Angew. Chem. Int. Ed. http://dx.doi.org/10.1002/anie.201106321 (2011)

Synthetic routes to enantiopure compounds often rely on resolution methods — removing the unwanted enantiomer — which limit the theoretical yield to 50%. In contrast, deracemization processes convert a mixture of enantiomers to a single enantiomer of product. There are several different classes of deracemization processes, but all of them rely on the transformations of a single stereocentre and thus provide access to only two possible products. Now, Nuno Maulide and co-workers from the Max-Planck-Institut für Kohlenforschung in Mülheim, Germany, have developed methodology that converts a racemic cyclobutene–lactone into any one of four diastereomeric disubstituted cyclobutene products. The conversion is highly selective and controlled by the choice of chiral ligand.

The methodology relies on a palladium-catalysed allylic alkylation (the Tsuji–Trost reaction) with a soft nucleophile. The expected mechanistic course of this type of reaction would be nucleophilic attack of the palladium catalyst on the lactone from the face opposite the carboxylate leaving group, resulting in ring opening and formation of π-allyl palladium species. In a second step, a soft nucleophile would be expected to attack the π-allyl from the face opposite the palladium. Such a double inversion leads to the cis-substituted product, and Maulide and co-workers could indeed make either enantiomer of this product using a chiral phosphoramidite ligand.

Using a phosphine–oxazoline ligand, however, resulted in the trans-product — again in highly selective fashion. Maulide and co-workers identified four possible scenarios that might lead to this outcome, and ultimately conclude that it is the result of an unusual attack of palladium from the same face as the carboxylate leaving group in the first step of the reaction.