Raman Spectroscopy as a PAT-Tool for Film-Coating Processes: In-Line Predictions Using one PLS Model for Different Cores

Although Raman spectroscopy has been described as a potential process analytical technique for tablet coating, it has rarely been transferred from academic studies to commercial manufacturing applications. The reasons for this are probably not only the high level of process understanding and experience with multivariate data analysis required, but also the product-dependent elaborate model-building. Hence, this study represents a feasibility study to investigate, whether subtraction of core spectra is a suitable approach to generate versatile models for one specific coating that can be applied on a multitude of different tablet cores. Raman spectroscopy was used to predict the application of coatings on three different tablet cores using PLS regression. The obtained spectra were preprocessed, and differential spectra were calculated by subtraction of the core spectrum from each inline spectrum. Normalization ensured comparability between the spectral data of the different cores. It was shown that in general it is possible to build models for a specific coating suspension that can predict the application of this suspension on different cores. In the presence of a strong Raman marker (TiO₂), promising results were obtained. Without the presence of a strong Raman marker this modeling approach is to be considered critical.