Partitioning of hydrophobic organic contaminants between polymer and lipids for two silicones and low density polyethylene

Chemosphere 186 (2017) 948e957

Contents lists available at ScienceDirect

Chemosphere journal homepage: www.elsevier.com/locate/chemosphere

Partitioning of hydrophobic organic contaminants between polymer and lipids for two silicones and low density polyethylene Foppe Smedes a, b, *, Tatsiana P. Rusina a, Henry Beeltje c, Philipp Mayer d a

Masaryk University, Faculty of Science, Research Centre for Toxic Compounds in the Environment (RECETOX), Kamenice 753/5, 625 00 Brno, Czech Republic b Deltares, P.O. Box 85467, 3508 AL Utrecht, The Netherlands c TNO, P.O. 80015, 3508 TA Utrecht, The Netherlands d Technical University of Denmark, Department of Environmental Engineering, Kongens Lyngby, Copenhagen, Denmark

h i g h l i g h t s

g r a p h i c a l a b s t r a c t


Relatively fast lipid diffusion in Silicone and LDPE.
Partition coefficients provided for 78 hydrophobic substances.
Lipid type and temperature does not affect partitioning.
Partitioning not affected by polymer lipid sorption.
Passive sampling results can be converted to lipid based equivalent.

a r t i c l e i n f o

a b s t r a c t

Article history: Received 10 June 2017 Received in revised form 8 August 2017 Accepted 9 August 2017 Available online 11 August 2017

Polymers are increasingly used for passive sampling of neutral hydrophobic organic substances (HOC) in environmental media including water, air, soil, sediment and even biological tissue. The equilibrium concentration of HOC in the polymer can be measured and then converted into equilibrium concentrations in other (defined) media, which however requires appropriate polymer to media partition coefficients. We determined thus polymer-lipid partition coefficients (KPL) of various PCB, PAH and organochlorine pesticides by equilibration of two silicones and low density polyethylene (LDPE) with fish oil and Triolein at 4  C and 20  C. We observed (i) that KPL was largely independent of lipid type and temperature, (ii) that lipid diffusion rates in the polymers were higher compared to predictions based on their molecular volume, (iii) that silicones showed higher lipid diffusion and lower lipid sorption compared to LDPE and (iv) that absorbed lipid behaved like a co-solute and did not affect the partitioning of HOC at least for the smaller molecular size HOC. The obtained KPL can convert measured equilibrium concentrations in passive sampling polymers into equilibrium concentrations in lipid, which then can be used (1) for environmental quality monitoring and assessment, (2) for thermodynamic exposure assessment and (3) for assessing the linkage between passive sampling and the traditionally measured lipid-normalized concentrations in biota. LDPE-lipid partition coefficients may also be of use for a thermodynamically sound risk assessment of HOC contained in microplastics. © 2017 Published by Elsevier Ltd.

Handling Editor: Keith Maruya Keywords: Passive sampling Partition coefficient Lipid Diffusion coefficient Microplastic

1. Introduction * Corresponding author. Masaryk University, Faculty of Science, Research Centre for Toxic Compounds in the Environment (RECETOX), Kamenice 753/5, 625 00 Brno, Czech Republic. E-mail address: [email protected] (F. Smedes). http://dx.doi.org/10.1016/j.chemosphere.2017.08.044 0045-6535/© 2017 Published by Elsevier Ltd.

Equilibrium passive sampling (EPS) in environmental media involves enrichment of organic pollutants into a micrometer thin

F. Smedes et al. / Chemosphere 186 (2017) 948e957

polymer while keeping matrix and interfering constituents outside the polymer (Mayer et al., 2003). Concentrations in polymers equilibrated with media (CP#media) can then be converted to equilibrium partitioning concentrations in other defined media, given the appropriate polymer to media partition coefficient is available (Gilbert et al., 2016). Dividing CP#media by polymerwater partition coefficients, for example, yields freely dissolved concentrations (Cfree) (Booij et al., 2016; Friedman et al., 2009; Kraaij et al., 2003); while dividing them with a polymer-lipid partition coefficient (KPL) would yield equilibrium concentrations €€ in lipids (CL#media) (Jahnke et al., 2014a; M€ aenpa a et al., 2011). The CL#media provides a basis for thermodynamic exposure assessment and bioaccumulation and toxicity evaluation. Recently, EPS has been extended to biological tissue using solid phase microextraction (SPME) fibers (Ossiander et al., 2008) and polydimethylsiloxane (PDMS) sheets (Jahnke et al., 2011), including in vivo applications (Allan et al., 2013). Analyte concentrations measured in polymers equilibrated with biological tissue can then be converted to equilibrium concentrations in lipids (CL#tissue) using KPL. A CL#tissue derived from in-tissue EPS showed good agreement with lipid-normalized concentrations obtained using conventional solvent extraction of tissue (Jahnke et al., 2011). Furthermore, concentrations in passive samplers equilibrated with soil and sediment have been shown to be proportional to lipidnormalized concentrations in organisms (CL) exposed to or originating from the same habitat (Friedman et al., 2009; Kraaij et al., 2003). Lipid-normalized PCB concentrations in chironomid larvae, for example, differed by less than a factor of two from CL#media €enp€ € et al., 2011). obtained by EPS of the habitat sediment (Ma aa However, derived CL#media should not be interpreted as predictions of actual lipid-based HOC concentrations within the organism; rather, they serve as a well-defined thermodynamic reference that quantifies the partitioning driven exposure level within a given media or habitat (Jahnke et al., 2014a). Actual levels may differ from equilibrium due to biomagnification, metabolism, natural variability, and other confounding factors. The KPL required to convert measured polymer concentrations into lipid-based concentrations are scarce in the literature, having only been reported for PCB, hexachlorobenzene, and a few organochlorine pesticides in silicone (Dürig et al., 2016; Jahnke et al., 2008). Reported partition coefficients showed negligible variation over the different lipids applied. However, it remained unclear as to whether observed polymer weight enhancements were caused by lipid adhering to the polymer surface or through lipid absorption into the polymer. The formation of a lipid film on the polymer would require a partition coefficient correction; while absorption into the polymer could change the polymer's partitioning properties. For PDMS-coated SPME fibers, for example, contact with lipid (and other matrices) has been shown to have little effect (10%) on their sorption partitioning properties (Jahnke and Mayer, 2010). The study aims are (1) to analytically determine KPL for a wide range of HOC for two silicone polymers and low density polyethylene (LDPE) (2) to quantify the influence of temperature and lipid nature on KPL, and (3) to clarify whether polymer-lipid contact affects its partition coefficient. In addition, we provide a short discussion on future applications for KPL. 2. Working principle An experimental system was designed that not only determined partition coefficients but also allowed the influence of adsorbed and absorbed lipid to be assessed. The upper disk of a polymer disk stack was brought into contact with an analyte-spiked lipid and allowed to equilibrate through uptake. An additional disk was immersed in the lipid (Fig. 1). The bottom disk in the stack (not in

949

Fig. 1. Experimental setup for the polymer-lipid partition experiment. A stack of polymer disks was placed onto a stainless-steel disk (SS) in the lid of a glass jar and brought into contact with lipid dosed with target substances. One disk was directly immersed in the lipid. The bottom Altesil disk (isolated in the LDPE incubation) was dosed with performance reference compounds.

contact with the lipid) was dosed with performance reference compounds (PRC) that equilibrate with the lipid in the opposite direction to the target compounds. This method was applied using triolein and fish oil at 4  C and 20  C for periods of between 8 and 42 days. After selected contact periods, all disks and the lipids were analyzed for spiked contaminants. Since HOC diffuse rapidly in polymers (Rusina et al., 2010; Rusina et al., 2007), we assumed that they would migrate faster than the lipid, and thus would attain equilibrium quicker. This setup would allow a comparison of partitioning by the lipid-dosed substances in the presence (immersed disk) and absence (bottom disk) of lipid in the polymer. Polymer diffusion of fish oil and triolein was also studied in a separate experiment, using two silicones (Altesil, SSP) and LDPE. 3. Method and materials 3.1. Materials Translucent Altesil silicone rubber sheets (300  300 cm and 0.5 mm thick) were purchased from Altec Products Limited (UK), 0.25 mm thick SSP-M823 silicone rubber membranes were obtained from Specialty Silicone Products (Ballston Spa, NY, USA), and LDPE (0.07 mm thick) was obtained from Brentwood Plastics (Brentwood, MO, USA). The polymers were pre-extracted in a Soxhlet apparatus (Altesil, SSP) or by shaking in ethyl acetate for four days (LDPE). Partitioning experiments were undertaken using the 16 EPA PAH, 22 OCP, and 24 PCB, including dioxin-like PCB. Fourteen PCB not occurring in industrial mixtures were used as PRC. (a full substance list is provided in the Supplementary information, (S1). These substances, substance mixtures, along with recovery internal standards (PCB 209, mirex, 13C-gamma-HCH, 13C2,4-DDT, 13C-PCB 52, 13C-PCB 180) and syringe internal standards (1,2,3,4-tetrachloronaphthalene (TCN), PCB 143), were obtained from various suppliers in The Netherlands. Olive oil (extra virgin) was purchased in a local supermarket (max. free fatty acids <1% as oleic acid); fish oil (Tobis omega-3 fish oil) was obtained from P.W. Health Supplies (Pharmacist Tracey Peake MRPharmS, UK). Triolein (glyceryl trioleate,  97.0% pure) and Tricosane (lipid analysis internal standard) were obtained from Sigma-Aldrich. All solvents

950

F. Smedes et al. / Chemosphere 186 (2017) 948e957

were pesticide grade or equivalent (Boom BV, Meppel, The Netherlands). 3.2. Lipid diffusion measurement Fish oil and olive oil diffusion coefficients (DP) in the polymers were measured using the film stack technique (Rusina et al., 2010), utilizing the setup and methods described in S2. Briefly, a stack of five polymer disks was brought into contact with a lipid saturated disk, from which all adhering lipid was removed by extensive wiping with paper tissue. After the selected contact time, the disks were separated and individually analyzed for lipid content. Diffusion of the lipid over the polymer stack was then modelled using Fick's second law of diffusion (Crank, 1957), and the model was fitted to the data with the diffusion coefficient as an adjustable parameter (Rusina et al., 2010; Rusina et al., 2007). 3.3. Polymer-lipid partitioning experiment Solvent-free dosing of 0.45e2.5 mg/g HOC to triolein and fish oil was performed through release from an immersed Altesil silicone sheet (30 h) that had been spiked with the appropriate amount of HOC. For the polymer-lipid partitioning experiment, multiple 42 mm diameter disks were stacked onto a stainless-steel disk inside the screw cap of a 100-mL amber glass jar (Fig. 1). To the jar 2 mL of HOC dosed lipid (triolein or fish oil) was added wherein one 24 mm polymer disk (Altesil, SSP, or LDPE) was immersed. The jar was then closed with the screw cap containing the stack of sheets. Two to four disks were included in the stack, depending on polymer type (S3). Starting from the bottom stainless steel disk, the disk order for Altesil (A) was a disk spiked with PRC (750e5350 ng/disk) through a water-methanol mixture(Booij et al., 2002) followed by two non-spiked disk; for SSP (B) the stack started with one PRCspiked Altesil disk followed by three non-spiked SSP disks; and for LDPE (C), the stack comprised one non-spiked LDPE disk followed by a PRC-spiked Altesil disk and an upper non-spiked LDPE disk. By placing the spiked Altesil disk between the LDPE disks, better contact was anticipated between the disks, preventing the LDPE disks from sliding. By turning the glass jar upside down, the spiked lipid flowed down, resulting in direct lipid contact with the upper (facing) sheet of the stack. The PRC-spiked Altesil disks in each experiment served as a kind of internal reference between the various equilibrations. Additionally, after incubation, the distribution of added PRC should provide information on equilibrium attainment. For each polymer-lipid combination, three exposures were performed at 4  C and 20  C for periods of between 9 and 40 days (S3). For LDPE-lipid incubations, longer exposure times were selected due to known slower substance diffusion in LDPE. After the selected contact period, the disks were retrieved and wiped with paper tissue to remove any lipid from the polymer surface. 3.4. Analysis The analytical procedures applied are described fully in S4 and briefly summarized below. Polymer disks were extracted with acetonitrile. The extract was cleaned on a C18-bound silica column to remove residual lipid, and after concentration transferred to hexane, followed by quantification of substances (HOC and PRC) by instrumental analyses. To quantify the total lipid absorbed by the polymer disk, first the lipid retained on the C18 clean-up column was eluted with acetone/hexane, where after this eluent was then used to extract any lipid remaining in the polymer disk. The lipid in the final extract was estimated using reversed phase HPLC, applying a gradient of acetone-water and evaporative light scattering detection (ELSD). Standard lipid solutions were used for

calibration, with triolein and tricosane used as internal standards for estimating fish oil and triolein (from olive oil), respectively (S2). The ELSD signal for the samples was brought into the appropriate calibration range by adjusting the final volume and associated internal standard amounts based on expected lipid content. Substances in the lipid were isolated from a pentane-diluted lipid sample using twofold extraction with acetonitrile at low temperature (18 to 10  C), lipid solubility being reduced in the acetonitrile phase. The acetonitrile extract was then treated similarly to the extracts of the polymer disks. Instrumental analysis for HOC and PRC was performed using GC-MS. All QA/QC measurements and results are provided in S4. 4. Results and discussion 4.1. Lipid diffusion coefficients in polymers Diffusion measurements indicated relatively rapid diffusion of lipids through the polymers (see S5 for diffusion profiles for triolein in Altesil and LDPE after 1.3 and 17 days, respectively). The lipid diffusion was faster through silicone polymers than LDPE, as earlier observed for PCB and PAH(Rusina et al., 2010). Triolein DP (S6) was measured using olive oil, in which triolein is the main lipid (65% peak area). As fish oil chromatograms showed many non-separated peaks (chromatograms shown in S7), diffusion coefficients (DP) could not be derived for individual lipid compounds; however, using the total peak area of the unresolved hump as an overall concentration measure, the DP obtained was found to be comparable with that for triolein. Lipid DP were temperature dependent and increased in Altesil by a factor of five (0.7 log unit) over a 2e47  C temperature range (S8). Compared with HOC such as PCB180 and indeno[123,-c,d]pyrene, lipid logDP in silicone and LDPE were lower by 0.4 and 1.5 log unit, respectively (S6). Extrapolation of the relationship for PCB and PAH between logDP (Rusina et al., 2010) and logMV (molecular volume, Molinspiration Cheminformatics Software, http://www.molinspira tion.com. 2017, S1), predicted lipid logDP for Altesil and LDPE lower by 1.3 and 2.4 log unit, respectively, than those observed (S9). The greater diffusion than expected from lipid MV may be associated with the lipid's flexible structure (Saleem et al., 1989). The lipid diffusion observed in LDPE does not agree with the MW cutoff at >600 Da suggested literature (Huckins et al., 2006) for entry of molecules into LDPE. 4.2. Lipid distribution in the polymer stacks Lipid distribution over the polymer stacks incubated with HOCdosed lipid was mostly complete within the applied incubation times (S10). For Altesil and SSP this agreed with model calculation using similar numerical fitting as for DP estimation (Section 3.2) applying constant lipid concentration at the polymer surface as the boundary condition. The calculation predicted 90% equilibrium for the bottom disk of the-stack after 8 d and 16 d at 20  C and 4  C exposures, respectively. For the 40-day incubations with LDPE we did not expect a full equilibration of lipids between the bulk lipid phase and the disks in the stack. Calculation showed 40% equilibration progress for the lipid-facing and 2% for the bottom disk, adopting negligible transport resistance for the Altesil disk situated in-between them. However, in practice lipid was found in all disks of the stack for most of the lipid-LDPE incubations (S10 C). Possibly, such rapid lipid distribution was caused by capillary forces which pulled lipid through the interface between glass and polymer. Oils, like lubricants, can enter interfaces even if there is hardly any space. Yet, disks contained only sorbed lipid as no visible lipid was noticed on

F. Smedes et al. / Chemosphere 186 (2017) 948e957

the surface of the disks separated after incubation, what also excluded leakage. The effect of absorbed lipid on substance partitioning could still be evaluated (Section 4.5) using three incubations in which the lipid content in the disks decreased towards the bottom of the stack (S10). Likely, the equilibration was slowed down because of inadequate disk contact, coincidently providing the desired lipid gradient. Silicones and LDPE absorbed 4e7 and 15e25 mg lipid g1 polymer, respectively (S10). The lipid content showed no consistent difference between immersed, lipid-facing and middle disks. Excluding the three incubations with inadequate disk contact the average ratio between immersed and middle disks with no lipid contact, was close to unity (0.98 ± 0.08 SE, n ¼ 14). Consequently, lipid concentrations (S10) mainly comprise lipid absorbed by polymers, there being no evidence of substantial adherence of lipid after wiping. Therefore, the general assumption (Dürig et al., 2016; Jahnke et al., 2008) that the weight increase in polymers exposed to lipid is caused by lipid adhering to the polymer surface was not supported. 4.3. Equilibrium attainment Once equilibrium is reached, KPL can be estimated from substance concentrations in the polymer (CP) and lipid (CL) using the

951

formula:

KPL ¼

CP CL

(1)

Equilibrium was attained when CP/CL for each disk in the stack was equal. As an example, CP/CL for Altesil (Fig. 2, panel A; lipiddosed PCB 153) showed little variation throughout the stack, indicating that equilibrium was reached. Likewise, CP/CL for SSP (Fig. 2, panel B) was mostly constant, only showing a clear decline towards the bottom sheets for the nine-day fish oil incubation at 4  C (Fig. 2, panel B, dashed blue line). The decline supports the suggested inadequate disk contact for this incubation (Section 4.2) as PCB153 would had been equilibrated in two days, and equivalent incubations did attain equilibrium. In Fig. 2 the distance of markers from left side of the bar indicates the lipid content and especially the SSP results in panel B shows that lower CP/CL is associated with lower lipid content. This association is made more visible by plotting CP/CL for different substances against disk lipid content (S11). This shows that CP/CL for substances with smaller molecular size and higher DP (Rusina et al., 2010) (pentachlorobenzene and phenanthrene), attained equilibrium without relationship with lipid content, while those with larger molecular sizes and lower DP (PCB 187 and indeno[1,2,3-cd]pyrene) still showed a decline over the stack. Similarly, for PRC dosed into the Altesil bottom disk,

Fig. 2. Polymer-lipid ratios for PCB 153 concentrations (CP/CL) in individual disks and lipid for different incubation conditions with three polymers: A) Altesil, B) SSP, and C) LDPE. Going from left to right within each of the three panels, CP/CL (y-axis) is depicted for the bottom, isolated, facing and immersed disk(s), respectively. CP/CL for Altesil, SSP, and LDPE are represented by squares, circles and triangles, respectively. Red and blue indicates whether the incubation was with triolein or fish oil, respectively, while open and filled markers indicate incubations at 4  C and 20  C, respectively. The numbers near the markers indicate the incubation time. Bar sections representing Altesil data were shaded gray to distinguish them from the target polymers in B) and C). The horizontal position of each marker within a bar section represents the lipid content in the polymer. From left to right, the bar width equals a scale of 0e10 mg ge1 for Altesil and SSP, and 0e30 mg ge1 for LDPE. Note that no lipid data were available for the nine-day incubation of LDPE with fish oil at 4  C. A graphical caption, to assist reading the Figure is provided in S13. The lines refer to discussion in the text and are to guide the eye. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

952

F. Smedes et al. / Chemosphere 186 (2017) 948e957

smaller molecules (biphenyl-d10) reached equilibrium while larger molecules (PCB 78) did not (S11; gray panels). For PRC, a decline in CP/CL from the bottom disk to the immersed disk is inversely related to lipid content as, to attain equilibrium, substance transport is in the opposite direction. Attainment of PRC equilibrium also requires more time than lipid-dosed substances, for reasons discussed below (Section 4.4). It is illustrative to compare the results for the nine-day SSP lipid incubations (S11) with those for 41-day (S12), where the lipid is fully distributed and all plotted substances, except PCB 78 attained full equilibrium. Overall, incubations with Altesil and SSP polymers were close to equilibrium. For the incubations with LDPE, no consistent influence of incubation time or temperature on CP/CL was observed (Fig. 2, panel C). The decline of CP/CL towards the bottom disk for the 41-day, 4  C triolein incubation with LDPE (dashed red line) coincided with a large decline for the lipid content (factor of ~10), indicating nonequilibrium because of inadequate disk contact. For the 35-day, 20  C fish oil incubation, with a lesser lipid decline towards the bottom of the stack (factor of 4), CP/CL in the bottom disk was less deviating from other incubations. CP/CL for the LDPE bottom disk was generally slightly lower than that for the lipid-facing and immersed disks (Fig. 2, panel C). However, this bottom disk CP/CL should also be considered already close to equilibrium as CP/CL for the Altesil disk (situated between bottom and lipid-facing LDPE disk) agreed well with the equilibrium CP/CL for Altesil only (Fig. 2, compare green lines in panel A and C). Equilibrium for the Altetesil disk means that the CP/CL of the lipid facing LDPE disk, also must represent equilibrium. 4.4. PRC distribution As discussed above, lipid-dosed substances generally distributed themselves rapidly through the stack. PRC dosed into the Altesil bottom disk however, often failed to reach equilibrium, particularly at higher MW. Biphenyl-d 10 (lowest MW), for example, showed no marked difference in CP/CL over the disk stack, indicating that equilibrium was attained (S13 A-C). PCB 78 (higher MW), on the other hand, was often still present at elevated concentrations in the dosing disk at the end of the incubation period (S13 D-F). Note that slower equilibration is not a general property of PRC; rather, it is inherent in the incubation design. In finite two phase systems, the sorption capacity ratio between the donor and acceptor phases influences the rate equilibrium is attained, as described by Booij and Tucca (2015). The lipid sorption capacity was 20e40 times higher than the polymer stack and, consequently, lipid-dosed substances require only 2e5% mass transfer to the polymer stack necessary for CP/CL to equal KPL. Furthermore, 95% equilibrium also means that KPL reached 95% of its final value. That this is not the case in a reversed situation what can best be understood from the extreme case where the receiving lipid phase has infinite capacity. Then CL remains zero during PRC release from the dosing polymer, what means that, for CP/CL to approach “KPL “, CP should also become zero, what takes infinite time. In this case 95% “equilibrium” means 95% released, but with 5% left in the polymer, CP is not even close to its final zero value. Consequently, the larger the lipid phase the more time will be needed for CP to reach its final value. A numerical example for the actual phase ratio is given in S14. Despite that for the PRC more time was needed for CP/CL to come near KPL, still sufficient situations where identified where disk and lipid PRC were in close equilibrium and KPL could be obtained. In the cases where dissipation from the dosed Altesil bottom sheet was still ongoing, equivalent CP/CL for the lipid-facing and immersed disks did indicate local equilibrium (S13 D). Similarly, for the SSP incubations high CP/CL PCB 78 occurred in the Altesil dosing

disk (S13 E). For the incubation with suspected poor disk contact (nine-day fish oil-SSP at 4  C), the Altesil bottom disk CP/CL exceeded the maximum scale. In the LDPE incubations (S13 F), the Altesil dosing disk CP/CL for PCB 78, was almost always higher than the ~0.065 observed for the Altesil only incubations (S13 D), confirming that slower transport is associated with lower diffusion rates in LDPE. For all incubations, the slightly lower CP/CL for lipiddosed substances in the bottom disk (indicating dis-equilibrium) always corresponded with markedly higher values for higher MW PRC in the Altesil dosing disk. 4.5. Assessing absorbed lipid effect on equilibrium partitioning To examine the effect of absorbed lipid on KPL, the lipid and stacked polymer equilibration was designed such that lipids would still be absent in the bottom-disk, while equilibrium was attained for lipid-dosed substances and bottom disk dosed PRC. Lipid migration through the stack was quicker than expected, and only the unintended inadequate disk contact in a few incubations provided the desired large differences in lipid content. Such a lipid gradient (2e5.3 mg g1 polymer) over the disk stack was observed for one SSP incubation (fish oil, 9 days at 4  C). A comparison of CP/ CL in bottom, middle, lipid facing, and the immersed disk, showed about equal values for substances of smaller molecular size (VM) throughout all substance groups, including PRC (Fig. 3, Panel A). So, for these substances the difference in lipid contents of a factor 2.6, did not affect CP/CL. Towards substances of VM larger than ~200 Å (VM listed above the bars), CP/CL slightly increased from bottom to the lipid immersed disk. To associate this increase with the parallel increasing lipid content, a similar increase also should have been observed for PRC of equal VM, which, however, showed a (larger) opposite effect, pointing to non-equilibrium. Indeed, the ratio of CP/CL in the bottom and lipid contact disks (RBF) increasingly deviated from unity for VM > 200 Å (S15). While VM is representing DP, for higher molecular weight PCB, PAH, and especially for most OCP the low RBF are evidently connected with slow diffusion. A similar evaluation was made for a LDPE incubation (fish oil, 35-day at 20  C), in which the lipid contents ranged from 3.9 to 15.5 mg g1, and bottom, lipid facing and immersed disks, showed about equal of CP/CL for substances of smaller and medium size PAHs (Fig. 3, panel B). Note that CP/CL in the bottom disk is often closer to that of the immersed disk, while the lipid facing disk shows a consistently higher CP/CL. This was likely caused by an analytical bias, since the CP/CL in the lipid facing disk was also standing out compared to other LDPE incubations (Fig. 2, panel C, blue line). Remarkedly, quite a high number of PRC demonstrated equilibrium, what likely is connected to the placement of the PRC dosed Altesil disk between the bottom and lipid facing LDPE disks. Like for SSP, low RBF for OCP and PCB are associated with high VM (S15), while lower PAH, chlorobenzenes and a number of PRC, are within analytical variability from unity, all with VM < 200 Å. Relatively large RBF differences for the four HCH isomers were observed for SSP and LDPE (S15 purple triangles at VM z185 Å)), while an equal VM was calculated. For these differences no explanation can be given, or it should be that, in spite of equal VM, the different spatial structures that apply, cause a different DP. For substances with a VM < 200 Å absorbed lipid did not measurably affect CP/CL for SSP and LDPE. Assumedly, what is valid for SSP will apply to Altesil too. For slower diffusing substances, we could not provide confirmation on, whether lipid presence in the polymer does, or does not, contribute to polymers ability to absorb substances. Jahnke et al. (2008) assumed that the lipid mass was adhering to the polymer surface and subsequently subtracted the “lipid added” substance concentration from CP, prior to calculating the partition coefficient.

F. Smedes et al. / Chemosphere 186 (2017) 948e957

953

Fig. 3. Comparison polymer lipid concentration ratios (CP/CL, y-axis) obtained for individual disk for the incubation of SSP with fish oil, 9 days at 4  C (left panel) and that of LDPE with fish oil for 35 days at 20  C (right panel). From top to down, the bars represent the CP/CL for the immersed, lipid facing, middle (only for SSP), and bottom disk. For LDPE incubation, the PRC dosing Altesil disk is the middle, empty bar in right panel. The numbers between brackets in the legend represent the lipid content (mg g1) and the numbers above the bars VM (Å) (S1).

The present work does not support such correction, as lipid is dissolved in the polymer (Section 4.1). Adjusting KPL would also imply a reversed correction if during sampling the polymer would absorb lipid. Consequently, since in this work and previously published work (Jahnke and Mayer, 2010), no effect of lipid on substances partitioning was demonstrated, no corrections for lipid presence were applied for calculation of KPL.

4.6. Polymer-lipid partition coefficients (KPL) As CP/CL was unaffected by lipid absorbed by the polymer, all measured values can be used for estimating KPL, providing equilibrium has been obtained. Median and mean KPL was estimated for each polymer-lipid combination at both 4  C and 20  C, while CP/CL for different time periods were merged (excluding those clearly associated with non-equilibrium). For lipid-dosed substances in the

954

F. Smedes et al. / Chemosphere 186 (2017) 948e957

Altesil and SSP incubations, all data were used to calculate median and mean KPL (S16, S17). For lipid-dosed substances in the LDPE incubations, the bottom disk often failed to attain full equilibrium and was thus not included when calculating median and mean KPL (S18). Inclusion of the lipid-facing disk was justified as CP/CL for the Altesil disk (situated just behind the lipid-facing LDPE disk agreed with that for Altesil only. While equilibrium was attained for all PRC of smaller MW (e.g. Biphenyl-d10), dis-equilibrium increased as MW increased. Consequently, several clear cases of non-equilibrium had to be removed before calculating KPL. Results for the immersed disk were leading for this selection, as rapid equilibrium is expected due to the surrounded contact with lipid, even if the lipid itself was not fully equilibrated with the stack. This was supported by the low coefficient of variation (CV) of 0.07 (n ¼ 6) of CP/CL for PRC PCB 204 (highest MW) from immersed disks in the different LDPE incubations. Going from low to high MW, as a rule, CP/CL from a lipid facing disk was not included anymore if differences with the immersed disk became systematic, i.e. negative for three following higher MW PRC. In several cases, this only left data from the immersed disk (S18). The average ratio between median and mean KPL for all substances (both HOC and PRC) equaled 1.00 (s ¼ 0.02) for SSP and Altesil, and 1.02 (s ¼ 0.04) for LDPE. Hence, it was decided to use mean KPL for further evaluation.

with average KPL values of 0.13, 0.08, and 0.05, respectively. For OCP and PCB, the LDPE KPL was one to five times higher than the SSP KPL, while that for PAH was two to sixty times higher, increasing from two-ring to six-ring PAH. Previous studies have also noted an increased affinity of higher MW PAH to LDPE rather than silicones (Rusina et al., 2007; Smedes et al., 2009). The KPL of PCB and PAH for Altesil and SSP decreased with increasing VM while KPL of PAH for LDPE only showed a small decrease or even increased (Fig. 4). Interestingly, for Altesil and SSP, KPL values for PCB of equal VM (and equal chlorine number) were higher with increasing chlorine substitution on ortho-positions. For LDPE, no such correlation was observed. Possibly, substance affinity to silicone polymers is related to spatial molecular structure. This seems also to be the case for HCH that show quite different KPL for the same VM, and a larger difference for silicones than LDPE. All other OCP showed similar levels for both, silicones and LDPE. The steeper KPL decline with increasing VM in silicones, also corresponds with their lower lipid uptake compared to LDPE. Overall, LDPE KPL was less dependent of molecular properties (including KOW; see S23). Equilibrium passive sampling with LDPE will be more biomimetic than silicone, which will be of importance for bioanalytical screening applications (Jin et al., 2013). 5. Applicability of polymer-lipid partitioning 5.1. Lipid-based concentrations across environmental media

4.7. No effect of lipid type and temperature on KPL Reported KPL of organochlorines for SSP were similar for different storage lipids (Jahnke et al., 2008), also observed for lipidwater partitioning for a range of other compounds (Geisler et al., 2012). Moreover, partitioning properties of PAHs were found to be similar between fish, olive, rapeseed and sunflower oil (Mayer et al., 2009). All these findings were confirmed by our data, and extended to more substances, to triolein and to two polymers, Altesil and LDPE. Mean KPL bi-plots for triolein and fish oil all showed scatter around unity (no significant difference) for all polymers at both 4  C and 20  C (S19). Likewise, KPL values were very similar at 4  C and 20  C (S20). This indicates that polymerlipid partitioning is little affected by temperature opposite to polymer-water partitioning, where partition coefficients are dependent on temperature (Jonker et al., 2015). 4.8. Final generic KPL Since KPL values show no dependence on temperature or storage lipid type, all mean values listed (S16-18) were combined for each polymer into generic mean KPL with relative 95% confidence intervals (Appendix A). Note that, while we used KPL in this work, KLP (Jahnke et al., 2008) have also been used as an alternative; hence, we included KLP values alongside those for KPL. A comparison of values for SSP-fish oil (Tobis oil) KPL obtained at 20  C in this work with those measured under identical conditions in the literature (Jahnke et al., 2008) show an average difference of <3%, with a CV of approx. 15% (S21). Generic KPL values for SSP showed no mean difference and even better agreement, with just 10% variation (S22), underlining that any variation caused by temperature or lipid type is lower than experimental variability. It should be noted that the above comparison used data uncorrected for adhering or absorbed lipid (Jahnke et al., 2008) as our results indicate that lipid that was absorbed (Section 4.2) had no effect on partitioning (Section 4.5), and correction would thus introduce bias. Our results indicate that the test substances generally show decreasing affinity to polymers in the order LDPE > Altesil > SSP,

The polymers Altesil, SSP, and LDPE, along with others such as POM (polyoxymethylene), are widely used for passive sampling in surface and pore water. While passive sampling is known for the measurement of very low chemical concentrations in surface and pore water, the HOC concentrations in polymers equilibrated with different environmental media allow for direct comparison between these media as a thermodynamic reference with equal units (Jahnke et al., 2014a; Jahnke et al., 2014b). Moreover, such a reference can be made independent of polymer type if HOC equilibrium concentrations in the polymer sampler (CP) are converted to lipidbased concentrations by dividing with KPL (CL#media). Thus, equilibrium concentrations in polymers exposed to different media (e.g. sediment or water) can be converted to lipid-based concentrations, i.e. CL#sediment and CL#water. For aqueous passive sampling in the kinetic mode, resulting freely dissolved HOC concentrations require transformation to CP at equilibrium through multiplying by their polymer-water partition coefficients, prior to conversion to CL#water. EPS has also been applied through polymer contact with biological tissue samples (e.g. fish), the CP obtained after equilibrium then being converted to a lipid-based concentration CL#tissue (Jahnke et al., 2011). Note that by using the KPL reported here, CL#media, and thus CL#tissue, are based on defined storage lipids, i.e. fish oil/triolein. This circumvents any issues connected with unknown lipid composition in biota, including uncertainties associated with lipid determination. For medium and lipid-rich tissues, CL#tissue agrees well with lipid-normalized tissue concentrations based on exhaustive extractions (CL) (Jahnke et al., 2011). For EPS in €enpa €€ €fer et al., other media, such as sediment (Ma a et al., 2011; Scha €enpa €a € et al., 2011), the obtained equilibrium CP 2015) or water (Ma after conversion to CL#media, represents the concentration in lipid at thermodynamic equilibrium with the media sampled (Jahnke et al., 2014a). Consequently, the CL#media represents the thermodynamic exposure level the organisms are exposed to, and the difference of the actual CL or CL#tissue with CL#media, can be used to identify natural biological variability, bio-magnification, and processes such as metabolism. In relation to the latter, CL#media can represent the exposure level for HOC that are not found in the organisms. Consequently, we advocate CL#media in abiotic media as a

F. Smedes et al. / Chemosphere 186 (2017) 948e957

955

Fig. 4. Lipid-polymer partition coefficients (KPL; kg kg1; y-axis) plotted against their molecular volume (VM,Å3; x-axis) for all three polymers investigated.

Temperature did not measurably affect the KPL and, in agreement with literature, similar KPL were found for different tested lipids. Lipid diffusion coefficients were higher in silicones than LDPE, but the latter showed a four times higher capacity to sorb lipid. This lipid was absorbed by the polymer and not present as surface film. We found no evidence that the lipid absorbed in the polymer affected the KPL, and thus did not correct for it when calculating KPL. A generic KPL can be applied for conversion of HOC polymer equilibrium concentrations to a well-defined lipid basis. Reported PRC KPL may be used for assessing the equilibrium status after in-tissue passive sampling from release of sampler dosed PRCs.

universal and comparable monitoring parameter, ready for use in thermodynamic exposure assessments over a range of matrices, thereby contributing to more balanced environmental quality assessment (Webster et al., 1999). As CL#media are expressed for a defined lipid, they are proportional to chemical activity, thereby providing a simple parameter for multi-media evaluation (Mackay and Arnot, 2011). 5.2. Microplastics The increasing appearance of plastics and microplastics (mp) within the aquatic environment, and the organic pollutants and additives within plastics that may be released and taken up by aquatic organisms (Teuten et al., 2009), is a subject of growing concern. Polyethylene is one of the most heavily used polymers today and it is consequently frequently found in the environment. By utilizing the LDPE KPL from this study, the substance concentration in microplastic LDPE can be converted to a lipid-based concentration at equilibrium with LDPE (CL#LDPEmp). This CL#LDPEmp can then be used to assess the microplastic contamination risk level in relation to CL#media in other environmental compartments, including biota.

Acknowledgment This work was supported through Czech Science Foundation Grant No. GACR 15-16512S “Investigation of accumulation of persistent bioaccumulative toxic organic substances into aquatic organisms”.

Appendix A Generic polymer-lipid partition coefficients (KPL) determined. The lipid-polymer partition coefficients (KLP) represent the reciprocal of KPL and are listed for convenience. All partition coefficients are given in kg kg-1. Confidence limits (95%) are given in relative units (%) and apply to both KPL and KLP.

6. Conclusions This study provided KPL for 78 substances (HOC and PRC) between Altesil, SSP silicones and LDPE and fish oil and triolein.

Altesil™

Hexachloro-1,3-butadiene Pentachlorobenzene Hexachlorobenzene PCB 18 PCB 28 PCB 31 PCB 44 PCB 49 PCB 52 PCB 101 PCB 118 PCB 138 PCB 153

SSP

KPL

KLP

0.538 0.135 0.107 0.131 0.106 0.0952 0.1113 0.0962 0.0965 0.0755 0.0584 0.0616 0.0555

1.86 7.38 9.35 7.62 9.47 10.51 8.99 10.4 10.36 13.24 17.1 16.24 18.02

0,95

CI%

17 10 7 9 10 6 8 6 9 6 7 4 3.6

LDPE

KPL

KLP

0.4 0.0985 0.0849 0.0819 0.067 0.0575 0.0634 0.0576 0.0585 0.0453 0.0329 0.0376 0.0359

2.49 10.15 11.78 12.2 14.9 17.4 15.8 17.4 17.1 22.1 30.4 26.6 27.9

0.95

CI%

28 9 8 8 13 8 9 9 8 9 9 10 10

KPL

KLP

0.95

0.26 0.252 0.309 0.088 0.114 0.094 0.0756 0.0852 0.079 0.0855 0.0952 0.0864 0.097

3.84 3.96 3.24 11.36 8.79 10.7 13.23 11.74 12.66 11.69 10.5 11.58 10.31

17 5 5 8 11 15 7 6 7 7 6 6 6

CI%

(continued on next page)

956

F. Smedes et al. / Chemosphere 186 (2017) 948e957

(continued ) Altesil™

SSP

KPL

KLP

PCB 170 PCB 180 PCB 187 PCB 77 PCB 81 PCB 105 PCB 114 PCB 123 PCB 126 PCB 156 PCB 157 PCB 167 PCB 169 PCB 189 Naphthalene Acenaphthylene Acenaphthene Fluorene Phenanthrene Anthracene Fluoranthene Pyrene Benz[a]anthracene Chrysene Benzo[b]fluoranthene Benzo[k]fluoranthene Benzo[a]pyrene Benzo[ghi]perylene Dibenz[a,h]anthracene Indeno[1,2,3-cd]pyrene cis-Chlordane trans-Chlordane alfa-HCH beta-HCH gamma-HCH delta-HCH 2,40 -DDD 2,40 -DDE 2,40 -DDT 4,40 -DDD 4,40 -DDE 4,40 -DDT Aldrin Dieldrin Endrin Isodrin Telodrin Heptachlor Heptachlor epoxide beta-Endosulfan Endosulfan sulfate

0.051 0.0483 0.0617 0.0539 0.059 0.0589 0.0553 0.062 0.0395 0.0437 0.0495 0.0457 0.0302 0.0355 0.193 0.1388 0.129 0.114 0.0838 0.0983 0.0602 0.0548 0.0418 0.0409 0.0234 0.0311 0.0276 0.0154 0.0204 0.0184 0.0632 0.0699 0.062 0.0076 0.075 0.018 0.0618 0.1155 0.0837 0.0374 0.0818 0.0447 0.176 0.137 0.115 0.135 0.099 0.135 0.097 0.0748 0.0556

19.63 20.7 16.2 18.6 17 17 18.1 16.1 25.3 22.9 20.2 21.9 33.1 28.1 5.17 7.2 7.73 8.77 11.9 10.17 16.6 18.26 23.9 24.4 42.8 32.1 36.2 64.9 49 54.3 15.8 14.3 16.1 131 13.3 55.6 16.2 8.66 12 26.7 12.2 22.4 5.67 7.3 8.7 7.39 10.1 7.41 10.3 13.4 18

Biphenyl-d10 (PRC) PCB 1 (PRC) PCB 2 (PRC) PCB 3 (PRC) PCB 10 (PRC) PCB 14 (PRC) PCB 21 (PRC) PCB 30 (PRC) PCB 50 (PRC) PCB 55 (PRC) PCB 78 (PRC) PCB 104 (PRC) PCB 145 (PRC) PCB 204 (PRC)

0.165 0.172 0.132 0.1238 0.147 0.11 0.1099 0.113 0.1125 0.093 0.0647 0.1023 0.1032 0.091

6.07 5.82 7.6 8.08 6.82 9.09 9.1 8.85 8.89 10.8 15.5 9.77 9.69 10.9

0,95

CI%

LDPE

KPL

KLP

4 5 7 11 10 10 8 11 11 11 8 10 10 11 6 2.4 8 7 9 6 2.7 1.9 6 5 9 10 13 13 12 11 9 9 15 27 12 17 8 8 9 8 9 11 8 14 11 9 11 12 14 10 12

0.0333 0.0322 0.0432 0.0266 0.0361 0.034 0.038 0.0348 0.0203 0.0274 0.029 0.0279 0.017 0.0237 0.135 0.093 0.0982 0.0783 0.0482 0.056 0.033 0.0306 0.018 0.0204 0.0109 0.0132 0.0115 0.0078 0.0068 0.008 0.0519 0.053 0.0343 0.0047 0.0299 0.00754 0.0324 0.0748 0.056 0.0183 0.0561 0.0274 0.149 0.0948 0.076 0.108 0.0818 0.1141 0.0722 0.0356 0.0236

30 31 23.1 37.6 27.7 29.4 26.3 28.8 49.2 36.5 34.5 35.8 58.7 42.1 7.4 10.8 10.18 12.8 20.8 17.7 30.3 32.7 55.6 49 92 75.9 87 128 147 125 19.3 18.9 29.1 214 33.5 133 30.8 13.4 17.9 54.6 17.8 36.5 6.71 10.55 13.1 9.24 12.2 8.77 13.9 28.1 42.3

13 12 12 8 11 7 7 9 8 14 8 9 8 11

0.1118 0.1178 0.0804 0.0797 0.1082 0.0629 0.071 0.0811 0.0769 0.057 0.0307 0.0768 0.075 0.0685

8.95 8.49 12.4 12.54 9.24 15.9 14.09 12.33 13.01 17.6 32.6 13 13.3 14.6

0.95

KPL

KLP

0.95

13 12 11 9 9 11 11 11 13 13 12 13 16 16 22 13 9 9 9 18 8 7 12 11 15 10 27 25 20 21 11 12 11 28 16 9 10 8 8 9 7 12 10 9 19 10 10 8 10 16 13

0.086 0.0995 0.0978 0.0903 0.106 0.0877 0.1096 0.1003 0.0878 0.0952 0.0954 0.1034 0.0911 0.1023 0.191 0.161 0.1853 0.165 0.152 0.188 0.183 0.203 0.1777 0.217 0.233 0.35 0.361 0.37 0.299 0.517 0.059 0.058 0.056 0.0281 0.052 0.0318 0.0484 0.0794 0.0656 0.0331 0.085 0.039 0.121 0.078 0.0745 0.11 0.0752 0.0831 0.0556 0.0449 0.0345

11.63 10.05 10.22 11.07 9.44 11.4 9.12 9.97 11.39 10.51 10.49 9.67 10.97 9.78 5.23 6.2 5.4 6.06 6.58 5.32 5.47 4.93 5.63 4.61 4.28 2.86 2.77 2.7 3.35 1.93 17 17.3 17.9 35.6 19.4 31.5 20.7 12.6 15.2 30.2 11.7 26 8.3 12.8 13.4 9.1 13.3 12 18 22.27 29

8 6 7 7 6 8 6 8 7 9 7 7 8 9 15 16 4 8 7 15 8 13 3.3 5 12 6 7 8 12 11 13 14 21 23 23 28 13 11 9 17 18 32 14 15 12 13 9 10 12 2.8 22

6 8 11 7 9 6 6 7 8 25 7 10 10 8

0.1293 0.124 0.1166 0.1289 0.115 0.1228 0.1135 0.1265 0.116 0.0966 0.0868 0.107 0.103 0.156

7.74 8.05 8.57 7.75 8.7 8.14 8.81 7.91 8.7 10.35 11.5 9.4 9.7 6.42

7 12 8 7 13 6 8 7 12 6 9 18 12 7

CI%

CI%

Supplementary information

References

Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.chemosphere.2017.08.044.

Allan, I.J., Bæk, K., Haugen, T.O., Hawley, K.L., Høgfeldt, A.S., Lillicrap, A.D., 2013. In vivo passive sampling of nonpolar contaminants in brown trout (Salmo trutta). Environ. Sci. Technol. 47, 11660e11667.

F. Smedes et al. / Chemosphere 186 (2017) 948e957 Booij, K., Smedes, F., van Weerlee, E.M., 2002. Spiking of performance reference compounds in low density polyethylene and silicone passive water samplers. Chemosphere 46, 1157e1161. Booij, K., Robinson, C.D., Burgess, R.M., Mayer, P., Roberts, C.A., Ahrens, L., Allan, I.J., Brant, J., Jones, L., Kraus, U.R., Larsen, M.M., Lepom, P., Petersen, J., Profrock, D., Roose, P., Schafer, S., Smedes, F., Tixier, C., Vorkamp, K., Whitehouse, P., 2016. Passive sampling in regulatory chemical monitoring of nonpolar organic compounds in the aquatic environment. Environ. Sci. Technol. 50, 3e17. Booij, K., Tucca, F., 2015. Passive samplers of hydrophobic organic chemicals reach equilibrium faster in the laboratory than in the field. Mar. Pollut. Bull. 98, 365e367. Crank, J., 1957. The Mathematics of Diffusion. University Press, Oxford. Dürig, W., Blakey, I., Grant, S., Chambers, L., Escher, B.I., Weijs, L., Gaus, C., 2016. New polymer passive sampler for sensitive biomonitoring of lipid-rich matrices. Environ. Sci. Technol. Lett. 3, 52e56. Friedman, C.L., Burgess, R.M., Perron, M.M., Cantwell, M.G., Ho, K.T., Lohmann, R., 2009. Comparing polychaete and polyethylene uptake to assess sediment resuspension effects on PCB bioavailability. Environ. Sci. Technol. 43, 2865e2870. Geisler, A., Endo, S., Goss, K., 2012. Partitioning of organic chemicals to storage lipids: elucidating the dependence on fatty acid composition and temperature. Environ. Sci. Technol. 46, 9519e9524. Gilbert, D., Witt, G., Smedes, F., Mayer, P., 2016. Polymers as reference partitioning phase: polymer calibration for an analytically operational approach to quantify multimedia phase partitioning. Anal. Chem. 88, 5818e5826. Huckins, J.N., Petty, J.D., Booij, K., 2006. Monitors of Organic Chemicals in the Environment: Semipermeable Membrane Devices. Springer, New York. Jahnke, A., McLachlan, M.S., Mayer, P., 2008. Equilibrium sampling: partitioning of organochlorine compounds from lipids into polydimethylsiloxane. Chemosphere 73, 1575e1581. Jahnke, A., MacLeod, M., Wickstrom, H., Mayer, P., 2014a. Equilibrium sampling to determine the thermodynamic potential for bioaccumulation of persistent organic pollutants from sediment. Environ. Sci. Technol. 48, 11352e11359. Jahnke, A., Mayer, P., 2010. Do complex matrices modify the sorptive properties of polydimethylsiloxane (PDMS) for non-polar organic chemicals? J. Chromatogr. A 1217, 4765e4770. Jahnke, A., Mayer, P., Adolfsson-Erici, M., McLachlan, M.S., 2011. Equilibrium sampling of environmental pollutants in fish: comparison with lipid-normalized concentrations and homogenization effects on chemical activity. Environ. Toxicol. Chem. 30, 1515e1521. Jahnke, A., Mayer, P., McLachlan, M.S., Wickstrom, H., Gilbert, D., MacLeod, M., 2014b. Silicone passive equilibrium samplers as 'chemometers' in eels and sediments of a Swedish lake. Environ. Sci. Process. Impacts 16, 464e472. Jin, L., Gaus, C., van Mourik, L., Escher, B.I., 2013. Applicability of passive sampling to bioanalytical screening of bioaccumulative chemicals in marine wildlife. Environ. Sci. Technol. 47, 7982e7988. Jonker, M.T., van der Heijden, S.A., Kotte, M., Smedes, F., 2015. Quantifying the effects of temperature and salinity on partitioning of hydrophobic organic

957

chemicals to silicone rubber passive samplers. Environ. Sci. Technol. 49, 6791e6799. Kraaij, R., Mayer, P., Busser, F.J.M., Bolscher, M.V., Seinen, W., Tolls, J., 2003. Measured pore-water concentrations make equilibrium partitioning work - a data analysis. Environ. Sci. Technol. 37, 268e274. Mackay, D., Arnot, J.A., 2011. The application of fugacity and activity to simulating the environmental fate of organic contaminants. J. Chem. Eng. Data 56, 1348e1355. €enpa €€ Ma a, K., Lepp€ anen, M.T., Reichenberg, F., Figueiredo, K., Mayer, P., 2011. Equilibrium sampling of persistent and bioaccumulative compounds in soil and sediment: comparison of two approaches to determine equilibrium partitioning concentrations in lipids. Environ. Sci. Technol. 45, 1041e1047. Mayer, P., Tolls, J., Hermens, J.L.M., Mackay, D., 2003. Equilibrium sampling devices. Environ. Sci. Technol. 37, 184Ae191A. Mayer, P., Torang, L., Glaesner, N., Jonsson, J.A., 2009. Silicone membrane equilibrator: measuring chemical activity of nonpolar chemicals with poly(dimethylsiloxane) microtubes immersed directly in tissue and lipids. Anal. Chem. 81, 1536e1542. Ossiander, L., Reichenberg, F., McLachlan, M.S., Mayer, P., 2008. Immersed solid phase microextraction to measure chemical activity of lipophilic organic contaminants in fatty tissue samples. Chemosphere 71, 1502e1510. Rusina, T.P., Smedes, F., Klanova, J., 2010. Diffusion coefficients of polychlorinated biphenyls and polycyclic aromatic hydrocarbons in polydimethylsiloxane and low-density polyethylene polymers. J. Appl. Polym. Sci. 116, 1803e1810. Rusina, T.P., Smedes, F., Klanova, J., Booij, K., Holoubek, I., 2007. Polymer selection for passive sampling: a comparison of critical properties. Chemosphere 68, 1344e1351. Saleem, M., Asfour, A.A., De Kee, D., 1989. Diffusion of organic penetrants through low density polyethylene (LDPE) films: effect of size and shape of the penetrant molecules. J. Appl. Polym. Sci. 37, 617e625. €fer, S., Antoni, C., Mo € hlenkamp, C., Claus, E., Reifferscheid, G., Heininger, P., Scha Mayer, P., 2015. Equilibrium sampling of polychlorinated biphenyls in River Elbe sediments e linking bioaccumulation in fish to sediment contamination. Chemosphere 138, 856e862. Smedes, F., Geertsma, R.W., van der Zande, T., Booij, K., 2009. Polymer-water partition coefficients of hydrophobic compounds for passive sampling: application of cosolvent models for validation. Environ. Sci. Technol. 43, 7047e7054. €rn, A., Teuten, E.L., Saquing, J.M., Knappe, D.R.U., Barlaz, M.A., Jonsson, S., Bjo Rowland, S.J., Thompson, R.C., Galloway, T.S., Yamashita, R., Ochi, D., Watanuki, Y., Moore, C., Viet, P.H., Tana, T.S., Prudente, M., Boonyatumanond, R., Zakaria, M.P., Akkhavong, K., Ogata, Y., Hirai, H., Iwasa, S., Mizukawa, K., Hagino, Y., Imamura, A., Saha, M., Takada, H., 2009. Transport and release of chemicals from plastics to the environment and to wildlife. Philos. Trans. R. Soc. Lond B Biol. Sci. 364, 2027e2045. Webster, E., Mackay, D., Qiang, K., 1999. Equilibrium lipid partitioning concentrations as a multi-media synoptic indicator of contaminant levels and trends in aquatic ecosystems. J. Gt. Lakes. Res. 25, 318e329.