Detection of microplastics in soil using near infrared spectroscopy and chemometrics

Microplastic pollution in soil have a vital impact on organisms living in soil. To quantify amount of microplastics in soil, the preparation approach need to be performed in order to separate plastic from soil. Therefore, the conventional methods such as chromatography are complicated and time-consuming method to identify and quantify microplastics in soils. In the study, the powerful of near infrared spectroscopy (NIRS) ranging 1100 – 2300 nm combined with chemometrics tools are demonstrated as a technique for rapidly monitoring microplastics in soil as the preparation step is unrequired. Polyethylene terephthalate (PET) and polystyrene (PS) were grinded by blender to generate the artificially microplastics. Size of grinded microplastics of PET and PS are in the range of 44.45 ± 66.68 and 103.39 ± 101.13 μm, respectively. Sets of artificially polluted soil samples were prepared by mixing the microplastics in soil with various concentrations (1-10 %w/w). The reflectance mode of NIR were used to acquire the NIR spectra of soil contaminated with microplastic. The models for microplastic prediction were generated using machine learning algorithms, partial least squares regression (PLSR), with data pre-processing in MATLAB®. The performance of models was evaluated showing that RMSEP for PET was 1.30 %w/w with R² = 0.8030 and RMSEP for PS was 1.19 %w/w with R² = 0.8253. After using variable selection method (shaving method), the RMSEP for PET was 1.35% w/w with R² = 0.7854 (47 variables) and RMSEP for PS was 1.09 %w/w with R² = 0.8582 (56 variables). The mixture of PET and PS microplastics in soil samples were quantified using external parameter orthogonalization (EPO) to extract only the reflectance spectra of target analyzes. The results presented RMSEP was 3.17 %w/w and 3.26 %w/w for PET and PS, respectively. It could be concluded that NIRS combined with chemometrics is potential technique for microplastics detection in soil.