Recently, it has been reported that

RAKSIT & JOHRI: JOURNAL OF AOAC INTERNATIONAL VOL. 84, NO. 5, 2001 1413 ENVIRONMENTAL ANALYSIS Determination of N-Nitrosodimethylamine in Environmental Aqueous Samples by Isotope-Dilution GC/MS SIM ASIT RAKSIT and SAIMA JOHRI ENTECH-Agri-Service Laboratory, Inc., Research Development Section, 6820 Kitimate Rd, Unit No. 4, Mississauga, ON, L5N 5M3, Canada An analytical method for the determination of N-nitrosodimethylamine (N-NDMA) in environmental aqueous samples has been developed using isotope dilution gas chromatography/mass spectrometry in the selected ion monitoring mode (GC/MS SIM). After deuterated N-nitrosodimethylamine (d 6 -N-NDMA) as surrogate standard was added to the samples, the analytes were extracted with methylene chloride, dried with anhydrous sodium sulfate, and concentrated in a rotary evaporator. The concentrated extracts were analyzed by GC/MS SIM after adding N-nitrosodiethylamine as an internal standard. The method detection limit for N-NDMA was 0.003 pg/ L and was validated by an analysis of a fortified water sample. The method was applied to real samples. Recently, it has been reported that N-nitrosodimethylamine (N-NDMA) is a suspected carcinogen and very toxic (1, 2). Many N-nitroso-compounds are used as solvents in fiber and plastic industries, as antioxidants in fuel, insect repellants, insecticides, fungicides, and lubricating oils (3 5). These compounds are discharged into the environment by human activities such as fertilizer industries, sewage output, and animal feedlots where nitrogen related-compounds are used (6, 7). N-NDMA is very soluble and volatile and therefore it can reach into our aquatic environment. N-NDMA has been detected at ppb levels in surfacewaters, seawaters, and soils (8, 9). There are a number of studies on N-nitrosamines (including N-NDMA) content in food (10, 11), alcoholic beverages (12, 13), and finished rubber products (14, 15). Most of these analytical methods have predominantly used a gas chromatography-thermal energy analyzer (GC TEA). Very seldom was gas chromatography/mass spectrometry (GC/MS) used for identification only, determining N-nitrosamines, including N-NDMA, at levels of 0.5 to 10 ng/g in a variety of matrixes. The GC TEA system has not been used to determine N-NDMA in environmental aqueous samples at low levels. Received January 26, 2001. Accepted by JS May 9, 2001. However, a method is needed to quantitatively detect N-NDMA at ppt levels in water samples. A bench-top GC/MS system is adequate to produce similar informations. Our goal was to develop a comprehensive analytical method for the identification and quantitation of N-NDMA in environmental aqueous samples. Isotope dilution has been used extensively for quantitative analysis in biological samples (16, 17). In this method, a deuterated analog of target analyte is directly added to the samples to serve as surrogate standard. An advantage of this method is that any loss of the analyte during the sample extraction, solvent concentration, and GC can be corrected relative to the surrogate standard recovery because the analyte and its labeled analog exhibit the same effects. We present a rapid, sensitive, and specific analytical method, which includes sample preparation, identification, and quantitation of N-NDMA in environmental aqueous samples using isotope dilution GC/MS in the selected ion monitoring (SIM) mode. Experimental Chemicals and Standards (a) Sodium chloride. Reagent grade, granular (J.T. Baker, Phillipsburg, NJ). (b) Sodium sulfate. Reagent grade, anhydrous, granular (J.T. Baker), rinsed further with methylene chloride and conditioned at 200 C. (c) Sodium hydroxide. Solid, ACS certified (Fisher Scientific, Nepean, Ontario, Canada). (d) Methylene chloride. Distilled in glass (Caledon Laboratory Chemicals, Georgetown, Ontario, Canada). (e) Methanol. Distilled in glass (Caledon Laboratory Chemicals). (f) Stock standard solutions. Stock solutions of N-NDMA at a concentration of 100 µg/ml each in methylene chloride and methanol were purchased with certification from Absolute Standards, Inc. (Hamden, CT). (g) Surrogate standard. Deuterated N-nitrosodimethylamine (d 6 -N-NDMA, purity 98%) was purchased from Cambridge Isotope Laboratories (Woburn, MA). Two surrogate standard solutions containing 10 mg/ml each were prepared separately by dissolving accurately weighed 100 mg in 10 ml methylene chloride and methanol in volumetric flasks. The

1414 RAKSIT & JOHRI: JOURNAL OF AOAC INTERNATIONAL VOL. 84, NO. 5, 2001 Figure 1. Total ion chromatogram (TIC) obtained after extraction of an 800 ml deionized water sample spiked at 0.20 pg/ L of each analyte. (A) Surrogate, d 6 -NDMA; (B) analyte, N-NDMA; (C) N-NDEA, the internal standard concentration was 500 pg/ L. methanol solution was used as surrogate spike standard in samples. (h) Internal standard solution. Stock solution of N-nitrosodiethylamine (N-NDEA) at a concentration of 100 µg/ml in methylene chloride was purchased with certification from Absolute Standards, Inc. All standard solutions were further diluted with methylene chloride and/or methanol to desired concentrations. Sample Preparation A known volume of sample was filtered and collected in a 1 L separatory funnel. A measured amount of deuterated N-NDMA solution in methanol as surrogate was added and mixed with the sample by shaking. The ph of the sample was checked with ph paper and adjusted to 12 with a 50% solution of NaOH in water. NaCl (14 g/l) was added to the sample and shaken to ensure the salt was dissolved. The sample was then extracted 3 with 40 ml aliquots of methylene chloride by shaking. The extract was dried, concentrated, and transferred into a culture tube fitted with a teflon-lined screw cap. The final sample extract volume of 2 or 4 ml (depending on color of extracts) was measured accurately with a Hamilton syringe. The internal standard at known concentration was added to each extract and analyzed by GC/MS SIM. Preparation of Spiked Test Sample A spike composite test sample containing N-NDMA and d 6 -N-NDMA in 800 ml deionzed water was prepared separately at a concentration of 0.10 pg/µl each for the determination of method detection limit (MDL). The sample was extracted with methylene chloride in a manner similar to that described above and the extract was concentrated to a volume of 4 ml. The extract contained 20 pg/µl equivalent of each compound. The internal standard at a concentration of 500 pg/µl was added to each extract and analyzed by GC/MS SIM. Preparation of Environmental Samples A number of aqueous industrial effluent samples were collected by our clients and sent to our laboratory for the identification and quantitation of N-NDMA. A 500 ml portion of each sample was filtered, spiked with surrogate standard, d 6 -NDMA, at a concentration of 0.20 pg/µl, and extracted with methylene chloride in a manner similar to that given ear-

RAKSIT & JOHRI: JOURNAL OF AOAC INTERNATIONAL VOL. 84, NO. 5, 2001 1415 Figure 2. Electron ionization mass spectra of (A) surrogate, d 6 -N-NDMA, and (B) analyte, N-NDMA. The asterisked ions represented the characteristic ion-pair of analyte and its labeled analog.

1416 RAKSIT & JOHRI: JOURNAL OF AOAC INTERNATIONAL VOL. 84, NO. 5, 2001 Figure 3. Calibration curve for the quantitation of N-NDMA in the range of 20 to 600 pg/ L. The relative response defined the ratio of primary ion peak of the analyte to surrogate standard [i.e., ratio m/z (74/80)] plotted versus concentration of analyte (pg/ L) injected. lier. The extract was concentrated toa2mlvolume containing 50 pg/µl surrogate, d 6 -NDMA. A 1 µl aliquot extract containing internal standard at a concentration of 500 pg/µl was analyzed by GC/MS SIM. GC/MS System Analyses were performed using an HP 6890 series GC coupled with an HP 5973 MSD and HP 7683 series autosampler (Agilent Technologies, Inc., Wilmington, DE). The capillary column was an HP-5MS, 30 m 0.25 mm id, 0.50 µm film thickness. (a) GC conditions. Helium was used as the carrier gas with a flow velocity of 30 cm/s and measured at 40 C. The injector temperature was maintained at 250 C. In split mode operation, a split ratio of 20:1 was used for sample injection of a 1 µl volume at initial oven temperature of 35 C. The split ratio was found adequate to separate the N-NDMA and labeled analog. They were not separated when the sample was injected in the splitless mode. The oven temperature was programed with an initial temperature of 35 C for 3 min, increased at 5 C/min to 100 C, held for 5 min, then increased at 15 C/min to 280 C. All 3 compounds were eluted before the column temperature reached 100 C. The second temperature Table 1. Method detection limit of N-nitrosodimethylamine Compound Spiked amount, pg/µl Mean value, pg/µl a Standard deviation MDL, pg/µl RSD, % N-NDMA 0.100 0.110 0.001 0.003 0.91 a Mean measured concentration of 8 replicate analyses.

RAKSIT & JOHRI: JOURNAL OF AOAC INTERNATIONAL VOL. 84, NO. 5, 2001 1417 Table 2. Validation data for method detection limit of N-NDMA Compound Quantity injected, pg/µl Quantity found, pg/µl RSD, % d 6 -NDMA 0.0250 0.0243 6 N-NDMA 0.0030 0.0025 10 Figure 4. Total ion chromatogram (TIC) obtained after extraction of a 500 ml real sample with methylene chloride. (A) Surrogate, d 6 -NDMA, and (B) analyte, N-NDMA. The concentration of spiked surrogate was 0.20 pg/ L. The concentration of internal standard (C), N-NDEA, was 500 pg/ L. The inserted peaks A 1 and B 1 shown at 10-fold increase. D represents the electron ionization mass spectra of N-NDMA found in the real sample.

1418 RAKSIT & JOHRI: JOURNAL OF AOAC INTERNATIONAL VOL. 84, NO. 5, 2001 Table 3. Confirmation of the presence of N-NDMA in the environmental aqueous samples Analytes Ion pair program was used to remove unwanted compounds from the column. (b) MS conditions. The MS quad-temperature was held at 150 C; MS source temperature was maintained at 230 C; electron energy was 70 ev, and the vacuum system was connected to a turbo-molecular pump. The calibration standards were analyzed first using GC/MS in the fullscan mode, then MS was setup for SIM mode. In SIM mode, the following characteristic ions were used for each compound: For N-nitrosodimethylamine, ions m/z = 75, 74, 59, 43, 42, 41; for d 6 -N-nitrosodimethylamine, ions m/z = 82, 80, 62, 48, 46; and for N-nitrosodiethylamine, ions m/z = 103, 102, 71, 57, 56, 42. The dwell time was 100 ms for all ions. Results and Discussion Standard vs samples Mass Spectra and Criteria of Compound Identification No. samples analyzed N-NDMA 74, 42 2.10 vs 2.20 30 d 6 -N-NDMA 80, 46 2.50 vs 2.40 Figure 1 shows the total ion chromatogram (TIC) for 800 ml of a water sample spiked with 0.20 pg/µl N-NDMA, its labeled analog, and the internal standard, N-NDEA. The major peaks were well separated as predicted under experimental conditions given above. There were no interferences with other organics presented either in solvents or in samples. The peak in the chromatogram at 7.40 min was so low that it could not be identified in fullscan mode even with the help of a National Institute of Standards and Technology (NIST) library search. The fragment ions for the peak were m/z = 59, 44, 43, 42, 41. Figures 2A and B show electron ionization mass spectra of d 6 -N-NDMA and N-NDMA, respectively. The selected characteristic ion pairs (shown with asterisks) for analyte and its deuterated analog were monitored using GC/MS SIM. They were used for quantitation and confirmation of N-NDMA in water samples. The primary ion ratio of the analyte (m/z = 74) to the surrogate standard (m/z = 80) was used for quantitation. The ion ratio of the primary ion (m/z = 74) to the secondary ion (m/z = 42; obtained from ion abundances after backgound correction) of the same analyte was used to confirm the presence/absence of N-NDMA in the aqueous samples. The quality control of the analysis was achieved by (1) monitoring surrogate recovery in each sample (the average percent recoveries were 100 ± 15), and (2) monitoring the internal standard response (peak area of primary ion, m/z = 102) for each sample and standard chromatogram. The responses were found within 15%. Quantitative Analysis Calibration standard solutions containing N-NDMA at concentrations of 20, 50, 100, 200, 400, and 600 pg/µl were prepared volumetrically from the stock solutions in methylene chloride. Each level of calibration standard contained surrogate, d 6 -N-NDMA, and internal standard, N-NDEA, at concentrations of 200 and 500 pg/µl, respectively. These calibration standards were analyzed by GC/MS SIM. The linear response range for compound was established by plotting the ratio of the primary ion (m/z = 74) peak area of the analyte to that of surrogate primary ion (m/z = 80) versus the concentration of N-NDMA (pg/µl) injected on the GC column. Figure 3 shows the calibration curve for N-NDMA using d 6 -N-NDMA as the isotopic diluent and shows linear relationship from 20 to 600 pg/µl. The relative response factor (RRF) for N-NDMA versus d 6 -NDMA was determined from the calibration data sets during a 10-week analysis period. The average RRF value was 0.96 ± 0.10 providing relative standard deviation (RSD) values of 10.4%, n = 25. The accuracy and precision of the method was determined for each spiked laboratory water at a concentration of 1.0 pg/µl from the recovery of N-NDMA concentration. The values were 10 to 20% depending upon the concentration of N-NDMA found after sample analysis. Regression lines that fit the calibration curve were used for automatic quantitative analysis of MDL of N-NDMA as well as in the analysis of real samples. The concentration of the analyte in the extract was determined from the calibration curve. The final concentration of analyte in the sample was computed using the volumes of the original sample and the final extract. Method Detection Limit The MDL for the analysis was determined using the guidelines established and given in refs. 18 and 19. Accordingly, the MDL for N-NDMA was calculated by analyzing eight 1 µl aliquots of above spiked test sample. The concentrations of analytes were determined using the 6-level calibration curve. The MDL value was calculated using Student s t-value appropriate to 99% confidence level and a standard deviation estimated. Three times of standard deviation provided MDL of N-NDMA. Mean measured concentrations observed ranged from 97 to 110% of the true value. The MDL value and the actual concentrations used are given in Table 1. The MDL value was validated by analyzing an 800 ml composite water sample containing N-NDMA and d 6 -N-NDMA at concentrations of 0.003 and 0.025 pg/µl, respectively. The fortified sample was extracted with methylene chloride according to the extraction procedure described earlier. The extract was concentrated to a volume of 0.50 ml and contained 5.0 and 40 pg/µl equivalent of N-NDMA and d 6 -N-NDMA, respectively. A calibration standard solution at similar concentrations was prepared volumetrically by diluting above stock solutions. The internal standard at a concentration of 80 pg/µl was added to these 2 solutions and analyzed by GC/MS SIM. The precision of the determinations was assessed by injections of 8 replicates and expressed as the RSD. The typical range of RSD

RAKSIT & JOHRI: JOURNAL OF AOAC INTERNATIONAL VOL. 84, NO. 5, 2001 1419 values was 6 10%. Validation data and the actual concentrations used are given in Table 2. Analysis of Real Environmental Samples We have been using the proposed method in our laboratory for the identification and quantitation of N-NDMA in real aqueous samples. These samples were extracted with methylene chloride in a similar manner as described above. The extracts were analyzed by GC/MS SIM. Figure 4 shows a typical TIC obtained for a 500 ml real sample after extraction with methylene chloride. Figure 4D shows the electron ionization mass spectra of N-NDMA found in a real sample. Figures 4A 1 and 4B 1 show an expanded plot of A and B data, respectively. The target peak identification was performed by comparing the retention time of the standard chromatogram and ion ratios given above. A 1.12 pg/µl concentration of N-NDMA was found in the real sample and the surrogate recovery was 90%. The other peaks in the chromatogram could not be assigned to any compounds because the TIC was obtained in SIM mode. Table 3 shows the consistency between analyte, N-NDMA ion ratios, and surrogate standard, d 6 -N-NDMA, ion ratios in standards and samples, confirming the presence of the target analyte in the environmental real samples. This method provides values within 15% of those expected, with good reproducibility. Conclusions The work presented here provided sample preparation, identification, and quantitative analysis of N-NDMA in aqueous samples. N-NDMA and its labeled analog were separated by 6.0 ± 0.5 s in TIC, indicating that the 2 peaks were well resolved. The characteristic ion ratio limits given in Table 3 could be used for the presence/absence of N-NDMA in any aqueous sample. The surrogate recoveries in the real samples showed that the analyte loss during the sample extraction, solvent concentration, and analysis was not singificant. Isotope-dilution GC/MS with electron-impact ionization in SIM mode acquisition generates chromatograms that help to determine and confirm the presence of N-NDMA in real samples. Acknowledgment We are grateful to M. Misra of Entech, Agri-Service Laboratory, Inc. (Mississauga, Ontario, Canada), for his support and interest in this work. References (1) Seventh Annual Report on Carcinogens (1994) Report PB95-109781, p. 286 (2) Haley, T.J. (1982) in Handbook of Carcinogens and Hazardous Substances: Chemical and Trace Analysis, M.C. Bowman (Ed.), Marcel Dekker, New York, NY, 1, pp 1 18 (3) Scientific and Technical Assessment Report on Nitrosamines (1977) Office of the Research and Development, U.S. EPA, Washington, DC, EPA-600/6 77-001 (4) The Merk Index (1996) 12th Ed., S. Budavani (Ed.), Rahway, NJ (5) Fleming, E.C., Pennington, J.C., Wachob, B.G., Howe, R.A., & Hill, D.O. (1996) J. Hazard. Mater. 51, 151 164 (6) Mills, A.L., & Alexander, M. (1976) J. Environ. Qual. 5, 437 446 (7) Mosier, A.R., & Torbit, S. (1976) J. Environ. Qual. 5, 465 468 (8) Sen, N.P., Baddoo, P.A., Water, D., & Boyle, M. (1994) Inter. J. Environ. Anal. Chem. 56, 149 163 (9) Jenkins, S.W.D., Koester, C.J., Taguchi, V.Y., Wang, D.T., Palmentier, J.P.F.P., & Hong, K.P. (1995) Envirom. Sci., Pollut. Res. 2, 207 210 (10) Song, P.J., & Hu, J.F. (1988) J. Chem. Toxic. 26, 205 208 (11) Pensabene, J.W., Fiddler, W., & Gates, R.A. (1992) J. AOAC Int. 75, 438 442 (12) Billedeau, S.M., Miller, B.J., & Thomson Jr, H.C. (1988) J. Food Sci. 53, 1696 1699 (13) Longo, M., Lionetti, C., & Cavallaro, A. (1995) J. Chromatogr. A 708, 303 307 (14) Fiddler, W., Pensabene, J.W., & Kimoto, W.J. (1985) Am. Ind. Hyg. Assoc. J. 46, 436 465 (15) Ireland, C.B., Hytrek, F.P., & Lasoski, B.A. (1980) Am. Ind. Hyg. Assoc. J. 41, 895 900 (16) Eiceman, G.A., Hill, H.H., Davani, B., & Garden-Torresday, J. (1996) Gas Chromatography, Anal. Chem. 68, 291R 308R (17) Pyon, K.H., Kracko, D.A., Strunk, M.R., Bechfold, W.E., Dahl, A.R., & Lewis, J.L. (1997) J. Anal. Toxic. 21, 363 368 (18) 40 CFR Part 136, Appendix B (1984) Federal Register, EPA, V 49, 198 199 (19) Glaser, J.A., Foerst, G.D., McKee, S.A., Quave, S.A., & Budde, W.L. (1981) Environ. Sci. Technol. 15, 1426 1429