If you’re at all involved in producing solid forms of compounds, you’re familiar with the concept of a polymorph. That, put simply, is a different crystalline form, and any given substance can have several. Or a lot more than several. I don’t know what the record is for any single compound, but it’s way up into the dozens, and you never know when a new one might show up. You can get these different forms through all sorts of variations in the way the compound is crystallized or precipitated: different solvents, different temperatures, different types of stirring (yes indeed), different concentrations. . .and we haven’t even gotten to hydrates and solvates, where molecules of water or other solvents are incorporated into the crystal structure as it forms. And those can have polymorphs in turn, or different ratios of water and solvent, and there’s just no end to it.
Polymorphs can behave completely differently from each other. The same substance, completely pure, can come out looking like several unmistakable crystal types (rods, prisms, plates), in different crystalline space groups, with completely different melting points and solubility behavior. Past blog posts on this are here, here, here, and here, and as that last one illustrates, this is definitely not something of just theoretical interest. The saga of the anti-HIV drug Ritonavir is the classic example. That altered solubility behavior is of tremendous importance for drug dosing, and in Ritonavir’s case, a more stable (and less desirable!) polymorph appeared and threatened to disrupt the entire supply of the drug, because after a while, it was appearing more and more in the production line. “More stable”, of course, can easily mean “so stable it would rather not dissolve and get into the bloodstream”. Polymorphs themselves can be patented separately over such meaningful differences, as can methods for their production.
Even very simple compounds can show grindingly complex behavior. Aspirin, for example, exists in two polymorphs (so far) that are so similar that they can both show up in the same single crystal, and it takes good X-ray data (and a good X-ray crystallographer) to work out what’s going on. One of those past blog post links is about the trickiness of phenylalanine, a substance you’d think had been studied to pieces by now. But as that link shows, D-ribose is in the same category: the crystallographic literature for years was all over the place.
How about something even more simple than those? I refer to the most simple amino acid of all, glycine. This new paper has some details on its polymorphs, and the ridiculously subtle effects that cause them to form. There are three that form under “normal” conditions from aqueous solution, and while the gamma one is the most thermodynamically stable, the alpha one is kinetically favored. Adding small amounts of simple alcohols to the aqueous solution starts producing mixtures of the alpha and beta forms, though, and eventually, as the alcohol concentration goes up, you only get the beta form. All classic polymorph behavior. But why? Here’s what the authors found:
Herein, by applying extremely sensitive pyroelectric measurements and impedance spectroscopy at different temperatures, we discovered that methanol, unlike water, is incorporated in a polar mode within the bulk of the growing a-polymorph. Consequently, the solvent molecules create polar domains, which exhibit pyroelectricity (i.e., they develop temporary voltage when exposed to temperature changes). As a result, it follows that the incorporated solvent molecules operate as conventional “tailor-made” inhibitors of this polymorph.
It appears that the alpha-form starts growing in a pyramidal shape, just as it usually does, but then starts incorporating methanol-solvated glycine molecules into its growing lattice. This buries methanol molecules into the crystal as it continues forming, and creates zones of higher polarity because of the different orientation of those solvated glycines (and the slight deformations in the nearest neighbor glycine arrangements as well). The normal arrangement in the crystal is a sort of “canceled-out” glycine dimer, but that gets disrupted into a more polarizable form. That’s the origin of the pyroelectric effect they’re able to detect.
Once this starts happening, that developing surface of the crystal ends up “poisoned”, as some molecular dynamic simulations suggest, by the methanols incorporated into the lattice. The new surface doesn’t recognize the usual glycine-dimer form so readily, and further growth is inhibited – you see truncated pyramids of the alpha-form that have stopped in their tracks. Meanwhile, the beta-form end up with one surface (a more hydrophilic one) similarly poisoned and inhibited, but another one (a more hydrophobic one) is totally unaffected, so the crystal grows more rapidly along that face and produces long rods of the beta-form.
This level of detail answers some questions and raises others. There are probably all sorts of metastable crystal forms (and crystal surfaces) showing up in complex polymorph situations, and the ability to see and characterize these things will (as in this case) give us a better physical picture of what’s going on. But predicting such behavior is another thing entirely. You’d hope as we start to pile up enough detail that predictive power will start to emerge (hello, machine learning). That’s a reasonable hope, but there’s no reason that things have to work out that way. The whole “Big Data” effort across various sciences is putting that view of the world to the test, and sometimes it’s going to work out and sometimes it isn’t. Stay tuned!