Vincent Mazel ^a, Joane Meynard ^a, Felipe Amado-Becker ^b, Pierre Tchoreloff ^a.

^a Univ. Bordeaux, CNRS, Arts et Metiers Institute of Technology, Bordeaux INP, INRAE, I2M Bordeaux, F-33400 Talence, France

^b Research and Development Division, F. Hoffmann-La Roche AG, Basel, Switzerland

Purpose.

Capping is a common defect that can occur during the manufacture of pharmaceutical tablets. Although its mechanism has already been described, a complete understanding of the influence of process parameters on capping is yet to be established. Furthermore, predicting capping for a given formulation remains challenging, despite the presence of various existing capping predictors in the literature. The first focus of this work was to better understand the influence of compaction speed on capping. In a second stage, we attempted to identify the properties that could be measured in order to predict tablet capping.

Methods.

All the experimental work was carried out using a Styl’one Evolution compaction simulator (Medelpharm, Beynost, France). A wide range of classical excipients as well as some common active pharmaceutical ingredients were studied. Machine learning algorithms, specifically decision trees, were employed to construct a predictive model.

Results.

It is well known that the speed of compaction has a significant impact on capping. However, the results obtained indicate that the speed at the end of decompression is the most critical factor, and slowing down only this part can mitigate capping.

The use of decision trees revealed that there are at least two cases that can result in capping. The first occurs when the formulation has both a high in-die elastic recovery and high residual die-wall pressure. The second situation corresponds to products that have very low compaction plastic energy.

Conclusions.

The speed at the end of the unloading process is of particular importance in preventing capping occurrences and could drive new developments in rotary presses. The use of decision trees could have practical applications in the future for conducting risk assessment analyses in relation to capping.