Abstract:In this paper calibration model building based on using an ATR-FTIR spectroscopy is investigated for in-situ measurement of the solution concentration during a cooling crystallization process. The cooling crystallization of L-glutamic Acid (LGA) as a case is studied here. It was found that using the metastable zone (MSZ) data for model calibration can contribute to the prediction accuracy, compared to the usage of undersaturated zone (USZ) spectra as traditionally practiced which may lead to undesired bias for prediction of the MSZ concentration. Calibration experiments were made for LGA solution under different concentrations. Four candidate calibration models were established using different zone data for comparison, based on a multivariate partial least-squares (PLS) regression algorithm for the collected spectra together with the corresponding temperature values. Experiments under varied concentrations and operational changes of temperature were conducted. The results indicate that using MSZ spectra for model calibration can give more accurate prediction of the solution concentration during the crystallization process. The primary reason was clarified as spectral nonlinearity for measurement between USZ and MSZ. In addition, an LGA cooling crystallization experiment was performed to verify the sensitivity of these calibration models for monitoring the crystal growth process.