On-Line Characterization of Organic Aerosols Formed from Biogenic

In this paper, we compare the results from the investigations of organic aerosols and artificially generated dimer cluster ions of the same compounds ...
0 downloads 0 Views 171KB Size
Download PDF
Anal. Chem. 2000, 72, 1905-1912

On-Line Characterization of Organic Aerosols Formed from Biogenic Precursors Using Atmospheric Pressure Chemical Ionization Mass Spectrometry Ulrich Ku 1 ckelmann, Bettina Warscheid, and Thorsten Hoffmann*

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

A method to investigate the chemical composition of organic aerosols formed from biogenic hydrocarbon oxidation using atmospheric pressure chemical ionization mass spectrometry (APCI/MS) is described. The method involves the direct introduction of aerosol particles into the ion source of the mass spectrometer. Using this technique, reaction monitoring experiments of r-pinene ozonolysis show the formation of hetero- and homomolecular cluster anions (dimers) of the primary oxidation products (multifunctional carboxylic acids). Since the formation of dimers plays a profound role in new particle formation processes by homogeneous nucleation in the atmosphere and, at the same time, is an intrinsic feature of APCI, it is essential to differentiate between both processes when on-line APCI/MS is applied. In this paper, we compare the results from the investigations of organic aerosols and artificially generated dimer cluster ions of the same compounds using identical ionization conditions. The clusters and their formation processes are characterized by varying the analyte concentration, investigating the thermal stability of dimers, and studying collisional activation properties of both ion species. The investigations show a significant difference in ion stability: dimer anions measured on-line have an estimated stability that is 20 kJ mol-1 higher than that of the corresponding artificially generated cluster ions. Hence, the technique provides the possibility to accurately characterize dimers as ionized reaction products from biogenic hydrocarbon oxidation and allows an insight into the process of new-particle formation by homogeneous nucleation. Atmospheric aerosols scatter sunlight, serve as condensation nuclei for cloud droplet formation, and participate in heterogeneous chemical reactions. Hence, they play an important role in global climate and atmospheric chemistry.1 Frequently, organic matter dominates the composition of sub-micrometer aerosols, especially over continental regions.2-4 A significant natural con(1) Andreae, M. O.; Crutzen, P. Science (Washington, D.C.) 1997, 276, 10521058. (2) Talbot, R. W.; Andreae, M. O.; Andreae, T. W.; Hariss, R. C. J. Geophys. Res. 1988, 93, 1499-1509. 10.1021/ac991178a CCC: $19.00 Published on Web 03/09/2000

© 2000 American Chemical Society

tribution to this organic fraction of atmospheric aerosol is believed to derive from the oxidation of biogenic volatile organic compounds (VOCs), since large quantities of reactive VOCs (monoterpenes, sesquiterpenes, as well as a series of oxygen-containing compounds) are emitted into the atmosphere by terrestrial vegetation.5,6 Their reactions with the principal atmospheric oxidizing agentssO3, OH, and nitrate radicals7sresult in the formation of products with a lower volatility, which finally undergo gas-to-particle conversion.8-10 One key aspect concerning the climatic relevance of terrestrial biogenic particles is the question of whether the oxidation products of VOCs just add to the tropospheric aerosol mass by condensation on preexisting particles or whether they also contribute to the aerosol number concentration by homogeneous nucleation of products with an extremely low volatility (Figure 1). Recent field studies on the aerosol formation at forested sites indeed indicated a link between new-particle formation processes and biogenic VOC oxidation.11-13 However, since the absolute amount of a condensable species required to form nanometer particles is extremely low, the identification of the actual nucleating species in the field is rather difficult and supplementary laboratory studies are essential to identify potential species for new-particle formation. The choice of a suitable analytical method should consider that likely (3) Artaxo, P.; Maenhaut, W.; Storms, H.; v. Grieken, R. J. Geophys. Res. 1990, 95, 16971-16978. (4) Hobbs, P. V. IGACtivities Newsletters 1998, 11, 3-5. (5) Fehsenfeld, F.; Calvert, J.; Fall, R.; Goldan, P.; Guenther, A. B.; Hewitt, C. N.; Lamb, B.; Liu, S.; Trainer, M.; Westberg; H.; Zimmerman, P. Global Biogeochem. Cycles 1992, 6, 389-430. (6) Guenther, A.; Hewitt, C. N.; Erickson, D.; Fall, R.; Geron, C.; Graedel, T.; Harley, P.; Klinger, L.; Lerdau, W. A.; McKay, W. A.; Pierce, T.; Scholes, B.; Steinbrecher, R.; Tallamraju, R.; Taylor, J.; Zimmerman, P. J. Geophys. Res. 1995, 100, 8873-8892. (7) Atkinson, R.; Arey, J. Acc. Chem. Res. 1998, 31, 574-583. (8) Seinfeld, J. H.; Pandis, S. N. Atmospheric Chemistry and Physics; John Wiley & Sons: 1998; Chapter 13. (9) Hoffmann, T.; Odum, J. R.; Bowman, F.; Collins, D.; Klockow, D.; Flagan, R. C.; Seinfeld, J. H. J. Atmos. Chem. 1997, 26, 189-222. (10) Odum, J. R.; Hoffmann, T.; Bowman, F.; Collins, D.; Flagan, R. C.; Seinfeld, J. H. Environ. Sci. Technol. 1996, 30, 2580-2585. (11) Kavouras, I. G.; Mihalopoulos, N.; Stephanou, E. G. Nature (London) 1998, 395, 683-686. (12) Ma¨kela¨, J. M.; Aalto, P.; Jokinen, V.; Pohja, T.; Nissinen, A.; Palmroth, S.; Markkanen, T.; Seitsonen, K.; Lihavainen, H.; Kulmala, M. Geophys. Res. Lett. 1997, 24, 1219-1222. (13) Marti, J. J.; Weber, R. J.; McMurry, P. H.; Eisele, F. L.; Tanner, D. J.; Jefferson, A. J. Geophys. Res. 1997, 102, 6331-6339.

Analytical Chemistry, Vol. 72, No. 8, April 15, 2000 1905

Figure 1. Reaction scheme of atmospheric R-pinene oxidation leading to secondary organic aerosols and various products observed from the ozonolysis of R-pinene.

candidates for nucleation can be assumed to possess very low vapor pressures. The most common methods utilized for the identification of VOC oxidation products in the particle phase involve preconcentration using filters or cartridges followed by solvent extraction and chromatographic analysis. The latter is either carried out after a derivatization step and GC separation14-16 or by analyzing the extracted products directly employing HPLC/MS.17 Since these techniques are multistep analytical procedures they possess several disadvantages. First, they are time-consuming and do not allow a direct insight into the particle-formation process. Second, analytical problems, such as sample degradation or incomplete analyte extraction, may arise when a preconcentration step is part of the procedure. Furthermore, intermolecular forces between the analytes, which might have an effect during the particle-formation process, will remain undiscovered when organic solvents are used during the sample preparation. Thus, we introduced an analytical on-line technique based on mass spectrometry for the molecular speciation of the newly formed particle phase. Choosing the R-pinene/ozone reaction as a model system for biogenic VOC oxidation and subsequent gas-to-particle conversion, an adequate ionization technique for mass spectrometric reactionproduct monitoring is required. Ideally, the technique has to provide the possibility to obtain information about the molecular composition of both gas and particle-phase ozonolysis products. For that reason and considering the specific aim of this work, to include also the analysis of compounds with a very low volatility, the ionization method of choice is atmospheric pressure chemical ionization (APCI).18-20 APCI/MS has been used before for the identification of products from biogenic VOC oxidation reactions in the gas phase21-23 and for introductory studies to investigate nucleation.24,25 (14) Yu, J.; Cocker, D. R.,III; Griffin, R. J.; Flagan, R. C.; Seinfeld, J. H J. Atmos. Chem. 1999, 34, 207-258. (15) Christofferson, T. S.; Molander, L. L.; Jensen, N. R.; Hjorth, J.; Kotzias, D.; Larsen, B. R. Atmos. Environ. 1998, 32, 1657-1661. (16) Jang, M.; Kamens, R. M. Atmos. Environ. 1998, 1999, 33, 459-474. (17) Glasius, M.; Duane, M.; Larsen, B. R. J. Chromatogr. A 1999, 833, 727735. (18) Kambara, H. Anal. Chem. 1982, 55, 143-146. (19) Horning, E. C.; Caroll, D. I.; Dzidic, I.; Haegele, K. D.; Horning, M. G.; Stillwell, R. N. J. Chromatogr. 1974, 99, 13. (20) Horning, E. C.; Caroll, D. I.; Dzidic, I.; Haegele, K. D.; Horning, M. G.; Stillwell, R. N. J. Chromatogr. Sci. 1974, 12, 725.

1906

Analytical Chemistry, Vol. 72, No. 8, April 15, 2000

One basic drawback of atmospheric pressure chemical ionization is the formation of cluster ions.26-29 For example, Carroll et al. already observed the dependence of benzene cluster formation (C12H12+) upon the benzene concentration.30 However, some analytical applications of APCI/MS take advantage of the fact that clustering reactions are occurring in the ionization region.31 In general, clusters are produced via a series of association reactions between an ion, I, and ligand, N, except in those instances where cluster ions are generated by the ionization of neutral clusters or by surface-bombardment techniques.32-34 The association process is considered to proceed via an intermediate complex (IN)* which has a characteristic lifetime against unimolecular decomposition back to the reactants I and N. Collision with a third body, M, removes excess energy and results in the formation of a stable cluster ion. Alternatively, collisions can activate a cluster ion, promoting the ion to a sufficiently higher energy level, enabling unimolecular dissociation to the reactants.35 The latter property of clusters was used by Kambara et al. to distinguish cluster ions from quasi-molecular ions working with collision induced dissociation (CID).27,28 They observed a drastic decrease of cluster ions’ relative abundances by increasing the drift voltage in the (21) Kwok, E. S. C.; Atkinson, R.; Arey, J. Environ. Sci. Technol. 1995, 29, 24672469. (22) Kwok, E. S. C.; Aschmann, S. M.; Arey, J.; Atkinson, R. Int. J. Chem. Kinet. 1996, 28, 925-934. (23) Aschmann, S. M.; Shu, Y.; Arey, J.; Atkinson, R. Atmos. Environ. 1997, 31, 3551-3560. (24) Hoffmann, T.; Bandur, R. Marggraf, U.; Linscheid, M. J. Geophys. Res. 1998, 103, 25569-25578. (25) Mikheev, V. B.; Pervukhin, V. V.; Laulainen, N. S. J. Aerosol Sci. 1998, 29, S89. (26) Covey, T. R.; Lee, E. D.; Bruins, A. P.; Henion, J. D. Anal. Chem. 1986, 58, 1451A. (27) Kambara, H.; Mitsui, Y.; Kanomata, I. Anal. Chem. 1979, 51, 1447-1452. (28) Kambara, H.; Kanomata, I. Anal. Chem. 1977, 49, 270-275. (29) Kambara, H.; Kanomata, I. Int. J. Mass Spectrom. Ion Phys. 1977, 25, 129136. (30) Caroll, D. I.; Dzidic, I.; Stillwell, R. N.; Haegele, K. D.; Horning, E. C. Anal. Chem. 1975, 47, 2369-2373. (31) Scott, A. D., Jr.; Hunter, E. J.; Ketkar, S. N. Anal. Chem. 1998, 70, 18021804. (32) Aberth, W. Anal. Chem. 1986, 58, 1221-1225. (33) Emery, W. B.; Cooks, R. G.; Toren, P. C. Anal. Chem. 1986, 58, 12181221. (34) Tondeur, Y.; Clifford, A. J.; Luigi, M.; DeLuca, M. Org. Mass Spectrom. 1985, 20, 159. (35) Castleman, A. W.; Keesee, R. G. Chem. Rev. 1986, 86, 589-618.

intermediate region between atmospheric pressure and analyzer vacuum. On the basis of these observations, they discussed the relation between cluster dissociation energy and the field strength critical for dissociation.28 Analogously, in electrospray ionization of biomolecules, the variation of the drift voltage, the so-called declustering,36,37 in the high-pressure region of the ion source is used to characterize macromolecular complexes.38 Moreover, the formation of clusters (in this case neutral clusters) play a fundamental role in atmospheric chemistry. For example, the accumulation of stable dimers in the gas phase may affect considerably the nucleation rate of condensable vapors.39 Here, the dimer formation can already be regarded as the first step in homogeneous nucleation, in which the buildup of small clusters (dimers, trimers, etc.) initiates the phase-transformation process. Therefore, if strong intermolecular forces between nucleating species (e.g., hydrogen bonds) shift the equilibrium between monomers and dimers, the initial step of the formation of a new phase is strongly influenced, resulting in a much higher tendency to form new particles than exists in systems in which nonassociated molecules are nucleating.40 The impetus of the present work is to demonstrate that online APCI/MS is a suitable analytical technique to obtain fundamental chemical information on aerosol-forming substances and that the results can give insight into nucleation processes in the atmosphere. To do so, the distinction between the processes outlined above, the differentiation between ionized dimers (in which the neutral dimers were already formed prior the ionization step) and equivalent clusters formed by ionization processes, becomes a crucial task. If the on-line-recorded spectra allow the interpretation that primary biogenic oxidation products (mainly multifunctional carboxylic acids) interact with each other directly following their formation and generate stable binary carboxylic acid adducts (see Figure 2), this information could indeed shed light on new-particle formation processes from natural sources. Otherwise, if the dimers are produced by ionization processes through association reactions in the ion source, the dimers prevent the accurate determination of product yields. However, it will be of great importance concerning the accuracy of the analytical results, whether the dimers are ionization artifacts or real reaction products. We try to solve this problem by comparing the mass spectrometric properties of artificially generated cluster ions of well-known oxidation products of R-pinene,14-17 namely cis-pinic (P) and cis-pinonic acid (Po), and the properties of the corresponding products observed on-line. Particularly, the influences of analyte concentration, vaporizer temperature, and so-called source CID voltage on relative ion abundance are investigated. In this study, especially the vaporizer temperature and source CID are fast and powerful techniques to get indispensable information about selected (not separated) ions simultaneously. (36) Smith, R. D.; Loo, J. A.; Orgorzalek-Loo, R. R.; Busman, M.; Udseth, H. R. Mass Spectrom. Rev. 1991, 10, 359. (37) Smith, R. D.; Light-Wahl, K. J. Biol. Mass Spectrom. 1993, 22, 493-501. (38) Przybylski, M.; Glocker, M. O. Angew. Chem. 1996, 108, 878-899. (39) Lushnikov, A. A.; Kulmala, M. Phys. Rev. E: Stat. Phys., Plasmas, Fluids, Relat. Interdiscip. Top. 1998, 58, 3157-3167. (40) Katz, J. L.; Fisk, J. A.; Rudek, M. M. In Nucleation and Atmospheric Aerosols, Proceedings of the 14th International Conference on Nucleation and Atmospheric Aerosols, Helsinki, Finland, August 26-30; Kulmala, M., Wagner, P. E., Eds.; Elsevier Science: New York, 1996; pp 1-10.

Figure 2. Hydrogen-bond pattern of the cis-norpinic/cis-pinic acid dimer (a) and the cis-pinic/cis-pinonic acid dimer anion (b) (gray for carbon, dark gray for oxygen, white for hydrogen).

EXPERIMENTAL SECTION On-Line Product Analysis of Organic Aerosols Formed in the r-Pinene/Ozone Reaction. The experiments were carried out in a 2-L spherical glass flask installed immediately in front of the APCI source of the mass spectrometer (see Figure 3). Prior to each experiment, the flask was flushed with humidified synthetic air (Messer-Griesheim, Germany; 20.5% O2, 79.5% N2, 60% rel humidity). R-Pinene (490 ppbv) was added from a dynamic test gas generator, which was based on an open tube diffusion technique. Ozone (570 ppbv) was produced by UV irradiation of an additional synthetic air supply and introduced into the flask. The total air flow through the flask was adjusted to 1 L min-1. Since all reactants (R-pinene, O3, humidified synthetic air) were introduced continuously into the reactor, the whole system can be considered as a constant organic aerosol generator with an average reaction time (mean residence time of the reactants) of 2 min. Because a condensation nucleus counter (CNC, TSI 3022, smallest detectable particle diameter, 7 nm) as particle detector was installed instead of the mass spectrometer, particle formation could be observed about 2 min after all reactants were introduced into the reaction flask. At approximately the same time, the first products could be measured in the negative-ion mode. Most of Analytical Chemistry, Vol. 72, No. 8, April 15, 2000

1907

Figure 3. Schematic diagram of the experimental setup.

the gas-phase hydrocarbons were removed in front of the APCI source using an activated charcoal diffusion denuder (efficiency > 0.99). However, reducing the flow through the diffusion denuder, hence increasing the removal efficiency, indicated that the small amount of remaining R-pinene has no influence on the results. Investigation of Artificial Cluster Formation in the Ion Source. At controlled rates (0.5-50 µg min-1), solutions of pinic acid (0.15 mg mL-1), pinonic acid (0.042 mg mL-1), and pinic + pinonic acid (0.03 mg mL-1 each) in water were delivered through a deactivated fused silica capillary into the APCI source using a syringe pump (Figure 3). The choice of the two reference compounds was based on their importance as reaction products in the R-pinene/O3 reaction as well as on their availability. Both substances are commercially available (Sigma/Aldrich (Rare Chemicals)). As is typical for APCI, a sheath gas flow (nitrogen) was used to transport the analytes into the vaporizer and ionization region. If necessary, the concentrations of the solutions given above were lowered to appropriate values. Mass Spectrometric Analysis. All data were recorded on a Finnigan LCQ ion-trap mass spectrometer. The instrument was equipped with an atmospheric pressure ion source, which operated in the chemical ionization mode (APCI). The commercial APCI source was slightly modified in order to allow the direct introduction of the aerosol flow. A glass tube (i.d. 1.1 mm, o.d. 1.8 mm) surrounded by the sheath gas tube was mounted in the center of a high-temperature tube (vaporizer) as a sample inlet. If no other value is mentioned, the vaporizer temperature was adjusted to 350 °C. The voltage applied to the corona discharge needle (3 kV) and the nitrogen sheath gas flow were constant during all experiments. Using a LCQ mass spectrometer, source CID occurs in the ion-optics region. This region is not a part of the atmospheric pressure ion source (see Figure 3). Because of the relatively low pressure (0.1 Pa), small differences in ion internal energies are detectable without ion separation with tandem MS techniques. Ions are activated by collisions of 2-20 1908

Analytical Chemistry, Vol. 72, No. 8, April 15, 2000

V (2-V steps) in the laboratory frame of reference (lab). Energies in the laboratory frame of reference can be converted to the center of mass frame (cm) with the following

Ecm ) Elab m/(M + m)

(1)

where Elab and Ecm are the relative translational energies of the ion and the neutral target in the laboratory and center of mass frame of reference, respectively.41-43 M and m are the masses of the ion and the neutral target, respectively. The target gas in the ion-optics region is an air/solvent mixture. RESULTS AND DISCUSSION It is a well-known fact that carboxylic acids produce anion dimers under relatively high pressure ionization conditions.41,44,45 Accordingly, under negative-ion APCI conditions, aqueous solutions of pinic and pinonic acid form homo- and heterodimeric cluster ions: [2P - H]- m/z 371, [2Po - H]- m/z 367, [P + Po - H]- m/z 369. The relative abundances of these cluster ions depend on the organic acid concentration (Figure 4). The higher the induced mass flux of pinic and/or pinonic acid, the higher is the relative abundance of the corresponding clusters [2P - H]-, [2Po - H]- and [P + Po - H]-. For pinic acid, pinonic acid, and their mixture, there is a threshold flux below which the production of cluster ions is negligible (see Figure 4). This threshold mass flux correlates with certain absolute intensities of the monomers [P - H]- and [Po - H]-. (41) Graul, S. T.; Schnute, M. E.; Squires, R. R. Int. J. Mass Spectrom. Ion Processes 1990, 96, 181-198. (42) Graul, S. T.; Squires, R. R. J. Am. Chem. Soc. 1990, 112, 2517-2529. (43) Busch, K. L.; Glish, G. L.; McLuckey, S. A. In Mass Spectrometry/Mass Spectrometry: Techniques and Application of Tandem Mass Spectrometry; VCH Publishers: New York, 1988; Chapter 3. (44) Wright, L. G.; McLuckey, S. A.; Cooks, R. G.; Wood, K. V. Int. J. Mass Spectrom. Ion Processes 1982, 42, 115-124. (45) Meot-Ner, M.; Sieck, L. W. J. Am. Chem. Soc. 1986, 108, 7525-7529.

Figure 5. Product mass spectrum of a pinic/pinonic acid solution (artificial cluster generation) (a) and in comparison with the mass spectrum of R-pinene/ozone reaction products (b). In both cases, the absolute intensity of the monomer anions is in the range of (2-3) × 104.

Figure 4. Monomer and dimer ion abundance as a function of the mass flux into the ion source: pinic acid (a), pinonic acid (b), mixture of pinic and pinonic acid (c) in water.

It is remarkable that dimer clusters were formed with strikingly high abundance whereas we did not observe significant formation of oligomers (tri- or tetramers). Additional investigations of the cluster-forming capability of other diacids without or with other configurational restrictions (adipic acid, trans-norpinic acid, data not shown) confirms this observation. Even in electrospray ionization there is no higher clustering noticeable. Obviously, the trans configuration complicates the cluster formation, since dimer ions of trans-norpinic acid are only produced with low abundance (maximum dimer/monomer ratio 1/2). In principle, it must be stressed here, that the basic structures of the pinic and/or pinonic acid dimers are determined through the cis configuration of the monomers. The formation of four intermolecular hydrogen bonds is sterically favored in the case of cis-pinic acid. The heterodimer (pinic/pinonic acid) can be associated with two hydrogen bonds and an additional hydrogen bond between the carbonyl group and the carboxylic OH proton-donating group. In Figure 2b, the hydrogen-bond patterns of the dimer anion [P + Po - H]- are demonstrated. Figure 5a shows the mass spectrum when an aqueous solution of a mixture of pinic acid and pinonic acid is delivered into the APCI/MS system.

The product spectra recorded on-line contain a large number of different dimers (Figure 5b). Besides the formation of dimers between pinic and pinonic acid, the observed variety of adducts between two monomers can be assumed to consist of other R-pinene reaction products which have been tentatively identified before,14-17 e.g., norpinonic acid (m/z 169), norpinic acid (m/z 171), and hydroxy pinonic acid (m/z 199). For instance, the signal appearing at m/z 357 can be associated with a remarkably stable dimer between cis-pinic and cis-norpinic acid (see Figure 2).24 However, because of the lack of reference compounds, we focus our investigations here on homo- and heterodimers of pinic and pinonic acid (m/z 369, m/z 371). Both dimers were characterized by tandem MS techniques (CID in the ion trap). The main channels for decay of these ions are the loss of a monomer, a minor channel the loss of water. The MS/MS spectra of the heterodimer m/z 369 and homodimer m/z 371 are shown in Figure 6. The low abundance of the daughter ion m/z 183, [Po - H]-, indicates the higher proton affinity of ketones in comparison with an equivalent carboxyl group.46 There is no significant difference between these spectra and the corresponding daughter-ion spectra of the artificially generated clusters m/z 369 and m/z 371, respectively. The first hint that the dimers found in the on-line monitoring experiments of R-pinene/O3 reaction are not ionization artifacts (46) Hunter, E. P.; Lias, S. G. In Proton Affinity Evaluation; Mallard, W. G., Linstrom, P. J., Eds.; NIST Chemistry WebBook, NIST Standard Reference Database No. 69; National Institute of Standards and Technology: Gaithersburg, MD, November 1998.

Analytical Chemistry, Vol. 72, No. 8, April 15, 2000

1909

Figure 6. MS/MS spectra of the dimer anions m/z 369 ([Po + P H]-) (a) and m/z 371 ([2P - H]-) (b) from R-pinene ozone experiments.

caused by clustering of the individual products in the ion source can be obtained by the comparison of the product spectra and the spectra of artificial generated clusters under the same ionization conditions. In Figure 4, we have shown the threshold intensities of the quasi-molecular ions [P - H]- and [Po - H]at that point, where cluster-ion abundance decreases to noise level (below 1 × 102). With regard to these threshold abundances, no cluster formation should be observable in the case of product analysis since even the highest signals of the corresponding monomers are too weak. However, in contrast to the behavior of the artificially generated clusters, product dimers appear in high abundance even at very low concentrations. This becomes obvious by downscaling (diluting the solution) the artificially generated cluster spectra to on-line product spectra level (Figure 5). Evidently, there seems to be a significant difference between ions with the same m/z values containing the same substances. In general, cyclic carboxylic diacids such as pinic acid are nonvolatile thermolabile compounds. The influence of temperature variation on the ionization process and especially on cluster formation should provide further information about the difference between the ion species with the same nominal mass. It is wellknown that clustering reactions in the ion source are favored when the temperature of the reactants in the gas phase decreases.47 Vice versa, thermal declustering can occur with increasing temperature.48,49 In Figure 7a, the relative abundance of the (47) Meot-Ner, M. In Gas-Phase Ion Chemistry; Bowers, M. T., Ed.; Academic Press: New York, 1979; Vol. 1, Chapter 6. (48) Guevremont, R.; Siu, K. W. M., Wang, J.; Ding, L. Anal. Chem. 1997, 69, 3959-3965.

1910 Analytical Chemistry, Vol. 72, No. 8, April 15, 2000

Figure 7. Relative abundance of m/z 183 ([Po - H]-), m/z 185 ([P - H]-), m/z 369 ([Po + P - H]-), and m/z 371 ([2P - H]-) as a function of the vaporizer temperature Tvap, artificially generated cluster ions (a), and R-pinene/ozone reaction products (b).

artificially generated cluster ions m/z 369, 371 as well as the relative abundance of the deprotonated acids m/z 183, 185 are shown as a function of the vaporizer temperature (100-600 °C). Up to 350 °C, the relative abundance of the cluster ions decreases nearly to noise level, whereas monomer abundance increases with increasing temperature. In contrast, Figure 7b shows the influence of temperature on monomer/dimer ratios when reaction products are introduced into the ion source. Here, the relative abundance of dimers increases with increasing temperature (up to 500 °C). At the same time, the relative abundance of monomer ions stays nearly constant. From 500 °C to 600 °C, the relative abundance of monomers and dimers decrease. The behavior of the monomer ions m/z 183 and m/z 185 gives a closer insight into the characteristics of the experiment itself, considering that the vaporizer temperature influences processes before ionization. In the case of artificial cluster generation, the increasing relative abundance of the monomer ions (between 200 and 350 °C), together with the decreasing abundance of cluster ions, indicates that cluster formation should not be considered as a pure gas-phase process. If adduct formation through association reactions would occur only in the gas phase, effective firstorder kinetics could be expected34 and monomer/dimer ion abundance should not be anticorrelated as observed, since the neutral reactant is typically in great excess of the ion concentration. Consequently, processes in the ion source can be assumed (49) Chen, Y. H.; Hill, H. H.; Wittmer, D. P. Int. J. Mass Spectrom. Ion Processes 1996, 154, 1-13.

to proceed via the following sequence: the solution of pinic and/ or pinonic acid is nebulized, the produced droplets are heated in the vaporizer region. The solvent is evaporated and neutral monomers and neutral dimers are entering the ionization region. The amount of dimers depends directly on the concentration of the corresponding acids in solution. Whether the dimers are already present in the condensed phase or are formed subsequent to the evaporation of the solvent in the gas phase cannot be decided from our experiments. However, gas-phase ion/molecule reactions in the ion source can have only a minor effect on the observed cluster-formation process, since the expected correlation between monomer and dimer ions is not observed. In contrast, the outcome of the on-line analysis of organic aerosols suggests that dimers always represent an essential fraction of the organic material within the aerosol particles formed in the R-pinene/O3 reaction. The monomer/dimer ratio cannot be comparatively changed by increasing or decreasing reactant concentrations, and dimers are always appearing together with the monomers, even at the lowest detectable concentrations. Intermolecular forces are strong, the thermal stability is high, and thermal declustering starts to play a role only at temperatures higher than 500 °C. A relatively high temperature is needed to induce the evaporation/ionization of the dimers (>250 °C). Moreover, the fact that no oligomer ions (e.g., trimers) can be observed indicates that the interactions between monomers and dimers are weak. This is consistent with the molecular structures suggested for the dimers (Figure 2), since all possible sites for strong intermolecular interactions are occupied. In light of these findings, an estimation of the energy difference between the artificially generated cluster ions and reaction products for dissociation becomes necessary. This has to be done in the ion optic region, since after isolation of the ions in the ion trap, they have enough time (ms) to rearrange to a more stable configuration in both cases. For that reason, the change of ion relative abundance (m/z 369, m/z 371) with increasing CID voltage in the ion optics region is monitored (Figure 8). As expected, the abundance of both ion species, artificially generated clusters (Figure 8a) and reaction products (Figure 8b), decrease in correspondence with this additional acceleration and collisional activation of the ions. The signals of the deprotonated monomers, [Po - H]- and [P - H]-, increase strongly (cluster generation) and slightly (products) in the range of 0-10 V; in the range of 10-20 V, they slightly begin to decrease. To interpret the obtained mass spectrometric data semiquantitatively, we have to make the following assumptions: (1) m/z 369 and m/z 371 are parent ions. If other ions decompose to these ions through collisional activation, the comparison of relative-ion abundance to determine energy differences will be meaningless. Since no possible precursor ions were detected, this assumption seems appropriate. Conversely, if the parent ions mainly dissociate into the monomers, competing fragmentations are negligible. In addition, some fundamental considerations concerning the gas-phase ion chemistry have to be made: (2)The relatively low pressure in the ion-optics region provides single-energy collisions in this range of additional axial energies. This assumption allows the conversion of translational energies (Elab to Ecm). At electron-volt collision energy activation, a relatively

Figure 8. Relative abundance of m/z 183 ([Po - H]-), m/z 185 ([P - H]-), m/z 369 ([Po + P - H]-), and m/z 371 ([2P - H]-) as a function of the CID voltage in the ion optics region, artificially generated cluster ions (a), and R-pinene/ozone reaction products (b).

large fraction of Ecm is transferred into internal energy.43 The internal energy deposition increases notably with increasing target pressure.50 Hence, an estimation of ion energies using Ecm under multicollision conditions would be misleading in the choice of range of energy. (3) The nominal kinetic energy of the parent ion is, in fact, a distribution of collision energies as a result of parent ion energy spread and thermal motion of the collision gas. We assume that parent ion energy spread and the thermal motion of the collision gas are constant. Then, an observed energy difference between ions with the same m/z values containing the same substances should provide a difference in the dissociation energies of the dimer ions. To obtain information about energy differences at an uniform level, the voltages (Elab) at 50% transmission of the parent ion are converted to relative translational energies in the center of mass frame of reference Ecm 50 using eq 1. The results are summarized in Table 1. The higher stability of the heterodimer anion [Po + P - H]- again corresponds with high ketone proton affinities.46 Binding energies in carboxylic acid dimers are probably in the range of 100-150 kJ mol-1 (dissociation at activation energies of 1-1.5 eV Ecm).41,45 According to these values, the estimated differences in relative translational energies (Table 1) are relatively low (0.21 eV or 20 kJ mol-1). To give a reasonable interpretation of these data, the hydrogen-bond pattern in Figure 2 has to be (50) Kentta¨maa, H. I.; Cooks, R. G. Int. J. Mass Spectrom. Ion Processes 1985, 64, 79-83.

Analytical Chemistry, Vol. 72, No. 8, April 15, 2000

1911

Table 1: Estimated Relative Translational Energies in the Center of Mass Frame of Reference (Ecm) at 50% Transmission

[Po + P - H][2P - H]-

Ecm 50(products) (eV)

Ecm 50(cluster) (eV)

∆Ecm (eV)

0.63 0.49

0.42 0.28

0.21 0.21

discussed in detail. The features (strength, bond distances, etc.) of a hydrogen bond can dramatically depend on the environment.51-53 Whether the clusters originate from the evaporation of the solvent or are produced in the condensed phase, the formation of hydrogen bonds can be retarded by protic solvent molecules protecting the functional groups of the monomers. Additionally, the acid/base equilibrium in solution hinders the association of neutral molecules. If the dimers as reaction products originate in the gas phase subsequent to the formation of the monomers, the influence of a polar solvent is negligible (60% rel humidity) and hydrogen bonds can be stabilized. Taking into account that resonance and charge are possible causes of the strength of a hydrogen bond,54 charge effects should be the same in both anion species. Therefore, a resonance stabilization with a more covalent and symmetrical O‚‚‚H‚‚‚O bond of the gas phase product dimer can be assumed. However, the lack of experimental or theoretical data concerning the investigated compounds prevents a more detailed quantitative assessment of the obtained results. Remaining specific questions, e.g., about structural differences between artificially generated clusters and reaction products or differences in the stability of distinct monomers, are beyond the scope of this work and can only be answered if a novel experimental setup is introduced. To answer these questions by the determination of activation energies, instrumental modifications would be necessary. (51) Garcia-Vilosa, M.; Gonzalez-Lafont, A.; Lluch, J. M. J. Am. Chem. Soc. 1997, 119, 1081-1086. (52) Perin, C. L.; Thoburn, J. D. J. Am. Chem. Soc. 1992, 114, 8559-8565. (53) Perrin, C. L. Science (Washington, D.C.) 1994, 266, 1665. (54) Gilli, P.; Bertolasi, V.; Ferretti, V.; Gilli, G. J. Am. Chem. Soc. 1994, 116, 909-915.

1912 Analytical Chemistry, Vol. 72, No. 8, April 15, 2000

CONCLUSIONS The direct introduction of organic aerosols into a slightly modified APCI source of an ion-trap mass spectrometer provides an attractive analytical approach for gaining reliable information on the chemical composition of condensable atmospheric species. Because of the possibility of following the particle-formation process on-line, the reaction conditions leading to aerosol-forming compounds can rapidly be changed and their influence on the chemical nature of the aerosols can be studied. Additionally, problems commonly encountered when multistep analytical procedures are employed, such as artifact formation during the sampling or preparation step, are greatly reduced when the compounds of interest are directly introduced into the analytical system. Furthermore, insight into fundamental processes in particle formation by homogeneous nucleation can be obtained by on-line APCI/MS. The dimers observed applying this technique could be characterized as products from gas-phase ozonolysis of R-pinene. Using adequate reference compounds and the specific properties of atmospheric pressure ionization to simulate clustering evidently shows that the artificially generated ions differ significantly from corresponding dimer anions measured on-line as reaction products. The experimental methods can easily be performed with a modern mass spectrometer. A comparison of cluster bond energies was carried out by collision-induced dissociation of different dimers. As far as the on-line APCI/MS analysis of other VOC oxidation reactions is concerned, the main difficulty is the unsatisfactory availability of adequate reference compounds. If reference compounds are available, appropriate analytical results can immediately be obtained by APCI/MS also for other VOC oxidation reactions. ACKNOWLEDGMENT This work was supported by the EC within the program Environment and Climate (ENV4-CT97-0391) and by the BMBF (Bundesministerium fu¨r Bildung, Wissenschaft, Forschung und Technologie) within the Aerosol Research Program. Received for review October 12, 1999. Accepted January 28, 2000. AC991178A