Application of multivariate methods to compression behavior evaluation of directly compressible materials

The present study is an approach to describe and predict compaction and tablet properties by a combination of a set of commonly used mathematical descriptors and multivariate methods based on continuous compression profiles. Effects of formulation and process parameters (e.g. composition, powder properties, compression speed) of well-known direct compression excipients of widely plastic, elastic, and fragmentary properties, and binary mixtures thereof were characterized. 2(3)-Full factorial designs with three centre points were applied for Avicel(R) PH 102, Starch 1500(R) and Spherolac(R) 100. Tablets (11 mm diameter) were compressed from hand-weighed powder (of constant true volume) at 104.1 +/- 0.2 MPa using a compaction simulator, yielding highly repeatable data. Heckel equation and work-related parameters were derived. Data were evaluated by multivariate analysis (principal component analysis (PCA) and partial least squares (PLS-2, PLS-1) models). The PCA indicated that Hausner ratio, work of compression (WoC), and tensile strength (TS) are negatively correlated to yield pressure of plastic (YPpl) and elastic deformation (YPel), Emcompress(R) fraction, helium-, bulk-, and tapped density, and particle size. PLS-2 model correlated all design variables, their interaction and square effects with all response variables. These correlations were further quantified for the most important responses (e.g. WoC, TS, YPpl, and YPel) by optimizing separate PLS-1 models. The results were found in accordance with expectations and show the ability of this approach to quantify compression behavior, as a step towards a 'formulation development tool' for tablets. (C) 2009 Published by Elsevier B.V.