Abstract: | A chemometrics approach, multivariate calibration in particular, was used to determine the polymorphism of a drug compound based on Fourier transform infrared (FTIR) spectroscopy. The partial least-squares projection to latent structure makes use of all of the data, and the latent variables created by the method make use of hidden or partially separated peaks for quantitation. This paper illustrates the usefulness of the partial least-squares multivariate calibration method as an efficient tool to determine the polymorphism of a drug. Also, the analysis suggests the use of information from the modeling as diagnostic tools to gain more insight from the data. In particular, the diagnostic tools allow an analyst to assess design characteristics and any shortcomings of a calibration experiment for the polymorphism of a drug compound. |