Mobility of biochar colloids from diverse feedstocks: Dominant control of zeta potential under varying ionic strength and Cd2+ presence.

The transport of biochar colloids (BCs) is a cross-media (water-soil-air) and cross-scale environmental process related to pollution diffusion, carbon cycling, and climate change. It requires multidisciplinary collaborative research to quantify its risks and formulate control strategies. In this study, column experiments compared the transport of BCs made from five feedstocks-Miscanthus Straw Pellets (MSP), Oil Seed Rape Pellets (OSR), Wheat Straw Pellets (WSP), Soft Wood Pellets (SWP), and Sewage Sludge Pellets (SS)-with/without cadmium (Cd2+). Published data were also used to create machine learning models predicting BC transport. At 1 mM ionic strength (IS), BC mobility decreased in the order: MSP (70.0 %) > OSR (65.2 %) > SWP (61.8 %) > WSP (58.5 %) > SS (51.2 %), correlating with zeta (ζ) potential (R2 = 0.98, P < 0.001). Mobility dropped to between 22.0 % and 32.0 % when IS was increased to 5 mM and decreasing further with Cd2+ addition (4.0 %-10.2 %). Random forest and linear regression analyses confirmed that the ζ potential, rather than the other feedstock-influenced properties, was the primary factor governing BCs transport under all tested conditions. The developed machine learning model, based ζ potential, demonstrated strong accuracy (R2 = 0.880, RMSE = 0.378) in predicting BCs mobility. These findings provide a valuable framework for predicting BCs mobility, enabling more effective biochar application in pollution remediation while minimizing secondary contamination risks.