Improving and testing geochemical speciation predictions of metal ions in natural waters.

The ability of WHAM VII and NICA-Donnan models to predict free-ion activities of Cu in natural waters was examined from two perspectives, (i) the presence of EDTA and NTA contaminants, (ii) the need to improve estimates of HA and FA concentrations. Potentiometric responses of a Cu(II) ion-selective electrode were investigated in five assays containing dissolved organic matter (DOM) isolated from a series of polluted (urban) and relatively unpolluted (upland) streams in northern England. The [Cu]/[DOC] ratio in these assays spanned an environmentally realistic range of ∼1-500 µmol/g. Reasonably good agreement between measured and predicted Cu(2+) activities was obtained with both WHAM VII and NICA-Donnan models, assuming 65% of DOM as fulvic acid and including the measured EDTA and NTA concentrations, but generally the models overestimated the activities by a factor of ∼2. In contrast, the models over-predicted the Cu(2+) activities by up to 2 orders of magnitude at low [Cu]/[DOC] ratios in urban waters if anthropogenic ligands were not included in the model simulations. Three-dimensional fluorescence excitation-emission matrix (EEM) spectroscopy was used to measure the functional properties of the isolated DOM and to estimate the fractions of FA and HA present. Using these fractions in the models gave improvements in predictions compared to the 65% FA assumption, as shown by higher correlations, reduced error and reduced bias. These results highlight various issues with the use of the available speciation models for predicting free ion concentrations in natural waters, such as the use of the Biotic Ligand Model (BLM) for the derivation of environmental standards. It is clearly necessary to measure EDTA and NTA in waters with urban influences, while fluorescence measurements offer the possibility of appreciably improving the accuracy of predictions.

Testing copper-speciation predictions in freshwaters over a wide range of metal-organic matter ratios.

The harsh chemical conditions involved in the isolation of fulvic acids (FA) and humic acids (HA) have been identified as a possible contributing factor to the significant mismatch between in situ measurements and model predictions of trace metal speciation in freshwaters, resulting from the use of isolated FA and HA in model calibration. A set of experimental assays were developed to enable Cu binding to DOM to be measured over the full range of [Cu]/[DOC] ratios (∼1-460 µmol g(-1)) observed in surface freshwaters. They were applied to the widely used and traditionally isolated Suwannee River HA and FA and to DOM isolated from headwater streams by a mild procedure using minimal chemical treatment. Good agreement was observed between measured free ion activities and those predicted using both WHAM/Model VII and NICA-Donnan speciation models for both traditionally and mildly isolated DOM. Agreement to within a factor of 2 for WHAM/Model VII contrasts with 100-fold differences previously reported between in situ Cu(2+) measurements and model predictions for a wide range of conditions. The results demonstrate that (a) existing speciation models are capable of accurately predicting Cu-humic binding in natural waters at environmentally realistic [Cu]/[DOC] ratios, under equilibrium conditions, and (b) that the isolation procedures traditionally used for HA and FA do not appreciably affect their binding characteristics.

Freshwater DOM quantity and quality from a two-component model of UV absorbance.

We present a model that considers UV-absorbing dissolved organic matter (DOM) to consist of two components (A and B), each with a distinct and constant spectrum. Component A absorbs UV light strongly, and is therefore presumed to possess aromatic chromophores and hydrophobic character, whereas B absorbs weakly and can be assumed hydrophilic. We parameterised the model with dissolved organic carbon concentrations [DOC] and corresponding UV spectra for c. 1700 filtered surface water samples from North America and the United Kingdom, by optimising extinction coefficients for A and B, together with a small constant concentration of non-absorbing DOM (0.80 mg DOCL⁻¹). Good unbiased predictions of [DOC] from absorbance data at 270 and 350 nm were obtained (r² = 0.98), the sum of squared residuals in [DOC] being reduced by 66% compared to a regression model fitted to absorbance at 270 nm alone. The parameterised model can use measured optical absorbance values at any pair of suitable wavelengths to calculate both [DOC] and the relative amounts of A and B in a water sample, i.e. measures of quantity and quality. Blind prediction of [DOC] was satisfactory for 9 of 11 independent data sets (181 of 213 individual samples).

On the acid-base properties of humic acid in soil.

Humic acid was isolated from three contrasting organic-rich soils and acid-base titrations performed over a range of ionic strengths. Results obtained were unlike most humic acid data sets; they showed a greater ionic strength dependency at low pH than at high pH. Forward- and back-titrations with the base and acid revealed hysteresis, particularly at low pH. Previous authors attributed this type of hysteresis to humic acid aggregates-created during the isolation procedure-being redissolved during titration as the pH increased and regarded the results as artificial. However, forward- and back-titrations with organic-rich soils also demonstrated a similar hysteretic behavior. These observations indicate (i) that titrations of humic acid in aggregated form (as opposed to the more usual dissolved form) are more representative of the acid-base properties of humic acid in soil and (ii) that the ionic strength dependency of proton binding in humic acid is related to its degree of aggregation. Thus, the current use of models based on data from dissolved humic substances to predictthe acid-base properties of humic acid in soil under environmental conditions may be flawed and could substantially overestimate their acid buffering capacity.