Environ. Sci. Technol. 2002, 36, 1061-1068
Comparison of Copper Speciation in Coastal Marine Waters Measured Using Analytical Voltammetry and Diffusion Gradient in Thin-Film Techniques M I C H A E L R . T W I S S * ,† A N D JAMES W. MOFFETT Department of Marine Chemistry and Geochemistry, Woods Hole Oceanographic Institution, Woods Hole, Massachusetts 02543
The diffusion gradient in thin-film hydrogel (DGT) probe is a promising tool for metal speciation work. Based on a passive sampling principle, it provides the potential for large data sets in complex regimes. DGT probes were deployed in waters characterized independently using competitive ligand exchange-adsorptive cathodic stripping voltammetry (CLE-ACSV). The CLE-ACSV used benzoyl acetone as the competitive ligand in discrete water samples collected during the deployment of the DGT probes. The DGT probes used a 15% polyacrylamide/0.4% bis-acrylamide crosslinker hydrogel and a Na-form of Chelex-100 to complex metal that fluxed into the probe through the hydrogel. Probes were deployed in locations characterized by the degree of pollution impact: the relatively pristine Vineyard Sound, MA, [Cu]total ≈ 6 nM, small seasonally active harbors on Cape Cod, MA, [Cu]total ) 12-64 nM, as well as a large polluted estuary, the Elizabeth River, VA, [Cu]total ) 4458 nM, and a large polluted port, San Diego Harbor, CA, [Cu]total ) 23-103 nM. This is the first study where DGT probes have been compared with an independent speciation technique in marine systems and used to establish the diffusion coefficient of Cu-complexing ligands in situ. Results showed that the probes produced highly precise data sets, with substantial differences in copper accumulation between contaminated and pristine waters. Comparison of DGT results with CLE-CSV indicate that at least 10-35% of the organically complexed copper derived by CLE-ACSV measurements was DGT-labile. Diffusion coefficients (corrected to 25 °C) of organically complexed DGTlabile Cu through the hydrogel ranged from 0.77 × 10-6 cm2 s-1 in Vineyard Sound to 2.16 × 10-6 cm2 s-1 in the Elizabeth River estuary. Accumulation rates of copper were substantially higher in contaminated waters than in pristine waters, suggesting that the probes in their current form may be useful as tracking tools to detect episodic sources of contamination.
Introduction Recent advances in in situ water quality sensors have led to the development of the diffusion gradient in thin-film * Corresponding author phone: (416) 979-5000, ext. 6565; fax: (416) 979-5044; e-mail:
[email protected]. † Current address: Department of Chemistry, Biology and Chemical Engineering, Ryerson Polytechnic University, 350 Victoria Street, Toronto, Ontario M5B 2K3, Canada. 10.1021/es0016553 CCC: $22.00 Published on Web 02/01/2002
2002 American Chemical Society
hydrogel (DGT) technique (1, 2) that, in principle, has the ability to integrate trace metal concentrations over prolonged periods. The basic principle underlying DGT is that the target analyte is accumulated at a rate that is proportional to the external concentration. A rationale for this approach centers on the principle drawback of protocols based on discrete sampling that may not accurately reflect metal levels in environments exhibiting high spatial and temporal variability. The use of DGT probes to determine metal concentrations is complicated by chemical speciation. For elements such as Mn, which in the dissolved form is essentially 100% Mn2+(aq), it is straightforward to determine total concentrations from DGT accumulation data. However, for elements such as Cu, where speciation is dominated by organic complexes spanning a range of stabilities and size, it is much more difficult. Recently, Zhang and Davison (3) showed that Cu complexes with humic and fulvic acids do accumulate in DGT probes and display differences in diffusion rate that are a characteristic function of gel composition. In this study, we deployed DGT probes in coastal marine waters from a variety of locations that displayed varying degrees of Cu contamination. We compared our data with Cu speciation data obtained by cathodic stripping voltammetry; we believe this is the first application of DGT probes in marine waters that has been characterized by an independent technique. The objectives were to study DGT accumulation properties in waters where Cu chemistry was significantly different, and to ascertain how organically complexed Cu in these systems was behaving. In general, Cu was most strongly complexed in noncontaminated locations and less strongly complexed in contaminated waters because, in the latter, most of the very strongest chelators were completely saturated. Consequently, we anticipated substantial differences in the responses of the probes among these waters. One application of DGT probes is as a tracking device to determine point sources of contamination in a receiving water. For such an application, an operationally defined measurement approximately proportional to total Cu is sufficient. However, an ultimate goal of our work is to provide a thermodynamically defined measurement by designing a probe that only accumulates inorganic Cu. Provided that only inorganic copper complexes diffuse through the gel and accumulate on the resin, free Cu2+ can be determined from stability constant data for the principal inorganic complexess an important objective because copper toxicity in natural waters is generally considered to be proportional to the free cupric ion concentration (4). One aim of this study was to establish how close we are to achieving that objective in marine systems. The DGT technique has been applied to aquatic (5, 6), edaphic (7), and sediment (8, 9) environments. Although the DGT technique has been applied in the marine environment (1, 10), extensive field testing in coastal marine systems has not been conducted. The DGT technique shows promise as a cost-effective means of measuring labile Cu in natural waters. It is based on Fick’s First Law of Diffusion. The following formula is used to calculate the concentration of trace metal in the ambient natural water that, during a defined deployment period, will diffuse through a hydrogel and be effectively complexed by a metal-chelating resin found at a discrete depth within the hydrogel (this metal is referred to herein as DGT-labile metal)
[M′] ) (M∆g)/(DAt) VOL. 36, NO. 5, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
(1) 9
1061
where [M′] ) concentration of DGT-labile trace metal in the bulk solution, mol cm-3; M ) mass of metal flux into the probe, mol; ∆g ) thickness of diffusive layer (diffusive gel plus protective membrane filter plus diffusive boundary layer [DBL]), cm; D ) diffusivity of metals in aqueous solution, cm2 s-1; A ) surface area of diffusion, cm2; t ) duration of deployment, s. In the current use of the DGT in natural systems, the term “DGT-labile Cu” applies to any Cu that is diffusible through the polyacrylamide hydrogel and is complexed by the Chelex resin. The concentration of this mobile and labile Cu is related to the bulk solution by applying a single diffusion coefficient based initially in this study on the diffusivity of inorganic Cu. However, DGT-labile Cu may be comprised of both inorganic and organically complexed Cu. Thus, flux of Cu into a DGT probe will be related to the concentration and respective diffusivity of the main Cu species (organic or inorganic) in solution. Although the diffusion coefficients of most metals in seawater are known (11), the diffusion coefficient of organic trace metal complexing ligands is not as well characterized. In unison with voltammetric analysis, DGT is used in the present study to establish the diffusion coefficients of natural Cu-complexing organic matter in various coastal regions. Competitive ligand exchange-adsorptive cathodic stripping voltammetry (CLE-ACSV) is a widely accepted technique for determining the organic speciation of several important bioactive trace metals in seawater including copper (1215). In CLE-ACSV, a known concentration of a model ligand (a well-characterized metal-binding ligand that forms an electroactive complex with the metal of interest) is added to the sample. After a suitable equilibration period, the concentration of electroactive metal-model ligand complex in the sample can then be determined voltammetrically using a mercury drop electrode. From thermodynamic and mass balance principles, and the analytical measurement of the ambient total dissolved concentration of metal, the concentration of the free metal cation in the water sample can be calculated. Although widely used, the technique is prone to a variety of interference, particularly the surface-active compounds present in many natural waters. Error propagation in the calculations can be large if the competing ligand is much stronger (or weaker) than the natural ligands. These issues are reviewed by Bruland et al. (16), who show that while determination of conditional stability constants and concentrations of ligands is influenced by these factors, agreement between different voltammetric protocols produces remarkably good agreement on free Cu2+ estimates. In this work, we consider CLE-ACSV to be a reliable technique for free Cu2+ determination for comparative purposes. We utilized a CLE-ACSV technique using benzoylacetone as the added model ligand (13, 17). Additional calibrations of the method were carried out so that it would provide an accurate estimate of [Cu2+] in coastal seawater characterized by high [DOC] and variable salinity. Study sites were chosen that represented a range in both copper contamination and DOC content so that the response of the DGT technique could be assessed under conditions most likely to be encountered in coastal marine and estuarine environments. In addition, several experiments were conducted in the laboratory using natural seawater that was amended with synthetic ligands in order to test various hypotheses regarding the flux of organically complexed copper through the hydrogel in the DGT probe. Field trials to compare the DGT and CLE-ASCV techniques were conducted in the water column of coastal waters chosen to represent a range of trace metal pollution and salinity. Both techniques were simultaneously applied to the same water body to provide as close a comparison as possible using these methods that are based on fundamentally different approaches to measuring copper speciation. Probes 1062
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 5, 2002
FIGURE 1. Diffusion gradient in thin-film hydrogel probe design showing the simple assembly of the probe. were deployed in locations characterized by the degree of copper contamination. Vineyard Sound, MA, with concentrations of copper around 6 nM, was the low-end member of our study. Two Cu-contaminated harbors adjacent to Vineyard Sound exhibit concentrations that are typically 5-10-fold higher, which has been attributed to seasonal inputs from antifouling paints (17). Two urban estuaries were also investigated, the Elizabeth River Estuary, Norfolk, VA, and San Diego Bay, CA.
Experimental Section Diffusion Gradient in Thin-Film. Hydrogels were retained in an acrylic piston-type device (2), referred to herein as a “probe” (Figure 1). The window on the probe provided an area for diffusion of 11.3 cm2. The diffusive gel was covered by a 0.0125 cm-thick 0.45 µm-pore size filter (polyvinylidene fluoride membrane filter); the filter type was selected for its low protein binding affinity, as stated by the producer (Millipore Corp.). The probes and membrane filters were soaked in 10% HCl and rinsed exhaustively with deionized water (18.2 MOhms•cm-1) water prior to use. Some deployments used larger probes with a 44.2 cm2 window. Gel Matrixes. Two hydrogel matrixes were used in this study: Fye hydrogel and APA hydrogel. In the preparation of all gels, the polymerized gels were cast on acid-clean glass plates with thickness established using plastic spacers between the plates. Gels were soaked in 0.1 M NaNO3 (prepared from a concentrated NaNO3 stock purified by passage through Chelex-100) for 1-3 days prior to use, with the solution changed at least three times. After the soaking step, Fye and APA gels cast at 0.0400 cm were measured using a microscope equipped with a micrometer to be 0.0411 and 0.0489 cm, respectively. Only one diffusive gel thickness was used in field deployments of DGT probes. A DBL of 0.031 cm has been estimated in the water column of a small quiescent lake using DGT probes with various diffusive layer thickness (18). Considering the high energy environment of the coastal marine environment (wind- and tide-driven currents), the DBL is assumed to have been present, yet smaller than the hydrogel and protective filter diffusive layer thickness (0.0525 cm) used in most of the DGT probes deployed in the field. A DBL thickness of 0.010 cm (95% confidence interval of 0.018-0.003 cm, r2 ) 0.995) was estimated for a deployment in Eel Pond Harbor (November 1997) that used variable diffusive gel thicknesses, as determined by plotting 1/M for
manganese flux into the DGT probes versus the combined thickness of the protective filter and diffusive hydrogel (the extrapolated intercept at 1/M ) 0 is the estimated thickness of the DBL; 2) The estimated DBL thickness was added to the diffusive gel and filter thickness in order to provide an estimation of ∆g; a DBL of 0.01 cm was assumed for all field deployments in this study. The Fye hydrogel (named after the Paul MacDonald Fye Laboratory, WHOI) formulation is 15% acrylamide with 0.4% bis-acrylamide as a cross-linker. The Fye hydrogel was prepared in the following proportions and manner: 10 mL of a 15% acrylamide and 0.4% bis-acrylamide solution at room temperature was degassed at
