Abstract
Optical spectroscopic techniques have been successfully employed in bioprocessing as process analytical technology for real-time process monitoring in numerous applications. The implementation of spectroscopy-based PAT techniques commonly necessitates the generation of representative process data used for calibration and validation of multivariate statistical models for analyzing the sample composition in real-time. To automate the generation of such data, we present a novel assembly of a commercially available chromatography system in combination with a Raman spectrometer for fast and accurate acquisition of Raman spectra. Using the ultra-/diafiltration (UF/DF) process as a case study, our methodology involved the preparation of representative calibration and validation mixtures of phosphate and citrate buffer and lysozyme as a model protein. Chemometric PLS models were calibrated and validated using these datasets, and applied to in-line recorded Raman spectra during a UF/DF experiment. The primary results demonstrated that the novel assembly provides robust and precise offline measurement of Raman spectra, which directly compare with in-line record data. The chemometric PLS models showed good alignment in calibration and validation datasets (R2 and Q2), and could be used to simultaneously monitor the buffer and protein concentrations in real-time during UF/DF. This study provides a simple, commercially available setup for automated acquisition of Raman spectra and demonstrates its straightforward application to bioprocess monitoring.