In this work, different process analytical technologies based on vibrational spectroscopy, i.e., attenuated total reflectance Fourier transform infrared (ATR-FTIR) and Raman spectroscopy, were applied by means of multivariate data analysis techniques. Wide applicability has been demonstrated by in situ monitoring of various crystallization processes, e.g., solubility curve measurement, cooling crystallization, and solventmediated polymorph transformation. A calibration strategy has been proposed to obtain accurate and robust estimations of the solute concentration by ATR-FTIR monitoring. Different calibration models and preprocessing techniques were applied and compared. It was shown that these methods allow for solute concentration monitoring of nonisothermal processes even for sparingly soluble substances such as L-glutamic acid in an aqueous environment. An extensive study has been performed to identify the underlying process parameters that influence the Raman signal, i.e., solid composition, solute concentration, suspension density, particle size and shape, and temperature. It is demonstrated that principal component analysis provides qualitative information for seeded and unseeded polymorphic transformations and enables end-point determination of a solid-state transformation process using L-glutamic acid. The multivariate calibration approach described in this work allows for quantitative application of Raman spectroscopy to a multiphase multicomponent dynamic process such as a solvent-mediated polymorphic transformation. Additionally, it was shown that multivariate analysis of Raman data allows for solute concentration estimation despite the fact that solute signals are weak and completely overlapping with signals related to the solid phase.

Design and optimization are important steps during the development of crystallization processes. The combined cooling/antisolvent crystallization of acetylsalicylic acid (ASA) in ethanol−water mixtures is studied by means of experiments and population balance modeling. Model-based approaches require accurate kinetics and thermodynamic data, which are obtained in this work using in situ process monitoring techniques such as attenuated total reflectance Fourier transform infrared (ATR-FTIR) spectroscopy and focused beam reflectance measurement (FBRM). Solubility is measured in situ as a function of temperature and solvent composition using a multivariate calibration model for the ATR-FTIR. Nucleation and growth kinetics are determined based on crystallization experiments by a combination of a population balance model and an integral parameter estimation technique. The model is finally used to calculate optimal cooling and antisolvent addition profiles of the combined cooling/antisolvent crystallization process using a multiobjective optimization approach which optimizes the process with respect to product properties, that is, particle size distribution, and performance, that is, process time. It was found that the solubility exhibits a maximum at 17 wt% water and that the growth rate correlates well with the solubility. No effect of the solvent composition on the nucleation rate could be identified. The optimized trajectories for cooling and antisolvent could greatly reduce the number of nuclei formed as shown through modeling and experiments. The study shows that combining cooling and antisolvent crystallization allows both improving productivity and reducing the formation of fines, and illustrates how process analytical tools and population balance modeling also are effective in crystallization processes where temperature and solvent composition change.

In this work, the polymorph transformation of the metastable R to the stable β polymorph of L-glutamic acid at 45 °C was monitored using in situ Raman spectroscopy. In a series of seeded transformation experiments, the effect of different operating conditions on the transformation was studied. Both increasing seed mass and increasing stirring rate decrease the transformation time, thus suggesting an attrition-based secondary nucleation mechanism of the β polymorph. Moreover, it was found that no pure seed crystals of the metastable R polymorph could be produced and that different sieve fractions of the R polymorph contained different amounts of the β polymorph, which was included within the R crystal. These inclusions had a significant effect on the transformation times meaning that in experiments with larger seeds the transformation was faster than in experiments with smaller seeds. Independent seeded batch desupersaturation experiments were conducted to determine the growth rate of the β polymorph. On the basis of this growth rate and of the seeded transformation experiments, the secondary nucleation rate of the β polymorph was estimated using a population balance model. Together with nucleation and growth kinetics of the R polymorph, which were measured previously, a fully descriptive model of the polymorph transformation process was developed.

No abstract

In this work, the effect of agglomeration on the final particle size distribution is investigated for batch precipitation processes carried out in stirred tank reactors. Agglomeration kinetics of R L-glutamic acid was determined based on seeded batch experiments by a combination with a population balance model and an integral parameter estimation technique. Different modeling approaches for the description of agglomeration are applied and assessed. The empirical model only takes into account the influence of supersaturation and stirring rate on the agglomeration process, while it neglects size dependencies. In the more rigorous modeling approaches, the agglomeration kernel is decomposed into a size-dependent collision frequency and an agglomeration probability. Computational fluid dynamics (CFD) is used to model the turbulent flow in the stirred reactor and to extract information about the shear rate distribution, which in turn can be used to incorporate the dependence of the agglomeration kernel on the local shear rate. The population balance model accounting for nucleation, growth, and agglomeration is used to predict the particle size distribution in precipitation experiments.