Quantum chemical calculations on the Peterson olefination with α-silyl ester enolates

Article abstract of DOI:10.1021/jo0262107

The reaction of stabilized Peterson reagents (α-silyl ester enolates) with ketones has been studied theoretically and experimentally. Enolate geometry was studied by trapping experiments and NMR spectroscopy and was found to differ markedly with the nature of the base (LiHMDS vs LDA vs KHMDS). The chelating effect of the lithium counterion was found to be critical for the reaction. For the two ketones studied, the combined weight of experimental and computational data assigns geometrical selectivity to the initial addition transition state, though in general there appears to be a fine balance between three possible choices for the rate-determining step.

Full text of DOI:10.1021/jo0262107

Qu a n tu m Ch em ica l Ca lcu la tion s on th e P eter son Olefin a tion w ith
r-Silyl Ester En ola tes
Malcolm B. Gillies, J anne E. Tønder, David Tanner, and Per-Ola Norrby*
Department of Chemistry, Organic Chemistry, Technical University of Denmark, Building 201,
Kemitorvet, DK-2800 Kgs. Lyngby, Denmark
pon@kemi.dtu.dk
Received J uly 18, 2002
The reaction of stabilized Peterson reagents (R-silyl ester enolates) with ketones has been studied
theoretically and experimentally. Enolate geometry was studied by trapping experiments and NMR
spectroscopy and was found to differ markedly with the nature of the base (LiHMDS vs LDA vs
KHMDS). The chelating effect of the lithium counterion was found to be critical for the reaction.
For the two ketones studied, the combined weight of experimental and computational data assigns
geometrical selectivity to the initial addition transition state, though in general there appears to
be a fine balance between three possible choices for the rate-determining step.
SCHEME 1. Olefin a tion Rea ction s
In tr od u ction
Olefination reactions are important tools in organic
synthesis, and since the pioneering work by Wittig,1 a
variety of reagents have been found for the conversion
of carbonyl compounds to alkenes. In addition to phos-
phorus species,2 there are important and widely used
reagents based on sulfur (J ulia reagents3) and silicon
(Peterson reagents4) (Scheme 1). Factors influencing the
selection of the appropriate olefination reagent5 for a
given application include chemo- and diastereoselectivity,
reactivity, and accessibility. The Peterson reagents offer
an interesting profile: their reactivity is high enough to
allow reaction with both ketones and aldehydes; the
reagents are frequently more Z-selective than the corre-
sponding stabilized phosphorus reagents;6 the silicon
reagents are available via several synthetic routes.
The present investigation was prompted by earlier
work on the application of olefination reactions in asym-
metric synthesis.7 In particular, the most efficient non-
SCHEME 2. Dia ster eocon ver gen t Syn th esis
Em p loyin g Asym m etr ic HWE Rea gen ts
enzymatic diastereoconvergent protocol to date could be
realized by combining the enantiodivergent8 asymmetric
Horner-Wadsworth-Emmons (HWE) reaction9 with a
stereospecific (E)-selective Pd-catalyzed allylic alkylation
(Scheme 2), converting both enantiomers of a racemic
aldehyde to one isomer of a γ-substituted R,â-unsaturated
ester in high yield and diastereomeric purity without the
need for separation of diastereomeric intermediates.10
In an attempt to widen the scope of the diastereocon-
vergent protocol, we have initiated an investigation of
the closest silicon analogue of the HWE reaction, addition
of R-silyl ester enolates (stabilized Peterson reagents) to
carbonyl compounds. Here we report the first results of
this study, a computational and experimental investiga-
* To whom correspondence should be addressed. Fax: (+45)
45933968.
(1) (a) Wittig, G.; Geissler, G. Liebigs Ann. Chem. 1953, 580, 44-
57. (b) Wittig, G.; Scho¨llkopf, U. Chem. Ber. 1954, 87, 1318-1330.
(2) Maryanoff, B. E.; Reitz, A. B. Chem. Rev. 1989, 89, 863-927.
(3) (a) J ulia, M.; Paris, J .-M. Tetrahedron Lett. 1973, 14, 4833-4836.
For
a review, see: (b) Kocienski, P. In Comprehensive Organic
Synthesis; Trost, B. M., Fleming, I., Eds.; Pergamon Press: Oxford,
1991; Vol. 6, pp 987-1000. Recent development: (c) Baudin, J . B.;
Hareau, G.; J ulia, S. A.; Lorne, R.; Ruel, O. Bull. Soc. Chim. Fr. 1993,
130, 856-878.
(4) (a) Peterson, D. J . J . Org. Chem. 1968, 33, 780-784. For reviews,
see: (b) Ager, D. J . Org. React. 1990, 38, 1-223. (c) Barrett, A. G. M.;
Hill, J . M.; Wallace, E. M.; Flygare, J . A. Synlett 1991, 764-770. (d)
van Staden, L. F.; Gravestock, D.; Ager, D. J . Chem. Soc. Rev. 2002,
31, 195-200.
(5) For a comparison of olefination methods, see: Kelly, S. E. In
Comprehensive Organic Synthesis; Trost, B. M., Fleming, I., Eds.;
Pergamon Press: Oxford, 1991; Vol. 1, pp 730-817.
(6) (a) Larson, G. L.; Prieto, J . A.; Herna´ndez, A. Tetrahedron Lett.
1981, 22, 1575-1578. (b) Kojima, S.; Inai, H.; Hidaka, T.; Ohkata, K.
Chem. Commun. 2000, 1795-1796.
(8) Faber, K. Chem. Eur. J . 2001, 7, 5004-5010.
(9) Pedersen, T. M.; J ensen, J . F.; Humble, R. E.; Rein, T.; Tanner,
D.; Bodmann, K.; Reiser, O. Org. Lett. 2000, 2, 535-538.
(10) Pedersen, T. M.; Hansen, E. L.; Kane, J .; Rein, T.; Helquist,
P.; Norrby, P.-O.; Tanner, D. J . Am. Chem. Soc. 2001, 123, 9738-9742.
(7) (a) Rein, T.; Pedersen, T. M. Synthesis 2002, 579-594. (b) Rein,
T.; Reiser, O. Acta Chem. Scand. 1996, 50, 369-379.
10.1021/jo0262107 CCC: $22.00 © 2002 American Chemical Society
Published on Web 09/18/2002
7378
J . Org. Chem. 2002, 67, 7378-7388

Peterson Olefination with R-Silyl Ester Enolates
SCHEME 3. Mod el P eter son Rea ction
SCHEME 4. P r op osed Mech a n ism for th e
P eter son Rea ction (cou n ter ion om itted ). P r e- a n d
P ost-Com p lexes (3a /3b a n d 6) Ar e Not Sh ow n
tion of the reaction path for a small model system, viz.
the reaction between methyl trimethylsilyl acetate and
acetone (Scheme 3). Characterization of the rate- and
selectivity-determining steps and the influence of the
enolate counterion have been of particular interest. We
have also compared the results with the previously
published study of the closely related HWE reaction.11,12
The goal of the current work was to increase our
understanding of the selectivity-determining step(s) in
the synthetically very useful Peterson reaction, for which
there is a surprising lack of theoretical studies. The
Wittig reaction and other phosphorus-based olefinations
have been subjected to several theoretical investiga-
tions,11-13 but as far as we are aware, for the Peterson
reaction previous computational work has been limited
to CNDO calculations.14 In future experimental work, an
appropriate chiral auxiliary will be used, and chiral,
racemic ketones will be investigated as substrates, in
close analogy to previously published HWE develop-
ments.7,9-12,15 To the best of our knowledge, there is but
a single example in the literature of a chiral ester being
used as an auxiliary in an asymmetric Peterson olefina-
tion reaction.16 In preparation for future studies of the
asymmetric version of this reaction, we have here inves-
tigated the reaction between the model Peterson reagent
and a simple, chiral ketone, 2-methyl cyclohexanone.
acetone as starting material 1. Single point energies of
selected stationary points were calculated with various
sized basis sets (6-31G*, 6-311+G**, 6-311+G(2df,2p))
at the MP2 correlated level. Geometry optimizations at
the B3LYP/6-31G* level were also carried out for transi-
tion states 9a ‡, 9b‡, 4‡, and 10‡ of the reaction with
2-methylcyclohexanone as starting material 1. All sta-
tionary points on the acetone reaction PES were opti-
mized using the polarizable continuum Poisson-Boltz-
mann solvation model available in J aguar,19 while single
point solvation energies were calculated for the remain-
ing gas-phase structures. We also determined gas-phase
stationary points and single-point solvation energies for
the acetone system with explicit coordination of two THF
molecules to lithium for 9a ‡, 9b‡, and 10‡. Cartesian
coordinates for all optimized structures are provided as
Supporting Information.
The pseudospectral J aguar program (version 4.1)20 was
used for all B3LYP calculations, and Gaussian 98 (revi-
sion A.11.3)21 was used for all MP2 calculations. The
solvent parameters were a dielectric coefficient of 7.43
and a probe radius of 252 pm, appropriate for THF; cavity
and surface area terms were not included in the solvation
calculations. The solution-phase calculations used set-
tings for “ultra-fine” SCF accuracy and “fine” DFT grids
Com p u ta tion a l Meth od s
Scheme 4 shows the proposed mechanism for the
Peterson reaction with all relevant intermediates, based
on the closely related HWE reaction.11 The enolate
starting material may be one of two diastereomers, cis
and trans,17 resulting in two convergent reaction path-
ways. The geometries of these ground-state structures,
together with the intervening transition states were
optimized using density functional calculations with the
B3LYP hybrid functional18 and the 6-31G* basis set, with
(11) (a) Brandt, P.; Norrby, P.-O.; Martin, I.; Rein, T. J . Org. Chem.
1998, 63, 1280-1289. (b) Rein, T.; Vares, L.; Kawasaki, I.; Pedersen,
T. M.; Norrby, P.-O.; Brandt, P.; Tanner, D. Phosphorus, Sulfur Silicon
1999, 144-146, 169-172.
(12) Norrby, P.-O.; Brandt, P.; Rein, T. J . Org. Chem. 1999, 64,
5845-5852.
(13) For recent examples, see: (a) Lu, W. C.; Wong, N. B.; Zhang,
R. Q. Theor. Chem. Acc. 2002, 107, 206-210. (b) Yamataka, H.; Nagase,
S. J . Am. Chem. Soc. 1998, 120, 7530-7536. (c) Armstrong, D. R.; Barr,
D.; Davidson, M. G.; Hutton, G.; O’Brien, P.; Snaith, R.; Warren, S. J .
Organomet. Chem. 1997, 529, 29-33.
(18) (a) Becke, A. D. J . Chem. Phys. 1993, 98, 5648-5652. (b) Lee,
C.; Yang, W.; Parr, R. G. Phys. Rev. B 1988, 37, 785-789.
(19) Marten, B.; Kim, K.; Cortis, C.; Friesner, R. A.; Murphy, R. B.;
Ringnalda, M. N.; Sitkoff, D.; Honig, B. J . Phys. Chem. 1996, 100,
11775-11788.
(14) Trindle, C.; Hwang, J .-T.; Carey, F. A. J . Org. Chem. 1973, 38,
8, 2664-2669.
(20) http://www.schrodinger.com.
(21) Frisch, M. J .; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.;
Robb, M. A.; Cheeseman, J . R.; Zakrzewski, V. G.; Montgomery, J . A.,
J r.; Stratmann, R. E.; Burant, J . C.; Dapprich, S.; Millam, J . M.;
Daniels, A. D.; Kudin, K. N.; Strain, M. C.; Farkas, O.; Tomasi, J .;
Barone, V.; Cossi, M.; Cammi, R.; Mennucci, B.; Pomelli, C.; Adamo,
C.; Clifford, S.; Ochterski, J .; Petersson, G. A.; Ayala, P. Y.; Cui, Q.;
Morokuma, K.; Salvador, P.; Dannenberg, J . J .; Malick, D. K.; Rabuck,
A. D.; Raghavachari, K.; Foresman, J . B.; Cioslowski, J .; Ortiz, J . V.;
Baboul, A. G.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.;
Komaromi, I.; Gomperts, R.; Martin, R. L.; Fox, D. J .; Keith, T.; Al-
Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Challacombe, M.; Gill,
P. M. W.; J ohnson, B.; Chen, W.; Wong, M. W.; Andres, J . L.; Gonzalez,
C.; Head-Gordon, M.; Replogle, E. S.; Pople, J . A. Gaussian, Inc.,
Pittsburgh, PA, 2001.
(15) For leading references, see: (a) Mendlik, M. T.; Cottard, M.;
Rein, T.; Helquist, P. Tetrahedron Lett. 1997, 38, 6375-6378. (b)
Kreuder, R.; Rein, T.; Reiser, O. Tetrahedron Lett. 1997, 38, 9035-
9038. (c) Tullis, J . S.; Helquist, P.; Rein, T. Phosphorus, Sulfur Silicon
1999, 144-146, 165-168. (d) Vares, L.; Rein, T. Org. Lett. 2000, 2,
2611-2614.
(16) Gais, H.-J .; Schmiedl, G.; Ossenkamp, R. K. L. Liebigs Ann./
Recueil 1997, 2419-2431.
(17) To avoid confusion resulting from counterion dependent priority
in the E/Z-nomenclature for enolates, cis and trans refer to the
relationship between silyl and the enolate oxygen. Thus, the cis enolate
2b corresponds to (E)-enolate with Li+ or no counterion, the (Z)-enolate
with K+, and gives the (Z)-enol ether on silyl trapping.
J . Org. Chem, Vol. 67, No. 21, 2002 7379

Gillies et al.
TABLE 2. Su p er im p osed Geom etr ies (a ll a tom s
exclu d in g Li) for Cor r esp on d in g Ga s- a n d
Solu tion -P h a se Lith iu m -Coor d in a ted Aceton e Rea ction
P a th s
TABLE 1. En er gies of th e Sta tion a r y P oin ts a lon g th e
Rea ction P a th fr om Aceton e
energy relative to lowest pre-reactive state kJ mol-1
Li
no Li
displacement (pm)
∆Ga
∆Eb
∆ESPc
∆Eb
9a ‡
9b‡
4a
4‡
4b
10‡
5
11‡
1 + 2a
1 + 2b
3a
0
19.5
6.0
20.4
40.8
54.5
18.5
33.9
12.8
29.3
44.6
0
11.5
20.2
29.2
52.9
0
4.8
3.3
6.0
rms
maximum
lithium
6
15
12
14
24
84
3
7
21
5
11
16
11
20
32
7
13
31
6
12
33
15
26
21
4.6
2.8
0
48
47.5
38.5
-
38.2
39.9
3b
9a ‡
9b‡
4a
33.4
24.6
TABLE 3. Sin gle P oin t MP 2 En er gies for Selected
Sta tion a r y P oin ts fr om th e Lith iu m -Coor d in a ted
Ga s-P h a se Aceton e Rea ction P a th
-10.1
-23.4
-9.1
-29.1
21.9
-54.4
-43.3
-132.2
-127.8
4‡
4.2
4b
-14.3
33.2
-41.4
-27.8
-122.8
-117.5
energy relative to 3a (kJ mol-1
)
10‡
5
62.7
-19.1
-7.1
-109.5
-140.1
-45.4
-35.8
-97.8
-107.1
basis set
6-31G*
6-311+G**
6-311+G(2df,2p)
3a
9a ‡
10‡
11‡
6
0
0
0
13.6
8.3
3.2
-23.4
-13.2
-31.8
7 + 8
a
Obtained from ∆E by addition of ZPE and ∆Gvib calculated at
b
the corresponding gas-phase geometry. Optimized in solvent,
including contributions from solvation but not from vibration.
TABLE 4. En er gies of Selected Sta tion a r y P oin ts a lon g
th e Rea ction P a th fr om 2-Meth ylcycloh exa n on e
c Energy in solvent, calculated at the gas-phase geometry.
energy (∆Ga) kJ mol-1
(see Supporting Information for J aguar input file ex-
amples). Whenever possible, all solution-phase stationary
points were also optimized in the gas phase and validated
as either minima or saddle points by analysis of the
analytical frequencies. Gas-phase DFT calculations were
performed with “accurate” SCF accuracy and “medium”
grid density.
Transition state optimizations were started from ge-
ometries derived from scans along the expected bond-
making/breaking reaction coordinates, with the analytical
gas-phase Hessians used initially. Transition states in
solvent were confirmed by the presence of a single
negative frequency in the updated Hessian.22 We per-
formed normal-mode analysis of all stationary points on
the gas-phase PES to give zero point energies and
vibrational free energy correction terms at the temper-
ature of the typical reaction conditions, 200 K. Energies
were calculated without scaling of the Hessian.23 For
stationary points that could be located in both solvent
and gas phase, and that showed strong similarity be-
tween the two structures,24 the ZPE and free energy
contributions from the gas-phase calculation were applied
to the corresponding solvated stationary point to yield a
final, composite free energy.
configuration
product
9a ‡
9b‡
4‡
10‡
b
RRR-eq
RRR-ax
RRS-eq
RRS-ax
RSR-eq
RSR-ax
RSS-eq
RSS-ax
(Z)
(Z)
(E)
(E)
(E)
(E)
(Z)
(Z)
27.8
33.6
37.2
24.3
9.1
29.7
4.3
28.7
36.6
32.9
39.3
36.4
33.1
27.3
28.8
27.9
-
51.3
34.6
46.2
37.4
53.8
29.4
56.0
36.5
9.9
37.2
36.9
13.7
29.4
0
b
-
a
Energy in solvent, calculated at the gas-phase geometry, with
Converged geometries for these
transition states could not be found.
b
addition of ZPE and ∆Gvib
.
solution-phase structures, Table 1 includes the corre-
sponding solution phase energies calculated at the gas-
phase geometries, and Table 2 demonstrates the small
structural deviation between gas-phase and solution-
structures for Li-coordinated species. Table 3 contains
the results of MP2 single point energy calculations on a
number of the B3LYP gas-phase stationary point geom-
etries, with various sizes of basis set.
Calculations on the reaction with 2-methylcyclohex-
anone as starting material are summarized in Table 4.
Two new chiral centers are formed in the initial TS in
addition to the center present in 1, necessitating the
investigation of four diastereomeric paths. Furthermore,
two low-energy chair conformations are accessible to the
cyclohexane moiety. We postulate that exchange between
the two chair conformations is feasible, but that crossover
between diastereomeric paths requires reversion to the
starting materials. Thus, we must consider eight paths
with pairwise crossover between conformers. We define
the (R) stereoisomer of the starting material as the
canonical configuration and list the stereocenter formed
in the ketone and enolate fragments afterward.
Resu lts
Energies for all stationary points along the reaction
path with acetone as substrate, with and without coor-
dinating Li+ counterion, can be found in Table 1. Paths
from trans and cis reagent are denoted a and b, respec-
tively.17 Note that at intermediate 4, the paths converge
via rotation TS 4‡. As a validation for the combination of
vibrational contributions from gas-phase geometries with
The reactivities of enolates with two different coun-
terions were tested in reactions with acetone (Scheme 3)
and with cyclohexanone. Under otherwise identical con-
ditions, the enolate derived from LHMDS and methyl
trimethylsilyl acetate (2) gave good yields of unsaturated
ester product 8 (96% in the case of methylcyclohexyli-
deneacetate),25 whereas the KHMDS-derived enolate
(22) Analytical second derivatives are not available for solvated
systems in the J aguar program, and accurate numerical frequency
calculations are computationally intractable for this size of system, so
we did not make this extra check of the transition states in solvent.
(23) Scott, A. P.; Radom, L. J . Phys. Chem. 1996, 100, 16502-16513.
(24) In the current work, we adhere to the following arbitrary
similarity criteria: all-atom overlay rms deviation excluding Li: <15
pm; energy from full optimization in solvent lower than single point
solvation at the gas-phase geometry, but not by more than 20 kJ mol-1
.
7380 J . Org. Chem., Vol. 67, No. 21, 2002

Peterson Olefination with R-Silyl Ester Enolates
showed very low reactivity: no product could be detected
in the reaction with acetone, while in the reaction with
cyclohexanone only a 3% yield could be detected by GC-
MS analysis (the major product, 61%, was instead the
enol ether, 1-trimethylsilyloxycyclohexene).
Discu ssion
Use of Com p osite F r ee En er gies. For all calcula-
tions involving the coordinating lithium cation (i.e.,
overall neutral systems), the geometries determined in
the gas phase and solvent correspond surprisingly well.
Thus, our best estimate of the total free energy in solution
(G) is obtained as a composite, by adding the vibrational
contributions calculated in the gas phase to the energies
of the corresponding stationary points obtained with the
solvation model. For verification, we also calculated the
total free energy surface based on the more conventional
method of adding all contributions calculated as single
points for the gas-phase geometries. The relative energies
on these two surfaces differ by no more than a few kJ
mol-1 (the difference is the same as that between ∆E and
∆ESP in Table 1).
F IGURE 1. Absolute energy difference between single-point
and fully relaxed solution potential energy surfaces (acetone
reaction).
ation is impractical, such as in the case of the 2-meth-
ylcyclohexanone reaction studied here.
For the anionic systems which have been used as a
model, for the case with a weakly coordinating counterion
like potassium11 or lithium-12-crown-4, a proper gas-
phase PES could not be obtained. Most stationary points
on the PES are dependent on some type of stabilization
of localized anionic charge, either by solvent or a coor-
dinating counterion. In the absence of both these stabiliz-
ing factors, the low-energy path along the reaction
coordinate displays a monotonic energy decrease, as has
been observed previously for other ionic reactions in the
gas phase.26 We must therefore neglect the vibrational
contributions for this reaction.22 All energy comparisons
including anionic structures are therefore performed at
the highest accessible level, potential energy in solvent
with added free energy solvent corrections, E (Table 1).
The close correspondence between the two systems
gives additional confidence in the reliability of the
solvated results and also provides the opportunity to
make a prediction of a realistic free energy surface.
Normal mode calculations on the gas-phase system give
corrections for zero point energy, and vibrational en-
thalpy and entropy, based on a harmonic approximation.
These corrections should be reasonable for the corre-
sponding solvated structures.
Accu r a cy of Rela tive F r ee En er gies. The accuracy
of the calculated composite free energies is subject to a
number of qualifications. The adequacy of the chosen
basis set, the treatment of electron correlation, and the
solvent model are matters for careful consideration. In
particular, transition state 10‡, which contains a pen-
tavalent silicon, may be treated inaccurately by relatively
low level calculations. The results in Table 3 showing the
relative energy of transition states 9a ‡ and 10‡ with
larger basis sets and treatment of correlation at the MP2
level indicate that this energy difference is not captured
well at the B3LYP/6-31G* level. The highest level
calculation with a 6-311+G(2df,2p) basis set gave a gas-
phase energy for 9a ‡ 35 kJ mol-1 higher than 10‡,
whereas B3LYP/6-31G* gave a difference of only 2 kJ
mol-1. The resulting ∆∆E can be applied to the difference
in composite free energy values to give a more realistic
picture of the identity of the rate-determining step.
The implicit solvation model does not give a complete
and accurate treatment of the effects of the coordination
of solvent molecules to the solute. This may be an
important omission in the case of the lithium counterion.
A more realistic treatment of solvation would be to
include an explicit solvation shell around Li. Such
calculations are very demanding, but we succeeded in
locating the TS 9‡ in the presence of two explicitly
coordinating THF molecules. Encouragingly, the geom-
etry of the TS is only marginally affected by the explicit
solvation, except for the position of the lithium itself in
the case of 9b‡. In the presence of explicit solvation, the
TS is in the expected six-membered ring chair of a regular
The calculations on the lithium-coordinated system in
the absence of solvent gave stationary point geometries
essentially identical to the solvated ones. All-atom su-
perpositions of the corresponding stationary point struc-
tures have an rms deviation of 15 pm or less, excluding
lithium in all cases (Table 2); the position of the coordi-
nating lithium is slightly more variable. The relative
energies of the stationary points are similar between the
gas phase and solvated calculations, with the exception
of the additional stabilization of intermediate 4 in the
solvated case. The single-point solution-phase energies
for the gas-phase minimum geometries present a picture
that corresponds closely to the PES from the solution-
phase minima. As expected, relaxation in solvent results
in lower energies, the absolute energy differences be-
tween the two solution-phase energy curves being shown
in Figure 1 (for clarity, only trans enolate pathway data
points are shown). The average difference in energies is
7 kJ mol-1, and the maximum is 18 kJ mol-1, with the
two curves close to parallel. These results support the
use of single-point solvation energies where full relax-
(25) Shimoji, K.; Taguchi, H.; Oshima, K.; Yamamoto, H.; Nozaki,
H. J . Am. Chem. Soc. 1974, 96, 1620-1621.
(26) Hagelin, H.; A° kermark, B.; Norrby, P.-O. Chem. Eur. J . 1999,
5, 902-909.
J . Org. Chem, Vol. 67, No. 21, 2002 7381

Gillies et al.
SCHEME 5. En ola te Tr a p p in g Exp er im en ts
ketone oxygen. Only results for the energetically most
favored gauche structures for 9‡ and 4 are reproduced
here.
F IGURE 2. Superimposition of TS 9b‡ structures calculated
with implicit THF solvation and with explicit coordination of
two THF molecules to lithium.
The difference in size of the initial activation barrier
between the coordinated and noncoordinated cases
amounts to 26 (26) kJ mol-1 (trans/cis) relative to the
precomplex when calculated in solution, without vibra-
tional corrections. In addition to the high initial activa-
tion energy, the noncoordinated reaction pathway differs
in the absence of a stationary point for the rotation
between 4a and 4b, and the intermediate dissociates
rather than passing this barrier. The additional stabili-
zation of intermediate 4 by Li+ is evident on comparison
of the calculated PES data, lying 47 (64) kJ mol-1 (4a /
4b) lower relative to the precomplex 3, and 21 (38) kJ
mol-1 lower relative to transition state 9, when Li
coordination is present.
aldol addition. Without the solvent shell, Li+ is attracted
to the electron density in the rest of the molecule, in
particular to the slightly negative silyl-bonded methyls,
giving a boatlike TS. The superimposition of the two
structures can be seen in Figure 2. Since the O-Li-O
moiety has no substituents in this model, there are none
of the usual negative effects of a boat conformation like
eclipsing or diaxial interactions. However, the proximity
of the Li cation to the organic moiety causes an artifac-
tual lowering of the total energy, composed partially of
electrostatic attraction and partially of basis set defi-
ciency effects. We can expect this error to cancel to a large
extent when comparing very similar structures, like
conformations, but to lower the predictivity when com-
paring structurally different complexes where the local
environment around lithium changes.
Role of th e Cou n ter ion . Most published procedures
for Peterson reactions of R-silyl enolates employ lithium
bases,4 and it has previously been observed that potas-
sium enolates are substantially less reactive than the
corresponding lithium enolates.27 We have verified that
this difference in reactivity also exists in our model
system. As mentioned previously, the potassium salt of
methyl trimethylsilyl acetate did not yield any observable
product with acetone and only 3% GC yield with cyclo-
hexanone. Similar results were obtained when the lithium
enolate was reacted in the presence of a slight excess of
12-crown-4 to disrupt the coordinating ability of lithium.28
In contrast, under otherwise identical conditions, the
lithium enolate reacted rapidly with both ketones, giving
a 96% yield in the reaction with cyclohexanone. These
results agree well with the computational study. In line
with previous studies on the HWE reaction,11 the free
enolate anion was used as a model for the system with
noncoordinating counterion. The most important points
on the reaction profiles are depicted in Figure 3. In the
absence of a coordinating metal, the approach of the
ketone to the Peterson reagent can occur with either anti
or one of two gauche relationships between silicon and
The much higher barrier in the absence of lithium
provides a rationalization for the lack of reactivity
observed in reactions between enolates of alkyl trialkyl-
silyl acetates and ketones under conditions lacking
strongly coordinating counterions. Only activated ketones
such as 2,3-epoxycyclohexanone have been reported to
react when poorly coordinating counterions are present,
the lack of reactivity of nonactivated ketones being
ascribed to increased ketone enolization.27 The same
trend was also observed here: the O-silylated enol ether
was the major product from reaction of cyclohexanone
with the potassium salt of enolate 2. Suppression of
ketone enolization and the presence of a strongly coor-
dinating cation thus appear to be vital for efficient
Peterson olefination.29
Con figu r a tion of En ola te. The selectivity in the
Peterson reaction is dependent on the configuration of
the starting enolate, 2. This has sometimes been assumed
to be cis in the literature,16 but our calculations indicate
that the trans form is substantially more stable. To
elucidate this critical point, we performed trapping
experiments to determine the favored form and the
configurational stability of the lithium enolate.
The Li enolates produced under varying conditions and
one example of a K enolate were trapped by O-silylation
using trimethylsilyl chloride (Scheme 5). The products
(27) Strekowski, L.; Visnick, M.; Battiste, M. A. Tetrahedron Lett.
1984, 25, 5603-5606.
(28) Tønder, J . E., unpublished results.
(29) For the use of cerium as counterion to circumvent enolization,
see: J ohnson, C. R.; Tait, B. D. J . Org. Chem. 1987, 52, 281-283. See
also ref. 4d.
7382 J . Org. Chem., Vol. 67, No. 21, 2002

Peterson Olefination with R-Silyl Ester Enolates
F IGURE 3. Reaction profiles with and without lithium counterion (reaction with acetone). Solution phase energies (∆E) relative
to precomplex 3. Only the pathway proceeding from the trans enolate 3a is shown. Gas-phase calculations failed to find any
stationary points for the free anionic system. The absence of both the stabilizing effect of solvent and of a coordinating lithium
gives a markedly different reaction pathway, and it appears that the entire pathway is downhill under these conditions.
some nondeprotonated ester in solution long enough for
the ester and ester enolate to equilibrate in an identity
reaction, giving the thermodynamically favored product
exclusively. This type of equilibration could also be
expected in the reaction mixture with an enolizable
ketone substrate, which could rationalize the observation
that enolates derived from LDA and LiHMDS give
similar product selectivities28 despite the differences in
enolate ratio from these two bases.
F IGURE 4. NOE results for 12a and 12b.
TABLE 5. En ola te-Tr a p p in g Exp er im en t. Resu lts Ar e
Ba sed on 1H NMR
Acet on e Olefin a t ion F r ee E n er gy Su r fa ce w it h
Li+. The lithium-coordinated solution-phase reaction
profile proceeding from acetone, including the vibrational
free energy correction, is shown in Figure 5, along with
the structures of the various stationary points. The
coordination of the ketone and ester oxygens to the
lithium ion largely predetermines the geometry of the
reacting species in this case. The presence of the lithium
ion also has a stabilizing effect on the various anionic
species, particularly the oxyanion intermediate 4 where
the negative charge is largely localized to the oxygen.
For both the gas-phase and the solvated Li-coordinated
systems the calculated difference in energy between
transition states 9b‡ and 10‡ at B3LYP/6-31G* is con-
siderably smaller than the expected uncertainty in rela-
tive energies of 10 kJ mol-1. Addition of the vibrational
correction terms increases the free energy of 10‡ by
approximately 10 kJ mol-1 relative to 9b‡, in line with
expectation for the more conformationally restricted
cyclic 10‡; the transition state 9a ‡ lies lower, 22 kJ mol-1
below 10‡. The rotational barrier between 4a and 4b is
29 kJ mol-1 lower than 10‡. If we apply a correction from
the MP2/6-311G+(2df,2p) results, 10‡ is lowered signifi-
cantly in energy relative to 9a ‡. Including solvation and
vibration terms, this gives a free energy for 9a ‡ 11 kJ
mol-1 above 10‡. Thus, different theoretical levels give
different predictions for the rate-determining step (RDS).
We can estimate from the identified sources of error that
a safe identification of the RDS would require at least
experiment
base
LDA
temperature (°C) 12a :12b yield (%)
1
2
3
4
5
-78
0
75:25
88:12
88
69
LDA
LiHMDS
LiHMDS
KHMDS
-78
100:00
100:00
25:75
>99
0
94
0
40
12 were assigned by 1D NOE experiments (Figure 4). The
ratios of 12a :12b are listed in Table 5.
The trans enolate 2a turned out to be favored in all
cases when lithium was used as counterion, sometimes
being formed exclusively (Table 5). Furthermore, unequal
ratios could be obtained even when the enolate was
formed at 0 °C (entries 2 and 4, Table 5), giving strong
indication that the enolate is configurationally stable
under the trapping conditions at -78 °C.
In agreement with the literature,30 the experiments
with LDA as base show that the trans ratio increases
when the enolate is formed at 0 °C, indicating that 2a is
the thermodynamically favored product, in good agree-
ment with the computational results. The mixture ob-
tained at -78 °C indicates that LDA is kinetically
undiscriminating. On the other hand, LiHMDS gives
exclusively trans enolate 2a independent of reaction
temperature. Interestingly enough, the potassium base
favors the cis enolate 12b. In line with the currently held
view of kinetic deprotonation of nonsilylated esters,31 it
is probable that LiHMDS acts as a slow base, leaving
(31) Ireland, R. E.; Daub, J . P. J . Org. Chem. 1981, 46, 479-485.
For a review see Evans, D. A.; Nelson, J . V.; Taber, T. R. Topics
Stereochem. 1982, 13, 1-115.
(30) Matsuda, I.; Murata, S.; Izumi, Y. J . Org. Chem. 1980, 45, 237-
240.
J . Org. Chem, Vol. 67, No. 21, 2002 7383

Gillies et al.
F IGURE 5. Free energy reaction profile (reaction with acetone).
explicit solvation on lithium, continuum solvation of the
full complex, and correlation treatment beyond MP2 with
at least a triple-ú basis set. Including all these factors in
one calculation is well beyond our current computational
resources. Instead, it is possible to identify the RDS by
a comparison with the very similar HWE reaction, where
the selectivity is also determined in transition states very
similar to 9‡ and 10‡.12 For the HWE, it was shown that
the two transition states were very similar in energy, but
with a very different sensitivity to steric crowding and
solvation. Thus, with sterically demanding reactants, the
ring closure/elimination TS (corresponding to 10‡) was
found to be rate determining, whereas for sterically
unencumbered reactions, the RDS was the addition (9‡).11
It has also been shown that addition of a mixed HWE/
Peterson reagent, which can choose elimination toward
phosphorus or silicon after the common addition step,
exclusively followed the Peterson elimination route for
all silyl groups smaller than triisopropylsilyl (TIPS).32
This clearly shows that the elimination activation barrier
is lower for silicon than for phosphorus. In conjunction
with what is known about the relative energy of the
addition and elimination steps for the HWE reaction, the
result gives a strong indication that 9‡ is higher in energy
than 10‡, in agreement with the MP2 but not with the
B3LYP/6-31G* energies.
P r ecom p lex. The solvated potential energy E of the
complex 3 between the anionic methyl TMS-acetate,
lithium, and acetone is substantially lower than that of
the separated starting materials 1 + 2, one of the major
reasons why the lithium ion lowers the barrier to
addition. However, on the free energy surface, the pre-
complex is slightly higher in energy than the reactants,
due mainly to the entropic penalty.
Tr a n sition Sta te 9. Transition state 9a ‡, which leads
to formation of a new C-C bond, is shown in Figure 6.
We note that the addition is strongly reminiscent of a
classical aldol addition. It has been observed that the
isolated reagent in vacuo prefers a four-membered ring
F IGURE 6. Transition state 9a ‡. Most hydrogen atoms have
been omitted for clarity. Carbon atoms are shown in black,
silicon in gray, and lithium and hydrogen in white. Oxygen is
stippled. Bond lengths labeled in pm. The remaining structures
follow the same conventions.
coordination of lithium,13c which has led to the proposal
of the so-called “butterfly” transition state for the Peter-
son reaction.32 However, as already noted by Warren and
co-workers for the corresponding Horner-Wittig reac-
tion,13c there is no Li-C coordination in the TS; even in
the gas phase, the addition proceeds via a six-membered
chairlike TS. The transition state is early, with a C-C
distance of 226 (239) pm (9a ‡/9b‡) in the solvated system.
The gas-phase geometries are essentially identical to the
solvated structures. The activation barrier is 10 (13) kJ
mol-1 lower in the presence of solvent. The solvated
transition state is somewhat later in the absence of
lithium, with a C-C distance of 204 (203) pm (9a ‡/9b‡),
and a substantially higher activation barrier as expected
from the Hammond postulate.33 The trend in activation
barriers, increasing in energy when either the coordinat-
ing lithium or the continuum solvation is removed, can
be understood from the Bell-Evans-Polanyi principle.33
(32) Waschbu¨sch, R.; Carran, J .; Savignac, P. Tetrahedron 1996, 52,
14199-14216.
(33) J ensen, F. Introduction to Computational Chemistry; Wiley:
Chichester, 1999.
7384 J . Org. Chem., Vol. 67, No. 21, 2002

Peterson Olefination with R-Silyl Ester Enolates
F IGURE 8. Transition state 10.
tion, TS 4‡ contains no partial bonds and is therefore
expected to be well described by relatively low level of
theory B3LYP. Transition states 9‡ and 10‡, on the other
hand, do contain partial bonds and could therefore be
expected to be lowered in energy relative to 4‡ by
inclusion of larger basis sets and/or higher correlation.
The current results indicate that 4‡ will not influence the
reaction selectivity (as opposed to the related reaction of
sulfur ylids with aldehydes35), but it cannot be excluded,
and a final verdict will have to await higher level
treatment or evaluation based on the selectivity profile.
To this end, we note that the pro-(E) ketone substituent
is eclipsed with the TMS group, with a C-Si distance of
325 pm. Thus, even at the currently calculated energy
level, 4‡ could be expected to function as a filter, prohibit-
ing formation of (E)-product from trans-enolate 1a .
Tr a n sition Sta te 10. Cyclic transition state 10‡
involves the concerted breaking of the Si-C bond to-
gether with the formation of a Si-O bond (Figure 8). The
transition state is relatively late, with the Si-C bond
weak at 227 pm for the solvated geometry. As with the
HWE reaction, the ring geometry is quite puckered,11 and
the Cester-C-C-Cpro-(Z) dihedral makes an angle of 44°.
TS 10‡ includes two partial bonds, partial hypervalency
at Si, variable coordination at Li, and also some close
lithium-methyl interactions. We expect it to be the most
difficult point on the PES to describe computationally.
We can see that the energy of this TS relative to that of
TS 9‡ is very sensitive both to level of theory and to basis
set. At B3LYP/6-31G*, including both solvation and
vibration contributions to the free energy, TS 10‡ seems
to be rate determining. However, TS 10‡ should yield
overall (E)-selectivity due to the close proximity between
the ester moiety and the pro-(Z) substituent, something
that is rarely seen in the Peterson reaction. Furthermore,
the MP2/6-311+G(2df,2p) correction lowers the energy
of TS 10‡ selectively. Thus, considering all sources of error
in the calculations and comparing to experimental re-
sults, we would not expect TS 10‡ to influence the
selectivity of reagents with the simple TMS substituent.
However, the picture is known to be different when
bulkier silyl substituents are employed,32 and like the
related HWE reaction, might also be influenced by the
choice of solvent.11
F IGURE 7. Transition state 4‡.
The reaction leads from a conjugated enolate anion, with
a fair gas-phase stability, to a localized oxyanion. Sta-
bilization of the latter by either coordination to a cation
or by continuum solvation will lower the energy of 4
relative to 3, and thus lower the barrier to reaction from
3 to 4.
Steric interactions in 9‡ are determined by interaction
between the ketone substituents and either the silyl or
the ester moiety. In 9a ‡, both substituents are gauche to
the silyl, but one is also crowded by being gauche to the
ester. The larger substituent can therefore be expected
to prefer the position anti to the ester moiety. Elimination
via 10‡ requires a 180 °C-C bond rotation, which would
eclipse the larger substituent with the ester moiety. We
should therefore expect Z selectivity from a rate-
determining 9a ‡. In 9b‡, both the silyl and the ester are
gauche to one substituent and anti to the other. It is also
possible for the TS to accommodate steric bulk by slight
rotation. It is therefore difficult to make a general
prediction for the selectivity in this TS. We will return
to a specific example later.
Oxya n ion In ter m ed ia te 4. Oxyanion intermediate
4 is computationally stable in both gas and solution phase
for the coordinated system. Without lithium, only the
solvated system was found to be a stationary point, in
line with the previous finding that the HWE oxyanion
intermediate was a nonstationary point at the B3LYP/
6-31G* level.11 The C-C bond is long, 163 (161) pm (4a /
4b) for the solvated Li-coordinated intermediate, similar
to the equivalent weak bond found for the HWE oxyanion
intermediate.11 In a Curtin-Hammett situation,34 the
relative energies of 4a and 4b have no influence on the
observed selectivities. Intermediate 4b is set up for direct
elimination via 10‡, whereas 4a must first convert to 4b
via TS 4‡.
Tor sion a l Tr a n sition Sta te 4. Torsional transition
state 4‡ is a simple torsional barrier between rotamers
4a and 4b (Figure 7), complicated by the coordination of
two oxygens to the lithium. The reaction eigenvector
includes a component of the lithium movement on a very
soft PES. This poses unexpected problems, both in
locating the TS for more substituted systems (vide infra)
and for evaluating the free energy components. In addi-
En ola te In ter m ed ia te 5 a n d Tr a n sition Sta te 11.
Formation of the enolate intermediate 5 is more exother-
mic than the preceding steps along the reaction path, and
(34) Maskill, H. The Physical Basis of Organic Chemistry; Oxford
University Press: Oxford; 1985.
(35) Aggarwal, V. K.; Harvey, J . N.; Richardson, J . J . Am. Chem.
Soc. 2002, 124, 5747-5756.
J . Org. Chem, Vol. 67, No. 21, 2002 7385

Gillies et al.
F IGURE 10. Axial and equatorial additions to cyclohexanone,
9a ‡.
should yield low (if any) selectivity, since one substituent
will be crowded by the silyl, and the other by the ester
moiety. However, TS 9b‡ is unlikely to influence the
selectivity, since the preferred form of the reactant is 2a ,
and equilibration of 2a and 2b can occur via reversible
protonation involving the substrate.
F IGURE 9. Transition state 11‡.
the subsequent barrier to transition state 11‡ is relatively
low. The O-C bond to the TMS group is weak at 152
pm, as previously seen for the corresponding step in the
HWE reaction.11 The lifetime of the intermediate is likely
to be short, making the likelihood of isomerization and
potential loss of steric control due to C-C bond rotation
fairly low. The transition state 11‡ for the final elimina-
tion of TMSO anion is shown in Figure 9. The subsequent
formation of the products is highly exothermic.
Or igin s of Selectivity. The Peterson reaction fre-
quently displays a high Z selectivity, even with moder-
ately encumbered ketones.27 To rationalize this selectiv-
ity, we must find the points on the free energy surface
where the pro-(E) ketone substituent is selectively
crowded. First of all, we can see that the final TS 11‡ is
very product like, with crowding only of the pro-(Z)
substituent. In any case where Z-selectivity is observed,
TS 11‡ cannot have an influence on the selectivity, and
thus elimination through 11‡ must be substantially faster
than the competing bond rotation.
2-Met h ylcycloh exa n on e Olefin a t ion R ea ct ion
P a th w a ys. To elucidate the nature of the rate-determin-
ing step, we have investigated the reaction of reagent 2
with 2-methylcyclohexanone in detail. At -78 °C, this
reaction gives a (Z):(E) ratio of 93:7 (slightly higher than
the ratio of 89:11 found earlier27), corresponding to a
barrier difference of 4.2 kJ mol-1. We expect to be able
to predict relative energies of similar transition states
with high confidence, but as has been already shown, we
cannot access a level of theory that will allow us to
determine the relative energies of different transition
states accurately enough. The enolate reagent has been
shown to be formed in trans configuration when LiHMDS
is used as base. We must therefore consider the sequence
9a ‡ f 4‡ f 10‡. We must also allow for the possibility of
reversion of 4b to cis enolate in situ via 9b‡. If such
reversion is feasible, readdition via other diastereomers
of 9b‡ must also be considered.
Product formation from 2a is achieved through three
additional transition states, 9a ‡, 4‡, and 10‡. For the
parent system, the rate- and selectivity-determining
barrier is 9a ‡, by a margin of 13 kJ mol-1 if the difference
between MP2/6-311G+(2df,2p) single point energies is
added as a correction to the free energies calculated at
the solvated B3LYP/6-31G* level. As was demonstrated
for the HWE reaction,12 when the reaction must pass
through a sequence of transition states with similar
activation energies, each TS will act as a filter, blocking
one or a few isomeric pathways while allowing others to
form product. A high observed selectivity might then be
the result of several transition states acting in concert.
The same type of serial selectivity has also recently been
demonstrated for the Aggarwal catalytic epoxidation of
aldehydes,35 where for the first time a torsional barrier
like 4‡ was found to affect the selectivity. In our case, 4‡
is completely eclipsed and therefore expected to disfavor
bulky groups in the pro-(E) substituent strongly and
selectively due to crowding with the silyl group. In TS
10‡, the strain is somewhat relieved, but the pro-(E)
substituent is still crowded by the silyl group. However,
the pro-(Z) substituent is now in closer proximity to the
ester moiety, decreasing the selectivity of 10‡ compared
to 4‡. In the initial addition, TS 9a ‡, the two substituents
are staggered around the silyl group, but the pro-(E)
substituent experiences additional crowding from the
ester moiety. This could rationalize the earlier observa-
tion that (Z)-selectivity increases with the size of the ester
moiety.36 Thus, all three transition states would be
expected to disfavor (E)-product. In contrast, TS 9b‡
Four diastereomeric pathways are possible, depending
on the face selectivity of both enolate and ketone in the
initial addition, and the 2-methyl substituent may be
axial or equatorial in each. Given that the reagent is
nonchiral in this model, we have investigated only the
(R)-enantiomer of the substrate. The enantiomeric form
of each stationary point, derived from the (S)-enantiomer
of the substrate, must have the same energy and result
in the same diastereomer of the product. The diastere-
omeric transition states and intermediates will be des-
ignated by the configuration of each chiral center (e.g.,
RRR), where the first label corresponds to the 2-position
in 2-methylcyclohexanone (always R), the second is the
configuration of the former carbonyl carbon, and the third
is the enolate R-position. The RRR and RSS diastereo-
mers will lead to (Z) product, whereas RRS and RSR will
lead to (E). Two chair forms with similar energies are
possible, with the methyl group in either axial or equato-
rial position (labeled ax or eq, respectively).
We will first investigate the initial addition, 9a ‡.
Somewhat unexpectedly for attack at a cyclohexanone,
there is a very strong preference for equatorial attack.
The TS is strongly oriented by the lithium, leading to
severe repulsion between the cyclohexanone ring and the
TMS group in any axial attack (Figure 10). We also see
that for equatorial attack the axial 2-positions of the
cyclohexanone are in close proximity to the TMS group.
(36) Larcheveque, M.; Legueut, C.; Debal, A.; Lallemand, J . Y.
Tetrahedron Lett. 1981, 22, 1595-1598.
7386 J . Org. Chem., Vol. 67, No. 21, 2002

Peterson Olefination with R-Silyl Ester Enolates
tioconvergent protocol.10 It is very important for the
correct selectivity that the carbonyl is strongly affected
by the existing chiral center, but that the enolate is not,
allowing for introduction of chiral auxiliaries in the
Peterson reagent.
Finally, as a note of caution, we wish to point out that
although the B3LYP/6-31G* level employed here is very
popular in mechanistic investigations, it is clear from the
present results that, for anionic systems, errors of at least
30 kJ mol-1 can be expected in the comparison of
electronically dissimilar transition states, even when
accounting for solvent effects. We would thus like to
stress the necessity of careful validation, preferably by
comparison of selected computational predictions with
experimental data.
F IGURE 11. The two favored forms of 9a ‡.
The energy penalty even for methyl substitution in these
positions is severe, much more so than for a mere axial
cyclohexane substituent.
All in all, only two of the eight possible addition modes
are energetically feasible, RSS-eq and RSR-eq (Figure
11). All other isomers of 9a ‡ are at least 20 kJ mol-1
higher than these two. The pro-(E) isomer, RSR-eq, is
higher in energy by 4.8 kJ mol-1, corresponding very well
to the experimental (Z)-preference.
Transition state 4‡ is in many ways similar to 9a ‡.
Interaction energies are higher, due to the shorter
interatomic distances and the fact that substituents are
now eclipsed. The absolute calculated barriers at our
highest accessible level usually differ by only a few kJ
mol-1 from the corresponding isomers of 9a ‡. However,
comparing the two favored isomers from 9a ‡, the energy
difference in 4‡ has been increased to 13.7 kJ mol-1, in
poor agreement with the experimental results. We there-
fore consider it unlikely that 4‡ influences the selectivity,
and that the high calculated energies are artifactual. A
likely source for the error is the application of a harmonic
approximation in the evaluation of the vibrational con-
tributions to the free energies of this floppy TS.
At this point, we will briefly consider TS 9b‡. In
agreement with the BEP principle,33 the TS leading to
the high energy cis enolate 2b is generally higher in
energy than the TS to 2a and always more than 20 kJ
mol-1 higher than the most favored form of 9a ‡, RSS-eq.
Formation of cis enolate in situ by the addition-reversion
route can therefore be excluded in this system.
Looking finally at TS 10‡, the calculated energies of
this TS are substantially higher than those for TS 9a ‡.
However, we have identified several reasons why this
high energy may be an artifact. Looking at the selectivity
profile expected from TS 10‡, we see that the major
difference between the isomeric forms lies in the interac-
tions of the methyl substituent. As stated for the acetone
model, all pro-(Z) forms will show an interaction between
the substituent and the ester moiety, whereas the sub-
stituent only interacts with the enolate hydrogen in the
pro-(E) forms. Interestingly, the methyl group prefers an
axial conformation for all diastereomers. RSR is lowest
in energy, with pro-(Z) forms RRR and RSS disfavored
by 5.3 and 6.9 kJ mol-1, respectively. A rate-determining
TS 10‡ is therefore at odds with the experimental picture.
Returning to the most likely RDS, 9a ‡, we wish to point
out that the existing chiral center has a strong influence
on the configuration of the former carbonyl carbon
(Felkin-Anh-Eisenstein-type selectivity), whereas the
enolate R-carbon is only weakly affected. The former
preference is > 20 kJ mol-1, whereas the latter is ca. 4
kJ mol-1. Both of these conditions are necessary for a
successful application of our recently introduced enan-
Con clu sion s
The preferred configuration of lithium enolates of
simple R-silyl esters is trans, 2a .17 The reaction of this
enolate with a ketone proceeds through four transition
states, three of which can act as a selective filters and
prohibit selective paths, in analogy with the previously
investigated HWE reaction11,12 and Aggarwal’s aldehyde
epoxidation.35 At least two of these transition states are
expected to disfavor (E)-product. The final elimination
of lithium siloxide is expected to be facile, and faster than
bond rotation and stereochemical scrambling in 5. Z
selectivity in the addition of 2-methylcyclohexanone
appears to be determined by the initial transition state
9a ‡, and the predicted product ratio is in good accordance
with experiment. The mechanistic insight provided by the
present study is currently being exploited for design of
stereoselective Peterson reagents suitable for asymmetric
synthesis. Results will be reported in due course.
Exp er im en ta l Section
Gen er a l. Tetrahydrofuran was distilled under nitrogen
from sodium/benzophenone prior to use. All reactions were
carried out in oven-dried glassware and under nitrogen.
Commercially available compounds were used as received.
P eter son Rea ction s w ith LiHMDS. Aceton e. LiHMDS
(1.0 M in THF, 1.40 mL, 1.4 mmol) was added over a period of
2 min to a solution of methyl trimethylsilyl acetate (0.25 mL,
1.4 mmol) in THF (6 mL) at 0 °C. The resulting yellow solution
was stirred for 30 min at 0 °C. The temperature was then
lowered to -78 °C, and acetone (0.11 mL, 1.5 mmol) was added
dropwise to the reaction mixture. Stirring was continued at
-78 °C for 2 h. Saturated ammonium chloride (10 mL) was
added, and the mixture was extracted with diethyl ether. GC-
MS of this extract showed the desired product (methyl 3,3-
dimethylacrylate) and hexamethyldisilazane. No attempt was
made to isolate the product due to its high volatility. Both
methyl 3,3-dimethylacrylate37 and hexamethyldisilazane38
showed MS data in accord with those previously reported.
Cycloh exa n on e. LiHMDS (1.0 M in THF, 1.40 mL, 1.4
mmol) was added over a period of 2 min to a solution of methyl
trimethylsilyl acetate (0.25 mL, 1.4 mmol) in THF (6 mL) at 0
°C. The resulting yellow solution was stirred for 30 min at 0
°C. The temperature was then lowered to -78 °C, and
cyclohexanone (0.16 mL, 1.5 mmol) was added dropwise to the
reaction mixture. Stirring was continued at -78 °C for 2 h.
(37) Leung-Toung, R.; Wentrup, C. Tetrahedron 1992, 48, 7641-
7654.
(38) Sharkey, A. G., J r.; Friedel, R. A.; Langer, S. H. Anal. Chem.
1957, 29, 770-776.
J . Org. Chem, Vol. 67, No. 21, 2002 7387

Gillies et al.
TABLE 6. 300 MHz NMR Da ta (CDCl3) for 12a a n d 12b
(p p m )
Saturated ammonium chloride (10 mL) was added, and the
mixture was extracted with diethyl ether. GC-MS of this
extract showed that 96% of the cyclohexanone had been
converted into the desired product (methyl cyclohexylidene-
acetate) and that 4% unreacted cyclohexanone was left. The
combined organic extracts were dried (MgSO4), and the solvent
was evaporated to give 0.208 g (96%) of methyl cyclohexyli-
1H NMR
OCH3
CH
OSi(CH3)3
Si(CH3)3
12a
12b
3.49
3.54
3.08
2.97
0.26
0.20
0.03
0.05
13C NMR
CO2
OCH3
CH
OSi(CH3)3
1
deneacetate with H NMR and MS spectral data in accord with
those reported earlier.39,40 The crude product was >95% pure,
12a
12b
161.59
164.67
72.78
63.94
54.00
55.05
0.17
0.49
1
according to H NMR.
2-Meth ylcycloh exa n on e. LiHMDS (1.0 M in THF, 1.40
mL, 1.4 mmol) was added to a 0 °C solution of methyl
trimethylsilyl acetate (0.25 mL, 1.5 mmol) in THF (3 mL). The
resulting yellow solution was stirred for 30 min at 0 °C, before
the temperature was lowered to -78 °C. 2-Methyl cyclohex-
anone (0.18 mL, 1.5 mmol) was added dropwise, and stirring
continued at -78 °C for 2 h. Ammonium chloride (sat. solution,
10 mL) was added, and the mixture was extracted with diethyl
ether, dried (MgSO4), and evaporated to give 164 mg of a
colorless oil. 1H NMR showed a mixture of the (Z)- and (E)-
isomer in a relationship of 92:8. Gas chromatography showed
a (Z/E) relationship of 93:7. Separation of the two isomers by
column chromatography was not possible.
LiHMDS, 0 °C. LiHMDS (1.0 M in THF, 1.5 mL, 1.5 mmol)
was added to a 0 °C solution of methyl trimethylsilyl acetate
(0.25 mL, 1.5 mmol) in THF (6 mL). The resulting yellow
solution was stirred for 30 min at 0 °C. TMSCl (0.29 mL, 2.25
mmol) was then added dropwise. Stirring was continued at 0
°C for 1 h before application of vacuum (2.5 mmHg) for removal
of the THF. CDCl3 was added to the solid residue and the
suspension filtered through cotton into an NMR tube. Apart
from 6% unconverted starting material, 1H NMR showed only
the trans isomer 12a .
LDA, -78 °C. Butyllithium (1.6 M in hexanes, 0.94 mL,
1.5 mmol) was added to a 0 °C solution of diisopropylamine
(0.21 mL, 1.5 mmol) in THF (6 mL), and stirring was continued
for 30 min. The solution was cooled to -78 °C, whereupon
methyl trimethylsilyl acetate (0.25 mL, 1.5 mmol) was slowly
added. The resulting yellow solution was stirred for 60 min at
-78 °C. TMSCl (0.29 mL, 2.25 mmol) was then added drop-
wise. Stirring was continued at -78 °C for 2 h before
application of vacuum (2.5 mmHg) for removal of the THF.
CDCl3 was added to the solid residue and the suspension
filtered through cotton into an NMR tube. In addition to 12%
unconverted starting material, 1H NMR showed the two
isomers (12a and 12b) in a 75:25 relationship (see Table 5 and
Table 6).
The 1H NMR data of the (Z)-isomer were in accordance with
those reported earlier.41 It was not possible to assign all signals
from the (E)-isomer as most of them were overlapped by the
signals from the (Z)-isomer. HRMS m/z calcd for C10H16O2 (M+)
168.1150, found 168.1156.
Attem p ted P eter son Rea ction s w ith KHMDS. Aceton e.
KHMDS (279 mg, 1.5 mmol) was added to a solution of methyl
trimethylsilyl acetate (0.25 mL, 1.4 mmol) in THF (6 mL) at 0
°C. The resulting yellow solution was stirred for 30 min at 0
°C. The temperature was then lowered to -78 °C, and acetone
(0.11 mL, 1.5 mmol) was added dropwise. Stirring was
continued at -78 °C for 2 h. Saturated ammonium chloride
(10 mL) was added, and the mixture was extracted with diethyl
ether. GC-MS of this extract showed only hexamethyldisila-
zane.38
Cycloh exa n on e. KHMDS (279 mg, 1.5 mmol) was added
to a solution of methyl trimethylsilyl acetate (0.25 mL, 1.4
mmol) in THF (6 mL) at 0 °C. The resulting yellow solution
was stirred for 30 min at 0 °C. The temperature was then
lowered to -78 °C, and cyclohexanone (0.16 mL, 1.5 mmol)
was added dropwise. Stirring was continued at -78 °C for 2
h. Saturated ammonium chloride (10 mL) was added, and the
mixture was extracted with diethyl ether. GC-MS of this
extract showed that only 3% of the desired product (methyl
cyclohexylideneacetate) had formed, whereas 36% cyclohex-
anone was left unchanged and 61% 1-trimethylsilyloxycyclo-
hexene had formed. The MS data for methyl cyclohexylidene-
acetate40 and 1-trimethylsilyloxycyclohexene42 were in accord
with those described earlier. No attempt was made to isolate
the products.
En ola te-Tr a p p in g Exp er im en ts. LiHMDS, -78 °C. LiH-
MDS (1.0 M in THF, 1.5 mL, 1.5 mmol) was added to a -78
°C solution of methyl trimethylsilyl acetate (0.25 mL, 1.5
mmol) in THF (6 mL). The resulting yellow solution was
stirred for 60 min at -78 °C. TMSCl (0.29 mL, 2.25 mmol)
was then added dropwise. Stirring was continued at -78 °C
for 2 h before application of vacuum (2.5 mmHg) for removal
of the THF. CDCl3 was added to the solid residue and the
suspension filtered through cotton into an NMR tube. 1H NMR
showed only one species, which was assigned as the trans-
isomer 12a by means of 1D differential NOE (Figure 4). The
1H and 13C NMR data are listed in Table 6.
LDA, 0 °C. Butyllithium (1.6 M in hexanes, 0.94 mL, 1.5
mmol) was added to a 0 °C solution of diisopropylamine (0.21
mL, 1.5 mmol) in THF (6 mL), and stirring was continued for
30 min. Methyl trimethylsilyl acetate (0.25 mL, 1.5 mmol) was
slowly added and the solution stirred for 30 min at 0 °C.
TMSCl (0.29 mL, 2.25 mmol) was then added dropwise and
stirring was continued at 0 °C for 1 h, before application of
vacuum (2.5 mmHg) for removal of the THF. CDCl3 was added
to the solid residue and the suspension filtered through cotton
into an NMR tube. In addition to 31% unconverted starting
1
material, H NMR showed the two isomers (12a and 12b) in
a 88:12 relationship (see Table 5 and Table 6).
KHMDS. KHMDS (299 mg, 1.5 mmol) was added to a 0 °C
solution of methyl trimethylsilyl acetate (0.25 mL, 1.5 mmol)
in THF (6 mL). The resulting mixture was stirred for 30 min
at 0 °C. TMSCl (0.29 mL, 2.25 mmol) was then added dropwise.
Stirring was continued at 0 °C for 1 h before application of
vacuum (2.5 mmHg) and removal of the THF. CDCl3 was
added to the residue and the suspension filtered through cotton
into an NMR tube. In addition to 60% unconverted starting
1
material, H NMR showed the two isomers (12a and 12b) in
a 25:75 relationship (see Table 5 and Table 6).
Ack n ow led gm en t . Financial support from the
Danish Natural Sciences Research Council, the Carls-
berg Foundation, the Lundbeck Foundation, and the
Torkil Holm Foundation is gratefully acknowledged.
J .E.T. thanks Novo Nordisk for support. We thank
Professor Tobias Rein for stimulating discussions.
(39) Hamdouchi, C.; Topolski, M.; Goedken, V.; Walborsky, H. M.
J . Org. Chem. 1993, 58, 3148-3155.
Su p p or tin g In for m a tion Ava ila ble: Examples of J aguar
input files. Cartesian coordinates and raw energies of compu-
tationally characterized species (ASCII). This material is
available free of charge via the Internet at http://pubs.acs.org.
(40) Hon, Y.-S.; Lu, L. Tetrahedron 1995, 51, 7937-7942.
(41) Duraisamy, M.; Walborsky, H. M. J . Am. Chem. Soc. 1983, 105,
3252-3264.
(42) Hydrio, J .; Van de Weghe, P.; Collin, J . Synthesis 1997, 1, 68-
72.
J O0262107
7388 J . Org. Chem., Vol. 67, No. 21, 2002