Here’s another paper with a reaction that would have looked like magic to me back when I first learned organic synthesis. The Wendtland group at MIT details a way to change tertiary carbon stereochemistries, flipping them/scrambling them through the use of a photochemical decatungstate-catalyzed radical reaction.

Examples of this are shown at right – all of these go in around 75-85% yield. Now, these these dimethylcyclohexanes provide an equilibrium ratio of the products corresponding to their thermodynamic stabilities, and that’s just what you’d expect, right? That also holds for things like epimerizing cis-decalin rings into trans (which is a reaction that I really could have used at one point in my grad school work, I can tell you). But that’s not always the case! They show other examples where thermodynamically less stable isomers are formed, and the only easy explanation for that is the production of different photochemically-induced species that apparently have different stabilities. 

The reaction can be carried out in the presence of functional groups like tertiary alcohols, methyl esters, ketones, and Boc-protected amines, which is very nice to see. There are numerous examples in the paper of taking complex frameworks such as sesquiterpenes and steroid derivatives and epimerizing single chiral centers. Cedrol, to pick one, has its isolated methyl group flipped in 50% yield, and there is absolutely no shorter way to get ahold of that epi-compound. The normal products of conjugate addition reactions, arene hydrogenations, radical cyclizations, Diels-Alder cycloadditions and other reactions can be epimerized as well, similarly giving access to stereoisomers that would be (in many cases) quite painful to obtain.

Overall, the reaction targets the most electron-rich CH bonds that it can reach without too much steric crowding, so all-carbon methine hydrogens (and particularly those involving methyl substituents) are your best bet. Other n-alkyl groups and even things like phenyl and isopropyl groups can be isomerized, but that only works efficiently if there aren’t more likely targets available on your substrate.

As the paper notes, this is a great opportunity to take the easier way for some transformations, even if it gives you the wrong stereoisomer, because you can just flip it around afterwards. I think it’s also a neat opportunity to generate new isomers of known compounds (natural products, of course, but many more as well, including med-chem SAR compounds), giving you some instant and relatively painless chemical diversity.