Understanding and predicting the adsorption and rejection of pesticides and metabolites by hollow fiber nanofiltration membranes

Abstract

Layer-by-layer Hollow fiber Nanofiltration (LbL HF-NF) membranes have shown great potential in the removal of micropollutants from water. Nonetheless, further research is required as the number of tested compounds remains limited. In this study, the adsorption and rejection of 164 micropollutants by two commercial HF-NF membranes with MWCO of 400 and 800 Da, were investigated. The investigation was conducted by evaluating the influence of 11 physicochemical micropollutants’ properties on the separation performances. Results showed that highly adsorbed compounds had average log D values above 3, while no or low adsorption was observed for compounds with negative log D values. Furthermore, adsorption was more frequently observed for neutral compounds compared to negatively charged ones. Micropollutants rejection results showed that steric exclusion plays the most important role in the rejection of micropollutants, and that the charge of micropollutants can also heavily influence their rejection. Likewise, negatively charged compounds were better rejected than neutral ones. Finally, the study demonstrated that adsorption and rejection of micropollutants can be predicted with good accuracy using the random forest algorithm. The prediction accuracy for adsorption was 80% and 73% for the 800 and 400 Da MWCO membranes, respectively. The RMSE for rejection predictions was 10.6% and 6% for the two membranes respectively.