In Search Of Good Recipes

That chemical preparations do not always work as reported is not new. Stephen K. Ritter wrote about the inflationary trend in yields not too long ago (C&EN, May 30, 2011, page 50). Also not new are efforts to report chemical preparations precisely so as to be consistently reproducible. The venerable publication Organic Syntheses has been providing rigorously vetted procedures, each with more detail than is typically supplied in scientific papers.

The Internet has spawned other means of checking recipes for synthesis. Lately the website Blog Syn has been causing a buzz.

Blog Syn uses the power of crowdsourcing to validate reactions reported in literature. What it does is effectively postpublication peer review, says blogger See Arr Oh, a Ph.D. chemist who works in industry. See Arr Oh runs the blog Just Like Cooking and contributes to C&EN’s blog The Haystack. He started Blog Syn and arranged for others to evaluate procedures that chemists, mostly working in pharma, say they can’t make work. At the end of the exercise, they rate the procedure’s reproducibility and share their findings. “We’re trying to help,” See Arr Oh tells me. “We just want all these awesome reactions to work.”

Blog Syn gained some attention when it launched in January and even more when it tackled an IBX-promoted benzylic oxidation (IBX is o-iodoxybenzoic acid). The reaction, supposedly a highly effective and highly selective means of forming aromatic carbonyl systems, came from Scripps Research Institute California in 2002 (J. Am. Chem. Soc., DOI: 10.1021/ja012127). Commenters at the blog Chemjobber had been lamenting its irreproducibility. See Arr Oh and two others examined the procedure, and in the course of their work, Scripps chemistry professor Phil S. Baran weighed in. Baran was a graduate student in the K. C. Nicolaou lab until 2001 and was a coauthor of the paper describing the reaction.

After much back-and-forth with Baran, See Arr Oh says, the special variables of reaction became clear: The starting materials must be purified; a specific blend of solvents needs to be heated in a certain way; IBX has to be freshly prepared; water must be present. After tabulating the deviations from the original procedure that made the reaction work, Blog Syn judged this reaction “reproducible with optimization.”

Another reason to test a reaction is the Wow! factor, because it is “way too cool,” See Arr Oh says. An example is an iron sulfide-catalyzed preparation of heterocycles (J. Am. Chem. Soc., DOI: 10.1021/ja311780a). An iron-sulfur cluster generated in situ from elemental iron and sulfur efficiently catalyzes “a redox/condensation cascade reaction between 2-amino/hydroxyl nitrobenzenes and activated methyl groups,” according to the paper. The result is an atom-economical route to building blocks for bioactive molecules and natural products, among others.

The reaction is “pharma’s dream,” See Arr Oh says. “It’s green, it doesn’t use toxic solvent, it uses cheap base metal.” When three volunteer experimenters ran it, however, they couldn’t get the published yields. Blog Syn has rated the reaction “moderately reproducible—yields much lower than anticipated.”

Blog Syn is getting a lot of nods for the service it is providing. Whether the work is sustainable is unclear. The ability to test reactions is limited by volunteers’ ability to piggyback on work they do for their day jobs. And the anonymity of most of the experimenters—graduate student Matthew Katcher of Princeton is the exception—could raise questions about credibility.

Nevertheless, chemists preparing to publish should heed See Arr Oh’s appeal: Be critical about reporting procedures; include variables and be specific enough so that a chemist with a bachelor’s degree can run it; aim procedures at the person most likely to do the chemistry; don’t write for other professors.

Some renowned chemists get high marks from See Arr Oh. Baran writes good preparations, as do Harvard’s E. J. Corey and Caltech’s Brian M. Stoltz, he says. They write deliberately, descriptively, specifically. “Sometimes Baran even has pictures, which is fantastic. If you can show me how the reaction looks at a certain stage and mine doesn’t look like it, mine’s not gonna work.”