Comparison of measured and modelled copper binding by natural organic matter in freshwaters

Fifteen freshwater samples containing significant concentrations of dissolved organic carbon-[DOC]-were titrated with copper under standardised conditions (pH 6 and 7), and concentrations of Cu(2+)-[Cu(2+)]-were measured with an ion-selective electrode. Measured values of [Cu(2+)], which were in the range 10(-11)-10(-5) moll(-1), were compared with those simulated using Humic Ion-Binding Models V and VI. It was assumed that copper speciation was controlled by the organic matter, represented by fulvic acid (FA), together with inorganic solution complexation (calculated with an inorganic speciation model). The models were calibrated by adjusting a single quantity, the concentration of FA. The optimised value-[FA](opt)-was that giving the best agreement, according to least squares, between measured and simulated [Cu(2+)]. The calculations took into account competition by other dissolved (filterable) metals (Mg, Al, Ca, Fe(II), Fe(III), Zn); in the case of Fe(III) it was assumed either that all the dissolved metal was truly in solution, or that the activity of Fe(3+) was controlled by equilibrium with Fe(OH)(3). The assumption about Fe(III) had relatively small effects on the fitting of Model V, but was significant for Model VI, because Model VI represents low-abundance, high-affinity binding sites in humic matter, which are sensitive to Fe(III) competition. Because of its inclusion of the high-affinity sites, Model VI provided better fits of the data than did Model V. Furthermore, Model VI with Fe(3+) activity controlled by Fe(OH)(3) gave smaller variation in the ratio of [FA](opt) to [DOC] than Model VI with all Fe(III) assumed to be in solution. The average [FA](opt)/[DOC] found from the Cu titrations was 1.30, which implies that 65% of the organic matter is 'active' with respect to metal binding. The average ratio of 1.30 is in reasonable agreement with ratios obtained by applying the model to field data sets for charge balance (1.22), Al speciation (1.56) and base titrations of Cu-amended waters (1.45). It is concluded that Model VI/Fe(OH)(3) provides the most reliable predictions of dissolved metal speciation in natural waters; at a total Cu concentration of 1 microM, the predicted concentration of Cu(2+) is expected to be correct to within a factor of 3.6 in 95% of cases.