Modeling engineered nanoparticles removal by conventional activated sludge treatment process in wastewater treatment plant

The production and applications of engineered nanomaterials (ENM) has increased rapidly in the last decade, with release of ENM to the environment through the sewer system and municipal wastewater treatment plants (WWTPs) being of concern. Currently, the literature on ENM release from WWTPs and removal of ENM by WWTPs is insufficient and disorganized. There is little quantitative data on the removal of multi-walled carbon nanotubes (MWCNTs), graphene oxide (GO), or few-layer graphene (FLG), from wastewater onto biomass. The removal of pristine and oxidized MWCNTs (O-MWCNTs), graphene oxide (GO), few-layer graphene (FLG) and Tween™ 20-coated Ag ENM by the interaction with biomass were determined by programmable thermal analysis (PTA) and UV-Vis spectrophotometry. The removal of pristine and O-MWCNTs was 96% from the water phase via aggregation and 30-min settling in presence or absence of biomass with an initial MWCNT concentration of 25 mg/L. The removal of 25 mg/L GO was 65% with biomass concentration at or above 1,000 mg TSS/L. The removal of 1 mg/L FLG was 16% with 50 mg TSS/L. The removal of Tween™ 20 Ag ENM with concentration from 0.97 mg/L to 2.6 mg/L was from 11% to 92% with biomass concentration of 500 mg TSS/L to 3,000 mg TSS/L, respectively.

A database of ENM removal by biomass was established by analyzing data from published papers, and non-linear solid-liquid distribution functions were built into the database. A conventional activated sludge (CAS) model was built based on a membrane bioreactor (MBR) model from a previous paper. An iterative numeric approach was adapted to the CAS model to calculate the result of non-linear adsorption of ENM by biomass in the CAS process. Kinetic studies of the CAS model showed the model performance changed mostly in the first 10 days after changing influent chemical oxygen demand (COD) concentration, and reached a steady state after 11 days. Over 60% of ENMs which have distribution coefficients in the database reached higher than 50% removal by the CAS model under general operational conditions. This result suggests that traditional WWTP which include the CAS process can remove many known types of ENMs in certain degree.