Determination of carbonyls using liquid chromatography-mass spectrometry with atmospheric pressure chemical ionization† Christine Kempter, Gabriela Zurek and Uwe Karst* Westfa�lische Wilhelms-Universita�t Mu� nster, Anorganisch-Chemisches Institut, Abteilung Analytische Chemie, Wilhelm-Klemm-Str. 8, D–48149 Mu�nster, Germany Received 7th April 1999, Accepted 3rd June 1999 A liquid chromatographic method for the determination of aldehydes and ketones based on mass spectrometric detection is described.Recently developed modular derivatizing agents are employed for analysis. These hydrazine reagents, e.g. 4-dimethylamino-6-(4-methoxy-1-naphthyl )-1,3,5-triazine-2-hydrazine (DMNTH), react with the carbonyl compounds with the formation of the respective hydrazones, which are separated by HPLC-MS with atmospheric pressure chemical ionization in the positive mode.Electrospray ionization may also be used for analysis. Particular focus is directed on various calibration approaches, including external calibration with standard solutions and internal calibration with a hydrazone standard of cyclobutanone, an aldehyde not likely to occur in real samples.A second approach for internal calibration is based on the 13C2-labelled acetaldehyde hydrazone standard. DiVerent calibration approaches may then be used for the analysis of real samples. Limits of detection range from 2×10-8 to 5×10-8 mol L-1 for a series of hydrazones, including hydrazones of saturated aldehydes with alkyl chain lengths from 1 to 7 carbon atoms, and hydrazones of selected unsaturated and aromatic aldehydes as well as ketone hydrazones.The determination of aldehydes and ketones is of great HPLC-MS instrumentation importance in various fields, including workplace monitoring, The HPLC-MS system from Shimadzu (Duisburg, Germany) emissions testing and process control. For the selective and consisted of the following components: controller unit simultaneous determination of a series of aldehydes and SCL-10Avp, degasser DGU-14A, two pumps LC-10ADvp, ketones, hydrazine reagents have been shown to be advantamixing chamber Model SUS (0.5 mL), autosampler SIL-10A, geous in recent years.1–10 They react readily with the carbonyl UV/VIS detector SPD-10AV, single quadrupole mass spec- compounds with the formation of the respective hydrazones, trometer LCMS QP8000 with atmospheric pressure ionization which are typically separated by reversed-phase liquid chromaand software Class 8000 Version 1.01.tography and detected photometrically1–8 or fluorescence spectroscopically. 9,10 Although mass spectrometric detection HPLC conditions appears to be promising to achieve a higher selectivity in combination with low limits of detection, only a few reports All separations were performed using a Discovery C18 column on the mass spectrometric determination of hydrazones have equipped with a guard column of the same material (Supelco, been published.11–13 Ko� lliker et al.12 have recently described Deisenhofen, Germany) with the following dimensions: particle the LC-MS-MS identification of 2,4-dinitrophenylhydrazones size, 5 mm; pore size, 100 A° ; length, 150 mm; id, 2.1 mm.using atmospheric pressure chemical ionization (APCI) in the Eluent A of the mobile phase was a solution of 1380 mL negative mode. Zurek et al.13 have reported the quantitative triethylamine and 557 mL acetic acid in 500 mL water analysis of these substances using standards based on stable (pH#7.5); eluent B was acetonitrile.A binary gradient at a isotopes.13 However, LC-MS publications on aldehyde deter- flow rate of 0.4 mL min-1 with the profile given in Table 1 mination have been limited up to now to the use of was used. The injection volume was 5 mL. UV detection was 2,4-dinitrophenylhydrazine (DNPH) derivatives. carried out at 313 nm. For fluorescence detection, the exci- Kempter et al.14 describe a modular derivatizing agent as a tation wavelength was 332 nm and the emission wavelength possible alternative to DNPH, because this new type of reagent was 395 nm.is more versatile and oVers easy access to fluorescence spectroscopic detection. The present work focuses on the development MS conditions of an HPLC-MS method with APCI and mass spectrometric All MS measurements were recorded using APCI in the detection to achieve higher selectivity of detection for the positive mode under the following conditions: nebulizer gas modular derivatizing agent.flow (N2), 2 L min-1; probe voltage, 4 kV; temperature of the APCI probe, 450 °C; curved desolvation line (CDL) voltage, Experimental -40 V; CDL temperature, 250 °C; deflector voltages, 55 V; detector gain, 1.5 kV.For SCAN mode measurements, a mass Chemicals range from 250 to 450 m/z was chosen; the integration time All chemicals were purchased from Aldrich Chemie (Steinheim, Germany) in the highest quality available. As solvent for LC, acetonitrile gradient grade from Merck (Darmstadt, Germany) Table 1 HPLC profile was used. Time/min 0.03 1 8.5 17 20.5 21.5 23.5 c(CH3CN) (%) 38 38 46 90 90 38 Stop †Presented at AIRMON’99, Geilo, Norway, February 10–14, 1999.J. Environ. Monit., 1999, 1, 307–311 307was 1.2 s. For selected ion monitoring (SIM) measurements, 2×10-8 to 2×10-5 mol L-1 were spiked with the internal standard solution resulting in concentrations of 10-5, 10-6 the integration time was 1 s. and 10-7 mol L-1 of the internal standard.Therefore, mixing ratios of the internal standards compared to the Linear range of the MS detection DMNThydrazones were obtained in the range of 10051 to A calibration curve for 4-dimethylamino-6-(4-methoxy- 152. All solutions were analysed by HPLC-MS using the same 1-naphthyl )-1,3,5-triazine-2-hydrazone (DMNThydrazone) time programme for the SIM traces as for the determination standards in acetonitrile was recorded three times in the range of the linear range of the MS detection.from 2×10-8 to 5×10-5 mol L-1. The mixture contained the The peak areas of the SIM traces were integrated separately. standards of formaldehyde, acetaldehyde, propanal, butanal, The response factors were calculated as the ratio of the peak pentanal, hexanal, heptanal, crotonaldehyde, acetone and p- areas of the DMNThydrazone to the internal standard with tolylaldehyde.The MS detection was carried out in SIM mode consideration of the respective concentration ratio. using the time programme given in Table 2 of the SIM traces. Preparation of the 4-dimethylamino-6-(4-methoxy- Synthesis and characterization of DMNThydrazones 1-naphthyl )-1,3,5-triazine-2-hydrazine (DMNTH) solution for The DMNThydrazones were prepared according to the real sample analysis procedure described in ref. 14. A 3.2×10-3 mol L-1 DMNTH solution was prepared by adding 100 mg DMNTH to 50 mL concentrated sulfuric acid in 10 mL distilled water and 89.95 mL acetonitrile. Sample preparation and air sampling procedure Air sampling was performed using a personal air sampler pump model I.H. (A.P. Buck, Inc., Orlando, FL, USA). The impingers contained 50 mL DMNTH solution. The sampling volume of the sample of a disinfected room was 10.0 L at a Ha Hb OCH3 N N N HN N C R1 R2 N H3C CH3 flow rate of 1.25 L min-1. The impinger was equipped with a backup impinger to control incomplete recovery. 13C2 Acetaldehyde DMNThydrazone. 1H NMR (200 MHz, CDCl3, TMS): d 1.24 (s, 1H, NH), 1.58, 2.06 (d, 3H, Analysis of the sample in workplace air after floor disinfection 13CH–13CH3), 3.27 (s, 6H, N(CH3)2), 4.02 (s, 3H, OCH3), Each sample was analysed in four diVerent ways: (i) SCAN 6.85 (d, 1H, Ar-Hb), 7.20 (m, 1H, NL13CH), 7.46 (m, 2H, mode with external calibration; (ii) SIM mode with addition Ar-H), 8.24 (m, 2H, Ar-H), 9.03 (d, 1H, Ar-Ha); MS (electron of 13C2 acetaldehyde DMNThydrazone as internal standard; impact, EI, 70 eV): m/z 338 (M+, 43%), 322 (M+-13CH3, (iii) UV/VIS detection at 313 nm with external calibration; 57%), 294 (M+-N13C2H4, 17%), 280 (M+-N213C2H4, 20%), and (iv) fluorescence detection (excitation; 332 nm; emission, 209 (10%), 184 (18%), 139 (100%), 96 (18%), 71 (15%), 55 395 nm) with external caSample preparation for (i), (8%); IR (KBr): 3440, 3228, 2998, 2936, 1575, 1529, 1511, (iii) and (iv): 5 mL of the sample solution was directly injected 1464, 1408, 1373, 1339, 1322, 1263, 1244, 1216, 1201, 1185, into the LC system without further dilution. Sample prep- 1160, 1132, 1033, 1019, 914, 878, 811, 771, 730, 724, 709, 653, aration for (ii): 100 mL of the sample was spiked with 100 mL 618, 590, 474 cm-1; analysis calc.for 13C2C16H20N6O: C, of the internal standard (5×10-5 mol L-1) and made up to 64.48%; H, 5.96%; N, 24.84%; found: C, 64.32%; H, 5.88%; 1 mL with acetonitrile. This mixture was analysed with N, 24.70%. HPLC-MS using the same time programme as described above for the determination of the response factor.Cyclobutanone DMNThydrazone. 1H NMR (200 MHz, CDCl3, TMS): d 1.71 (s, 1H, NH), 1.99 (m, 1H, CH2–CH–CH2), 2.86, 2.92 (2 m, 4H, CH2–CH–CH2), 3.27 Results and discussion (s, 6H, N(CH3)2), 4.02 (s, 3H, OCH3), 6.84 (m, 2H, Ar-Hb, The reaction of aldehydes and ketones with the derivatizing NLCH), 7.49 (m, 2H, Ar-H), 8.23 (m, 2H, Ar-H), 8.99 (d, agent DMNTH with the formation of the respective hydrazone 1H, Ar-Ha); MS (EI, 70 eV): m/z 362 (M+, 11%), 333 is presented in Fig. 1. First investigations were carried out (M+-C2H5, 100%), 319 (M+-C4H7, 35%), 184 (16%), 140 regarding the use of APCI and electrospray ionization (ESI) (41%), 96 (18%), 71 (27%), 55 (24%); IR (KBr): 3434, 3279, as interfaces to liquid chromatography. It is obvious from the 3075, 2994, 2925, 1674, 1621, 1564, 1522, 1510, 1470, 1425, structures of both DMNTH and its hydrazones that these 1402, 1324, 1273, 1244, 1204, 1190, 1163, 1125, 1093, 1028, molecules contain several basic, but no acidic functional 912, 809, 772, 729, 714, 663, 618, 471 cm-1; analysis calc.for groups. Therefore, protonation of these substances appeared C20H19N6O: C, 66.28%; H, 6.12%; N, 23.19%; found: C, to be more likely than deprotonation, thus leaving the positive 66.01%; H, 6.37%; N, 23.29%.ionization mode as the more promising approach. Experiments with both interfaces combined with both the positive and Response factors using internal standards negative mode led to the following conclusions: As expected A stock solution of the internal standards 13C2 acetaldehyde from the above considerations, APCI and ESI are equally DMNThydrazone and cyclobutanone DMNThydrazone suitable for the ionization of the derivatives in the positive was prepared at a concentration of 5×10-5 mol L-1 mode, while there is almost no signal obtained in the negative DMNThydrazone standards in the concentration range mode.The proposed ionization mechanism for APCI(+) and ESI(+) is also depicted in Fig. 1. We have selected APCI(+) Table 2 MS detection programme as the favourable ionization technique for all further measurements, as the APCI interface is more compatible with higher Time/min 6.5–10.1 10.1–13.0 13.0–14.9 14.9–19.0 flow rates from the LC system compared to the ESI interface. m/z 323; 337; 351; 363 363; 365 379; 393; This allows easier downscaling of the chromatographic separa- 339; 351; 407; 413 tion from column diameters of 4.6 mm and flow rates of 308 J.Environ. Monit., 1999, 1, 307–311UV/VIS and fluorescence detection have been recorded using a diVerent chromatographic system to that of MS detection. Due to diVerent void volumes, diVerent retention times are observed. The fluorescence detector is connected in series to the UV/VIS detector.Therefore, slightly longer retention times and broader peaks are observed. It should be noted that two isomers are observed for most of the unsymmetrical hydrazones. This increases quantification problems in complex matrices containing several aldehydes. All other chromatograms represent extracts of single masses from the total ion chromatogram (TIC). It is obvious that selectivity is signifi- cantly increased, as even saturated and unsaturated aldehyde hydrazones with the same alkyl chain length may be determined selectively.The two peaks in the chromatogram of the m/z 351 trace represent acetone and propanal, which are very well separated chromatographically. The SIM mode has been used for quantification, as it is more sensitive compared to the TIC extracts. Time programming of the SIM traces has been employed as described in the Experimental section to further improve the limits of detection.Quantification in liquid chromatography by MS detection is typically associated with larger relative standard deviations of the results compared to UV/VIS detection, when external calibration is used. This is due to possible changes in the Fig. 1 Reaction of DMNTH with carbonyls and protonation of the ionization conditions, which may for example be caused by formed hydrazones in LC-MS during atmospheric pressure chemical coelutions of other compounds which may influence ionization. ionization in the positive mode. Changes in the mass spectrometric system, e.g., the vacuum, will also aVect the reproducibility of the system.15,16 We have therefore focused on diVerent calibration 1.5 mL min-1 to column diameters of 2.1 mm and flow rates approaches.External calibration with solutions of hydrazone of 0.4 mL min-1. standards is typically used in combination with UV/VIS and In Fig. 2, the APCI(+) mass spectrum of pentanal fluorescence detection of the derivatives. The use of internal DMNThydrazone is presented.In addition to the (M+H)+ standards is generally more critical in HPLC than in GC, as peak, which is the base peak, almost no fragmentation is the peak capacity of a typical chromatogram is significantly observed under the selected ionization conditions. If desired, lower, and it is more diYcult to separate chromatographically some structural information may be obtained by the use of the internal standard from the analytes.To avoid these prob- cone fragmentation, but mild ionization conditions were seleclems, we have employed two diVerent techniques of internal ted in this work to allow easy quantification of unfragmented standardization. First, a carbonyl hydrazone, which is not (M+H)+ peaks. likely to occur in real samples, has been synthesized.We have Chromatograms of DMNTH and a series of the respective selected cyclobutanone DMNThydrazone, as it elutes signifi- hydrazones are presented in Fig. 3. A chromatogram with cantly in advance of the other C4-carbonyl hydrazones and it UV/VIS detection at a wavelength of 313 nm and a chromatois characterized by the identical mass as the a,b-unsaturated gram with fluorescence detection (excitation wavelength, C4-DMNThydrazone. Using these precautions, the coelution 332 nm; emission wavelength, 395 nm) prove that low selecof the internal standard with a hydrazone of identical mass tivity with both detection techniques will be obtained in the case of complex real samples.The chromatograms with can be excluded. Second, the 13C2-labelled acetaldehyde Fig. 2 APCI mass spectrum (positive mode) of pentanal DMNThydrazone. J. Environ. Monit., 1999, 1, 307–311 309Fig. 3 Chromatograms of the separation of DMNTH and a series of DMNThydrazones using UV/VIS detection (first line), fluorescence detection (second line) and APCI(+)-MS detection at selected masses. The individual mass traces have been recorded as extracts of the TIC.Concentration of the hydrazones: 2.3×10-5 mol L-1. Concentration of DMNTH: 1.1×10-4 mol L-1. 1, DMNTH; DMNThydrazones of: 2, formaldehyde; 3, acetaldehyde; 4, 13C2 acetaldehyde; 5, acetone; 6, propanal; 7, cyclobutanone; 8, crotonaldehyde; 9, butanal; 10, pentanal; 11, hexanal; 12, p-tolualdehyde; 13, heptanal. DMNThydrazone has been synthesized. It coelutes with the Calibration curves have been recorded for a series of hydrazones using the separation and mass spectrometric con- non-labelled acetaldehyde DMNThydrazone, but is discriminated by its mass.Due to coelution, identical ionization ditions stated above. The calibration functions for four representative hydrazones are provided in Fig. 4. It is obvious conditions for both internal standard and analyte should yield excellent recoveries.In addition, the isotope-labelled standard that similar, although not identical, limits of detection are obtained for the hydrazones. Table 3 summarizes the limits of may also be employed as classical internal standard as in the case of the cyclobutanone hydrazone. detection determined as a signal-to-noise ratio 351. Limits Fig. 4 Calibration function for selected DMNThydrazones. 310 J. Environ. Monit., 1999, 1, 307–311Table 3 Limits of detection (LOD) for HPLC-MS determination of four ways: UV/VIS detection (3.8 mg L-1), fluorescence detecselected DMNThydrazones tion (4.2 mg L-1), MS detection with external calibration (3.4 mg L-1) and MS detection with internal calibration DMNThydrazone of LOD (APCI-MS)/mol L-1 (4.2 mg L-1).While the data obtained by fluorescence detection and MS detection with internal calibration correlate well, Formaldehyde 5×10-8 Acetaldehyde 2×10-8 lower values are observed for UV/VIS detection and MS Propanal 5×10-8 detection with external calibration. This may be due to elution Butanal 2×10-8 of the formaldehyde hydrazone peak on the tailing of the Pentanal 2×10-8 excess reagent peak in the case of UV/VIS detection and due Hexanal 2×10-8 to limited reproducibility in the case of MS detection with Heptanal 2×10-8 external calibration.13 Acetone 2×10-8 Crotonaldehyde 5×10-8 The mass spectrometric detection of DMNThydrazones with p-Tolualdehyde 2×10-8 diVerent calibration techniques is therefore a promising alternative to UV/VIS and fluorescence detection approaches. Table 4 Response factors for internal calibration using the DMNThydrazones of 13C2-acetaldehyde and cyclobutanone Acknowledgements DMNThydrazone of 13C2 acetaldehyde Cyclobutanone Financial support by the Deutsche Forschungsgemeinschaft (DFG) under project numbers Ka1093/2–1 and Ka1093/2–2 Formaldehyde 0.55 0.79 is gratefully acknowledged.C. K.thanks the Deutsche Acetaldehyde 1.03 1.52 Bundesstiftung Umwelt for a scholarship, G. Z. thanks the Propanal 0.85 1.17 Stiftung der Deutschen Wirtschaft for a scholarship. Butanal 0.86 1.14 Pentanal 1.07 1.48 Hexanal 0.99 1.35 References Heptanal 1.07 1.43 Acetone 0.88 1.26 1 R. H. Beasley, C. E. HoVmann, M. L. Rueppel and J. W. Worley, Crotonaldehyde 0.75 1.00 Anal. Chem., 1980, 52, 1110.p-Tolualdehyde 1.06 1.47 2 K. Kuwata, M. Uebori and H. J. Yamasaki, J. Chromatogr. Sci., 1979, 17, 264. 3 D. Grosjean and K. Fung, Anal. Chem., 1982, 54, 1221. of quantification are three times higher. Linearity was observed 4 W.Po� tter and U. Karst, Anal. Chem., 1996, 68, 3354. 5 J.-O. Levin, K. Andersson, R. Lindahl and C. A. Nilsson, Anal. up to 5×10-5 mol L-1, being limited by the solubility of the Chem., 1985, 57, 1032.substances. 6 R. R. Miksch, D. W. Anthon, L. Z. Fanning, C. D. Hollowell, For the use of internal standards, response factors have K. Revzan and J. Glanville, Anal. Chem., 1981, 53, 2118. been calculated for all investigated hydrazones based on the 7 I. Ahonen, E. Priha and M.-L. A� ija�la�, Chemosphere, 1984, 13, 521. two internal standard compounds.The response factors are 8 C. H. Risner and P. J. Martin, Chromatogr. Sci., 1994, 32, 76. listed in Table 4. All response factors may be applied for the 9 W. Schmied, M. Przewosnik and K. Ba�chmann, Fresenius’ Z. Anal. Chem., 1989, 335, 464. complete concentration range mentioned above. It should be 10 D. R. Rodier, L. Nondek and J. W. Birks, Environ.Sci. Technol., noted, however, that the concentrations of the internal stan- 1993, 27, 2814. dards should optimally be in the same concentration range as 11 K. L. Olson and S. J. Swarin, J. Chromatogr., 1985, 333, 337. the analyte concentrations. Best results are typically obtained 12 S. Ko� lliker, M. Oehme and C. Dye, Anal. Chem., 1998, 70, 1979. when the analyte and standard concentrations do not diVer 13 G. Zurek, H. Luftmann and U. Karst, 1999, submitted for by more than a factor of 5. The internal standard concentration publication. 14 C. Kempter, W. Po� tter, N. Binding, H. Kla�ning, U. Witting and should not be too low, especially in the case of low analyte U. Karst, 1999, submitted for publication. concentrations, as deviations caused by the integration of 15 R. Willoughby, E. Sheehan and S. Mitrovich, A Global View of small peaks for both analyte and standard may end up with LC/MS, Pittsburgh, Global View Publications, 1998, p. 351. very high total relative standard deviations (RSDs). 16 R. Baldwin, R. A. Bethem, R. K. Boyol, W. L. Budde, T. Cairns, A real air sample taken immediately after floor disinfection R. D. Gibbens, J. D. Henion, M. A. Kaiser, D. L. Lewis, of a small room with poor ventilation with aldehyde-containing J. E. Matusik, J. A. Sphon, R. W. Stephany and R. K. Trubey, J. Am. Soc. Mass Spectrom., 1997, 8, 1180. disinfectants was obtained using impingers. Significant formaldehyde concentrations were detected in the air sample. No breakthrough of the analyte into the backup impinger was observed. The formaldehyde concentration was determined in Paper 9/02766A J. Environ. Monit., 1999, 1, 3