Objectives: Sticking is one of the most common and damaging issues that occur during tablet manufacturing. Sticking is the adhesion of powder onto tooling surfaces during compression. Because of the numerous factors involved in its occurrence, understanding tablet sticking requires the simultaneous investigation of these factors to clarify their possible interactions. However, conducting such a study experimentally can present a significant financial and technical burden. In this study, we aimed to leverage the large amount of data that is usually generated during industrial manufacturing to gain insights into sticking.
Methods: This was achieved by collecting and analyzing a total of 71 historical batches that used an ibuprofen-based formulation. We associate each batch with a hundred parameters, including a qualitative descriptor of sticking, and employ a predefined methodology based primarily on multivariate data analysis.
Results and conclusions: Our results highlight the role of lubrication, water content, and the low melting point of ibuprofen in its sticking tendency. Based on these findings, we propose and discuss an industrial manufacturing data analysis approach to sticking and its associated systematic methodology, consisting of collection, exploration, and data modeling.
Keywords: Tablet sticking; linear discriminant analysis; multivariate data analysis; partial least squares-discriminant analysis; powder compaction; principal component analysis.