Research On The Application Of Raman Spectroscopy In The Crystallization Process Of A-K

In recent years, study of Raman spectroscopy in industrial production monitoring and control is taking more and more attentions, especially in the industrial crystallization process monitoring and control. Industrial crystallization is a solid-liquid separation technique that involved many problems and the degree of supersaturation plays a key role in the crystallization process.In this paper, Raman spectroscopy was used to measure the A-K solution concentration and set mathematical model which can predict the solution concentration,then a seeded cooling crystallization process of A-K water solution was monitored by Raman spectroscopy, in this way we can study the changes in the whole cooling crystallization process,for instance, the degree of supersaturation.To determine the optimum model for predicting the solution information, two different methods of setting models was compared and researched in this paper. The first one is the linear regression method, this method need the feature peak was selected and finally a relationship was built between the peak ratio and sample solution concentration, this method is simple and useful; Partial least squares method (PLS) is the second method, using this method, we can research relationship between the Raman spectral information and multiple solution property simultaneously. The optimum model was found by choosing proper pre-processing method, proper spectral range (1550～1750cm-1)and number of PLS factors. Through comparing, the most proper pre-processing method is adopting smoothing filtering first and then taking the second derivative, and under this pre-processing method, the PLS factor choose2to build the model. Compared two methods, external validation was implemented to prove the higher accuracy of PLS method than the linear regression method.The optimum model has been built, then it was used to monitor the seeded cooling crystallization process of A-K water solution. Through processing the spectrum information, the change of A-K water solution concentration was obtained, then the solubility as a function of temperature in this process was calculated according to the literatures. However, the real-time information of supersaturation was provided using Raman spectroscopy. In addition, the function relationship between seed loading and the final CSD was established in this study. The results show that the PLS model can be applied to monitor the crystallization process effectively, and remarkable effect of seed loading was found for the final CSD.