuances might apply to other catalyst systems. Our investigations of the conjugate addition of aldehydes to nitroalkenes6,7 and the α-chlorination7 and α-selenylation8 of aldehydes employed kinetic profiling by in situ reaction calorimetric monitoring along with a battery of NMR spectroscopic tools for identification of catalytic intermediates. We begin by describing the mechanistic puzzles that led us to probe for insights beyond the conventional steric model I, followed by a review of our results leading to a new paradigm for stereocontol supporting a role for “downstream intermediate species” in enamine catalysis.
Scheme 4. Organocatalyzed α-Chlorination of Aldehydes
■
BACKGROUND Our first puzzle arises from studies of the conjugate addition of aldehydes to nitroalkenes, transformations in which using diarylprolinol ether catalysts are particularly efficient with linear aldehydes as substrates (Scheme 3a). This reaction has become so widely employed that it has been termed a “benchmark” for assessing new chiral amine catalysts.9 However, when α,α-disubstituted aldehydes are used as substrates, the results are more complex, as is also shown in Scheme 3. Reaction rate is virtually completely suppressed for α,α-disubstituted aldehydes, a fact that remains unexplained by the cycle of Scheme 1, especially because a similar sluggishness is not observed for α,α-disubstituted aldehydes in reactions with other electrophiles, such as α-amination with DEAD,6b,10 as shown in Scheme 3b. Puzzle 1. Does the cycle of Scheme 1 with model I of Scheme 2 (shown with a general electrophile, E) explain why the reaction rate is nearly completely suppressed in reactions of α,α-disubstituted aldehydes with E = nitroalkenes but not with E = DEAD? A second puzzle arises from the highly vaunted generality of the diarylprolinol ethers for enamine catalysis of a wide range of electrophiles. In work mainly by the groups of Jørgensen and Hayashi, these catalysts have been shown to give excellent enantioselectivity not only in conjugate addition to nitroalkenes and α-aminations as shown in Schemes 2 and 3 but also in αaddition to vinyl ketones, Mannich reactions, α-sulfenylations, α-fluorinations, and α-brominations.4 Jorgensen had shown that 2,5-diphenylpyrrolidine 8 is an excellent catalyst for α-
Table 1. α-Chlorination Reactions
a
catalyst
substrate
“Cl+”
solvent, T (°C)
% ee
ref
7 7 8 3
1c 1c 1d a
2c 2d 2c a
CHCl3, 4 CHCl3, −30 CH2Cl2, 0 a
19 92 95 a
12 12 11 a
Not reported.
7 using a quinone-based chlorination agent but not with NCS12 (Scheme 4, Table 1). The lack of a report for α-chlorination using diarylprolinol ethers is puzzling, given the success of this and other α-halogenations for other pyrrolidine catalysts. Puzzle 2. Does the cycle of Scheme 1 with model I of Scheme 2 explain why α-chlorination has not been reported using diarylprolinol ether catalysts, given the success of other α-halogenations with catalysts thought to obey model I? Can model I account for dif ferent performance f rom dif ferent chlorinating agents as electrophiles? A third puzzle addresses the role of the reaction medium in model I. High enantioselectivities in the α-selenylation of aldehydes using diarylprolinol ethers has been reported separately by the groups of Melchiorre13 and Cordova14 (Scheme 5). Interestingly, Cordova and co-workers14 found a B
DOI: 10.1021/acs.accounts.5b00394 Acc. Chem. Res. XXXX, XXX, XXX−XXX
Article
Accounts of Chemical Research Scheme 5. α-Selenylation of Aldehydes14
Scheme 6. Identification of Catalyst Resting States in the Reactions of Schemes 3, 4, and 5
striking solvent effect in the reaction: they reported an inversion in the sense of enantioselectivity when the solvent was switched from toluene (95% ee) to acetonitrile (−30% ee). The authors noted that while the reaction medium may alter selectivities, this reversal was unexpected. The fundamental mechanism relating the solvent to model I of Scheme 2 remains unclear. Puzzle 3. Does the cycle of Scheme 1 with model I of Scheme 2 rationalize an inversion in enantioselectivity caused simply by changing the solvent? These three puzzles were addressed in a series of kinetic and mechanistic studies summarized below.
■
RESULTS AND DISCUSSION
Kinetic Profiles
An initial clue to solving the three puzzles presented above came from kinetic studies of the reactions in Schemes 3a, 4, and 5. Figure 1 reveals that all three reactions exhibit kinetic profiles structures of the intermediates identified in each case. In the conjugate addition reactions of Scheme 3a, a single cyclobutane diastereomer 10 accounts for virtually all of the catalyst for both linear and α,α-disubstituted aldehydes.6,7 Under reaction conditions for linear aldehydes, 10 gives way at very high conversion to the product enamine E-11. It should be noted that 11 is unable to form in the case of the α,α-disubstituted aldehydes. In the α-chlorination and α-selenylation reactions, we found the catalyst resting state to be a pair of diastereoisomeric intermediates that persist throughout the reaction. Covalent addition of 2c across the enamine double bond in αchlorination yields one rapidly equilibrating syn/anti pair, 12, as the resting state. The connectivity but not the stereochemistry of these species was established by low temperature NMR.7a In α-selenylation, product enamine species E-13 and Z-13 are present in approximately equal concentrations in reactions carried out in CH2Cl2, while in toluene only E-13 is observed.8 Note that in each case, the resting state contains both the aldehyde and the electrophile substrates, as predicted by the zero order kinetics in the reaction profiles of Figure 1. Once the predominant catalytic species in these reactions have been identified in each case, the next mechanistic task is to demonstrate whether the species lies on the catalytic cycle or is simply a spectator. We treat each of the “puzzles” raised above in turn to probe the role of these species in each reaction.
Figure 1. Kinetic profiles via reaction calorimetry of three reactions catalyzed by 3:6−8 (a, magenta) Scheme 3a, conjugate addition, 1a (1.2 M), 2a (1.0 M), 3 (0.1 M) CH3COOH (0.025 M), toluene, 25 °C; (b, green) Scheme 4, α-chlorination, 1d (0.63 M), 2c (0.48 M, 3 (0.048 M), CH3COOH (0.01 M), CHCl3, −11.4 °C; (c, blue) Scheme 5, αselenylation. 1d (0.075 M), 2e (0.025 M), 3 (0.0025 M), CH2Cl2, 25 °C.
that are notable for their similar and highly unusual features.15 In each case shown in Figure 1, a rapid initial rate was followed by a regime of zero order kinetics in both aldehyde and electrophile concentrations. Observation of a regime of constant rate is typically attributed to “saturation kinetics” in both substrates, suggesting that the rate-determining step in the cycle for each reaction involves a catalyst resting state that contains both substrates. The rapid initial rate is characteristic of a transient period that occurs while the concentration of this resting state is building up.
Puzzle 1: Conjugate Addition to Nitroalkenes
While it was originally suggested that the cyclobutane species 10, which was identified simultaneously by us and by a collaboration of the groups of Hayashi and Seebach,16 is a “parasitic” off-cycle species, a number of observations taken together led us to propose that for the case of linear aldehydes, both 10 and E-11 lie on the catalytic cycle.
Downstream Intermediates
Our mechanistic studies then turned to identification of the catalyst resting state highlighted in each of these reactions, which was accomplished via a detailed series of variable temperature and 2-D NMR experiments. Scheme 6 shows the C
DOI: 10.1021/acs.accounts.5b00394 Acc. Chem. Res. XXXX, XXX, XXX−XXX
Article
Accounts of Chemical Research First, enamine formation was shown to be rapid and reversible for both linear and α,α-disubstituted aldehydes, confirming that the suppression of product formation for the latter must occur at a step further along in the cycle. Second, crossover experiments demonstrated that electrophile addition is also rapid and reversible. Formation of the cyclobutane resting state 10 proceeds rapidly for either R = Ph or R = H, but its formation appears to signal a dead end for reaction of R = Ph but not R = H. Third, the importance of R = H is further highlighted by the finding of a small normal deuterium KIE using α-deuterated aldehyde 1a. Given that the ratedetermining step occurs at some step after rapid, reversible enamine formation, a role for this aldehyde proton is difficult to envision in the cycle of Scheme 1. To reconcile these observations, we proposed the reaction network shown in Scheme 7, highlighting the role of the downstream species in red.
Scheme 8. Rationalization for Substrate Reactivity Differences in the Reactions of Scheme 2
Scheme 7. Catalytic Cycle for the Conjugate Addition to Nitroalkenes of Scheme 3a
provide a clue: reactions of DEAD with 1a and 1b both show first order kinetics in aldehyde concentration and zero order kinetics in DEAD concentration. A simple explanation for such behavior is that enamine formation is the rate-determining step when DEAD is the electrophile. The free catalyst is the resting state, and no stable downstream species exist under catalytic conditions. This electrophile thus provides a classic example of model I, with the effectiveness of the catalyst dictated by the irreversible reaction of enamine with electrophile as illustrated in Scheme 2. The complexities uncovered in the mechanism of the reactions of Scheme 3 require further discussion in the context of model I of Scheme 2. Indeed, the observed reversibility of the enamine−electrophile interaction17 raises the possibility, at least theoretically, that high product enantioselectivity could ultimately be achieved even if selection at the enamine− electrophile level is very poor. A Curtin−Hammett scenario17 may be outlined where imperfect selectivity of enamine addition to electrophile 2a leads to reversible formation of two (or more) hypothetical diastereomeric cyclobutane intermediates in rapid equilibrium, with enantioselectivity being a function of the relative stability of the intermediates and their relative reactivity onward to product (via product enamines). This is illustrated in Scheme 9. This scenario presents an alternate view of key catalyst design parameters compared with model I of Scheme 2. A highly selective catalyst in the reaction of Scheme 3a need not be selective in the enamine−electrophile interaction, if either the stability of any alternate cyclobutane formed is low (Kminor is small) or its productive rate constant kminor is small. The reversibility of the network offers a “second chance” to any system exhibiting poor selectivity at the stereogenic bondforming step.18 High product enantiomeric excess can be achieved simply by channeling the reaction to a single highly stable downstream species that leads to a single product. Other Mechanistic Proposals. Alternatives to the catalytic cycle through cyclobutane 10a have been proposed. Seebach and co-workers19 initially questioned this route on the intuitive assumption that the elementary steps from the cyclobutane to the product enamine (through fleeting species, an enaminonitronate or a 1,2-oxazine, not shown in Scheme 7) would be energetically uphill; later calculations led them to retract this viewpoint.19b,20 Pihko and co-workers21 suggested that in the case of α-alkyl substituted nitroalkenes (not studied in our
Scheme 7 incorporates all of the experimental clues noted above to suggest that with R = H, the cyclobutane resting state 10 provides a direct route to product 4a via product enamine 11, providing a role for the proton R = H and a rationalization for a difference in rates when R = D. Diastereoselective protonation of 11 is supported by the observation of a selective relationship between E-11 and the major enantiomer of 4a as demonstrated by EXSY-NMR.7 However, when R = Me, species 10 is necessarily a parasitic off-cycle species because no product enamine species 11 is possible, and therefore the route to product 4b through 10 is blocked. The sluggishness of the rate is attributed to the fact that most of the catalyst is sequestered in an off-cycle reservoir 10, as generalized in Scheme 8. Puzzle 1 also asked why the reaction of Scheme 3 does not exhibit the stark difference in reactivity observed in comparing linear and α,α-disubstituted aldehydes when DEAD is employed as the electrophile. Once again, kinetic features D
DOI: 10.1021/acs.accounts.5b00394 Acc. Chem. Res. XXXX, XXX, XXX−XXX
Article
Accounts of Chemical Research Puzzle 2: α-Chlorination
Scheme 9. Curtin−Hammett Scenario for Rationalization of Enantioselectivity in a Reversible Network
Turning to the second set of puzzles, those raised by the αchlorination reaction of Scheme 4, we now consider the role of the downstream intermediate species identified in this case as one pair of syn and anti isomers, 12. We found that formation of this species is general for Cl−Y electrophiles where Y = succinimide or phthalimide with a variety of catalysts including 3, 7, 8, and diarylprolinol ether 15 (3 with Ph groups replaced by Ar = 3,5-(CF3)2-C6H4-). Carrying out the catalytic reaction of Scheme 4 using 3, we obtained 42% ee (Table 2, entry 1),7a a Table 2. Diastereomeric Ratios of 12 and Enantiomeric Ratios of 6a7a
1 2 3 4 5 6 7
work), oxazines are more stable than the corresponding cyclobutanes and are the resting state on the catalytic cycle. Indeed, that proposal is not in conflict with, but rather is in support of, the key concepts we describe here: they report a stable intermediate species formed reversibly downstream from the stereogenic bond-forming step that lies on the cycle. As in our model, their proposal indicates that the relative stability and reactivity of downstream species dictates enantioselectivity, rather than the encounter at the enamine−electrophile interaction, where even poor selectivity could result in high ultimate enantiomeric excess. Thus, the general scenario of Scheme 9 applies to Pihko’s mechanism for reactions of α-alkyl substituted nitroalkenes. Benchmark Reaction. These findings suggest that for many catalyst candidates, the “benchmark” reaction of Scheme 3a may present complications beyond the simple model I of Scheme 2. The results of this reaction may not in fact provide an assessment of the selectivity of the enamine−electrophile interaction. In such a case, designing an effective catalyst for the reaction of Scheme 3a should involve consideration of how the catalyst structure may affect the relative stability of potential diastereomers of 10. While some of the same factors that apply to Scheme 2 (assessing the approach of the electrophile to one face or the other of the enamine) may also apply to assessing the relative stability of these diastereomers, this need not be the case a priori, since the former invokes kinetics while the latter thermodynamics. If an assessment of catalyst’s selectivity in the stereogenic bond-forming step itself is the primary task of a benchmark reaction, a more appropriate candidate for such a benchmark would be a reaction in which this step is irreversible. Kinetic profiles can provide a rapid means of assessing the suitability of a reaction as a potential benchmark. Observation of overall zero order kinetics, as in the three cases of Figure 1, is a hallmark of a reaction exhibiting a downstream resting state, implying that factors subsequent to formation of the stereogenic bondforming step may play a role. By contrast, observation of overall positive order kinetics suggests that the resting state lies at either the free catalyst or the enamine, both of which indicate that the enamine−electrophile interaction is likely to be irreversible and thus key to enantioselectivity.
catalyst
Y
dr,a syn-12/anti-12
er,b R-6a/S-6a
3 15 8 7 3 15 8
succinimide succinimide succinimide succinimide phthalimide phthalimide phthalimide
30:70 16:84
Explaining Anomalies in Enamine Catalysis: “Downstream Species” as a New Paradigm for Stereocontrol Jordi Burés,†,‡ Alan Armstrong,† and Donna G. Blackmond*,†,‡ †
Department of Chemistry, The Scripps Research Institute, La Jolla, California 92037 United States Department of Chemistry, Imperial College London, London SW72AZ, U.K.
‡
CONSPECTUS: Enantioselective organocatalysis by diarylprolinol ethers is remarkably selective and efficient for a wide range of transformations involving a number of different activation modes, including HOMO activation via enamines. A simple steric model based on facial discrimination in the attack of an enamine on an electrophile has been invoked to rationalize high enantioselectivity. In a number of reactions, however, experimental observations have persistently left us with mechanistic puzzles that fail to fit neatly into this simple picture. Further studies involving both kinetic profiling of reaction networks and NMR spectroscopic characterization of the structures of intermediate species helped us to address these puzzles. This work led to the proposal of a new paradigm for stereocontrol in asymmetric aminocatalysis, demonstrating that the ultimate stereochemical outcome may not, in fact, be determined solely in the stereogenic bond-forming step between enamine and electrophile. The identification of stable species occurring downstream from the addition of an electrophile to an enamine, and the discovery of kinetic features that are diagnostic of the presence of such species, allows development of a new mechanistic framework that reveals a hierarchical selection. Both kinetic and thermodynamic processes associated with downstream intermediates can exert an influence on the ultimate enantioselectivity. Interestingly, the role of these species may be either to enhance or to erode selectivity established at the enamine−electrophile step. The reversibility of steps preceding and subsequent to the stereogenic bond-forming step is an important factor, as are reaction parameters that may stabilize or destabilize intermediates, including the nature of the electrophile counterion and the solvent. These concepts hold implications for the future design and optimization of asymmetric catalytic processes, because such design does not necessarily feature the same parameters at the second hierarchical level as it does at the first. Examples presented to highlight these issues include the conjugate addition of aldehydes to nitroalkenes and the α-chlorination and α-selenylation of aldehdyes using diarylprolinol ether catalysts commonly assumed to follow a steric model for enantioselectivity. While the new paradigm for stereocontrol involving downstream intermediates is developed here for enamine catalysis, the same concepts may hold for other organocatalytic modes of activation.
■
INTRODUCTION
been, there remain puzzling experimental observations from a number of different reactions that are not easily explained by the model shown in Schemes 1 and 2. Indeed, our detailed mechanistic studies6−8 of a number of reactions using these catalysts revealed general experimental features that suggest a more complex mechanism may operate for enamine catalysis by diarylprolinol ethers. In some cases, we found that enantioselectivity is ultimately determined not at the step of electrophile−enamine interaction but instead further along the catalytic cycle after the initial stereogenic bond-forming step (in steps not elaborated in Scheme 1), leading us to highlight the importance of what we term “downstream intermediate species”. The concept holds important implications for the design and optimization of reactions based on diarylprolinol
The development of a wide range of highly selective transformations using pyrrolidine-based organocatalysts has led to their elevation as “privileged” structures in asymmetric catalysis.1 The diarylprolinol ethers introduced independently by the groups of Jørgensen2 and Hayashi3 enjoy a particularly broad usage because of their ability to employ several different activation modes.4 Scheme 1 shows the conventionally accepted mechanism for enamine activation. Diarylprolinol ethers contain a bulky catalyst side chain designed to offer excellent facial differentiation, illustrated in Scheme 2 for enamine intermediate I. Scheme 2 shows that enantiocontrol in the steric model I is in the opposite sense as found for proline enamine II directed by Bronsted acid cocatalysis. The simple steric model I rationalizes the experimental results in a remarkably wide variety of reactions catalyzed by diarylprolinol ethers.5 However, as successful as the model has © XXXX American Chemical Society
Received: August 25, 2015
A
DOI: 10.1021/acs.accounts.5b00394 Acc. Chem. Res. XXXX, XXX, XXX−XXX
Article
Accounts of Chemical Research Scheme 1. Enamine Mechanism
Scheme 3. Diarylprolinol Ether Catalyzed Reactions
Scheme 2. Transition State Models
chlorination (Scheme 4, Table 1),11 and MacMillan reported high enantioselectivities using imidazolidinone catalysts such as
ethers, and it should be considered that such mechanistic nuances might apply to other catalyst systems. Our investigations of the conjugate addition of aldehydes to nitroalkenes6,7 and the α-chlorination7 and α-selenylation8 of aldehydes employed kinetic profiling by in situ reaction calorimetric monitoring along with a battery of NMR spectroscopic tools for identification of catalytic intermediates. We begin by describing the mechanistic puzzles that led us to probe for insights beyond the conventional steric model I, followed by a review of our results leading to a new paradigm for stereocontol supporting a role for “downstream intermediate species” in enamine catalysis.
Scheme 4. Organocatalyzed α-Chlorination of Aldehydes
■
BACKGROUND Our first puzzle arises from studies of the conjugate addition of aldehydes to nitroalkenes, transformations in which using diarylprolinol ether catalysts are particularly efficient with linear aldehydes as substrates (Scheme 3a). This reaction has become so widely employed that it has been termed a “benchmark” for assessing new chiral amine catalysts.9 However, when α,α-disubstituted aldehydes are used as substrates, the results are more complex, as is also shown in Scheme 3. Reaction rate is virtually completely suppressed for α,α-disubstituted aldehydes, a fact that remains unexplained by the cycle of Scheme 1, especially because a similar sluggishness is not observed for α,α-disubstituted aldehydes in reactions with other electrophiles, such as α-amination with DEAD,6b,10 as shown in Scheme 3b. Puzzle 1. Does the cycle of Scheme 1 with model I of Scheme 2 (shown with a general electrophile, E) explain why the reaction rate is nearly completely suppressed in reactions of α,α-disubstituted aldehydes with E = nitroalkenes but not with E = DEAD? A second puzzle arises from the highly vaunted generality of the diarylprolinol ethers for enamine catalysis of a wide range of electrophiles. In work mainly by the groups of Jørgensen and Hayashi, these catalysts have been shown to give excellent enantioselectivity not only in conjugate addition to nitroalkenes and α-aminations as shown in Schemes 2 and 3 but also in αaddition to vinyl ketones, Mannich reactions, α-sulfenylations, α-fluorinations, and α-brominations.4 Jorgensen had shown that 2,5-diphenylpyrrolidine 8 is an excellent catalyst for α-
Table 1. α-Chlorination Reactions
a
catalyst
substrate
“Cl+”
solvent, T (°C)
% ee
ref
7 7 8 3
1c 1c 1d a
2c 2d 2c a
CHCl3, 4 CHCl3, −30 CH2Cl2, 0 a
19 92 95 a
12 12 11 a
Not reported.
7 using a quinone-based chlorination agent but not with NCS12 (Scheme 4, Table 1). The lack of a report for α-chlorination using diarylprolinol ethers is puzzling, given the success of this and other α-halogenations for other pyrrolidine catalysts. Puzzle 2. Does the cycle of Scheme 1 with model I of Scheme 2 explain why α-chlorination has not been reported using diarylprolinol ether catalysts, given the success of other α-halogenations with catalysts thought to obey model I? Can model I account for dif ferent performance f rom dif ferent chlorinating agents as electrophiles? A third puzzle addresses the role of the reaction medium in model I. High enantioselectivities in the α-selenylation of aldehydes using diarylprolinol ethers has been reported separately by the groups of Melchiorre13 and Cordova14 (Scheme 5). Interestingly, Cordova and co-workers14 found a B
DOI: 10.1021/acs.accounts.5b00394 Acc. Chem. Res. XXXX, XXX, XXX−XXX
Article
Accounts of Chemical Research Scheme 5. α-Selenylation of Aldehydes14
Scheme 6. Identification of Catalyst Resting States in the Reactions of Schemes 3, 4, and 5
striking solvent effect in the reaction: they reported an inversion in the sense of enantioselectivity when the solvent was switched from toluene (95% ee) to acetonitrile (−30% ee). The authors noted that while the reaction medium may alter selectivities, this reversal was unexpected. The fundamental mechanism relating the solvent to model I of Scheme 2 remains unclear. Puzzle 3. Does the cycle of Scheme 1 with model I of Scheme 2 rationalize an inversion in enantioselectivity caused simply by changing the solvent? These three puzzles were addressed in a series of kinetic and mechanistic studies summarized below.
■
RESULTS AND DISCUSSION
Kinetic Profiles
An initial clue to solving the three puzzles presented above came from kinetic studies of the reactions in Schemes 3a, 4, and 5. Figure 1 reveals that all three reactions exhibit kinetic profiles structures of the intermediates identified in each case. In the conjugate addition reactions of Scheme 3a, a single cyclobutane diastereomer 10 accounts for virtually all of the catalyst for both linear and α,α-disubstituted aldehydes.6,7 Under reaction conditions for linear aldehydes, 10 gives way at very high conversion to the product enamine E-11. It should be noted that 11 is unable to form in the case of the α,α-disubstituted aldehydes. In the α-chlorination and α-selenylation reactions, we found the catalyst resting state to be a pair of diastereoisomeric intermediates that persist throughout the reaction. Covalent addition of 2c across the enamine double bond in αchlorination yields one rapidly equilibrating syn/anti pair, 12, as the resting state. The connectivity but not the stereochemistry of these species was established by low temperature NMR.7a In α-selenylation, product enamine species E-13 and Z-13 are present in approximately equal concentrations in reactions carried out in CH2Cl2, while in toluene only E-13 is observed.8 Note that in each case, the resting state contains both the aldehyde and the electrophile substrates, as predicted by the zero order kinetics in the reaction profiles of Figure 1. Once the predominant catalytic species in these reactions have been identified in each case, the next mechanistic task is to demonstrate whether the species lies on the catalytic cycle or is simply a spectator. We treat each of the “puzzles” raised above in turn to probe the role of these species in each reaction.
Figure 1. Kinetic profiles via reaction calorimetry of three reactions catalyzed by 3:6−8 (a, magenta) Scheme 3a, conjugate addition, 1a (1.2 M), 2a (1.0 M), 3 (0.1 M) CH3COOH (0.025 M), toluene, 25 °C; (b, green) Scheme 4, α-chlorination, 1d (0.63 M), 2c (0.48 M, 3 (0.048 M), CH3COOH (0.01 M), CHCl3, −11.4 °C; (c, blue) Scheme 5, αselenylation. 1d (0.075 M), 2e (0.025 M), 3 (0.0025 M), CH2Cl2, 25 °C.
that are notable for their similar and highly unusual features.15 In each case shown in Figure 1, a rapid initial rate was followed by a regime of zero order kinetics in both aldehyde and electrophile concentrations. Observation of a regime of constant rate is typically attributed to “saturation kinetics” in both substrates, suggesting that the rate-determining step in the cycle for each reaction involves a catalyst resting state that contains both substrates. The rapid initial rate is characteristic of a transient period that occurs while the concentration of this resting state is building up.
Puzzle 1: Conjugate Addition to Nitroalkenes
While it was originally suggested that the cyclobutane species 10, which was identified simultaneously by us and by a collaboration of the groups of Hayashi and Seebach,16 is a “parasitic” off-cycle species, a number of observations taken together led us to propose that for the case of linear aldehydes, both 10 and E-11 lie on the catalytic cycle.
Downstream Intermediates
Our mechanistic studies then turned to identification of the catalyst resting state highlighted in each of these reactions, which was accomplished via a detailed series of variable temperature and 2-D NMR experiments. Scheme 6 shows the C
DOI: 10.1021/acs.accounts.5b00394 Acc. Chem. Res. XXXX, XXX, XXX−XXX
Article
Accounts of Chemical Research First, enamine formation was shown to be rapid and reversible for both linear and α,α-disubstituted aldehydes, confirming that the suppression of product formation for the latter must occur at a step further along in the cycle. Second, crossover experiments demonstrated that electrophile addition is also rapid and reversible. Formation of the cyclobutane resting state 10 proceeds rapidly for either R = Ph or R = H, but its formation appears to signal a dead end for reaction of R = Ph but not R = H. Third, the importance of R = H is further highlighted by the finding of a small normal deuterium KIE using α-deuterated aldehyde 1a. Given that the ratedetermining step occurs at some step after rapid, reversible enamine formation, a role for this aldehyde proton is difficult to envision in the cycle of Scheme 1. To reconcile these observations, we proposed the reaction network shown in Scheme 7, highlighting the role of the downstream species in red.
Scheme 8. Rationalization for Substrate Reactivity Differences in the Reactions of Scheme 2
Scheme 7. Catalytic Cycle for the Conjugate Addition to Nitroalkenes of Scheme 3a
provide a clue: reactions of DEAD with 1a and 1b both show first order kinetics in aldehyde concentration and zero order kinetics in DEAD concentration. A simple explanation for such behavior is that enamine formation is the rate-determining step when DEAD is the electrophile. The free catalyst is the resting state, and no stable downstream species exist under catalytic conditions. This electrophile thus provides a classic example of model I, with the effectiveness of the catalyst dictated by the irreversible reaction of enamine with electrophile as illustrated in Scheme 2. The complexities uncovered in the mechanism of the reactions of Scheme 3 require further discussion in the context of model I of Scheme 2. Indeed, the observed reversibility of the enamine−electrophile interaction17 raises the possibility, at least theoretically, that high product enantioselectivity could ultimately be achieved even if selection at the enamine− electrophile level is very poor. A Curtin−Hammett scenario17 may be outlined where imperfect selectivity of enamine addition to electrophile 2a leads to reversible formation of two (or more) hypothetical diastereomeric cyclobutane intermediates in rapid equilibrium, with enantioselectivity being a function of the relative stability of the intermediates and their relative reactivity onward to product (via product enamines). This is illustrated in Scheme 9. This scenario presents an alternate view of key catalyst design parameters compared with model I of Scheme 2. A highly selective catalyst in the reaction of Scheme 3a need not be selective in the enamine−electrophile interaction, if either the stability of any alternate cyclobutane formed is low (Kminor is small) or its productive rate constant kminor is small. The reversibility of the network offers a “second chance” to any system exhibiting poor selectivity at the stereogenic bondforming step.18 High product enantiomeric excess can be achieved simply by channeling the reaction to a single highly stable downstream species that leads to a single product. Other Mechanistic Proposals. Alternatives to the catalytic cycle through cyclobutane 10a have been proposed. Seebach and co-workers19 initially questioned this route on the intuitive assumption that the elementary steps from the cyclobutane to the product enamine (through fleeting species, an enaminonitronate or a 1,2-oxazine, not shown in Scheme 7) would be energetically uphill; later calculations led them to retract this viewpoint.19b,20 Pihko and co-workers21 suggested that in the case of α-alkyl substituted nitroalkenes (not studied in our
Scheme 7 incorporates all of the experimental clues noted above to suggest that with R = H, the cyclobutane resting state 10 provides a direct route to product 4a via product enamine 11, providing a role for the proton R = H and a rationalization for a difference in rates when R = D. Diastereoselective protonation of 11 is supported by the observation of a selective relationship between E-11 and the major enantiomer of 4a as demonstrated by EXSY-NMR.7 However, when R = Me, species 10 is necessarily a parasitic off-cycle species because no product enamine species 11 is possible, and therefore the route to product 4b through 10 is blocked. The sluggishness of the rate is attributed to the fact that most of the catalyst is sequestered in an off-cycle reservoir 10, as generalized in Scheme 8. Puzzle 1 also asked why the reaction of Scheme 3 does not exhibit the stark difference in reactivity observed in comparing linear and α,α-disubstituted aldehydes when DEAD is employed as the electrophile. Once again, kinetic features D
DOI: 10.1021/acs.accounts.5b00394 Acc. Chem. Res. XXXX, XXX, XXX−XXX
Article
Accounts of Chemical Research Puzzle 2: α-Chlorination
Scheme 9. Curtin−Hammett Scenario for Rationalization of Enantioselectivity in a Reversible Network
Turning to the second set of puzzles, those raised by the αchlorination reaction of Scheme 4, we now consider the role of the downstream intermediate species identified in this case as one pair of syn and anti isomers, 12. We found that formation of this species is general for Cl−Y electrophiles where Y = succinimide or phthalimide with a variety of catalysts including 3, 7, 8, and diarylprolinol ether 15 (3 with Ph groups replaced by Ar = 3,5-(CF3)2-C6H4-). Carrying out the catalytic reaction of Scheme 4 using 3, we obtained 42% ee (Table 2, entry 1),7a a Table 2. Diastereomeric Ratios of 12 and Enantiomeric Ratios of 6a7a
1 2 3 4 5 6 7
work), oxazines are more stable than the corresponding cyclobutanes and are the resting state on the catalytic cycle. Indeed, that proposal is not in conflict with, but rather is in support of, the key concepts we describe here: they report a stable intermediate species formed reversibly downstream from the stereogenic bond-forming step that lies on the cycle. As in our model, their proposal indicates that the relative stability and reactivity of downstream species dictates enantioselectivity, rather than the encounter at the enamine−electrophile interaction, where even poor selectivity could result in high ultimate enantiomeric excess. Thus, the general scenario of Scheme 9 applies to Pihko’s mechanism for reactions of α-alkyl substituted nitroalkenes. Benchmark Reaction. These findings suggest that for many catalyst candidates, the “benchmark” reaction of Scheme 3a may present complications beyond the simple model I of Scheme 2. The results of this reaction may not in fact provide an assessment of the selectivity of the enamine−electrophile interaction. In such a case, designing an effective catalyst for the reaction of Scheme 3a should involve consideration of how the catalyst structure may affect the relative stability of potential diastereomers of 10. While some of the same factors that apply to Scheme 2 (assessing the approach of the electrophile to one face or the other of the enamine) may also apply to assessing the relative stability of these diastereomers, this need not be the case a priori, since the former invokes kinetics while the latter thermodynamics. If an assessment of catalyst’s selectivity in the stereogenic bond-forming step itself is the primary task of a benchmark reaction, a more appropriate candidate for such a benchmark would be a reaction in which this step is irreversible. Kinetic profiles can provide a rapid means of assessing the suitability of a reaction as a potential benchmark. Observation of overall zero order kinetics, as in the three cases of Figure 1, is a hallmark of a reaction exhibiting a downstream resting state, implying that factors subsequent to formation of the stereogenic bondforming step may play a role. By contrast, observation of overall positive order kinetics suggests that the resting state lies at either the free catalyst or the enamine, both of which indicate that the enamine−electrophile interaction is likely to be irreversible and thus key to enantioselectivity.
catalyst
Y
dr,a syn-12/anti-12
er,b R-6a/S-6a
3 15 8 7 3 15 8
succinimide succinimide succinimide succinimide phthalimide phthalimide phthalimide
30:70 16:84
