On-line Real-time Characterization Of Crystal Shape,Multidimensional Sizes And Facet Growth Kinetics In Crystallization Processes

Crystallization is an important unit operation that is widely used in the manufacture and purification of high value-added pharmaceuticals, fine chemicals and materials. On-line real-time spectroscopic instrument, also known as process analytical technology(PAT) has been employed in crystallization process optimization and control with the objective of obtaining high quality crystalline products. PAT has been applied to crystallization for the measurement of solution concentration and supersaturation using ATR FTIR(attenuated total reflectance Fourier transform infrared spectroscopy) and UVvis(Ultraviolet–visible spectroscopy), for characterization of the size of crystals using FBRM(Focused Beam Reflectance Measurement) and ultrasound spectroscopy, and for identification of polymorphs using Raman spectroscopy and XRD. In recent years, on-line imaging and image analysis has also attracted much attention for characterization of crystal shape and shape distribution. Crystal shape, shape distribution and multidimensional sizes are vital properties for crystalline products as they can have profound impact on dissolution rate, processing ability(e.g. filtration efficiency and flowability), and drug bioavailability. However, since crystals are moving and rotating in a stirred tank crystallizer, the current instrument based on 2D imaging cannot accurately characterize the shape and size accurately. The purpose of this study is to research on stereo imaging technique for accurate characterization of crystal shape, multidimensional sizes and facet specific growth kinetics. The key research contents are described below:(1) By reference to L-glutamic acid crystallization that produces needle shaped form crystals, non-invasive stereo imaging technique was investigated for real-time characterization of the growth rate in the length direction, i.e., the change of the length as a function of time and operational condition. The result was compared with off-line characterization of the final dried crystals. Off-line characterization was made using a commercial 2D imaging instrument, Morphologi G3, for final dried crystals, during the process, dry crystals were laid down on a slice of glass, avoiding measurement errors in length. The on-line stereo imaging result is also compared with on-line 2D. It was found that on-line stereo imaging gave estimation of crystal that is very close to the off-line result, while on-line 2D imaging gave much smaller size. The work fully demonstrated the feasibility of applying 3D imaging for crystallization monitoring, and the advantage of 3D imaging over 2D. The 3D reconstruction method is based on a triangulation approach, which is effective only for simple crystal structures such as needle, shape and rod like crystals.(2) Based on the study of crystallization of potash alum in a small(200 ml) crystallizer that produces crystals of 26 faces, study was carried out on applying stereo imaging for monitoring crystallization processes that generate complex crystal structures, in particular on the methods of 3D crystal shape reconstruction from 2D images. A stereo imaging camera model was proposed for 3D shape reconstruction of crystals of complex structure. It was found that the stereo imaging camera model gives more accurate 3D shape reconstruction than the camera model approach that was proposed in literature. On the basis the new and more powerful crystal shape reconstruction approach, the growth behavior of the three independent faces {111},{110} and {100} were investigated, and their face specific growth kinetics were established by correlating the growth rates with supersaturation.(3) Stereo imaging based on the newly proposed stereo imaging camera model for 3D crystal shape reconstruction was then applied to stirred tank crystallizer, still based on crystallization of potash alum. It was found that it is still feasible to obtain crystal 3D shape for the dynamically moving and rotating crystals. The work also revealed limitations of the method in the application. In the stirred tank crystallizer, as a result of the reduction in the quality of images, segmentation of crystals from image background and crystal image edge detection can have bring errors,leading to errors in 3D crystal shape reconstruction. This gives clear pointers to future research in further development of the technique. The study on the face growth behaviour found that the {100} face grows the fastest causing the face area rapid reduction, while the {111} face grows the slowest resulting in its area getting larger and larger. The knowledge is valuable information for the control of crystal morphology.