One-year dynamics of atmospheric organophosphate esters: Size-fractionated distribution, seasonal trends, and machine learning-based driver identification

Organophosphate esters (OPEs) have increasingly raised global concern due to their pervasive presence in airborne particles and potential risks they pose to human health. This study systematically characterized OPE concentrations in PM 2.5 , PM 10 , and TSP over a one-year period. The results revealed that OPEs preferentially partitioned to smaller particles and displayed higher total concentrations during the warm season. Among alkyl-, halogenated-, and aryl-OPEs, halogenated-OPEs exhibited the highest concentrations while aryl-OPEs ranked highest in ToxPi-based prioritization. To identify key environmental drivers, Random Forest model combined with SHAP analysis and structural equation model were employed. The RF-SHAP model demonstrated higher explanatory power for halogenated-OPEs compared to alkyl- and aryl-OPEs, with temperature, atmospheric pressure, and the interaction effect of temperature and relative humidity identified as dominant influencing factors. Furthermore, the findings suggest that OPEs may undergo atmospheric transformation under complex environmental conditions. Overall, this study provides a comprehensive understanding of the particle-size distribution, temporal dynamics, environmental determinants, and health risk profiles associated with particle-bound OPEs. These insights are invaluable for informing future regulations concerning emerging contaminants in the atmosphere.