Stereoselective cycloaddition and epoxidation of enol ethers by α-peroxy lactone as a function of steric and stereoelectronic effects

Article abstract of DOI:10.1021/jo961645d

The reaction of mono- and dioxy-substituted olefins 2 with dimethyl α-peroxy lactone 1 affords the cycloaddition products 3 and the epoxides 4 with a high degree of stereoretention of the initial olefin configuration. Only for the pyran 2c is the ene product 5c obtained. When the reaction is run in methanol as cosolvent, additionally the trapping products 6 are observed. The S_N2 reaction is found to be highly regioselective in all cases, as displayed by the cycloadducts 3 and the trapping products 6. The preferred reaction mode has been found to be sensitive to steric effects. The product distribution is rationalized in terms of the diastereomeric 1,4-zwitterionic epoxonium intermediates syn- and anti-C, which are proposed to arise from a side-differentiated S_N2 attack of the enol ether double bond on the peroxide bond of the α-peroxy lactone 1 through a perpendicular spiro-configurated transition state geometry. When the α-peroxy lactone 1 approaches the enol ether 2 from the oxy-substituted side, the syn-C epoxonium intermediate is formed, which leads to the epoxide 4 after release of the corresponding α-lactone. The latter oligomerizes to the polyester 8 or is trapped in methanol as the α-methoxy acid 9. On the contrary, the anti-C epoxonium intermediate results by approach of the α-peroxy lactone 1 from the non-oxy-substituted side of the enol ether 2, but the electronic repulsion between the lone pairs of the epoxonium and enol ether oxygens leads by ring opening of the epoxonium species to the coiled 1,6-zwitterion (U conformation). The latter is too short-lived for stereorandomization and closes to the cycloadducts 3 under high retention of the initial enol ether configuration, but is sufficiently long-lived to be trapped in methanol stereoselectively in form of the adducts 6. These unprecedented results in the peroxide-olefin reaction are contrasted with the previously reported α-peroxy lactone 1 oxidation of alkenes. While the enol ethers 2 lead to the cycloadducts 3 with a high degree of stereoretention and the alkenes lead to extensive loss of the initial olefin geometry, for both trapping by methanol in form of the adducts 6 takes place, again with high stereoselectivity for the enol ethers but not for the alkenes. This mechanistic dichotomy requires different intermediates, namely, the epoxonium species C for the enol ethers and the stretched 1,6-dipole (W conformation) A for the alkenes, which both lead to the cycloadducts 3, the former by way of the coiled 1,6-dipole (U conformation) D. For the enol ethers the epoxonium intermediate C is the precursor to the epoxide, while for the alkenes an independent concerted "butterfly" transition state geometry B applies in the epoxidation.

Full text of DOI:10.1021/jo961645d

8432
J . Org. Chem. 1996, 61, 8432-8438
Ster eoselective Cycloa d d ition a n d Ep oxid a tion of En ol Eth er s by
r-P er oxy La cton e a s a F u n ction of Ster ic a n d Ster eoelectr on ic
Effects
Waldemar Adam and Llu´ıs Blancafort*
Institute of Organic Chemistry, University of Wu¨rzburg, Am Hubland, D-97074 Wu¨rzburg, Germany
Received August 26, 1996X
The reaction of mono- and dioxy-substituted olefins 2 with dimethyl R-peroxy lactone 1 affords the
cycloaddition products 3 and the epoxides 4 with a high degree of stereoretention of the initial
olefin configuration. Only for the pyran 2c is the ene product 5c obtained. When the reaction is
run in methanol as cosolvent, additionally the trapping products 6 are observed. The SN2 reaction
is found to be highly regioselective in all cases, as displayed by the cycloadducts 3 and the trapping
products 6. The preferred reaction mode has been found to be sensitive to steric effects. The product
distribution is rationalized in terms of the diastereomeric 1,4-zwitterionic epoxonium intermediates
syn - and a n ti-C, which are proposed to arise from a side-differentiated SN2 attack of the enol
ether double bond on the peroxide bond of the R-peroxy lactone 1 through a perpendicular spiro-
configurated transition state geometry. When the R-peroxy lactone 1 approaches the enol ether 2
from the oxy-substituted side, the syn -C epoxonium intermediate is formed, which leads to the
epoxide 4 after release of the corresponding R-lactone. The latter oligomerizes to the polyester 8
or is trapped in methanol as the R-methoxy acid 9. On the contrary, the a n ti-C epoxonium
intermediate results by approach of the R-peroxy lactone 1 from the non-oxy-substituted side of
the enol ether 2, but the electronic repulsion between the lone pairs of the epoxonium and enol
ether oxygens leads by ring opening of the epoxonium species to the coiled 1,6-zwitterion (U
conformation). The latter is too short-lived for stereorandomization and closes to the cycloadducts
3 under high retention of the initial enol ether configuration, but is sufficiently long-lived to be
trapped in methanol stereoselectively in form of the adducts 6. These unprecedented results in
the peroxide-olefin reaction are contrasted with the previously reported R-peroxy lactone 1 oxidation
of alkenes. While the enol ethers 2 lead to the cycloadducts 3 with a high degree of stereoretention
and the alkenes lead to extensive loss of the initial olefin geometry, for both trapping by methanol
in form of the adducts 6 takes place, again with high stereoselectivity for the enol ethers but not
for the alkenes. This mechanistic dichotomy requires different intermediates, namely, the
epoxonium species C for the enol ethers and the stretched 1,6-dipole (W conformation) A for the
alkenes, which both lead to the cycloadducts 3, the former by way of the coiled 1,6-dipole (U
conformation) D. For the enol ethers the epoxonium intermediate C is the precursor to the epoxide,
while for the alkenes an independent concerted “butterfly” transition state geometry B applies in
the epoxidation.
In tr od u ction
For the epoxidation we postulated a concerted mecha-
nism with a “butterfly” transition state geometry3
B
We have recently reported1 on the oxidation of di-, tri-,
and tetrasubstituted alkenes by the R-peroxy lactone 1
(Scheme 1). As expected, the R-peroxy lactone 1 is
considerably more reactive than the corresponding 1,2-
dioxetanes2 due to the activating polarization of the
peroxide bond by the adjacent carbonyl group. Two
different reaction modes were observed. On the one
hand, cycloadducts and ene products were formed through
the stretched 1,6-dipole (W conformation) A proposed to
arise from SN2 attack of the π nucleophile on the peroxide
bond of R-peroxy lactone 1, which is responsible for the
extensive loss of the initial alkene geometry in the
cycloaddition. On the other hand, more sterically en-
cumbered alkenes gave epoxidation. The intervention of
the 1,6-dipole is mandatory in view of the loss of stereo-
chemistry in the cycloaddition, the formation of ene
products, and the trapping in the presence of methanol.
(Scheme 1) rather than a stepwise pathway through a
bona fide 1,4-dipolar epoxonium intermediate.
To gain further mechanistic insight into the complexi-
ties and intricacies of the peroxide-olefin reaction,4 we
decided to examine the R-peroxy lactone 1 oxidation of
the considerably more electron-rich enol ethers 2a -h . We
* To whom correspondence should be addressed. Fax: +49 931
8884756. E-mail: adam@chemie.uni-wuerzburg.de.
X Abstract published in Advance ACS Abstracts, November 1, 1996.
(1) Adam, W.; Blancafort, L. J . Am. Chem. Soc. 1996, 118, 4778-
4787.
(2) Adam, W.; Andler, S.; Heil, M. Angew. Chem., Int. Ed. Engl.
1991, 30, 1365-1366.
anticipated that the higher nucleophilicity of the enol
(3) (a) Bartlett, P. D. Rec. Chem. Progr. 1950, 2, 47-51. (b) Baum-
stark, A. L.; McCloskey, C. J . Tetrahedron Lett. 1987, 28, 3311-3314.
S0022-3263(96)01645-3 CCC: $12.00 © 1996 American Chemical Society

Cycloaddition and Epoxidation of Enol Ethers
J . Org. Chem., Vol. 61, No. 24, 1996 8433
Sch em e 1. Mech a n ism for th e Rea ction of r-P er oxy La cton e 1 w ith Alk en es in th e Absen ce a n d P r esen ce
of Meth a n ol
ethers 2 would favor the heterolysis of the peroxide bond,
for which purpose we decided to compare the monooxy-
substituted olefins 2a -c with the dioxy-substituted
derivatives 2d -h . Moreover, the Z/E isomer pairs 2b
and 2d were chosen to assess the stereochemical course
of the cycloaddition, while the dioxenes 2e-h were
selected to study the influence on the product distribution
of steric effects on the lateral sides of the double bond of
the π nucleophile, by comparison of the methyl-substi-
tuted with the hydrogen-substituted enol ethers 2e and
2f and the flattening effect of the benzo substitution in
substrates 2g and 2h . Furthermore, the methyl-substi-
tuted derivatives 2b,f,h should be particularly prone for
the ene reaction through proton abstraction. The unsym-
metrical enol ethers 2a -c should provide information on
the regioselectivity of the oxidation. Analogous to the
alkene nucleophiles, trapping experiments in methanol
should allow us to probe the participation of any polar
intermediate. Herein we present unprecedented results
in the reaction of the R-peroxy lactone 1 with enol ethers
2, which disclose marked differences in their mechanistic
features compared with those of the alkenes. The most
salient conclusion is that, in contrast to the previously
studied alkenes,1 for the enol ethers the intervention of
the epoxonium intermediate C best rationalizes the
initial encounter with the R-peroxy lactone 1.
obtained in the reaction with enol ethers 2, namely, the
cycloadducts 3 and the epoxides 4 (Scheme 2). Independ-
ent synthesis of the epoxides 4a , 4c, cis-4d , 4e, 4f, and
4h 6 by dioxirane epoxidation of the corresponding olefins
allowed their identification or confirmation of their
absence in the R-peroxy lactone oxidations by comparison
of their spectral data with that of the authentic material.
Epoxides tr a n s-4d and 4g were identified by the char-
1
acteristic H NMR signals at δ 4.69 (singlet) and 5.06
(singlet). Isolation and full characterization of the ep-
oxides 4 were not possible due to their thermal lability
and decomposition during Florisil chromatography. The
cycloadducts 3 were isolated and their structures as-
signed on the basis of their spectral and analytical data.
Thus, the lactone functionality of the 1,4-dioxan-2-ones
3 displays the characteristic IR band at 1720-1765 cm-1
and a carbonyl 13C NMR resonance between δ 170.7 and
173.0, while the acetal ring positions show typical chemi-
1
cal shifts between δ 5 and 6 in the H NMR and between
δ 90 and 105 in the 13C NMR spectra.
To determine the regioselectivity of the cycloaddition
for the unsymmetrical enol ethers 2a -c, the cycloadduct
3a was allowed to react with an excess of phenylmagne-
sium bromide, which in accordance to an established
literature precedent7 yielded the acid 10a from nucleo-
philic attack of the Grignard reagent at the alkoxy-
acyloxy acetal functionality. The regiochemical assign-
Resu lts
R-Peroxy lactone 1 was synthesized according to the
published procedure5 and was obtained as deuterio-
chloroform or 1,1,1-trichloroethane solution. CDCl3 was
used for the determination of the product distribution at
the analytical scale, while CH3CCl3 was used for pre-
parative purposes. The trapping experiments were run
in a 1:1 mixture of the R-peroxy lactone solution and
methanol. The results of the analytical scale reactions
are summarized in Table 1.
ment of the cis- and tr a n s-3b cycloadducts was
determined by analogy of its spectral data with those of
the cycloadduct 3a ; the alkoxy-acyloxy acetal 13C NMR
resonances for cis- and tr a n s-3b appear at δ 102 and
105 vs 100.3 for 3a . For comparison, the alkoxy-alkoxy
acetal carbon atoms for cis- and tr a n s-3d show 13C NMR
shifts at δ 93. For cycloadduct 3c, beside the similarity
Beside the catalytic decomposition of the R-peroxy
lactone 1 into acetone, two kinds of products were
(4) (a) Greene, F. D.; Rees, W. W. J . Am. Chem. Soc. 1958, 80, 3432-
3437. (b) Adam, W.; Griesbeck, A.; Kappes, D. J . Org. Chem. 1986,
51, 4479-4481. (c) Adam, W.; Hadjiarapoglou, L.; Curci, R.; Mello, R.
(6) Adam, W.; Hadjiarapoglou, L.; Wang, X. Tetrahedron Lett. 1991,
32, 1295-1298.
(7) (a) Shostakovskii, M. F.; Kulibekov, M. R. Zh. Obshch. Khim.
1958, 28, 2339-2341. (b) Nu¨tzel, K. In Methoden der Organischen
Chemie; Mu¨ller, E., Ed.; Georg Thieme Verlag: Stuttgart, 1973; Bd.
XIII/2a, p 329.
In Organic Peroxides; Ando, W., Ed.; J ohn Wiley
Chichester, 1992; pp 195-220.
&
Sons Ltd.:
(5) Adam, W.; Alze´rreca, A.; Liu, J .-C.; Yany, F. J . Am. Chem. Soc.
1977, 99, 5768-5773.

8434 J . Org. Chem., Vol. 61, No. 24, 1996
Adam and Blancafort
Ta ble 1. P r od u ct Stu d iesa of th e Rea ction of r-P er oxy La cton e 1 w ith En ol Eth er s 2
a
b
At -20 °C, in CH3CCl3 or CDCl3; for trapping experiments in 1:1 CH3CCl3/CH3OH. Product distribution in CDCl3 determined by 1H
NMR spectroscopy with hexamethyldisiloxane as internal standard (in CH3CCl3 by 1H NMR analysis of the crude product mixture after
distillation of the solvent). c Mass balance (m.b.) refers to 100% consumption of R-peroxy lactone 1. Cis:trans ratio of the cycloadducts
d
indicated in parenthesis. e Contains 22% of ene product 5c (cf. Scheme 2). f Cis:trans ratio (d.r.) of the trapping products. Oligomeric
g
h
ester 8 not quantified due to overlapping with other signals. Yield estimated in terms of methoxy ester 9, trapping product of the R-lactone.
Stereochemistry of the diastereomers not determined. Contains 13% of undetermined side products. Also 29% of oligomeric ester 8
detected. Yield estimated in terms of oligomeric ester 8; epoxide not directly quantified due to overlapped signals with other products.
Sch em e 2. Con d ition s a n d P r od u cts for th e Rea ction betw een r-P er oxy La cton e 1 a n d En ol Eth er s 2
i
k
l
m
of its 13C NMR shift data with that of derivative 3e,
additional structural proof is provided by NOE experi-
ments (Figure 1).
The cycloaddition was stereoselective in all cases and
took place under retention of the olefin configuration.
While the cyclic enol ethers 2c (Table 1, entries 4 and 5)
and 2e-h (Table 1, entries 12-16) yielded exclusively
the cis cycloadducts, only a small degree of inversion (ca.
10-20%) was found for the acyclic enol ethers 2b (Table
1, entries 2 and 3) and 2d (Table 1, entries 6-11).
The stereochemical course of the cycloaddition was
established on the basis of NOE experiments and the

Cycloaddition and Epoxidation of Enol Ethers
J . Org. Chem., Vol. 61, No. 24, 1996 8435
F igu r e 1. Structural assignments of the cycloadducts 3b,c,d ,e,g and the trapping products 6c,e by NOE experiments
(enhancements in percent; methyl group effects were not determined for cis- and tr a n s-3b) and coupling constants.
coupling constants of the cycloadducts 3b-h (Figure 1).
For the cis- and tr a n s-3d cycloadducts, the preferred
conformation is dictated by the anomeric effect, such that
the relative position of each of the acetal protons at the
C-5 and C-6 positions is very similar in both diastereo-
mers and similar NOE effects and J 5,6 coupling constants
are observed. Fortunately, for the cis-3d isomer, an
appreciable NOE effect of 3.0% operates between the
axial methyl substituent and the 6-H acetal proton, which
allows us to identify the major diastereomer as cis-3d
in the cycloaddition with Z-2d .
Additionally, a third type of product was obtained only
in the case of enol ether 2c, namely, the ene product 5c
(Table 1, entry 4, footnote e). While the presence of the
acid functionality is evidenced both by the IR bands at
3600-2300 and 1700 cm-1 and the 13C NMR resonance
at δ 177.4, the dioxy-substituted double bond displays
characteristic resonances at δ 6.47 (1H NMR) and δ 133.2
and 137.6 (13C NMR).
When the reaction of R-peroxy lactone 1 with the enol
ethers Z- and E-2d was carried out in deuterioacetonitrile
as cosolvent (Table 1, entries 7 and 10), the change in
the polarity of the medium had only a moderate influence
on the product distribution. Thus, the epoxide yield
decreased, while more acetone was formed due to cata-
lytic decomposition, whereas the stereoselectivity of the
cycloaddition decreased slightly.
The trapping experiments were run in methanol as
cosolvent (Scheme 2 and Table 1, entries 5, 8, 11 and 13).
The trapping products 6 were identified on the basis of
their spectral data, for which especially the IR bands at
3500-2500 and 1730 cm-1, as well as the 13C NMR
resonances at δ 176, were definitive. This allows us to
exclude the alternative hydroxy ester structure 6d ′.
As far as the cycloadducts 3f and 3h are concerned,
the assignment of the configurations was encumbered by
1
the fact that the H NMR chemical shifts of the methyl
bridgehead substituents overlap severely with the methyl
groups of the lactone ring. Furthermore, all attempts
failed to grow suitable crystals for an X-ray structure
determination. However, the fact that in all other
cycloadditions the olefin geometry is conserved prompts
us to assume that retention also applies for the deriva-
tives 2f and 2h .
In the cases for which epoxidation took place, namely,
for the enol ethers 2d (Table 1, entries 6-11), 2g (Table
1, entry 15), and 2h (Table 1, entry 16), the corresponding
oligomeric ester 8 derived from the R-lactone of 18 was
observed and quantified for olefins 2g and 2h . In the
case of 2g, the yield of epoxide 4g matched the yield of
oligomeric ester 8, which establishes that both products
are formed, as expected, in the same amounts. For the
Z,E-diastereomeric enol ethers 2d , formation of the
oligomeric ester 8 could not be confirmed due to the
Furthermore, additional proof was provided in favor of
the regioisomer 6d by esterification with diazomethane.
In these methanol trapping experiments the cycloaddi-
tion remained stereoselective; in fact, for the acyclic enol
ethers Z- and E-2d (Table 1, entries 8 and 11) even a
higher degree of diastereoselectivity was found. Ad-
ditionally, an acceleration of the reaction was observed;
in the case of dihydropyran 2c (Table 1, entries 4 and 5)
1
overlap of its H NMR signals with those corresponding
to the cycloadduct 3d ; nevertheless, its presence is
inferred from the fact that the epoxides 4d are observed.
(8) Chapman, O. L.; Wojtowski, P. W.; Adam, W.; Rodr´ıguez, O.;
Ruckta¨schel, R. J . Am. Chem. Soc. 1972, 94, 1365-1367.

8436 J . Org. Chem., Vol. 61, No. 24, 1996
Adam and Blancafort
the reaction times decreased from 15 to 5 min in
methanol. Furthermore, in the presence of methanol, the
R-lactone derived from the R-peroxy lactone 1 on deoxy-
genation was trapped as the R-methoxy acid 9 (Scheme
2).
tion to the 1,4-dioxanones 3 mainly with retention of
initial configuration and oxidation of the enol ethers to
the corresponding epoxides 4 (Table 1 and Scheme 2).
The epoxidation mode only occurs with the 1,2-dioxy-
substituted olefins Z- and E-2d , 2g, and 2h (Table 1,
entries 6-11, 15, and 16). In the presence of methanol,
the trapping products 6 are also observed, again mainly
with retention of configuration.
To rationalize this complex product distribution mecha-
nistically, it is instructive to recall our previous results
on the oxidation of alkenes with the R-peroxy lactone 1.1
For the latter case, two independent pathways were
postulated for the formation of cycloadducts, the first via
the formation of ene products and epoxides and the
second by nucleophilic attack of the alkene double bond
on the alkyl-substituted oxygen of the peroxide bond of
the R-peroxy lactone 1 (Scheme 1). Mainly steric factors
decide whether the end-on, unsymmetrical attack leads
to the stretched 1,6-zwitterion A (W conformation) as the
precursor to the cycloadducts 3 or the centered, sym-
metrical attack leads to the epoxide 4 through a concerted
process with a “butterfly” transition state B. Beside
cyclization under loss of the alkene configuration and
formation of the ene products by intramolecular H
transfer, the 1,6-zwitterion A could be trapped when the
reaction was performed in methanol as cosolvent.
The present methanol trapping experiments for the
enol ethers 2 (Table 1, entries 5, 8, 11, and 13) exclude,
despite the observed retention of configuration, concerted
cycloaddition and suggest the intervention of a zwitter-
ionic intermediate. The charge separation which ac-
companies the peroxide bond breakage to generate this
zwitterionic intermediate is in part alleviated by the polar
methanol, as exemplified by the rate acceleration for the
dihydropyran 2c (Table 1, entries 4 and 5). The question
of mechanistic relevance is the structural nature of this
intermediate, which exercises extensive stereomemory
but can be trapped! It certainly cannot be the stretched
1,6-dipole A (W conformation) proposed in the reaction
of the R-peroxy lactone 1 with alkenes (Scheme 1),
because the trappable dipole A does not preserve its
configuration. Moreover, to propose that the cycloaddi-
tion is concerted in nonpolar (CDCl3, CH3CCl3) but
stepwise in polar (CH3CN, MeOH) solvents is not justified
because the configuration of the cycloadducts is largely
retained also in acetonitrile (Table 1, entries 7 and 10)
and methanol (Table 1, entries 8 and 11) as cosolvents
compared to chloroform (Table 1, entries 6 and 9).
Consequently, the results in Table 1 demand a cyclo-
addition intermediate which largely retains its configu-
ration and can be trapped, i.e., different from the
stretched 1,6-dipole A for the alkene oxidation (Scheme
1). Furthermore, a satisfactory mechanistic interpreta-
tion must account for the fact that all enol ethers 2 afford
cycloadducts 3, but only for certain ones, namely, the
olefins Z,E-2d , 2g, and 2h , epoxides 4 are also obtained.
We propose that the reaction proceeds through the
epoxonium 1,4-dipole (Scheme 3), whose structure is side-
differentiated in terms of the two diastereomeric spiro
configurations syn - and a n ti-C. The stereolabels refer
to the location of the onium substituent with respect to
the dioxy-substituted side of the substrate. Both struc-
tures arise from nucleophilic attack of the enol ether 2
on the peroxide bond of the R-peroxy lactone 1 in a central
and symmetric approach. Thus, in contrast to the alkene
nucleophiles, the two lateral sides of the π bond are
electronically differentiated, with the sole exception of
the centrosymmetric substrate E-2d . In addition we
In regard to the configurations of the trapping products
6, the cyclic enol ether 2c yielded the diastereomeric
mixture 6c in a ratio of 87:13 (Table 1, entry 5). The
major diastereomer tr a n s-6c could be isolated, while the
minor cis-6c diastereomer was identified in the 1H NMR
spectrum of the crude product mixture by a doublet at δ
4.57 (J ) 2.3 Hz) for the 2-H' acetal proton and a singlet
at δ 3.46 for the methoxy group. The configuration of
the diastereomers was determined by NMR spectroscopy.
While the NOE effects measured for tr a n s-6c were not
conclusive, the definitive assignment was made on the
basis of the coupling constants between the protons of
the alkoxy-substituted positions (Figure 1). Thus, the
major diastereomer tr a n s-6c has a coupling constant of
5.6 Hz, which is an average value of its two possible
equilibrating conformers (ca. 10 Hz for the axial-axial
and ca. 2 Hz for the equatorial-equatorial coupling
constants).9 For the minor isomer cis-6c, whose two
conformers should show a small axial-equatorial cou-
pling constant, a value of 2.3 Hz was measured, which
confirms its cis configuration. Unfortunately, the con-
figuration of the acyclic diastereomers 6d (Table 1,
entries 8 and 11) could not be assigned. Nevertheless,
the spectra of the crude trapping mixtures for the
individual Z and E isomers of the enol ether 2d revealed
that opposite ratios of diastereomers were obtained,
which confirms that the intermediate precursor for the
trapped products 6d retains its initial configuration to a
certain extent. As for the configuration of the 1,4-dioxene
trapping product diastereomers 6e, which were obtained
in a ratio of 80:20 (Table 1, entry 13), an NOE experiment
(Figure 1) allowed no distinction between the two iso-
mers.
Furthermore, to probe that the trapping products 6 did
not arise from methanolysis of the cycloadducts 3, a
control experiment for 3d in deuteriomethanol was
carried out. Even after 1 d at room temperature, no
reaction of the cycloadduct 3d was observed, which
confirms that the cycloadduct 3d persists methanolysis
under the reaction conditions for its formation from the
R-peroxy lactone 1.
Finally, the mass balances were satisfactory (>80%)
in most of the cases. Although in the reaction of R-peroxy
lactone 1 with the enol ether 2a (Table 1, entry 1) the
mass balance was only 74%, no additional products
beside the cycloadduct 3a and acetone could be detected.
The low mass balance of identified products (71%) for the
reaction with enol ether E-2d in deuterioacetonitrile as
cosolvent (Table 1, entry 10) must be corrected for
1
unknown products (13%), which were detected in the H
NMR spectrum of the crude reaction mixture. Finally,
the low mass balance (60%) observed also in the trapping
experiment for this enol ether (Table 1, entry 11) is
tentatively attributed to catalytic decomposition of the
R-peroxy lactone 1 to acetone.
Discu ssion
The present study on the reaction of R-peroxy lactone
1 with a series of mono- and dioxygen-substituted olefins
2 reveals two reaction modes, namely, formal cycloaddi-
(9) Sweet, F.; Brown, R. K. Can. J . Chem. 1966, 44, 1571-1576.

Cycloaddition and Epoxidation of Enol Ethers
J . Org. Chem., Vol. 61, No. 24, 1996 8437
Sch em e 3. Mech a n ism for th e Rea ction of r-P er oxy La cton e 1 w ith En ol Eth er s 2 a n d Meth a n ol Tr a p p in g
Illu str a ted for th e cis-Dioxy-Su bstitu ted Ca ses
propose that the syn -C intermediate is sufficiently long-
lived to release the epoxide 4 by intramolecular nucleo-
philic substitution of the carboxylate ion on the R carbon
atom of the incipient R-lactone. In contrast, the a n ti-C
intermediate opens up to the coiled 1,6-zwitterion D (U
conformation), which due to the proximity of the car-
boxylate and the carbocation centers is too short-lived
to stereorandomize and cyclizes to the cycloadducts 3
meric pair, Z,E-2d diastereomeric mixtures of the
with retention of the enol ether configuration. In fact,
trapping products 6 are obtained with opposite dia-
an analogous, U-shaped intermediate has been postu-
stereomeric ratios, which shows that the trapped inter-
lated in the reaction of tetracyanoethylene with enol
mediate exercises a certain degree of stereomemory.
ethers,10 for which cycloaddition under stereoretention
The driving forces for the ring opening of the a n ti-C
epoxonium dipole are the electronic repulsion between
the lone pair electrons of the onium center and the enol
ether oxygen substituents, which are located on the same
and stereoselective trapping of the intermediate are
observed. Furthermore, interconversion of the diastereo-
meric syn - and a n ti-C can be excluded on the basis of
computational results for analogous perepoxide interme-
diates.11 Thus, MINDO/3 calculations on the addition of
singlet oxygen to 1,3-butadiene11a give an activation
barrier of 24 kcal/mol for the interconversion of the
perepoxide diastereomers, while in a multiconfiguration
CASSCF calculation with a double-ê basis set on the
singlet oxygen addition to ethylene11b the inversion
barrier was estimated to be as high as 78 kcal/mol.
1,4-Dipoles syn - and a n ti-C, as well as the coiled 1,6-
zwitterion D, may be trapped stereoselectively by metha-
nol to afford the adduct 6. Thus, for dihydropyran 2c
SN2 attack takes place at the epoxonium C preferentially
to afford tr a n s-6c as the major trapping product. The
small (13%) amount of the cis-6c diastereomer may
derive from the methanol aggregate D (MeOH), analo-
gous to the one proposed for the trapping of a zwitterionic
intermediate in the reaction of tetracyanoethylene with
ethyl cis-propenyl ether.10b Also for the open diastereo-
side of the intermediate a n ti-C, and the stabilization of
the incipient carbocation by the adjacent O-substituent.
The possibility as to whether the proposed D species,
instead of a bona fide intermediate, may be a transition
state on the way to the cycloadducts 3 is unlikely in view
of the partial stereoisomerization observed for the cy-
cloadducts 3 and the trapping products 6 (Table 1, entries
6-11). Additionally, the formation of the cycloadducts
3 directly from the 1,4-dipoles syn - or a n ti-C through
SN2 attack at the epoxonium function can be excluded
for geometric reasons.
The observed epoxidation for the 1,2-dioxy-substituted
substrates Z- and E-2d , 2g, and 2h is accounted for by
their nucleophilic attack on the R-peroxy lactone 1 to
afford the syn -C epoxonium intermediate (Scheme 3).
The electronic repulsion argued for a n ti-C species is
absent in syn -C, so that the syn -C epoxonium structure
does not open but leads to the epoxide by intramolecular
SN2 attack. In contrast, for the analogous 2e and 2f
derivatives, which give no epoxides (Table 1, entries 12-
14), their central, symmetric approach to generate a
syn -C intermediate is sterically blocked by the 5,6-
(10) (a) Huisgen, R. Acc. Chem. Res. 1977, 10, 117-124. (b) Huisgen,
R. Acc. Chem. Res. 1977, 10, 199-206.
(11) (a) Dewar, M. J . S.; Thiel, W. J . Am. Chem. Soc. 1977, 99, 2338-
2339. (b) Hotokka, M.; Roos, B.; Siegbahn, P. J . Am. Chem. Soc. 1983,
105, 5263-5269.

8438 J . Org. Chem., Vol. 61, No. 24, 1996
Adam and Blancafort
substituents in view of their puckered conformations;
thus, only cycloaddition through the a n ti-C route instead
of epoxidation occurs.
cited,13 which is a concerted process for alkenes but a
dipolar one for the more electron-rich enol ethers and
enamines. Thus, the degree of charge transfer from the
double bond to the peroxide bond is higher for the enol
ethers than the alkenes, so that nucleophilic substitution
occurs in which the carboxy oxygen of the peroxide bond
in the R-peroxy lactone is displaced as a carboxylate ion
and the epoxonium-type intermediate C is formed instead
of concerted epoxidation. Moreover, although stepwise
oxidation processes with peroxides in nonpolar solvents
have been claimed to be disfavored by the high-energy
barriers for the charge separation when the developing
charge is not stabilized,14 in the present case there is
indeed a stabilization due to the fact that the formed
oxyanion is a carboxylate. Furthermore, it has to be
explained why no end-on attack leading to an open-chain
1,6-dipole of the A type occurs with the enol ethers. We
consider this an additional consequence of the high
degree of charge transfer in the early stage of the
reaction, since the alkoxy-substituted oxygen becomes
more nucleophilic and traps more effectively the incipient
carbocation through back-donation to form the bridged
dipole C.
In conclusion, the high degree of stereomemory for the
enol ethers but extensive stereorandomization for the
alkenes in the cycloadduct formation, yet with methanol
trapping for both substrates, demands two distinct
intermediates, namely, the epoxonium-type 1,4-dipole C
(Scheme 3) and the stretched 1,6-dipole A (Scheme 1).
While the epoxonium species C rearranges by way of the
coiled 1,6-dipole D highly stereoselectively to the cyclo-
adduct 3, the stretched 1,6-dipole A suffers substantial
stereochemical loss prior to cyclization. Moreover, for the
enol ethers the epoxonium intermediate C serves as a
precursor to the epoxide, while for the alkenes the
concerted epoxidation through the concerted transition
state B is reasonable. Attempts to provide some insight
into this mechanistic dichotomy through semiempirical
(AM1, PM3) and low-level (3-21G*) ab initio calculations
unfortunately failed; however, detailed ab initio calcula-
tions14 might be informative.
For the cis-configurated olefins Z-2d , 2g, and 2h , the
dioxy-substituted side is accessible for the R-peroxy
lactone 1 electrophile, because in the acyclic substrate
Z-2d the ethoxy groups are conformationally free to
rotate out of the way, while for the benzo-substituted
derivatives 2g,h the flat benzene ring presents no steric
obstacles. Indeed, the electron-rich arene moiety may
provide through charge transfer complexation an ad-
ditional incentive for the syn -C attack of the electrophilic
oxidant in analogy to the effect proposed for electrophilic
additions to aryl-substituted benzobicyclo[2.2.2]octadienes
and 11-isopropylidenedibenzonorbornadienes,12 where
the electrophile approaches the double bond from its
benzo-substituted face in those cases in which other
effects do not intervene.
In the case of the centrosymmetric, trans-configurated
E-2d substrate only one epoxonium species is possible,
which partitions between epoxidation and cycloaddition.
Also some Grob-type fragmentation into acetone and
presumably CO2 with regeneration of olefin E-2d (entries
9 and 10) takes place. In the presence of methanol,
significant amounts of trapping product 6d (Table 1,
entry 11) are observed with extensive loss of stereo-
chemistry.
The mechanistic analysis in Scheme 3 also applies for
the monooxy-substituted olefins 2a -c. For these the
sterically less hindered side is preferentially approached
to afford the a n ti-C epoxonium intermediate. The latter
opens up to the 1,6-dipole D and subsequently cyclizes
stereoselectively to the cycloadducts 3 with predominant
retention for the acyclic enol ethers Z- and E-2b (Table
1, entries 2 and 3) and exclusive cis cycloadduct 3c for
the dihydropyran 2c (Table 1, entries 4 and 5). For the
latter substrate also the methanol trapping to the adduct
6c proceeds with a large degree of stereoselectivity, which
implies that the methanol was added by a back-side
attack directly to the epoxonium intermediate a n ti-C
and/or to the transient U-shaped 1,6-dipole D to yield
the trans diastereomer. That this dipole is sufficiently
long-lived for trapping is substantiated by the observed
H abstraction (in the dipolar intermediate D for 2c R4 )
H, Scheme 3) to afford the ene product 5c (Scheme 2).
Ack n ow led gm en t. We thank Prof. G. To´th (Buda-
pest) for his valuable help in the structural assignment
of adducts cis- and tr a n s-3d and 6c by NMR spectros-
copy. We thank the Deutsche Forschungsgemeinschaft
(Schwerpunktprogramm “Peroxidchemie: mechanis-
tische und pra¨parative Aspekte des Sauerstofftrans-
fers”) for generous financial support, and L.B. thanks
the Deutscher Akademischer Austauschdienst (DAAD)
for a doctoral fellowship (1992-1996).
Su p p or tin g In for m a tion Ava ila ble: Experimental sec-
tion on the synthesis and complete characterization of the
products (16 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.
At first glance, it may seem a contradiction to postulate
for the R-peroxy lactone 1 oxidation of the enol ethers 2
a mechanism based on a 1,4-zwitterionic epoxonium
intermediate C (Scheme 3) when in the alkene case
concerted epoxidation (Scheme 1) was preferred. How-
ever, the higher nucleophilicity of the electron-rich enol
ethers compared to alkenes seems to be decisive here.
As an analogy, the addition of olefins to ketenes may be
J O961645D
(13) (a) Huisgen, R.; Feiler, L. A. Chem. Ber. 1969, 102, 3391-3404.
(b) Huisgen, R.; Feiler, L. A.; Otto, P. Chem. Ber. 1969, 102, 3405-
3427. (c) Feiler, L. A.; Huisgen, R. Chem. Ber. 1969, 102, 3428-3443.
(14) (a) Bach, R. D.; Owensby, A. L.; Gonza´lez, C.; Schlegel, H. B.
J . Am. Chem. Soc. 1991, 113, 2338-2339. (b) Bach, R. D.; Owensby,
A. L.; Gonza´lez, C.; Schlegel, H. B.; McDouall, J . J . W. J . Am. Chem.
Soc. 1991, 113, 6001-6011. (c) Bach, R. D.; Andre´s, J . L.; Owensby,
A. L.; Schlegel, H. B.; McDouall, J . J . W. J . Am. Chem. Soc. 1992, 114,
7207-7217.
(12) (a) Paquette, L. A.; Bellamy, F.; Wells, G. J .; Bo¨hm, M. C.;
Gleiter, R. J . Am. Chem. Soc. 1981, 103, 7122-7133. (b) Paquette, L.
A.; Klinger, F.; Hertel, L. W. J . Org. Chem. 1981, 46, 4403-4413.