Article pubs.acs.org/est
Molecular Evidence of Heavy-Oil Weathering Following the M/V Cosco Busan Spill: Insights from Fourier Transform Ion Cyclotron Resonance Mass Spectrometry Karin L. Lemkau,*,†,‡ Amy M. McKenna,§ David C. Podgorski,§ Ryan P. Rodgers,§ and Christopher M. Reddy† †
Department of Marine Chemistry and Geochemistry, Woods Hole Oceanographic Institution, 360 Woods Hole Rd., Woods Hole, Massachusetts 02543, United States ‡ Department of Earth Science and Marine Science Institute, University of California, Santa Barbara, California 93106-9630, United States § National High Magnetic Field Laboratory, Florida State University, 1800 East Paul Dirac Drive, Tallahassee, Florida 32310-4005, United States S Supporting Information *
ABSTRACT: Recent studies have highlighted a critical need to investigate oil weathering beyond the analytical window afforded by conventional gas chromatography (GC). In particular, techniques capable of detecting polar and higher molecular weight (HMW; > 400 Da) components abundant in crude and heavy fuel oils (HFOs) as well as transformation products. Here, we used atmospheric pressure photoionization Fourier transform ion cyclotron resonance mass spectrometry (APPI FT-ICR MS) to identify molecular transformations in oil-residue samples from the 2007 M/V Cosco Busan HFO spill (San Francisco, CA). Over 617 days, the abundance and diversity of oxygen-containing compounds increased relative to the parent HFO, likely from bio- and photodegradation. HMW, highly aromatic, alkylated compounds decreased in relative abundance concurrent with increased relative abundance of less alkylated stable aromatic structures. Combining these results with GC-based data yielded a more comprehensive understanding of oil spill weathering. For example, dealkylation trends and the overall loss of HMW species observed by FT-ICR MS has not previously been documented and is counterintuitive given losses of lower molecular weight species observed by GC. These results suggest a region of relative stability at the interface of these techniques, which provides new indicators for studying long-term weathering and identifying sources.
■
INTRODUCTION
range of weathered oil samples, were also not amenable to GC analysis.4 If a light sweet crude, such as the Macondo well oil, contains a large fraction of native oil compounds (and generates transformation products) beyond the analytical window of GC then this fraction is likely much larger for heavy crude oils or some refined petroleum products.3−4 For example, HFOs used in place of diesel fuels in power generating engines are prepared from high viscosity refinery residues enriched in high-boiling and heteroatom-containing compounds that are not amenable to GC-analysis.5 These residues are blended with low viscosity distillates to meet vessel specifications and enable transport and use. The long-term fate and environmental impact of non-GC amenable compounds following an oil spill remains largely unknown, although chemical transformation products of this
In a recent study, Incardona et al. (2012) attributed the unexpectedly high toxicity of the M/V Cosco Busan heavy fuel oil (HFO) spill (San Francisco Bay, CA; 2007) to photochemical transformation of the “chemically uncharacterized remains of crude oil refinement” and emphasized the critical need for advanced analytical techniques in oil spill science beyond conventional gas chromatography (GC).1,2 This need was further demonstrated following the Deepwater Horizon disaster where a comprehensive GC-based analysis of Macondo well reservoir fluid (oil and gas), totaling 140 target analytes (including C1−C5 hydrocarbons), could only account for ∼50% of the mass of the reservoir fluid.3 The remaining 50% of mass were therefore nontargeted or unidentifiable compounds as well as those not amenable to GC-analysis, such as some compounds containing heteroatoms (e.g., oxygen, nitrogen, or sulfur). In addition to compounds in the native oil released from the Macondo well, Aeppli et al., (2012) demonstrated that many oxygenated transformation products, found in a wide © 2014 American Chemical Society
Received: Revised: Accepted: Published: 3760
August 26, 2013 February 14, 2014 February 21, 2014 February 21, 2014 dx.doi.org/10.1021/es403787u | Environ. Sci. Technol. 2014, 48, 3760−3767
Environmental Science & Technology
Article
617 days post spill. Because each oil patch was sampled in its entirety, we were unable to sample the same exact oil patch at each time point. As a result and despite efforts to minimize heterogeneity, small variations in sample location (relative to tide line), immediate surroundings (sun/shade), and sample history (sedimentary/debris content) are expected. Here, we focus on the longer term weathering trends as revealed by the complete set of samples taken as a whole. Comprehensive two-dimensional gas chromatography (GC × GC) and various other techniques were employed to fingerprint samples and provide ancillary data for support and comparison with FT-ICR MS results. Biomarker ratios obtained from GC × GC confirmed that the analyzed field samples were from the M/V Cosco Busan spill (SI Table S1).31 GC × GC data was also used to estimate biodegradation indices.32 Elemental analysis and thin-layer chromatography with flame ionization detection (TLC-FID) were performed on all samples. TLC-FID results were normalized to the conserved biomarker 17α(H),21β(H)-hopane,33 measured in our samples via GC × GC, to provide absolute trends in fractional content of our samples.4 Details of instrumental methods for all analyses are provided in the SI (S1). Fourier Transform Ion Cyclotron Resonance Mass Spectrometry. Briefly, negative ion atmospheric pressure photoionization was coupled to FT-ICR MS for sample analysis. Because negative-ion APPI produces radical anions and deprotonated ions from polar and nonpolar species simultaneously, it is well suited for examining native or weathered oils rich in oxygen that challenge conventional ionization techniques used with FT-ICR MS (e.g., electrospray ionization) due to salt-adduct formation.34−39 Molecular ions detected by negative ion APPI FT-ICR MS of petroleum samples include carboxylic acids, five- and six-membered ring (pyrrolic and pyridinic) nitrogen, aromatic sulfur (e.g., thiophenes), and polyaromatic hydrocarbons (e.g., PAHs, PASH, PANH, PAOH). Only those hydrocarbon species with acidic protons that can stabilize an anion or radical (e.g., fluorene, pKa ∼23) will be detected. FT-ICR MS experimental methods and selection of ionization mode are detailed in the SI (S2 and S3).
fraction have been implicated in oil spill toxicity even beyond that observed by Incardona et al., (2012).6−8 There are numerous approaches to expand beyond the analytical window of traditional GC-based techniques, including liquid chromatography, high temperature gas chromatography, nuclear magnetic resonance, infrared spectroscopy, stable isotopes, natural abundance radiocarbon, elemental analysis, and thin layer chromatography (TLC).4,9−11 However, all of these techniques are limited by resolution and spectral complexity when considering complex mixtures. A recent paper by McKenna et al., (2013) argued that ultrahigh resolution Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR MS) is ideally suited to investigate molecular-level changes of non-GC amenable compounds within complex mixtures.12 This technique requires no prior sample separation and provides the resolving power (m/ Δm50%, of 1 400 000 at m/z 400) needed to assign molecular formulas to the tens of thousands of compounds within a complex petroleum sample.12,13 Although FT-ICR MS has been used extensively to address heavy whole crudes,12−16 refinery deposits15,17−19 and changes due to various refinery operations,18,20−24 application to environmentally weathered oil samples has been limited.24−29 Following the recommendations of Incardona et al., (2012) and previous findings on the Macondo well oil, we sought to expand our analytical window to examine the M/V Cosco Busan oil spill using FT-ICR MS. The M/V Cosco Busan spill occurred in November 2007 when the vessel struck the San FranciscoOakland Bay Bridge and released approximately 200 000 L of HFO into San Francisco Bay. We used negative ion atmospheric pressure photoionization (APPI) FT-ICR MS to analyze a timeseries of oil samples collected over 617 days following the spill. This is a particularly beneficial opportunity to consider new approaches as we have already surveyed this spill using traditional GC methods.30 Using FT-ICR MS, we identified molecular transformations as a function of environmental exposure relative to the parent HFO and highlight compositional trends as a function of time. We also compared insights from FT-ICR MS to our previous GC-based work on this spill.30 In doing so, we characterize molecular changes in the oil from the M/V Cosco Busan spill both in the traditionally examined GC-amenable compounds and in the typically overlooked higher MW compounds.
■
RESULTS AND DISCUSSION We used bulk analyses, GC-derived weathering indicators, and ultrahigh resolution FT-ICR MS to provide a comprehensive understanding of weathering following the M/V Cosco Busan HFO spill. Bulk and GC-based analyses supplied basic weathering information commonly gathered after a spill and offered a point of comparison for trends observed by the more novel application of FT-ICR MS. Unlike traditional bulk techniques, application of FT-ICR MS to the parent HFO and weathered samples allowed us to highlight and identify compositional changes at the molecular level for large (>∼200 Da) heavily oxidized compounds not previously examined to gain insight into the complex transformations of hydrocarbons in the environment. Biodegradation Indices. Samples were categorized on the Peters and Moldowan (PM) biodegradation scale based on the presence/absence of compounds found in oil residues relative to the native oil.32 This scale ranges from 0 (nondegraded) to 10 (severely degraded). Though developed to examine biodegradation of reservoir fluids, it has been used extensively to compare the degree of degradation in environmental samples, where multiple degradation pathways (abiotic and
■
MATERIALS AND METHODS Sample Collection. Oil samples were collected from a rocky shoreline at Shorebird Park (Berkeley, CA; Supporting Information (SI) Figure S1) following the 2007 M/V Cosco Busan oil spill, as described elsewhere.30 Oil residues were scraped from boulders with a stainless steel spatula and stored in combusted Al-foil envelopes. Numerous scrapings from an individual oil patch were combined and considered as one sample to overcome sample heterogeneity. Samples were transported to Woods Hole, MA, frozen (−20 °C) and subsequently shipped to the National High Magnetic Field Laboratory in Tallahassee, FL. The unweathered parent HFO was provided by the National Oceanic and Atmospheric Administration.30 Gas Chromatography and Bulk Oil Composition Analyses. Six representative samples were selected for FTICR MS analysis based on general weathering trends observed by GC with flame ionization detection (FID) in a previous study.30 They were oil residues obtained 55, 139, 292, 511, and 3761
dx.doi.org/10.1021/es403787u | Environ. Sci. Technol. 2014, 48, 3760−3767
Environmental Science & Technology
Article
Table 1. Saturate, Aromatic, And Oxygenated Fractional Composition As Determined by TLC-FID,4 Elemental Composition (Weight Percent by Mass) And Biodegradation Indices of the Parent HFO and Field Samples saturatea aromatica oxygenated 1a oxygenated 2a carbon hydrogen nitrogen sulfur oxygen ash H:C ratio biodegradation indexb a
parent HFO
day 55
day 139
day 296
day 402
day 511
day 617
22% 47% 12% 19% 85% 10% 0.53% 1.8% 1.8%
18% 36% 22% 24%
19% 35% 21% 26% 67% 8.6% 0.59% 1.8% 10.5% 9.4% 1.54 2
17% 18% 31% 34% 79% 9.5% 0.76% 1.9% 2.6% 6.6% 1.43 4/5
17% 15% 30% 38% 80% 9.3% 1.0% 1.5% 6.2% 2.3% 1.39 4/5
16% 30% 26% 28% 79% 9.4% 0.77% 1.8% 5.3% 3.0% 1.42 5
15% 7% 34% 44% 77% 9.2% 1.0% 1.7% 6.6% 3.9% 142 5/6
1.43 0
2
Average errors based on triplicate analysis are ±5%. bDeveloped by Peters et al. (2005) based on hydrocarbon biomarkers.32
biotic) occur.32,40 We observed that heavily weathered samples based on GC analyses were more biodegraded, according to the PM scale (Table 1). After 55 and 139 days, samples had signs of light biodegradation (biodegradation index of 2). By 296 and 402 days postspill, samples were moderately degraded with indices of 4−5. After 617 days, field samples were moderately to heavily biodegraded (5/6).32 Elemental Analysis. Bulk elemental analysis detected global changes in total weight percent carbon, hydrogen, nitrogen, oxygen, and sulfur (Table 1). Considering the ±0.3% certainty in these measurements, the total mass fraction of nitrogen within the parent HFO (0.53% by weight) did not change with extended time post spill (1.0% at 617 days). Similarly, sulfur content of field samples (1.5−1.9%) did not vary from that of the parent HFO value (1.8%). The largest exposure-related shift was the increase in oxygen. The oxygen content in the parent HFO (1.8% by weight) more than tripled by 617 days post spill (6.6% by weight). This observed increase in oxygen content was consistent with oxygen-species formation through environmental exposure.4,7,41,42 The sample collected after 139 days contained nearly twice the oxygen content of the 617 day sample (10.5 versus 6.6%), inconsistent with other measures of oxygenation for this sample. Based on the high ash content of this sample (9.4%), we believe the elevated oxygen is from calcium carbonate contamination. Accounting for the latter, this sample had an oxygen content of ∼7.5%, more in line with data across the timeseries (see SI S4 for a more detailed discussion on this anomaly). Thin-Layer Chromatography. TLC-FID analysis also supported an increased prominence of oxygenated fractions with weathering, consistent with previous studies.4,43−46 Similar to Aeppli et al., we used TLC-FID to classify sample extracts into four fractions based on order of least retention (see SI S1 and Aeppli et al., 2012 for further details): saturates, aromatics and the oxygen-containing fractions (oxygenated 1 and 2). The saturate and aromatic fractions are considered GC-amenable. While some minor components within the oxygenated fractions can be detected using GC (e.g., esters and ketones), in general, these fractions are not amenable to GC analysis.4 Table 1 reports weight percent of each fraction for parent HFO and field samples. An increase in the oxygenated fractions occurred across the timeseries. Oxygenated fractions accounted for 31% of the total mass of the parent HFO. In samples collected 296 and 617 days after the spill, 65 and 78% of the total mass eluted in the oxygenated fractions, respectively.
The increase in oxygen-containing compounds may have derived from photo- and biodegradative chemical transformations.4,7,41,46 This increase in the oxygenated fractions was also consistent with the observed increase in oxygen content via elemental analysis and with biodegradation indices. These bulk measurements support general oxidation with environmental exposure but do not provide information regarding the molecular-level changes occurring within our samples. We used ultrahigh resolution FT-ICR MS to investigate these compositional changes on a molecular level. Ultrahigh Resolution Fourier Transform Ion Cyclotron Resonance Mass Spectrometry. Here we used FT-ICR MS to detect oxygenated compounds, as well as many others, providing a molecular snapshot at the compositional changes within our samples. For example, the molecular weight distribution observable within the raw broadband mass spectra shifted to from a distribution centered on ∼650 m/z in the parent HFO to one at ∼325 m/z by 617 days post spill (SI S5, Figure S2). Heteroatom Class Distribution. Using assigned molecular formulas for each m/z peak, compounds were grouped into “classes” within a sample based on their heteroatom content. Compositional trends that among samples were visualized with heteroatom-class distribution graphs. Figure 1a shows the heteroatom class distribution for select species of >1% relative abundance in the APPI FT-ICR mass spectra for the parent HFO and weathered oil samples (55−617 days). Because there is no “hopane” equivalent for FT-ICR MS that allows normalization across the data set,33 mass spectral peaks were normalized to the most abundant peak in each mass spectrum. Assuming similar ionization efficiencies of compound classes across samples (i.e., minimal matrix effects), observed changes across heteroatom classes show trends related to greater environmental exposure. Though APPI is not immune to matrix effects, it is less susceptible than other ionization techniques. Observation of trends consistent with TLC-FID and elemental analysis suggests that matrix effects are not overwhelming the signature of our samples. Compounds containing one nitrogen atom (N1 class) were the most abundant in the parent HFO, followed by the O1, N1O1, and O2 classes (Figure 1a). However after 617 days the O2 class was the most abundant. A decrease in the relative abundance of low heteroatom-containing compounds (e.g., hydrocarbons and N1) also positively correlated with increased time post spill. 3762
dx.doi.org/10.1021/es403787u | Environ. Sci. Technol. 2014, 48, 3760−3767
Environmental Science & Technology
Article
abundance and diversity of SxOy and NxOy classes also occurred with weathering. The relative decrease of the O1 class corresponded to a relative increase in the O2 class, a trend consistent with environmental oxidation of alcohols, aldehydes, and/or ketones to form acids.25 Compositional Degradation of the Parent Heavy Fuel Oil. FT-ICR MS routinely measures masses to within 50−300 ppb and when combined with Kendrick mass sorting, which uses homologous series to extend accurate mass assignments to higher molecular weights than otherwise possible, and isotopic fine structure enables assignment of unique elemental compositions to each molecule.47 Double bond equivalents (DBE; the number of rings plus double bonds to carbon) can be useful in examining compositional differences across samples. Because each ring or double bond in a molecule reduces its number of hydrogen atoms by two, the DBE of a molecule can be calculated from its elemental composition.48 Isoabundance contoured plots of DBE versus carbon number for the three most abundant classes in the parent HFO (N1, N1O1, and hydrocarbon classes) highlight compositional changes that occurred over 617 days (Figure 2). Each of these classes showed a decrease across the timeseries in the average carbon number weighted by relative-abundance (hereafter referred to as average carbon number). For example in the N1 class, the average carbon number decreased from C49 in the parent HFO to C32 in the 617 day sample, a loss of 17 carbon atoms per structure (Table 2). Aromaticity remained essentially constant with an average DBE weighted by relative abundance (hereafter referred to as average DBE) of 17 in the parent HFO decreasing to 16 in the 617 day sample. However, though average aromaticity remained stable, we observed the formation of local maxima around specific DBEs (Figure 2; Table 2). The winnowing down of the diverse structures within the parent HFO indicates that environmental weathering results in preservation and/or formation of relatively stable compounds with specific aromaticity characteristics. The observed changes in each compound class are the result of a balance of processes that both degrade and form new relatively stable compounds with weathering. Where multiple stable structures were observed, they often differed by three DBE units, equivalent to the loss or addition of an aromatic ring to an already aromatic system. For a given DBE, the spread of
Figure 1. Heteroatom class distribution (heteroatom content) of select species for the parent HFO and field samples from negativeAPPI FT-ICR MS (a) select species of >1% relative abundance in the parent HFO and (b) expanded view of oxygen-containing heteroatom classes.
Oxygen. Heteroatom classes that contain one or more oxygen atoms are highlighted in Figure 1b for the parent HFO and the 296 and 617 day samples. Weathered field samples contained higher-order oxygenated species (up to O8) not detected in the parent HFO. An increase in the relative
Figure 2. Isoabundance-contoured plots of double bond equivalents (DBE) versus carbon number for the most abundant heteroatom classes (shown on right) within the parent HFO for the parent HFO and field samples collected from 55 to 617 days postspill. Each compositional image is normalized to the most abundant species within that heteroatom class for that sample. The three heteroatom classes presented are the N1, N1O1 and HC classes. Average carbon number and DBE values for parent HFO and the 617 day sample are listed in Table 2 3763
dx.doi.org/10.1021/es403787u | Environ. Sci. Technol. 2014, 48, 3760−3767
Environmental Science & Technology
Article
Table 2. Summary Table of Average Carbon Number (C #), Average Neutral Species DBE and Core Structures Found in Parent HFO and 617 Day Samples for Each Heteroatom Class parent HFO
617 days
heteroatom class
average C # (range)
average DBE (range)
average C # (range)
average DBE (range)
N1 N1O1 HC O1 O2 O3 O4
49 (19−84) 48 (15−82) 49 (23−88) 48 (15−82) 46 (14−80) 33 (12−57) ndab
17 (8−33) 18 (8−32) 19 (9−32) 14 (5−31) 14 (2−31) 14 (2−28) ndb
32 16−58) 29 (15−56) ndc 27 (13−55) 27 (12−55) 27 (12−55) 27 (12−55)
16 (10−27) 16 (8−32) ndc 14 (3−29) 10 (2−28) 12 (2−27) 12 (2−27)
a
nd = not detected. bAverage C# and DBE (and ranges) at 296 days post spill were 33 (12−73) and 13(2−28), respectively. cAverage C# and DBE (and ranges) at 511 days post spill were 35 (20−60) and 18 (10−30), respectively.
Figure 3. Isoabundance-contoured plots of double bond equivalents (DBE) versus carbon number for the O1 to O4 heteroatom classes (shown on right) for the parent HFO and field samples collected from 55 to 617 days postspill. Each compositional image is normalized to the most abundant species within that heteroatom class for that sample. Average carbon number and DBE values for parent HFO and the 617 day sample are listed in Table 2
highly alkyl-substituted core structures present in the parent HFO. Increased Abundance of Polar Compounds. Oxygencontaining compounds increase in abundance with extent of weathering,4,7,43 however, molecular characterization of newly formed oxygen containing compounds has been limited to the small fraction that is accessible with traditional GC-based techniques. Figure 3 shows DBE versus carbon number plots for the O1, O2, O3, and O4 classes for the parent HFO and weathered oil over 617 days. Compounds that contain one oxygen atom (O1) showed no overall shift in average DBE (14 in both the parent HFO and after 617 days) but carbon number decreased by an average of 19 carbons per molecule. The O2 class more than doubled in relative abundance (Figure 1) from 5 to ∼11% after 617 days. Within the O2 class, both carbon number and DBE showed marked decreases from the parent HFO values (C46 and 14 DBE) to 617 days post spill (C27 and 9 DBE). A bimodal distribution was observed in the O2 class (Figure 3), with compounds with 2 DBE corresponding to naphthenic acids (naturally occurring alkyl-substituted saturated cyclic and noncyclic carboxylic acids), as previously observed in heavily biodegraded crude.25 This bimodal distribution is thought to be the result of two different contributions to the O2 class. First, higher DBE species
carbon numbers represents changes in alkylation; thus the N1 species of highest relative abundance after 617 days correspond to dealkylated homologues of compounds in the parent HFO with equivalent DBE values. The N1O1 and hydrocarbon classes showed similar weathering trends to those observed in the N1 class. The average carbon number of the N1O1 class in the parent HFO decreased from C48 to C29 by 617 days post spill. The average DBE also decreased slightly from 18 to 16. In the parent HFO, compounds in the hydrocarbon class have an average of 19 DBE that decreased to 18 by day 511. Hydrocarbons present after 511 days contained an average of 14 fewer carbons than those within the parent HFO across the same DBE range, a trend consistent with dealkylation of hydrocarbon compounds with weathering. The trend of decreasing carbon number with weathering indicates that lower molecular weight compounds observed by FT-ICR MS are more recalcitrant than their higher molecular weight alkylated homologues present within our samples. Because oil spill studies based on GC analysis generally show loss of lower molecular weight species this result is counterintuitive and is a weathering trend not previously documented following an oil spill. In each of these three most abundant classes, the compounds present after 617 days were condensed, aromatic homologues of more aliphatic, 3764
dx.doi.org/10.1021/es403787u | Environ. Sci. Technol. 2014, 48, 3760−3767
Environmental Science & Technology
Article
observed within the parent HFO follow the weathering trend of all the aromatic species as observed in the other classes. Second, oxidation of the aliphatic hydrocarbons also occurs over time and results in formation of oxygenated compounds with lower DBE values. The decrease in average carbon number is less prominent in the O3 and O4 classes than in other classes, with average carbon number decreasing only by 5 for the O3 class. O4 species, while not detected in the parent HFO, were present in the 296 day sample, and a decrease in average carbon number (from 33 to 27) was observed by 617 days post spill. Both classes showed overall decreases in average DBE and formation/preservation of stable structures (Table 2). Compounds that contain O3 and O4 constituted less than 3% each of the total abundance in the parent HFO but increased to 10 and 8% respectively, by 617 days post spill. These newly formed, highly oxygenated compounds are relatively resistant to degradation,4 comprised of mixed phenol, ketone and carboxylic acid species, and likely correspond to compounds formed through the biotic and abiotic oxidation of native HFO compounds, as previously reported.4,7,41,43,49 Weathering Processes. Previous, GC-based analysis of field samples from Shorebird Park showed signs of biodegradation (decreasing n-C18/phytane ratios) beginning about one month post spill, with no significant photodegradation detected at this site.30 However, while traditional GC-based alkane and PAH ratio techniques provide a strong basis for understanding the oil weathering, these ratios can become unusable within only a few months after a spill, and as suggested by the work of Incardona et al., they likely do not tell the complete story.30,50,51 For example, larger aromatic compounds outside the analytical window of GC may be more susceptible to photodegradation due to their greater absorbance of environmentally relevant wavelengths. In the sample set presented here using FT-ICR MS, dealkylation trends and increasing oxygen content consistent with biodegradation and/or photodegradation are observable through 617 days post spill. Evidence for Biodegradation. In addition to increased oxygen content as measured by elemental analysis and TLCFID, with FT-ICR MS we also observed molecular-level changes consistent with biodegradation. The relative abundance of oxygen-rich compound classes increased, indicating that not only did the oxygen content as a whole increase but that many compounds contained more than one oxygen atom. Dealkylation trends consistent with biodegradation, were also observed across all classes examined. Dealkylation resulting from biodegradation has been previously identified by FT-ICR MS in genetically related reservoirs with varying degree of biodegradation,26 but over much longer time scales. Here FTICR MS provided molecular-level insight into weathering trends consistent with biodegradation long after traditional GCbased indicators (e.g., pristine, phytane, and associated nalkanes) have been degraded. Because of the expanded analytical window afforded by FT-ICR MS, compositional changes consistent with biodegradation were observed through 600 days post spill. Evidence of Photodegradation. Dealkylation is also consistent with photodegradation, which preferentially degrades alkyl-substituted PAHs.7,4145 This weathering pattern was present in all examined compound classes. Also, the increased abundance of O2 over time could correlate with carboxylic acid formation through photodegradation.41,52
Photodegradation has been implicated in the unexpected toxicity of the M/V Cosco Busan spill,2 though photodegradation indicators were not observed by GC-based techniques.30 FT-ICR MS accesses higher carbon number, more aromatic species that will have greater absorption of environmentally relevant wavelengths and thus (barring matrix effects) be more susceptible to photodegradation, than lower molecular weight, GC-amenable PAHs.53 Therefore, these larger compounds detectable using FT-ICR MS may provide more sensitive indicators of photodegradation than currently available indicator ratios which utilize smaller PAHs. This study is the first temporal, molecular-level examination of the non-GC amenable oil fraction following a spill. Through a combined technique approach, FT-ICR MS and GC-based characterization provides unprecedented qualitative and quantitative molecular level insight into heavy oil transformation as a function of time. With environmental weathering, GC-based trends generally showed a loss of lower molecular weight (
