The Wharton rearrangement presents an opportunity to convert one α,β-unsaturated ketone to its regioisomeric allylic alcohol. To this end Ohloff, et al.,1,2 explored the transformation of α-ionone 1 into α-damascone 2. Epoxide 3 was exposed to hydrazine in methanol to afford bicyclic allylic alcohol 4 (30%) and the uncyclized allylic alcohols 5 and 6 (20%, ~1:1). The proposed mechanism (3 -> 7 -> 8 -> 9) goes through the vinyldiazene 8, which loses nitrogen to form reactive species. Ohloff proposed a mechanism invoking the vinyl anion 9, which is either protonated by methanol to afford acyclic allylic alcohols 5 and 6 or undergoes cyclization with the endocyclic double bond followed protonation of the resultant secondary anion. The bothersome aspect of this mechanism is, can a vinyl anion in a sea of methanol survive long enough to avoid protonation and undergo cyclization? The same point can be raised about the traditional Wolff-Kishner reaction run at elevated temperature in protic ethylene glycol.
Stork and Williard3 explored the scope of the cyclization under Ohloff conditions finding that both 5- and 6-membered rings can be formed. Cyclization of epoxyketones 10a and 10b formed 11a (85%) and 11b (60%), respectively, as mixture of stereoisomers. In the former example, little of the uncyclized product was formed. These authors suggested a radical-based mechanism because in a typical reaction, the ratio of cyclized-to-uncyclization product was not markedly affected by employing the more acidic β,β,β-trifluoroethanol or the less acidic tert-butyl alcohol as a solvent. The authors argue that if a vinyl anion were involved, the more acidic solvent would be expected to increase the amount of uncyclized product at the expense of cyclized material. An intense yellow color developed which faded with time upon mixing the epoxide and hydrazine independent of whether or not cyclization occurred. A strong UV absorption at 232 nm and a weak one at 409 nm were attributed to the vinyldiazene. A concerted vinyldiazene cyclization was considered but with less enthusiasm.
Taber and Stachel4 designed a set of experiments to resolve this dichotomy based on the known stereochemistry of radical and anionic cyclizations. Tri-n-butylstannane-initiated cyclization of bromide 12a gave an excess of cis-cyclopentane 13a over trans-14a in accord with the cyclization of bromide 12b to 13b and 14b (13b>14b).5 Alternatively, metal-halogen exchange of iodide 15a with t-BuLi gave more of trans-cyclopentane 14a over the cis-isomer 13a in an anionic cyclization in accord with the results of Bailey6 for iodide 15b. The Wolff-Kishner reduction of ketone 16a in triethyleneglycol (TEG) afforded an 11:1 ratio of trans-14a over cis-13a. This result led the authors to conclude that the Wolff-Kishner cyclization is anionic as opposed to radical in nature.
While it is not necessary that the Wharton rearrangement and the Wolff-Kishner reduction must proceed via the same mechanism, in my opinion there is ample opportunity for more definitive results. Should an occasion arise where only the radical or anionic mechanism would apply, then the definitive experiment has been conducted.
1) G. Ohloff and G. Uhde, Helv. Chim. Acta, 1970, 53, 531.
2) K. H. Schulte-Elte, V. Rautenstrauch and G. Ohloff, Helv. Chim. Acta, 1971, 54, 1805.
3) G. Stork and P. G. Williard, J. Am. Chem. Soc., 1977, 99, 7067.
4) D. F. Taber and S. J. Stachel, Tetrahedron Lett., 1992, 33, 903.
5) A. L. J. Beckwith, I. Blair and G. Phillipou, J. Am. Chem. Soc., 1974, 96, 1613.
6) W. F. Bailey, T. T. Nurmi, J. J. Patricia and W. Wang, J. Am. Chem. Soc., 1987, 109, 2442.