Angewandte
Chemie
as opposed to a purely catalytic reaction.[7] We sought to
determine the cause of this kinetic phenomenon.
For the first injection of CHD, Pd(OAc)2 and Pd(OBz)2
gave identical rate profiles (Figure 2). In each case, the rate
(which is proportional to heat flow) had a profile resembling
that of an autocatalytic-like reaction. This result confirmed
our earlier findings (Figure 1). For the reaction starting with
the [Pd(DA)2] complex (5), the maximum rate was observed
from the onset of the reaction.
The diastereoselectivity for the formation of the bisben-
zoate 1 changed over the course of the reaction (Figure 1).
These observations (change in rate and selectivity) suggested
a change in the nature of the catalyst early in the reaction (to
give rise to competing reaction pathways), rather than simply
the liberation of more catalyst into solution (through the
breaking up of aggregates).[8] From our observations, it
appears that there is more than one catalytic species capable
of producing the product. One catalyst (active at t = 0)
produces the benzoate 1 slowly and with low stereoselectivity.
A second catalyst, generated during the reaction, reacts at a
faster rate with greater selectivity. We sought to elucidate
which reaction products could be associated with this
apparent change in catalyst structure.
1H NMR spectroscopic analysis in [D6]acetone indicated
that the carboxylates exchanged rapidly on palladium under
the reaction conditions (Pd(OAc)2, PhCO2H, p-benzoqui-
none), as evidenced by the formation of AcOH. However,
there were no significant interactions observed between
Pd(OAc)2 and either the cis bisbenzoate 1 or the isolated
Diels–Alder adduct 4. Hydroquinone (HQ) showed a weak
interaction with Pd(OAc)2,[9] although substantial complex-
ation was only observed with a significant excess of hydro-
quinone (> 20:1 HQ/Pd). This observation seemed inconsis-
tent with the rate changes that occurred early in the reaction.
If the reaction proceeds through a PdII–Pd0 cycle, as
originally proposed,[1a] interactions of the reaction products
(e.g. benzoate 1, hydroquinone, Diels–Alder adduct 4 (DA))
with palladium(0) may also be possible. When [Pd2-
(dba)3]·CHCl3 and the isolated Diels–Alder adduct 4 were
combined in [D6]acetone, only the unbound dibenzylidenea-
Figure 2. Rate (r) of product formation (cis and trans) versus time with
~
^
Pd(OAc)2 (&), Pd(OBz)2 ( ), and [Pd(DA)2] (5; ). Reactions were
carried out at 40.08C with 1.25 mol% of the palladium compound.
Subsequent injections of CHD for all catalysts (Pd(OAc)2,
Pd(OBz)2, and [Pd(DA)2]) led to positive-order rate profiles
without an induction periods, and significantly all displayed
similar profiles to the initial injection with the [Pd(DA)2]
catalyst 5. Furthermore, the reaction rates with all three
catalysts were now superimposable; that is, each reaction now
contained an identical amount of the same catalytically active
species.
1
cetone (dba) was observable by H NMR analysis; a green
solid was also noted to have precipitated from the solution.[10]
Characterization of this material by NMR spectroscopy in
CDCl3 showed significantly shifted enone resonances and the
apparent desymmetrization of the Diels–Alder system. This
result was consistent with an 18-electron 2:1 Diels–Alder
adduct/palladium(0) complex,[11] as confirmed by HRMS. The
[Pd(DA)2] complex (5) was formed readily on a multigram
scale in either acetone or diethyl ether (Scheme 2).[12]
We used reaction calorimetry to compare the catalytic
activity of the new complex 5 to that of Pd(OAc)2 and
Pd(OBz)2 catalsyts.[13] The reaction heat flow was observed
upon consecutive injections of CHD to a preequilibrated
solution of the palladium catalyst, benzoic acid, and p-
benzoquinone in acetone.[14]
The formation of benzoate 1 was observed by HPLC
methods (under identical conditions to those used for our
initial experiments) using our new catalyst. It was found that
the reaction with the complex [Pd(DA)2] (5) had a linear
reaction profile (was truly catalytic) and proceeded at a much
higher rate than the reactions with the other catalysts
(Figure 3; conversion at 210 min is 60% with 5 versus 15%
with Pd(OBz)2 (compare Figure 1)). Furthermore, the use of
this catalyst led to a consistent diastereomeric ratio and less of
the Diels–Alder side product 4 upon complete reaction. The
above data suggests that the [Pd(DA)2] complex directly
produces the true active catalyst for the oxidation reaction;
presumably, the active catalyst is “[PdDA]” (6), which is
formed from 5 through the decomplexation of one DA ligand
(Scheme 3).
We propose that the bidentate Diels–Alder adduct forces
the reaction to proceed through a cationic pathway, thus
enhancing the rate of oxidation (Scheme 3). Initial computa-
tional analysis[15,16] of this process indicates that the Diels–
Alder adduct encapsulates the palladium atom to a significant
extent to produce an environment in which only one
carboxylate moiety can be present at the stage of diene
activation (complex 8). The forced cationic nature of this
complex may therefore explain the enhanced reactivity of this
catalyst. The reaction displays a zero-order dependence on p-
Scheme 2. Formation of the [Pd(DA)2] complex (5). [Pd2(dba)3] was
substituted for [Pd2(dba)3]·CHCl3 for larger-scale syntheses.
Angew. Chem. Int. Ed. 2009, 48, 5958 –5961
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
5959