Since this procedure was first published in this series the use of methylsulfinyl carbanion (as the base) in dimethyl sulfoxide
(as solvent) has been found to give better results in many examples than the base-solvent pair described here.3
The Wittig reaction appears to proceed more rapidly in dimethyl sulfoxide
and the yields of olefin are frequently superior. For purposes of comparison with the present procedure the procedure of Greenwald, Chaykovsky, and Corey3
was repeated several times with the following deviations from the published procedure. Triphenylmethylphosphonium bromide from two different batches was purchased from the Aldrich Chemical Co., Inc.
This material contained a small amount (ca.
1–2 percent) of benzene
(as indicated by nmr analysis). Drying as described in part A reduced the benzene
content to less than 1 percent. The apparatus was set up and the purification of cyclohexanone
was carried out as described in the preparation of methylenecyclohexane oxide
(this volume, p. 755
). With these modifications the procedure resembled that described by Monson.4
In several experiments the apparent
) of methylenecyclohexane
clustered about the 8.10 g.
) reported by Greenwald, Chaykovsky, and Corey;3
however, analysis by nmr and glpc showed that the product was a mixture of methylenecyclohexane
in the ratio of 4 to 1. Methylenecyclohexane
may be separated quite easily by distillation (using a spinning-band column
) or by preparative gas chromatography on any of several columns (SE-30, di-2-ethylhexyl sebacate
, silver nitrate
). Thus, the true
yield of methylenecyclohexane
. This is a substantial improvement over the yield obtained by the present procedure. The source of the benzene
was not determined. Using different batches of the various reagents did not alter the results. Omitting the washing of the sodium hydride
resulted in a 50–60% reduction in apparent yield (private communication from D. G. McMahan and H. E. Baumgarten).