PERFORMANCE OF PASSIVE SAMPLERS FOR MONITORING ESTUARINE WATERCOLUMN CONCENTRATIONS: 2. EMERGING CONTAMINANTS

MONIQUE M. PERRON,*y ROBERT M. BURGESS,z ERIC M. SUUBERG,x MARK G. CANTWELL,z and KELLY G. PENNELLkyNational Research Council, US Environmental Protection Agency, ORD/NHEERL, Narragansett, Rhode Island

zUS Environmental Protection Agency, ORD/NHEERL, Narragansett, Rhode IslandxSchool of Engineering, Brown University, Providence, Rhode Island, USA

kCivil and Environmental Engineering Department, University of Massachusetts Dartmouth, North Dartmouth, Massachusetts, USA

(Submitted 17 January 2013; Returned for Revision 8 February 2013; Accepted 8 April 2013)

Abstract: Measuring dissolved concentrations of emerging contaminants, such as polybrominated diphenyl ethers (PBDEs) and triclosan,can be challenging due to their physicochemical properties resulting in low aqueous solubilities and association with particles. Passivesampling methods have been applied to assess dissolved concentrations in water and sediments primarily for legacy contaminants.Although the technology is applicable to some emerging contaminants, the use of passive samplers with emerging contaminants is limited.In the present study, the performance of 3 common passive samplers was evaluated for sampling PBDEs and triclosan. Passive samplingpolymers included low-density polyethylene (PE) and polyoxymethylene (POM) sheets, and polydimethylsiloxane (PDMS)-coated solid-phase microextraction (SPME) fibers. Dissolved concentrations were calculated using measured sampler concentrations and laboratory-derived partition coefficients. Dissolved tri-, tetra-, and pentabrominated PBDE congeners were detected at several of the study sites atvery low pg/L concentrations using PE and POM. Calculated dissolved water concentrations of triclosan ranged from 1.7 ng/L to 18 ng/Lfor POM and 8.8 ng/L to 13 ng/L for PE using performance reference compound equilibrium adjustments. Concentrations in SPMEwerenot reported due to lack of detectable chemical in the PDMS polymer deployed. Although both PE and POM were found to effectivelyaccumulate emerging contaminants from thewater column, further research is needed to determine their utility as passive sampling devicesfor emerging contaminants. Environ Toxicol Chem 2013;32:2190–2196. # 2013 SETAC

Keywords: Passive sampling Emerging contaminants Polyethylene Polyoxymethylene Solid-phase microextraction fibers

INTRODUCTION

The occurrence of new or emerging contaminants in theaquatic environment has been a growing concern [1]. Theseemerging contaminants are not regularly monitored in theenvironment, but many are biologically active and are suspectedof causing adverse ecological or human effects, or both (http://toxics.usgs.gov/regional/emc). Emerging contaminants encom-pass a wide range of compounds and may be classified aspharmaceuticals, pesticides or insecticides, personal careproducts, or hormones [2]. More specifically, these includepolybrominated diphenyl ethers (PBDEs), which have been usedfor years as flame retardants in clothing textiles and furniture [3],and triclosan, an antimicrobial compound used in variousconsumer products [4].

Polybrominated diphenyl ethers are typically producedcommercially as flame retardants in 3 mixtures. Pentabromo-diphenyl ether (penta-BDE) is a product consisting primarily ofPBDE congeners 47, 99, 100, 153, and 154. Octabromodiphenylether (octa-BDE) is a mixture of hexa- through nona-brominatedcongeners. The deca-product is composed almost entirely ofPBDE congener 209 [5]. Due to their widespread global use,PBDEs have been detected in numerous media including air,water, sediments, aquatic organisms, and humans [3,6,7]. Theypersist in the environment, have the capability to bioaccumulateand biomagnify in the food chain, and have been found to cause a

variety of adverse ecotoxicological effects including enzymedisruption, hepatic disease, thyroid modifications, and acutetoxicity [6,7].

The antimicrobial agent triclosan is used in householdproducts, such as shampoo, toothpaste, soap, and deodorant. Inaddition, triclosan is a component of plastic products, includingkitchen utensils and children’s toys [4]. Given its extensive use,triclosan has been detected in wastewater, surface water, sewagesludge, and sediments [1,8–14]. It has also been found to causeacute and chronic effects in aquatic organisms [15,16].

Measuring dissolved concentrations of emerging contami-nants, such as PBDEs and triclosan, can be challenging due totheir physicochemical properties resulting in low aqueoussolubilities and association with particles. For example, theaqueous solubility of PBDE congeners range from approxi-mately 2 � 10�3 mg/L to 0.1 mg/L [17,18], with log KOW

values of 5.74 to almost 10 [18–20]. Triclosan is more soluble,with a solubility of 12 mg/L and log KOW of 4.8 [21]. As aconsequence of their physicochemical properties, both PBDEsand triclosan would be expected to occur primarily in sedimentswithin estuarine systems. In comparison with other contami-nants of concern (e.g., polychlorinated biphenyls [PCBs],polycyclic aromatic hydrocarbons [PAHs]), relatively fewinvestigations have measured these emerging contaminantsdissolved in the estuarine water column. The dissolved concen-tration is critical because it is considered to be the best measureof bioavailability in aquatic systems [22] and is necessary toperform accurate risk assessments. Recently, passive samplingmethods have been applied for measuring dissolved concen-trations in water and sediments using semipermeable membranedevices, polyethylene (PE), polydimethylsiloxane (PDMS), and

All Supplemental Data may be found in the online version of this article.* Address correspondence to perron.monique@epa.gov.Published online 17 April 2013 in Wiley Online Library

(wileyonlinelibrary.com).DOI: 10.1002/etc.2248

Environmental Toxicology and Chemistry, Vol. 32, No. 10, pp. 2190–2196, 2013# 2013 SETAC

Printed in the USA

2190

polyoxymethylene (POM) [23–26]. To date, the emphasis ofpassive sampler development and use has been on monitoringlegacy contaminants (e.g., PAHs, PCBs, chlorinated pesticides).The technology should be applicable to some emergingcontaminants, although data for this class of chemicals arecurrently very limited [27–29].

The results presented in the present study are components of a2-part assessment of the use of passive samplers in the marineenvironment. For the present study, the performance of 3common passive samplers (PE, PDMS, and POM) wasevaluated for sampling PBDEs (congeners 17, 47, 71, 99,100, 183, and 209) and triclosan in the natural waters of severalstations in the temperate estuary Narragansett Bay, RhodeIsland, USA. Narragansett Bay has a history of industrialactivity, which has contaminated the sediments in portions of thebay with legacy contaminants [30,31]. Results of the perfor-mance of the passive samplers with legacy contaminants arediscussed in Perron et al. [32]. Narragansett Bay is surroundedby suburban and urban areas, and contamination by emergingcontaminants is highly likely. Dissolved concentrations of eachemerging contaminant were determined using the partitioningcalculations described in Perron et al. [32] based on relationshipsdeveloped by others (e.g., Huckins et al. [23]). This investigationhad 2 specific objectives: 1) to compare passive samplereffectiveness for water column monitoring of selected emergingcontaminants in the marine environment, and 2) to derivepartition coefficients for triclosan and PBDEs.

MATERIALS AND METHODS

Materials

Chemical stock solutions of triclosan and methyl triclosanwere prepared by Ultra Scientific in acetone. Neat PBDEs werepurchased from Cambridge Isotope Laboratories. FluorinatedPBDEs (FBDEs) in methanol were purchased from Sigma-Aldrich. Labeled methyl triclosan in nonane was purchased fromWellington Laboratories. Supplemental Data, Table S1, pro-vides a list of target chemicals, performance referencecompounds (PRCs), and internal standards.

Low-density PE (25-mm thickness; Covalence Plastics)and POM (75 mm; CS Hyde) were cut into strips of 15 cm �40 cm and 6 cm � 40 cm, respectively. Strips of PE andPOM were precleaned by soaking in acetone for 24 h andthen in dichloromethane (DCM) for 24 h. Solid-phasemicroextraction fibers (SPME: 200-mm inner silica corewith 10-mm outer PDMS; Fiberguide Industries) were cutto 2.5 cm in length and precleaned as described above for PEand POM.

Field site locations and deployment

Strips of PE and POM were soaked in a PRC solution (80:20methanol:water with FBDEs and 13C-methyl triclosan) for atleast 28 d on an orbital shaking table. Each PRC jar contained4 sampler strips and 900 mL aqueous PRC solution. Aftersoaking, strips of PE and POM were removed from the PRCsolution, 1 strip of each sampler was taken from each PRCsolution jar for measuring predeployment PRC concentrations,and the remaining strips were attached to stainless steel wire(diameter ¼ 0.032 in; Malin) inside galvanized extendedminnow traps (diameter ¼ 22 cm, length ¼ 30 in; TackleFactory). Minnow traps were used to protect samplers frombiotic and abiotic threats. Fifteen precut SPME fibers wereplaced inside a fine copper mesh (TWP Inc) hand-made

envelope, and the envelope was attached to the inside of thetrap by stainless steel wire. Inside each trap, 3 strips of PE andPOMwere attached to maximize their surface area. Three SPMEenvelopes were also attached to the inside of each trap. Passivesampler traps were deployed approximately 1 m above thesediment surface for 21 d at 6 site locations in Narragansett Bay(Supplemental Data, Table S2). Four deployments (GreenwichBay, Bristol Harbor, Mount Hope Bay, and Newport Harbor)were from water-based US Coast Guard navigational buoys.Two deployments (US Environmental Protection Agency[USEPA] and Providence River) were from docks at sitelocations.

Partition coefficient derivation

Partition coefficients were obtained by plating chemicalsolutions onto the sides of a 250-mL amber jar before adding250 mL of Milli-Q water and approximately 5 mg of a sampler.Plating involved discharging chemical solutions prepared in anorganic solvent onto the jar interior wall while slowly rotatingthe jar and allowing the solvent to evaporate, leaving thechemical residue. All operations with the stock solutions wereperformed under low-level laboratory lighting away from directsunlight to avoid any photodegradation, and sodium azide wasadded to each jar to prevent contaminant microbial degradation.Jars were then placed on an orbital shaking table for 56 d, whichwas found to be sufficient for equilibration in prior kineticsstudies (data not shown). Water and samplers were extracted asdescribed below.

Extraction

Passive samplers were weighed and extracted sequentiallywith acetone and DCM for 24 h each. Supplemental Data,Table S1, indicates the internal standards used for eachcontaminant type. Samplers retrieved from the field were wipedclean of water and epiphytes before extraction. Acetone andDCMextracts were combined, solvent exchanged to hexane, andvolume reduced to 1 mL under a stream of nitrogen gas. Watersamples from partition coefficient derivation studies wereextracted twice with pentane. Extracts were combined, treatedwith sodium sulfate, solvent exchanged to hexane, and volumereduced to 1 mL. For triclosan analysis, extract subsamples(100 mL) were evaporated to dryness under a nitrogen gasstream and derivatized for 1 h with 50 mL of N-tert-butyldimethylsilyl-N-methyltrifluoroacetamide. Extract vol-umes were brought up to 100 mL using ethyl acetate andanalyzed within 24 h of derivatization [16].

Instrumental analysis

Analyses of triclosan andmethyl triclosan were performed onan Agilent 7890 gas chromatograph equipped with a 5975 massselective detector (Agilent Technologies) operated in selectiveion monitoring mode. Analytes and internal standards wereseparated using an Agilent DB-5MS capillary column (30 mlength; 250 mm diameter; 0.25 mm thickness) and quantifiedwith a 5- or 6-point calibration curve. Analyses of PBDEs andFBDEs were performed on an Agilent 6890 gas chromatographequipped with a 5973 mass selective detector (AgilentTechnologies) operated in selective ion monitoring modeutilizing negative chemical ionization. Analytes and internalstandards were separated using an Agilent DB-5 capillarycolumn (15 m length; 250 mm diameter; 0.1 mm thickness) andquantified with a 5-point calibration curve.

Passive sampler performance for emerging contaminants Environ Toxicol Chem 32, 2013 2191

Calculation of dissolved concentrations

Dissolved water concentrations (CD) were calculated usingthe following equation

CD ¼ Csampler

ð1� e�ketÞ � Ksampler

where Ksampler is the chemical-specific passive sampler-waterpartition coefficient adjusted for the presence of salt usingSetschenow constants [33], Csampler is the measured samplerconcentrations, and ke is the mass transfer coefficients obtainedfrom PRC equilibration [32]. The FBDEs and labeled methyltriclosan were used as PRCs for PBDEs and triclosan,respectively (Supplemental Data, Table S1). When PRCcorrections were not used (see Discussion), the equation abovewas simplified by having the ð1� e�ketÞ term removed from thedenominator and CD was calculated as the ratio of the measuredsampler concentration (Csampler) and chemical-specific passivesampler-water partition coefficient (Ksampler) adjusted for thepresence of salt using Setschenow constants.

RESULTS AND DISCUSSION

Partition coefficients

Partition coefficients for PE, POM, and SPME were obtainedfor PBDEs, triclosan, and methyl triclosan (Table 1). ForPBDEs,Ksampler values for each sampler were plotted against logKOW values, and the findings are discussed below. Valuesobtained for triclosan and methyl triclosan are discussedindividually.

For PE, a good correlation between chemical-specific PEpartition coefficients (KPE) and KOW was observed for most ofthe PBDE congeners with the exception of 183 and 209(r ¼ 0.997 with removal of these congeners; Figure 1). These 2congeners have log KOW values over 8 and would be susceptibleto energy differences needed for cavity formation in octanolversus the passive sampler polymer. That is, the Gibbs freeenergy required for cavity formation in the passive sampler

polymer becomes much higher for larger chemicals, while theamount of energy needed for cavity formation in octanol changeslittle for the same chemicals [34]. Bao et al. [20] observed asimilar linear relationship for most of the 23 PBDE congenersthey studied (Table 2), such that log KPE values increased withincreasing log KOW until a plateau at log KOW of approximately8.3 was reached followed by a decrease in log KPE withincreasing log KOW. Sacks and Lohmann [29] obtained log KPE

values for individual PBDE congeners (Table 2), but due to thepresence of a third nonane phase in their partitioning experi-ments, apparent partition coefficients were much lower thanthose derived in the present study and in that of Bao et al. [20].After accounting for partitioning into this third nonane phase,adjusted values from Sacks and Lohmann [29] were much higherthan the range observed in the present study (Table 1) and Baoet al. [20].

A linear relationship was observed again with SPMEfor PBDEs with a plateau at log KOW of approximately 8.3(Figure 1a). Good correlation was observed for congeners belowthis log KOW (r ¼ 0.992; Figure 1b). ter Laak et al. [35] reportedlog SPME-water partition coefficient (KSPME) values for 3 of thesame congeners measured in the present study. All 3 values werefound to be approximately a magnitude lower than the valuesobtained in the present study. A similar linear relationship wasagain observed using POM with the same plateau as PE andSPME (Figure 1a), and a good correlation was found below theplateau (r ¼ 0.996; Figure 1b). To our knowledge, this is the firststudy to report POM-water partition coefficient (KPOM) valuesfor PBDEs; therefore, we were unable to compare our valueswith those obtained in other studies.

When partition coefficients are compared for each congeneracross passive sampler polymers (e.g.,KPE vsKPOM vsKSPME forPBDE congener 100), no significant differences (p ¼ 0.05) areobserved except for PBDE congener 17 (Table 1). This findingindicates that for most of the PBDE congeners measured,congener-specific partition coefficients were approximately thesame regardless of the type of passive sampler polymer utilized.Because these partition coefficients were derived within the samelaboratory, interlaboratory variability was removed, and this

Table 1. Log octanol–water (log KOW) and log passive sampler–water partition coefficient values (log KPS; L/kg) derived in the laboratorya

Chemical

Octanol–water partitioncoefficient(log KOW)

PE–water partitioncoefficient(log KPE)

POM–water partitioncoefficient(log KPOM)

SPME–water partitioncoefficient(log KSPME) ANOVA p value

Polybrominated diphenyl ethers (PBDEs)PBDE 17 5.74b 5.17 � 0.04 5.42 � 0.03 5.92 � 0.15 0.0002PBDE 47 6.81b 6.20 � 0.20 6.29 � 0.13 6.61 � 0.42 0.26PBDE 71 6.54b 5.98 � 0.15 6.03 � 0.10 6.52 � 0.37 0.059PBDE 99 7.32b 6.74 � 0.54 6.78 � 0.27 7.16 � 0.35 0.42PBDE 100 7.24b 6.55 � 0.49 6.60 � 0.23 7.01 � 0.35 0.33PBDE 183 8.27b 6.22 � 1.16 6.54 � 0.62 7.63 � 0.13 0.14PBDE 209 9.97c 5.11 � 1.04 5.76 � 0.19 6.79 � 0.42 0.054

TriclosanTriclosan 4.80d 2.44 � 0.40 3.79 � 0.26 4.06 � 0.13

MethyltriclosanMethyltriclosan 5.20e 4.41 � 0.10 5.27 � 0.18 4.90 � 0.15

a Values were derived for polyethylene (PE), polyoxymethylene (POM), and solid-phase microextraction (SPME) fibers with polydimethylsiloxane (PDMS)coating. Mean and 1 standard deviation are presented. Statistical p values are presented for the comparison of passive sampler partition coefficients for eachcongener (i.e., log KPE vs log KPOM vs log KSPME).b Braekevelt et al. [19].c Wania and Dugani [18].d Reiss et al. [21].e Dann and Hontela [15].ANOVA ¼ analysis of variance.

2192 Environ Toxicol Chem 32, 2013 M.M. Perron et al.

comparison can be performed. The lack of statistical significancebetween passive sampler partition coefficients suggests thatdespite substantial differences in their chemical structures andcompositions, the partitioning of nonionic organic chemicalsfrom aqueous solution occurs with the same degree of affinitybetween polymers. For example, relative to chemical composi-tion, POM contains oxygen along with carbon and hydrogen, theonly atoms in PE, whereas PDMS also contains silicon andoxygen atoms.

Although triclosan and methyl triclosan values were notplotted, it should be noted that only one other investigation has

published partition coefficients for these chemicals. Sacks andLohmann [28] found log KPE values of 3.34 and 4.53 fortriclosan and methyl triclosan, respectively. The value obtainedin the present study for methyl triclosan was only slightly lowerat 4.41, but the log KPE for triclosan was much lower, at 2.44.This difference may be because of the polarity of the chemical orthe differences in chemical analysis used in the 2 studies, or bothfactors. In the present study, derivatization was used to analyzetriclosan, whereas Sacks and Lohmann [28] did not derivatize.Derivatization is known to improve instrument sensitivity whentriclosan is measured [36–38]. Of the chemicals investigated inthe present study, triclosan is the only one that has an ionic form(i.e., pKa ¼ 8.2). It is possible that this characteristic may resultin greater variability and differences between studies. As far aswe know, this is the first study to report KPOM and KSPME valuesfor triclosan and methyl triclosan; therefore, we are unable tocompare our values with those obtained in other studies.

Dissolved concentrations of PBDEs

Because PBDEs are used in a wide range of commercialproducts, the waste resulting from these materials is believed tobe the main source of PBDEs in the environment [6]. Thisincludes discharge from wastewater treatment plants, wasteincineration, and leaching from landfill sites [6]. In NarragansettBay, all of these may be considered potential sources of PBDEs.In addition, in the aquatic environment, it is known that thesetypes of organic contaminants, which over time have accumu-lated in the sediments, can be released into the water columnunder the right conditions (e.g., storms, shipping-relatedresuspension), making the sediments a nonpoint source.

Dissolved PBDE concentrations calculated using themeasured PE sampler concentrations (Supplemental Data,Figure S1) with PRC adjustments and the chemical-specificpartition coefficients described above (Supplemental Data,Table S1) yielded very low levels (pg/L; Figure 2, SupplementalData, Table S3). Results are only shown for PE and POMbecause contaminants were not detectable in the SPMEsamplers. The lack of detectable chemicals in the SPMEreplicates suggests that not enough PDMS polymer wasdeployed. As a result of a tropical storm-damaged cage, POMreplicates were lost at the Bristol Harbor site, and no data areshown. Tri-, tetra-, and pentabrominated congeners (17, 47, 71,99, and 100) were detected at several of the study sites, whereasthe hepta- and decabrominated congeners (183 and 209) werenot detected at any of them. A study by Sacks and Lohmann [29]in the same estuary measured dissolved PBDE concentrationsusing PE (25-mm thickness) and found similar results. Themajority of the congeners detected in their study were tetra- andpentabrominated congeners. Dissolved concentrations of theirmost abundant congener (PBDE 47) ranged from 0.18 pg/L to2.3 pg/L. In the present study, dissolved concentrations of thesame congener based on PE were measured at slightly higherconcentrations (1.94–11.9 pg/L; Supplemental Data, Table S3).

Although water concentrations could be calculated for 5PBDE congeners using PRC-adjusted PE data, concentrationscould not be obtained in the same manner using POM because ofpoor PRC behavior, indicating that equilibration was less than10%. Three fluorinated PBDEs (FBDE 28, 100, and 208) wereused as PRCs to account for nonequilibration of the samplersafter retrieval from the field. With POM, issues arose with thesePRCs, such that very low or no equilibration was observed for allof the deployments. As a result, dissolved concentrations couldnot be calculated using the measured PBDE concentrations inthe POM and PRC adjustments. The cause of this abnormal

4

5

6

7

8

4 5 6 7 8 9 10 11

Log

Ksa

mpl

er(L

/kg)

Log Kow (L/kg)

PEy = 0.9557 (±0.04) x - 0.304 (±0.27)

r² = 0.99

POMy = 0.8327 (±0.04) x + 0.6202 (±0.28)

r² = 0.99

SPMEy = 0.7534 (±0.06) x + 1.5734 (±0.38)

r² = 0.98

4

5

6

7

8

5 6 7 8

Log

Ksa

mpl

er(L

/kg)

Log Kow (L/kg)

a

b

Figure 1. (a) Polybrominated diphenyl ether (PBDE) sampler-water partitioncoefficients (log Ksampler; L/kg) for polyethylene (PE; *), polyoxymethy-lene (POM; &), and solid-phase microextraction fibers (SPME; ~) plottedagainst octanol–water partition coefficients (log KOW; L/kg) and (b) thelinear portions of the log Ksampler and log KOW relationships.

Table 2. Experimentally derived polyethylene–water partition coefficients(log KPE; L/kg) measured for polybrominated diphenyl ethers (PBDEs)

obtained from the literaturea

PBDEcongener

Baoet al. [20]

Sacks andLohmann [29](apparent)b

Sacks andLohmann [29](accounting forthird phase)b

17 5.4347 6.25 5.01 7.1871 6.0299 6.88 5.02 7.79100 6.82 5.02 8.05183 7.39 5.07 9.11209 5.61

a For values from the present study, see Table 1.b Converted to L/kg using rPE ¼ 0.92 g/mL.

Passive sampler performance for emerging contaminants Environ Toxicol Chem 32, 2013 2193

behavior of FBDEs with POM is unknown, but it could beattributed to the structure of the POM. The POM contains arepeating oxygen-containing group (i.e., ether), which mayresult in unexpected intermolecular interactions with compoundslike FBDEs. For example, in general, nonionic organic chemicaland polymer intermolecular interactions are dominated by vander Waals dispersive forces [33]. This is especially the case forPE, which is composed of only carbon and hydrogen. In contrast,the ether groups in the POM may result in electron donor–acceptor interactions between the oxygens and the FBDEs thatare stronger than the dispersive forces [33]. Consequently, therate of PRC release from the polymer would not be the same asthe uptake of measured contaminants by the polymer. Perronet al. [32] reported similar behavior by PRCs with PCBs usingPOM. Recognizing that the PRCs were not functioning withthe PBDEs, non-PRC–adjusted POM concentrations, based onthe assumption of equilibrium conditions, yielded dissolvedconcentrations ranging from 89 pg/L to 105 pg/L, which were1.7 times to 3.6 times higher than those observed with PE.

Dissolved concentrations of triclosan

Many personal care products and consumer goods that aremost often disposed of in household wastewater drains containtriclosan. As a result, triclosan primarily enters the environmentthrough municipal wastewater treatment effluent [9]. Removal

efficiencies vary depending on the type of removal processesused by the wastewater treatment plant [39]. There are 35wastewater treatment plants that discharge into the NarragansettBay watershed (http://www.savebay.org/page.aspx?pid¼335).For the present study, 2 stations were located near wastewatertreatment plants. The state’s largest wastewater treatmentfacility, Fields Point, provides preliminary, primary, andsecondary treatment to Providence, Rhode Island, USA andthe surrounding area. This facility’s effluent discharge is locatednear the Providence River site. The East Greenwich WastewaterTreatment Plant is located near the Greenwich Bay site. Thissmaller facility provides up to tertiary treatment and dischargesdirectly into Greenwich Bay.

Due to cost restraints, the PRC for triclosan (13C-labeledmethyl triclosan) was only used at 3 stations (USEPA,Greenwich Bay, and Providence River); therefore triclosanconcentrations are only shown for these stations. As noted withPBDEs, contaminants were not detectable in SPME, and POMreplicates were lost at the Bristol Harbor site; therefore no dataare shown. Calculated triclosan water concentrations usingmeasured PE concentrations (Supplemental Data, Figure S2) andPRC adjustments were approximately the same at all 3 sites (8.8–13 ng/L). Conversely, calculated dissolved water concentrationsof triclosan ranged from 1.7 ng/L to 18 ng/L for POM adjustedwith PRC data (Figure 3, Supplemental Data, Table S3). Non-PRC–adjusted POM resulted in lower triclosan concentrations atall 3 sites (�70% decrease from PRC-adjusted POM), but still

Figure 2. Calculated mean dissolved concentrations (pg/L) of totalpolybrominated diphenyl ethers (PBDEs) in polyethylene (PE) andpolyoxymethylene (POM) at the 6 study sites in Narragansett Bay, RhodeIsland, USA. Concentrations derived from PE adjusted with performancereference compounds (PRCs) (black bars), and non-PRC–adjusted POM(hatched bars) are presented. ND ¼ no data for the Bristol Harbor site wherePOM replicates were lost due to a tropical storm. Error bars represent 1standard deviation (SD).

Figure 3. Calculated mean dissolved concentrations (ng/L) of triclosan inpolyethylene (PE) and polyoxymethylene (POM) at 3 of the study sites inNarragansett Bay, Rhode Island, USA. Concentrations derived from PEadjusted with performance reference compounds (PRCs; black bars), PRC-adjusted POM (white bars), and non-PRC–adjusted POM (hatched bars) arepresented. Error bars represent 1 standard deviation (SD).

2194 Environ Toxicol Chem 32, 2013 M.M. Perron et al.

displayed a range from 0.5 ng/L to 5.6 ng/L. Water concen-trations of triclosan would be anticipated to vary between sitesgiven that the Providence River and Greenwich Bay sites arenear wastewater treatment plant outfalls, which are directsources of triclosan into Narragansett Bay [40]. In contrast,expected triclosan levels would be lower at the USEPA site,which is not located near a wastewater treatment plant and iscloser to open ocean conditions. The differences between PE andPOM for this compound are likely due again to the polymerstructure of the samplers. As mentioned above, POM contains arepeating polar ether group (and the potential for the associatedelectron donor–acceptor interactions) that is not present in theexclusively carbon and hydrogen structure of PE, which mayresult in POM having a higher affinity for polar chemicals suchas triclosan [27]. Consequently, it may be a more analyticallysensitive sampler for this compound because of its preferentialaccumulation in the polymer. This is also displayed by the higherpartition coefficient for triclosan with POM (log KPOM ¼ 3.79)compared with PE (log KPE ¼ 2.44). Whether POM is acting asan equilibrium sampler for relatively polar organic contaminantscould be arguable. Given these interactions, it might be behavingas an integrative sampler [23].

Sacks and Lohmann [28] conducted a study in NarragansettBay at similar study sites and were unable to detect triclosan inthe water using PE. We would expect that triclosan would bedetected by the passive sampler. Dissolved concentrations oftriclosan have been measured in the water column of the samestudy area using large-volume water extraction methods [40] atlevels ranging from 0.5 ng/L to 7.4 ng/L, which are comparableto the dissolved concentrations calculated in the present study.As discussed earlier, derivatization was not utilized in triclosananalyses by Sacks and Lohmann [28], which can result inreduced analytical sensitivity. In addition, as observed in thepresent study, PEmay not be the best suited sampler for triclosandue to its relatively low affinity for the polymer.

CONCLUSIONS

Water column deployments demonstrated the strengths andweaknesses of the passive samplers evaluated. First, relative toselection of the type of passive sampler for water columndeployments, both PE and POM were effective at accumulatingemerging contaminants, but the use of SPME in the fiberconfiguration was problematic due to their small size, theirfragility, and the large masses needed to achieve acceptableanalytical sensitivity. Use of SPME, and more specificallyPDMS, in configurations with greater surface areas and masses(e.g., sheets, rings) is worth investigating for water columndeployments. Second, to ensure confidence in the dissolvedconcentrations, it is critical that the contaminants and samplersachieve equilibrium in the water column or that equilibriumconditions can be estimated. Here, equilibrium conditions wereestimated based on the use of PRCs; however, we observed thatissues can arise when PRCs are used with certain polymers. Forexample, although POM accumulated contaminants, usingPRCs with POM was problematic with the PBDEs. Applicationof PRCs with POM is not as commonly practiced as with PE, andtypically equilibrium is assumed for calculating dissolvedconcentrations with POM. Given the importance of establishingequilibrium conditions, the performance of water column studiescomparing dissolved concentrations from passive samplers withmeasured dissolved concentrations from large-volume extrac-tions would be beneficial. Finally, analytical sensitivity to andpolymer affinity for the target chemicals can influence selection

of the type of passive sampler. For example, POM veryeffectively accumulated triclosan as compared with PE, whichimproved the analytical sensitivity. Understanding the strengthsandweaknesses of passive samplers for specific applications willassist in designing scientifically robust and effective studieswhile also advancing the acceptance of passive samplers asenvironmental monitoring tools.

SUPPLEMENTAL DATA

Figures S1–S2. (2.7 MB DOC).

Acknowledgment—The authors thank D. Katz, J. Sullivan, P. Pelletier, andD. Cobb for their help with field and laboratory work. We appreciate thetechnical reviews of the manuscript provided by K. Ho, P. Pelletier, and B.Bergen. The present study was supported by grant P42ES013660 (BrownUniversity Superfund Research Program) from the National Institute ofEnvironmental Health Sciences.

Disclaimer—The content of this document is solely the responsibility of theauthors and does not necessarily represent the views of the National Instituteof Environmental Health Sciences or the National Institutes of Health. Thisreport has been reviewed by the US Environmental Protection Agency’sOffice of Research and Development, National Health and EnvironmentalEffects Research Laboratory, Atlantic Ecology Division and approved forpublication (ORD Tracking No. ORD-003227). Approval does not signifythat the contents necessarily reflect the views of the Agency. Mention oftrade names or commercial products does not constitute endorsement orrecommendation for use.

REFERENCES

1. Novak PJ, ArnoldWA, Blazer VS, Halden RU, Klaper RD, Kolpin DW,Kriebel D, Love NG, Martinovic´-Weigelt D, Patisaul HB, Snyder SA,vom Saal FS, Weisbrod AV, Swackhamer DL. 2011. On the need for anational (U.S.) research program to elucidate the potential risks tohuman health and the environment posed by contaminants of emergingconcern. Environ Sci Technol 45:3829–3830.

2. Richardson SD. 2007. Water analysis: Emerging contaminants andcurrent issues. Anal Chem 79:4295–4324.

3. Rahman F, Langford KH, Scrimshaw MD, Lester JM. 2001.Polybrominated diphenyl ether (PBDE) flame retardants. Sci TotalEnviron 275:1–17.

4. Daughton CG, Ternes T. 1999. Pharmaceuticals and personal careproducts in the environment: Agents of subtle change? Environ HealthPerspect 107:907–937.

5. Hites RA. 2004. Polybrominated diphenyl ethers in the environment andin people: A meta-analysis of concentrations. Environ Sci Technol 38:945–956.

6. Darnerud PO, Eriksen GS, Johannesson T, Larsen PB, Viluksela M.2001. Polybrominated diphenyl ethers: Occurrence, dietary exposure,and toxicology. Environ Health Perspect 109:49–68.

7. de Wit CA. 2002. An overview of brominated flame retardants in theenvironment. Chemosphere 46:583–624.

8. McAvoy DC, Schatowitz B, Jacob M, Hauk A, Eckhoff WS. 2002.Measurement of triclosan in wastewater treatment systems. EnvironToxicol Chem 21:1323–1329.

9. Halden RU, Paull DH. 2005. Co-occurrence of triclocarban and triclosanin US water resources. Environ Sci Technol 39:1420–1426.

10. Kinney CA, Furlong ET, Kolpin DW, Burkhardt MR, Steven D, ZauggSD, Werner SL, Bossio JP, Benotti MJ. 2008. Bioaccumulation ofpharmaceuticals and other anthropogenic waste indicators in earth-worms from agricultural soil amended with biosolid or swine manure.Environ Sci Technol 42:1863–1870.

11. Singer H, Muller S, Tixier C, Pillonel L. 2002. Triclosan: Occurrenceand fate of a widely used biocide in the aquatic environment: Fieldmeasurements in wastewater treatment plants, surface waters, and lakesediments. Environ Sci Technol 36:4998–5004.

12. Morales S, Canosa P, Rodriguez I, Rubi E, Cela R. 2005. Microwaveassisted extraction followed by gas chromatography with tandem massspectrometry for the determination of triclosan and two relatedchlorophenols in sludge and sediments. J Chromatogr A 1082:128–135.

13. Miller TR, Heidler J, Chillrud SN, DeLaquil A, Ritchie JC, Mihalic JN,Bopp R, Halden RU. 2008. Fate of triclosan and evidence for reductivedechlorination of triclocarban in estuarine sediments. Environ SciTechnol 42:4570–4576.

Passive sampler performance for emerging contaminants Environ Toxicol Chem 32, 2013 2195

14. Cantwell MG, Wilson BA, Zhu J, Wallace GT, King JW, Olsen CR,Burgess RM, Smith JP. 2010. Temporal trends of triclosan contamina-tion in dated sediment cores from four urbanized estuaries: Evidence ofpreservation and accumulation. Chemosphere 78:347–352.

15. Dann AB, Hontela A. 2011. Triclosan: Environmental exposure, toxicityand mechanisms of action. J Appl Toxicol 31:285–311.

16. Perron MM, Ho KT, Cantwell MG, Burgess RM, Pelletier MC. 2012.Effects of triclosan onmarine benthic and epibenthic organisms.EnvironChem Toxicol 31:1861–1866.

17. Tittlemier SA, Halldorson T, Stern GA, Tomy GT. 2002. Vaporpressures, aqueous solubilities, and Henry’s law constants of somebrominated flame retardants. Environ Toxicol Chem 21:1804–1810.

18. Wania F, Dugani CB. 2003. Assessing the long-range transport potentialof polybrominated diphenyl ethers: A comparison of four multimediamodels. Environ Toxicol Chem 22:1252–1261.

19. Braekevelt E, Tittlemier SA, TomyGT. 2003. Direct measure of octanol-water partition coefficients of some environmentally relevant brominateddiphenyl ether congeners. Chemosphere 51:563–567.

20. Bao L, You J, Zeng EY. 2011. Sorption of PBDE in low-densitypolyethylene film: Implications for bioavailability of BDE-209. EnvironToxicol Chem 30:1731–1738.

21. Reiss R,Mackay N, Habig C, Griffin J. 2002. Ecological risk assessmentfor triclosan in lotic systems following discharge from wastewatertreatment plants in the United States. Environ Toxicol Chem 21:2483–2492.

22. Di Toro DM, Zarba CS, Hansen DJ, Berry WJ, Swartz RC, Cowan CE,Pavlou SP, Allen HE, Thomas NA, Paquin PR. 1991. Technical basis forestablishing sediment quality criteria for nonionic organic chemicalsusing equilibrium partitioning. Environ Toxicol Chem 10:1541–1583.

23. Huckins JN, Petty JD, Booij K. 2006.Monitors of Organic Chemicals inthe Environment. Springer, New York, NY, USA.

24. Hawthorne SB, Grabanski CB, Miller DJ, Kreitinger JP. 2005. Solid-phase microextraction measurement of parent and alkyl polycyclicaromatic hydrocarbons in milliliter sediment pore water samples anddetermination of KDOC values. Environ Sci Technol 39:2795–2803.

25. Maruya KA, Zeng EY, Tsukada D, Bay SM. 2009. A passive samplerbased on solid-phase microextraction for quantifying hydrophobicorganic contaminants in sediment pore water. Environ Toxicol Chem28:733–740.

26. Gschwend PM, MacFarlane JK, Reible DD, Lu X, Hawthorne SB,Nakles DV, Thompson T. 2011. Comparison of polymeric samplersfor accurately assessing PCBs in pore waters. Environ Toxicol Chem30:1288–1296.

27. Endo S, Hale SE, Goss K, Arp HPH. 2011. Equilibrium partitioncoefficients of diverse polar and nonpolar organic compounds to

polyoxymethylene (POM) passive sampling devices. Environ SciTechnol 45:10124–10132.

28. Sacks VP, Lohmann R. 2011. Development and use of polyethylenepassive samplers to detect triclosans and alkylphenols in an urbanestuary. Environ Sci Technol 45:2270–2277.

29. Sacks VP, Lohmann R. 2012. Freely dissolved PBDEs in water andporewater of an urban estuary. Environ Pollut 162:287–293.

30. Hartmann PC, Quinn JG, Cairns RW, King JW. 2005. Depositionalhistory of organic contaminants in Narragansett Bay, Rhode Island,USA. Mar Pollut Bull 50:388–395.

31. Latimer JS, Quinn JG. 1996. Historical trends and current inputs ofhydrophobic organic compounds in an urban estuary: The sedimentaryrecord. Environ Sci Technol 30:623–633.

32. Perron MM, Burgess RM, Suuberg EM, Cantwell MG, Pennell KG.Performance of passive samplers for monitoring estuarine water columnconcentrations: 1. Contaminants of concern. Environ Toxicol Chem32:2182–2189.

33. Schwarzenbach RP, Gschwend PM, ImbodenDM. 2003.EnvironmentalOrganic Chemistry, 2nd ed. John Wiley, Hoboken, NJ, USA.

34. Yang Z, Zhao Y, Tao F, Ran Y, Mai B, Zeng EY. 2007. Physicalorigin for the nonlinear sorption of very hydrophobic organicchemicals in a membrane-like polymer film. Chemosphere 69:1518–1524.

35. ter Laak TL, Busser FJM, Hermens JLM. 2008. Poly(dimethylsiloxane)as passive sampler material for hydrophobic chemicals: Effect ofchemical properties and sampler characteristics on partitioning andequilibration times. Anal Chem 80:3859–3866.

36. Canosa P, Rodriguez I, Rubi E, Cela R. 2005. Optimization of solid-phase microextraction conditions for the determination of triclosanand possible related compounds in water samples. J Chromatogr A1072:107–115.

37. Pietrogrande MC, Basaglia G. 2007. GC–MS analytical methods for thedetermination of personal-care products in water matrices. Trends AnalChem 26:1086–1094.

38. Yu Y, Wu L. 2012. Analysis of endocrine disrupting compounds,pharmaceuticals and personal care products in sewage sludge bychromatography-mass spectrometry. Talanta 89:258–263.

39. Sabaliunas D, Webb SF, Hauk A, Jacob M, Eckhoff WS. 2003.Environmental fate of triclosan in the river Aire basin, UK. Water Res37:3145–3154.

40. Katz D, Cantwell MG, Sullivan J, Perron MM, Burgess RM, Ho KT,Charpentier MA. 2013. Factors regulating the accumulation and spatialdistribution of the emerging contaminant triclosan in the sediments ofan urbanized estuary: Greenwich Bay, Rhode Island, USA. Sci TotalEnviron 443:123–133.

2196 Environ Toxicol Chem 32, 2013 M.M. Perron et al.