Solid-Phase Microextraction: Determination of

Russian Chemical Reviews 2015 84, 653-664 ... Determination of aniline derivatives in water samples after preconcentration with oxidized multiwalled c...
0 downloads 0 Views 147KB Size
Anal. Chem. 2004, 76, 1028-1038

In Situ Derivatization/Solid-Phase Microextraction: Determination of Polar Aromatic Amines Thomas Zimmermann, Wolfgang J. Ensinger, and Torsten C. Schmidt*

Department of Chemistry, Philipps-University Marburg, Hans-Meerwein-Strasse, 35032 Marburg, Germany

A solid-phase microextraction GC/MS method for the trace determination of a wide variety of polar aromatic amines in aqueous samples was developed. Prior to extraction the analytes were derivatized directly in the aqueous solution by diazotation and subsequent iodination in a one-pot reaction. The derivatives were extracted by direct-SPME using a PDMS/DVB fiber and analyzed by GC/MS in the full-scan mode. By diazotation/iodination, the polarity of the analytes was significantly decreased and as a consequence extraction yields were dramatically improved. The derivatization proved to be suitable for strongly deactivated aromatic amines and even the very polar diamino compounds can efficiently be enriched after derivatization. We investigated 18 anilines comprising a wide range of functional groups, which could be determined simultaneously. The method was thoroughly validated, and the precision at a concentration of 0.5 µg/L was 3.8-11% relative standard deviation for nonnitrated analytes using aniline-d5 as internal standard and 3.7-10% for nitroaromatic amines without internal standard. The in situ derivatization/SPME/GC/MS method was calibrated over the whole analytical procedure and was linear over 2 orders of magnitude. Using 10-mL samples, detection limits of 2-13 ng/L were achieved for 15 of the 18 analytes. For two aminodinitrotoluene isomers and a diaminonitrotoluene, detection limits ranged from 27 to 38 ng/L. By allowing quantification at the 0.1 µg/L level, analysis of all target compounds meets EU drinking water regulations. The method provides high sensitivity, robustness, and high sample throughput by automation. Finally, the method was applied to various real water samples and in wastewater from a former ammunition plant the contents of several aromatic amines were quantified. Aromatic amines are widespread chemicals in several industries. They are used in the manufacture of rubber chemicals, pesticides, dyes, pharmaceuticals, and photographic chemicals.1 Their major use, however, is in the production of rigid polyure* Corresponding author: (phone) + 49-7071-29-7 31 47; (fax) + 49-7071-2951 39; (e-mail) [email protected]. Present address: Center for Applied Geosciences, Eberhard-Karls-University Tuebingen, Wilhelmstr. 56, D-72074 Tuebingen. (1) Vogt, P. F.; Geralis, J. J. In Ullmann’s Encyclopedia of Industrial Chemistry, 5th ed.; Gerhartz, W., Yamamoto, Y. S., Campbell, F. T., Pfefferkorn, R., Rounsaville, J. F., Eds.; VCH: Weinheim, 1985; Vol. A2, p 35.

1028 Analytical Chemistry, Vol. 76, No. 4, February 15, 2004

thanes and reaction-injection-molded parts for the construction, automotive, and durable goods industries. During production, use, and disposal of these goods, emissions of aromatic amines may occur. Of equal importance is the formation of aromatic amines in the environment due to degradation of precursors, e.g., by microbially mediated reduction of nitroaromatic compounds (NACs).2,3 NACs are among the most widely used anthropogenic chemicals, and according to the OECD, ∼70 NACs currently are high-production-volume chemicals with a production of more than 1000 t per year in at least one country. Besides contamination due to their use, NACs are formed and released during incomplete combustion processes.4-6 Furthermore, aromatic amines are released during hydrolysis of azo dyes7 and pesticides.8 Up to now, more than 30 aromatic amines have been identified in the environment as metabolites of anilides, carbamates, nitrophenols, or phenylurea pesticides.8-10 The global annual emission of 4-chloroaniline alonesa compound that has been classified as possibly carcinogenic to humanssis estimated at 100010 000 t.11 The toxicological properties of arylamines are mainly characterized by their ability to form DNA adducts. Currently, the International Agency for Research on Cancer (IARC) has classified six aromatic amines as carcinogenic or probably carcinogenic to humans12 (IARC list 1 and 2A), but several other anilines also have been found carcinogenic in animal experiments.13 Many aromatic amines cause damage to DNA and reacted positive in mutagenicity tests.13 As a consequence, these substances are suspected to be harmful to humans and need to be monitored regularly. (2) Spain, J. C., Ed. Biodegradation of Nitroaromatic Compounds; Plenum Press: New York, 1995. (3) Larson, R. A.; Weber, E. J. Reaction Mechanisms in Environmental Organic Chemistry; Lewis: Boca Raton, FL, 1994. (4) Gibson, T. L. Mutat. Res. 1983, 121, 115-121. (5) Tokiwa, H.; Ohnishi, Y. Crit. Rev. Toxicol. 1986, 17, 23-69. (6) Zwirner-Baier, I.; Neumann, H. G. Mutat. Res. 1999, 441, 135-144. (7) Clarke, E. A.; Anliker, R. Organic Dyes and Pigments. In Handbook of Environmental Chemistry; Hutzinger, O., Ed.; Springer: Heidelberg, 1980; Vol. 3A. (8) Domsch, K. H. Pestizide im Boden; VCH: Weinheim, 1992. (9) Dorfler, U.; Scheunert, I. Verbleib von Pflanzenchutzmitteln in der Umwelt; Umweltbundesamt: Berlin, 1989. (10) Barcelo, D.; Hennion, M. C. Trace determination of pesticides and their degradation products in water; Elsevier: Amsterdam, 1997. (11) Rippen, G. Handbuch der Umweltchemikalien; Loseblattsammlung, ecomed: Landsberg, 1990. (12) See: http://www.iarc.fr. (13) Gold, L. S.; Zeiger, E. Handbook of carcinogenic potency and genotoxicity databases; CRC Press: Boca Raton, 1997. 10.1021/ac035098p CCC: $27.50

© 2004 American Chemical Society Published on Web 01/15/2004

Analysis of aromatic amines is often performed by classical reversed-phase HPLC with UV detection.14-16 Using an electrochemical14,17 or mass spectrometric detector,18 sensitivity has been increased. Nevertheless, HPLC is hampered by its low peak capacity that does not allow complete separation of the large number of analytes that require monitoring. Although GC separations of halogenated and alkylated anilines have been reported,19-22 especially less volatile arylamines require derivatization prior to analysis.23 The most important class of derivatizing reagents are carboxylic acid anhydrides, and by using perfluorated reagents in combination with ECD or NCI/MS, a highly sensitive detection can be realized.24-26 To achieve low detection limits, an enrichment step prior to analysis is essential. Typically this is done by liquid-liquid extraction (LLE)14,19,21,27 or solid-phase extraction (SPE).14,15,16,18,19,22,28-31 There are several applications using LLE with dichloromethane14,19,27 or SPE with GCB,18 SCX,29,30 C18,16,19,27 or PS/ DVB14,15,19,22,28,31 cartridges. The methods have shown to be capable for several alkylated, halogenated, and nitrated aromatic amines and diamines. On the other hand, problems have been mentioned using SCX cartridges due to large amounts of inorganic cations30 as well as losses during evaporation.21 LLE requires large amounts of toxic organic solvents, and phase separation is hampered by the formation of emulsions. Both types of extraction require large sample volumes and therefore sampling, transport, storage, and sample preparation are elaborate. The mentioned problems can be overcome by miniaturized techniques such as mLLE,20 on-line SPE-HPLC,15,17 and solid-phase microextraction (SPME).32-36 While classical methods use only a rather small amount of the eluate, these techniques transfer the (14) Lewin, U.; Wennrich, L.; Efer, J.; Engewald, W. Chromatographia 1997, 45, 91-98. (15) Kruppa, J.; Preiss, A.; Levsen, K.; Kabus, H. P. Acta Hydrochim. Hydrobiol. 1996, 24, 226-231. (16) Zhao, S. L.; Wei, F. S.; Zou, H. F.; Xu, X. B. Chemosphere 1998, 36, 73-78. (17) Patsias, J.; Papadopoulou-Mourkidou, E. J. Chromatogr., A 2000, 904, 171188. (18) Di Corcia, A.; Costantino, A.; Crescenzi, C.; Samperi, R. J. Chromatogr., A 1999, 852, 465-474. (19) Oostdyk, T. S.; Grob, R. L.; Snyder, J. L.; McNally, M. E. J. Environ. Sci. Health 1994, A29, 1607-1628. (20) Vreuls, R. J. J.; Romijn, E.; Brinkman, U. A. T. J. Microcolumn Sep. 1998, 10, 581-588. (21) Tekel, J.; Schultzova, K.; Kovacicova, J. J. High Resolut. Chromatogr. 1993, 16, 126-128. (22) Lacorte, S.; Guiffard, I.; Fraise, D.; Barcelo, D. Anal. Chem. 2000, 72, 14301440. (23) Kataoka, H. J. Chromatogr., A 1996, 733, 19-34. (24) Maurino, V.; Minero, C.; Pelizzetti, E.; Angelino, S.; Vincenti, M. J. Am. Soc. Mass Spectrom. 1999, 10, 1328-1336. (25) Onuska, F. I.; Terry, K. A.; Maguire, R. J. Water Qual. Res. J. Can. 2000, 35, 245-261. (26) Longo, M.; Cavallaro, A. J. Chromatogr., A 1996, 753, 91-100. (27) Levsen, K.; Mussmann, P.; Berger-Preiss, E.; Preiss, A.; Volmer, D.; Wu¨nsch, G. Acta Hydrochim. Hydrobiol. 1993, 21, 153-166. (28) Brede, C.; Skjevrak, I.; Herikstad, H. J. Chromatogr., A 2003, 983, 35-42. (29) Jeevan, R. J. G.; Bhaskar, M.; Chandrasekar, R.; Radhakrishnan, G. Electrophoresis 2002, 23, 584-590. (30) Nielen, M. W. F.; Frei, R. W.; Brinkman, U. A. T. J. Chromatogr. 1984, 317, 557-567. (31) Less, M.; Schmidt, T. C.; von Lo ¨w, E.; Stork, G. J. Chromatogr., A 1998, 810, 173-182. (32) Mu ¨ ller, L.; Fattore, E.; Benfenati, E. J. Chromatogr., A 1997, 791, 221230. (33) Barshick, S. A.; Griest, W. H. Anal. Chem. 1998, 70, 3015-3020. (34) Wu, Y. C.; Huang, S. D. Anal. Chem. 1999, 71, 310-318.

extracted amount quantitatively to the analytical column. Therefore, they are more efficient, minimizing sample preparation. The risk of losses and contamination can be reduced and small samples can be analyzed as well. These methods can be easily automated providing high sample throughput. Unfortunately, the extraction of polar compounds is difficult,17,20,22 and most applications for SPME deal with the less polar alkylated and halogenated arylamines.32,35,36 Best results were obtained with chlorinated anilines,32,36 but for alkylated or nitrated compounds, the detection limits increased substantially.33-35 As can be seen from Mu¨ller et al., the extraction efficiency from headspace is rather low32 and therefore analytes are typically extracted by direct-SPME following analysis by gas chromatography. Determination by SPME-HPLC is limited to one publication due to well-known problems such as peak broadening and swelling of the SPME fiber in organic solvents.34 Up to now, there is no example for in situ derivatization/SPME for aromatic amines. This methodology was introduced by Pan and Pawliszyn in 1997,37 but publications dealing with derivatization in aqueous solution are limited to the determination of carboxylic acids38-40 or amphetamines41,42 focusing on bioanalytical applications. Benzyl bromide,38,39 chloroformates,38,41,40 and benzoyl chlorides42 were used for derivatization, but problems due to hydrolysis of the reagent,38,40 large reagent excess,38 and low sensitivity40 were reported. The aim of our study was therefore to develop an easy and robust derivatization that is carried out directly in the aqueous solution. In previous studies of our group, the diazotation with subsequent iodination was carried out after SPE prior to analysis by GC-ECD.43,47 However, sample preparation was rather time-consuming, and limits of detection were only in the lowmicrogram per liter range. In this paper, we present substantially improved limits of detection by derivatization prior to extraction. Lowering the polarity of the analytes using in situ derivatization and subsequent SPME in combination with GC/MS enables detection limits in the low-nanogram per liter range. To the best of our knowledge, this is the first in situ derivatization/SPME method for aqueous solutions dealing with a large number of analytes covering a broad range of chemical properties. (35) van Doorn, H.; Grabanski, C. B.; Miller, D. J.; Hawthorne, S. B. J. Chromatogr., A 1998, 829, 223-233. (36) Paul, S.; Lienig, D.; Worch, E. Vom Wasser 1997, 88, 273-283. (37) Pan, L.; Pawliszyn, J. Anal. Chem. 1997, 69, 196-205. (38) Wittmann, G.; Van Langenhove, H.; Dewulf, J. J. Chromatogr., A 2000, 874, 225-234. (39) Nilsson, T.; Baglio, D.; Galdo-Miguez, I.; Madsen, J. O.; Facchetti, S. J. Chromatogr., A 1998, 826, 211-216. (40) Henriksen, T.; Svensmark, B.; Lindhardt, B.; Juhler, R. K. Chemosphere 2001, 44, 1531-1539. (41) Ugland, H. G.; Krogh, M.; Rasmussen, K. E. J. Chromatogr., B 1997, 701, 29-38. (42) Koster, E. H. M.; Bruins, C. H. P.; Wemes, C.; de Jong, G. J. J. Sep. Sci. 2001, 24, 116-122. (43) Haas, R.; Schmidt, T. C.; Steinbach, K.; von Lo¨w, E. Fresenius J. Anal. Chem. 1997, 359, 497-501. (44) Dugay, J.; Miege, C.; Hennion, M. C. J. Chromatogr., A 1998, 795, 27-42. (45) Yang, Y.; Miller, D. J.; Hawthorne, S. B. J. Chromatogr., A 1998, 800, 257266. (46) Eisert, R.; Levsen, K. Fresenius J. Anal. Chem. 1995, 351, 555-562. (47) Schmidt, T. C.; Less, M.; Haas, R.; von Lo ¨w, E.; Steinbach, K.; Stork, G. J. Chromatogr., A 1998, 810, 161-172.

Analytical Chemistry, Vol. 76, No. 4, February 15, 2004

1029

Table 1. Investigated Compounds with Abbreviations, Their Iodinated Derivatives, Retention Times tR, Nominal Mass (MW), and Masses Used for Detection compound

abbrev

derivative

tR (min)

MW (m/z)

detected m/za

aniline 4-aminotoluene 4-chloroaniline 3,4-dichloroaniline 2,4,5-trichloroaniline 3-chloro-4-fluoroaniline 2,4-diaminotoluene 2,6-diaminotoluene 2-naphthylamine 2-aminobiphenyl 4-aminobiphenyl 4-nitroaniline 2-amino-4-nitrotoluene 2-amino-6-nitrotoluene 4-amino-2-nitrotoluene 2-amino-4,6-dinitrotoluene 4-amino-2,6-dinitrotoluene 2,4-diamino-6-nitrotoluene aniline-d5

A 4AT 4CA 3,4DCA 2,4,5TCA 3C4FA 2,4DAT 2,6DAT 2NaA 2ABP 4ABP 4NA 2A4NT 2A6NT 4A2NT 2A4,6DNT 4A2,6DNT 2,4DA6NT A-d5

iodobenzene 4-iodotoluene 1-chloro-4-iodobenzene 1,2-dichloro-4-iodobenzene 1,2,4-trichloro-5-iodobenzene 2-chloro-4-iodo-1-fluorobenzene 2,4-diiodotoluene 2,6-diiodotoluene 2-iodonaphthaline 2-iodobiphenyl 4-iodobiphenyl 1-iodo-4-nitrobenzene 2-iodo-4-nitrotoluene 2-iodo-6-nitrotoluene 4-iodo-2-nitrotoluene 2-iodo-4,6-dinitrotoluene 4-iodo-2,6-dinitrotoluene 2,4-diiodo-6-nitrotoluene iodobenzene-d5

8.36 9.54 9.98 11.55 13.41 10.04 13.22 13.31 14.59 15.20 16.98 13.22 15.69 14.05 14.44 18.49 17.76 18.00 8.35

204 218 238 272 306 256 344 344 254 280 280 249 263 263 263 308 308 389 209

204, 77, 127 218, 91, 65 238, 111, 75 272, 145, 109 306, 179, 143 256, 129, 109 344, 217, 90 344, 217, 90 254, 127, 74 280, 152, 127 280, 152, 127 249, 219, 203 263, 90, 105 246, 89, 119 246, 89, 119 291, 164, 89 291, 89, 63 372, 344, 216 209, 82, 127

a

Base peak used for quantification is underlined.

EXPERIMENTAL SECTION Chemicals and Reagents. Reference substances and the internal standard aniline-d5 were obtained from various suppliers: Aldrich (Steinheim, Germany), Fluka (Neu-Ulm, Germany), Merck (Darmstadt, Germany), Promochem (Wesel, Germany), Mallinckrodt-Baker (Griesheim, Germany), and Riedel-de-Hae¨n (Seelze, Germany) in the highest purity available. The compounds and abbreviations used throughout the text as well as the corresponding derivatives, their molecular weights, and typical fragment ions are given in Table 1. Sodium nitrite was purchased from Riedel-de-Hae¨n, hydroiodic acid (ACS reagent, unstabilized, 55%), and amidosulfonic acid were from Aldrich, and ethyl acetate (HPLC grade), methanol (HPLC gradient grade), sodium sulfite, disodium hydrogen phosphate, potassium dihydrogen phosphate, and sodium hydroxide were from Merck, all in the highest purity available. Nanopure water was generated with a Direct-Q laboratory water purification system from Millipore (Schwalbach, Germany). Standards. Stock solutions of each analyte were prepared in methanol by weighing ∼10 mg of analyte into a 10-mL volumetric flask and diluting to volume containing the analytes at a concentration of 1 g/L. Aliquots of each stock solution were mixed and diluted with methanol to concentrations of 10, 50, 100, 500, 1000, and 5000 µg/L of each component. These solutions were further diluted by a factor of 10 with Nanopure water to obtain spike solutions in water/methanol 9/1 (v/v). Separate stock and spike solutions of the internal standard aniline-d5 were prepared. The concentrations were the same as mentioned above. The spike solutions were stored at -20 °C and used for a period of one month. Water Samples. Nanopure water was used for validation and optimization experiments. The 10-mL aliquots of the water sample in 25-mL screw-cap vials were spiked with 100 µL of the internal standard (IS; concentration 50 µg/L) to obtain an IS concentration of 0.5 µg/L in the sample. A total of 100 µL of the spike solution (concentration 50 µg/L) was added. The content of organic solvent in the sample was kept as low as possible to minimize possible 1030

Analytical Chemistry, Vol. 76, No. 4, February 15, 2004

side reactions. Furthermore, the spike volume was kept small and constant since the composition may influence partitioning of the analytes. Thus, during calibration, the spike volume was 100 µL while the concentration of the spike solution was varied between 50 and 5000 µg/L. These samples were stored at 4 °C until derivatization. Real water samples were taken from the drain of a wastewater disposal site (P 2, pH 5.7, conductivity 282 µS/cm, pale yellow) and a well (ASB 3, pH 5.9, conductivity 153 µS/cm) at the former ammunition plant Stadtallendorf, Hessen, on February 6, 2002. The surface water sample was withdrawn from the river Ohm near Marburg, Hessen, on May, 7, 2001. The drinking water sample was taken from the water conduit in Marburg on January 26, 2002. The samples were stored at 4 °C in brown glass bottles and analyzed within 4 weeks. Samples were filtered through membrane filters of 0.45-µm pore size (Sartorius, Go¨ttingen, Germany) before derivatization as depicted below. Derivatization. A 10-mL sample of water was acidified with 0.2 mL of hydroiodic acid. The solution was mixed with 0.5 mL of sodium nitrite in water (concentration 10 g/L) and shaken. After a reaction time of 20 min, 1 mL of amidosulfonic acid in water (concentration 50 g/L) was added to destroy the surplus of nitrite and the mixture was vigorously shaken for 45 min. The solution was heated for 5 min in a water bath at 100 °C and afterward cooled to room temperature. The surplus of iodine was destroyed with 0.25 mL of a saturated aqueous solution of sodium sulfite. The pH was adjusted by addition of 0.25 mL of a 0.25 mol/L solution of dipotassium hydrogen phosphate and 0.4 mL of a 5 mol/L solution of sodium hydroxide to a final pH value of ∼8. The mixtures were completely filled into 13-mL crimp-top vials without any headspace above the water sample. The vials were sealed with aluminum foil and stored at 4 °C until extraction with SPME. For optimization of chromatographic conditions, a sample containing all analytes and the internal standard aniline-d5 at a concentration of 1 mg/L was derivatized as described above. Subsequently, 2 mL of ethyl acetate was added and the sample

was shaken for 5 min. After phase separation, an aliquot of the organic phase was transferred to an autosampler vial. The organic phase was stored in a 1.5-mL autosampler vial at -20 °C and showed sufficient stability over several months. Solid-Phase Microextraction Procedure. All extractions were carried out in 13-mL crimp-top vials sealed with aluminum foil to prevent distortion of the fiber during the automated extraction process. At the same time, contamination with phthalates was reduced. Prior to use, the fibers were conditioned according to the recommendations of the manufacturer. Various parameters influencing the SPME process were optimized: The extraction procedure was optimized with five different commercially available fibers: 100-µm poly(dimethylsiloxane) (PDMS), 65-µm poly(dimethylsiloxane)/divinylbenzene (PDMS/DVB), 75 µm carboxen/poly(dimethylsiloxane) (Car/PDMS), 70-µm Carbowax/divinylbenzene (CW/DVB), and 50/30-µm divinylbenzene/carboxen/poly(dimethylsiloxane) (DVB/Car/PDMS) (all from Supelco, Deisenhofen, Germany). Extraction times were varied between 5 and 90 min. In the optimized SPME procedure, a 65-µm PDMS/DVB fiber was immersed in the water sample for 30 min. The sample was mixed by vibration of the SPME fiber. Once the extraction was completed, the fiber was withdrawn back inside the fiber holder and injected into the GC. The fibers were thermally desorbed for 5 min in a Siltek deactivated SPME liner with 0.8-mm i.d. (Restek, Bad Homburg, Germany). Because of depletion of analytes, each sample was used for only one extraction. Typically, SPME was run in the “prep ahead” mode extracting the following sample while the GC program was running in order to minimize total analysis time. Instrumentation. The gas chromatographic system consisted of a gas chromatograph 3800 equipped with an autosampler 8200, a programmable 1079 split/splitless injector, and a Saturn 2000 ion trap mass spectrometer (all Varian, Darmstadt, Germany). The whole SPME procedure was automated using the additional SPME Kit III (Varian). The temperature of the injection block was initially held at 50 °C and increased to 250 °C within 1 min after insertion of the SPME fiber. After 3 min, the split ratio was set to 1:100. For optimization of the chromatographic conditions, 1 µL of the extract in ethyl acetate was injected. The parameters used are the same as described for solid-phase microextraction. Separation of the analytes was performed with a Siltek deactivated column, 30 m, 0.25-mm i.d., 0.25-µm df (Stx-CLPesticides from Restek). Additionally, two times 1.5 m of a Siltek deactivated fused-silica capillary, 0.25-mm i.d. (Restek) was used as retention gap and transfer line. Carrier gas was helium 5.0 (Messer, Griesheim, Germany) at a constant flow rate of 2 mL/ min. The following temperature program was used: 40 °C for 3 min, 15 °C/min to 130 °C, 30 °C/min to 160 °C, 160 °C for 5 min, 30 °C/min to 200 °C, and 20 °C/min to 250 °C. The ion trap mass spectrometer was operated in the full-scan mode with electron impact at 70 eV with a filament/multiplier delay time of 6 min. The scan range under observation was 60450 m/z with 0.35 s/scan. Because of low chemical noise, the emission current was raised from typical 15 to 40 µA and the multiplier offset was +100 V. Ionization times were set by automatic gain control. The temperature of the ion trap, the transfer line, and the manifold were 200, 280, and 55 °C, respectively.

Figure 1. Reaction scheme for the iodination of aromatic amines.

Optimization of the derivatization reaction was done by reversed-phase HPLC. The chromatographic system consisted of an M 480 pump with on-line degasser ERC-3315, autosampler GINA 50 and UV detector UVD 160S (all from Gynkotek, Germering, Germany). Analytes were separated on a PurospherRP18 column (125 × 3 mm, 5 µm, from Merck) with the following elution gradient: methanol/phosphate buffer (10 mM, pH 7) 30/ 70 (v/v) for 10 min, raised to methanol/buffer 80/20 (v/v) within 20 min, and kept for additional 15 min. The flow rate was set to 0.8 mL/min and the injection volume to 100 µL. Chromatograms were recorded and peak areas evaluated at detection wavelengths of 230 and 254 nm. The conversion rate during the derivatization reaction was determined by standard addition. RESULTS AND DISCUSSION Derivatization. Derivatization reactions are typically carried out in order to improve the chromatographic properties or the sensitivity. For that purpose, stability toward hydrolysis is of minor priority. The most frequently used reagents for the derivatization of aromatic amines are acylation reagents, e.g., carboxylic acid anhydrides, silylation reagents, chloroformates, dinitrophenylation reagents, carbonyl compounds forming Schiff bases as well as sulfonyl chlorides and phosphoryl chlorides.23 Numerous publications on analytical methods using these reagents are available. However, all of these reagents are more or less labile to hydrolysis and the exclusion of water is often a prerequisite. As the reagents are electrophiles reacting with the nucleophilic amines, there is always a competition with the likewise nucleophilic water. On the other hand, increasing the stability of the reagent leads to less hydrolysis but at the same time to a lower reactivity toward the aromatic amines. For this reason, up to now there is no publication dealing with the derivatization of deactivated aromatic amines in the presence of water. However, to increase extractability, the derivatization has to be carried out directly in the aqueous solution. Therefore, we chose a reaction that follows a different mechanism and is typically carried out in water: the diazotation of the amino group followed by subsequent substitution of the diazo group by iodine in a one-pot reaction (Figure 1). The derivatization is very simple, and the amines are quickly converted to aromatic iodine compounds. A total of 18 different aromatic amines with various functional groups were chosen as model compounds. The analytes are known degradation products of pesticides or nitroaromatic compounds. They cover a wide range of polarity as indicated in Table 2 by their corresponding log Kow values. The high reactivity is illustrated by the fact that after derivatization for all substances except 2A4,6DNT less than 5% of the arylamines remained. For 2A4,6DNT, only 77% were converted. These results were obtained using a mass balance approach by directly injecting the aqueous solution into HPLC (a) prior to derivatization, (b) after derivatization, and (c) after spiking the Analytical Chemistry, Vol. 76, No. 4, February 15, 2004

1031

Table 2. Octanol-Water Partition Constants of Anilines and Their Iodinated Derivatives compound

log Kow aromatic amine calca/exp

log Kow derivative calca/exp

aniline 4AT 4CA 3,4DCA 2,4,5TCA 3C4FA 2,4DAT 2,6DAT 2NaA 2ABP 4ABP 4NA 2A4NT 2A6NT 4A2NT 2A4,6DNT 4A2,6DNT 2,4DA6NT

0.92/0.90b 1.41/1.39b 1.91/1.91f 2.60/2.55g 3.35/3.69h 2.15/2.06i 0.14/0.09/2.28/2.28b 2.80/2.84b 2.80/2.88b 1.26/1.39e 1.71/1.87c 1.63/1.68/1.58/1.85d 1.55/2.10d 0.58/0.79d

3.27/3.25j 3.76/3.98/4.12g 4.57/5.28/4.12/4.59/4.29/4.44/4.65/5.15/3.01/3.21/3.13/3.43/2.87/3.09/4.25/-

a Calculated with clogP from Daylight Chemical Information Systems (http://www.daylight.com). b Reference 58. c Reference 59. d Reference 60. e Reference 61. f Reference 62. g Reference 63. h Reference 64. i Reference 65. j Reference 66.

derivatized solution with the arylamines for quantification. The experiments were carried out at a concentration of 10 mg/L to be able to detect traces of remaining anilines. The derivatization was carried out as mentioned above with sodium nitrite at a concentration of 0.5 g/L in the sample. As can be seen from Figure 2, in contrast to other derivatization reagents, at higher nitrite concentrations even the highly deactivated 2A4,6DNT can be derivatized almost quantitatively. Therefore, the remarkably high reactivity even in the presence of water is one of the benefits of the derivatization reaction. In general, byproducts appeared only at trace levels. Phenol, cresol, naphthol, and hydroxybiphenyls, which may be formed in a hydroxy dediazoniation with water as competing nucleophile to I3-, were identified by GC/MS. If the derivatization was carried out in the presence of methanol, the alcohol acted as nucleophile and resulted in additional formation of anisol, methylanisol, methoxynaphthalene, and methoxybiphenyls. This is the major reason spike solutions were diluted with water instead of methanol prior to use. For the diamino compounds, several mixed products were identified. However, compared with arylamines, the reactivity of other nucleophiles such as water or methanol is low and their reaction requires elevated temperatures. Furthermore, removal of the amino group occurred leading to chlorobenzene, dichlorobenzene, trichlorobenzene, chlorofluorobenzene, and nitrotoluenes. While the formation of phenols happened to the most reactive unsubstituted arylamines, removal of the amino group occurred with the less reactive chlorinated and nitrated amines. For all analytes, the estimated sum of byproducts does not exceed 5% of the iodinated target derivative. Extractability. For the extractability of organic compounds, their polarity, measured, for example, as octanol-water partition constants, is of great importance. In Table 2, experimental log Kow values for the anilines and the corresponding aromatic iodine compounds are given. If no experimental data were available, values calculated with clogP from Daylight Chemical Information 1032

Analytical Chemistry, Vol. 76, No. 4, February 15, 2004

Systems (http://www.daylight.com) are included. Unfortunately, for many iodinated compounds, no experimental data were available but predicted and experimental values agree very well with a mean deviation of 8.6%. The log Kow values for the anilines are in the range 0.09-3.35, for the derivatives in the range 3.05.3. This means a strong shift of Kow between the arylamines and their corresponding derivatives. The increase is between 1.29 and 4.45 orders of magnitude with an average increase of 1.92 orders of magnitude per amino group. However, a correlation between the extracted amount in SPME and log Kow is inconsistent as some authors have shown.44-46 This is especially the case if divinylbenzene is incorporated in the fiber and adsorption takes place in addition to partitioning. Nevertheless, there is a clear tendency that polar compounds are more difficult to extract while best results for SPME were obtained for nonpolar compounds. Therefore lowering the polarity of the analytes by derivatization certainly improves the extraction efficiency. Separation and Detection. For the optimization of the chromatographic and mass spectrometric conditions, an extract of the derivatives in ethyl acetate was used. During injection, a PTV-temperature program was applied to slowly vaporize solvent residues because the fast expansion during evaporation may harm the coating. As a consequence, in comparison with isothermal desorption, peaks are slightly broadened. On the other hand, thermal stress of the SPME fiber was reduced and the fiber could be used up to 80 times. The chromatographic parameters could be directly transferred from liquid injection to SPME without any change in the retention times. A typical chromatogram obtained with the extract in ethyl acetate is shown in Figure 3. Nearly all peaks are baseline separated, and differentiation of coeluting substances is possible by using the corresponding ion traces and by characteristic fragment ions. The retention times of the derivatives are summarized in Table 1. The tailing typically observed for chromatographic peaks of underivatized aromatic amines disappeared almost completely. The mass spectrometer was operated in the full-scan mode in order to allow detection and identification of nontarget analytes. In contrast to earlier results obtained on a quadrupole MS,47 the derivatives showed remarkably little fragmentation under the chosen conditions. This is one reason for the excellent sensitivity of the method. As a consequence, the base peak was normally identical with the molecular ion peak as can be seen from Table 1 (among the main mass fragments the base peak is underlined). Only in the case of the derivatives of 2A6NT, 4A2NT, 2A4,6DNT, 4A2,6DNT, and 2,4DA6NT the base peak is the [M - 17]+ peak due to the well-known ortho effect in fragmentation of nitrotoluenes. For all analytes, the base peak has been used for quantification. Furthermore, all derivatives show a fragment with mass 127 belonging to the iodine substituent but its intensity is often low. Derivatization shifts the molecular mass of the target compound by 111 amu per amino group. This is of great advantage because the number of unspecific fragments in the mass spectra decreases with increasing mass. As a consequence the chemical noise was significantly reduced and the emission current could be increased to 40 µA. In combination with low fragmentation, the low noise led to a further improvement of the sensitivity.

Figure 2. Effect of nitrite concentration on the derivatization yield of the highly deactivated 2A4,6DNT.

Figure 3. GC/MS chromatogram in the full-scan mode of a standard mixture of anilines after derivatization to the corresponding iodobenzenes. The analytes were extracted by LLE with ethyl acetate. The injected amount of each derivative corresponds to 5 ng of the precursor aromatic amine. For the abbreviations, see Table 1.

Selection of SPME Fiber. To the best of our knowledge, this is the first time aromatic iodides have been investigated by SPME. Therefore, we performed optimization of the SPME parameters in detail, i.e., the type of fiber, the extraction time, and the pH. The addition of salt was not taken into account because adverse effects are described in particular in direct-immersion mode48 due (48) Ferrari, R.; Nilsson, T.; Arena, R.; Arlati, P.; Bartolucci, C.; Basla, R.; Cioni, F.; Del Carlo, G.; Dellavedova, P.; Fattore, E.; Fungi, M.; Grote, C.; Guidotti, M.; Morgillo, S.; Muller, L.; Volante, M. J. Chromatogr., A 1998, 795, 371376.

to a rapid destruction of the fiber and precipitation of salt in the injector block. Five different commercially available SPME fibers covering a wide range of properties were evaluated. The polarities of the fiber coating are described as nonpolar (PDMS 100 µm), midpolar (Car/PDMS 75 µm, DVB/Car/PDMS 50/30 µm, PDMS/DVB 65 µm), and polar (CW/DVB, 65 µm).49 Furthermore, the fibers differ in the sorption mechanism: In the case of PDMS, absorption (49) Kataoka, H.; Lord, H. L.; Pawliszyn, J. J. Chromatogr., A 2000, 880, 35-62.

Analytical Chemistry, Vol. 76, No. 4, February 15, 2004

1033

(partitioning) in the bulk phase is dominant whereas adsorption on the surface dominates for DVB-based coatings, which consist of porous DVB particles with a surface of 750 m2/g held together by PDMS or CW.50 Fiber selectivities were determined with 10-mL samples containing the analytes at a concentration of 0.5 µg/L. For each type of fiber, five samples were derivatized independently. Derivatization and extraction were carried out as mentioned above. In all cases, the same injector conditions were used to allow for a comparison of results. Details of the SPME optimization procedure will be reported elsewhere. The extraction efficiency of the Car/PDMS coating was low. Only for a few derivatives did this fiber perform satisfactorily. The PDMS fiber on the other hand showed only good results for the least polar derivatives with log Kow > 3.5 whereas the derivatives of aniline and all nitro-containing compounds were only enriched to a low extent. All fibers based on DVB, especially PDMS/DVB and DVB/Car/PDMS, showed good extraction efficiencies for all analytes. In all cases best results were obtained with the PDMS/ DVB fiber, especially for the more polar derivates of the nitrosubstituted amines. As a consequence, PDMS/DVB fibers were used for all further experiments. They showed excellent performance and could be used for up to 80 extractions. The findings may possibly be ascribed to the different sorption mechanism and the larger surface area of the DVB-based fibers. However, sorption phenomena during extraction with regard to the porosity of the coating and the number of available sorption sites have not been investigated in detail so far. The surface area of SPME fibers is not characterized, nor are fibers with different surface areas available. Equilibration Time. For determination of the sampling time, 10-mL samples containing the analytes at a concentration of 0.5 µg/L were derivatized and extracted as described above. The extraction times were varied from 5 to 90 min, and each measurement was carried out in duplicate. As equilibrium was not reached after 90 min for any of the analytes, the equilibration times were estimated with the use of a fitting function. Based on work by Ai, the equilibration times could be described as the exponential function n(t) ) n0(1 - e-at) + C.51 The parameter a represents the slope of the equilibration curve, indicating how fast equilibrium is reached. The amounts of analyte extracted at equilibrium and before reaching equilibrium are characterized by n0 and n(t), respectively, and C is a constant. It is assumed that equilibrium is reached when 95% of the equilibrium amount is extracted.52 The equilibration times are between 135 and 345 min for all analytes. Equilibration times of several hours are known for PAHs,53,54 which have log Kow values very similar to those of the iodinated derivatives.53 Using an extraction time of 30 min, an average of 37% of the maximum amount after achieving equilibrium was sorbed to the SPME fiber. While equilibration for the compounds without nitro groups takes ∼180 min, the (50) Pawliszyn, J. Applications of solid-phase microextraction; Royal Society of Chemistry: Cambridge, 1999. (51) Ai, J. Anal. Chem. 1997, 69, 1230-1236. (52) Pawliszyn, J. Solid-phase microextraction: theory and practice; Wiley-VCH: New York, 1997. (53) Doong, R. A.; Chang, S. M. Anal. Chem. 2000, 72, 3647-3652. (54) Langenfeld, J. J.; Hawthorne, S. B.; Miller, D. J. Anal. Chem. 1996, 68, 144-155.

1034 Analytical Chemistry, Vol. 76, No. 4, February 15, 2004

Table 3. Correlation Coefficients, Detection Limits, and Precision for the in Situ Derivatization/SPME/GC/ MS Method in Nanopure Water

compound

correl coeff Ra

LOD (ng/L)

aniline 4AT 4CA 3,4DCA 2,4,5TCA 3C4FA 2,4DAT 2,6DAT 2NaA 2AB 4AB 4NA 2A4NT 2A6NT 4A2NT 2A4,6DNT 4A2,6DNT 2,4DA6NT

0.9987 0.9996 0.9997 0.9994 0.9995 0.9996 0.9967 0.9989 0.9987 0.9935 0.9991 0.9916 0.9923 0.9944 0.9946 0.9934 0.9924 0.9952

4 12 2 3 6 3 13 7 11 5 9 5 8 2 3 38 27 30

precisionc with ISd (without) 6.0 4.3 3.8 6.7 11 4.3 11 9.2 7.4 14 11 20 16 14 16 20 13 16

(15) (19) (16) (17) (15) (17) (17) (16) (20) (23) (10) (3.7) (4.3) (6.2) (5.0) (4.2) (10) (4.5)

a Determined with n ) 15 samples in the linear range 50-5000 ng/L. b Limit of detection for S/N ) 3, n ) 9 samples, and concentration 0.5 µg/L. c Precision expressed as RSD (%) at the 0.5 µg/L level, n ) 8 replicates. d Aniline-d5 at a concentration of 0.5 µg/L.

median equilibration time for nitro-substituted derivatives is 265 min. Validation. Calibration was accomplished according to the principles described by Kromidas55 and the calibration range selected in accordance with the EU regulations for toxic organic pollutants in drinking water, which comprises arylamines formed as metabolites of pesticides. For the determination of the validation parameters, the whole optimized method as described in the Experimental Section was used. The samples were spiked at five different concentrations (0.05, 0.1, 0.5, 1, and 5 µg/L) and at each level three replicates were prepared. To each sample aniline-d5 was added as internal standard to obtain a final concentration of 0.5 µg/L. The calibration curves were linear over a range of at least 2 orders of magnitude for all analytes under investigation, with correlation coefficients R between 0.9919 and 0.9997. The correlation coefficients of the nitro-substituted analytes were slightly worse due to lower suitability of the internal standard for these compounds as is also indicated by the lower precision (see below). For the determination of the detection limits, nine samples were analyzed containing the analytes at the 0.5 µg/L level and compared with two blanks. After detection in full-scan mode, the signal-to-noise ratios were calculated with the Varian software. The detection limits were calculated as the average amount of analyte yielding a response that is three times the noise. As can be seen from Table 3, the LODs for all analytes except 2A4,6DNT (38 ng/ L), 4A2,6DNT (27 ng/L), and 2,4DA6NT (30 ng/L) are below 13 ng/L, underlining the high sensitivity of the method. The three analytes with higher LODs are the three latest eluting compounds, and the lower sensitivity can be attributed to elevated oven temperatures and column bleeding leading to higher noise levels. Nevertheless, the analytical method for all target compounds meets the EU regulatory levels for drinking water of 0.1 µg/L. (55) Kromidas, S. Validierung in der Analytik; Wiley-VCH: Weinheim, 1999.

Figure 4. Chromatogram of 1-chloro-4-iodobenzene after in situ derivatization/SPME of 4-chloroaniline at a concentration of 10 ng/L. In (c), the single ion trace 238 is shown. In (a) and (b), the mass spectrum in the peak maximum and the reference spectrum are shown.

The indicated detection limits were confirmed by analyzing three samples containing the analytes at a concentration of 10 ng/L. Even at this low level, a good agreement with reference mass spectra was found. For example, for the derivative of 4CA, in addition to the molecular peak at 238 m/z, the characteristic mass traces 127 and [M - 127]+ are clearly observable (Figure 4). These results confirmed the excellent sensitivity of the presented derivatization/SPME method. For the chlorinated compounds, the results are at least comparable to other SPME publications for aromatic amines.32,36 For the more polar alkylated

and nitrated aromatic amines, the method is at least 1 order of magnitude more sensitive.33-35 Many analytes were analyzed by SPME for the first time. Detection limits in the lower nanogram per liter level have been reported for SPE,18,22 but the required large sample volumes involved the problems mentioned above. Especially for the diamino compounds under investigation, analysis is difficult with SPE19,22,31 due to their high polarity causing rapid breakthrough. For the determination of precision, eight samples containing the analytes and aniline-d5 at a concentration of 0.5 µg/L were Analytical Chemistry, Vol. 76, No. 4, February 15, 2004

1035

Table 4. Recoveries in Spiked Water Samples of Different Origin with and without Aniline-d5 as Internal Standard IS mean recovery (%)a drinking water b compound aniline 4AT 4CA 3,4DCA 2,4,5TCA 3C4FA 2,4DAT 2,6DAT 2NaA 2AB 4AB 4NA 2A4NT 2A6NT 4A2NT 2A4,6DNT 4A2,6DNT 2,4DA6NT

with

ISe

119 115 103 107 118 103 87 98 105 97 115 95 90 91 94 84 86 85

surface waterb

without 107 103 95 98 110 95 80 90 95 89 106 88 83 84 87 77 79 78

with

ISe

94 93 98 98 107 98 77 87 91 77 101 100 94 90 97 96 124 104

well waterc,d

without 84 83 87 87 96 88 69 78 81 69 91 91 86 81 88 88 112 94

with

ISe

103 101 101 103 103 101 135 121 121 127 131 149 149 138 146 ndf nd 160

without 70 68 68 70 70 68 91 82 82 88 90 100 101 93 98 nd nd 108

a Referring to Nanopure water, determined at the 0.5 mg.L spike level. b n ) 9. c n ) 6. d Sample from a former ammunition plant highly contaminated with NACs. e Aniline-d5 at a concentration of 0.5 µg/L. f nd, not determined.

derivatized and analyzed as described above. Incorporating the IS, the precision expressed as relative standard deviation (RSD) is between 3.8 and 20% as can be seen from Table 3. Apart from the nitro-substituted compounds, the average is 8.0% whereas for the amines with nitro groups worse results were obtained. For these compounds, aniline-d5 proved unsuitable as IS as can be seen from the evaluation without internal standard. In the latter case, the average RSD for nitro-substituted arylamines is only 5.4% while all other RSDs are significantly higher. Therefore, for further experiments, an appropriate internal standard such as 4-nitroaniline-d4 should be used for arylamines containing nitro groups. To compare run-to-run and day-to-day precision, 18 additional samples were analyzed on three subsequent days. With a onesided F-test, no significant differences were found between both series except for aniline at the 95% confidence level. As SPME is strongly influenced by pH, salt concentration, and dissolved organic carbon content of the sample matrix, three water samples of different origin were investigated. Performance of the overall method for a drinking water sample, a surface water sample, and a groundwater sample from a former ammunition plant highly contaminated with NACs was compared with that for Nanopure water. The recoveries were calculated from n ) 6 or n ) 9 samples spiked at a concentration of 0.5 µg/L (Table 4). In comparison with the Nanopure water samples (Table 3), the precision did not change significantly and matrix compounds did not hamper peak integration in any case. The recoveries of the spiked real water samples are in good agreement with the Nanopure water samples and differ by ∼11.6%. Apart from few exceptions, the overall recoveries show differences compared with Nanopure water below 20%. Since there is only a minor influence of sample matrix, quantification by external calibration proved to be acceptable, too. The deviations tend to increase with the complexity of the matrix. As the matrix constituents have a compound-specific influence on the extraction of the derivatives, this effect is difficult to compensate by a single internal standard. 1036 Analytical Chemistry, Vol. 76, No. 4, February 15, 2004

For instance, Kelz demonstrated that extraction of nitro-substituted iodinated compounds is increased by the addition of salt while extraction of compounds without nitro groups is hampered.56 On the other hand, as indicated by the recoveries, despite much higher amounts of NACs in the groundwater sample, saturation of the fiber did not occur. Saturation due to the limited number of sorption sites had been identified by Gorecki et al. as the disadvantage of the DVB-coated SPME fibers.57 Analysis of Water Samples. The in situ derivatization/ SPME/GC/MS method was applied to the analysis of several groundwater samples from the former ammunition plant in Stadtallendorf/Hessen. The quantification was based on a fourpoint external standard calibration curve that was generated by spiking Nanopure water samples with the analytes. Derivatization and extraction of the calibration and the real water samples was performed under exactly identical conditions. Each sample was analyzed five times using the optimized derivatization/SPME method. Because of better reproducibility for the nitroaromatic compounds under investigation, quantification was done without aniline-d5. Figure 5 shows a chromatogram in the full-scan mode of a highly contaminated sample. The pale yellow color of the (56) Kelz, H. Ph.D. thesis, Marburg, 1999. (57) Gorecki, T.; Yu, X. M.; Pawliszyn, J. Analyst 1999, 124, 643-649. (58) Smith, C. J.; Hansch, C. Food Chem. Toxicol. 2000, 38, 637-646. (59) Inoue, S.; Ogino, A.; Kise, M.; Kitano, M.; Tsuchiya, S.; Fujita, T. Chem. Pharm. Bull. Jpn. 1974, 22, 2064-2068. (60) Elovitz, M. S.; Weber, E. J. Environ. Sci. Technol. 1999, 33, 2617-2625. (61) Haderlein, S. B.; Weissmahr, K. W.; Schwarzenbach, R. P. Environ. Sci. Technol. 1996, 30, 612-622. (62) Mahmud, R.; Tingle, M. D.; Maggs, J. L.; Cronin, M. T. D.; Dearden, J. C.; Park, B. K. Toxicology 1997, 117, 1-11. (63) Wu, C. D.; Wei, D. B.; Liu, X. H.; Wang, L. S. Bull. Environ. Contam. Toxicol. 2001, 66, 777-783. (64) de Wolf, W.; de Bruijn, J. H. M.; Seinen, W.; Hermens, J. L. M. Environ. Sci. Technol. 1992, 26, 1197-1201. (65) Eadsworth, C. V.; Moser, P. Chemosphere 1983, 12, 1459-1475. (66) Chiou, C. T.; Freed, V. H.; Schmedding, D. W.; Kohnert, R. L. Environ. Sci. Technol. 1977, 11, 475-478.

Figure 5. GC/MS chromatogram of the sample P2 from the former ammunition plant in Stadtallendorf (a) in the full-scan mode and (b) single ions (218 + 246 + 249 + 263 + 291 + 372 m/z). The sample was derivatized and extracted by SPME as described. Compound identification: nitrotoluenes (NT), nitroxylenes (NX), dinitrotoluenes (DNT), trinitrotoluene (TNT), and aminonitrotoluenes (ANT); for further abbreviations see Table 1.

sample indicated large amounts of NACs, and several nitrotoluenes, dinitrotoluenes, and trinitrotoluene as well as some nitroxylenes were identified by their characteristic mass spectra. In the samples under investigation, the contents of several aminonitrotoluenes arising from microbial reduction of the nitroaromatic explosives were quantified. Table 5 summarizes the results obtained for different groundwater samples from Stadtallendorf. Apart from compounds present near the limit of quantification, the precision is good regarding the complex matrix. Only the quantification of 2,4DA6NT in sample P2 is hampered by high amounts of trinitrotoluene exceeding the capacity of the ion trap. In addition to external calibration, quantification was done by standard addition. To that end, six replicates of the sample ASB3 were spiked at a concentration of 0.5 µg/L and analyzed in the same way as described above. In Table 5, the results obtained by external calibration and standard addition are compared. The amounts determined by both approaches are comparable considering the low concentrations of the analytes.

In previous studies, similar samples from Stadtallendorf were extracted by SPE and analyzed with GC-ECD.47 Apart from seasonal fluctuations and differences between the sampling points, the former values are comparable to the results obtained from the derivatization/SPME method. Although only 1% of the sample volume was required, the sample preparation procedure was much less tedious, and the sensitivity was even better. CONCLUSIONS In this paper, we describe a new approach to overcome problems during sample preparation due to the polarity of the analytes. With the diazotation/iodination, a simple derivatization was presented to increase the extractability of aromatic amines by solid-phase microextraction. In general, this is the first in situ derivatization/SPME method suitable for a larger number of analytes bearing various functional groups. The method was successfully used even for the analysis of diamino compounds and strongly deactivated anilines. By iodination, the molecular Analytical Chemistry, Vol. 76, No. 4, February 15, 2004

1037

Table 5. Detected Nitroaromatic Amines in Groundwater Samples from the Former Ammunition Plant in Stadtallendorf P2

ASB 3

external calibration

external calibration

standard additiona

compound

concn (µg/L)

RSD (%)b

concn (µg/L)

RSD (%)c

concn (µg/L)

RSD (%)b

4NA 2A4NT 2A6NT 4A2NT 2A4,6DNT 4A2,6DNT 2,4DA6NT

0.27 0.61 1.32 1.40 6.41 14.5 0.24

10 3.0 4.9 5.9 2.6 4.4 19

0.05 0.02 0.03 0.04 2.22 4.44 0.02

14 15 13 14 5.1 6.0 14

0.01 0.01 0.02 0.02 3.12 5.05 0.05

15 13 17 8.4 4.0 7.2 11

a

lower nanogram per liter range, and precision was similar to other SPME methods. Compared with earlier results, sensitivity is significantly improved. In view of small sample volumes in combination with automation, the method has several benefits for routine analysis. The in situ derivatization/SPME/GC/MS method was successfully applied to several real water samples. ACKNOWLEDGMENT We thank the Restek Corp. for providing the Stx CLPesticides column and Klaus Steinbach for his support. We also acknowledge the Institute for Environmental Hygiene Marburg where most experiments were carried out and who facilitated this work until their closure. In particular, Thorben Bonarius, Henrik Kelz, Melanie Less, and Eberhard von Lo¨w are to be named.

Samples were spiked at the 0.5 µg/L level. b n ) 5. c n ) 4.

Received for review September 18, 2003. Accepted November 27, 2003. mass is shifted by 111 amu and sensitivity benefits from little fragmentation and low noise. The limits of detection were in the

1038 Analytical Chemistry, Vol. 76, No. 4, February 15, 2004

AC035098P