(538f) Integration of Raman Spectroscopy into Upstream Processes As a PAT (Process Analytical Technology) Tool for Real-Time Monitoring

Raman spectroscopy is used in chemistry to provide a fingerprint by which molecules can be identified and quantified. In the present case, the same principle for the determination of temporal profiles of key biochemicals was integrated into a Pichia fermentation process development. There are three stages involved in this study: 1) Evaluation of the feasibility in fermentation process development by building calibration models, and testing their performance.  2) Prediction of new batches under similar condition using a different Raman unit 3) Utilization of the model for different conditions within a fermentation process platform. Calibration models were developed using partial least-squares (PLS) regression employing concentration values obtained offline and spectra collected inline during the upstream processes.

Four compounds were investigated: glycerol, methanol and sorbitol, as the main biochemicals to sustain growth and productivity; and ammonium as the sole nitrogen source. Under routine development conditions, 14 different fermentation conditions, including changes of the strain, media components and process parameter set-points were used for both modeling and test. All models showed high R² and low error from cross validation. Additionally 4 independent batches using new Raman probes and analyzer were used for prediction with a remaining low error. Finally, models were tested on fermentation batches performed under platform process conditions but expressing a different product. Prediction showed low error proving the robustness of the models based on Raman Technology.