There have been a number of successful chiral catalysts that have been developed for the asymmetric synthesis of epoxides.   The same is not true for epoxide’s aza cousins, aziridines.  When we became interested in this subject in 1996 there were no catalytic asymmetric aziridination reactions known that worked for a variety of substrates.  We became interested in the synthesis of aziridines from the reactions of imines with diazo compounds mediated by a Lewis acid.  The first report of general method for the generation of an aziridine from the reaction of an imine and ethyl diazoacetate with simple Lewis acids appeared in 1996 by Brookhart and Templeton. 

Inspired by the work of Brookhart and Templeton, we decided to attempt to render this reaction asymmetric by the use of a chiral Lewis acid prepared from the VANOL or VAPOL ligand [1]  The aluminum catalyst that we had found to be effective in the asymmetric Diels-Alder reaction described above was only moderately successful.  After considerable experimentation with other Lewis acids in combination with VAPOL we eventually found that an extremely effective and general catalyst could be prepared from triphenylborate and VAPOL.  The catalyst is prepared by heating VAPOL and triphenylborate at 55 ^oC for one hour and then removing all volatiles at this same temperature under vacuum.  As indicated by the data in Table 4, the reaction is highly diastereoselective for the formation of cis-aziridines with up to ≥ 50 : 1 cis to trans selectivity and in 55 - 91 % isolated yields for nine different imine substrates.  The enantioselectivity is remarkably high (90 - 98 % ee) over a range of imines which include those prepared from electron-poor and electron-rich aryl aldehydes and from branched and unbranched aliphatic aldehdyes. The reaction is rapid giving 20 turnovers an hour at room temperature for the imine from benzaldehyde which provides the aziridine 185a.  The reaction can be catalyzed with as little as 0.5 mole percent catalyst to give 185a in 80 % yield and  98 % ee after 24 hours.

It was remarkable to find that catalysts derived from both VANOL and VAPOL are effective in this reaction.  The variation between the two catalysts is quite small.  In all other reactions that we have examined to this point, the VANOL and VAPOL ligands have always led to disparate levels of induction.  In the present case, the asymmetric inductions track nearly identically with each substrate and only small differences are seen.   For some reactions slightly higher inductions were posted by the VANOL catalyst and for others the higher induction was observed with the VAPOL catalyst.  What is remarkable is that for both ligands and for all substrates, the asymmetric induction for all combinations falls in the range of 90 – 98 % ee. The diastereoselectivites are very high for both catalysts but the VANOL catalyst gave a selectivity of greater than 50 : 1 for every substrate.

Based on the information that we have at this time, our current working model assumes the formation of a one to one adduct of the VAPOL ligand and triphenyl borate with two of the phenoxy groups replaced by the phenol functions of the VAPOL ligand. We have found that the benzhydryl substituent on the imine is crucial.  With the catalyst generated from VAPOL we found that the benzhydryl imine 189 was far superior to the benzyl imine in terms of yield, rate and asymmetric induction (Scheme XXIV).  The trityl imine was unreactive, presumably due to steric hindrance in the approach to the chiral VAPOL boron complex.