Secondary nucleation in stirred vessels

Abstract

The secondary nucleation of crystals in a stirred tank has been studied, both theoretically and experimentally. A theoretical model for the secondary nucleation caused by collisions in a stirred tank crystallizer has been developed. The model considers nucleation to result from crystal-impeller,crystal-crystallizer surfaces and crystal-crystal collisions. It was assumed that the nucleation rate could be calculated from the product of two terms; one giving the collision frequency, the other giving the number of nuclei generated per collision. The resultant expression gives'the nucleation rate as a function of supersaturation, agitation conditions, am.ount of crystals present and crystal size. The model was found to be consistent with correlations from the literature for nucleation rates in continuous crystallizers. Experimentally a technique was developed using batch crystallizers to allow accurate and reproducible measureFients of secondary nucleation rates in agitated crystallizers. The collision nucleation of magnesium sulphate heptahydrate v;as studied in a battery of six identical 600 cm3 crystallizers using this technique. Altogether forty six separate runs were performed. The effects of the following variables were investigated: (1) the number of seed crystals, between one and one hundred (2) the size of the seed crystal, 0.5 to 1.6 mm (3) the solution supersaturation, 0.2 to 2.1 kg of hydrate per 100 kg of solution (4) the stirring speed, 310 to 460 rev/min. Most of the data were obtained using steel impellers. Nine runs were performed using soft polypropylene impellers. The results showed that the nucleation rate with steel impellers was linearly dependent on the number of seed crystals and very strongly dependent on the supersaturation (2.5 power), stirring speed (4.2 power) and crystal size (4.0 power). The nucleation rate with the soft polypropylene impellers was much less (by a factor of one third to one ninth) than the nucleation rate with the steel impellers. This difference between the two impellers vras greatest at the low stirring speed. The results from these batch tests were compared with reported results forcontinuous crystallizers and with the model for collision nucleation. To assist in the use of these correlations in the design of unseeded continuous MSMPR crystallizers a generalized graphical design chart vms developed. This chart allows the direct calculation of the volume of the crystallizer required to provide a specified yield (or product of specified size). Alternatively, the chart can be used to evaluate the yield (and mean product size) from a given crystallizer. A power law dependence on supersaturation for both growrth rate and nucleation was assumed. The nucleation rate (secondary) was also assumed to depend on the amount of solid phase present in the suspension and both crystal surface area and suspension density were considered. The transients of the cystallizer have been solved for certain conditions.