(208a) Process Design for Crystal Property Control, Using Model-Based Optimal Feedback Control

Grover, M. - Presenter, Georgia Tech
Crystal properties such as size, shape, and form are critical to performance in downstream processes and final products. However, it is not straightforward to determine the process design and recipe required to generate target properties reliably. Three barriers are (1) the lack of understanding of the underlying crystallization process, (2) the stochastic nature of crystallization (especially nucleation), and (3) the limited availability of in situ sensors that measure crystal properties directly. Here we present an approach to controlling crystal properties using feedback control. The feedback policy is specifically calculated to achieve target properties, using a dynamic model of the process together with in situ sensors for feedback. If fundamental physical models are available, they can be used to calculate the policy, but empirical models learned from data can alternatively be used and may provide a more direct path to implementation of crystal property control.