Direct Quantitative Analysis of Organic Compounds in the Gas and

Feb 15, 2003 - transferred into the vaporizer unit of the ion source via a deactivated fused-silica capillary by using the sheath liquid syringe pump,...
14 downloads 0 Views 196KB Size
Download PDF
Anal. Chem. 2003, 75, 1410-1417

Direct Quantitative Analysis of Organic Compounds in the Gas and Particle Phase Using a Modified Atmospheric Pressure Chemical Ionization Source in Combination with Ion Trap Mass Spectrometry Bettina Warscheid, Ulrich Ku 1 ckelmann, and Thorsten Hoffmann*

Institute of Spectrochemistry and Applied Spectroscopy, ISAS, P.O.Box 101352, 44013 Dortmund, Germany

A slightly modified atmospheric pressure chemical ionization source is employed for direct quantitative analysis of volatile or semivolatile organic compounds in air. The method described here is based on the direct introduction of an analyte in the gas or particle phase, or both, into the ion source of a commercial ion trap mass spectrometer. For quantitation, a standard solution is directly transferred into the vaporizer unit of the ion source via a deactivated fused-silica capillary by using the sheath liquid syringe pump, which is part of the mass spectrometer. The standard addition procedure is conducted by varying the pump rate of a diluted solution of the standard compound in methanol/water. A N2 sheath gas flow is applied for optimal vaporization and mixing with the analyte gas stream. By performing detailed reagent ion monitoring experiments, it is shown that the relative signal intensity of [M + H]+ ions is dependent on the relative humidity of the analyte gas stream as well as the composition and concentration of CI reagent ions. The method is validated by a comparison of the standard addition results with a calibration test gas of known concentration. To demonstrate the potential of atmospheric pressure chemical ionization mass spectrometry as a quantitative analytical technique for on-line investigations, a tropospherically relevant reaction is carried out in a 493-L reaction chamber at atmospheric pressure and 296 K in synthetic air at 50% relative humidity. Finally, the applicability of the technique to rapidly differentiate between analytes in the gas and particle phase is demonstrated. Gaseous organic compounds represent key species for the understanding and control of a variety of technical and environmental processes. Areas such as industrial process analysis and medical or environmental research rely on the development of analytical techniques to monitor trace amounts of organic species in complex matrixes with a high time resolution. In this context, environmental chemistry is an especially challenging field, since hundreds of different organic compounds are emitted into the atmosphere by anthropogenic or natural sources, which directly * Corresponding author. E-mail: [email protected].

1410 Analytical Chemistry, Vol. 75, No. 6, March 15, 2003

affect air quality and impact the tropospheric oxidation capacity. Furthermore, these compounds also go through several oxidation processessa circumstance that even enhances the complexity of the analytical problem. For example, the reaction of volatile organic compounds (VOCs) with atmospheric oxidizing species (ozone, OH radicals) leads to the formation of less volatile products that can undergo a phase transition (gas-to-particle conversion).1-4 The necessity to differentiate between concentrations of the analytes in the gas and particle phase, e.g., for toxicologic or climate related studies, adds further to the analytical challenge. In general, off-line techniques are applied for the identification and quantitation of organic compounds either in the gas or in the particle phase, employing preconcentration steps using cryotraps,5 cartridges,6-9 or denuder/filter packs10-12 followed by thermal desorption or solvent extraction and chromatographic separation using GC or LC. However, the main drawbacks of applying offline techniques are high time consumption per analysis, and an enhanced risk of positive and negative artifacts by conducting a multistep analytical procedure.13,14 Therefore, several alternative analytical approaches have been used to improve the analysis of airborne organic compounds. For example, in situ long-path FTIR spectroscopy was applied for the quantitative determination of (1) Atkinson, R. J. Phys. Chem. Ref. Data 1997, 26, 215. (2) Jenkin, M. E.; Clemitshaw, K. C. Atmos. Environ. 2000, 34, 2499. (3) Andreae, M. O.; Crutzen, P. Science 1997, 276, 1052. (4) Hoffmann, T.; Bandur. R.; Marggraf, U.; Linscheid, M. J. Geophys. Res. 1998, 103, 25569. (5) Yokouchi, Y.; Ambe Y. Atmos. Environ. 1985, 19, 1271. (6) Hakola, H.; Arey, J.; Aschmann, S. A.; Atkinson, R. J. Atmos. Chem. 1994, 18, 75. (7) Grosjean, D.; Williams, E. L.; Seinfeld, J. H. Environ. Sci. Technol. 1992, 26, 1526. (8) Kotzias, D.; Nicollin, B.; Duane, M.; Daiber, R.; Eijk, J. V.; Rogora, L.; Schlitt, H. Naturwissenschaften 1991, 78, 38. (9) Calogirou, A.; Kotzias, D.; Kettrup, A. Atmos. Environ. 1996, 31, 283. (10) Yu, J.; Cocker, D. R., III; Griffin, R. J.; Flagan, R. C.; Seinfeld, J. H. J. Atmos. Chem. 1999, 34, 207. (11) Glasius, M.; Lahaniati, M.; Calogirou, A.; Di Bella, D.; Jensen, N. R.; Hjorth, J.; Kotzias, D.; Larsen, B. R. Environ. Sci. Technol. 2000, 34, 1001. (12) Glasius, M.; Duane, M.; Larsen, B. R. J. Chromatogr., A 1999, 833, 121. (13) Eatough, D. J.; Wadsworth, A.; Eatough, D. A.; Crawford, J. W.; Hansen, L. D.; Lewis, E. A. Atmos. Environ. 1993, 27A, 1213. (14) Eatough, D. J.; Eathough, D. A.; Lewis, L.; Lewis, E. A. J. Geophys. Res. 1996, 101, 19515. 10.1021/ac025788d CCC: $25.00

© 2003 American Chemical Society Published on Web 02/15/2003

gas-phase organic compounds in environmental chamber studies.15,16 Carbonyl products are especially appropriate analytes due to their strong CdO absorption bands. However, since multiple carbonyl compounds can be present, the carbonyl stretching region is not characteristic to specifically quantify a single compound. Furthermore, reliable FT-IR measurements are demanding in the presence of higher relative humidities. Besides infrared spectroscopy, mass spectrometry can be utilized as an on-line technique to obtain information about airborne organics in real time. A selective ionization technique can be applied to avoid a chromatographic separation step. Protontransfer reaction mass spectrometry (PTR-MS) has been successfully implemented for in situ on-line monitoring of airborne VOCs down to the 10 ppt (v/v) range,17-22 usually in combination with quadrupole-MS. Here, H3O+ ions are produced in a separate ion source, which subsequently protonates analytes with a higher proton affinity than water in a selected ion flow tube (SIFT). A variant of SIFT-MS, flowing afterglow mass spectrometry (FAMS), was recently applied for on-line determination of the deuterium abundance in breath water.23 Resonance-enhanced multiphoton ionization (REMPI), which often has been combined with time-of-flight MS, represents an entirely different ionization method for real-time studies. REMPI was shown to be an attractive tool particularly for the analysis of airborne organics,24-26 due to high selectivity and, for a number of species, high sensitivity. The mass spectrometric techniques mentioned above, such as PTR-MS, SIFT-MS, or REMPI-TOF, all utilize fairly specific instrumentation, which is usually not available in standard laboratories. In addition, the target analytes have to be known, and therefore, the measurement of unidentified compounds, e.g., in reaction mixtures, is not straightforward. In contrast, the analytical technique presented here is based on the application of a benchtop MS instrument equipped with a slightly modified commercial APCI source to enable on-line monitoring of organics. Enhanced selectivity for the investigation of complex organic mixtures is provided by MSn experiments in an ion trap mass spectrometer. In previous papers, we have demonstrated the potential of on-line APCI ion trap mass spectrometry as a tool for structure elucidation of low volatile oxidation products directly from complex organic matrixes.27,28 Recently, a similar approach (15) Winterhalter, R.; Neeb, P.; Grossmann, D.; Kolhoff, A.; Horie, O.; Moortgat, G. K. J. Atmos. Chem. 2000, 35, 165. (16) Orlando, J. J.; Noziere, B.; Tyndall, G. S.; Orzechowska, G. E.; Paulson, S. E.; Rudich, Y. J. Geophys. Res. 2000, 105, 11561. (17) Lindinger, W.; Hansel, A.; Jordan, A. Int. J. Mass Spectrom. Ion Processes 1998, 173, 191. (18) Lindinger, W.; Hansel, A.; Jordan, A. Chem. Soc. Rev. 1998, 27, 347. (19) Reiner, T.; Mo ¨hler, O.; Arnold, F. J. Geophys. Res. 1998, 103, 31309. (20) Crutzen, P. J.; Williams, J.; Poeschl, U.; Hoor, P.; Fischer, H.; Warneke, C.; Holzinger, R.; Hansel, A.; Lindinger, W.; Scheeren, B.; Lelieveld, J. Atmos. Environ. 2000, 34, 1161. (21) Karl, T.; Fall, R.; Jordan, A.; Lindinger, W. Environ. Sci. Technol. 2001, 35, 2926. (22) DeGauw, J.; Howard, C. J.; Custer, T. G.; Baker, B. M.; Field, R. Environ. Sci. Technol. 2000, 34, 2640. (23) Hue, N.; Serani, L.; Laprevote, O. Rapid Commun. Mass Spectrom. 2001, 15, 203. (24) Zimmermann, R.; Herger, H. J.; Kettrup, A. Fresenius J. Anal. Chem. 1999, 363, 720. (25) Schmidt, S.; Appel, M. F.; Garnica, R. M.; Schindler, R. N.; Benter, T. Anal. Chem. 1999, 71, 3721. (26) Boesl, U. J. Mass Spectrom. 2000, 35, 289. (27) Warscheid, B.; Hoffmann, T. Rapid Commun. Mass Spectrom. 2001, 15, 2259.

was presented by Cooks and co-workers focusing on semivolatile organics, such as methyl salicylate and dimethyl methylphosphonate.29 The present work is focused on the development of a highspeed quantitative method by using on-line APCI-MS. The occurrence of matrix effects has to be taken into account, even though APCI is less susceptible to matrix effects for quantitative work than electrospray ionization (ESI) as reported in off-line studies.30-32 In this work, matrix effects are examined at different ion source conditions, as no chromatographic separation step is employed and the entire reaction mixture is introduced directly into the APCI source. The method presented here is directed to the rapid quantitative analysis of semivolatile organic compounds present in the gas and aerosol phases. Therefore, a standard addition method has been developed using a diluted standard compound solution. The standard solution is directly transferred into the APCI source at certain flow rates by the built-in syringe pump of the instrument and mixed to the gaseous reaction mixture before ionization takes place. The performance of the quantitation procedure is demonstrated for the direct determination of a C9-ketone (sabinaketone) derived from the gas-phase ozonolysis of a monoterpene (sabinene) at dry (e1% relative humidity) and humid (50% relative humidity) reaction conditions. In addition, the partitioning of sabinaketone between the gas and particle phases was estimated using a charcoal denuder system prior to analysis, which separates gas-phase components from the secondary organic aerosol (SOA) formed from the biogenic hydrocarbon/ozone reaction in the gas phase. The objective of this work is to demonstrate the potential and usefulness of on-line APCI-MS for reaction monitoring, mechanistic related studies, and rapid product quantitation in particular. EXPERIMENTAL SECTION On-Line APCI-MS Operating Conditions. All data were acquired using a Finnigan LCQ ion trap mass spectrometer (LCQ Classic, Thermo Finnigan, San Jose, CA) operating in the positive ion mode. The instrument was used with the original APCI source, just the gas and analyte inlet lines were modified. A sketch of the adapted APCI source used for qualitative and quantitative investigations is shown in Figure 1. For qualitative investigations, the N2 sheath gas was adjusted to 3.1 L min-1, whereas for quantitative measurements the N2 sheath gas flow was adjusted in the range of 3-10.7 L min-1. The following APCI source parameters were adjusted to ensure optimal vaporization and ionization conditions for analytes in air and standards in the liquid phase: vaporizer temperature 450 °C, corona discharge voltage 3 kV, and plasma current 3 µA. Capillary voltage and capillary temperature were adjusted to 6 V and 200 °C, respectively. In some experiments, transfer octapole collisioninduced dissociation (CID) took place on [2M + H]+ cluster ions of nopinone at 12-V collision energy. (28) Warscheid, B.; Hoffmann, T. Rapid Commun. Mass Spectrom. 2002, 16, 496. (29) Charles, L.; Riter, L. S.; Cooks, R. G. Anal. Chem. 2001, 73, 5061. (30) Jewett, B. N.; Ramaley, L.; Kwak, J. C. T. J. Am. Soc. Mass Spectrom. 1999, 10, 529. (31) Henion, J. D.; Thomson, B. A.; Dawson, P. H. Anal. Chem. 1982, 54, 451. (32) Blount, B. C.; Milgram, K. E.; Silva, M. J.; Malek, N. A.; Reidy, J. A.; Needham, L. L.; Brock, J. W. Anal. Chem. 2000, 72, 4127.

Analytical Chemistry, Vol. 75, No. 6, March 15, 2003

1411

Figure 1. Sketch of the experimental setup for direct quantitative analysis of organic compounds in the gas and particle phase.

Quantitation Procedure. To demonstrate the quantitation capabilities of the method, nopinone (C9H14O; MW ) 138), the main oxidation product from the tropospheric oxidation of β-pinene, was chosen as a model compound. A calibration gas mixture of this compound was directly transferred from a dynamic test gas generator, which was based on an open tube diffusion technique, into the ion source via a synthetic air gas stream (20.5% O2, 79.5% N2). The overall output of the nopinone standard was determined by weighing and resulted in a nopinone mixing ratio of 105 ( 2 ppb (v/v) under the conditions applied. The built-in syringe pump of the commercial mass spectrometer was used to deliver the sheath liquid via a deactivated fusedsilica capillary into the vaporizer unit of the APCI source. Both a water/methanol (50:50, v/v) solution and pure water were applied as sheath liquids. For the standard addition quantification procedure, a water/methanol (50:50, v/v) solution of nopinone was introduced into the ion source via the sheath liquid inlet line. The concentration of the standard was varied by adjusting different flow rates ranging from 6 to 20 µL min-1. By using a standard solution of 0.0965 µg µL-1 nopinone, the mass flow rate of nopinone into the ion source could be adjusted between 0.58 and 1.93 µg min-1. On-Line Investigations of Oxidation Products. Experiments were performed in a 493-L cylindrical reaction chamber made of glass at a temperature of 296 ( 2 K. Prior to each experiment, the reaction vessel was flushed with dry synthetic air (20.5% O2, 79.5% N2). The overall flow through the reaction chamber was adjusted to 7.82 L min-1. To investigate the formation of the semivolatile compound sabinaketone, which represents the main oxidation product from the ozonolysis of sabinene,6 typically, ozone was added first into the reaction vessel. Ozone was produced by UV irradiation of another synthetic air supply and measured with a Dasibi Environmental Corp. O3 analyzer (model 1008-RS, Glendale, CA). When a constant ozone mixing ratio of 300 ppb (v/v) was reached in the chamber, the biogenic hydrocarbon was added from a second dynamic test gas generator into the reaction chamber. The hydrocarbon output of the test gas generator was adjusted to 12.18 µg min-1, resulting in a steady-state hydrocarbon mixing ratio of ∼280 ppb (v/v) in the reaction chamber in the absence of ozone. Two sets of experiments were performed: (1) the investigation of the ozonolysis reaction at dry conditions (e 1% relative humidity) and (2) at a relative humidity of 50%. The relative humidity in the reaction chamber was measured with a humidity sensor (KPC 1/1-115). 1412

Analytical Chemistry, Vol. 75, No. 6, March 15, 2003

In general, the reaction chamber was operated by continuously introducing the reactants into the chamber, hence representing a continuous stirred tank reactor system. The reaction chamber was directly connected to the slightly modified APCI source of the mass spectrometer by 1.5-m Teflon tubing (i.d. 5 mm). The sample flow into the APCI source was adjusted between 0.45 and 1.5 L min-1 and mixed with a N2-sheath gas flow of 9.7-10.7 L min-1 by reaching the vaporizer unit of the ion source. A second pump stage was installed at the APCI source to ensure a constant gas flow from the reaction chamber through the ion source. Quantitation of sabinaketone was performed at steady-state conditions ∼5 h after starting the ozonolysis reaction as described above. Chemicals. The monoterpene, sabinene, and the ozonolysis product, nopinone, were purchased from Sigma Aldrich at a purity of 99 and 98%, respectively. Both compounds were of highest purity commercially available and were used without further purification. Since the oxidation product sabinaketone was not commercially available, nopinone was applied for the standard addition procedure performed here. RESULTS AND DISCUSSION Nopinone in the Gas Phase. Figure 2 shows the on-line APCI-MS spectrum of nopinone (C9H14O) as a gaseous standard compound. As can be seen in the figure, ions at m/z 139, 121, and 277 were formed, which are proposed to correspond to [M + H]+ ions, [M + H - H2O]+ fragment ions, and [2M + H]+ ions of nopinone. The C9H13+ ions (m/z 121) can either be formed by spontaneous fragmentation of [M + H]+ or by collision-induced fragmentation under the conditions present in the atmospheric pressure ion source, since loss of H2O (18 Da) also represents the most efficient fragmentation pathway of protonated nopinone molecules as a result of low-energy CID (see MS/MS spectrum insets in Figure 2). In addition, clustering reactions are well-known processes for chemical ionization at atmospheric pressure.33-39 In accordance, CID of the proposed dimer ions at m/z 277 yielded (33) Covey, T. R.; Lee, E. D.; Bruins, A. P.; Henion, J. D. Anal. Chem. 1986, 58, 1451A. (34) Kambara, H.; Mitsiu, Y.; Kanomata, I. Anal. Chem. 1979, 51, 1447. (35) Kambara, H.; Kanomata, I. Anal. Chem. 1977, 49, 270. (36) Kambara, H.; Kanomata, I. Int. J. Mass Spectrom. Ion Phys. 1977, 25, 129. (37) Herrmans, C. E. M.; van der Hoeven, R. A. M.; Niessen, W. M. A.; Tjaden, U. R.; Van der Greef, U. R. J. Org. Mass Spectrom. 1989, 24, 109. (38) Ma¨rk, T. D. Int. J. Mass Spectrom. Ion Processes 1987, 79, 1. (39) Scott,. A. D., Jr.; Hunter, E. J.; Ketkar, S. N. Anal. Chem. 1998, 70, 1802.

Figure 2. MS and MS2 spectra of nopinone ([M + H]+ ) m/z 139) continuously introduced into the APCI source of an ion trap mass spectrometer. Ions observed at m/z 121 and 277 represent [M + H - H2O]+ fragment ions and [2M + H]+ cluster ions of nopinone, respectively.

Figure 3. Relative signal intensities of (I) protonated nopinone (m/z 139) and (II) proton-bound nopinone dimers (m/z 277). To consider the entire quantity of nopinone, the signal intensities for (III) were calculated according to eq 1.

the exclusive formation of protonated monomers of nopinone at m/z 139 (not shown). However, the formation of proton-bound dimers of nopinone (C18H29O2+) was observed to not be linearly dependent on the adjusted concentration of nopinone (C9H14O), which was varied between a mass flux of 0.7 and 2.2 µg min-1 according to a mixing ratio of 124 to 390 ppb (v/v) during these experiments (Figure 3). It was observed that the dimer-to-monomer ratio of protonated nopinone increased with an increase of the amount of nopinone introduced into the APCI source. Therefore, the relative signal intensities of nopinone (Inopinone) were calculated by using both protonated monomer and proton-bound dimer ion signal intensities:

Inopinone ) Im/z139 + 2(Im/z277)

(1)

Since each dimer comprises two molecules of nopinone, the signal intensity of ions at m/z 277 was multiplied by a factor of 2 and added to the signal intensity of monomer ions at m/z 139 to account for the quantity of nopinone at each concentration adjusted here. As a result, the coefficient of determination was enhanced from R2 ) 0.9979 to R2 ) 0.9995.

Matrix Effects. Chemical ionization processes at atmospheric pressuresand hence the signal intensities of the target analytess depend on the nature and concentration of reactant ions generated by corona discharge. For example, the water content of a gaseous sample influences not only the abundance of the ionizing protonated water clusters but also the water cluster size distribution (H3O+ × (H2O)n). The latter in turn influences the proton affinity of the cluster ions and, therefore, the proton-transfer reaction to the analyte itself. In addition, even small concentration changes of compounds with high gas-phase basicities (e.g., certain nitrogencontaining organics) can strongly influence the concentration of reactant ions, again affecting the analyte signal. As a result, the target ion signal intensity is not necessarily directly proportional to its concentration in the gas phase, if the sample composition is changing. To illustrate the effect of these various factors, it is worthwhile to examine certain fundamentals of chemical ionization at atmospheric pressure. Figure 4a shows traces of different reactant ions produced at different conditions in the APCI source (positive ion mode). Figure 4b displays the relative signal abundance of ions between m/z 20 and 100 when dry synthetic air (relative humidity e1%; 20.5% O2, 79.5% N2) was used as the carrier gas, N2 was used as sheath gas, and no analyte was added. The spectrum is dominated by m/z 37 (H+(H2O)2), which probably represents the most important reactant ions. Under the conditions chosen, the signal with a lower relative abundance at m/z 46 is assumed to represent N2•+(H2O) primary ions40,41 (Figure 3b), since its relative abundance shows a characteristic humidity dependence (data not shown). Therefore, other possible alternative ion compositions, such as NO2+, appear less likely. When at t ) 1 min, 105 ( 2 ppb (v/v) nopinone was continuously added to the carrier gas stream, the relative abundance of both primary ions (H+(H2O)2 and N2•+(H2O)) decreased by ∼25%, certainly as a result of losses of the primary ions due to their reaction with the carbonyl compound. However, when 2 µL min-1 of a water/methanol solution (50: 50 v/v) was introduced into the ion source at t ) 2 min, the composition and abundance of the reactant ions were completely altered. As can be seen in Figure 4a, a strong depletion of both H+(H2O)2 ions and N2•+(H2O) ions took place, and at the same (40) Pavlik, M.; Skalny, J. D. Rapid Commun. Mass Spectrom. 1997, 11, 1757. (41) Kebarle, P. J. Am. Soc. Mass Spectrom. 1992, 3, 1.

Analytical Chemistry, Vol. 75, No. 6, March 15, 2003

1413

Figure 4. (a) Mass traces of cluster ions and protonated nopinone (105 ( 2 ppb (v/v)) observed at different conditions in the APCI source: dry synthetic air with a relative humidity of e1% (t ) 0-1 min), dry synthetic air plus nopinone (t ) 1-2 min), dry synthetic air plus nopinone plus a water/methanol (50:50, v/v) solution introduced with different flow rates from 2 to 18 µL min-1 into the ion source (t ) 2-11 min). (b) and (c) Sections of mass spectra observed at (b) t ) 0.5 min and (c) t ) 10 min, respectively.

time, the formation of H+(CH3OH) and H+(CH3OH)2 cluster ions was highly favored. Ions observed at m/z 47 in the mass spectrum can be rationalized by the formation of H+(N2)(H2O) cluster ions. Here, an alternative explanation would be the loss of water from the methanol dimer. However, in this case, a better correlation with m/z 65 would be expected. By the stepwise increase of the sheath liquid flow rate from 4 to 18 µL min-1, the signal intensities of most ions, reactant and analyte ions, were essentially constant (for details of the analyte signal intensity at small sheath liquid flow rates, see below). Only H+(CH3OH)2 cluster ions were roughly doubled, reaching about the same intensities as the protonated methanol monomers at the end of the experiment (Figure 4a and c). Interestingly, a reduced formation of the proton-bound nopinone dimer was observed with increasing methanol concentrations (data not shown). One explanation for this decrease would be an enhanced dissociation of analyte dimers as a result of a higher collision rate with methanol molecules in the APCI source. On the basis of these experiments, we conclude that the use of a sheath liquid is highly beneficial for the efficiency of the proton-transfer reaction. However, more important than the composition and abundance of reactant ions is the quantitative behavior of the analyte signal as a result of a changing matrix composition. Therefore, the following discussion concentrates on the overall signal intensities of nopinone, present as protonated monomer and proton-bound dimer, at different ion source and experimental conditions. Figure 5 shows the development of the relative signal intensities of nopinone at different sheath liquid (H2O/CH3OH; 50:50, v/v) flow rates in the range of 0-18 µL min-1 by using dry synthetic air (relative humidity of e1%) as carrier gas for the 1414 Analytical Chemistry, Vol. 75, No. 6, March 15, 2003

Figure 5. Signal intensities of protonated nopinone, which were calculated according to eq 1, as a function of the sheath liquid flow rate between 0 and 18 µL min-1 by using a dry carrier gas stream (relative humidity e1%).

carbonyl compound. As can already be expected from the short discussion above, the altered composition and abundance of primary reactant ions by the addition of sheath liquid strongly influences the signal intensities of the analyte at small sheath liquid flow rates, when the conditions abruptly changed from dry (sheath liquid flow zero) to conditions where water was present. Evidently, when the same experiment was performed using humidified synthetic air as the carrier gas for the volatile analyte, the effect was less pronounced. However, the behavior of the analyte signal intensity at flow rates below 4 µL min-1 demonstrates that changes in the humidity of the gas phase to be investigated have to be considered for on-line quanitation procedures using APCI. Nevertheless, since the signal intensity levels

Figure 6. Signal intensities of protonated nopinone (calculated according to eq 1) as a function of the nopinone mass flux into the ion source for two different sheath gas flows. Standard addition was performed by introducing a liquid solution of 0.059 µg µL-1 nopinone in methanol/water (50:50, v/v) at flow rates ranging from 0 to 20 µL min-1 into the APCI source. The carrier gas stream was adjusted to 3.6 L min-1 and the N2 sheath gas flow to either 3 or 9 L min-1.

off at higher flow rates (see Figure 5), the addition of a supplementary sheath liquid flow larger than 4 µL min-1 can overcome the influence of humidity on the analyte signal intensity. These results are in accordance with previous studies, which reported an increased sensitivity and the restraint of matrix effects due to different water contents in air samples by addition of water vapor.42-44 Nopinone in the Liquid Phase. Nopinone was added to the sheath liquid solution (H2O/CH3OH; 50:50, v/v) to perform standard addition. Since the APCI source used here was developed as an interface for LC/MS, the vaporizer unit was designed for the nebulization and evaporation of solvents introduced at flow rates in the lower milliliter per minute range. Therefore, we had to make sure that the standard solution was accurately transferred to the gas phase at the microliter-range liquid flow rates applied here. It was found that sheath gas flow rates between 3 and 10 L min-1 were appropriate to ensure reliable conditions for both desolvation and evaporation of nopinone present in the sheath liquid solution. For reasons of simplicity, the concentration of the standard, which was directly introduced into the vaporizer unit by using a deactivated fused-silica capillary, was changed by varying the sheath liquid flow rates between 5 and 20 µL min-1. Figure 6 shows the signal intensities of protonated nopinone (calculated according to eq 1) displayed as a function of the nopinone mass flux into the ion source. The carrier gas stream (dry synthetic air) was adjusted to 3.6 L min-1, while the N2 sheath gas flow was either adjusted to 3 or 9 L min-1. The relative signal intensities of the analyte deviate from a linear behavior at mass fluxes higher than ∼0.7 µg min-1 at the lower sheath gas flow of 3 L min-1, as is obvious from Figure 6. This limitation of the upper value of the linear dynamic range relates to the ratio of reactant ions to analyte ions, which at last yields the saturation of the ion (42) Taylor, A. J.; Linforth, R. S.; Harvey, B. A.; Blake, A. Food Chem. 2000, 71, 327. (43) Sunner, J.; Nicol, G.; Kebarle, P. Anal. Chem. 1988, 60, 1300. (44) Zehentbauer, G.; Krick, T.; Reineccius, G. A. J. Agric. Food Chem. 2000, 48, 5389.

Figure 7. Direct quantitative determination of gaseous nopinone generated by a test gas source by performing standard additions using the same compound in the liquid phase.

current at analyte concentrations in the low-ppm (v/v) range.45 However, by increasing the sheath gas flow, an appropriate dilution of higher concentrated samples is possible. Then, nopinone addition into the ion source shows a linear relationship between signal intensity and mass flux up to values greater than 1 µg min-1. Based on the statistical evaluation of the calibration curve46 for a sheath gas flow of 9 L min-1, the limit of detection (corresponding to 3.29σ of the signal intensity of the blank) and limit of determination (corresponding to 10σ of the signal intensity of the blank) were estimated to be about 100 and 300 ppt (v/v), respectively. Figure 7 shows the direct quantitative determination of nopinone present in the gas phase from the dynamic test gas generator by performing standard addition using the same compound as standard compound in the liquid phase. Before each standard addition, a pure H2O/CH3OH sheath liquid solution (50:50, v/v) at a flow rate of 6 µL min-1 was introduced into the ion source to counteract the influence of the additional sheath liquid on the nopinone signal intensities discussed above (see Figure 5). Based on the results shown in Figure 7, a nopinone concentration of the gaseous nopinone reference from the test gas generator of 99 ( 4 ppb (v/v) was determined. This value is in good agreement with the value of 105 ( 2 ppb (v/v) determined by gravimetric measurements. The slightly smaller concentration determined by the standard addition procedure might be explained by some losses of the semivolatile C9-carbonyl on its way from the dynamic test gas generator to the ion source within the 1.5-m Teflon tubing. Example Quantification of a Semivolatile Reaction Product. An important advantage of APCI-MS applied as an on-line technique is its high time resolution of ∼1 s. This analytical technique can therefore provide valuable information on the product distribution even under rapidly changing reaction conditions. In addition, certain physicochemical properties of compounds in air, e.g., their distribution between the gas and particle phases, can promptly be investigated by using on-line APCI(45) Benoit, F. M.; Davidson, W. R.; Lovett, A. M.; Nacson, S.; Ngo, A. Anal. Chem. 1983, 55, 805. (46) Miller, J. C.; Miller, J. N. Statistics for Analytical Chemistry, 2nd ed.; John Wiley & Sons: New York, 1988; Chapter 5.

Analytical Chemistry, Vol. 75, No. 6, March 15, 2003

1415

Figure 8. (a) Traces of educt (m/z 137) and product ions (m/z 139, 277) after starting gas-phase ozonolysis by mixing sabinene into the reaction chamber at t ) 0. (b) Mass spectrum observed at a reaction time of t ) 120 min, which shows protonated sabinene at m/z 137 (educt), protonated sabinaketone at m/z 139 (product), and both [M + H - H2O]+ fragment ions and [2M + H]+ cluster ions of sabinaketone at m/z 121 and 277, respectively.

Figure 9. Direct quantitative determination of sabinaketone derived from the gas-phase ozonolysis of sabinene at a relative humidity of 50%. Temporal behavior of the relative abundance of (a) [M + H]+ ) m/z 139 and (b) [2M + H]+ ) m/z 277 ions of sabinaketone during the standard addition procedure.

ITMS.4,47 The main goal of the present work is to show the applicability of on-line APCI-MS for quantitative measurements of a target analyte present in a complex reaction mixture. For this purpose, the yield of sabinaketone (C9H14O), the main semivolatile oxidation product of the gas-phase ozonolysis of sabinene, was to be determined. Figure 8a shows traces of both [M + H]+ ions of the parent hydrocarbon sabinene (m/z 137) and the product sabinaketone (m/z 139) as well as the trace of the proton-bound cluster ions of sabinaketone observed at m/z 277. In addition, the decay of ozone as a result of constantly mixing sabinene at t ) 0 min into the 493-L reaction chamber is illustrated as a function of the reaction time. An almost instantaneous increase of the reaction product ion signal intensities (m/z 139) is observed, when the monoterpene ([M + H]+ ) m/z 137) is introduced into the chamber (Figure 8). Protonated cluster ions of sabinaketone (m/z 277) were (47) Ku ¨ ckelmann, U.; Warscheid, B.; Hoffmann, T. Anal. Chem. 2000, 72, 1905.

1416 Analytical Chemistry, Vol. 75, No. 6, March 15, 2003

detected at ∼15 min, as a result of ion-molecule reactions at higher product concentrations. Steady-state conditions with respect to product formation processes were observed at a reaction time of ∼300 min. Figure 8b shows the full on-line APCI-MS spectrum recorded at a reaction time of 120 min. At this time, protonated sabinaketone molecules (m/z 139) represent the ions of highest signal intensity (100% relative abundance), while the relative abundance of the proton-bound dimers was roughly 5%. In addition, [M + H]+ ions of sabinene (m/z 137) are still present, and ions at m/z 121 correspond to [M + H - H2O]+ fragment ions of sabinaketone. To perform a direct quantitative analysis of sabinaketone detected here, the dynamic range of the standard addition was tested under reaction chamber conditions adjusted in this experiment. Since sabinaketone is not commercially available, nopinone was used as internal reference compound. As both compounds are saturated bicyclic C9-ketones with the same molecular mass

Figure 10. Direct quantitative determination of sabinaketone derived from the gas-phase ozonolysis of sabinene at a relative humidity of 50% (gas and particle phases). Note, the relative signal intensities of both protonated sabinaketone and nopinone were calculated according to eq 1. Table 1. Molar Yields of Sabinaketone Present either in the Particle Phase or Both Gas and Particle Phases Derived from the Sabinene/Ozone Reaction at a Relative Humidity of E1 and 50%, Respectively molar yield sabinaketone C9H14O

at 50% relative humidity

at e1% relative humidity

gas and particle phase particle phase

0.64 ( 0.08 0.14 ( 0.02

0.33 ( 0.05 0.10 ( 0.01

(see Figures 2 and 8), proton affinity values of these compounds can be expected to be very similar. In consideration of the results discussed above, standard solution flow rates between 5 and 15 µL min-1 were adjusted for the direct quantitative analysis of sabinaketone in the gas and particle phases that derived from the ozonolysis of sabinene at 50% relative humidity (Figure 9). The average 5-95% rise and fall times were 40 and 30 s, respectively. However, the rise time was mainly determined by the transport of the standard solution through the fused-silica capillary, as the average 5-95% rise time for gaseous compounds was ∼15 s. Considering both monomer (m/z 139) and dimer (m/z 277) ion relative abundances, a mass flux of 0.283 µg min-1 sabinaketone from the reaction chamber into the APCI source was determined (Figure 10), which corresponds to the formation of 111 ppb (v/v) sabinaketone at a relative humidity of 50%. At dry reaction conditions, however, the formation of ∼55 ppb (v/v) sabinaketone was observed by applying the same quantification procedure. The molar yield of sabinaketone was therefore roughly reduced by a factor of 2 (see Table 1). The strong influence of humidity on the yield of the carbonyl compound can be explained by the reaction of water molecules with biradicals

formed as intermediates in the gas-phase ozonolysis of alkenes.48 Additionally, a charcoal denuder system was mounted directly in front of the ion source to separate gas-phase from particle-phase products derived from the sabinene/ozone reaction. For the conditions chosen, between 22 (50% relative humidity) and 30% (dry conditions) sabinaketone was present in the particle phase. Certainly, more experimental data on the gas-particle partitioning of semi- and low-volatile organic compounds are needed to reveal the underlying processes. However, these kinds of data are crucial to successfully model and predict the formation of secondary organic aerosols in the ambient atmosphere. In this context, the analytical on-line technique presented here might be useful not only for the identification and quantitation of tropospherically relevant VOC oxidation products but also for the measurement of fundamental physicochemical data. CONCLUSIONS On-line APCI-MS has been shown to be a valuable real-time technique for both qualitative and quantitative investigations of organic compounds. The method allows the direct introduction of both gaseous compounds and aerosols into a slightly modified ion source of a commercial ion trap mass spectrometer. A time resolution of ∼1 s is reached. In the present work, the technique has been applied for the direct quantification of a semivolatile reaction product in air by simultaneously introducing a liquid standard solution into the ion source. The quantification procedure was performed by the addition of the standard solution at different flow rates using the built-in syringe pump of the mass spectrometer. Both pure water and a water/methanol (50:50, v/v) mixture provide suitable matrixes for the efficient ionization of the analytes by proton transfer. However, the main advantage of the application of a standard addition method for the quantification of organic compounds in air by using a liquid standard solution is given by the fact that low-volatile organic species, which often represent significant constituents of airborne aerosol particles, also would be principally accessible for direct quantitative analysis. The method presented here has been shown to enable the quantification of a single product from a complex organic gas-phase reaction mixture in the presence of water vapor. Furthermore, working with or without an active charcoal denuder system in front of the APCI source provides the means either to measure both gas and particle phases or to focus exclusively on the chemical composition of the particulate matter. ACKNOWLEDGMENT This work was supported by the EC within the program Environment and Climate (EVK2-CT-1999-00016) and by the BMBF (Bundesministerium fu¨r Bildung, Wissenschaft, Forschung und Technologie) within the Aerosol Research Program. Received for review May 21, 2002. Accepted January 14, 2003. AC025788D (48) Warscheid, B.; Hoffmann, T. Atmos. Environ. 2001, 35, 2927.

Analytical Chemistry, Vol. 75, No. 6, March 15, 2003

1417