Determination of Extractable Organic Chlorine and Bromine by Probe

Tone Normann Asp,*,† Stig Pedersen-Bjergaard,‡ and Tyge Greibrokk§. Department of ... emission was measured for the determination of EOCl and. EO...
1 downloads 0 Views 148KB Size
Anal. Chem. 1997, 69, 3558-3564

Determination of Extractable Organic Chlorine and Bromine by Probe Injection Dual-Microplasma Atomic Emission Spectrometry Tone Normann Asp,*,† Stig Pedersen-Bjergaard,‡ and Tyge Greibrokk§

Department of Food Hygiene, Norwegian College of Veterinary Medicine, P.O. Box 8146-Dep, 0033 Oslo, Norway, School of Pharmacy, University of Oslo, P.O. Box 1068 Blindern, 0316 Oslo, Norway, and Department of Chemistry, University of Oslo, P.O. Box 1033 Blindern, 0315 Oslo, Norway

A probe injection dual-microplasma spectrometer is evaluated as a low-cost alternative for the determination of extractable organic chlorine and bromine (EOCl and EOBr). The system consists of two 350 kHz plasmas sustained in the same stream of helium and a probe for sample application in the interplasma region. The sample was applied with a microsyringe into a small cup on the sample probe. Subsequently, the extraction solvent was gently evaporated, and the sample cup was pushed into the interplasma region. The first plasma was in direct contact with the sample probe and served to rapidly vaporize the sample material. The vaporized sample was then transferred to the second plasma, where atomic emission was measured for the determination of EOCl and EOBr. For both Cl and Br, 120 pg detection limits and 1000:1 halogen-to-carbon selectivities were obtained, and responses were linear over 3 orders of magnitude. The release of halogenated organic compounds to the environment has received increasing interest during the last decades. Most of these compounds are of xenobiotic nature and, therefore, are potential environmental hazards. Because of the large number of halogenated compounds with different chemical and physical properties, individual analysis of each of them would be impractical and presently also impossible. Sum parameters such as extractable organic halogen/chlorine/bromine (EOX/EOCl/EOBr) are, therefore, often used to estimate the total load of halogenated compounds into the environment. Several different extraction methods have been developed to determine the sum parameters.1-8 Most of them utilize either microcoulometric titration or neutron activation analysis (NAA) as detection methods. The microcoulometric titration method is relatively inexpensive but cannot distinguish between chlorine and bromine (EOCl and EOBr). †

Norwegian College of Veterinary Medicine. School of Pharmacy, University of Oslo. § Department of Chemistry, University of Oslo. (1) Wegman, R. C. C.; Greve, P. A. Sci. Total Environ. 1977, 7, 235-245. (2) Ahnoff, M.; Josefsson, B.; Lunde, G.; Andersson, G. Water Res. 1979, 13, 1233-1237. (3) Gether, J.; Lunde, G. Anal. Chim. Acta 1979, 108, 137-147. (4) Fawkes, J.; Albro, P. W.; Walters, D. B.; McKinney, J. D. Anal. Chem. 1982, 54, 1866-1871. (5) Bjørseth, A.; Carlberg, G. E.; Baumann Ofstad, E.; Rambæk, J. P.; Halvorsen, C. Anal. Chim. Acta 1984, 160, 257-262. (6) Martinsen, K.; Kringstad, A.; Carlberg, G. Water Sci. Technol. 1988, 20, 13-24. (7) Grøn, C. Vatten 1988, 44, 205-212. (8) van Strien, A.; Apples, J. Int. Lab. 1996, May, 9-10. ‡

3558 Analytical Chemistry, Vol. 69, No. 17, September 1, 1997

Although this is possible with NAA, this technique can hardly be characterized as a routine method since very special equipment is required for neutron irradiation of the samples. In addition to the two classical methods, electrothermal vaporization inductively coupled plasma mass spectrometry (ETV-ICPMS) has recently been reported as a new technique for the determination of EOCl and EOBr.9,10 Unfortunately, although the general interest for ICPMS has increased rapidly, both the instrumental costs and the operating costs are very high. Thus, no low-cost alternative is currently available for differentiation between EOCl and EOBr. From a theoretical point of view, also plasma atomic emission spectrometry (AES) should be an attractive technique for the determination of EOCl and EOBr, owing to the high sensitivity and multielemental selectivity and to the linear and compoundindependent nature of elemental responses. Most work on AES is currently performed with inductively coupled plasmas sustained in argon (Ar-ICP). Unfortunately, Ar-ICP systems are inefficient for the determination of chlorine and bromine, and consequently, the potential of this technique is low for the determination of EOCl and EOBr. With helium discharges, in contrast, chlorine and bromine may be successfully detected both with ICP systems11-15 and with the small and more popular microwave-induced plasmas (He-MIP).16-23 However, most research in this field has been focused on inorganic chlorine and bromine, while reports on the determination of EOCl and EOBr are scarce. To develop a new, low-cost approach for the determination of EOCl and EOBr, attention was focused on plasma atomic emission spectrometry (AES) in the present work. A helium plasma was selected for its high excitation potential, and it was decided to (9) Richner, P.; Wunderli, S. J. Anal. At. Spectrom. 1993, 8, 45-49. (10) Manninen, P. K. G. J. Anal. At. Spectrom. 1994, 9, 209-211. (11) Wolnik, K. A.; Miller, D. C.; Selskar, C. J.; Fricke, F. L. Appl. Spectrosc. 1985, 39, 930-935. (12) Chan, S.-K.; Montaser, A. Spectrochim. Acta 1985, 40B, 1467-1472. (13) Chan, S.-K.; Van Hoven, R. L.; Montaser, A. Anal. Chem. 1986, 58, 23422343. (14) Chan, S.-K.; Montaser, A. Spectrochim. Acta 1987, 42B, 591-597. (15) Tan, H.; Chan, S.-K.; Montaser, A. Anal. Chem. 1988, 60, 2542-2544. (16) Beenakker, C. I. M.; Boumans, P. W. J. M.; Rommers, P. J. Philips Technol. Rev. 1980, 39, 65-77. (17) van Dalen, H. P. J.; Kwee, B. G.; de Galan, L. Anal. Chim. Acta 1982, 142, 159-171. (18) Rait, N.; Golightly, D. W.; Massoni, C. J. Spectrochim. Acta 1984, 39B, 931937. (19) Abdillahi, M. M.; Snook, R. D. Analyst 1986, 111, 265-267. (20) Matousek, J. P.; Orr, B. J.; Selby, M. Talanta 1986, 33, 875-882. (21) Kitagawa, K.; Mizutani, A.; Yanagisawa, M. Anal. Sci. 1989, 5, 539-544. (22) Wu, M.; Carnahan, J. W. Appl. Spectrosc. 1990, 44, 673-678. (23) Nakahara, T.; Morimoto, S.; Wasa, T. J. Anal. At. Spectrom. 1992, 7, 211217. S0003-2700(96)01219-X CCC: $14.00

© 1997 American Chemical Society

miniaturize the discharge system as compared with ICP to reduce the operational costs. Microwaves have frequently been applied to sustain small helium plasmas, but the energy coupling is relatively complicated and requires the use of special cavities. Therefore, in the present work, a 350 kHz helium plasma was utilized due to the straightforward energy coupling and because of the excellent analytical properties, as demonstrated for gas chromatographic detection.24-37 To enhance the sensitivity of the method, the extraction solvent was evaporated prior to analysis. The solid sample was subsequently introduced at the plasma inlet by a sample probe. To ensure a rapid and quantitative transfer of sample, the sample probe and the plasma gas were heated by a second plasma. This probe injection dual-microplasma spectrometer was optimized and validated in the present work for the determination of EOCl and EOBr. EXPERIMENTAL SECTION The instrumental setup of the probe injection dual-microplasma spectrometer is illustrated in Figure 1. Two silica tubes (40-50 mm, 1 mm i.d., 1/8 in. o.d.) were utilized to contain the vaporizing plasma and the analytical plasma. Both were connected to a 1/8 in. Swagelok cross union (Solon, OH) fitted with 1/8 in. graphite ferrules. To extend the vaporizing plasma to the sample probe, the corresponding inlet on the cross union was drilled to give an internal diameter of 1/8 in. The inlet of the vaporizing plasma was connected to the gas line (for plasma and dopant gases) with a 1/8-1/16 in. reducing union (Swagelok) fitted with graphite ferrules. Because this union also functioned as rf electrode for the vaporizing plasma, the gases were introduced through a nonconducting 320 µm o.d. fused silica capillary to ensure electrical insulation of the connection. The probe was a massive stainless steel rod (150 mm, 1/16 in. o.d.) with a 2 µL sample cup drilled in the middle. The probe was placed in the cross union perpendicular to the stream of helium through the two discharges. To avoid leakage of helium from the cross union, this was fitted with a 1/8 in. vespel ferrule at the left inlet and a rubber septum in the right inlet. Both sealings were carefully installed to allow horizontal movement of the probe.

Figure 1. Probe injection dual-radio frequency plasma system. (A) General view of instrument and (B) probe concept.

(24) Skelton, R. J.; Markides, K. E.; Farnsworth, P. B.; Lee, M. L. J. High Resolut. Chromatogr. 1988, 11, 75-81. (25) Skelton, R. J., Jr.; Chang, H.-C. K.; Farnsworth, P. B.; Markides, K. E.; Lee, M. L. Anal. Chem. 1989, 61, 2292-2298. (26) Skelton, R. J., Jr.; Markides, K. E.; Lee, M. L.; Farnsworth, P. B. Appl. Spectrosc. 1990, 44, 853-857. (27) Chang, H.-C. K.; Skelton, R. J., Jr.; Markides, K. E.; Lee, M. L. Polycycl. Aromat. Compd. 1990, 1, 251-264. (28) Wu, M.; Lee, M. L.; Farnsworth, P. B. J. Anal. At. Spectrom. 1992, 7, 197200. (29) Wu, M.; Liu, Z.; Farnsworth, P. B.; Lee, M. L. Anal. Chem. 1993, 65, 21852188. (30) Farnsworth, P. B.; Wu, M.; Tacquard, M.; Lee, M. L. Appl. Spectrosc. 1994, 48, 742-746. (31) Pedersen-Bjergaard, S.; Greibrokk, T. J. High Resolut. Chromatogr. 1992, 15, 677-681. (32) Pedersen-Bjergaard, S.; Greibrokk, T. Anal. Chem. 1993, 65, 1998-2002. (33) Pedersen-Bjergaard, S.; Greibrokk, T. J. Microcolumn Sep. 1994, 6, 1118. (34) Pedersen-Bjergaard, S.; Greibrokk, T. J. Chromatogr. A 1994, 686, 109119. (35) Pedersen-Bjergaard, S.; Greibrokk, T. J. High Resolut. Chromatogr. 1995, 18, 1-5. (36) Asp, T. N.; Pedersen-Bjergaard, S.; Greibrokk, T. J. Chromatogr. A 1996, 736, 157-164. (37) Asp, T. N.; Pedersen-Bjergaard, S.; Greibrokk, T. J. High Resolut. Chromatogr. 1997, 20, 201-207.

Two HPG-2 radio frequency power supplies (ENI Power Systems, Rochester, NY) were utilized to sustain the dualmicroplasma system. The vaporizing discharge was sustained between the grounded cross union/sample probe and the reducing union at the inlet, with the latter working as the rf electrode. In a similar way, the analytical plasma was maintained utilizing a 2 mm o.d. steel rod at the outlet as the rf electrode and the cross union as ground. With two rf power supplies, individual operation of the two plasmas was possible. The optical part of the system has been described in detail elsewhere in connection with gas chromatographic detection.31-37 The analytical plasma was viewed side-on through the wall of the silica tube. Atomic emission was measured in the near-infrared part of the spectrum by a Model H-20 IR monochromator (Instruments SA, Metuchen, NJ) equipped with 500 µm slits. A long-pass filter with a 595 nm cutoff (Melles Griot, Irvine, CA) was used to reject second- and third-order radiation, and a pair of achromatic lenses (f ) 58 mm, 12.7 mm diameter; Newport Corp., Fountain Valley, CA) was used to focus the emission light in front of the monochromator. The photomultiplier tube (R2658) was Analytical Chemistry, Vol. 69, No. 17, September 1, 1997

3559

Table 1. Chlorinated and Brominated Model Compounds Used in the Experiments symbol

name

formula

A B C D E F G H I J

3,5-dichlorophenol 4,4′-dichlorobiphenyl 9,10-dichloroanthracene 4-(2,4-dichlorophenoxy)butyric acid Rhodamine B 4-bromophenol 4,4′-dibromobiphenyl bis(4-bromophenyl) ether 9,10-dibromoanthracene 2-bromohexadecanoic acid

C6H4Cl2O C12H8Cl2 C14H8Cl2 C10H10Cl2O3 C28H31ClN2O3 C6H5BrO C12H8Br2 C12H8Br2O C14H8Br2 C16H31BrO2

operated with a Model C665 dc power supply (both Hamamatsu, Shizuoka-ken, Japan). Signals from the photomultiplier tube were recorded with an SR6335 strip chart recorder from Graphtec (Yokohama, Japan) or with an HP3395 A integrator (HewlettPackard, Avondale PA). The emission signals were measured at 837.6 nm for chlorine, 827.2 nm for bromine, and 940.5 nm for carbon detection. The plasma gas used was 99.9999% helium (Hydro, Oslo, Norway) at a flow rate of 200 mL/min. Traces of 99.999% oxygen (AGA, Oslo, Norway) were utilized as dopant gas, and pressurized air (AGA) was used to cool the probe before each sample application. Tetrahydrofuran (Fisons, Loughborough, England), cyclohexane (Rathburn Chemicals Limited, Walkerburn, Scotland), ethyl acetate (Rathburn Chemicals), and pentane (Kebo Lab, Oslo, Norway), used as solvents, were all of HPLC or p.a. grade. The chlorinated and brominated model compounds used (Table 1) were commercially available with reported purities exceeding 99%. The soil sample was extracted with cyclohexane, while the crude oil was diluted with the same solvent. Both samples were spiked with a mixture of dichlorophenol, dichloroanthracene, and 4-(2,4dichlorophenoxy)butyric acid, 12.1, 17.1, and 17.6 ng/µL, respectively, for the oil and 19.4, 22.8, and 17.8 ng/µL, respectively, for the soil extract. Calculations of Detection Limit, Dynamic Range, and Selectivity. The detection limit is defined as the amount of element required to produce a peak 3 times the standard deviation of the background, while the dynamic range is defined as the range of sample concentration over which the response factor (area per unit mass) varies by less than 10%. The selectivities for the halogens with respect to carbon are defined as the ratio of the peak response per gram of halogen to the peak response per gram of carbon. RESULTS AND DISCUSSION Basic Idea and Instrumental Considerations. The idea was to utilize the elemental selectivity, the high sensitivity, and the compound-independent nature of plasma atomic emission spectrometry for the determination of extractable organic chlorine and bromine (EOCl and EOBr). Because organic solvents inherently quench the excitation ability of plasma systems, it was decided to evaporate the solvent prior to analysis and to introduce the sample into the plasma in the solid state. Evaporation of the solvent before measurement was straightforward because most of the organic solvents utilized for the determination of EOCl and 3560

Analytical Chemistry, Vol. 69, No. 17, September 1, 1997

EOBr are quite volatile and evaporate quickly at room temperature. Thus, although some losses of highly volatile components may occur, solvent evaporation prior to analysis was expected to dramatically improve the detection limit of the method. The probe injection dual-microplasma spectrometer is illustrated in Figure 1. The central part of the system comprised a probe for sample introduction, and two microplasmas were sustained in the same stream of helium: one plasma for heating the sample (vaporizing plasma) and the second for the optical measurements (analytical plasma). The sample was applied with a microsyringe into the drilled 2 µL sample cup on the steel probe. Subsequently, the solvent was gently evaporated at room temperature, and the sample cup was pushed into the interplasma region. The vaporizing plasma was in direct contact with the steel probe and served to vaporize the sample material into the stream of helium. Since the heat conductivity of the probe was high, this evaporation occurred immediately. Because the helium stream was extensively heated by the vaporizing plasma, the vaporized sample was effectively transferred to the analytical plasma, where atomic emission was measured for the determination of EOCl and EOBr. Although the position of the sample cup changed from sample application to injection, the sample probe was always present perpendicular to the helium stream in the interplasma region. As the sample was introduced very close to the analytical plasma, the discharge was not in direct contact with the moving steel probe. This prevented destabilization of the analytical plasma during horizontal movements of the sample probe, and minor shifts in the baseline were observed only at very high sensitivity settings. Two rf power supplies were used to generate the two microplasmas. Although this increased the price of the present setup, individual optimization and control was a major advantage during the present preliminary experiments. Sample Application. The sample volume used in this work was 1 µL, which was manually injected with a microsyringe into the small sample cup without difficulties. Since the volume of the sample cup was approximately 2 µL, injection volumes up to 2 µL were possible in order to enhance elemental responses. In cases of trace analysis, where injection volumes exceeding 2 µL may be necessary, several 1-2 µL applications with solvent evaporation in between enhanced the concentration detection limit of the method. Plasma Operation and Optimization. Prior to application in the small sample cup, the sample probe was cooled in a stream of pressurized air for 40 s in order to avoid losses of volatile components. After sample application, the organic solvent was allowed to evaporate at room temperature for 25-40 s. The exact evaporation time depended on the volatility of the solvent used and was experimentally determined by monitoring the carbon response following injection of pure solvent. With the simple cooling procedure employed, which was found to be acceptable for this preliminary study, it was necessary to turn off both discharges during sample application and solvent evaporation. The heat from the plasma region would otherwise increase the temperature of the probe/sample cup, with subsequent losses of volatile sample constituents. Since the atomization/excitation conditions of the analytical plasma stabilized a few seconds after ignition, this discharge was ignited only 10 s prior to injection.

Figure 2. Effect of plasma gas flow rate on responses for (A) chlorine and (B) bromine. The results were based on three repetitive applications of 50 ng of halogen. The model compounds used were 4,4′-dichlorobiphenyl (9), 4-(2,4-dichlorophenoxy)butyric acid (b), 4,4′dibromobiphenyl (2), and 2-bromohexanoic acid (1).

The vaporizing plasma was ignited immediately after the introduction of the sample cup. Although the temperature of the sample probe was low during introduction, it increased rapidly after ignition because the tail of the vaporizing plasma was in direct contact with the steel probe. Consequently, signals arose in the analytical plasma immediately after ignition of the vaporizing plasma. Both discharges were turned off after each measurement, and the cooling procedure discussed above was repeated prior to a new analysis. The time between samples was approximately 2 min. For the optimization of helium flow and power level for the two plasmas, model compounds were carefully selected to cover a broad range regarding both volatility and polarity. For the optimization studies in the chlorine-selective mode (EOCl), dichlorobiphenyl was selected as a nonpolar compound of relatively high volatility, while (dichlorophenoxy)butyric acid was chosen as a model for polar components of low volatility. For similar reasons, dibromobiphenyl and monobromohexadecanoic acid were selected for the EOBr experiments. Tetrahydrofuran was utilized as solvent, and the model compounds were present at the 50 ng of halogen/µL level. As shown in Figure 2, the helium flow rate significantly affected signals in both the chlorine- and bromine-selective modes. For the volatile model compounds, the signal decreased continuously as the flow of helium was increased from 100 to 600 mL/min.

Figure 3. Effect of power level in the analytical plasma on responses for (A) chlorine and (B) bromine. The results were based on three repetitive applications of 50 ng of halogen. The model compounds used were 4,4′-dichlorobiphenyl (9), 4-(2,4-dichlorophenoxy)butyric acid (b), 4,4′-dibromobiphenyl (2), and 2-bromohexanoic acid (1).

This effect arose from the dilution of the discharge with helium and from a decrease in the residence time of the analytes. For the nonvolatile and polar components, however, a sensitivity enhancement was observed when the helium flow was increased from 100 to 200-300 mL/min. This effect probably arose as a result of evaporation and transport problems at low flow rates. As the helium flow was increased above 200-300 mL/ min, signals for the nonvolatile components decreased as for the volatile ones. Based on the results in Figure 2, 200 mL/min was utilized to sustain the two discharges throughout the present study. In addition to the helium flow rate, the energy load was optimized for both the analytical plasma and the vaporizing plasma. As illustrated in Figure 3, the signals measured in the analytical plasma were only slightly affected as the power (energy reaching the plasma) was varied from 37 to 65 W. Because the most efficient energy transfer was obtained at moderate power levels, the analytical plasma was sustained at 50 W. In a subsequent experiment, signals were measured from a 50 W analytical plasma while the energy load of the vaporizing plasma was varied from 45 to 95 W. As illustrated in Figure 4, the signals from the volatile model compounds were almost unaffected by the power level of the vaporizing plasma. Obviously, even at low power settings, Analytical Chemistry, Vol. 69, No. 17, September 1, 1997

3561

Figure 5. Signals obtained for the different chlorinated compounds listed in Table 1. All the compounds were applied at the 20 ng of chlorine level.

Figure 4. Effect of power level in the vaporizing plasma on responses for (A) chlorine and (B) bromine. The results were based on three repetitive applications of 50 ng of halogen. The model compounds used were 4,4′-dichlorobiphenyl (9), 4-(2,4-dichlorophenoxy)butyric acid (b), 4,4′-dibromobiphenyl (2), and 2-bromohexanoic acid (1).

these compounds were effectively transferred to the analytical plasma. However, for nonvolatile sample constituents, signals decreased as the vaporizing plasma was operated at low power levels. This effect arose from insufficient heating of the sample probe. Therefore, to effectively cover a broad range of compounds while minimizing the amount of reflected power, the vaporizing plasma was operated at 75 W throughout the present study. Dopant Gas. Traces of O2 were added to the helium plasma gas in order to prevent formation and deposition of elemental carbon within the small plasma tubes. In addition to this important effect, the O2 dopant served to suppress spectral interferences caused by changes in the carbon continuum. This was advantageous in the present work because of the low optical resolution of the monochromator. Because only traces of O2 were required for effective plasma doping, the exact flow rate of O2 was unknown. However, the oxygen level was easily controlled by the intensity ratio between the oxygen line at 777.2 nm and the helium line at 706.5 nm. In the present study, IO/IHe ) 3-4 was found to be optimal. Solvent Considerations. Considering the diversity of environmental samples, a variety of different extraction solvents have 3562 Analytical Chemistry, Vol. 69, No. 17, September 1, 1997

been reported for the determination of EOCl and EOBr.7,38 Because the extraction solvent was evaporated prior to the AES analysis, the analytical responses were found to be unaffected by the solvent used. Thus, as long as volatile solvents are used, the type of solvent may be selected to optimize extraction recoveries without emphasis on the final AES measurement. Response Factors. The potential of plasma AES for the determination of EOCl and EOBr relies on compound-independent emission signals. Thus, 10 ng of chlorine or bromine should give the same signal, regardless of the compound or sample composition. To verify this experimentally, the responses from several different halogenated compounds dissolved in tetrahydrofuran at the 20 ng of halogen/µL level were investigated. As illustrated in Table 1, the model components selected differed substantially in terms of both polarity and volatility in order to reflect the chemical diversity of environmental samples. Because of these differences, peak shapes differed significantly, as illustrated for the chlorinated compounds in Figure 5. Generally, volatile and nonpolar compounds were rapidly transferred to the analytical plasma, while the vaporization time increased for the polar and high-boiling components. Thus, high and narrow peaks were obtained for the volatile compounds, while broader signals were observed for components of high molecular weight. Fortunately, although peak shapes differed significantly, the corresponding areas varied within 21% for the chlorinated compounds (Figure 6A) and 25% for the brominated components (Figure 6B). Relevant studies for comparison are scarce in the literature, but the response variations observed in the present work were in accordance with similar studies on organic compounds performed with gas chromatographic introduction to different plasma systems.37,39-42 The excellent response of Rhodamine B supported the fact that the method is applicable also for chlorinated (38) Martinsen, K.; Kringstad, A.; Carlberg, G. E. Water Sci. Technol. 1988, 20, 13-24. (39) Uchida, H.; Berthod, A.; Winefordner, J. D. Analyst 1990, 115, 933-937. (40) Pedersen-Bjergaard, S.; Asp, T. N.; Greibrokk, T. Anal. Chim. Acta 1992, 265, 87-92. (41) Kovacic, N.; Rasmus, T. L. J. Anal. At. Spectrom. 1992, 7, 999-1005. (42) Pedersen-Bjergaard, S.; Semb, S. I.; Brevik, E. M.; Greibrokk, T. J. Chromatogr. A 1996, 723, 337-347.

Figure 6. Area responses for (A) different chlorinated compounds and (B) different brominated compounds. The results were based on five repetitive applications of 20 ng of halogen. The test compounds used are listed in Table 1.

compounds not amenable to gas chromatography. Unfortunately, highly volatile compounds like dichloro- and dibromobenzene were not detected with the dual-plasma system. For these compounds, the cooling of the probe to room temperature was insufficient, and the solutes evaporated before the sample cup was pushed into the plasma region. Although this was a disadvantage, most highly volatile compounds present in environmental samples probably evaporate during sampling and extraction. Thus, the discrimination observed with the present detection concept represented only a minor problem for practical work. However, work is currently in progress to expand the application field also to highly volatile compounds by further cooling of the sample probe. Detection Limits and Linearity. For both chlorine and bromine, detection limits at the 120 pg level were obtained with 1 µL injections utilizing dichloro- and dibromobiphenyl as model compounds. For the EOCl methods based on microcoulometry, neutron activation, and inductively coupled plasma mass spectrometry, corresponding values between 5 and 50 ng1,2,10,38,43 have been reported for chlorine. With electrothermal vaporization MIP as a relevant and alternative atomic spectroscopic technique, the halogens have been detected from the 180-1000 pg level (3σ),19,22 but direct comparison with the present work is difficult because most ETV-MIP work has focused on inorganic halogens. With organic compounds, most work of relevance has been performed with gas chromatographic introduction, where chlorine and bromine have been detected down to the 50 pg level.24,44,45 Thus, the present approach was highly sensitive for the determination of EOCl and, consequently, required only small amounts of extract to achieve low detection limits. In addition, the responses for both chlorine and bromine were linear within 3 orders of magnitude, which simplified quantification. Repeatability. In Figure 7, five repetitive measurements of 20 ng of Cl from both dichlorobiphenyl and dichloroanthracene are illustrated. The RSDs for the area responses were 4.2 and 5.7%, respectively, which were considered acceptable for manual injections. Applications. Although the spectrometer consisted of a single, low-resolution monochromator without spectral background correction, 1000:1 selectivities relative to carbon were (43) Riehl, E.; Gandee, L. A. Am. Lab. 1990, 22, 29-30, 32-33. (44) Huang, D.; Blades, M. W. Appl. Spectrosc. 1991, 45, 1468-1477. (45) Quimby, B. D.; Sullivan, J. J. Anal. Chem. 1990, 62, 1027-1034.

Figure 7. Five applications of 20 ng of Cl from dichlorobiphenyl and dichloroanthracene. The RSDs for the area response were 4.2 and 5.7%, respectively.

obtained for both chlorine and bromine. For relatively simple samples and standard solutions, this selectivity was sufficient to avoid interferences from carbon. For applications where large amounts of carbon were present in addition to the halogens of interest, interfering signals were observed on the halogen lines due to elevation of the carbon continuum. Thus, background correction was necessary to correct the quantitative results. This was accomplished for each sample by an initial measurement of the carbon content by monitoring the carbon channel and utilizing biphenyl as reference. Based on highly accurate halogen-tocarbon selectivities determined in advance, the interference signal on the halogen channel was easily calculated from the estimated carbon content for each sample. With this type of simple background correction, quantitative data of high quality were obtained, even for very complicated samples, with the probe injection dual-microplasma spectrometer. This was confirmed by two experiments where extracts of crude oil (1000 ng C/µL) and soil (200 ng C/µL) were spiked with dichlorophenol, dichloroanthracene, and 4-(2,4-dichlorophenoxy)butyric acid at the 15-20 ng/µL level of total chlorine. The experimental error using dichloroanthracene as reference was, in both cases, below 5%.

CONCLUSIONS The present work has demonstrated a new concept for the determination of extractable organic chlorine and bromine (EOCl and EOBr) based on atomic emission spectrometry. Because of the simplicity and miniaturized nature of the probe injection dualmicroplasma spectrometer, both the instrumental costs and the operating costs were relatively low. The equipment was capable of differentiating between chlorine and bromine, and with a simple change of wavelength, other elements may be measured as well. Compared with those of existing techniques like microcoulometric titration, neutron activation, and inductively coupled plasma mass spectrometry, detection limits were significantly lower in the present work. Thus, probe injection dual-microplasma spectrometry should be an attractive alternative for the determination of EOCl and EOBr. Analytical Chemistry, Vol. 69, No. 17, September 1, 1997

3563

Although the results reported above were very promising, work is in progress to further develop the probe injection dualmicroplasma spectrometer. To improve the capability for highly volatile compounds, improved cooling of the sample probe is planned, together with experiments utilizing different probe materials. In addition, replacement of the cross union in steel with a material of lower activity is considered in order to avoid possible interactions between sample constituents and the hot steel surface. While two plasma generators were used in the present work to enable individual optimization of the two plasmas,

3564

Analytical Chemistry, Vol. 69, No. 17, September 1, 1997

work is currently in progress to develop a single rf source capable of powering both discharges. This may significantly reduce the instrumental cost.

Received for review December 3, 1996. Accepted May 6, 1997.X AC961219S X

Abstract published in Advance ACS Abstracts, July 1, 1997.