Interpretation of metal speciation data in coastal waters: the effects of humic substances on copper binding as a test case

Hypotheses regarding the nature and behavior of metal-binding ligands present in natural waters are often based solely on analysis of metal complexation data. In this study, we examined the interpretation of metal titration data in systems where metal binding both by a heterogeneous mixture of ligands, such as humic substances, as well as by a homogeneous class of strong ligands are significant. As a test case, we created simulated competitive ligand exchange data sets based on the copper complexation characteristics of humic substances and other ligand types thought to dominate copper speciation in coastal and fresh waters, and used various data fitting techniques to extract best fit ligand parameters (conditional binding constants and total ligand concentrations) from these data. Different data fitting techniques produced comparable, accurate descriptions of the variation of free copper concentration as a function of total copper over the range of total copper probed by each simulated data set. However, we show that internal calibrations must be applied with extreme caution, since the effect of intermediate-strength binding by humic substances or similar ligands will be indistinguishable from surfactant effects in voltammetric titration data. While the different fitting techniques produced comparable results as long as the data were accurate (e.g., not distorted by internal calibration errors), the presence of the heterogeneous ligands (humic substances) prevented extraction of accurate ligand parameters for the strong discrete ligands. In fact, different sets of fitting parameters produced equally good fits to the data, emphasizing that distinctions between strongly binding (“L1”) and weakly binding (“L2”) ligand classes are arbitrary in waters where heterogeneous mixtures of ligands are present. Total ligand concentrations extracted from fits for which conditional binding constants were pre-specified allowed limited interpretation of the behavior of different ligand types present in the model samples.