Chiral phosphoramide-catalyzed aldol additions of ketone enolates. Preparative aspects

Article abstract of DOI:10.1021/ja984123j

Trichlorosilyl enolates of ketones (enoxytrichlorosilanes) were demonstrated to be highly reactive aldol addition reagents. Trichlorosilyl enolates of cyclohexanone (E-enolate) and propiophenone (Z-enolate) reacted readily at room temperature with a wide

Full text of DOI:10.1021/ja984123j

4982
J. Am. Chem. Soc. 1999, 121, 4982-4991
Chiral Phosphoramide-Catalyzed Aldol Additions of Ketone Enolates.
Preparative Aspects
Scott E. Denmark,* Robert A. Stavenger, Ken-Tsung Wong, and Xiping Su
Contribution from the Roger Adams Laboratory, Department of Chemistry, UniVersity of Illinois,
Urbana, Illinois 61801
ReceiVed NoVember 30, 1998
Abstract: Trichlorosilyl enolates of ketones (enoxytrichlorosilanes) were demonstrated to be highly reactive
aldol addition reagents. Trichlorosilyl enolates of cyclohexanone (E-enolate) and propiophenone (Z-enolate)
reacted readily at room temperature with a wide variety of aldehydes to afford aldol addition products in high
yield and diastereoselectivity (E f syn, Z f anti). These reactions were shown to be highly susceptible to
acceleration by catalytic quantities of chiral phosphoramides. In particular, a phosphoramide derived from
(S,S)-stilbenediamine was remarkably effective not only in accelerating the reaction but also in modulating
the diastereoselectivity and in providing the aldol addition products in good to excellent enantioselectivity.
The diastereoselectivity of the unpromoted process has been interpreted as a consequence of reaction via a
pentacoordinate, trigonal bipyramidal (tbp) silicon complex through a boatlike transition structure. The
phosphoramide-catalyzed reactions are more complicated and are believed to proceed via hexacoordinate,
octahedral complexes through chairlike transition structures. A systematic examination of the influence of
solvent, concentration, addition rate, and catalyst loading on rate and stereoselectivity is also described.
Introduction
enolates4 and stable enoxysilane species.5 After a detailed
examination of the methods currently available for enantiose-
lective aldolizations, it was apparent that there was still an
opportunity for a fundamental conceptual advance that could
satisfy the long sought goal of high stereoselectivity from readily
available precursors under asymmetric catalysis. Our objective
was to develop a new type of aldol addition that embodied the
most advantageous features of existing approaches and also
obviated their limitations. Detailed herein is a full account of
our studies on the invention of an aldol addition reaction that
provides high diastereo- and enantioselectivity from easily
prepared6 (or in situ prepared)7d,f trichlorosilyl enolates (enoxy-
trichlorosilanes) in the presence of catalytic quantities of chiral
phosphoramides (Lewis bases).7 This paper is concerned
primarily with the preparative aspects of the reaction and its
current scope and limitations. The attendant studies on the
mechanism and origin of selectivity will be addressed in a
subsequent disclosure.
The aldol addition reaction has achieved the venerable status
of a “strategy-level reaction” in organic synthesis.1 The general-
ity, versatility, and selectivity associated with this construction
has been the subject of countless reviews and authoritative
summaries.2 In addition, the historians of chemistry will no
doubt recognize that the revolution in organic synthesis that
ushered in the era of reagent-controlled strategies for total
synthesis can be associated with the development of stereose-
lective aldol addition reactions.3
Our interest in the aldol addition has focused on an
understanding of the origins of stereochemical control in the
most common variants of the reaction involving both metal
(1) Norcross, R. D.; Paterson, I. Chem. ReV. 1995, 95, 2041.
(2) (a) Evans, D. A.; Nelson, J. V.; Taber, T. R. In Topics in
Stereochemistry; Eliel, E. L., Wilen, S. H., Eds.; Wiley-Interscience: New
York, 1982; Vol. 13, p 1. (b) Heathcock, C. H. In ComprehensiVe Carbanion
Chemistry; Buncel, E., Durst, T., Eds.; Elsevier: New York, 1984; Vol.
5B, p 177. (c) Heathcock, C. H. In Asymmetric Synthesis; Morrison, J. D.,
Ed.; Academic Press: New York, 1984; Vol 3, Chapter 2. (d) Mukaiyama,
T. Org. React. 1982, 28, 203. (e) Braun, M. Angew. Chem., Int. Ed. 1987,
26, 24. (f) Heathcock, C. H. In ComprehesiVe Organic Synthesis: Additions
to C-X π-Bonds Part 2; Heathcock, C. H., Ed.; Pergamon Press: Oxford,
1991; Chapter 1.5. (g) Heathcock, C. H. In ComprehesiVe Organic
Synthesis: Additions to C-X π-Bonds Part 2; Heathcock, C. H., Ed.;
Pergamon Press: Oxford, 1991; Chapter 1.6. (h) Kim, B. M.; Williams, S.
F.; Masamune, S. In ComprehesiVe Organic Synthesis: Additions to C-X
π-Bonds Part 2; Heathcock, C. H., Ed.; Pergamon Press: Oxford, 1991;
Chapter 1.7. (i) Rathke, M. W.; Weipert, P. In ComprehesiVe Organic
Synthesis: Additions to C-X π-Bonds Part 2; Heathcock, C. H., Ed.;
Pergamon Press: Oxford, 1991; Chapter 1.8. (j) Paterson, I. In Compre-
hesiVe Organic Synthesis: Additions to C-X π-Bonds Part 2; Heathcock,
C. H., Ed.; Pergamon Press: Oxford, 1991; Chapter 1.9. (k) Gennari, C. In
ComprehesiVe Organic Synthesis: Additions to C-X π-Bonds Part 2;
Heathcock, C. H., Ed.; Pergamon Press: Oxford, 1991; Chapter 2.4. (l)
Mukaiyama, T.; Kobayashi, S. Org. React. 1994, 46, 1. (m) Cowden, C. J.;
Paterson, I. Org. React. 1997, 51, 1.
To provide the conceptual framework for the invention of
the process described herein, we will first outline briefly the
status of the existing methods with particular attention to the
advantages and disadvantages which the new advance was
designed to address.
(4) Denmark, S. E.; Henke, B. R. J. Am. Chem. Soc. 1991, 113, 2177.
(5) (a) Denmark, S. E.; Lee, W. J. Org. Chem. 1994, 59, 707. (b)
Denmark, S. E.; Griedel, B. D.; Coe, D. M. J. Org. Chem. 1993, 58, 988.
(c) Denmark, S. E.; Chen, C.-T. Tetrahedron Lett. 1994, 35, 4327. (d)
Denmark, S. E.; Griedel, B. D.; Coe, D. M.; Schnute, M. E. J. Am. Chem.
Soc. 1994, 116, 7026. (e) Denmark, S. E.; Griedel, B. D. J. Org. Chem.
1994, 59, 5136.
(6) Denmark, S. E.; Stavenger, R. A.; Winter, S. B. D.; Wong, K.-T.;
Barsanti, P. A. J. Org. Chem. 1998, 63, 9517.
(7) (a) Denmark, S. E.; Winter, S. B. D.; Su, X.; Wong, K.-T. J. Am.
Chem. Soc. 1996, 118, 7404. (b) Denmark, S. E.; Winter, S. B. D. Synlett
1997, 1087. (c) Denmark, S. E.; Wong, K.-T.; Stavenger, R. A. J. Am.
Chem. Soc. 1997, 119, 2333. (d) Denmark, S. E.; Stavenger, R. A.; Wong,
K.-T. J. Org. Chem. 1998, 63, 918. (e) Denmark, S. E.; Stavenger, R. A.;
Wong, K.-T. Tetrahedron 1998, 54, 10389. (f) Denmark, S. E.; Stavenger,
R. A. J. Org. Chem. 1998, 63, 9524.
(3) Compare for example the strategic approaches to erythronolide B
and 6-deoxyerythronolide B. (a) Corey, E. J. et al. J. Am. Chem. Soc. 1978,
100, 4618, 4620. (b) Masamune, S.; Hirama, M.; Mori, S.; Ali, S. A.;
Garvey, D. S. J. Am. Chem. Soc. 1981, 103, 1568.
10.1021/ja984123j CCC: $18.00 © 1999 American Chemical Society
Published on Web 05/15/1999

Aldol Additions by Lewis Base Catalysis
J. Am. Chem. Soc., Vol. 121, No. 21, 1999 4983
Background
Catalytic processes have, however, been developed for the
aldol addition reaction, Scheme 212. These reactions take
advantage of the well-known Mukaiyama directed-aldol addition
reaction of enoxysilane derivatives of ketones, esters, thioesters,
and amides in combination with aldehydes activated by chiral
Lewis acids.2d,2k Some of the more commonly used and selective
chiral Lewis acids are: diamine complexes of tin(II) triflate,13
borane complexes of a monoester of tartaric acid (CAB
catalysts),14 sulfonamido amino acid borane complexes,15
titanium binaphthol16 and binaphthylimine complexes,17 ferro-
cenylphosphine-gold18a and BINAP-silver18b complexes, and
most recently, copper(II) bis-oxazoline complexes.19
Some of the earliest examples of asymmetric aldol addition
reactions involved lithium enolates of chiral carbonyl com-
pounds that reacted with aldehydes, presumably through orga-
nized transition structures to give good diastereoselectivities.8
Because the enolates were chiral, these translated to enantio-
merically enriched products once the auxiliaries were destroyed
or removed. Although high selectivities were obtained, these
reactions were not ideal from a practical point of view because
they required a stoichiometric amount of covalently bound
auxiliaries. Moreover, the highly reactive lithium enolates
employed do not necessarily require the pre-assembly of
aldehyde and enolate in the reactive intermediates or transition
structures4 which attenuates the stereochemical information
transfer from the covalently bound auxiliary.
Scheme 2
One of the greatest advances in aldol technology was the use
of less reactive metalloenolates (boron or titanium) which do
facilitate the association of aldehyde, enolate, and auxiliary in
the closed transition structure, Scheme 1.
These variants of the aldol reaction have a number of key
features in common: (1) the additions have been demonstrated
for aldehydes and enol metal derivatives with catalytic loading
of the chiral Lewis acid, (2) the diastereo- and enantioselectivity
is variable although can be high in certain cases, and (3) these
reactions are not responsive to prostereogenic features, i.e., when
the configuration of the enolsilane nucleophile changes, the
diastereoselectivity of the product does not change.20
Scheme 1
A very recently developed class of aldol addition involves
the use of chirally modified metalloids in a catalytic process.18b,21
In these reactions, a metal/phosphine complex is proposed to
undergo transmetalation with TMS enol ethers or tributylstannyl
ketones to provide chiral metalloid enolates in situ. Aldol
addition then proceeds, with turnover of the metalloid species
to another latent enol donor. Other approaches have used chiral
fluoride sources for Nakamura/Kuwajima/Noyori22 aldol addi-
Some of the most powerful reagents for asymmetric aldol
addition employ auxiliary modified enolates in which the chiral
appendage is attached through an acyl linkage or directly around
the metal of the enolate. Reactions of the geometrically defined
enolates with aldehydes give, with extremely high stereochem-
ical felicity, the syn- or anti-diastereomers with high enantio-
meric excess after cleavage of the controlling group. Some of
these powerful reagents are the acyl oxazolidinone boron
enolates,2a the diazaborolidine derived enolates,9 titanium eno-
lates derived from diacetone glucose,10 the diisiopinylcamphenyl
boron enolates for ketone aldolizations2m and proline-derived
silanes for N,O-ketene acetals.11
(12) For recent reviews of catalytic enantioselective aldol additions see:
(a) Nelson, S. G. Tetrahedron: Asymmetry, 1998, 9, 357. (b) Gro¨ger, H.;
Vogl, E. M.; Shibasaki, M. Chem.sEur. J. 1998, 4, 1137.
(13) Kobayashi, S.; Uchiro, H.; Shiina, I.; Mukaiyama, T. Tetrahedron
1993, 49, 1761.
(14) (a) Ishihara, K.; Maruyama, T.; Mouri, M.; Gao, Q.; Furuta, K.;
Yamamoto, H. Bull. Chem. Soc. Jpn. 1993, 66, 3483. (b) Sato, M.; Sunami,
S.; Sugita, Y.; Kaneko, C. Chem. Pharm. Bull. 1994, 42, 839.
(15) (a) Parmee, E. R.; Tempkin, O.; Masamune, S.; Abiko, A. J. Am.
Chem. Soc. 1991, 113, 9365. (b) Kiyooka, S.-i.; Kaneko, Y.; Komura, M.;
Matsuo, H.; Nakano, M. J. Org. Chem. 1991, 56, 2276. (c) Corey, E. J.;
Cywin, C. L.; Roper, T. D. Tetrahedron Lett. 1992, 33, 6907.
(16) (a) Mikami, K.; Matsukawa, S. J. Am. Chem. Soc. 1994, 116, 4077.
(b) Keck, G. E.; Krishnamurthy, D. J. Am. Chem. Soc. 1995, 117, 2363.
(17) (a) Singer, R. A.; Carreira, E. M.; Lee, W. J. Am. Chem. Soc. 1994,
116, 8837. (b) Carreira, E. M.; Singer, R. A. J. Am. Chem. Soc. 1995, 117,
12360.
(18) (a) Sawamura, M.; Ito, Y. In Catalytic Asymmetric Synthesis, Ojima,
I., Ed.; VCH: New York, 1993; pp 367-388. (b) Yanagisawa, A.;
Matsumoto, Y.; Nakashima, H.; Asakawa, K.; Yamamoto, H. J. Am. Chem.
Soc. 1997, 119, 9319.
(19) (a) Evans, D. A.; Kozlowski, M.; Murry, J.; Burgey, C.; Campos,
K.; Connel, B.; Staples, R. J. Am. Chem. Soc. 1999, 121, 669. (b) Evans,
D. A.; Burgey, C.; Kozlowski, M.; Tregay, S. J. Am. Chem. Soc. 1999,
121, 686.
(20) For examples of primarily anti-selective catalytic enantioselective
aldol additions, see: (a) Evans, D. A.; MacMillan, D. W. C.; Campos, K.
R. J. Am. Chem. Soc. 1997, 119, 10859. (b) Kobayashi, S.; Horibe, M.;
Hachiya, I. Tetrahedron Lett. 1995, 36, 3173.
(21) (a) Sodeoka, M.; Tokunoh, R.; Miyazaki, F.; Hagiwara, E.;
Shibasaki, M. Synlett 1997, 463. (b) Krueger, J.; Carreira, E. M. J. Am.
Chem. Soc. 1998, 120, 837.
(22) For leading references, see: (a) Kuwajima, I.; Nakamura, E.;
Shimizu, M. J. Am. Chem. Soc. 1982, 104, 1025. (b) Noyori, R.; Nishida,
I.; Sakata, J. J. Am. Chem. Soc. 1983, 105, 1598. (c) Nakamura, E.; Shimizu,
M.; Kuwajima, I.; Sakata, J.; Yokoyama, K.; Noyori, R. J. Org. Chem.
1983, 48, 932. (d) Hertler, W. R.; Reddy, G. S.; Sogah, D. Y. J. Org. Chem.
1988, 53, 3532.
The key features common to these agents are: (1) the metal
serves as an organizational center, (2) the electrophile, nucleo-
phile, and asymmetric modifier are held in close proximity
around the coordination sphere of the metal, ensuring high
stereochemical information transfer, (3) the geometry of the
enolate translates with high stereochemical responsiveness to
diastereoselectivity in the aldol product. Despite these powerful
advantages one of the most significant disadvantages is that these
reactions have never been rendered catalytic, and in fact it is
the high degree of metal affinity between aldehyde, enolate,
and chiral auxiliary that interferes with the turnover.
(8) (a) Masamune, S.; Ali, S. A.; Snittman, D. L.; Garvey, D. S. Angew.
Chem., Int. Ed. Engl. 1980, 19, 557. (b) Lynch, J. E.; Volante, R. P.; Wattley,
R. V.; Shinkai, I. Tetrahedron Lett. 1987, 28, 1385. (c) Oppolzer, W. Pure
Appl. Chem. 1988, 60, 39. (d) Kim, B. H.; Curran, D. P. Tetrahedron 1993,
49, 293.
(9) Corey, E. J.; Kim, S. S. J. Am. Chem. Soc. 1990, 112, 4976.
(10) Duthaler, R. O.; Hafner, A. Chem. ReV. 1992, 92, 807.
(11) (a) Myers, A. G.; Widdowson, K. L.; Kukkola, P. J. J. Am. Chem.
Soc. 1992, 114, 2765. (b) Myers, A. G.; Kephart, S. E.; Chen, H. J. Am.
Chem. Soc. 1992, 114, 7922.

4984 J. Am. Chem. Soc., Vol. 121, No. 21, 1999
Denmark et al.
Scheme 3
environment with single-point attachment. Candidates for the
G* Lewis basic group would include phosphine oxides and
derivatives such as phosphoramides or phosphonates, N-oxides,
and sulfoxides but not negatively charged alkoxides or amides
and carboxylates. It is viewed that these later groups would be
too nucleophilic and effect the cleavage of the O-MLn bond.5d
Thus, to reduce this to practice we envisioned the use of a
new class of aldol reagents, trichlorosilyl enolates, in conjunction
with perhaps the most Lewis basic of all of the groups
considered, the phosphoramides, Scheme 4. These can be seen
as chiral analogues of HMPA, the Lewis basicity of which is
well documented.25 To test the feasibility of this proposal we
needed a ready and efficient access to this unusual class of
enoxysilane derivatives, trichlorosilyl enol ethers.
Scheme 4
tions of ammonium enolates23 and heterobimetallic catalysts
which, through dual Lewis acid and Brønsted base activation,
promote the addition of unmodified ketones to aldehydes.24
We envisioned the possibility of devising a new aldol addition
reaction that uses Lewis base catalysis for activation of the
nucleophile. The challenges that face the development of
nucleophilic catalysis of the aldol addition are outlined in
Scheme 3. In this case it is the enoxymetal derivative which is
activated by preassociation with a chiral Lewis basic group (G*)
bearing a nonbonding pair of electrons. The -ate complex must
be more reactive than the free enolate for the ligand-accelerated
catalysis to be observed. Next, association of this -ate complex
with the Lewis basic carbonyl oxygen of the aldehyde produces
a hyper-reactive complex in which the metal has expanded its
valence by two. It is expected that this association complex
between enolate, aldehyde, and the chiral Lewis basic group
reacts through a closed-type transition structure with a high
degree of information transfer to produce the metal aldolate
product. This represents a single reaction event, and for turnover
to be observed, the aldolate must undergo the expulsion of the
Lewis basic G* group with the formation of the chelated metal
aldolate product. Thus, Lewis base-catalysis involves simulta-
neous activation of the nucleophile and the electrophile within
the coordination sphere of the metal. The reaction must take
place in a closed array and be capable of releasing the activating
group by chelation or change in the Lewis acidity.
To invent such a process one must consider the design criteria
for the enoxy metal and the G* group. For the metal, the MLn
subunit must be able expand its valence by two and balance
nucleophilicity of the enolate with electrophilicity to coordinate
both the Lewis basic aldehyde and the chiral G* group. Such
metals that would satisfy these criteria are those which can
expand their valence such as silicon, tin, titanium, zirconium,
and aluminum. To accommodate the valence expansion and
impart sufficient Lewis acidity to that metal group such that
two Lewis basic atoms may associate, the ligands (L) should
be small and strongly electron-withdrawing such as halogen or
carboxyl groups. The criteria necessary for the chiral Lewis basic
group G* are that it must be able to activate the addition without
cleaving the O-MLn linkage and provide an effective asymmetric
A variety of methods for the synthesis of chlorosilyl enolates
1-2 (Chart 1) from stannyl ketones, TMS enol ethers, and
directly from cyclohexanone have been recently discussed in
detail.6 Thus, the current presentation will focus on the
aldolization of these reagents, the experimental aspects of yield
and selectivity optimization (catalyst structure and loading,
solvent, addition rate), and the scope of the reaction.
Chart 1
Results
1. Uncatalyzed Reactions. 1.1. Cyclohexanone-Derived
Enolate. With the knowledge that appropriately electrophilic
silyl enolates can be reactive toward aldehydes in the absence
of external promoters5d,e,12,26 we attempted similar “unpromoted”
aldol additions with this new class of ketone-derived silyl
(25) HMPA and related phosphoramides are known to have the highest
donicities of common solvents, see: (a) Reichardt, C. SolVents and SolVent
Effects in Organic Chemistry, 2nd ed.; VCH: Weinheim, 1988; pp 17-
27. (b) Gritzner, G.; Ho¨rzenberger, F. J. Chem. Soc., Faraday Trans. 1995,
91, 3843. (c) Gritzner, G. Z. Phys. Chem. Neue Folge 1988, 158, 99. (d)
Gritzner, G. J. Mol. Liq. 1997, 487. (e) Sandstro¨m, M.; Persson, I.; Persson,
P. Acta Chem. Scand. 1990, 44, 653. (f) Maria, P.-C.; Gal, J.-F.; Franceschi,
J. d.; Fargin, E. J. Am. Chem. Soc. 1987, 109, 483. (g) Bassindale, A. R.;
Lau, J. C.-Y.; Taylor, P. G. J. Organomet. Chem. 1995, 490, 75. (h)
Bassindale, A. R.; Lau, J. C.-Y.; Taylor, P. G. J. Organomet. Chem. 1995,
499, 137.
(26) (a) Miura, K.; Sato, H.; Tamaki, K.; Ito, H.; Hosomi, A. Tetrahedron
Lett. 1998, 39, 2585. (b) Kiyooka, S.-i.; Shimizu, A.; Torii, S. Tetrahedron
Lett. 1998, 39, 5237. (c) In situ generated dimethyl(triflato)silyl enolates
are claimed to engage in uncatalyzed aldol reactions, see: Kobayashi, S.;
Nishio, K. J. Org. Chem. 1993, 58, 2647.
(23) Ando, A.; Miura, T.; Tatematsu, T.; Shioiri, T. Tetrahedron Lett.
1993, 34, 1507.
(24) Yamada, Y. M. A.; Yoshikawa, N.; Sasai, M.; Shibasaki, M. Angew.
Chem., Int. Ed. Engl. 1997, 36, 1871.

Aldol Additions by Lewis Base Catalysis
J. Am. Chem. Soc., Vol. 121, No. 21, 1999 4985
Table 2. Uncatalyzed Aldol Additions of in-Situ Generated 1a
enolates. The preliminary examination of reactivity employed
the cyclohexanone-derived trichlorosilyl enolate 1 (as a proto-
typical E-configured enolate) and benzaldehyde (a). These
reactants cleanly provided the aldol adduct 3a in high yield and
with high diastereoselectivity, favoring the syn isomer, Table
1.27 The transformation of such an E-configured enolate to a
syn aldol product suggested that a closed, boatlike transition
structures may be operative, as have previously been proposed
for the reaction of Lewis-acidic silicon enolates.5d,e,12
To explore the generality of this aldolization, a representative
selection of aldehyde acceptors (Chart 1) were combined with
1 under similar conditions, Table 1. All aldehyde classes
examined reacted cleanly, and the adducts could be isolated in
good to high yield. All of the reactions, save that with
cyclohexanecarboxaldehyde (k) (entry 7), were syn-diastereo-
selective, to varying degrees. Hindered, conjugated aldehydes
(entry 4) and aliphatic aldehydes (entries 6 and 7) provided
modest selectivity, while benzaldehyde, cinnamaldehyde, and
the propargyl aldehyde h provided excellent levels of diaste-
reoselectivity (entry 5). Sterically congested aromatic aldehydes
and aliphatic aldehydes tended to react more slowly than smaller,
conjugated aldehydes. In the case of cinnamaldehyde (e),
R-methyl cinnamaldehyde (f), and phenyl propargyl aldehyde
(h), exclusive 1,2-addition was observed; the corresponding 1,4-
addition product was not detected.
entry
aldehyde
product
syn/antib
yield,c %
1
2
3
4
a
e
f
4a
4e
4f
>50/1
38/1
16/1
66
59
64
52
d
4d
1.8/1
a (1) 1.5 equiv cyclohexanone/1.6 equiv Cl3SiOTf/1.1 equiv i-Pr2NEt/
0.5 M/20 min. (2) 1.0 equiv RCHO/0.5 M/16 h. b Determined by H
1
NMR analysis. c Chromatographically homogeneous material.
The preliminary observation that Et2O provided significantly
higher diastereoselectivity than the other solvents prompted a
brief optimization of the in situ aldolization in this medium.
Gratifyingly, when the reaction concentration was raised to 0.5
M and reaction time extended to 16 h, respectable yields of the
aldol adducts could be obtained, again with high diastereose-
lectivity, Table 2, entry 1. That the yields were still only
moderate may reflect destruction of the enolate under the
reaction conditions or the participation of competitive pathways.
1
Analysis of the reaction mixture by H NMR before addition
of aldehyde indicated that clean, essentially quantitative conver-
sion to 1 had occurred.
Table 1. Uncatalyzed Aldol Reactions of 1a
With a more useful protocol in hand, a representative subset
of aldehydes (Chart 1) was surveyed, Table 2. Unfortunately,
aliphatic (hydrocinnamaldehyde (j)) and hindered (trimethylac-
etaldehyde) aldehydes did not react under these conditions.
However, both unsaturated aldehydes e and f provided high
diastereoselectivity, while 1-naphthaldehyde (d) was nearly
unselective. The trend in Table 2, entries 3 and 4 should be
contrasted to that found in Table 1, entries 1 and 3.
1.3. Propiophenone-Derived Enolate. To determine the
effect that enolate geometry would have on the unpromoted aldol
process, the Z-configured trichlorosilyl enolate derived from
propiophenone ((Z)-2)6 was combined with a variety of alde-
hydes (Chart 1) under the standard conditions, Table 3. As was
previously observed with enolate 1, the reactions proceeded in
high yield, except with hindered, conjugated aldehydes, Table
4, entry 5. Not unexpectedly,5d reactions with (Z)-2 were weakly
anti-selective as was observed previously with other Z-config-
ured silyl enolates, again suggestive of a boatlike transition
structure. As before, only 1,2-addition was observed for R,â-
unsaturated aldehydes.
entry
aldehyde
time, h
product
syn/antib
yield, %
1
2
3
4
5
6
7
8
a
d
e
f
h
i
6
8
1
11
2
12
12
36
3a
3d
3e
3f
3h
3i
49/1
16/1
49/1
5.7/1
36/1
7.3/1
5.3/1
1/1
92c
90d
83c
86d
91c
78c
82d
92d
j
k
3j
3k
a 0.5 M/0 °C. b Determined by 1H NMR analysis. c Analytically pure
material. d Chromatographically homogeneous material.
1.2. In Situ Generated Enolates. To avoid the isolation and
handling of the sensitive trichlorosilyl enolates, we investigated
the possibility of performing these useful reactions by in situ
generation of 1. Following the analogy of boron-based2m
(R2BOTf/amine base) and tin(II)-based2l (Sn(OTf)2/amine base)
enolate generation systems, we examined the use of the
trichlorosilyl triflate28 (Cl3SiOTf)/amine base system for the
direct enolsilylation of cyclohexanone. Enolization (Cl3SiOTf/
i-Pr2NEt/0 °C) in CH2Cl2, followed by addition of benzaldehyde
led to a disappointing 13/1 syn/anti ratio in modest yield. We
surmised that the CH2Cl2-soluble ammonium salts were interfer-
ing in the reaction and therefore surveyed the effect of other
solvents in the overall reaction. Although the yields were all
low, the diastereoselectivities were dependent on solvent, with
the highest syn-selectivity being obtained in solvents in which
the ammonium salts had precipitated from solution (Et2O and
pentane).
Table 3. Uncatalyzed Aldol Reactions of (Z)-2a
entry
aldehyde
time, h
product
syn/antib
yield, %
1
2
3
4
5
6
7
a
b
d
e
f
g
h
10
10
16
10
12
16
11
4a
4b
4d
4e
4f
4g
4h
1/2.3
1/2.9
1/1.3
1/1.9
1/2.2
1/1.9
1/2.2
97c
93d
95c
95c
64d
89c
89d
(27) The presence of 4 Å molecular sieves in the reaction mixture has
no apparent effect on the selectivity of the process, although they may
prevent elimination of the trichlorosilyl aldolate.
a 0.5 M/0 °C. b Determined by 1H NMR analysis. c Chromatographi-
cally homogeneous material. d Analytically pure material.
(28) (a) Bassindale, A. R.; Stout, T. J. Organomet. Chem. 1984, 271,
C1. (b) Ru¨dinger, C.; Beruda, H.; Schmidbaur, H. Z. Naturforsch., B: Chem.
Sci. 1994, 49b, 1348.
2. Catalyzed Aldol Additions. 2.1. Cyclohexanone-Derived
Enolate. 2.1.1. Background Reaction. To assess the efficiency

4986 J. Am. Chem. Soc., Vol. 121, No. 21, 1999
Denmark et al.
Table 5. Chiral Phosphoramide Catalyzed Addition of 1 to
of chiral phosphoramides in promoting the addition of 1 to
benzaldehyde (a), a control experiment without phosphoramides
was required. Two data points were collected at 8 min and at 2
h for the reaction of 1 with a at -78 °C (0.1 M CH2Cl2); by 1H
NMR analysis of the crude reaction mixtures, only 2 and 19%
conversions, respectively, were observed (Table 4). With the
less volatile 4-tert-butylbenzaldehyde (c), after 8 min at -78
°C, 2% of the syn aldol product 3c was isolated along with
96% recovery of the starting aldehyde c. After 2 h at -78 °C,
the isolated yield of syn-3c and recovered c were 19 and 78%,
respectively. Although appreciable background reaction was
observed with a and c, the reaction between 1 and 1-naphthy-
laldehyde (d) gave negligible background; after 2 h at -78 °C,
96% of the aldehyde d was recovered.
Benzaldehydea
entry
catalyst
syn/antib
er,c synd
er,c antie
yield,f %
1
2
3
4
5
(S,S)-5
(R,R)-6
(R)-7
8
1/50
nd
27.6/1
1/3.00
3.55/1
1/1.22
1/1.15
94
91
87
94
96
1/2.0
3.2/1
1/1.1
1/3.1
1.00/1
1/3.00
1.08/1
1.06/1
9
a 10 mol % cat./2 h. b Determined by 1H NMR analysis. c Determined
by CSP HPLC analysis. d Ratio (2S,1′S)/(2R,1′R) isomer. e Ratio (2R,1′S)/
(2S,1′R) isomer. f Chromatographically homogeneous material.
Table 4. Background Control Experiments
moderate. Proline derived phosphoramides 8 and 9 were slightly
anti-selective, but the enantioselectivities were very low.
2.1.2.2. Configuration of Aldol Products. The absolute
configuration of anti-3a derived from the reaction using (S,S)-5
was established to be (2R,1′S) (as drawn) by single-crystal X-ray
analysis of the corresponding 4-bromobenzoate.30
recovery
aldehyde
time, min
conv., %a
yield 3, %b
aldehyde, %b
To establish the absolute configuration of the major enanti-
omer of syn-3a, a highly enantiomerically enriched sample of
syn-3a was prepared by epimerization of anti-3a. Thus, enan-
tioenriched (-)-anti-3a was epimerized in methanolic potassium
carbonate at 0 °C for 1.5 h, Scheme 5. Chromatographic
isolation provided (-)-syn-3a in 21% yield and 19.0/1 enan-
tiomeric ratio along with 61% of the anti-diastereomer. Extended
reaction times resulted in further dehydration and decomposition
of 3a. Acylation of the levorotatory syn-3a with 4-bromobenzoyl
chloride in the presence of excess of DMAP and triethylamine
gave the crystalline dextrorotatory 4-bromobenzoate 10 in 98%
yield, which was recrystallized to enantiomeric purity. Single
crystals suitable for X-ray analysis established that the dex-
trorotatory ester has (2S,1′S) configuration.30 Thus, the major
enantiomer in the syn manifold formed with catalyst (S,S)-5 is
(2S,1′S)-3a by correlation of elution order on CSP HPLC.
a
a
c
c
d
8
120
8
120
120
2
19
nd
nd
nd
ndc
nd
2
19
nd
nd
nd
96
78
96
a Determined by H NMR analysis. b Chromatographically homo-
1
geneous material. c Not determined.
Chart 2
Scheme 5
2.1.2. Optimization of Aldol Addition. 2.1.2.1. Promoter
Structure. Clearly, the primary factor in effecting fast enanti-
oselective aldol additions is the selection of an appropriate
catalyst. Thus, the initial optimization focused on a broad survey
of phosphoramide catalyst structure. A selection of phosphora-
mides derived from chiral, enantiopure amines is shown in Chart
2.29 We chose the reaction of 1 with a at -78 °C (0.1 M
CH2Cl2) as the test system. All of the phosphoramides 5-9
proved to be effective in catalyzing the addition (Table 5). In
the presence of 10 mol % of the phosphoramides, high yields
of the aldol addition products were obtained. The phosphoramide
(S,S)-5 gave the best results as gauged by both diastereomeric
and enantiomeric ratios of the products. Phosphoramide (R,R)-6
gave low diastereoselectivity and moderate enantioselectivity
of the anti diastereomer. The binaphthyldiamine-derived phos-
phoramide (R)-7 was the only syn-selective phosphoramide, but
the enantioselectivity in both the syn and anti manifolds was
2.1.2.3. Reaction Conditions. After performing the initial
survey of catalyst structure, we briefly examined the effect of
different solvents and on both the diastereo- and enantioselec-
tivity of the aldol additions catalyzed by (S,S)-5. This reaction,
though efficient and always highly selective, suffered from
variable diastereoselectivity. Although the dr of the product was
always >20/1 anti/syn, the exact ratio seemed to vary with the
batch of catalyst, enolate, and aldehyde. Some of the reasons
(30) All X-ray crystal structure data has been deposited in the Cambridge
Crystallographic Data Center as supplementary publication Nos. CCDC-
111716 ((+)-anti-10), CCDC-111826 ((+)-syn-10)), CCDC-111715 ((+)-
syn-4b)). Copies of the data can be obtained free of charge on application
to CCDC, 12 Union Road, Cambridge CB21EZ, UK (fax: (+44)1223-
336-033; e-mail: deposit@ccdc.cam.ac.uk).
(29) Denmark, S. E.; Su, X.; Nishigaichi, Y. M.; Coe, D. M.; Wong,
K.-T.; Winter, S. B. D.; Choi, J.-Y. J. Org. Chem. 1999, 64, 1958.

Aldol Additions by Lewis Base Catalysis
J. Am. Chem. Soc., Vol. 121, No. 21, 1999 4987
for this variability have since been identified and will be
presented in a mechanistic analysis of the reaction.10 Conse-
quently, in each of these optimization studies the benchmark
reaction (1, a, and 10 mol % (S,S)-5) was performed side-by-
side with the other experiments in that particular study. Thus,
in the following tables several examples of this specific reaction
will be shown, each with slightly different results. This slight
variability does not undermine the conclusions presented in each
study, nor is it currently relevant due to improvements in
reaction protocol (vide infra).
In the first series of experiments, benzaldehyde was added
quickly to a cold (-78 °C) solution of phosphoramide (S,S)-5
and 1 in various solvents, Table 6. The benchmark reaction
proceeded well, with high (35/1) anti/syn selectivity and high
er in the anti manifold. The use of the less polar solvents toluene
and pentane provided significantly diminished dr and er, entries
3 and 4. The relatively low yield of the reaction in pentane
most likely represents the limited solubility of the phosphora-
mide in this solvent at low temperature. When Et2O was used,
the diastereoselectivity of the process increased (entry 2), similar
to the response of the unpromoted reaction (although in the
opposite sense). However, the enantioselectivity of the anti
process decreased appreciably, relative to that obtained in
CH2Cl2.
2.1.3. Survey of Aldehyde. With a set of optimal conditions
selected for the reaction, a survey of aldehyde structures in the
reaction of 1 was undertaken, the results of which are sum-
marized in Table 8. The phosphoramide (S,S)-5 catalyzed all
of the reactions admirably, adducts being formed in uniformly
high yield. In all cases, save that with the sterically nonde-
manding propargylic aldehyde h, excellent anti-selectivity was
observed. In addition, the enantioselectivity in the anti manifold
was very high, e.g., the 1-naphthaldehyde-derived product being
formed in 65.7/1 er. Unfortunately, the use of enolizable
aldehydes did not afford the corresponding aldol adducts,
probably due to competing deprotonation by the basic enolate/
phosphoramide complex. Although in certain systems (specif-
ically acetate7a and methyl ketone-derived7d trichlorosilyl eno-
lates) the use of enolizable aldehydes is possible, in the present
system (and with enolate (Z)-2) this remains a limitation.
Table 8. Aldol Additions of 1 Catalyzed by 10 mol % (S,S)-5a
entry
aldehyde
product
syn/antib
er,c antid
yield,e %
1
2
3
4
5
a
d
e
f
3a
3d
3e
3f
1/61
<1/99
<1/99
<1/99
1/5.3
27.6/1
65.7/1
15.7/1f
24.0/1
10.1/1
95
94
94
98
90
Table 6. Solvent Effects in Catalyzed Aldol Additions of 1a
h
3h
a 10 mol % (S,S)-5/0.1 M/2 h. b Determined by H NMR analysis.
c Determined by CSP HPLC analysis. d Ratio (2S,1′S)/(2R,1′R) isomer.
e Analytically pure material. f Ratio (2R,1′R)/(2S,1′S) isomer due to
priority change.
1
entry
solvent
syn/antib
er,c synd
er,c antie
yield,f %
1
2
3
4
CH2Cl2
Et2O
toluene
pentane
1/35
1/>50
1/6.2
1/2
nd
nd
1.43/1
1.56/1
25.3/1
8.52/1
5.02/1
1.88/1
85
82
84
70
The absolute configurations of syn-3a and anti-3a have
already been unambiguously assigned (Section 2.1.2.2.). The
configurations of syn- and anti-3d-h were assigned by analogy.
2.1.4. Rate of Mixing. During the course of the optimization
of related additions with cyclopentanone- and cycloheptanone-
derived trichlorosilyl enolates, the rate of aldehyde addition to
the reaction mixture was found to have dramatic consequences
on diastereoselectivity.7e By dropwise addition of benzaldehyde
to the reaction mixture (1 and (S,S)-5) over 1 h such that the
reaction is essentially complete upon addition of the aldehyde,
the anti-selectivity not only increased, but became much more
reproducible as well, consistently providing anti/syn ratios of
>50/1. Notably, this reaction variable (like most others with
catalyst (S,S)-5) had little effect on the enantioselectivity of the
reaction. With this modified reaction protocol in hand, the effect
of catalyst loading was reinvestigated. The same trend was
apparent, with the anti-selectivity decreasing with decreasing
catalyst loading. Also, the reaction was considerably slower at
very low loading, providing only 53% yield at 0.5 mol %. This
is due to poor conversion, and, if desired, the reaction conditions
could probably be altered to provide a high yield in this case
as well. Again, the er of anti-3a formed was unchanged over a
20-fold change in catalyst concentration, e.g., even 0.5 mol %
of (S,S)-5 provided anti-3a in 20.7/1 er albeit with a modest
(5/1) anti/syn ratio. Indeed, with this new protocol, high (28/1)
dr can now be routinely obtained with 2 mol % of (S,S)-5, while
with the fast addition protocol, 10 mol % of (S,S)-5 was required
to obtain such selectivity.
a 0.1 M/2 h. b Determined by H NMR analysis. c Determined by
CSP HPLC analysis. d Ratio (2S,1′S)/(2R,1′R) isomer. e Ratio (2R,1′S)/
(2S,1′R) isomer. f Chromatographically homogeneous material.
1
Next, the influences of overall reaction concentration and
catalyst loading were investigated again with a protocol involv-
ing fast addition of aldehyde. The results are compiled in Table
7. Increasing the concentration of the reaction solution from
0.1 to 0.5 M had a rather strong effect on diastereoselectivity
(compare entries 1 and 4). Furthermore, on lowering the catalyst
loading from 10 to 2 mol %, the diastereoselectivity dropped
sharply, from 14/1 to 2.4/1 anti/syn. Importantly, the enantio-
meric ratio of the products did not change significantly, either
on changing the concentration or on lowering the loading of
the phosphoramide.
Table 7. Phosphoramide Loading Effect in Aldol Addition with 1
entry loading conc., M syn/antia er,b sync er,b antid yield,e %
1
2
3
4
10%
5%
2%
0.5
0.5
0.5
0.1
1/14
1/10
1/2.4
1/28
1.46/1
1.29/1
1.31/1
1.47/1
16.9/1
17.9/1
14.9/1
20.7/1
94
90
84
90
10%
With the advent of the slow addition/low catalyst loading
protocol, the effect of solvent on the addition was reexamined,
Table 9. The results of reactions in CH2Cl2, Et2O, and toluene
a Determined by 1H NMR analysis. b Determined by CSP HPLC
analysis. c Ratio (2S,1′S)/(2R,1′R) isomer. d Ratio (2R,1′S)/(2S,1′R)
isomer. e Chromatographically homogeneous material.

4988 J. Am. Chem. Soc., Vol. 121, No. 21, 1999
Denmark et al.
Table 10. Chiral Phosphoramide Catalyzed Addition of (Z)-2 to
were similar to those obtained previously (Table 6). However,
the use of propionitrile as solvent provided slightly better
diastereoselectivity (31/1 vs 28/1) and attenuated enantioselec-
tivity (14.4/1 vs 22.8/1) compared to the use of CH2Cl2. The
compatibility of such a Lewis basic solvent with a process
catalyzed by Lewis bases demonstrates the dramatic accelerating
effect the phosphoramide catalysts have on this reaction.
Benzaldehydea
entry
catalyst
syn/antic
er,d syne
er,d antif
yield,g %
Table 9. Solvent Effects in Aldol Additions of 1 Catalyzed by 2
mol % (S,S)-5a
1
(S,S)-5
(S,S)-5
(R,R)-6
(R)-7
11.5/1
18/1
1/2.7
1/1.5
32.3/1
39.0/1
1/1.38
1/2.23
1.22/1
1.23/1
1.08/1
1.13/1
77
95
72
61
2b
3
4
a 10 mol % catalyst/0.1 M/6 h. b 15 mol % (S,S)-5 used. c Determined
1
by H NMR analysis. d Determined by CSP HPLC analysis. e Ratio
(2S,3S)/(2R,3R)-isomer. f Configuration of major enantiomer not as-
signed. g Chromatographically homogeneous material.
entry
solvent
CH2Cl2
Et2O
toluene
CH3CH2CN
syn/antib er,d synd er,c antie yield,f %
1
2
3
4
1/28
1/49
1/5.2
1/31
1.03/1
1.46/1
1.06/1
1.35/1
22.8/1
7.00/1
4.56/1
14.4/1
91
59
51
85
Table 11. Aldol Additions of (Z)-2 Catalyzed by (S,S)-5a
a 1 h addition/final concentration 0.1 M. b Determined by 1H NMR
analysis of the crude reaction mixture. c Determined by CSP SFC
analysis. d Ratio (2S,1′S)/(2R,1′R) isomer. e Ratio (2R,1′S)/(2S,1′R)
isomer. f Chromatographically homogeneous material.
entry aldehyde product syn/antib er,c synd er,c antie yield,f %
1
2
3
4
5
6
a
b
d
e
g
h
4a
4b
4d
4e
4g
4h
18/1
12/1
3.0/1
9.4/1
7.0/1
1/3.5
39.0/1
49.0/1
11.5/1
24.0/1g
21.2/1g
3.76/1
1.23/1
nd
1.44/1
1.27/1
1.22/1
1.22/1
95
89
96
97
94
92
2.2. Propiophenone-Derived Enolates. To provide support
for our hypothesis that the Lewis base-catalyzed aldol additions
do indeed proceed via hypercoordinate siliconate transition
structures, we examined the reaction of a Z-configured enolate,
that derived from propiophenone, 2. The addition of (Z)-2 to
benzaldehyde is much slower than the corresponding reaction
of enolate 1 (Scheme 6). In the absence of an external promoter
only 1% of the product 4a was isolated after 6 h at -78 °C.
a 15 mol % (S,S)-5/0.1 M/6 h. b Determined by H NMR analysis.
c Determined by CSP HPLC analysis. d Ratio (2S,3S)/(2R,3R)-isomer.
e Configuration of major enantiomer not assigned. f Analytically pure
material. g Ratio (2S,3R)/(2R,3S) isomer due to priority change.
1
Scheme 6
established by single X -ray analysis.30 The absolute configura-
tions of the other syn products were assigned by analogy.
Discussion
1. Uncatalyzed Reactions. 1.1. Enolate Reactivity. Chlo-
rosilyl enolates are unique in their ability to undergo both direct
and catalyzed aldol additions under mild conditions. They are
distinguished from common alkylsilyl enol ethers in their ability
to react with aldehydes under mild thermal conditions.31 It is
important to realize, however, that this is not due to a higher
nucleophilicity, as might be presumed in the context of “enolate
chemistry”. The term enolate is used throughout this work
primarily to distinguish the remarkable and divergent reactivity
of these chlorosilyl enolic species form their alkylsilyl coun-
terparts. In reality they are better considered as enol ethers, rather
than true enolates which are typically highly charge-localized
species associated with a cation (metal, ammonium, or sulfonium
salt) and whose reactivity is usually governed by the nucleo-
philicity of the double bond. Rather, this reactivity is due to
the highly electrophilic nature of the silicon atom, and in fact,
the reagents themselves are non-nucleophilic (certainly less
nucleophilic than the corresponding trimethylsilyl enol ethers)
due to the inductive polarization of electron density from the
enol system to the electropositive silicon center. The relatively
high reactivity toward aldehydes is, therefore, most likely due
to the activation accorded both partners when the aldehyde is
coordinated to the Lewis acidic silicon center. The aldehyde is
activated by virtue of the increased electrophilicity of the
Fortunately, this chlorosilyl enolate, like 1, was susceptible
to Lewis base catalysis. Phosphoramides 5, 6, and 7 catalyzed
the addition of (Z)-2 to benzaldehyde to give good yields of
the aldol addition product 4a (Table 10). The stilbenediamine
derived phosphoramide (S,S)-5 again gave the best results; a
syn/anti ratio of 11.5/1 and the er in the syn manifold up to
32.3/1. Phosphoramides 6 and 7 were poorly selective catalysts
in both diastereo- and enantioselective senses. The lower yields
of the reactions may be a reflection of slower reaction with
enolate 1. Increasing the catalyst loading from 10 to 15 mol %
increased the reaction yield to 95% and also improved the
diastereoselectivity (as expected, cf. Table 11) but not the
enantioselectivity.
The generality of the catalyzed reaction of enolate (Z)-2 was
surveyed with various aldehydes, Table 11. In the presence of
15 mol % of (S,S)-5, (Z)-2 reacted with aromatic, olefinic and
acetylenic aldehydes at -78 °C to give high yields of the aldol
adducts 4. The diastereoselectivity of the reactions varied from
moderate to high depending on the structure of the aldehyde.
The syn isomer is generally preferred except for the acetylenic
aldehyde h. The enantioselectivity for the major syn diastere-
omer was very high except for aldehyde h. In contrast, the
enantioselectivity for the minor anti diastereomer was uniformly
low. The absolute configuration of syn-4b was unambiguously
(31) For reactions of alkylsilyl enol ethers under mild conditions, see:
(a) Lubineau, A.; Auge´, J.; Queneau, Y. Synthesis 1994, 741. (b) RajanBabu,
T. V. J. Org. Chem. 1984, 49, 2083.

Aldol Additions by Lewis Base Catalysis
J. Am. Chem. Soc., Vol. 121, No. 21, 1999 4989
Scheme 7
carbonyl function upon coordination to the Lewis acidic silicon
center. The enol species is concurrently activated by making
the silyl enolate more electron rich overall and therefore more
nucleophilic. This finds good analogy in the reaction of boron
enolates2m with aldehydes and in reactions of strained-ring alkyl
silyl enolates.5d
six-membered transition structures with aldehyde coordinated
to the metal center and so-called “open-chain” transition
structures often associated with Lewis acid-promoted aldol
additions.4 Open transition structures are characterized by
variable, though often high syn-selectivity, which is effectively
independent of starting enol geometry. Although the results with
1 (and the corresponding cyclopentanone- and cycloheptanone-
derived enolates)7e could be rationalized in terms of an open-
chain structure, the uniformly moderate-to-high syn-selectivity
with enolate 1 and the uniformly low anti-selectivity with enolate
2 are better explained by invoking a closed transition structure
in agreement with the dual activation proposal above. Given
the stereochemical course of the reactions with 1, (E f syn) a
boatlike structure is predicted to be favored in such a model
which could also explain the modest Z f anti selectivity
observed with enolate 2. Indeed, these stereochemical outcomes
are in complete agreement with previous work on silacyclobutyl-
derived ketene acetals, wherein intramolecular silyl group
transfer (necessitating a closed transition structure) was dem-
onstrated by a double label crossover study.5d In this case as
well, the E f syn correlation was excellent whereas the Z f
anti correlation was not.
Thus, although trichlorosilyl enolates are reactive toward
aldehydes, they are unreactive when combined with non-Lewis
basic electrophiles, such as alkylating agents.32 The reaction
potential of these reagents is only enabled upon coordination
with an aldehyde (or other suitable Lewis basic electrophile).
As such, dual activation of substrates is needed and the reaction
most likely takes place via a closed, six-membered transition
structure as has been proposed for numerous other enol- and
allylmetal species.4 Such organizational control which is com-
monly thought of in the chemistry of boron, titanium, and other
metal enolates, although not associated with the Lewis acid-
promoted reactions of silyl enol ethers,4,5d can lead to high,
geometry-divergent stereoselectivity in both a relative and (when
a chiral auxiliary is part of the assembly) an absolute sense,
Scheme 7.
Although the reactions bear striking mechanistic similarity
to reactions involving boron enolates, the present aldolizations
are slower. This is most likely due to the greater Lewis acidity
of the enol boronates relative to the trichlorosilyl enolates. In
addition, it is likely that aldehyde activation dominates the
reactivity of the species, as silacyclobutyl ketene acetals
(certainly more nucleophilic species) react with aldehydes more
slowly than the trichlorosilyl enolates of ketones and much more
slowly than with trichlorosilyl enolates derived from methyl
acetate. In a computational study of the silicon-directed aldol
addition of a trihydridosilyl enolate to acetaldehyde, Gung
concluded the nucleophilicity of the enol double bond is more
important in the bond-forming event than is aldehyde activa-
tion.33 However, such a hydridosilyl species would be both more
nucleophilic and less electrophilic than the trichlorosilyl enolates
employed herein. Hence, we favor an explanation wherein
aldehyde activation of primary importance in the reaction
pathway.
The high syn-selectivity arising from boatlike transition
structures with 1 (E-enolate) is reasonable since there are few
unfavorable steric interactions in these assemblies; however, why
then are Z-enolates relatively unselective in these processes. One
reason that can be identified is that when the configuration
around silicon is assumed to be trigonal bipyramidal (tbp), with
aldehyde binding in the apical position, there are severe steric
interactions between the enolic Z-substituent and a substituent
in the basal plane of the trigonal bipyramidal silicon complex,
Figure 1. In addition, the Z-substituent also eclipses the
aldehydic H, an apparently unimportant interaction with Z-H
enolates (1), though potentially relevant with increasing size of
the Z-substituent (i.e., methyl in (Z)-2). Without a clearer picture
of the exact conformations in the transition structure it is difficult
to rationalize the selectivity differences when different aldehydes
are used in the uncatalyzed reactions. In addition, the effect of
solvents is challenging to understand. Considering the proposed
closed nature of the assembly, less polar solvents would lead
to higher selectivity, due to “solvent-driven compression”34 of
the transition structure. This was found not to be the case,
however, and it is unclear what role solvation has on these
1.2. Stereochemical Consequences and Transition Struc-
ture. The uncatalyzed reactions of enolate 1 (E-configured) are
all highly syn-selective, while those of enolate 2 (Z-configured)
are slightly anti-selective. Two limiting global arrangements of
the reacting partners in the aldol addition are possible; closed,
(32) Winter, S. D. W.; Stavenger, R. A., unpublished results from these
laboratories.
(33) Gung, B. W.; Zhu, Z.; Fouch, R. J. Org. Chem. 1995, 60, 2860.
(34) (a) Dubois, J. E.; Dubois, M. J. Chem. Soc., Chem. Commun. 1968,
1567. (b) Evans, D. A.; Nelson, J. V.; Vogel, E.; Taber, T. R. J. Am. Chem.
Soc. 1981, 103, 3099.

4990 J. Am. Chem. Soc., Vol. 121, No. 21, 1999
Denmark et al.
reactions, especially given the relatively polar nature of the
hypervalent silicon intermediate.
Figure 1. Possible boatlike transition structures.
2. Catalyzed Reactions. 2.1. Phosphoramide Catalysis and
Global Transition Structure Picture. The concepts and
practicality of Lewis base catalysis have recently been re-
viewed;35 therefore, only salient points will be made here. The
most distinguishing factor in Lewis base (compared to Lewis
acid) catalysis of the aldol addition, is that substrate(s) may be
activated in either electrophilic, nucleophilic, or synergistic
manner in the former, whereas electrophilic activation is the
rule for the later. Although Lewis-base-catalyzed aldol additions
have been previously reported, in all cases they involve the use
of strong, anionic Lewis bases (primarily fluoride) which cleave
the enol/silicon bond to provide ammonium or tris(amino)-
sulfonium (“free”) enolates which then react via an open-chain
transition structure. The distinguishing point of neutral Lewis
base catalysis is the potential of reaction through an associative
(“closed”) transition structure with enol, aldehyde, and catalyst
simultaneously organized around a metal center, thereby provid-
ing good, stereodiVergent internal diastereoselection and control
of absolute configuration. This concept is closely related to the
fluoride-catalyzed additions of crotyltrifluorosilanes36 and the
formamide,37,38c phosphoramide,38a,b,d and N-oxide-catalyzed39
additions of crotyltrichlorosilanes, all of which are believed to
proceed through a closed transition structure organized around
a hexacoordinate silicon center.
A detailed study and analysis of the mechanism of the
phosphoramide-catalyzed aldol addition has been completed and
will be elaborated in detail in a forthcoming paper. As an aid
for further discussion, the current mechanistic hypothesis will
be presented without the supporting documentation to allow
analysis of the reaction features presented above. We currently
believe that the reaction of 1 and benzaldehyde catalyzed by
phosphoramide (S,S)-5 proceeds through a chairlike transition
structure, with aldehyde, enol, and two chiral phosphoramide
molecules organized around a cationic, hexacoordinate silicon
atom, Figure 2.40 This hypothesis best explains all of the results
at hand. In addition, there is a competing pathway, also involving
a cationic silicon species, but with only one phosphoramide
bound to the silicon, leading to a pentacoordinate, boatlike
transition structure. Full discussion of rate acceleration (cataly-
sis), effect of phosphoramide loading, and the effect of the rate
of aldehyde addition will be deferred to a subsequent paper,
with only practical, preparative aspects being considered here.
Figure 2. Unified mechanistic scheme.
2.2. Stereochemical Consequences. The reactions of 1
catalyzed by (S,S)-5 proceeded with varying levels of diaste-
reoselectivity, all favoring the anti product. Conversely, (Z)-2
provided the syn-configured adducts, albeit with generally lower
diastereoselectivity. This stereodivergence is stronger than that
associated with the uncatalyzed reaction and is not consistent
with reaction through an open-chain transition structure. Rather,
this stereochemical course strongly suggests that the reaction
proceeds through a closed, chairlike transition structure, as
presented above. The lower diastereoselectivity of the Z-enolate
can again be explained by consideration of nonbonded interac-
tions of the enolic Z-substituent with groups on the silicon,
although inspection of molecular models suggests that these
groups are more distant in a hexacoordinate silicon chair, thus
leading to a reasonable degree of selectivity. The reason for
the switch from boatlike to chairlike transition structures with
configuration at silicon (pentacoordinate or hexacoordinate) is
unclear at this time.
In simplistic terms the changes in diastereoselectivity with
both catalyst loading and slow aldehyde addition is due to the
2/1, phosphoramide/enolate ratio present in the transition
structure for the formation of anti-3a. The related cationic 1/1
complex, which would involve a pentacoordinate silicon species,
is then considered to provide syn-3a. Thus, higher phosphora-
mide loading favors 2/1 coordination, leading to better E f
anti correlation. Similarly, when aldehyde is added slowly to
enolate and phosphoramide, a high catalyst loading (relative to
aldehyde) protocol is mimicked. If the rate of the reaction is
fast relative to the aldehyde addition rate, one would expect
that the results obtained from slow addition of aldehyde would
mimic the results obtained from high catalyst loading experi-
ments, whatever that result (higher E f anti correlation) or its
origin (2/1, phosphoramide/enolate ratio in the transition
structure) may be.
The effect of solvent in the catalyzed reaction is more difficult
to explain, due to the multiple transition structures potentially
operable (i.e., pentacoordinate chairs and boats, hexacoordinate
chairs and boats). It seems that very nonpolar solvents (toluene
and pentane) lower the energy difference between competing
pathways, both between chairs and boats (leading to reduced
diastereoselection) and between the major, chairlike pathways
(leading to reduced enantioselection). Relatively polar solvents
(relative to CH2Cl2) lead to higher diastereoselectivity, with only
small changes in enantioselectivity. Neither of these trends can
be definitively explained at the present time, given our limited
knowledge of the conformation of the proposed cationic
transition structures and the generally poor understanding of
solvation effects on the stereochemical course of such reactions.
Other chiral phosphoramide structures were uniformly less
successful catalysts compared to the stilbenediamine-derived
(35) Stavenger, R. A.; Denmark, S. E. Angew. Chem., Int. Ed., manuscript
submitted.
(36) (a) Kira, M.; Kobayashi, M.; Sakurai, H. Tetrahedron Lett. 1987,
28, 4081. (b) Sato, K.; Kira, M.; Sakurai, H. J. Am. Chem. Soc. 1989, 111,
6429.
(37) Kobayashi, S.; Nishio, K. J. Org. Chem. 1994, 59, 6620.
(38) (a) Denmark, S. E.; Coe, D. M.; Pratt, N. E.; Greidel, B. G. J. Org.
Chem. 1994, 59, 6161. (b) Iseki, K.; Kuroki, Y.; Takahashi, M.; Kishimoto,
S.; Kobayashi, Y. Tetrahedron 1997, 53, 3513-3526. (c) Iseki, K.; Mizuno,
S.; Kuroki, Y.; Kobayashi, Y. Tetrahedron Lett. 1998, 39, 2767. (d) For a
related asymmetric allenylation, see: Iseki, K.; Kuroki, Y.; Kobayashi, Y.
Tetrahedron: Asymmetry 1998, 9, 2889.
(39) Nakajima, M.; Saito, M.; Shiro, M.; Hashimoto, S.-i. J. Am. Chem.
Soc. 1998, 120, 6419.
(40) Denmark, S. E.; Su, X.; Nishigaichi, Y. J. Am. Chem. Soc. 1998,
120, 12990.

Aldol Additions by Lewis Base Catalysis
J. Am. Chem. Soc., Vol. 121, No. 21, 1999 4991
Figure 3. Factors leading to observed stereoselectivity.
(S,S)-5. The reasons for this are unclear, although phosphora-
mides 7-9 have vastly different structures. The poor perfor-
mance of the cyclohexanediamine-derived phosphoramide 6 is
more surprising since the local structure (piperidinodiazaphos-
pholidine oxide) is similar to (S,S)-5, although the overall shape
is fairly different due to the dihedral angles between the C(4)
and C(5) substituents. Clearly, this change is enough to
significantly alter both the overall composition of the transition
structure (1 or 2 phosphoramides, leading to boat- or chairlike
transition structures) and the more subtle differences between
competing boat- or chairlike structures (leading to diminished
enantioselectivity). Differences between the aldol reaction and
the related allylation of aldehydes with allyltrichlorosilanes are
exemplified by the lack of reactivity of catalyst 7 in the later
reaction and the very successful use of 6, 8, and 9. By
comparison, these phosphoramides afforded only meager se-
lectivities in the aldol reactions described herein.
The absolute stereochemical course of the reactions with
(S,S)-5 does offer clues about the overall arrangement of the
components in the transition structure and the effects the
phosphoramide catalysts have on the reaction, Figure 3. Given
the absolute configurations of the major products, it can be
concluded that the face of the enolate is preserved, whether an
E- or Z-enolate is used. Also, the aldehyde face is preserved,
with the change in the stereochemical course of the reaction
being simply due to change in configuration of the starting enol
double bond. This suggests that, whatever the mode of dif-
ferentiation, in the hexacoordinate array, the Si-C(1)41 face of
the enolate is blocked, and the configuration of the products is
then determined by (a) enol geometry, (b) the intrinsic prefer-
ence of hexacoordinate silicon enols for a chairlike transition
structure, and (c) the configuration of the chiral catalyst.42 The
use of (S,S)-5 effectively blocks the Si-C(1) face of the enolate;
placement of the aldehyde in a chairlike transition structure then
correctly predicts both the predominant diastereomer formed
and the observed absolute configuration of the major diastere-
omer, regardless of the geometry/conformation around the
silicon atom and in the actual transition structure. Thus, cyclic
enolates (E-configured) typically provide excellent anti-selectiv-
ity in reactions catalyzed by (S,S)-5, while Z-enolates provide
moderate to high syn-selectivity in the analogous reactions. In
both cases (and in the case of methyl ketone- and methyl acetate-
derived trichlorosilyl enolates) the newly formed hydroxyl-
bearing center is of the S-configuration if the aldehyde
substituent has priority over the enol backbone.
Conclusions
The chemistry of a new class of aldol addition reagents,
enoxytrichlorosilanes (trichlorosilyl enolates), has been inves-
tigated. These novel species are reactivity-inverted enolates in
that their chemistry is dominated by the electrophilicity of the
silicon unit. Trichlorosilyl enolates of cyclohexanone and
propiophenone react spontaneously with a wide variety of
aldehydes to afford the aldol addition products via a trigonal
bipyramidal siliconate complex through predominantly boatlike
transition structures. Most importantly, the aldol addition of
trichlorosilyl enolates with aldehydes is highly susceptible to
catalysis by chiral phosphoramides which function as Lewis-
basic activators. In the presence of a stilbenediamine-derived
phosphoramide, the aldol additions are highly diastereo- and
enantioselective. The predominant products can be interpreted
as arising from chairlike assemblies of phosphoramide, aldehyde,
and enolate around a hexacoordinate siliconate complex. Both
E- and Z-configured enolates react with high diastereoselectivity
and enantioselectivity to the respective anti and syn aldol
products with nonenolizable aldehydes. This represents a unique
illustration of the dual activation afforded by Lewis base
complexation of electrophilic species and augurs well for
broader applications in the vast domain of organosilicon
chemistry.
Experimental Section
See Supporting Information.
Acknowledgment. We are grateful to the National Science
Foundation for generous financial support (CHE 9500397 and
9803124). R.A.S. thanks the Eastman Chemical Co. for a
graduate fellowship. X.S. thanks the University of Illinois for
graduate fellowships.
Supporting Information Available: Full experimental
details, general procedures for aldol addition reactions and full
characterization data for all aldol products and derivatives
described are provided (PDF). This material is available free
of charge via the Internet at http://pubs.acs.org.
(41) This notation is used to describe the faces of the enol double bond
relative to the C(1) (oxygen-bearing) position, rather than the more typical
C(2) position which is actually transformed into the resultant tetrahedral
center.
(42) For a discussion of the solid stae structures of [(S,S)-5]2‚SnCl4 and
related complexes, see: Denmark, S. E.; Su, X. Tetrahedron 1999, in press.
JA984123J