Fitting the Site Densities

The measured pH increased steadily during the adsorption step and decreased with the start of the desorption step. The decrease was sharp in the case of the desorption of arsenate and chromate. In the case of phosphate, the decrease occmred with fluctuations. 

Column experiments were conducted to simulate the transport of oxoanion aquifers containing iron hydroxides. The Riedel de Haen Hfo material used in these column experiments exhibits a low specific surface area and contains a significant proportion of goethite. PHREEQC2 is equipped to model surface complexation of oxoanions onto iron hydroxides but a consistent data set of surface complexation constants is only available for amorphous hydrous ferric oxides. Tests were conducted to determine whether it is possible to use PHREEQC2 with this data set to model the oxoanion transport in the columns. If the data set of surface complexation constants is also suitable for the Riedel de Haen Hfo material than it should be sufficient to adjust the site density of the iron hydroxide surface in the model. 

At first the data of the column experiments were compared with model predictions achieved by using the site density of amorphous iron hydroxide (0.2 mol sites/mol Fe( ,oii(i)) and by using the site density of goethite (0.016 mol sites/mol F6(soiid))- The PHREEQC2 model runs with a site density typical for amorphous iron hydroxide overestimated the retention of the three tested oxoanions by a factor of 2.1 to 4.7 and the model runs with the site density of goethite underestimated the retention by a factor of 2.2 to 4.6. The relative deviation between the observed breakthrough of the oxoanions and the model predictions is quite similar for all three oxoanions. Best approximations of the model predictions were achieved with site densities (mol sites/mol Fe( ,oii(j)) of 0.054 (chromate), 0.044 (phosphate) and 0.074 (arsenate). 

This result appears quite consistent. It indicates that the complexation constants derived for amorphous iron hydroxide may also be suitable for the Riedel de Haen Hfo material used and that only the surface site density has to be adjusted. Nevertheless the difference between the best-fit value for the number of sites differs between phosphate (minimum) and arsenate (maximum) is about 70 %. An entirely consistent result would require the same result of the fitting for all three oxoanions. 

One characteristic of the surface complexation modelling with PHREEQC2 is that the pH of the column outflow is also predicted while considering several reactions. As a result it must be stated that this modelling of the pH seems to be a problem. In all three column experiments a steady increase in pH was measured during the adsorption step and a steady decrease during the desorption step. The general effect of proton con- 

---