Adam et al.
polarization of the CC double bond. Electron-rich sub-
stituents in the R region increase the electron density in
the â region through conjugation (+M effect) and the
attack will be preferred there. This implies an asym-
metric perepoxy-like structure for the transition state,
with the O-1 atom closer to the double bond in the â
region. For the ene reaction, the steric interaction with
the substituent R1 is less severe, because the terminal
O-2 atom is closer to the methyl group for the hydrogen
abstraction, whereas in the [2+2] cycloaddition, the
terminal O-2 atom must come into the R region to cyclize
the dioxetane ring and the interaction with R1 is much
more intensive.
Dia ster eoselectivity in th e DMD a n d m CP BA
Ep oxid a tion s. The epoxidation of the enecarbamate
1j-l showed also a high degree of diastereoselectivity
(Table 3), although not as high as for the photooxygen-
ations (Table 2). The diastereoselectivity of the epoxida-
tion is again controlled by the R1 substituent of the
oxazolidinone ring, since the epoxidation of the parent
achiral enecarbamate 1h (R1 ) H) is unselective (Table
3, entries 1 and 2). Moreover, the configuration of
stereogenic center in the oxazolidinone ring determines
the absolute configuration of the epoxide product, that
is, the R configuration in the enecarbamate Z-1j (entries
3 and 4, Table 3) leads mainly to the (4R,2′S,3′S)-4j
product, whereas the S configuration in the substrates
Z-1k (entries 5 and 6) and Z-1l (entries 7 und 8) affords
mainly the corresponding 2′R,3′R epoxides.
To account for the fact that the size of the R1 substitu-
ent dictates the extent of the diastereoselectivity of the
epoxidation, we propose the mechanism in Figure 8 for
the oxygen transfer, in which for convenience of com-
parison the attacks from above and below for both the E
and the Z diastereomers are given. When X ) H, the
steric hindrance for the 1′Re,2′Re attack is more or less
the same as that for the 1′Si,2′Si attack, since a favorable
conformation may be selected in which the methyl groups
are out of the way of the incoming oxidant. When,
however, R1 is a tert-butyl group (X ) CH3 in Figure 9)
as in the enecarbamate Z-1l (entries 7 and 8, Table 3),
the 1′Re,2′Re attack is severely obstructed and the
highest 1′Si,2′Si diastereoselectivity (dr ) 7:93) is ob-
tained; the (2′R,3′R)-4l epoxide is formed almost exclu-
sively for both oxidants DMD and mCPBA.
The preferably attacked π face remains the same for
the Z- or E-configured enecarbamates, i.e., the 1′Si,2′Si
face for the Z and the 1′Si,2′Re face for the E isomer in
Figure 8; however, the extent of the diastereofacial
differentiation depends decisively on the double bond
configuration, as manifested by the fact that the epoxi-
dation of the Z isomers is always more selective than that
of the E isomers. This difference is due to the conforma-
tional effects in the Z and E isomers, which express the
shielding efficacy of the double bond in the enecarbamate.
As may be seen from the lowest energy conformations of
the E-1l and Z-1l isomers calculated by the DFT method
(see the Supporting Information), the shielding of the
upper face of the double bond, i.e., 1′Re,2′Re for Z-1l and
the 1′Re,2′Si for E-1l (Figure 8), by the tBu group is more
effective in the case of the Z than the E diastereomer.
Expectedly, the 1′Si,2′Si attack should be facilitated for
the Z case, as is documented by the higher dr values for
the Z-1l substrate (Table 3, entries 7 and 8) versus those
F IGURE 7. Fukui’s polarity model of singlet oxygen applied
to the photooxygenation of the enecarbamate substrate.
is too remote to interact sterically and/or electronically
with the attacking 1O2. Thus, the high diastereoselectivity
in the dioxetane formation from the Z-1i-l enecarbam-
ates (Table 2, entries 20-23) may be explained in terms
of the synergistic interplay between the directing vinylic
nitrogen effect and the effective shielding of one of the π
faces of the CC double bond by the R1 substituent
(Scheme 7). Unfortunately, the diastereoselectivity of the
[2+2] cycloaddition for the enecarbamates E-1i-l could
not be assessed (Table 2, entries 16-19), because these
dioxetanes decompose during the photooxygenation.
Dia ster eoselectivity in th e En e Rea ction of Sin -
glet Oxygen . The diastereoselectivity of the ene reaction,
like the mode selectivity, depends strongly on the double
bond configuration of the enecarbamate (Figure 3). In the
case of the Z-1i-l derivatives (entries 20-23, Table 2),
relatively little (13-25%) ene product is formed and the
diastereoselectivity depends strongly on the steric hin-
drance of the R1 substituent, which follows the order Me
t
iPr < Ph , Bu. In fact, to obtain the ene product, an
anti attack of singlet oxygen must occur with respect to
the R1(Ph) substituent (Scheme 6), because the methyl
group with the abstractable hydrogen atoms is on the
opposite side and less likely accessible. In the case of the
methyl and isopropyl R1 substituents (Table 2, entries
20 and 21), the steric shielding is not sufficiently effective
to obtain appreciable diastereofacial differentiation in the
anti attack (entries 20 and 21, Table 2). On the contrary,
the phenyl and tert-butyl R1 substituents (entries 22 and
23) are sufficiently spacious to cover up one π face of the
double bond and block hydrogen abstraction from the anti
direction.
In the case of the E-1 substrates, the oxazolidinone ring
is on the same side of the double bond as the methyl
group (Scheme 6). Thus, through the steering effect of
the vinylic nitrogen, the singlet oxygen enters the double
bond from the side of the oxazolidinone ring (syn attack)
and encounters more severe steric interactions with the
R1 substituent. Accordingly, the diastereoselectivity in
the ene reaction of the E-1 substrates is generally higher
(entries 16-19, Table 2) than that for the corresponding
Z isomer (entries 20-23) and already a methyl substitu-
ent gives about the same diastereoselectivity (88:12) as
a tert-butyl substituent (91:9).
The generally lower diastereoselectivity of the ene
reaction with respect to the [2+2] cycloaddition depends
on the distinct steric interactions between the incoming
singlet oxygen and the R1 substituent for these two
reaction modes. According to Fukui’s theoretical analy-
sis,15 the alignment of the approaching O-1 atom (Figure
7) determines whether the R or â region of the double
bond will be attacked, which in turn depends on the
1712 J . Org. Chem., Vol. 69, No. 5, 2004