Isomerisation of palladium π-allyl complexes

Abstract

The palladium-catalysed asymmetric allylic alkylation is a mild and versatile bond forming reaction between a nucleophile and allylic electrophile. The wide scope of nucleophiles used, and the high regio- and stereoselectivity obtainable renders this transformation an important technique in enantioselective synthesis.

The mechanism is known to go via a key palladium π-allyl intermediate, followed by nucleophilic addition occurring at the terminal allylic carbon. Both the formation of the palladium π-allyl, and the nucleophilic addition to generate the alkylated product and palladium(0) proceed with high levels of inversion of stereochemistry, and both provide an opportunity for the induction of stereochemistry.

However in the case of ligand controlled nucleophilic addition memory effects have been observed. The epimerisation of the palladium π-allyl before nucleophilic attack is key to achieving high levels of selectivity when racemic starting materials and chiral ligands are employed. Previous work in the Lloyd-Jones group has determined that prolonging the lifetime of the palladium π-allyl species, either by the use of weakly coordinating counter ions or slow addition of the nucleophile reduces this memory effect, however increasing the rate of epimerisation would have a result in a similar effect.

One of the mechanisms resulting in the epimerisation of the palladium π-allyl species is mediated by palladium(0), however the details of the mechanism are not well understood.

We describe the synthesis of a diastereotopic palladium cyclohexenyl ester and labelled the complex with 108palladium and d3 at the cyclohexenyl ester. Using simultaneous 31P NMR and mass spectrometry, we have acquired strong evidence against mechanisms involving a single electron transfer, as proposed by Stille, of formation of a dinuclear palladium(I) species followed by an inversion event, and we have gained evidence supporting the direct nucleophilic addition of the palladium(0), resulting in inversion of stereochemistry. The differences in rates of nucleophilic attack involving monodentate and bidentate phosphine ligands on both the palladium I-cyclohexenyl ester have also been explored.

Throughout the mechanistic investigation, it was noted that the 31P NMR spectroscopy experiment used gave non-quantitative results, and in fact the differences in quantification of the species varied with the spectrometer used. We also describe our investigations into where these differences arise from and an optimum set of parameters for quantitative 31P NMR spectroscopy. The conclusions are also applicable to other heternuclear NMR spectroscopic experiments.