2667

Environmental Toxicology and Chemistry, Vol. 18, No. 12, pp. 2667–2673, 1999q 1999 SETAC

Printed in the USA0730-7268/99 $9.00 1 .00

Environmental Chemistry

SILVER SPECIATION IN WASTEWATER EFFLUENT, SURFACE WATERS, ANDPORE WATERS

NICHOLAS W.H. ADAMS and JAMES R. KRAMER*Department of Geology, McMaster University, Hamilton, Ontario L8S 4M1, Canada

(Received 19 August 1998; Accepted 10 February 1999)

Abstract—Silver, inorganic sulfide, and thiol compounds were measured in municipal wastewater effluent, receiving waters, andpore waters from an anoxic lake sediment in order to predict silver speciation in these systems. We found submicromolar concen-trations of inorganic sulfide even in fully oxic surface water. This inorganic sulfide is likely to exist in the form of colloidal metalsulfides, which have been shown to be stable under oxidizing conditions for periods of several hours. Inorganic sulfide in both thewastewater effluent and receiving waters was found to be 200 to 300 times in excess of silver concentrations, whereas inorganicsulfide in the pore waters was 1,000 to 15,000 times in excess of silver concentrations. With sulfide in excess of silver, we predictsilver sulfide complexes to dominate silver speciation. Thiols were present at low nanomolar levels in pore waters but were notdetectable (,1 nM) in wastewater effluent or receiving waters. Thiols do not appear to be important to silver speciation in thesefreshwater systems. Partitioning of silver into particulate (.0.45 mm), colloidal (10 kDa, 0.45 mm), and dissolved (,10 kDa) sizefractions showed that a significant proportion of silver is in the colloidal (30–35%) and dissolved phases (15–20%). Dissolvedphase concentrations were relatively constant in the treatment plant effluent and receiving waters, suggesting that silver in the ,10-kDa size fraction is strongly complexed by ligands that are not significantly affected by aggregation or sorption processes.

Keywords—Silver Sulfide Thiols Speciation Wastewater

INTRODUCTION

Silver is found in unperturbed aquatic systems at ultratracelevels (5–100 pM) [1–5], whereas silver concentrations nearthe outfalls of municipal wastewater treatment plants (WTP)tend to be enhanced by at least one to two orders of magnitude[6]. It is of concern that these enhanced silver concentrationsmay affect aquatic biota.

The bioavailability and toxicity of metals are integrallylinked to their chemical forms [7,8]. In order to conduct mean-ingful toxicity experiments, we must therefore have a knowl-edge of how silver speciates in the environment. To date, fewstudies exist that deal with the chemical forms of silver innatural waters. Based on the available thermodynamic data,we previously postulated that silver(I) is likely to be com-plexed by sulfides under almost all natural conditions. Silver(I)forms very strong complexes with S(2II) species. The for-mation constant for Ag1 1 HS2 ø AgHS0 3 log K ; 13 [9].Formation constants of similar magnitude have been deter-mined for several organic sulfide compounds (i.e., thiols) [10].Compounds containing oxygen and nitrogen groups displaymuch weaker silver binding (e.g., ethylenediaminetetraaceticacid [EDTA], log K ; 6) [11].

In marine and estuarine pore waters, sulfide is present inmillimolar concentrations [12,13]. Other reduced sulfur spe-cies are present at lower concentrations, including a numberof thiol compounds [14]. In addition, sulfides have been shownto occur at nanomolar and picomolar concentrations in theocean [15,16]. Little equivalent data exist for freshwater. T.F.Rozan and G. Benoit (personal communication) recently mea-sured metal sulfide species at nanomolar concentrations in ox-ygenated river waters. Wang et al. [17] have also measured

* To whom correspondence may be addressed(kramer@mcmaster.ca).

various sulfur species in pore waters from anoxic lake sedi-ment.

Our thermodynamic calculations suggest that silver will becomplexed by S(2II), as long as S(2II) concentrations arehigher than silver concentrations [18]. Under this condition,the free silver ion concentration should be negligible, becauseof the strength of the silver–sulfur(2II) bond.

In this article, we present the results of a study examiningsilver, sulfide, and thiol concentrations in pore water, surfacewater, and WTP effluent samples. We also present data on thedistribution of these species among the particulate, colloidal,and dissolved phases.

METHODS

Surface water and wastewater effluents were sampledthroughout the spring and summer of 1997 and 1998. Waste-water effluents were sampled at the Dundas and Burlington(Ontario, Canada) municipal WTPs. Surface water sampleswere taken at several locations downstream from the Dundasplant in the Desjardin Canal.

Both treatment plants use a tertiary activated sludge treat-ment process and add ferric chloride to enhance floc formationand setting in the clarifier tanks. Disinfection is accomplishedthrough chlorination, and treated effluent from the Dundasplant also passes through a sand filter as the final step in thetreatment process.

Low-level trace metal techniques were used to avoid con-tamination and to provide sensitive and reliable analyses. Pro-cedures for the preparation of sampling apparatus, processingof water samples, and analysis by inductively coupled plasmamass spectrometry (ICP-MS) were adapted from protocolsused by the Lake Michigan Tributary Monitoring Project [19].All surfaces contacting water samples were acid-cleaned in20% ACS grade HNO3 and 1% high-purity TMA HNO3 acid

2668 Environ. Toxicol. Chem. 18, 1999 N.W.H. Adams and J.R. Kramer

Table 1. Inductively coupled plasma–mass spectroscopy operatingconditions

RF power (W)Argon plasma gas (L/min)Argon aux gas (L/min)Argon neb gas (L/min)Operating vacuum (Torr)Cones

1,2001221.31 3 1025

NickelNebulizer Cetac 5100AT ultrasonicSample uptake (ml/min) 1–1.5Quad mode Peak hop, 200 ms dwell,

three points per peakAcquisition 90 s, five cycles, 10 replicatesRinse 300 s, 2% HNO3

Internal standard 4 mg/L In

(J.T. Baker, Phillipsburg, NJ, USA) for a minimum of 3 d.Materials were then rinsed with Milli-Qt water (Millipore,Bedford, MA, USA) and dried in a high-efficiency particulateair–filtered environment. Samples for metal, sulfide, and thiolanalysis were collected in 1-L Teflont bottles stored in zip-lock polyethylene bags. Wastewater samples were collecteddirectly into Teflon bottles by personnel wearing arm-lengthpolyvinyl chloride powderless gloves. Water samples from theDesjardin Canal were collected from a small rubber raft. Inall cases, samples were collected with no headspace to limitoxidation. Samples were returned to the laboratory (,1 h) andplaced in a N2-filled (O2-scrubbed) glovebag. Aliquots of eachTeflon bottle were processed for metal analysis. Subsamplesfor metal analysis were then further processed in the high-efficiency particulate air–filtered cleanhood, whereas subsam-ples for sulfide and thiol analyses were processed in the N2-filled glovebag. In each case, subsamples were filtered using25-mm Millipore 0.45-mm polysulfone syringe filters, alongwith Millipore 10-kDa Ultrafreet-15 centrifugal filters. Filterswere prepared by rinsing twice with 1% TMA HNO3 acid andMilli-Q water in the high-efficiency particulate air enclosure.

Sediment core samples were collected in Hamilton Harbour,Lake Ontario, Ontario, Canada, on August 20, 1997, at threestations with coordinates N4792473.8 E592328.5 (A),N4792258.8 E591758.0 (B), and N4792996.5 E594001.8 (C).Water depth at these locations was between 15 and 20 m.Sediment from station A was sandy compared to that fromstations B and C, which were predominantly silty. Core sam-ples were obtained by pushing 10 3 50-cm2 polypropylenetubes into a box core collected from a barge. Tubes were back-filled with interfacial water and sealed immediately. Two coresamples were collected from each of the box cores taken atthe three stations. Sediments were extruded from the tubes in10-cm increments using a pneumatic extruder. Sediment sub-samples were placed in 125-ml high-density polyethylene bot-tles, cleaned using the trace metal protocol, and backfilled withargon (.99.99%). Subsamples were centrifuged (IEC, Need-ham Heights, MA, USA) at 3,000 rpm (K2,000 g) for 20 min.Pore waters were processed for metal, sulfide, and thiol anal-ysis as described for the surface water and wastewater effluentsamples. No unfiltered samples were analyzed for the porewaters because of its high particulate content.

For metal analysis, 10 ml of the unfiltered, 0.45-mm, and10-kDa filtered subsamples were stored in high-density poly-ethylene, 15-ml centrifuge tubes and acidified with 200-mlUltrex II HNO3 (J.T. Baker). Unfiltered subsamples were dou-ble-bagged in zip-lock polyethylene bags and heated at 608Cfor 24 h. Immediately before analysis, samples were acidifiedwith an additional 200 ml 2 N HNO3 containing 4 mg/L In asan internal standard. Ultratrace metals analyses were per-formed on a PE-Sciex Elan-500 ICP-MS (Perkin Elmer, Nor-walk, CT, USA), dedicated to low-level samples, for a periodof 2 weeks. Because our ICP-MS facility is used for a varietyof sample types, we found it necessary to thoroughly cleanthe instrument before analysis. By following strict protocols,we were able to eliminate contamination and hysteresis prob-lems, which commonly affect ultratrace level ICP-MS. Table1 summarizes ICP-MS operating conditions. Instrumental andprocedural blanks and metal standards were placed randomlywithin a run as part of our quality assurance program. Sampleconcentrations were determined based on ultrapure calibrationstandards (Plasma Chem, Bradley Beach, NJ, USA).

Sulfide concentrations were determined colorimetrically

[20]. The colorimetric reagent was made by mixing 2.25 gN,N-dimethyl-p-phenylenediamine oxalate (Sigma, St. Louis,MO, USA) in 660 ml H2SO4, 340 ml Milli-Q, and 5.4 gFeCl3·6H2O (BDH, Toronto, ON, Canada) in 100 ml HCl and100 ml Milli-Q. Twenty-five milliliters of the unfiltered, 0.45-mm, and 10-kDa filtered subsamples were added to amber glassbottles, which were sealed with Telfon-coated silicone septa.Two milliliters of the colorimetric reagent was added to eachbottle, and the responses were measured 1 h later using aSpectronic 20 (Bauch & Lomb, Rochester, NY, USA) at 670nm. Calibration standards were made from a Na2S solutionstandardized iodimetrically against a thiosulfate standard (An-achemia, Montreal, PQ, Canada).

Thiols were determined by reverse phase high-performanceliquid chromatography. Following the method outlined byMopper and Delmas [21], samples were derivatized using anortho-phthalaldehyde/2-aminoethanol reagent followed byhigh-performance liquid chromatography and fluorometric de-tection. A Bio-Rad model 700 gradient module was used asthe pumping system (Bio-Rad, Richmond, CA, USA). Sepa-ration was carried out with a 250 3 4.6–mm2 Vydact (218TP)C18 reverse phase column (Bio-Rad) containing 5-m packing.A Bio-Sil ODS-10 guard column (Bio-Rad) was placed in frontof the analytical column. Detection was accomplished with aHewlett-Packard 1046A Fluorometric Detector (Hewlett-Pack-ard, Avondale, PA, USA) with the monochromators set to 350nm (excitation) and 420 nm (emission). Chromatograms werecollected on a Hewlett-Packard 3390A Integrator.

Mobile phases consisted of 0.05 M sodium acetate (pH 5.7)(phase A) and acetonitrile (phase B) degassed with helium(.99.999%). Reagents consisted of 20 mg of ortho-phthal-aldehyde (Sigma) in 1 ml of acetonitrile and 25 ml of 2-ami-noethanol (Sigma) in 1 ml of 0.1 M sodium borate. Both re-agents were then diluted 100-fold in the same solvents. Stocksolutions of thiol standards were prepared: glutathione, L-cys-teine, 3-mercaptopropionic acid, 2-mercaptoethanol, thiolacticacid, and ethanethiol. The stock solutions were prepared in 5mM sodium acetate buffer (pH 4.0) degassed with helium andcontaining 0.1 mM EDTA. Mixed standards were preparedfrom stock solutions and diluted to 10 to 1,000 nM with de-gassed Milli-Q water before derivatization. Twenty-five mi-croliters of each reagent was added to 1 ml of standard orsample. Solutions (100 ml loop) Xwere injected after precisely1 min. The 0.45-mm and 10-kDa samples were reacted andinjected with no further treatment. Separations were performedat 258C and at a flow rate of 1 ml/min. The gradient used wasas follows: isocratic at 10% B for 1 min, 10 to 50% B in 10

Silver speciation in water and effluents Environ. Toxicol. Chem. 18, 1999 2669

Table 2. Characteristics of wastewater, surface water, and pore watera

Sample

Temper-ature(8C) pH

Specificconduc-

tance(mS/cm)

Oxygen(mg/L)(% sat) DOC (mg/L)

Secondary clarifierinflow (Dundas) 23.9 7.5 828 8.8 (105) 14.3

Secondary clarifieroutflow (Dundas) 24 7.5 838 9 (107) 10.7

Plant outflow (Dundas) 24.2 7.5 835 9.1 (108) 10.4Plant outfall

(Desjardin Canal) 23.2 7.5 856 8.9 (105) 7.1250 m downstream

(Desjardin Canal) 22.5 7.3 795 8.4 (98) 6.5250 m downstream (bottom

water) (Desjardin Canal) 22.5 7.2 812 6.5 (76) 6.8500 m downstream

(Desjardin Canal) 23.4 7.4 742 8.9 (105) 6.1

a Data collected on 05/06/97.

Table 3. Spike recoveries in percenta

Location Unfiltered0.45 mmfiltered

10 kDafiltered

Dundas wastewater treatmentplant

Desjardin CanalHamilton Harbour pore water

92 6 994 6 1197 6 8

94 6 599 6 495 6 10

102 6 8105 6 13

96 6 14

a 0.5 nM AgNO3 spikes.

min, isocratic at 50% B for 10 min, isocratic at 100% B for5 min, 100 to 10%, B in 1 min, and at 10% B for 10 min.

Unfiltered pore water, wastewater, and surface water sam-ples were reacted with a 1 mM solution of NaBH4 to liberateany thiol compounds bound to particulate matter through di-sulfide linkages. Samples were then filtered, and the pH wasadjusted to near neutral with 0.1 N HCl. This method of re-ducing disulfide linkages was tested on a 1 mM solution ofglutathione disulfide, which was fully recovered as glutathione.Other samples were reacted with 1 mM EDTA to liberate thiolsbound by metals.

Temperature, pH (Ross combination electrode, Orion, Bev-erly, MA, USA), conductivity (Perolin-Bird Archer, Cobourg,ON, Canada), and dissolved oxygen (Model 600, ESD, New-ark, DE, USA) were measured at each sampling location. Dis-solved organic carbon (total organic carbon 2 dissolved in-organic carbon) was determined by infrared detection of CO2

following persulfate–ultraviolet digestion on a Dorhmann-DC180 automated system (Table 2).

RESULTS AND DISCUSSION

Quality assurance for silver data

The instrumental detection limit for ICP-MS analysis ofsilver (three of five blank replicates) ranged from 0.01 to 0.05nM under conditions outlined in Table 1 (laboratory controlblanks consisted of Milli-Q water spiked with the equivalentUltrex IIt HNO3 [J.T. Baker] and in solution used for samples).Field quality assurance for silver measurements included 10%procedural blanks, 25% sample replicates, and 10% AgNO3

spiked samples. Procedural blanks, carried to the field andprocessed concurrently with the samples, were not significantlydifferent from the laboratory control blanks. Individual pro-cedural blanks were always within the 95% confidence intervalof the control blank. Precision relative standard deviation onreplicated samples ranged between 7.5 and 15.1% for low-range samples (0.05–0.25 nM), and between 1.6 and 8.1% forhigher-range samples (.0.25 nM). Samples from each sam-pling location were collected, filtered, and spiked with AgNO3

to determine recoveries. Spiked recoveries are summarized inTable 3.

Quality assurance for sulfide and thiol measurements

Instrumental detection limit for the sulfide analyses was 1to 2 nM (laboratory control blanks consisted of the colori-

metric reagent added to degassed Milli-Q). Procedural blankscarried to the field and processed concurrently with the sampleswere not significantly different from control blanks. Also, fieldsamples with no colorimetric reagent added were not signifi-cantly different from control blanks (i.e., indigenous materialin the samples did not display absorbance at the measurementwavelength, 670 nm). Precision relative standard deviation onall sulfide measurements was greater than 10%. Recoveries ofNa2S spikes from the three filter fractions were 97 6 15%(unfiltered), 96 6 8% (0.45-m–filtered), and 99 6 7% (10-kDa–filtered). Additionally, a 50 nM sulfide solution (pH 8)was filtered through the 0.45-m and 10-kDa filters. Recoverieswere found to be excellent from the 0.45-m filtration (100 65%) but tended to be low and quite variable (74 6 19%) fromthe 10-kDa centrifugal filtration. Even with the special pre-cautions taken in this study, the centrifugation step causedsignificant losses of sulfide.

Instrumental detection limit for the thiol analyses varied,depending on the compound (glutathione and L-cysteine, 25nM; 3-mercaptopropionic acid, 2-mercaptoethanol, thiolacticacid, and ethanethiol, 1–2 nM; laboratory control blanks con-sisted of the fluorometric reagent added to degassed Milli-Q).Procedural blanks were not significantly different from controlblanks. To determine recovery efficiencies for the thiol ana-lyses, 0.45-mm–filtered samples were spiked with 50 nM 3-mercaptopropionic acid (recoveries were 95 6 12%).

Silver in treatment plants

Although overall removal efficiencies for silver in munic-ipal WTPs are generally greater than 95% [6], concentrationsof silver in WTP effluents are enhanced by at least one to twoorders of magnitude, relative to background levels. In thisstudy, we focused on the small fraction of silver that is not

2670 Environ. Toxicol. Chem. 18, 1999 N.W.H. Adams and J.R. Kramer

Table 4. Silver concentrations in the wastewater treatment plants in nanomoles per litera

SampleParticulate

(.0.45 mm)Colloidal

(10 kDa–0.45 mm)Dissolved(,10 kDa)

Secondary clarifier inflow (Dundas)Secondary clarifier outflow (Dundas)Plant outflow (Dundas)Plant outflow (Burlington)

0.95 6 0.130.79 6 0.160.78 6 0.10

1.03

1.12/0.800.97/0.570.89/0.65

—

0.57 6 0.090.41 6 0.030.39 6 0.01

0.5

0.69/0.470.44/0.360.41/0.38

—

0.32 6 0.070.29 6 0.060.28 6 0.04

0.26

0.42/0.250.38/0.230.32/0.23

—

a Mean values 61 SD; high/low values; data compiled from three sampling trips to the Dundas plant (1997–1998).

Table 5. Silver concentrations downstream from the Dundas wastewater treatment plants in nanomoles per litera

LocationParticulate

(.0.45 mm)Colloidal

(10 kDa–0.45 mm)Dissolved(,10 kDa)

Plant outfall250 m downstream250 m downstream (bottom water)500 m downstream

0.84 6 0.380.51 6 0.060.42 6 0.070.45 6 0.10

1.54/0.390.60/0.400.51/0.340.56/0.29

0.42 6 0.040.42 6 0.070.35 6 0.070.40 6 0.04

0.47/0.360.49/0.320.44/0.340.48/0.28

0.25 6 0.090.29 6 0.100.27 6 0.090.31 6 0.07

0.40/0.140.45/0.170.38/0.150.45/0.25

a Mean values 61 SD; high/low values; data compiled from five sampling trips to the Desjardin Canal (1997–1998).

removed during primary treatment. Silver was measured atthree locations in the Dundas plant. Grab samples were takenat the inflow and outflow of the secondary clarifier and at theoutflow from the plant after chlorine addition and sand filtra-tion. Silver concentrations measured in the plant are sum-marized in Table 4.

Particulate-bound (.0.45 mm) silver concentrations in theplant averaged 0.8 to 1 nM. A small decrease in silver con-centrations was observed between the inflow to the secondaryclarifier and the outflow from the plant, although the proportionof particulate silver remained constant (Table 4). The 20%removal observed between the secondary clarifier and the ef-fluent leaving the plant is presumably due to particle settlingin the secondary clarifier and entrapment of particulate matterduring sand filtration. The proportion of particle-bound silverin the plant outflow averaged 54 6 5%, which compares fa-vorably with values obtained by Shafer et al. (60 6 16%),who examined silver partitioning in a number of WTPs locatedin Wisconsin, USA [6].

Colloidally bound (10 kDa, 0.45 mm) silver showed a 30%decrease between the secondary clarifier inflow and secondaryclarifier outflow, suggesting scavenging by larger particles oraggregation of colloidal particles. No significant change in thecolloidal bound silver was observed between the secondaryclarifier outflow and the plant outflow. Colloids containingsilver do not appear to undergo further aggregation or sorptionafter leaving the secondary clarifier and do not appear to beremoved by the sand filter.

Ultrafiltration of samples from the three sample locationswithin the plant showed that dissolved (,10 kDa) silver con-centrations remained constant (Table 4). Dissolved silver doesnot appear to be affected by aggregation or sorption processesin the final stages of treatment. In a concurrent study, we foundhigh affinity sites for silver in the ,10-kDa filter fraction ofthe secondary clarifier [22]. Complexation of silver in the ,10-kDa filter fraction explains its lack of reactivity toward par-ticulate matter, which have previously been shown to havelarge partition coefficients, Kd [4,6].

Although ultrafiltration studies conducted by Shafer et al.[6] found that only 2% of total silver passes through the 10-kDa membrane, we found significantly higher proportions ofdissolved silver (16–30%) in plant effluent. We found an av-

erage of 0.28 6 0.04 nM silver in the ,10-kDa filter fractioncompared to only 0.03 nM measured by Shafer et al. [6]. Asingle set of samples from the Burlington plant also showeda higher proportion of silver in the ,10-kDa filter fraction(15%). During the April 1998 trip to the Dundas plant, wecompared the 10-kDa filtered concentrations, determined usingthe Millipore Ultrafree-15t centrifugal filters, and separationscarried out on a Millipore cross-flow membrane system. Bothmethods gave similar results. In a study by Pham and Garnier[23], river water samples were tagged with radioisotopes ofseveral different metals, including silver. The authors foundthat for the ,0.45-mm size fractions, up to 25% of the silverwas in the ,10-kDa fraction (although some samples did notcontain ,10 kDa of silver).

Silver downstream from the the Dundas treatment plant

The fate of silver leaving the Dundas plant was examinedby sampling at three locations downstream from the plant inthe Desjardin Canal (plant effluent contributes 100% of thebase flow in the canal). Grab samples, taken a few centimetersbelow the water surface, were collected 5 m downstream fromthe submerged plant outflow pipe, midway along the canal(250 m downstream), and immediately upstream of where thecanal discharges to Cootes Paradise marsh (500 m down-stream). Midway along the canal, bottom water was also col-lected at a depth of 3 m through 0.635-cm o.d. Teflon tubinginto a Teflon bottle using a 50-ml HDPP syringe fitted with atwo-way valve. Silver concentrations measured downstreamof the Dundas plant are summarized in Table 5.

Particulate-bound silver (0.8 6 0.4 nM) was quite variableat the plant outfall, displaying extremes not captured duringsampling trips to the treatment plant. Particulate-bound silverconcentrations decreased by ø40% between the outfall andthe second sampling location 250 m downstream and a further10% at 500 m downstream (Table 5). Because the effluentfrom the Dundas plant represents the only major input intothe Desjardin Canal, losses of silver in the particulate phasemust be due to nonconservative processes such as particlesettling, sorption, and uptake by biota. In contrast, concentra-tions of colloidally bound (10 kDa, 0.45 mm) and dissolvedsilver (,10 kDa) remained relatively constant over all threesampling locations. Silver in these filter fractions appears un-

Silver speciation in water and effluents Environ. Toxicol. Chem. 18, 1999 2671

Table 6. Inorganic sulfide concentrations in the wastewater treatment plants in nanomoles per litera

SampleParticulate

(.0.45 mm)Colloidal

(10 kDa–0.45 mm)Dissolved(,10 kDa)

Secondary clarifier inflow (Dundas)Secondary clarifier outflow (Dundas)Plant outflow (Dundas)Plant outflow (Burlington)

268 6 36250 6 47230 6 37

240

306/220290/184260/178

—

195 6 32170 6 25153 6 20

170

240/165204/144180/132

—

61 6 569 6 470 6 22

110

65/5475/6598/45

—

a Mean values 61 SD; high/low values; data compiled from three sampling trips to the Dundas plant (1997–1998).

Table 7. Inorganic sulfide concentrations downstream from the Dundas wastewater treatment plant in nanomoles per litera

LocationParticulate

(.0.45 mm)Colloidal

(10 kDa–0.45 mm)Dissolved(,10 kDa)

Plant outfall250 m downstream250 m downstream (bottom water)500 m downstream

145 6 45155 6 59163 6 46137 6 46

205/102258/102243/126213/90

130 6 41121 6 49130 6 45115 6 38

205/96213/84213/96184/82

72 6 1765 6 1476 6 1869 6 20

102/5490/49

112/6490/49

a Mean values 61 SD; high/low values; data compiled from five sampling trips to the Desjardin Canal (1997–1998).

affected by aggregation or sorption processes over the lengthof the canal (residence time for water in the canal is ø48 h).As with the ,10-kDa filter fraction in the treatment plant, thisuniformity in the silver concentration suggests that the silverin these fractions must be associated with strong complexants.Further, the concentrations of these complexants must remainrelatively constant in the different filter fractions. Otherwise,changes in the distribution of complexants would be mirroredin the distribution of silver among the different filter fractions.As will be shown in the next section, inorganic sulfide con-centrations do remain relatively constant over the length ofthe canal.

Sulfide and thiols in wastewater effluents and surfacewaters

Sulfide concentrations determined by the methylene bluecolorimetric method include the sulfide species: S22, HS2, andH2S (aq) (HS2 is the predominant factor that determines thepH of these waters). However, the addition of an acidic reagentsolution to the sample decreases the pH and causes a varietyof colloidal metal sulfides to decompose, allowing liberatedsulfide to react with the colorimetric reagents. Amorphous FeSreacts quantitatively under these conditions (FeS → Fe 1 SS),as does ZnS and CuS, although at different rates (CuS , FeS; ZnS). Polysulfides, Sn

22, also react to yield one equivalentS(II). Thiol compounds, on the other hand, do not decomposeunder these conditions and cannot be measured by this tech-nique [24].

Measurable sulfide concentrations were present in both thewastewater effluent samples as well as the fully oxic surfacewater samples (Tables 6 and 7). Particulate sulfide concentra-tions in the Dundas WTP decreased from the secondary clar-ifier inflow (270 nM) to the plant outflow (230 nM). Therewas a dramatic drop between the plant outflow and the outfallin the canal (145 nM). No change was observed between theoutfall and 500 m downstream. Sulfide concentrations in thecolloidal phase decreased from 200 nM at the secondary clar-ifier inflow to 150 nM at the plant outfall and decreased from130 nM at the outfall to 115 nM, 500 m downstream. In con-trast, dissolved sulfide (,10 kDa) concentrations remainedremarkably constant (60–70 nM). As observed in the distri-bution of silver among the different size fractions, the distri-

bution of inorganic sulfide also remained relatively constant(Tables 6 and 7). Simpson et al. [25] showed that several metalsulfides can be stable under oxic conditions for periods ofseveral hours (CdS, CuS, PbS, and ZnS), whereas FeS andMnS oxidize extremely rapidly under the same conditions. Theoxidation rate of free sulfide (HS2) was found to be slowerthan that of FeS and MnS, but faster when compared to theother metal sulfides examined. The sulfide measured in thisstudy is likely to be present in the form of these more stablemetal sulfides. Measurement of copper and zinc concentrationsin the plant and in the canal indicated that there is always anexcess of these metals over measured sulfide concentrations.

Thiol compounds were not measurable in any of the waste-water or surface water samples we examined. Reactions ofunfiltered samples with the equivalent of 0.1 mM NaBH4 toliberate thiols bound through disulfide linkages and with theequivalent of 0.1 mM EDTA to liberate thiols bound to metalsalso did not produce any thiol peaks in the chromatograms.This finding suggests that concentrations of simple thiols ofthe kinds used as standards in this study are at least 100 timeslower than inorganic sulfide concentrations.

Silver, sulfide, and thiols in pore waters

Silver concentrations were not detectable (,0.01 nM) inthe pore waters from sampling locations B and C. Silver wasalso not detectable in interfacial water samples collected fromlocations B and C. However, at location A, low concentrationsof silver in the colloidal fraction (but not in the ,10-kDa filterfraction) were measured in both the pore water and interfacialwater (Fig. 1). Sulfide concentrations were higher than mea-sured in the wastewater and surface water samples. Inorganicsulfide concentrations in the colloidal filter fraction rangedfrom 350 to 1,000 nM and from 50 to 150 nM in the ,10-kDa filter fraction. Two small peaks, corresponding to 3-mer-captopropionic acid and mercaptoethanol, were present in sev-eral chromatograms from location A. Mean concentrations forthe 10 to 20- and 20 to 30-cm depth intervals were 9.1 and15.5, and 7.0 and 10.4 nM, respectively. Reactions with NaBH4

and EDTA on unfiltered pore waters did not yield any increasein thiol concentrations. Thiol concentrations in freshwater porewaters are considerably lower than in marine and estuarine

2672 Environ. Toxicol. Chem. 18, 1999 N.W.H. Adams and J.R. Kramer

Fig. 1. Silver and inorganic sulfide pore water concentrations (col-loidal fraction) at Hamilton Harbour location A. Error bars denote61SE.

Fig. 2. Total sulfide versus total silver concentrations at Dundas waste-water treatment plant (n) and Desjardin Canal (●) and colloidal filterfractions at Hamilton Harbour location A (m).

pore waters, where thiol concentrations up to 10 mM havebeen reported [14].

Acid-volatile sulfide measurements on sediments from lo-cation A (intervals 0–10 and 40–50 cm) gave average valuesof 54.4 and 32.8 mol/g, respectively (unpublished work). Therelatively large amounts of acid-volatile sulfide in these sed-iments and low silver concentrations in the pore water suggestthat silver is effectively scavenged by the sulfides present inthe sediment.

Silver speciation

Our investigation of wastewater effluents, surface waters,and pore waters has demonstrated the ubiquity of inorganicsulfide at the nanomolar levels even in fully oxic waters. Underthese conditions, sulfide concentrations will always be in ex-cess of silver present at ultratrace concentrations (,1 nM).This is shown by a plot of the total silver concentration againstthat of inorganic sulfide for all the samples collected duringthis study (Fig. 2). Because silver forms a very strong sulfidecomplex compared to other metals (Hg21 . Ag1 . Cu1 .Pb21 . Zn21), it is reasonable to assume that even if sulfide,determined by the methylene blue method, is in the form ofmetal sulfides, silver should outcompete other metals for theavailable sulfide. The fact that sulfide was measurable in allthree size fractions we examined suggests that silver in thesedifferent size fractions is likely to exist in the form of silver

sulfide complexes. However, colloidal metal sulfides are prob-ably entrapped in complex networks of organic material, andsilver is probably a coprecipitate or a surface complex in thesecomplex aggregates. This is evidenced by the correlation be-tween dissolved organic carbon (Table 2), particulate sulfide(Tables 6 and 7), and particulate silver (Tables 4 and 5). Shaferet al. [6] also observed a correlation between silver and dis-solved organic carbon in the effluents of several WisconsinWTPs. To explain this correlation, the authors suggested thatsilver was complexed by dissolved organic carbon. However,sulfide to silver ratios of 100 to 1,000 at the Burlington andDundas WTPs indicate that sulfide is responsible for com-plexing silver instead.

Our investigation of thiols in wastewater, surface water, andpore water suggests that this form of sulfur(2II) is probablynot an important species for complexation of metals in fresh-waters. Inorganic sulfide concentrations are far too much inexcess for thiol compounds to be of importance. However, wehave not yet investigated the abundance of sulfhydryl sites inhigher-molecular-weight substances (i.e., humics and fulvics).It should also be noted that thiol compounds, namely, gluta-thione and metal chelating polypeptides (metallothioneins andphytochelatins) probably play an important role in regulatingsilver and other metal concentrations in living organisms [26].Although silver sulfide complexes possess large log K values,silver is quite labile between sulfur(2II) species [27,28]. It ispossible, therefore, to envision transfer of silver to biota viaan exchange between an inorganic sulfide species in the aque-ous phase and thiolic sites in the organism.

Acknowledgement—We would like to thank the staff at the CanadaCentre for Inland Waters for their help in collecting the sediment coresamples. We thank Kodak Canada and the Natural Sciences and En-gineering Research Council of Canada for providing the funding forthis research project. Nicholas Adams also acknowledges receipt ofa Natural Sciences and Engineering Research Council of Canada andOntario Graduate Scholarship.

REFERENCES

1. Smith GJ, Flegal AR. 1993. Silver in San Francisco Bay estuarinewaters. Estuaries 16:547–558.

2. Sanudo-Wilhelmy SA, Flegal AR. 1992. Anthropogenic silver inthe southern California Bight: A new tracer of sewage in coastalwaters. Environ Sci Technol 26:2147–2151.

3. Shafer MM, Armstrong DE, Overdier JT, Walker MT. 1994. Par-titioning and fate of silver in background streams and effluent-receiving streams. Proceedings, 2nd International Argentum Con-ference, Madison, WI, USA, September 11–14, pp 99–105.

4. Wen L, Santschi PH, Gill GA, Paternostro CL, Lehman RD. 1997.Colloidal and particulate silver in river and estuarine waters ofTexas. Geochim Cosmochim Acta 52:1505–1519.

5. Rozan TF, Hunter KS, Benoit G. 1995. Silver in fresh water:Sources, transport and fate in Connecticut rivers. Proceedings,3rd International Argentum Conference, Washington, DC, USA,August 6–9, pp 181–184.

6. Shafer MM, Overdier JT, Armstrong DE. 1998. Removal, parti-tioning, and fate of silver and other metals in wastewater treatmentplants and effluent-receiving streams. Environ Toxicol Chem 17:630–641.

7. Kramer JR, et al. 1997. Chemical speciation and metal toxicityin surface freshwaters. In Bergman HL, Dorward-King EJ, eds,Reassessment of Metals Criteria for Aquatic Life Protection.SETAC Technical Publication. Society of Environmental Toxi-cology and Chemistry, Pensacola, FL, USA, pp 57–70.

8. Connell DB, Sanders JG, Riedel GF, Abbe GR. 1991. Pathwaysof silver uptake and trophic transfer in estuarine organisms. En-viron Sci Technol 25:921–924.

9. Schwarzenbach G, Widmer M. 1966. Die loslichkeit von me-tallsulfiden II. Silbersulfid[1]. Helv Chim 49:111–123.

Silver speciation in water and effluents Environ. Toxicol. Chem. 18, 1999 2673

10. Adams NWH, Kramer JR. 1999. Potentiometric determination ofsilver thiolate formation constants. Aquat Geochem 5:1–11.

11. Smith RM, Martell AE. 1997. NIST critically selected stabilityconstants of metal constants databases. NIST Database 46, 3.0,Standard Reference Program. National Institute of Standards andTechnology, Gaitherburg, MD, USA.

12. Boulegue J, Ciabrini JP, Fouillac C, Michard G, Ouzounian G.1979. Field titrations of dissolved sulfur species in anoxic en-vironments—Geochemistry of Puzzichello waters (Corsica,France). Chem Geol 25:19–29.

13. Luther GW, Giblin AE, Varsolona R. 1985. Polarographic analysisof sulfur species in marine pore waters. Limnol Oceanogr 30:727–736.

14. MacCrehan W, Shea D. 1994. Temporal relationship of thiols toinorganic sulfur compounds in anoxic chesapeake bay sedimentpore water. In Vairavamurthy A, Schoonen MAA, eds, Geochem-ical Transformations of Sedimentary Sulfur, American ChemicalSociety, Washington, DC, pp 294–310.

15. Cutter GA, Oatts TJ. 1987. Determination of dissolved sulfideand sedimentary sulfur speciation using gas chromatography–photoionization detection. Anal Chem 59:717–721.

16. Luther GW, Tsamakis E. 1989. Concentration and form of dis-solved sulfide in the oxic water column of the ocean. Mar Chem27:165–177.

17. Wang F, Tessier A, Buffle J. 1998. Voltammetric determinationof elemental sulfur in pore waters. Limnol Oceanogr 43:1353–1361.

18. Kramer JR. 1995. Aqueous silver in the environment. Proceed-ings, 3rd International Argentum Conference, Washington, DC,August 6–9, pp 19–30.

19. Hurley JP, Shafer MM, Cowell SE, Overdier JT, Hughes PE,

Armstrong DE. 1996. Trace metal assessment of Lake Michigantributaries using low-level techniques. Environ Sci Technol 30:2093–2098.

20. Cline JD. 1969. Spectrophotometric determination of hydrogensulfide in natural waters. Limnol Oceanogr 14:454–458.

21. Mopper K, Delmas D. 1984. Trace determination of biologicalthiols by liquid chromatography and precolumn fluorometric la-beling with o-Phthalaldehyde. Anal Chem 56:2557–2560.

22. Adams NWH, Kramer JR. 1999. Determination of silver speci-ation in wastewater and receiving waters by competitive ligandequilibration/solvent extraction. Environ Toxicol Chem 18:2674–2680.

23. Pham MK, Garnier J. 1998. Distribution of trace elements as-sociated with dissolved compounds (,0.45 mm–1 nm) in fresh-water using coupled (frontal cascade) ultrafiltration and chro-matographic separations. Environ Sci Technol 32:440–449.

24. Kramer JR, Wu F. 1998. Use of AVS in interpreting Ag fate andtoxicity. Proceedings, 5th International Argentum Conference,Hamilton, ON, Canada, September 28–October 1, 1997, pp 109–113.

25. Simpson SL, Apte SC, Batley GE. 1998. Effects of short-termresuspension events on trace metal speciation in polluted anoxicsediments. Environ Sci Technol 32:620–625.

26. Wang W-X, Fisher NS. 1997. Modeling metal bioavailability formarine mussels. Rev Environ Contam Toxicol 151:39–65.

27. Adams NWH, Kramer JR. 1998. Reactivity of Ag1 ion with thiolligands in the presence of iron sulfide. Environ Toxicol Chem17:625–629.

28. Bell RA, Kramer JR. 1999. Structural chemistry and geochem-istry of silver–sulfur compounds: Critical review. Environ Tox-icol Chem 18:9–22.