Communications
Table 1: Effect of catalysts on the reaction of 1a with 2a.[a]
reaction between ROHand R ’Hwould be an ideal process for
À
C C bond formation because preparation of the reactive
materials would not be required and only H2O would be
generated as a side product (Figure 1). This process would
Entry
Catalyst
Yield [%]
Entry
Catalyst
Yield [%]
1
2
3
4
[Pd(PPh3)4]
HCl[b]
InCl3
0
0
5
6
7
8
BF3·OEt2
AlCl3
GaCl3
0
6
52 (49)[c]
45
18
InBr3
Sc(OTf)3
25[d]
[a] All reactions were performed using 1a (1 mmol), 2a (1 mmol), and
catalyst (0.05 mmol) in toluene (2 mL). OTf=trifluoromethanesulfo-
nate. [b] HCl in dioxane. [c] Isolated yield (8.6 g) after distillation from a
Figure 1.
synthesis performed on
a large scale (50 mmol scale, 15 h).
[d] PhCOCH2CHPhCH(CO2Et)2 was obtained in 25% alongside the
desired product 3aa (25%).
significantly save energy during multistep transformations
and separation of the product from the salt. Thus, a catalyst
for the reaction is required that activates either ROHor R ’H
and has low oxophilicity and insensitivity to water.
The Tsuji–Trost coupling reaction between allylic acetate
or its derivatives with a carbanion derived from active
methylenes[1] is an important synthetic tool. Recently, direct
use of allylic alcohols and active methylenes for the Tsuji–
Trost-type reactions has been intensively studied using
palladium complexes in the presence of a base or acid
cocatalyst.[2,3] Copper(i) can accelerate the reaction in the
(entry 3, Table 1). The product was easily isolated by distil-
lation as there was no contamination by salt side-products.
We then explored the generality of the InCl3-catalyzed
reaction by varying the alcohol and active methylene
substrates (Table 2). The reaction of allylic alcohol 1a with
active methylenes 2a–c including diesters, keto esters, and
diketones smoothly proceeded to give the corresponding
alkylated products (entries 1–3, Table 2). The primary allylic
alcohol 1b afforded a regioisomeric mixture (entry 4). The
cyclic allylic alcohol 1c gave the desired product 3cc in high
yield (entry 6, Table 2). Interestingly, the benzylic alcohols 1d
and 1e were applicable in this system as electrophiles
(entries 7–11, Table 2),[8] while such benzylic species are not
suitable in the Tsuji–Trost reaction based on p-allyl chemis-
try.[9] When simple monoketones such as acetophenone or
propiophenone were used as nucleophiles, no desired product
was obtained. Surprisingly, the reaction using a-methoxyke-
tone 2d gave the alkylated product 3ad (entry 12, Table 2).
Higher loadings of 2 improved the yield for the reactions that
initially gave low yields (entries 5 and 13). In all cases,
selective monoalkylation was accomplished without dialky-
lation.[3] Thus, the present system has potential for a wide
range of nucleophiles and electrophiles.
absence of
a cocatalyst but an equimolar amount is
required.[4] Although the catalytic reaction using a cobalt
species was reported,[5] active methylenes were limited to b-
diketones or keto esters, while diesters, which display a much
lower reactivity owing to lower acidity (enolizable ability),
could not be used. Those reactions based on transition-metal
chemistry involve only allylic alcohols because the key
intermediates are a p-allylic species. In this context, expan-
sion of substrates for this type of reaction to a wide range of
active methylenes and other classes of alcohols is desired.
Catalytic activation of alcohols is generally difficult
because of the inefficient leaving-group ability of the hydroxy
group. We have been studying catalytic activation with silyl
nucleophiles and found that indium catalysts are quite
effective.[6] We now report the indium-catalyzed reaction of
allylic and benzylic alcohols with various active methylenes,
alkoxyketones, and indoles, in the absence of a cocatalyst/
activator.
Scheme 1 shows plausible pathways for the InCl3-cata-
lyzed reaction,[10] with the most probable being a direct path
through alkylation of indium-activated alcohol.[6] We also
First, we examined the reaction of the allylic alcohol 1a
with diethyl malonate 2a using various catalysts (Table 1).
Neither a palladium catalyst nor HCl gave the product
(entries 1 and 2). Gratifyingly, InCl3 was found to act as a
catalyst to give 3aa in 52% yield (entry 3, Table 1). InBr3 also
led to formation of the product (entry 4), while no products
were observed using In(OAc)3, In(OH)3, In(acac)3 (acac =
acetylacetonate), or In as a catalyst. Other Group 13 Lewis
acids such as BF3·OEt2, AlCl3, and GaCl3 gave lower yields
(entries 5–7, Table 1). When Sc(OTf)3 was used as a catalyst,
the product was contaminated with that of a redox side
reaction (entry 8).[7] A large-scale synthesis of 3aa was
successful using InCl3 without the need for a solvent
Scheme 1. Plausible reaction courses for the catalyzed reaction.
propose paths A and B from the observation of bis(diphe-
nylmethyl) ether 4d by NMR spectroscopy during the
formation of 3dc in the InCl3-catalyzed reaction of 1d with
2c. A dimeric ether 4 could be an intermediate of the reaction
794
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2006, 45, 793 –796
Yasuda, Makoto
