Selective Crystallization of the Metastable α-Form of l-Glutamic Acid using Concentration Feedback Control

Abstract

A systematic methodology is presented for the selective crystallization of the metastable form of a monotropic dimorph, l-glutamic acid, for batch cooling crystallization. Attenuated total reflection-Fourier transform infrared (ATR-FTIR) spectroscopy coupled with chemometrics was used to determine the solute concentration and solubility curves of both α- and β-forms of l-glutamic acid in aqueous solution. The metastable limit associated with secondary nucleation for a seeded system was determined using laser backscattering (focused beam reflectance measurement, FBRM). Batch crystallizations seeded with the metastable α-form crystals following various preset supersaturation profiles were implemented using concentration feedback control which regulated the cooling rate based on the in situ measurement of solute concentration. Batch crystallizations operated at constant relative supersaturation in an appropriate temperature range prevented secondary nucleation of both polymorph types and were successful in selectively growing large metastable crystals with uniform size.

The crystal growth kinetics of alpha calcium sulfate hemihydrate in concentrated CaCl2–HCl solutions

“…As it can be evaluated the supersaturation line remained practically flat especially over the period 20-60 min. This is a satisfactory regulation of supersaturation despite not using a feed-back control system as done in other studies [4] given the non-availability of corrosion-resistant probes for the current highly corrosive crystallization solution. The required equilibrium concentrations and the activity coefficients necessary for the calculation of the relative supersaturation were modeled with the help of the thermodynamic modeling software package OLI Stream Analyzer 3.2 [38].…”

“…For example, the modulus of rupture of commercial α-hemihydrate was reported to be approximately 2 MN/m 2 higher than that of the more commonly used β-hemihydrate. The latter material is produced normally by dry calcination while the former requires the presence of water either as vapor (steam) or liquid phase as such they have different crystallization behavior than typical competing polymorphic systems [4]. In this context it should be added that both forms of hemihydrate, namely the α and β ones have the same crystal structure but the alpha form having superior plaster properties due to its potential to form well developed crystals with fewer cracks and pores [5].…”

“…ATR-FTIR [4,31,32) or electrochemical in-situ monitoring (e.g. ionselective electrodes [33][34][35] or conductivity probes).…”

Comparative study on ATR-FTIR calibration models for monitoring solution concentration in cooling crystallization

“…The recent references [13,15] presented rapid PLS-based calibration methods based on the USZ spectra for real-time measurement of the ammonium sulphate and urea crystallization processes, respectively. By comparison, only a few references [16][17][18][19] suggested that the metastable zone (MSZ) spectra should be taken into account for model calibration because the crystallization process is mainly operated in MSZ. It was proposed [16,17] to combine the USZ spectra with the MSZ spectra for calibration to improve prediction accuracy on the solution concentration of potassium dihydrogen phosphate (KDP) and LGA during crystallization, therefore reducing the averaged prediction error to 0.37 g/L solvent for LGA solution.…”

“…By comparison, only a few references [16][17][18][19] suggested that the metastable zone (MSZ) spectra should be taken into account for model calibration because the crystallization process is mainly operated in MSZ. It was proposed [16,17] to combine the USZ spectra with the MSZ spectra for calibration to improve prediction accuracy on the solution concentration of potassium dihydrogen phosphate (KDP) and LGA during crystallization, therefore reducing the averaged prediction error to 0.37 g/L solvent for LGA solution. Cornel et al [18] used the collected USZ and MSZ spectra together to build a calibration model in terms of the PCR and PLSR algorithms for LGA solution, indicating smaller root mean square error of cross validation (RMSECV) by using PLSR, i.e.…”