Key points are not available for this paper at this time.
This study explores the application of electro-membrane filtration (EMF) for effective removal of nanoplastics (NPs). It assesses the impact of feed variations and the underlying mechanisms related to feed characteristics, to improve efficiency and selectivity in Ti3C2Tx EMF. Additionally, 4 machine learning (ML) were applied to investigate the flux and rejection prediction capability on unseen data, namely, polynomial linear regression (LR), decision trees, random forest, and gradient boosting classifier. The study systematically investigates pH ranging from 5 to 9, NP sizes 25 to 100 nm, and voltage ON/OFF modes comprehensively. Experiments are replicated for polystyrene (PS) and polymethyl methacrylate (PMMA) NPs, addressing their distinct characteristics. As a result of applied voltage, the flux increases by up to 36.9 % for PS, and 29.5 % for PMMA. Additionally, a rejection of up to 95.4 % and 97 % for PS and PMMA NP, respectively was reported. The results show that pH influences flux differently for PS and PMMA, with basic conditions enhancing flux for PS and acidic conditions favored for PMMA. Rejection is significantly affected by NP size and pH, with larger NPs and lower pH leading to higher rejection due to size exclusion and electrostatic interactions. Finally, the ML models performed well with a maximum R2 of 0.95, 0.96, 0.96, and 0.98 for predicting PS flux, PS rejection, PMMA flux, and PMMA rejection, respectively. The study underscores the complex interplay of factors influencing filtration performance and provides valuable insights for designing efficient Ti3C2Tx EMF systems.
Ouda et al. (Wed,) studied this question.