Image analysis: unlocking potential to optimize chemical reactions
Image analysis is a topic that has gained significant attention across various industries, but what potential does it hold for process chemists? By utilizing algorithms to scan a series of images, scientists can extract valuable data that can accelerate successful process development when optimizing reaction workup. The information extracted from the images can help identify important variables such as reaction rate, yield, and purity, ultimately leading to the identification of the most promising pathways for further research.
![]() |
This technology can also be useful for monitoring ongoing reactions, enabling scientists to identify any changes or irregularities in the process. This helps prevent the formation of unwanted byproducts and ensures that the reaction proceeds as intended. The use of image analysis technology can help process chemists improve the rate of experiments, compare experiment conditions, and capture more information to avoid unwanted surprises at scale-up.
Improving API Workup Optimization
During active pharmaceutical ingredient (API) development, process chemists spend significant time on workup optimization but collect much less information during reaction optimization. The cooling rate, dosing rate, and seeding conditions play a critical role in separation efficiency, making the close monitoring of their effects crucial. A recent study at a contract research organization (CRO) evaluated the ability of METTLER TOLEDO’s EasyViewer™ probe and iC Vision™ software to improve workup optimization. Turbidity IA is a powerful image analysis method that runs natively in iC Vision and produces sensitive measurements that can be tracked over time. The new system substantially reduced the cycle time for workup via automated analysis of the inline images under three seeding conditions.
Seeding Study
Turbidity IA trends provided immediate process understanding to improve workup optimization. While the reaction was predicted to require eight hours, the first pass with EasyViewer and iC Vision detected completion after 4.5. This allowed a second experiment to be run the same day and reduced the risk of batch failures at scale from conditions changing during an unnecessary hold. Further examination of the turbidity trend indicated additional process events that could be investigated using the EasyViewer images (Figure 1). Comparison of the same time point across seeding conditions enabled scientists at the CRO to measure the impact of seeding on the reaction workup (Figure 2).
Hidden Risks
The purpose of seeding is to control precipitation and maximize quality. Notable from Figures 1 and 2 is spontaneous precipitation when anti-solvent dosing begins. All three experiments are characterized by a short-lived immiscible phase, quickly followed by spontaneous precipitation of a polymorphic form 1, which differs from the seed polymorph form 2. When the scientists saw spontaneous precipitation of a new polymorphic form, they were concerned about the impact of this on purity and filterability. Based on this finding, they decided that further work should be done to reduce risk and develop a seeding strategy that avoids this risky mechanism.
Advanced image analysis reduces risk by automatically detecting undesired events that could lead to downstream impurities. Using a simple workup sensor that leverages the power of imaging and image analysis, process chemists can easily improve the rate of experiments with real-time endpoint determination, quickly compare experiment conditions, and capture more information—including hidden risks in their process—to avoid unwanted surprises at scale-up.