Enantioselective Ring-Opening

Ever since Nugent first reported a practical, catalytic method for the enantioselective opening of meso-epoxides with trimethylsilyl azide (TMSN3), [1] such desymmetrization reactions of epoxides and aziridines using a variety of nucleophiles have been the subject of extensive research. The less developed ring-opening reactions, those of meso-aziridines by carbon and nitrogen nucleophiles, give direct access to enantiopure b-amino acids and 1,2-diamines—two classes of compounds which have broad chemical and pharmaceutical relevance. Li, Fern ndez, and Jacobsen first reported enantioselective ring-opening of meso-aziridines with TMSN3, which were catalyzed by Cr III complexes of tridentate Schiff bases. Then Shibasaki et al. reported Y and Gd complexes as catalysts for related reactions with TMSN3 [3g] and TMSCN. The opening of aziridine rings with TMSN3 under Brønsted acid catalysis, and a similar reaction with aryl amines catalyzed by Nb complexes are also noteworthy. Herein we report the synthesis and application of readily available, discrete dimeric yttrium–salen complexes that catalyze highly enantioselective desymmetrization of mesoaziridines with both TMSN3 and TMSCN. For comparable substrates, the enantioselectivity in the TMSN3-mediated reactions exceed the highest values reported to date. In previous work, we reported that Y alkoxides and salen complexes are exceptionally efficient catalysts for transacylation of secondary alcohols and for ring-opening of epoxides by TMSCN or TMSN3: [4c,5] only 0.01 mol% catalyst is required (substrate/catalyst ratio of 10000:1) in the epoxide ring-opening reactions. Even though the enantioselectivity in these reactions [Eq. (1) in Figure 1; up to 77% ee for TMSCN-mediated ring-opening of epoxycyclohexane] never matched the best reported results (91% ee at a substrate/catalyst ratio of about 9:1), the catalytic efficiency is unparalleled among epoxide ring-opening reactions, and may require a different model to describe the transition state. To explain the second-order dependence of the catalyst on Crand Yb-catalyzed ring-opening reactions, Jacobsen has quite convincingly suggested the involvement of two molecules of the catalyst: one each for the activation of the nucleophile and the electrophile. However, by looking at the structure of our yttrium catalyst (a distorted trigonal bipyramid with yttrium at the apex, see Figure 1), it is difficult to envision how two individual molecules of the yttrium complex can be involved (as indicated by the transition state suggested by Jacobsen) in a reaction that proceeds with such high efficiency. An attractive alternative would be to invoke a dimeric catalyst along the lines of one suggested by McCleland, Nugent, and Finn to explain the observations related to the (alkanolamine)Zr-mediated ring-opening of epoxides by TMSN3 (Scheme 1). Anecdotal support for such a hypothesis comes from the ease with which early transition metals, inclusive of yttrium, form anion-bridged dimers, including an OH-dimer from 2a which we have isolated and characterized. Changes seen in the IR spectrum of a mixture the yttrium–salen complex 2a (Figure 1) and excess TMSCN are also indicative of bridged CN-structures. Kinetic studies based on in situ IR spectroscopy, though tentative, do not rule out such a possibility. The veracity of such an idea notwithstanding, we decided to examine the ring-opening reactions of N-4-nitrobenzoylaziridines catalyzed by fully characterized dimeric yttrium–salen complexes. A variety of yttrium compounds, among them monomeric complexes 1a, 1b, 2a, 2b, and 2c (Figure 2), were prepared from Y[N(SiHMe2)2]3·2THF and the corresponding ligands, as previously described [Eq. (2)]. The dimeric complexes Figure 1. Epoxide ring-opening reaction and structures of yttrium catalysts.