Detection limits of soil microplastics using mid-infrared spectroscopy

Abstract Microplastic (MP) pollution in agricultural soils presents emerging risks to soil health, crop productivity, and food safety, yet detecting MPs at low concentrations remains difficult. This study establishes an innovative mid-infrared (MIR) spectroscopy framework combining spectral analysis and wavelength selection to detect MPs in soils. Two common MPs, low-density polyethylene (LDPE) and polyethylene terephthalate (PET), were spiked into sand, loam, and clay soils at concentrations ranging from 0.01 to 0.6%, after which MIR spectra were acquired. The spectra were analysed by comparing selected plastic-indicative wavelengths (PIWs) with soil-indicative wavelengths (SIWs). Soil spectral detection limits (SSDLs) were subsequently derived using SIW: PIW band ratios and MP concentration relationships combined with a Partial Least Squares Regression (PLSR) - Cubist modelling approach. Results showed that soil matrix effects are more influential than polymer identity in controlling SSDL. Sand and loam soils produced the best performance, with low SSDLs and comparatively high R² values for both polymers. Clay, however, caused strong spectral interference, reducing predictive accuracy, although several PET and PE absorption features remained relatively detectable at low concentrations. These trends demonstrate the dominant role of soil texture and mineralogy in shaping the sensitivity and reliability of MIR detection. Overall, integrating soil-specific calibration, targeted wavelength selection, spectral processing, and a hybrid PLSR-Cubist framework enables detection of MPs at low concentrations, with performance dependent on soil mineralogy and texture. These findings highlight MIR spectroscopy, particularly with texture-stratified modelling, as a promising method for low-level MP detection in agricultural soils.