Full Paper
to
give
intermediate
8
(Scheme 2). Decomposition of
intermediate 8 will release the
Diels–Alder adduct and regener-
ate catalyst 1. Herein, we report
the application of carbocations
as extremely efficient, mild, and
selective Lewis acid catalysts for
the Diels–Alder reaction, with
catalyst loadings down to ppm
Scheme 1. Trityl- versus silyl-cation-catalyzed Mukaiyama aldol reaction.
Lewis acids, for example, trimethylsilyl trifluoromethanesul-
fonate (TMSOTf), are frequently used as catalysts. The appli-
cation of silyl cations as powerful catalysts for Diels–Alder
reactions has recently gained attention, as demonstrated
by the groups of Sawamura[11] and Oestreich.[12]
2) Decomposition of the trityl cation to form a Brønsted acid
that catalyzes the reaction.
In a kinetic study, Denmark et al. found support for the car-
bocationic mechanism (Scheme 1, Tr-route).[13] However, a thor-
ough mechanistic study by Bosnich et al. contradicted the re-
sults of Denmark et al.[14] The investigation by Bosnich et al. ex-
cludes the possibility of a Brønsted acid catalyzed pathway
Scheme 2. Trityl-cation-catalyzed Diels–Alder reaction.
levels. We demonstrate how the reactivity and selectivity of
the reaction can be altered by tuning the Lewis acidity of the
carbocation. We have strong support for carbocationic catalysis
because we were able to exclude the involvement of compet-
ing Brønsted acid catalysis.
and points toward
a
silyl-cation-catalyzed mechanism
(Scheme 1, TMS-route). In 1997, Chen et al. reported on at-
tempts towards an asymmetric Mukaiyama aldol reaction,
mediated by stoichiometric amounts of a chiral trityl carbocat-
ion.[15] The aldol addition products were isolated with 3–50%
enantiomeric excess (ee), depending on the counterion and ar- Results and Discussion
omatic groups. However, when catalytic amounts of the chiral
Diels–Alder reactions
carbocation were used, no enantioselectivity was observed.
These results also support the suggestion that catalysis occurs
through a trityl-cation-initiated formation of a silyl cation (for
example, Scheme 1, TMS-route). In contradiction, Mukaiyama
et al. reported a trityl-catalyzed aldol addition of vinylacetate
to aldehydes.[16] Thus, removal of the silyl groups from the
system points towards trityl-cation catalysis, although, no con-
trol experiments were carried out to exclude Brønsted acid cat-
alysis. Kagan and co-workers developed a chiral ferrocenyl car-
bocation that was used as a catalyst in an attempted asym-
metric Diels–Alder reaction.[17] However, the absence of enan-
tioselectivity and further mechanistic investigations verified
that in situ degradation of the carbocation led to formation of
a Brønsted acid, which proved to be the actual catalyst.[18]
Despite the previously reported inconclusive results on the
application of carbocations as Lewis acid catalysts, we set out
to verify this concept by designing conditions that would ex-
clude the involvement of competing silyl-based Lewis acid or
Brønsted acid catalysis. The Diels–Alder reaction was chosen as
a model reaction to test the catalytic activity of the trityl carbo-
cations,[17,19] avoiding the problem of possible silyl-cation catal-
ysis. In our proposed catalytic cycle, the dienophile (the a,b-
unsaturated aldehyde) will react with carbocation 1, forming
intermediate oxonium ion 7, thus lowering the LUMO of the
dienophile and enabling the pericyclic reaction with the diene
Triphenylcarbenium tetrafluoroborate or trityl tetrafluoroborate
(TrBF4) is an inexpensive commercially available carbocation
that is stable enough to handle without any special precau-
tions. It also constitutes a rather unique mode of carbon-cen-
tered Lewis acidity, with extensive possibilities for relatively
easy tuning of the electronic properties of the carbocation
through variation of the aromatic groups.[5] To our satisfaction,
we found that TrBF4 is an extremely efficient catalyst for the
Diels–Alder reaction between acrolein and cyclohexadiene in
CH2Cl2; addition of only 0.5 mol% gave full conversion in less
than 1 h to selectively provide the endo product in quantitative
yields (Table 1, entry 3). After further optimization, we found
that the catalyst loading could be decreased to 0.1 mol% and
still maintain good activity (Table 1, entry 1). However, turnover
of the catalyst stopped after 48 h and 85% conversion, most
likely owing to catalyst decomposition or inhibition. The opti-
mal catalyst loading was found to be 0.2 mol%, providing the
endo adduct as the only observable isomer, in 94% yield after
48 h (Table 1 entry 2).[20] It is important to stress the order of
addition of the starting materials and the catalyst. When only
cyclohexadiene was added to a solution of TrBF4 in CH2Cl2, the
reaction mixture turned black immediately and complete de-
composition or polymerization of the diene was observed by
1H NMR spectroscopy. However, adding the dienophile to a so-
Chem. Eur. J. 2014, 20, 1066 – 1072
1067
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim