SILVER COMPLEXATION IN SURFACE FRESHWATER AND EFFLUENT

Russell T. Herrin*, Anders W. Andren, Martin M. Shafer, and David E. Armstrong (Water Chemistry Program, University of Wisconsin-Madison, 660 N. Park St., Madison, WI 53706. Corresponding author e-mail: rherrin@students.wisc.edu)

ABSTRACT

Results of competing ligand equilibration experiments indicate that the majority of Ag(I) in the filtered phase of river water and sewage treatment plant effluent is strongly complexed to ligands present in those systems. Furthermore, appreciable fractions of these Ag(I) complexes adsorb to Teflon surfaces in unacidified samples. These complexes do not, however, adsorb to glass surfaces. Oxidation of river water and effluent reduce the fraction of Teflon-adsorbed Ag to undetectable levels. These observations indicate that Ag(I) in river waters and effluents is present in the form of strong complexes that are hydrophobic in nature. Organic matter containing thiol functionalities is likely to cause this behavior. Formation of hydrophobic complexes may enhance the bioavailability of Ag(I).

INTRODUCTION

Increasing numbers of researchers and risk managers have agreed that total metal concentration or filterable metal concentration are likely to be incomplete indicators of the risks that metals present in natural waters. Toxicity, trophic transfer, and fate of silver and other metals in aquatic systems are strongly affected by the forms (e.g., the complexes and redox species) in which metals are present. Relatively little research has been performed to determine the speciation of cationic silver (Ag(I)) in fresh waters. One previous study (Adams and Kramer 1999) exploited a competing ligand equilibration–silver titration approach. In contrast, an investigation of Ag(I) speciation in ocean and estuarine waters using the same technique indicated that no ligands strong enough to compete with DDC were present, and that silver-chloride complexes likely dominated Ag(I) chemistry in these systems (Miller and Bruland 1994).

We performed Ag(I) speciation investigations on three rivers not directly impacted by treatment plant effluents, and on one undiluted treatment plant effluent. Our goal was to determine an effective natural ligand concentration and natural ligand-Ag(I) stability constant. While it is possible that several chemically distinct ligands affect the speciation of filterable Ag(I) in a given water, determination of a single concentration-stability constant pair provides valuable information on the fraction of Ag(I) that is likely to be present as free Ag⁺. Results of our experiments (Herrin 1999) indicated that filterable Ag(I) speciation in rivers is dominated by complexation by sub-nM concentrations of ligands with very high stability constants with respect to Ag(I) (Table 1). The magnitude of these stability constants point toward sulfides or thiol groups on dissolved organic species as important silver ligands. Results of competing ligand tests performed on the treatment plant effluent did not allow calculation of a concentration and stability constant for ligands present in the effluent, but comparison to river water results indicates that the treatment plant effluent contains a larger concentration of ligands with a lower effective stability constant (Herrin 1999).

Additional information on the ligands with which Ag(I) associates came from an unexpected source. Quality control tests performed during the competing ligand experiment revealed that the material from which sample containers were constructed had a dramatic effect on the fraction of silver recovered during competing ligand equilibration experiments. In experiments in which Teflon sample containers were used, a significant fraction of total silver was found associated with bottle walls. When glass containers were used, however, all detectable silver was found in the aqueous phase or associated with the competing ligand. These differences in silver adsorption, in view of the surface properties of Teflon and glass, indicate that the majority of strong ligands with which Ag(I) associates in river waters are likely to have a hydrophobic character, and are therefore likely organic.

METHODS

Water for speciation experiments was collected from three different rivers, and from the effluent stream of a sewage treatment plant. Chemical characteristics of these systems are collected in Table 2. Samples were collected and experiments were performed using trace-metal clean procedures. Samples were filtered in the field through a cartridge filter with a pore size of 0.45 mm. Filtered water was composited in 5 or 2.5-L glass or Teflon containers and then divided into 500-mL glass or Teflon bottles. Silver concentration in samples was increased by 0.46 nM by spiking with AgNO₃(aq) before experiments were initiated.

Two competing-ligand procedures were performed on each sample. In the first, the functional group on a measured mass of Chelex resin was allowed to come to equilibrium with a sample. Sample and resin were separated by filtration, and the metal associated with the resin was eluted with 1-M HNO₃. Characterization of Chelex as a competing ligand is described in Herrin (1999). Mass of silver associated with the resin and mass in the aqueous phase were determined. In the second procedure, DDC(aq) was equilibrated with the sample. Silver complexed by DDC was extracted from aqueous solution with a small volume of CHCl₃, and subsequently back-extracted into 6-M HNO₃ for concentration analysis. A subsample of the aqueous phase was retained for concentration analysis as well. This procedure is similar to published methods (Miller and Bruland 1994, Adams and Kramer 1999). In both cases, experiments were also performed on samples that had been amended with hydrogen peroxide (H₂O₂) and illuminated with ultraviolet light for 24 hours or more to oxidize most organic matter and any other oxidizable species.

Silver concentrations were determined using graphite furnace atomic absorption spectrophotometry (GFAAS). Aqueous phases, eluates, and extracts were analyzed for silver concentration. In addition, bottles in which Chelex equilibrations were performed and separatory funnels in which DDC equilibrations were performed were shaken for 12 hours with 10 mL of 1-M HNO₃. This solution was analyzed for silver concentration in an effort to quantify mass of silver adsorbed to bottle and funnel surfaces.

RESULTS AND DISCUSSION

In DDC/solvent extraction experiments performed on water from all samples, silver concentrations in acid rinses of the separatory funnels were undetectable. Results indicated that all available silver could be accounted for as Ag(aq) or extracted silver, though a correction was required for an analytical interferent (likely CHCl₃) in the aqueous phase. These observations held whether glass or Teflon separatory funnels were used in experiments. Similar results were observed when Chelex equilibrations were performed in glass bottles. Recoveries were in all cases 90 percent or better (Fig. 1), and silver concentrations in 1-M HNO₃ bottle rinses were undetectable (<0.3 nM).

When Chelex equilibrations were performed in Teflon on the same sample matrices, however, the sum of silver masses in the aqueous and Chelex-associated phases represented only 2 to 70 percent of total silver. Five to 60 percent was found to be associated with bottle surfaces. This observation agrees well with those of Wen et al. (1997), who noted that if samples were stored unacidified in Teflon for long periods, quantitative recovery of silver was not possible without UV oxidation. Our storage times (~24 h) were much shorter than those of Wen et al (~2 months), thus extending their conclusions to shorter storage times. Recoveries in the aqueous and Chelex-associated phases were better in the treatment plant effluent, but 8 percent of total silver was found in the bottle rinse. It is likely that 1-M HNO₃ bottle rinses did not quantitatively remove silver from the bottle surfaces, leading to incomplete total recoveries in cases where bottle sorption was significant (Fig. 1).

Silver losses to Teflon surfaces were evident only in Chelex experiments performed on unoxidized river water or effluent samples. When samples were oxidized with H₂O₂ and UV, greater than 85 percent was found in the aqueous and Chelex phases, and less than 1 percent was found in bottle-association tests. In addition, good recoveries and little evidence of bottle sorption were observed in results of Chelex experiments on laboratory-created solutions containing ultrapure water, NaHCO₃, AgNO₃, and in some cases NaCN (Fig. 1, “CN expt”).

These results show a pattern in which a significant fraction of Ag(I) in solution associates with Teflon bottle surfaces when filterable, oxidizable river-water species (or, to a lesser extent, species resulting from sewage treatment) are in solution. Thus, it appears that Ag(I) associated with bottle walls is in the form of complexes with river-water species. These complexes do not, however associate with glass surfaces, indicating that they have some hydrophobic character. Interestingly, the river system lowest in dissolved organic matter concentration (Black Earth Creek) is the system for which losses to Teflon were the most extreme. Quantity of organic matter is therefore not a good predictor for the extent of Ag(I)-ligand losses to Teflon. The chemical nature (e.g., the hydrophobicity) of ligands binding Ag(I) is likely the important determining factor.

The apparent hydrophobicity of Ag(I) complexes with natural ligands may have important implications for the interaction of Ag(I) with aquatic biota. Recent experiments have shown that uncharged AgCl complexes pass through phytoplankton membranes more easily than Ag⁺ or AgCl₂^- (Reinfelder and Chang 1999). If hydrophobic silver complexes in rivers and other aquatic systems are sufficiently small to be transported across a cell membrane, they may enhance the bioavailaibility (though not necessarily the toxicity) of Ag(I).

REFERENCES

Adams N.W.H., Kramer J.R. (1999) Environ. Toxicol. Chem.18: 2674-2680.

Herrin R.T. (1999) Ph. D. Dissertation, The University of Wisconsin, Madison, WI.

Miller L.A., Bruland K.W. (1994) Anal. Chim. Acta 284: 573-586.

Reinfelder J.R., Chang S.I. (1999) Environ. Sci. Technol. 33: 1860-1863.

Wen L.S., Tang D., Lehman R., Gill G., Santschi P. (1997) In: Proc. Fifth Intl. Conf. on Transport, Fate, Effects of Silver in the Environ. (A.W. Andren, T.W. Bober, Editors), Madison, University of Wisconsin Sea Grant, pp. 415-420.

Table 1: Ligand stability constants and concentrations calculated based on comparison of experiments in which DDC was the competing ligand and Chelex was the competing ligand. All silver was associated with DDC in MMSD effluent, so the model could not be applied to this system. See Table 2 footnote for meaning of abbreviations.

System	Batch Chelex, DDC		[S(II-)]filterable (nM)
	log K (AgL)	[L]tot (nM)
BEC	14.3	0.62	12
MMSD	-----	-----	71
Blk R.	15.4-16.2	0.41-0.42	16
Miss. R.	14.9	0.52	13

Table 2: Selected characteristics of systems studied.

System^*	Sampling Date	pH	Conductivity (mS cm^-1)	DOC (mg L^-1)	[Ag]filterable (pM)	[Ag]total (pM)	SPM (mg L^-1)
BEC 1	2/10/99	8.1	625	3.1	20	7.9	17.7
BEC 2	4/12/99	8.1	620	2.5			17.7
MMSD	5/26/99	7.8	1730	6.0	310	1300	12.5
Blk. R.	6/29/99	8.3	110	15.4	21	28	5.2
Miss. R.	8/2/99	8.4	340	9.9	9	44	17.1

^* BEC 1: Black Earth Creek, first sampling; BEC 2: Black Earth Creek, second sampling;

MMSD: Madison Metropolitan Sewerage Treatment Plant effluent; Blk. R.: Black River;

Miss. R.: Mississippi River.

Fig. 1: Mass balance results for Chelex competing ligand experiments on

laboratory samples (CN expt), river waters and treatment plant effluent.

See footnote to Table 2 for meaning of abbreviations.