The Birch reduction is pretty interesting to run, especially the first time you do it. Liquid ammonia is not a typical reaction solvent, and condensing it off a cold finger always looks a bit like a magic trick. You’ll be standing there with a beaker of sodium or lithium metal pieces (sitting under solvent!), which were likely carved off with a knife from a larger piece that lives its life in a jar of mineral oil. Sodium metal itself looks like a grey lump or cylinder under those conditions, but as you dig into it (the stuff has a consistency in between cold butter and a hard salami), the fresh surface is brilliantly shiny and metallic, as if somebody had invented spreadable chrome for car fenders. But only for an instant; it tarnishes while you watch. Robert Frost was surely right that nothing gold can stay, but nothing silver hangs around for long either in that column of the periodic table, not if there’s any oxygen or water around.
Or any ammonia. As you add your alkali metal to the solution, you get a startling blue color that forms around each piece of dissolving metal – actual solvated electrons – and at least at first, the blue probably dissipates as it heads out into the solution and meets your substrate. We’re taught in chemistry that reduction is the addition of electrons to a substance, and here you are, reducing your starting material by dunking it in a literal bath of electrons. Eventually, though, it holds on, and you have a frosty blue reaction flask that looks like a very refreshing drink for an alien on a hot day. (Any lab that you’d care to work in would be a hot day for liquid-ammonia-drinking aliens – I always figured that the different additives and substrates in the Birch reactions I’ve run over the years would be like different cocktail recipes for them. . .) Note: more laboratory blue candidates here.
The final part of the workup is simple, anyway: you just pull the flask out of the cold bath and come back in a while when all the ammonia has boiled away. The reason you go to all this trouble, of course, is that the electron-dunk provides you with transformations and structures that can be hard to realize any other way. There aren’t too many good options for taking down an aromatic ring by one double bond’s worth of oxidation state, for example. But no one wants to do this on large scale, which can present a real impasse. Here’s what happens when there’s no way out and you have to do the Birch on an industrial scale: significant time, expense, and effort.
There’s a new paper from the Baran lab, though, that tries to get around these problems. I’ve written about Birch alternatives before, and there have been many, but they all suffer from some of the defects of the original (or add new ones). When you’re talking about pumping electrons directly into your substrate, though, electrochemistry should come to mind. There have been many reports of this sort of thing, naturally, going back to Arthur Birch himself, but electrochemistry itself has generally been an unattractive proposition for synthetic organic chemists, with its own reputation for wonky unscalability. Baran’s group has been trying to change that, and this paper is another step in that process.
The paper really shows you how tricky such optimization can be. Along the way, they noticed a metallic substance plating out on the cathode, and realized that this was lithium metal itself. This was a problem in the early lithium-ion battery days, so the group turned to the sorts of methods that have been used in the battery industry to keep this from happening, specifically tris(pyrrolidino)phosphoramide (TPPA), a (nontoxic) phosphoramide that’s also been used as an HMPA replacement in organic synthesis. The proton source for the reaction needed optimization (dimethylurea worked the best), as did the material of the anode (magnesium instead of aluminum), and the cathode itself needed to be made physically smaller (to increase the current density). That’s the sort of electrochemical thinking that has to be learned; if you’ve never done this stuff before it could be a while before that last one occurs to you.
In the end (after detailed study of the reaction), it doesn’t appear that lithium metal itself or any solvated electrons are the active species in this reaction – rather, the substrates are getting reduced directly on the cathode surface, and that the flow of electrons from that surface is the rate-limiting step. Overall, the reaction looks like a single-electron reduction, then a protonation, then both steps again. A lithium cation/dimethylurea complex is a key part of system, probably sitting right next to the radical anion intermediate, which helps explain the dependence on the proton donor component.
Synthetically, this reaction seems to behave very similarly to the classic Birch, but with no ammonia and no alkali metals, of course. A wide range of substituted aryls react just the way that they’re supposed to, and the carbocyclic rings of heteraryls react preferentially, too, as they should. Very interestingly, it appears that a range of other synthetic transformations that depend on dissolving-metal conditions (ketone reduction, McMurry coupling, reductive cyclizations and ring openings of various kinds) can also run under the same conditions – which is not true of the Birch reaction itself, nor of any of its alternatives.
And finally, the whole thing seems to be easily scaled up. The paper demonstrates this by just stacking more electrochemical modules together and running them in either batch or flow mode. This take it from milligram scale, to 10 grams, to 100-gram scale with virtually identical yields, which makes me think that a commercial Birchomatic machine (Baranomatic?) is in our futures. I wouldn’t mind that at all, and I suspect that there are many other chemists who would be happy to never roll out an ammonia tank again. . .