(610a) From Batch to Continuous Reactive Crystallization: A Case Study on Beta-Lactam Antibiotics

Authors: 
McDonald, M. A., Georgia Institute of Technology
Grover, M. A., Georgia Institute of Technology
Rousseau, R. W., Georgia Institute of Technology
Bommarius, A. S., Georgia Institute of Technology
Using data from experiments on batch crystallization to design a continuous system is recognized to be a challenge. Additional complications exist when crystallization is coupled with a chemical reaction and reaction and crystallization kinetics are linked through the generation of a supersaturated environment. Should this coupling of kinetics lead to a Class I behavior (i.e. one in which there is significant supersaturation in the product stream), then the product yield and crystal quality, especially as measured by crystal size distribution, become linked [1].

In the present work, penicillin G acylase (PGA) catalyzes the formation of the antibiotic but also catalyzes the degradation of the product into a slightly soluble by-product. In batch, the PGA-catalyzed reactive crystallization process improved selectivity and productivity by 50% and 25%, respectively [2]. Adapting the batch process for continuous operation can improve productivity further while ensuring better quality consistency, as measured by purity and size distribution. However, PGA must be retained within the crystallizer for the process to be economically feasible, complicating product removal. As a stepping-stone that circumvents this issue, a fed-batch process has been used to tune models accounting for nucleation and growth kinetics, reaction kinetics, and catalyst deactivation kinetics.

Achieving high yields while controlling crystal size distribution requires independent control of the residence times of the solutes (reactants) and solids. Issues encountered include washout, where nucleation or reaction or both are insufficient to maintain a stable slurry; enzyme loss, where catalytic activity diminishes due to either PGA escaping the crystallizer or permanently unfolding; and fines-rich size distributions, where crystal residence time is too short for desired growth. Control over the enzyme reaction rate can be used to reach the desired set point within a feasible space defined by the batch operation. The reaction rate of PGA can be tuned by adjusting temperature, pH, and reactant concentration, however all three of these have been found to impact the antibiotic solubility or crystallization kinetics, further complicating the development of a continuous process. A model of an MSMPR with independent crystal and liquor residence times has been used to optimize the continuous process, and an experimental continuous apparatus has been demonstrated to validate the model while operating under more constrained conditions (imposed by issues of scale and equipment). Compared to the previously published batch process, the in silico productivity of the continuous process increases by approximately 2-fold at the sacrifice of approximately 10% of the conversion. The in silico average crystal size is small, on the order of tens of microns, comparable to that obtained in batch. The in vitro experimental results confirm a decrease in conversion compared to batch; however better purity, consistency, and productivity offset this loss. Experimental crystal sizes were also small, however it may be possible to alleviate this issue with fines destruction.

References

[1] Tavare, N. S. (2013) Industrial crystallization: process simulation analysis and design, Springer Science & Business Media.

[2] McDonald, M. A., Bommarius, A. S., and Rousseau, R. W. (2017), Chem. Eng. Sci. 165, 81-88.

Topics: