Ultra-trace determination of Persistent Organic Pollutants in Arctic ice using stir barsorptive extraction and gas chromatography coupled to massspectrometry
S. Lacorte1*,J. Quintana2, R. Tauler1, F. Ventura2, A.Tovar-Sánchez3 and C. M. Duarte3
1 Department of Environmental Chemistry, IDAEA-CSIC, Jordi Girona 18-26, 08034 Barcelona, Catalonia, Spain.
2 Aigües de Barcelona, Av. General Batet 4, 08028 Barcelona, Catalonia, Spain.
3Department of Global Change Research, IMEDEA-CSIC-UIB, Miquel Marqués 21, 07190 Esporles, Mallorca, Spain.
* corresponding author:
Abstract
The present study presents the optimization and application of an analytical method based on the use of stir bar sorptive extraction (SBSE) gas chromatography coupled to mass spectrometry (GC-MS) for the ultratrace analysis of POPs (Persistent Organic Pollutants) in Arctic ice. In a first step, the mass-spectrometry conditions were optimized to quantify 48 compounds (polycyclic aromatic hydrocarbons, brominated diphenyl ethers, chlorinated biphenyls, and organochlorinated pesticides) at the low pg/L level. In a second step, the performance of this analytical method was evaluated to determine POPsin Arctic cores collected during anoceanographic campaign. Using a calibration range from 1 to 1800 pg/L and by adjusting acquisition parameters, limits of detection at the 0.1-99 and 102-891 pg/L for organohalogenated compounds and polycyclic aromatic hydrocarbons, respectively, were obtained by extracting 200 mL of unfilteredice-water. α-hexachlorocyclohexane, DDTs,chlorinated biphenyl congeners 28, 101 and 118 and brominated diphenyl ethers congeners 47 and 99 were detected in ice cores at levels between 0.5 to 258 pg/L. We emphasise the advantages and disadvantages of in situ SBSE in comparison with traditional extraction techniques used to analyze POPs in ice.
1. Introduction
Because of long range transport and dispersion throughout the environment, Persistent Organic Pollutants (POPs) have been detected in remote areas such as the Arctic[1]and Antarctic[2] ecosystems. The main sources of these compounds in polar environments are atmospheric transport and continental run-off. Although the concentrations encountered in ice-water are at the low pg/L level[3, 4], there is evidence that these compounds are released upon snow and ice melt [5] and are accumulated in apical predators in polar food webs, such as seals, whales, polar bears [6] and humans [7]. There are reasons for concern on the potential risks they may pose for fauna and, ultimately, for human health.
Yet, data on contaminant loads in Arctic ice is very scarce. To date, resolving POP levels in Arctic ice is particularly important due to accelerated rates of ice melting [8], which releases the POPs trapped in ice into the surrounding waters [9].
Considering the low concentration of POPs in ice samples, analytical methods should be sufficiently sensitive and selective to meet quantification limits at the pg/L level. The measurement of such low levels is analytically complex, especially when performed in the field (e.g. on board of an oceanographic vessel) where sampling, processing and storage require a rigorous analytical control to reach the required sensitivity and at the same time avoid any external source of contamination [4].Traditionally, large ice volumes are extracted to detect pg/L concentration. In a very early study performed in 1983, Tanabe et al. extracted 200-1200 Lof ice melt water using an Amberlite XAD-2 resin column which were reconstituted in 100 µL of hexane and analyzed by gas chromatography coupled to mass spectrometry (GC-MS) using a packed column, measuring concentrations of 1500-4900 pg/L in Antarctic ice and snow [10]. Donald et al. melted 20 L of water equivalents of snow and ice and liquid-liquid extracted (LLE) target compounds, yielding limits of detection (LODs) of 2 pg/L for organochlorinated (OC) compounds [6]. Villa et al. (2003) LLE-extracted 1-5 L of ice water to obtain limits of detection of 0.25-1 ng/L for several OCpesticides [11]. Gustafsson et al. (2005) collected 200 L of ice which were melted on an ice-melter and using LLE,levels of 0.005 to 0.44 pg/L of polychlorinated biphenyls (PCBs) were detectedin Arctic ice and snow [4].To avoid the use of large solvent volumes in LLE, Solid Phase Extraction (SPE) base techniques have been deployed. Using C18 cartridges, 1-6 L of snow were preconcentrated and yieldedLODs between 0.63 and 27 pg/L for polybromo diphenyl ethers (PBDEs) [12]. Speedisks were also evaluated to process high sample volumes (up to 50 L) without the need for sample filtration and provided mean recoveries of 68% and LODs between 0.2 and 124.8 pg/L for 75 organic compounds in snow samples [13].
Progress in sample-prep techniquesand technological improvements, especially in the analytical instrumentation has led to minimize extracted sample volumes and solvents or either use solventless analytical procedures. Semi-permeable membrane devices (SPMDs) have been identified as an alternative to extract POPs in snow [14]and provide an integrated measure of freely dissolved contaminants,reaching LOD of 0.2-0.4 ng after exposure of SPMD for 12-20 days. Solid Phase Microextraction (SPME) has been used to determine OC pesticides in Himalayan ice using only 35 mL of water [15]. This technique is solvent-free and is characterized by the fact that analytes are extracted on a fiber which is then injected in a GC-MS, minimizing sample manipulation and increasing in sensitivity since all extracted analytes are detected. The outcome of Stir Bar Sorptive Extraction (SBSE) followed by thermal desorption and GC-MS further improves the method sensitivity since it has higher capacity than SPME. By using 100 mL of water, all preconcentrated compounds are detected, lowering the LODs to ng/L for PAHs, PCBs and pesticides [16]. This technique has additional advantages such as minimal sample manipulation, implying minimal external contamination risk and isalso solventless. This recent development was timely, as its application could be instrumental in achieving an increase of knowledge on pollutant loads in the polar ice sought as one of the aims of the International Polar Year (IPY 2007-08).
The aim of the present study was to develop an ultra sensitive methodology based in SBSE-GC-MS to identify a large number of POPs in Arctic ice cores collected during the 2007 ATOS oceanographic campaign. First, in situ extraction conditions (considering boat movement and limited laboratory facilities) were optimized in an attempt to: (i) reduce the amount of ice extracted in comparison to state of the art methods; (ii) avoid the burden to store and transport large water volumes and(iii) increase the sample throughput.Second, the analytical conditions were carefully optimized to reach the 0.1 pg/L sensitivity for 4 chemical families of POPs. We report here the new approach and its performance, as well as its utility to determine the levels of contaminants in Arctic cores, which are then compared with values reported in the literature using other analytical methods.
2. Experimental
2.1. Chemicals and Reagents
Compounds analyzed are indicated in Table 1. Sixteen Environmental Protection Agency (EPA) PAHs were purchased from AccuStandard (New Haven, CT, USA) as a mix solution at 200 mg/L in methanol. The internal standard solution used to quantify these compounds contained naphthalene-d8, acenaphthylene-d8, acenaphthene-d10, phenanthrene-d10, chrysene-d12and perylene-d12 at 2 mg/L in methanol (Supelco, Bellefonte, USA). PCBs were purchased as a mix solution at 10 mg/L in iso-octane and OCpesticidesat 100 mg/L in methanol (Dr. Ehrenstorfer, Augsburg, Germany). The main PBDE congeners studied were bromodiphenyl ether (BDE)28, BDE 47, BDE 100, BDE 99, BDE 154, BDE 153 and BDE 183, purchased at 1 µg/mL in nonane(Cambridge Isotope Laboratories, Inc., Andover, MA, USA). BDE 209 was not analyzed since it implied a separate GC method. PCB 65 and 200 (Dr. Ehrenstorfer) were used as internal standards for all halogenated compounds. Methanol and HPLC grade water were from Merck (Darmstadt, Germany).
2.2. Sampling area
The cruise was conducted on board of the Spanish Research Vessel (R/V) “Hespérides” from June 28th to July 28th, 2007, sailing from Iceland to the FramStrait. Seven ice stations were sampled along the cruise track (Figure 1), corresponding to multi-year ice ranging in thickness from 2 to 3 m. At each station, 1 mdeep x7.25 cm diameter ice cores were collected using a motorized Mark III coring device (Kovacs Enterprise Inc.), removing the unconsolidated surface snow and ice before sampling. The blades of the coring device were stainless steel and were rinsed with MilliQ water prior to each sampling event. The individual ice cores were inserted inside a precleaned PVC core holder and transported to the research vessel, where they were kept at – 12 º C until sectioned, typically within two hours after sampling. From each 1 m long ice core, the 20 cm extremes were cut using an acetone precleaned knife and placed in a PFTE bag, sealed totally and left in a cooler at the side of the ship (0-5 ºC) until melted.
2.3 Extraction procedure
SBSE extraction was performed in situ in the vessel laboratories.100 mLice melt water was transferred in a water-methanol-acetone pre-washed Erlenmeyerflasks where 10 mL of MeOH were added together with 100-500 pg of the internal standards.At this step, new precleaned stir bars (or Twisters) were added in the Erlenmeyer flask which were immediately capped and placed on the 15 position magnetic stirrer (Gerstel, GmbH, Mülheuim a/d Ruhr, Germany) at room temperature, in the dark.Extractions were carried out with new 20 mm length × 1.0 mmfilm thickness polydimethylsiloxane (PDMS)coated stir bars which corresponded to 126 µL of phase. Each sample was extracted in duplicate. Samples were agitated at 900 rpm during 24 h to reach an equilibrium partitioning between the dissolved chemical and the PDMS phase of the stir-bar. The extraction of solutes from aqueous phase into PDMS phase is controlled by the PDMS/water partition coefficient (approximated by the octanol water coefficient, log Kow) to the mass of analyte present in the aqueous sample of a known volume, according to:
Where K PDMS/w is the distribution coefficient between polydimethylsiloxane and water; C PDMS and Cw is the concentration of a solute in the polydimethylsiloxane phase and in the water; m PDMS y m w is the mass of the solute in the polydimethylsiloxane phase and in the aqueous phase and ß is the phase ratio (ß = VS/VPDMS, which represents the volume of the PDMS coated Twister and the volume of water, respectively [17, 18]). All target analytes studied exhibit Kowthat reflect high hydrophobicity and thus have a high tendency to diffuse onto the PDMS phase. A theoretical percent recovery for a given analyte i initially dissolved in water is given by:
If the KSBSEfor any specific compound is substituted by its Kow, the theoretical percentage recovery can be calculated. In our specific case, and using BDE 47 as an example, considering the Kow = 5.9 x 106, the sample volume of 100 mL and the volume of the PDMS fiber of 126 µL, substituting these values to the above equation, we obtain:
This calculation predicts that the recovery for this specific analyte would be of 100%. As demonstrated by other authors, Kow higher than 3.5 ensures an efficient partitioning of solutes to the PDMS phase within 2 h, and partitioning increases with longer extraction times, leading tohigher sensitivities[19, 20]. After 24 h extraction, time chosen for the above mentioned conditions, stir bars were removed with tweezers, rinsed with HPLC grade water, dried with a lint-free tissue and placed into the insert of a 2 mL vial andcapped. The pre-concentrated SBSE bars were kept at 4º C in the refrigerator of the boat during 4 months, time that took the ship to reach Spain. Once samples were gathered from the ship, they were immediately processed in the land-based laboratory.
To prevent any external source of contamination and to ensure full recovery of target analytes, some precautions were taken in the extraction and storage steps in the boat conditions: (i) ice cores pieces were placed from the holders into the Teflon bags inside a cool room avoiding any contact with hands; (ii) teflon bags were sealed until melted; (iii) 100 mL of meltwater (in duplicate) were place directly inside the precleaned Erlenmeyer flask and capped immediately after so that there was no contact with ship atmosphere; (iv) storage conditions were controlled by measuring the recoveries of internal standards in each sample.
2.4. Instrumental analysis
An Agilent 6890GC/5975B MS system (Agilent Technologies, Palo Alto, CA, USA) equipped with a programmed-temperature vaporization (PTV) injector was used. Two stir bars (corresponding each to 100 mL of extracted meltwater) were placed inside a precleaned Twister Desorption Liners (Gerstel), capped with a sealed Transportation Adapter and placed on a Autosample Tray. Stir bars were thermally desorbed in a thermal desorption unit (TDU from Gerstel) connected to the PTV injector CIS-4 (Gerstel) by a heated transfer line. TD was performed from 15 ºC (holding time 0.8 min) and then increased at 60 ºC/min to 280 ºC held during 7 min (desorption parameters). Helium flow was set at 50 mL/min. The PTV injector temperature was held at 8 °C during 0.1 min and then increased to 325 ºC at 10 ºC/s and finally held during 7 min. An Agilent HP-5MS (30 m × 0.25 mm i.d. × 0.25 μm film thickness) capillary column was used. The oven temperature was programmed from 70 ºC(holding time 2 min) to 150 ºC at 25 ºC/min, to 200 ºC at 3 ºC/min and finally to 280 ºC at 8 ºC/min, keeping the final temperature for 10 min. Transfer line and ion source temperatures were 280 ºC and 230 ºC, respectively.
Data acquisition was performed simultaneously using full scan conditions over a mass range of 44 to 750 amu and time scheduled Selected Ion Monitoring (SIM) using three or four ions per compound (Table 2). To enhance sensitivity, the SIM program was optimized using the autoSIM option and resulted in 27 chromatographic windows where 1 to 7 compounds were included, thus diminishing the number of ions displayed in each window and therefore, increasing in sensitivity. The sum of the two most abundant ions per compound was used for quantification. Peak detection and integration was carried out using MSD ChemStation (Agilent) software using external standard quantification. The concentration of target analytes was corrected by the recovery of each surrogate standard in cases where recoveries were lower than 70% (Table 3).
2.5. Quality Control/Quality Assurance
To prevent contamination and to obtain reliable POPs data, special care was given to blank analysis and to sensitivity and identification criteria.
As for blank analysis, HPLC grade water was extracted in boat conditions to evaluate possible external contributions of any of the target compounds. We also performed laboratory blanks using HPLC grade water and we evaluated the memory effect of empty Twister DesorptionLiners used in the TDU of the GC to evaluate carry over effects among samples.
Method optimization was performed with HPLC grade water. A calibration curve at 1, 5, 10, 20, 50, 90, 180, 460, 900, 1360 and 1800 pg/L (the last 2 concentrations were measured only for PAHs) with internal standards at 500 pg for deuterated PAHs and 100 pg for PCBs was used.Recoveries were tested in HPLC water spiked at 10 pg/L level. LODs for organohalogenated compounds were calculated by dividing the sum of the intercept value plus 3 times its standard deviation by the slope, both obtained from the calibration curve. This technique relies on the overall performance of the calibration, not just the response at one concentration. For PAHs, since boat blanks contained traces between 98-300 pg/L, LODs were calculated using 3 times the standard deviation of 3 blanks.
All target compounds should undergo the following identification and confirmation criteria: (i) each compound was identified using at least 3 specific ions; (ii) the retention time of target compounds should be within 3 s to that of a standard; (iii) the isotope ratio of the two ions monitored per congener should be within 20% of the theoretical isotopic ratio, and (iv) the signal to noise ratio for the sum of 2 ions of a specific compound should be S/N=3 or higher.
3. Results and discussion
3.1. GC-EI-MS performance
GC-quadrupole mass spectrometer (Q-MS) with electron impact (EI) ionization has been identified as the technique most often applied to the analysis of a large number of POPs given their easy calibration and operational features [21, 22]. The multiresidual SBSE method herein developed included the main OC pesticides, PCB and PBDE congeners according to their use in technical formulations. The chromatographic conditions were optimized to resolve 48 compounds in 40 min (Figure. 2). Due to the nature of compounds, 3 coelutions were observed: (i) PCB101 and α-endosulfan, which could be resolved at their specific m/z and fully identified using 3 or 4 acquisition ions; (ii) benzo(a)anthracene and chrysene and (iii) 4,4’-DDD and 2,4’-DDT which had the same ions and their concentration is given as the sum of both compounds.
Calibration range was performed at an ultra-low level, from 1 to 1800 pg/L (900 pg/L for organohalogenated compounds since they are expected at the low pg/L concentration in Arctic waters). Table 2 provides the calibration parameters obtained using external standard quantification. In this specific case, the surrogate standards were used to determine recovery efficiency and the stability of the compounds stored in the SBSE bars, but were not used for quantification purposes since their concentration exceeded the concentration levels of target compounds in the samples. In general, good linear calibration curves were obtained (typically, R2 > 0.990) (Table 2). Overall, these results imply that a linear range was maintained over at least 2 orders of magnitude, which ensures a good reliability to quantify compounds present at the pg/L level.
3.2. Sensitivity and detection limits
The method was developed and finely refined to increase its sensitivity as much as possible, while maintaining the identification capabilities of the EI ion source. By optimizing the time scheduled SIM conditions, 27 retention time windows were obtained with 4 to 14 ions per window. By decreasing the number of ions per window, it is possible to enhance up to 10 times the sensitivity of the method by increasing the dwell time of each ion. Another feature of the method is that the base peak and the second most abundant ion were summed to increase the ion sensitivity. By doing so, it is possible to almost double the signal of any particular ion, depending on the intensity of each. Table 2 provides the LODs obtained after pre-concentrating 200 mL of water spiked over 1 to 1800pg/L (900 pg/L for organohalogenated compounds). These LODs range from 0.1 to 99 pg/Lfor PCBs, PBDEs and chlorinated pesticides. For PAHs, given that a small contribution was observed from blanks performed on board of the vessel, the LODs were calculated using 3 times the standard deviation of three boat blanks. Levels ranged from 102 to 891 pg/L, which are also adequate to determine PAHs in ice since they arepresent at higher levels than organohalogenated compounds. Comparing to other studies where SBSE was optimized [16, 19, 20], herein we provide LODs 10-100 times lower due to the calibration optimization at ultra-trace levels.