assume that the expected cis products 2 and 4 arise from the
well-established double inversion mechanism for Tsuji–Trost
reactions, involving formation of a p-allyl complex on the face
opposite to the leaving group followed by attack of the
nucleophile anti to palladium (global retention, Scheme 2 and
Scheme 3). It is anticipated that such a reaction would
proceed through a symmetric p-allyl palladium complex 6
(Scheme 7), implying that the enantioselectivity of the
process directly reflects the ability of the chiral ligand to
Scheme 7. cis-Selective deuterium labeling experiment employing sub-
strate [D]-1. The yield was determined by NMR spectroscopy (internal
standard). See Supporting Information.
Scheme 6. a) Scope of malonate nucleophiles in the trans-selective
diastereodivergent deracemization. The ee values were measured on
the corresponding benzamides. The d.r. values were determined by
NMR analysis of the crude mixture. b) Scope of additional nucleo-
philes for the trans-selective process. Yields refer to analytically pure,
isolated compounds. The yield of 5j was determined by NMR
spectroscopy (internal standard). For 5 and 5i the ee values were
determined for the corresponding benzamides, and for 5j on the
dimethylester; these values refer to the major diastereoisomer. The d.r.
values were determined by NMR analysis of the crude mixture. For 5h,
5I, and 5j, exclusively trans-disubstituted cyclobutenes were obtained,
the d.r. values refer to the center marked with *. See Supporting
Information.
direct the incoming nucleophile to only one of the two
enantiotopic faces of the complex 6. Clear evidence in support
of this mechanism was obtained by treating deuterio-lactone
[D]-1 with a malonate salt in the presence of phosphoramidite
ligand L1b, whereupon complete deuterium scrambling was
found in the final product (see Supporting Information).
Concerning the unprecedented enantioselective trans-
selective process mediated by ligand L2a, it is clear that a
stereo-anomalous step must be involved. Four scenarios can
in principle be envisaged: 1) syn-addition of the catalyst (i.e.,
retention upon oxidative addition), 2) isomerization of the
complexed intermediate, 3) nucleophilic addition onto palla-
dium followed by reductive elimination (analogous to what is
observed for “hard” nucleophiles, see Scheme 2), or 4) quan-
titative epimerization of an intermediate cis-product.
Scenario 4 can be ruled out as we found the cis-products 5
to be perfectly stable under these reaction conditions and, in
addition, the enantiomeric excess of the trace amounts of cis-
isomer 2 formed using ligand L2a was always much lower
than that observed for the trans analogue 5 (See Supporting
Information). A direct nucleophilic addition of a malonate
anion to palladium (Scenario 3) would be difficult to ration-
alize on electronic grounds. Scenarios 1 and 2 are, therefore,
the most plausible ones.
with remarkable enantioselectivities strongly suggest that a
true deracemization process is operative. To our knowledge,
there are no precedents of palladium-catalyzed asymmetric
allylic alkylation with global inversion of configuration for
“soft” nucleophiles, let alone processes that lend themselves
to both inversion and retention of configuration, at will.
Following these results, we were intrigued by the ability of
ligand L2a to bias the Tsuji–Trost reaction of lactone 1
towards formation of the trans product diastereoisomer with
“soft” nucleophiles. Indeed, other classes of nucleophiles,
such as b-ketoesters and azlactones also led to selective
generation of the corresponding trans products (Scheme 6b).
As depicted, b-ketoesters led to good levels of diastereose-
lection with excellent enantioselection for the major isomer.
In the case of azlactones, the intramolecular cyclization that
allows formation of an azabicyclic product in the case of the
cis-selective process is geometrically impossible.[13] The trans
cylobutene aminodiacid 5j was therefore the product of this
reaction.
Considering possible isomerization mechanisms for palla-
dium allyl complexes (Scenario 2), Bꢀckvall and others have
proposed[6d,7b,9] the direct displacement of palladium from a
metal allyl complex by another palladium center. Such a
mechanism, however, requires a strong dependence of the
reaction outcome on the concentration of metal (i.e., the
catalyst loading). In control experiments, we did not detect
any significant change in enantio- or diastereoselectivity upon
Concerning the mechanistic details of this ligand-con-
trolled diastereodivergent deracemization of lactone 1, we
Angew. Chem. Int. Ed. 2011, 50, 12631 –12635
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim