Electronic Thesis and Dissertation Repository


Doctor of Philosophy


Chemical and Biochemical Engineering


Sohrab Rohani


Nucleation is a crucial step in the solution crystallization process. Despite their good development, classical nucleation theory and two-step nucleation theory cannot explain all the nucleation phenomena, especially for the non-classical nucleation phenomena which include oiling out, gelation and non-monotonic nucleation. Accordingly, for the non-classical nucleation systems, the crystallization processes are seldom designed based on the nucleation monitoring and supervision. In this thesis, crystallization process optimization was conducted to study the mechanism of non-classical nucleation phenomena and in-line process monitoring technology development.

Two kinds of non-classical nucleation phenomena with non-monotonic nucleation rate and gel formation were investigated, and accordingly, two nucleation pathways that self-induced nucleation and jellylike phase mediated nucleation were proposed based on the analysis of in-line spectral monitoring and off-line sample characterizations. Results indicated the agitation level would affect the pre-nucleation clusters’ existence in the non-monotonic nucleation system, and the properties of solvent determined the formation of jellylike phase and the transformation to crystals. Motion-based objects tracking model and the state-of-the-art neural network Mask R-CNN were introduced to monitor the onset of nucleation and following the crystallization process. Combined with a cost-effective camera probe, the developed real-time tracking system can detect the nucleation onset accurately even with ultrasonic irradiation and can extract much more information during the whole crystallization process. Subsequently, ultrasonic irradiation and seeding were used to optimize a non-classical nucleation system that accompanied oiling out phenomenon. Different frequencies and intensities of ultrasonic irradiation and seeds addition time were screened to optimize the nucleation step, which proved their effectiveness of promoting nucleation and narrowing the metastable zone widths of oiling out and nucleation. A fine-tuning of nucleation step was carried out in a mixed suspension mixed product removal (MSMPR)-tubular crystallizer series. The nucleation step was optimized in the MSMPR stage with the aid of principal component analysis, which enabled the growth of crystals in the tubular crystallizer with preferred polymorphism, shape, and size. The study in this thesis provides insights into non-classical nucleation mechanism and nucleation based crystallization process design and optimization.