LAUSR

Abstract The objective of this research is to understand how the properties of magnesium stearate (MgSt) affect product performance in a quantitative manner using a multivariate modeling approach. In addition, we explored the feasibility of using NIR and Raman spectra as a surrogate measurement of physiochemical properties in prediction of performance in tablet direct compression. Partial least square models to predict performance attributes (PAs) from MgSt properties or spectra were developed and validated. The model input variables are MgSt physiochemical properties, spectra, key formulation and process parameters. Material physiochemical properties include fatty acid composition, loss on drying, densities, particle size distribution, specific surface area, and solid state properties. The key formulation and process parameters include MgSt concentration, filler type and compression force. The output variables are PAs including tablet ejection force, breaking force and disintegration time. It was found that the prediction of MgSt performance from its properties greatly depends on filler type and PAs of interest. NIR spectra successfully predicted lubricant performance in lactose tablet; however, predictions from Raman spectra were not acceptable. In the cases that the contributing physiochemical properties in performance prediction are sufficiently captured in the spectra, the spectra can be used as an alternative tool to predict excipient performance.