Using (in)solubility to pull an equilibrium mixture of product diastereoisomers over to the desired

26 January 2004

This old, but extremely powerful process chemistry trick was very elegantly illustrated in a recent symposium presentation*.

The desired aminonitrile (B) can be made by a Strecker reaction with ketone (A).

A screen of a range of different amines RNH 2 indicated that the stereoselectivity of the Strecker reaction was poor. However, with benzylamine, the desired diastereoisomer (stereochemistry as indicated) crystallised out of the reaction mixture in 80% yield. In contrast, nothing crystallised out of the analogous Strecker reactions with p-methoxy- or p-nitro-benzylamine, and poor diastereomeric product ratios of 1:1 and 1.6:1 respectively were observed in these cases.

The above was using racemic ketone (A). Unfortunately, the trick failed with the non-racemic counterpart, owing to the higher solubility of the single enantiomer product compared with its racemic form. In the single enantiomer case, an in situ diastereomeric product ratio of 1.5:1 was observed with benzylamine, confirming that the equilibrium diastereoselectivity for the Strecker reaction is poor, and leading to the conclusion that in the racemic case, crystallisation of one of the diastereomers was the driving force pulling the equilibrium over the desired product.

Preparation of the desired single enantiomer was ultimately achieved by reacting racemic ketone with homochiral a -methylbenzylamine. Upon seeding, the desired homochiral diastereoisomer crystallised out in 37% yield (i.e. 74% of theory).

So the next time you have a stereogenic reaction which is reversible and can give a mixture of diastereomers, and you are screening different solvents to influence the stereoselectivity of the reaction, don't forget to check whether there is a suitable reaction solvent from which your desired stereoisomer readily crystallises.....

* E.D. Moher (Lilly, USA), "Synthesis Studies on a Conformationally Restricted Glutamic Acid Analog", Presentation given at the 21 st SCI Process Development Symposium, Cambridge, UK, 10-12 December 2003.