J. Am. Chem. Soc. doi:10.1021/ja9085512 (2009)

At some point in their careers, most chemists have scratched a beaker to induce crystallization, but how this trick works is still mysterious. Amanda Page and Richard Sear of the University of Surrey in Guildford, UK, have studied the process using computer simulations.

They modelled scratches as wedge-shaped grooves and found that when the angle of the wedge is optimal, the rate of crystal nucleation is orders of magnitude higher in the wedge than on a flat surface, as is seen in experiments.

Nucleation is fastest when this angle allows a defect-free piece of crystal to fit perfectly in the wedge. So by tuning a wedge angle to fit a particular crystal polymorph, the creation of this form could be favoured over others, the authors say.