Origins of Stereocontrol in the [2 + 2] Cycloaddition between Achiral Ketenes and Chiral α-Alkoxy Aldehydes. A Pericyclic Alternative to the Aldol Reaction

Article abstract of DOI:10.1021/jo9806512

Ab Initio calculations predict that the thermal [2 + 2] cycloaddition reaction between C_2v-symmetric ketenes and enantiopure aldehydes takes place with high enough stereocontrol for preparative purposes. The sense of induction is predicted to be non-Felkin. It is also found that the [2 + 2] cycloaddition involving nonactivated ketenes is facilitated by using 5 M solutions of lithium perchlorate in diethyl ether. It is found that both the purely thermal and lithium-assisted [2 + 2] cycloadditions result in the same type of stereocontrol. This method constitutes a general route for the synthesis of homochiral 2-oxetanones and related compounds, thus providing a pericyclic alternative to the aldol reaction.

Full text of DOI:10.1021/jo9806512

5216
J . Org. Chem. 1998, 63, 5216-5227
Or igin s of Ster eocon tr ol in th e [2 + 2] Cycloa d d ition betw een
Ach ir a l Keten es a n d Ch ir a l r-Alk oxy Ald eh yd es. A P er icyclic
Alter n a tive to th e Ald ol Rea ction
Begon˜a Lecea, Ana Arrieta, Iosune Arrastia, and Fernando P. Coss´ıo*
Kimika Fakultatea, Euskal Herriko Unibertsitatea, P.K. 1072, San Sebastia´n-Donostia, Spain, and
Farmazi Fakultatea, Euskal Herriko Unibertsitatea, P.K. 450, 010080 Vitoria-Gasteiz, Spain
Received April 8, 1998
Ab Initio calculations predict that the thermal [2 + 2] cycloaddition reaction between C2v-symmetric
ketenes and enantiopure aldehydes takes place with high enough stereocontrol for preparative
purposes. The sense of induction is predicted to be non-Felkin. It is also found that the [2 + 2]
cycloaddition involving nonactivated ketenes is facilitated by using 5 M solutions of lithium
perchlorate in diethyl ether. It is found that both the purely thermal and lithium-assisted [2 + 2]
cycloadditions result in the same type of stereocontrol. This method constitutes a general route
for the synthesis of homochiral 2-oxetanones and related compounds, thus providing a pericyclic
alternative to the aldol reaction.
In tr od u ction
Since the pioneering studies of Wilsmore and
Staudinger,1 the exceptional ability of ketenes to partici-
pate in thermal cycloadditions is well recognized. Indeed,
in his recent monograph on ketenes,2 Tidwell has written
that “cycloaddition has remained as the most distinctive,
useful, and intellectually challenging, aspect of ketene
chemistry.” In particular, these heterocumulenes are
very versatile reagents for thermal [2 + 2] cycloaddi-
tions.2,3 Given that this kind of reaction is in general
difficult under thermal conditions, the electronic pecu-
liarities of ketenes attracted considerable attention in the
course of the development of the theory of pericyclic
reactions.4 Therefore, [2 + 2] cycloadditions between
ketenes and imines,5 alkenes,6 alkynes,7 nitroso,8 azo,9
and carbonyl10 compounds have been described, and some
of these reactions have proven to be very useful in the
chemical synthesis of pharmacologically valuable com-
pounds.11
Although all the above-mentioned reactions can be
formally envisaged as [2 + 2] cycloadditions, the actual
F igu r e 1. Schematic representation of the reactive sites of a
ketene toward nucleophiles (ipso interaction) and electrophiles
(R interaction). The curved line represents an intramolecular
connection between the nucleophilic and electrophilic sites.
(1) (a) Wilsmore, N. T. M. J . Chem. Soc. 1907, 91, 1938. (b)
Staudinger, H.; Kleuer, H. W. Chem. Ber. 1906, 39, 968.
(2) Tidwell, T. T. Ketenes; Wiley: New York, 1995; pp 459-571 and
references therein.
(3) (a) Hyatt, J . A.; Raynolds, P. W. Ketene Cycloadditions. In
Organic Reactions; Paquette, L., et al., Eds.; Wiley: New York, 1994;
Vol. 45, pp 159-646. (b) Ghosez, L.; Marchand Brynaert, J . In
Comprehensive Organic Chemistry; Trost, B. M., Fleming, I., Eds.;
Pergamon: Oxford, 1991; Vol. 5, pp 86-89. (c) Oligaruso, M. A.; Wolfe,
J . F. In Synthesis of Lactones and Lactams; Patai, S., Rappoport, Z.,
Eds.; Wiley: Chichester, 1993; pp 63-76, 303.
mechanisms operating in these transformations can be
different. This variable behavior can be rationalized by
taking into account that the π-frontier orbitals of ketenes
lie in orthogonal planes.2 Thus, a nucleophile can
interact with the ipso sp-hybridized carbon atom of the
ketene by means of the in-plane interaction depicted in
Figure 1. On the other hand, an electrophile must
interact via the R carbon atom of the ketene. The
trajectory of this latter attack is determined by the
HOMO plane of the ketene, which is perpendicular to
its molecular plane (Figure 1).
In a concerted reaction, the ketene and the π-system
must suffer both interactions in a singlesbut not neces-
sarily synchronoussstep, thus leading to a distorted non-
supra-supra geometry of the corresponding transition
state, such as the [π2s + π2a] mechanism proposed by
(4) Wang, X.; Houk, K. N. J . Am. Chem. Soc. 1990, 112, 1754 and
footnote 10 therein.
(5) Georg, G. I.; Ravikumar, V. T. In The Organic Chemistry of
â-Lactams; Georg, G.-L., Ed.; Werlag Chemie: New York, 1993; pp
295-331.
(6) Snider, B. B. Chem. Rev. 1988, 88, 793.
(7) See, for example: (a) Hassner, A.; Dillon, J . Synthesis 1979, 689.
(b) Danheiser, R. L.; Savariar, S. Tetrahedron Lett. 1987, 28, 3299. (c)
Fishbein, P. L.; Moore, H. W. J . Org. Chem. 1985, 50, 3226.
(8) (a) Kerber, R. C.; Cann, M. C. J . Org. Chem. 1974, 39, 2552. (b)
Kresze, G.; Trede, A. Tetrahedron 1963, 19, 133.
(9) (a) Fahar, E.; Fisher, W.; J ung, A.; Savel, L. Tetrahedron Lett.
1967, 161. (b) Hall, J . H.; Kellogg, R. M. J . Org. Chem. 1966, 31, 1079.
(10) Pommier, A.; Pons, J .-M. Synthesis 1993, 441.
(11) Pommier, A.; Pons, J .-M. Synthesis 1995, 729.
S0022-3263(98)00651-3 CCC: $15.00 © 1998 American Chemical Society
Published on Web 07/08/1998

Pericyclic Alternative to the Aldol Reaction
J . Org. Chem., Vol. 63, No. 15, 1998 5217
Sch em e 1a
a “pericyclic” alternative to the aldol reaction, the latter
being of paramount importance in organic synthesis.19
In effect, the product of this [2 + 2] cycloaddition, namely
a 2-oxetanone (â-lactone), and the product of an aldol
reaction, for instance, a â-hydroxy ester, can be inter-
converted into each other via alcoholysis or cyclization,
respectively (Scheme 1). However, two drawbacks have
hampered the development of this reaction: First, in
general, nonnucleophilic ketenes are quite unreactive
toward carbonyl compounds unless suitable catalysts are
employed.3,11 Second, the stereoselectivity of the cycload-
dition is in general poor, although efficient stereocontrol
for the catalyzed version has been achieved in several
cases.3,11,20
a
The possible substituents at the different positions are not
specified.
Within this context, the aim of the present work has
been to explore computationally the possibility of achiev-
ing high stereocontrol in the [2 + 2] cycloaddition
between C2v-symmetric ketenes and homochiral alde-
hydes. In addition, since we have developed recently21
a method that allows the [2 + 2] cycloaddition between
unactivated ketenes and aromatic or R,â-unsaturated
aldehydes, we have verified the predictions made by
theoretical methods. The final goal of the work has been
to improve our understanding of the reasons underlying
the stereocontrol in this kind of cycloadditions, thus
yielding a complementary alternative to aldol-type reac-
tions.
Woodward and Hoffmann12 or, more recently the [π2s +
(π2s + 2s)] mechanism.13
π
According to the above-mentioned scheme, electrophilic
ketenes, namely those incorporating electron-withdraw-
ing groups, will be especially prone to the ipso interaction
represented in Scheme 1. If the other partner is in turn
nucleophilic enough, the relative importance of the ipso
interaction is quite large and the cycloaddition is actually
a two-step reaction in which the Cipso-Y bond is formed
first, followed by a ring closure to form the CR-X bond.
This is the case in the reaction between ketenes and
imines, also known as the Staudinger reaction. This
nonconcerted mechanism has been studied by our group14
and by others.15 In other cycloadditions, the interaction
between ketenes and π-systems is concerted, although
quite asynchronous. In general, the CR-X interaction is
developed to a lesser extent.
Com p u ta tion a l Meth od s
All ab initio calculations described in this work have
been carried out using the GAUSSIAN 9422 suite of
programs. The 6-31G* basis set23 was used, and the
geometries of all the stationary points were fully opti-
mized using analytical gradient techniques.24
Most of the saddle points were located and optimized
using the eigenvector-following algorithm.25 In these
cases, the possible conformers were explored at the HF/
6-31G* and HF/AM126 levels. The conformer distribution
of the diastereomeric cycloadducts was explored by means
of Monte Carlo simulations27 using the MM2* force field28
Recently, we have published two papers concerning the
thermal [2 + 2] cycloaddition between ketenes and
carbonyl compounds.16 This reaction has also been
explored computationally by Rajzmann17 and Yamabe.18
It has been found that this reaction is concerted and
follows a [π2s + (π2s + π2s)] mechanism in the absence of
Lewis acid catalysts. The preparative potential of this
reaction has been explored by different groups.10,11 This
is not surprising since the reaction can be considered as
(19) See, for example: (a) Evans, D. A.; Nelson, J . V.; Taber, T. R.
Top. Stereochem. 1982, 13, 1. (b) Kaiama, T.; Kobayashi, S. In Organic
Reactions; Paquette, L., et al., Eds.; Wiley: New York, 1994; Vol. 46.
(20) (a) Pommier, A.; Pons, J .-M.; Kocienski, P. J . J . Org. Chem.
1995, 60, 7334. (b) Pommier, A.; Pons, J .-M.; Kocienski, P. J .; Wong,
L. Synthesis 1995, 1294. (c) Song, C. E.; Ryu, T. H. R.; Roh, E. J .; Kim,
I. O.; Ha, H.-J . Tetrahedron: Asymmetry 1994, 5, 1215. (d) Tamai, Y.;
Yoshiwara, H.; Someya, M.; Fukumoto, J .; Miyano, S. J . Chem. Soc.,
Chem. Commun. 1994, 2281. (e) Palomo, C.; Miranda, J . I.; Cuevas,
C.; Odriozola, J . M. J . Chem. Soc., Chem. Commun. 1995, 1735. (f)
Tamai, Y.; Someya, M.; Fukumoto, J .: Miyano, S. J . Chem. Soc., Perkin
Trans. 1 1994, 1549.
(21) Arrastia, I.; Coss´ıo, F. P. Tetrahedron Lett. 1996, 37, 7143.
(22) GAUSSIAN 94, Revision B.2: Frisch, M. J .; Trucks, G. W.;
Schlegel, H. B.; Gill, P. M. W.; J ohnson, B. G.; Robbb, M. A.;
Cheeseman, J . R.; Keith, T.; Peterson, G. A.; Montgomery, J . A.;
Raghavachari, K.; Al-Laham, M. A.; Zakrzewski, V. G.; Ortiz, J . V.;
Foresman, J . B.; Peng, C. Y.; Ayala, P. Y.; Chen, W.; Wong, M. W.,
Andres, J . L.; Replogle, E. S.; Gomperts, R.; Martin, R. L.; Fox, D. J .;
Binkley, J . S.; Defrees, D. J .; Baker, J .; Stewart, J . P.; Head-Gordon,
M.; Gonzalez, C.; Pople, J . A. Gaussian, Inc., Pittsburgh, PA, 1995.
(23) Hehre, W. J .; Radom, L.; Schleyer, P.v. R.; Pople, J . A. Ab Initio
Molecular Orbital Theory; Wiley: New York, 1986; p 82 and references
therein.
(12) Woodward, R. B.; Hoffmann, R. The Conservation of Orbital
Symmetry; Verlag Chemie-Academic Press: New York, 1970; p 163.
(13) (a) Pasto, D. J . J . Am. Chem. Soc. 1979, 101, 37. (b) Valent´ı,
E.; Perica`s, M. A.; Moyano, A. J . Org. Chem. 1990, 55, 3582. (c)
Krabbenhoft, H. O. J . Org. Chem. 1978, 43, 1305.
(14) (a) Coss´ıo, F. P.; Ugalde, J . M.; Lopez, X.; Lecea, B.; Palomo,
C. J . Am. Chem. Soc. 1993, 115, 995. (b) Coss´ıo, F. P.; Arrieta, A.;
Lecea, B.; Ugalde, J . M. J . Am. Chem. Soc. 1994, 116, 2085. (c) Arrieta,
A.; Ugalde, J . M.; Coss´ıo, F. P.; Lecea, B. Tetrahedron Lett. 1994, 35,
4465. (d) Arrastia, I.; Arrieta, A.; Ugalde, J . M.; Coss´ıo, F. P.; Lecea,
B. Tetrahedron Lett. 1994, 35, 7825. (e) Lecea, B.; Arrastia, I.; Arrieta,
A.; Roa, G.; Lopez, X.; Arriortua, M. I.; Ugalde, J . M.; Coss´ıo, F. P. J .
Org. Chem. 1996, 61, 3070.
(15) (a) Sordo, J . A.; Gonza´lez, J .; Sordo, T. L. J . Am. Chem. Soc.
1992, 114, 6249. (b) Lo´pez, R.; Sordo, T. L.; Sordo, J . A.; Gonza´lez, J .
J . Org. Chem. 1993, 58, 7036. (c) Assfeld, X.; Ruiz-Lo´pez, M. F.;
Gonza´lez, J .; Lo´pez, R.; Sordo, J . A.; Sordo, T. L. J . Comput. Chem.
1994, 15, 479. (d) Lo´pez, R.; Ruiz-Lo´pez. M. F.; Rinaldi, D.; Sordo, J .
A.; Sordo, T. L. J . Phys. Chem. 1996, 100, 10600. (e) Boyd, D. B.
Computer-Assisted Molecular Design Studies of â-Lactam Antibiotics.
In Frontiers on Antibiotic Research; Umezawa, H., Ed.; Academic
Press: Tokyo, 1987; pp 339-356. (f) Cooper, R. D. G.; Daugherty, B.
W.; Boyd, D. B. Pure Appl. Chem. 1987, 59, 485.
(16) (a) Lecea, B.; Arrieta, A.; Roa, G.; Ugalde, J . M.; Coss´ıo, F. P.
J . Am. Chem. Soc. 1994, 116, 9613. (b) Lecea, B.; Arrieta, A.; Lopez,
X.; Ugalde, J . M.; Coss´ıo, F. P. J . Am. Chem. Soc. 1995, 117, 12314.
(17) (a) Pons, J .-M.; Pommier, A.; Rajzmann, M.; Liotard, D.
THEOCHEM 1994, 313, 361. (b) Pons, J .-M.; Oblin, M.; Pommier, A.;
Rajzmann, M.; Liotard, D. J . Am. Chem. Soc. 1997, 119, 3333.
(18) Yamabe, S.; Minato, T.; Osamura, Y. J . Chem. Soc., Chem.
Commun. 1993, 450.
(24) See ref 23, pp 55, 56 and references therein.
(25) Baker, J . J . Comput. Chem. 1986, 7, 385.
(26) Dewar, M. J . S.; Zoebish, E. G.; Healy, E. F.; Stewart, J . J . P.
J . Am. Chem. Soc. 1985, 107, 3902.
(27) (a) Chang, G.; Guida, W. C.; Still, W. C. J . Am. Chem. Soc. 1989,
111, 4379. (b) Saunders: M.; Houk, K. N.; Wu, Y.-D.; Still, W. C.;
Lipton, M.; Chang, G.; Guida, W. J . Am. Chem. Soc. 1990, 112, 1419.
(28) Allinger, N. L. J . Am. Chem. Soc. 1977, 99, 8127.

5218 J . Org. Chem., Vol. 63, No. 15, 1998
Lecea et al.
as implemented in the Macro Model package.29 The
selected minima were further reoptimized at the RHF/
AM1 and HF/6-31G* levels. Unless otherwise stated,
conformers included in the next section represent the
lowest energy conformation for a given reaction path.
All the saddle points located and characterized at the
HF/6-31G* level were further optimized using density
functional theory30 (DFT). For this purpose, we have
used the hybrid three-parameter functional developed by
Becke,31 which has proven to be quite reliable in pericyclic
reactions.32 This functional combines the Becke’s gradi-
ent corrected functional and the Lee-Yang-Parr and
Vosko-Wilk-Nusair correlation functionals31 with part
of the exact Hartree-Fock exchange energy. This method
is denoted in the text as B3LYP. All stationary points
were characterized by harmonic analysis.33 Zero-point
energies obtained at the HF/6-31G* levels were scaled34
by a factor of 0.89. The ZPVEs obtained at the B3LYP
/6-31G* level were not scaled.
Sch em e 2
rupole and higher multipoles. Therefore, we have also
calculated the relative energies of several stationary
points using the SCIPCM model.40 This approach can
be considered as equivalent to an infinite expansion in
the multipole series. Most of the calculations included
in this work have been computed at ꢀ ) 9.08, which
corresponds to dichloromethane.
The differences in energy were computed at the B3LYP/
6-31G* level, including ZPVE corrections. Aside from the
DFT calculations, electron correlation has been also
estimated at the second-order Møller-Plesset (MP2)35
level on HF/6-31G* geometries.
Donor-acceptor interactions have been calculated by
means of the natural bond orbital (NBO) model.41 Ac-
cording to the NBO decomposition, the energy of a
donor-acceptor interaction is given by
Given the importance of solvent effects in cycloaddition
reactions,36 several calculations were also performed
taking into account these effects. Solvent effects were
considered using the self-consistent reaction field (SCRF)
method.37 In one of the SCRF methods included in our
calculations, the solute is described by its dipole moment
and an spherical cavity of radius ao. The solvent is
described by its dielectric constant ꢀ. This method is
usually mentioned as the Onsager model.38 In this work,
it will be denoted as L1A1 (multipole expansion truncated
at lmax ) 1, spherical cavity).38 In a previous paper,39 we
have found that the description provided by the L1A1
model is not accurate enough for describing diastereo-
meric structures in which the relative orientation of the
dipoles can promote substantial variations in the quad-
2
φ*|Fˆ|φ
∆E(φ2φ)* ) -nφ
(1)
ꢀ
φ* - ꢀφ
where nφ is the occupation number of the donor orbital,
Fˆ is the Fock operator, φ and φ* are two filled and unfilled
NBOs, respectively, and ꢀφ and ꢀ* are their respective
φ
energies. The Wiberg bond indices42 have also been
computed on an NBO basis.
Resu lts a n d Discu ssion
As representative model compounds, we selected the
ketenes and aldehydes shown in Scheme 2. Apart from
ketene 1b itself, we have selected dichloroketene 1a as
a model of an activated electrophilic ketene and dimeth-
ylketene 1c as a representative case of a stabilized
nonelectrophilic ketene.43 (S)-2-Methoxypropionaldehyde
2a was considered as a computationally tractable ana-
logue of the synthetically more useful (S)-2-(benzyloxy)-
propionaldehyde 2b. The possible syn and anti cycload-
ducts 3a -f and 4a -f, respectively, are also collected in
Scheme 2.
(29) MacroModel V5.0: Mohamadi, F.; Richards, N. G. J .; Guida,
W. C.; Liskamp, R.; Lipton, M.; Caufield, C.; Chang, G.; Hendrickson,
T.; Still, W. C. J . Comput. Chem. 1990, 11, 440.
(30) Parr, R. G.; Yang, W. Density-Functional Theory of Atoms and
Molecules; Oxford: New York, 1989.
(31) (a) Becke, A. D. J . Chem. Phys. 1993, 98, 5648. (b) Becke, A. D.
Phys. Rev. A 1988, 38, 3098. (c) Lee, C.; Yang, W.; Parr, R. G. Phys.
Rev. B 1980, 37, 785. (d) Vosko, S. H.; Wilk, L.; Nusair, M. Can. J .
Phys. 1980, 58, 1200.
(32) See, for example: (a) Wiest, O.; Black, K. A.; Houk, K. N. J .
Am. Chem. Soc. 1994, 116, 10336. (b) Wiest, O.; Houk, K. N.; Black,
K. A.; Thomas, B., IV. J . Am. Chem. Soc. 1995, 117, 8594. (c) Goldstein,
E.; Beno, B.; Houk, K. N. J . Am. Chem. Soc. 1996, 118, 6036. (d) Yoo,
H. Y.; Houk, K. N. J . Am. Chem. Soc. 1997, 119, 2877.
(33) McIver, J . W.; Komornicki, A. K. J . Am. Chem. Soc. 1972, 94,
2625.
In a previous work,39 we have reported a computational
study on the different conformations of (S)-2-methoxy-
propanal 2a both in the gas phase and in solution. This
aldehyde has also been studied by Frenking et al.44 At
the B3LYP/6-31G*, B3LYP(L1A1)/6-31G*, and MP2-
(SCIPCM)/6-31G*//HF(L1A1)/6-31G* levels, the relative
(34) Pople, J . A.; Schleyer, B.; Krishnan, R.; DeFrees, D. J .; Binkley,
J . S.; Frisch, H.; Whiteside, R.; Hout, R. F., J r.; Hehre, W. J . Int. J .
Quantum Chem. Symp. 1981, 15, 269.
(35) (a) Binkley, J . S.; Pople, J . A. Int. J . Quantum. Chem. 1975, 9,
229. (b) Pople, J . A.; Binkley, J . S.; Seeger, R. Int. J . Quantum Chem.
Symp. 1976, 10, 1.
(36) Reichardt, C. Solvents and Solvent Effects in Organic Chemistry;
VCH: Weinheim, 1990; pp 147-170.
(37) (a) Tomasi, J .; Persico, M. Chem. Rev. 1994, 94, 2027. (b)
Simkin, B. Ya.; Sheikhet, I. Quantum Chemical and Statistical Theory
of Solutions-A Computational Approach; Ellis Horwood: London, 1995;
pp 78-101. (c) Cramer, C. J .; Trhular, D., D. G. In Reviews in
Computational Chemistry; Lipkowitz, K. B., Boyd, D. B., Eds.; VCH
Publishers: New York, 1995; Vol. VI, pp 1-72.
(38) (a) Wong, M. W.; Wiberg, K. B.; Frisch, M. J . J . Am. Chem.
Soc. 1992, 114, 523. (b) Wong, M. W.; Wiberg, K. B.; Frisch, M. J . J .
Am. Chem. Soc. 1992, 114, 1645.
(40) (a) Wiberg, K. B.; Castejon, H.; Keith, T. A. J . Comput. Chem.
1996, 17, 185. (b) Wiberg, K. B.; Keith, T. A.; Frisch, M. J .; Murcko,
M. J . Phys. Chem. 1995, 99, 9072. (c) Wiberg, K. B.; Rablen, P. R.;
Rush, D. J .; Keith, T. A. J . Am. Chem. Soc. 1995, 117, 4261.
(41) Glendening, E. D.; Reed, A. E.; Carpenter, J . E.; Weinhold, F.
NBO Version 3.1, Gaussian, Inc., 1995.
(42) Wiberg, K. B. Tetrahedron 1968, 24, 1083.
(43) Gong, L.; McAllister, M. A.; Tidwell, T. T. J . Am. Chem. Soc.
1991, 113, 6021.
(39) Lecea, B.; Arrieta, A.; Coss´ıo, F. P. J . Org. Chem. 1997, 62,
6485.
(44) Frenking, G.; Ko¨hler, K. F.; Reetz, M. T. Tetrahedron 1991, 47,
8991.

Pericyclic Alternative to the Aldol Reaction
J . Org. Chem., Vol. 63, No. 15, 1998 5219
F igu r e 2. Six conformations of (S)-2-methoxypropanal at the
B3LYP/6-31G* + ∆ZPVE, B3LYP(L1A1)/6-31G* + ∆ZPVE (in
parentheses), and MP2(SCIPCM)/6-31G*//HF(L1A1)/6-31G* +
∆ZPVE (in square brackets) levels. Numbers are relative
energies in kcal/mol. The data in solution refer to ꢀ ) 9.08
(dichloromethane).
F igu r e 4. Ball and stick representations of diastereomeric
saddle points TSa and TS′a . Numbers correspond to the
Wiberg bond indices, computed at several theoretical levels.
In this and Figures 5-8, unless otherwise stated, atoms are
represented by increasing order of shadowing as follows: H,
C, O.
F igu r e 3. Two ion-molecule complexes resulting from the
interaction between (S)-3-methoxypropanal and lithium cation.
Numbers correspond to the relative energies (in kcal/mol) at
the B3LYP/6-31G* + ∆ZPVE and MP2/6-31G*//HF/6-31G* +
∆ZPVE level (in parentheses).
of ca. 78:22 at 25 °C. Inclusion of solvent effects45 results
in lower computed activation energies and also in lower
stereocontrol (see Table 2). For instance, the calculated
activation energy at the B3LYP(SCIPCM)/6-31G*//B3LYP-
(L1A1)/6-31G* is ca. 0.8 kcal/mol lower than that pre-
dicted in the gas phase. In addition, the ∆∆Ea value
calculated at the same theoretical level is ca. 0.3 kcal/
mol lower than that obtained in vacuo. This result is
even more pronounced when electron correlation was
approximately evaluated using the MP2 approach. Thus,
energies of the different conformers are at those reported
in Figure 2. According to these results, conformer A is
the most stable one both in the gas phase and in
dichloromethane solution, although the L1A1 model
overestimates the stability of conformer B. Similarly, we
have found39 that when cations such as lithium are
present, the chelated structure H is largely the most
stable one, as shown in Figure 3.
(45) One reviewer has suggested that hydrogen bonding between
Et3N+H (formed in the course of the generation of the ketene, vide
infra) and the sp3-hybridized oxygen atom of the aldehyde could be
more important than solvent effects. To check this hypothesis, we have
optimized the complex between ammonium cation and hydroxyacetal-
dehyde. We have found that at the HF/6-31G* level the cation forms
a hydrogen bond with the hydroxylic oxygen. However, at the B3LYP/
6-31G* level, hydrogen bonding occurs through the oxygen of the
carbonyl group. These interactions should be lower at the TS and with
bulkier substituents. The electrostatic effect, if any, should be not very
directional in character and will not affect significantly the stereo-
chemical outcome.
We have studied first the possible reaction paths
associated with the interaction between 2a and dichlo-
roketene 1a to form either syn- of anti-3a (see Scheme
2). We have located two diastereomeric saddle points
denoted as syn -TSa and a n ti-TSa in Figure 4. The
corresponding activation and reaction energies, as well
as the relative energies, are shown in Table 2. According
to our results, syn -TSa is the lowest energy saddle point.
Thus, at the MP2/6-31G*//HF/6-31G* and B3LYP/6-31G*
levels, it is predicted that syn -TSa is 0.56 and 0.76 kcal/
mol more stable than a n ti-TSa , respectively. The latter
energy difference corresponds to a diastereomeric ratio

5220 J . Org. Chem., Vol. 63, No. 15, 1998
Lecea et al.
Ta ble 1. Activa tion En er giesa ,b (∆Ea , k ca l/m ol), a n d Rea ction En er giesa ,b (∆Er xn , k ca l/m ol), Associa ted w ith th e [2 + 2]
Cycloa d d ition Rea ction betw een Keten es 1a -c a n d Ald eh yd e 2a (Con for m er s A a n d H) To Yield 2-Oxeta n on es syn - a n d
a n ti-3a ,c,e
syn products
anti products
level
MP2/6-31G*//HF/6-31G*
B3LYP/6-31G*
MP2(L1A1)/6-31G*//HF(L1A1)/6-31G*
B3LYP(L1A1)/6-31G*
HF(SCIPCM)/6-31G*//HF(L1A1)/6-31G*
MP2(SCIPCM)/6-31G*//HF(L1A1)/6-31G*
B3LYP(SCIPCM)/6-31G*//B3LYP(L1A1)/6-31G*
∆Ea
∆E′a
∆Erxn
∆Ea
∆∆Ea
∆E′a
∆∆E′a
∆Erxn
∆∆Erxn
1a + 2a f 3a
8.34
16.38
23.02
19.62
22.56
18.68
37.03
23.34
18.82
-42.19
23.58
20.38
22.67
19.20
37.14
23.42
19.25
+0.56
+0.76
+0.11
+0.52
+0.11
+0.08
+0.43
14.37
21.77
+6.03
+5.39
-43.00
-32.50
-44.02
-33.30
-29.74
-42.76
-31.99
-0.81
-0.45
-1.90
-1.48
-1.26
-0.96
-0.69
-32.05
-42.12
-31.82
-28.48
-41.80
-31.30
1b + 2a f 3c
MP2/6-31G*//HF/6-31G*
B3LYP/6-31G*
26.91
26.96
7.38
-30.35
-24.11
30.23
29.22
+3.32
+2.26
8.58
11.26
+1.20
+0.12
-30.87
-24.45
-0.52
-0.34
11.14
1c + 2a f 3e
MP2/6-31G*//HF/6-31G*
B3LYP/6-31G*
27.49
28.23
-0.36
-34.33
-24.37
30.52
29.94
+3.03
+1.71
2.62
9.39
+2.98
+1.08
-39.36
-25.70
-5.03
-1.33
8.31
a
b
ZPVE corrections, computed at the optimization level. ∆∆Ea and ∆∆Erxn are the relative energies with respect to the syn cycloadduct.
Ta ble 2. Ma in Secon d -Or d er En er gies (∆E(2)), k ca l/m ol)
dihedral angles lie in the range 174.7-179.3°. In con-
in th e Tr a n sition Str u ctu r es TSa ,c,e a n d TS′a ,c,e
trast, the syn-2-oxetanones show minimum energy con-
level
φXC f φ* φ* f φ* φXC f φ* φ* f φ*
* XC * XC
formations in which the same dihedral angle varies from
62.3° to 69.7°, thus indicating that both groups are
gauche to each other. In summary, both thermodynamic
control and enolate methodology yield the opposite ster-
eochemical outcome with respect to the [2 + 2] cycload-
dition under kinetic control.
syn -TSa
a n ti-TSa
B3LYP/6-31G*
B3LYP(L1A1)/6-31G*
3.71a
1.47b
2.05b
12.46c
0.00
0.00
2.87a
12.69c
syn -TS′a
a n ti-TS′a
B3LYP/6-31G*
B3LYP/6-31G*
B3LYP/6-31G*
B3LYP/6-31G*
B3LYP/6-31G*
2.18f
1.22g
3.14d
1.60e
syn -TSc
a n ti-TSc
Given the high electrophilicity of ketene 1a , the ipso
interaction is the predominant one in these transition
structures (see Table 1 of the Supporting Information for
geometrical details). Thus, in both isomers of TSa the
bond indices between the O1 and C2 atoms are ca. 0.70,
whereas those associated with the formation of the C3-
C4 bond are ca. 0.25 (see Figure 4, structures syn - and
a n ti-TSa ). The main difference between syn - and a n ti-
TSa is that in the former the C6-O8 bond is almost
antiperiplanar to the C3-C4 bond being formed, the
dihedral angle φ ) C3-C4-C6-O8 being in the range
170.7-178.5° (see Table 1 of the Supporting Information).
In a n ti-TSa , the C6-H7 bond is almost eclipsed to the
C3-C4 bond, the dihedral angle ê ) C3-C4-C6-H7
varying from 1.0 to 10.3°. On the other hand, if we define
the nucleophilic angle of attack in the form RN ) C3-
C4-O1, we observe that in both TSs this value is always
lower than 90°. Therefore, the values of RN do not
correspond to the Bu¨rgi-Dunitz trajectories48 usually
observed in 1,2-nucleophilic additions to carbonyl com-
pounds. If chelation effects are not involved, these latter
processes are usually associated with Felkin-Ahn dia-
stereoselectivities.49
4.42a
1.87b
1.82d
3.39e
syn -TS′c
a n ti-TS′c
1.64d
2.34e
2.42d
3.00e
syn -TSe
a n ti-TSe
4.42a
1.94b
8.42c
1.04b
syn -TS′e
a n ti-TS′e
1.78f
1.14g
2.82d
2.30e
a
b
∆E(2)[σ6,8 f π1*,4].
∆E(2)[π1,4 f σ* ]. c ∆E(2)[σ6,7 f π1*,4].
∆E(2)[σ6,8
f
σ* ]. e ∆E(2)[σ3,4
f
σ*6,68,.8] f ∆E(2)[σ6,9
f σ* ].
3,4
d
3,4
∆E(2)[σ3,4 f σ* .]
g
6,9
at the MP2(SCIPCM)/6-31G*//HF(L1A1)/6-31G* level,
virtually no stereoselection is predicted (∆∆Ea ) 0.08
kcal/mol), a result that is not in agreement with the
experimental evidence (vide infra). On the basis of this
result, we conclude that the B3LYP method is more
appropriate for the evaluation of stereocontrol in this
kind of reactions. The major isomer predicted for this
cycloaddition is syn-3a , which corresponds to a non-
Felkin diastereoselectivity. Interestingly, when 2-oxe-
tanones are synthesized using the enolate methodology
developed by Danheiser,46 anti-2-oxetanones are prefer-
ably formed with excellent stereocontrol, as we have
shown previously.47
Our calculations also predict that anti cycloadducts are
more stable than their syn isomers, as shown in Table
1. It is noteworthy that the minimum energy conformers
in both diastereomers are quite different (see Figure 5).
In the anticycloadducts, the two C-O bonds are in an
antiperiplanar relationship and the φ ) O8-C6-C4-O1
The origins of the stereochemical outcome in this
reaction can be rationalized as the summation of steric,
electrostatic, and stereoelectronic effects. The latter ones
can in turn be understood by examining the substituent
effects on the MOs of the parent system. Figure 6 shows
the canonical FMOs of the transition structure associated
with the simplest case, namely [2 + 2] cycloaddition
(46) (a) Danheiser, R. L.; Nowick, J . S. J . Org. Chem. 1991, 56, 1176.
(b) Danheiser, R. L.; Choi, Y. M.; Menichincheri, M.; Stoner, E. J . J .
Org. Chem. 1993, 58, 322.
(47) Arrastia, I.; Lecea, B.; Cossı´o, F. P. Tetrahedron Lett. 1996, 37,
245.
(48) (a) Bu¨rgi, H. B.; Dunitz, J . D.; Scheffer, E. J . Am. Chem. Soc.
1973, 95, 5065. (b) Bu¨rgi, H. B.; Du¨nitz, J . D.; Lehn, J . M. Tetrahedron
1974, 30, 1563.
(49) Gawley, R. E.; Aube´, J . Principles of Asymmetric Synthesis;
Pergamon: Oxford, 1996; pp 121-160 and references therein.

Pericyclic Alternative to the Aldol Reaction
J . Org. Chem., Vol. 63, No. 15, 1998 5221
F igu r e 6. Computer plot of the FMO’s of the transition
structure corresponding to the [2 + 2] cycloaddition between
ketene and formaldehyde, calculated at the B3LYP/6-31G*.
Contour plot has been fixed at 0.065 e/bohr3.
F igu r e 5. Ball and stick representations of the lowest energy
conformations of syn- and anti-2-oxetanones 3a ,c,e. Bond
distances and angles are given in Å and deg, respectively. φ is
the O8-C6-C4-O1 dihedral angle and is reported as an
absolute value.
therefore, the C3‚‚‚C4 bond can act as a very efficient
donor. Similarly, the energy of the LUMO of this TS is
-1.43 eV at the B3LYP/6-31G* level, and therefore, the
C3 atom constitutes a strong acceptor. Consequently, the
C3 atom will be very sensitive to the effect of donating
substituent of the aldehyde moiety and the C3-C4 bond
being formed will interact efficiently with σ-acceptors at
C3.
To quantify the above-mentioned stereoelectronic ef-
fects involved in syn - and a n ti-TSa , we have analyzed
the most relevant two-electron interactions between the
corresponding NBOs. In some cases, the NPA assigns
between ketene and formaldehyde.50 As it can be seen,
both FMOs result mainly from the two possible combina-
tions between the LUMO of formaldehyde (basically a
π* orbital) and the distorted HOMO of ketene (the 2b1
CO
MO within the C2v symmetry). The HOMO of the parent
TS is dominated by the distorted σ interaction between
C3 and C4. This MO has an energy of -6.83 eV, and
(50) This TS has been calculated previously at other theoretical
levels (see refs 16-18). Its main geometrical features at the B3LYP/
6-31G* level are the following (bond distances are given in Å and
numbers in parentheses correspond to the bond indices):
the acceptor character of C3 to the π* localized orbital.
3,4
This is consistent with the high participation of this
orbital in the FMOs represented in Figure 6. We have
separated in two groups the stereoelectronic effects
between the chiral group XC, namely the (S)-1-methoxy-
ethyl group, and the ring being formed. In the first one,
we have included the φX f φ* interactions between
C
q
filled NBOs of XC and vacant NBOs localized on the ring
being formed. In the second group, we have collected the

5222 J . Org. Chem., Vol. 63, No. 15, 1998
Lecea et al.
In a preliminary communication,21 we have reported
that the [2 + 2] cycloaddition between ketenes and
aldehydes can be efficiently promoted using 5 M solutions
of lithium perchlorate in diethyl ether (5 M LPDE). This
polar solvent has proved to be useful in [4 + 2] cycload-
dition reactions and other processes.52 Therefore, we
performed the reaction between dichloroacetyl chloride
and 2b in 5 M LPDE and in the presence of 1.1 equiv of
triethylamine. Under these conditions, cycloadduct syn -
3b was obtained as a single stereoisomer in 70% yield
after 30 min of reaction at 0-5 °C. This cycloadduct
exhibited identical spectroscopic properties with respect
to those previously found for syn -3b. Therefore, the
sense of stereocontrol in dichloromethane and in 5 M
LPDE was found to be the same.
Sch em e 3
The origins of the efficiency of 5 M LPDE in promoting
thermally allowed cycloadditions have been the subject
of a vivid debate. Thus, several authors52 have attributed
this effect to the high internal pressure generated by the
solvent as well as to other nonspecific solvent effects.
Others investigators,53 however, explain the observed
acceleration in terms of Lewis acid catalysis by the
lithium cation. In this respect, we have recently re-
ported54 that in several Diels-Alder reactions between
nitroalkenes and dienes 5 M LPDE (and 4 M lithium
perchlorate in nitromethane as well) promotes formation
of the cycloadducts that are obtained under purely
thermal conditions. This result is significant, since it has
been reported that nitroalkenes can act as dienes if Lewis
acids are present, to yield cyclic nitronates, which in turn
can suffer a variety of interesting transformations.55
Despite these precedents, Lewis acid complexation of
the lithium cation cannot be in principle ruled out in the
[2 + 2] cycloaddition between carbonyl compounds and
ketenes. Although complexation of a solvated lithium
cation to a carbonyl compound is not a favored process
in an ethereal solvent,56 the resulting chelated structure
can be more reactive than the uncomplexed carbonyl
compound. Therefore, we decided to explore computa-
tionally the reaction between 2a complexed with a
lithium cation57 and dichloroketene 1a . The two transi-
tion structures syn -TS′a and a n ti-TS′a are shown in
Figure 4, and their geometrical and energetic features
are reported in Table 1 and in Table 1 of the Supporting
Information. As can be observed, in both saddle points
the lithium atom is coordinated to the O1 and O8 atoms,
similar to the chelated structure H depicted in Figure 3.
Nonchelated structures spontaneously converged to the
chelated ones upon optimization.
most significant interactions of type φq f φ* , namely
XC
between a filled NBO of the ring being formed and an
antibonding NBO associated with the chiral group. The
result of this analysis is reported in Table 2. From the
data included in Table 2, it is clear that in both syn - and
a n ti-TSa the former type of interaction is the predomi-
nant one. Thus, in syn -TSa the main term is the σ6,8
f
π* contribution, whereas in a n ti-TSa the most impor-
1,4
tant term corresponds to the σ6,7 f π* interaction.
1,4
Given that the C-H bond is a better donor than the C-O
bond,39 the ∆E(2) terms associated with the σ6,7 f π*
1,4
interaction are much more important than those corre-
sponding to the σ6,8 f π* donation (see Figure 4).
1,4
However, syn -TSa is calculated to be of lower energy
than a n ti-TSa (vide supra). It can be therefore con-
cluded that two-electron stereoelectronic effects do not
determine the stereocontrol of the reaction. Given that
in these transition structures the [π2s + (π2s
+ π2s)]
geometry imposes a significant torsion around the C2-
C3 bond, in a n ti-TSa there is a steric repulsive interac-
tion between the methyl group and the endo chlorine
atom, whereas in syn -TSa the methyl group is at the
opposite side of the endo chlorine atom (see Figure 4).
In addition, there is a repulsive Coulombic interaction
between the exo chlorine and the O8 atom in a n ti-TSa .
The superposition of all these effects results in the
preferential formation of syn -3a under kinetic control.
To verify the validity of this prediction, we carried out
the reaction between 1a , generated in situ by treatment
of dichloroacetyl chloride with triethylamine, and (S)-2-
(benzyloxy)propionaldehyde 2b. After 30 min of reaction
at 0-5 °C, cycloadduct 3b was isolated in 70% yield as a
single stereoisomer after purification by flash chroma-
tography (see Scheme 3). Inspection of the crude reaction
1
Our results indicate that in the case of the syn
diastereomer lithium chelation induces a lowering in the
activation energy of 3.24 kcal/mol at the B3LYP/6-31G*
+ ∆ZPVE level. At the MP2/6-31G*//HF/6-31G* +
∆ZPVE level, we have obtained a lowering of 14.68 kcal/
mol, which is probably too large.16 In any case, the
former value is in qualitative agreement with the ac-
celeration experimentally observed by us. In addition,
the energy difference between both transition structures
mixture by H NMR revealed that only one isomer was
formed in the course of the [2 + 2] cycloaddition. To
verify the stereochemistry of the cycloadduct, the benzyl
group was removed by treatment with TiCl4, followed by
transesterification to form the γ-lactone 4,47 which was
further transformed into its solid 4-nitrobenzoyl ester
derivative 5. Both butyrolactones 4 and 5 exhibited low
coupling constants between the protons at C4 and C5 (3.2
and 3.4 Hz, respectively, see Experimental section),
together with a low nuclear Overhauser effect (NOE) over
the proton at C4 when the methyl group was presatu-
rated. These spectral data were assigned to a cis
configuration for these products51 and, therefore, to a syn
configuration for the starting propiolactone 3b.
(52) (a) Grieco, P. A.; Nunes, J . J .; Gaul, M. D. J . Am. Chem. Soc.
1990, 112, 4595. (b) Waldmann, H. Angew. Chem., Int. Ed. Engl. 1991,
30, 1306. (c) Grieco, P. A.; Handy, S. T.; Beck, J . P. Tetrahedron Lett.
1994, 35, 2663. (d) Grieco, P. A. Aldrichim. Acta 1991, 24, 59.
(53) Forman, M. A.; Dailey, W. P. J . Am. Chem. Soc. 1991, 113, 2761.
(54) Ayerbe, M.; Coss´ıo, F. P. Tetrahedron Lett. 1995, 36, 4447.
(55) Denmark, S. E.; Thorarensen, A. Chem. Rev. 1996, 96, 137.
(56) Chen, X.; Hortelano, E. R.; Eliel, E. L.; Frye, S. V. J . Am. Chem.
Soc. 1992, 114, 1778.
(51) J aime, C.; Segura, C.; Dinare´s, I.; Font, J . J . Org. Chem. 1993,
58, 154.

Pericyclic Alternative to the Aldol Reaction
J . Org. Chem., Vol. 63, No. 15, 1998 5223
is calculated to be 6.03 and 5.39 kcal/mol in favor of the
syn cycloadduct at the MP2/6-31G*//HF/6-31G* + ∆ZPVE
and B3LYP/6-31G* + ∆ZPVE levels, respectively (see the
∆∆E′a values in Table 1). Therefore, these calculations
predict the same sense of stereoselection observed for the
1a + 2a f 3a process.
Both syn - and a n ti-TS′a diastereomeric structures
show important differences with respect to their non-
lithiated analogues (see Figure 4 and Table 1 of the
Supporting Information). Thus, the C3‚‚‚C4 bond is now
largely more developed than the O1‚‚‚C2 bond. For
example, in syn -TS′a , the respective bond distances are
calculated to be 1.774 and 2.573 Å at the B3LYP/6-31G*
level, which correspond to Wiberg NBA bond indices of
0.654 and 0.065, respectively. Therefore, the reaction can
be described as a two-stage process, in which the main
interaction in the transition structure corresponds to a
nucleophilic attack of the ketene over the chelated (and
more electrophilic) aldehyde. The RN value is found to
be 105.3° at the B3LYP/6-31G* level, which is now in
line with the Bu¨rgi-Dunitz trajectory. In summary, the
stereochemistry of the lithium-assisted version of the
reaction is what one would expect from the cyclic Cram
model. Thus, in syn -TS′a , the attack of the ketene takes
place through the less hindered side of the chelated
aldehyde, whereas in a n ti-TS′a the attack occurs through
the side of the methyl group, thus resulting in a desta-
bilizing steric interaction between the methyl group and
a chlorine atom, as well as a distortion in the cyclic
structure of the chelated aldehyde (see Figure 4). There-
fore, in this case, a more reactive species (H in Figure 3)
is actually more selective (compare the ∆∆Ea and ∆∆E′a
values in Table 1), as is predicted in the cyclic Cram
model.56,58 It is important to note that in these calcula-
tions with saddle points syn - and a n ti-TS′a solvation of
the lithium cation by the ethereal solvent has not been
considered. However, it is expected that the effect of this
simplification is similar for both diastereomeric struc-
tures and therefore will not affect significantly their
relative energies. In summary, we can conclude that if
lithium perchlorate accelerates the cycloaddition rate via
ion-molecule complexation the sense of stereoselection
is the same that observed for the purely thermal cycload-
dition. Therefore, both reaction paths are in agreement
with our experimental findings.
F igu r e 7. Ball and stick representations of diastereomeric
saddle points TSc and TS′c. Numbers correspond to the
Wiberg bond indices.
2a to yield 2-oxetanones syn - and a n ti-3c. The chief
geometric features of the transition structures syn - and
a n ti-TSc are reported in Table 1 of the Supporting
Information and in Figure 7. Given the lower electro-
philicity of ketene with respect to dichloroketene, the
O1-C2 and C3-C4 bond distances in these TSs are
larger and shorter than in the preceding case, respec-
tively. The B3LYP/6-31G* method predicts a higher
degree of advancement in the former interaction, par-
ticularly in the case of syn -TSc.
Since the interaction of the endo-hydrogen atoms of
ketene and the substituent of the aldehyde is lower than
in the preceding case, there is a remarkable freedom for
the chiral group of 2a to adopt the optimal conformation.
Thus, in syn -TSc the value of φ is almost 180°, thus
indicating that the C6-O8 and C3‚‚‚C4 bonds are per-
fectly antiperiplanar to each other (see Table 1 of the
Supporting Information). Inclusion of electron correla-
tion in this saddle point induces an enlargement of the
C3‚‚‚C4 bond distance, the energy associated to the π1,4
The next step of our study was to explore computa-
tionally the behavior of ketene 1b itself toward aldehyde
(57) One reviewer has suggested that lithium complexation could
also take place through the oxygen atom of the ketene. Attempts to
locate saddle points in which the lithium was complexed to O5
invariably converged to the reported TSs upon optimization. On the
other hand, the ∆E value of the following isodesmic equation
f σ* interaction being only 1.87 kcal/mol at the
6,8
B3LYP/6-31G* level (see Table 2). The highest stabiliz-
ing donation for this saddle point is 4.42 kcal/mol, a value
that is associated with the reverse σ6,8 f π* donation.
1,4
In the case of a n ti-TSc, the main two electron interaction
is the σ3,4 f σ* donation, with a second-order energy
6,8
was found to be -38.2 kcal/mol at the B3LYP/6-31G* + ∆ZPVE level,
thus suggesting preferential aldehyde complexation also in the sub-
strates. This is in agreement with the well-known participation of the
resonance form II in ketenes (see ref 2, p 33):
value of 3.39 kcal/mol at the B3LYP/6-31G* level. Since
in a n ti-TSc the C3‚‚‚C4 bond distance is shorter than
in syn -TSc, the C3‚‚‚C4 and C6-C8 bonds cannot adopt
an antiperiplanar conformation, and therefore, the mag-
nitude of the σ3,4 f σ* interaction, although signifi-
6,8
cant, is not optimal.
The global result of all these combined interactions is
that a n ti-TSc is calculated to be 3.32 and 2.26 kcal/mol
(58) Cram, D. J .; Kopecky, K. R. J . Am. Chem. Soc. 1959, 81, 2748.

5224 J . Org. Chem., Vol. 63, No. 15, 1998
Lecea et al.
Sch em e 4
tion of 2-oxetanone 3d was observed after 4 h of reaction
at room temperature. Compound 3d was unequivocally
characterized by a characteristic carbonyl resonance in
1
its IR spectrum at 1823 cm-1 and by the 300 MHz H
NMR spectrum of the crude reaction mixture (see Figure
2 of the Supporting Information). Treatment of syn -3d
with magnesium dibromide gave butyrolactone 6 in a
55% overall yield from 2b. Compound 6 was treated with
TiCl4 and then with diisopropylethylamine at -20 °C for
5 h to give (+)-angelica lactone 7 in 42% yield. The
structure of this naturally occurring compound was
verified from the reported physical and spectroscopic
data.60 Therefore, the reaction between aldehyde 2b and
ketene 1b was proved to be feasible, provided that the
reaction was carried out using 5 M LPDE as solvent.
Under these conditions, a single diastereomer was de-
tected, and its structure was assigned to syn -3d on the
basis of the stereochemistry of compound 6 and is again
in agreement with the computational results. On the
other hand, our method provides a straightforward
synthesis of (+)-angelica lactone 7 with total stereocon-
trol. It is noteworthy that in this case the enolate
approach does not yield satisfactory results. For in-
stance, reaction between 2b and lithium enolate derived
from tert-butyl acetate yields the two possible syn and
anti â-hydroxy esters with virtually no stereoselection.61
We have also studied computationally the interaction
between dimethylketene 1c and (S)-2-methoxypropanal
2a . The two diastereomeric transition structures, syn -
and a n ti-TSe, are depicted in Figure 8. The shape of
these saddle points closely resembles those already
described for the 1b + 2a f 3c process. The character-
istic angles RN, ê, and φ are also similar for the two
diastereomeric saddle points. In this case, the low
electrophilicity of the ketene results in an enlargement
of the activation energy. Thus, the activation energy
computed for the 1c + 2a f syn -3e process is found to
be of 4.47 and 8.61 kcal/mol higher than that computed
for the 1a + 2a f syn -3a process at the MP2/6-31G*//
HF/6-31G* + ∆ZPVE and B3LYP/6-31G* + ∆ZPVE
levels, respectively. This means that the reactions using
dimethylketene should be even more difficult than those
using ketene. On the other hand, the interaction be-
tween the endo methyl group and the (S)-1-methoxyethyl
group is larger than in the preceding case, and therefore,
the difference in energy between the two diastereomeric
saddle points is larger than between syn - and a n ti-TSc.
Thus, syn -TSe is predicted to be 3.03 and 1.71 kcal/mol
more stable than a n ti-TSe at the MP2/6-31G*//HF/6-
31G* + ∆ZPVE and B3LYP/6-31G* + ∆ZPVE levels,
respectively. In the case of the lithium-assisted saddle
points, the ∆∆E′a values are 2.98 and 1.08 kcal/mol,
respectively, both of them in favor of the formation of
syn -3e.
less stable than syn -TSc at MP2/6-31G*//HF/6-31G* +
∆ZPVE and B3LYP/6-31G* + ∆ZPVE levels, respec-
tively. It is also noteworthy that the calculated activation
energies are higher than in the preceding case. Thus,
the ∆Ea value for the 1b + 2a f syn -3c process is
calculated to be 7.34 kcal/mol larger than that found for
the 1a + 2a f syn -3a process at the B3LYP/6-31G* +
∆ZPVE level (see Table 1). Therefore, the 1b + 2a f
3c transformation is predicted to be significantly more
difficult than the former one, although the computed ∆Ea
values are probably too low at these theoretical levels59
(vide infra).
We have also calculated the lithium-assisted version
of the reaction between ketene and (S)-2-methoxypropa-
nal. We have located two diastereomeric saddle points
syn -TS′c and a n ti-TS′c associated with the reaction
coordinate. Their chief geometrical features are quite
similar to those found for the diastereomeric TS′a saddle
points. The C3‚‚‚C4 bond is again substantially more
developed than the O1‚‚‚O2 bond (see the corresponding
bond orders in Figure 7). The RN values of syn - and a n ti-
TS′c are of ca. 107°, and once again, their shape is
concordant with the cyclic (chelated) Cram model. At the
MP2/6-31G*//HF/6-31G* + ∆ZPVE level, it is predicted
that syn -TS′c is 1.20 kcal/mol lower in energy than a n ti-
TS′c. This should correspond to a ca. 89:11 ratio between
syn - and a n ti-3c. At the B3LYP/6-31G* level, the
difference between syn - and a n ti-TS′c is lower (see Table
1). We have found that inclusion of solvent effects does
not modify the sense of stereoselection. Following a
suggestion made by one reviewer, we have included two
water molecules around the lithium cation to model the
ethereal solvent, and we have computed the relative
energies. At the B3LYP(SCIPCM)/6-31G*//HF/6-31G* +
∆ZPVE level, we have found that syn -TS′c‚2H2O is 1.06
kcal/mol more stable than a n ti-TS′c‚2H2O (see Figure
1 and Table 2 of the Supporting Information).
We have studied experimentally the cycloaddition
between ketene 1b, generated in situ from acetyl chloride
and aldehyde 2b (Scheme 4). After 12 h of reaction at
room temperature, no cycloaddition products were de-
tected from inspection of the reaction mixture. This
result seems to support that the ∆Ea values computed
for this reaction are systematically too low. In contrast,
when the reaction was performed in 5 M LPDE, forma-
The verification of these predictions is shown in
Scheme 5. Reaction of 2-methylpropionyl chloride with
2b in the presence of triethylamine and 5 M LPDE gave
cycloadduct syn -3f as the only stereoisomer detectable
by NMR analysis on the crude reaction mixture. Depro-
(60) (a) Ortun˜o, R. M.; Alonso, D.; Cardellach, J .; Font, J . Tetrahe-
dron 1987, 43, 2191. For the synthesis of the (-)-(R)-angelica lactone
from D-ribonolactone, see: (b) Camps, P.; Cardellach, J .; Corbera, J .;
Font, J .; Ortun˜o, R. M.; Ponsat´ı, O. Tetrahedron 1983, 39, 395. (c) Chen,
S.-Y.; J oullie´, M. J . Org. Chem. 1984, 49, 2168.
(61) Reetz, M. T. Angew. Chem., Int. Ed. Engl. 1984, 23, 1984 and
references therein. See also: Braun, M. Angew. Chem., Int. Ed. Engl.
1987, 26, 24.
(59) As we have previously reported, in this reaction the magnitudes
of the activation energies are underestimated at the MP2/6-31G* level
(see ref 16). Probably, this deficiency can be extended to the B3LYP/
6-31G* level. For recent studies on this problem, see: (a) Fabian, W.
M. F.; Kollenz, G. J . Org. Chem. 1997, 62, 8497. (b) Durant, J . L. Chem.
Phys. Lett. 1996, 256, 595. (c) Glukhovtsev, M. N.; Bach, R. D.; Pross,
A.; Radom, L. Chem. Phys. Lett. 1996, 260, 558.

Pericyclic Alternative to the Aldol Reaction
J . Org. Chem., Vol. 63, No. 15, 1998 5225
hand, reaction between 2-pyridyl isobutanethioate and
2b in the presence of LDA under the conditions reported
by Danheiser gave 2-oxetanone a n ti-3f in 61% yield.
When the same reaction was performed using 2b and
phenyl isobutanethioate, the yield of a n ti-3f was only
39%. This compound, when subjected to the deprotec-
tion-transesterification sequence (see Scheme 5), yielded
γ-lactone a n ti-8 in 79% yield after purification by flash
chromatography. The coupling constant between the
C4-H and C5-H methinic protons was found to be 7.9 Hz,
a value again in excellent agreement with that obtained
by J aime51 (J 4,5 ) 7.4 Hz). Therefore, the enolate and [2
+ 2] approaches yield opposite stereochemistries, with
complete stereocontrol in both cases, thus showing the
complementary character of both methodologies.
Con clu sion s
From the combined computational and experimental
results reported in this study the following conclusions
can be drawn:
(i) The [2 + 2] cycloaddition between C2v symmetric
ketenes and homochiral aldehydes takes place through
a concerted [π2s + (π2s + 2s)] mechanism.
π
(ii) The transition structures associated with these
processes are highly asynchronous. The degree of ad-
vancement of the σ bonds being formed is sensitive to
the method used in the optimization of the saddle point.
A correlated method such as B3LYP predicts that the O1‚
‚‚C2 bond is more advanced than the C3‚‚‚C4 bond.
(iii) The [π2s + (π2s
+ π2s)] mechanism favors the
formation of syn cycloadducts via transition structures
whose geometry can be schematized as depicted in Figure
9. The origins of the stereoselection lie in the interaction
between the endo substituent of the ketene and the chiral
template. In the syn transition structure, this interaction
is minimized and the large group is antiperiplanar to the
C3‚‚‚C4 bond being formed. In the anti transition
structure, there is a destabilizing interaction between the
endo substituent and the medium-size substituent of the
carbonyl compound. The magnitude of these interactions
is also controlled by the C3‚‚‚C4 bond distance, which in
turn is related to the nucleophilic or electrophilic char-
acter of the ketene. It is noteworthy that the geometries
of these syn saddle points closely resemble the 1,3-allylic
strain model proposed by Kishi62,64 for the cis hydroxyl-
ation of chiral allylic ethers. Similarly, the anti saddle
points are related to those proposed by Vedejs63,64 for the
same reaction.
F igu r e 8. Ball and stick representations of diastereomeric
saddle points TSe and TS′e. Numbers correspond to the
Wiberg bond indices.
Sch em e 5
(iv) In the lithium-assisted version of the reaction, the
aldehyde and the lithium cation can form a chelated ion-
molecule complex. The main features of this complex are
kept in the corresponding transition structure, thus
leading again to the preferential formation of the syn
cycloadduct as schematized in Figure 9. The features of
these transition structures are those expected from the
cyclic Cram model, in which the ketene attacks the
chelated lithium-aldehyde complex through its less
hindered side.
tection of the benzyloxy group followed by in situ trans-
esterification gave the butyrolactone syn -8 in excellent
yield. The syn-stereochemistry of this compound was
unequivocally established on the basis of the coupling
constant between the C4-H and the C5-H methinic
protons. The experimental value of this coupling con-
stant is 3.7 Hz, a value in excellent agreement with that
predicted by J aime et al.51 (J 4,5 ) 3.8 Hz). On the other
(v) Calculations predict that thermodynamic control
favors preferential formation of anti cycloadducts.
(62) Cha, J . K.; Christ, W. J .; Kishi, Y. Tetrahedron 1984, 40, 2247.
(63) Vedejs, E.; McClure, C. K. J . Am. Chem. Soc. 1986, 108, 1094.
(64) For a recent computational study on the stereochemistry of this
reaction, see: Haller, J .; Strassner, T.; Houk, K. N. J . Am. Chem. Soc.
1997, 119, 8031.

5226 J . Org. Chem., Vol. 63, No. 15, 1998
Lecea et al.
chromatography purifications were performed using silica gel
(230-400 mesh) and 1:20 AcOEt/hexanes mixtures.
The glassware used in the cycloaddition reactions was oven
or flame dried and assembled under a slightly positive argon
pressure. Melting points are uncorrected. Chemical shifts in
the 1H NMR spectra are reported as δ values (ppm) relative
to internal tetramethylsilane (δ ) 0.00 ppm). The NOE
experiments were run at 300 MHz by preirradiating the
desired signals for 15 s with the decoupler channel turned on
at 15 db below 1 W and acquiring the spectrum with the
decoupler turned off. The enhancements were directly mea-
sured by integration of the signals resulting from the respec-
tive difference spectra.
Gen er a l P r oced u r e for Cycloa d d ition Rea ction s in 5
M LP DE. Triethylamine (1.1 equiv) was added at 0 °C (ice-
water bath) under argon atmosphere over a vigorously stirred
solution of the corresponding acyl chloride (2 or 4 equiv, see
the Supporting Information) and (S)-2-(benzyloxy)propanal 2b
(0.16 g, 1 mmol) in 2 mL of 5 M LPDE. After being stirred at
room temperature or 0-5 °C for a certain time, the reaction
mixture was diluted with dichloromethane (15 mL), and the
resulting solution was washed successively with water, 1 N
HCl, and a saturated solution of sodium bicarbonate. The
organic layer was dried, the solvent was evaporated under
reduced pressure, and the resulting residue was purified by
flash chromatography or bulb-to-bulb distillation.
(R )-4-[(S )-1-(B e n zy lox y)e t h yl]-3,3-d im e t h y l-2-o x e -
ta n on e (syn -3f) was obtained from isopropyl chloride (1 equiv)
following the general procedure. The crude product was
purified by flash chromatography as a colorless oil: yield 58%;
1
[R]25 +26.0 (c ) 0.4, CH2Cl2); IR (film) 1825 cm-1; H NMR
D
(CDCl3) δ 7.38-7.24 (m, 5H), 4.68 (d, 1H, J ) 11.8 Hz), 4.64
(d, 1H, J ) 11.8 Hz), 4.15 (d, 1H, J ) 8.6 Hz), 3.74 (dq, 1H, J
) 6.2 Hz, J ′ ) 8.6 Hz), 1.42 (s, 3H), 1.28 (s, 3H), 1.19 (d, 3H,
J ) 6.2 Hz); 13C NMR(CDCl3) δ 174.9, 138.1, 128.4, 127.6, 85.7,
73.4, 71.7, 22.8, 16.9, 16.4.
F igu r e 9. Schematic representation of the possible transition
structures leading to the formation of syn- and anti-2-oxe-
tanones. S, M, and L stand for the small, medium, and large
substituents.
(R )-4-[(S )-1-(B e n zy lox y )e t h yl]-3,3-d im e t h y l-2-o x e -
ta n on e (a n ti-3f). A cooled solution (0-5 °C) in THF (10 mL)
of diisopropylamine (0.31 mL, 2.2 mmol) and a catalytic
amount of 1,10 phenantroline was treated with a few drops of
n-butyllithium until the solution had a dark red color. Then,
a 1.6 M solution of n-butyllithium in hexanes (1.38 mL, 2.2
mmol) was added via syringe over 2 min. After 15 min, the
resulting mixture was cooled to -78 °C, and 2-pyridyl isobu-
tanethioate (0.36 g, 2 mmol) or phenyl isobutanethioate (0.36
g, 2 mmol) was added dropwise via syringe over 5 min. After
30 min, a solution of (S)-(benzyloxy)propanal (0.33 g, 2 mmol)
in THF (5 mL) was added dropwise over 20 min via a cannula
that was cooled at -78 °C (dry ice-acetone bath) by means of
a glass tube with rubber septum. The reaction mixture was
stirred at -78 °C for 30 min and at 0-5 °C for 75 min and
then allowed to reach room temperature. Half-saturated
NH4Cl solution (10 mL) was then added, and the resulting
mixture was partitioned between 15 mL of water and 15 mL
of diethyl ether. The organic layer was washed with 10%
K2CO3 solution (2 × 25 mL) and a saturated solution of NaCl
(25 mL) and dried (MgSO4), and the solvent was evaporated
under reduced pressure to yield the crude product, which was
purified by flash chromatography: yields 39% (with phenyl
(vi) The predictions made on the basis of our compu-
tational studies have been verified experimentally. Re-
action between (S)-2-(benzyloxy)propanal and dichlo-
roketene, ketene, and dimethylketene leads to the
corresponding syn-2-oxetanones with virtually complete
stereocontrol. It is also found that the cycloaddition is
facilitated when the reaction is carried out using 5 M
LPDE as solvent.
(vii) The [2 + 2] cycloaddition between carbonyl
compounds and ketenes can be considered as a pericyclic
alternative to the nonchelation-controlled aldol reaction,
with a complementary stereochemical outcome.
Exp er im en ta l Section
Gen er a l Meth od s. Commercially available compounds
were used as such without further purification. Solvents were
anhydrized and distilled according to standard protocols.65
Diethyl ether was further dried and distilled over sodium-
acetophenone. Lithium perchlorate, purchased from Aldrich
(A.C.S. reagent), was dried at 160 °C and 0.05 mmHg with
silica gel for 4 h in a Ku¨gelrohr apparatus prior to use. We
have found that this treatment is essential to carry out
successful and reproducible experiments (CAUTION: perchlo-
rates are potentially explosive compounds. Under the condi-
tions described above, lithium perchlorate can be manipulated
safely). Acyl chlorides66 and (S)-(benzyloxy)propanal 2b67 were
prepared by means of previously reported procedures. Flash
isobutanethioate) and 61% (with 2-pyridyl isobutanethioate).
1
[R]25 +26.5 (c ) 1.3, CH2Cl2); IR (film) 1825 cm-1; H NMR
D
(CDCl3) δ 7.36-7.26 (m, 5H), 4.66 (d, 1H, J ) 11.2 Hz), 4.43
(d, 1H, J ) 11.2 Hz), 4.02 (d, 1H, J ) 8.7 Hz), 3.78 (dq, 1H, J
) 8.7 Hz, J ′ ) 6.0 Hz), 1.44 (s, 3H), 1.35 (d, 3H, J ) 6.0 Hz),
1.32 (s, 3H); 13C NMR (CDCl3) δ 175.2, 137.6, 128.5, 127.9,
104.4, 83.3, 73.2, 70.6, 23.1, 16.7, 16.1.
Gen er a l P r oced u r e for t h e Con ver sion of 2-Oxe-
ta n on es (3) in to Bu tyr ola cton es. Titanium tetrachloride
(1 M solution in CH2Cl2, 1 mL) was added over a solution of
the corresponding 2-oxetanone 3 (1 mmol) in dichloromethane
(2.5 mL). The reaction mixture was stirred for a certain time
at different temperatures and was monitored by thin-layer
chromatography. Water (7 mL) was added, and the resulting
mixture was extracted with CH2Cl2 (3 × 7 mL). The organic
layers were combined and filtered twice through a Celite pad.
(65) Perrin, D. D.; Armarego, W. L. F. Purification of Laboratory
Chemicals; Pergamon: Oxford, 1988.
(66) Baumgarten, H. E.; Bower, F. A.; Okamoto, T. T. J . Am. Chem.
Soc. 1957, 79, 3145.
(67) Zemribo, R.; Romo, D. Tetrahedron Lett. 1995, 36, 4159.

Pericyclic Alternative to the Aldol Reaction
J . Org. Chem., Vol. 63, No. 15, 1998 5227
The filtrate was dried (MgSO4), and the solvent was evapo-
rated under reduced pressure. The resulting oily residue was
purified by flash chromatography.
romethane, 1 mL) was added at room temperature over a
solution of compound 6 (1 mmol) in dichloromethane (4 mL).
The resulting mixture was stirred at the same temperature
for 30 min. Water was then added, and the aqueous layer was
extracted with dichloromethane (3 × 25 mL). The organic
layers were combined and dried (Mg SO4). Evaporation of the
solvent gave an oily residue that was redissolved in dichlo-
romethane (4 mL). Over this mixture, a solution of diisopropyl
ethylamine (0.4 mL, 2.5 mmol) in dichloromethane (1 mL) was
added at -20 °C. The resulting mixture was stirred for 5 h
at -20 °C. Then, citric acid (10% solution, 20 mL) was added.
The organic layer was separated and washed with NaHCO3
(saturated solution, 20 mL) and water (20 mL). The organic
layer was dried (MgSO4), and evaporation of the solvent under
reduced pressure gave an oily residue that was purified by
distillation in a Ku¨gelrohr apparatus: yield 42%; bp 100 °C/
(4S,5S)-4-Hyd r oxy-3,3,5-tr im eth yl-2-oxola n on e (syn -8)
was prepared from syn -3f following the general procedure. The
reaction was conducted at -78 °C for 30 min. The product
was obtained as a colorless oil: yield 96%; [R]25 -56.0 (c )
D
0.62, CH2Cl2); IR (film) 1750 cm-1
;
1H NMR (CDCl3) δ 4.68
(dq, 1H, J ) 3.7 Hz, J ′ ) 6.5 Hz), 3.92 (d, 1H, J ) 3.7 Hz), 2.2
(sb, 1H), 1.26 (s, 6H), 1.42 (d, 3H, J ) 6.5 Hz); 13C NMR (CDCl3)
δ 181.6, 77.4, 77.2, 45.9, 22.8, 17.9, 13.8.
(4R,5S)-4-Hyd r oxy-3,3,5-tr im eth yl-2-oxola n on e (a n ti-
8) was prepared from a n ti-3f following the general procedure.
The reaction was conducted at room temperature for 12 h. The
product was obtained as colorless crystals: yield 79%; mp 70-
71 °C (ethyl acetate-hexanes); [R]25 -46.0 (c ) 0.1, CH2Cl2);
D
IR (film) 1736 cm-1; H NMR (CDCl3) δ 4.24 (dq, 1H, J ) 6.2
1
15 Torr (lit.56a bp 98-100/15 Torr); [R]25 +93.53 (c ) 0.5,
D
Hz, J ′ ) 7.9 Hz), 3.75 (dd, 1H, J ) 5.9 Hz, J ′ ) 7.9 Hz), 2.58
(d, 1H, J ) 6.2 Hz), 1.47 (d, 3H, J ) 6.2 Hz), 1.26 (s, 3H), 1.18
(s, 3H); 13C NMR (CDCl3) δ 180.2, 81.5, 78.0, 43.8, 22.7, 18.2,
17.6. Anal. Calcd for C7H12O3: C, 59.30; H, 8.40; O, 33.29.
Found: C, 59.35; H, 8.38; O, 33.31%.
CH2Cl2) (lit.56a [R]25 +93.83 (c ) 0.5, CHCl3). The spectro-
D
scopic properties of the title compound were coincident with
those previously reported.56a
(4S,5S)-4-(Ben zyloxy)-5-m eth yl-2-oxola n on e (6). Com-
pound syn -3d was prepared from acetyl chloride (4 equiv) and
2b following the general procedure (vide supra). The crude
[2 + 2] cycloadduct was characterized by its spectroscopic
features (see Figure 1 of the Supporting Information). At-
tempts to purify this cycloadduct failed, and so it was used as
such in the following steps. Magnesium dibromide (3.3 g, 18
mmol) was added over a solution of crude syn -3d (9 mmol,
theoretical from 2b) in dichloromethane (18 mL). The result-
ing mixture was stirred for 30 min at room temperature, and
then water (20 mL) was added. The aqueous layer was washed
with dichloromethane (2 × 20 mL). The organic layers were
combined and dried (MgSO4), and the solvent was evaporated
under reduced pressure. Purification by flash chromatography
Ack n ow led gm en t. The present work has been
supported by the Secretar´ıa de Estado de Universidades,
Investigacio´n y Desarrollo (Project PB96-1481), and by
Gobierno Vasco/Eusko J aurlaritza (Project GV 170.215-
EX97/11). A grant from the Gobierno Vasco/Eusko
J aurlaritza to I.A. is gratefully acknowledged. We also
thank the Plan Nacional de I+D (CICYT) and the
CIEMAT for a generous gift of computing time at the
Cray YMP-EL supercomputer.
Su p p or tin g In for m a tion Ava ila ble: Additional experi-
mental details and characterization of compounds syn-3b, 4,
and 5; 1H NMR and 13C NMR spectra of all products described
in the text; tables including Cartesian coordinates, total
energies, and zero-point vibrational energies of all the transi-
tion structures and cycloadducts reported in this paper (37
pages). This material is contained in libraries on microfiche,
immediately follows this article in the microfilm version of the
journal, and can be ordered from the ACS; see any current
masthead page for ordering information.
gave the title product as a colorless oil: yield 55% (overal from
1
2b); [R]25 +26.2 (c ) 1.04, CH2Cl2); IR (film) 1709 cm-1; H
D
NMR (CDCl3) δ 7.35-7.26 (m, 5H), 4.61 (d, 1H, J ) 11.6 Hz),
4.54 (d, 1H, J ) 11.6 Hz), 4.33 (ddd, 1H, J ) 9.1 Hz, J ′ ) 6.1
Hz, J ′′ ) 4.5 Hz), 3.68 (q, 1H, J ) 6.1 Hz), 3.16 (dd, 1H, J )
16.8 Hz, J ′ ) 4.5 Hz), 2.87 (dd, 1H, J ) 16.8 Hz, J ′ ) 9.1 Hz),
1.33 (d, 3H, J ) 6.1 Hz); 13C NMR (CDCl3) δ 176.6, 137.6,
128.4, 127.8, 76.9, 71.3, 51.1, 40.0, 17.2.
(+)-(S)-5-Meth yloxol-3-en -2-on e ((+)-(S)-â-An gelica Lac-
ton e 7). Titanium tetrachloride (1 M solution in dichlo-
J O9806512