The development of the first catalyzed reaction of ketenes and imines: Catalytic, asymmetric synthesis of β-lactams

Article abstract of DOI:10.1021/ja0258226

We report practical methodology for the catalytic, asymmetric synthesis of β-lactams resulting from the development of a catalyzed reaction of ketenes (or their derived zwitterionic enolates) and imines. The products of these asymmetric reactions can serve as precursors to a number of enzyme inhibitors and drug candidates as well as valuable synthetic intermediates. We present a detailed study of the mechanism of the β-lactam forming reaction with proton sponge as the stoichiometric base, including kinetics and isotopic labeling studies. Stereochemical models based on molecular mechanics (MM) calculations are also presented to account for the observed stereoregular sense of induction in our reactions and to provide a guidepost for the design of other catalyst systems.

Full text of DOI:10.1021/ja0258226

Published on Web 05/16/2002
The Development of the First Catalyzed Reaction of Ketenes
and Imines: Catalytic, Asymmetric Synthesis of â-Lactams
Andrew E. Taggi, Ahmed M. Hafez, Harald Wack, Brandon Young,
Dana Ferraris, and Thomas Lectka*
Contribution from the Department of Chemistry, Johns Hopkins UniVersity,
3400 North Charles Street, Baltimore, Maryland 21218
Received February 5, 2002
Abstract: We report practical methodology for the catalytic, asymmetric synthesis of â-lactams resulting
from the development of a catalyzed reaction of ketenes (or their derived zwitterionic enolates) and imines.
The products of these asymmetric reactions can serve as precursors to a number of enzyme inhibitors
and drug candidates as well as valuable synthetic intermediates. We present a detailed study of the
mechanism of the â-lactam forming reaction with proton sponge as the stoichiometric base, including kinetics
and isotopic labeling studies. Stereochemical models based on molecular mechanics (MM) calculations
are also presented to account for the observed stereoregular sense of induction in our reactions and to
provide a guidepost for the design of other catalyst systems.
Introduction
this structural motif a worthwhile goal for the synthetic organic
chemist,10 thus the synthesis of these nonantibiotic â-lactams
The clinical relevance of â-lactams continues to expand at a
surprising rate. Although their use as antibiotics is being
compromised to some extent by bacterial resistance pressures,1
recently â-lactams (especially nonnatural ones) have achieved
many important nonantibiotic uses. Some of the most notable
advances concern the development of mechanism-based serine
protease inhibitors2 of elastase,3 cytomegalovirus protease,4
thrombin,5 prostate specific antigen,6 â-lactamase,7 and cell
metastasis.8 As a testament to the continuing importance of
â-lactams to pharmaceutical science, a recent issue of Tetra-
hedron was devoted exclusively to â-lactam chemistry and
synthesis. In the preface of this issue, a call was given to
pharmaceutical companies to reintensify efforts to develop new
types and uses of â-lactams, and refocus efforts away from
antibiotic use.9 The sheer importance of â-lactams has made
will be the focus of this contribution. While considerable effort
has been put into synthetic methodology to construct the basic
â-lactam skeleton, there have been few general methods
proposed for their enantioselective synthesis. Current asym-
metric methodology mainly uses chiral auxiliary-based systems
that while effective, require an auxiliary that necessitates
additional synthetic operations to install and remove. A general
methodology built around a useful catalytic asymmetric reaction,
as opposed to chiral auxiliaries, would be extremely useful. We
herein summarize general and efficient methodology for the
synthesis of the monobactam skeleton, which rests on the
development of the first catalyzed reaction of ketenes (or their
synthetic equivalents) and imines. We recently accomplished
this mechanistically distinct, catalyzed reaction by making the
ketene nucleophilic (through generation of a zwitterionic
intermediate),11 and the imine component nonnucleophilic, as
in electron deficient R-imino ester 5a (X ) Ts, R ) Et).12,13
Many new classes of extremely useful â-lactams result from a
successful asymmetric reaction of this type of imines with
ketenes, including thrombin, prostate specific antigen, and
cytomegalovirus protease inhibitors (Scheme 1).
* Corresponding author. E-mail: lectka@jhunix.hcf.jhu.edu.
(1) For some reviews on â-lactam antibiotics and fungicides see: (a) The
Organic Chemistry of â-Lactams; Georg, G. I., Ed.; VCH Publications:
New York, 1993, and references therein. (b) Neuhaus, F. C.; Georgopa-
padaku, N. H. In Emerging Targets in Antibacterial and Antifungal
Chemotherapy; Sutcliffe, J., Georgopapadakou, N. H., Eds.; Chapman and
Hall: New York, 1992.
(2) Haley, T. M.; Angier, S. J.; Borthwick, A. D.; Singh, R.; Micetich, R. G.
I Drugs 2000, 3, 512-517.
(3) Skiles, J. W.; Sorcek, R.; Jacober, S.; Miao, C.; Mui, P. W.; McNeil, D.;
Rosenthal, A. S. Bioorg. Med. Chem. Lett. 1993, 3, 773-778.
(4) (a) Ogilvie, W. W.; Yoakim, C.; Do, F.; Hache, B.; Lagace, L.; Naud, J.;
O’Meara, J. A.; Deziel, R. Bioorg. Med. Chem. 1999, 7, 1521-1531. (b)
Borthwick, A. D.; Weingarten, G.; Haley, T. M.; Tomaszewski, M.; Wang,
W.; Hu, Z.; Bedard, J.; Jin, H.; Yuen, L.; Mansour, T. S. Bioorg. Med.
Chem. Lett. 1998, 8, 365-370.
(5) Han, W. T.; Trehan, A. K.; Wright, J. J. K.; Federici, M. E.; Seiler, S. M.;
Meanwell, N. A. Bioorg. Med. Chem. 1995, 3, 1123-1143.
(6) Adlington, R. M.; Baldwin, J. E.; Chen, B.; Cooper, S. L.; McCoull, W.;
Pritchard, G. J.; Howe, T. J. Bioorg. Med. Chem. Lett. 1997, 7, 1689-
1694.
(7) Danelon, G. O.; Mata, E. G.; Mascaretti, O. A.; Girardini, J.; Marro, M.;
Roveri, O. A. Bioorg. Med. Chem. Lett. 1995, 5, 2037-2040.
(8) Tsuruoka, T.; Nakabayashi, S.; Fukuyasu, H.; Ishii, Y.; Tsuruoka, T.;
Yamamoto, H.; Inouye, S.; Kondo, S. European Patent Application, 1989,
31 pp.
There have been a few reported cases of catalytic, enantio-
selective syntheses of â-lactams. Alper reported a carbonylation
of chiral and meso aziridines to produce optically enriched
products in modest enantioselectivity (ee).14 Tomioka et al.
(9) Miller, M. J. Tetrahedron 2000, 56, preface.
(10) A literature search found over 1600 articles describing the synthesis of
â-lactams over the last 25 years.
(11) Taggi, A. E.; Hafez, A. M.; Wack, H.; Young, B.; Drury, W. J., III; Lectka,
T. J. Am. Chem. Soc. 2000, 122, 7831-7832.
(12) Imine 5a was first popularized in work by: Tschaen, D. H.; Turos, E.;
Weinreb, S. M. J. Org. Chem. 1984, 49, 5058-5064.
(13) We have also used N-acyl imino esters to form â-lactams as intermediates
in the synthesis of substituted aspartic acids: Dudding, T.; Hafez, A. M.;
Taggi, A. E.; Wagerle, T. R.; Lectka, T. Org. Lett. 2002, 4, 387-390.
9
6626
J. AM. CHEM. SOC. 2002, 124, 6626-6635
10.1021/ja0258226 CCC: $22.00 © 2002 American Chemical Society

Catalytic, Asymmetric Synthesis of â-Lactams
A R T I C L E S
Scheme 1. â-Lactam Disconnection into Imines and Ketenes
Scheme 2. Nucleophilic Catalysis with Ketenes
described an interesting approach in which a chiral ether was
used as a chelating catalyst for lithium enolates in their ring
forming reactions with imines to form optically enriched
â-lactams.15 Finally, Doyle et al. have exploited the rhodium-
catalyzed C-H insertion reaction of diazoacetates to construct
the â-lactam skeleton for some specialized cases.16 Biocatalytic
methodology has also played a notable role in the synthesis of
optically pure â-lactams.17 These pioneering methods represent
initial forays into this area, and outline a number of re-
maining challenges, including enantioselectivity and substrate
generality.
While there are many known syntheses of â-lactams, the
Staudinger reaction is widely accepted as one of the simplest
methods. It is a process that proceeds without a catalyst,18 and
its rate depends on the nucleophilicity of the imine. It is
generally thought to be a stepwise [2+2] cycloaddition that is
initiated by an imine nucleophile attacking a ketene, followed
by cyclization of the azadiene intermediate (eq 1).19 This
with achiral imines to achieve asymmetric induction through
the use of a chiral nucleophilic catalyst.
Results and Discussion
Asymmetric Catalysis with Ketenes. There has been a surge
of interest in nucleophilic asymmetric catalysis as organic
chemists have come to see the benefits and practicality offered
by totally organic catalyst systems, especially their ease of
handling and recovery.22 In fact, some of the earliest discoveries
in the field of asymmetric catalysis involved the use of organic
nucleophiles. One pioneering study occurred in the labs of
Pracejus, who used alkaloids to catalyze the alcoholysis of
disubstituted ketenes to afford chiral alcohols in moderate
enantioselectivity (ee).23 In the 1980s, Wynberg developed an
asymmetric â-lactone synthesis using ketenes, an activated
aldehyde component, and cinchona alkaloid catalysts.24 Tamai
and Romo have also used ketenes for the asymmetric synthesis
of â-lactones.25 This list would possibly be much longer were
it not for the fact that ketenes are often difficult to handle. It is
clear that efficient, wide-ranging ketene generation would render
many problems in asymmetric catalysis and mechanistic organic
chemistry approachable for the first time.
Ketene Generation. Ketenes are versatile, well-known
substrates in organic synthesis. Their cumulated double bond
structure imparts a unique spectrum of reactivity, and gives rise
to a large number of mechanistically interesting transformations.
Ketenes usually add substrates avidly across their CdC bond,
a reaction which can be catalyzed by various nucleophilic Lewis
bases (Scheme 2),26 suggesting to us a strategy whereby the
zwitterionic enolate could react with an electrophilic imine, were
we to have in hand a generally useful ketene synthesis. There
are many ways to synthesize ketenes through thermolytic, and
photolytic reactions, although from a synthetic standpoint these
methods suffer from drawbacks, requiring cumbersome ap-
paratus, tricky conditions, or esoteric starting materials.27 In
addition, many methodological studies limit their scope to
readily available (but less useful) disubstituted ketenes. In
synthetic chemistry, ketenes are most often produced from
inexpensive acid chlorides through dehydrohalogenation reac-
tions with tertiary amines. This method works well for distillable
disubstituted ketenes. More synthetically useful and chemically
challenging monosubstituted ketenes are generally much more
reactive, and must, with rare exception, be made in situ, where
reaction has a very low energy of activation and thus proceeds
rapidly even at reduced temperatures. Due to the wide range of
available ketene and imine combinations, this method is
considered one of the most useful and economical, although it
is not apparently amenable to catalysis.
Many asymmetric versions of the Staudinger reaction involve
the use of a chiral oxazolidinone substituted ketene.20 Bose et
al. have developed a diastereoselective â-lactam synthesis using
a chiral auxiliary attached to the imine.21 While these and other
related procedures may be efficient in cases where the chiral
auxiliary is part of the final target molecule, in most cases it is
not and must be removed. In our system, achiral ketenes react
(14) (a) Calet, S.; Urso, F.; Alper, H. J. Am. Chem. Soc. 1989, 111, 931-934.
(b) Davoli, P.; Moretti, I.; Prati, F.; Alper, H. J. Org. Chem. 1999, 64,
518-521.
(15) Fujieda, H.; Kanai, M.; Kambara, A. I.; Tomioka, K. J. Am. Chem. Soc.
1997, 119, 2060-2061.
(16) Doyle, M. P.; Kalinin, A. V. Synlett 1995, 10, 1075-1076.
(17) Cheetham, P. S. J. Applied Biocatalysis, 2nd ed.; Zylepsis, Ltd.: Kent, 2000;
pp 93-152.
(18) Asymmetry in the Staudinger reaction can be achieved by attachment of a
suitable chiral auxiliary to the imine or to the terminal carbon atom of the
ketene: (a) Palomo, C.; Aizpurua, J. M.; Ganboa, I.; Oiarbide, M. Eur. J.
Org. Chem. 1999, 12, 3223-3235. (b) Cooper, R. D. G.; Daugherty, B.
W.; Boyd, D. B. Pure Appl. Chem. 1987, 59, 485-492.
(19) (a) Banik, B. K.; Becker, F. F. Tetrahedron Lett. 2000, 41, 6551-6554.
(b) Bandini, E.; Martelli, G.; Spunta, G.; Bongini, A.; Panunzio, M.
Tetrahedron Lett. 1996, 37, 4409-4412. (c) Jayaraman, M.; Srirajan, V.;
Deshmukh, A. R. A. S.; Bhawal, R. M. Tetrahedron 1996, 52, 3741-
3756. (d) Cossio, F. P.; Ugalde, J. M.; Lopez, X.; Lecea, B.; Palomo, C.
J. Am. Chem. Soc. 1993, 115, 995-1004. (e) Lynch, J. E.; Riseman, S.
M.; Laswell, W. L.; Tschaen, D. M.; Volante, R. P.; Smith, G. B.; Shinkai,
I. J. Org. Chem. 1989, 54, 3792-3796.
(22) For a comprehensive review see: Dalko, P. I.; Moisan, L. Angew. Chem.,
Int. Ed. 2001, 40, 3726-3748.
(23) Pracejus, H.; Kohl, G. Justus Liebigs Ann. Chem. 1969, 722, 1-11.
(24) Wynberg, H.; Staring, E. G. J. J. Am. Chem. Soc. 1982, 104, 166-168.
(25) (a) Tamai, Y.; Someya, M.; Fikumoto, J.; Miyano, S. J. Chem. Soc., Perkin
Trans. 1994, 12, 1549-1550. (b) Tennyson, R.; Romo, D. J. Org. Chem.
2000, 65, 7248-7252.
(26) Qiao, G. G.; Andraos, J.; Wentrup, C. J. Am. Chem. Soc. 1996, 118, 5634-
(20) Evans, D. A.; Sjogren, E. B. Tetrahedron Lett. 1985, 26, 3783-3786.
(21) Bose, A. K.; Manhas, M. S.; van der Veen, J. M.; Bari, S. S.; Wagle, D.
R. Tetrahedron 1992, 48, 4831-4844.
5638.
(27) (a) Tidwell, T. T. Ketenes; John Wiley & Sons: New York, 1995. (b)
Tidwell, T. T. Acc. Chem. Res. 1990, 23, 273-279.
9
J. AM. CHEM. SOC. VOL. 124, NO. 23, 2002 6627

A R T I C L E S
Taggi et al.
Scheme 3. Shuttle Deprotonation with Kinetic and
Thermodynamic Bases
they are invariably stirred with precipitated and partially
dissolved ammonium salt byproducts.28 In some cases, the
presence of these trialkylammonium byproducts presents no
problem. However, we found that in many instances (vide infra),
trialkylammonium salts were deleterious to established reac-
tions.29 Trialkylammonium salts have an appreciable solubility
in almost all organic solvents and can serve as unwanted
Brønsted acid catalysts. They are in equilibrium with the
corresponding free amine bases that can act as nucleophiles or
ligands; and because they are achiral, they may interfere with
the desired reaction, producing racemic products.
One possible solution is to use polymer-bound bases such as
the highly basic resin BEMP 3a,30 a triaminophosphonamide
imine bound to a polymeric support,31 to form ketenes 2 from
acid halides 1.32 We found that BEMP produces many ketenes
rapidly when THF solutions of suitable acid chlorides 1 are
passed through an addition funnel at -78 °C containing the
polymer (eq 2).33,34 Alternatively, the BEMP polymer can be
Our first approach to the shuttle base synthesis of ketenes in
situ was the use of the strong organic base proton sponge (3b)
as a nonnucleophilic thermodynamic base. We found that simply
mixing 1 equiv of 3b and acid chlorides 1 at low temperature
usually does not produce detectable amounts of ketene. Although
3b is a strong thermodynamic base, it is hindered and in most
cases kinetically slow at deprotonating carbon-based acids.37
We found that a nucleophile such as benzoylquinine (BQ,
4a), an inexpensive38 and versatile asymmetric catalyst,39 serves
as an excellent shuttle base (base k) when added to a solution
of various acid chlorides 1 in toluene at low temperature. In
the case of diphenylketene 2b, a yellow solution is formed along
with a white precipitate over the course of a few minutes.40 In
an appropriate solvent like toluene, the ammonium salt of proton
sponge was found to not interfere in many asymmetric reactions
catalyzed by chiral nucleophiles. The main drawbacks to the
use of the proton sponge-shuttle procedure include economical
aspects (although proton sponge is a moderately priced chemical,
in large quantities its use may be a cost factor) as well as the
possibility that in rare instances the sponge itself may react in
undesirable ways.33
added to a solution of the acid chloride at low temperature. After
a few minutes, filtration of the solid-supported base produces
the desired solution of pure ketene. We previously investigated
other polymer-supported bases such as guanidines and tertiary
amines but found them to be ineffective for ketene generation.35
Although they are attractive reagents, the primary drawback to
the use of polymeric bases involves their expense (5 g of BEMP
costs about $120).
Shuttle Deprotonation. The use of tertiary amines for
dehydrohalogenations of acid halides 1 to form ketenes for our
reactions is complicated by the fact that they are usually too
nucleophilic. The use of a stoichiometric base that is thermo-
dynamically strong, but kinetically nonnucleophilic, could
overcome this problem. This strategy, which we term “shuttle
deprotonation”, utilizes a catalytic, chiral nucleophile, which
is kinetically active (base k), to dehydrohalogenate the acid
chloride in the first step (Scheme 3). Exploiting the premise
that proton transfers between heteroatoms are inherently fast,36
the kinetically favored base k then rapidly transfers its proton
to base t, the thermodynamically active, but kinetically restricted
base, to regenerate itself for another catalytic cycle.
Carbonate and Hydride Shuttle Bases. We have also looked
at powdered carbonates as stoichiometric heterogeneous bases
(37) Remarkably, there are some cases where the R-proton of the acid chloride
is acidic enough to make it possible for proton sponge to perform the
dehydrohalogenation without the assistance of BQ. See: Tidwell, T. T.;
Fenwick, M. H. Eur. J. Org. Chem. 2001, 3415-3419.
(38) Multigram quantities of benzoylquinine are easily synthesized in one step
from benzoyl chloride and quinine. Also, BQ is benchtop stable indefinitely.
(39) Cinchona alkaloids have the status as general nucleophilic catalysts and
ligands for many organic reactions, for example see: (a) McDaid, P.; Chen,
Y.; Deng, L. Angew. Chem., Int. Ed. 2002, 41, 338-340. (b) Hang, J.;
Tian, S.-K.; Tang, L.; Deng, L. J. Am. Chem. Soc. 2001, 123, 12696-
12697. (c) Bolm, C.; Schiffers, I.; Dinter, C. L.; Defre`re, L.; Gerlach, A.;
Raabe, G. Synthesis 1719-1730. (d) Kacprzak, K.; Gawronski, J. Synthesis,
2001, 7, 961-998. (e) Bolm, C.; Schiffers, I.; Dinter, C. L.; Gerlach, A. J.
Org. Chem. 2000, 65, 6984-6991. (f) Zhang, F.-Y.; Corey, E. J. Org. Lett.
2000, 2, 1097-1100. (g) Alvarez, R.; Hourdin, M.-A.; Cave, C.; d’Angelo,
J.; Chaminade, P. Tetrahedron Lett. 1999, 40, 7091-7094. (h) Martyres,
D. Synlett 1999, 9, 1508-1515. (i) Studer, M.; Blaser, H.-U.; Okafor, V.
J. Chem. Soc., Chem. Commun. 1998, 9, 1053-1054. (j) Corey, E. J.; Noe,
M. C.; Grogan, M. J. Tetrahedron Lett. 1996, 37, 4899-4902. (k) Kolb,
H. C.; Andersson, P. G.; Sharpless, K. B. J. Am. Chem. Soc. 1994, 116,
1278-1291. (l) Wang, L.; Sharpless, K. B. J. Am. Chem. Soc. 1992, 114,
7568-7570. (m) Hiemstra, H.; Wynberg, H. J. Am. Chem. Soc. 1981, 103,
417-430.
(28) Even after filtration, solubilized salts may remain.
(29) Hegedus has found that trialkylammonium salts can interfere with the
diastereoselectivity of the Staudinger reaction to produce â-lactams:
Hegedus, L. S.; Montgomery, J.; Narukawa, Y.; Snustad, D. C. J. Am.
Chem. Soc. 1991, 113, 5784-5791.
(30) The pKa of the conjugate acid of BEMP in DMSO is 16.2. See: O’Donnell,
M. J.; Delgado, F.; Hostettler, C.; Schwesinger, R. Tetrahedron Lett. 1998,
39, 8775-8778.
(31) Schwesinger, R.; Willaredt, J.; Schlemper, H.; Keller, M.; Schmitt, D.; Fritz,
H. Chem. Ber. 1994, 127, 2435-2454.
(32) Hafez, A. M.; Taggi, A. E.; Wack, H.; Drury, W. J., III; Lectka, T. Org.
Lett. 2000, 2, 3963-3965.
(33) Wack, H.; Taggi, A. E.; Hafez, A. M.; Drury, W. J., III; Lectka, T. J. Am.
Chem. Soc. 2001, 123, 1531-1532.
(34) For scaled up reactions, mechanical agitation of the polymer in the addition
funnel is beneficial.
(35) Hafez, A. M.; Taggi, A. E.; Dudding, T.; Lectka, T. J. Am. Chem. Soc.
2001, 123, 10853-10859.
(40) Tidwell has recently used our proton sponge shuttle base methodology to
observe and study the reactivity of bisketenes: Allen, A. D.; Fenwick, M.
H.; Jabri, A.; Rangwala, H.; Saidi, K.; Tidwell, T. T. Org. Lett. 2001, 3,
4095-4098.
(36) (a) Bell, R. P. The Proton in Chemistry; Cornell University Press: Ithaca,
NY, 1973. (b) For discussions concerning proton sponge basicity, see:
Alder, R. W. Chem. ReV. 1989, 89, 1215-1270.
9
6628 J. AM. CHEM. SOC. VOL. 124, NO. 23, 2002

Catalytic, Asymmetric Synthesis of â-Lactams
A R T I C L E S
Scheme 5. Sodium Hydride Shuttle Deprotonation System
Figure 1. Schematic of the double reaction flask used with K2CO3.
Scheme 4. Carbonate Shuttle Deprotonation System
for monosubstituted ketene formation from acid halides 1 using
a shuttle deprotonation strategy.41 Carbonate salts are very
inexpensive commodity chemicals,42 they are nontoxic, and they
are easily disposed of after use. Their insolubility in organic
media can also be used to advantage.43 Fine mesh K2CO3 is in
contact with the organic phase, in which a small amount of a
chiral nucleophile and an acid chloride are dissolved (Scheme
4). The catalytic shuttle base dehydrohalogenates the acid
chloride to form ketene 2 and the ammonium salt of the base.
The carbonate then deprotonates the ammonium salt, thus
regenerating the shuttle base. Once the ketene has formed
stoichiometrically, it can easily be separated from the solid phase
by filtration if necessary.
We have simplified the above process by designing a piece
of glassware that facilitates the ketene generation experiment
(Figure 1). The apparatus consists of two recovery flasks linked
by a fritted disc. The disc is located as low as possible in the
assembly to allow it to be conveniently submerged in a cold
bath. Simply by canting the assembly in one direction, the
reaction liquor can be transferred from one side to another
without removing the piece from the bath.
formation of the relatively stable diphenylketene both visually
and spectroscopically. The use of NaH results in a much cleaner
reaction, with NaCl and H2 as benign byproducts. Our
working hypothesis for the mechanism is presented in Scheme
5. Once again, a catalytic shuttle base, such as BQ, effects
dehydrohalogenation of an acid chloride to the corresponding
ketene. The acid chloride either forms an acylammonium salt
with BQ and is deprotonated by NaH or the BQ deprotonates
the acid chloride (possibly as its ammonium salt) and then
shuttles the proton to the NaH, thus regenerating BQ for another
catalytic cycle.
Catalyzing the Reaction. Having established several methods
of ketene synthesis, we turned our attention to â-lactam forming
reactions. In our preliminary work, we reported the catalytic,
asymmetric reaction of imine 5a46 and ketenes 2 to produce
â-lactams in high enantio- and diastereomeric excess, employing
chiral cinchona alkaloid derivatives as catalysts.11 Due to the
fact that imine 5a is electrophilic, the background rate of reaction
with various ketenes is fortunately minimal. Imine electrophi-
licity thus necessitates an “umpolung” of ketene polarity; hence
reaction with a nucleophile converts the ketene into the putative
zwitterionic enolate. Due to this role reversal, this reaction is
mechanistically distinct from the Staudinger reaction and thus
should not be referred to as such (an idea kindly pointed out
by the referees of our original publication). In an initial screen,
we found that a diverse array of catalysts, including metal-based
nucleophiles (Cp2Co, NaCo(CO)4),47 tributyl phosphite, and
amine-based nucleophiles such as triethylamine, promote the
reaction of a ketene with R-imino ester 5a. Initially we chose
to examine catalyzed diastereoselectivity. For this purpose we
chose methylphenylketene 2a, which reacted with imine 5a to
We recently improved upon our use of solid inorganic bases
for ketene generation by developing a procedure using inex-
pensive NaH.44 We found that using 10 mol % 15-crown-5 as
a phase transfer cocatalyst to help solubilize the NaH,45 in
addition to 10 mol % BQ, allowed for the efficient formation
of ketenes. Ketene formation was confirmed by observing the
(41) Hafez, A. M.; Taggi, A. E.; Wack, H.; Esterbrook, J.; Lectka, T. Org. Lett.
2001, 3, 2049-2051.
(42) Fine mesh size K2CO3 is available from Aldrich and many other comp-
anies.
(43) (a) Nelson, A. Angew. Chem., Int. Ed. Engl. 1999, 38, 1583-1585. (b)
Combined amine/carbonate base application: Tanabe, Y.; Yamamoto, H.;
Yoshida, Y.; Miyawaki, T.; Utsumi, N. Bull. Chem. Soc. Jpn. 1995, 68,
297-300. (c) Sasson, Y.; Bilman, N. J. Chem. Soc., Perkin Trans. 2 1989,
2029-2033.
(44) Taggi, A. E.; Wack, H.; Hafez, A. M.; France, S. Org. Lett. 2002, 4, 627-
629.
(46) We have used R-imino ester 5a in the catalytic, asymmetric synthesis of
R-amino acid derivatives: (a) Drury, W. J. III; Ferraris, D.; Cox, C.; Young,
B.; Lectka, T. J. Am. Chem. Soc. 1998, 120, 11006-11007. (b) Ferraris,
D.; Young, B.; Cox, C.; Drury, W. J. III; Dudding, T.; Lectka, T. J. Org.
Chem. 1998, 63, 6090-6091. (c) Ferraris, D.; Young, B.; Dudding, T.;
Lectka, T. J. Am. Chem. Soc. 1998, 120, 4548-4549.
(45) In the case of â-lactams, the use of more polar solvents such as propionitrile
and THF was found to decrease the diastereoselectivity greatly, possibly
due to solubilized salts.
(47) Wack, H.; Drury, W. J. III; Taggi, A. E.; Ferraris, D.; Lectka, T. Org.
Lett. 1999, 1, 1985-1988.
9
J. AM. CHEM. SOC. VOL. 124, NO. 23, 2002 6629

A R T I C L E S
Taggi et al.
Table 1. Diastereoselectivity in the Nucleophile-Catalyzed
Reaction of Methylphenylketene 2a with Imine 5a
a Yield after column chromatography. b dr (cis/trans) was determined by
1H NMR of the crude residue.
Figure 2. Representative â-lactams synthesized using proton sponge.
form the cis and trans diastereomers (Table 1 and eq 3). We
identified two sets of catalysts: one that produced predominantly
the trans â-lactam, and another group that produced the cis
diastereomer of â-lactam 6a. Despite the fact that simple tertiary
°C. To further extend the scope and applicability of this reaction,
we turned our attention to more reactive, synthetically useful
monosubstituted ketenes. For example, the use of Hu¨nig’s base
to generate phenylketene in situ afforded product 6c in low yield
and enantioselectivity with BQ catalyst 4a at -78 °C in THF
or toluene. It was obvious to us that Hu¨nig’s base was effectively
competing with BQ as a catalyst in a nonselective fashion. At
this point the necessity for a nonnucleophilic, yet thermody-
namically strong base became apparent and as a result, we
devised the shuttle base strategy for in situ ketene synthesis.
We employed proton sponge 3b as a nonnucleophilic “proton
sink” and BQ (4a) as the shuttle base in toluene at -78 °C.
The reaction afforded cis-lactam 6c in 99% ee and good yield
(65% yield, 99/1 dr, eq 4). In this reaction, 4a plays two distinct
amines such as triethylamine usually catalyze the reaction in a
nonselective fashion, we found that a bifunctional catalyst (4h,
Table 1, entry 3) containing a nucleophilic center in tandem
with a hydrogen bond donor may lead to a potentially rigidified
activated complex affording products in a cis/trans ratio of
3:97.48 When a small amount of DMSO was titrated into the
reactions involving this catalyst, the reaction diastereoselectivity
eroded,49 implying a potential hydrogen bond interaction
between the N-H of the amide and the ketene adduct with the
nucleophilic center of the catalyst. Competitive H-bonding to
DMSO apparently disrupts the rigidity of the activated complex.
Surprisingly, we found that benzimidazole (4j) forms the
opposite diastereomer (98/2 dr, entry 7).
Enantioselective Reactions. Next we investigated how our
catalyst leads could be translated into a chiral environment. Due
to our success with amine-based nucleophiles, we decided to
investigate various alkaloids as potential catalysts. As a first
step, we found that benzoylquinine (BQ, 4a), a cinchona
alkaloid, catalyzes the addition of diphenylketene 2b to imino
ester 5a in 36% yield and 99% ee in toluene solvent at -78
catalytic roles: namely as a dehydrohalogenation agent and as
a nucleophilic catalyst. Additionally, we found that in all cases
using benzoylquinidine 4b instead of 4a inverted the stereo-
chemistry of the product. For example ent-6c was formed using
4b in similar yield and selectivity as when 4a was used (64%
yield, 99% ee, 99/1 dr, Figure 2).
To exemplify the wide array of employable functionalities,
â-lactams 6d-l were synthesized from a variety of in situ
generated ketenes in moderate to good yields (50-65%). All
the products are formed with high ee and dr (cis/trans ratio),
and were chosen in part for their potential utility in pharma-
ceutically active â-lactam synthesis (Figure 2). For example,
lactam 6d is an elastase inhibitor originally reported by DuPont-
(48) (a) Hydrogen bond contacts have been postulated to play similar roles in
other catalytic reactions: Ameer, F.; Drewes, S. E.; Freese, S.; Kaye, P.
T. Synth. Commun. 1988, 18, 495-500. (b) Vasbinder, M. M.; Jarvo, E.
R.; Miller, S. J. Angew. Chem., Int. Ed. 2001, 40, 2824-2827.
(49) For example, as DMSO is titrated in, the reaction diastereoselectivity erodes
dramatically. H-bonding between DMSO and the catalyst can be discerned
in the IR spectrum.
9
6630 J. AM. CHEM. SOC. VOL. 124, NO. 23, 2002

Catalytic, Asymmetric Synthesis of â-Lactams
A R T I C L E S
Table 2. A Comparison of the Four Ketene Generation Methods
for 6i
c
entry
base
rel costa
% yieldb
%ee
drd
1
2
3
4
BEMP
proton sponge/BQ
K2CO3/BQ
863
11
2
60
57
56
60
99
99
93
99
99/1
99/1
8/1
NaH/15-crown-5/BQ
1
25/1
a Relative cost based on the least expensive $/g from the 2000/2001
Sigma Aldrich catalog. b Isolated yield after column chromatography.
c Enantioselectivity determined by chiral HPLC. d dr (cis/trans) was deter-
Figure 3. Homology between the structure of achiral bifunctional catalyst
and the quinuclidine and amide groups of 4c.
1
mined by H NMR of the crude residue.
Scheme 6. The Use of N,O-Acetals To Form â-Lactams
Merck that has undergone clinical trials.50 Furthermore, while
3-oxo-substituted ketenes are not amenable to the standard chiral
auxiliary methodology, our process allows acetoxy, phenoxy,
and benzyloxy â-lactams (6g-i) to be synthesized in good yield
and with high selectivity. The azido-substituted â-lactam 6k
allows for the synthesis of 3-amino-substituted â-lactams, which
are prevalent in pharmaceutical chemistry.51 As a third example
of a heteroatom substituent, we were able to use bromoacetyl
chloride to form the corresponding â-lactam 6l with similar
selectivity and yield (96% ee, 98/2 dr). Formation of vinyl-
substituted â-lactam 6j (98% ee, 99/1 dr) occurs in preference
to a theoretically possible [4+2] cycloaddition to form a six-
membered lactam product. Enantiomers (for example, ent-6c)
of our â-lactam compounds can be made with similar selec-
tivity using the “pseudoenantiomeric” benzoylquinidine 4b. The
azido group on â-lactam 6k provides a convenient handle for
functionalization reactions. The azido group is known to be
easily converted with retention of configuration to an amine or
an amide substituent by catalytic hydrogenation followed by
acylation (eq 5).52
eleven times less expensive than proton sponge, and oVer 850
times less expensive than BEMP.
Due to its similarity with trans-selective catalyst 4h (entry
3, Table 1), quinine amide 4c, made in three steps from quinine,
was also investigated (Figure 3). We found that 4c behaved
similarly with respect to BQ in several reactions. For example,
phenyl-substituted â-lactam 6b, was obtained with somewhat
decreased selectivity (89% ee, 10:1 dr) although not as the trans
diastereomer. It appears that the other substituents on the
skeleton of the alkaloid catalyst are effectively controlling the
diastereoselectivity.
For many purposes tosyl groups are not ideal substituents
for the nitrogen atom of â-lactams. Other sulfonyl groups would
be more desirable for many applications. Along these lines, we
have developed an alternative procedure for conducting the
reaction that potentially allows for the use of other sulfonyl
substituents (Scheme 6). Beginning with easy-to-make N,O-
acetal 7 (readily available from glyoxylate and sulfonamide) in
situ silylation with TMSCl presumably leads to intermediate 8.
We found in an earlier work that 8 can be converted to the
imine in situ upon which it can react in an enantioselective
fashion to form the desired products.53 For example, 7 leads to
product 6c (58% yield, 95% ee, and 8/1 dr). In this work, we
successfully employed a variety of different sulfonyl groups in
our N,O-acetals.
Similarly, the Ts group of many of our â-lactam products
can be removed by treatment with 2 equiv of SmI2 in THF to
afford deprotected â-lactams such as 6n in high yield (eq 6).
We compared our other three ketene generation methods to
determine the possible effect on yield and selectivity. As
expected, we found that the use of BEMP provided product 6i
with similar results (Table 2, entry 2, 58% yield, 99% ee, 99/1
dr) to that of proton sponge. When K2CO3 was used as the
stoichiometric base, the enantioselectivity and diastereoselec-
tivity were decreased somewhat (entry 3, 56% yield, 93% ee,
8/1 dr). Finally, we found that the use of NaH led to 6i with
similarly high selectivity (entry 4, 60% yield, 99% ee, 25/1 dr).
Of the four stoichiometric bases employed NaH was by far the
least expensive, being two times less expensive than K2CO3,
In an effort to improve the overall yield, we investigated the
effect of Lewis acids on the standard reaction of 1c and 5a
catalyzed by BQ. To our surprise, we found that the simple
addition of 10 mol % of inexpensive In(OTf)3 to the 10 mol %
BQ/proton sponge shuttle system afforded â-lactam 6c in 92%
yield and comparable selectiVity (98% ee, 60:1 dr).54
Stereochemical Models. Given the size of the putative
reactive complexes between the cinchona alkaloid derivatives
and ketenes, their conformational flexibility, and all-organic
nature, molecular mechanics (MM) calculations were judged
(50) Firestone, R. A.; Barker, P. L.; Pisano, J. M.; Ashe, B. M.; Dahlgren, M.
E. Tetrahedron 1990, 46, 2255-2262.
(53) Ferraris, D.; Dudding, T.; Young, B.; Drury, W. J., III; Lectka, T. J. Org.
Chem. 1999, 64, 2168-2169.
(54) A preliminary screen shows that this yield enhancement holds true for
aliphatic and oxygen-substituted ketenes as well France, S.; Wack, H.;
Hafez, A. M.; Taggi, A. E.; Witsil, D. R.; Lectka, T. Org.Lett. 2002, 4,
1603-1605.
(51) Azidoacetyl chloride has previously been used to make azido-substituted
â-lactams by Bose. See ref 21.
(52) (a) Shibuya, M.; Jinbo, Y.; Kubota, S. Chem. Pharm. Bull. 1982, 32, 1303-
1312. (b) Ojima, I.; Suga, S.; Abe, R. Tetrahedron Lett. 1980, 21, 3907-
3910.
9
J. AM. CHEM. SOC. VOL. 124, NO. 23, 2002 6631

A R T I C L E S
Taggi et al.
Figure 5. Stereochemical models of the putative zwitterionic intermediate
of BEQ (4e) and phenylketene.
Figure 4. Stereochemical models of the putative zwitterionic intermediates
of BQ, deoxy BQ, and phenylketene.
to be the best method to formulate useful stereochemical models
for the â-lactam forming reactions. For example, Monte Carlo
calculations employing the Macromodel program (AMBER
force field) on the putative zwitterionic intermediate of BQ and
phenylketene afforded several closely structurally related en-
ergetic minima represented by ketene-4a, a ketene-catalyst
adduct in which the quinine moiety is preferably trans to the
substituent across the CdC bond (Figure 4). The model shows
that the re-face (top face) of the ketene CdC bond (highlighted
in green) is completely open to the approach of an electrophile.
The model correctly predicts the stereoregular sense of induction
observed in the â-lactam forming reactions, and is consistent
with other models proposed for cinchona alkaloid catalyzed
processes.24 In stark contrast, the lowest energy minimum in
which the si-face is exposed is calculated to be almost 7 kcal/
mol higher in energy.
Figure 6. Stereochemical models of the putative zwitterionic intermediate
of BC and phenylketene.
Another surprising effect that we observed concerns the
catalyst benzoylcinchonidine (BC, 4f), in which the methoxy
group of complex ketene-4a is absent. MM calculations on the
complex of 4f with phenylketene (Figure 6) reveal two very
closely spaced low-energy minima. One minimum has its si-
face exposed, the other, only 0.13 kcal higher in energy, has its
re-face exposed. For all purposes, the calculation predicts that
a racemic â-lactam product should result. In this event we found
that the product of the reaction of imino ester 5a with
phenylketene is indeed racemic. We therefore conclude that the
presence of the methoxy group is critical for selectiVity.
Operationally, it appears that the methoxy group serves to anchor
the quinoline moiety proximate to the ketene-derived enolate.
When the methoxy group is absent, the quinoline moiety may
spin away to another isoenergetic conformation, allowing the
face of the ketene to flip. We have proposed that quinine amide
4c engages in hydrogen bonding with the ketene enolate in the
activated complex ketene-4c, yielding product with the same
sense of induction as BQ. Macromodel calculations were
performed on model complex ketene-4c, and a low-energy
complex was found in which an intramolecular hydrogen bond
stabilized the oxygen of the ketene enolate, leaving the re-face
exposed to attack of an electrophile (Figure 7). The correspond-
ing low-energy structure with the si-face exposed is 3.84 kcal/
mol higher in energy. The energy difference between the two
faces is a factor of 2 smaller than that of BQ, explaining the
decreased selectivity when using 4c (89% ee). This calculation
suggests, however, that hydrogen bonding can be used as an
organizing principle in the catalyst design of ketene reactions.
The performance of various cinchona alkaloid derivatives in
Similar calculations on the adduct of phenylketene and
deoxyquinine 4d predict the same sense of induction (as
observed, Figure 4) except for the fact that the lowest si-face
conformation is now about 2 kcal/mol higher in energy, a
smaller gap reflected in the diminished ee (72%) that we
observed. This calculation suggests that the presence of the
“oxy” stereogenic center in BQ is not critical to the sense of
induction, and it is the steric bulk of the benzoyl group that is
most critical in enhancing selectivity.
To confirm the validity of this conclusion, we investigated
the consequence of epimerizing the “oxy” stereogenic center R
to the ester oxygen (affording benzoylepiquinine, or BEQ).
Macromodel calculations predict the same sense of induction
as that seen in BQ-catalyzed reactions. As seen in model ketene-
4e, the re-face of the ketene zwitterion is still exposed (Figure
5). The lowest energy corresponding si-face exposed conformer
is some 6.69 kcal/mol higher in energy. When we performed a
â-lactam forming reaction using BEQ (phenylacetyl chloride
1c, imino ester 5a) under standard conditions with proton sponge
as the stoichiometric base, we obtained a 57% yield of the cis
diastereomer 6c in 97% ee, confirming our prediction. Thus,
we can conclude that the orientation of the “oxy” center is not
critical to selectivity.
9
6632 J. AM. CHEM. SOC. VOL. 124, NO. 23, 2002

Catalytic, Asymmetric Synthesis of â-Lactams
A R T I C L E S
Scheme 7. Proposed Mechanism of â-Lactam Formation with
Proton Sponge
Figure 7. Stereochemical model of the zwitterionic intermediate when using
benzamidoquinine 4c.
Table 3. Summary of the Selectivities Imparted by Cinchona
Alkaloid-Derived Catalysts
several kilocalories higher in energy than the cis, reflecting our
experimental results. In this case, the CdC enolate bond and
the CdN imine bond are gauche, rather than anti to each other,
once again an orientation that has been proposed to be higher
in energy. The alternative trans “anti” conformation suffers from
apparent close contact of the carboethoxy and phenyl groups.
Mechanism of â-Lactam Formation with Proton Sponge
as the Base. We investigated the mechanism of the â-lactam
formation reaction, with proton sponge as the base, through
rate studies and deuterium-labeling experiments. The studies
outlined below allow us to propose a detailed mechanism,
shown in Scheme 7. We were especially interested in two
questions: namely, what is the rate determining step of the
reaction, and are ketenes necessarily formed in all cases, or is
it possible that in certain instances, the reaction preempts ketene
formation to proceed directly to the â-lactam from a zwitterionic
intermediate?
catalyst
% ee
99
dr
benzoylquinine 4a
99/1
99/1
99/1
10/1
2/1
benzoylquinidine 4b
benzoyl-epi-quinine 4e
benzamidoquinine 4c
deoxyquinine 4d
99 (ent)
99
89
72
5
benzoylcinchonidine 4f
5/1
Figure 8. Stereochemical models of cis and trans diastereoselectivity.
For example, if k4[zwitterion][imino ester] > k3[zwitterion]
(where k3 encompasses the possible rate dependence on any
other species) then free ketene will not, or need not, be formed
to any appreciable extent. Of course, if the imino ester 5a is
withheld from the reaction, then free ketene should be formed.
By varying the concentrations of acid chloride, imine, proton
sponge, and catalyst, we found that the initial rate of reaction
is dependent on the concentrations of acid chloride and catalyst.
The concentrations of proton sponge and imine have no effect
on the rate of product formation. A plausible rate determining
step for the reaction involves one of three scenarios: (1)
formation of the acylammonium salt as an initial intermediate,
(2) a concerted dehydrohalogenation of the acid chloride by
BQ to yield the ketene directly, or (3) the (rate determining)
deprotonation of the acylammonium salt by a second molecule
of BQ. Scenario 3 can be ruled out since we know that the rate
of reaction is first order in BQ (we would expect second order
behavior for this step in the former scenario) and zero order in
proton sponge. The first and second scenarios are more difficult
to distinguish, although the second should give rise to a
substantial primary kinetic isotope effect (KIE) for an irreVers-
ible ketene forming reaction.
our reactions is summarized in Table 3. Depending on the nature
of substituents on the alkaloid catalyst, selectivities range from
>99% ee to virtually racemic. As a final comment, these
calculations demonstrate that the selectivity of cinchona alkaloid
catalyzed ketene reactions can, in certain circumstances, be
predicted with some accuracy. They also suggest that on this
basis other effective nucleophilic catalysts can be designed and
synthesized as desired.
Models for Diastereoselectivity. One of the hallmarks of
these BQ-catalyzed reactions is predominant cis diastereose-
lectivity for a wide variety of ketenes. To shed light on these
results, we performed MM calculations (Macromodel, AMBER
force field) in which we dock imino ester 5a to the zwitter-
ionic enolate (derived from BQ and phenylketene) at 2.2 Å
to represent, albeit in the crudest fashion, a transition state
distance.
The lowest energy assembly is derived from interaction with
the re-faces of 4a and 5a, leading to the cis product. The
structures represented in Figure 8 show interaction between the
enolate, in model form, and 5a. The imine CdN bond in the
cis assembly is antiperiplanar to the enolate CdC bond (in
green), a low-energy “open” transition state orientation is
proposed to account for the diastereoselectivity of the Mu-
kaiyama aldol reaction.55 The lowest energy trans assembly is
Consequently, we sought also to illuminate mechanistic details
by a series of kinetic isotope studies and deuterium-labeling
experiments. We measured a KIE of only 1.1 for the reaction
of 1c-d2 as compared to 1c, a fact which, upon the surface,
appears to rule out scenario 2. To our surprise, when 2,2-d2-
phenylacetyl chloride (1c-d2) was allowed to react to form
â-lactam product to completion under standard conditions, the
(55) (a) Otera, J.; Fujita, Y.; Sakuta, N.; Fujita, M.; Fukuzumi, S. J. Org. Chem.
1996, 61, 2951-2962. (b) Denmark, S. E.; Lee, W. J. Org. Chem. 1994,
59, 707-709.
9
J. AM. CHEM. SOC. VOL. 124, NO. 23, 2002 6633

A R T I C L E S
Taggi et al.
products contained only trace amounts of deuterium (eq 7).
However, when 1c-d2 was treated with BEMP resin to form
phenylketene directly, labeled product 6c-d1 was isolated in good
reaction where we replaced BQ with a nucleophile that is
unlikely to act as a competent base. When the reaction was run
under standard conditions with PPh3 (instead of BQ), proton
57
sponge, and 1c, only a trace of product 6c was observed. This
lends credence to the idea that a second molecule of BQ could
be acting as a shuttle base and transferring the proton from the
acylammonium salt to proton sponge, regenerating the catalyst.
We have also examined the formation of diphenylketene from
proton sponge and BQ by ReactIR. In the absence of imine 5a,
an equilibrium between acid chloride and free ketene is
established over time, as witnessed by the observed IR frequen-
cies for the ketene and acyl chloride CdO stretches, a spec-
troscopic conformation of the kinetic data that we assembled
for phenylketene. If the experiment is performed under identical
conditions in the presence of imine, no free ketene is observed
(although we obviously cannot strictly rule out its intermediacy
in low concentrations). What allows us to conclude that free
ketene need not be involved in this particular reaction is the
fact that the rate of ketene formation in experiment one is less
than the rate of product formation in experiment two. Whether
this conclusion holds for all manner of ketenes cannot be
concluded at this point. HoweVer, we can conclusiVely state that
scenario 2, which requires the formation of free ketene, cannot
be the predominate reaction mode in this case. Although the
question of whether free ketene is formed in the reaction is
mainly semantic, there could be a practical consequence. For
example, if free ketenes are not formed, then reactions such as
ketene dimerization, which in some cases may require the
presence of free ketene, should be suppressed.
yield (eq 8). We interpret this result to imply that, with PS as
the stoichiometric base, the ketene (or zwitterionic enolate) is
formed reversibly. The pKa difference between the enolate and
proton sponge is expected to be small enough in an organic
solvent to ensure reversibility. It should be noted that when
scrupulous regard is paid to excluding moisture, the reaction
of eq 7 yields 6c-d1 predominately. Consequently, the isotope
effect of 1.1 that we measured is more likely a normal
equilibrium effect than a KIE (involving formation of an sp2-
bound H or D) and scenario 2 cannot be ruled out on this basis.
Precedent suggests that acylammonium salt intermediates are
involved in the synthesis of the majority of ketenes; however,
for acid halides possessing certain electron-withdrawing groups
in the R-position, direct dehydrohalogenation (scenario 2) may
operate.37,56
We can conclude that in the presence of trace amounts of
water (or other Brønsted acids HX), the deuteron in PS‚DCl
exchanges for a proton (Scheme 8), forming PS‚HCl, which
can protonate the enolate, thus effecting the exchange. Under
the conditions of our reaction, we obviously have enough
adventitious HX or water present to promote an almost complete
exchange. When solid-phase based BEMP (a powerful base) is
employed, ketene is formed irreversibly, so the deuterium label
is preserved in the product.
Spectroscopic Observation of a Zwitterionic Enolate.
Reaction of a ketene and an amine base to form a zwitterionic
intermediate is a reversible process whose equilibrium constant
should be dependent on the basicity of the amine as well as the
electrophilicity of the ketene. Considering the reaction of BQ
with diphenylketene, for example, the equilibrium constant is
almost certainly much less than one, making the derived enolate
difficult to observe spectroscopically (eq 9). To observe stable,
Scheme 8. R-Proton Exchange with Proton Sponge
characterizable zwitterionic intermediates, we employed the
much more electrophilic ketene 2p in order to favor formation
of the intermediate. Dicarboethoxyketene 2p is known to form
adducts with DMAP,58 and consequently we studied its reaction
with BQ through IR and NMR spectroscopy. A preliminary DFT
calculation we performed (B3LYP/6-31G*) revealed a stable
adduct 9 of an electronically similar electrophilic ketene with a
quinuclidine core, suggesting the viability of an experiment to
observe the enolate intermediate directly. Interestingly, the
distance from the quinuclidine N to the ketene carbonyl carbon
As a confirming experiment to determine if a second molecule
of BQ could be acting as a shuttle base (scenario 1), we ran a
(56) Brady, W. T.; Scherubel, G. A. J. Org. Chem. 1974, 39, 3790-3791. (b)
Brady, W. T.; Scherubel, G. A. J. Am. Chem. Soc. 1973, 95, 7447-7449.
(c) Walborsky, H. M. J. Am. Chem. Soc. 1952, 74, 4962-4963.
(57) We had previously found that triphenylphosphine catalyzed the reaction
of 2b with 5a.
(58) Gompper, R.; Wolf, U. Liebigs Ann. Chem. 1979, 1388-1405.
9
6634 J. AM. CHEM. SOC. VOL. 124, NO. 23, 2002

Catalytic, Asymmetric Synthesis of â-Lactams
A R T I C L E S
ketenes can be precursors in the reaction (as we believe they
must be for the carbonate and hydride shuttle deprotonation
procedures). For example, when diazoketone 10 is photolyzed,
it produces phenylketene 2c through a standard Wolff rear-
rangement (eq 10). We photolyzed 10 in the presence of imino
ester 5a and BQ catalyst to form lactam 6c in 52% yield (dr
5/1, ee 93%). The yield and product selectivity obtained are
similar, if slightly less than the proton sponge promoted
experiments, and suggest that zwitterionic intermediates can be
formed from chemically distinct precursors.
Conclusion
We have reported a catalytic, asymmetric route to â-lactams
using chiral amines to catalyze a reaction between ketenes (or
derived zwitterionic enolates) and corresponding electron-
deficient imines. Kinetics experiments, labeling studies, and
direct observation of intermediates, as well as stereochemical
models, have helped to formulate a proposed mechanism for
the reaction. Further work will center on a trans-selective
asymmetric â-lactam forming process.
Figure 9. Diagnostic ketene stretch at 2155 cm-1 for the 4a-2p adduct.
Experimental Section
General Procedure for â-Lactams 6 with Proton Sponge.61 To a
solution of phenylacetyl chloride (20 mg, 0.129 mmol) in toluene (0.5
mL) at -78 °C62 was added proton sponge 3b (31 mg, 0.142 mmol) in
toluene (0.5 mL) immediately followed by benzoylquinine 4a (6 mg,
0.0129 mmol) and R-imino ester 5a (33 mg, 0.129 mmol). The reaction
was allowed to stir for 5 h as it slowly warmed to room temperature.
The solvent was removed under reduced pressure and the crude mixture
was subjected to column chromatography (15% EtOAc/hexanes) on a
plug of silica gel to yield 6c (65% yield, 33 mg).
is calculated to be long (1.70 Å), whereas the zwitterionic C-O-
bond in 9a is very short (1.19 Å)svirtually a CdO bond, facts
perhaps better represented by 9b. Other adducts derived from
less electrophilic ketenes are often not computationally stable
and dissociate during the calculation.
A solution of the ketene in toluene revealed a carbonyl stretch
at 2155 cm-1 (Figure 9). As BQ was added, this peak diminished
in intensity, and the ester carbonyl peaks shifted to lower
wavenumber, consistent with formation of a zwitterion.59 A new
resonance of strong intensity was observed at 1730 cm-1, in
good agreement with the calculated vibrational frequency of
1741 cm-1 (scaled)60 for a coupled stretch of an ester carbonyl
group and the “C-O” bond of the ketene zwitterion. The
reaction of BQ with ketene 2p is not instantaneous, but
equilibrium is reached after about 5 min; whereupon the IR
spectrum remains stable, contraindicating a catalytic chemical
reaction. Quenching the reaction with MeOH affords a quantita-
tive yield of the corresponding triester, affirming that another
competing chemical process is not occurring.
In summary, the kinetics experiments, precedent, and IR
spectroscopic data we have gathered are all consistent with the
formation of an acylammonium salt in the rate determining step
and subsequent zwitterion formation, followed by fast reaction
with imine 5a (Scheme 7). Formation of the crucial zwitterionic
intermediate is reversible, and putative â-lactam ring closure
and ejection of the catalytic nucleophile to form the product
also must be fast steps in the reaction.
Acknowledgment. T.L. thanks the NIH, DuPont, Eli Lilly,
the NSF Career Program for support, the Dreyfus Foundation
for a Teacher-Scholar Award, and the Alfred P. Sloan Founda-
tion for a Fellowship. H.W. thanks Johns Hopkins for a
Chambers and Sonneborn Fellowship. A.T. thanks the Organic
Division of the American Chemical Society for a Graduate
Fellowship sponsored by Organic Reactions, Inc. (2001-2002).
The authors also thank Professor Kenneth Karlin for use of his
ReactIR spectrometer, Professor John Toscano for use of his
Rayonet reactor, and Drs. Victor G. Young, Jr., and Maren Pink
(Minnesota) for obtaining the crystal structure of 6j.
Supporting Information Available: General procedures for
using BEMP, K2CO3, and NaH as the stoichiometric base, com-
pound characterization, and kinetic data (PDF). This material
is available free of charge via the Internet at http://pubs.acs.org.
JA0258226
(61) A large-scale reaction (2.6 mmol) was carried out to form 6c by multiplying
the amounts of reagents and solvents by a factor of 20. Similar values for
ee, dr, and yield were received.
(62) It was found that the optimal conditions for in situ generation of ketenes
2b, 2h, and 2i were to stir the solution of proton sponge and acid chloride
at 0 °C for 30 min and then reduce the temperature to -78 °C proceeding
with the standard reaction conditions. These conditions better effect
dehydrohalogenation for more stable ketenes.
Photochemical Generation of Phenylketene. Another ex-
periment conclusively determined that free monosubstituted
(59) See Supporting Information for the full IR spectrum.
(60) Scott, A. P.; Radom, L. J. Phys. Chem. 1996, 100, 16502-16513.
9
J. AM. CHEM. SOC. VOL. 124, NO. 23, 2002 6635