Microplastics detection in agricultural soil combining 3D Laser Scanning Confocal Microscopy with machine learning

Abstract Low-density plastics of different origins are a major source of microplastic (MP) contamination in agricultural soil systems. Although several plastic entry pathways are well known, such as the fragmentation of plastic materials used in so-called plasticulture or the contamination of organic fertilisers, including compost and sewage sludge, quantifying the MP contamination of these soil systems remains challenging and time-consuming. This study developed and rigorously tested a hazard-free workflow to overcome these limitations and expand the capabilities for detecting MP. The workflow combines 3D Laser Scanning Confocal Microscopy (Keyence VK-X1000, Japan) with machine-learning-based data analysis and was evaluated using three agricultural topsoils spiked with transparent and black low-density polyethylene and polypropylene particles (<53 µm, 53-100 µm, 100-250 µm) and polypropylene fibres (1000 µm). The method reliably detects both transparent and black MP ≥53 µm in soils with low particulate organic matter content, achieving a mean recovery rate of 80% ± 28%. Transparent MPs were reliably identified, whereas black MPs and fibres were influenced by particulate organic matter. Beyond particle count and size, the approach quantifies surface morphology using high-resolution 3D data. Four 25 g samples (100 g total soil) can be processed within three days, providing a fast, accurate, and environmentally safe tool for MP analysis in agricultural soils.