Comparison of analytical techniques for dynamic trace metal speciation in natural freshwaters.

Several techniques for speciation analysis of Cu, Zn, Cd, Pb, and Ni are used in freshwater systems and compared with respect to their performance and to the metal species detected. The analytical techniques comprise the following: (i) diffusion gradients in thin-film gels (DGT); (ii) gel integrated microelectrodes combined to voltammetric in situ profiling system (GIME-VIP); (iii) stripping chronopotentiometry (SCP); (iv) flow-through and hollow fiber permeation liquid membranes (FTPLM and HFPLM); (v) Donnan membrane technique (DMT); (vi) competitive ligand-exchange/stripping voltammetry (CLE-SV). All methods could be used both under hardwater and under softwater conditions, although in some cases problems with detection limits were encountered at the low total concentrations. The detected Cu, Cd, and Pb concentrations decreased in the order DGT > or = GIME-VIP > or = FTPLM > or = HFPLM approximately = DMT (>CLE-SV for Cd), detected Zn decreased as DGT > or = GIME-VIP and Ni as DGT > DMT, in agreement with the known dynamic features of these techniques. Techniques involving in situ measurements (GIME-VIP) or in situ exposure (DGT, DMT, and HFPLM) appear to be appropriate in avoiding artifacts which may occur during sampling and sample handling.

Model predictions of metal speciation in freshwaters compared to measurements by in situ techniques.

Measurements of trace metal species in situ in a softwater river, a hardwater lake, and a hardwater stream were compared to the equilibrium distribution of species calculated using two models, WHAM 6, incorporating humic ion binding model VI and visual MINTEQ incorporating NICA-Donnan. Diffusive gradients in thin films (DGT) and voltammetry at a gel integrated microelectrode (GIME) were used to estimate dynamic species that are both labile and mobile. The Donnan membrane technique (DMT) and hollow fiber permeation liquid membrane (HFPLM) were used to measure free ion activities. Predictions of dominant metal species using the two models agreed reasonably well, even when colloidal oxide components were considered. Concentrations derived using GIME were generally lower than those from DGT, consistent with calculations of the lability criteria that take into account the smaller time window available forthe fluxto GIME. Model predictions of free ion activities generally did not agree with measurements, highlighting the need for further work and difficulties in obtaining appropriate input data.