(116d) Separation of Enantiomers by Coupled Preferential Crystallization - Impact of Initial Conditions On Process Performance

Separating enantiomers is of persistent importance due to the possible differences in biological effects of either one of the species or the racemate. Extending the array of available chiral separation techniques is thus valuable to tackle the respective separation task.

This contribution focuses on the separation of enantiomers from a racemic mixture by means of preferential crystallization (PC). It has been shown to be a feasible and elegant approach to resolve conglomerate and compound forming systems [1]. In this work D-/L-threonine/H₂O serves as a model system for conglomerates to investigate coupled PC with the emphasis on how varying initial conditions influence the process performance.

Figure 1 illustrates simple batch and coupled PC in a ternary phase diagram as well as a schematic representation of the process setup. The simplest case of batch PC makes use of one crystallizer containing an initially racemic solution (e.g. point A). Upon subcooling at a defined rate to a temperature T_cryst the system enters the metastable zone yielding a clear solution which is not yet subject to spontaneous nucleation of either enantiomer. Addition of homochiral seed crystals (preferred or p-enantiomer) starts the isothermal process which in the beginning is governed by kinetics. These lead to growth and secondary nucleation of the added material. The liquid phase composition follows a trajectory extending into the left (seeded with E₂) or the right half (seeded with E₁) of the phase diagram. At a certain point the counter species (c-enantiomer) undergoes primary nucleation which gradually causes a contamination of the solid product [2]. Stopping the process at final state FS^(A) or FS^(B) respectively is thus mandatory to meet purity demands. However, a high mass of the respective p-enantiomer remains dissolved in the liquid phase resulting in a low product yield.

Figure 1: Principle of preferential crystallization illustrated in a ternary phase diagram. Simple batch (blue and red lines) and coupled PC are considered (green line). The schematic process setup represents a configuration for coupled PC. Setting the pumps inactive allows simple batch operation in each crystallizer.

Figure 2 Comparison of symmetric (a) and asymmetric (b) coupled PC. Shown are liquid phase masses of the respective p-enantiomers as well as the final CSD of L-threonine obtained after each experiment.

Great improvements could be achieved by coupling two vessels seeded with opposite enantiomers and exchanging solid free liquid [3,4]. As a result the driving forces, i.e. the supersaturation in each crystallizer, are raised with respect to the p-enantiomer and lowered regarding nucleation of the contaminant. Under ideally symmetric initial conditions, mainly determined by the amount and size distribution of the seed crystals introduced into each system, the liquid phase will remain racemic up to equilibrium and solid product with 100 % purity can obtained [3,4].

It is however challenging, if not impossible, to have seed crystals of opposite optical activity which are otherwise identical. Even when equal mass and crystal size distribution (CSD) are guaranteed, still the surfaces of each population will differ. As a consequence the seeds grow at different rates producing an asymmetric process course, which in turn can have a negative effect on the final product purity and quality.

In this study asymmetric initial conditions were created by using seed masses differing by a factor of four (2.0 g L-Thr, 0.5 g D-Thr). All other parameters such as liquid phase composition, stirrer speed and temperature were identical in both vessels. Observation of the process evolution was done using density measurement and polarimetry in each crystallizer to obtain the composition of the liquid phase. Solid phase analysis consisted of purity measurement by chiral HPLC and the determination of the CSD.

First results indicate that the impact of the initial asymmetry on product purity of each preferred species is less significant than assumed. In case of a reference run using 0.5 g of both enantiomers as seed material it was greater than 98 % and remained above this value when asymmetric initial conditions were chosen. However, differences in liquid phase composition regarding the p-enantiomers and changes in CSD can be observed as depicted in Figure 2.

A distinct bimodal distribution was obtained under symmetric seeding which is less pronounced in the asymmetric case. The concentration decrease regarding the p-enantiomer in each crystallizer is nearly identical for the first three hours in Figure 2a (symmetric seeding). When different seed masses are used (Figure 2b) a deviation occurs from the very beginning until the system approaches equilibrium. It is worth noting that not only L-threonine leaves the liquid phase at a faster rate but also D-threonine although its seed mass remained the same. A transient increase of supersaturation caused by the fast mass transfer of L-Thr is apparently occurring.

These preliminary results reveal a certain robustness of the process towards disturbances. It is however of great interest to restore symmetry for unequal initial conditions in light of batch to batch reproducibility and to be able to track optimized trajectories leading to high yield and purity while staying within productivity bounds.

This contribution therefore presents results of an investigation of the application of model predictive control (MPC) to accomplish this task. One crystallizer is operated under isothermal conditions serving as the master concerning the rate at which the p-enantiomer leaves the liquid phase. MPC is then used to manipulate the temperature in the other vessel accordingly to adjust the rate of mass transfer regarding its preferred species. The impact on the solid product and overall process performance will be scrutinized in this contribution which will give rise to further improvements of coupled preferential crystallization.

[1]     Lorenz H., Perlberg A., Sapoundjiev D., Elsner M.P., Seidel-Morgenstern A., 2006, Crystallization of enantiomers, Chemical Engineering and Processing 45, 863-873.

[2]     Profir, V.M., Matsuoka, M. (2000), Processes and phenomena of purity decrease during the optical resolution of DL-threonine by preferential crystallization, Colloids and Surfaces A: Physicochemical and Engineering Aspects 164, 315

[3]     Elsner, M.P., Ziomek, G., Seidel-Morgenstern, A. (2007), Simultaneous preferential crystallization in a coupled, batch operation mode. Part I: Theoretical analysis and optimization, Chemical Engineering Science 62 (17), 4760-4769

[4]     Elsner, M.P., Ziomek, G., Seidel-Morgenstern, A. (2009), Efficient separation of enantiomers by preferential crystallization in two coupled vessels, AIChE Journal 55 (3), 640-649