Quantitative Analysis of Mixed Halogen Dioxins and Furans in Fire

Article pubs.acs.org/ac

Quantitative Analysis of Mixed Halogen Dioxins and Furans in Fire Debris Utilizing Atmospheric Pressure Ionization Gas Chromatography-Triple Quadrupole Mass Spectrometry Kari L. Organtini,† Anne L. Myers,‡ Karl J. Jobst,‡,§ Eric J. Reiner,‡,∥ Brian Ross,⊥ Adam Ladak,# Lauren Mullin,# Douglas Stevens,# and Frank L. Dorman*,† †

Department of Biochemistry, Microbiology, and Molecular Biology, The Pennsylvania State University, 107 Althouse Laboratory, University Park, Pennsylvania 16802, United States ‡ Ontario Ministry of the Environment and Climate Change, 125 Resources Road, Toronto, Ontario, Canada M9P 3V6 § Department of Chemistry, McMaster University, 1280 Main St. W., Hamilton, Ontario, Canada L8S 4M1 ∥ Department of Chemistry, University of Toronto, 80 St. George Street, Toronto, Ontario, Canada M5S 3H6 ⊥ Fire and Emergency Services Training Institute, P.O. Box 6031, 2025 Courtneypark Drive East, Toronto, Ontario, Canada L8S 4M1 # Waters Corporation, 34 Maple Street, Milford, Massachusetts 01757, United States S Supporting Information *

ABSTRACT: Residential and commercial fires generate a complex mixture of volatile, semivolatile, and nonvolatile compounds. This study focused on the semi/nonvolatile components of fire debris to better understand firefighter exposure risks. Using the enhanced sensitivity of gas chromatography coupled to atmospheric pressure ionization-tandem mass spectrometry (APGC-MS/MS), complex fire debris samples collected from simulation fires were analyzed for the presence of potentially toxic polyhalogenated dibenzop-dioxins and dibenzofurans (PXDD/Fs and PBDD/Fs). Extensive method development was performed to create multiple reaction monitoring (MRM) methods for a wide range of PXDD/Fs from dihalogenated through hexahalogenated homologue groups. Higher halogenated compounds were not observed due to difficulty eluting them off the long column used for analysis. This methodology was able to identify both polyhalogenated (mixed bromo-/ chloro- and polybromo-) dibenzo-p-dioxins and dibenzofurans in the simulated burn study samples collected, with the dibenzofuran species being the dominant compounds in the samples. Levels of these compounds were quantified as total homologue groups due to the limitations of commercial congener availability. Concentration ranges in household simulation debris were observed at 0.01−5.32 ppb (PXDFs) and 0.18−82.11 ppb (PBDFs). Concentration ranges in electronics simulation debris were observed at 0.10−175.26 ppb (PXDFs) and 0.33−9254.41 ppb (PBDFs). Samples taken from the particulate matter coating the firefighters’ helmets contained some of the highest levels of dibenzofurans, ranging from 4.10 ppb to 2.35 ppm. The data suggest that firefighters and first responders at fire scenes are exposed to a complex mixture of potentially hundreds to thousands of different polyhalogenated dibenzo-p-dioxins and dibenzofurans that could negatively impact their health. he career of firefighting is considered a dangerous occupation in many aspects. On site deaths and injuries are always at the forefront of consideration. More recently, increasing awareness is being given to the health effects that firefighters experience from exposure to fire debris created during fire incidents. Volatile organic compounds (VOCs) are the most commonly studied emissions from fire sites. A comprehensive study of VOCs from structural fires performed by Austin et. al identified benzene, toluene, naphthalene, propene, 1,3-butadiene, and styrene as the VOCs most commonly present in such samples.1 Many of these compounds are regulated and are known or possible human carcinogens. It is important to investigate the semivolatile and nonvolatile content of fire debris and the particulate matter it generates as

T

© XXXX American Chemical Society

these compounds may also significantly contribute to the longterm health of firefighters. Several epidemiological studies have been published on the cancer risks experienced in the firefighter population. LeMasters et al. compiled 32 of these studies to determine overall cancer risk.2 They concluded that there was a significant increase in risk to firefighters in association with several cancers including multiple myeloma, non-Hodgkin’s lymphoma, prostate, and testicular cancers. Other cancers showed probable risk. A study by Zeig-Owens et al. published in 2011 also Received: July 1, 2015 Accepted: September 26, 2015

A

DOI: 10.1021/acs.analchem.5b02463 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

spectrometry has been demonstrated in the application to the analysis of PCDD/Fs by Reiner et al.17 and a small range of PXDD/Fs by Myers et al.18 Atmospheric pressure ionization provides a soft ionization technique, preserving the molecular ion for additional sensitivity. By taking advantage of the benefits provided by utilization of APGC-MS/MS, trace levels of dioxin compounds can be easily, accurately, and confidently identified in complex fire debris samples.

concluded that there was a 19% increase in total cancer rates among World Trade Center debris exposed New York City firefighters after the September 11th attacks.3 The increased risks of occupational exposure to toxic compounds in fire debris makes it exceedingly important to more fully understand the types of compounds that are being created. The heavy load of flame retardants in modern consumer products may be a contributing cause for the increase in longterm detrimental health problems in firefighters. Brominated flame retardants (BFRs), more specifically polybrominated diphenyl ethers (PBDEs), have been among the most commonly used brominated flame retardants on the market. Certain PBDE formulations are starting to be phased out due to potential toxicity, but they have not all been replaced or phased out yet and importation of products containing PBDEs is not prohibited.4 Furthermore, many products containing PBDEs are still in current use. It has been recognized that, under thermal stress, PBDEs and other BFRs form polybrominated dibenzo-p-dioxins (PBDDs) and dibenzofurans (PBDFs).5 With a source of chlorine present in the product materials, polyhalogenated (Br/Cl) dibenzo-p-dioxins and dibenzofurans can also be created (PXDD/Fs).6 PXDD/Fs have been found in the fly ash resulting from large-scale combustion of consumer products in municipal waste incinerators, indicating the potential for their formation under accidental fire conditions as well.7−9 Dibenzo-p-dioxins and dibenzofurans have long been considered a persistent environmental pollutant of concern due to their toxicological effects and classification as a potential human carcinogen.10 Dioxin compounds were found to be present at the World Trade Center site following the initial collapse of the towers and subsequent smoldering of the debris.11 In fact, elevated blood serum levels of polychlorinated dibenzo-p-dioxins and dibenzofurans have been reported in firefighters in a few separate incidences, including those involved in the World Trade Center rescues.12−14 Further understanding of the creation and exposure of halogenated dibenzo-p-dioxin and dibenzofuran congeners needs to be undertaken. Currently, only 17 chlorinated congeners are regulated and monitored for, which does not fully represent the exposure hazards to samples, such as fire debris, where these compounds are also likely generated.15 Initial studies into the generation of PXDD/Fs in fire debris created at a firefighter training institute used GCxGC-TOFMS to begin to understand the complexity of the samples.16 The previous work identified a range of PXDFs and PBDFs in the fire debris samples, but some samples analyzed contained little detectable content. Although the previous study demonstrated the generation of mixed halogenated dioxins in simulated fire debris, it indicated that an analytical technique providing greater sensitivity was needed to fully characterize the PXDD/F content of the fire debris samples to fully understand environmental and human exposure. Tandem mass spectrometry (MS/MS) employing a triple quadrupole mass spectrometer equipped with atmospheric pressure chemical ionization (APGC-MS/MS) was employed for further, highly sensitive, analysis of the samples. Multiple reaction monitoring (MRM) allowed for selection of only the analytes of interest and improved sensitivity through the reduction of chemical noise contributed by other components in the sample extract. The mass spectrometer can switch between MRM transitions quickly, allowing for development of a single method that monitors for a large range of compounds, which is crucial for analysis of PXDD/Fs. The technique of tandem mass

■

EXPERIMENTAL SECTION Standards and Chemicals. All standards used were obtained from Wellington Laboratories, Inc. (Guelph, Ontario, Canada) and were >99% pure. Polyhalogenated dibenzo-pdioxin and dibenzofuran (PXDD/F) standard mixes were prepared in the laboratory using individual PXDD/F standards (Table S1). Polybrominated dibenzo-p-dioxin and dibenzofuran (PBDD/F) standard mixes consisting of monobromo through octabromo congeners of both dibenzo-p-dioxins and dibenzofurans were prepared in the laboratory using individual PBDD/F standards. EPA1613-CSS was used as a cleanup recovery standard and was added to sample extracts after extraction, prior to any sample cleanup. This standard consists of 37Cl4-2,3,7,8-TCDD. EPA1613-ISS was used as an injection standard and was spiked into sample extracts immediately prior to sample analysis. This standard consists of 13C-1,2,3,4,-TCDD and 13C-1,2,3,7,8,9HexaCDD. A mixture of 13C labeled PCDD/F and PXDD/F standards were spiked into samples prior to extraction for isotope dilution quantification. EPA1613-LCS was used as the 13 C-PCDD/F standards, consisting of 15 13C-labeled compounds. A mixture of 13C-labeled PXDD/F was prepared in the laboratory using individual 13C-PXDD/F standards (Table S1). The isotope dilution method of quantification was utilized to provide for an increase in accuracy by accounting for percent recovery of the native compounds being quantified. Organic solvents (toluene, hexane, dichloromethane, acetone) were obtained from Avantor Performance materials (formerly JT Baker, Center Valley, PA) and were ultra resianalyzed grade. Nonane was obtained from Acros Organics (New Jersey) and was 99% pure. High grade sulfuric acid and sodium hydroxide pellets were obtained from Avantor Performance materials (formerly JT Baker, Center Valley, PA). Activated carbon used for sample cleanup was AmocoPX21 and was obtained from Ontario Ministry of Environment, as it is no longer in production. Specific procedures have previously been developed using this material for “dioxin and dioxin-like” extract cleanup.16 Silica gel for sample cleanup was obtained from Sigma−Aldrich (St. Louis, MO) and was high purity grade with a pore size of 60°A. Burn Study. To generate fire debris samples, burn studies were conducted at the Fire and Emergency Services Training Institute (FESTI) in Toronto, Ontario, Canada. Studies were developed and implemented as described previously.16 Debris was created at two burn simulations. The contents and procedure of the first burn simulation are described in detail previously.16 The second burn simulation study was similar in nature, given that it simulated both a residential (household burn) and commercial (electronics burn) fire. The household fire consisted of two fabric sofa chairs with various cushions, two rugs, six pillows, a foam mattress pad, a wooden coffee table, and two vinyl covered chairs. The electronics fire consisted of two computer towers, two flat screen computer monitors, two keyboards, two used ink cartridges, two B


Article


temperature program was as follows: initial oven temperature, 120 °C hold for 1 min; 35 °C/min to 200 °C no hold; 4.5 °C/ min to 280 °C hold for 8 min; 20 °C/min to 330 °C hold for 15 min. The Xevo TQ-S tandem quadrupole mass spectrometer source was run under dry conditions to promote charge transfer ionization and maintained at a temperature of 150 °C. Nitrogen was supplied by an NM32LA nitrogen generator from Peak Scientific (Billerica, MA) and was used as the auxiliary gas, maintained at a flow rate of 400 L/h. Argon was utilized as the collision gas at a flow rate of 0.18 mL/min. Cone gas flow was turned off for the first 8 min of the analysis and then maintained at a flow rate of 215 L/h until the end of the analysis. Corona current was 20 μA for the first 8 min of the analysis and then maintained at 4.0 μA until the end. Cone voltage was maintained at 30 V for all compounds. The mass spectrometer transfer line was heated to 360 °C. The mass spectrometer was operated in positive ion mode using the multiple reaction monitoring (MRM) mode. Between three and six transitions were utilized for each compound monitored. One MRM served as a quantification ion, and the remaining MRMs were qualifier ions. For the dibenzofuran species, MRMs monitored included −COBr, −COBr2, and −COBrCl. For the dibenzo-p-dioxin species, MRMs monitored included −COBr and −(CO)2BrCl. MRM transitions for the 13 C labeled PXDD/F standards only utilized the −13COBr transition. The PXDD/F method also contained 13C labeled PCDD/Fs. Two MRMs were monitored for each compound selecting for the −13COCl precursor > product transition. Specific MRM information for the PXDD/F method, including retention time windows, collision energies, and dwell times utilized, can be found in Table S2. A separate method was developed for analysis of the PBDD/ F species. Up to three MRM transitions were monitored for each homologue group. Fragment losses monitored included −COBr, −COBr2, (dibenzofurans only), and −(CO)2Br (dibenzo-p-dioxins only). Specific MRM information for the PBDD/F method, including retention time windows, collision energies, and dwell times utilized, can be found in Table S3.

televisions, a microwave, and various VHS tapes in a plastic holder. In addition, sheets of aluminum foil were secured to the walls of each burn container to collect airborne particulates. A road flare was used to directly ignite both fires. The doors of the burn cells remained open during the fires to allow greater oxygen flow to produce a more complete burn than previously achieved in the first study, which was allowed to smolder for a longer period of time. Both fires were allowed to burn for approximately 30 min before being completely extinguished with water. Samples were collected following extinguishing of each fire. Any solid debris samples that remained were collected from each of the items in both fires. A general sampling of mixed debris was also collected from the floor of the container from both fires. The only aluminum foil sheets that remained intact after extinguishment were those in the electronics fire and these were collected for analysis. A scraping of burnt ash was collected from the household fire. Toluene soaked cellulose wipe samples were taken from the helmets of all three firefighters assisting with the burns after the completion of both fires. All samples were collected and secured in either 500 mL amber screw top jars or plastic zip bags, depending on the size and nature of the sample. Sample collection vessels were new and therefore were not tested for background contaminants. Prior to sampling, control samples were collected of the aluminum foil and cellulose wipes used for sample collection to verify the absence of target compounds. Sample Extraction and Cleanup. Sample extraction and cleanup was performed in accordance with the Ontario Ministry of Environment Method E3418.19 The procedure was completed as previously described.16 Briefly, samples were air-dried for at least 24 h, until no visible signs of condensation remained. Sample extraction was performed using Soxhlet extraction with hexane as the extraction solvent. Prior to extraction, samples were fortified with a mixture of 13C labeled PXDD/Fs and PCDD/Fs in order to perform quantification by isotope dilution. Samples were extracted for approximately 22 h. Extracts were concentrated prior to cleanup using a rotary evaporator. Prior to sample cleanup, a cleanup recovery standard (37Cl4-2,3,7,8-TCDD; EPA1613-CSS) was spiked into each sample. Two stages of sample cleanup were performed. First, the sample was eluted with hexane through an acid−base silica column. Next, the extract was eluted through a reversible 5% carbon/silica (w/w) column using toluene as the final elution solvent. The cleaned sample extract was then concentrated to 100 μL using a rotary evaporator and evaporation under a nitrogen stream. Immediately prior to instrumental analysis, the samples were diluted 1:10 in toluene and spiked with an internal standard solution (EPA1613-ISS). APGC-MS/MS Analysis. Sample analysis was performed using a Xevo TQ-S equipped with the APGC source (Waters Corporation, Milford, MA) and an Agilent 7890A gas chromatograph and Agilent 7693A autosampler (Agilent Technologies, Santa Clara, CA). A 60 m × 0.18 mm × 0.10 μm Rtx Dioxin-2 (Restek, Bellefonte, PA) column was used for the analysis. Approximately 1.0 m × 0.32 mm stainless steel Sulfinert tubing (Restek, Bellefonte, PA) was coupled to the column exit and was installed through the transfer line of the instrument. Helium carrier gas was used, and the GC was operated in splitless mode at a constant flow rate of 1.1 mL/ min. The injector was maintained at a temperature of 290 °C and utilized a 4.0 mm drilled hole UnilinerTM (Restek, Bellefonte, PA). Samples were injected using the 7693A ALS (Agilent Technologies) at a volume of 0.5 μL. The GC oven

■

RESULTS AND DISCUSSION MRM Method Development. For analysis of PXDD/Fs, MRMs had to be systematically developed, in most cases, without the use of a pure standard. Commercial availability of PXDD/F standards is extremely limited. In order to develop the MRMs needed for analysis of these compounds, a sample extract previously analyzed using GCxGC-TOFMS was used.16 The sample used contained both the largest range of PXDD/Fs identified and among the highest concentration of PXDD/Fs of the available samples. A list of potential precursor > product fragment losses was generated using mass spectra generated on the GCxGC-TOFMS instrument with an EI source. A comprehensive list of the MRM’s tested during method development is listed in Table S4. The transition of the most abundant mass in molecular ion cluster (typically M + 2 or M + 4) > M-COBr was chosen as the quantitation ion for all compounds. This MRM typically produced the largest response. The next most abundant mass (+2 or −2 amu from most abundant mass) was also chosen as an MRM monitoring its loss of −COBr as the first qualifier ion, as these are unique loss fragments for these molecules. A variety of other MRMs were chosen for each mixed halogenated class of congeners based on the fragment ions of −COBr, −COBr2, C


Article


Figure 1. Chromatogram of electronics fire sample extract demonstrating different MRMs monitored for the BrCl dibenzofuran homologue group.

Figure 2. Total ion chromatogram of the Br2Cl dibenzofuran homologue group in an electronics fire burnt dry wall sample. The letters represent the peaks for which peak-to-peak signal/noise (S/N) values were calculated. S/N values are as follows (A) 46, (B) 20, (C) 89, (D) 62, and (E) 12.

−COBrCl, and − (CO)2BrCl. More qualifier ions were used for dibenzo-p-dioxins and dibenzofurans that were barely detectable or not detected at all using prior GCxGC-TOFMS analysis. Elution time windows were also set based on the window in which the compounds eluted in the method development sample extract. Figure 1 demonstrates three different MRM transitions monitored for the BrCl dibenzofuran homologue group. A commercial standard does not exist for this particular group.

The peak patterns in all three MRM transitions match well, indicating these are valid transitions to use for further analysis. Similar confirmation was performed for all of the MRMs developed for each homologue group. The chromatograms of these samples are a result of a complex mixture of chemically similar compounds. Therefore, most peaks are unresolved from other peaks of the same homologue group using this technique. The inability to completely resolve these chemically similar compounds creates D


Article


Figure 3. Total ion chromatogram of the BrCl3 dibenzo-p-dioxin homologue group in an electronics burnt wire fire debris sample. The inset shows the isotope ratio match of the molecular ion cluster comparing the experimental mass spectrum (top) and the predicted mass spectrum (bottom).

using the GCxGC-TOFMS. The same sample extracts were analyzed using the APGC-MS/MS with the MRM method discussed above. APGC-MS/MS analysis supported the findings from the GCxGC-TOFMS analysis; however, it proved to be a much more sensitive technique by approximately a thousand-fold. A larger number of peaks were identified in each congener homologue group monitored. Figure 2 demonstrates the magnitude of enhanced detectability using the APGC-MS/ MS. The chromatogram shows the Br2Cl DF homologue group in an electronics fire burnt dry wall sample. Approximately 20 peaks are identifiable in the figure, although only one peak was detected in the previous TOFMS analysis.16 The increase in detected peaks demonstrates the significant improvement in sensitivity by coupling APGC with MS/MS. The peak-to-peak signal-to-noise values calculated for selected peaks in the chromatogram (labeled A through E) range from 12:1 to 89:1, demonstrating that a substantial signal is still produced from trace levels of these compounds. Using the enhanced sensitivity of the APGC-MS/MS system, polyhalogenated and polybrominated dibenzo-p-dioxins were identified in a few of the electronics fire samples from the first burn study. No polyhalogenated dibenzo-p-dioxins were detectable in the second burn study, likely as a result of more complete combustion in the second round of simulations. PXDDs identified include BrCl2, BrCl3, Br2Cl, Br2Cl2, Br2Cl3, and Br3Cl dibenzo-p-dioxin. PBDDs identified ranged from Br through Br5 dibenzo-p-dioxin. Identifications were verified by authentic standard, if available. Otherwise, the isotope ratio pattern of the molecular ion cluster was utilized for identification, as demonstrated in Figure 3. The figure

a chromatogram containing a large number of unresolved peaks, as demonstrated in Figure 1. Although information on single peaks cannot generally be inferred, the pattern of the group of peaks as a whole can be useful. Also, the entire area of the complex of peaks can be summed and quantified to generate a total congener group concentration. A method for PBDD/F analysis was also developed in a similar manner as described, although with the aid of standards, instead of using a sample extract. Elution windows for each class of PBDD/F congener were established with a sample extract. APGC-MS/MS Is an Extremely Sensitive Technique for Fire Debris Analysis. An interlaboratory study was previously performed in which the APGC-MS/MS instrument was directly compared to a magnetic sector high resolution mass spectrometer.20 The APGC-MS/MS system demonstrated equivalent, or better, performance than the GC-HRMS system for chlorinated dibenzo-p-dioxin and dibenzofuran analysis. It also demonstrated increased sensitivity with the ability to significantly expand upon the dioxin analysis methods that only monitor for 17 polychlorinated congeners without sacrificing sensitivity. The developed APGC-MS/MS technique was also compared to GCxGC-TOFMS analysis which was initially used to analyze samples to get a better understanding of what compounds were present. Data from the GCxGC-TOFMS analysis of the first burn study identified a wide range of mixed halogenated (Br/ Cl) dibenzofurans and polybrominated dibenzofurans, although some samples contained little detectable content.16 Furthermore, no polyhalogenated dibenzo-p-dioxins were identified E


Article


Figure 4. Graphs comparing the PXDD/F (A, B, E) and PBDD/F (C, D, F) contents of selected fire debris samples. Normalized peak area was calculated on the basis of the 13C-1,2,3,4-TCDD injection standard spiked into all samples prior to analysis. Note that the y-axes are shown on a logarithmic scale due to large variances in normalized peak areas between congener groups.

the trend holds that burn study 1 generated a larger amount of PXDFs (Figure 4E). Conversely, the PBDF trend is the opposite. In the same electronics samples, general debris and foil, burn study 2 generated larger numbers of PBDFs (Figure 4C,D). Interestingly, when the average of each homologue group is taken across all samples analyzed, the trend is not so straightforward. The dibenzofurans with a lower bromination level (Br−Br3) were more prominent in burn study 2; both burn studies were fairly even in overall Br4 levels, and the dibenzofurans with a higher bromination level (Br5−Br7) were more prominent in burn study 1. There were two samples of interest that did not follow these general trends. A circuit board was collected and extracted from the electronics fire during each burn study. The circuit board sample from burn 1 had high levels of both PXDFs and PBDFs whereas the circuit board sample from burn 2 had very low PXDF and PBDF content. Also, the helmet wipes collected from the firefighters during each study were very different as well. All three helmet wipe samples taken during the second burn study contained relatively high levels of PXDFs and PBDFs, whereas the firefighter helmet wipe collected from the first burn study had much lower contents of both PXDFs and PBDFs. The qualitative differences between the two burn studies support the theory that dibenzo-p-dioxin and dibenzofuran generation is dependent upon the starting contents of the fire and the properties of the actual combustion (duration, temperature, etc.). The differences seen in this study indicate that the materials burned in each separate study contained differing loads of flame retardants. This is particularly demonstrated in the overall pattern of PBDFs generated. The

demonstrates the BrCl3 DD homologue group function with several peaks identified. The subset in the figure demonstrates the isotope ratio match of the indicated peak, comparing the experimental mass spectrum with the predicted mass spectrum generated from the MassLynx software of a compound with the molecular formula of C12H4O2BrCl3. The experimental mass spectrum only contains the particular masses monitored in the MRM transitions. Therefore, it does not contain the full isotope pattern of the molecular ion cluster in the predicted mass spectrum. Qualitative Comparison of Burn Studies. A quantitative comparison cannot be performed between the samples generated at each burn study as 13C-labeled internal standards for isotope dilution quantification were not available during the extraction of the first burn study samples. However, an injection standard was utilized at a constant concentration in all samples prior to APGC-MS/MS analysis. 13C-1,2,3,4-TCDD was used as an injection standard; therefore, total peak areas of each homologue group could be normalized to the 13C-1,2,3,4TCDD area. The calculated ratio was used for comparison purposes to discover trends among the generation of fire debris samples generated by two separate experiments. Overall, the samples generated in the first burn study had higher levels of PXDFs, as demonstrated in Figure 4A,B,E. Figure 4A,B shows the trend in two individual samples, general electronics debris and electronics foil, respectively. The electronics debris graph (Figure 4A) does indicate that two congener groups, BrCl and Br2Cl, show the opposite trend, with higher normalized peak areas in burn study 2. Generally, though, when the total peak areas for each homologue group are averaged among all samples analyzed for each burn study, F


Article


Table 1. Isotope Dilution Semiquantification of Extracted Household Fire Debris Samples Generated from a Simulated Controlled Burn Experiment in ng/g paper-like material BrCl DF BrCl2 DF BrCl3 DF Br2Cl DF BrCl4 DF Br2Cl2 DF Br2Cl3 DF Br3Cl DF Br3Cl2 DF Br4Cl DF Br4Cl2 DF Br5Cl DF Br DF Br2 DF Br3 DF Br4 DF Br5 DF Br6 DF Br7 DF

dry wall

0.20 0.03 0.01

0.01 0.01 0.01

0.0008

0.0015

0.32

0.01 0.16

0.04

0.02 0.00 0.01 0.46 0.72 1.49 1.34 0.39 0.18

0.78 2.21 3.59 4.00 0.57 0.19

door scraping

sofa 1

1.42 0.76 0.58 7.63 0.07 5.11 0.15 5.32 0.02 0.24

0.16 0.02

0.35 11.84 48.67 27.17 3.16 0.29

0.91 1.55 1.37 0.77 0.41

sofa 2

vinyl chair

house ash 1A

house ash 1B

0.47 15.35 3.51 4.45 3.75

2.39 0.64 0.81 0.84

0.74

0.04

22.53 5.26 5.95 4.77 1.45

40.88 82.11 50.40 30.35 7.40

Table 2. Isotope Dilution Semiquantification of Extracted Electronics Fire Debris Samples Generated from a Simulated Controlled Burn Experiment in ng/g

BrCl DF BrCl2 DF BrCl3 DF Br2Cl DF BrCl4 DF Br2Cl2 DF Br2Cl3 DF Br3Cl DF Br3Cl2 DF Br4Cl DF Br4Cl2 DF Br5Cl DF Br DF Br2 DF Br3 DF Br4 DF Br5 DF Br6 DF Br7 DF

foil 1

foil 2

foil 3

firefighter helmet 1



circuit board

VHS tapes

microwave debris

electronics debris 1A

electronics debris 1B

4.58 2.45 2.84 27.62 1.20 41.10 3.57 70.72 4.95 20.13 3.88 3.70 4.69 65.42 506.12 730.19 287.03 51.86 6.93

2.31 0.54 0.60 9.29 0.16 10.36 0.86 13.35 0.42 2.50 0.56

21.48 10.30 9.53 88.26 2.65 96.25 7.21 140.88 8.39 24.98 5.34 6.36 4.90 136.75 630.99 870.37 349.28 59.12 11.84

6.51 1.94 2.30 25.44 0.77 36.13 3.12 62.34 4.45 19.44 6.54 6.97 11.15 51.11 198.41 385.37 326.61 417.82 974.01

3.41 1.06 0.95 12.68 0.26 10.07 1.05 14.34 0.88 4.32 1.29 1.80 4.10 67.07 255.59 340.96 153.83 72.34

16.44 6.75 10.05 76.57 5.05 103.56 16.65 175.26 25.84 135.50 43.48 56.62 6.62 117.78 604.68 1176.01 1235.80 1560.32 2349.78

0.10

2.15 0.26 0.48 5.60 0.09 4.77 2.53 20.67 2.43 11.43 3.66 11.68 12.45 78.07 192.54 147.83 164.77 152.98 239.54

1.88 0.26 2.16 5.08 0.13 3.53 9.19 7.85 0.36 2.33 0.25 0.77 24.36 187.89 658.23 749.61 203.37 27.35

6.13 1.05 2.54 14.50 0.43 11.94 2.93 29.18 1.83 14.23 2.64 4.59 189.00 1354.03 5274.86 6094.49 2012.30 495.08 298.80

10.07 2.99 6.37 33.01 2.13 35.37 5.75 66.09 3.79 18.01 2.23 4.08 170.05 1468.09 6040.79 9254.41 2725.79 470.21 120.18

4.93 71.84 238.76 210.29 70.06 8.93

PBDF pattern suggests that the second burn study possibly contained materials containing flame retardants of lower bromination levels and the first burn study contained materials with flame retardants of higher bromination levels. The pattern could also be a result of the second burn study being more of a complete combustion event than the first resulting in the creation of differing types of compounds. The difference in starting materials and burn time/completeness of combustion may also explain the lack of dibenzo-p-dioxin formation in the second burn study. It has been shown that dioxin formation is increased during events of incomplete combustion.21 Although there were indications that polyhalo- (Br/Cl) and polybromodibenzo-p-dioxins may be present in some of the samples from the second burn study, including a firefighter helmet wipe, the

0.48

0.33 1.68 4.60 6.58 2.88

results were inconclusive due to the peaks being close to background levels in the chromatogram. Quantification of PXDD/Fs and PBDD/Fs in Simulated Fire Debris. Quantification of PXDD/F and PBDD/F levels in the simulation fire debris samples was performed on the electronics debris samples created during the second burn study. These samples were spiked with a mixture of 13C-labeled dibenzo-p-dioxins and dibenzofurans prior to extraction in order to quantify by isotope dilution. Sample blanks (for foil and wipe matrices) and method blank samples were analyzed using the same analytical methods used for the collected burn samples. Dibenzo-p-dioxin and dibenzofuran levels in all blank samples were below the limit of detection (LOD) of the APGC-MS/MS. LODs were previously determined to range G


Article


Figure 5. Total ion chromatograms of the different tetra- substituted homologue groups identified in a firefighter helmet wipe sample.

of the second sofa chair and vinyl covered chair. These particular samples had PBDF levels more similar to the electronics fire samples. This pattern indicates that these particular pieces of furniture most likely were manufactured when the use of brominated flame retardants (most likely in the polyurethane foam) was commonly accepted. It is also interesting to note the PBDF levels in the other sofa chair burned were approximately 5−15-fold lower. The lower level of PBDFs identified also points to differences in starting materials. The door scraping taken from the household fire burn cell also had elevated levels of PBDFs, as well as the highest concentrations of PXDFs. This sample represented the particulate matter present in the smoke and ash that coated the surface of the debris, as well as what was transported from the fire through air. Studying these types of samples is significant because they are a representation of what firefighters and first responders are exposed to through inhalation and contact, as well as a representation of what is being deposited in the surrounding environments.

from 0.15 to 1.4 pg/g for tetra- through octa- halogenated dioxins and furans.20 The quantification of compounds performed is still considered semiquantitative for two reasons. 13C-labeled standards are only commercially available for a limited number of compounds analyzed in this study. Therefore, a true isotope dilution study cannot be performed because a labeled version of every compound is not available. Second, the quantification was performed as a sum of all peaks present in a homologue group instead of on an individual peak basis. This approach was used due to the complex mixture of peaks present in each homologue group resulting in chromatograms where optimal peak resolution is impossible to attain. Details on how quantification was performed can be found in the Supporting Information. The results of quantification of the household fire debris samples are tabulated in Table 1, and the electronics fire debris can be located in Table 2. As indicated in the quantification tables, the burn study 2 household fire samples contained much lower levels of both PXDFs and PBDFs than the electronics fire, with the exception H


Article


By performing two separate sets of burn simulation experiments, trends may be able to be identified in the difference of PXDD/F and PBDD/F generation possibly due to differences in combustion materials. Although levels varied between the two experiments, polyhalogenated dibenzofurans were generated in all simulated fires indicating the high likelihood that these compounds are generated in most fire debris. These results are concerning due to the potential of constant exposure of firefighters to these types of compounds. Samples removed directly from the firefighters active in the burn simulation experiments were shown to contain high levels of PXDFs and PBDFs. The ability for these compounds to be transported by particulate matter and ash generated in large amounts during fire scenes also suggest potential impacts to surrounding environments. The presence of dibenzo-p-dioxins and dibenzofurans of all halogenation levels in these samples compels the inclusion of an expanded range of dibenzo-pdioxins and dibenzofurans in regulations and routine environmental screening procedures. Excluding all but 17 chlorinated congeners may be severely underestimating the toxicity implications of environmental samples. Furthermore, the safety procedures and standards for firefighters in contact with fire debris need to be reevaluated in order to better protect them from exposure to these compounds in an effort to reduce or eliminate adverse health effects.

The quantification of the burn study 2 electronics debris samples produced much higher levels of both PXDFs and PBDFs when compared to the household fire samples. The only sample that did not have much detectable content was the circuit board sample. The PBDFs were present in higher concentrations than the PXDFs The most interesting samples were those that represented direct exposure to firefighters through particulate matter, including the foil samples (hung on the walls of the burn cell) and the firefighter helmet wipes. Besides the general electronics debris sample collected from the ashes in the bottom of the burn cell, the foil and helmet wipes contained the largest quantities of halogenated dibenzofurans (Table 2). This is concerning as it indicates that firefighters are directly exposed to high levels of a complex mixture of these potentially dangerous compounds through two sources. One source of exposure is inhalation of smoke and ash after removing their respirators. The second route of exposure is through contact exposure of touching contaminated pieces of equipment and the subsequent transport of these materials until proper cleaning of their suits and equipment. Furthermore, the water used to clean firefighter equipment is not released into the environment without any treatment, acting as another route of environmental deposition of dibenzop-dioxins and dibenzofurans into the environment. These results indicate that the halogenated dibenzo-p-dioxins and dibenzofurans may favor adsorption to small particulate matter rather than remaining in the actual burning debris. The preference of adsorption to particulate matter has been shown to be true of the polychlorinated analogs (PCDD/Fs).22,23 Therefore, these compounds may be easily transported, causing exposure concerns. Figure 5 gives a visual representation of the extent of complexity of halogenated dibenzofurans identified in the firefighter helmet wipes. The figure shows a range of all of the tetrahalogenated dibenzofurans identified in the helmet samples, ranging from tetrachloro- through tetrabromo- with all mixed bromo/chloro- homologue groups represented in between. Although the figure only shows a small subset of the halogenated planar contents of the sample, it represents the range of complexity in the samples. It is likely that there are numerous toxic congeners present in this “tetra-group”. Since the helmet wipe samples were taken directly off of active firefighters, it accurately represents their risk of being exposed to fire debris.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.analchem.5b02463. Native and 13C labeled polyhalogenated dibenzofurans and dibenzo-p-dioxins standards used. MRM transitions monitored, retention time windows, dwell times cone voltages, and collision energies for APGC-MS/MS analysis. Fragment ions tested for method development. Details of sample quantification. (PDF)

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: fl[email protected]. Phone: 814-863-6805. Notes

The authors declare no competing financial interest.

■

■

CONCLUSIONS A method for applying the benefits of gas chromatography coupled to atmospheric pressure ionization-tandem mass spectrometry was successfully developed in the absence of all but a few authentic standards. The method was used to analyze complex fire debris samples for a wide range of polyhalogenated dibenzo-p-dioxins and dibenzofurans from various simulated burn studies. Polyhalogenated (Br/Cl) and polybrominated dibenzo-p-dioxins and dibenzofurans were identified in the samples, with the dibenzofuran species dominating the samples. The use of the APGC-MS/MS technique provided a significant boost in sensitivity which was necessary for identification of these compounds that individually were present at trace levels. When quantified as halogenated homologue groups, the total levels of each homologue group were on the parts per billion level. Many of the individual peaks identified with the APGCMS/MS analysis had previously been overlooked with a less sensitive instrument.

ACKNOWLEDGMENTS The authors would like to acknowledge the following: (1) the staff at the Fire and Emergency Services Training Institute (FESTI), specifically Mike Hutchinson, Sam Marshall, Pike Krpan, Drew Morris, Zach Tomkinson, and Cody Williams, for providing their facilities and assistance during the burn simulation study; (2) Waters Corporation for the Xevo APGC-TQS system; (3) Brock Chittim, Alex Konstantinov, and Jeff Klein at Wellington Laboratories for technical assistance; (4) Restek Corporation for various chromatographic consumables and columns; (5) Terry Kolic from the Ontario Ministry of the Environment and Climate Change (MOECC) for being instrumental in obtaining the Amoco PX-21 Carbon and training on the extractions and cleanup procedures.

■

REFERENCES

(1) Austin, C. C.; Wang, D.; Ecobichon, D. J.; Dussault, G. J. Toxicol. Environ. Health, Part A 2001, 63, 437−458.

I


Article

Analytical Chemistry (2) LeMasters, G. K.; Genaidy, A. M.; Succop, P.; Deddens, J.; Sobeih, T.; Barriera-Viruet, H.; Dunning, K.; Lockey, J. J. Occup. Environ. Med. 2006, 48, 1189−1202. (3) Zeig-Owens, R.; Webber, M. P.; Hall, C. B.; Schwartz, T.; Jaber, N.; Weakley, J.; Rohan, T. E.; Cohen, H. W.; Derman, O.; Aldrich, T. K.; Kelly, K.; Prezant, D. J. Lancet 2011, 378, 898−905. (4) USEPA. Polybrominated diphenyl ethers (PBDEs) Project Plan; USEPA: Washington, DC, March 2006. (5) Eljarrat, E.; Feo, M. L.; Barcelo, D. In The Handbook of Environmental Chemistry: Brominated Flame Retardants; Elajarrat, E., Barcelo, D., Eds.; Springer: Heidelberg, 2011; pp 187−202. (6) Weber, R.; Kuch, B. Environ. Int. 2003, 29, 699−710. (7) Tong, H. Y.; Monson, S. J.; Gross, M. L.; Huang, L. Q. Anal. Chem. 1991, 63, 2697−2705. (8) Sovocool, G. W.; Donnelly, J. R.; Munslow, W. D.; Vonnahme, T. L.; Nunn, N. J.; Tondeur, Y.; Mitchum, R. K. Chemosphere 1989, 18, 193−200. (9) Chatkittikunwong, W.; Creaser, C. S. Chemosphere 1994, 29, 559−566. (10) Steenland, K.; Bertazzi, P.; Baccarelli, A.; Kogevinas, M. Environ. Health Perspect. 2004, 112, 1265−1268. (11) USEPA. Exposure and human health evaluation of airborne pollution form the World Trade Center disaster; USEPA: Washington, DC, October 2002. (12) Shaw, S. D.; Berger, M. L.; Harris, J. H.; Yun, S. H.; Wu, Q.; Liao, C.; Blum, A.; Stefani, A.; Kannan, K. Chemosphere 2013, 91, 1386−1394. (13) Horii, Y.; Jiang, Q.; Hanari, N.; Lam, P. K. S.; Yamashita, N.; Jansing, R.; Aldous, K. M.; Mauer, M. P.; Eadon, G. A.; Kannan, K. Environ. Sci. Technol. 2010, 44, 5188−5194. (14) Edelman, P.; Osterloh, J.; Pirkle, J.; Caudill, S.; Grainger, J.; Jones, R.; Blount, B.; Calafat, A.; Turner, W.; Feldman, D.; Baron, S.; Bernard, B.; Lushniak, B. D.; Kelly, K.; Prezant, D. Environ. Health Perspect. 2003, 111, 1906−1911. (15) Van den Berg, M.; Birnbaum, L. S.; Denison, M.; De Vito, M.; Farland, W.; Feeley, M.; Fiedler, H.; Hakansson, H.; Hanberg, A.; Haws, L.; Rose, M.; Safe, S.; Schrenk, D.; Tohyama, C.; Tritscher, A.; Tuomisto, J.; Tysklind, M.; Walker, N.; Peterson, R. E. Toxicol. Sci. 2006, 93, 223−241. (16) Organtini, K. L.; Myers, A. L.; Jobst, K. J.; Cochran, J.; Ross, B.; McCarry, B.; Reiner, E. J.; Dorman, F. L. J. Chromatogr. A 2014, 1369, 138−146. (17) Reiner, E. J.; Schellenberg, D. H.; Taguchi, V. Y.; Mercer, R. S.; Townsend, J. A.; Thompson, T. S.; Clement, R. E. Chemosphere 1990, 20, 1385−1392. (18) Myers, A. L.; Mabury, S. A.; Reiner, E. J. Chemosphere 2012, 87, 1063−1069. (19) Ontario Ministry of Environment Method DFPCB-E3418; Laboratory Services Branch, Ontario Ministry of the Environment: Toronto, ON, 2010. (20) Organtini, K. L.; Haimovici, L.; Jobst, K. J.; Reiner, E. J.; Ladak, A.; Stevens, D.; Cochran, J.; Dorman, F. L. Anal. Chem. 2015, 87, 7902−7908. (21) Aurell, J.; Marklund, S. Chemosphere 2009, 75, 667−673. (22) McKay, G. Chem. Eng. J. 2002, 86, 343−368. (23) Lohmann, R.; Jones, K. C. Sci. Total Environ. 1998, 219, 53−81.

J


Quantitative Analysis of Mixed Halogen Dioxins and Furans in Fire

Recommend Documents