Communication
and high diastereoselectivity (Table 2, entry 2). With the same
enol ether, the reaction of either dimethyl or diphenylketene
with enol ether 4 was much more efficient, both in terms of
yield and diastereoselectivity: the cis-cyclobutanones 10 and
11 were obtained in good to excellent yields with good to ex-
cellent diastereoselectivities (Table 2, entries 3 and 4). Similar
results were obtained with ethyl butenyl ether 5 (Table 2, en-
tries 5 and 6). Finally, silyl enol ethers 6 and 7 were also con-
verted into the corresponding cis-cyclobutanones 14 and 15
with only traces of the trans products (Table 2, entries 7 and 8).
Overall, in [2+2] cycloaddition involving ketenes, diastereo-
merically enriched cyclobutanones can easily be obtained re-
gardless of the stereochemical purity of the starting olefins.
Having experimentally proven the kinetic non-equivalence
of Z/E-olefin in [2+2] cycloadditions, it seemed important to
gain insight into the mechanistic origin of this dichotomy. To
the best of our knowledge, no theoretical studies have been
realized on this Z/E reactivity since the first theoretical study
on the cycloaddition of ketene with alkene.[3] The mechanistic
basis of ketene–alkene [2+2] cycloaddition were set by the
groups of Houk[8a] and Moyano[8b] in the 1990s. They first
showed the importance of the nearly perpendicular approach
of the ketene to the alkene, and also that the quasi-pericyclic
[p2s+(p2s+p2s)] reaction mechanism is more relevant than
a concerted asynchronous (p2s+p2a) cycloaddition.[8,9] Their re-
activity investigation was realized using respectively ab initio
(HF and MP2 for the simplest examples) and semi-empirical
calculations (AM1). In these works, the difference of reactivity
of Z/E-alkenes was not addressed. However, the enhanced re-
activity of the Z-alkene over the E-alkene has tentatively been
justified by simple steric interactions arguments[10] or by as-
suming that the energetic effects of the alkenes substituents
would be additive.[8a] Applying this shortcut would lead to op-
posite results based on Houk or Moyano’s results (inversion of
relative stability between TS3 and TS4: Table 3, entry 1 and 2).
It was thus worthwhile examining simple cycloaddition of ke-
tenes and Z/E-alkenes at the DFT level.
Table 3. Relative activation energy [kcalmolꢀ1] for the transition states of
propene with ketenes.
Entry
R
TS1
TS2
TS3
TS4
1
2
3
4
5
6
H[a]
0
0
0
0
0
0
4.9
1.2
2.9
4.3
3.5
4.3
4.5
6.8
5.2
4.8
5.5
5.7
6.9
6.0
6.0
8.6
9.1
10.7
H[b]
H[c]
Me[c]
Cl[c]
Ph[c]
[a] Results reported by Houk et al. (RHF/3-31G).[8a] [b] Results reported by
Moyano et al. (AM1).[9b] [c] Our calculations (B3LYP/6-31+G(d,p)).
tion are additive is not satisfactory. Moreover, both steric repul-
sion and the effect of positive charge stabilization are inter-
twined, and therefore cannot justify that the Z-alkenes are
more reactive than the E-alkenes. Calculations on the cycload-
dition between symmetrically substituted ketene and disubsti-
tuted Z/E-but-2-ene were therefore next performed. For each
isomer, there are two approaches: one, called “anti”, in which
the alkene molecule approaches the ketene with its substitu-
ents as far as possible from the ketene substituents, and
a second, called “syn”, in which the olefin approach takes place
with alkene substituents close to the ketene substituents. The
results showed that the transition states of disubstituted al-
kenes are very similar to those of mono-substituted alkenes,
with the ketene approaching in a slightly bent perpendicular
fashion (Table 4). The calculated difference of energy of activa-
tion between E- and Z-olefins correlates nicely with the experi-
¼
mental results: the lowest activation energy (DG ) is always as-
sociated with the TS Z anti (Table 4). The increase of size of the
ketene results in an increase of this difference of activation
energy: from 1.2 to 5.5 kcalmolꢀ1 with nonsubstituted- and di-
phenylketene, respectively. It turned out that the steric repul-
sion between the alkene substituents either at C3 or at C4 and
the smaller substituent of the ketene, which points toward the
alkene, is of crucial importance, much more important than
the carbonyl interaction with the alkene. The E anti and E syn
transition states are relatively close in energy, especially for
bulkier ketenes, in which a more perpendicular approach leads
to the merge of the two trajectories. Finally, the TS Z syn is en-
ergetically the highest one since it combines steric repulsions
between the ketene substituent with both C3 and C4 alkene
substituents.
Performing calculations using Gaussian 09 D.01 at DFT
B3LYP/6-31+G(d,p) level,[11] a trend of reactivity similar to the
results of Houk and Moyano was found for all four transition
states (TS). These calculations confirm, without surprise, the
importance of the regioselectivity of the process through stabi-
lization of the positive charge by the methyl group (TS1-2<
TS3-4), and a particular unfavorable steric interaction between
the substituent of the alkene and the ketene (TS1<TS2 and
TS3<TS4; Table 3). These results also confirm that TS3 is lower
than TS4, following the previous order reported by Houk, but
opposite to the one reported by Moyano (Table 3, entries 1
and 2). In addition, these results also show that TS2 is always
lower in energy than TS3, as reported by Moyano, but oppo-
site to the results of Houk: hence, the interaction of the alkene
substituent is greater with the ketene substituent compared
with the ketene carbonyl (in combination with the less favora-
ble initial charge development at C3; Table 3).
As enol ethers are commonly used in the [2+2] cycloaddi-
tion with ketene, and also proved to be good substrate for ki-
netic resolution, the computational study of their Z/E differ-
ence of reactivity in [2+2] cycloaddition was next investigated.
With enol ethers, their s-cis or s-trans conformation should be
considered, but it has previously been shown for Diels–Alder
and [3+2] cycloaddition that in the transition state, the s-trans
conformation is always preferred, regardless of the ground
state.[14] For [2+2] cycloaddition, we also confirm that despite
a preference for the s-cis conformation for E-enol ethers with
Simply based on these results, it would be very unreliable to
predict the outcome of the cycloaddition for disubstituted al-
kenes: assuming that the energetic effects of methyl substitu-
Chem. Eur. J. 2015, 21, 1 – 6
3
ꢂ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ