Discrimination between Single Crystals and Agglomerates during the Crystallization Process

Abstract

A setup for online measurement of crystallization processes was developed, where particle images were taken and in combination with artificial neural networks (ANNs) were used to convert the particle size distribution into populations of single crystals and agglomerates. For generation of an ANN that discriminates all particle sizes equally well, the training set composition was investigated while proportional similarity was applied for image descriptor selection. Adipic acid/water served as a model system. The effect of the measurement setup on experiments performed for normal cooling crystallization was evaluated. It was found that the major challenge of measuring crystal agglomerates lies in superimposing aggregates that falsify the online measurement results, as offline measurements demonstrated.

In situ Image Processing and Data Binning Strategy for Particle Engineering Applications

“…It could provide a feedback strategy and decision‐making tool in addition to the direct, classic monitoring function. The approach may be also improved for more general applications using sophisticated machine‐vision processing steps and neural networks .…”

Towards Continuous Primary Manufacturing Processes—Particle Design through Combined Crystallization and Particle Isolation

“…By the application of a peristaltic pump (Watson Marlow, d i = 6 mm, Pharmed) installed behind the flow cell to avoid solid attrition or breakage before measuring, the suspension was pumped through the cell (gap: 2 mm; 1024 × 1024 pixel resolution; module M6) and analyzed by the laser sensor. The image acquisition and quantification of PSD were executed by artificial neural networks implemented in MATLAB [30,31]. Furthermore, the distinction of single and non-single (aggregated & agglomerated) crystals as subpopulations of the overall PSD via particle classifier was made [30,31].…”

“…The image acquisition and quantification of PSD were executed by artificial neural networks implemented in MATLAB [30,31]. Furthermore, the distinction of single and non-single (aggregated & agglomerated) crystals as subpopulations of the overall PSD via particle classifier was made [30,31]. In this case, collectives of primary particles arranged and connected via crystal bridges starting from a supersaturated suspension was named agglomerate, while, in the case of aggregates, weak interactions at the edges of the particles could be cemented together to form an agglomerate or could be disrupted again [32].…”

“…networks implemented in MATLAB [30,31]. Furthermore, the distinction of single and nonsingle (aggregated & agglomerated) crystals as subpopulations of the overall PSD via particle classifier was made [30,31].…”

“…Figure 7 shows the results of the averaged overall PSDs (left) of the seed crystals (seeds), after the SFC (SFC OUT ) and after the CVSF (CVSF OUT ) as well as the corresponding subpopulations (right) of single crystals (sc) and non-single crystals (agg). Particle size d eq [µm] CP-Q2 The grey area shows the analyzation limit for particle classification of 80.6 µ m [30]. Operating conditions as shown in Table 1.…”