Mid-Infrared Classification for post-consumer plastics combining Infrared spectroscopy and open-source data mining.

Plastics are essential in everyday life due to their unique chemical and physical properties. However, plastic pollution has become one of the most pressing environmental challenges, driven by the rapid increase in plastic production and consumption, as well as the accumulation of non-degradable post-consumer plastic waste [1-4]. While recycling technologies offer a promising solution for managing this waste, the presence of organic contaminants in recycled resins often hinders their reuse and limits their compliance with quality standards [3,4]. One of the primary obstacles to the widespread adoption of recycled materials in fostering a circular and sustainable economy is the presence of unpleasant odors that frequently characterize the granules at the end of the recycling process. These odors originate from volatile substances that are not entirely removed during recycling. The implementation of innovative selective and/or specific deodorization technologies presents an effective strategy for eliminating such contaminants from the granules. This approach is closely linked to the advancement of plastic classification techniques that can also identify the presence of specific volatile organic compounds (VOCs). The mid-infrared (MIR) region has recently attracted attention due to its inherent advantages over the VIS and Near-IR regions, which are commonly used for optical screening of plastics. The fundamental vibrational modes in this spectral region make MIR frequencies particularly well-suited for high-fidelity machine learning (ML) classification, offering potential solutions for high-throughput classification due to their scalability, accuracy, and capability. Pioneering efforts in high-throughput polymer characterization using MIR spectra are underway, as demonstrated by the works [5-7]. To date, published MIR datasets for preliminary studies do not reflect the molecular heterogeneity of recycled plastics, let alone their complex composition in relation to contaminants. For example, the extensive literature on microplastic databases consists of samples that have been chemically and physically altered by environmental factors such as oxidation and UV irradiation, over which there is no control. Here, we studied various recycled plastics from mechanical extrusion through Attenuated Total Reflectance Infrared (ATR-IR) spectroscopy. An MIR database was generated, curated, and evaluated for real-world post-consumer plastics. Starting from MIR database, we designed and trained ML automatic and rapid recognition method for four plastic classes of interest: high-density polyethylene type B (HDPE-B), high-density polyethylene type P (HDPE-P), low-density polyethylene (LDPE), and polypropylene (PP). We established quantifiable metrics for evaluating the performance. Finally, taking advantage of our extensive research activity in the field of VOCs ultra-sensitive monitoring [8-12], we performed a preliminary identifier and predictive recognition of VOCs quantities present of the surface of post-consumer plastic granules responsible for the odors. Although this represents a preliminary result, it holds significant industrial relevance, particularly within the context of the circular economy. Specifically, the ability to identify the molecules responsible for the undesirable odors, which concurrently impede the performance of recycled plastics, offers the potential to implement targeted strategies aimed at the removal of these specific molecules. References 1. A. Cabanes, F.J. Valdés, A. Fullana. SM&T, 2020, 25, e00179. 2. R. Geyer, J. R. Jambeck and K. L. Law. Sci. Adv., 2017, 3, e1700782. 3. M. MacLeod, H. P. H. Arp, M. B. Tekman and A. Jahnke. Science, 2021, 373, 61–65. 4. J. Demarteau, A. R. Epstein, P. R. Christensen, M. Abubekerov, H. Wang, S. J. Teat, T. J. Seguin, C. W. Chan, C. D. Scown, T. P. Russell, J. D. Keasling, K. A. Persson and B. A. Helms. Sci. Adv., 2022, 8, eabp8823. 5. S. Jiang, Z. Xu, M. Kamran, S. Zinchik, S. Paheding, A. G. McDonald, E. Bar-Ziv and V. M. Zavala. Comput. Chem. Eng., 2021, 155, 107547. 6. F. Long, S. Jiang, A. G. Adekunle, V. M Zavala and E. Bar-Ziv. ACS Sustainable Chem. Eng., 2022, 10, 16064–16069. 7. N. Stavinski, V Maheshkar, S. Thomas, K. Dantu, L. Velarde. Environ. Sci.: Adv., 2023, 2, 1099. 8. V. Galstyan, A. D’Arco, M. Di Fabrizio, N. Poli, S. Lupi, E. Comini. Reviews in Analytical Chemistry 2021, 40 (1), 33-57. 9. A. D’Arco, D. Rocco, F. P. Magboo, C. Moffa, G. Della Ventura, A. Marcelli, L. Palumbo, L. Mattiello, S. Lupi, M. Petrarca. Optics Express 2022, 30(11): 19005-19016. 10. A. D’Arco, T. Mancini, M.C. Paolozzi, S. Macis, L. Mosesso, A. Marcelli, M. Petrarca, F. Radica, G. Tranfo, S. Lupi, G. Della Ventura. Sensors 2022, 22(15), 5624. 11. F. Radica, G. Della Ventura, L. Malfatti, M. Cestelli Guidi, A. D’Arco, A. Grilli, A. Marcelli, P. Innocenzi. Talanta 2021, 233:122510 12. T. Mancini, F. Radica, L. Mosesso, M.C. Paolozzi, S. Macis, A. Marcelli, S. Tamascelli, G. Tranfo, G. Della Ventura, S. Lupi, A. D’Arco. JECE, 2025, online.