Identification and classification of microplastics by a micro-hyperspectral imaging scanner prototype

Plastic pollution has emerged as a critical global issue, exerting significant ecological impacts on both aquatic and terrestrial ecosystems. However, the detection of microplastics (MPs), i.e., particles smaller than 5 mm, remains a complex task due to their microscopic size and diverse chemical compositions. Hyperspectral imaging (HSI) has recently gained attention as a powerful technique to address this challenge, offering precise differentiation of MPs by leveraging their distinct spectral signatures. Moreover, HSI integrates imaging with spectral analysis, enabling rapid identification and significantly reducing processing times compared to conventional spectroscopic methods. In this study, a novel approach for the rapid detection and classification of MPs using a custom-built near-infrared (NIR: 1000–1650 nm) micro-spectral scanner is proposed. The innovative micro-HSI spectral system, specifically designed for MPs recognition, combines hyperspectral technology with microscopy to achieve a spatial resolution of 10 μm/pixel. The scanner’s sample holder is mounted on a motorized 2-axes system that moves in the X and Y directions, enabling the scanning of areas up to 20 x 26 mm. The scanning process involves parallel acquisitions with a 3.37 mm field of view (FOV) of the IR camera, which are automatically mosaicked by the acquisition software. To evaluate the performance of the micro-HSI device, reference samples were prepared by grinding three common polymer types typically found in marine and terrestrial environments: polystyrene (PS), polypropylene (PP), and high-density polyethylene (HDPE). The obtained MPs, ranging in size from 1 mm to 30 μm, were collected on nine adhesive carbon tabs (three tabs per polymer type) with a diameter of 1 cm. Hyperspectral images of the samples were acquired and pre-processed to enhance spectral features and reduce noise caused by light scattering and MPs morphology and then evaluated using Principal Component Analysis (PCA), with particular attention to spectral variations as particle size decreased. Based on the features identified by PCA, a hierarchical classification approach was developed to automatically and rapidly identify MPs, achieving accurate classification for particles greater than 100–150 μm, confirming the device’s ability to detect a wide range of MP sizes with high precision. The results demonstrated that the innovative micro-HSI system, when combined with a dedicated classification models, enabled significantly faster analysis of MPs compared to conventional techniques such as micro-Fourier Transform Infrared Spectroscopy (micro-FTIR). Additionally, the system’s configuration supports automated scanning and mosaicking, further reducing analysis time. This advancement can be a starter point to improve the micro-HSI system for efficient, high-resolution MPs analysis, paving the way for broader applications in environmental monitoring and plastic pollution research.