Assessment of Prediction Models for Punch Sticking in Tablet Formulations

Abstract

Punchsticking is a common tablet compression manufacturing issue experienced duringlate-stage large-scale manufacturing. Prediction of punch sticking propensityand identification of the sticking component is important for early-stageformulation development. Application of novel predictive capabilities offers early-stagesticking propensity assessment. 16 API compounds were utilised to assess punchsticking prediction using removable punch tip tooling. API descriptors weretested for sticking correlation using multivariate analysis. NIR imaging, SEM-EDXand Raman microscopy were used to examine the material adhered to the punch tips.Predictive modelling using linear and non-linear equations proved inaccurate inpunch sticking mass prediction.  PCA analysisidentified sticking correlated physical descriptors and provided a dataset andmethod for further descriptor studies. Raman microscopy was identified as asuitable technique for chemical identification of punch sticking material, whichoffers insight towards a mechanistic understanding.