Container Glass Cullet Analysis and Differentiation by Laser‐Induced Breakdown Spectroscopy

ABSTRACT Silica‐based glasses are the most widespread glass family, being soda‐lime glass containers extensively used in the industry. Even though glass can be infinitely recycled, the glass container manufactured using cullet requires some standards to enable the recycling, such as the glass color and the presence of non‐compatible materials. This study explores the spectroscopic information obtained by laser‐induced breakdown spectroscopy (LIBS) combined with a support vector machine (SVM) with a k ‐fold cross‐validation approach for the differentiation of materials in cullet. Bottles from two distinct beer brands were selected for the glass families amber, green, and flint/clear, along with borosilicate glass and ceramic samples, for classification as non‐compatible materials frequently found in cullet. Using a 10‐fold or 4‐fold cross‐validation and a linear kernel, the approach enables the classification to reach an accuracy of 0.9997 or 0.9994, respectively, between the samples, considering bottles/glasses of different brand beers and non‐compatible materials. Considering a dataset where the glasses were grouped by glass color and not by beer brand, the approach reached a mean accuracy of 1 for a 10‐fold or 4‐fold cross‐validation. The remarkable differences in the chemical compositions of the borosilicate glass, ceramic materials, and the presence of specific elements explain the accuracy.