Environ. Sci. Technol. 2003, 37, 138-146
In Situ Trace Metal Speciation in Lake Surface Waters Using DGT, Dialysis, and Filtration JENNY GIMPEL,† HAO ZHANG,† W I L L I A M D A V I S O N , * ,† A N D ANTHONY C. EDWARDS‡ Environmental Sciences, IENS, Lancaster University, Lancaster LA1 4YQ, UK, and Macauley Institute, Craigiebuckler, Aberdeen AB15 8QH
In situ measurements of Fe and Mn by dialysis and diffusive gradients in thin-films (DGT) in 5 lakes (pH 4.77.5, ionic strength 0.3-5 mmol l-1) and Cu and Zn in an acidic and circumneutral lake were compared to results from on site filtration. For the most acidic lake (pH 4.7) all measurements agreed, indicating an absence of colloids and negligible complexation by organic matter. There was little difference in the Mn concentrations measured by the three techniques for any lake, consistent with it being free from complexation. Zn measured by dialysis in circumneutral water was only slightly higher than DGT measurements, appropriate to only partial complexation. Substantial differences between dialysis and DGT for Cu were consistent with complexation by fulvic and humic substances, though not to the extent predicted by the speciation code WHAM. To achieve a good fit it was necessary to adjust the pK for Cu-fulvic binding from 0.8 to 1.3 and to assume that fulvic substances dominated. The presence of low molecular weight strong binding ligands would also be consistent with the data. Differences between the three measurement methods were greatest for Fe, attributable to the presence of large oxyhydroxide colloids, organic complexation and low molecular weight, reactive hydrolysis products. Fe and Mn concentrations measured by DGT on samples returned to the laboratory were much lower than in situ concentrations, illustrating the need for in situ measurements. While use of two in situ techniques provided useful information on the speciation of these natural waters, further refinements are required for unambiguous characterization of the solution. The use of DGT with a more restricted gel that excludes complexes with humic substances should provide complementary information to in situ dialysis.
Introduction The problems of determining concentrations of metals that are truly in solution from analysis of water samples are well documented. Metal speciation may change during sampling and storage. Laxen and Chandler (1, 2) found concentrations of filtered metals changed within a 2 h delay between sampling and filtration. Benes and Steinnes (3) measured a * Corresponding author phone: 0044 1524 593935; fax: 0044 1524 593985; e-mail:
[email protected]. † Lancaster University. ‡ Macauley Institute. 138
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 1, 2003
35% drop in Mn concentrations after 1 day’s storage of lake water samples collected by dialysis. Buffle and Leppard (4, 5) list the processes thought to affect metal speciation during storage. These include adsorption to container walls and increased microbial activity that can trigger both aggregation of small colloids into larger particles and oxidation of elements such as Fe and Mn. Artifacts associated with filtration that can affect speciation have been reported (4-6). Horowitz et al. (7, 8) investigated factors that led to over- or underestimation of the dissolved iron concentration in water and suspended sediment samples. For a constant pore size, varying the brand or diameter of the filter or the volume of water filtered was shown to alter significantly the filtrate concentration. Processes thought to be responsible were listed as (a) sorption/desorption of trace elements from the filter or from solids retained by it, (b) inclusion/exclusion of colloidally associated trace elements, and (c) filter clogging. To that list could be added (d) sample contamination when metals are measured at trace levels. Metal concentrations in natural waters are governed by dynamic interchanges rather than reflecting a simple equilibrium state. Removal of a volume of water from such a system isolates it from potential metal sources and sinks (e.g. particle settling) and can alter or cut off the finely balanced physical and chemical processes responsible for the distribution of metal species. Techniques that perform a separation or measurement in situ, without removal of a water sample, offer the best means of determining accurately the components present in true solution, as they circumvent any problems associated with post-sampling changes, such as aggregation or oxidation. Few in situ techniques exist that are easy to use. None measure the full range of metal species in solution but rather provide information about particular fractions of species that are operationally defined. Simultaneous use of different techniques such as filtration, diffusive gradients in thin-films (DGT), anodic stripping voltammetry (ASV), and dialysis can provide complementary information that results in a fuller description of the solution. DGT is a robust, versatile, and easy to use tool that has been deployed in situ in lakes, river, and seawater (9-11). It was used in conjunction with ASV to study copper speciation in a stream of high dissolved organic carbon (DOC) content (12). Samples had to be modified prior to measurement by ASV to ensure adequate supporting electrolyte and pH control. Such problems may be overcome by using microelectrodes coated with a hydrogel (13). Dialysis has been used in situ on a number of occasions (14, 15). More recently, dialysis was deployed in situ in combination with DGT devices in a Canadian lake (16). In this exceptionally low ionic strength water (I < 0.1 mmol L-1) DGT was found to measure more metal in solution than dialysis. This phenomenon was attributed to counter diffusing sodium ions enhancing the diffusion coefficients of metals in the DGT units. In the only other reported codeployment of DGT and dialysis, in water samples collected from a stream, concentrations measured by DGT were generally less than those measured by dialysis (17). This work examines our current ability to characterize accurately the distribution of submicron metal species in lake waters. It systematically compares results, with the aid of the speciation model WHAM (version 1), from in situ measurements of Fe, Mn, Cu, and Zn made by DGT and dialysis with those by on site filtration. Five lakes are studied with ionic strengths ranging from 0.3 to 5 mmol L-1 and pH from 4.7 to 7.5. 10.1021/es0200995 CCC: $25.00
2003 American Chemical Society Published on Web 12/03/2002
FIGURE 1. Diagram of DGT and dialysis assemblies.
Experimental Method DGT and Dialysis. Diffusive gradients in thin films (DGT) employ a binding agent that accumulates solutes after their passage through a well-defined diffusion layer (9). An ionpermeable polyacrylamide hydrogel, of known thickness (∆g) and exposure area (A), is commonly used as the diffusive layer. Chelex 100, incorporated into a second gel layer, serves as the binding agent for trace metals. The gels are enclosed in a small plastic device (Figure 1) that is immersed in solution. The diffusion coefficient (D) of metal ions within the diffusive gel can be independently measured. Fick’s first law of diffusion can then be used to determine the concentration (C) of metal in the solution from the measured mass (M) of metal accumulated in the resin in a known deployment time (t), as shown in equation 1.
C )M∆g/(DtA)
(1)
In this work calculations were initially performed using this simple equation and assuming that all metal species diffused at the same rate as the free metal ions. The resulting value of C reflects the concentration of all metal species in solution that can dissociate as they pass through the gel layer (labile) and that are sufficiently small to pass through the gel. Metal humic species diffuse more slowly than free metal ions and so their contribution will be underestimated. After modeling the metal speciation in solution, calculations were repeated to account for the slower diffusion (lower D values) of metal complexes with humic substances. The diffusive gel in DGT has usually been prepared to a standard formula of 15 vol % acrylamide and 0.3 vol % agarose-derived cross-linker (18), but a change in the agarosederived cross linker since 1998 resulted in all subsequent gels made with this composition having lower diffusion coefficients (19). A gel of alternative composition, which permitted freer diffusion, was prepared, namely 15 vol % acrylamide and 0.12 vol % agarose-derived cross-linker (DGT Research Ltd.). The diffusion coefficients of Cd and Cu in the gel immersed in a solution of 0.1 M NaNO3 at pH 6 was measured using a diaphragm cell in the pseudo-steady-state mode (19) and found to be 94% of those in water. Gels were cast according to published procedures (18). Standard DGT solution deployment devices were used (Figure 1) (18), obtainable from DGT Research Ltd. In this study diffusive gel thicknesses of 0.92 and 1.16 mm were used. Concentrations calculated from devices with these different thicknesses were not significantly different and generally within 10% of each other. This is consistent with there being a negligibly small diffusive boundary layer on the surface of the DGT devices, as would be expected for the wind-mixed surface waters of lakes (10). A protective 0.45 µm cellulose nitrate filter (Whatmans) separated the diffusive gel from solution. This 135 µm thick membrane has been found to behave like an extension of the diffusive layer (9). Some retardation of metal ion diffusion in the filter has been observed in recent unpublished work, but as a 20% diminution in the diffusion coefficient in the filter would affect the calculated concentration by
