Mapping the polymorphic forms of fexofenadine in pharmaceutical tablets using near infrared chemical imaging

This study presents a comparison of partial least squares (PLS) and multivariate curve resolution (MCR-ALS) in the quantification of polymorphic forms I and II of fexofenadine.HCl in pharmaceutical tablets using near infrared chemical imaging (NIR-CI). The PLS model built using a standard normal variate (SNV) pre-processing method resulted in satisfactory fits between the reference and predicted values, with a root mean square error of prediction (RMSEP) for both polymorphic forms below 1.5% (w/w). The MCR-ALS results were obtained using an augmented matrix and SNV pre-processing method. The lack of fit value for decomposition was 0.13%, the correlation coefficient between the pure spectra and the obtained spectral profiles was 99.94% and the RMSEP was below 6% (w/w). The MCR-ALS model efficiently quantified the polymorphic forms and generated distribution maps; however, the PLS model exhibited better recovery of the concentrations.