ARTICLE pubs.acs.org/est
Development and Use of Polyethylene Passive Samplers To Detect Triclosans and Alkylphenols in an Urban Estuary Victoria P. Sacks† and Rainer Lohmann*,† †
University of Rhode Island Graduate School of Oceanography, South Ferry Road, Narragansett, Rhode Island 02882, United States
bS Supporting Information ABSTRACT: To be able to use polyethylene passive samplers (PE) in the field, the partitioning constants between PE and water (KPEw) of the compounds examined must be known. The KPEws of triclosan (TCS), methyl-triclosan (MTCS), n-nonylphenol (n-NP), nonylphenol-technical mix (NP-tech), n-octylphenol (n-OP), and t-octylphenol (t-OP) were measured as a function of pH, temperature, and salinity, and a salt effect was calculated for TCS, n-OP, and t-OP. Log KPEws used for calculating dissolved concentrations were taken from 20 °C studies taking salt into account: 3.42 (TCS), 4.53 (MTCS), 4.20 (n-NP), 3.69 (n-OP), and 2.87 (t-OP). The KPEw of hydroxyl-group containing compounds were strongly affected by pH, whereas MTCS with its methoxy-group was not. Measured KPEws could not be estimated from octanol-water partitioning constants due to the semipolar makeup of the compounds investigated. Instead, a good correlation (KPEw = 0.679 Khdw þ 1.033, r2 = 0.984, p = 0.001) was obtained with hexadecane-water partitioning constants (Khdw) predicted from COSMOtherm. During deployments in Narragansett Bay (RI) in the fall of 2009, concentrations of MTCS and t-OP in surface and bottom waters ranged from 40-225 pg L-1 and 3.5-11 ng L-1, respectively. These concentrations are far below EC50 values for rainbow trout. Surface/bottom and bottom/porewater activity ratios were calculated. These indicated surface waters as the main source of MTCS, while surface water as well as sediments were sources of t-OP.
’ INTRODUCTION Polyethylene passive samplers (PE) have been widely used to concentrate trace hydrophobic organic contaminants (HOCs) in the environment (e.g. ref 1). Nonpolar persistent organic pollutants such as polychlorinated biphenyls (PCBs), polycyclic aromatic hydrocarbons (PAHs), etc. have previously been measured using this sampling approach. Due to the hydrophobic nature of these compounds and PE’s nonpolar structure, HOCs partition strongly into the PE matrix (104-109 times) and accumulate in the plastic until equilibrium is reached.2 This process concentrates the analytes prior to extraction resulting in lower detection limits, easier/less expensive laboratory methods, and reduced organic solvent use. Beyond traditional HOCs, emerging contaminants from pharmaceuticals and personal care products (PPCSs) are entering the environment. An understanding of these contaminants’ transport and accumulation in the environment is essential for maintaining a healthy environment and food web. The PPCPs of interest include endocrine disrupting compounds (EDCs) containing polar functional groups. EDCs have been increasingly gaining attention due to their capacity to interfere with natural hormone cycles in humans and wildlife, potentially affecting metabolism, development, reproduction, and growth.3 Effects of these chemicals have been observed at extremely low concentrations. PE samplers provide a r 2011 American Chemical Society
simple way to determine low concentrations from environmental samples. In addition, results from PE samplers represent timeintegrated concentrations, unavailable by active water sampling. Here we present results on the use of PE passive samplers to determine concentrations of selected semipolar contaminants in the aqueous environment. Triclosan (TCS) (5-chloro-2-(2,4-dichlorophenoxy)phenol) is widely used as an antibacterial/antimicrobial agent in personal care products and consumer items (soaps, toothpastes, etc.). TCS has received attention recently due to its potential as an EDC, and it has been shown to produce chronic or acute toxicity in aquatic species (e.g., ref 4). TCS is also bioaccumulative and has been detected in aquatic species tissue.5 Wastewater treatment does not completely remove TCS, and significant amounts are potentially released into effluent and surface waters.6 Once in surface waters, TCS has been shown to undergo both thermal and photochemical degradation7 and can form toxic polychlorinated dibenzo-p-dioxins.8 Methyl triclosan (MTCS) (2,4,40 -trichloro-20 -methoxydiphenyl ether) is Received: December 7, 2010 Accepted: January 20, 2011 Revised: January 15, 2011 Published: February 22, 2011 2270
dx.doi.org/10.1021/es1040865 | Environ. Sci. Technol. 2011, 45, 2270–2277
Environmental Science & Technology
ARTICLE
most likely a biological degradation product produced during wastewater treatment. MTCS is more lipophilic (log Kow of 5.4 compared to 4.8 (TCS)9) and has been shown to bioaccumulate in aquatic organisms. Both TCS and MTCS have been widely detected in surface waters in the United States.10,11 Alkylphenols (AP) are man-made chemicals used mainly in the production of alkylphenol ethoxylates, major components of nonionic surfactants found in many personal care products, detergents, paints, herbicides, and pesticides; they are also degradation products of alkylphenol ethoxylates.12 APs include a variety of chemicals, though nonylphenol (NP) and octylphenol (OP) are the most commercially important. Due to the polar hydroxyl group on the phenolic end, NP and OP are more soluble in water than nonpolar HOCs and a fraction can pass through wastewater treatment processes to enter into the environment. They have been found in sewage sludge, wastewater, sediments, and surface water (e.g., ref 13). NP and OP have demonstrated estrogenic activity in vitro and in vivo at μg L-1 concentrations,14 for example causing increased vitellogenin and decreased serum testosterone levels in male fish. The freely dissolved concentration of contaminants measured by passive samplers is the fraction that is directly bioavailable, as opposed to the fraction of contaminants sorbed to particulate or colloidal material.15 When using a PE passive sampler, the freely dissolved concentration of contaminant in the water (Cw) (e.g., ng L-1) is calculated via eq 1 as long as the partitioning behavior of the chemical (the PE-water partitioning constant, KPEw) and the concentration in the PE at equilibrium (CPE) (converted to ng L-1 using the density of PE) are known CPE Cw ¼ KPEw
CðPRCÞPE, t CðPRCÞPE, o
ð1Þ
ð2Þ
where C(PRC)PE,t is the concentration of PRC in the PE at time t (e.g., ng mL-1), and C(PRC)PE,o is the initial PE concentration of PRC (e.g., ng mL-1). Dissolved concentration, Cw, is then determined using eq 3 Cw ¼
CPE flost KPEw
’ METHODS Reagents and Chemicals. Standards of TCS,
For nonpolar HOCs, KPEws have been found to correlate closely with octanol-water equilibrium partitioning constants (Kow),16 which are readily available in the literature. In the case of polar contaminants (TCS and APs), KPEws cannot be reliably estimated from Kows because they have enhanced solubility in octanol compared to PE due to additional polar interactions. In the laboratory, KPEw values are derived via eq 1 rearranged, where CPE and Cw (e.g., ng L-1) are at equilibrium. In field deployments, the dissolved concentration, Cw, is calculated using the measured or derived KPEws. To do this, one must be certain that the CPE value used is the concentration in PE at equilibrium. To correct for the equilibrium concentration of contaminants in PEs deployed in the field, samplers are pretreated with performance reference compounds (PRCs). PRCs structurally resemble the compounds of interest but are not naturally found in the environment. As both the uptake and the release of organic contaminants are governed by molecular diffusion,17 the fraction of PRCs lost (flost) (eq 2) can be used to determine the extent of equilibrium reached for the compounds of interest flost ¼ 1 -
As part of this study, we deployed samplers at various sites throughout Narragansett Bay in the early fall of 2009. Results were compared to those obtained using the commercially available polar organic chemical integrative sampler (POCIS). POCIS is a widely used passive sampler consisting of a solid sorbent material contained between two microporous membranes which measures polar or hydrophilic organic chemicals from aqueous environments.18 At two sites, sediment grab samples were collected, as sediments can potentially act as secondary sources of accumulated historical pollution. Sediments were incubated with PE samplers to measure porewater concentrations and derive porewater-water column gradients of the contaminants of interest. In summary, we undertook this study to (i) determine the PE-water partitioning behavior of emerging contaminants of concern - triclosan (TCS), methyl triclosan (MTCS), n-nonylphenol (n-NP), nonylphenol-technical mixture (NP-tech), n-octylphenol (n-OP), and tert-octylphenol (t-OP); (ii) compare different approaches in predicting KPEw by using hexadecanewater partitioning constants (Khdw); (iii) assess the effects of pH, temperature, and salt on KPEw; (iv) determine freely dissolved concentrations of these chemicals using PE samplers and POCIS in surface and bottom waters of Narragansett Bay (RI); and (v) measure porewater concentrations and derive environmental gradients in the field.
ð3Þ
13
C12 TCS, MTCS, C12 MTCS, n-NP, NP-technical mixture, 13C n-NP, and p-terphenyl-d14 were obtained from Cambridge Isotope Laboratories Inc. (Andover, MA, USA). n-OP, and OP-d17 were obtained from Chiron via RT Corporation (Laramie, WY, USA), and NP-d4 was obtained from Isotech Inc. (Miamisburg, OH, USA). OP-d17 and NP-d4 were used as PRCs in this study. Clean water used for laboratory partitioning experiments was obtained using a Milli-Q water purification system (Millipore). Laboratory Partitioning Experiments. Low density PE was cut into 0.05 g pieces from 25.4 μm plastic drop cloth (Covalence Plastics, Minneapolis, MN) and precleaned twice by extraction in dichloromethane and hexane for 24 h. Each piece of polyethylene was punctured with a small stainless steel ring to avoid floatation and placed in a precleaned 250 mL glass round-bottom flask (muffled and solvent rinsed in acetone and hexane). Each sample was filled with 250 mL of milli-Q water and spiked with 1000 ng of TCS, MTCS, n-NP, NP-tech, n-OP, and t-OP. Triplicate samples and blanks using PRC-loaded-PE were shaken in climate controlled chambers under four conditions: 5 °C/0 psu, 5 °C/ 93 psu, 20 °C/0 psu, and 20 °C/93 psu for 6 weeks on an orbital shaker table. Replicates at pH 7-12 (1 sample for each pH) were prepared with PE and PPCPs as above using Tris (National Diagnostics), HCl (J.T. Baker), NaHCO3, KCl (Fischer Scientific), and NaOH (Mallinckrodt) to adjust to a given pH. pKa values used in this study were taken from Tixier et al., 2002 (TCS) and Muller and Schlatter, 1998 (n-NP, n-OP).7,19 PE was extracted twice in n-hexane using sealed glass containers, and extracts were concentrated to 1 mL using a Turbovap (Turbovap II, Caliper Sciences, Inc.; 38 °C, 5 psi). Samples were further concentrated to 50 μL with a gentle stream of high-purity nitrogen gas. Water was liquid-liquid extracted with dichloromethane following a modified version of EPA method 3510C and further concentrated to 50 μL as above. Glass flasks were 13
2271
dx.doi.org/10.1021/es1040865 |Environ. Sci. Technol. 2011, 45, 2270–2277
Environmental Science & Technology
ARTICLE
Table 1. Laboratory-Derived Polyethylene-Water Partitioning Constants (KPEw) and Salt-Corrected KPEw Values Used for Field Deployments of Selected Trace Organic Contaminants partitioning constants 20 °C/0 PSU replicates (6 week, pH replicates) n
KPEw
stdev
salt corrected (field samples) log KPEw
KPEw
log KPEw
triclosan
TCS
4
2167
261
3.34
2634
3.42
methyl triclosan
MTCS
8
33737
13051
4.53
33737
4.53
n-nonylphenol nonylphenol-tech
n-NP NP-tech
7 7
15699 3227
5043 757
4.20 3.51
15699 3227
4.20 3.51
n-octylphenol
n-OP
6
4325
912
3.64
4953
3.69
t-octylphenol
t-OP
3
550
134
2.74
747
2.87
nonylphenol-d4
n-NP-d4
3
8794
1467
3.94
8794
3.94
octylphenol-d17
n-OP-d17
3
8327
3900
3.92
9537
3.98
extracted three times with 20 mL of dichloromethane after removal of the water and PE to evaluate the effects of glass adsorption. Comparison of KPEw to Khdw. Khdw values were (i) calculated via Abraham descriptors and the relevant poly parameter linear free energy relationship (pp-LFER) shown in eq 420 log Khdw ¼ 0:667E-1:617S-3:587A-4:869B þ 4:433V þ 0:087 ð4Þ
where E and V describe nonspecific molecular interactions such as van der Waals forces and cavity formation, S describes polarizability and dipolarity, and A and B represent the hydrogen bond donor and acceptor capacity. These parameters were estimated using ABSOLV, a commercial software which is based on a group contribution approach for property estimations (personal communication, Dr. Dominic Di Toro, University of Delaware). In addition, Khdw values were (ii) obtained from the SPARC online calculator.21 SPARC is chemical property estimation software run by the U.S. EPA where Khdw values are predicted using the SMILES string indicating the molecular structure. SPARC uses solvation models that describe the intermolecular interaction upon placing an organic solute molecule in a solvent system.22 Lastly, Khdw values were (iii) calculated using the commercial COSMOtherm software (personal communication, Dr. Kai-Uwe Goss, Helmholtz Centre for Environmental Research UFZ) (version C2.1 release 01.08 in combination with Turbomole 5.10), which calculates solvation properties using statistical thermodynamic approaches.23 Log Khdw values are included in SI Table 1. Field Sampler Preparation. Polyethylene was cut into approximately 2 g sheets from 25.4 μm plastic drop cloth and precleaned twice in dichloromethane and hexane. The PEs were then dried, placed in a solution of deuterated PRCs, and stirred in darkness for three weeks using a modified method from Booij et al.17 (80:20, water:methanol). “Spiked” PE samplers were then dried using Kimwipes, strung on stainless steel wire, wrapped in precombusted aluminum foil, and stored at -20 °C until use. Field Collection and Experimental Design. Two spatial deployments of Narragansett Bay were executed at eight water quality buoys which make up the Narragansett Bay Fixed-Site Monitoring Network24 (see the SI and SI Figure 1 for details). For the field deployment, PE samplers were strung on stainless steel wire, attached to a line 1 m below the surface and 1 m above the sediment, and allowed to equilibrate for four weeks in the early fall of 2009. Upon retrieval, all samplers were wrapped in cleaned foil and placed in a freezer at -20 °C until extraction.
Prior to extraction, PEs were wiped free of biofouling and dried using Kimwipes. PE was twice extracted in n-hexane using sealed glass containers and extracts concentrated to 50 μL as above. Laboratory and field blanks showed concentrations of contaminants of interest well below 10% of the sample concentrations so results were not blank corrected. At two sites POCIS, consisting of a solid sorbent material contained between two microporous polyethersulfone membranes, were deployed. POCIS were rented from and extracted by Environmental Sampling Technologies, Inc. (EST, St. Joseph, MO, USA) (see the SI for details). Sediment Porewater. The surface 3 cm of sediment at two sites (Quonset Point, and Conimicut Point) was collected at the time of PE recovery using a standard grab sampler. PE samplers were cut into approximately 0.2, 0.6, and 1.1 g pieces and cleaned twice in dichloromethane for 48 h. PE was dried, punctured with a small stainless steel ring, and placed in 250 mL precleaned round-bottom flasks with 100 g of fully homogenized sediment (approximately 50 g dry weight), 7 mg of bactericide (sodium azide), and milli-Q water such that there was minimum airspace in the flasks upon closure (n = 6). Assuming desorption from sediment is the rate-limiting step,25 larger pieces of PE take longer to come to equilibrium. Equilibrium is confirmed when different sizes of PE show the same concentration. At the end of incubations (6 and 9 weeks), PEs were wiped free of sediment with milli-Q water and Kimwipes, twice extracted in hexane, and concentrated to 50 μL using a Turbovap and a gentle stream of nitrogen gas. Analysis. Details of the simultaneous analysis of TCS, MTCS, n-NP, NP-tech, n-OP, and t-OP have not previously been reported. The first step in this work was to develop a GC/MS method. Derivatization with n-(t-butyldimethylsilyl)N-methyltrifluoroacetamide (MTBSTFA) was explored but ultimately deemed unnecessary because 10s pg L-1 detection limits were attainable using underivatized compounds (see the SI), albeit at the cost of frequent GC maintenance. Gas chromatography/mass spectrometry analyses were conducted with an Agilent 6890 coupled with a 5973 MS using a DB-5MS column (Agilent J&W GC Columns, 122-5532, length 30 m, ID 0.250 mm, film 0.25 μm) with splitless injection (275 °C), electron-impact ionization (EI, 70 eV, transfer line 300 °C) in selected-ion-monitoring mode (SIM) (SI Table 2). The GC temperature program was as follows, 80 °C held for 1 min, 12 °C min-1 to 220 °C, 10 °C min-1 to 280 °C, and held for 12 min. The injection volume was 1 μL, and the helium carrier gas flow was maintained at 1 mL min-1. 2272
dx.doi.org/10.1021/es1040865 |Environ. Sci. Technol. 2011, 45, 2270–2277
Environmental Science & Technology
ARTICLE
Figure 1. Correlations of measured KPEw (20 °C/0 psu) for selected trace organic contaminants against their Khdw values estimated using Abraham descriptors, SPARC, and COSMOtherm. For comparison, measured KPEw for various pesticides from Hale et al. (2010) are included.26
Quality Control. For quality assurance and recovery monitoring of extractions, internal standards of 13C labeled surrogate compounds were added prior to extraction (50 ng). 13C TCS was used as an internal standard for TCS, 13C MTCS for MTCS, and 13C n-NP for n-NP, NP-tech, n-OP, and t-OP. Furthermore, p-terphenyl-d14 was added to final extracts as an injection standard to monitor analytical and instrumental variability (50 ng). A calibration standard curve was used to derive response factors of the analytes relative to the appropriate isotope-labeled internal standards. It was run for every 40 samples or as deemed necessary determined by a check standard analyzed every 10 samples. Recoveries of the internal standards were calculated relative to the injection standard which was added prior to injection into the GC-MS. Recoveries of 13C labeled internal standards for deployed PE typically ranged from 73% to 122%. For partitioning experiments ∼60 to 120% of added compounds were accounted for in PE, water, and glass extracts (SI Table 3).
’ RESULTS AND DISCUSSION Partitioning Experiments - KPEw. Log KPEw (average of 6 week equilibration and pH experiments, Table 1) displayed a wide range, from 2.7 for t-OP to 4.5 for MTCS. While these values are well below KPEw observed for HOCs, they still indicate significant enrichment in PE, on the order of 550->30,000 times relative to freely dissolved concentrations. As expected, the less polar compound, MTCS (4.5), displayed a higher KPEw than TCS (3.3). Within the alkylphenols, the longer chained n-NP (4.2) showed higher KPEw than n-OP (3.6), and the linear chain n-OP had a higher KPEw than the branched t-OP (2.7). As the diverging results for the different alkylphenols seemed surprising at first, we needed to verify that equilibrium had indeed been reached in our experiments. For 6 week partitioning experiments, equilibrium was confirmed with both PRCs and by comparison to the mid range pH replicates. The value of KPEw for n-NP-d4 (PRC) agreed with KPEw for n-NP (analyte) in the 5 °C/0 psu, 20 °C/93 psu, and 20 °C/0 psu experiments (SI Table 4). The PRC n-OP-d17 is significantly heavier than n-OP and more closely resembles n-NP in molecular weight. Comparable results for n-OP-d17 and n-NP for the same conditions further confirmed equilibrium. Due to the smaller size of n-OP, we concluded that n-OP had also reached equilibrium. In the 5 °C/93 psu experiment, PRCs displayed slightly higher KPEw values than native compounds (by 7 and 23% for NP-d4 and OP-d17, respectively). The results were within uncertainties suggesting that the experiments were close to equilibrium, as was confirmed by the comparison of the 6 week and pH partitioning experiments.
Partitioning Experiments - Khdw Comparison. Due to the additional interaction of the hydroxyl group with octanol in comparison to PE, a correlation cannot be made between KPEw and Kow for the compounds studied. In contrast, hexadecane, a long chain hydrocarbon, is a better choice to represent PE. Following a study by Hale et al. (2010), we aimed to see whether the available data for Khdw (hexadecane-water partitioning constants) resulted in good correlations with our measured KPEw.26 A significant correlation is the first step toward relying on Khdw to predict KPEw for further compounds. Khdw values were calculated using (i) Abraham descriptors, (ii) SPARC, and (iii) COSMOtherm software. Similar to the conclusion reached by Hale et al. (2010), a significantly better correlation was observed using Khdw values predicted with COSMOtherm software (r2 = 0.98, p = 0.001) than with Abraham’s pp-LFER (r2 = 0.56, p = 0.14) or with SPARC generated values (r2 = 0.73, p = 0.07) (Figure 1). The KPEws derived in this study correlated even better with Khdws predicted from COSMOtherm than those measured previously for various pesticides.26 This further supports the use of COSMOtherm as a tool for predicting Khdw values. The poor correlation with the Abraham pp-LFER might be because parameters were estimated with the ABSOLV software rather than measured experimentally. We note however that even measured Abraham parameters at times need to be refined.27 Once a sufficient number of KPEws are experimentally determined, they can be used to derive improved Abraham parameters. Our results support previous results that suggest SPARC is not as good in predicting partitioning properties as the COSMOtherm software. While additional work using a wider range of compounds is needed, the COSMOtherm results look very promising and could be useful for predicting KPEws in the future. These KPEw in turn could be used for an initial risk assessment when PE is deployed to screen for EDCs. Partitioning Experiments - Effect of pH. The change of partitioning as a function of pH (n = 1 for each pH) was examined experimentally (Figure 2). As expected, the KPEwS of hydroxyl-group containing compounds were strongly affected by pH, whereas MTCS, with its methoxy-group, was not. For comparison to our measurements, we predicted partitioning as a function of pH (KPEw(pH)) assuming that KPEw only accumulates neutral molecules (eq 5)
log KPEw ðpHÞ ¼ log KPEw ðpH ¼ 7Þ þ log
1
1 þ 10ðpH - pKaÞ
ð5Þ
The change in KPEw is consistent with predictions such that at a compounds’ pKa value, its KPEw has decreased by 50% (Figure 2). 2273
dx.doi.org/10.1021/es1040865 |Environ. Sci. Technol. 2011, 45, 2270–2277
Environmental Science & Technology
ARTICLE
Figure 2. Measured versus predicted log KPEw as a function of pH for methyl triclosan (MTCS), n-nonylphenol (n-NP), n-octylphenol (n-OP), nonylphenol technical mixture (NP-tech), and triclosan (TCS).
For the hydroxyl-containing compounds (TCS, n-NP, NP-tech, and n-OP), the KPEw of the protonated (neutral) molecules exceeded the deprotonated (charged) species by ∼2 orders of magnitude. This confirms that PE is primarily a sampling medium for neutral, apolar compounds. At higher pH (e.g., > 11) the measured log KPEw for TCS was higher (∼1) than the predicted value (1 at most locations indicating a higher activity in surface waters than in bottom waters. Ratios were considered significant when they differed by more than three standard deviations from unity (observed at all locations except Mount
View, Bullocks Neck, and Conimicut Point). The bottom/ porewater activity ratios (defined by bottom water PE concentration divided by porewater PE concentration) were established to indicate a gradient into or out of the sediment. The ratios for MTCS were small yet significantly >1 at Quonset Point; MTCS was not detected in porewater at Conimicut Point. These findings demonstrate a top-down activity gradient showing surface water as the source of MTCS in Narragansett Bay, which agrees with our assumption that it might derive from WWTP effluents. The water column appeared close to equilibrium at all locations for t-OP as determined by surface/bottom activity ratios with the exception of Mount View and Bullocks Neck (located at the mouth of Greenwich Bay and below the outfall from a large WWTP) where activity gradients were small, yet significantly >1. Bottom/porewater activity ratios for t-OP were
