Prediction of Maximum Yield in the Crystallization of Multicomponent Isomeric Systems

We have developed a computer program (CRYSS), which may be used to rank the performance of a series of crystallization experiments. The method, at its simplest, allows the chemist to take composition data from partially purified crystallized solids and mother liquors; when this is coupled with initial composition data, the program can predict the maximum yield the system is capable of delivering after optimization. In this paper the underlying theoretical basis of the method is investigated by means of the crystallization of a three-component blend of hydrobenzoin isomers, along with the investigation of a four-component system based on aminoindanol. In the latter case, the modulation of purification performance via salt formation is demonstrated.