Autonomous Control of Crystal Size and Shape in Dense Suspensions via Imaging

This article presents a novel approach to control the particle size and shape distribution (PSSD) during cooling crystallization at solid loadings representative of industrial processes. The methodology combines stereoscopic imaging with an automated dilution system for high quality size and shape characterization of particles. This data is fed into a process control scheme that cools, heats, and performs wet milling, to reach both a specified particle size and a minimum yield. The effectiveness and adaptability of the control strategy were demonstrated using cooling crystallization of mannitol in water as a case study. Key process parameters, including target particle size, seed loading, and seed population, were systematically varied to assess robustness and reproducibility of the control strategy. The results demonstrate that this method consistently reaches the desired particle size and yield. Different product PSSDs were obtained from a single seed population and consistent product outcomes were achieved using different seed populations and seed loadings, demonstrating the ability of the controller to adapt to process conditions and targets. A significant achievement of this work is the successful operation of the automated dilution and imaging system under supersaturated conditions at suspension densities typical of pharmaceutical manufacturing (5-10 wt%) for extended durations without human intervention. This work advances the state of crystallization control by providing a versatile, compound-agnostic methodology capable of real-time particle characterization and process control under conditions relevant to industrial crystallization. The combination of high-quality particle size and shape characterization with process control has the potential of enhancing reproducibility and reducing process development time.