Terahertz spectroscopy for the classification of multilayer versus monolayer consumer plastic waste for enhanced recycling

Abstract

The technology for the identification and sorting of polymers is critical in revalorizing consumer plastic waste for a circular economy. Here, the use of THz Time-Domain Spectroscopy (THz-TDS) is proposed, for the first time, as a new method for discriminating between multilayer and monomaterial samples in the process of recycling plastic packaging waste. Most non-polar polymers are semi-transparent in the THz band, and thus, these waves can be used to inspect multilayer plastic objects and retrieve volumetric information independently of their color. The information retrieved from THz waves has been combined with a K-nearest neighbours (KNN) classifier to determine if the sample is made of a single material (typically, monolayer) or if it is a multi-material multilayer plastic packaging. The machine learning classifier was trained using plastic waste samples with a known structure. Experimental results with plastic waste classified on a conveyor layer at 13 m/min show a success rate for the multilayer vs monomaterial classifier of 89.4 %. The experimental results suggest that the proposed technology has the potential to enhance the efficiency of the recycling industry, increasing the monetization of the waste through an increased purity in the recovered fractions.