In 1990s new anti-HIV drugs were being rushed into production to fight the then relatively new disease HIV. One of these was Ritonavir. It was approved in 1996. Like most drugs, it was sold as tablets that contain crystals of the drug molecule. All was well.
Then in 1998 the tablets started to change. The problem was that Ritonavir forms more than one type of crystal. This is not unusual, most molecules form more than one type of crystal.
For example, there are about ten different types of ice – ten different crystal forms of water. In each form the molecules are arranged in a regular repeating crystal lattice but there are many many different types of lattice, and so the same molecule can be arranged into different lattices. For example, the ice in your gin and tonic is the type sometimes called hexagonal ice because if viewed along a particular direction, the molecules are packed into little hexagons – this is why snowflakes have sixfold (hexagonal) symmetry. Another type of ice is called cubic ice because in it the water molecules are packed into little cubes.
Anyway, back to Ritonavir. Its safety and dosing had been carefully tested in form I, so it was known to be safe in form I. But then the world’s supply started changing to form II – which had not been tested and whose safety was unknown. This was a bit of a disaster. In the end they had to resort to producing it as gel capsules.
This focused minds in the pharmaceutical industry on the question of how to get the crystal form you want, and avoid the ones you do not. Control over crystal form is now well known as an important but tricky problem. I have been thinking about it bit recently.
The formation of a crystal starts off with nucleation: the formation of a tiny, microscopic, nucleus of the crystal. The conventional theory says that the rates of nucleation are sensitive to pretty much everything and can occur with a huge range of values. This suggests that it should be easy to make the rate of the form you want say 10,000 times higher than of one that you don’t want, and then you’re done.
But it isn’t that easy, the conventional theory must be missing something. It may be that the rates are not well defined, there is some evidence for this. Then for each form of the crystal there may not be a single rate but a whole range of them, and so to control which crystal forms we probably have to somehow make sure the two ranges of rates don’t overlap. I am a bit clueless on how to do this but sometimes asking the right the question is half the battle.