Resolving the Unresolved Complex Mixture in ... - ACS Publications

Resolved peaks are not expected in the Narragansett Bay UCM hump because motor ...... Victoria Daskalou , Polona Vreča , Gregor Muri , Constantine St...
1 downloads 0 Views 539KB Size
Environ. Sci. Technol. 2003, 37, 1653-1662

Resolving the Unresolved Complex Mixture in Petroleum-Contaminated Sediments G L E N N S . F R Y S I N G E R , * ,† RICHARD B. GAINES,† LI XU,‡ AND CHRISTOPHER M. REDDY‡ Department of Science, United States Coast Guard Academy, New London, Connecticut 06320, and Department of Marine Chemistry and Geochemistry, Woods Hole Oceanographic Institution, Woods Hole, Massachusetts 02543

Comprehensive two-dimensional gas chromatography (GC×GC) was used to investigate the chemical composition of the unresolved complex mixture (UCM) of hydrocarbons in petroleum-contaminated marine sediments. The UCM hydrocarbons were extracted and separated with silica and silver-impregnated silica gel chromatography to yield four fractions (branched alkanes and cycloalkanes, monoaromatics, naphthalenes, and multi-ring PAHs) prior to GC×GC analysis. GC×GC separations used a poly(dimethylsiloxane) stationary phase for volatility selectivity on the first dimension and a 14% cyanopropylphenyl polysiloxane phase for polarity selectivity on the second dimension to fully resolve monoaromatic, naphthalene, and multi-ring PAH compounds from the UCM. A chiral γ-cyclodextrin phase was used for shape selectivity on the second GC×GC dimension to resolve individual branched alkanes and cycloalkanes in the saturates fraction of the UCM. The ability of GC×GC to resolve thousands of individual chemical components from the UCM will facilitate an understanding of the sources, weathering, and toxicity of UCM hydrocarbons.

Introduction The term “unresolved complex mixture” (UCM) describes the raised baseline hump that is often observed in gas chromatograms of petroleum. The first reference to the UCM is uncertain, but in 1970, Blumer et al. (1) examined petroleum hydrocarbons extracted from diesel fuel-contaminated shellfish and described chromatograms containing “ ... a broad unresolved background from cycloparaffins and aromatics ...”. He noted that this unresolved background was “... typical for chromatograms of crude oil or crude oil distillates that are analyzed on a column of relatively low efficiency”. Blumer et al.’s packed-column gas chromatogram contained a baseline hump with a carbon number range from n-C12 to n-C22. The n-alkane and isoprenoid biomarker compounds were degraded and reduced in concentration relative to fresh diesel fuel so the baseline hump of unresolved compounds was enhanced (Figure 1). By the early 1970s, the specific phrase unresolved complex mixture and the acronym UCM were used routinely to describe hump-shaped chromatograms as well as to describe a broad class of petroleum* Corresponding author phone: (860)444-8656; fax: (860)701-6147; e-mail: [email protected]. † United States Coast Guard Academy. ‡ Woods Hole Oceanographic Institution. 10.1021/es020742n CCC: $25.00 Published on Web 03/12/2003

 2003 American Chemical Society

based environmental contaminants (2-5). Inherent in the term unresolved complex mixture is the acceptance that existing analytical technology cannot describe the chemical composition of the sample in any other way. There are multiple sources of petroleum that can contribute to the UCM in the extracts of environmental samples. These include acute and chronic oil spills, urban runoff, atmospheric deposition, and waste treatment plants. The National Academy of Sciences estimated that U.S. coastal regions annually receive 2.7 t of fresh or weathered petroleum hydrocarbons including crude oil and refined products such as kerosene, home-heating oil, diesel fuel, industrial heating fuel, bunker oil, and motor lubricating oils (6). Two examples of UCM contamination in marine sediments are shown in Figure 2 (7, 8). Each sediment sample was extracted and analyzed with a high-resolution open-tubular capillary column. Each chromatogram contains a prominent UCM hump resulting from the chromatographic overlap of thousands of compounds. Petroleum compounds present at high concentrations produce discernible peaks on top of the hump, but even these peaks are likely to consist of multiple overlapping compounds. While both chromatograms in Figure 2 have a UCM hump, they are distinct. The Wild Harbor River sediment UCM (Figure 2a) has a carbon number range from n-C12 to n-C24 while the Narragansett Bay sediment UCM (Figure 2b) has a range from about n-C16 to n-C36. Each sediment has a different source and type of petroleum contamination. The Wild Harbor River sediment was collected from a salt marsh that was contaminated by a diesel fuel spill in 1969 (1, 2, 7, 9-11). The expected diesel fuel n-alkane peaks do not appear in the chromatogram because they were preferentially lost by microbial degradation (12, 13). The shape and range of this UCM hump is almost identical to that observed by Blumer at the same site 30 years earlier (see Figure 1) (1, 7). The Narragansett Bay sediment has been impacted by multiple sources, but used motor oil from a combination of urban runoff and sewage overflow predominates (14). Resolved peaks are not expected in the Narragansett Bay UCM hump because motor oil chromatograms have few distinct peaks (15). Peaks visible on top of the UCM are from other abundant contaminants such as polycyclic aromatic hydrocarbons (PAHs), polychlorinated biphenyls (PCBs), and substituted benzotriazoles (8, 16, 17). As was the case here, if the petroleum source and type are different, the carbon number range and shape of the UCM will be different. In some instances when sediments have multiple sources of petroleum contamination, there can be multiple UCM humps in the chromatogram (18). This “humpfitting” approach to distinguishing different sources of the UCM has been used for over 30 years. Unfortunately, this method provides little information about the chemical composition of the UCM. Additional information is needed to accurately determine the source of UCM contamination, to evaluate the extent of UCM weathering or biodegradation, and to assess the residual toxicity of UCM compounds. A number of analytical methods have been developed to explore the chemical composition of the UCM. These include wet-chemical methods such as open-column silica gel chromatography, silver-impregnated silica gel chromatography, thin-layer chromatography (TLC) (19), molecular sieves, and urea or thiourea adductions (19, 20) to separate the UCM petroleum compounds into distinct chemical fractions or groups. Other chemical methods include chromic acid (19, 21, 22) or ruthenium tetroxide oxidations (23, 24) to chemically modify UCM fractions prior to analysis. VOL. 37, NO. 8, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1653

FIGURE 1. Gas chromatogram of petroleum hydrocarbons extracted from oysters (Crassostrea virginica) harvested from Wild Harbor River, West Falmouth, MA, in November 1969 following the Florida oil spill. Column: 10 ft, 1/8 in., 2.2% Apiezon L on Chromosorb W, 70-80 mesh. Temperature program: 100 °C to 300 °C at 6 °C min-1. T marks the position of isoprenoid alkanes. Figure adapted from Blumer et al. (1). Spectroscopic methods for UCM analysis include infrared (IR) (2, 4, 19), fluorescence (25), nuclear magnetic resonance (NMR) (4, 19), and high-resolution mass spectrometry (26). The most successful petroleum UCM analyses have come from instrumented separation methods such as gas chromatography (GC) (27, 28) and high-performance liquid chromatography (HPLC) (29, 30) or hyphenated methods such as gas chromatography/mass spectrometry (GC/MS) (31-35). These analytical methods have produced an understanding of UCM chemistry. First, the chemical composition of the UCM is derived from the composition of its petroleum source or sources. Therefore, the UCM may contain a full array of alkanes, branched alkanes, cycloalkanes, monoaromatics, multi-ring aromatics, heteroatomic aromatics, steranes, and cyclic triterpenoids, etc. that are part of crude or refined petroleum. Second, evaporation, dissolution, and chemical and biological oxidation weather the petroleum. Evaporation removes the most volatile compounds; dissolution removes the more polar and water-soluble compounds; and biodegradation generally attacks the linear alkanes, branched alkanes, and then the cycloalkanes and aromatics (36). Recalcitrant petroleum compounds that are the most resistant to degradation remain to form the characteristic UCM baseline hump. Killops and Al-Juboori analyzed one such UCM sample with a variety of methods

and found that it contained about 90% saturates and only about 10% aromatics (19). The UCM saturates class was thought to include monosubstituted “T”-branched alkane isomers (21, 37); substituted cycloalkanes (mostly singlering) (19); and monoaromatics with linear, branched, and cycloalkyl-substituted benzenes or tetrahydronaphthalenes (38, 39). In this work, the novel analytical method of comprehensive two-dimensional gas chromatography (GC×GC) was used to explore the complete chemical composition of the UCM. GC×GC separations were optimized to fully resolve all chemical groups of the UCM including the alkanes; branched alkanes; one- and multi-ring cycloalkanes; and one-, two-, and multi-ring aromatics. Open-column silica gel and silver-impregnated silica gel liquid chromatographic separations were used to isolate individual chemical classes from the UCM prior to GC×GC separation.

Materials and Methods Sediment Samples. A sediment core was collected from a salt marsh near the Wild Harbor River (site M-1) in West Falmouth, MA. The site was contaminated with diesel fuel from a 1969 spill (1, 2, 7, 9-11). The most heavily contaminated horizon of the core (14-16 cm) was air-dried and extracted by pressurized fluid extraction (ASE 200 Extraction System, Dionex Corp.) with dichloromethane at 60 °C and 1000 psi (40). The total extract was concentrated by evaporation with nitrogen gas at 40 °C and then separated with a silica gel column (100-200 mesh, 0.5 cm i.d., 18 cm). The saturated hydrocarbon fraction (F1) was eluted with 10 mL of hexane. The unsaturated fraction (F2) was eluted with 10 mL of hexane/dichloromethane (1:1 v/v). Each of these fractions was further separated with silver-impregnated silica gel column chromatography. The silver-impregnated silica gel was prepared by mixing silica gel (100-200 mesh, 80 g) with silver nitrate (16 g) in 200 mL of methanol/H2O (4:1 v/v). The slurry was dried, activated in the dark at 120 °C overnight, packed in glass columns (0.5 cm i.d., 18 cm), and charged with the F1 and F2 fractions, respectively. The F1 fraction was eluted with 5 mL of hexane to produce the F11 fraction containing saturates (branched alkanes and cycloalkanes), followed by 10 mL of dichloromethane to produce the F12 fraction containing monoaromatics (alkylbenzenes).

FIGURE 2. Gas chromatograms of sediment extracts from (a) Wild Harbor River, West Falmouth, MA, and (b) Narragansett Bay, RI. Column: 30 m, 0.25 mm, 0.25 µm 5% phenyl poly(dimethylsiloxane) phase. Temperature program: 50 °C (3 min hold) to 320 °C at 5 °C min-1. Panel a data from Reddy et al. (7). Panel b data from Frysinger et al. (8). 1654

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 8, 2003

FIGURE 3. UCM fractionation flowchart. The F2 fraction was eluted with 10 mL of hexane/dichloromethane (9:1 v/v) to produce the F21 fraction containing tworing aromatics (alkylnaphthalenes), followed by 10 mL of dichloromethane to produce the F22 fraction containing multi-ring PAHs (Figure 3). Additional UCM-contaminated sediment samples were collected from Narragansett Bay, RI by push core (17) and extracted by pressurized fluid extraction with dichloromethane/methanol (90:10 v/v) at 100 °C and 2000 psi. Extracts were concentrated, treated with copper powder to remove elemental sulfur, and cleaned up with a silica gel column to remove polar compounds (i.e. alcohols and fatty acids). Gas Chromatography/Mass Spectrometry. The GC/MS system was a Hewlett-Packard GCD Plus quadrupole mass spectrometer. Separations were performed on a 5% phenyl poly(dimethylsiloxane) column (Hewlett-Packard HP-5, 30 m, 0.25 mm i.d., 0.25 µm film) and temperature programmed from 50 (3 min hold) to 320 °C at 5.0 °C min-1. Helium was the carrier gas with a constant flow of 1.0 mL min-1. A 1.0-µL sediment extract sample was injected into a 250 °C splitless injector (1.0 min purge time). The quadrupole was scanned from m/z 10 to m/z 450 to acquire mass spectra at about 2 Hz. Comprehensive Two-Dimensional Gas Chromatography. The acronym GC×GC is used for comprehensive twodimensional gas chromatography because orthogonal gas chromatographic separations are used in both analytical dimensions. Orthogonality is achieved by using stationary phases with different selectivity. GC×GC relies on modulation to effectively transfer all analytes from the first-dimension column to the second-dimension column; therefore, it is comprehensive (41-44). The GC×GC system used to analyze the sediment extracts was a Hewlett-Packard 6890 gas chromatograph configured with a split/splitless injector, two chromatography columns, a rotating thermal modulator (Zoex Corp.), and a flame ionization detector (45). Two additional column ovens were installed in the GC so that the temperature of the first column, modulator, and second column could be independently programmed. A 1.0-µL sediment extract sample was injected into a 250 °C splitless injector (1.0 min purge time). The first-dimension column was a nonpolar poly(dimethylsiloxane) phase (Quadrex 0071, 9.5 m, 0.10 mm i.d., 0.5 µm film) that was temperature programmed from 40 to 290 °C at 0.5 °C min-1. The modulation column was poly(dimethylsiloxane) (0.08 m, 0.10 mm i.d., 0.5 µm film) and temperature programmed from 30 (30 min hold) to 265 °C at 0.5 °C min-1. Initial experiments used a polar 14% cyanopropylphenyl polysiloxane seconddimension column (Quadrex 007-1701, 0.75 m, 0.10 mm i.d., 0.14 µm film) that was temperature programmed from 40 to

250 °C at 0.5 °C min-1 (80 min hold). Follow-up experiments used a γ-cyclodextrin second-dimension column (Restek RtγDEXsa, 2.0 m, 0.10 mm i.d., 0.10 µm film) that was temperature programmed from 60 to 250 °C at 0.5 °C min-1 (100 min hold). A 0.8-m guard column and a 0.25-m detector transfer line of 0.10 mm i.d. deactivated fused-silica column were connected with glass press-fit connectors. Hydrogen was the carrier gas with a constant flow of 0.5 mL min-1. The thermal modulator was heated to 100 °C above the temperature of the modulator column. The heater rotated at 0.25 revolutions s-1 through an angle of 160° to desorb analytes from the entire length of the modulator column trap and inject them into the second column. After a 0.2-s pause at the peak launch position, the heater returned to a home position off the column until the start of the next modulator cycle. The modulator period was 20 s. The FID was sampled at 100 Hz. The serial FID data was sliced into modulator-periodlength segments and written into a two-dimensional array. The array was visualized as an interpolated color image (Transform, Research Systems, Inc.) (8, 41). Individual GC×GC chromatogram peaks were integrated after baseline subtraction. Baseline subtraction is possible because modulation produces periods of detector baseline in each seconddimension chromatogram. The baseline average was subtracted, and the remaining FID signal was integrated for each peak. In the same manner, groups of peaks or entire regions of the GC×GC chromatogram were integrated after baseline subtraction. Since the goal of region integration was to compare the mass of analytes with the same carbon number in adjacent regions of the chromatogram, FID response factors were not determined.

Results and Discussion Gas Chromatography/Mass Spectrometry. Conventional GC/MS methods were used to explore the chemical composition of the Wild Harbor River sediment UCM. An examination of the extracted ion chromatogram for the threecarbon-substituted (C3-) naphthalene isomers (m/z 170) (Figure 4) shows distributions in the UCM that are typical for petroleum fuels (31). There were also abundant C4- (m/z 184), C5- (m/z 198), and trace C6-naphthalenes in the UCM. It is expected that, after 30 years of weathering, a combination of evaporation, dissolution, and biodegradation mechanisms removed naphthalene, the methyl naphthalenes, and the ethyl- and dimethylnaphthalenes from the UCM. Alkylphenanthrenes and alkyldibenzothiophenes were also present in the Wild Harbor sediment UCM. The extracted ion chromatogram for the methylphenanthrenes (m/z 192) has prominent peaks for two isomers, 3-methylphenanthrene and 9-methylphenanthrene. This pattern is different from the four methylphenanthrene isomer peaks typically observed in petroleum fuels; this suggests preferential degradation of selected methylphenanthrene isomers in the Wild Harbor sediment UCM (46). There is a significant concentration of C2- (m/z 206) and C3-phenanthrenes (m/z 220) in the UCM. These are more abundant than would normally be expected for diesel fuel because the preferential weathering of other compounds has increased the relative concentration of the alkylphenanthrenes. The methyldibenzothiophenes (m/z 198), C2- (m/z 212), and C3-dibenzothiophenes (m/z 226) are also present in the UCM. The parent compounds, phenanthrene and dibenzothiophene, do not remain in the UCM because they are more readily biodegraded than their alkylated counterparts (33). While the extracted ion chromatogram approach was very successful for examining the multi-ring aromatic composition of the Wild Harbor UCM, it is not as well suited for exploring the monoaromatics or saturates. Knowledge of the monoaromatic composition, specifically the alkylbenzenes, is VOL. 37, NO. 8, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1655

FIGURE 4. Extracted ion chromatograms obtained by GC/MS of a whole sediment extract from Wild Harbor River, West Falmouth, MA. CxN indicates alkylnaphthalenes, CxDBT indicates alkyldibenzothiophenes, and CxPh indicates alkylphenanthrenes.

FIGURE 5. Extracted ion chromatograms obtained by GC/MS of a whole sediment extract from Wild Harbor River, West Falmouth, MA. (a) Alkylbenzenes, m/z 91, (b) alkanes m/z 85, and (c) isoprenoid biomarkers m/z 113. important because they are known to contribute to the residual toxicity of the UCM (38, 39). If extracted ion chromatograms representing alkylbenzenes (m/z 91 or 105) are examined, a UCM hump is observed that suggests that there are numerous alkylbenzenes present in the UCM (Figure 5a). However, since the m/z 91 ion is an alkylbenzene fragment, little is known about the isomer distribution of the parent compounds. It is also possible that unrelated compounds produce the same fragment. Extracted ion chromatograms for parent alkylbenzenes are often directly examined in petroleum, for example, m/z 120, 134, and 148 for the C3-, C4-, and C5-alkylbenzenes, etc. (32), but the weathered Wild Harbor sample contains only C8- and higher alkylbenzenes for which the extracted ion chromatograms are not unique to alkylbenzenes. 1656

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 8, 2003

The saturates present an even greater challenge for GC/ MS analysis. For example, branched alkane composition can be examined with a m/z 57 or 85 extracted ion chromatogram (Figure 5b), but a UCM hump results. Since alkanes fragment extensively, information on the hundreds or thousands of individual alkanes is limited. Similarly, the m/z 83 and 97 fragments are not suitable for exploring individual cycloalkane compounds. One class of saturates that does have a characteristic fragment ion is the acyclic isoprenoid biomarkers, thus an examination of m/z 113 shows peaks that correspond to pristane and phytane (Figure 5c) (47). Comprehensive Two-Dimensional Gas Chromatography. The two-dimensional GC×GC chromatogram for the Wild Harbor River whole sediment extract is shown in Figure 6. The GC×GC separation was able to resolve hundreds to

FIGURE 6. GC×GC chromatogram of Wild Harbor River (MA) whole sediment extract. The x-axis is retention (min) on the nonpolar (poly(dimethylsiloxane)) first-dimension column. The y-axis is relative retention (s) on the polar (14% cyanopropylphenyl polydimenthylsiloxane) second-dimension column. The background is blue. Low abundance compound peaks are white, and high abundance peaks are red to black. BA, branched alkanes; C1A, one-ring alkylcycloalkanes; C2A, two-ring alkylcycloalkanes; B, alkylbenzenes; N, alkylnaphthalenes; P, alkylphenanthrenes and alkyldibenzothiophenes. thousands of compounds from the previously unresolved complex mixture (see Figure 2a). A volatility-based separation with a nonpolar poly(dimethylsiloxane) column produced a boiling point elution order across the x-axis. The homologous series of n-alkanes, which is typically seen in GC×GC chromatograms, is missing in this weathered fuel oil sample. However, a band containing numerous branched alkanes (BA), including the isoprenoids pristane and phytane, is distributed along the first dimension of the GC×GC chromatogram. A polarity-based separation with a 14% cyanopropylphenyl poly(dimethylsiloxane) column produced classtype separation along the y-axis. The identification and ordering of chemical classes has been determined in petroleum products ranging from gasoline (48-51) to diesel fuel (41, 52-54) and crude oil (55). GC×GC peaks are best identified by comparing the two-dimensional retention times with those for authentic standards or by interfacing the GC×GC separation with a mass spectrometer (56, 57). The branched alkanes (BA) have the least second-dimension retention. The cycloalkanes appear in bands just above the branched alkanes. The first band contains the numerous one-ring alkylcycloalkanes (C1A), including the alkylcyclohexanes and alkylcyclopentanes. The next highest band contains two-ring alkylcycloalkanes (C2A), including the alkyldecahydronaphthalenes and alkyltetrahydroindans. The peaks above the cycloalkanes are the alkylbenzenes (B). Also located with the alkylbenzenes are the alkylindans and the alkyltetrahydronaphthalenes. The peaks labeled (N) are predominantly naphthalenes. After 30 years of weathering, only the C3- and higher alkylnaphthalenes remain. Other petroleum components of interest located with the naphthalenes are the alkylbiphenyls and the sulfur-containing alkylbenzothiophenes. The peaks at greatest second-dimension retention (P) are the alkylphenanthrenes and alkyldibenzothiophenes. Because the chemical classes are well-separated and grouped in the GC×GC chromatogram, we were able to determine the mass percent of individual chemical classes by integrating large areas of the GC×GC chromatogram. Integration of the saturates (below the dotted line), including the branched alkanes (BA), alkylcycloalkanes (C1A and C2A), and the alkylbenzene monoaromatics (B), indicates that they comprise about 85% of the UCM mass. The multi-ring aromatics (above the dotted line), including the naphthalenes (N) and phenanthrenes (P), comprise only 15%. These

percentages agree well with the 90% saturates and 10% aromatics composition determined by Killops and Al-Juboori (19). This means that conventional GC/MS methods are used to explore only about 15% of the UCM of middle distillate fuels, which leaves most of the UCM (especially the saturates) unexplored. UCM Fractionation. To more completely explore the chemical composition of the UCM with GC×GC, the sediment extract sample was separated with successive silica and silverimpregnated silica gel open-column liquid chromatography to isolate individual three-ring (F22), two-ring (F21), one-ring aromatic (F12), and saturates (F11) fractions of the UCM (see Figure 3). Figure 7 shows the GC×GC chromatogram of the threering aromatics fraction (F22). Numbered circles on the chromatogram mark the two-dimensional retention time of selected chemical standards (Table 1). A retention time match in two dimensions should be considered reliable for peak identification because coelution is less likely than in onedimensional gas chromatography. Coelution in two dimensions generally results from overlap of related isomers such as m- and p-xylene (50, 51). In Figure 7, the phenanthrenes are marked by circles 1-4. Phenanthrene (circle 1) is weathered below detection limits in our analysis. The compounds 3- and 9-methylphenanthrene (circles 2 and 3) define a band of C1-phenanthrenes. The compound 3,6-dimethylphenanthrene (circle 4) defines the C2-phenanthrene band. The dibenzothiophenes are marked by circles 5-7. Again, the parent compound, dibenzothiophene (circle 5), is not detected. 4-Methyldibenzothiophene (circle 6) defines the C1-dibenzothiophene band, and 4,6-dimethyldibenzothiophene (circle 7) defines the C2-dibenzothiophenes band. Although no chemical standards were available, the C3-phenanthrene and C3dibenzothiophene regions of the chromatogram can be identified on the basis of the repeating nature of GC×GC chromatogram bands (8, 41). Note that the phenanthrenes and dibenzothiophenes are intermingled in the GC×GC analysis. This is because, for the particular column stationary phases used here, the volatility and polarity retention of these groups are similar. For these groups of UCM compounds, GC/MS is more selective because phenanthrenes and dibenzothiophenes are easily separated by mass (see Figure 3). VOL. 37, NO. 8, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1657

FIGURE 7. GC×GC chromatogram of Wild Harbor F22 saturates fraction. Numbered peaks identified in Table 1.

TABLE 1. GC×GC Chromatogram Peak Identities for Figures 7-10 1 2 3 4 5 6 7 8 9 10 11 12

phenanthrene 3-methylphenanthrene 9-methylphenanthrene 3,6-dimethylphenanthrene dibenzothiophene 4-methyldibenzthiophene 4,6-dimethyldibenzothiophene fluorene 1-methylfluorene 2-isopropylnaphthalene 2,3,6-trimethylnaphthalene 1,4,6,7-tetramethylnaphthalene

13 14 15 16 17 18 19 20 21 22 23 24

2,6-diisopropylnaphthalene n-octylbenzene n-nonylbenzene n-decylbenzene n-undecylbenzene n-dodecylbenzene n-tridecylbenzene n-tetradecylbenzene norpristane pristane phytane n-octylcyclohexane

FIGURE 8. GC×GC chromatogram of Wild Harbor F21 saturates fraction. Numbered peaks identified in Table 1. In Figure 7, circles 8 and 9 mark fluorene and 1-methylfluorene. These compounds contain two aromatic rings and one saturated ring. They were eluted in the F22 fraction even though they are less polar than the three-ring phenanthrenes. Their reduced polarity is also evident in the GC×GC analysis because they are less retained on the second column. Figure 8 shows the GC×GC chromatogram of the tworing aromatics fraction (F21). The naphthalenes are identified by circles 10-13. 2-Isopropylnaphthalene (circle 10) and 2,3,6-trimethylnaphthalene (circle 11) define the C3-naphthalene band. The branched and monosubstituted isomers elute first because of their lower boiling point, and the trimethylnaphthalenes elute last. The compound 1,4,6,7tetramethylnaphthalene (circle12) defines the C4-naphthalene band, and 2,6-diisopropylnaphthalene defines the C6naphthalene band. No C5-naphthalene standard was available, 1658

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 8, 2003

but the C5-naphthalene band position can be inferred by its position between the C4- and C6-naphthalene bands. Again, there are other types of UCM compounds present in or near the identified alkylnaphthalene bands. These most likely include alkylbiphenyls and alkylbenzothiophenes, which both have similar retention and thus position in the GC×GC separation (41). The presence of other compounds in the bands suggests that GC/MS analysis may be preferred because of its mass selectivity and its ability to quantify known PAH compounds, but the GC×GC chromatogram image provides the advantage of being able to see all of the isomer bands at once. This is useful for source fingerprinting and for observing small differences between samples. Figure 9 shows the GC×GC chromatogram of the onering aromatics fraction (F12). The bands in the GC×GC image contain the alkylbenzenes as well as naphthenoaromatics

FIGURE 9. GC×GC chromatogram of Wild Harbor F12 saturates fraction. Numbered peaks identified in Table 1.

FIGURE 10. GC×GC chromatogram of Wild Harbor F11 saturates fraction. Numbered peaks are identified in Table 1. Dotted lines separate branched alkanes (bottom) including norpristane (21), pristane (22), and phytane (23); one-ring alkylcycloalkanes (middle); and two-ring alkylcycloalkanes (top). such as the alkylindans and alkyltetrahydronaphthalenes. The isolation and identification of these alkylbenzene bands is important because these compounds have been linked to shellfish toxicity (38, 39). Circles 14-20 mark the homologous series of linear alkyl benzenes from n-octylbenzene to n-tetradecylbenzene. Each of these compounds is positioned near the midpoint of an alkylbenzene band that contains numerous isomers. The separation in the GC×GC chromatogram allows the relative concentration and distribution of individual alkyl benzene isomers to be examined. Other peaks observed farther up in the GC×GC chromatogram (Figure 9) at about 10 s retention on the second dimension are some alkylnaphthalenes that eluted in the F12 fraction. The open-column silica and silver-impregnated silica gel separations did not produce absolute separation of classes, so some classes appear in more than one fraction. Some alkylbenzenes appeared in the F21 fraction as well (see Figure 8). Future work will use HPLC separations to produce purer fractions. Figure 10 shows the GC×GC chromatogram of the saturates fraction (F11). The line of peaks across the bottom of the chromatogram are the branched alkanes. The large peaks marked 21-23 are the isoprenoid biomarkers norpristane, pristane, and phytane, respectively. The peaks above the branched alkanes are the cycloalkanes. The position of n-octylcyclohexane is marked (circle 24). Other peaks at about the same second-dimension retention are alkylcyclohexane or alkylcyclopentane isomers. Peaks at greater second-

dimension retention are multi-ring cycloalkanes. The abundance of the one-ring alkylcycloalkanes is greater than the two-ring cycloalkanes. Integration of the areas defined by the dotted lines in Figure 10 shows that the composition of the F11 fraction is about 20% branched alkanes, 50% onering alkylcycloalkanes, and 30% two-ring and greater alkylcycloalkanes. This is in agreement with the UCM studies of Killops and Al-Juboori (19), which found that one-ring cycloalkanes are the most abundant in the UCM. Unlike the other UCM fractions, the branched alkanes and cycloalkanes in this GC×GC chromatogram are not ordered into distinct sloping bands with carbon number spacing. This suggests that the selectivity of the GC×GC separation is not properly tuned to the dimensionality of the mixture. In addition, there is a lot of unused peak capacity (open space) in the two-dimensional retention time plane. Since the open-column silica gel and silver-impregnated silica gel chromatography steps used a polarity-based selectivity to produce the F11 fraction, a second polarity-based separation in the GC×GC analysis is not effective. A polarity-byvolatility-by-polarity separation results; these separation dimensions are not orthogonal. To improve the separation of the F11 saturates fraction, a 2.0-m section of a chiral γ-cyclodextrin column was substituted for the polar second column. This “shape” selective column was better able to separate the branched alkanes and cycloalkanes from one another. Figure 11 shows the resulting polarity-by-volatilityby-shape chromatogram that contains many separated peaks VOL. 37, NO. 8, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1659

FIGURE 11. GC×GC chromatogram of Wild Harbor F11 saturates fraction. Chiral (γ-cyclodextrin) second-dimension column. The expected positions of the n-C12-n-C22 alkanes are marked with circles.

FIGURE 12. GC×GC chromatogram for a Narragansett Bay, RI, whole sediment extract. The peak labeled 1 is C1-benzotriazole, and the peak labeled 2 is C10-benzotriazole and C8-Cl-benzotriazole (see Reddy et al., 17). The peaks labeled 3-12 are PAHs identified with chemical standards: (3) phenanthrene; (4) anthracene; (5) fluoranthene; (6) pyrene; (7) benz[a]anthracene; (8) chrysene; (9) benzo[b]fluoranthene/ benzo[j]fluoranthrene; (10) benzo[a]pyrene; (11) indeno[1,2,3-cd]pyrene; (12) benzo[ghi]perylene; (13-15) 22S,22R-homohopanes, bishomohopanes, and trishomohopanes, respectively. GC×GC conditions: first-dimension, poly(dimethylsiloxane) (9.5 m, 0.10 mm i.d., 0.5 µm film) 40-320 °C at 3 °C min-1. Modulation column: poly(dimethylsiloxane) (0.08 m, 0.10 mm i.d., 0.5 µm film) 0-280 °C at 3 °C min-1. Second-dimension, trifluoropropylmethyl polysiloxane (1.0 m, 0.10 mm i.d., 0.1 µm film) 40-320 °C at 3 °C min-1. organized into numerous groups that are spread across the entire second dimension of the GC×GC chromatogram. The branched alkanes, which had multiple coelutions in the previous GC×GC separation (Figure 10), are now separated and organized into numerous diagonal bands. We hypothesize that the different bands arise from different alkane substitution patterns. For example, all monomethyl-substituted alkanes form one band, and all dimethyl alkanes may form the adjacent band. Circles on the chromatogram indicate the two-dimensional retention time positions of the n-alkanes. It is important to note that, in this sample, the microbial degradation of n-alkanes is nearly complete. Only trace peaks remain at each n-alkane position. The trace quantities of n-alkanes would be quite difficult to determine by other analytical methods. This degree of separation could produce an accurate calculation of n-C17/pristane ratios for samples where these peaks are normally lost in the UCM hump. The number of resolved cycloalkane compounds is also significantly increased in this sample. The cycloalkane compounds are now grouped into bands and spread across the full two-dimensional retention time plane of the gas chromatogram. The identity is uncertain for most of these peaks, but we hypothesize that alkylcyclohexane, alkyl1660

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 8, 2003

cyclopentane, alkyltetrahydroindan, and alkyldecahydronaphthalene isomers as well as other more unusual combinations of cycloalkane rings form unique bands in the GC×GC chromatogram. The hundreds of peaks in the F11 fraction GC×GC chromatogram (Figure 11) are from just one of the four fractions of the original sediment extract. By selecting an appropriate second-dimension column, we were able to tune the GC×GC conditions to separate the saturates fraction of the UCM. Considering that the GC×GC separation came after open-column liquid chromatography fractionation, it can be argued that a true three-dimensional chromatographic separation with a polarity-by-volatility-by-shape selectivity was achieved. In total, the optimized GC×GC separations of the four UCM fractions (Figures 7-9 and Figure 11) show the full resolution of thousands of individual UCM compounds. Efforts to identify these compounds are underway. Chemical standards are impractical for a task of this magnitude, but they can be used to map the location of particular groups of compounds as was demonstrated in this paper. More specific identification of particular compounds in the GC×GC chromatogram will be completed with a time-of-flight mass

spectrometer. The ability to separate and identify the chemical composition of the UCM creates many avenues for future research. Separation of the saturates is significant because they comprise the majority of the UCM, but they are not often studied because they were analytically inaccessible. All UCMs Are Not The Same. The term UCM is used to describe the GC hump of unresolved compounds regardless of contaminant source. The Wild Harbor sediment UCM resulted from diesel fuel contamination, and it was extensively separated by GC×GC. Samples with different or multiple UCM sources will present greater analytical challenges. For example, the volatility-by-polarity GC×GC chromatogram for a Narragansett Bay sediment sample is shown in Figure 12 (8). This UCM is dominated mainly by inputs of used motor oil (14). In the GC×GC chromatogram, a smear of mostly unresolved peaks extends over a carbon number range from n-C20 to n-C40. This range is consistent with that for motor oil. Motor oils are known to consist primarily of saturated petroleum compounds, especially numerous branched and cycloalkane isomers (58). The GC×GC resolution of this saturates band is poor given the particular GC×GC conditions used, but it is not surprising since the number of isomers for these saturates classes increases dramatically with carbon number. For example, a branched alkane with 30 carbon atoms has over 1 billion possible isomers, and a cycloalkane will have even more (59). If even a fraction of these isomers are present in motor oil, it is not realistic to separate them with GC×GC as was done for the diesel range UCM. Even so, GC×GC provides insight into the chemical composition of this UCM sample. The n-C25-n-C40 alkane peaks are visible and line up with the carbon number axis. The odd-numbered alkanes appear more abundant, which suggests that these n-alkanes have a biogenic source (60). Sterane and hopane biomarkers are also visible in the GC×GC image. A band of peaks running from about 75 to 85 min and at about 0.75 s retention on the second dimension is the sterane biomarkers, and the more clearly resolved band running from about 80 to 90 min and at about 1.0 s on the second dimension is hopane biomarkers. The three pairs of hopanes (peaks 13-15) are the 22S- and 22R- homohopanes, bishomohopanes, and trishomohopanes, respectively (55). The presence of these biomarker peaks is significant because they are often used as indicators of motor oil contamination in sediments (60, 61). In addition to the used motor oil, there are additional peaks in the GC×GC chromatogram that result from other sources of contamination. The peaks labeled 1 and 2 in the GC×GC chromatogram are UV light inhibitors called benzotriazoles that were released from a chemical plant a few kilometers away from the sampling site (16). Numerous PAHs are also found in this sample, and they form a band across the GC×GC chromatogram from phenanthrene (peak 3) to benzo[ghi]perylene (peak 12). In addition to the parent PAHs, there are many less abundant alkylated PAH isomers that are organized into bands in the GC×GC image. The distribution of alkylated PAHs relative to parent PAHs is often used to apportion petrogenic and pyrogenic sources of these compounds (62, 63). The Narragansett Bay sediment UCM also shows numerous PCB congeners as well as p-nonylphenols. In this example, GC×GC was unable to resolve the motor oil UCM, but GC×GC was especially effective for screening the sediment for a wide range of environmental contaminants (8).

Acknowledgments We would like to thank Drs. James Quinn and John Farrington for their invaluable assistance in this project. This work was supported in part by the Robert T. Alexander Coast Guard Academy Science and Teaching Graduate Fellowship. This

paper has not been subject to official U.S. Coast Guard review. Mention of trade names in the paper does not indicate U.S. Coast Guard endorsement of any kind. Additional support was provided by a grant from the National Science Foundation to C.M.R. (CHE-0089172). This is WHOI Contribution No. 10834.

Literature Cited (1) Blumer, M.; Souza, G.; Sass, J. Mar. Biol. 1970, 5, 195-202. (2) Blumer, M.; Ehrhardt M.; Jones, J. H. Deep-Sea Res. 1973, 20, 239-259. (3) Farrington, J. W.; Quinn, J. G. Estuarine Coastal Mar. Sci. 1973, 1, 71-79. (4) Zafiriou, O. C. Estuarine Coastal Mar. Sci. 1973, 1, 81-87. (5) Farrington, J. W.; Quinn, J. G. J. Water Pollut. Control Fed. 1973, 45, 704-712. (6) National Academy of Sciences. Oil in the Sea: National Academy Press: Washington, DC, 1985. (7) Reddy, C. M.; Eglinton, T. I.; Hounshell, A.; White, H. K.; Xu, L.; Gaines, R. B.; Frysinger, G. S. Environ. Sci. Technol. 2002, 36, 4754-4760. (8) Frysinger, G. S.; Gaines, R. B.; Reddy, C. M. Environ. Forensics 2002, 3, 27-34. (9) Sanders, H. L.; Grassle, J. F.; Hampson, G. R.; Morse, L. S.; GarnerPrice, S.; Jones, C. J. Mar. Res. 1980, 38, 265-380. (10) Teal, J. M.; Farrington, J. W.; Burns, K. A.; Stegeman, J. J.; Tripp, B. W.; Woodin, B.; Phinney, C. Mar. Pollut. Bull. 1992, 24, 607614. (11) Burns, K. A.; Teal, J. M. Estuarine Coastal Mar. Sci. 1979, 8, 349-360. (12) Atlas, R. M. Microbiol. Rev. 1981, 45, 180-209. (13) Atlas, R. M. Mar. Pollut. Bull. 1995, 31, 178-182. (14) Latimer, J. S.; Hoffman, E. J.; Hoffman, G.; Fasching, J. L.; Quinn, J. G. Water Air Soil Pollut. 1990, 52, 1-21. (15) Pruell, R. J.; Quinn, J. G. Environ. Pollut. 1988, 49, 89-97. (16) Latimer, J. S.; Quinn, J. G. Environ. Sci. Technol. 1996, 30, 623633. (17) Reddy, C. M.; Quinn, J. G.; King, J. Environ. Sci. Technol. 2000, 34, 973-979. (18) Thompson S.; Eglinton, G. Mar. Pollut. Bull. 1978, 9, 133-136. (19) Killops, S. D.; Al-Juboori, M. A. H. A. Org. Geochem. 1990, 15, 147-160. (20) Chukwuemeka, N. A.; Nwobodo, I. F. Fuel 1994, 73, 779-782. (21) Gough, M. A.; Rowland, S. J. Nature 1990, 344, 648-650. (22) Revill, A. T.; Carr, M. R.; Rowland, S. J. J. Chromatogr. 1992, 589, 281-286. (23) Warton, B.; Alexander, R.; Kagi, R. I. Org. Geochem. 1999, 30, 1255-1272. (24) Warton, B.; Alexander, R.; Kagi, R. I. Org. Geochem. 2000, 31 249. (25) Mason, R. P. Mar. Pollut. Bull. 1987, 18, 528-533. (26) Rodgers, R. P.; Blumer, E. N.; Freitas, M. A.; Marshall, A. G. Anal. Chem. 1999, 71, 5171-5176. (27) U.S. EPA Method 8015B: Nonhalogenated Organics Using GC/ FID. Test Methods for Evaluating Solid Waste: Physical/Chemical Methods (SW-846), 3rd ed., Vol. 1B; Final Update III; U.S. Government Printing Office: Washington, DC, 1996, (28) Standard Test Methods for Comparison of Waterborne Petroleum Oils by Gas Chromatography; D-3328-90; American Society for Testing and Materials: Philadelphia, 1990. (29) Killops, S. D.; Readman, J. W. Org. Geochem. 1985, 8, 247-257. (30) Lundanes, E.; Greibrokk, T. J. High Resolut. Chromatogr. 1994, 17, 197-202. (31) Standard Practice for Oil Spill Source Identification by GasChromatography and Positive Ion Electron Impact Low Resolution Mass Spectrometry; D-5739-00; American Society for Testing and Materials: Philadelphia, 1990. (32) Wang, Z.; Fingas, M. J. Chromatogr. A 1995, 712, 321-343. (33) Wang, Z.; Fingas, M. J. Chromatogr. A 1997, 774, 51-78. (34) Boehm, P. D.; Douglas, G. S.; Burns, W. A.; Mankiewicz, P. J.; Page, D. S.; Bence, A. E. Mar. Pollut. Bull. 1997, 34, 599-613. (35) Wang, Z.; Fingas, M.; Page, D. S. J. Chromatogr. A 1999, 843, 369-411. (36) Volkman, J. K. Org. Geochem. 1984, 6, 619-632. (37) Gough, M. A.; Rhead, M. M.; Rowland, S. J. Org. Geochem. 1992, 18, 17-22. (38) Rowland, S.; Donkin, P.; Smith, E.; Wraige, E. Environ. Sci. Technol. 2001, 35, 2640-2644. (39) Smith, E.; Wraige, E.; Donkin, P.; Rowland, S. Environ. Toxicol. Chem. 2001, 20, 2428-2432. (40) Richter, B. E. J. Chromatogr. A 2000, 874, 217-224. VOL. 37, NO. 8, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1661

(41) Gaines, R. B.; Frysinger, G. S.; Hendrick-Smith, M. A.; Stuart, J. D. Environ. Sci. Technol. 1999, 33, 2106-2112. (42) Philips J. B.; Beens, J. J. Chromatogr. A. 1999, 856, 331-347. (43) Bertsch, W. J. High Resolut. Chromatogr. 2000, 23, 167-181. (44) Liu, Z.; Lee, M. L. J. Microcolumn Sep. 2000, 12, 241-254. (45) Phillips, J. B.; Gaines, R. B.; Blomberg, J.; van der Wielen, F. W. M.; Dimandja, J.-M.; Green, V.; Granger, J.; Patterson, D.; Racovalis, L.; de Geus, H.-J.; de Boer, J.; Haglund, P.; Lipsky, J.; Sinha, V.; Ledford, E. B. J. High Resolut. Chromatogr. 1999, 22, 3-10. (46) Wang, Z.; Fingas, M.; Blenkinsopp, S.; Sergy, G.; Landriault, M.; Sigouin, L.; Foght, J.; Semple, K.; Westlake, D. W. S. J. Chromatogr. A. 1998, 809, 89-107. (47) Albaiges, J.; Albrecht, P. Int. J. Environ. Anal. Chem. 1979, 6, 171-190. (48) Frysinger, G. S.; Gaines, R. B.; Ledford, E. B., Jr. J. High Resolut. Chromatogr. 1999, 22, 195-200. (49) Frysinger, G. S.; Gaines, R. B. J. High Resolut. Chromatogr. 2000, 23, 197-201. (50) Gaines, R. B.; Ledford, E. B., Jr.; Stuart, J. D. J. Microcolumn Sep. 1998, 10, 597-604. (51) Frysinger, G. S.; Gaines, R. B. J. Forensic Sci. 2002, 47, 471-482. (52) Beens, J.; Blomberg, J.; Schoenmakers, P. J. J. High Resolut. Chromatogr. 2000, 23, 182-188. (53) Blomberg, J.; Schoenmakers, P. J.; Beens, J.; Tijssen, R. J. High Resolut. Chromatogr. 1997, 20, 539-544.

1662

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 8, 2003

(54) Venkatramani, C. J.; Phillips, J. B. J. Microcolumn Sep. 1993, 5, 511-516. (55) Frysinger, G. S.; Gaines, R. B. J. Sep. Sci. 2001, 24, 87-96. (56) Frysinger, G. S.; Gaines, R. B. J. High Resolut. Chromatogr. 1999, 22, 251-255. (57) van Deursen, M.; Beens, M, J.; Reijenga, J.; Lipman, P.; Cramers, C. J. High. Resolut. Chromatogr. 2000, 23, 507-510. (58) Potter, T. L.; Simmons, K. E. TPH Working Group Series, Vol. 2: Composition of Petroleum Mixtures; Amherst Scientific Publishers: Amherst, MA, 1998. (59) Beens, J.; Brinkman, U. A. Th. Trends Anal. Chem. 2000, 19, 260-275. (60) Volkman, J. K.; Holdsworth, D. G.; Neill, G. P.; Bavor, H. J., Jr. Sci. Total Environ. 1992, 112, 203-219. (61) Bieger, T.; Hellou, J.; Abrajano, T. A., Jr. Mar. Pollut. Bull. 1996, 32, 270-274. (62) Hites, R. A.; LaFlamme, R. E.; Windsor, J. G.; Farrington, J. W.; Deuser, W. G. Geochim. Cosmochim. Acta 1980, 44, 873-878. (63) Youngblood, W. W.; Blumer, M. Geochim. Cosmochim. Acta 1975, 39, 1303-1314.

Received for review May 20, 2002. Revised manuscript received January 27, 2003. Accepted January 28, 2003. ES020742N