Far from crystal clear

A PhD student working with a colleague and I has some great data on crystallisation of a molecule that can crystallise into not one but two different types of crystals (usually called crystal polymorphs). The molecule is a small amino acid called glycine, and the two different crystals we see are the alpha and gamma polymorphs. The motivation is kind of to see how you would get one polymorph or the other, and to understand what is going on. Following earlier workers we have found that if we add salt to the solution we get more gamma and less alpha. So that is relatively straightforward.

But if you look a bit deeper at the data, it is clear that life is quite complex. Naively, you might think that if with salt you get more gamma and less alpha, this is due to some combination of the rate of crystallisation into gamma speeding up, and the rate into alpha slowing down. This is not really what is happening. For (at least) four reasons.

The first reason is that both rates are varying with time, they are not constant. For example, without salt the rate at which alpha crystals forms starts high, then decreases. For both polymorphs, the rates of formation are not numbers, they are functions. The second reason is that these functions change when we add salt. Salt does not just speed up or slow down crystallisation into the alpha form, it radically changes its time dependence, it goes from initially fast and then slowing, to very slow at the start, and then accelerating.

The third reason is that the two processes are entangled, we cannot assume they are independent of each other. The fourth is quite subtle. It is that the two competing processes mask each other: if one of our samples forms a crystal of the alpha polymorph, it cannot then form a gamma crystal as well, so for that sample we have no direct information on gamma crystallisation.

These four problems add up to quite a mess. In order, they say you are trying to find not just two numbers but two functions, that these two functions are in turn functions of salt concentration, that these two functions are not independent of each, and that we are missing a lot of information.

I think the data is good enough to try to for a top journal, but we are going to have to think carefully how to write it. It is an understatement to say that we are asking more questions than we are answering.