Diferenciação de adesivos hot melt por análise multivariada: uma abordagem por FTIR-ATR para controle de qualidade

Abstract

Hot melt adhesives are widely used in the industry, but their complex formulations, which are mixtures of polymers and resins, generate infrared (FTIR) spectra with intense band overlap, making differentiation between suppliers impractical by visual inspection. This work aimed to develop and validate a rapid and non-destructive method, based on FTIR-ATR spectroscopy and chemometrics, to differentiate brands of adhesives. Samples from two distinct brands of hot melt adhesive were analysed. After preprocessing the spectra, PCA was performed, revealing a clear separation of groups by PC1 (45.59% of the variance). The loadings analysis identified the chemical markers of the separation: Brand B was characterized by C=O (carbonyl) stretching band at approximately 1762.5 cm-1, and Band H by C-H (aliphatic) stretching bands at approximately 2911.7 cm-1.Subsequently, a DD-SIMCA classification model was developed. The dataset was divided (70% calibration / 30% prediction) using the Kennard-Stone algorithm, and single-class models were built with PCs. In the external validation step, the DD-SIMCA model achieved 100% accuracy, sensitivity, and specificity, with no classification errors. The results demonstrate that the FTIR-ATR methodology, combined with DD-SIMCA, is a robust and accurate tool with direct applicability in industrial quality control for verifying the identity of raw materials.