Quantification of brominated flame retardants in soil

To evaluate the method’s fitness for purpose and the contamination levels in soil from common land uses, six samples were obtained representing various contamination histories: from an open waste burning site, from a location known to have an ash burden, from a chicken run, from an orchard, from arable land and from a raspberry plantation. The first three soil samples were taken from the surface, at a depth of up to 3 cm, in the expectation of collecting substantially contaminated soil. The sampling depth on land under cultivation was 40 cm, and was specified in order to obtain soil contaminated at a level as low as possible, especially to sample soil free of potential atmospheric deposition of burning ash, which may contain BFRs.

Standard solutions for 10 PBDE congeners and 8 nBFRs were used along with their isotopically labelled homologues. The only exceptions were BDE-49 and decabromodiphenyl ethane, as labelled equivalents were not commercially available when the analysis was conducted. Standard solutions of PBDE, hexabromobenzene, pentabromoethylbenzene, 2-ethylhexyl-2,3,4,5-tetrabromobenzoate, 1,2-bis(2,4,6-tribromophenoxy)ethane, bis(2-ethylhexyl) tetrabromophthalate, decabromodiphenyl ethane and ¹³C₆ hexabromobenzene came from Cambridge Isotope Laboratories (Tewksbury, MA, USA). Standard solutions of TBX and PBT were delivered by Tokyo Chemical Industry (Tokyo, Japan) and Merck (Darmstadt, Germany). Isotopically labelled TBX and PBT came from Toronto Research Chemicals (Toronto, ON, Canada), and the labelled 2-ethylhexyl-2,3,4,5-tetrabromobenzoate, 1,2-bis(2,4,6-tribromophenoxy)ethane and bis(2-ethylhexyl) tetrabromophthalate were delivered by Wellington Laboratories (Guelph, ON, Canada). The standard solutions were prepared in isooctane, toluene and n-nonane. Table 1 provides a detailed list of the analytes.

Table 1.

	Name	CAS No.	Structural formula	Acronym	m/z
nBFR	2,3,5,6-Tetrabromo-p-xylene	23488-38-2		TBX	419.7182421.7162
	2,3,4,5,6-Pentabromotoluene	87-83-2		PBT	491.6307499.6282
	2,3,4,5,6-Pentabromoethylbenzene	85-22-3		PBEB	485.6105487.6085
	Hexabromobenzene	87-82-1		HBB	547.5074549.5054
	2-Ethylhexyl 2,3,4,5-tetrabromobenzoate	183658-27-7		EH-TBB	418.6735420.6714
	1,2-Bis(2,4,6-tribromophenoxy)ethane	37853-59-1		BTBPE	356.7948358.7928
	Bis(2-ethylhexyl) tetrabromophthalate	26040-51-7		BEH-TBPH	462.6633464.6613
	1,2-Bis(2,3,4,5,6-pentabromophenyl)ethane	84852-53-9		DBDPE	484.6032486.6000
	2,4,4′-2,4,4′-Tribromodiphenylether	41318-75-6		BDE-28	405.8027407.8002
PBDE	2,2′,4,4′-Tetrabromodiphenyl ether	5436-43-1		BDE-47	483.7132485.7111
	2,2′,4,5′-Tetrabromodiphenyl ether	243982-82-3		BDE-49	483.7132485.7111
	2,2′,4,4′,5-Pentabromodiphenyl ether	60348-60-9		BDE-99	492.9691563.6216
	2,2′,4,4′,6-Pentabromodiphenyl ether	189084-64-8		BDE-100	492.9691563.6216
	2,2,3,4,4,5-Hexabromodiphenyl ether	67888-98-6		BDE-138	481.6980483.6960
	2,2′,4,4′,5,5′-Hexabromodiphenyl ether	68631-49-2		BDE-153	481.6980483.6960
	2,2′,4,4′,5,6′-Hexabromodiphenyl ether	207122-15-4		BDE-154	481.6980483.6960
	2,2′,3,4,4′,5′,6-Heptabromodiphenyl ether	207122-16-5		BDE-183	561.6060563.6040
	Decabromodiphenyl ether	1163-19-5		BDE-209	799.3330801.3310

Copper powder (<75 μm) was purchased from Sigma-Aldrich (St. Louis, MO, USA). Supelco aluminium oxide 150 acidic, Brockmann I, 37% HCl, dichloromethane, n-hexane (Chemsolute) and sulfuric acid (VI) were supplied by Merck. Silica gel was delivered by Fluka (Buchs, Switzerland). Florisil (100–200 mesh) was provided by LGC Standards (Wesel, Germany). Diatomaceous earth was sourced from Restek (Bellefonte, PA, USA). Helium (99.9999%) and nitrogen (99.999%) gases were delivered by Messer (Gumpoldskirchen, Austria). Freshwater sediment ERM-CC537a was purchased from the EU Directorate General Joint Research Center (Geel, Belgium).

About 500 g of each sample was dried at 80°C for 48 h in a UM 200 oven (Memmert, Schwabach, Germany). The dried samples were sieved through a mesh size of 1 mm and next through 0.5 mm. The samples obtained in this way were homogenised using a Turbula 3D Shaker Mixer (WAB, Muttenz, Switzerland). The moisture content was subsequently checked using an MA45C moisture analyser (Sartorius, Göttingen, Germany). If the moisture content exceeded 1%, the entire sample was dried again.

A 10 g mass of sample was mixed with 0.2 g of powdered copper previously activated using 37% HCl and diatomaceous earth. Then the mixture was quantitatively transferred to the extraction cell, and isotopically labelled quantitative standards (¹³C and/or ²D) were added. The analytes were extracted with dichloromethane : n-hexane (1 : 1 v/v) using accelerated solvent extraction (Dionex, Sunnyvale, CA, USA) at 100 bar and 100°C. The extract was concentrated and transferred to a 50 mL centrifuge tube, topped up to 30 mL with dichloromethane : n-hexane, and 8 mL of concentrated H₂SO₄ was added. The mixture was shaken for 20 min and centrifuged, and the supernatant was transferred to a second tube and sulfuric acid (VI) was added again. The supernatant was then evaporated, and the residue was dissolved in 3 mL of n-hexane, purified and fractionated using normal-phase liquid chromatography. The obtained solution was applied to a column containing 4.4 g of 70–230 mesh silica gel impregnated with 44% H₂SO₄ (Supelco, now part of Merck). Some of the analytes were eluted with 20 mL of n-hexane and directly applied to a column containing 3 g of activated acidic alumina (150 mesh, Brockmann I, Merck). The analytes were eluted from the alumina with 20 mL of dichloromethane : n-hexane (50 : 50 v/v) mixture and 10 mL of dichloromethane to a 100 mL round-bottom F1 flask. The remaining analytes were eluted from the silica column with 25 mL of dichloromethane. The eluate was evaporated to dryness, dissolved in n-hexane and then purified using a column containing 1 g of deactivated Florisil (LGC). A round-bottom F1 flask was placed under the column, and then the analytes were eluted with 25 mL of dichloromethane. The obtained eluate was concentrated; the ¹³C₁₂ BDE-139 recovery standard was added and the solution was evaporated to dryness under a gentle stream of nitrogen. The residue was dissolved in 100 μL of toluene, and then 1 μL was analysed using a DFS magnetic sector gas-chromatography high-resolution mass spectrometer (Thermo Fisher Scientific, Waltham, MA, USA).

The analytes were separated using an RTX-1614 15m × 0.25 mm × 0.1 μm capillary chromatography column (Restek). Instrumental analyses were performed separately for PBDE and nBFR using different temperature gradients. The PBDE and nBFR samples were injected using a programmable-temperature vaporisation injector. Its starting temperature was set at 120°C and the sample was injected in the splitless mode for 2 min, then the injector temperature was increased by 5°C·s⁻¹ to 315°C and held for 18 min. The chromatograph oven temperature programme for PBDE determination was as follows: isothermal at 60°C for 1 min, 30°C/min to 220°C held for 3 min, 8°C/min to 240°C held for 2 min, 30°C/min to 300°C, and 2.5°C/min to 320°C held for 12.5 min. The helium flow rate was a constant 1.2 mL/min and the transfer line was maintained at 290°C. The chromatograph oven temperature programme for nBFRs determination was as follows: isothermal at 60°C for 1 min, 30°C/min to 200°C, 3.1°C/min to 210°C held for 2 min, 12°C/min to 240°C held for 1 min, 30°C/min to 275°C held for 2 min and 30°C/min to 315°C held for 5 min. The helium flow rate was once again a constant 1.2 mL/min and the transfer line was maintained at 315°C.

The mass spectrometer was calibrated by constant injection of the mass reference compounds: per-fluorotributylamine (FC-43) for nBFRs and perfluorokerosene (PFK) for PBDEs. The ion-source temperature was maintained at 280°C. The ionisation energy was 47.5 eV and the electron current was 1.0 mA. The quantifications were performed using multiple ion detection at a resolution equal or superior to 10,000 at 10% peak height.

Before routine analysis, the purity of all sorbents and solvents was verified. Throughout the chromatographic process, mass spectrometry resolution was consistently maintained at 10,000. Five standard concentrations were used to calibrate the HRGC-HRMS system. A blank sample and ERM-CC537 freshwater sediment were analysed with every sample series, which was of 12 samples or fewer. The recovery of ¹³C₁₂ PBDE met US EPA 1614 method criteria requirements: 25–150% for BDE-28 to BDE-183, and 20–200% for BDE-209 (42). The recovery of the isotopically labelled nBFRs was in the acceptable range 20–150%. Solvent blank samples were added to each samples series and examined in the same way as matrix samples to control ingredient purity, within-lab interference and potential cross-contamination.

A validation study was performed in terms of the instrumental limit of quantification (iLOQ), limit of quantification (LOQ), limit of detection (LOD), specificity, linearity, accuracy (recovery) and precision (repeatability and within-laboratory reproducibility). These were determined using a blank soil sample spiked at a level near the LOD of the method. Ten samples prepared in the same way were spiked and analysed in two series. The LOD and LOQ were calculated based on standard deviation (SD) multiplied by 3 and 10, respectively. Calibration curves were constructed based on a standard solution with constant labelled analogues and six different concentrations of analytes. The linearity of the method was evaluated by plotting curves from spiked samples at four concentration levels. A comparison of the results of six replicates prepared on the same day at three different concentrations of PBDEs and nBFRs served as the assessment of repeatability. Table 2 shows the PBDE concentrations and Table 3 those of nBFRs. Within-laboratory reproducibility of the procedure was estimated by comparing the results when the same sample was prepared and analysed on three different days. Coefficients of variation were calculated for each analyte concentration level, and the analyte concentrations were calculated based on the native-to-labelled homologue ratio; therefore, their concentrations were not corrected for recovery. The procedure trueness was calculated by the formula (measured level/fortified level) × 100%. The relative standard uncertainty was calculated as the square root of repeatability, within-laboratory reproducibility, and the uncertainty of the certified reference material. The expanded uncertainty (U) was calculated as the ratio of the coverage factor (k = 2).

After initial level evaluation, a soil sample collected from a depth of 40 cm was selected for method validation. The LOQs for individual PBDE congeners ranged from 0.002 to 0.15 ng g⁻¹ dry weight (d.w.), and for nBFRs from 0.002 to 0.061 ng g⁻¹ d.w. The method’s trueness was determined, which, assaying fortified (spiked) samples, ranged from -18% to +12% and –12% to +7% for PBDE congeners and their sum, respectively (Tables 2 and 4). For nBFRs and their sum, it ranged from –15% to +33% and –15% to +12%, respectively (Tables 3 and 4). Determination of the test procedure accuracy (n = 6) from the reference material showed it to range from –8% to +33% for individual PBDE congeners. The variation under repeatability conditions ranged from 1.4% to 12% and from 2.8% to 20.0% for PBDE congeners and nBFRs, respectively. Individual compound groups’ sum contents were compared and in no case had variability exceeding 9% (Table 4). The coefficient of variation under reproducibility conditions ranged from –2.2% to 8.7% for PBDEs and from 4.3% to 16.0% for nBFRs, and did not exceed 8% for the sum of the tested compounds. The mean recoveries of labelled analogues ranged from 38.3% to 116.0% for PBDEs and from 41.5 to 112.0% for nBFRs. A range from 14 to 65% for PBDE congeners and one from 13 to 71% for nBFRs indicated the estimated expanded uncertainty (k = 2) using reference material or analyte-enhanced samples. Validation confirmed that the developed test method was suitable for its intended use. It allowed the analysis of soils for the content of 10 PBDE congeners and 8 nBFRs and their sums. The analytical procedure facilitated analyte quantification in the range from 0.16 to 1,700 ng·g⁻¹ d.w. for ∑ PBDEs\sum {{\rm{PBDEs}}} and 0.072 to 1,130 ng·g⁻¹ d.w. for ∑ nBFRs\sum {{\rm{nBFRs}}} .

Table 4.

	∑ PBDE\sum {{\rm{PBDE}}} without BDE-209	∑ PBDE\sum {{\rm{PBDE}}}	∑ nBFR\sum {{\rm{nBFR}}} without DBDPE	∑ nBFR\sum {{\rm{nBFR}}}	∑ BFR\sum {{\rm{BFR}}}
LOD (ng/g. d.m.)	0.014	0.049	0.022	0.022	0.057
LOQ (ng/g d.m.)	0.045	0.162	0.072	0.072	0.192
Level 1
Fortification (ng/g. d.m.)	24	34	1.3	2.3	36
Repeatability
Trueness (%)	9	7	5	6	7
CV (%)	6.2	5.4	1.2	5.0	5.1
Reproducibility
Trueness (%)	–1	1	3	2	1
CV (%)	0.7	0.9	3.6	7.9	5.1
Uncertainty (%)	27	23	12	21	23
Level 2
Fortification (ng/g. d.m.)	48	68	25	45	113
Trueness (%)	5	4	–5	12	7
CV (%)	3.6	4.2	3.6	2.9	3.0
Uncertainty (%)	16	16	15	27	19
Level 3
Fortification (ng/g. d.m.)	1200	1700	625	1130	2830
Trueness (%)	–13	–12	–9	–15	–13
CV (%)	0.9	2.0	4.5	8.8	2.6
Uncertainty (%)	26	26	23	40	28
R2	0.99995	0.99996	1.00000	0.99986	0.99993
β0	0.87	0.87	0.91	0.85	0.86
β1	4.58	6.01	0.41	4.12	10.24

∑ PBDE\sum {{\rm{PBDE}}} – sum of polybrominated diphenyl ethers; ∑ nBFR\sum {{\rm{nBFR}}} – sum of novel brominated flame retardants; DBDPE – 1,2-Bis(2,3,4,5,6-pentabromophenyl)ethane; ∑ BFR\sum {{\rm{BFR}}} – sum of brominated flame retardants; LOD – limit of detection; LOQ – limit of quantification; Level 1/2/3 – enrichment levels; d.m. – dry matter; CV – coefficient of variation; R² – coefficient of determination; β₀ and β₁ – are dimensional parameter vector in the equation y=β₀ + β₁x

Polybrominated diphenyl ether content ranged from 0.23 ng·g⁻¹ dry matter (d.m.) to 485.7 ng·g⁻¹ d.m. in the investigated soil samples. The concentration of nBFRs was at markedly lower levels and was found in the range of 0.11–0.81 ng·g⁻¹ d.m. As expected, soil samples with limited exposure to atmospheric precipitation, e.g. soil collected from a depth of 40 cm or from ground under cover, were characterised by the lowest concentrations of BFRs.