Chiral phosphoramide-catalyzed aldol additions of ketone trichlorosilyl enolates. Mechanistic aspects

Article abstract of DOI:10.1021/jo060243v

The mechanism of the catalytic, enantioselective addition of trichlorosilyl enolates to aldehydes has been investigated. Kinetic studies using ReactIR and rapid injection NMR (RINMR) spectroscopy have confirmed the simultaneous operation of dual mechanist

Full text of DOI:10.1021/jo060243v

Chiral Phosphoramide-Catalyzed Aldol Additions of Ketone
Trichlorosilyl Enolates. Mechanistic Aspects
Scott E. Denmark,* Son M. Pham, Robert A. Stavenger, Xiping Su, Ken-Tsung Wong, and
Yutaka Nishigaichi
Department of Chemistry, Roger Adams Laboratory, UniVersity of Illinois, Urbana, Illinois 61801
denmark@scs.uiuc.edu
ReceiVed February 6, 2006
The mechanism of the catalytic, enantioselective addition of trichlorosilyl enolates to aldehydes has been
investigated. Kinetic studies using ReactIR and rapid injection NMR (RINMR) spectroscopy have
confirmed the simultaneous operation of dual mechanistic pathways involving either one or two
phosphoramides bound to a siliconium ion organizational center. This mechanistic dichotomy was initially
postulated on the basis of catalyst loading studies and nonlinear effects studies. This duality explains the
difference in reactivity and stereoselectivity of various classes of phosphoramides. Determination of
Arrhenius activation parameters revealed that aldol addition occurs through the reversible albeit unfavorable
formation of an activated complex, and natural-abundance 13C NMR kinetic isotope effect (KIE) studies
have determined that the turnover limiting step is the aldol addition. A thorough examination of a range
of phosphoramides has established empirical structure-activity selectivity relationships. In addition, the
effects of catalyst loading, rate of addition, solvents, and additives have been studied and together allow
the formulation of a unified mechanistic picture for the aldol addition.
Introduction and Background
to a wide range of aryl and unsaturated aldehydes.5
A
particularly important feature among substituted enolates in this
reaction is the configurational correlation between enolate
geometry and reaction diastereoselectivity. Using geometrically
defined trichlorosilyl enolates of either E- or Z-configuration,
both syn and anti products may be obtained through the selection
of an appropriate phosphoramide.6 More recently, we have
expanded this field to encompass trichlorosilyl enolates bearing
either R- and â-oxygen substituents with R- and â-stereogenic
centers with excellent results.7 In all cases, high double
diastereoselection is achieved with no deleterious influence from
the resident heteroatom. Additional progress has been made in
In recent years, our laboratories have been actively engaged
in the development of Lewis base catalyzed stereoselective
reactions of organotrichlorosilanes.1 The reactions promoted by
these reagents are highly responsive to the action of chiral Lewis
bases, enhancing both reactivity and stereoselectivity for a broad
range of useful transformations including aldol additions,2
allylations,3 and epoxide openings.4
Primary among these has been the development of enantio-
selective aldol additions catalyzed by chiral phosphoramides.
These endeavors have provided new avenues for the stereo-
selective addition of methyl and cyclic trichlorosilyl enolates
(5) (a) Denmark, S. E.; Stavenger, R. A.; Wong, K.-T.; Su, X. J. Am.
Chem. Soc. 1999, 121, 4982-4991. (b) Denmark, S. E.; Stavenger, R. A.
J. Am. Chem. Soc. 2000, 122, 8837-8847. (c) Denmark, S. E.; Stavenger,
R. A.; Wong, K.-T. Tetrahedron 1998, 54, 10389-10402.
(1) Denmark, S. E.; Fujimori, S. In Modern Aldol Reactions; Mahrwald,
R., Ed.; Wiley-VCH: Weinheim, 2004; Vol. 2, Chapter 7.
(2) Denmark, S. E.; Stavenger, R. A. Acc. Chem. Res. 2000, 33, 432-
440.
(6) (a) Denmark, S. E.; Su, X.; Nishigaichi, Y. J. Am. Chem. Soc. 1998,
120, 12990-12991. (b) Denmark, S. E.; Pham, S. M. J. Org. Chem. 2003,
68, 5045-5055.
(3) (a) Denmark, S. E.; Fu, J.; Coe, D. M.; Su, X.; Pratt N. E.; Griedel,
B. D. J. Org. Chem. 2006, 71, 1513-1522. (b) Denmark, S. E.; Fu, J.;
Lawler, M. J. J. Org. Chem. 2006, 71, 1523-1536. (c) Denmark, S. E.;
Fu, J. Chem. Commun. 2003, 167-170. (d) Denmark, S. E.; Fu, J. J. Am.
Chem. Soc. 2000, 122, 12021-12022. (e) Denmark, S. E.; Fu, J. J. Am.
Chem. Soc. 2001, 123, 9488-9489.
(7) (a) Denmark, S. E.; Fujimori, S.; Pham, S. M. J. Org. Chem. 2005,
70, 10823-10840. (b) Denmark, S. E.; Fujimori, S. Org. Lett. 2002, 4,
3477-3480. (c) Denmark, S. E.; Fujimori, S. Org. Lett. 2002, 4, 3473-
3476. (d) Denmark, S. E.; Fujimori, S. Synlett 2001, 1024-1029. (e)
Denmark, S. E.; Pham, S. M. Org. Lett. 2001, 3, 2201-2204. (f) Denmark,
S. E.; Stavenger, R. A. J. Org. Chem. 1998, 63, 9524-9527.
(4) Denmark, S. E.; Barsanti, P. A.; Wong, K.-T.; Stavenger, R. A. J.
Org. Chem. 1998, 63, 2428-2429.
10.1021/jo060243v CCC: $33.50 © 2006 American Chemical Society
Published on Web 04/08/2006
3904
J. Org. Chem. 2006, 71, 3904-3922

Aldol Additions of Ketone Trichlorosilyl Enolates
the field of crossed aldehyde-aldehyde aldol additions8 as well
as acetate-ketone aldol additions.9
The conceptual framework for the design and implementation
of the novel class of reactions has been described previously in
a full account.1 From its inception, the Lewis base catalysis
concept was mechanistically unique among asymmetric trans-
formations, and indeed, a number of papers disclosing critical
kinetic10 spectroscopic11 and mechanistic6a insights have ap-
peared. The objective of this full account is to present an
integrated analysis of all the foregoing studies together with
new structure/selectivity studies, medium effects and additional
mechanistic studies and therefrom a unified mechanistic picture
of the Lewis-base-catalyzed aldol addition reaction of trichlo-
rosilyl enolates.
FIGURE 1. Plot of ln[aldehyde] versus time for the uncatalyzed
addition of 1 to 2. The graph depicts a linear fit, f(x) ) mx + b (m )
0.025, R2 ) 0.999), for a first-order reaction.
Results
Determination of Arrhenius activation parameters was com-
pleted in the usual manner by measuring the rate constant of
the uncatalyzed reaction as a function of temperature between
25 and 60 °C.13 Interestingly, analysis of the activation
parameters (calculated for T ) 303 K) reveals an unexpectedly
large entropic contribution (∆Sq ) -56.7 ( <0.1 cal mol-1
K-1) to the free energy of activation (∆Gq) 23.0 ( <0.1 kcal
mol-1). Although a large entropic contribution suggests a highly
ordered transition structure, its magnitude does not preclude an
unfavorable preequilibrium binding between the trichlorosilyl
enolate and aldehyde prior to aldolization. The relatively small
value for the enthalpy of activation (∆Hq ) 5.8 ( <0.1 kcal
mol-1) illustrates the extreme reactivity of trichlorosilyl enolates.
Although kinetic analysis by aliquot sampling is reliable for
reactions run at or above room temperature, it is generally
unsuitable for fast reactions performed at subambient temper-
atures. With this in mind, kinetic analysis of catalyzed reactions
was determined by in situ monitoring using IR spectroscopy.
The kinetic analysis described above was confirmed using a
Mettler-Toledo ReactIR14 for the acetone enolate/pivalaldehyde
system without adamantane. Gratifyingly, the kinetic results
from IR analysis were in excellent agreement with those from
GC analysis,10a and we could next turn our attention to the aldol
system of the cyclohexanone-derived trichlorosilyl enolate 412
and benzaldehyde (5a), Scheme 2. Although kinetic analysis
from the rates of formation of diastereomers of i remained a
concern, the spectroscopic coincidence of the syn and anti
diastereomers removed this complication.
1. Unpromoted Reactions. 1.1. Kinetic Studies. Initial
kinetics studies were conducted with aliphatic aldehyde partners
because of retro-aldolization of aromatic aldol adducts during
gas chromatographic analysis. To further simplify the kinetic
analysis, the use of an unsubstituted trichlorosilyl enolate elimi-
nated the need to analyze mixtures of diastereomers. Following
a brief survey, the combination of acetone derived trichlorosilyl
enolate 112 and pivalaldehyde (2) was selected for the initial
kinetic runs, Scheme 1.10a To accurately measure substrate
concentrations during the course of the aldol addition, it was
necessary to employ a nonvolatile, unreactive internal standard
for which precise response factors could be determined. Ada-
mantane fit the desired criteria and was used as the internal
standard (Int. Std.) in all subsequent GC kinetic runs. An addi-
tional albeit minor deviation from the standard reaction condi-
tions involved using 1,2-dichloroethane (DCE) as a reaction
medium to allow for a larger temperature range which would
be necessary for the determination of Arrhenius parameters.
SCHEME 1
The overall reaction order could be determined for the
unpromoted reaction at 30 °C, by employing pseudo-order
conditions. The reaction was shown to be first order in enolate,
using 10 equiv of 2, by virtue of a best fit analysis of first and
second-order functions, Figure 1. Likewise, the reaction was
determined to be first order in aldehyde using 10 equiv of 1.
Thus, the uncatalyzed aldol addition was determined to be
second-order overall and first order in each component.
SCHEME 2
(8) (a) Denmark, S. E.; Ghosh, S. K. Angew. Chem., Int. Ed. 2001, 40,
4759-4762. (b) Denmark, S. E.; Bui, T. Proc. Nat. Acad. Sci. U.S.A. 2004,
101, 5439-5444. (c) Denmark, S. E.; Bui, T. J. Org. Chem., 2005, 70,
10393-10399.
(9) (a) Denmark, S. E.; Winter, S. B. D.; Su, X.; Wong, K.-T. J. Am.
Chem. Soc. 1996, 118, 7404-7405. (b) Denmark, S. E.; Fan, Y. J. Am.
Chem. Soc. 2002, 124, 4233-4235. (c) Denmark, S. E.; Fan, Y.; Eastgate,
M. D. J. Org. Chem. 2005, 70, 5235-5248.
(10) (a) Denmark, S. E.; Pham, S. M. HelV. Chim. Acta 2000, 83, 1846-
1853. (b) Pham, S. M. Ph.D. Thesis, University of Illinois at Urbana-
Champaign, 2002.
The Arrhenius parameters were again determined for the
uncatalyzed addition of cyclohexanone trichlorosilyl enolate 4
to benzaldehyde using in situ IR monitoring (calculated for T
) 303 K). Again, the activation parameters reveal a large entro-
pic contribution (∆Sq ) -58.2 ( 1.6 cal mol-1 K-1) and a small
enthalpic contribution (∆Hq ) 2.4 ( 0.5 kcal mol-1) to the free
(13) All kinetic runs were performed in triplicate to ensure accuracy and
reproducibility. See the Supporting Information for complete details.
(14) ReactIR 1000 fitted with a 5/8 in. DiComp Probe, running software
version 2.1a. Mettler-Toledo AutoChem Inc., 7075 Samuel Morse Dr.,
Columbia, MD, 21046, or visit http://us.mt.com.
(11) (a) Denmark, S. E.; Su, X. Tetrahedron 1999, 55, 8727-8738. (b)
Denmark, S. E.; Fu, J. J. Am. Chem. Soc. 2003, 125, 2208-2216.
(12) Denmark, S. E.; Stavenger, R. A.; Winter, S. B. D.; Wong, K.-T.;
Barsanti, P. A. J. Org. Chem. 1998, 63, 9517-9523.
J. Org. Chem, Vol. 71, No. 10, 2006 3905

Denmark et al.
CHART 1
energy of activation (∆Gq ) 20.0 ( 1.0 kcal mol-1). Compared
with the acetone-pivalaldehyde example, ∆Gq is smaller for
addition of cyclohexanone enolate to benzaldehyde and in
agreement with our preparative examples which demonstrate
that additions to hindered aliphatic aldehydes such as pivalal-
dehyde are slower with respect to additions to benzaldehyde.
1.2. Background Reactions at -78 °C. To assess the
efficiency of phosphoramides in promoting the aldol reactions,
a control reaction without promoters needed to be performed.
Although trichlorosilyl enolate 4 did not react with 1-naphthal-
dehyde at -78 °C (96% recovery of 1-naphthaldehyde was
obtained after 2 h), the reaction between 4 and 5a or 5b resulted
appreciable amount of products syn-6a and syn-6b. For example,
after 2 h at -78 °C, both 5a and 5b gave 19% of the isolated
syn products 6a and 6b, and the aldehyde 5b was recovered in
78% yield. When the reactions were quenched after 8 min at
-78 °C, only 2% conversion had occurred for both 5a and 5b
as judged by 1H NMR. On the other hand, the promoted
reactions were very fast. Using 10 mol % of chiral phosphor-
amide 7a, Chart 1, the reaction was complete within 8 min at
-78 °C, Scheme 3. Even quenching the reaction after 10 s
resulted in a greater than 50% conversion.
2. Promoted Reactions. 2.1. Survey of Chiral Phosphor-
amides. 2.1.1. Reaction with Cyclohexanone-Derived Enolate
4. Initial studies with cyclohexanone-derived enolate 4 showed
that N,N′-dimethylstilbene-1,2-diamine derived phosphoramide
7a is the best catalyst for the aldol reactions.5 To understand
the structural features that are important for rate as well as
selectivity and to further optimize the structure of this catalyst,
we undertook a survey of relevant phosphoramide structures
around the basic skeleton of 7a (Chart 1). The phosphoramides
shown in Chart 1 were all prepared in enantiomerically pure
form as has been described previously.15
2.1.1.1. Survey of External Substituents. Although it had
been established early on that the piperidinyl group is critical
for high stereoselectivities, the ease with which catalysts bearing
modifications on the external nitrogen can be prepared allowed
us to readily examine the effects of these peripheral modifica-
tions. The results are listed in Table 1.
All of the phosphoramides 7m through 7q effectively
catalyzed the reaction to give good yields of aldol products,
however both the diastereo- and the enantioselectivities varied
dramatically. Whereas 7a was highly anti diastereoselective, 7m
and 7q were less so (Table 1, entries 1 and 5) and 7p was only
slightly anti selective (Table 3, entry 4). On the other hand with
7n and 7o the diastereoselectivity of the reaction was reversed
and highly syn selective reactions were observed (Table 1,
entries 2 and 3)! In the anti manifold, phosphoramide 7a was
highly enantioselective, but all of the other catalysts were less
selective. There is a general trend in the enantiomeric excess
of the anti products and the syn/anti ratio of the reactions; i.e.,
SCHEME 3
(15) Denmark, S. E.; Su, X.; Nishigaichi, Y.; Coe, D. M.; Wong, K.-T.;
Winter S. B. D.; Choi, J. Y. J. Org. Chem. 1999, 64, 1958-1967.
3906 J. Org. Chem., Vol. 71, No. 10, 2006

Aldol Additions of Ketone Trichlorosilyl Enolates
TABLE 1. Influence of External Substituents in Catalyst 7
creased the formation of syn product. Additionally, as exempli-
fied by the use of 7k, extremely bulky substituents proximal to
the Lewis-basic oxygen diminish the relative rates of aldol
additions. Again, the parent phosphoramide 7a was the best
catalyst in terms of enantioselectivity (Table 2, entry 2). Ethyl
substituents at the R2 position in 7 resulted in only a slight
decrease in enantioselectivity of the anti product (with the same
sense of asymmetric induction), but dramatically decreased
diastereoselectivity (Table 2, entry 3). Aromatic groups on this
position constituted highly syn-selective catalysts (Table 2,
entries 4 and 5), although only moderate enantioselectivity was
observed in the syn manifold. Phosphoramide 7k appeared
syn erb
(config)
anti erb
(config)
yield,c
%
entry
1
cat.
7md NMe2
NR3
syn/antia
1/19
2
61.5/38.5
93.3/6.7
91
89
90
86
89
94
(2S,1′S)
52.4/47.6 81.1/18.9
(2R,1′R) (2R,1′S)
50.0/50.0 60.0/40.0
(2R,1′S)
50.0/50.0 70.6/29.4
(2S,1′R)
(2R,1′S)
2
3
4
5
6
7nd
7od
7pe
7qd
7ad
N(n-Pr)2
N(i-Pr)2
NPh2
16/1
32/1
1/1.9
1/8.9
1/60
1
highly congested, and indeed its H NMR, 13C NMR, and 31P
NMR clearly showed the coexistence of several rotamers at
room temperature. This also explains the slow reaction in the
system and low yield. Even with 1 equiv of 7k, the reaction
only gave 52% yield of the syn product with 46% ee (Table 4,
entry 6). Hydrogen substitutions at the internal position (7h)
resulted in many side reactions including deprotonation of the
phosphoramide, and a very low yield of the syn product was
obtained with poor enantioselectivity (Table 2, entry 1).
2.1.1.3. Survey of the Backbone Substituents. Having
surveyed the effects of modifications to the external nitrogen,
in addition to the ring substituents proximal to the Lewis-basic
oxygen, we next turned our attention to studying the effects on
rate and selectivity by modification of the chiral backbone. Table
3 compiles the results with different phosphoramides.
N(CH2)4
N(CH2)5
54.6/45.4 88.0/12.0
(2S,1′S)
(2R,1′S)
96.0/4.0
(2R,1′S)
a Determined by 1H NMR analysis of the crude product mixture.
b Determined by CSP HPLC analysis; absolute configuration was assigned
by the elution order from CSP HPLC. c Chromatographically homogeneous
material. d (S,S)-Catalyst was used. e (R,R)-Catalyst was used.
higher enantioselectivity corresponds to a lower syn/anti ratio
except in entry 2, which was highly syn selective but with
moderate enantioselectivity of the anti product. The enantiose-
lectivities of the syn products were low in all reactions, even
in highly syn selective reactions.
2.1.1.2. Survey of Internal Substituents. The effects of the
internal substituents on the nitrogens of the 2,5-diazaphospho-
lidine ring were next to be examined. While maintaining the
external piperidinyl unit, stilbene-1,2-diamine derivatives bear-
ing modifications at the N,N′ positions were employed to provide
a series of phosphoramides containing variations around the 2,5-
diazaphospholidine skeleton. The results are collected in Table
2.
TABLE 3. Influence of the Backbone Substituents in Catalyst 7
syn erb
(config)
anti erb yield,c
entry cat.
R1
syn/antia
1/46
(config)
%
1
2
3
4
5
7ad Ph
58.3/41.7 95.5/4.5
(2S,1′S) (2R,1′S)
88
TABLE 2. Influence of the Internal Nitrogen Substituents in
Catalyst (S,S)-7
7be 4-CF3C6H4
1/2.5
1/53
1/10
69/1
60.0/40.0 92.3/7.7
(2R,1′R) (2S,1′R)
65.5/34.5 96.0/4.0
(2R,1′R) (2S,1′R)
69.7/30.3 90.5/9.5
(2S,1′S)
52.4/47.6
(2R,1′R)
80
80
98
71
7ce
7dd 3,5-Me2C6H3
4-MeOC6H4
(2R,1′S)
7ed
cyclohexyl
syn erb
(config)
anti erb
(config)
yield,c
%
R2
syn/antia
6
7
7ff
1-naphthyl
1/3.0
1/56
82
76
entry cat.
7 ge 2-naphthyl
96.0/4.0
1
7h
7a
7i
H
>50/1
54.6/45.4
(2R,1′R)
26
94
92
94
31
52
(2S,1′R)
a Determined by 1H NMR analysis of the crude product mixture.
b Determined by CSP HPLC analysis; absolute configuration was assigned
by the elution order from CSP HPLC. c Chromatographically homogeneous
material. d (S,S)-Catalyst was used. e (R,R)-Catalyst was used. f Racemic
catalyst was used.
2
Me
Et
1/60
1/2.0
97/1
44/1
45/1
96.0/4.0
(2R,1′S)
3
52.4/47.6 94.4/5.6
(2R,1′R)
76.7/23.3
(2R,1′R)
64.3/35.7
(2R,1′R)
73.0/27.0
(2R,1′R)
(2R,1′S)
4
7j
Ph
5
7k
7k
1-naphthyl
1-naphthyl
All the phosphoramides used in this study were efficient
catalysts and gave good to high yields of products. Aromatic
substituents on the R1 position all provided anti-selective
reactions, while the cyclohexyl group gave a highly syn-selective
reaction with essentially no enantioselectivity (Table 3, entry
5). Catalysts bearing electron-withdrawing groups such as in
7b appear to be less anti selective than their counterparts
containing electron-donating groups 7c (Table 3, entries 2 and
3). Phosphoramides with sterically demanding groups such as
3,5-dimethylphenyl and 1-naphthyl were also less anti selective
(Table 3, entries 4 and 6). The enantioselectivity of the anti
6d
a Determined by 1H NMR analysis of the crude product mixture.
b Determined by CSP HPLC analysis; absolute configuration was assigned
by the elution order from CSP HPLC. c Chromatographically homogeneous
material. d 100 mol % of 7k was used.
With the exception of 7h (R2 ) H), which afforded poor
yields and little to no enantioselectivity, increasing the size of
the N,N2 substituents on the 2,5-diazaphospholidine ring in-
J. Org. Chem, Vol. 71, No. 10, 2006 3907

Denmark et al.
products was high in all cases with the same sense of asymmetric
induction. Slightly lower enantioselectivities were observed for
7b and 7d (92.3/7.7 and 90.5/9.5 er, respectively). Parallel
experiments showed that both 7c and 7g gave nearly identical
results as the parent phosphoramide 7a in terms of both
diastereoselectivity and enantioselectivity (Table 3, entries 3 and
7 vs entry 1). The enantioselectivity in the syn manifold was
low, but the same sense of asymmetric induction was observed
for entries 1-4 in Table 5. Phosphoramide 7f was highly
congested as judged by rotamers observed by NMR spectros-
copy at room temperature and 60 °C. Because of the low
selectivity obtained with 7f, resolution of the precursor diamine
was not attempted.
observed previously, increasing the steric bulk around the Lewis-
basic oxygen produces a shift in diastereoselectivity. While both
7a and 7m gave syn products, the sterically more demanding
catalysts 7n and 7i gave anti products preferentially (Table 4,
entries 3 and 4). The enantioselectivities of the syn products
were high for 7a (97.0/3.0 er) and 7m (90.9/9.1 er) and only
moderate for the relatively bulky, anti-selective catalysts 7n and
7i. In fact, the enantioselectivities for the anti products were
low for all the catalysts (from 54.6/45.4 to 65.5/34.5 er). In the
syn manifold the major enantiomer had the absolute configu-
ration (2S,3S) except 7n, which gave the syn product with
(2R,3R) configuration. All the anti products had the (2R,3S)
configuration although the enantioselectivities were low.
2.1.1.4. Phenethylamine-Derived Phosphoramides. Chiral
phosphoramides 8 derived from (S)-1-phenethylamine were also
synthesized and used in the asymmetric aldol reaction, Chart
2. Both catalysts afforded aldol products 6a in greater than 97%
yields. With methyl groups on the internal nitrogens, 8a was
slightly anti selective while 8b, with sterically more demanding
phenyl groups in the internal positions, was highly syn selective,
providing 1/4.2 and 59/1 syn/anti, respectively. Enantioselec-
tivities for both reactions were low, less than 58.3.4/41.7 er,
and a switch of absolute configuration of the anti product was
observed from 8a to 8b.
2.2. Survey of Achiral Phosphoramides. In view of the
dramatic effects of substituents demonstrated for the chiral
catalysts, a series of structurally simpler achiral phosphoramides
was prepared. Because the level of diastereo- and enantiose-
lectivities were generally not coupled within the phosphoramide-
catalyzed aldol additions, a diverse series of achiral agents
should allow for a better understanding of the factors governing
the mode of diastereoselection.
2.2.1. Reaction with Cyclohexanone-Derived Enolate 4.
HMPA and two series of achiral phosphoramides 11 and 12
having the basic diazaphospholidine skeleton were used in the
reaction of enolate 4 and benzaldehyde, Table 5. By analogy
to the studies involving chiral phosphoramides, increasing steric
bulk around the site of coordination decreases the relative rate
of reaction and alters the diastereoselectivity from anti to syn.
Thus, phosphoramides 11a, 11c, 11d, and 12a were faster acting
catalysts than the acyclic phosphoramide HMPA, whereas 11b
was comparable to HMPA (Table 5, entry 3) and 12b was
significantly slower than HMPA (Table 5, entry 7). It is
interesting to note that phosphoramide 11d, which is more
sterically hindered than 11b, actually gave a higher conversion
than 11b after 8 min at -78 °C (Table 5, entry 5 vs entry 3).
Only 11a and 12a gave moderately anti selective reactions
(Table 5, entries 2 and 6) while the remainder of the catalysts
were syn selective (syn/anti ranging from 3.3/1 to 38/1). The
synthetic utility of these catalysts was also demonstrated in
preparative experiments that were allowed to proceed to
CHART 2
2.1.2. Reaction with Propiophenone-Derived Enolate 9. To
evaluate the effects of different catalysts on the aldol addition
of enolates bearing the opposite geometric configuration, several
phosphoramides were examined in the reaction of (Z)-912 and
5a, Table 4.
All the phosphoramides, with the exception of 7q, gave good
yields of the aldol products with the parent phosphoramide 7a
still affording the best selectivities. In general, the diastereo-
selectivity for enolate (Z)-9 was less than that of enolate 4. As
TABLE 4. Influence of Phosphoramide Structure on the Selectivities of Reaction with (Z)-9
syn erb
(config)
anti erc
(config)
yield,d
%
entry
1
cat.
R1/R2
syn/antia
12/1
7a
(CH2)5/Me
97.0/3.0
(2S,3S)
90.9/9.1
(2S,3S)
70.6/29.4
(2R,3R)
50.0/50.0
54.6/45.4
(2R,3S)
60.0/40.0
(2R,3S)
63.0/37.0
(2R,3S)
65.5/34.5
(2R,3S)
77
74
77
41
74
2
3
4
5
7m
7n
7q
7i
Me/Me
5.7/1
1/9.0
1/1.6
1/1.3
i-Pr/Me
(CH2)4/Me
(CH2)5/Et
72.2/27.8
(2S,3S)
56.5/43.5
(2R,3S)
a Determined by 1H NMR analysis of the crude product mixture. b Determined by CSP HPLC analysis; absolute configuration was assigned by the elution
order from CSP HPLC. c Absolute configuration was tentatively assigned by comparison to the literature value of optical rotation.16,17 d Chromatographically
homogeneous material.
3908 J. Org. Chem., Vol. 71, No. 10, 2006

Aldol Additions of Ketone Trichlorosilyl Enolates
completion. All of the catalysts gave high yield of the aldol
products after 1.5 to 6 h at -78 °C with essentially the same
selectivities (see the last two columns in Table 5).
that the syn and anti pathways are mechanistically distinct. In
other words, although the syn and anti pathways are governed
by the concentration of catalyst, the enantioselectivity of the
individual pathways remains constant regardless of catalyst
concentration. This possibility had been previously encountered
in our work with cyclopentanone and cycloheptanone derived
enolates wherein the rate of addition of benzaldehyde was
critical to obtain high diastereoselectivity.5c When the aldehyde
was added slowly, high and reproducible anti-selectivity was
obtained; with fast addition, lower and variable diastereoselec-
tivity was observed. In this case as well, the enantiomeric ratio
of the anti-configured products was essentially constant over a
range of reaction conditions, suggesting the presence of two
mechanistically distinct pathways.
TABLE 5. Achiral Phosphoramides in Catalyzed Aldol Additions
of 4
conversion,a
%
time,
h
yield,c
%
entry
cat.
syn/antib
syn/antib
TABLE 6. Phosphoramide Loading Effects in Aldol Addition with
7a
1
2
3
4
5
6
7
HMPA
11a
11b
11c
11d
38
82
35
76
83
87
8
3.3/1
1/2.6
38/1
16/1
19/1
1/1.6
20/1
1.5
6.0
1.5
1.5
3.0
1/2.8
27/1
31/1
40/1
1/2.7
99
93
96
95
96
12a
12b
a Determined as the ratio of remaining benzaldehyde and products by
1H NMR analysis of the crude product mixture after 8 min. b Determined
by 1H NMR analysis of the crude product mixture. c Chromatographically
homogeneous material.
loading, conc,
yield,d
%
entry
mol %
M
syn/antia
er (syn)b
er (anti)b,c
1
2
3
4
5
10
5
2
10
100
0.5
0.5
0.5
0.1
0.1
1/14
1/10
1/2.4
1/28
1/95
60.0/40.0
56.5/43.5
56.5/43.5
60.0/40.0
77.8/22.2
94.4/5.6
94.7/5.3
93.8/6.2
95.5/4.5
95.7/4.3
94
90
84
90
97
2.2.2. Reaction with Propiophenone-Derived Enolate 9. As
with the chiral phosphoramides, the effect of the achiral catalysts
on the reaction of Z-enolates was evaluated. Thus, phosphor-
amides 11d and 12a were also used in the reaction of (Z)-9
with benzaldehyde. With 15 mol % of 11d or 12a, the aldol
products 10 were obtained in 58% or 82% yields with moderate
anti selectivity, Scheme 4.
a Determined by 1H NMR analysis. b Determined by CSP HPLC analysis.
c (2R,1′S)/(2S,1′R). d Chromatographically homogeneous material.
Under the hypothesis that the “slow addition effect” should
be relevant in the reaction with enolate 4 as well, another catalyst
loading study with 7a was conducted, this time with the
aldehyde added slowly over roughly 1 h, Table 7. Clearly there
is a dramatic effect, as the level of diastereoselectivity changed
from 1/2.4 to 1/28 with 2 mol % catalyst for fast and slow
additions of benzaldehyde, respectively. In fact, even with 0.5
mol % catalyst, the reaction is moderately anti selective (syn/
anti ) 1/5). As before, the enantioselectivity of the anti process
remained unchanged, even over a 20-fold change in phosphora-
mide loading.
SCHEME 4
2.3. Influence of Substrate/Catalyst Stoichiometry. 2.3.1.
Chiral Phosphoramides. In the process of optimizing the
conditions for the asymmetric aldol addition of 4 to benzalde-
hyde, an interesting catalyst loading effect was discovered. Initial
studies with an acetone derived enolate (vide infra) suggested
that reactions at low catalyst loadings and 0.1 M were slow, so
the present phosphoramide loading study with enolate 4 was
performed at 0.5 M concentration, Table 6. Interestingly, as the
catalyst loading was reduced the process became less anti
selective. In addition, simply changing the concentration of the
reaction also significantly lowered the diastereomeric ratio of
the product (compare entries 1 and 4 in Table 6). Even more
interesting was the fact that the enantiomeric ratio of the anti-
isomer remained essentially unchanged throughout the range
of catalyst loadings examined. This suggests that the diastereo-
and enantioselectivity determining factors are independent and
TABLE 7. Catalyst Loading Effect with 7a and Slow Addition of
Benzaldehyde
entry loading, mol % syn/antib er (syn)c er (anti)c,d yield,e %
1
2
3
10
2
0.5
<1/50
1/28
1/5
58.3/41.7
52.4/47.6
52.4/47.6
95.5/4.5
95.7/4.3
95.5/4.5
94
96
53
a 1 h addition time to a final concentration of 0.1 M. b Determined by
1H NMR analysis. c Determined by CSP SFC analysis. d (2R,1′S)/(2S,1′R).
e Chromatographically homogeneous material.
In a more complete catalyst loading study involving the
acetone-derived trichlorosilyl enolate 1, a less dramatic effect
was observed, Table 8. Above a ceiling catalyst loading (roughly
5 mol %) there was little variation in the enantiomeric ratio of
(16) Ishihara, K.; Maruyama, T.; Mouri, M.; Gao, Q.; Furuta, K.;
Yamamoto, H. Bull. Chem. Soc. Jpn. 1993, 66, 3483-3491.
(17) Juaristi, E.; Beck, A. K.; Hansen, J.; Matt, T.; Mukhopadhyay, T.;
Simson, M.; Seebach, D. Synthesis 1993, 1271-1290.
J. Org. Chem, Vol. 71, No. 10, 2006 3909

Denmark et al.
TABLE 9. Loading Effect for the Addition of 4 to Benzaldehyde
Using Catalyst 11ca
the aldol adduct even up to 20 mol % of phosphoramide.
However, when even lower phosphoramide loadings (down to
1 mol %) were used, the enantiomeric ratio of the product
dropped slightly, though the yield of the overall process
remained high when the reaction was performed at 0.5 M. This
effect stands in contrast to the lack of an enantioselectivity
dependence on catalyst loading in the aldol additions with the
cyclohexanone-derived enolate 4 above; perhaps suggesting the
origin of the diminished enantioselectivity at lower catalyst
loading in 1 is connected to the factors controlling diastereo-
selectivity for reactions of 4 (vide infra).18
entry
loading, mol %
dr,b syn/anti
yield,c %
1
2
3
4
5
6
7
8
1.8
2.1
9.1
9.1
46
130/1
107/1
31/1
95
97
96
97
83
95
97
96
22/1
4.8/1
4.4/1
1.4/1
1.3/1
TABLE 8. Catalyst Loading Effect with 7aa
46
194
201
a 1 h addition time to a final concentration of 0.1 M. b Determined by
1H NMR analysis. c Chromatographically homogeneous material.
At lower catalyst loading, higher syn/anti ratio was observed.
A straight line with good correlation (R2 ) 0.997) was obtained
when the syn/anti ratio was plotted against 1/[11c]. This
dramatic influence of the catalyst loading on selectivity clearly
implies the simultaneous operation of competitive pathways in
the reaction (vide infra).
Phosphoramide 12a also showed a loading dependence in the
reaction of 4 and 5a (Table 10), albeit a much smaller effect
was observed, Figure 3. The catalyst 12a was moderately anti
selective (syn/anti 1/7-8) at high catalyst loadings while slightly
syn selective (syn/anti 1.2/1) at lower catalyst loadings (2.1 mol
%). Again a linear relationship can be derived between the syn/
anti ratio and 1/[12a] (R2 ) 0.988).
entry
loading, mol %
conc, M
erb (S/R)
yield,c %
1
2
3
4
5
6
7
20
10
10d
5
3
2
0.1
0.1
0.5
0.5
0.5
0.5
0.5
92.3/7.7
91.7/8.3
7.7/92.3
92.3/7.7
89.6/10.4
88.6/11.4
83.6/16.4
82
90
87
88
86
88
92
1
a 1 h addition time to a final concentration of 0.1 M. b Determined by
CSP HPLC analysis of the corresponding 2,4-dinitrophenyl carbamates.
c Chromatographically homogeneous material. d Performed with 10 mol %
of (R,R)-7a.
2.3.2. Achiral Phosphoramides. 2.3.2.1. Reaction with
Cyclohexanone-Derived Enolate 4. The dependence of the
diastereomeric ratio of the aldol products on the catalyst loading
has also been observed with the achiral catalysts. In the reaction
of 4 and 5a, the ratio of phosphoramide 11c to the enolate 4
was very important in determining the syn/anti ratio of product
6a. Under otherwise identical conditions, the syn/anti ratio of
6a was 130/1 when 2 mol % of 11c was used, and the ratio
dropped to 2/1 when 200 mol % of the same catalyst was used,
Table 9. A detailed study on this intriguing loading effect was
undertaken, and the results are depicted in Figure 2.
TABLE 10. Loading Effect for the Addition of 4 to Benzaldehyde
Using Catalyst 12aa
entry
loading, mol %
dr,b syn/anti
yield,c %
1
2
3
4
5
6
7
8
9
10
2.1
2.4
1.2/1
1.0/1
1/1.6
1/2.7
1/2.4
1/4.3
1/7.9
1/8.1
1/6.8
1/6.1
96
92
98
96
97
99
90
93
91
ND
5.4
9.4
9.5
19.7
46
49
195
199
a 1 h addition time to a final concentration of 0.1 M. b Determined by
1H NMR analysis. c Chromatographically homogeneous material.
FIGURE 2. Dependence of product syn/anti ratio on the loading of
11c.
(18) It should be noted that methyl ketone derived trichlorosilyl enolates
are less reactive than the cyclohexanone derived enolate 4. In a control
experiment, the trichlorosilyl enolate derived from methyl n-butyl ketone
and 4-phenylbenzaldehyde were stirred at -78 °C for 2 h, providing only
4% of the corresponding aldol adduct, with 95% recovery of the unchanged
aldehyde after column chromatography.
FIGURE 3. Loading effects of 12a in the reaction of 4 with
benzaldehyde.
3910 J. Org. Chem., Vol. 71, No. 10, 2006

Aldol Additions of Ketone Trichlorosilyl Enolates
2.3.2.2. Propiophenone-Derived Enolate 9. Whereas the
reaction with enolate 4 showed dramatic dependence of product
syn/anti ratio on catalyst loading, the reaction with enolate 9
was essentially independent of catalyst loading. With catalyst
11d, in a 7-fold range of the catalyst loadings, 14-98 mol %,
the reaction produced essentially the same syn/anti ratio, 1/1.4
to 1/1.5. With catalyst 12a the syn/anti ratio increased slightly
(1/5.7 to 1/4.1) in a 6.7-fold increase of catalyst loadings from
15 to 100 mol %.
of benzoic acid was minimal. To understand the nature of these
effects, we initiated studies to investigate the influence of
benzoic acid and other acids in the reaction.
In the reaction of enolate 4 and freshly distilled benzaldehyde,
known quantities of acids were added, and the results are listed
in Table 12. Without benzoic acid, the reaction of 4 and 5a
was catalyzed by 2 mol % of phosphoramide 12a to give high
yield of the aldol products 6a in a syn/anti ratio of 1.2/1 (Table
12, entry 1). In the presence of 2 mol % benzoic acid the reaction
yield was still good and the syn/anti ratio increased to 89/1
(Table 12, entry 2). With (S)-mandelic acid (10 mol %) and
12a (9.6 mol %), the reaction was again highly syn selective
and afforded a high yield of product, but with no detectable
enantiomeric enrichment (Table 12, entry 3). Strong acids such
as triflic acid inhibited the catalytic ability of 12a, and the results
observed were essentially the same as the background reaction
(Table 12, entry 4). With phosphoramide 11c a similar increase
in the syn selectivity by benzoic acid was observed (Table 12,
entries 5 and 7 vs entries 6 and 8) although the effects were
much less dramatic as in the case of 12a. In the absence of
phosphoramide, benzoic acid itself (2 mol %) did not promote
the reaction and only a background reaction was observed (Table
12, entry 9).
2.3.3. The Effects of Acids on the Aldol Addition Reac-
tions. In the course of these loading effect studies we noticed
that a small amount of benzoic acid (13), which was present as
a contaminant in the benzaldehyde used, dramatically altered
the syn/anti ratio of the reaction with enolate 4. Freshly distilled
benzaldehyde was stored under nitrogen in the freezer, but after
several days a small amount of benzoic acid may accumulate
1
and can be subsequently detected by H NMR. Table 11 lists
the results of a loading study with 12a and obtained with
benzaldehyde spiked with 2% of benzoic acid.
TABLE 11. Benzoic Acid Effects on Syn/Anti Ratio of the Aldol
Addition Reactions
TABLE 12. Acid Effects in the Reaction between 4 and
Benzaldehydea
entry
loading 12a, mol %
syn/antia
yield,b %
1
2
3
4
5
6
2.0
3.6
5.6
7.3
9.1
72/1
33/1
23/1
4.0/1
1/1.1
1/2.9
81
95
85
90
98
95
entry cat.a (mol %)
additivea (mol %)
syn/antib yield,c %
18.4
1
2
3
4
5
6
7
8
9
12a (2.1)
12a (2.0)
12a (9.6)
12a (2.1)
11c (2.1)
11c (2.0)
11c (9.1)
11c (9.8)
none
none
PhCO2H (2.0)
(S)-mandelic acid (10)
CF3SO3H (2.1)
none
1.2/1
89/1
96
87
84d
25
97
97
97
92
25
a Determined by 1H NMR analysis of the crude product mixture.
b Chromatographically homogeneous material.
76/1
140/1
107/1
200/1
22/1
With catalyst 12a, which was unselective in the reaction of
enolate 4 with benzaldehyde (see Table 5, entry 6), the reaction
became highly syn selective at low loadings in the presence of
benzoic acid (Table 11, entry 1). The yields of all the reactions
were high. A comparison of the loading effects with and without
benzoic acid is plotted in Figure 4. This dramatic increase in
the stereoselectivity only manifests itself at lower catalyst
loading. At higher catalyst loading the impact of small amount
PhCO2H (2.0)
none
PhCO2H (9.8)
PhCO2H (2.0)
45/1
>200/1
a Relative to [4]. b Determined by 1H NMR analysis of the crude mixture.
c Chromatographically homogeneous material. d Racemic.
The reaction of enolate (Z)-9 also showed a moderate benzoic
acid effect, Scheme 5. In the absence of benzoic acid, the
reaction of the (Z)-9 with benzaldehyde gave an 82% yield of
the aldol product with a syn/anti ratio of 1/5.6, while the addition
of 15 mol % benzoic acid increased the syn/anti ratio to 1/1.5.
Thus, with both Z- and E-enolates, the effect of benzoic acid
was to increase the syn selectivity of the reactions.
SCHEME 5
FIGURE 4. Loading effects of phosphoramide 12a in the presence
and absence of benzoic acid ([ without benzoic acid, 9 with benzoic
acid).
Pronounced benzoic acid effects have also been found in the
aldol addition reaction of 4 and 5a catalyzed by 7a, Table 13.
J. Org. Chem, Vol. 71, No. 10, 2006 3911

Denmark et al.
With freshly distilled benzaldehyde, the reaction was highly anti
selective and highly enantioselective in the anti manifold (Table
13, entry 1). With 1 mol % of benzoic acid, only a slight
decrease in anti selectivity and yield was observed (Table 13,
entry 2 vs entry 1). With 10 mol % benzoic acid (equivalent to
phosphoramide 7a), the reaction becomes syn selective and
much slower than the reaction without benzoic acid (only 24%
conversion after 8 min, Table 13, entry 3). The syn product
from the reaction with 10 mol % of benzoic acid was essentially
racemic while the anti product had decreased enantioselectivity
compared to entry 1. The reaction with 20 mol % of benzoic
acid was also rather slow. After extended time (2 h) the reaction
was highly syn selective although the conversion was still low
(62%, Table 13, entry 4). The syn product from the reaction
was racemic, while the anti product showed moderate enanti-
oselectivity (77.3/22.7 er).
employing 10 mol % of phosphoramide 7a, Figure 5. Thus,
with 40% ee of catalyst 7a, the anti product 6a was obtained in
53% ee (76.7/23.3 er).20
TABLE 13. Benzoic Acid Effect in the Reaction Promoted by
Chiral Catalyst 7a
FIGURE 5. Relationship between the enantiopurity of anti-6a and
(R,R)-7a.
PhCO2H, time,
entry mol %
min syn/antib
conv, syn erd
anti erd yield,e
c
%
(config) (config) %
1a
0
1
8
8
1/28
1/17
4.1/1
18/1
100 54.6/45.4 95.5/4.5
(2S) (2R)
91 54.6/45.4 95.5/4.5
(2S) (2R)
96
90
18
60
The above data fits very well with Kagan’s two-ligand
model.21 Thus, assuming two chiral ligands are involved in the
stereodetermining transition structure and assuming a statistical
distribution of the enantiomeric ligands, the enantioselectivity
of the product (eeprod) and the enantiopurity of the ligands (eecat)
are related by the following equation
2
3
10
20
8
24 50.0/50.0 89.0/11.0
(2R)
62 50.0/50.0 77.3/22.7
(2R)
4
120
a Parallel with other experiments in the table. b Determined by 1H NMR
analysis of the crude product mixture. c Determined as the ratio of remaining
benzaldehyde and products by 1H NMR analysis of the crude product
mixture. d Determined by CSP HPLC analysis; absolute configuration was
assigned by the elution order from CSP HPLC. e Chromatographically
homogeneous material.
2
eeprod ) ee0ee
(1)
cat 1 + g + (1 - g)eecat2
where ee0 is the product enantioselectivity when eecat equals
100% and g ) kRS/kRR, where kRS/kRR is the ratio of the rate
constants through heterodimeric (kRS) and homodimeric (kRR
)
2.4. Nonlinear Effects. From the loading effect studies, we
hypothesized that catalyst concentration dependent stereoselec-
tivity is indicative of competing reaction pathways involving
different molecularity in the transition structure assembly. We
reasoned that if more than one phosphoramide were indeed
involved in one of the pathways (syn- or anti-selective) then
nonlinear effects19 could be observed with nonracemic phos-
phoramide 7a. Moreover, if one of the two pathways proceeds
through a transition structure involving only one phosphoramide
bound to silicon, then a linear relationship should exist between
the enantiomeric composition of the aldol product and the
catalyst enantiopurity.
transition states. An iterative line fitting of the data in Figure 5
to the above equation gave g ) 0.277 with R2 ) 1.000.
Therefore, the calculated reaction rate constant of a het-
erodimeric transition structure is about 28% of that of a
homodimeric transition structure.
2.4.2. The Syn Manifold. Enantioenriched phosphoramide
7j was prepared in a similar manner as above. Thus, the
calculated amounts of the corresponding (S,S)-7j and (()-7j
were weighed into a reaction vessel which was then dried under
high vacuum. The phosphoramides were then dissolved in
methylene chloride under argon and employed in aldol addi-
tions.
2.4.1. The Anti Manifold. Enantioenriched samples of the
phosphoramide were prepared by weighing the calculated
amounts of (S,S)-7a and (R,R)-7a into the reaction vessel which
was then dried under high vacuum. The phosphoramides were
then dissolved in methylene chloride under argon and used in
aldol additions of 4 to 5a at -78 °C. The results for 20, 40, 60,
and 80% ee catalyst showed a positive nonlinear effect when
The results for 0, 20, 40, 60, 80, and 100% ee of catalyst 7j
were plotted in Figure 6, and indeed, a linear relationship
between the syn product enantioselectivity and the enantiopurity
(20) For discussion of nonlinear effects, the use of enantiomeric excess
(ee) is preferred. For an disucssion of the relative merits of er vs ee, see:
(a) Kagan, H. B. Recl. TraV. Chem. Pays-Bas 1995, 114, 203-205. (b)
Schurig, V. Enantiomer 1996, 1, 139-143. Gawley, R. E. J. Org. Chem.
2006, 73, 2411-2416.
(19) (a) Puchot, C.; Samuel, O.; Dun˜ach, E.; Zhao, S.; Agami, C.; Kagan,
H. B. J. Am. Chem. Soc. 1986, 108, 2353-2357. (b) Guillaneux, D.; Zhao,
S.-H.; Samuel, O.; Rainford, D.; Kagan, H. B. J. Am. Chem. Soc. 1994,
116, 9430-9439.
(21) (a) Fenwick, D. R.; Kagan, H. B. Top. Stereochem. 1999, 22, 257-
296. (b) Girard, C.; Kagan, H. B. Angew. Chem., Int. Ed. 1998, 22, 2922-
2959. (c) Avalos, M.; Babiano, R.; Cintas, P.; Jime´nez, J. L.; Palacios, J.
C. Tetrahedron: Asymmetry 1997, 8, 2997-3017.
3912 J. Org. Chem., Vol. 71, No. 10, 2006

Aldol Additions of Ketone Trichlorosilyl Enolates
of catalyst 7j was observed with R2 ) 0.999. These observations
strongly support the hypothesis that two phosphoramide mol-
ecules are involved in the transition structure in the anti-selective
pathway, whereas only one phosphoramide molecule is present
in the transition structure leading to the syn product.
“reservoir effect” has been invoked by Noyori and co-workers
to explain the highly positive nonlinear effect in the diethylzinc
addition of benzaldehyde catalyzed by (-)-3-exo-(dimethyl-
amino)isoborneol (DAIB).23 In contrast to the results obtained
by Noyori and co-workers, essentially no dependence of the
degree of nonlinearity on the concentration of catalyst 7a was
observed (Table 14, entry 1 vs 5). Also, a reservoir effect cannot
explain the change of diastereoselectivity on the catalyst loading.
Thus, taken together, these results strongly support a transition
structure with more than one phosphoramide molecule. In
addition, these control experiments clearly rule out other
interpretations for nonlinear effects and support the hypothesis
that two phosphoramide molecules are present in the transition
structure and are responsible for the nonlinear effects observed.
2.4.3.2. Low-Temperature NMR Studies. Several attempts
were made to reveal the nature of the silicon-phosphoramide
association in the solution state.24 Low-temperature experiments
were designed to probe the existence of multiple complexes of
phosphoramides with SiCl4. Unfortunately, low-temperature 31
P
NMR studies could not reveal any distinct signals indicative of
complex formation. Rather, the presence of broad peaks and
the disappearance of the original phosphorus signal upon
addition of SiCl4 suggest complexation is highly dynamic,
possibly involving the rapid equilibration of multiple phosphorus-
silicon complexes.25
FIGURE 6. Relationship between the enantiopurity of syn-6a and
(S,S)-7j.
2.4.3. Control Experiments. 2.4.3.1. Fast Quenching Ex-
periments. An alternative rationale for the nonlinear effect is
the potential interaction between the catalyst and the product.
Although this “product-inhibition” process can lead to nonlinear
effects, it also affects a change in enantioselectivity as a function
of time. To rule out the possibility that the product from the
reaction may be involved in the stereochemical determining
step22 or in the nonlinear effect observed in Figure 5, the
enantiomeric composition of the product as a function of
conversion was determined, Table 14. With catalyst 7a in 40%
ee the conversion of the reaction was 55% after 10 s and with
essentially the same enantioselectivity as at 100% conversion
(Table 14, entries 1 and 3). The reaction with 10 mol % of 5a
was employed to mimic the reaction at 10% conversion and
gave anti-6a with similar enantioselectivity (Table 14, entry 4).
In the presence of 1 equiv of phosphoramide 7a, the possibility
that one of the enantiomers of the phosphoramide 7a 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.
2.5. Solvent Effects and Salt Effects. 2.5.1. Solvent Effects.
Given that the transition structure for these catalyzed aldoliza-
tions is proposed to involve cationic, hexacoordinate silicon
species, it is reasonable to expect that the nature of the reaction
medium would strongly influence both the overall rate and
selectivity of the transformation. To investigate this, two solvent
surveys were performed, both with fast and slow additions of
5a (Table 15). In the first series of experiments, 5a was added
quickly to a cold (-78 °C) solution of phosphoramide (S,S)-7a
(10 mol %) and 4 in various solvents (entries 1, 3, 5, and 7).
The benchmark reaction proceeded with high diastereoselectivity
(anti) and with high er in the anti manifold. The use of the less
polar solvents toluene and pentane provided significantly dimin-
ished dr and er, entries 5 and 7. The relatively low yield of the
reaction in pentane most likely represents the limited solubility
of the phosphoramide in this solvent at low temperature. When
Et2O was used, the diastereoselectivity of the process increased
(entry 3), However, the enantioselectivity of the anti process
decreased appreciably relative to that obtained in CH2Cl2.
Subsequent reactions were carried out with 2 mol % of
phosphoramide (Table 15, entries 2, 4, 6, and 8). Clearly, solvent
does have an effect on both the diastereoselectivity and the
enantioselectivity of the resultant aldol adducts. Additionally,
the trends are the same regardless of the reaction protocol used.
Diethyl ether provided higher diastereoselectivity relative to
CH2Cl2, though enantioselectivity was lowered. Toluene and
pentane provided low levels of both relative and absolute
stereochemical control. The use of propionitrile, however,
provided both high diastereoselectivity and only slightly at-
TABLE 14. Control Experiments Using Enantioenriched 7a
erc
entry 5a, equiv 7a, equiv time, s conv,a % yield,b % (anti-6a)
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
10
480
480
100
63
55
100
100
95
61
52
96
99
76.7/23.3
76.9/23.1
76.6/23.4
76.6/23.4
77.0/23.0
(22) Shibata, T.; Morioka, H.; Hayase, T.; Choji, K.; Soai, K. J. Am.
Chem. Soc. 1996, 118, 471-472.
(23) Kitamura, M.; Okada, S.; Suga, S.; Noyori, R. J. Am. Chem. Soc.
1989, 111, 4028-4036.
a Determined by 1H NMR analysis. b Chromatographically homogeneous
material. c Determined by CSP SFC analysis.
(24) For X-ray crystallographic studies of phosphoramides with tin
tetrachloride, see ref 11a.
(25) For a solution NMR studies of phosphoramides with silicon
tetrachloride, see: Denmark, S. E.; Beutner, G. L.; Wynn, T.; Eastgate, M.
D. J. Am. Chem. Soc. 2005, 127, 3774-3789.
Yet another explanation for a nonlinear effect is the existence
of aggregates in the resting state of the catalyst. This so-called
J. Org. Chem, Vol. 71, No. 10, 2006 3913

Denmark et al.
tenuated enantioselectivity in comparison with CH2Cl2. This
result is rather surprising given the relatively high Lewis basicity
of this solvent. Clearly such a nitrile donor cannot compete with
the phosphoramide catalyst, even when used as the reaction
solvent. At lower catalyst loadings, a difference in rate was also
apparent, with the reactions in Et2O and toluene being signifi-
cantly slower (based on yield, balance is unreacted aldehyde)
than experiments performed in either CH2Cl2 or propionitrile.
rivaling CH2Cl2 in both yield and enantioselectivity, though as
above with 4, the selectivity was slightly attenuated.
2.5.2. Salt Effects. All of the above studies indicated the
operation of competitive pathways involving one or two
phosphoramide molecules in the transition structure for pro-
moted aldolization of trichlorosilyl enolates. If two phosphora-
mide molecules and an aldehyde coordinate to the silicon of
the enolate, then ionization of a chloride ion from the trichlo-
rosilyl enolate is required to maintain hexacoordination.26 To
probe the necessity of ionization, the effects of tetrabutylam-
monium salts on the rate and selectivities of the reaction were
investigated. In the reaction promoted by phosphoramide 7a,
tetrabutylammonium chloride or triflate gave similar results, but
the reaction was too fast to conclude if these salts affect the
reaction. With the slower acting catalyst 7j (which gave 44%
conversion to the syn product 6a with 53% ee after 8 min at
-78 °C), it was clearly shown that Bu4N+Cl- inhibits the
reaction whereas Bu4N+OTf- accelerated the reaction, Scheme
6. Thus, in the presence of 1.2 equiv of Bu4N+Cl- or
Bu4N+OTf-, conversion of the reaction was 8% (63.0/37.0 er)
and 92% (77.3/22.7 er), respectively. Thus, the rates of aldol
addition of trichlorosilyl enolates catalyzed by phosphoramides
are greatly influenced by the presence of tetrabutylammonium
salts.
TABLE 15. Solvent Effects in Aldol Additions of 4 Catalyzed by
7a
loading,
entry mol %
solvent
er
er
(anti)
yield,c
%
(ꢀ)a
syn/antib (syn)
1
2
3
4
5
6
7
8
10d
CH2Cl2
(9.1)
CH2Cl2
(9.1)
Et2O
(4.3)
Et2O
(4.3)
toluene
(2.4)
toluene
(2.4)
1/35
1/28
1/>50
1/49
1/6.2
1/5.2
1/2
ND
1/1g
ND
96.2/3.8e
95.8/4.2g
89.5/10.5e
85
91
82
59
84
51
70
85
2f
10d
2f
1.5/1g 87.5/12.5g
1.4/1e 83.3/16.7e
1.1/1g 87.2/12.8g
1.2/1e 65.5/34.5e
1.4/1g 93.3/6.7g
SCHEME 6
10d
2f
10d
2f
pentane
(1.8)
CH3CH2CN
(37.5)
1/31
a At 20 °C, taken from the CRC Handbook of Chemistry and Physics,
72nd ed. b Determined by 1H NMR analysis. c Chromatographically ho-
mogeneous material. d PhCHO added over 1 min/0.1 M/2 h. e Determined
by CSP HPLC analysis. f PhCHO added over 1 h/0.1 M. g Determined by
CSP SFC analysis.
2.6. Kinetics. Initial kinetic experiments were performed
using catalyst 7j. The decision to start with this phosphoramide
allowed for the study of a slower acting catalyst prior to
evaluating the more active catalyst 7a. Moreover, previous
studies have suggested that 7j proceeds through a single
phosphoramide pathway, which should provide a more tractable
system for initial investigation.
2.6.1. Reactions Catalyzed by Phosphoramide 7j. Unlike
determining the kinetic order for the uncatalyzed reactions,
determining order in catalyst could not be realistically ac-
complished using pseudo-order conditions. Rather, the rate
constant for the catalyzed reaction was measured as a function
of catalyst concentration. Manipulation of standard rate equa-
tions reveals that under these conditions the slope of log(k) vs
log[catalyst] is equivalent to the kinetic order of the species in
question, in this case the catalyst.
When a survey of reaction solvents was performed with
acetone-derived enolate 1, a similar trend was found in the
enantioselectivity of the adducts, Table 16. Both yield and
selectivity were lowered dramatically on switching from
CH2Cl2 to Et2O, THF, or toluene. Trichloroethylene (Table 16,
entry 2) provided a good yield, though with lower enantio-
selectivity. Again, propionitrile was a very effective solvent,
TABLE 16. Survey of Solvents on the Addition of 1 to
Benzaldehyde
The order in phosphoramide was determined by measuring
the observed rate constant (kobs) for the aldol reaction between
cyclohexanone trichlorosilyl enolate 4 and benzaldehyde as a
function of concentration near typical catalyst loadings by
ReactIR analysis. Presentation of the results in a log-log plot
revealed, as expected, a linear relationship with a slope nearly
equal to 1, m ) 1.014, Figure 7.
Interestingly, inspection of this function suggested that the
rate did not approach zero as the concentration of catalyst
approached zero. Careful extrapolation to 0 mol % catalyst gave
an unrealistically high value for the theoretical background
entry
solvent
CH2Cl2
trichloroethylene
Et2O
toluene
THF
erb
yield,c %
1
2
3
4
92.9/7.1
92
88
37
48
59
88
80.0/20.0
68.8/31.2
64.3/35.7
77.3/22.7
10.3/89.7
5
6d
CH3CH2CN
a 5a added over 1 min, 0.1 M/2 h. b Determined by CSP HPLC analysis
of the corresponding 2,4-dinitrophenyl carbamates. c Chromatographically
homogeneous material. d Performed with 10 mol % (R,R)-7a.
(26) Short, J. D.; Attenoux, S.; Berrisford, D. J. Tetrahedron Lett. 1997,
38, 2351-2354.
3914 J. Org. Chem., Vol. 71, No. 10, 2006

Aldol Additions of Ketone Trichlorosilyl Enolates
rates by decreasing the temperature resulted in unstable baselines
and inconsistent results. Furthermore, attempts to reduce the
reaction rates by dilution of the reaction mixture gave poor
signal-to-noise ratios. Although kinetics studies using ReactIR
were inconclusive, Rapid Injection NMR (RINMR)27 analysis
was able to address the limitations of the previous methods. To
verify the reliability of the RINMR apparatus, activation
parameters for the aldol addition of 4 to 6a catalyzed by the
more bulky phosphoramide 7j were repeated using this instru-
ment. Careful analysis revealed good agreement with the
previous ReactIR studies, Table 17.
Thus secured, the more challenging task of studying reactions
promoted using catalyst 7a was undertaken. 1H NMR analysis
permitted operation at lower concentrations than previously
possible for the reaction of 4 with 5a. As described previously,
a plot of log(kobs) vs log[catalyst] allowed for the determination
of order in catalyst for catalyst concentrations between 0.009
and 0.016 M. When plotted, the slope of the logarithmic function
(m) was determined to be 2.113, Figure 9.
FIGURE 7. Plot of log(kobs - B) versus log([catalyst]) for the addition
of 4 to 5a catalyzed by 7a at T ) -35 °C. The graph depicts the linear
fit to f(x) ) mx + b (m ) 1.014, R2 ) 1.000).
reaction, as determined by earlier preparative studies. This
prompted us to examine the reaction rates at much lower catalyst
concentrations, Figure 8. The sharp change in slope at low
catalyst loadings is consistent with a change in mechanism for
this aldol addition. Indeed, these results fully support the
hypothesis that the promoted pathway involves ionization of
chloride and utilizes cationic siliconate species whereas the
unpromoted pathway does not. Inspection of Figure 8 also
reveals a lower catalyst loading limit of approximately 2 mol
% catalyst loading, at which the undesirable unpromoted
pathway becomes competitive with the promoted pathway.
FIGURE 9. Plot of log(kobs) versus log([catalyst]) for the addition of
4 to 6a catalyzed by 7a at T ) -80 °C. The graph depicts the linear
fit to f(x) ) mx + b (m ) 2.113, R2 ) 0.992).
Arrhenius activation energies were then determined using 10
mol % of catalyst over a 20° temperature range, between -70
and -90 °C (Table 18). Attempts to extend this range by
increasing the temperature above -70 °C gave reaction rates
which exceeded the ability of the RINMR. Furthermore, it was
decided not to decrease the operational temperature beyond -90
°C to avoid the possibility of sample freezing.
FIGURE 8. Plot of kobs versus catalyst loading for the addition of 4
to 5a, catalyzed by 7j.
Activation parameters for the catalyzed reaction could be
determined as before, now over a temperature range of -11 to
-65 °C, Table 17. As was observed for the uncatalyzed addition
of 4 to 5a, ∆Sq remains the predominant contributor to ∆Gq
The nearly vanishing value for ∆Hqreveals a greater level of
activation for the catalyzed reaction.
TABLE 18. Arrhenius Activation Energies for the Aldol Addition
of 4 to 5a Promoted by 7a As Determined by RINMR Analysisa
activation parameters
∆Hq (kcal mol-1
∆Sq (cal mol-1 K-1
∆Gq (kcal mol-1
activation values
)
1.5 ( 0.5
-51.9 ( 1.0
17.3 ( 1.0
)
TABLE 17. Arrhenius Activation Energies for the Aldol Addition
of 4 to 5a Catalyzed by 7j as Determined by Kinetic Analysisa
)
a Activation parameters were calculated for T ) 303 K.
activation values
(ReactIR)
activation values
(RINMR)
activation parameters
∆Hq (kcal mol-1
∆Sq (cal mol-1 K-1
∆Gq (kcal mol-1
2.7. Determination of the Turnover-Limiting Step. Through
extensive kinetics studies using GC, IR, and NMR analysis, the
molecularity for the Lewis base catalyzed aldol addition
)
0.3 ( 0.1
-67.1 ( 0.7
20.6 ( 0.2
1.2
-63.3
20.4
)
)
a Activation parameters were calculated for T ) 303 K.
(27) For examples on the use of rapid injection NMR techniques, see:
(a) McGarrity, J. F.; Prodolliet, J.; Smyth, T. Org. Magn. Reson. 1981, 17,
59-65. (b) Palmer, C. A.; Ogle, C. A.; Arnett, E. M. J. Am. Chem. Soc.
1992, 114, 5619-5625. (c) Reetz, M. T.; Raguse, B.; Marth, C. F.; Hu¨gel,
H. M.; Bach, T.; Fox, D. N. A. Tetrahedron 1992, 48, 5731-5742. (d) For
an in depth discussion of the design, construction and application of the
RINMR apparatus, see ref 10b.
2.6.2. Reactions Catalyzed by Phosphoramide 7a. Initial
attempts to study relations catalyzed by phosphoramide 7a using
the ReactIR techniques proved inconclusive due to extremely
fast reaction rates at -65 °C. Attempts to attenuate the reaction
J. Org. Chem, Vol. 71, No. 10, 2006 3915

Denmark et al.
involving trichlorosilyl enolates has been elucidated and shows
a rate expression as rate ) k[cat][enolate][aldehyde] for catalyst
7j and rate ) k[cat]2[enolate][aldehyde] for catalyst 7a.
Although the order of each component at the rate-determining
step is known, either complexation or aldolization is turnover
limiting. The experimentally determined reaction order is
consistent with both scenarios, and while the Arrhenius activa-
tion parameters may suggest complexation is turnover limiting
(large negative ∆Sq, small ∆Hq), it does not discount aldolization
for this event. To conclusively determine the turnover-limiting
step, kinetic isotope effects (KIE) from natural-abundance 13C
NMR analysis (as pioneered by Singleton) were utilized.28
Because the formation of diastereomers would be readily evident
in the analysis required for natural abundance kinetic KIE using
NMR spectroscopy, a methyl ketone enolate was chosen to
simplify the kinetic analysis. 2-Hexanone trichlorosilyl enolate,
16, was selected for its reactivity and ease of preparation,12
Scheme 7.
Discussion
1. Unpromoted Reaction. By employing pseudo-first order
conditions, the molecularity for the unpromoted aldol addition
was determined to be first order in both trichlorosilyl enolate
and aldehyde. This affords the classic bimolecular rate expres-
sion of d[product]/dt ) k[enolate][aldehyde]. A combination
of the molecularity and the stereoregularity for the uncatalyzed
reaction allows the reasonable hypothesis that aldolization occurs
through a pentacoordinate silicon species whose ligands are
arranged in a trigonal bipyramidal manner, Figure 11. For
intramolecular addition to take place, the aldehyde and the enol
oxygen must be in an apical/basal cis configuration. Two such
diastereomeric complexes are possible; in the more stable (ii),
the aldehyde occupies the apical position with a chloride
occupying the remaining one. Thus the two remaining chlorides
and the enol ether would be positioned around the basal plane.
However, this is not the more reactive of the two. For the
components to be activated in the absence of catalyst, a
pseudorotation to place the aldehyde in a basal position (sp2)
and the enol ether in an apical position (3-center-4-electron
hybrid) is necessary (iii).30
SCHEME 7
Fortunately, 13C NMR analysis of the aldol product between
enolate 16 and 5a at 5% conversion afforded excellent results.
To obtain sufficient material for 13C NMR analysis, the aldol
addition was performed on a 10 mmol scale and was stopped
at 5% conversion by using only 0.05 equiv of benzaldehyde
with respect to trichlorosilyl enolate. To corroborate the results,
an additional reaction was performed using 0.05 equiv of enolate
with respect to benzaldehyde. The product from each reaction
was isolated and purified in a manner similar to all preparative-
type reactions. The results from a reaction that was stopped at
5% conversion were compared with those from a standard
reaction taken to full conversion. The integration ratios are
shown in parts a and c of Figure 10 and reveal a depletion in
the 13C isotope at the reactive center of the excess reagent in
both cases. The KIEs29 are shown in parts b and d of Figure 10
and clearly show a KIE at the reactive trichlorosilyl enolate
carbon and the reactive aldehydic center.
FIGURE 11. Pentacoordinate silicon in an uncatalyzed reaction.
The Arrhenius parameters for the unpromoted aldol addition
revealed ∆Gq to be primarily a function of ∆Sq, whereas ∆Hq
was a minor contributor. Such a large negative ∆Sq is typically
indicative of a highly organized transition structure or an
unfavorable preequilibrium.31 Stereochemical studies implicate
a boatlike transition structure as dictated by the correlation of
(E)-enolate f syn-aldolate and (Z)-enolate f anti-aldolate.
Although the closed nature of the transition structure certainly
contributes to the ∆Sq it is unlikely that rotational and vibrational
degrees of freedom could alone afford such large negative
entropies of activation. It is therefore likely that complexation
of the aldehyde to the trichlorosilyl enolate is an independent
and reversible step toward aldolization. Furthermore, a surpris-
ingly small ∆Hq is suggestive of a high level of activation in
this binary complex. This conclusion is understandable consid-
ering the rate at which the trichlorosilyl enolates react with
aldehydes at room temperature.
(28) Singleton, D. A.; Thomas, A. A. J. Am. Chem. Soc. 1995, 117,
9357-2358.
(29) For a detailed discussion involving the determination of 13C KIEs
from NMR integrations and error analysis, see ref 28.
(30) The trigonal bipyramidal silicon structure is created from one
hypervalent (3-center-4-electron) hybrid orbital and three sp2 hybrids (basal
plane). The enolate will experience the greatest activation in the apical
position because of the concentration of electron density at the ligands in
hypervalent bonds. Correspondingly, the aldehyde will experience the
greatest activation in the basal position because the sp2 orbitals have the
most s-character and therefore are the most electronegative sites. For a
description of this bonding scheme, see: Curnow, O. J. J. Chem. Educ.
1998, 75, 910-915.
(31) For additional examples of kinetic studies on aldol additions, see:
(a) Myers, A. G.; Widdowson, K. L. J. Am. Chem. Soc. 1990, 112, 9672-
9674. (b) Myers, A. G.; Widdowson, K. L.; Kukkola, P. J. J. Am. Chem.
Soc. 1992, 114, 2765-2767. For an ab initio MO study, see: (c) Gung, B.
W.; Zhu, Z.; Fouch, R. A. J. Org. Chem. 1995, 60, 2860-2864.
FIGURE 10. (a) 13C isotopic composition of aldol adduct 17 isolated
from a reaction taken to 5% conversion using limited aldehyde. (b)
13
12
13
C KIEs (k C/k C) for a reaction taken to 5% conversion using limited
aldehyde. (c) 13C isotopic composition of aldol adduct 17 isolated from
a reaction taken to 5% conversion using limited enol ether. (d) 13C
12
13
KIEs (k C/k C) for a reaction taken to 5% conversion using limited
enol ether.
3916 J. Org. Chem., Vol. 71, No. 10, 2006

Aldol Additions of Ketone Trichlorosilyl Enolates
2. Promoted Reaction. 2.1. Kinetics. The discovery of
primary 13C KIEs under conditions of limiting reagents leaves
little doubt that the turnover limiting step can only be aldoliza-
tion following assembly of the reactive complex. If complexation
were turnover limiting, then no 13C KIE would be observed in
the natural abundance 13C NMR studies, because no isotopic
partitioning for carbon is expected during this associative step.
The observance of a KIE greater than 1 for k C/k C is consistent
with the manner in which the experiment was established. The
use of excess trichlorosilyl enolate allows for preferential aldol
addition of the 12C enolate, enriching the amount of 12C in the
isolated aldol adduct. The result observed by 13C NMR is a
decrease in relative area at the reactive carbon, and ultimately
a primary KIE value greater than 1.0. Furthermore, it is unlikely
that a primary KIE would be observed should the rate-limiting
step have been the complexation of ligands or even catalyst
turnover.
The activation energies for a reaction using phosphoramide
7a are representative of what has been observed for the previous
reaction sets. Again, the large negative value for ∆Sq implies
an unfavorable preequilibrium, although ∆Sq is notably smaller
for reactions catalyzed by phosphoramide 7a than for the
previous two examples. This is striking considering assembly
of the activated complex involves coordination of four inde-
pendent species. The fact that ∆Sq is smaller may be the effect
of solvation or other interactions with the reaction medium. In
any case, this two-phosphoramide complex is more kinetically
competent than the single phosphoramide complex. This trans-
lates into a lower ∆G‡ and ultimately into faster reaction rates
as compared to the more bulky phosphoramide 7j, Figure 12.
12
13
The use of in-situ monitoring using ReactIR to study
promoted reactions proved to be invaluable compared to
traditional GC analysis. Determining the change in the observed
rate constant (kobs) as a function of catalyst concentration readily
established that promoted aldol additions are first order de-
pendent on catalyst 7j (m ) 1.014). This behavior provided
direct evidence for the presence of a single catalyst molecule
in the transition structure when a bulky phosphoramide is
employed. The fact that the value for the slope is nearly one
also suggests that little if any product yielding side reactions
occur, of which, the most important is the background, or
catalyst free reaction. Certainly, if this were significant, the value
of the logarithmic relationship would most likely be much less
than one to account for reactions proceeding in the absence of
phosphoramide. Additionally, should the 2:1 catalyst/enolate
pathway be competitive, a value between 1 and 2 would be
expected depending upon the kinetic competition between
the 1:1 and 2:1 pathways. A slope close to 1.0 clearly shows
no product-yielding side reactions are competitive, and thus,
all products arise from a single, kinetically competent path-
way.
As in the unpromoted aldol addition, the Arrhenius activation
parameters determined using phosphoramide 7j show that ∆Gq
is again dominated by ∆Sq and even less so by ∆Hq. The nearly
vanishing ∆Hq is indicative of the greater activation in the
phosphoramide-promoted complex compared to the unpromoted
complex. Once again the large negative ∆Sq suggests reversible
preassembly of the activated complex involving enolate, alde-
hyde and phosphoramide prior to aldolization.
FIGURE 12. Free energy profiles for aldol additions using phos-
phoramide 7j and 7a.
2.2. Overview of Chemical Results. The remarkable rate
acceleration observed when reactions are performed in the
presence of phosphoramides is clearly the basis for the high
level of stereoinduction enjoyed with specific phosphoramides.
It is of no surprise that the level of enantioselectivity is
proportional to the activity of the catalyst. It is also noteworthy
to recognize that the direction of stereoinduction is tunable based
upon the structure of the phosphoramide. With regard to
E-trichlorosilyl enolates (such as 4), bulky phosphoramides such
7j and 7k are selective for the syn aldol adduct, Efsyn, through
a one-phosphoramide pathway. Conversely, smaller phosphora-
mides such as 7a and 7m are selective in the anti manifold,
Efanti, via a two phosphoramide pathway. Although any
modifications around the phosphoramide structure will likely
affect a change in reactivity and stereoselectivity, it is predomi-
nantly the substituents on the nitrogens of the diazaphospholidine
skeleton that determine the relative bulk of the catalyst.
Specifically, methyl and ethyl substituents are considered small
whereas aryl substituents such as phenyl and 1-naphthyl may
be considered large on the basis of the previous analysis.
The challenges of determining kinetic parameters for rela-
tively fast reactions at subambient temperatures were greatly
overcome by the use of RINMR spectroscopy. The apparatus
was crucial for providing meaningful, real-time kinetics of
reactions with half-lives on the order of tens of seconds.
Measurement of kobs as a function of the concentration revealed
second order dependence in phosphoramide 7a. Although
previous experimental results have suggested mechanistic dif-
ferences, this provided the first direct evidence for the existence
of a mechanistic dichotomy between catalysts 7a and 7j, namely
that reactions using 7a proceed through a two-phosphoramide
pathway whereas reactions using 7j proceed through a one-
phosphoramide pathway. Additionally, a two phosphoramide
mechanism provides unambiguous substantiation for a ligand
effect as the origin of nonlinear asymmetric induction observed
for 7a.
J. Org. Chem, Vol. 71, No. 10, 2006 3917

Denmark et al.
In addition to the shape of the phosphoramide, the amount
of the catalyst also has a profound effect on stereoselectivity.
Again considering (E)-trichlorosilyl enolates, decreasing the
loading of catalyst generally increases formation of the syn
isomer by promoting the one-phosphoramide pathway. To
enhance the anti manifold, higher loadings of phosphoramides
may be required which activate the two-phosphoramide path-
way. Indeed this was also achieved by slow addition of the
aldehyde component. In this manner, the relative concentration
of the phosphoramide catalyst remained high with respect to
the reactive partners allowing reactions to proceed through the
more selective two-phosphoramide mechanism. Although subtle,
this unexpected loading effect illustrates the mechanistic
dichotomy for various phosphoramides (vide supra).
Perhaps a more striking detail regarding the stereochemical
effects of different phosphoramides lies in the remarkable
difference between 7a (dimethylstilbene-1,2-diamine) and 7j
(diphenylstilbene-1,2-diamine) with regard to the diastereose-
lectivity. As clearly highlighted in the nonlinear effect studies,
7a displays a positive nonlinear effect allowing stereochemical
inductions greater than the actual enantiopurity of the catalyst
used. Subsequent studies have ruled out product inhibition, and
a reservoir effect and indeed kinetic studies have defined a
multiple ligand model as the basis for nonlinear induction.
Conversely, 7j displayed linear induction, proceeding through
a one-phosphoramide pathway as determined by kinetic studies.
The ability of various ionic additives such as tetrabutylam-
monium chloride (Bu4N+Cl-) and tetrabutylammonium triflate
(Bu4N+OTf-) to affect the rates of reaction underscores the
nature of the silicon center during phosphoramide-catalyzed
aldol additions of trichlorosilyl enolates to aldehydes. Attenu-
ation of rate from the addition of Bu4N+Cl- is indicative of a
common ion effect,32 suggesting reactions proceed through a
cationic intermediate created by the dissociation of chloride.33
The addition of Bu4N+OTf- enhances the ionic strength of the
medium allowing stabilization of the ionic intermediates and
thus provides a rate acceleration resulting in an increased
conversion from 44% to 92% after only 8 min.34
protonate the phosphoramide. Interestingly, the strongest acid,
triflic acid, completely suppressed the catalyzed pathway, and
only background reaction was observed.
2.3. Mechanistic Hypothesis. Kinetic analysis has revealed
that a mechanistic dichotomy exists between the reactions
catalyzed by phosphoramides 7a and 7j. Under typical loadings
(5-15 mol %), phosphoramide 7j catalyzes the aldol addition
of trichlorosilyl enolates to aldehydes through the coordination
of a single phosphoramide. Alternatively, 7a acts through a two-
phosphoramide mechanism. Furthermore, results taken from the
salt effect studies would imply that aldol addition proceeds
through a cationic silicon species created by the coordination
of one or two phosphoramides and the subsequent ionization
of chloride. It is likely that coordination of an aldehyde to this
cationic siliconate activates it toward enolate addition, Scheme
8.
SCHEME 8
3. Integration of Results. The most influential component
in the rate of the aldol addition of trichlorosilyl enolates to
aldehydes is the structure of the phosphoramide. A survey of
various substituents at different locations around the phosphor-
amide revealed that large substituents on the nitrogens of the
2,5-diazaphospholidine skeleton greatly attenuated the rate of
aldolization. For the addition of 4 to benzaldehyde, yields
diminished to 31% when using the naphthyl-substituted phos-
phoramide 7k. Even the use of stoichiometric amounts of this
phosphoramide could only enhance yields to 52%. Similar
results were observed for the achiral phosphoramides. Although
for some aldol partners, use of the naphthyl-substituted achiral
catalyst 11c showed only minimal rate attenuation compared
to 11a, use of a branched aliphatic group in 11b afforded
products in only 35% yields. Unlike the nitrogen substituents
on the diazaphospholidine ring, modification of the chiral
backbone of the phosphoramides had only a modest effect on
rates and selectivities of reactions. Although yields were lower
when employing the bulky naphthyl phosphoramides such as
7f and 7g, the overall effect was minimal when compared to
effects from substitution at the phospholidine nitrogens. More-
over, the addition of electron-withdrawing and electron-donating
groups on the phenyl rings did not have a major effect on the
rate of addition. Moving away from the aryl-based backbone
to the cyclohexyldiamine-derived phosphoramide 7e led to the
largest rate decrease. Last, modification of the substituents on
the external nitrogen of the chiral phosphoramides had little
effect on the rate of addition of 4 to benzaldehyde. However, a
pronounced effect was observed when employing enolate 9.
Contracting the piperidinyl substituent to a pyrrolidinyl unit
caused yields to drop significantly from 77% to 41% for
phosphoramides 7a and 7q, respectively. Taken together, these
results reveal that bulky substituents on the nitrogen atoms of
the phospholidine ring reduce the propensity for multiple
phosphoramide coordination forcing reactions through the
slower acting one-phosphoramide pathway.
The addition of Brønsted acids greatly increased the syn
selectivity for the addition of cyclohexanone-derived enolate 4
to benzaldehyde for both achiral and chiral phosphoramides.
The origin of this effect is not certain, but most likely arises
from an enhancement in the reaction flux through the one-
phosphoramide pathway (Efsyn). Because this effect was most
pronounced at low catalyst loading, it is possible that the
Brønsted acids function to remove a fraction of catalyst from
the trichlorosilyl enolate binding equilibrium by protonation of
the oxygen, although their pKa’s are not low enough. Alterna-
tively, the Brønsted acids may interact with the trichlorosilyl
enolates to form chlorosilylcarboxylates and HCl which can
(32) (a) Bateman, L. C.; Hughes, E. D.; Ingold, C. K. J. Chem. Soc.
1940, 974, 1017-1029. (b) Streitwieser, A. SolVolytic Displacement
Reactions; McGraw-Hill: New York, 1962. (c) Thornton, E. SolVolysis
Mechanisms; Ronald Press: New York, 1964.
(33) Cationic silicon species have been proposed as intermediates in other
Lewis base promoted reactions; see refs 3b, 4, 11a, 25, and: (a) Chojnowski,
J.; Cypryk, M.; Michalski, J.; Wozniak, L. J. Organomet. Chem. 1985, 288,
275-282. (b) Corriu, R. J. P.; Dabosi, G.; Martineau, M. J. Organomet.
Chem. 1980, 186, 25-37. (c) Bassindale, A. R.; Lau, J. C.-Y.; Taylor, P.
G. J. Organomet. Chem. 1995, 490, 75-82. (d) Bassindale, A. R.; Lau, J.
C.-Y.; Taylor, P. G. J. Organomet. Chem. 1995, 499, 137-141.
(34) A similar effect has been noted for Bu4N+I- (ref 25). This
observation also rules out the possibility that the retarding effect of
Bu4N+Cl- arises from the role of chloride as a competitive ligand for the
complex.
3918 J. Org. Chem., Vol. 71, No. 10, 2006

Aldol Additions of Ketone Trichlorosilyl Enolates
Another significant influence on the rate of reaction was the
effect of salt additives. As discussed previously, Bu4N+OTf-
accelerated the addition of 4 to 5a in the presence of 7j. This
was interpreted as stabilization of ionic intermediates on the
mechanistic pathway arising from the enhanced ionic strength
of the medium. Furthermore, the decrease in rate from the
addition of Bu4N+Cl- suggests not only the intermediacy of
ionic intermediates but also that cationic siliconates arise from
the ionization of chloride. The common ion effect certainly
implicates chloride as a key participant in the aldol addition of
trichlorosilyl enolates to aldehydes and also explains the low
reactivity of aliphatic aldehydes by chlorosilyl ether formation.8
The use of geometrically defined enolates such as 4 and (Z)-9
allows for additional mechanistic insights based on the resulting
stereochemical outcomes. The stereochemical correlation shows
that the addition proceeds through a closed transition structure
centered around a siliconium ion. This stems from the observa-
tion that a switch in syn-anti manifolds occurs when employing
either 4 or (Z)-9 in aldol additions using similar phosphoramides.
The switch in diastereoselectivity can best be explained from
configurational correlation via aldol additions through closed
transition structures.35 In the case of enolate 4, this would imply
addition through a boatlike transition structure to give syn aldol
products in the presence of catalysts such as 7j (diphenylstil-
bene-1,2-diamine). Likewise, anti products would arise from
additions through a chairlike transition structure from catalysts
similar to 7a (dimethylstilbene-1,2-diamine).
Together with kinetic studies, the stereochemical information
from the addition of geometrically defined enolates allows a
clearer mechanistic picture to emerge. For phosphoramide 7j,
kinetic and nonlinear effect studies revealed that a single
phosphoramide is present in the transition structure for the
stereochemical-determining step. Because this phosphoramide
affords syn products from E-enolates, it follows that 7j acts in
a 1:1 manner with trichlorosilyl enolates to afford aldol products
through a boatlike transition structure. As is the case for other
syn-selective phosphoramides such as 7k and 7p, steric bulk
proximal to the coordinating oxygen limits complexation to a
1:1 mode with aldolization proceeding through a boatlike
transition structure.11a
amounts of the anti product. Moreover, the observation that slow
addition of the aldehyde leads to enhanced anti diastereoselec-
tivity from the reaction of 4 with 5a in the presence of 7a
demonstrates that by maintaining a relatively high concentration
of catalyst with respect to available substrates, aldol addition
proceeds predominantly by a 2:1 pathway, through a chairlike
transition structure.
The preference for the 1:1 or 2:1 (phosphoramide/silicon)
complexes to adopt either boatlike or chairlike transition
structures is likely controlled by the nature of the two different
silicon complexes. As in the unpromoted pathway, the 1:1
manifold involves five ligands positioned in a trigonal bipyra-
midal manner around the now cationic silicon center (Figure
13). As before, the components must be proximal for reaction
to occur and it is likely that in the more stable complexes, the
aldehyde occupies one of the apical positions and either the
phosphoramide or one of the chlorides in the other. On the basis
of X-ray crystallographic studies of five-coordinate silicon36 and
tin complexes,11a,37 we suspect that the phosphoramide occupies
the apical position, leaving the enol ether and the two chlorides
to orient along the basal plane. This positions the enolate and
phosphoramide in relatively close proximity to the aldehyde,
iv. A more reactive complex would arise from a Berry pseudo-
rotation that interchanges the enolate and aldehyde moieties,
v.30 In either of these complexes, the preference for the boat
geometry is rooted in the small bond angle at silicon (90°) and
the placement of the hydrogens on both the aldehyde carbonyl
group and the E-enolate toward the chlorine atom. In any
chairlike arrangement, either the aldehyde residue or the enolate
spectator group will experience unfavorable steric interactions.
For the less bulky phosphoramide 7a, kinetic and nonlinear
effect studies revealed the aldol additions were second order in
catalyst, clearly implicating a transition structure containing two
phosphoramides for the stereochemistry determining step in the
aldol addition. Additionally, the stereochemical consequences
would dictate that aldol addition through this 2:1 (phosphor-
amide/silicon) pathway proceeds via a chairlike structure to
afford anti products from enolate 4. Indeed, unlike the more
sterically congested catalysts, phosphoramides bearing small
substituents on the internal nitrogens are able to achieve 2:1
complexation.11a This divergence in mechanism is further
highlighted by the remarkable loading effect observed for the
diastereoselectivity of aldol additions in the presence 11c. At
low catalyst loadings where 1:1 complexation is most likely,
the reactions are highly syn selective with enolate 4 and catalyst
11c. Increasing the amount of phosphoramide up to 2.0 equiv
allows the 2:1 pathway to become competitive, affording greater
FIGURE 13. Isomeric 1:1 cationic complexes.
In the 2:1 pathway, the cationic siliconate must now adopt
an octahedral geometry to accommodate the additional phos-
phoramide ligand. Although multiple isomeric complexes are
possible, the aldehyde and enolate must be bound in a
cis-manner for reaction to take place (Figure 14). This factor
reduces the number of geometrically isomeric complexes to
four: three with cis-configured phosphoramides (vi-viii) and
one with trans-configured groups (ix).38 Although we are unable
to unequivocally identify which or how many of these com-
plexes contribute to the overall reaction flux, several factors
allow qualitative conclusions. First, given the high enantiose-
lectivity observed with 7a it is unlikely that multiple complexes
(36) Reviews on hypercoordinate silicon compounds: (a) Kost, D.;
Kalikhman, I. In The Chemistry of Organic Silicon Compounds; Rappoport,
Z., Apeloig, Y., Eds.; Wiley: Chichester, 1998; Vol. 2, Part 2. (b) Holmes,
R. R. Chem. ReV. 1996, 96, 927-950. (c) Chuit, C.; Corriu, R. J. P.; Reye,
C.; Young, J. C. Chem. ReV. 1993, 93, 1371-1448. (d) Tandura, S. N.;
Voronkov, M. G.; Alekseev, N. V. Top. Curr. Chem. 1986, 131, 99-189.
(37) Aslanov, L. A.; Attiya, V. M.; Ionov, V. M.; Permin, A. B.;
Petrosyan, V. S. Zh. Strukt. Khim. 1977, 18, 1113-1118.
(35) (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, Chapter 1. (b) Braun, M. In StereoselectiVe Synthesis,
Methods of Organic Chemistry (Houben-Weyl), E21 ed.; Helmchen, G.,
Hoffman, R., Mulzer, J., Schaumann, E., Eds.; Thieme: Stuttgart, 1996;
Vol. 3; pp 1603-1612.
(38) Complexes vi and vii also exist as two limiting diastereomeric chair
complexes, but will not be considered at this level of analysis.
J. Org. Chem, Vol. 71, No. 10, 2006 3919

Denmark et al.
contribute similarly. Moreover, recent computational analysis
of the transition structures for a trichlorosilyl ketene acetal
reacting with a ketone under catalysis by a bis-pyridine N-oxide
suggests that complexes that orient the carbonyl oxygen trans
to a chlorine atom are not stable relative to those with trans
carbonyl-ligand geometries.39 Thus, complexes vi and ix are
disregarded.40 Of the two remaining complexes, vii is expected
to be more reactive because the enolate participates in 3-center-
4-electron bonding with the chlorine rendering it more nucleo-
philic, whereas the aldehyde participates in bonding via an sp
hybrid which activates by electron withdrawal.
Unlike electron-withdrawing groups, electron-donating groups
such as the p-methoxy substituents in 7c had little effect on the
stereochemical course of reaction. Both the diastereo- and
enantioselectivities remained high as compared to the parent
phosphoramide. Surprisingly, the use 3,5-dimethylphenylstil-
benediamine as the chiral backbone in 7d, afforded aldol
products with attenuated stereoselectivity. The origin of these
somewhat subtle effects is unclear and must await a detailed
computational analysis of the reaction pathways.
As with modifications to the chiral backbone, substitutions
at the external nitrogen greatly effect stereoselectivities. Indeed,
only the small N,N′-dimethyl substituents (7m) and the piper-
idinyl group (7a) afforded high enantioselectivities for reactions
of 4 to benzaldehyde. Introducing larger substituents such as
isopropyl (7p), n-propyl (7n), or phenyl groups (7q) greatly
diminished enantioselectivities. In fact, these modifications led
to an inversion in diastereoselectivity to the syn manifold,
suggesting that reaction now proceeds through a 1:1 pathway.
Even the removal of a single methylene unit from the piperidinyl
ring to afford a pyrrolidinyl-substituted phosphoramide (7q)
resulted in attenuated stereoselectivities. Although the individual
effects of substitutions around the phosphoramide structure are
not always predictable, it is clear that a delicate balance is
necessary to achieve high stereoselectivities.
Taken together, the chemical and kinetic (and spectroscopic)
results provide a unified mechanism to be formulated for the
phosphoramide-catalyzed aldol additions of trichlorosilyl eno-
lates, Figure 15. Although the kinetic data clearly indicate
aldolization to be turnover limiting the experimental results
cannot adequately predict the order of events prior to aldol
addition. Initial complexation between enolate 4 and a phos-
phoramide can generate the activated siliconium intermediate
x.41 With bulky phosphoramides such as 7j or at low catalyst
loading, the intermediate x directly binds the aldehyde substrate
and then reacts via a cationic trigonal bipyramidal complex (e.g.,
v) in a boat conformation to afford the syn aldol product.
Although the enantioselectivities are only modest, the major
enantiomer consistently arises from attach on the Si face of the
enolate when using the (S,S)-7j catalyst. With smaller phos-
phoramides, such as 7a, or at higher catalyst loadings, x can
bind a second molecule of phosphoramide to form a new
siliconium intermediate xi. This species retains sufficient
electrophilicity to bind the substrate aldehyde and then react
via a cationic octahedral complex (e.g., vii) in a chair conforma-
tion to afford the anti aldol product. In this manifold, the
enantioselectivities are much higher, but again the major
enantiomer arises from attack on the Si face of the enolate with
the (S,S)-7a catalyst. Thus, the dramatic change in diastereo-
selectivity results from a change in the preference for reaction
via a chair versus a boat transition structure which in turn arises
FIGURE 14. Isomeric 2:1 cationic complexes.
Although both the 1:1 and 2:1 manifolds afford aldol products
with high diastereoselectivities, higher enantioselectivities are
generally observed through the 2:1 complex. The improved
enantioselectivities may in part arise from the increased reactiv-
ity of the 2:1 pathway, causing the background reaction to be
less competitive. The enhanced reactivity of the 2:1 complex
over the 1:1 complex may stem from an increased activation
of the aldehyde component in the octahedral complex compared
to the trigonal bipyramidal complex (tbp). The increased
s-character in the aldehyde-silicon bond of the octahedral
complex (sp hybrid) versus the sp2 hybrid in the tbp complex
can lead to greater electrophilic activation. Additionally, the 2:1
or 6-coordinate complex is better able to impart stereochemical
information by creating a more dissymmetric space around the
reactive partners compared to a 1:1 complex.
The aryl groups in the stilbene-1,2-diamine phosphoramide
backbone afforded the highest enantioselectivities (Table 5). By
switching to an aliphatic backbone such as cyclohexane-1,2-
diamine, 7e, enantioselectivities deteriorated. Although elec-
tronic variations imparted little effect on rates of aldol additions,
their effect on stereoselectivity is more pronounced. Electron-
withdrawing groups such as the p-trifluoromethyl substituents
in 7b had a detrimental effect on the diastereoselectivity for
the addition of 4 to benzaldehyde. In addition, the enantiose-
lectivity was attenuated as well, though not to the same degree.
(41) We cannot exclude the existence of a preassociation between
trichlorosilyl enolate and the aldehyde. Considering the spontaneity of the
uncatalyzed reaction at ambient temperatures, this association certainly takes
place albeit with greatly diminished consequences at -78 °C. Moreover,
as illustrated in Figure 12, we cannot rule out association of a second ligand
prior to ionization of the chloride. Species x is a highly reactive tetravalent
siliconium ion and may not be formed without a second ligand. In the case
of the one-phosphoramide pathway, ionization would occur after aldehyde
binding but prior to aldolization and would be needed for activation of the
aldehyde through a cationic species. In the two-phosphoramide pathway,
the ionization would occur after the binding of the second phosphoramide,
also shown in Figure 15. In both of these cases, the silicon atom would
always be Lewis-satisfied (i.e., in [EL2]+ the silicon has a half share of 9
electrons). It never has to become a very high energy cationic 7 electron
species and remains hypervalent and formally electron rich.
(39) The octahedral silicon structure is created from two hypervalent
(3-center-4-electron) hybrid orbitals and one sp hybrid. For a description
of this bonding scheme, see refs 9c and 30.
(40) The high tendency of group 14 complexes to be cis configured also
rules out complex ix. Ruzicka, S. J.; Merbach, A. E. Inorg. Chim. Acta
1976, 20, 221-229.
3920 J. Org. Chem., Vol. 71, No. 10, 2006

Aldol Additions of Ketone Trichlorosilyl Enolates
FIGURE 15. Grand unified mechanistic scheme for phosphoramide-catalyzed aldolization.
from a change in the structure of the siliconium complex, itself
a consequence of the size of the phosphoramide ligand. The
origin of the preference for a chairlike transition structure in
the octahedral complexes remains obscure and will require
computational analysis.
Experimental Section
General Experimental Procedures. See the Supporting Infor-
mation.
General Procedure for Catalyzed Aldol Additions of Trichlo-
rosilyl Enolates 4 and (Z)-9 with Fast Addition of Aldehyde:
(-)-(2R,1′S)-2-(Hydroxyphenylmethyl)cyclohexanone (anti-6a).
Catalyst (S,S)-7a (7.3 mg, 0.02 mmol, 0.1 equiv) was dissolved in
CH2Cl2 (2 mL) and was cooled to -78 °C. Trichlorosilyl enolate
4b (40 µL, 0.22 mmol, 1.1 equiv) was added dropwise over 1 min.
Benzaldehyde (20.3 µL, 1.0 mmol) was then added neat, over 20
s. The reaction mixture was stirred at -78 °C for 2 h, satd aq
NaHCO3 solution (5 mL) was added quickly, and the mixture was
allowed to warm to rt. The phases were separated, and the aqueous
phase was extracted with CH2Cl2 (3 × 10 mL). The organic phases
were combined, dried over Na2SO4, filtered, and concentrated in
vacuo. The syn/anti ratio was determined by 1H NMR (400 MHz)
analysis to be 1/28. The crude material was purified by column
chromatography (SiO2, hexane/EtOAc, 6/1) to give 1.3 mg of (-)-
syn-6a as an oil and 35.3 mg (90% total) of (-)-anti-6a as a clear
oil. Analytical data for (-)-anti-6a: mp 41-42 °C; 1H NMR (400
MHz, CDCl3) 7.36-7.26 (m, 5 H, Ph), 4.78 (dd, J ) 9.2 and 2.4,
1 H, PhCHOH), 3.98 (d, J ) 2.4, 1 H, OH), 2.64-2.58 (m, 1 H,
C(2)Hax), 2.50-2.44 (m, 1 H, C(6)Heq), 2.39-2.31 (m, 1 H, C(6)-
Hax), 2.11-2.03 (m, 1 H, C(5)Heq), 1.80-1.74 (m, 1 H, C(4)Heq),
1.71-1.47 (m, 3 H, C(5)Hax, C(3)Heq, C(4)Hax), 1.34-1.23 (m, 1
H, C(3)Hax); 13C NMR (100.6 MHz, CDCl3) 215.53 (CdO), 140.84
(ipso-Ph), 128.30 (m-Ph), 127.81 (p-Ph), 126.95 (o-Ph), 74.64
(CHOH), 57.35 (C(2)), 42.60 (C(6)), 30.76 (C(3)), 27.74 (C(5)),
24.63 (C(4)); IR (CHCl3): 3536 (m), 3066 (m), 3033 (m), 2945
(s), 2904 (m), 2868 (m), 1697 (s), 1497 (m), 1450 (s), 1426 (m),
1400 (m), 1322 (m), 1312 (m), 1297(m), 1226 (s), 1201(s), 1191
(m), 1130 (s), 1101 (m), 1064 (m), 1041 (s), 1029 (m), 1016 (m),
846 (m), 777 (m), 731 (m), 724 (m), 707 (m); MS (EI, 70 eV) 204
(M+, 6), 186 (M+ - H2O, 21), 106 (M+ - C6H10O, 40), 98 (M+
- C7H6O, 100), 70 (48), 55 (33); TLC Rf 0.24 (hexane/EtOAc,
Conclusions and Outlook
We have developed a mechanistic rationale to explain the
origin of rate enhancements and mode of stereoselectivity for
the phosphoramide-catalyzed aldol additions of trichlorosilyl
enolates to aldehydes. Structure-activity-selectivity relationships
were developed for a variety of phosphoramides detailing the
propensity for bulky phosphoramides to participate via a 1:1
phosphoramide/enolate pathway through boatlike transition
structures centered around a 5-coordinate cationic siliconate.
Likewise, smaller phosphoramides are able to engage the
trichlorosilyl enolate in a 2:1 manner, reacting through 6-co-
ordinate, chairlike transition structure. The mechanistic inter-
pretation fully explains the remarkable loading effects and
nonlinear effects observed.
Clearly, the operation of dual mechanistic pathways that are
dependent on catalyst size and concentration present significant
preparative problems. The less selective 1:1 pathway will
intervene as catalyst loading drops (to improve turnover) or as
catalyst size increases (to improve enantioselectivity). The
obvious solution is to covalently tether two catalyst molecules
and study the cooperativity of the two subunits. This strategy
has yielded excellent results for enantioselective allylation3b,11b
and for reactions promoted by silicon tetrachloride.25 Initial
studies in this direction for aldolization are promising, and
further refinements and improvements will be reported in due
course.42
3/1); [R]24 -24.2 (c ) 1.03, CHCl3); HPLC tR (2R,1′S)-6a 16.2
D
min (4.6%); tR (2S,1′R)-6a 19.6 min (95.4%) (Chiracel OJ, 90/10
hexane/i-PrOH, 0.5 mL min-1); HPLC tR (2S,1′S)-6a 16.8 min
(40.5%); tR (2R,1′R)-6a 22.3 min (59.5%) (Chiracel OJ, 90/10
(42) We have demonstrated a cooperative effect on the stereoselectivity
of aldol additions with acyclic trichlrosilyl enolated using tethered phos-
phoramides: see ref 6b.
J. Org. Chem, Vol. 71, No. 10, 2006 3921

Denmark et al.
hexane/i-PrOH, 0.5 mL min-1). Anal. Calcd for C13H16O2
(204.27): C, 76.44; H, 7.90. Found: C, 76.45; H, 7.80.
min (59.0%) (Chiralcel OJ, 150 bar, 40 °C, 6% CH3OH in CO2,
2.5 mL min-1). Analytical Data for (-)-anti-6a: SFC tR (2S,1′R)-
6a 2.23 min (4.5%); tR (2R,1′S)-6a 2.54 min (95.5%) (Chiralcel
OJ, 150 bar, 40 °C, 6% CH3OH in CO2, 2.5 mL min-1).
General Procedure for Catalyzed Aldol Additions of 4 with
Slow Addition of Aldehyde: (-)-(2R,1′S)-2-(Hydroxyphenyl-
methyl)cyclohexanone (anti-6a). Catalyst (S,S)-7a (37.6 mg, 0.1
mmol, 0.1 equiv) was dried under vacuum (0.05 mmHg) for 12 at
rt, CH2Cl2 (5 mL) was added, and the solution was cooled to -75
°C (internal). Trichlorosilyl enolate 4 (200 µL, 1.1 mmol, 1.1 equiv)
was added dropwise over 2 min. A solution of benzaldehyde (102.0
µL, 1.0 mmol) in CH2Cl2 (5 mL) was then added to the first
solution, dropwise, via cannula over 45 min During the addition,
the temperature remained at -75 °C. The reaction mixture was
stirred at -75 °C for 30 min, it was quickly poured into cold (0
°C) satd aq NaHCO3 solution (10 mL), and the slurry was stirred
for 15 min. The two-phase mixture was filtered through Celite, the
phases were separated, and the aqueous phase was extracted with
CH2Cl2 (3 × 50 mL). The organic phases were combined, dried
over Na2SO4, filtered, and concentrated. The syn/anti ratio was
determined by 1H NMR (500 MHz) analysis to be 1/>50. The crude
material was purified by column chromatography (SiO2, hexane/
EtOAc, 6/1) to give 2.4 mg of (-)-syn-6a as an oil and 189.8 mg
(94% total) of (-)-anti-6a as a clear oil. Analytical Data for (-)-
syn-6a: SFC tR (2S,1′S)-6a 2.62 min (41.0%); tR (2R,1′R)-6a 3.26
Acknowledgment. We are grateful to the National Science
Foundation for generous financial support (CHE 0105205 and
0414440). S.M.P. thanks the Eastman-Kodak Co. for a graduate
fellowship. X.S. thanks the University of Illinois for a graduate
fellowship. Y.N. thanks the Ministry of Science and Education
of Japan for a postdoctoral fellowship. We thank Mr. Bruce
Williams for assistance in the design and construction of the
RINMR apparatus and Dr. Martin Eastgate for helpful discus-
sions.
Supporting Information Available: Detailed procedures for
the preparation and characterization of phosphoramide catalysts,
aldolization experiments, nonlinear effect studies, and all kinetic
(GC, ReactIR, and RINMR) experiments. This material is available
free of charge via the Internet at http://pubs.acs.org.
JO060243V
3922 J. Org. Chem., Vol. 71, No. 10, 2006