Removal of Pesticides from Aqueous Solution: Quantitative Relationship Between Activated Carbon Characteristics and Adsorption Properties

Abstract

The adsorption of pesticides (atrazin, atrazin-desethyl and triflusulfuron-methyl) from aqueous solution is performed by activated carbon fibers (ACF) and granular activated carbons (GAC) in static and dynamic reactors, in order to study the co-influence of adsorbent and adsorbate characteristics on the adsorption mechanisms. First, mono-component adsorption equilibrium is carried out in a batch reactor for a wide range of concentrations (from 5 μg l−1 to 21.4 mg l−1). Classic models, like Freundlich and Langmuir equations, are applied: the maximum adsorption capacities are high, ranging between 63 and 509 mg g−1. The comparison of single-solute isotherms tends to confirm the decisive role of the adsorbent properties in the adsorption capacity of pesticides by the activated carbons: the performance of ACF is significantly higher than that of GAC due to a narrower pore size distribution of fibers in the area of micropores. Furthermore, their small diameter (10 μm compared with 1 mm for grains) enables faster adsorption kinetics because of the larger surface area exposed to the fluid. The influence of adsorbate size is also demonstrated. A multiple linear regression enables the co-influence of adsorbent and adsorbate properties to be quantified, a relationship being assessed between Langmuir maximum adsorption capacity and pesticide molecular weight and adsorbent diameter (R2 = 0.90). Secondly, the adsorption of the three pesticides is studied in a dynamic reactor: in this case, the influence of operating conditions (inlet concentration C0 , flow velocity U0 ) is also taken into account. As the initial concentration or flow velocity decreases, the column performance significantly improves. Both operating factors are included in a multiple linear regression (R2 = 0.91) used to predict saturation adsorption capacity, with molecular weight and particle diameter being again designed as influent explicative variables.