Microplastics emissions from occupant activities in residential spaces: Experimental quantification and relative importance analysis of particle generation factors

Occupant activities generate indoor microplastics (MPs), yet their activity-specific proportions have been insufficiently quantified. Herein, indoor MPs emissions associated with walking friction, clothing fiber detachment, and chair leg cover friction were investigated in two residences using one-month field sampling, activity simulations, and supplementary machine-learning analysis of friction factors. Airborne particles were collected using air purifier filters, while generated MPs were isolated and quantified using an imaging algorithm. For walking-induced emissions, 88 datasets were constructed for each foot-contact condition over normal load and friction cycle levels. Extreme gradient boosting (XGBoost) models predicted MPs concentrations for ethylene vinyl acetate (EVA) slipper and barefoot conditions, achieving R² values of 0.974 and 0.982, respectively. SHapley additive explanations (SHAP) analysis showed that normal load and friction cycles contributed 46.78% and 53.22%, respectively, under EVA slipper contact, and 42.31% and 57.69%, respectively, under barefoot contact, thereby confirming the greater effect of repeated friction. Walking-induced concentrations increased from 27 to 96 items/m³ at 5000 cycles to 66–154 items/m³ at 10,000 cycles, with the EVA slippers producing higher emissions. Monthly airborne MPs concentrations were 19.90 and 23.15 items/m³ in residential units in mixed-use and townhouse buildings, respectively. Clothing fiber detachment was the largest quantified source, accounting for 14.57–15.16%, whereas walking and chair-leg friction contributed 1.64–3.87% and 0.78–1.66%, respectively. Study established an activity-specific experimental framework for occupant behavioral factors, achieved machine learning-based interpretation, and supported exposure reduction strategies.