Coupling hyperspectral imaging with machine learning algorithms for detecting polyethylene (PE) and polyamide (PA) in soils

Authors: CHEN, HUAN - Clemson University; Shin, Tae-Sung; Park, Bosoon; Ro, Kyoung; JEONG, CHANGYOON - Louisiana State University; JEON, HWANG-JU - Louisiana State University; TAN, PEI-LING - Clemson University

Submitted to: Journal of Hazardous Materials
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 4/17/2024
Publication Date: 4/18/2024
Citation: Huan, C., Shin, T., Park, B., Ro, K.S., Jeong, C., Jeon, H., Tan, P. 2024. Coupling hyperspectral imaging with machine learning algorithms for detecting microplastics in soils. Journal of Hazardous Materials. https://doi.org/10.1016/j.jhazmat.2024.134346.
DOI: https://doi.org/10.1016/j.jhazmat.2024.134346

Interpretive Summary: Microplastics (MPs) have emerged as persistent pollutants and attracted global attention because they are small enough to be taken by biota and can enter the food chain. The MPs in agricultural soils mainly comes from the application of soil amendments (e.g., biosolids and compost) and plastic mulching films, treated-wastewater irrigation, and atmospheric deposition. MPs have a large surface area that can act as an efficient adsorbent/carrier for the micropollutants such as pesticides, antibiotics, and heavy metal(loid)s in agricultural soils, resulting in a significant accumulation of these micropollutants. Despite the critical need for studying the presence and the fate of MPs in our soils, it has not been adequately understood due to the difficulty in detecting and quantifying MPs in soils. Current methods to detect MPs are labor intensive and time–consuming; therefore, it is critical to develop a rapid and effective method to detect MPs directly without complicated procedures. Hyperspectral imaging (HSI) technology has been used for rapidly detecting and identifying contaminants in poultry: It has the potential to directly detect and distinguish different types of MPs in soils. In this study, we identified essential wavelengths for optimal MP detection and demonstrated that the HSI method was able to detect and quantify different MPs by characterizing the spectral profiles of soil–MP mixtures.

Technical Abstract: Microplastics (MPs) as the carrier of micropollutants could transport from soil to crops and affect food safety adversely. It is critical to develop a rapid and efficient method for their identification in soils without pretreatments. Here, we coupled hyperspectral imaging (HSI) systems [i.e., visible near infrared (VNIR: 400–1000 nm), indium gallium arsenide (InGaAs: 800–1600 nm), and mercury cadmium telluride (MCT: 1000–2500 nm)] with machine learning algorithms to distinguish soils spiked with synthetic MPs (commercial PE and PA with an average size of 50 and 300 µm, respectively). The soil-normalized shortwave infrared (SWIR) spectra unveiled large spectral differences not only between control soil without MPs and pure MPs [i.e., polyethylene (PE) 100% and polyamide (PA) 100%], but also among 5 soil–MP mixtures (i.e., PE 1.6%, PE 6.9%, PA 5.0%, and PA 11.3%). This was primarily attributable to the 1st–3rd overtones and combination bands of C-H groups in MPs. Feature reduction techniques visually demonstrated the separability of the 7 sample types by SWIR and the inseparability of 5 soil–MP mixtures by VNIR. Moreover, classification models showed the accuracies of 92–100% by InGaAs, 97–100% by MCT, and 44–87% by VNIR. Our study indicated the feasibility of SWIR HSI systems in detecting PE (low as 1.6%) and PA (low as 5.0%) in soils.