WE ARE WRITING to report on an accident that occurred in the chemistry department at Northwestern University on Dec. 3, 2010. Unfortunately, one of our advisees was seriously injured. The accident—a reaction mixture detonation—occurred during an attempt to synthesize 2-(tert-butylsulfonyl)iodosylbenzene, a partially soluble form of iodosylbenzene that is particularly convenient for use as an oxygen source in studies of catalytic chemical oxidations, such as olefin to epoxide reactions. The synthesis had been performed about a dozen times previously at Northwestern without incident.
The synthesis procedure was a modified version of a procedure first described by Dainius Macikenas and coworkers (J. Am. Chem. Soc., DOI: 10.1021/ja991094j), which in turn had been adapted from a tested “Organic Syntheses” preparation (Sharefkin, J. G. and H. Saltzman, in “Organic Syntheses”; H. C. Baumgarten, Ed.; New York: John Wiley & Sons, 1973; Collection Vol. 5, page 660). One modification was the use of a higher H2O2/iodobenzene ratio (25 instead of 2.8) while maintaining a similar H2O2 concentration. Likely more relevant was a second modification: the use of 35% by weight (freshly opened) hydrogen peroxide, rather than the 30 wt % solution indicated in the Macikenas procedure and used previously at Northwestern.
We do not know with any certainty what caused the explosion. However, the procedure entails combining aqueous H2O2 with acetic anhydride to form peracetic acid. The water component of the aqueous H2O2 solution should serve to remove excess acetic anhydride. We speculate that if some acetic anhydride remained after conversion of the majority to peracetic acid (the desired intermediate compound) or acetic acid (side product), the anhydride could have combined with peracetic acid to form diacetyl peroxide. This organic peroxide is known to be a shock-sensitive explosive.
If our reasoning is correct, the amount of diacetyl peroxide that potentially can form is greater in the modified reaction. Presumably, the less water initially present the greater the chance of forming the unstable organic peroxide. For a given amount of H2O2, the number of moles of water present in 35 wt % hydrogen peroxide is about 21% less than the number present in 30 wt % hydrogen peroxide. It is sobering to realize that even with 35 wt % hydrogen peroxide, the combined number of moles of water and hydrogen peroxide likely exceeded the number of moles of acetic anhydride initially present—and yet an explosion occurred. It is unclear what the margin of error is with regard to water and hydrogen peroxide concentration versus acetic anhydride concentration. However, we believe that at least some diacetyl peroxide is formed under all reaction conditions.
We emphasize that the above “explanation” and discussion are speculative. Nevertheless, there is support from the patent literature. (See, for example, U.S. Patent No. 3,079,443, “Production of a Solution of Diacetyl Peroxide in Acetic Anhydride.”) In the patented process the coreactant is aqueous H2O2.
At least until the cause of the explosion can be determined, we strongly encourage researchers to consider using alternative, nonperoxide, routes to 2-(t-butylsulfonyl)iodosylbenzene, iodobenzene diacetate, and related compounds (J. Am. Chem. Soc., DOI: 10.1021/ja1069773). More generally, we recommend that aqueous H2O2 and acetic anhydride never be combined—despite the fact that, until now, this has been a commonly used reagent combination in oxidation chemistry.
I e-mailed Hupp and Nguyen to see if they wanted to add anything else. Hupp replied that “Safety glasses did successfully protect the advisee’s eyes.”