Curtin−Hammett Paradigm for Stereocontrol in Organocatalysis
by Diarylprolinol Ether Catalysts
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 SW7 2AZ, U.K.
*S Supporting Information
ABSTRACT: Detailed mechanistic study of two reactions
catalyzed by diarylprolinol ether catalysts, the conjugate addition
of aldehydes to nitro-oleﬁns and the α-chlorination of
aldehydes, leads to the proposal that the stereochemical outcome
in these cases is not determined by the transition state of the step
in which the stereogenic center is formed from enamine attack on
the electrophile but instead is correlated with the relative stability
and reactivity of diastereomeric intermediates downstream in the
catalytic cycle. This combination of kinetic and thermodynamic
factors illustrates a remarkable Curtin−Hammett scenario that
can result in either an enhancement or an erosion of the selectivity that would be predicted by the transition state for enamine attack
on the electrophile. Evidence is oﬀered to suggest that this concept may represent a general phenomenon for pyrrolidine-based
catalysts lacking an acidic directing proton. Implications for catalyst and reaction design are discussed.
■ INTRODUCTION
Since the contemporaneous reports in 2000 of the
intermolecular aldol reaction catalyzed by proline1
and the
imidazolidinone-catalyzed Diels−Alder reaction,2
research in
the area of asymmetric aminocatalysis has burgeoned, with a
variety of eﬃcient and selective catalysts being introduced for
large number of transformations. Mechanistic models aiming to
rationalize stereoselectivity for reactions thought to follow
an enamine mechanism have invoked a role for Bronsted acid
co-catalysis in the case of proline and related catalysts, supported
by theoretical calculations in what is now known as the Houk−
List model (I, Scheme 1).3
A complementary model for catalysts
lacking a directing proton proposes that selectivity is determined
by the steric bulk of catalyst substituents (II, Scheme 1).4
Opposite product stereochemistry is predicted in the two models
for the reaction of enamines with X = Y electrophiles.
The results we present in this work lead to the development
of a general concept for selectivity in asymmetric aminocatalysis
that incorporates additional factors not considered in the
transition state models shown in Scheme 1. We document two
examples of reactions in which stable, equilibrated intermediate
species appear in the catalytic cycle downstream from the
stereogenic center-forming step. This allows us to propose a
selectivity paradigm invoking Curtin−Hammett5
control, modulated
by the relative stability of diastereomeric intermediate
species. The concept is discussed in the context of the conjugate
addition of linear aldehydes to nitro-oleﬁns and the α-chlorination
of aldehydes, both catalyzed by pyrrolidine-based systems
that are thought to follow the stereochemical model II. The
general implications of such a selectivity paradigm in asymmetric
organocatalysis both for mechanistic understanding and for catalyst
and reaction design are discussed.
■ RESULTS AND DISCUSSION
Conjugate Addition. Since their introduction by Hayashi6
and by Jorgensen,7
diarylprolinol ethers such as 3 have enjoyed
spectacular success in a variety of enamine- and iminium-based
organocatalytic transformations.4
Most notably, they ﬁgure prominently
in cascade sequences where Michael additions typically
appear as the ﬁrst transformation in combinations such as
En-Im-En consecutive cycles.8
In addition, the conjugate addition
of linear aldehydes to nitro-oleﬁns (Scheme 2, top) has
become a benchmark reaction for probing the eﬃciency and
selectivity of new aminocatalysts.
While model II oﬀers a rationale for the sense of the stereochemical
outcome of the attack of the enamine on electrophiles
such as 2a, several features of the reactions shown in Scheme 2
raise questions that are diﬃcult to rationalize solely by the
proposed steric arguments. First, reactions of linear aldehydes
Received: January 13, 2012
Published: March 27, 2012
Scheme 1. Transition-State Models for the Stereogenic
Center-Forming Step in Enamine Catalysis
Article
pubs.acs.org/JACS
© 2012 American Chemical Society 6741 dx.doi.org/10.1021/ja300415t | J. Am. Chem. Soc. 2012, 134, 6741−6750
with nitro-oleﬁns proceed not only with high enantioselectivity
but also with a high level of facial selectivity with respect to the
nitro-oleﬁn electrophile, a feature that is not readily explained
by model II. Selectivity typically follows Seebach’s topological
rule, and it has been conceded that the actual mechanism may
be more complicated.9
Facial selectivity in this reaction using
primary amine thiourea catalysts has been explained by
invoking enzyme-like eﬀects of approximation and hydrogen
bonding of the nitroalkene to the catalyst,10
but such eﬀects are
not applicable to catalyst systems following model II.
Second, while the reaction of 2a with 1a is extremely eﬃcient,
reaction with 1b (Scheme 2, bottom) is extremely sluggish.
Enamine formation between 3 and either 1a or 1b occurs in
minutes under these conditions,11
and near quantitative conversion
to 4a occurs on the order of minutes, but only minimal
formation of product 4b is observed even after 24 h. This stark
diﬀerence in reactivity is diﬃcult to explain solely by the increase
in steric bulk aﬀorded by exchange of R2
= H for R2
= CH3 in an
enamine reacting with an X = Y electrophile as envisioned in
model II.
We uncovered a further mechanistic puzzle in studies of
deuterium isotope eﬀects in the reaction of propanal 1a with
nitrostyrene 2a catalyzed by 3. Figure 1 shows that kinetic
isotope studies employing deuterated acetic acid gave a normal
isotope eﬀect, albeit of small magnitude, with kH/kD = 1.28.
Both our work12
and that of Hayashi and Seebach13
have quantiﬁed
rate acceleration upon addition of acid in this reaction, in
accordance with many reports showing that addition of acid leads
to higher yields. Surprisingly, as also shown in Figure 1, employing
the deuterated substrate d2
-1a (deuterated at the α-H positions)
together with CD3COOD gave a much stronger isotope eﬀect, for a
cumulative kH/kD = 3.25. Although enamine formation with catalyst
3 is accelerated by acid, this step is ruled out as rate-determining in
the reactions of Scheme 2 because of the observation of zero-order
kinetics in [1a]. Thus observation of such an isotope eﬀect in a
substrate that undergoes non-rate-determining deprotonation is
diﬃcult to reconcile with a reaction known to be accelerated by
acid. In addition, all reactions regardless of deuterium content
exhibit identical rapid initial rates, accounting, as previously
reported, for ca. 1 turnover of the catalyst prior to a transition to
a slower regime of overall zero-order kinetics.12
NMR spectroscopy provides further clues concerning the
role of protonation and deprotonation of intermediates. Our
recent studies12
of this reaction and those of Hayashi and
Seebach13
using catalyst 3 identiﬁed stable cyclobutane species
5 (Scheme 3) existing as a single diastereomer as the resting
state in the system, resulting from reversible attack of the
enamine on the nitro-oleﬁn. Further NOESY experiments
carried out on the cyclobutane species 5a (R1
= CH3; R2
= Ph)
prepared from 1a, 2a, an excess of 3, and molecular sieves (to
suppress hydrolysis to product 4a) revealed EXSY cross-peaks
for a minor species 6 in equilibrium with the cyclobutane
(Figure 2). The same species was detected, albeit to a slightly
lesser extent, during the catalytic reaction under standard
reaction conditions.11
Importantly, no EXSY cross-peaks were
detected for the stable cyclobutane species 5b derived from the
α-branched aldehyde 1b with a methyl group in place of the
proton at the 2′ position (Scheme 4).11
Scheme 2. Relative Reactivity and Selectivity of 1a and 1b in
Reactions with 2a Catalyzed by 311
Figure 1. Deuterium isotope eﬀects monitored as fraction conversion
vs time for the reaction of 1a with 2a catalyzed by 3 in toluene at 25 °C
in the presence of normal and deuterated acetic acid and normal
and deuterated 1a. Reaction conditions: [1a]0 and [d2
-1a]0 = 1.2 M;
[2a]0 = 1.0 M; [3] = 0.1 M; [CH3COOH] or [CD3COOD] = 0.1 M;
[H2O] or [D2O] = 0.5 M.11
Scheme 3. Reversible Formation of Cyclobutane 512
Figure 2. NOESY experiment of 5a identifying EXSY cross-peaks for
new species 6 as shown in Scheme 4. Peak numbers refer to carbon
atoms of 5a (blue) and 6 (red) in Scheme 4.11
Journal of the American Chemical Society Article
dx.doi.org/10.1021/ja300415t | J. Am. Chem. Soc. 2012, 134, 6741−67506742
The observed EXSY cross-peaks allow a comparison of the shifts
of the signals for this new species to those of the cyclobutane
5a.11
No EXSY correlation with the new species is observed for
the proton at 2′ on 5a, suggesting that 6 lacks this proton. In addition,
the protons attached to the 1′ and 4′ carbon atoms of 5a are
the most aﬀected in the exchange experiment. This leads to the
proposal for the identity of this new minor species as 6 shown in
Scheme 4, where the 1′ and 2′ carbon atoms change in hybridization
from sp3
to sp2
upon formation of the enamine nitronate 6 after
deprotonation at 2′. Such a species cannot be formed from
cyclobutane 5b, which explains the lack of EXSY cross-peaks in this
case. Although 6 cannot be isolated, the EXSY correlation provides
strong support for its intermediacy between 5a and 7, both of which
species have been characterized and reported previously by us.12
The signiﬁcance of the presence (or lack thereof) of a proton at
the 2′ position of cyclobutane 5 extends to the catalytic and kinetic
isotope results. While it has been suggested that this species is a
“parasitic” oﬀ-cycle reservoir, we may rationalize the isotope eﬀects
of both CD3COOD and d2
-1a by proposing that deprotonation
of the resting state 5a to form the minor species 6 occurs on
the catalytic cycle, followed by protonation to give the product
enamine, which is formed exclusively as E-7, as shown in Scheme 5.
All catalytic species shown in the modiﬁed catalytic cycle presented
in Scheme 5 have been observed experimentally. Some of the steps
shown may represent sequences of elementary reactions combined
into one kinetically meaningful step.14
As we have shown previously,
the fast initial rate (Figure 1, dashed box) represents saturation
of species 5a in steps that rapidly become equilibrated, explaining
the insensitivity of initial rate to deuterium isotope eﬀects.
The subsequent deprotonation and protonation steps shown in
Scheme 5 both contribute to the overall observed kinetics of the
zero-order regime that dominates the remainder of the reaction, and
the isotope eﬀect (kH
)obs/(kD
)obs will exhibit a complex dependence
on the elementary rate constants associated with these steps.15
The proposal of a highly selective reaction Pathway through
enamine 7 may at ﬁrst appear to be counterintuitive, given that
this step destroys the stereocenter formed in the addition step.
However, our observation of enamine E-7 exclusively, and the
fact that the syn:anti product ratio remains constant during
reaction turnover, are in accord with the suggestion that a
stereospeciﬁc relationship exists between E-7 and the major syn
product. We recently demonstrated stereospeciﬁc enamine
formation between 3 and 2-phenyl propionaldehyde.16
EXSY
experiments carried out on reaction of the thermodynamic
syn:anti product mixture with 3 demonstrate that E-7 correlates
preferentially with the syn adduct, as shown in Figure 3.11
We have shown that the syn:anti product ratio erodes when
product 4a remains in contact with the catalyst for an extended
period after reaction turnover is complete and the free catalyst
is no longer sequestered as 5a (Scheme 6). Figure 4 shows that,
when the product with syn:anti = 93:7 is mixed with 3, a low
steady-state concentration of enamine E-7 is rapidly established,
Scheme 4. Assignment of Intermediate Species 6 from EXSY
Shifts11
Scheme 5. Modiﬁed Catalytic Cycle Proceeding through 5a,
6, and 7a
a
All catalytic species in the cycle have been observed experimentally.
Figure 3. 1
H NMR spectrum of an equilibrated mixture of 4a (0.87 M,
syn:anti = 60:40) with 3 (0.36 M) and acetic acid (0.66 M) forming
E-7 at 25 °C. Inset: EXSY peaks from irradiation of E-7 (oleﬁnic proton)
using a 1-D GOESY sequence (600 MHz ﬁeld) at diﬀerent mixing
times τ as shown.
Scheme 6. Reversible Enamine Formation between 3 and
syn-4a and anti-4a
Journal of the American Chemical Society Article
dx.doi.org/10.1021/ja300415t | J. Am. Chem. Soc. 2012, 134, 6741−67506743
and equilibration between syn-4a and anti-4a continues until
the thermodynamic ratio of ca. 60:40 is reached.17
The equilibrium
constants for enamine formation from syn and anti, Keq,S
and Keq,A, may be estimated from the ﬁnal concentrations in
Figure 4, as shown in Scheme 6 and eq 1.
= =
‐
‐
=
−
K
k
k
E
syn
7
4a 3[H O]
[ ]
[ ] [ ]
3.2eq,S
s
s 2
eq
eq eq (1a)
= =
‐
‐
=
−
K
k
k
E
anti
7
4a 3[H O]
[ ]
[ ] [ ]
4.9eq,A
a
a 2
eq
eq eq (1b)
Non-stereoselective protonation of E-7 would exhibit identical
rates for forming syn-4a and anti-4a, or a selectivity ratio s =
k−s/k−a = 1. However, as shown in Figure 5, the experimentally
observed temporal change in the syn:anti ratio cannot be accurately
described for non-selective enamine protonation. Kinetic model-
ing11,18
of the process shown in Scheme 6 accommodates the
experimental data with a selectivity ratio of s > 39, revealing
that enamine protonation is a highly selective process. Since
protonation of 6 is an irreversible step on the catalytic cycle, the
stereochemistry at C-2′ that was set in the addition step in
Scheme 5 is preserved in the initial formation of product 4a
through stereoselective product enamine formation and
protonation.
The modiﬁed cycle proposed in Scheme 5 allows us to
address the features of the reactions of Scheme 2 that were
diﬃcult to explain solely by the steric model II. First, the route
through cyclobutane 5a provides a rationalization of the high
facial selectivity observed with regard to nitrostyrene,
attributing this to the high stability of the single diastereomer
5a and to the stereospeciﬁcity of its reaction, rather than calling
simply on speciﬁcity in the approach of the nitro-oleﬁn to the
enamine. The enhanced stereocontrol aﬀorded by the reaction
pathway through the stable species 5a provides an alternative to
Seebach’s acyclic synclinal model9
invoking a transition state
with favorable electrostatic interactions between the nitrogen of
the enamine 8 and the nitro group of 2a.
Scheme 5 also provides an explanation of the second curious
feature noted earlier concerning the near-inertness of αbranched
aldehydes such as 1b in the conjugate addition with
nitrostyrene 2a. This may be rationalized by the fact that these
reactions cannot proceed by the newly proposed pathway
through species 6 and 7; the lack of a proton at the 2′ position
precludes the possibility of formation of enamine 7 from 5b. As
in the case of linear aldehydes, cyclobutane 5b is formed rapidly
and remains the catalyst resting state; however, in the reaction
of 1b, cyclobutane 5b exists as a true oﬀ-cycle reservoir that
sequesters most of the catalyst. Therefore, the concentration
active on-cycle species is vanishingly small, resulting in a severe
suppression of the rate of turnover of 1b.
The role of the cyclobutane intermediate 5a in the catalytic
cycle described in Scheme 5 has implications for the proposed
mechanistic models of Scheme 1. In any competitive reaction
network, selectivity is determined by ΔΔG⧧
, the diﬀerence in
energies of the relevant transition states. Our results suggest
that the transition state relevant for the network of Scheme 5
lies subsequent to the step in which the stereogenic center is
formed, providing a counterpoint to the classic transition-state
model II. Figure 6 compares model II (Pathway A) to the novel
proposed mechanistic pathway through 5a (Pathway B). For
this theoretical comparison, the two pathways are shown as
exhibiting the same ΔΔG⧧
. Numbered intermediates are based
in Scheme 5. Figure 6b shows parallel networks for products 4a
and 4a′, which are enantiomeric at the C-2 stereocenter formed
in the addition step.
It is widely assumed in enamine-based organocatalysis for
both models I and II shown in Scheme 1 that product
enantioselectivity is determined as shown in Pathway A (Figure 6a).
Product enantiomeric ratio (er) in this classic case is given
by the ratio of rate constants for an irreversible reaction step
between enamine 8a and electrophile 2a, or k2/k2′ (Figure 6c).
If, however, stable intermediates such as 5a are reversibly
formed on the cycle subsequent to the stereogenic centerforming
stepas in Scheme 5 and Pathway B in Figure 6all
preceding intermediates appearing in the catalytic cycle become
equilibrated, because 5a serves as the resting state, or what may
be termed the “hold-up” point in the cycle. In Pathway B the
relevant transition state for determining product er is not the
stereogenic center-forming step shown in model II; rather,
product er depends on two separate factors revealed in the
equation in Figure 6c: (i) a kinetic component given by the
relative rate constants for the ﬁrst irreversible step and (ii) a
thermodynamic component given by the relative equilibrium
constants for all the equilibrated diastereomeric intermediates
appearing prior to the ﬁrst irreversible step in the mirror
image networks. Thus, Figure 6a shows that the ΔΔG⧧
value
determining product selectivity via Pathway B derives from a
combination of the relative heights of the transition states for
the ﬁrst irreversible step and the relative depths of the energy
wells for the preceding intermediates.
Selectivity in Pathway B can diﬀer from that in Pathway A
because the relative rates of electrophile attack by the common
enamine 8 (k2/k2′) may not necessarily correlate with the relative
stability of species formed after that attack.19
Indeed, it may
Figure 5. Experimental and calculated syn/(syn + anti) for the data
shown in Figure 4. For nonselective enamine protonation, k−s = k−a.
Experimental data ﬁt to k−s/k−a ≥ 39.11,18
Figure 4. Temporal proﬁles of the reaction of 4a with 3 monitored by
NMR spectroscopy. [4a]0 = 1.0 M (syn:anti = 93:7); [3]0 = 0.115 M.
Journal of the American Chemical Society Article
dx.doi.org/10.1021/ja300415t | J. Am. Chem. Soc. 2012, 134, 6741−67506744
be seen from the equation in Figure 6c that a highly selective
reaction could result from Pathway B even if the attack of the
enamine on the electrophile is itself unselective.19
Pathway B introduces an intriguing Curtin−Hammett scenario
for stereocontrol. The Curtin−Hammett principle, invoked for
cases where rapidly equilibrating conformers or intermediate
species exist on parallel pathways to reaction products, states that
the product ratio is controlled only by the diﬀerence in standard
free energies (ΔΔG⧧
) derived from the respective transition states
and cannot be determined a priori solely from knowledge of the
relative concentrations of intermediate species. However, the
IUPAC deﬁnition acknowledges that the ΔΔG⧧
value obtained
under Curtin−Hammett conditions embodies both kinetic and
thermodynamic components of selectivity, as we demonstrated
in Figure 6 for Pathway B: product composition is “formally
related to the relative concentrations” of intermediates and to
“the respective rate constants of their reactions”.20
A pioneering example of Curtin−Hammett selectivity control
in asymmetric catalysis was revealed by Landis and Halpern’s
studies of asymmetric hydrogenation of enamides catalyzed by
chiral Rh phosphines under low H2 pressure.21,22
Enantioselectivity
was ultimately governed by the ratio of rate constants
for the irreversible oxidative addition of hydrogen to
Rh-enamide complexes, although the major product was formed
through the pathway of the less stable, minor intermediate. In
that “major−minor” case, the sense of the contribution of the
thermodynamic component was overridden by that of the
kinetic component of the Curtin−Hammett relation for product
ratio. Conversely, it is also possible that higher intermediate
stability can outweigh lower reactivity as the determining
factor in competitive reactions in a scenario that has been termed
“monopolizing kinetics”.23
In the present case, the thermodynamic
component is found to dictate selectivity.
Another noteworthy distinction in the example presented
here compared to the Landis−Halpern case is the role of an
explicit reversible connectivity between the enantiomeric cycles
that occurs subsequent to the step in which the stereogenic
center is initially formed. Our example also diﬀers from
examples of stereodivergent networks where enantiodiﬀerentiation
occurs after the ﬁrst irreversible step, as demonstrated in
several cases including Schiﬀ base (metal)-catalyzed epoxidations
and aziridinations24
and Pd(binap)-catalyzed arylation of
dihydrofuran.25
The selectivity scenario illustrated in Figure 6 is,
to the best of our knowledge, revealed here for the ﬁrst time in
organocatalysis.
In all of these examples, mechanistic understanding as well as
reaction improvement ultimately relies on an accurate assessment
of the step in the cycle representing the relevant transition
state for stereocontrol. In the present example, examination
of the energy diagram of Figure 6a, the network of Figure 6b, and
the equation for er shown in Figure 6c reveals a key consequence
of this scenario, which is that a network obeying Pathway B
aﬀords a number of diﬀerent potential points in the cycle where
stereocontrol may be induced. This concept may have important
general implications for catalyst and reaction design. Rather than
placing an exclusive focus on the transition states of the stereogenic
center-forming step, strategies for the discovery and design
of eﬃcient catalyst/substrate combinations that would conventionally
be assumed to follow the steric model II might beneﬁt
from consideration of means for optimization of the relative
activity or stability of downstream intermediates, such as 5a/5a′,26
as a design tool to enhance er.
α-Chlorination. The ﬁrst highly enantioselective, direct αchlorination
of aldehydes was reported by MacMillan and co-
workers27
using imidazolidinones with acid co-catalysts. A protonmediated
cyclic transition state corresponding to our Pathway A
was proposed as the stereo-deﬁning step in that work. Jorgensen28
Figure 6. Comparison of proposed mechanistic pathways for the two enantiomeric cycles of the reaction of Scheme 5, with compound numbering as
in Scheme 5. Pathway A/A′ represents selectivity as would be determined according to model II by the transition state for reaction between 2a and 8a.
Pathway B/B′ represents the novel pathway proposed in Scheme 5, where selectivity is determined by the relative stabilities and activities of
downstream intermediates. (a) Reaction pathway qualitatively described by the energy diagram from the point of bifurcation at addition of the
electrophile to the enamine. Structures for experimentally observed species 5a, 6a, and 8a are shown. Dashed lines are given for Pathway B′ where no
species have been experimentally observed. (b) Reaction pathway described by rate and equilibrium steps, with proposed rate-determining species
for each pathway highlighted in boxes. Only kinetically meaningful species are included. (c) Mathematical description of the enantiomeric ratio (er)
for each pathway, equating the transition-state energy diﬀerence to the appropriate relationship to the kinetic and equilibrium constants.
Journal of the American Chemical Society Article
dx.doi.org/10.1021/ja300415t | J. Am. Chem. Soc. 2012, 134, 6741−67506745
reported highly enantioselective chlorination using 2,5-diphenylpyrrolidine
10, proposing that the reaction proceeds via an
unusual initial N-chlorination followed by a 1,3-sigmatropic shift
of the chlorine atom to the enamine carbon atom. Interestingly,
α-chlorination using diarylprolinol ethers 3 or 9 has not been
reported (Scheme 7), although these catalysts undergo selective
α-ﬂuorination and α-bromination.4,7
We examined the kinetic proﬁle of the reaction of Scheme 7
using catalyst 3 and chlorinating sources 2b and 2c, monitoring
conversion by reaction calorimetry. Figure 7 shows that the
kinetic proﬁle exhibits features that closely resemble those observed
for the conjugate addition of aldehydes to nitrostyrene with
this catalyst. A rapid initial rate corresponding to ca. one turnover
of the catalyst was followed by a slower zero-order regime, demonstrating
that the reaction does not depend on the concentration
of either of the two reactants. Curiously, however, Figure 7
also shows that, although the rate in the second regime is insensitive
to concentrations, its absolute magnitude is inﬂuenced
by the nature of the chlorinating reagent, with the bulkier
N-chlorophthalimide 2c being less reactive.
In analogy with our results for the conjugate addition to nitrostyrene
using 3, this kinetic proﬁle suggests the rapid buildup
within the cycle of a stable intermediate species that contains
both substrates. The rate-determining step occurs downstream
from this resting state, which lies after the step in which the
electrophile X−Y is added to form the stereogenic center.
Variable-temperature NMR spectroscopic studies of the
interaction of 1c with N-chlorosuccinimide 2b and catalyst 3
were carried out to probe the identity of the intermediate
species implied from the kinetic data. Observation of a single
peak at room temperature that resolves into two peaks below
ca. −30 °C, as shown in Figure 8 (top), suggests a rapid and
reversible equilibrium between two diastereomers, which is
conﬁrmed by EXSY experiments as shown in Figure 8 (bottom),
allowing assignment of aminal 5c analogous to a species reported
by Jorgensen using pyrrolidine.28b
NMR studies using catalysts 9
and 10 identiﬁed analogous aminal diastereomers. In none of
these examples was any evidence observed for the N-chlorinated
intermediate proposed by Jorgensen.28b
Taken together, the
kinetic and spectroscopic results suggest that the rapid initial
rate corresponds to the buildup of 5c as the resting state in the
catalytic system.
Table 1 shows the ratio of the two NMR peaks for diastereomeric
species 5c formed from the interaction of 1c with 2b
or 2c and the catalysts 3, 9, and 10. The ratio of diastereomers
Figure 7. Conversion vs time for the reaction of Scheme 6 with 2b and
2c (0.63 M) and 1c (0.48 M) using 10 mol % catalyst 3 and 2 mol %
CH3COOH at −11.4 °C in CHCl3.11
Figure 8. (Top) Variable-temperature 1
H NMR spectra of 1c (0.65
M), 2b (0.5 M), and 3 (0.1 M) in CDCl3 showing an apparent
single resting state at 20 °C that splits into two diastereomers below
−30 °C. (Bottom) EXSY cross-peaks reveal the connectivity of the
two diastereomers of 5c existing at room temperature in a rapid
equilibrium.
Scheme 7. α-Chlorination of Aldehydes with Diﬀerent
Catalysts and Chlorinating Agents
Journal of the American Chemical Society Article
dx.doi.org/10.1021/ja300415t | J. Am. Chem. Soc. 2012, 134, 6741−67506746
depends on the catalyst employed, ranging from 30:70 for 3
with 2b to the case of catalyst 10, where a single diastereomer is
observed with either electrophile. Strikingly, Table 1 also reveals
that an excellent correlation exists between the ratio of
diastereomeric species 5c and the er of the reaction product 4c
for all catalysts and both electrophiles.
The results of Table 1 imply that the diastereomeric species
5c are directly involved in catalyst turnover, rather than existing
as parasitic oﬀ-cycle reservoirs. In support of this proposal,
consideration of species 5c as the resting state and invoking
rate-determining elimination of the leaving group Y from this
species helps to explain the seemingly contradictory observations
of overall zero-order kinetics and a diﬀerence in reactivity
for diﬀerent chlorinating reagents.
The close relationship between the relative concentrations of
the diastereomers of 5c and the er of product 4c requires
careful analysis of potential mechanistic pathways that might
involve these species. The atom connectivity identiﬁed by our
NMR studies allows four possible diastereomers of 5c,
consisting of two sets of syn:anti pairs formed from attack at
opposite faces of the enamine. If addition of chloride to the
enamine is considered to be irreversible, the observed equilibration
of the diastereomers of 5c could occur via reversible
dissociation of the leaving group Y (Y = succinimide or phthalimide),
as shown in Scheme 8. This scheme shows the
conventionally proposed catalytic route within the dotted lines,
where the product is derived directly from the iminium ion
species formed upon addition of Cl, analogous to a Pathway A
scenario in our previous example. The scheme also proposes a
catalytic route arising from diastereomeric intermediates formed
from this iminium ion. This route exhibits some features of our
previously described Pathway B, with kinetically meaningful
intermediates being formed on-cycle and equilibrating downstream
from the explicit stereogenic center-forming step.
In the conventional route, stereocontrol is set in the Cl
addition step to enamine 8c forming the iminium ion 11c (or
11c′), which directly undergoes hydrolysis to form the product.
In this scenario the diastereomers 5c and 5c′ lie oﬀ-cycle and
are unrelated to the stereochemical outcome; the product er
will be related directly to the facial selectivity in the Cl addition
step. However, the observed relationship between diastereomer
stability and product er revealed in Table 1 cannot be predicted
by this mechanism.
A second diﬃculty with the conventionally proposed pathway
is that irreversible Cl addition in Scheme 8 implies that the
observed equilibration between the two oﬀ-cycle diastereomers
occurs between those formed from addition of Cl to the same
face of the enamine. The fact that we observe only two out
of the four possible diastereomers implies near-perfect facial
selectivity in the addition of Cl to the enamine 8c. Thus, observation
of a non-perfect er in a reaction following the conventional
route should be accompanied by observation of
all four diastereomeric intermediates of 5c, in contrast to the
experimental result.
The alternate proposed route places these diastereomers on
the catalytic cycle itself, with the steps to product through these
Table 1. Diastereomeric Ratios (dr) for Species 5c Identiﬁed
by NMR and Correlation with the Stereochemical Outcome
of the Catalytic Reaction11
entry catalyst Cl−Y dra
5c:5c′ (Keq
dias
) erb
R-4c:S-4c
1 3 2b 30:70 29:71
2 9 2b 16:84 16:84
3 10 2b <5:95c
3:97d
4 3 2c 26:74 23:77
5 9 2c 21:79 21:79e
6 10 2c <5:95c
3:97d
a
Keq
dias
= [syn-5c]/[anti-5c] or [syn-5c′]/[anti-5c′], determined from
NMR spectroscopy at −54 °C. b
Determined by chiral GC. c
Minor
diastereomer not detected. d
From ref 28b. e
Ca. 5% conversion.
Scheme 8. Proposed Catalytic Cycles for α-Chlorination
with Catalysts 3, 9, and 10
Journal of the American Chemical Society Article
dx.doi.org/10.1021/ja300415t | J. Am. Chem. Soc. 2012, 134, 6741−67506747
species envisioned as shown in Scheme 8. Facial attack of the
enamine proceeds exclusively by either the left or the right
pathway, to form 11c or 11c′, respectively, which equilibrates
between one of the two possible syn and one of the two
possible anti diastereomers of 5c. E2 elimination of the leaving
group Y from these species would yield the Z-enamine from
syn-5c and the E-enamine from anti-5c. Invoking stereospeciﬁc
reaction of these enamines, as we demonstrated for analogous
product enamine species in the conjugate addition to nitrooleﬁns,
implies that product R-4c results from the former and
product S-4c from the latter, as shown in Scheme 8. No species
have been observed downstream from the diastereomers 5c,
supporting the suggestion that elimination of the leaving group
Y is the rate-determining step in a pathway to product through
these diastereomers. As suggested previously, this mechanism is
consistent with both the diﬀerent absolute rates for 2b and 2c
as electrophiles and the zero-order kinetics in electrophile
concentration.
This reaction scheme represents a further example of a
Curtin−Hammett scenario, where the kinetic and thermodynamic
components of product enantioselectivity are shown in
eq 2 for the network in Scheme 8. The er is given by the
relative concentrations of the two diastereomers of 5c and the
relative rates of Y-elimination to form the product enamines.
Since the absolute stereochemistry of the observed diastereomers
is unknown, eq 2 is written for each of the two possible
syn:anti pairs arising from opposite facial selectivity in enamine
attack on Cl.
This Curtin−Hammett scenario provides a reasonable
mechanistic alternative to model II for rationalizing the differences
in enantioselectivity observed comparing catalysts 3, 9,
and 10 in Table 1. The close correlation observed between the
er and Keq
dias
in all examples in Table 1 implies that the rate of
elimination of the leaving group is similar for the syn and anti
diastereomers, or ksyn ≈ kanti in eq 2. In this case the relative
stability of the two diastereomers is in direct proportion to the
er and serves as the primary stereocontrolling factor. This helps
to explain the higher selectivity achieved with catalyst 10, less
bulky but arguably sterically more complex than catalysts 3 or
9, which is diﬃcult to rationalize by model II.
Thus we may conclude that the near-perfect facial selectivity
initially achieved in the addition of Cl−Y in Scheme 8, required
to rationalize observation of only two of the four diastereomers
of 5c, in fact can give way to an erosion of product er downstream,
due to the opposite stereochemistry predicted from
reaction of the syn and anti diastereomeric intermediates.
Generality. The preceding examples oﬀer experimental
evidence illustrating two subtly diﬀerent scenarios where equilibrating
intermediate species formed downstream from the
stereogenic center-forming step may inﬂuence the stereochemical
outcome. In the conjugate addition of linear aldehydes to nitrooleﬁns,
product stereoselectivity was found to be dictated by a
single highly stable diastereomer existing as the on-cycle resting
state equilibrated with upstream intermediates. In principle, such a
scenario would allow imperfect facial selectivity to be enhanced by
channeling intermediates from both sides of the cycle to
product through this stable species on one side of the network.
In the second example of the α-chlorination of aldehydes, the
opposite scenario was revealed: near-perfect facial selectivity in
attack of the enamine could be eroded by the competitive
reactions of equilibrated intermediates formed subsequent to the
irreversible addition of the electrophile. It is noteworthy that in
both cases, simple consideration of the transition state for the
stereogenic center-forming step was found to be inadequate to
explain the observed stereochemical outcome. Evocation of the
Curtin−Hammett principle, in which both relative stability and
reactivity of kinetically competitive species must be considered,
allows rationalization of the experimental results.
These results raise a question of the generality of this
Curtin−Hammett paradigm in organocatalysis. Precedent exists
in a number of other organocatalytic reactions with the potential
to exhibit intermediates downstream from the stereogenic
center-forming step. Early model calculations by Houk3a
predicted
that analogous oxetane species would form from the
collapse of iminium ion intermediates in amine-catalyzed aldol
reactions (Scheme 9a). Azetidine intermediates have been
isolated in stoichiometric reactions between enamines and imines
(Scheme 9b),29
suggesting that such intermediates may be viable
in catalytic Mannich reactions. Gellman reported observation of a
dihydropyran in the aldehyde−enone Michael addition catalyzed
by a MacMillan imidazolidinone (Scheme 9c).30
We identiﬁed a
single stable cyclic intermediate in the reaction of a maleimide with
8, the enamine formed from 1a and 3 (Scheme 9d),11
a reaction
that has been shown to proceed with high enantioselectivity.31
While the cases shown in Scheme 9 all reveal cyclic intermediates
in addition reactions with X = Y electrophiles, our
results in the α-chlorination of aldehydes imply a more general
role may also exist in enamine catalysis for downstream
intermediates that are not cyclic and in reactions that involve
substitution with X−Y electrophiles. The potential for such
species to inﬂuence the stereochemical outcome of the reaction
is currently unexplored and will depend in each case both
on the relative stability and reactivity of the diastereomers
formed and on whether they lie on or oﬀ the catalytic cycle.
Scheme 9. Intermediates Identiﬁed Subsequent to the
Stereogenic Center-Forming Step in Reactions between
Enamines and Electrophiles: (a) Ref 3a, (b) Ref 29, (c) Ref
30, and (d) This Work11
Journal of the American Chemical Society Article
dx.doi.org/10.1021/ja300415t | J. Am. Chem. Soc. 2012, 134, 6741−67506748
Further kinetic and spectroscopic investigations of these and
other reaction systems may allow these questions to be answered.
In cases where such intermediates may play a role in the
stereochemical outcome, the potential exists for exploitation
of these features via rational catalyst and reaction design that
may ultimately lead to even more eﬃcient and selective
asymmetric organocatalytic processes than might be envisioned
by consideration of the precepts of model II alone.
■ CONCLUSIONS
In summary, NMR spectroscopic identiﬁcation of intermediate
species coupled with kinetic studies help to rationalize
the stereochemical outcome in two separate organocatalytic
reactions, the conjugate addition of linear aldehydes to nitrooleﬁns
and the α-chlorination of aldehydes, in a manner that
could not be explained by a simple steric model of enamine
attack on an electrophile as in model II. These ﬁndings address
general reactivity and selectivity concepts with the proposal of a
novel reaction paradigm that combines a classic Curtin−Hammett
scenario with the concept of reversibility in or subsequent to the
stereogenic center-forming step. Selectivity in these examples is
rationalized not by comparison of transition states for formation
of the stereogenic center but by the relative stability and reactivity
of equilibrated downstream intermediates in the separate
branches of a competitive reaction network. Such networks oﬀer
the opportunity to tune selectivity at a number of diﬀerent points
in the catalytic cycle and introduce additional possibilities for
rational design of active and selective substrate/catalyst combinations.
Experimental results are oﬀered to suggest that this
concept may apply to related reactions and thus may represent a
general phenomenon for amine catalysts lacking an acidic
directing proton.
■ ASSOCIATED CONTENT
*S Supporting Information
Experimental procedures, kinetic studies, and structural NMR
studies. This material is available free of charge via the Internet
at http://pubs.acs.org.
■ AUTHOR INFORMATION
Corresponding Author
blackmond@scripps.edu
Notes
The authors declare no competing ﬁnancial interest.
■ ACKNOWLEDGMENTS
A.A. and D.G.B. acknowledge funding from the EPSRC. J.B.
acknowledges a postdoctoral fellowship from the Education
Ministry of Spain (EX2009-0687) and FECYT. We acknowledge
D.-H. Huang and L. Pasternack (TSRI NMR Facility) for
valuable assistance with NMR spectroscopy.
■ REFERENCES
(1) List, B.; Lerner, R. A.; Barbas, C. F., III. J. Am. Chem. Soc. 2000,
122, 2395.
(2) Ahrendt, K. A.; Borths, C. J.; MacMillan, D. W. C. J. Am. Chem.
Soc. 2000, 122, 4243.
(3) (a) Bahmanyar, S.; Houk, K. N. J. Am. Chem. Soc. 2001, 123,
11273. (b) Bahmanyar, S.; Houk, K. N.; Martin, H. J.; List, B. J. Am.
Chem. Soc. 2003, 125, 2475. (c) Hoang, L.; Bahmanyar, S.; Houk,
K. N.; List, B. J. Am. Chem. Soc. 2003, 125, 16. (d) List, B.; Hoang, L.;
Martin, H. J. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 5839.
(e) Clemente, F. R.; Houk, K. N. Angew. Chem., Int. Ed. 2004, 43,
5765.
(4) (a) Palomo, C.; Mielgo, A. Angew. Chem., Int. Ed. 2006, 45, 7876.
(b) Jensen, K. L.; Dickmeiss, G.; Jiang, H.; Albrecht, L.; Jorgensen,
K.-A. Acc. Chem. Res. 2012, 45, 248.
(5) Seeman, J. I. Chem. Rev. 1983, 83, 83.
(6) Hayashi, Y.; Gotoh, H.; Hayashi, T.; Shoji, M. Angew. Chem., Int.
Ed. 2005, 44, 4212.
(7) Franzen, J.; Marigo, M.; Fielenbach, D.; Wabnitz, T. C.;
Kjaersgaard, A.; Jorgensen, K. A. J. Am. Chem. Soc. 2005, 127, 18296.
(8) Grondal, C.; Jeanty, M.; Enders, D. Nat. Chem. 2010, 2, 167.
(9) (a) Blarer, S. J.; Seebach, D. Chem. Ber. 1983, 116, 3086.
(b) Seebach, D.; Beck, A. K.; Golinski, J.; Hay, J. N.; Laube, T. Helv.
Chim. Acta 1985, 68, 162. (c) Seebach, D.; Golinski, J. Helv. Chim.
Acta 1981, 64, 1413.
(10) Lalonde, M. P.; Chen, Y.; Jacobsen, E. N. Angew. Chem., Int. Ed.
2006, 45, 6366.
(11) See Supporting Information for details
(12) Bures, J.; Armstrong, A.; Blackmond, D. G. J. Am. Chem. Soc.
2011, 133, 8822.
(13) Patora-Komisarska, K.; Benohoud, M.; Ishikawa, H.; Seebach,
D.; Hayashi, Y. Helv. Chim. Acta 2011, 94, 719.
(14) The reversible addition of enamine 8a to nitrostyrene 2a to
form cyclobutane 5a shown in Scheme 5 is likely to proceed via several
sequential, but kinetically indistinguishable, elementary steps that have
been combined as equilibrium constant Keq,5a. Similar arguments apply
to other steps in the cycle, including the deprotonation of cyclobutane
5a with equilibrium constant Keq,6a.
(15) For H/D isotope eﬀects in carbon acid deprotonations, see:
Watt, C. I. F. J. Phys. Org. Chem. 2010, 23, 561. See Supporting
Information for a summary of the mathematical analysis pertaining to
the results presented here.
(16) Bures, J.; Armstrong, A.; Blackmond, D. G. Chem. Sci. 2012, 3,
1273.
(17) The reverse of the catalytic cycle does not proceed further from
E-7, whose formation in the forward cycle is irreversible under catalytic
conditions.
(18) Hoops, S.; Sahle, S.; Gauges, R.; Lee, C.; Pahle, J.; Simus, N.;
Singhal, M.; Xu, L.; Mendes, P.; Kummer, U. Bioinformatics 2006, 22,
3067.
(19) For example, a highly selective outcome is predicted for
Pathway B in the case where the ratio Keq,5a/Keq,5a′ is large, and each of
the other ratios in the equation describing the er for Pathway B is unity
(unselective).
(20) The IUPAC deﬁnition of the Curtin−Hammett principle: Gold,
V.; Loening, K. L.; McNaught, A. D.; Shemi, P. IUPAC Compendium of
Chemical Terminology, 2nd ed.; Blackwell Science: Oxford, 1997.
Because the principle states that product selectivity depends only on
ΔΔG⧧
, it is commonly, although incorrectly, stated as a corollary that
product selectivity is independent of the relative stability of
intermediates. See Supporting Information for a correction of this
misconception.
(21) Landis, C. R.; Halpern, J. A. J. Am. Chem. Soc. 1987, 109, 1746.
(22) Halpern, J. Science 1982, 217, 401.
(23) Ferretti, A. C.; Mathew, J. S.; Ashworth, I.; Purdy, M.; Brennan,
C.; Blackmond, D. G. Adv. Synth. Catal. 2008, 350, 1007.
(24) Zhang, W.; Lee, N. H.; Jacobsen, E. N. J. Am. Chem. Soc. 1994,
116, 425.
(25) Ozawa, F.; Kubo, A.; Matsumoto, Y.; Hayashi, T. Organometallics
1993, 12, 4188.
(26) There are a number of diastereomeric possibilities for
cyclobutane 5a′, which is not observed experimentally. We suggest
that the most likely lowest energy alternative exhibits all conﬁgurations
in the cyclobutane ring reversed from that of 5a.
(27) Brochu, M. P.; Brown, S. P.; MacMillan, D. W. C. J. Am. Chem.
Soc. 2004, 126, 4108.
(28) (a) Halland, N.; Braunton, A.; Bachmann, S.; Marigo, M.;
Jorgensen, K. A. J. Am. Chem. Soc. 2004, 126, 4790. (b) Halland, N.;
Journal of the American Chemical Society Article
dx.doi.org/10.1021/ja300415t | J. Am. Chem. Soc. 2012, 134, 6741−67506749
Alstrup Lie, M.; Kjaersgaard, A.; Marigo, M.; Schiott, B.; Jorgensen, K.
A. Chem.Eur. J. 2005, 11, 7083.
(29) Alben, R. W. M.; Smit, R.; Scheeren, J. W. J. Org. Chem. 1987,
52, 365.
(30) Peelen, T. J.; Chi, Y.; Gellman, S. H. J. Am. Chem. Soc. 2005,
127, 11598.
(31) Zhao, G. L.; Xu, Y.; Sunden, H.; Eriksson, L.; Sayah, M.;
Cordova, A. Chem. Commun. 2007, 734.
Journal of the American Chemical Society Article
dx.doi.org/10.1021/ja300415t | J. Am. Chem. Soc. 2012, 134, 6741−67506750