PAT monitoring and PBE modelling of batch cooling solution crystallization in the presence of impurities
Résumé
Hindering effects of impurities on the crystal growth are usually assumed to result of the adsorption of impurity species on the crystal surface. In the presence of impurities the growth rate does not depend on supersaturation only, but also on the concentration of impurities and on the time a given particle spent in contact with inhibiting species (unsteady-state adsorption mechanisms). Few kinetic models describe such phenomena. Moreover, published models are derived from data obtained from specific experiments performed on singles crystals, which makes their application to real industrial crystallizers rather questionable. Indeed, for process engineering purposes, the available kinetic inhibition models accounting for the effect of impurities (e.g. Cabrera-Vermilyea or Kubota-Mullin's approaches), have to be evaluated in industrial situations where complex and distributed features of the crystallizing suspensions are involved (e.g. during batch solution crystallization). Population Balance Equations (PBE) modelling offers an invaluable simulation tool for such evaluation. With this aim in view, a comprehensive modelling approach based on in situ continuous and dispersed phase measurements, and specific PBE simulation was developed to represent and better understand the effect of impurities on the development of batch crystallization processes. The cooling solution crystallization of Ammonium Oxalate (AO) in water in the presence of various concentration of Nickel Sulphate was selected as a model system during this study. In situ measurements of supersaturation were performed using ATRFTIR spectroscopy and the CSD was assessed thanks to in situ image acquisition followed by off-line image processing. The experimental results were simulated after estimating crystallization kinetic parameters, including parameters of models describing the inhibiting adsorption of impurity on the growing crystal surfaces. Nonlinear optimization techniques were used to fit the experimental data to the simulated ones.
Origine : Fichiers produits par l'(les) auteur(s)