Evaluation of crystallization kinetics of adipic acid in an oscillatory baffled crystallizer

Brown, C. J. and Lee, Y. C. and Nagy, Z. K. and Ni, X. (2014) Evaluation of crystallization kinetics of adipic acid in an oscillatory baffled crystallizer. CrystEngComm, 16 (34). pp. 8008-8014. ISSN 1466-8033 (https://doi.org/10.1039/c4ce00192c)

[thumbnail of Brown-etal-CrystEngComm-2015-Evaluation-of-crystallization-kinetics-of-adipic-acid-in-an-oscillatory]
Preview
Text. Filename: Brown_etal_CrystEngComm_2015_Evaluation_of_crystallization_kinetics_of_adipic_acid_in_an_oscillatory.pdf
Accepted Author Manuscript

Download (1MB)| Preview

Abstract

For solution crystallization, nucleation can be characterized by the maximum sub-cooling (or metastable limit), which is known to vary with numerous process parameters. The relationship between the metastable limit and cooling rate is of particular interest, as it can be utilized to derive nucleation kinetic parameters. However, this relationship is open to interpretation. This work presents the application of three such interpretations (Nývlt, Kubota and a population balance based method) to a cooling crystallization of adipic acid in an oscillatory baffled crystallizer, a relatively new type of crystallizer with increased studies and applications in continuous plug flow operation. It also considers the role the device employed to detect nucleation events plays in the derived kinetic parameters. The result of this study shows that although all three interpretations can reasonably predict the maximum sub-cooling over a tested range of cooling rates, the linear assumptions in the Nývlt and Kubota interpretations give increased deviations from the experimental data, in particular for faster cooling rates. In contrast to the two aforementioned models, the population balance based method maintains a minimal deviation across the whole range of cooling rates used. In addition, although the population balance method does not consider the sensitivity of detection tools in its implementation, while the Kubota method does, the sensitivity of nucleation detection is reflected in the derived nucleation rate constants. This journal is