Identification of recycled polyethylene and virgin polyethylene based on untargeted migrants

With increasing attention to recycled polyethylene (PE), its safety as a food contact material has become non-negligible. In this study, we report an untargeted gas chromatography coupled to mass spectrometer (GC-MS) analysis combined with multivariate data analysis for discriminating recycled PE from virgin PE. 80 (semi-) volatile migrants were identified, including 10 hydrocarbons, 29 esters, 3 aldehydes, 9 alcohols, 2 ethers, 4 acids, 4 benzene derivatives, 4 ketones, 3 amides, 2 piperazine derivatives and 10 unknowns. The hydrocarbons in virgin samples showed a greater variety and abundance than those in recycled samples. Based on orthogonal partial least squares discriminant analysis (OPLS-DA) and non-parametric test, 38 markers were selected to establish a classification model to identify recycled and virgin PE. These markers mainly derived from additives, daily chemical products related and food related input pollutants, most of which were rich in recycled samples. Furthermore, in classification models, linear discrimination analysis (LDA) showed higher stability and predictability than soft independent modelling of class analogy (SIMCA). The average accuracy of training and prediction set reached 100% and 92%, respectively, in LDA model. The identification of recycled PE and virgin PE based on migration of untargeted substances is feasible.

Automated summary

This study successfully discriminated recycled PE from virgin PE using untargeted gas chromatography coupled to mass spectrometer analysis and multivariate data analysis, identifying markers for establishing a classification model. Additives and pollutants were found to be important factors for distinguishing between the two, with linear discrimination analysis showing better stability and predictability in the classification model.