1

SUPPLEMENTARY INFORMATION

Analytical procedure

All glass materials used in the analytical procedure and instruments used in the sampling campaigns were previously rinsed with acetone and n-hexane (Sigma Aldrich, Steinheim, Germany) to avoid any external source of contamination. Frozen soil samples were lyophilised (Instrument Edward 1001) for 24 h at a final temperature of 5 °C, and powdered soils were conserved in closed glass pots that had been previously rinsed with acetone and n-hexane.

Around 20 g of dry soil were placed in cellulose thimbles (25 x 100 mm, Whatman, England) and then extracted using a Soxhlet apparatus (Falc Instruments, Lurano, Italy) for 12 h using 100 mL of acetone/n-hexane (1:1 v/v). Between two extractions, Soxhlet apparatus was washed for 8 h, using the same solvent mixture. PUF-disks were not lyophilised because already dry, and extracted with an excess of Na2SO4 anhydrous directly in the Soxhlet apparatus without cellulose thimbles. 20-mL aliquot of p,p’–DDE D8 (Dr. Ehrenstrofer, Ausburg, Germany) was added to the thimbles and to PUF disks as a recovery standard before extraction to monitor analyte losses. The extracts were concentrated in a rotary evaporator (RV 06-LR, IKA, Staufen, Germany) to a small volume, transferred in conic glass flask and resuspended in 3 mL of n-hexane and subjected to overnight digestion with 6 mL of 95-97% sulphuric acid to oxidise and dehydrate the organic matter. The supernatant was separated from the acid solution, brought to a volume of 1 mL and purified using a two-layer column of 10 g of Silica gel with a 70-230 mesh (activated overnight at 130 °C and subsequently deactivated with water at 5% w/w) and 10 g of Florisil (activated for 16 h at 650 °C), both purchased from Sigma Aldrich, Steinheim, Germany. Columns were first washed with n-hexane/acetone/dichloromethane (8:1:1 v/v/v), and extracts were then loaded at the top. Elution was first performed with 50 mL of n-hexane and then with 50 mL of n-hexane/dichloromethane (1:1 v/v). A 1-mL aliquot of isooctane was added. The purified extracts were then concentrated to 1 mL using a rotary evaporator and gentle nitrogen flow. All solvents used were from Sigma Aldrich, Steinheim, Germany.

GC-MS-MS analysis

An aliquot of 2 L of the cleanup extract was injected into a gas chromatograph (Trace GC Ultra, Thermo-Electron, Texas Instruments, USA) equipped with a programmed temperature vaporiser injector (PTV) and coupled to a PolarisQ Ion Trap mass spectrometer that was equipped with an AS 2000 auto sampler (Thermo Electron). The chromatographic separation was performed with an Rtx-5MS capillary column with the following dimensions: 30 m length, 0.25 mm I.D., 0.25 mm film thickness (Restek, Bellafonte, PA, USA). The PCB congeners 18, 28+31, 52, 44, 101, 149, 118, 153, 138, 180, 170, 194 and 209 were determined by EI MS/MS-SIM mode. Chromatographic conditions were as follows: PTV in splitless mode; carrier gas helium at 1 mL min-1; surge pressure of 100 kPa; initial oven temperature starting at 100°C and maintained for 1 min, then ramped to 180°C (no hold time) at 20°C min-1, to 200°C (no hold time) at 1.5°C min-1, to 250°C (no hold time) at 3°C min-1 and finally to 300°C (held 5 min) at 30°C min-1. Mass spectrometry conditions were as follows: EI mode with standard electron energy of 70 eV; transfer line was maintained at 270°C, damping gas at 1 mL min-1 and ion source at 250°C. Details of the monitored ions for PCB detection are available from the authors.

Quality assurance (QA), quality control (QC) and limits of quantification (LOQs) of the analyses

PCB quantification in the final purified extract was performed using an external standard method with a calibration line from 0.05 ng mL-1 to 10 ng mL-1 for each congener. Regression coefficients of the fitting lines were all above 0.98. Each standard and sample were injected twice to evaluate instrumental repeatability. The mean instrumental response (mean of the two automatically integrated peak areas) was used for quantification. Peaks were integrated automatically using the dedicated Xcalibur software (Thermo Electron Corporation, San Jose, California). The analytical repeatability was evaluated by analysing three sub-samples, resulting in a value below 20%. Accuracy of the method was evaluated externally (among laboratories) by analysing a standard reference material (NIST, Gaithersburg, MD, USA) and internally (within the laboratory) by means of recovery analyses. The results obtained for the reference material were within the range of certified values specified for the target compounds. For recovery analyses, three sub-samples of soil and three PUF-disks, spiked with three increasing amounts of PCB standard (1, 2 and 5 ng of each PCB congener), were analysed together with an un-spiked sub-sample of soil and an un-spiked PUF-disk. Mean recoveries were above 80%, with a maximum variability of 25%.

Samples and blanks (one for 5 samples) were spiked with the recovery standard p,p’ –DDE-D8. DDE-D8 recoveries in soil samples ranged between 81% and 118% with mean of 99.5% ± 11% (standard deviations of the DDE-D8 recoveries in all analysed samples). DDE-D8 recoveries in PUF-disk samples ranged between 82% and 118% with just one exception of 52%. Because PUF-disk analyses were not repeatable, this analyseis was not repeated. The same isotope labelled DDE (DDE-D8), used as recovery standard in each sample, was also used for recovery analyses of each PCB congener (internal standard method) in addition to the external standard procedure (described above). During this trial, recoveries calculated using DDE-D8 as internal standard were consistent to those calculated by external standard method, allowing the use of DDE-D8 as a representative recovery standard for all analysed PCBs in each sample.

All blanks were treated to the same analytical procedure as the samples, and no detectable peaks, corresponding to the analysed PCB congeners, were present in the blanks above the standard thresholds for peak integration (Xcalibur software, Thermo Electron Corporation, San Jose, California). For this reason, LODs were not calculated by the “mean concentration in blank + 10-fold the standard deviation of blanks” but based on the lowest injected standard (quantification within the range of the calibration line). Considering the lowest injected standard (0.05 ng mL-1) and the minimum extracted weight of 10 g of dry soil, the limits of quantification (LOQs) were 0.005 ng g-1 d.w. for soil and 0.05 ng PUF-1 for PUF disk samples.

REFERENCES SI

Dunnivant FM, Elzerman AW, Jurs PC, Hasan MN (1992) Quantitative Structure-Property Relationships for aqueous solubilities and Henry’s law constants of polychlorinated biphenyls. Environ Sci Technol 26:1567-1573.

Hawker DW, Connell DW (1988) Octanol-water partition coefficients of polychlorinated biphenyl congeners. Environ Sci Technol 22:382-387.

Li N, Wania F, Lei YD, Daly GL (2003) A comprehensive and critical compilation, evaluation, and selection of physical–chemical property data for selected polychlorinated biphenyls. J Chem Eng Data 50:742-768.

Murphy TJ, Mullin MD, Meyer JA (1987). Equilibration of polychlorinated biphenyls and toxaphene with air and water. Environ Sci Technol 21:155-162.

Shoeib M, Harner T (2002) Characterization and comparison of three passive air samplers for persistent organic pollutants. Environ Sci Technol 36:4142-4151.