The chemistry of trichlorosilyl enolates. 6. Mechanistic duality in the lewis base-catalyzed aldol addition reaction [18]

Full text of DOI:10.1021/ja982993v

12990
J. Am. Chem. Soc. 1998, 120, 12990-12991
Table 1. Diastereoselectivity Dependence on Bulk of the
N-Substituent in 5a
The Chemistry of Trichlorosilyl Enolates. 6.
Mechanistic Duality in the Lewis Base-Catalyzed
Aldol Addition Reaction
phosphoramide
time, h
syn/antib
yield,c %
Scott E. Denmark,* Xiping Su, and Yutaka Nishigaichi
5a
5b
5c
5d
1.5
6.0
1.5
1.5
1/2.8
27/1
31/1
99
93
96
95
Roger Adams Laboratory, Department of Chemistry
UniVersity of Illinois, Urbana, Illinois 61801
40/1
ReceiVed August 20, 1998
1
a Reaction with 10 mol % 5 at -75 °C. b Determined by H NMR
analysis. c Isolated, purified product.
Recent disclosures from these laboratories have demonstrated
the concept of Lewis base catalysis in asymmetric aldol additions1
as illustrated in the reaction of trichlorosilyl enolates and
aldehydes in the presence of a catalytic amount of chiral
phosphoramides.2 For example, trichlorosilyl enolate 1 reacts with
benzaldehyde in very high enantio- and diastereoselectivity with
10 mol % of phosphoramide (S,S)-4, Scheme 1. On the basis of
the geometry-dependent diastereoselectivity observed, these reac-
tions are proposed to proceed via closed, chairlike transition
structures organized about a hexacoordinate siliconate.2b We now
report evidence in support of a mechanistic hypothesis that a
hexacoordinate, cationic silicon species with two chiral phos-
phoramide molecules is on the pathway from E-enolates to anti
aldol products.
Scheme 1
Figure 1. Loading dependence of selectivity.
Phosphoramide (S,S)-4 dramatically accelerated the reaction
in Scheme 1. In the absence of the catalyst the reaction between
1 and 2 was very slow; after 8 min at -78 °C only 2.3%
conversion to the aldol product was detected by NMR (2%
isolated). On the other hand, with 10 mol % of (S,S)-4 the reaction
proceeded to 100% conversion in 8 min at -78 °C (95% isolated).
To evaluate the effects of catalyst structure on rate and
selectivity, we surveyed the series of achiral phosphoramides
5a-d (Chart 1) with regard to their ability to promote the aldol
process in Scheme 1. The results, Table 1, showed that increasing
the bulk of the substituent had a dramatic effect on the diaste-
reoselectivity of the reaction.
Figure 2. Ee dependence of catalysts 4 and 6.
decreased dramatically with increased loading of 5c, Figure 1.
The influence of the catalyst loading on selectivity is difficult to
understand unless the reaction is viewed as proceeding by
competitive pathways to the two diastereomers which respond
differently to catalyst concentration. Since the anti diastereomer
is prevalent with the less hindered catalysts (4 and 5a) as well as
with 5c at higher loadings, we postulated that this isomer arises
from a pathway that involves two phosphoramide molecules. This
pathway apparently proceeds through a chairlike transition
structure on the basis of the E f anti correlation. If the syn
diastereomer arises from an independent pathway that involves
only one phosphoramide (consistent with its preferential formation
from hindered catalysts and at low catalyst concentration), then
the diastereomeric ratio could be loading dependent.3
To substantiate this proposition, we made use of nonlinear
effects to identify a higher order molecularity of the catalyst in
the pathway to the anti diastereomer.4 The results with 10 mol %
of 4, shown in Figure 2 (b), clearly demonstrated a positive
nonlinear effect and thus support the hypothesis of a transition
structure with more than one phosphoramide molecule. These data
fit very well with Kagan’s two-ligand model,4b and assuming a
Chart 1
This intriguing switch of diastereoselectivity provided an
important mechanistic insight when we discovered that it was
also highly dependent on the catalyst loading. The syn/anti
selectivity of the reaction between 1 and 2 catalyzed by 5c
(1) For recent reviews see: (a) Nelson, S. G. Tetrahedron: Asymmetry 1998,
9, 357 and referenced cited therein. (b) Gro¨ger, H.; Vogl, E. M.; Shibasaki,
M. Chem. Eur. J. 1998, 4, 1137. For recent advances, see: (c) Kru¨ger, J.;
Carreira, E. M. J. Am. Chem. Soc. 1998, 120, 837. (d) Yanagisawa, A.;
Matsumoto, Y.; Nakashima, H.; Asakawa, K.; Yamamoto, H. J. Am. Chem.
Soc. 1997, 119, 9319.
(2) (a) Denmark, S. E.; Wong, K.-T.; Stavenger, R. A. J. Am. Chem. Soc.
1997, 119, 2333. (b) Denmark, S. E.; Stavenger, R. A.; Wong, K.-T. J. Org.
Chem. 1998, 63, 918. (c) Denmark, S. E.; Stavenger, R. A.; Wong, K.-T.
Tetrahedron 1998, 54, 10389.
(3) Given these divergent pathways and assuming one pathway is highly
syn diastereoselective, we expect a linear relationship of syn/anti ratio with
[6]-1 as seen in Figure 1, see Supporting Information for details.
10.1021/ja982993v CCC: $15.00 © 1998 American Chemical Society
Published on Web 12/01/1998

Communications to the Editor
J. Am. Chem. Soc., Vol. 120, No. 49, 1998 12991
Table 2. Conversion Dependence of Enantiomeric Puritya
2,
equiv
4,b
equiv
time,
s
conversion,
%
yield,c
%
anti-3,
entry
ee,d %
1
2
3
4
5
1.0
1.0
1.0
0.1
1.0
0.1
0.1
0.1
0.1
1.0
480
30
100
63
95
61
52
96
99
53.3
53.7
53.2
53.2
53.9
10
55
480
480
100
100
a Reaction of 1 and 2, Scheme 1. b 40% ee catalyst used. c Isolated,
purified product. d Determined by chiral stationary phase, supercritical
fluid (CO2) chromatographic analysis.
Figure 3. Penta- and hexacoordinate cationic silicon assemblies.
incorporate two phosphoramide molecules in the transition
structure leading from 1 to the anti diastereomers. Both of the
deficiencies with the earlier model can be accommodated if we
posit the intermediacy of cationic silicon species, Figure 3.6
Phosphoramide-promoted ionization of chloride to produce such
a cationic silicon intermediate7 would both explain the activation
and allow for singly- (pentacoordinate, TBP) and doubly-
(hexacoordinate, octahedral) complexed species.
Thus, to probe the possibility of chloride ionization, we
investigated the influence of various salts on the rate of the
promoted aldol reaction. Catalyst 6 gave 44% conversion of 1
and 2 to the product syn-3 with 53% ee after 8 min at -78 °C.
In the presence of 1.2 equiv of Bu4N+Cl-, the aldol addition was
inhibited (8% conversion (26% ee)), while in the presence of an
equivalent amount of Bu4N+OTf- the reaction was accelerated
(92% conversion (55% ee)). These results are fully consistent
with the hypothesis of prior ionization of the enolate 1 by the
action of phosphoramides. The resulting cationic silicon species
should have much higher Lewis acidity than 1 itself to activate
the aldehyde for nucleophilic attack. Accordingly, Bu4N+Cl-
inhibited the ionization of 1 through the common ion effect,8
whereas the triflate facilitated the ionization by increasing the
ionic strength of the medium.9
In summary, we have demonstrated that the origin of the rate
acceleration in the aldol addition reactions between trichlorosilyl
enolates and aldehydes stems from the ionization of the enolate
promoted by the phosphoramides. Sterically demanding phos-
phoramides bind to the enolate in a 1/1 fashion and the resulting
pentacoordinate cationic siliconate favors a boatlike arrangement.
Sterically less demanding phosphoramides can bind in a 2/1 fash-
ion, and the resulting hexacoordinate cationic siliconate favors a
chairlike arrangement. Further development of highly enantiose-
lective catalysts and studies of the reaction kinetics are underway.
statistical distribution of the enantiomeric ligands, the calculated
reaction rate of a homochiral catalyst assembly is about 3.6 times
that of a meso catalyst assembly.
For this phenomenon to be responsible for the loading-
dependent diastereoselectivity, the pathway to the syn diastere-
omer must have a different molecularity in catalyst. If the syn-
selective pathway (for the E-enolate) proceeds through an
intermediate with only one phosphoramide binding to silicon, then
a linear relationship should exist between the ee of the syn aldol
product and the ee of the catalyst. For this purpose we devised
the hindered, phosphoramide 6 (Chart 1), and gratifyingly found
that it was indeed a syn-diastereoselective catalyst, Scheme 1,
albeit with still modest enantioselectivity.5 Nevertheless, we could
still make use of this compound for mechanistic purposes, and
as clearly illustrated in Figure 2 (9), a linear relationship was in
fact observed with an excellent correlation coefficient.
It is necessary to consider other factors that can give rise to a
nonlinear dependence, such as the reservoir effect4e and interaction
of the catalytic species with the forming product. We feel it is
unlikely that phosphoramides are tightly aggregated in the ground
state, but even if this were so, such an effect cannot explain the
change of diastereoselectivity on the catalyst loading. In addition,
if a reservoir effect were responsible for the observed nonlinear
effect, then its magnitude should be catalyst concentration
dependent. Comparison of entries 1 and 5 in Table 2 clearly shows
that with 4 of 40% ee, the ee of anti-3 is independent of catalyst
concentration over a 10-fold range.
To rule out the possibility that the reaction product may be
involved in the stereochemical determining step or in the nonlinear
effect, we have examined the enantiomeric composition of the
product as a function of conversion, Table 2. With the catalyst 4
of 40% ee, the conversion of the reaction after 10 s was 55%
and the ee was essentially the same as at 100% conversion (entries
1 and 3). The reaction with 10 mol % 2 was employed to mimic
the reaction at 10 mol % conversion and gave product anti-3 with
similar enantiomeric excess (entry 4). With 1 equiv of 4, the
possibility that one of the enantiomers of the phosphoramide may
be selectively bound to the product to give rise to the nonlinear
effects should be minimized and again essentially the same
enantioselectivity was obtained. Taken together, these control
experiments rule out other interpretations and support the
hypothesis that two phosphoramide molecules in the transition
structure are responsible for the nonlinear effects observed.
In our original formulation for the mechanism of this reaction
we postulated pentacoordinate and hexacoordinate silicon inter-
mediates for the unpromoted and promoted reactions, respec-
tively.2 This view must be modified for several reasons: (1) the
simple change in coordination number or geometry at silicon
cannot easily account for the dramatic rate acceleration the
phosphoramide provides and (2) there must be a way to
Acknowledgment. We are grateful to the National Science Foundation
for generous financial support (CHE 9500397 and 9803124). Y.N. thanks
the Ministry of Science and Education of Japan for a postdoctoral
fellowship. We thank Mr. R. A. Stavenger for stimulating discussions.
Supporting Information Available: Procedures for the preparation
and full characterization of 5a-d and 6 and representative procedures
for aldol addition reactions and nonlinear effect studies and kinetic analysis
of nonlinear effects and loading effects (17 pages, print/PDF). See any
current masthead page for ordering information and Web access
instructions.
JA982993V
(6) Cationic silicon species have been proposed as intermediates in other
Lewis base-promoted reactions, see: (a) Denmark, S. E.; Barsanti, P. A.;
Wong, K.-T.; Stavenger, R. A. J. Org. Chem. 1998, 63, 2428. (b) Chojnowski,
J.; Cypryk, M.; Michalski, J.; Wozniak, L. J. Organomet. Chem. 1985, 288,
275. (c) Corriu, R. J. P.; Dabosi, G.; Martineau, M. J. Organomet. Chem.
1980, 186, 25. (d) Bassindale, A. R.; Lau, J. C.-Y.; Taylor, P. G. J. Organomet.
Chem. 1995, 490, 75. (e) Bassindale, A. R.; Lau, J. C.-Y.; Taylor, P. G. J.
Organomet. Chem. 1995, 499, 137.
(7) Berrisford and co-workers have recently postulated that Lewis bases
promote the ionization of allyltrichlorosilanes. Short, J. D.; Attenoux, S.;
Berrisford, D. J. Tetrahedron Lett. 1997, 38, 2351. See also: Nakajima, M.;
Saito, M.; Shiro, M.; Hashimoto, S.-i. J. Am. Chem. Soc. 1998, 120, 6419.
(8) (a) Bateman, L. C.; Hughes, E. D.; Ingold, C. K. J. Chem. Soc. 1940,
974, 1017. (b) Streitwieser, A. SolVolytic Displacement Reactions; McGraw-
Hill: New York, 1962. (c) Thornton, E. SolVolysis Mechanisms; Ronald
Press: New York, 1964.
(4) Nonlinear effect: (a) Puchot, C.; Samuel, O.; Dun˜ach, E.; Zhao, S.;
Agami, C.; Kagan, H. B. J. Am. Chem. Soc. 1986, 108, 2353. (b) Guillaneux,
D.; Zhao, S.-H.; Samuel, O.; Rainford, D.; Kagan, H. B. J. Am. Chem. Soc.
1994, 116, 9430. (c) Avalos, M.; Babiano, R.; Cintas, P.; Jime´nez, J. L.;
Palacios, J. C. Tetrahedron: Asymmetry 1997, 8, 2997. (d) Kagan, H. B.;
Fenwick, D. Top. Stereochem. 1999, in press. Reservoir effect: (e) Kitamura,
M.; Suga, S.; Oka, H.; Noyori, R. J. Am. Chem. Soc. 1998, 120, 9800.
(5) The absolute configuraion of (-)-syn-3a was determined to be (2S,1′S)
by X-ray crystal structure analysis of the 4-bromobenzoate ester.
(9) This result is in contrast to the observations made by Berrisford et al.
in the allylation reactions.7a We cannot rule out the inhibition of reaction by
chloride ion serving as a competitive ligand for the trichlorosilyl enolate.