Toward in Situ Identification of Microplastics in Water Using Raman Spectroscopy and Machine Learning

Plastic pollution is a significant environmental issue, with microplastics (particles <5 mm) being particularly challenging to detect and quantify due to their small size. This study shows preliminary results on developing an in situ system for the real-time detection, characterization, and identification of microplas-tics in oceans. The proposed system will integrate multiple techniques, including Raman spectroscopy, laser-induced fluorescence, photo-acoustic imaging, acoustic imaging, and optical imaging, to achieve rapid and accurate measurements. This paper focuses on the potential of Raman spectroscopy for in situ microplastic characterization and on leveraging machine learning models for plastics classification. We collected Raman spectra from various plastic samples, including pristine and weathered microplastics. We trained a Support Vector Machine (SVM) classifier on open-access Ra-man libraries, comparing models trained on the full Raman spectrum versus the fingerprint region. The full-spectrum model achieved higher accuracy, correctly identifying all pristine samples and 13 out of 18 environmental samples. Our findings demonstrate the feasibility of using Raman spectroscopy combined with machine learning for a potential real-time microplastic detection system.