Future Goals.
For basic industrial control of shape, the enhanced image analysis feedback signal may be all that is required. One can envision using even simple proportional feedback control to adjust the impurity level to push the crystal shape back to the desired form once an excursion is detected. Even with the purposeful deletion of unclassifiable crystals, our sampling rate and image analysis is adequate to maintain control of the model system. These operational difficulties will only become less troublesome as the video cameras and imaging software continue to improve. We wish to push the analysis further, however. We understand the role of the impurity in the crystal structure. The essential understanding of the crystal structure comes from studies by Sherwood and coworkers [S, 71. With this microscopic understanding, we can model the dynamics of the transition of crystal shape in a distribution of crystals. We are constructing a population balance model with two internal coordinates that tracks crystal shape changes. This model is applicable to typical industrial crystallization processes. As far as we know, these results are the first demonstration of real time control of crystal shape. Although serious challenges remain, and the prototype process is somewhat idealistic, we hope these results inspire practitioners to think seriously about application of these principles in suitable industrial processes.