Predicting the physical properties of tablets from ATR-FTIR spectra using partial least squares regression

Abstract

Context: The formulation of a new tablet is a time-consuming activity involving the preparation and testing of many different formulations with the aim of identifying one with the desired properties. In complex formulations it may not be clear which excipient is responsible for eliciting a particular property.

Objective: To investigate partial least squares (PLS) regression analysis of ATR-FTIR spectra of tablets as a predictive and investigative tool in the formulation of novel tablet formulations.

Materials: Magnesium stearate, lactose, acetylsalicylic acid and Ac-Di-Sol.

Results and discussion: ATR-FTIR spectra of a simple aspirin tablet formulation with varying amounts of the lubricant magnesium stearate were obtained. PLS models were built using the spectral data as the multivariate variable and various physical properties of the tablets as the univariate variables. PLS models that allowed good predications to be made for samples not included in the training set were obtained for tablet hardness and disintegration time. It was clear from PLS model regression coefficients that magnesium stearate was responsible for the variation in the tablets’ physical properties.

Conclusion: PLS regression in combination with ATR-FTIR spectroscopy has been shown to be a useful approach for the prediction of the physical properties of tablets.

Keywords::

• Multivariate
• formulation
• prediction
• spectroscopy