Adsorption of Arsenic in Smelter Flue Dust

However, as we noted early in this chapter, numerous assumptions are employed in the field applications of surface complexation models. Davis et al. (1998) noted that surface complexation models are mainly developed from well-controlled laboratory experiments. It is unclear how the models can be applied to soil and sediments where the double layers of the heterogenous particles may interact and the competitive adsorption of many different ions can cause significant changes in the electrical properties of mineral-water interfaces. 

Doyle et al. (1994) used the triple layer model of Davis et al. (1978) to model surface adsorption of arsenic onto amorphous ferric oxides. Copper smelting has produced 

The authors first used minteqa2 to evaluate the solubility control of As concentrations in the effluents of column experiments. They found that it was necessary to replace the log K values for the arsenic aqueous species and the mineral scorodite [FeAsCU 2H2O] in the minteqa2 database with those for the Fe-As-SC 4 systems from Robins (1990) in order to match the simulation and experimental results. From the microprobe study they found 1 p.m-sized, As-bearing phases having a stoichiometric composition close to scorodite. 

The authors found that the model-predicted As concentration is close to the leachate concentrations from the column packed with dust from the continuous reactor (1 120 H.gL-1 versus 1 330 xgL-1), when the solubility product of scorodite from Robins (1990) and the triple layer model is used. Only 11% As in the system was sorbed onto hydrous ferric oxide surfaces. Arsenic concentrations in the leachate are largely controlled by scorodite solubility. It should also be pointed out that simulations using solubility only and without including surface adsorption resulted in a closer match (1 270 n-gL-1 versus 1330 piglA1). For the simulation of the column experiments using wastes from the batch-reactor, the triple layer model predicted too low an As concentration (33 jigL-1 versus 120 pigL-1). 

---