Abstrag& The reduction of a-aminonitriles with sodium proceeds via an a-aminoradical, which can be engaged in intramolecular cyclization. Reduction does not involve prior dissociation of cyanide to form an iminium species.
The utility of a-amino&riles as versatile intermediates in organic synthesis is well-established.'
In addition to the chemical reactivity characteristic of the individual functional groups, a-aminonitriles participate in a number of reactions where the CN moiety is displaced by hydrogen or an alkyl group.* Thus, reduction with sodium metal,*jsk NaBH ,+*a or LiAlH$e leads to substitution of H for CN, while the Bruylants reaction284f with Grignard reagents effects the alkylative replacement of CN. These reactions have generally been formulated as proceeding through dissociation of the a-amino&rile to an iminium ion plus cyanide, followed by reduction or Grignard addition (Eq. 1).
-CN' Eq. 1
We became interested in this process because of reports that certain a-aminoniniles formed dimeric vicinal diamines under the Bntylants conditions.2dj This result suggests that a-amino radicals might be involved in the cyanide displacement reactions. Since cyclizations of alkenyl u-ammo radicals have demonstrated utility in the preparation of nitrogen heterccycles,3 we began an investigation to determine whether a-amino&riles might serve as useful precursors to a-amino radicals. While our work was in progress, Martin and coworkers reported their studies on radicals derived from reduction of iminium salts4 Recently, K&ma and associates have shown that cyanide displacement as in the Bruylants reaction can be effected with various hetemaryllithium reagents.
with the very interesting observation that a-aminonitriles do not react equivalently to the corresponding iminium species5 This is quite puzzling if the mechanism of Fq. 1 is operable. We have gathered evidence that a-aminonitriles can be reduced directly to aamino radicals without formation of the iminium ion, and report our results here.
In order to detect a radical intermediate, we utilized an a-aminonitrile incorporating the 5-hexenyl cyclizable probe. Thus, alkylation of the anion of N,N-dimethylphenylacetoninilela-c with 5-bromopentene6 gave 1 (Scheme I) in 93% yield. When 1 was treated with N~/NIYI+~ clean and rapid conversion to the benzylamine 2 occurred, with no evidence for formation of cyclized products.
Apparently, any u-amino radical which is produced is rapidly reduced to the anion and pmtonated by solvent. With sodium in refluxing toluene or benzene. on the other hand, a mixture of 2 (30%). 9 (trace), 4a (49%), and 4h (16%) was obtained. The last two, cyclized products are those expected from ring closure of the a-amino radical ai). followed by hydrogen atom abstraction. The structure