We wish to report the successful development and application of a new ionization source for the on-line analysis of condensed-phase organic compounds by mass spectrometry. Photoelectron resonance capture ionization (PERCI) provides improved real-time and on-line identification capabilities for condensed-phase organics over alternate ionization methods such as conventional electron impact, resonance-enhanced multiphoton ionization (REMPI), and chemical ionization. The relatively simple mass spectra produced by PERCI make it well suited to study heterogeneous reactions on organic particles. PERCI operates on the principle of a tunable, laser-induced photoelectric effect to controllably generate low-energy electrons (<1 eV). Ionization occurs when these electrons resonantly attach, through associative and/or dissociative mechanisms, to nearby gas-phase molecules.1, 2 PERCI provides an exceptionally soft and sensitive ionization, which generates primarily intact molecular ions. Reducing the amount of fragmentation not only simplifies component identification, but also provides an added inherent sensitivity since the analyte signal is concentrated to fewer ion peaks. When used in conjunction with mass spectrometry, PERCI should have wide ranging applications from identification and measurement of ultratrace quantities of organic pollutants to the direct study of heterogeneous reactions occurring on atmospheric particles.
A relatively new class of mass spectrometers has been under development with the unique ability to analyze particles on-line and in real time.3 Traditionally more successful for the analysis of inorganic particles, this new class of instruments has recently been applied to organic compounds;4 however, fundamental limitations, primarily resulting from extensive molecular fragmentation, still exist for the identification of some important classes of organic particles.5 The integration of PERCI into particle mass spectrometers is a promising method for overcoming these limitations and providing complete qualitative pictures of the organic composition of these particles.
Here, for the first time, the exceptionally soft and sensitive nature of PERCI is exemplified by monitoring the reactive uptake of atmospheric ozone by particle-phase oleic acid, which is a well-studied heterogeneous system and has direct relevance to atmospheric chemistry. This chemical system is often used as a model for complex multicomponent atmospheric particles. Furthermore, surface-active organic components, such as oleic acid, are ubiquitous components of atmospheric particles,6 at times making up 50% of the total particulate mass.7 These organic components impact direct and indirect radiative properties of the particles,8, 9 reactively take up atmospheric trace gases (e.g., O3, OH., and NOx) through heterogeneous chemical reactions,10–12 and can pose severe human health hazards.13, 14 As examples of the relevance of this system, oxidative processing of the components of these particles may improve their ability to act as cloud condensation nuclei (CCN) and disrupt the balance of atmospheric oxidants.15
Photoelectron resonance capture ionization mass spectrometry
A schematic of the PERCI source is shown in Fig. 1. Briefly, it consists of a low-energy (sub-mJ) pulsed (10 Hz), tunable (235–300 nm) ultraviolet laser (Opotek Inc., Carlsbad, CA, USA) gently focused onto the surface of a pure aluminum photocathode. The pulsed laser generates a short (7 ns) burst of photoelectrons with a measured quantum efficiency of 6 × 10−4 photoelectrons/incident photon. The photoelectron energy is nominally equal to the difference between the incident photon energy and the photocathode metal work function. Photoelectron flux densities as high as 1021 cm−2 s−1 have been obtained,16 resulting in sensitivities more than an order of magnitude greater than conventional photoionization methods commonly used for particle mass spectrometry.17, 18 Additional details on the PERCI source and a demonstration of its ionization capabilities for gas-phase organics are described elsewhere.16
In order to extend PERCI to particle analysis, efficient particle vaporization is required prior to ionization. A variety of vaporization schemes, both thermal and laser-based, have been developed for particle mass spectrometry and described elsewhere.4, 19, 20 For the experiments discussed here, particle vaporization is achieved thermally by placing a resistively heated filament in the particle beam path. By aligning the photocathode in close proximity to the filament, the molecular vapors from the particle can be efficiently ionized. Two modes of vaporization are used to obtain the results presented here: analysis can be done in real time as the particles impact the hot filament or, to improve signal to noise, after collecting for a short time on the cool filament. The firing of the laser is not synchronized with the particle arrival and subsequent vaporization.
It must be noted that the first-generation vaporization source used in these first experiments presents an obstacle to providing quantitative measurements using our instrument. The filament temperature, particle deposition efficiency, and vaporization efficiency are examples of three variables with this particular setup that are not accurately controlled and which are necessary for successful quantitative analysis.
Mass analysis of the PERCI anions is achieved with a time-of-flight mass spectrometer (R. M. Jordan, Inc., Grass Valley, CA, USA) operating in reflectron mode. Each spectrum is an accumulation of 50 laser shots. The shot-to-shot variation for signals measured on gas-phase samples range from 10–30%, attributed mostly to fluctuations in laser intensity. Data was acquired at 1 GS/s using a digital oscilloscope (WavePro 7000, LeCroy, Chestnut Ridge, NY, USA). The working mass range of the instrument at this sampling rate is m/z 0–500. The measured resolution (m/Δm) for the mass spectrum shown in Fig. 2 is 510 at m/z 187 and 310 at m/z 281.
Particle generation and flow reactor
Polydisperse oleic acid (9-octadecenoic acid) particles are produced by nebulization (concentric pneumatic nebulizer, J.E. Meinhard Associates, Santa Ana, CA, USA) of a dilute solution (∼500 ppm) of oleic acid in 15% v/v ethanol in water. The particles pass through a diffusion drier to remove residual solvent. The size distribution of the dried aerosol was found to have a geometric mean of 155 nm with a geometric standard deviation of 2.00. Particle number densities were on the order of 105 cm−3. The particles are introduced into a flow reactor via a glass tube (0.32 cm i.d.), centered within a 2.54 cm i.d. flow reactor, which acts as the aerosol injector. Ozone is generated by passing air through an electric discharge ozonator (model 8340, Mathesen). The concentration of ozone is determined photometrically in a 100 mm quartz cell by measuring the absorbance of ozone at 253.65 nm (σ = 1141 × 10−20 cm2).21 The ozone is introduced into the flow reactor (2.54 cm i.d.) upstream of the aerosol injector. The flow of the gas is laminar (Re < 100) at one atmosphere total pressure within the flow reactor. A typical partial pressure of the ozone in these experiments was 1.8 × 10−4 atm, high enough relative to the particle concentration for these reactions to follow pseudo-first-order kinetics. The ozone exposure time of the particles is dictated by the position of the aerosol injector at a given flow rate of the particle and gas phases. The flow rate in the reactor for these experiments was held constant at 0.5 L/min, such that positioning the central particle injector tube anywhere from 1–25 cm from the end of the flow reactor results in reaction times from ∼1–10 s.
Reacted particles are introduced into the mass spectrometer through a differentially pumped inlet and focused into a beam using a series of aerodynamic lenses. A 220 μm diameter critical orifice at the entrance of the inlet keeps the sampling flow rate constant at 0.70 L/min. Characterization of the inlet and further details have been reported elsewhere.22 A short length of Teflon tubing (43 cm, 4 mm i.d.) was used to couple the flow reactor to the particle inlet. The flow through this tubing added an additional 0.5 s to the particle reaction time.
RESULTS AND DISCUSSION
Figure 2 underscores the exceptionally soft and sensitive nature or PERCI, as applied to the heterogeneous oxidation of oleic acid particles by gas-phase ozone. This spectrum was acquired by collecting reacted oleic acid particles for 1 min on the cold filament before vaporizing. The major oxidation products measured by PERCI-MS, detected as [MH]−, are predicted by two reaction pathways during the ozonolysis of oleic acid. As reported in the literature23–25 and illustrated in Scheme 1, ozone adds across the double bond in oleic acid to form a primary ozonide (1), which can decompose by two alternative pathways to produce 1-nonanal (142 Da) (2), and azelaic acid (188 Da) (3) or nonanoic acid (172 Da) (4) and 9-oxononanoic acid (158 Da) (5). Azelaic acid and nonanoic acid are formed by rearrangement of the Criegee intermediates (I) and (II), respectively. Unreacted oleic acid is also present as the [MH]− peak at m/z 281.
All four of the expected oxidized products are clearly identified in Fig. 2 from their [MH]− ions, including 1-nonanal, which to the authors' knowledge has been identified only once previously in the particle phase.26 The absence of 1-nonanal in the particle-phase mass spectra from previous studies was generally ascribed to its greater volatility, and hence loss to the gas phase. This does not appear to be the case, and some other loss mechanism may be operative. It must be acknowledged, however, that differences in particle concentration, a factor in the gas-particle partitioning, from those used by other workers in previous experiments may also be a reason for the observed particle-phase 1-nonanal in our experiments.
The sensitivity of PERCI toward the unfragmented molecular ions greatly simplifies the mass spectrum and enables the direct measurement of previously unreported reaction products. While there are a number of minor peaks present in addition to the four major ions, of particular interest are the higher mass peaks observed, as shown in the expanded spectrum in Fig. 3, at m/z 285, 299, 301, 311, 313, and 315. These products, having molecular weights greater than that of oleic acid, allude to the occurrence of a complex series of products, intermediates, and mechanisms involved in this system beyond that presented in the simplified Scheme 1. These product ions have been tentatively assigned as higher order oxygenates, such as trioxalanes and peroxides, in accord with the Criegee mechanism for the ozonolysis of alkenes.27 For example, as illustrated in Scheme 2, the ion fragment at m/z 299 is hypothesized to be the [MH]− ion from the secondary ozonide formed from a 1,3-dipolar addition reaction of 1-nonanal (2) with the alkyl-terminated Criegee intermediate (II). Though hydration of the oleic acid double bond would also lead to a product with the same m/z, this reaction is unlikely since acidic conditions are necessary for this reaction to occur to any significant extent.
What is particularly fascinating about the presence of this secondary ozonide product is that the two reactants in this process arise from separate cleavage routes of the primary ozonide (1), exposing the reaction pathways shown in Scheme 1 as overly simplified. The implications of these findings are that multiple routes exist for both the formation and loss of the four major products and additional highly oxidized products, which will significantly alter the final composition of the particle from what is predicted based on Scheme 1. An in-depth discussion of the additional product ions and the corresponding mechanisms and pathways leading to their formation is beyond the scope of this communication and will be the focus of a future publication.
The high sensitivity and fast temporal response of PERCI-MS can also be used to great advantage to monitor fast heterogeneous reactions in real time. Figure 4 shows the evolution of the oleic acid particle mass spectrum with increasing exposure time to a constant ozone partial pressure of 1.8 × 10−4 atm. This series of spectra was acquired by vaporizing the particles in real time as they came in contact with the continually heated filament. This experiment paints a clear picture of the evolution of the heterogeneous chemical reaction taking place on the particle as the oleic acid peak at m/z 281 decreases and molecular product peaks at m/z 141, 157, 171, and 187 increase with increasing O3 exposure.
From this data an estimate of the reactive uptake coefficient was determined to be 5 × 10−5 using the method for a polydisperse aerosol described by Hearn and Smith.26 This is roughly 1–2 orders of magnitude lower than that reported elsewhere.12, 23–26 A significant source of error here is likely the difference between the measured particle size distribution and that of the particles responsible for the signal, as a characterization of the analytical figures of merit for both the PERCI source and the vaporization source has not yet been completed. The low estimated uptake coefficient alludes to a bias favoring the high diameter particles. Work is underway to characterize and improve the analytical capabilities of the PERCI particle mass spectrometer. One aspect that must be addressed, for example, is the relative merits of collecting particles on the cool filament vs. immediate vaporization as the particles impact the heated filament. Sampling artifacts appear to result from the choice of vaporization method since there is an observed difference in the relative peak heights of the ions at m/z 171 to those at m/z 187 in Fig. 2 vs. Fig. 4. Despite these current quantitative limitations, it is apparent that the data in Fig. 4 are at least semiquantitative since there is a clear trend of increased product yields with increased ozone exposure.
Finally, a brief comment on the sensitivity of PERCI is warranted. Recently, other methods of on-line particle mass spectrometry have been applied to the study of oleic acid ozonolysis.24, 25 While successful at measuring total organic aerosol, however, a direct measurement of reaction products was greatly limited by excessive fragmentation of the molecular products during the ionization step. Previous attempts to measure this reaction24, 25 in real time have yielded unreacted oleic acid peaks that made up less than 1% and 15% of the total ion signal. By comparison, in the unreacted oleic acid spectrum obtained by PERCI-MS (Fig. 4), the [MH]− peak at m/z 281 is nearly 100% of the total ion signal.
We believe that PERCI-MS represents a significant analytical advance in the study of particle-phase organics and the important atmospheric reactions taking place on them. PERCI-MS has proven to be a useful tool for the identification of reaction products from the heterogeneous reaction of ozone with oleic acid particles. The simplified mass spectrum obtained allows for the identification of minor reaction products in addition to the predicted and previously observed major products, providing new insight into a well-studied heterogeneous system. The results demonstrate the great potential of PERCI to permit the future study of less well-understood, or unknown, chemical reactions by its ability to directly identify the reaction products.
A natural extension of this work is a more quantitative treatment of the kinetic data; however, some barriers currently exist to operation of the PERCI mass spectrometer in a quantitative fashion. The main obstacle is the design of a suitable vaporization source which can be reproducibly aligned with both the particle beam and the PERCI source. A full characterization of the analytical figures of merit of the method is underway and will follow in a future publication. We are in the process of optimizing our PERCI-MS system and hope to approach the heterogeneous reaction kinetics from the standpoint of each product separately. This complete kinetic picture will be crucial for understanding the ultimate fate of the reacted particles and, therefore, the broader implications of these reactions in terms of environmental impact as well as human health.
The authors gratefully acknowledge the financial support of the American Society for Mass Spectrometry, Vermont EPSCoR and the National Center for Environmental Research (NCER) STAR Program, EPA (Fellowship No. 91615301-0).