Quantitative Detection of Aromatic Compounds in Single Aerosol

Anal. Chem. 2001, 73, 2317-2322

Quantitative Detection of Aromatic Compounds in Single Aerosol Particle Mass Spectrometry Ephraim Woods, III, Geoffrey D. Smith, Yury Dessiaterik, Tomas Baer,* and Roger E. Miller*

Department of Chemistry, University of North Carolina, Chapel Hill, North Carolina 27599-3290

Most laser-based aerosol mass spectrometers rely on a single ultraviolet laser to both ablate and ionize the aerosol particle. This technique produces complex and fragmented mass spectra, especially for organic compounds. The approach presented here achieves a more robust and quantitative analysis using a CO2 laser to evaporate the aerosol particle and a vacuum ultraviolet laser to ionize the vapor plume. Vacuum ultraviolet laser ionization produces little fragmentation in the mass spectra, making the identification of an aerosol particle’s constituents more straightforward. An analysis of simple, threecomponent mixtures of aniline, benzyl alcohol, and mnitrotoluene shows that the technique also provides a quantitative analysis for all the components of the mixture. Furthermore, the detection of predominantly parent ion signal from anthracene particles demonstrates the utility of the technique in the analysis of lower vapor pressure, solid-phase aerosols. Finally, we discuss the potential and limitations of this technique in analyzing organic atmospheric aerosols. The development of instruments capable of analyzing both the size and composition of single atmospheric aerosol particles is an active area of research. Mass spectrometry is the principle tool of these efforts since it is fast enough to provide on-line analysis and has sufficient sensitivity to detect minor components of the aerosols. In recent years, a number of spectrometer configurations have been developed using different aerosol sampling techniques, particle sizing methods, and detection/ionization schemes.1,2 Here, we focus on the latter. Several researchers3-5 employ flash vaporization of the aerosol particle followed by electron impact ionization and quadrupole mass analysis of the vapor cloud. This arrangement is technically simple and quantitative. However, the quadrupole mass spectrometer can only monitor a single mass for any single aerosol particle. Other groups6-8 use laser ionization coupled with time-of-flight mass spectrometry, a technique that * Corresponding authors: (fax) (919)962-2388; (e-mail) [email protected]; [email protected]. (1) Johnston, M. V. J. Mass Spectrom. 2000, 35, 585-595. (2) McMurry, P. H. Atmos. Environ. 1999, 34, 1959-1999. (3) Allen, J.; Gould, R. K. Rev. Sci. Instrum. 1981, 52, 804-809. (4) Jayne, J. T.; Leard, D. C.; Zhang, X.; Davidovits, P.; Smith, K. A.; Kolb, C. E.; Worsnop, D. R. Aerosol Sci. Technol. 2000, 33, 49-70. (5) Tobias, H. J.; Kooiman, P. M.; Docherty K. S.; Ziemann, P. J. Aerosol Sci. Technol. 2000, 33, 170-190. (6) Prather, K. A.; Nordmeyer, T.; Salt, K. Anal. Chem. 1994, 66, 1403-1407. (7) Carson, P. G.; Neubauer, K. R.; Johnston, M. V.; Wexler, A. S. J. Aerosol Sci. 1995, 26, 535-545. (8) Murphy, D. M.; Thomson, D. S. Aerosol Sci. Technol. 1995, 22, 237-249. 10.1021/ac001166l CCC: $20.00 Published on Web 04/06/2001

© 2001 American Chemical Society

can measure an entire mass spectrum for a single aerosol particle. These laser-based methods are important because they can characterize a single, multicomponent aerosol particle in a single laser shot. One of the primary goals of laser-based single-particle mass spectroscopy is to develop a more robust and quantitative ionization scheme. In many designs, a single UV laser both ablates and ionizes the aerosol particle.6,7 Although this scheme is simple and powerful in the qualitative description of single aerosol particles, it falls short in a few areas. First, since the ions are born in a dense plume in this single-laser approach, there is the potential for ion-neutral molecule reactions, which prevent the ultimate ion concentrations from quantitatively reflecting the composition of the particle. These ion-molecule processes, referred to as “matrix effects”, often skew the relative intensity of features in mass spectra in favor of low-ionization-potential components.9,10 Second, a single photon of ultraviolet light is insufficient to ionize most molecules, so that this single-laser approach depends on multiphoton ionization. In favorable cases, multiphoton ionization may result in interpretable spectra, but in general, it produces complex, highly fragmented spectra. Third, incomplete vaporization of micrometer-sized particles causes the technique to be more sensitive to the outermost layers of the particle than the core.11,12 In general, single-laser ablation techniques place aerosol particles into broad categories, such as organic, sulfate, sea salt, or some combination thereof, but quantitative speciation is difficult or impossible, especially for organic molecules. The work of Prather and co-workers13 and Zelenyuk et al.14 explores the possibility of separating the evaporation and ionization steps using a two-laser approach. In these experiments, a CO2 laser evaporates the aerosol particle prior to multiphoton ionization of the resulting vapor plume by the 266-nm radiation from a frequency-quadrupled Nd:YAG laser13 or the 193-nm light from an ArF excimer laser.14 The time delay between the evaporation and ionization lasers allows the vapor to expand to low density before it is ionized, eliminating some problems associated with (9) Ge, Z.; Wexler, A. S.; Johnston, M. V. Environ. Sci. Technol. 1998, 32, 32183223. (10) Reilly, P. T. A.; Lazar, A. C.; Gieray, R. A.; Whitten, W. B.; Ramsey, J. M. Aerosol Sci. Technol. 2000, 33, 135-152. (11) Carson, P. G.; Johnston, M. V.; Wexler, A. S. Aerosol Sci. Technol. 1997, 26, 291-300. (12) Ge, Z.; Wexler, A. S.; Johnston, M. V. J. Phys. Chem. A 1998, 102, 173180. (13) Morrical, B. D.; Fergenson, D. P.; Prather, K. A. J. Am. Soc. Mass Spectrom. 1998, 9, 1068-1073. (14) Zelenyuk, A.; Cabalo, J.; Baer, T.; Miller, R. E. Anal. Chem. 1999, 71, 18021808.

Analytical Chemistry, Vol. 73, No. 10, May 15, 2001 2317

matrix effects. Also, the lower threshold for gas-phase ionization, compared to the condensed phase,14 permits the use of low ionization laser fluences resulting in less fragmentation in the mass spectra. Nevertheless, multiphoton ionization schemes do not lend themselves to quantitative analysis. In particular, it is difficult to control the number of absorbed photons, which is crucial in determining relative ionization efficiencies and fragmentation patterns. For these reasons, other groups have suggested13,15 employing a vacuum ultraviolet (VUV) laser to perform the ionization. Near-threshold VUV laser photoionization is a useful tool because it produces ions with little internal energy and causes less fragmentation than techniques such as electron impact and multiphoton ionization. In particular, 10.5-eV radiation (118 nm), produced by frequency tripling the third harmonic (355 nm) of a Nd:YAG laser in Xe gas, has been shown to be both experimentally accessible and applicable to many classes of organic molecules. Chen and co-workers demonstrated fragment free mass spectra for several aromatic compounds.16 Van Bramer and Johnston17,18 measured mass spectra for a large variety of aliphatic molecules, which typically fragment more easily than aromatic compounds. They found that n-alkanes, alkenes, ketones, carboxylic acids, and ethers all produce predominant molecular ion signals. Aldehydes and amines fragment to a greater extent but still show significant parent ion features. Branched alkanes, dienes, alcohols, and esters produce little or no molecular ion but usually a single, dominant fragment ion. In each case, the VUV photoionization mass spectra show larger molecular ion signals, less fragmentation, and a higher fraction of low-energy rearrangement, structurally significant fragment ions than 12-eV electron impact ionization. Single-photon ionization using VUV has the added benefit that it does not depend on intermediate electronic resonances and is, therefore, more general than laser multiphoton ionization. Also, the 10.5-eV photoionization cross sections of many organic molecules are similar,16 ensuring comparable sensitivities. Some groups already make use of these advantages in laser desorption/photoionization of bulk materials. Hanley and co-workers used this combination to analyze organic, biological, and polymeric thin films,19 and de Vries and co-workers20 demonstrated fragment-free mass spectra using VUV ionization of laser-desorbed, jet-cooled bulk solids. No such demonstration has been reported for the laser desorption/ ionization analysis of aerosol particles. We report here the first scheme to incorporate both CO2 laser evaporation and VUV laser ionization to analyze the composition of aerosol particles. The CO2 laser evaporation minimizes matrix effects by ensuring that the ions are born at low density. The VUV light produces little or no fragmentation, allowing for a more accurate identification of an aerosol’s components. In this study, we characterize the operation of this instrument by analyzing particles generated from prepared mixtures of organic liquids. In (15) Cabalo, J.; Zelenyuk, A.; Baer, T.; Miller, R. E. Aerosol Sci. Technol. 2000, 33, 3-19. (16) Arps, J. H.; Chen C. H.; McCann M. P.; Datskou I. Appl. Spectrosc. 2001, 43, 1211-1214. (17) van Bramer, S. E.; Johnston, M. V. J. Am. Soc. Mass Spectrom. 1989, 1, 419-426. (18) van Bramer, S. E.; Johnston, M. V. Appl. Spectrosc. 1991, 46, 255-261. (19) Kornienko, O.; Ada, E. T.; Jillian, T.; Wijesundara, M. B. J.; Hanley, L. Anal. Chem. 2001, 70, 1208-1213. (20) Nir, E.; Huntziker, H. E.; de Vries, M. S. Anal. Chem. 1999, 71, 16741678.

2318 Analytical Chemistry, Vol. 73, No. 10, May 15, 2001

Figure 1. Schematic diagram of the experimental apparatus. It consists of an aerodynamic lens inlet, two stages of differential pumping, a light scattering station for particle sizing and laser timing, and a time-of-flight mass spectrometer. The inset shows the overlap of the CO2 (square) and VUV (circle) laser beams.

addition, we analyze anthracene particles to demonstrate the ability of the method to detect lower vapor pressure (Tboil ) 613 K) solid particles as well. EXPERIMENTAL SECTION The arrangement of the present experiment, shown schematically in Figure 1, is similar to that described in previous papers from this laboratory14,15 and to other laser aerosol time-of-flight mass spectrometers used for on-line analysis of atmospheric aerosol particles.21 The system comprises an aerodynamic lens inlet, two stages of differential pumping, a light scattering station, and a laser-based time-of-flight mass spectrometer. A glass nebulizer (Meinhard) generates liquid aerosol particles directly from the liquid sample and solid particles from a dilute solution of the solid in a volatile solvent. The particles (1-3 mm in diameter) from an external gas stream enter the aerodynamic lens through a 100-µm, flow-limiting orifice. The aerodynamic lens consists of a series of orifices of successively decreasing diameter, a design based on the work of McMurry and co-workers.22,23 The lens focuses aerosol particles onto a well-defined axis, greatly increasing the efficiency with which we detect them. The focused particles accelerate through two stages of differential pumping to speeds of ∼100 m/s before entering the chamber containing the time-of-flight mass spectrometer. The particles then pass through two 532-nm diode laser beams placed 10 cm apart. Separate photomultiplier tubes detect the scattered light from the diode lasers, and a digital timing circuit calculates the velocity of the particle based on the time delay between the two scattered light signals. The circuit then triggers the pulsed lasers to fire when the particle arrives in the mass spectrometer. A pulsed TEA-CO2 laser (Lumonics) producing 30-500 mJ/ pulse of radiation near 10.6 µm vaporizes the particle prior to ionization. A NaCl lens focuses this light into the chamber to a spot size of ∼1 mm. After a delay of 2-3 µs, a 118.5-nm laser (21) Noble, C. A.; Prather, K. A. Environ. Sci. Technol. 1996, 30, 2667-2680. (22) Liu, P.; Ziemann, P. J.; Kittelson, D. B.; McMurry, P. H. Aerosol Sci. Technol. 1995, 22, 293-313. (23) Liu, P.; Ziemann, P. J.; Kittelson, D. B.; McMurry, P. H. Aerosol Sci. Technol. 1995, 22, 314-324.

beam, produced by frequency tripling 10 mJ of the 355-nm output of a Nd:YAG laser (Continuum) in a mixture of Xe and Ar gas, ionizes the vapor cloud for time-of-flight mass analysis. The main chamber and the Xe cell are coupled by a LiF lens that recollimates the 355-nm light and loosely focuses the copropagating 118.5-nm light into the chamber. Based on the position of the 355-nm focus in the Xe cell and the curvature of the LiF lens, we calculate a spot size of 1 mm for the VUV. The unfocused, residual 355-nm light does not appear to contribute to the signals we observe. The position of the VUV laser focus is offset from the CO2 laser focus by 0.25 mm along the particle beam axis to account for the motion of the particle center of mass, as shown by the inset of Figure 1. The following section describes in more detail how these delay and alignment parameters are chosen. Since particles of different sizes, and thus velocities, travel different distances during the 2-3-µs delay, the 0.25-mm offset is not optimal for every particle. Consequently, vapor plumes originating from particles of different size do experience slightly different VUV intensities. However, the difference in the distance traveled by the smallest and largest particles that we have detected (300 nm-5 µm) is only ∼0.3 mm, well within the 1-mm spot of the VUV laser. One could employ a variable delay to account for this polydispersity since the particle’s velocity is measured in advance of the laser triggers, though in these experiments it is not necessary. Incomplete evaporation by the CO2 laser, as a result of either power instabilities or particle trajectories that sample less intense portions of the CO2 laser beam, could also contribute to shot-toshot fluctuations in intensity. None of these issues appear to be a problem in our present study, as the aerosols are fairly monodisperse and the CO2 energy is far enough above the threshold for complete evaporation in nearly every shot. We find absolute shotto-shot intensity fluctuations are only ∼15%, likely owing to power fluctuations in VUV intensity alone. In general, the particles of interest may contain both ionic and neutral components. The CO2 laser promotes the ionic species directly into the gas phase, and the instrument detects them without ionization by the VUV laser. Similar behavior has been observed for laser desorption in thin liquid jets.24 Thus, to ensure that all spectral features have the same origin in time, we keep the ionization region field free during the application of the laser pulses, and pulse the extraction field of the time-of-flight spectrometer 500 ns after the VUV laser is fired. The relative detection efficiency for ionic and neutral species is a complex topic that will be addressed in a later paper. Since the focus here is on the characterization of the VUV ionization, we will restrict our discussion to the neutrals that are evaporated by the CO2 laser and ionized by the VUV laser. Each firing of the CO2 and VUV laser produces an entire mass spectrum corresponding to a single aerosol particle. A digital oscilloscope (HP, Infinium) digitizes each mass spectrum and transfers it to a PC through a GPIB connection. While the experiment operates at an average frequency of 10 Hz, aerosol particles arrive into the chamber at irregular intervals. If the digital timing circuit does not receive a light scattering signal within 100 ms, it fires the CO2 laser and the flashlamps of the Nd:YAG laser so that their temperatures remain constant. These blank firings produce empty spectra since the (24) Watterburg, A.; Sobott, F.; Barth, H. D.; Brutschy, B. Eur. Mass Spectrom. 1999, 5, 71-76.

Figure 2. Mass spectra of m-nitrotoluene particles as a function of the CO2 laser fluence. The inset shows the integrated intensity of the signal vs CO2 laser fluence with the dashed vertical line indicating the threshold fluence of 5.0 × 107 W/cm2. The leftmost point on the inset plot is the origin (0,0), and the other four points correspond to the four spectra shown in the figure.

lasers do not hit a particle, but the data acquisition is still triggered. The control software sorts out the empty time-of-flight spectra after the data collection is complete. Typically, we record a single mass spectrum by transferring 500 single-particle spectra to the PC and averaging the 100-250 spectra that contain data. RESULTS AND DISCUSSION Optimization of Laser Parameters. A consequence of using two lasers is that there are a large number of parameters available for optimizing the performance. In particular, these include the power, focal volume, alignment, and timing of both the CO2 and VUV lasers. To complicate matters further, many of these parameters are coupled. For example, as mentioned in the Experimental Section, the focal positions of the two lasers must be offset to account for the motion of the center of mass of the particle/vapor plume. Since we require that the particle is fully evaporated before firing the VUV laser, and the evaporation rate depends on the power of the CO2 laser, the optimum delay (and thus alignment) is a function of the CO2 laser power. To converge on a standard set of parameters, it is prudent to fix the most critical one and optimize the others. Since we are concerned here with detecting species with little fragmentation, we choose to fix the CO2 laser power such that the particle is completely (or nearly completely) vaporized while imparting a minimum of internal energy to the gas molecules. Figure 2 illustrates how the mass spectrum of m-nitrotoluene particles, generated by CO2 laser evaporation and VUV laser ionization, depends on the energy of the CO2 laser. Each spectrum represents an average of ∼100 single-particle spectra. The inset shows that, beyond a minimum laser power, the total integrated intensity of the spectra is nearly constant with respect to the power of the CO2 laser. This constant intensity implies that the evaporation is nearly complete above this threshold energy. The mass spectra also show that the amount of fragmentation increases as Analytical Chemistry, Vol. 73, No. 10, May 15, 2001

2319

Figure 3. Energy level diagram showing the two pathways to fragmentation in the mass spectra. In the case of high CO2 laser fluence, chemistry occurs on the ground potential energy surface (M f A + B), and the VUV laser ionizes the neutral reaction product, A. In the case of moderate CO2 laser fluence, the neutral reaction does not occur but the ions have enough energy to dissociate also producing the fragment ion A+. Small CO2 laser fluences produce signal from only the parent ion, M+.

Figure 4. (A) Mass spectrum obtained by averaging the CO2 plus VUV laser spectra of 100 aerosol particles generated from a mixture of aniline, benzyl alcohol, and m-nitrotoluene. (B) Representative single laser shot mass spectrum of the same mixture. The shot-toshot fluctuation of the relative intensities is (15%.

the CO2 laser power increases. Fragmentation can arise from either neutral chemistry initiated by the CO2 laser or ion decomposition, which is a function of the ion’s internal energy. Figure 3 depicts these two pathways to fragmentation. In the case of neutral chemistry, the CO2 laser heats the molecules sufficiently to enable them to surmount one or more reaction barriers, and the VUV laser ionizes the neutral reaction products that enter the gas phase. For molecules with high reaction barriers, the CO2 laser may add sufficient energy to evaporate the particle without initiating chemical reactions. Nevertheless, the evaporated neutral molecules may have greater internal energy than room-temperature molecules. When the VUV laser ionizes these excited molecules, the ions may be born with sufficient internal energy to undergo unimolecular decomposition on the ion potential energy surface. Since the dissociation energy for the m-nitrotoluene ion (1.6 eV)25 is considerably less than for the neutral molecule (3.3 eV, estimated from known bond enthalpies), the fragments we observe in this case likely come from the ion dissociation. The results in Figure 2 show that a CO2 laser power of 5 × 107 W/cm2 is sufficient to evaporate the particle without producing excessive fragmentation in the spectra, and we typically collect data at slightly higher values (∼6 × 107 W/cm2) to ensure shotto-shot stability. We further find, by examining the signal magnitude as a function of the delay between lasers, that a 2-3µs delay is sufficient to ensure that the particle is completely evaporated for this laser power. The alignment offset between the CO2 laser and VUV laser along the particle beam axis is set to be 0.25 mm, because, on average, the center of mass of the particle/ vapor plume travels 0.25 mm during this 2-3-µs delay. These results are specific to m-nitrotoluene and are not necessarily representative of all molecules. Other molecules may absorb the

CO2 laser radiation more efficiently, necessitating the use of lower CO2 laser powers. In addition, molecules may have lower ionization potentials or lower energy pathways to ion decomposition. However, the results do suggest that these settings produce vapor plumes with only modest amounts of energy, and it is logical to expect significant parent ion signals for many molecules. Analysis of Known Mixtures. Analyzing a mixture of organic compounds containing m-nitrotoluene (m-NT), aniline (An), and benzyl alcohol (BA) is a test of this approach. These compounds represent three different functional groups and therefore absorb the CO2 radiation with differing efficiencies. Figure 4 shows a mass spectrum of a particle generated from a mixture having a molar ratio of 0.14 (m-NT):0.37 (An):0.49 (BA). The bottom trace is a single-particle spectrum and the top trace is the result of averaging ∼100 single-particle spectra. This single-shot spectrum is well representative of the averaged spectrum. In fact, we measure the shot-to-shot standard deviation in the absolute integrated mass spectral feature intensities to be only 15%. (Note that while the line widths for features in the averaged spectrum increase with mass as expected, this trend happens to be reversed for this particular single-shot spectrum. Thus, the maximums do not reflect the integrated intensity.) The spectra are easy to assign, since the parent ion signal is dominant in each case. Only the benzyl alcohol fragments to an observable extent, producing a feature at mass 79 through the loss of HCO. Since the appearance potential of the HCO loss channel (10.3 eV)26 is less than the energy of the VUV photon (10.5 eV), we expect to see this daughter ion. It is useful to compare the above results with those obtained by using 266-nm multiphoton ionization rather than VUV ionization. Figure 5 shows the mass spectra of the same mixture vaporized by the CO2 laser and ionized with two different powers

(25) Baer, T.; Morrow, J. C.; Shao, J. D.; Olesik, S. J. Am. Chem. Soc. 1988, 110, 5633-5638.

(26) Selim, E. T. M.; Rabbih, M. A.; Fahmey, M. A. Indian J. Pure Appl. Phys. 1987, 25, 451.

2320 Analytical Chemistry, Vol. 73, No. 10, May 15, 2001

Figure 5. Mass spectra obtained by using 266-nm multiphoton ionization at fluences of 2 × 107 (lower trace) and 6 × 107 W/cm2 (top trace) rather than VUV ionization. Although there is little fragmentation at the lower fluence, there is no signal from mnitrotoluene. At a fluence only 3 times higher, the spectrum is highly fragmented. Table 1. Measured Relative Photoionization Cross Sections at 10.5 eV relative cross section aniline benzyl alcohol m-nitrotoluene a

2.5 ( 0.9 1.5 ( 0.5 1a

Values normalized to cross section for m-nitrotoluene.

of 266-nm light. Using 100 mJ/pulse (2 × 107 W/cm2) of UV results in the spectrum shown in the bottom trace of Figure 5. Although there is little fragmentation in the spectrum, mnitrotoluene does not appear in the spectrum at all. Two photons of 266-nm light are insufficient to ionize m-nitrotoluene; thus, it does not appear at this UV flux where three photon excitation is unlikely. Raising the power by less than 1 order of magnitude results in the top trace of Figure 5. In that case, the spectrum is highly fragmented making a qualitative, much less a quantitative, description difficult or impossible. Clearly, the VUV ionization is a more robust tool in this case. To obtain a quantitative analysis from the VUV ionization, we must account for the differences in the relative ionization cross sections for the three molecules. The relative cross sections were measured by independently introducing a known pressure of each component’s vapor into the spectrometer and firing the VUV laser to produce a mass spectrum for the room-temperature gas. The integrated intensity of these spectra, normalized to the pressure, provides the relative, thermally averaged ionization cross section for each gas. We use an ion gauge to measure the pressure and assume that its response is the same for all three gases. Based on known ion gauge sensitivities for other substituted benzenes,27 this assumption could cause as much as 25% error in our measured cross sections. Table 1 shows the results for the three components of the test mixture. Comparing the relative intensity of the features in the spectrum, adjusted with the ionization cross section, directly yields the relative concentration of each component in the mixture. (27) Summers, R. L. NASA Technical Note TD-N-5285, 1969.

Figure 6. Plot comparing the measured mole fraction to the known mole fraction for a given component of one of four aniline/benzyl alcohol/m-nitrotoluene mixtures.(b, aniline; 4, benzyl alcohol; 0, m-nitrotoluene) The solid line represents identical measured and known concentrations, and the error bars reflect only the uncertainty in the ionization cross-section measurements. The dashed lines indicate the three different components of one of the four mixtures.

Figure 6 compares the measured concentrations with the actual ones for four different mixtures of aniline, m-nitrotoluene, and benzyl alcohol. The plot shows 12 data points, one for each of the three components of the four mixtures, as well as a line of unity slope. The error bars reflect only the uncertainty in the ionization cross sections. The trend in the data is consistent with the ideal line, indicating that the detection scheme is robust with respect to both the identity of the component and its concentration. It follows that the CO2 laser evaporation step produces a vapor plume whose composition is representative of the particle’s composition. As another test of the applicability of our approach, we analyze anthracene particles. Since anthracene’s boiling point (613 K) is much higher than those of the organic liquids used in the previous experiment, it provides a critical test of the CO2 laser’s ability to evaporate the lower vapor pressure components of particles found in the atmosphere. Anthracene is a polycyclic aromatic hydrocarbon (PAH), a class of molecules that constitutes an important component of organic aerosols because of its potential heath risk to humans.28 These particles are generated by making aerosols from a dilute solution of anthracene in 1,2-dichloroethane and evaporating the solvent molecules prior to analysis. Figure 7 shows a mass spectrum resulting from CO2 laser evaporation and VUV laser ionization of a single anthracene particle under the same experimental conditions used for the mixture analysis. Here again, the dominant feature in the spectrum corresponds to the parent mass (m/z ) 178). These results demonstrate the potential utility of VUV ionization in the identification of lower vapor pressure components, PAHs in particular, in atmospheric aerosols. (28) Finlayson-Pitts, B. J.; Pitts, J. N. Chemistry of the Upper and Lower Atmosphere: Theory, Experiments and Applications; Academic Press: New York, 2000.

Analytical Chemistry, Vol. 73, No. 10, May 15, 2001

2321

Figure 7. Single-shot mass spectrum of an anthracene particle. The spectrum shows predominantly parent ion signal.

APPLICATION TO ATMOSPHERIC AEROSOLS This work shows the potential for detecting organic molecules in atmospheric aerosol particles; however, it is only the first step. There are several considerations that dictate the applicability of the technique, including sensitivity, the concentration and types of organic compounds found in typical aerosols, and the complexity of those real mixtures. With regard to the latter issue, we would not expect to analyze real aerosols as completely as we have done for our test aerosols. While the analysis would be straightforward if all molecules produced only parent ion signal, many molecules (for example, those containing aliphatic hydrocarbon chains) have lower ion dissociation energies than the aromatic compounds in this study. These molecules could fragment and cause some ambiguity in assigning features in the mass spectrum. However, difficulties in analyzing complex mixtures are inherent in any form of mass spectrometry. Our results show that this technique is promising and better able to analyze mixtures than existing methods. Sensitivity is another important practical issue. We show here that the aerosols are nearly completely evaporated by the CO2 laser, and on the basis of our previous measurement of the velocity of the molecules in the vapor plume,15 the 1-mm spot of the VUV is large enough to encompass a large fraction of the plume at a 2-3-µs delay. Therefore, the sensitivity of this technique rests on the ionization efficiency of the VUV laser. On the basis of measured ion currents from molecules with known 118-nm ionization cross sections (for example, NO), we estimate that the laser produces between 1010 and 1011 VUV photons per laser shot. This results in an ionization efficiency of 10-6-10-5 ions/molecule for molecules with a typical ionization cross section on the order of 10-18 cm2 and a probe volume of ∼10-3 cm3. Thus, an aerosol particle must contain a minimum of 106 molecules of one particular component to produce a signal in a single laser shot. (This result is consistent with our signal magnitudes; however, the excessive electrical noise of our CO2 laser hampers our signal-to-noise ratio as is evident in Figure 4.) As a practical example, we consider (29) Sweet, C. W.; Gatz, D. F. Atmos. Environ. 1998, 32, 1129-1133. (30) Hepburn, J. W. In Laser Techniques In Chemistry; Myers, A. B., Rizzo, T. R., Eds.; John Wiley and Sons: New York, 1995; Chapter 5.

2322

Analytical Chemistry, Vol. 73, No. 10, May 15, 2001

PAH molecules. The concentration of individual, particulate-phase PAHs in the troposphere is ∼1 ng/m3,28 or one part in 104 of the total aerosol mass of ∼10 µg/m3.29 This means that a single particle having this average PAH concentration would have to contain about 104 × 106 ) 1010 molecules to produce a detectable signal. This corresponds to a particle that is 1-2 µm in diameter. Other typical organic components, such as organic acids and ketones, may have concentrations up to 5 ng/m3 and be detected in smaller particles. It would be advantageous to produce higher fluences of VUV; however, more efficient VUV generation techniques, such as resonant difference frequency mixing,30 are perhaps more technically demanding than is practical for a field instrument. Still, this sensitivity is comparable to other two-step desorption/ionization schemes currently used for field measurements. For example, Jayne et al.4 reported an efficiency of 10-6 ions/molecule for dioctyl phthalate in their flash vaporization/electron impact ionization system. One must make further considerations to assess the potential of the technique to be quantitative for atmospheric aerosols. As with all mass spectrometric techniques, a quantitative analysis of mixtures requires a calibration of the ionization efficiency for each component. These sensitivities may be either measured or calculated. In either case, quantization by VUV photoionization is a marked improvement over existing single-particle mass spectrometry techniques that rely on single-laser, multiphoton, ablation-ionization schemes. A potentially more limiting problem concerns the relative detection efficiencies of neutral and ionic components of a particular mixture. As we stated in the Experimental Section, ionic components of a mixture may be directly promoted into the gas phase and detected without the use of an ionizing laser. Preliminary results from our laboratory show that the spectral intensities for these “evaporated ions” are not linear with concentration and may depend on the size of the particle. We are currently investigating this interesting problem in more detail. CONCLUSION Single-particle mass spectrometry using CO2 laser evaporation of the particle prior to ionization by a VUV laser is an improvement over previous single-laser and two-laser configurations. This approach enables superior qualitative analysis of the composition of organic aerosols because the near-threshold, VUV ionization scheme produces little fragmentation in the mass spectra. Analyzing known mixtures of organic compounds shows that the technique is also quantitative and robust with respect to the identity of the component and its concentration. We further demonstrate the applicability of the technique to low vapor pressure organic atmospheric aerosols by analyzing anthracene particles. Work is presently underway to apply this method to mixtures of both ionic and neutral species, as well as to heterogeneous aerosol particles. ACKNOWLEDGMENT We gratefully acknowledge support from AFOSR, Grant F49620-99-1-0064, NSF, Grant CHE 9727788, and DARPA, Grant F49620-98-1-0268. We also thank Dr. Doug Worsnop and Professor Peter McMurry for the use of their aerodynamic lens design. Received for review September 28, 2000. Accepted February 13, 2001. AC001166L