Abstract:
Emerging contaminants are released into the water bodies from various industrial and
agricultural, trade and commerce activities and from use of personal care products in our
daily lives. The discharge of these chemicals in the natural water bodies being in low
concentration ranging from (ng/L to µg/L) make them escape from the conventional
wastewater treatment plants (CWWTP). Therefore, advanced treatment strategies such as the
use of adsorbents are employed to remove the contaminants of emerging concern from
municipal and industrial water systems. In this research, partitioning models were developed
to understand what properties of chemicals govern the adsorption behavior and to predict the
adsorption potential of widely used adsorbents. Conventional and emerging adsorbents were
evaluated individually using diverse set of chemicals with using the multivariant linear
regression (MLR) algorithm and cross-validation tests. The model was developed for 165
neutral organic compounds in which adsorption capacity was estimated on four different
types of adsorbents including carbon nanotubes (CNTs), granular activated carbon (GAC),
biochar, and resins. A relationship was developed between various parameters like the
surface area of the adsorbent (BET), total pore volume (V t ), the equilibrium concentration of
adsorbate (Ce), octanol-water coefficient (K ow ), and air-water coefficient (K aw ). The four-
partition models of CNTs, GAC, biochar, and resins were stable, robust, and accurate with
statistical indices such as (R 2 =0.79 ̵ 0.69) and Root Mean Square Error (RMSE) ranging from
(0.323 ̵ 0.5064). The correlation coefficient for the adsorbent resin is small among all other
adsorbents because of considering total pore volume (V t ) instead of the macropore volume
(V m ) and micropore volume (V mi ) to maintain uniformity in parameters for all models. The
sensitivity analysis inferred that the interaction of the parameters with each other greatly
affects the adsorption phenomena, and each model has its own influential parameter for that
specific adsorbent. The four models do not only offer a theoretical method for predicting
adsorption capacity of adsorbents but also have the prediction power of adsorption behavior
of all organic pollutants in an aqueous environment reducing the need for further
experiments. These models for sorbents can be further improved by adding more authentic
experimental data and can be utilized in the future for estimating the adsorption behavior of
ionized emerging contaminants in real water.
