Implementation and Application of Image Analysis-Based Turbidity Direct Nucleation Control for Rapid Agrochemical Crystallization Process Design and Scale-Up

Traditionally, the crystallization unit operation and process variables (e.g., temperature and/or antisolvent addition profile, agitation rate, and seeding parameters) are optimized using the design of experiments (DOE) methodology within the quality by-design (QbD) framework. To achieve rapid process design, reduce the number of experiments, and minimize personnel exposure to toxic chemicals, quality-by-control (QbC)-based methods (i.e., direct design or model-free methods) such as supersaturation control and direct nucleation control can be applied to quickly determine an operating profile of a process, leading the system to the desired critical quality attributes (CQAs). In this work, various direct design approaches were implemented to investigate the impact on the particle length and filtration time of high-aspect-ratio particles. In addition, an alternative image-based direct design approach, called turbidity direct nucleation control (TDNC), was implemented, which improved the filtration time by 2.5 times compared to that in the standard process. The robustness, usability, and scalability of the TDNC approach were investigated by varying the solvent composition and running open loop scale-up experiments. The improved procedure was applied to a commercial-scale crystallizer, resulting in similar improvements in the filtration rate. The commercial scale-up also resulted in a surprising reduction in wash efficiency requiring centrifuge optimization to be addressed.

Automated summary

Traditionally, the design of experiments (DOE) methodology within the quality by-design (QbD) framework is used to optimize the crystallization process. In this study, various direct design approaches and an image-based direct design method called turbidity direct nucleation control (TDNC) were implemented to improve the particle length and filtration time of high-aspect-ratio particles. The TDNC method showed promising results and was further investigated for its robustness, usability, and scalability.