Z.-X. Yu, Y. Tang et al.
desired products in good yields. The reactions of cyclopro-
panes bearing styryl, vinyl, and 2-furyl groups also worked
well, affording [3+2] products 3i–k exclusively, which could
easily be further functionalized and transformed. Next, a
series of enol silyl ethers with a different type or size of ring
system were evaluated with the 2-furyl-substituted cyclopro-
hydrogen atoms in a cis configuration, or as a base to ab-
stract the methine proton at the a-position of the silylated
carbonyl group (via transition state TS3-cis) to generate by-
product 5. Similarly, from I-trans, the other [3+2] cycload-
duct 4 with two methine hydrogen atoms in a trans configu-
ration and byproduct 5 can be generated. Therefore, the se-
lectivity of this reaction depends on the relative energies in
these transformations. Our DFT calculations indicate that
the steric effect of substituents on the ester and silyl groups
is responsible for obtaining high diastereoselectivities in the
[3+2] cycloaddition reaction.[13]
In reaction A (R=Me, Si=TMS, Figure 1), the genera-
tion of zwitterionic intermediates I-cis and I-trans through
nucleophilic attack of enol silyl ether 2 on the copper(II) ac-
tivated cyclopropane in complex C is rate determining and
irreversible,[14] requiring activation free energies of 15.7 and
15.4 kcalmolÀ1 in CH2Cl2, respectively (Figure 1, blue line).
The very close energies suggest that these two intermediates
will be formed in nearly equal amounts. From I-cis, the
[3+2] cycloadduct 3 will be generated exclusively through
TS2-cis because the transition state TS3-cis, which gives by-
product 5, is higher in energy than TS2-cis by 7.0 kcalmolÀ1
(Figure 1). However, cycloadduct 4 along with a minor
amount of byproduct 5 will be generated from I-trans. This
is because the activation free energy of the transformation
of I-trans into 5 is just 1.3 kcalmolÀ1 higher than that of the
intramolecular cyclization through TS2-trans (14.2 versus
12.9 kcalmolÀ1, Figure 1). Therefore, the computational re-
sults show that the major products of reaction A are the
[3+2] cycloadducts 3 and 4, but the ratio of 3 to 4 is poor
(about 1:1). This computational result is in good agreement
with the experimental results (Table 1, entry 16).
pane by using CuACTHNUTRGNEUNG(SbF6)2/L1 (5 mol%) as the catalyst
system. It was found that all of these reactions diastereose-
lectively gave the desired [3+2] cycloadducts as various
fused-ring systems in moderate to good yields.
Highly functionalized ring-fused five-membered carbocy-
cles are potentially useful in organic synthesis and pharma-
ceutical science. For example, the furyl group in 3k can be
easily oxidized to form a carboxyl group by treatment with
RuCl3/NaIO4 (Scheme 3). The DIBAL-H reduction of 3k,
followed by a Horner–Wadsworth–Emmons reaction gave
product 10 with four contiguous stereocenters as a single
diastereomer (Scheme 3). The relative configuration of alde-
hyde intermediate 9 was confirmed by X-ray diffraction
In reaction B (R=iPr, Si=TBDPS, Figure 1), there is still
no clear preference for the generation of zwitterionic inter-
mediate I-cis or I-trans. This is because the free energy dif-
ference between the two nucleophilic attack transition states
TS1-cis and TS1-trans is only 0.8 kcalmolÀ1 (20.9 versus
21.7 kcalmolÀ1, Figure 1, red line). In contrast to reaction A,
for which the rate-determining step is the first step (nucleo-
philic attack), reaction B has the intramolecular cyclization
as the rate-limiting step. As a result, the diastereoselectivity
of this reaction will be determined by the free-energy differ-
ence between the two cyclization transition states TS2-cis
and TS2-trans. According to the DFT calculations, transition
state TS2-cis, leading to product 3, is 2.8 kcalmolÀ1 lower in
energy than transition state TS2-trans, giving product 4
(24.4 versus 27.2 kcalmolÀ1, Figure 1), which predicts a 3/4
ratio of 110:1 at 298 K. This is consistent with the experi-
mentally observed excellent diastereoselectivity (d.r.>99:1,
Table 1, entry 5). Furthermore, the DFT calculations indi-
cate that a 21% yield of byproduct 5 will also be generated
from I-trans (via TS3-trans, 25.2 kcalmolÀ1, 0.8 kcalmolÀ1
higher than TS2-cis, Figure 1), which is also close to the ex-
perimental result.
Scheme 3. Transformations of product 3k.
analysis.[9]
We applied DFT calculations by using the (U)B3LYP
method[10] to study the reaction mechanisms of two repre-
sentative reactions A and B (Figure 1) and rationalize the
method by which the substituents on the ester and silyl
groups influence the diastereoselectivity of the [3+2] cyclo-
addition reaction.[11] Our mechanistic studies started from
the commonly accepted complex C because the formation
of C from cyclopropane-1,1-dicarboxylate 1 and the copper
catalyst Cu2+L2 is highly exergonic.[12] It is interesting to
À
note that the C1 C2 bond in the cyclopropane in complex C
À
is about 0.15 ꢀ longer than the C1 C3 bond (Figure 1, and
for DFT-calculated structures, see Figure S3 in the Support-
ing Information). This indicates that the C1 C2 bond will
À
be broken much more easily. DFT calculations further re-
vealed that complex C can react at the C2 position with
enol silyl ether 2 to generate two different zwitterionic inter-
mediates I-cis and I-trans through transition states TS1-cis
and TS1-trans, respectively (Figure 1). In the zwitterionic in-
termediate I-cis, the carbanion functions as either a nucleo-
phile to attack the silylcarboxonium (via transition state
TS2-cis) to form one [3+2] cycloadduct 3 with two methine
In both reactions A and B, the selectivity of the first step
(nucleophilic attack) is poor, but the selectivity of the
second step (intramolecular cyclization) is high. However, in
2198
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 2196 – 2201