Solution crystallization processes are challenged by the need to control crystal quality attributes such as the crystal size distribution (CSD). Emulsion solution crystallization is an attractive process intensification strategy to control crystal quality attributes through miniaturization. Droplets of an emulsion can act as tiny crystallizers by confining crystals so that crystal nucleation and growth are limited by the droplet size with a certain amount of supersaturation. This thesis presents a novel process concept that is based on the integration of membrane emulsification and solution cooling crystallization to achieve intensified crystallization.

First, a proof of concept is developed by constructing and characterizing an experimental setup of the integrated process. A water-in-o...[ Read more ]

Solution crystallization processes are challenged by the need to control crystal quality attributes such as the crystal size distribution (CSD). Emulsion solution crystallization is an attractive process intensification strategy to control crystal quality attributes through miniaturization. Droplets of an emulsion can act as tiny crystallizers by confining crystals so that crystal nucleation and growth are limited by the droplet size with a certain amount of supersaturation. This thesis presents a novel process concept that is based on the integration of membrane emulsification and solution cooling crystallization to achieve intensified crystallization.

First, a proof of concept is developed by constructing and characterizing an experimental setup of the integrated process. A water-in-oil emulsion is created through membrane emulsification and glycine is crystallized inside the droplets of the emulsion by cooling. The process is characterized in terms of the attainable CSD and the crystal number density as a function of the emulsification method and supersaturation. Large and monodisperse droplets obtained from membrane emulsification can achieve a narrow and predictable CSD with higher productivity compared to a mechanical emulsification method. The crystal number density is strongly affected by the initial supersaturation when using membrane emulsification but not the CSD. In contrast, the CSD changes with supersaturation when applying a mechanical emulsification method. The CSD obtained from a conventional bulk crystallization is broader and lacks the control imposed by the uniform droplets from membrane emulsification.

The proof-of-concept experiments utilize a so-called crash cooling strategy in which the emulsion is cooled down as fast as possible. However, it is expected that the cooling profile will substantially affect the crystallization performance, which needs to be optimized. Therefore, second, several cooling profiles are designed for emulsion solution crystallization. The experimental setup is extended with a tubular heat exchanger downstream of the membrane to facilitate precise control over the initial temperature. The results show that the cooling profile for emulsion solution crystallization needs to be designed qualitatively differently from conventional bulk cooling crystallization, given the dominance of primary nucleation in emulsion crystallization. The crystal number density is highly sensitive to the cooling profile due to the sensitivity of primary nucleation with respect to supersaturation. However, a narrow and predictable CSD is achieved regardless of the cooling profile due to the confinement of droplets, which restricts crystal growth. In contrast, a broader CSD with a higher crystal number density is achieved in bulk crystallization when using the same temperature profiles, which is consistent with increased secondary nucleation and the absence of crystal confinement.

Given the importance of primary nucleation in emulsion solution crystallization, methods to determine the primary nucleation rates in emulsion solution crystallization systems are needed. Therefore, third, a novel approach is developed to obtain a kinetic model for primary nucleation in emulsion solution crystallization. Two different strategies are explored. First, high-throughput induction-time measurements are conducted in parallel 1-mL batches. The solution in these batches mimics the solution inside the emulsion droplets including the possible presence of surfactants. Subsequently, the obtained nucleation rates are scaled down to a microscale agitated droplet of an emulsion system. The benefit of such a scale-down method is that the nucleation rates as a function of the solution conditions can be obtained rapidly with specialized equipment. However, the accuracy of scaling down nucleation rates is unknown for this type of application. Therefore, second, a direct counting method is developed as an alternative strategy and is compared to the scale-down method. The direct counting method is based on the counting of crystals inside emulsion droplets as a function of time under the actual process conditions of the emulsion solution crystallization process. The comparison reveals that the direct counting method is a substantially more accurate way to obtain a kinetic model for primary nucleation in an emulsion solution crystallization system, which is attributed to a possible impact of the liquid-liquid interface, which is present in the emulsion process but not in the 1-mL model batch system, and the differences in the hydrodynamic conditions.

Process models are important to support the design and operation of future membraneassisted emulsion solution crystallization processes. However, such process models are currently not well available. Therefore, the primary nucleation rate model developed in Chapter 4 is extended to a full process model for the model compound glycine in Chapter 5 of the thesis. The novel stochastic-deterministic process model for emulsion solution crystallization is based on a Markov process to describe stochastic primary nucleation and a deterministic model to describe crystal growth, which are combined with material balances to model the changes of the solution inside the droplets. A growth rate equation is proposed that accounts for the confinement effect of droplets. The growth rate parameters are fitted from experimental data. Subsequently, the validated model is used to characterize the process performance in terms of the CSD and crystal number density as a function of the droplet size distribution (DSD). New insights are revealed on how to manipulate the DSD to get a desired crystal quality. Furthermore, the simulations of emulsion solution crystallization are compared to a conventional bulk crystallization process, modeled using the population balance equation, which reveals the different crystallization behavior between an emulsion and bulk system due to the supersaturation distribution in an emulsion system and the chemical and spatial confinement in the droplets of the emulsion. These factors are caused by the DSD, which is a unique variable for an emulsion system, while only supersaturation is the main variable to control the crystal quality attributes in a bulk system. Especially, the membrane-assisted emulsion solution crystallization process has the flexibility to control the CSD in that DSD can be flexibly designed by changing the membrane properties and the DSD can shape the CSD.