A Cautionary Tale On Reproducibility

It's a chemistry professor's worst nightmare: A promising and published reaction inexplicably stops working.

John J. Eisch of the State University of New York, Binghamton, found himself in the middle of such a nightmare when Ph.D. student Somnath Dutta graduated and passed along his project on the thermal cleavage of bibenzylic C–C bonds with di-n-butylzirconium diethoxide to John N. Gitua.

To their dismay, Eisch and Gitua tried to repeat Dutta's procedure but they found no C–C bond cleavage whatsoever. So began an eight-month investigation that uncovered two causes of their reproducibility problems (Organometallics, DOI: 10.1021/om061009y).

First, they learned that iron salts catalyze the reaction, suggesting that some metal had leached out of Dutta's well-used magnetic stir bar. Second, they found that the cleavage was photochemical rather than thermal. Dutta's lab bench, 1.5 meters from the overhead lights, was better positioned for light to initiate the reaction, whereas Gitua's bench, 2.8 meters from the lights, didn't get enough illumination.

"What sustained my faith and my determination to ferret out the unknown causes of apparent irreproducibility was that both students are superb and conscientious experimentalists," Eisch says.