Advanced approaches for modeling trace metal sorption in aqueous systems

An extensive collection of new macroscopic and spectroscopic data was used to assess the ability of the modified triple-layer model (TLM) to predict single-solute lead (Pb) and zinc (Zn) sorption onto 2-line ferrihydrite over a wide range of conditions in both single- and multi-equilibrium-stage systems. Regression of constant-pH isotherm data, together with potentiometric titration and pH edge data, was found to be a much more rigorous test of a surface complexation model than fitting pH edge data alone. When combined with spectroscopic data, the choices of feasible surface species/site types were limited to a few. Excellent fits of the Pb sorption data were obtained with a two-species, one-site model using the species pairs (≡FeO)₂Pb/≡FeOHPb 2+ and (≡FeO)₂Pb/≡FeOPb+-NO ₃−. For Zn, a one-species, one-Zn-sorption-site model using the bidentate complex, (≡FeO)₂Zn, provided excellent fits of the data. Surprisingly, best-fit equilibrium “constants” for the Pb and Zn surface complexes and, for Zn, the density of Zn sorption sites had to be varied with pH to fit the data. A surface activity coefficient term was also required for both metals to reduce the ionic-strength dependence of sorption.; Experimental data and modeling results showed that a multistage sorption system can significantly reduce Pb and Zn effluent concentrations for the same total amount of sorbent or, alternatively, dramatically lower sorbent consumption for the same effluent concentration. Model predictions were made using the modified TLM integrated into a steady-state, multistage, equilibrium adsorber model within the OLI Software (OLI Systems, Inc., Morris Plains, NJ). Engineering screening evaluations showed that a 2- to 3-stage sorption process can provide significant economic savings when compared to a 1-stage process operating with the same target effluent metal concentration. Additional equilibrium stages beyond 2 or 3 provide diminishing economic returns.; Finally, using a novel uncertainty analysis module within the OLI Software, error propagation through the modified TLM was studied for both metals using isotherm, pH edge, and multistage treatment data. When coupled with the nonlinear equation solver's gain matrix, the linearized local extrapolation and error propagation models in the OLI code provided a satisfactory alternative to a more rigorous, time-consuming Monte Carlo simulation.