A sample wetting strategy to overcome differences in physical morphology between lab-prepared training samples and pharmaceutical process samples for reliable quantitative Raman spectroscopic analysis

Abstract

As a training set for multivariate quantitative analysis, lab-blended powder samples are frequently used due to the facile concentration variation of the individual components. Real pharmaceutical process samples, however, are often granular in form. The different morphologies between training and process samples result in dissimilar Raman spectral features, thereby seriously deteriorating the accuracy of the analysis. To overcome this hurdle, an effective and simple strategy to make physical presentations of both training and process samples similar, called water wetting, was demonstrated in this study. Wetting potentially dismantles process granule samples into smaller aggregates, whereas lab powder samples could be somewhat aggregated by water-bridging. Thus, water wetting would produce similar morphologies of both samples. To evaluate the strategy, samples containing esomeprazole magnesium dihydrate (EMD) as an active pharmaceutical ingredient (API) and five additional components were employed. Initially, training samples were mixed with the individual components at the intended concentration ranges. A partial least squares (PLS) model developed using the Raman spectra of the training samples was inaccurate in determining EMD concentrations in the real process samples. When both training and process samples were wetted prior to analysis, accurate concentration determination was realized due to no significant difference in Raman spectral features between samples. The similar morphologies and crystallinities of the two types of wetted samples were confirmed by analyzing the corresponding SEM images and XRD patterns. The proposed strategy is versatile and expandable for further vibrational spectroscopic analysis of various other samples with different morphologies.