Angewandte
Chemie
our next objective was to investigate if enantioselectivity
could be realized on a very different alkene substrate.
We chose to investigate hydroxycarbonyation of norbor-
nene using the dimeric catalysts. The control of the chemo-
selectivity using norbornene is a challenge, and this control
must be accomplished in addition to control of exo/endo
selectivity and enantioselectivity. During the course of our
work, the first promising results for formation of racemic
methyl exo norbornate, as well as an asymmetric synthesis
(40% ee) were published.[7] However, a successful hydrox-
ycarbonylation protocol has not yet been developed.
Our early attempts with norbornene were plagued by
poor chemoselectivity and the formation of low molecular
weight oligomers, thus reducing the selectivity of the reaction.
This problem was also found when using triphenylphosphine-
based catalysts. After some optimization, the new chiral
catalysts delivered high conversion, reasonable chemoselec-
tivity (50 to 65%), high exo/endo selectivity, and up to 95% ee
in the hydroxycarbonylation of norbornene (Scheme 2 and
Figure 1. X-ray structure of complex 2di.
somehow accessed more readily from the dipalladium pre-
catalysts. If the dipalladium complexes form a monomeric
catalyst, then it seems likely that some form of “PdX2” would
be released presumably in a soluble form since the reactions
are homogeneous. An experiment in which the opposite
enantiomer of phanephos was added to the dipalladium
precatalyst was carried out, because it seemed likely that
upon release of “PdX2” the other enantiomer of phanephos
would coordinate to palladium and then deliver a racemic
product. The results showed almost no loss of enantioselec-
tivity in this experiment, but a significant decrease in the yield
(Table S3 in the Supporting Information). The conversion of
the dimer into the active catalyst is therefore rather more
complex, and an intriguing possibility that we cannot rule out
thus far is that the catalytic cycle utilizes dimetallic inter-
mediates having bridging diphosphines. Carbon monoxide,
hydride, and halide ligands are all very competent bridging
ligands that could facilitate this type of cycle. A halide-
bridged dimetallic complex of a monophosphine has been
isolated from a carbonylation experiment, but in this case was
considered to be part of a nonproductive catalytic pathway.[8]
However, given that the monophosphine dimer did promote
the carbonylation in low yield, it is possible that mono-
phosphine dipalladium species are not as long-lived as
diphosphine dipalladium catalysts. Examination of dimetallic
complexes of other chiral bridging phosphines used in other
types of asymmetric catalysis may prove to be an interesting
area for additional study.[5d] It is possible that distinct halide
ligands within the chiral [Pd2Cl2(m-Cl)2] core could be
selectively exchanged for CO and hydride ligands, thereby
making migratory insertion into the alkene complex favored
on one face of the complexed alkene (somewhat reminiscent
of the proposed selective ligand exchanges in the Sharpless
epoxidation using dimeric titanium complexes[9]). Alterna-
tively, given that dual metal systems such as [PdCl2(PPh3)2]
and SnCl2 are enhanced carbonylation catalysts relative to a
mono-metal system, and are postulated to form Pd/SnCl3
species,[10] then it is possible that some form of [PdLxCly] is
released from the dimer, but returns to act as an anionic
ligand for the chiral palladium intermediates at some point in
the cycle, therefore giving the rate enhancements observed. It
is likely that hydrolysis/alcoholysis of the acyl species is the
Scheme 2. An example of one of the first successful hydroxycarbonyla-
tions of norbornene (changing to the S-configured catalyst delivers the
(1S, 2S, 4R)-4 (exo-(2S)-4).
Tables S5 and S6 in the Supporting Information). For this
substrate, both the dipalladium catalysts and monomers give
similar enantioselectivities, and in some cases better reactivity
was encountered using the dipalladium catalysts. The high
enantioselectivity and evidence that the substrate turns over
are very promising for future studies on the applicability of
this class of catalyst.
The use of a precatalyst in which the diphosphine bridges
two metals is extremely rare, and an in-depth study of this
catalyst and its mode of action in carbonylation reaction will
be the focus of future studies. However, to completely
confirm the bimetallic nature of the precatalyst, we have
determined the structure of [Pd2Cl2(m-Cl)2(m-xyl-phanephos)]
using X-ray crystallography. The crystal structure is shown in
Figure 1. Amongst several interesting features, the dimer
possesses a shorter Pd–Pd bond [2.9136(19) ꢀ] than has been
observed in other dimeric palladium halides. The terminal
chloride ligands are located gauche to each other, which is in
contrast to the syn arrangement reported in the study of the
palladium dimers of tritycene diphosphines.[6b] The [Pd2Cl2(m-
Cl)2] core can be best described as possessing an unusual form
of axial chirality, and the mirror image of this core (presum-
ably formed from the opposite enantiomer of diphosphine)
places the terminal chlorines in opposing directions to those
shown in Figure 1.[5c] Our initial working model was that the
real active catalyst would be a monomeric species that is
Angew. Chem. Int. Ed. 2010, 49, 9197 –9200
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim