You might be surprised to know how little we chemists know about what our reactions are really doing. A case in point is the “on water” field. Water is generally not the greatest solvent for a lot of classic organic chemistry reactions, since the reactants, reagents, and products are often not very soluble (or are outright reactive with the water itself). But water would still be a great solvent for many of them if it were possible, since it’s cheap, nontoxic, and has the potential advantage that product isolation can be made easier by the insolubility.
When I was in grad school, Breslow’s group reported acceleration of the classic Diels-Alder reaction when some examples were run under aqueous conditions, and the thought at the time was “solvent pressure” – that the reactants were being literally squeezed together by the hydrogen-bonding network of the water. That sounds weird, but the Diels-Alder is one of those reactions whose transition state actually has a smaller volume than the reactants or the product, so putting pressure on it speeds it up (by lowering the energy needed to get to that transition state, thermodynamically speaking). People have put Diels-Alders in all kinds of high-pressure rigs, and some years ago I wrote about a fun example where the reaction was run in a Teflon tube that was frozen in water ice in a sealed bomb reactor, which gave it a brutally high pressure without special equipment. (I can’t seem to find a link to that one, though!)
Then just a few years ago, Barry Sharpless and co-workers introduced “on water” chemistry, which was defined as reactions that were accelerated by occurring on the water/organic solvent interface. Water at that boundary is most definitely not good old water in the bulk phase, any more than the water molecules in the pockets or around the surfaces of proteins are. There are some reactions that speed by up to hundreds of times under such conditions, and as this new paper from the Zare group at Stanford shows, there is still no consensus about just how this happens:
At first, on-water acceleration was ascribed by Breslow and co-workers primarily to the accumulation of hydrophobic species at the air-water interface. Later, Jung and Marcus put forward the idea that the acceleration was caused by “dangling” OH bonds in the hydrophobic phase surrounding water. However, calculations by Thomas, Tirado-Rives, and Jorgensen were unable to ascribe the rate acceleration to unusual participation of water molecule on the water surface. Since then, Beattie, McErlean, and Phippen have proposed that reaction with water at the interface results in both the protonated substrate and free OH-, which is stabilized by its strong adsorption at the interface. We also know from past experiments on microdroplet chemistry that additional factors causing acceleration may be the lack of three-dimensional solvation at the surface of the water droplet, the speed of two-dimensional diffusion, the large electric field at the water-air interface, and the presence of charged species that preferentially accumulate on the water surface.
Yep, it’s a different world, and the more surface you can provide for such weirdo effects to have their way, the more new chemistry you might discover. This latest paper takes things about as far as they can go, because they’re looking at the microdroplets produced by electrospray ionization, as used in mass spectrometry work. They’re using a model system, the cycloaddition of quadricyclane and diethylazodicarboxylate, that has been used by several others in the on-water field as a benchmark. If you stir those two in toluene, the reaction is pretty slow (24% yield after 24 hours), but Sharpless and co-workers found that the “on water” conditions led to a 40% yield after 4 hours, and almost 70% after 18 hours. Zare’s group finds that the reaction “on droplets” is accelerated by another 115-fold over the Sharpless conditions. Interestingly, if you run the same reaction in deuterated water, you slow down by a factor of five, which is basically the same slowdown that the Sharpless group observed – a pretty strong solvent isotope effect that shows that it’s the O-H bonds that are a big factor somehow.
I hope that it’s possible to find out more about just what’s going on here, so we can manipulate reaction conditions accordingly. Speeding things up by this amount is always welcome, and there’s also the possibility for completely new reactions. This paper, for example, also noted accelerated formation of byproducts from the breakdown of the DEAD itself, and you have to wonder what other things are waiting out there. Running reactions under electrospray conditions like this gives you an instant profile of what’s going on (by feeding the results directly into the mass spec), so this should make reaction discovery much easier.
You might object that this would be hard to scale up, but on the other hand, it’s also (it seems) a classic flow-chemistry situation, where the reaction itself is taking place in a relatively small zone. I could imagine a flow setup with some electrospray apparatus in the middle of it – if you can get things up to a first-pass-is-enough level, it could be pretty interesting. And beyond the practical applications, you do wonder what’s going on here. The details of such effects could prove useful in understanding the solvation shells around ligands and proteins for modeling purposes, and who knows what else.