Continuous and Discontinuous Anti-Solvent Crystallization: An Integrated Real-time Process Monitoring and Modeling Approach

Continuous manufacturing has been widely used in large-scale chemical and petrochemical industry for several decades. In the past several years, the pharmaceutical industry has started its adaptation of continuous manufacturing. Research and development effort in the area of continuous pharmaceutical manufacturing has demonstrated its promising, technical feasibility, and certain advantages. Challenges and opportunities are coexisting during the implementation of continuous pharmaceutical manufacturing. From a process engineering point of view, thoroughly studying continuous process for each critical unit operation at the initial stage of implementation can certainly help to pave a solid foundation for a wholly integrated continuous manufacturing process line. In addition, integrated real-time PAT process monitoring and modeling approach can provide a powerful strategy for us to gain in-depth understanding of the process thermodynamics, kinetics, and dynamics, thus to help rationale process design and product design, and ultimately the  exercise of feedback and feed forward process control. In this presentation, we will report our results on developing an integrated real-time process monitoring and modeling approach for comparing the continuous and discontinuous anti-solvent addition for a model multi-component pharmaceutical crystallization system (naproxen-Eudragit L100-alcohol-water). The model crystallization system was monitored in real-time via Focused Beam Reflectance Measurement (FBRM) and Particle Vision Microscopy (PVM). The nucleation induction time calculated via our previously established method using real-time FBRM chord-length-distribution (CLD) profile was then analyzed using a modified Classic Nucleation Theory (CNT). Our study demonstrated that, as expected, under otherwise identical conditions, the nucleation induction time for discontinuous antisolvent crystallization process is longer than its counterpart of continuous antisolvent crystallization. Real-time PVM image acquired for batch crystallization demonstrated that a wide crystal size distribution (CSD) is obtained, which is unfavorable to subsequent processing. Real-time PVM image acquired for continuous crystallization demonstrated that a narrow crystal size distribution (CSD) is obtained, which is favorable to subsequent processing and better quality. Therefore, our study demonstrated certain technical differences between continuous and discontinuous anti-solvent addition for a model pharmaceutical crystallization system, their impacts on certain critical quality attributes (CQAs), and how those differences can be interpreted via the classic nucleation theory (CNT).