2003 The Japan Society for Analytical Chemistry 1365 Determination of Benzene, Toluene, Ethylbenzene and Xylene in River Water by Solid-Phase Extraction and Gas Chromatography Mohammad A. MOTTALEB,* Mohammad Z. ABEDIN,** and Mohammad S. ISLAM** *Department of Chemistry, University of Rajshahi, Rajshahi 6205, Bangladesh **Department of Chemical Technology and Polymer Sciences, Shahjalal University of Science and Technology, Sylhet 3114, Bangladesh A rapid and reproducible method is described that employs solid-phase extraction (SPE) using dichloromethane, followed by gas chromatography (GC) with flame ionization detection for the determination of benzene, toluene, ethylbenzene, xylene and cumene (BTEXC) from Buriganga River water of Bangladesh. The method was applied to detect BTEXC in a sample collected from the surface, or 5 cm depth of water. Two-hundred milliliters of n-hexane-pretreated and filtered water samples were applied directly to a C 18 SPE column. BTEXC were extracted with dichloromethane and the BTEX concentrations were obtained to be 0.1 to 0.37 µg ml 1. The highest concentration of benzene was found as 0.37 µg ml 1 with a relative standard deviation (RSD) of 6.2; cumene was not detected. The factors influencing SPE e.g., adsorbent types, sample load volume, eluting solvent, headspace and temperatures, were investigated. A cartridge containing a C 18 adsorbent and using dichloromethane gave a better performance for the extraction of BTEXC from water. Average recoveries exceeding 90 could be achieved for cumene at 4 C with a 2.7 RSD. (Received October 1, 2002; Accepted July 17, 2003) Introduction Volatile organic compounds (VOC), e.g., benzene, toluene, ethylbenzene, xylene and cumene (BTEXC), are important environmental contaminants because of their high toxicity and widespread occurrence. They are present in aviation fuel (gasoline) and are widely used as industrial solvents and raw materials for the production of different commodities. 1 Benzene, toluene and ethylbenzene are among compounds designated as priority pollutants by the US EPA, and the action and risk levels of benzene, toluene, ethylbenzene, and xylene are described in the Dutch Government Quality Standards for the Assessment of Soil and Water Contamination. 2 The determination of BTEXC in environmental matrices is difficult because of their trace-level presence and losses incurred during sample handling, extraction etc. Recently, there has been considerable interest in the development of SPE columns for the clean-up, extraction, and pre-concentration of liquid samples. 3 SPE applications are found in different environmental areas, such as soils and sediments, 4,5 industrial effluents 6 and water samples. 7 9 SPE has been used for the extraction of BTEXC 10 and pesticides 11 from soils and sediments. A few SPE applications have also To whom correspondence should be addressed. E-mail: Mottaleb.Mohammad@epamail.epa.gov M. A. M. present address: EPA/NRC Postdoctoral Research Associate, Environmental Chemistry Branch, Environmental Sciences Division, National Exposure Research Laboratory, U.S. Environmental Protection Agency, P.O. Box 93478, Las Vegas, NV 89193-3478, USA. appeared for VOC components in the BTEXC analysis of water samples; a cartridge containing C 18 adsorbent was used to extract BTEX from industrial effluents 6 and benzene and toluene from seawater. 7 Although solid-phase micro-extraction (SPME) was also applied to isolate BTEX from water samples either by direct adsorption from the liquid 8 or via headspace sampling, 9 the limitation of SPME for the quantitation of sulfurbased VOC has been reported. 12 The VOCs analyzed from air and water samples by open-tubular, wall-coated columns, 13 solvent trapping 14 and semi-voc from air by atmosphericpressure chemical ionization mass spectrometry 15 were also described. Buriganga river water is important because it flows through the capital of Bangladesh and has an enormous impact on the socio-economic development of the country, especially industrial and shipping-marine activities. Tanker-washing sewage, shipping scrap particles and oil spillage are common features on the river at different ghats (marine terminals). As a result, the water is continually polluted by various organic compounds, especially hydrocarbons. In addition, aromatic solvents are increasingly used in industry, 1 and the wastes are sometimes disposed of in the aquatic environment, which becomes increasingly contaminated. To monitor contaminants, we recently determined the concentrations of various normal saturated hydrocarbons in Buriganga river water 16 and pesticides in soil 17 by GC, using liquid-liquid extraction and SPE methods, respectively. We also reported on the development of a chromatography method for the determination and characterization of anionic detergents in river water. 18 The present paper describes the concentration levels of benzene, toluene, ethylbenzene, xylene and cumene in water samples collected at two depths from the Sadarghat, one of the biggest
1366 ANALYTICAL SCIENCES OCTOBER 2003, VOL. 19 Table 1 Concentration of a BTEXC standard solution used for GC optimization Solution Concentration of BTEXC standard solution/µg ml 1 Benzene Toluene Ethylbenzene Xylene Cumene Standard 1 8.77 8.67 8.67 8.61 8.61 Standard 2 17.14 17.34 17.35 17.22 17.22 Standard 3 26.31 26.01 26.01 25.83 25.83 Standard 4 35.08 34.68 34.68 34.44 34.44 marine terminals, on the Buriganga river. Also discussed are the recoveries of BTEXC using different adsorbents for SPE columns and factors influencing SPE adsorption, such as the sample load volume, eluting solvent, headspace and temperatures. Experimental Apparatus and reagents A Varian gas chromatograph (Model 3300) equipped with a flame ionization detector (FID) was used in this study. A DB-1 fused-silica mega-bore analytical column (15 m 0.53 mm i.d., 1.5 µm thickness) and a phenyl-methyl deactivated guard column were used. An integrator (Varian Model 4290) gave the peak area and retention time of the peaks separated by GC-FID. Methanol (HPLC grade), hexane, dichloromethane and chloroform were obtained from Merck Ltd. (Germany). VOCs and internal standards were purchased from Sigma Chemicals Ltd., USA. The C 18, C 8 and phenyl (PH) (500 mg, 3 ml) SPE cartridges were obtained from the Supelco Ltd. Stock and working standard solutions Stock solution. An aliquot (10 µl) of each of the BTEXC constituents and internal standard dichlorobenzene (DCB) were dissolved in 100 ml of dichloromethane. The concentrations of the BTEXC components corresponded to 87.7, 86.7, 86.7, 86.1, 86.1 µg ml 1, respectively and the concentration of DCB corresponded to 130.5 µg ml 1. Working standard solutions. Four standard solutions were made for measuring the linearity of the GC response. The concentrations of the standard solutions are given in Table 1. Collection of river water samples Contaminated water samples were collected in 1-l dark glass bottles on September 21, 1999 from the Sadarghat. A map of the Buriganga river and the location of the Sadarghat area have been reported. 16 Cleaned bottles were rinsed with sample water prior to sample collection. Ten-liter samples were collected at the surface and ten samples were also collected at a depth of 5 cm. The distance of the sample collection point from the river bank was about 200 m. Method development work was carried out using double-distilled water containing BTEXC at the same concentrations as those found in the river water. Extraction of VOC from synthetic and river water samples A 10 ml volume of a synthetic sample was extracted with 10 ml of hexane in a 30 ml vial; the layers were allowed to separate. Prior to SPE work, 2 to 4 ml of the organic layer was removed and stored in a sealed glass vial at 4 C. The extraction of a river-water sample was carried out essentially according to a reported method. 16 Each (500 ml) water sample was shaken Fig. 1 GC-FID detection chromatogram for a standard solution of BTEXC, and internal standard (dichlorobenzene). Conditions: gas flow rates: N 2 (carrier gas), 4 ml min 1 ; H 2, 33 ml min 1 ; air, 330 ml min 1 ; detector temperature, 180 C; injector temperature, 180 C; mega-bore, DB-1 fused-silica analytical column; oven temperature program: 35 C for 5 min, to 70 C at 5 C min 1, to 180 C at 15 C min 1 and hold 10 min at 180 C; and sample injected volume, 1 µl. vigorously for 30 min with 50 ml of hexane at 4 C. The aqueous layer was separated and extracted again with 25 ml of hexane. The combined extracts were then stored at 4 C for SPE. Prior to direct use of the water sample in SPE, the sample (200 ml) was filtered by a 0.45 µm nylon membrane. Solid-phase extraction The column was activated with 3 ml 50 methanol and preequilibrated with 3 ml 1 methanol. The river-water (100 ml) sample was loaded on the column at 3 ml min 1. Elution was carried out with 2 portions of 2 ml aqueous 1 methanol. Finally, solutes were eluted with two aliquots of 2 ml of dichloromethane. Similar elution profiles were obtained for recovery experiments. DCB (200 µl or 5) was added as an internal standard prior to a GC analysis. Passing samples through a dryer containing sodium sulfate only eliminated traces of water. Calculation of response factor and concentration of components The relative response factor of a component (R F) to the internal standard of DCB is given by C DCB A DCB R F = Cc, where C DCB and C c represent the concentrations of DCB and the component analyte, respectively, in terms of µg/ml. The terms A DCB and A c indicate the peak-area counts from the integrator for DCB and the component analyte, respectively. The response factors for all components were calculated as mentioned above, and the concentration of each component (C c) was calculated as follows: A c A DCB A c C DCB C c =. RF
1367 Table 2 SPE of river-water samples at 4 C (C 18 column, sample volume 200 ml) Concentration of BTEXC component in river water/µg ml 1 Analyte Average conc. 0 cm depth SD a RSD, b Average conc. 5 cm depth SD a RSD, b Benzene 0.350 ± 0.034 9.8 0.372 ± 0.0230 6.2 Toluene 0.234 ± 0.033 14.3 0.167 ± 0.0289 17.4 Ethylbenzene 0.145 ± 0.0230 15.8 0.104 ± 0.0209 20.2 Xylene 0.124 ± 0.0206 16.1 0.144 ± 0.0135 9.36 Cumene a. SD represents standard deviation, which was calculated from each of three measurements. b. RSD means relative standard deviation. Fig. 2 Calibration curve for optimization of the GC-FID system for the analysis of BTEXC in river-water samples. Concentration of the injected BTEXC solution (please see Experimental, preparation of stock and working standard solutions). The operating conditions are the same as in Fig. 1. Fig. 3 GC-FID chromatograms for the river-water samples. For (A), water was collected from the surface, and for (B), water was collected from 5 cm depth. The operating conditions are the same as in Fig. 1. Results and Discussion GC optimization The GC-FID system used was optimized before the VOC measurement. Separations were achieved with different temperature programs. A good separation of individual BTEXC constituents, including DCB, was obtained under the following conditions: injector temperature, 180 C; column oven temperature, 35 C for 5 min, to 70 C at 5 C min 1, to 180 C at 15 C min 1 and hold at 180 C for 10 min. The detector temperature was 180 C and the flow rate of the nitrogen carrier gas was 4 ml min 1. Figure 1 is a GC-FID chromatogram of a standard solution of the BTEXC components, showing that the components and DCB were well-resolved. The linearity of the detector response was also demonstrated by injecting the working standard solutions into the GC-FID instrument. Figure 2 depicts the calibration graphs of the peak area versus the concentration. The FID gave good linearity of the response for the detection of each of the BTEXC constituents. Hence, it was decided that the above conditions could be used for the determination of BTEXC from the river-water samples. Presence of VOC in water samples Table 2 summarizes the concentration of BTEXC in the riverwater samples analyzed by SPE-GC-FID. The presence of ethylbenzene, xylene and cumene was not detected using the experimental conditions described previously, although trace levels of benzene and toluene were found. To detect the other constituents of the BTEXC family, an increased volume of 200 ml of water sample was directly applied to the SPE at 4 C. An appreciable amount of benzene and toluene, including trace levels of ethylbenzene and xylene, were obtained in the riverwater sample. However, cumene was never found. Figure 3 shows a representative GC-FID chromatogram of the riverwater samples. The chromatograms in Figs. 3(A) and 3(B) correspond to the surface and 5 cm depth of water, collected from the Buriganga river. These were obtained when 2 µl SPE eluted samples were injected into the GC and showed a similar chromatographic elution pattern with different magnitudes of the BTEX components peak. Blank experiments were performed prior to sample injection. Selection of adsorbents and eluting solvent To select the suitability of adsorbents and eluting solvents, the percentage recovery of BTEXC constituents was investigated using C 18, C 8 and PH cartridges with CH 2Cl 2 and CHCl 3 solvents. The recoveries obtained when 2 ml portions of standard VOC solutions were passed through different SPE columns and eluted with two portions of 2 ml of dichloromethane or chloroform at 4 C. The recovery results are presented in Table 3. The extractions were performed simultaneously for each solvent. Regardless of the solvents used, higher recoveries were obtained for xylene and cumene. This may have been due to a less evaporative loss of the two components because of their higher boiling point. Moreover, in comparison between CH 2Cl 2 and CHCl 3 solvents, it was observed that slightly better recoveries were obtained when CH 2Cl 2 was used as the eluting solvent (Table 3). This is probably due to a more non-polar interaction between a bonded
1368 ANALYTICAL SCIENCES OCTOBER 2003, VOL. 19 Table 3 Recovery of BTEXC from different SPE columns eluted with CH 2Cl 2 and CHCl 3 at a temperature of 4 C Solvent SPE column Recoveries () ± standard deviation (SD) a Benzene Toluene Ethylbenzene Xylene Cumene CH 2Cl 2 C 18 76.0 ± 2.5 80.5 ± 2.5 83.7 ± 3.5 87.1 ± 4.0 90.3 ± 3.3 C 8 70.0 ± 3.5 68.3 ± 2.0 74.6 ± 3.5 83.7 ± 1.5 79.6 ± 1.4 PH 73.0 ± 1.5 77.8 ± 3.5 70.4 ± 4.0 76.3 ± 1.5 82.5 ± 2.3 CHCl 3 C 18 74.0 ± 3.0 78.0 ± 2.5 80.0 ± 2.0 84.0 ± 3.0 87.0 ± 4.0 C 8 69.0 ± 3.5 67.0 ± 2.5 72.0 ± 3.0 80.0 ± 2.5 77.0 ± 3.0 PH 71.2 ± 3.0 76.0 ± 3.0 69.0 ± 4.0 75.0 ± 3.0 78.0 ± 2.5 a. SD values were calculated from each of five measurements. Table 4 Effect of the temperature on the recovery of BTEXC with CH 2Cl 2 using a C 18 cartridge Analytes At a temperature of 20 C Recovery, SD a RSD b, At a temperature of 4 C Recovery, SD a RSD b, Benzene 76 ± 3.0 3.9 79 ± 2.8 3.5 Toluene 82 ± 2.9 3.5 86 ± 2.5 2.9 Ethylbenzene 84 ± 3.3 3.9 91 ± 3.2 3.5 Xylene 87 ± 3.5 4.0 88 ± 2.7 3.1 Cumene 89 ± 2.9 3.3 91 ± 2.5 2.7 a. SD represents standard deviation, which was calculated from each of five measurements. b. RSD means relative standard deviation. Fig. 4 Effect of the sample load volume on the recoveries of the BTEXC constituents with a C 18 SPE column using dichloromethane solvent. phase and the CH 2Cl 2 system. Effect of the sample load volume The effect of the sample volume on the SPE recovery is one of the most important factors, because the SPE performance is affected by the amount of sample volume loaded on a particular column. 6,10 The break-through volume of a C 18 column (500 mg, 3 ml capacity) was determined by passing a number of VOC standard solutions to a volume of up to 500 ml. Known masses of the analytes were introduced. There were no appreciable changes in the recovery rates up to a sample volume of 400 ml. The percentage recovery of benzene and toluene decreases more rapidly than ethylbenzene, xylene and cumene. This observation confirms the fact that ethylbenzene, xylene and cumene possess higher breakthrough volumes than benzene and toluene (Fig. 4). Similar break-through volume curves were obtained for C 8 and phenyl substituted SPE columns. These are not shown. Factors affecting the SPE performance investigations Temperature. To investigate the effect of the temperature on SPE performance, a cartridge containing C 18 material was employed with CH 2Cl 2 solvent at temperatures of 20 C and 4 C. An effect of temperature on the SPE recovery of BTEXC constituents was found to occur (Table 4). It has been observed that both liquid-liquid and solid-phase extractions provided slightly better recoveries when experiments were carried out at Table 5 Effect of the headspace on the recovery of BTEXC with CH 2Cl 2 using a C 18 cartridge at 4 C Analyte Headspace No headspace Recovery, SD a Recovery, SD a Recovery difference, Benzene 67.0 ±2.0 76.0 ±3.0 11.8 Toluene 72.0 ±3.5 81.0 ±2.5 11.1 Ethlbenzene 74.0 ±3.5 84.0 ±2.5 11.9 Xylene 79.0 ±2.7 87.0 ±4.0 9.2 Cumene 82.0 ±2.8 90.0 ±3.5 8.9 a. SD represents the standard deviation, which was calculated from each of five measurements. a temperature of 4 C. The relative standard deviations (RSD) were calculated for each of the BTEXC constituents and temperatures. At 20 C, the RSD of recoveries were between 3.3 and 4.0; however, at 4 C improved recoveries of the constituents were achieved with RSD values of 2.7 to 3.5. Headspace. In order to investigate the effect of headspace, recovery experiments for each of the BTEXC components were performed using a C 18 column in an environment at a temperature of 4 C. This temperature provided recoveries of 79 to 91 of the constituents (Table 4). Parallel extractions of the components with and without headspace were performed. The results are presented Table 5. It can be seen that the incurred losses of the BTEXC constituents occurred in the range of 9 to 12 due to allowing a headspace. Since trace levels of BTEXC components could be found in water, or other real samples, the above losses (9 12) are quite significant. Thus, it was decided that the existence of a headspace increases the
1369 evaporative losses, which decreases the percentage recovery. This study was conducted while allowing no headspace in the sampling and at a temperature of 4 C. Acknowledgements The authors would like to thank the Bangladesh University grants commission for financial support to carry out this research. References 1. J. J. G. Klist, in Chemistry and Analysis of VOCs in the Environment, ed. H. J. T. Bloemen and J. Burn, 1993, Glasgow. 2. Environmental Quality Standards for Soil and Water, Netherlands Ministry of Housing, Physical Planning and Environment, 1991, Leidschendam. 3. L. A. Berrueta, B. Gallo, and F. Vicente, Chromatographia, 1995, 40, 474. 4. P. Loconto, LC-GC Int., 1991, 4, 10. 5. Z. Zhang and J. Pawliszyn, Anal. Chem., 1995, 67, 34. 6. I. S. Deans, C. M. Davidson, D. Littlejohn, and L. Brown, Analyst, 1993, 118, 1375. 7. W. A. Saner, J. R. Djadamec, R. W. Sager, and T. J. Kleen, Anal. Chem., 1979, 51, 2180. 8. C. L. Auther, J. Pawliszyn, and P. R. Belardi, J. High. Res. Chromatogr., 1992, 15, 741. 9. Z. Zhang and J. Pawliszyn, Anal. Chem., 1993, 65, 1843. 10. K. M. Meney, C. M. Davidson, and D. Littlejohn, Analyst, 1998, 123, 195. 11. M. J. Redondo, M. J. Ruiz, B. Boluda, and G. Font, Chromatographia, 1993, 36, 147. 12. R. A. Murray, Anal. Chem., 2001, 73, 1646. 13. B. C. D. Tan, P. J. Marriott, H. K. Lee, and P. D. Morrison, Analyst, 2000, 125, 469. 14. M. A. Stone and L. T. Taylor, Anal. Chem., 2000, 72, 1268. 15. L. Charles, L. S. Riter, and R. G. Cooks, Anal. Chem., 2001, 73, 5061. 16. M. A. Mottaleb, M. Ferdous, M. S. Islam, and M. A. Hossain, Anal. Sci., 1999, 15, 995. 17. M. A. Mottaleb and M. Z. Abedin, Anal. Sci., 1999, 15, 283. 18. M. A. Mottaleb, Mikrochim. Acta, 1999, 132, 31.