Anal. Chem. 2007, 79, 2386-2393
Compound-Specific Nitrogen and Carbon Isotope Analysis of Nitroaromatic Compounds in Aqueous Samples Using Solid-Phase Microextraction Coupled to GC/IRMS Michael Berg,† Jakov Bolotin,‡ and Thomas B. Hofstetter*,‡
Eawag, Swiss Federal Institute of Aquatic Science and Technology, 8600 Dubendorf, Switzerland, and Institute of Biogeochemistry and Pollutant Dynamics (IBP), ETH Zurich, 8092 Zurich, Switzerland
Solid-phase microextraction (SPME) coupled to gas chromatography/isotope ratio mass spectrometry was used to determine the δ15N and δ13C signatures of selected nitroaromatic contaminants such as the explosive 2,4,6trinitrotoluene (TNT) for derivation of isotopic enrichment factors of contaminant transformation. Parameters for efficient extraction of nitroaromatic compounds (NACs) and substituted anilines from water samples were evaluated by SPME-GC/MS. δ13C signatures determined by SPME-GC/IRMS and elemental analyzer IRMS (EA-IRMS) were in good agreement, generally within (0.7‰, except for 2,4-dinitrotoluene (2,4-DNT) and TNT, which showed slight deviations (99.5%) from Scharlau (Barcelona, Spain) were used to prepare stock solutions. SPME of Water Samples. Three different fiber coatings (stationary phases) were tested for SPME extraction, namely, polyacrylate (PA), carbowax-divinylbenzene, and carboxendivinylbenzene. Among them, the PA fiber (85 µm, Supelco) was the best suited for aqueous extraction of NACs, which is in agreement with studies reported in the literature.32,35 Moreover, compared to alternative fiber materials, PA enabled more reproducible measurements and the material was robust enough to achieve, on average, 100 injections/fiber. Direct aqueous immersion SPME was performed, and the following parameters were systematically evaluated: extraction time, temperature, NaCl concentration, and desorption time. The analytes were desorbed from the SPME fiber at 270 °C in a split/splitless injector equipped with a deactivated liner. The optimized SPME settings applied (26) Shouakar-Stash, O.; Drimmie, R. J.; Zhang, M.; Frape, S. K. Appl. Geochem. 2006, 21, 766-781. (27) Dias, R. F.; Freeman, K. H. Anal. Chem. 1997, 69, 944-950. (28) Dayan, H.; Abrajano, T.; Sturchio, N. C.; Winsor, L. Org. Geochem. 1999, 30, 755-763. (29) Halasz, A.; Groom, C.; Zhou, E.; Paquet, L.; Beaulieu, C.; Deschamps, S.; Corriveau, A.; Thiboutot, S.; Ampleman, G.; Dubois, C.; Hawari, J. J. Chromatogr., A 2002, 963, 411-418. (30) Mayfield, H. T.; Burr, E.; Cantrell, M. Anal. Lett. 2006, 39, 1463-1474. (31) Conder, J. M.; La Point, T. W.; Lotufo, G. R.; Steevens, J. A. Environ. Sci. Technol. 2003, 37, 1625-1632. (32) Monteil-Rivera, F.; Beaulieu, C.; Hawari, J. J. Chromatogr., A 2005, 1066, 177-187. (33) Hofstetter, T. B.; Heijman, C. G.; Haderlein, S. B.; Holliger, C.; Schwarzenbach, R. P. Environ. Sci. Technol. 1999, 33, 1479-1487. (34) Hofstetter, T. B.; Schwarzenbach, R. P.; Haderlein, S. B. Environ. Sci. Technol. 2003, 37, 519-528. (35) Calderara, S.; Gardebas, D.; Martinez, F. Forensic Sci. Int. 2003, 137, 6-12.
Analytical Chemistry, Vol. 79, No. 6, March 15, 2007
2387
Table 1. Names, Abbreviations, and Suppliers of the NACs and Substituted Anilines Investigated in This Work compound name
abbrev
NACs 2-methylnitrobenzene 4-methylnitrobenzene 2-chloronitrobenzene 4-chloronitrobenzene 2,4-dinitrotoluene 2,6-dinitrotoluene 2,4,6-trinitrotoluene
2-CH3-NB 4-CH3-NB 2-Cl-NB 4-Cl-NB 2,4-DNT 2,6-DNT 2,4,6-TNT
anilines 2-methylaniline 4-methylaniline 2-chloroaniline 4-chloroaniline
2-CH3-An 4-CH3-An 2-Cl-An 4-Cl-An
MW
target iona (m/z)
qualifier iona (m/z)
Aldrich, >99% Fluka, >98% Fluka, >99% Fluka, >98% Fluka, >98% Aldrich, 98% Ems-Dottikon
137.14 137.14 157.56 157.56 182.14 182.14 227.13
120 137 157 157 165 165 210
91 107 111 111 89 89 89
Merck, for analysis Merck, for analysis Aldrich, >99.5% Fluka, >99%
107.15 107.15 127.57 127.57
106 106 127 127
77 77 92 92
supplier, purity
a Experimental conditions were 4 M NaCl, extraction time 35 min, extraction temperature 45 °C, desorption time 3 min, and desorption temperature 270 °C.
for the quantification of the investigated NACs in water samples as well as the determination of δ15N and δ13C values were as follows: 1.3 mL of the aqueous solution was transferred into 2-mL autosampler glass vials containing 0.30 g of NaCl (final concentration 4.0 M) and immediately closed with Teflon-sealed screw caps. Subsequently, the samples were shaken on a Vortex shaker to dissolve NaCl completely. The vials were equilibrated by the autosampler at 40 °C followed by direct immersion SPME for 45 min. After the extraction, the analytes were thermally desorbed for 3 min in the split/splitless injector (270 °C). Each measurement was conducted in triplicate. The SPME fiber was conditioned for 30 min at 270 °C after 20-30 samples and replaced after ∼100 injections. The influence of SPME parameters on the signal intensities obtained from SPME-GC/MS are illustrated in Figure S1, Supporting Information. GC/MS Analysis. Evaluation of the SPME method and concentration measurements in aqueous samples were conducted with a GC/MS system (Trace GC/Trace DSQ EI 250, Thermo Electron Corp., Waltham, MA). The GC was equipped with a cold on-column injector, a split/splitless injector with a Merlin Microseal (Merlin Instrument Co.), a deactivated guard column (0.5 m × 0.53 mm, BGB, Anwil, Switzerland), and a CombiPAL autosampler (CTC, Zwingen, Switzerland), allowing liquid injections as well as solid-phase microextraction followed by thermal desorption. Helium was used as carrier gas at a constant column head pressure of 100 kPa. The analytes were separated on a 30 m × 0.32 mm fused-silica column (Zebron, ZB-5-ms, 0.25 µm, Phenomenex, Torrence, CA) with the following temperature program: 1 min at 50 °C, with 10 °C/min to 250 °C, and 5 min at 250 °C. The compound peaks were recorded in the positive electron impact mode and selected ion monitoring, using the mass traces of the target and qualifier ions listed in Table 1. External calibration was applied to quantify the analyte concentrations. Cold on-column injections, allowing for the introduction of defined amounts of analyte into the GC/MS, were used to determine SPME extraction efficiencies and instrumental detection limits. For this purpose, ethyl acetate solutions containing 6.5 mg L-1 of the standard compounds were prepared and 1 µL was injected in the cold oncolumn injector (25 °C). 2388
Analytical Chemistry, Vol. 79, No. 6, March 15, 2007
Isotope Ratio Measurements Using GC/IRMS. δ13C signatures were determined using a Trace GC (Thermo Electron Corp.) coupled to an isotope ratio mass spectrometer (IRMS; DeltaPLUS XL, Thermo Electron Corp.) via a combustion interface (GC Combustion III).25 Oxidation and reduction reactors were maintained at 940 and 650 °C, respectively. The GC was equipped with the same column and autosampler as for GC/MS analysis, and the identical temperature program and He carrier gas pressure (100 kPa) were used. The NiO/CuO/Pt wires in the combustion unit were oxidized with O2 for 12 h at 940 °C prior to use. Isotope signatures were measured after SPME extraction of the analytes and, for selected measurements, after on-column injection. All δ13C signatures of the analytes are reported relative to Vienna PeeDee Belemnite (δ13CVPDB).36 Nitrogen isotope signatures were determined in separate measurements with the same setup used for δ13C analysis with the following adjustments: the oxidation reactor was used without preoxidation at 980 °C, and the reduction reactor temperature was 650 °C.15-17 Liquid N2 was used to trap CO2 produced from analyte combustion, thus preventing it from entering the ion source and causing interferences on the measured N2 isotopologues. All δ15N signatures are given relative to air (δ15NAir). To verify δ15N and δ13C values for potential isotope fractionation during preconcentration by SPME, CSIA, or both, the isotope signatures of all analytes were measured by elemental analyzer (Carlo Erba) coupled to an IRMS (EA-IRMS, Fisons Optima37) to obtain reference isotope ratios for comparisons. Stable Isotope Signatures Determined in Water Samples with SPME-GC/IRMS. δ15N and δ13C signatures of monosubstituted NACs were measured by SPME-GC/IRMS together with the corresponding substituted anilines. Di- and trinitrotoluenes were analyzed as single compounds to avoid potential interferences from competitive adsorption of the analytes to the SPME fiber. An example chromatogram for δ15N measurements by (36) Gro ¨ning, M. In Handbook of stable isotope analytical techniques; de Groot, P. A., Ed.; Elsevier: Amsterdam, 2004; Vol. 1, pp 874-906. (37) Lehmann, M. F.; Bernasconi, S. M.; Barbieri, A.; McKenzie, J. A. Geochim. Cosmochim. Acta 2002, 66, 3573-3584.
Figure 1. Chromatogram of an SPME-GC/IRMS analysis of the δ15N isotope signature of 4-nitrotoluene (4-CH3-NB) and the corresponding 4-methylaniline (4-CH3-An): (a) mass 28 (14N2+) and mass 29 (15N14N+) ion currents and (b) partial chromatogram of instantaneous mass 29/28 ion-current ratios.
SPME-GC/IRMS is shown in Figure 1 for 4-CH3-NB and 4-CH3An including the instantaneous mass 29/28 ion-current ratios. To evaluate the SPME-GC/IRMS method for the determination of stable isotope signatures of NACs, we studied the influence of selected SPME parameters with regard to its possible N isotope fractionation caused by the enrichment technique. Lowered SPME extraction efficiencies, as observed during GC/MS measurements using NaCl concentrations below 4 M, extraction times shorter than 20 min, or extraction temperatures below 30 °C resulted in higher standard deviations of δ15N (σ ( 2‰, data not shown). Deviations of δ15N values measured by SPME-GC/IRMS under these conditions from isotope signatures measured by EA-IRMS were small compared to the loss of precision and showed a bias toward more negative δ15N values. Uncertainty of isotope signatures determined by CSIA, which can arise from variations of analyte masses introduced into the mass spectrometer (nonlinearity effects),38,39 were investigated for both δ13C and δ15N. We measured the changes of δ15N over variable concentration ranges by GC/IRMS using both on-column (38) Merritt, D. A.; Hayes, J. M. Anal. Chem. 1994, 66, 2336-2347. (39) Jochmann, M. A.; Blessing, M.; Haderlein, S. B.; Schmidt, T. C. Rapid Commun. Mass Spectrom. 2006, 20, 3639-3648.
Figure 2. δ13C and δ15N isotope signature of 4-CH3-NB as a function of IRMS peak amplitudes (m/z 44 for 12CO2+, m/z 28 for 14N +) for on-column injection (OC) to GC/IRMS and for SPME-GC/ 2 IRMS (errors are (1σ). The horizontal bar represents isotopic reference values of the pure compound analyzed with EA-IRMS ( 1σ.
injection and SPME to examine the concentration ranges suited for accurate analysis. Comparisons were made on the basis of signal intensities (i.e., peak amplitudes at ion currents of the more abundant isotopologues). According to the instrument specification, the minimum analyte mass injected on the GC column for accurate isotope ratio determination is 0.8 nmol of C (9.6 ng of C) and 1.5 nmol of N2 (42 ng of N), respectively. Consequently, the range of investigated analyte concentrations ranged from 10 µg L-1 (SPME) to 1 g L-1 (on-column) depending on the considered element, injection technique, and SPME extraction efficiency of the analyte. As illustrated for the analysis of 4-CH3-NB in Figure 2, δ13C and δ15N values of variable NAC concentrations and, consequently, peak amplitudes, deviate from the isotope ratios obtained EA-IRMS at low analyte masses introduced to the GC. Amounts of C masses transferred to the GC corresponded to 0.610 nmol (corresponding to peak amplitudes between 0.4 and 4 V) and resulted in δ13C values, which were within the experimental error of the reference signature determined by EA-IRMS. Analysis of δ15N by GC/IRMS was less sensitive owing to the lower abundance of heavy isotope (average of isotope ratios are 0.0112 for 13C/12C and 0.00366 for 15N/14N23), the lower ionization Analytical Chemistry, Vol. 79, No. 6, March 15, 2007
2389
Table 2. Absolute Recoveries, Extraction Efficiencies, and Detection Limits of the SPME-GC/MS and SPME-GC/IRMS Methods recoveriesa
extraction efficiencies
polluted groundwater (n ) 5) unspiked
spiked with 3.8 µg L-1
concn, µg L-1
concn, µg L-1
NACs 2-CH3-NB 4-CH3-NB 2-Cl-NB 4-Cl-NB 2,4-DNT 2,6-DNT 2,4,6-TNT
0.01 (20) 0.01 (31) ndg 0.41 (5) 1.2 (10) 0.50 (6) 83 (5)
anilines 2-CH3-An 4-CH3-An 2-Cl-An 4-Cl-An
0.40 (18) nd 0.03 (20) 0.16 (20)
detection limits SPMEGC/MS
SPME
SPMEGC/IRMS
spiked with 38 µg L-1 δ13C LOD,e mg L-1
δ15N LOD,f mg L-1
rec, %
extraction efficiency, %
enrichment factor
LOD,b µg L-1
IDL,c pg
MDL,d µg L-1
36.5 (7) 36.0 (5) nd 34.0 (4) 38.8 (1) 40.7 (2) 121 (3)
96 (6) 9 (4) nd 89 (4) 101 (1) 102 (2) 100 (10)
32 43 42 46 17 36 15
420 560 550 600 220 470 200
0.011 0.012 0.010 0.013 0.082 0.045 0.36
5 7 6 8 18 21 70
0.006 0.010 nd 0.060 0.36 0.090 12
0.073 0.19 0.16 0.16 0.60 0.22 0.78
1.6 2.5 4.8 2.3 9.6 5.1 22
40.8 (2) nd 39.9 (2) 34.3 (2)
106 (2) nd 105 (2) 90 (2)
5.5 5.4 21 13
72 70 270 170
0.029 0.025 0.011 0.023
2 2 3 4
0.22 nd 0.018 0.10
0.97 1.6 0.32 0.40
nah na na na
rec, %
concn, µg L-1
3.4 (10) 3.3 (6) nd 2.8 (6) 4.3 (4) 3.5 (3) nd
89 (9) 87 (5) nd 74 (4) 82 (4) 79 (3) nd
4.0 (9) nd 3.8 (6) 3.0 (6)
95 (9) nd 100 (6) 79 (5)
a Numbers in parentheses are RSDs. b Limit of detection (concentration equivalent to IDL). c Instrumental detection limit (signal/noise ) 3). Method detection limit (3 times standard deviation of concentration determined in unspiked groundwater). e Corresponding to 0.8 nmol of C on GC. f Corresponding to 1.5 nmol of N2 on GC. g nd, not determined. h na, not applicable (see text).
d
efficiencies of N2 compared to CO2, and the fact that the generation of one detectable ion (N2+) requires conversion of two analyte molecules (mono-NACs). Therefore, peak amplitudes between 0.4 and 0.5 V corresponding to 1.5 nmol of N2 loaded on the GC yielded reliable δ15N values within the range of values observed by EA-IRMS. δ15N measured by GC/IRMS were accurate to signal intensities of up to 3 V (10% were only resulting in measurements of unspiked groundwater for concentrations being very close to the MDL. The extraction efficiencies were generally higher for the NACs (32-46%) than for di- and trinitrotoluenes (15-36%) and substituted anilines (5.4-21%). We hypothesize that the latter observation is due to the anilines’ higher polarity as illustrated by their higher water solubility (e.g., 10-1 vs 10-3 M for substituted anilines
Figure 3. Comparison of δ13C and δ15N isotope signature measured for substituted NACs and anilines with SPME-GC/IRMS and EA-IRMS (all values (1σ). (a) δ13C values relative to VPDB and (b) δ15N values relative to Air.
vs NACs41) and lower affinity for more apolar phases (e.g., lower octanol-water partitioning constants of substituted anilines41). The corresponding enrichment factors of the analytes indicate the increase of GC/MS signal intensities using the SPME method compared to on-column injection per mole of analyte. The detection limits are dependent on the SPME extraction efficiency of the individual compounds as well as the noise of the mass traces monitored. With the exception of the two DNTs and TNT, very low LODs of
