The ESDs consist of PDMS layers deposited on the internal wall of sample glass vials. The ESDs are equivalent in all respects apart from the thickness of the PDMS. This in vial-sample preparation minimises sample handling, and parallel application of the different thicknesses allows validation of the equilibrium sampling to measure chemical activity. The procedure consists of three parts:

1. Vials with 3–12 μm thin layers of PDMS are equilibrated with samples of the investigated soil.

2. Solvent extracts of the PDMS are analysed for the content of PAHs to determine C_PDMS.

3. The chemical activity is found as the product of C_PDMS and an activity coefficient (Equation 1).

After sampling the soil in PDMS coated vials, the quantities of PAHs (n_PAH) in the PDMS were extracted to methanol, measured by HPLC, and plotted against the PDMS volumes (Figure 3). Linear regressions (least square fit, forced [0;0]) with r² exceeding 0.90 were obtained for 7 PAHs. The graphs clearly demonstrate that n_PAH is proportional to V_PDMS, which is consistent with artefact-free equilibrium sampling.

Equilibrium partitioning concentrations in the PDMS (C_PDMS, Table 1) were calculated as the arithmetic mean of the 9 coated vials. The relative standard error of the arithmetic mean ranged from 2 to 4 %, a very high precision. Additionally, C_PDMS was also determined as the slope of the linear regression. These slope estimates and the arithmetic means hardly differed (2–4 %), and both calculation principles are equally suitable. A high accuracy in the C_PDMS measurements is expected, since the potential for systematic errors associated with the input parameters (n_PAH, V_MeOH, V_PDMS) are limited.

C_PDMS is proportional to the freely dissolved concentration (C_free), chemical activity and also fugacity 6. C_free can be determined with the analyte specific PDMS-water partition ratio K_PDMS/aq:

C f r e e = C P D M S K P D M S / a q MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacPC6xNi=xI8qiVKYPFjYdHaVhbbf9v8qqaqFr0xc9vqFj0dXdbba91qpepeI8k8fiI+fsY=rqGqVepae9pg0db9vqaiVgFr0xfr=xfr=xc9adbaqaaeGaciGaaiaabeqaaeqabiWaaaGcbaGaem4qam0aaSbaaSqaaiabcAgaMjabckhaYjabcwgaLjabcwgaLbqabaGccqGH9aqpjuaGdaWcaaqaaiabdoeadnaaBaaabaGaeiiuaaLaeiiraqKaeiyta0Kaei4uamfabeaaaeaacqWGlbWsdaWgaaqaaiabccfaqjabcseaejabc2eanjabcofatjabc+caViabcggaHjabcghaXbqabaaaaaaa@438A@

This calibration towards C_free has already been covered in several publications (e.g. 17223132), and the present paper focuses on the determination of chemical activities from C_PDMS (Equation 1).

The activity coefficients in Table 3 were determined at solubility (Equation 3). At this limit, the chemical activity of the solute PAH in the saturated solvent is usually equal to the activity of the PAH in its pure solid crystal state, a_xstal. In turn, a_xstal is exponentially related to the free energy difference between the compound's liquid and solid states at the given temperature. This difference can be determined from the melting point temperature, heat of fusion and heat capacity as detailed elsewhere 3334. However, a_xstal were instead estimated through Equation 4, for reasons of consistency. For similar reasons, the activity coefficients were based on solubility in methanol and the partition ratio to the PDMS. First, S_(s)MeOH and K_PDMS/MeOH could be determined with sufficient precision (Table 3 and Table A in Additional File 1). Second, S_(s)MeOH for several PAHs are published and can be used to validate experimental accuracy (Figure A in Additional File 1). Third, the use of partitioning solutions and solubilities in the calibration of ESDs is consistent with current practice 51115172235 and carries certain advantages.

The in vial-ESD format and A/V-approach has unique and important features. Most important is the integrated QA/QC of the measurements. This extends the applicability towards very difficult or very dirty samples that might cause problems during equilibration or analysis due to their composition. Examples of troublesome sample components include soot particles with a high PAH content 25 and (micro)droplets of non-aqueous phase liquids (NAPLs) that may foul the PDMS surface. But, when using the in vial-ESD format, equilibrium sampling flaws can be inspected by plotting n_a against V_PDMS. This integrated validation makes the method reliable at its applicability limit, whereas other equilibrium sampling methods require a greater margin of safety. Second, the integration of thin polymer layers within sample vials 27 reduces the risk of measurement error. Handling procedures are minimised as sample material can be secured, transported, stored and equilibrated, under seal and in a single container. The subsequent liquid extract of the PDMS makes the method compatible with many standard analytical procedures. For example, addition of internal standards, clean-up and/or pre-concentration steps can be included in future equilibrium sampling method developments. The in vial-ESDs are both practical and versatile. Third, the design of the devices is also adaptable. The ESD dimensions can be tuned towards faster equilibration by reducing the polymer coating thickness 27. Alternatively, detection limits can be lowered by increasing the V_PDMS to sample a greater amount of analyte for detection. All of these advantages make ESDs with multiple A/V ratios attractive.

3–12 μm layers of PDMS were prepared in 20 mL glass vials by dispersion coating in the following manner. The appropriate mass of medical adhesive, Silastic^® PDMS Silicone Type A (Dow Corning, Seneffe, BE; density 1.15 kg/L 28) paste was dispersed in 10 times its volume in n-pentane by wrist shaking and sonication. The dispersion was filtered (glass fiber, 0.7 μm nominal, Pall Corp., MI) and diluted in a geometric series. All polymer dispersions were used at the day of preparation and should be prepared using disposable glassware and materials to avoid inadvertent formation of insoluble polymer layers.

After careful tare on an analytical balance (Sartorius BP 210S, d = 0.1 mg), the vials to be coated were placed horizontally on a roller-mixer (Stuart sci., Stone, UK). While rotated (33 rpm) 3.0 mL of the appropriate dispersion was added to each vial. The solvent was evaporated under a gentle stream of nitrogen and the vials were left to polymerise for 72 h in a fume hood. Siloxane oligomers were removed from the cross-linked PDMS by three extractions with ethyl acetate (3 mL, 10 min vigorous shaking). The vials were again balanced to determine the mass of PDMS now covering their inner sidewall surface.

PDMS-coated vials with different coating thicknesses were filled with 18–20 g of the test soil (n = 9). 10 mL aqueous solution of sodium azide (0.5 g/L) was added to each vial to inhibit microbial activity and to create a soil suspension for the sampling. Vials were capped with metal lined screw caps and placed horizontally in a customised holder rotated (6 rpm) on a roller-mixer for 122 h.

Soil content was removed, the vials were rinsed with a little distilled water on a whirly mixer before being wiped internally with lint-free tissue. To each vial was then added 1.00 mL methanol. The vials were again capped and rotated for 12 h. The methanol extracts were collected and stored in freezer (-20°C) until analysed for PAHs.

Methanol extracts of PDMS were analysed for PAHs by HPLC-fluorescence detection (Agilent 1100 system with G1321A FLD (Ex. 260 nm; Em. 350, 420, 440 and 500 nm). Separation column: CP-Ecospher 4 PAH (Varian Inc., Palo Alto, CA) operated at 0.5 ml/min (28°C, 10 μL injection); Mobile phase: methanol, SUPER-Q treated water (Millipore, MA); Gradient (in %methanol, by weight): t = 0 min 80%; t = 5–30 min linear gradient 80–100%; t = 30–45 min 100%. Quantification was accomplished by a five-point external standard curve. Extract analysis was carried out within 2 weeks after sampling.

Solubilities were measured by placing excess PAH crystals together with methanol in PTFE-sealed glass vials and equilibrating these in a thermostated water bath at 25.0 ± 0.1°C for at least 3 days. Attainment of the solubility limit was verified by repetitive measurements on consecutive days. Aliquots of the saturated methanol solutions were passed through 0.2 μm PTFE-filters into volumetric flasks and quantitatively diluted with methanol to reach suitable concentrations before analysis by the HPLC-FLD method described above.

Partition ratios (K_PDMS/MeOH) between Silastic^® PDMS polymer and methanol was measured in a separate experiment. First, PDMS sheets were prepared by filling a polyethylene mold (0.25 mm high) with Silastic^® paste. The top surface was pressed even, and covered with wet tissue and a glass plate. After one week, the cross-linked PDMS was cut into sheets weighing approximately 1 g, Soxhlet extracted (ethyl acetate, 100 h) and then washed with methanol. A clean sheet and an excess of methanol solution containing all the analyte PAHs (approx. 25 μg/L) were placed in a sealed amber glass container. After equilibration by shaking for > 1 week at 20°C, the methanol and PDMS were separately taken for extraction and analysis of the PAH contents by GC-MS as described in 37. The respective, volume based, concentrations in the Silastic^® PDMS sheet and the methanol solution were divided to calculate the partition ratios.

A risk assessment of the pollution in the test material soil is warranted by the present soil quality criterion for PAHs in Denmark. The legislation dictates that the sum of fluoranthene, benzo [a]pyrene, benzo [k]fluoranthene, and indeno [1,2,3-cd]pyrene may not exceed 4 mg/kg dry soil 38.

Several concerns regarding the risk posed by the PAHs in the polluted soil can be assessed with the chemical activities reported in Table 1. First, a_soil quantifies the chemical pollutant's potential for partitioning and distribution to other materials in contact with the soil. This makes the activity relevant because it is first after release and transfer to a susceptible target that a PAH may cause detriment of health or effects on the environment. Second, a_soil provides a measure for comparison of chemical contamination levels. The activity reflects a chemical's affinity (e.g. partial molar free energy of sorption) to the soil material. Soils are dynamic, heterogeneous and unpredictable geosorbents. Therefore, C_soil is harder to interpret and compare than are concentrations in a well-defined, stable and homogenous phase such as PDMS. In consequence, equilibrium sampling measurements are useful because the results can easily be compared to the levels in other measured soils, in air, water or biological tissue, as well as to effect levels determined in toxicity studies, whenever such are available.

For example, comparison of different exposure routes may be part of health risk assessments. Cases may include, for instance, risk of systemic up-take after ingestion of polluted soil and inhalation of urban air. To illustrate, Figure 4 depicts chemical activities of several priority pollutant PAHs, both in the tested soil and in the gas phase of an urban atmosphere 39(Additional File 1). Comparatively, PAHs in the air seem to have greater chemical potential for diffusion through biological membranes, such as gastrointestinal or alveolar epithelium.

PAHs of ≥98% purity were obtained from Sigma-Aldrich (Copenhagen, DK) or Cambridge Isotope Lab. (Andover, MA).

P.a. grade n-pentane, ethyl acetate, sodium azide and HPLC-grade methanol were obtained from Merck (Darmstadt, DE).

The PAH contaminated soil were excavated from a former manufactured gas plant site (Gaswerk Tiefstark, Klosterwall 2, Freie Hansestadt Hamburg, DE) and then compost treated by A/S Bioteknisk Jordrens (Esbjerg, DK) prior to this study. It contained 14 % water, 5–7 % organic carbon, 11–16 % silt, 3–4 % clay and weathered coal-tar, petroleum residues, soot and ash.

The bioremediation of the soil had removed a fraction of the PAHs; whereas those associated with the black carbon 40 likely remained. The residual total concentration of the 16 EPA PAHs was 120 mg/kg_soil. This test material was selected on account of the PAH-containing soot that is claimed to complicate equilibrium sampling 25. Prior to sampling, the soil was sieved to 2 mm (steel mesh) and manually homogenized.

The qualification criteria for PAHs extracted from the ESDs included retention times within 5% of a known sample and correct fluorescence emission wavelength(s). Signal integration was performed with HP Chemstation software (A.06.03, Agilent Tech.) and corrected by hand as necessary. About 7% standards and solvent blanks were included in the analytical sample sequences. ESD blanks were considered unnecessary since positive blank values would be eliminated during the equilibration. The measured results were tested by least squares linear regression of n_PAH on V_PDMS, forced through the origin. An r² > 0.90 goodness of fit, was set as a necessary but not sufficient equilibrium sampling criterion. One outlier in the data set for anthracene was defined as the measured value of n_ANT differed by a factor > 2 from the predicted (n = 9). Equilibrium concentrations in PDMS (C_PDMS, Table 1) were calculated as the arithmetic mean of the measurements (n = 9-8).

For each PAH in Table 3, the activity coefficient in the Silastic^® PDMS (γ_PDMS) that coated the vials was determined from partitioning and solubility data:

γ P D M S = a x s t a l S ( s ) M e O H K P D M S / M e O H , MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacPC6xNi=xI8qiVKYPFjYdHaVhbbf9v8qqaqFr0xc9vqFj0dXdbba91qpepeI8k8fiI+fsY=rqGqVepae9pg0db9vqaiVgFr0xfr=xfr=xc9adbaqaaeGaciGaaiaabeqaaeqabiWaaaGcbaGaeq4SdC2aaSbaaSqaaiabccfaqjabcseaejabc2eanjabcofatbqabaGccqGH9aqpjuaGdaWcaaqaaiabdggaHnaaBaaabaGaeiiEaGNaei4CamNaeiiDaqNaeiyyaeMaeiiBaWgabeaaaeaacqGGtbWudaWgaaqaaiabcIcaOiabcohaZjabcMcaPiabc2eanjabcwgaLjabc+eapjabcIeaibqabaGaem4saS0aaSbaaeaacqGGqbaucqGGebarcqGGnbqtcqGGtbWucqGGVaWlcqGGnbqtcqGGLbqzcqGGpbWtcqGGibasaeqaaaaacqGGSaalaaa@51F9@

where a_xstal is the activity of the PAH in its crystal state, S_(s)MeOH the solubility in methanol and K_PDMS/MeOH is the Silastic^® PDMS-methanol partition ratio.

The PAHs' activities in their pure solid crystals, also known as fugacity ratios 41, at 25°C (T = 298 K) were estimated from melting point temperatures (T_m, K), assuming their entropy of melting to be 56 Jmole^-1K^-1 (i.e. Walden's rule) as suggested by Yalkowsky et al. 42:

a x s t a l = exp ⁡ ( 6.8 [ 1 − T m T ] ) MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacPC6xNi=xI8qiVKYPFjYdHaVhbbf9v8qqaqFr0xc9vqFj0dXdbba91qpepeI8k8fiI+fsY=rqGqVepae9pg0db9vqaiVgFr0xfr=xfr=xc9adbaqaaeGaciGaaiaabeqaaeqabiWaaaGcbaGaemyyae2aaSbaaSqaaiabcIha4jabcohaZjabcsha0jabcggaHjabcYgaSbqabaGccqGH9aqpcyGGLbqzcqGG4baEcqGGWbaCcqGGOaakcqaI2aGncqGGUaGlcqaI4aaodaWadaqaaiabigdaXiabgkHiTKqbaoaalaaabaGaeiivaq1aaSbaaeaacqGGTbqBaeqaaaqaaiabcsfaubaaaOGaay5waiaaw2faaiabcMcaPaaa@46C6@