Anal. Chem. 1996, 68, 144-155
Quantitative Analysis of Fuel-Related Hydrocarbons in Surface Water and Wastewater Samples by Solid-Phase Microextraction John J. Langenfeld, Steven B. Hawthorne,* and David J. Miller
Energy and Environmental Research Center, P.O. Box 9018, University of North Dakota, Grand Forks, North Dakota 58202
Solid-phase microextraction (SPME) parameters were examined on water contaminated with hydrocarbons including benzene and alkylbenzenes, n-alkanes, and polycyclic aromatic hydrocarbons (PAHs). Absorption equilibration times ranged from several minutes for low molecular weight compounds such as benzene to 5 h for high molecular weight compounds such as benzo[a]pyrene. Under equilibrium conditions, SPME analysis with GC/FID was linear over 3-6 orders of magnitude, with linear correlation coefficients (r2) greater than 0.96. Experimentally determined FID detection limits ranged from ∼30 ppt (w/w hydrocarbon/sample water) for high molecular weight PAHs (e.g., MW > 202) to ∼1 ppb for low molecular weight aromatic hydrocarbons. Experimental distribution constants (K) were different with 100and 7-µm poly(dimethylsiloxane) fibers, and poor correlations with previously published values suggest that K depends on the fiber coating thickness and the sorbent preparation method. The sensitivity of SPME analysis is not significantly enhanced by larger sample volumes, since increasing the water volume (e.g., from 1 to 100 mL) has little effect on the number of analyte molecules absorbed by the fiber, especially for compounds with K < 500. Water sample storage should utilize silanized glassware, since hydrocarbon losses up to 70% could be attributed to unsilanized glassware walls when samples were stored for 48 h. Hydrocarbon losses at part-perbillion concentrations also occurred with surface waters due to partitioning onto part-per-thousand concentrations of suspended solids. Quantitative determinations of aromatic and aliphatic hydrocarbons (e.g., in gasolinecontaminated water) can be performed using GC/MS with deuterated internal standard or standard addition calibration as long as the target components or standards had unique ions for quantitation or sufficient chromatographic resolution from interferences. SPME analysis gave good quantitative performance with surface waters having high suspended sediment contents, as well as with coal gasification wastewater which contained matrix organics at 106-fold higher concentrations than the target aromatic hydrocarbons. Good agreement was obtained between a 45-min SPME and methylene chloride extraction for the determination of PAH concentrations in creosote-contaminated water, demonstrating that SPME is a useful 144 Analytical Chemistry, Vol. 68, No. 1, January 1, 1996
technique for the rapid determination of hydrocarbons in complex water matrices. Determination of organic pollutant identities and concentrations in surface water, groundwater, seawater, and wastewater is an important topic in environmental analysis. Although many man-made organic chemicals have been detected in the aquatic environment, petroleum hydrocarbons including BTEX (benzene, toluene, ethylbenzene, and xylenes), n-alkanes, and polycyclic aromatic hydrocarbons (PAHs) represent some of the most common pollutants found in water from the United States and other industrialized countries where petroleum products are used heavily. Current techniques available for the determination of volatile hydrocarbons in water include methods such as purge and trap or static headspace, while semi- and nonvolatile petroleumrelated compounds (e.g., PAHs) are usually extracted by liquidliquid extraction or solid-phase extraction (SPE). There are drawbacks to these methods, including leaks and carryover during purge and trap, the large consumption of organic solvents (i.e., methylene chloride) with liquid-liquid extraction, and the plugging and channeling problems, as well as the large sample sizes, during SPE of water with high suspended solids content.1-3 Solid-phase microextraction (SPME) has the potential to overcome many difficulties associated with conventional methods for the extraction of organics from water.4 SPME is advantageous compared to liquid-liquid extraction because it completely eliminates solvents from the extraction procedure and does not require bulky glassware. SPME is also not subject to breakthrough, plugging, and channeling problems that are often encountered during SPE. While liquid-liquid and solid-phase extraction rely on exhaustive removal of the analytes from the water sample, SPME is an equilibrium process that relies on analyte partitioning between the water and a polymeric extraction phase, which has most often been poly(dimethylsiloxane) coated onto a fused silica fiber with coating thicknesses ranging from 7 to 100 µm (although other geometries have been reported).5 SPME allows extraction and concentration to be performed in a single step and direct analyte transfer to a gas chromatograph by thermal desorption inside a heated injection port. As a result, (1) Markell, C.; Hagen, D. F.; Brunelle, V. A. LC-GC 1991, 9, 332. (2) Ho, J. S.; Tang, P. H.; Eichelberger, J. W.; Budde, W. L. J. Chromatogr. Sci. 1995, 33, 1. (3) Blevins, D. D.; Schultheis, S. K. LC-GC 1994, 12, 12. (4) Boyd-Boland, A. A.; Chai, M.; Luo, Y. Z.; Zhang, Z.; Yang, M. J.; Pawliszyn, J.; Gorecki, T. Environ. Sci. Technol. 1994, 28, 569A. (5) Wittkamp, B.; Hawthorne, S. B.; Tilotta, D. Manuscript in preparation. 0003-2700/96/0368-0144$12.00/0
© 1995 American Chemical Society
excellent detection limits have been reported for hydrophobic organics.6 The technique is simple to perform, solvent-free, and rapid enough that sample preparation time can be reduced to several minutes. This report describes the effects of SPME parameters that determine its sensitivity for the determination of hydrocarbons in water samples. Parameters that will be studied include the time to reach equilibrium and the effects of water volume, fiber coating thickness, glassware silanization, suspended solids content, and multiple extractions from the same sample. Finally, SPME will be used to identify and quantitate the hydrocarbon components in complex water samples using external standard calibration, internal standard calibration, and standard addition as well as comparison to liquid-liquid extraction. Problems that were encountered during the analysis of the real-world samples will also be discussed, along with their solutions.
EXPERIMENTAL SECTION Water Samples. Five different water samples were spiked with the petroleum hydrocarbons and used in this study as test matrices to optimize SPME conditions. HPLC-grade water was obtained from Fisher Scientific (Pittsburgh, PA). River water samples were obtained from the Red River near Grand Forks, ND, and the Little Missouri River, approximately 0.5 mi west of Medora, ND. Based on gravimetric analysis, the suspended solids contents were 430 and 1450 mg/L, respectively, and the total organic carbon (TOC) contents were 22 and 10 mg/L, respectively. Wetland water was obtained from a prairie pothole approximately 15 mi west of Larimore, ND, and contained 825 mg/L suspended solids and 27 mg/L TOC. Discharge water from a sewage treatment facility was obtained from the City of Grand Forks secondary wastewater treatment holding pond and contained 1460 mg/L suspended solids and 24 mg/L TOC. Four water samples were used that contained complex mixtures of organics. Gasoline-contaminated water was prepared by fortifying HPLC-grade water with a standard gasoline obtained from the American Petroleum Institute (API, Washington, DC). The gasoline was obtained with a certificate (Certificate No. 11105008, unleaded gas [API 91-1], 10-24-91) reporting the concentrations of over 200 gasoline components. Water contaminated by creosote-treated wood (14 mg/L TOC and 480 mg/L suspended solids) was collected from a wetland adjacent to a set of railroad tracks. Two different coal gasification wastewaters were obtained from experimental reactors utilizing low-rank coals at the Energy and Environmental Research Center and used without any other preparation. Standards. Three separate 1 mg/mL (each individual component) stock standards of the target pollutants were prepared. One consisted of benzene, toluene, ethylbenzene, and o-, m-, and p-xylene (BTEX) prepared in acetone. A second standard was prepared in acetone and contained PAHs including naphthalene, phenanthrene, anthracene, fluoranthene, pyrene, benz[a]anthracene, chrysene, and benzo[a]pyrene. A third stock standard of n-alkanes, including n-hexane, n-octane, and n-decane, was also prepared in acetone. Aliquots of the three stock standards were combined into single standards and diluted in acetone to concentrations of 200, 100, 50, 10, 1, and 0.1 µg/mL of each compound. (6) Arthur, C. L.; Pawliszyn, J. Anal. Chem. 1990, 62, 2145.
A separate 1 mg/mL standard containing benzene-d6, toluene-d8, ethylbenzene-d10, m-xylene-d10, p-xylene-d10, o-xylene-d10, naphthalened8, phenanthrene-d10, anthracene-d10, fluoranthene-d10, pyrene-d10, benz[a]anthracene-d12, chrysene-d12, and benzo[a]pyrene-d12 (Cambridge Isotopes, Woburn, MA) was prepared in acetone and diluted for gas chromatography/mass spectrometry (GC/MS) determinations. Samples for SPME optimization were prepared by spiking 2-8 µL of the appropriate standard into 2-38 mL of water inside either a 2-mL autosampler vial with a crimp-top septum (Hewlett-Packard, Avondale, PA) or a 40-mL screw-cap vial with a Teflon-lined septum (Supelco Corp., Bellefonte, PA). Aliquots of gasoline-contaminated water (38 mL), creosote-contaminated water (2 mL), or coal gasification wastewater (2 mL) were placed inside either a 2-mL autosampler vial or a 40-mL vial, and an internal standard or standard addition standard was added when appropriate. A magnetic stir bar was added to each vial (8 mm long × 1.5 mm diameter and 1 in. long × 5/16 in. diameter for 2- and 40-mL vials, respectively), and the vials were immediately sealed to prevent volatilization losses (note that there was no headspace above the water once the magnetic stir bar was added to the vials). Glassware silanization was performed by soaking the glassware overnight in a 15% (v/v) mixture of dichlorodimethylsilane (Aldrich Chemicals, Milwaukee, WI) in toluene. The glassware was rinsed in toluene and methanol and oven-dried for 1 h at 150 °C. Solid-Phase Microextraction Technique. SPME was performed with a commercially available 100- or 7-µm film thickness poly(dimethylsiloxane) fiber housed in its manual holder (Supelco Corp.). The vial septum was pierced with the fiber withdrawn inside the needle of the holder. Once the needle had penetrated the septum, the plunger on the fiber holder was depressed to expose the fiber to the water sample for 1-480 min. During the extraction, the water samples were continuously agitated with a magnetic stir bar on a stir plate revolving at 1000 rpm. Once the extraction was completed, the fiber was withdrawn back inside the fiber holder, removed from the water sample, and analyzed immediately. Analysis of the extracted components in the fiber was performed with a Hewlett-Packard 5890 Series II gas chromatograph equipped with a split/splitless injection port and a flame ionization detector (FID). The column was a 60-m DB-5 with an internal diameter of 0.32 mm and a stationary phase thickness of 0.25 µm (J&W Scientific, Folsom, CA). Hydrogen was used as the carrier gas, and the linear velocity was maintained at 50 cm/s. The chromatographic conditions, including the trapping temperature, fiber desorption temperature, and desorption time, were optimized previously.7 The following steps were used to transfer the extracted analytes from the SPME fiber to the chromatographic column: (1) the injection port septum was pierced with the fiber retracted inside the needle of the SPME device (the adjustable needle on the SPME manual holder was set to 3.4 cm, since this was the length of the syringe needle designed specifically for use with the Hewlett-Packard split/splitless injector); (2) the plunger on the SPME holder was depressed to expose the fiber inside the hot split/splitless injection port, which was maintained at 300 °C in the splitless mode (note that the maximum desorption (7) Langenfeld, J. J.; Hawthorne, S. B.; Miller, D. J. J. Chromatogr., in press.
Analytical Chemistry, Vol. 68, No. 1, January 1, 1996
145
temperature recommended by the manufacturer for a 100-µm poly(dimethylsiloxane) fiber was 220 °C, but temperatures up to 300 °C were used without any detectable degradation in fiber performance or significant chromatographic background); (3) the analytes were thermally desorbed inside the injection port for 3 min and swept into the column by the carrier gas, where they were trapped and focused at the column inlet, which was maintained at -30 °C by cryogenically cooling the GC oven with liquid nitrogen; (4) the fiber was retracted back into the fiber holder assembly and removed from the GC injection port; and (5) the chromatographic separation was performed in the normal manner. The column temperature program was -30 °C for 3 min (note that the 3-min holding time at -30 °C is when the thermal desorption occurs in the splitless mode, and once the thermal desorption is complete, the split vent is opened for the remainder of the chromatographic separation), ramped at 70 °C/min to 0 °C, ramped at 10 °C/min to 280 °C, ramped at 70 °C/min to 320 °C, and held at 320 °C for 10 min. The injector and detector temperatures were maintained at 300 °C and 320 °C, respectively, throughout the entire chromatographic separation. Prior to SPME analyses, the GC/FID response was calibrated by solvent injections of the prepared standards into a cool on-column injector so that K for each compound could be determined, and the FID response was found to be linear (r2 > 0.99) over 5 orders of magnitude (ranging from 100 pg to 10 µg for all of the BTEX, n-alkane, and PAH components). Analysis by GC/MS was performed with a Hewlett-Packard 5985 GC/MS equipped with a 5890 Series II gas chromatograph and a split/splitless injection port. The column used was either a 60-m DB-5 (0.32 mm i.d., 0.25 µm film thickness) or a 30-m DB-1 (0.32 mm i.d., 5 µm film thickness). The mass spectrometer was operated in the full-scan mode between 50 and 300 amu, and ionization was carried out in the electron impact mode (70 eV). The transfer line temperature was maintained at 280 °C, the ionsource temperature was at 200 °C, and the electron multiplier voltage was set at 2600 V. All other chromatographic conditions were identical to those described earlier for GC/FID.
RESULTS AND DISCUSSION Sampling Equilibration Times. Since SPME is an equilibrium extraction method, the time to reach absorption equilibrium determines the maximum amount of analyte that can be extracted by the fiber and, therefore, controls the sensitivity of the method. The absorption equilibration time can be determined by extracting a sample for different times and noting the amount of time it takes for the amount extracted by the fiber to remain constant (as determined by GC analysis of the desorbed compounds from the fiber). Previous reports have shown that the BTEX components reach equilibrium in 95% of the equilibrium extractable mass) were ∼15 min for (8) Potter, D. W.; Pawliszyn, J. J. Chromatogr. 1992, 625, 247.
146 Analytical Chemistry, Vol. 68, No. 1, January 1, 1996
Figure 1. Equilibration time profiles of n-alkanes and PAHs from a 1 ppb water standard with a 100-µm poly(dimethylsiloxane) fiber. The break in the graph at 100 min was used to expand the features of the initial stages of the extraction. The percent extracted for each compound was normalized to the quantity extracted after 8 h.
BTEX components, in agreement with a previous report.8 However, the equilibration times for n-alkanes and PAHs were longer. All of the n-alkanes required approximately 30 min to reach equilibrium. Equilibration times for the PAHs generally increased with molecular weight and ranged between 10 min (naphthalene), 45 min (phenanthrene), 60 min (anthracene and pyrene), 90 min (fluoranthene), and 300 min (benz[a]anthracene, chrysene, and benzo[a]pyrene). Further SPME extractions were performed at times as long as 1 week without significant increases in the amounts of hydrocarbon that were extracted, which indicates that all of the hydrocarbons were at equilibrium after a 5-h extraction time. While the large differences between the equilibrium times for PAHs appear to be related to the value of K (i.e., compounds with higher K take longer to come to equilibrium), the long equilibration times actually reflect the kinetics of the SPME process (i.e., diffusion in the fiber coating),9 since high molecular weight compounds have lower diffusion coefficients. SPME Linearity and FID Detection Limits. Hydrocarbon detection limits and linear ranges were examined with a 100-µm poly(dimethylsiloxane) fiber by extracting eight water standards ranging from 0.027 ppb (w/v) to 235 000 ppb (38 mL of water) with SPME under equilibrium conditions (5 h). The experimentally determined linear range of SPME, the water solubility of each hydrocarbon, and the linear correlation coefficients (r2) for the linear range determinations are shown in Table 1. SPME was linear over approximately 3-6 orders of magnitude for all of the hydrocarbons that were tested, with linear correlation coefficients (r2) greater than 0.96 in all cases. In general, the linearity of SPME ranged from a practical detection limit (conservatively measured as the hydrocarbon concentration in water that gave an FID signal-to-noise ratio [S/N] of at least 12:1) that was measured in the mid-part-per-trillion range for most of the hydrocarbons to the highest concentrations tested for each compound (i.e., 16-80% of their maximum solubility in water). Since the relative response factors for hydrocarbons on an FID (9) Louch, D.; Motlagh, S.; Pawliszyn, J. Anal. Chem. 1992, 64, 1187.
Table 1. Experimentally Determined Equilibrium Partition Coefficients, Linear Ranges, and Correlation Coefficients for SPME of Hydrocarbons from Pure Water Using GC/FID
benzene toluene ethylbenzene m + p-xylene o-xylene n-hexane n-octane n-decane naphthalene phenanthrene anthracene fluoranthene pyrene benz[a]anthracene chrysene benzo[a]pyrene
100 µm
K 7 µm
published
lowest standard tested (detection limit, ppb)f
highest standard tested (ppb)
water solubility (ppb)g
r2
70 147 308 317 262 1 940 3 560 2 070 705 2 550 1 380 12 800 11 700 6 710 9 380 2 980
97 383 519 448 365 2 230 10 000 11 530 514 26 310 9 380 24 080 27 730 28 920 52 600 18 390
199a 126b 758,a 340b 2137a, 528b 2041a,c 1819a, 654b nad nad nad 1023e nad 12 589e nad nad 91 201e naa 72 443e
3.6 0.93 0.31 0.36 0.36 0.16 0.11 0.07 0.42 0.14 0.39 0.03 0.03 0.05 0.05 0.07
510 000 235 000 30 548 31 634 29 904 4 160 327 4.7 12 356 253 12.3 58 61 3.6 1.6 13.2
1 780 000 515 000 152 000 198 000h 175 000 13 000 700 9 30 000 1 100 40 260 140 14 2 50
0.99 0.99 0.99 0.99 0.99 0.98 0.98 0.96 0.99 0.99 0.99 0.98 0.99 0.97 0.96 0.97
a Values of K using an optical fiber coated with 100-µm poly(dimethylsiloxane) from Polymicro Technologies.8 b Values of K using an optical fiber coated with 56-µm poly(dimethylsiloxane) from Polymicro Technologies.10 c Values of K for m-xylene only. d Data not available for these compounds. e Values of K using an optical fiber coated with 15-µm poly(dimethylsiloxane) from Polymicro Technologies.11 f Detection limits were based on the lowest analyte concentration in water that was tested. At the concentrations stated in the table, each analyte gave a S/N ratio of at least 12:1 after 5-h SPME with a 100-µm poly(dimethylsiloxane) fiber. g Water solubilities at 20 °C.17 h Solubility of p-xylene.
are nearly identical on a mass basis, the detection limits parallel the values of K. For example, K for phenanthrene is approximately 3 times higher than that for naphthalene (with the 100-µm fiber), and the detection limit for phenanthrene is ∼3 times lower than that for naphthalene. The linear ranges and practical detection limits reported in Table 1 were determined under equilibrium conditions, which were unreasonably long for the high molecular weight PAHs. However, as shown in Figure 1, decreasing the extraction time from 5 h to 45 min only lowered the amount extracted by ∼4050% for the PAHs with the longest equilibration times (e.g., chrysene and benzo[a]pyrene). Therefore, shorter extraction times can be utilized while still maintaining low to sub-part-perbillion detection limits using FID for all of the test compounds in Table 1. Distribution Constants (K) and Effect of Fiber Coating. Table 1 also shows a comparison of experimentally determined values of K for 100- and 7-µm poly(dimethylsiloxane) fibers determined from the slope of the linear range (equilibrated for 5 h). K ranged from 70 for benzene to 12 800 for fluoranthene when measured with a 100-µm poly(dimethylsiloxane) fiber. When values of K were determined with a 7-µm poly(dimethylsiloxane) fiber, they were generally higher than those determined with the 100-µm fiber, most notably for the high molecular weight compounds. For example, K was 2-10 times higher with the 7-µm fiber for n-alkanes (except n-hexane) and PAHs with molecular weights greater than that of naphthalene (MW ) 128). Although K would be expected to be the same for a particular system (i.e., poly(dimethylsiloxane)-water) regardless of the coating thickness, this was not the case. A possible explanation is that the 100-µm poly(dimethylsiloxane) fiber was not at equilibrium. However, as previously discussed, equilibrations as long as 1 week were performed without detectable increases in extractable mass compared to a 5-h equilibration. Furthermore, any effects of water saturation and glassware losses can be eliminated because values of K were calculated from the linear range determinations which extended up to 6 orders of magnitude with linear correlation
coefficients greater than 0.96 (as opposed to determinations of K based on single concentration measurements). The values of K determined in this study were also compared with previously published values of K for the same compounds, which are also shown in Table 1. Reasonable agreement was obtained between the values of K for BTEX compounds determined in this study (7-µm fiber) and those determined previously with a homemade 56-µm fiber.10 However, previously published values of K for BTEX were found to differ up to a factor of 4 between themselves. For example, Potter and Pawliszyn8 used a homemade 100-µm poly(dimethylsiloxane) fiber and determined values of K for toluene, ethylbenzene, and o-xylene to be 758, 2137, and 1819, respectively, while Arthur et al.10 used a homemade 56-µm poly(dimethylsiloxane) fiber and reported K for the same compounds to be 340, 528, and 654. Although no previous data have been reported for n-alkanes, values of K for low molecular weight PAHs (naphthalene and anthracene) in this study (7-µm fiber) showed reasonable agreement with previously published values determined with a homemade 15-µm poly(dimethylsiloxane) fiber,11 but values of K for high molecular weight PAHs (benz[a]anthracene and benzo[a]pyrene with 100- or 7-µm fibers) in this study were significantly lower than the previously reported values. It is also not likely that the acetone spiking solvent affected values of K since the typical acetone concentration was
