2011 INTERNATIONAL OIL SPILL CONFERENCE
1 2011-428
Tracking and Modeling the Degradation of a 30 Year Old Fuel Oil Spill with
Comprehensive Two-Dimensional Gas Chromatography
March 14, 2011
Glenn S. Frysinger1, Gregory J. Hall
1, Ariana L. Pourmonir
1, Heather N. Bischel
2, Emily E.
Peacock2, Robert N. Nelson
2, Christopher M. Reddy
2
1Department of Science, U.S. Coast Guard Academy, 27 Mohegan Ave. New London, CT 06320
2Department of Marine Chemistry & Geochemistry, Woods Hole Oceanographic Institution, 266
Woods Hole Rd. Woods Hole, MA. 02543
ABSTRACT
On October 9, 1974, the barge Bouchard 65, carrying 12 million liters of No. 2 fuel oil,
spilled an undetermined amount of oil off the west entrance of the Cape Cod Canal in Buzzards
Bay, Massachusetts, USA. Wind patterns and currents caused significant oiling of the nearby
Winsor Cove salt marsh. Fortunately, an original Bouchard 65 cargo oil sample was retained
from the spill which offers a unique opportunity to compare in vitro weathering experiments with
petroleum that has been weathered naturally for over thirty years. Samples of the original
product were bio-weathered in the lab over a period of days, and then these samples, plus a
contemporary sample from Winsor Cove, were analyzed by Comprehensive two-dimensional gas
chromatography with Time of Flight Mass Spectrometry detection (GC×GC-MS). The data
from the laboratory experiment was used to create a Principal Components Regression model to
predict amount of weathering. The environmental sample was projected onto the regression
model and fit most well at approximately 13 days of laboratory weathering time. Examination of
the regression vector from PCR shows mass ions related to polar and volatile compounds
decreased first, while mass ions and peaks related to sesquiterpanes persisted during the
weathering process.
INTRODUCTION
On October 9, 1974, the barge Bouchard 65, carrying 12 million liters of No. 2 fuel oil,
spilled an undetermined amount of oil off the west entrance of the Cape Cod Canal in Buzzards
Bay, Massachusetts, USA. Wind patterns and currents caused significant oiling of the nearby
Winsor Cove salt marsh, (Hampson and Moul, 1978). Studies conducted in the first three years
after the spill found that oil persisted in the marsh, but some oil components such as alkyl
substituted PAHs had decreased (Teal et al., 1978). Sediment cores collected between 2001 and
2005, 30 years after the spill, showed that heavily weathered and degraded oil was still
detectable in the upper 4 cm of the marsh sediment. Total petroleum hydrocarbons (TPHs) were
found to be 8.7 mg g-1
(dry weight) and polycyclic aromatic hydrocarbons (PAHs) were 16.7 μg
g-1
for total alkyl-substituted naphthalenes and phenanthrenes (Peacock et al., 2007).
Fortunately, an original Bouchard 65 cargo oil sample was retained from the 1974 spill so
comparisons can be made between the chemical composition of the original oil and the degraded
oil remaining in the Winsor Cove sediments after 30 years. In laboratory experiments, the
Bouchard 65 No. 2 fuel oil was biodegraded under aerobic, nutrient-rich conditions in order to
2011 INTERNATIONAL OIL SPILL CONFERENCE
2 2011-428
simulate the biodegradation that occurred at Winsor Cove. A degradation model can be
constructed from the laboratory experiments to calibrate the extent of natural weathering of
petroleum compounds in Winsor Cove sediments.
Comprehensive two-dimensional gas chromatography (GC×GC) produces a high resolution
separation in order to identify and quantify individual compounds in complex samples like
petroleum. GC×GC is a two column gas chromatography method where the first column is
typically nonpolar for a volatility-based separation, and the second column is polar for a polarity-
based separation. Coeluting analytes from the first column are periodically trapped and
concentrated with a modulator and then injected (3 s cycle time) into the second column for a
rapid separation via a different retention mechanism. While single column GC can resolve 100s
of peaks, GC×GG separations have a peak capacity that is a product of the two columns so 1000s
of peaks are routinely resolved. A GC and GC×GC-MS total ion chromatogram of a fuel oil is
shown in Figure 1. The x-axis shows a volatility based separation. The C9 to C21 n-alkanes are
seen ranging from left to right as large equally spaced peaks. The y-axis shows the rapid polarity
based separations that effectively resolve the first column coelutions. Peaks are reconstructed as
a color contour plot from the total ion signal of hundreds of second-column chromatograms. As
second-column retention time increases, the one-and two-ring cycloalkanes, and the one- two-
and three-ring aromatics are separated. Examination of each vertical slice shows that multiple
coeluters have been resolved. In addition to producing more separated peaks per time, GC×GC
is known for its ability to separate and group chemical compounds by class (Dimandja, 2004,
Gaines et al, 2007). If coupled to a time-of-flight mass spectrometer, there is a full scan mass
spectrum for each GC×GC separated peak. The mass spectra are interference free because
components are chromatographically separated prior to mass spectrum measurement.
The data density of GC×GC-MS chromatograms is high. A 90 minute chromatogram as shown
in Figure 1 will consist of 540,000 individual mass spectra (100 Hz, m/z 45-250) for more than
110 million data points per chromatogram. Chemometric approaches that use multivariate
statistical analyses are a proven method to analyze large data sets and produce an unbiased
determination of which chromatogram data describe the variance between multiple GC×GC-MS
chromatograms. Both principal component analysis (PCA) and principal component regression
(PCR) methods will be used to analyze the fuel oil biodegradation time series. PCA is a method
that reduces data dimensionality to identify a reduced number of orthogonal variables (PCs) that
describe the majority of the data variance. The results are displayed as two- or three-dimensional
scores plots that frequently show grouping of related samples. In addition to the scores, the
loadings identify which variables (such as which chromatogram peaks) contribute most to the
variance. In some cases, samples are not distinctly ordered in the PCA scores plot according to
the specific characteristic of interest, so regression (PCR) is required to determine the linear
combination of the PCs that captures the variance in the regression variable, which is typically
called the “Y block”. Many studies have been published using chemometric techniques for
fingerprinting or classification, only a few have focused on regression (Pierce et al., 2008;
Mispaleer et al., 2003).
The goals of the described experiments include:
1.) The use of GC×GC-MS analysis to describe the weathering loss of chemical
compounds from a fuel oil after 30 years of natural degradation.
2011 INTERNATIONAL OIL SPILL CONFERENCE
3 2011-428
2.) The completion of laboratory degradation experiments to simulate the biodegradation
of fuel oil for comparison to petroleum weathered in the environment.
3.) The application of PCA and PCR chemometric methods to assist these analyses and to
calibrate the rate of degradation in laboratory experiments.
METHODS
Incubations
Live and Control incubations of the Bouchard 65 No. 2 fuel oil cargo were prepared in 4-
L amber glass bottles with 200-300 g of 1 mm-sieved sediment (dry weight) and 700-1000 mL
autoclaved, 0.2 μm-filtered seawater. Oil-free sediment samples containing natural oil-degrading
microbes were collected from Wild Harbor Marsh site M-1. Nutrients as 0.25g NH4Cl and 0.2g
KH2PO4 were introduced to each incubation. The Control incubation was autoclaved (40 min/40
min exhaust) and spiked with 50 g azide dissolved in seawater (10g added initially, 40g added
after 2 weeks). Live and Control incubations were spiked with 1.2 g neat Bouchard 65 No. 2
fuel oil. The incubations were covered with foil to block out light, stirred for the duration of the
experiment, and kept under humidified air (approximate flow rate 180 mL min-1
). Duplicate
sediment samples were taken from each incubation at Day 0, 6, 11, 20, 36 and 46. One gram of
air-dried sediment was spiked with Hexatriacontane (n-C36, 15 μg) and extracted with
dichloromethane:methanol (9:1) by accelerated solvent extraction (100 C, 1000 psi). Extracts
were concentrated by rotary evaporation and percolated through a sodium sulfate column eluted
with dichloromethane to remove water. Extracts were reduced in volume (≤ 1mL).
GC Analysis
The samples were analyzed on an Agilent 6890 series gas chromatograph with a Flame
Ionization Detector (FID). A 1 uL sample was injected splitless to a polydimethylsiloxane
capillary column (J&W DB-5MS, 60.0m, 0.32-mm I.D., 0.25-um film) with GC oven
temperature, 60°C (1 min), 20°C min-1
to 80°C (3 min), 5°C min-1
to 320°C (20 min). Helium
was the carrier gas at 1.5 ml min-1
constant flow. Total Petroleum Hydrocarbon (TPH)
quantification was determined with calibration solutions of n-C36, stearyl palmitate and
Bouchard 65 No. 2 fuel oil.
GC×GC-MS
The GCGC-MS system used to analyze the sediment extracts was a Leco Pegasus 4D
system consisting of an Agilent 6890 gas chromatograph configured with a split injector, two
chromatography columns, a liquid nitrogen-cooled pulsed jet modulator, and a unit mass
resolution time-of-flight mass spectrometer. A 1.0 μL sediment extract sample was injected into
a 250 C split injector with a 20:1 split ratio. The first-dimension separation was performed on a
nonpolar polydimethylsiloxane phase (Phenomenex, ZB-1, 30.0 m, 0.25 mm I.D., 1.0-μm film)
with GC oven temperature 40°C (1 min) , 3 C min-1
to 300 C. The modulation column was
deactivated column (1.0 m, 0.10 mm I.D.) with temperature 140°C (1 min), 3 C min-1
to 400
C. The second-dimension separation was performed on a polar 50% phenyl equivalent
polysiloxane phase (SGE, BPX-50, 3.5 m, 0.10 mm I.D., 0.1 μm film) with GC oven temperature
40°C (1 min), 3 C min-1
to 300 C. Hydrogen was the carrier gas at 1.2 mL min-1
constant
flow. The GC×GC modulator period was 3 s. The ToF detector collected spectra from m/z 45 to
250 at 100Hz.
2011 INTERNATIONAL OIL SPILL CONFERENCE
4 2011-428
Chemometrics
GC×GC-MS data was baseline corrected with CGImage, v2.1 and imported as comma
separated value (.csv) file to MATLAB v 2010b where it was compiled into a single dataset
object (DSO) with PLS_Toolbox®
v6.0.1 published by Eigenvector Research Inc. A smaller
region from 46.4 to 57.6 min range on the first dimension and 0.2 to 2.5 s on the second
dimension was selected for PCA and PCR analysis. The reduced range focused on a region of
the chromatogram with high peak density and significant changes caused by biodegradation. In
the mass spectra, only ions that represent specific petroleum classes and useful for oil
fingerprinting were included in the data set (ASTM, 2006; ASTM, 2010). The ions and
associated chemical classes are listed in Table 1. The final DSO contained seven GC×GC-MS
chromatograms with 224 points in the first dimension by 231 points in the second dimension by
27 selected mass ions. Prior to the fit of a chemometric model the data was pre-processed by
unit normalization to offer equal leverage to each sample, and then mean-centered to center the
data at the origin. The regression or “Y block” (predicted variable) was mean-centered to match.
RESULTS / DISCUSSION
Figures 2a-c show the gas chromatograms for the Live incubation results for the
Bouchard 65 fuel oil sample at Day 0, 11 and 46 respectively. Prevalent n-alkane and branched
alkane peaks apparent in Day 0 are significantly smaller by Day 11 and nearly indiscernible by
Day 46 of the experiment. As the prominent n-alkanes are degraded, the chromatogram
transforms to an unresolved complex mixture (UCM) hump with few chromatographically
resolved peaks. Figures 3a-b for the Control incubation show persistence of the n-alkanes
throughout the 46 day incubation period. Comparison of the Live and Control experiment can be
used to determine the factors contributing to degradation. The Control experiment shows an
overall loss of fuel oil mass and a preferential loss of the more volatile, lower molecular weight
compounds. This weathering loss can be attributed to evaporation. The Live experiment with
the loss of n-alkanes shows the additional contribution of microbial degradation. A plot of the n-
C18 to phytane peak ratios versus time in Figure 4 reveals a significant decrease in the ratio for
the Live incubation within the first 11 days of the experiment and a continued downward slope
through Day 46 suggesting strong microbe activity. Branched isoprenoid alkanes such as
phytane are generally more difficult for microorganisms to degrade than straight chain straight
chain alkanes, so a decrease of n-C18 relative to phytane is a classic indication of microbial
activity. The ratio remains relatively constant in the Control incubation, confirming the absence
of microorganisms.
Two-dimensional GC×GC-MS total ion chromatograms are shown in Figures 5a-d. Panel 5a is
the Bouchard 65 cargo oil, panels 5b and 5c are the Day 11 and Day 46 Live incubations, and
panel 5d is the Winsor Cove (0-1 cm) sediment sample. Comparison of the Bouchard 65 cargo
oil with the Day 11 incubation shows that there is a near total loss of compounds smaller than n-
C15 and a complete loss of alkyl substituted benzenes, naphthalenes, and phenanthrenes across
the whole chromatogram. The Day 46 incubation shows even more significant weathering. The
most prominent peaks remaining are the pristane and phytane isoprenoids (marked with solid
circles) and a series of C5- and C6-subsituted decalins in the C15 – C16 retention index range
(marked with dashed circle). These compounds are commonly known as sesquiterpanes. This
chromatogram region is expanded in Figures 6a-c as m/z 123 extracted ion chromatograms. The
sesquiterpanes are not among the most abundant compounds in the fresh oil sample, but after 46
2011 INTERNATIONAL OIL SPILL CONFERENCE
5 2011-428
days of evaporation and biodegradation they are the most volatile compounds still remaining and
are among the most intense peaks on the chromatogram (Fig 6b). Comparison of these
chromatograms to a Winsor Cove (0-1 cm) sample 30 years after the 1974 spill confirms the
persistence of these classes of compounds in the marsh sediment (Fig 6c). The recalcitrance of
these compounds despite a loss of equally and less volatile compounds in the laboratory
degradation experiment as well in field conditions leads us to the question of why the molecules
are especially persistent. Structural complexity may limit the availability of these compounds to
microorganisms, as the methyl groups surrounding the decalin backbone appear to serve as
protection from microbial degradation. The sesquiterpanes serve as a useful biomarker for fuel
oil fingerprinting, even after significant weathering has occurred (Gaines et al., 1999, Gaines et
al., 2006, Stout et al., 2005, Wang et al., 2005).
While these qualitative comparisons are informative, a quantitative method for comparing
laboratory weathering to that in the environment would also be useful. For that comparison,
chemometric regression techniques were used. The first step in the chemometric analysis was to
fit a Principal Components Analysis (PCA) model to the data to capture the primary directions of
variance in the data. During this step the number of Principal Components (PCs) chosen in the
model was based on the total variance captured by successive additional PCs. The six GCxGC-
MS chromatograms for the laboratory incubation data experiment were fit with 4 PCs that
captured 95.97 percent of the variance in the data. The scores of PC1, PC2, and PC3 for the
incubation experiment are shown in Figure 7. The Winsor Cove sample is projected on the PCs.
There is no apparent grouping or trend in the data based on the PCA analysis. This is not
unexpected because all samples represented were chemically identical prior to laboratory
incubation or 30 years of natural weathering.
Principal Components Regression (PCR) is a second step after PCA. A vector comprised of a
linear combination of the PCs was calculated to capture the highest variance in the Y block.
Figure 8 shows the PCR results. The x-axis is the measured duration of weathering, a
combination of evaporation and biodegradation for the Live Bouchard 65 oil incubation. The y-
axis is predicted weathering from the PCR model. The regression vector fit to the six solid
diamond data points in this model captured 99.71 percent of the “Y block” variance. The open
squares show the cross-validated fit of each data point to the PCR model with error bars
calculated for each sample based on the residuals for each sample scaled by each sample’s
leverage on the overall model using the method described by Faber and Bro (2002). The
naturally weathered Winsor Cove sample was projected on the PCR model and had weathering
most closely described by 13 days of laboratory incubation. This means that the chemical
composition of the Winsor Cove sample was most similar to that obtained after 13 days of
evaporation and nutrient-rich aerobic biodegradation in the laboratory. During the entire fitting
process the three dimensional data must be “unfolded” into one long vector (1,397,088 data
points for each sample) in order for PCR to be calculated properly. Since the loadings are
vectors in the same variable space as the data, the loadings can be refolded after analysis and
visualized the same way as the original GC×GC-MS chromatogram data.
When interpreting the loadings, it should be noted that each pixel can have either a positive or
negative value. The relationship between the signs of these values is the important
consideration, not necessarily the sign itself. For example, a peak with a positive value will be
2011 INTERNATIONAL OIL SPILL CONFERENCE
6 2011-428
changing opposite to the direction a peak with a negative value is changing. The effect of unit
normalization in this data is to make peaks that are unchanging with weathering seem positive
since they are becoming more important with respect to the total mass left in the sample.
Negative peaks are those compounds disappearing first as a result of the weathering. Figure 9a
shows the GCxGC-MS total ion (TIC) loadings. The blue peaks indicate peaks lost during the
laboratory fuel oil weathering, the red peaks mark compounds retained in the sample. The
predominance of blue peaks is expected because weathering was significant in the chromatogram
region used for the model (see Figures 5a-c). The few large red peaks observed in the TIC
loadings (Fig 9a) are also observed in the m/z 123 loadings (Fig 9b). This confirms that the
compound peaks identified as being degradation resistant using the PCR model were
sesquiterpanes. Figure 9c is the original GC×GC-MS m/z 123 data for the Day 46
biodegradation sample scaled equivalently for comparison.
Principal Components Regression of GC×GC-MS data has provided a quantitative calibration
between laboratory weathering experiments and naturally weathered oil. This method is more
informative than simple peak ratios, but also is consistent with the results obtained by those
ratios. The examination of regression vectors of these multivariate data sets can show important
chemical information that can be used for future investigations into both fingerprinting and
weathering extent.
REFERENCES
ASTM Standard D5739. 2006. " Standard practice for oil spill identification by gas
chromatography and positive ion electron impact low resolution mass spectrometry,” ASTM
International, West Conshohocken, PA, 2006.
ASTM Standard E1618. 2010. “Standard test method for ignitable liquid residues in extracts
from fire debris samples by gas chromatography-mass spectrometry,” ASTM International, West
Conshohocken, PA, 2010.
Dimandja, J-M.D. 2004. Comprehensive 2-D GC provides high-performance separations in
terms of selectivity, sensitivity, speed, and structure. Analytical Chemistry 76:167A–174A
Faber, N.M. and R. Bro. 2002. Standard error for multiway PLS: 1. Background and a simulation
study. Chemomem. and Intell. Syst. 61:133–149
Gaines, R.B., G.S. Frysinger, M.S. Hendrick-Smith, and J.D. Stuart. 1999. Oil spill identification
by comprehensive two-dimensional gas chromatography (GC×GC). Environmental Science
&Technology 33:2106–112.
Gaines, R.B., G.J. Hall, G.S. Frysinger, W.R. Gronlund, and K.L. Juaire. 2006. Chemometric
determination of target compounds used to fingerprint unweathered diesel fuels. Environmental
Forensics 7:1–12.
Gaines, R.B., G.S. Frysinger, C.M. Reddy, and R.K. Nelson. 2007. Oil spill identification by
comprehensive two-dimensional gas chromatography (GC × GC), Chapter 5, in Oil Spill
2011 INTERNATIONAL OIL SPILL CONFERENCE
7 2011-428
Environmental Forensics – Fingerprinting and Source Identification. Z. Wang, S.A. Stout Eds.
Elsevier.
Hampson, G.R., and E.T. Moul. 1978. No. 2. Fuel oil spill in Bourne, Massachusetts: immediate
assessment of the effects on marine invertebrates and a 3-year study of growth and recovery of a
salt marsh. Journal of the Fisheries Research Board of Canada 35:731–744.
Peacock, E.E., G.R. Hampson, R.K. Nelson, L. Xu, G.S. Frysinger, R.B. Gaines, J.W.
Farrington, B.W. Tripp, and C.M. Reddy. 2007. The 1974 spill of the Bouchard 65 oil barge:
Petroleum hydrocarbons persist in Winsor Cove salt marsh sediments. Marine Pollution Bulletin
54: 214–225
Pierce K.M., J.C. Hoggard, R.E. Mohler, and R.E. Synovec. 2008. Recent advancements
in comprehensive two-dimensional separations with chemometrics. Journal of Chromatography
A. 1184:341–52.
Stout, S.A., A.D. Uhler, and K.J McCarthy. 2005. Middle distillate fuel fingerprinting using
drimane-based bicyclic sesquiterpanes. Environmental Forensics 6:241–251.
Teal, J.M., K.A. Burns, and J.W. Farrington. 1978. Analyses of aromatic hydrocarbons in
intertidal sediments resulting from two spills of No. 2 fuel oil in Buzzards Bay, Massachusetts.
Journal of the Fisheries Research Board of Canada. 35:510–520.
Mispelaar, V.G.V., A.C. Tas, A.K. Smilde, P.J. Schoenmakers, and A.C.V. Asten. 2003.
Quantitative analysis of target components by comprehensive two-dimensional gas
chromatography. Journal of Chromatography A. 1019:15–29.
Wang, Z., C. Yang, M. Fingas, B. Hollebone, X. Peng, A.B. Hansen, and J.H. Christensen.
2005. Characterization, weathering, and application of sesquiterpanes to source identification of
spilled lighter petroleum products. Environmental Science &Technology 39:8700–8707.
2011 INTERNATIONAL OIL SPILL CONFERENCE
8 2011-428
Figure 1. GC and GC×GC-MS chromatogram of No. 2 fuel oil. The 2D retention position of
significant petroleum classes is identified.
2011 INTERNATIONAL OIL SPILL CONFERENCE
9 2011-428
Figure 2. Live biodegradation experiment results including neat Bouchard 65 No. 2 fuel oil (a),
Day 11 (b), and Day 46 (c), GC chromatograms.
Figure 3. Control biodegradation experiment results including neat Bouchard 65 No. 2 fuel oil
(a), and Day 46 (b), GC chromatograms.
2011 INTERNATIONAL OIL SPILL CONFERENCE
10 2011-428
Figure 4. n-C18 to phytane ratio for biodegradation experiment. Triangle indicates ratio for
neat Bouchard 65 No. 2 fuel oil. Diamonds show non-varying ratio for Control experiment.
Square show decrease in ratio of Live biodegradation.
2011 INTERNATIONAL OIL SPILL CONFERENCE
11 2011-428
Figure 5. GC×GC-MS total ion chromatograms for neat Bouchard 65 No. 2 fuel oil (a), Day 11
(b), Day 46 (c), and Winsor Cove sediment extracts (d). Solid circles in 5c mark positions of
undegraded pristine and phytane isoprenoid biomarkers. Dashed circle marks position of
sesquiterpanes. Square indicates chromatogram region used for PCA and PCR chemometric
analysis.
2011 INTERNATIONAL OIL SPILL CONFERENCE
12 2011-428
Figure 6. GC×GC-MS m/z 123 extracted ion chromatograms for neat Bouchard 65 oil, No. 2
fuel oil (a), Day 46 (b), and Winsor Cove sediment extracts (c). Dashed lines in 6b mark the
series of C5 (M+ = 208) and C6 (M
+ = 222) substituted sesquiterpanes. The mass spectra for
compound peaks labeled A and B are provided.
2011 INTERNATIONAL OIL SPILL CONFERENCE
13 2011-428
Figure 7. PCA Scores plot for biodegradation time series model. Winsor Cove sample has been
projected onto the model and not used for initial fit.
2011 INTERNATIONAL OIL SPILL CONFERENCE
14 2011-428
Figure 8. PCR for biodegradation time series model. The x-axis is the measured weathering
duration for Live Bouchard 65 oil incubation. The y-axis is predicted weathering from the PCR
model. Regression captured 99.71 percent of the “Y block” variance. Open squares show the
cross-validated fit of each data point to the PCR model with error bars. Winsor Cove sample has
only been predicted, and has been moved to the regression line (approx. 13 days) for
visualization.
2011 INTERNATIONAL OIL SPILL CONFERENCE
15 2011-428
Figure 9. PCR Loadings for biodegradation time series model. Blue peaks in TIC loading (a)
indicate peaks lost during laboratory weathering; red peaks (circled) mark compounds retained
after weathering. Extracted ion m/z 123 loading (b) shows that retained peaks have 123 ion and
belong to sesquiterpane class. GC×GC-MS . Extracted ion m/z 123 data provided for
comparison.
2011 INTERNATIONAL OIL SPILL CONFERENCE
16 2011-428
Table 1 – Mass Spectrum Ions Included in Chemometric Analysis
Chemical Class m/z
Alkanes 57, 71, 85
Acyclic Isoprenoids 113
Alkylcyclohexanes 55, 67, 81, 83, 97
Sesquiterpanes 123, 179, 193, 207
Alkylbenzenes 91, 92, 105, 162, 148, 134, 120
Alkylnaphthalenes 128, 142, 156, 170
Alkylfluoranthenes 166, 180, 194