(470b) Analysis and Optimization of Different Configurations for Preferential Crystallization

Many of organic molecules being interesting for the pharmaceutical, food and agricultural industry are chiral. Usually, only one of the enantiomers shows the desired properties [1, 2]. Besides, the most commonly used classical resolution via formation of diastereomers, direct crystallization methods have become increasingly important in recent years. An attractive process is enantioselective preferential crystallization [3]. One major limitation of this concept is that the racemic mixture must crystallize as a conglomerate, i. e. a physical mixture of crystals where each crystal is enantiomerically pure. In solution such systems tend to reach an equilibrium state in which the liquid phase will have racemic composition and the solid phase will consist of a mixture of crystals of both enantiomers. However, before approaching this state, it is possible to preferentially produce just one of the enantiomers after seeding with homochiral crystals. The process is based on the different initial crystal surface areas of both enantiomers and the specific driving forces due to different supersaturations. It is worth noting that preferential crystallization is used up to now only for a few substances in an industrial scale. From chemical engineering point of view processes have been mostly developed empirically. Perfectly mixed batch crystallizers, typically used for preferential crystallization, can be described mathematically in a simplified manner using a dynamic, one dimensional model which includes experimentally determined kinetic parameters. Such a model was applied in order to describe the process for the threonine-H₂O system [4]. Based on the simplified approach, attractive and more effective operation modes using several crystallizers can be studied (see e.g. Fig. 1). Besides qualifying each mode individually a comparison between optimized rivaling concepts is possible. This requires optimization of each crystallizer configuration. Typically the productivity is taken as objective function. In our work a modified NELDER-MEAD simplex method [5] was applied. This algorithm was found to be more reliable compared to conventional deterministic methods. Using this optimizer it is possible to determine for each configuration important process variables like mass of seeds, mass of racemate, initial seed size distribution, exchange flow rate between several crystallizers, temperature etc. As an example it will be shown that the productivity of threonine crystallization using the mode shown in Fig. 1 with an initial enantiomeric excess of 2% and under the assumption of trouble-free and simultaneous operation reveals an optimum regarding the exchange. The reason for the occurrence of this optimum is that the initial enantiomeric excess must be first decomposed before both liquid phases are mixed in order to achieve overall growth rates as high as possible and to exploit effectively the "enhancement effect" of this crystalliser configuration.

[1] COLLINS, A.N., SHELDRAKE, G.N., CROSBY, J. (1994): Chirality in Industry: The Commercial Manufacture and Applications of Optically Active Compounds, John Wiley & Sons

[2] COLLINS, A.N., SHELDRAKE, G.N., CROSBY, J. (1997): Chirality in Industry II: Developments in the Manufacture and Applications of Optically Active Compounds, John Wiley & Sons

[3] JACQUES, J.; COLLET, A.; WILEN, S.H. (1994): Enantiomers, racemates and resolutions, Krieger, Malabar

[4] ELSNER, M.P., FERNÁNDEZ MENÉNDEZ, D., ALONSO MUSLERA, E., SEIDEL-MORGENSTERN, A. (2005): Experimental study and simplified mathematical description of preferential crystallization, Chirality 17 (S1), S183-S195

[5] NELDER, J.A., MEAD; R. (1965): A simplex method for function minimization, Comput. J., 7, 308-313

Fig. 1: Simultaneous preferential crystallization process in two separated vessels which are coupled via the liquid phase.