We chemists spend a lot of time doing things in the solution phase. It makes sense – if you want things to react, getting all the partners dissolved in some medium where they can roam around and contact each other is surely the way to go, most of the time. But it’s also true that we don’t really have a good grasp of what’s happening in solution at the molecular level. Solvation shells, nanoscale aggregates and clusters, effects of interfacial regions: the idealized picture of things floating around is complicated by a lot of real-world effects, which can be very important. They’re particularly so when it comes to dissolution and its opposite, precipitation and crystallization.
One example is the formation of microcrystals. I know that I had a mental picture of these that is probably incorrect. As I imagined it, all sorts of nucleation events happened in solution but the size of the resulting crystals was constrained (perhaps by amount of available solute, or by effects of stirring) so that they just didn’t grow very large. But in many systems, it appears that what happens first is a liquid-liquid phase separation (sound familiar?), what’s called spinodal decomposition into microscopic solute-rich and solute-poor droplets, and the solute-rich ones then crystallize.
This new paper provides another example of liquid-phase weirdness. The authors are looking at a very well-known reaction, reduction of a metal salt to the elemental metal. That sentence hides a lot of complications: for example, what form does this metal arrive in? You can get a mirror deposit on the inside wall of the vessel (as with the famous old Tollens reagent or the other more stable mirror-silvering mixtures), but more often you get some sort of fine black powder. Microscopic examination of that, though, reveals a huge range of particle sizes and morphologies depending on conditions, and it’s just those changes that can make a big difference in (for example) catalyst performance. Finely divided high-surface-area metals (palladium/platinum on carbon, Raney nickel, Rieke metals) are very important in synthetic chemistry, and their properties vary tremendously.
This work provides an ingenious variation. The authors take an aqueous solution of the metal salt (silver nitrate in the first example), freeze it in liquid nitrogen, and then let this ice chunk slowly dissolve in a cold solution of the reducing agent (sodium borohydride). That gives, under the right conditions, essentially an atom-by-atom release of silver ions into the solution, and what you get out is atomically dispersed metal. By electron microscopy, the great bulk of the material is produced as what are apparently single silver atoms, not clusters, aggregates, etc. (there’s also further characterization by EXAFS). I’m surprised that it doesn’t form those on standing, but as I said before, there are a lot of things about the solution phase that my mental models don’t handle well. The authors believe (on the basis of modeling calculations) that the solvation shell around the single metal atoms is a barrier to forming such nanoparticles, and that the release of individual metal atoms is thus key to the whole process. Dimers of silver atoms and the like seem to be just fine in aqueous solution if they can form (or if they’re whittled down to that size); this synthesis just doesn’t let it happen.
The paper demonstrates the same sort of chemistry with a whole list of metals – platinum, palladium, cobalt, nickel, copper, iridium, gold and more. It certainly seems to be a general process, and the resulting atomically dispersed metals would, you’d have to imagine, be very active catalysts in further reactions. What’s more, this entire method would seem to have applications in many other reactions, too, where the slow release of one reactant into an excess of another is important. I’m actually kicking myself not to have thought of something like this – in hindsight it seems like a perfectly simple idea, but try coming up with these in the forward direction!