The widespread use of plastic materials, owing to their several advantageous properties, has resulted in a considerable increase in plastic consumption. Consequently, the production of primary and secondary microplastics has also increased. To identify, categorize, and quantify microplastics, several analytical methods, such as thermal analysis and spectroscopic methods, have been developed. They generally offer little insight into the size and shape of microplastics, require time-consuming sample preparation and classification, and are susceptible to background interference. Herein, we created a macroscale hyperspectral Raman method to quickly quantify and characterize large volumes of plastics. Using this approach, we successfully obtained Raman spectra of five different types of microplastics scattered over an area of 12.4 mm × 12.4 mm within just 550 s and perfectly classified these microplastics using a machine learning method. Additionally, we demonstrated that our system is effective for obtaining Raman spectra, even when the microplastics are suspended in aquatic environments or bound to metal-mesh nets. These results highlight the considerable potential of our proposed method for real-world applications.
Keywords: Hyperspectral imaging; Line scanning; Machine learning; Microplastics; Raman spectroscopy.
Copyright © 2023 Elsevier B.V. All rights reserved.