Automatic Identification of Individual Nanoplastics by Raman Spectroscopy Based on Machine Learning

This study combined Raman spectroscopyand machine learningto accurately identify nanoplastics, with successful application tocomplex environmental samples. The increasing prevalence of nanoplasticsin the environmentunderscoresthe need for effective detection and monitoring techniques. Currentmethods mainly focus on microplastics, while accurate identificationof nanoplastics is challenging due to their small size and complexcomposition. In this work, we combined highly reflective substratesand machine learning to accurately identify nanoplastics using Ramanspectroscopy. Our approach established Raman spectroscopy data setsof nanoplastics, incorporated peak extraction and retention data processing,and constructed a random forest model that achieved an average accuracyof 98.8% in identifying nanoplastics. We validated our method withtap water spiked samples, achieving over 97% identification accuracy,and demonstrated the applicability of our algorithm to real-worldenvironmental samples through experiments on rainwater, detectingnanoscale polystyrene (PS) and polyvinyl chloride (PVC). Despite thechallenges of processing low-quality nanoplastic Raman spectra andcomplex environmental samples, our study demonstrated the potentialof using random forests to identify and distinguish nanoplastics fromother environmental particles. Our results suggest that the combinationof Raman spectroscopy and machine learning holds promise for developingeffective nanoplastic particle detection and monitoring strategies.