Abstract

Time-domain nuclear magnetic resonance (TD-NMR) is a powerful technique for polysaccharide characterization because it provides insights on molecular mobility and hydration-related domains; however, differentiating pectin samples produced by distinct extraction and purification procedures remains challenging when their structural features are highly similar. Therefore, a multivariate analysis strategy based on Saturation Recovery-Magic Sandwich Echo (SR-MSE) signals was developed to improve the differentiation of pectin samples from sugar beet pulp obtained through distinct production methods. SR-MSE signals acquired at four recovery delays were transformed into 1H NMR spectra, which were subsequently decomposed via multivariate curve resolution-alternating least squares (MCR-ALS). The proposed strategy was benchmarked against conventional Carr-Purcell-Meiboom-Gill (CPMG) relaxation analysis via principal component analysis (PCA), MCR-ALS and linear discriminant analysis (LDA). The CPMG relaxation curves mainly reflected broad differences in water mobility and provided limited clustering capability among the most similar samples. In contrast, the SR-MSE derived 1H NMR spectral decomposition enhanced sample clustering and revealed spectral contributions associated with proton domains of different domains. The proposed MCR-ALS strategy also improved the discrimination of production-related factors, particularly purification route and maltodextrin incorporation. Overall, the proposed methodology provides an alternative analysis of TD-NMR signals than conventional relaxation analysis, offering a practical chemometric workflow for differentiating structurally similar samples using low-field NMR.