Other Methods
Why we were doing this chemistry
Known Methodology
Summary of Results

Our efforts to understand the mechanism began with a study of the effect of benzyne on the endoxide, since we initally believed benzyne to be responsible through possible radical interaction with the endoxide. Benzyne was involved, but not through interaction with the substrate!

Benzyne, by itself was shown not to be the cause of this transformation by the following experiment. Benzyne generated by the decomposition of anthranilic acid by isoamylnitrite in the presence of endoxide does not give naphthalene, but instead returns the endoxide and triphenylene. No naphthalene could be detected.


We thus began to consider that either the Grignard reagent, the Mg metal itself, through loss of MgO, or a metal-aromatic charge transfer complex might be involved.

We further narrowed our search by systematically eliminating these possibilities. Freshly purchased, powdered Mg in the presence of 1,4-dihydro-1,4-epoxynaphthalene in refluxing THF, over several days, failed to produce any naphthalene. This turned out to be deceptive, because we were later to find that the Mg simply wasn't in an active enough form!

Second, magnesium dihalides were applied as mimics for the possible action of MgBrF which we thought might be setting up a complex with the heteroatom of the endoxide, eventually resulting in its removal. These experiments did not result in the formation of naphthalene.

Lastly, an e- charge transfer mechanism was eliminated by refluxing the endoxide shown with 2,3-dimethylnaphthalene and powdered Mg. There was no reaction. At the same time, we had found that refluxing the endoxides in the presence of excess Grignard reagents gave naphthalene, but still believed that this result was proceeding by a different mechanism then that of the benzyne chemistry.

Reinvestigation of the original experiment, being more careful this time not to lose any volatile products, revealed a great deal about the mechanism.

The GC/MS trace of the non-purified reactions (simply quenched, so as not to lose any volatile products), revealed 2-bromo- and 2-fluoro-biphenyl.

Reinvestigation of the Grignard reactions in this way also revealed the presence of, in every case, the bis-organic Wurtz-like product of the given Grignard. We confirmed our GC/MS determinations with proton and carbon magnetic resonance experiments.

At this point, we realized that the mechanisms of both the Grignard initiated and 'benzyne' involved route were probably similar. Deoxygenation of the endoxide was probably accomplished by the formation of a highly activated Mg species generated. We believe this to be explained through the action of the Schlenk equilibrium on Grignard reagents.

We believe that the Schlenk equilibrium must not proceed very readily to Mg(0) and organic coupling products very much, because the formation of the coupling product acts as a sink. It could be imagined that only 2 equivalents of the Grignard to endoxide would be needed to complete this reaction, but it doesn't do so, at least not in a reasonable amount of time. Excess Grignard reagent seems to be needed to ensure that enough Mg(0) is formed to carry out the reaction, the residual being lost on workup.

Interestingly, the reactivity of endoxides to Grignard reagents observed here is very different to the reactivity of non-benzo-fused 1,4-epoxides with Grignard reagents. (Lautens, M.; Ma, S. J. Org. Chem. 1996, 61, 7246.)

Back to top of page.