Detection of chlorpyrifos in rice fields soil based on terahertz spectroscopic techniques

Chlorpyrifos, an organophosphorus insecticide widely utilized in rice cultivation, poses environmental and health risks due to the accumulation of pesticide residues in soil. Traditional detection methods for chlorpyrifos residues, such as high-performance liquid chromatography and gas chromatography-mass spectrometry (GC-MS), are often limited by complex pretreatment processes and high equipment costs. In this study, a microalgae-based sensing strategy combining microalgal adsorption with terahertz spectroscopy was developed. Changes in the spectral characteristics of microalgal metabolites (β-carotene, lipids, and starch) in the terahertz range were analyzed. Principal component analysis was used solely to assess spectral clustering at different exposure times. Three machine learning models, namely Random Forest, Support Vector Machine, and one-dimensional convolutional neural networks (1D-CNN), were developed and comparatively evaluated for classification of chlorpyrifos concentration levels, with 1D-CNN achieving the best performance. Experimental results showed that this method required only 10 mL of sample and reduced the detection time to within 12 min. Additionally, soil samples from five distinct regions were analyzed, showing a classification accuracy of 89% relative to GC-MS reference results.