Comprehensive Profiling of Coal Tar and Crude Oil ... - ACS Publications

Mar 19, 2012 - Differences between SIE and spectral deconvolution of GC/MS data based on multiple fragmentation patterns per homologue ranged from a f...
1 downloads 11 Views 2MB Size
Article pubs.acs.org/est

Comprehensive Profiling of Coal Tar and Crude Oil to Obtain Mass Spectra and Retention Indices for Alkylated PAH Shows Why Current Methods Err Christian D. Zeigler and Albert Robbat, Jr.* Tufts University, Department of Chemistry, Medford, Massachusetts, United States of America 02155 S Supporting Information *

ABSTRACT: Investigators use C1 to C4 substituted polycyclic aromatic hydrocarbons (PAH) to assess ecological risk and to track fossil fuels and related pollutants in the environment. To quantify these compounds gas chromatography/ mass spectrometry (GC/MS) is used. This work demonstrates single ion monitoring (SIM) or extraction (SIE) of full scan data produces inaccurate and imprecise concentration estimates due to incorrect homologue peak assignments. Profiling of coal tar and crude oil by automated sequential GC−GC/MS provided the retention windows and spectral patterns for each homologue to correctly quantify these compounds. Simultaneous pulsed flame photometric (sulfurspecific) detection differentiated PAH from polycyclic aromatic sulfur heterocycles and their alkylated homologues when they eluted within the same retention windows and had common ions. Differences between SIE and spectral deconvolution of GC/MS data based on multiple fragmentation patterns per homologue ranged from a few percent for C1 compounds to hundreds of percent for the higher alkylated homologues. Findings show current methods produce poor quality data adversely affecting forensic investigations, risk assessments, and weathering studies.



the expected partitioning behavior of these compounds.25,26 Thus, alkylated PAH concentrations impact site investigations, ecological risk assessments, remedial actions, and, ultimately, continuing land and water management. Similarly, forensic scientists are interested in alkylated PAHs because they serve as useful indicators of petroleum, coal tar, and creosote weathering in the environment.27,28 Investigators study PAH transport and migration pathways as well as the rate these pollutants degrade. For example, the ratio of alkylated 2ring and 3-ring PAHs to 4-ring and 5-ring PAHs suggests the extent source materials have weathered in the environment and whether it is likely the remaining compounds will evaporate, dissolve, or degrade due to local conditions.29 To this end, we showed when analysts quantify alkylated PAH in diesel fuel based on selected ion monitoring (SIM) or extraction (SIE) from full scan mass spectrometry data, estimates of concentration are biased high.30 Since most analysts use only the molecular ion for identification and quantitation, matrix components that possess the same m/z will be included in the signal, resulting in false positives. The findings also demonstrated false negatives occur when one fragmentation pattern per homologue is used. In this case, estimates of ΣPAH34 are biased low. C 4 naphthalene

INTRODUCTION Polycyclic aromatic hydrocarbons (PAH) are ubiquitous environmental pollutants. They are in fossil fuels and their byproducts,1 and are found in the environment from natural seeps2 or runoff from asphalt.3,4 Incomplete combustion of organic materials can result in transporting these compounds over long distances as gaseous molecules or organically bound particulate matter.5,6 Notwithstanding the devastating explosion of British Petroleum’s Deepwater Horizon oil rig in the Gulf of Mexico,7 oil releases occur frequently during exploration, recovery, and transport.8 For example, the ruptured pipeline in Marshall, Michigan (U.S.) released more than 800 000 gal of oil only 11 days after the Deepwater Horizon accident in the Gulf.9,10 In addition, there are tens of thousands of coal tar contaminated gas plants worldwide that are and will continue to contribute to PAH pollution.11,12 PAHs also end up in the meat, fish, vegetables, and beverages we eat and drink because they can accumulate in animals and plants.13,14 Some PAHs are toxic,15,16 mutagenic,17,18 and carcinogenic,19,20 and therefore pose risks to human health and the environment. Alkylated PAHs have been shown to contribute substantially to the toxicity of PAH mixtures, in some cases accounting for 80% of the toxic burden.21 The U.S. EPA provides guidelines toward estimating the hazards posed by contaminated soils and sediments based on the concentration of 18 parent PAH and 16 C1 to C4 alkylated homologues.22−24 Risk to the environment is quantified as a toxic unit (TU) and is calculated from PAH concentrations and © 2012 American Chemical Society

Received: Revised: Accepted: Published: 3935

September 2, 2011 February 20, 2012 March 7, 2012 March 19, 2012 dx.doi.org/10.1021/es2030824 | Environ. Sci. Technol. 2012, 46, 3935−3942

Environmental Science & Technology

Article

Figure 1. (a) GC/FID traces of coal tar (left) and crude oil (right). TIC heart-cut chromatograms correspond to samples (b) cut 20, (c) cut 34, and (d) cut 44.

exemplifies the importance of knowing all fragmentation patterns in a homologue. The relative abundance of the M-15 ion to the molecular ion is 70%, 115%, and 210% for 1,4,5,8tetramethyl, 1,2,3,4-tetramethyl, and 1-methyl-7-isopropylnaphthalene, respectively. In contrast to the first PAH, the M-15 ion is the base ion for the last two compounds. No matter which fragmentation pattern is selected, at least two isomers will be missed, which leads to lower concentration estimates for the family. On the basis of this work, it should be evident that meaningful ecological and forensic studies require multiple fragmentation patterns per homologue (MFPPH) to accurately obtain so-called PAH34 concentrations. Mass spectra for only nine of the 110 possible C 4 naphthalenes are in the literature or standard mass spec library such as NIST or Wiley. To obtain this information, we developed a library-building process that identified alkylated PAH fragmentation patterns in coal tar and crude oil based on automated sequential, multidimensional gas chromatography/ mass spectrometry (GC−GC/MS). This technique produced clean spectral patterns by transferring one minute sample

portions from the first to the second column, whose stationary phases differed in polarity. Subsequent injections occurred after each preceding analysis finished.30 Our first objective was to produce a library that contained the minimum number of spectral patterns capable of identifying all PAH34 compounds. Our second objective was to learn the magnitude of the concentration misestimates for all homologues, which was not possible in the diesel fuel study. The library reported herein gives scientists, engineers, and regulators the ability to quantify these compounds more accurately, which should lead to more defensible assessments of toxicity, fate, and transport. Results show SIM is inaccurate and imprecise compared to MFPPH.



EXPERIMENTAL SECTION Samples. A total of eight samples were analyzed. Four coal tar samples came from two different manufactured gas plants (MGP) in North Carolina and two from the same MGP release into the Hudson River (New York). Purchased from ONTA (Toronto, Ontario, Canada) were Merey and Orinoco crude 3936

dx.doi.org/10.1021/es2030824 | Environ. Sci. Technol. 2012, 46, 3935−3942

Environmental Science & Technology

Article

process. Our first objective was to identify as many alkylated PAH as possible using the least number of fragmentation patterns per homologue. Table S1 in Supporting Information, SI, lists the molecular ions, confirming ions, and relative abundances needed to detect them. Isomers reported in the literature31−36 that differed from those found in this work are also included for completeness. Listed together are those isomers that have the same fragmentation pattern, such as 1ethyl and 2-ethylnaphthalene. Table S2 of the SI lists the retention windows found for each homologue. Library Building. Figure 1a compares the GC/FID response obtained from the 2-dimensional separation of coal tar (left) and Arabian crude oil (right). The red vertical lines in the figure show the 1-min sample transfers (heart-cut), see schematic Figure S1 of the SI. Figures 1b−d are the total ion current chromatograms (TIC) for heart-cuts 20, 34, and 44. Oil cuts 34 and 44 reveal the presence of an n-alkane, whose signal is much smaller compared to other compounds in the sample, see the unfractionated sample trace in Figure S2 of the SI for comparison. Although most of the regularly spaced n-alkanes were removed during fractionation, they still pose potential matrix interferences for library building. The coal tar heart-cuts illustrate the following: (1) the successful removal of essentially all of the aliphatics and polar compounds from the aromatic fraction; (2) fewer matrix interferences than oil; (3) higher concentrations of alkylated PAH than oil; and (4) lower concentrations of unresolved complex mixture. Therefore, coal tar served as the primary source material to identify PAH fragmentation patterns. For some alkylated PAH identification without deconvolution is possible, see C2 naphthalene in cut 20. For others spectral deconvolution is necessary, see cuts 34 and 44. For example, C2 phenanthrene coelutes with the C2 and C3 3-ring PASH in cut 34. The combination of the sulfur-specific detector and the deconvolution software provided complementary means of identification. Once the fragmentation patterns were established, the PFPD was not needed to quantify the PAH by GC/MS. The criteria for the library-building process were as follows: (1) Inspection of each heart-cut peak determined if the mass spectrum at each peak scan was invariant. (2) When sample ions matched a known alkylated PAH spectrum from NIST, Wiley, or the literature, fragmentation patterns were cataloged independent of whether their relative abundances (RA) were the same as long as their reconstructed ion current (RIC) traces comaximized. (3) When signal from the matrix appeared to interfere with alkylated PAH ions, deconvolution either “cleaned” the spectra or confirmed the compound was not an alkylated PAH. (4) For those ion signals that did not match known spectra, we compared the homologue’s molecular ion plus five or more confirming ions, selected combinatorially, against the sample spectrum. In most cases, there were seven unique ions that met the relative abundance criterion. When the ion signals for these ions comaximized and produced RA differences ≥30% compared to known ion ratios, the pattern was new and cataloged. A more thorough description of deconvolution and its use for library building is in the SI. Alkylated PAH Fragmentation Patterns. Several investigators37,38 report the mechanism for the benzylium-tropylium species from toluene, ethyl benzene, and xylenes. Although the literature suggests there is competition39 between two different pathways that lead to an equilibrium40−42 of the benzylium and tropylium species, for purposes of library building, the two

oils. Zhendi Wang from Environment Canada (Ottawa, Ontario, Canada) provided a 23.2% by weight weathered Arabian crude oil. Obtained from Alberta Innovates (Edmonton, Alberta, Canada) was a 12% bitumen Athabasca oil sand sample. Instrumentation. The Gerstel (Mülheim an der Ruhr, Germany) MPS2 autosampler, CIS6 injector, MPC mass flow controller, CTS1 freeze trap/thermal desorber, and Maestro software in conjunction with Agilent Technologies (Santa Clara, CA) GC/MS models 6890/5973 and Shimadzu, Inc. (Columbia, MD) models GC2010/QP2010plus were used to construct the GC-GC/MS-PFPD, GC/MS-PFPD, and GC/MS instruments. The pulsed flame photometric detector (PFPD) was from OI Analytical (Houston, TX). Obtained from Restek (Bellefonte, PA) were two 30 m × 0.25 mm ID × 0.25 μm columns with Rxi-17Sil-ms and Rtx-5 ms stationary phases. Sample Preparation and Analysis. Details of the coal tar and oil fractionation procedure, materials purchased, and analytical standards are in Supporting Information. Also described are the setup, operation of the instruments, and the deconvolution software used to analyze the data. NIST05 and Wiley mass spectral libraries were used to identify alkylated PAH where possible.



RESULTS AND DISCUSSION Analysis of eight fresh and weathered crude oil and coal tar samples by automated sequential, multidimensional GC-GC/ MS provided the data to produce new alkylated PAH fragmentation patterns and retention windows. Simultaneous sulfur detection differentiated PAH from PASH isomers, which helped to correctly assign each peak during the library building

Figure 2. Schematic of a generalized PAH electron impact fragmentation leading to the formation of the benzylium-tropylium PAH species. R1, R2, R3 are H or (CH2)nCH3. R can be 0−3 methyl groups or an ethyl group at any site on the ring depending on R1, R2, and R3. R3 migration (1) precedes radical formation (2) of the norcaradiene-like ion. Ring-opening (3) and loss of R2 (4) creates the tropylium ion. Alternatively, cleavage of •R3 leads directly to the benzylium-PAH (5). Expulsion of acetylene from the tropylium ring (6) results in an “indene-like” PAH. 3937

dx.doi.org/10.1021/es2030824 | Environ. Sci. Technol. 2012, 46, 3935−3942

Environmental Science & Technology

Article

Figure 3. (a) Direct cleavage and (b) McLafferty rearrangement for electron impact ionization of 2-butylnaphthalene.

Figure 4. Matrix interferences in Arabian Crude Oil that affect the detection of C1 fluorene. (a) Molecular ion traces of m/z 180 (blue) and m/z 194 (purple) for C1 and C2 fluorene, respectively. Red lines indicate literature retention windows. (b) Inspection of the ion traces for m/z 180 and m/z 165 (green, confirming ion) between 22.4 and 26.4 min. The four red x peaks are not C1 fluorene and should not be included in the peak area used to determine concentration. (c) Adding another confirming ion eliminates four more peaks from the total peak area shown in (a).

channels lead to the same m/z value. Figure 2 explains the generalized benzylium-tropylium PAH formation from the molecular ion. For alkylated PAH, detection of the molecular ion always occurs but it is not necessarily the base ion, see Figure S3b of the SI. For example, multiple patterns are required to detect C2 naphthalenes, since ion ratios differ enough that a single fragmentation pattern cannot identify all isomers. Figure 2 depicts R3 fragmentation (5) forming the benzylium-PAH, or isomerization and subsequent cleavage producing the tropylium species. For the latter, the R3 group transfers (1) from the α-carbon to the ipso position before radical bond formation creates the norcaradiene-PAH (2). Ring-opening (3) creates the cycloheptatriene-like PAH ion and •R2 loss (4) produces the tropylium ion. R3 is a hydrogen for all isomers except tertiary butyl PAH; direct cleavage is the more likely mechanism when R1, R2, and R3 are methyl groups.43 For C1 isomers, it is clear why (M-1)+ is a prominent ion and (M-15)+ has a low RA (