This study proposes a simple approach for the recognition of polyamide 6.9 samples differing in impurity amounts and viscosities (modulated during the synthesis), which are parameters plausibly variable in polymers' manufacturing processes. Infrared spectroscopy (ATR-FTIR) was combined with chemometrics, applying statistical methods to experimental data. Both non-supervised and supervised methods have been used (PCA and PLS-DA), and a predictive model that could assess the polyamide type of unknown samples was created. Chemometric tools led to a satisfying degree of discrimination among samples, and the predictive model resulted in a great classification of unknown samples with an accuracy of 88.89%. Traditional physical-chemical characterizations (such as thermal and mechanical tests) showed their limits in the univocal identification of sample types, and additionally, they resulted in time-consuming procedures and specimen destruction. The spectral modifications have been investigated to understand the main signals that are more likely to affect the discrimination process. The proposed hybrid methodology represents a potential support for quality control activities within the production sector, especially when the spectra of compounds with the same nominal composition show almost identical signals.
Keywords: chemometrics; infrared spectroscopy; polyamide 6.9.